drm/i915/guc: Fix lockdep due to log relay channel handling under struct_mutex
Subsystems (2)
PATCH 1/4 drm/i915/guc: Fix lockdep due to log relay channel handling under struct_mutex
This patch fixes lockdep issue due to circular locking dependency of struct_mutex, i_mutex_key, mmap_sem, relay_channels_mutex.
For GuC log relay channel we create debugfs file that requires i_mutex_key lock and we are doing that under struct_mutex. So we introduced newer dependency as:
&dev->struct_mutex --> &sb->s_type->i_mutex_key#3 --> &mm->mmap_sem
However, there is dependency from mmap_sem to struct_mutex. Hence we separate the relay create/destroy operation from under struct_mutex.
======================================================
WARNING: possible circular locking dependency detected
4.15.0-rc6-CI-Patchwork_7614+ #1 Not tainted
------------------------------------------------------
debugfs_test/1388 is trying to acquire lock:
(&dev->struct_mutex){+.+.}, at: [<00000000d5e1d915>] i915_mutex_lock_interruptible+0x47/0x130 [i915]
but task is already holding lock:
(&mm->mmap_sem){++++}, at: [<0000000029a9c131>] __do_page_fault+0x106/0x560
which lock already depends on the new lock.
the existing dependency chain (in reverse order) is:
-> #3 (&mm->mmap_sem){++++}:
_copy_to_user+0x1e/0x70
filldir+0x8c/0xf0
dcache_readdir+0xeb/0x160
iterate_dir+0xdc/0x140
SyS_getdents+0xa0/0x130
entry_SYSCALL_64_fastpath+0x1c/0x89-> #2 (&sb->s_type->i_mutex_key#3){++++}:
start_creating+0x59/0x110
__debugfs_create_file+0x2e/0xe0
relay_create_buf_file+0x62/0x80
relay_late_setup_files+0x84/0x250
guc_log_late_setup+0x4f/0x110 [i915]
i915_guc_log_register+0x32/0x40 [i915]
i915_driver_load+0x7b6/0x1720 [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-> #1 (relay_channels_mutex){+.+.}:
relay_open+0x12c/0x2b0
intel_guc_log_runtime_create+0xab/0x230 [i915]
intel_guc_init+0x81/0x120 [i915]
intel_uc_init+0x29/0xa0 [i915]
i915_gem_init+0x182/0x530 [i915]
i915_driver_load+0xaa9/0x1720 [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){+.+.}:
__mutex_lock+0x81/0x9b0
i915_mutex_lock_interruptible+0x47/0x130 [i915]
i915_gem_fault+0x201/0x790 [i915]
__do_fault+0x15/0x70
__handle_mm_fault+0x677/0xdc0
handle_mm_fault+0x14f/0x2f0
__do_page_fault+0x2d1/0x560
page_fault+0x4c/0x60other info that might help us debug this:
Chain exists of: &dev->struct_mutex --> &sb->s_type->i_mutex_key#3 --> &mm->mmap_sem Possible unsafe locking scenario:
CPU0 CPU1
---- ----
lock(&mm->mmap_sem);
lock(&sb->s_type->i_mutex_key#3);
lock(&mm->mmap_sem); lock(&dev->struct_mutex);
*** DEADLOCK ***
1 lock held by debugfs_test/1388:
#0: (&mm->mmap_sem){++++}, at: [<0000000029a9c131>] __do_page_fault+0x106/0x560
stack backtrace: CPU: 2 PID: 1388 Comm: debugfs_test Not tainted 4.15.0-rc6-CI-Patchwork_7614+ #1 Hardware name: To Be Filled By O.E.M. To Be Filled By O.E.M./J4205-ITX, BIOS P1.10 09/29/2016 Call Trace: dump_stack+0x5f/0x86 print_circular_bug.isra.18+0x1d0/0x2c0 __lock_acquire+0x14ae/0x1b60 ? lock_acquire+0xaf/0x200 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+0x201/0x790 [i915] __do_fault+0x15/0x70 ? _raw_spin_unlock+0x29/0x40 __handle_mm_fault+0x677/0xdc0 handle_mm_fault+0x14f/0x2f0 __do_page_fault+0x2d1/0x560 ? page_fault+0x36/0x60 page_fault+0x4c/0x60
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=104693
Testcase: igt/debugfs_test/read_all_entries # with enable_guc=1 and guc_log_level=1
--- drivers/gpu/drm/i915/i915_debugfs.c | 12 +--- drivers/gpu/drm/i915/i915_drv.c | 2 +- drivers/gpu/drm/i915/i915_gem.c | 4 +- drivers/gpu/drm/i915/intel_guc.c | 6 +- drivers/gpu/drm/i915/intel_guc_log.c | 111 ++++++++++++++++++++++++++--------- drivers/gpu/drm/i915/intel_guc_log.h | 9 +++ drivers/gpu/drm/i915/intel_uc.c | 26 ++++++-- drivers/gpu/drm/i915/intel_uc.h | 4 +- 8 files changed, 125 insertions(+), 49 deletions(-)
drivers/gpu/drm/i915/i915_debugfs.c +1 -11
@@ -2467,7 +2467,6 @@ static int i915_guc_log_control_get(void *data, u64 *val)static int i915_guc_log_control_set (void *data,u64 val) {struct drm_i915_private *dev_priv = data; -int ret;if (!HAS_GUC (dev_priv))return -ENODEV ;
@@ -2475,16 +2474,7 @@ static int i915_guc_log_control_set(void *data, u64 val)if (!dev_priv->guc.log.vma)return -EINVAL ; - ret =mutex_lock_interruptible (&dev_priv->drm.struct_mutex); -if (ret) -return ret; - -intel_runtime_pm_get (dev_priv); - ret =i915_guc_log_control (dev_priv, val); -intel_runtime_pm_put (dev_priv); - -mutex_unlock (&dev_priv->drm.struct_mutex); -return ret; +return i915_guc_log_control (dev_priv, val); } DEFINE_SIMPLE_ATTRIBUTE(i915_guc_log_control_fops,
drivers/gpu/drm/i915/i915_drv.c +1 -1
@@ -626,7 +626,7 @@ 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); +intel_uc_fini_misc (dev_priv);i915_gem_cleanup_userptr (dev_priv);i915_gem_drain_freed_objects (dev_priv);
drivers/gpu/drm/i915/i915_gem.c +2 -2
@@ -5193,7 +5193,7 @@ int i915_gem_init(struct drm_i915_private *dev_priv)if (ret)return ret; - ret =intel_uc_init_wq (dev_priv); + ret =intel_uc_init_misc (dev_priv);if (ret)return ret;
@@ -5289,7 +5289,7 @@ int i915_gem_init(struct drm_i915_private *dev_priv)intel_uncore_forcewake_put (dev_priv,FORCEWAKE_ALL );mutex_unlock (&dev_priv->drm.struct_mutex); -intel_uc_fini_wq (dev_priv); +intel_uc_fini_misc (dev_priv);if (ret != -EIO )i915_gem_cleanup_userptr (dev_priv);
drivers/gpu/drm/i915/intel_guc.c +5 -1
@@ -87,8 +87,10 @@ int intel_guc_init_wq(struct intel_guc *guc)*/ guc->log.runtime.flush_wq =alloc_ordered_workqueue ("i915-guc_log" ,WQ_HIGHPRI |WQ_FREEZABLE ); -if (!guc->log.runtime.flush_wq) +if (!guc->log.runtime.flush_wq) { +DRM_ERROR ("Couldn't allocate workqueue for GuC log\n" );return -ENOMEM ; + }/* * Even though both sending GuC action, and adding a new workitem to
@@ -109,6 +111,8 @@ int intel_guc_init_wq(struct intel_guc *guc) WQ_HIGHPRI);if (!guc->preempt_wq) {destroy_workqueue (guc->log.runtime.flush_wq); +DRM_ERROR ("Couldn't allocate workqueue for GuC " +"preemption\n" );return -ENOMEM ; } }
drivers/gpu/drm/i915/intel_guc_log.c +84 -27
229 lines not on this page · raw message
drivers/gpu/drm/i915/intel_guc_log.h +9 -0
@@ -32,6 +32,13 @@struct drm_i915_private ;struct intel_guc ; +/* +* The first page is to save log buffer state. Allocate one +* extra page for others in case for overlap +*/ +#define GUC_LOG_SIZE ((1 + GUC_LOG_DPC_PAGES + 1 + GUC_LOG_ISR_PAGES + \ + 1 + GUC_LOG_CRASH_PAGES + 1) << PAGE_SHIFT) +struct intel_guc_log {u32 flags;struct i915_vma *vma;
@@ -52,6 +59,8 @@ struct intel_guc_log {int intel_guc_log_create (struct intel_guc *guc);void intel_guc_log_destroy (struct intel_guc *guc); +int intel_guc_log_relay_create (struct intel_guc *guc); +void intel_guc_log_relay_destroy (struct intel_guc *guc);int i915_guc_log_control (struct drm_i915_private *dev_priv,u64 control_val);void i915_guc_log_register (struct drm_i915_private *dev_priv);void i915_guc_log_unregister (struct drm_i915_private *dev_priv);
drivers/gpu/drm/i915/intel_uc.c +21 -5
@@ -236,28 +236,44 @@ 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 intel_uc_init_misc (struct drm_i915_private *dev_priv) { +struct intel_guc *guc = &dev_priv->guc;int ret;if (!USES_GUC (dev_priv))return 0 ; - ret =intel_guc_init_wq (&dev_priv->guc); + ret =intel_guc_init_wq (guc);if (ret) {DRM_ERROR ("Couldn't allocate workqueues for GuC\n" ); -return ret; + goto err; + } + + ret =intel_guc_log_relay_create (guc); +if (ret) { +DRM_ERROR ("Couldn't allocate relay for GuC log\n" ); + goto err_relay; }return 0 ; + +err_relay: +intel_guc_fini_wq (guc); +err: +return ret; } -void intel_uc_fini_wq (struct drm_i915_private *dev_priv) +void intel_uc_fini_misc (struct drm_i915_private *dev_priv) { +struct intel_guc *guc = &dev_priv->guc; +if (!USES_GUC (dev_priv))return ; -intel_guc_fini_wq (&dev_priv->guc); +intel_guc_fini_wq (guc); + +intel_guc_log_relay_destroy (guc); }int intel_uc_init (struct drm_i915_private *dev_priv)
drivers/gpu/drm/i915/intel_uc.h +2 -2
@@ -33,8 +33,8 @@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_misc (struct drm_i915_private *dev_priv); +void intel_uc_fini_misc (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);int intel_uc_init (struct drm_i915_private *dev_priv);
Signature · 6 lines
1.9.1 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH 2/4 drm/i915/guc: Grab RPM wakelock while disabling GuC interrupts
Disabling GuC interrupts involves access to GuC IRQ control registers hence ensure device is RPM awake.
v2: Add comment about need to synchronize flush work and log runtime
destroy
v3: Moved patch earlier in the series and removed comment about future work. (Tvrtko)
v4-v5: Rebase.
