[PATCH 1/2] drm/amdgpu: add amdgpu_gfx_sched_mask and amdgpu_compute_sched_mask debugfs

[Huang, Tim]

[Public]

Hi Jesse,

> -----Original Message-----
> From: amd-gfx <amd-gfx-bounces at lists.freedesktop.org> On Behalf Of
> Jesse.zhang at amd.com
> Sent: Friday, October 18, 2024 10:31 AM
> To: amd-gfx at lists.freedesktop.org
> Cc: Deucher, Alexander <Alexander.Deucher at amd.com>; Koenig, Christian
> <Christian.Koenig at amd.com>; Zhang, Jesse(Jie) <Jesse.Zhang at amd.com>
> Subject: [PATCH 1/2] drm/amdgpu: add amdgpu_gfx_sched_mask and
> amdgpu_compute_sched_mask debugfs
>
> compute/gfx may have multiple rings on some hardware.
> In some cases, userspace wants to run jobs on a specific ring for validation
> purposes.
> This debugfs entry helps to disable or enable submitting jobs to a specific ring.
> This entry is populated only if there are at least two or more cores in the
> gfx/compute ip.
>
> Signed-off-by: Jesse Zhang <jesse.zhang at amd.com> Suggested-by:Alex
> Deucher <alexander.deucher at amd.com>
> ---
>  drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c |   2 +
>  drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c     | 142
> ++++++++++++++++++++
>  drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h     |   2 +
>  3 files changed, 146 insertions(+)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c
> b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c
> index 37d8657f0776..6e3f657cab9c 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c
> @@ -2096,6 +2096,8 @@ int amdgpu_debugfs_init(struct amdgpu_device
> *adev)
>               amdgpu_debugfs_umsch_fwlog_init(adev, &adev->umsch_mm);
>
>       amdgpu_debugfs_jpeg_sched_mask_init(adev);
> +     amdgpu_debugfs_gfx_sched_mask_init(adev);
> +     amdgpu_debugfs_compute_sched_mask_init(adev);
>
>       amdgpu_ras_debugfs_create_all(adev);
>       amdgpu_rap_debugfs_init(adev);
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
> b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
> index b6acbe923b6b..29997c9f68b6 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c
> @@ -1868,3 +1868,145 @@ void
> amdgpu_gfx_enforce_isolation_ring_end_use(struct amdgpu_ring *ring)
>       }
>       mutex_unlock(&adev->enforce_isolation_mutex);
>  }
> +
> +/*
> + * debugfs for to enable/disable gfx job submission to specific core.
> + */
> +#if defined(CONFIG_DEBUG_FS)
> +static int amdgpu_debugfs_gfx_sched_mask_set(void *data, u64 val) {
> +     struct amdgpu_device *adev = (struct amdgpu_device *)data;
> +     u32 i;
> +     u64 mask = 0;
> +     struct amdgpu_ring *ring;
> +
> +     if (!adev)
> +             return -ENODEV;
> +
> +     mask = (1 << adev->gfx.num_gfx_rings) - 1;
> +     if ((val & mask) == 0)
> +             return -EINVAL;
> +
> +     for (i = 0; i < adev->gfx.num_gfx_rings; ++i) {
> +             ring = &adev->gfx.gfx_ring[i];
> +             if (val & (1 << i))
> +                     ring->sched.ready = true;
> +             else
> +                     ring->sched.ready = false;
> +     }
> +     /* publish sched.ready flag update effective immediately across smp */
> +     smp_rmb();
> +     return 0;
> +}
> +
> +static int amdgpu_debugfs_gfx_sched_mask_get(void *data, u64 *val) {
> +     struct amdgpu_device *adev = (struct amdgpu_device *)data;
> +     u32 i;
> +     u64 mask = 0;
> +     struct amdgpu_ring *ring;
> +
> +     if (!adev)
> +             return -ENODEV;
> +     for (i = 0; i < adev->gfx.num_gfx_rings; ++i) {
> +             ring = &adev->gfx.gfx_ring[i];
> +             if (ring->sched.ready)
> +                     mask |= 1 << i;
> +     }
> +
> +     *val = mask;
> +     return 0;
> +}
> +
> +DEFINE_DEBUGFS_ATTRIBUTE(amdgpu_debugfs_gfx_sched_mask_fops,
> +                      amdgpu_debugfs_gfx_sched_mask_get,
> +                      amdgpu_debugfs_gfx_sched_mask_set, "%llx\n");
> +
> +#endif
> +
> +void amdgpu_debugfs_gfx_sched_mask_init(struct amdgpu_device *adev) {
> +#if defined(CONFIG_DEBUG_FS)
> +     struct drm_minor *minor = adev_to_drm(adev)->primary;
> +     struct dentry *root = minor->debugfs_root;
> +     char name[32];
> +
> +     if (!(adev->gfx.num_gfx_rings > 1))
> +             return;
> +     sprintf(name, "amdgpu_gfx_sched_mask");

I recommend using the literal string 'amdgpu_gfx_sched_mask' as the argument for `debugfs_create_file` instead of a variable name. The same applies to the other instances.

