https://bugs.freedesktop.org/show_bug.cgi?id=49030
Bug #: 49030 Summary: Possible recursive locking detected in r600g Classification: Unclassified Product: Mesa Version: git Platform: Other OS/Version: All Status: NEW Severity: minor Priority: medium Component: Drivers/Gallium/r600 AssignedTo: dri-devel@lists.freedesktop.org ReportedBy: alexandre.f.demers@gmail.com
Here is what I see under dmesg:
============================================= [ INFO: possible recursive locking detected ] 3.4.0-rc3+ #120 Tainted: G C --------------------------------------------- gnome-shell/1120 is trying to acquire lock: (&vm->mutex){+.+...}, at: [<ffffffffa034be9a>] radeon_vm_unbind+0x24/0x40 [radeon]
but task is already holding lock: (&vm->mutex){+.+...}, at: [<ffffffffa035b84c>] radeon_cs_ioctl+0x434/0x598 [radeon]
other info that might help us debug this: Possible unsafe locking scenario:
CPU0 ---- lock(&vm->mutex); lock(&vm->mutex);
*** DEADLOCK ***
May be due to missing lock nesting notation
2 locks held by gnome-shell/1120: #0: (&mutex->mutex){+.+.+.}, at: [<ffffffffa035b44e>] radeon_cs_ioctl+0x36/0x598 [radeon] #1: (&vm->mutex){+.+...}, at: [<ffffffffa035b84c>] radeon_cs_ioctl+0x434/0x598 [radeon]
stack backtrace: Pid: 1120, comm: gnome-shell Tainted: G C 3.4.0-rc3+ #120 Call Trace: [<ffffffff8108acd9>] __lock_acquire+0xbf6/0xd11 [<ffffffff8108a5a3>] ? __lock_acquire+0x4c0/0xd11 [<ffffffff8108b211>] lock_acquire+0x92/0x104 [<ffffffffa034be9a>] ? radeon_vm_unbind+0x24/0x40 [radeon] [<ffffffff8108a5a3>] ? __lock_acquire+0x4c0/0xd11 [<ffffffff814d4460>] __mutex_lock_common+0x48/0x34e [<ffffffffa034be9a>] ? radeon_vm_unbind+0x24/0x40 [radeon] [<ffffffffa034be9a>] ? radeon_vm_unbind+0x24/0x40 [radeon] [<ffffffff8108b69e>] ? trace_hardirqs_on_caller+0x123/0x17f [<ffffffff814d4839>] mutex_lock_nested+0x2f/0x36 [<ffffffffa035b84c>] ? radeon_cs_ioctl+0x434/0x598 [radeon] [<ffffffffa034be9a>] radeon_vm_unbind+0x24/0x40 [radeon] [<ffffffffa034c2f9>] radeon_vm_bind+0xb9/0x1c4 [radeon] [<ffffffffa035b857>] radeon_cs_ioctl+0x43f/0x598 [radeon] [<ffffffffa01fa9f9>] drm_ioctl+0x2d6/0x3c0 [drm] [<ffffffff8128dfed>] ? rcu_read_unlock+0x1c/0x1e [<ffffffffa035b418>] ? radeon_cs_finish_pages+0x91/0x91 [radeon] [<ffffffff8128e44a>] ? avc_has_perm_flags+0x6b/0x7f [<ffffffff8113c20c>] vfs_ioctl+0x24/0x2f [<ffffffff8113caf7>] do_vfs_ioctl+0x412/0x455 [<ffffffff8113cb90>] sys_ioctl+0x56/0x7a [<ffffffff814dd3ad>] system_call_fastpath+0x1a/0x1f
It seems to be only a warning. However, if possible, it should be silenced if there is no reason to have it. I know I reported something similar a couple of months ago and it was fixed by moving some code around.