linux_dsm_epyc7002/drivers/ata/pata_it821x.c

785 lines
22 KiB
C
Raw Normal View History

/*
* pata_it821x.c - IT821x PATA for new ATA layer
* (C) 2005 Red Hat Inc
* Alan Cox <alan@redhat.com>
* (C) 2007 Bartlomiej Zolnierkiewicz
*
* based upon
*
* it821x.c
*
* linux/drivers/ide/pci/it821x.c Version 0.09 December 2004
*
* Copyright (C) 2004 Red Hat <alan@redhat.com>
*
* May be copied or modified under the terms of the GNU General Public License
* Based in part on the ITE vendor provided SCSI driver.
*
* Documentation available from
* http://www.ite.com.tw/pc/IT8212F_V04.pdf
* Some other documents are NDA.
*
* The ITE8212 isn't exactly a standard IDE controller. It has two
* modes. In pass through mode then it is an IDE controller. In its smart
* mode its actually quite a capable hardware raid controller disguised
* as an IDE controller. Smart mode only understands DMA read/write and
* identify, none of the fancier commands apply. The IT8211 is identical
* in other respects but lacks the raid mode.
*
* Errata:
* o Rev 0x10 also requires master/slave hold the same DMA timings and
* cannot do ATAPI MWDMA.
* o The identify data for raid volumes lacks CHS info (technically ok)
* but also fails to set the LBA28 and other bits. We fix these in
* the IDE probe quirk code.
* o If you write LBA48 sized I/O's (ie > 256 sector) in smart mode
* raid then the controller firmware dies
* o Smart mode without RAID doesn't clear all the necessary identify
* bits to reduce the command set to the one used
*
* This has a few impacts on the driver
* - In pass through mode we do all the work you would expect
* - In smart mode the clocking set up is done by the controller generally
* but we must watch the other limits and filter.
* - There are a few extra vendor commands that actually talk to the
* controller but only work PIO with no IRQ.
*
* Vendor areas of the identify block in smart mode are used for the
* timing and policy set up. Each HDD in raid mode also has a serial
* block on the disk. The hardware extra commands are get/set chip status,
* rebuild, get rebuild status.
*
* In Linux the driver supports pass through mode as if the device was
* just another IDE controller. If the smart mode is running then
* volumes are managed by the controller firmware and each IDE "disk"
* is a raid volume. Even more cute - the controller can do automated
* hotplug and rebuild.
*
* The pass through controller itself is a little demented. It has a
* flaw that it has a single set of PIO/MWDMA timings per channel so
* non UDMA devices restrict each others performance. It also has a
* single clock source per channel so mixed UDMA100/133 performance
* isn't perfect and we have to pick a clock. Thankfully none of this
* matters in smart mode. ATAPI DMA is not currently supported.
*
* It seems the smart mode is a win for RAID1/RAID10 but otherwise not.
*
* TODO
* - ATAPI and other speed filtering
* - RAID configuration ioctls
*/
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/pci.h>
#include <linux/init.h>
#include <linux/blkdev.h>
#include <linux/delay.h>
#include <scsi/scsi_host.h>
#include <linux/libata.h>
#define DRV_NAME "pata_it821x"
#define DRV_VERSION "0.3.8"
struct it821x_dev
{
unsigned int smart:1, /* Are we in smart raid mode */
timing10:1; /* Rev 0x10 */
u8 clock_mode; /* 0, ATA_50 or ATA_66 */
u8 want[2][2]; /* Mode/Pri log for master slave */
/* We need these for switching the clock when DMA goes on/off
The high byte is the 66Mhz timing */
u16 pio[2]; /* Cached PIO values */
u16 mwdma[2]; /* Cached MWDMA values */
u16 udma[2]; /* Cached UDMA values (per drive) */
u16 last_device; /* Master or slave loaded ? */
};
#define ATA_66 0
#define ATA_50 1
#define ATA_ANY 2
#define UDMA_OFF 0
#define MWDMA_OFF 0
/*
* We allow users to force the card into non raid mode without
* flashing the alternative BIOS. This is also necessary right now
* for embedded platforms that cannot run a PC BIOS but are using this
* device.
*/
static int it8212_noraid;
/**
* it821x_program - program the PIO/MWDMA registers
* @ap: ATA port
* @adev: Device to program
* @timing: Timing value (66Mhz in top 8bits, 50 in the low 8)
*
* Program the PIO/MWDMA timing for this channel according to the
* current clock. These share the same register so are managed by
* the DMA start/stop sequence as with the old driver.
*/
static void it821x_program(struct ata_port *ap, struct ata_device *adev, u16 timing)
{
struct pci_dev *pdev = to_pci_dev(ap->host->dev);
struct it821x_dev *itdev = ap->private_data;
int channel = ap->port_no;
u8 conf;
/* Program PIO/MWDMA timing bits */
if (itdev->clock_mode == ATA_66)
conf = timing >> 8;
else
conf = timing & 0xFF;
pci_write_config_byte(pdev, 0x54 + 4 * channel, conf);
}
/**
* it821x_program_udma - program the UDMA registers
* @ap: ATA port
* @adev: ATA device to update
* @timing: Timing bits. Top 8 are for 66Mhz bottom for 50Mhz
*
* Program the UDMA timing for this drive according to the
* current clock. Handles the dual clocks and also knows about
* the errata on the 0x10 revision. The UDMA errata is partly handled
* here and partly in start_dma.
*/
static void it821x_program_udma(struct ata_port *ap, struct ata_device *adev, u16 timing)
{
struct it821x_dev *itdev = ap->private_data;
struct pci_dev *pdev = to_pci_dev(ap->host->dev);
int channel = ap->port_no;
int unit = adev->devno;
u8 conf;
/* Program UDMA timing bits */
if (itdev->clock_mode == ATA_66)
conf = timing >> 8;
else
conf = timing & 0xFF;
if (itdev->timing10 == 0)
pci_write_config_byte(pdev, 0x56 + 4 * channel + unit, conf);
else {
/* Early revision must be programmed for both together */
pci_write_config_byte(pdev, 0x56 + 4 * channel, conf);
pci_write_config_byte(pdev, 0x56 + 4 * channel + 1, conf);
}
}
/**
* it821x_clock_strategy
* @ap: ATA interface
* @adev: ATA device being updated
*
* Select between the 50 and 66Mhz base clocks to get the best
* results for this interface.
*/
static void it821x_clock_strategy(struct ata_port *ap, struct ata_device *adev)
{
struct pci_dev *pdev = to_pci_dev(ap->host->dev);
struct it821x_dev *itdev = ap->private_data;
u8 unit = adev->devno;
struct ata_device *pair = ata_dev_pair(adev);
int clock, altclock;
u8 v;
int sel = 0;
/* Look for the most wanted clocking */
if (itdev->want[0][0] > itdev->want[1][0]) {
clock = itdev->want[0][1];
altclock = itdev->want[1][1];
} else {
clock = itdev->want[1][1];
altclock = itdev->want[0][1];
}
/* Master doesn't care does the slave ? */
if (clock == ATA_ANY)
clock = altclock;
/* Nobody cares - keep the same clock */
if (clock == ATA_ANY)
return;
/* No change */
if (clock == itdev->clock_mode)
return;
/* Load this into the controller */
if (clock == ATA_66)
itdev->clock_mode = ATA_66;
else {
itdev->clock_mode = ATA_50;
sel = 1;
}
pci_read_config_byte(pdev, 0x50, &v);
v &= ~(1 << (1 + ap->port_no));
v |= sel << (1 + ap->port_no);
pci_write_config_byte(pdev, 0x50, v);
/*
* Reprogram the UDMA/PIO of the pair drive for the switch
* MWDMA will be dealt with by the dma switcher
*/
if (pair && itdev->udma[1-unit] != UDMA_OFF) {
it821x_program_udma(ap, pair, itdev->udma[1-unit]);
it821x_program(ap, pair, itdev->pio[1-unit]);
}
/*
* Reprogram the UDMA/PIO of our drive for the switch.
* MWDMA will be dealt with by the dma switcher
*/
if (itdev->udma[unit] != UDMA_OFF) {
it821x_program_udma(ap, adev, itdev->udma[unit]);
it821x_program(ap, adev, itdev->pio[unit]);
}
}
/**
* it821x_passthru_set_piomode - set PIO mode data
* @ap: ATA interface
* @adev: ATA device
*
* Configure for PIO mode. This is complicated as the register is
* shared by PIO and MWDMA and for both channels.
*/
static void it821x_passthru_set_piomode(struct ata_port *ap, struct ata_device *adev)
{
/* Spec says 89 ref driver uses 88 */
static const u16 pio[] = { 0xAA88, 0xA382, 0xA181, 0x3332, 0x3121 };
static const u8 pio_want[] = { ATA_66, ATA_66, ATA_66, ATA_66, ATA_ANY };
struct it821x_dev *itdev = ap->private_data;
int unit = adev->devno;
int mode_wanted = adev->pio_mode - XFER_PIO_0;
/* We prefer 66Mhz clock for PIO 0-3, don't care for PIO4 */
itdev->want[unit][1] = pio_want[mode_wanted];
itdev->want[unit][0] = 1; /* PIO is lowest priority */
itdev->pio[unit] = pio[mode_wanted];
it821x_clock_strategy(ap, adev);
it821x_program(ap, adev, itdev->pio[unit]);
}
/**
* it821x_passthru_set_dmamode - set initial DMA mode data
* @ap: ATA interface
* @adev: ATA device
*
* Set up the DMA modes. The actions taken depend heavily on the mode
* to use. If UDMA is used as is hopefully the usual case then the
* timing register is private and we need only consider the clock. If
* we are using MWDMA then we have to manage the setting ourself as
* we switch devices and mode.
