[net] net/x25: do not hold the cpu too long in x25_new_lci()

Message ID 20190208204105.16314-1-edumazet@google.com
State Accepted
Delegated to: David Miller
Headers show
Series
  • [net] net/x25: do not hold the cpu too long in x25_new_lci()
Related show

Commit Message

Eric Dumazet Feb. 8, 2019, 8:41 p.m.
Due to quadratic behavior of x25_new_lci(), syzbot was able
to trigger an rcu stall.

Fix this by not blocking BH for the whole duration of
the function, and inserting a reschedule point when possible.

If we care enough, using a bitmap could get rid of the quadratic
behavior.

syzbot report :

rcu: INFO: rcu_preempt self-detected stall on CPU
rcu:    0-...!: (10500 ticks this GP) idle=4fa/1/0x4000000000000002 softirq=283376/283376 fqs=0
rcu:     (t=10501 jiffies g=383105 q=136)
rcu: rcu_preempt kthread starved for 10502 jiffies! g383105 f0x0 RCU_GP_WAIT_FQS(5) ->state=0x402 ->cpu=0
rcu: RCU grace-period kthread stack dump:
rcu_preempt     I28928    10      2 0x80000000
Call Trace:
 context_switch kernel/sched/core.c:2844 [inline]
 __schedule+0x817/0x1cc0 kernel/sched/core.c:3485
 schedule+0x92/0x180 kernel/sched/core.c:3529
 schedule_timeout+0x4db/0xfd0 kernel/time/timer.c:1803
 rcu_gp_fqs_loop kernel/rcu/tree.c:1948 [inline]
 rcu_gp_kthread+0x956/0x17a0 kernel/rcu/tree.c:2105
 kthread+0x357/0x430 kernel/kthread.c:246
 ret_from_fork+0x3a/0x50 arch/x86/entry/entry_64.S:352
NMI backtrace for cpu 0
CPU: 0 PID: 8759 Comm: syz-executor2 Not tainted 5.0.0-rc4+ #51
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Call Trace:
 <IRQ>
 __dump_stack lib/dump_stack.c:77 [inline]
 dump_stack+0x172/0x1f0 lib/dump_stack.c:113
 nmi_cpu_backtrace.cold+0x63/0xa4 lib/nmi_backtrace.c:101
 nmi_trigger_cpumask_backtrace+0x1be/0x236 lib/nmi_backtrace.c:62
 arch_trigger_cpumask_backtrace+0x14/0x20 arch/x86/kernel/apic/hw_nmi.c:38
 trigger_single_cpu_backtrace include/linux/nmi.h:164 [inline]
 rcu_dump_cpu_stacks+0x183/0x1cf kernel/rcu/tree.c:1211
 print_cpu_stall kernel/rcu/tree.c:1348 [inline]
 check_cpu_stall kernel/rcu/tree.c:1422 [inline]
 rcu_pending kernel/rcu/tree.c:3018 [inline]
 rcu_check_callbacks.cold+0x500/0xa4a kernel/rcu/tree.c:2521
 update_process_times+0x32/0x80 kernel/time/timer.c:1635
 tick_sched_handle+0xa2/0x190 kernel/time/tick-sched.c:161
 tick_sched_timer+0x47/0x130 kernel/time/tick-sched.c:1271
 __run_hrtimer kernel/time/hrtimer.c:1389 [inline]
 __hrtimer_run_queues+0x33e/0xde0 kernel/time/hrtimer.c:1451
 hrtimer_interrupt+0x314/0x770 kernel/time/hrtimer.c:1509
 local_apic_timer_interrupt arch/x86/kernel/apic/apic.c:1035 [inline]
 smp_apic_timer_interrupt+0x120/0x570 arch/x86/kernel/apic/apic.c:1060
 apic_timer_interrupt+0xf/0x20 arch/x86/entry/entry_64.S:807
 </IRQ>
RIP: 0010:__read_once_size include/linux/compiler.h:193 [inline]
RIP: 0010:queued_write_lock_slowpath+0x13e/0x290 kernel/locking/qrwlock.c:86
Code: 00 00 fc ff df 4c 8d 2c 01 41 83 c7 03 41 0f b6 45 00 41 38 c7 7c 08 84 c0 0f 85 0c 01 00 00 8b 03 3d 00 01 00 00 74 1a f3 90 <41> 0f b6 55 00 41 38 d7 7c eb 84 d2 74 e7 48 89 df e8 6c 0f 4f 00
RSP: 0018:ffff88805f117bd8 EFLAGS: 00000206 ORIG_RAX: ffffffffffffff13
RAX: 0000000000000300 RBX: ffffffff89413ba0 RCX: 1ffffffff1282774
RDX: 0000000000000000 RSI: 0000000000000004 RDI: ffffffff89413ba0
RBP: ffff88805f117c70 R08: 1ffffffff1282774 R09: fffffbfff1282775
R10: fffffbfff1282774 R11: ffffffff89413ba3 R12: 00000000000000ff
R13: fffffbfff1282774 R14: 1ffff1100be22f7d R15: 0000000000000003
 queued_write_lock include/asm-generic/qrwlock.h:104 [inline]
 do_raw_write_lock+0x1d6/0x290 kernel/locking/spinlock_debug.c:203
 __raw_write_lock_bh include/linux/rwlock_api_smp.h:204 [inline]
 _raw_write_lock_bh+0x3b/0x50 kernel/locking/spinlock.c:312
 x25_insert_socket+0x21/0xe0 net/x25/af_x25.c:267
 x25_bind+0x273/0x340 net/x25/af_x25.c:705
 __sys_bind+0x23f/0x290 net/socket.c:1505
 __do_sys_bind net/socket.c:1516 [inline]
 __se_sys_bind net/socket.c:1514 [inline]
 __x64_sys_bind+0x73/0xb0 net/socket.c:1514
 do_syscall_64+0x103/0x610 arch/x86/entry/common.c:290
 entry_SYSCALL_64_after_hwframe+0x49/0xbe
RIP: 0033:0x457e39
Code: ad b8 fb ff c3 66 2e 0f 1f 84 00 00 00 00 00 66 90 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 0f 83 7b b8 fb ff c3 66 2e 0f 1f 84 00 00 00 00
RSP: 002b:00007fafccd0dc78 EFLAGS: 00000246 ORIG_RAX: 0000000000000031
RAX: ffffffffffffffda RBX: 0000000000000003 RCX: 0000000000457e39
RDX: 0000000000000012 RSI: 0000000020000240 RDI: 0000000000000004
RBP: 000000000073bf00 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000246 R12: 00007fafccd0e6d4
R13: 00000000004bdf8b R14: 00000000004ce4b8 R15: 00000000ffffffff
Sending NMI from CPU 0 to CPUs 1:
NMI backtrace for cpu 1
CPU: 1 PID: 8752 Comm: syz-executor4 Not tainted 5.0.0-rc4+ #51
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
RIP: 0010:__x25_find_socket+0x78/0x120 net/x25/af_x25.c:328
Code: 89 f8 48 c1 e8 03 80 3c 18 00 0f 85 a6 00 00 00 4d 8b 64 24 68 4d 85 e4 74 7f e8 03 97 3d fb 49 83 ec 68 74 74 e8 f8 96 3d fb <49> 8d bc 24 88 04 00 00 48 89 f8 48 c1 e8 03 0f b6 04 18 84 c0 74
RSP: 0018:ffff8880639efc58 EFLAGS: 00000246
RAX: 0000000000040000 RBX: dffffc0000000000 RCX: ffffc9000e677000
RDX: 0000000000040000 RSI: ffffffff863244b8 RDI: ffff88806a764628
RBP: ffff8880639efc80 R08: ffff8880a80d05c0 R09: fffffbfff1282775
R10: fffffbfff1282774 R11: ffffffff89413ba3 R12: ffff88806a7645c0
R13: 0000000000000001 R14: ffff88809f29ac00 R15: 0000000000000000
FS:  00007fe8d0c58700(0000) GS:ffff8880ae900000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 0000001b32823000 CR3: 00000000672eb000 CR4: 00000000001406e0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Call Trace:
 x25_new_lci net/x25/af_x25.c:357 [inline]
 x25_connect+0x374/0xdf0 net/x25/af_x25.c:786
 __sys_connect+0x266/0x330 net/socket.c:1686
 __do_sys_connect net/socket.c:1697 [inline]
 __se_sys_connect net/socket.c:1694 [inline]
 __x64_sys_connect+0x73/0xb0 net/socket.c:1694
 do_syscall_64+0x103/0x610 arch/x86/entry/common.c:290
 entry_SYSCALL_64_after_hwframe+0x49/0xbe
RIP: 0033:0x457e39
Code: ad b8 fb ff c3 66 2e 0f 1f 84 00 00 00 00 00 66 90 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 0f 83 7b b8 fb ff c3 66 2e 0f 1f 84 00 00 00 00
RSP: 002b:00007fe8d0c57c78 EFLAGS: 00000246 ORIG_RAX: 000000000000002a
RAX: ffffffffffffffda RBX: 0000000000000003 RCX: 0000000000457e39
RDX: 0000000000000012 RSI: 0000000020000200 RDI: 0000000000000004
RBP: 000000000073bf00 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000246 R12: 00007fe8d0c586d4
R13: 00000000004be378 R14: 00000000004ceb00 R15: 00000000ffffffff

Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: syzbot <syzkaller@googlegroups.com>
Cc: Andrew Hendry <andrew.hendry@gmail.com>
Cc: linux-x25@vger.kernel.org
---
 net/x25/af_x25.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

Comments

David Miller Feb. 11, 2019, 9:20 p.m. | #1
From: Eric Dumazet <edumazet@google.com>
Date: Fri,  8 Feb 2019 12:41:05 -0800

> Due to quadratic behavior of x25_new_lci(), syzbot was able
> to trigger an rcu stall.
> 
> Fix this by not blocking BH for the whole duration of
> the function, and inserting a reschedule point when possible.
> 
> If we care enough, using a bitmap could get rid of the quadratic
> behavior.
> 
> syzbot report :
 ...
> Signed-off-by: Eric Dumazet <edumazet@google.com>
> Reported-by: syzbot <syzkaller@googlegroups.com>

Applied, thanks.

Patch

diff --git a/net/x25/af_x25.c b/net/x25/af_x25.c
index 5121729b8b631f45d81d3e0332b0d2e4346d3c5d..ec3a828672ef53991146bf94a2266932a5f25872 100644
--- a/net/x25/af_x25.c
+++ b/net/x25/af_x25.c
@@ -352,17 +352,15 @@  static unsigned int x25_new_lci(struct x25_neigh *nb)
 	unsigned int lci = 1;
 	struct sock *sk;
 
-	read_lock_bh(&x25_list_lock);
-
-	while ((sk = __x25_find_socket(lci, nb)) != NULL) {
+	while ((sk = x25_find_socket(lci, nb)) != NULL) {
 		sock_put(sk);
 		if (++lci == 4096) {
 			lci = 0;
 			break;
 		}
+		cond_resched();
 	}
 
-	read_unlock_bh(&x25_list_lock);
 	return lci;
 }