2002-11-01 23:33:34

by Khalid Aziz

Subject: [PATCH] ide-scsi driver starts DMA too soon

ide-scsi driver starts DMA as soon as it writes the ATAPI PACKET command
in command register and before sending the ATAPI command. This will
cause problems on many drives. Right way to do it is to start DMA after
sending the ATAPI command. I am attaching a patch that fixes this. This
patch will allow many more CD-RW drives to work reliably in DMA mode
than do today.

Marcelo, please apply.

--
Khalid

====================================================================
Khalid Aziz Linux and Open Source Lab
(970)898-9214 Hewlett-Packard
[email protected] Fort Collins, CO


-----------CUT HERE-------------CUT HERE-------------CUT HERE-----------
--- linux-2.4.19/drivers/scsi/ide-scsi.c Fri Aug 2 18:39:44 2002
+++ linux-2.4.19.cd_dma/drivers/scsi/ide-scsi.c Fri Nov 1 16:31:30 2002
@@ -397,6 +397,8 @@
idescsi_pc_t *pc = scsi->pc;
byte ireason;
ide_startstop_t startstop;
+ struct request *rq = pc->rq;
+ int dma_ok = 0;

if (ide_wait_stat (&startstop,drive,DRQ_STAT,BUSY_STAT,WAIT_READY)) {
printk (KERN_ERR "ide-scsi: Strange, packet command initiated yet DRQ isn't asserted\n");
@@ -407,8 +409,14 @@
printk (KERN_ERR "ide-scsi: (IO,CoD) != (0,1) while issuing a packet command\n");
return ide_do_reset (drive);
}
+ if (drive->using_dma && rq->bh)
+ dma_ok=!HWIF(drive)->dmaproc(test_bit (PC_WRITING, &pc->flags) ? ide_dma_write : ide_dma_read, drive);
ide_set_handler(drive, &idescsi_pc_intr, get_timeout(pc), NULL); /* Set the interrupt routine */
atapi_output_bytes (drive, scsi->pc->c, 12); /* Send the actual packet */
+ if (dma_ok) {
+ set_bit (PC_DMA_IN_PROGRESS, &pc->flags);
+ (void) (HWIF(drive)->dmaproc(ide_dma_begin, drive));
+ }
return ide_started;
}

@@ -437,10 +445,6 @@
OUT_BYTE (bcount >> 8,IDE_BCOUNTH_REG);
OUT_BYTE (bcount & 0xff,IDE_BCOUNTL_REG);

- if (dma_ok) {
- set_bit (PC_DMA_IN_PROGRESS, &pc->flags);
- (void) (HWIF(drive)->dmaproc(ide_dma_begin, drive));
- }
if (test_bit (IDESCSI_DRQ_INTERRUPT, &scsi->flags)) {
ide_set_handler (drive, &idescsi_transfer_pc, get_timeout(pc), NULL);
OUT_BYTE (WIN_PACKETCMD, IDE_COMMAND_REG); /* Issue the packet command */

2002-11-02 00:49:04

by Alan

Subject: Re: [PATCH] ide-scsi driver starts DMA too soon

On Fri, 2002-11-01 at 23:39, Khalid Aziz wrote:
> ide-scsi driver starts DMA as soon as it writes the ATAPI PACKET command
> in command register and before sending the ATAPI command. This will
> cause problems on many drives. Right way to do it is to start DMA after
> sending the ATAPI command. I am attaching a patch that fixes this. This
> patch will allow many more CD-RW drives to work reliably in DMA mode
> than do today.
>
> Marcelo, please apply.

Marcelo this is in 2.5, and 2.4-ac. Khalid is certainly correct although
making such a change in -rc rather than a pre does mean it wants extra
thought

2002-11-06 15:20:17

by Marcelo Tosatti

Subject: Re: [PATCH] ide-scsi driver starts DMA too soon

On 2 Nov 2002, Alan Cox wrote:

> On Fri, 2002-11-01 at 23:39, Khalid Aziz wrote:
> > ide-scsi driver starts DMA as soon as it writes the ATAPI PACKET command
> > in command register and before sending the ATAPI command. This will
> > cause problems on many drives. Right way to do it is to start DMA after
> > sending the ATAPI command. I am attaching a patch that fixes this. This
> > patch will allow many more CD-RW drives to work reliably in DMA mode
> > than do today.
> >
> > Marcelo, please apply.
>
> Marcelo this is in 2.5, and 2.4-ac. Khalid is certainly correct although
> making such a change in -rc rather than a pre does mean it wants extra
> thought

I prefer being safe and saving this one for 2.4.21-pre.

Thanks Khalid!