Home » Mailing list archives » linux-netdev » 2008 » October » 17

2.6.27-05178-g2e532d6: list_add corruption

Previous thread: [PATCH] igb: add IGB_DCA instead of selecting INTEL_IOATDMA by Jeff Kirsher on Friday, October 17, 2008 - 11:08 am. (2 messages)

Next thread: Re: 2.6.27 seems to break something with DSL (fwd) by Olon on Friday, October 17, 2008 - 1:11 pm. (4 messages)
[view original message]
From: Alexander Beregalov
Subject: 2.6.27-05178-g2e532d6: list_add corruption
Date: Friday, October 17, 2008 - 12:20 pm

I am afraid I do not know how to reproduce this.

WARNING: at lib/list_debug.c:30 __list_add+0x44/0x5c()
list_add corruption. prev->next should be next (f6638448), but was
f26f078c. (prev=f26f078c).
Modules linked in: w83627hf hwmon_vid i2c_nforce2
Pid: 1933, comm: rtorrent Not tainted 2.6.27-05178-g2e532d6 #6
Call Trace:
 [<c011de4b>] warn_slowpath+0x4b/0x69
 [<c0137a03>] ? trace_hardirqs_off+0xb/0xd
 [<c01070ab>] ? native_sched_clock+0x76/0x88
 [<c0138cc5>] ? trace_hardirqs_on+0xb/0xd
 [<c0138c93>] ? trace_hardirqs_on_caller+0xe6/0x10d
 [<c031aa43>] ? release_sock+0xac/0xb4
 [<c0138cc5>] ? trace_hardirqs_on+0xb/0xd
 [<c0139d4e>] ? __lock_acquire+0xaeb/0xafa
 [<c0139d4e>] ? __lock_acquire+0xaeb/0xafa
 [<c0137a39>] ? put_lock_stats+0xd/0x21
 [<c026a404>] __list_add+0x44/0x5c
 [<c0186be0>] ep_poll_callback+0x5b/0xb4
 [<c011735b>] __wake_up_common+0x35/0x5b
 [<c0117c6d>] __wake_up_sync+0x31/0x44
 [<c031ae39>] sock_def_readable+0x38/0x63
 [<c03467eb>] tcp_rcv_established+0x37c/0x49d
 [<c034c2cc>] tcp_v4_do_rcv+0x25/0x151
 [<c0386269>] ? _spin_lock_nested+0x51/0x59
 [<c034c6e9>] tcp_v4_rcv+0x2f1/0x4a2
 [<c0336a65>] ip_local_deliver+0x86/0xd6
 [<c03369c1>] ip_rcv+0x361/0x37f
 [<c0322ad7>] netif_receive_skb+0x1af/0x1cf
 [<c02e5892>] nv_napi_poll+0x201/0x4f8
 [<c0138cc5>] ? trace_hardirqs_on+0xb/0xd
 [<c03217f0>] net_rx_action+0x98/0x155
 [<c0121d94>] __do_softirq+0x6a/0xf1
 [<c0121e4a>] do_softirq+0x2f/0x47
 [<c0121fb9>] irq_exit+0x3b/0x84
 [<c01046fb>] do_IRQ+0x6d/0x82
 [<c01038d0>] common_interrupt+0x28/0x30
---[ end trace 71f919bde0b515ac ]---
------------[ cut here ]------------
WARNING: at lib/list_debug.c:30 __list_add+0x44/0x5c()
list_add corruption. prev->next should be next (f6638448), but was
f26f078c. (prev=f26f078c).
Modules linked in: w83627hf hwmon_vid i2c_nforce2
Pid: 1933, comm: rtorrent Tainted: G        W 2.6.27-05178-g2e532d6 #6
Call Trace:
 [<c011de4b>] warn_slowpath+0x4b/0x69
 [<c0160063>] ? shmem_writepage+0x147/0x1ad
 [<c0139de8>] ? ...
[ message continues ]
[view original message]
From: Davide Libenzi
Subject: Re: 2.6.27-05178-g2e532d6: list_add corruption
Date: Friday, October 17, 2008 - 4:19 pm

This is very likely to be caused by a patch that went in a few days ago. 
Already posted a fix to Andrew.



- Davide


--


[ message continues ]
[view original message]
From: Alexander Beregalov
Subject: Re: 2.6.27-05178-g2e532d6: list_add corruption
Date: Friday, October 17, 2008 - 11:41 pm
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 ...
[ message continues ]
[view original message]
From: Davide Libenzi
Subject: Re: 2.6.27-05178-g2e532d6: list_add corruption
Date: Saturday, October 18, 2008 - 7:38 am

Hard to be sure, but certainly a double insert can create a similar 
problem. The only way to tell is to get the fix and retry.


- Davide


--


[ message continues ]
[view original message]
From: Alexander Beregalov
Subject: Re: 2.6.27-05178-g2e532d6: list_add corruption
Date: Saturday, October 18, 2008 - 8:50 am

I have already applied your patch and the kernel works better, no
errors anymore.
Thanks
--


[ message continues ]
Previous thread: [PATCH] igb: add IGB_DCA instead of selecting INTEL_IOATDMA by Jeff Kirsher on Friday, October 17, 2008 - 11:08 am. (2 messages)

Next thread: Re: 2.6.27 seems to break something with DSL (fwd) by Olon on Friday, October 17, 2008 - 1:11 pm. (4 messages)