Received-SPF: pass (google.com: domain of linux-kernel+bounces-103606-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.80.249 as permitted sender) client-ip=147.75.80.249;
Date: Thu, 14 Mar 2024 13:09:48 -0400
From: Johannes Weiner <hannes@cmpxchg.org>
To: Christian =?iso-8859-1?Q?K=F6nig?= <ckoenig.leichtzumerken@gmail.com>
Cc: Alex Deucher <alexander.deucher@amd.com>,
	"Sharma, Shashank" <Shashank.Sharma@amd.com>,
	Christian =?iso-8859-1?Q?K=F6nig?= <christian.koenig@amd.com>,
	amd-gfx@lists.freedesktop.org, dri-devel@lists.freedesktop.org,
	linux-kernel@vger.kernel.org
Subject: Re: [PATCH] drm/amdgpu: fix deadlock while reading mqd from debugfs
Message-ID: <20240314170948.GA581298@cmpxchg.org>
References: <20240307221609.7651-1-hannes@cmpxchg.org>
 <c411dce6-faaf-46c3-8bb6-8c4db871e598@gmail.com>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <c411dce6-faaf-46c3-8bb6-8c4db871e598@gmail.com>

Hello,

On Fri, Mar 08, 2024 at 12:32:33PM +0100, Christian K?nig wrote:
> Am 07.03.24 um 23:07 schrieb Johannes Weiner:
> > Lastly I went with an open loop instead of a memcpy() as I wasn't
> > sure if that memory is safe to address a byte at at time.

Shashank pointed out to me in private that byte access would indeed be
safe. However, after actually trying it it won't work because memcpy()
doesn't play nice with mqd being volatile:

/home/hannes/src/linux/linux/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c: In function 'amdgpu_debugfs_mqd_read':
/home/hannes/src/linux/linux/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c:550:22: warning: passing argument 1 of '__builtin_dynamic_object_size' discards 'volatil' qualifier from pointer target type [-Wdiscarded-qualifiers]
  550 |         memcpy(kbuf, mqd, ring->mqd_size);

So I would propose leaving the patch as-is. Shashank, does that sound
good to you?

(Please keep me CC'd on replies, as I'm not subscribed to the graphics
lists.)

Thanks!