Re: [PATCH v3 2/6] drm/xe/xe_ras: Add support to query page offline queue and list

["Ghimiray, Himal Prasad" <himal.prasad.ghimiray@intel.com>]

On 28-09-2026 11:48, Riana Tauro wrote:
> Add support to query page offline list and queue from firmware
> during module load. The page offline list command retrieves pages that
> are already offlined by the firmware. The page offline queue command
> retrieves the pages pending to be offlined by the firmware.
> 
> Cc: Tejas Upadhyay <tejas.upadhyay@intel.com>
> Signed-off-by: Riana Tauro <riana.tauro@intel.com>
> ---
> v2: rebase
>      store total pages once per response (Sashiko)
> 
> v3: common function for offline and queue (Himal)
> ---
>   drivers/gpu/drm/xe/xe_ras.c                   | 74 +++++++++++++++++++
>   drivers/gpu/drm/xe/xe_ras_types.h             | 35 +++++++++
>   drivers/gpu/drm/xe/xe_sysctrl_mailbox_types.h |  4 +
>   3 files changed, 113 insertions(+)
> 
> diff --git a/drivers/gpu/drm/xe/xe_ras.c b/drivers/gpu/drm/xe/xe_ras.c
> index 1225c561a872..752754f09314 100644
> --- a/drivers/gpu/drm/xe/xe_ras.c
> +++ b/drivers/gpu/drm/xe/xe_ras.c
> @@ -335,6 +335,77 @@ static bool ras_counter_is_valid(struct xe_device *xe, struct xe_ras_error_class
>   	return true;
>   }
>   
> +static void get_offline_pages(struct xe_device *xe, u32 cmd, void *req, size_t req_size,
> +			      void *resp, size_t resp_size,
> +			      struct xe_ras_offline_common *common, bool offline)
> +{
> +	struct xe_sysctrl_mailbox_command command = {0};
> +	struct xe_ras_offline_list_request *list_req;
> +	u32 total_pages = 0, count = 0;
> +	ssize_t rlen;
> +	int ret, i;
> +
> +	list_req = req ? req : NULL;
> +
> +	xe_sysctrl_create_command(&command, XE_SYSCTRL_GROUP_GFSP, cmd, req, req_size, resp,
> +				  resp_size);
> +
> +	do {
> +		memset(resp, 0, resp_size);
> +
> +		if (list_req)
> +			list_req->index = count;
> +
> +		ret = xe_sysctrl_send_command(&xe->sc, &command, &rlen);
> +		if (ret) {
> +			xe_log_err(xe, SYSCTRL, ret, "failed to get page offline data, cmd=%#x\n",
> +				   cmd);
> +			return;
> +		}
> +
> +		if (rlen != resp_size) {
> +			xe_log_err(xe, SYSCTRL, -EINVAL,
> +				   "unexpected page offline response length %zu (expected %zu), cmd=%#x\n",
> +				   rlen, resp_size, cmd);
> +			return;
> +		}
> +
> +		for (i = 0; i < common->pages_returned && i < XE_RAS_NUM_PAGES; i++)
> +			handle_page_offline(xe, common->page_addresses[i], offline);
> +
> +		count += common->pages_returned;
> +		if (!common->pages_returned)
> +			break;
> +
> +		if (!total_pages)
> +			total_pages = common->total_pages;
> +
> +		if (count > total_pages) {
> +			xe_log_err(xe, SYSCTRL, -EINVAL,
> +				   "Pages returned exceed total pages %u, returned %u, cmd=%#x\n",
> +				   total_pages, count, cmd);
> +			return;
> +		}
> +	} while (common->additional_data);
> +}
> +
> +static void get_queued_pages(struct xe_device *xe)
> +{
> +	struct xe_ras_offline_common response = {0};
> +
> +	get_offline_pages(xe, XE_SYSCTRL_CMD_GET_OFFLINE_QUEUE, NULL, 0, &response,
> +			  sizeof(response), &response, true);
> +}
> +
> +static void get_offlined_list(struct xe_device *xe)
> +{
> +	struct xe_ras_offline_list_response response = {0};
> +	struct xe_ras_offline_list_request request = {0};
> +
> +	get_offline_pages(xe, XE_SYSCTRL_CMD_GET_OFFLINE_LIST, &request, sizeof(request),
> +			  &response, sizeof(response), &response.common, false);
> +}
> +
>   static struct pci_dev *find_usp_dev(struct pci_dev *pdev)
>   {
>   	struct pci_dev *vsp;
> @@ -1049,6 +1120,9 @@ void xe_ras_init(struct xe_device *xe)
>   	if (IS_ENABLED(CONFIG_PCIEAER))
>   		ras_usp_aer_init(xe);
>   
> +	get_queued_pages(xe);
> +	get_offlined_list(xe);

I believe these also needs to be called during resume in case of 
suspend-resume flow.

> +
>   	ret = devm_device_add_group(xe->drm.dev, &gpu_health_group);
>   	if (ret)
>   		xe_err(xe, "Failed to create GPU health sysfs, err=%d\n", ret);
> diff --git a/drivers/gpu/drm/xe/xe_ras_types.h b/drivers/gpu/drm/xe/xe_ras_types.h
> index f119489bcdf2..021ffbd6d4e2 100644
> --- a/drivers/gpu/drm/xe/xe_ras_types.h
> +++ b/drivers/gpu/drm/xe/xe_ras_types.h
> @@ -10,6 +10,7 @@
>   
>   #define XE_RAS_NUM_COUNTERS			16
>   #define XE_RAS_NUM_ERROR_ARR			3
> +#define XE_RAS_NUM_PAGES			25
>   /* Error bits in IEH global error status register */
>   #define XE_RAS_SOC_IEH_PUNIT			BIT(1)
>   /* Device memory error categories */
> @@ -330,6 +331,40 @@ struct xe_ras_page_offline_response {
>   	u32 reserved;
>   } __packed;
>   
> +/**
> + * struct xe_ras_offline_common - Common structure for offline list and queue
> + */
> +struct xe_ras_offline_common {
> +	/** @total_pages: Total number of queued pages */
> +	u32 total_pages;
> +	/** @pages_returned: Number of pages returned in this response */
> +	u32 pages_returned;
> +	/** @page_addresses: Array of page addresses (4KB aligned) */
> +	u64 page_addresses[XE_RAS_NUM_PAGES];
> +	/** @additional_data: Indicates if more data is available */
> +	u8 additional_data;
> +	/** @reserved: Reserved for future use */
> +	u8 reserved[3];
> +} __packed;
> +
> +/**
> + * struct xe_ras_offline_list_request - Request for get offline list command
> + */
> +struct xe_ras_offline_list_request {
> +	/** @index: Zero-based index into the offline page list */
> +	u32 index;
> +} __packed;
> +
> +/**
> + * struct xe_ras_offline_list_response - Response from get offline list command
> + */
> +struct xe_ras_offline_list_response {
> +	/** @max_entries: Total no of pages that can be stored in flash */
> +	u32 max_entries;
> +	/** @common: Common offline page information */
> +	struct xe_ras_offline_common common;
> +} __packed;
> +
>   /**
>    * struct xe_ras_get_health_request - Request structure for obtaining gpu health
>    */
> diff --git a/drivers/gpu/drm/xe/xe_sysctrl_mailbox_types.h b/drivers/gpu/drm/xe/xe_sysctrl_mailbox_types.h
> index a01576bf2e73..3a71ed446949 100644
> --- a/drivers/gpu/drm/xe/xe_sysctrl_mailbox_types.h
> +++ b/drivers/gpu/drm/xe/xe_sysctrl_mailbox_types.h
> @@ -31,6 +31,8 @@ enum xe_sysctrl_group {
>    * @XE_SYSCTRL_CMD_SET_THRESHOLD: Set error threshold
>    * @XE_SYSCTRL_CMD_GET_PENDING_EVENT: Retrieve pending event
>    * @XE_SYSCTRL_CMD_PAGE_OFFLINE: Instruct firmware to offline/remove a page
> + * @XE_SYSCTRL_CMD_GET_OFFLINE_LIST: Retrieve list of all offlined pages from flash
> + * @XE_SYSCTRL_CMD_GET_OFFLINE_QUEUE: Retrieve list of offlined queued pages from firmware
>    * @XE_SYSCTRL_CMD_GET_HEALTH: Retrieve gpu health
>    * @XE_SYSCTRL_CMD_SET_HEALTH: Set gpu health
>    */
> @@ -42,6 +44,8 @@ enum xe_sysctrl_gfsp_cmd {
>   	XE_SYSCTRL_CMD_SET_THRESHOLD		= 0x06,
>   	XE_SYSCTRL_CMD_GET_PENDING_EVENT	= 0x07,
>   	XE_SYSCTRL_CMD_PAGE_OFFLINE		= 0x08,
> +	XE_SYSCTRL_CMD_GET_OFFLINE_LIST		= 0x09,
> +	XE_SYSCTRL_CMD_GET_OFFLINE_QUEUE	= 0x0A,
>   	XE_SYSCTRL_CMD_GET_HEALTH		= 0x0B,
>   	XE_SYSCTRL_CMD_SET_HEALTH		= 0x0C,
>   };