[PATCH][WIP][RFC] powerpc: fixup lwsync at runtime

[Kumar Gala]

This is a work in progress towards make lwsync fixed up at runtime.  The
patch is based (requires) the module refactoring patch.  Some of this code
should be reworked based on the code patching work from Michael.  Also,
ppc64 and vdso support needs a bit of cleaning up.

Some questions:
* How do we determine at runtime if we need to convert sync to lwsync.
This is simliar to cpu_feature but has the issue that cpu_feature is about
either having code in or not.

* Do we simplify the section of fixup information and remove the mask and
value since we don't use them and they double the size of the section

- k

 arch/powerpc/kernel/cputable.c    |   17 +++++++++++++++--
 arch/powerpc/kernel/module.c      |    5 +++++
 arch/powerpc/kernel/setup_32.c    |    3 +++
 arch/powerpc/kernel/vmlinux.lds.S |    6 ++++++
 include/asm-powerpc/synch.h       |   26 +++++++++++++++++---------
 include/asm-powerpc/system.h      |    2 +-
 6 files changed, 47 insertions(+), 12 deletions(-)

diff --git a/arch/powerpc/kernel/cputable.c b/arch/powerpc/kernel/cputable.c
index c5397c1..4905f1d 100644
--- a/arch/powerpc/kernel/cputable.c
+++ b/arch/powerpc/kernel/cputable.c
@@ -1600,7 +1600,7 @@ struct cpu_spec * __init identify_cpu(unsigned long offset, unsigned int pvr)
 	return NULL;
 }

-void do_feature_fixups(unsigned long value, void *fixup_start, void *fixup_end)
+static inline void __do_feature_fixups(unsigned long value, void *fixup_start, void *fixup_end, u32 insn)
 {
 	struct fixup_entry {
 		unsigned long	mask;
@@ -1625,7 +1625,7 @@ void do_feature_fixups(unsigned long value, void *fixup_start, void *fixup_end)
 		pend = ((unsigned int *)fcur) + (fcur->end_off / 4);

 		for (p = pstart; p < pend; p++) {
-			*p = 0x60000000u;
+			*p = insn;
 			asm volatile ("dcbst 0, %0" : : "r" (p));
 		}
 		asm volatile ("sync" : : : "memory");
@@ -1634,3 +1634,16 @@ void do_feature_fixups(unsigned long value, void *fixup_start, void *fixup_end)
 		asm volatile ("sync; isync" : : : "memory");
 	}
 }
+
+void do_feature_fixups(unsigned long value, void *fixup_start, void *fixup_end)
+{
+	__do_feature_fixups(value, fixup_start, fixup_end, 0x60000000u);
+}
+
+#define PPC_LWSYNC_INSTR	0x7c2004ac
+void do_lwsync_fixups(void *fixup_start, void *fixup_end)
+{
+#if defined(__powerpc64__) || defined(CONFIG_PPC_E500MC)
+	__do_feature_fixups(0, fixup_start, fixup_end, PPC_LWSYNC_INSTR);
+#endif
+}
diff --git a/arch/powerpc/kernel/module.c b/arch/powerpc/kernel/module.c
index 40dd52d..34905b8 100644
--- a/arch/powerpc/kernel/module.c
+++ b/arch/powerpc/kernel/module.c
@@ -86,6 +86,11 @@ int module_finalize(const Elf_Ehdr *hdr,
 				  (void *)sect->sh_addr + sect->sh_size);
 #endif

+	sect = find_section(hdr, sechdrs, "__lwsync_fixup");
+	if (sect != NULL)
+		do_lwsync_fixups((void *)sect->sh_addr,
+				 (void *)sect->sh_addr + sect->sh_size);
+
 	return 0;
 }

diff --git a/arch/powerpc/kernel/setup_32.c b/arch/powerpc/kernel/setup_32.c
index 9e83add..d1e498f 100644
--- a/arch/powerpc/kernel/setup_32.c
+++ b/arch/powerpc/kernel/setup_32.c
@@ -101,6 +101,9 @@ unsigned long __init early_init(unsigned long dt_ptr)
 			  PTRRELOC(&__start___ftr_fixup),
 			  PTRRELOC(&__stop___ftr_fixup));

+	do_lwsync_fixups(PTRRELOC(&__start___lwsync_fixup),
+			 PTRRELOC(&__stop___lwsync_fixup));
+
 	return KERNELBASE + offset;
 }

diff --git a/arch/powerpc/kernel/vmlinux.lds.S b/arch/powerpc/kernel/vmlinux.lds.S
index 0c3000b..d4d351a 100644
--- a/arch/powerpc/kernel/vmlinux.lds.S
+++ b/arch/powerpc/kernel/vmlinux.lds.S
@@ -127,6 +127,12 @@ SECTIONS
 		*(__ftr_fixup)
 		__stop___ftr_fixup = .;
 	}
+	. = ALIGN(8);
+	__lwsync_fixup : AT(ADDR(__lwsync_fixup) - LOAD_OFFSET) {
+		__start___lwsync_fixup = .;
+		*(__lwsync_fixup)
+		__stop___lwsync_fixup = .;
+	}
 #ifdef CONFIG_PPC64
 	. = ALIGN(8);
 	__fw_ftr_fixup : AT(ADDR(__fw_ftr_fixup) - LOAD_OFFSET) {
diff --git a/include/asm-powerpc/synch.h b/include/asm-powerpc/synch.h
index 42a1ef5..c031808 100644
--- a/include/asm-powerpc/synch.h
+++ b/include/asm-powerpc/synch.h
@@ -3,20 +3,28 @@
 #ifdef __KERNEL__

 #include <linux/stringify.h>
+#include <asm/asm-compat.h>

-#if defined(__powerpc64__) || defined(CONFIG_PPC_E500MC)
-#define __SUBARCH_HAS_LWSYNC
-#endif
+#ifndef __ASSEMBLY__
+extern unsigned int __start___lwsync_fixup, __stop___lwsync_fixup;
+extern void do_lwsync_fixups(void *fixup_start, void *fixup_end);
+#endif /* __ASSEMBLY__ */

-#ifdef __SUBARCH_HAS_LWSYNC
-#    define LWSYNC	lwsync
-#else
-#    define LWSYNC	sync
-#endif
+#define BEGIN_LWSYNC_SECTION_NESTED(label)	label:
+#define BEGIN_LWSYNC_SECTION		BEGIN_LWSYNC_SECTION_NESTED(97)
+#define END_LWSYNC_SECTION_NESTED(msk, val, label) \
+	MAKE_FTR_SECTION_ENTRY(msk, val, label, __lwsync_fixup)
+#define END_LWSYNC_SECTION \
+	END_LWSYNC_SECTION_NESTED(1, 1, 97)
+
+#    define LWSYNC		\
+	BEGIN_LWSYNC_SECTION;	\
+	sync;			\
+	END_LWSYNC_SECTION;

 #ifdef CONFIG_SMP
 #define ISYNC_ON_SMP	"\n\tisync\n"
-#define LWSYNC_ON_SMP	__stringify(LWSYNC) "\n"
+#define LWSYNC_ON_SMP	stringify_in_c(LWSYNC) "\n"
 #else
 #define ISYNC_ON_SMP
 #define LWSYNC_ON_SMP
diff --git a/include/asm-powerpc/system.h b/include/asm-powerpc/system.h
index df781ad..15218bb 100644
--- a/include/asm-powerpc/system.h
+++ b/include/asm-powerpc/system.h
@@ -34,7 +34,7 @@
  * SMP since it is only used to order updates to system memory.
  */
 #define mb()   __asm__ __volatile__ ("sync" : : : "memory")
-#define rmb()  __asm__ __volatile__ (__stringify(LWSYNC) : : : "memory")
+#define rmb()  __asm__ __volatile__ (stringify_in_c(LWSYNC) : : : "memory")
 #define wmb()  __asm__ __volatile__ ("sync" : : : "memory")
 #define read_barrier_depends()  do { } while(0)

-- 
1.5.5.1