Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752713AbXLCVk4 (ORCPT ); Mon, 3 Dec 2007 16:40:56 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1751018AbXLCVks (ORCPT ); Mon, 3 Dec 2007 16:40:48 -0500 Received: from baron.miniblade.com ([65.14.39.133]:16431 "EHLO coraid.com" rhost-flags-OK-FAIL-OK-FAIL) by vger.kernel.org with ESMTP id S1751160AbXLCVkr (ORCPT ); Mon, 3 Dec 2007 16:40:47 -0500 Date: Mon, 3 Dec 2007 16:38:37 -0500 From: "Ed L. Cashin" To: Andrew Morton Cc: jnelson-kernel-bugzilla@jamponi.net, bugme-daemon@bugzilla.kernel.org, nickpiggin@yahoo.com.au, clameter@sgi.com, linux-kernel@vger.kernel.org, rjw@sisk.pl Subject: Re: [Bugme-new] [Bug 9482] New: kernel GPF in 2.6.24 (g09f345da) Message-ID: <20071203213837.GE27094@coraid.com> References: <20071201122302.10dd9c44.akpm@linux-foundation.org> <20071203162137.GB25251@coraid.com> <20071203113459.d36c1a01.akpm@linux-foundation.org> <20071203210005.GB27094@coraid.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20071203210005.GB27094@coraid.com> User-Agent: Mutt/1.5.16 (2007-06-11) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 5500 Lines: 122 On Mon, Dec 03, 2007 at 04:00:05PM -0500, Ed L. Cashin wrote: ... > I'll keep looking at this, but at a glance it looks like the cpu > number is valid, because I don't trip a BUG_ON when I make the change > below (the badval variable is noise, sorry). > > --- lx/lib/percpu_counter.c.20071130 2007-12-03 15:43:19.000000000 -0500 > +++ lx/lib/percpu_counter.c 2007-12-03 15:47:38.000000000 -0500 > @@ -33,7 +33,9 @@ void __percpu_counter_add(struct percpu_ > s64 count; > s32 *pcount; > int cpu = get_cpu(); > + u64 badval = 0xffffffffffffffffULL; > > + BUG_ON(!cpu_possible(cpu)); > pcount = per_cpu_ptr(fbc->counters, cpu); > count = *pcount + amount; > if (count >= batch || count <= -batch) { It appears that the fbc->counters pointer is NULL. I added the line, BUG_ON(!fbc->counters); ... (on line 39 in my percpu_counter.c), and it results in the trace below. It looks like when it's NULL, percpu_ptr passes it to __percpu_disguise, which makes it all ones and then tries to dereference 0xffffffffffffffff to access to the "ptrs" member of the struct percpu_data. ------------[ cut here ]------------ kernel BUG at lib/percpu_counter.c:39! invalid opcode: 0000 [1] SMP CPU 0 Modules linked in: aoe Pid: 3354, comm: bash Not tainted 2.6.24-rc3-47dbg #10 RIP: 0010:[] [] __percpu_counter_add+0x2a/0x8f RSP: 0018:ffff810075031aa8 EFLAGS: 00010046 RAX: 00000000ffffffff RBX: ffff81007fd19bd8 RCX: 0000000000000000 RDX: 0000000000000010 RSI: 0000000000000001 RDI: 0000000000000000 RBP: ffff810075031ac8 R08: ffff81007cc077b0 R09: ffffffff802ae5ee R10: ffff810075031aa8 R11: ffff8100750318e8 R12: ffff81007c81c380 R13: ffff810073ce8250 R14: 0000000000000200 R15: ffff8100755016b0 FS: 00002b3e5c052db0(0000) GS:ffffffff8078b000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b CR2: 00002b7f44fb64e0 CR3: 000000007c4b1000 CR4: 00000000000006e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 Process bash (pid: 3354, threadinfo ffff810075030000, task ffff81007b4da040) Stack: ffff810075031ac8 ffff81007fd19bd8 ffff81007c81c380 0000000000000000 ffff810075031af8 ffffffff802ae682 0000100075031ae8 ffff8100755016b0 0000000000000200 ffff81007fd19bd8 ffff810075031b18 ffffffff802ae75c Call Trace: [] __set_page_dirty+0xdc/0x121 [] mark_buffer_dirty+0x95/0x99 [] __block_commit_write+0x72/0xac [] block_write_end+0x4f/0x5b [] blkdev_write_end+0x1b/0x38 [] generic_file_buffered_write+0x1c0/0x648 [] current_fs_time+0x22/0x29 [] __generic_file_aio_write_nolock+0x358/0x3c2 [] filemap_fault+0x1c4/0x320 [] unlock_page+0x2d/0x31 [] generic_file_aio_write_nolock+0x3b/0x8d [] do_sync_write+0xe2/0x126 [] autoremove_wake_function+0x0/0x38 [] do_page_fault+0x3f8/0x7bb [] fd_install+0x5f/0x68 [] vfs_write+0xae/0x137 [] sys_write+0x47/0x70 [] system_call+0x7e/0x83 Code: 0f 0b eb fe 0f a3 3d 7e 08 4f 00 19 c0 85 c0 75 04 0f 0b eb RIP [] __percpu_counter_add+0x2a/0x8f RSP BUG: sleeping function called from invalid context at kernel/rwsem.c:20 in_atomic():0, irqs_disabled():1 INFO: lockdep is turned off. Call Trace: [] debug_show_held_locks+0x1b/0x24 [] __might_sleep+0xc7/0xc9 [] down_read+0x1d/0x4a [] exit_mm+0x34/0xf7 [] do_exit+0x247/0x75b [] kernel_math_error+0x0/0x7e [] do_trap+0x101/0x110 [] do_invalid_op+0x91/0x9a [] __percpu_counter_add+0x2a/0x8f [] :aoe:aoeblk_make_request+0x1c3/0x1d0 [] io_schedule+0x28/0x34 [] error_exit+0x0/0x9a [] __set_page_dirty+0x48/0x121 [] __percpu_counter_add+0x2a/0x8f [] __set_page_dirty+0xdc/0x121 [] mark_buffer_dirty+0x95/0x99 [] __block_commit_write+0x72/0xac [] block_write_end+0x4f/0x5b [] blkdev_write_end+0x1b/0x38 [] generic_file_buffered_write+0x1c0/0x648 [] current_fs_time+0x22/0x29 [] __generic_file_aio_write_nolock+0x358/0x3c2 [] filemap_fault+0x1c4/0x320 [] unlock_page+0x2d/0x31 [] generic_file_aio_write_nolock+0x3b/0x8d [] do_sync_write+0xe2/0x126 [] autoremove_wake_function+0x0/0x38 [] do_page_fault+0x3f8/0x7bb [] fd_install+0x5f/0x68 [] vfs_write+0xae/0x137 [] sys_write+0x47/0x70 [] system_call+0x7e/0x83 -- Ed L Cashin -- 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/