Skip to contentSkip to conversation

drm/sched: Fix kernel NULL pointer dereference error

PATCH Superseded Arvind Yadav 4 participants 4 messages Sep 30, 2022 – Oct 8, 2022

Subsystems (4)

v1 of 4 Superseded Latest is v4 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 drm/sched: Fix kernel NULL pointer dereference error

BUG: kernel NULL pointer dereference, address: 0000000000000088
 #PF: supervisor read access in kernel mode
 #PF: error_code(0x0000) - not-present page
 PGD 0 P4D 0
 Oops: 0000 [#1] PREEMPT SMP NOPTI
 CPU: 2 PID: 0 Comm: swapper/2 Not tainted 6.0.0-rc2-custom #1
 Arvind : [dma_fence_default_wait _START] timeout = -1
 Hardware name: AMD Dibbler/Dibbler, BIOS RDB1107CC 09/26/2018
 RIP: 0010:drm_sched_job_done.isra.0+0x11/0x140 [gpu_sched]
 Code: 8b fe ff ff be 03 00 00 00 e8 7b da b7 e3 e9 d4 fe ff ff 66 0f 1f 44 00 00 0f 1f 44 00 00 55 48 89 e5 41 55 41 54 49 89 fc 53 <48> 8b 9f 88 00 00 00 f0 ff 8b f0 00 00 00 48 8b 83 80 01 00 00 f0
 RSP: 0018:ffffb1b1801d4d38 EFLAGS: 00010087
 RAX: ffffffffc0aa48b0 RBX: ffffb1b1801d4d70 RCX: 0000000000000018
 RDX: 000036c70afb7c1d RSI: ffff8a45ca413c60 RDI: 0000000000000000
 RBP: ffffb1b1801d4d50 R08: 00000000000000b5 R09: 0000000000000000
 R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000000
 R13: ffffb1b1801d4d70 R14: ffff8a45c4160000 R15: ffff8a45c416a708
 FS:  0000000000000000(0000) GS:ffff8a48a0a80000(0000) knlGS:0000000000000000
 CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
 CR2: 0000000000000088 CR3: 000000014ad50000 CR4: 00000000003506e0
 Call Trace:
  <IRQ>
  drm_sched_job_done_cb+0x12/0x20 [gpu_sched]
  dma_fence_signal_timestamp_locked+0x7e/0x110
  dma_fence_signal+0x31/0x60
  amdgpu_fence_process+0xc4/0x140 [amdgpu]
  gfx_v9_0_eop_irq+0x9d/0xd0 [amdgpu]
  amdgpu_irq_dispatch+0xb7/0x210 [amdgpu]
  amdgpu_ih_process+0x86/0x100 [amdgpu]
  amdgpu_irq_handler+0x24/0x60 [amdgpu]
  __handle_irq_event_percpu+0x4b/0x190
  handle_irq_event_percpu+0x15/0x50
  handle_irq_event+0x39/0x60
  handle_edge_irq+0xaf/0x210
  __common_interrupt+0x6e/0x110
  common_interrupt+0xc1/0xe0
  </IRQ>
  <TASK>
---
 drivers/gpu/drm/scheduler/sched_main.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)
drivers/gpu/drm/scheduler/sched_main.c +6 -1
index 6684d88463b4..390272f6b126 100644
@@ -172,7 +172,12 @@ drm_sched_rq_select_entity(struct drm_sched_rq *rq)
 static void drm_sched_job_done(struct drm_sched_job *s_job)
 {
 	struct drm_sched_fence *s_fence = s_job->s_fence;
-	struct drm_gpu_scheduler *sched = s_fence->sched;
+	struct drm_gpu_scheduler *sched;
+
+	if (!s_fence)
+		return;
+
+	sched = s_fence->sched;
 
 	atomic_dec(&sched->hw_rq_count);
 	atomic_dec(sched->score);
Signature · 1 line
2.25.1
#1 lore raw

Am 30.09.22 um 10:48 schrieb Arvind Yadav:

Quoted 29 lines
BUG: kernel NULL pointer dereference, address: 0000000000000088
  #PF: supervisor read access in kernel mode
  #PF: error_code(0x0000) - not-present page
  PGD 0 P4D 0
  Oops: 0000 [#1] PREEMPT SMP NOPTI
  CPU: 2 PID: 0 Comm: swapper/2 Not tainted 6.0.0-rc2-custom #1
  Arvind : [dma_fence_default_wait _START] timeout = -1
  Hardware name: AMD Dibbler/Dibbler, BIOS RDB1107CC 09/26/2018
  RIP: 0010:drm_sched_job_done.isra.0+0x11/0x140 [gpu_sched]
  Code: 8b fe ff ff be 03 00 00 00 e8 7b da b7 e3 e9 d4 fe ff ff 66 0f 1f 44 00 00 0f 1f 44 00 00 55 48 89 e5 41 55 41 54 49 89 fc 53 <48> 8b 9f 88 00 00 00 f0 ff 8b f0 00 00 00 48 8b 83 80 01 00 00 f0
  RSP: 0018:ffffb1b1801d4d38 EFLAGS: 00010087
  RAX: ffffffffc0aa48b0 RBX: ffffb1b1801d4d70 RCX: 0000000000000018
  RDX: 000036c70afb7c1d RSI: ffff8a45ca413c60 RDI: 0000000000000000
  RBP: ffffb1b1801d4d50 R08: 00000000000000b5 R09: 0000000000000000
  R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000000
  R13: ffffb1b1801d4d70 R14: ffff8a45c4160000 R15: ffff8a45c416a708
  FS:  0000000000000000(0000) GS:ffff8a48a0a80000(0000) knlGS:0000000000000000
  CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
  CR2: 0000000000000088 CR3: 000000014ad50000 CR4: 00000000003506e0
  Call Trace:
   <IRQ>
   drm_sched_job_done_cb+0x12/0x20 [gpu_sched]
   dma_fence_signal_timestamp_locked+0x7e/0x110
   dma_fence_signal+0x31/0x60
   amdgpu_fence_process+0xc4/0x140 [amdgpu]
   gfx_v9_0_eop_irq+0x9d/0xd0 [amdgpu]
   amdgpu_irq_dispatch+0xb7/0x210 [amdgpu]
   amdgpu_ih_process+0x86/0x100 [amdgpu]
   amdgpu_irq_handler+0x24/0x60 [amdgpu]
Quoted 8 lines
   __handle_irq_event_percpu+0x4b/0x190
   handle_irq_event_percpu+0x15/0x50
   handle_irq_event+0x39/0x60
   handle_edge_irq+0xaf/0x210
   __common_interrupt+0x6e/0x110
   common_interrupt+0xc1/0xe0
   </IRQ>
   <TASK>

How is this triggered any why haven't we seen it before?

Christian

Quoted 23 lines
---
  drivers/gpu/drm/scheduler/sched_main.c | 7 ++++++-
  1 file changed, 6 insertions(+), 1 deletion(-)
drivers/gpu/drm/scheduler/sched_main.c +6 -1
index 6684d88463b4..390272f6b126 100644
@@ -172,7 +172,12 @@ drm_sched_rq_select_entity(struct drm_sched_rq *rq)
  static void drm_sched_job_done(struct drm_sched_job *s_job)
  {
  	struct drm_sched_fence *s_fence = s_job->s_fence;
-	struct drm_gpu_scheduler *sched = s_fence->sched;
+	struct drm_gpu_scheduler *sched;
+
+	if (!s_fence)
+		return;
+
+	sched = s_fence->sched;
 
  	atomic_dec(&sched->hw_rq_count);
  	atomic_dec(sched->score);
#2 lore raw
Quoted 39 lines · Christian König

Am 30.09.22 um 10:48 schrieb Arvind Yadav:

