Re: [PATCH 0/2] mpt2sas,mpt3sas - PCI master abort fixups

[James Bottomley <James.Bottomley@HansenPartnership.com>]

On Sun, 2015-04-12 at 20:11 -0400, Joe Lawrence wrote:
> On 12/30/2014 09:07 AM, Joe Lawrence wrote:
> > A colleague noticed that the mpt2 and mpt3sas drivers do not correctly
> > check the PCI master abort pattern in _base_wait_for_doorbell_ack.  This
> > pattern should be checked *prior* to any valid bit patterns, which would
> > always return true since a PCI read on master abort sets all bits high.
> > 
> > The second patch adds similar checking to _base_wait_for_doorbell_int and
> > _base_wait_for_doorbell_not_used to avoid potentially long loops around
> > PCI reads.
> > 
> > Joe Lawrence (2):
> >   mpt2sas,mpt3sas: correct master-abort checking in doorbell ack
> >   mpt2sas,mpt3sas: additional master abort checks
> > 
> >  drivers/scsi/mpt2sas/mpt2sas_base.c |   17 ++++++++++++-----
> >  drivers/scsi/mpt3sas/mpt3sas_base.c |   17 ++++++++++++-----
> >  2 files changed, 24 insertions(+), 10 deletions(-)
> > 
> 
> Avago ping?
> 
> This one was pretty straightforward: check 0xFFFFFFFF *before* any
> individual bit(s), i.e. before reading the doorbell register.

OK, Joe, explain why this patch is important: what problems could result
from it not being present?  If you convince everyone then no more mpt2/3
sas patches until this is at least commented on and a plan of action
proposed.

James