Skip to content

Instantly share code, notes, and snippets.

@xrivendell7
Created March 21, 2024 08:32
Show Gist options
  • Star 0 You must be signed in to star a gist
  • Fork 0 You must be signed in to fork a gist
  • Save xrivendell7/f829da1471add4307c9a435d88de5ecc to your computer and use it in GitHub Desktop.
Save xrivendell7/f829da1471add4307c9a435d88de5ecc to your computer and use it in GitHub Desktop.
TITLE: possible deadlock in sock_hash_delete_elem
CORRUPTED: false ()
MAINTAINERS (TO): [davem@davemloft.net edumazet@google.com kuba@kernel.org linux-kernel@vger.kernel.org pabeni@redhat.com]
MAINTAINERS (CC): [bpf@vger.kernel.org jakub@cloudflare.com john.fastabend@gmail.com netdev@vger.kernel.org]
WARNING: possible circular locking dependency detected
6.8.0 #1 Not tainted
------------------------------------------------------
80aaaa/8172 is trying to acquire lock:
ffff88814e959c20 (&htab->buckets[i].lock
){+...}-{2:2}, at: spin_lock_bh include/linux/spinlock.h:356 [inline]
){+...}-{2:2}, at: sock_hash_delete_elem+0xb5/0x320 net/core/sock_map.c:939
but task is already holding lock:
ffff8880b952c8d8 (
hrtimer_bases.lock
){-.-.}-{2:2}
, at: lock_hrtimer_base kernel/time/hrtimer.c:175 [inline]
, at: hrtimer_start_range_ns+0xea/0xc90 kernel/time/hrtimer.c:1303
which lock already depends on the new lock.
the existing dependency chain (in reverse order) is:
-> #2 (
hrtimer_bases.lock
){-.-.}-{2:2}
:
lock_acquire+0x1e4/0x530 kernel/locking/lockdep.c:5754
__raw_spin_lock_irqsave include/linux/spinlock_api_smp.h:110 [inline]
_raw_spin_lock_irqsave+0xd5/0x120 kernel/locking/spinlock.c:162
lock_hrtimer_base kernel/time/hrtimer.c:175 [inline]
hrtimer_start_range_ns+0xea/0xc90 kernel/time/hrtimer.c:1303
hrtimer_start include/linux/hrtimer.h:275 [inline]
run_page_cache_worker kernel/rcu/tree.c:3341 [inline]
kvfree_call_rcu+0x5e6/0x790 kernel/rcu/tree.c:3446
rtnl_register_internal+0x4ad/0x5d0 net/core/rtnetlink.c:265
rtnl_register+0x3b/0x80 net/core/rtnetlink.c:315
ip_rt_init+0x302/0x3b0 net/ipv4/route.c:3719
ip_init+0x13/0x20 net/ipv4/ip_output.c:1664
inet_init+0x3f1/0x5a0 net/ipv4/af_inet.c:2022
do_one_initcall+0x250/0x880 init/main.c:1241
do_initcall_level+0x164/0x220 init/main.c:1303
do_initcalls+0x48/0x90 init/main.c:1319
kernel_init_freeable+0x455/0x5f0 init/main.c:1557
kernel_init+0x22/0x2c0 init/main.c:1446
ret_from_fork+0x52/0x80 arch/x86/kernel/process.c:147
ret_from_fork_asm+0x1a/0x30 arch/x86/entry/entry_64.S:243
-> #1
(
krc.lock
){..-.}-{2:2}
:
lock_acquire+0x1e4/0x530 kernel/locking/lockdep.c:5754
__raw_spin_lock include/linux/spinlock_api_smp.h:133 [inline]
_raw_spin_lock+0x2e/0x40 kernel/locking/spinlock.c:154
krc_this_cpu_lock kernel/rcu/tree.c:2960 [inline]
add_ptr_to_bulk_krc_lock kernel/rcu/tree.c:3359 [inline]
kvfree_call_rcu+0x18a/0x790 kernel/rcu/tree.c:3444
sock_hash_free_elem net/core/sock_map.c:900 [inline]
sock_hash_delete_from_link net/core/sock_map.c:923 [inline]
sock_map_unlink net/core/sock_map.c:1566 [inline]
sock_map_remove_links+0x497/0x580 net/core/sock_map.c:1578
sock_map_close+0x12b/0x2f0 net/core/sock_map.c:1643
inet_release+0x191/0x210 net/ipv4/af_inet.c:437
__sock_release net/socket.c:659 [inline]
sock_close+0xc3/0x250 net/socket.c:1421
__fput+0x442/0x8d0 fs/file_table.c:423
task_work_run+0x25c/0x320 kernel/task_work.c:180
exit_task_work include/linux/task_work.h:38 [inline]
do_exit+0xa46/0x28a0 kernel/exit.c:878
do_group_exit+0x20b/0x2c0 kernel/exit.c:1027
get_signal+0x17ef/0x18d0 kernel/signal.c:2907
arch_do_signal_or_restart+0x9b/0x890 arch/x86/kernel/signal.c:310
exit_to_user_mode_loop kernel/entry/common.c:105 [inline]
exit_to_user_mode_prepare include/linux/entry-common.h:328 [inline]
__syscall_exit_to_user_mode_work kernel/entry/common.c:201 [inline]
syscall_exit_to_user_mode+0xc9/0x360 kernel/entry/common.c:212
do_syscall_64+0x10a/0x240 arch/x86/entry/common.c:89
entry_SYSCALL_64_after_hwframe+0x6d/0x75
-> #0
(
&htab->buckets[i].lock
){+...}-{2:2}
:
check_prev_add kernel/locking/lockdep.c:3134 [inline]
check_prevs_add kernel/locking/lockdep.c:3253 [inline]
validate_chain+0x18cb/0x58e0 kernel/locking/lockdep.c:3869
__lock_acquire+0x1346/0x1fd0 kernel/locking/lockdep.c:5137
lock_acquire+0x1e4/0x530 kernel/locking/lockdep.c:5754
__raw_spin_lock_bh include/linux/spinlock_api_smp.h:126 [inline]
_raw_spin_lock_bh+0x35/0x50 kernel/locking/spinlock.c:178
spin_lock_bh include/linux/spinlock.h:356 [inline]
sock_hash_delete_elem+0xb5/0x320 net/core/sock_map.c:939
bpf_prog_2c29ac5cdc6b1842+0x42/0x46
bpf_dispatcher_nop_func include/linux/bpf.h:1234 [inline]
__bpf_prog_run include/linux/filter.h:657 [inline]
bpf_prog_run include/linux/filter.h:664 [inline]
__bpf_trace_run kernel/trace/bpf_trace.c:2381 [inline]
bpf_trace_run2+0x227/0x470 kernel/trace/bpf_trace.c:2420
trace_hrtimer_start include/trace/events/timer.h:222 [inline]
debug_activate kernel/time/hrtimer.c:479 [inline]
enqueue_hrtimer+0x34a/0x3b0 kernel/time/hrtimer.c:1090
__hrtimer_start_range_ns kernel/time/hrtimer.c:1265 [inline]
hrtimer_start_range_ns+0xad5/0xc90 kernel/time/hrtimer.c:1305
hrtimer_start_expires include/linux/hrtimer.h:289 [inline]
hrtimer_sleeper_start_expires kernel/time/hrtimer.c:1972 [inline]
do_nanosleep+0x15d/0x620 kernel/time/hrtimer.c:2048
hrtimer_nanosleep+0x22c/0x470 kernel/time/hrtimer.c:2104
__do_sys_clock_nanosleep kernel/time/posix-timers.c:1396 [inline]
__se_sys_clock_nanosleep+0x33e/0x3d0 kernel/time/posix-timers.c:1373
do_syscall_64+0xfb/0x240
entry_SYSCALL_64_after_hwframe+0x6d/0x75
other info that might help us debug this:
Chain exists of:
&htab->buckets[i].lock --> krc.lock --> hrtimer_bases.lock
Possible unsafe locking scenario:
CPU0 CPU1
---- ----
lock(hrtimer_bases.lock);
lock(krc.lock);
lock(hrtimer_bases.lock);
lock(&htab->buckets[i].lock);
*** DEADLOCK ***
2 locks held by 80aaaa/8172:
#0: ffff8880b952c8d8 (hrtimer_bases.lock){-.-.}-{2:2}, at: lock_hrtimer_base kernel/time/hrtimer.c:175 [inline]
#0: ffff8880b952c8d8 (hrtimer_bases.lock){-.-.}-{2:2}, at: hrtimer_start_range_ns+0xea/0xc90 kernel/time/hrtimer.c:1303
#1: ffffffff8e531920 (rcu_read_lock){....}-{1:2}, at: rcu_lock_acquire include/linux/rcupdate.h:298 [inline]
#1: ffffffff8e531920 (rcu_read_lock){....}-{1:2}, at: rcu_read_lock include/linux/rcupdate.h:750 [inline]
#1: ffffffff8e531920 (rcu_read_lock){....}-{1:2}, at: __bpf_trace_run kernel/trace/bpf_trace.c:2380 [inline]
#1: ffffffff8e531920 (rcu_read_lock){....}-{1:2}, at: bpf_trace_run2+0x119/0x470 kernel/trace/bpf_trace.c:2420
stack backtrace:
CPU: 2 PID: 8172 Comm: 80aaaa Not tainted 6.8.0 #1
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.16.2-debian-1.16.2-1 04/01/2014
Call Trace:
<TASK>
__dump_stack lib/dump_stack.c:88 [inline]
dump_stack_lvl+0x1f4/0x2f0 lib/dump_stack.c:106
check_noncircular+0x36a/0x4a0 kernel/locking/lockdep.c:2187
check_prev_add kernel/locking/lockdep.c:3134 [inline]
check_prevs_add kernel/locking/lockdep.c:3253 [inline]
validate_chain+0x18cb/0x58e0 kernel/locking/lockdep.c:3869
__lock_acquire+0x1346/0x1fd0 kernel/locking/lockdep.c:5137
lock_acquire+0x1e4/0x530 kernel/locking/lockdep.c:5754
__raw_spin_lock_bh include/linux/spinlock_api_smp.h:126 [inline]
_raw_spin_lock_bh+0x35/0x50 kernel/locking/spinlock.c:178
spin_lock_bh include/linux/spinlock.h:356 [inline]
sock_hash_delete_elem+0xb5/0x320 net/core/sock_map.c:939
bpf_prog_2c29ac5cdc6b1842+0x42/0x46
bpf_dispatcher_nop_func include/linux/bpf.h:1234 [inline]
__bpf_prog_run include/linux/filter.h:657 [inline]
bpf_prog_run include/linux/filter.h:664 [inline]
__bpf_trace_run kernel/trace/bpf_trace.c:2381 [inline]
bpf_trace_run2+0x227/0x470 kernel/trace/bpf_trace.c:2420
trace_hrtimer_start include/trace/events/timer.h:222 [inline]
debug_activate kernel/time/hrtimer.c:479 [inline]
enqueue_hrtimer+0x34a/0x3b0 kernel/time/hrtimer.c:1090
__hrtimer_start_range_ns kernel/time/hrtimer.c:1265 [inline]
hrtimer_start_range_ns+0xad5/0xc90 kernel/time/hrtimer.c:1305
hrtimer_start_expires include/linux/hrtimer.h:289 [inline]
hrtimer_sleeper_start_expires kernel/time/hrtimer.c:1972 [inline]
do_nanosleep+0x15d/0x620 kernel/time/hrtimer.c:2048
hrtimer_nanosleep+0x22c/0x470 kernel/time/hrtimer.c:2104
__do_sys_clock_nanosleep kernel/time/posix-timers.c:1396 [inline]
__se_sys_clock_nanosleep+0x33e/0x3d0 kernel/time/posix-timers.c:1373
do_syscall_64+0xfb/0x240
entry_SYSCALL_64_after_hwframe+0x6d/0x75
RIP: 0033:0x470d35
Code: 24 0c 89 3c 24 48 89 4c 24 18 e8 f6 92 fe ff 4c 8b 54 24 18 48 8b 54 24 10 41 89 c0 8b 74 24 08
RSP: 002b:00007ffc3469c0a0 EFLAGS: 00000293 ORIG_RAX: 00000000000000e6
RAX: ffffffffffffffda RBX: 00007ffc3469c338 RCX: 0000000000470d35
RDX: 00007ffc3469c0e0 RSI: 0000000000000000 RDI: 0000000000000000
RBP: 00007ffc3469c120 R08: 0000000000000000 R09: 0000000000000226
R10: 0000000000000000 R11: 0000000000000293 R12: 0000000000000001
R13: 00007ffc3469c328 R14: 0000000000000001 R15: 0000000000000001
</TASK>
TITLE: kernel panic: kernel: panic_on_warn set ...
CORRUPTED: false ()
MAINTAINERS (TO): [davem@davemloft.net edumazet@google.com kuba@kernel.org linux-kernel@vger.kernel.org pabeni@redhat.com]
MAINTAINERS (CC): [bpf@vger.kernel.org jakub@cloudflare.com john.fastabend@gmail.com netdev@vger.kernel.org]
cpuidle_idle_call kernel/sched/idle.c:191 [inline]
do_idle+0x22f/0x5d0 kernel/sched/idle.c:332
cpu_startup_entry+0x42/0x60 kernel/sched/idle.c:430
start_secondary+0x100/0x100 arch/x86/kernel/smpboot.c:313
common_startup_64+0x13e/0x147
</TASK>
Kernel panic - not syncing: kernel: panic_on_warn set ...
CPU: 1 PID: 0 Comm: swapper/1 Not tainted 6.8.0 #1
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.16.2-debian-1.16.2-1 04/01/2014
Call Trace:
<IRQ>
__dump_stack lib/dump_stack.c:88 [inline]
dump_stack_lvl+0x1f4/0x2f0 lib/dump_stack.c:106
panic+0x35a/0x890 kernel/panic.c:344
__warn+0x33c/0x4d0
__report_bug lib/bug.c:199 [inline]
report_bug+0x2ca/0x520 lib/bug.c:219
handle_bug+0x3e/0x70 arch/x86/kernel/traps.c:239
exc_invalid_op+0x1a/0x50 arch/x86/kernel/traps.c:260
asm_exc_invalid_op+0x1a/0x20 arch/x86/include/asm/idtentry.h:621
RIP: 0010:__local_bh_disable_ip+0x1f6/0x220 kernel/softirq.c:307
Code: 2c 0b 00 65 48 8b 04 25 28 00 00 00 48 3b 84 24 80 00 00 00 75 33 48 8d 65 d8 5b 41 5c 41 5d 45
RSP: 0018:ffffc90000a28b00 EFLAGS: 00010006
RAX: 0000000080010003 RBX: ffffffff898a95d5 RCX: 0000000080010003
RDX: dffffc0000000000 RSI: 0000000000000201 RDI: ffffffff898a95d5
RBP: ffffc90000a28bc8 R08: 0000000000000005 R09: ffffffff898adbb7
R10: 000000000000000c R11: ffff888140ee0040 R12: dffffc0000000000
R13: 1ffff92000145164 R14: ffffc90000a28b60 R15: 0000000000000201
__raw_spin_lock_bh include/linux/spinlock_api_smp.h:125 [inline]
_raw_spin_lock_bh+0x1c/0x50 kernel/locking/spinlock.c:178
spin_lock_bh include/linux/spinlock.h:356 [inline]
sock_hash_delete_elem+0xb5/0x320 net/core/sock_map.c:939
bpf_prog_2c29ac5cdc6b1842+0x42/0x46
bpf_dispatcher_nop_func include/linux/bpf.h:1234 [inline]
__bpf_prog_run include/linux/filter.h:657 [inline]
bpf_prog_run include/linux/filter.h:664 [inline]
__bpf_trace_run kernel/trace/bpf_trace.c:2381 [inline]
bpf_trace_run2+0x227/0x470 kernel/trace/bpf_trace.c:2420
trace_hrtimer_start include/trace/events/timer.h:222 [inline]
debug_activate kernel/time/hrtimer.c:479 [inline]
enqueue_hrtimer+0x34a/0x3b0 kernel/time/hrtimer.c:1090
__run_hrtimer kernel/time/hrtimer.c:1709 [inline]
__hrtimer_run_queues+0x6dc/0xd50 kernel/time/hrtimer.c:1756
hrtimer_interrupt+0x38a/0x990 kernel/time/hrtimer.c:1818
local_apic_timer_interrupt arch/x86/kernel/apic/apic.c:1032 [inline]
__sysvec_apic_timer_interrupt+0x107/0x3a0 arch/x86/kernel/apic/apic.c:1049
instr_sysvec_apic_timer_interrupt arch/x86/kernel/apic/apic.c:1043 [inline]
sysvec_apic_timer_interrupt+0xa1/0xc0 arch/x86/kernel/apic/apic.c:1043
</IRQ>
<TASK>
asm_sysvec_apic_timer_interrupt+0x1a/0x20 arch/x86/include/asm/idtentry.h:702
RIP: 0010:native_irq_disable arch/x86/include/asm/irqflags.h:37 [inline]
RIP: 0010:arch_local_irq_disable arch/x86/include/asm/irqflags.h:72 [inline]
RIP: 0010:default_idle+0x13/0x20 arch/x86/kernel/process.c:743
Code: 29 c2 e9 72 ff ff ff 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 f3 0f 1e fa 66 90 0f 00 20
RSP: 0018:ffffc90000197dc8 EFLAGS: 000002c6
RAX: 5572eec3e4b0d000 RBX: ffffffff816e07af RCX: 000000000001cae9
RDX: 0000000000000001 RSI: ffffffff8beac260 RDI: ffffffff8c3ec620
RBP: ffffc90000197f20 R08: ffff88823bc37d0b R09: 1ffff11047786fa1
R10: dffffc0000000000 R11: ffffed1047786fa2 R12: 1ffff92000032fd2
R13: 1ffff110281dc008 R14: 0000000000000001 R15: dffffc0000000000
default_idle_call+0x74/0xb0 kernel/sched/idle.c:117
cpuidle_idle_call kernel/sched/idle.c:191 [inline]
do_idle+0x22f/0x5d0 kernel/sched/idle.c:332
cpu_startup_entry+0x42/0x60 kernel/sched/idle.c:430
start_secondary+0x100/0x100 arch/x86/kernel/smpboot.c:313
common_startup_64+0x13e/0x147
</TASK>
TITLE: WARNING in sock_hash_delete_elem
CORRUPTED: false ()
MAINTAINERS (TO): [davem@davemloft.net edumazet@google.com kuba@kernel.org linux-kernel@vger.kernel.org pabeni@redhat.com]
MAINTAINERS (CC): [bpf@vger.kernel.org jakub@cloudflare.com john.fastabend@gmail.com netdev@vger.kernel.org]
------------[ cut here ]------------
WARNING: CPU: 1 PID: 0 at kernel/softirq.c:307 __local_bh_disable_ip+0x1f6/0x220 kernel/softirq.c:307
Modules linked in:
CPU: 1 PID: 0 Comm: swapper/1 Not tainted 6.8.0 #1
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.16.2-debian-1.16.2-1 04/01/2014
RIP: 0010:__local_bh_disable_ip+0x1f6/0x220 kernel/softirq.c:307
Code: 2c 0b 00 65 48 8b 04 25 28 00 00 00 48 3b 84 24 80 00 00 00 75 33 48 8d 65 d8 5b 41 5c 41 5d 45
RSP: 0018:ffffc90000a28b00 EFLAGS: 00010006
RAX: 0000000080010003 RBX: ffffffff898a95d5 RCX: 0000000080010003
RDX: dffffc0000000000 RSI: 0000000000000201 RDI: ffffffff898a95d5
RBP: ffffc90000a28bc8 R08: 0000000000000005 R09: ffffffff898adbb7
R10: 000000000000000c R11: ffff888140ee0040 R12: dffffc0000000000
R13: 1ffff92000145164 R14: ffffc90000a28b60 R15: 0000000000000201
FS: 0000000000000000(0000) GS:ffff88823bc00000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00005624dad956b0 CR3: 0000000029fd0000 CR4: 0000000000750ef0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
PKRU: 55555554
Call Trace:
<IRQ>
__raw_spin_lock_bh include/linux/spinlock_api_smp.h:125 [inline]
_raw_spin_lock_bh+0x1c/0x50 kernel/locking/spinlock.c:178
spin_lock_bh include/linux/spinlock.h:356 [inline]
sock_hash_delete_elem+0xb5/0x320 net/core/sock_map.c:939
bpf_prog_2c29ac5cdc6b1842+0x42/0x46
bpf_dispatcher_nop_func include/linux/bpf.h:1234 [inline]
__bpf_prog_run include/linux/filter.h:657 [inline]
bpf_prog_run include/linux/filter.h:664 [inline]
