From: "Christian König" <christian.koenig@amd.com> To: Maarten Lankhorst <maarten.lankhorst@canonical.com>, <airlied@linux.ie> Cc: <thellstrom@vmware.com>, <nouveau@lists.freedesktop.org>, <linux-kernel@vger.kernel.org>, <dri-devel@lists.freedesktop.org>, <bskeggs@redhat.com>, <alexander.deucher@amd.com> Subject: Re: [PATCH 09/19] drm/radeon: handle lockup in delayed work, v2 Date: Mon, 4 Aug 2014 10:36:59 +0200 [thread overview] Message-ID: <53DF462B.2060102@amd.com> (raw) In-Reply-To: <53DBD269.80807@canonical.com> Hi Maarten, Sorry for the delay. I've got way to much todo recently. Am 01.08.2014 um 19:46 schrieb Maarten Lankhorst: > > On 01-08-14 18:35, Christian König wrote: >> Am 31.07.2014 um 17:33 schrieb Maarten Lankhorst: >>> Signed-off-by: Maarten Lankhorst <maarten.lankhorst@canonical.com> >>> --- >>> V1 had a nasty bug breaking gpu lockup recovery. The fix is not >>> allowing radeon_fence_driver_check_lockup to take exclusive_lock, >>> and kill it during lockup recovery instead. >> That looks like the delayed work starts running as soon as we submit a fence, and not when it's needed for waiting. >> >> Since it's a backup for failing IRQs I would rather put it into radeon_irq_kms.c and start/stop it when the IRQs are started/stoped. > The delayed work is not just for failing irq's, it's also the handler that's used to detect lockups, which is why I trigger after processing fences, and reset the timer after processing. The idea was turning the delayed work on and off when we turn the irq on and off as well, processing of the delayed work handler can still happen in radeon_fence.c > > Specifically what happened was this scenario: > > - lock up occurs > - write lock taken by gpu_reset > - delayed work runs, tries to acquire read lock, blocks > - gpu_reset tries to cancel delayed work synchronously > - has to wait for delayed work to finish -> deadlock Why do you want to disable the work item from the lockup handler in the first place? Just take the exclusive lock in the work item, when it concurrently runs with the lockup handler it will just block for the lockup handler to complete. Regards, Christian. > ~Maarten
WARNING: multiple messages have this Message-ID (diff)
From: "Christian König" <christian.koenig@amd.com> To: Maarten Lankhorst <maarten.lankhorst@canonical.com>, airlied@linux.ie Cc: thellstrom@vmware.com, nouveau@lists.freedesktop.org, linux-kernel@vger.kernel.org, dri-devel@lists.freedesktop.org, bskeggs@redhat.com, alexander.deucher@amd.com Subject: Re: [PATCH 09/19] drm/radeon: handle lockup in delayed work, v2 Date: Mon, 4 Aug 2014 10:36:59 +0200 [thread overview] Message-ID: <53DF462B.2060102@amd.com> (raw) In-Reply-To: <53DBD269.80807@canonical.com> Hi Maarten, Sorry for the delay. I've got way to much todo recently. Am 01.08.2014 um 19:46 schrieb Maarten Lankhorst: > > On 01-08-14 18:35, Christian König wrote: >> Am 31.07.2014 um 17:33 schrieb Maarten Lankhorst: >>> Signed-off-by: Maarten Lankhorst <maarten.lankhorst@canonical.com> >>> --- >>> V1 had a nasty bug breaking gpu lockup recovery. The fix is not >>> allowing radeon_fence_driver_check_lockup to take exclusive_lock, >>> and kill it during lockup recovery instead. >> That looks like the delayed work starts running as soon as we submit a fence, and not when it's needed for waiting. >> >> Since it's a backup for failing IRQs I would rather put it into radeon_irq_kms.c and start/stop it when the IRQs are started/stoped. > The delayed work is not just for failing irq's, it's also the handler that's used to detect lockups, which is why I trigger after processing fences, and reset the timer after processing. The idea was turning the delayed work on and off when we turn the irq on and off as well, processing of the delayed work handler can still happen in radeon_fence.c > > Specifically what happened was this scenario: > > - lock up occurs > - write lock taken by gpu_reset > - delayed work runs, tries to acquire read lock, blocks > - gpu_reset tries to cancel delayed work synchronously > - has to wait for delayed work to finish -> deadlock Why do you want to disable the work item from the lockup handler in the first place? Just take the exclusive lock in the work item, when it concurrently runs with the lockup handler it will just block for the lockup handler to complete. Regards, Christian. > ~Maarten _______________________________________________ dri-devel mailing list dri-devel@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/dri-devel
next prev parent reply other threads:[~2014-08-04 8:37 UTC|newest] Thread overview: 69+ messages / expand[flat|nested] mbox.gz Atom feed top 2014-07-31 15:32 [PATCH 01/19] fence: add debugging lines to fence_is_signaled for the callback Maarten Lankhorst 2014-07-31 15:32 ` Maarten Lankhorst 2014-07-31 15:32 ` [PATCH 02/19] drm/ttm: add interruptible parameter to ttm_eu_reserve_buffers Maarten Lankhorst 2014-07-31 15:32 ` Maarten Lankhorst 2014-07-31 15:33 ` [PATCH 03/19] drm/ttm: kill off some members to ttm_validate_buffer Maarten Lankhorst 2014-07-31 15:33 ` Maarten Lankhorst 2014-07-31 15:33 ` [PATCH 04/19] drm/nouveau: add reservation to nouveau_gem_ioctl_cpu_prep Maarten Lankhorst 2014-07-31 15:33 ` [PATCH 05/19] drm/nouveau: require reservations for nouveau_fence_sync and nouveau_bo_fence Maarten Lankhorst 2014-07-31 15:33 ` [PATCH 06/19] drm/ttm: call ttm_bo_wait while inside a reservation Maarten Lankhorst 2014-07-31 15:33 ` Maarten Lankhorst 2014-07-31 15:33 ` [PATCH 07/19] drm/ttm: kill fence_lock Maarten Lankhorst 2014-07-31 15:33 ` Maarten Lankhorst 2014-07-31 15:33 ` [PATCH 08/19] drm/nouveau: rework to new fence interface Maarten Lankhorst 2014-07-31 15:33 ` Maarten Lankhorst 2014-07-31 15:33 ` [PATCH 09/19] drm/radeon: handle lockup in delayed work, v2 Maarten Lankhorst 2014-07-31 15:33 ` Maarten Lankhorst 2014-08-01 16:35 ` Christian König 2014-08-01 16:35 ` Christian König 2014-08-01 17:46 ` Maarten Lankhorst 2014-08-01 17:46 ` Maarten Lankhorst 2014-08-04 8:36 ` Christian König [this message] 2014-08-04 8:36 ` Christian König 2014-08-04 8:55 ` Maarten Lankhorst 2014-08-04 8:55 ` Maarten Lankhorst 2014-08-04 11:57 ` Christian König 2014-08-04 11:57 ` Christian König 2014-08-04 13:34 ` Maarten Lankhorst 2014-08-04 13:34 ` Maarten Lankhorst 2014-08-04 14:37 ` Christian König 2014-08-04 14:37 ` Christian König 2014-08-04 14:40 ` Maarten Lankhorst 2014-08-04 14:40 ` Maarten Lankhorst 2014-08-04 14:45 ` Christian König 2014-08-04 14:45 ` Christian König 2014-08-04 14:58 ` Maarten Lankhorst 2014-08-04 14:58 ` Maarten Lankhorst 2014-08-04 15:04 ` Christian König 2014-08-04 15:04 ` Christian König 2014-08-04 15:09 ` Maarten Lankhorst 2014-08-04 15:09 ` Maarten Lankhorst 2014-08-04 17:04 ` Christian König 2014-08-04 17:04 ` Christian König 2014-08-05 8:16 ` Daniel Vetter 2014-08-05 8:16 ` Daniel Vetter 2014-08-05 9:34 ` Maarten Lankhorst 2014-08-05 9:34 ` Maarten Lankhorst 2014-07-31 15:33 ` [PATCH 10/19] drm/radeon: add timeout argument to radeon_fence_wait_seq Maarten Lankhorst 2014-07-31 15:33 ` Maarten Lankhorst 2014-07-31 15:33 ` [PATCH 11/19] drm/radeon: use common fence implementation for fences, v2 Maarten Lankhorst 2014-07-31 15:34 ` [PATCH 12/19] drm/qxl: rework to new fence interface Maarten Lankhorst 2014-07-31 15:34 ` Maarten Lankhorst 2014-07-31 15:34 ` [PATCH 13/19] drm/vmwgfx: get rid of different types of fence_flags entirely Maarten Lankhorst 2014-07-31 15:34 ` Maarten Lankhorst 2014-07-31 15:34 ` [PATCH 14/19] drm/vmwgfx: rework to new fence interface Maarten Lankhorst 2014-07-31 15:34 ` Maarten Lankhorst 2014-07-31 15:34 ` [PATCH 15/19] drm/ttm: flip the switch, and convert to dma_fence Maarten Lankhorst 2014-07-31 15:34 ` [PATCH 16/19] drm/nouveau: use rcu in nouveau_gem_ioctl_cpu_prep Maarten Lankhorst 2014-07-31 15:34 ` [PATCH 17/19] drm/radeon: use rcu waits in some ioctls Maarten Lankhorst 2014-07-31 15:34 ` Maarten Lankhorst 2014-08-01 8:27 ` Michel Dänzer 2014-08-01 10:12 ` Maarten Lankhorst 2014-08-01 14:13 ` Michel Dänzer 2014-08-01 17:07 ` Maarten Lankhorst 2014-08-04 8:42 ` Michel Dänzer 2014-08-04 8:56 ` Maarten Lankhorst 2014-08-04 9:25 ` Michel Dänzer 2014-08-04 9:30 ` Maarten Lankhorst 2014-07-31 15:34 ` [PATCH 18/19] drm/vmwgfx: use rcu in vmw_user_dmabuf_synccpu_grab Maarten Lankhorst 2014-07-31 15:34 ` [PATCH 19/19] drm/ttm: use rcu in core ttm Maarten Lankhorst
Reply instructions: You may reply publicly to this message via plain-text email using any one of the following methods: * Save the following mbox file, import it into your mail client, and reply-to-all from there: mbox Avoid top-posting and favor interleaved quoting: https://en.wikipedia.org/wiki/Posting_style#Interleaved_style * Reply using the --to, --cc, and --in-reply-to switches of git-send-email(1): git send-email \ --in-reply-to=53DF462B.2060102@amd.com \ --to=christian.koenig@amd.com \ --cc=airlied@linux.ie \ --cc=alexander.deucher@amd.com \ --cc=bskeggs@redhat.com \ --cc=dri-devel@lists.freedesktop.org \ --cc=linux-kernel@vger.kernel.org \ --cc=maarten.lankhorst@canonical.com \ --cc=nouveau@lists.freedesktop.org \ --cc=thellstrom@vmware.com \ /path/to/YOUR_REPLY https://kernel.org/pub/software/scm/git/docs/git-send-email.html * If your mail client supports setting the In-Reply-To header via mailto: links, try the mailto: linkBe sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.