possible deadlock in tcm_alloc

From: Kyungtae Kim
Date: Tue Nov 03 2020 - 15:01:15 EST

Next message: Lyude Paul: "Re: [PATCH] drm/edid: Fix uninitialized variable in drm_cvt_modes()"
Previous message: Mimi Zohar: "Re: [PATCH v8 1/4] powerpc: Refactor kexec functions to move arch independent code to drivers/of"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

We report a potential deadlock while setting up a tcm gadget as follows.

==================================================================
WARNING: possible circular locking dependency detected
5.8.13 #7 Not tainted
------------------------------------------------------
c1_fuzz/722 is trying to acquire lock:
ffffffff87cc2828 (tpg_instances_lock){+.+.}-{3:3}, at:
tcm_alloc+0x1f/0x2b0 drivers/usb/gadget/function/f_tcm.c:2300

but task is already holding lock:
ffff88806396b310 (&gi->lock){+.+.}-{3:3}, at:
config_usb_cfg_link+0x70/0x3a0 drivers/usb/gadget/configfs.c:413

which lock already depends on the new lock.

the existing dependency chain (in reverse order) is:

-> #3 (&gi->lock){+.+.}-{3:3}:
__mutex_lock_common kernel/locking/mutex.c:956 [inline]
__mutex_lock+0x14f/0x1380 kernel/locking/mutex.c:1103
mutex_lock_nested+0x16/0x20 kernel/locking/mutex.c:1118
config_usb_cfg_link+0x70/0x3a0 drivers/usb/gadget/configfs.c:413
configfs_symlink+0x4f5/0xf10 fs/configfs/symlink.c:202
vfs_symlink fs/namei.c:3953 [inline]
vfs_symlink+0x304/0x540 fs/namei.c:3939
do_symlinkat+0x104/0x1a0 fs/namei.c:3980
__do_sys_symlinkat fs/namei.c:3994 [inline]
__se_sys_symlinkat fs/namei.c:3991 [inline]
__x64_sys_symlinkat+0x6e/0xb0 fs/namei.c:3991
do_syscall_64+0x54/0xa0 arch/x86/entry/common.c:392
entry_SYSCALL_64_after_hwframe+0x44/0xa9

-> #2 (&sb->s_type->i_mutex_key#17){++++}-{3:3}:
down_write+0x8f/0x140 kernel/locking/rwsem.c:1531
inode_lock include/linux/fs.h:810 [inline]
configfs_depend_item_unlocked+0xba/0x470 fs/configfs/dir.c:1240
usbg_make_tpg+0x43a/0x5e0 drivers/usb/gadget/function/f_tcm.c:1373
target_fabric_make_tpg+0xa7/0x9d0
drivers/target/target_core_fabric_configfs.c:837
configfs_mkdir+0x3f8/0xb40 fs/configfs/dir.c:1344
vfs_mkdir+0x38f/0x640 fs/namei.c:3641
do_mkdirat+0x113/0x1e0 fs/namei.c:3664
__do_sys_mkdir fs/namei.c:3680 [inline]
__se_sys_mkdir fs/namei.c:3678 [inline]
__x64_sys_mkdir+0x57/0x80 fs/namei.c:3678
do_syscall_64+0x54/0xa0 arch/x86/entry/common.c:392
entry_SYSCALL_64_after_hwframe+0x44/0xa9

-> #1 (&opts->dep_lock){+.+.}-{3:3}:
__mutex_lock_common kernel/locking/mutex.c:956 [inline]
__mutex_lock+0x14f/0x1380 kernel/locking/mutex.c:1103
mutex_lock_nested+0x16/0x20 kernel/locking/mutex.c:1118
usbg_make_tpg+0x145/0x5e0 drivers/usb/gadget/function/f_tcm.c:1365
target_fabric_make_tpg+0xa7/0x9d0
drivers/target/target_core_fabric_configfs.c:837
configfs_mkdir+0x3f8/0xb40 fs/configfs/dir.c:1344
vfs_mkdir+0x38f/0x640 fs/namei.c:3641
do_mkdirat+0x113/0x1e0 fs/namei.c:3664
__do_sys_mkdir fs/namei.c:3680 [inline]
__se_sys_mkdir fs/namei.c:3678 [inline]
__x64_sys_mkdir+0x57/0x80 fs/namei.c:3678
do_syscall_64+0x54/0xa0 arch/x86/entry/common.c:392
entry_SYSCALL_64_after_hwframe+0x44/0xa9

-> #0 (tpg_instances_lock){+.+.}-{3:3}:
check_prev_add kernel/locking/lockdep.c:2496 [inline]
check_prevs_add kernel/locking/lockdep.c:2601 [inline]
validate_chain kernel/locking/lockdep.c:3218 [inline]
__lock_acquire+0x2fe7/0x5120 kernel/locking/lockdep.c:4395
lock_acquire+0x18d/0xb00 kernel/locking/lockdep.c:4982
__mutex_lock_common kernel/locking/mutex.c:956 [inline]
__mutex_lock+0x14f/0x1380 kernel/locking/mutex.c:1103
mutex_lock_nested+0x16/0x20 kernel/locking/mutex.c:1118
tcm_alloc+0x1f/0x2b0 drivers/usb/gadget/function/f_tcm.c:2300
usb_get_function+0x58/0xc0 drivers/usb/gadget/functions.c:61
config_usb_cfg_link+0x1ed/0x3a0 drivers/usb/gadget/configfs.c:435
configfs_symlink+0x4f5/0xf10 fs/configfs/symlink.c:202
vfs_symlink fs/namei.c:3953 [inline]
vfs_symlink+0x304/0x540 fs/namei.c:3939
do_symlinkat+0x104/0x1a0 fs/namei.c:3980
__do_sys_symlinkat fs/namei.c:3994 [inline]
__se_sys_symlinkat fs/namei.c:3991 [inline]
__x64_sys_symlinkat+0x6e/0xb0 fs/namei.c:3991
do_syscall_64+0x54/0xa0 arch/x86/entry/common.c:392
entry_SYSCALL_64_after_hwframe+0x44/0xa9

other info that might help us debug this:

Chain exists of:
tpg_instances_lock --> &sb->s_type->i_mutex_key#17 --> &gi->lock

Possible unsafe locking scenario:

CPU0 CPU1
---- ----
lock(&gi->lock);
lock(&sb->s_type->i_mutex_key#17);
lock(&gi->lock);
lock(tpg_instances_lock);

*** DEADLOCK ***

3 locks held by c1_fuzz/722:
#0: ffff8880662ac438 (sb_writers#15){.+.+}-{0:0}, at: sb_start_write
include/linux/fs.h:1675 [inline]
#0: ffff8880662ac438 (sb_writers#15){.+.+}-{0:0}, at:
mnt_want_write+0x3c/0xa0 fs/namespace.c:354
#1: ffff8880568afaf0 (&sb->s_type->i_mutex_key#17){++++}-{3:3}, at:
inode_lock include/linux/fs.h:810 [inline]
#1: ffff8880568afaf0 (&sb->s_type->i_mutex_key#17){++++}-{3:3}, at:
configfs_symlink+0x388/0xf10 fs/configfs/symlink.c:193
#2: ffff88806396b310 (&gi->lock){+.+.}-{3:3}, at:
config_usb_cfg_link+0x70/0x3a0 drivers/usb/gadget/configfs.c:413

stack backtrace:
CPU: 2 PID: 722 Comm: c1_fuzz Not tainted 5.8.13 #7
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Bochs 01/01/2011
Call Trace:
__dump_stack lib/dump_stack.c:77 [inline]
dump_stack+0xa7/0xea lib/dump_stack.c:118
print_circular_bug.cold.46+0x15a/0x169 kernel/locking/lockdep.c:1703
check_noncircular+0x327/0x3e0 kernel/locking/lockdep.c:1827
check_prev_add kernel/locking/lockdep.c:2496 [inline]
check_prevs_add kernel/locking/lockdep.c:2601 [inline]
validate_chain kernel/locking/lockdep.c:3218 [inline]
__lock_acquire+0x2fe7/0x5120 kernel/locking/lockdep.c:4395
lock_acquire+0x18d/0xb00 kernel/locking/lockdep.c:4982
__mutex_lock_common kernel/locking/mutex.c:956 [inline]
__mutex_lock+0x14f/0x1380 kernel/locking/mutex.c:1103
mutex_lock_nested+0x16/0x20 kernel/locking/mutex.c:1118
tcm_alloc+0x1f/0x2b0 drivers/usb/gadget/function/f_tcm.c:2300
usb_get_function+0x58/0xc0 drivers/usb/gadget/functions.c:61
config_usb_cfg_link+0x1ed/0x3a0 drivers/usb/gadget/configfs.c:435
configfs_symlink+0x4f5/0xf10 fs/configfs/symlink.c:202
vfs_symlink fs/namei.c:3953 [inline]
vfs_symlink+0x304/0x540 fs/namei.c:3939
do_symlinkat+0x104/0x1a0 fs/namei.c:3980
__do_sys_symlinkat fs/namei.c:3994 [inline]
__se_sys_symlinkat fs/namei.c:3991 [inline]
__x64_sys_symlinkat+0x6e/0xb0 fs/namei.c:3991
do_syscall_64+0x54/0xa0 arch/x86/entry/common.c:392
entry_SYSCALL_64_after_hwframe+0x44/0xa9
RIP: 0033:0x439d2a
Code: ff 4c 8b 7c 24 30 64 c7 03 22 00 00 00 41 bc 22 00 00 00 e9 1a
fd ff ff 45 31 ed e9 4d fe ff ff 48 63 f6 b8 0a 01 00 00 0f 05 <48> 3d
00 f0 ff ff 77 02 f3 c3 48 c7 c2 c0 ff ff ff f7 d8 64 89 02
RSP: 002b:00007ffc01673278 EFLAGS: 00000206 ORIG_RAX: 000000000000010a
RAX: ffffffffffffffda RBX: 00000000004002b0 RCX: 0000000000439d2a
RDX: 00007ffc016732e0 RSI: 0000000000000004 RDI: 00007ffc016732a0
RBP: 00007ffc01673320 R08: 0000000000000000 R09: 00007ffc01674ee1
R10: 0000000000000005 R11: 0000000000000206 R12: 0000000000000000
R13: 0000000000403200 R14: 0000000000403290 R15: 0000000000000000
==================================================================

Regards,
Kyungtae Kim

Next message: Lyude Paul: "Re: [PATCH] drm/edid: Fix uninitialized variable in drm_cvt_modes()"
Previous message: Mimi Zohar: "Re: [PATCH v8 1/4] powerpc: Refactor kexec functions to move arch independent code to drivers/of"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]