Bug 111915

https://intel-gfx-ci.01.org/tree/drm-tip/drmtip_382/fi-elk-e7500/igt@gem_mmap_gtt@basic-small-copy.html
<4> [42.467859] ======================================================
<4> [42.467863] WARNING: possible circular locking dependency detected
<4> [42.467869] 5.4.0-rc1-g78b99839aed8-drmtip_382+ #1 Tainted: G     U           
<4> [42.467874] ------------------------------------------------------
<4> [42.467878] kworker/u8:5/155 is trying to acquire lock:
<4> [42.467882] ffff8d348559dbb8 (&mapping->i_mmap_rwsem){++++}, at: unmap_mapping_pages+0x48/0x130
<4> [42.467893] 
but task is already holding lock:
<4> [42.467897] ffff8d34856093a0 (&vm->mutex){+.+.}, at: i915_vma_unbind+0xe6/0x4a0 [i915]
<4> [42.468015] 
which lock already depends on the new lock.

<4> [42.468019] 
the existing dependency chain (in reverse order) is:
<4> [42.468024] 
-> #2 (&vm->mutex){+.+.}:
<4> [42.468031]        __mutex_lock+0x9a/0x9d0
<4> [42.468089]        i915_vma_unbind+0xa4/0x4a0 [i915]
<4> [42.468146]        i915_gem_object_unbind+0x153/0x1c0 [i915]
<4> [42.468202]        userptr_mn_invalidate_range_start+0x9f/0x200 [i915]
<4> [42.468207]        __mmu_notifier_invalidate_range_start+0xa3/0x180
<4> [42.468212]        unmap_vmas+0x143/0x150
<4> [42.468216]        unmap_region+0xa3/0x100
<4> [42.468220]        __do_munmap+0x25d/0x490
<4> [42.468223]        __vm_munmap+0x6e/0xc0
<4> [42.468227]        __x64_sys_munmap+0x12/0x20
<4> [42.468231]        do_syscall_64+0x4f/0x210
<4> [42.468236]        entry_SYSCALL_64_after_hwframe+0x49/0xbe
<4> [42.468240] 
-> #1 (mmu_notifier_invalidate_range_start){+.+.}:
<4> [42.468247]        page_mkclean_one+0xda/0x210
<4> [42.468251]        rmap_walk_file+0xff/0x260
<4> [42.468255]        page_mkclean+0x9f/0xb0
<4> [42.468259]        clear_page_dirty_for_io+0xa2/0x300
<4> [42.468264]        mpage_submit_page+0x1a/0x70
<4> [42.468268]        mpage_process_page_bufs+0xe7/0x110
<4> [42.468272]        mpage_prepare_extent_to_map+0x1d2/0x2b0
<4> [42.468277]        ext4_writepages+0x592/0x1230
<4> [42.468280]        do_writepages+0x46/0xe0
<4> [42.468285]        __filemap_fdatawrite_range+0xc6/0x100
<4> [42.468289]        file_write_and_wait_range+0x3c/0x90
<4> [42.468293]        ext4_sync_file+0x154/0x500
<4> [42.468298]        do_fsync+0x33/0x60
<4> [42.468302]        __x64_sys_fsync+0xb/0x10
<4> [42.468306]        do_syscall_64+0x4f/0x210
<4> [42.468309]        entry_SYSCALL_64_after_hwframe+0x49/0xbe
<4> [42.468313] 
-> #0 (&mapping->i_mmap_rwsem){++++}:
<4> [42.468320]        __lock_acquire+0x1328/0x15d0
<4> [42.468324]        lock_acquire+0xa7/0x1c0
<4> [42.468328]        down_write+0x33/0x70
<4> [42.468332]        unmap_mapping_pages+0x48/0x130
<4> [42.468390]        i915_vma_revoke_mmap+0x81/0x1b0 [i915]
<4> [42.468448]        i915_vma_unbind+0xee/0x4a0 [i915]
<4> [42.468506]        i915_vma_destroy+0x31/0x2f0 [i915]
<4> [42.468561]        __i915_gem_free_objects+0xb8/0x4b0 [i915]
<4> [42.468567]        process_one_work+0x26a/0x620
<4> [42.468571]        worker_thread+0x37/0x380
<4> [42.468575]        kthread+0x119/0x130
<4> [42.468579]        ret_from_fork+0x3a/0x50
<4> [42.468582] 
other info that might help us debug this:

<4> [42.468587] Chain exists of:
  &mapping->i_mmap_rwsem --> mmu_notifier_invalidate_range_start --> &vm->mutex

<4> [42.468596]  Possible unsafe locking scenario:

<4> [42.468601]        CPU0                    CPU1
<4> [42.468604]        ----                    ----
<4> [42.468607]   lock(&vm->mutex);
<4> [42.468610]                                lock(mmu_notifier_invalidate_range_start);
<4> [42.468616]                                lock(&vm->mutex);
<4> [42.468620]   lock(&mapping->i_mmap_rwsem);
<4> [42.468624] 
 *** DEADLOCK ***