linux-kernel - BUG: using smp_processor_id() in preemptible [00000000] during put_task

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Date:   Fri, 9 Dec 2016 15:28:20 +0100
From:   Michal Hocko <mhocko@...nel.org>
To:     Andy Lutomirski <luto@...capital.net>
Cc:     LKML <linux-kernel@...r.kernel.org>
Subject: BUG: using smp_processor_id() in preemptible [00000000] during
 put_task_stack

Hi Andy,
I am hitting the following
[  570.715345] BUG: using smp_processor_id() in preemptible [00000000] code: umount/6193
[  570.716880] caller is debug_smp_processor_id+0x17/0x19
[  570.717876] CPU: 2 PID: 6193 Comm: umount Tainted: G        W       4.9.0-rc8-nofstest4-next-20161209-00012-g0db618dfb2cf #1017
[  570.720162] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.9.3-20161025_171302-gandalf 04/01/2014
[  570.720862] Call Trace:
[  570.720862]  dump_stack+0x68/0x92
[  570.720862]  check_preemption_disabled+0xce/0xe0
[  570.720862]  debug_smp_processor_id+0x17/0x19
[  570.720862]  vfree_atomic+0x2f/0x65
[  570.720862]  put_task_stack+0xd0/0x13c
[  570.720862]  kthread_stop+0x108/0x29a
[  570.720862]  destroy_workqueue+0x167/0x1fb
[  570.720862]  ext4_put_super+0x44/0x2fa
[  570.720862]  generic_shutdown_super+0x6a/0xeb
[  570.720862]  kill_block_super+0x27/0x67
[  570.720862]  deactivate_locked_super+0x30/0x68
[  570.720862]  deactivate_super+0x3e/0x41
[  570.720862]  cleanup_mnt+0x58/0x76
[  570.720862]  __cleanup_mnt+0x12/0x14
[  570.720862]  task_work_run+0x77/0xa0
[  570.720862]  exit_to_usermode_loop+0x67/0x92
[  570.720862]  do_syscall_64+0x16c/0x197
[  570.720862]  entry_SYSCALL64_slow_path+0x25/0x25

it seems that vfre_atomic doesn't like it has preemption enabled in
__vfree_deferred which calls this_cpu_ptr. I haven't studied the code
deeply but I guess what we want is to put vfree_atomic into the atomic
context. Something like the following or at least disable preemption
around vfree_atomic. Or maybe I am completely missing what the above
warning tries to tell...

diff --git a/kernel/fork.c b/kernel/fork.c
index 8f63bf912195..1c69561eec55 100644
--- a/kernel/fork.c
+++ b/kernel/fork.c
@@ -227,9 +227,9 @@ static inline void free_thread_stack(struct task_struct *tsk)
 			local_irq_restore(flags);
 			return;
 		}
+		vfree_atomic(tsk->stack);
 		local_irq_restore(flags);
 
-		vfree_atomic(tsk->stack);
 		return;
 	}
 #endif
-- 
Michal Hocko
SUSE Labs