On Tue, 20 Sep 2005, Kai Makisara wrote:
...
> If I try 5121 kB, write fails as is expected but not completely nicely.
> Here are some test printks:
>
> scsi_req_map_sg: q->back_merge failed, i=10
> st 1:0:5:0: extraneous data discarded.
> st 1:0:5:0: COMMAND FAILED (87 0 1).
> sym: cmd: 0x0a 0x00 0x50 0x04 0x00 0x00
> st: cmd=0x0a result=0x70000 resid=0 sense[0]=0x00 sense[2]=0x00
>
> scsi_req_map_sg fails as it should but still a bogus SCSI command is sent.
> I think the reason for this is simple but I don't want to delay the good
> news by trying to debug this.
>
The reason was simple: scsi_req_map_sg did not set err when q->back_merge
failed. The patch at the end fixes the problem (I am not sure what would
be the correct error code).
--
Kai
--- linux-2.6.14-rc1-blk3/drivers/scsi/scsi_lib.c.old 2005-09-22 22:57:53.000000000 +0300
+++ linux-2.6.14-rc1-blk3/drivers/scsi/scsi_lib.c 2005-09-22 22:58:14.000000000 +0300
@@ -360,6 +360,7 @@
blk_rq_bio_prep(q, rq, bio);
else if (!q->back_merge_fn(q, rq, bio)) {
bio_endio(bio, bio->bi_size, 0);
+ err = -ENOMEM;
goto free_bios;
} else {
rq->biotail->bi_next = bio;
-
: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
