[PATCH] Kexec: Kexec on panic fix with nmi watchdog enabled

o Problem: Kexec on panic hangs if first kernel is booted with nmi_watchdog
  command line parameter. This problem occurs because kexec crash shutdown
  code replaces the NMI callback handler. This handler saves the cpu register
  states and halts the cpu. If system is booted with nmi_watchdog parameter,
  then crashing cpu also runs this nmi handler and halts itself.

o This patch fixes the problem by keeping a track of crashing cpu and not
  executing the new nmi handler on crashing cpu.

o There is a dependence on smp_processor_id() function which might return
  insane value for cpu, if cpu field of thread_info is corrupted.

Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

diff --git a/arch/i386/kernel/crash.c b/arch/i386/kernel/crash.c
index 31e077bb0caad729df75f65b4bcb7a1e3cb9b9da..a021681d21f8a61b5ad2ae180a9182c8a0f514ce 100644 (file)
--- a/arch/i386/kernel/crash.c
+++ b/arch/i386/kernel/crash.c
@@ -28,6 +28,8 @@
 
 
 note_buf_t crash_notes[NR_CPUS];
+/* This keeps a track of which one is crashing cpu. */
+static int crashing_cpu;
 
 static u32 *append_elf_note(u32 *buf,
        char *name, unsigned type, void *data, size_t data_len)
@@ -113,6 +115,13 @@ static atomic_t waiting_for_crash_ipi;
 static int crash_nmi_callback(struct pt_regs *regs, int cpu)
 {
        struct pt_regs fixed_regs;
+
+       /* Don't do anything if this handler is invoked on crashing cpu.
+        * Otherwise, system will completely hang. Crashing cpu can get
+        * an NMI if system was initially booted with nmi_watchdog parameter.
+        */
+       if (cpu == crashing_cpu)
+               return 1;
        local_irq_disable();
 
        /* CPU does not save ss and esp on stack if execution is already
@@ -187,6 +196,9 @@ void machine_crash_shutdown(void)
         */
        /* The kernel is broken so disable interrupts */
        local_irq_disable();
+
+       /* Make a note of crashing cpu. Will be used in NMI callback.*/
+       crashing_cpu = smp_processor_id();
        nmi_shootdown_cpus();
        lapic_shutdown();
 #if defined(CONFIG_X86_IO_APIC)