From: "Christian König" <christian.koenig@amd.com>
To: Ellen Pan <yunru.pan@amd.com>, amd-gfx@lists.freedesktop.org
Cc: Alexander.Deucher@amd.com, Shravankumar.Gande@amd.com
Subject: Re: [PATCH 5/6] drm/amdgpu: Add logic for VF ipd and VF bios to init from dynamic crit_region offsets
Date: Thu, 9 Oct 2025 14:31:56 +0200 [thread overview]
Message-ID: <1fa7760d-913f-4826-869d-818b60fc74ad@amd.com> (raw)
In-Reply-To: <20251009033106.25022-1-yunru.pan@amd.com>
On 09.10.25 05:31, Ellen Pan wrote:
> 1. Added VF logic to init IP discovery using the offsets from dynamic(v2) critical regions;
> 2. Added VF logic to init bios image using the offsets from dynamic(v2) critical regions;
>
> Signed-off-by: Ellen Pan <yunru.pan@amd.com>
> ---
> drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c | 12 ++++-
> drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c | 5 ++
> drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c | 47 +++++++++++++++++++
> drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h | 3 ++
> 4 files changed, 66 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c
> index 00e96419fcda..2cbb24ede86e 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c
> @@ -114,7 +114,17 @@ static bool amdgpu_read_bios_from_vram(struct amdgpu_device *adev)
>
> adev->bios = NULL;
> vram_base = pci_resource_start(adev->pdev, 0);
> - bios = ioremap_wc(vram_base, size);
> +
> + if (amdgpu_sriov_vf(adev) && adev->virt.init_data_done) {
> + resource_size_t bios_offset;
> +
> + if (amdgpu_virt_get_bios_info(adev, &bios_offset, &size))
> + return false;
> +
> + bios = ioremap_wc(vram_base + bios_offset, size);
> + } else
> + bios = ioremap_wc(vram_base, size);
> +
It would probably better to give offset as parameter to amdgpu_read_bios_from_vram().
There is already a comment saying "this is required for SR-IOV" where that code would make perfect sense.
> if (!bios)
> return false;
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c
> index 73401f0aeb34..0dd9ff576cdc 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c
> @@ -292,6 +292,11 @@ static int amdgpu_discovery_read_binary_from_mem(struct amdgpu_device *adev,
> }
> }
>
> + if ((adev->virt.req_init_data_ver == GPU_CRIT_REGION_V2) && adev->virt.init_data_done) {
> + ret = amdgpu_virt_init_ip_discovery(adev, binary);
> + return ret;
> + }
> +
It is probably better to rework the code here and put all the SRIOV handling into the calling function.
And then call either amdgpu_discovery_read_binary_from_mem() or amdgpu_virt_init_ip_discovery() and not delegate the call to another function after calling the first one.
> vram_size = RREG32(mmRCC_CONFIG_MEMSIZE);
> if (!vram_size || vram_size == U32_MAX)
> sz_valid = false;
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c
> index e9dbab53cb06..9181acef4e9b 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c
> @@ -954,6 +954,53 @@ int amdgpu_virt_init_critical_region(struct amdgpu_device *adev)
> return r;
> }
>
> +int amdgpu_virt_init_ip_discovery(struct amdgpu_device *adev, uint8_t *binary)
> +{
> + uint32_t ip_discovery_offset =
> + adev->virt.crit_region_offsets[AMD_SRIOV_MSG_IPD_TABLE_ID];
> + uint32_t ip_discovery_size =
> + adev->virt.crit_region_sizes_kb[AMD_SRIOV_MSG_IPD_TABLE_ID] << 10;
> + uint64_t pos = 0;
> +
> + dev_info(adev->dev, "use ip discovery information copied from dynamic crit_region_table at offset 0x%x with size of 0x%x bytes.\n", ip_discovery_offset, ip_discovery_size);
That line is to long, please use checkpatch.pl.
> +
> + if (!IS_ALIGNED(ip_discovery_offset, 4) || !IS_ALIGNED(ip_discovery_size, 4)) {
> + DRM_ERROR("IP discovery data not aligned to 4 bytes\n");
> + return -EINVAL;
> + }
> +
> + if (ip_discovery_size > DISCOVERY_TMR_SIZE) {
> + DRM_ERROR("Invalid IP discovery size: 0x%x\n", ip_discovery_size);
> + return -EINVAL;
> + }
> +
> + pos = (uint64_t)ip_discovery_offset;
> + amdgpu_device_vram_access(adev, pos, (uint32_t *)binary,
> + ip_discovery_size, false);
> +
> + return 0;
> +}
> +
> +int amdgpu_virt_get_bios_info(struct amdgpu_device *adev,
> + resource_size_t *bios_offset, resource_size_t *bios_size)
> +{
> + uint32_t vbios_offset = adev->virt.crit_region_offsets[AMD_SRIOV_MSG_VBIOS_IMG_TABLE_ID];
> + uint32_t vbios_size =
> + adev->virt.crit_region_sizes_kb[AMD_SRIOV_MSG_VBIOS_IMG_TABLE_ID] << 10;
> +
> + dev_info(adev->dev, "use bios information copied from dynamic crit_region_table at offset 0x%x with size of 0x%x bytes.\n", vbios_offset, vbios_size);
Same here.
Regards,
Christian.
> +
> + if (vbios_size > *bios_size) {
> + DRM_ERROR("Invalid vbios size: 0x%x\n", vbios_size);
> + return -EINVAL;
> + }
> +
> + *bios_offset = vbios_offset;
> + *bios_size = vbios_size;
> +
> + return 0;
> +}
> +
> void amdgpu_virt_init(struct amdgpu_device *adev)
> {
> bool is_sriov = false;
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h
> index 5f6014b2f349..d122347ff666 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h
> @@ -434,6 +434,9 @@ void amdgpu_virt_fini_data_exchange(struct amdgpu_device *adev);
> void amdgpu_virt_init(struct amdgpu_device *adev);
>
> int amdgpu_virt_init_critical_region(struct amdgpu_device *adev);
> +int amdgpu_virt_init_ip_discovery(struct amdgpu_device *adev, uint8_t *binary);
> +int amdgpu_virt_get_bios_info(struct amdgpu_device *adev,
> + resource_size_t *bios_offset, resource_size_t *bios_size);
>
> bool amdgpu_virt_can_access_debugfs(struct amdgpu_device *adev);
> int amdgpu_virt_enable_access_debugfs(struct amdgpu_device *adev);
next prev parent reply other threads:[~2025-10-09 12:32 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-10-09 3:31 [PATCH 5/6] drm/amdgpu: Add logic for VF ipd and VF bios to init from dynamic crit_region offsets Ellen Pan
2025-10-09 12:31 ` Christian König [this message]
2025-10-09 15:24 ` Lazar, Lijo
-- strict thread matches above, loose matches on Subject: below --
2025-10-10 4:43 [PATCH 1/6] drm/amdgpu: Updated naming of SRIOV critical region offsets/sizes with _V1 suffix Ellen Pan
2025-10-10 4:43 ` [PATCH 5/6] drm/amdgpu: Add logic for VF ipd and VF bios to init from dynamic crit_region offsets Ellen Pan
2025-10-10 10:46 ` Lazar, Lijo
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1fa7760d-913f-4826-869d-818b60fc74ad@amd.com \
--to=christian.koenig@amd.com \
--cc=Alexander.Deucher@amd.com \
--cc=Shravankumar.Gande@amd.com \
--cc=amd-gfx@lists.freedesktop.org \
--cc=yunru.pan@amd.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox