[Intel-xe] [PATCH] drm/xe: destroy clients engine and vm xarrays on close

[Mika Kuoppala]

Lucas De Marchi <lucas.demarchi at intel.com> writes:

> On Wed, Apr 12, 2023 at 02:09:23PM +0300, Mika Kuoppala wrote:
>>xe_file_close cleanups the xarrays but forgets
>>to destroy them causing a memleak in xarray internals.
>>Found with kmemleak.
>>
>>Signed-off-by: Mika Kuoppala <mika.kuoppala at linux.intel.com>
>>---
>> drivers/gpu/drm/xe/xe_device.c | 2 ++
>> 1 file changed, 2 insertions(+)
>>
>>diff --git a/drivers/gpu/drm/xe/xe_device.c b/drivers/gpu/drm/xe/xe_device.c
>>index 45d6e5ff47fd..30415015db4a 100644
>>--- a/drivers/gpu/drm/xe/xe_device.c
>>+++ b/drivers/gpu/drm/xe/xe_device.c
>>@@ -70,6 +70,7 @@ static void xe_file_close(struct drm_device *dev, struct drm_file *file)
>> 		xe_engine_put(e);
>> 	}
>> 	mutex_unlock(&xef->engine.lock);
>>+	xa_destroy(&xef->engine.xa);
>> 	mutex_destroy(&xef->engine.lock);
>> 	device_kill_persistent_engines(xe, xef);
>>
>>@@ -77,6 +78,7 @@ static void xe_file_close(struct drm_device *dev, struct drm_file *file)
>> 	xa_for_each(&xef->vm.xa, idx, vm)
>> 		xe_vm_close_and_put(vm);
>> 	mutex_unlock(&xef->vm.lock);
>>+	xa_destroy(&xef->vm.xa);
>
> looks good to me considering the xe_file_open() initialized it. But
> shouldn't we print a warning or something on !xa_empty()?

The xe_vm_close_and_put() does not remove from the list.
If I add warn here, it gets triggered.
So some work is needed on vm side if we want to enforce
it being empty at this point.

-Mika

>
> Lucas De Marchi
>
>> 	mutex_destroy(&xef->vm.lock);
>>
>> 	kfree(xef);
>>-- 
>>2.34.1
>>