RISC-V: Fixup schedule out issue in machine_crash_shutdown()
authorXianting Tian <xianting.tian@linux.alibaba.com>
Thu, 11 Aug 2022 07:41:47 +0000 (15:41 +0800)
committerPalmer Dabbelt <palmer@rivosinc.com>
Thu, 11 Aug 2022 15:54:46 +0000 (08:54 -0700)
commitad943893d5f1d0aeea892bf7b781cf8062b36d58
tree2b8c06458cc8c1cded882b76f4a7ee6fe7a236a9
parent59c026c359c30f116fef6ee958e24d04983efbb0
RISC-V: Fixup schedule out issue in machine_crash_shutdown()

Current task of executing crash kexec will be schedule out when panic is
triggered by RCU Stall, as it needs to wait rcu completion. It lead to
inability to enter the crash system.

The implementation of machine_crash_shutdown() is non-standard for RISC-V
according to other Arch's implementation(eg, x86, arm64), we need to send
IPI to stop secondary harts.

[224521.877268] rcu: INFO: rcu_preempt detected stalls on CPUs/tasks:
[224521.883471] rcu:  0-...0: (3 GPs behind) idle=cfa/0/0x1 softirq=3968793/3968793 fqs=2495
[224521.891742]  (detected by 2, t=5255 jiffies, g=60855593, q=328)
[224521.897754] Task dump for CPU 0:
[224521.901074] task:swapper/0     state:R  running task   stack:  0 pid:  0 ppid:   0 flags:0x00000008
[224521.911090] Call Trace:
[224521.913638] [<ffffffe000c432de>] __schedule+0x208/0x5ea
[224521.918957] Kernel panic - not syncing: RCU Stall
[224521.923773] bad: scheduling from the idle thread!
[224521.928571] CPU: 2 PID: 0 Comm: swapper/2 Kdump: loaded Tainted: G   O  5.10.113-yocto-standard #1
[224521.938658] Call Trace:
[224521.941200] [<ffffffe00020395c>] walk_stackframe+0x0/0xaa
[224521.946689] [<ffffffe000c34f8e>] show_stack+0x32/0x3e
[224521.951830] [<ffffffe000c39020>] dump_stack_lvl+0x7e/0xa2
[224521.957317] [<ffffffe000c39058>] dump_stack+0x14/0x1c
[224521.962459] [<ffffffe000243884>] dequeue_task_idle+0x2c/0x40
[224521.968207] [<ffffffe000c434f4>] __schedule+0x41e/0x5ea
[224521.973520] [<ffffffe000c43826>] schedule+0x34/0xe4
[224521.978487] [<ffffffe000c46cae>] schedule_timeout+0xc6/0x170
[224521.984234] [<ffffffe000c4491e>] wait_for_completion+0x98/0xf2
[224521.990157] [<ffffffe00026d9e2>] __wait_rcu_gp+0x148/0x14a
[224521.995733] [<ffffffe0002761c4>] synchronize_rcu+0x5c/0x66
[224522.001307] [<ffffffe00026f1a6>] rcu_sync_enter+0x54/0xe6
[224522.006795] [<ffffffe00025a436>] percpu_down_write+0x32/0x11c
[224522.012629] [<ffffffe000c4266a>] _cpu_down+0x92/0x21a
[224522.017771] [<ffffffe000219a0a>] smp_shutdown_nonboot_cpus+0x90/0x118
[224522.024299] [<ffffffe00020701e>] machine_crash_shutdown+0x30/0x4a
[224522.030483] [<ffffffe00029a3f8>] __crash_kexec+0x62/0xa6
[224522.035884] [<ffffffe000c3515e>] panic+0xfa/0x2b6
[224522.040678] [<ffffffe0002772be>] rcu_sched_clock_irq+0xc26/0xcb8
[224522.046774] [<ffffffe00027fc7a>] update_process_times+0x62/0x8a
[224522.052785] [<ffffffe00028d522>] tick_sched_timer+0x9e/0x102
[224522.058533] [<ffffffe000280c3a>] __hrtimer_run_queues+0x16a/0x318
[224522.064716] [<ffffffe0002812ec>] hrtimer_interrupt+0xd4/0x228
[224522.070551] [<ffffffe0009a69b6>] riscv_timer_interrupt+0x3c/0x48
[224522.076646] [<ffffffe000268f8c>] handle_percpu_devid_irq+0xb0/0x24c
[224522.083004] [<ffffffe00026428e>] __handle_domain_irq+0xa8/0x122
[224522.089014] [<ffffffe00062f954>] riscv_intc_irq+0x38/0x60
[224522.094501] [<ffffffe000201bd4>] ret_from_exception+0x0/0xc
[224522.100161] [<ffffffe000c42146>] rcu_eqs_enter.constprop.0+0x8c/0xb8

With the patch, it can enter crash system when RCU Stall occur.

Fixes: e53d28180d4d ("RISC-V: Add kdump support")
Signed-off-by: Xianting Tian <xianting.tian@linux.alibaba.com>
Link: https://lore.kernel.org/r/20220811074150.3020189-4-xianting.tian@linux.alibaba.com
Cc: stable@vger.kernel.org
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
arch/riscv/kernel/machine_kexec.c