* [PATCH] 2.5.67-ac2 fix new PIO handlers
@ 2003-04-27 18:38 Bartlomiej Zolnierkiewicz
2003-04-28 1:45 ` Osamu Tomita
0 siblings, 1 reply; 3+ messages in thread
From: Bartlomiej Zolnierkiewicz @ 2003-04-27 18:38 UTC (permalink / raw)
To: Alan Cox; +Cc: linux-kernel
[-- Attachment #1: Type: TEXT/PLAIN, Size: 509 bytes --]
Hi,
During rewrite of bio walking patch to use rq->cbio instead
of rq->hard_bio I've realized I had screwed orig 2.5.66 patches :\
I somehow forgot to update task_map_rq(), it should use
__bio_kmap_irq(rq->bio, current_idx, flags) instead of
bio_kmap_irq(rq->bio, flags). The result is that you get
bio corruption (-> data corruption) when using PIO multiple
with sectors multiply > 8 (PAGE_SIZE == bio_vec size).
Attached patch fixes it and fixes TASKFILE ioctl.
Please apply to next -ac.
--
Bartlomiej
[-- Attachment #2: Type: TEXT/PLAIN, Size: 5226 bytes --]
# Fix PIO handlers (!).
#
# Detailed changelog:
# - introduce proper task_(un)map_rq() - critical fix
# - fix process_that_request_first() for rq->buffer (non-bio) based requests
# - set rq->hard_nr_sectors and rq->hard_cur_sectors in ide_diag_taskfile()
# - fix PIO handlers for rq->buffer based requests (TASKFILE ioctl)
# - revert ->prehandler() handling for flagged taskfiles as they have their
# own PIO handlers
#
# Bartlomiej Zolnierkiewicz <bzolnier@elka.pw.edu.pl>
diff -uNr linux-2.5.67-ac2/drivers/block/ll_rw_blk.c linux/drivers/block/ll_rw_blk.c
--- linux-2.5.67-ac2/drivers/block/ll_rw_blk.c Sun Apr 27 19:54:40 2003
+++ linux/drivers/block/ll_rw_blk.c Sun Apr 27 19:54:58 2003
@@ -2265,9 +2265,11 @@
nsect = min_t(unsigned int, req->current_nr_sectors,
nr_sectors);
req->current_nr_sectors -= nsect;
- req->nr_bio_sectors -= nsect;
nr_sectors -= nsect;
- blk_rq_next_segment(req);
+ if (req->bio) {
+ req->nr_bio_sectors -= nsect;
+ blk_rq_next_segment(req);
+ }
}
return 1;
}
diff -uNr linux-2.5.67-ac2/drivers/ide/ide-taskfile.c linux/drivers/ide/ide-taskfile.c
--- linux-2.5.67-ac2/drivers/ide/ide-taskfile.c Sun Apr 27 19:54:40 2003
+++ linux/drivers/ide/ide-taskfile.c Sun Apr 27 19:58:41 2003
@@ -59,9 +59,6 @@
#define DTF(x...)
