Many warnings on 2.6.29 from your sync_single_inode warning
Quoted 5 lines · Stephen Hemminger
The
WARN_ONyou introduced in
fs: new inodei_statecorruption fixTriggers every time during our regression test. The regression test runs on a VM under VMWare. Any ideas?
Thanks. Adding linux-fsdevel.
I guess you're mounting with -osync or -odirsync, or opening with O_SYNC?
I think it's a false positive. The filesystem code syncs the inode to disk just after creating it but before doing an unlock_new_inode on it to clear I_NEW. I believe this should not result in any race, because it is called by the guy who has created the inode and he never clears I_NEW. The race came about when pdflush would be able to perform writeout in this window, and non-atomic updates of i_state would race with the creator when they are messing with i_state (without inode_lock held, because inode is new).
We could add some convoluted checks to try to allow for this, but I think we might as well just rip out the checks...
Would anyone prefer me to try to keep them? If not, I'll send a patch to rip them out.
Quoted 120 lines
May 26 02:20:59 vyatta ntpd[2792]: ntpd [email protected] Tue Mar 10 02:08:20 UTC 2009 (1)May 26 02:20:59 vyatta ntpd[2793]: Warning: the "dynamic" keyword has been obsoleted and will be removed in the next releaseMay 26 02:20:59 vyatta last message repeated 3 timesMay 25 19:21:55 vyatta kernel: [ 99.490294] ------------[ cut here ]------------May 25 19:21:55 vyatta kernel: [ 99.490314] WARNING: at fs/fs-writeback.c:292 __writeback_single_inode+0x150/0x35f()May 25 19:21:55 vyatta kernel: [ 99.490349] Hardware name: VMware Virtual PlatformMay 25 19:21:55 vyatta kernel: [ 99.490386] Modules linked in: iptable_nat ip6table_filter iptable_filter ip6table_raw ip6_tables xt_NOTRACK iptable_raw ip_tables x_tables nf_nat_pptp nf_conntrack_pptp nf_conntrack_proto_gre nf_nat_h323 nf_conntrack_h323 nf_nat_sip nf_conntrack_sip nf_nat_proto_gre nf_nat_tftp nf_nat_ftp nf_nat nf_conntrack_ipv4 nf_defrag_ipv4 nf_conntrack_tftp nf_conntrack_ftp nf_conntrack ipv6 md_mod parport_pc parport container psmouse serio_raw pcspkr i2c_piix4 i2c_core button shpchp pci_hotplug intel_agp ac agpgart evdev vfat fat ext2 mbcache battery squashfs loop unionfs nls_utf8 isofs nls_base zlib_inflate sg sr_mod cdrom floppy mptspi mptscsih mptbase scsi_transport_spi pcnet32 mii ata_piix pata_acpi ata_generic libata scsi_mod thermal processor fan thermal_sysMay 25 19:21:55 vyatta kernel: [ 99.490713] Pid: 3483, comm: mkdir Not tainted 2.6.29-1-586-vyatta #1May 25 19:21:55 vyatta kernel: [ 99.490755] Call Trace:May 25 19:21:55 vyatta kernel: [ 99.491060] [<c0122288>] warn_slowpath+0x80/0xb6May 25 19:21:55 vyatta kernel: [ 99.491093] [<c02d8a8d>] schedule+0x688/0x6f3May 25 19:21:55 vyatta kernel: [ 99.491103] [<c0137eea>] getnstimeofday+0x51/0xd7May 25 19:21:55 vyatta kernel: [ 99.491111] [<c01343ad>] ktime_get_ts+0x22/0x49May 25 19:21:55 vyatta