[PATCH][BUG] Incorrect SCSI transfer length computation from odd sized scsi_execute_async() transfers.

[PATCH][BUG] Incorrect SCSI transfer length computation from odd sized scsi_execute_async() transfers.

[Jeremy Linton]

Any function which use scsi_execute_async() and transfers "odd" sized 
data that doesn't align correctly with the segment sizes may have its 
transfer length padded out to the closest segment size.

For writes, this results in unnecessary data being transfered to the 
SCSI target. For reads, it affects the residual data length being 
returned to the application since the residual length will be based on 
the padded transfer size rather than the actual request size.

The easiest way to see this is by trying to read using the SG_IO ioctl a 
large (>32k) buffer size from a tape device that only has a few bytes of 
data stored for the current block. The resulting resid will generally be 
incorrect.

I've fixed this simply by changing scsi_req_map_sg() so that it places 
the requested transfer length in rq->data_len rather than the sum of all 
the sg segments.

This patch applies against scsi_lib.c in 2.6.22.

Signed-off-by: Jeremy Linton <jli@greshamstorage.com>


--- linux-2.6.22/drivers/scsi/scsi_lib.c.orig   2007-07-11 
19:07:06.000000000 -0500
+++ linux-2.6.22/drivers/scsi/scsi_lib.c        2007-07-11 
18:43:36.000000000 -0500
@@ -350,7 +350,7 @@ static int scsi_req_map_sg(struct reques
         }

         rq->buffer = rq->data = NULL;
-       rq->data_len = data_len;
+       rq->data_len = bufflen;
         return 0;

  free_bios:

Re: [PATCH][BUG] Incorrect SCSI transfer length computation from odd sized scsi_execute_async() transfers.

[Mike Christie]

Jeremy Linton wrote:
> Any function which use scsi_execute_async() and transfers "odd" sized 
> data that doesn't align correctly with the segment sizes may have its 
> transfer length padded out to the closest segment size.
> 
> For writes, this results in unnecessary data being transfered to the 
> SCSI target. For reads, it affects the residual data length being 
> returned to the application since the residual length will be based on 
> the padded transfer size rather than the actual request size.
> 
> The easiest way to see this is by trying to read using the SG_IO ioctl a 
> large (>32k) buffer size from a tape device that only has a few bytes of 
> data stored for the current block. The resulting resid will generally be 
> incorrect.
> 
> I've fixed this simply by changing scsi_req_map_sg() so that it places 
> the requested transfer length in rq->data_len rather than the sum of all 
> the sg segments.
> 
> This patch applies against scsi_lib.c in 2.6.22.
> 
> Signed-off-by: Jeremy Linton <jli@greshamstorage.com>
> 
> 
> --- linux-2.6.22/drivers/scsi/scsi_lib.c.orig   2007-07-11 
> 19:07:06.000000000 -0500
> +++ linux-2.6.22/drivers/scsi/scsi_lib.c        2007-07-11 
> 18:43:36.000000000 -0500
> @@ -350,7 +350,7 @@ static int scsi_req_map_sg(struct reques
>         }
> 
>         rq->buffer = rq->data = NULL;
> -       rq->data_len = data_len;
> +       rq->data_len = bufflen;
>         return 0;
> 

I think you needed some other bits in there. See this patch 
http://marc.info/?l=linux-scsi&m=117392208211297&w=2

I tried just setting the bufflen first, and that still had problems. 
Could you try the patch here

http://marc.info/?l=linux-scsi&m=117392208211297&w=2

Re: [PATCH][BUG] Incorrect SCSI transfer length computation from odd sized scsi_execute_async() transfers.

[Jeremy Linton]

Mike Christie wrote:
> I think you needed some other bits in there. See this patch 
> I tried just setting the bufflen first, and that still had problems. 
> Could you try the patch here
> 
> http://marc.info/?l=linux-scsi&m=117392208211297&w=2


I just read the thread.. I didn't see any strange retries with my test 
case. I will try duplicating the problem tomorrow. Then I will apply 
your patch and rerun my test.

I'm curious if this has been known since 2.6.19 why the patch hasn't 
propagated to the main kernel tree?