| Message ID | 20180328140631.13944-1-zlang@redhat.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Wed, Mar 28, 2018 at 10:06:31PM +0800, Zorro Lang wrote: > There's a situation where the directory structure and the inobt > thinks the inode is free, but the inode on disk thinks it is still > in use. XFS should detect it and prevent the kernel from oopsing > on lookup. > > Signed-off-by: Zorro Lang <zlang@redhat.com> > --- > > Hi, > > There's a weird issue: > > When run this case on upstream general kernel(4.16-rc6 without > XFS_WARN/XFS_DEBUG config), it trigger a soft lockup bug[1], > and the case block there. But if I use Dave's patch: > (https://marc.info/?l=linux-xfs&m=152161877728015&w=2) > test passed. I don't know if this softlockup bug is what > Dave tried to fix in his patch too? > > If I test on upstream kernel with XFS_WARN, I didn't hit this > soft lockup issue, just below issue as expected: > XFS: Assertion failed: ip->i_d.di_nblocks == 0, file: fs/xfs/xfs_inode.c > > When I test on RHEL-7 debug kernel (with XFS_WARN), trigger the > soft lockup bug again. > > Thanks, > Zorro > > [1] > [ 455.751099] watchdog: BUG: soft lockup - CPU#12 stuck for 22s! [umount:2631] > [ 455.781145] Modules linked in: sunrpc coretemp intel_rapl sb_edac x86_pkg_temp_thermal intel_powerclamp kvm_intel kvm irqbypass crct10dif_pclmul crc32_pclmul ghash_clmulni_intel pcbc aesni > _intel aes_x86_64 crypto_simd glue_helper cryptd intel_cstate hpilo intel_rapl_perf wmi ipmi_si iTCO_wdt hpwdt iTCO_vendor_support ipmi_devintf sg ipmi_msghandler acpi_power_meter ioatdma pcs > pkr shpchp i2c_i801 pcc_cpufreq dca lpc_ich ip_tables xfs libcrc32c uas usb_storage sd_mod tg3 hwmon mgag200 xhci_pci ptp crc32c_intel serio_raw xhci_hcd hpsa ttm pps_core scsi_transport_sas > dm_mirror dm_region_hash dm_log dm_mod dax ipv6 crc_ccitt autofs4 > [ 456.029470] CPU: 12 PID: 2631 Comm: umount Tainted: G L 4.16.0-rc6+ #3 > [ 456.058306] Hardware name: HP ProLiant DL360 Gen9, BIOS P89 05/06/2015 > [ 456.081804] RIP: 0010:fsnotify_unmount_inodes+0xcc/0x100 > [ 456.099735] RSP: 0018:ffffc900074b3e50 EFLAGS: 00000246 ORIG_RAX: ffffffffffffff12 > [ 456.127922] RAX: 0000000000000000 RBX: ffff88045cecd178 RCX: 000000000000001b > [ 456.154306] RDX: 0000000000000001 RSI: ffffc900074b3d30 RDI: ffff88045cecd200 > [ 456.180539] RBP: 0000000000000000 R08: 000000000000000f R09: ffffc900074b3db8 > [ 456.206731] R10: 000000000000035c R11: 0000000000000018 R12: ffff880465c1cd88 > [ 456.232869] R13: ffff880465c1c800 R14: ffff880465c1cd80 R15: 0000000000000000 > [ 456.259048] FS: 00007f698e06b880(0000) GS:ffff88046f500000(0000) knlGS:0000000000000000 > [ 456.292396] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 > [ 456.314274] CR2: 000055ae574a4628 CR3: 00000004699d6002 CR4: 00000000001606e0 > [ 456.340388] Call Trace: > [ 456.345439] generic_shutdown_super+0x32/0x110 > [ 456.359532] kill_block_super+0x21/0x50 > [ 456.370883] deactivate_locked_super+0x3f/0x70 > [ 456.384883] cleanup_mnt+0x3b/0x70 > [ 456.394269] task_work_run+0x92/0xb0 > [ 456.404408] exit_to_usermode_loop+0x6c/0x99 > [ 456.417663] do_syscall_64+0xf5/0x130 > [ 456.428266] entry_SYSCALL_64_after_hwframe+0x42/0xb7 > [ 456.445027] RIP: 0033:0x7f698d2ddb87 > [ 456.455141] RSP: 002b:00007fffb980d058 EFLAGS: 00000246 ORIG_RAX: 00000000000000a6 > [ 456.483339] RAX: 0000000000000000 RBX: 000055ae5749c080 RCX: 00007f698d2ddb87 > [ 456.509478] RDX: 0000000000000001 RSI: 0000000000000000 RDI: 000055ae574a3460 > [ 456.535573] RBP: 000055ae574a3460 R08: 000055ae574a3480 R09: 0000000000000000 > [ 456.561797] R10: 00007fffb980cae0 R11: 0000000000000246 R12: 00007f698de58d58 > [ 456.588281] R13: 0000000000000000 R14: 000055ae5749c270 R15: 000055ae5749c080 > [ 456.614425] Code: 8d 98 e0 fe ff ff 74 2c 48 8d bb 88 00 00 00 e8 5b fa 52 00 f6 83 a0 00 00 00 38 75 0e 8b 83 58 01 00 00 85 c0 0f 85 74 ff ff ff <c6> 83 88 00 00 00 00 eb c1 41 c6 85 80 05 00 00 00 48 85 ed 74 > > > > tests/xfs/444 | 126 ++++++++++++++++++++++++++++++++++++++++++++++++++++++ > tests/xfs/444.out | 2 + > tests/xfs/group | 1 + > 3 files changed, 129 insertions(+) > create mode 100755 tests/xfs/444 > create mode 100644 tests/xfs/444.out > > diff --git a/tests/xfs/444 b/tests/xfs/444 > new file mode 100755 > index 00000000..58848f4f > --- /dev/null > +++ b/tests/xfs/444 > @@ -0,0 +1,126 @@ > +#! /bin/bash > +# FS QA Test 444 > +# > +# Test a corruption when the directory structure and the inobt thinks the inode > +# is free, but the inode on disk thinks it is still in use. > +# > +#----------------------------------------------------------------------- > +# Copyright (c) 2018 YOUR NAME HERE. All Rights Reserved. Nice patch Mr. HERE. > +# > +# This program is free software; you can redistribute it and/or > +# modify it under the terms of the GNU General Public License as > +# published by the Free Software Foundation. > +# > +# This program is distributed in the hope that it would be useful, > +# but WITHOUT ANY WARRANTY; without even the implied warranty of > +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the > +# GNU General Public License for more details. > +# > +# You should have received a copy of the GNU General Public License > +# along with this program; if not, write the Free Software Foundation, > +# Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA > +#----------------------------------------------------------------------- > +# > + > +seq=`basename $0` > +seqres=$RESULT_DIR/$seq > +echo "QA output created by $seq" > + > +here=`pwd` > +tmp=/tmp/$$ > +status=1 # failure is the default! > +trap "_cleanup; exit \$status" 0 1 2 3 15 > + > +_cleanup() > +{ > + cd / > + rm -f $tmp.* > +} > + > +# get standard environment, filters and checks > +. ./common/rc > +. ./common/filter > + > +# remove previous $seqres.full before test > +rm -f $seqres.full > + > +# real QA test starts here > + > +# Modify as appropriate. > +_supported_fs xfs > +_supported_os Linux > +_require_scratch_nocheck > +_require_no_xfs_bug_on_assert > + > +_filter_dmesg() > +{ > + local warn1="Internal error xfs_trans_cancel.*fs/xfs/xfs_trans\.c.*" > + local warn2="WARNING:.*fs/xfs/xfs_message\.c:.*assfail.*" > + > + sed -e "s#$warn1#Intentional error in xfs_trans_cancel#" \ > + -e "s#$warn2#Intentional warnings in assfail#" > +} > +# If the expected behivor is kernel warning, dissable dmesg, need more check! > +#_disable_dmesg_check Why is this commented out? Can it go away? > + > +# Use crc=0, due to this crash is only possible on v4 XFS or v5 XFS mounted > +# with the ikeep mount option. For all other V5 XFS, this problem cannot > +# occur because we don't read inodes we are allocating from disk - we simply > +# overwrite them with the new inode information. > +_scratch_mkfs_xfs -m crc=0 >> $seqres.full 2>&1 > +blksz=$(_scratch_xfs_get_sb_field blocksize) > +agcount=$(_scratch_xfs_get_sb_field agcount) > + > +_scratch_mount > +# Create a directory for later allocation in same AG (AG 0, due to this's an > +# empty XFS for now) > +mkdir $SCRATCH_MNT/dir > + > +# Allocate 1 block for testfile > +$XFS_IO_PROG -fc 'pwrite 0 $blksz' -c fsync $SCRATCH_MNT/dir/testfile >> $seqres.full > +_scratch_unmount > + > +# We only have one file in one directory (it's generally in AGI 0). So only > +# one AG has free inodes (XFS allocates inodes in chunks of 64), so the > +# AG which has the testfile, its freecount should not be 0. > +for ((agi=0; agi<agcount; agi++)); do > + freecount=$(_scratch_xfs_get_metadata_field freecount "agi $agi") > + if [ "$freecount" != "0" ]; then > + break > + fi > +done > +# Make sure we found the AG contains the testfile > +if [ $agi -gt $agcount ]; then > + _fail "Can't find testfile in which AG" > +fi Can't we figure out which AG the testfile inode is in from the inode number directly? > +# Due to we only allocate 1 block for