#endif
-#define task_map_rq(rq, flags) ide_map_buffer((rq), (flags))
-#define task_unmap_rq(rq, buf, flags) ide_unmap_buffer((rq), (buf), (flags))
-
static void ata_bswap_data (void *buffer, int wcount)
{
u16 *p = buffer;
@@ -406,6 +403,11 @@
while (rq->hard_bio != rq->bio)
if (!DRIVER(drive)->end_request(drive, 1, bio_sectors(rq->hard_bio)))
return ide_stopped;
+ /* Complete rq->buffer based request (ioctls). */
+ if (!rq->bio && !rq->nr_sectors) {
+ ide_end_drive_cmd(drive, stat, HWIF(drive)->INB(IDE_ERROR_REG));
+ return ide_stopped;
+ }
rq->errors = 0;
task_sectors(drive, rq, 1, PIO_IN);
@@ -452,6 +454,11 @@
while (rq->hard_bio != rq->bio)
if (!DRIVER(drive)->end_request(drive, 1, bio_sectors(rq->hard_bio)))
return ide_stopped;
+ /* Complete rq->buffer based request (ioctls). */
+ if (!rq->bio && !rq->nr_sectors) {
+ ide_end_drive_cmd(drive, stat, HWIF(drive)->INB(IDE_ERROR_REG));
+ return ide_stopped;
+ }
rq->errors = 0;
do {
@@ -510,6 +517,11 @@
while (rq->hard_bio != rq->bio)
if (!DRIVER(drive)->end_request(drive, 1, bio_sectors(rq->hard_bio)))
return ide_stopped;
+ /* Complete rq->buffer based request (ioctls). */
+ if (!rq->bio && !rq->nr_sectors) {
+ ide_end_drive_cmd(drive, stat, HWIF(drive)->INB(IDE_ERROR_REG));
+ return ide_stopped;
+ }
/* Still data left to transfer. */
ide_set_handler(drive, &task_out_intr, WAIT_WORSTCASE, NULL);
@@ -569,6 +581,11 @@
while (rq->hard_bio != rq->bio)
if (!DRIVER(drive)->end_request(drive, 1, bio_sectors(rq->hard_bio)))
return ide_stopped;
+ /* Complete rq->buffer based request (ioctls). */
+ if (!rq->bio && !rq->nr_sectors) {
+ ide_end_drive_cmd(drive, stat, HWIF(drive)->INB(IDE_ERROR_REG));
+ return ide_stopped;
+ }
/* Still data left to transfer. */
ide_set_handler(drive, &task_mulout_intr, WAIT_WORSTCASE, NULL);
@@ -952,11 +969,12 @@
*/
if (args->command_type != IDE_DRIVE_TASK_NO_DATA) {
if (data_size == 0)
- rq.current_nr_sectors = rq.nr_sectors = (args->hobRegister[IDE_NSECTOR_OFFSET_HOB] << 8) | args->tfRegister[IDE_NSECTOR_OFFSET];
- /* rq.hard_cur_sectors */
+ rq.nr_sectors = (args->hobRegister[IDE_NSECTOR_OFFSET_HOB] << 8) | args->tfRegister[IDE_NSECTOR_OFFSET];
else
- rq.current_nr_sectors = rq.nr_sectors = data_size / SECTOR_SIZE;
- /* rq.hard_cur_sectors */
+ rq.nr_sectors = data_size / SECTOR_SIZE;
+
+ rq.hard_nr_sectors = rq.nr_sectors;
+ rq.hard_cur_sectors = rq.current_nr_sectors = rq.nr_sectors;
}
if (args->tf_out_flags.all == 0) {
@@ -1549,14 +1567,10 @@
if (task->handler == NULL)
return ide_stopped;
- if (task->prehandler != NULL) {
- hwif->OUTBSYNC(drive, taskfile->command, IDE_COMMAND_REG);
- ndelay(400); /* FIXME */
- return task->prehandler(drive, HWGROUP(drive)->rq);
- }
-
/* Issue the command */
ide_execute_command(drive, taskfile->command, task->handler, WAIT_WORSTCASE, NULL);
+ if (task->prehandler != NULL)
+ return task->prehandler(drive, HWGROUP(drive)->rq);
}
return ide_started;
diff -uNr linux-2.5.67-ac2/include/linux/ide.h linux/include/linux/ide.h
--- linux-2.5.67-ac2/include/linux/ide.h Sun Apr 27 19:54:40 2003
+++ linux/include/linux/ide.h Sun Apr 27 20:00:28 2003
@@ -832,6 +832,26 @@
ide_startstop_t (*ide_dma_queued_start)(ide_drive_t *drive);
} ide_dma_ops_t;
+static inline void *task_map_rq(struct request *rq, unsigned long *flags)
+{
+ /*
+ * fs request
+ */
+ if (rq->bio)
+ return rq_map_buffer(rq, flags);
+
+ /*
+ * task request
+ */
+ return rq->buffer + blk_rq_offset(rq);
+}
+
+static inline void task_unmap_rq(struct request *rq, char *buffer, unsigned long *flags)
+{
+ if (rq->bio)
+ rq_unmap_buffer(buffer, flags);
+}
+
/*
* mapping stuff, prepare for highmem...
