drm/i915: Use the full hammer when shutting down the rcu tasks
To flush all call_rcu() tasks (here from i915_gem_free_object()) we need
to call rcu_barrier() (not synchronize_rcu()). If we don't then we may
still have objects being freed as we continue to teardown the driver -
in particular, the recently released rings may race with the memory
manager shutdown resulting in sporadic:
[ 142.217186] WARNING: CPU: 7 PID: 6185 at drivers/gpu/drm/drm_mm.c:932 drm_mm_takedown+0x2e/0x40
[ 142.217187] Memory manager not clean during takedown.
[ 142.217187] Modules linked in: i915(-) x86_pkg_temp_thermal intel_powerclamp coretemp crct10dif_pclmul crc32_pclmul ghash_clmulni_intel lpc_ich snd_hda_codec_realtek snd_hda_codec_generic mei_me mei snd_hda_codec_hdmi snd_hda_codec snd_hwdep snd_hda_core snd_pcm e1000e ptp pps_core [last unloaded: snd_hda_intel]
[ 142.217199] CPU: 7 PID: 6185 Comm: rmmod Not tainted 4.9.0-rc2-CI-Trybot_242+ #1
[ 142.217199] Hardware name: LENOVO 10AGS00601/SHARKBAY, BIOS FBKT34AUS 04/24/2013
[ 142.217200] ffffc90002ecfce0 ffffffff8142dd65 ffffc90002ecfd30 0000000000000000
[ 142.217202] ffffc90002ecfd20 ffffffff8107e4e6 000003a40778c2a8 ffff880401355c48
[ 142.217204] ffff88040778c2a8 ffffffffa040f3c0 ffffffffa040f4a0 00005621fbf8b1f0
[ 142.217206] Call Trace:
[ 142.217209] [<ffffffff8142dd65>] dump_stack+0x67/0x92
[ 142.217211] [<ffffffff8107e4e6>] __warn+0xc6/0xe0
[ 142.217213] [<ffffffff8107e54a>] warn_slowpath_fmt+0x4a/0x50
[ 142.217214] [<ffffffff81559e3e>] drm_mm_takedown+0x2e/0x40
[ 142.217236] [<ffffffffa035c02a>] i915_gem_cleanup_stolen+0x1a/0x20 [i915]
[ 142.217246] [<ffffffffa034c581>] i915_ggtt_cleanup_hw+0x31/0xb0 [i915]
[ 142.217253] [<ffffffffa0310311>] i915_driver_cleanup_hw+0x31/0x40 [i915]
[ 142.217260] [<ffffffffa0312001>] i915_driver_unload+0x141/0x1a0 [i915]
[ 142.217268] [<ffffffffa031c2c4>] i915_pci_remove+0x14/0x20 [i915]
[ 142.217269] [<ffffffff8147d214>] pci_device_remove+0x34/0xb0
[ 142.217271] [<ffffffff8157b14c>] __device_release_driver+0x9c/0x150
[ 142.217272] [<ffffffff8157bcc6>] driver_detach+0xb6/0xc0
[ 142.217273] [<ffffffff8157abe3>] bus_remove_driver+0x53/0xd0
[ 142.217274] [<ffffffff8157c787>] driver_unregister+0x27/0x50
[ 142.217276] [<ffffffff8147c265>] pci_unregister_driver+0x25/0x70
[ 142.217287] [<ffffffffa03d764c>] i915_exit+0x1a/0x71 [i915]
[ 142.217289] [<ffffffff811136b3>] SyS_delete_module+0x193/0x1e0
[ 142.217291] [<ffffffff818174ae>] entry_SYSCALL_64_fastpath+0x1c/0xb1
[ 142.217292] ---[ end trace 6fd164859c154772 ]---
[ 142.217505] [drm:show_leaks] *ERROR* node [6b6b6b6b6b6b6b6b + 6b6b6b6b6b6b6b6b]: inserted at
[<ffffffff81559ff3>] save_stack.isra.1+0x53/0xa0
[<ffffffff8155a98d>] drm_mm_insert_node_in_range_generic+0x2ad/0x360
[<ffffffffa035bf23>] i915_gem_stolen_insert_node_in_range+0x93/0xe0 [i915]
[<ffffffffa035c855>] i915_gem_object_create_stolen+0x75/0xb0 [i915]
[<ffffffffa036a51a>] intel_engine_create_ring+0x9a/0x140 [i915]
[<ffffffffa036a921>] intel_init_ring_buffer+0xf1/0x440 [i915]
[<ffffffffa036be1b>] intel_init_render_ring_buffer+0xab/0x1b0 [i915]
[<ffffffffa0363d08>] intel_engines_init+0xc8/0x210 [i915]
[<ffffffffa0355d7c>] i915_gem_init+0xac/0xf0 [i915]
[<ffffffffa0311454>] i915_driver_load+0x9c4/0x1430 [i915]
[<ffffffffa031c2f8>] i915_pci_probe+0x28/0x40 [i915]
[<ffffffff8147d315>] pci_device_probe+0x85/0xf0
[<ffffffff8157b7ff>] driver_probe_device+0x21f/0x430
[<ffffffff8157baee>] __driver_attach+0xde/0xe0
In particular note that the node was being poisoned as we inspected the
list, a clear indication that the object is being freed as we make the
assertion.
v2: Don't loop, just assert that we do all the work required as that
will be better at detecting further errors.
Fixes: fbbd37b36fa5 ("drm/i915: Move object release to a freelist + worker")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20161101084843.3961-1-chris@chris-wilson.co.uk
diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c
index 6a99544..3b9bfd2 100644
--- a/drivers/gpu/drm/i915/i915_drv.c
+++ b/drivers/gpu/drm/i915/i915_drv.c
@@ -544,7 +544,7 @@ static void i915_gem_fini(struct drm_i915_private *dev_priv)
i915_gem_context_fini(&dev_priv->drm);
mutex_unlock(&dev_priv->drm.struct_mutex);
- synchronize_rcu();
+ rcu_barrier();
flush_work(&dev_priv->mm.free_work);
WARN_ON(!list_empty(&dev_priv->context_list));
diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index 1e5d2bf..b512745 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -4787,6 +4787,8 @@ void i915_gem_load_cleanup(struct drm_device *dev)
{
struct drm_i915_private *dev_priv = to_i915(dev);
+ WARN_ON(!llist_empty(&dev_priv->mm.free_list));
+
kmem_cache_destroy(dev_priv->requests);
kmem_cache_destroy(dev_priv->vmas);
kmem_cache_destroy(dev_priv->objects);