On Wed, Nov 4, 2015 at 5:55 PM, Cody P Schafer <d...@codyps.com> wrote: > Ideas as to what could cause this would be appreciated. > > This consistently is triggered shortly after boot (I presume due to > conmand calling fsync on a file). > > Note that I'm not quite running 4.2.5, but none of the changes I have > additionally applied are to btrfs or atomics. > > Let me know if there is a way for me to get you more info. > > It looks like the line is: > > mutex_lock(&inode->i_mutex); > atomic_inc(&root->log_batch); <<<< > full_sync = test_bit(BTRFS_INODE_NEEDS_FULL_SYNC, > &BTRFS_I(inode)->runtime_flags); > > > addr2line of the trapping instruction: > addr2line -e > ./work/beaglebone-poky-linux-gnueabi/linux-yocto-ikabit/4.2.5+gitAUTOINC+c29ac18888-r1/linux-beaglebone-standard-build/arch/arm/boot/vmlinux > c0217a34 -i > > /home/cody/obj/y/tmp/work-shared/beaglebone/kernel-source/arch/arm/include/asm/atomic.h:194 > > /home/cody/obj/y/tmp/work-shared/beaglebone/kernel-source/fs/btrfs/file.c:1886 > > fault log: > > [ 11.488382] Alignment trap: not handling instruction e1993f9f at > [<c0217a34>] > [ 11.560558] Unhandled fault: alignment exception (0x001) at 0x0000026d > [ 11.607301] pgd = dc090000 > [ 11.610166] [0000026d] *pgd=9c063831, *pte=00000000, *ppte=00000000 > [ 11.665548] Internal error: : 1 [#1] PREEMPT ARM > [ 11.670388] Modules linked in: omaplfb(O) bufferclass_ti(O) pvrsrvkm(O) > [ 11.677341] CPU: 0 PID: 248 Comm: connmand Tainted: G O > 4.2.5-yocto-standard #1 > [ 11.686172] Hardware name: Generic AM33XX (Flattened Device Tree) > [ 11.692551] task: dc00d100 ti: dc068000 task.ti: dc068000 > [ 11.698219] PC is at btrfs_sync_file+0x104/0x3f4 > [ 11.703051] LR is at btrfs_sync_file+0x100/0x3f4 > [ 11.707883] pc : [<c0217a38>] lr : [<c0217a34>] psr: 60060013 > [ 11.707883] sp : dc069e98 ip : dc069e98 fp : dc069f1c > [ 11.719897] r10: 00000000 r9 : 0000026d r8 : dd746b40 > [ 11.725363] r7 : dcef8dcc r6 : 00000000 r5 : dcef8d68 r4 : 00000001 > [ 11.732192] r3 : 7fffffff r2 : 00000000 r1 : 00000000 r0 : dcef8dcc > [ 11.739024] Flags: nZCv IRQs on FIQs on Mode SVC_32 ISA ARM Segment > user > [ 11.746491] Control: 10c5387d Table: 9c090019 DAC: 00000015 > [ 11.752502] Process connmand (pid: 248, stack limit = 0xdc068210) > [ 11.758878] Stack: (0xdc069e98 to 0xdc06a000) > [ 11.763436] 9e80: > ffffffff 7fffffff > [ 11.771998] 9ea0: dc069f38 c014582c 0000b30b 00000000 00000000 > 000037f6 000081a4 d5012f68 > [ 11.780559] 9ec0: 00000000 80000000 00000000 00000000 00000000 > 00000000 0000008d 00000000 > [ 11.789121] 9ee0: 00000400 00000000 00000001 20b04e34 563924e6 > dd746b40 dce2beb8 00000000 > [ 11.797683] 9f00: 00000000 00000000 dc068000 00000000 dc069f54 > dc069f20 c0171d00 c0217940 > [ 11.806245] 9f20: ffffffff 7fffffff 00000000 dc069f38 c0145b6c > dd746b40 00000000 dd746b40 > [ 11.814807] 9f40: 00000076 c000f7a4 dc069f74 dc069f58 c0171d3c > c0171c40 ffffffff 7fffffff > [ 11.823369] 9f60: 00000000 c015cc9c dc069f94 dc069f78 c0171d7c > c0171d14 00000000 bebe2bf8 > [ 11.831931] 9f80: 00e64c4d b6a104d0 dc069fa4 dc069f98 c017204c > c0171d50 00000000 dc069fa8 > [ 11.840492] 9fa0: c000f600 c017203c bebe2bf8 00e64c4d 00000009 > bebe2bf8 0000008d 00000000 > [ 11.849054] 9fc0: bebe2bf8 00e64c4d b6a104d0 00000076 00e64810 > 00e60cd0 00000009 00000000 > [ 11.857616] 9fe0: 00000000 bebe2bd4 b6e63384 b6dcef00 60060010 > 00000009 044a1103 099b0303 > [ 11.866199] [<c0217a38>] (btrfs_sync_file) from [<c0171d00>] > (vfs_fsync_range+0xcc/0xd4) > [ 11.874674] [<c0171d00>] (vfs_fsync_range) from [<c0171d3c>] > (vfs_fsync+0x34/0x3c) > [ 11.882600] [<c0171d3c>] (vfs_fsync) from [<c0171d7c>] (do_fsync+0x38/0x54) > [ 11.889890] [<c0171d7c>] (do_fsync) from [<c017204c>] (SyS_fsync+0x1c/0x20) > [ 11.897188] [<c017204c>] (SyS_fsync) from [<c000f600>] > (ret_fast_syscall+0x0/0x3c) > [ 11.905116] Code: e14b05fc e1a00007 eb1251f7 e1993f9f (e2833001) > [ 13.418969] ---[ end trace d7bcd93aea7d243c ]--- > [ 13.442420] Kernel panic - not syncing: Fatal exception
I looked into this a bit further, and it looks like my issue is that the btrfs_root* from BTRFS_I(inode)->root is somehow `1` instead of an actual pointer value, so it appears this isn't quite an alignment issue. Kernel starts without issue if I just stop doing the overlay (and as a result stop using btrfs at all). I've added some debugging, and the actual file it's trying to fsync appears to be: `/var/lib/connman/settings.4CAC7X` That directory is a overlayfs dir with a btrfs upper and a squashfs lower. Ideas on why the btrfs_root pointer is 0x1? -- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
