1
linux/drivers/scsi/53c700.c

2180 lines
70 KiB
C
Raw Normal View History

/* -*- mode: c; c-basic-offset: 8 -*- */
/* NCR (or Symbios) 53c700 and 53c700-66 Driver
*
* Copyright (C) 2001 by James.Bottomley@HansenPartnership.com
**-----------------------------------------------------------------------------
**
** 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., 675 Mass Ave, Cambridge, MA 02139, USA.
**
**-----------------------------------------------------------------------------
*/
/* Notes:
*
* This driver is designed exclusively for these chips (virtually the
* earliest of the scripts engine chips). They need their own drivers
* because they are missing so many of the scripts and snazzy register
* features of their elder brothers (the 710, 720 and 770).
*
* The 700 is the lowliest of the line, it can only do async SCSI.
* The 700-66 can at least do synchronous SCSI up to 10MHz.
*
* The 700 chip has no host bus interface logic of its own. However,
* it is usually mapped to a location with well defined register
* offsets. Therefore, if you can determine the base address and the
* irq your board incorporating this chip uses, you can probably use
* this driver to run it (although you'll probably have to write a
* minimal wrapper for the purpose---see the NCR_D700 driver for
* details about how to do this).
*
*
* TODO List:
*
* 1. Better statistics in the proc fs
*
* 2. Implement message queue (queues SCSI messages like commands) and make
* the abort and device reset functions use them.
* */
/* CHANGELOG
*
* Version 2.8
*
* Fixed bad bug affecting tag starvation processing (previously the
* driver would hang the system if too many tags starved. Also fixed
* bad bug having to do with 10 byte command processing and REQUEST
* SENSE (the command would loop forever getting a transfer length
* mismatch in the CMD phase).
*
* Version 2.7
*
* Fixed scripts problem which caused certain devices (notably CDRWs)
* to hang on initial INQUIRY. Updated NCR_700_readl/writel to use
* __raw_readl/writel for parisc compatibility (Thomas
* Bogendoerfer). Added missing SCp->request_bufflen initialisation
* for sense requests (Ryan Bradetich).
*
* Version 2.6
*
* Following test of the 64 bit parisc kernel by Richard Hirst,
* several problems have now been corrected. Also adds support for
* consistent memory allocation.
*
* Version 2.5
*
* More Compatibility changes for 710 (now actually works). Enhanced
* support for odd clock speeds which constrain SDTR negotiations.
* correct cacheline separation for scsi messages and status for
* incoherent architectures. Use of the pci mapping functions on
* buffers to begin support for 64 bit drivers.
*
* Version 2.4
*
* Added support for the 53c710 chip (in 53c700 emulation mode only---no
* special 53c710 instructions or registers are used).
*
* Version 2.3
*
* More endianness/cache coherency changes.
*
* Better bad device handling (handles devices lying about tag
* queueing support and devices which fail to provide sense data on
* contingent allegiance conditions)
*
* Many thanks to Richard Hirst <rhirst@linuxcare.com> for patiently
* debugging this driver on the parisc architecture and suggesting
* many improvements and bug fixes.
*
* Thanks also go to Linuxcare Inc. for providing several PARISC
* machines for me to debug the driver on.
*
* Version 2.2
*
* Made the driver mem or io mapped; added endian invariance; added
* dma cache flushing operations for architectures which need it;
* added support for more varied clocking speeds.
*
* Version 2.1
*
* Initial modularisation from the D700. See NCR_D700.c for the rest of
* the changelog.
* */
#define NCR_700_VERSION "2.8"
#include <linux/kernel.h>
#include <linux/types.h>
#include <linux/string.h>
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h percpu.h is included by sched.h and module.h and thus ends up being included when building most .c files. percpu.h includes slab.h which in turn includes gfp.h making everything defined by the two files universally available and complicating inclusion dependencies. percpu.h -> slab.h dependency is about to be removed. Prepare for this change by updating users of gfp and slab facilities include those headers directly instead of assuming availability. As this conversion needs to touch large number of source files, the following script is used as the basis of conversion. http://userweb.kernel.org/~tj/misc/slabh-sweep.py The script does the followings. * Scan files for gfp and slab usages and update includes such that only the necessary includes are there. ie. if only gfp is used, gfp.h, if slab is used, slab.h. * When the script inserts a new include, it looks at the include blocks and try to put the new include such that its order conforms to its surrounding. It's put in the include block which contains core kernel includes, in the same order that the rest are ordered - alphabetical, Christmas tree, rev-Xmas-tree or at the end if there doesn't seem to be any matching order. * If the script can't find a place to put a new include (mostly because the file doesn't have fitting include block), it prints out an error message indicating which .h file needs to be added to the file. The conversion was done in the following steps. 1. The initial automatic conversion of all .c files updated slightly over 4000 files, deleting around 700 includes and adding ~480 gfp.h and ~3000 slab.h inclusions. The script emitted errors for ~400 files. 2. Each error was manually checked. Some didn't need the inclusion, some needed manual addition while adding it to implementation .h or embedding .c file was more appropriate for others. This step added inclusions to around 150 files. 3. The script was run again and the output was compared to the edits from #2 to make sure no file was left behind. 4. Several build tests were done and a couple of problems were fixed. e.g. lib/decompress_*.c used malloc/free() wrappers around slab APIs requiring slab.h to be added manually. 5. The script was run on all .h files but without automatically editing them as sprinkling gfp.h and slab.h inclusions around .h files could easily lead to inclusion dependency hell. Most gfp.h inclusion directives were ignored as stuff from gfp.h was usually wildly available and often used in preprocessor macros. Each slab.h inclusion directive was examined and added manually as necessary. 6. percpu.h was updated not to include slab.h. 7. Build test were done on the following configurations and failures were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my distributed build env didn't work with gcov compiles) and a few more options had to be turned off depending on archs to make things build (like ipr on powerpc/64 which failed due to missing writeq). * x86 and x86_64 UP and SMP allmodconfig and a custom test config. * powerpc and powerpc64 SMP allmodconfig * sparc and sparc64 SMP allmodconfig * ia64 SMP allmodconfig * s390 SMP allmodconfig * alpha SMP allmodconfig * um on x86_64 SMP allmodconfig 8. percpu.h modifications were reverted so that it could be applied as a separate patch and serve as bisection point. Given the fact that I had only a couple of failures from tests on step 6, I'm fairly confident about the coverage of this conversion patch. If there is a breakage, it's likely to be something in one of the arch headers which should be easily discoverable easily on most builds of the specific arch. Signed-off-by: Tejun Heo <tj@kernel.org> Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org> Cc: Ingo Molnar <mingo@redhat.com> Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 01:04:11 -07:00
#include <linux/slab.h>
#include <linux/ioport.h>
#include <linux/delay.h>
#include <linux/spinlock.h>
#include <linux/completion.h>
#include <linux/init.h>
#include <linux/proc_fs.h>
#include <linux/blkdev.h>
#include <linux/module.h>
#include <linux/interrupt.h>
#include <linux/device.h>
#include <asm/dma.h>
#include <asm/system.h>
#include <asm/io.h>
#include <asm/pgtable.h>
#include <asm/byteorder.h>
#include <scsi/scsi.h>
#include <scsi/scsi_cmnd.h>
#include <scsi/scsi_dbg.h>
#include <scsi/scsi_eh.h>
#include <scsi/scsi_host.h>
#include <scsi/scsi_tcq.h>
#include <scsi/scsi_transport.h>
#include <scsi/scsi_transport_spi.h>
#include "53c700.h"
/* NOTE: For 64 bit drivers there are points in the code where we use
* a non dereferenceable pointer to point to a structure in dma-able
* memory (which is 32 bits) so that we can use all of the structure
* operations but take the address at the end. This macro allows us
* to truncate the 64 bit pointer down to 32 bits without the compiler
* complaining */
#define to32bit(x) ((__u32)((unsigned long)(x)))
#ifdef NCR_700_DEBUG
#define STATIC
#else
#define STATIC static
#endif
MODULE_AUTHOR("James Bottomley");
MODULE_DESCRIPTION("53c700 and 53c700-66 Driver");
MODULE_LICENSE("GPL");
/* This is the script */
#include "53c700_d.h"
STATIC int NCR_700_queuecommand(struct scsi_cmnd *, void (*done)(struct scsi_cmnd *));
STATIC int NCR_700_abort(struct scsi_cmnd * SCpnt);
STATIC int NCR_700_bus_reset(struct scsi_cmnd * SCpnt);
STATIC int NCR_700_host_reset(struct scsi_cmnd * SCpnt);
STATIC void NCR_700_chip_setup(struct Scsi_Host *host);
STATIC void NCR_700_chip_reset(struct Scsi_Host *host);
STATIC int NCR_700_slave_alloc(struct scsi_device *SDpnt);
STATIC int NCR_700_slave_configure(struct scsi_device *SDpnt);
STATIC void NCR_700_slave_destroy(struct scsi_device *SDpnt);
static int NCR_700_change_queue_depth(struct scsi_device *SDpnt, int depth, int reason);
static int NCR_700_change_queue_type(struct scsi_device *SDpnt, int depth);
STATIC struct device_attribute *NCR_700_dev_attrs[];
STATIC struct scsi_transport_template *NCR_700_transport_template = NULL;
static char *NCR_700_phase[] = {
"",
"after selection",
"before command phase",
"after command phase",
"after status phase",
"after data in phase",
"after data out phase",
"during data phase",
};
static char *NCR_700_condition[] = {
"",
"NOT MSG_OUT",
"UNEXPECTED PHASE",
"NOT MSG_IN",
"UNEXPECTED MSG",
"MSG_IN",
"SDTR_MSG RECEIVED",
"REJECT_MSG RECEIVED",
"DISCONNECT_MSG RECEIVED",
"MSG_OUT",
"DATA_IN",
};
static char *NCR_700_fatal_messages[] = {
"unexpected message after reselection",
"still MSG_OUT after message injection",
"not MSG_IN after selection",
"Illegal message length received",
};
static char *NCR_700_SBCL_bits[] = {
"IO ",
"CD ",
"MSG ",
"ATN ",
"SEL ",
"BSY ",
"ACK ",
"REQ ",
};
static char *NCR_700_SBCL_to_phase[] = {
"DATA_OUT",
"DATA_IN",
"CMD_OUT",
"STATE",
"ILLEGAL PHASE",
"ILLEGAL PHASE",
"MSG OUT",
"MSG IN",
};
/* This translates the SDTR message offset and period to a value
* which can be loaded into the SXFER_REG.
