Re: [PATCH 2.6.13] scsi: sd fails to copy cmd_len on SG_IO

[Mike Christie <michaelc@cs.wisc.edu>]

Timothy Thelin wrote:
> > -----Original Message-----
> > From: Mike Christie [mailto:michaelc@cs.wisc.edu]
> > Sent: Tuesday, September 13, 2005 10:49 AM
> > To: Timothy Thelin
> > Cc: James Bottomley; SCSI Mailing List; Linux Kernel; Andrew Morton
> > Subject: Re: [PATCH 2.6.13] scsi: sd fails to copy cmd_len on SG_IO
> >
> >
> > Timothy Thelin wrote:
> >
> > > This fixes an issue when doing SG_IO on an sd device: the
> > > sd driver fails to copy the request's cmd_len to the scsi
> > > command's cmd_len when initializing the command.
> >
> > Do you need the same fix to st, sr, and scsi_lib (in the 
> > scsi_generic_done path)?
>
>
> I just looked, and st and sr look like they need the same fix, but i'm
> unaware of where scsi_lib might need it (I'm new to the Linux scsi stack).
> Mind elaborating on your thoughts of the scsi_generic_done path? (I cant
> find
> the symbol in drivers/scsi/*)

look at the bottom of scsi_prep_fn() in scsi_lib.c. SCSI-ml will do 
block_pc commands (which come down like block layer SG_IO ones) for 
things like inquiry and report luns when scanning a host. At that time 
there is no upper layer driver like sd attached so scsi_generic_done() 
is used as a callback for finishing up commands.


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/