*/
static void it821x_passthru_set_dmamode(struct ata_port *ap, struct ata_device *adev)
{
static const u16 dma[] = { 0x8866, 0x3222, 0x3121 };
static const u8 mwdma_want[] = { ATA_ANY, ATA_66, ATA_ANY };
static const u16 udma[] = { 0x4433, 0x4231, 0x3121, 0x2121, 0x1111, 0x2211, 0x1111 };
static const u8 udma_want[] = { ATA_ANY, ATA_50, ATA_ANY, ATA_66, ATA_66, ATA_50, ATA_66 };
struct pci_dev *pdev = to_pci_dev(ap->host->dev);
struct it821x_dev *itdev = ap->private_data;
int channel = ap->port_no;
int unit = adev->devno;
u8 conf;
if (adev->dma_mode >= XFER_UDMA_0) {
int mode_wanted = adev->dma_mode - XFER_UDMA_0;
itdev->want[unit][1] = udma_want[mode_wanted];
itdev->want[unit][0] = 3; /* UDMA is high priority */
itdev->mwdma[unit] = MWDMA_OFF;
itdev->udma[unit] = udma[mode_wanted];
if (mode_wanted >= 5)
itdev->udma[unit] |= 0x8080; /* UDMA 5/6 select on */
/* UDMA on. Again revision 0x10 must do the pair */
pci_read_config_byte(pdev, 0x50, &conf);
if (itdev->timing10)
conf &= channel ? 0x9F: 0xE7;
else
conf &= ~ (1 << (3 + 2 * channel + unit));
pci_write_config_byte(pdev, 0x50, conf);
it821x_clock_strategy(ap, adev);
it821x_program_udma(ap, adev, itdev->udma[unit]);
} else {
int mode_wanted = adev->dma_mode - XFER_MW_DMA_0;
itdev->want[unit][1] = mwdma_want[mode_wanted];
itdev->want[unit][0] = 2; /* MWDMA is low priority */
itdev->mwdma[unit] = dma[mode_wanted];
itdev->udma[unit] = UDMA_OFF;
/* UDMA bits off - Revision 0x10 do them in pairs */
pci_read_config_byte(pdev, 0x50, &conf);
if (itdev->timing10)
conf |= channel ? 0x60: 0x18;
else
conf |= 1 << (3 + 2 * channel + unit);
pci_write_config_byte(pdev, 0x50, conf);
it821x_clock_strategy(ap, adev);
}
}
/**
* it821x_passthru_dma_start - DMA start callback
* @qc: Command in progress
*
* Usually drivers set the DMA timing at the point the set_dmamode call
* is made. IT821x however requires we load new timings on the
* transitions in some cases.
*/
static void it821x_passthru_bmdma_start(struct ata_queued_cmd *qc)
{
struct ata_port *ap = qc->ap;
struct ata_device *adev = qc->dev;
struct it821x_dev *itdev = ap->private_data;
int unit = adev->devno;
if (itdev->mwdma[unit] != MWDMA_OFF)
it821x_program(ap, adev, itdev->mwdma[unit]);
else if (itdev->udma[unit] != UDMA_OFF && itdev->timing10)
it821x_program_udma(ap, adev, itdev->udma[unit]);
ata_bmdma_start(qc);
}
/**
* it821x_passthru_dma_stop - DMA stop callback
* @qc: ATA command
*
* We loaded new timings in dma_start, as a result we need to restore
* the PIO timings in dma_stop so that the next command issue gets the
* right clock values.
*/
static void it821x_passthru_bmdma_stop(struct ata_queued_cmd *qc)
{
struct ata_port *ap = qc->ap;
struct ata_device *adev = qc->dev;
struct it821x_dev *itdev = ap->private_data;
int unit = adev->devno;
ata_bmdma_stop(qc);
if (itdev->mwdma[unit] != MWDMA_OFF)
it821x_program(ap, adev, itdev->pio[unit]);
}
/**
* it821x_passthru_dev_select - Select master/slave
* @ap: ATA port
* @device: Device number (not pointer)
*
* Device selection hook. If necessary perform clock switching
*/
static void it821x_passthru_dev_select(struct ata_port *ap,
unsigned int device)
{
struct it821x_dev *itdev = ap->private_data;
if (itdev && device != itdev->last_device) {
struct ata_device *adev = &ap->link.device[device];
it821x_program(ap, adev, itdev->pio[adev->devno]);
itdev->last_device = device;
}
ata_sff_dev_select(ap, device);
}
/**
* it821x_smart_qc_issue - wrap qc issue prot
* @qc: command
*
* Wrap the command issue sequence for the IT821x. We need to
* perform out own device selection timing loads before the
* usual happenings kick off
*/
static unsigned int it821x_smart_qc_issue(struct ata_queued_cmd *qc)
{
switch(qc->tf.command)
{
/* Commands the firmware supports */
case ATA_CMD_READ:
case ATA_CMD_READ_EXT:
case ATA_CMD_WRITE:
case ATA_CMD_WRITE_EXT:
case ATA_CMD_PIO_READ:
case ATA_CMD_PIO_READ_EXT:
case ATA_CMD_PIO_WRITE:
case ATA_CMD_PIO_WRITE_EXT:
case ATA_CMD_READ_MULTI:
case ATA_CMD_READ_MULTI_EXT:
case ATA_CMD_WRITE_MULTI:
case ATA_CMD_WRITE_MULTI_EXT:
case ATA_CMD_ID_ATA:
/* Arguably should just no-op this one */
case ATA_CMD_SET_FEATURES:
return ata_sff_qc_issue(qc);
}
printk(KERN_DEBUG "it821x: can't process command 0x%02X\n", qc->tf.command);
return AC_ERR_DEV;
}
/**
* it821x_passthru_qc_issue - wrap qc issue prot
* @qc: command
*
* Wrap the command issue sequence for the IT821x. We need to
* perform out own device selection timing loads before the
* usual happenings kick off
*/
static unsigned int it821x_passthru_qc_issue(struct ata_queued_cmd *qc)
{
it821x_passthru_dev_select(qc->ap, qc->dev->devno);
return ata_sff_qc_issue(qc);
}
/**
* it821x_smart_set_mode - mode setting
* @link: interface to set up
* @unused: device that failed (error only)
*
* Use a non standard set_mode function. We don't want to be tuned.
