Re: [PATCH] ce/gf100: fix incorrect CE0 address calculation on some GPUs

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]<

 



On Wed, Nov 3, 2021 at 2:11 AM Ben Skeggs <skeggsb@xxxxxxxxx> wrote:
>
> From: Ben Skeggs <bskeggs@xxxxxxxxxx>
>
> The code which constructs the modules for each engine present on the GPU
> passes -1 for 'instance' on non-instanced engines, which affects how the
> name for a sub-device is generated.  This is then stored as 'instance 0'
> in nvkm_subdev.inst, so code can potentially be shared with earlier GPUs
> that only had a single instance of an engine.
>
> However, GF100's CE constructor uses this value to calculate the address
> of its falcon before it's translated, resulting in CE0 getting the wrong
> address.
>
> This slightly modifies the approach, always passing a valid instance for
> engines that *can* have multiple copies, and having the code for earlier
> GPUs explicitly ask for non-instanced name generation.
>
> Bug: https://gitlab.freedesktop.org/drm/nouveau/-/issues/91
>
> Fixes: 50551b15c760 ("drm/nouveau/ce: switch to instanced constructor")
> Cc: <stable@xxxxxxxxxxxxxxx> # v5.12+
> Signed-off-by: Ben Skeggs <bskeggs@xxxxxxxxxx>
> ---
>  drivers/gpu/drm/nouveau/nvkm/engine/ce/gt215.c    | 2 +-
>  drivers/gpu/drm/nouveau/nvkm/engine/device/base.c | 3 +--
>  2 files changed, 2 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/ce/gt215.c b/drivers/gpu/drm/nouveau/nvkm/engine/ce/gt215.c
> index 704df0f2d1f1..09a112af2f89 100644
> --- a/drivers/gpu/drm/nouveau/nvkm/engine/ce/gt215.c
> +++ b/drivers/gpu/drm/nouveau/nvkm/engine/ce/gt215.c
> @@ -78,6 +78,6 @@ int
>  gt215_ce_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst,
>              struct nvkm_engine **pengine)
>  {
> -       return nvkm_falcon_new_(&gt215_ce, device, type, inst,
> +       return nvkm_falcon_new_(&gt215_ce, device, type, -1,
>                                 (device->chipset != 0xaf), 0x104000, pengine);
>  }
> diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/device/base.c b/drivers/gpu/drm/nouveau/nvkm/engine/device/base.c
> index ca75c5f6ecaf..b51d690f375f 100644
> --- a/drivers/gpu/drm/nouveau/nvkm/engine/device/base.c
> +++ b/drivers/gpu/drm/nouveau/nvkm/engine/device/base.c
> @@ -3147,8 +3147,7 @@ nvkm_device_ctor(const struct nvkm_device_func *func,
>         WARN_ON(device->chip->ptr.inst & ~((1 << ARRAY_SIZE(device->ptr)) - 1));             \
>         for (j = 0; device->chip->ptr.inst && j < ARRAY_SIZE(device->ptr); j++) {            \
>                 if ((device->chip->ptr.inst & BIT(j)) && (subdev_mask & BIT_ULL(type))) {    \
> -                       int inst = (device->chip->ptr.inst == 1) ? -1 : (j);                 \
> -                       ret = device->chip->ptr.ctor(device, (type), inst, &device->ptr[j]); \
> +                       ret = device->chip->ptr.ctor(device, (type), (j), &device->ptr[j]);  \
>                         subdev = nvkm_device_subdev(device, (type), (j));                    \
>                         if (ret) {                                                           \
>                                 nvkm_subdev_del(&subdev);                                    \
> --
> 2.31.1
>

Reviewed-by: Karol Herbst <kherbst@xxxxxxxxxx>




[Index of Archives]     [Linux ARM Kernel]     [Linux ARM]     [Linux Omap]     [Fedora ARM]     [IETF Annouce]     [Security]     [Bugtraq]     [Linux]     [Linux OMAP]     [Linux MIPS]     [eCos]     [Asterisk Internet PBX]     [Linux API]

  Powered by Linux