From: Brad Peters <bpeters@redhat.com>
Date: Tue, 29 Jul 2008 14:53:12 -0400
Subject: [ppc] adds crashdump shutdown hooks
Message-id: 20080729185312.8726.44776.sendpatchset@squad5-lp1.lab.bos.redhat.com
O-Subject: [PATCH RHEL 5.3 2/5 v2] Adds crashdump shutdown hooks (into default_machine_crash_shutdown)
Bugzilla: 442409
RH-Acked-by: David Howells <dhowells@redhat.com>

RHBZ#:
======
https://bugzilla.redhat.com/show_bug.cgi?id=442409

Description:
===========
This adds hooks into the default_machine_crash_shutdown so drivers can
register a function to be run in the first kernel before we hand off
to the second kernel.  This should only be used in exceptional
circumstances, like where the device can't be reset in the second
kernel alone (as is the case with eHEA).  To emphasize this, the
number of handles allowed to be registered is currently #def to 1.

This uses the setjmp/longjmp code to call out to the registered hooks,
so any bogus exceptions we encounter will hopefully be recoverable.

This patch was originally written by Michael Neuling. Backported this
patch to RHEL 5.x kernel base.

Upstream Status:
================
Patch posted:
http://ozlabs.org/pipermail/linuxppc-dev/2008-January/050140.html

===============================================================

Brad Peters 1-978-392-1000 x 23183
IBM on-site partner.

Proposed Patch:
===============
This patch is based on 2.6.18-94.el5

diff --git a/arch/powerpc/kernel/crash.c b/arch/powerpc/kernel/crash.c
index f1f602e..f35cb59 100644
--- a/arch/powerpc/kernel/crash.c
+++ b/arch/powerpc/kernel/crash.c
@@ -33,6 +33,8 @@
 #include <asm/lmb.h>
 #include <asm/firmware.h>
 #include <asm/smp.h>
+#include <asm/system.h>
+#include <asm/setjmp.h>
 
 #ifdef DEBUG
 #include <asm/udbg.h>
@@ -101,6 +103,11 @@ static void crash_save_this_cpu(struct pt_regs *regs, int cpu)
 	final_note(buf);
 }
 
+#define CRASH_HANDLER_MAX 1
+/* NULL terminated list of shutdown handles */
+static crash_shutdown_t crash_shutdown_handles[CRASH_HANDLER_MAX+1];
+static DEFINE_SPINLOCK(crash_handlers_lock);
+
 #ifdef CONFIG_SMP
 static atomic_t enter_on_soft_reset = ATOMIC_INIT(0);
 
@@ -341,9 +348,72 @@ static inline void crash_kexec_stop_spus(void)
 }
 #endif /* CONFIG_SPU_BASE */
 
+/*
+ * Register a function to be called on shutdown.  Only use this if you
+ * can't reset your device in the second kernel.
+ */
+int crash_shutdown_register(crash_shutdown_t handler)
+{
+	unsigned int i, rc;
+
+	spin_lock(&crash_handlers_lock);
+	for (i = 0 ; i < CRASH_HANDLER_MAX; i++)
+		if (!crash_shutdown_handles[i]) {
+			/* Insert handle at first empty entry */
+			crash_shutdown_handles[i] = handler;
+			rc = 0;
+			break;
+		}
+
+	if (i == CRASH_HANDLER_MAX) {
+		printk(KERN_ERR "Crash shutdown handles full, "
+		       "not registered.\n");
+		rc = 1;
+	}
+
+	spin_unlock(&crash_handlers_lock);
+	return rc;
+}
+EXPORT_SYMBOL(crash_shutdown_register);
+
+int crash_shutdown_unregister(crash_shutdown_t handler)
+{
+	unsigned int i, rc;
+
+	spin_lock(&crash_handlers_lock);
+	for (i = 0 ; i < CRASH_HANDLER_MAX; i++)
+		if (crash_shutdown_handles[i] == handler)
+			break;
+
+	if (i == CRASH_HANDLER_MAX) {
+		printk(KERN_ERR "Crash shutdown handle not found\n");
+		rc = 1;
+	} else {
+		/* Shift handles down */
+		for (; crash_shutdown_handles[i]; i++)
+			crash_shutdown_handles[i] =
+				crash_shutdown_handles[i+1];
+		rc = 0;
+	}
+
+	spin_unlock(&crash_handlers_lock);
+	return rc;
+}
+EXPORT_SYMBOL(crash_shutdown_unregister);
+
+static unsigned long crash_shutdown_buf[JMP_BUF_LEN];
+
+static int handle_fault(struct pt_regs *regs)
+{
+	longjmp(crash_shutdown_buf, 1);
+	return 0;
+}
+
 void default_machine_crash_shutdown(struct pt_regs *regs)
 {
-	unsigned int irq;
+	unsigned int i;
+	int (*old_handler)(struct pt_regs *regs);
+
 
 	/*
 	 * This function is only called after the system
@@ -357,15 +427,41 @@ void default_machine_crash_shutdown(struct pt_regs *regs)
 	 */
 	local_irq_disable();
 
-	for_each_irq(irq) {
-		struct irq_desc *desc = irq_desc + irq;
+	for_each_irq(i) {
+		struct irq_desc *desc = irq_desc + i;
 
 		if (desc->status & IRQ_INPROGRESS)
-			desc->chip->eoi(irq);
+			desc->chip->eoi(i);
 
 		if (!(desc->status & IRQ_DISABLED))
-			desc->chip->disable(irq);
+			desc->chip->disable(i);
+	}
+
+	/*
+	 * Call registered shutdown routines savely.  Swap out
+	 * __debugger_fault_handler, and replace on exit.
+	 */
+	old_handler = __debugger_fault_handler;
+	__debugger_fault_handler = handle_fault;
+	for (i = 0; crash_shutdown_handles[i]; i++) {
+		if (setjmp(crash_shutdown_buf) == 0) {
+			/*
+			 * Insert syncs and delay to ensure
+			 * instructions in the dangerous region don't
+			 * leak away from this protected region.
+			 */
+			asm volatile("sync; isync");
+			/* dangerous region */
+			crash_shutdown_handles[i]();
+			asm volatile("sync; isync");
+			/* 
+			 * wait a little while to see if we get a
+			 * machine check 
+			 */
+			__delay(200);
+		}
 	}
+	__debugger_fault_handler = old_handler;
 
 	/*
 	 * Make a note of crashing cpu. Will be used in machine_kexec
diff --git a/include/asm-powerpc/kexec.h b/include/asm-powerpc/kexec.h
index 11cbdf8..a6178dd 100644
--- a/include/asm-powerpc/kexec.h
+++ b/include/asm-powerpc/kexec.h
@@ -125,6 +125,9 @@ struct pt_regs;
 extern void default_machine_kexec(struct kimage *image);
 extern int default_machine_kexec_prepare(struct kimage *image);
 extern void default_machine_crash_shutdown(struct pt_regs *regs);
+typedef void (*crash_shutdown_t)(void);
+extern int crash_shutdown_register(crash_shutdown_t handler);
+extern int crash_shutdown_unregister(crash_shutdown_t handler);
 
 extern void machine_kexec_simple(struct kimage *image);
 extern void crash_kexec_secondary(struct pt_regs *regs);