From patchwork Tue Nov 1 16:43:03 2016 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Patchwork-Submitter: Mika Kuoppala X-Patchwork-Id: 9407651 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 0E39160585 for ; Tue, 1 Nov 2016 16:43:43 +0000 (UTC) Received: from mail.wl.linuxfoundation.org (localhost [127.0.0.1]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id F03F72993E for ; Tue, 1 Nov 2016 16:43:42 +0000 (UTC) Received: by mail.wl.linuxfoundation.org (Postfix, from userid 486) id E2258299E6; Tue, 1 Nov 2016 16:43:42 +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 5964E2993E for ; Tue, 1 Nov 2016 16:43:41 +0000 (UTC) Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id B56CC89DCF; Tue, 1 Nov 2016 16:43:40 +0000 (UTC) X-Original-To: intel-gfx@lists.freedesktop.org Delivered-To: intel-gfx@lists.freedesktop.org Received: from mga04.intel.com (mga04.intel.com [192.55.52.120]) by gabe.freedesktop.org (Postfix) with ESMTPS id 2976B89DCF for ; Tue, 1 Nov 2016 16:43:40 +0000 (UTC) Received: from fmsmga001.fm.intel.com ([10.253.24.23]) by fmsmga104.fm.intel.com with ESMTP; 01 Nov 2016 09:43:40 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos; i="5.31,580,1473145200"; d="scan'208"; a="1062414170" Received: from rosetta.fi.intel.com ([10.237.72.176]) by fmsmga001.fm.intel.com with ESMTP; 01 Nov 2016 09:43:37 -0700 Received: by rosetta.fi.intel.com (Postfix, from userid 1000) id 24080840311; Tue, 1 Nov 2016 18:43:05 +0200 (EET) From: Mika Kuoppala To: intel-gfx@lists.freedesktop.org Date: Tue, 1 Nov 2016 18:43:03 +0200 Message-Id: <1478018583-5816-1-git-send-email-mika.kuoppala@intel.com> X-Mailer: git-send-email 2.7.4 MIME-Version: 1.0 Subject: [Intel-gfx] [PATCH] drm/i915: Move hangcheck code out from i915_irq.c 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 Create new file for hangcheck specific code, intel_hangcheck.c, and move all related code in it. v2: s/intel_engine_hangcheck/intel_engine (Chris) No functional changes. Cc: Chris Wilson Cc: Joonas Lahtinen Cc: Tvrtko Ursulin Signed-off-by: Mika Kuoppala Reviewed-by: Chris Wilson --- drivers/gpu/drm/i915/Makefile | 1 + drivers/gpu/drm/i915/i915_drv.c | 1 + drivers/gpu/drm/i915/i915_drv.h | 1 + drivers/gpu/drm/i915/i915_irq.c | 417 ------------------------------ drivers/gpu/drm/i915/intel_engine_cs.c | 5 - drivers/gpu/drm/i915/intel_hangcheck.c | 450 +++++++++++++++++++++++++++++++++ 6 files changed, 453 insertions(+), 422 deletions(-) create mode 100644 drivers/gpu/drm/i915/intel_hangcheck.c diff --git a/drivers/gpu/drm/i915/Makefile b/drivers/gpu/drm/i915/Makefile index 240ce9a..0857e50 100644 --- a/drivers/gpu/drm/i915/Makefile +++ b/drivers/gpu/drm/i915/Makefile @@ -47,6 +47,7 @@ i915-y += i915_cmd_parser.o \ i915_trace_points.o \ intel_breadcrumbs.o \ intel_engine_cs.o \ + intel_hangcheck.o \ intel_lrc.o \ intel_mocs.o \ intel_ringbuffer.o \ diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c index 48f4d21..9e5a547 100644 --- a/drivers/gpu/drm/i915/i915_drv.c +++ b/drivers/gpu/drm/i915/i915_drv.c @@ -830,6 +830,7 @@ static int i915_driver_init_early(struct drm_i915_private *dev_priv, intel_init_dpio(dev_priv); intel_power_domains_init(dev_priv); intel_irq_init(dev_priv); + intel_hangcheck_init(dev_priv); intel_init_display_hooks(dev_priv); intel_init_clock_gating_hooks(dev_priv); intel_init_audio_hooks(dev_priv); diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h index b8e72cd..6c0b0a6 100644 --- a/drivers/gpu/drm/i915/i915_drv.h +++ b/drivers/gpu/drm/i915/i915_drv.h @@ -3002,6 +3002,7 @@ extern bool intel_has_gpu_reset(struct drm_i915_private *dev_priv); extern void i915_reset(struct drm_i915_private *dev_priv); extern int intel_guc_reset(struct drm_i915_private *dev_priv); extern void intel_engine_init_hangcheck(struct intel_engine_cs *engine); +extern void intel_hangcheck_init(struct drm_i915_private *dev_priv); extern unsigned long i915_chipset_val(struct drm_i915_private *dev_priv); extern unsigned long i915_mch_val(struct drm_i915_private *dev_priv); extern unsigned long i915_gfx_val(struct drm_i915_private *dev_priv); diff --git a/drivers/gpu/drm/i915/i915_irq.c b/drivers/gpu/drm/i915/i915_irq.c index ecd06d3..6d7505b 100644 --- a/drivers/gpu/drm/i915/i915_irq.c +++ b/drivers/gpu/drm/i915/i915_irq.c @@ -2848,420 +2848,6 @@ static void gen8_disable_vblank(struct drm_device *dev, unsigned int pipe) spin_unlock_irqrestore(&dev_priv->irq_lock, irqflags); } -static bool -ipehr_is_semaphore_wait(struct intel_engine_cs *engine, u32 ipehr) -{ - if (INTEL_GEN(engine->i915) >= 8) { - return (ipehr >> 23) == 0x1c; - } else { - ipehr &= ~MI_SEMAPHORE_SYNC_MASK; - return ipehr == (MI_SEMAPHORE_MBOX | MI_SEMAPHORE_COMPARE | - MI_SEMAPHORE_REGISTER); - } -} - -static struct intel_engine_cs * -semaphore_wait_to_signaller_ring(struct intel_engine_cs *engine, u32 ipehr, - u64 offset) -{ - struct drm_i915_private *dev_priv = engine->i915; - struct intel_engine_cs *signaller; - enum intel_engine_id id; - - if (INTEL_GEN(dev_priv) >= 8) { - for_each_engine(signaller, dev_priv, id) { - if (engine == signaller) - continue; - - if (offset == signaller->semaphore.signal_ggtt[engine->hw_id]) - return signaller; - } - } else { - u32 sync_bits = ipehr & MI_SEMAPHORE_SYNC_MASK; - - for_each_engine(signaller, dev_priv, id) { - if(engine == signaller) - continue; - - if (sync_bits == signaller->semaphore.mbox.wait[engine->hw_id]) - return signaller; - } - } - - DRM_DEBUG_DRIVER("No signaller ring found for %s, ipehr 0x%08x, offset 0x%016llx\n", - engine->name, ipehr, offset); - - return ERR_PTR(-ENODEV); -} - -static struct intel_engine_cs * -semaphore_waits_for(struct intel_engine_cs *engine, u32 *seqno) -{ - struct drm_i915_private *dev_priv = engine->i915; - void __iomem *vaddr; - u32 cmd, ipehr, head; - u64 offset = 0; - int i, backwards; - - /* - * This function does not support execlist mode - any attempt to - * proceed further into this function will result in a kernel panic - * when dereferencing ring->buffer, which is not set up in execlist - * mode. - * - * The correct way of doing it would be to derive the currently - * executing ring buffer from the current context, which is derived - * from the currently running request. Unfortunately, to get the - * current request we would have to grab the struct_mutex before doing - * anything else, which would be ill-advised since some other thread - * might have grabbed it already and managed to hang itself, causing - * the hang checker to deadlock. - * - * Therefore, this function does not support execlist mode in its - * current form. Just return NULL and move on. - */ - if (engine->buffer == NULL) - return NULL; - - ipehr = I915_READ(RING_IPEHR(engine->mmio_base)); - if (!ipehr_is_semaphore_wait(engine, ipehr)) - return NULL; - - /* - * HEAD is likely pointing to the dword after the actual command, - * so scan backwards until we find the MBOX. But limit it to just 3 - * or 4 dwords depending on the semaphore wait command size. - * Note that we don't care about ACTHD here since that might - * point at at batch, and semaphores are always emitted into the - * ringbuffer itself. - */ - head = I915_READ_HEAD(engine) & HEAD_ADDR; - backwards = (INTEL_GEN(dev_priv) >= 8) ? 5 : 4; - vaddr = (void __iomem *)engine->buffer->vaddr; - - for (i = backwards; i; --i) { - /* - * Be paranoid and presume the hw has gone off into the wild - - * our ring is smaller than what the hardware (and hence - * HEAD_ADDR) allows. Also handles wrap-around. - */ - head &= engine->buffer->size - 1; - - /* This here seems to blow up */ - cmd = ioread32(vaddr + head); - if (cmd == ipehr) - break; - - head -= 4; - } - - if (!i) - return NULL; - - *seqno = ioread32(vaddr + head + 4) + 1; - if (INTEL_GEN(dev_priv) >= 8) { - offset = ioread32(vaddr + head + 12); - offset <<= 32; - offset |= ioread32(vaddr + head + 8); - } - return semaphore_wait_to_signaller_ring(engine, ipehr, offset); -} - -static int semaphore_passed(struct intel_engine_cs *engine) -{ - struct drm_i915_private *dev_priv = engine->i915; - struct intel_engine_cs *signaller; - u32 seqno; - - engine->hangcheck.deadlock++; - - signaller = semaphore_waits_for(engine, &seqno); - if (signaller == NULL) - return -1; - - if (IS_ERR(signaller)) - return 0; - - /* Prevent pathological recursion due to driver bugs */ - if (signaller->hangcheck.deadlock >= I915_NUM_ENGINES) - return -1; - - if (i915_seqno_passed(intel_engine_get_seqno(signaller), seqno)) - return 1; - - /* cursory check for an unkickable deadlock */ - if (I915_READ_CTL(signaller) & RING_WAIT_SEMAPHORE && - semaphore_passed(signaller) < 0) - return -1; - - return 0; -} - -static void semaphore_clear_deadlocks(struct drm_i915_private *dev_priv) -{ - struct intel_engine_cs *engine; - enum intel_engine_id id; - - for_each_engine(engine, dev_priv, id) - engine->hangcheck.deadlock = 0; -} - -static bool instdone_unchanged(u32 current_instdone, u32 *old_instdone) -{ - u32 tmp = current_instdone | *old_instdone; - bool unchanged; - - unchanged = tmp == *old_instdone; - *old_instdone |= tmp; - - return unchanged; -} - -static bool subunits_stuck(struct intel_engine_cs *engine) -{ - struct drm_i915_private *dev_priv = engine->i915; - struct intel_instdone instdone; - struct intel_instdone *accu_instdone = &engine->hangcheck.instdone; - bool stuck; - int slice; - int subslice; - - if (engine->id != RCS) - return true; - - intel_engine_get_instdone(engine, &instdone); - - /* There might be unstable subunit states even when - * actual head is not moving. Filter out the unstable ones by - * accumulating the undone -> done transitions and only - * consider those as progress. - */ - stuck = instdone_unchanged(instdone.instdone, - &accu_instdone->instdone); - stuck &= instdone_unchanged(instdone.slice_common, - &accu_instdone->slice_common); - - for_each_instdone_slice_subslice(dev_priv, slice, subslice) { - stuck &= instdone_unchanged(instdone.sampler[slice][subslice], - &accu_instdone->sampler[slice][subslice]); - stuck &= instdone_unchanged(instdone.row[slice][subslice], - &accu_instdone->row[slice][subslice]); - } - - return stuck; -} - -static enum intel_engine_hangcheck_action -head_stuck(struct intel_engine_cs *engine, u64 acthd) -{ - if (acthd != engine->hangcheck.acthd) { - - /* Clear subunit states on head movement */ - memset(&engine->hangcheck.instdone, 0, - sizeof(engine->hangcheck.instdone)); - - return HANGCHECK_ACTIVE; - } - - if (!subunits_stuck(engine)) - return HANGCHECK_ACTIVE; - - return HANGCHECK_HUNG; -} - -static enum intel_engine_hangcheck_action -engine_stuck(struct intel_engine_cs *engine, u64 acthd) -{ - struct drm_i915_private *dev_priv = engine->i915; - enum intel_engine_hangcheck_action ha; - u32 tmp; - - ha = head_stuck(engine, acthd); - if (ha != HANGCHECK_HUNG) - return ha; - - if (IS_GEN2(dev_priv)) - return HANGCHECK_HUNG; - - /* Is the chip hanging on a WAIT_FOR_EVENT? - * If so we can simply poke the RB_WAIT bit - * and break the hang. This should work on - * all but the second generation chipsets. - */ - tmp = I915_READ_CTL(engine); - if (tmp & RING_WAIT) { - i915_handle_error(dev_priv, 0, - "Kicking stuck wait on %s", - engine->name); - I915_WRITE_CTL(engine, tmp); - return HANGCHECK_KICK; - } - - if (INTEL_GEN(dev_priv) >= 6 && tmp & RING_WAIT_SEMAPHORE) { - switch (semaphore_passed(engine)) { - default: - return HANGCHECK_HUNG; - case 1: - i915_handle_error(dev_priv, 0, - "Kicking stuck semaphore on %s", - engine->name); - I915_WRITE_CTL(engine, tmp); - return HANGCHECK_KICK; - case 0: - return HANGCHECK_WAIT; - } - } - - return HANGCHECK_HUNG; -} - -/* - * This is called when the chip hasn't reported back with completed - * batchbuffers in a long time. We keep track per ring seqno progress and - * if there are no progress, hangcheck score for that ring is increased. - * Further, acthd is inspected to see if the ring is stuck. On stuck case - * we kick the ring. If we see no progress on three subsequent calls - * we assume chip is wedged and try to fix it by resetting the chip. - */ -static void i915_hangcheck_elapsed(struct work_struct *work) -{ - struct drm_i915_private *dev_priv = - container_of(work, typeof(*dev_priv), - gpu_error.hangcheck_work.work); - struct intel_engine_cs *engine; - enum intel_engine_id id; - unsigned int hung = 0, stuck = 0; - int busy_count = 0; -#define BUSY 1 -#define KICK 5 -#define HUNG 20 -#define ACTIVE_DECAY 15 - - if (!i915.enable_hangcheck) - return; - - if (!READ_ONCE(dev_priv->gt.awake)) - return; - - /* As enabling the GPU requires fairly extensive mmio access, - * periodically arm the mmio checker to see if we are triggering - * any invalid access. - */ - intel_uncore_arm_unclaimed_mmio_detection(dev_priv); - - for_each_engine(engine, dev_priv, id) { - bool busy = intel_engine_has_waiter(engine); - u64 acthd; - u32 seqno; - u32 submit; - - semaphore_clear_deadlocks(dev_priv); - - /* We don't strictly need an irq-barrier here, as we are not - * serving an interrupt request, be paranoid in case the - * barrier has side-effects (such as preventing a broken - * cacheline snoop) and so be sure that we can see the seqno - * advance. If the seqno should stick, due to a stale - * cacheline, we would erroneously declare the GPU hung. - */ - if (engine->irq_seqno_barrier) - engine->irq_seqno_barrier(engine); - - acthd = intel_engine_get_active_head(engine); - seqno = intel_engine_get_seqno(engine); - submit = intel_engine_last_submit(engine); - - if (engine->hangcheck.seqno == seqno) { - if (i915_seqno_passed(seqno, submit)) { - engine->hangcheck.action = HANGCHECK_IDLE; - } else { - /* We always increment the hangcheck score - * if the engine is busy and still processing - * the same request, so that no single request - * can run indefinitely (such as a chain of - * batches). The only time we do not increment - * the hangcheck score on this ring, if this - * engine is in a legitimate wait for another - * engine. In that case the waiting engine is a - * victim and we want to be sure we catch the - * right culprit. Then every time we do kick - * the ring, add a small increment to the - * score so that we can catch a batch that is - * being repeatedly kicked and so responsible - * for stalling the machine. - */ - engine->hangcheck.action = - engine_stuck(engine, acthd); - - switch (engine->hangcheck.action) { - case HANGCHECK_IDLE: - case HANGCHECK_WAIT: - break; - case HANGCHECK_ACTIVE: - engine->hangcheck.score += BUSY; - break; - case HANGCHECK_KICK: - engine->hangcheck.score += KICK; - break; - case HANGCHECK_HUNG: - engine->hangcheck.score += HUNG; - break; - } - } - - if (engine->hangcheck.score >= HANGCHECK_SCORE_RING_HUNG) { - hung |= intel_engine_flag(engine); - if (engine->hangcheck.action != HANGCHECK_HUNG) - stuck |= intel_engine_flag(engine); - } - } else { - engine->hangcheck.action = HANGCHECK_ACTIVE; - - /* Gradually reduce the count so that we catch DoS - * attempts across multiple batches. - */ - if (engine->hangcheck.score > 0) - engine->hangcheck.score -= ACTIVE_DECAY; - if (engine->hangcheck.score < 0) - engine->hangcheck.score = 0; - - /* Clear head and subunit states on seqno movement */ - acthd = 0; - - memset(&engine->hangcheck.instdone, 0, - sizeof(engine->hangcheck.instdone)); - } - - engine->hangcheck.seqno = seqno; - engine->hangcheck.acthd = acthd; - busy_count += busy; - } - - if (hung) { - char msg[80]; - unsigned int tmp; - int len; - - /* If some rings hung but others were still busy, only - * blame the hanging rings in the synopsis. - */ - if (stuck != hung) - hung &= ~stuck; - len = scnprintf(msg, sizeof(msg), - "%s on ", stuck == hung ? "No progress" : "Hang"); - for_each_engine_masked(engine, dev_priv, hung, tmp) - len += scnprintf(msg + len, sizeof(msg) - len, - "%s, ", engine->name); - msg[len-2] = '\0'; - - return i915_handle_error(dev_priv, hung, msg); - } - - /* Reset timer in case GPU hangs without another request being added */ - if (busy_count) - i915_queue_hangcheck(dev_priv); -} - static void ibx_irq_reset(struct drm_device *dev) { struct drm_i915_private *dev_priv = to_i915(dev); @@ -4583,9 +4169,6 @@ void intel_irq_init(struct drm_i915_private *dev_priv) if (INTEL_INFO(dev_priv)->gen >= 8) dev_priv->rps.pm_intr_keep |= GEN8_PMINTR_REDIRECT_TO_GUC; - INIT_DELAYED_WORK(&dev_priv->gpu_error.hangcheck_work, - i915_hangcheck_elapsed); - if (IS_GEN2(dev_priv)) { /* Gen2 doesn't have a hardware frame counter */ dev->max_vblank_count = 0; diff --git a/drivers/gpu/drm/i915/intel_engine_cs.c b/drivers/gpu/drm/i915/intel_engine_cs.c index 94de3d6..841f8d1 100644 --- a/drivers/gpu/drm/i915/intel_engine_cs.c +++ b/drivers/gpu/drm/i915/intel_engine_cs.c @@ -220,11 +220,6 @@ void intel_engine_init_global_seqno(struct intel_engine_cs *engine, u32 seqno) intel_engine_wakeup(engine); } -void intel_engine_init_hangcheck(struct intel_engine_cs *engine) -{ - memset(&engine->hangcheck, 0, sizeof(engine->hangcheck)); -} - static void intel_engine_init_timeline(struct intel_engine_cs *engine) { engine->timeline = &engine->i915->gt.global_timeline.engine[engine->id]; diff --git a/drivers/gpu/drm/i915/intel_hangcheck.c b/drivers/gpu/drm/i915/intel_hangcheck.c new file mode 100644 index 0000000..53df5b1 --- /dev/null +++ b/drivers/gpu/drm/i915/intel_hangcheck.c @@ -0,0 +1,450 @@ +/* + * Copyright © 2016 Intel Corporation + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS + * IN THE SOFTWARE. + * + */ + +#include "i915_drv.h" + +static bool +ipehr_is_semaphore_wait(struct intel_engine_cs *engine, u32 ipehr) +{ + if (INTEL_GEN(engine->i915) >= 8) { + return (ipehr >> 23) == 0x1c; + } else { + ipehr &= ~MI_SEMAPHORE_SYNC_MASK; + return ipehr == (MI_SEMAPHORE_MBOX | MI_SEMAPHORE_COMPARE | + MI_SEMAPHORE_REGISTER); + } +} + +static struct intel_engine_cs * +semaphore_wait_to_signaller_ring(struct intel_engine_cs *engine, u32 ipehr, + u64 offset) +{ + struct drm_i915_private *dev_priv = engine->i915; + struct intel_engine_cs *signaller; + enum intel_engine_id id; + + if (INTEL_GEN(dev_priv) >= 8) { + for_each_engine(signaller, dev_priv, id) { + if (engine == signaller) + continue; + + if (offset == signaller->semaphore.signal_ggtt[engine->hw_id]) + return signaller; + } + } else { + u32 sync_bits = ipehr & MI_SEMAPHORE_SYNC_MASK; + + for_each_engine(signaller, dev_priv, id) { + if(engine == signaller) + continue; + + if (sync_bits == signaller->semaphore.mbox.wait[engine->hw_id]) + return signaller; + } + } + + DRM_DEBUG_DRIVER("No signaller ring found for %s, ipehr 0x%08x, offset 0x%016llx\n", + engine->name, ipehr, offset); + + return ERR_PTR(-ENODEV); +} + +static struct intel_engine_cs * +semaphore_waits_for(struct intel_engine_cs *engine, u32 *seqno) +{ + struct drm_i915_private *dev_priv = engine->i915; + void __iomem *vaddr; + u32 cmd, ipehr, head; + u64 offset = 0; + int i, backwards; + + /* + * This function does not support execlist mode - any attempt to + * proceed further into this function will result in a kernel panic + * when dereferencing ring->buffer, which is not set up in execlist + * mode. + * + * The correct way of doing it would be to derive the currently + * executing ring buffer from the current context, which is derived + * from the currently running request. Unfortunately, to get the + * current request we would have to grab the struct_mutex before doing + * anything else, which would be ill-advised since some other thread + * might have grabbed it already and managed to hang itself, causing + * the hang checker to deadlock. + * + * Therefore, this function does not support execlist mode in its + * current form. Just return NULL and move on. + */ + if (engine->buffer == NULL) + return NULL; + + ipehr = I915_READ(RING_IPEHR(engine->mmio_base)); + if (!ipehr_is_semaphore_wait(engine, ipehr)) + return NULL; + + /* + * HEAD is likely pointing to the dword after the actual command, + * so scan backwards until we find the MBOX. But limit it to just 3 + * or 4 dwords depending on the semaphore wait command size. + * Note that we don't care about ACTHD here since that might + * point at at batch, and semaphores are always emitted into the + * ringbuffer itself. + */ + head = I915_READ_HEAD(engine) & HEAD_ADDR; + backwards = (INTEL_GEN(dev_priv) >= 8) ? 5 : 4; + vaddr = (void __iomem *)engine->buffer->vaddr; + + for (i = backwards; i; --i) { + /* + * Be paranoid and presume the hw has gone off into the wild - + * our ring is smaller than what the hardware (and hence + * HEAD_ADDR) allows. Also handles wrap-around. + */ + head &= engine->buffer->size - 1; + + /* This here seems to blow up */ + cmd = ioread32(vaddr + head); + if (cmd == ipehr) + break; + + head -= 4; + } + + if (!i) + return NULL; + + *seqno = ioread32(vaddr + head + 4) + 1; + if (INTEL_GEN(dev_priv) >= 8) { + offset = ioread32(vaddr + head + 12); + offset <<= 32; + offset |= ioread32(vaddr + head + 8); + } + return semaphore_wait_to_signaller_ring(engine, ipehr, offset); +} + +static int semaphore_passed(struct intel_engine_cs *engine) +{ + struct drm_i915_private *dev_priv = engine->i915; + struct intel_engine_cs *signaller; + u32 seqno; + + engine->hangcheck.deadlock++; + + signaller = semaphore_waits_for(engine, &seqno); + if (signaller == NULL) + return -1; + + if (IS_ERR(signaller)) + return 0; + + /* Prevent pathological recursion due to driver bugs */ + if (signaller->hangcheck.deadlock >= I915_NUM_ENGINES) + return -1; + + if (i915_seqno_passed(intel_engine_get_seqno(signaller), seqno)) + return 1; + + /* cursory check for an unkickable deadlock */ + if (I915_READ_CTL(signaller) & RING_WAIT_SEMAPHORE && + semaphore_passed(signaller) < 0) + return -1; + + return 0; +} + +static void semaphore_clear_deadlocks(struct drm_i915_private *dev_priv) +{ + struct intel_engine_cs *engine; + enum intel_engine_id id; + + for_each_engine(engine, dev_priv, id) + engine->hangcheck.deadlock = 0; +} + +static bool instdone_unchanged(u32 current_instdone, u32 *old_instdone) +{ + u32 tmp = current_instdone | *old_instdone; + bool unchanged; + + unchanged = tmp == *old_instdone; + *old_instdone |= tmp; + + return unchanged; +} + +static bool subunits_stuck(struct intel_engine_cs *engine) +{ + struct drm_i915_private *dev_priv = engine->i915; + struct intel_instdone instdone; + struct intel_instdone *accu_instdone = &engine->hangcheck.instdone; + bool stuck; + int slice; + int subslice; + + if (engine->id != RCS) + return true; + + intel_engine_get_instdone(engine, &instdone); + + /* There might be unstable subunit states even when + * actual head is not moving. Filter out the unstable ones by + * accumulating the undone -> done transitions and only + * consider those as progress. + */ + stuck = instdone_unchanged(instdone.instdone, + &accu_instdone->instdone); + stuck &= instdone_unchanged(instdone.slice_common, + &accu_instdone->slice_common); + + for_each_instdone_slice_subslice(dev_priv, slice, subslice) { + stuck &= instdone_unchanged(instdone.sampler[slice][subslice], + &accu_instdone->sampler[slice][subslice]); + stuck &= instdone_unchanged(instdone.row[slice][subslice], + &accu_instdone->row[slice][subslice]); + } + + return stuck; +} + +static enum intel_engine_hangcheck_action +head_stuck(struct intel_engine_cs *engine, u64 acthd) +{ + if (acthd != engine->hangcheck.acthd) { + + /* Clear subunit states on head movement */ + memset(&engine->hangcheck.instdone, 0, + sizeof(engine->hangcheck.instdone)); + + return HANGCHECK_ACTIVE; + } + + if (!subunits_stuck(engine)) + return HANGCHECK_ACTIVE; + + return HANGCHECK_HUNG; +} + +static enum intel_engine_hangcheck_action +engine_stuck(struct intel_engine_cs *engine, u64 acthd) +{ + struct drm_i915_private *dev_priv = engine->i915; + enum intel_engine_hangcheck_action ha; + u32 tmp; + + ha = head_stuck(engine, acthd); + if (ha != HANGCHECK_HUNG) + return ha; + + if (IS_GEN2(dev_priv)) + return HANGCHECK_HUNG; + + /* Is the chip hanging on a WAIT_FOR_EVENT? + * If so we can simply poke the RB_WAIT bit + * and break the hang. This should work on + * all but the second generation chipsets. + */ + tmp = I915_READ_CTL(engine); + if (tmp & RING_WAIT) { + i915_handle_error(dev_priv, 0, + "Kicking stuck wait on %s", + engine->name); + I915_WRITE_CTL(engine, tmp); + return HANGCHECK_KICK; + } + + if (INTEL_GEN(dev_priv) >= 6 && tmp & RING_WAIT_SEMAPHORE) { + switch (semaphore_passed(engine)) { + default: + return HANGCHECK_HUNG; + case 1: + i915_handle_error(dev_priv, 0, + "Kicking stuck semaphore on %s", + engine->name); + I915_WRITE_CTL(engine, tmp); + return HANGCHECK_KICK; + case 0: + return HANGCHECK_WAIT; + } + } + + return HANGCHECK_HUNG; +} + +/* + * This is called when the chip hasn't reported back with completed + * batchbuffers in a long time. We keep track per ring seqno progress and + * if there are no progress, hangcheck score for that ring is increased. + * Further, acthd is inspected to see if the ring is stuck. On stuck case + * we kick the ring. If we see no progress on three subsequent calls + * we assume chip is wedged and try to fix it by resetting the chip. + */ +static void i915_hangcheck_elapsed(struct work_struct *work) +{ + struct drm_i915_private *dev_priv = + container_of(work, typeof(*dev_priv), + gpu_error.hangcheck_work.work); + struct intel_engine_cs *engine; + enum intel_engine_id id; + unsigned int hung = 0, stuck = 0; + int busy_count = 0; +#define BUSY 1 +#define KICK 5 +#define HUNG 20 +#define ACTIVE_DECAY 15 + + if (!i915.enable_hangcheck) + return; + + if (!READ_ONCE(dev_priv->gt.awake)) + return; + + /* As enabling the GPU requires fairly extensive mmio access, + * periodically arm the mmio checker to see if we are triggering + * any invalid access. + */ + intel_uncore_arm_unclaimed_mmio_detection(dev_priv); + + for_each_engine(engine, dev_priv, id) { + bool busy = intel_engine_has_waiter(engine); + u64 acthd; + u32 seqno; + u32 submit; + + semaphore_clear_deadlocks(dev_priv); + + /* We don't strictly need an irq-barrier here, as we are not + * serving an interrupt request, be paranoid in case the + * barrier has side-effects (such as preventing a broken + * cacheline snoop) and so be sure that we can see the seqno + * advance. If the seqno should stick, due to a stale + * cacheline, we would erroneously declare the GPU hung. + */ + if (engine->irq_seqno_barrier) + engine->irq_seqno_barrier(engine); + + acthd = intel_engine_get_active_head(engine); + seqno = intel_engine_get_seqno(engine); + submit = intel_engine_last_submit(engine); + + if (engine->hangcheck.seqno == seqno) { + if (i915_seqno_passed(seqno, submit)) { + engine->hangcheck.action = HANGCHECK_IDLE; + } else { + /* We always increment the hangcheck score + * if the engine is busy and still processing + * the same request, so that no single request + * can run indefinitely (such as a chain of + * batches). The only time we do not increment + * the hangcheck score on this ring, if this + * engine is in a legitimate wait for another + * engine. In that case the waiting engine is a + * victim and we want to be sure we catch the + * right culprit. Then every time we do kick + * the ring, add a small increment to the + * score so that we can catch a batch that is + * being repeatedly kicked and so responsible + * for stalling the machine. + */ + engine->hangcheck.action = + engine_stuck(engine, acthd); + + switch (engine->hangcheck.action) { + case HANGCHECK_IDLE: + case HANGCHECK_WAIT: + break; + case HANGCHECK_ACTIVE: + engine->hangcheck.score += BUSY; + break; + case HANGCHECK_KICK: + engine->hangcheck.score += KICK; + break; + case HANGCHECK_HUNG: + engine->hangcheck.score += HUNG; + break; + } + } + + if (engine->hangcheck.score >= HANGCHECK_SCORE_RING_HUNG) { + hung |= intel_engine_flag(engine); + if (engine->hangcheck.action != HANGCHECK_HUNG) + stuck |= intel_engine_flag(engine); + } + } else { + engine->hangcheck.action = HANGCHECK_ACTIVE; + + /* Gradually reduce the count so that we catch DoS + * attempts across multiple batches. + */ + if (engine->hangcheck.score > 0) + engine->hangcheck.score -= ACTIVE_DECAY; + if (engine->hangcheck.score < 0) + engine->hangcheck.score = 0; + + /* Clear head and subunit states on seqno movement */ + acthd = 0; + + memset(&engine->hangcheck.instdone, 0, + sizeof(engine->hangcheck.instdone)); + } + + engine->hangcheck.seqno = seqno; + engine->hangcheck.acthd = acthd; + busy_count += busy; + } + + if (hung) { + char msg[80]; + unsigned int tmp; + int len; + + /* If some rings hung but others were still busy, only + * blame the hanging rings in the synopsis. + */ + if (stuck != hung) + hung &= ~stuck; + len = scnprintf(msg, sizeof(msg), + "%s on ", stuck == hung ? "No progress" : "Hang"); + for_each_engine_masked(engine, dev_priv, hung, tmp) + len += scnprintf(msg + len, sizeof(msg) - len, + "%s, ", engine->name); + msg[len-2] = '\0'; + + return i915_handle_error(dev_priv, hung, msg); + } + + /* Reset timer in case GPU hangs without another request being added */ + if (busy_count) + i915_queue_hangcheck(dev_priv); +} + +void intel_engine_init_hangcheck(struct intel_engine_cs *engine) +{ + memset(&engine->hangcheck, 0, sizeof(engine->hangcheck)); +} + +void intel_hangcheck_init(struct drm_i915_private *i915) +{ + INIT_DELAYED_WORK(&i915->gpu_error.hangcheck_work, + i915_hangcheck_elapsed); +}