On Thu 2022-12-01 05:27:13, Zhang, Qiang1 wrote:
> Greeting,
>
> FYI, we noticed BUG:scheduling_while_atomic due to commit (built with clang-14):
>
> commit: 8bdbdd7f43cd74c7faca6add8a62d541503ae21d ("printk: Prepare for SRCU console list protection")
> https://git.kernel.org/cgit/linux/kernel/git/next/linux-next.git master
>
> in testcase: boot
>
> on test machine: qemu-system-x86_64 -enable-kvm -cpu SandyBridge -smp 2 -m 16G
>
> caused below changes (please refer to attached dmesg/kmsg for entire log/backtrace):
>
>
> [ 8.561823][ T0] BUG: scheduling while atomic: swapper/0/0x00000002
> [ 8.569154][ T0] no locks held by swapper/0.
> [ 8.571934][ T0] Modules linked in:
> [ 8.573001][ T0] CPU: 0 PID: 0 Comm: swapper Not tainted 6.1.0-rc1-00015-g8bdbdd7f43cd #1 9e431b4e696756e99f94acf7e74ac6e512df80ce
> [ 8.576740][ T0] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.16.0-debian-1.16.0-4 04/01/2014
> [ 8.579942][ T0] Call Trace:
> [ 8.581143][ T0] <TASK>
> [ 8.582054][ T0] dump_stack_lvl (??:?)
> [ 8.583312][ T0] ? netdev_notice (??:?)
> [ 8.584753][ T0] ? lockdep_print_held_locks (lockdep.c:?)
> [ 8.586563][ T0] __schedule_bug (core.c:?)
> [ 8.588171][ T0] ? trace_sched_switch (core.c:?)
> [ 8.589753][ T0] ? save_trace (lockdep.c:?)
> [ 8.591135][ T0] schedule_debug (core.c:?)
> [ 8.592582][ T0] __schedule (core.c:?)
> [ 8.593902][ T0] ? __sched_text_start (core.c:?)
> [ 8.595356][ T0] ? add_chain_block (lockdep.c:?)
> [ 8.596847][ T0] ? find_held_lock (lockdep.c:?)
> [ 8.598368][ T0] schedule (??:?)
> [ 8.599564][ T0] schedule_timeout (??:?)
> [ 8.600937][ T0] ? console_conditional_schedule (??:?)
> [ 8.602773][ T0] do_wait_for_common (build_utility.c:?)
> [ 8.604522][ T0] ? console_conditional_schedule (??:?)
> [ 8.606462][ T0] ? bit_wait_io_timeout (build_utility.c:?)
> [ 8.608196][ T0] ? _raw_spin_lock_irq (??:?)
> [ 8.609935][ T0] ? lockdep_hardirqs_on (??:?)
> [ 8.611646][ T0] wait_for_completion (??:?)
> [ 8.613253][ T0] synchronize_srcu (??:?)
Hmm, I do not understand it. It seems that this is called
in normal context. There was console_unlock() right
before synchronize_srcu(). It means that we took
semaphore in the same context.
Also all the calls above looks correct. wait_for_completion()
seems to be called in normal context as well.
Best Regards,
Petr
> [ 8.614825][ T0] ? srcu_gp_start_if_needed (??:?)
> [ 8.616664][ T0] ? rcu_read_lock_any_held (??:?)
> [ 8.618281][ T0] ? console_trylock_spinning (??:?)
> [ 8.620181][ T0] unregister_console (??:?)
> [ 8.621875][ T0] register_console (??:?)
> [ 8.623401][ T0] ? serial8250_isa_init_ports (8250_core.c:?)
> [ 8.625281][ T0] ? parse_options (super.c:?)
> [ 8.626887][ T0] univ8250_console_init (8250_core.c:?)
> [ 8.628583][ T0] console_init (??:?)
> [ 8.630025][ T0] start_kernel (??:?)
> [ 8.631558][ T0] secondary_startup_64_no_verify (??:?)
> [ 8.633502][ T0] </TASK>
> [ 8.634624][ T0] ------------[ cut here ]------------
> [ 8.636289][ T0] releasing a pinned lock
> [ 8.638693][ T0] WARNING: CPU: 0 PID: 0 at kernel/locking/lockdep.c:5352 lock_release (??:?)
