On 31-08-2026 08:41 pm, Alex Deucher wrote:
On Mon, Aug 31, 2026 at 11:02 AM Sunil Khatri <[email protected]> wrote:
Add the proper checks for some of the gfx11 variant IP's
for firmware support of userqueues.
Signed-off-by: Sunil Khatri <[email protected]>
---
drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c | 6 +++++-
1 file changed, 5 insertions(+), 1 deletion(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c
b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c
index 2c160ec9dc57..ca142c69cc28 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c
@@ -1670,7 +1670,11 @@ static int gfx_v11_0_sw_init(struct amdgpu_ip_block
*ip_block)
case IP_VERSION(11, 5, 2):
case IP_VERSION(11, 5, 3):
/* add firmware version checks here */
- if (0 && !adev->gfx.disable_uq) {
+ if (!adev->gfx.disable_uq &&
+ adev->gfx.me_fw_version >= 3090 &&
+ adev->gfx.pfp_fw_version >= 3190 &&
+ adev->gfx.mec_fw_version >= 3450 &&
+ adev->mes.fw_version[0] >= 147) {
We should verify this on all of these APUs first. I'm not sure if the
fixes have landed for phoenix parts yet. We may need to split the
checks for stix vs phx depending on the firmware versions.
ok, so for now i will drop this patch and will pick up later when APUs
will be tested... My understanding was to use the latest version which
should work for those APUs too, but yes its possible the fw does not
support these yet and hard to validate with device availability and
support as open questions...
Regards
Sunil Khatri
Alex
adev->userq_funcs[AMDGPU_HW_IP_GFX] = &userq_mes_funcs;
adev->userq_funcs[AMDGPU_HW_IP_COMPUTE] =
&userq_mes_funcs;
}
--
2.34.1