kernel: [ 99.491119] [<c0152d90>] delayacct_end+0x70/0x77May 25 19:21:55 vyatta kernel: [ 99.491123] [<c0152de8>] __delayacct_blkio_end+0x2f/0x32May 25 19:21:55 vyatta kernel: [ 99.491127] [<c02d8b63>] io_schedule+0x6b/0x87May 25 19:21:55 vyatta kernel: [ 99.491131] [<c02d8e69>] __wait_on_bit+0x50/0x58May 25 19:21:55 vyatta kernel: [ 99.491142] [<c015abb8>] sync_page+0x0/0x36May 25 19:21:55 vyatta kernel: [ 99.491148] [<c019020c>] __writeback_single_inode+0x150/0x35fMay 25 19:21:55 vyatta kernel: [ 99.491152] [<c01317dd>] wake_bit_function+0x0/0x3cMay 25 19:21:55 vyatta kernel: [ 99.491157] [<c015fa39>] write_one_page+0xa3/0xb8May 25 19:21:55 vyatta kernel: [ 99.491161] [<c0190434>] sync_inode+0x19/0x23May 25 19:21:55 vyatta kernel: [ 99.491562] [<d09bdfe4>] ext2_sync_inode+0x26/0x2c [ext2]May 25 19:21:55 vyatta kernel: [ 99.491577] [<d09bca65>] ext2_make_empty+0x165/0x17b [ext2]May 25 19:21:55 vyatta kernel: [ 99.491589] [<d09bfc8a>] ext2_mkdir+0x9c/0x105 [ext2]May 25 19:21:55 vyatta kernel: [ 99.491607] [<c0183231>] vfs_mkdir+0x51/0x95May 25 19:21:55 vyatta kernel: [ 99.491613] [<c01842e8>] sys_mkdirat+0x97/0xd0May 25 19:21:55 vyatta kernel: [ 99.491621] [<c016d78b>] remove_vma+0x3e/0x43May 25 19:21:55 vyatta kernel: [ 99.491625] [<c016e351>] do_munmap+0x1ed/0x207May 25 19:21:55 vyatta kernel: [ 99.491633] [<c0184330>] sys_mkdir+0xf/0x13May 25 19:21:55 vyatta kernel: [ 99.491638] [<c0102f12>] syscall_call+0x7/0xbMay 25 19:21:55 vyatta kernel: [ 99.491746] ---[ end trace 70fa05fb7550612c ]---May 25 19:21:55 vyatta kernel: [ 99.513627] ------------[ cut here ]------------May 25 19:21:55 vyatta kernel: [ 99.513633] WARNING: at fs/fs-writeback.c:317 __writeback_single_inode+0x202/0x35f()May 25 19:21:55 vyatta kernel: [ 99.513636] Hardware name: VMware Virtual PlatformMay 25 19:21:55 vyatta kernel: [ 99.513638] Modules linked in: iptable_nat ip6table_filter iptable_filter ip6table_raw ip6_tables xt_NOTRACK iptable_raw ip_tables x_tables nf_nat_pptp nf_conntrack_pptp nf_conntrack_proto_gre nf_nat_h323 nf_conntrack_h323 nf_nat_sip nf_conntrack_sip nf_nat_proto_gre nf_nat_tftp nf_nat_ftp nf_nat nf_conntrack_ipv4 nf_defrag_ipv4 nf_conntrack_tftp nf_conntrack_ftp nf_conntrack ipv6 md_mod parport_pc parport container psmouse serio_raw pcspkr i2c_piix4 i2c_core button shpchp pci_hotplug intel_agp ac agpgart evdev vfat fat ext2 mbcache battery squashfs loop unionfs nls_utf8 isofs nls_base zlib_inflate sg sr_mod cdrom floppy mptspi mptscsih mptbase scsi_transport_spi pcnet32 mii ata_piix pata_acpi ata_generic libata scsi_mod thermal processor fan thermal_sysMay 25 19:21:55 vyatta kernel: [ 99.513784] Pid: 3483, comm: mkdir Tainted: G W 2.6.29-1-586-vyatta #1May 25 19:21:55 vyatta kernel: [ 99.513788] Call Trace:May 25 19:21:55 vyatta kernel: [ 99.513794] [<c0122288>] warn_slowpath+0x80/0xb6May 25 19:21:55 vyatta kernel: [ 99.513799] [<c015a7a6>] find_get_pages_tag+0x35/0xd5May 25 19:21:55 vyatta kernel: [ 