Quoted 29 lines
BUG: kernel NULL pointer dereference, address: 0000000000000088
  #PF: supervisor read access in kernel mode
  #PF: error_code(0x0000) - not-present page
  PGD 0 P4D 0
  Oops: 0000 [#1] PREEMPT SMP NOPTI
  CPU: 2 PID: 0 Comm: swapper/2 Not tainted 6.0.0-rc2-custom #1
  Arvind : [dma_fence_default_wait _START] timeout = -1
  Hardware name: AMD Dibbler/Dibbler, BIOS RDB1107CC 09/26/2018
  RIP: 0010:drm_sched_job_done.isra.0+0x11/0x140 [gpu_sched]
  Code: 8b fe ff ff be 03 00 00 00 e8 7b da b7 e3 e9 d4 fe ff ff 66 0f 1f 44 00 00 0f 1f 44 00 00 55 48 89 e5 41 55 41 54 49 89 fc 53 <48> 8b 9f 88 00 00 00 f0 ff 8b f0 00 00 00 48 8b 83 80 01 00 00 f0
  RSP: 0018:ffffb1b1801d4d38 EFLAGS: 00010087
  RAX: ffffffffc0aa48b0 RBX: ffffb1b1801d4d70 RCX: 0000000000000018
  RDX: 000036c70afb7c1d RSI: ffff8a45ca413c60 RDI: 0000000000000000
  RBP: ffffb1b1801d4d50 R08: 00000000000000b5 R09: 0000000000000000
  R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000000
  R13: ffffb1b1801d4d70 R14: ffff8a45c4160000 R15: ffff8a45c416a708
  FS:  0000000000000000(0000) GS:ffff8a48a0a80000(0000) knlGS:0000000000000000
  CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
  CR2: 0000000000000088 CR3: 000000014ad50000 CR4: 00000000003506e0
  Call Trace:
   <IRQ>
   drm_sched_job_done_cb+0x12/0x20 [gpu_sched]
   dma_fence_signal_timestamp_locked+0x7e/0x110
   dma_fence_signal+0x31/0x60
   amdgpu_fence_process+0xc4/0x140 [amdgpu]
   gfx_v9_0_eop_irq+0x9d/0xd0 [amdgpu]
   amdgpu_irq_dispatch+0xb7/0x210 [amdgpu]
   amdgpu_ih_process+0x86/0x100 [amdgpu]
   amdgpu_irq_handler+0x24/0x60 [amdgpu]
Quoted 8 lines
   __handle_irq_event_percpu+0x4b/0x190
   handle_irq_event_percpu+0x15/0x50
   handle_irq_event+0x39/0x60
   handle_edge_irq+0xaf/0x210
   __common_interrupt+0x6e/0x110
   common_interrupt+0xc1/0xe0
   </IRQ>
   <TASK>
Quoted 1 line

How is this triggered any why haven't we seen it before?

IGT has few 'amdgpu' specific testcases which is not related  to fence.

while running those test cases I have got this crash but this crash is not always reproducible.

~Arvind

Quoted 25 lines

Christian

Quoted 22 lines

---
  drivers/gpu/drm/scheduler/sched_main.c | 7 ++++++-

  1 file changed, 6 insertions(+), 1 deletion(-)
drivers/gpu/drm/scheduler/sched_main.c +6 -1
index 6684d88463b4..390272f6b126 100644
@@ -172,7 +172,12 @@ drm_sched_rq_select_entity(struct drm_sched_rq *rq)
   static void drm_sched_job_done(struct drm_sched_job *s_job)
   {
       struct drm_sched_fence *s_fence = s_job->s_fence;
-    struct drm_gpu_scheduler *sched = s_fence->sched;
+    struct drm_gpu_scheduler *sched;
+
+    if (!s_fence)
+        return;
+
+    sched = s_fence->sched;
         atomic_dec(&sched->hw_rq_count);
       atomic_dec(sched->score);
Tested-by
#3 lore raw
Quoted 44 lines · Arvind
Quoted 42 lines · Christian König

Am 30.09.22 um 10:48 schrieb Arvind Yadav:

