Skip to contentSkip to conversation

drm/nouveau: Fix out-of-bounds access when deferencing MMU type

PATCH In review Thomas Zimmermann 5 participants 8 messages Nov 10, 2020 – Nov 13, 2020

Subsystems (4)
lore
Share & tools
lore

Thomas Zimmermann

author maintainer
#0 lore raw

PATCH 1/1 drm/nouveau: Fix out-of-bounds access when deferencing MMU type

The value of struct drm_device.ttm.type_vram can become -1 for unknown types of memory (see nouveau_ttm_init()). This leads to an out-of-bounds error when accessing struct nvif_mmu.type[]:

  [   18.304116] ==================================================================
  [   18.311649] BUG: KASAN: slab-out-of-bounds in nouveau_ttm_io_mem_reserve+0x17a/0x7e0 [nouveau]
  [   18.320415] Read of size 1 at addr ffff88810ffac1fe by task systemd-udevd/342
  [   18.327681]
  [   18.329208] CPU: 1 PID: 342 Comm: systemd-udevd Tainted: G            E     5.10.0-rc2-1-default+ #581
  [   18.338681] Hardware name: Dell Inc. OptiPlex 9020/0N4YC8, BIOS A24 10/24/2018
  [   18.346032] Call Trace:
  [   18.348536]  dump_stack+0xae/0xe5
  [   18.351919]  print_address_description.constprop.0+0x17/0xf0
  [   18.357787]  ? nouveau_ttm_io_mem_reserve+0x17a/0x7e0 [nouveau]
  [   18.363818]  __kasan_report.cold+0x20/0x38
  [   18.368099]  ? nouveau_ttm_io_mem_reserve+0x17a/0x7e0 [nouveau]
  [   18.374133]  kasan_report+0x3a/0x50
  [   18.377789]  nouveau_ttm_io_mem_reserve+0x17a/0x7e0 [nouveau]
  <...>
  [   18.767690] Allocated by task 342:
  [   18.773087]  kasan_save_stack+0x1b/0x40
  [   18.778890]  __kasan_kmalloc.constprop.0+0xbf/0xd0
  [   18.785646]  __kmalloc_track_caller+0x1be/0x390
  [   18.792165]  kstrdup_const+0x46/0x70
  [   18.797686]  kobject_set_name_vargs+0x2f/0xb0
  [   18.803992]  kobject_init_and_add+0x9d/0xf0
  [   18.810117]  ttm_mem_global_init+0x12c/0x210 [ttm]
  [   18.816853]  ttm_bo_global_init+0x4a/0x160 [ttm]
  [   18.823420]  ttm_bo_device_init+0x39/0x220 [ttm]
  [   18.830046]  nouveau_ttm_init+0x2c3/0x830 [nouveau]
  [   18.836929]  nouveau_drm_device_init+0x1b4/0x3f0 [nouveau]
  <...>
  [   19.105336] ==================================================================

Fix this error, by not using type_vram as an index if it's negative. Assume default values instead.

The error was seen on Nvidia G72 hardware.

Fixes 1cf65c45183a ("drm/ttm: add caching state to ttm_bus_placement")
Cc Christian König <[email protected]>
Cc Michael J. Ruhl <[email protected]>
Cc Maarten Lankhorst <[email protected]>
Cc Maxime Ripard <[email protected]>
Cc Thomas Zimmermann <[email protected]>
Cc David Airlie <[email protected]>
Cc Daniel Vetter <[email protected]>
Cc Ben Skeggs <[email protected]>
Cc Dave Airlie <[email protected]>
Cc Gerd Hoffmann <[email protected]>
Cc Alex Deucher <[email protected]>
Cc "Christian König" <[email protected]>
Cc VMware Graphics <[email protected]>
Cc Roland Scheidegger <[email protected]>
Cc Huang Rui <[email protected]>
Cc Felix Kuehling <[email protected]>
Cc Hawking Zhang <[email protected]>
Cc Jason Gunthorpe <[email protected]>
Cc Likun Gao <[email protected]>
---
 drivers/gpu/drm/nouveau/nouveau_bo.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
drivers/gpu/drm/nouveau/nouveau_bo.c +4 -1
index 8133377d865d..fe15299d417e 100644
@@ -1142,9 +1142,12 @@ nouveau_ttm_io_mem_reserve(struct ttm_bo_device *bdev, struct ttm_resource *reg)
 	struct nvkm_device *device = nvxx_device(&drm->client.device);
 	struct nouveau_mem *mem = nouveau_mem(reg);
 	struct nvif_mmu *mmu = &drm->client.mmu;
-	const u8 type = mmu->type[drm->ttm.type_vram].type;
+	u8 type = 0;
 	int ret;
 
+	if (drm->ttm.type_vram >= 0)
+		type = mmu->type[drm->ttm.type_vram].type;
+
 	mutex_lock(&drm->ttm.io_reserve_mutex);
 retry:
 	switch (reg->mem_type) {
Signature · 6 lines
2.29.2

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
Acked-by on v1 1/1
#1 lore raw

Am 10.11.20 um 14:36 schrieb Thomas Zimmermann:

Quoted 57 lines

The value of struct drm_device.ttm.type_vram can become -1 for unknown types of memory (see nouveau_ttm_init()). This leads to an out-of-bounds error when accessing struct nvif_mmu.type[]:

   [   18.304116] ==================================================================
   [   18.311649] BUG: KASAN: slab-out-of-bounds in nouveau_ttm_io_mem_reserve+0x17a/0x7e0 [nouveau]
   [   18.320415] Read of size 1 at addr ffff88810ffac1fe by task systemd-udevd/342
   [   18.327681]
   [   18.329208] CPU: 1 PID: 342 Comm: systemd-udevd Tainted: G            E     5.10.0-rc2-1-default+ #581
   [   18.338681] Hardware name: Dell Inc. OptiPlex 9020/0N4YC8, BIOS A24 10/24/2018
   [   18.346032] Call Trace:
   [   18.348536]  dump_stack+0xae/0xe5
   [   18.351919]  print_address_description.constprop.0+0x17/0xf0
   [   18.357787]  ? nouveau_ttm_io_mem_reserve+0x17a/0x7e0 [nouveau]
   [   18.363818]  __kasan_report.cold+0x20/0x38
   [   18.368099]  ? nouveau_ttm_io_mem_reserve+0x17a/0x7e0 [nouveau]
   [   18.374133]  kasan_report+0x3a/0x50
   [   18.377789]  nouveau_ttm_io_mem_reserve+0x17a/0x7e0 [nouveau]
   <...>
   [   18.767690] Allocated by task 342:
   [   18.773087]  kasan_save_stack+0x1b/0x40
   [   18.778890]  __kasan_kmalloc.constprop.0+0xbf/0xd0
   [   18.785646]  __kmalloc_track_caller+0x1be/0x390
   [   18.792165]  kstrdup_const+0x46/0x70
   [   18.797686]  kobject_set_name_vargs+0x2f/0xb0
   [   18.803992]  kobject_init_and_add+0x9d/0xf0
   [   18.810117]  ttm_mem_global_init+0x12c/0x210 [ttm]
   [   18.816853]  ttm_bo_global_init+0x4a/0x160 [ttm]
   [   18.823420]  ttm_bo_device_init+0x39/0x220 [ttm]
   [   18.830046]  nouveau_ttm_init+0x2c3/0x830 [nouveau]
   [   18.836929]  nouveau_drm_device_init+0x1b4/0x3f0 [nouveau]
   <...>
   [   19.105336] ==================================================================

Fix this error, by not using type_vram as an index if it's negative. Assume default values instead.

The error was seen on Nvidia G72 hardware.

Fixes 1cf65c45183a ("drm/ttm: add caching state to ttm_bus_placement")
Cc Christian König <[email protected]>
Cc Michael J. Ruhl <[email protected]>
Cc Maarten Lankhorst <[email protected]>
Cc Maxime Ripard <[email protected]>
Cc Thomas Zimmermann <[email protected]>
Cc David Airlie <[email protected]>
Cc Daniel Vetter <[email protected]>
Cc Ben Skeggs <[email protected]>
Cc Dave Airlie <[email protected]>
Cc Gerd Hoffmann <[email protected]>
Cc Alex Deucher <[email protected]>
Cc "Christian König" <[email protected]>
Cc VMware Graphics <[email protected]>
Cc Roland Scheidegger <[email protected]>
Cc Huang Rui <[email protected]>
Cc Felix Kuehling <[email protected]>
Quoted 8 lines
Quoted 12 lines
---
  drivers/gpu/drm/nouveau/nouveau_bo.c | 5 ++++-
  1 file changed, 4 insertions(+), 1 deletion(-)
drivers/gpu/drm/nouveau/nouveau_bo.c +0 -0
index 8133377d865d..fe15299d417e 100644
@@ -1142,9 +1142,12 @@ nouveau_ttm_io_mem_reserve(struct ttm_bo_device *bdev, struct ttm_resource *reg)
  	struct nvkm_device *device = nvxx_device(&drm->client.device);
  	struct nouveau_mem *mem = nouveau_mem(reg);
  	struct nvif_mmu *mmu = &drm->client.mmu;

drivers/gpu/drm/nouveau/nouveau_bo.c nouveau_ttm_io_mem_reserve(struct ttm_bo_device *bdev, struct ttm_resource *reg) · line 1153

Quoted 10 lines
drivers/gpu/drm/nouveau/nouveau_bo.c +4 -1
-	const u8 type = mmu->type[drm->ttm.type_vram].type;
+	u8 type = 0;
  	int ret;
 
+	if (drm->ttm.type_vram >= 0)
+		type = mmu->type[drm->ttm.type_vram].type;
+
  	mutex_lock(&drm->ttm.io_reserve_mutex);
  retry:
  	switch (reg->mem_type) {

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

#2 lore raw
In reply to Thomas Zimmermann
Quoted 21 lines

-----Original Message-----
From: Thomas Zimmermann <[email protected]>
Sent: Tuesday, November 10, 2020 8:37 AM
To: [email protected]; [email protected]; [email protected]; Ruhl, Michael J <[email protected]>; [email protected]
Cc: [email protected]; [email protected]; Thomas Zimmermann <[email protected]>; Maarten Lankhorst <[email protected]>; Maxime Ripard <[email protected]>; Dave Airlie <[email protected]>; Gerd Hoffmann <[email protected]>; Alex Deucher <[email protected]>; VMware Graphics <[email protected]>; Roland Scheidegger <[email protected]>; Huang Rui <[email protected]>; Felix Kuehling <[email protected]>; Hawking Zhang <[email protected]>; Jason Gunthorpe <[email protected]>; Likun Gao <[email protected]>; [email protected]; spice- [email protected]; [email protected]
Subject: [PATCH] drm/nouveau: Fix out-of-bounds access when deferencing MMU type

Quoted 3 lines

The value of struct drm_device.ttm.type_vram can become -1 for unknown types of memory (see nouveau_ttm_init()). This leads to an out-of-bounds error when accessing struct nvif_mmu.type[]:

Would this make more sense to just set the type_vram = 0 instead of -1?

Mike

Quoted 93 lines
 [   18.304116]
===========================================================
=======
 [   18.311649] BUG: KASAN: slab-out-of-bounds in
nouveau_ttm_io_mem_reserve+0x17a/0x7e0 [nouveau]
 [   18.320415] Read of size 1 at addr ffff88810ffac1fe by task systemd-
udevd/342
 [   18.327681]
 [   18.329208] CPU: 1 PID: 342 Comm: systemd-udevd Tainted: G            E
5.10.0-rc2-1-default+ #581
 [   18.338681] Hardware name: Dell Inc. OptiPlex 9020/0N4YC8, BIOS A24
10/24/2018
 [   18.346032] Call Trace:
 [   18.348536]  dump_stack+0xae/0xe5
 [   18.351919]  print_address_description.constprop.0+0x17/0xf0
 [   18.357787]  ? nouveau_ttm_io_mem_reserve+0x17a/0x7e0 [nouveau]
 [   18.363818]  __kasan_report.cold+0x20/0x38
 [   18.368099]  ? nouveau_ttm_io_mem_reserve+0x17a/0x7e0 [nouveau]
 [   18.374133]  kasan_report+0x3a/0x50
 [   18.377789]  nouveau_ttm_io_mem_reserve+0x17a/0x7e0 [nouveau]
 <...>
 [   18.767690] Allocated by task 342:
 [   18.773087]  kasan_save_stack+0x1b/0x40
 [   18.778890]  __kasan_kmalloc.constprop.0+0xbf/0xd0
 [   18.785646]  __kmalloc_track_caller+0x1be/0x390
 [   18.792165]  kstrdup_const+0x46/0x70
 [   18.797686]  kobject_set_name_vargs+0x2f/0xb0
 [   18.803992]  kobject_init_and_add+0x9d/0xf0
 [   18.810117]  ttm_mem_global_init+0x12c/0x210 [ttm]
 [   18.816853]  ttm_bo_global_init+0x4a/0x160 [ttm]
 [   18.823420]  ttm_bo_device_init+0x39/0x220 [ttm]
 [   18.830046]  nouveau_ttm_init+0x2c3/0x830 [nouveau]
 [   18.836929]  nouveau_drm_device_init+0x1b4/0x3f0 [nouveau]
 <...>
 [   19.105336]

===========================================================
=======

Fix this error, by not using type_vram as an index if it's negative. Assume default values instead.

The error was seen on Nvidia G72 hardware.

