From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from gabe.freedesktop.org (gabe.freedesktop.org [131.252.210.177]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature ECDSA (prime256v1) server-digest SHA256) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 0B84A606A32A for ; Wed, 18 Nov 2026 10:02:50 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=intel.com header.i=@intel.com header.a=rsa-sha256 header.s=Intel header.b=n/jiPnTp Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id C3F1810E511; Wed, 18 Nov 2026 10:02:44 +0000 (UTC) Received: from mgamail.intel.com (mgamail.intel.com [198.175.65.16]) by gabe.freedesktop.org (Postfix) with ESMTPS id ECE9010E511 for ; Wed, 18 Nov 2026 10:02:38 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1794996159; x=1826532159; h=date:from:to:cc:subject:message-id:references: mime-version:in-reply-to; bh=i3AXuSy+bLFD+Y9hPVMlhleXH4MnuffkHWzH/15XMeM=; b=n/jiPnTp1x7koMrKZWySN6fTvL9pL1HVnxrg02oCc5qOZ5kefVAzM0Cc /dH/PwIpMC3YiH1fvXumiEk7LMS/JFMZBx38BU/giG6NyairigAWHNqJH GHPomwO/tMVykwjf6vQneMwy0bdopshKlj3Bu6GuNXnG6Xc8NV3uGaXSB KLEJHbg59Zmxt0xNRln2Y7ub+Lnl92CileIaUllFBCYhuf7dfR43uEmfU 6taIwUodjIagBRh1pHTzayhairXfEk22C+HDB+VKHvFqwt6krV2ds5hvt 9vLr/9ztMJ7ipS0c3BJY6GzbAdyMpHUHRccKJ/PuhVePjQnWYHfpQpQii A==; X-CSE-ConnectionGUID: v/LEQa1VRTm0TLey8D9cTQ== X-CSE-MsgGUID: OqQP+IsfSsSLKUdcC0odwQ== X-IronPort-AV: E=McAfee;i="6800,10657,11969"; a="3791937" X-IronPort-AV: E=Sophos;i="6.27,219,1787036400"; d="scan'208";a="3791937" Received: from orviesa003.jf.intel.com ([10.64.159.143]) by orvoesa108.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 18 Nov 2026 02:02:38 -0800 X-CSE-ConnectionGUID: gTVjSLNGTgeJQuhkbHRkGg== X-CSE-MsgGUID: bjg1EX1tThG+YuU+lyKfbg== X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="6.27,219,1787036400"; d="scan'208";a="10015164" Received: from black.igk.intel.com ([10.91.253.5]) by orviesa003.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 18 Nov 2026 02:02:36 -0800 Date: Wed, 18 Nov 2026 11:02:33 +0100 From: Raag Jadav To: Badal Nilawar Cc: intel-xe@lists.freedesktop.org, anshuman.gupta@intel.com, rodrigo.vivi@intel.com, daniele.ceraolospurio@intel.com, riana.tauro@intel.com, mallesh.koujalagi@intel.com, aravind.iddamsetty@intel.com, himal.prasad.ghimiray@intel.com, arvind.yadav@intel.com, syed.abdul.muqthyar.ahmed@intel.com, nitin.r.gote@intel.com, matthew.d.roper@intel.com Subject: Re: [PATCH v7 2/3] drm/xe/xe_ras: Add support to retrieve info queue data Message-ID: References: <20261117072118.4051706-5-badal.nilawar@intel.com> <20261117072118.4051706-7-badal.nilawar@intel.com> X-Archived-By: MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20261117072118.4051706-7-badal.nilawar@intel.com> X-BeenThere: intel-xe@lists.freedesktop.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: Intel Xe graphics driver List-Archive: List-Post: List-Help: List-Subscribe: , Sender: "Intel-xe" X-Spam-Status: No, score=-15.8 required=3.0 tests=BAYES_00,DKIMWL_WL_HIGH, DKIM_SIGNED,DKIM_VALID,DKIM_VALID_AU,HEADER_FROM_DIFFERENT_DOMAINS, INCLUDES_CR_TRAILER,INCLUDES_PATCH,MAILING_LIST_MULTI,SPF_HELO_NONE, SPF_PASS autolearn=ham autolearn_force=no version=4.0.1 X-Spam-Checker-Version: SpamAssassin 4.0.1 (2024-03-26) on aws-us-west-2-korg-listserv-alma10-1 On Tue, Nov 17, 2026 at 12:51:21PM +0530, Badal Nilawar wrote: > Retrieve the RAS info queue data, in multiple chunks, and assemble it into > a flat buffer. > > Signed-off-by: Badal Nilawar > --- > v6: > - Replace