DMARC-Filter: OpenDMARC Filter v1.3.2 mx1.redhat.com 0CEE161D12
Date: Wed, 12 Apr 2017 12:14:26 +0800
From: Peter Xu <peterx@redhat.com>
To: Alex Williamson <alex.williamson@redhat.com>
Cc: kvm@vger.kernel.org, eric.auger@redhat.com, kwankhede@nvidia.com,
        linux-kernel@vger.kernel.org, slp@redhat.com
Subject: Re: [PATCH v3] vfio/type1: Remove locked page accounting workqueue
Message-ID: <20170412041426.GC16464@pxdev.xzpeter.org>
References: <20170411192644.2131.68702.stgit@gimli.home>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
In-Reply-To: <20170411192644.2131.68702.stgit@gimli.home>
User-Agent: Mutt/1.5.24 (2015-08-30)
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1263
Lines: 26

On Tue, Apr 11, 2017 at 01:28:06PM -0600, Alex Williamson wrote:
> If the mmap_sem is contented then the vfio type1 IOMMU backend will
> defer locked page accounting updates to a workqueue task.  This has a
> few problems and depending on which side the user tries to play, they
> might be over-penalized for unmaps that haven't yet been accounted or
> race the workqueue to enter more mappings than they're allowed.  The
> original intent of this workqueue mechanism seems to be focused on
> reducing latency through the ioctl, but we cannot do so at the cost
> of correctness.  Remove this workqueue mechanism and update the
> callers to allow for failure.  We can also now recheck the limit under
> write lock to make sure we don't exceed it.
> 
> vfio_pin_pages_remote() also now necessarily includes an unwind path
> which we can jump to directly if the consecutive page pinning finds
> that we're exceeding the user's memory limits.  This avoids the
> current lazy approach which does accounting and mapping up to the
> fault, only to return an error on the next iteration to unwind the
> entire vfio_dma.
> 
> Cc: stable@vger.kernel.org
> Signed-off-by: Alex Williamson <alex.williamson@redhat.com>

Reviewed-by: Peter Xu <peterx@redhat.com>

-- 
Peter Xu