testfile, and this's the only one data > +# block we use. So we use single level inobt, So the ${agi}->root->recs[1] > +# should be the only one record points the chunk which contains testfile's > +# inode. > +# An exmaple of inode record is as below: > +# recs[1] = [startino,freecount,free] 1:[1024,59,0xffffffffffffffe0] > +freecount=$(_scratch_xfs_get_metadata_field "recs[1].freecount" \ > + "agi $agi" "addr root") > +fmask=$(_scratch_xfs_get_metadata_field "recs[1].free" "agi $agi" "addr root") > + > +# fmask shift right 1 bit, and freecount++, to mark testfile inode as free in > +# inobt. (But the inode itself isn't freed, it still has allocated block) > +freecount="$((freecount + 1))" > +fmask="$((fmask / 2))" TBH I was expecting this to find testfile's inode number, set freecount=1, and then reset the freemask so that testfile is the only free inode in the chunk, thereby forcing(?) the next allocation to end up with testfile's inode and reproduce the crash. Not sure why we're shifting right by one bit? tldr: I'm confused :) > +_scratch_xfs_set_metadata_field "recs[1].freecount" "$freecount" \ > + "agi $agi" "addr root" >/dev/null > +_scratch_xfs_set_metadata_field "recs[1].free" "$fmask" \ > + "agi $agi" "addr root" >/dev/null > + > +# Mount again and create a new inode cover that inode we just 'freed' from inobt > +_scratch_mount > +$XFS_IO_PROG -fc 'pwrite 0 $blksz' -c fsync $SCRATCH_MNT/dir/newfile 2>&1 | \ > + grep -i "Structure needs cleaning" | _filter_scratch How often does this fail to allocate the inode we've messed with? --D > + > +# filter a intentional internal errors > +_check_dmesg _filter_dmesg > + > +# success, all done > +status=0 > +exit > diff --git a/tests/xfs/444.out b/tests/xfs/444.out > new file mode 100644 > index 00000000..2daaf2fc > --- /dev/null > +++ b/tests/xfs/444.out > @@ -0,0 +1,2 @@ > +QA output created by 444 > +SCRATCH_MNT/dir/newfile: Structure needs cleaning > diff --git a/tests/xfs/group b/tests/xfs/group > index e2397fe6..831f2cfa 100644 > --- a/tests/xfs/group > +++ b/tests/xfs/group > @@ -441,3 +441,4 @@ > 441 auto quick clone quota > 442 auto stress clone quota > 443 auto quick ioctl fsr > +444 auto quick > -- > 2.14.3 > > -- > To unsubscribe from this list: send the line "unsubscribe fstests" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html -- To unsubscribe from this list: send the line "unsubscribe fstests" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Wed, Mar 28, 2018 at 09:24:37AM -0700, Darrick J. Wong wrote: > On Wed, Mar 28, 2018 at 10:06:31PM +0800, Zorro Lang wrote: > > There's a situation where the directory structure and the inobt > > thinks the inode is free, but the inode on disk thinks it is still > > in use. XFS should detect it and prevent the kernel from oopsing > > on lookup. > > > > Signed-off-by: Zorro Lang <zlang@redhat.com> > > --- > > > > Hi, > > > > There's a weird issue: > > > > When run this case on upstream general kernel(4.16-rc6 without > > XFS_WARN/XFS_DEBUG config), it trigger a soft lockup bug[1], > > and the case block there. But if I use Dave's patch: > > (https://marc.info/?l=linux-xfs&m=152161877728015&w=2) > > test passed. I don't know if this softlockup bug is what > > Dave tried to fix in his patch too? > > > > If I test on upstream kernel with XFS_WARN, I didn't hit this > > soft lockup issue, just below issue as expected: > > XFS: Assertion failed: ip->i_d.di_nblocks == 0, file: fs/xfs/xfs_inode.c > > > > When I test on RHEL-7 debug kernel (with XFS_WARN), trigger the > > soft lockup bug again. > > > > Thanks, > > Zorro > > > > [1] > > [ 455.751099] watchdog: BUG: soft lockup - CPU#12 stuck for 22s! [umount:2631] > > [ 455.781145] Modules linked in: sunrpc coretemp intel_rapl sb_edac x86_pkg_temp_thermal intel_powerclamp kvm_intel kvm irqbypass crct10dif_pclmul crc32_pclmul ghash_clmulni_intel pcbc aesni > > _intel aes_x86_64 crypto_simd glue_helper cryptd intel_cstate hpilo intel_rapl_perf wmi ipmi_si iTCO_wdt hpwdt iTCO_vendor_support ipmi_devintf sg ipmi_msghandler acpi_power_meter ioatdma pcs > > pkr shpchp i2c_i801 pcc_cpufreq dca lpc_ich ip_tables xfs libcrc32c uas usb_storage sd_mod tg3 hwmon mgag200 xhci_pci ptp crc32c_intel serio_raw xhci_hcd hpsa ttm pps_core scsi_transport_sas > > dm_mirror dm_region_hash dm_log dm_mod dax ipv6 crc_ccitt autofs4 > > [ 456.029470] CPU: 12 PID: 2631 Comm: umount Tainted: G L 4.16.0-rc6+ #3 > > [ 456.058306] Hardware name: HP ProLiant DL360 Gen9, BIOS P89 05/06/2015 > > [ 456.081804] RIP: 0010:fsnotify_unmount_inodes+0xcc/0x100 > > [ 456.099735] RSP: 0018:ffffc900074b3e50 EFLAGS: 00000246 ORIG_RAX: ffffffffffffff12 > > [ 456.127922] RAX: 0000000000000000 RBX: ffff88045cecd178 RCX: 000000000000001b > > [ 456.154306] RDX: 0000000000000001 RSI: ffffc900074b3d30 RDI: ffff88045cecd200 > > [ 456.180539] RBP: 0000000000000000 R08: 000000000000000f R09: ffffc900074b3db8 > > [ 456.206731] R10: 000000000000035c R11: 0000000000000018 R12: ffff880465c1cd88 > > [ 456.232869] R13: ffff880465c1c800 R14: ffff880465c1cd80 R15: 0000000000000000 > > [ 456.259048] FS: 00007f698e06b880(0000) GS:ffff88046f500000(0000) knlGS:0000000000000000 > > [ 456.292396] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 > > [ 456.314274] CR2: 000055ae574a4628 CR3: 00000004699d6002 CR4: 00000000001606e0 > > [ 456.340388] Call Trace: > > [ 456.345439] generic_shutdown_super+0x32/0x110 > > [ 456.359532] kill_block_super+0x21/0x50 > > [ 456.370883] deactivate_locked_super+0x3f/0x70 > > [ 456.384883] cleanup_mnt+0x3b/0x70 > > [ 456.394269] task_work_run+0x92/0xb0 > > [ 456.404408] exit_to_usermode_loop+0x6c/0x99 > > [ 456.417663] do_syscall_64+0xf5/0x130 > > [ 456.428266] entry_SYSCALL_64_after_hwframe+0x42/0xb7 > > [ 456.445027] RIP: 0033:0x7f698d2ddb87 > > [ 456.455141] RSP: 002b:00007fffb980d058 EFLAGS: 00000246 ORIG_RAX: 00000000000000a6 > > [ 456.483339] RAX: 0000000000000000 RBX: 000055ae5749c080 RCX: 00007f698d2ddb87 > > [ 456.509478] RDX: 0000000000000001 RSI: 0000000000000000 RDI: 000055ae574a3460 > > [ 456.535573] RBP: 000055ae574a3460 R08: 000055ae574a3480 R09: 0000000000000000 > > [ 456.561797] R10: 00007fffb980cae0 R11: 0000000000000246 R12: 00007f698de58d58 > > [ 456.588281] R13: 0000000000000000 R14: 000055ae5749c270 R15: 000055ae5749c080 > > [ 456.614425] Code: 8d 98 e0 fe ff ff 74 2c 48 8d bb 88 00 00 00 e8 5b fa 52 00 f6 83 a0 00 00 00 38 75 0e 8b 83 58 01 00 00 85 c0 0f 85 74 ff ff ff <c6> 83 88 00 00 00 00 eb c1 41 c6 85 80 05 00 00 00 48 85 ed 74 Any idea about if this's https://marc.info/?l=linux-xfs&m=152161877728015&w=2 try to fix? > > > > > > > > tests/xfs/444 | 126 ++++++++++++++++++++++++++++++++++++++++++++++++++++++ > > tests/xfs/444.out | 2 + > > tests/xfs/group | 1 + > > 3 files changed, 129 insertions(+) > > create mode 100755 tests/xfs/444 > > create mode 100644 tests/xfs/444.out > > > > diff --git a/tests/xfs/444 b/tests/xfs/444 > > new file mode 100755 > > index 00000000..58848f4f > > --- /dev/null > > +++ b/tests/xfs/444 > > @@ -0,0 +1,126 @@ > > +#! /bin/bash > > +# FS QA Test 444 > > +# > > +# Test a corruption when the directory structure and the inobt thinks the inode > > +# is free, but the inode on disk thinks it is still in use. > > +# > > +#----------------------------------------------------------------------- > > +# Copyright (c) 2018 YOUR NAME HERE. All Rights Reserved. > > Nice patch Mr. HERE. Ah, I always forgot changing this in V1 patch... > > > +# > > +# This program is free software; you can redistribute it and/or > > +# modify it under the terms of the GNU General Public License as > > +# published by the Free Software Foundation. > > +# > > +# This program is distributed in the hope that it would be useful, > > +# but WITHOUT ANY WARRANTY; without even the implied warranty of > > +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the > > +# GNU General Public License for more details. > > +# > > +# You should have received a copy of the GNU General Public License > > +# along with this program; if not, write the Free Software Foundation, > > +# Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA > > +#----------------------------------------------------------------------- > > +# > > + > > +seq=`basename $0` > > +seqres=$RESULT_DIR/$seq > > +echo "QA output created by $seq" > > + > > +here=`pwd` > > +tmp=/tmp/$$ > > +status=1 # failure is the default! > > +trap "_cleanup; exit \$status" 0 1 2 3 15 > > + > > +_cleanup() > > +{ > > + cd / > > + rm -f $tmp.* > > +} > > + > > +# get standard environment, filters and checks > > +. ./common/rc > > +. ./common/filter > > + > > +# remove previous $seqres.full before test > > +rm -f $seqres.full > > + > > +# real QA test starts here > > + > > +# Modify as appropriate. > > +_supported_fs xfs > > +_supported_os Linux > > +_require_scratch_nocheck > > +_require_no_xfs_bug_on_assert > > + > > +_filter_dmesg() > > +{ > > + local warn1="Internal error xfs_trans_cancel.*fs/xfs/xfs_trans\.c.*" > > + local warn2="WARNING:.*fs/xfs/xfs_message\.c:.*assfail.*" > > + > > + sed -e "s#$warn1#Intentional error in xfs_trans_cancel#" \ > > + -e "s#$warn2#Intentional warnings in assfail#" > > +} > > +# If the expected behivor is kernel warning, dissable dmesg, need more check! > > +#_disable_dmesg_check > > Why is this commented out? Can it go away? Yeah, it should be removed. > > > + > > +# Use crc=0, due to this crash is only possible on v4 XFS or v5 XFS mounted > > +# with the ikeep mount option. For all other V5 XFS, this problem cannot > > +# occur because we don't read inodes we are allocating from disk - we simply > > +# overwrite them with the new inode information. > > +_scratch_mkfs_xfs -m crc=0 >> $seqres.full 2>&1 > > +blksz=$(_scratch_xfs_get_sb_field blocksize) > > +agcount=$(_scratch_xfs_get_sb_field agcount) > > + > > +_scratch_mount > > +# Create a directory for later allocation in same AG (AG 0, due to this's an > > +# empty XFS for now) > > +mkdir $SCRATCH_MNT/dir > > + > > +# Allocate 1 block for testfile > > +$XFS_IO_PROG -fc 'pwrite 0 $blksz' -c fsync $SCRATCH_MNT/dir/testfile >> $seqres.full > > +_scratch_unmount > > + > > +# We only have one file in one directory (it's generally in AGI 0). So only > > +# one AG has free inodes (XFS allocates inodes in chunks of 64), so the > > +# AG which has the testfile, its freecount should not be 0. > > +for ((agi=0; agi<agcount; agi++)); do > > + freecount=$(_scratch_xfs_get_metadata_field freecount "agi $agi") > > + if [ "$freecount" != "0" ]; then > > + break > > + fi > > +done > > +# Make sure we found the AG contains the testfile > > +if [ $agi -gt $agcount ]; then > > + _fail "Can't find testfile in which AG" > > +fi > > Can't we figure out which AG the testfile inode is in from the inode > number directly? Sure, thanks for you told me how to do that:) > > > +# Due to we only allocate 1 block for testfile, and this's the only one data > > +# block we use. So we use single level inobt, So the ${agi}->root->recs[1] > > +# should be the only one record points the chunk which contains testfile's > > +# inode. > > +# An exmaple of inode record is as below: > > +# recs[1] = [startino,freecount,free] 1:[1024,59,0xffffffffffffffe0] > > +freecount=$(_scratch_xfs_get_metadata_field "recs[1].freecount" \ > > + "agi $agi" "addr root") > > +fmask=$(_scratch_xfs_get_metadata_field "recs[1].free" "agi $agi" "addr root") > > + > > +# fmask shift right 1 bit, and freecount++, to mark testfile inode as free in > > +# inobt. (But the inode itself isn't freed, it still has allocated block) > > +freecount="$((freecount + 1))" > > +fmask="$((fmask / 2))" > > TBH I was expecting this to find testfile's inode number, set > freecount=1, and then reset the freemask so that testfile is the only > free inode in the chunk, thereby