drm/amdgpu: fix a possible NULL dereference when attempting reset
Subsystems (2)
PATCH 1/1 drm/amdgpu: fix a possible NULL dereference when attempting reset
[drm:gfx_v8_0_priv_reg_irq] *ERROR* Illegal register access in command stream [drm] IP block:gmc_v8_0 is hung!
[drm] IP block:gfx_v8_0 is hung!
BUG: unable to handle kernel NULL pointer dereference at 00000000000000d8 IP: amd_sched_hw_job_reset+0x3c/0x9a PGD 3aedd8067 P4D 3aedd8067 PUD 3aedd9067 PMD 0 Oops: 0000 [#1] PREEMPT SMP Modules linked in: cpufreq_conservative bnep bluetooth ecdh_generic serial_ir snd_hrtimer snd_seq_dummy snd_seq_midi snd_rawmidi snd_seq_oss snd_seq_midi_event snd_seq snd_seq_device nct6775 em28xx_rc tda18271 cxd2820r joydev em28xx_dvb usb_storage em28xx tveeprom snd_hda_codec_realtek snd_hda_codec_generic snd_hda_codec_hdmi snd_hda_intel snd_hda_codec snd_hda_core snd_pcm_oss snd_mixer_oss sp5100_tco sg snd_pcm snd_timer CPU: 0 PID: 34 Comm: kworker/0:1 Not tainted 4.14.0-rc4+ #3 Hardware name: System manufacturer System Product Name/A88X-PRO, BIOS 1602 12/04/2014 Workqueue: events amdgpu_irq_reset_work_func task: ffff88041caa44c0 task.stack: ffffc90000144000 RIP: 0010:amd_sched_hw_job_reset+0x3c/0x9a RSP: 0018:ffffc90000147de8 EFLAGS: 00010293 RAX: ffff88031adee850 RBX: ffff88031adee800 RCX: 0000000000000001 RDX: 0000000000000001 RSI: 0000000000000000 RDI: ffff88040db14c08 RBP: ffffc90000147e08 R08: ffff88040dac71d0 R09: ffff88040dac71c0 R10: 0000000000000000 R11: ffff880409575038 R12: ffff88040db14c08 R13: ffff88040db14bf8 R14: ffff88040db14b50 R15: ffff8803e1575580 FS: 0000000000000000(0000) GS:ffff88041ec00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00000000000000d8 CR3: 00000003aec3e000 CR4: 00000000000406f0 Call Trace: amdgpu_gpu_reset+0x9b/0x55b ? _raw_spin_unlock_irq+0x12/0x24 amdgpu_irq_reset_work_func+0x16/0x18 process_one_work+0x124/0x1db ? rescuer_thread+0x26a/0x26a worker_thread+0x19d/0x250 ? rescuer_thread+0x26a/0x26a kthread+0xf1/0xf6 Code: 00 41 54 4c 8d a7 b8 00 00 00 53 4c 89 e7 e8 d9 a9 26 00 49 8b 86 b0 00 00 00 48 8d 58 b0 48 8d 43 50 4c 39 e8 74 51 48 8b 73 10 <48> 8b be d8 00 00 00 48 85 ff 74 37 48 81 c6 c0 00 00 00 e8 6c RIP: amd_sched_hw_job_reset+0x3c/0x9a RSP: ffffc90000147de8
CR2: 00000000000000d8
--- drivers/gpu/drm/amd/scheduler/gpu_scheduler.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/amd/scheduler/gpu_scheduler.c +1 -1
@@ -427,7 +427,7 @@ void amd_sched_hw_job_reset(struct amd_gpu_scheduler *sched)spin_lock (&sched->job_list_lock);list_for_each_entry_reverse (s_job, &sched->ring_mirror_list, node) { -if (s_job->s_fence->parent && +if (s_job->s_fence && s_job->s_fence->parent &&dma_fence_remove_callback (s_job->s_fence->parent, &s_job->s_fence->cb)) {dma_fence_put (s_job->s_fence->parent);
Signature · 6 lines
2.11.0 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
A message that is not in the archive (sent off-list or not yet received): 20171017184134.5171-1-devspam-kUQ7/2e57v/LS5A4iEQ01Vpr/1R2p/[email protected]
Yeah, we already stumbled over that internally as well.
The patch is incorrect, the problem is that we forgot to keep an extra reference on the s_fence to avoid freeing it to early.
