Re: [PATCH v4 5/6] drm/amdgpu: Add logic for VF ipd and VF bios to init from dynamic crit_region offsets

["Lazar, Lijo" <lijo.lazar@amd.com>]

On 10/15/2025 1:43 AM, Ellen Pan wrote:
> 1. Added VF logic in amdgpu_virt to init IP discovery using the offsets from dynamic(v2) critical regions;
> 2. Added VF logic in amdgpu_virt 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      | 36 ++++++++---
>   drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c | 23 +++++--
>   drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c      | 63 +++++++++++++++++++
>   drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h      |  2 +
>   4 files changed, 111 insertions(+), 13 deletions(-)
> 
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c
> index 00e96419fcda..5960ab1be4d8 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c
> @@ -96,11 +96,12 @@ void amdgpu_bios_release(struct amdgpu_device *adev)
>    * part of the system bios.  On boot, the system bios puts a
>    * copy of the igp rom at the start of vram if a discrete card is
>    * present.
> - * For SR-IOV, the vbios image is also put in VRAM in the VF.
> + * For SR-IOV, if dynamic critical region is not enabled,
> + * the vbios image is also put at the start of VRAM in the VF.
>    */
>   static bool amdgpu_read_bios_from_vram(struct amdgpu_device *adev)
>   {
> -	uint8_t __iomem *bios;
> +	uint8_t __iomem *bios = NULL;
>   	resource_size_t vram_base;
>   	resource_size_t size = 256 * 1024; /* ??? */
>   
> @@ -114,18 +115,35 @@ 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 (!bios)
> -		return false;
> +
> +	/* For SR-IOV, if dynamic critical region is enabled,
> +	* the vbios image is put at a dynamic offset of VRAM in the VF.
> +	* If dynamic critical region is disabled, follow the same seq as on baremetal.
> +	*/
> +	if (!(amdgpu_sriov_vf(adev) && adev->virt.is_dynamic_crit_regn_enabled)) {
> +		bios = ioremap_wc(vram_base, size);
> +		if (!bios)
> +				return false;
> +	}
>   
>   	adev->bios = kmalloc(size, GFP_KERNEL);
>   	if (!adev->bios) {
> -		iounmap(bios);
> -		return false;
> +			if (bios)
> +				iounmap(bios);
> +			return false;
>   	}
> +
> +	if (amdgpu_sriov_vf(adev) && adev->virt.is_dynamic_crit_regn_enabled) {
> +		if (amdgpu_virt_get_dynamic_data_info(adev,
> +					AMD_SRIOV_MSG_VBIOS_IMG_TABLE_ID, adev->bios, &size))
> +			return false;
> +	}
> +

It may be cleaner to keep this code up - allocate, read the binary and 
then use a label to jump to verification (check_atom_bios) for a 
successful read.

>   	adev->bios_size = size;
> -	memcpy_fromio(adev->bios, bios, size);
> -	iounmap(bios);
> +	if (bios) {
> +		memcpy_fromio(adev->bios, bios, size);
> +		iounmap(bios);
> +	}
>   
>   	if (!check_atom_bios(adev, size)) {
>   		amdgpu_bios_release(adev);
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c
> index 73401f0aeb34..e035dba96790 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c
> @@ -283,7 +283,7 @@ static int amdgpu_discovery_read_binary_from_mem(struct amdgpu_device *adev,
>   		 * wait for this to complete.  Once the C2PMSG is updated, we can
>   		 * continue.
>   		 */
> -
> +		
>   		for (i = 0; i < 2000; i++) {
>   			msg = RREG32(mmMP0_SMN_C2PMSG_33);
>   			if (msg & 0x80000000)
> @@ -299,13 +299,28 @@ static int amdgpu_discovery_read_binary_from_mem(struct amdgpu_device *adev,
>   		vram_size <<= 20;
>   
>   	if (sz_valid) {
> -		uint64_t pos = vram_size - DISCOVERY_TMR_OFFSET;
> -		amdgpu_device_vram_access(adev, pos, (uint32_t *)binary,
> -					  adev->mman.discovery_tmr_size, false);
> +		if (amdgpu_sriov_vf(adev) && adev->virt.is_dynamic_crit_regn_enabled) {
> +			/* For SRIOV VFs with dynamic critical region enabled,
> +			 * we will get the IPD binary via below call.
> +			 * If dynamic critical is disabled, fall through to normal seq.
> +			 */
> +			valid_size = vram_size;

Only mman.discovery_tmr_size is the allocation size of binary.
> +			if (amdgpu_virt_get_dynamic_data_info(adev,
> +						AMD_SRIOV_MSG_IPD_TABLE_ID, binary, &valid_size)) {
> +				ret = -EINVAL;
> +				goto exit;
> +			}
> +		} else {
> +			uint64_t pos = vram_size - DISCOVERY_TMR_OFFSET;
> +
> +			amdgpu_device_vram_access(adev, pos, (uint32_t *)binary,
> +					adev->mman.discovery_tmr_size, false);
> +		}
>   	} else {
>   		ret = amdgpu_discovery_read_binary_from_sysmem(adev, binary);
>   	}
>   
> +exit:
>   	if (ret)
>   		dev_err(adev->dev,
>   			"failed to read discovery info from memory, vram size read: %llx",
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c
> index 820dab538164..fef4ebb0f879 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c
> @@ -940,6 +940,14 @@ int amdgpu_virt_init_critical_region(struct amdgpu_device *adev)
>   	adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_BAD_PAGE_INFO_TABLE_ID].size_kb =
>   		init_data_hdr->bad_page_size_in_kb;
>   
> +	/* Validation for critical region info */
> +	 if (adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_IPD_TABLE_ID].size_kb > DISCOVERY_TMR_SIZE) {
> +		dev_err(adev->dev, "Invalid IP discovery size: 0x%x\n",
> +				adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_IPD_TABLE_ID].size_kb);
> +		r = -EINVAL;
> +		goto out;
> +	}
> +
>   	/* reserved memory starts from crit region base offset with the size of 5MB */
>   	adev->mman.fw_vram_usage_start_offset = adev->virt.crit_regn.offset;
>   	adev->mman.fw_vram_usage_size = adev->virt.crit_regn.size_kb << 10;
> @@ -958,6 +966,61 @@ int amdgpu_virt_init_critical_region(struct amdgpu_device *adev)
>   	return r;
>   }
>   
> +int amdgpu_virt_get_dynamic_data_info(struct amdgpu_device *adev,
> +	int data_id, uint8_t *binary, uint64_t *size)
> +{
> +	uint32_t data_offset = 0;
> +	uint32_t data_size = 0;
> +	enum amd_sriov_msg_table_id_enum data_table_id = data_id;
> +	char *data_name;
> +
> +	if (data_table_id >= AMD_SRIOV_MSG_MAX_TABLE_ID)
> +		return -EINVAL;
> +
> +	data_offset = adev->virt.crit_regn_tbl[data_table_id].offset;
> +	data_size = adev->virt.crit_regn_tbl[data_table_id].size_kb << 10;
> +
> +	switch (data_id) {
> +	case AMD_SRIOV_MSG_IPD_TABLE_ID:
> +		data_name = "IPD";
> +		if (!IS_ALIGNED(data_offset, 4) || !IS_ALIGNED(data_size, 4)) {
> +			dev_err(adev->dev, "IP discovery data not aligned to 4 bytes\n");
> +			return -EINVAL;
> +		}

I think this should also be part of table verification.

> +
> +		amdgpu_device_vram_access(adev,
> +				(uint64_t)data_offset, (uint32_t *)binary, data_size, false);
> +		if (!binary)
> +			return -EINVAL;

This NULL check should be before reading any content to the binary.> +
> +		if (((uint64_t)data_offset + (uint64_t)data_size) > *size)
> +			return -EINVAL;

This doesn't look correct either.

You may generically do like -
	// not valid binary or not enough size allocated
	if (!binary || !size || *size < data_size)
		return -EINVAL;

	// otherwise, copy the content
	amdgpu_device_vram_access(adev, (uint64_t)data_offset, (uint32_t 
*)binary, data_size, false);
	*size = (uint64_t)data_size; // update the size as out param.
> +
> +		*size = (uint64_t)data_size;
> +
> +		break;
> +
> +	case AMD_SRIOV_MSG_VBIOS_IMG_TABLE_ID:
> +		data_name = "BIOS";

If this name is preferred for printing, probably return as 
amdgpu_virt_get_dyn_table_name(tbl_id) so that it can be used at other 
places as well.
> +		if (data_size > *size) {
> +			dev_err(adev->dev, "Invalid vbios size: 0x%x\n", data_size);
> +			return -EINVAL;
> +		}
> +
> +		amdgpu_device_vram_access(adev,
> +				(uint64_t)data_offset, (uint32_t *)binary, data_size, false);
> +
> +		*size = (uint64_t)data_size;
> +		break;
> +	}
> +
> +	dev_info(adev->dev,
> +		"Got %s info from dynamic crit_region_table at offset 0x%x with size of 0x%x bytes.\n",
> +		data_name, data_offset, data_size);

This could be kept as dev_dbg().

Thanks,
Lijo

> +
> +	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 5c1dce9731e1..a3ae1ff40e84 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h
> @@ -440,6 +440,8 @@ 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_get_dynamic_data_info(struct amdgpu_device *adev,
> +	int data_id, uint8_t *binary, uint64_t *size);
>   
>   bool amdgpu_virt_can_access_debugfs(struct amdgpu_device *adev);
>   int amdgpu_virt_enable_access_debugfs(struct amdgpu_device *adev);