drm/amdgpu: Add lockdep annotations for lock ordering validation

Add lockdep annotations to teach lockdep the correct lock hierarchy
and catch ordering violations during development. This follows the
pattern established by dma-resv in drivers/dma-buf/dma-resv.c.

Lock ordering hierarchy (outermost to innermost):

1. userq_sch_mutex   - Global userq scheduler (enforce_isolation)
2. userq_mutex       - Per-context userq (held across queue create/destroy)
3. notifier_lock     - MMU notifier synchronization
4. vram_lock         - VRAM memory allocator
5. reset_domain->sem - GPU reset synchronization
6. reset_lock        - Reset control mutex
7. srbm_mutex        - SRBM register access
8. grbm_idx_mutex    - GRBM index register access
9. mmio_idx_lock     - MMIO index access (spinlock)

The implementation provides:
- Lock ordering training at module init (amdgpu_lockdep_init)
- Lock class association for real driver locks (amdgpu_lockdep_set_class)

Dummy locks are associated with the same class keys as real driver locks
via lockdep_set_class(), ensuring lockdep connects the training ordering
with actual runtime locks.

Testing:
  Build the kernel with CONFIG_PROVE_LOCKING=y (enables CONFIG_LOCKDEP):
    scripts/config --enable PROVE_LOCKING
    scripts/config --enable DEBUG_LOCKDEP
    make -j$(nproc)

  On boot, dmesg should show:
    AMDGPU: Lockdep annotations initialized (9 lock levels)

  The companion IGT test (tests/amdgpu/amd_lockdep) exercises lock-heavy
  GPU code paths concurrently to trigger lockdep warnings on violations:
    sudo ./build/tests/amdgpu/amd_lockdep
    sudo dmesg | grep -A 50 "circular locking dependency"

  IGT subtests:
    concurrent-reset-and-submit  - reset_sem vs submission locks
    concurrent-mmap-and-evict    - mmap_lock vs vram_lock
    concurrent-userptr-and-reset - notifier_lock vs reset_sem
    stress-all-paths             - all of the above simultaneously

  A clean dmesg (no "circular locking dependency" or "possible recursive
  locking detected" messages) confirms no lock ordering violations.

  For CI integration, the test should be run on kernels compiled with
  CONFIG_LOCKDEP=y; dmesg is scanned post-run for lockdep splats.

v2: (Christian)
- Move notifier_lock and vram_lock before reset locks in hierarchy.
  HMM invalidation holds notifier_lock and can wait for GPU reset
  completion, so notifier_lock must be outer to reset_domain->sem.
- Associate dummy locks with lock class keys via lockdep_set_class()
  so lockdep connects training with real driver locks.
- Update commit message to list all 9 lock levels.

Requires CONFIG_PROVE_LOCKING=y to activate.

Cc: Christian Konig <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Vitaly Prosyak <vitaly.prosyak@amd.com>
Reviewed-by: Christian Konig <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This commit is contained in:
Vitaly Prosyak 2026-05-13 16:08:30 -04:00 committed by Alex Deucher
parent 92a8dba246
commit 1d0f5838b1
6 changed files with 242 additions and 1 deletions

View File

@ -69,7 +69,7 @@ amdgpu-y += amdgpu_device.o amdgpu_reg_access.o amdgpu_doorbell_mgr.o amdgpu_kms
amdgpu_vm_sdma.o amdgpu_discovery.o amdgpu_ras_eeprom.o amdgpu_nbio.o \
amdgpu_umc.o smu_v11_0_i2c.o amdgpu_fru_eeprom.o amdgpu_rap.o \
amdgpu_fw_attestation.o amdgpu_securedisplay.o \
amdgpu_eeprom.o amdgpu_mca.o amdgpu_psp_ta.o amdgpu_lsdma.o \
amdgpu_eeprom.o amdgpu_mca.o amdgpu_psp_ta.o amdgpu_lsdma.o amdgpu_lockdep.o \
amdgpu_ring_mux.o amdgpu_xcp.o amdgpu_seq64.o amdgpu_aca.o amdgpu_dev_coredump.o \
amdgpu_cper.o amdgpu_userq_fence.o amdgpu_eviction_fence.o amdgpu_ip.o

View File

@ -105,6 +105,7 @@
#include "amdgpu_mca.h"
#include "amdgpu_aca.h"
#include "amdgpu_ras.h"
#include "amdgpu_lockdep.h"
#include "amdgpu_cper.h"
#include "amdgpu_xcp.h"
#include "amdgpu_seq64.h"