The correct fix should be on Alex public branch by the end of today.
Regards,
Christian.
Am 17.10.2017 um 20:41 schrieb Darren Salt:
Quoted 46 lines
[drm:gfx_v8_0_priv_reg_irq] *ERROR* Illegal register access in command stream [drm] IP block:gmc_v8_0 is hung!
[drm] IP block:gfx_v8_0 is hung!BUG: unable to handle kernel NULL pointer dereference at 00000000000000d8 IP: amd_sched_hw_job_reset+0x3c/0x9a PGD 3aedd8067 P4D 3aedd8067 PUD 3aedd9067 PMD 0 Oops: 0000 [#1] PREEMPT SMP Modules linked in: cpufreq_conservative bnep bluetooth ecdh_generic serial_ir snd_hrtimer snd_seq_dummy snd_seq_midi snd_rawmidi snd_seq_oss snd_seq_midi_event snd_seq snd_seq_device nct6775 em28xx_rc tda18271 cxd2820r joydev em28xx_dvb usb_storage em28xx tveeprom snd_hda_codec_realtek snd_hda_codec_generic snd_hda_codec_hdmi snd_hda_intel snd_hda_codec snd_hda_core snd_pcm_oss snd_mixer_oss sp5100_tco sg snd_pcm snd_timer CPU: 0 PID: 34 Comm: kworker/0:1 Not tainted 4.14.0-rc4+ #3 Hardware name: System manufacturer System Product Name/A88X-PRO, BIOS 1602 12/04/2014 Workqueue: events amdgpu_irq_reset_work_func task: ffff88041caa44c0 task.stack: ffffc90000144000 RIP: 0010:amd_sched_hw_job_reset+0x3c/0x9a RSP: 0018:ffffc90000147de8 EFLAGS: 00010293 RAX: ffff88031adee850 RBX: ffff88031adee800 RCX: 0000000000000001 RDX: 0000000000000001 RSI: 0000000000000000 RDI: ffff88040db14c08 RBP: ffffc90000147e08 R08: ffff88040dac71d0 R09: ffff88040dac71c0 R10: 0000000000000000 R11: ffff880409575038 R12: ffff88040db14c08 R13: ffff88040db14bf8 R14: ffff88040db14b50 R15: ffff8803e1575580 FS: 0000000000000000(0000) GS:ffff88041ec00000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00000000000000d8 CR3: 00000003aec3e000 CR4: 00000000000406f0 Call Trace: amdgpu_gpu_reset+0x9b/0x55b ? _raw_spin_unlock_irq+0x12/0x24 amdgpu_irq_reset_work_func+0x16/0x18 process_one_work+0x124/0x1db ? rescuer_thread+0x26a/0x26a worker_thread+0x19d/0x250 ? rescuer_thread+0x26a/0x26a kthread+0xf1/0xf6 Code: 00 41 54 4c 8d a7 b8 00 00 00 53 4c 89 e7 e8 d9 a9 26 00 49 8b 86 b0 00 00 00 48 8d 58 b0 48 8d 43 50 4c 39 e8 74 51 48 8b 73 10 <48> 8b be d8 00 00 00 48 85 ff 74 37 48 81 c6 c0 00 00 00 e8 6c RIP: amd_sched_hw_job_reset+0x3c/0x9a RSP: ffffc90000147de8CR2:
00000000000000d8Signed-off-by Darren Salt <[email protected]>--- drivers/gpu/drm/amd/scheduler/gpu_scheduler.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)drivers/gpu/drm/amd/scheduler/gpu_scheduler.c +0
-0
drivers/gpu/drm/amd/scheduler/gpu_scheduler.c void amd_sched_hw_job_reset(struct amd_gpu_scheduler *sched) · line 433
Quoted 8 lines
drivers/gpu/drm/amd/scheduler/gpu_scheduler.c +1
-1@@ -427,7 +427,7 @@ void amd_sched_hw_job_reset(struct amd_gpu_scheduler *sched)spin_lock (&sched->job_list_lock);list_for_each_entry_reverse (s_job, &sched->ring_mirror_list, node) { -if (s_job->s_fence->parent && +if (s_job->s_fence && s_job->s_fence->parent &&dma_fence_remove_callback (s_job->s_fence->parent, &s_job->s_fence->cb)) {dma_fence_put (s_job->s_fence->parent);
_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx