* null-ptr-deref due to "ext4: fix potential race between online resizing and write operations"
@ 2020-02-21 14:02 Qian Cai
2020-02-21 19:58 ` Jitindar SIngh, Suraj
0 siblings, 1 reply; 3+ messages in thread
From: Qian Cai @ 2020-02-21 14:02 UTC (permalink / raw)
To: Suraj Jitindar Singh
Cc: Theodore Ts'o, Andreas Dilger, linux-ext4, linux-kernel,
Paul E. McKenney
Reverted the linux-next commit c20bac9bf82c ("ext4: fix potential race between
s_flex_groups online resizing and access") fixed the crash below (with line
numbers),
struct flex_groups *flex_group = sbi_array_rcu_deref(EXT4_SB(sb),
s_flex_groups, g);
[ 575.924527][T13183] LTP: starting fanotify13
[ 576.010554][T31835] /dev/zero: Can't open blockdev
[ 576.867392][T31835] EXT4-fs (loop0): mounting ext3 file system using the ext4
subsystem
[ 576.919604][T31835] EXT4-fs (loop0): mounted filesystem with ordered data
mode. Opts: (null)
[ 576.920112][T31835] ext3 filesystem being mounted at /tmp/ltp-
ZMONVGlgwi/o0A0RE/mntpoint supports timestamps until 2038 (0x7fffffff)
[ 576.948501][T31854] BUG: Kernel NULL pointer dereference on read at
0x00000070
[ 576.948550][T31854] Faulting instruction address: 0xc008000010501bfc
[ 576.948573][T31854] Oops: Kernel access of bad area, sig: 11 [#1]
[ 576.948575][ C2] irq event stamp: 107073312
[ 576.948583][ C2] hardirqs last enabled at (107073312):
[<c00000000099a174>] _raw_spin_unlock_irqrestore+0x94/0xd0
[ 576.948595][T31854] LE PAGE_SIZE=64K MMU=Radix SMP NR_CPUS=256
DEBUG_PAGEALLOC NUMA PowerNV
[ 576.948598][T31854] Modules linked in: brd ext4 crc16 mbcache jbd2 loop
ip_tables x_tables xfs sd_mod bnx2x ahci libahci mdio libata tg3 libphy
firmware_class dm_mirror dm_region_hash dm_log dm_mod
[ 576.948614][ C2] hardirqs last disabled at (107073311):
[<c000000000999e0c>] _raw_spin_lock_irqsave+0x3c/0xa0
[ 576.948646][T31854] CPU: 52 PID: 31854 Comm: fanotify13 Not tainted 5.6.0-
rc2-next-20200221 #7
[ 576.948689][ C2] softirqs last enabled at (107073296):
[<c000000000113b3c>] irq_enter+0x8c/0xc0
[ 576.948693][ C2] softirqs last disabled at (107073297):
[<c000000000113cdc>] irq_exit+0x16c/0x1d0
[ 576.948754][T31854] NIP: c008000010501bfc LR: c008000010501d94 CTR:
c0000000001f1e30
[ 576.948758][T31854] REGS: c00000129f56f700 TRAP: 0300 Not tainted (5.6.0-
rc2-next-20200221)
[ 576.948945][T31854] MSR: 9000000000009033 <SF,HV,EE,ME,IR,DR,RI,LE> CR:
24004224 XER: 20040000
[ 576.948982][T31854] CFAR: c008000010501d9c DAR: 0000000000000070 DSISR:
40000000 IRQMASK: 0
[ 576.948982][T31854] GPR00: c008000010501d94 c00000129f56f990 c0080000105c1600
0000000000000001
[ 576.948982][T31854] GPR04: c000000001510808 0000000000000008 0000000005cf0ca2
fffffffe5ca98558
[ 576.948982][T31854] GPR08: 0000000000000001 0000000000000070 0000000000000000
c00800001057b690
[ 576.948982][T31854] GPR12: c0000000001f1e30 c000001ffffd5600 000000000000000e
00000000000007ff
[ 576.948982][T31854] GPR16: c00000129f56fa20 000000000000fff5 0000000000000001
0000000000001dbc
[ 576.948982][T31854] GPR20: 0000000000000000 000000000000002e 0000000000000800
0000000000000020
[ 576.948982][T31854] GPR24: 000000000000000e 0000000000000000 0000000000000000
c000000001510808
[ 576.948982][T31854] GPR28: c000001206b8d000 c0080000105d8227 c00000129f56fa20
0000000000000001
[ 576.949200][T31854] NIP [c008000010501bfc] get_orlov_stats+0x114/0x390 [ext4]
get_orlov_stats at fs/ext4/ialloc.c:373 (discriminator 11)
[ 576.949232][T31854] LR [c008000010501d94] get_orlov_stats+0x2ac/0x390 [ext4]
[ 576.949243][T31854] Call Trace:
[ 576.949260][T31854] [c00000129f56f990] [c008000010501d94]
get_orlov_stats+0x2ac/0x390 [ext4] (unreliable)
get_orlov_stats at fs/ext4/ialloc.c:373 (discriminator 11)
[ 576.949301][T31854] [c00000129f56f9f0] [c00800001050231c]
find_group_orlov+0x4a4/0x6b0 [ext4]
find_group_orlov at fs/ext4/ialloc.c:467
[ 576.949334][T31854] [c00000129f56fae0] [c0080000105055c8]
__ext4_new_inode+0x1450/0x23c0 [ext4]
[ 576.949367][T31854] [c00000129f56fc50] [c008000010547f2c]
ext4_mkdir+0x104/0x590 [ext4]
[ 576.949399][T31854] [c00000129f56fd60] [c0000000004cbc64]
vfs_mkdir+0x114/0x210
[ 576.949432][T31854] [c00000129f56fda0] [c0000000004d1a70]
do_mkdirat+0xb0/0x1a0
[ 576.949454][T31854] [c00000129f56fe20] [c00000000000b378]
system_call+0x5c/0x68
[ 576.949465][T31854] Instruction dump:
[ 576.949473][T31854] 3c620000 e8638730 7f44d378 38630068 48078ccd e8410018
60000000 60000000
[ 576.949497][T31854] 60000000 73490001 4182019c 7b091f24 <7f59482a> 4807a0d1
e8410018 2fa30000
[ 576.949522][T31854] ---[ end trace de4acb29e0d7791c ]---
[ 577.200573][T31854]
[ 578.200652][T31854] Kernel panic - not syncing: Fatal exception
[ 579
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: null-ptr-deref due to "ext4: fix potential race between online resizing and write operations"
2020-02-21 14:02 null-ptr-deref due to "ext4: fix potential race between online resizing and write operations" Qian Cai
@ 2020-02-21 19:58 ` Jitindar SIngh, Suraj
2020-02-22 0:33 ` Theodore Y. Ts'o
0 siblings, 1 reply; 3+ messages in thread
From: Jitindar SIngh, Suraj @ 2020-02-21 19:58 UTC (permalink / raw)
To: cai; +Cc: adilger.kernel, linux-ext4, linux-kernel, paulmck, tytso
On Fri, 2020-02-21 at 09:02 -0500, Qian Cai wrote:
> Reverted the linux-next commit c20bac9bf82c ("ext4: fix potential
> race between
> s_flex_groups online resizing and access") fixed the crash below
> (with line
> numbers),
Good catch, this is a bug where the dereference of the array
s_flex_groups needs to happen after the "if (flex_size > 1)" if
statement in fs/ext4/ialloc.c:373
>
> struct flex_groups *flex_group = sbi_array_rcu_deref(EXT4_SB(sb),
> s_flex_groups,
> g);
>
> [ 575.924527][T13183] LTP: starting fanotify13
> [ 576.010554][T31835] /dev/zero: Can't open blockdev
> [ 576.867392][T31835] EXT4-fs (loop0): mounting ext3 file system
> using the ext4
> subsystem
> [ 576.919604][T31835] EXT4-fs (loop0): mounted filesystem with
> ordered data
> mode. Opts: (null)
> [ 576.920112][T31835] ext3 filesystem being mounted at /tmp/ltp-
> ZMONVGlgwi/o0A0RE/mntpoint supports timestamps until 2038
> (0x7fffffff)
> [ 576.948501][T31854] BUG: Kernel NULL pointer dereference on read
> at
> 0x00000070
> [ 576.948550][T31854] Faulting instruction address:
> 0xc008000010501bfc
> [ 576.948573][T31854] Oops: Kernel access of bad area, sig: 11 [#1]
> [ 576.948575][ C2] irq event stamp: 107073312
> [ 576.948583][ C2] hardirqs last enabled at (107073312):
> [<c00000000099a174>] _raw_spin_unlock_irqrestore+0x94/0xd0
> [ 576.948595][T31854] LE PAGE_SIZE=64K MMU=Radix SMP NR_CPUS=256
> DEBUG_PAGEALLOC NUMA PowerNV
> [ 576.948598][T31854] Modules linked in: brd ext4 crc16 mbcache jbd2
> loop
> ip_tables x_tables xfs sd_mod bnx2x ahci libahci mdio libata tg3
> libphy
> firmware_class dm_mirror dm_region_hash dm_log dm_mod
> [ 576.948614][ C2] hardirqs last disabled at (107073311):
> [<c000000000999e0c>] _raw_spin_lock_irqsave+0x3c/0xa0
> [ 576.948646][T31854] CPU: 52 PID: 31854 Comm: fanotify13 Not
> tainted 5.6.0-
> rc2-next-20200221 #7
> [ 576.948689][ C2] softirqs last enabled at (107073296):
> [<c000000000113b3c>] irq_enter+0x8c/0xc0
> [ 576.948693][ C2] softirqs last disabled at (107073297):
> [<c000000000113cdc>] irq_exit+0x16c/0x1d0
> [ 576.948754][T31854] NIP: c008000010501bfc LR: c008000010501d94
> CTR:
> c0000000001f1e30
> [ 576.948758][T31854] REGS: c00000129f56f700 TRAP: 0300 Not
> tainted (5.6.0-
> rc2-next-20200221)
> [ 576.948945][T31854] MSR: 9000000000009033
> <SF,HV,EE,ME,IR,DR,RI,LE> CR:
