From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from gabe.freedesktop.org (gabe.freedesktop.org [131.252.210.177]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.lore.kernel.org (Postfix) with ESMTPS id 26580CCFA1A for ; Fri, 7 Nov 2025 18:13:31 +0000 (UTC) Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id C77BC10EB82; Fri, 7 Nov 2025 18:13:30 +0000 (UTC) Authentication-Results: gabe.freedesktop.org; dkim=pass (2048-bit key; unprotected) header.d=intel.com header.i=@intel.com header.b="MSHSnNeo"; dkim-atps=neutral Received: from mgamail.intel.com (mgamail.intel.com [192.198.163.14]) by gabe.freedesktop.org (Postfix) with ESMTPS id 31B7F10EB78 for ; Fri, 7 Nov 2025 18:13:29 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1762539209; x=1794075209; h=from:to:cc:subject:date:message-id:in-reply-to: references:mime-version:content-transfer-encoding; bh=y2dNeE9uhAWPLmzoYu40GOgpAFdEpWxJ0bR2W1GPmSg=; b=MSHSnNeoKCtaZCsjW7q11NizLj/4AvWRPfBZmFjxTqHFLRbIVY/V69wr FOaNsGwk51rycOJoia8+Wv9OwGr7z9sMQGD9XCsUX4RSxtrr93PWXe6tT /iao9iCzY8ikaQTquB0WlyIu5eYOJuO40IDJHe0jc1e6EVGBmV6lLoYQF dwPoxfTVZXm9EIAolFYQKsm2sCtoTYESnlLBedX0S/GAXC6F2s0Ek1taW 1yamOgkL3l0HyZX4ImXlfyMbVeAlCi++1doVee3Xfq+/5ByPy4WM74W+G Gmb0kf025OSR88lqOkMvLVpuiRL1/0Y2uX1S3l8EfxmPKuYc8531NTVJn w==; X-CSE-ConnectionGUID: l1DxEoIvRaGtdYrAWqbvzg== X-CSE-MsgGUID: AkjMJPLkSwSMkCFiqvA6EQ== X-IronPort-AV: E=McAfee;i="6800,10657,11606"; a="64730184" X-IronPort-AV: E=Sophos;i="6.19,287,1754982000"; d="scan'208";a="64730184" Received: from orviesa005.jf.intel.com ([10.64.159.145]) by fmvoesa108.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 07 Nov 2025 10:13:29 -0800 X-CSE-ConnectionGUID: LusWqpDyRhK0ueSiHtsLmQ== X-CSE-MsgGUID: 88PRti4IT0OnRFdYwG+WNA== X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="6.19,287,1754982000"; d="scan'208";a="193271157" Received: from mdroper-desk1.fm.intel.com ([10.1.39.133]) by orviesa005-auth.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 07 Nov 2025 10:13:28 -0800 From: Matt Roper To: intel-xe@lists.freedesktop.org Cc: matthew.d.roper@intel.com Subject: [PATCH 04/33] drm/xe/forcewake: Create dedicated type for forcewake references Date: Fri, 7 Nov 2025 10:13:20 -0800 Message-ID: <20251107181315.631642-39-matthew.d.roper@intel.com> X-Mailer: git-send-email 2.51.1 In-Reply-To: <20251107181315.631642-35-matthew.d.roper@intel.com> References: <20251107181315.631642-35-matthew.d.roper@intel.com> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-BeenThere: intel-xe@lists.freedesktop.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: Intel Xe graphics driver List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: intel-xe-bounces@lists.freedesktop.org Sender: "Intel-xe" xe_force_wake_get() currently returns an integer mask of power domains that were successfully awoken; both this mask and a pointer to the force wake collection must be passed to xe_force_wake_put() to release the wake reference. Create a dedicated structure type to hold both the mask and the