*
* NOTE: According to SCSI-2, the true transfer period (in ns) is
* actually four times this period value */
static inline __u8
NCR_700_offset_period_to_sxfer(struct NCR_700_Host_Parameters *hostdata,
__u8 offset, __u8 period)
{
int XFERP;
__u8 min_xferp = (hostdata->chip710
? NCR_710_MIN_XFERP : NCR_700_MIN_XFERP);
__u8 max_offset = (hostdata->chip710
? NCR_710_MAX_OFFSET : NCR_700_MAX_OFFSET);
if(offset == 0)
return 0;
if(period < hostdata->min_period) {
printk(KERN_WARNING "53c700: Period %dns is less than this chip's minimum, setting to %d\n", period*4, NCR_700_MIN_PERIOD*4);
period = hostdata->min_period;
}
XFERP = (period*4 * hostdata->sync_clock)/1000 - 4;
if(offset > max_offset) {
printk(KERN_WARNING "53c700: Offset %d exceeds chip maximum, setting to %d\n",
offset, max_offset);
offset = max_offset;
}
if(XFERP < min_xferp) {
XFERP = min_xferp;
}
return (offset & 0x0f) | (XFERP & 0x07)<<4;
}
static inline __u8
NCR_700_get_SXFER(struct scsi_device *SDp)
{
struct NCR_700_Host_Parameters *hostdata =
(struct NCR_700_Host_Parameters *)SDp->host->hostdata[0];
return NCR_700_offset_period_to_sxfer(hostdata,
spi_offset(SDp->sdev_target),
spi_period(SDp->sdev_target));
}
struct Scsi_Host *
NCR_700_detect(struct scsi_host_template *tpnt,
struct NCR_700_Host_Parameters *hostdata, struct device *dev)
{
dma_addr_t pScript, pSlots;
__u8 *memory;
__u32 *script;
struct Scsi_Host *host;
static int banner = 0;
int j;
if(tpnt->sdev_attrs == NULL)
tpnt->sdev_attrs = NCR_700_dev_attrs;
memory = dma_alloc_noncoherent(hostdata->dev, TOTAL_MEM_SIZE,
&pScript, GFP_KERNEL);
if(memory == NULL) {
printk(KERN_ERR "53c700: Failed to allocate memory for driver, detatching\n");
return NULL;
}
script = (__u32 *)memory;
hostdata->msgin = memory + MSGIN_OFFSET;
hostdata->msgout = memory + MSGOUT_OFFSET;
hostdata->status = memory + STATUS_OFFSET;
/* all of these offsets are L1_CACHE_BYTES separated. It is fatal
* if this isn't sufficient separation to avoid dma flushing issues */
BUG_ON(!dma_is_consistent(hostdata->dev, pScript) && L1_CACHE_BYTES < dma_get_cache_alignment());
hostdata->slots = (struct NCR_700_command_slot *)(memory + SLOTS_OFFSET);
hostdata->dev = dev;
pSlots = pScript + SLOTS_OFFSET;
/* Fill in the missing routines from the host template */
tpnt->queuecommand = NCR_700_queuecommand;
tpnt->eh_abort_handler = NCR_700_abort;
tpnt->eh_bus_reset_handler = NCR_700_bus_reset;
tpnt->eh_host_reset_handler = NCR_700_host_reset;
tpnt->can_queue = NCR_700_COMMAND_SLOTS_PER_HOST;
tpnt->sg_tablesize = NCR_700_SG_SEGMENTS;
tpnt->cmd_per_lun = NCR_700_CMD_PER_LUN;
tpnt->use_clustering = ENABLE_CLUSTERING;
tpnt->slave_configure = NCR_700_slave_configure;
tpnt->slave_destroy = NCR_700_slave_destroy;
tpnt->slave_alloc = NCR_700_slave_alloc;
tpnt->change_queue_depth = NCR_700_change_queue_depth;
tpnt->change_queue_type = NCR_700_change_queue_type;
if(tpnt->name == NULL)
tpnt->name = "53c700";
if(tpnt->proc_name == NULL)
tpnt->proc_name = "53c700";
host = scsi_host_alloc(tpnt, 4);
if (!host)
return NULL;
memset(hostdata->slots, 0, sizeof(struct NCR_700_command_slot)
* NCR_700_COMMAND_SLOTS_PER_HOST);
for (j = 0; j < NCR_700_COMMAND_SLOTS_PER_HOST; j++) {
dma_addr_t offset = (dma_addr_t)((unsigned long)&hostdata->slots[j].SG[0]
- (unsigned long)&hostdata->slots[0].SG[0]);
hostdata->slots[j].pSG = (struct NCR_700_SG_List *)((unsigned long)(pSlots + offset));
if(j == 0)
hostdata->free_list = &hostdata->slots[j];
else
hostdata->slots[j-1].ITL_forw = &hostdata->slots[j];
hostdata->slots[j].state = NCR_700_SLOT_FREE;
}
for (j = 0; j < ARRAY_SIZE(SCRIPT); j++)
script[j] = bS_to_host(SCRIPT[j]);
/* adjust all labels to be bus physical */
for (j = 0; j < PATCHES; j++)
script[LABELPATCHES[j]] = bS_to_host(pScript + SCRIPT[LABELPATCHES[j]]);
/* now patch up fixed addresses. */
script_patch_32(hostdata->dev, script, MessageLocation,
pScript + MSGOUT_OFFSET);
script_patch_32(hostdata->dev, script, StatusAddress,
pScript + STATUS_OFFSET);
script_patch_32(hostdata->dev, script, ReceiveMsgAddress,
pScript + MSGIN_OFFSET);
hostdata->script = script;
hostdata->pScript = pScript;
dma_sync_single_for_device(hostdata->dev, pScript, sizeof(SCRIPT), DMA_TO_DEVICE);
hostdata->state = NCR_700_HOST_FREE;
hostdata->cmd = NULL;
host->max_id = 8;
host->max_lun = NCR_700_MAX_LUNS;
BUG_ON(NCR_700_transport_template == NULL);
host->transportt = NCR_700_transport_template;
host->unique_id = (unsigned long)hostdata->base;
hostdata->eh_complete = NULL;
host->hostdata[0] = (unsigned long)hostdata;
/* kick the chip */
NCR_700_writeb(0xff, host, CTEST9_REG);
if (hostdata->chip710)
hostdata->rev = (NCR_700_readb(host, CTEST8_REG)>>4) & 0x0f;
else
hostdata->rev = (NCR_700_readb(host, CTEST7_REG)>>4) & 0x0f;
hostdata->fast = (NCR_700_readb(host, CTEST9_REG) == 0);
if (banner == 0) {
printk(KERN_NOTICE "53c700: Version " NCR_700_VERSION " By James.Bottomley@HansenPartnership.com\n");
banner = 1;
}
printk(KERN_NOTICE "scsi%d: %s rev %d %s\n", host->host_no,
hostdata->chip710 ? "53c710" :
(hostdata->fast ? "53c700-66" : "53c700"),
hostdata->rev, hostdata->differential ?
"(Differential)" : "");
/* reset the chip */
NCR_700_chip_reset(host);
if (scsi_add_host(host, dev)) {
dev_printk(KERN_ERR, dev, "53c700: scsi_add_host failed\n");
scsi_host_put(host);
return NULL;
}
spi_signalling(host) = hostdata->differential ? SPI_SIGNAL_HVD :
SPI_SIGNAL_SE;
return host;
}
int
NCR_700_release(struct Scsi_Host *host)
{
struct NCR_700_Host_Parameters *hostdata =
(struct NCR_700_Host_Parameters *)host->hostdata[0];
dma_free_noncoherent(hostdata->dev, TOTAL_MEM_SIZE,
hostdata->script, hostdata->pScript);
return 1;
}
static inline __u8
NCR_700_identify(int can_disconnect, __u8 lun)
{
return IDENTIFY_BASE |
((can_disconnect) ? 0x40 : 0) |
(lun & NCR_700_LUN_MASK);
}
/*
* Function : static int data_residual (Scsi_Host *host)
*
* Purpose : return residual data count of what's in the chip. If you
* really want to know what this function is doing, it's almost a
* direct transcription of the algorithm described in the 53c710
* guide, except that the DBC and DFIFO registers are only 6 bits
* wide on a 53c700.
*
* Inputs : host - SCSI host */
static inline int
NCR_700_data_residual (struct Scsi_Host *host) {
struct NCR_700_Host_Parameters *hostdata =
(struct NCR_700_Host_Parameters *)host->hostdata[0];
int count, synchronous = 0;
unsigned int ddir;
if(hostdata->chip710) {
count = ((NCR_700_readb(host, DFIFO_REG) & 0x7f) -
(NCR_700_readl(host, DBC_REG) & 0x7f)) & 0x7f;
} else {
count = ((NCR_700_readb(host, DFIFO_REG) & 0x3f) -
(NCR_700_readl(host, DBC_REG) & 0x3f)) & 0x3f;
}
if(hostdata->fast)
synchronous = NCR_700_readb(host, SXFER_REG) & 0x0f;
/* get the data direction */
ddir = NCR_700_readb(host, CTEST0_REG) & 0x01;
if (ddir) {
/* Receive */
if (synchronous)
count += (NCR_700_readb(host, SSTAT2_REG) & 0xf0) >> 4;
else
if (NCR_700_readb(host, SSTAT1_REG) & SIDL_REG_FULL)
++count;
} else {
/* Send */
__u8 sstat = NCR_700_readb(host, SSTAT1_REG);
if (sstat & SODL_REG_FULL)
++count;
if (synchronous && (sstat & SODR_REG_FULL))
++count;
}
#ifdef NCR_700_DEBUG
if(count)
printk("RESIDUAL IS %d (ddir %d)\n", count, ddir);
#endif
return count;
}
/* print out the SCSI wires and corresponding phase from the SBCL register
* in the chip */
static inline char *
sbcl_to_string(__u8 sbcl)
{
int i;
static char ret[256];
ret[0]='\0';
for(i=0; i<8; i++) {
if((1<<i) & sbcl)
strcat(ret, NCR_700_SBCL_bits[i]);
}
strcat(ret, NCR_700_SBCL_to_phase[sbcl & 0x07]);
return ret;
}
static inline __u8
bitmap_to_number(__u8 bitmap)
{
__u8 i;
for(i=0; i<8 && !(bitmap &(1<<i)); i++)
;
return i;
}
/* Pull a slot off the free list */
STATIC struct NCR_700_command_slot *
find_empty_slot(struct NCR_700_Host_Parameters *hostdata)
{
struct NCR_700_command_slot *slot = hostdata->free_list;
if(slot == NULL) {
/* sanity check */
if(hostdata->command_slot_count != NCR_700_COMMAND_SLOTS_PER_HOST)
printk(KERN_ERR "SLOTS FULL, but count is %d, should be %d\n", hostdata->command_slot_count, NCR_700_COMMAND_SLOTS_PER_HOST);
return NULL;
}
if(slot->state != NCR_700_SLOT_FREE)
/* should panic! */
printk(KERN_ERR "BUSY SLOT ON FREE LIST!!!\n");
hostdata->free_list = slot->ITL_forw;
slot->ITL_forw = NULL;
/* NOTE: set the state to busy here, not queued, since this
* indicates the slot is in use and cannot be run by the IRQ
* finish routine. If we cannot queue the command when it
* is properly build, we then change to NCR_700_SLOT_QUEUED */
slot->state = NCR_700_SLOT_BUSY;
slot->flags = 0;
hostdata->command_slot_count++;
return slot;
}
STATIC void
free_slot(struct NCR_700_command_slot *slot,
struct NCR_700_Host_Parameters *hostdata)
{
if((slot->state & NCR_700_SLOT_MASK) != NCR_700_SLOT_MAGIC) {
printk(KERN_ERR "53c700: SLOT %p is not MAGIC!!!\n", slot);
}
if(slot->state == NCR_700_SLOT_FREE) {
printk(KERN_ERR "53c700: SLOT %p is FREE!!!\n", slot);
}
slot->resume_offset = 0;
slot->cmnd = NULL;
