Re: [PATCH v4 11/13] drm/xe/xe_ras: Add support for page offline list and queue commands

["Tauro, Riana" <riana.tauro@intel.com>]

Hi Tejas

On 4/21/2026 11:49 AM, Upadhyay, Tejas wrote:
>
>> -----Original Message-----
>> From: Tauro, Riana <riana.tauro@intel.com>
>> Sent: 17 April 2026 14:28
>> To: intel-xe@lists.freedesktop.org
>> Cc: Tauro, Riana <riana.tauro@intel.com>; Gupta, Anshuman
>> <anshuman.gupta@intel.com>; Vivi, Rodrigo <rodrigo.vivi@intel.com>;
>> aravind.iddamsetty@linux.intel.com; Nilawar, Badal
>> <badal.nilawar@intel.com>; Jadav, Raag <raag.jadav@intel.com>;
>> Koppuravuri, Ravi Kishore <ravi.kishore.koppuravuri@intel.com>; Koujalagi,
>> Mallesh <mallesh.koujalagi@intel.com>; Purkait, Soham
>> <soham.purkait@intel.com>; Upadhyay, Tejas <tejas.upadhyay@intel.com>;
>> Ghimiray, Himal Prasad <himal.prasad.ghimiray@intel.com>
>> Subject: [PATCH v4 11/13] drm/xe/xe_ras: Add support for page offline list
>> and queue commands
>>
>> Add handling for page offline list and queue sysctrl commands. 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>
>> Cc: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
>> Signed-off-by: Riana Tauro <riana.tauro@intel.com>
>> ---
>>   drivers/gpu/drm/xe/xe_ras.c                   | 85 +++++++++++++++++++
>>   drivers/gpu/drm/xe/xe_ras_types.h             | 39 +++++++++
>>   drivers/gpu/drm/xe/xe_sysctrl_mailbox_types.h |  6 +-
>>   3 files changed, 129 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/gpu/drm/xe/xe_ras.c b/drivers/gpu/drm/xe/xe_ras.c index
>> cd1ac6441b92..42ec27c05e9a 100644
>> --- a/drivers/gpu/drm/xe/xe_ras.c
>> +++ b/drivers/gpu/drm/xe/xe_ras.c
>> @@ -319,6 +319,87 @@ int xe_ras_page_offline(struct xe_device *xe, enum
>> xe_ras_page_action action, u6
>>   	return 0;
>>   }
>>
>> +static void get_queued_pages(struct xe_device *xe) {
>> +	struct xe_sysctrl_mailbox_command command = {0};
>> +	struct xe_ras_page_offline_queue response = {0};
>> +	u32 count = 0;
>> +	size_t rlen;
>> +	int ret;
>> +
>> +	/* Supported only on platforms with system controller */
>> +	if (!xe->info.has_sysctrl)
>> +		return;
>> +
>> +	prepare_sysctrl_command(&command,
>> XE_SYSCTRL_CMD_GET_OFFLINE_QUEUE, NULL, 0,
>> +				&response, sizeof(response));
>> +
>> +	do {
>> +		memset(&response, 0, sizeof(response));
>> +
>> +		ret = xe_sysctrl_send_command(&xe->sc, &command, &rlen);
>> +		if (ret) {
>> +			xe_err(xe, "[RAS]: sysctrl: error ret %d\n", ret);
>> +			return;
>> +		}
>> +
>> +		if (rlen != sizeof(response)) {
>> +			xe_err(xe, "[RAS]: sysctrl: response size mismatch.
>> Expected %zu, got %zu\n",
>> +			       sizeof(response), rlen);
>> +			return;
>> +		}
>> +
>> +		/* TODO: Process pages and offline them */
>> +
>> +		count += response.pages_returned;
>> +		if (count > response.total_pages) {
>> +			xe_err(xe, "[RAS]: sysctrl: Pages returned exceed total
>> pages %u, returned %u\n",
>> +			       response.total_pages, count);
>> +			return;
>> +		}
>> +	} while (response.additional_data);
>> +}
>> +
>> +static void get_offlined_list(struct xe_device *xe) {
>> +	struct xe_sysctrl_mailbox_command command = {0};
>> +	struct xe_ras_page_offline_list response = {0};
>> +	int ret, count = 0;
>> +	size_t rlen;
>> +
>> +	/* Supported only on platforms with system controller */
>> +	if (!xe->info.has_sysctrl)
>> +		return;
>> +
>> +	prepare_sysctrl_command(&command,
>> XE_SYSCTRL_CMD_GET_OFFLINE_LIST, NULL, 0,
>> +				&response, sizeof(response));
>> +
>> +	do {
>> +		memset(&response, 0, sizeof(response));
>> +
>> +		ret = xe_sysctrl_send_command(&xe->sc, &command, &rlen);
>> +		if (ret) {
>> +			xe_err(xe, "[RAS]: sysctrl: error ret %d\n", ret);
>> +			return;
>> +		}
>> +
>> +		if (rlen != sizeof(response)) {
>> +			xe_err(xe, "[RAS]: sysctrl: response size mismatch.
>> Expected %zu, got %zu\n",
>> +			       sizeof(response), rlen);
>> +			return;
>> +		}
>> +
>> +		/* TODO: Process pages and offline them */
>> +
>> +		count += response.pages_returned;
>> +		if (count > response.total_pages) {
>> +			xe_err(xe, "[RAS]: sysctrl: Pages returned exceed total
>> pages %u, returned %u\n",
>> +			       response.total_pages, count);
>> +			return;
>> +		}
>> +	} while (response.additional_data);
>> +}
>> +
>>   #ifdef CONFIG_PCIEAER
>>   static void aer_unmask_and_downgrade_internal_error(struct xe_device *xe)
>> { @@ -394,4 +475,8 @@ void xe_ras_init(struct xe_device *xe)  #ifdef
>> CONFIG_PCIEAER
>>   	aer_unmask_and_downgrade_internal_error(xe);
>>   #endif
>> +
>> +	/* Get any pages that need to be offlined from firmware and reserve
>> them */
> Should be executed only on CRI, right?

