linux_dsm_epyc7002/drivers/dma/iop-adma.c

1727 lines
48 KiB
C
Raw Normal View History

dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
/*
* offload engine driver for the Intel Xscale series of i/o processors
* Copyright © 2006, Intel Corporation.
*
* This program is free software; you can redistribute it and/or modify it
* under the terms and conditions of the GNU General Public License,
* version 2, as published by the Free Software Foundation.
*
* This program is distributed in the hope it will be useful, but WITHOUT
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
* FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
* more details.
*
* You should have received a copy of the GNU General Public License along with
* this program; if not, write to the Free Software Foundation, Inc.,
* 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
*
*/
/*
* This driver supports the asynchrounous DMA copy and RAID engines available
* on the Intel Xscale(R) family of I/O Processors (IOP 32x, 33x, 134x)
*/
#include <linux/init.h>
#include <linux/module.h>
#include <linux/delay.h>
#include <linux/dma-mapping.h>
#include <linux/spinlock.h>
#include <linux/interrupt.h>
#include <linux/platform_device.h>
#include <linux/memory.h>
#include <linux/ioport.h>
#include <linux/raid/pq.h>
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h percpu.h is included by sched.h and module.h and thus ends up being included when building most .c files. percpu.h includes slab.h which in turn includes gfp.h making everything defined by the two files universally available and complicating inclusion dependencies. percpu.h -> slab.h dependency is about to be removed. Prepare for this change by updating users of gfp and slab facilities include those headers directly instead of assuming availability. As this conversion needs to touch large number of source files, the following script is used as the basis of conversion. http://userweb.kernel.org/~tj/misc/slabh-sweep.py The script does the followings. * Scan files for gfp and slab usages and update includes such that only the necessary includes are there. ie. if only gfp is used, gfp.h, if slab is used, slab.h. * When the script inserts a new include, it looks at the include blocks and try to put the new include such that its order conforms to its surrounding. It's put in the include block which contains core kernel includes, in the same order that the rest are ordered - alphabetical, Christmas tree, rev-Xmas-tree or at the end if there doesn't seem to be any matching order. * If the script can't find a place to put a new include (mostly because the file doesn't have fitting include block), it prints out an error message indicating which .h file needs to be added to the file. The conversion was done in the following steps. 1. The initial automatic conversion of all .c files updated slightly over 4000 files, deleting around 700 includes and adding ~480 gfp.h and ~3000 slab.h inclusions. The script emitted errors for ~400 files. 2. Each error was manually checked. Some didn't need the inclusion, some needed manual addition while adding it to implementation .h or embedding .c file was more appropriate for others. This step added inclusions to around 150 files. 3. The script was run again and the output was compared to the edits from #2 to make sure no file was left behind. 4. Several build tests were done and a couple of problems were fixed. e.g. lib/decompress_*.c used malloc/free() wrappers around slab APIs requiring slab.h to be added manually. 5. The script was run on all .h files but without automatically editing them as sprinkling gfp.h and slab.h inclusions around .h files could easily lead to inclusion dependency hell. Most gfp.h inclusion directives were ignored as stuff from gfp.h was usually wildly available and often used in preprocessor macros. Each slab.h inclusion directive was examined and added manually as necessary. 6. percpu.h was updated not to include slab.h. 7. Build test were done on the following configurations and failures were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my distributed build env didn't work with gcov compiles) and a few more options had to be turned off depending on archs to make things build (like ipr on powerpc/64 which failed due to missing writeq). * x86 and x86_64 UP and SMP allmodconfig and a custom test config. * powerpc and powerpc64 SMP allmodconfig * sparc and sparc64 SMP allmodconfig * ia64 SMP allmodconfig * s390 SMP allmodconfig * alpha SMP allmodconfig * um on x86_64 SMP allmodconfig 8. percpu.h modifications were reverted so that it could be applied as a separate patch and serve as bisection point. Given the fact that I had only a couple of failures from tests on step 6, I'm fairly confident about the coverage of this conversion patch. If there is a breakage, it's likely to be something in one of the arch headers which should be easily discoverable easily on most builds of the specific arch. Signed-off-by: Tejun Heo <tj@kernel.org> Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org> Cc: Ingo Molnar <mingo@redhat.com> Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 15:04:11 +07:00
#include <linux/slab.h>
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
#include <mach/adma.h>
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
#include "dmaengine.h"
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
#define to_iop_adma_chan(chan) container_of(chan, struct iop_adma_chan, common)
#define to_iop_adma_device(dev) \
container_of(dev, struct iop_adma_device, common)
#define tx_to_iop_adma_slot(tx) \
container_of(tx, struct iop_adma_desc_slot, async_tx)
/**
* iop_adma_free_slots - flags descriptor slots for reuse
* @slot: Slot to free
* Caller must hold &iop_chan->lock while calling this function
*/
static void iop_adma_free_slots(struct iop_adma_desc_slot *slot)
{
int stride = slot->slots_per_op;
while (stride--) {
slot->slots_per_op = 0;
slot = list_entry(slot->slot_node.next,
struct iop_adma_desc_slot,
slot_node);
}
}
static void
iop_desc_unmap(struct iop_adma_chan *iop_chan, struct iop_adma_desc_slot *desc)
{
struct dma_async_tx_descriptor *tx = &desc->async_tx;
struct iop_adma_desc_slot *unmap = desc->group_head;
struct device *dev = &iop_chan->device->pdev->dev;
u32 len = unmap->unmap_len;
enum dma_ctrl_flags flags = tx->flags;
u32 src_cnt;
dma_addr_t addr;
dma_addr_t dest;
src_cnt = unmap->unmap_src_cnt;
dest = iop_desc_get_dest_addr(unmap, iop_chan);
if (!(flags & DMA_COMPL_SKIP_DEST_UNMAP)) {
enum dma_data_direction dir;
if (src_cnt > 1) /* is xor? */
dir = DMA_BIDIRECTIONAL;
else
dir = DMA_FROM_DEVICE;
dma_unmap_page(dev, dest, len, dir);
}
if (!(flags & DMA_COMPL_SKIP_SRC_UNMAP)) {
while (src_cnt--) {
addr = iop_desc_get_src_addr(unmap, iop_chan, src_cnt);
if (addr == dest)
continue;
dma_unmap_page(dev, addr, len, DMA_TO_DEVICE);
}
}
desc->group_head = NULL;
}
static void
iop_desc_unmap_pq(struct iop_adma_chan *iop_chan, struct iop_adma_desc_slot *desc)
{
struct dma_async_tx_descriptor *tx = &desc->async_tx;
struct iop_adma_desc_slot *unmap = desc->group_head;
struct device *dev = &iop_chan->device->pdev->dev;
u32 len = unmap->unmap_len;
enum dma_ctrl_flags flags = tx->flags;
u32 src_cnt = unmap->unmap_src_cnt;
dma_addr_t pdest = iop_desc_get_dest_addr(unmap, iop_chan);
dma_addr_t qdest = iop_desc_get_qdest_addr(unmap, iop_chan);
int i;
if (tx->flags & DMA_PREP_CONTINUE)
src_cnt -= 3;
if (!(flags & DMA_COMPL_SKIP_DEST_UNMAP) && !desc->pq_check_result) {
dma_unmap_page(dev, pdest, len, DMA_BIDIRECTIONAL);
dma_unmap_page(dev, qdest, len, DMA_BIDIRECTIONAL);
}
if (!(flags & DMA_COMPL_SKIP_SRC_UNMAP)) {
dma_addr_t addr;
for (i = 0; i < src_cnt; i++) {
addr = iop_desc_get_src_addr(unmap, iop_chan, i);
dma_unmap_page(dev, addr, len, DMA_TO_DEVICE);
}
if (desc->pq_check_result) {
dma_unmap_page(dev, pdest, len, DMA_TO_DEVICE);
dma_unmap_page(dev, qdest, len, DMA_TO_DEVICE);
}
}
desc->group_head = NULL;
}
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
static dma_cookie_t
iop_adma_run_tx_complete_actions(struct iop_adma_desc_slot *desc,
struct iop_adma_chan *iop_chan, dma_cookie_t cookie)
{
struct dma_async_tx_descriptor *tx = &desc->async_tx;
BUG_ON(tx->cookie < 0);
if (tx->cookie > 0) {
cookie = tx->cookie;
tx->cookie = 0;
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
/* call the callback (must not sleep or submit new
* operations to this channel)
*/
if (tx->callback)
tx->callback(tx->callback_param);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
/* unmap dma addresses
* (unmap_single vs unmap_page?)
