drm/xe: Add vm snapshot mutex for easily taking a vm snapshot during devcoredump

The devcoredump is done in fence signaling context. Because of this, we
cannot take any of the normal mutexes or we would invert.

Normal: Take vm->lock, dma_fence_wait()
Devcoredump: from dma_fence_wait() context, take vm->lock.

This doesn't work, and we only care about integrity, so take the locks
around additions and removals of vma's.

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240221133024.898315-5-maarten.lankhorst@linux.intel.com

diff --git a/drivers/gpu/drm/xe/xe_vm.c b/drivers/gpu/drm/xe/xe_vm.c
index df1e3841005d48a6c76e28b3f5bfc10e5728c266..5c8cbaf0b8d985e673efff43e3cde3ca138e2cb5 100644 (file)
--- a/drivers/gpu/drm/xe/xe_vm.c
+++ b/drivers/gpu/drm/xe/xe_vm.c
@@ -1055,7 +1055,9 @@ static int xe_vm_insert_vma(struct xe_vm *vm, struct xe_vma *vma)
        xe_assert(vm->xe, xe_vma_vm(vma) == vm);
        lockdep_assert_held(&vm->lock);
 
+       mutex_lock(&vm->snap_mutex);
        err = drm_gpuva_insert(&vm->gpuvm, &vma->gpuva);
+       mutex_unlock(&vm->snap_mutex);
        XE_WARN_ON(err);        /* Shouldn't be possible */
 
        return err;
@@ -1066,7 +1068,9 @@ static void xe_vm_remove_vma(struct xe_vm *vm, struct xe_vma *vma)
        xe_assert(vm->xe, xe_vma_vm(vma) == vm);
        lockdep_assert_held(&vm->lock);
 
+       mutex_lock(&vm->snap_mutex);
        drm_gpuva_remove(&vma->gpuva);
+       mutex_unlock(&vm->snap_mutex);
        if (vm->usm.last_fault_vma == vma)
                vm->usm.last_fault_vma = NULL;
 }
@@ -1293,6 +1297,7 @@ struct xe_vm *xe_vm_create(struct xe_device *xe, u32 flags)
        vm->flags = flags;
 
        init_rwsem(&vm->lock);
+       mutex_init(&vm->snap_mutex);
 
        INIT_LIST_HEAD(&vm->rebind_list);
 
@@ -1418,6 +1423,7 @@ err_close:
        return ERR_PTR(err);
 
 err_no_resv:
+       mutex_destroy(&vm->snap_mutex);
        for_each_tile(tile, xe, id)
                xe_range_fence_tree_fini(&vm->rftree[id]);
        kfree(vm);
@@ -1517,6 +1523,8 @@ void xe_vm_close_and_put(struct xe_vm *vm)
 
        up_write(&vm->lock);
 
+       mutex_destroy(&vm->snap_mutex);
+
        mutex_lock(&xe->usm.lock);
        if (vm->flags & XE_VM_FLAG_FAULT_MODE)
                xe->usm.num_vm_in_fault_mode--;

diff --git a/drivers/gpu/drm/xe/xe_vm_types.h b/drivers/gpu/drm/xe/xe_vm_types.h
index a975ac83eccae45a403297f9ae88bef34a3c5957..7d4f810f9c0465aa0b61f2b8144ebf080cf57590 100644 (file)
--- a/drivers/gpu/drm/xe/xe_vm_types.h
+++ b/drivers/gpu/drm/xe/xe_vm_types.h
@@ -160,6 +160,11 @@ struct xe_vm {
         * VM
         */
        struct rw_semaphore lock;
+       /**
+        * @snap_mutex: Mutex used to guard insertions and removals from gpuva,
+        * so we can take a snapshot safely from devcoredump.
+        */
+       struct mutex snap_mutex;
 
        /**
         * @rebind_list: list of VMAs that need rebinding. Protected by the