__bpf_trace_run kernel/trace/bpf_trace.c:2381 [inline]
bpf_trace_run2+0x227/0x470 kernel/trace/bpf_trace.c:2420
trace_hrtimer_start include/trace/events/timer.h:222 [inline]
debug_activate kernel/time/hrtimer.c:479 [inline]
enqueue_hrtimer+0x34a/0x3b0 kernel/time/hrtimer.c:1090
__run_hrtimer kernel/time/hrtimer.c:1709 [inline]
__hrtimer_run_queues+0x6dc/0xd50 kernel/time/hrtimer.c:1756
hrtimer_interrupt+0x38a/0x990 kernel/time/hrtimer.c:1818
local_apic_timer_interrupt arch/x86/kernel/apic/apic.c:1032 [inline]
__sysvec_apic_timer_interrupt+0x107/0x3a0 arch/x86/kernel/apic/apic.c:1049
instr_sysvec_apic_timer_interrupt arch/x86/kernel/apic/apic.c:1043 [inline]
sysvec_apic_timer_interrupt+0xa1/0xc0 arch/x86/kernel/apic/apic.c:1043
</IRQ>
<TASK>
asm_sysvec_apic_timer_interrupt+0x1a/0x20 arch/x86/include/asm/idtentry.h:702
RIP: 0010:native_irq_disable arch/x86/include/asm/irqflags.h:37 [inline]
RIP: 0010:arch_local_irq_disable arch/x86/include/asm/irqflags.h:72 [inline]
RIP: 0010:default_idle+0x13/0x20 arch/x86/kernel/process.c:743
Code: 29 c2 e9 72 ff ff ff 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 f3 0f 1e fa 66 90 0f 00 20
RSP: 0018:ffffc90000197dc8 EFLAGS: 000002c6
RAX: 5572eec3e4b0d000 RBX: ffffffff816e07af RCX: 000000000001cae9
RDX: 0000000000000001 RSI: ffffffff8beac260 RDI: ffffffff8c3ec620
RBP: ffffc90000197f20 R08: ffff88823bc37d0b R09: 1ffff11047786fa1
R10: dffffc0000000000 R11: ffffed1047786fa2 R12: 1ffff92000032fd2
R13: 1ffff110281dc008 R14: 0000000000000001 R15: dffffc0000000000
default_idle_call+0x74/0xb0 kernel/sched/idle.c:117
cpuidle_idle_call kernel/sched/idle.c:191 [inline]
do_idle+0x22f/0x5d0 kernel/sched/idle.c:332
cpu_startup_entry+0x42/0x60 kernel/sched/idle.c:430
start_secondary+0x100/0x100 arch/x86/kernel/smpboot.c:313
common_startup_64+0x13e/0x147
</TASK>
#define _GNU_SOURCE
#include <endian.h>
#include <errno.h>
#include <pthread.h>
#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sys/syscall.h>
#include <sys/types.h>
#include <time.h>
#include <unistd.h>
#include <linux/futex.h>
#ifndef __NR_bpf
#define __NR_bpf 321
#endif
static void sleep_ms(uint64_t ms) {
usleep(ms * 1000);
}
static uint64_t current_time_ms(void) {
struct timespec ts;
if (clock_gettime(CLOCK_MONOTONIC, &ts))
exit(1);
return (uint64_t)ts.tv_sec * 1000 + (uint64_t)ts.tv_nsec / 1000000;
}
static void thread_start(void* (*fn)(void*), void* arg) {
pthread_t th;
pthread_attr_t attr;
pthread_attr_init(&attr);
pthread_attr_setstacksize(&attr, 128 << 10);
int i = 0;
for (; i < 100; i++) {
if (pthread_create(&th, &attr, fn, arg) == 0) {
pthread_attr_destroy(&attr);
return;
}
if (errno == EAGAIN) {
usleep(50);
continue;
}
break;
}
exit(1);
}
#define BITMASK(bf_off, bf_len) (((1ull << (bf_len)) - 1) << (bf_off))
#define STORE_BY_BITMASK(type, htobe, addr, val, bf_off, bf_len) \
*(type*)(addr) = \
htobe((htobe(*(type*)(addr)) & ~BITMASK((bf_off), (bf_len))) | \
(((type)(val) << (bf_off)) & BITMASK((bf_off), (bf_len))))
typedef struct {
int state;
} event_t;
static void event_init(event_t* ev) {
ev->state = 0;
}
static void event_reset(event_t* ev) {
ev->state = 0;
}
static void event_set(event_t* ev) {
if (ev->state)
exit(1);
__atomic_store_n(&ev->state, 1, __ATOMIC_RELEASE);
syscall(SYS_futex, &ev->state, FUTEX_WAKE | FUTEX_PRIVATE_FLAG, 1000000);
}
static void event_wait(event_t* ev) {
while (!__atomic_load_n(&ev->state, __ATOMIC_ACQUIRE))
syscall(SYS_futex, &ev->state, FUTEX_WAIT | FUTEX_PRIVATE_FLAG, 0, 0);
}
static int event_isset(event_t* ev) {
return __atomic_load_n(&ev->state, __ATOMIC_ACQUIRE);
}
static int event_timedwait(event_t* ev, uint64_t timeout) {
uint64_t start = current_time_ms();
uint64_t now = start;
for (;;) {