*/
if (desc->group_head && desc->unmap_len) {
if (iop_desc_is_pq(desc))
iop_desc_unmap_pq(iop_chan, desc);
else
iop_desc_unmap(iop_chan, desc);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
}
}
/* run dependent operations */
dma_run_dependencies(tx);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
return cookie;
}
static int
iop_adma_clean_slot(struct iop_adma_desc_slot *desc,
struct iop_adma_chan *iop_chan)
{
/* the client is allowed to attach dependent operations
* until 'ack' is set
*/
if (!async_tx_test_ack(&desc->async_tx))
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
return 0;
/* leave the last descriptor in the chain
* so we can append to it
*/
if (desc->chain_node.next == &iop_chan->chain)
return 1;
dev_dbg(iop_chan->device->common.dev,
"\tfree slot: %d slots_per_op: %d\n",
desc->idx, desc->slots_per_op);
list_del(&desc->chain_node);
iop_adma_free_slots(desc);
return 0;
}
static void __iop_adma_slot_cleanup(struct iop_adma_chan *iop_chan)
{
struct iop_adma_desc_slot *iter, *_iter, *grp_start = NULL;
dma_cookie_t cookie = 0;
u32 current_desc = iop_chan_get_current_descriptor(iop_chan);
int busy = iop_chan_is_busy(iop_chan);
int seen_current = 0, slot_cnt = 0, slots_per_op = 0;
dev_dbg(iop_chan->device->common.dev, "%s\n", __func__);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
/* free completed slots from the chain starting with
* the oldest descriptor
*/
list_for_each_entry_safe(iter, _iter, &iop_chan->chain,
chain_node) {
pr_debug("\tcookie: %d slot: %d busy: %d "
"this_desc: %#x next_desc: %#x ack: %d\n",
iter->async_tx.cookie, iter->idx, busy,
iter->async_tx.phys, iop_desc_get_next_desc(iter),
async_tx_test_ack(&iter->async_tx));
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
prefetch(_iter);
prefetch(&_iter->async_tx);
/* do not advance past the current descriptor loaded into the
* hardware channel, subsequent descriptors are either in
* process or have not been submitted
*/
if (seen_current)
break;
/* stop the search if we reach the current descriptor and the
* channel is busy, or if it appears that the current descriptor
* needs to be re-read (i.e. has been appended to)
*/
if (iter->async_tx.phys == current_desc) {
BUG_ON(seen_current++);
if (busy || iop_desc_get_next_desc(iter))
break;
}
/* detect the start of a group transaction */
if (!slot_cnt && !slots_per_op) {
slot_cnt = iter->slot_cnt;
slots_per_op = iter->slots_per_op;
if (slot_cnt <= slots_per_op) {
slot_cnt = 0;
slots_per_op = 0;
}
}
if (slot_cnt) {
pr_debug("\tgroup++\n");
if (!grp_start)
grp_start = iter;
slot_cnt -= slots_per_op;
}
/* all the members of a group are complete */
if (slots_per_op != 0 && slot_cnt == 0) {
struct iop_adma_desc_slot *grp_iter, *_grp_iter;
int end_of_chain = 0;
pr_debug("\tgroup end\n");
/* collect the total results */
if (grp_start->xor_check_result) {
u32 zero_sum_result = 0;
slot_cnt = grp_start->slot_cnt;
grp_iter = grp_start;
list_for_each_entry_from(grp_iter,
&iop_chan->chain, chain_node) {
zero_sum_result |=
iop_desc_get_zero_result(grp_iter);
pr_debug("\titer%d result: %d\n",
grp_iter->idx, zero_sum_result);
slot_cnt -= slots_per_op;
if (slot_cnt == 0)
break;
}
pr_debug("\tgrp_start->xor_check_result: %p\n",
grp_start->xor_check_result);
*grp_start->xor_check_result = zero_sum_result;
}
/* clean up the group */
slot_cnt = grp_start->slot_cnt;
grp_iter = grp_start;
list_for_each_entry_safe_from(grp_iter, _grp_iter,
&iop_chan->chain, chain_node) {
cookie = iop_adma_run_tx_complete_actions(
grp_iter, iop_chan, cookie);
slot_cnt -= slots_per_op;
end_of_chain = iop_adma_clean_slot(grp_iter,
iop_chan);
if (slot_cnt == 0 || end_of_chain)
break;
}
/* the group should be complete at this point */
BUG_ON(slot_cnt);
slots_per_op = 0;
grp_start = NULL;
if (end_of_chain)
break;
else
continue;
} else if (slots_per_op) /* wait for group completion */
continue;
/* write back zero sum results (single descriptor case) */
if (iter->xor_check_result && iter->async_tx.cookie)
*iter->xor_check_result =
iop_desc_get_zero_result(iter);
cookie = iop_adma_run_tx_complete_actions(
iter, iop_chan, cookie);
if (iop_adma_clean_slot(iter, iop_chan))
break;
}
if (cookie > 0) {
iop_chan->common.completed_cookie = cookie;
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
pr_debug("\tcompleted cookie %d\n", cookie);
}
}
static void
iop_adma_slot_cleanup(struct iop_adma_chan *iop_chan)
{
spin_lock_bh(&iop_chan->lock);
__iop_adma_slot_cleanup(iop_chan);
spin_unlock_bh(&iop_chan->lock);
}
static void iop_adma_tasklet(unsigned long data)
{
struct iop_adma_chan *iop_chan = (struct iop_adma_chan *) data;
/* lockdep will flag depedency submissions as potentially
* recursive locking, this is not the case as a dependency
* submission will never recurse a channels submit routine.
* There are checks in async_tx.c to prevent this.
*/
spin_lock_nested(&iop_chan->lock, SINGLE_DEPTH_NESTING);
__iop_adma_slot_cleanup(iop_chan);
spin_unlock(&iop_chan->lock);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
}
static struct iop_adma_desc_slot *
iop_adma_alloc_slots(struct iop_adma_chan *iop_chan, int num_slots,
int slots_per_op)
{
struct iop_adma_desc_slot *iter, *_iter, *alloc_start = NULL;
LIST_HEAD(chain);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
int slots_found, retry = 0;
/* start search from the last allocated descrtiptor
* if a contiguous allocation can not be found start searching
* from the beginning of the list
*/
retry:
slots_found = 0;
if (retry == 0)
iter = iop_chan->last_used;
else
iter = list_entry(&iop_chan->all_slots,
struct iop_adma_desc_slot,
slot_node);
list_for_each_entry_safe_continue(
iter, _iter, &iop_chan->all_slots, slot_node) {
prefetch(_iter);
prefetch(&_iter->async_tx);
if (iter->slots_per_op) {
/* give up after finding the first busy slot
* on the second pass through the list
*/
if (retry)
break;
slots_found = 0;
continue;
}
/* start the allocation if the slot is correctly aligned */
if (!slots_found++) {
if (iop_desc_is_aligned(iter, slots_per_op))
alloc_start = iter;
else {
slots_found = 0;
continue;
}
}
if (slots_found == num_slots) {
struct iop_adma_desc_slot *alloc_tail = NULL;
struct iop_adma_desc_slot *last_used = NULL;
iter = alloc_start;
while (num_slots) {
int i;
dev_dbg(iop_chan->device->common.dev,
"allocated slot: %d "
"(desc %p phys: %#x) slots_per_op %d\n",
iter->idx, iter->hw_desc,
iter->async_tx.phys, slots_per_op);
/* pre-ack all but the last descriptor */
if (num_slots != slots_per_op)
async_tx_ack(&iter->async_tx);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
list_add_tail(&iter->chain_node, &chain);
alloc_tail = iter;
iter->async_tx.cookie = 0;
iter->slot_cnt = num_slots;
iter->xor_check_result = NULL;
for (i = 0; i < slots_per_op; i++) {
iter->slots_per_op = slots_per_op - i;
last_used = iter;
iter = list_entry(iter->slot_node.next,
struct iop_adma_desc_slot,
slot_node);
}
num_slots -= slots_per_op;
}
alloc_tail->group_head = alloc_start;
alloc_tail->async_tx.cookie = -EBUSY;
list_splice(&chain, &alloc_tail->tx_list);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
iop_chan->last_used = last_used;
iop_desc_clear_next_desc(alloc_start);
iop_desc_clear_next_desc(alloc_tail);
return alloc_tail;
}
}
if (!retry++)
goto retry;
/* perform direct reclaim if the allocation fails */
__iop_adma_slot_cleanup(iop_chan);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
return NULL;
}
static void iop_adma_check_threshold(struct iop_adma_chan *iop_chan)
{
dev_dbg(iop_chan->device->common.dev, "pending: %d\n",
iop_chan->pending);
if (iop_chan->pending >= IOP_ADMA_THRESHOLD) {
iop_chan->pending = 0;
iop_chan_append(iop_chan);
}
}
static dma_cookie_t
iop_adma_tx_submit(struct dma_async_tx_descriptor *tx)
{
struct iop_adma_desc_slot *sw_desc = tx_to_iop_adma_slot(tx);
struct iop_adma_chan *iop_chan = to_iop_adma_chan(tx->chan);
struct iop_adma_desc_slot *grp_start, *old_chain_tail;
int slot_cnt;
int slots_per_op;
dma_cookie_t cookie;
dma_addr_t next_dma;
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
grp_start = sw_desc->group_head;
slot_cnt = grp_start->slot_cnt;
slots_per_op = grp_start->slots_per_op;
spin_lock_bh(&iop_chan->lock);
cookie = dma_cookie_assign(tx);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
old_chain_tail = list_entry(iop_chan->chain.prev,
struct iop_adma_desc_slot, chain_node);
list_splice_init(&sw_desc->tx_list,
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
&old_chain_tail->chain_node);
/* fix up the hardware chain */
next_dma = grp_start->async_tx.phys;
iop_desc_set_next_desc(old_chain_tail, next_dma);
BUG_ON(iop_desc_get_next_desc(old_chain_tail) != next_dma); /* flush */
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
/* check for pre-chained descriptors */
iop_paranoia(iop_desc_get_next_desc(sw_desc));
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
/* increment the pending count by the number of slots
* memcpy operations have a 1:1 (slot:operation) relation
* other operations are heavier and will pop the threshold
* more often.
