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 F1A0DC7EE23 for ; Mon, 15 May 2023 14:41:18 +0000 (UTC) Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id C90E110E20D; Mon, 15 May 2023 14:41:18 +0000 (UTC) Received: from mga02.intel.com (mga02.intel.com [134.134.136.20]) by gabe.freedesktop.org (Postfix) with ESMTPS id 6F67110E1F8 for ; Mon, 15 May 2023 14:41:11 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1684161671; x=1715697671; h=from:to:cc:subject:date:message-id:in-reply-to: references:mime-version:content-transfer-encoding; bh=1jy4nh0Dp9Q85Oh1rYv/P4VIll51YXXKlTsytusCN7M=; b=Z+4S38nUBD0XCSYLDxjzGVCJjqe26tLvACdpq/oN74JnnvmwiOJdwacy WgH/Wix+UtzjHbZDrQgyMZK8D+ohIDmIaBljV6ZnIglpaQOghMfrhkmsW vIuxMzcYWj37vm9TIpI1VzJriqFnfAPPfmI696jp+ZRsYy4OQ7rxYT1MH CyH77L5MDrwhWNB2Ev7fyMZNsXKJHXUWt2ZAxWpc8UVyX2sC7DVSwdjIx rFuFuonjScxrgcbIxSw050OrgIp95b1zejzq/os8mgJsIAiClw9Ge5URH gZ1CBe0s8eusCSrDozorEhLnIciD6GV8+x4aRtZTtuPNswktc2KBcjTyx Q==; X-IronPort-AV: E=McAfee;i="6600,9927,10711"; a="340569359" X-IronPort-AV: E=Sophos;i="5.99,276,1677571200"; d="scan'208";a="340569359" Received: from orsmga002.jf.intel.com ([10.7.209.21]) by orsmga101.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 15 May 2023 07:41:10 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=McAfee;i="6600,9927,10711"; a="700974579" X-IronPort-AV: E=Sophos;i="5.99,276,1677571200"; d="scan'208";a="700974579" Received: from kaelanhx-mobl1.ger.corp.intel.com (HELO mwauld-desk1.intel.com) ([10.252.19.149]) by orsmga002-auth.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 15 May 2023 07:41:09 -0700 From: Matthew Auld To: intel-xe@lists.freedesktop.org Date: Mon, 15 May 2023 15:40:26 +0100 Message-Id: <20230515144027.82857-5-matthew.auld@intel.com> X-Mailer: git-send-email 2.40.1 In-Reply-To: <20230515144027.82857-1-matthew.auld@intel.com> References: <20230515144027.82857-1-matthew.auld@intel.com> MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Subject: [Intel-xe] [PATCH v3 5/6] drm/xe: fix xe_device_mem_access_get() races 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: , Cc: Rodrigo Vivi Errors-To: intel-xe-bounces@lists.freedesktop.org Sender: "Intel-xe" It looks like there is at least one race here, given that the pm_runtime_suspended() check looks to return false if we are in the process of suspending the device (RPM_SUSPENDING vs RPM_SUSPENDED). We later also do xe_pm_runtime_get_if_active(), but since the device is suspending or has now suspended, this doesn't do anything either. Following from this we can potentially return from xe_device_mem_access_get() with the device suspended or about to be, leading to broken behaviour. Attempt to fix this by always grabbing the runtime ref when our internal ref transitions from 0 -> 1, and then wrap the whole thing with a lock to ensure callers are serialized. v2: - ct->lock looks to be primed with fs_reclaim, so holding that and then allocating memory will cause lockdep to complain. Now that we unconditionally grab the mem_access.lock around mem_access_{get,put}, we need to change the ordering wrt to grabbing the ct->lock, since some of the runtime_pm routines can allocate memory (or at least that's what lockdep seems to suggest). Hopefully not a big deal. It might be that there were already issues with this, just that the atomics where "hiding" the potential issues. v3: - Use Thomas Hellström' idea with tracking the active task that is executing in the resume or suspend callback, in order to avoid recursive resume/suspend calls deadlocking on itself. - Split the ct->lock change. Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/258 Signed-off-by: Matthew Auld Cc: Rodrigo Vivi Cc: Thomas Hellström Cc: Matthew Brost Signed-off-by: Matthew Auld --- drivers/gpu/drm/xe/xe_device.c | 30 ++++++++++++++----- drivers/gpu/drm/xe/xe_device.h | 6 ++++ drivers/gpu/drm/xe/xe_device_types.h | 11 ++++++- drivers/gpu/drm/xe/xe_pm.c | 45 +++++++++++++++------------- drivers/gpu/drm/xe/xe_pm.h | 2 +- 5 files changed, 64 insertions(+), 30 deletions(-) diff --git a/drivers/gpu/drm/xe/xe_device.c b/drivers/gpu/drm/xe/xe_device.c index d0297ee432b2..123343bc33b0 100644 --- a/drivers/gpu/drm/xe/xe_device.c +++ b/drivers/gpu/drm/xe/xe_device.c @@ -405,22 +405,38 @@ u32 xe_device_ccs_bytes(struct xe_device *xe, u64 size) void xe_device_mem_access_get(struct xe_device *xe) { - bool resumed = xe_pm_runtime_resume_if_suspended(xe); + /* + * This looks racy, but should be fine since the pm_callback_task only + * transitions from NULL -> current (and back to NULL again), during the + * runtime_resume() or runtime_suspend() callbacks, for which there can + * only be a single one running for our device. We only need to prevent + * recursively calling the runtime_get or runtime_put from those + * callbacks, as well as preventing triggering any access_ongoing + * asserts. + */ + if (READ_ONCE(xe->pm_callback_task) == current) + return; + /* + * It's possible to get rid of the mem_access.lock here, however lockdep + * has an easier time digesting the mem_access.lock -> runtime_pm chain + * (and any locks the caller is holding), on all transitions, and not + * for example just on the 0 -> 1. + */ mutex_lock(&xe->mem_access.lock); - if (xe->mem_access.ref++ == 0) - xe->mem_access.hold_rpm = xe_pm_runtime_get_if_active(xe); + if (xe->mem_access.ref == 0) + xe->mem_access.hold_rpm = xe_pm_runtime_resume_and_get(xe); + xe->mem_access.ref++; mutex_unlock(&xe->mem_access.lock); - /* The usage counter increased if device was immediately resumed */ - if (resumed) - xe_pm_runtime_put(xe); - XE_WARN_ON(xe->mem_access.ref == S32_MAX); } void xe_device_mem_access_put(struct xe_device *xe) { + if (READ_ONCE(xe->pm_callback_task) == current) + return; + mutex_lock(&xe->mem_access.lock); if (--xe->mem_access.ref == 0 && xe->mem_access.hold_rpm) xe_pm_runtime_put(xe); diff --git a/drivers/gpu/drm/xe/xe_device.h b/drivers/gpu/drm/xe/xe_device.h index 9ab7e6134f89..7a4ba5212d9c 100644 --- a/drivers/gpu/drm/xe/xe_device.h +++ b/drivers/gpu/drm/xe/xe_device.h @@ -92,6 +92,9 @@ void xe_device_mem_access_put(struct xe_device *xe); static inline void xe_device_assert_mem_access(struct xe_device *xe) { + if (READ_ONCE(xe->pm_callback_task) == current) + return; + XE_WARN_ON(!xe->mem_access.ref); } @@ -99,6 +102,9 @@ static inline bool xe_device_mem_access_ongoing(struct xe_device *xe) { bool ret; + if (READ_ONCE(xe->pm_callback_task) == current) + return true; + mutex_lock(&xe->mem_access.lock); ret = xe->mem_access.ref; mutex_unlock(&xe->mem_access.lock); diff --git a/drivers/gpu/drm/xe/xe_device_types.h b/drivers/gpu/drm/xe/xe_device_types.h index d91193c41d2e..0f4f00fa4d18 100644 --- a/drivers/gpu/drm/xe/xe_device_types.h +++ b/drivers/gpu/drm/xe/xe_device_types.h @@ -256,7 +256,10 @@ struct