From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail333.us4.mandrillapp.com ([205.201.137.77]:43800 "EHLO mail333.us4.mandrillapp.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1756125AbcCAX4z convert rfc822-to-8bit (ORCPT ); Tue, 1 Mar 2016 18:56:55 -0500 Received: from pmta03.dal05.mailchimp.com (127.0.0.1) by mail333.us4.mandrillapp.com id hqom4e174no2 for ; Tue, 1 Mar 2016 23:54:45 +0000 (envelope-from ) From: Greg Kroah-Hartman Subject: [PATCH 4.4 187/342] drm/amdgpu: hold reference to fences in amdgpu_sa_bo_new (v2) To: Cc: Greg Kroah-Hartman , , =?utf-8?Q?Nicolai=20H=C3=A4hnle?= , Alex Deucher , =?utf-8?Q?Christian=20K=C3=B6nig?= Message-Id: <20160301234533.988821471@linuxfoundation.org> In-Reply-To: <20160301234527.990448862@linuxfoundation.org> References: <20160301234527.990448862@linuxfoundation.org> Date: Tue, 01 Mar 2016 23:54:45 +0000 MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 8BIT Sender: stable-owner@vger.kernel.org List-ID: 4.4-stable review patch. If anyone has any objections, please let me know. ------------------ From: Nicolai Hähnle commit a8d81b36267366603771431747438d18f32ae2d5 upstream. An arbitrary amount of time can pass between spin_unlock and fence_wait_any_timeout, so we need to ensure that nobody frees the fences from under us. A stress test (rapidly starting and killing hundreds of glxgears instances) ran into a deadlock in fence_wait_any_timeout after about an hour, and this race condition appears to be a plausible cause. v2: agd: rebase on upstream Signed-off-by: Nicolai Hähnle Reviewed-by: Alex Deucher Reviewed-by: Christian König Signed-off-by: Greg Kroah-Hartman --- drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c @@ -354,12 +354,15 @@ int amdgpu_sa_bo_new(struct amdgpu_sa_ma for (i = 0, count = 0; i < AMDGPU_MAX_RINGS; ++i) if (fences[i]) - fences[count++] = fences[i]; + fences[count++] = fence_get(fences[i]); if (count) { spin_unlock(&sa_manager->wq.lock); t = fence_wait_any_timeout(fences, count, false, MAX_SCHEDULE_TIMEOUT); + for (i = 0; i < count; ++i) + fence_put(fences[i]); + r = (t > 0) ? 0 : t; spin_lock(&sa_manager->wq.lock); } else {