--- drivers/gpu/drm/i915/intel_guc_log.c | 3 +++ 1 file changed, 3 insertions(+)
drivers/gpu/drm/i915/intel_guc_log.c +3 -0
@@ -694,7 +694,10 @@ void i915_guc_log_unregister(struct drm_i915_private *dev_priv)mutex_lock (&dev_priv->drm.struct_mutex);/* GuC logging is currently the only user of Guc2Host interrupts */ +intel_runtime_pm_get (dev_priv);gen9_disable_guc_interrupts (dev_priv); +intel_runtime_pm_put (dev_priv); +guc_log_runtime_destroy (guc);mutex_unlock (&dev_priv->drm.struct_mutex);
Signature · 6 lines
1.9.1 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH 2/4] drm/i915/guc: Grab RPM wakelock while disabling GuC interrupts
Quoting Sagar Arun Kamble (2018-01-22 08:26:02)
Quoted 26 lines
Disabling GuC interrupts involves access to GuC IRQ control registers hence ensure device is RPM awake.
v2: Add comment about need to synchronize flush work and log runtime
destroyv3: Moved patch earlier in the series and removed comment about future work. (Tvrtko)
v4-v5: Rebase.
Signed-off-by Sagar Arun Kamble <[email protected]>Cc Michal Wajdeczko <[email protected]>Cc Daniele Ceraolo Spurio <[email protected]>Cc Tvrtko Ursulin <[email protected]>Cc Chris Wilson <[email protected]>Cc Joonas Lahtinen <[email protected]>--- drivers/gpu/drm/i915/intel_guc_log.c | 3 +++ 1 file changed, 3 insertions(+)drivers/gpu/drm/i915/intel_guc_log.c +0
-0
drivers/gpu/drm/i915/intel_guc_log.c void i915_guc_log_unregister(struct drm_i915_private *dev_priv) · line 698
Quoted 5 lines
drivers/gpu/drm/i915/intel_guc_log.c +1
-0@@ -694,7 +694,10 @@ void i915_guc_log_unregister(struct drm_i915_private *dev_priv)mutex_lock (&dev_priv->drm.struct_mutex);/* GuC logging is currently the only user of Guc2Host interrupts */ +intel_runtime_pm_get (dev_priv);gen9_disable_guc_interrupts (dev_priv);
Do we have/want assert_rpm_wakelock_held() ingen9_disable_guc_interrupts() ?
Are we really locking the interrupts on/off with struct_mutex? -Chris
_______________________________________________
Intel-gfx mailing list [email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH 2/4] drm/i915/guc: Grab RPM wakelock while disabling GuC interrupts
Quoted 28 lines · Chris Wilson
Quoting Sagar Arun Kamble (2018-01-22 08:26:02)
Quoted 27 lines
Disabling GuC interrupts involves access to GuC IRQ control registers hence ensure device is RPM awake.
v2: Add comment about need to synchronize flush work and log runtime
destroyv3: Moved patch earlier in the series and removed comment about future work. (Tvrtko)
v4-v5: Rebase.
Signed-off-by Sagar Arun Kamble <[email protected]>Cc Michal Wajdeczko <[email protected]>Cc Daniele Ceraolo Spurio <[email protected]>Cc Tvrtko Ursulin <[email protected]>Cc Chris Wilson <[email protected]>Cc Joonas Lahtinen <[email protected]>--- drivers/gpu/drm/i915/intel_guc_log.c | 3 +++ 1 file changed, 3 insertions(+)drivers/gpu/drm/i915/intel_guc_log.c +0
-0@@ -694,7 +694,10 @@ void i915_guc_log_unregister(struct drm_i915_private *dev_priv)
Quoted 6 lines
drivers/gpu/drm/i915/intel_guc_log.c void i915_guc_log_unregister(struct drm_i915_private *dev_priv) · line 698
Quoted 4 lines
drivers/gpu/drm/i915/intel_guc_log.c +1
-0mutex_lock (&dev_priv->drm.struct_mutex);/* GuC logging is currently the only user of Guc2Host interrupts */ +intel_runtime_pm_get (dev_priv);gen9_disable_guc_interrupts (dev_priv);Do we have/want
assert_rpm_wakelock_held()ingen9_disable_guc_interrupts()?
We don't have. Will add it.
Quoted 1 line
Are we really locking the interrupts on/off with
struct_mutex?
Yes.
Quoted 1 line
-Chris
Thanks,
Sagar
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH 3/4 drm/i915/guc: Enable interrupts before resuming GuC during runtime resume
In reply to Sagar Arun KambleGuC log streaming needs interrupts enabled prior to GuC resume but runtime pm interrupt setup was happening post GuC resume. Fix it. While at it, fix the unwinding of steps in the runtime suspend path.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=104695
--- drivers/gpu/drm/i915/i915_drv.c | 13 +++++++++---- 1 file changed, 9 insertions(+), 4 deletions(-)
drivers/gpu/drm/i915/i915_drv.c +9 -4
@@ -2600,6 +2600,11 @@ static int intel_runtime_suspend(struct device *kdev)intel_runtime_pm_enable_interrupts (dev_priv); +intel_guc_resume (dev_priv); + +i915_gem_init_swizzling (dev_priv); +i915_gem_restore_fences (dev_priv); +enable_rpm_wakeref_asserts (dev_priv);return ret;
@@ -2665,8 +2670,6 @@ static int intel_runtime_resume(struct device *kdev)if (intel_uncore_unclaimed_mmio (dev_priv))DRM_DEBUG_DRIVER ("Unclaimed access during suspend, bios?\n" ); -intel_guc_resume (dev_priv); -if (IS_GEN9_LP (dev_priv)) {bxt_disable_dc9 (dev_priv);bxt_display_core_init (dev_priv, true);
@@ -2681,6 +2684,10 @@ static int intel_runtime_resume(struct device *kdev)intel_uncore_runtime_resume (dev_priv); +intel_runtime_pm_enable_interrupts (dev_priv); + +intel_guc_resume (dev_priv); +/* * No point of rolling back things in case of an error, as the best * we can do is to hope that things will still work (and disable RPM).
@@ -2688,8 +2695,6 @@ static int intel_runtime_resume(struct device *kdev)i915_gem_init_swizzling (dev_priv);i915_gem_restore_fences (dev_priv); -intel_runtime_pm_enable_interrupts (dev_priv); -/* * On VLV/CHV display interrupts are part of the display * power well, so hpd is reinitialized from there. For
Signature · 6 lines
1.9.1 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH 3/4] drm/i915/guc: Enable interrupts before resuming GuC during runtime resume
Quoting Sagar Arun Kamble (2018-01-22 08:26:03)
Quoted 11 lines
GuC log streaming needs interrupts enabled prior to GuC resume but runtime pm interrupt setup was happening post GuC resume. Fix it. While at it, fix the unwinding of steps in the runtime suspend path.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=104695
Signed-off-by Sagar Arun Kamble <[email protected]>Cc Chris Wilson <[email protected]>Cc Michal Wajdeczko <[email protected]>Cc Michał Winiarski <[email protected]>Cc Joonas Lahtinen <[email protected]>Cc Marta Lofstedt <[email protected]>
Ok, moving the interrupts around shouldn't be too much of an issue during rpm suspend/resume as we hold the rpm wakeref for the operation.
-Chris
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH 4/4 Revert "drm/i915/guc: Keep GuC log disabled by default"
In reply to Sagar Arun KambleComment DRM_ERROR_RATELIMIT outputs as log capturer won't be running.
This reverts commit bd724318b682587ad2f989ab8e0f7b3d4486ced5.
--- drivers/gpu/drm/i915/i915_params.h | 2 +- drivers/gpu/drm/i915/intel_guc_log.c | 4 ++-- 2 files changed, 3 insertions(+), 3 deletions(-)
drivers/gpu/drm/i915/i915_params.h +1 -1
@@ -48,7 +48,7 @@ param(int, enable_ips, 1) \ param(int, invert_brightness, 0) \ param(int, enable_guc, 0) \ - param(int, guc_log_level, 0) \ + param(int, guc_log_level, -1) \ param(char *, guc_firmware_path, NULL) \ param(char *, huc_firmware_path, NULL) \param (int, mmio_debug,0 ) \
drivers/gpu/drm/i915/intel_guc_log.c +2 -2
@@ -228,7 +228,7 @@ static bool guc_check_log_buf_overflow(struct intel_guc *guc,/* buffer_full_cnt is a 4 bit counter */ guc->log.total_overflow_count[type] +=16 ; } -DRM_ERROR_RATELIMITED ("GuC log buffer overflow\n" ); +//DRM_ERROR_RATELIMITED("GuC log buffer overflow\n"); }return overflow;
@@ -341,7 +341,7 @@ static void guc_read_update_log_buffer(struct intel_guc *guc)/* Used rate limited to avoid deluge of messages, logs might be * getting consumed by User at a slow rate. */ -DRM_ERROR_RATELIMITED ("no sub-buffer to capture logs\n" ); +//DRM_ERROR_RATELIMITED("no sub-buffer to capture logs\n"); guc->log.capture_miss_count++; } }
Signature · 6 lines
1.9.1 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
I recommend that this is tested with a HACK to enable the GUC logs again, so that we can see if it really fixes the issue.
Quoted 632 lines
-----Original Message-----
From: Kamble, Sagar A
Sent: Monday, January 22, 2018 10:26 AM
To: [email protected]
Cc: Kamble, Sagar A <[email protected]>; Wajdeczko, Michal <[email protected]>; Ceraolo Spurio, Daniele <[email protected]>; Ursulin, Tvrtko <[email protected]>; Chris Wilson <[email protected]>; Joonas Lahtinen <[email protected]>; Lofstedt, Marta <[email protected]>; Winiarski, Michal <[email protected]>
Subject: [PATCH 1/4] drm/i915/guc: Fix lockdep due to log relay channel handling understruct_mutexThis patch fixes lockdep issue due to circular locking dependency of
struct_mutex,i_mutex_key,mmap_sem,relay_channels_mutex.
