On Fri, Oct 10, 2025 at 12:43 AM Ellen Pan <[email protected]> wrote:
>
> 1. Added enum amd_sriov_crit_region_version to support multi versions
> 2. Added logic in SRIOV mailbox to regonize crit_region version during
> req_gpu_init_data
>
> Signed-off-by: Ellen Pan <[email protected]>
> ---
> drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c | 3 ++-
> drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h | 8 +++++++
> drivers/gpu/drm/amd/amdgpu/amdgv_sriovmsg.h | 5 +++++
> drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c | 23 ++++++++++++++-------
> 4 files changed, 31 insertions(+), 8 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c
> b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c
> index e95adf0407a0..3a6b0e1084d7 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c
> @@ -150,7 +150,8 @@ void amdgpu_virt_request_init_data(struct amdgpu_device
> *adev)
> virt->ops->req_init_data(adev);
>
> if (adev->virt.req_init_data_ver > 0)
> - DRM_INFO("host supports REQ_INIT_DATA handshake\n");
> + DRM_INFO("host supports REQ_INIT_DATA handshake of
> critical_region_version %d\n",
> + adev->virt.req_init_data_ver);
dev_info() is preferred.
> else
> DRM_WARN("host doesn't support REQ_INIT_DATA handshake\n");
Change this to dev_warn() while you are here.
> }
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h
> b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h
> index d1172c8e58c4..51ff73b1fcd5 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h
> @@ -262,6 +262,11 @@ struct amdgpu_virt_ras {
>
> DECLARE_ATTR_CAP_CLASS(amdgpu_virt, AMDGPU_VIRT_CAPS_LIST);
>
> +struct amdgpu_virt_region {
> + uint64_t offset;
> + uint64_t size_kb;
> +};
> +
> /* GPU virtualization */
> struct amdgpu_virt {
> uint32_t caps;
> @@ -289,6 +294,9 @@ struct amdgpu_virt {
> bool ras_init_done;
> uint32_t reg_access;
>
> + /* dynamic(v2) critical regions */
> + struct amdgpu_virt_region init_data_header;
> +
> /* vf2pf message */
> struct delayed_work vf2pf_work;
> uint32_t vf2pf_update_interval_ms;
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgv_sriovmsg.h
> b/drivers/gpu/drm/amd/amdgpu/amdgv_sriovmsg.h
> index 3b35154e2df6..b53caab5b706 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgv_sriovmsg.h
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgv_sriovmsg.h
> @@ -65,6 +65,11 @@
> (AMD_SRIOV_MSG_VBIOS_SIZE_KB_V1 +
> AMD_SRIOV_MSG_DATAEXCHANGE_SIZE_KB_V1 + \
> AMD_SRIOV_MSG_RAS_TELEMETRY_SIZE_KB_V1)
>
> +enum amd_sriov_crit_region_version {
> + GPU_CRIT_REGION_V1 = 1,
> + GPU_CRIT_REGION_V2 = 2,
> +};
> +
> /*
> * PF2VF history log:
> * v1 defined in amdgim
> diff --git a/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c
> b/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c
> index e5282a5d05d9..cd5b2f07edb8 100644
> --- a/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c
> +++ b/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c
> @@ -222,12 +222,20 @@ static int
> xgpu_nv_send_access_requests_with_param(struct amdgpu_device *adev,
> adev->virt.req_init_data_ver = 0;
> } else {
> if (req == IDH_REQ_GPU_INIT_DATA) {
> - adev->virt.req_init_data_ver =
> -
> RREG32_NO_KIQ(mmMAILBOX_MSGBUF_RCV_DW1);
> -
> - /* assume V1 in case host doesn't set version
> number */
> - if (adev->virt.req_init_data_ver < 1)
> - adev->virt.req_init_data_ver = 1;
> + switch
> (RREG32_NO_KIQ(mmMAILBOX_MSGBUF_RCV_DW1)) {
> + case GPU_CRIT_REGION_V2:
> + adev->virt.req_init_data_ver =
> GPU_CRIT_REGION_V2;
> + adev->virt.init_data_header.offset =
> +
> RREG32_NO_KIQ(mmMAILBOX_MSGBUF_RCV_DW2);
> + adev->virt.init_data_header.size_kb =
> +
> RREG32_NO_KIQ(mmMAILBOX_MSGBUF_RCV_DW3);
> + break;
> + default:
> + adev->virt.req_init_data_ver =
> GPU_CRIT_REGION_V1;
> + adev->virt.init_data_header.offset =
> -1;
> + adev->virt.init_data_header.size_kb =
> 0;
> + break;
> + }
> }
> }
>
> @@ -285,7 +293,8 @@ static int xgpu_nv_release_full_gpu_access(struct
> amdgpu_device *adev,
>
> static int xgpu_nv_request_init_data(struct amdgpu_device *adev)
> {
> - return xgpu_nv_send_access_requests(adev, IDH_REQ_GPU_INIT_DATA);
> + return xgpu_nv_send_access_requests_with_param(adev,
> IDH_REQ_GPU_INIT_DATA,
> + 0, GPU_CRIT_REGION_V2, 0);
What is this change for? This parameter doesn't seem to be used.
Alex
> }
>
> static int xgpu_nv_mailbox_ack_irq(struct amdgpu_device *adev,
> --
> 2.34.1
>