> [ 8.641591][ T0] Modules linked in:
> [ 8.642864][ T0] CPU: 0 PID: 0 Comm: swapper Tainted: G W 6.1.0-rc1-00015-g8bdbdd7f43cd #1 9e431b4e696756e99f94acf7e74ac6e512df80ce
> [ 8.646469][ T0] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.16.0-debian-1.16.0-4 04/01/2014
> [ 8.649578][ T0] RIP: 0010:lock_release (??:?)
> [ 8.651243][ T0] Code: 00 00 e9 0f fe ff ff 48 83 05 06 f6 ff 06 01 e8 91 4e 2d 03 e9 67 fe ff ff 48 c7 c7 20 8c cf 84 e8 90 3f ec ff 48 8b 54 24 08 <0f> 0b 48 83 05 a9 f6 ff 06 01 e9 66 fc ff ff e8 67 c9 99 01 85 c0
> All code
> ========
> 0: 00 00 add %al,(%rax)
> 2: e9 0f fe ff ff jmpq 0xfffffffffffffe16
> 7: 48 83 05 06 f6 ff 06 addq $0x1,0x6fff606(%rip) # 0x6fff615
> e: 01
> f: e8 91 4e 2d 03 callq 0x32d4ea5
> 14: e9 67 fe ff ff jmpq 0xfffffffffffffe80
> 19: 48 c7 c7 20 8c cf 84 mov $0xffffffff84cf8c20,%rdi
> 20: e8 90 3f ec ff callq 0xffffffffffec3fb5
> 25: 48 8b 54 24 08 mov 0x8(%rsp),%rdx
> 2a:* 0f 0b ud2 <-- trapping instruction
> 2c: 48 83 05 a9 f6 ff 06 addq $0x1,0x6fff6a9(%rip) # 0x6fff6dd
> 33: 01
> 34: e9 66 fc ff ff jmpq 0xfffffffffffffc9f
> 39: e8 67 c9 99 01 callq 0x199c9a5
> 3e: 85 c0 test %eax,%eax
>
> Code starting with the faulting instruction
> ===========================================
> 0: 0f 0b ud2
> 2: 48 83 05 a9 f6 ff 06 addq $0x1,0x6fff6a9(%rip) # 0x6fff6b3
> 9: 01
> a: e9 66 fc ff ff jmpq 0xfffffffffffffc75
> f: e8 67 c9 99 01 callq 0x199c97b
> 14: 85 c0 test %eax,%eax
> [ 8.656638][ T0] RSP: 0000:ffffffff862077c0 EFLAGS: 00010086
> [ 8.658249][ T0] RAX: 0000000000000017 RBX: ffffffff86244244 RCX: ffffffff8631a420
> [ 8.660383][ T0] RDX: ffffffff86244218 RSI: 0000000000000008 RDI: 0000000000000001
> [ 8.662798][ T0] RBP: ffffffff862078e0 R08: dffffc0000000000 R09: 00000000035545d9
> [ 8.665205][ T0] R10: dffff7fff0c40e7a R11: 0000000000000001 R12: ffffffff862fcf58
> [ 8.667420][ T0] R13: 1ffffffff0c40f06 R14: ffffffff86244220 R15: dffffc0000000000
> [ 8.669824][ T0] FS: 0000000000000000(0000) GS:ffffffff862ab000(0000) knlGS:0000000000000000
> [ 8.672552][ T0] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> [ 8.674557][ T0] CR2: ffff88843ffff000 CR3: 0000000006235000 CR4: 00000000000406b0
> [ 8.677082][ T0] Call Trace:
> [ 8.678220][ T0] <TASK>
> [ 8.679230][ T0] ? __lock_acquire (??:?)
> [ 8.680709][ T0] ? kvm_sched_clock_read (kvmclock.c:?)
> [ 8.682309][ T0] ? sched_clock_cpu (??:?)
> [ 8.683854][ T0] ? kvm_sched_clock_read (kvmclock.c:?)