slot->state = NCR_700_SLOT_FREE;
slot->ITL_forw = hostdata->free_list;
hostdata->free_list = slot;
hostdata->command_slot_count--;
}
/* This routine really does very little. The command is indexed on
the ITL and (if tagged) the ITLQ lists in _queuecommand */
STATIC void
save_for_reselection(struct NCR_700_Host_Parameters *hostdata,
struct scsi_cmnd *SCp, __u32 dsp)
{
/* Its just possible that this gets executed twice */
if(SCp != NULL) {
struct NCR_700_command_slot *slot =
(struct NCR_700_command_slot *)SCp->host_scribble;
slot->resume_offset = dsp;
}
hostdata->state = NCR_700_HOST_FREE;
hostdata->cmd = NULL;
}
STATIC inline void
NCR_700_unmap(struct NCR_700_Host_Parameters *hostdata, struct scsi_cmnd *SCp,
struct NCR_700_command_slot *slot)
{
if(SCp->sc_data_direction != DMA_NONE &&
SCp->sc_data_direction != DMA_BIDIRECTIONAL)
scsi_dma_unmap(SCp);
}
STATIC inline void
NCR_700_scsi_done(struct NCR_700_Host_Parameters *hostdata,
struct scsi_cmnd *SCp, int result)
{
hostdata->state = NCR_700_HOST_FREE;
hostdata->cmd = NULL;
if(SCp != NULL) {
struct NCR_700_command_slot *slot =
(struct NCR_700_command_slot *)SCp->host_scribble;
dma_unmap_single(hostdata->dev, slot->pCmd,
[SCSI] Let scsi_cmnd->cmnd use request->cmd buffer - struct scsi_cmnd had a 16 bytes command buffer of its own. This is an unnecessary duplication and copy of request's cmd. It is probably left overs from the time that scsi_cmnd could function without a request attached. So clean that up. - Once above is done, few places, apart from scsi-ml, needed adjustments due to changing the data type of scsi_cmnd->cmnd. - Lots of drivers still use MAX_COMMAND_SIZE. So I have left that #define but equate it to BLK_MAX_CDB. The way I see it and is reflected in the patch below is. MAX_COMMAND_SIZE - means: The longest fixed-length (*) SCSI CDB as per the SCSI standard and is not related to the implementation. BLK_MAX_CDB. - The allocated space at the request level - I have audit all ISA drivers and made sure none use ->cmnd in a DMA Operation. Same audit was done by Andi Kleen. (*)fixed-length here means commands that their size can be determined by their opcode and the CDB does not carry a length specifier, (unlike the VARIABLE_LENGTH_CMD(0x7f) command). This is actually not exactly true and the SCSI standard also defines extended commands and vendor specific commands that can be bigger than 16 bytes. The kernel will support these using the same infrastructure used for VARLEN CDB's. So in effect MAX_COMMAND_SIZE means the maximum size command scsi-ml supports without specifying a cmd_len by ULD's Signed-off-by: Boaz Harrosh <bharrosh@panasas.com> Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
2008-04-30 01:19:47 -07:00
MAX_COMMAND_SIZE, DMA_TO_DEVICE);
if (slot->flags == NCR_700_FLAG_AUTOSENSE) {
char *cmnd = NCR_700_get_sense_cmnd(SCp->device);
#ifdef NCR_700_DEBUG
printk(" ORIGINAL CMD %p RETURNED %d, new return is %d sense is\n",
SCp, SCp->cmnd[7], result);
scsi_print_sense("53c700", SCp);
#endif
dma_unmap_single(hostdata->dev, slot->dma_handle,
SCSI_SENSE_BUFFERSIZE, DMA_FROM_DEVICE);
/* restore the old result if the request sense was
* successful */
if (result == 0)
result = cmnd[7];
/* restore the original length */
SCp->cmd_len = cmnd[8];
} else
NCR_700_unmap(hostdata, SCp, slot);
free_slot(slot, hostdata);
#ifdef NCR_700_DEBUG
if(NCR_700_get_depth(SCp->device) == 0 ||
NCR_700_get_depth(SCp->device) > SCp->device->queue_depth)
printk(KERN_ERR "Invalid depth in NCR_700_scsi_done(): %d\n",
NCR_700_get_depth(SCp->device));
#endif /* NCR_700_DEBUG */
NCR_700_set_depth(SCp->device, NCR_700_get_depth(SCp->device) - 1);
SCp->host_scribble = NULL;
SCp->result = result;
SCp->scsi_done(SCp);
} else {
printk(KERN_ERR "53c700: SCSI DONE HAS NULL SCp\n");
}
}
STATIC void
NCR_700_internal_bus_reset(struct Scsi_Host *host)
{
/* Bus reset */
NCR_700_writeb(ASSERT_RST, host, SCNTL1_REG);
udelay(50);
NCR_700_writeb(0, host, SCNTL1_REG);
}
STATIC void
NCR_700_chip_setup(struct Scsi_Host *host)
{
struct NCR_700_Host_Parameters *hostdata =
(struct NCR_700_Host_Parameters *)host->hostdata[0];
__u8 min_period;
__u8 min_xferp = (hostdata->chip710 ? NCR_710_MIN_XFERP : NCR_700_MIN_XFERP);
if(hostdata->chip710) {
__u8 burst_disable = 0;
__u8 burst_length = 0;
switch (hostdata->burst_length) {
case 1:
burst_length = BURST_LENGTH_1;
break;
case 2:
burst_length = BURST_LENGTH_2;
break;
case 4:
burst_length = BURST_LENGTH_4;
break;
case 8:
burst_length = BURST_LENGTH_8;
break;
default:
burst_disable = BURST_DISABLE;
break;
}
hostdata->dcntl_extra |= COMPAT_700_MODE;
NCR_700_writeb(hostdata->dcntl_extra, host, DCNTL_REG);
NCR_700_writeb(burst_length | hostdata->dmode_extra,
host, DMODE_710_REG);
NCR_700_writeb(burst_disable | hostdata->ctest7_extra |
(hostdata->differential ? DIFF : 0),
host, CTEST7_REG);
NCR_700_writeb(BTB_TIMER_DISABLE, host, CTEST0_REG);
NCR_700_writeb(FULL_ARBITRATION | ENABLE_PARITY | PARITY
| AUTO_ATN, host, SCNTL0_REG);
} else {
NCR_700_writeb(BURST_LENGTH_8 | hostdata->dmode_extra,
host, DMODE_700_REG);
NCR_700_writeb(hostdata->differential ?
DIFF : 0, host, CTEST7_REG);
if(hostdata->fast) {
/* this is for 700-66, does nothing on 700 */
NCR_700_writeb(LAST_DIS_ENBL | ENABLE_ACTIVE_NEGATION
| GENERATE_RECEIVE_PARITY, host,
CTEST8_REG);
} else {
NCR_700_writeb(FULL_ARBITRATION | ENABLE_PARITY
| PARITY | AUTO_ATN, host, SCNTL0_REG);
}
}
NCR_700_writeb(1 << host->this_id, host, SCID_REG);
NCR_700_writeb(0, host, SBCL_REG);
NCR_700_writeb(ASYNC_OPERATION, host, SXFER_REG);
NCR_700_writeb(PHASE_MM_INT | SEL_TIMEOUT_INT | GROSS_ERR_INT | UX_DISC_INT
| RST_INT | PAR_ERR_INT | SELECT_INT, host, SIEN_REG);
NCR_700_writeb(ABORT_INT | INT_INST_INT | ILGL_INST_INT, host, DIEN_REG);
NCR_700_writeb(ENABLE_SELECT, host, SCNTL1_REG);
if(hostdata->clock > 75) {
printk(KERN_ERR "53c700: Clock speed %dMHz is too high: 75Mhz is the maximum this chip can be driven at\n", hostdata->clock);
/* do the best we can, but the async clock will be out
* of spec: sync divider 2, async divider 3 */
DEBUG(("53c700: sync 2 async 3\n"));
NCR_700_writeb(SYNC_DIV_2_0, host, SBCL_REG);
NCR_700_writeb(ASYNC_DIV_3_0 | hostdata->dcntl_extra, host, DCNTL_REG);
hostdata->sync_clock = hostdata->clock/2;
} else if(hostdata->clock > 50 && hostdata->clock <= 75) {
/* sync divider 1.5, async divider 3 */
DEBUG(("53c700: sync 1.5 async 3\n"));
NCR_700_writeb(SYNC_DIV_1_5, host, SBCL_REG);
NCR_700_writeb(ASYNC_DIV_3_0 | hostdata->dcntl_extra, host, DCNTL_REG);
hostdata->sync_clock = hostdata->clock*2;
hostdata->sync_clock /= 3;
} else if(hostdata->clock > 37 && hostdata->clock <= 50) {
/* sync divider 1, async divider 2 */
DEBUG(("53c700: sync 1 async 2\n"));
NCR_700_writeb(SYNC_DIV_1_0, host, SBCL_REG);
NCR_700_writeb(ASYNC_DIV_2_0 | hostdata->dcntl_extra, host, DCNTL_REG);
hostdata->sync_clock = hostdata->clock;
} else if(hostdata->clock > 25 && hostdata->clock <=37) {
/* sync divider 1, async divider 1.5 */
DEBUG(("53c700: sync 1 async 1.5\n"));
NCR_700_writeb(SYNC_DIV_1_0, host, SBCL_REG);
NCR_700_writeb(ASYNC_DIV_1_5 | hostdata->dcntl_extra, host, DCNTL_REG);
hostdata->sync_clock = hostdata->clock;
} else {
DEBUG(("53c700: sync 1 async 1\n"));
NCR_700_writeb(SYNC_DIV_1_0, host, SBCL_REG);
NCR_700_writeb(ASYNC_DIV_1_0 | hostdata->dcntl_extra, host, DCNTL_REG);
/* sync divider 1, async divider 1 */
hostdata->sync_clock = hostdata->clock;
}
/* Calculate the actual minimum period that can be supported
* by our synchronous clock speed. See the 710 manual for
* exact details of this calculation which is based on a
* setting of the SXFER register */
min_period = 1000*(4+min_xferp)/(4*hostdata->sync_clock);
hostdata->min_period = NCR_700_MIN_PERIOD;
if(min_period > NCR_700_MIN_PERIOD)
hostdata->min_period = min_period;
}
STATIC void
NCR_700_chip_reset(struct Scsi_Host *host)
{
struct NCR_700_Host_Parameters *hostdata =
(struct NCR_700_Host_Parameters *)host->hostdata[0];
if(hostdata->chip710) {
NCR_700_writeb(SOFTWARE_RESET_710, host, ISTAT_REG);
udelay(100);
NCR_700_writeb(0, host, ISTAT_REG);
} else {
NCR_700_writeb(SOFTWARE_RESET, host, DCNTL_REG);
udelay(100);
NCR_700_writeb(0, host, DCNTL_REG);
}
mdelay(1000);
NCR_700_chip_setup(host);
}
/* The heart of the message processing engine is that the instruction
* immediately after the INT is the normal case (and so must be CLEAR
* ACK). If we want to do something else, we call that routine in
* scripts and set temp to be the normal case + 8 (skipping the CLEAR
* ACK) so that the routine returns correctly to resume its activity
* */
STATIC __u32
process_extended_message(struct Scsi_Host *host,
struct NCR_700_Host_Parameters *hostdata,
struct scsi_cmnd *SCp, __u32 dsp, __u32 dsps)
{
__u32 resume_offset = dsp, temp = dsp + 8;
__u8 pun = 0xff, lun = 0xff;
if(SCp != NULL) {
pun = SCp->device->id;
lun = SCp->device->lun;
}
switch(hostdata->msgin[2]) {
case A_SDTR_MSG:
if(SCp != NULL && NCR_700_is_flag_set(SCp->device, NCR_700_DEV_BEGIN_SYNC_NEGOTIATION)) {
