* Yoshihiro YUNOMAE <yoshihiro.yunomae.ez@xxxxxxxxxxx> wrote: > Prevent crash_kexec() from deadlocking of ioapic_lock. s/of/on > When crash_kexec() is executed on a cpu, the cpu will get > ioapic_lock in disable_IO_APIC(). So if the cpu gets NMI > while locking ioapic_lock, a deadlock wiil happen. In s/will > this patch, ioapic_lock is initialized before > disable_IO_APIC(). > > To confirm this deadlocking, you'll set up as follows: s/deadlocking/deadlock > 1. Add mdelay(1000) after raw_spin_lock_irqsave() in > native_ioapic_set_affinity()@arch/x86/kernel/apic/io_apic.c > > Although the deadlocking can occur without this modification, it will > increase the potential of the deadlocking problem. s/deadlocking/deadlock > > 2. Build and install the kernel > > 3. Set up the OS which will run panic() and kexec when NMI is injected > # echo "kernel.unknown_nmi_panic=1" >> /etc/sysctl.conf > # vim /etc/default/grub > add "nmi_watchdog=0 crashkernel=256M" in GRUB_CMDLINE_LINUX line > # grub2-mkconfig > > 4. Reboot the OS > > 5. Run following command for each vcpu on the guest > # while true; do echo <CPU num> > /proc/irq/<IO-APIC-edge or IO-APIC-fasteoi>/smp_affinitity; done; > By running this command, cpus will get ioapic_lock for setting affinity. > > 6. Inject NMI (push a dump button or execute 'virsh inject-nmi <domain>' if you > use VM) > After injecting NMI, panic() is called in an nmi-handler context. > Then, kexec will normally run in panic(), but the operation will be stopped > by deadlock of ioapic_lock in crash_kexec()->machine_crash_shutdown()-> s/of/on > native_machine_crash_shutdown()->disable_IO_APIC()->clear_IO_APIC()-> > clear_IO_APIC_pin()->ioapic_read_entry(). I suppose we could do this patch if it's a common occurance. A few minor details need fixing: > > Signed-off-by: Yoshihiro YUNOMAE <yoshihiro.yunomae.ez@xxxxxxxxxxx> > Cc: Thomas Gleixner <tglx@xxxxxxxxxxxxx> > Cc: Ingo Molnar <mingo@xxxxxxxxxx> > Cc: "H. Peter Anvin" <hpa@xxxxxxxxx> > Cc: x86@xxxxxxxxxx > Cc: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> > Cc: Andi Kleen <ak@xxxxxxxxxxxxxxx> > Cc: Seiji Aguchi <seiji.aguchi@xxxxxxx> > Cc: Konrad Rzeszutek Wilk <konrad.wilk@xxxxxxxxxx> > Cc: Sebastian Andrzej Siewior <sebastian@xxxxxxxxxxxxx> > Cc: Joerg Roedel <joro@xxxxxxxxxx> > Cc: Zhang Yanfei <zhangyanfei@xxxxxxxxxxxxxx> > Cc: "Eric W. Biederman" <ebiederm@xxxxxxxxxxxx> > Cc: Gleb Natapov <gleb@xxxxxxxxxx> > Cc: Marcelo Tosatti <mtosatti@xxxxxxxxxx> > Cc: linux-kernel@xxxxxxxxxxxxxxx > Cc: stable@xxxxxxxxxxxxxxx > --- > arch/x86/include/asm/apic.h | 2 ++ > arch/x86/kernel/apic/io_apic.c | 5 +++++ > arch/x86/kernel/crash.c | 4 ++++ > 3 files changed, 11 insertions(+) > > diff --git a/arch/x86/include/asm/apic.h b/arch/x86/include/asm/apic.h > index f8119b5..ddb06af 100644 > --- a/arch/x86/include/asm/apic.h > +++ b/arch/x86/include/asm/apic.h > @@ -715,4 +715,6 @@ static inline void exiting_ack_irq(void) > ack_APIC_irq(); > } > > +extern void ioapic_lock_init(void); > + > #endif /* _ASM_X86_APIC_H */ > diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c > index 9ed796c..2816c07 100644 > --- a/arch/x86/kernel/apic/io_apic.c > +++ b/arch/x86/kernel/apic/io_apic.c > @@ -1534,6 +1534,11 @@ void intel_ir_io_apic_print_entries(unsigned int apic, > } > } > > +void ioapic_lock_init(void) > +{ > + raw_spin_lock_init(&ioapic_lock); > +} Please name this ioapic_zap_locks() to make clear that this is crash handling related. > + > __apicdebuginit(void) print_IO_APIC(int ioapic_idx) > { > union IO_APIC_reg_00 reg_00; > diff --git a/arch/x86/kernel/crash.c b/arch/x86/kernel/crash.c > index 74467fe..ea039d5 100644 > --- a/arch/x86/kernel/crash.c > +++ b/arch/x86/kernel/crash.c > @@ -129,6 +129,10 @@ void native_machine_crash_shutdown(struct pt_regs *regs) > > lapic_shutdown(); > #if defined(CONFIG_X86_IO_APIC) Please enhance this #ifdef while at it. > + /* > + * Prevent crash_kexec() from deadlocking of ioapic_lock. > + */ s/of/on. Also, single-line comment can go /* here */. > + ioapic_lock_init(); > disable_IO_APIC(); > #endif > #ifdef CONFIG_HPET_TIMER > Thanks, Ingo -- To unsubscribe from this list: send the line "unsubscribe stable" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html