> [ 8.685268][ T0] _raw_spin_unlock (??:?)
> [ 8.686730][ T0] dequeue_task_idle (build_policy.c:?)
> [ 8.688259][ T0] ? set_next_task_idle (build_policy.c:?)
> [ 8.689822][ T0] ? update_rq_clock (??:?)
> [ 8.691387][ T0] __schedule (core.c:?)
> [ 8.692782][ T0] ? __sched_text_start (core.c:?)
> [ 8.694287][ T0] ? add_chain_block (lockdep.c:?)
> [ 8.695869][ T0] ? find_held_lock (lockdep.c:?)
> [ 8.697314][ T0] schedule (??:?)
> [ 8.698529][ T0] schedule_timeout (??:?)
> [ 8.701929][ T0] ? console_conditional_schedule (??:?)
> [ 8.703826][ T0] do_wait_for_common (build_utility.c:?)
> [ 8.705395][ T0] ? console_conditional_schedule (??:?)
> [ 8.707189][ T0] ? bit_wait_io_timeout (build_utility.c:?)
> [ 8.708779][ T0] ? _raw_spin_lock_irq (??:?)
> [ 8.710371][ T0] ? lockdep_hardirqs_on (??:?)
> [ 8.711982][ T0] wait_for_completion (??:?)
> [ 8.713487][ T0] synchronize_srcu (??:?)
> [ 8.715020][ T0] ? srcu_gp_start_if_needed (??:?)
> [ 8.716772][ T0] ? rcu_read_lock_any_held (??:?)
> [ 8.718477][ T0] ? console_trylock_spinning (??:?)
> [ 8.720179][ T0] unregister_console (??:?)
> [ 8.721644][ T0] register_console (??:?)
> [ 8.728846][ T0] ? serial8250_isa_init_ports (8250_core.c:?)
> [ 8.730692][ T0] ? parse_options (super.c:?)
> [ 8.732180][ T0] univ8250_console_init (8250_core.c:?)
> [ 8.733762][ T0] console_init (??:?)
> [ 8.735096][ T0] start_kernel (??:?)
> [ 8.736425][ T0] secondary_startup_64_no_verify (??:?)
> [ 8.738088][ T0] </TASK>
> [ 8.739057][ T0] irq event stamp: 494
> [ 8.740376][ T0] hardirqs last enabled at (493): dump_stack_lvl (??:?)
> [ 8.743081][ T0] hardirqs last disabled at (494): __schedule (core.c:?)
> [ 8.745797][ T0] softirqs last enabled at (0): 0x0
> [ 8.747882][ T0] softirqs last disabled at (0): 0x0
> [ 8.749889][ T0] ---[ end trace 0000000000000000 ]---
> [ 8.751520][ T0] bad: scheduling from the idle thread!
> [ 8.753351][ T0] CPU: 0 PID: 0 Comm: swapper Tainted: G W 6.1.0-rc1-00015-g8bdbdd7f43cd #1 9e431b4e696756e99f94acf7e74ac6e512df80ce
> [ 8.757566][ T0] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.16.0-debian-1.16.0-4 04/01/2014
> [ 8.760783][ T0] Call Trace:
> [ 8.761970][ T0] <TASK>
> [ 8.763067][ T0] dump_stack_lvl (??:?)
> [ 8.764611][ T0] ? netdev_notice (??:?)
> [ 8.766213][ T0] ? lockdep_hardirqs_on_prepare (??:?)
> [ 8.768214][ T0] ? print_irqtrace_events (??:?)
> [ 8.769911][ T0] ? kvm_sched_clock_read (kvmclock.c:?)
> [ 8.771550][ T0] dequeue_task_idle (build_policy.c:?)
> [ 8.773087][ T0] ? set_next_task_idle (build_policy.c:?)
> [ 8.774647][ T0] ? update_rq_clock (??:?)
> [ 8.776280][ T0] __schedule (core.c:?)
> [ 8.777729][ T0] ? __sched_text_start (core.c:?)
> [ 8.779238][ T0] ? add_chain_block (lockdep.c:?)
> [ 8.780922][ T0] ? find_held_lock (lockdep.c:?)
> [ 8.782560][ T0] schedule (??:?)
> [ 8.783988][ T0] schedule_timeout (??:?)
> [ 8.785515][ T0] ? console_conditional_schedule (??:?)
> [ 8.787500][ T0] do_wait_for_common (build_utility.c:?)
> [ 8.789162][ T0] ? console_conditional_schedule (??:?)
> [ 8.790997][ T0] ? bit_wait_io_timeout (build_utility.c:?)
> [ 8.792567][ T0] ? _raw_spin_lock_irq (??:?)
> [ 8.794292][ T0] ? lockdep_hardirqs_on (??:?)
> [ 8.796008][ T0] wait_for_completion (??:?)
> [ 8.797520][ T0] synchronize_srcu (??:?)
> [ 8.799001][ T0] ? srcu_gp_start_if_needed (??:?)
> [ 8.800696][ T0] ? rcu_read_lock_any_held (??:?)
> [ 8.802508][ T0] ? console_trylock_spinning (??:?)
> [ 8.804316][ T0] unregister_console (??:?)
> [ 8.805810][ T0] register_console (??:?)
> [ 8.807364][ T0] ? serial8250_isa_init_ports (8250_core.c:?)
> [ 8.811167][ T0] ? parse_options (super.c:?)
> [ 8.812693][ T0] univ8250_console_init (8250_core.c:?)
> [ 8.814269][ T0] console_init (??:?)
> [ 8.815765][ T0] start_kernel (??:?)
> [ 8.817296][ T0] secondary_startup_64_no_verify (??:?)
> [ 8.819246][ T0] </TASK>
>
>
>
> Through the config file, it can be found that the following options are enabled
>
> CONFIG_TINY_RCU=y
> CONFIG_RCU_EXPERT=y
> CONFIG_SRCU=y
> CONFIG_TINY_SRCU=y
>
> This problem should have been fixed:
>
> https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/?id=0bf43fcbf8ebbf84a2b1b8770eaefcdb4a99afd6
>
> commit dbc6ca150842650d513705f26e3e6b7a4e182ce9
> Author: Zqiang <qiang1.zhang@xxxxxxxxx>
> Date: Wed Nov 9 15:36:38 2022 +0800
>
> srcu: Make Tiny synchronize_srcu() check for readers
>
> This commit adds lockdep checks for illegal use of synchronize_srcu()
> within same-type SRCU read-side critical sections and within normal
> RCU read-side critical sections. It also makes synchronize_srcu()
> be a no-op during early boot.
>
> These changes bring Tiny synchronize_srcu() into line with both Tree
> synchronize_srcu() and Tiny synchronize_rcu().
>
> Signed-off-by: Zqiang <qiang1.zhang@xxxxxxxxx>
> Signed-off-by: Paul E. McKenney <paulmck@xxxxxxxxxx>
>
>
> Thanks
> Zqiang
>
>
>
>
>
>
>
> If you fix the issue, kindly add following tag
> | Reported-by: kernel test robot <yujie.liu@xxxxxxxxx>
> | Link: https://lore.kernel.org/oe-lkp/202211302358.ef0db537-yujie.liu@xxxxxxxxx
>
>
> To reproduce:
>
> # build kernel
> cd linux
> cp config-6.1.0-rc1-00015-g8bdbdd7f43cd .config
> make HOSTCC=clang-14 CC=clang-14 ARCH=x86_64 olddefconfig prepare modules_prepare bzImage modules
> make HOSTCC=clang-14 CC=clang-14 ARCH=x86_64 INSTALL_MOD_PATH=<mod-install-dir> modules_install
> cd <mod-install-dir>
> find lib/ | cpio -o -H newc --quiet | gzip > modules.cgz
>
>
> git clone https://github.com/intel/lkp-tests.git
> cd lkp-tests
> bin/lkp qemu -k <bzImage> -m modules.cgz job-script # job-script is attached in this email
>
> # if come across any failure that blocks the test,
> # please remove ~/.lkp and /lkp dir to run from a clean state.
>
>
> --
> 0-DAY CI Kernel Test Service
> https://01.org/lkp
