Re: Lockdep show 6.6-rc regression in Xen HVM CPU hotplug

[Juergen Gross <jgross@xxxxxxxx>]

On 23.10.23 22:50, David Woodhouse wrote:
> Since commit 87797fad6cce ("xen/events: replace evtchn_rwlock with
> RCU"), I can no longer hotplug CPUs under Xen with lockdep enabled.
>
> (This is real Xen 4.17.1; annoyingly I have different failure modes
> with Xen guests under qemu/kvm and I'll deal with those next.)
>
> Offlining complains thus:
>
> [root@localhost cpu1]# echo 0 > online
> [   52.936265]
> [   52.936271] =============================
> [   52.936274] WARNING: suspicious RCU usage
> [   52.936277] 6.6.0-rc5+ #1357 Not tainted
> [   52.936280] -----------------------------
> [   52.936282] lib/maple_tree.c:856 suspicious rcu_dereference_check() usage!
> [   52.936287]
> [   52.936287] other info that might help us debug this:
> [   52.936287]
> [   52.936289]
> [   52.936289] RCU used illegally from offline CPU!
> [   52.936289] rcu_scheduler_active = 2, debug_locks = 1
> [   52.936293] 1 lock held by swapper/1/0:
> [   52.936296]  #0: ffffffff89c03820 (rcu_read_lock){....}-{1:3}, at: 
> mtree_load+0x90/0x590
> [   52.936321]
> [   52.936321] stack backtrace:
> [   52.936324] CPU: 1 PID: 0 Comm: swapper/1 Not tainted 6.6.0-rc5+ #1357
> [   52.936329] Hardware name: Xen HVM domU, BIOS 4.17.1 09/26/2023
> [   52.936332] Call Trace:
> [   52.936335]  <TASK>
> [   52.936340]  dump_stack_lvl+0x5b/0x90
> [   52.936350]  lockdep_rcu_suspicious+0x15a/0x1c0
> [   52.936366]  mtree_load+0x49e/0x590
> [   52.936385]  irq_get_irq_data+0xe/0x20
> [   52.936394]  xen_send_IPI_one+0xa4/0x100
> [   52.936404]  __xen_send_IPI_mask+0x1b/0x50
> [   52.936414]  generic_exec_single+0x35/0x1c0
> [   52.936423]  smp_call_function_single+0xc2/0x140
> [   52.936436]  ? cpuhp_report_idle_dead+0x42/0x60
> [   52.936444]  do_idle+0xda/0xe0
> [   52.936451]  cpu_startup_entry+0x2a/0x30
> [   52.936456]  start_secondary+0x123/0x140
> [   52.936465]  secondary_startup_64_no_verify+0x178/0x17b
> [   52.936490]  </TASK>
I'm puzzled. This path doesn't contain any of the RCU usage I've added in
commit 87797fad6cce.

Are you sure that with just reverting commit 87797fad6cce the issue doesn't
manifest anymore? I'd rather expect commit 721255b9826b having caused this
behavior, just telling from the messages above.


Juergen

Attachment: OpenPGP_0xB0DE9DD628BF132F.asc
Description: OpenPGP public key

Attachment: OpenPGP_signature.asc
Description: OpenPGP digital signature