Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1422706AbWJPQMX (ORCPT ); Mon, 16 Oct 2006 12:12:23 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1422710AbWJPQMX (ORCPT ); Mon, 16 Oct 2006 12:12:23 -0400 Received: from alpha.polcom.net ([83.143.162.52]:61382 "EHLO alpha.polcom.net") by vger.kernel.org with ESMTP id S1422706AbWJPQMX (ORCPT ); Mon, 16 Oct 2006 12:12:23 -0400 Date: Mon, 16 Oct 2006 18:12:15 +0200 (CEST) From: Grzegorz Kulewski To: linux-kernel@vger.kernel.org Cc: shaggy@austin.ibm.com, dm-devel@redhat.com, dm-crypt@saout.de, mingo@elte.hu, akpm@osdl.org Subject: Strange SIGSEGV problem around dmcrypt, evms and jfs Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII; format=flowed Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 5939 Lines: 100 Hi, I was begining to play with dmcrypt, evms and jfs on one spare disk I have (currently empty and only for tests). I produced some partitions with evms and made volumes on them. Nothing strange, normal configuration. The partition layout seems ok. Then I used dmcrypt mappings on top of two of them to make encrypted swaps and swapon'ed them. Still everything was ok. Then I tested different ciphers performance by doing dmcrypt mappings on top of some other volume with different settings and dd'ed data from and to them to test the speed. Then I choosen one cipher setup and and did the final mapping and created and mounted jfs on it. Then I copied one large (like 4GB) file on it several times to make sure everything is ok. I checked sha1sums and everything was indeed ok. But then all big applications (firefox, oo2, acroread, ..., opera was the notable exception) couldn't start being killed by SIGSEGVs out of nowhere. I reproduced it two time already (after a clean reboot): today and yesterday. Maybe someone knows what is happening? For me it looks like something broken some kernel memory and the kernel started doing stupid things. But nothing strange has shown in logs. One time I couldn't even shut down the machine normally, only SysRQ-B worked (shutdown scripts were probably killed too or something). Every application works ok (and did so for at least a year) before I will start playing with dmcrypt and jfs. I am not sure where exactly the problems start but will be investigating it shortly. I am rather sure that my hardware is ok. Everything was and is fine till I will start doing these tests. Including that testing disk (tested with smart and dd and some others). My setup is: - Athlon (Barton) XP 2000MHz - Abit KW7 KT880 board - 1GB DDR 133 - main disk is 80GB Samsung @ IDE (VIA southbridge) - testing disk is 250GB Seagate @ SATA (VIA southbridge). This behavoiur was observed on 2.6.18-ck1 + vesafb-tng patch. Kernel was tainted by nvidia and kqemu modules. Now I am trying to recreate this problem with 2.6.18.1 with nearly all kernel debuging options turned on and without any proprietary modules loaded. But since I don't know exactly how to reproduce the problem it may take some time so any suggestions what can be wrong are welcome. Further info available on request. BTW. Why booting my machine with 2.6.18.1 with nearly all debuging on I got the following. While I am nearly sure it is not the problem I am writing about I will report it: Oct 16 17:29:33 kangur [ 74.485627] ============================================= Oct 16 17:29:33 kangur [ 74.485767] [ INFO: possible recursive locking detected ] Oct 16 17:29:33 kangur [ 74.485840] --------------------------------------------- Oct 16 17:29:33 kangur [ 74.485912] evms_activate/2346 is trying to acquire lock: Oct 16 17:29:33 kangur [ 74.485985] (&md->io_lock){----}, at: [] dm_request+0x18/0x150 [dm_mod] Oct 16 17:29:33 kangur [ 74.486269] Oct 16 17:29:33 kangur [ 74.486270] but task is already holding lock: Oct 16 17:29:33 kangur [ 74.486406] (&md->io_lock){----}, at: [] dm_request+0x18/0x150 [dm_mod] Oct 16 17:29:33 kangur [ 74.486673] Oct 16 17:29:33 kangur [ 74.486674] other info that might help us debug this: Oct 16 17:29:33 kangur [ 74.486813] 1 lock held by evms_activate/2346: Oct 16 17:29:33 kangur [ 74.486883] #0: (&md->io_lock){----}, at: [] dm_request+0x18/0x150 [dm_mod] Oct 16 17:29:33 kangur [ 74.487191] Oct 16 17:29:33 kangur [ 74.487192] stack backtrace: Oct 16 17:29:33 kangur [ 74.487475] [] show_trace_log_lvl+0x18d/0x1b0 Oct 16 17:29:33 kangur [ 74.487606] [] show_trace+0x12/0x20 Oct 16 17:29:33 kangur [ 74.487728] [] dump_stack+0x19/0x20 Oct 16 17:29:33 kangur [ 74.487851] [] __lock_acquire+0x813/0xd80 Oct 16 17:29:33 kangur [ 74.488044] [] lock_acquire+0x75/0xa0 Oct 16 17:29:33 kangur [ 74.488230] [] down_read+0x3a/0x50 Oct 16 17:29:33 kangur [ 74.488413] [] dm_request+0x18/0x150 [dm_mod] Oct 16 17:29:33 kangur [ 74.488543] [] generic_make_request+0x147/0x1c0 Oct 16 17:29:33 kangur [ 74.489020] [] __map_bio+0x4d/0xa0 [dm_mod] Oct 16 17:29:33 kangur [ 74.489141] [] __split_bio+0x34a/0x380 [dm_mod] Oct 16 17:29:33 kangur [ 74.489263] [] dm_request+0xd4/0x150 [dm_mod] Oct 16 17:29:33 kangur [ 74.489384] [] generic_make_request+0x147/0x1c0 Oct 16 17:29:33 kangur [ 74.489745] [] submit_bio+0x72/0x120 Oct 16 17:29:33 kangur [ 74.490112] [] submit_bh+0xca/0x120 Oct 16 17:29:33 kangur [ 74.490358] [] block_read_full_page+0x258/0x2d0 Oct 16 17:29:33 kangur [ 74.490602] [] blkdev_readpage+0xf/0x20 Oct 16 17:29:33 kangur [ 74.490851] [] __do_page_cache_readahead+0x1b0/0x260 Oct 16 17:29:33 kangur [ 74.491071] [] blockable_page_cache_readahead+0x62/0xe0 Oct 16 17:29:33 kangur [ 74.491288] [] page_cache_readahead+0x11d/0x1d0 Oct 16 17:29:33 kangur [ 74.491504] [] do_generic_mapping_read+0x462/0x4e0 Oct 16 17:29:33 kangur [ 74.491718] [] __generic_file_aio_read+0xee/0x220 Oct 16 17:29:33 kangur [ 74.491929] [] generic_file_read+0x8d/0xb0 Oct 16 17:29:33 kangur [ 74.492141] [] vfs_read+0xad/0x180 Oct 16 17:29:33 kangur [ 74.492377] [] sys_read+0x3d/0x70 Oct 16 17:29:33 kangur [ 74.492616] [] sysenter_past_esp+0x56/0x8d Oct 16 17:29:33 kangur [ 74.492736] [] 0xb7f19410 Thanks in advance, Grzegorz Kulewski - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/