> 24004224 XER: 20040000
> [ 576.948982][T31854] CFAR: c008000010501d9c DAR: 0000000000000070
> DSISR:
> 40000000 IRQMASK: 0
> [ 576.948982][T31854] GPR00: c008000010501d94 c00000129f56f990
> c0080000105c1600
> 0000000000000001
> [ 576.948982][T31854] GPR04: c000000001510808 0000000000000008
> 0000000005cf0ca2
> fffffffe5ca98558
> [ 576.948982][T31854] GPR08: 0000000000000001 0000000000000070
> 0000000000000000
> c00800001057b690
> [ 576.948982][T31854] GPR12: c0000000001f1e30 c000001ffffd5600
> 000000000000000e
> 00000000000007ff
> [ 576.948982][T31854] GPR16: c00000129f56fa20 000000000000fff5
> 0000000000000001
> 0000000000001dbc
> [ 576.948982][T31854] GPR20: 0000000000000000 000000000000002e
> 0000000000000800
> 0000000000000020
> [ 576.948982][T31854] GPR24: 000000000000000e 0000000000000000
> 0000000000000000
> c000000001510808
> [ 576.948982][T31854] GPR28: c000001206b8d000 c0080000105d8227
> c00000129f56fa20
> 0000000000000001
> [ 576.949200][T31854] NIP [c008000010501bfc]
> get_orlov_stats+0x114/0x390 [ext4]
> get_orlov_stats at fs/ext4/ialloc.c:373 (discriminator 11)
> [ 576.949232][T31854] LR [c008000010501d94]
> get_orlov_stats+0x2ac/0x390 [ext4]
> [ 576.949243][T31854] Call Trace:
> [ 576.949260][T31854] [c00000129f56f990] [c008000010501d94]
> get_orlov_stats+0x2ac/0x390 [ext4] (unreliable)
> get_orlov_stats at fs/ext4/ialloc.c:373 (discriminator 11)
> [ 576.949301][T31854] [c00000129f56f9f0] [c00800001050231c]
> find_group_orlov+0x4a4/0x6b0 [ext4]
> find_group_orlov at fs/ext4/ialloc.c:467
> [ 576.949334][T31854] [c00000129f56fae0] [c0080000105055c8]
> __ext4_new_inode+0x1450/0x23c0 [ext4]
> [ 576.949367][T31854] [c00000129f56fc50] [c008000010547f2c]
> ext4_mkdir+0x104/0x590 [ext4]
> [ 576.949399][T31854] [c00000129f56fd60] [c0000000004cbc64]
> vfs_mkdir+0x114/0x210
> [ 576.949432][T31854] [c00000129f56fda0] [c0000000004d1a70]
> do_mkdirat+0xb0/0x1a0
> [ 576.949454][T31854] [c00000129f56fe20] [c00000000000b378]
> system_call+0x5c/0x68
> [ 576.949465][T31854] Instruction dump:
> [ 576.949473][T31854] 3c620000 e8638730 7f44d378 38630068 48078ccd
> e8410018
> 60000000 60000000
> [ 576.949497][T31854] 60000000 73490001 4182019c 7b091f24 <7f59482a>
> 4807a0d1
> e8410018 2fa30000
> [ 576.949522][T31854] ---[ end trace de4acb29e0d7791c ]---
> [ 577.200573][T31854]
> [ 578.200652][T31854] Kernel panic - not syncing: Fatal exception
> [ 579
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: null-ptr-deref due to "ext4: fix potential race between online resizing and write operations"
2020-02-21 19:58 ` Jitindar SIngh, Suraj
@ 2020-02-22 0:33 ` Theodore Y. Ts'o
0 siblings, 0 replies; 3+ messages in thread
From: Theodore Y. Ts'o @ 2020-02-22 0:33 UTC (permalink / raw)
To: Jitindar SIngh, Suraj
Cc: cai, adilger.kernel, linux-ext4, linux-kernel, paulmck
On Fri, Feb 21, 2020 at 07:58:01PM +0000, Jitindar SIngh, Suraj wrote:
> On Fri, 2020-02-21 at 09:02 -0500, Qian Cai wrote:
> > Reverted the linux-next commit c20bac9bf82c ("ext4: fix potential
> > race between
> > s_flex_groups online resizing and access") fixed the crash below
> > (with line
> > numbers),
>
> Good catch, this is a bug where the dereference of the array
> s_flex_groups needs to happen after the "if (flex_size > 1)" if
> statement in fs/ext4/ialloc.c:373
Cai, thanks for noting the problem! Suraj, I've fixed up the patch on
the ext4.git tree.
- Ted
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2020-02-22 0:34 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-02-21 14:02 null-ptr-deref due to "ext4: fix potential race between online resizing and write operations" Qian Cai
2020-02-21 19:58 ` Jitindar SIngh, Suraj
2020-02-22 0:33 ` Theodore Y. Ts'o
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.