Fixes 1cf65c45183a ("drm/ttm: add caching state to ttm_bus_placement")
Cc Christian König <[email protected]>
Cc Michael J. Ruhl <[email protected]>
Cc Maarten Lankhorst <[email protected]>
Cc Maxime Ripard <[email protected]>
Cc Thomas Zimmermann <[email protected]>
Cc David Airlie <[email protected]>
Cc Daniel Vetter <[email protected]>
Cc Ben Skeggs <[email protected]>
Cc Dave Airlie <[email protected]>
Cc Gerd Hoffmann <[email protected]>
Cc Alex Deucher <[email protected]>
Cc "Christian König" <[email protected]>
Cc VMware Graphics <[email protected]>
Cc Roland Scheidegger <[email protected]>
Cc Huang Rui <[email protected]>
Cc Felix Kuehling <[email protected]>
Cc Hawking Zhang <[email protected]>
Cc Jason Gunthorpe <[email protected]>
Cc Likun Gao <[email protected]>

---
drivers/gpu/drm/nouveau/nouveau_bo.c | 5 ++++-

1 file changed, 4 insertions(+), 1 deletion(-)
drivers/gpu/drm/nouveau/nouveau_bo.c +4 -1
index 8133377d865d..fe15299d417e 100644
@@ -1142,9 +1142,12 @@ nouveau_ttm_io_mem_reserve(struct
 ttm_bo_device *bdev, struct ttm_resource *reg)
 	struct nvkm_device *device = nvxx_device(&drm->client.device);
 	struct nouveau_mem *mem = nouveau_mem(reg);
 	struct nvif_mmu *mmu = &drm->client.mmu;
-	const u8 type = mmu->type[drm->ttm.type_vram].type;
+	u8 type = 0;
 	int ret;
 
+	if (drm->ttm.type_vram >= 0)
+		type = mmu->type[drm->ttm.type_vram].type;
+
 	mutex_lock(&drm->ttm.io_reserve_mutex);

retry:

Quoted 3 lines
	switch (reg->mem_type) {

--
2.29.2

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

author maintainer
#3 lore raw

Hi

Am 10.11.20 um 16:27 schrieb Ruhl, Michael J:

Quoted 21 lines
Quoted 21 lines

-----Original Message-----
From: Thomas Zimmermann <[email protected]>
Sent: Tuesday, November 10, 2020 8:37 AM
To: [email protected]; [email protected]; [email protected]; Ruhl, Michael J <[email protected]>; [email protected]
Cc: [email protected]; [email protected]; Thomas Zimmermann <[email protected]>; Maarten Lankhorst <[email protected]>; Maxime Ripard <[email protected]>; Dave Airlie <[email protected]>; Gerd Hoffmann <[email protected]>; Alex Deucher <[email protected]>; VMware Graphics <[email protected]>; Roland Scheidegger <[email protected]>; Huang Rui <[email protected]>; Felix Kuehling <[email protected]>; Hawking Zhang <[email protected]>; Jason Gunthorpe <[email protected]>; Likun Gao <[email protected]>; [email protected]; spice- [email protected]; [email protected]
Subject: [PATCH] drm/nouveau: Fix out-of-bounds access when deferencing MMU type

Quoted 5 lines
Quoted 3 lines

The value of struct drm_device.ttm.type_vram can become -1 for unknown types of memory (see nouveau_ttm_init()). This leads to an out-of-bounds error when accessing struct nvif_mmu.type[]:

Would this make more sense to just set the type_vram = 0 instead of -1?

From what I understand, these indices refer to an internal type of MMU, rsp the MMU's capabilities. However, my hardware (pre-NV50) does not have an MMU at all.

I agree that it would be nice to have a cleaner design that incorporates this case, but resolving that would apparently require more than a bugfix.

Best regards
Thomas

Quoted 100 lines

Mike

Quoted 96 lines
 [   18.304116]
===========================================================
=======
 [   18.311649] BUG: KASAN: slab-out-of-bounds in
nouveau_ttm_io_mem_reserve+0x17a/0x7e0 [nouveau]
 [   18.320415] Read of size 1 at addr ffff88810ffac1fe by task systemd-
udevd/342
 [   18.327681]
 [   18.329208] CPU: 1 PID: 342 Comm: systemd-udevd Tainted: G            E
5.10.0-rc2-1-default+ #581
 [   18.338681] Hardware name: Dell Inc. OptiPlex 9020/0N4YC8, BIOS A24
10/24/2018
 [   18.346032] Call Trace:
 [   18.348536]  dump_stack+0xae/0xe5
 [   18.351919]  print_address_description.constprop.0+0x17/0xf0
 [   18.357787]  ? nouveau_ttm_io_mem_reserve+0x17a/0x7e0 [nouveau]
 [   18.363818]  __kasan_report.cold+0x20/0x38
 [   18.368099]  ? nouveau_ttm_io_mem_reserve+0x17a/0x7e0 [nouveau]
 [   18.374133]  kasan_report+0x3a/0x50
 [   18.377789]  nouveau_ttm_io_mem_reserve+0x17a/0x7e0 [nouveau]
 <...>
 [   18.767690] Allocated by task 342:
 [   18.773087]  kasan_save_stack+0x1b/0x40
 [   18.778890]  __kasan_kmalloc.constprop.0+0xbf/0xd0
 [   18.785646]  __kmalloc_track_caller+0x1be/0x390
 [   18.792165]  kstrdup_const+0x46/0x70
 [   18.797686]  kobject_set_name_vargs+0x2f/0xb0
 [   18.803992]  kobject_init_and_add+0x9d/0xf0
 [   18.810117]  ttm_mem_global_init+0x12c/0x210 [ttm]
 [   18.816853]  ttm_bo_global_init+0x4a/0x160 [ttm]
 [   18.823420]  ttm_bo_device_init+0x39/0x220 [ttm]
 [   18.830046]  nouveau_ttm_init+0x2c3/0x830 [nouveau]
 [   18.836929]  nouveau_drm_device_init+0x1b4/0x3f0 [nouveau]
 <...>
 [   19.105336]

===========================================================
=======

Fix this error, by not using type_vram as an index if it's negative. Assume default values instead.

The error was seen on Nvidia G72 hardware.

Fixes 1cf65c45183a ("drm/ttm: add caching state to ttm_bus_placement")
Cc Christian König <[email protected]>
Cc Michael J. Ruhl <[email protected]>
Cc Maarten Lankhorst <[email protected]>
Cc Maxime Ripard <[email protected]>
Cc Thomas Zimmermann <[email protected]>
Cc David Airlie <[email protected]>
Cc Daniel Vetter <[email protected]>
Cc Ben Skeggs <[email protected]>
Cc Dave Airlie <[email protected]>
Cc Gerd Hoffmann <[email protected]>
Cc Alex Deucher <[email protected]>
Cc "Christian König" <[email protected]>
Cc VMware Graphics <[email protected]>
Cc Roland Scheidegger <[email protected]>
Cc Huang Rui <[email protected]>
Cc Felix Kuehling <[email protected]>
Cc Hawking Zhang <[email protected]>
Cc Jason Gunthorpe <[email protected]>
Cc Likun Gao <[email protected]>

---
drivers/gpu/drm/nouveau/nouveau_bo.c | 5 ++++-

1 file changed, 4 insertions(+), 1 deletion(-)
drivers/gpu/drm/nouveau/nouveau_bo.c +4 -1
index 8133377d865d..fe15299d417e 100644
@@ -1142,9 +1142,12 @@ nouveau_ttm_io_mem_reserve(struct
 ttm_bo_device *bdev, struct ttm_resource *reg)
 	struct nvkm_device *device = nvxx_device(&drm->client.device);
 	struct nouveau_mem *mem = nouveau_mem(reg);
 	struct nvif_mmu *mmu = &drm->client.mmu;
-	const u8 type = mmu->type[drm->ttm.type_vram].type;
+	u8 type = 0;
 	int ret;
 
+	if (drm->ttm.type_vram >= 0)
+		type = mmu->type[drm->ttm.type_vram].type;
+
 	mutex_lock(&drm->ttm.io_reserve_mutex);

retry:

	switch (reg->mem_type) {

--
2.29.2

Signature · 10 lines
Thomas Zimmermann
Graphics Driver Developer
SUSE Software Solutions Germany GmbH
Maxfeldstr. 5, 90409 Nürnberg, Germany
(HRB 36809, AG Nürnberg)
Geschäftsführer: Felix Imendörffer
_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
Reviewed-by
#4 lore raw
Quoted 47 lines

-----Original Message-----
From: Thomas Zimmermann <[email protected]>
Sent: Wednesday, November 11, 2020 7:08 AM
To: Ruhl, Michael J <[email protected]>; [email protected]; [email protected]; [email protected]; [email protected]
Cc: [email protected]; [email protected]; Maarten Lankhorst <[email protected]>; Maxime Ripard <[email protected]>; Dave Airlie <[email protected]>; Gerd Hoffmann <[email protected]>; Alex Deucher <[email protected]>; VMware Graphics <[email protected]>; Roland Scheidegger <[email protected]>; Huang Rui <[email protected]>; Felix Kuehling <[email protected]>; Hawking Zhang <[email protected]>; Jason Gunthorpe <[email protected]>; Likun Gao <[email protected]>; [email protected]; spice- [email protected]; [email protected]
Subject: Re: [PATCH] drm/nouveau: Fix out-of-bounds access when deferencing MMU type

Hi

Am 10.11.20 um 16:27 schrieb Ruhl, Michael J:

Quoted 8 lines
Quoted 8 lines

-----Original Message-----
From: Thomas Zimmermann <[email protected]>
Sent: Tuesday, November 10, 2020 8:37 AM
To: [email protected]; [email protected]; [email protected]; Ruhl, Michael J <[email protected]>; [email protected]
Cc: [email protected]; [email protected];

Thomas

Quoted 8 lines
Quoted 8 lines

Zimmermann <[email protected]>; Maarten Lankhorst <[email protected]>; Maxime Ripard <[email protected]>; Dave Airlie <[email protected]>; Gerd Hoffmann <[email protected]>; Alex Deucher <[email protected]>; VMware Graphics <[email protected]>; Roland Scheidegger <[email protected]>; Huang Rui <[email protected]>; Felix Kuehling <[email protected]>; Hawking Zhang <[email protected]>; Jason Gunthorpe <[email protected]>; Likun

Gao

Quoted 3 lines
Quoted 3 lines

<[email protected]>; [email protected]; spice- [email protected]; [email protected]
Subject: [PATCH] drm/nouveau: Fix out-of-bounds access when

deferencing

Quoted 3 lines
Quoted 3 lines

MMU type

The value of struct drm_device.ttm.type_vram can become -1 for

unknown

Quoted 8 lines
Quoted 4 lines
Quoted 2 lines

types of memory (see nouveau_ttm_init()). This leads to an out-of-bounds error when accessing struct nvif_mmu.type[]:

Would this make more sense to just set the type_vram = 0 instead of -1?

From what I understand, these indices refer to an internal type of MMU, rsp the MMU's capabilities. However, my hardware (pre-NV50) does not have an MMU at all.

Yeah, and upon further review I see that my comment was completely wrong (value vs. index).

A better suggestion would have been, create an entry in the array that means, "unsupported type" with a value of 0, but...

Quoted 2 lines

I agree that it would be nice to have a cleaner design that incorporates this case, but resolving that would apparently require more than a bugfix.

I agree. The -1 index is a special case for the platform path (platform != NV_DEVICE_INFO_V0_SOC). This is a fix for the issue, but not a complete solution.

If you need it:

Thanks,
Mike

Quoted 106 lines

Best regards
Thomas

Quoted 6 lines

Mike

Quoted 3 lines
 [   18.304116]

===========================================================

Quoted 26 lines
Quoted 26 lines

=======

 [   18.311649] BUG: KASAN: slab-out-of-bounds in
nouveau_ttm_io_mem_reserve+0x17a/0x7e0 [nouveau]
 [   18.320415] Read of size 1 at addr ffff88810ffac1fe by task systemd-
udevd/342
 [   18.327681]
 [   18.329208] CPU: 1 PID: 342 Comm: systemd-udevd Tainted: G            E
5.10.0-rc2-1-default+ #581
 [   18.338681] Hardware name: Dell Inc. OptiPlex 9020/0N4YC8, BIOS A24
10/24/2018
 [   18.346032] Call Trace:
 [   18.348536]  dump_stack+0xae/0xe5
 [   18.351919]  print_address_description.constprop.0+0x17/0xf0
 [   18.357787]  ? nouveau_ttm_io_mem_reserve+0x17a/0x7e0 [nouveau]
 [   18.363818]  __kasan_report.cold+0x20/0x38
 [   18.368099]  ? nouveau_ttm_io_mem_reserve+0x17a/0x7e0 [nouveau]
 [   18.374133]  kasan_report+0x3a/0x50
 [   18.377789]  nouveau_ttm_io_mem_reserve+0x17a/0x7e0 [nouveau]
 <...>
 [   18.767690] Allocated by task 342:
 [   18.773087]  kasan_save_stack+0x1b/0x40
 [   18.778890]  __kasan_kmalloc.constprop.0+0xbf/0xd0
 [   18.785646]  __kmalloc_track_caller+0x1be/0x390
 [   18.792165]  kstrdup_const+0x46/0x70
 [   18.797686]  kobject_set_name_vargs+0x2f/0xb0
 [   18.803992]  kobject_init_and_add+0x9d/0xf0
Quoted 8 lines
Quoted 8 lines
 [   18.810117]  ttm_mem_global_init+0x12c/0x210 [ttm]
 [   18.816853]  ttm_bo_global_init+0x4a/0x160 [ttm]
 [   18.823420]  ttm_bo_device_init+0x39/0x220 [ttm]
 [   18.830046]  nouveau_ttm_init+0x2c3/0x830 [nouveau]
 [   18.836929]  nouveau_drm_device_init+0x1b4/0x3f0 [nouveau]
 <...>
 [   19.105336]

===========================================================

Quoted 51 lines
Quoted 51 lines

=======

Fix this error, by not using type_vram as an index if it's negative. Assume default values instead.

