2022-11-18 12:10:55

by Denis Arefev

[permalink] [raw]
Subject: [PATCH v2] amdgpu: nbio_v7_4: Add pointer check

Return value of a function 'amdgpu_ras_find_obj' is dereferenced
at nbio_v7_4.c:325 without checking for null

Found by Linux Verification Center (linuxtesting.org) with SVACE.

Signed-off-by: Denis Arefev <[email protected]>
---
drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c | 3 +++
1 file changed, 3 insertions(+)

diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c
index eadc9526d33f..0f2ac99de864 100644
--- a/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c
+++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c
@@ -304,6 +304,9 @@ static void nbio_v7_4_handle_ras_controller_intr_no_bifring(struct amdgpu_device
struct ras_err_data err_data = {0, 0, 0, NULL};
struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);

+ if (!obj)
+ return;

bif_doorbell_intr_cntl = RREG32_SOC15(NBIO, 0, mmBIF_DOORBELL_INT_CNTL);
if (REG_GET_FIELD(bif_doorbell_intr_cntl,
BIF_DOORBELL_INT_CNTL, RAS_CNTLR_INTERRUPT_STATUS)) {
--
2.25.1



2022-11-18 15:36:47

by Alex Deucher

[permalink] [raw]
Subject: Re: [PATCH v2] amdgpu: nbio_v7_4: Add pointer check

On Fri, Nov 18, 2022 at 9:11 AM Denis Arefev <[email protected]> wrote:
>
> Return value of a function 'amdgpu_ras_find_obj' is dereferenced
> at nbio_v7_4.c:325 without checking for null
>
> Found by Linux Verification Center (linuxtesting.org) with SVACE.
>
> Signed-off-by: Denis Arefev <[email protected]>
> ---
> drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c | 3 +++
> 1 file changed, 3 insertions(+)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c
> index eadc9526d33f..0f2ac99de864 100644
> --- a/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c
> +++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_4.c
> @@ -304,6 +304,9 @@ static void nbio_v7_4_handle_ras_controller_intr_no_bifring(struct amdgpu_device
> struct ras_err_data err_data = {0, 0, 0, NULL};
> struct amdgpu_ras *ras = amdgpu_ras_get_context(adev);
>
> + if (!obj)
> + return;
>

I don't think we could ever get here with a NULL obj. This is part of
the interrupt handler for NBIO RAS, so the interrupt would never be
enabled in the first place if the obj didn't exist.

Alex

> bif_doorbell_intr_cntl = RREG32_SOC15(NBIO, 0, mmBIF_DOORBELL_INT_CNTL);
> if (REG_GET_FIELD(bif_doorbell_intr_cntl,
> BIF_DOORBELL_INT_CNTL, RAS_CNTLR_INTERRUPT_STATUS)) {
> --
> 2.25.1
>