Re: WARNING: at net/ipv4/tcp_input.c:2927 tcp_ack+0xd55/0x1991()

[Jonathan Corbet]

On Sat, 28 Mar 2009 10:55:14 +0100
Markus Trippelsdorf <markus@xxxxxxxxxxxxxxx> wrote:

> Yes, you might be right, because running with CONFIG_PROVE_LOCKING and 
> CONFIG_DETECT_SOFTLOCKUP enabled points to a possible bug in the BKL
> removal patches (fasync) by Jonathan Corbet. (I wasn't able so far to
> reproduce the original WARNING.)
> 
> Here is one example:
> 
> =========================================================
> [ INFO: possible irq lock inversion dependency detected ]
> 2.6.29-03321-gbe0ea69 #7
> ---------------------------------------------------------
> swapper/0 just changed the state of lock:
>  (fasync_lock){..+.}, at: [<ffffffff8028a2ac>] kill_fasync+0x24/0x45
> but this lock took another, hard-irq-unsafe lock in the past:
>  (&f->f_lock){--..}

That's not a bug; f_lock will never be taken in IRQ mode.  There's a
fix for the warning in linux-next now; my plan is to get it upstream
before -rc1.

Thanks,

jon

Jonathan Corbet / LWN.net / corbet@xxxxxxx
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/