Re: [PATCH v3 1/2] drm/xe/uapi: disallow bind queue sharing

[Matthew Auld <matthew.auld@intel.com>]

On 20/11/2025 15:34, Lucas De Marchi wrote:
> On Thu, Nov 20, 2025 at 01:27:29PM +0000, Matthew Auld wrote:
>> Currently this is very broken if someone attempts to create a bind
>> queue and share it across multiple VMs. For example currently we assume
>> it is safe to acquire the user VM lock to protect some of the bind queue
>> state, but if allow sharing the bind queue with multiple VMs then this
>> quickly breaks down.
>>
>> To fix this reject using a bind queue with any VM that is not the same
>> VM that was originally passed when creating the bind queue. This a uAPI
>> change, however this was more of an oversight on kernel side that we
>> didn't reject this, and expectation is that userspace shouldn't be using
>> bind queues in this way, so in theory this change should go unnoticed.
>>
>> Based on a patch from Matt Brost.
>>
>> v2 (Matt B):
>>  - Hold the vm lock over queue create, to ensure it can't be closed as
>>    we attach the user_vm to the queue.
>>  - Make sure we actually check for NULL user_vm in destruction path.
>> v3:
>>  - Fix error path handling.
>>
>> Fixes: dd08ebf6c352 ("drm/xe: Introduce a new DRM driver for Intel GPUs")
>> Reported-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
>> Signed-off-by: Matthew Auld <matthew.auld@intel.com>
>> Cc: José Roberto de Souza <jose.souza@intel.com>
>> Cc: Matthew Brost <matthew.brost@intel.com>
>> Cc: Michal Mrozek <michal.mrozek@intel.com>
>> Cc: Carl Zhang <carl.zhang@intel.com>
>> Cc: <stable@vger.kernel.org> # v6.8+
> 
> we never had any platform officially supported back in 6.8. Let's make
> it 6.12 to avoid useless backporting work.
> 
>> Acked-by: José Roberto de Souza <jose.souza@intel.com>
> 
> Michal / Carl, can you also ack compute/media are ok with this change?

Ping on this? I did a cursory grep for DRM_XE_ENGINE_CLASS_VM_BIND and 
found no users in compute-runtime or media-driver in upstream. This 
change should only be noticeable if you directly use 
DRM_XE_ENGINE_CLASS_VM_BIND to create a dedicated bind queue, which you 
then pass into vm_bind.

> 
> Lucas De Marchi