For GuC log relay channel we create debugfs file that requiresi_mutex_keylock and we are doing that understruct_mutex. So we introduced newer dependency as:&dev->struct_mutex --> &sb->s_type->i_mutex_key#3 --> &mm-Quoted 1 line
mmap_semHowever, there is dependency frommmap_semto
struct_mutex. Hence we separate the relay create/destroy operation from understruct_mutex.======================================================
WARNING: possible circular locking dependency detected 4.15.0-rc6-CI- Patchwork_7614+ #1 Not tainted
------------------------------------------------------
debugfs_test/1388 is trying to acquire lock:
(&dev->struct_mutex){+.+.}, at: [<00000000d5e1d915>]i915_mutex_lock_interruptible+0x47/0x130[i915]but task is already holding lock:
(&mm->mmap_sem){++++}, at: [<0000000029a9c131>]__do_page_fault+0x106/0x560which lock already depends on the new lock.
the existing dependency chain (in reverse order) is:
-> #3 (&mm->mmap_sem){++++}:
_copy_to_user+0x1e/0x70 filldir+0x8c/0xf0 dcache_readdir+0xeb/0x160 iterate_dir+0xdc/0x140 SyS_getdents+0xa0/0x130 entry_SYSCALL_64_fastpath+0x1c/0x89-> #2 (&sb->s_type->i_mutex_key#3){++++}:
start_creating+0x59/0x110 __debugfs_create_file+0x2e/0xe0 relay_create_buf_file+0x62/0x80 relay_late_setup_files+0x84/0x250 guc_log_late_setup+0x4f/0x110 [i915] i915_guc_log_register+0x32/0x40 [i915] i915_driver_load+0x7b6/0x1720 [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-> #1 (relay_channels_mutex){+.+.}:
relay_open+0x12c/0x2b0 intel_guc_log_runtime_create+0xab/0x230 [i915] intel_guc_init+0x81/0x120 [i915] intel_uc_init+0x29/0xa0 [i915] i915_gem_init+0x182/0x530 [i915] i915_driver_load+0xaa9/0x1720 [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){+.+.}:
__mutex_lock+0x81/0x9b0 i915_mutex_lock_interruptible+0x47/0x130 [i915] i915_gem_fault+0x201/0x790 [i915] __do_fault+0x15/0x70 __handle_mm_fault+0x677/0xdc0 handle_mm_fault+0x14f/0x2f0 __do_page_fault+0x2d1/0x560 page_fault+0x4c/0x60other info that might help us debug this:
Chain exists of: &dev->struct_mutex --> &sb->s_type->i_mutex_key#3 --> &mm-Quoted 1 line
mmap_semPossible unsafe locking scenario:CPU0 CPU1 ---- ---- lock(&mm->mmap_sem); lock(&sb->s_type->i_mutex_key#3); lock(&mm->mmap_sem);
lock(&dev->struct_mutex);*** DEADLOCK ***
1 lock held by debugfs_test/1388:
#0: (&mm->mmap_sem){++++}, at: [<0000000029a9c131>]__do_page_fault+0x106/0x560 stack backtrace: CPU: 2 PID: 1388 Comm: debugfs_test Not tainted 4.15.0-rc6-CI- Patchwork_7614+ #1 Hardware name: To Be Filled By O.E.M. To Be Filled By O.E.M./J4205-ITX, BIOS P1.10 09/29/2016 Call Trace: dump_stack+0x5f/0x86 print_circular_bug.isra.18+0x1d0/0x2c0 __lock_acquire+0x14ae/0x1b60 ? lock_acquire+0xaf/0x200 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+0x201/0x790 [i915] __do_fault+0x15/0x70 ? _raw_spin_unlock+0x29/0x40 __handle_mm_fault+0x677/0xdc0 handle_mm_fault+0x14f/0x2f0 __do_page_fault+0x2d1/0x560 ? page_fault+0x36/0x60 page_fault+0x4c/0x60Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=104693
Testcase: igt/debugfs_test/read_all_entries # with enable_guc=1 and guc_log_level=1Signed-off-by Sagar Arun Kamble <[email protected]>Cc Michal Wajdeczko <[email protected]>Cc Daniele Ceraolo Spurio <[email protected]>Cc Tvrtko Ursulin <[email protected]>Cc Chris Wilson <[email protected]>Cc Joonas Lahtinen <[email protected]>Cc Marta Lofstedt <[email protected]>Cc Michal Winiarski <[email protected]>--- drivers/gpu/drm/i915/i915_debugfs.c | 12 +--- drivers/gpu/drm/i915/i915_drv.c | 2 +- drivers/gpu/drm/i915/i915_gem.c | 4 +- drivers/gpu/drm/i915/intel_guc.c | 6 +- drivers/gpu/drm/i915/intel_guc_log.c | 111 ++++++++++++++++++++++++++--------- drivers/gpu/drm/i915/intel_guc_log.h | 9 +++ drivers/gpu/drm/i915/intel_uc.c | 26 ++++++-- drivers/gpu/drm/i915/intel_uc.h | 4 +- 8 files changed, 125 insertions(+), 49 deletions(-)drivers/gpu/drm/i915/i915_debugfs.c +1
-11@@ -2467,7 +2467,6 @@ static int i915_guc_log_control_get(void *data, u64 *val) static int i915_guc_log_control_set(void *data, u64 val) {struct drm_i915_private *dev_priv = data; -int ret;if (!HAS_GUC (dev_priv))return -ENODEV ;@@ -2475,16 +2474,7 @@ static int i915_guc_log_control_set(void *data, u64 val)if (!dev_priv->guc.log.vma)return -EINVAL ; - ret =mutex_lock_interruptible (&dev_priv- >drm.struct_mutex); -if (ret) -return ret; - -intel_runtime_pm_get (dev_priv); - ret =i915_guc_log_control (dev_priv, val); -intel_runtime_pm_put (dev_priv); - -mutex_unlock (&dev_priv->drm.struct_mutex); -return ret; +return i915_guc_log_control (dev_priv, val); }DEFINE_SIMPLE_ATTRIBUTE(i915_guc_log_control_fops,
(quoted diff) +0
-0
b/drivers/gpu/drm/i915/i915_drv.cindex 1fbe378..29048b9 100644drivers/gpu/drm/i915/i915_drv.c +1
-1@@ -626,7 +626,7 @@ 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); +intel_uc_fini_misc (dev_priv);i915_gem_cleanup_userptr (dev_priv);drivers/gpu/drm/i915/i915_drv.c +0
-0i915_gem_drain_freed_objects (dev_priv);(quoted diff) +0
-0
b/drivers/gpu/drm/i915/i915_gem.cindex 7f0684c..34d0d51 100644drivers/gpu/drm/i915/i915_gem.c +2
-2@@ -5193,7 +5193,7 @@ int i915_gem_init(struct drm_i915_private *dev_priv)if (ret)return ret; - ret =intel_uc_init_wq (dev_priv); + ret =intel_uc_init_misc (dev_priv);if (ret)return ret;@@ -5289,7 +5289,7 @@ int i915_gem_init(struct drm_i915_private *dev_priv)intel_uncore_forcewake_put (dev_priv,FORCEWAKE_ALL );mutex_unlock (&dev_priv->drm.struct_mutex); -intel_uc_fini_wq (dev_priv); +intel_uc_fini_misc (dev_priv);if (ret != -EIO )drivers/gpu/drm/i915/i915_gem.c +0
-0i915_gem_cleanup_userptr (dev_priv);drivers/gpu/drm/i915/intel_guc.c +3
-1@@ -87,8 +87,10 @@ int intel_guc_init_wq(struct intel_guc *guc)*/ guc->log.runtime.flush_wq =alloc_ordered_workqueue ("i915-guc_log" ,WQ_HIGHPRI |WQ_FREEZABLE ); -if (!guc->log.runtime.flush_wq) +if (!guc->log.runtime.flush_wq) { + DRM_ERROR("Couldn't allocate workqueue for GuC log\n"); return -ENOMEM; + }drivers/gpu/drm/i915/intel_guc.c +0
-0/* * Even though both sending GuC action, and adding a new workitem to @@ -109,6 +111,8 @@ int
intel_guc_init_wq(struct intel_guc *guc)WQ_HIGHPRI);drivers/gpu/drm/i915/intel_guc.c +0
-0if (!guc->preempt_wq) { destroy_workqueue(guc-Quoted 1 line
log.runtime.flush_wq);
drivers/gpu/drm/i915/intel_guc.c +1
-0+ DRM_ERROR("Couldn't allocateworkqueue for GuC "
drivers/gpu/drm/i915/intel_guc.c +1
-0+ "preemption\n");return -ENOMEM ; } }drivers/gpu/drm/i915/intel_guc_log.c +5
-2@@ -359,16 +359,21 @@ static bool guc_log_has_runtime(struct intel_guc *guc)return guc->log.runtime.buf_addr !=NULL ; } +static bool guc_log_has_relay (struct intel_guc *guc) { +return guc->log.runtime.relay_chan !=NULL ; } +static int guc_log_runtime_create (struct intel_guc *guc) {struct drm_i915_private *dev_priv =guc_to_i915 (guc);void *vaddr; -struct rchan *guc_log_relay_chan; -size_t n_subbufs, subbuf_size;int ret;lockdep_assert_held (&dev_priv->drm.struct_mutex); +GEM_BUG_ON (!guc_log_has_relay (guc)); +GEM_BUG_ON (guc_log_has_runtime (guc)); ret = i915_gem_object_set_to_wc_domain(guc->log.vma-Quoted 1 line
obj, true); @@ -387,8 +392,38 @@ static int guc_log_runtime_create(struct
intel_guc*guc)drivers/gpu/drm/i915/intel_guc_log.c +1
-0guc->log.runtime.buf_addr = vaddr; + INIT_WORK(&guc->log.runtime.flush_work,
capture_logs_work);drivers/gpu/drm/i915/intel_guc_log.c +7
-0+ +return 0 ; +} + +static void guc_log_runtime_destroy (struct intel_guc *guc) { +/* +* It's possible that the runtime stuff was never allocated because
drivers/gpu/drm/i915/intel_guc_log.c +21
-1+* GuC log was disabled at the boot time. +**/ +if (!guc_log_has_runtime (guc)) +return ; + +i915_gem_object_unpin_map (guc->log.vma->obj); + guc->log.runtime.buf_addr =NULL ; +} + +int intel_guc_log_relay_create (struct intel_guc *guc) { +struct drm_i915_private *dev_priv =guc_to_i915 (guc); +struct rchan *guc_log_relay_chan; +size_t n_subbufs, subbuf_size; +int ret; + +if (!i915_modparams.guc_log_level) +return 0 ; + +GEM_BUG_ON (guc_log_has_relay (guc)); +/* Keep the size of sub buffers same as shared log buffer */ - subbuf_size = guc->log.vma->obj->base.size; + subbuf_size =GUC_LOG_SIZE ;/* Store up to 8 snapshots, which is large enough to buffer sufficient
* boot time logs and provides enough leeway to User, in
terms of @@ -407,33 +442,31 @@ static intguc_log_runtime_create(struct intel_guc *guc)
DRM_ERROR("Couldn't create relay chan for
GuC logging\n");drivers/gpu/drm/i915/intel_guc_log.c +1