Quoted 40 lines
BUG: kernel NULL pointer dereference, address: 0000000000000088
  #PF: supervisor read access in kernel mode
  #PF: error_code(0x0000) - not-present page
  PGD 0 P4D 0
  Oops: 0000 [#1] PREEMPT SMP NOPTI
  CPU: 2 PID: 0 Comm: swapper/2 Not tainted 6.0.0-rc2-custom #1
  Arvind : [dma_fence_default_wait _START] timeout = -1
  Hardware name: AMD Dibbler/Dibbler, BIOS RDB1107CC 09/26/2018
  RIP: 0010:drm_sched_job_done.isra.0+0x11/0x140 [gpu_sched]
  Code: 8b fe ff ff be 03 00 00 00 e8 7b da b7 e3 e9 d4 fe ff ff 66

0f 1f 44 00 00 0f 1f 44 00 00 55 48 89 e5 41 55 41 54 49 89 fc 53 <48> 8b 9f 88 00 00 00 f0 ff 8b f0 00 00 00 48 8b 83 80 01 00 00 f0
  RSP: 0018:ffffb1b1801d4d38 EFLAGS: 00010087
  RAX: ffffffffc0aa48b0 RBX: ffffb1b1801d4d70 RCX: 0000000000000018
  RDX: 000036c70afb7c1d RSI: ffff8a45ca413c60 RDI: 0000000000000000
  RBP: ffffb1b1801d4d50 R08: 00000000000000b5 R09: 0000000000000000
  R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000000
  R13: ffffb1b1801d4d70 R14: ffff8a45c4160000 R15: ffff8a45c416a708 FS:  0000000000000000(0000) GS:ffff8a48a0a80000(0000) knlGS:0000000000000000
  CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
  CR2: 0000000000000088 CR3: 000000014ad50000 CR4: 00000000003506e0

  Call Trace:
   <IRQ>
   drm_sched_job_done_cb+0x12/0x20 [gpu_sched]
   dma_fence_signal_timestamp_locked+0x7e/0x110
   dma_fence_signal+0x31/0x60
   amdgpu_fence_process+0xc4/0x140 [amdgpu]
   gfx_v9_0_eop_irq+0x9d/0xd0 [amdgpu]
   amdgpu_irq_dispatch+0xb7/0x210 [amdgpu]
   amdgpu_ih_process+0x86/0x100 [amdgpu]
   amdgpu_irq_handler+0x24/0x60 [amdgpu]
   __handle_irq_event_percpu+0x4b/0x190
   handle_irq_event_percpu+0x15/0x50
   handle_irq_event+0x39/0x60
   handle_edge_irq+0xaf/0x210
   __common_interrupt+0x6e/0x110
   common_interrupt+0xc1/0xe0
   </IRQ>
   <TASK>
Quoted 6 lines
Quoted 1 line

How is this triggered any why haven't we seen it before?

IGT has few 'amdgpu' specific testcases which is not related  to fence.

while running those test cases I have got this crash but this crash is not always reproducible.

I've encountered this crash twice (out of two attempts) with kernel built from mainline commit 4c86114194e6 when I started firefox.

The patch below fixed the crash and does not seem to introduce any apparent issue.

Quoted 24 lines
Quoted 24 lines
Quoted 23 lines

---
  drivers/gpu/drm/scheduler/sched_main.c | 7 ++++++-

  1 file changed, 6 insertions(+), 1 deletion(-)
drivers/gpu/drm/scheduler/sched_main.c +6 -1
index 6684d88463b4..390272f6b126 100644
@@ -172,7 +172,12 @@ drm_sched_rq_select_entity(struct drm_sched_rq *rq)
   static void drm_sched_job_done(struct drm_sched_job *s_job)
   {
       struct drm_sched_fence *s_fence = s_job->s_fence;
-    struct drm_gpu_scheduler *sched = s_fence->sched;
+    struct drm_gpu_scheduler *sched;
+
+    if (!s_fence)
+        return;
+
+    sched = s_fence->sched;
         atomic_dec(&sched->hw_rq_count);
       atomic_dec(sched->score);