From patchwork Wed Dec 13 12:50:40 2017 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Patchwork-Submitter: =?utf-8?q?Micha=C5=82_Winiarski?= X-Patchwork-Id: 10109849 Return-Path: Received: from mail.wl.linuxfoundation.org (pdx-wl-mail.web.codeaurora.org [172.30.200.125]) by pdx-korg-patchwork.web.codeaurora.org (Postfix) with ESMTP id 1D89E60327 for ; Wed, 13 Dec 2017 12:51:54 +0000 (UTC) Received: from mail.wl.linuxfoundation.org (localhost [127.0.0.1]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id 1162E28F0F for ; Wed, 13 Dec 2017 12:51:54 +0000 (UTC) Received: by mail.wl.linuxfoundation.org (Postfix, from userid 486) id 0648328F2B; Wed, 13 Dec 2017 12:51:54 +0000 (UTC) X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on pdx-wl-mail.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-4.2 required=2.0 tests=BAYES_00, RCVD_IN_DNSWL_MED autolearn=ham version=3.3.1 Received: from gabe.freedesktop.org (gabe.freedesktop.org [131.252.210.177]) (using TLSv1.2 with cipher DHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by mail.wl.linuxfoundation.org (Postfix) with ESMTPS id F2DBF28F0F for ; Wed, 13 Dec 2017 12:51:52 +0000 (UTC) Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id A52636E458; Wed, 13 Dec 2017 12:51:52 +0000 (UTC) X-Original-To: intel-gfx@lists.freedesktop.org Delivered-To: intel-gfx@lists.freedesktop.org Received: from mga05.intel.com (mga05.intel.com [192.55.52.43]) by gabe.freedesktop.org (Postfix) with ESMTPS id 68D636E458 for ; Wed, 13 Dec 2017 12:51:51 +0000 (UTC) X-Amp-Result: SKIPPED(no attachment in message) X-Amp-File-Uploaded: False Received: from fmsmga007.fm.intel.com ([10.253.24.52]) by fmsmga105.fm.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 13 Dec 2017 04:51:50 -0800 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.45,397,1508828400"; d="scan'208";a="2079721" Received: from irsmsx102.ger.corp.intel.com ([163.33.3.155]) by fmsmga007.fm.intel.com with ESMTP; 13 Dec 2017 04:51:49 -0800 Received: from localhost (172.28.172.64) by IRSMSX102.ger.corp.intel.com (163.33.3.155) with Microsoft SMTP Server (TLS) id 14.3.319.2; Wed, 13 Dec 2017 12:51:49 +0000 From: =?UTF-8?q?Micha=C5=82=20Winiarski?= To: Date: Wed, 13 Dec 2017 13:50:40 +0100 Message-ID: <20171213125046.1735-2-michal.winiarski@intel.com> X-Mailer: git-send-email 2.14.3 In-Reply-To: <20171213125046.1735-1-michal.winiarski@intel.com> References: <20171213125046.1735-1-michal.winiarski@intel.com> MIME-Version: 1.0 X-Originating-IP: [172.28.172.64] Subject: [Intel-gfx] [PATCH v2 2/8] drm/i915/guc: Move GuC workqueue allocations outside of the mutex X-BeenThere: intel-gfx@lists.freedesktop.org X-Mailman-Version: 2.1.18 Precedence: list List-Id: Intel graphics driver community testing & development List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: intel-gfx-bounces@lists.freedesktop.org Sender: "Intel-gfx" X-Virus-Scanned: ClamAV using ClamSMTP This gets rid of the following lockdep splat: Reviewed-by: Chris Wilson ====================================================== WARNING: possible circular locking dependency detected 4.15.0-rc2-CI-Patchwork_7428+ #1 Not tainted ------------------------------------------------------ debugfs_test/1351 is trying to acquire lock: (&dev->struct_mutex){+.+.}, at: [<000000009d90d1a3>] i915_mutex_lock_interruptible+0x47/0x130 [i915] but task is already holding lock: (&mm->mmap_sem){++++}, at: [<000000005df01c1e>] __do_page_fault+0x106/0x560 which lock already depends on the new lock. the existing dependency chain (in reverse order) is: -> #6 (&mm->mmap_sem){++++}: __might_fault+0x63/0x90 _copy_to_user+0x1e/0x70 filldir+0x8c/0xf0 dcache_readdir+0xeb/0x160 iterate_dir+0xe6/0x150 SyS_getdents+0xa0/0x130 entry_SYSCALL_64_fastpath+0x1c/0x89 -> #5 (&sb->s_type->i_mutex_key#5){++++}: lockref_get+0x9/0x20 -> #4 ((completion)&req.done){+.+.}: wait_for_common+0x54/0x210 devtmpfs_create_node+0x130/0x150 device_add+0x5ad/0x5e0 device_create_groups_vargs+0xd4/0xe0 device_create+0x35/0x40 msr_device_create+0x22/0x40 cpuhp_invoke_callback+0xc5/0xbf0 cpuhp_thread_fun+0x167/0x210 smpboot_thread_fn+0x17f/0x270 kthread+0x173/0x1b0 ret_from_fork+0x24/0x30 -> #3 (cpuhp_state-up){+.+.}: cpuhp_issue_call+0x132/0x1c0 __cpuhp_setup_state_cpuslocked+0x12f/0x2a0 __cpuhp_setup_state+0x3a/0x50 page_writeback_init+0x3a/0x5c start_kernel+0x393/0x3e2 secondary_startup_64+0xa5/0xb0 -> #2 (cpuhp_state_mutex){+.+.}: __mutex_lock+0x81/0x9b0 __cpuhp_setup_state_cpuslocked+0x4b/0x2a0 __cpuhp_setup_state+0x3a/0x50 page_alloc_init+0x1f/0x26 start_kernel+0x139/0x3e2 secondary_startup_64+0xa5/0xb0 -> #1 (cpu_hotplug_lock.rw_sem){++++}: cpus_read_lock+0x34/0xa0 apply_workqueue_attrs+0xd/0x40 __alloc_workqueue_key+0x2c7/0x4e1 intel_guc_submission_init+0x10c/0x650 [i915] intel_uc_init_hw+0x29e/0x460 [i915] i915_gem_init_hw+0xca/0x290 [i915] i915_gem_init+0x115/0x3a0 [i915] i915_driver_load+0x9a8/0x16c0 [i915] i915_pci_probe+0x2e/0x90 [i915] pci_device_probe+0x9c/0x120 driver_probe_device+0x2a3/0x480 __driver_attach+0xd9/0xe0 bus_for_each_dev+0x57/0x90 bus_add_driver+0x168/0x260 driver_register+0x52/0xc0 do_one_initcall+0x39/0x150 do_init_module+0x56/0x1ef load_module+0x231c/0x2d70 SyS_finit_module+0xa5/0xe0 entry_SYSCALL_64_fastpath+0x1c/0x89 -> #0 (&dev->struct_mutex){+.+.}: lock_acquire+0xaf/0x200 __mutex_lock+0x81/0x9b0 i915_mutex_lock_interruptible+0x47/0x130 [i915] i915_gem_fault+0x201/0x760 [i915] __do_fault+0x15/0x70 __handle_mm_fault+0x85b/0xe40 handle_mm_fault+0x14f/0x2f0 __do_page_fault+0x2d1/0x560 page_fault+0x22/0x30 other info that might help us debug this: Chain exists of: &dev->struct_mutex --> &sb->s_type->i_mutex_key#5 --> &mm->mmap_sem Possible unsafe locking scenario: CPU0 CPU1 ---- ---- lock(&mm->mmap_sem); lock(&sb->s_type->i_mutex_key#5); lock(&mm->mmap_sem); lock(&dev->struct_mutex); *** DEADLOCK *** 1 lock held by debugfs_test/1351: #0: (&mm->mmap_sem){++++}, at: [<000000005df01c1e>] __do_page_fault+0x106/0x560 stack backtrace: CPU: 2 PID: 1351 Comm: debugfs_test Not tainted 4.15.0-rc2-CI-Patchwork_7428+ #1 Hardware name: /NUC6i5SYB, BIOS SYSKLi35.86A.0057.2017.0119.1758 01/19/2017 Call Trace: dump_stack+0x5f/0x86 print_circular_bug+0x230/0x3b0 check_prev_add+0x439/0x7b0 ? lockdep_init_map_crosslock+0x20/0x20 ? unwind_get_return_address+0x16/0x30 ? __lock_acquire+0x1385/0x15a0 __lock_acquire+0x1385/0x15a0 lock_acquire+0xaf/0x200 ? i915_mutex_lock_interruptible+0x47/0x130 [i915] __mutex_lock+0x81/0x9b0 ? i915_mutex_lock_interruptible+0x47/0x130 [i915] ? i915_mutex_lock_interruptible+0x47/0x130 [i915] ? i915_mutex_lock_interruptible+0x47/0x130 [i915] i915_mutex_lock_interruptible+0x47/0x130 [i915] ? __pm_runtime_resume+0x4f/0x80 i915_gem_fault+0x201/0x760 [i915] __do_fault+0x15/0x70 __handle_mm_fault+0x85b/0xe40 handle_mm_fault+0x14f/0x2f0 __do_page_fault+0x2d1/0x560 page_fault+0x22/0x30 RIP: 0033:0x7f98d6f49116 RSP: 002b:00007ffd6ffc3278 EFLAGS: 00010283 RAX: 00007f98d39a2bc0 RBX: 0000000000000000 RCX: 0000000000001680 RDX: 0000000000001680 RSI: 00007ffd6ffc3400 RDI: 00007f98d39a2bc0 RBP: 00007ffd6ffc33a0 R08: 0000000000000000 R09: 00000000000005a0 R10: 000055e847c2a830 R11: 0000000000000002 R12: 0000000000000001 R13: 000055e847c1d040 R14: 00007ffd6ffc3400 R15: 00007f98d6752ba0 v2: Init preempt_work unconditionally (Chris) Testcase: igt/debugfs_test/read_all_entries Signed-off-by: Michał Winiarski Cc: Chris Wilson Cc: Joonas Lahtinen Cc: Michal Wajdeczko --- drivers/gpu/drm/i915/i915_drv.c | 1 + drivers/gpu/drm/i915/i915_gem.c | 4 ++ drivers/gpu/drm/i915/intel_guc.c | 57 +++++++++++++++++++++++ drivers/gpu/drm/i915/intel_guc.h | 2 + drivers/gpu/drm/i915/intel_guc_log.c | 23 ---------- drivers/gpu/drm/i915/intel_guc_submission.c | 70 +++++++---------------------- drivers/gpu/drm/i915/intel_guc_submission.h | 2 + drivers/gpu/drm/i915/intel_uc.c | 26 +++++++++++ drivers/gpu/drm/i915/intel_uc.h | 2 + 9 files changed, 110 insertions(+), 77 deletions(-) diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c index 721ccce1832f..285c8b238bff 100644 --- a/drivers/gpu/drm/i915/i915_drv.c +++ b/drivers/gpu/drm/i915/i915_drv.c @@ -621,6 +621,7 @@ static void i915_gem_fini(struct drm_i915_private *dev_priv) i915_gem_contexts_fini(dev_priv); mutex_unlock(&dev_priv->drm.struct_mutex); + intel_uc_fini_wq(dev_priv); i915_gem_cleanup_userptr(dev_priv); i915_gem_drain_freed_objects(dev_priv); diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c index 8c3d801696b7..4b2ca43a610f 100644 --- a/drivers/gpu/drm/i915/i915_gem.c +++ b/drivers/gpu/drm/i915/i915_gem.c @@ -5160,6 +5160,10 @@ int i915_gem_init(struct drm_i915_private *dev_priv) if (ret) return ret; + ret = intel_uc_init_wq(dev_priv); + if (ret) + return ret; + /* This is just a security blanket to placate dragons. * On some systems, we very sporadically observe that the first TLBs * used by the CS may be stale, despite us poking the TLB reset. If diff --git a/drivers/gpu/drm/i915/intel_guc.c b/drivers/gpu/drm/i915/intel_guc.c index 92ed22f38fc4..3c6bf5a34c3c 100644 --- a/drivers/gpu/drm/i915/intel_guc.c +++ b/drivers/gpu/drm/i915/intel_guc.c @@ -69,6 +69,63 @@ void intel_guc_init_early(struct intel_guc *guc) guc->notify = gen8_guc_raise_irq; } +int intel_guc_init_wq(struct intel_guc *guc) +{ + struct drm_i915_private *dev_priv = guc_to_i915(guc); + + /* + * GuC log buffer flush work item has to do register access to + * send the ack to GuC and this work item, if not synced before + * suspend, can potentially get executed after the GFX device is + * suspended. + * By marking the WQ as freezable, we don't have to bother about + * flushing of this work item from the suspend hooks, the pending + * work item if any will be either executed before the suspend + * or scheduled later on resume. This way the handling of work + * item can be kept same between system suspend & rpm suspend. + */ + guc->log.runtime.flush_wq = alloc_ordered_workqueue("i915-guc_log", + WQ_HIGHPRI | WQ_FREEZABLE); + if (!guc->log.runtime.flush_wq) + return -ENOMEM; + + /* + * Even though both sending GuC action, and adding a new workitem to + * GuC workqueue are serialized (each with its own locking), since + * we're using mutliple engines, it's possible that we're going to + * issue a preempt request with two (or more - each for different + * engine) workitems in GuC queue. In this situation, GuC may submit + * all of them, which will make us very confused. + * Our preemption contexts may even already be complete - before we + * even had the chance to sent the preempt action to GuC!. Rather + * than introducing yet another lock, we can just use ordered workqueue + * to make sure we're always sending a single preemption request with a + * single workitem. + */ + if (HAS_LOGICAL_RING_PREEMPTION(dev_priv) && + USES_GUC_SUBMISSION(dev_priv)) { + guc->preempt_wq = alloc_ordered_workqueue("i915-guc_preempt", + WQ_HIGHPRI); + if (!guc->preempt_wq) { + destroy_workqueue(guc->log.runtime.flush_wq); + return -ENOMEM; + } + } + + return 0; +} + +void intel_guc_fini_wq(struct intel_guc *guc) +{ + struct drm_i915_private *dev_priv = guc_to_i915(guc); + + if (HAS_LOGICAL_RING_PREEMPTION(dev_priv) && + USES_GUC_SUBMISSION(dev_priv)) + destroy_workqueue(guc->preempt_wq); + + destroy_workqueue(guc->log.runtime.flush_wq); +} + static int guc_shared_data_create(struct intel_guc *guc) { struct i915_vma *vma; diff --git a/drivers/gpu/drm/i915/intel_guc.h b/drivers/gpu/drm/i915/intel_guc.h index 81659e223e11..52856a97477d 100644 --- a/drivers/gpu/drm/i915/intel_guc.h +++ b/drivers/gpu/drm/i915/intel_guc.h @@ -119,6 +119,8 @@ static inline u32 guc_ggtt_offset(struct i915_vma *vma) void intel_guc_init_early(struct intel_guc *guc); void intel_guc_init_send_regs(struct intel_guc *guc); void intel_guc_init_params(struct intel_guc *guc); +int intel_guc_init_wq(struct intel_guc *guc); +void intel_guc_fini_wq(struct intel_guc *guc); int intel_guc_init(struct intel_guc *guc); void intel_guc_fini(struct intel_guc *guc); int intel_guc_send_nop(struct intel_guc *guc, const u32 *action, u32 len); diff --git a/drivers/gpu/drm/i915/intel_guc_log.c b/drivers/gpu/drm/i915/intel_guc_log.c index 1a2c5eed9929..eaedd63e3819 100644 --- a/drivers/gpu/drm/i915/intel_guc_log.c +++ b/drivers/gpu/drm/i915/intel_guc_log.c @@ -411,30 +411,8 @@ static int guc_log_runtime_create(struct intel_guc *guc) guc->log.runtime.relay_chan = guc_log_relay_chan; INIT_WORK(&guc->log.runtime.flush_work, capture_logs_work); - - /* - * GuC log buffer flush work item has to do register access to - * send the ack to GuC and this work item, if not synced before - * suspend, can potentially get executed after the GFX device is - * suspended. - * By marking the WQ as freezable, we don't have to bother about - * flushing of this work item from the suspend hooks, the pending - * work item if any will be either executed before the suspend - * or scheduled later on resume. This way the handling of work - * item can be kept same between system suspend & rpm suspend. - */ - guc->log.runtime.flush_wq = alloc_ordered_workqueue("i915-guc_log", - WQ_HIGHPRI | WQ_FREEZABLE); - if (!guc->log.runtime.flush_wq) { - DRM_ERROR("Couldn't allocate the wq for GuC logging\n"); - ret = -ENOMEM; - goto err_relaychan; - } - return 0; -err_relaychan: - relay_close(guc->log.runtime.relay_chan); err_vaddr: i915_gem_object_unpin_map(guc->log.vma->obj); guc->log.runtime.buf_addr = NULL; @@ -450,7 +428,6 @@ static void guc_log_runtime_destroy(struct intel_guc *guc) if (!guc_log_has_runtime(guc)) return; - destroy_workqueue(guc->log.runtime.flush_wq); relay_close(guc->log.runtime.relay_chan); i915_gem_object_unpin_map(guc->log.vma->obj); guc->log.runtime.buf_addr = NULL; diff --git a/drivers/gpu/drm/i915/intel_guc_submission.c b/drivers/gpu/drm/i915/intel_guc_submission.c index c020560c395e..8f4b274d66a7 100644 --- a/drivers/gpu/drm/i915/intel_guc_submission.c +++ b/drivers/gpu/drm/i915/intel_guc_submission.c @@ -1187,57 +1187,15 @@ static void guc_ads_destroy(struct intel_guc *guc) i915_vma_unpin_and_release(&guc->ads_vma); } -static int guc_preempt_work_create(struct intel_guc *guc) -{ - struct drm_i915_private *dev_priv = guc_to_i915(guc); - struct intel_engine_cs *engine; - enum intel_engine_id id; - - /* - * Even though both sending GuC action, and adding a new workitem to - * GuC workqueue are serialized (each with its own locking), since - * we're using mutliple engines, it's possible that we're going to - * issue a preempt request with two (or more - each for different - * engine) workitems in GuC queue. In this situation, GuC may submit - * all of them, which will make us very confused. - * Our preemption contexts may even already be complete - before we - * even had the chance to sent the preempt action to GuC!. Rather - * than introducing yet another lock, we can just use ordered workqueue - * to make sure we're always sending a single preemption request with a - * single workitem. - */ - guc->preempt_wq = alloc_ordered_workqueue("i915-guc_preempt", - WQ_HIGHPRI); - if (!guc->preempt_wq) - return -ENOMEM; - - for_each_engine(engine, dev_priv, id) { - guc->preempt_work[id].engine = engine; - INIT_WORK(&guc->preempt_work[id].work, inject_preempt_context); - } - - return 0; -} - -static void guc_preempt_work_destroy(struct intel_guc *guc) -{ - struct drm_i915_private *dev_priv = guc_to_i915(guc); - struct intel_engine_cs *engine; - enum intel_engine_id id; - - for_each_engine(engine, dev_priv, id) - cancel_work_sync(&guc->preempt_work[id].work); - - destroy_workqueue(guc->preempt_wq); - guc->preempt_wq = NULL; -} - /* * Set up the memory resources to be shared with the GuC (via the GGTT) * at firmware loading time. */ int intel_guc_submission_init(struct intel_guc *guc) { + struct drm_i915_private *dev_priv = guc_to_i915(guc); + struct intel_engine_cs *engine; + enum intel_engine_id id; int ret; if (guc->stage_desc_pool) @@ -1256,20 +1214,18 @@ int intel_guc_submission_init(struct intel_guc *guc) if (ret < 0) goto err_stage_desc_pool; - ret = guc_preempt_work_create(guc); - if (ret) - goto err_log; - GEM_BUG_ON(!guc->preempt_wq); - ret = guc_ads_create(guc); if (ret < 0) - goto err_wq; + goto err_log; GEM_BUG_ON(!guc->ads_vma); + for_each_engine(engine, dev_priv, id) { + guc->preempt_work[id].engine = engine; + INIT_WORK(&guc->preempt_work[id].work, inject_preempt_context); + } + return 0; -err_wq: - guc_preempt_work_destroy(guc); err_log: intel_guc_log_destroy(guc); err_stage_desc_pool: @@ -1279,8 +1235,14 @@ int intel_guc_submission_init(struct intel_guc *guc) void intel_guc_submission_fini(struct intel_guc *guc) { + struct drm_i915_private *dev_priv = guc_to_i915(guc); + struct intel_engine_cs *engine; + enum intel_engine_id id; + + for_each_engine(engine, dev_priv, id) + cancel_work_sync(&guc->preempt_work[id].work); + guc_ads_destroy(guc); - guc_preempt_work_destroy(guc); intel_guc_log_destroy(guc); guc_stage_desc_pool_destroy(guc); } diff --git a/drivers/gpu/drm/i915/intel_guc_submission.h b/drivers/gpu/drm/i915/intel_guc_submission.h index 021fe85c8f71..fb081cefef93 100644 --- a/drivers/gpu/drm/i915/intel_guc_submission.h +++ b/drivers/gpu/drm/i915/intel_guc_submission.h @@ -77,5 +77,7 @@ int intel_guc_submission_init(struct intel_guc *guc); int intel_guc_submission_enable(struct intel_guc *guc); void intel_guc_submission_disable(struct intel_guc *guc); void intel_guc_submission_fini(struct intel_guc *guc); +int intel_guc_preempt_work_create(struct intel_guc *guc); +void intel_guc_preempt_work_destroy(struct intel_guc *guc); #endif diff --git a/drivers/gpu/drm/i915/intel_uc.c b/drivers/gpu/drm/i915/intel_uc.c index 3040a0e00142..785850838a44 100644 --- a/drivers/gpu/drm/i915/intel_uc.c +++ b/drivers/gpu/drm/i915/intel_uc.c @@ -188,6 +188,32 @@ static void guc_disable_communication(struct intel_guc *guc) guc->send = intel_guc_send_nop; } +int intel_uc_init_wq(struct drm_i915_private *dev_priv) +{ + int ret; + + if (!USES_GUC(dev_priv)) + return 0; + + ret = intel_guc_init_wq(&dev_priv->guc); + if (ret) { + DRM_ERROR("Couldn't allocate workqueues for GuC\n"); + return ret; + } + + return 0; +} + +void intel_uc_fini_wq(struct drm_i915_private *dev_priv) +{ + if (!USES_GUC(dev_priv)) + return; + + GEM_BUG_ON(!HAS_GUC(dev_priv)); + + intel_guc_fini_wq(&dev_priv->guc); +} + int intel_uc_init_hw(struct drm_i915_private *dev_priv) { struct intel_guc *guc = &dev_priv->guc; diff --git a/drivers/gpu/drm/i915/intel_uc.h b/drivers/gpu/drm/i915/intel_uc.h index 7a59e2486e9e..53edfeaf56b0 100644 --- a/drivers/gpu/drm/i915/intel_uc.h +++ b/drivers/gpu/drm/i915/intel_uc.h @@ -33,6 +33,8 @@ void intel_uc_init_early(struct drm_i915_private *dev_priv); void intel_uc_init_mmio(struct drm_i915_private *dev_priv); void intel_uc_init_fw(struct drm_i915_private *dev_priv); void intel_uc_fini_fw(struct drm_i915_private *dev_priv); +int intel_uc_init_wq(struct drm_i915_private *dev_priv); +void intel_uc_fini_wq(struct drm_i915_private *dev_priv); int intel_uc_init_hw(struct drm_i915_private *dev_priv); void intel_uc_fini_hw(struct drm_i915_private *dev_priv);