-1ret = -ENOMEM ; - goto err_vaddr; + goto err; } GEM_BUG_ON(guc_log_relay_chan->subbuf_size <
subbuf_size);drivers/gpu/drm/i915/intel_guc_log.c +0
-1guc->log.runtime.relay_chan = guc_log_relay_chan; - INIT_WORK(&guc->log.runtime.flush_work,
capture_logs_work);drivers/gpu/drm/i915/intel_guc_log.c +4
-5return 0 ; -err_vaddr: -i915_gem_object_unpin_map (guc->log.vma->obj); - guc->log.runtime.buf_addr =NULL ; +err: +/* logging will be off */ + i915_modparams.guc_log_level =0 ;return ret; } -static void guc_log_runtime_destroy (struct intel_guc *guc) +void intel_guc_log_relay_destroy (struct intel_guc *guc) {/* -* It's possible that the runtime stuff was never allocated because
drivers/gpu/drm/i915/intel_guc_log.c +3
-3+* It's possible that the relay was never allocated because * GuC log was disabled at the boot time. */ -if (!guc_log_has_runtime (guc)) +if (!guc_log_has_relay (guc))return ;relay_close (guc->log.runtime.relay_chan); -i915_gem_object_unpin_map (guc->log.vma->obj); - guc->log.runtime.buf_addr =NULL ; + guc->log.runtime.relay_chan =NULL ; } static int guc_log_late_setup(struct intel_guc *guc) @@ -441,17 +474,26 @@static int
guc_log_late_setup(struct intel_guc *guc)drivers/gpu/drm/i915/intel_guc_log.c +0
-2struct drm_i915_private *dev_priv =guc_to_i915 (guc);int ret; -lockdep_assert_held (&dev_priv->drm.struct_mutex); -if (!guc_log_has_runtime (guc)) {/* * If log was disabled at boot time, then setup needed to handle
* log buffer flush interrupts would not have
been done yet, sodrivers/gpu/drm/i915/intel_guc_log.c +16
-8* do that now. */ - ret =guc_log_runtime_create (guc); +GEM_BUG_ON (guc_log_has_relay (guc)); + + ret =intel_guc_log_relay_create (guc);if (ret) goto err; + +mutex_lock (&dev_priv->drm.struct_mutex); +intel_runtime_pm_get (dev_priv); + ret =guc_log_runtime_create (guc); +intel_runtime_pm_put (dev_priv); +mutex_unlock (&dev_priv->drm.struct_mutex); + +if (ret) + goto err_relay; } ret =guc_log_relay_file_create (guc);@@ -461,7 +503,11 @@ static int guc_log_late_setup(struct intel_guc *guc)return 0 ; err_runtime: +mutex_lock (&dev_priv->drm.struct_mutex);guc_log_runtime_destroy (guc); +mutex_unlock (&dev_priv->drm.struct_mutex); +err_relay: +intel_guc_log_relay_destroy (guc); err:/* logging will remain off */ i915_modparams.guc_log_level =0 ;@@ -509,17 +555,10 @@ int intel_guc_log_create(struct intel_guc *guc)struct i915_vma *vma;unsigned long offset;u32 flags; -u32 size;int ret;GEM_BUG_ON (guc->log.vma); -/* The first page is to save log buffer state. Allocate one -* extra page for others in case for overlap */ - size = (1 +GUC_LOG_DPC_PAGES +1 + -GUC_LOG_ISR_PAGES +1 + -GUC_LOG_CRASH_PAGES +1 ) <<PAGE_SHIFT ; -/* We require SSE 4.1 for fast reads from the GuC log buffer and
drivers/gpu/drm/i915/intel_guc_log.c +9
-1* it should be present on the chipsets supporting GuC based * submisssions. @@ -529,7 +568,7 @@ int intel_guc_log_create(struct intel_guc *guc) goto err; } - vma =intel_guc_allocate_vma (guc, size); + vma =intel_guc_allocate_vma (guc,GUC_LOG_SIZE );if (IS_ERR (vma)) { ret =PTR_ERR (vma); goto err;@@ -584,7 +623,15 @@ int i915_guc_log_control(struct drm_i915_private *dev_priv, u64 control_val)return 0 ; verbosity = enable_logging ? control_val -1 :0 ; + + ret =mutex_lock_interruptible (&dev_priv- >drm.struct_mutex); +if (ret) +return ret; +intel_runtime_pm_get (dev_priv); ret =guc_log_control (guc, enable_logging, verbosity); +intel_runtime_pm_put (dev_priv); +mutex_unlock (&dev_priv->drm.struct_mutex); +if (ret <0 ) {DRM_DEBUG_DRIVER("guc_logging_control
action failed %d\n", ret);drivers/gpu/drm/i915/intel_guc_log.c +0
-0return ret;@@ -605,7 +652,11 @@ int i915_guc_log_control(struct drm_i915_private *dev_priv, u64 control_val) }/* GuC logging is currently the only user of Guc2Host interrupts */
drivers/gpu/drm/i915/intel_guc_log.c +4
-0+mutex_lock (&dev_priv->drm.struct_mutex); +intel_runtime_pm_get (dev_priv);gen9_enable_guc_interrupts (dev_priv); +intel_runtime_pm_put (dev_priv); +mutex_unlock (&dev_priv->drm.struct_mutex); }else {/* * Once logging is disabled, GuC won't generate logs & send an @@ -613,7 +664,11 @@ int
i915_guc_log_control(struct drm_i915_private *dev_priv, u64 control_val)
* which is yet to be captured. So request GuC to
update the log
* buffer state and then collect the left over
logs.drivers/gpu/drm/i915/intel_guc_log.c +4
-0*/ +mutex_lock (&dev_priv->drm.struct_mutex); +intel_runtime_pm_get (dev_priv);guc_flush_logs (guc); +intel_runtime_pm_put (dev_priv); +mutex_unlock (&dev_priv->drm.struct_mutex);/* As logging is disabled, update log level to reflect that */
drivers/gpu/drm/i915/intel_guc_log.c +2
-2i915_modparams.guc_log_level =0 ;@@ -627,19 +682,21 @@ void i915_guc_log_register(struct drm_i915_private *dev_priv)if (!USES_GUC_SUBMISSION (dev_priv) || !i915_modparams.guc_log_level)return ; -mutex_lock (&dev_priv->drm.struct_mutex);guc_log_late_setup (&dev_priv->guc); -mutex_unlock (&dev_priv->drm.struct_mutex); }void i915_guc_log_unregister (struct drm_i915_private *dev_priv) { +struct intel_guc *guc = &dev_priv->guc; +if (!USES_GUC_SUBMISSION (dev_priv))return ;mutex_lock (&dev_priv->drm.struct_mutex);/* GuC logging is currently the only user of Guc2Host interrupts */
drivers/gpu/drm/i915/intel_guc_log.c +3
-1gen9_disable_guc_interrupts (dev_priv); -guc_log_runtime_destroy (&dev_priv->guc); +guc_log_runtime_destroy (guc);mutex_unlock (&dev_priv->drm.struct_mutex); + +intel_guc_log_relay_destroy (guc); }drivers/gpu/drm/i915/intel_guc_log.h +8
-0@@ -32,6 +32,13 @@struct drm_i915_private ;struct intel_guc ; +/* +* The first page is to save log buffer state. Allocate one +* extra page for others in case for overlap */ +#define GUC_LOG_SIZE ((1 + GUC_LOG_DPC_PAGES + 1 + GUC_LOG_ISR_PAGES + \ + 1 + GUC_LOG_CRASH_PAGES + 1) << PAGE_SHIFT) +struct intel_guc_log {u32 flags;struct i915_vma *vma;@@ -52,6 +59,8 @@ struct intel_guc_log {int intel_guc_log_create (struct intel_guc *guc);void intel_guc_log_destroy (struct intel_guc *guc); +int intel_guc_log_relay_create (struct intel_guc *guc);void +intel_guc_log_relay_destroy (struct intel_guc *guc); int i915_guc_log_control(struct drm_i915_private *dev_priv, u64
control_val); voidi915_guc_log_register(struct drm_i915_private *dev_priv); voidi915_guc_log_unregister(struct drm_i915_private *dev_priv); diff --gita/drivers/gpu/drm/i915/intel_uc.cb/drivers/gpu/drm/i915/intel_uc.cindex f78a17b..e3f3509 100644drivers/gpu/drm/i915/intel_uc.c +21
-5@@ -236,28 +236,44 @@ 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 intel_uc_init_misc (struct drm_i915_private *dev_priv) { +struct intel_guc *guc = &dev_priv->guc;int ret;if (!USES_GUC (dev_priv))return 0 ; - ret =intel_guc_init_wq (&dev_priv->guc); + ret =intel_guc_init_wq (guc);if (ret) {DRM_ERROR ("Couldn't allocate workqueues for GuC\n"); -return ret; + goto err; + } + + ret =intel_guc_log_relay_create (guc); +if (ret) { +DRM_ERROR ("Couldn't allocate relay for GuC log\n"); + goto err_relay; }return 0 ; + +err_relay: +intel_guc_fini_wq (guc); +err: +return ret; } -void intel_uc_fini_wq (struct drm_i915_private *dev_priv) +void intel_uc_fini_misc (struct drm_i915_private *dev_priv) { +struct intel_guc *guc = &dev_priv->guc; +if (!USES_GUC (dev_priv))return ; -intel_guc_fini_wq (&dev_priv->guc); +intel_guc_fini_wq (guc); + +intel_guc_log_relay_destroy (guc);}
int
intel_uc_init(struct drm_i915_private *dev_priv)diff --gita/drivers/gpu/drm/i915/intel_uc.hb/drivers/gpu/drm/i915/intel_uc.hindex 8a72497..f2984e0 100644drivers/gpu/drm/i915/intel_uc.h +0
-0@@ -33,8 +33,8 @@void intel_uc_init_mmio (struct drm_i915_private *dev_priv);void
intel_uc_init_fw(struct drm_i915_private *dev_priv); voidintel_uc_fini_fw(struct drm_i915_private *dev_priv); -intintel_uc_init_wq(struct drm_i915_private *dev_priv); -voidintel_uc_fini_wq(struct drm_i915_private *dev_priv);drivers/gpu/drm/i915/intel_uc.h +2
-0+int intel_uc_init_misc (struct drm_i915_private *dev_priv);void +intel_uc_fini_misc (struct drm_i915_private *dev_priv);int intel_uc_init_hw (struct drm_i915_private *dev_priv);void
Quoted 4 lines
intel_uc_fini_hw(struct drm_i915_private *dev_priv); intintel_uc_init(struct drm_i915_private *dev_priv);
--
1.9.1
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 1 line · Marta
I recommend that this is tested with a HACK to enable the GUC logs again, so that we can see if it really fixes the issue.
Yes. Patch 4 in the series enables the GuC log for testing.
Quoted 627 lines
Quoted 627 lines
-----Original Message-----
From: Kamble, Sagar A
Sent: Monday, January 22, 2018 10:26 AM
To: [email protected]
Cc: Kamble, Sagar A <[email protected]>; Wajdeczko, Michal <[email protected]>; Ceraolo Spurio, Daniele <[email protected]>; Ursulin, Tvrtko <[email protected]>; Chris Wilson <[email protected]>; Joonas Lahtinen <[email protected]>; Lofstedt, Marta <[email protected]>; Winiarski, Michal <[email protected]>
Subject: [PATCH 1/4] drm/i915/guc: Fix lockdep due to log relay channel handling understruct_mutexThis patch fixes lockdep issue due to circular locking dependency of
struct_mutex,i_mutex_key,mmap_sem,relay_channels_mutex.
For GuC log relay channel we create debugfs file that requiresi_mutex_keylock and we are doing that understruct_mutex. So we introduced newer dependency as:&dev->struct_mutex --> &sb->s_type->i_mutex_key#3 --> &mm-Quoted 1 line
mmap_semHowever, there is dependency frommmap_semto
struct_mutex. Hence we separate the relay create/destroy operation from understruct_mutex.======================================================
WARNING: possible circular locking dependency detected 4.15.0-rc6-CI- Patchwork_7614+ #1 Not tainted
------------------------------------------------------
debugfs_test/1388 is trying to acquire lock:
(&dev->struct_mutex){+.+.}, at: [<00000000d5e1d915>]i915_mutex_lock_interruptible+0x47/0x130[i915]but task is already holding lock:
(&mm->mmap_sem){++++}, at: [<0000000029a9c131>]__do_page_fault+0x106/0x560which lock already depends on the new lock.
the existing dependency chain (in reverse order) is:
-> #3 (&mm->mmap_sem){++++}:
_copy_to_user+0x1e/0x70 filldir+0x8c/0xf0 dcache_readdir+0xeb/0x160 iterate_dir+0xdc/0x140 SyS_getdents+0xa0/0x130 entry_SYSCALL_64_fastpath+0x1c/0x89-> #2 (&sb->s_type->i_mutex_key#3){++++}:
start_creating+0x59/0x110 __debugfs_create_file+0x2e/0xe0 relay_create_buf_file+0x62/0x80 relay_late_setup_files+0x84/0x250 guc_log_late_setup+0x4f/0x110 [i915] i915_guc_log_register+0x32/0x40 [i915] i915_driver_load+0x7b6/0x1720 [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-> #1 (relay_channels_mutex){+.+.}:
relay_open+0x12c/0x2b0 intel_guc_log_runtime_create+0xab/0x230 [i915] intel_guc_init+0x81/0x120 [i915] intel_uc_init+0x29/0xa0 [i915] i915_gem_init+0x182/0x530 [i915] i915_driver_load+0xaa9/0x1720 [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){+.+.}:
__mutex_lock+0x81/0x9b0 i915_mutex_lock_interruptible+0x47/0x130 [i915] i915_gem_fault+0x201/0x790 [i915] __do_fault+0x15/0x70 __handle_mm_fault+0x677/0xdc0 handle_mm_fault+0x14f/0x2f0 __do_page_fault+0x2d1/0x560 page_fault+0x4c/0x60other info that might help us debug this:
Chain exists of: &dev->struct_mutex --> &sb->s_type->i_mutex_key#3 --> &mm-Quoted 1 line
mmap_semPossible unsafe locking scenario:CPU0 CPU1 ---- ---- lock(&mm->mmap_sem); lock(&sb->s_type->i_mutex_key#3); lock(&mm->mmap_sem);
lock(&dev->struct_mutex);*** DEADLOCK ***
1 lock held by debugfs_test/1388:
#0: (&mm->mmap_sem){++++}, at: [<0000000029a9c131>]__do_page_fault+0x106/0x560 stack backtrace: CPU: 2 PID: 1388 Comm: debugfs_test Not tainted 4.15.0-rc6-CI- Patchwork_7614+ #1 Hardware name: To Be Filled By O.E.M. To Be Filled By O.E.M./J4205-ITX, BIOS P1.10 09/29/2016 Call Trace: dump_stack+0x5f/0x86 print_circular_bug.isra.18+0x1d0/0x2c0 __lock_acquire+0x14ae/0x1b60 ? lock_acquire+0xaf/0x200 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+0x201/0x790 [i915] __do_fault+0x15/0x70 ? _raw_spin_unlock+0x29/0x40 __handle_mm_fault+0x677/0xdc0 handle_mm_fault+0x14f/0x2f0 __do_page_fault+0x2d1/0x560 ? page_fault+0x36/0x60 page_fault+0x4c/0x60Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=104693
Testcase: igt/debugfs_test/read_all_entries # with enable_guc=1 and guc_log_level=1Signed-off-by Sagar Arun Kamble <[email protected]>Cc Michal Wajdeczko <[email protected]>Cc Daniele Ceraolo Spurio <[email protected]>Cc Tvrtko Ursulin <[email protected]>Cc Chris Wilson <[email protected]>Cc Joonas Lahtinen <[email protected]>Cc Marta Lofstedt <[email protected]>Cc Michal Winiarski <[email protected]>--- drivers/gpu/drm/i915/i915_debugfs.c | 12 +--- drivers/gpu/drm/i915/i915_drv.c | 2 +- drivers/gpu/drm/i915/i915_gem.c | 4 +- drivers/gpu/drm/i915/intel_guc.c | 6 +- drivers/gpu/drm/i915/intel_guc_log.c | 111 ++++++++++++++++++++++++++--------- drivers/gpu/drm/i915/intel_guc_log.h | 9 +++ drivers/gpu/drm/i915/intel_uc.c | 26 ++++++-- drivers/gpu/drm/i915/intel_uc.h | 4 +- 8 files changed, 125 insertions(+), 49 deletions(-)drivers/gpu/drm/i915/i915_debugfs.c +1
-11@@ -2467,7 +2467,6 @@ static int i915_guc_log_control_get(void *data, u64 *val) static int i915_guc_log_control_set(void *data, u64 val) {struct drm_i915_private *dev_priv = data; -int ret;if (!HAS_GUC (dev_priv))return -ENODEV ;@@ -2475,16 +2474,7 @@ static int i915_guc_log_control_set(void *data, u64 val)if (!dev_priv->guc.log.vma)return -EINVAL ; - ret =mutex_lock_interruptible (&dev_priv- > drm.struct_mutex); -if (ret) -return ret; - -intel_runtime_pm_get (dev_priv); - ret =i915_guc_log_control (dev_priv, val); -intel_runtime_pm_put (dev_priv); - -mutex_unlock (&dev_priv->drm.struct_mutex); -return ret; +return i915_guc_log_control (dev_priv, val); }DEFINE_SIMPLE_ATTRIBUTE(i915_guc_log_control_fops,
(quoted diff) +0
-0
b/drivers/gpu/drm/i915/i915_drv.cindex 1fbe378..29048b9 100644drivers/gpu/drm/i915/i915_drv.c +1
-1@@ -626,7 +626,7 @@ 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); +intel_uc_fini_misc (dev_priv);i915_gem_cleanup_userptr (dev_priv);
i915_gem_drain_freed_objects(dev_priv);(quoted diff) +0
-0
b/drivers/gpu/drm/i915/i915_gem.cindex 7f0684c..34d0d51 100644drivers/gpu/drm/i915/i915_gem.c +2
-2@@ -5193,7 +5193,7 @@ int i915_gem_init(struct drm_i915_private *dev_priv)if (ret)return ret; - ret =intel_uc_init_wq (dev_priv); + ret =intel_uc_init_misc (dev_priv);if (ret)return ret;@@ -5289,7 +5289,7 @@ int i915_gem_init(struct drm_i915_private *dev_priv)intel_uncore_forcewake_put (dev_priv,FORCEWAKE_ALL );mutex_unlock (&dev_priv->drm.struct_mutex); -intel_uc_fini_wq (dev_priv); +intel_uc_fini_misc (dev_priv);if (ret != -EIO )i915_gem_cleanup_userptr(dev_priv);drivers/gpu/drm/i915/intel_guc.c +3
-1@@ -87,8 +87,10 @@ int intel_guc_init_wq(struct intel_guc *guc)*/ guc->log.runtime.flush_wq =alloc_ordered_workqueue ("i915-guc_log" ,WQ_HIGHPRI |WQ_FREEZABLE ); -if (!guc->log.runtime.flush_wq) +if (!guc->log.runtime.flush_wq) { + DRM_ERROR("Couldn't allocate workqueue for GuC log\n"); return -ENOMEM; + }/*
* Even though both sending GuC action, and adding a new
workitem to @@ -109,6 +111,8 @@ intintel_guc_init_wq(struct intel_guc *guc)WQ_HIGHPRI);if (!guc->preempt_wq) {
destroy_workqueue(guc-Quoted 1 line
log.runtime.flush_wq);
drivers/gpu/drm/i915/intel_guc.c +1
-0+ DRM_ERROR("Couldn't allocateworkqueue for GuC "
drivers/gpu/drm/i915/intel_guc.c +1
-0+ "preemption\n");return -ENOMEM ; } }drivers/gpu/drm/i915/intel_guc_log.c +5
-2@@ -359,16 +359,21 @@ static bool guc_log_has_runtime(struct intel_guc *guc)return guc->log.runtime.buf_addr !=NULL ; } +static bool guc_log_has_relay (struct intel_guc *guc) { +return guc->log.runtime.relay_chan !=NULL ; } +static int guc_log_runtime_create (struct intel_guc *guc) {struct drm_i915_private *dev_priv =guc_to_i915 (guc);void *vaddr; -struct rchan *guc_log_relay_chan; -size_t n_subbufs, subbuf_size;int ret;lockdep_assert_held (&dev_priv->drm.struct_mutex); +GEM_BUG_ON (!guc_log_has_relay (guc)); +GEM_BUG_ON (guc_log_has_runtime (guc)); ret = i915_gem_object_set_to_wc_domain(guc->log.vma-Quoted 1 line
obj, true); @@ -387,8 +392,38 @@ static int guc_log_runtime_create(struct
intel_guc*guc)guc->log.runtime.buf_addr = vaddr;drivers/gpu/drm/i915/intel_guc_log.c +1
-0+ INIT_WORK(&guc->log.runtime.flush_work,
capture_logs_work);drivers/gpu/drm/i915/intel_guc_log.c +7
-0+ +return 0 ; +} + +static void guc_log_runtime_destroy (struct intel_guc *guc) { +/* +* It's possible that the runtime stuff was never allocated because
drivers/gpu/drm/i915/intel_guc_log.c +21
-1+* GuC log was disabled at the boot time. +**/ +if (!guc_log_has_runtime (guc)) +return ; + +i915_gem_object_unpin_map (guc->log.vma->obj); + guc->log.runtime.buf_addr =NULL ; +} + +int intel_guc_log_relay_create (struct intel_guc *guc) { +struct drm_i915_private *dev_priv =guc_to_i915 (guc); +struct rchan *guc_log_relay_chan; +size_t n_subbufs, subbuf_size; +int ret; + +if (!i915_modparams.guc_log_level) +return 0 ; + +GEM_BUG_ON (guc_log_has_relay (guc)); +/* Keep the size of sub buffers same as shared log buffer */ - subbuf_size = guc->log.vma->obj->base.size; + subbuf_size =GUC_LOG_SIZE ;/* Store up to 8 snapshots, which is large enough to buffer sufficient
* boot time logs and provides enough leeway to User, in
terms of @@ -407,33 +442,31 @@ static intguc_log_runtime_create(struct intel_guc *guc)
DRM_ERROR("Couldn't create relay chan for
GuC logging\n");ret = -ENOMEM;drivers/gpu/drm/i915/intel_guc_log.c +1
-1- goto err_vaddr; + goto err; } GEM_BUG_ON(guc_log_relay_chan->subbuf_size <
subbuf_size);
guc->log.runtime.relay_chan=guc_log_relay_chan;drivers/gpu/drm/i915/intel_guc_log.c +0
-1- INIT_WORK(&guc->log.runtime.flush_work,
capture_logs_work);
return 0;drivers/gpu/drm/i915/intel_guc_log.c +4
-5-err_vaddr: -i915_gem_object_unpin_map (guc->log.vma->obj); - guc->log.runtime.buf_addr =NULL ; +err: +/* logging will be off */ + i915_modparams.guc_log_level =0 ;return ret; } -static void guc_log_runtime_destroy (struct intel_guc *guc) +void intel_guc_log_relay_destroy (struct intel_guc *guc) {/* -* It's possible that the runtime stuff was never allocated because
drivers/gpu/drm/i915/intel_guc_log.c +3
-3+* It's possible that the relay was never allocated because * GuC log was disabled at the boot time. */ -if (!guc_log_has_runtime (guc)) +if (!guc_log_has_relay (guc))return ;relay_close (guc->log.runtime.relay_chan); -i915_gem_object_unpin_map (guc->log.vma->obj); - guc->log.runtime.buf_addr =NULL ; + guc->log.runtime.relay_chan =NULL ; } static int guc_log_late_setup(struct intel_guc *guc) @@ -441,17 +474,26 @@static int
guc_log_late_setup(struct intel_guc *guc)
structdrm_i915_private*dev_priv=guc_to_i915(guc); int ret;drivers/gpu/drm/i915/intel_guc_log.c +0
-2-lockdep_assert_held (&dev_priv->drm.struct_mutex); -if (!guc_log_has_runtime (guc)) {/* * If log was disabled at boot time, then setup needed to handle
* log buffer flush interrupts would not have
been done yet, so
* do that now.
