diff options
author | Chris Wilson <chris@chris-wilson.co.uk> | 2016-09-06 08:38:44 +0100 |
---|---|---|
committer | Chris Wilson <chris@chris-wilson.co.uk> | 2016-09-06 09:25:03 +0100 |
commit | 19eb9189c55715b8e15401fb6d862c5766724220 (patch) | |
tree | bce1e31258c233915d5205647fdde91ddbf7fd3a | |
parent | fc32de9358c0a7b6f7c4639a3a8056b10a9b7669 (diff) | |
download | linux-19eb9189c55715b8e15401fb6d862c5766724220.tar.bz2 |
drm/i915: Don't wait for a spinlock inside error capture
If we can't grab the breadcrumb's spinlock, possibly due to a driver
deadlock inside the waiters, ignore them. Like hangcheck, error
capturing must work no matter how the driver/GPU dies.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: http://patchwork.freedesktop.org/patch/msgid/20160906073844.22561-1-chris@chris-wilson.co.uk
Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
-rw-r--r-- | drivers/gpu/drm/i915/i915_gpu_error.c | 25 |
1 files changed, 20 insertions, 5 deletions
diff --git a/drivers/gpu/drm/i915/i915_gpu_error.c b/drivers/gpu/drm/i915/i915_gpu_error.c index aed55e4f100f..d45a959b7c53 100644 --- a/drivers/gpu/drm/i915/i915_gpu_error.c +++ b/drivers/gpu/drm/i915/i915_gpu_error.c @@ -489,7 +489,10 @@ int i915_error_state_to_str(struct drm_i915_error_state_buf *m, } } - if (ee->num_waiters) { + if (IS_ERR(ee->waiters)) { + err_printf(m, "%s --- ? waiters [unable to acquire spinlock]\n", + dev_priv->engine[i].name); + } else if (ee->num_waiters) { err_printf(m, "%s --- %d waiters\n", dev_priv->engine[i].name, ee->num_waiters); @@ -648,7 +651,8 @@ static void i915_error_state_free(struct kref *error_ref) i915_error_object_free(ee->wa_ctx); kfree(ee->requests); - kfree(ee->waiters); + if (!IS_ERR_OR_NULL(ee->waiters)) + kfree(ee->waiters); } i915_error_object_free(error->semaphore); @@ -933,7 +937,14 @@ static void error_record_engine_waiters(struct intel_engine_cs *engine, ee->num_waiters = 0; ee->waiters = NULL; - spin_lock(&b->lock); + if (RB_EMPTY_ROOT(&b->waiters)) + return; + + if (!spin_trylock(&b->lock)) { + ee->waiters = ERR_PTR(-EDEADLK); + return; + } + count = 0; for (rb = rb_first(&b->waiters); rb != NULL; rb = rb_next(rb)) count++; @@ -947,9 +958,13 @@ static void error_record_engine_waiters(struct intel_engine_cs *engine, if (!waiter) return; - ee->waiters = waiter; + if (!spin_trylock(&b->lock)) { + kfree(waiter); + ee->waiters = ERR_PTR(-EDEADLK); + return; + } - spin_lock(&b->lock); + ee->waiters = waiter; for (rb = rb_first(&b->waiters); rb; rb = rb_next(rb)) { struct intel_wait *w = container_of(rb, typeof(*w), node); |