With or without this change,
This series are,

Reviewed-by: Tim Huang <tim.huang at amd.com>

Best Regards,
Tim



> +     debugfs_create_file(name, 0600, root, adev,
> +                         &amdgpu_debugfs_gfx_sched_mask_fops);
> +#endif
> +}
> +
> +/*
> + * debugfs for to enable/disable compute job submission to specific core.
> + */
> +#if defined(CONFIG_DEBUG_FS)
> +static int amdgpu_debugfs_compute_sched_mask_set(void *data, u64 val) {
> +     struct amdgpu_device *adev = (struct amdgpu_device *)data;
> +     u32 i;
> +     u64 mask = 0;
> +     struct amdgpu_ring *ring;
> +
> +     if (!adev)
> +             return -ENODEV;
> +
> +     mask = (1 << adev->gfx.num_compute_rings) - 1;
> +     if ((val & mask) == 0)
> +             return -EINVAL;
> +
> +     for (i = 0; i < adev->gfx.num_compute_rings; ++i) {
> +             ring = &adev->gfx.compute_ring[i];
> +             if (val & (1 << i))
> +                     ring->sched.ready = true;
> +             else
> +                     ring->sched.ready = false;
> +     }
> +
> +     /* publish sched.ready flag update effective immediately across smp */
> +     smp_rmb();
> +     return 0;
> +}
> +
> +static int amdgpu_debugfs_compute_sched_mask_get(void *data, u64 *val)
> +{
> +     struct amdgpu_device *adev = (struct amdgpu_device *)data;
> +     u32 i;
> +     u64 mask = 0;
> +     struct amdgpu_ring *ring;
> +
> +     if (!adev)
> +             return -ENODEV;
> +     for (i = 0; i < adev->gfx.num_compute_rings; ++i) {
> +             ring = &adev->gfx.compute_ring[i];
> +             if (ring->sched.ready)
> +                     mask |= 1 << i;
> +     }
> +
> +     *val = mask;
> +     return 0;
> +}
> +
> +DEFINE_DEBUGFS_ATTRIBUTE(amdgpu_debugfs_compute_sched_mask_fops
> ,
> +                      amdgpu_debugfs_compute_sched_mask_get,
> +                      amdgpu_debugfs_compute_sched_mask_set, "%llx\n");
> +
> +#endif
> +
> +void amdgpu_debugfs_compute_sched_mask_init(struct amdgpu_device
> *adev)
> +{ #if defined(CONFIG_DEBUG_FS)
> +     struct drm_minor *minor = adev_to_drm(adev)->primary;
> +     struct dentry *root = minor->debugfs_root;
> +     char name[32];
> +
> +     if (!(adev->gfx.num_compute_rings > 1))
> +             return;
> +     sprintf(name, "amdgpu_compute_sched_mask");
> +     debugfs_create_file(name, 0600, root, adev,
> +                         &amdgpu_debugfs_compute_sched_mask_fops);
> +#endif
> +}
> +
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
> b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
> index f710178a21bc..9275c02c94c6 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
> @@ -582,6 +582,8 @@ void amdgpu_gfx_sysfs_isolation_shader_fini(struct
> amdgpu_device *adev);  void amdgpu_gfx_enforce_isolation_handler(struct
> work_struct *work);  void
> amdgpu_gfx_enforce_isolation_ring_begin_use(struct amdgpu_ring *ring);
> void amdgpu_gfx_enforce_isolation_ring_end_use(struct amdgpu_ring *ring);
> +void amdgpu_debugfs_gfx_sched_mask_init(struct amdgpu_device *adev);
> +void amdgpu_debugfs_compute_sched_mask_init(struct amdgpu_device
> +*adev);
>
>  static inline const char *amdgpu_gfx_compute_mode_desc(int mode)  {
> --
> 2.25.1