* The BIOS configured everything. Our job is not to fiddle. We
* read the dma enabled bits from the PCI configuration of the device
* and respect them.
*/
static int it821x_smart_set_mode(struct ata_link *link, struct ata_device **unused)
{
struct ata_device *dev;
ata_link_for_each_dev(dev, link) {
if (ata_dev_enabled(dev)) {
/* We don't really care */
dev->pio_mode = XFER_PIO_0;
dev->dma_mode = XFER_MW_DMA_0;
/* We do need the right mode information for DMA or PIO
and this comes from the current configuration flags */
if (ata_id_has_dma(dev->id)) {
ata_dev_printk(dev, KERN_INFO, "configured for DMA\n");
dev->xfer_mode = XFER_MW_DMA_0;
dev->xfer_shift = ATA_SHIFT_MWDMA;
dev->flags &= ~ATA_DFLAG_PIO;
} else {
ata_dev_printk(dev, KERN_INFO, "configured for PIO\n");
dev->xfer_mode = XFER_PIO_0;
dev->xfer_shift = ATA_SHIFT_PIO;
dev->flags |= ATA_DFLAG_PIO;
}
}
}
return 0;
}
/**
* it821x_dev_config - Called each device identify
* @adev: Device that has just been identified
*
* Perform the initial setup needed for each device that is chip
* special. In our case we need to lock the sector count to avoid
* blowing the brains out of the firmware with large LBA48 requests
*
* FIXME: When FUA appears we need to block FUA too. And SMART and
* basically we need to filter commands for this chip.
*/
static void it821x_dev_config(struct ata_device *adev)
{
unsigned char model_num[ATA_ID_PROD_LEN + 1];
ata_id_c_string(adev->id, model_num, ATA_ID_PROD, sizeof(model_num));
if (adev->max_sectors > 255)
adev->max_sectors = 255;
if (strstr(model_num, "Integrated Technology Express")) {
/* RAID mode */
printk(KERN_INFO "IT821x %sRAID%d volume",
adev->id[147]?"Bootable ":"",
adev->id[129]);
if (adev->id[129] != 1)
printk("(%dK stripe)", adev->id[146]);
printk(".\n");
}
/* This is a controller firmware triggered funny, don't
report the drive faulty! */
adev->horkage &= ~ATA_HORKAGE_DIAGNOSTIC;
}
/**
* it821x_ident_hack - Hack identify data up
* @ap: Port
*
* Walk the devices on this firmware driven port and slightly
* mash the identify data to stop us and common tools trying to
* use features not firmware supported. The firmware itself does
* some masking (eg SMART) but not enough.
*
* This is a bit of an abuse of the cable method, but it is the
* only method called at the right time. We could modify the libata
* core specifically for ident hacking but while we have one offender
* it seems better to keep the fallout localised.
*/
static int it821x_ident_hack(struct ata_port *ap)
{
struct ata_device *adev;
ata_link_for_each_dev(adev, &ap->link) {
if (ata_dev_enabled(adev)) {
adev->id[84] &= ~(1 << 6); /* No FUA */
adev->id[85] &= ~(1 << 10); /* No HPA */
adev->id[76] = 0; /* No NCQ/AN etc */
}
}
return ata_cable_unknown(ap);
}
/**
* it821x_check_atapi_dma - ATAPI DMA handler
* @qc: Command we are about to issue
*
* Decide if this ATAPI command can be issued by DMA on this
* controller. Return 0 if it can be.
*/
static int it821x_check_atapi_dma(struct ata_queued_cmd *qc)
{
struct ata_port *ap = qc->ap;
struct it821x_dev *itdev = ap->private_data;
/* Only use dma for transfers to/from the media. */
if (ata_qc_raw_nbytes(qc) < 2048)
return -EOPNOTSUPP;
/* No ATAPI DMA in smart mode */
if (itdev->smart)
return -EOPNOTSUPP;
/* No ATAPI DMA on rev 10 */
if (itdev->timing10)
return -EOPNOTSUPP;
/* Cool */
return 0;
}
/**
* it821x_port_start - port setup
* @ap: ATA port being set up
*
* The it821x needs to maintain private data structures and also to
* use the standard PCI interface which lacks support for this
* functionality. We instead set up the private data on the port
* start hook, and tear it down on port stop
*/
static int it821x_port_start(struct ata_port *ap)
{
struct pci_dev *pdev = to_pci_dev(ap->host->dev);
struct it821x_dev *itdev;
u8 conf;
int ret = ata_sff_port_start(ap);
if (ret < 0)
return ret;
itdev = devm_kzalloc(&pdev->dev, sizeof(struct it821x_dev), GFP_KERNEL);
if (itdev == NULL)
return -ENOMEM;
ap->private_data = itdev;
pci_read_config_byte(pdev, 0x50, &conf);
if (conf & 1) {
itdev->smart = 1;
/* Long I/O's although allowed in LBA48 space cause the
onboard firmware to enter the twighlight zone */
/* No ATAPI DMA in this mode either */
}
/* Pull the current clocks from 0x50 */
if (conf & (1 << (1 + ap->port_no)))
itdev->clock_mode = ATA_50;
else
itdev->clock_mode = ATA_66;
itdev->want[0][1] = ATA_ANY;
itdev->want[1][1] = ATA_ANY;
itdev->last_device = -1;
if (pdev->revision == 0x10) {
itdev->timing10 = 1;
/* Need to disable ATAPI DMA for this case */
if (!itdev->smart)
printk(KERN_WARNING DRV_NAME": Revision 0x10, workarounds activated.\n");
}
return 0;
}
static struct scsi_host_template it821x_sht = {
ATA_BMDMA_SHT(DRV_NAME),
};
static struct ata_port_operations it821x_smart_port_ops = {
libata: implement and use ops inheritance libata lets low level drivers build ata_port_operations table and register it with libata core layer. This allows low level drivers high level of flexibility but also burdens them with lots of boilerplate entries. This becomes worse for drivers which support related similar controllers which differ slightly. They share most of the operations except for a few. However, the driver still needs to list all operations for each variant. This results in large number of duplicate entries, which is not only inefficient but also error-prone as it becomes very difficult to tell what the actual differences are. This duplicate boilerplates all over the low level drivers also make updating the core layer exteremely difficult and error-prone. When compounded with multi-branched development model, it ends up accumulating inconsistencies over time. Some of those inconsistencies cause immediate problems and fixed. Others