2023-12-05 13:07:20

by Marek Szyprowski

[permalink] [raw]
Subject: [PATCH] drm/debugfs: fix potential NULL pointer dereference

encoder->funcs entry might be NULL, so check it first before calling its
methods. This fixes NULL pointer dereference observed on Rasberry Pi
3b/4b boards.

Fixes: caf525ed45b4 ("drm/encoder: register per-encoder debugfs dir")
Signed-off-by: Marek Szyprowski <[email protected]>
---
This fixes the following issue observed on Raspberry Pi 4b:

vc4-drm gpu: bound fe400000.hvs (ops vc4_hvs_ops [vc4])
vc4-drm gpu: bound fef00700.hdmi (ops vc4_hdmi_ops [vc4])
vc4-drm gpu: bound fef05700.hdmi (ops vc4_hdmi_ops [vc4])
vc4-drm gpu: bound fe004000.txp (ops vc4_txp_ops [vc4])
vc4-drm gpu: bound fe206000.pixelvalve (ops vc4_crtc_ops [vc4])
vc4-drm gpu: bound fe207000.pixelvalve (ops vc4_crtc_ops [vc4])
vc4-drm gpu: bound fe20a000.pixelvalve (ops vc4_crtc_ops [vc4])
vc4-drm gpu: bound fe216000.pixelvalve (ops vc4_crtc_ops [vc4])
8<--- cut here ---
Unable to handle kernel NULL pointer dereference at virtual address 00000010 when read
[00000010] *pgd=00000000
Internal error: Oops: 5 [#1] SMP ARM
Modules linked in: sha256_arm raspberrypi_hwmon cfg80211(+) hci_uart btbcm bluetooth vc4(+) ecdh_generic ecc libaes snd_soc_hdmi_codec snd_soc_core v3d drm_shmem_helper ac97_bus snd_pcm_dmaengine snd_pcm genet(+) gpu_sched snd_timer snd bcm2711_thermal soundcore drm_dma_helper
CPU: 1 PID: 221 Comm: systemd-udevd Not tainted 6.7.0-rc4-next-20231205 #14267
Hardware name: BCM2711
PC is at drm_debugfs_encoder_add+0x6c/0x98
LR is at 0x0
...
drm_debugfs_encoder_add from drm_encoder_register_all+0x20/0x60
drm_encoder_register_all from drm_modeset_register_all+0x34/0x70
drm_modeset_register_all from drm_dev_register+0x24c/0x28c
drm_dev_register from vc4_drm_bind+0x21c/0x2d0 [vc4]
vc4_drm_bind [vc4] from try_to_bring_up_aggregate_device+0x160/0x1bc
try_to_bring_up_aggregate_device from component_master_add_with_match+0xc4/0xf8
component_master_add_with_match from vc4_platform_drm_probe+0xa0/0xc0 [vc4]
vc4_platform_drm_probe [vc4] from platform_probe+0x5c/0xb8
platform_probe from really_probe+0xc8/0x2dc
really_probe from __driver_probe_device+0x88/0x19c
__driver_probe_device from driver_probe_device+0x30/0x104
driver_probe_device from __driver_attach+0x90/0x174
__driver_attach from bus_for_each_dev+0x6c/0xb4
bus_for_each_dev from bus_add_driver+0xcc/0x1cc
bus_add_driver from driver_register+0x7c/0x118
driver_register from vc4_drm_register+0x44/0x1000 [vc4]
vc4_drm_register [vc4] from do_one_initcall+0x40/0x1e0
do_one_initcall from do_init_module+0x50/0x1e4
do_init_module from init_module_from_file+0x90/0xbc
init_module_from_file from sys_finit_module+0x144/0x258
sys_finit_module from ret_fast_syscall+0x0/0x54
Exception stack(0xf0cf1fa8 to 0xf0cf1ff0)
...
---[ end trace 0000000000000000 ]---

---
drivers/gpu/drm/drm_debugfs.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/drm_debugfs.c b/drivers/gpu/drm/drm_debugfs.c
index 02e7481758c0..f4715a67e340 100644
--- a/drivers/gpu/drm/drm_debugfs.c
+++ b/drivers/gpu/drm/drm_debugfs.c
@@ -638,7 +638,7 @@ void drm_debugfs_encoder_add(struct drm_encoder *encoder)
debugfs_create_file("bridges", 0444, root, encoder,
&bridges_fops);

- if (encoder->funcs->debugfs_init)
+ if (encoder->funcs && encoder->funcs->debugfs_init)
encoder->funcs->debugfs_init(encoder, root);
}

--
2.34.1


2023-12-05 13:21:10

by Dmitry Baryshkov

[permalink] [raw]
Subject: Re: [PATCH] drm/debugfs: fix potential NULL pointer dereference

On Tue, 5 Dec 2023 at 15:06, Marek Szyprowski <[email protected]> wrote:
>
> encoder->funcs entry might be NULL, so check it first before calling its
> methods. This fixes NULL pointer dereference observed on Rasberry Pi
> 3b/4b boards.
>
> Fixes: caf525ed45b4 ("drm/encoder: register per-encoder debugfs dir")
> Signed-off-by: Marek Szyprowski <[email protected]>