uint64_t remain = timeout - (now - start);
struct timespec ts;
ts.tv_sec = remain / 1000;
ts.tv_nsec = (remain % 1000) * 1000 * 1000;
syscall(SYS_futex, &ev->state, FUTEX_WAIT | FUTEX_PRIVATE_FLAG, 0, &ts);
if (__atomic_load_n(&ev->state, __ATOMIC_ACQUIRE))
return 1;
now = current_time_ms();
if (now - start > timeout)
return 0;
}
}
struct thread_t {
int created, call;
event_t ready, done;
};
static struct thread_t threads[16];
static void execute_call(int call);
static int running;
static void* thr(void* arg) {
struct thread_t* th = (struct thread_t*)arg;
for (;;) {
event_wait(&th->ready);
event_reset(&th->ready);
execute_call(th->call);
__atomic_fetch_sub(&running, 1, __ATOMIC_RELAXED);
event_set(&th->done);
}
return 0;
}
static void loop(void) {
int i, call, thread;
for (call = 0; call < 11; call++) {
for (thread = 0; thread < (int)(sizeof(threads) / sizeof(threads[0]));
thread++) {
struct thread_t* th = &threads[thread];
if (!th->created) {
th->created = 1;
event_init(&th->ready);
event_init(&th->done);
event_set(&th->done);
thread_start(thr, th);
}
if (!event_isset(&th->done))
continue;
event_reset(&th->done);
th->call = call;
__atomic_fetch_add(&running, 1, __ATOMIC_RELAXED);
event_set(&th->ready);
event_timedwait(&th->done, 50 + (call == 0 ? 500 : 0) +
(call == 5 ? 500 : 0) +
(call == 10 ? 500 : 0));
break;
}
}
for (i = 0; i < 100 && __atomic_load_n(&running, __ATOMIC_RELAXED); i++)
sleep_ms(1);
}
uint64_t r[5] = {0xffffffffffffffff, 0xffffffffffffffff, 0xffffffffffffffff,
0xffffffffffffffff, 0xffffffffffffffff};
void execute_call(int call) {
intptr_t res = 0;
switch (call) {
case 0:
*(uint32_t*)0x20002c80 = 3;
*(uint32_t*)0x20002c84 = 5;
*(uint64_t*)0x20002c88 = 0x20000340;
sprintf((char*)0x20000340, "%020llu", (long long)0);
*(uint64_t*)0x20002c90 = 0x20000080;
memcpy((void*)0x20000080, "GPL\000", 4);
*(uint32_t*)0x20002c98 = 4;
*(uint32_t*)0x20002c9c = 0;
*(uint64_t*)0x20002ca0 = 0;
*(uint32_t*)0x20002ca8 = 0;
*(uint32_t*)0x20002cac = 0;
memset((void*)0x20002cb0, 0, 16);
*(uint32_t*)0x20002cc0 = 0;
*(uint32_t*)0x20002cc4 = 0;
*(uint32_t*)0x20002cc8 = -1;
*(uint32_t*)0x20002ccc = 8;
*(uint64_t*)0x20002cd0 = 0;
*(uint32_t*)0x20002cd8 = 0;
*(uint32_t*)0x20002cdc = 0x10;
*(uint64_t*)0x20002ce0 = 0;
*(uint32_t*)0x20002ce8 = 0;
*(uint32_t*)0x20002cec = 0;
*(uint32_t*)0x20002cf0 = 0;
*(uint32_t*)0x20002cf4 = 0;
*(uint64_t*)0x20002cf8 = 0;
*(uint64_t*)0x20002d00 = 0;
*(uint32_t*)0x20002d08 = 0x10;
*(uint32_t*)0x20002d0c = 0;
res =
syscall(__NR_bpf, /*cmd=*/5ul, /*arg=*/0x20002c80ul, /*size=*/0x90ul);
if (res != -1)
r[0] = res;
break;
case 1:
*(uint32_t*)0x200009c0 = 0x12;
*(uint32_t*)0x200009c4 = 4;
*(uint32_t*)0x200009c8 = 8;
*(uint32_t*)0x200009cc = 8;
*(uint32_t*)0x200009d0 = 0;
*(uint32_t*)0x200009d4 = -1;
*(uint32_t*)0x200009d8 = 0;
memset((void*)0x200009dc, 0, 16);
*(uint32_t*)0x200009ec = 0;
*(uint32_t*)0x200009f0 = -1;
*(uint32_t*)0x200009f4 = 0;
*(uint32_t*)0x200009f8 = 0;
*(uint32_t*)0x200009fc = 0;
*(uint64_t*)0x20000a00 = 0;
res =
syscall(__NR_bpf, /*cmd=*/0ul, /*arg=*/0x200009c0ul, /*size=*/0x48ul);
if (res != -1)
r[1] = res;
break;
case 2:
res = syscall(__NR_socket, /*domain=*/0xaul, /*type=*/2ul, /*proto=*/0);
if (res != -1)
r[2] = res;
break;
case 3:
*(uint32_t*)0x20000080 = r[1];
*(uint64_t*)0x20000088 = 0x20000a40;
memcpy((void*)0x20000a40, "\x90\xb4", 2);
*(uint64_t*)0x20000090 = 0x20000040;
*(uint32_t*)0x20000040 = r[2];
*(uint64_t*)0x20000098 = 0;
syscall(__NR_bpf, /*cmd=*/2ul, /*arg=*/0x20000080ul, /*size=*/0x20ul);
break;
case 4:
*(uint32_t*)0x200000c0 = 0x11;
*(uint32_t*)0x200000c4 = 0xc;
*(uint64_t*)0x200000c8 = 0x20000440;
*(uint8_t*)0x20000440 = 0x18;
STORE_BY_BITMASK(uint8_t, , 0x20000441, 0, 0, 4);
STORE_BY_BITMASK(uint8_t, , 0x20000441, 0, 4, 4);
*(uint16_t*)0x20000442 = 0;
*(uint32_t*)0x20000444 = 0;
*(uint8_t*)0x20000448 = 0;
*(uint8_t*)0x20000449 = 0;
*(uint16_t*)0x2000044a = 0;
*(uint32_t*)0x2000044c = 0;
*(uint8_t*)0x20000450 = 0x18;
STORE_BY_BITMASK(uint8_t, , 0x20000451, 1, 0, 4);
STORE_BY_BITMASK(uint8_t, , 0x20000451, 1, 4, 4);
*(uint16_t*)0x20000452 = 0;
*(uint32_t*)0x20000454 = r[1];
*(uint8_t*)0x20000458 = 0;
*(uint8_t*)0x20000459 = 0;
*(uint16_t*)0x2000045a = 0;
*(uint32_t*)0x2000045c = 0;
STORE_BY_BITMASK(uint8_t, , 0x20000460, 7, 0, 3);
STORE_BY_BITMASK(uint8_t, , 0x20000460, 0, 3, 1);
STORE_BY_BITMASK(uint8_t, , 0x20000460, 0xb, 4, 4);
STORE_BY_BITMASK(uint8_t, , 0x20000461, 8, 0, 4);
STORE_BY_BITMASK(uint8_t, , 0x20000461, 0, 4, 4);
*(uint16_t*)0x20000462 = 0;
*(uint32_t*)0x20000464 = 0;
STORE_BY_BITMASK(uint8_t, , 0x20000468, 3, 0, 3);
STORE_BY_BITMASK(uint8_t, , 0x20000468, 3, 3, 2);
STORE_BY_BITMASK(uint8_t, , 0x20000468, 3, 5, 3);
STORE_BY_BITMASK(uint8_t, , 0x20000469, 0xa, 0, 4);
STORE_BY_BITMASK(uint8_t, , 0x20000469, 8, 4, 4);
*(uint16_t*)0x2000046a = 0xfff8;
*(uint32_t*)0x2000046c = 0;
STORE_BY_BITMASK(uint8_t, , 0x20000470, 7, 0, 3);
STORE_BY_BITMASK(uint8_t, , 0x20000470, 1, 3, 1);
STORE_BY_BITMASK(uint8_t, , 0x20000470, 0xb, 4, 4);
STORE_BY_BITMASK(uint8_t, , 0x20000471, 2, 0, 4);
STORE_BY_BITMASK(uint8_t, , 0x20000471, 0xa, 4, 4);
*(uint16_t*)0x20000472 = 0;
*(uint32_t*)0x20000474 = 0;
STORE_BY_BITMASK(uint8_t, , 0x20000478, 7, 0, 3);
STORE_BY_BITMASK(uint8_t, , 0x20000478, 0, 3, 1);
STORE_BY_BITMASK(uint8_t, , 0x20000478, 0, 4, 4);
STORE_BY_BITMASK(uint8_t, , 0x20000479, 2, 0, 4);
STORE_BY_BITMASK(uint8_t, , 0x20000479, 0, 4, 4);
*(uint16_t*)0x2000047a = 0;
*(uint32_t*)0x2000047c = 0xfffffff8;
STORE_BY_BITMASK(uint8_t, , 0x20000480, 7, 0, 3);
STORE_BY_BITMASK(uint8_t, , 0x20000480, 0, 3, 1);
STORE_BY_BITMASK(uint8_t, , 0x20000480, 0xb, 4, 4);
STORE_BY_BITMASK(uint8_t, , 0x20000481, 3, 0, 4);
STORE_BY_BITMASK(uint8_t, , 0x20000481, 0, 4, 4);
*(uint16_t*)0x20000482 = 0;
*(uint32_t*)0x20000484 = 8;
STORE_BY_BITMASK(uint8_t, , 0x20000488, 7, 0, 3);
STORE_BY_BITMASK(uint8_t, , 0x20000488, 0, 3, 1);
STORE_BY_BITMASK(uint8_t, , 0x20000488, 0xb, 4, 4);
STORE_BY_BITMASK(uint8_t, , 0x20000489, 4, 0, 4);
STORE_BY_BITMASK(uint8_t, , 0x20000489, 0, 4, 4);
*(uint16_t*)0x2000048a = 0;
*(uint32_t*)0x2000048c = 0;
*(uint8_t*)0x20000490 = 0x85;
*(uint8_t*)0x20000491 = 0;
*(uint16_t*)0x20000492 = 0;
*(uint32_t*)0x20000494 = 3;
*(uint8_t*)0x20000498 = 0x95;
*(uint8_t*)0x20000499 = 0;
*(uint16_t*)0x2000049a = 0;
*(uint32_t*)0x2000049c = 0;
*(uint64_t*)0x200000d0 = 0x20000240;
memcpy((void*)0x20000240, "GPL\000", 4);
*(uint32_t*)0x200000d8 = 0;
*(uint32_t*)0x200000dc = 0;
*(uint64_t*)0x200000e0 = 0;
*(uint32_t*)0x200000e8 = 0;
*(uint32_t*)0x200000ec = 0;
memset((void*)0x200000f0, 0, 16);
*(uint32_t*)0x20000100 = 0;
*(uint32_t*)0x20000104 = 0;
*(uint32_t*)0x20000108 = -1;
*(uint32_t*)0x2000010c = 0;
*(uint64_t*)0x20000110 = 0;
*(uint32_t*)0x20000118 = 0;
*(uint32_t*)0x2000011c = 0;
*(uint64_t*)0x20000120 = 0;
*(uint32_t*)0x20000128 = 0;
*(uint32_t*)0x2000012c = 0;
*(uint32_t*)0x20000130 = 0;
*(uint32_t*)0x20000134 = 0;
*(uint64_t*)0x20000138 = 0;
*(uint64_t*)0x20000140 = 0;
*(uint32_t*)0x20000148 = 0;
*(uint32_t*)0x2000014c = 0;
res =
syscall(__NR_bpf, /*cmd=*/5ul, /*arg=*/0x200000c0ul, /*size=*/0x90ul);
if (res != -1)
r[3] = res;
break;
case 5:
*(uint64_t*)0x20000000 = 0x20000640;
memcpy((void*)0x20000640, "ext4_es_find_extent_range_enter\000", 32);
*(uint32_t*)0x20000008 = r[3];
syscall(__NR_bpf, /*cmd=*/0x11ul, /*arg=*/0x20000000ul, /*size=*/0x10ul);
break;
case 6:
*(uint32_t*)0x20000a80 = r[0];
*(uint32_t*)0x20000a84 = 0xe0;
*(uint64_t*)0x20000a88 = 0x200008c0;
*(uint32_t*)0x200008f4 = 8;
*(uint64_t*)0x200008f8 = 0x20000680;
*(uint32_t*)0x20000928 = 9;
*(uint32_t*)0x2000092c = 1;
*(uint64_t*)0x20000930 = 0x200006c0;
*(uint64_t*)0x20000938 = 0x20000740;
*(uint32_t*)0x20000944 = 0x8b;
*(uint64_t*)0x20000948 = 0x20000780;
*(uint32_t*)0x20000950 = 0x48;
*(uint32_t*)0x20000954 = 0x10;
*(uint64_t*)0x20000958 = 0x20000800;
*(uint64_t*)0x20000960 = 0x20000840;
*(uint32_t*)0x20000968 = 8;
*(uint32_t*)0x2000096c = 0xc7;
*(uint32_t*)0x20000970 = 8;
*(uint32_t*)0x20000974 = 8;
*(uint64_t*)0x20000978 = 0x20000880;
syscall(__NR_bpf, /*cmd=*/0xful, /*arg=*/0x20000a80ul, /*size=*/0x10ul);
break;
case 7:
res = syscall(__NR_pipe, /*pipefd=*/0x20000000ul);
if (res != -1)
r[4] = *(uint32_t*)0x20000004;
break;
case 8:
*(uint64_t*)0x20000100 = 0x20000580;
memset((void*)0x20000580, 128, 1);
*(uint64_t*)0x20000108 = 1;
syscall(__NR_vmsplice, /*fd=*/r[4], /*vec=*/0x20000100ul, /*vlen=*/1ul,
/*f=*/0ul);
break;
case 9:
*(uint32_t*)0x20000040 = r[4];
*(uint16_t*)0x20000044 = 0;
*(uint16_t*)0x20000046 = 0;
syscall(__NR_poll, /*fds=*/0x20000040ul, /*nfds=*/1ul,
/*timeout=*/0x8001);
break;
case 10:
*(uint64_t*)0x20000280 = 0x20000240;
memcpy((void*)0x20000240, "hrtimer_start\000", 14);
*(uint32_t*)0x20000288 = r[3];
syscall(__NR_bpf, /*cmd=*/0x11ul, /*arg=*/0x20000280ul, /*size=*/0x10ul);
break;
}
}
int main(void) {
syscall(__NR_mmap, /*addr=*/0x1ffff000ul, /*len=*/0x1000ul, /*prot=*/0ul,
/*flags=MAP_FIXED|MAP_ANONYMOUS|MAP_PRIVATE*/ 0x32ul, /*fd=*/-1,
/*offset=*/0ul);
syscall(__NR_mmap, /*addr=*/0x20000000ul, /*len=*/0x1000000ul,
/*prot=PROT_WRITE|PROT_READ|PROT_EXEC*/ 7ul,
/*flags=MAP_FIXED|MAP_ANONYMOUS|MAP_PRIVATE*/ 0x32ul, /*fd=*/-1,
/*offset=*/0ul);
syscall(__NR_mmap, /*addr=*/0x21000000ul, /*len=*/0x1000ul, /*prot=*/0ul,
/*flags=MAP_FIXED|MAP_ANONYMOUS|MAP_PRIVATE*/ 0x32ul, /*fd=*/-1,
/*offset=*/0ul);
loop();
return 0;
}
r0 = bpf$BPF_PROG_RAW_TRACEPOINT_LOAD(0x5, &(0x7f0000002c80)={0x3, 0x5, &(0x7f0000000340)=ANY=[@ANYRESDEC=0x0], &(0x7f0000000080)='GPL\x00', 0x4}, 0x90)
r1 = bpf$MAP_CREATE(0x0, &(0x7f00000009c0)=@base={0x12, 0x4, 0x8, 0x8}, 0x48)
r2 = socket$inet6_udp(0xa, 0x2, 0x0)
bpf$MAP_UPDATE_ELEM(0x2, &(0x7f0000000080)={r1, &(0x7f0000000a40)="90b4", &(0x7f0000000040)=@udp6=r2}, 0x20)
r3 = bpf$PROG_LOAD(0x5, &(0x7f00000000c0)={0x11, 0xc, &(0x7f0000000440)=@framed={{}, [@ringbuf_output={{0x18, 0x1, 0x1, 0x0, r1}, {}, {}, {}, {}, {}, {}, {0x85, 0x0, 0x0, 0x3}}]}, &(0x7f0000000240)='GPL\x00', 0x0, 0x0, 0x0, 0x0, 0x0, '\x00', 0x0, 0x0, 0xffffffffffffffff, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0}, 0x90)
bpf$BPF_RAW_TRACEPOINT_OPEN(0x11, &(0x7f0000000000)={&(0x7f0000000640)='ext4_es_find_extent_range_enter\x00', r3}, 0x10)
bpf$BPF_GET_PROG_INFO(0xf, &(0x7f0000000a80)={r0, 0xe0, &(0x7f00000008c0)={0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x8, &(0x7f0000000680)=[0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0], ""/16, 0x0, 0x0, 0x0, 0x0, 0x9, 0x1, &(0x7f00000006c0)=[0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0], &(0x7f0000000740)=[0x0], 0x0, 0x8b, &(0x7f0000000780)=[{}, {}, {}, {}, {}, {}, {}, {}, {}], 0x48, 0x10, &(0x7f0000000800), &(0x7f0000000840), 0x8, 0xc7, 0x8, 0x8, &(0x7f0000000880)}}, 0x10)
pipe(&(0x7f0000000000)={0xffffffffffffffff, <r4=>0xffffffffffffffff})
vmsplice(r4, &(0x7f0000000100)=[{&(0x7f0000000580)="80", 0x1}], 0x1, 0x0)
poll(&(0x7f0000000040)=[{r4}], 0x1, 0x8001)
bpf$BPF_RAW_TRACEPOINT_OPEN(0x11, &(0x7f0000000280)={&(0x7f0000000240)='hrtimer_start\x00', r3}, 0x10)
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment