On Sat, 18 Oct 2008 20:07:47 +0200
Borislav Petkov [off-list ref] wrote:

quoted

would you please take a look at this bug caused by

http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff_plain;h=e5318b531b008c79d2a0c0df06a7b8628da38e2f

Did he confirm that this patch causes this bug?

But, yes, this patch might be the cause.

Using the dma safe check for all the non fs commands is the right
thing. But blk_queue_update_dma_pad() has side effects. It might be
the cause (but I'm not sure about it since I don't know how the ide
code completes requests).

IDE simply disables DMA for unaligned requests (unlike libata) so
there is no point to tell the block layer about the dma padding
limitation.

Valerio, can you try this patch?

Thanks,

diff --git a/drivers/ide/ide-cd.c b/drivers/ide/ide-cd.c
index 3308b1c..6961877 100644
--- a/drivers/ide/ide-cd.c
+++ b/drivers/ide/ide-cd.c

@@ -1250,7 +1250,7 @@ static void cdrom_do_block_pc(ide_drive_t *drive, struct request *rq)
 		 * NOTE! The "len" and "addr" checks should possibly have
 		 * separate masks.
 		 */
-		alignment = queue_dma_alignment(q) | q->dma_pad_mask;
+		alignment = queue_dma_alignment(q) | 15;
 		if ((unsigned long)buf & alignment || rq->data_len & alignment
 		    || object_is_on_stack(buf))
 			drive->dma = 0;

@@ -1981,7 +1981,6 @@ static int ide_cdrom_setup(ide_drive_t *drive)
 
 	blk_queue_prep_rq(drive->queue, ide_cdrom_prep_fn);
 	blk_queue_dma_alignment(drive->queue, 31);
-	blk_queue_update_dma_pad(drive->queue, 15);
 	drive->queue->unplug_delay = (1 * HZ) / 1000;
 	if (!drive->queue->unplug_delay)
 		drive->queue->unplug_delay = 1;