Reviewed-by: Dmitry Baryshkov <[email protected]>

> ---
> This fixes the following issue observed on Raspberry Pi 4b:
>
> vc4-drm gpu: bound fe400000.hvs (ops vc4_hvs_ops [vc4])
> vc4-drm gpu: bound fef00700.hdmi (ops vc4_hdmi_ops [vc4])
> vc4-drm gpu: bound fef05700.hdmi (ops vc4_hdmi_ops [vc4])
> vc4-drm gpu: bound fe004000.txp (ops vc4_txp_ops [vc4])
> vc4-drm gpu: bound fe206000.pixelvalve (ops vc4_crtc_ops [vc4])
> vc4-drm gpu: bound fe207000.pixelvalve (ops vc4_crtc_ops [vc4])
> vc4-drm gpu: bound fe20a000.pixelvalve (ops vc4_crtc_ops [vc4])
> vc4-drm gpu: bound fe216000.pixelvalve (ops vc4_crtc_ops [vc4])
> 8<--- cut here ---
> Unable to handle kernel NULL pointer dereference at virtual address 00000010 when read
> [00000010] *pgd=00000000
> Internal error: Oops: 5 [#1] SMP ARM
> Modules linked in: sha256_arm raspberrypi_hwmon cfg80211(+) hci_uart btbcm bluetooth vc4(+) ecdh_generic ecc libaes snd_soc_hdmi_codec snd_soc_core v3d drm_shmem_helper ac97_bus snd_pcm_dmaengine snd_pcm genet(+) gpu_sched snd_timer snd bcm2711_thermal soundcore drm_dma_helper
> CPU: 1 PID: 221 Comm: systemd-udevd Not tainted 6.7.0-rc4-next-20231205 #14267
> Hardware name: BCM2711
> PC is at drm_debugfs_encoder_add+0x6c/0x98
> LR is at 0x0
> ...
> drm_debugfs_encoder_add from drm_encoder_register_all+0x20/0x60
> drm_encoder_register_all from drm_modeset_register_all+0x34/0x70
> drm_modeset_register_all from drm_dev_register+0x24c/0x28c
> drm_dev_register from vc4_drm_bind+0x21c/0x2d0 [vc4]
> vc4_drm_bind [vc4] from try_to_bring_up_aggregate_device+0x160/0x1bc
> try_to_bring_up_aggregate_device from component_master_add_with_match+0xc4/0xf8
> component_master_add_with_match from vc4_platform_drm_probe+0xa0/0xc0 [vc4]
> vc4_platform_drm_probe [vc4] from platform_probe+0x5c/0xb8
> platform_probe from really_probe+0xc8/0x2dc
> really_probe from __driver_probe_device+0x88/0x19c
> __driver_probe_device from driver_probe_device+0x30/0x104
> driver_probe_device from __driver_attach+0x90/0x174
> __driver_attach from bus_for_each_dev+0x6c/0xb4
> bus_for_each_dev from bus_add_driver+0xcc/0x1cc
> bus_add_driver from driver_register+0x7c/0x118
> driver_register from vc4_drm_register+0x44/0x1000 [vc4]
> vc4_drm_register [vc4] from do_one_initcall+0x40/0x1e0
> do_one_initcall from do_init_module+0x50/0x1e4
> do_init_module from init_module_from_file+0x90/0xbc
> init_module_from_file from sys_finit_module+0x144/0x258
> sys_finit_module from ret_fast_syscall+0x0/0x54
> Exception stack(0xf0cf1fa8 to 0xf0cf1ff0)
> ...
> ---[ end trace 0000000000000000 ]---
>
> ---
> drivers/gpu/drm/drm_debugfs.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)


--
With best wishes
Dmitry

2023-12-06 14:54:57

by Nishanth Menon

[permalink] [raw]
Subject: Re: [PATCH] drm/debugfs: fix potential NULL pointer dereference

On 15:19-20231205, Dmitry Baryshkov wrote:
> On Tue, 5 Dec 2023 at 15:06, Marek Szyprowski <[email protected]> wrote:
> >
> > encoder->funcs entry might be NULL, so check it first before calling its
> > methods. This fixes NULL pointer dereference observed on Rasberry Pi
> > 3b/4b boards.
> >
> > Fixes: caf525ed45b4 ("drm/encoder: register per-encoder debugfs dir")
> > Signed-off-by: Marek Szyprowski <[email protected]>
>
> Reviewed-by: Dmitry Baryshkov <[email protected]>

Tested-by: Nishanth Menon <[email protected]>

Fixed multiple platform boot fails on TI K3 SoC family including
BeaglePlay, BeagleBone-AI64.

--
Regards,
Nishanth Menon
Key (0xDDB5849D1736249D) / Fingerprint: F8A2 8693 54EB 8232 17A3 1A34 DDB5 849D 1736 249D

2023-12-07 10:33:34

by Dmitry Baryshkov

[permalink] [raw]
Subject: Re: [PATCH] drm/debugfs: fix potential NULL pointer dereference

On Tue, 05 Dec 2023 14:06:31 +0100, Marek Szyprowski wrote:
> encoder->funcs entry might be NULL, so check it first before calling its
> methods. This fixes NULL pointer dereference observed on Rasberry Pi
> 3b/4b boards.
>
>

Applied to drm-misc-next, thanks!

[1/1] drm/debugfs: fix potential NULL pointer dereference
commit: 76385d493c2137460ee7735a5d3a494099c35188

Best regards,
--
Dmitry Baryshkov <[email protected]>