From mboxrd@z Thu Jan 1 00:00:00 1970 From: linux@armlinux.org.uk (Russell King - ARM Linux) Date: Tue, 8 May 2018 12:45:17 +0100 Subject: [PATCH REPOST] ARM: smp: Move clear_tasks_mm_cpumask() call to __cpu_die() In-Reply-To: <20180504112939.25493-1-bigeasy@linutronix.de> References: <20180504112939.25493-1-bigeasy@linutronix.de> Message-ID: <20180508114516.GE16141@n2100.armlinux.org.uk> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On Fri, May 04, 2018 at 01:29:39PM +0200, Sebastian Andrzej Siewior wrote: > From: Grygorii Strashko > > Suspending a CPU on a RT kernel results in the following backtrace: > > | Disabling non-boot CPUs ... > | BUG: sleeping function called from invalid context at kernel/locking/rtmutex.c:917 > | in_atomic(): 1, irqs_disabled(): 128, pid: 18, name: migration/1 > | INFO: lockdep is turned off. > | irq event stamp: 122 > | hardirqs last enabled at (121): [] _raw_spin_unlock_irqrestore+0x88/0x90 > | hardirqs last disabled at (122): [] _raw_spin_lock_irq+0x28/0x5c > | CPU: 1 PID: 18 Comm: migration/1 Tainted: G W 4.1.4-rt3-01046-g96ac8da #204 > | Hardware name: Generic DRA74X (Flattened Device Tree) > | [] (unwind_backtrace) from [] (show_stack+0x20/0x24) > | [] (show_stack) from [] (dump_stack+0x88/0xdc) > | [] (dump_stack) from [] (___might_sleep+0x198/0x2a8) > | [] (___might_sleep) from [] (rt_spin_lock+0x30/0x70) > | [] (rt_spin_lock) from [] (find_lock_task_mm+0x9c/0x174) > | [] (find_lock_task_mm) from [] (clear_tasks_mm_cpumask+0xb4/0x1ac) > | [] (clear_tasks_mm_cpumask) from [] (__cpu_disable+0x98/0xbc) > | [] (__cpu_disable) from [] (take_cpu_down+0x1c/0x50) > | [] (take_cpu_down) from [] (multi_cpu_stop+0x11c/0x158) > | [] (multi_cpu_stop) from [] (cpu_stopper_thread+0xc4/0x184) > | [] (cpu_stopper_thread) from [] (smpboot_thread_fn+0x18c/0x324) > | [] (smpboot_thread_fn) from [] (kthread+0xe8/0x104) > | [] (kthread) from [] (ret_from_fork+0x14/0x3c) > | CPU1: shutdown > > The root cause of above backtrace is task_lock() which takes a sleeping > lock on -RT. > > To fix the issue, move clear_tasks_mm_cpumask() call from __cpu_disable() > to __cpu_die() which is called on the thread which is asking for a target > CPU to be shutdown. In addition, this change restores CPU hotplug > functionality on ARM CPU1 can be unplugged/plugged many times. This looks fine to me - all the paths that use the mm_cpumask() check that the CPU is still marked online. Please send to the patch system, thanks. > > Signed-off-by: Grygorii Strashko > Cc: > Cc: Sekhar Nori > Cc: Russell King > Link: http://lkml.kernel.org/r/1441995683-30817-1-git-send-email-grygorii.strashko at ti.com > Signed-off-by: Thomas Gleixner > [bigeasy: slighty edited the commit message] > Signed-off-by: Sebastian Andrzej Siewior > --- > arch/arm/kernel/smp.c | 3 +-- > 1 file changed, 1 insertion(+), 2 deletions(-) > > --- a/arch/arm/kernel/smp.c > +++ b/arch/arm/kernel/smp.c > @@ -236,8 +236,6 @@ int __cpu_disable(void) > flush_cache_louis(); > local_flush_tlb_all(); > > - clear_tasks_mm_cpumask(cpu); > - > return 0; > } > > @@ -255,6 +253,7 @@ void __cpu_die(unsigned int cpu) > } > pr_debug("CPU%u: shutdown\n", cpu); > > + clear_tasks_mm_cpumask(cpu); > /* > * platform_cpu_kill() is generally expected to do the powering off > * and/or cutting of clocks to the dying CPU. Optionally, this may -- RMK's Patch system: http://www.armlinux.org.uk/developer/patches/ FTTC broadband for 0.8mile line in suburbia: sync at 8.8Mbps down 630kbps up According to speedtest.net: 8.21Mbps down 510kbps up