Hi Steve, Boris

On 11/27/25 16:08, Boris Brezillon wrote:
On Thu, 27 Nov 2025 16:02:15 +0000
Steven Price <[email protected]> wrote:

On 27/11/2025 08:12, Akash Goel wrote:
This commit prevents the possibility of a use after free issue in the
GROUP_CREATE ioctl function, which arose as pointer to the group is
accessed in that ioctl function after storing it in the Xarray.
A malicious userspace can second guess the handle of a group and try
to call GROUP_DESTROY ioctl from another thread around the same time
as GROUP_CREATE ioctl.

To prevent the use after free exploit, this commit uses a mark on an
entry of group pool Xarray which is added just before returning from
the GROUP_CREATE ioctl function. The mark is checked for all ioctls
that specify the group handle and so userspace won't be abe to delete
a group that isn't marked yet.

Co-developed-by: Boris Brezillon <[email protected]>
Signed-off-by: Akash Goel <[email protected]>

Reviewed-by: Steven Price <[email protected]>

I *think* this should have a...

Fixes: d2624d90a0b7 ("drm/panthor: assign unique names to queues")

... as I don't believe it was a problem before the rearrangement that
happened there.

Oh, yeah, I didn't notice the commit was missing a Fixes tag, and
you're correct about the offending commit.


Sorry for not adding the Fixes tag.


I think the problem has been present since the beginning and the Fixes tag should be

Fixes: de85488138247 ("drm/panthor: Add the scheduler logical block")


Initially the code was like this,


ret = xa_alloc(&gpool->xa, &gid, group, XA_LIMIT(1, MAX_GROUPS_PER_POOL), GFP_KERNEL);
        if (ret)
                goto err_put_group;

        mutex_lock(&sched->reset.lock);
        if (atomic_read(&sched->reset.in_progress)) {
                panthor_group_stop(group);
        } else {
                mutex_lock(&sched->lock);
                list_add_tail(&group->run_node,
                              &sched->groups.idle[group->priority]);
                mutex_unlock(&sched->lock);
        }
        mutex_unlock(&sched->reset.lock);

        return gid;

If the GROUP_CREATE ioctl thread somehow gets preempted immediately after xa_alloc(), then another thread might succeed in freeing the group through GROUP_DESTROY ioctl.

Initially it would have been very difficult to trigger the UAF, but
d2624d90a0b7 ("drm/panthor: assign unique names to queues") would have made the code more susceptible to UAF.

Please kindly correct me if I interpreted things incorrectly.

Will accordingly send a v2.

Best regards
Akash



Thanks,
Steve

---
  drivers/gpu/drm/panthor/panthor_sched.c | 19 +++++++++++++++----
  1 file changed, 15 insertions(+), 4 deletions(-)

diff --git a/drivers/gpu/drm/panthor/panthor_sched.c 
b/drivers/gpu/drm/panthor/panthor_sched.c
index b834123a6560..a6b8024e1a3c 100644
--- a/drivers/gpu/drm/panthor/panthor_sched.c
+++ b/drivers/gpu/drm/panthor/panthor_sched.c
@@ -779,6 +779,12 @@ struct panthor_job_profiling_data {
   */
  #define MAX_GROUPS_PER_POOL 128
+/*
+ * Mark added on an entry of group pool Xarray to identify if the group has
+ * been fully initialized and can be accessed elsewhere in the driver code.
+ */
+#define GROUP_REGISTERED XA_MARK_1
+
  /**
   * struct panthor_group_pool - Group pool
   *
@@ -3007,7 +3013,7 @@ void panthor_fdinfo_gather_group_samples(struct 
panthor_file *pfile)
                return;
xa_lock(&gpool->xa);
-       xa_for_each(&gpool->xa, i, group) {
+       xa_for_each_marked(&gpool->xa, i, group, GROUP_REGISTERED) {
                guard(spinlock)(&group->fdinfo.lock);
                pfile->stats.cycles += group->fdinfo.data.cycles;
                pfile->stats.time += group->fdinfo.data.time;
@@ -3727,6 +3733,8 @@ int panthor_group_create(struct panthor_file *pfile,
group_init_task_info(group); + xa_set_mark(&gpool->xa, gid, GROUP_REGISTERED);
+
        return gid;
err_erase_gid:
@@ -3744,6 +3752,9 @@ int panthor_group_destroy(struct panthor_file *pfile, u32 
group_handle)
        struct panthor_scheduler *sched = ptdev->scheduler;
        struct panthor_group *group;
+ if (!xa_get_mark(&gpool->xa, group_handle, GROUP_REGISTERED))
+               return -EINVAL;
+
        group = xa_erase(&gpool->xa, group_handle);
        if (!group)
                return -EINVAL;
@@ -3769,12 +3780,12 @@ int panthor_group_destroy(struct panthor_file *pfile, 
u32 group_handle)
  }
static struct panthor_group *group_from_handle(struct panthor_group_pool *pool,
-                                              u32 group_handle)
+                                              unsigned long group_handle)
  {
        struct panthor_group *group;
xa_lock(&pool->xa);
-       group = group_get(xa_load(&pool->xa, group_handle));
+       group = group_get(xa_find(&pool->xa, &group_handle, group_handle, 
GROUP_REGISTERED));
        xa_unlock(&pool->xa);
return group;
@@ -3861,7 +3872,7 @@ panthor_fdinfo_gather_group_mem_info(struct panthor_file 
*pfile,
                return;
xa_lock(&gpool->xa);
-       xa_for_each(&gpool->xa, i, group) {
+       xa_for_each_marked(&gpool->xa, i, group, GROUP_REGISTERED) {
                stats->resident += group->fdinfo.kbo_sizes;
                if (group->csg_id >= 0)
                        stats->active += group->fdinfo.kbo_sizes;


Reply via email to