struct scsi_target *starget = SCp->device->sdev_target;
__u8 period = hostdata->msgin[3];
__u8 offset = hostdata->msgin[4];
if(offset == 0 || period == 0) {
offset = 0;
period = 0;
}
spi_offset(starget) = offset;
spi_period(starget) = period;
if(NCR_700_is_flag_set(SCp->device, NCR_700_DEV_PRINT_SYNC_NEGOTIATION)) {
spi_display_xfer_agreement(starget);
NCR_700_clear_flag(SCp->device, NCR_700_DEV_PRINT_SYNC_NEGOTIATION);
}
NCR_700_set_flag(SCp->device, NCR_700_DEV_NEGOTIATED_SYNC);
NCR_700_clear_flag(SCp->device, NCR_700_DEV_BEGIN_SYNC_NEGOTIATION);
NCR_700_writeb(NCR_700_get_SXFER(SCp->device),
host, SXFER_REG);
} else {
/* SDTR message out of the blue, reject it */
shost_printk(KERN_WARNING, host,
"Unexpected SDTR msg\n");
hostdata->msgout[0] = A_REJECT_MSG;
dma_cache_sync(hostdata->dev, hostdata->msgout, 1, DMA_TO_DEVICE);
script_patch_16(hostdata->dev, hostdata->script,
MessageCount, 1);
/* SendMsgOut returns, so set up the return
* address */
resume_offset = hostdata->pScript + Ent_SendMessageWithATN;
}
break;
case A_WDTR_MSG:
printk(KERN_INFO "scsi%d: (%d:%d), Unsolicited WDTR after CMD, Rejecting\n",
host->host_no, pun, lun);
hostdata->msgout[0] = A_REJECT_MSG;
dma_cache_sync(hostdata->dev, hostdata->msgout, 1, DMA_TO_DEVICE);
script_patch_16(hostdata->dev, hostdata->script, MessageCount,
1);
resume_offset = hostdata->pScript + Ent_SendMessageWithATN;
break;
default:
printk(KERN_INFO "scsi%d (%d:%d): Unexpected message %s: ",
host->host_no, pun, lun,
NCR_700_phase[(dsps & 0xf00) >> 8]);
spi_print_msg(hostdata->msgin);
printk("\n");
/* just reject it */
hostdata->msgout[0] = A_REJECT_MSG;
dma_cache_sync(hostdata->dev, hostdata->msgout, 1, DMA_TO_DEVICE);
script_patch_16(hostdata->dev, hostdata->script, MessageCount,
1);
/* SendMsgOut returns, so set up the return
* address */
resume_offset = hostdata->pScript + Ent_SendMessageWithATN;
}
NCR_700_writel(temp, host, TEMP_REG);
return resume_offset;
}
STATIC __u32
process_message(struct Scsi_Host *host, struct NCR_700_Host_Parameters *hostdata,
struct scsi_cmnd *SCp, __u32 dsp, __u32 dsps)
{
/* work out where to return to */
__u32 temp = dsp + 8, resume_offset = dsp;
__u8 pun = 0xff, lun = 0xff;
if(SCp != NULL) {
pun = SCp->device->id;
lun = SCp->device->lun;
}
#ifdef NCR_700_DEBUG
printk("scsi%d (%d:%d): message %s: ", host->host_no, pun, lun,
NCR_700_phase[(dsps & 0xf00) >> 8]);
spi_print_msg(hostdata->msgin);
printk("\n");
#endif
switch(hostdata->msgin[0]) {
case A_EXTENDED_MSG:
resume_offset = process_extended_message(host, hostdata, SCp,
dsp, dsps);
break;
case A_REJECT_MSG:
if(SCp != NULL && NCR_700_is_flag_set(SCp->device, NCR_700_DEV_BEGIN_SYNC_NEGOTIATION)) {
/* Rejected our sync negotiation attempt */
spi_period(SCp->device->sdev_target) =
spi_offset(SCp->device->sdev_target) = 0;
NCR_700_set_flag(SCp->device, NCR_700_DEV_NEGOTIATED_SYNC);
NCR_700_clear_flag(SCp->device, NCR_700_DEV_BEGIN_SYNC_NEGOTIATION);
} else if(SCp != NULL && NCR_700_get_tag_neg_state(SCp->device) == NCR_700_DURING_TAG_NEGOTIATION) {
/* rejected our first simple tag message */
scmd_printk(KERN_WARNING, SCp,
"Rejected first tag queue attempt, turning off tag queueing\n");
/* we're done negotiating */
NCR_700_set_tag_neg_state(SCp->device, NCR_700_FINISHED_TAG_NEGOTIATION);
hostdata->tag_negotiated &= ~(1<<scmd_id(SCp));
SCp->device->tagged_supported = 0;
scsi_deactivate_tcq(SCp->device, host->cmd_per_lun);
} else {
shost_printk(KERN_WARNING, host,
"(%d:%d) Unexpected REJECT Message %s\n",
pun, lun,
NCR_700_phase[(dsps & 0xf00) >> 8]);
/* however, just ignore it */
}
break;
case A_PARITY_ERROR_MSG:
printk(KERN_ERR "scsi%d (%d:%d) Parity Error!\n", host->host_no,
pun, lun);
NCR_700_internal_bus_reset(host);
break;
case A_SIMPLE_TAG_MSG:
printk(KERN_INFO "scsi%d (%d:%d) SIMPLE TAG %d %s\n", host->host_no,
pun, lun, hostdata->msgin[1],
NCR_700_phase[(dsps & 0xf00) >> 8]);
/* just ignore it */
break;
default:
printk(KERN_INFO "scsi%d (%d:%d): Unexpected message %s: ",
host->host_no, pun, lun,
NCR_700_phase[(dsps & 0xf00) >> 8]);
spi_print_msg(hostdata->msgin);
printk("\n");
/* just reject it */
hostdata->msgout[0] = A_REJECT_MSG;
dma_cache_sync(hostdata->dev, hostdata->msgout, 1, DMA_TO_DEVICE);
script_patch_16(hostdata->dev, hostdata->script, MessageCount,
1);
/* SendMsgOut returns, so set up the return
* address */
resume_offset = hostdata->pScript + Ent_SendMessageWithATN;
break;
}
NCR_700_writel(temp, host, TEMP_REG);
/* set us up to receive another message */
dma_cache_sync(hostdata->dev, hostdata->msgin, MSG_ARRAY_SIZE, DMA_FROM_DEVICE);
return resume_offset;
}
STATIC __u32
process_script_interrupt(__u32 dsps, __u32 dsp, struct scsi_cmnd *SCp,
struct Scsi_Host *host,
struct NCR_700_Host_Parameters *hostdata)
{
__u32 resume_offset = 0;
__u8 pun = 0xff, lun=0xff;
if(SCp != NULL) {
pun = SCp->device->id;
lun = SCp->device->lun;
}
if(dsps == A_GOOD_STATUS_AFTER_STATUS) {
DEBUG((" COMMAND COMPLETE, status=%02x\n",
hostdata->status[0]));
/* OK, if TCQ still under negotiation, we now know it works */
if (NCR_700_get_tag_neg_state(SCp->device) == NCR_700_DURING_TAG_NEGOTIATION)
NCR_700_set_tag_neg_state(SCp->device,
NCR_700_FINISHED_TAG_NEGOTIATION);
/* check for contingent allegiance contitions */
if(status_byte(hostdata->status[0]) == CHECK_CONDITION ||
status_byte(hostdata->status[0]) == COMMAND_TERMINATED) {
struct NCR_700_command_slot *slot =
(struct NCR_700_command_slot *)SCp->host_scribble;
if(slot->flags == NCR_700_FLAG_AUTOSENSE) {
/* OOPS: bad device, returning another
* contingent allegiance condition */
scmd_printk(KERN_ERR, SCp,
"broken device is looping in contingent allegiance: ignoring\n");
NCR_700_scsi_done(hostdata, SCp, hostdata->status[0]);
} else {
char *cmnd =
NCR_700_get_sense_cmnd(SCp->device);
#ifdef NCR_DEBUG
scsi_print_command(SCp);
printk(" cmd %p has status %d, requesting sense\n",
SCp, hostdata->status[0]);
#endif
/* we can destroy the command here
* because the contingent allegiance
* condition will cause a retry which
* will re-copy the command from the
* saved data_cmnd. We also unmap any
* data associated with the command
* here */
NCR_700_unmap(hostdata, SCp, slot);
dma_unmap_single(hostdata->dev, slot->pCmd,
[SCSI] Let scsi_cmnd->cmnd use request->cmd buffer - struct scsi_cmnd had a 16 bytes command buffer of its own. This is an unnecessary duplication and copy of request's cmd. It is probably left overs from the time that scsi_cmnd could function without a request attached. So clean that up. - Once above is done, few places, apart from scsi-ml, needed adjustments due to changing the data type of scsi_cmnd->cmnd. - Lots of drivers still use MAX_COMMAND_SIZE. So I have left that #define but equate it to BLK_MAX_CDB. The way I see it and is reflected in the patch below is. MAX_COMMAND_SIZE - means: The longest fixed-length (*) SCSI CDB as per the SCSI standard and is not related to the implementation. BLK_MAX_CDB. - The allocated space at the request level - I have audit all ISA drivers and made sure none use ->cmnd in a DMA Operation. Same audit was done by Andi Kleen. (*)fixed-length here means commands that their size can be determined by their opcode and the CDB does not carry a length specifier, (unlike the VARIABLE_LENGTH_CMD(0x7f) command). This is actually not exactly true and the SCSI standard also defines extended commands and vendor specific commands that can be bigger than 16 bytes. The kernel will support these using the same infrastructure used for VARLEN CDB's. So in effect MAX_COMMAND_SIZE means the maximum size command scsi-ml supports without specifying a cmd_len by ULD's Signed-off-by: Boaz Harrosh <bharrosh@panasas.com> Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
2008-04-30 01:19:47 -07:00
MAX_COMMAND_SIZE,
DMA_TO_DEVICE);
cmnd[0] = REQUEST_SENSE;
cmnd[1] = (SCp->device->lun & 0x7) << 5;
cmnd[2] = 0;
cmnd[3] = 0;
cmnd[4] = SCSI_SENSE_BUFFERSIZE;
cmnd[5] = 0;
/* Here's a quiet hack: the
* REQUEST_SENSE command is six bytes,
* so store a flag indicating that
* this was an internal sense request
* and the original status at the end
* of the command */
cmnd[6] = NCR_700_INTERNAL_SENSE_MAGIC;
cmnd[7] = hostdata->status[0];
cmnd[8] = SCp->cmd_len;
SCp->cmd_len = 6; /* command length for
* REQUEST_SENSE */
slot->pCmd = dma_map_single(hostdata->dev, cmnd, MAX_COMMAND_SIZE, DMA_TO_DEVICE);
slot->dma_handle = dma_map_single(hostdata->dev, SCp->sense_buffer, SCSI_SENSE_BUFFERSIZE, DMA_FROM_DEVICE);
slot->SG[0].ins = bS_to_host(SCRIPT_MOVE_DATA_IN | SCSI_SENSE_BUFFERSIZE);
slot->SG[0].pAddr = bS_to_host(slot->dma_handle);
slot->SG[1].ins = bS_to_host(SCRIPT_RETURN);
slot->SG[1].pAddr = 0;
slot->resume_offset = hostdata->pScript;
dma_cache_sync(hostdata->dev, slot->SG, sizeof(slot->SG[0])*2, DMA_TO_DEVICE);
dma_cache_sync(hostdata->dev, SCp->sense_buffer, SCSI_SENSE_BUFFERSIZE, DMA_FROM_DEVICE);
/* queue the command for reissue */