forcing(?) the next allocation to end > up with testfile's inode and reproduce the crash. Not sure why we're > shifting right by one bit? > > tldr: I'm confused :) Hmmm... a little confused at here. Do you mean this: # stat -c %i /mnt/test/dir/testfile 1028 # umount $dev # xfs_db -x $dev xfs_db> inode 1028 xfs_db> convert inode 1028 agno 0x0 (0) xfs_db> agi 0 xfs_db> addr root xfs_db> p magic = 0x49414254 level = 0 numrecs = 1 leftsib = null rightsib = null recs[1] = [startino,freecount,free] 1:[1024,59,0xffffffffffffffe0] xfs_db> write recs[1].startino 1028 recs[1].startino = 1028 xfs_db> write recs[1].freecount 1 recs[1].freecount = 1 xfs_db> write recs[1].free 1 recs[1].free = 0x1 xfs_db> q But after mount this XFS again, and tried to do `touch /mnt/test/dir/newfile`, I got this warning: [47420.479191] XFS: Assertion failed: fs_is_ok, file: fs/xfs/libxfs/xfs_ialloc.c, line: 1156 [45/9735] [47420.520226] ------------[ cut here ]------------ [47420.543399] WARNING: CPU: 13 PID: 2267 at fs/xfs/xfs_message.c:105 asswarn+0x33/0x40 [xfs] .... [47421.791340] XFS (dm-2): Internal error XFS_WANT_CORRUPTED_GOTO at line 1156 of file fs/xfs/libxfs/xfs_ialloc.c. Caller xfs_dialloc_ag+0x6e/0x360 [xfs] .... Hmm... I'm confused. > > > +_scratch_xfs_set_metadata_field "recs[1].freecount" "$freecount" \ > > + "agi $agi" "addr root" >/dev/null > > +_scratch_xfs_set_metadata_field "recs[1].free" "$fmask" \ > > + "agi $agi" "addr root" >/dev/null > > + > > +# Mount again and create a new inode cover that inode we just 'freed' from inobt > > +_scratch_mount > > +$XFS_IO_PROG -fc 'pwrite 0 $blksz' -c fsync $SCRATCH_MNT/dir/newfile 2>&1 | \ > > + grep -i "Structure needs cleaning" | _filter_scratch > > How often does this fail to allocate the inode we've messed with? Everytime in my test Thanks, Zorro. > > --D > > > + > > +# filter a intentional internal errors > > +_check_dmesg _filter_dmesg > > + > > +# success, all done > > +status=0 > > +exit > > diff --git a/tests/xfs/444.out b/tests/xfs/444.out > > new file mode 100644 > > index 00000000..2daaf2fc > > --- /dev/null > > +++ b/tests/xfs/444.out > > @@ -0,0 +1,2 @@ > > +QA output created by 444 > > +SCRATCH_MNT/dir/newfile: Structure needs cleaning > > diff --git a/tests/xfs/group b/tests/xfs/group > > index e2397fe6..831f2cfa 100644 > > --- a/tests/xfs/group > > +++ b/tests/xfs/group > > @@ -441,3 +441,4 @@ > > 441 auto quick clone quota > > 442 auto stress clone quota > > 443 auto quick ioctl fsr > > +444 auto quick > > -- > > 2.14.3 > > > > -- > > To unsubscribe from this list: send the line "unsubscribe fstests" in > > the body of a message to majordomo@vger.kernel.org > > More majordomo info at http://vger.kernel.org/majordomo-info.html > -- > To unsubscribe from this list: send the line "unsubscribe fstests" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html -- To unsubscribe from this list: send the line "unsubscribe fstests" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/tests/xfs/444 b/tests/xfs/444 new file mode 100755 index 00000000..58848f4f --- /dev/null +++ b/tests/xfs/444 @@ -0,0 +1,126 @@ +#! /bin/bash +# FS QA Test 444 +# +# Test a corruption when the directory structure and the inobt thinks the inode +# is free, but the inode on disk thinks it is still in use. +# +#----------------------------------------------------------------------- +# Copyright (c) 2018 YOUR NAME HERE. All Rights Reserved. +# +# This program is free software; you can redistribute it and/or +# modify it under the terms of the GNU General Public License as +# published by the Free Software Foundation. +# +# This program is distributed in the hope that it would be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +# GNU General Public License for more details. +# +# You should have received a copy of the GNU General Public License +# along with this program; if not, write the Free Software Foundation, +# Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA +#----------------------------------------------------------------------- +# + +seq=`basename $0` +seqres=$RESULT_DIR/$seq +echo "QA output created by $seq" + +here=`pwd` +tmp=/tmp/$$ +status=1 # failure is the default! +trap "_cleanup; exit \$status" 0 1 2 3 15 + +_cleanup() +{ + cd / + rm -f $tmp.