We do have has_sysctrl check in the start of the ras_init function.
This check is also there in every function too.

Riana

>
> Tejas
>> +	get_offlined_list(xe);
>> +	get_queued_pages(xe);
>>   }
>> diff --git a/drivers/gpu/drm/xe/xe_ras_types.h
>> b/drivers/gpu/drm/xe/xe_ras_types.h
>> index d76310866da5..1a5de6cd16a1 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_ERROR_ARR		3
>>   #define XE_RAS_MAX_ERROR_DETAILS	16
>> +#define XE_RAS_NUM_PAGES		25
>>   #define XE_RAS_IEH_PUNIT_ERROR		BIT(1)
>>
>>   /**
>> @@ -297,4 +298,42 @@ struct xe_ras_page_offline_response {
>>   	/** @reserved: Reserved for future use */
>>   	u32 reserved;
>>   } __packed;
>> +
>> +/**
>> + * struct xe_ras_page_offline_list - Response from get offline list
>> +command
>> + *
>> + * This structure provides the details of offlined pages from flash.
>> + */
>> +struct xe_ras_page_offline_list {
>> +	/** @max_entries: Total no of pages that can be stored in flash */
>> +	u32 max_entries;
>> +	/** @total_pages: Total number of permanently offlined pages */
>> +	u32 total_pages;
>> +	/** @pages_returned: Number of pages returned in this response */
>> +	u32 pages_returned;
>> +	/** @page_addresses: Array of permanently offlined 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_page_offline_queue - Response from get offline queue
>> +command
>> + *
>> + * This structure provides the details of queued offlined pages from firmware.
>> + */
>> +struct xe_ras_page_offline_queue {
>> +	/** @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 offlined 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;
>>   #endif
>> diff --git a/drivers/gpu/drm/xe/xe_sysctrl_mailbox_types.h
>> b/drivers/gpu/drm/xe/xe_sysctrl_mailbox_types.h
>> index 2cafa8a14cc3..b6139ac0eaca 100644
>> --- a/drivers/gpu/drm/xe/xe_sysctrl_mailbox_types.h
>> +++ b/drivers/gpu/drm/xe/xe_sysctrl_mailbox_types.h
>> @@ -15,10 +15,14 @@
>>    *
>>    * @XE_SYSCTRL_CMD_GET_SOC_ERROR: Get basic error information
>>    * @XE_SYSCTRL_CMD_PAGE_OFFLINE: Instruct firmware to offline/decline a
>> page
>> + * @XE_SYSCTRL_CMD_GET_OFFLINE_LIST: Get list of all offlined pages
>> + from flash
>> + * @XE_SYSCTRL_CMD_GET_OFFLINE_QUEUE: Get list of offlined queued
>> pages
>> + from firmware
>>    */
>>   enum xe_sysctrl_mailbox_command_id {
>>   	XE_SYSCTRL_CMD_GET_SOC_ERROR		= 0x01,
>> -	XE_SYSCTRL_CMD_PAGE_OFFLINE		= 0x08
>> +	XE_SYSCTRL_CMD_PAGE_OFFLINE		= 0x08,
>> +	XE_SYSCTRL_CMD_GET_OFFLINE_LIST		= 0x09,
>> +	XE_SYSCTRL_CMD_GET_OFFLINE_QUEUE	= 0x0A
>>   };
>>
>>   enum xe_sysctrl_group {
>> --
>> 2.47.1