destination buffer checks with xe_assert() (Raag) > - Extract get_first_chunk() and get_remaining_chunks() helpers (Raag) > v7: > - Move RAS has_info_queue check from CPER to RAS (Raag) > - Also check that the reported total size is consistent across all chunks > - Limit info queue fetches to 2x XE_SYSCTRL_FLOOD_LIMIT (Raag) > --- > drivers/gpu/drm/xe/xe_ras.c | 184 ++++++++++++++++++ > drivers/gpu/drm/xe/xe_ras.h | 6 +- > drivers/gpu/drm/xe/xe_ras_types.h | 91 ++++++++- > drivers/gpu/drm/xe/xe_sysctrl_mailbox_types.h | 2 + > 4 files changed, 280 insertions(+), 3 deletions(-) > > diff --git a/drivers/gpu/drm/xe/xe_ras.c b/drivers/gpu/drm/xe/xe_ras.c > index e8984c966876..0a5bac375d8c 100644 > --- a/drivers/gpu/drm/xe/xe_ras.c > +++ b/drivers/gpu/drm/xe/xe_ras.c > @@ -1123,6 +1123,190 @@ int xe_ras_set_threshold(struct xe_device *xe, u8 severity, u8 component, u32 th > return 0; > } > > +static int get_info_queue(struct xe_device *xe, > + struct xe_ras_get_info_queue_request *request, > + struct xe_ras_get_info_queue_response *response) > +{ > + struct xe_sysctrl_mailbox_command command = {0}; > + const struct xe_ras_info_queue_header *hdr; > + size_t rlen; > + int ret; > + > + xe_sysctrl_create_command(&command, XE_SYSCTRL_GROUP_GFSP, XE_SYSCTRL_CMD_GET_INFO_QUEUE, > + request, sizeof(*request), response, sizeof(*response)); > + > + ret = xe_sysctrl_send_command(&xe->sc, &command, &rlen); > + if (ret) { > + xe_err(xe, "sysctrl: failed to get info queue %d\n", ret); > + return ret; > + } > + > + if (rlen != sizeof(*response)) { > + xe_err(xe, "sysctrl: unexpected get info queue response length %zu (expected %zu)\n", > + rlen, sizeof(*response)); > + return -EIO; > + } > + > + ret = ras_status_to_errno(response->status); > + if (ret) { > + xe_err(xe, "sysctrl: get info queue command failed with status %#x\n", > + response->status); > + return ret; > + } > + > + hdr = &response->qdata_response.header; > + xe_dbg(xe, "[RAS] get info queue: total_size=%u chunk_offset=%u chunk_size=%u flags=%#x num_headers=%u\n", Please make this consistent with other similar logs (total=%u offset=%u ...) > + hdr->total_size, hdr->chunk_offset, hdr->chunk_size, hdr->flags, hdr->num_headers); > + > + return 0; > +} > + > +static bool chunk_is_valid(const struct xe_ras_info_queue_header *hdr, > + u32 expected_offset, u32 total_size, u32 dest_size) > +{ > + u32 end; > + > + return hdr->total_size == total_size && total_size && > + hdr->chunk_offset == expected_offset && > + hdr->chunk_size && hdr->chunk_size <= XE_RAS_INFO_QUEUE_MAX_CHUNK_SIZE && > + !check_add_overflow(hdr->chunk_offset, hdr->chunk_size, &end) && > + end <= min_t(u32, total_size, XE_RAS_INFO_QUEUE_MAX_TOTAL_SIZE) && > + end <= dest_size; > +} > + > +static u32 get_first_chunk(struct xe_device *xe, > + const struct xe_ras_get_counter_response *counter_resp, > + u8 *dest_buf, u32 dest_size, u32 *flags) > +{ > + const struct xe_ras_info_queue_header *hdr = &counter_resp->first.header; > + > + if (!chunk_is_valid(hdr, 0, hdr->total_size, dest_size)) { > + xe_warn(xe, "[RAS]: invalid first info queue chunk (total=%u offset=%u size=%u, expected offset=0)\n", > + hdr->total_size, hdr->chunk_offset, hdr->chunk_size); > + return 0; > + } > + > + memcpy(dest_buf, counter_resp->first.data, hdr->chunk_size); > + *flags = hdr->flags; > + > + return hdr->chunk_size; > +} > + > +static u32 get_remaining_chunks(struct xe_device *xe, > + const struct xe_ras_get_counter_response *counter_resp, > + u8 *dest_buf, u32 dest_size, u32 offset, u32 flags) > +{ > + u32 requests_left = XE_SYSCTRL_FLOOD_LIMIT * 2; > + int ret; > + > + xe_assert(xe, offset < dest_size); This should be inside the loop and done for each iteration... > + while ((flags & XE_RAS_INFO_QUEUE_FLAG_MORE_DATA) && offset < dest_size) { ... so we may also want to remove it from while() condition. And don't worry about infinite loop, this is exactly why we have requests_left in place. > + struct xe_ras_info_queue_data_request *qdata_request; > + struct xe_ras_get_info_queue_response response = {}; > + struct xe_ras_get_info_queue_request request = {}; > + struct xe_ras_info_queue_header *hdr; > + > + if (!requests_left) { This can be at the end of the loop and can be consolidated into if (!--requests_left) > + xe_warn(xe, "[RAS] info queue drain stopped after %u requests with data remaining\n", 'requests' is arbitrary and not useful in the logs. Rather, "[RAS] info queue flooding, aborted at offset=%u\n" > + XE_SYSCTRL_FLOOD_LIMIT * 2); > + break; > + } > + requests_left--; > + > + qdata_request = &request.qdata_request; > + qdata_request->offset = offset; > + qdata_request->size = min(XE_RAS_INFO_QUEUE_MAX_CHUNK_SIZE, min_t() > + dest_size - offset); Can be less lines? > + qdata_request->session_id = counter_resp->counter; > + > + ret = get_info_queue(xe, &request, &response); > + if (ret) { > + xe_err(xe, "[RAS]: info queue drain aborted: fetch at offset=%u, ret = %d\n", > + offset, ret); > + return offset; > + } > + > + hdr = &response.qdata_response.header; > + > + if (!chunk_is_valid(hdr, offset, counter_resp->first.header.total_size, > + dest_size)) { > + xe_warn(xe, "[RAS]: invalid info queue chunk (total=%u offset=%u size=%u, expected offset=%u)\n", > + hdr->total_size, hdr->chunk_offset, hdr->chunk_size, offset); > + return offset; > + } > + > + memcpy(dest_buf + offset, response.qdata_response.data, hdr->chunk_size); > + offset += hdr->chunk_size; > + flags = hdr->flags; > + } Let's also sanity check on loop exist. if (offset != dest_size) { xe_warn(xe, "[RAS] info queue drain incomplete, missed %u size\n", dest_size - offset); return offset; } > + return min(offset, counter_resp->first.header.total_size); min_t() > +} > + > +/** > + * xe_ras_get_info_queue_size() - Get the total size of the RAS info queue > + * @response: Counter response carrying the first info queue chunk > + * > + * Return: Info queue size in bytes, capped at the supported maximum, or zero > + * if no usable info queue is present. > + */ > +u32 xe_ras_get_info_queue_size(struct xe_device *xe, > + const struct xe_ras_get_counter_response *response) > +{ > + u32 total_size; > + > + if (!response->has_info_queue) > + return 0; > + > + total_size = response->first.header.total_size; > + if (!total_size) { > + xe_warn(xe, "[RAS] info queue reports zero size\n"); > + return 0; > + } > + > + if (total_size > XE_RAS_INFO_QUEUE_MAX_TOTAL_SIZE) { > + xe_warn(xe, "[RAS] info queue size %u exceeds max size %u, truncating\n", > + total_size, XE_RAS_INFO_QUEUE_MAX_TOTAL_SIZE); > + total_size = XE_RAS_INFO_QUEUE_MAX_TOTAL_SIZE; > + } > + return total_size; > +} > + > +/** > + * xe_ras_drain_info_queue() - Drain the full RAS info queue into a flat buffer > + * @xe: xe device > + * @response: counter response carrying the first embedded chunk > + * @dest_buf: destination buffer supplied by the caller > + * @dest_size: size of @dest_buf in bytes > + * > + * Copies the first chunk already embedded in @response, then loops > + * issuing GET_INFO_QUEUE to fetch any remaining chunks until the queue > + * signals no more data or a transport/bounds error is encountered. On > + * transport or bounds errors the function stops and returns whatever has > + * been assembled so far. > + * > + * Returns: Number of valid bytes written into @dest_buf. > + */ > +u32 xe_ras_drain_info_queue(struct xe_device *xe, > + const struct xe_ras_get_counter_response *response, > + u8 *dest_buf, u32 dest_size) > +{ > + u32 flags, offset; > + > + if (!response->has_info_queue) Now that we have xe_ras_get_info_queue_size() in place, we shouldn't be at this point and this can be xe_assert(). > + return 0; > + > + xe_assert(xe, dest_buf); > + xe_assert(xe, dest_size); > + > + offset = get_first_chunk(xe, response, dest_buf, dest_size, &flags); > + if (!offset) > + return 0; I believe this should be if (!offset || offset == dest_size) return offset; > + return get_remaining_chunks(xe, response, dest_buf, dest_size, offset, flags); > +} > + > static ssize_t gpu_health_show(struct device *dev, struct device_attribute *attr, char *buf) > { > struct xe_ras_get_health_response response = {0}; > diff --git a/drivers/gpu/drm/xe/xe_ras.h b/drivers/gpu/drm/xe/xe_ras.h > index f7b587ac84e3..bca460b5f954 100644 > --- a/drivers/gpu/drm/xe/xe_ras.h > +++ b/drivers/gpu/drm/xe/xe_ras.h > @@ -25,7 +25,11 @@ bool xe_ras_get_disable_vram_page_offline(struct xe_device *xe); > u32 xe_ras_get_max_pages(struct xe_device *xe); > int xe_ras_get_counter_response(struct xe_device *xe, const struct xe_ras_error_class *counter, > struct xe_ras_get_counter_response *response); > +u32 xe_ras_get_info_queue_size(struct xe_device *xe, > + const struct xe_ras_get_counter_response *response); > bool xe_ras_is_aer_internal_error(struct xe_device *xe); > void xe_ras_set_aer_internal_error(struct xe_device *xe); > - Stray churn. Please avoid this. Raag > +u32 xe_ras_drain_info_queue(struct xe_device *xe, > + const struct xe_ras_get_counter_response *response, > + u8 *dest_buf, u32 dest_size); > #endif > diff --git a/drivers/gpu/drm/xe/xe_ras_types.h b/drivers/gpu/drm/xe/xe_ras_types.h > index 0ff5803d495b..c6480a7035f6 100644 > --- a/drivers/gpu/drm/xe/xe_ras_types.h > +++ b/drivers/gpu/drm/xe/xe_ras_types.h > @@ -19,6 +19,10 @@ > #define XE_RAS_MEMORY_POISON BIT(2) > #define XE_RAS_MEMORY_DATA_PARITY BIT(5) > > +#define XE_RAS_INFO_QUEUE_MAX_CHUNK_SIZE 200 > +#define XE_RAS_INFO_QUEUE_MAX_TOTAL_SIZE 5120 > +#define XE_RAS_INFO_QUEUE_FLAG_MORE_DATA 0x02 > + > /** > * enum xe_ras_page_action - Page offline actions for page offline request > * > @@ -110,6 +114,83 @@ struct xe_ras_threshold_crossed { > struct xe_ras_error_class counters[XE_RAS_NUM_COUNTERS]; > } __packed; > > +/** > + * struct xe_ras_info_queue_header - Metadata for large info queue data transfers > + * > + * Provides chunk metadata for commands that support extended info queue > + * functionality. Used when the total data exceeds a single mailbox response. > + */ > +struct xe_ras_info_queue_header { > + /** @total_size: Total size of the complete info queue data in bytes */ > + u32 total_size; > + /** @chunk_offset: Offset of this chunk within the total data in bytes */ > + u32 chunk_offset; > + /** @chunk_size: Size of the data in this chunk in bytes */ > + u32 chunk_size; > + /** @reserved0: Reserved for future use */ > + u32 reserved0; > + /** @flags: Info queue control flags (XE_RAS_INFO_QUEUE_FLAG_*) */ > + u32 flags:8; > + /** @reserved1: Reserved for future use */ > + u32 reserved1:4; > + /** @num_headers: Number of detailed counter headers at start of the data */ > + u32 num_headers:5; > + /** @reserved2: Reserved for future use */ > + u32 reserved2:15; > + /** @reserved3: Reserved for future use */ > + u32 reserved3; > +} __packed; > + > +/** > + * struct xe_ras_info_queue_data_request - Request for a specific chunk of info queue data > + * > + * Allows the driver to request continuation of large info queue transfers > + * by specifying an offset and size within the full data set. > + */ > +struct xe_ras_info_queue_data_request { > + /** @offset: Byte offset of the requested data chunk */ > + u32 offset; > + /** @size: Maximum size of the requested chunk in bytes */ > + u32 size; > + /** @session_id: Session ID to correlate multi-chunk transfers */ > + struct xe_ras_error_class session_id; > + /** @reserved: Reserved for future use */ > + u32 reserved; > +} __packed; > + > +/** > + * struct xe_ras_info_queue_data_response - Generic response for commands with info queues > + * > + * Standard response format for any command that returns an info queue > + * payload. May be embedded in a command-specific response structure. > + */ > +struct xe_ras_info_queue_data_response { > + /** @header: Info queue metadata for this chunk */ > + struct xe_ras_info_queue_header header; > + /** @data: Info queue data for this chunk */ > + u8 data[XE_RAS_INFO_QUEUE_MAX_CHUNK_SIZE]; > +} __packed; > + > +/** > + * struct xe_ras_get_info_queue_request - Request for XE_SYSCTRL_CMD_GET_INFO_QUEUE > + */ > +struct xe_ras_get_info_queue_request { > + /** @qdata_request: Info queue request parameters */ > + struct xe_ras_info_queue_data_request qdata_request; > + /** @reserved: Reserved for future use */ > + u32 reserved[16]; > +} __packed; > + > +/** > + * struct xe_ras_get_info_queue_response - Response for XE_SYSCTRL_CMD_GET_INFO_QUEUE > + */ > +struct xe_ras_get_info_queue_response { > + /** @status: Status of the retrieval operation */ > + u32 status; > + /** @qdata_response: Info queue data chunk */ > + struct xe_ras_info_queue_data_response qdata_response; > +} __packed; > + > /** > * struct xe_ras_get_counter_request - Request structure for get counter > */ > @@ -132,8 +213,14 @@ struct xe_ras_get_counter_response { > u64 timestamp; > /** @threshold: Threshold value for the counter */ > u32 threshold; > - /** @reserved: Reserved */ > - u32 reserved[57]; > + /** @reserved: Reserved for future use */ > + u32 reserved:9; > + /** @has_info_queue: Set if info queue is available */ > + u32 has_info_queue:1; > + /** @reserved1: Reserved for future use */ > + u32 reserved1:22; > + /** @first: First info queue chunk, if @has_info_queue is set */ > + struct xe_ras_info_queue_data_response first; > } __packed; > > /** > diff --git a/drivers/gpu/drm/xe/xe_sysctrl_mailbox_types.h b/drivers/gpu/drm/xe/xe_sysctrl_mailbox_types.h > index cdacceebca06..50b1d9abaf1f 100644 > --- a/drivers/gpu/drm/xe/xe_sysctrl_mailbox_types.h > +++ b/drivers/gpu/drm/xe/xe_sysctrl_mailbox_types.h > @@ -37,6 +37,7 @@ enum xe_sysctrl_group { > * @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 > + * @XE_SYSCTRL_CMD_GET_INFO_QUEUE: Retrieve a chunk of info queue data > */ > enum xe_sysctrl_gfsp_cmd { > XE_SYSCTRL_CMD_GET_SOC_ERROR = 0x01, > @@ -50,6 +51,7 @@ enum xe_sysctrl_gfsp_cmd { > XE_SYSCTRL_CMD_GET_OFFLINE_QUEUE = 0x0A, > XE_SYSCTRL_CMD_GET_HEALTH = 0x0B, > XE_SYSCTRL_CMD_SET_HEALTH = 0x0C, > + XE_SYSCTRL_CMD_GET_INFO_QUEUE = 0x0D, > }; > > /** > -- > 2.56.0 >