Re: [PATCH v2 6/7] mm, slab: call kvfree_rcu_barrier() from kmem_cache_destroy()

[Keith Busch <kbusch@kernel.org>]

On Tue, Feb 25, 2025 at 10:57:38AM +0100, Vlastimil Babka wrote:
> I tried to create a kunit test for it, but it doesn't trigger anything. Maybe
> it's too simple, or racy, and thus we are not flushing any of the queues from
> kvfree_rcu_barrier()?

Thanks, your test readily triggers it for me, but only if I load
rcutorture at the same time.

[  142.223220] CPU: 11 UID: 0 PID: 186 Comm: kworker/u64:11 Tainted: G            E    N 6.13.0-04839-g5e7b40f0ddce-dirty #831
[  142.223222] Tainted: [E]=UNSIGNED_MODULE, [N]=TEST
[  142.223223] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS rel-1.16.3-0-ga6ed6b701f0a-prebuilt.qemu.org 04/01/2014
[  142.223224] Workqueue: test_kfree_rcu_destroy_wq cache_destroy_workfn [slub_kunit]
[  142.223230] Call Trace:
[  142.223231]  <TASK>
[  142.223233]  dump_stack_lvl+0x64/0x90
[  142.223239]  print_circular_bug+0x2c5/0x400
[  142.223243]  check_noncircular+0x103/0x120
[  142.223246]  ? save_trace+0x3e/0x360
[  142.223249]  __lock_acquire+0x1481/0x24b0
[  142.223252]  lock_acquire+0xaa/0x2a0
[  142.223253]  ? console_lock_spinning_enable+0x3e/0x60
[  142.223255]  ? lock_release+0xb3/0x250
[  142.223257]  console_lock_spinning_enable+0x5a/0x60
[  142.223258]  ? console_lock_spinning_enable+0x3e/0x60
[  142.223260]  console_flush_all+0x2b1/0x490
[  142.223262]  ? console_flush_all+0x29/0x490
[  142.223264]  console_unlock+0x49/0xf0
[  142.223266]  vprintk_emit+0x12b/0x300
[  142.223269]  ? kfree_rcu_shrink_scan+0x120/0x120
[  142.223270]  _printk+0x48/0x50
[  142.223272]  ? kfree_rcu_shrink_scan+0x120/0x120
[  142.223273]  __warn_printk+0xb4/0xe0
[  142.223276]  ? 0xffffffffa05d6000
[  142.223278]  ? kfree_rcu_shrink_scan+0x120/0x120
[  142.223279]  check_flush_dependency.part.0+0xad/0x100
[  142.223282]  __flush_work+0x38a/0x4a0
[  142.223284]  ? find_held_lock+0x2b/0x80
[  142.223287]  ? flush_rcu_work+0x26/0x40
[  142.223289]  ? lock_release+0xb3/0x250
[  142.223290]  ? __mutex_unlock_slowpath+0x2c/0x270
[  142.223292]  flush_rcu_work+0x30/0x40
[  142.223294]  kvfree_rcu_barrier+0xe9/0x130
[  142.223296]  kmem_cache_destroy+0x2b/0x1f0
[  142.223297]  cache_destroy_workfn+0x20/0x40 [slub_kunit]
[  142.223299]  process_one_work+0x1cd/0x560
[  142.223302]  worker_thread+0x183/0x310
[  142.223304]  ? rescuer_thread+0x330/0x330
[  142.223306]  kthread+0xd8/0x1d0
[  142.223308]  ? ret_from_fork+0x17/0x50
[  142.223310]  ? lock_release+0xb3/0x250
[  142.223311]  ? kthreads_online_cpu+0xf0/0xf0
[  142.223313]  ret_from_fork+0x2d/0x50
[  142.223315]  ? kthreads_online_cpu+0xf0/0xf0
[  142.223317]  ret_from_fork_asm+0x11/0x20
[  142.223321]  </TASK>
[  142.371052] workqueue: WQ_MEM_RECLAIM test_kfree_rcu_destroy_wq:cache_destroy_workfn [slub_kunit] is flushing !WQ_MEM_RECLAIM events_unbound:kfree_rcu_work
[  142.371072] WARNING: CPU: 11 PID: 186 at kernel/workqueue.c:3715 check_flush_dependency.part.0+0xad/0x100
[  142.375748] Modules linked in: slub_kunit(E) rcutorture(E) torture(E) kunit(E) iTCO_wdt(E) iTCO_vendor_support(E) intel_uncore_frequency_common(E) skx_edac_common(E) nfit(E) libnvdimm(E) kvm_intel(E) kvm(E) evdev(E) bochs(E) serio_raw(E) drm_kms_helper(E) i2c_i801(E) e1000e(E) i2c_smbus(E) intel_agp(E) intel_gtt(E) lpc_ich(E) agpgart(E) mfd_core(E) drm_shm]