slot->state = NCR_700_SLOT_QUEUED;
slot->flags = NCR_700_FLAG_AUTOSENSE;
hostdata->state = NCR_700_HOST_FREE;
hostdata->cmd = NULL;
}
} else {
// Currently rely on the mid layer evaluation
// of the tag queuing capability
//
//if(status_byte(hostdata->status[0]) == GOOD &&
// SCp->cmnd[0] == INQUIRY && SCp->use_sg == 0) {
// /* Piggy back the tag queueing support
// * on this command */
// dma_sync_single_for_cpu(hostdata->dev,
// slot->dma_handle,
// SCp->request_bufflen,
// DMA_FROM_DEVICE);
// if(((char *)SCp->request_buffer)[7] & 0x02) {
// scmd_printk(KERN_INFO, SCp,
// "Enabling Tag Command Queuing\n");
// hostdata->tag_negotiated |= (1<<scmd_id(SCp));
// NCR_700_set_flag(SCp->device, NCR_700_DEV_BEGIN_TAG_QUEUEING);
// } else {
// NCR_700_clear_flag(SCp->device, NCR_700_DEV_BEGIN_TAG_QUEUEING);
// hostdata->tag_negotiated &= ~(1<<scmd_id(SCp));
// }
//}
NCR_700_scsi_done(hostdata, SCp, hostdata->status[0]);
}
} else if((dsps & 0xfffff0f0) == A_UNEXPECTED_PHASE) {
__u8 i = (dsps & 0xf00) >> 8;
scmd_printk(KERN_ERR, SCp, "UNEXPECTED PHASE %s (%s)\n",
NCR_700_phase[i],
sbcl_to_string(NCR_700_readb(host, SBCL_REG)));
scmd_printk(KERN_ERR, SCp, " len = %d, cmd =",
SCp->cmd_len);
scsi_print_command(SCp);
NCR_700_internal_bus_reset(host);
} else if((dsps & 0xfffff000) == A_FATAL) {
int i = (dsps & 0xfff);
printk(KERN_ERR "scsi%d: (%d:%d) FATAL ERROR: %s\n",
host->host_no, pun, lun, NCR_700_fatal_messages[i]);
if(dsps == A_FATAL_ILLEGAL_MSG_LENGTH) {
printk(KERN_ERR " msg begins %02x %02x\n",
hostdata->msgin[0], hostdata->msgin[1]);
}
NCR_700_internal_bus_reset(host);
} else if((dsps & 0xfffff0f0) == A_DISCONNECT) {
#ifdef NCR_700_DEBUG
__u8 i = (dsps & 0xf00) >> 8;
printk("scsi%d: (%d:%d), DISCONNECTED (%d) %s\n",
host->host_no, pun, lun,
i, NCR_700_phase[i]);
#endif
save_for_reselection(hostdata, SCp, dsp);
} else if(dsps == A_RESELECTION_IDENTIFIED) {
__u8 lun;
struct NCR_700_command_slot *slot;
__u8 reselection_id = hostdata->reselection_id;
struct scsi_device *SDp;
lun = hostdata->msgin[0] & 0x1f;
hostdata->reselection_id = 0xff;
DEBUG(("scsi%d: (%d:%d) RESELECTED!\n",
host->host_no, reselection_id, lun));
/* clear the reselection indicator */
SDp = __scsi_device_lookup(host, 0, reselection_id, lun);
if(unlikely(SDp == NULL)) {
printk(KERN_ERR "scsi%d: (%d:%d) HAS NO device\n",
host->host_no, reselection_id, lun);
BUG();
}
if(hostdata->msgin[1] == A_SIMPLE_TAG_MSG) {
struct scsi_cmnd *SCp = scsi_find_tag(SDp, hostdata->msgin[2]);
if(unlikely(SCp == NULL)) {
printk(KERN_ERR "scsi%d: (%d:%d) no saved request for tag %d\n",
host->host_no, reselection_id, lun, hostdata->msgin[2]);
BUG();
}
slot = (struct NCR_700_command_slot *)SCp->host_scribble;
DDEBUG(KERN_DEBUG, SDp,
"reselection is tag %d, slot %p(%d)\n",
hostdata->msgin[2], slot, slot->tag);
} else {
struct scsi_cmnd *SCp = scsi_find_tag(SDp, SCSI_NO_TAG);
if(unlikely(SCp == NULL)) {
sdev_printk(KERN_ERR, SDp,
"no saved request for untagged cmd\n");
BUG();
}
slot = (struct NCR_700_command_slot *)SCp->host_scribble;
}
if(slot == NULL) {
printk(KERN_ERR "scsi%d: (%d:%d) RESELECTED but no saved command (MSG = %02x %02x %02x)!!\n",
host->host_no, reselection_id, lun,
hostdata->msgin[0], hostdata->msgin[1],
hostdata->msgin[2]);
} else {
if(hostdata->state != NCR_700_HOST_BUSY)
printk(KERN_ERR "scsi%d: FATAL, host not busy during valid reselection!\n",
host->host_no);
resume_offset = slot->resume_offset;
hostdata->cmd = slot->cmnd;
/* re-patch for this command */
script_patch_32_abs(hostdata->dev, hostdata->script,
CommandAddress, slot->pCmd);
script_patch_16(hostdata->dev, hostdata->script,
CommandCount, slot->cmnd->cmd_len);
script_patch_32_abs(hostdata->dev, hostdata->script,
SGScriptStartAddress,
to32bit(&slot->pSG[0].ins));
/* Note: setting SXFER only works if we're
* still in the MESSAGE phase, so it is vital
* that ACK is still asserted when we process
* the reselection message. The resume offset
* should therefore always clear ACK */
NCR_700_writeb(NCR_700_get_SXFER(hostdata->cmd->device),
host, SXFER_REG);
dma_cache_sync(hostdata->dev, hostdata->msgin,
MSG_ARRAY_SIZE, DMA_FROM_DEVICE);
dma_cache_sync(hostdata->dev, hostdata->msgout,
MSG_ARRAY_SIZE, DMA_TO_DEVICE);
/* I'm just being paranoid here, the command should
* already have been flushed from the cache */
dma_cache_sync(hostdata->dev, slot->cmnd->cmnd,
slot->cmnd->cmd_len, DMA_TO_DEVICE);
}
} else if(dsps == A_RESELECTED_DURING_SELECTION) {
/* This section is full of debugging code because I've
* never managed to reach it. I think what happens is
* that, because the 700 runs with selection
* interrupts enabled the whole time that we take a
* selection interrupt before we manage to get to the
* reselected script interrupt */
__u8 reselection_id = NCR_700_readb(host, SFBR_REG);
struct NCR_700_command_slot *slot;
/* Take out our own ID */
reselection_id &= ~(1<<host->this_id);
/* I've never seen this happen, so keep this as a printk rather
* than a debug */
printk(KERN_INFO "scsi%d: (%d:%d) RESELECTION DURING SELECTION, dsp=%08x[%04x] state=%d, count=%d\n",
host->host_no, reselection_id, lun, dsp, dsp - hostdata->pScript, hostdata->state, hostdata->command_slot_count);
{
/* FIXME: DEBUGGING CODE */
__u32 SG = (__u32)bS_to_cpu(hostdata->script[A_SGScriptStartAddress_used[0]]);
int i;
for(i=0; i< NCR_700_COMMAND_SLOTS_PER_HOST; i++) {
if(SG >= to32bit(&hostdata->slots[i].pSG[0])
&& SG <= to32bit(&hostdata->slots[i].pSG[NCR_700_SG_SEGMENTS]))
break;
}
printk(KERN_INFO "IDENTIFIED SG segment as being %08x in slot %p, cmd %p, slot->resume_offset=%08x\n", SG, &hostdata->slots[i], hostdata->slots[i].cmnd, hostdata->slots[i].resume_offset);
SCp = hostdata->slots[i].cmnd;
}
if(SCp != NULL) {
slot = (struct NCR_700_command_slot *)SCp->host_scribble;
/* change slot from busy to queued to redo command */
slot->state = NCR_700_SLOT_QUEUED;
}
hostdata->cmd = NULL;
if(reselection_id == 0) {
if(hostdata->reselection_id == 0xff) {
printk(KERN_ERR "scsi%d: Invalid reselection during selection!!\n", host->host_no);
return 0;
} else {
printk(KERN_ERR "scsi%d: script reselected and we took a selection interrupt\n",
host->host_no);
reselection_id = hostdata->reselection_id;
}
} else {
/* convert to real ID */
reselection_id = bitmap_to_number(reselection_id);
}
hostdata->reselection_id = reselection_id;
/* just in case we have a stale simple tag message, clear it */
hostdata->msgin[1] = 0;
dma_cache_sync(hostdata->dev, hostdata->msgin,
MSG_ARRAY_SIZE, DMA_BIDIRECTIONAL);
if(hostdata->tag_negotiated & (1<<reselection_id)) {
resume_offset = hostdata->pScript + Ent_GetReselectionWithTag;
} else {
resume_offset = hostdata->pScript + Ent_GetReselectionData;
}
} else if(dsps == A_COMPLETED_SELECTION_AS_TARGET) {
/* we've just disconnected from the bus, do nothing since
* a return here will re-run the queued command slot
* that may have been interrupted by the initial selection */
DEBUG((" SELECTION COMPLETED\n"));
} else if((dsps & 0xfffff0f0) == A_MSG_IN) {
resume_offset = process_message(host, hostdata, SCp,
dsp, dsps);
} else if((dsps & 0xfffff000) == 0) {
__u8 i = (dsps & 0xf0) >> 4, j = (dsps & 0xf00) >> 8;
printk(KERN_ERR "scsi%d: (%d:%d), unhandled script condition %s %s at %04x\n",
host->host_no, pun, lun, NCR_700_condition[i],
NCR_700_phase[j], dsp - hostdata->pScript);
if(SCp != NULL) {
struct scatterlist *sg;
scsi_print_command(SCp);
scsi_for_each_sg(SCp, sg, scsi_sg_count(SCp) + 1, i) {
printk(KERN_INFO " SG[%d].length = %d, move_insn=%08x, addr %08x\n", i, sg->length, ((struct NCR_700_command_slot *)SCp->host_scribble)->SG[i].ins, ((struct NCR_700_command_slot *)SCp->host_scribble)->SG[i].pAddr);
}
}
NCR_700_internal_bus_reset(host);
} else if((dsps & 0xfffff000) == A_DEBUG_INTERRUPT) {
printk(KERN_NOTICE "scsi%d (%d:%d) DEBUG INTERRUPT %d AT %08x[%04x], continuing\n",
host->host_no, pun, lun, dsps & 0xfff, dsp, dsp - hostdata->pScript);
resume_offset = dsp;
} else {
printk(KERN_ERR "scsi%d: (%d:%d), unidentified script interrupt 0x%x at %04x\n",
host->host_no, pun, lun, dsps, dsp - hostdata->pScript);
NCR_700_internal_bus_reset(host);
}
return resume_offset;
}
/* We run the 53c700 with selection interrupts always enabled. This
* means that the chip may be selected as soon as the bus frees. On a
* busy bus, this can be before the scripts engine finishes its
* processing. Therefore, part of the selection processing has to be
* to find out what the scripts engine is doing and complete the
* function if necessary (i.e. process the pending disconnect or save
* the interrupted initial selection */
STATIC inline __u32
process_selection(struct Scsi_Host *host, __u32 dsp)
{
__u8 id = 0; /* Squash compiler warning */
int count = 0;
__u32 resume_offset = 0;
struct NCR_700_Host_Parameters *hostdata =
(struct NCR_700_Host_Parameters *)host->hostdata[0];
struct scsi_cmnd *SCp = hostdata->cmd;
__u8 sbcl;
for(count = 0; count < 5; count++) {
id = NCR_700_readb(host, hostdata->chip710 ?