collection pointer. While this change does little on its own, it will make it easier for us to add scope-based cleanup of forcewake in the future. FIXME: For ease of review, this patch contains only the manual changes to add the structure and change the get/put function definitions; it does not build on its own since the rest of the driver is still trying to call the get/put functions with the old signature. The next patch contains the coccinelle-generated changes necessary elsewhere in the driver to adapt to the new interface. The two patches will be squashed together when applied, remain separate for now to help reviewers. Signed-off-by: Matt Roper --- drivers/gpu/drm/xe/xe_debugfs.c | 23 ++++++++++++++++++----- drivers/gpu/drm/xe/xe_drm_client.c | 4 ++-- drivers/gpu/drm/xe/xe_eu_stall.c | 2 +- drivers/gpu/drm/xe/xe_force_wake.c | 19 ++++++++++++------- drivers/gpu/drm/xe/xe_force_wake.h | 11 ++++++----- drivers/gpu/drm/xe/xe_force_wake_types.h | 15 +++++++++++++++ drivers/gpu/drm/xe/xe_oa_types.h | 2 +- drivers/gpu/drm/xe/xe_pmu.c | 6 ++---- 8 files changed, 57 insertions(+), 25 deletions(-) diff --git a/drivers/gpu/drm/xe/xe_debugfs.c b/drivers/gpu/drm/xe/xe_debugfs.c index e91da9589c5f..2d858110922b 100644 --- a/drivers/gpu/drm/xe/xe_debugfs.c +++ b/drivers/gpu/drm/xe/xe_debugfs.c @@ -198,7 +198,7 @@ static int forcewake_open(struct inode *inode, struct file *file) struct xe_device *xe = inode->i_private; struct xe_gt *gt; u8 id, last_gt; - unsigned int fw_ref; + struct xe_force_wake_ref fw_ref; xe_pm_runtime_get(xe); for_each_gt(gt, xe, id) { @@ -213,10 +213,19 @@ static int forcewake_open(struct inode *inode, struct file *file) err_fw_get: for_each_gt(gt, xe, id) { + struct xe_force_wake_ref all_fw_ref; + + /* + * A bit of a hack since we didn't save the actual forcewake + * reference above. + */ + all_fw_ref.fw = gt_to_fw(gt); + all_fw_ref.domains = XE_FORCEWAKE_ALL; + if (id < last_gt) - xe_force_wake_put(gt_to_fw(gt), XE_FORCEWAKE_ALL); + xe_force_wake_put(all_fw_ref); else if (id == last_gt) - xe_force_wake_put(gt_to_fw(gt), fw_ref); + xe_force_wake_put(fw_ref); else break; } @@ -228,11 +237,15 @@ static int forcewake_open(struct inode *inode, struct file *file) static int forcewake_release(struct inode *inode, struct file *file) { struct xe_device *xe = inode->i_private; + struct xe_force_wake_ref all_fw_ref; struct xe_gt *gt; u8 id; - for_each_gt(gt, xe, id) - xe_force_wake_put(gt_to_fw(gt), XE_FORCEWAKE_ALL); + all_fw_ref.domains = XE_FORCEWAKE_ALL; + for_each_gt(gt, xe, id) { + all_fw_ref.fw = gt_to_fw(gt); + xe_force_wake_put(all_fw_ref); + } xe_pm_runtime_put(xe); return 0; diff --git a/drivers/gpu/drm/xe/xe_drm_client.c b/drivers/gpu/drm/xe/xe_drm_client.c index f931ff9b1ec0..60a6ea7c88e4 100644 --- a/drivers/gpu/drm/xe/xe_drm_client.c +++ b/drivers/gpu/drm/xe/xe_drm_client.c @@ -287,7 +287,7 @@ static struct xe_hw_engine *any_engine(struct xe_device *xe) static bool force_wake_get_any_engine(struct xe_device *xe, struct xe_hw_engine **phwe, - unsigned int *pfw_ref) + struct xe_force_wake_ref *pfw_ref) { enum xe_force_wake_domains domain; unsigned int fw_ref; @@ -322,7 +322,7 @@ static void show_run_ticks(struct drm_printer *p, struct drm_file *file) struct xe_hw_engine *hwe; struct xe_exec_queue *q; u64 gpu_timestamp; - unsigned int fw_ref; + struct xe_force_wake_ref fw_ref; /* * RING_TIMESTAMP registers are inaccessible in VF mode. diff --git a/drivers/gpu/drm/xe/xe_eu_stall.c b/drivers/gpu/drm/xe/xe_eu_stall.c index 97dfb7945b7a..8b3da9ae6888 100644 --- a/drivers/gpu/drm/xe/xe_eu_stall.c +++ b/drivers/gpu/drm/xe/xe_eu_stall.c @@ -49,7 +49,7 @@ struct xe_eu_stall_data_stream { wait_queue_head_t poll_wq; size_t data_record_size; size_t per_xecore_buf_size; - unsigned int fw_ref; + struct xe_force_wake_ref fw_ref; struct xe_gt *gt; struct xe_bo *bo; diff --git a/drivers/gpu/drm/xe/xe_force_wake.c b/drivers/gpu/drm/xe/xe_force_wake.c index c59a9b330697..2e675536b36e 100644 --- a/drivers/gpu/drm/xe/xe_force_wake.c +++ b/drivers/gpu/drm/xe/xe_force_wake.c @@ -169,11 +169,12 @@ static int domain_sleep_wait(struct xe_gt *gt, * Return: opaque reference to woken domains or zero if none of requested * domains were awake. */ -unsigned int __must_check xe_force_wake_get(struct xe_force_wake *fw, - enum xe_force_wake_domains domains) +struct xe_force_wake_ref __must_check xe_force_wake_get(struct xe_force_wake *fw, + enum xe_force_wake_domains domains) { struct xe_gt *gt = fw->gt; struct xe_force_wake_domain *domain; + struct xe_force_wake_ref fw_ref; unsigned int ref_incr = 0, awake_rqst = 0, awake_failed = 0; unsigned int tmp, ref_rqst; unsigned long flags; @@ -208,7 +209,10 @@ unsigned int __must_check xe_force_wake_get(struct xe_force_wake *fw, if (domains == XE_FORCEWAKE_ALL && ref_incr == fw->initialized_domains) ref_incr |= XE_FORCEWAKE_ALL; - return ref_incr; + fw_ref.fw = fw; + fw_ref.domains = ref_incr; + + return fw_ref; } /** @@ -221,8 +225,9 @@ unsigned int __must_check xe_force_wake_get(struct xe_force_wake *fw, * and waits for acknowledgment for domain to sleep within 50 milisec timeout. * Warns in case of timeout of ack from domain. */ -void xe_force_wake_put(struct xe_force_wake *fw, unsigned int fw_ref) +void xe_force_wake_put(struct xe_force_wake_ref fw_ref) { + struct xe_force_wake *fw = fw_ref.fw; struct xe_gt *gt = fw->gt; struct xe_force_wake_domain *domain; unsigned int tmp, sleep = 0; @@ -233,14 +238,14 @@ void xe_force_wake_put(struct xe_force_wake *fw, unsigned int fw_ref) * Avoid unnecessary lock and unlock when the function is called * in error path of individual domains. */ - if (!fw_ref) + if (!fw_ref.domains) return; if (xe_force_wake_ref_has_domain(fw_ref, XE_FORCEWAKE_ALL)) - fw_ref = fw->initialized_domains; + fw_ref.domains = fw->initialized_domains; spin_lock_irqsave(&fw->lock, flags); - for_each_fw_domain_masked(domain, fw_ref, fw, tmp) { + for_each_fw_domain_masked(domain, fw_ref.domains, fw, tmp) { xe_gt_assert(gt, domain->ref); if (!