just remain there dormant making maintenance increasingly difficult. To rectify the problem, this patch implements ata_port_operations inheritance. To allow LLDs to easily re-use their own ops tables overriding only specific methods, this patch implements poor man's class inheritance. An ops table has ->inherits field which can be set to any ops table as long as it doesn't create a loop. When the host is started, the inheritance chain is followed and any operation which isn't specified is taken from the nearest ancestor which has it specified. This operation is called finalization and done only once per an ops table and the LLD doesn't have to do anything special about it other than making the ops table non-const such that libata can update it. libata provides four base ops tables lower drivers can inherit from - base, sata, pmp, sff and bmdma. To avoid overriding these ops accidentaly, these ops are declared const and LLDs should always inherit these instead of using them directly. After finalization, all the ops table are identical before and after the patch except for setting .irq_handler to ata_interrupt in drivers which didn't use to. The .irq_handler doesn't have any actual effect and the field will soon be removed by later patch. * sata_sx4 is still using old style EH and currently doesn't take advantage of ops inheritance. Signed-off-by: Tejun Heo <htejun@gmail.com>
2008-03-25 10:22:49 +07:00
.inherits = &ata_bmdma_port_ops,
.check_atapi_dma= it821x_check_atapi_dma,
.qc_issue = it821x_smart_qc_issue,
libata: implement and use ops inheritance libata lets low level drivers build ata_port_operations table and register it with libata core layer. This allows low level drivers high level of flexibility but also burdens them with lots of boilerplate entries. This becomes worse for drivers which support related similar controllers which differ slightly. They share most of the operations except for a few. However, the driver still needs to list all operations for each variant. This results in large number of duplicate entries, which is not only inefficient but also error-prone as it becomes very difficult to tell what the actual differences are. This duplicate boilerplates all over the low level drivers also make updating the core layer exteremely difficult and error-prone. When compounded with multi-branched development model, it ends up accumulating inconsistencies over time. Some of those inconsistencies cause immediate problems and fixed. Others just remain there dormant making maintenance increasingly difficult. To rectify the problem, this patch implements ata_port_operations inheritance. To allow LLDs to easily re-use their own ops tables overriding only specific methods, this patch implements poor man's class inheritance. An ops table has ->inherits field which can be set to any ops table as long as it doesn't create a loop. When the host is started, the inheritance chain is followed and any operation which isn't specified is taken from the nearest ancestor which has it specified. This operation is called finalization and done only once per an ops table and the LLD doesn't have to do anything special about it other than making the ops table non-const such that libata can update it. libata provides four base ops tables lower drivers can inherit from - base, sata, pmp, sff and bmdma. To avoid overriding these ops accidentaly, these ops are declared const and LLDs should always inherit these instead of using them directly. After finalization, all the ops table are identical before and after the patch except for setting .irq_handler to ata_interrupt in drivers which didn't use to. The .irq_handler doesn't have any actual effect and the field will soon be removed by later patch. * sata_sx4 is still using old style EH and currently doesn't take advantage of ops inheritance. Signed-off-by: Tejun Heo <htejun@gmail.com>
2008-03-25 10:22:49 +07:00
.cable_detect = it821x_ident_hack,
.set_mode = it821x_smart_set_mode,
.dev_config = it821x_dev_config,
.port_start = it821x_port_start,
};
static struct ata_port_operations it821x_passthru_port_ops = {
libata: implement and use ops inheritance libata lets low level drivers build ata_port_operations table and register it with libata core layer. This allows low level drivers high level of flexibility but also burdens them with lots of boilerplate entries. This becomes worse for drivers which support related similar controllers which differ slightly. They share most of the operations except for a few. However, the driver still needs to list all operations for each variant. This results in large number of duplicate entries, which is not only inefficient but also error-prone as it becomes very difficult to tell what the actual differences are. This duplicate boilerplates all over the low level drivers also make updating the core layer exteremely difficult and error-prone. When compounded with multi-branched development model, it ends up accumulating inconsistencies over time. Some of those inconsistencies cause immediate problems and fixed. Others just remain there dormant making maintenance increasingly difficult. To rectify the problem, this patch implements ata_port_operations inheritance. To allow LLDs to easily re-use their own ops tables overriding only specific methods, this patch implements poor man's class inheritance. An ops table has ->inherits field which can be set to any ops table as long as it doesn't create a loop. When the host is started, the inheritance chain is followed and any operation which isn't specified is taken from the nearest ancestor which has it specified. This operation is called finalization and done only once per an ops table and the LLD doesn't have to do anything special about it other than making the ops table non-const such that libata can update it. libata provides four base ops tables lower drivers can inherit from - base, sata, pmp, sff and bmdma. To avoid overriding these ops accidentaly, these ops are declared const and LLDs should always inherit these instead of using them directly. After finalization, all the ops table are identical before and after the patch except for setting .irq_handler to ata_interrupt in drivers which didn't use to. The .irq_handler doesn't have any actual effect and the field will soon be removed by later patch. * sata_sx4 is still using old style EH and currently doesn't take advantage of ops inheritance. Signed-off-by: Tejun Heo <htejun@gmail.com>
