x86/hpet: Read HPET directly if panic in progress

zhaoxin inclusion
category: other

-------------------

When the clocksource of the system is HPET，a CPU executing read_hpet might
be interrupted by #GP/#PF to executing the panic，this may lead to
read_hpet dead loops:

CPU x				    CPU x
----                                ----
read_hpet()
  arch_spin_trylock(&hpet.lock)
  [CPU x got the hpet.lock]         #GP/#PF happened
                                    panic()
                                      kmsg_dump()
                                        pstore_dump()
                                          pstore_record_init()
                                            ktime_get_real_fast_ns()
                                              read_hpet()
                                              [dead loops]

To avoid this dead loops, read HPET directly if panic in progress.

Signed-off-by: leoliu-oc <[email protected]>

arch/x86/kernel/hpet.c

@@ -804,6 +804,12 @@ static u64 read_hpet(struct clocksource *cs)
 	if (in_nmi())
 		return (u64)hpet_readl(HPET_COUNTER);
 
+	/*
+	 * Read HPET directly if panic in progress.
+	 */
+	if (unlikely(atomic_read(&panic_cpu) != PANIC_CPU_INVALID))
+		return (u64)hpet_readl(HPET_COUNTER);
+
 	/*
 	 * Read the current state of the lock and HPET value atomically.
 	 */