99.513805] [<c02d8b63>] io_schedule+0x6b/0x87May 25 19:21:55 vyatta kernel: [ 99.513810] [<c0161053>] pagevec_lookup_tag+0x1b/0x22May 25 19:21:55 vyatta kernel: [ 99.513814] [<c015b4ac>] wait_on_page_writeback_range+0x9d/0xdeMay 25 19:21:55 vyatta kernel: [ 99.513819] [<c0194e37>] __wait_on_buffer+0x16/0x18May 25 19:21:55 vyatta kernel: [ 99.513827] [<c01951e3>] sync_dirty_buffer+0x54/0x87May 25 19:21:55 vyatta kernel: [ 99.513841] [<d09be355>] ext2_update_inode+0x28a/0x2c4 [ext2]May 25 19:21:55 vyatta kernel: [ 99.513846] [<c01902be>] __writeback_single_inode+0x202/0x35fMay 25 19:21:55 vyatta kernel: [ 99.513851] [<c015fa39>] write_one_page+0xa3/0xb8May 25 19:21:55 vyatta kernel: [ 99.513855] [<c0190434>] sync_inode+0x19/0x23May 25 19:21:55 vyatta kernel: [ 99.513867] [<d09bdfe4>] ext2_sync_inode+0x26/0x2c [ext2]May 25 19:21:55 vyatta kernel: [ 99.513879] [<d09bca65>] ext2_make_empty+0x165/0x17b [ext2]May 25 19:21:55 vyatta kernel: [ 99.513890] [<d09bfc8a>] ext2_mkdir+0x9c/0x105 [ext2]May 25 19:21:55 vyatta kernel: [ 99.513897] [<c0183231>] vfs_mkdir+0x51/0x95May 25 19:21:55 vyatta kernel: [ 99.513901] [<c01842e8>] sys_mkdirat+0x97/0xd0May 25 19:21:55 vyatta kernel: [ 99.513905] [<c016d78b>] remove_vma+0x3e/0x43May 25 19:21:55 vyatta kernel: [ 99.513909] [<c016e351>] do_munmap+0x1ed/0x207May 25 19:21:55 vyatta kernel: [ 99.513914] [<c0184330>] sys_mkdir+0xf/0x13May 25 19:21:55 vyatta kernel: [ 99.513918] [<c0102f12>] syscall_call+0x7/0xbMay 25 19:21:55 vyatta kernel: [ 99.513921] ---[ end trace 70fa05fb7550612d ]---
Quoted 20 lines · On Thu, 28 May 2009 08:03:33 +0200 Nick Piggin <[email protected]> wrote:
Quoted 5 lines · Stephen Hemminger
The
WARN_ONyou introduced in
fs: new inodei_statecorruption fixTriggers every time during our regression test. The regression test runs on a VM under VMWare. Any ideas?
Thanks. Adding linux-fsdevel.
I guess you're mounting with -osync or -odirsync, or opening with
O_SYNC?I think it's a false positive. The filesystem code syncs the inode to disk just after creating it but before doing an
unlock_new_inodeon it to clearI_NEW. I believe this should not result in any race, because it is called by the guy who has created the inode and he never clearsI_NEW. The race came about when pdflush would be able to perform writeout in this window, and non-atomic updates ofi_statewould race with the creator when they are messing withi_state(withoutinode_lockheld, because inode is new).
Quoted 5 lines
We could add some convoluted checks to try to allow for this, but I think we might as well just rip out the checks...
Would anyone prefer me to try to keep them? If not, I'll send a patch to rip them out.
You are right the file system is mounted on a virtual floppy device with -o sync. That explains it.
This also explains why it only shows up in regression system (developers usually install the software). The floppy is useful for saving configuration when running off live cd.