*/drivers/gpu/drm/i915/intel_guc_log.c +16
-8- ret =guc_log_runtime_create (guc); +GEM_BUG_ON (guc_log_has_relay (guc)); + + ret =intel_guc_log_relay_create (guc);if (ret) goto err; + +mutex_lock (&dev_priv->drm.struct_mutex); +intel_runtime_pm_get (dev_priv); + ret =guc_log_runtime_create (guc); +intel_runtime_pm_put (dev_priv); +mutex_unlock (&dev_priv->drm.struct_mutex); + +if (ret) + goto err_relay; } ret =guc_log_relay_file_create (guc);@@ -461,7 +503,11 @@ static int guc_log_late_setup(struct intel_guc *guc)return 0 ; err_runtime: +mutex_lock (&dev_priv->drm.struct_mutex);guc_log_runtime_destroy (guc); +mutex_unlock (&dev_priv->drm.struct_mutex); +err_relay: +intel_guc_log_relay_destroy (guc); err:/* logging will remain off */ i915_modparams.guc_log_level =0 ;@@ -509,17 +555,10 @@ int intel_guc_log_create(struct intel_guc *guc)struct i915_vma *vma;unsigned long offset;u32 flags; -u32 size;int ret;GEM_BUG_ON (guc->log.vma); -/* The first page is to save log buffer state. Allocate one -* extra page for others in case for overlap */ - size = (1 +GUC_LOG_DPC_PAGES +1 + -GUC_LOG_ISR_PAGES +1 + -GUC_LOG_CRASH_PAGES +1 ) <<PAGE_SHIFT ; -/* We require SSE 4.1 for fast reads from the GuC log buffer and
* it should be present on the chipsets supporting GuC based
* submisssions.drivers/gpu/drm/i915/intel_guc_log.c +9
-1@@ -529,7 +568,7 @@ int intel_guc_log_create(struct intel_guc *guc) goto err; } - vma =intel_guc_allocate_vma (guc, size); + vma =intel_guc_allocate_vma (guc,GUC_LOG_SIZE );if (IS_ERR (vma)) { ret =PTR_ERR (vma); goto err;@@ -584,7 +623,15 @@ int i915_guc_log_control(struct drm_i915_private *dev_priv, u64 control_val)return 0 ; verbosity = enable_logging ? control_val -1 :0 ; + + ret =mutex_lock_interruptible (&dev_priv- > drm.struct_mutex); +if (ret) +return ret; +intel_runtime_pm_get (dev_priv); ret =guc_log_control (guc, enable_logging, verbosity); +intel_runtime_pm_put (dev_priv); +mutex_unlock (&dev_priv->drm.struct_mutex); +if (ret <0 ) {DRM_DEBUG_DRIVER("guc_logging_control
action failed %d\n", ret);
return ret;drivers/gpu/drm/i915/intel_guc_log.c +0
-0@@ -605,7 +652,11 @@ int i915_guc_log_control(struct drm_i915_private *dev_priv, u64 control_val) }/* GuC logging is currently the only user of Guc2Host interrupts */
drivers/gpu/drm/i915/intel_guc_log.c +4
-0+mutex_lock (&dev_priv->drm.struct_mutex); +intel_runtime_pm_get (dev_priv);gen9_enable_guc_interrupts (dev_priv); +intel_runtime_pm_put (dev_priv); +mutex_unlock (&dev_priv->drm.struct_mutex); }else {/* * Once logging is disabled, GuC won't generate logs & send an @@ -613,7 +664,11 @@ int
i915_guc_log_control(struct drm_i915_private *dev_priv, u64 control_val)
* which is yet to be captured. So request GuC to
update the log
* buffer state and then collect the left over
logs.
*/drivers/gpu/drm/i915/intel_guc_log.c +4
-0+mutex_lock (&dev_priv->drm.struct_mutex); +intel_runtime_pm_get (dev_priv);guc_flush_logs (guc); +intel_runtime_pm_put (dev_priv); +mutex_unlock (&dev_priv->drm.struct_mutex);/* As logging is disabled, update log level to reflect that */
i915_modparams.guc_log_level = 0;drivers/gpu/drm/i915/intel_guc_log.c +2
-2@@ -627,19 +682,21 @@ void i915_guc_log_register(struct drm_i915_private *dev_priv)if (!USES_GUC_SUBMISSION (dev_priv) || !i915_modparams.guc_log_level)return ; -mutex_lock (&dev_priv->drm.struct_mutex);guc_log_late_setup (&dev_priv->guc); -mutex_unlock (&dev_priv->drm.struct_mutex); }void i915_guc_log_unregister (struct drm_i915_private *dev_priv) { +struct intel_guc *guc = &dev_priv->guc; +if (!USES_GUC_SUBMISSION (dev_priv))return ;mutex_lock (&dev_priv->drm.struct_mutex);/* GuC logging is currently the only user of Guc2Host interrupts */
gen9_disable_guc_interrupts(dev_priv);drivers/gpu/drm/i915/intel_guc_log.c +3
-1-guc_log_runtime_destroy (&dev_priv->guc); +guc_log_runtime_destroy (guc);mutex_unlock (&dev_priv->drm.struct_mutex); + +intel_guc_log_relay_destroy (guc); }drivers/gpu/drm/i915/intel_guc_log.h +8
-0@@ -32,6 +32,13 @@struct drm_i915_private ;struct intel_guc ; +/* +* The first page is to save log buffer state. Allocate one +* extra page for others in case for overlap */ +#define GUC_LOG_SIZE ((1 + GUC_LOG_DPC_PAGES + 1 + GUC_LOG_ISR_PAGES + \ + 1 + GUC_LOG_CRASH_PAGES + 1) << PAGE_SHIFT) +struct intel_guc_log {u32 flags;struct i915_vma *vma;@@ -52,6 +59,8 @@ struct intel_guc_log {int intel_guc_log_create (struct intel_guc *guc);void intel_guc_log_destroy (struct intel_guc *guc); +int intel_guc_log_relay_create (struct intel_guc *guc);void +intel_guc_log_relay_destroy (struct intel_guc *guc); int i915_guc_log_control(struct drm_i915_private *dev_priv, u64
control_val); voidi915_guc_log_register(struct drm_i915_private *dev_priv); voidi915_guc_log_unregister(struct drm_i915_private *dev_priv); diff --gita/drivers/gpu/drm/i915/intel_uc.cb/drivers/gpu/drm/i915/intel_uc.cindex f78a17b..e3f3509 100644drivers/gpu/drm/i915/intel_uc.c +21
-5@@ -236,28 +236,44 @@ 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 intel_uc_init_misc (struct drm_i915_private *dev_priv) { +struct intel_guc *guc = &dev_priv->guc;int ret;if (!USES_GUC (dev_priv))return 0 ; - ret =intel_guc_init_wq (&dev_priv->guc); + ret =intel_guc_init_wq (guc);if (ret) {DRM_ERROR ("Couldn't allocate workqueues for GuC\n"); -return ret; + goto err; + } + + ret =intel_guc_log_relay_create (guc); +if (ret) { +DRM_ERROR ("Couldn't allocate relay for GuC log\n"); + goto err_relay; }return 0 ; + +err_relay: +intel_guc_fini_wq (guc); +err: +return ret; } -void intel_uc_fini_wq (struct drm_i915_private *dev_priv) +void intel_uc_fini_misc (struct drm_i915_private *dev_priv) { +struct intel_guc *guc = &dev_priv->guc; +if (!USES_GUC (dev_priv))return ; -intel_guc_fini_wq (&dev_priv->guc); +intel_guc_fini_wq (guc); + +intel_guc_log_relay_destroy (guc);}int
intel_uc_init(struct drm_i915_private *dev_priv)diff --gita/drivers/gpu/drm/i915/intel_uc.hb/drivers/gpu/drm/i915/intel_uc.hindex 8a72497..f2984e0 100644drivers/gpu/drm/i915/intel_uc.h +0
-0@@ -33,8 +33,8 @@void intel_uc_init_mmio (struct drm_i915_private *dev_priv);void
intel_uc_init_fw(struct drm_i915_private *dev_priv); voidintel_uc_fini_fw(struct drm_i915_private *dev_priv); -intintel_uc_init_wq(struct drm_i915_private *dev_priv); -void
Quoted 8 lines
Quoted 8 lines
intel_uc_fini_wq(struct drm_i915_private *dev_priv);drivers/gpu/drm/i915/intel_uc.h +2
-0+int intel_uc_init_misc (struct drm_i915_private *dev_priv);void +intel_uc_fini_misc (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); intintel_uc_init(struct drm_i915_private *dev_priv);
--
1.9.1
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Patchwork
CI✓ Fi.CI.BAT: success for series starting with [1/4] drm/i915/guc: Fix lockdep due to log relay channel handling under struct_mutex
In reply to Sagar Arun Kamble== Series Details ==
== Series Details ==
Series: series starting with [1/4] drm/i915/guc: Fix lockdep due to log relay channel handling under struct_mutex
URL : https://patchwork.freedesktop.org/series/36875/
State : success
== Summary ==
