[PATCH 2/5] drm/amdkfd: add multi-process debugging support for GC v9.4.3

[Kim, Jonathan]

[Public]

> -----Original Message-----
> From: Huang, JinHuiEric <JinHuiEric.Huang at amd.com>
> Sent: Wednesday, June 28, 2023 5:23 PM
> To: amd-gfx at lists.freedesktop.org
> Cc: Kim, Jonathan <Jonathan.Kim at amd.com>; Kim, Jonathan
> <Jonathan.Kim at amd.com>; Kuehling, Felix <Felix.Kuehling at amd.com>;
> Huang, JinHuiEric <JinHuiEric.Huang at amd.com>
> Subject: [PATCH 2/5] drm/amdkfd: add multi-process debugging support for
> GC v9.4.3
>
> From: Jonathan Kim <jonathan.kim at amd.com>
>
> Similar to GC v9.4.2, GC v9.4.3 should use the 5-Dword extended
> MAP_PROCESS packet to support multi-process debugging.  Update the
> mutli-process debug support list so that the KFD updates the runlist
> on debug mode setting and that it allocates enough GTT memory during
> KFD device initialization.
>
> Signed-off-by: Jonathan Kim <jonathan.kim at amd.com>
> Reviewed-by: Felix Kuehling <felix.kuehling at amd.com>
> Signed-off-by: Eric Huang <jinhuieric.huang at amd.com>
> ---
>  drivers/gpu/drm/amd/amdkfd/kfd_debug.h | 5 +++--
>  1 file changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_debug.h
> b/drivers/gpu/drm/amd/amdkfd/kfd_debug.h
> index a289e59ceb79..a0afc6a7b6c4 100644
> --- a/drivers/gpu/drm/amd/amdkfd/kfd_debug.h
> +++ b/drivers/gpu/drm/amd/amdkfd/kfd_debug.h
> @@ -76,8 +76,9 @@ int kfd_dbg_send_exception_to_runtime(struct
> kfd_process *p,
>
>  static inline bool kfd_dbg_is_per_vmid_supported(struct kfd_node *dev)
>  {
> -     return KFD_GC_VERSION(dev) == IP_VERSION(9, 4, 2) ||
> -            KFD_GC_VERSION(dev) >= IP_VERSION(11, 0, 0);
> +     return (KFD_GC_VERSION(dev) == IP_VERSION(9, 4, 2) ||
> +             KFD_GC_VERSION(dev) == IP_VERSION(9, 4, 3) ||
> +             KFD_GC_VERSION(dev) >= IP_VERSION(11, 0, 0));
>  }

This should probably be the last patch in the series since the rest of the series is required to enable debugging correctly.
Also did we ever close on enabling the trap temporaries regardless of debug mode?
IIRC, core dump will require this to capture consistent wave data.
That should probably be done in this patch since this check will also trigger the ttmp setup call on process creation.

Thanks,

Jon

>
>  void debug_event_write_work_handler(struct work_struct *work);
> --
> 2.34.1