[Dle-develop] [RFC][PATCH -mmotm 2/4] Remove mutex_lock from pstore in panic path

[Seiji A. <sei...@hd...>]

Hi,
 
This patch removes a function call chain of kmsg_dump and mutex_lock of pstore from panic path 
 so that both pstore and APEI storage backend can work reliably in panic path.

 kernel/kexec.c
   - Move kmsg_dump(KMSG_DUMP_PANIC) behind machine_crash_shutdown() so that pstore can
     work with one cpu.

 fs/pstore/platform.c
   pstore_dump()
   - Remove mutex_lock from panic path

 kernel/printk.c
   - Add spin_lock_init(&logbuf) for avoiding dead lock in panic path

TODO:
  APEI storage backend will work with this patch.
  However, I don't have any access to servers capable of APEI storage backend.
  Please help to test my patch.

 Signed-off-by: Seiji Aguchi <sei...@hd...>

---
 fs/pstore/platform.c |    3 ++-
 kernel/panic.c       |    4 ++--
 2 files changed, 4 insertions(+), 3 deletions(-)

diff --git a/fs/pstore/platform.c b/fs/pstore/platform.c index 85e0a9c..061911c 100644
--- a/fs/pstore/platform.c
+++ b/fs/pstore/platform.c
@@ -75,6 +75,7 @@ static void pstore_dump(struct kmsg_dumper *dumper,
 		why = "Unknown";
 
 	switch (reason) {
+	case KMSG_DUMP_PANIC:
 	case KMSG_DUMP_KEXEC:
 		/* Skip if there is no driver or there is a driver calling
 		   pstore_register() */
@@ -114,7 +115,7 @@ static void pstore_dump(struct kmsg_dumper *dumper,
 		total += l1_cpy + l2_cpy;
 	}
 
-	if (reason != KMSG_DUMP_KEXEC)
+	if (reason != KMSG_DUMP_PANIC && reason != KMSG_DUMP_KEXEC)
 		mutex_unlock(&psinfo->buf_mutex);
 }
 
diff --git a/kernel/panic.c b/kernel/panic.c index 6923167..a9dce52 100644
--- a/kernel/panic.c
+++ b/kernel/panic.c
@@ -11,6 +11,7 @@
 #include <linux/debug_locks.h>
 #include <linux/interrupt.h>
 #include <linux/kmsg_dump.h>
+#include <linux/pstore.h>
 #include <linux/kallsyms.h>
 #include <linux/notifier.h>
 #include <linux/module.h>
@@ -88,14 +89,13 @@ NORET_TYPE void panic(const char * fmt, ...)
 	 */
 	crash_kexec(NULL);
 
-	kmsg_dump(KMSG_DUMP_PANIC);
-
 	/*
 	 * Note smp_send_stop is the usual smp shutdown function, which
 	 * unfortunately means it may not be hardened to work in a panic
 	 * situation.
 	 */
 	smp_send_stop();
+	pstore_kmsg_dump_in_interrupt(KMSG_DUMP_PANIC);
 
 	atomic_notifier_call_chain(&panic_notifier_list, 0, buf);
 
--
1.7.1