From mboxrd@z Thu Jan 1 00:00:00 1970 From: James Bottomley Subject: Re: [PATCH 0/2] mpt2sas,mpt3sas - PCI master abort fixups Date: Sun, 12 Apr 2015 17:54:05 -0700 Message-ID: <1428886445.2196.43.camel@HansenPartnership.com> References: <1419948455-31624-1-git-send-email-joe.lawrence@stratus.com> <552B0998.1050808@stratus.com> Mime-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 7bit Return-path: Received: from bedivere.hansenpartnership.com ([66.63.167.143]:40482 "EHLO bedivere.hansenpartnership.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751776AbbDMAyH (ORCPT ); Sun, 12 Apr 2015 20:54:07 -0400 In-Reply-To: <552B0998.1050808@stratus.com> Sender: linux-scsi-owner@vger.kernel.org List-Id: linux-scsi@vger.kernel.org To: Joe Lawrence Cc: linux-scsi@vger.kernel.org, Nagalakshmi Nandigama , Praveen Krishnamoorthy , Sreekanth Reddy , Abhijit Mahajan , Christoph Hellwig 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