Re: PROBLEM: 2.6test11 kernel panic on "head -1 /proc/net/tcp"

[lkml-031128@amos.mailshell.com]

OGAWA Hirofumi wrote:
> lkml-031128@amos.mailshell.com writes:
> 
> 
>>After I added Ogawa's line, "head -1 /proc/net/tcp" stopped
>>freezing my machine but PPP failed to work.
>>
>>Also after adding Ogawa's line, PPP works fine (as it is now, as
>>I write this message) as long as I don't try "head -1 /proc/net/tcp"
>>after boot. If I'll try "head -1 /proc/net/tcp" now PPP will stop
>>working.
> 
> 
> Can you reproduce the fail of PPP? I couldn't reproduce it.
> What reason is the fail of PPP? (the "debug" option of pppd may be helpful)

Sorry. I just tried:

1. From a multi-user mode, after an uptime of 5 days (test11 with your
fix).
2. killed the ppp daemon (/etc/init.d/ppp stop). Made sure the ppp0
interface is down.
3. did "head -1 /proc/net/tcp" and "cat /proc/net/tcp". Passed fine.
4. re-startted ppp daemon.
5. System is fine. No kernel errors. PPP works flowlessly.

So I think my linking of PPP to the fix was wrong. Maybe the PPP failure
was unrelated to this.

> 
> Of course, the following message is easy reproducible. But it's
> debugging message, not the real problem. And probably it's unrelated
> to the fail of PPP.
> 
> 
>>>>>Badness in local_bh_enable at kernel/softirq.c:121
>>>>>Call Trace:
>>>>> [<c011df25>] local_bh_enable+0x85/0x90
>>>>> [<c02315e2>] ppp_async_push+0xa2/0x180
>>>>> [<c0230efd>] ppp_asynctty_wakeup+0x2d/0x60
>>>>> [<c0202638>] pty_unthrottle+0x58/0x60
>>>>> [<c01ff0fd>] check_unthrottle+0x3d/0x40
>>>>> [<c01ff1a3>] n_tty_flush_buffer+0x13/0x60
>>>>> [<c0202a47>] pty_flush_buffer+0x67/0x70
>>>>> [<c01fba41>] do_tty_hangup+0x3f1/0x460