Series 36875v1 series starting with [1/4] drm/i915/guc: Fix lockdep due to log relay channel handling under struct_mutex
https://patchwork.freedesktop.org/api/1.0/series/36875/revisions/1/mbox/
Test gem_ringfill:
Subgroup basic-default:
skip -> PASS (fi-bsw-n3050)
Test kms_pipe_crc_basic:
Subgroup suspend-read-crc-pipe-b:
incomplete -> PASS (fi-snb-2520m) fdo#103713fdo#103713 https://bugs.freedesktop.org/show_bug.cgi?id=103713
fi-bdw-5557u total:288 pass:267 dwarn:0 dfail:0 fail:0 skip:21 time:435s fi-bdw-gvtdvm total:288 pass:264 dwarn:0 dfail:0 fail:0 skip:24 time:437s fi-blb-e6850 total:288 pass:223 dwarn:1 dfail:0 fail:0 skip:64 time:385s fi-bsw-n3050 total:288 pass:242 dwarn:0 dfail:0 fail:0 skip:46 time:523s fi-bwr-2160 total:288 pass:183 dwarn:0 dfail:0 fail:0 skip:105 time:295s fi-bxt-dsi total:288 pass:258 dwarn:0 dfail:0 fail:0 skip:30 time:500s fi-bxt-j4205 total:288 pass:259 dwarn:0 dfail:0 fail:0 skip:29 time:502s fi-byt-j1900 total:288 pass:253 dwarn:0 dfail:0 fail:0 skip:35 time:496s fi-byt-n2820 total:288 pass:249 dwarn:0 dfail:0 fail:0 skip:39 time:479s fi-elk-e7500 total:224 pass:168 dwarn:10 dfail:0 fail:0 skip:45 fi-gdg-551 total:288 pass:179 dwarn:0 dfail:0 fail:1 skip:108 time:303s fi-glk-1 total:288 pass:260 dwarn:0 dfail:0 fail:0 skip:28 time:525s fi-hsw-4770 total:288 pass:261 dwarn:0 dfail:0 fail:0 skip:27 time:398s fi-hsw-4770r total:288 pass:261 dwarn:0 dfail:0 fail:0 skip:27 time:407s fi-ilk-650 total:288 pass:228 dwarn:0 dfail:0 fail:0 skip:60 time:426s fi-ivb-3520m total:288 pass:259 dwarn:0 dfail:0 fail:0 skip:29 time:473s fi-ivb-3770 total:288 pass:255 dwarn:0 dfail:0 fail:0 skip:33 time:421s fi-kbl-7500u total:288 pass:263 dwarn:1 dfail:0 fail:0 skip:24 time:465s fi-kbl-7560u total:288 pass:269 dwarn:0 dfail:0 fail:0 skip:19 time:504s fi-kbl-7567u total:288 pass:268 dwarn:0 dfail:0 fail:0 skip:20 time:461s fi-kbl-r total:288 pass:261 dwarn:0 dfail:0 fail:0 skip:27 time:511s fi-pnv-d510 total:288 pass:222 dwarn:1 dfail:0 fail:0 skip:65 time:631s fi-skl-6260u total:288 pass:268 dwarn:0 dfail:0 fail:0 skip:20 time:443s fi-skl-6600u total:288 pass:261 dwarn:0 dfail:0 fail:0 skip:27 time:521s fi-skl-6700hq total:288 pass:262 dwarn:0 dfail:0 fail:0 skip:26 time:538s fi-skl-6700k2 total:288 pass:264 dwarn:0 dfail:0 fail:0 skip:24 time:500s fi-skl-6770hq total:288 pass:268 dwarn:0 dfail:0 fail:0 skip:20 time:512s fi-skl-gvtdvm total:288 pass:265 dwarn:0 dfail:0 fail:0 skip:23 time:446s fi-snb-2520m total:288 pass:248 dwarn:0 dfail:0 fail:0 skip:40 time:549s fi-snb-2600 total:288 pass:248 dwarn:0 dfail:0 fail:0 skip:40 time:406s Blacklisted hosts: fi-cfl-s2 total:288 pass:256 dwarn:0 dfail:0 fail:3 skip:26 time:565s fi-glk-dsi total:288 pass:258 dwarn:0 dfail:0 fail:0 skip:30 time:483s fi-skl-guc total:288 pass:260 dwarn:0 dfail:0 fail:0 skip:28 time:423s
a1f34f9415a41d5a4d219bcbbe5cf39a0b07cc22 drm-tip: 2018y-01m-20d-19h-27m-32s UTC integration manifest d79027e481e0 Revert "drm/i915/guc: Keep GuC log disabled by default"cce3e79ffe73 drm/i915/guc: Enable interrupts before resuming GuC during runtime resume 1ae00207a925 drm/i915/guc: Grab RPM wakelock while disabling GuC interrupts 3ff2d9df1077 drm/i915/guc: Fix lockdep due to log relay channel handling under struct_mutex
== Logs ==
For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_7735/issues.html
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoting Sagar Arun Kamble (2018-01-22 08:26:01)
Quoted 34 lines
drivers/gpu/drm/i915/intel_guc_log.c +14
-5+int intel_guc_log_relay_create (struct intel_guc *guc) +{ +struct drm_i915_private *dev_priv =guc_to_i915 (guc); +struct rchan *guc_log_relay_chan; +size_t n_subbufs, subbuf_size; +int ret; + +if (!i915_modparams.guc_log_level) +return 0 ; + +GEM_BUG_ON (guc_log_has_relay (guc)); +/* Keep the size of sub buffers same as shared log buffer */ - subbuf_size = guc->log.vma->obj->base.size; + subbuf_size =GUC_LOG_SIZE ;/* Store up to 8 snapshots, which is large enough to buffer sufficient * boot time logs and provides enough leeway to User, in terms of @@ -407,33 +442,31 @@ static int guc_log_runtime_create(struct intel_guc *guc)DRM_ERROR ("Couldn't create relay chan for GuC logging\n" ); ret = -ENOMEM ; - goto err_vaddr; + goto err; }GEM_BUG_ON (guc_log_relay_chan->subbuf_size < subbuf_size); guc->log.runtime.relay_chan = guc_log_relay_chan; -INIT_WORK (&guc->log.runtime.flush_work, capture_logs_work);return 0 ; -err_vaddr: -i915_gem_object_unpin_map (guc->log.vma->obj);
drivers/gpu/drm/i915/intel_guc_log.c static int guc_log_runtime_create(struct intel_guc *guc) · line 460
Quoted 10 lines
drivers/gpu/drm/i915/intel_guc_log.c +4
-2- guc->log.runtime.buf_addr =NULL ; +err: +/* logging will be off */ + i915_modparams.guc_log_level =0 ;return ret; } -static void guc_log_runtime_destroy (struct intel_guc *guc) +void intel_guc_log_relay_destroy (struct intel_guc *guc) {
Now exposed to multiple users, we need to document what the locking requirements are here. Or add some local locking. Looks like at the moment, _create is using struct_mutex, which as we are only handling the relay may be overkill (but since this is one-off init, fair enough). -Chris
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 42 lines · Chris Wilson
Quoting Sagar Arun Kamble (2018-01-22 08:26:01)
Quoted 41 lines
drivers/gpu/drm/i915/intel_guc_log.c +17
-6+int intel_guc_log_relay_create (struct intel_guc *guc) +{ +struct drm_i915_private *dev_priv =guc_to_i915 (guc); +struct rchan *guc_log_relay_chan; +size_t n_subbufs, subbuf_size; +int ret; + +if (!i915_modparams.guc_log_level) +return 0 ; + +GEM_BUG_ON (guc_log_has_relay (guc)); +/* Keep the size of sub buffers same as shared log buffer */ - subbuf_size = guc->log.vma->obj->base.size; + subbuf_size =GUC_LOG_SIZE ;/* Store up to 8 snapshots, which is large enough to buffer sufficient * boot time logs and provides enough leeway to User, in terms of @@ -407,33 +442,31 @@ static int guc_log_runtime_create(struct intel_guc *guc)DRM_ERROR ("Couldn't create relay chan for GuC logging\n" ); ret = -ENOMEM ; - goto err_vaddr; + goto err; }GEM_BUG_ON (guc_log_relay_chan->subbuf_size < subbuf_size); guc->log.runtime.relay_chan = guc_log_relay_chan; -INIT_WORK (&guc->log.runtime.flush_work, capture_logs_work);return 0 ; -err_vaddr: -i915_gem_object_unpin_map (guc->log.vma->obj); - guc->log.runtime.buf_addr =NULL ; +err: +/* logging will be off */ + i915_modparams.guc_log_level =0 ;return ret; }
Quoted 5 lines
drivers/gpu/drm/i915/intel_guc_log.c static int guc_log_runtime_create(struct intel_guc *guc) · line 460
Quoted 3 lines
drivers/gpu/drm/i915/intel_guc_log.c +1
-1-static void guc_log_runtime_destroy (struct intel_guc *guc) +void intel_guc_log_relay_destroy (struct intel_guc *guc) {Now exposed to multiple users, we need to document what the locking requirements are here. Or add some local locking.
Do you mean locking between relay_create and relay_destroy?
Quoted 2 lines
Looks like at the
moment,_createis usingstruct_mutex,
relay_create and relay_destroy are now to be done outside of struct_mutex. I will add this documentation to the functions.
Quoted 3 lines
which as we are only handling the
relay may be overkill (but since this is one-off init, fair enough). -Chris
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoting Sagar Arun Kamble (2018-01-22 10:38:10)
Quoted 45 lines · Chris Wilson
Quoted 42 lines
Quoting Sagar Arun Kamble (2018-01-22 08:26:01)
Quoted 41 lines
drivers/gpu/drm/i915/intel_guc_log.c +17
-6+int intel_guc_log_relay_create (struct intel_guc *guc) +{ +struct drm_i915_private *dev_priv =guc_to_i915 (guc); +struct rchan *guc_log_relay_chan; +size_t n_subbufs, subbuf_size; +int ret; + +if (!i915_modparams.guc_log_level) +return 0 ; + +GEM_BUG_ON (guc_log_has_relay (guc)); +/* Keep the size of sub buffers same as shared log buffer */ - subbuf_size = guc->log.vma->obj->base.size; + subbuf_size =GUC_LOG_SIZE ;/* Store up to 8 snapshots, which is large enough to buffer sufficient * boot time logs and provides enough leeway to User, in terms of @@ -407,33 +442,31 @@ static int guc_log_runtime_create(struct intel_guc *guc)DRM_ERROR ("Couldn't create relay chan for GuC logging\n" ); ret = -ENOMEM ; - goto err_vaddr; + goto err; }GEM_BUG_ON (guc_log_relay_chan->subbuf_size < subbuf_size); guc->log.runtime.relay_chan = guc_log_relay_chan; -INIT_WORK (&guc->log.runtime.flush_work, capture_logs_work);return 0 ; -err_vaddr: -i915_gem_object_unpin_map (guc->log.vma->obj); - guc->log.runtime.buf_addr =NULL ; +err: +/* logging will be off */ + i915_modparams.guc_log_level =0 ;return ret; }
Quoted 6 lines
Quoted 5 lines
drivers/gpu/drm/i915/intel_guc_log.c static int guc_log_runtime_create(struct intel_guc *guc) · line 460
Quoted 3 lines
drivers/gpu/drm/i915/intel_guc_log.c +1
-1-static void guc_log_runtime_destroy (struct intel_guc *guc) +void intel_guc_log_relay_destroy (struct intel_guc *guc) {Now exposed to multiple users, we need to document what the locking requirements are here. Or add some local locking.
Do you mean locking between
relay_createandrelay_destroy?
We need a lock around guc->log.runtime.relay_chan as the destroy path is not ostensibly serialised between multiple potential callers. Ordinarily I would have said that serialisation for create/destroy/access of relay_chan was guaranteed by init/fini ordering, but that's no longer clear (based on a 5min read of the patch).
The most important question is "can relay_destroy be called while some user still has access to the relay_chan?"
Quoted 4 lines
Quoted 2 lines
Looks like at the
moment,_createis usingstruct_mutex,
relay_createandrelay_destroyare now to be done outside ofstruct_mutex. I will add this documentation to the functions.
(lockdep_assert_held :)
-Chris
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 50 lines · Chris Wilson
Quoting Sagar Arun Kamble (2018-01-22 10:38:10)
Quoted 49 lines · Chris Wilson
Quoted 47 lines
Quoting Sagar Arun Kamble (2018-01-22 08:26:01)
Quoted 34 lines
drivers/gpu/drm/i915/intel_guc_log.c +14
-5+int intel_guc_log_relay_create (struct intel_guc *guc) +{ +struct drm_i915_private *dev_priv =guc_to_i915 (guc); +struct rchan *guc_log_relay_chan; +size_t n_subbufs, subbuf_size; +int ret; + +if (!i915_modparams.guc_log_level) +return 0 ; + +GEM_BUG_ON (guc_log_has_relay (guc)); +/* Keep the size of sub buffers same as shared log buffer */ - subbuf_size = guc->log.vma->obj->base.size; + subbuf_size =GUC_LOG_SIZE ;/* Store up to 8 snapshots, which is large enough to buffer sufficient * boot time logs and provides enough leeway to User, in terms of @@ -407,33 +442,31 @@ static int guc_log_runtime_create(struct intel_guc *guc)DRM_ERROR ("Couldn't create relay chan for GuC logging\n" ); ret = -ENOMEM ; - goto err_vaddr; + goto err; }GEM_BUG_ON (guc_log_relay_chan->subbuf_size < subbuf_size); guc->log.runtime.relay_chan = guc_log_relay_chan; -INIT_WORK (&guc->log.runtime.flush_work, capture_logs_work);return 0 ; -err_vaddr: -i915_gem_object_unpin_map (guc->log.vma->obj);drivers/gpu/drm/i915/intel_guc_log.c static int guc_log_runtime_create(struct intel_guc *guc) · line 460
Quoted 10 lines
drivers/gpu/drm/i915/intel_guc_log.c +4
-2- guc->log.runtime.buf_addr =NULL ; +err: +/* logging will be off */ + i915_modparams.guc_log_level =0 ;return ret; } -static void guc_log_runtime_destroy (struct intel_guc *guc) +void intel_guc_log_relay_destroy (struct intel_guc *guc) {Now exposed to multiple users, we need to document what the locking requirements are here. Or add some local locking.
Quoted 5 lines
Quoted 1 line
Do you mean locking between
relay_createandrelay_destroy?We need a lock around
guc->log.runtime.relay_chanas the destroy path is not ostensibly serialised between multiple potential callers. Ordinarily I would have said that serialisation for create/destroy/access ofrelay_chanwas guaranteed by init/fini ordering,
I was also initially thinking that init/fini ordering should take care of synchronization. Checked further and I see that relay_open and relay_destroy are synchronized by relay_channels_mutex and internally if needed through debugfs inode synchronization. So I feel we need not add new lock.
Quoted 12 lines
but that's no longer
clear (based on a 5min read of the patch).The most important question is "can
relay_destroybe called while some user still has access to therelay_chan?"Quoted 4 lines
Quoted 2 lines
Looks like at the
moment,_createis usingstruct_mutex,
relay_createandrelay_destroyare now to be done outside ofstruct_mutex. I will add this documentation to the functions.(
lockdep_assert_held:)
-Chris
Signature · 7 lines
Thanks, Sagar _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 48 lines · Sagar Arun Kamble
Quoted 45 lines · Chris Wilson
Quoting Sagar Arun Kamble (2018-01-22 10:38:10)
Quoted 44 lines · Chris Wilson
Quoted 42 lines
Quoting Sagar Arun Kamble (2018-01-22 08:26:01)
Quoted 31 lines
drivers/gpu/drm/i915/intel_guc_log.c +12
-0+int intel_guc_log_relay_create (struct intel_guc *guc) +{ + struct drm_i915_private *dev_priv = guc_to_i915(guc); + struct rchan *guc_log_relay_chan; + size_t n_subbufs, subbuf_size; + int ret; + + if (!i915_modparams.guc_log_level) + return 0; + + GEM_BUG_ON(guc_log_has_relay(guc)); +/* Keep the size of sub buffers same as shared log buffer */ -
subbuf_size=guc->log.vma->obj->base.size;
+subbuf_size=GUC_LOG_SIZE;
/* Store up to 8 snapshots, which is large enough to buffer sufficient
* boot time logs and provides enough leeway to User, in terms ofdrivers/gpu/drm/i915/intel_guc_log.c +1
-1@@ -407,33 +442,31 @@ static int guc_log_runtime_create(struct intel_guc *guc) DRM_ERROR("Couldn't create relay chan for GuC logging\n"); ret = -ENOMEM; - goto err_vaddr; + goto err; } GEM_BUG_ON(guc_log_relay_chan->subbuf_size < subbuf_size); guc->log.runtime.relay_chan = guc_log_relay_chan; - INIT_WORK(&guc->log.runtime.flush_work, capture_logs_work); return 0;-err_vaddr:
-i915_gem_object_unpin_map(guc->log.vma->obj);
-guc->log.runtime.buf_addr= NULL;Quoted 8 lines
+err:
+ /* logging will be off */
+ i915_modparams.guc_log_level = 0;
return ret;
}
-static voidguc_log_runtime_destroy(struct intel_guc *guc)+voidintel_guc_log_relay_destroy(struct intel_guc *guc){Now exposed to multiple users, we need to document what the locking requirements are here. Or add some local locking.
Quoted 8 lines
Quoted 5 lines
Quoted 1 line
Do you mean locking between
relay_createandrelay_destroy?We need a lock around
guc->log.runtime.relay_chanas the destroy path is not ostensibly serialised between multiple potential callers. Ordinarily I would have said that serialisation for create/destroy/access ofrelay_chanwas guaranteed by init/fini ordering,I was also initially thinking that init/fini ordering should take care of synchronization. Checked further and I see that
relay_openandrelay_destroyare synchronized byrelay_channels_mutexand internally if needed through debugfs inode synchronization. So I feel we need not add new lock.
Sorry. after some more thinking, I am now convinced that lock will be needed as guc_log_has_relay was accessing it without any locking. Will share new rev. Thanks for the review.
Quoted 13 lines
Quoted 12 lines
but that's no longer
clear (based on a 5min read of the patch).The most important question is "can
relay_destroybe called while some user still has access to therelay_chan?"Quoted 4 lines
Quoted 2 lines
Looks like at the
moment,_createis usingstruct_mutex,
relay_createandrelay_destroyare now to be done outside ofstruct_mutex. I will add this documentation to the functions.(
lockdep_assert_held:)
-Chris
Signature · 7 lines
Thanks, Sagar _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Patchwork
CI✗ Fi.CI.IGT: failure for series starting with [1/4] drm/i915/guc: Fix lockdep due to log relay channel handling under struct_mutex
In reply to Sagar Arun Kamble== Series Details ==
== Series Details ==
Series: series starting with [1/4] drm/i915/guc: Fix lockdep due to log relay channel handling under struct_mutex
URL : https://patchwork.freedesktop.org/series/36875/
State : failure
== Summary ==
Test drv_suspend:
Subgroup forcewake:
skip -> PASS (shard-hsw)
Test kms_frontbuffer_tracking:
Subgroup fbc-1p-primscrn-pri-indfb-draw-render:
fail -> PASS (shard-apl) fdo#101623 +1
Test kms_flip:
Subgroup 2x-flip-vs-expired-vblank-interruptible:
pass -> FAIL (shard-hsw)
Subgroup rcs-wf_vblank-vs-dpms-interruptible:
dmesg-warn -> PASS (shard-hsw) fdo#102614
Subgroup flip-vs-expired-vblank-interruptible:
fail -> PASS (shard-apl) fdo#102887
Subgroup modeset-vs-vblank-race-interruptible:
pass -> FAIL (shard-apl) fdo#103060
Subgroup vblank-vs-suspend-interruptible:
incomplete -> PASS (shard-hsw) fdo#100368fdo#101623 https://bugs.freedesktop.org/show_bug.cgi?id=101623
fdo#102614 https://bugs.freedesktop.org/show_bug.cgi?id=102614
fdo#102887 https://bugs.freedesktop.org/show_bug.cgi?id=102887
fdo#103060 https://bugs.freedesktop.org/show_bug.cgi?id=103060
fdo#100368 https://bugs.freedesktop.org/show_bug.cgi?id=100368
shard-apl total:2702 pass:1669 dwarn:1 dfail:0 fail:25 skip:1006 time:14465s shard-hsw total:2780 pass:1721 dwarn:1 dfail:0 fail:15 skip:1042 time:15536s shard-snb total:2780 pass:1316 dwarn:1 dfail:0 fail:13 skip:1450 time:8109s
Blacklisted hosts:
shard-kbl total:2776 pass:1834 dwarn:1 dfail:0 fail:24 skip:916 time:10705s
== Logs ==
For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_7735/shards.html
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx