<div dir="ltr"><div>There is no GDS shadowing info in the device info uapi, so userspace can't create any GDS buffer and thus can't have any GDS va. It's a uapi issue, not what firmware wants to do.<br></div><div><br></div><div>Marek<br></div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Thu, Apr 6, 2023 at 6:31 AM Christian König <<a href="mailto:ckoenig.leichtzumerken@gmail.com" target="_blank">ckoenig.leichtzumerken@gmail.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
<div>
That's what I thought as well, but Mitch/Hans insisted on that.<br>
<br>
We should probably double check internally.<br>
<br>
Christian.<br>
<br>
<div>Am 06.04.23 um 11:43 schrieb Marek
Olšák:<br>
</div>
<blockquote type="cite">
<div dir="ltr">
<div>GDS memory isn't used on gfx11. Only GDS OA is used.<br>
</div>
<div><br>
</div>
<div>Marek<br>
</div>
</div>
<br>
<div class="gmail_quote">
<div dir="ltr" class="gmail_attr">On Thu, Apr 6, 2023 at 5:09 AM
Christian König <<a href="mailto:christian.koenig@amd.com" target="_blank">christian.koenig@amd.com</a>>
wrote:<br>
</div>
<blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
<div> Why that?<br>
<br>
This is the save buffer for GDS, not the old style GDS BOs.<br>
<br>
Christian.<br>
<br>
<div>Am 06.04.23 um 09:36 schrieb Marek Olšák:<br>
</div>
<blockquote type="cite">
<div dir="ltr">
<div>gds_va is unnecessary.</div>
<div><br>
</div>
<div>Marek<br>
</div>
</div>
<br>
<div class="gmail_quote">
<div dir="ltr" class="gmail_attr">On Thu, Mar 30, 2023
at 3:18 PM Alex Deucher <<a href="mailto:alexander.deucher@amd.com" target="_blank">alexander.deucher@amd.com</a>>
wrote:<br>
</div>
<blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">For GFX11, the UMD
needs to allocate some shadow buffers<br>
to be used for preemption. The UMD allocates the
buffers<br>
and passes the GPU virtual address to the kernel since
the<br>
kernel will program the packet that specified these<br>
addresses as part of its IB submission frame.<br>
<br>
v2: UMD passes shadow init to tell kernel when to
initialize<br>
the shadow<br>
<br>
Reviewed-by: Christian König <<a href="mailto:christian.koenig@amd.com" target="_blank">christian.koenig@amd.com</a>><br>
Signed-off-by: Alex Deucher <<a href="mailto:alexander.deucher@amd.com" target="_blank">alexander.deucher@amd.com</a>><br>
---<br>
include/uapi/drm/amdgpu_drm.h | 10 ++++++++++<br>
1 file changed, 10 insertions(+)<br>
<br>
diff --git a/include/uapi/drm/amdgpu_drm.h
b/include/uapi/drm/amdgpu_drm.h<br>
index b6eb90df5d05..3d9474af6566 100644<br>
--- a/include/uapi/drm/amdgpu_drm.h<br>
+++ b/include/uapi/drm/amdgpu_drm.h<br>
@@ -592,6 +592,7 @@ struct drm_amdgpu_gem_va {<br>
#define AMDGPU_CHUNK_ID_SCHEDULED_DEPENDENCIES 0x07<br>
#define AMDGPU_CHUNK_ID_SYNCOBJ_TIMELINE_WAIT 0x08<br>
#define AMDGPU_CHUNK_ID_SYNCOBJ_TIMELINE_SIGNAL 0x09<br>
+#define AMDGPU_CHUNK_ID_CP_GFX_SHADOW 0x0a<br>
<br>
struct drm_amdgpu_cs_chunk {<br>
__u32 chunk_id;<br>
@@ -708,6 +709,15 @@ struct drm_amdgpu_cs_chunk_data {<br>
};<br>
};<br>
<br>
+#define
AMDGPU_CS_CHUNK_CP_GFX_SHADOW_FLAGS_INIT_SHADOW
0x1<br>
+<br>
+struct drm_amdgpu_cs_chunk_cp_gfx_shadow {<br>
+ __u64 shadow_va;<br>
+ __u64 csa_va;<br>
+ __u64 gds_va;<br>
+ __u64 flags;<br>
+};<br>
+<br>
/*<br>
* Query h/w info: Flag that this is integrated
(a.h.a. fusion) GPU<br>
*<br>
-- <br>
2.39.2<br>
<br>
</blockquote>
</div>
</blockquote>
<br>
</div>
</blockquote>
</div>
</blockquote>
<br>
</div>
</blockquote></div>