On 10/8/26 08:50, Jesse Zhang wrote:
> Since amdgpu_drm_ioctl() was removed, amdgpu_info_ioctl() no longer
> holds a runtime PM reference. When the GPU is runtime suspended, the
> queries that touch the hardware run against a powered down device:
>
> - AMDGPU_INFO_READ_MMR_REG reads all ones,
> - AMDGPU_INFO_TIMESTAMP returns ~0ull,
> - AMDGPU_INFO_DEV_INFO reports 0 for the engine and memory clock
> limits because the SMU is down,
> - AMDGPU_INFO_SENSOR fails with -ENOENT.
>
> On Navi31 with an idle GPU in BAMACO, GRBM_STATUS reads 0xffffffff
> instead of 0x0000302c, the timestamp is 0xffffffffffffffff, and the
> min/max sclk is reported as 0/0 kHz instead of 500000/1880000 kHz.
>
> Resume the GPU for these queries, the same way the CS ioctl does.
>
> Fixes: a66aee8f3057 ("drm/amdgpu: get rid of amdgpu_drm_ioctl")
> Signed-off-by: Jesse Zhang <[email protected]>
Yeah, I already expected that we have missed some.
Reviewed-by: Christian König <[email protected]>
Regards,
Christian.
> ---
> drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c | 28 ++++++++++++++++++++++++-
> 1 file changed, 27 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
> b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
> index f34150d20a2f..1589b6a6575f 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
> @@ -780,9 +780,16 @@ int amdgpu_info_ioctl(struct drm_device *dev, void
> *data, struct drm_file *filp)
>
> return copy_to_user(out, &count, min(size, 4u)) ? -EFAULT : 0;
> }
> - case AMDGPU_INFO_TIMESTAMP:
> + case AMDGPU_INFO_TIMESTAMP: {
> + PM_RUNTIME_ACQUIRE_IF_ENABLED_AUTOSUSPEND(dev->dev, pm);
> +
> + ret = PM_RUNTIME_ACQUIRE_ERR(&pm);
> + if (ret)
> + return ret;
> +
> ui64 = amdgpu_gfx_get_gpu_clock_counter(adev);
> return copy_to_user(out, &ui64, min(size, 8u)) ? -EFAULT : 0;
> + }
> case AMDGPU_INFO_FW_VERSION: {
> struct drm_amdgpu_info_firmware fw_info;
>
> @@ -906,6 +913,12 @@ int amdgpu_info_ioctl(struct drm_device *dev, void
> *data, struct drm_file *filp)
> if (info->read_mmr_reg.count > 128)
> return -EINVAL;
>
> + PM_RUNTIME_ACQUIRE_IF_ENABLED_AUTOSUSPEND(dev->dev, pm);
> +
> + ret = PM_RUNTIME_ACQUIRE_ERR(&pm);
> + if (ret)
> + return ret;
> +
> regs = kmalloc_array(info->read_mmr_reg.count, sizeof(*regs),
> GFP_KERNEL);
> if (!regs)
> @@ -940,6 +953,13 @@ int amdgpu_info_ioctl(struct drm_device *dev, void
> *data, struct drm_file *filp)
> uint64_t vm_size;
> uint32_t pcie_gen_mask, pcie_width_mask;
>
> + /* The clock limits are only available while the SMU is up. */
> + PM_RUNTIME_ACQUIRE_IF_ENABLED_AUTOSUSPEND(dev->dev, pm);
> +
> + ret = PM_RUNTIME_ACQUIRE_ERR(&pm);
> + if (ret)
> + return ret;
> +
> dev_info = kzalloc_obj(*dev_info);
> if (!dev_info)
> return -ENOMEM;
> @@ -1167,6 +1187,12 @@ int amdgpu_info_ioctl(struct drm_device *dev, void
> *data, struct drm_file *filp)
> }
> }
> case AMDGPU_INFO_SENSOR: {
> + PM_RUNTIME_ACQUIRE_IF_ENABLED_AUTOSUSPEND(dev->dev, pm);
> +
> + ret = PM_RUNTIME_ACQUIRE_ERR(&pm);
> + if (ret)
> + return ret;
> +
> if (!adev->pm.dpm_enabled)
> return -ENOENT;
>