*/
iop_chan->pending += slot_cnt;
iop_adma_check_threshold(iop_chan);
spin_unlock_bh(&iop_chan->lock);
dev_dbg(iop_chan->device->common.dev, "%s cookie: %d slot: %d\n",
__func__, sw_desc->async_tx.cookie, sw_desc->idx);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
return cookie;
}
static void iop_chan_start_null_memcpy(struct iop_adma_chan *iop_chan);
static void iop_chan_start_null_xor(struct iop_adma_chan *iop_chan);
/**
* iop_adma_alloc_chan_resources - returns the number of allocated descriptors
* @chan - allocate descriptor resources for this channel
* @client - current client requesting the channel be ready for requests
*
* Note: We keep the slots for 1 operation on iop_chan->chain at all times. To
* avoid deadlock, via async_xor, num_descs_in_pool must at a minimum be
* greater than 2x the number slots needed to satisfy a device->max_xor
* request.
* */
static int iop_adma_alloc_chan_resources(struct dma_chan *chan)
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
{
char *hw_desc;
int idx;
struct iop_adma_chan *iop_chan = to_iop_adma_chan(chan);
struct iop_adma_desc_slot *slot = NULL;
int init = iop_chan->slots_allocated ? 0 : 1;
struct iop_adma_platform_data *plat_data =
iop_chan->device->pdev->dev.platform_data;
int num_descs_in_pool = plat_data->pool_size/IOP_ADMA_SLOT_SIZE;
/* Allocate descriptor slots */
do {
idx = iop_chan->slots_allocated;
if (idx == num_descs_in_pool)
break;
slot = kzalloc(sizeof(*slot), GFP_KERNEL);
if (!slot) {
printk(KERN_INFO "IOP ADMA Channel only initialized"
" %d descriptor slots", idx);
break;
}
hw_desc = (char *) iop_chan->device->dma_desc_pool_virt;
slot->hw_desc = (void *) &hw_desc[idx * IOP_ADMA_SLOT_SIZE];
dma_async_tx_descriptor_init(&slot->async_tx, chan);
slot->async_tx.tx_submit = iop_adma_tx_submit;
INIT_LIST_HEAD(&slot->tx_list);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
INIT_LIST_HEAD(&slot->chain_node);
INIT_LIST_HEAD(&slot->slot_node);
hw_desc = (char *) iop_chan->device->dma_desc_pool;
slot->async_tx.phys =
(dma_addr_t) &hw_desc[idx * IOP_ADMA_SLOT_SIZE];
slot->idx = idx;
spin_lock_bh(&iop_chan->lock);
iop_chan->slots_allocated++;
list_add_tail(&slot->slot_node, &iop_chan->all_slots);
spin_unlock_bh(&iop_chan->lock);
} while (iop_chan->slots_allocated < num_descs_in_pool);
if (idx && !iop_chan->last_used)
iop_chan->last_used = list_entry(iop_chan->all_slots.next,
struct iop_adma_desc_slot,
slot_node);
dev_dbg(iop_chan->device->common.dev,
"allocated %d descriptor slots last_used: %p\n",
iop_chan->slots_allocated, iop_chan->last_used);
/* initialize the channel and the chain with a null operation */
if (init) {
if (dma_has_cap(DMA_MEMCPY,
iop_chan->device->common.cap_mask))
iop_chan_start_null_memcpy(iop_chan);
else if (dma_has_cap(DMA_XOR,
iop_chan->device->common.cap_mask))
iop_chan_start_null_xor(iop_chan);
else
BUG();
}
return (idx > 0) ? idx : -ENOMEM;
}
static struct dma_async_tx_descriptor *
iop_adma_prep_dma_interrupt(struct dma_chan *chan, unsigned long flags)
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
{
struct iop_adma_chan *iop_chan = to_iop_adma_chan(chan);
struct iop_adma_desc_slot *sw_desc, *grp_start;
int slot_cnt, slots_per_op;
dev_dbg(iop_chan->device->common.dev, "%s\n", __func__);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
spin_lock_bh(&iop_chan->lock);
slot_cnt = iop_chan_interrupt_slot_count(&slots_per_op, iop_chan);
sw_desc = iop_adma_alloc_slots(iop_chan, slot_cnt, slots_per_op);
if (sw_desc) {
grp_start = sw_desc->group_head;
iop_desc_init_interrupt(grp_start, iop_chan);
grp_start->unmap_len = 0;
sw_desc->async_tx.flags = flags;
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
}
spin_unlock_bh(&iop_chan->lock);
return sw_desc ? &sw_desc->async_tx : NULL;
}
static struct dma_async_tx_descriptor *
iop_adma_prep_dma_memcpy(struct dma_chan *chan, dma_addr_t dma_dest,
dma_addr_t dma_src, size_t len, unsigned long flags)
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
{
struct iop_adma_chan *iop_chan = to_iop_adma_chan(chan);
struct iop_adma_desc_slot *sw_desc, *grp_start;
int slot_cnt, slots_per_op;
if (unlikely(!len))
return NULL;
BUG_ON(len > IOP_ADMA_MAX_BYTE_COUNT);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
dev_dbg(iop_chan->device->common.dev, "%s len: %u\n",
__func__, len);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
spin_lock_bh(&iop_chan->lock);
slot_cnt = iop_chan_memcpy_slot_count(len, &slots_per_op);
sw_desc = iop_adma_alloc_slots(iop_chan, slot_cnt, slots_per_op);
if (sw_desc) {
grp_start = sw_desc->group_head;
iop_desc_init_memcpy(grp_start, flags);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
iop_desc_set_byte_count(grp_start, iop_chan, len);
iop_desc_set_dest_addr(grp_start, iop_chan, dma_dest);
iop_desc_set_memcpy_src_addr(grp_start, dma_src);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
sw_desc->unmap_src_cnt = 1;
sw_desc->unmap_len = len;
sw_desc->async_tx.flags = flags;
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
}
spin_unlock_bh(&iop_chan->lock);
return sw_desc ? &sw_desc->async_tx : NULL;
}
static struct dma_async_tx_descriptor *
iop_adma_prep_dma_memset(struct dma_chan *chan, dma_addr_t dma_dest,
int value, size_t len, unsigned long flags)
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
{
struct iop_adma_chan *iop_chan = to_iop_adma_chan(chan);
struct iop_adma_desc_slot *sw_desc, *grp_start;
int slot_cnt, slots_per_op;
if (unlikely(!len))
return NULL;
BUG_ON(len > IOP_ADMA_MAX_BYTE_COUNT);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
dev_dbg(iop_chan->device->common.dev, "%s len: %u\n",
__func__, len);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
spin_lock_bh(&iop_chan->lock);
slot_cnt = iop_chan_memset_slot_count(len, &slots_per_op);
sw_desc = iop_adma_alloc_slots(iop_chan, slot_cnt, slots_per_op);
if (sw_desc) {
grp_start = sw_desc->group_head;
iop_desc_init_memset(grp_start, flags);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
iop_desc_set_byte_count(grp_start, iop_chan, len);
iop_desc_set_block_fill_val(grp_start, value);
iop_desc_set_dest_addr(grp_start, iop_chan, dma_dest);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
sw_desc->unmap_src_cnt = 1;
sw_desc->unmap_len = len;
sw_desc->async_tx.flags = flags;
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
}
spin_unlock_bh(&iop_chan->lock);
return sw_desc ? &sw_desc->async_tx : NULL;
}
static struct dma_async_tx_descriptor *
iop_adma_prep_dma_xor(struct dma_chan *chan, dma_addr_t dma_dest,
dma_addr_t *dma_src, unsigned int src_cnt, size_t len,
unsigned long flags)
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
{
struct iop_adma_chan *iop_chan = to_iop_adma_chan(chan);
struct iop_adma_desc_slot *sw_desc, *grp_start;
int slot_cnt, slots_per_op;
if (unlikely(!len))
return NULL;
BUG_ON(len > IOP_ADMA_XOR_MAX_BYTE_COUNT);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
dev_dbg(iop_chan->device->common.dev,
"%s src_cnt: %d len: %u flags: %lx\n",
__func__, src_cnt, len, flags);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
spin_lock_bh(&iop_chan->lock);
slot_cnt = iop_chan_xor_slot_count(len, src_cnt, &slots_per_op);
sw_desc = iop_adma_alloc_slots(iop_chan, slot_cnt, slots_per_op);
if (sw_desc) {
grp_start = sw_desc->group_head;
iop_desc_init_xor(grp_start, src_cnt, flags);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
iop_desc_set_byte_count(grp_start, iop_chan, len);
iop_desc_set_dest_addr(grp_start, iop_chan, dma_dest);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
sw_desc->unmap_src_cnt = src_cnt;
sw_desc->unmap_len = len;
sw_desc->async_tx.flags = flags;
while (src_cnt--)
iop_desc_set_xor_src_addr(grp_start, src_cnt,
dma_src[src_cnt]);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
}
spin_unlock_bh(&iop_chan->lock);
return sw_desc ? &sw_desc->async_tx : NULL;
}
static struct dma_async_tx_descriptor *
iop_adma_prep_dma_xor_val(struct dma_chan *chan, dma_addr_t *dma_src,
unsigned int src_cnt, size_t len, u32 *result,
unsigned long flags)
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
{
struct iop_adma_chan *iop_chan = to_iop_adma_chan(chan);
struct iop_adma_desc_slot *sw_desc, *grp_start;
int slot_cnt, slots_per_op;
if (unlikely(!len))
return NULL;
dev_dbg(iop_chan->device->common.dev, "%s src_cnt: %d len: %u\n",
__func__, src_cnt, len);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
spin_lock_bh(&iop_chan->lock);
slot_cnt = iop_chan_zero_sum_slot_count(len, src_cnt, &slots_per_op);
sw_desc = iop_adma_alloc_slots(iop_chan, slot_cnt, slots_per_op);
if (sw_desc) {
grp_start = sw_desc->group_head;
iop_desc_init_zero_sum(grp_start, src_cnt, flags);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
iop_desc_set_zero_sum_byte_count(grp_start, len);
grp_start->xor_check_result = result;
pr_debug("\t%s: grp_start->xor_check_result: %p\n",
__func__, grp_start->xor_check_result);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
sw_desc->unmap_src_cnt = src_cnt;
sw_desc->unmap_len = len;
sw_desc->async_tx.flags = flags;
while (src_cnt--)
iop_desc_set_zero_sum_src_addr(grp_start, src_cnt,
dma_src[src_cnt]);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
}
spin_unlock_bh(&iop_chan->lock);
return sw_desc ? &sw_desc->async_tx : NULL;
}
static struct dma_async_tx_descriptor *
iop_adma_prep_dma_pq(struct dma_chan *chan, dma_addr_t *dst, dma_addr_t *src,
unsigned int src_cnt, const unsigned char *scf, size_t len,
unsigned long flags)
{
struct iop_adma_chan *iop_chan = to_iop_adma_chan(chan);
struct iop_adma_desc_slot *sw_desc, *g;
int slot_cnt, slots_per_op;
int continue_srcs;
if (unlikely(!len))
return NULL;
BUG_ON(len > IOP_ADMA_XOR_MAX_BYTE_COUNT);
dev_dbg(iop_chan->device->common.dev,
"%s src_cnt: %d len: %u flags: %lx\n",
__func__, src_cnt, len, flags);
if (dmaf_p_disabled_continue(flags))
continue_srcs = 1+src_cnt;
else if (dmaf_continue(flags))
continue_srcs = 3+src_cnt;
else
continue_srcs = 0+src_cnt;
spin_lock_bh(&iop_chan->lock);
slot_cnt = iop_chan_pq_slot_count(len, continue_srcs, &slots_per_op);
sw_desc = iop_adma_alloc_slots(iop_chan, slot_cnt, slots_per_op);
if (sw_desc) {
int i;
g = sw_desc->group_head;
iop_desc_set_byte_count(g, iop_chan, len);
/* even if P is disabled its destination address (bits
* [3:0]) must match Q. It is ok if P points to an
* invalid address, it won't be written.
*/
if (flags & DMA_PREP_PQ_DISABLE_P)
dst[0] = dst[1] & 0x7;
iop_desc_set_pq_addr(g, dst);
sw_desc->unmap_src_cnt = src_cnt;
sw_desc->unmap_len = len;
sw_desc->async_tx.flags = flags;
for (i = 0; i < src_cnt; i++)
iop_desc_set_pq_src_addr(g, i, src[i], scf[i]);
/* if we are continuing a previous operation factor in
* the old p and q values, see the comment for dma_maxpq
* in include/linux/dmaengine.h
*/
if (dmaf_p_disabled_continue(flags))
iop_desc_set_pq_src_addr(g, i++, dst[1], 1);
else if (dmaf_continue(flags)) {
iop_desc_set_pq_src_addr(g, i++, dst[0], 0);
iop_desc_set_pq_src_addr(g, i++, dst[1], 1);
iop_desc_set_pq_src_addr(g, i++, dst[1], 0);
}
iop_desc_init_pq(g, i, flags);
}
spin_unlock_bh(&iop_chan->lock);
return sw_desc ? &sw_desc->async_tx : NULL;
}
static struct dma_async_tx_descriptor *
iop_adma_prep_dma_pq_val(struct dma_chan *chan, dma_addr_t *pq, dma_addr_t *src,
unsigned int src_cnt, const unsigned char *scf,
size_t len, enum sum_check_flags *pqres,
unsigned long flags)
{
struct iop_adma_chan *iop_chan = to_iop_adma_chan(chan);
struct iop_adma_desc_slot *sw_desc, *g;
int slot_cnt, slots_per_op;
if (unlikely(!len))
return NULL;
BUG_ON(len > IOP_ADMA_XOR_MAX_BYTE_COUNT);
dev_dbg(iop_chan->device->common.dev, "%s src_cnt: %d len: %u\n",
__func__, src_cnt, len);
spin_lock_bh(&iop_chan->lock);
slot_cnt = iop_chan_pq_zero_sum_slot_count(len, src_cnt + 2, &slots_per_op);
sw_desc = iop_adma_alloc_slots(iop_chan, slot_cnt, slots_per_op);
if (sw_desc) {
/* for validate operations p and q are tagged onto the
* end of the source list
*/
int pq_idx = src_cnt;
g = sw_desc->group_head;
iop_desc_init_pq_zero_sum(g, src_cnt+2, flags);
iop_desc_set_pq_zero_sum_byte_count(g, len);
g->pq_check_result = pqres;
pr_debug("\t%s: g->pq_check_result: %p\n",
__func__, g->pq_check_result);
sw_desc->unmap_src_cnt = src_cnt+2;
sw_desc->unmap_len = len;
sw_desc->async_tx.flags = flags;
while (src_cnt--)
iop_desc_set_pq_zero_sum_src_addr(g, src_cnt,
src[src_cnt],
scf[src_cnt]);
iop_desc_set_pq_zero_sum_addr(g, pq_idx, src);
}
spin_unlock_bh(&iop_chan->lock);
return sw_desc ? &sw_desc->async_tx : NULL;
}
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
static void iop_adma_free_chan_resources(struct dma_chan *chan)
{
struct iop_adma_chan *iop_chan = to_iop_adma_chan(chan);
struct iop_adma_desc_slot *iter, *_iter;
int in_use_descs = 0;
iop_adma_slot_cleanup(iop_chan);
spin_lock_bh(&iop_chan->lock);
list_for_each_entry_safe(iter, _iter, &iop_chan->chain,
chain_node) {
in_use_descs++;
list_del(&iter->chain_node);
}
list_for_each_entry_safe_reverse(
iter, _iter, &iop_chan->all_slots, slot_node) {
list_del(&iter->slot_node);
kfree(iter);
iop_chan->slots_allocated--;
}
iop_chan->last_used = NULL;
dev_dbg(iop_chan->device->common.dev, "%s slots_allocated %d\n",
__func__, iop_chan->slots_allocated);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
spin_unlock_bh(&iop_chan->lock);
/* one is ok since we left it on there on purpose */
if (in_use_descs > 1)
printk(KERN_ERR "IOP: Freeing %d in use descriptors!\n",
in_use_descs - 1);
}
/**
* iop_adma_status - poll the status of an ADMA transaction
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
* @chan: ADMA channel handle
* @cookie: ADMA transaction identifier
* @txstate: a holder for the current state of the channel or NULL
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
*/
static enum dma_status iop_adma_status(struct dma_chan *chan,
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
dma_cookie_t cookie,
struct dma_tx_state *txstate)
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
{
struct iop_adma_chan *iop_chan = to_iop_adma_chan(chan);
int ret;
ret = dma_cookie_status(chan, cookie, txstate);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
if (ret == DMA_SUCCESS)
return ret;
iop_adma_slot_cleanup(iop_chan);
return dma_cookie_status(chan, cookie, txstate);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
}
static irqreturn_t iop_adma_eot_handler(int irq, void *data)
{
struct iop_adma_chan *chan = data;
dev_dbg(chan->device->common.dev, "%s\n", __func__);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
tasklet_schedule(&chan->irq_tasklet);
iop_adma_device_clear_eot_status(chan);
return IRQ_HANDLED;
}
static irqreturn_t iop_adma_eoc_handler(int irq, void *data)
{
struct iop_adma_chan *chan = data;
dev_dbg(chan->device->common.dev, "%s\n", __func__);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
tasklet_schedule(&chan->irq_tasklet);
iop_adma_device_clear_eoc_status(chan);
return IRQ_HANDLED;
}
static irqreturn_t iop_adma_err_handler(int irq, void *data)
{
struct iop_adma_chan *chan = data;
unsigned long status = iop_chan_get_status(chan);
dev_printk(KERN_ERR, chan->device->common.dev,
"error ( %s%s%s%s%s%s%s)\n",
iop_is_err_int_parity(status, chan) ? "int_parity " : "",
iop_is_err_mcu_abort(status, chan) ? "mcu_abort " : "",
iop_is_err_int_tabort(status, chan) ? "int_tabort " : "",
iop_is_err_int_mabort(status, chan) ? "int_mabort " : "",
iop_is_err_pci_tabort(status, chan) ? "pci_tabort " : "",
iop_is_err_pci_mabort(status, chan) ? "pci_mabort " : "",
iop_is_err_split_tx(status, chan) ? "split_tx " : "");
iop_adma_device_clear_err_status(chan);
BUG();
return IRQ_HANDLED;
}
static void iop_adma_issue_pending(struct dma_chan *chan)
{
struct iop_adma_chan *iop_chan = to_iop_adma_chan(chan);
if (iop_chan->pending) {
iop_chan->pending = 0;
iop_chan_append(iop_chan);
}
}
/*
* Perform a transaction to verify the HW works.
*/
#define IOP_ADMA_TEST_SIZE 2000
static int __devinit iop_adma_memcpy_self_test(struct iop_adma_device *device)
{
int i;
void *src, *dest;
dma_addr_t src_dma, dest_dma;
struct dma_chan *dma_chan;
dma_cookie_t cookie;
struct dma_async_tx_descriptor *tx;
int err = 0;
struct iop_adma_chan *iop_chan;
dev_dbg(device->common.dev, "%s\n", __func__);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
src = kmalloc(IOP_ADMA_TEST_SIZE, GFP_KERNEL);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
if (!src)
return -ENOMEM;
dest = kzalloc(IOP_ADMA_TEST_SIZE, GFP_KERNEL);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
if (!dest) {
kfree(src);
return -ENOMEM;
}
/* Fill in src buffer */
for (i = 0; i < IOP_ADMA_TEST_SIZE; i++)
((u8 *) src)[i] = (u8)i;
/* Start copy, using first DMA channel */
dma_chan = container_of(device->common.channels.next,
struct dma_chan,
device_node);
if (iop_adma_alloc_chan_resources(dma_chan) < 1) {
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
err = -ENODEV;
goto out;
}
dest_dma = dma_map_single(dma_chan->device->dev, dest,
IOP_ADMA_TEST_SIZE, DMA_FROM_DEVICE);
src_dma = dma_map_single(dma_chan->device->dev, src,
IOP_ADMA_TEST_SIZE, DMA_TO_DEVICE);
tx = iop_adma_prep_dma_memcpy(dma_chan, dest_dma, src_dma,
IOP_ADMA_TEST_SIZE,
DMA_PREP_INTERRUPT | DMA_CTRL_ACK);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
cookie = iop_adma_tx_submit(tx);
iop_adma_issue_pending(dma_chan);
msleep(1);
if (iop_adma_status(dma_chan, cookie, NULL) !=
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
DMA_SUCCESS) {
dev_printk(KERN_ERR, dma_chan->device->dev,
"Self-test copy timed out, disabling\n");
err = -ENODEV;
goto free_resources;
}
iop_chan = to_iop_adma_chan(dma_chan);
dma_sync_single_for_cpu(&iop_chan->device->pdev->dev, dest_dma,
IOP_ADMA_TEST_SIZE, DMA_FROM_DEVICE);
if (memcmp(src, dest, IOP_ADMA_TEST_SIZE)) {
dev_printk(KERN_ERR, dma_chan->device->dev,
"Self-test copy failed compare, disabling\n");
err = -ENODEV;
goto free_resources;
}
free_resources:
iop_adma_free_chan_resources(dma_chan);
out:
kfree(src);
kfree(dest);
return err;
}
#define IOP_ADMA_NUM_SRC_TEST 4 /* must be <= 15 */
static int __devinit
iop_adma_xor_val_self_test(struct iop_adma_device *device)
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
{
int i, src_idx;
struct page *dest;
struct page *xor_srcs[IOP_ADMA_NUM_SRC_TEST];
struct page *zero_sum_srcs[IOP_ADMA_NUM_SRC_TEST + 1];
dma_addr_t dma_srcs[IOP_ADMA_NUM_SRC_TEST + 1];
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
dma_addr_t dma_addr, dest_dma;
struct dma_async_tx_descriptor *tx;
struct dma_chan *dma_chan;
dma_cookie_t cookie;
u8 cmp_byte = 0;
u32 cmp_word;
u32 zero_sum_result;
int err = 0;
struct iop_adma_chan *iop_chan;
dev_dbg(device->common.dev, "%s\n", __func__);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
for (src_idx = 0; src_idx < IOP_ADMA_NUM_SRC_TEST; src_idx++) {
xor_srcs[src_idx] = alloc_page(GFP_KERNEL);
if (!xor_srcs[src_idx]) {
while (src_idx--)
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
__free_page(xor_srcs[src_idx]);
return -ENOMEM;
}
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
}
dest = alloc_page(GFP_KERNEL);
if (!dest) {
while (src_idx--)
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
__free_page(xor_srcs[src_idx]);
return -ENOMEM;
}
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
/* Fill in src buffers */
for (src_idx = 0; src_idx < IOP_ADMA_NUM_SRC_TEST; src_idx++) {
u8 *ptr = page_address(xor_srcs[src_idx]);
for (i = 0; i < PAGE_SIZE; i++)
ptr[i] = (1 << src_idx);
}
for (src_idx = 0; src_idx < IOP_ADMA_NUM_SRC_TEST; src_idx++)
cmp_byte ^= (u8) (1 << src_idx);
cmp_word = (cmp_byte << 24) | (cmp_byte << 16) |
(cmp_byte << 8) | cmp_byte;
memset(page_address(dest), 0, PAGE_SIZE);
dma_chan = container_of(device->common.channels.next,
struct dma_chan,
device_node);
if (iop_adma_alloc_chan_resources(dma_chan) < 1) {
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
err = -ENODEV;
goto out;
}
/* test xor */
dest_dma = dma_map_page(dma_chan->device->dev, dest, 0,
PAGE_SIZE, DMA_FROM_DEVICE);
for (i = 0; i < IOP_ADMA_NUM_SRC_TEST; i++)
dma_srcs[i] = dma_map_page(dma_chan->device->dev, xor_srcs[i],
0, PAGE_SIZE, DMA_TO_DEVICE);
tx = iop_adma_prep_dma_xor(dma_chan, dest_dma, dma_srcs,
IOP_ADMA_NUM_SRC_TEST, PAGE_SIZE,
DMA_PREP_INTERRUPT | DMA_CTRL_ACK);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
cookie = iop_adma_tx_submit(tx);
iop_adma_issue_pending(dma_chan);
msleep(8);
if (iop_adma_status(dma_chan, cookie, NULL) !=
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
DMA_SUCCESS) {
dev_printk(KERN_ERR, dma_chan->device->dev,
"Self-test xor timed out, disabling\n");
err = -ENODEV;
goto free_resources;
}
iop_chan = to_iop_adma_chan(dma_chan);
dma_sync_single_for_cpu(&iop_chan->device->pdev->dev, dest_dma,
PAGE_SIZE, DMA_FROM_DEVICE);
for (i = 0; i < (PAGE_SIZE / sizeof(u32)); i++) {
u32 *ptr = page_address(dest);
if (ptr[i] != cmp_word) {
dev_printk(KERN_ERR, dma_chan->device->dev,
"Self-test xor failed compare, disabling\n");
err = -ENODEV;
goto free_resources;
}
}
dma_sync_single_for_device(&iop_chan->device->pdev->dev, dest_dma,
PAGE_SIZE, DMA_TO_DEVICE);
/* skip zero sum if the capability is not present */
if (!dma_has_cap(DMA_XOR_VAL, dma_chan->device->cap_mask))
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
goto free_resources;
/* zero sum the sources with the destintation page */
for (i = 0; i < IOP_ADMA_NUM_SRC_TEST; i++)
zero_sum_srcs[i] = xor_srcs[i];
zero_sum_srcs[i] = dest;
zero_sum_result = 1;
for (i = 0; i < IOP_ADMA_NUM_SRC_TEST + 1; i++)
dma_srcs[i] = dma_map_page(dma_chan->device->dev,
zero_sum_srcs[i], 0, PAGE_SIZE,
DMA_TO_DEVICE);
tx = iop_adma_prep_dma_xor_val(dma_chan, dma_srcs,
IOP_ADMA_NUM_SRC_TEST + 1, PAGE_SIZE,
&zero_sum_result,
DMA_PREP_INTERRUPT | DMA_CTRL_ACK);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
cookie = iop_adma_tx_submit(tx);
iop_adma_issue_pending(dma_chan);
msleep(8);
if (iop_adma_status(dma_chan, cookie, NULL) != DMA_SUCCESS) {
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
dev_printk(KERN_ERR, dma_chan->device->dev,
"Self-test zero sum timed out, disabling\n");
err = -ENODEV;
goto free_resources;
}
if (zero_sum_result != 0) {
dev_printk(KERN_ERR, dma_chan->device->dev,
"Self-test zero sum failed compare, disabling\n");
err = -ENODEV;
goto free_resources;
}
/* test memset */
dma_addr = dma_map_page(dma_chan->device->dev, dest, 0,
PAGE_SIZE, DMA_FROM_DEVICE);
tx = iop_adma_prep_dma_memset(dma_chan, dma_addr, 0, PAGE_SIZE,
DMA_PREP_INTERRUPT | DMA_CTRL_ACK);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
cookie = iop_adma_tx_submit(tx);
iop_adma_issue_pending(dma_chan);
msleep(8);
if (iop_adma_status(dma_chan, cookie, NULL) != DMA_SUCCESS) {
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
dev_printk(KERN_ERR, dma_chan->device->dev,
"Self-test memset timed out, disabling\n");
err = -ENODEV;
goto free_resources;
}
for (i = 0; i < PAGE_SIZE/sizeof(u32); i++) {
u32 *ptr = page_address(dest);
if (ptr[i]) {
dev_printk(KERN_ERR, dma_chan->device->dev,
"Self-test memset failed compare, disabling\n");
err = -ENODEV;
goto free_resources;
}
}
/* test for non-zero parity sum */
zero_sum_result = 0;
for (i = 0; i < IOP_ADMA_NUM_SRC_TEST + 1; i++)
dma_srcs[i] = dma_map_page(dma_chan->device->dev,
zero_sum_srcs[i], 0, PAGE_SIZE,
DMA_TO_DEVICE);
tx = iop_adma_prep_dma_xor_val(dma_chan, dma_srcs,
IOP_ADMA_NUM_SRC_TEST + 1, PAGE_SIZE,
&zero_sum_result,
DMA_PREP_INTERRUPT | DMA_CTRL_ACK);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
cookie = iop_adma_tx_submit(tx);
iop_adma_issue_pending(dma_chan);
msleep(8);
if (iop_adma_status(dma_chan, cookie, NULL) != DMA_SUCCESS) {
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
dev_printk(KERN_ERR, dma_chan->device->dev,
"Self-test non-zero sum timed out, disabling\n");
err = -ENODEV;
goto free_resources;
}
if (zero_sum_result != 1) {
dev_printk(KERN_ERR, dma_chan->device->dev,
"Self-test non-zero sum failed compare, disabling\n");
err = -ENODEV;
goto free_resources;
}
free_resources:
iop_adma_free_chan_resources(dma_chan);
out:
src_idx = IOP_ADMA_NUM_SRC_TEST;
while (src_idx--)
__free_page(xor_srcs[src_idx]);
__free_page(dest);
return err;
}
#ifdef CONFIG_RAID6_PQ
static int __devinit
iop_adma_pq_zero_sum_self_test(struct iop_adma_device *device)
{
/* combined sources, software pq results, and extra hw pq results */
struct page *pq[IOP_ADMA_NUM_SRC_TEST+2+2];
/* ptr to the extra hw pq buffers defined above */
struct page **pq_hw = &pq[IOP_ADMA_NUM_SRC_TEST+2];
/* address conversion buffers (dma_map / page_address) */
void *pq_sw[IOP_ADMA_NUM_SRC_TEST+2];
dma_addr_t pq_src[IOP_ADMA_NUM_SRC_TEST+2];
dma_addr_t *pq_dest = &pq_src[IOP_ADMA_NUM_SRC_TEST];
int i;
struct dma_async_tx_descriptor *tx;
struct dma_chan *dma_chan;
dma_cookie_t cookie;
u32 zero_sum_result;
int err = 0;
struct device *dev;
dev_dbg(device->common.dev, "%s\n", __func__);
for (i = 0; i < ARRAY_SIZE(pq); i++) {
pq[i] = alloc_page(GFP_KERNEL);
if (!pq[i]) {
while (i--)
__free_page(pq[i]);
return -ENOMEM;
}
}
/* Fill in src buffers */
for (i = 0; i < IOP_ADMA_NUM_SRC_TEST; i++) {
pq_sw[i] = page_address(pq[i]);
memset(pq_sw[i], 0x11111111 * (1<<i), PAGE_SIZE);
}
pq_sw[i] = page_address(pq[i]);
pq_sw[i+1] = page_address(pq[i+1]);
dma_chan = container_of(device->common.channels.next,
struct dma_chan,
device_node);
if (iop_adma_alloc_chan_resources(dma_chan) < 1) {
err = -ENODEV;
goto out;
}
dev = dma_chan->device->dev;
/* initialize the dests */
memset(page_address(pq_hw[0]), 0 , PAGE_SIZE);
memset(page_address(pq_hw[1]), 0 , PAGE_SIZE);
/* test pq */
pq_dest[0] = dma_map_page(dev, pq_hw[0], 0, PAGE_SIZE, DMA_FROM_DEVICE);
pq_dest[1] = dma_map_page(dev, pq_hw[1], 0, PAGE_SIZE, DMA_FROM_DEVICE);
for (i = 0; i < IOP_ADMA_NUM_SRC_TEST; i++)
pq_src[i] = dma_map_page(dev, pq[i], 0, PAGE_SIZE,
DMA_TO_DEVICE);
tx = iop_adma_prep_dma_pq(dma_chan, pq_dest, pq_src,
IOP_ADMA_NUM_SRC_TEST, (u8 *)raid6_gfexp,
PAGE_SIZE,
DMA_PREP_INTERRUPT |
DMA_CTRL_ACK);
cookie = iop_adma_tx_submit(tx);
iop_adma_issue_pending(dma_chan);
msleep(8);
if (iop_adma_status(dma_chan, cookie, NULL) !=
DMA_SUCCESS) {
dev_err(dev, "Self-test pq timed out, disabling\n");
err = -ENODEV;
goto free_resources;
}
raid6_call.gen_syndrome(IOP_ADMA_NUM_SRC_TEST+2, PAGE_SIZE, pq_sw);
if (memcmp(pq_sw[IOP_ADMA_NUM_SRC_TEST],
page_address(pq_hw[0]), PAGE_SIZE) != 0) {
dev_err(dev, "Self-test p failed compare, disabling\n");
err = -ENODEV;
goto free_resources;
}
if (memcmp(pq_sw[IOP_ADMA_NUM_SRC_TEST+1],
page_address(pq_hw[1]), PAGE_SIZE) != 0) {
dev_err(dev, "Self-test q failed compare, disabling\n");
err = -ENODEV;
goto free_resources;
}
/* test correct zero sum using the software generated pq values */
for (i = 0; i < IOP_ADMA_NUM_SRC_TEST + 2; i++)
pq_src[i] = dma_map_page(dev, pq[i], 0, PAGE_SIZE,
DMA_TO_DEVICE);
zero_sum_result = ~0;
tx = iop_adma_prep_dma_pq_val(dma_chan, &pq_src[IOP_ADMA_NUM_SRC_TEST],
pq_src, IOP_ADMA_NUM_SRC_TEST,
raid6_gfexp, PAGE_SIZE, &zero_sum_result,
DMA_PREP_INTERRUPT|DMA_CTRL_ACK);
cookie = iop_adma_tx_submit(tx);
iop_adma_issue_pending(dma_chan);
msleep(8);
if (iop_adma_status(dma_chan, cookie, NULL) !=
DMA_SUCCESS) {
dev_err(dev, "Self-test pq-zero-sum timed out, disabling\n");
err = -ENODEV;
goto free_resources;
}
if (zero_sum_result != 0) {
dev_err(dev, "Self-test pq-zero-sum failed to validate: %x\n",
zero_sum_result);
err = -ENODEV;
goto free_resources;
}
/* test incorrect zero sum */
i = IOP_ADMA_NUM_SRC_TEST;
memset(pq_sw[i] + 100, 0, 100);
memset(pq_sw[i+1] + 200, 0, 200);
for (i = 0; i < IOP_ADMA_NUM_SRC_TEST + 2; i++)
pq_src[i] = dma_map_page(dev, pq[i], 0, PAGE_SIZE,
DMA_TO_DEVICE);
zero_sum_result = 0;
tx = iop_adma_prep_dma_pq_val(dma_chan, &pq_src[IOP_ADMA_NUM_SRC_TEST],
pq_src, IOP_ADMA_NUM_SRC_TEST,
raid6_gfexp, PAGE_SIZE, &zero_sum_result,
DMA_PREP_INTERRUPT|DMA_CTRL_ACK);
cookie = iop_adma_tx_submit(tx);
iop_adma_issue_pending(dma_chan);
msleep(8);
if (iop_adma_status(dma_chan, cookie, NULL) !=
DMA_SUCCESS) {
dev_err(dev, "Self-test !pq-zero-sum timed out, disabling\n");
err = -ENODEV;
goto free_resources;
}
if (zero_sum_result != (SUM_CHECK_P_RESULT | SUM_CHECK_Q_RESULT)) {
dev_err(dev, "Self-test !pq-zero-sum failed to validate: %x\n",
zero_sum_result);
err = -ENODEV;
goto free_resources;
}
free_resources:
iop_adma_free_chan_resources(dma_chan);
out:
i = ARRAY_SIZE(pq);
while (i--)
__free_page(pq[i]);
return err;
}
#endif
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
static int __devexit iop_adma_remove(struct platform_device *dev)
{
struct iop_adma_device *device = platform_get_drvdata(dev);
struct dma_chan *chan, *_chan;
struct iop_adma_chan *iop_chan;
struct iop_adma_platform_data *plat_data = dev->dev.platform_data;
dma_async_device_unregister(&device->common);
dma_free_coherent(&dev->dev, plat_data->pool_size,
device->dma_desc_pool_virt, device->dma_desc_pool);
list_for_each_entry_safe(chan, _chan, &device->common.channels,
device_node) {
iop_chan = to_iop_adma_chan(chan);
list_del(&chan->device_node);
kfree(iop_chan);
}
kfree(device);
return 0;
}
static int __devinit iop_adma_probe(struct platform_device *pdev)
{
struct resource *res;
int ret = 0, i;
struct iop_adma_device *adev;
struct iop_adma_chan *iop_chan;
struct dma_device *dma_dev;
struct iop_adma_platform_data *plat_data = pdev->dev.platform_data;
res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
if (!res)
return -ENODEV;
if (!devm_request_mem_region(&pdev->dev, res->start,
resource_size(res), pdev->name))
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
return -EBUSY;
adev = kzalloc(sizeof(*adev), GFP_KERNEL);
if (!adev)
return -ENOMEM;
dma_dev = &adev->common;
/* allocate coherent memory for hardware descriptors
* note: writecombine gives slightly better performance, but
* requires that we explicitly flush the writes
*/
if ((adev->dma_desc_pool_virt = dma_alloc_writecombine(&pdev->dev,
plat_data->pool_size,
&adev->dma_desc_pool,
GFP_KERNEL)) == NULL) {
ret = -ENOMEM;
goto err_free_adev;
}
dev_dbg(&pdev->dev, "%s: allocated descriptor pool virt %p phys %p\n",
__func__, adev->dma_desc_pool_virt,
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
(void *) adev->dma_desc_pool);
adev->id = plat_data->hw_id;
/* discover transaction capabilites from the platform data */
dma_dev->cap_mask = plat_data->cap_mask;
adev->pdev = pdev;
platform_set_drvdata(pdev, adev);
INIT_LIST_HEAD(&dma_dev->channels);
/* set base routines */
dma_dev->device_alloc_chan_resources = iop_adma_alloc_chan_resources;
dma_dev->device_free_chan_resources = iop_adma_free_chan_resources;
dma_dev->device_tx_status = iop_adma_status;
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
dma_dev->device_issue_pending = iop_adma_issue_pending;
dma_dev->dev = &pdev->dev;
/* set prep routines based on capability */
if (dma_has_cap(DMA_MEMCPY, dma_dev->cap_mask))
dma_dev->device_prep_dma_memcpy = iop_adma_prep_dma_memcpy;
if (dma_has_cap(DMA_MEMSET, dma_dev->cap_mask))
dma_dev->device_prep_dma_memset = iop_adma_prep_dma_memset;
if (dma_has_cap(DMA_XOR, dma_dev->cap_mask)) {
dma_dev->max_xor = iop_adma_get_max_xor();
dma_dev->device_prep_dma_xor = iop_adma_prep_dma_xor;
}
if (dma_has_cap(DMA_XOR_VAL, dma_dev->cap_mask))
dma_dev->device_prep_dma_xor_val =
iop_adma_prep_dma_xor_val;
if (dma_has_cap(DMA_PQ, dma_dev->cap_mask)) {
dma_set_maxpq(dma_dev, iop_adma_get_max_pq(), 0);
dma_dev->device_prep_dma_pq = iop_adma_prep_dma_pq;
}
if (dma_has_cap(DMA_PQ_VAL, dma_dev->cap_mask))
dma_dev->device_prep_dma_pq_val =
iop_adma_prep_dma_pq_val;
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
if (dma_has_cap(DMA_INTERRUPT, dma_dev->cap_mask))
dma_dev->device_prep_dma_interrupt =
iop_adma_prep_dma_interrupt;
iop_chan = kzalloc(sizeof(*iop_chan), GFP_KERNEL);
if (!iop_chan) {
ret = -ENOMEM;
goto err_free_dma;
}
iop_chan->device = adev;
iop_chan->mmr_base = devm_ioremap(&pdev->dev, res->start,
resource_size(res));
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
if (!iop_chan->mmr_base) {
ret = -ENOMEM;
goto err_free_iop_chan;
}
tasklet_init(&iop_chan->irq_tasklet, iop_adma_tasklet, (unsigned long)
iop_chan);
/* clear errors before enabling interrupts */
iop_adma_device_clear_err_status(iop_chan);
for (i = 0; i < 3; i++) {
irq_handler_t handler[] = { iop_adma_eot_handler,
iop_adma_eoc_handler,
iop_adma_err_handler };
int irq = platform_get_irq(pdev, i);
if (irq < 0) {
ret = -ENXIO;
goto err_free_iop_chan;
} else {
ret = devm_request_irq(&pdev->dev, irq,
handler[i], 0, pdev->name, iop_chan);
if (ret)
goto err_free_iop_chan;
}
}
spin_lock_init(&iop_chan->lock);
INIT_LIST_HEAD(&iop_chan->chain);
INIT_LIST_HEAD(&iop_chan->all_slots);
iop_chan->common.device = dma_dev;
dma_cookie_init(&iop_chan->common);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
list_add_tail(&iop_chan->common.device_node, &dma_dev->channels);
if (dma_has_cap(DMA_MEMCPY, dma_dev->cap_mask)) {
ret = iop_adma_memcpy_self_test(adev);
dev_dbg(&pdev->dev, "memcpy self test returned %d\n", ret);
if (ret)
goto err_free_iop_chan;
}
if (dma_has_cap(DMA_XOR, dma_dev->cap_mask) ||
dma_has_cap(DMA_MEMSET, dma_dev->cap_mask)) {
ret = iop_adma_xor_val_self_test(adev);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
dev_dbg(&pdev->dev, "xor self test returned %d\n", ret);
if (ret)
goto err_free_iop_chan;
}
if (dma_has_cap(DMA_PQ, dma_dev->cap_mask) &&
dma_has_cap(DMA_PQ_VAL, dma_dev->cap_mask)) {
#ifdef CONFIG_RAID6_PQ
ret = iop_adma_pq_zero_sum_self_test(adev);
dev_dbg(&pdev->dev, "pq self test returned %d\n", ret);
#else
/* can not test raid6, so do not publish capability */
dma_cap_clear(DMA_PQ, dma_dev->cap_mask);
dma_cap_clear(DMA_PQ_VAL, dma_dev->cap_mask);
ret = 0;
#endif
if (ret)
goto err_free_iop_chan;
}
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
dev_printk(KERN_INFO, &pdev->dev, "Intel(R) IOP: "
"( %s%s%s%s%s%s%s)\n",
async_tx: add support for asynchronous GF multiplication [ Based on an original patch by Yuri Tikhonov ] This adds support for doing asynchronous GF multiplication by adding two additional functions to the async_tx API: async_gen_syndrome() does simultaneous XOR and Galois field multiplication of sources. async_syndrome_val() validates the given source buffers against known P and Q values. When a request is made to run async_pq against more than the hardware maximum number of supported sources we need to reuse the previous generated P and Q values as sources into the next operation. Care must be taken to remove Q from P' and P from Q'. For example to perform a 5 source pq op with hardware that only supports 4 sources at a time the following approach is taken: p, q = PQ(src0, src1, src2, src3, COEF({01}, {02}, {04}, {08})) p', q' = PQ(p, q, q, src4, COEF({00}, {01}, {00}, {10})) p' = p + q + q + src4 = p + src4 q' = {00}*p + {01}*q + {00}*q + {10}*src4 = q + {10}*src4 Note: 4 is the minimum acceptable maxpq otherwise we punt to synchronous-software path. The DMA_PREP_CONTINUE flag indicates to the driver to reuse p and q as sources (in the above manner) and fill the remaining slots up to maxpq with the new sources/coefficients. Note1: Some devices have native support for P+Q continuation and can skip this extra work. Devices with this capability can advertise it with dma_set_maxpq. It is up to each driver how to handle the DMA_PREP_CONTINUE flag. Note2: The api supports disabling the generation of P when generating Q, this is ignored by the synchronous path but is implemented by some dma devices to save unnecessary writes. In this case the continuation algorithm is simplified to only reuse Q as a source. Cc: H. Peter Anvin <hpa@zytor.com> Cc: David Woodhouse <David.Woodhouse@intel.com> Signed-off-by: Yuri Tikhonov <yur@emcraft.com> Signed-off-by: Ilya Yanok <yanok@emcraft.com> Reviewed-by: Andre Noll <maan@systemlinux.org> Acked-by: Maciej Sosnowski <maciej.sosnowski@intel.com> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2009-07-15 02:20:36 +07:00
dma_has_cap(DMA_PQ, dma_dev->cap_mask) ? "pq " : "",
dma_has_cap(DMA_PQ_VAL, dma_dev->cap_mask) ? "pq_val " : "",
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
dma_has_cap(DMA_XOR, dma_dev->cap_mask) ? "xor " : "",
dma_has_cap(DMA_XOR_VAL, dma_dev->cap_mask) ? "xor_val " : "",
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
dma_has_cap(DMA_MEMSET, dma_dev->cap_mask) ? "fill " : "",
dma_has_cap(DMA_MEMCPY, dma_dev->cap_mask) ? "cpy " : "",
dma_has_cap(DMA_INTERRUPT, dma_dev->cap_mask) ? "intr " : "");
dma_async_device_register(dma_dev);
goto out;
err_free_iop_chan:
kfree(iop_chan);
err_free_dma:
dma_free_coherent(&adev->pdev->dev, plat_data->pool_size,
adev->dma_desc_pool_virt, adev->dma_desc_pool);
err_free_adev:
kfree(adev);
out:
return ret;
}
static void iop_chan_start_null_memcpy(struct iop_adma_chan *iop_chan)
{
struct iop_adma_desc_slot *sw_desc, *grp_start;
dma_cookie_t cookie;
int slot_cnt, slots_per_op;
dev_dbg(iop_chan->device->common.dev, "%s\n", __func__);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
spin_lock_bh(&iop_chan->lock);
slot_cnt = iop_chan_memcpy_slot_count(0, &slots_per_op);
sw_desc = iop_adma_alloc_slots(iop_chan, slot_cnt, slots_per_op);
if (sw_desc) {
grp_start = sw_desc->group_head;
list_splice_init(&sw_desc->tx_list, &iop_chan->chain);
async_tx_ack(&sw_desc->async_tx);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
iop_desc_init_memcpy(grp_start, 0);
iop_desc_set_byte_count(grp_start, iop_chan, 0);
iop_desc_set_dest_addr(grp_start, iop_chan, 0);
iop_desc_set_memcpy_src_addr(grp_start, 0);
dmaengine: fix cookie handling in iop-adma.c and ppc4xx/adma.c Dan Williams said: > > Russell King wrote: > > Firstly, we have DMA_MIN_COOKIE which has value 1 - so any cookies below > > that aren't valid.  That seems sane. > > > > We seem to have different behaviours: > > > > -       cookie = c->cookie; > > -       cookie++; > > -       if (cookie < 0) > > -               cookie = 1; > > -       c->cookie = cookie; > > -       tx->cookie = cookie; > > > > c->cookie here is initialized to zero, so the first cookie given out will > > be 1.  This is how most DMA engine drivers implement this. > > > > Then we have this: > > > >                cookie = chan->common.cookie; > >                cookie++; > >                if (cookie <= 1) > >                        cookie = 2; > > > >                /* initialize the completed cookie to be less than > >                 * the most recently used cookie > >                 */ > >                chan->common.completed_cookie = cookie - 1; > >                chan->common.cookie = sw_desc->async_tx.cookie = cookie; > > > > Again, chan->common.cookie starts off at 0.  The first cookie given out > > will be 2, and 1 will never be used.  There are three drivers which > > implement it this way. > > > > Why is there this difference, and can these three be corrected to behave > > the same way as the first (and therefore the assignment of cookies > > consolidated?) > > Yes, they should be consolidated, and I believe they have drifted only > because there were no good common helpers and murphy's law took over. So lets fix this up to use the common dma_cookie_assign() helper. Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk> Tested-by: Linus Walleij <linus.walleij@linaro.org> Reviewed-by: Linus Walleij <linus.walleij@linaro.org> Acked-by: Jassi Brar <jassisinghbrar@gmail.com> [imx-sdma.c & mxs-dma.c] Tested-by: Shawn Guo <shawn.guo@linaro.org> Signed-off-by: Vinod Koul <vinod.koul@linux.intel.com>
2012-03-07 05:36:07 +07:00
cookie = dma_cookie_assign(&sw_desc->async_tx);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
/* initialize the completed cookie to be less than
* the most recently used cookie
*/
iop_chan->common.completed_cookie = cookie - 1;
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
/* channel should not be busy */
BUG_ON(iop_chan_is_busy(iop_chan));
/* clear any prior error-status bits */
iop_adma_device_clear_err_status(iop_chan);
/* disable operation */
iop_chan_disable(iop_chan);
/* set the descriptor address */
iop_chan_set_next_descriptor(iop_chan, sw_desc->async_tx.phys);
/* 1/ don't add pre-chained descriptors
* 2/ dummy read to flush next_desc write
*/
BUG_ON(iop_desc_get_next_desc(sw_desc));
/* run the descriptor */
iop_chan_enable(iop_chan);
} else
dev_printk(KERN_ERR, iop_chan->device->common.dev,
"failed to allocate null descriptor\n");
spin_unlock_bh(&iop_chan->lock);
}
static void iop_chan_start_null_xor(struct iop_adma_chan *iop_chan)
{
struct iop_adma_desc_slot *sw_desc, *grp_start;
dma_cookie_t cookie;
int slot_cnt, slots_per_op;
dev_dbg(iop_chan->device->common.dev, "%s\n", __func__);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
spin_lock_bh(&iop_chan->lock);
slot_cnt = iop_chan_xor_slot_count(0, 2, &slots_per_op);
sw_desc = iop_adma_alloc_slots(iop_chan, slot_cnt, slots_per_op);
if (sw_desc) {
grp_start = sw_desc->group_head;
list_splice_init(&sw_desc->tx_list, &iop_chan->chain);
async_tx_ack(&sw_desc->async_tx);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
iop_desc_init_null_xor(grp_start, 2, 0);
iop_desc_set_byte_count(grp_start, iop_chan, 0);
iop_desc_set_dest_addr(grp_start, iop_chan, 0);
iop_desc_set_xor_src_addr(grp_start, 0, 0);
iop_desc_set_xor_src_addr(grp_start, 1, 0);
dmaengine: fix cookie handling in iop-adma.c and ppc4xx/adma.c Dan Williams said: > > Russell King wrote: > > Firstly, we have DMA_MIN_COOKIE which has value 1 - so any cookies below > > that aren't valid.  That seems sane. > > > > We seem to have different behaviours: > > > > -       cookie = c->cookie; > > -       cookie++; > > -       if (cookie < 0) > > -               cookie = 1; > > -       c->cookie = cookie; > > -       tx->cookie = cookie; > > > > c->cookie here is initialized to zero, so the first cookie given out will > > be 1.  This is how most DMA engine drivers implement this. > > > > Then we have this: > > > >                cookie = chan->common.cookie; > >                cookie++; > >                if (cookie <= 1) > >                        cookie = 2; > > > >                /* initialize the completed cookie to be less than > >                 * the most recently used cookie > >                 */ > >                chan->common.completed_cookie = cookie - 1; > >                chan->common.cookie = sw_desc->async_tx.cookie = cookie; > > > > Again, chan->common.cookie starts off at 0.  The first cookie given out > > will be 2, and 1 will never be used.  There are three drivers which > > implement it this way. > > > > Why is there this difference, and can these three be corrected to behave > > the same way as the first (and therefore the assignment of cookies > > consolidated?) > > Yes, they should be consolidated, and I believe they have drifted only > because there were no good common helpers and murphy's law took over. So lets fix this up to use the common dma_cookie_assign() helper. Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk> Tested-by: Linus Walleij <linus.walleij@linaro.org> Reviewed-by: Linus Walleij <linus.walleij@linaro.org> Acked-by: Jassi Brar <jassisinghbrar@gmail.com> [imx-sdma.c & mxs-dma.c] Tested-by: Shawn Guo <shawn.guo@linaro.org> Signed-off-by: Vinod Koul <vinod.koul@linux.intel.com>
2012-03-07 05:36:07 +07:00
cookie = dma_cookie_assign(&sw_desc->async_tx);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
/* initialize the completed cookie to be less than
* the most recently used cookie
*/
iop_chan->common.completed_cookie = cookie - 1;
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
/* channel should not be busy */
BUG_ON(iop_chan_is_busy(iop_chan));
/* clear any prior error-status bits */
iop_adma_device_clear_err_status(iop_chan);
/* disable operation */
iop_chan_disable(iop_chan);
/* set the descriptor address */
iop_chan_set_next_descriptor(iop_chan, sw_desc->async_tx.phys);
/* 1/ don't add pre-chained descriptors
* 2/ dummy read to flush next_desc write
*/
BUG_ON(iop_desc_get_next_desc(sw_desc));
/* run the descriptor */
iop_chan_enable(iop_chan);
} else
dev_printk(KERN_ERR, iop_chan->device->common.dev,
"failed to allocate null descriptor\n");
spin_unlock_bh(&iop_chan->lock);
}
static struct platform_driver iop_adma_driver = {
.probe = iop_adma_probe,
.remove = iop_adma_remove,
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
.driver = {
.owner = THIS_MODULE,
.name = "iop-adma",
},
};
module_platform_driver(iop_adma_driver);
dmaengine: driver for the iop32x, iop33x, and iop13xx raid engines The Intel(R) IOP series of i/o processors integrate an Xscale core with raid acceleration engines. The capabilities per platform are: iop219: (2) copy engines iop321: (2) copy engines (1) xor and block fill engine iop33x: (2) copy and crc32c engines (1) xor, xor zero sum, pq, pq zero sum, and block fill engine iop34x (iop13xx): (2) copy, crc32c, xor, xor zero sum, and block fill engines (1) copy, crc32c, xor, xor zero sum, pq, pq zero sum, and block fill engine The driver supports the features of the async_tx api: * asynchronous notification of operation completion * implicit (interupt triggered) handling of inter-channel transaction dependencies The driver adapts to the platform it is running by two methods. 1/ #include <asm/arch/adma.h> which defines the hardware specific iop_chan_* and iop_desc_* routines as a series of static inline functions 2/ The private platform data attached to the platform_device defines the capabilities of the channels 20070626: Callbacks are run in a tasklet. Given the recent discussion on LKML about killing tasklets in favor of workqueues I did a quick conversion of the driver. Raid5 resync performance dropped from 50MB/s to 30MB/s, so the tasklet implementation remains until a generic softirq interface is available. Changelog: * fixed a slot allocation bug in do_iop13xx_adma_xor that caused too few slots to be requested eventually leading to data corruption * enabled the slot allocation routine to attempt to free slots before returning -ENOMEM * switched the cleanup routine to solely use the software chain and the status register to determine if a descriptor is complete. This is necessary to support other IOP engines that do not have status writeback capability * make the driver iop generic * modified the allocation routines to understand allocating a group of slots for a single operation * added a null xor initialization operation for the xor only channel on iop3xx * support xor operations on buffers larger than the hardware maximum * split the do_* routines into separate prep, src/dest set, submit stages * added async_tx support (dependent operations initiation at cleanup time) * simplified group handling * added interrupt support (callbacks via tasklets) * brought the pending depth inline with ioat (i.e. 4 descriptors) * drop dma mapping methods, suggested by Chris Leech * don't use inline in C files, Adrian Bunk * remove static tasklet declarations * make iop_adma_alloc_slots easier to read and remove chances for a corrupted descriptor chain * fix locking bug in iop_adma_alloc_chan_resources, Benjamin Herrenschmidt * convert capabilities over to dma_cap_mask_t * fixup sparse warnings * add descriptor flush before iop_chan_enable * checkpatch.pl fixes * gpl v2 only correction * move set_src, set_dest, submit to async_tx methods * move group_list and phys to async_tx Cc: Russell King <rmk@arm.linux.org.uk> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2007-01-03 03:52:26 +07:00
MODULE_AUTHOR("Intel Corporation");
MODULE_DESCRIPTION("IOP ADMA Engine Driver");
MODULE_LICENSE("GPL");
MODULE_ALIAS("platform:iop-adma");