View File

@ -3757,6 +3757,9 @@ int amdgpu_device_init(struct amdgpu_device *adev,
mutex_init(&adev->pm.stable_pstate_ctx_lock);
mutex_init(&adev->benchmark_mutex);
mutex_init(&adev->gfx.reset_sem_mutex);
/* Associate locks with lockdep classes for ordering validation */
amdgpu_lockdep_set_class(adev);
/* Initialize the mutex for cleaner shader isolation between GFX and compute processes */
mutex_init(&adev->enforce_isolation_mutex);
for (i = 0; i < MAX_XCP; ++i) {

View File

@ -3158,6 +3158,9 @@ static int __init amdgpu_init(void)
{
int r;
/* Train lockdep on correct lock ordering */
amdgpu_lockdep_init();
r = amdgpu_sync_init();
if (r)
return r;

View File

@ -0,0 +1,195 @@
// SPDX-License-Identifier: MIT
/*
* Copyright 2024 Advanced Micro Devices, Inc.
*
* Lockdep annotation for AMDGPU lock ordering
*
* This module teaches lockdep the correct lock ordering to catch
* potential deadlocks at development time rather than runtime.
*
* Based on dma-resv lockdep approach from:
* drivers/dma-buf/dma-resv.c:dma_resv_lockdep()
*/
#include "amdgpu.h"
#include "amdgpu_reset.h"
#ifdef CONFIG_LOCKDEP
/* Lock class keys for associating with real driver locks */
static struct lock_class_key amdgpu_userq_sch_mutex_key;
static struct lock_class_key amdgpu_userq_mutex_key;
static struct lock_class_key amdgpu_notifier_lock_key;
static struct lock_class_key amdgpu_vram_lock_key;
static struct lock_class_key amdgpu_reset_sem_key;
static struct lock_class_key amdgpu_reset_lock_key;
static struct lock_class_key amdgpu_srbm_lock_key;
static struct lock_class_key amdgpu_grbm_lock_key;
static struct lock_class_key amdgpu_mmio_lock_key;
/**
* amdgpu_lockdep_set_class - Associate lock class keys with real locks
* @adev: AMDGPU device
*
* Call during device init to associate lock classes with actual locks
* so lockdep can track them properly.
*/
void amdgpu_lockdep_set_class(struct amdgpu_device *adev)
{
lockdep_set_class(&adev->gfx.userq_sch_mutex,
&amdgpu_userq_sch_mutex_key);
lockdep_set_class(&adev->notifier_lock, &amdgpu_notifier_lock_key);
lockdep_set_class(&adev->srbm_mutex, &amdgpu_srbm_lock_key);
lockdep_set_class(&adev->grbm_idx_mutex, &amdgpu_grbm_lock_key);
lockdep_set_class(&adev->mmio_idx_lock, &amdgpu_mmio_lock_key);
if (adev->reset_domain)
lockdep_set_class(&adev->reset_domain->sem,
&amdgpu_reset_sem_key);
}
/**
* amdgpu_lockdep_init - Teach lockdep the correct lock ordering
*
* Instantiates dummy objects and takes locks in the correct order to
* train lockdep. This helps catch lock ordering violations during
* development.
*
* Lock ordering hierarchy (outermost to innermost):
*
* 1. userq_sch_mutex - Global userq scheduler (enforce_isolation)
* 2. userq_mutex - Per-context userq (held across queue create/destroy)
* 3. notifier_lock - MMU notifier lock
* 4. vram_lock - VRAM allocator lock
* 5. reset_domain->sem - GPU reset synchronization
* 6. reset_lock - Reset control lock
* 7. srbm_mutex - SRBM register access
* 8. grbm_idx_mutex - GRBM index access
* 9. mmio_idx_lock - MMIO index access (spinlock)
*
* Evidence:
* - userq_sch_mutex -> userq_mutex: amdgpu_gfx_kfd_sch_ctrl() calls
* amdgpu_userq_stop_sched_for_enforce_isolation() which takes userq_mutex
* - userq_mutex -> notifier_lock: userq paths may trigger MMU notifier
* invalidation which acquires notifier_lock
* - notifier_lock -> reset_domain->sem: HMM invalidation callback holds
* notifier_lock and can wait for GPU reset completion, so notifier_lock
* must be outer to reset_domain->sem
* - vram_lock -> reset_domain->sem: VRAM management paths may need to
* wait for ongoing reset to complete
*
* Note: mmap_lock ordering relative to GPU locks is already taught
* by dma-resv (drivers/dma-buf/dma-resv.c).
*/
int amdgpu_lockdep_init(void)
{
struct amdgpu_reset_domain *reset_domain = NULL;
struct amdgpu_reset_control reset_ctl;
struct mutex userq_sch_mutex;
struct mutex userq_mutex;
struct mutex notifier_lock;
struct mutex vram_lock;
struct mutex srbm_mutex;
struct mutex grbm_idx_mutex;
spinlock_t mmio_idx_lock;
unsigned long flags;
/*
* Initialize dummy reset domain
*/
reset_domain = amdgpu_reset_create_reset_domain(SINGLE_DEVICE,
"lockdep_test");
if (!reset_domain)
return -ENOMEM;
/* Initialize dummy locks */
mutex_init(&userq_sch_mutex);
mutex_init(&userq_mutex);
mutex_init(&notifier_lock);
mutex_init(&vram_lock);
mutex_init(&reset_ctl.reset_lock);
mutex_init(&srbm_mutex);
mutex_init(&grbm_idx_mutex);
spin_lock_init(&mmio_idx_lock);
/*
* Associate dummy locks with the same class keys used for real
* driver locks. This ensures lockdep connects the ordering learned
* here with the actual locks used at runtime.
*/
lockdep_set_class(&userq_sch_mutex, &amdgpu_userq_sch_mutex_key);
lockdep_set_class(&userq_mutex, &amdgpu_userq_mutex_key);
lockdep_set_class(&notifier_lock, &amdgpu_notifier_lock_key);
lockdep_set_class(&vram_lock, &amdgpu_vram_lock_key);
lockdep_set_class(&reset_domain->sem, &amdgpu_reset_sem_key);
lockdep_set_class(&reset_ctl.reset_lock, &amdgpu_reset_lock_key);
lockdep_set_class(&srbm_mutex, &amdgpu_srbm_lock_key);
lockdep_set_class(&grbm_idx_mutex, &amdgpu_grbm_lock_key);
lockdep_set_class(&mmio_idx_lock, &amdgpu_mmio_lock_key);
/*
* Take locks in the correct order to train lockdep.
* This establishes the dependency chain.
*/
/* Level 1: Global userq scheduler mutex (outermost) */
mutex_lock(&userq_sch_mutex);
/* Level 2: Per-context userq mutex */
mutex_lock(&userq_mutex);
/* Level 3: MMU notifier lock */
mutex_lock(&notifier_lock);
/* Level 4: VRAM allocator lock */
mutex_lock(&vram_lock);
/* Level 5: Reset domain semaphore */
down_read(&reset_domain->sem);
/* Level 6: Reset control lock */
mutex_lock(&reset_ctl.reset_lock);
/*
* Mark potential memory reclaim boundary.
* GPU operations might trigger memory allocation/reclaim.
*/
fs_reclaim_acquire(GFP_KERNEL);
/* Level 7: SRBM register access */
mutex_lock(&srbm_mutex);
/* Level 8: GRBM index access */
mutex_lock(&grbm_idx_mutex);
/* Level 9: MMIO index access (innermost lock, spinlock) */
spin_lock_irqsave(&mmio_idx_lock, flags);
/*
* All locks acquired in order.
* Lockdep has now learned the valid dependency chain.
*/
/* Release in reverse order */
spin_unlock_irqrestore(&mmio_idx_lock, flags);
mutex_unlock(&grbm_idx_mutex);
mutex_unlock(&srbm_mutex);
fs_reclaim_release(GFP_KERNEL);
mutex_unlock(&reset_ctl.reset_lock);
up_read(&reset_domain->sem);
mutex_unlock(&vram_lock);
mutex_unlock(&notifier_lock);
mutex_unlock(&userq_mutex);
mutex_unlock(&userq_sch_mutex);
/* Cleanup */
amdgpu_reset_put_reset_domain(reset_domain);
pr_info("AMDGPU: Lockdep annotations initialized (9 lock levels)\n");
return 0;
}
#endif /* CONFIG_LOCKDEP */

View File

@ -0,0 +1,39 @@
/* SPDX-License-Identifier: MIT */
/*
* Copyright 2024 Advanced Micro Devices, Inc.
*
* Lockdep annotation interface for AMDGPU
*/
#ifndef __AMDGPU_LOCKDEP_H__
#define __AMDGPU_LOCKDEP_H__
#include <linux/lockdep.h>
struct amdgpu_device;
#ifdef CONFIG_LOCKDEP
/**
* amdgpu_lockdep_init - Train lockdep on correct lock ordering
*
* Call once during module init to establish the lock dependency chain.
*/
int amdgpu_lockdep_init(void);
/**
* amdgpu_lockdep_set_class - Associate lock class keys with real locks
* @adev: AMDGPU device
*
* Call during device init to associate lock classes with actual locks.
*/
void amdgpu_lockdep_set_class(struct amdgpu_device *adev);
#else /* !CONFIG_LOCKDEP */
static inline int amdgpu_lockdep_init(void) { return 0; }
static inline void amdgpu_lockdep_set_class(struct amdgpu_device *adev) {}
#endif /* CONFIG_LOCKDEP */
#endif /* __AMDGPU_LOCKDEP_H__ */