--domain->ref) { diff --git a/drivers/gpu/drm/xe/xe_force_wake.h b/drivers/gpu/drm/xe/xe_force_wake.h index 0e3e84bfa51c..86e9bca7cac9 100644 --- a/drivers/gpu/drm/xe/xe_force_wake.h +++ b/drivers/gpu/drm/xe/xe_force_wake.h @@ -15,9 +15,9 @@ void xe_force_wake_init_gt(struct xe_gt *gt, struct xe_force_wake *fw); void xe_force_wake_init_engines(struct xe_gt *gt, struct xe_force_wake *fw); -unsigned int __must_check xe_force_wake_get(struct xe_force_wake *fw, - enum xe_force_wake_domains domains); -void xe_force_wake_put(struct xe_force_wake *fw, unsigned int fw_ref); +struct xe_force_wake_ref __must_check xe_force_wake_get(struct xe_force_wake *fw, + enum xe_force_wake_domains domains); +void xe_force_wake_put(struct xe_force_wake_ref fw_ref); static inline int xe_force_wake_ref(struct xe_force_wake *fw, @@ -56,9 +56,10 @@ xe_force_wake_assert_held(struct xe_force_wake *fw, * Return: true if domain is refcounted. */ static inline bool -xe_force_wake_ref_has_domain(unsigned int fw_ref, enum xe_force_wake_domains domain) +xe_force_wake_ref_has_domain(struct xe_force_wake_ref fw_ref, + enum xe_force_wake_domains domain) { - return fw_ref & domain; + return fw_ref.domains & domain; } #endif diff --git a/drivers/gpu/drm/xe/xe_force_wake_types.h b/drivers/gpu/drm/xe/xe_force_wake_types.h index 9cfa28faf7bc..26df4adba4c5 100644 --- a/drivers/gpu/drm/xe/xe_force_wake_types.h +++ b/drivers/gpu/drm/xe/xe_force_wake_types.h @@ -107,4 +107,19 @@ struct xe_force_wake { struct xe_force_wake_domain domains[XE_FW_DOMAIN_ID_COUNT]; }; +/** + * struct xe_force_wake_ref - Xe force wake reference + * + * Represents a wakeref for a subset of the power domains belonging to an + * xe_force_wake collection. Returned by xe_force_wake_get() and passed + * to xe_force_wake_put(). + */ +struct xe_force_wake_ref { + /** @fw: back pointer to force wake collection */ + struct xe_force_wake *fw; + + /** @domains: mask of individual domains held by this reference */ + unsigned int domains; +}; + #endif diff --git a/drivers/gpu/drm/xe/xe_oa_types.h b/drivers/gpu/drm/xe/xe_oa_types.h index cf080f412189..84bd5018d0f3 100644 --- a/drivers/gpu/drm/xe/xe_oa_types.h +++ b/drivers/gpu/drm/xe/xe_oa_types.h @@ -266,6 +266,6 @@ struct xe_oa_stream { struct xe_sync_entry *syncs; /** @fw_ref: Forcewake reference */ - unsigned int fw_ref; + struct xe_force_wake_ref fw_ref; }; #endif diff --git a/drivers/gpu/drm/xe/xe_pmu.c b/drivers/gpu/drm/xe/xe_pmu.c index c63335eb69e5..dbd95327f9fc 100644 --- a/drivers/gpu/drm/xe/xe_pmu.c +++ b/drivers/gpu/drm/xe/xe_pmu.c @@ -214,12 +214,10 @@ static bool event_param_valid(struct perf_event *event) static void xe_pmu_event_destroy(struct perf_event *event) { struct xe_device *xe = container_of(event->pmu, typeof(*xe), pmu.base); - struct xe_gt *gt; - unsigned int *fw_ref = event->pmu_private; + struct xe_force_wake_ref *fw_ref = event->pmu_private; if (fw_ref) { - gt = xe_device_get_gt(xe, config_to_gt_id(event->attr.config)); - xe_force_wake_put(gt_to_fw(gt), *fw_ref); + xe_force_wake_put(*fw_ref); kfree(fw_ref); event->pmu_private = NULL; } -- 2.51.1