CTEST9_REG : SFBR_REG);
/* Take out our own ID */
id &= ~(1<<host->this_id);
if(id != 0)
break;
udelay(5);
}
sbcl = NCR_700_readb(host, SBCL_REG);
if((sbcl & SBCL_IO) == 0) {
/* mark as having been selected rather than reselected */
id = 0xff;
} else {
/* convert to real ID */
hostdata->reselection_id = id = bitmap_to_number(id);
DEBUG(("scsi%d: Reselected by %d\n",
host->host_no, id));
}
if(hostdata->state == NCR_700_HOST_BUSY && SCp != NULL) {
struct NCR_700_command_slot *slot =
(struct NCR_700_command_slot *)SCp->host_scribble;
DEBUG((" ID %d WARNING: RESELECTION OF BUSY HOST, saving cmd %p, slot %p, addr %x [%04x], resume %x!\n", id, hostdata->cmd, slot, dsp, dsp - hostdata->pScript, resume_offset));
switch(dsp - hostdata->pScript) {
case Ent_Disconnect1:
case Ent_Disconnect2:
save_for_reselection(hostdata, SCp, Ent_Disconnect2 + hostdata->pScript);
break;
case Ent_Disconnect3:
case Ent_Disconnect4:
save_for_reselection(hostdata, SCp, Ent_Disconnect4 + hostdata->pScript);
break;
case Ent_Disconnect5:
case Ent_Disconnect6:
save_for_reselection(hostdata, SCp, Ent_Disconnect6 + hostdata->pScript);
break;
case Ent_Disconnect7:
case Ent_Disconnect8:
save_for_reselection(hostdata, SCp, Ent_Disconnect8 + hostdata->pScript);
break;
case Ent_Finish1:
case Ent_Finish2:
process_script_interrupt(A_GOOD_STATUS_AFTER_STATUS, dsp, SCp, host, hostdata);
break;
default:
slot->state = NCR_700_SLOT_QUEUED;
break;
}
}
hostdata->state = NCR_700_HOST_BUSY;
hostdata->cmd = NULL;
/* clear any stale simple tag message */
hostdata->msgin[1] = 0;
dma_cache_sync(hostdata->dev, hostdata->msgin, MSG_ARRAY_SIZE,
DMA_BIDIRECTIONAL);
if(id == 0xff) {
/* Selected as target, Ignore */
resume_offset = hostdata->pScript + Ent_SelectedAsTarget;
} else if(hostdata->tag_negotiated & (1<<id)) {
resume_offset = hostdata->pScript + Ent_GetReselectionWithTag;
} else {
resume_offset = hostdata->pScript + Ent_GetReselectionData;
}
return resume_offset;
}
static inline void
NCR_700_clear_fifo(struct Scsi_Host *host) {
const struct NCR_700_Host_Parameters *hostdata
= (struct NCR_700_Host_Parameters *)host->hostdata[0];
if(hostdata->chip710) {
NCR_700_writeb(CLR_FIFO_710, host, CTEST8_REG);
} else {
NCR_700_writeb(CLR_FIFO, host, DFIFO_REG);
}
}
static inline void
NCR_700_flush_fifo(struct Scsi_Host *host) {
const struct NCR_700_Host_Parameters *hostdata
= (struct NCR_700_Host_Parameters *)host->hostdata[0];
if(hostdata->chip710) {
NCR_700_writeb(FLUSH_DMA_FIFO_710, host, CTEST8_REG);
udelay(10);
NCR_700_writeb(0, host, CTEST8_REG);
} else {
NCR_700_writeb(FLUSH_DMA_FIFO, host, DFIFO_REG);
udelay(10);
NCR_700_writeb(0, host, DFIFO_REG);
}
}
/* The queue lock with interrupts disabled must be held on entry to
* this function */
STATIC int
NCR_700_start_command(struct scsi_cmnd *SCp)
{
struct NCR_700_command_slot *slot =
(struct NCR_700_command_slot *)SCp->host_scribble;
struct NCR_700_Host_Parameters *hostdata =
(struct NCR_700_Host_Parameters *)SCp->device->host->hostdata[0];
__u16 count = 1; /* for IDENTIFY message */
if(hostdata->state != NCR_700_HOST_FREE) {
/* keep this inside the lock to close the race window where
* the running command finishes on another CPU while we don't
* change the state to queued on this one */
slot->state = NCR_700_SLOT_QUEUED;
DEBUG(("scsi%d: host busy, queueing command %p, slot %p\n",
SCp->device->host->host_no, slot->cmnd, slot));
return 0;
}
hostdata->state = NCR_700_HOST_BUSY;
hostdata->cmd = SCp;
slot->state = NCR_700_SLOT_BUSY;
/* keep interrupts disabled until we have the command correctly
* set up so we cannot take a selection interrupt */
hostdata->msgout[0] = NCR_700_identify((SCp->cmnd[0] != REQUEST_SENSE &&
slot->flags != NCR_700_FLAG_AUTOSENSE),
SCp->device->lun);
/* for INQUIRY or REQUEST_SENSE commands, we cannot be sure
* if the negotiated transfer parameters still hold, so
* always renegotiate them */
if(SCp->cmnd[0] == INQUIRY || SCp->cmnd[0] == REQUEST_SENSE ||
slot->flags == NCR_700_FLAG_AUTOSENSE) {
NCR_700_clear_flag(SCp->device, NCR_700_DEV_NEGOTIATED_SYNC);
}
/* REQUEST_SENSE is asking for contingent I_T_L(_Q) status.
* If a contingent allegiance condition exists, the device
* will refuse all tags, so send the request sense as untagged
* */
if((hostdata->tag_negotiated & (1<<scmd_id(SCp)))
&& (slot->tag != SCSI_NO_TAG && SCp->cmnd[0] != REQUEST_SENSE &&
slot->flags != NCR_700_FLAG_AUTOSENSE)) {
count += scsi_populate_tag_msg(SCp, &hostdata->msgout[count]);
}
if(hostdata->fast &&
NCR_700_is_flag_clear(SCp->device, NCR_700_DEV_NEGOTIATED_SYNC)) {
count += spi_populate_sync_msg(&hostdata->msgout[count],
spi_period(SCp->device->sdev_target),
spi_offset(SCp->device->sdev_target));
NCR_700_set_flag(SCp->device, NCR_700_DEV_BEGIN_SYNC_NEGOTIATION);
}
script_patch_16(hostdata->dev, hostdata->script, MessageCount, count);
script_patch_ID(hostdata->dev, hostdata->script,
Device_ID, 1<<scmd_id(SCp));
script_patch_32_abs(hostdata->dev, hostdata->script, CommandAddress,
slot->pCmd);
script_patch_16(hostdata->dev, hostdata->script, CommandCount,
SCp->cmd_len);
/* finally plumb the beginning of the SG list into the script
* */
script_patch_32_abs(hostdata->dev, hostdata->script,
SGScriptStartAddress, to32bit(&slot->pSG[0].ins));
NCR_700_clear_fifo(SCp->device->host);
if(slot->resume_offset == 0)
slot->resume_offset = hostdata->pScript;
/* now perform all the writebacks and invalidates */
dma_cache_sync(hostdata->dev, hostdata->msgout, count, DMA_TO_DEVICE);
dma_cache_sync(hostdata->dev, hostdata->msgin, MSG_ARRAY_SIZE,
DMA_FROM_DEVICE);
dma_cache_sync(hostdata->dev, SCp->cmnd, SCp->cmd_len, DMA_TO_DEVICE);
dma_cache_sync(hostdata->dev, hostdata->status, 1, DMA_FROM_DEVICE);
/* set the synchronous period/offset */
NCR_700_writeb(NCR_700_get_SXFER(SCp->device),
SCp->device->host, SXFER_REG);
NCR_700_writel(slot->temp, SCp->device->host, TEMP_REG);
NCR_700_writel(slot->resume_offset, SCp->device->host, DSP_REG);
return 1;
}
irqreturn_t
IRQ: Maintain regs pointer globally rather than passing to IRQ handlers Maintain a per-CPU global "struct pt_regs *" variable which can be used instead of passing regs around manually through all ~1800 interrupt handlers in the Linux kernel. The regs pointer is used in few places, but it potentially costs both stack space and code to pass it around. On the FRV arch, removing the regs parameter from all the genirq function results in a 20% speed up of the IRQ exit path (ie: from leaving timer_interrupt() to leaving do_IRQ()). Where appropriate, an arch may override the generic storage facility and do something different with the variable. On FRV, for instance, the address is maintained in GR28 at all times inside the kernel as part of general exception handling. Having looked over the code, it appears that the parameter may be handed down through up to twenty or so layers of functions. Consider a USB character device attached to a USB hub, attached to a USB controller that posts its interrupts through a cascaded auxiliary interrupt controller. A character device driver may want to pass regs to the sysrq handler through the input layer which adds another few layers of parameter passing. I've build this code with allyesconfig for x86_64 and i386. I've runtested the main part of the code on FRV and i386, though I can't test most of the drivers. I've also done partial conversion for powerpc and MIPS - these at least compile with minimal configurations. This will affect all archs. Mostly the changes should be relatively easy. Take do_IRQ(), store the regs pointer at the beginning, saving the old one: struct pt_regs *old_regs = set_irq_regs(regs); And put the old one back at the end: set_irq_regs(old_regs); Don't pass regs through to generic_handle_irq() or __do_IRQ(). In timer_interrupt(), this sort of change will be necessary: - update_process_times(user_mode(regs)); - profile_tick(CPU_PROFILING, regs); + update_process_times(user_mode(get_irq_regs())); + profile_tick(CPU_PROFILING); I'd like to move update_process_times()'s use of get_irq_regs() into itself, except that i386, alone of the archs, uses something other than user_mode(). Some notes on the interrupt handling in the drivers: (*) input_dev() is now gone entirely. The regs pointer is no longer stored in the input_dev struct. (*) finish_unlinks() in drivers/usb/host/ohci-q.c needs checking. It does something different depending on whether it's been supplied with a regs pointer or not. (*) Various IRQ handler function pointers have been moved to type irq_handler_t. Signed-Off-By: David Howells <dhowells@redhat.com> (cherry picked from 1b16e7ac850969f38b375e511e3fa2f474a33867 commit)
2006-10-05 06:55:46 -07:00
NCR_700_intr(int irq, void *dev_id)
{
struct Scsi_Host *host = (struct Scsi_Host *)dev_id;
struct NCR_700_Host_Parameters *hostdata =
(struct NCR_700_Host_Parameters *)host->hostdata[0];
__u8 istat;
__u32 resume_offset = 0;
__u8 pun = 0xff, lun = 0xff;
unsigned long flags;
int handled = 0;
/* Use the host lock to serialise access to the 53c700
* hardware. Note: In future, we may need to take the queue
* lock to enter the done routines. When that happens, we
* need to ensure that for this driver, the host lock and the
* queue lock point to the same thing. */
spin_lock_irqsave(host->host_lock, flags);
if((istat = NCR_700_readb(host, ISTAT_REG))
& (SCSI_INT_PENDING | DMA_INT_PENDING)) {
__u32 dsps;
__u8 sstat0 = 0, dstat = 0;
__u32 dsp;
struct scsi_cmnd *SCp = hostdata->cmd;
enum NCR_700_Host_State state;
handled = 1;
state = hostdata->state;
SCp = hostdata->cmd;
if(istat & SCSI_INT_PENDING) {
udelay(10);
sstat0 = NCR_700_readb(host, SSTAT0_REG);
}
if(istat & DMA_INT_PENDING) {
udelay(10);
dstat = NCR_700_readb(host, DSTAT_REG);
}
dsps = NCR_700_readl(host, DSPS_REG);
dsp = NCR_700_readl(host, DSP_REG);
DEBUG(("scsi%d: istat %02x sstat0 %02x dstat %02x dsp %04x[%08x] dsps 0x%x\n",
host->host_no, istat, sstat0, dstat,
(dsp - (__u32)(hostdata->pScript))/4,
dsp, dsps));
if(SCp != NULL) {
pun = SCp->device->id;
lun = SCp->device->lun;
}
if(sstat0 & SCSI_RESET_DETECTED) {
struct scsi_device *SDp;
int i;
hostdata->state = NCR_700_HOST_BUSY;
printk(KERN_ERR "scsi%d: Bus Reset detected, executing command %p, slot %p, dsp %08x[%04x]\n",
host->host_no, SCp, SCp == NULL ? NULL : SCp->host_scribble, dsp, dsp - hostdata->pScript);
scsi_report_bus_reset(host, 0);
/* clear all the negotiated parameters */
__shost_for_each_device(SDp, host)
NCR_700_clear_flag(SDp, ~0);
/* clear all the slots and their pending commands */
for(i = 0; i < NCR_700_COMMAND_SLOTS_PER_HOST; i++) {
struct scsi_cmnd *SCp;
struct NCR_700_command_slot *slot =
&hostdata->slots[i];
if(slot->state == NCR_700_SLOT_FREE)
continue;
SCp = slot->cmnd;
printk(KERN_ERR " failing command because of reset, slot %p, cmnd %p\n",
slot, SCp);
free_slot(slot, hostdata);
SCp->host_scribble = NULL;
NCR_700_set_depth(SCp->device, 0);
/* NOTE: deadlock potential here: we
* rely on mid-layer guarantees that
* scsi_done won't try to issue the
* command again otherwise we'll
* deadlock on the
* hostdata->state_lock */
SCp->result = DID_RESET << 16;
SCp->scsi_done(SCp);
}
mdelay(25);
NCR_700_chip_setup(host);
hostdata->state = NCR_700_HOST_FREE;
hostdata->cmd = NULL;
/* signal back if this was an eh induced reset */
if(hostdata->eh_complete != NULL)
complete(hostdata->eh_complete);
goto out_unlock;
} else if(sstat0 & SELECTION_TIMEOUT) {
DEBUG(("scsi%d: (%d:%d) selection timeout\n",
host->host_no, pun, lun));
NCR_700_scsi_done(hostdata, SCp, DID_NO_CONNECT<<16);
} else if(sstat0 & PHASE_MISMATCH) {
struct NCR_700_command_slot *slot = (SCp == NULL) ? NULL :
(struct NCR_700_command_slot *)SCp->host_scribble;
if(dsp == Ent_SendMessage + 8 + hostdata->pScript) {
/* It wants to reply to some part of
* our message */
#ifdef NCR_700_DEBUG
__u32 temp = NCR_700_readl(host, TEMP_REG);
int count = (hostdata->script[Ent_SendMessage/4] & 0xffffff) - ((NCR_700_readl(host, DBC_REG) & 0xffffff) + NCR_700_data_residual(host));
printk("scsi%d (%d:%d) PHASE MISMATCH IN SEND MESSAGE %d remain, return %p[%04x], phase %s\n", host->host_no, pun, lun, count, (void *)temp, temp - hostdata->pScript, sbcl_to_string(NCR_700_readb(host, SBCL_REG)));
#endif
resume_offset = hostdata->pScript + Ent_SendMessagePhaseMismatch;
} else if(dsp >= to32bit(&slot->pSG[0].ins) &&
dsp <= to32bit(&slot->pSG[NCR_700_SG_SEGMENTS].ins)) {
int data_transfer = NCR_700_readl(host, DBC_REG) & 0xffffff;
int SGcount = (dsp - to32bit(&slot->pSG[0].ins))/sizeof(struct NCR_700_SG_List);
int residual = NCR_700_data_residual(host);
int i;
#ifdef NCR_700_DEBUG
__u32 naddr = NCR_700_readl(host, DNAD_REG);
printk("scsi%d: (%d:%d) Expected phase mismatch in slot->SG[%d], transferred 0x%x\n",
host->host_no, pun, lun,
SGcount, data_transfer);
scsi_print_command(SCp);
if(residual) {
printk("scsi%d: (%d:%d) Expected phase mismatch in slot->SG[%d], transferred 0x%x, residual %d\n",
host->host_no, pun, lun,
SGcount, data_transfer, residual);
}
#endif
data_transfer += residual;
if(data_transfer != 0) {
int count;
__u32 pAddr;
SGcount--;
count = (bS_to_cpu(slot->SG[SGcount].ins) & 0x00ffffff);
DEBUG(("DATA TRANSFER MISMATCH, count = %d, transferred %d\n", count, count-data_transfer));
slot->SG[SGcount].ins &= bS_to_host(0xff000000);
slot->SG[SGcount].ins |= bS_to_host(data_transfer);
pAddr = bS_to_cpu(slot->SG[SGcount].pAddr);
pAddr += (count - data_transfer);
#ifdef NCR_700_DEBUG
if(pAddr != naddr) {
printk("scsi%d (%d:%d) transfer mismatch pAddr=%lx, naddr=%lx, data_transfer=%d, residual=%d\n", host->host_no, pun, lun, (unsigned long)pAddr, (unsigned long)naddr, data_transfer, residual);
}
#endif
slot->SG[SGcount].pAddr = bS_to_host(pAddr);
}
/* set the executed moves to nops */
for(i=0; i<SGcount; i++) {
slot->SG[i].ins = bS_to_host(SCRIPT_NOP);
slot->SG[i].pAddr = 0;
}
dma_cache_sync(hostdata->dev, slot->SG, sizeof(slot->SG), DMA_TO_DEVICE);
/* and pretend we disconnected after
* the command phase */
resume_offset = hostdata->pScript + Ent_MsgInDuringData;
/* make sure all the data is flushed */
NCR_700_flush_fifo(host);
} else {
__u8 sbcl = NCR_700_readb(host, SBCL_REG);
printk(KERN_ERR "scsi%d: (%d:%d) phase mismatch at %04x, phase %s\n",
host->host_no, pun, lun, dsp - hostdata->pScript, sbcl_to_string(sbcl));
NCR_700_internal_bus_reset(host);
}
} else if(sstat0 & SCSI_GROSS_ERROR) {
printk(KERN_ERR "scsi%d: (%d:%d) GROSS ERROR\n",
host->host_no, pun, lun);
NCR_700_scsi_done(hostdata, SCp, DID_ERROR<<16);
} else if(sstat0 & PARITY_ERROR) {
printk(KERN_ERR "scsi%d: (%d:%d) PARITY ERROR\n",
host->host_no, pun, lun);
NCR_700_scsi_done(hostdata, SCp, DID_ERROR<<16);
} else if(dstat & SCRIPT_INT_RECEIVED) {
DEBUG(("scsi%d: (%d:%d) ====>SCRIPT INTERRUPT<====\n",
host->host_no, pun, lun));
resume_offset = process_script_interrupt(dsps, dsp, SCp, host, hostdata);
} else if(dstat & (ILGL_INST_DETECTED)) {
printk(KERN_ERR "scsi%d: (%d:%d) Illegal Instruction detected at 0x%08x[0x%x]!!!\n"
" Please email James.Bottomley@HansenPartnership.com with the details\n",
host->host_no, pun, lun,
dsp, dsp - hostdata->pScript);
NCR_700_scsi_done(hostdata, SCp, DID_ERROR<<16);
} else if(dstat & (WATCH_DOG_INTERRUPT|ABORTED)) {
printk(KERN_ERR "scsi%d: (%d:%d) serious DMA problem, dstat=%02x\n",
host->host_no, pun, lun, dstat);
NCR_700_scsi_done(hostdata, SCp, DID_ERROR<<16);
}
/* NOTE: selection interrupt processing MUST occur
* after script interrupt processing to correctly cope
* with the case where we process a disconnect and
* then get reselected before we process the
* disconnection */
if(sstat0 & SELECTED) {
/* FIXME: It currently takes at least FOUR
* interrupts to complete a command that
* disconnects: one for the disconnect, one
* for the reselection, one to get the
* reselection data and one to complete the
* command. If we guess the reselected
* command here and prepare it, we only need
* to get a reselection data interrupt if we
* guessed wrongly. Since the interrupt
* overhead is much greater than the command
* setup, this would be an efficient
* optimisation particularly as we probably
* only have one outstanding command on a
* target most of the time */
resume_offset = process_selection(host, dsp);
}
}
if(resume_offset) {
if(hostdata->state != NCR_700_HOST_BUSY) {
printk(KERN_ERR "scsi%d: Driver error: resume at 0x%08x [0x%04x] with non busy host!\n",
host->host_no, resume_offset, resume_offset - hostdata->pScript);
hostdata->state = NCR_700_HOST_BUSY;
}
DEBUG(("Attempting to resume at %x\n", resume_offset));
NCR_700_clear_fifo(host);
NCR_700_writel(resume_offset, host, DSP_REG);
}
/* There is probably a technical no-no about this: If we're a
* shared interrupt and we got this interrupt because the
* other device needs servicing not us, we're still going to
* check our queued commands here---of course, there shouldn't
* be any outstanding.... */
if(hostdata->state == NCR_700_HOST_FREE) {
int i;
for(i = 0; i < NCR_700_COMMAND_SLOTS_PER_HOST; i++) {
/* fairness: always run the queue from the last
* position we left off */
int j = (i + hostdata->saved_slot_position)
% NCR_700_COMMAND_SLOTS_PER_HOST;
if(hostdata->slots[j].state != NCR_700_SLOT_QUEUED)
continue;
if(NCR_700_start_command(hostdata->slots[j].cmnd)) {
DEBUG(("scsi%d: Issuing saved command slot %p, cmd %p\t\n",
host->host_no, &hostdata->slots[j],
hostdata->slots[j].cmnd));
hostdata->saved_slot_position = j + 1;
}
break;
}
}
out_unlock:
spin_unlock_irqrestore(host->host_lock, flags);
return IRQ_RETVAL(handled);
}
STATIC int
NCR_700_queuecommand(struct scsi_cmnd *SCp, void (*done)(struct scsi_cmnd *))
{
struct NCR_700_Host_Parameters *hostdata =
(struct NCR_700_Host_Parameters *)SCp->device->host->hostdata[0];
__u32 move_ins;
enum dma_data_direction direction;
struct NCR_700_command_slot *slot;
if(hostdata->command_slot_count >= NCR_700_COMMAND_SLOTS_PER_HOST) {
/* We're over our allocation, this should never happen
* since we report the max allocation to the mid layer */
printk(KERN_WARNING "scsi%d: Command depth has gone over queue depth\n", SCp->device->host->host_no);
return 1;
}
/* check for untagged commands. We cannot have any outstanding
* commands if we accept them. Commands could be untagged because:
*
* - The tag negotiated bitmap is clear
* - The blk layer sent and untagged command
*/
if(NCR_700_get_depth(SCp->device) != 0
&& (!(hostdata->tag_negotiated & (1<<scmd_id(SCp)))
|| !blk_rq_tagged(SCp->request))) {
CDEBUG(KERN_ERR, SCp, "has non zero depth %d\n",
NCR_700_get_depth(SCp->device));
return SCSI_MLQUEUE_DEVICE_BUSY;
}
if(NCR_700_get_depth(SCp->device) >= SCp->device->queue_depth) {
CDEBUG(KERN_ERR, SCp, "has max tag depth %d\n",
NCR_700_get_depth(SCp->device));
return SCSI_MLQUEUE_DEVICE_BUSY;
}
NCR_700_set_depth(SCp->device, NCR_700_get_depth(SCp->device) + 1);
/* begin the command here */
/* no need to check for NULL, test for command_slot_count above
* ensures a slot is free */
slot = find_empty_slot(hostdata);
slot->cmnd = SCp;
SCp->scsi_done = done;
SCp->host_scribble = (unsigned char *)slot;
SCp->SCp.ptr = NULL;
SCp->SCp.buffer = NULL;
#ifdef NCR_700_DEBUG
printk("53c700: scsi%d, command ", SCp->device->host->host_no);
scsi_print_command(SCp);
#endif
if(blk_rq_tagged(SCp->request)
&& (hostdata->tag_negotiated &(1<<scmd_id(SCp))) == 0
&& NCR_700_get_tag_neg_state(SCp->device) == NCR_700_START_TAG_NEGOTIATION) {
scmd_printk(KERN_ERR, SCp, "Enabling Tag Command Queuing\n");
hostdata->tag_negotiated |= (1<<scmd_id(SCp));
NCR_700_set_tag_neg_state(SCp->device, NCR_700_DURING_TAG_NEGOTIATION);
}
/* here we may have to process an untagged command. The gate
* above ensures that this will be the only one outstanding,
* so clear the tag negotiated bit.