2008-03-25 10:22:49 +07:00
.inherits = &ata_bmdma_port_ops,
.check_atapi_dma= it821x_check_atapi_dma,
.dev_select = it821x_passthru_dev_select,
.bmdma_start = it821x_passthru_bmdma_start,
.bmdma_stop = it821x_passthru_bmdma_stop,
.qc_issue = it821x_passthru_qc_issue,
libata: implement and use ops inheritance libata lets low level drivers build ata_port_operations table and register it with libata core layer. This allows low level drivers high level of flexibility but also burdens them with lots of boilerplate entries. This becomes worse for drivers which support related similar controllers which differ slightly. They share most of the operations except for a few. However, the driver still needs to list all operations for each variant. This results in large number of duplicate entries, which is not only inefficient but also error-prone as it becomes very difficult to tell what the actual differences are. This duplicate boilerplates all over the low level drivers also make updating the core layer exteremely difficult and error-prone. When compounded with multi-branched development model, it ends up accumulating inconsistencies over time. Some of those inconsistencies cause immediate problems and fixed. Others just remain there dormant making maintenance increasingly difficult. To rectify the problem, this patch implements ata_port_operations inheritance. To allow LLDs to easily re-use their own ops tables overriding only specific methods, this patch implements poor man's class inheritance. An ops table has ->inherits field which can be set to any ops table as long as it doesn't create a loop. When the host is started, the inheritance chain is followed and any operation which isn't specified is taken from the nearest ancestor which has it specified. This operation is called finalization and done only once per an ops table and the LLD doesn't have to do anything special about it other than making the ops table non-const such that libata can update it. libata provides four base ops tables lower drivers can inherit from - base, sata, pmp, sff and bmdma. To avoid overriding these ops accidentaly, these ops are declared const and LLDs should always inherit these instead of using them directly. After finalization, all the ops table are identical before and after the patch except for setting .irq_handler to ata_interrupt in drivers which didn't use to. The .irq_handler doesn't have any actual effect and the field will soon be removed by later patch. * sata_sx4 is still using old style EH and currently doesn't take advantage of ops inheritance. Signed-off-by: Tejun Heo <htejun@gmail.com>
2008-03-25 10:22:49 +07:00
.cable_detect = ata_cable_unknown,
.set_piomode = it821x_passthru_set_piomode,
.set_dmamode = it821x_passthru_set_dmamode,
.port_start = it821x_port_start,
};
static void it821x_disable_raid(struct pci_dev *pdev)
{
/* Reset local CPU, and set BIOS not ready */
pci_write_config_byte(pdev, 0x5E, 0x01);
/* Set to bypass mode, and reset PCI bus */
pci_write_config_byte(pdev, 0x50, 0x00);
pci_write_config_word(pdev, PCI_COMMAND,
PCI_COMMAND_PARITY | PCI_COMMAND_IO |
PCI_COMMAND_MEMORY | PCI_COMMAND_MASTER);
pci_write_config_word(pdev, 0x40, 0xA0F3);
pci_write_config_dword(pdev,0x4C, 0x02040204);
pci_write_config_byte(pdev, 0x42, 0x36);
pci_write_config_byte(pdev, PCI_LATENCY_TIMER, 0x20);
}
static int it821x_init_one(struct pci_dev *pdev, const struct pci_device_id *id)
{
u8 conf;
libata: clean up SFF init mess The intention of using port_mask in SFF init helpers was to eventually support exoctic configurations such as combination of legacy and native port on the same controller. This never became actually necessary and the related code always has been subtly broken one way or the other. Now that new init model is in place, there is no reason to make common helpers capable of handling all corner cases. Exotic cases can simply dealt within LLDs as necessary. This patch removes port_mask handling in SFF init helpers. SFF init helpers don't take n_ports argument and interpret it into port_mask anymore. All information is carried via port_info. n_ports argument is dropped and always two ports are allocated. LLD can tell SFF to skip certain port by marking it dummy. Note that SFF code has been treating unuvailable ports this way for a long time until recent breakage fix from Linus and is consistent with how other drivers handle with unavailable ports. This fixes 1-port legacy host handling still broken after the recent native mode fix and simplifies SFF init logic. The following changes are made... * ata_pci_init_native_host() and ata_init_legacy_host() both now try to initialized whatever they can and mark failed ports dummy. They return 0 if any port is successfully initialized. * ata_pci_prepare_native_host() and ata_pci_init_one() now doesn't take n_ports argument. All info should be specified via port_info array. Always two ports are allocated. * ata_pci_init_bmdma() exported to be used by LLDs in exotic cases. * port_info handling in all LLDs are standardized - all port_info arrays are const stack variable named ppi. Unless the second port is different from the first, its port_info is specified as NULL (tells libata that it's identical to the last non-NULL port_info). * pata_hpt37x/hpt3x2n: don't modify static variable directly. Make an on-stack copy instead as ata_piix does. * pata_uli: It has 4 ports instead of 2. Don't use ata_pci_prepare_native_host(). Allocate the host explicitly and use init helpers. It's simple enough. Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-05-04 17:43:58 +07:00
static const struct ata_port_info info_smart = {
.flags = ATA_FLAG_SLAVE_POSS,
.pio_mask = 0x1f,
.mwdma_mask = 0x07,
.port_ops = &it821x_smart_port_ops
};
libata: clean up SFF init mess The intention of using port_mask in SFF init helpers was to eventually support exoctic configurations such as combination of legacy and native port on the same controller. This never became actually necessary and the related code always has been subtly broken one way or the other. Now that new init model is in place, there is no reason to make common helpers capable of handling all corner cases. Exotic cases can simply dealt within LLDs as necessary. This patch removes port_mask handling in SFF init helpers. SFF init helpers don't take n_ports argument and interpret it into port_mask anymore. All information is carried via port_info. n_ports argument is dropped and always two ports are allocated. LLD can tell SFF to skip certain port by marking it dummy. Note that SFF code has been treating unuvailable ports this way for a long time until recent breakage fix from Linus and is consistent with how other drivers handle with unavailable ports. This fixes 1-port legacy host handling still broken after the recent native mode fix and simplifies SFF init logic. The following changes are made... * ata_pci_init_native_host() and ata_init_legacy_host() both now try to initialized whatever they can and mark failed ports dummy. They return 0 if any port is successfully initialized. * ata_pci_prepare_native_host() and ata_pci_init_one() now doesn't take n_ports argument. All info should be specified via port_info array. Always two ports are allocated. * ata_pci_init_bmdma() exported to be used by LLDs in exotic cases. * port_info handling in all LLDs are standardized - all port_info arrays are const stack variable named ppi. Unless the second port is different from the first, its port_info is specified as NULL (tells libata that it's identical to the last non-NULL port_info). * pata_hpt37x/hpt3x2n: don't modify static variable directly. Make an on-stack copy instead as ata_piix does. * pata_uli: It has 4 ports instead of 2. Don't use ata_pci_prepare_native_host(). Allocate the host explicitly and use init helpers. It's simple enough. Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-05-04 17:43:58 +07:00
static const struct ata_port_info info_passthru = {
.flags = ATA_FLAG_SLAVE_POSS,
.pio_mask = 0x1f,
.mwdma_mask = 0x07,
.udma_mask = ATA_UDMA6,
.port_ops = &it821x_passthru_port_ops
};
libata: clean up SFF init mess The intention of using port_mask in SFF init helpers was to eventually support exoctic configurations such as combination of legacy and native port on the same controller. This never became actually necessary and the related code always has been subtly broken one way or the other. Now that new init model is in place, there is no reason to make common helpers capable of handling all corner cases. Exotic cases can simply dealt within LLDs as necessary. This patch removes port_mask handling in SFF init helpers. SFF init helpers don't take n_ports argument and interpret it into port_mask anymore. All information is carried via port_info. n_ports argument is dropped and always two ports are allocated. LLD can tell SFF to skip certain port by marking it dummy. Note that SFF code has been treating unuvailable ports this way for a long time until recent breakage fix from Linus and is consistent with how other drivers handle with unavailable ports. This fixes 1-port legacy host handling still broken after the recent native mode fix and simplifies SFF init logic. The following changes are made... * ata_pci_init_native_host() and ata_init_legacy_host() both now try to initialized whatever they can and mark failed ports dummy. They return 0 if any port is successfully initialized. * ata_pci_prepare_native_host() and ata_pci_init_one() now doesn't take n_ports argument. All info should be specified via port_info array. Always two ports are allocated. * ata_pci_init_bmdma() exported to be used by LLDs in exotic cases. * port_info handling in all LLDs are standardized - all port_info arrays are const stack variable named ppi. Unless the second port is different from the first, its port_info is specified as NULL (tells libata that it's identical to the last non-NULL port_info). * pata_hpt37x/hpt3x2n: don't modify static variable directly. Make an on-stack copy instead as ata_piix does. * pata_uli: It has 4 ports instead of 2. Don't use ata_pci_prepare_native_host(). Allocate the host explicitly and use init helpers. It's simple enough. Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-05-04 17:43:58 +07:00
const struct ata_port_info *ppi[] = { NULL, NULL };
static char *mode[2] = { "pass through", "smart" };
int rc;
rc = pcim_enable_device(pdev);
if (rc)
return rc;
/* Force the card into bypass mode if so requested */
if (it8212_noraid) {
printk(KERN_INFO DRV_NAME ": forcing bypass mode.\n");
it821x_disable_raid(pdev);
}
pci_read_config_byte(pdev, 0x50, &conf);
conf &= 1;
printk(KERN_INFO DRV_NAME ": controller in %s mode.\n", mode[conf]);
if (conf == 0)
libata: clean up SFF init mess The intention of using port_mask in SFF init helpers was to eventually support exoctic configurations such as combination of legacy and native port on the same controller. This never became actually necessary and the related code always has been subtly broken one way or the other. Now that new init model is in place, there is no reason to make common helpers capable of handling all corner cases. Exotic cases can simply dealt within LLDs as necessary. This patch removes port_mask handling in SFF init helpers. SFF init helpers don't take n_ports argument and interpret it into port_mask anymore. All information is carried via port_info. n_ports argument is dropped and always two ports are allocated. LLD can tell SFF to skip certain port by marking it dummy. Note that SFF code has been treating unuvailable ports this way for a long time until recent breakage fix from Linus and is consistent with how other drivers handle with unavailable ports. This fixes 1-port legacy host handling still broken after the recent native mode fix and simplifies SFF init logic. The following changes are made... * ata_pci_init_native_host() and ata_init_legacy_host() both now try to initialized whatever they can and mark failed ports dummy. They return 0 if any port is successfully initialized. * ata_pci_prepare_native_host() and ata_pci_init_one() now doesn't take n_ports argument. All info should be specified via port_info array. Always two ports are allocated. * ata_pci_init_bmdma() exported to be used by LLDs in exotic cases. * port_info handling in all LLDs are standardized - all port_info arrays are const stack variable named ppi. Unless the second port is different from the first, its port_info is specified as NULL (tells libata that it's identical to the last non-NULL port_info). * pata_hpt37x/hpt3x2n: don't modify static variable directly. Make an on-stack copy instead as ata_piix does. * pata_uli: It has 4 ports instead of 2. Don't use ata_pci_prepare_native_host(). Allocate the host explicitly and use init helpers. It's simple enough. Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-05-04 17:43:58 +07:00
ppi[0] = &info_passthru;
else
libata: clean up SFF init mess The intention of using port_mask in SFF init helpers was to eventually support exoctic configurations such as combination of legacy and native port on the same controller. This never became actually necessary and the related code always has been subtly broken one way or the other. Now that new init model is in place, there is no reason to make common helpers capable of handling all corner cases. Exotic cases can simply dealt within LLDs as necessary. This patch removes port_mask handling in SFF init helpers. SFF init helpers don't take n_ports argument and interpret it into port_mask anymore. All information is carried via port_info. n_ports argument is dropped and always two ports are allocated. LLD can tell SFF to skip certain port by marking it dummy. Note that SFF code has been treating unuvailable ports this way for a long time until recent breakage fix from Linus and is consistent with how other drivers handle with unavailable ports. This fixes 1-port legacy host handling still broken after the recent native mode fix and simplifies SFF init logic. The following changes are made... * ata_pci_init_native_host() and ata_init_legacy_host() both now try to initialized whatever they can and mark failed ports dummy. They return 0 if any port is successfully initialized. * ata_pci_prepare_native_host() and ata_pci_init_one() now doesn't take n_ports argument. All info should be specified via port_info array. Always two ports are allocated. * ata_pci_init_bmdma() exported to be used by LLDs in exotic cases. * port_info handling in all LLDs are standardized - all port_info arrays are const stack variable named ppi. Unless the second port is different from the first, its port_info is specified as NULL (tells libata that it's identical to the last non-NULL port_info). * pata_hpt37x/hpt3x2n: don't modify static variable directly. Make an on-stack copy instead as ata_piix does. * pata_uli: It has 4 ports instead of 2. Don't use ata_pci_prepare_native_host(). Allocate the host explicitly and use init helpers. It's simple enough. Signed-off-by: Tejun Heo <htejun@gmail.com> Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-05-04 17:43:58 +07:00
ppi[0] = &info_smart;
return ata_pci_sff_init_one(pdev, ppi, &it821x_sht, NULL);
}
#ifdef CONFIG_PM
static int it821x_reinit_one(struct pci_dev *pdev)
{
struct ata_host *host = dev_get_drvdata(&pdev->dev);
int rc;
rc = ata_pci_device_do_resume(pdev);
if (rc)
return rc;
/* Resume - turn raid back off if need be */
if (it8212_noraid)
it821x_disable_raid(pdev);
ata_host_resume(host);
return rc;
}
#endif
static const struct pci_device_id it821x[] = {
{ PCI_VDEVICE(ITE, PCI_DEVICE_ID_ITE_8211), },
{ PCI_VDEVICE(ITE, PCI_DEVICE_ID_ITE_8212), },
{ },
};
static struct pci_driver it821x_pci_driver = {
.name = DRV_NAME,
.id_table = it821x,
.probe = it821x_init_one,
.remove = ata_pci_remove_one,
#ifdef CONFIG_PM
.suspend = ata_pci_device_suspend,
.resume = it821x_reinit_one,
#endif
};
static int __init it821x_init(void)
{
return pci_register_driver(&it821x_pci_driver);
}
static void __exit it821x_exit(void)
{
pci_unregister_driver(&it821x_pci_driver);
}
MODULE_AUTHOR("Alan Cox");
MODULE_DESCRIPTION("low-level driver for the IT8211/IT8212 IDE RAID controller");
MODULE_LICENSE("GPL");
MODULE_DEVICE_TABLE(pci, it821x);
MODULE_VERSION(DRV_VERSION);
module_param_named(noraid, it8212_noraid, int, S_IRUGO);
MODULE_PARM_DESC(noraid, "Force card into bypass mode");
module_init(it821x_init);
module_exit(it821x_exit);