* +} + +# get standard environment, filters and checks +. ./common/rc +. ./common/filter + +# remove previous $seqres.full before test +rm -f $seqres.full + +# real QA test starts here + +# Modify as appropriate. +_supported_fs xfs +_supported_os Linux +_require_scratch_nocheck +_require_no_xfs_bug_on_assert + +_filter_dmesg() +{ + local warn1="Internal error xfs_trans_cancel.*fs/xfs/xfs_trans\.c.*" + local warn2="WARNING:.*fs/xfs/xfs_message\.c:.*assfail.*" + + sed -e "s#$warn1#Intentional error in xfs_trans_cancel#" \ + -e "s#$warn2#Intentional warnings in assfail#" +} +# If the expected behivor is kernel warning, dissable dmesg, need more check! +#_disable_dmesg_check + +# Use crc=0, due to this crash is only possible on v4 XFS or v5 XFS mounted +# with the ikeep mount option. For all other V5 XFS, this problem cannot +# occur because we don't read inodes we are allocating from disk - we simply +# overwrite them with the new inode information. +_scratch_mkfs_xfs -m crc=0 >> $seqres.full 2>&1 +blksz=$(_scratch_xfs_get_sb_field blocksize) +agcount=$(_scratch_xfs_get_sb_field agcount) + +_scratch_mount +# Create a directory for later allocation in same AG (AG 0, due to this's an +# empty XFS for now) +mkdir $SCRATCH_MNT/dir + +# Allocate 1 block for testfile +$XFS_IO_PROG -fc 'pwrite 0 $blksz' -c fsync $SCRATCH_MNT/dir/testfile >> $seqres.full +_scratch_unmount + +# We only have one file in one directory (it's generally in AGI 0). So only +# one AG has free inodes (XFS allocates inodes in chunks of 64), so the +# AG which has the testfile, its freecount should not be 0. +for ((agi=0; agi<agcount; agi++)); do + freecount=$(_scratch_xfs_get_metadata_field freecount "agi $agi") + if [ "$freecount" != "0" ]; then + break + fi +done +# Make sure we found the AG contains the testfile +if [ $agi -gt $agcount ]; then + _fail "Can't find testfile in which AG" +fi + +# Due to we only allocate 1 block for testfile, and this's the only one data +# block we use. So we use single level inobt, So the ${agi}->root->recs[1] +# should be the only one record points the chunk which contains testfile's +# inode. +# An exmaple of inode record is as below: +# recs[1] = [startino,freecount,free] 1:[1024,59,0xffffffffffffffe0] +freecount=$(_scratch_xfs_get_metadata_field "recs[1].freecount" \ + "agi $agi" "addr root") +fmask=$(_scratch_xfs_get_metadata_field "recs[1].free" "agi $agi" "addr root") + +# fmask shift right 1 bit, and freecount++, to mark testfile inode as free in +# inobt. (But the inode itself isn't freed, it still has allocated block) +freecount="$((freecount + 1))" +fmask="$((fmask / 2))" +_scratch_xfs_set_metadata_field "recs[1].freecount" "$freecount" \ + "agi $agi" "addr root" >/dev/null +_scratch_xfs_set_metadata_field "recs[1].free" "$fmask" \ + "agi $agi" "addr root" >/dev/null + +# Mount again and create a new inode cover that inode we just 'freed' from inobt +_scratch_mount +$XFS_IO_PROG -fc 'pwrite 0 $blksz' -c fsync $SCRATCH_MNT/dir/newfile 2>&1 | \ + grep -i "Structure needs cleaning" | _filter_scratch + +# filter a intentional internal errors +_check_dmesg _filter_dmesg + +# success, all done +status=0 +exit diff --git a/tests/xfs/444.out b/tests/xfs/444.out new file mode 100644 index 00000000..2daaf2fc --- /dev/null +++ b/tests/xfs/444.out @@ -0,0 +1,2 @@ +QA output created by 444 +SCRATCH_MNT/dir/newfile: Structure needs cleaning diff --git a/tests/xfs/group b/tests/xfs/group index e2397fe6..831f2cfa 100644 --- a/tests/xfs/group +++ b/tests/xfs/group @@ -441,3 +441,4 @@ 441 auto quick clone quota 442 auto stress clone quota 443 auto quick ioctl fsr +444 auto quick
