mirror of
https://github.com/AuxXxilium/linux_dsm_epyc7002.git
synced 2024-12-05 13:16:39 +07:00
03fc7f9c99
The commit719f6a7040
("printk: Use the main logbuf in NMI when logbuf_lock is available") brought back the possible deadlocks in printk() and NMI. The check of logbuf_lock is done only in printk_nmi_enter() to prevent mixed output. But another CPU might take the lock later, enter NMI, and: + Both NMIs might be serialized by yet another lock, for example, the one in nmi_cpu_backtrace(). + The other CPU might get stopped in NMI, see smp_send_stop() in panic(). The only safe solution is to use trylock when storing the message into the main log-buffer. It might cause reordering when some lines go to the main lock buffer directly and others are delayed via the per-CPU buffer. It means that it is not useful in general. This patch replaces the problematic NMI deferred context with NMI direct context. It can be used to mark a code that might produce many messages in NMI and the risk of losing them is more critical than problems with eventual reordering. The context is then used when dumping trace buffers on oops. It was the primary motivation for the original fix. Also the reordering is even smaller issue there because some traces have their own time stamps. Finally, nmi_cpu_backtrace() need not longer be serialized because it will always us the per-CPU buffers again. Fixes:719f6a7040
("printk: Use the main logbuf in NMI when logbuf_lock is available") Cc: stable@vger.kernel.org Link: http://lkml.kernel.org/r/20180627142028.11259-1-pmladek@suse.com To: Steven Rostedt <rostedt@goodmis.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Cc: Sergey Senozhatsky <sergey.senozhatsky.work@gmail.com> Cc: linux-kernel@vger.kernel.org Cc: stable@vger.kernel.org Acked-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Petr Mladek <pmladek@suse.com>
111 lines
3.0 KiB
C
111 lines
3.0 KiB
C
// SPDX-License-Identifier: GPL-2.0
|
|
/*
|
|
* NMI backtrace support
|
|
*
|
|
* Gratuitously copied from arch/x86/kernel/apic/hw_nmi.c by Russell King,
|
|
* with the following header:
|
|
*
|
|
* HW NMI watchdog support
|
|
*
|
|
* started by Don Zickus, Copyright (C) 2010 Red Hat, Inc.
|
|
*
|
|
* Arch specific calls to support NMI watchdog
|
|
*
|
|
* Bits copied from original nmi.c file
|
|
*/
|
|
#include <linux/cpumask.h>
|
|
#include <linux/delay.h>
|
|
#include <linux/kprobes.h>
|
|
#include <linux/nmi.h>
|
|
#include <linux/cpu.h>
|
|
#include <linux/sched/debug.h>
|
|
|
|
#ifdef arch_trigger_cpumask_backtrace
|
|
/* For reliability, we're prepared to waste bits here. */
|
|
static DECLARE_BITMAP(backtrace_mask, NR_CPUS) __read_mostly;
|
|
|
|
/* "in progress" flag of arch_trigger_cpumask_backtrace */
|
|
static unsigned long backtrace_flag;
|
|
|
|
/*
|
|
* When raise() is called it will be passed a pointer to the
|
|
* backtrace_mask. Architectures that call nmi_cpu_backtrace()
|
|
* directly from their raise() functions may rely on the mask
|
|
* they are passed being updated as a side effect of this call.
|
|
*/
|
|
void nmi_trigger_cpumask_backtrace(const cpumask_t *mask,
|
|
bool exclude_self,
|
|
void (*raise)(cpumask_t *mask))
|
|
{
|
|
int i, this_cpu = get_cpu();
|
|
|
|
if (test_and_set_bit(0, &backtrace_flag)) {
|
|
/*
|
|
* If there is already a trigger_all_cpu_backtrace() in progress
|
|
* (backtrace_flag == 1), don't output double cpu dump infos.
|
|
*/
|
|
put_cpu();
|
|
return;
|
|
}
|
|
|
|
cpumask_copy(to_cpumask(backtrace_mask), mask);
|
|
if (exclude_self)
|
|
cpumask_clear_cpu(this_cpu, to_cpumask(backtrace_mask));
|
|
|
|
/*
|
|
* Don't try to send an NMI to this cpu; it may work on some
|
|
* architectures, but on others it may not, and we'll get
|
|
* information at least as useful just by doing a dump_stack() here.
|
|
* Note that nmi_cpu_backtrace(NULL) will clear the cpu bit.
|
|
*/
|
|
if (cpumask_test_cpu(this_cpu, to_cpumask(backtrace_mask)))
|
|
nmi_cpu_backtrace(NULL);
|
|
|
|
if (!cpumask_empty(to_cpumask(backtrace_mask))) {
|
|
pr_info("Sending NMI from CPU %d to CPUs %*pbl:\n",
|
|
this_cpu, nr_cpumask_bits, to_cpumask(backtrace_mask));
|
|
raise(to_cpumask(backtrace_mask));
|
|
}
|
|
|
|
/* Wait for up to 10 seconds for all CPUs to do the backtrace */
|
|
for (i = 0; i < 10 * 1000; i++) {
|
|
if (cpumask_empty(to_cpumask(backtrace_mask)))
|
|
break;
|
|
mdelay(1);
|
|
touch_softlockup_watchdog();
|
|
}
|
|
|
|
/*
|
|
* Force flush any remote buffers that might be stuck in IRQ context
|
|
* and therefore could not run their irq_work.
|
|
*/
|
|
printk_safe_flush();
|
|
|
|
clear_bit_unlock(0, &backtrace_flag);
|
|
put_cpu();
|
|
}
|
|
|
|
bool nmi_cpu_backtrace(struct pt_regs *regs)
|
|
{
|
|
int cpu = smp_processor_id();
|
|
|
|
if (cpumask_test_cpu(cpu, to_cpumask(backtrace_mask))) {
|
|
if (regs && cpu_in_idle(instruction_pointer(regs))) {
|
|
pr_warn("NMI backtrace for cpu %d skipped: idling at %pS\n",
|
|
cpu, (void *)instruction_pointer(regs));
|
|
} else {
|
|
pr_warn("NMI backtrace for cpu %d\n", cpu);
|
|
if (regs)
|
|
show_regs(regs);
|
|
else
|
|
dump_stack();
|
|
}
|
|
cpumask_clear_cpu(cpu, to_cpumask(backtrace_mask));
|
|
return true;
|
|
}
|
|
|
|
return false;
|
|
}
|
|
NOKPROBE_SYMBOL(nmi_cpu_backtrace);
|
|
#endif
|