the lock validator i'm working on found another item:
============================
[ BUG: illegal lock usage! ]
----------------------------
illegal {enabled-softirqs} -> {used-in-softirq} usage.
hackbench/8407 [HC0[0]:SC1[2]:HE1:SE0] takes:
{&state[i].lock} [<c0bf7d08>] wrandom_flush+0x2a/0xb3
{enabled-softirqs} state was registered at:
[<c0befc33>] fn_hash_insert+0x565/0x5c3
hardirqs last enabled at: [<c0d29a36>] _spin_unlock_irq+0xd/0x10
softirqs last enabled at: [<c012d0a0>] irq_exit+0x36/0x38
other info that might help us debug this:
locks held by hackbench/8407: <none>
stack backtrace:
[<c0103e86>] show_trace+0xd/0xf
[<c0103e9d>] dump_stack+0x15/0x17
[<c013eb40>] print_usage_bug+0x16d/0x177
[<c013f130>] mark_lock+0xe3/0x248
[<c013f65d>] debug_lock_chain+0x3c8/0xb1d
[<c013fde3>] debug_lock_chain_spin+0x31/0x48
[<c0410c4e>] _raw_spin_lock+0x34/0x7f
[<c0d298fd>] _spin_lock+0x8/0xa
[<c0bf7d08>] wrandom_flush+0x2a/0xb3
[<c0bc3738>] rt_cache_flush+0x3f/0xd8
[<c0bc3843>] rt_secret_rebuild+0x11/0x26
[<c0131165>] run_timer_softirq+0x143/0x19e
[<c012d384>] __do_softirq+0x84/0xff
[<c0104d76>] do_softirq+0x52/0xbb
=======================
[<c012d0a0>] irq_exit+0x36/0x38
[<c0118809>] smp_apic_timer_interrupt+0x4e/0x51
[<c010376f>] apic_timer_interrupt+0x27/0x2c
[<c01670b2>] kmem_cache_alloc+0x28/0xa6
[<c0b80be6>] __alloc_skb+0x25/0xee
[<c0b7f144>] sock_alloc_send_skb+0x5d/0x192
[<c0c29bdf>] unix_stream_sendmsg+0x131/0x33f
[<c0b7e30a>] do_sock_write+0xbd/0xc6
[<c0b7e441>] sock_aio_write+0x56/0x64
[<c016a9a4>] do_sync_write+0xb1/0xe6
[<c016adf0>] vfs_write+0xbd/0x155
[<c016b657>] sys_write+0x3b/0x60
[<c0102c25>] syscall_call+0x7/0xb
the culprit seems to be:
(gdb) list *0xc0befc33
0xc0befc33 is in fn_hash_insert (net/ipv4/fib_hash.c:527).
522 list_add_tail(&new_fa->fa_list,
523 (fa ? &fa->fa_list : &f->fn_alias));
524 fib_hash_genid++;
525 write_unlock_bh(&fib_hash_lock);
526
527 if (new_f)
528 fz->fz_nent++;
529 rt_cache_flush(-1);
530
531 rtmsg_fib(RTM_NEWROUTE, key, new_fa, z, tb->tb_id, n, req);
(gdb)
which enabled softirqs with {&state[i].lock} still held. If the
rt_secret_rebuild() would hit that codepath at that point then it could
cause a deadlock, right?
Ingo
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
[Index of Archives]
[Kernel Newbies]
[Netfilter]
[Bugtraq]
[Photo]
[Stuff]
[Gimp]
[Yosemite News]
[MIPS Linux]
[ARM Linux]
[Linux Security]
[Linux RAID]
[Video 4 Linux]
[Linux for the blind]
[Linux Resources]