The error was seen on Nvidia G72 hardware.

Fixes 1cf65c45183a ("drm/ttm: add caching state to ttm_bus_placement")
Cc Christian König <[email protected]>
Cc Michael J. Ruhl <[email protected]>
Cc Maarten Lankhorst <[email protected]>
Cc Maxime Ripard <[email protected]>
Cc Thomas Zimmermann <[email protected]>
Cc David Airlie <[email protected]>
Cc Daniel Vetter <[email protected]>
Cc Ben Skeggs <[email protected]>
Cc Dave Airlie <[email protected]>
Cc Gerd Hoffmann <[email protected]>
Cc Alex Deucher <[email protected]>
Cc "Christian König" <[email protected]>
Cc VMware Graphics <[email protected]>
Cc Roland Scheidegger <[email protected]>
Cc Huang Rui <[email protected]>
Cc Felix Kuehling <[email protected]>
Cc Hawking Zhang <[email protected]>
Cc Jason Gunthorpe <[email protected]>
Cc Likun Gao <[email protected]>

---
drivers/gpu/drm/nouveau/nouveau_bo.c | 5 ++++-

1 file changed, 4 insertions(+), 1 deletion(-)
drivers/gpu/drm/nouveau/nouveau_bo.c +1 -1
index 8133377d865d..fe15299d417e 100644
@@ -1142,9 +1142,12 @@ nouveau_ttm_io_mem_reserve(struct
 ttm_bo_device *bdev, struct ttm_resource *reg)
 	struct nvkm_device *device = nvxx_device(&drm->client.device);
 	struct nouveau_mem *mem = nouveau_mem(reg);
 	struct nvif_mmu *mmu = &drm->client.mmu;
-	const u8 type = mmu->type[drm->ttm.type_vram].type;
+	u8 type = 0;
 	int ret;
 
Quoted 9 lines
Quoted 8 lines
drivers/gpu/drm/nouveau/nouveau_bo.c +3 -0
+	if (drm->ttm.type_vram >= 0)
+		type = mmu->type[drm->ttm.type_vram].type;
+
 	mutex_lock(&drm->ttm.io_reserve_mutex);

retry:

	switch (reg->mem_type) {

--
2.29.2

Quoted 7 lines

--
Thomas Zimmermann
Graphics Driver Developer
SUSE Software Solutions Germany GmbH
Maxfeldstr. 5, 90409 Nürnberg, Germany
(HRB 36809, AG Nürnberg)
Geschäftsführer: Felix Imendörffer

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

#5 lore raw
Quoted 66 lines · Michael J
Quoted 48 lines

-----Original Message-----
From: Thomas Zimmermann <[email protected]>
Sent: Wednesday, November 11, 2020 7:08 AM
To: Ruhl, Michael J <[email protected]>; [email protected]; [email protected]; [email protected]; [email protected]
Cc: [email protected]; [email protected]; Maarten Lankhorst <[email protected]>; Maxime Ripard <[email protected]>; Dave Airlie <[email protected]>; Gerd Hoffmann <[email protected]>; Alex Deucher <[email protected]>; VMware Graphics <[email protected]>; Roland Scheidegger <[email protected]>; Huang Rui <[email protected]>; Felix Kuehling <[email protected]>; Hawking Zhang <[email protected]>; Jason Gunthorpe <[email protected]>; Likun Gao <[email protected]>; [email protected]; spice- [email protected]; [email protected]
Subject: Re: [PATCH] drm/nouveau: Fix out-of-bounds access when deferencing MMU type

Hi

Am 10.11.20 um 16:27 schrieb Ruhl, Michael J:

Quoted 8 lines
Quoted 8 lines

-----Original Message-----
From: Thomas Zimmermann <[email protected]>
Sent: Tuesday, November 10, 2020 8:37 AM
To: [email protected]; [email protected]; [email protected]; Ruhl, Michael J <[email protected]>; [email protected]
Cc: [email protected]; [email protected];

Thomas

Quoted 8 lines
Quoted 8 lines

Zimmermann <[email protected]>; Maarten Lankhorst <[email protected]>; Maxime Ripard <[email protected]>; Dave Airlie <[email protected]>; Gerd Hoffmann <[email protected]>; Alex Deucher <[email protected]>; VMware Graphics <[email protected]>; Roland Scheidegger <[email protected]>; Huang Rui <[email protected]>; Felix Kuehling <[email protected]>; Hawking Zhang <[email protected]>; Jason Gunthorpe <[email protected]>; Likun

Gao

Quoted 3 lines
Quoted 3 lines

<[email protected]>; [email protected]; spice- [email protected]; [email protected]
Subject: [PATCH] drm/nouveau: Fix out-of-bounds access when

deferencing

Quoted 3 lines
Quoted 3 lines

MMU type

The value of struct drm_device.ttm.type_vram can become -1 for

unknown

Quoted 8 lines
Quoted 4 lines
Quoted 2 lines

types of memory (see nouveau_ttm_init()). This leads to an out-of-bounds error when accessing struct nvif_mmu.type[]:

Would this make more sense to just set the type_vram = 0 instead of -1?

From what I understand, these indices refer to an internal type of MMU, rsp the MMU's capabilities. However, my hardware (pre-NV50) does not have an MMU at all.

Yeah, and upon further review I see that my comment was completely wrong (value vs. index).

A better suggestion would have been, create an entry in the array that means, "unsupported type" with a value of 0, but...

Quoted 2 lines

I agree that it would be nice to have a cleaner design that incorporates this case, but resolving that would apparently require more than a bugfix.

Quoted 6 lines

I agree. The -1 index is a special case for the platform path (platform != NV_DEVICE_INFO_V0_SOC). This is a fix for the issue, but not a complete solution.

If you need it:

I've put an alternate fix for this here[1], and will get it into drm-fixes later today.

Ben.

