s390/nmi: use smp_emergency_stop instead of smp_send_stop

The smp_send_stop() function can be called from s390_handle_damage
while DAT is off. This happens if a machine check indicates that
kernel gprs or control registers can not be restored. The function
smp_send_stop reenables DAT via __load_psw_mask. That should work
for the case of lost kernel gprs and the system will do the expected
stop of all CPUs. But if control registers are lost, in particular
CR13 with the home space ASCE, interesting secondary crashes may
occur.

Make smp_emergency_stop callable from nmi.c and remove the cpumask
argument. Replace the smp_send_stop call with smp_emergency_stop in
the s390_handle_damage function.

In addition add notrace and NOKPROBE_SYMBOL annotations for all
functions required for the emergency shutdown.

Reviewed-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

diff --git a/arch/s390/include/asm/smp.h b/arch/s390/include/asm/smp.h
index 3deb134587b744f41164c58255d9f2bba3c92dc0..3470274a985cbe29b690b5922d500ac2c4bf86c4 100644 (file)
--- a/arch/s390/include/asm/smp.h
+++ b/arch/s390/include/asm/smp.h
@@ -27,6 +27,7 @@ extern void arch_send_call_function_ipi_mask(const struct cpumask *mask);
 
 extern void smp_call_online_cpu(void (*func)(void *), void *);
 extern void smp_call_ipl_cpu(void (*func)(void *), void *);
+extern void smp_emergency_stop(void);
 
 extern int smp_find_processor_id(u16 address);
 extern int smp_store_status(int cpu);
@@ -52,6 +53,10 @@ static inline void smp_call_online_cpu(void (*func)(void *), void *data)
        func(data);
 }
 
+static inline void smp_emergency_stop(void)
+{
+}
+
 static inline int smp_find_processor_id(u16 address) { return 0; }
 static inline int smp_store_status(int cpu) { return 0; }
 static inline int smp_vcpu_scheduled(int cpu) { return 1; }

diff --git a/arch/s390/kernel/nmi.c b/arch/s390/kernel/nmi.c
index 31d03a84126c5a96545b6496628ea38a5cd046dc..15e28eefe7e9247dff483ef8d936e23b053ae66b 100644 (file)
--- a/arch/s390/kernel/nmi.c
+++ b/arch/s390/kernel/nmi.c
@@ -12,6 +12,7 @@
 #include <linux/init.h>
 #include <linux/errno.h>
 #include <linux/hardirq.h>
+#include <linux/kprobes.h>
 #include <linux/time.h>
 #include <linux/module.h>
 #include <linux/sched/signal.h>
@@ -38,12 +39,13 @@ struct mcck_struct {
 
 static DEFINE_PER_CPU(struct mcck_struct, cpu_mcck);
 
-static void s390_handle_damage(void)
+static notrace void s390_handle_damage(void)
 {
-       smp_send_stop();
+       smp_emergency_stop();
        disabled_wait((unsigned long) __builtin_return_address(0));
        while (1);
 }
+NOKPROBE_SYMBOL(s390_handle_damage);
 
 /*
  * Main machine check handler function. Will be called with interrupts enabled
@@ -275,6 +277,7 @@ static int notrace s390_validate_registers(union mci mci, int umode)
 
        return kill_task;
 }
+NOKPROBE_SYMBOL(s390_validate_registers);
 
 /*
  * Backup the guest's machine check info to its description block
@@ -300,6 +303,7 @@ static void notrace s390_backup_mcck_info(struct pt_regs *regs)
        mcck_backup->failing_storage_address
                        = S390_lowcore.failing_storage_address;
 }
+NOKPROBE_SYMBOL(s390_backup_mcck_info);
 
 #define MAX_IPD_COUNT  29
 #define MAX_IPD_TIME   (5 * 60 * USEC_PER_SEC) /* 5 minutes */
@@ -443,6 +447,7 @@ void notrace s390_do_machine_check(struct pt_regs *regs)
        clear_cpu_flag(CIF_MCCK_GUEST);
        nmi_exit();
 }
+NOKPROBE_SYMBOL(s390_do_machine_check);
 
 static int __init machine_check_init(void)
 {

diff --git a/arch/s390/kernel/smp.c b/arch/s390/kernel/smp.c
index cc04b74fbb84c69a2c2c8aead224450da90f0034..2dba3e88a972d08497b05b7e50e77155dcb233b1 100644 (file)
--- a/arch/s390/kernel/smp.c
+++ b/arch/s390/kernel/smp.c
@@ -36,6 +36,7 @@
 #include <linux/sched/task_stack.h>
 #include <linux/crash_dump.h>
 #include <linux/memblock.h>
+#include <linux/kprobes.h>
 #include <asm/asm-offsets.h>
 #include <asm/diag.h>
 #include <asm/switch_to.h>
@@ -422,13 +423,17 @@ void smp_yield_cpu(int cpu)
  * Send cpus emergency shutdown signal. This gives the cpus the
  * opportunity to complete outstanding interrupts.
  */
-static void smp_emergency_stop(cpumask_t *cpumask)
+void notrace smp_emergency_stop(void)
 {
+       cpumask_t cpumask;
        u64 end;
        int cpu;
 
+       cpumask_copy(&cpumask, cpu_online_mask);
+       cpumask_clear_cpu(smp_processor_id(), &cpumask);
+
        end = get_tod_clock() + (1000000UL << 12);
-       for_each_cpu(cpu, cpumask) {
+       for_each_cpu(cpu, &cpumask) {
                struct pcpu *pcpu = pcpu_devices + cpu;
                set_bit(ec_stop_cpu, &pcpu->ec_mask);
                while (__pcpu_sigp(pcpu->address, SIGP_EMERGENCY_SIGNAL,
@@ -437,21 +442,21 @@ static void smp_emergency_stop(cpumask_t *cpumask)
                        cpu_relax();
        }
        while (get_tod_clock() < end) {
-               for_each_cpu(cpu, cpumask)
+               for_each_cpu(cpu, &cpumask)
                        if (pcpu_stopped(pcpu_devices + cpu))
-                               cpumask_clear_cpu(cpu, cpumask);
-               if (cpumask_empty(cpumask))
+                               cpumask_clear_cpu(cpu, &cpumask);
+               if (cpumask_empty(&cpumask))
                        break;
                cpu_relax();
        }
 }
+NOKPROBE_SYMBOL(smp_emergency_stop);
 
 /*
  * Stop all cpus but the current one.
  */
 void smp_send_stop(void)
 {
-       cpumask_t cpumask;
        int cpu;
 
        /* Disable all interrupts/machine checks */
@@ -459,17 +464,16 @@ void smp_send_stop(void)
        trace_hardirqs_off();
 
        debug_set_critical();
-       cpumask_copy(&cpumask, cpu_online_mask);
-       cpumask_clear_cpu(smp_processor_id(), &cpumask);
 
        if (oops_in_progress)
-               smp_emergency_stop(&cpumask);
+               smp_emergency_stop();
 
        /* stop all processors */
-       for_each_cpu(cpu, &cpumask) {
-               struct pcpu *pcpu = pcpu_devices + cpu;
-               pcpu_sigp_retry(pcpu, SIGP_STOP, 0);
-               while (!pcpu_stopped(pcpu))
+       for_each_online_cpu(cpu) {
+               if (cpu == smp_processor_id())
+                       continue;
+               pcpu_sigp_retry(pcpu_devices + cpu, SIGP_STOP, 0);
+               while (!pcpu_stopped(pcpu_devices + cpu))
                        cpu_relax();
        }
 }