*
^ permalink raw reply [flat|nested] 3+ messages in thread* Re: [PATCH] 2.5.67-ac2 fix new PIO handlers
2003-04-27 18:38 [PATCH] 2.5.67-ac2 fix new PIO handlers Bartlomiej Zolnierkiewicz
@ 2003-04-28 1:45 ` Osamu Tomita
2003-04-28 13:10 ` Bartlomiej Zolnierkiewicz
0 siblings, 1 reply; 3+ messages in thread
From: Osamu Tomita @ 2003-04-28 1:45 UTC (permalink / raw)
To: Bartlomiej Zolnierkiewicz; +Cc: Alan Cox, linux-kernel
Hi,
Bartlomiej Zolnierkiewicz wrote:
>
> Hi,
>
> During rewrite of bio walking patch to use rq->cbio instead
> of rq->hard_bio I've realized I had screwed orig 2.5.66 patches :\
>
> I somehow forgot to update task_map_rq(), it should use
> __bio_kmap_irq(rq->bio, current_idx, flags) instead of
> bio_kmap_irq(rq->bio, flags). The result is that you get
> bio corruption (-> data corruption) when using PIO multiple
> with sectors multiply > 8 (PAGE_SIZE == bio_vec size).
>
> Attached patch fixes it and fixes TASKFILE ioctl.
>
> Please apply to next -ac.
Could you please tell me how to apply your patch. Aginst 2.5.62-ac2?
I have a ext2 fs corruption problem at PIO mode on 2.5.67-ac2.
There is no problem on vanilla 2.5.67.
I tryed the all patches below.
tf-ioctls-1.diff
tf-ioctls-2a.diff
tf-ioctls-2b.diff
tf-ioctls-3.diff
tf-ioctls-4.diff
tf-dio-1.diff
tf-dio-2.diff
tf-dio-3.diff
tf-dio-4.diff
masked-irq.diff
They didn't solve problem.
This pio-fixes.diff can't be applied after them?
Regards,
Osamu Tomita
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH] 2.5.67-ac2 fix new PIO handlers
2003-04-28 1:45 ` Osamu Tomita
@ 2003-04-28 13:10 ` Bartlomiej Zolnierkiewicz
0 siblings, 0 replies; 3+ messages in thread
From: Bartlomiej Zolnierkiewicz @ 2003-04-28 13:10 UTC (permalink / raw)
To: Osamu Tomita; +Cc: Alan Cox, linux-kernel
On Mon, 28 Apr 2003, Osamu Tomita wrote:
> Hi,
>
> Bartlomiej Zolnierkiewicz wrote:
> >
> > Hi,
> >
> > During rewrite of bio walking patch to use rq->cbio instead
> > of rq->hard_bio I've realized I had screwed orig 2.5.66 patches :\
> >
> > I somehow forgot to update task_map_rq(), it should use
> > __bio_kmap_irq(rq->bio, current_idx, flags) instead of
> > bio_kmap_irq(rq->bio, flags). The result is that you get
> > bio corruption (-> data corruption) when using PIO multiple
> > with sectors multiply > 8 (PAGE_SIZE == bio_vec size).
Hmmm, not only > 8, also 3, 5.
> > Attached patch fixes it and fixes TASKFILE ioctl.
> >
> > Please apply to next -ac.
> Could you please tell me how to apply your patch. Aginst 2.5.62-ac2?
> I have a ext2 fs corruption problem at PIO mode on 2.5.67-ac2.
Did you use multi PIO? You can check with 'hdparm -m /dev/hdx'.
If yes this patch should fix the corruption.
> There is no problem on vanilla 2.5.67.
> I tryed the all patches below.
> tf-ioctls-1.diff
> tf-ioctls-2a.diff
> tf-ioctls-2b.diff
> tf-ioctls-3.diff
> tf-ioctls-4.diff
> tf-dio-1.diff
> tf-dio-2.diff
> tf-dio-3.diff
> tf-dio-4.diff
> masked-irq.diff
> They didn't solve problem.
No, they don't.
> This pio-fixes.diff can't be applied after them?
>
> Regards,
> Osamu Tomita
Could you please apply only pio_fixes.diff on top of the vanilla
2.5.67-ac2 and tell me if it solves the problem? It should.
Thanks,
--
Bartlomiej
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2003-04-28 12:59 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2003-04-27 18:38 [PATCH] 2.5.67-ac2 fix new PIO handlers Bartlomiej Zolnierkiewicz
2003-04-28 1:45 ` Osamu Tomita
2003-04-28 13:10 ` Bartlomiej Zolnierkiewicz
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).