There's a situation where the directory structure and the inobt thinks the inode is free, but the inode on disk thinks it is still in use. XFS should detect it and prevent the kernel from oopsing on lookup. Signed-off-by: Zorro Lang <zlang@redhat.com> --- Hi, There's a weird issue: When run this case on upstream general kernel(4.16-rc6 without XFS_WARN/XFS_DEBUG config), it trigger a soft lockup bug[1], and the case block there. But if I use Dave's patch: (https://marc.info/?l=linux-xfs&m=152161877728015&w=2) test passed. I don't know if this softlockup bug is what Dave tried to fix in his patch too? If I test on upstream kernel with XFS_WARN, I didn't hit this soft lockup issue, just below issue as expected: XFS: Assertion failed: ip->i_d.di_nblocks == 0, file: fs/xfs/xfs_inode.c When I test on RHEL-7 debug kernel (with XFS_WARN), trigger the soft lockup bug again. Thanks, Zorro [1] [ 455.751099] watchdog: BUG: soft lockup - CPU#12 stuck for 22s! [umount:2631] [ 455.781145] Modules linked in: sunrpc coretemp intel_rapl sb_edac x86_pkg_temp_thermal intel_powerclamp kvm_intel kvm irqbypass crct10dif_pclmul crc32_pclmul ghash_clmulni_intel pcbc aesni _intel aes_x86_64 crypto_simd glue_helper cryptd intel_cstate hpilo intel_rapl_perf wmi ipmi_si iTCO_wdt hpwdt iTCO_vendor_support ipmi_devintf sg ipmi_msghandler acpi_power_meter ioatdma pcs pkr shpchp i2c_i801 pcc_cpufreq dca lpc_ich ip_tables xfs libcrc32c uas usb_storage sd_mod tg3 hwmon mgag200 xhci_pci ptp crc32c_intel serio_raw xhci_hcd hpsa ttm pps_core scsi_transport_sas dm_mirror dm_region_hash dm_log dm_mod dax ipv6 crc_ccitt autofs4 [ 456.029470] CPU: 12 PID: 2631 Comm: umount Tainted: G L 4.16.0-rc6+ #3 [ 456.058306] Hardware name: HP ProLiant DL360 Gen9, BIOS P89 05/06/2015 [ 456.081804] RIP: 0010:fsnotify_unmount_inodes+0xcc/0x100 [ 456.099735] RSP: 0018:ffffc900074b3e50 EFLAGS: 00000246 ORIG_RAX: ffffffffffffff12 [ 456.127922] RAX: 0000000000000000 RBX: ffff88045cecd178 RCX: 000000000000001b [ 456.154306] RDX: 0000000000000001 RSI: ffffc900074b3d30 RDI: ffff88045cecd200 [ 456.180539] RBP: 0000000000000000 R08: 000000000000000f R09: ffffc900074b3db8 [ 456.206731] R10: 000000000000035c R11: 0000000000000018 R12: ffff880465c1cd88 [ 456.232869] R13: ffff880465c1c800 R14: ffff880465c1cd80 R15: 0000000000000000 [ 456.259048] FS: 00007f698e06b880(0000) GS:ffff88046f500000(0000) knlGS:0000000000000000 [ 456.292396] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 456.314274] CR2: 000055ae574a4628 CR3: 00000004699d6002 CR4: 00000000001606e0 [ 456.340388] Call Trace: [ 456.345439] generic_shutdown_super+0x32/0x110 [ 456.359532] kill_block_super+0x21/0x50 [ 456.370883] deactivate_locked_super+0x3f/0x70 [ 456.384883] cleanup_mnt+0x3b/0x70 [ 456.394269] task_work_run+0x92/0xb0 [ 456.404408] exit_to_usermode_loop+0x6c/0x99 [ 456.417663] do_syscall_64+0xf5/0x130 [ 456.428266] entry_SYSCALL_64_after_hwframe+0x42/0xb7 [ 456.445027] RIP: 0033:0x7f698d2ddb87 [ 456.455141] RSP: 002b:00007fffb980d058 EFLAGS: 00000246 ORIG_RAX: 00000000000000a6 [ 456.483339] RAX: 0000000000000000 RBX: 000055ae5749c080 RCX: 00007f698d2ddb87 [ 456.509478] RDX: 0000000000000001 RSI: 0000000000000000 RDI: 000055ae574a3460 [ 456.535573] RBP: 000055ae574a3460 R08: 000055ae574a3480 R09: 0000000000000000 [ 456.561797] R10: 00007fffb980cae0 R11: 0000000000000246 R12: 00007f698de58d58 [ 456.588281] R13: 0000000000000000 R14: 000055ae5749c270 R15: 000055ae5749c080 [ 456.614425] Code: 8d 98 e0 fe ff ff 74 2c 48 8d bb 88 00 00 00 e8 5b fa 52 00 f6 83 a0 00 00 00 38 75 0e 8b 83 58 01 00 00 85 c0 0f 85 74 ff ff ff <c6> 83 88 00 00 00 00 eb c1 41 c6 85 80 05 00 00 00 48 85 ed 74 tests/xfs/444 | 126 ++++++++++++++++++++++++++++++++++++++++++++++++++++++ tests/xfs/444.out | 2 + tests/xfs/group | 1 + 3 files changed, 129 insertions(+) create mode 100755 tests/xfs/444 create mode 100644 tests/xfs/444.out