2007-05-08 07:33:32 +07:00
|
|
|
/*
|
|
|
|
* Driver for OHCI 1394 controllers
|
2006-12-20 07:58:35 +07:00
|
|
|
*
|
|
|
|
* Copyright (C) 2003-2006 Kristian Hoegsberg <krh@bitplanet.net>
|
|
|
|
*
|
|
|
|
* This program is free software; you can redistribute it and/or modify
|
|
|
|
* it under the terms of the GNU General Public License as published by
|
|
|
|
* the Free Software Foundation; either version 2 of the License, or
|
|
|
|
* (at your option) any later version.
|
|
|
|
*
|
|
|
|
* This program is distributed in the hope that 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., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
|
|
|
|
*/
|
|
|
|
|
2007-08-21 02:58:30 +07:00
|
|
|
#include <linux/compiler.h>
|
2006-12-20 07:58:35 +07:00
|
|
|
#include <linux/delay.h>
|
2006-12-28 05:36:37 +07:00
|
|
|
#include <linux/dma-mapping.h>
|
2007-08-21 02:40:30 +07:00
|
|
|
#include <linux/gfp.h>
|
2007-08-21 02:41:22 +07:00
|
|
|
#include <linux/init.h>
|
|
|
|
#include <linux/interrupt.h>
|
|
|
|
#include <linux/kernel.h>
|
2007-05-16 02:36:10 +07:00
|
|
|
#include <linux/mm.h>
|
2007-08-21 02:41:22 +07:00
|
|
|
#include <linux/module.h>
|
2008-03-21 04:04:36 +07:00
|
|
|
#include <linux/moduleparam.h>
|
2007-08-21 02:41:22 +07:00
|
|
|
#include <linux/pci.h>
|
2007-08-21 02:40:30 +07:00
|
|
|
#include <linux/spinlock.h>
|
2006-12-28 05:36:37 +07:00
|
|
|
|
2007-08-21 02:40:30 +07:00
|
|
|
#include <asm/page.h>
|
2007-08-25 19:08:19 +07:00
|
|
|
#include <asm/system.h>
|
2006-12-20 07:58:35 +07:00
|
|
|
|
2008-03-01 08:42:56 +07:00
|
|
|
#ifdef CONFIG_PPC_PMAC
|
|
|
|
#include <asm/pmac_feature.h>
|
|
|
|
#endif
|
|
|
|
|
2006-12-20 07:58:35 +07:00
|
|
|
#include "fw-ohci.h"
|
2007-08-21 02:41:22 +07:00
|
|
|
#include "fw-transaction.h"
|
2006-12-20 07:58:35 +07:00
|
|
|
|
2007-05-08 07:33:35 +07:00
|
|
|
#define DESCRIPTOR_OUTPUT_MORE 0
|
|
|
|
#define DESCRIPTOR_OUTPUT_LAST (1 << 12)
|
|
|
|
#define DESCRIPTOR_INPUT_MORE (2 << 12)
|
|
|
|
#define DESCRIPTOR_INPUT_LAST (3 << 12)
|
|
|
|
#define DESCRIPTOR_STATUS (1 << 11)
|
|
|
|
#define DESCRIPTOR_KEY_IMMEDIATE (2 << 8)
|
|
|
|
#define DESCRIPTOR_PING (1 << 7)
|
|
|
|
#define DESCRIPTOR_YY (1 << 6)
|
|
|
|
#define DESCRIPTOR_NO_IRQ (0 << 4)
|
|
|
|
#define DESCRIPTOR_IRQ_ERROR (1 << 4)
|
|
|
|
#define DESCRIPTOR_IRQ_ALWAYS (3 << 4)
|
|
|
|
#define DESCRIPTOR_BRANCH_ALWAYS (3 << 2)
|
|
|
|
#define DESCRIPTOR_WAIT (3 << 0)
|
2006-12-20 07:58:35 +07:00
|
|
|
|
|
|
|
struct descriptor {
|
|
|
|
__le16 req_count;
|
|
|
|
__le16 control;
|
|
|
|
__le32 data_address;
|
|
|
|
__le32 branch_address;
|
|
|
|
__le16 res_count;
|
|
|
|
__le16 transfer_status;
|
|
|
|
} __attribute__((aligned(16)));
|
|
|
|
|
2007-02-17 05:34:40 +07:00
|
|
|
struct db_descriptor {
|
|
|
|
__le16 first_size;
|
|
|
|
__le16 control;
|
|
|
|
__le16 second_req_count;
|
|
|
|
__le16 first_req_count;
|
|
|
|
__le32 branch_address;
|
|
|
|
__le16 second_res_count;
|
|
|
|
__le16 first_res_count;
|
|
|
|
__le32 reserved0;
|
|
|
|
__le32 first_buffer;
|
|
|
|
__le32 second_buffer;
|
|
|
|
__le32 reserved1;
|
|
|
|
} __attribute__((aligned(16)));
|
|
|
|
|
2007-05-08 07:33:35 +07:00
|
|
|
#define CONTROL_SET(regs) (regs)
|
|
|
|
#define CONTROL_CLEAR(regs) ((regs) + 4)
|
|
|
|
#define COMMAND_PTR(regs) ((regs) + 12)
|
|
|
|
#define CONTEXT_MATCH(regs) ((regs) + 16)
|
2007-02-07 02:49:31 +07:00
|
|
|
|
2007-02-07 02:49:30 +07:00
|
|
|
struct ar_buffer {
|
2006-12-20 07:58:35 +07:00
|
|
|
struct descriptor descriptor;
|
2007-02-07 02:49:30 +07:00
|
|
|
struct ar_buffer *next;
|
|
|
|
__le32 data[0];
|
|
|
|
};
|
2006-12-20 07:58:35 +07:00
|
|
|
|
2007-02-07 02:49:30 +07:00
|
|
|
struct ar_context {
|
|
|
|
struct fw_ohci *ohci;
|
|
|
|
struct ar_buffer *current_buffer;
|
|
|
|
struct ar_buffer *last_buffer;
|
|
|
|
void *pointer;
|
2007-02-07 02:49:31 +07:00
|
|
|
u32 regs;
|
2006-12-20 07:58:35 +07:00
|
|
|
struct tasklet_struct tasklet;
|
|
|
|
};
|
|
|
|
|
2007-02-17 05:34:39 +07:00
|
|
|
struct context;
|
|
|
|
|
|
|
|
typedef int (*descriptor_callback_t)(struct context *ctx,
|
|
|
|
struct descriptor *d,
|
|
|
|
struct descriptor *last);
|
firewire: fw-ohci: Dynamically allocate buffers for DMA descriptors
Previously, the fw-ohci driver used fixed-length buffers for storing
descriptors for isochronous receive DMA programs. If an application
(such as libdc1394) generated a DMA program that was too large, fw-ohci
would reach the limit of its fixed-sized buffer and return an error to
userspace.
This patch replaces the fixed-length ring-buffer with a linked-list of
page-sized buffers. Additional buffers can be dynamically allocated and
appended to the list when necessary. For a particular context, buffers
are kept around after use and reused as necessary, so there is no
allocation taking place after the DMA program is generated for the first
time.
In addition, the buffers it uses are coherent for DMA so there is no
syncing required before and after writes. This syncing wasn't properly
done in the previous version of the code.
-
This is the fourth version of my patch that replaces a fixed-length
buffer for DMA descriptors with a dynamically allocated linked-list of
buffers.
As we discovered with the last attempt, new context programs are
sometimes queued from interrupt context, making it unacceptable to call
tasklet_disable() from context_get_descriptors().
This version of the patch uses ohci->lock for all locking needs instead
of tasklet_disable/enable. There is a new requirement that
context_get_descriptors() be called while holding ohci->lock. It was
already held for the AT context, so adding the requirement for the iso
context did not seem particularly onerous. In addition, this has the
side benefit of allowing iso queue to be safely called from concurrent
user-space threads, which previously was not safe.
Signed-off-by: David Moore <dcm@acm.org>
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
-
Fixes the following issues:
- Isochronous reception stopped prematurely if an application used a
larger buffer. (Reproduced with coriander.)
- Isochronous reception stopped after one or a few frames on VT630x
in OHCI 1.0 mode. (Fixes reception in coriander, but dvgrab still
doesn't work with these chips.)
Patch update: struct member alignment, whitespace nits
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2008-01-07 05:21:41 +07:00
|
|
|
|
|
|
|
/*
|
|
|
|
* A buffer that contains a block of DMA-able coherent memory used for
|
|
|
|
* storing a portion of a DMA descriptor program.
|
|
|
|
*/
|
|
|
|
struct descriptor_buffer {
|
|
|
|
struct list_head list;
|
|
|
|
dma_addr_t buffer_bus;
|
|
|
|
size_t buffer_size;
|
|
|
|
size_t used;
|
|
|
|
struct descriptor buffer[0];
|
|
|
|
};
|
|
|
|
|
2007-02-17 05:34:39 +07:00
|
|
|
struct context {
|
2007-03-04 20:45:18 +07:00
|
|
|
struct fw_ohci *ohci;
|
2007-02-17 05:34:39 +07:00
|
|
|
u32 regs;
|
firewire: fw-ohci: Dynamically allocate buffers for DMA descriptors
Previously, the fw-ohci driver used fixed-length buffers for storing
descriptors for isochronous receive DMA programs. If an application
(such as libdc1394) generated a DMA program that was too large, fw-ohci
would reach the limit of its fixed-sized buffer and return an error to
userspace.
This patch replaces the fixed-length ring-buffer with a linked-list of
page-sized buffers. Additional buffers can be dynamically allocated and
appended to the list when necessary. For a particular context, buffers
are kept around after use and reused as necessary, so there is no
allocation taking place after the DMA program is generated for the first
time.
In addition, the buffers it uses are coherent for DMA so there is no
syncing required before and after writes. This syncing wasn't properly
done in the previous version of the code.
-
This is the fourth version of my patch that replaces a fixed-length
buffer for DMA descriptors with a dynamically allocated linked-list of
buffers.
As we discovered with the last attempt, new context programs are
sometimes queued from interrupt context, making it unacceptable to call
tasklet_disable() from context_get_descriptors().
This version of the patch uses ohci->lock for all locking needs instead
of tasklet_disable/enable. There is a new requirement that
context_get_descriptors() be called while holding ohci->lock. It was
already held for the AT context, so adding the requirement for the iso
context did not seem particularly onerous. In addition, this has the
side benefit of allowing iso queue to be safely called from concurrent
user-space threads, which previously was not safe.
Signed-off-by: David Moore <dcm@acm.org>
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
-
Fixes the following issues:
- Isochronous reception stopped prematurely if an application used a
larger buffer. (Reproduced with coriander.)
- Isochronous reception stopped after one or a few frames on VT630x
in OHCI 1.0 mode. (Fixes reception in coriander, but dvgrab still
doesn't work with these chips.)
Patch update: struct member alignment, whitespace nits
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2008-01-07 05:21:41 +07:00
|
|
|
int total_allocation;
|
2007-03-04 20:45:18 +07:00
|
|
|
|
firewire: fw-ohci: Dynamically allocate buffers for DMA descriptors
Previously, the fw-ohci driver used fixed-length buffers for storing
descriptors for isochronous receive DMA programs. If an application
(such as libdc1394) generated a DMA program that was too large, fw-ohci
would reach the limit of its fixed-sized buffer and return an error to
userspace.
This patch replaces the fixed-length ring-buffer with a linked-list of
page-sized buffers. Additional buffers can be dynamically allocated and
appended to the list when necessary. For a particular context, buffers
are kept around after use and reused as necessary, so there is no
allocation taking place after the DMA program is generated for the first
time.
In addition, the buffers it uses are coherent for DMA so there is no
syncing required before and after writes. This syncing wasn't properly
done in the previous version of the code.
-
This is the fourth version of my patch that replaces a fixed-length
buffer for DMA descriptors with a dynamically allocated linked-list of
buffers.
As we discovered with the last attempt, new context programs are
sometimes queued from interrupt context, making it unacceptable to call
tasklet_disable() from context_get_descriptors().
This version of the patch uses ohci->lock for all locking needs instead
of tasklet_disable/enable. There is a new requirement that
context_get_descriptors() be called while holding ohci->lock. It was
already held for the AT context, so adding the requirement for the iso
context did not seem particularly onerous. In addition, this has the
side benefit of allowing iso queue to be safely called from concurrent
user-space threads, which previously was not safe.
Signed-off-by: David Moore <dcm@acm.org>
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
-
Fixes the following issues:
- Isochronous reception stopped prematurely if an application used a
larger buffer. (Reproduced with coriander.)
- Isochronous reception stopped after one or a few frames on VT630x
in OHCI 1.0 mode. (Fixes reception in coriander, but dvgrab still
doesn't work with these chips.)
Patch update: struct member alignment, whitespace nits
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2008-01-07 05:21:41 +07:00
|
|
|
/*
|
|
|
|
* List of page-sized buffers for storing DMA descriptors.
|
|
|
|
* Head of list contains buffers in use and tail of list contains
|
|
|
|
* free buffers.
|
|
|
|
*/
|
|
|
|
struct list_head buffer_list;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Pointer to a buffer inside buffer_list that contains the tail
|
|
|
|
* end of the current DMA program.
|
|
|
|
*/
|
|
|
|
struct descriptor_buffer *buffer_tail;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* The descriptor containing the branch address of the first
|
|
|
|
* descriptor that has not yet been filled by the device.
|
|
|
|
*/
|
|
|
|
struct descriptor *last;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* The last descriptor in the DMA program. It contains the branch
|
|
|
|
* address that must be updated upon appending a new descriptor.
|
|
|
|
*/
|
|
|
|
struct descriptor *prev;
|
2007-02-17 05:34:39 +07:00
|
|
|
|
|
|
|
descriptor_callback_t callback;
|
|
|
|
|
2007-03-04 20:45:18 +07:00
|
|
|
struct tasklet_struct tasklet;
|
2007-02-17 05:34:39 +07:00
|
|
|
};
|
|
|
|
|
2007-05-08 07:33:35 +07:00
|
|
|
#define IT_HEADER_SY(v) ((v) << 0)
|
|
|
|
#define IT_HEADER_TCODE(v) ((v) << 4)
|
|
|
|
#define IT_HEADER_CHANNEL(v) ((v) << 8)
|
|
|
|
#define IT_HEADER_TAG(v) ((v) << 14)
|
|
|
|
#define IT_HEADER_SPEED(v) ((v) << 16)
|
|
|
|
#define IT_HEADER_DATA_LENGTH(v) ((v) << 16)
|
2006-12-20 07:58:35 +07:00
|
|
|
|
|
|
|
struct iso_context {
|
|
|
|
struct fw_iso_context base;
|
2007-02-17 05:34:39 +07:00
|
|
|
struct context context;
|
2007-12-19 15:09:18 +07:00
|
|
|
int excess_bytes;
|
2007-02-17 05:34:44 +07:00
|
|
|
void *header;
|
|
|
|
size_t header_length;
|
2006-12-20 07:58:35 +07:00
|
|
|
};
|
|
|
|
|
|
|
|
#define CONFIG_ROM_SIZE 1024
|
|
|
|
|
|
|
|
struct fw_ohci {
|
|
|
|
struct fw_card card;
|
|
|
|
|
2007-02-17 05:34:49 +07:00
|
|
|
u32 version;
|
2006-12-20 07:58:35 +07:00
|
|
|
__iomem char *registers;
|
|
|
|
dma_addr_t self_id_bus;
|
|
|
|
__le32 *self_id_cpu;
|
|
|
|
struct tasklet_struct bus_reset_tasklet;
|
2007-01-26 12:38:04 +07:00
|
|
|
int node_id;
|
2006-12-20 07:58:35 +07:00
|
|
|
int generation;
|
|
|
|
int request_generation;
|
2007-03-08 00:12:56 +07:00
|
|
|
u32 bus_seconds;
|
2008-03-01 08:47:15 +07:00
|
|
|
bool old_uninorth;
|
2006-12-20 07:58:35 +07:00
|
|
|
|
2007-05-08 07:33:32 +07:00
|
|
|
/*
|
|
|
|
* Spinlock for accessing fw_ohci data. Never call out of
|
|
|
|
* this driver with this lock held.
|
|
|
|
*/
|
2006-12-20 07:58:35 +07:00
|
|
|
spinlock_t lock;
|
|
|
|
u32 self_id_buffer[512];
|
|
|
|
|
|
|
|
/* Config rom buffers */
|
|
|
|
__be32 *config_rom;
|
|
|
|
dma_addr_t config_rom_bus;
|
|
|
|
__be32 *next_config_rom;
|
|
|
|
dma_addr_t next_config_rom_bus;
|
|
|
|
u32 next_header;
|
|
|
|
|
|
|
|
struct ar_context ar_request_ctx;
|
|
|
|
struct ar_context ar_response_ctx;
|
2007-03-08 00:12:49 +07:00
|
|
|
struct context at_request_ctx;
|
|
|
|
struct context at_response_ctx;
|
2006-12-20 07:58:35 +07:00
|
|
|
|
|
|
|
u32 it_context_mask;
|
|
|
|
struct iso_context *it_context_list;
|
|
|
|
u32 ir_context_mask;
|
|
|
|
struct iso_context *ir_context_list;
|
|
|
|
};
|
|
|
|
|
2007-01-23 01:17:37 +07:00
|
|
|
static inline struct fw_ohci *fw_ohci(struct fw_card *card)
|
2006-12-20 07:58:35 +07:00
|
|
|
{
|
|
|
|
return container_of(card, struct fw_ohci, card);
|
|
|
|
}
|
|
|
|
|
2007-02-17 05:34:40 +07:00
|
|
|
#define IT_CONTEXT_CYCLE_MATCH_ENABLE 0x80000000
|
|
|
|
#define IR_CONTEXT_BUFFER_FILL 0x80000000
|
|
|
|
#define IR_CONTEXT_ISOCH_HEADER 0x40000000
|
|
|
|
#define IR_CONTEXT_CYCLE_MATCH_ENABLE 0x20000000
|
|
|
|
#define IR_CONTEXT_MULTI_CHANNEL_MODE 0x10000000
|
|
|
|
#define IR_CONTEXT_DUAL_BUFFER_MODE 0x08000000
|
2006-12-20 07:58:35 +07:00
|
|
|
|
|
|
|
#define CONTEXT_RUN 0x8000
|
|
|
|
#define CONTEXT_WAKE 0x1000
|
|
|
|
#define CONTEXT_DEAD 0x0800
|
|
|
|
#define CONTEXT_ACTIVE 0x0400
|
|
|
|
|
|
|
|
#define OHCI1394_MAX_AT_REQ_RETRIES 0x2
|
|
|
|
#define OHCI1394_MAX_AT_RESP_RETRIES 0x2
|
|
|
|
#define OHCI1394_MAX_PHYS_RESP_RETRIES 0x8
|
|
|
|
|
|
|
|
#define FW_OHCI_MAJOR 240
|
|
|
|
#define OHCI1394_REGISTER_SIZE 0x800
|
|
|
|
#define OHCI_LOOP_COUNT 500
|
|
|
|
#define OHCI1394_PCI_HCI_Control 0x40
|
|
|
|
#define SELF_ID_BUF_SIZE 0x800
|
2007-02-07 02:49:30 +07:00
|
|
|
#define OHCI_TCODE_PHY_PACKET 0x0e
|
2007-02-17 05:34:49 +07:00
|
|
|
#define OHCI_VERSION_1_1 0x010010
|
2007-01-26 12:38:49 +07:00
|
|
|
|
2006-12-20 07:58:35 +07:00
|
|
|
static char ohci_driver_name[] = KBUILD_MODNAME;
|
|
|
|
|
2008-03-21 04:04:36 +07:00
|
|
|
#ifdef CONFIG_FIREWIRE_OHCI_DEBUG
|
|
|
|
|
|
|
|
#define OHCI_PARAM_DEBUG_IRQS 1
|
|
|
|
#define OHCI_PARAM_DEBUG_SELFIDS 2
|
|
|
|
#define OHCI_PARAM_DEBUG_AT_AR 4
|
|
|
|
|
|
|
|
static int param_debug;
|
|
|
|
module_param_named(debug, param_debug, int, 0644);
|
|
|
|
MODULE_PARM_DESC(debug, "Verbose logging (default = 0"
|
|
|
|
", IRQs = " __stringify(OHCI_PARAM_DEBUG_IRQS)
|
|
|
|
", self-IDs = " __stringify(OHCI_PARAM_DEBUG_SELFIDS)
|
|
|
|
", AT/AR events = " __stringify(OHCI_PARAM_DEBUG_AT_AR)
|
|
|
|
", or a combination, or all = -1)");
|
|
|
|
|
|
|
|
static void log_irqs(u32 evt)
|
|
|
|
{
|
|
|
|
if (likely(!(param_debug & OHCI_PARAM_DEBUG_IRQS)))
|
|
|
|
return;
|
|
|
|
|
|
|
|
printk(KERN_DEBUG KBUILD_MODNAME ": IRQ %08x%s%s%s%s%s%s%s%s%s%s%s\n",
|
|
|
|
evt,
|
|
|
|
evt & OHCI1394_selfIDComplete ? " selfID" : "",
|
|
|
|
evt & OHCI1394_RQPkt ? " AR_req" : "",
|
|
|
|
evt & OHCI1394_RSPkt ? " AR_resp" : "",
|
|
|
|
evt & OHCI1394_reqTxComplete ? " AT_req" : "",
|
|
|
|
evt & OHCI1394_respTxComplete ? " AT_resp" : "",
|
|
|
|
evt & OHCI1394_isochRx ? " IR" : "",
|
|
|
|
evt & OHCI1394_isochTx ? " IT" : "",
|
|
|
|
evt & OHCI1394_postedWriteErr ? " postedWriteErr" : "",
|
|
|
|
evt & OHCI1394_cycleTooLong ? " cycleTooLong" : "",
|
|
|
|
evt & OHCI1394_cycle64Seconds ? " cycle64Seconds" : "",
|
|
|
|
evt & ~(OHCI1394_selfIDComplete | OHCI1394_RQPkt |
|
|
|
|
OHCI1394_RSPkt | OHCI1394_reqTxComplete |
|
|
|
|
OHCI1394_respTxComplete | OHCI1394_isochRx |
|
|
|
|
OHCI1394_isochTx | OHCI1394_postedWriteErr |
|
|
|
|
OHCI1394_cycleTooLong | OHCI1394_cycle64Seconds)
|
|
|
|
? " ?" : "");
|
|
|
|
}
|
|
|
|
|
|
|
|
static const char *speed[] = {
|
|
|
|
[0] = "S100", [1] = "S200", [2] = "S400", [3] = "beta",
|
|
|
|
};
|
|
|
|
static const char *power[] = {
|
|
|
|
[0] = "+0W", [1] = "+15W", [2] = "+30W", [3] = "+45W",
|
|
|
|
[4] = "-3W", [5] = " ?W", [6] = "-3..-6W", [7] = "-3..-10W",
|
|
|
|
};
|
|
|
|
static const char port[] = { '.', '-', 'p', 'c', };
|
|
|
|
|
|
|
|
static char _p(u32 *s, int shift)
|
|
|
|
{
|
|
|
|
return port[*s >> shift & 3];
|
|
|
|
}
|
|
|
|
|
|
|
|
static void log_selfids(int generation, int self_id_count, u32 *s)
|
|
|
|
{
|
|
|
|
if (likely(!(param_debug & OHCI_PARAM_DEBUG_SELFIDS)))
|
|
|
|
return;
|
|
|
|
|
|
|
|
printk(KERN_DEBUG KBUILD_MODNAME ": %d selfIDs, generation %d\n",
|
|
|
|
self_id_count, generation);
|
|
|
|
|
|
|
|
for (; self_id_count--; ++s)
|
|
|
|
if ((*s & 1 << 23) == 0)
|
|
|
|
printk(KERN_DEBUG "selfID 0: %08x, phy %d [%c%c%c] "
|
|
|
|
"%s gc=%d %s %s%s%s\n",
|
|
|
|
*s, *s >> 24 & 63, _p(s, 6), _p(s, 4), _p(s, 2),
|
|
|
|
speed[*s >> 14 & 3], *s >> 16 & 63,
|
|
|
|
power[*s >> 8 & 7], *s >> 22 & 1 ? "L" : "",
|
|
|
|
*s >> 11 & 1 ? "c" : "", *s & 2 ? "i" : "");
|
|
|
|
else
|
|
|
|
printk(KERN_DEBUG "selfID n: %08x, phy %d "
|
|
|
|
"[%c%c%c%c%c%c%c%c]\n",
|
|
|
|
*s, *s >> 24 & 63,
|
|
|
|
_p(s, 16), _p(s, 14), _p(s, 12), _p(s, 10),
|
|
|
|
_p(s, 8), _p(s, 6), _p(s, 4), _p(s, 2));
|
|
|
|
}
|
|
|
|
|
|
|
|
static const char *evts[] = {
|
|
|
|
[0x00] = "evt_no_status", [0x01] = "-reserved-",
|
|
|
|
[0x02] = "evt_long_packet", [0x03] = "evt_missing_ack",
|
|
|
|
[0x04] = "evt_underrun", [0x05] = "evt_overrun",
|
|
|
|
[0x06] = "evt_descriptor_read", [0x07] = "evt_data_read",
|
|
|
|
[0x08] = "evt_data_write", [0x09] = "evt_bus_reset",
|
|
|
|
[0x0a] = "evt_timeout", [0x0b] = "evt_tcode_err",
|
|
|
|
[0x0c] = "-reserved-", [0x0d] = "-reserved-",
|
|
|
|
[0x0e] = "evt_unknown", [0x0f] = "evt_flushed",
|
|
|
|
[0x10] = "-reserved-", [0x11] = "ack_complete",
|
|
|
|
[0x12] = "ack_pending ", [0x13] = "-reserved-",
|
|
|
|
[0x14] = "ack_busy_X", [0x15] = "ack_busy_A",
|
|
|
|
[0x16] = "ack_busy_B", [0x17] = "-reserved-",
|
|
|
|
[0x18] = "-reserved-", [0x19] = "-reserved-",
|
|
|
|
[0x1a] = "-reserved-", [0x1b] = "ack_tardy",
|
|
|
|
[0x1c] = "-reserved-", [0x1d] = "ack_data_error",
|
|
|
|
[0x1e] = "ack_type_error", [0x1f] = "-reserved-",
|
|
|
|
[0x20] = "pending/cancelled",
|
|
|
|
};
|
|
|
|
static const char *tcodes[] = {
|
|
|
|
[0x0] = "QW req", [0x1] = "BW req",
|
|
|
|
[0x2] = "W resp", [0x3] = "-reserved-",
|
|
|
|
[0x4] = "QR req", [0x5] = "BR req",
|
|
|
|
[0x6] = "QR resp", [0x7] = "BR resp",
|
|
|
|
[0x8] = "cycle start", [0x9] = "Lk req",
|
|
|
|
[0xa] = "async stream packet", [0xb] = "Lk resp",
|
|
|
|
[0xc] = "-reserved-", [0xd] = "-reserved-",
|
|
|
|
[0xe] = "link internal", [0xf] = "-reserved-",
|
|
|
|
};
|
|
|
|
static const char *phys[] = {
|
|
|
|
[0x0] = "phy config packet", [0x1] = "link-on packet",
|
|
|
|
[0x2] = "self-id packet", [0x3] = "-reserved-",
|
|
|
|
};
|
|
|
|
|
|
|
|
static void log_ar_at_event(char dir, int speed, u32 *header, int evt)
|
|
|
|
{
|
|
|
|
int tcode = header[0] >> 4 & 0xf;
|
|
|
|
char specific[12];
|
|
|
|
|
|
|
|
if (likely(!(param_debug & OHCI_PARAM_DEBUG_AT_AR)))
|
|
|
|
return;
|
|
|
|
|
|
|
|
if (unlikely(evt >= ARRAY_SIZE(evts)))
|
|
|
|
evt = 0x1f;
|
|
|
|
|
|
|
|
if (header[0] == ~header[1]) {
|
|
|
|
printk(KERN_DEBUG "A%c %s, %s, %08x\n",
|
|
|
|
dir, evts[evt], phys[header[0] >> 30 & 0x3],
|
|
|
|
header[0]);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
switch (tcode) {
|
|
|
|
case 0x0: case 0x6: case 0x8:
|
|
|
|
snprintf(specific, sizeof(specific), " = %08x",
|
|
|
|
be32_to_cpu((__force __be32)header[3]));
|
|
|
|
break;
|
|
|
|
case 0x1: case 0x5: case 0x7: case 0x9: case 0xb:
|
|
|
|
snprintf(specific, sizeof(specific), " %x,%x",
|
|
|
|
header[3] >> 16, header[3] & 0xffff);
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
specific[0] = '\0';
|
|
|
|
}
|
|
|
|
|
|
|
|
switch (tcode) {
|
|
|
|
case 0xe: case 0xa:
|
|
|
|
printk(KERN_DEBUG "A%c %s, %s\n",
|
|
|
|
dir, evts[evt], tcodes[tcode]);
|
|
|
|
break;
|
|
|
|
case 0x0: case 0x1: case 0x4: case 0x5: case 0x9:
|
|
|
|
printk(KERN_DEBUG "A%c spd %x tl %02x, "
|
|
|
|
"%04x -> %04x, %s, "
|
|
|
|
"%s, %04x%08x%s\n",
|
|
|
|
dir, speed, header[0] >> 10 & 0x3f,
|
|
|
|
header[1] >> 16, header[0] >> 16, evts[evt],
|
|
|
|
tcodes[tcode], header[1] & 0xffff, header[2], specific);
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
printk(KERN_DEBUG "A%c spd %x tl %02x, "
|
|
|
|
"%04x -> %04x, %s, "
|
|
|
|
"%s%s\n",
|
|
|
|
dir, speed, header[0] >> 10 & 0x3f,
|
|
|
|
header[1] >> 16, header[0] >> 16, evts[evt],
|
|
|
|
tcodes[tcode], specific);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
#else
|
|
|
|
|
|
|
|
#define log_irqs(evt)
|
|
|
|
#define log_selfids(generation, self_id_count, sid)
|
|
|
|
#define log_ar_at_event(dir, speed, header, evt)
|
|
|
|
|
|
|
|
#endif /* CONFIG_FIREWIRE_OHCI_DEBUG */
|
|
|
|
|
2007-01-23 01:17:37 +07:00
|
|
|
static inline void reg_write(const struct fw_ohci *ohci, int offset, u32 data)
|
2006-12-20 07:58:35 +07:00
|
|
|
{
|
|
|
|
writel(data, ohci->registers + offset);
|
|
|
|
}
|
|
|
|
|
2007-01-23 01:17:37 +07:00
|
|
|
static inline u32 reg_read(const struct fw_ohci *ohci, int offset)
|
2006-12-20 07:58:35 +07:00
|
|
|
{
|
|
|
|
return readl(ohci->registers + offset);
|
|
|
|
}
|
|
|
|
|
2007-01-23 01:17:37 +07:00
|
|
|
static inline void flush_writes(const struct fw_ohci *ohci)
|
2006-12-20 07:58:35 +07:00
|
|
|
{
|
|
|
|
/* Do a dummy read to flush writes. */
|
|
|
|
reg_read(ohci, OHCI1394_Version);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int
|
|
|
|
ohci_update_phy_reg(struct fw_card *card, int addr,
|
|
|
|
int clear_bits, int set_bits)
|
|
|
|
{
|
|
|
|
struct fw_ohci *ohci = fw_ohci(card);
|
|
|
|
u32 val, old;
|
|
|
|
|
|
|
|
reg_write(ohci, OHCI1394_PhyControl, OHCI1394_PhyControl_Read(addr));
|
2007-07-13 03:24:19 +07:00
|
|
|
flush_writes(ohci);
|
2006-12-20 07:58:35 +07:00
|
|
|
msleep(2);
|
|
|
|
val = reg_read(ohci, OHCI1394_PhyControl);
|
|
|
|
if ((val & OHCI1394_PhyControl_ReadDone) == 0) {
|
|
|
|
fw_error("failed to set phy reg bits.\n");
|
|
|
|
return -EBUSY;
|
|
|
|
}
|
|
|
|
|
|
|
|
old = OHCI1394_PhyControl_ReadData(val);
|
|
|
|
old = (old & ~clear_bits) | set_bits;
|
|
|
|
reg_write(ohci, OHCI1394_PhyControl,
|
|
|
|
OHCI1394_PhyControl_Write(addr, old));
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2007-02-07 02:49:30 +07:00
|
|
|
static int ar_context_add_page(struct ar_context *ctx)
|
2006-12-20 07:58:35 +07:00
|
|
|
{
|
2007-02-07 02:49:30 +07:00
|
|
|
struct device *dev = ctx->ohci->card.device;
|
|
|
|
struct ar_buffer *ab;
|
2008-03-14 06:27:49 +07:00
|
|
|
dma_addr_t uninitialized_var(ab_bus);
|
2007-02-07 02:49:30 +07:00
|
|
|
size_t offset;
|
|
|
|
|
2008-03-13 04:43:26 +07:00
|
|
|
ab = dma_alloc_coherent(dev, PAGE_SIZE, &ab_bus, GFP_ATOMIC);
|
2007-02-07 02:49:30 +07:00
|
|
|
if (ab == NULL)
|
|
|
|
return -ENOMEM;
|
|
|
|
|
2007-05-10 06:23:14 +07:00
|
|
|
memset(&ab->descriptor, 0, sizeof(ab->descriptor));
|
2007-05-08 07:33:35 +07:00
|
|
|
ab->descriptor.control = cpu_to_le16(DESCRIPTOR_INPUT_MORE |
|
|
|
|
DESCRIPTOR_STATUS |
|
|
|
|
DESCRIPTOR_BRANCH_ALWAYS);
|
2007-02-07 02:49:30 +07:00
|
|
|
offset = offsetof(struct ar_buffer, data);
|
|
|
|
ab->descriptor.req_count = cpu_to_le16(PAGE_SIZE - offset);
|
|
|
|
ab->descriptor.data_address = cpu_to_le32(ab_bus + offset);
|
|
|
|
ab->descriptor.res_count = cpu_to_le16(PAGE_SIZE - offset);
|
|
|
|
ab->descriptor.branch_address = 0;
|
|
|
|
|
2007-05-23 05:55:48 +07:00
|
|
|
ctx->last_buffer->descriptor.branch_address = cpu_to_le32(ab_bus | 1);
|
2007-02-07 02:49:30 +07:00
|
|
|
ctx->last_buffer->next = ab;
|
|
|
|
ctx->last_buffer = ab;
|
|
|
|
|
2007-05-08 07:33:35 +07:00
|
|
|
reg_write(ctx->ohci, CONTROL_SET(ctx->regs), CONTEXT_WAKE);
|
2006-12-20 07:58:35 +07:00
|
|
|
flush_writes(ctx->ohci);
|
2007-02-07 02:49:30 +07:00
|
|
|
|
|
|
|
return 0;
|
2006-12-20 07:58:35 +07:00
|
|
|
}
|
|
|
|
|
2008-03-01 08:47:15 +07:00
|
|
|
#if defined(CONFIG_PPC_PMAC) && defined(CONFIG_PPC32)
|
|
|
|
#define cond_le32_to_cpu(v) \
|
|
|
|
(ohci->old_uninorth ? (__force __u32)(v) : le32_to_cpu(v))
|
|
|
|
#else
|
|
|
|
#define cond_le32_to_cpu(v) le32_to_cpu(v)
|
|
|
|
#endif
|
|
|
|
|
2007-02-07 02:49:30 +07:00
|
|
|
static __le32 *handle_ar_packet(struct ar_context *ctx, __le32 *buffer)
|
2006-12-20 07:58:35 +07:00
|
|
|
{
|
|
|
|
struct fw_ohci *ohci = ctx->ohci;
|
2007-01-26 12:37:57 +07:00
|
|
|
struct fw_packet p;
|
|
|
|
u32 status, length, tcode;
|
|
|
|
|
2008-03-01 08:47:15 +07:00
|
|
|
p.header[0] = cond_le32_to_cpu(buffer[0]);
|
|
|
|
p.header[1] = cond_le32_to_cpu(buffer[1]);
|
|
|
|
p.header[2] = cond_le32_to_cpu(buffer[2]);
|
2007-01-26 12:37:57 +07:00
|
|
|
|
|
|
|
tcode = (p.header[0] >> 4) & 0x0f;
|
|
|
|
switch (tcode) {
|
|
|
|
case TCODE_WRITE_QUADLET_REQUEST:
|
|
|
|
case TCODE_READ_QUADLET_RESPONSE:
|
2007-02-07 02:49:30 +07:00
|
|
|
p.header[3] = (__force __u32) buffer[3];
|
2007-01-26 12:37:57 +07:00
|
|
|
p.header_length = 16;
|
2007-02-07 02:49:30 +07:00
|
|
|
p.payload_length = 0;
|
2007-01-26 12:37:57 +07:00
|
|
|
break;
|
|
|
|
|
|
|
|
case TCODE_READ_BLOCK_REQUEST :
|
2008-03-01 08:47:15 +07:00
|
|
|
p.header[3] = cond_le32_to_cpu(buffer[3]);
|
2007-02-07 02:49:30 +07:00
|
|
|
p.header_length = 16;
|
|
|
|
p.payload_length = 0;
|
|
|
|
break;
|
|
|
|
|
|
|
|
case TCODE_WRITE_BLOCK_REQUEST:
|
2007-01-26 12:37:57 +07:00
|
|
|
case TCODE_READ_BLOCK_RESPONSE:
|
|
|
|
case TCODE_LOCK_REQUEST:
|
|
|
|
case TCODE_LOCK_RESPONSE:
|
2008-03-01 08:47:15 +07:00
|
|
|
p.header[3] = cond_le32_to_cpu(buffer[3]);
|
2007-01-26 12:37:57 +07:00
|
|
|
p.header_length = 16;
|
2007-02-07 02:49:30 +07:00
|
|
|
p.payload_length = p.header[3] >> 16;
|
2007-01-26 12:37:57 +07:00
|
|
|
break;
|
|
|
|
|
|
|
|
case TCODE_WRITE_RESPONSE:
|
|
|
|
case TCODE_READ_QUADLET_REQUEST:
|
2007-02-07 02:49:30 +07:00
|
|
|
case OHCI_TCODE_PHY_PACKET:
|
2007-01-26 12:37:57 +07:00
|
|
|
p.header_length = 12;
|
2007-02-07 02:49:30 +07:00
|
|
|
p.payload_length = 0;
|
2007-01-26 12:37:57 +07:00
|
|
|
break;
|
|
|
|
}
|
2006-12-20 07:58:35 +07:00
|
|
|
|
2007-02-07 02:49:30 +07:00
|
|
|
p.payload = (void *) buffer + p.header_length;
|
|
|
|
|
|
|
|
/* FIXME: What to do about evt_* errors? */
|
|
|
|
length = (p.header_length + p.payload_length + 3) / 4;
|
2008-03-01 08:47:15 +07:00
|
|
|
status = cond_le32_to_cpu(buffer[length]);
|
2007-02-07 02:49:30 +07:00
|
|
|
|
|
|
|
p.ack = ((status >> 16) & 0x1f) - 16;
|
|
|
|
p.speed = (status >> 21) & 0x7;
|
|
|
|
p.timestamp = status & 0xffff;
|
|
|
|
p.generation = ohci->request_generation;
|
2006-12-20 07:58:35 +07:00
|
|
|
|
2008-03-21 04:04:36 +07:00
|
|
|
log_ar_at_event('R', p.speed, p.header, status >> 16 & 0x1f);
|
|
|
|
|
2007-05-08 07:33:32 +07:00
|
|
|
/*
|
|
|
|
* The OHCI bus reset handler synthesizes a phy packet with
|
2006-12-20 07:58:35 +07:00
|
|
|
* the new generation number when a bus reset happens (see
|
|
|
|
* section 8.4.2.3). This helps us determine when a request
|
|
|
|
* was received and make sure we send the response in the same
|
|
|
|
* generation. We only need this for requests; for responses
|
|
|
|
* we use the unique tlabel for finding the matching
|
2007-05-08 07:33:32 +07:00
|
|
|
* request.
|
|
|
|
*/
|
2006-12-20 07:58:35 +07:00
|
|
|
|
2007-01-26 12:37:57 +07:00
|
|
|
if (p.ack + 16 == 0x09)
|
2008-02-23 18:24:17 +07:00
|
|
|
ohci->request_generation = (p.header[2] >> 16) & 0xff;
|
2006-12-20 07:58:35 +07:00
|
|
|
else if (ctx == &ohci->ar_request_ctx)
|
2007-01-26 12:37:57 +07:00
|
|
|
fw_core_handle_request(&ohci->card, &p);
|
2006-12-20 07:58:35 +07:00
|
|
|
else
|
2007-01-26 12:37:57 +07:00
|
|
|
fw_core_handle_response(&ohci->card, &p);
|
2006-12-20 07:58:35 +07:00
|
|
|
|
2007-02-07 02:49:30 +07:00
|
|
|
return buffer + length + 1;
|
|
|
|
}
|
2006-12-20 07:58:35 +07:00
|
|
|
|
2007-02-07 02:49:30 +07:00
|
|
|
static void ar_context_tasklet(unsigned long data)
|
|
|
|
{
|
|
|
|
struct ar_context *ctx = (struct ar_context *)data;
|
|
|
|
struct fw_ohci *ohci = ctx->ohci;
|
|
|
|
struct ar_buffer *ab;
|
|
|
|
struct descriptor *d;
|
|
|
|
void *buffer, *end;
|
|
|
|
|
|
|
|
ab = ctx->current_buffer;
|
|
|
|
d = &ab->descriptor;
|
|
|
|
|
|
|
|
if (d->res_count == 0) {
|
|
|
|
size_t size, rest, offset;
|
2008-03-26 03:47:16 +07:00
|
|
|
dma_addr_t start_bus;
|
|
|
|
void *start;
|
2007-02-07 02:49:30 +07:00
|
|
|
|
2007-05-08 07:33:32 +07:00
|
|
|
/*
|
|
|
|
* This descriptor is finished and we may have a
|
2007-02-07 02:49:30 +07:00
|
|
|
* packet split across this and the next buffer. We
|
2007-05-08 07:33:32 +07:00
|
|
|
* reuse the page for reassembling the split packet.
|
|
|
|
*/
|
2007-02-07 02:49:30 +07:00
|
|
|
|
|
|
|
offset = offsetof(struct ar_buffer, data);
|
2008-03-26 03:47:16 +07:00
|
|
|
start = buffer = ab;
|
|
|
|
start_bus = le32_to_cpu(ab->descriptor.data_address) - offset;
|
2007-02-07 02:49:30 +07:00
|
|
|
|
|
|
|
ab = ab->next;
|
|
|
|
d = &ab->descriptor;
|
|
|
|
size = buffer + PAGE_SIZE - ctx->pointer;
|
|
|
|
rest = le16_to_cpu(d->req_count) - le16_to_cpu(d->res_count);
|
|
|
|
memmove(buffer, ctx->pointer, size);
|
|
|
|
memcpy(buffer + size, ab->data, rest);
|
|
|
|
ctx->current_buffer = ab;
|
|
|
|
ctx->pointer = (void *) ab->data + rest;
|
|
|
|
end = buffer + size + rest;
|
|
|
|
|
|
|
|
while (buffer < end)
|
|
|
|
buffer = handle_ar_packet(ctx, buffer);
|
|
|
|
|
2008-03-13 04:43:26 +07:00
|
|
|
dma_free_coherent(ohci->card.device, PAGE_SIZE,
|
2008-03-26 03:47:16 +07:00
|
|
|
start, start_bus);
|
2007-02-07 02:49:30 +07:00
|
|
|
ar_context_add_page(ctx);
|
|
|
|
} else {
|
|
|
|
buffer = ctx->pointer;
|
|
|
|
ctx->pointer = end =
|
|
|
|
(void *) ab + PAGE_SIZE - le16_to_cpu(d->res_count);
|
|
|
|
|
|
|
|
while (buffer < end)
|
|
|
|
buffer = handle_ar_packet(ctx, buffer);
|
|
|
|
}
|
2006-12-20 07:58:35 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
static int
|
2007-02-07 02:49:31 +07:00
|
|
|
ar_context_init(struct ar_context *ctx, struct fw_ohci *ohci, u32 regs)
|
2006-12-20 07:58:35 +07:00
|
|
|
{
|
2007-02-07 02:49:30 +07:00
|
|
|
struct ar_buffer ab;
|
2006-12-20 07:58:35 +07:00
|
|
|
|
2007-02-07 02:49:31 +07:00
|
|
|
ctx->regs = regs;
|
|
|
|
ctx->ohci = ohci;
|
|
|
|
ctx->last_buffer = &ab;
|
2006-12-20 07:58:35 +07:00
|
|
|
tasklet_init(&ctx->tasklet, ar_context_tasklet, (unsigned long)ctx);
|
|
|
|
|
2007-02-07 02:49:30 +07:00
|
|
|
ar_context_add_page(ctx);
|
|
|
|
ar_context_add_page(ctx);
|
|
|
|
ctx->current_buffer = ab.next;
|
|
|
|
ctx->pointer = ctx->current_buffer->data;
|
|
|
|
|
2007-05-31 06:06:35 +07:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void ar_context_run(struct ar_context *ctx)
|
|
|
|
{
|
|
|
|
struct ar_buffer *ab = ctx->current_buffer;
|
|
|
|
dma_addr_t ab_bus;
|
|
|
|
size_t offset;
|
|
|
|
|
|
|
|
offset = offsetof(struct ar_buffer, data);
|
2007-06-24 01:28:17 +07:00
|
|
|
ab_bus = le32_to_cpu(ab->descriptor.data_address) - offset;
|
2007-05-31 06:06:35 +07:00
|
|
|
|
|
|
|
reg_write(ctx->ohci, COMMAND_PTR(ctx->regs), ab_bus | 1);
|
2007-05-08 07:33:35 +07:00
|
|
|
reg_write(ctx->ohci, CONTROL_SET(ctx->regs), CONTEXT_RUN);
|
2007-02-07 02:49:30 +07:00
|
|
|
flush_writes(ctx->ohci);
|
2006-12-20 07:58:35 +07:00
|
|
|
}
|
2007-03-04 20:45:18 +07:00
|
|
|
|
firewire: OHCI 1.0 Isochronous Receive support
Third rendition of FireWire OHCI 1.0 Isochronous Receive support, using a
zer-copy method similar to OHCI 1.1 which puts the IR data payload directly
into the userspace buffer. The zero-copy implementation eliminates the
video artifacts, audio popping, and buffer underrun problems seen with
version 1 of this patch, as well as fixing a regression in OHCI 1.1 support
introduced by version 2 of this patch.
Successfully tested in OHCI 1.1 mode on the following chipsets:
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
- Ti TSB41AB2 (rev 01), OHCI 1.1 (PCI on SB Audigy)
- Apple UniNorth 2 (rev 81), OHCI 1.1 (PowerBook G4 onboard)
Successfully tested in OHCI 1.0 mode on the following chipsets:
- Agere FW323 (rev 06), OHCI 1.0 (Mac Mini onboard)
- Agere FW323 (rev 06), OHCI 1.0 (PCI)
- Via VT6306 (rev 46), OHCI 1.0 (PCI)
- NEC OrangeLink (rev 01), OHCI 1.0 (PCI)
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
The bulk of testing was done in an x86_64 system, but was also successfully
sanity-tested on other systems, including a PPC(32) PowerBook G4 and an i686
EPIA M10k. Crude benchmarking (watching top during capture) puts the cpu
utilization during capture on the EPIA's 1GHz Via C3 processor around 13%,
which is down from 30% with the v1 code.
Some implementation details:
To maintain the same userspace API as dual-buffer mode, we set up two
descriptors for every incoming packet. The first is an INPUT_MORE descriptor,
pointing to a buffer large enough to hold just the packet's iso headers,
immediately followed by an INPUT_LAST descriptor, pointing to a chunk of the
userspace buffer big enough for the packet's data payload. With this setup,
each incoming packet fills in these two descriptors in a manner that very
closely emulates dual-buffer receive, to the point where the bulk of the
handle_ir_* code is now identical between the two (and probably primed for
some restructuring to share code between them).
The only caveat I have at the moment is that neither of my OHCI 1.0 Via
VT6307-based FireWire controllers work particularly well with this code
for reasons I have yet to figure out.
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2007-12-04 01:43:12 +07:00
|
|
|
static struct descriptor *
|
|
|
|
find_branch_descriptor(struct descriptor *d, int z)
|
|
|
|
{
|
|
|
|
int b, key;
|
|
|
|
|
|
|
|
b = (le16_to_cpu(d->control) & DESCRIPTOR_BRANCH_ALWAYS) >> 2;
|
|
|
|
key = (le16_to_cpu(d->control) & DESCRIPTOR_KEY_IMMEDIATE) >> 8;
|
|
|
|
|
|
|
|
/* figure out which descriptor the branch address goes in */
|
|
|
|
if (z == 2 && (b == 3 || key == 2))
|
|
|
|
return d;
|
|
|
|
else
|
|
|
|
return d + z - 1;
|
|
|
|
}
|
|
|
|
|
2007-02-17 05:34:39 +07:00
|
|
|
static void context_tasklet(unsigned long data)
|
|
|
|
{
|
|
|
|
struct context *ctx = (struct context *) data;
|
|
|
|
struct descriptor *d, *last;
|
|
|
|
u32 address;
|
|
|
|
int z;
|
firewire: fw-ohci: Dynamically allocate buffers for DMA descriptors
Previously, the fw-ohci driver used fixed-length buffers for storing
descriptors for isochronous receive DMA programs. If an application
(such as libdc1394) generated a DMA program that was too large, fw-ohci
would reach the limit of its fixed-sized buffer and return an error to
userspace.
This patch replaces the fixed-length ring-buffer with a linked-list of
page-sized buffers. Additional buffers can be dynamically allocated and
appended to the list when necessary. For a particular context, buffers
are kept around after use and reused as necessary, so there is no
allocation taking place after the DMA program is generated for the first
time.
In addition, the buffers it uses are coherent for DMA so there is no
syncing required before and after writes. This syncing wasn't properly
done in the previous version of the code.
-
This is the fourth version of my patch that replaces a fixed-length
buffer for DMA descriptors with a dynamically allocated linked-list of
buffers.
As we discovered with the last attempt, new context programs are
sometimes queued from interrupt context, making it unacceptable to call
tasklet_disable() from context_get_descriptors().
This version of the patch uses ohci->lock for all locking needs instead
of tasklet_disable/enable. There is a new requirement that
context_get_descriptors() be called while holding ohci->lock. It was
already held for the AT context, so adding the requirement for the iso
context did not seem particularly onerous. In addition, this has the
side benefit of allowing iso queue to be safely called from concurrent
user-space threads, which previously was not safe.
Signed-off-by: David Moore <dcm@acm.org>
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
-
Fixes the following issues:
- Isochronous reception stopped prematurely if an application used a
larger buffer. (Reproduced with coriander.)
- Isochronous reception stopped after one or a few frames on VT630x
in OHCI 1.0 mode. (Fixes reception in coriander, but dvgrab still
doesn't work with these chips.)
Patch update: struct member alignment, whitespace nits
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2008-01-07 05:21:41 +07:00
|
|
|
struct descriptor_buffer *desc;
|
2007-02-17 05:34:39 +07:00
|
|
|
|
firewire: fw-ohci: Dynamically allocate buffers for DMA descriptors
Previously, the fw-ohci driver used fixed-length buffers for storing
descriptors for isochronous receive DMA programs. If an application
(such as libdc1394) generated a DMA program that was too large, fw-ohci
would reach the limit of its fixed-sized buffer and return an error to
userspace.
This patch replaces the fixed-length ring-buffer with a linked-list of
page-sized buffers. Additional buffers can be dynamically allocated and
appended to the list when necessary. For a particular context, buffers
are kept around after use and reused as necessary, so there is no
allocation taking place after the DMA program is generated for the first
time.
In addition, the buffers it uses are coherent for DMA so there is no
syncing required before and after writes. This syncing wasn't properly
done in the previous version of the code.
-
This is the fourth version of my patch that replaces a fixed-length
buffer for DMA descriptors with a dynamically allocated linked-list of
buffers.
As we discovered with the last attempt, new context programs are
sometimes queued from interrupt context, making it unacceptable to call
tasklet_disable() from context_get_descriptors().
This version of the patch uses ohci->lock for all locking needs instead
of tasklet_disable/enable. There is a new requirement that
context_get_descriptors() be called while holding ohci->lock. It was
already held for the AT context, so adding the requirement for the iso
context did not seem particularly onerous. In addition, this has the
side benefit of allowing iso queue to be safely called from concurrent
user-space threads, which previously was not safe.
Signed-off-by: David Moore <dcm@acm.org>
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
-
Fixes the following issues:
- Isochronous reception stopped prematurely if an application used a
larger buffer. (Reproduced with coriander.)
- Isochronous reception stopped after one or a few frames on VT630x
in OHCI 1.0 mode. (Fixes reception in coriander, but dvgrab still
doesn't work with these chips.)
Patch update: struct member alignment, whitespace nits
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2008-01-07 05:21:41 +07:00
|
|
|
desc = list_entry(ctx->buffer_list.next,
|
|
|
|
struct descriptor_buffer, list);
|
|
|
|
last = ctx->last;
|
2007-02-17 05:34:39 +07:00
|
|
|
while (last->branch_address != 0) {
|
firewire: fw-ohci: Dynamically allocate buffers for DMA descriptors
Previously, the fw-ohci driver used fixed-length buffers for storing
descriptors for isochronous receive DMA programs. If an application
(such as libdc1394) generated a DMA program that was too large, fw-ohci
would reach the limit of its fixed-sized buffer and return an error to
userspace.
This patch replaces the fixed-length ring-buffer with a linked-list of
page-sized buffers. Additional buffers can be dynamically allocated and
appended to the list when necessary. For a particular context, buffers
are kept around after use and reused as necessary, so there is no
allocation taking place after the DMA program is generated for the first
time.
In addition, the buffers it uses are coherent for DMA so there is no
syncing required before and after writes. This syncing wasn't properly
done in the previous version of the code.
-
This is the fourth version of my patch that replaces a fixed-length
buffer for DMA descriptors with a dynamically allocated linked-list of
buffers.
As we discovered with the last attempt, new context programs are
sometimes queued from interrupt context, making it unacceptable to call
tasklet_disable() from context_get_descriptors().
This version of the patch uses ohci->lock for all locking needs instead
of tasklet_disable/enable. There is a new requirement that
context_get_descriptors() be called while holding ohci->lock. It was
already held for the AT context, so adding the requirement for the iso
context did not seem particularly onerous. In addition, this has the
side benefit of allowing iso queue to be safely called from concurrent
user-space threads, which previously was not safe.
Signed-off-by: David Moore <dcm@acm.org>
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
-
Fixes the following issues:
- Isochronous reception stopped prematurely if an application used a
larger buffer. (Reproduced with coriander.)
- Isochronous reception stopped after one or a few frames on VT630x
in OHCI 1.0 mode. (Fixes reception in coriander, but dvgrab still
doesn't work with these chips.)
Patch update: struct member alignment, whitespace nits
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2008-01-07 05:21:41 +07:00
|
|
|
struct descriptor_buffer *old_desc = desc;
|
2007-02-17 05:34:39 +07:00
|
|
|
address = le32_to_cpu(last->branch_address);
|
|
|
|
z = address & 0xf;
|
firewire: fw-ohci: Dynamically allocate buffers for DMA descriptors
Previously, the fw-ohci driver used fixed-length buffers for storing
descriptors for isochronous receive DMA programs. If an application
(such as libdc1394) generated a DMA program that was too large, fw-ohci
would reach the limit of its fixed-sized buffer and return an error to
userspace.
This patch replaces the fixed-length ring-buffer with a linked-list of
page-sized buffers. Additional buffers can be dynamically allocated and
appended to the list when necessary. For a particular context, buffers
are kept around after use and reused as necessary, so there is no
allocation taking place after the DMA program is generated for the first
time.
In addition, the buffers it uses are coherent for DMA so there is no
syncing required before and after writes. This syncing wasn't properly
done in the previous version of the code.
-
This is the fourth version of my patch that replaces a fixed-length
buffer for DMA descriptors with a dynamically allocated linked-list of
buffers.
As we discovered with the last attempt, new context programs are
sometimes queued from interrupt context, making it unacceptable to call
tasklet_disable() from context_get_descriptors().
This version of the patch uses ohci->lock for all locking needs instead
of tasklet_disable/enable. There is a new requirement that
context_get_descriptors() be called while holding ohci->lock. It was
already held for the AT context, so adding the requirement for the iso
context did not seem particularly onerous. In addition, this has the
side benefit of allowing iso queue to be safely called from concurrent
user-space threads, which previously was not safe.
Signed-off-by: David Moore <dcm@acm.org>
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
-
Fixes the following issues:
- Isochronous reception stopped prematurely if an application used a
larger buffer. (Reproduced with coriander.)
- Isochronous reception stopped after one or a few frames on VT630x
in OHCI 1.0 mode. (Fixes reception in coriander, but dvgrab still
doesn't work with these chips.)
Patch update: struct member alignment, whitespace nits
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2008-01-07 05:21:41 +07:00
|
|
|
address &= ~0xf;
|
|
|
|
|
|
|
|
/* If the branch address points to a buffer outside of the
|
|
|
|
* current buffer, advance to the next buffer. */
|
|
|
|
if (address < desc->buffer_bus ||
|
|
|
|
address >= desc->buffer_bus + desc->used)
|
|
|
|
desc = list_entry(desc->list.next,
|
|
|
|
struct descriptor_buffer, list);
|
|
|
|
d = desc->buffer + (address - desc->buffer_bus) / sizeof(*d);
|
firewire: OHCI 1.0 Isochronous Receive support
Third rendition of FireWire OHCI 1.0 Isochronous Receive support, using a
zer-copy method similar to OHCI 1.1 which puts the IR data payload directly
into the userspace buffer. The zero-copy implementation eliminates the
video artifacts, audio popping, and buffer underrun problems seen with
version 1 of this patch, as well as fixing a regression in OHCI 1.1 support
introduced by version 2 of this patch.
Successfully tested in OHCI 1.1 mode on the following chipsets:
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
- Ti TSB41AB2 (rev 01), OHCI 1.1 (PCI on SB Audigy)
- Apple UniNorth 2 (rev 81), OHCI 1.1 (PowerBook G4 onboard)
Successfully tested in OHCI 1.0 mode on the following chipsets:
- Agere FW323 (rev 06), OHCI 1.0 (Mac Mini onboard)
- Agere FW323 (rev 06), OHCI 1.0 (PCI)
- Via VT6306 (rev 46), OHCI 1.0 (PCI)
- NEC OrangeLink (rev 01), OHCI 1.0 (PCI)
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
The bulk of testing was done in an x86_64 system, but was also successfully
sanity-tested on other systems, including a PPC(32) PowerBook G4 and an i686
EPIA M10k. Crude benchmarking (watching top during capture) puts the cpu
utilization during capture on the EPIA's 1GHz Via C3 processor around 13%,
which is down from 30% with the v1 code.
Some implementation details:
To maintain the same userspace API as dual-buffer mode, we set up two
descriptors for every incoming packet. The first is an INPUT_MORE descriptor,
pointing to a buffer large enough to hold just the packet's iso headers,
immediately followed by an INPUT_LAST descriptor, pointing to a chunk of the
userspace buffer big enough for the packet's data payload. With this setup,
each incoming packet fills in these two descriptors in a manner that very
closely emulates dual-buffer receive, to the point where the bulk of the
handle_ir_* code is now identical between the two (and probably primed for
some restructuring to share code between them).
The only caveat I have at the moment is that neither of my OHCI 1.0 Via
VT6307-based FireWire controllers work particularly well with this code
for reasons I have yet to figure out.
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2007-12-04 01:43:12 +07:00
|
|
|
last = find_branch_descriptor(d, z);
|
2007-02-17 05:34:39 +07:00
|
|
|
|
|
|
|
if (!ctx->callback(ctx, d, last))
|
|
|
|
break;
|
|
|
|
|
firewire: fw-ohci: Dynamically allocate buffers for DMA descriptors
Previously, the fw-ohci driver used fixed-length buffers for storing
descriptors for isochronous receive DMA programs. If an application
(such as libdc1394) generated a DMA program that was too large, fw-ohci
would reach the limit of its fixed-sized buffer and return an error to
userspace.
This patch replaces the fixed-length ring-buffer with a linked-list of
page-sized buffers. Additional buffers can be dynamically allocated and
appended to the list when necessary. For a particular context, buffers
are kept around after use and reused as necessary, so there is no
allocation taking place after the DMA program is generated for the first
time.
In addition, the buffers it uses are coherent for DMA so there is no
syncing required before and after writes. This syncing wasn't properly
done in the previous version of the code.
-
This is the fourth version of my patch that replaces a fixed-length
buffer for DMA descriptors with a dynamically allocated linked-list of
buffers.
As we discovered with the last attempt, new context programs are
sometimes queued from interrupt context, making it unacceptable to call
tasklet_disable() from context_get_descriptors().
This version of the patch uses ohci->lock for all locking needs instead
of tasklet_disable/enable. There is a new requirement that
context_get_descriptors() be called while holding ohci->lock. It was
already held for the AT context, so adding the requirement for the iso
context did not seem particularly onerous. In addition, this has the
side benefit of allowing iso queue to be safely called from concurrent
user-space threads, which previously was not safe.
Signed-off-by: David Moore <dcm@acm.org>
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
-
Fixes the following issues:
- Isochronous reception stopped prematurely if an application used a
larger buffer. (Reproduced with coriander.)
- Isochronous reception stopped after one or a few frames on VT630x
in OHCI 1.0 mode. (Fixes reception in coriander, but dvgrab still
doesn't work with these chips.)
Patch update: struct member alignment, whitespace nits
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2008-01-07 05:21:41 +07:00
|
|
|
if (old_desc != desc) {
|
|
|
|
/* If we've advanced to the next buffer, move the
|
|
|
|
* previous buffer to the free list. */
|
|
|
|
unsigned long flags;
|
|
|
|
old_desc->used = 0;
|
|
|
|
spin_lock_irqsave(&ctx->ohci->lock, flags);
|
|
|
|
list_move_tail(&old_desc->list, &ctx->buffer_list);
|
|
|
|
spin_unlock_irqrestore(&ctx->ohci->lock, flags);
|
|
|
|
}
|
|
|
|
ctx->last = last;
|
2007-02-17 05:34:39 +07:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
firewire: fw-ohci: Dynamically allocate buffers for DMA descriptors
Previously, the fw-ohci driver used fixed-length buffers for storing
descriptors for isochronous receive DMA programs. If an application
(such as libdc1394) generated a DMA program that was too large, fw-ohci
would reach the limit of its fixed-sized buffer and return an error to
userspace.
This patch replaces the fixed-length ring-buffer with a linked-list of
page-sized buffers. Additional buffers can be dynamically allocated and
appended to the list when necessary. For a particular context, buffers
are kept around after use and reused as necessary, so there is no
allocation taking place after the DMA program is generated for the first
time.
In addition, the buffers it uses are coherent for DMA so there is no
syncing required before and after writes. This syncing wasn't properly
done in the previous version of the code.
-
This is the fourth version of my patch that replaces a fixed-length
buffer for DMA descriptors with a dynamically allocated linked-list of
buffers.
As we discovered with the last attempt, new context programs are
sometimes queued from interrupt context, making it unacceptable to call
tasklet_disable() from context_get_descriptors().
This version of the patch uses ohci->lock for all locking needs instead
of tasklet_disable/enable. There is a new requirement that
context_get_descriptors() be called while holding ohci->lock. It was
already held for the AT context, so adding the requirement for the iso
context did not seem particularly onerous. In addition, this has the
side benefit of allowing iso queue to be safely called from concurrent
user-space threads, which previously was not safe.
Signed-off-by: David Moore <dcm@acm.org>
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
-
Fixes the following issues:
- Isochronous reception stopped prematurely if an application used a
larger buffer. (Reproduced with coriander.)
- Isochronous reception stopped after one or a few frames on VT630x
in OHCI 1.0 mode. (Fixes reception in coriander, but dvgrab still
doesn't work with these chips.)
Patch update: struct member alignment, whitespace nits
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2008-01-07 05:21:41 +07:00
|
|
|
/*
|
|
|
|
* Allocate a new buffer and add it to the list of free buffers for this
|
|
|
|
* context. Must be called with ohci->lock held.
|
|
|
|
*/
|
|
|
|
static int
|
|
|
|
context_add_buffer(struct context *ctx)
|
|
|
|
{
|
|
|
|
struct descriptor_buffer *desc;
|
2008-03-14 06:27:49 +07:00
|
|
|
dma_addr_t uninitialized_var(bus_addr);
|
firewire: fw-ohci: Dynamically allocate buffers for DMA descriptors
Previously, the fw-ohci driver used fixed-length buffers for storing
descriptors for isochronous receive DMA programs. If an application
(such as libdc1394) generated a DMA program that was too large, fw-ohci
would reach the limit of its fixed-sized buffer and return an error to
userspace.
This patch replaces the fixed-length ring-buffer with a linked-list of
page-sized buffers. Additional buffers can be dynamically allocated and
appended to the list when necessary. For a particular context, buffers
are kept around after use and reused as necessary, so there is no
allocation taking place after the DMA program is generated for the first
time.
In addition, the buffers it uses are coherent for DMA so there is no
syncing required before and after writes. This syncing wasn't properly
done in the previous version of the code.
-
This is the fourth version of my patch that replaces a fixed-length
buffer for DMA descriptors with a dynamically allocated linked-list of
buffers.
As we discovered with the last attempt, new context programs are
sometimes queued from interrupt context, making it unacceptable to call
tasklet_disable() from context_get_descriptors().
This version of the patch uses ohci->lock for all locking needs instead
of tasklet_disable/enable. There is a new requirement that
context_get_descriptors() be called while holding ohci->lock. It was
already held for the AT context, so adding the requirement for the iso
context did not seem particularly onerous. In addition, this has the
side benefit of allowing iso queue to be safely called from concurrent
user-space threads, which previously was not safe.
Signed-off-by: David Moore <dcm@acm.org>
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
-
Fixes the following issues:
- Isochronous reception stopped prematurely if an application used a
larger buffer. (Reproduced with coriander.)
- Isochronous reception stopped after one or a few frames on VT630x
in OHCI 1.0 mode. (Fixes reception in coriander, but dvgrab still
doesn't work with these chips.)
Patch update: struct member alignment, whitespace nits
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2008-01-07 05:21:41 +07:00
|
|
|
int offset;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* 16MB of descriptors should be far more than enough for any DMA
|
|
|
|
* program. This will catch run-away userspace or DoS attacks.
|
|
|
|
*/
|
|
|
|
if (ctx->total_allocation >= 16*1024*1024)
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
desc = dma_alloc_coherent(ctx->ohci->card.device, PAGE_SIZE,
|
|
|
|
&bus_addr, GFP_ATOMIC);
|
|
|
|
if (!desc)
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
offset = (void *)&desc->buffer - (void *)desc;
|
|
|
|
desc->buffer_size = PAGE_SIZE - offset;
|
|
|
|
desc->buffer_bus = bus_addr + offset;
|
|
|
|
desc->used = 0;
|
|
|
|
|
|
|
|
list_add_tail(&desc->list, &ctx->buffer_list);
|
|
|
|
ctx->total_allocation += PAGE_SIZE;
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2007-02-17 05:34:39 +07:00
|
|
|
static int
|
|
|
|
context_init(struct context *ctx, struct fw_ohci *ohci,
|
firewire: fw-ohci: Dynamically allocate buffers for DMA descriptors
Previously, the fw-ohci driver used fixed-length buffers for storing
descriptors for isochronous receive DMA programs. If an application
(such as libdc1394) generated a DMA program that was too large, fw-ohci
would reach the limit of its fixed-sized buffer and return an error to
userspace.
This patch replaces the fixed-length ring-buffer with a linked-list of
page-sized buffers. Additional buffers can be dynamically allocated and
appended to the list when necessary. For a particular context, buffers
are kept around after use and reused as necessary, so there is no
allocation taking place after the DMA program is generated for the first
time.
In addition, the buffers it uses are coherent for DMA so there is no
syncing required before and after writes. This syncing wasn't properly
done in the previous version of the code.
-
This is the fourth version of my patch that replaces a fixed-length
buffer for DMA descriptors with a dynamically allocated linked-list of
buffers.
As we discovered with the last attempt, new context programs are
sometimes queued from interrupt context, making it unacceptable to call
tasklet_disable() from context_get_descriptors().
This version of the patch uses ohci->lock for all locking needs instead
of tasklet_disable/enable. There is a new requirement that
context_get_descriptors() be called while holding ohci->lock. It was
already held for the AT context, so adding the requirement for the iso
context did not seem particularly onerous. In addition, this has the
side benefit of allowing iso queue to be safely called from concurrent
user-space threads, which previously was not safe.
Signed-off-by: David Moore <dcm@acm.org>
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
-
Fixes the following issues:
- Isochronous reception stopped prematurely if an application used a
larger buffer. (Reproduced with coriander.)
- Isochronous reception stopped after one or a few frames on VT630x
in OHCI 1.0 mode. (Fixes reception in coriander, but dvgrab still
doesn't work with these chips.)
Patch update: struct member alignment, whitespace nits
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2008-01-07 05:21:41 +07:00
|
|
|
u32 regs, descriptor_callback_t callback)
|
2007-02-17 05:34:39 +07:00
|
|
|
{
|
|
|
|
ctx->ohci = ohci;
|
|
|
|
ctx->regs = regs;
|
firewire: fw-ohci: Dynamically allocate buffers for DMA descriptors
Previously, the fw-ohci driver used fixed-length buffers for storing
descriptors for isochronous receive DMA programs. If an application
(such as libdc1394) generated a DMA program that was too large, fw-ohci
would reach the limit of its fixed-sized buffer and return an error to
userspace.
This patch replaces the fixed-length ring-buffer with a linked-list of
page-sized buffers. Additional buffers can be dynamically allocated and
appended to the list when necessary. For a particular context, buffers
are kept around after use and reused as necessary, so there is no
allocation taking place after the DMA program is generated for the first
time.
In addition, the buffers it uses are coherent for DMA so there is no
syncing required before and after writes. This syncing wasn't properly
done in the previous version of the code.
-
This is the fourth version of my patch that replaces a fixed-length
buffer for DMA descriptors with a dynamically allocated linked-list of
buffers.
As we discovered with the last attempt, new context programs are
sometimes queued from interrupt context, making it unacceptable to call
tasklet_disable() from context_get_descriptors().
This version of the patch uses ohci->lock for all locking needs instead
of tasklet_disable/enable. There is a new requirement that
context_get_descriptors() be called while holding ohci->lock. It was
already held for the AT context, so adding the requirement for the iso
context did not seem particularly onerous. In addition, this has the
side benefit of allowing iso queue to be safely called from concurrent
user-space threads, which previously was not safe.
Signed-off-by: David Moore <dcm@acm.org>
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
-
Fixes the following issues:
- Isochronous reception stopped prematurely if an application used a
larger buffer. (Reproduced with coriander.)
- Isochronous reception stopped after one or a few frames on VT630x
in OHCI 1.0 mode. (Fixes reception in coriander, but dvgrab still
doesn't work with these chips.)
Patch update: struct member alignment, whitespace nits
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2008-01-07 05:21:41 +07:00
|
|
|
ctx->total_allocation = 0;
|
|
|
|
|
|
|
|
INIT_LIST_HEAD(&ctx->buffer_list);
|
|
|
|
if (context_add_buffer(ctx) < 0)
|
2007-02-17 05:34:39 +07:00
|
|
|
return -ENOMEM;
|
|
|
|
|
firewire: fw-ohci: Dynamically allocate buffers for DMA descriptors
Previously, the fw-ohci driver used fixed-length buffers for storing
descriptors for isochronous receive DMA programs. If an application
(such as libdc1394) generated a DMA program that was too large, fw-ohci
would reach the limit of its fixed-sized buffer and return an error to
userspace.
This patch replaces the fixed-length ring-buffer with a linked-list of
page-sized buffers. Additional buffers can be dynamically allocated and
appended to the list when necessary. For a particular context, buffers
are kept around after use and reused as necessary, so there is no
allocation taking place after the DMA program is generated for the first
time.
In addition, the buffers it uses are coherent for DMA so there is no
syncing required before and after writes. This syncing wasn't properly
done in the previous version of the code.
-
This is the fourth version of my patch that replaces a fixed-length
buffer for DMA descriptors with a dynamically allocated linked-list of
buffers.
As we discovered with the last attempt, new context programs are
sometimes queued from interrupt context, making it unacceptable to call
tasklet_disable() from context_get_descriptors().
This version of the patch uses ohci->lock for all locking needs instead
of tasklet_disable/enable. There is a new requirement that
context_get_descriptors() be called while holding ohci->lock. It was
already held for the AT context, so adding the requirement for the iso
context did not seem particularly onerous. In addition, this has the
side benefit of allowing iso queue to be safely called from concurrent
user-space threads, which previously was not safe.
Signed-off-by: David Moore <dcm@acm.org>
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
-
Fixes the following issues:
- Isochronous reception stopped prematurely if an application used a
larger buffer. (Reproduced with coriander.)
- Isochronous reception stopped after one or a few frames on VT630x
in OHCI 1.0 mode. (Fixes reception in coriander, but dvgrab still
doesn't work with these chips.)
Patch update: struct member alignment, whitespace nits
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2008-01-07 05:21:41 +07:00
|
|
|
ctx->buffer_tail = list_entry(ctx->buffer_list.next,
|
|
|
|
struct descriptor_buffer, list);
|
|
|
|
|
2007-02-17 05:34:39 +07:00
|
|
|
tasklet_init(&ctx->tasklet, context_tasklet, (unsigned long)ctx);
|
|
|
|
ctx->callback = callback;
|
|
|
|
|
2007-05-08 07:33:32 +07:00
|
|
|
/*
|
|
|
|
* We put a dummy descriptor in the buffer that has a NULL
|
2007-02-17 05:34:39 +07:00
|
|
|
* branch address and looks like it's been sent. That way we
|
firewire: fw-ohci: Dynamically allocate buffers for DMA descriptors
Previously, the fw-ohci driver used fixed-length buffers for storing
descriptors for isochronous receive DMA programs. If an application
(such as libdc1394) generated a DMA program that was too large, fw-ohci
would reach the limit of its fixed-sized buffer and return an error to
userspace.
This patch replaces the fixed-length ring-buffer with a linked-list of
page-sized buffers. Additional buffers can be dynamically allocated and
appended to the list when necessary. For a particular context, buffers
are kept around after use and reused as necessary, so there is no
allocation taking place after the DMA program is generated for the first
time.
In addition, the buffers it uses are coherent for DMA so there is no
syncing required before and after writes. This syncing wasn't properly
done in the previous version of the code.
-
This is the fourth version of my patch that replaces a fixed-length
buffer for DMA descriptors with a dynamically allocated linked-list of
buffers.
As we discovered with the last attempt, new context programs are
sometimes queued from interrupt context, making it unacceptable to call
tasklet_disable() from context_get_descriptors().
This version of the patch uses ohci->lock for all locking needs instead
of tasklet_disable/enable. There is a new requirement that
context_get_descriptors() be called while holding ohci->lock. It was
already held for the AT context, so adding the requirement for the iso
context did not seem particularly onerous. In addition, this has the
side benefit of allowing iso queue to be safely called from concurrent
user-space threads, which previously was not safe.
Signed-off-by: David Moore <dcm@acm.org>
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
-
Fixes the following issues:
- Isochronous reception stopped prematurely if an application used a
larger buffer. (Reproduced with coriander.)
- Isochronous reception stopped after one or a few frames on VT630x
in OHCI 1.0 mode. (Fixes reception in coriander, but dvgrab still
doesn't work with these chips.)
Patch update: struct member alignment, whitespace nits
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2008-01-07 05:21:41 +07:00
|
|
|
* have a descriptor to append DMA programs to.
|
2007-05-08 07:33:32 +07:00
|
|
|
*/
|
firewire: fw-ohci: Dynamically allocate buffers for DMA descriptors
Previously, the fw-ohci driver used fixed-length buffers for storing
descriptors for isochronous receive DMA programs. If an application
(such as libdc1394) generated a DMA program that was too large, fw-ohci
would reach the limit of its fixed-sized buffer and return an error to
userspace.
This patch replaces the fixed-length ring-buffer with a linked-list of
page-sized buffers. Additional buffers can be dynamically allocated and
appended to the list when necessary. For a particular context, buffers
are kept around after use and reused as necessary, so there is no
allocation taking place after the DMA program is generated for the first
time.
In addition, the buffers it uses are coherent for DMA so there is no
syncing required before and after writes. This syncing wasn't properly
done in the previous version of the code.
-
This is the fourth version of my patch that replaces a fixed-length
buffer for DMA descriptors with a dynamically allocated linked-list of
buffers.
As we discovered with the last attempt, new context programs are
sometimes queued from interrupt context, making it unacceptable to call
tasklet_disable() from context_get_descriptors().
This version of the patch uses ohci->lock for all locking needs instead
of tasklet_disable/enable. There is a new requirement that
context_get_descriptors() be called while holding ohci->lock. It was
already held for the AT context, so adding the requirement for the iso
context did not seem particularly onerous. In addition, this has the
side benefit of allowing iso queue to be safely called from concurrent
user-space threads, which previously was not safe.
Signed-off-by: David Moore <dcm@acm.org>
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
-
Fixes the following issues:
- Isochronous reception stopped prematurely if an application used a
larger buffer. (Reproduced with coriander.)
- Isochronous reception stopped after one or a few frames on VT630x
in OHCI 1.0 mode. (Fixes reception in coriander, but dvgrab still
doesn't work with these chips.)
Patch update: struct member alignment, whitespace nits
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2008-01-07 05:21:41 +07:00
|
|
|
memset(ctx->buffer_tail->buffer, 0, sizeof(*ctx->buffer_tail->buffer));
|
|
|
|
ctx->buffer_tail->buffer->control = cpu_to_le16(DESCRIPTOR_OUTPUT_LAST);
|
|
|
|
ctx->buffer_tail->buffer->transfer_status = cpu_to_le16(0x8011);
|
|
|
|
ctx->buffer_tail->used += sizeof(*ctx->buffer_tail->buffer);
|
|
|
|
ctx->last = ctx->buffer_tail->buffer;
|
|
|
|
ctx->prev = ctx->buffer_tail->buffer;
|
2007-02-17 05:34:39 +07:00
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2007-02-17 05:34:44 +07:00
|
|
|
static void
|
2007-02-17 05:34:39 +07:00
|
|
|
context_release(struct context *ctx)
|
|
|
|
{
|
|
|
|
struct fw_card *card = &ctx->ohci->card;
|
firewire: fw-ohci: Dynamically allocate buffers for DMA descriptors
Previously, the fw-ohci driver used fixed-length buffers for storing
descriptors for isochronous receive DMA programs. If an application
(such as libdc1394) generated a DMA program that was too large, fw-ohci
would reach the limit of its fixed-sized buffer and return an error to
userspace.
This patch replaces the fixed-length ring-buffer with a linked-list of
page-sized buffers. Additional buffers can be dynamically allocated and
appended to the list when necessary. For a particular context, buffers
are kept around after use and reused as necessary, so there is no
allocation taking place after the DMA program is generated for the first
time.
In addition, the buffers it uses are coherent for DMA so there is no
syncing required before and after writes. This syncing wasn't properly
done in the previous version of the code.
-
This is the fourth version of my patch that replaces a fixed-length
buffer for DMA descriptors with a dynamically allocated linked-list of
buffers.
As we discovered with the last attempt, new context programs are
sometimes queued from interrupt context, making it unacceptable to call
tasklet_disable() from context_get_descriptors().
This version of the patch uses ohci->lock for all locking needs instead
of tasklet_disable/enable. There is a new requirement that
context_get_descriptors() be called while holding ohci->lock. It was
already held for the AT context, so adding the requirement for the iso
context did not seem particularly onerous. In addition, this has the
side benefit of allowing iso queue to be safely called from concurrent
user-space threads, which previously was not safe.
Signed-off-by: David Moore <dcm@acm.org>
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
-
Fixes the following issues:
- Isochronous reception stopped prematurely if an application used a
larger buffer. (Reproduced with coriander.)
- Isochronous reception stopped after one or a few frames on VT630x
in OHCI 1.0 mode. (Fixes reception in coriander, but dvgrab still
doesn't work with these chips.)
Patch update: struct member alignment, whitespace nits
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2008-01-07 05:21:41 +07:00
|
|
|
struct descriptor_buffer *desc, *tmp;
|
2007-02-17 05:34:39 +07:00
|
|
|
|
firewire: fw-ohci: Dynamically allocate buffers for DMA descriptors
Previously, the fw-ohci driver used fixed-length buffers for storing
descriptors for isochronous receive DMA programs. If an application
(such as libdc1394) generated a DMA program that was too large, fw-ohci
would reach the limit of its fixed-sized buffer and return an error to
userspace.
This patch replaces the fixed-length ring-buffer with a linked-list of
page-sized buffers. Additional buffers can be dynamically allocated and
appended to the list when necessary. For a particular context, buffers
are kept around after use and reused as necessary, so there is no
allocation taking place after the DMA program is generated for the first
time.
In addition, the buffers it uses are coherent for DMA so there is no
syncing required before and after writes. This syncing wasn't properly
done in the previous version of the code.
-
This is the fourth version of my patch that replaces a fixed-length
buffer for DMA descriptors with a dynamically allocated linked-list of
buffers.
As we discovered with the last attempt, new context programs are
sometimes queued from interrupt context, making it unacceptable to call
tasklet_disable() from context_get_descriptors().
This version of the patch uses ohci->lock for all locking needs instead
of tasklet_disable/enable. There is a new requirement that
context_get_descriptors() be called while holding ohci->lock. It was
already held for the AT context, so adding the requirement for the iso
context did not seem particularly onerous. In addition, this has the
side benefit of allowing iso queue to be safely called from concurrent
user-space threads, which previously was not safe.
Signed-off-by: David Moore <dcm@acm.org>
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
-
Fixes the following issues:
- Isochronous reception stopped prematurely if an application used a
larger buffer. (Reproduced with coriander.)
- Isochronous reception stopped after one or a few frames on VT630x
in OHCI 1.0 mode. (Fixes reception in coriander, but dvgrab still
doesn't work with these chips.)
Patch update: struct member alignment, whitespace nits
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2008-01-07 05:21:41 +07:00
|
|
|
list_for_each_entry_safe(desc, tmp, &ctx->buffer_list, list)
|
|
|
|
dma_free_coherent(card->device, PAGE_SIZE, desc,
|
|
|
|
desc->buffer_bus -
|
|
|
|
((void *)&desc->buffer - (void *)desc));
|
2007-02-17 05:34:39 +07:00
|
|
|
}
|
|
|
|
|
firewire: fw-ohci: Dynamically allocate buffers for DMA descriptors
Previously, the fw-ohci driver used fixed-length buffers for storing
descriptors for isochronous receive DMA programs. If an application
(such as libdc1394) generated a DMA program that was too large, fw-ohci
would reach the limit of its fixed-sized buffer and return an error to
userspace.
This patch replaces the fixed-length ring-buffer with a linked-list of
page-sized buffers. Additional buffers can be dynamically allocated and
appended to the list when necessary. For a particular context, buffers
are kept around after use and reused as necessary, so there is no
allocation taking place after the DMA program is generated for the first
time.
In addition, the buffers it uses are coherent for DMA so there is no
syncing required before and after writes. This syncing wasn't properly
done in the previous version of the code.
-
This is the fourth version of my patch that replaces a fixed-length
buffer for DMA descriptors with a dynamically allocated linked-list of
buffers.
As we discovered with the last attempt, new context programs are
sometimes queued from interrupt context, making it unacceptable to call
tasklet_disable() from context_get_descriptors().
This version of the patch uses ohci->lock for all locking needs instead
of tasklet_disable/enable. There is a new requirement that
context_get_descriptors() be called while holding ohci->lock. It was
already held for the AT context, so adding the requirement for the iso
context did not seem particularly onerous. In addition, this has the
side benefit of allowing iso queue to be safely called from concurrent
user-space threads, which previously was not safe.
Signed-off-by: David Moore <dcm@acm.org>
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
-
Fixes the following issues:
- Isochronous reception stopped prematurely if an application used a
larger buffer. (Reproduced with coriander.)
- Isochronous reception stopped after one or a few frames on VT630x
in OHCI 1.0 mode. (Fixes reception in coriander, but dvgrab still
doesn't work with these chips.)
Patch update: struct member alignment, whitespace nits
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2008-01-07 05:21:41 +07:00
|
|
|
/* Must be called with ohci->lock held */
|
2007-02-17 05:34:39 +07:00
|
|
|
static struct descriptor *
|
|
|
|
context_get_descriptors(struct context *ctx, int z, dma_addr_t *d_bus)
|
|
|
|
{
|
firewire: fw-ohci: Dynamically allocate buffers for DMA descriptors
Previously, the fw-ohci driver used fixed-length buffers for storing
descriptors for isochronous receive DMA programs. If an application
(such as libdc1394) generated a DMA program that was too large, fw-ohci
would reach the limit of its fixed-sized buffer and return an error to
userspace.
This patch replaces the fixed-length ring-buffer with a linked-list of
page-sized buffers. Additional buffers can be dynamically allocated and
appended to the list when necessary. For a particular context, buffers
are kept around after use and reused as necessary, so there is no
allocation taking place after the DMA program is generated for the first
time.
In addition, the buffers it uses are coherent for DMA so there is no
syncing required before and after writes. This syncing wasn't properly
done in the previous version of the code.
-
This is the fourth version of my patch that replaces a fixed-length
buffer for DMA descriptors with a dynamically allocated linked-list of
buffers.
As we discovered with the last attempt, new context programs are
sometimes queued from interrupt context, making it unacceptable to call
tasklet_disable() from context_get_descriptors().
This version of the patch uses ohci->lock for all locking needs instead
of tasklet_disable/enable. There is a new requirement that
context_get_descriptors() be called while holding ohci->lock. It was
already held for the AT context, so adding the requirement for the iso
context did not seem particularly onerous. In addition, this has the
side benefit of allowing iso queue to be safely called from concurrent
user-space threads, which previously was not safe.
Signed-off-by: David Moore <dcm@acm.org>
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
-
Fixes the following issues:
- Isochronous reception stopped prematurely if an application used a
larger buffer. (Reproduced with coriander.)
- Isochronous reception stopped after one or a few frames on VT630x
in OHCI 1.0 mode. (Fixes reception in coriander, but dvgrab still
doesn't work with these chips.)
Patch update: struct member alignment, whitespace nits
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2008-01-07 05:21:41 +07:00
|
|
|
struct descriptor *d = NULL;
|
|
|
|
struct descriptor_buffer *desc = ctx->buffer_tail;
|
|
|
|
|
|
|
|
if (z * sizeof(*d) > desc->buffer_size)
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
if (z * sizeof(*d) > desc->buffer_size - desc->used) {
|
|
|
|
/* No room for the descriptor in this buffer, so advance to the
|
|
|
|
* next one. */
|
2007-02-17 05:34:39 +07:00
|
|
|
|
firewire: fw-ohci: Dynamically allocate buffers for DMA descriptors
Previously, the fw-ohci driver used fixed-length buffers for storing
descriptors for isochronous receive DMA programs. If an application
(such as libdc1394) generated a DMA program that was too large, fw-ohci
would reach the limit of its fixed-sized buffer and return an error to
userspace.
This patch replaces the fixed-length ring-buffer with a linked-list of
page-sized buffers. Additional buffers can be dynamically allocated and
appended to the list when necessary. For a particular context, buffers
are kept around after use and reused as necessary, so there is no
allocation taking place after the DMA program is generated for the first
time.
In addition, the buffers it uses are coherent for DMA so there is no
syncing required before and after writes. This syncing wasn't properly
done in the previous version of the code.
-
This is the fourth version of my patch that replaces a fixed-length
buffer for DMA descriptors with a dynamically allocated linked-list of
buffers.
As we discovered with the last attempt, new context programs are
sometimes queued from interrupt context, making it unacceptable to call
tasklet_disable() from context_get_descriptors().
This version of the patch uses ohci->lock for all locking needs instead
of tasklet_disable/enable. There is a new requirement that
context_get_descriptors() be called while holding ohci->lock. It was
already held for the AT context, so adding the requirement for the iso
context did not seem particularly onerous. In addition, this has the
side benefit of allowing iso queue to be safely called from concurrent
user-space threads, which previously was not safe.
Signed-off-by: David Moore <dcm@acm.org>
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
-
Fixes the following issues:
- Isochronous reception stopped prematurely if an application used a
larger buffer. (Reproduced with coriander.)
- Isochronous reception stopped after one or a few frames on VT630x
in OHCI 1.0 mode. (Fixes reception in coriander, but dvgrab still
doesn't work with these chips.)
Patch update: struct member alignment, whitespace nits
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2008-01-07 05:21:41 +07:00
|
|
|
if (desc->list.next == &ctx->buffer_list) {
|
|
|
|
/* If there is no free buffer next in the list,
|
|
|
|
* allocate one. */
|
|
|
|
if (context_add_buffer(ctx) < 0)
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
desc = list_entry(desc->list.next,
|
|
|
|
struct descriptor_buffer, list);
|
|
|
|
ctx->buffer_tail = desc;
|
|
|
|
}
|
2007-02-17 05:34:39 +07:00
|
|
|
|
firewire: fw-ohci: Dynamically allocate buffers for DMA descriptors
Previously, the fw-ohci driver used fixed-length buffers for storing
descriptors for isochronous receive DMA programs. If an application
(such as libdc1394) generated a DMA program that was too large, fw-ohci
would reach the limit of its fixed-sized buffer and return an error to
userspace.
This patch replaces the fixed-length ring-buffer with a linked-list of
page-sized buffers. Additional buffers can be dynamically allocated and
appended to the list when necessary. For a particular context, buffers
are kept around after use and reused as necessary, so there is no
allocation taking place after the DMA program is generated for the first
time.
In addition, the buffers it uses are coherent for DMA so there is no
syncing required before and after writes. This syncing wasn't properly
done in the previous version of the code.
-
This is the fourth version of my patch that replaces a fixed-length
buffer for DMA descriptors with a dynamically allocated linked-list of
buffers.
As we discovered with the last attempt, new context programs are
sometimes queued from interrupt context, making it unacceptable to call
tasklet_disable() from context_get_descriptors().
This version of the patch uses ohci->lock for all locking needs instead
of tasklet_disable/enable. There is a new requirement that
context_get_descriptors() be called while holding ohci->lock. It was
already held for the AT context, so adding the requirement for the iso
context did not seem particularly onerous. In addition, this has the
side benefit of allowing iso queue to be safely called from concurrent
user-space threads, which previously was not safe.
Signed-off-by: David Moore <dcm@acm.org>
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
-
Fixes the following issues:
- Isochronous reception stopped prematurely if an application used a
larger buffer. (Reproduced with coriander.)
- Isochronous reception stopped after one or a few frames on VT630x
in OHCI 1.0 mode. (Fixes reception in coriander, but dvgrab still
doesn't work with these chips.)
Patch update: struct member alignment, whitespace nits
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2008-01-07 05:21:41 +07:00
|
|
|
d = desc->buffer + desc->used / sizeof(*d);
|
2007-05-10 06:23:14 +07:00
|
|
|
memset(d, 0, z * sizeof(*d));
|
firewire: fw-ohci: Dynamically allocate buffers for DMA descriptors
Previously, the fw-ohci driver used fixed-length buffers for storing
descriptors for isochronous receive DMA programs. If an application
(such as libdc1394) generated a DMA program that was too large, fw-ohci
would reach the limit of its fixed-sized buffer and return an error to
userspace.
This patch replaces the fixed-length ring-buffer with a linked-list of
page-sized buffers. Additional buffers can be dynamically allocated and
appended to the list when necessary. For a particular context, buffers
are kept around after use and reused as necessary, so there is no
allocation taking place after the DMA program is generated for the first
time.
In addition, the buffers it uses are coherent for DMA so there is no
syncing required before and after writes. This syncing wasn't properly
done in the previous version of the code.
-
This is the fourth version of my patch that replaces a fixed-length
buffer for DMA descriptors with a dynamically allocated linked-list of
buffers.
As we discovered with the last attempt, new context programs are
sometimes queued from interrupt context, making it unacceptable to call
tasklet_disable() from context_get_descriptors().
This version of the patch uses ohci->lock for all locking needs instead
of tasklet_disable/enable. There is a new requirement that
context_get_descriptors() be called while holding ohci->lock. It was
already held for the AT context, so adding the requirement for the iso
context did not seem particularly onerous. In addition, this has the
side benefit of allowing iso queue to be safely called from concurrent
user-space threads, which previously was not safe.
Signed-off-by: David Moore <dcm@acm.org>
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
-
Fixes the following issues:
- Isochronous reception stopped prematurely if an application used a
larger buffer. (Reproduced with coriander.)
- Isochronous reception stopped after one or a few frames on VT630x
in OHCI 1.0 mode. (Fixes reception in coriander, but dvgrab still
doesn't work with these chips.)
Patch update: struct member alignment, whitespace nits
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2008-01-07 05:21:41 +07:00
|
|
|
*d_bus = desc->buffer_bus + desc->used;
|
2007-02-17 05:34:39 +07:00
|
|
|
|
|
|
|
return d;
|
|
|
|
}
|
|
|
|
|
2007-02-17 05:34:40 +07:00
|
|
|
static void context_run(struct context *ctx, u32 extra)
|
2007-02-17 05:34:39 +07:00
|
|
|
{
|
|
|
|
struct fw_ohci *ohci = ctx->ohci;
|
|
|
|
|
2007-05-08 07:33:35 +07:00
|
|
|
reg_write(ohci, COMMAND_PTR(ctx->regs),
|
firewire: fw-ohci: Dynamically allocate buffers for DMA descriptors
Previously, the fw-ohci driver used fixed-length buffers for storing
descriptors for isochronous receive DMA programs. If an application
(such as libdc1394) generated a DMA program that was too large, fw-ohci
would reach the limit of its fixed-sized buffer and return an error to
userspace.
This patch replaces the fixed-length ring-buffer with a linked-list of
page-sized buffers. Additional buffers can be dynamically allocated and
appended to the list when necessary. For a particular context, buffers
are kept around after use and reused as necessary, so there is no
allocation taking place after the DMA program is generated for the first
time.
In addition, the buffers it uses are coherent for DMA so there is no
syncing required before and after writes. This syncing wasn't properly
done in the previous version of the code.
-
This is the fourth version of my patch that replaces a fixed-length
buffer for DMA descriptors with a dynamically allocated linked-list of
buffers.
As we discovered with the last attempt, new context programs are
sometimes queued from interrupt context, making it unacceptable to call
tasklet_disable() from context_get_descriptors().
This version of the patch uses ohci->lock for all locking needs instead
of tasklet_disable/enable. There is a new requirement that
context_get_descriptors() be called while holding ohci->lock. It was
already held for the AT context, so adding the requirement for the iso
context did not seem particularly onerous. In addition, this has the
side benefit of allowing iso queue to be safely called from concurrent
user-space threads, which previously was not safe.
Signed-off-by: David Moore <dcm@acm.org>
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
-
Fixes the following issues:
- Isochronous reception stopped prematurely if an application used a
larger buffer. (Reproduced with coriander.)
- Isochronous reception stopped after one or a few frames on VT630x
in OHCI 1.0 mode. (Fixes reception in coriander, but dvgrab still
doesn't work with these chips.)
Patch update: struct member alignment, whitespace nits
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2008-01-07 05:21:41 +07:00
|
|
|
le32_to_cpu(ctx->last->branch_address));
|
2007-05-08 07:33:35 +07:00
|
|
|
reg_write(ohci, CONTROL_CLEAR(ctx->regs), ~0);
|
|
|
|
reg_write(ohci, CONTROL_SET(ctx->regs), CONTEXT_RUN | extra);
|
2007-02-17 05:34:39 +07:00
|
|
|
flush_writes(ohci);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void context_append(struct context *ctx,
|
|
|
|
struct descriptor *d, int z, int extra)
|
|
|
|
{
|
|
|
|
dma_addr_t d_bus;
|
firewire: fw-ohci: Dynamically allocate buffers for DMA descriptors
Previously, the fw-ohci driver used fixed-length buffers for storing
descriptors for isochronous receive DMA programs. If an application
(such as libdc1394) generated a DMA program that was too large, fw-ohci
would reach the limit of its fixed-sized buffer and return an error to
userspace.
This patch replaces the fixed-length ring-buffer with a linked-list of
page-sized buffers. Additional buffers can be dynamically allocated and
appended to the list when necessary. For a particular context, buffers
are kept around after use and reused as necessary, so there is no
allocation taking place after the DMA program is generated for the first
time.
In addition, the buffers it uses are coherent for DMA so there is no
syncing required before and after writes. This syncing wasn't properly
done in the previous version of the code.
-
This is the fourth version of my patch that replaces a fixed-length
buffer for DMA descriptors with a dynamically allocated linked-list of
buffers.
As we discovered with the last attempt, new context programs are
sometimes queued from interrupt context, making it unacceptable to call
tasklet_disable() from context_get_descriptors().
This version of the patch uses ohci->lock for all locking needs instead
of tasklet_disable/enable. There is a new requirement that
context_get_descriptors() be called while holding ohci->lock. It was
already held for the AT context, so adding the requirement for the iso
context did not seem particularly onerous. In addition, this has the
side benefit of allowing iso queue to be safely called from concurrent
user-space threads, which previously was not safe.
Signed-off-by: David Moore <dcm@acm.org>
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
-
Fixes the following issues:
- Isochronous reception stopped prematurely if an application used a
larger buffer. (Reproduced with coriander.)
- Isochronous reception stopped after one or a few frames on VT630x
in OHCI 1.0 mode. (Fixes reception in coriander, but dvgrab still
doesn't work with these chips.)
Patch update: struct member alignment, whitespace nits
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2008-01-07 05:21:41 +07:00
|
|
|
struct descriptor_buffer *desc = ctx->buffer_tail;
|
2007-02-17 05:34:39 +07:00
|
|
|
|
firewire: fw-ohci: Dynamically allocate buffers for DMA descriptors
Previously, the fw-ohci driver used fixed-length buffers for storing
descriptors for isochronous receive DMA programs. If an application
(such as libdc1394) generated a DMA program that was too large, fw-ohci
would reach the limit of its fixed-sized buffer and return an error to
userspace.
This patch replaces the fixed-length ring-buffer with a linked-list of
page-sized buffers. Additional buffers can be dynamically allocated and
appended to the list when necessary. For a particular context, buffers
are kept around after use and reused as necessary, so there is no
allocation taking place after the DMA program is generated for the first
time.
In addition, the buffers it uses are coherent for DMA so there is no
syncing required before and after writes. This syncing wasn't properly
done in the previous version of the code.
-
This is the fourth version of my patch that replaces a fixed-length
buffer for DMA descriptors with a dynamically allocated linked-list of
buffers.
As we discovered with the last attempt, new context programs are
sometimes queued from interrupt context, making it unacceptable to call
tasklet_disable() from context_get_descriptors().
This version of the patch uses ohci->lock for all locking needs instead
of tasklet_disable/enable. There is a new requirement that
context_get_descriptors() be called while holding ohci->lock. It was
already held for the AT context, so adding the requirement for the iso
context did not seem particularly onerous. In addition, this has the
side benefit of allowing iso queue to be safely called from concurrent
user-space threads, which previously was not safe.
Signed-off-by: David Moore <dcm@acm.org>
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
-
Fixes the following issues:
- Isochronous reception stopped prematurely if an application used a
larger buffer. (Reproduced with coriander.)
- Isochronous reception stopped after one or a few frames on VT630x
in OHCI 1.0 mode. (Fixes reception in coriander, but dvgrab still
doesn't work with these chips.)
Patch update: struct member alignment, whitespace nits
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2008-01-07 05:21:41 +07:00
|
|
|
d_bus = desc->buffer_bus + (d - desc->buffer) * sizeof(*d);
|
2007-02-17 05:34:39 +07:00
|
|
|
|
firewire: fw-ohci: Dynamically allocate buffers for DMA descriptors
Previously, the fw-ohci driver used fixed-length buffers for storing
descriptors for isochronous receive DMA programs. If an application
(such as libdc1394) generated a DMA program that was too large, fw-ohci
would reach the limit of its fixed-sized buffer and return an error to
userspace.
This patch replaces the fixed-length ring-buffer with a linked-list of
page-sized buffers. Additional buffers can be dynamically allocated and
appended to the list when necessary. For a particular context, buffers
are kept around after use and reused as necessary, so there is no
allocation taking place after the DMA program is generated for the first
time.
In addition, the buffers it uses are coherent for DMA so there is no
syncing required before and after writes. This syncing wasn't properly
done in the previous version of the code.
-
This is the fourth version of my patch that replaces a fixed-length
buffer for DMA descriptors with a dynamically allocated linked-list of
buffers.
As we discovered with the last attempt, new context programs are
sometimes queued from interrupt context, making it unacceptable to call
tasklet_disable() from context_get_descriptors().
This version of the patch uses ohci->lock for all locking needs instead
of tasklet_disable/enable. There is a new requirement that
context_get_descriptors() be called while holding ohci->lock. It was
already held for the AT context, so adding the requirement for the iso
context did not seem particularly onerous. In addition, this has the
side benefit of allowing iso queue to be safely called from concurrent
user-space threads, which previously was not safe.
Signed-off-by: David Moore <dcm@acm.org>
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
-
Fixes the following issues:
- Isochronous reception stopped prematurely if an application used a
larger buffer. (Reproduced with coriander.)
- Isochronous reception stopped after one or a few frames on VT630x
in OHCI 1.0 mode. (Fixes reception in coriander, but dvgrab still
doesn't work with these chips.)
Patch update: struct member alignment, whitespace nits
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2008-01-07 05:21:41 +07:00
|
|
|
desc->used += (z + extra) * sizeof(*d);
|
|
|
|
ctx->prev->branch_address = cpu_to_le32(d_bus | z);
|
|
|
|
ctx->prev = find_branch_descriptor(d, z);
|
2007-02-17 05:34:39 +07:00
|
|
|
|
2007-05-08 07:33:35 +07:00
|
|
|
reg_write(ctx->ohci, CONTROL_SET(ctx->regs), CONTEXT_WAKE);
|
2007-02-17 05:34:39 +07:00
|
|
|
flush_writes(ctx->ohci);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void context_stop(struct context *ctx)
|
|
|
|
{
|
|
|
|
u32 reg;
|
2007-02-17 05:34:42 +07:00
|
|
|
int i;
|
2007-02-17 05:34:39 +07:00
|
|
|
|
2007-05-08 07:33:35 +07:00
|
|
|
reg_write(ctx->ohci, CONTROL_CLEAR(ctx->regs), CONTEXT_RUN);
|
2007-02-17 05:34:42 +07:00
|
|
|
flush_writes(ctx->ohci);
|
2007-02-17 05:34:39 +07:00
|
|
|
|
2007-02-17 05:34:42 +07:00
|
|
|
for (i = 0; i < 10; i++) {
|
2007-05-08 07:33:35 +07:00
|
|
|
reg = reg_read(ctx->ohci, CONTROL_SET(ctx->regs));
|
2007-02-17 05:34:42 +07:00
|
|
|
if ((reg & CONTEXT_ACTIVE) == 0)
|
|
|
|
break;
|
|
|
|
|
|
|
|
fw_notify("context_stop: still active (0x%08x)\n", reg);
|
2007-07-13 03:25:14 +07:00
|
|
|
mdelay(1);
|
2007-02-17 05:34:42 +07:00
|
|
|
}
|
2007-02-17 05:34:39 +07:00
|
|
|
}
|
2006-12-20 07:58:35 +07:00
|
|
|
|
2007-03-08 00:12:49 +07:00
|
|
|
struct driver_data {
|
|
|
|
struct fw_packet *packet;
|
|
|
|
};
|
2006-12-20 07:58:35 +07:00
|
|
|
|
2007-05-08 07:33:32 +07:00
|
|
|
/*
|
|
|
|
* This function apppends a packet to the DMA queue for transmission.
|
2007-03-08 00:12:49 +07:00
|
|
|
* Must always be called with the ochi->lock held to ensure proper
|
2007-05-08 07:33:32 +07:00
|
|
|
* generation handling and locking around packet queue manipulation.
|
|
|
|
*/
|
2007-03-08 00:12:49 +07:00
|
|
|
static int
|
|
|
|
at_context_queue_packet(struct context *ctx, struct fw_packet *packet)
|
2006-12-20 07:58:35 +07:00
|
|
|
{
|
|
|
|
struct fw_ohci *ohci = ctx->ohci;
|
2007-10-21 16:20:07 +07:00
|
|
|
dma_addr_t d_bus, uninitialized_var(payload_bus);
|
2007-03-08 00:12:49 +07:00
|
|
|
struct driver_data *driver_data;
|
|
|
|
struct descriptor *d, *last;
|
|
|
|
__le32 *header;
|
2006-12-20 07:58:35 +07:00
|
|
|
int z, tcode;
|
2007-03-08 00:12:49 +07:00
|
|
|
u32 reg;
|
2006-12-20 07:58:35 +07:00
|
|
|
|
2007-03-08 00:12:49 +07:00
|
|
|
d = context_get_descriptors(ctx, 4, &d_bus);
|
|
|
|
if (d == NULL) {
|
|
|
|
packet->ack = RCODE_SEND_ERROR;
|
|
|
|
return -1;
|
2006-12-20 07:58:35 +07:00
|
|
|
}
|
|
|
|
|
2007-05-08 07:33:35 +07:00
|
|
|
d[0].control = cpu_to_le16(DESCRIPTOR_KEY_IMMEDIATE);
|
2007-03-08 00:12:49 +07:00
|
|
|
d[0].res_count = cpu_to_le16(packet->timestamp);
|
|
|
|
|
2007-05-08 07:33:32 +07:00
|
|
|
/*
|
|
|
|
* The DMA format for asyncronous link packets is different
|
2006-12-20 07:58:35 +07:00
|
|
|
* from the IEEE1394 layout, so shift the fields around
|
|
|
|
* accordingly. If header_length is 8, it's a PHY packet, to
|
2007-05-08 07:33:32 +07:00
|
|
|
* which we need to prepend an extra quadlet.
|
|
|
|
*/
|
2007-03-08 00:12:49 +07:00
|
|
|
|
|
|
|
header = (__le32 *) &d[1];
|
2006-12-20 07:58:35 +07:00
|
|
|
if (packet->header_length > 8) {
|
2007-03-08 00:12:49 +07:00
|
|
|
header[0] = cpu_to_le32((packet->header[0] & 0xffff) |
|
|
|
|
(packet->speed << 16));
|
|
|
|
header[1] = cpu_to_le32((packet->header[1] & 0xffff) |
|
|
|
|
(packet->header[0] & 0xffff0000));
|
|
|
|
header[2] = cpu_to_le32(packet->header[2]);
|
2006-12-20 07:58:35 +07:00
|
|
|
|
|
|
|
tcode = (packet->header[0] >> 4) & 0x0f;
|
|
|
|
if (TCODE_IS_BLOCK_PACKET(tcode))
|
2007-03-08 00:12:49 +07:00
|
|
|
header[3] = cpu_to_le32(packet->header[3]);
|
2006-12-20 07:58:35 +07:00
|
|
|
else
|
2007-03-08 00:12:49 +07:00
|
|
|
header[3] = (__force __le32) packet->header[3];
|
|
|
|
|
|
|
|
d[0].req_count = cpu_to_le16(packet->header_length);
|
2006-12-20 07:58:35 +07:00
|
|
|
} else {
|
2007-03-08 00:12:49 +07:00
|
|
|
header[0] = cpu_to_le32((OHCI1394_phy_tcode << 4) |
|
|
|
|
(packet->speed << 16));
|
|
|
|
header[1] = cpu_to_le32(packet->header[0]);
|
|
|
|
header[2] = cpu_to_le32(packet->header[1]);
|
|
|
|
d[0].req_count = cpu_to_le16(12);
|
2006-12-20 07:58:35 +07:00
|
|
|
}
|
|
|
|
|
2007-03-08 00:12:49 +07:00
|
|
|
driver_data = (struct driver_data *) &d[3];
|
|
|
|
driver_data->packet = packet;
|
2007-03-27 06:18:19 +07:00
|
|
|
packet->driver_data = driver_data;
|
firewire: OHCI 1.0 Isochronous Receive support
Third rendition of FireWire OHCI 1.0 Isochronous Receive support, using a
zer-copy method similar to OHCI 1.1 which puts the IR data payload directly
into the userspace buffer. The zero-copy implementation eliminates the
video artifacts, audio popping, and buffer underrun problems seen with
version 1 of this patch, as well as fixing a regression in OHCI 1.1 support
introduced by version 2 of this patch.
Successfully tested in OHCI 1.1 mode on the following chipsets:
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
- Ti TSB41AB2 (rev 01), OHCI 1.1 (PCI on SB Audigy)
- Apple UniNorth 2 (rev 81), OHCI 1.1 (PowerBook G4 onboard)
Successfully tested in OHCI 1.0 mode on the following chipsets:
- Agere FW323 (rev 06), OHCI 1.0 (Mac Mini onboard)
- Agere FW323 (rev 06), OHCI 1.0 (PCI)
- Via VT6306 (rev 46), OHCI 1.0 (PCI)
- NEC OrangeLink (rev 01), OHCI 1.0 (PCI)
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
The bulk of testing was done in an x86_64 system, but was also successfully
sanity-tested on other systems, including a PPC(32) PowerBook G4 and an i686
EPIA M10k. Crude benchmarking (watching top during capture) puts the cpu
utilization during capture on the EPIA's 1GHz Via C3 processor around 13%,
which is down from 30% with the v1 code.
Some implementation details:
To maintain the same userspace API as dual-buffer mode, we set up two
descriptors for every incoming packet. The first is an INPUT_MORE descriptor,
pointing to a buffer large enough to hold just the packet's iso headers,
immediately followed by an INPUT_LAST descriptor, pointing to a chunk of the
userspace buffer big enough for the packet's data payload. With this setup,
each incoming packet fills in these two descriptors in a manner that very
closely emulates dual-buffer receive, to the point where the bulk of the
handle_ir_* code is now identical between the two (and probably primed for
some restructuring to share code between them).
The only caveat I have at the moment is that neither of my OHCI 1.0 Via
VT6307-based FireWire controllers work particularly well with this code
for reasons I have yet to figure out.
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2007-12-04 01:43:12 +07:00
|
|
|
|
2007-03-08 00:12:49 +07:00
|
|
|
if (packet->payload_length > 0) {
|
|
|
|
payload_bus =
|
|
|
|
dma_map_single(ohci->card.device, packet->payload,
|
|
|
|
packet->payload_length, DMA_TO_DEVICE);
|
|
|
|
if (dma_mapping_error(payload_bus)) {
|
|
|
|
packet->ack = RCODE_SEND_ERROR;
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
|
|
|
d[2].req_count = cpu_to_le16(packet->payload_length);
|
|
|
|
d[2].data_address = cpu_to_le32(payload_bus);
|
|
|
|
last = &d[2];
|
|
|
|
z = 3;
|
2006-12-20 07:58:35 +07:00
|
|
|
} else {
|
2007-03-08 00:12:49 +07:00
|
|
|
last = &d[0];
|
|
|
|
z = 2;
|
2006-12-20 07:58:35 +07:00
|
|
|
}
|
|
|
|
|
2007-05-08 07:33:35 +07:00
|
|
|
last->control |= cpu_to_le16(DESCRIPTOR_OUTPUT_LAST |
|
|
|
|
DESCRIPTOR_IRQ_ALWAYS |
|
|
|
|
DESCRIPTOR_BRANCH_ALWAYS);
|
2006-12-20 07:58:35 +07:00
|
|
|
|
2007-03-08 00:12:49 +07:00
|
|
|
/* FIXME: Document how the locking works. */
|
|
|
|
if (ohci->generation != packet->generation) {
|
2007-08-30 00:40:28 +07:00
|
|
|
if (packet->payload_length > 0)
|
|
|
|
dma_unmap_single(ohci->card.device, payload_bus,
|
|
|
|
packet->payload_length, DMA_TO_DEVICE);
|
2007-03-08 00:12:49 +07:00
|
|
|
packet->ack = RCODE_GENERATION;
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
|
|
|
context_append(ctx, d, z, 4 - z);
|
2006-12-20 07:58:35 +07:00
|
|
|
|
2007-03-08 00:12:49 +07:00
|
|
|
/* If the context isn't already running, start it up. */
|
2007-05-08 07:33:35 +07:00
|
|
|
reg = reg_read(ctx->ohci, CONTROL_SET(ctx->regs));
|
2007-04-11 05:11:17 +07:00
|
|
|
if ((reg & CONTEXT_RUN) == 0)
|
2007-03-08 00:12:49 +07:00
|
|
|
context_run(ctx, 0);
|
|
|
|
|
|
|
|
return 0;
|
2006-12-20 07:58:35 +07:00
|
|
|
}
|
|
|
|
|
2007-03-08 00:12:49 +07:00
|
|
|
static int handle_at_packet(struct context *context,
|
|
|
|
struct descriptor *d,
|
|
|
|
struct descriptor *last)
|
2006-12-20 07:58:35 +07:00
|
|
|
{
|
2007-03-08 00:12:49 +07:00
|
|
|
struct driver_data *driver_data;
|
2006-12-20 07:58:35 +07:00
|
|
|
struct fw_packet *packet;
|
2007-03-08 00:12:49 +07:00
|
|
|
struct fw_ohci *ohci = context->ohci;
|
|
|
|
dma_addr_t payload_bus;
|
2006-12-20 07:58:35 +07:00
|
|
|
int evt;
|
|
|
|
|
2007-03-08 00:12:49 +07:00
|
|
|
if (last->transfer_status == 0)
|
|
|
|
/* This descriptor isn't done yet, stop iteration. */
|
|
|
|
return 0;
|
2006-12-20 07:58:35 +07:00
|
|
|
|
2007-03-08 00:12:49 +07:00
|
|
|
driver_data = (struct driver_data *) &d[3];
|
|
|
|
packet = driver_data->packet;
|
|
|
|
if (packet == NULL)
|
|
|
|
/* This packet was cancelled, just continue. */
|
|
|
|
return 1;
|
2007-02-07 02:49:32 +07:00
|
|
|
|
2007-03-08 00:12:49 +07:00
|
|
|
payload_bus = le32_to_cpu(last->data_address);
|
|
|
|
if (payload_bus != 0)
|
|
|
|
dma_unmap_single(ohci->card.device, payload_bus,
|
2006-12-20 07:58:35 +07:00
|
|
|
packet->payload_length, DMA_TO_DEVICE);
|
|
|
|
|
2007-03-08 00:12:49 +07:00
|
|
|
evt = le16_to_cpu(last->transfer_status) & 0x1f;
|
|
|
|
packet->timestamp = le16_to_cpu(last->res_count);
|
2006-12-20 07:58:35 +07:00
|
|
|
|
2008-03-21 04:04:36 +07:00
|
|
|
log_ar_at_event('T', packet->speed, packet->header, evt);
|
|
|
|
|
2007-03-08 00:12:49 +07:00
|
|
|
switch (evt) {
|
|
|
|
case OHCI1394_evt_timeout:
|
|
|
|
/* Async response transmit timed out. */
|
|
|
|
packet->ack = RCODE_CANCELLED;
|
|
|
|
break;
|
2006-12-20 07:58:35 +07:00
|
|
|
|
2007-03-08 00:12:49 +07:00
|
|
|
case OHCI1394_evt_flushed:
|
2007-05-08 07:33:32 +07:00
|
|
|
/*
|
|
|
|
* The packet was flushed should give same error as
|
|
|
|
* when we try to use a stale generation count.
|
|
|
|
*/
|
2007-03-08 00:12:49 +07:00
|
|
|
packet->ack = RCODE_GENERATION;
|
|
|
|
break;
|
2006-12-20 07:58:35 +07:00
|
|
|
|
2007-03-08 00:12:49 +07:00
|
|
|
case OHCI1394_evt_missing_ack:
|
2007-05-08 07:33:32 +07:00
|
|
|
/*
|
|
|
|
* Using a valid (current) generation count, but the
|
|
|
|
* node is not on the bus or not sending acks.
|
|
|
|
*/
|
2007-03-08 00:12:49 +07:00
|
|
|
packet->ack = RCODE_NO_ACK;
|
|
|
|
break;
|
2006-12-20 07:58:35 +07:00
|
|
|
|
2007-03-08 00:12:49 +07:00
|
|
|
case ACK_COMPLETE + 0x10:
|
|
|
|
case ACK_PENDING + 0x10:
|
|
|
|
case ACK_BUSY_X + 0x10:
|
|
|
|
case ACK_BUSY_A + 0x10:
|
|
|
|
case ACK_BUSY_B + 0x10:
|
|
|
|
case ACK_DATA_ERROR + 0x10:
|
|
|
|
case ACK_TYPE_ERROR + 0x10:
|
|
|
|
packet->ack = evt - 0x10;
|
|
|
|
break;
|
2006-12-20 07:58:35 +07:00
|
|
|
|
2007-03-08 00:12:49 +07:00
|
|
|
default:
|
|
|
|
packet->ack = RCODE_SEND_ERROR;
|
|
|
|
break;
|
|
|
|
}
|
2006-12-20 07:58:35 +07:00
|
|
|
|
2007-03-08 00:12:49 +07:00
|
|
|
packet->callback(packet, &ohci->card, packet->ack);
|
2006-12-20 07:58:35 +07:00
|
|
|
|
2007-03-08 00:12:49 +07:00
|
|
|
return 1;
|
2006-12-20 07:58:35 +07:00
|
|
|
}
|
|
|
|
|
2007-05-08 07:33:35 +07:00
|
|
|
#define HEADER_GET_DESTINATION(q) (((q) >> 16) & 0xffff)
|
|
|
|
#define HEADER_GET_TCODE(q) (((q) >> 4) & 0x0f)
|
|
|
|
#define HEADER_GET_OFFSET_HIGH(q) (((q) >> 0) & 0xffff)
|
|
|
|
#define HEADER_GET_DATA_LENGTH(q) (((q) >> 16) & 0xffff)
|
|
|
|
#define HEADER_GET_EXTENDED_TCODE(q) (((q) >> 0) & 0xffff)
|
2007-01-26 12:38:26 +07:00
|
|
|
|
|
|
|
static void
|
|
|
|
handle_local_rom(struct fw_ohci *ohci, struct fw_packet *packet, u32 csr)
|
|
|
|
{
|
|
|
|
struct fw_packet response;
|
|
|
|
int tcode, length, i;
|
|
|
|
|
2007-05-08 07:33:35 +07:00
|
|
|
tcode = HEADER_GET_TCODE(packet->header[0]);
|
2007-01-26 12:38:26 +07:00
|
|
|
if (TCODE_IS_BLOCK_PACKET(tcode))
|
2007-05-08 07:33:35 +07:00
|
|
|
length = HEADER_GET_DATA_LENGTH(packet->header[3]);
|
2007-01-26 12:38:26 +07:00
|
|
|
else
|
|
|
|
length = 4;
|
|
|
|
|
|
|
|
i = csr - CSR_CONFIG_ROM;
|
|
|
|
if (i + length > CONFIG_ROM_SIZE) {
|
|
|
|
fw_fill_response(&response, packet->header,
|
|
|
|
RCODE_ADDRESS_ERROR, NULL, 0);
|
|
|
|
} else if (!TCODE_IS_READ_REQUEST(tcode)) {
|
|
|
|
fw_fill_response(&response, packet->header,
|
|
|
|
RCODE_TYPE_ERROR, NULL, 0);
|
|
|
|
} else {
|
|
|
|
fw_fill_response(&response, packet->header, RCODE_COMPLETE,
|
|
|
|
(void *) ohci->config_rom + i, length);
|
|
|
|
}
|
|
|
|
|
|
|
|
fw_core_handle_response(&ohci->card, &response);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
handle_local_lock(struct fw_ohci *ohci, struct fw_packet *packet, u32 csr)
|
|
|
|
{
|
|
|
|
struct fw_packet response;
|
|
|
|
int tcode, length, ext_tcode, sel;
|
|
|
|
__be32 *payload, lock_old;
|
|
|
|
u32 lock_arg, lock_data;
|
|
|
|
|
2007-05-08 07:33:35 +07:00
|
|
|
tcode = HEADER_GET_TCODE(packet->header[0]);
|
|
|
|
length = HEADER_GET_DATA_LENGTH(packet->header[3]);
|
2007-01-26 12:38:26 +07:00
|
|
|
payload = packet->payload;
|
2007-05-08 07:33:35 +07:00
|
|
|
ext_tcode = HEADER_GET_EXTENDED_TCODE(packet->header[3]);
|
2007-01-26 12:38:26 +07:00
|
|
|
|
|
|
|
if (tcode == TCODE_LOCK_REQUEST &&
|
|
|
|
ext_tcode == EXTCODE_COMPARE_SWAP && length == 8) {
|
|
|
|
lock_arg = be32_to_cpu(payload[0]);
|
|
|
|
lock_data = be32_to_cpu(payload[1]);
|
|
|
|
} else if (tcode == TCODE_READ_QUADLET_REQUEST) {
|
|
|
|
lock_arg = 0;
|
|
|
|
lock_data = 0;
|
|
|
|
} else {
|
|
|
|
fw_fill_response(&response, packet->header,
|
|
|
|
RCODE_TYPE_ERROR, NULL, 0);
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
|
|
|
sel = (csr - CSR_BUS_MANAGER_ID) / 4;
|
|
|
|
reg_write(ohci, OHCI1394_CSRData, lock_data);
|
|
|
|
reg_write(ohci, OHCI1394_CSRCompareData, lock_arg);
|
|
|
|
reg_write(ohci, OHCI1394_CSRControl, sel);
|
|
|
|
|
|
|
|
if (reg_read(ohci, OHCI1394_CSRControl) & 0x80000000)
|
|
|
|
lock_old = cpu_to_be32(reg_read(ohci, OHCI1394_CSRData));
|
|
|
|
else
|
|
|
|
fw_notify("swap not done yet\n");
|
|
|
|
|
|
|
|
fw_fill_response(&response, packet->header,
|
2007-05-10 06:23:14 +07:00
|
|
|
RCODE_COMPLETE, &lock_old, sizeof(lock_old));
|
2007-01-26 12:38:26 +07:00
|
|
|
out:
|
|
|
|
fw_core_handle_response(&ohci->card, &response);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
2007-03-08 00:12:49 +07:00
|
|
|
handle_local_request(struct context *ctx, struct fw_packet *packet)
|
2007-01-26 12:38:26 +07:00
|
|
|
{
|
|
|
|
u64 offset;
|
|
|
|
u32 csr;
|
|
|
|
|
2007-03-08 00:12:55 +07:00
|
|
|
if (ctx == &ctx->ohci->at_request_ctx) {
|
|
|
|
packet->ack = ACK_PENDING;
|
|
|
|
packet->callback(packet, &ctx->ohci->card, packet->ack);
|
|
|
|
}
|
2007-01-26 12:38:26 +07:00
|
|
|
|
|
|
|
offset =
|
|
|
|
((unsigned long long)
|
2007-05-08 07:33:35 +07:00
|
|
|
HEADER_GET_OFFSET_HIGH(packet->header[1]) << 32) |
|
2007-01-26 12:38:26 +07:00
|
|
|
packet->header[2];
|
|
|
|
csr = offset - CSR_REGISTER_BASE;
|
|
|
|
|
|
|
|
/* Handle config rom reads. */
|
|
|
|
if (csr >= CSR_CONFIG_ROM && csr < CSR_CONFIG_ROM_END)
|
|
|
|
handle_local_rom(ctx->ohci, packet, csr);
|
|
|
|
else switch (csr) {
|
|
|
|
case CSR_BUS_MANAGER_ID:
|
|
|
|
case CSR_BANDWIDTH_AVAILABLE:
|
|
|
|
case CSR_CHANNELS_AVAILABLE_HI:
|
|
|
|
case CSR_CHANNELS_AVAILABLE_LO:
|
|
|
|
handle_local_lock(ctx->ohci, packet, csr);
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
if (ctx == &ctx->ohci->at_request_ctx)
|
|
|
|
fw_core_handle_request(&ctx->ohci->card, packet);
|
|
|
|
else
|
|
|
|
fw_core_handle_response(&ctx->ohci->card, packet);
|
|
|
|
break;
|
|
|
|
}
|
2007-03-08 00:12:55 +07:00
|
|
|
|
|
|
|
if (ctx == &ctx->ohci->at_response_ctx) {
|
|
|
|
packet->ack = ACK_COMPLETE;
|
|
|
|
packet->callback(packet, &ctx->ohci->card, packet->ack);
|
|
|
|
}
|
2007-01-26 12:38:26 +07:00
|
|
|
}
|
2007-01-26 12:38:04 +07:00
|
|
|
|
2006-12-20 07:58:35 +07:00
|
|
|
static void
|
2007-03-08 00:12:49 +07:00
|
|
|
at_context_transmit(struct context *ctx, struct fw_packet *packet)
|
2006-12-20 07:58:35 +07:00
|
|
|
{
|
|
|
|
unsigned long flags;
|
2007-03-08 00:12:49 +07:00
|
|
|
int retval;
|
2006-12-20 07:58:35 +07:00
|
|
|
|
|
|
|
spin_lock_irqsave(&ctx->ohci->lock, flags);
|
|
|
|
|
2007-05-08 07:33:35 +07:00
|
|
|
if (HEADER_GET_DESTINATION(packet->header[0]) == ctx->ohci->node_id &&
|
2007-01-26 12:38:04 +07:00
|
|
|
ctx->ohci->generation == packet->generation) {
|
2007-01-26 12:38:26 +07:00
|
|
|
spin_unlock_irqrestore(&ctx->ohci->lock, flags);
|
|
|
|
handle_local_request(ctx, packet);
|
|
|
|
return;
|
2007-01-26 12:38:04 +07:00
|
|
|
}
|
2006-12-20 07:58:35 +07:00
|
|
|
|
2007-03-08 00:12:49 +07:00
|
|
|
retval = at_context_queue_packet(ctx, packet);
|
2006-12-20 07:58:35 +07:00
|
|
|
spin_unlock_irqrestore(&ctx->ohci->lock, flags);
|
|
|
|
|
2007-03-08 00:12:49 +07:00
|
|
|
if (retval < 0)
|
|
|
|
packet->callback(packet, &ctx->ohci->card, packet->ack);
|
firewire: OHCI 1.0 Isochronous Receive support
Third rendition of FireWire OHCI 1.0 Isochronous Receive support, using a
zer-copy method similar to OHCI 1.1 which puts the IR data payload directly
into the userspace buffer. The zero-copy implementation eliminates the
video artifacts, audio popping, and buffer underrun problems seen with
version 1 of this patch, as well as fixing a regression in OHCI 1.1 support
introduced by version 2 of this patch.
Successfully tested in OHCI 1.1 mode on the following chipsets:
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
- Ti TSB41AB2 (rev 01), OHCI 1.1 (PCI on SB Audigy)
- Apple UniNorth 2 (rev 81), OHCI 1.1 (PowerBook G4 onboard)
Successfully tested in OHCI 1.0 mode on the following chipsets:
- Agere FW323 (rev 06), OHCI 1.0 (Mac Mini onboard)
- Agere FW323 (rev 06), OHCI 1.0 (PCI)
- Via VT6306 (rev 46), OHCI 1.0 (PCI)
- NEC OrangeLink (rev 01), OHCI 1.0 (PCI)
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
The bulk of testing was done in an x86_64 system, but was also successfully
sanity-tested on other systems, including a PPC(32) PowerBook G4 and an i686
EPIA M10k. Crude benchmarking (watching top during capture) puts the cpu
utilization during capture on the EPIA's 1GHz Via C3 processor around 13%,
which is down from 30% with the v1 code.
Some implementation details:
To maintain the same userspace API as dual-buffer mode, we set up two
descriptors for every incoming packet. The first is an INPUT_MORE descriptor,
pointing to a buffer large enough to hold just the packet's iso headers,
immediately followed by an INPUT_LAST descriptor, pointing to a chunk of the
userspace buffer big enough for the packet's data payload. With this setup,
each incoming packet fills in these two descriptors in a manner that very
closely emulates dual-buffer receive, to the point where the bulk of the
handle_ir_* code is now identical between the two (and probably primed for
some restructuring to share code between them).
The only caveat I have at the moment is that neither of my OHCI 1.0 Via
VT6307-based FireWire controllers work particularly well with this code
for reasons I have yet to figure out.
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2007-12-04 01:43:12 +07:00
|
|
|
|
2006-12-20 07:58:35 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
static void bus_reset_tasklet(unsigned long data)
|
|
|
|
{
|
|
|
|
struct fw_ohci *ohci = (struct fw_ohci *)data;
|
2007-01-26 12:38:04 +07:00
|
|
|
int self_id_count, i, j, reg;
|
2006-12-20 07:58:35 +07:00
|
|
|
int generation, new_generation;
|
|
|
|
unsigned long flags;
|
2007-07-26 00:18:08 +07:00
|
|
|
void *free_rom = NULL;
|
|
|
|
dma_addr_t free_rom_bus = 0;
|
2006-12-20 07:58:35 +07:00
|
|
|
|
|
|
|
reg = reg_read(ohci, OHCI1394_NodeID);
|
|
|
|
if (!(reg & OHCI1394_NodeID_idValid)) {
|
2007-08-30 05:11:40 +07:00
|
|
|
fw_notify("node ID not valid, new bus reset in progress\n");
|
2006-12-20 07:58:35 +07:00
|
|
|
return;
|
|
|
|
}
|
2007-08-30 05:11:40 +07:00
|
|
|
if ((reg & OHCI1394_NodeID_nodeNumber) == 63) {
|
|
|
|
fw_notify("malconfigured bus\n");
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
ohci->node_id = reg & (OHCI1394_NodeID_busNumber |
|
|
|
|
OHCI1394_NodeID_nodeNumber);
|
2006-12-20 07:58:35 +07:00
|
|
|
|
2008-03-20 03:40:32 +07:00
|
|
|
reg = reg_read(ohci, OHCI1394_SelfIDCount);
|
|
|
|
if (reg & OHCI1394_SelfIDCount_selfIDError) {
|
|
|
|
fw_notify("inconsistent self IDs\n");
|
|
|
|
return;
|
|
|
|
}
|
2007-05-08 07:33:32 +07:00
|
|
|
/*
|
|
|
|
* The count in the SelfIDCount register is the number of
|
2006-12-20 07:58:35 +07:00
|
|
|
* bytes in the self ID receive buffer. Since we also receive
|
|
|
|
* the inverted quadlets and a header quadlet, we shift one
|
2007-05-08 07:33:32 +07:00
|
|
|
* bit extra to get the actual number of self IDs.
|
|
|
|
*/
|
2008-03-20 03:40:32 +07:00
|
|
|
self_id_count = (reg >> 3) & 0x3ff;
|
2008-03-20 04:05:02 +07:00
|
|
|
if (self_id_count == 0) {
|
|
|
|
fw_notify("inconsistent self IDs\n");
|
|
|
|
return;
|
|
|
|
}
|
2008-03-01 08:47:15 +07:00
|
|
|
generation = (cond_le32_to_cpu(ohci->self_id_cpu[0]) >> 16) & 0xff;
|
2007-08-25 19:08:19 +07:00
|
|
|
rmb();
|
2006-12-20 07:58:35 +07:00
|
|
|
|
|
|
|
for (i = 1, j = 0; j < self_id_count; i += 2, j++) {
|
2008-03-20 03:40:32 +07:00
|
|
|
if (ohci->self_id_cpu[i] != ~ohci->self_id_cpu[i + 1]) {
|
|
|
|
fw_notify("inconsistent self IDs\n");
|
|
|
|
return;
|
|
|
|
}
|
2008-03-01 08:47:15 +07:00
|
|
|
ohci->self_id_buffer[j] =
|
|
|
|
cond_le32_to_cpu(ohci->self_id_cpu[i]);
|
2006-12-20 07:58:35 +07:00
|
|
|
}
|
2007-08-25 19:08:19 +07:00
|
|
|
rmb();
|
2006-12-20 07:58:35 +07:00
|
|
|
|
2007-05-08 07:33:32 +07:00
|
|
|
/*
|
|
|
|
* Check the consistency of the self IDs we just read. The
|
2006-12-20 07:58:35 +07:00
|
|
|
* problem we face is that a new bus reset can start while we
|
|
|
|
* read out the self IDs from the DMA buffer. If this happens,
|
|
|
|
* the DMA buffer will be overwritten with new self IDs and we
|
|
|
|
* will read out inconsistent data. The OHCI specification
|
|
|
|
* (section 11.2) recommends a technique similar to
|
|
|
|
* linux/seqlock.h, where we remember the generation of the
|
|
|
|
* self IDs in the buffer before reading them out and compare
|
|
|
|
* it to the current generation after reading them out. If
|
|
|
|
* the two generations match we know we have a consistent set
|
2007-05-08 07:33:32 +07:00
|
|
|
* of self IDs.
|
|
|
|
*/
|
2006-12-20 07:58:35 +07:00
|
|
|
|
|
|
|
new_generation = (reg_read(ohci, OHCI1394_SelfIDCount) >> 16) & 0xff;
|
|
|
|
if (new_generation != generation) {
|
|
|
|
fw_notify("recursive bus reset detected, "
|
|
|
|
"discarding self ids\n");
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* FIXME: Document how the locking works. */
|
|
|
|
spin_lock_irqsave(&ohci->lock, flags);
|
|
|
|
|
|
|
|
ohci->generation = generation;
|
2007-03-08 00:12:49 +07:00
|
|
|
context_stop(&ohci->at_request_ctx);
|
|
|
|
context_stop(&ohci->at_response_ctx);
|
2006-12-20 07:58:35 +07:00
|
|
|
reg_write(ohci, OHCI1394_IntEventClear, OHCI1394_busReset);
|
|
|
|
|
2007-05-08 07:33:32 +07:00
|
|
|
/*
|
|
|
|
* This next bit is unrelated to the AT context stuff but we
|
2006-12-20 07:58:35 +07:00
|
|
|
* have to do it under the spinlock also. If a new config rom
|
|
|
|
* was set up before this reset, the old one is now no longer
|
|
|
|
* in use and we can free it. Update the config rom pointers
|
|
|
|
* to point to the current config rom and clear the
|
2007-05-08 07:33:32 +07:00
|
|
|
* next_config_rom pointer so a new udpate can take place.
|
|
|
|
*/
|
2006-12-20 07:58:35 +07:00
|
|
|
|
|
|
|
if (ohci->next_config_rom != NULL) {
|
2007-06-06 06:27:05 +07:00
|
|
|
if (ohci->next_config_rom != ohci->config_rom) {
|
|
|
|
free_rom = ohci->config_rom;
|
|
|
|
free_rom_bus = ohci->config_rom_bus;
|
|
|
|
}
|
2006-12-20 07:58:35 +07:00
|
|
|
ohci->config_rom = ohci->next_config_rom;
|
|
|
|
ohci->config_rom_bus = ohci->next_config_rom_bus;
|
|
|
|
ohci->next_config_rom = NULL;
|
|
|
|
|
2007-05-08 07:33:32 +07:00
|
|
|
/*
|
|
|
|
* Restore config_rom image and manually update
|
2006-12-20 07:58:35 +07:00
|
|
|
* config_rom registers. Writing the header quadlet
|
|
|
|
* will indicate that the config rom is ready, so we
|
2007-05-08 07:33:32 +07:00
|
|
|
* do that last.
|
|
|
|
*/
|
2006-12-20 07:58:35 +07:00
|
|
|
reg_write(ohci, OHCI1394_BusOptions,
|
|
|
|
be32_to_cpu(ohci->config_rom[2]));
|
|
|
|
ohci->config_rom[0] = cpu_to_be32(ohci->next_header);
|
|
|
|
reg_write(ohci, OHCI1394_ConfigROMhdr, ohci->next_header);
|
|
|
|
}
|
|
|
|
|
2008-02-29 02:54:43 +07:00
|
|
|
#ifdef CONFIG_FIREWIRE_OHCI_REMOTE_DMA
|
|
|
|
reg_write(ohci, OHCI1394_PhyReqFilterHiSet, ~0);
|
|
|
|
reg_write(ohci, OHCI1394_PhyReqFilterLoSet, ~0);
|
|
|
|
#endif
|
|
|
|
|
2006-12-20 07:58:35 +07:00
|
|
|
spin_unlock_irqrestore(&ohci->lock, flags);
|
|
|
|
|
2007-07-26 00:18:08 +07:00
|
|
|
if (free_rom)
|
|
|
|
dma_free_coherent(ohci->card.device, CONFIG_ROM_SIZE,
|
|
|
|
free_rom, free_rom_bus);
|
|
|
|
|
2008-03-21 04:04:36 +07:00
|
|
|
log_selfids(generation, self_id_count, ohci->self_id_buffer);
|
|
|
|
|
2007-01-26 12:38:04 +07:00
|
|
|
fw_core_handle_bus_reset(&ohci->card, ohci->node_id, generation,
|
2006-12-20 07:58:35 +07:00
|
|
|
self_id_count, ohci->self_id_buffer);
|
|
|
|
}
|
|
|
|
|
|
|
|
static irqreturn_t irq_handler(int irq, void *data)
|
|
|
|
{
|
|
|
|
struct fw_ohci *ohci = data;
|
2007-03-08 00:12:56 +07:00
|
|
|
u32 event, iso_event, cycle_time;
|
2006-12-20 07:58:35 +07:00
|
|
|
int i;
|
|
|
|
|
|
|
|
event = reg_read(ohci, OHCI1394_IntEventClear);
|
|
|
|
|
2007-06-10 00:31:14 +07:00
|
|
|
if (!event || !~event)
|
2006-12-20 07:58:35 +07:00
|
|
|
return IRQ_NONE;
|
|
|
|
|
|
|
|
reg_write(ohci, OHCI1394_IntEventClear, event);
|
2008-03-21 04:04:36 +07:00
|
|
|
log_irqs(event);
|
2006-12-20 07:58:35 +07:00
|
|
|
|
|
|
|
if (event & OHCI1394_selfIDComplete)
|
|
|
|
tasklet_schedule(&ohci->bus_reset_tasklet);
|
|
|
|
|
|
|
|
if (event & OHCI1394_RQPkt)
|
|
|
|
tasklet_schedule(&ohci->ar_request_ctx.tasklet);
|
|
|
|
|
|
|
|
if (event & OHCI1394_RSPkt)
|
|
|
|
tasklet_schedule(&ohci->ar_response_ctx.tasklet);
|
|
|
|
|
|
|
|
if (event & OHCI1394_reqTxComplete)
|
|
|
|
tasklet_schedule(&ohci->at_request_ctx.tasklet);
|
|
|
|
|
|
|
|
if (event & OHCI1394_respTxComplete)
|
|
|
|
tasklet_schedule(&ohci->at_response_ctx.tasklet);
|
|
|
|
|
2007-02-17 05:34:36 +07:00
|
|
|
iso_event = reg_read(ohci, OHCI1394_IsoRecvIntEventClear);
|
2006-12-20 07:58:35 +07:00
|
|
|
reg_write(ohci, OHCI1394_IsoRecvIntEventClear, iso_event);
|
|
|
|
|
|
|
|
while (iso_event) {
|
|
|
|
i = ffs(iso_event) - 1;
|
2007-02-17 05:34:39 +07:00
|
|
|
tasklet_schedule(&ohci->ir_context_list[i].context.tasklet);
|
2006-12-20 07:58:35 +07:00
|
|
|
iso_event &= ~(1 << i);
|
|
|
|
}
|
|
|
|
|
2007-02-17 05:34:36 +07:00
|
|
|
iso_event = reg_read(ohci, OHCI1394_IsoXmitIntEventClear);
|
2006-12-20 07:58:35 +07:00
|
|
|
reg_write(ohci, OHCI1394_IsoXmitIntEventClear, iso_event);
|
|
|
|
|
|
|
|
while (iso_event) {
|
|
|
|
i = ffs(iso_event) - 1;
|
2007-02-17 05:34:39 +07:00
|
|
|
tasklet_schedule(&ohci->it_context_list[i].context.tasklet);
|
2006-12-20 07:58:35 +07:00
|
|
|
iso_event &= ~(1 << i);
|
|
|
|
}
|
|
|
|
|
2007-08-21 02:58:30 +07:00
|
|
|
if (unlikely(event & OHCI1394_postedWriteErr))
|
|
|
|
fw_error("PCI posted write error\n");
|
|
|
|
|
2007-12-23 04:14:52 +07:00
|
|
|
if (unlikely(event & OHCI1394_cycleTooLong)) {
|
|
|
|
if (printk_ratelimit())
|
|
|
|
fw_notify("isochronous cycle too long\n");
|
|
|
|
reg_write(ohci, OHCI1394_LinkControlSet,
|
|
|
|
OHCI1394_LinkControl_cycleMaster);
|
|
|
|
}
|
|
|
|
|
2007-03-08 00:12:56 +07:00
|
|
|
if (event & OHCI1394_cycle64Seconds) {
|
|
|
|
cycle_time = reg_read(ohci, OHCI1394_IsochronousCycleTimer);
|
|
|
|
if ((cycle_time & 0x80000000) == 0)
|
|
|
|
ohci->bus_seconds++;
|
|
|
|
}
|
|
|
|
|
2006-12-20 07:58:35 +07:00
|
|
|
return IRQ_HANDLED;
|
|
|
|
}
|
|
|
|
|
2007-05-31 06:06:35 +07:00
|
|
|
static int software_reset(struct fw_ohci *ohci)
|
|
|
|
{
|
|
|
|
int i;
|
|
|
|
|
|
|
|
reg_write(ohci, OHCI1394_HCControlSet, OHCI1394_HCControl_softReset);
|
|
|
|
|
|
|
|
for (i = 0; i < OHCI_LOOP_COUNT; i++) {
|
|
|
|
if ((reg_read(ohci, OHCI1394_HCControlSet) &
|
|
|
|
OHCI1394_HCControl_softReset) == 0)
|
|
|
|
return 0;
|
|
|
|
msleep(1);
|
|
|
|
}
|
|
|
|
|
|
|
|
return -EBUSY;
|
|
|
|
}
|
|
|
|
|
2006-12-20 07:58:35 +07:00
|
|
|
static int ohci_enable(struct fw_card *card, u32 *config_rom, size_t length)
|
|
|
|
{
|
|
|
|
struct fw_ohci *ohci = fw_ohci(card);
|
|
|
|
struct pci_dev *dev = to_pci_dev(card->device);
|
|
|
|
|
2007-05-31 06:06:35 +07:00
|
|
|
if (software_reset(ohci)) {
|
|
|
|
fw_error("Failed to reset ohci card.\n");
|
|
|
|
return -EBUSY;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Now enable LPS, which we need in order to start accessing
|
|
|
|
* most of the registers. In fact, on some cards (ALI M5251),
|
|
|
|
* accessing registers in the SClk domain without LPS enabled
|
|
|
|
* will lock up the machine. Wait 50msec to make sure we have
|
|
|
|
* full link enabled.
|
|
|
|
*/
|
|
|
|
reg_write(ohci, OHCI1394_HCControlSet,
|
|
|
|
OHCI1394_HCControl_LPS |
|
|
|
|
OHCI1394_HCControl_postedWriteEnable);
|
|
|
|
flush_writes(ohci);
|
|
|
|
msleep(50);
|
|
|
|
|
|
|
|
reg_write(ohci, OHCI1394_HCControlClear,
|
|
|
|
OHCI1394_HCControl_noByteSwapData);
|
|
|
|
|
|
|
|
reg_write(ohci, OHCI1394_LinkControlSet,
|
|
|
|
OHCI1394_LinkControl_rcvSelfID |
|
|
|
|
OHCI1394_LinkControl_cycleTimerEnable |
|
|
|
|
OHCI1394_LinkControl_cycleMaster);
|
|
|
|
|
|
|
|
reg_write(ohci, OHCI1394_ATRetries,
|
|
|
|
OHCI1394_MAX_AT_REQ_RETRIES |
|
|
|
|
(OHCI1394_MAX_AT_RESP_RETRIES << 4) |
|
|
|
|
(OHCI1394_MAX_PHYS_RESP_RETRIES << 8));
|
|
|
|
|
|
|
|
ar_context_run(&ohci->ar_request_ctx);
|
|
|
|
ar_context_run(&ohci->ar_response_ctx);
|
|
|
|
|
|
|
|
reg_write(ohci, OHCI1394_SelfIDBuffer, ohci->self_id_bus);
|
|
|
|
reg_write(ohci, OHCI1394_PhyUpperBound, 0x00010000);
|
|
|
|
reg_write(ohci, OHCI1394_IntEventClear, ~0);
|
|
|
|
reg_write(ohci, OHCI1394_IntMaskClear, ~0);
|
|
|
|
reg_write(ohci, OHCI1394_IntMaskSet,
|
|
|
|
OHCI1394_selfIDComplete |
|
|
|
|
OHCI1394_RQPkt | OHCI1394_RSPkt |
|
|
|
|
OHCI1394_reqTxComplete | OHCI1394_respTxComplete |
|
|
|
|
OHCI1394_isochRx | OHCI1394_isochTx |
|
2007-12-23 04:14:52 +07:00
|
|
|
OHCI1394_postedWriteErr | OHCI1394_cycleTooLong |
|
|
|
|
OHCI1394_cycle64Seconds | OHCI1394_masterIntEnable);
|
2007-05-31 06:06:35 +07:00
|
|
|
|
|
|
|
/* Activate link_on bit and contender bit in our self ID packets.*/
|
|
|
|
if (ohci_update_phy_reg(card, 4, 0,
|
|
|
|
PHY_LINK_ACTIVE | PHY_CONTENDER) < 0)
|
|
|
|
return -EIO;
|
|
|
|
|
2007-05-08 07:33:32 +07:00
|
|
|
/*
|
|
|
|
* When the link is not yet enabled, the atomic config rom
|
2006-12-20 07:58:35 +07:00
|
|
|
* update mechanism described below in ohci_set_config_rom()
|
|
|
|
* is not active. We have to update ConfigRomHeader and
|
|
|
|
* BusOptions manually, and the write to ConfigROMmap takes
|
|
|
|
* effect immediately. We tie this to the enabling of the
|
|
|
|
* link, so we have a valid config rom before enabling - the
|
|
|
|
* OHCI requires that ConfigROMhdr and BusOptions have valid
|
|
|
|
* values before enabling.
|
|
|
|
*
|
|
|
|
* However, when the ConfigROMmap is written, some controllers
|
|
|
|
* always read back quadlets 0 and 2 from the config rom to
|
|
|
|
* the ConfigRomHeader and BusOptions registers on bus reset.
|
|
|
|
* They shouldn't do that in this initial case where the link
|
|
|
|
* isn't enabled. This means we have to use the same
|
|
|
|
* workaround here, setting the bus header to 0 and then write
|
|
|
|
* the right values in the bus reset tasklet.
|
|
|
|
*/
|
|
|
|
|
2007-06-06 06:27:05 +07:00
|
|
|
if (config_rom) {
|
|
|
|
ohci->next_config_rom =
|
|
|
|
dma_alloc_coherent(ohci->card.device, CONFIG_ROM_SIZE,
|
|
|
|
&ohci->next_config_rom_bus,
|
|
|
|
GFP_KERNEL);
|
|
|
|
if (ohci->next_config_rom == NULL)
|
|
|
|
return -ENOMEM;
|
2006-12-20 07:58:35 +07:00
|
|
|
|
2007-06-06 06:27:05 +07:00
|
|
|
memset(ohci->next_config_rom, 0, CONFIG_ROM_SIZE);
|
|
|
|
fw_memcpy_to_be32(ohci->next_config_rom, config_rom, length * 4);
|
|
|
|
} else {
|
|
|
|
/*
|
|
|
|
* In the suspend case, config_rom is NULL, which
|
|
|
|
* means that we just reuse the old config rom.
|
|
|
|
*/
|
|
|
|
ohci->next_config_rom = ohci->config_rom;
|
|
|
|
ohci->next_config_rom_bus = ohci->config_rom_bus;
|
|
|
|
}
|
2006-12-20 07:58:35 +07:00
|
|
|
|
2007-06-06 06:27:05 +07:00
|
|
|
ohci->next_header = be32_to_cpu(ohci->next_config_rom[0]);
|
2006-12-20 07:58:35 +07:00
|
|
|
ohci->next_config_rom[0] = 0;
|
|
|
|
reg_write(ohci, OHCI1394_ConfigROMhdr, 0);
|
2007-06-06 06:27:05 +07:00
|
|
|
reg_write(ohci, OHCI1394_BusOptions,
|
|
|
|
be32_to_cpu(ohci->next_config_rom[2]));
|
2006-12-20 07:58:35 +07:00
|
|
|
reg_write(ohci, OHCI1394_ConfigROMmap, ohci->next_config_rom_bus);
|
|
|
|
|
|
|
|
reg_write(ohci, OHCI1394_AsReqFilterHiSet, 0x80000000);
|
|
|
|
|
|
|
|
if (request_irq(dev->irq, irq_handler,
|
2007-03-06 09:19:51 +07:00
|
|
|
IRQF_SHARED, ohci_driver_name, ohci)) {
|
2006-12-20 07:58:35 +07:00
|
|
|
fw_error("Failed to allocate shared interrupt %d.\n",
|
|
|
|
dev->irq);
|
|
|
|
dma_free_coherent(ohci->card.device, CONFIG_ROM_SIZE,
|
|
|
|
ohci->config_rom, ohci->config_rom_bus);
|
|
|
|
return -EIO;
|
|
|
|
}
|
|
|
|
|
|
|
|
reg_write(ohci, OHCI1394_HCControlSet,
|
|
|
|
OHCI1394_HCControl_linkEnable |
|
|
|
|
OHCI1394_HCControl_BIBimageValid);
|
|
|
|
flush_writes(ohci);
|
|
|
|
|
2007-05-08 07:33:32 +07:00
|
|
|
/*
|
|
|
|
* We are ready to go, initiate bus reset to finish the
|
|
|
|
* initialization.
|
|
|
|
*/
|
2006-12-20 07:58:35 +07:00
|
|
|
|
|
|
|
fw_core_initiate_bus_reset(&ohci->card, 1);
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int
|
|
|
|
ohci_set_config_rom(struct fw_card *card, u32 *config_rom, size_t length)
|
|
|
|
{
|
|
|
|
struct fw_ohci *ohci;
|
|
|
|
unsigned long flags;
|
2007-07-26 00:18:08 +07:00
|
|
|
int retval = -EBUSY;
|
2006-12-20 07:58:35 +07:00
|
|
|
__be32 *next_config_rom;
|
2008-03-14 06:27:49 +07:00
|
|
|
dma_addr_t uninitialized_var(next_config_rom_bus);
|
2006-12-20 07:58:35 +07:00
|
|
|
|
|
|
|
ohci = fw_ohci(card);
|
|
|
|
|
2007-05-08 07:33:32 +07:00
|
|
|
/*
|
|
|
|
* When the OHCI controller is enabled, the config rom update
|
2006-12-20 07:58:35 +07:00
|
|
|
* mechanism is a bit tricky, but easy enough to use. See
|
|
|
|
* section 5.5.6 in the OHCI specification.
|
|
|
|
*
|
|
|
|
* The OHCI controller caches the new config rom address in a
|
|
|
|
* shadow register (ConfigROMmapNext) and needs a bus reset
|
|
|
|
* for the changes to take place. When the bus reset is
|
|
|
|
* detected, the controller loads the new values for the
|
|
|
|
* ConfigRomHeader and BusOptions registers from the specified
|
|
|
|
* config rom and loads ConfigROMmap from the ConfigROMmapNext
|
|
|
|
* shadow register. All automatically and atomically.
|
|
|
|
*
|
|
|
|
* Now, there's a twist to this story. The automatic load of
|
|
|
|
* ConfigRomHeader and BusOptions doesn't honor the
|
|
|
|
* noByteSwapData bit, so with a be32 config rom, the
|
|
|
|
* controller will load be32 values in to these registers
|
|
|
|
* during the atomic update, even on litte endian
|
|
|
|
* architectures. The workaround we use is to put a 0 in the
|
|
|
|
* header quadlet; 0 is endian agnostic and means that the
|
|
|
|
* config rom isn't ready yet. In the bus reset tasklet we
|
|
|
|
* then set up the real values for the two registers.
|
|
|
|
*
|
|
|
|
* We use ohci->lock to avoid racing with the code that sets
|
|
|
|
* ohci->next_config_rom to NULL (see bus_reset_tasklet).
|
|
|
|
*/
|
|
|
|
|
|
|
|
next_config_rom =
|
|
|
|
dma_alloc_coherent(ohci->card.device, CONFIG_ROM_SIZE,
|
|
|
|
&next_config_rom_bus, GFP_KERNEL);
|
|
|
|
if (next_config_rom == NULL)
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
spin_lock_irqsave(&ohci->lock, flags);
|
|
|
|
|
|
|
|
if (ohci->next_config_rom == NULL) {
|
|
|
|
ohci->next_config_rom = next_config_rom;
|
|
|
|
ohci->next_config_rom_bus = next_config_rom_bus;
|
|
|
|
|
|
|
|
memset(ohci->next_config_rom, 0, CONFIG_ROM_SIZE);
|
|
|
|
fw_memcpy_to_be32(ohci->next_config_rom, config_rom,
|
|
|
|
length * 4);
|
|
|
|
|
|
|
|
ohci->next_header = config_rom[0];
|
|
|
|
ohci->next_config_rom[0] = 0;
|
|
|
|
|
|
|
|
reg_write(ohci, OHCI1394_ConfigROMmap,
|
|
|
|
ohci->next_config_rom_bus);
|
2007-07-26 00:18:08 +07:00
|
|
|
retval = 0;
|
2006-12-20 07:58:35 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
spin_unlock_irqrestore(&ohci->lock, flags);
|
|
|
|
|
2007-05-08 07:33:32 +07:00
|
|
|
/*
|
|
|
|
* Now initiate a bus reset to have the changes take
|
2006-12-20 07:58:35 +07:00
|
|
|
* effect. We clean up the old config rom memory and DMA
|
|
|
|
* mappings in the bus reset tasklet, since the OHCI
|
|
|
|
* controller could need to access it before the bus reset
|
2007-05-08 07:33:32 +07:00
|
|
|
* takes effect.
|
|
|
|
*/
|
2006-12-20 07:58:35 +07:00
|
|
|
if (retval == 0)
|
|
|
|
fw_core_initiate_bus_reset(&ohci->card, 1);
|
2007-07-26 00:18:08 +07:00
|
|
|
else
|
|
|
|
dma_free_coherent(ohci->card.device, CONFIG_ROM_SIZE,
|
|
|
|
next_config_rom, next_config_rom_bus);
|
2006-12-20 07:58:35 +07:00
|
|
|
|
|
|
|
return retval;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void ohci_send_request(struct fw_card *card, struct fw_packet *packet)
|
|
|
|
{
|
|
|
|
struct fw_ohci *ohci = fw_ohci(card);
|
|
|
|
|
|
|
|
at_context_transmit(&ohci->at_request_ctx, packet);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void ohci_send_response(struct fw_card *card, struct fw_packet *packet)
|
|
|
|
{
|
|
|
|
struct fw_ohci *ohci = fw_ohci(card);
|
|
|
|
|
|
|
|
at_context_transmit(&ohci->at_response_ctx, packet);
|
|
|
|
}
|
|
|
|
|
2007-02-07 02:49:32 +07:00
|
|
|
static int ohci_cancel_packet(struct fw_card *card, struct fw_packet *packet)
|
|
|
|
{
|
|
|
|
struct fw_ohci *ohci = fw_ohci(card);
|
2007-03-08 00:12:49 +07:00
|
|
|
struct context *ctx = &ohci->at_request_ctx;
|
|
|
|
struct driver_data *driver_data = packet->driver_data;
|
|
|
|
int retval = -ENOENT;
|
2007-02-07 02:49:32 +07:00
|
|
|
|
2007-03-08 00:12:49 +07:00
|
|
|
tasklet_disable(&ctx->tasklet);
|
2007-02-07 02:49:32 +07:00
|
|
|
|
2007-03-08 00:12:49 +07:00
|
|
|
if (packet->ack != 0)
|
|
|
|
goto out;
|
2007-02-07 02:49:32 +07:00
|
|
|
|
2008-03-21 04:04:36 +07:00
|
|
|
log_ar_at_event('T', packet->speed, packet->header, 0x20);
|
2007-03-08 00:12:49 +07:00
|
|
|
driver_data->packet = NULL;
|
|
|
|
packet->ack = RCODE_CANCELLED;
|
|
|
|
packet->callback(packet, &ohci->card, packet->ack);
|
|
|
|
retval = 0;
|
2007-02-07 02:49:32 +07:00
|
|
|
|
2007-03-08 00:12:49 +07:00
|
|
|
out:
|
|
|
|
tasklet_enable(&ctx->tasklet);
|
2007-02-07 02:49:32 +07:00
|
|
|
|
2007-03-08 00:12:49 +07:00
|
|
|
return retval;
|
2007-02-07 02:49:32 +07:00
|
|
|
}
|
|
|
|
|
2006-12-20 07:58:35 +07:00
|
|
|
static int
|
|
|
|
ohci_enable_phys_dma(struct fw_card *card, int node_id, int generation)
|
|
|
|
{
|
2008-02-29 02:54:43 +07:00
|
|
|
#ifdef CONFIG_FIREWIRE_OHCI_REMOTE_DMA
|
|
|
|
return 0;
|
|
|
|
#else
|
2006-12-20 07:58:35 +07:00
|
|
|
struct fw_ohci *ohci = fw_ohci(card);
|
|
|
|
unsigned long flags;
|
2007-01-24 03:11:43 +07:00
|
|
|
int n, retval = 0;
|
2006-12-20 07:58:35 +07:00
|
|
|
|
2007-05-08 07:33:32 +07:00
|
|
|
/*
|
|
|
|
* FIXME: Make sure this bitmask is cleared when we clear the busReset
|
|
|
|
* interrupt bit. Clear physReqResourceAllBuses on bus reset.
|
|
|
|
*/
|
2006-12-20 07:58:35 +07:00
|
|
|
|
|
|
|
spin_lock_irqsave(&ohci->lock, flags);
|
|
|
|
|
|
|
|
if (ohci->generation != generation) {
|
|
|
|
retval = -ESTALE;
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
2007-05-08 07:33:32 +07:00
|
|
|
/*
|
|
|
|
* Note, if the node ID contains a non-local bus ID, physical DMA is
|
|
|
|
* enabled for _all_ nodes on remote buses.
|
|
|
|
*/
|
2007-01-24 03:11:43 +07:00
|
|
|
|
|
|
|
n = (node_id & 0xffc0) == LOCAL_BUS ? node_id & 0x3f : 63;
|
|
|
|
if (n < 32)
|
|
|
|
reg_write(ohci, OHCI1394_PhyReqFilterLoSet, 1 << n);
|
|
|
|
else
|
|
|
|
reg_write(ohci, OHCI1394_PhyReqFilterHiSet, 1 << (n - 32));
|
|
|
|
|
2006-12-20 07:58:35 +07:00
|
|
|
flush_writes(ohci);
|
|
|
|
out:
|
2007-01-22 02:46:45 +07:00
|
|
|
spin_unlock_irqrestore(&ohci->lock, flags);
|
2006-12-20 07:58:35 +07:00
|
|
|
return retval;
|
2008-02-29 02:54:43 +07:00
|
|
|
#endif /* CONFIG_FIREWIRE_OHCI_REMOTE_DMA */
|
2006-12-20 07:58:35 +07:00
|
|
|
}
|
2007-03-04 20:45:18 +07:00
|
|
|
|
2007-03-08 00:12:56 +07:00
|
|
|
static u64
|
|
|
|
ohci_get_bus_time(struct fw_card *card)
|
|
|
|
{
|
|
|
|
struct fw_ohci *ohci = fw_ohci(card);
|
|
|
|
u32 cycle_time;
|
|
|
|
u64 bus_time;
|
|
|
|
|
|
|
|
cycle_time = reg_read(ohci, OHCI1394_IsochronousCycleTimer);
|
|
|
|
bus_time = ((u64) ohci->bus_seconds << 32) | cycle_time;
|
|
|
|
|
|
|
|
return bus_time;
|
|
|
|
}
|
|
|
|
|
2007-02-17 05:34:46 +07:00
|
|
|
static int handle_ir_dualbuffer_packet(struct context *context,
|
|
|
|
struct descriptor *d,
|
|
|
|
struct descriptor *last)
|
2006-12-20 07:58:35 +07:00
|
|
|
{
|
2007-02-17 05:34:40 +07:00
|
|
|
struct iso_context *ctx =
|
|
|
|
container_of(context, struct iso_context, context);
|
|
|
|
struct db_descriptor *db = (struct db_descriptor *) d;
|
2007-03-15 04:34:53 +07:00
|
|
|
__le32 *ir_header;
|
2007-02-17 05:34:44 +07:00
|
|
|
size_t header_length;
|
2007-03-15 04:34:53 +07:00
|
|
|
void *p, *end;
|
|
|
|
int i;
|
2007-02-17 05:34:46 +07:00
|
|
|
|
2008-02-23 18:24:57 +07:00
|
|
|
if (db->first_res_count != 0 && db->second_res_count != 0) {
|
2007-12-19 15:09:18 +07:00
|
|
|
if (ctx->excess_bytes <= le16_to_cpu(db->second_req_count)) {
|
|
|
|
/* This descriptor isn't done yet, stop iteration. */
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
ctx->excess_bytes -= le16_to_cpu(db->second_req_count);
|
|
|
|
}
|
2007-02-17 05:34:40 +07:00
|
|
|
|
2007-03-15 04:34:53 +07:00
|
|
|
header_length = le16_to_cpu(db->first_req_count) -
|
|
|
|
le16_to_cpu(db->first_res_count);
|
|
|
|
|
|
|
|
i = ctx->header_length;
|
|
|
|
p = db + 1;
|
|
|
|
end = p + header_length;
|
|
|
|
while (p < end && i + ctx->base.header_size <= PAGE_SIZE) {
|
2007-05-08 07:33:32 +07:00
|
|
|
/*
|
|
|
|
* The iso header is byteswapped to little endian by
|
2007-04-11 05:11:16 +07:00
|
|
|
* the controller, but the remaining header quadlets
|
|
|
|
* are big endian. We want to present all the headers
|
|
|
|
* as big endian, so we have to swap the first
|
2007-05-08 07:33:32 +07:00
|
|
|
* quadlet.
|
|
|
|
*/
|
2007-04-11 05:11:16 +07:00
|
|
|
*(u32 *) (ctx->header + i) = __swab32(*(u32 *) (p + 4));
|
|
|
|
memcpy(ctx->header + i + 4, p + 8, ctx->base.header_size - 4);
|
2007-03-15 04:34:53 +07:00
|
|
|
i += ctx->base.header_size;
|
2007-12-19 15:09:18 +07:00
|
|
|
ctx->excess_bytes +=
|
2008-02-23 18:24:57 +07:00
|
|
|
(le32_to_cpu(*(__le32 *)(p + 4)) >> 16) & 0xffff;
|
2007-03-15 04:34:53 +07:00
|
|
|
p += ctx->base.header_size + 4;
|
|
|
|
}
|
|
|
|
ctx->header_length = i;
|
2007-02-17 05:34:44 +07:00
|
|
|
|
2007-12-19 15:09:18 +07:00
|
|
|
ctx->excess_bytes -= le16_to_cpu(db->second_req_count) -
|
|
|
|
le16_to_cpu(db->second_res_count);
|
|
|
|
|
2007-05-08 07:33:35 +07:00
|
|
|
if (le16_to_cpu(db->control) & DESCRIPTOR_IRQ_ALWAYS) {
|
2007-03-15 04:34:53 +07:00
|
|
|
ir_header = (__le32 *) (db + 1);
|
|
|
|
ctx->base.callback(&ctx->base,
|
|
|
|
le32_to_cpu(ir_header[0]) & 0xffff,
|
2007-02-17 05:34:44 +07:00
|
|
|
ctx->header_length, ctx->header,
|
2007-02-17 05:34:40 +07:00
|
|
|
ctx->base.callback_data);
|
2007-02-17 05:34:44 +07:00
|
|
|
ctx->header_length = 0;
|
|
|
|
}
|
2006-12-20 07:58:35 +07:00
|
|
|
|
2007-02-17 05:34:40 +07:00
|
|
|
return 1;
|
2006-12-20 07:58:35 +07:00
|
|
|
}
|
|
|
|
|
firewire: OHCI 1.0 Isochronous Receive support
Third rendition of FireWire OHCI 1.0 Isochronous Receive support, using a
zer-copy method similar to OHCI 1.1 which puts the IR data payload directly
into the userspace buffer. The zero-copy implementation eliminates the
video artifacts, audio popping, and buffer underrun problems seen with
version 1 of this patch, as well as fixing a regression in OHCI 1.1 support
introduced by version 2 of this patch.
Successfully tested in OHCI 1.1 mode on the following chipsets:
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
- Ti TSB41AB2 (rev 01), OHCI 1.1 (PCI on SB Audigy)
- Apple UniNorth 2 (rev 81), OHCI 1.1 (PowerBook G4 onboard)
Successfully tested in OHCI 1.0 mode on the following chipsets:
- Agere FW323 (rev 06), OHCI 1.0 (Mac Mini onboard)
- Agere FW323 (rev 06), OHCI 1.0 (PCI)
- Via VT6306 (rev 46), OHCI 1.0 (PCI)
- NEC OrangeLink (rev 01), OHCI 1.0 (PCI)
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
The bulk of testing was done in an x86_64 system, but was also successfully
sanity-tested on other systems, including a PPC(32) PowerBook G4 and an i686
EPIA M10k. Crude benchmarking (watching top during capture) puts the cpu
utilization during capture on the EPIA's 1GHz Via C3 processor around 13%,
which is down from 30% with the v1 code.
Some implementation details:
To maintain the same userspace API as dual-buffer mode, we set up two
descriptors for every incoming packet. The first is an INPUT_MORE descriptor,
pointing to a buffer large enough to hold just the packet's iso headers,
immediately followed by an INPUT_LAST descriptor, pointing to a chunk of the
userspace buffer big enough for the packet's data payload. With this setup,
each incoming packet fills in these two descriptors in a manner that very
closely emulates dual-buffer receive, to the point where the bulk of the
handle_ir_* code is now identical between the two (and probably primed for
some restructuring to share code between them).
The only caveat I have at the moment is that neither of my OHCI 1.0 Via
VT6307-based FireWire controllers work particularly well with this code
for reasons I have yet to figure out.
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2007-12-04 01:43:12 +07:00
|
|
|
static int handle_ir_packet_per_buffer(struct context *context,
|
|
|
|
struct descriptor *d,
|
|
|
|
struct descriptor *last)
|
|
|
|
{
|
|
|
|
struct iso_context *ctx =
|
|
|
|
container_of(context, struct iso_context, context);
|
2007-12-20 03:26:38 +07:00
|
|
|
struct descriptor *pd;
|
firewire: OHCI 1.0 Isochronous Receive support
Third rendition of FireWire OHCI 1.0 Isochronous Receive support, using a
zer-copy method similar to OHCI 1.1 which puts the IR data payload directly
into the userspace buffer. The zero-copy implementation eliminates the
video artifacts, audio popping, and buffer underrun problems seen with
version 1 of this patch, as well as fixing a regression in OHCI 1.1 support
introduced by version 2 of this patch.
Successfully tested in OHCI 1.1 mode on the following chipsets:
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
- Ti TSB41AB2 (rev 01), OHCI 1.1 (PCI on SB Audigy)
- Apple UniNorth 2 (rev 81), OHCI 1.1 (PowerBook G4 onboard)
Successfully tested in OHCI 1.0 mode on the following chipsets:
- Agere FW323 (rev 06), OHCI 1.0 (Mac Mini onboard)
- Agere FW323 (rev 06), OHCI 1.0 (PCI)
- Via VT6306 (rev 46), OHCI 1.0 (PCI)
- NEC OrangeLink (rev 01), OHCI 1.0 (PCI)
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
The bulk of testing was done in an x86_64 system, but was also successfully
sanity-tested on other systems, including a PPC(32) PowerBook G4 and an i686
EPIA M10k. Crude benchmarking (watching top during capture) puts the cpu
utilization during capture on the EPIA's 1GHz Via C3 processor around 13%,
which is down from 30% with the v1 code.
Some implementation details:
To maintain the same userspace API as dual-buffer mode, we set up two
descriptors for every incoming packet. The first is an INPUT_MORE descriptor,
pointing to a buffer large enough to hold just the packet's iso headers,
immediately followed by an INPUT_LAST descriptor, pointing to a chunk of the
userspace buffer big enough for the packet's data payload. With this setup,
each incoming packet fills in these two descriptors in a manner that very
closely emulates dual-buffer receive, to the point where the bulk of the
handle_ir_* code is now identical between the two (and probably primed for
some restructuring to share code between them).
The only caveat I have at the moment is that neither of my OHCI 1.0 Via
VT6307-based FireWire controllers work particularly well with this code
for reasons I have yet to figure out.
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2007-12-04 01:43:12 +07:00
|
|
|
__le32 *ir_header;
|
2007-12-20 03:26:38 +07:00
|
|
|
void *p;
|
|
|
|
int i;
|
firewire: OHCI 1.0 Isochronous Receive support
Third rendition of FireWire OHCI 1.0 Isochronous Receive support, using a
zer-copy method similar to OHCI 1.1 which puts the IR data payload directly
into the userspace buffer. The zero-copy implementation eliminates the
video artifacts, audio popping, and buffer underrun problems seen with
version 1 of this patch, as well as fixing a regression in OHCI 1.1 support
introduced by version 2 of this patch.
Successfully tested in OHCI 1.1 mode on the following chipsets:
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
- Ti TSB41AB2 (rev 01), OHCI 1.1 (PCI on SB Audigy)
- Apple UniNorth 2 (rev 81), OHCI 1.1 (PowerBook G4 onboard)
Successfully tested in OHCI 1.0 mode on the following chipsets:
- Agere FW323 (rev 06), OHCI 1.0 (Mac Mini onboard)
- Agere FW323 (rev 06), OHCI 1.0 (PCI)
- Via VT6306 (rev 46), OHCI 1.0 (PCI)
- NEC OrangeLink (rev 01), OHCI 1.0 (PCI)
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
The bulk of testing was done in an x86_64 system, but was also successfully
sanity-tested on other systems, including a PPC(32) PowerBook G4 and an i686
EPIA M10k. Crude benchmarking (watching top during capture) puts the cpu
utilization during capture on the EPIA's 1GHz Via C3 processor around 13%,
which is down from 30% with the v1 code.
Some implementation details:
To maintain the same userspace API as dual-buffer mode, we set up two
descriptors for every incoming packet. The first is an INPUT_MORE descriptor,
pointing to a buffer large enough to hold just the packet's iso headers,
immediately followed by an INPUT_LAST descriptor, pointing to a chunk of the
userspace buffer big enough for the packet's data payload. With this setup,
each incoming packet fills in these two descriptors in a manner that very
closely emulates dual-buffer receive, to the point where the bulk of the
handle_ir_* code is now identical between the two (and probably primed for
some restructuring to share code between them).
The only caveat I have at the moment is that neither of my OHCI 1.0 Via
VT6307-based FireWire controllers work particularly well with this code
for reasons I have yet to figure out.
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2007-12-04 01:43:12 +07:00
|
|
|
|
2007-12-20 03:26:38 +07:00
|
|
|
for (pd = d; pd <= last; pd++) {
|
|
|
|
if (pd->transfer_status)
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
if (pd > last)
|
firewire: OHCI 1.0 Isochronous Receive support
Third rendition of FireWire OHCI 1.0 Isochronous Receive support, using a
zer-copy method similar to OHCI 1.1 which puts the IR data payload directly
into the userspace buffer. The zero-copy implementation eliminates the
video artifacts, audio popping, and buffer underrun problems seen with
version 1 of this patch, as well as fixing a regression in OHCI 1.1 support
introduced by version 2 of this patch.
Successfully tested in OHCI 1.1 mode on the following chipsets:
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
- Ti TSB41AB2 (rev 01), OHCI 1.1 (PCI on SB Audigy)
- Apple UniNorth 2 (rev 81), OHCI 1.1 (PowerBook G4 onboard)
Successfully tested in OHCI 1.0 mode on the following chipsets:
- Agere FW323 (rev 06), OHCI 1.0 (Mac Mini onboard)
- Agere FW323 (rev 06), OHCI 1.0 (PCI)
- Via VT6306 (rev 46), OHCI 1.0 (PCI)
- NEC OrangeLink (rev 01), OHCI 1.0 (PCI)
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
The bulk of testing was done in an x86_64 system, but was also successfully
sanity-tested on other systems, including a PPC(32) PowerBook G4 and an i686
EPIA M10k. Crude benchmarking (watching top during capture) puts the cpu
utilization during capture on the EPIA's 1GHz Via C3 processor around 13%,
which is down from 30% with the v1 code.
Some implementation details:
To maintain the same userspace API as dual-buffer mode, we set up two
descriptors for every incoming packet. The first is an INPUT_MORE descriptor,
pointing to a buffer large enough to hold just the packet's iso headers,
immediately followed by an INPUT_LAST descriptor, pointing to a chunk of the
userspace buffer big enough for the packet's data payload. With this setup,
each incoming packet fills in these two descriptors in a manner that very
closely emulates dual-buffer receive, to the point where the bulk of the
handle_ir_* code is now identical between the two (and probably primed for
some restructuring to share code between them).
The only caveat I have at the moment is that neither of my OHCI 1.0 Via
VT6307-based FireWire controllers work particularly well with this code
for reasons I have yet to figure out.
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2007-12-04 01:43:12 +07:00
|
|
|
/* Descriptor(s) not done yet, stop iteration */
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
i = ctx->header_length;
|
2007-12-20 03:26:38 +07:00
|
|
|
p = last + 1;
|
firewire: OHCI 1.0 Isochronous Receive support
Third rendition of FireWire OHCI 1.0 Isochronous Receive support, using a
zer-copy method similar to OHCI 1.1 which puts the IR data payload directly
into the userspace buffer. The zero-copy implementation eliminates the
video artifacts, audio popping, and buffer underrun problems seen with
version 1 of this patch, as well as fixing a regression in OHCI 1.1 support
introduced by version 2 of this patch.
Successfully tested in OHCI 1.1 mode on the following chipsets:
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
- Ti TSB41AB2 (rev 01), OHCI 1.1 (PCI on SB Audigy)
- Apple UniNorth 2 (rev 81), OHCI 1.1 (PowerBook G4 onboard)
Successfully tested in OHCI 1.0 mode on the following chipsets:
- Agere FW323 (rev 06), OHCI 1.0 (Mac Mini onboard)
- Agere FW323 (rev 06), OHCI 1.0 (PCI)
- Via VT6306 (rev 46), OHCI 1.0 (PCI)
- NEC OrangeLink (rev 01), OHCI 1.0 (PCI)
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
The bulk of testing was done in an x86_64 system, but was also successfully
sanity-tested on other systems, including a PPC(32) PowerBook G4 and an i686
EPIA M10k. Crude benchmarking (watching top during capture) puts the cpu
utilization during capture on the EPIA's 1GHz Via C3 processor around 13%,
which is down from 30% with the v1 code.
Some implementation details:
To maintain the same userspace API as dual-buffer mode, we set up two
descriptors for every incoming packet. The first is an INPUT_MORE descriptor,
pointing to a buffer large enough to hold just the packet's iso headers,
immediately followed by an INPUT_LAST descriptor, pointing to a chunk of the
userspace buffer big enough for the packet's data payload. With this setup,
each incoming packet fills in these two descriptors in a manner that very
closely emulates dual-buffer receive, to the point where the bulk of the
handle_ir_* code is now identical between the two (and probably primed for
some restructuring to share code between them).
The only caveat I have at the moment is that neither of my OHCI 1.0 Via
VT6307-based FireWire controllers work particularly well with this code
for reasons I have yet to figure out.
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2007-12-04 01:43:12 +07:00
|
|
|
|
2007-12-20 03:26:38 +07:00
|
|
|
if (ctx->base.header_size > 0 &&
|
|
|
|
i + ctx->base.header_size <= PAGE_SIZE) {
|
firewire: OHCI 1.0 Isochronous Receive support
Third rendition of FireWire OHCI 1.0 Isochronous Receive support, using a
zer-copy method similar to OHCI 1.1 which puts the IR data payload directly
into the userspace buffer. The zero-copy implementation eliminates the
video artifacts, audio popping, and buffer underrun problems seen with
version 1 of this patch, as well as fixing a regression in OHCI 1.1 support
introduced by version 2 of this patch.
Successfully tested in OHCI 1.1 mode on the following chipsets:
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
- Ti TSB41AB2 (rev 01), OHCI 1.1 (PCI on SB Audigy)
- Apple UniNorth 2 (rev 81), OHCI 1.1 (PowerBook G4 onboard)
Successfully tested in OHCI 1.0 mode on the following chipsets:
- Agere FW323 (rev 06), OHCI 1.0 (Mac Mini onboard)
- Agere FW323 (rev 06), OHCI 1.0 (PCI)
- Via VT6306 (rev 46), OHCI 1.0 (PCI)
- NEC OrangeLink (rev 01), OHCI 1.0 (PCI)
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
The bulk of testing was done in an x86_64 system, but was also successfully
sanity-tested on other systems, including a PPC(32) PowerBook G4 and an i686
EPIA M10k. Crude benchmarking (watching top during capture) puts the cpu
utilization during capture on the EPIA's 1GHz Via C3 processor around 13%,
which is down from 30% with the v1 code.
Some implementation details:
To maintain the same userspace API as dual-buffer mode, we set up two
descriptors for every incoming packet. The first is an INPUT_MORE descriptor,
pointing to a buffer large enough to hold just the packet's iso headers,
immediately followed by an INPUT_LAST descriptor, pointing to a chunk of the
userspace buffer big enough for the packet's data payload. With this setup,
each incoming packet fills in these two descriptors in a manner that very
closely emulates dual-buffer receive, to the point where the bulk of the
handle_ir_* code is now identical between the two (and probably primed for
some restructuring to share code between them).
The only caveat I have at the moment is that neither of my OHCI 1.0 Via
VT6307-based FireWire controllers work particularly well with this code
for reasons I have yet to figure out.
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2007-12-04 01:43:12 +07:00
|
|
|
/*
|
|
|
|
* The iso header is byteswapped to little endian by
|
|
|
|
* the controller, but the remaining header quadlets
|
|
|
|
* are big endian. We want to present all the headers
|
|
|
|
* as big endian, so we have to swap the first quadlet.
|
|
|
|
*/
|
|
|
|
*(u32 *) (ctx->header + i) = __swab32(*(u32 *) (p + 4));
|
|
|
|
memcpy(ctx->header + i + 4, p + 8, ctx->base.header_size - 4);
|
2007-12-20 03:26:38 +07:00
|
|
|
ctx->header_length += ctx->base.header_size;
|
firewire: OHCI 1.0 Isochronous Receive support
Third rendition of FireWire OHCI 1.0 Isochronous Receive support, using a
zer-copy method similar to OHCI 1.1 which puts the IR data payload directly
into the userspace buffer. The zero-copy implementation eliminates the
video artifacts, audio popping, and buffer underrun problems seen with
version 1 of this patch, as well as fixing a regression in OHCI 1.1 support
introduced by version 2 of this patch.
Successfully tested in OHCI 1.1 mode on the following chipsets:
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
- Ti TSB41AB2 (rev 01), OHCI 1.1 (PCI on SB Audigy)
- Apple UniNorth 2 (rev 81), OHCI 1.1 (PowerBook G4 onboard)
Successfully tested in OHCI 1.0 mode on the following chipsets:
- Agere FW323 (rev 06), OHCI 1.0 (Mac Mini onboard)
- Agere FW323 (rev 06), OHCI 1.0 (PCI)
- Via VT6306 (rev 46), OHCI 1.0 (PCI)
- NEC OrangeLink (rev 01), OHCI 1.0 (PCI)
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
The bulk of testing was done in an x86_64 system, but was also successfully
sanity-tested on other systems, including a PPC(32) PowerBook G4 and an i686
EPIA M10k. Crude benchmarking (watching top during capture) puts the cpu
utilization during capture on the EPIA's 1GHz Via C3 processor around 13%,
which is down from 30% with the v1 code.
Some implementation details:
To maintain the same userspace API as dual-buffer mode, we set up two
descriptors for every incoming packet. The first is an INPUT_MORE descriptor,
pointing to a buffer large enough to hold just the packet's iso headers,
immediately followed by an INPUT_LAST descriptor, pointing to a chunk of the
userspace buffer big enough for the packet's data payload. With this setup,
each incoming packet fills in these two descriptors in a manner that very
closely emulates dual-buffer receive, to the point where the bulk of the
handle_ir_* code is now identical between the two (and probably primed for
some restructuring to share code between them).
The only caveat I have at the moment is that neither of my OHCI 1.0 Via
VT6307-based FireWire controllers work particularly well with this code
for reasons I have yet to figure out.
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2007-12-04 01:43:12 +07:00
|
|
|
}
|
|
|
|
|
2007-12-20 03:26:38 +07:00
|
|
|
if (le16_to_cpu(last->control) & DESCRIPTOR_IRQ_ALWAYS) {
|
|
|
|
ir_header = (__le32 *) p;
|
firewire: OHCI 1.0 Isochronous Receive support
Third rendition of FireWire OHCI 1.0 Isochronous Receive support, using a
zer-copy method similar to OHCI 1.1 which puts the IR data payload directly
into the userspace buffer. The zero-copy implementation eliminates the
video artifacts, audio popping, and buffer underrun problems seen with
version 1 of this patch, as well as fixing a regression in OHCI 1.1 support
introduced by version 2 of this patch.
Successfully tested in OHCI 1.1 mode on the following chipsets:
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
- Ti TSB41AB2 (rev 01), OHCI 1.1 (PCI on SB Audigy)
- Apple UniNorth 2 (rev 81), OHCI 1.1 (PowerBook G4 onboard)
Successfully tested in OHCI 1.0 mode on the following chipsets:
- Agere FW323 (rev 06), OHCI 1.0 (Mac Mini onboard)
- Agere FW323 (rev 06), OHCI 1.0 (PCI)
- Via VT6306 (rev 46), OHCI 1.0 (PCI)
- NEC OrangeLink (rev 01), OHCI 1.0 (PCI)
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
The bulk of testing was done in an x86_64 system, but was also successfully
sanity-tested on other systems, including a PPC(32) PowerBook G4 and an i686
EPIA M10k. Crude benchmarking (watching top during capture) puts the cpu
utilization during capture on the EPIA's 1GHz Via C3 processor around 13%,
which is down from 30% with the v1 code.
Some implementation details:
To maintain the same userspace API as dual-buffer mode, we set up two
descriptors for every incoming packet. The first is an INPUT_MORE descriptor,
pointing to a buffer large enough to hold just the packet's iso headers,
immediately followed by an INPUT_LAST descriptor, pointing to a chunk of the
userspace buffer big enough for the packet's data payload. With this setup,
each incoming packet fills in these two descriptors in a manner that very
closely emulates dual-buffer receive, to the point where the bulk of the
handle_ir_* code is now identical between the two (and probably primed for
some restructuring to share code between them).
The only caveat I have at the moment is that neither of my OHCI 1.0 Via
VT6307-based FireWire controllers work particularly well with this code
for reasons I have yet to figure out.
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2007-12-04 01:43:12 +07:00
|
|
|
ctx->base.callback(&ctx->base,
|
|
|
|
le32_to_cpu(ir_header[0]) & 0xffff,
|
|
|
|
ctx->header_length, ctx->header,
|
|
|
|
ctx->base.callback_data);
|
|
|
|
ctx->header_length = 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
2007-02-17 05:34:39 +07:00
|
|
|
static int handle_it_packet(struct context *context,
|
|
|
|
struct descriptor *d,
|
|
|
|
struct descriptor *last)
|
2006-12-20 07:58:35 +07:00
|
|
|
{
|
2007-02-17 05:34:39 +07:00
|
|
|
struct iso_context *ctx =
|
|
|
|
container_of(context, struct iso_context, context);
|
2007-03-04 20:45:18 +07:00
|
|
|
|
2007-02-17 05:34:39 +07:00
|
|
|
if (last->transfer_status == 0)
|
|
|
|
/* This descriptor isn't done yet, stop iteration. */
|
|
|
|
return 0;
|
|
|
|
|
2007-05-08 07:33:35 +07:00
|
|
|
if (le16_to_cpu(last->control) & DESCRIPTOR_IRQ_ALWAYS)
|
2007-02-17 05:34:44 +07:00
|
|
|
ctx->base.callback(&ctx->base, le16_to_cpu(last->res_count),
|
|
|
|
0, NULL, ctx->base.callback_data);
|
2007-02-17 05:34:39 +07:00
|
|
|
|
|
|
|
return 1;
|
2006-12-20 07:58:35 +07:00
|
|
|
}
|
|
|
|
|
2007-02-17 05:34:39 +07:00
|
|
|
static struct fw_iso_context *
|
2007-03-15 04:34:54 +07:00
|
|
|
ohci_allocate_iso_context(struct fw_card *card, int type, size_t header_size)
|
2006-12-20 07:58:35 +07:00
|
|
|
{
|
|
|
|
struct fw_ohci *ohci = fw_ohci(card);
|
|
|
|
struct iso_context *ctx, *list;
|
2007-02-17 05:34:39 +07:00
|
|
|
descriptor_callback_t callback;
|
2007-02-17 05:34:40 +07:00
|
|
|
u32 *mask, regs;
|
2006-12-20 07:58:35 +07:00
|
|
|
unsigned long flags;
|
2007-02-17 05:34:44 +07:00
|
|
|
int index, retval = -ENOMEM;
|
2006-12-20 07:58:35 +07:00
|
|
|
|
|
|
|
if (type == FW_ISO_CONTEXT_TRANSMIT) {
|
|
|
|
mask = &ohci->it_context_mask;
|
|
|
|
list = ohci->it_context_list;
|
2007-02-17 05:34:39 +07:00
|
|
|
callback = handle_it_packet;
|
2006-12-20 07:58:35 +07:00
|
|
|
} else {
|
2007-03-04 20:45:18 +07:00
|
|
|
mask = &ohci->ir_context_mask;
|
|
|
|
list = ohci->ir_context_list;
|
firewire: OHCI 1.0 Isochronous Receive support
Third rendition of FireWire OHCI 1.0 Isochronous Receive support, using a
zer-copy method similar to OHCI 1.1 which puts the IR data payload directly
into the userspace buffer. The zero-copy implementation eliminates the
video artifacts, audio popping, and buffer underrun problems seen with
version 1 of this patch, as well as fixing a regression in OHCI 1.1 support
introduced by version 2 of this patch.
Successfully tested in OHCI 1.1 mode on the following chipsets:
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
- Ti TSB41AB2 (rev 01), OHCI 1.1 (PCI on SB Audigy)
- Apple UniNorth 2 (rev 81), OHCI 1.1 (PowerBook G4 onboard)
Successfully tested in OHCI 1.0 mode on the following chipsets:
- Agere FW323 (rev 06), OHCI 1.0 (Mac Mini onboard)
- Agere FW323 (rev 06), OHCI 1.0 (PCI)
- Via VT6306 (rev 46), OHCI 1.0 (PCI)
- NEC OrangeLink (rev 01), OHCI 1.0 (PCI)
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
The bulk of testing was done in an x86_64 system, but was also successfully
sanity-tested on other systems, including a PPC(32) PowerBook G4 and an i686
EPIA M10k. Crude benchmarking (watching top during capture) puts the cpu
utilization during capture on the EPIA's 1GHz Via C3 processor around 13%,
which is down from 30% with the v1 code.
Some implementation details:
To maintain the same userspace API as dual-buffer mode, we set up two
descriptors for every incoming packet. The first is an INPUT_MORE descriptor,
pointing to a buffer large enough to hold just the packet's iso headers,
immediately followed by an INPUT_LAST descriptor, pointing to a chunk of the
userspace buffer big enough for the packet's data payload. With this setup,
each incoming packet fills in these two descriptors in a manner that very
closely emulates dual-buffer receive, to the point where the bulk of the
handle_ir_* code is now identical between the two (and probably primed for
some restructuring to share code between them).
The only caveat I have at the moment is that neither of my OHCI 1.0 Via
VT6307-based FireWire controllers work particularly well with this code
for reasons I have yet to figure out.
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2007-12-04 01:43:12 +07:00
|
|
|
if (ohci->version >= OHCI_VERSION_1_1)
|
|
|
|
callback = handle_ir_dualbuffer_packet;
|
|
|
|
else
|
|
|
|
callback = handle_ir_packet_per_buffer;
|
2006-12-20 07:58:35 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
spin_lock_irqsave(&ohci->lock, flags);
|
|
|
|
index = ffs(*mask) - 1;
|
|
|
|
if (index >= 0)
|
|
|
|
*mask &= ~(1 << index);
|
|
|
|
spin_unlock_irqrestore(&ohci->lock, flags);
|
|
|
|
|
|
|
|
if (index < 0)
|
|
|
|
return ERR_PTR(-EBUSY);
|
|
|
|
|
2007-03-04 20:45:18 +07:00
|
|
|
if (type == FW_ISO_CONTEXT_TRANSMIT)
|
|
|
|
regs = OHCI1394_IsoXmitContextBase(index);
|
|
|
|
else
|
|
|
|
regs = OHCI1394_IsoRcvContextBase(index);
|
|
|
|
|
2006-12-20 07:58:35 +07:00
|
|
|
ctx = &list[index];
|
2007-05-10 06:23:14 +07:00
|
|
|
memset(ctx, 0, sizeof(*ctx));
|
2007-02-17 05:34:44 +07:00
|
|
|
ctx->header_length = 0;
|
|
|
|
ctx->header = (void *) __get_free_page(GFP_KERNEL);
|
|
|
|
if (ctx->header == NULL)
|
|
|
|
goto out;
|
|
|
|
|
firewire: fw-ohci: Dynamically allocate buffers for DMA descriptors
Previously, the fw-ohci driver used fixed-length buffers for storing
descriptors for isochronous receive DMA programs. If an application
(such as libdc1394) generated a DMA program that was too large, fw-ohci
would reach the limit of its fixed-sized buffer and return an error to
userspace.
This patch replaces the fixed-length ring-buffer with a linked-list of
page-sized buffers. Additional buffers can be dynamically allocated and
appended to the list when necessary. For a particular context, buffers
are kept around after use and reused as necessary, so there is no
allocation taking place after the DMA program is generated for the first
time.
In addition, the buffers it uses are coherent for DMA so there is no
syncing required before and after writes. This syncing wasn't properly
done in the previous version of the code.
-
This is the fourth version of my patch that replaces a fixed-length
buffer for DMA descriptors with a dynamically allocated linked-list of
buffers.
As we discovered with the last attempt, new context programs are
sometimes queued from interrupt context, making it unacceptable to call
tasklet_disable() from context_get_descriptors().
This version of the patch uses ohci->lock for all locking needs instead
of tasklet_disable/enable. There is a new requirement that
context_get_descriptors() be called while holding ohci->lock. It was
already held for the AT context, so adding the requirement for the iso
context did not seem particularly onerous. In addition, this has the
side benefit of allowing iso queue to be safely called from concurrent
user-space threads, which previously was not safe.
Signed-off-by: David Moore <dcm@acm.org>
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
-
Fixes the following issues:
- Isochronous reception stopped prematurely if an application used a
larger buffer. (Reproduced with coriander.)
- Isochronous reception stopped after one or a few frames on VT630x
in OHCI 1.0 mode. (Fixes reception in coriander, but dvgrab still
doesn't work with these chips.)
Patch update: struct member alignment, whitespace nits
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2008-01-07 05:21:41 +07:00
|
|
|
retval = context_init(&ctx->context, ohci, regs, callback);
|
2007-02-17 05:34:44 +07:00
|
|
|
if (retval < 0)
|
|
|
|
goto out_with_header;
|
2006-12-20 07:58:35 +07:00
|
|
|
|
|
|
|
return &ctx->base;
|
2007-02-17 05:34:44 +07:00
|
|
|
|
|
|
|
out_with_header:
|
|
|
|
free_page((unsigned long)ctx->header);
|
|
|
|
out:
|
|
|
|
spin_lock_irqsave(&ohci->lock, flags);
|
|
|
|
*mask |= 1 << index;
|
|
|
|
spin_unlock_irqrestore(&ohci->lock, flags);
|
|
|
|
|
|
|
|
return ERR_PTR(retval);
|
2006-12-20 07:58:35 +07:00
|
|
|
}
|
|
|
|
|
2007-03-15 04:34:54 +07:00
|
|
|
static int ohci_start_iso(struct fw_iso_context *base,
|
|
|
|
s32 cycle, u32 sync, u32 tags)
|
2006-12-20 07:58:35 +07:00
|
|
|
{
|
2007-03-04 20:45:18 +07:00
|
|
|
struct iso_context *ctx = container_of(base, struct iso_context, base);
|
2007-02-17 05:34:39 +07:00
|
|
|
struct fw_ohci *ohci = ctx->context.ohci;
|
2007-03-29 01:26:10 +07:00
|
|
|
u32 control, match;
|
2006-12-20 07:58:35 +07:00
|
|
|
int index;
|
|
|
|
|
2007-02-17 05:34:40 +07:00
|
|
|
if (ctx->base.type == FW_ISO_CONTEXT_TRANSMIT) {
|
|
|
|
index = ctx - ohci->it_context_list;
|
2007-03-29 01:26:10 +07:00
|
|
|
match = 0;
|
|
|
|
if (cycle >= 0)
|
|
|
|
match = IT_CONTEXT_CYCLE_MATCH_ENABLE |
|
2007-02-17 05:34:40 +07:00
|
|
|
(cycle & 0x7fff) << 16;
|
2007-02-17 05:34:50 +07:00
|
|
|
|
2007-02-17 05:34:40 +07:00
|
|
|
reg_write(ohci, OHCI1394_IsoXmitIntEventClear, 1 << index);
|
|
|
|
reg_write(ohci, OHCI1394_IsoXmitIntMaskSet, 1 << index);
|
2007-03-29 01:26:10 +07:00
|
|
|
context_run(&ctx->context, match);
|
2007-02-17 05:34:40 +07:00
|
|
|
} else {
|
|
|
|
index = ctx - ohci->ir_context_list;
|
firewire: OHCI 1.0 Isochronous Receive support
Third rendition of FireWire OHCI 1.0 Isochronous Receive support, using a
zer-copy method similar to OHCI 1.1 which puts the IR data payload directly
into the userspace buffer. The zero-copy implementation eliminates the
video artifacts, audio popping, and buffer underrun problems seen with
version 1 of this patch, as well as fixing a regression in OHCI 1.1 support
introduced by version 2 of this patch.
Successfully tested in OHCI 1.1 mode on the following chipsets:
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
- Ti TSB41AB2 (rev 01), OHCI 1.1 (PCI on SB Audigy)
- Apple UniNorth 2 (rev 81), OHCI 1.1 (PowerBook G4 onboard)
Successfully tested in OHCI 1.0 mode on the following chipsets:
- Agere FW323 (rev 06), OHCI 1.0 (Mac Mini onboard)
- Agere FW323 (rev 06), OHCI 1.0 (PCI)
- Via VT6306 (rev 46), OHCI 1.0 (PCI)
- NEC OrangeLink (rev 01), OHCI 1.0 (PCI)
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
The bulk of testing was done in an x86_64 system, but was also successfully
sanity-tested on other systems, including a PPC(32) PowerBook G4 and an i686
EPIA M10k. Crude benchmarking (watching top during capture) puts the cpu
utilization during capture on the EPIA's 1GHz Via C3 processor around 13%,
which is down from 30% with the v1 code.
Some implementation details:
To maintain the same userspace API as dual-buffer mode, we set up two
descriptors for every incoming packet. The first is an INPUT_MORE descriptor,
pointing to a buffer large enough to hold just the packet's iso headers,
immediately followed by an INPUT_LAST descriptor, pointing to a chunk of the
userspace buffer big enough for the packet's data payload. With this setup,
each incoming packet fills in these two descriptors in a manner that very
closely emulates dual-buffer receive, to the point where the bulk of the
handle_ir_* code is now identical between the two (and probably primed for
some restructuring to share code between them).
The only caveat I have at the moment is that neither of my OHCI 1.0 Via
VT6307-based FireWire controllers work particularly well with this code
for reasons I have yet to figure out.
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2007-12-04 01:43:12 +07:00
|
|
|
control = IR_CONTEXT_ISOCH_HEADER;
|
|
|
|
if (ohci->version >= OHCI_VERSION_1_1)
|
|
|
|
control |= IR_CONTEXT_DUAL_BUFFER_MODE;
|
2007-03-29 01:26:10 +07:00
|
|
|
match = (tags << 28) | (sync << 8) | ctx->base.channel;
|
|
|
|
if (cycle >= 0) {
|
|
|
|
match |= (cycle & 0x07fff) << 12;
|
|
|
|
control |= IR_CONTEXT_CYCLE_MATCH_ENABLE;
|
|
|
|
}
|
2006-12-20 07:58:35 +07:00
|
|
|
|
2007-02-17 05:34:40 +07:00
|
|
|
reg_write(ohci, OHCI1394_IsoRecvIntEventClear, 1 << index);
|
|
|
|
reg_write(ohci, OHCI1394_IsoRecvIntMaskSet, 1 << index);
|
2007-05-08 07:33:35 +07:00
|
|
|
reg_write(ohci, CONTEXT_MATCH(ctx->context.regs), match);
|
2007-03-29 01:26:10 +07:00
|
|
|
context_run(&ctx->context, control);
|
2007-02-17 05:34:40 +07:00
|
|
|
}
|
2006-12-20 07:58:35 +07:00
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2007-02-17 05:34:42 +07:00
|
|
|
static int ohci_stop_iso(struct fw_iso_context *base)
|
|
|
|
{
|
|
|
|
struct fw_ohci *ohci = fw_ohci(base->card);
|
2007-03-04 20:45:18 +07:00
|
|
|
struct iso_context *ctx = container_of(base, struct iso_context, base);
|
2007-02-17 05:34:42 +07:00
|
|
|
int index;
|
|
|
|
|
|
|
|
if (ctx->base.type == FW_ISO_CONTEXT_TRANSMIT) {
|
|
|
|
index = ctx - ohci->it_context_list;
|
|
|
|
reg_write(ohci, OHCI1394_IsoXmitIntMaskClear, 1 << index);
|
|
|
|
} else {
|
|
|
|
index = ctx - ohci->ir_context_list;
|
|
|
|
reg_write(ohci, OHCI1394_IsoRecvIntMaskClear, 1 << index);
|
|
|
|
}
|
|
|
|
flush_writes(ohci);
|
|
|
|
context_stop(&ctx->context);
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2006-12-20 07:58:35 +07:00
|
|
|
static void ohci_free_iso_context(struct fw_iso_context *base)
|
|
|
|
{
|
|
|
|
struct fw_ohci *ohci = fw_ohci(base->card);
|
2007-03-04 20:45:18 +07:00
|
|
|
struct iso_context *ctx = container_of(base, struct iso_context, base);
|
2006-12-20 07:58:35 +07:00
|
|
|
unsigned long flags;
|
|
|
|
int index;
|
|
|
|
|
2007-02-17 05:34:42 +07:00
|
|
|
ohci_stop_iso(base);
|
|
|
|
context_release(&ctx->context);
|
2007-02-17 05:34:44 +07:00
|
|
|
free_page((unsigned long)ctx->header);
|
2007-02-17 05:34:42 +07:00
|
|
|
|
2006-12-20 07:58:35 +07:00
|
|
|
spin_lock_irqsave(&ohci->lock, flags);
|
|
|
|
|
|
|
|
if (ctx->base.type == FW_ISO_CONTEXT_TRANSMIT) {
|
|
|
|
index = ctx - ohci->it_context_list;
|
|
|
|
ohci->it_context_mask |= 1 << index;
|
|
|
|
} else {
|
|
|
|
index = ctx - ohci->ir_context_list;
|
|
|
|
ohci->ir_context_mask |= 1 << index;
|
|
|
|
}
|
|
|
|
|
|
|
|
spin_unlock_irqrestore(&ohci->lock, flags);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int
|
2007-02-17 05:34:40 +07:00
|
|
|
ohci_queue_iso_transmit(struct fw_iso_context *base,
|
|
|
|
struct fw_iso_packet *packet,
|
|
|
|
struct fw_iso_buffer *buffer,
|
|
|
|
unsigned long payload)
|
2006-12-20 07:58:35 +07:00
|
|
|
{
|
2007-03-04 20:45:18 +07:00
|
|
|
struct iso_context *ctx = container_of(base, struct iso_context, base);
|
2007-02-17 05:34:39 +07:00
|
|
|
struct descriptor *d, *last, *pd;
|
2006-12-20 07:58:35 +07:00
|
|
|
struct fw_iso_packet *p;
|
|
|
|
__le32 *header;
|
2007-02-17 05:34:38 +07:00
|
|
|
dma_addr_t d_bus, page_bus;
|
2006-12-20 07:58:35 +07:00
|
|
|
u32 z, header_z, payload_z, irq;
|
|
|
|
u32 payload_index, payload_end_index, next_page_index;
|
2007-02-17 05:34:39 +07:00
|
|
|
int page, end_page, i, length, offset;
|
2006-12-20 07:58:35 +07:00
|
|
|
|
2007-05-08 07:33:32 +07:00
|
|
|
/*
|
|
|
|
* FIXME: Cycle lost behavior should be configurable: lose
|
|
|
|
* packet, retransmit or terminate..
|
|
|
|
*/
|
2006-12-20 07:58:35 +07:00
|
|
|
|
|
|
|
p = packet;
|
2007-02-17 05:34:38 +07:00
|
|
|
payload_index = payload;
|
2006-12-20 07:58:35 +07:00
|
|
|
|
|
|
|
if (p->skip)
|
|
|
|
z = 1;
|
|
|
|
else
|
|
|
|
z = 2;
|
|
|
|
if (p->header_length > 0)
|
|
|
|
z++;
|
|
|
|
|
|
|
|
/* Determine the first page the payload isn't contained in. */
|
|
|
|
end_page = PAGE_ALIGN(payload_index + p->payload_length) >> PAGE_SHIFT;
|
|
|
|
if (p->payload_length > 0)
|
|
|
|
payload_z = end_page - (payload_index >> PAGE_SHIFT);
|
|
|
|
else
|
|
|
|
payload_z = 0;
|
|
|
|
|
|
|
|
z += payload_z;
|
|
|
|
|
|
|
|
/* Get header size in number of descriptors. */
|
2007-05-10 06:23:14 +07:00
|
|
|
header_z = DIV_ROUND_UP(p->header_length, sizeof(*d));
|
2006-12-20 07:58:35 +07:00
|
|
|
|
2007-02-17 05:34:39 +07:00
|
|
|
d = context_get_descriptors(&ctx->context, z + header_z, &d_bus);
|
|
|
|
if (d == NULL)
|
|
|
|
return -ENOMEM;
|
2006-12-20 07:58:35 +07:00
|
|
|
|
|
|
|
if (!p->skip) {
|
2007-05-08 07:33:35 +07:00
|
|
|
d[0].control = cpu_to_le16(DESCRIPTOR_KEY_IMMEDIATE);
|
2006-12-20 07:58:35 +07:00
|
|
|
d[0].req_count = cpu_to_le16(8);
|
|
|
|
|
|
|
|
header = (__le32 *) &d[1];
|
2007-05-08 07:33:35 +07:00
|
|
|
header[0] = cpu_to_le32(IT_HEADER_SY(p->sy) |
|
|
|
|
IT_HEADER_TAG(p->tag) |
|
|
|
|
IT_HEADER_TCODE(TCODE_STREAM_DATA) |
|
|
|
|
IT_HEADER_CHANNEL(ctx->base.channel) |
|
|
|
|
IT_HEADER_SPEED(ctx->base.speed));
|
2006-12-20 07:58:35 +07:00
|
|
|
header[1] =
|
2007-05-08 07:33:35 +07:00
|
|
|
cpu_to_le32(IT_HEADER_DATA_LENGTH(p->header_length +
|
2006-12-20 07:58:35 +07:00
|
|
|
p->payload_length));
|
|
|
|
}
|
|
|
|
|
|
|
|
if (p->header_length > 0) {
|
|
|
|
d[2].req_count = cpu_to_le16(p->header_length);
|
2007-05-10 06:23:14 +07:00
|
|
|
d[2].data_address = cpu_to_le32(d_bus + z * sizeof(*d));
|
2006-12-20 07:58:35 +07:00
|
|
|
memcpy(&d[z], p->header, p->header_length);
|
|
|
|
}
|
|
|
|
|
|
|
|
pd = d + z - payload_z;
|
|
|
|
payload_end_index = payload_index + p->payload_length;
|
|
|
|
for (i = 0; i < payload_z; i++) {
|
|
|
|
page = payload_index >> PAGE_SHIFT;
|
|
|
|
offset = payload_index & ~PAGE_MASK;
|
|
|
|
next_page_index = (page + 1) << PAGE_SHIFT;
|
|
|
|
length =
|
|
|
|
min(next_page_index, payload_end_index) - payload_index;
|
|
|
|
pd[i].req_count = cpu_to_le16(length);
|
2007-02-17 05:34:38 +07:00
|
|
|
|
|
|
|
page_bus = page_private(buffer->pages[page]);
|
|
|
|
pd[i].data_address = cpu_to_le32(page_bus + offset);
|
2006-12-20 07:58:35 +07:00
|
|
|
|
|
|
|
payload_index += length;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (p->interrupt)
|
2007-05-08 07:33:35 +07:00
|
|
|
irq = DESCRIPTOR_IRQ_ALWAYS;
|
2006-12-20 07:58:35 +07:00
|
|
|
else
|
2007-05-08 07:33:35 +07:00
|
|
|
irq = DESCRIPTOR_NO_IRQ;
|
2006-12-20 07:58:35 +07:00
|
|
|
|
2007-02-17 05:34:39 +07:00
|
|
|
last = z == 2 ? d : d + z - 1;
|
2007-05-08 07:33:35 +07:00
|
|
|
last->control |= cpu_to_le16(DESCRIPTOR_OUTPUT_LAST |
|
|
|
|
DESCRIPTOR_STATUS |
|
|
|
|
DESCRIPTOR_BRANCH_ALWAYS |
|
2007-02-17 05:34:35 +07:00
|
|
|
irq);
|
2006-12-20 07:58:35 +07:00
|
|
|
|
2007-02-17 05:34:39 +07:00
|
|
|
context_append(&ctx->context, d, z, header_z);
|
2006-12-20 07:58:35 +07:00
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
2007-03-04 20:45:18 +07:00
|
|
|
|
2007-02-17 05:34:40 +07:00
|
|
|
static int
|
2007-02-17 05:34:46 +07:00
|
|
|
ohci_queue_iso_receive_dualbuffer(struct fw_iso_context *base,
|
|
|
|
struct fw_iso_packet *packet,
|
|
|
|
struct fw_iso_buffer *buffer,
|
|
|
|
unsigned long payload)
|
2007-02-17 05:34:40 +07:00
|
|
|
{
|
|
|
|
struct iso_context *ctx = container_of(base, struct iso_context, base);
|
|
|
|
struct db_descriptor *db = NULL;
|
|
|
|
struct descriptor *d;
|
|
|
|
struct fw_iso_packet *p;
|
|
|
|
dma_addr_t d_bus, page_bus;
|
|
|
|
u32 z, header_z, length, rest;
|
2007-03-15 04:34:53 +07:00
|
|
|
int page, offset, packet_count, header_size;
|
2007-03-04 20:45:18 +07:00
|
|
|
|
2007-05-08 07:33:32 +07:00
|
|
|
/*
|
|
|
|
* FIXME: Cycle lost behavior should be configurable: lose
|
|
|
|
* packet, retransmit or terminate..
|
|
|
|
*/
|
2007-02-17 05:34:40 +07:00
|
|
|
|
|
|
|
p = packet;
|
|
|
|
z = 2;
|
|
|
|
|
2007-05-08 07:33:32 +07:00
|
|
|
/*
|
|
|
|
* The OHCI controller puts the status word in the header
|
|
|
|
* buffer too, so we need 4 extra bytes per packet.
|
|
|
|
*/
|
2007-03-15 04:34:53 +07:00
|
|
|
packet_count = p->header_length / ctx->base.header_size;
|
|
|
|
header_size = packet_count * (ctx->base.header_size + 4);
|
|
|
|
|
2007-02-17 05:34:40 +07:00
|
|
|
/* Get header size in number of descriptors. */
|
2007-05-10 06:23:14 +07:00
|
|
|
header_z = DIV_ROUND_UP(header_size, sizeof(*d));
|
2007-02-17 05:34:40 +07:00
|
|
|
page = payload >> PAGE_SHIFT;
|
|
|
|
offset = payload & ~PAGE_MASK;
|
|
|
|
rest = p->payload_length;
|
|
|
|
|
|
|
|
/* FIXME: make packet-per-buffer/dual-buffer a context option */
|
|
|
|
while (rest > 0) {
|
|
|
|
d = context_get_descriptors(&ctx->context,
|
|
|
|
z + header_z, &d_bus);
|
|
|
|
if (d == NULL)
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
db = (struct db_descriptor *) d;
|
2007-05-08 07:33:35 +07:00
|
|
|
db->control = cpu_to_le16(DESCRIPTOR_STATUS |
|
|
|
|
DESCRIPTOR_BRANCH_ALWAYS);
|
2007-03-15 04:34:53 +07:00
|
|
|
db->first_size = cpu_to_le16(ctx->base.header_size + 4);
|
2007-12-19 15:09:18 +07:00
|
|
|
if (p->skip && rest == p->payload_length) {
|
|
|
|
db->control |= cpu_to_le16(DESCRIPTOR_WAIT);
|
|
|
|
db->first_req_count = db->first_size;
|
|
|
|
} else {
|
|
|
|
db->first_req_count = cpu_to_le16(header_size);
|
|
|
|
}
|
2007-02-17 05:34:45 +07:00
|
|
|
db->first_res_count = db->first_req_count;
|
2007-05-10 06:23:14 +07:00
|
|
|
db->first_buffer = cpu_to_le32(d_bus + sizeof(*db));
|
2007-03-04 20:45:18 +07:00
|
|
|
|
2007-12-19 15:09:18 +07:00
|
|
|
if (p->skip && rest == p->payload_length)
|
|
|
|
length = 4;
|
|
|
|
else if (offset + rest < PAGE_SIZE)
|
2007-02-17 05:34:40 +07:00
|
|
|
length = rest;
|
|
|
|
else
|
|
|
|
length = PAGE_SIZE - offset;
|
|
|
|
|
2007-02-17 05:34:45 +07:00
|
|
|
db->second_req_count = cpu_to_le16(length);
|
|
|
|
db->second_res_count = db->second_req_count;
|
2007-02-17 05:34:40 +07:00
|
|
|
page_bus = page_private(buffer->pages[page]);
|
|
|
|
db->second_buffer = cpu_to_le32(page_bus + offset);
|
|
|
|
|
2007-02-17 05:34:47 +07:00
|
|
|
if (p->interrupt && length == rest)
|
2007-05-08 07:33:35 +07:00
|
|
|
db->control |= cpu_to_le16(DESCRIPTOR_IRQ_ALWAYS);
|
2007-02-17 05:34:47 +07:00
|
|
|
|
2007-02-17 05:34:40 +07:00
|
|
|
context_append(&ctx->context, d, z, header_z);
|
|
|
|
offset = (offset + length) & ~PAGE_MASK;
|
|
|
|
rest -= length;
|
2007-12-19 15:09:18 +07:00
|
|
|
if (offset == 0)
|
|
|
|
page++;
|
2007-02-17 05:34:40 +07:00
|
|
|
}
|
|
|
|
|
2007-02-17 05:34:46 +07:00
|
|
|
return 0;
|
|
|
|
}
|
2007-02-17 05:34:50 +07:00
|
|
|
|
firewire: OHCI 1.0 Isochronous Receive support
Third rendition of FireWire OHCI 1.0 Isochronous Receive support, using a
zer-copy method similar to OHCI 1.1 which puts the IR data payload directly
into the userspace buffer. The zero-copy implementation eliminates the
video artifacts, audio popping, and buffer underrun problems seen with
version 1 of this patch, as well as fixing a regression in OHCI 1.1 support
introduced by version 2 of this patch.
Successfully tested in OHCI 1.1 mode on the following chipsets:
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
- Ti TSB41AB2 (rev 01), OHCI 1.1 (PCI on SB Audigy)
- Apple UniNorth 2 (rev 81), OHCI 1.1 (PowerBook G4 onboard)
Successfully tested in OHCI 1.0 mode on the following chipsets:
- Agere FW323 (rev 06), OHCI 1.0 (Mac Mini onboard)
- Agere FW323 (rev 06), OHCI 1.0 (PCI)
- Via VT6306 (rev 46), OHCI 1.0 (PCI)
- NEC OrangeLink (rev 01), OHCI 1.0 (PCI)
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
The bulk of testing was done in an x86_64 system, but was also successfully
sanity-tested on other systems, including a PPC(32) PowerBook G4 and an i686
EPIA M10k. Crude benchmarking (watching top during capture) puts the cpu
utilization during capture on the EPIA's 1GHz Via C3 processor around 13%,
which is down from 30% with the v1 code.
Some implementation details:
To maintain the same userspace API as dual-buffer mode, we set up two
descriptors for every incoming packet. The first is an INPUT_MORE descriptor,
pointing to a buffer large enough to hold just the packet's iso headers,
immediately followed by an INPUT_LAST descriptor, pointing to a chunk of the
userspace buffer big enough for the packet's data payload. With this setup,
each incoming packet fills in these two descriptors in a manner that very
closely emulates dual-buffer receive, to the point where the bulk of the
handle_ir_* code is now identical between the two (and probably primed for
some restructuring to share code between them).
The only caveat I have at the moment is that neither of my OHCI 1.0 Via
VT6307-based FireWire controllers work particularly well with this code
for reasons I have yet to figure out.
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2007-12-04 01:43:12 +07:00
|
|
|
static int
|
|
|
|
ohci_queue_iso_receive_packet_per_buffer(struct fw_iso_context *base,
|
|
|
|
struct fw_iso_packet *packet,
|
|
|
|
struct fw_iso_buffer *buffer,
|
|
|
|
unsigned long payload)
|
|
|
|
{
|
|
|
|
struct iso_context *ctx = container_of(base, struct iso_context, base);
|
|
|
|
struct descriptor *d = NULL, *pd = NULL;
|
2007-12-20 03:26:38 +07:00
|
|
|
struct fw_iso_packet *p = packet;
|
firewire: OHCI 1.0 Isochronous Receive support
Third rendition of FireWire OHCI 1.0 Isochronous Receive support, using a
zer-copy method similar to OHCI 1.1 which puts the IR data payload directly
into the userspace buffer. The zero-copy implementation eliminates the
video artifacts, audio popping, and buffer underrun problems seen with
version 1 of this patch, as well as fixing a regression in OHCI 1.1 support
introduced by version 2 of this patch.
Successfully tested in OHCI 1.1 mode on the following chipsets:
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
- Ti TSB41AB2 (rev 01), OHCI 1.1 (PCI on SB Audigy)
- Apple UniNorth 2 (rev 81), OHCI 1.1 (PowerBook G4 onboard)
Successfully tested in OHCI 1.0 mode on the following chipsets:
- Agere FW323 (rev 06), OHCI 1.0 (Mac Mini onboard)
- Agere FW323 (rev 06), OHCI 1.0 (PCI)
- Via VT6306 (rev 46), OHCI 1.0 (PCI)
- NEC OrangeLink (rev 01), OHCI 1.0 (PCI)
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
The bulk of testing was done in an x86_64 system, but was also successfully
sanity-tested on other systems, including a PPC(32) PowerBook G4 and an i686
EPIA M10k. Crude benchmarking (watching top during capture) puts the cpu
utilization during capture on the EPIA's 1GHz Via C3 processor around 13%,
which is down from 30% with the v1 code.
Some implementation details:
To maintain the same userspace API as dual-buffer mode, we set up two
descriptors for every incoming packet. The first is an INPUT_MORE descriptor,
pointing to a buffer large enough to hold just the packet's iso headers,
immediately followed by an INPUT_LAST descriptor, pointing to a chunk of the
userspace buffer big enough for the packet's data payload. With this setup,
each incoming packet fills in these two descriptors in a manner that very
closely emulates dual-buffer receive, to the point where the bulk of the
handle_ir_* code is now identical between the two (and probably primed for
some restructuring to share code between them).
The only caveat I have at the moment is that neither of my OHCI 1.0 Via
VT6307-based FireWire controllers work particularly well with this code
for reasons I have yet to figure out.
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2007-12-04 01:43:12 +07:00
|
|
|
dma_addr_t d_bus, page_bus;
|
|
|
|
u32 z, header_z, rest;
|
2007-12-20 03:26:38 +07:00
|
|
|
int i, j, length;
|
|
|
|
int page, offset, packet_count, header_size, payload_per_buffer;
|
firewire: OHCI 1.0 Isochronous Receive support
Third rendition of FireWire OHCI 1.0 Isochronous Receive support, using a
zer-copy method similar to OHCI 1.1 which puts the IR data payload directly
into the userspace buffer. The zero-copy implementation eliminates the
video artifacts, audio popping, and buffer underrun problems seen with
version 1 of this patch, as well as fixing a regression in OHCI 1.1 support
introduced by version 2 of this patch.
Successfully tested in OHCI 1.1 mode on the following chipsets:
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
- Ti TSB41AB2 (rev 01), OHCI 1.1 (PCI on SB Audigy)
- Apple UniNorth 2 (rev 81), OHCI 1.1 (PowerBook G4 onboard)
Successfully tested in OHCI 1.0 mode on the following chipsets:
- Agere FW323 (rev 06), OHCI 1.0 (Mac Mini onboard)
- Agere FW323 (rev 06), OHCI 1.0 (PCI)
- Via VT6306 (rev 46), OHCI 1.0 (PCI)
- NEC OrangeLink (rev 01), OHCI 1.0 (PCI)
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
The bulk of testing was done in an x86_64 system, but was also successfully
sanity-tested on other systems, including a PPC(32) PowerBook G4 and an i686
EPIA M10k. Crude benchmarking (watching top during capture) puts the cpu
utilization during capture on the EPIA's 1GHz Via C3 processor around 13%,
which is down from 30% with the v1 code.
Some implementation details:
To maintain the same userspace API as dual-buffer mode, we set up two
descriptors for every incoming packet. The first is an INPUT_MORE descriptor,
pointing to a buffer large enough to hold just the packet's iso headers,
immediately followed by an INPUT_LAST descriptor, pointing to a chunk of the
userspace buffer big enough for the packet's data payload. With this setup,
each incoming packet fills in these two descriptors in a manner that very
closely emulates dual-buffer receive, to the point where the bulk of the
handle_ir_* code is now identical between the two (and probably primed for
some restructuring to share code between them).
The only caveat I have at the moment is that neither of my OHCI 1.0 Via
VT6307-based FireWire controllers work particularly well with this code
for reasons I have yet to figure out.
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2007-12-04 01:43:12 +07:00
|
|
|
|
|
|
|
/*
|
|
|
|
* The OHCI controller puts the status word in the
|
|
|
|
* buffer too, so we need 4 extra bytes per packet.
|
|
|
|
*/
|
|
|
|
packet_count = p->header_length / ctx->base.header_size;
|
2007-12-20 03:26:38 +07:00
|
|
|
header_size = ctx->base.header_size + 4;
|
firewire: OHCI 1.0 Isochronous Receive support
Third rendition of FireWire OHCI 1.0 Isochronous Receive support, using a
zer-copy method similar to OHCI 1.1 which puts the IR data payload directly
into the userspace buffer. The zero-copy implementation eliminates the
video artifacts, audio popping, and buffer underrun problems seen with
version 1 of this patch, as well as fixing a regression in OHCI 1.1 support
introduced by version 2 of this patch.
Successfully tested in OHCI 1.1 mode on the following chipsets:
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
- Ti TSB41AB2 (rev 01), OHCI 1.1 (PCI on SB Audigy)
- Apple UniNorth 2 (rev 81), OHCI 1.1 (PowerBook G4 onboard)
Successfully tested in OHCI 1.0 mode on the following chipsets:
- Agere FW323 (rev 06), OHCI 1.0 (Mac Mini onboard)
- Agere FW323 (rev 06), OHCI 1.0 (PCI)
- Via VT6306 (rev 46), OHCI 1.0 (PCI)
- NEC OrangeLink (rev 01), OHCI 1.0 (PCI)
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
The bulk of testing was done in an x86_64 system, but was also successfully
sanity-tested on other systems, including a PPC(32) PowerBook G4 and an i686
EPIA M10k. Crude benchmarking (watching top during capture) puts the cpu
utilization during capture on the EPIA's 1GHz Via C3 processor around 13%,
which is down from 30% with the v1 code.
Some implementation details:
To maintain the same userspace API as dual-buffer mode, we set up two
descriptors for every incoming packet. The first is an INPUT_MORE descriptor,
pointing to a buffer large enough to hold just the packet's iso headers,
immediately followed by an INPUT_LAST descriptor, pointing to a chunk of the
userspace buffer big enough for the packet's data payload. With this setup,
each incoming packet fills in these two descriptors in a manner that very
closely emulates dual-buffer receive, to the point where the bulk of the
handle_ir_* code is now identical between the two (and probably primed for
some restructuring to share code between them).
The only caveat I have at the moment is that neither of my OHCI 1.0 Via
VT6307-based FireWire controllers work particularly well with this code
for reasons I have yet to figure out.
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2007-12-04 01:43:12 +07:00
|
|
|
|
|
|
|
/* Get header size in number of descriptors. */
|
|
|
|
header_z = DIV_ROUND_UP(header_size, sizeof(*d));
|
|
|
|
page = payload >> PAGE_SHIFT;
|
|
|
|
offset = payload & ~PAGE_MASK;
|
2007-12-20 03:26:38 +07:00
|
|
|
payload_per_buffer = p->payload_length / packet_count;
|
firewire: OHCI 1.0 Isochronous Receive support
Third rendition of FireWire OHCI 1.0 Isochronous Receive support, using a
zer-copy method similar to OHCI 1.1 which puts the IR data payload directly
into the userspace buffer. The zero-copy implementation eliminates the
video artifacts, audio popping, and buffer underrun problems seen with
version 1 of this patch, as well as fixing a regression in OHCI 1.1 support
introduced by version 2 of this patch.
Successfully tested in OHCI 1.1 mode on the following chipsets:
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
- Ti TSB41AB2 (rev 01), OHCI 1.1 (PCI on SB Audigy)
- Apple UniNorth 2 (rev 81), OHCI 1.1 (PowerBook G4 onboard)
Successfully tested in OHCI 1.0 mode on the following chipsets:
- Agere FW323 (rev 06), OHCI 1.0 (Mac Mini onboard)
- Agere FW323 (rev 06), OHCI 1.0 (PCI)
- Via VT6306 (rev 46), OHCI 1.0 (PCI)
- NEC OrangeLink (rev 01), OHCI 1.0 (PCI)
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
The bulk of testing was done in an x86_64 system, but was also successfully
sanity-tested on other systems, including a PPC(32) PowerBook G4 and an i686
EPIA M10k. Crude benchmarking (watching top during capture) puts the cpu
utilization during capture on the EPIA's 1GHz Via C3 processor around 13%,
which is down from 30% with the v1 code.
Some implementation details:
To maintain the same userspace API as dual-buffer mode, we set up two
descriptors for every incoming packet. The first is an INPUT_MORE descriptor,
pointing to a buffer large enough to hold just the packet's iso headers,
immediately followed by an INPUT_LAST descriptor, pointing to a chunk of the
userspace buffer big enough for the packet's data payload. With this setup,
each incoming packet fills in these two descriptors in a manner that very
closely emulates dual-buffer receive, to the point where the bulk of the
handle_ir_* code is now identical between the two (and probably primed for
some restructuring to share code between them).
The only caveat I have at the moment is that neither of my OHCI 1.0 Via
VT6307-based FireWire controllers work particularly well with this code
for reasons I have yet to figure out.
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2007-12-04 01:43:12 +07:00
|
|
|
|
|
|
|
for (i = 0; i < packet_count; i++) {
|
|
|
|
/* d points to the header descriptor */
|
2007-12-20 03:26:38 +07:00
|
|
|
z = DIV_ROUND_UP(payload_per_buffer + offset, PAGE_SIZE) + 1;
|
firewire: OHCI 1.0 Isochronous Receive support
Third rendition of FireWire OHCI 1.0 Isochronous Receive support, using a
zer-copy method similar to OHCI 1.1 which puts the IR data payload directly
into the userspace buffer. The zero-copy implementation eliminates the
video artifacts, audio popping, and buffer underrun problems seen with
version 1 of this patch, as well as fixing a regression in OHCI 1.1 support
introduced by version 2 of this patch.
Successfully tested in OHCI 1.1 mode on the following chipsets:
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
- Ti TSB41AB2 (rev 01), OHCI 1.1 (PCI on SB Audigy)
- Apple UniNorth 2 (rev 81), OHCI 1.1 (PowerBook G4 onboard)
Successfully tested in OHCI 1.0 mode on the following chipsets:
- Agere FW323 (rev 06), OHCI 1.0 (Mac Mini onboard)
- Agere FW323 (rev 06), OHCI 1.0 (PCI)
- Via VT6306 (rev 46), OHCI 1.0 (PCI)
- NEC OrangeLink (rev 01), OHCI 1.0 (PCI)
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
The bulk of testing was done in an x86_64 system, but was also successfully
sanity-tested on other systems, including a PPC(32) PowerBook G4 and an i686
EPIA M10k. Crude benchmarking (watching top during capture) puts the cpu
utilization during capture on the EPIA's 1GHz Via C3 processor around 13%,
which is down from 30% with the v1 code.
Some implementation details:
To maintain the same userspace API as dual-buffer mode, we set up two
descriptors for every incoming packet. The first is an INPUT_MORE descriptor,
pointing to a buffer large enough to hold just the packet's iso headers,
immediately followed by an INPUT_LAST descriptor, pointing to a chunk of the
userspace buffer big enough for the packet's data payload. With this setup,
each incoming packet fills in these two descriptors in a manner that very
closely emulates dual-buffer receive, to the point where the bulk of the
handle_ir_* code is now identical between the two (and probably primed for
some restructuring to share code between them).
The only caveat I have at the moment is that neither of my OHCI 1.0 Via
VT6307-based FireWire controllers work particularly well with this code
for reasons I have yet to figure out.
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2007-12-04 01:43:12 +07:00
|
|
|
d = context_get_descriptors(&ctx->context,
|
2007-12-20 03:26:38 +07:00
|
|
|
z + header_z, &d_bus);
|
firewire: OHCI 1.0 Isochronous Receive support
Third rendition of FireWire OHCI 1.0 Isochronous Receive support, using a
zer-copy method similar to OHCI 1.1 which puts the IR data payload directly
into the userspace buffer. The zero-copy implementation eliminates the
video artifacts, audio popping, and buffer underrun problems seen with
version 1 of this patch, as well as fixing a regression in OHCI 1.1 support
introduced by version 2 of this patch.
Successfully tested in OHCI 1.1 mode on the following chipsets:
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
- Ti TSB41AB2 (rev 01), OHCI 1.1 (PCI on SB Audigy)
- Apple UniNorth 2 (rev 81), OHCI 1.1 (PowerBook G4 onboard)
Successfully tested in OHCI 1.0 mode on the following chipsets:
- Agere FW323 (rev 06), OHCI 1.0 (Mac Mini onboard)
- Agere FW323 (rev 06), OHCI 1.0 (PCI)
- Via VT6306 (rev 46), OHCI 1.0 (PCI)
- NEC OrangeLink (rev 01), OHCI 1.0 (PCI)
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
The bulk of testing was done in an x86_64 system, but was also successfully
sanity-tested on other systems, including a PPC(32) PowerBook G4 and an i686
EPIA M10k. Crude benchmarking (watching top during capture) puts the cpu
utilization during capture on the EPIA's 1GHz Via C3 processor around 13%,
which is down from 30% with the v1 code.
Some implementation details:
To maintain the same userspace API as dual-buffer mode, we set up two
descriptors for every incoming packet. The first is an INPUT_MORE descriptor,
pointing to a buffer large enough to hold just the packet's iso headers,
immediately followed by an INPUT_LAST descriptor, pointing to a chunk of the
userspace buffer big enough for the packet's data payload. With this setup,
each incoming packet fills in these two descriptors in a manner that very
closely emulates dual-buffer receive, to the point where the bulk of the
handle_ir_* code is now identical between the two (and probably primed for
some restructuring to share code between them).
The only caveat I have at the moment is that neither of my OHCI 1.0 Via
VT6307-based FireWire controllers work particularly well with this code
for reasons I have yet to figure out.
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2007-12-04 01:43:12 +07:00
|
|
|
if (d == NULL)
|
|
|
|
return -ENOMEM;
|
|
|
|
|
2007-12-20 03:26:38 +07:00
|
|
|
d->control = cpu_to_le16(DESCRIPTOR_STATUS |
|
|
|
|
DESCRIPTOR_INPUT_MORE);
|
|
|
|
if (p->skip && i == 0)
|
|
|
|
d->control |= cpu_to_le16(DESCRIPTOR_WAIT);
|
firewire: OHCI 1.0 Isochronous Receive support
Third rendition of FireWire OHCI 1.0 Isochronous Receive support, using a
zer-copy method similar to OHCI 1.1 which puts the IR data payload directly
into the userspace buffer. The zero-copy implementation eliminates the
video artifacts, audio popping, and buffer underrun problems seen with
version 1 of this patch, as well as fixing a regression in OHCI 1.1 support
introduced by version 2 of this patch.
Successfully tested in OHCI 1.1 mode on the following chipsets:
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
- Ti TSB41AB2 (rev 01), OHCI 1.1 (PCI on SB Audigy)
- Apple UniNorth 2 (rev 81), OHCI 1.1 (PowerBook G4 onboard)
Successfully tested in OHCI 1.0 mode on the following chipsets:
- Agere FW323 (rev 06), OHCI 1.0 (Mac Mini onboard)
- Agere FW323 (rev 06), OHCI 1.0 (PCI)
- Via VT6306 (rev 46), OHCI 1.0 (PCI)
- NEC OrangeLink (rev 01), OHCI 1.0 (PCI)
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
The bulk of testing was done in an x86_64 system, but was also successfully
sanity-tested on other systems, including a PPC(32) PowerBook G4 and an i686
EPIA M10k. Crude benchmarking (watching top during capture) puts the cpu
utilization during capture on the EPIA's 1GHz Via C3 processor around 13%,
which is down from 30% with the v1 code.
Some implementation details:
To maintain the same userspace API as dual-buffer mode, we set up two
descriptors for every incoming packet. The first is an INPUT_MORE descriptor,
pointing to a buffer large enough to hold just the packet's iso headers,
immediately followed by an INPUT_LAST descriptor, pointing to a chunk of the
userspace buffer big enough for the packet's data payload. With this setup,
each incoming packet fills in these two descriptors in a manner that very
closely emulates dual-buffer receive, to the point where the bulk of the
handle_ir_* code is now identical between the two (and probably primed for
some restructuring to share code between them).
The only caveat I have at the moment is that neither of my OHCI 1.0 Via
VT6307-based FireWire controllers work particularly well with this code
for reasons I have yet to figure out.
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2007-12-04 01:43:12 +07:00
|
|
|
d->req_count = cpu_to_le16(header_size);
|
|
|
|
d->res_count = d->req_count;
|
2007-12-20 03:26:38 +07:00
|
|
|
d->transfer_status = 0;
|
firewire: OHCI 1.0 Isochronous Receive support
Third rendition of FireWire OHCI 1.0 Isochronous Receive support, using a
zer-copy method similar to OHCI 1.1 which puts the IR data payload directly
into the userspace buffer. The zero-copy implementation eliminates the
video artifacts, audio popping, and buffer underrun problems seen with
version 1 of this patch, as well as fixing a regression in OHCI 1.1 support
introduced by version 2 of this patch.
Successfully tested in OHCI 1.1 mode on the following chipsets:
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
- Ti TSB41AB2 (rev 01), OHCI 1.1 (PCI on SB Audigy)
- Apple UniNorth 2 (rev 81), OHCI 1.1 (PowerBook G4 onboard)
Successfully tested in OHCI 1.0 mode on the following chipsets:
- Agere FW323 (rev 06), OHCI 1.0 (Mac Mini onboard)
- Agere FW323 (rev 06), OHCI 1.0 (PCI)
- Via VT6306 (rev 46), OHCI 1.0 (PCI)
- NEC OrangeLink (rev 01), OHCI 1.0 (PCI)
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
The bulk of testing was done in an x86_64 system, but was also successfully
sanity-tested on other systems, including a PPC(32) PowerBook G4 and an i686
EPIA M10k. Crude benchmarking (watching top during capture) puts the cpu
utilization during capture on the EPIA's 1GHz Via C3 processor around 13%,
which is down from 30% with the v1 code.
Some implementation details:
To maintain the same userspace API as dual-buffer mode, we set up two
descriptors for every incoming packet. The first is an INPUT_MORE descriptor,
pointing to a buffer large enough to hold just the packet's iso headers,
immediately followed by an INPUT_LAST descriptor, pointing to a chunk of the
userspace buffer big enough for the packet's data payload. With this setup,
each incoming packet fills in these two descriptors in a manner that very
closely emulates dual-buffer receive, to the point where the bulk of the
handle_ir_* code is now identical between the two (and probably primed for
some restructuring to share code between them).
The only caveat I have at the moment is that neither of my OHCI 1.0 Via
VT6307-based FireWire controllers work particularly well with this code
for reasons I have yet to figure out.
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2007-12-04 01:43:12 +07:00
|
|
|
d->data_address = cpu_to_le32(d_bus + (z * sizeof(*d)));
|
|
|
|
|
2007-12-20 03:26:38 +07:00
|
|
|
rest = payload_per_buffer;
|
|
|
|
for (j = 1; j < z; j++) {
|
|
|
|
pd = d + j;
|
|
|
|
pd->control = cpu_to_le16(DESCRIPTOR_STATUS |
|
|
|
|
DESCRIPTOR_INPUT_MORE);
|
|
|
|
|
|
|
|
if (offset + rest < PAGE_SIZE)
|
|
|
|
length = rest;
|
|
|
|
else
|
|
|
|
length = PAGE_SIZE - offset;
|
|
|
|
pd->req_count = cpu_to_le16(length);
|
|
|
|
pd->res_count = pd->req_count;
|
|
|
|
pd->transfer_status = 0;
|
|
|
|
|
|
|
|
page_bus = page_private(buffer->pages[page]);
|
|
|
|
pd->data_address = cpu_to_le32(page_bus + offset);
|
|
|
|
|
|
|
|
offset = (offset + length) & ~PAGE_MASK;
|
|
|
|
rest -= length;
|
|
|
|
if (offset == 0)
|
|
|
|
page++;
|
|
|
|
}
|
firewire: OHCI 1.0 Isochronous Receive support
Third rendition of FireWire OHCI 1.0 Isochronous Receive support, using a
zer-copy method similar to OHCI 1.1 which puts the IR data payload directly
into the userspace buffer. The zero-copy implementation eliminates the
video artifacts, audio popping, and buffer underrun problems seen with
version 1 of this patch, as well as fixing a regression in OHCI 1.1 support
introduced by version 2 of this patch.
Successfully tested in OHCI 1.1 mode on the following chipsets:
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
- Ti TSB41AB2 (rev 01), OHCI 1.1 (PCI on SB Audigy)
- Apple UniNorth 2 (rev 81), OHCI 1.1 (PowerBook G4 onboard)
Successfully tested in OHCI 1.0 mode on the following chipsets:
- Agere FW323 (rev 06), OHCI 1.0 (Mac Mini onboard)
- Agere FW323 (rev 06), OHCI 1.0 (PCI)
- Via VT6306 (rev 46), OHCI 1.0 (PCI)
- NEC OrangeLink (rev 01), OHCI 1.0 (PCI)
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
The bulk of testing was done in an x86_64 system, but was also successfully
sanity-tested on other systems, including a PPC(32) PowerBook G4 and an i686
EPIA M10k. Crude benchmarking (watching top during capture) puts the cpu
utilization during capture on the EPIA's 1GHz Via C3 processor around 13%,
which is down from 30% with the v1 code.
Some implementation details:
To maintain the same userspace API as dual-buffer mode, we set up two
descriptors for every incoming packet. The first is an INPUT_MORE descriptor,
pointing to a buffer large enough to hold just the packet's iso headers,
immediately followed by an INPUT_LAST descriptor, pointing to a chunk of the
userspace buffer big enough for the packet's data payload. With this setup,
each incoming packet fills in these two descriptors in a manner that very
closely emulates dual-buffer receive, to the point where the bulk of the
handle_ir_* code is now identical between the two (and probably primed for
some restructuring to share code between them).
The only caveat I have at the moment is that neither of my OHCI 1.0 Via
VT6307-based FireWire controllers work particularly well with this code
for reasons I have yet to figure out.
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2007-12-04 01:43:12 +07:00
|
|
|
pd->control = cpu_to_le16(DESCRIPTOR_STATUS |
|
|
|
|
DESCRIPTOR_INPUT_LAST |
|
|
|
|
DESCRIPTOR_BRANCH_ALWAYS);
|
2007-12-20 03:26:38 +07:00
|
|
|
if (p->interrupt && i == packet_count - 1)
|
firewire: OHCI 1.0 Isochronous Receive support
Third rendition of FireWire OHCI 1.0 Isochronous Receive support, using a
zer-copy method similar to OHCI 1.1 which puts the IR data payload directly
into the userspace buffer. The zero-copy implementation eliminates the
video artifacts, audio popping, and buffer underrun problems seen with
version 1 of this patch, as well as fixing a regression in OHCI 1.1 support
introduced by version 2 of this patch.
Successfully tested in OHCI 1.1 mode on the following chipsets:
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
- Ti TSB41AB2 (rev 01), OHCI 1.1 (PCI on SB Audigy)
- Apple UniNorth 2 (rev 81), OHCI 1.1 (PowerBook G4 onboard)
Successfully tested in OHCI 1.0 mode on the following chipsets:
- Agere FW323 (rev 06), OHCI 1.0 (Mac Mini onboard)
- Agere FW323 (rev 06), OHCI 1.0 (PCI)
- Via VT6306 (rev 46), OHCI 1.0 (PCI)
- NEC OrangeLink (rev 01), OHCI 1.0 (PCI)
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
The bulk of testing was done in an x86_64 system, but was also successfully
sanity-tested on other systems, including a PPC(32) PowerBook G4 and an i686
EPIA M10k. Crude benchmarking (watching top during capture) puts the cpu
utilization during capture on the EPIA's 1GHz Via C3 processor around 13%,
which is down from 30% with the v1 code.
Some implementation details:
To maintain the same userspace API as dual-buffer mode, we set up two
descriptors for every incoming packet. The first is an INPUT_MORE descriptor,
pointing to a buffer large enough to hold just the packet's iso headers,
immediately followed by an INPUT_LAST descriptor, pointing to a chunk of the
userspace buffer big enough for the packet's data payload. With this setup,
each incoming packet fills in these two descriptors in a manner that very
closely emulates dual-buffer receive, to the point where the bulk of the
handle_ir_* code is now identical between the two (and probably primed for
some restructuring to share code between them).
The only caveat I have at the moment is that neither of my OHCI 1.0 Via
VT6307-based FireWire controllers work particularly well with this code
for reasons I have yet to figure out.
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2007-12-04 01:43:12 +07:00
|
|
|
pd->control |= cpu_to_le16(DESCRIPTOR_IRQ_ALWAYS);
|
|
|
|
|
|
|
|
context_append(&ctx->context, d, z, header_z);
|
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2007-02-17 05:34:40 +07:00
|
|
|
static int
|
|
|
|
ohci_queue_iso(struct fw_iso_context *base,
|
|
|
|
struct fw_iso_packet *packet,
|
|
|
|
struct fw_iso_buffer *buffer,
|
|
|
|
unsigned long payload)
|
|
|
|
{
|
2007-02-17 05:34:49 +07:00
|
|
|
struct iso_context *ctx = container_of(base, struct iso_context, base);
|
firewire: fw-ohci: Dynamically allocate buffers for DMA descriptors
Previously, the fw-ohci driver used fixed-length buffers for storing
descriptors for isochronous receive DMA programs. If an application
(such as libdc1394) generated a DMA program that was too large, fw-ohci
would reach the limit of its fixed-sized buffer and return an error to
userspace.
This patch replaces the fixed-length ring-buffer with a linked-list of
page-sized buffers. Additional buffers can be dynamically allocated and
appended to the list when necessary. For a particular context, buffers
are kept around after use and reused as necessary, so there is no
allocation taking place after the DMA program is generated for the first
time.
In addition, the buffers it uses are coherent for DMA so there is no
syncing required before and after writes. This syncing wasn't properly
done in the previous version of the code.
-
This is the fourth version of my patch that replaces a fixed-length
buffer for DMA descriptors with a dynamically allocated linked-list of
buffers.
As we discovered with the last attempt, new context programs are
sometimes queued from interrupt context, making it unacceptable to call
tasklet_disable() from context_get_descriptors().
This version of the patch uses ohci->lock for all locking needs instead
of tasklet_disable/enable. There is a new requirement that
context_get_descriptors() be called while holding ohci->lock. It was
already held for the AT context, so adding the requirement for the iso
context did not seem particularly onerous. In addition, this has the
side benefit of allowing iso queue to be safely called from concurrent
user-space threads, which previously was not safe.
Signed-off-by: David Moore <dcm@acm.org>
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
-
Fixes the following issues:
- Isochronous reception stopped prematurely if an application used a
larger buffer. (Reproduced with coriander.)
- Isochronous reception stopped after one or a few frames on VT630x
in OHCI 1.0 mode. (Fixes reception in coriander, but dvgrab still
doesn't work with these chips.)
Patch update: struct member alignment, whitespace nits
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2008-01-07 05:21:41 +07:00
|
|
|
unsigned long flags;
|
|
|
|
int retval;
|
2007-02-17 05:34:49 +07:00
|
|
|
|
firewire: fw-ohci: Dynamically allocate buffers for DMA descriptors
Previously, the fw-ohci driver used fixed-length buffers for storing
descriptors for isochronous receive DMA programs. If an application
(such as libdc1394) generated a DMA program that was too large, fw-ohci
would reach the limit of its fixed-sized buffer and return an error to
userspace.
This patch replaces the fixed-length ring-buffer with a linked-list of
page-sized buffers. Additional buffers can be dynamically allocated and
appended to the list when necessary. For a particular context, buffers
are kept around after use and reused as necessary, so there is no
allocation taking place after the DMA program is generated for the first
time.
In addition, the buffers it uses are coherent for DMA so there is no
syncing required before and after writes. This syncing wasn't properly
done in the previous version of the code.
-
This is the fourth version of my patch that replaces a fixed-length
buffer for DMA descriptors with a dynamically allocated linked-list of
buffers.
As we discovered with the last attempt, new context programs are
sometimes queued from interrupt context, making it unacceptable to call
tasklet_disable() from context_get_descriptors().
This version of the patch uses ohci->lock for all locking needs instead
of tasklet_disable/enable. There is a new requirement that
context_get_descriptors() be called while holding ohci->lock. It was
already held for the AT context, so adding the requirement for the iso
context did not seem particularly onerous. In addition, this has the
side benefit of allowing iso queue to be safely called from concurrent
user-space threads, which previously was not safe.
Signed-off-by: David Moore <dcm@acm.org>
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
-
Fixes the following issues:
- Isochronous reception stopped prematurely if an application used a
larger buffer. (Reproduced with coriander.)
- Isochronous reception stopped after one or a few frames on VT630x
in OHCI 1.0 mode. (Fixes reception in coriander, but dvgrab still
doesn't work with these chips.)
Patch update: struct member alignment, whitespace nits
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2008-01-07 05:21:41 +07:00
|
|
|
spin_lock_irqsave(&ctx->context.ohci->lock, flags);
|
2007-02-17 05:34:40 +07:00
|
|
|
if (base->type == FW_ISO_CONTEXT_TRANSMIT)
|
firewire: fw-ohci: Dynamically allocate buffers for DMA descriptors
Previously, the fw-ohci driver used fixed-length buffers for storing
descriptors for isochronous receive DMA programs. If an application
(such as libdc1394) generated a DMA program that was too large, fw-ohci
would reach the limit of its fixed-sized buffer and return an error to
userspace.
This patch replaces the fixed-length ring-buffer with a linked-list of
page-sized buffers. Additional buffers can be dynamically allocated and
appended to the list when necessary. For a particular context, buffers
are kept around after use and reused as necessary, so there is no
allocation taking place after the DMA program is generated for the first
time.
In addition, the buffers it uses are coherent for DMA so there is no
syncing required before and after writes. This syncing wasn't properly
done in the previous version of the code.
-
This is the fourth version of my patch that replaces a fixed-length
buffer for DMA descriptors with a dynamically allocated linked-list of
buffers.
As we discovered with the last attempt, new context programs are
sometimes queued from interrupt context, making it unacceptable to call
tasklet_disable() from context_get_descriptors().
This version of the patch uses ohci->lock for all locking needs instead
of tasklet_disable/enable. There is a new requirement that
context_get_descriptors() be called while holding ohci->lock. It was
already held for the AT context, so adding the requirement for the iso
context did not seem particularly onerous. In addition, this has the
side benefit of allowing iso queue to be safely called from concurrent
user-space threads, which previously was not safe.
Signed-off-by: David Moore <dcm@acm.org>
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
-
Fixes the following issues:
- Isochronous reception stopped prematurely if an application used a
larger buffer. (Reproduced with coriander.)
- Isochronous reception stopped after one or a few frames on VT630x
in OHCI 1.0 mode. (Fixes reception in coriander, but dvgrab still
doesn't work with these chips.)
Patch update: struct member alignment, whitespace nits
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2008-01-07 05:21:41 +07:00
|
|
|
retval = ohci_queue_iso_transmit(base, packet, buffer, payload);
|
2007-02-17 05:34:49 +07:00
|
|
|
else if (ctx->context.ohci->version >= OHCI_VERSION_1_1)
|
firewire: fw-ohci: Dynamically allocate buffers for DMA descriptors
Previously, the fw-ohci driver used fixed-length buffers for storing
descriptors for isochronous receive DMA programs. If an application
(such as libdc1394) generated a DMA program that was too large, fw-ohci
would reach the limit of its fixed-sized buffer and return an error to
userspace.
This patch replaces the fixed-length ring-buffer with a linked-list of
page-sized buffers. Additional buffers can be dynamically allocated and
appended to the list when necessary. For a particular context, buffers
are kept around after use and reused as necessary, so there is no
allocation taking place after the DMA program is generated for the first
time.
In addition, the buffers it uses are coherent for DMA so there is no
syncing required before and after writes. This syncing wasn't properly
done in the previous version of the code.
-
This is the fourth version of my patch that replaces a fixed-length
buffer for DMA descriptors with a dynamically allocated linked-list of
buffers.
As we discovered with the last attempt, new context programs are
sometimes queued from interrupt context, making it unacceptable to call
tasklet_disable() from context_get_descriptors().
This version of the patch uses ohci->lock for all locking needs instead
of tasklet_disable/enable. There is a new requirement that
context_get_descriptors() be called while holding ohci->lock. It was
already held for the AT context, so adding the requirement for the iso
context did not seem particularly onerous. In addition, this has the
side benefit of allowing iso queue to be safely called from concurrent
user-space threads, which previously was not safe.
Signed-off-by: David Moore <dcm@acm.org>
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
-
Fixes the following issues:
- Isochronous reception stopped prematurely if an application used a
larger buffer. (Reproduced with coriander.)
- Isochronous reception stopped after one or a few frames on VT630x
in OHCI 1.0 mode. (Fixes reception in coriander, but dvgrab still
doesn't work with these chips.)
Patch update: struct member alignment, whitespace nits
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2008-01-07 05:21:41 +07:00
|
|
|
retval = ohci_queue_iso_receive_dualbuffer(base, packet,
|
2007-02-17 05:34:46 +07:00
|
|
|
buffer, payload);
|
2007-02-17 05:34:49 +07:00
|
|
|
else
|
firewire: fw-ohci: Dynamically allocate buffers for DMA descriptors
Previously, the fw-ohci driver used fixed-length buffers for storing
descriptors for isochronous receive DMA programs. If an application
(such as libdc1394) generated a DMA program that was too large, fw-ohci
would reach the limit of its fixed-sized buffer and return an error to
userspace.
This patch replaces the fixed-length ring-buffer with a linked-list of
page-sized buffers. Additional buffers can be dynamically allocated and
appended to the list when necessary. For a particular context, buffers
are kept around after use and reused as necessary, so there is no
allocation taking place after the DMA program is generated for the first
time.
In addition, the buffers it uses are coherent for DMA so there is no
syncing required before and after writes. This syncing wasn't properly
done in the previous version of the code.
-
This is the fourth version of my patch that replaces a fixed-length
buffer for DMA descriptors with a dynamically allocated linked-list of
buffers.
As we discovered with the last attempt, new context programs are
sometimes queued from interrupt context, making it unacceptable to call
tasklet_disable() from context_get_descriptors().
This version of the patch uses ohci->lock for all locking needs instead
of tasklet_disable/enable. There is a new requirement that
context_get_descriptors() be called while holding ohci->lock. It was
already held for the AT context, so adding the requirement for the iso
context did not seem particularly onerous. In addition, this has the
side benefit of allowing iso queue to be safely called from concurrent
user-space threads, which previously was not safe.
Signed-off-by: David Moore <dcm@acm.org>
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
-
Fixes the following issues:
- Isochronous reception stopped prematurely if an application used a
larger buffer. (Reproduced with coriander.)
- Isochronous reception stopped after one or a few frames on VT630x
in OHCI 1.0 mode. (Fixes reception in coriander, but dvgrab still
doesn't work with these chips.)
Patch update: struct member alignment, whitespace nits
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2008-01-07 05:21:41 +07:00
|
|
|
retval = ohci_queue_iso_receive_packet_per_buffer(base, packet,
|
firewire: OHCI 1.0 Isochronous Receive support
Third rendition of FireWire OHCI 1.0 Isochronous Receive support, using a
zer-copy method similar to OHCI 1.1 which puts the IR data payload directly
into the userspace buffer. The zero-copy implementation eliminates the
video artifacts, audio popping, and buffer underrun problems seen with
version 1 of this patch, as well as fixing a regression in OHCI 1.1 support
introduced by version 2 of this patch.
Successfully tested in OHCI 1.1 mode on the following chipsets:
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
- Ti TSB41AB2 (rev 01), OHCI 1.1 (PCI on SB Audigy)
- Apple UniNorth 2 (rev 81), OHCI 1.1 (PowerBook G4 onboard)
Successfully tested in OHCI 1.0 mode on the following chipsets:
- Agere FW323 (rev 06), OHCI 1.0 (Mac Mini onboard)
- Agere FW323 (rev 06), OHCI 1.0 (PCI)
- Via VT6306 (rev 46), OHCI 1.0 (PCI)
- NEC OrangeLink (rev 01), OHCI 1.0 (PCI)
- NEC uPD72847 (rev 01), OHCI 1.1 (PCI)
- Ti XIO2200(A) (rev 01), OHCI 1.1 (PCIe)
The bulk of testing was done in an x86_64 system, but was also successfully
sanity-tested on other systems, including a PPC(32) PowerBook G4 and an i686
EPIA M10k. Crude benchmarking (watching top during capture) puts the cpu
utilization during capture on the EPIA's 1GHz Via C3 processor around 13%,
which is down from 30% with the v1 code.
Some implementation details:
To maintain the same userspace API as dual-buffer mode, we set up two
descriptors for every incoming packet. The first is an INPUT_MORE descriptor,
pointing to a buffer large enough to hold just the packet's iso headers,
immediately followed by an INPUT_LAST descriptor, pointing to a chunk of the
userspace buffer big enough for the packet's data payload. With this setup,
each incoming packet fills in these two descriptors in a manner that very
closely emulates dual-buffer receive, to the point where the bulk of the
handle_ir_* code is now identical between the two (and probably primed for
some restructuring to share code between them).
The only caveat I have at the moment is that neither of my OHCI 1.0 Via
VT6307-based FireWire controllers work particularly well with this code
for reasons I have yet to figure out.
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2007-12-04 01:43:12 +07:00
|
|
|
buffer,
|
|
|
|
payload);
|
firewire: fw-ohci: Dynamically allocate buffers for DMA descriptors
Previously, the fw-ohci driver used fixed-length buffers for storing
descriptors for isochronous receive DMA programs. If an application
(such as libdc1394) generated a DMA program that was too large, fw-ohci
would reach the limit of its fixed-sized buffer and return an error to
userspace.
This patch replaces the fixed-length ring-buffer with a linked-list of
page-sized buffers. Additional buffers can be dynamically allocated and
appended to the list when necessary. For a particular context, buffers
are kept around after use and reused as necessary, so there is no
allocation taking place after the DMA program is generated for the first
time.
In addition, the buffers it uses are coherent for DMA so there is no
syncing required before and after writes. This syncing wasn't properly
done in the previous version of the code.
-
This is the fourth version of my patch that replaces a fixed-length
buffer for DMA descriptors with a dynamically allocated linked-list of
buffers.
As we discovered with the last attempt, new context programs are
sometimes queued from interrupt context, making it unacceptable to call
tasklet_disable() from context_get_descriptors().
This version of the patch uses ohci->lock for all locking needs instead
of tasklet_disable/enable. There is a new requirement that
context_get_descriptors() be called while holding ohci->lock. It was
already held for the AT context, so adding the requirement for the iso
context did not seem particularly onerous. In addition, this has the
side benefit of allowing iso queue to be safely called from concurrent
user-space threads, which previously was not safe.
Signed-off-by: David Moore <dcm@acm.org>
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
-
Fixes the following issues:
- Isochronous reception stopped prematurely if an application used a
larger buffer. (Reproduced with coriander.)
- Isochronous reception stopped after one or a few frames on VT630x
in OHCI 1.0 mode. (Fixes reception in coriander, but dvgrab still
doesn't work with these chips.)
Patch update: struct member alignment, whitespace nits
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2008-01-07 05:21:41 +07:00
|
|
|
spin_unlock_irqrestore(&ctx->context.ohci->lock, flags);
|
|
|
|
|
|
|
|
return retval;
|
2007-02-17 05:34:40 +07:00
|
|
|
}
|
|
|
|
|
2007-01-14 21:29:07 +07:00
|
|
|
static const struct fw_card_driver ohci_driver = {
|
2006-12-20 07:58:35 +07:00
|
|
|
.name = ohci_driver_name,
|
|
|
|
.enable = ohci_enable,
|
|
|
|
.update_phy_reg = ohci_update_phy_reg,
|
|
|
|
.set_config_rom = ohci_set_config_rom,
|
|
|
|
.send_request = ohci_send_request,
|
|
|
|
.send_response = ohci_send_response,
|
2007-02-07 02:49:32 +07:00
|
|
|
.cancel_packet = ohci_cancel_packet,
|
2006-12-20 07:58:35 +07:00
|
|
|
.enable_phys_dma = ohci_enable_phys_dma,
|
2007-03-08 00:12:56 +07:00
|
|
|
.get_bus_time = ohci_get_bus_time,
|
2006-12-20 07:58:35 +07:00
|
|
|
|
|
|
|
.allocate_iso_context = ohci_allocate_iso_context,
|
|
|
|
.free_iso_context = ohci_free_iso_context,
|
|
|
|
.queue_iso = ohci_queue_iso,
|
2007-02-17 05:34:41 +07:00
|
|
|
.start_iso = ohci_start_iso,
|
2007-02-17 05:34:42 +07:00
|
|
|
.stop_iso = ohci_stop_iso,
|
2006-12-20 07:58:35 +07:00
|
|
|
};
|
|
|
|
|
2008-03-01 08:42:56 +07:00
|
|
|
#ifdef CONFIG_PPC_PMAC
|
2008-03-01 18:35:29 +07:00
|
|
|
static void ohci_pmac_on(struct pci_dev *dev)
|
|
|
|
{
|
2008-03-01 08:42:56 +07:00
|
|
|
if (machine_is(powermac)) {
|
|
|
|
struct device_node *ofn = pci_device_to_OF_node(dev);
|
|
|
|
|
|
|
|
if (ofn) {
|
|
|
|
pmac_call_feature(PMAC_FTR_1394_CABLE_POWER, ofn, 0, 1);
|
|
|
|
pmac_call_feature(PMAC_FTR_1394_ENABLE, ofn, 0, 1);
|
|
|
|
}
|
|
|
|
}
|
2008-03-01 18:35:29 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
static void ohci_pmac_off(struct pci_dev *dev)
|
|
|
|
{
|
|
|
|
if (machine_is(powermac)) {
|
|
|
|
struct device_node *ofn = pci_device_to_OF_node(dev);
|
|
|
|
|
|
|
|
if (ofn) {
|
|
|
|
pmac_call_feature(PMAC_FTR_1394_ENABLE, ofn, 0, 0);
|
|
|
|
pmac_call_feature(PMAC_FTR_1394_CABLE_POWER, ofn, 0, 0);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
#else
|
|
|
|
#define ohci_pmac_on(dev)
|
|
|
|
#define ohci_pmac_off(dev)
|
2008-03-01 08:42:56 +07:00
|
|
|
#endif /* CONFIG_PPC_PMAC */
|
|
|
|
|
2008-03-01 18:35:29 +07:00
|
|
|
static int __devinit
|
|
|
|
pci_probe(struct pci_dev *dev, const struct pci_device_id *ent)
|
|
|
|
{
|
|
|
|
struct fw_ohci *ohci;
|
|
|
|
u32 bus_options, max_receive, link_speed;
|
|
|
|
u64 guid;
|
|
|
|
int err;
|
|
|
|
size_t size;
|
|
|
|
|
|
|
|
ohci_pmac_on(dev);
|
|
|
|
|
2007-05-10 06:23:14 +07:00
|
|
|
ohci = kzalloc(sizeof(*ohci), GFP_KERNEL);
|
2006-12-20 07:58:35 +07:00
|
|
|
if (ohci == NULL) {
|
|
|
|
fw_error("Could not malloc fw_ohci data.\n");
|
|
|
|
return -ENOMEM;
|
|
|
|
}
|
|
|
|
|
|
|
|
fw_card_initialize(&ohci->card, &ohci_driver, &dev->dev);
|
|
|
|
|
2007-05-10 06:23:15 +07:00
|
|
|
err = pci_enable_device(dev);
|
|
|
|
if (err) {
|
2006-12-20 07:58:35 +07:00
|
|
|
fw_error("Failed to enable OHCI hardware.\n");
|
2008-02-25 00:59:55 +07:00
|
|
|
goto fail_free;
|
2006-12-20 07:58:35 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
pci_set_master(dev);
|
|
|
|
pci_write_config_dword(dev, OHCI1394_PCI_HCI_Control, 0);
|
|
|
|
pci_set_drvdata(dev, ohci);
|
|
|
|
|
2008-03-01 08:47:15 +07:00
|
|
|
#if defined(CONFIG_PPC_PMAC) && defined(CONFIG_PPC32)
|
|
|
|
ohci->old_uninorth = dev->vendor == PCI_VENDOR_ID_APPLE &&
|
|
|
|
dev->device == PCI_DEVICE_ID_APPLE_UNI_N_FW;
|
|
|
|
#endif
|
2006-12-20 07:58:35 +07:00
|
|
|
spin_lock_init(&ohci->lock);
|
|
|
|
|
|
|
|
tasklet_init(&ohci->bus_reset_tasklet,
|
|
|
|
bus_reset_tasklet, (unsigned long)ohci);
|
|
|
|
|
2007-05-10 06:23:15 +07:00
|
|
|
err = pci_request_region(dev, 0, ohci_driver_name);
|
|
|
|
if (err) {
|
2006-12-20 07:58:35 +07:00
|
|
|
fw_error("MMIO resource unavailable\n");
|
2007-05-10 06:23:15 +07:00
|
|
|
goto fail_disable;
|
2006-12-20 07:58:35 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
ohci->registers = pci_iomap(dev, 0, OHCI1394_REGISTER_SIZE);
|
|
|
|
if (ohci->registers == NULL) {
|
|
|
|
fw_error("Failed to remap registers\n");
|
2007-05-10 06:23:15 +07:00
|
|
|
err = -ENXIO;
|
|
|
|
goto fail_iomem;
|
2006-12-20 07:58:35 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
ar_context_init(&ohci->ar_request_ctx, ohci,
|
|
|
|
OHCI1394_AsReqRcvContextControlSet);
|
|
|
|
|
|
|
|
ar_context_init(&ohci->ar_response_ctx, ohci,
|
|
|
|
OHCI1394_AsRspRcvContextControlSet);
|
|
|
|
|
firewire: fw-ohci: Dynamically allocate buffers for DMA descriptors
Previously, the fw-ohci driver used fixed-length buffers for storing
descriptors for isochronous receive DMA programs. If an application
(such as libdc1394) generated a DMA program that was too large, fw-ohci
would reach the limit of its fixed-sized buffer and return an error to
userspace.
This patch replaces the fixed-length ring-buffer with a linked-list of
page-sized buffers. Additional buffers can be dynamically allocated and
appended to the list when necessary. For a particular context, buffers
are kept around after use and reused as necessary, so there is no
allocation taking place after the DMA program is generated for the first
time.
In addition, the buffers it uses are coherent for DMA so there is no
syncing required before and after writes. This syncing wasn't properly
done in the previous version of the code.
-
This is the fourth version of my patch that replaces a fixed-length
buffer for DMA descriptors with a dynamically allocated linked-list of
buffers.
As we discovered with the last attempt, new context programs are
sometimes queued from interrupt context, making it unacceptable to call
tasklet_disable() from context_get_descriptors().
This version of the patch uses ohci->lock for all locking needs instead
of tasklet_disable/enable. There is a new requirement that
context_get_descriptors() be called while holding ohci->lock. It was
already held for the AT context, so adding the requirement for the iso
context did not seem particularly onerous. In addition, this has the
side benefit of allowing iso queue to be safely called from concurrent
user-space threads, which previously was not safe.
Signed-off-by: David Moore <dcm@acm.org>
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
-
Fixes the following issues:
- Isochronous reception stopped prematurely if an application used a
larger buffer. (Reproduced with coriander.)
- Isochronous reception stopped after one or a few frames on VT630x
in OHCI 1.0 mode. (Fixes reception in coriander, but dvgrab still
doesn't work with these chips.)
Patch update: struct member alignment, whitespace nits
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2008-01-07 05:21:41 +07:00
|
|
|
context_init(&ohci->at_request_ctx, ohci,
|
2007-03-08 00:12:49 +07:00
|
|
|
OHCI1394_AsReqTrContextControlSet, handle_at_packet);
|
2006-12-20 07:58:35 +07:00
|
|
|
|
firewire: fw-ohci: Dynamically allocate buffers for DMA descriptors
Previously, the fw-ohci driver used fixed-length buffers for storing
descriptors for isochronous receive DMA programs. If an application
(such as libdc1394) generated a DMA program that was too large, fw-ohci
would reach the limit of its fixed-sized buffer and return an error to
userspace.
This patch replaces the fixed-length ring-buffer with a linked-list of
page-sized buffers. Additional buffers can be dynamically allocated and
appended to the list when necessary. For a particular context, buffers
are kept around after use and reused as necessary, so there is no
allocation taking place after the DMA program is generated for the first
time.
In addition, the buffers it uses are coherent for DMA so there is no
syncing required before and after writes. This syncing wasn't properly
done in the previous version of the code.
-
This is the fourth version of my patch that replaces a fixed-length
buffer for DMA descriptors with a dynamically allocated linked-list of
buffers.
As we discovered with the last attempt, new context programs are
sometimes queued from interrupt context, making it unacceptable to call
tasklet_disable() from context_get_descriptors().
This version of the patch uses ohci->lock for all locking needs instead
of tasklet_disable/enable. There is a new requirement that
context_get_descriptors() be called while holding ohci->lock. It was
already held for the AT context, so adding the requirement for the iso
context did not seem particularly onerous. In addition, this has the
side benefit of allowing iso queue to be safely called from concurrent
user-space threads, which previously was not safe.
Signed-off-by: David Moore <dcm@acm.org>
Signed-off-by: Kristian Høgsberg <krh@redhat.com>
Signed-off-by: Jarod Wilson <jwilson@redhat.com>
-
Fixes the following issues:
- Isochronous reception stopped prematurely if an application used a
larger buffer. (Reproduced with coriander.)
- Isochronous reception stopped after one or a few frames on VT630x
in OHCI 1.0 mode. (Fixes reception in coriander, but dvgrab still
doesn't work with these chips.)
Patch update: struct member alignment, whitespace nits
Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2008-01-07 05:21:41 +07:00
|
|
|
context_init(&ohci->at_response_ctx, ohci,
|
2007-03-08 00:12:49 +07:00
|
|
|
OHCI1394_AsRspTrContextControlSet, handle_at_packet);
|
2006-12-20 07:58:35 +07:00
|
|
|
|
|
|
|
reg_write(ohci, OHCI1394_IsoRecvIntMaskSet, ~0);
|
|
|
|
ohci->it_context_mask = reg_read(ohci, OHCI1394_IsoRecvIntMaskSet);
|
|
|
|
reg_write(ohci, OHCI1394_IsoRecvIntMaskClear, ~0);
|
|
|
|
size = sizeof(struct iso_context) * hweight32(ohci->it_context_mask);
|
|
|
|
ohci->it_context_list = kzalloc(size, GFP_KERNEL);
|
|
|
|
|
|
|
|
reg_write(ohci, OHCI1394_IsoXmitIntMaskSet, ~0);
|
|
|
|
ohci->ir_context_mask = reg_read(ohci, OHCI1394_IsoXmitIntMaskSet);
|
|
|
|
reg_write(ohci, OHCI1394_IsoXmitIntMaskClear, ~0);
|
|
|
|
size = sizeof(struct iso_context) * hweight32(ohci->ir_context_mask);
|
|
|
|
ohci->ir_context_list = kzalloc(size, GFP_KERNEL);
|
|
|
|
|
|
|
|
if (ohci->it_context_list == NULL || ohci->ir_context_list == NULL) {
|
|
|
|
fw_error("Out of memory for it/ir contexts.\n");
|
2007-05-10 06:23:15 +07:00
|
|
|
err = -ENOMEM;
|
|
|
|
goto fail_registers;
|
2006-12-20 07:58:35 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
/* self-id dma buffer allocation */
|
|
|
|
ohci->self_id_cpu = dma_alloc_coherent(ohci->card.device,
|
|
|
|
SELF_ID_BUF_SIZE,
|
|
|
|
&ohci->self_id_bus,
|
|
|
|
GFP_KERNEL);
|
|
|
|
if (ohci->self_id_cpu == NULL) {
|
|
|
|
fw_error("Out of memory for self ID buffer.\n");
|
2007-05-10 06:23:15 +07:00
|
|
|
err = -ENOMEM;
|
|
|
|
goto fail_registers;
|
2006-12-20 07:58:35 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
bus_options = reg_read(ohci, OHCI1394_BusOptions);
|
|
|
|
max_receive = (bus_options >> 12) & 0xf;
|
|
|
|
link_speed = bus_options & 0x7;
|
|
|
|
guid = ((u64) reg_read(ohci, OHCI1394_GUIDHi) << 32) |
|
|
|
|
reg_read(ohci, OHCI1394_GUIDLo);
|
|
|
|
|
2007-05-10 06:23:15 +07:00
|
|
|
err = fw_card_add(&ohci->card, max_receive, link_speed, guid);
|
|
|
|
if (err < 0)
|
|
|
|
goto fail_self_id;
|
2006-12-20 07:58:35 +07:00
|
|
|
|
2007-02-17 05:34:49 +07:00
|
|
|
ohci->version = reg_read(ohci, OHCI1394_Version) & 0x00ff00ff;
|
2007-02-17 05:34:43 +07:00
|
|
|
fw_notify("Added fw-ohci device %s, OHCI version %x.%x\n",
|
2007-02-17 05:34:49 +07:00
|
|
|
dev->dev.bus_id, ohci->version >> 16, ohci->version & 0xff);
|
2006-12-20 07:58:35 +07:00
|
|
|
return 0;
|
2007-05-10 06:23:15 +07:00
|
|
|
|
|
|
|
fail_self_id:
|
|
|
|
dma_free_coherent(ohci->card.device, SELF_ID_BUF_SIZE,
|
|
|
|
ohci->self_id_cpu, ohci->self_id_bus);
|
|
|
|
fail_registers:
|
|
|
|
kfree(ohci->it_context_list);
|
|
|
|
kfree(ohci->ir_context_list);
|
|
|
|
pci_iounmap(dev, ohci->registers);
|
|
|
|
fail_iomem:
|
|
|
|
pci_release_region(dev, 0);
|
|
|
|
fail_disable:
|
|
|
|
pci_disable_device(dev);
|
2008-02-25 00:59:55 +07:00
|
|
|
fail_free:
|
|
|
|
kfree(&ohci->card);
|
2007-05-10 06:23:15 +07:00
|
|
|
|
|
|
|
return err;
|
2006-12-20 07:58:35 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
static void pci_remove(struct pci_dev *dev)
|
|
|
|
{
|
|
|
|
struct fw_ohci *ohci;
|
|
|
|
|
|
|
|
ohci = pci_get_drvdata(dev);
|
2007-03-08 00:12:38 +07:00
|
|
|
reg_write(ohci, OHCI1394_IntMaskClear, ~0);
|
|
|
|
flush_writes(ohci);
|
2006-12-20 07:58:35 +07:00
|
|
|
fw_core_remove_card(&ohci->card);
|
|
|
|
|
2007-05-08 07:33:32 +07:00
|
|
|
/*
|
|
|
|
* FIXME: Fail all pending packets here, now that the upper
|
|
|
|
* layers can't queue any more.
|
|
|
|
*/
|
2006-12-20 07:58:35 +07:00
|
|
|
|
|
|
|
software_reset(ohci);
|
|
|
|
free_irq(dev->irq, ohci);
|
2007-05-10 06:23:15 +07:00
|
|
|
dma_free_coherent(ohci->card.device, SELF_ID_BUF_SIZE,
|
|
|
|
ohci->self_id_cpu, ohci->self_id_bus);
|
|
|
|
kfree(ohci->it_context_list);
|
|
|
|
kfree(ohci->ir_context_list);
|
|
|
|
pci_iounmap(dev, ohci->registers);
|
|
|
|
pci_release_region(dev, 0);
|
|
|
|
pci_disable_device(dev);
|
2008-02-25 00:59:55 +07:00
|
|
|
kfree(&ohci->card);
|
2008-03-01 18:35:29 +07:00
|
|
|
ohci_pmac_off(dev);
|
2008-03-01 08:42:56 +07:00
|
|
|
|
2006-12-20 07:58:35 +07:00
|
|
|
fw_notify("Removed fw-ohci device.\n");
|
|
|
|
}
|
|
|
|
|
2007-05-31 06:06:35 +07:00
|
|
|
#ifdef CONFIG_PM
|
2008-03-01 18:35:29 +07:00
|
|
|
static int pci_suspend(struct pci_dev *dev, pm_message_t state)
|
2007-05-31 06:06:35 +07:00
|
|
|
{
|
2008-03-01 18:35:29 +07:00
|
|
|
struct fw_ohci *ohci = pci_get_drvdata(dev);
|
2007-05-31 06:06:35 +07:00
|
|
|
int err;
|
|
|
|
|
|
|
|
software_reset(ohci);
|
2008-03-01 18:35:29 +07:00
|
|
|
free_irq(dev->irq, ohci);
|
|
|
|
err = pci_save_state(dev);
|
2007-05-31 06:06:35 +07:00
|
|
|
if (err) {
|
2007-06-10 00:26:22 +07:00
|
|
|
fw_error("pci_save_state failed\n");
|
2007-05-31 06:06:35 +07:00
|
|
|
return err;
|
|
|
|
}
|
2008-03-01 18:35:29 +07:00
|
|
|
err = pci_set_power_state(dev, pci_choose_state(dev, state));
|
2007-09-06 14:50:30 +07:00
|
|
|
if (err)
|
|
|
|
fw_error("pci_set_power_state failed with %d\n", err);
|
2008-03-01 18:35:29 +07:00
|
|
|
ohci_pmac_off(dev);
|
2008-03-01 08:42:56 +07:00
|
|
|
|
2007-05-31 06:06:35 +07:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2008-03-01 18:35:29 +07:00
|
|
|
static int pci_resume(struct pci_dev *dev)
|
2007-05-31 06:06:35 +07:00
|
|
|
{
|
2008-03-01 18:35:29 +07:00
|
|
|
struct fw_ohci *ohci = pci_get_drvdata(dev);
|
2007-05-31 06:06:35 +07:00
|
|
|
int err;
|
|
|
|
|
2008-03-01 18:35:29 +07:00
|
|
|
ohci_pmac_on(dev);
|
|
|
|
pci_set_power_state(dev, PCI_D0);
|
|
|
|
pci_restore_state(dev);
|
|
|
|
err = pci_enable_device(dev);
|
2007-05-31 06:06:35 +07:00
|
|
|
if (err) {
|
2007-06-10 00:26:22 +07:00
|
|
|
fw_error("pci_enable_device failed\n");
|
2007-05-31 06:06:35 +07:00
|
|
|
return err;
|
|
|
|
}
|
|
|
|
|
2007-06-06 06:27:05 +07:00
|
|
|
return ohci_enable(&ohci->card, NULL, 0);
|
2007-05-31 06:06:35 +07:00
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
2006-12-20 07:58:35 +07:00
|
|
|
static struct pci_device_id pci_table[] = {
|
|
|
|
{ PCI_DEVICE_CLASS(PCI_CLASS_SERIAL_FIREWIRE_OHCI, ~0) },
|
|
|
|
{ }
|
|
|
|
};
|
|
|
|
|
|
|
|
MODULE_DEVICE_TABLE(pci, pci_table);
|
|
|
|
|
|
|
|
static struct pci_driver fw_ohci_pci_driver = {
|
|
|
|
.name = ohci_driver_name,
|
|
|
|
.id_table = pci_table,
|
|
|
|
.probe = pci_probe,
|
|
|
|
.remove = pci_remove,
|
2007-05-31 06:06:35 +07:00
|
|
|
#ifdef CONFIG_PM
|
|
|
|
.resume = pci_resume,
|
|
|
|
.suspend = pci_suspend,
|
|
|
|
#endif
|
2006-12-20 07:58:35 +07:00
|
|
|
};
|
|
|
|
|
|
|
|
MODULE_AUTHOR("Kristian Hoegsberg <krh@bitplanet.net>");
|
|
|
|
MODULE_DESCRIPTION("Driver for PCI OHCI IEEE1394 controllers");
|
|
|
|
MODULE_LICENSE("GPL");
|
|
|
|
|
2007-05-06 04:17:13 +07:00
|
|
|
/* Provide a module alias so root-on-sbp2 initrds don't break. */
|
|
|
|
#ifndef CONFIG_IEEE1394_OHCI1394_MODULE
|
|
|
|
MODULE_ALIAS("ohci1394");
|
|
|
|
#endif
|
|
|
|
|
2006-12-20 07:58:35 +07:00
|
|
|
static int __init fw_ohci_init(void)
|
|
|
|
{
|
|
|
|
return pci_register_driver(&fw_ohci_pci_driver);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void __exit fw_ohci_cleanup(void)
|
|
|
|
{
|
|
|
|
pci_unregister_driver(&fw_ohci_pci_driver);
|
|
|
|
}
|
|
|
|
|
|
|
|
module_init(fw_ohci_init);
|
|
|
|
module_exit(fw_ohci_cleanup);
|