[ofa-general] Re: [PATCH 2.6.30.4] Fix for NULL pointer dereference by SRP initiator triggered by a SCSI reset after the SRP connection has been closed

[James Bottomley]

On Wed, 2009-08-05 at 19:54 +0200, Bart Van Assche wrote:
> On Wed, Aug 5, 2009 at 7:44 PM, Roland Dreier<rdreier at cisco.com> wrote:
> >
> >  > The NULL pointer dereference happens when srp_reset_device() calls
> >  > srp_send_tsk_mgmt(target, req, SRP_TSK_LUN_RESET) with
> >  > req->scmnd->device == NULL. When the sg_reset command issues an
> >  > SG_SCSI_RESET ioctl, scsi_reset_provider() is invoked and allocates an
> >  > scmnd structure and sets scmnd->device to NULL. It is this scmnd
> >  > structure that is passed to srp_reset_device(). What I'm not sure
> >  > about is whether scsi_reset_provider() should set req->scmnd->device
> >  > to a non-NULL value or whether srp_send_tsk_mgmt() should be able to
> >  > handle the condition req->scmnd->device == NULL.
> >
> > Well, I don't see how the reset ioctl can do anything useful unless it
> > passes a device in with the scsi command -- otherwise for example
> > srp_reset_device() has no idea what LUN to try and reset.
> 
> (added linux-scsi in CC)
> 
> I hope one of the SCSI people can tell us whether the behavior that
> scsi_reset_provider()
> passes the value NULL in req->scmnd->device to
> scsi_try_bus_device_reset() is correct ?

Need more information.

cmd->device is supposed to be initialised in scsi_get_command(), which
scsi_reset_provider() calls ... why do you think it got set to null?

James