*
* FIXME: This will royally screw up on multiple LUN devices
* */
if(!blk_rq_tagged(SCp->request)
&& (hostdata->tag_negotiated &(1<<scmd_id(SCp)))) {
scmd_printk(KERN_INFO, SCp, "Disabling Tag Command Queuing\n");
hostdata->tag_negotiated &= ~(1<<scmd_id(SCp));
}
if((hostdata->tag_negotiated &(1<<scmd_id(SCp)))
&& scsi_get_tag_type(SCp->device)) {
slot->tag = SCp->request->tag;
CDEBUG(KERN_DEBUG, SCp, "sending out tag %d, slot %p\n",
slot->tag, slot);
} else {
slot->tag = SCSI_NO_TAG;
/* must populate current_cmnd for scsi_find_tag to work */
SCp->device->current_cmnd = SCp;
}
/* sanity check: some of the commands generated by the mid-layer
* have an eccentric idea of their sc_data_direction */
if(!scsi_sg_count(SCp) && !scsi_bufflen(SCp) &&
SCp->sc_data_direction != DMA_NONE) {
#ifdef NCR_700_DEBUG
printk("53c700: Command");
scsi_print_command(SCp);
printk("Has wrong data direction %d\n", SCp->sc_data_direction);
#endif
SCp->sc_data_direction = DMA_NONE;
}
switch (SCp->cmnd[0]) {
case REQUEST_SENSE:
/* clear the internal sense magic */
SCp->cmnd[6] = 0;
/* fall through */
default:
/* OK, get it from the command */
switch(SCp->sc_data_direction) {
case DMA_BIDIRECTIONAL:
default:
printk(KERN_ERR "53c700: Unknown command for data direction ");
scsi_print_command(SCp);
move_ins = 0;
break;
case DMA_NONE:
move_ins = 0;
break;
case DMA_FROM_DEVICE:
move_ins = SCRIPT_MOVE_DATA_IN;
break;
case DMA_TO_DEVICE:
move_ins = SCRIPT_MOVE_DATA_OUT;
break;
}
}
/* now build the scatter gather list */
direction = SCp->sc_data_direction;
if(move_ins != 0) {
int i;
int sg_count;
dma_addr_t vPtr = 0;
struct scatterlist *sg;
__u32 count = 0;
sg_count = scsi_dma_map(SCp);
BUG_ON(sg_count < 0);
scsi_for_each_sg(SCp, sg, sg_count, i) {
vPtr = sg_dma_address(sg);
count = sg_dma_len(sg);
slot->SG[i].ins = bS_to_host(move_ins | count);
DEBUG((" scatter block %d: move %d[%08x] from 0x%lx\n",
i, count, slot->SG[i].ins, (unsigned long)vPtr));
slot->SG[i].pAddr = bS_to_host(vPtr);
}
slot->SG[i].ins = bS_to_host(SCRIPT_RETURN);
slot->SG[i].pAddr = 0;
dma_cache_sync(hostdata->dev, slot->SG, sizeof(slot->SG), DMA_TO_DEVICE);
DEBUG((" SETTING %08lx to %x\n",
(&slot->pSG[i].ins),
slot->SG[i].ins));
}
slot->resume_offset = 0;
slot->pCmd = dma_map_single(hostdata->dev, SCp->cmnd,
[SCSI] Let scsi_cmnd->cmnd use request->cmd buffer - struct scsi_cmnd had a 16 bytes command buffer of its own. This is an unnecessary duplication and copy of request's cmd. It is probably left overs from the time that scsi_cmnd could function without a request attached. So clean that up. - Once above is done, few places, apart from scsi-ml, needed adjustments due to changing the data type of scsi_cmnd->cmnd. - Lots of drivers still use MAX_COMMAND_SIZE. So I have left that #define but equate it to BLK_MAX_CDB. The way I see it and is reflected in the patch below is. MAX_COMMAND_SIZE - means: The longest fixed-length (*) SCSI CDB as per the SCSI standard and is not related to the implementation. BLK_MAX_CDB. - The allocated space at the request level - I have audit all ISA drivers and made sure none use ->cmnd in a DMA Operation. Same audit was done by Andi Kleen. (*)fixed-length here means commands that their size can be determined by their opcode and the CDB does not carry a length specifier, (unlike the VARIABLE_LENGTH_CMD(0x7f) command). This is actually not exactly true and the SCSI standard also defines extended commands and vendor specific commands that can be bigger than 16 bytes. The kernel will support these using the same infrastructure used for VARLEN CDB's. So in effect MAX_COMMAND_SIZE means the maximum size command scsi-ml supports without specifying a cmd_len by ULD's Signed-off-by: Boaz Harrosh <bharrosh@panasas.com> Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
2008-04-30 01:19:47 -07:00
MAX_COMMAND_SIZE, DMA_TO_DEVICE);
NCR_700_start_command(SCp);
return 0;
}
STATIC int
NCR_700_abort(struct scsi_cmnd * SCp)
{
struct NCR_700_command_slot *slot;
scmd_printk(KERN_INFO, SCp,
"New error handler wants to abort command\n\t");
scsi_print_command(SCp);
slot = (struct NCR_700_command_slot *)SCp->host_scribble;
if(slot == NULL)
/* no outstanding command to abort */
return SUCCESS;
if(SCp->cmnd[0] == TEST_UNIT_READY) {
/* FIXME: This is because of a problem in the new
* error handler. When it is in error recovery, it
* will send a TUR to a device it thinks may still be
* showing a problem. If the TUR isn't responded to,
* it will abort it and mark the device off line.
* Unfortunately, it does no other error recovery, so
* this would leave us with an outstanding command
* occupying a slot. Rather than allow this to
* happen, we issue a bus reset to force all
* outstanding commands to terminate here. */
NCR_700_internal_bus_reset(SCp->device->host);
/* still drop through and return failed */
}
return FAILED;
}
STATIC int
NCR_700_bus_reset(struct scsi_cmnd * SCp)
{
DECLARE_COMPLETION_ONSTACK(complete);
struct NCR_700_Host_Parameters *hostdata =
(struct NCR_700_Host_Parameters *)SCp->device->host->hostdata[0];
scmd_printk(KERN_INFO, SCp,
"New error handler wants BUS reset, cmd %p\n\t", SCp);
scsi_print_command(SCp);
/* In theory, eh_complete should always be null because the
* eh is single threaded, but just in case we're handling a
* reset via sg or something */
spin_lock_irq(SCp->device->host->host_lock);
while (hostdata->eh_complete != NULL) {
spin_unlock_irq(SCp->device->host->host_lock);
msleep_interruptible(100);
spin_lock_irq(SCp->device->host->host_lock);
}
hostdata->eh_complete = &complete;
NCR_700_internal_bus_reset(SCp->device->host);
spin_unlock_irq(SCp->device->host->host_lock);
wait_for_completion(&complete);
spin_lock_irq(SCp->device->host->host_lock);
hostdata->eh_complete = NULL;
/* Revalidate the transport parameters of the failing device */
if(hostdata->fast)
spi_schedule_dv_device(SCp->device);
spin_unlock_irq(SCp->device->host->host_lock);
return SUCCESS;
}
STATIC int
NCR_700_host_reset(struct scsi_cmnd * SCp)
{
scmd_printk(KERN_INFO, SCp, "New error handler wants HOST reset\n\t");
scsi_print_command(SCp);
spin_lock_irq(SCp->device->host->host_lock);
NCR_700_internal_bus_reset(SCp->device->host);
NCR_700_chip_reset(SCp->device->host);
spin_unlock_irq(SCp->device->host->host_lock);
return SUCCESS;
}
STATIC void
NCR_700_set_period(struct scsi_target *STp, int period)
{
struct Scsi_Host *SHp = dev_to_shost(STp->dev.parent);
struct NCR_700_Host_Parameters *hostdata =
(struct NCR_700_Host_Parameters *)SHp->hostdata[0];
if(!hostdata->fast)
return;
if(period < hostdata->min_period)
period = hostdata->min_period;
spi_period(STp) = period;
spi_flags(STp) &= ~(NCR_700_DEV_NEGOTIATED_SYNC |
NCR_700_DEV_BEGIN_SYNC_NEGOTIATION);
spi_flags(STp) |= NCR_700_DEV_PRINT_SYNC_NEGOTIATION;
}
STATIC void
NCR_700_set_offset(struct scsi_target *STp, int offset)
{
struct Scsi_Host *SHp = dev_to_shost(STp->dev.parent);
struct NCR_700_Host_Parameters *hostdata =
(struct NCR_700_Host_Parameters *)SHp->hostdata[0];
int max_offset = hostdata->chip710
? NCR_710_MAX_OFFSET : NCR_700_MAX_OFFSET;
if(!hostdata->fast)
return;
if(offset > max_offset)
offset = max_offset;
/* if we're currently async, make sure the period is reasonable */
if(spi_offset(STp) == 0 && (spi_period(STp) < hostdata->min_period ||
spi_period(STp) > 0xff))
spi_period(STp) = hostdata->min_period;
spi_offset(STp) = offset;
spi_flags(STp) &= ~(NCR_700_DEV_NEGOTIATED_SYNC |
NCR_700_DEV_BEGIN_SYNC_NEGOTIATION);
spi_flags(STp) |= NCR_700_DEV_PRINT_SYNC_NEGOTIATION;
}
STATIC int
NCR_700_slave_alloc(struct scsi_device *SDp)
{
SDp->hostdata = kzalloc(sizeof(struct NCR_700_Device_Parameters),
GFP_KERNEL);
if (!SDp->hostdata)
return -ENOMEM;
return 0;
}
STATIC int
NCR_700_slave_configure(struct scsi_device *SDp)
{
struct NCR_700_Host_Parameters *hostdata =
(struct NCR_700_Host_Parameters *)SDp->host->hostdata[0];
/* to do here: allocate memory; build a queue_full list */
if(SDp->tagged_supported) {
scsi_set_tag_type(SDp, MSG_ORDERED_TAG);
scsi_activate_tcq(SDp, NCR_700_DEFAULT_TAGS);
NCR_700_set_tag_neg_state(SDp, NCR_700_START_TAG_NEGOTIATION);
} else {
/* initialise to default depth */
scsi_adjust_queue_depth(SDp, 0, SDp->host->cmd_per_lun);
}
if(hostdata->fast) {
/* Find the correct offset and period via domain validation */
if (!spi_initial_dv(SDp->sdev_target))
spi_dv_device(SDp);
} else {
spi_offset(SDp->sdev_target) = 0;
spi_period(SDp->sdev_target) = 0;
}
return 0;
}
STATIC void
NCR_700_slave_destroy(struct scsi_device *SDp)
{
kfree(SDp->hostdata);
SDp->hostdata = NULL;
}
static int
NCR_700_change_queue_depth(struct scsi_device *SDp, int depth, int reason)
{
if (reason != SCSI_QDEPTH_DEFAULT)
return -EOPNOTSUPP;
if (depth > NCR_700_MAX_TAGS)
depth = NCR_700_MAX_TAGS;
scsi_adjust_queue_depth(SDp, scsi_get_tag_type(SDp), depth);
return depth;
}
static int NCR_700_change_queue_type(struct scsi_device *SDp, int tag_type)
{
int change_tag = ((tag_type ==0 && scsi_get_tag_type(SDp) != 0)
|| (tag_type != 0 && scsi_get_tag_type(SDp) == 0));
struct NCR_700_Host_Parameters *hostdata =
(struct NCR_700_Host_Parameters *)SDp->host->hostdata[0];
scsi_set_tag_type(SDp, tag_type);
/* We have a global (per target) flag to track whether TCQ is
* enabled, so we'll be turning it off for the entire target here.
* our tag algorithm will fail if we mix tagged and untagged commands,
* so quiesce the device before doing this */
if (change_tag)
scsi_target_quiesce(SDp->sdev_target);
if (!tag_type) {
/* shift back to the default unqueued number of commands
* (the user can still raise this) */
scsi_deactivate_tcq(SDp, SDp->host->cmd_per_lun);
hostdata->tag_negotiated &= ~(1 << sdev_id(SDp));
} else {
/* Here, we cleared the negotiation flag above, so this
* will force the driver to renegotiate */
scsi_activate_tcq(SDp, SDp->queue_depth);
if (change_tag)
NCR_700_set_tag_neg_state(SDp, NCR_700_START_TAG_NEGOTIATION);
}
if (change_tag)
scsi_target_resume(SDp->sdev_target);
return tag_type;
}
static ssize_t
NCR_700_show_active_tags(struct device *dev, struct device_attribute *attr, char *buf)
{
struct scsi_device *SDp = to_scsi_device(dev);
return snprintf(buf, 20, "%d\n", NCR_700_get_depth(SDp));
}
static struct device_attribute NCR_700_active_tags_attr = {
.attr = {
.name = "active_tags",
.mode = S_IRUGO,
},
.show = NCR_700_show_active_tags,
};
STATIC struct device_attribute *NCR_700_dev_attrs[] = {
&NCR_700_active_tags_attr,
NULL,
};
EXPORT_SYMBOL(NCR_700_detect);
EXPORT_SYMBOL(NCR_700_release);
EXPORT_SYMBOL(NCR_700_intr);
static struct spi_function_template NCR_700_transport_functions = {
.set_period = NCR_700_set_period,
.show_period = 1,
.set_offset = NCR_700_set_offset,
.show_offset = 1,
};
static int __init NCR_700_init(void)
{
NCR_700_transport_template = spi_attach_transport(&NCR_700_transport_functions);
if(!NCR_700_transport_template)
return -ENODEV;
return 0;
}
static void __exit NCR_700_exit(void)
{
spi_release_transport(NCR_700_transport_template);
}
module_init(NCR_700_init);
module_exit(NCR_700_exit);