[1] https://github.com/skeggsb/nouveau/commit/4590f7120c2f1f4aea9d8b93a2dae43b312d35ad

Quoted 117 lines

Thanks,
Mike

Quoted 106 lines

Best regards
Thomas

Quoted 6 lines

Mike

Quoted 3 lines
 [   18.304116]

===========================================================

Quoted 26 lines
Quoted 26 lines

=======

 [   18.311649] BUG: KASAN: slab-out-of-bounds in
nouveau_ttm_io_mem_reserve+0x17a/0x7e0 [nouveau]
 [   18.320415] Read of size 1 at addr ffff88810ffac1fe by task systemd-
udevd/342
 [   18.327681]
 [   18.329208] CPU: 1 PID: 342 Comm: systemd-udevd Tainted: G            E
5.10.0-rc2-1-default+ #581
 [   18.338681] Hardware name: Dell Inc. OptiPlex 9020/0N4YC8, BIOS A24
10/24/2018
 [   18.346032] Call Trace:
 [   18.348536]  dump_stack+0xae/0xe5
 [   18.351919]  print_address_description.constprop.0+0x17/0xf0
 [   18.357787]  ? nouveau_ttm_io_mem_reserve+0x17a/0x7e0 [nouveau]
 [   18.363818]  __kasan_report.cold+0x20/0x38
 [   18.368099]  ? nouveau_ttm_io_mem_reserve+0x17a/0x7e0 [nouveau]
 [   18.374133]  kasan_report+0x3a/0x50
 [   18.377789]  nouveau_ttm_io_mem_reserve+0x17a/0x7e0 [nouveau]
 <...>
 [   18.767690] Allocated by task 342:
 [   18.773087]  kasan_save_stack+0x1b/0x40
 [   18.778890]  __kasan_kmalloc.constprop.0+0xbf/0xd0
 [   18.785646]  __kmalloc_track_caller+0x1be/0x390
 [   18.792165]  kstrdup_const+0x46/0x70
 [   18.797686]  kobject_set_name_vargs+0x2f/0xb0
 [   18.803992]  kobject_init_and_add+0x9d/0xf0
Quoted 8 lines
Quoted 8 lines
 [   18.810117]  ttm_mem_global_init+0x12c/0x210 [ttm]
 [   18.816853]  ttm_bo_global_init+0x4a/0x160 [ttm]
 [   18.823420]  ttm_bo_device_init+0x39/0x220 [ttm]
 [   18.830046]  nouveau_ttm_init+0x2c3/0x830 [nouveau]
 [   18.836929]  nouveau_drm_device_init+0x1b4/0x3f0 [nouveau]
 <...>
 [   19.105336]

===========================================================

Quoted 51 lines
Quoted 51 lines

=======

Fix this error, by not using type_vram as an index if it's negative. Assume default values instead.

The error was seen on Nvidia G72 hardware.

Fixes 1cf65c45183a ("drm/ttm: add caching state to ttm_bus_placement")
Cc Christian König <[email protected]>
Cc Michael J. Ruhl <[email protected]>
Cc Maarten Lankhorst <[email protected]>
Cc Maxime Ripard <[email protected]>
Cc Thomas Zimmermann <[email protected]>
Cc David Airlie <[email protected]>
Cc Daniel Vetter <[email protected]>
Cc Ben Skeggs <[email protected]>
Cc Dave Airlie <[email protected]>
Cc Gerd Hoffmann <[email protected]>
Cc Alex Deucher <[email protected]>
Cc "Christian König" <[email protected]>
Cc VMware Graphics <[email protected]>
Cc Roland Scheidegger <[email protected]>
Cc Huang Rui <[email protected]>
Cc Felix Kuehling <[email protected]>
Cc Hawking Zhang <[email protected]>
Cc Jason Gunthorpe <[email protected]>
Cc Likun Gao <[email protected]>

---
drivers/gpu/drm/nouveau/nouveau_bo.c | 5 ++++-

1 file changed, 4 insertions(+), 1 deletion(-)
drivers/gpu/drm/nouveau/nouveau_bo.c +1 -1
index 8133377d865d..fe15299d417e 100644
@@ -1142,9 +1142,12 @@ nouveau_ttm_io_mem_reserve(struct
 ttm_bo_device *bdev, struct ttm_resource *reg)
    struct nvkm_device *device = nvxx_device(&drm->client.device);
    struct nouveau_mem *mem = nouveau_mem(reg);
    struct nvif_mmu *mmu = &drm->client.mmu;
-   const u8 type = mmu->type[drm->ttm.type_vram].type;
+   u8 type = 0;
    int ret;
 
Quoted 9 lines
Quoted 8 lines
drivers/gpu/drm/nouveau/nouveau_bo.c +3 -0
+   if (drm->ttm.type_vram >= 0)
+           type = mmu->type[drm->ttm.type_vram].type;
+
    mutex_lock(&drm->ttm.io_reserve_mutex);

retry:

    switch (reg->mem_type) {

--
2.29.2

Quoted 7 lines

--
Thomas Zimmermann
Graphics Driver Developer
SUSE Software Solutions Germany GmbH
Maxfeldstr. 5, 90409 Nürnberg, Germany
(HRB 36809, AG Nürnberg)
Geschäftsführer: Felix Imendörffer

Quoted 4 lines

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

Reviewed-by
#6 lore raw
Quoted 100 lines

-----Original Message-----
From: Ben Skeggs <[email protected]>
Sent: Wednesday, November 11, 2020 9:39 PM
To: Ruhl, Michael J <[email protected]>
Cc: Thomas Zimmermann <[email protected]>; [email protected]; [email protected]; [email protected]; [email protected]; amd- [email protected]; [email protected]; dri- [email protected]; [email protected]; Roland Scheidegger <[email protected]>; Jason Gunthorpe <[email protected]>; Huang Rui <[email protected]>; VMware Graphics <linux-graphics- [email protected]>; Gerd Hoffmann <[email protected]>; spice- [email protected]; Alex Deucher <[email protected]>; Dave Airlie <[email protected]>; Likun Gao <[email protected]>; Felix Kuehling <[email protected]>; Hawking Zhang <[email protected]>
Subject: Re: [PATCH] drm/nouveau: Fix out-of-bounds access when deferencing MMU type

Quoted 14 lines · Michael J
Quoted 14 lines

-----Original Message-----
From: Thomas Zimmermann <[email protected]>
Sent: Wednesday, November 11, 2020 7:08 AM
To: Ruhl, Michael J <[email protected]>; [email protected]; [email protected]; [email protected]; [email protected]
Cc: [email protected]; [email protected]; Maarten Lankhorst <[email protected]>; Maxime Ripard <[email protected]>; Dave Airlie <[email protected]>; Gerd Hoffmann <[email protected]>; Alex Deucher <[email protected]>; VMware Graphics <[email protected]>; Roland Scheidegger <[email protected]>; Huang Rui <[email protected]>; Felix Kuehling <[email protected]>; Hawking Zhang <[email protected]>; Jason Gunthorpe <[email protected]>; Likun

Gao

Quoted 7 lines
Quoted 7 lines

<[email protected]>; [email protected]; spice- [email protected]; [email protected]
Subject: Re: [PATCH] drm/nouveau: Fix out-of-bounds access when deferencing MMU type

Hi

Quoted 7 lines
Quoted 7 lines

Am 10.11.20 um 16:27 schrieb Ruhl, Michael J:

Quoted 6 lines
Quoted 6 lines

-----Original Message-----
From: Thomas Zimmermann <[email protected]>
Sent: Tuesday, November 10, 2020 8:37 AM
To: [email protected]; [email protected]; [email protected]; Ruhl,