xe_device { * triggering additional actions when they occur. */ struct { - /** @lock: protect the ref count */ + /** + * @lock: Serialize xe_device_mem_access users, + * and protect the below internal state, like @ref. + */ struct mutex lock; /** @ref: ref count of memory accesses */ s32 ref; @@ -264,6 +267,12 @@ struct xe_device { bool hold_rpm; } mem_access; + /** + * @pm_callback_task: Track the active task that is running in either + * the runtime_suspend or runtime_resume callbacks. + */ + struct task_struct *pm_callback_task; + /** @d3cold_allowed: Indicates if d3cold is a valid device state */ bool d3cold_allowed; diff --git a/drivers/gpu/drm/xe/xe_pm.c b/drivers/gpu/drm/xe/xe_pm.c index b7b57f10ba25..6f25268ca492 100644 --- a/drivers/gpu/drm/xe/xe_pm.c +++ b/drivers/gpu/drm/xe/xe_pm.c @@ -141,39 +141,46 @@ int xe_pm_runtime_suspend(struct xe_device *xe) { struct xe_gt *gt; u8 id; - int err; + int err = 0; + + if (xe->d3cold_allowed && xe_device_mem_access_ongoing(xe)) + return -EBUSY; + + /* Disable access_ongoing asserts and prevent recursive pm calls */ + WRITE_ONCE(xe->pm_callback_task, current); if (xe->d3cold_allowed) { - if (xe_device_mem_access_ongoing(xe)) - return -EBUSY; - err = xe_bo_evict_all(xe); if (err) - return err; + goto out; } for_each_gt(gt, xe, id) { err = xe_gt_suspend(gt); if (err) - return err; + goto out; } xe_irq_suspend(xe); - - return 0; +out: + WRITE_ONCE(xe->pm_callback_task, NULL); + return err; } int xe_pm_runtime_resume(struct xe_device *xe) { struct xe_gt *gt; u8 id; - int err; + int err = 0; + + /* Disable access_ongoing asserts and prevent recursive pm calls */ + WRITE_ONCE(xe->pm_callback_task, current); if (xe->d3cold_allowed) { for_each_gt(gt, xe, id) { err = xe_pcode_init(gt); if (err) - return err; + goto out; } /* @@ -182,7 +189,7 @@ int xe_pm_runtime_resume(struct xe_device *xe) */ err = xe_bo_restore_kernel(xe); if (err) - return err; + goto out; } xe_irq_resume(xe); @@ -193,10 +200,11 @@ int xe_pm_runtime_resume(struct xe_device *xe) if (xe->d3cold_allowed) { err = xe_bo_restore_user(xe); if (err) - return err; + goto out; } - - return 0; +out: + WRITE_ONCE(xe->pm_callback_task, NULL); + return err; } int xe_pm_runtime_get(struct xe_device *xe) @@ -210,14 +218,9 @@ int xe_pm_runtime_put(struct xe_device *xe) return pm_runtime_put_autosuspend(xe->drm.dev); } -/* Return true if resume operation happened and usage count was increased */ -bool xe_pm_runtime_resume_if_suspended(struct xe_device *xe) +bool xe_pm_runtime_resume_and_get(struct xe_device *xe) { - /* In case we are suspended we need to immediately wake up */ - if (pm_runtime_suspended(xe->drm.dev)) - return !pm_runtime_resume_and_get(xe->drm.dev); - - return false; + return !pm_runtime_resume_and_get(xe->drm.dev); } int xe_pm_runtime_get_if_active(struct xe_device *xe) diff --git a/drivers/gpu/drm/xe/xe_pm.h b/drivers/gpu/drm/xe/xe_pm.h index 6a885585f653..1b4c15b5e71a 100644 --- a/drivers/gpu/drm/xe/xe_pm.h +++ b/drivers/gpu/drm/xe/xe_pm.h @@ -19,7 +19,7 @@ int xe_pm_runtime_suspend(struct xe_device *xe); int xe_pm_runtime_resume(struct xe_device *xe); int xe_pm_runtime_get(struct xe_device *xe); int xe_pm_runtime_put(struct xe_device *xe); -bool xe_pm_runtime_resume_if_suspended(struct xe_device *xe); +bool xe_pm_runtime_resume_and_get(struct xe_device *xe); int xe_pm_runtime_get_if_active(struct xe_device *xe); #endif -- 2.40.1