Michael J

Quoted 6 lines
Quoted 6 lines
Quoted 2 lines
Quoted 2 lines

<[email protected]>; [email protected]
Cc: [email protected]; [email protected];

Thomas

Quoted 3 lines
Quoted 3 lines

Zimmermann <[email protected]>; Maarten Lankhorst <[email protected]>; Maxime Ripard <[email protected]>; Dave Airlie <[email protected]>; Gerd

Hoffmann

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

<[email protected]>; Alex Deucher <[email protected]>; VMware Graphics <[email protected]>; Roland Scheidegger <[email protected]>; Huang Rui

<[email protected]>;

Quoted 4 lines
Quoted 4 lines
Quoted 2 lines
Quoted 2 lines

Felix Kuehling <[email protected]>; Hawking Zhang <[email protected]>; Jason Gunthorpe <[email protected]>; Likun

Gao

Quoted 1 line
Quoted 1 line

<[email protected]>; [email protected];

spice-

Quoted 8 lines
Quoted 8 lines
Quoted 2 lines
Quoted 2 lines

[email protected]; [email protected]
Subject: [PATCH] drm/nouveau: Fix out-of-bounds access when

deferencing

Quoted 3 lines
Quoted 3 lines

MMU type

The value of struct drm_device.ttm.type_vram can become -1 for

unknown

Quoted 1 line
Quoted 1 line

types of memory (see nouveau_ttm_init()). This leads to an out-of-

bounds

Quoted 9 lines
Quoted 7 lines
Quoted 3 lines
Quoted 1 line

error when accessing struct nvif_mmu.type[]:

Would this make more sense to just set the type_vram = 0 instead of -1?

From what I understand, these indices refer to an internal type of MMU, rsp the MMU's capabilities. However, my hardware (pre-NV50) does not have an MMU at all.

Yeah, and upon further review I see that my comment was completely

wrong

Quoted 3 lines

(value vs. index).

A better suggestion would have been, create an entry in the array that

means,

Quoted 11 lines

"unsupported type" with a value of 0, but...

Quoted 2 lines

I agree that it would be nice to have a cleaner design that incorporates this case, but resolving that would apparently require more than a bugfix.

I agree. The -1 index is a special case for the platform path (platform != NV_DEVICE_INFO_V0_SOC). This is a fix for the issue, but not a complete solution.

If you need it:

Quoted 8 lines

I've put an alternate fix for this here[1], and will get it into drm-fixes later today.

Ben.

[1]
https://github.com/skeggsb/nouveau/commit/4590f7120c2f1f4aea9d8b93a2d
ae43b312d35ad

This makes a lot of sense. I spent some time trying to reconcile the platform info
that was not being used in this case, but didn't see the solution like this. This is
pretty clean.

If you would like:

For this solution as well.

Mike

Quoted 57 lines
Quoted 8 lines

Thanks,
Mike

Quoted 4 lines

Best regards
Thomas

Quoted 1 line
Quoted 7 lines
Quoted 5 lines
Quoted 5 lines

Mike

Quoted 3 lines
 [   18.304116]

==========================================================

=

Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines

=======

 [   18.311649] BUG: KASAN: slab-out-of-bounds in
nouveau_ttm_io_mem_reserve+0x17a/0x7e0 [nouveau]
 [   18.320415] Read of size 1 at addr ffff88810ffac1fe by task systemd-
udevd/342
 [   18.327681]
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
 [   18.329208] CPU: 1 PID: 342 Comm: systemd-udevd Tainted: G            E
5.10.0-rc2-1-default+ #581
 [   18.338681] Hardware name: Dell Inc. OptiPlex 9020/0N4YC8, BIOS A24
10/24/2018
 [   18.346032] Call Trace:
 [   18.348536]  dump_stack+0xae/0xe5
 [   18.351919]  print_address_description.constprop.0+0x17/0xf0
 [   18.357787]  ? nouveau_ttm_io_mem_reserve+0x17a/0x7e0

[nouveau]

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
 [   18.363818]  __kasan_report.cold+0x20/0x38
 [   18.368099]  ? nouveau_ttm_io_mem_reserve+0x17a/0x7e0

[nouveau]

Quoted 19 lines
Quoted 10 lines
Quoted 10 lines
Quoted 10 lines
 [   18.374133]  kasan_report+0x3a/0x50
 [   18.377789]  nouveau_ttm_io_mem_reserve+0x17a/0x7e0 [nouveau]
 <...>
 [   18.767690] Allocated by task 342:
 [   18.773087]  kasan_save_stack+0x1b/0x40
 [   18.778890]  __kasan_kmalloc.constprop.0+0xbf/0xd0
 [   18.785646]  __kmalloc_track_caller+0x1be/0x390
 [   18.792165]  kstrdup_const+0x46/0x70
 [   18.797686]  kobject_set_name_vargs+0x2f/0xb0
 [   18.803992]  kobject_init_and_add+0x9d/0xf0
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
 [   18.810117]  ttm_mem_global_init+0x12c/0x210 [ttm]
 [   18.816853]  ttm_bo_global_init+0x4a/0x160 [ttm]
 [   18.823420]  ttm_bo_device_init+0x39/0x220 [ttm]
 [   18.830046]  nouveau_ttm_init+0x2c3/0x830 [nouveau]
 [   18.836929]  nouveau_drm_device_init+0x1b4/0x3f0 [nouveau]
 <...>
 [   19.105336]
Quoted 1 line

==========================================================

=

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

=======

Quoted 71 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines
Quoted 7 lines

Fix this error, by not using type_vram as an index if it's negative. Assume default values instead.

The error was seen on Nvidia G72 hardware.

Fixes 1cf65c45183a ("drm/ttm: add caching state to

ttm_bus_placement")

Quoted 63 lines
Quoted 52 lines
Quoted 42 lines
Quoted 42 lines
Cc Christian König <[email protected]>
Cc Michael J. Ruhl <[email protected]>
Cc Maarten Lankhorst <[email protected]>
Cc Maxime Ripard <[email protected]>
Cc Thomas Zimmermann <[email protected]>
Cc David Airlie <[email protected]>
Cc Daniel Vetter <[email protected]>
Cc Ben Skeggs <[email protected]>
Cc Dave Airlie <[email protected]>
Cc Gerd Hoffmann <[email protected]>
Cc Alex Deucher <[email protected]>
Cc "Christian König" <[email protected]>
Cc VMware Graphics <[email protected]>
Cc Roland Scheidegger <[email protected]>
Cc Huang Rui <[email protected]>
Cc Felix Kuehling <[email protected]>
Cc Hawking Zhang <[email protected]>
Cc Jason Gunthorpe <[email protected]>
Cc Likun Gao <[email protected]>

---
drivers/gpu/drm/nouveau/nouveau_bo.c | 5 ++++-

1 file changed, 4 insertions(+), 1 deletion(-)
drivers/gpu/drm/nouveau/nouveau_bo.c +1 -1
index 8133377d865d..fe15299d417e 100644
@@ -1142,9 +1142,12 @@ nouveau_ttm_io_mem_reserve(struct
 ttm_bo_device *bdev, struct ttm_resource *reg)
    struct nvkm_device *device = nvxx_device(&drm->client.device);
    struct nouveau_mem *mem = nouveau_mem(reg);
    struct nvif_mmu *mmu = &drm->client.mmu;
-   const u8 type = mmu->type[drm->ttm.type_vram].type;
+   u8 type = 0;
    int ret;
 
Quoted 9 lines
Quoted 8 lines
drivers/gpu/drm/nouveau/nouveau_bo.c +3 -0
+   if (drm->ttm.type_vram >= 0)
+           type = mmu->type[drm->ttm.type_vram].type;
+
    mutex_lock(&drm->ttm.io_reserve_mutex);

retry:

    switch (reg->mem_type) {

--
2.29.2

Quoted 7 lines

--
Thomas Zimmermann
Graphics Driver Developer
SUSE Software Solutions Germany GmbH
Maxfeldstr. 5, 90409 Nürnberg, Germany
(HRB 36809, AG Nürnberg)
Geschäftsführer: Felix Imendörffer

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

#7 lore raw

Am 12.11.20 um 15:20 schrieb Ruhl, Michael J:

Quoted 97 lines
Quoted 97 lines

-----Original Message-----
From: Ben Skeggs <[email protected]>
Sent: Wednesday, November 11, 2020 9:39 PM
To: Ruhl, Michael J <[email protected]>
Cc: Thomas Zimmermann <[email protected]>; [email protected]; [email protected]; [email protected]; [email protected]; amd- [email protected]; [email protected]; dri- [email protected]; [email protected]; Roland Scheidegger <[email protected]>; Jason Gunthorpe <[email protected]>; Huang Rui <[email protected]>; VMware Graphics <linux-graphics- [email protected]>; Gerd Hoffmann <[email protected]>; spice- [email protected]; Alex Deucher <[email protected]>; Dave Airlie <[email protected]>; Likun Gao <[email protected]>; Felix Kuehling <[email protected]>; Hawking Zhang <[email protected]>
Subject: Re: [PATCH] drm/nouveau: Fix out-of-bounds access when deferencing MMU type

Quoted 5 lines · Michael J
Quoted 5 lines

-----Original Message-----
From: Thomas Zimmermann <[email protected]>
Sent: Wednesday, November 11, 2020 7:08 AM
To: Ruhl, Michael J <[email protected]>; [email protected]; [email protected]; [email protected]; [email protected]

Quoted 8 lines
Quoted 8 lines

Cc: [email protected]; [email protected]; Maarten Lankhorst <[email protected]>; Maxime Ripard <[email protected]>; Dave Airlie <[email protected]>; Gerd Hoffmann <[email protected]>; Alex Deucher <[email protected]>; VMware Graphics <[email protected]>; Roland Scheidegger <[email protected]>; Huang Rui <[email protected]>; Felix Kuehling <[email protected]>; Hawking Zhang <[email protected]>; Jason Gunthorpe <[email protected]>; Likun

Gao

Quoted 5 lines
Quoted 5 lines

<[email protected]>; [email protected]; spice- [email protected]; [email protected]
Subject: Re: [PATCH] drm/nouveau: Fix out-of-bounds access when deferencing MMU type

Quoted 8 lines
Quoted 8 lines

Hi

Am 10.11.20 um 16:27 schrieb Ruhl, Michael J:

Quoted 5 lines
Quoted 5 lines

-----Original Message-----
From: Thomas Zimmermann <[email protected]>
Sent: Tuesday, November 10, 2020 8:37 AM
To: [email protected]; [email protected]; [email protected]; Ruhl,

Michael J

Quoted 6 lines
Quoted 6 lines
Quoted 2 lines
Quoted 2 lines

<[email protected]>; [email protected]
Cc: [email protected]; [email protected];

Thomas

Quoted 3 lines
Quoted 3 lines

Zimmermann <[email protected]>; Maarten Lankhorst <[email protected]>; Maxime Ripard <[email protected]>; Dave Airlie <[email protected]>; Gerd

Hoffmann

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

<[email protected]>; Alex Deucher <[email protected]>; VMware Graphics <[email protected]>; Roland Scheidegger <[email protected]>; Huang Rui

<[email protected]>;

Quoted 4 lines
Quoted 4 lines
Quoted 2 lines
Quoted 2 lines

Felix Kuehling <[email protected]>; Hawking Zhang <[email protected]>; Jason Gunthorpe <[email protected]>; Likun

Gao

Quoted 1 line
Quoted 1 line

<[email protected]>; [email protected];

spice-

Quoted 8 lines
Quoted 8 lines
Quoted 2 lines
Quoted 2 lines

[email protected]; [email protected]
Subject: [PATCH] drm/nouveau: Fix out-of-bounds access when

deferencing

Quoted 3 lines
Quoted 3 lines

MMU type

The value of struct drm_device.ttm.type_vram can become -1 for

unknown

Quoted 1 line
Quoted 1 line

types of memory (see nouveau_ttm_init()). This leads to an out-of-

bounds

Quoted 6 lines
Quoted 5 lines
Quoted 2 lines
Quoted 1 line

error when accessing struct nvif_mmu.type[]:

Would this make more sense to just set the type_vram = 0 instead of -1?

From what I understand, these indices refer to an internal type of MMU, rsp the MMU's capabilities. However, my hardware (pre-NV50) does not have an MMU at all.

Yeah, and upon further review I see that my comment was completely

wrong

Quoted 3 lines

(value vs. index).

A better suggestion would have been, create an entry in the array that

means,

Quoted 10 lines

"unsupported type" with a value of 0, but...

Quoted 2 lines

I agree that it would be nice to have a cleaner design that incorporates this case, but resolving that would apparently require more than a bugfix.

I agree. The -1 index is a special case for the platform path (platform != NV_DEVICE_INFO_V0_SOC). This is a fix for the issue, but not a complete solution.

If you need it:

I've put an alternate fix for this here[1], and will get it into drm-fixes later today.

Quoted 8 lines
Quoted 5 lines

Ben.

[1]
https://github.com/skeggsb/nouveau/commit/4590f7120c2f1f4aea9d8b93a2d
ae43b312d35ad

This makes a lot of sense. I spent some time trying to reconcile the platform info
that was not being used in this case, but didn't see the solution like this. This is
pretty clean.

I was already wondering why the old code never hit that problem, but this explains it properly and also fixes it up cleanly.

Quoted 4 lines

If you would like:

Feel free to add an Reviewed-by: Christian König <[email protected]> as well.

Regards,
Christian.

Quoted 5 lines

For this solution as well.

Mike

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx