From: Janice M. Girouard <jgirouar@redhat.com>
Subject: [RHEL 5.1 FEATURE] bz #228099 1/12 PPC: Cell Platform Base kernel  support
Date: Tue, 17 Apr 2007 14:54:12 -0400 (Eastern Daylight Time)
Bugzilla: 228099
Message-Id: <Pine.WNT.4.64.0704161557170.5540@IBM-3MTQI3AXJFW>
Changelog: [ppc64] Cell Platform Base kernel support



RHBZ#:
------
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=228099

Description:
------------
1 of 12 patches by Ben Herrenschmidt of IBM required for supporting IBM 
customers purchasing Cell blade configurations.  

Without this patch, there are two copies of the same functions that apply 
feature fixups, one for CPU features and one for firmware features.  In 
addition, they are both in assembly and with separate implementations for 
32 and 64 bit platforms.  identify_cpu() is also implemented in assembly 
and separate for 32 and 64 bit.  

This patch replaces duplicate functions with a pair of C functions.  The 
call sites are slightly moved on ppc64 as well to be called from C instead 
of fro assembly, though it's a very small change, and thus shouldn't cause 
any problem.

RHEL Version Found:
-------------------
feature enhancement for 5.1.

Upstream Status:
----------------
This has been accepted into the kernel and can be viewed in the 2.6.20.3 
kernel.

Test Status:
------------
This patch was tested by Benjamin herrenschmidt of IBM.  It has been
ack'd by Olof Johansson and signed off by Paul Mackerras.

Proposed Patch:
----------------
Please review and ACK for RHEL 5.1.
-

Index: linux-2.6/arch/powerpc/kernel/cputable.c
===================================================================
--- linux-2.6.orig/arch/powerpc/kernel/cputable.c	2007-01-24 16:01:50.000000000 +1100
+++ linux-2.6/arch/powerpc/kernel/cputable.c	2007-01-24 16:02:12.000000000 +1100
@@ -18,6 +18,7 @@
 
 #include <asm/oprofile_impl.h>
 #include <asm/cputable.h>
+#include <asm/prom.h>		/* for PTRRELOC on ARCH=ppc */
 
 struct cpu_spec* cur_cpu_spec = NULL;
 EXPORT_SYMBOL(cur_cpu_spec);
@@ -67,7 +68,7 @@
 #define PPC_FEATURE_SPE_COMP	0
 #endif
 
-struct cpu_spec	cpu_specs[] = {
+static struct cpu_spec cpu_specs[] = {
 #ifdef CONFIG_PPC64
 	{	/* Power3 */
 		.pvr_mask		= 0xffff0000,
@@ -1120,3 +1121,72 @@
 #endif /* !CLASSIC_PPC */
 #endif /* CONFIG_PPC32 */
 };
+
+struct cpu_spec *identify_cpu(unsigned long offset)
+{
+	struct cpu_spec *s = cpu_specs;
+	struct cpu_spec **cur = &cur_cpu_spec;
+	unsigned int pvr = mfspr(SPRN_PVR);
+	int i;
+
+	s = PTRRELOC(s);
+	cur = PTRRELOC(cur);
+
+	if (*cur != NULL)
+		return PTRRELOC(*cur);
+
+	for (i = 0; i < ARRAY_SIZE(cpu_specs); i++,s++)
+		if ((pvr & s->pvr_mask) == s->pvr_value) {
+			*cur = cpu_specs + i;
+#ifdef CONFIG_PPC64
+			/* ppc64 expects identify_cpu to also call setup_cpu
+			 * for that processor. I will consolidate that at a
+			 * later time, for now, just use our friend #ifdef.
+			 * we also don't need to PTRRELOC the function pointer
+			 * on ppc64 as we are running at 0 in real mode.
+			 */
+			if (s->cpu_setup) {
+				s->cpu_setup(offset, s);
+			}
+#endif /* CONFIG_PPC64 */
+			return s;
+		}
+	BUG();
+	return NULL;
+}
+
+void do_feature_fixups(unsigned long offset, unsigned long value,
+		       void *fixup_start, void *fixup_end)
+{
+	struct fixup_entry {
+		unsigned long	mask;
+		unsigned long	value;
+		unsigned int	*start;
+		unsigned int	*end;
+	} *fcur, *fend;
+
+	fcur = fixup_start;
+	fend = fixup_end;
+
+	for (; fcur < fend; fcur++) {
+		unsigned int *pstart, *pend, *p;
+
+		if ((value & fcur->mask) == fcur->value)
+			continue;
+
+		/* These PTRRELOCs will disappear once the new scheme for
+		 * modules and vdso is implemented
+		 */
+		pstart = PTRRELOC(fcur->start);
+		pend = PTRRELOC(fcur->end);
+
+		for (p = pstart; p < pend; p++) {
+			*p = 0x60000000u;
+			asm volatile ("dcbst 0, %0" : : "r" (p));
+		}
+		asm volatile ("sync" : : : "memory");
+		for (p = pstart; p < pend; p++)
+			asm volatile ("icbi 0,%0" : : "r" (p));
+		asm volatile ("sync; isync" : : : "memory");
+	}
+}
Index: linux-2.6/arch/powerpc/kernel/head_64.S
===================================================================
--- linux-2.6.orig/arch/powerpc/kernel/head_64.S	2007-01-24 16:01:50.000000000 +1100
+++ linux-2.6/arch/powerpc/kernel/head_64.S	2007-01-24 16:02:12.000000000 +1100
@@ -1552,11 +1552,6 @@
 	li	r0,0
 	stdu	r0,-STACK_FRAME_OVERHEAD(r1)
 
-	LOAD_REG_IMMEDIATE(r3,cpu_specs)
-	LOAD_REG_IMMEDIATE(r4,cur_cpu_spec)
-	li	r5,0
-	bl	.identify_cpu
-
 	LOAD_REG_IMMEDIATE(r2,__toc_start)
 	addi	r2,r2,0x4000
 	addi	r2,r2,0x4000
@@ -1925,13 +1920,6 @@
 	addi	r2,r2,0x4000
 	add	r2,r2,r26
 
-	LOAD_REG_IMMEDIATE(r3, cpu_specs)
-	add	r3,r3,r26
-	LOAD_REG_IMMEDIATE(r4,cur_cpu_spec)
-	add	r4,r4,r26
-	mr	r5,r26
-	bl	.identify_cpu
-
 	/* Save some low level config HIDs of CPU0 to be copied to
 	 * other CPUs later on, or used for suspend/resume
 	 */
@@ -1967,12 +1955,6 @@
 	li	r0,0
 	stdu	r0,-STACK_FRAME_OVERHEAD(r1)
 
-	/* Apply the CPUs-specific fixups (nop out sections not relevant
-	 * to this CPU
-	 */
-	li	r3,0
-	bl	.do_cpu_ftr_fixups
-
 	/* ptr to current */
 	LOAD_REG_IMMEDIATE(r4, init_task)
 	std	r4,PACACURRENT(r13)
Index: linux-2.6/arch/powerpc/kernel/misc_32.S
===================================================================
--- linux-2.6.orig/arch/powerpc/kernel/misc_32.S	2007-01-24 16:01:50.000000000 +1100
+++ linux-2.6/arch/powerpc/kernel/misc_32.S	2007-01-24 16:02:12.000000000 +1100
@@ -102,80 +102,6 @@
 	blr
 
 /*
- * identify_cpu,
- * called with r3 = data offset and r4 = CPU number
- * doesn't change r3
- */
-_GLOBAL(identify_cpu)
-	addis	r8,r3,cpu_specs@ha
-	addi	r8,r8,cpu_specs@l
-	mfpvr	r7
-1:
-	lwz	r5,CPU_SPEC_PVR_MASK(r8)
-	and	r5,r5,r7
-	lwz	r6,CPU_SPEC_PVR_VALUE(r8)
-	cmplw	0,r6,r5
-	beq	1f
-	addi	r8,r8,CPU_SPEC_ENTRY_SIZE
-	b	1b
-1:
-	addis	r6,r3,cur_cpu_spec@ha
-	addi	r6,r6,cur_cpu_spec@l
-	sub	r8,r8,r3
-	stw	r8,0(r6)
-	blr
-
-/*
- * do_cpu_ftr_fixups - goes through the list of CPU feature fixups
- * and writes nop's over sections of code that don't apply for this cpu.
- * r3 = data offset (not changed)
- */
-_GLOBAL(do_cpu_ftr_fixups)
-	/* Get CPU 0 features */
-	addis	r6,r3,cur_cpu_spec@ha
-	addi	r6,r6,cur_cpu_spec@l
-	lwz	r4,0(r6)
-	add	r4,r4,r3
-	lwz	r4,CPU_SPEC_FEATURES(r4)
-
-	/* Get the fixup table */
-	addis	r6,r3,__start___ftr_fixup@ha
-	addi	r6,r6,__start___ftr_fixup@l
-	addis	r7,r3,__stop___ftr_fixup@ha
-	addi	r7,r7,__stop___ftr_fixup@l
-
-	/* Do the fixup */
-1:	cmplw	0,r6,r7
-	bgelr
-	addi	r6,r6,16
-	lwz	r8,-16(r6)	/* mask */
-	and	r8,r8,r4
-	lwz	r9,-12(r6)	/* value */
-	cmplw	0,r8,r9
-	beq	1b
-	lwz	r8,-8(r6)	/* section begin */
-	lwz	r9,-4(r6)	/* section end */
-	subf.	r9,r8,r9
-	beq	1b
-	/* write nops over the section of code */
-	/* todo: if large section, add a branch at the start of it */
-	srwi	r9,r9,2
-	mtctr	r9
-	add	r8,r8,r3
-	lis	r0,0x60000000@h	/* nop */
-3:	stw	r0,0(r8)
-	andi.	r10,r4,CPU_FTR_SPLIT_ID_CACHE@l
-	beq	2f
-	dcbst	0,r8		/* suboptimal, but simpler */
-	sync
-	icbi	0,r8
-2:	addi	r8,r8,4
-	bdnz	3b
-	sync			/* additional sync needed on g4 */
-	isync
-	b	1b
-
-/*
  * call_setup_cpu - call the setup_cpu function for this cpu
  * r3 = data offset, r24 = cpu number
  *
Index: linux-2.6/arch/powerpc/kernel/setup_32.c
===================================================================
--- linux-2.6.orig/arch/powerpc/kernel/setup_32.c	2007-01-24 16:01:50.000000000 +1100
+++ linux-2.6/arch/powerpc/kernel/setup_32.c	2007-01-24 16:02:12.000000000 +1100
@@ -95,6 +95,7 @@
 unsigned long __init early_init(unsigned long dt_ptr)
 {
 	unsigned long offset = reloc_offset();
+	struct cpu_spec *spec;
 
 	/* First zero the BSS -- use memset_io, some platforms don't have
 	 * caches on yet */
@@ -104,8 +105,11 @@
 	 * Identify the CPU type and fix up code sections
 	 * that depend on which cpu we have.
 	 */
-	identify_cpu(offset, 0);
-	do_cpu_ftr_fixups(offset);
+	spec = identify_cpu(offset);
+
+	do_feature_fixups(offset, spec->cpu_features,
+			  PTRRELOC(&__start___ftr_fixup),
+			  PTRRELOC(&__stop___ftr_fixup));
 
 	return KERNELBASE + offset;
 }
Index: linux-2.6/arch/powerpc/kernel/setup_64.c
===================================================================
--- linux-2.6.orig/arch/powerpc/kernel/setup_64.c	2007-01-24 16:01:50.000000000 +1100
+++ linux-2.6/arch/powerpc/kernel/setup_64.c	2007-01-24 16:03:37.000000000 +1100
@@ -176,6 +176,9 @@
 
 void __init early_setup(unsigned long dt_ptr)
 {
+	/* Identify CPU type */
+	identify_cpu(0);
+
 	/* Assume we're on cpu 0 for now. Don't write to the paca yet! */
 	setup_paca(0);
 
@@ -354,6 +357,12 @@
 {
 	DBG(" -> setup_system()\n");
 
+	/* Apply the CPUs-specific and firmware specific fixups to kernel
+	 * text (nop out sections not relevant to this CPU or this firmware)
+	 */
+	do_feature_fixups(0, cur_cpu_spec->cpu_features,
+			  &__start___ftr_fixup, &__stop___ftr_fixup);
+
 	/*
 	 * Unflatten the device-tree passed by prom_init or kexec
 	 */
Index: linux-2.6/arch/powerpc/platforms/iseries/setup.c
===================================================================
--- linux-2.6.orig/arch/powerpc/platforms/iseries/setup.c	2007-01-24 16:01:50.000000000 +1100
+++ linux-2.6/arch/powerpc/platforms/iseries/setup.c	2007-01-24 16:02:12.000000000 +1100
@@ -654,6 +654,11 @@
 	if (!of_flat_dt_is_compatible(root, "IBM,iSeries"))
 		return 0;
 
+	/* Identify CPU type. This is done again by the common code later
+	 * on but calling this function multiple times is fine.
+	 */
+	identify_cpu(0);
+
 	powerpc_firmware_features |= FW_FEATURE_ISERIES;
 	powerpc_firmware_features |= FW_FEATURE_LPAR;
 
Index: linux-2.6/arch/ppc/kernel/misc.S
===================================================================
--- linux-2.6.orig/arch/ppc/kernel/misc.S	2007-01-24 16:01:50.000000000 +1100
+++ linux-2.6/arch/ppc/kernel/misc.S	2007-01-24 16:02:12.000000000 +1100
@@ -110,80 +110,6 @@
 	blr
 
 /*
- * identify_cpu,
- * called with r3 = data offset and r4 = CPU number
- * doesn't change r3
- */
-_GLOBAL(identify_cpu)
-	addis	r8,r3,cpu_specs@ha
-	addi	r8,r8,cpu_specs@l
-	mfpvr	r7
-1:
-	lwz	r5,CPU_SPEC_PVR_MASK(r8)
-	and	r5,r5,r7
-	lwz	r6,CPU_SPEC_PVR_VALUE(r8)
-	cmplw	0,r6,r5
-	beq	1f
-	addi	r8,r8,CPU_SPEC_ENTRY_SIZE
-	b	1b
-1:
-	addis	r6,r3,cur_cpu_spec@ha
-	addi	r6,r6,cur_cpu_spec@l
-	sub	r8,r8,r3
-	stw	r8,0(r6)
-	blr
-
-/*
- * do_cpu_ftr_fixups - goes through the list of CPU feature fixups
- * and writes nop's over sections of code that don't apply for this cpu.
- * r3 = data offset (not changed)
- */
-_GLOBAL(do_cpu_ftr_fixups)
-	/* Get CPU 0 features */
-	addis	r6,r3,cur_cpu_spec@ha
-	addi	r6,r6,cur_cpu_spec@l
-	lwz	r4,0(r6)
-	add	r4,r4,r3
-	lwz	r4,CPU_SPEC_FEATURES(r4)
-
-	/* Get the fixup table */
-	addis	r6,r3,__start___ftr_fixup@ha
-	addi	r6,r6,__start___ftr_fixup@l
-	addis	r7,r3,__stop___ftr_fixup@ha
-	addi	r7,r7,__stop___ftr_fixup@l
-
-	/* Do the fixup */
-1:	cmplw	0,r6,r7
-	bgelr
-	addi	r6,r6,16
-	lwz	r8,-16(r6)	/* mask */
-	and	r8,r8,r4
-	lwz	r9,-12(r6)	/* value */
-	cmplw	0,r8,r9
-	beq	1b
-	lwz	r8,-8(r6)	/* section begin */
-	lwz	r9,-4(r6)	/* section end */
-	subf.	r9,r8,r9
-	beq	1b
-	/* write nops over the section of code */
-	/* todo: if large section, add a branch at the start of it */
-	srwi	r9,r9,2
-	mtctr	r9
-	add	r8,r8,r3
-	lis	r0,0x60000000@h	/* nop */
-3:	stw	r0,0(r8)
-	andi.	r10,r4,CPU_FTR_SPLIT_ID_CACHE@l
-	beq	2f
-	dcbst	0,r8		/* suboptimal, but simpler */
-	sync
-	icbi	0,r8
-2:	addi	r8,r8,4
-	bdnz	3b
-	sync			/* additional sync needed on g4 */
-	isync
-	b	1b
-
-/*
  * call_setup_cpu - call the setup_cpu function for this cpu
  * r3 = data offset, r24 = cpu number
  *
Index: linux-2.6/arch/ppc/kernel/setup.c
===================================================================
--- linux-2.6.orig/arch/ppc/kernel/setup.c	2007-01-24 16:01:50.000000000 +1100
+++ linux-2.6/arch/ppc/kernel/setup.c	2007-01-24 16:02:12.000000000 +1100
@@ -38,6 +38,7 @@
 #include <asm/nvram.h>
 #include <asm/xmon.h>
 #include <asm/ocp.h>
+#include <asm/prom.h>
 
 #define USES_PPC_SYS (defined(CONFIG_85xx) || defined(CONFIG_83xx) || \
 		      defined(CONFIG_MPC10X_BRIDGE) || defined(CONFIG_8260) || \
@@ -53,8 +54,6 @@
 
 extern void platform_init(unsigned long r3, unsigned long r4,
 		unsigned long r5, unsigned long r6, unsigned long r7);
-extern void identify_cpu(unsigned long offset, unsigned long cpu);
-extern void do_cpu_ftr_fixups(unsigned long offset);
 extern void reloc_got2(unsigned long offset);
 
 extern void ppc6xx_idle(void);
@@ -298,6 +297,7 @@
 {
  	unsigned long phys;
 	unsigned long offset = reloc_offset();
+	struct cpu_spec *spec;
 
  	/* Default */
  	phys = offset + KERNELBASE;
@@ -310,8 +310,10 @@
 	 * Identify the CPU type and fix up code sections
 	 * that depend on which cpu we have.
 	 */
-	identify_cpu(offset, 0);
-	do_cpu_ftr_fixups(offset);
+	spec = identify_cpu(offset);
+	do_feature_fixups(offset, spec->cpu_features,
+			  PTRRELOC(&__start___ftr_fixup),
+			  PTRRELOC(&__stop___ftr_fixup));
 
 	return phys;
 }
Index: linux-2.6/include/asm-powerpc/cputable.h
===================================================================
--- linux-2.6.orig/include/asm-powerpc/cputable.h	2007-01-24 16:01:50.000000000 +1100
+++ linux-2.6/include/asm-powerpc/cputable.h	2007-01-24 16:02:12.000000000 +1100
@@ -85,8 +85,11 @@
 
 extern struct cpu_spec		*cur_cpu_spec;
 
-extern void identify_cpu(unsigned long offset, unsigned long cpu);
-extern void do_cpu_ftr_fixups(unsigned long offset);
+extern unsigned int __start___ftr_fixup, __stop___ftr_fixup;
+
+extern struct cpu_spec *identify_cpu(unsigned long offset);
+extern void do_feature_fixups(unsigned long offset, unsigned long value,
+			      void *fixup_start, void *fixup_end);
 
 #endif /* __ASSEMBLY__ */
 
Index: linux-2.6/arch/powerpc/kernel/misc_64.S
===================================================================
--- linux-2.6.orig/arch/powerpc/kernel/misc_64.S	2007-01-24 16:01:50.000000000 +1100
+++ linux-2.6/arch/powerpc/kernel/misc_64.S	2007-01-24 16:02:12.000000000 +1100
@@ -246,85 +246,6 @@
 	isync
 	blr
 
-/*
- * identify_cpu and calls setup_cpu
- * In:	r3 = base of the cpu_specs array
- *	r4 = address of cur_cpu_spec
- *	r5 = relocation offset
- */
-_GLOBAL(identify_cpu)
-	mfpvr	r7
-1:
-	lwz	r8,CPU_SPEC_PVR_MASK(r3)
-	and	r8,r8,r7
-	lwz	r9,CPU_SPEC_PVR_VALUE(r3)
-	cmplw	0,r9,r8
-	beq	1f
-	addi	r3,r3,CPU_SPEC_ENTRY_SIZE
-	b	1b
-1:
-	sub	r0,r3,r5
-	std	r0,0(r4)
-	ld	r4,CPU_SPEC_SETUP(r3)
-	cmpdi	0,r4,0
-	add	r4,r4,r5
-	beqlr
-	ld	r4,0(r4)
-	add	r4,r4,r5
-	mtctr	r4
-	/* Calling convention for cpu setup is r3=offset, r4=cur_cpu_spec */
-	mr	r4,r3
-	mr	r3,r5
-	bctr
-
-/*
- * do_cpu_ftr_fixups - goes through the list of CPU feature fixups
- * and writes nop's over sections of code that don't apply for this cpu.
- * r3 = data offset (not changed)
- */
-_GLOBAL(do_cpu_ftr_fixups)
-	/* Get CPU 0 features */
-	LOAD_REG_IMMEDIATE(r6,cur_cpu_spec)
-	sub	r6,r6,r3
-	ld	r4,0(r6)
-	sub	r4,r4,r3
-	ld	r4,CPU_SPEC_FEATURES(r4)
-	/* Get the fixup table */
-	LOAD_REG_IMMEDIATE(r6,__start___ftr_fixup)
-	sub	r6,r6,r3
-	LOAD_REG_IMMEDIATE(r7,__stop___ftr_fixup)
-	sub	r7,r7,r3
-	/* Do the fixup */
-1:	cmpld	r6,r7
-	bgelr
-	addi	r6,r6,32
-	ld	r8,-32(r6)	/* mask */
-	and	r8,r8,r4
-	ld	r9,-24(r6)	/* value */
-	cmpld	r8,r9
-	beq	1b
-	ld	r8,-16(r6)	/* section begin */
-	ld	r9,-8(r6)	/* section end */
-	subf.	r9,r8,r9
-	beq	1b
-	/* write nops over the section of code */
-	/* todo: if large section, add a branch at the start of it */
-	srwi	r9,r9,2
-	mtctr	r9
-	sub	r8,r8,r3
-	lis	r0,0x60000000@h	/* nop */
-3:	stw	r0,0(r8)
-	andi.	r10,r4,CPU_FTR_SPLIT_ID_CACHE@l
-	beq	2f
-	dcbst	0,r8		/* suboptimal, but simpler */
-	sync
-	icbi	0,r8
-2:	addi	r8,r8,4
-	bdnz	3b
-	sync			/* additional sync needed on g4 */
-	isync
-	b	1b
-
 #if defined(CONFIG_PPC_PMAC) || defined(CONFIG_PPC_MAPLE)
 /*
  * Do an IO access in real mode

RHBZ#:
------
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=228099

Description:
------------
1 of 12 patches developed by Benjamin Herrenschmidt of IBM required for 
supportin IBM customers purchasing Cell blade configurations.

This patch adds some macros that can be used with an explicit label in 
order to nest cpu features.  This should be used very carefully but is 
necessary for the upcoming cell TB fixup.

RHEL Version Found:
-------------------
feature enhancement for 5.1.

Upstream Status:
----------------
This has been accepted into the kernel and can be viewed in the 2.6.20.3 
kernel.    There are slight changes, when the 5.1 code differs from the 
kernel.  

Test Status:
------------
This patch was tested by Benjamin herrenschmidt of IBM.  It has been
ack'd by Olof Johansson and signed off by Paul Mackerras.

Proposed Patch:
----------------
Please review and ACK for RHEL 5.1.
-

diff --git a/include/asm-powerpc/cputable.h b/include/asm-powerpc/cputable.h
index 4d22218..65faf32 100644
--- a/include/asm-powerpc/cputable.h
+++ b/include/asm-powerpc/cputable.h
@@ -434,30 +434,34 @@ static inline int cpu_has_feature(unsigned long feature)
 
 #ifdef __ASSEMBLY__
 
-#define BEGIN_FTR_SECTION		98:
+#define BEGIN_FTR_SECTION_NESTED(label)	label:
+#define BEGIN_FTR_SECTION		BEGIN_FTR_SECTION_NESTED(98)
 
 #ifndef __powerpc64__
-#define END_FTR_SECTION(msk, val)		\
+#define END_FTR_SECTION_NESTED(msk, val, label) \
 99:						\
 	.section __ftr_fixup,"a";		\
 	.align 2;				\
 	.long msk;				\
 	.long val;				\
-	.long 98b;				\
+	.long label##b;				\
 	.long 99b;				\
 	.previous
 #else /* __powerpc64__ */
-#define END_FTR_SECTION(msk, val)		\
+#define END_FTR_SECTION_NESTED(msk, val, label) \
 99:						\
 	.section __ftr_fixup,"a";		\
 	.align 3;				\
 	.llong msk;				\
 	.llong val;				\
-	.llong 98b;				\
+	.llong label##b;				\
 	.llong 99b;	 			\
 	.previous
 #endif /* __powerpc64__ */
 
+#define END_FTR_SECTION(msk, val)		\
+	END_FTR_SECTION_NESTED(msk, val, 98)
+
 #define END_FTR_SECTION_IFSET(msk)	END_FTR_SECTION((msk), (msk))
 #define END_FTR_SECTION_IFCLR(msk)	END_FTR_SECTION((msk), 0)
 #endif /* __ASSEMBLY__ */

RHBZ#:
------
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=228099

Description:
------------
1 of 12 patches developed by Benjamin Herrenschmidt of IBM required for 
supportin IBM customers purchasing Cell blade configurations.

This patch reworks the feature fixup mecanism so vdso's can be fixed up.
The main issue was that the construct:

            .long   label  (or .llong on 64 bits)

will not work in the case of a shared library like the vdso. It will 
generate an empty placeholder in the fixup table along with a reloc, which 
is not something we can deal with in the vdso.

The idea here (thanks Alan Modra !) is to instead use something like:

    1:
            .long   label - 1b

That is, the feature fixup tables no longer contain addresses of bits 
of code to patch, but offsets of such code from the fixup table entry 
itself. That is properly resolved by ld when building the .so's. I've
modified the fixup mecanism generically to use that method for the rest
of the kernel as well.

Another trick is that the 32 bits vDSO included in the 64 bits kernel  
need to have a table in the 64 bits format. However, gas does not support 
32 bits code with a statement of the form:

            .llong  label - 1b  (Or even just .llong label)

That is, it cannot emit the right fixup/relocation for the linker to use 
to assign a 32 bits address to an .llong field. Thus, in the specific case 
of the 32 bits vdso built as part of the 64 bits kernel, we are using a 
modified macro that generates:

            .long   0xffffffff
            .llong  label - 1b

Note that this assumes that the value is negative which is enforced by the 
.lds (those offsets are always negative as the .text is always before the 
fixup table and gas doesn't support emiting the reloc the other way around).

RHEL Version Found:
-------------------
feature enhancement for 5.1.

Upstream Status:
----------------
This has been accepted into the kernel and can be viewed in the 2.6.20.3 
kernel.    

Test Status:
------------
This patch was tested by Benjamin herrenschmidt of IBM.  It has been
signed off by Olof Johansson and Paul Mackerras.

Proposed Patch:
----------------
Please review and ACK for RHEL 5.1.
-
Index: linux-2.6/arch/powerpc/kernel/cputable.c
===================================================================
--- linux-2.6.orig/arch/powerpc/kernel/cputable.c	2007-01-24 16:02:12.000000000 +1100
+++ linux-2.6/arch/powerpc/kernel/cputable.c	2007-01-24 16:04:30.000000000 +1100
@@ -1155,14 +1155,13 @@
 	return NULL;
 }
 
-void do_feature_fixups(unsigned long offset, unsigned long value,
-		       void *fixup_start, void *fixup_end)
+void do_feature_fixups(unsigned long value, void *fixup_start, void *fixup_end)
 {
 	struct fixup_entry {
 		unsigned long	mask;
 		unsigned long	value;
-		unsigned int	*start;
-		unsigned int	*end;
+		long		start_off;
+		long		end_off;
 	} *fcur, *fend;
 
 	fcur = fixup_start;
@@ -1177,8 +1176,8 @@
 		/* These PTRRELOCs will disappear once the new scheme for
 		 * modules and vdso is implemented
 		 */
-		pstart = PTRRELOC(fcur->start);
-		pend = PTRRELOC(fcur->end);
+		pstart = ((unsigned int *)fcur) + (fcur->start_off / 4);
+		pend = ((unsigned int *)fcur) + (fcur->end_off / 4);
 
 		for (p = pstart; p < pend; p++) {
 			*p = 0x60000000u;
Index: linux-2.6/arch/powerpc/kernel/setup_32.c
===================================================================
--- linux-2.6.orig/arch/powerpc/kernel/setup_32.c	2007-01-24 16:02:12.000000000 +1100
+++ linux-2.6/arch/powerpc/kernel/setup_32.c	2007-01-24 16:04:30.000000000 +1100
@@ -107,7 +107,7 @@
 	 */
 	spec = identify_cpu(offset);
 
-	do_feature_fixups(offset, spec->cpu_features,
+	do_feature_fixups(spec->cpu_features,
 			  PTRRELOC(&__start___ftr_fixup),
 			  PTRRELOC(&__stop___ftr_fixup));
 
Index: linux-2.6/arch/powerpc/kernel/setup_64.c
===================================================================
--- linux-2.6.orig/arch/powerpc/kernel/setup_64.c	2007-01-24 16:03:37.000000000 +1100
+++ linux-2.6/arch/powerpc/kernel/setup_64.c	2007-01-24 16:04:56.000000000 +1100
@@ -360,7 +360,7 @@
 	/* Apply the CPUs-specific and firmware specific fixups to kernel
 	 * text (nop out sections not relevant to this CPU or this firmware)
 	 */
-	do_feature_fixups(0, cur_cpu_spec->cpu_features,
+	do_feature_fixups(cur_cpu_spec->cpu_features,
 			  &__start___ftr_fixup, &__stop___ftr_fixup);
 
 	/*
Index: linux-2.6/arch/powerpc/kernel/vdso.c
===================================================================
--- linux-2.6.orig/arch/powerpc/kernel/vdso.c	2007-01-24 16:01:50.000000000 +1100
+++ linux-2.6/arch/powerpc/kernel/vdso.c	2007-01-24 16:10:35.000000000 +1100
@@ -36,6 +36,8 @@
 #include <asm/vdso.h>
 #include <asm/vdso_datapage.h>
 
+#include "setup.h"
+
 #undef DEBUG
 
 #ifdef DEBUG
@@ -587,6 +589,31 @@
 	return 0;
 }
 
+
+static __init int vdso_fixup_features(struct lib32_elfinfo *v32,
+				      struct lib64_elfinfo *v64)
+{
+	void *start32;
+	unsigned long size32;
+
+#ifdef CONFIG_PPC64
+	void *start64;
+	unsigned long size64;
+
+	start64 = find_section64(v64->hdr, "__ftr_fixup", &size64);
+	if (start64)
+		do_feature_fixups(cur_cpu_spec->cpu_features,
+				  start64, start64 + size64);
+#endif /* CONFIG_PPC64 */
+
+	start32 = find_section32(v32->hdr, "__ftr_fixup", &size32);
+	if (start32)
+		do_feature_fixups(cur_cpu_spec->cpu_features,
+				  start32, start32 + size32);
+
+	return 0;
+}
+
 static __init int vdso_fixup_alt_funcs(struct lib32_elfinfo *v32,
 				       struct lib64_elfinfo *v64)
 {
@@ -635,6 +662,9 @@
 	if (vdso_fixup_datapage(&v32, &v64))
 		return -1;
 
+	if (vdso_fixup_features(&v32, &v64))
+		return -1;
+
 	if (vdso_fixup_alt_funcs(&v32, &v64))
 		return -1;
 
@@ -715,6 +745,7 @@
 	 * Setup the syscall map in the vDOS
 	 */
 	vdso_setup_syscall_map();
+
 	/*
 	 * Initialize the vDSO images in memory, that is do necessary
 	 * fixups of vDSO symbols, locate trampolines, etc...
Index: linux-2.6/arch/powerpc/kernel/vdso32/vdso32.lds.S
===================================================================
--- linux-2.6.orig/arch/powerpc/kernel/vdso32/vdso32.lds.S	2007-01-24 16:01:50.000000000 +1100
+++ linux-2.6/arch/powerpc/kernel/vdso32/vdso32.lds.S	2007-01-24 16:06:01.000000000 +1100
@@ -32,6 +32,11 @@
   PROVIDE (_etext = .);
   PROVIDE (etext = .);
 
+  . = ALIGN(8);
+  __ftr_fixup : {
+    *(__ftr_fixup)
+  }
+
   /* Other stuff is appended to the text segment: */
   .rodata		: { *(.rodata .rodata.* .gnu.linkonce.r.*) }
   .rodata1		: { *(.rodata1) }
Index: linux-2.6/arch/powerpc/kernel/vdso64/vdso64.lds.S
===================================================================
--- linux-2.6.orig/arch/powerpc/kernel/vdso64/vdso64.lds.S	2007-01-24 16:01:50.000000000 +1100
+++ linux-2.6/arch/powerpc/kernel/vdso64/vdso64.lds.S	2007-01-24 16:05:35.000000000 +1100
@@ -31,6 +31,11 @@
   PROVIDE (_etext = .);
   PROVIDE (etext = .);
 
+  . = ALIGN(8);
+  __ftr_fixup : {
+    *(__ftr_fixup)
+  }
+
   /* Other stuff is appended to the text segment: */
   .rodata         : { *(.rodata .rodata.* .gnu.linkonce.r.*) }
   .rodata1        : { *(.rodata1) }
Index: linux-2.6/arch/ppc/kernel/setup.c
===================================================================
--- linux-2.6.orig/arch/ppc/kernel/setup.c	2007-01-24 16:02:12.000000000 +1100
+++ linux-2.6/arch/ppc/kernel/setup.c	2007-01-24 16:04:30.000000000 +1100
@@ -311,7 +311,7 @@
 	 * that depend on which cpu we have.
 	 */
 	spec = identify_cpu(offset);
-	do_feature_fixups(offset, spec->cpu_features,
+	do_feature_fixups(spec->cpu_features,
 			  PTRRELOC(&__start___ftr_fixup),
 			  PTRRELOC(&__stop___ftr_fixup));
 
Index: linux-2.6/include/asm-powerpc/asm-compat.h
===================================================================
--- linux-2.6.orig/include/asm-powerpc/asm-compat.h	2007-01-24 16:01:50.000000000 +1100
+++ linux-2.6/include/asm-powerpc/asm-compat.h	2007-01-24 16:04:30.000000000 +1100
@@ -14,6 +14,58 @@
 #  define ASM_CONST(x)		__ASM_CONST(x)
 #endif
 
+
+/*
+ * Feature section common macros
+ *
+ * Note that the entries now contain offsets between the table entry
+ * and the code rather than absolute code pointers in order to be
+ * useable with the vdso shared library. There is also an assumption
+ * that values will be negative, that is, the fixup table has to be
+ * located after the code it fixes up.
+ */
+#ifdef CONFIG_PPC64
+#ifdef __powerpc64__
+/* 64 bits kernel, 64 bits code */
+#define MAKE_FTR_SECTION_ENTRY(msk, val, label, sect)	\
+99:							\
+	.section sect,"a";				\
+	.align 3;					\
+98:						       	\
+	.llong msk;					\
+	.llong val;					\
+	.llong label##b-98b;				\
+	.llong 99b-98b;		 			\
+	.previous
+#else /* __powerpc64__ */
+/* 64 bits kernel, 32 bits code (ie. vdso32) */
+#define MAKE_FTR_SECTION_ENTRY(msk, val, label, sect)	\
+99:							\
+	.section sect,"a";				\
+	.align 3;					\
+98:						       	\
+	.llong msk;					\
+	.llong val;					\
+	.long 0xffffffff;      				\
+	.long label##b-98b;				\
+	.long 0xffffffff;	       			\
+	.long 99b-98b;		 			\
+	.previous
+#endif /* !__powerpc64__ */
+#else /* CONFIG_PPC64 */
+/* 32 bits kernel, 32 bits code */
+#define MAKE_FTR_SECTION_ENTRY(msk, val, label, sect)	\
+99:						       	\
+	.section sect,"a";			       	\
+	.align 2;				       	\
+98:						       	\
+	.long msk;				       	\
+	.long val;				       	\
+	.long label##b-98b;			       	\
+	.long 99b-98b;				       	\
+	.previous
+#endif /* !CONFIG_PPC64 */
+
 #ifdef __powerpc64__
 
 /* operations for longs and pointers */
Index: linux-2.6/include/asm-powerpc/cputable.h
===================================================================
--- linux-2.6.orig/include/asm-powerpc/cputable.h	2007-01-24 16:04:13.000000000 +1100
+++ linux-2.6/include/asm-powerpc/cputable.h	2007-01-24 16:27:42.000000000 +1100
@@ -88,8 +88,8 @@
 extern unsigned int __start___ftr_fixup, __stop___ftr_fixup;
 
 extern struct cpu_spec *identify_cpu(unsigned long offset);
-extern void do_feature_fixups(unsigned long offset, unsigned long value,
-			      void *fixup_start, void *fixup_end);
+extern void do_feature_fixups(unsigned long value, void *fixup_start,
+			      void *fixup_end);
 
 #endif /* __ASSEMBLY__ */
 
@@ -426,32 +426,11 @@
 #ifdef __ASSEMBLY__
 
 #define BEGIN_FTR_SECTION_NESTED(label)	label:
-#define BEGIN_FTR_SECTION		BEGIN_FTR_SECTION_NESTED(98)
-
-#ifndef __powerpc64__
-#define END_FTR_SECTION_NESTED(msk, val, label) \
-99:						\
-	.section __ftr_fixup,"a";		\
-	.align 2;				\
-	.long msk;				\
-	.long val;				\
-	.long label##b;				\
-	.long 99b;				\
-	.previous
-#else /* __powerpc64__ */
+#define BEGIN_FTR_SECTION		BEGIN_FTR_SECTION_NESTED(97)
 #define END_FTR_SECTION_NESTED(msk, val, label) \
-99:						\
-	.section __ftr_fixup,"a";		\
-	.align 3;				\
-	.llong msk;				\
-	.llong val;				\
-	.llong label##b;				\
-	.llong 99b;	 			\
-	.previous
-#endif /* __powerpc64__ */
-
+	MAKE_FTR_SECTION_ENTRY(msk, val, label, __ftr_fixup)
 #define END_FTR_SECTION(msk, val)		\
-	END_FTR_SECTION_NESTED(msk, val, 98)
+	END_FTR_SECTION_NESTED(msk, val, 97)
 
 #define END_FTR_SECTION_IFSET(msk)	END_FTR_SECTION((msk), (msk))
 #define END_FTR_SECTION_IFCLR(msk)	END_FTR_SECTION((msk), 0)
Index: linux-2.6/include/asm-powerpc/timex.h
===================================================================
--- linux-2.6.orig/include/asm-powerpc/timex.h	2007-01-24 16:01:50.000000000 +1100
+++ linux-2.6/include/asm-powerpc/timex.h	2007-01-24 16:27:42.000000000 +1100
@@ -30,13 +30,15 @@
 	ret = 0;
 
 	__asm__ __volatile__(
-		"98:	mftb %0\n"
+		"97:	mftb %0\n"
 		"99:\n"
 		".section __ftr_fixup,\"a\"\n"
+		".align 2\n"
+		"98:\n"
 		"	.long %1\n"
 		"	.long 0\n"
-		"	.long 98b\n"
-		"	.long 99b\n"
+		"	.long 97b-98b\n"
+		"	.long 99b-98b\n"
 		".previous"
 		: "=r" (ret) : "i" (CPU_FTR_601));
 #endif

RHBZ#:
------
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=228099

Description:
------------
1 of 12 patches developed by Benjamin herrenschmidt of IBM  required for 
IBM customers purchasing Cell blade configurations.

This patch adds support for feature fixups in modules.  This involves 
adding support for R_PPC64_REL64 relocs to the 64 bits module loader.  It 
also modifies modpost.c to ignore the powerpc fixup sections (or it would 
warn when used in .init.text).

RHEL Version Found:
-------------------
feature enhancement for 5.1.

Upstream Status:
----------------
This has been accepted into the kernel and can be viewed in the 2.6.20.3 
kernel.    

Test Status:
------------
This patch was tested by Benjamin herrenschmidt of IBM.  It has been
signed off by Olof Johansson and Paul Mackerras.

Proposed Patch:
----------------
Please review and ACK for RHEL 5.1.
-

Index: linux-2.6/arch/powerpc/kernel/module_32.c
===================================================================
--- linux-2.6.orig/arch/powerpc/kernel/module_32.c	2007-01-24 16:00:26.000000000 +1100
+++ linux-2.6/arch/powerpc/kernel/module_32.c	2007-01-24 16:13:46.000000000 +1100
@@ -24,6 +24,8 @@
 #include <linux/kernel.h>
 #include <linux/cache.h>
 
+#include "setup.h"
+
 #if 0
 #define DEBUGP printk
 #else
@@ -269,33 +271,50 @@
 	return 0;
 }
 
+static const Elf_Shdr *find_section(const Elf_Ehdr *hdr,
+				    const Elf_Shdr *sechdrs,
+				    const char *name)
+{
+	char *secstrings;
+	unsigned int i;
+
+	secstrings = (char *)hdr + sechdrs[hdr->e_shstrndx].sh_offset;
+	for (i = 1; i < hdr->e_shnum; i++)
+		if (strcmp(secstrings+sechdrs[i].sh_name, name) == 0)
+			return &sechdrs[i];
+	return NULL;
+}
+
 int module_finalize(const Elf_Ehdr *hdr,
 		    const Elf_Shdr *sechdrs,
 		    struct module *me)
 {
-	char *secstrings;
-	unsigned int i;
+	const Elf_Shdr *sect;
 
 	me->arch.bug_table = NULL;
 	me->arch.num_bugs = 0;
 
 	/* Find the __bug_table section, if present */
-	secstrings = (char *)hdr + sechdrs[hdr->e_shstrndx].sh_offset;
-	for (i = 1; i < hdr->e_shnum; i++) {
-		if (strcmp(secstrings+sechdrs[i].sh_name, "__bug_table"))
-			continue;
-		me->arch.bug_table = (void *) sechdrs[i].sh_addr;
-		me->arch.num_bugs = sechdrs[i].sh_size / sizeof(struct bug_entry);
-		break;
+	sect = find_section(hdr, sechdrs, "__bug_table");
+	if (sect != NULL) {
+		me->arch.bug_table = (void *) sect->sh_addr;
+		me->arch.num_bugs = sect->sh_size / sizeof(struct bug_entry);
 	}
 
-	/*
+ 	/*
 	 * Strictly speaking this should have a spinlock to protect against
 	 * traversals, but since we only traverse on BUG()s, a spinlock
 	 * could potentially lead to deadlock and thus be counter-productive.
 	 */
 	list_add(&me->arch.bug_list, &module_bug_list);
 
+	/* Apply feature fixups */
+	sect = find_section(hdr, sechdrs, "__ftr_fixup");
+	if (sect != NULL)
+		do_feature_fixups(cur_cpu_spec->cpu_features,
+				  (void *)sect->sh_addr,
+				  (void *)sect->sh_addr + sect->sh_size);
+
 	return 0;
 }
 
Index: linux-2.6/arch/powerpc/kernel/module_64.c
===================================================================
--- linux-2.6.orig/arch/powerpc/kernel/module_64.c	2007-01-24 16:00:26.000000000 +1100
+++ linux-2.6/arch/powerpc/kernel/module_64.c	2007-01-24 16:14:30.000000000 +1100
@@ -22,6 +22,9 @@
 #include <linux/vmalloc.h>
 #include <asm/module.h>
 #include <asm/uaccess.h>
+#include <asm/firmware.h>
+
+#include "setup.h"
 
 /* FIXME: We don't do .init separately.  To do this, we'd need to have
    a separate r2 value in the init and core section, and stub between
@@ -400,6 +403,11 @@
 				| (value & 0x03fffffc);
 			break;
 
+		case R_PPC64_REL64:
+			/* 64 bits relative (used by features fixups) */
+			*location = value - (unsigned long)location;
+			break;
+
 		default:
 			printk("%s: Unknown ADD relocation: %lu\n",
 			       me->name,
@@ -413,23 +421,33 @@
 
 LIST_HEAD(module_bug_list);
 
-int module_finalize(const Elf_Ehdr *hdr,
-		const Elf_Shdr *sechdrs, struct module *me)
+static const Elf_Shdr *find_section(const Elf_Ehdr *hdr,
+				    const Elf_Shdr *sechdrs,
+				    const char *name)
 {
 	char *secstrings;
 	unsigned int i;
 
+	secstrings = (char *)hdr + sechdrs[hdr->e_shstrndx].sh_offset;
+	for (i = 1; i < hdr->e_shnum; i++)
+		if (strcmp(secstrings+sechdrs[i].sh_name, name) == 0)
+			return &sechdrs[i];
+	return NULL;
+}
+
+int module_finalize(const Elf_Ehdr *hdr,
+		const Elf_Shdr *sechdrs, struct module *me)
+{
+	const Elf_Shdr *sect;
+
 	me->arch.bug_table = NULL;
 	me->arch.num_bugs = 0;
 
 	/* Find the __bug_table section, if present */
-	secstrings = (char *)hdr + sechdrs[hdr->e_shstrndx].sh_offset;
-	for (i = 1; i < hdr->e_shnum; i++) {
-		if (strcmp(secstrings+sechdrs[i].sh_name, "__bug_table"))
-			continue;
-		me->arch.bug_table = (void *) sechdrs[i].sh_addr;
-		me->arch.num_bugs = sechdrs[i].sh_size / sizeof(struct bug_entry);
-		break;
+	sect = find_section(hdr, sechdrs, "__bug_table");
+	if (sect != NULL) {
+		me->arch.bug_table = (void *) sect->sh_addr;
+		me->arch.num_bugs = sect->sh_size / sizeof(struct bug_entry);
 	}
 
 	/*
@@ -439,6 +457,13 @@
 	 */
 	list_add(&me->arch.bug_list, &module_bug_list);
 
+	/* Apply feature fixups */
+	sect = find_section(hdr, sechdrs, "__ftr_fixup");
+	if (sect != NULL)
+		do_feature_fixups(cur_cpu_spec->cpu_features,
+				  (void *)sect->sh_addr,
+				  (void *)sect->sh_addr + sect->sh_size);
+
 	return 0;
 }
 
Index: linux-2.6/scripts/mod/modpost.c
===================================================================
--- linux-2.6.orig/scripts/mod/modpost.c	2007-01-24 16:00:26.000000000 +1100
+++ linux-2.6/scripts/mod/modpost.c	2007-01-24 16:14:04.000000000 +1100
@@ -910,6 +910,7 @@
 		".fixup",
 		".smp_locks",
 		".plt",  /* seen on ARCH=um build on x86_64. Harmless */
+		"__ftr_fixup",		/* powerpc cpu feature fixup */
 		NULL
 	};
 	/* Start of section names */

RHBZ#:
------
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=228099

Description:
------------
1 of 12 patches developed by Ben Herrenschmidt required for supporting IBM 
customers purchasing Cell blade configurations.

The Cell CPU timebase has an erratum.  When reading the entire 64 bits of 
the timebase with one mftb instruction, there is a handful of cycles 
window during which one might read a value with the low order 32 bits 
already reset to 0x00000000 but the high order bits not yet incremented by 
one.  This fixes it by reading the timebase again until the low order 32 
bits are no longer 0.  That might introduce occasional latencies if 
hitting mftb just at the wrong time, but no more than 70ns on a cell 
blade, and that was considered acceptable.

RHEL Version Found:
-------------------
feature enhancement for 5.1.

Upstream Status:
----------------
This has been accepted into the kernel and can be viewed in the 2.6.20.3 
kernel.    

Test Status:
------------
This patch was tested by Benjamin herrenschmidt of IBM.  It has been
signed off by Olof Johansson and Paul Mackerras.

Proposed Patch:
----------------
Please review and ACK for RHEL 5.1.
-

Index: linux-2.6/arch/powerpc/kernel/vdso64/gettimeofday.S
===================================================================
--- linux-2.6.orig/arch/powerpc/kernel/vdso64/gettimeofday.S	2007-01-22 14:38:56.000000000 +1100
+++ linux-2.6/arch/powerpc/kernel/vdso64/gettimeofday.S	2007-01-24 15:21:02.000000000 +1100
@@ -229,8 +229,10 @@
 	xor	r0,r8,r8		/* create dependency */
 	add	r3,r3,r0
 
-	/* Get TB & offset it */
-	mftb	r7
+	/* Get TB & offset it. We use the MFTB macro which will generate
+	 * workaround code for Cell.
+	 */
+	MFTB(r7)
 	ld	r9,CFG_TB_ORIG_STAMP(r3)
 	subf	r7,r9,r7
 
Index: linux-2.6/include/asm-powerpc/cputable.h
===================================================================
--- linux-2.6.orig/include/asm-powerpc/cputable.h	2007-01-22 14:38:59.000000000 +1100
+++ linux-2.6/include/asm-powerpc/cputable.h	2007-01-24 15:22:28.000000000 +1100
@@ -140,6 +140,7 @@
 #define CPU_FTR_CI_LARGE_PAGE		LONG_ASM_CONST(0x0000100000000000)
 #define CPU_FTR_PAUSE_ZERO		LONG_ASM_CONST(0x0000200000000000)
 #define CPU_FTR_PURR			LONG_ASM_CONST(0x0000400000000000)
+#define CPU_FTR_CELL_TB_BUG		LONG_ASM_CONST(0x0000800000000000)
 
 #ifndef __ASSEMBLY__
 
@@ -327,7 +328,8 @@
 #define CPU_FTRS_CELL	(CPU_FTR_SPLIT_ID_CACHE | CPU_FTR_USE_TB | \
 	    CPU_FTR_HPTE_TABLE | CPU_FTR_PPCAS_ARCH_V2 | \
 	    CPU_FTR_ALTIVEC_COMP | CPU_FTR_MMCRA | CPU_FTR_SMT | \
-	    CPU_FTR_CTRL | CPU_FTR_PAUSE_ZERO | CPU_FTR_CI_LARGE_PAGE)
+	    CPU_FTR_CTRL | CPU_FTR_PAUSE_ZERO | CPU_FTR_CI_LARGE_PAGE | \
+	    CPU_FTR_CELL_TB_BUG )
 #define CPU_FTRS_COMPATIBLE	(CPU_FTR_SPLIT_ID_CACHE | CPU_FTR_USE_TB | \
 	    CPU_FTR_HPTE_TABLE | CPU_FTR_PPCAS_ARCH_V2)
 #endif
Index: linux-2.6/include/asm-powerpc/ppc_asm.h
===================================================================
--- linux-2.6.orig/include/asm-powerpc/ppc_asm.h	2007-01-22 14:38:59.000000000 +1100
+++ linux-2.6/include/asm-powerpc/ppc_asm.h	2007-01-24 15:21:02.000000000 +1100
@@ -30,9 +30,9 @@
 	mfspr	ra,SPRN_PURR;		/* get processor util. reg */	\
 END_FTR_SECTION_IFSET(CPU_FTR_PURR);					\
 BEGIN_FTR_SECTION;							\
-	mftb	ra;			/* or get TB if no PURR */	\
+	MFTB(ra);			/* or get TB if no PURR */	\
 END_FTR_SECTION_IFCLR(CPU_FTR_PURR);					\
-	ld	rb,PACA_STARTPURR(r13);				\
+	ld	rb,PACA_STARTPURR(r13);					\
 	std	ra,PACA_STARTPURR(r13);					\
 	subf	rb,rb,ra;		/* subtract start value */	\
 	ld	ra,PACA_USER_TIME(r13);					\
@@ -45,9 +45,9 @@
 	mfspr	ra,SPRN_PURR;		/* get processor util. reg */	\
 END_FTR_SECTION_IFSET(CPU_FTR_PURR);					\
 BEGIN_FTR_SECTION;							\
-	mftb	ra;			/* or get TB if no PURR */	\
+	MFTB(ra);			/* or get TB if no PURR */	\
 END_FTR_SECTION_IFCLR(CPU_FTR_PURR);					\
-	ld	rb,PACA_STARTPURR(r13);				\
+	ld	rb,PACA_STARTPURR(r13);					\
 	std	ra,PACA_STARTPURR(r13);					\
 	subf	rb,rb,ra;		/* subtract start value */	\
 	ld	ra,PACA_SYSTEM_TIME(r13);				\
@@ -274,6 +274,16 @@
 #define ISYNC_601
 #endif
 
+#ifdef CONFIG_PPC_CELL
+#define MFTB(dest)			\
+90:	mftb  dest;			\
+BEGIN_FTR_SECTION_NESTED(96);		\
+	cmpwi dest,0;			\
+	beq-  90b;			\
+END_FTR_SECTION_NESTED(CPU_FTR_CELL_TB_BUG, CPU_FTR_CELL_TB_BUG, 96)
+#else
+#define MFTB(dest)			mftb dest
+#endif
 
 #ifndef CONFIG_SMP
 #define TLBSYNC
Index: linux-2.6/include/asm-powerpc/reg.h
===================================================================
--- linux-2.6.orig/include/asm-powerpc/reg.h	2007-01-22 14:38:59.000000000 +1100
+++ linux-2.6/include/asm-powerpc/reg.h	2007-01-24 15:21:02.000000000 +1100
@@ -617,10 +617,35 @@
 				: "=r" (rval)); rval;})
 #define mtspr(rn, v)	asm volatile("mtspr " __stringify(rn) ",%0" : : "r" (v))
 
+#ifdef __powerpc64__
+#ifdef CONFIG_PPC_CELL
+#define mftb()		({unsigned long rval;				\
+			asm volatile(					\
+				"90:	mftb %0;\n"			\
+				"97:	cmpwi %0,0;\n"			\
+				"	beq- 90b;\n"			\
+				"99:\n"					\
+				".section __ftr_fixup,\"a\"\n"		\
+				".align 3\n"				\
+				"98:\n"					\
+				"	.llong %1\n"			\
+				"	.llong %1\n"			\
+				"	.llong 97b-98b\n"		\
+				"	.llong 99b-98b\n"		\
+				".previous"				\
+			: "=r" (rval) : "i" (CPU_FTR_CELL_TB_BUG)); rval;})
+#else
 #define mftb()		({unsigned long rval;	\
 			asm volatile("mftb %0" : "=r" (rval)); rval;})
+#endif /* !CONFIG_PPC_CELL */
+
+#else /* __powerpc64__ */
+
 #define mftbl()		({unsigned long rval;	\
 			asm volatile("mftbl %0" : "=r" (rval)); rval;})
+#define mftbu()		({unsigned long rval;	\
+			asm volatile("mftbu %0" : "=r" (rval)); rval;})
+#endif /* !__powerpc64__ */
 
 #define mttbl(v)	asm volatile("mttbl %0":: "r"(v))
 #define mttbu(v)	asm volatile("mttbu %0":: "r"(v))
Index: linux-2.6/include/asm-powerpc/time.h
===================================================================
--- linux-2.6.orig/include/asm-powerpc/time.h	2007-01-22 14:38:59.000000000 +1100
+++ linux-2.6/include/asm-powerpc/time.h	2007-01-24 15:21:03.000000000 +1100
@@ -78,30 +78,35 @@
 #define __USE_RTC()	0
 #endif
 
-/* On ppc64 this gets us the whole timebase; on ppc32 just the lower half */
+#ifdef CONFIG_PPC64
+
+/* For compatibility, get_tbl() is defined as get_tb() on ppc64 */
+#define get_tbl		get_tb
+
+#else
+
 static inline unsigned long get_tbl(void)
 {
-	unsigned long tbl;
-
 #if defined(CONFIG_403GCX)
+	unsigned long tbl;
 	asm volatile("mfspr %0, 0x3dd" : "=r" (tbl));
+	return tbl;
 #else
-	asm volatile("mftb %0" : "=r" (tbl));
+	return mftbl();
 #endif
-	return tbl;
 }
 
 static inline unsigned int get_tbu(void)
 {
+#ifdef CONFIG_403GCX
 	unsigned int tbu;
-
-#if defined(CONFIG_403GCX)
 	asm volatile("mfspr %0, 0x3dc" : "=r" (tbu));
+	return tbu;
 #else
-	asm volatile("mftbu %0" : "=r" (tbu));
+	return mftbu();
 #endif
-	return tbu;
 }
+#endif /* !CONFIG_PPC64 */
 
 static inline unsigned int get_rtcl(void)
 {
@@ -127,7 +132,7 @@
 {
 	return mftb();
 }
-#else
+#else /* CONFIG_PPC64 */
 static inline u64 get_tb(void)
 {
 	unsigned int tbhi, tblo, tbhi2;
@@ -140,7 +145,7 @@
 
 	return ((u64)tbhi << 32) | tblo;
 }
-#endif
+#endif /* !CONFIG_PPC64 */
 
 static inline void set_tb(unsigned int upper, unsigned int lower)
 {
Index: linux-2.6/include/asm-powerpc/timex.h
===================================================================
--- linux-2.6.orig/include/asm-powerpc/timex.h	2007-01-22 14:38:59.000000000 +1100
+++ linux-2.6/include/asm-powerpc/timex.h	2007-01-24 15:21:03.000000000 +1100
@@ -8,6 +8,7 @@
  */
 
 #include <asm/cputable.h>
+#include <asm/reg.h>
 
 #define CLOCK_TICK_RATE	1024000 /* Underlying HZ */
 
@@ -15,13 +16,11 @@
 
 static inline cycles_t get_cycles(void)
 {
-	cycles_t ret;
-
 #ifdef __powerpc64__
-
-	__asm__ __volatile__("mftb %0" : "=r" (ret) : );
-
+	return mftb();
 #else
+	cycles_t ret;
+
 	/*
 	 * For the "cycle" counter we use the timebase lower half.
 	 * Currently only used on SMP.
@@ -39,9 +38,8 @@
 		"	.long 99b\n"
 		".previous"
 		: "=r" (ret) : "i" (CPU_FTR_601));
-#endif
-
 	return ret;
+#endif
 }
 
 #endif	/* __KERNEL__ */

RHBZ#:
------
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=228099

Description:
------------
1 of 12 patches developed by Banjamin Herrenschmidt of IBM required for 
IBM customers purchasing Cell blade configurations.

This patch reworks the cell iic interrupt handling so that:

- Node ID is back in the interrupt number (only one IRQ host is created 
for all nodes). This allows interrupts from sources on another node to be 
routed non-locally. This will allow possibly one day to fix maxcpus=1 or 
2 and still get interrupts from devices on BE 1. (A bit more fixing is 
needed for that) and it will allow us to implement actual affinity 
control of external interrupts.

- Added handling of the IO exceptions interrupts (badly named, but Ben 
re-used the name initially used by STI). Those are the interrupts exposed 
by IIC_ISR and IIC_IRR, such as the IOC translation exception, performance 
monitor, etc... Those get their special numbers in the IRQ number space 
and are internally implemented as a cascade on unit 0xe, class 1 of each node.

RHEL Version Found:
-------------------
feature enhancement for 5.1.

Upstream Status:
----------------
This has been accepted into the kernel and can be viewed in the 2.6.20.3 
kernel.    

Test Status:
------------
This patch was tested by Benjamin herrenschmidt of IBM.  It has been
signed off by Olof Johansson and Paul Mackerras.

Proposed Patch:
----------------
Please review and ACK for RHEL 5.1.
-

Index: linux-malta-RH5/arch/powerpc/platforms/cell/interrupt.c
===================================================================
--- linux-malta-RH5.orig/arch/powerpc/platforms/cell/interrupt.c	2007-01-10 15:10:07.000000000 +1100
+++ linux-malta-RH5/arch/powerpc/platforms/cell/interrupt.c	2007-01-10 15:11:06.000000000 +1100
@@ -21,6 +21,12 @@
  * You should have received a copy of the GNU General Public License
  * along with this program; if not, write to the Free Software
  * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ *
+ * TODO:
+ * - Fix various assumptions related to HW CPU numbers vs. linux CPU numbers
+ *   vs node numbers in the setup code
+ * - Implement proper handling of maxcpus=1/2 (that is, routing of irqs from
+ *   a non-active node to the active node)
  */
 
 #include <linux/interrupt.h>
@@ -44,24 +50,25 @@ struct iic {
 	u8 target_id;
 	u8 eoi_stack[16];
 	int eoi_ptr;
-	struct irq_host *host;
+	struct device_node *node;
 };
 
 static DEFINE_PER_CPU(struct iic, iic);
 #define IIC_NODE_COUNT	2
-static struct irq_host *iic_hosts[IIC_NODE_COUNT];
+static struct irq_host *iic_host;
 
 /* Convert between "pending" bits and hw irq number */
 static irq_hw_number_t iic_pending_to_hwnum(struct cbe_iic_pending_bits bits)
 {
 	unsigned char unit = bits.source & 0xf;
+	unsigned char node = bits.source >> 4;
+	unsigned char class = bits.class & 3;
 
+	/* Decode IPIs */
 	if (bits.flags & CBE_IIC_IRQ_IPI)
-		return IIC_IRQ_IPI0 | (bits.prio >> 4);
-	else if (bits.class <= 3)
-		return (bits.class << 4) | unit;
+		return IIC_IRQ_TYPE_IPI | (bits.prio >> 4);
 	else
-		return IIC_IRQ_INVALID;
+		return (node << IIC_IRQ_NODE_SHIFT) | (class << 4) | unit;
 }
 
 static void iic_mask(unsigned int irq)
@@ -86,21 +93,70 @@ static struct irq_chip iic_chip = {
 	.eoi = iic_eoi,
 };
 
+
+static void iic_ioexc_eoi(unsigned int irq)
+{
+}
+
+static void iic_ioexc_cascade(unsigned int irq, struct irq_desc *desc,
+			    struct pt_regs *regs)
+{
+	struct cbe_iic_regs __iomem *node_iic = (void __iomem *)desc->handler_data;
+	unsigned int base = (irq & 0xffffff00) | IIC_IRQ_TYPE_IOEXC;
+	unsigned long bits, ack;
+	int cascade;
+
+	for (;;) {
+		bits = in_be64(&node_iic->iic_is);
+		if (bits == 0)
+			break;
+		/* pre-ack edge interrupts */
+		ack = bits & IIC_ISR_EDGE_MASK;
+		if (ack)
+			out_be64(&node_iic->iic_is, ack);
+		/* handle them */
+		for (cascade = 63; cascade >= 0; cascade--)
+			if (bits & (0x8000000000000000UL >> cascade)) {
+				unsigned int cirq =
+					irq_linear_revmap(iic_host,
+							  base | cascade);
+				if (cirq != NO_IRQ)
+					generic_handle_irq(cirq, regs);
+			}
+		/* post-ack level interrupts */
+		ack = bits & ~IIC_ISR_EDGE_MASK;
+		if (ack)
+			out_be64(&node_iic->iic_is, ack);
+	}
+	desc->chip->eoi(irq);
+}
+
+
+static struct irq_chip iic_ioexc_chip = {
+	.typename = " CELL-IOEX",
+	.mask = iic_mask,
+	.unmask = iic_unmask,
+	.eoi = iic_ioexc_eoi,
+};
+
 /* Get an IRQ number from the pending state register of the IIC */
 static unsigned int iic_get_irq(struct pt_regs *regs)
 {
-  	struct cbe_iic_pending_bits pending;
- 	struct iic *iic;
-
- 	iic = &__get_cpu_var(iic);
- 	*(unsigned long *) &pending =
- 		in_be64((unsigned long __iomem *) &iic->regs->pending_destr);
- 	iic->eoi_stack[++iic->eoi_ptr] = pending.prio;
- 	BUG_ON(iic->eoi_ptr > 15);
-	if (pending.flags & CBE_IIC_IRQ_VALID)
-		return irq_linear_revmap(iic->host,
- 					 iic_pending_to_hwnum(pending));
-	return NO_IRQ;
+	struct cbe_iic_pending_bits pending;
+	struct iic *iic;
+	unsigned int virq;
+
+	iic = &__get_cpu_var(iic);
+	*(unsigned long *) &pending =
+		in_be64((unsigned long __iomem *) &iic->regs->pending_destr);
+	if (!(pending.flags & CBE_IIC_IRQ_VALID))
+		return NO_IRQ;
+	virq = irq_linear_revmap(iic_host, iic_pending_to_hwnum(pending));
+	if (virq == NO_IRQ)
+		return NO_IRQ;
+	iic->eoi_stack[++iic->eoi_ptr] = pending.prio;
+	BUG_ON(iic->eoi_ptr > 15);
+	return virq;
 }
 
 #ifdef CONFIG_SMP
@@ -108,12 +164,7 @@ static unsigned int iic_get_irq(struct p
 /* Use the highest interrupt priorities for IPI */
 static inline int iic_ipi_to_irq(int ipi)
 {
-	return IIC_IRQ_IPI0 + IIC_NUM_IPIS - 1 - ipi;
-}
-
-static inline int iic_irq_to_ipi(int irq)
-{
-	return IIC_NUM_IPIS - 1 - (irq - IIC_IRQ_IPI0);
+	return IIC_IRQ_TYPE_IPI + 0xf - ipi;
 }
 
 void iic_setup_cpu(void)
@@ -123,7 +174,7 @@ void iic_setup_cpu(void)
 
 void iic_cause_IPI(int cpu, int mesg)
 {
-	out_be64(&per_cpu(iic, cpu).regs->generate, (IIC_NUM_IPIS - 1 - mesg) << 4);
+	out_be64(&per_cpu(iic, cpu).regs->generate, (0xf - mesg) << 4);
 }
 
 u8 iic_get_target_id(int cpu)
@@ -134,9 +185,7 @@ EXPORT_SYMBOL_GPL(iic_get_target_id);
 
 struct irq_host *iic_get_irq_host(int node)
 {
-	if (node < 0 || node >= IIC_NODE_COUNT)
-		return NULL;
-	return iic_hosts[node];
+	return iic_host;
 }
 EXPORT_SYMBOL_GPL(iic_get_irq_host);
 
@@ -149,34 +198,20 @@ static irqreturn_t iic_ipi_action(int ir
 
 	return IRQ_HANDLED;
 }
-
 static void iic_request_ipi(int ipi, const char *name)
 {
-	int node, virq;
+	int virq;
 
-	for (node = 0; node < IIC_NODE_COUNT; node++) {
-		char *rname;
-		if (iic_hosts[node] == NULL)
-			continue;
-		virq = irq_create_mapping(iic_hosts[node],
-					  iic_ipi_to_irq(ipi));
-		if (virq == NO_IRQ) {
-			printk(KERN_ERR
-			       "iic: failed to map IPI %s on node %d\n",
-			       name, node);
-			continue;
-		}
-		rname = kzalloc(strlen(name) + 16, GFP_KERNEL);
-		if (rname)
-			sprintf(rname, "%s node %d", name, node);
-		else
-			rname = (char *)name;
-		if (request_irq(virq, iic_ipi_action, IRQF_DISABLED,
-				rname, (void *)(long)ipi))
-			printk(KERN_ERR
-			       "iic: failed to request IPI %s on node %d\n",
-			       name, node);
+	virq = irq_create_mapping(iic_host, iic_ipi_to_irq(ipi));
+	if (virq == NO_IRQ) {
+		printk(KERN_ERR
+		       "iic: failed to map IPI %s\n", name);
+		return;
 	}
+	if (request_irq(virq, iic_ipi_action, IRQF_DISABLED, name,
+			(void *)(long)ipi))
+		printk(KERN_ERR
+		       "iic: failed to request IPI %s\n", name);
 }
 
 void iic_request_IPIs(void)
@@ -193,16 +228,24 @@ void iic_request_IPIs(void)
 
 static int iic_host_match(struct irq_host *h, struct device_node *node)
 {
-	return h->host_data != NULL && node == h->host_data;
+	return device_is_compatible(node,
+				    "IBM,CBEA-Internal-Interrupt-Controller");
 }
 
 static int iic_host_map(struct irq_host *h, unsigned int virq,
 			irq_hw_number_t hw)
 {
-	if (hw < IIC_IRQ_IPI0)
-		set_irq_chip_and_handler(virq, &iic_chip, handle_fasteoi_irq);
-	else
+	switch (hw & IIC_IRQ_TYPE_MASK) {
+	case IIC_IRQ_TYPE_IPI:
 		set_irq_chip_and_handler(virq, &iic_chip, handle_percpu_irq);
+		break;
+	case IIC_IRQ_TYPE_IOEXC:
+		set_irq_chip_and_handler(virq, &iic_ioexc_chip,
+					 handle_fasteoi_irq);
+		break;
+	default:
+		set_irq_chip_and_handler(virq, &iic_chip, handle_fasteoi_irq);
+	}
 	return 0;
 }
 
@@ -211,11 +254,39 @@ static int iic_host_xlate(struct irq_hos
 			   irq_hw_number_t *out_hwirq, unsigned int *out_flags)
 
 {
-	/* Currently, we don't translate anything. That needs to be fixed as
-	 * we get better defined device-trees. iic interrupts have to be
-	 * explicitely mapped by whoever needs them
-	 */
-	return -ENODEV;
+	unsigned int node, ext, unit, class;
+	const u32 *val;
+
+	if (!device_is_compatible(ct,
+				     "IBM,CBEA-Internal-Interrupt-Controller"))
+		return -ENODEV;
+	if (intsize != 1)
+		return -ENODEV;
+	val = get_property(ct, "#interrupt-cells", NULL);
+	if (val == NULL || *val != 1)
+		return -ENODEV;
+
+	node = intspec[0] >> 24;
+	ext = (intspec[0] >> 16) & 0xff;
+	class = (intspec[0] >> 8) & 0xff;
+	unit = intspec[0] & 0xff;
+
+	/* Check if node is in supported range */
+	if (node > 1)
+		return -EINVAL;
+
+	/* Build up interrupt number, special case for IO exceptions */
+	*out_hwirq = (node << IIC_IRQ_NODE_SHIFT);
+	if (unit == IIC_UNIT_IIC && class == 1)
+		*out_hwirq |= IIC_IRQ_TYPE_IOEXC | ext;
+	else
+		*out_hwirq |= IIC_IRQ_TYPE_NORMAL |
+			(class << IIC_IRQ_CLASS_SHIFT) | unit;
+
+	/* Dummy flags, ignored by iic code */
+	*out_flags = IRQ_TYPE_EDGE_RISING;
+
+	return 0;
 }
 
 static struct irq_host_ops iic_host_ops = {
@@ -225,7 +296,7 @@ static struct irq_host_ops iic_host_ops 
 };
 
 static void __init init_one_iic(unsigned int hw_cpu, unsigned long addr,
-				struct irq_host *host)
+				struct device_node *node)
 {
 	/* XXX FIXME: should locate the linux CPU number from the HW cpu
 	 * number properly. We are lucky for now
@@ -237,19 +308,19 @@ static void __init init_one_iic(unsigned
 
 	iic->target_id = ((hw_cpu & 2) << 3) | ((hw_cpu & 1) ? 0xf : 0xe);
 	iic->eoi_stack[0] = 0xff;
-	iic->host = host;
+	iic->node = of_node_get(node);
 	out_be64(&iic->regs->prio, 0);
 
-	printk(KERN_INFO "IIC for CPU %d at %lx mapped to %p, target id 0x%x\n",
-	       hw_cpu, addr, iic->regs, iic->target_id);
+	printk(KERN_INFO "IIC for CPU %d target id 0x%x : %s\n",
+	       hw_cpu, iic->target_id, node->full_name);
 }
 
 static int __init setup_iic(void)
 {
 	struct device_node *dn;
 	struct resource r0, r1;
-	struct irq_host *host;
-	int found = 0;
+	unsigned int node, cascade, found = 0;
+	struct cbe_iic_regs __iomem *node_iic;
  	u32 *np;
 
 	for (dn = NULL;
@@ -270,19 +341,37 @@ static int __init setup_iic(void)
 			of_node_put(dn);
 			return -ENODEV;
 		}
-		host = NULL;
-		if (found < IIC_NODE_COUNT) {
-			host = irq_alloc_host(IRQ_HOST_MAP_LINEAR,
-					      IIC_SOURCE_COUNT,
-					      &iic_host_ops,
-					      IIC_IRQ_INVALID);
-			iic_hosts[found] = host;
-			BUG_ON(iic_hosts[found] == NULL);
-			iic_hosts[found]->host_data = of_node_get(dn);
-			found++;
-		}
-		init_one_iic(np[0], r0.start, host);
-		init_one_iic(np[1], r1.start, host);
+		found++;
+		init_one_iic(np[0], r0.start, dn);
+		init_one_iic(np[1], r1.start, dn);
+
+		/* Setup cascade for IO exceptions. XXX cleanup tricks to get
+		 * node vs CPU etc...
+		 * Note that we configure the IIC_IRR here with a hard coded
+		 * priority of 1. We might want to improve that later.
+		 */
+		node = np[0] >> 1;
+		node_iic = cbe_get_cpu_iic_regs(np[0]);
+		cascade = node << IIC_IRQ_NODE_SHIFT;
+		cascade |= 1 << IIC_IRQ_CLASS_SHIFT;
+		cascade |= IIC_UNIT_IIC;
+		cascade = irq_create_mapping(iic_host, cascade);
+		if (cascade == NO_IRQ)
+			continue;
+		/*
+		 * irq_data is a generic pointer that gets passed back
+		 * to us later, so the forced cast is fine.
+		 */
+		set_irq_data(cascade, (void __force *)node_iic);
+		set_irq_chained_handler(cascade , iic_ioexc_cascade);
+		out_be64(&node_iic->iic_ir,
+			 (1 << 12)		/* priority */ |
+			 (node << 4)		/* dest node */ |
+			 IIC_UNIT_THREAD_0	/* route them to thread 0 */);
+		/* Flush pending (make sure it triggers if there is
+		 * anything pending
+		 */
+		out_be64(&node_iic->iic_is, 0xfffffffffffffffful);
 	}
 
 	if (found)
@@ -293,6 +382,12 @@ static int __init setup_iic(void)
 
 void __init iic_init_IRQ(void)
 {
+	/* Setup an irq host data structure */
+	iic_host = irq_alloc_host(IRQ_HOST_MAP_LINEAR, IIC_SOURCE_COUNT,
+				  &iic_host_ops, IIC_IRQ_INVALID);
+	BUG_ON(iic_host == NULL);
+	irq_set_default_host(iic_host);
+
 	/* Discover and initialize iics */
 	if (setup_iic() < 0)
 		panic("IIC: Failed to initialize !\n");
@@ -303,3 +398,19 @@ void __init iic_init_IRQ(void)
 	/* Enable on current CPU */
 	iic_setup_cpu();
 }
+
+void iic_set_interrupt_routing(int cpu, int thread, int priority)
+{
+	struct cbe_iic_regs __iomem *iic_regs = cbe_get_cpu_iic_regs(cpu);
+	u64 iic_ir = 0;
+	int node = cpu >> 1;
+
+	/* Set which node and thread will handle the next interrupt */
+	iic_ir |= CBE_IIC_IR_PRIO(priority) |
+		  CBE_IIC_IR_DEST_NODE(node);
+	if (thread == 0)
+		iic_ir |= CBE_IIC_IR_DEST_UNIT(CBE_IIC_IR_PT_0);
+	else
+		iic_ir |= CBE_IIC_IR_DEST_UNIT(CBE_IIC_IR_PT_1);
+	out_be64(&iic_regs->iic_ir, iic_ir);
+}
Index: linux-malta-RH5/arch/powerpc/platforms/cell/interrupt.h
===================================================================
--- linux-malta-RH5.orig/arch/powerpc/platforms/cell/interrupt.h	2007-01-10 15:10:07.000000000 +1100
+++ linux-malta-RH5/arch/powerpc/platforms/cell/interrupt.h	2007-01-10 15:11:06.000000000 +1100
@@ -2,48 +2,76 @@
 #define ASM_CELL_PIC_H
 #ifdef __KERNEL__
 /*
- * Mapping of IIC pending bits into per-node
- * interrupt numbers.
+ * Mapping of IIC pending bits into per-node interrupt numbers.
  *
- * IRQ     FF CC SS PP   FF CC SS PP	Description
+ * Interrupt numbers are in the range 0...0x1ff where the top bit
+ * (0x100) represent the source node. Only 2 nodes are supported with
+ * the current code though it's trivial to extend that if necessary using
+ * higher level bits
  *
- * 00-3f   80 02 +0 00 - 80 02 +0 3f	South Bridge
- * 00-3f   80 02 +b 00 - 80 02 +b 3f	South Bridge
- * 41-4a   80 00 +1 ** - 80 00 +a **	SPU Class 0
- * 51-5a   80 01 +1 ** - 80 01 +a **	SPU Class 1
- * 61-6a   80 02 +1 ** - 80 02 +a **	SPU Class 2
- * 70-7f   C0 ** ** 00 - C0 ** ** 0f	IPI
+ * The bottom 8 bits are split into 2 type bits and 6 data bits that
+ * depend on the type:
  *
- *    F flags
- *    C class
- *    S source
- *    P Priority
- *    + node number
- *    * don't care
+ * 00 (0x00 | data) : normal interrupt. data is (class << 4) | source
+ * 01 (0x40 | data) : IO exception. data is the exception number as
+ *                    defined by bit numbers in IIC_SR
+ * 10 (0x80 | data) : IPI. data is the IPI number (obtained from the priority)
+ *                    and node is always 0 (IPIs are per-cpu, their source is
+ *                    not relevant)
+ * 11 (0xc0 | data) : reserved
  *
- * A node consists of a Cell Broadband Engine and an optional
- * south bridge device providing a maximum of 64 IRQs.
- * The south bridge may be connected to either IOIF0
- * or IOIF1.
- * Each SPE is represented as three IRQ lines, one per
- * interrupt class.
- * 16 IRQ numbers are reserved for inter processor
- * interruptions, although these are only used in the
- * range of the first node.
+ * In addition, interrupt number 0x80000000 is defined as always invalid
+ * (that is the node field is expected to never extend to move than 23 bits)
  *
- * This scheme needs 128 IRQ numbers per BIF node ID,
- * which means that with the total of 512 lines
- * available, we can have a maximum of four nodes.
  */
 
 enum {
-	IIC_IRQ_INVALID		= 0xff,
-	IIC_IRQ_MAX		= 0x3f,
-	IIC_IRQ_EXT_IOIF0	= 0x20,
-	IIC_IRQ_EXT_IOIF1	= 0x2b,
-	IIC_IRQ_IPI0		= 0x40,
-	IIC_NUM_IPIS    	= 0x10, /* IRQs reserved for IPI */
-	IIC_SOURCE_COUNT	= 0x50,
+	IIC_IRQ_INVALID		= 0x80000000u,
+	IIC_IRQ_NODE_MASK	= 0x100,
+	IIC_IRQ_NODE_SHIFT	= 8,
+	IIC_IRQ_MAX		= 0x1ff,
+	IIC_IRQ_TYPE_MASK	= 0xc0,
+	IIC_IRQ_TYPE_NORMAL	= 0x00,
+	IIC_IRQ_TYPE_IOEXC	= 0x40,
+	IIC_IRQ_TYPE_IPI	= 0x80,
+	IIC_IRQ_CLASS_SHIFT	= 4,
+	IIC_IRQ_CLASS_0		= 0x00,
+	IIC_IRQ_CLASS_1		= 0x10,
+	IIC_IRQ_CLASS_2		= 0x20,
+	IIC_SOURCE_COUNT	= 0x200,
+
+	/* Here are defined the various source/dest units. Avoid using those
+	 * definitions if you can, they are mostly here for reference
+	 */
+	IIC_UNIT_SPU_0		= 0x4,
+	IIC_UNIT_SPU_1		= 0x7,
+	IIC_UNIT_SPU_2		= 0x3,
+	IIC_UNIT_SPU_3		= 0x8,
+	IIC_UNIT_SPU_4		= 0x2,
+	IIC_UNIT_SPU_5		= 0x9,
+	IIC_UNIT_SPU_6		= 0x1,
+	IIC_UNIT_SPU_7		= 0xa,
+	IIC_UNIT_IOC_0		= 0x0,
+	IIC_UNIT_IOC_1		= 0xb,
+	IIC_UNIT_THREAD_0	= 0xe, /* target only */
+	IIC_UNIT_THREAD_1	= 0xf, /* target only */
+	IIC_UNIT_IIC		= 0xe, /* source only (IO exceptions) */
+
+	/* Base numbers for the external interrupts */
+	IIC_IRQ_EXT_IOIF0	=
+		IIC_IRQ_TYPE_NORMAL | IIC_IRQ_CLASS_2 | IIC_UNIT_IOC_0,
+	IIC_IRQ_EXT_IOIF1	=
+		IIC_IRQ_TYPE_NORMAL | IIC_IRQ_CLASS_2 | IIC_UNIT_IOC_1,
+
+	/* Base numbers for the IIC_ISR interrupts */
+	IIC_IRQ_IOEX_TMI	= IIC_IRQ_TYPE_IOEXC | IIC_IRQ_CLASS_1 | 63,
+	IIC_IRQ_IOEX_PMI	= IIC_IRQ_TYPE_IOEXC | IIC_IRQ_CLASS_1 | 62,
+	IIC_IRQ_IOEX_ATI	= IIC_IRQ_TYPE_IOEXC | IIC_IRQ_CLASS_1 | 61,
+	IIC_IRQ_IOEX_MATBFI	= IIC_IRQ_TYPE_IOEXC | IIC_IRQ_CLASS_1 | 60,
+	IIC_IRQ_IOEX_ELDI	= IIC_IRQ_TYPE_IOEXC | IIC_IRQ_CLASS_1 | 59,
+
+	/* Which bits in IIC_ISR are edge sensitive */
+	IIC_ISR_EDGE_MASK	= 0x4ul,
 };
 
 extern void iic_init_IRQ(void);
@@ -52,9 +80,10 @@ extern void iic_request_IPIs(void);
 extern void iic_setup_cpu(void);
 
 extern u8 iic_get_target_id(int cpu);
-extern struct irq_host *iic_get_irq_host(int node);
 
 extern void spider_init_IRQ(void);
 
+extern void iic_set_interrupt_routing(int cpu, int thread, int priority);
+
 #endif
 #endif /* ASM_CELL_PIC_H */
Index: linux-malta-RH5/arch/powerpc/platforms/cell/spider-pic.c
===================================================================
--- linux-malta-RH5.orig/arch/powerpc/platforms/cell/spider-pic.c	2007-01-10 15:10:07.000000000 +1100
+++ linux-malta-RH5/arch/powerpc/platforms/cell/spider-pic.c	2007-01-10 15:11:06.000000000 +1100
@@ -243,7 +243,6 @@ static unsigned int __init spider_find_c
 	u32 *imap, *tmp;
 	int imaplen, intsize, unit;
 	struct device_node *iic;
-	struct irq_host *iic_host;
 
 #if 0 /* Enable that when we have a way to retreive the node as well */
 	/* First, we check wether we have a real "interrupts" in the device
@@ -289,11 +288,11 @@ static unsigned int __init spider_find_c
 	 * the iic host from the iic OF node, but that way I'm still compatible
 	 * with really really old old firmwares for which we don't have a node
 	 */
-	iic_host = iic_get_irq_host(pic->node_id);
-	if (iic_host == NULL)
-		return NO_IRQ;
 	/* Manufacture an IIC interrupt number of class 2 */
-	virq = irq_create_mapping(iic_host, 0x20 | unit);
+	virq = irq_create_mapping(NULL,
+				  (pic->node_id << IIC_IRQ_NODE_SHIFT) |
+				  (2 << IIC_IRQ_CLASS_SHIFT) |
+				  unit);
 	if (virq == NO_IRQ)
 		printk(KERN_ERR "spider_pic: failed to map cascade !");
 	return virq;
Index: linux-malta-RH5/arch/powerpc/platforms/cell/spu_base.c
===================================================================
--- linux-malta-RH5.orig/arch/powerpc/platforms/cell/spu_base.c	2007-01-10 15:10:07.000000000 +1100
+++ linux-malta-RH5/arch/powerpc/platforms/cell/spu_base.c	2007-01-10 15:11:06.000000000 +1100
@@ -568,24 +568,23 @@ static void spu_unmap(struct spu *spu)
 /* This function shall be abstracted for HV platforms */
 static int __init spu_map_interrupts(struct spu *spu, struct device_node *np)
 {
-	struct irq_host *host;
 	unsigned int isrc;
 	u32 *tmp;
 
-	host = iic_get_irq_host(spu->node);
-	if (host == NULL)
-		return -ENODEV;
-
-	/* Get the interrupt source from the device-tree */
+	/* Get the interrupt source unit from the device-tree */
 	tmp = (u32 *)get_property(np, "isrc", NULL);
 	if (!tmp)
 		return -ENODEV;
-	spu->isrc = isrc = tmp[0];
+	isrc = tmp[0];
+
+	/* Add the node number */
+	isrc |= spu->node << IIC_IRQ_NODE_SHIFT;
+	spu->isrc = isrc;
 
 	/* Now map interrupts of all 3 classes */
-	spu->irqs[0] = irq_create_mapping(host, 0x00 | isrc);
-	spu->irqs[1] = irq_create_mapping(host, 0x10 | isrc);
-	spu->irqs[2] = irq_create_mapping(host, 0x20 | isrc);
+	spu->irqs[0] = irq_create_mapping(NULL, IIC_IRQ_CLASS_0 | isrc);
+	spu->irqs[1] = irq_create_mapping(NULL, IIC_IRQ_CLASS_1 | isrc);
+	spu->irqs[2] = irq_create_mapping(NULL, IIC_IRQ_CLASS_2 | isrc);
 
 	/* Right now, we only fail if class 2 failed */
 	return spu->irqs[2] == NO_IRQ ? -EINVAL : 0;
Index: linux-malta-RH5/arch/powerpc/platforms/cell/cbe_regs.h
===================================================================
--- linux-malta-RH5.orig/arch/powerpc/platforms/cell/cbe_regs.h	2007-01-10 15:10:07.000000000 +1100
+++ linux-malta-RH5/arch/powerpc/platforms/cell/cbe_regs.h	2007-01-10 15:11:06.000000000 +1100
@@ -102,18 +102,28 @@ struct cbe_iic_regs {
 
 	/* IIC interrupt registers */
 	struct	cbe_iic_thread_regs thread[2];			/* 0x0400 */
-	u64     iic_ir;						/* 0x0440 */
-	u64     iic_is;						/* 0x0448 */
+
+	u64	iic_ir;						/* 0x0440 */
+#define CBE_IIC_IR_PRIO(x)      (((x) & 0xf) << 12)
+#define CBE_IIC_IR_DEST_NODE(x) (((x) & 0xf) << 4)
+#define CBE_IIC_IR_DEST_UNIT(x) ((x) & 0xf)
+#define CBE_IIC_IR_IOC_0        0x0
+#define CBE_IIC_IR_IOC_1S       0xb
+#define CBE_IIC_IR_PT_0         0xe
+#define CBE_IIC_IR_PT_1         0xf
+
+	u64	iic_is;						/* 0x0448 */
+#define CBE_IIC_IS_PMI		0x2
 
 	u8	pad_0x0450_0x0500[0x0500 - 0x0450];		/* 0x0450 */
 
 	/* IOC FIR */
 	u64	ioc_fir_reset;					/* 0x0500 */
-	u64	ioc_fir_set;
-	u64	ioc_checkstop_enable;
-	u64	ioc_fir_error_mask;
-	u64	ioc_syserr_enable;
-	u64	ioc_fir;
+	u64	ioc_fir_set;					/* 0x0508 */
+	u64	ioc_checkstop_enable;				/* 0x0510 */
+	u64	ioc_fir_error_mask;				/* 0x0518 */
+	u64	ioc_syserr_enable;				/* 0x0520 */
+	u64	ioc_fir;					/* 0x0528 */
 
 	u8	pad_0x0530_0x1000[0x1000 - 0x0530];		/* 0x0530 */
 };

RHBZ#:
------
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=228099

Description:
------------
7 of 12 patches by Ben Herrenschmidt of IBM required for supporting IBM 
customers purchasing Cell blade configurations.  

This patch adds new dcr_map, dcr_read, and dcr_write accessors for DCRs 
that can be used by drivers to transparently address either native DCRs or 
memory mapped DCRs.  The implementation for memory mapped DCRs is done 
after the binding being currently worked on fro SLOF and the Axon chipset.  
This patch enables it for the cell native platform.

RHEL Version Found:
-------------------
feature enhancement for 5.1.

Upstream Status:
----------------
This has been accepted into the kernel and can be viewed in the 2.6.20.3 
kernel.

Test Status:
------------
This patch was tested by Benjamin Herrenschmidt of IBM.  It has been
ack'd by Olof Johansson and signed off by Paul Mackerras.

Proposed Patch:
----------------
Please review and ACK for RHEL 5.1.
-

Index: linux-malta-RH5/arch/powerpc/Kconfig
===================================================================
--- linux-malta-RH5.orig/arch/powerpc/Kconfig	2007-01-10 14:46:06.000000000 +1100
+++ linux-malta-RH5/arch/powerpc/Kconfig	2007-01-10 15:24:53.000000000 +1100
@@ -160,9 +160,11 @@ config PPC_86xx
 
 config 40x
 	bool "AMCC 40x"
+	select PPC_DCR_NATIVE
 
 config 44x
 	bool "AMCC 44x"
+	select PPC_DCR_NATIVE
 
 config 8xx
 	bool "Freescale 8xx"
@@ -208,6 +210,19 @@ config PPC_FPU
 	bool
 	default y if PPC64
 
+config PPC_DCR_NATIVE
+	bool
+	default n
+
+config PPC_DCR_MMIO
+	bool
+	default n
+
+config PPC_DCR
+	bool
+	depends on PPC_DCR_NATIVE || PPC_DCR_MMIO
+	default y
+
 config BOOKE
 	bool
 	depends on E200 || E500
@@ -424,6 +439,7 @@ config PPC_CELL
 config PPC_CELL_NATIVE
 	bool
 	select PPC_CELL
+	select PPC_DCR_MMIO
 	default n
 
 config PPC_IBM_CELL_BLADE
Index: linux-malta-RH5/arch/powerpc/kernel/Makefile
===================================================================
--- linux-malta-RH5.orig/arch/powerpc/kernel/Makefile	2007-01-10 14:46:06.000000000 +1100
+++ linux-malta-RH5/arch/powerpc/kernel/Makefile	2007-01-10 15:18:58.000000000 +1100
@@ -62,6 +62,7 @@ obj-$(CONFIG_BOOTX_TEXT)	+= btext.o
 obj-$(CONFIG_SMP)		+= smp.o
 obj-$(CONFIG_KPROBES)		+= kprobes.o
 obj-$(CONFIG_PPC_UDBG_16550)	+= legacy_serial.o udbg_16550.o
+
 module-$(CONFIG_PPC64)		+= module_64.o
 obj-$(CONFIG_MODULES)		+= $(module-y)
 
Index: linux-malta-RH5/arch/powerpc/sysdev/Makefile
===================================================================
--- linux-malta-RH5.orig/arch/powerpc/sysdev/Makefile	2007-01-10 14:46:07.000000000 +1100
+++ linux-malta-RH5/arch/powerpc/sysdev/Makefile	2007-01-10 15:21:59.000000000 +1100
@@ -5,8 +5,8 @@ endif
 obj-$(CONFIG_MPIC)		+= mpic.o
 obj-$(CONFIG_PPC_INDIRECT_PCI)	+= indirect_pci.o
 obj-$(CONFIG_PPC_MPC106)	+= grackle.o
-obj-$(CONFIG_BOOKE)		+= dcr.o
-obj-$(CONFIG_40x)		+= dcr.o
+obj-$(CONFIG_PPC_DCR)		+= dcr.o
+obj-$(CONFIG_PPC_DCR_NATIVE)	+= dcr-low.o
 obj-$(CONFIG_U3_DART)		+= dart_iommu.o
 obj-$(CONFIG_MMIO_NVRAM)	+= mmio_nvram.o
 obj-$(CONFIG_FSL_SOC)		+= fsl_soc.o
Index: linux-malta-RH5/arch/powerpc/sysdev/dcr-low.S
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-malta-RH5/arch/powerpc/sysdev/dcr-low.S	2007-01-10 15:18:58.000000000 +1100
@@ -0,0 +1,39 @@
+/*
+ * "Indirect" DCR access
+ *
+ * Copyright (c) 2004 Eugene Surovegin <ebs@ebshome.net>
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under  the terms of  the GNU General Public License as published by the
+ * Free Software Foundation;  either version 2 of the License, or (at your
+ * option) any later version.
+ */
+
+#include <asm/ppc_asm.h>
+#include <asm/processor.h>
+
+#define DCR_ACCESS_PROLOG(table) \
+	rlwinm  r3,r3,4,18,27;   \
+	lis     r5,table@h;      \
+	ori     r5,r5,table@l;   \
+	add     r3,r3,r5;        \
+	mtctr   r3;              \
+	bctr
+
+_GLOBAL(__mfdcr)
+	DCR_ACCESS_PROLOG(__mfdcr_table)
+
+_GLOBAL(__mtdcr)
+	DCR_ACCESS_PROLOG(__mtdcr_table)
+
+__mfdcr_table:
+	mfdcr  r3,0; blr
+__mtdcr_table:
+	mtdcr  0,r4; blr
+
+dcr     = 1
+        .rept   1023
+	mfdcr   r3,dcr; blr
+	mtdcr   dcr,r4; blr
+	dcr     = dcr + 1
+	.endr
Index: linux-malta-RH5/arch/powerpc/sysdev/dcr.c
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-malta-RH5/arch/powerpc/sysdev/dcr.c	2007-01-10 15:25:32.000000000 +1100
@@ -0,0 +1,137 @@
+/*
+ * (c) Copyright 2006 Benjamin Herrenschmidt, IBM Corp.
+ *                    <benh@kernel.crashing.org>
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+
+#undef DEBUG
+
+#include <linux/kernel.h>
+#include <asm/prom.h>
+#include <asm/dcr.h>
+
+unsigned int dcr_resource_start(struct device_node *np, unsigned int index)
+{
+	unsigned int ds;
+	u32 *dr = (u32 *)get_property(np, "dcr-reg", &ds);
+
+	if (dr == NULL || ds & 1 || index >= (ds / 8))
+		return 0;
+
+	return dr[index * 2];
+}
+
+unsigned int dcr_resource_len(struct device_node *np, unsigned int index)
+{
+	unsigned int ds;
+	u32 *dr = (u32 *)get_property(np, "dcr-reg", &ds);
+
+	if (dr == NULL || ds & 1 || index >= (ds / 8))
+		return 0;
+
+	return dr[index * 2 + 1];
+}
+
+#ifndef CONFIG_PPC_DCR_NATIVE
+
+static struct device_node * find_dcr_parent(struct device_node * node)
+{
+	struct device_node *par, *tmp;
+	u32 *p;
+
+	for (par = of_node_get(node); par;) {
+		if (get_property(par, "dcr-controller", NULL))
+			break;
+		p = (u32 *)get_property(par, "dcr-parent", NULL);
+		tmp = par;
+		if (p == NULL)
+			par = of_get_parent(par);
+		else
+			par = of_find_node_by_phandle(*p);
+		of_node_put(tmp);
+	}
+	return par;
+}
+
+u64 of_translate_dcr_address(struct device_node *dev,
+			     unsigned int dcr_n,
+			     unsigned int *out_stride)
+{
+	struct device_node *dp;
+	u32 *p;
+	unsigned int stride;
+	u64 ret;
+
+	dp = find_dcr_parent(dev);
+	if (dp == NULL)
+		return OF_BAD_ADDR;
+
+	/* Stride is not properly defined yet, default to 0x10 for Axon */
+	p = (u32 *)get_property(dp, "dcr-mmio-stride", NULL);
+	stride = (p == NULL) ? 0x10 : *p;
+
+	/* XXX FIXME: Which property name is to use of the 2 following ? */
+	p = (u32 *)get_property(dp, "dcr-mmio-range", NULL);
+	if (p == NULL)
+		p = (u32 *)get_property(dp, "dcr-mmio-space", NULL);
+	if (p == NULL)
+		return OF_BAD_ADDR;
+
+	/* Maybe could do some better range checking here */
+	ret = of_translate_address(dp, p);
+	if (ret != OF_BAD_ADDR)
+		ret += (u64)(stride) * (u64)dcr_n;
+	if (out_stride)
+		*out_stride = stride;
+	return ret;
+}
+
+dcr_host_t dcr_map(struct device_node *dev, unsigned int dcr_n,
+		   unsigned int dcr_c)
+{
+	dcr_host_t ret = { .token = NULL, .stride = 0 };
+	u64 addr;
+
+	pr_debug("dcr_map(%s, 0x%x, 0x%x)\n",
+		 dev->full_name, dcr_n, dcr_c);
+
+	addr = of_translate_dcr_address(dev, dcr_n, &ret.stride);
+	pr_debug("translates to addr: 0x%lx, stride: 0x%x\n",
+		 addr, ret.stride);
+	if (addr == OF_BAD_ADDR)
+		return ret;
+	pr_debug("mapping 0x%x bytes\n", dcr_c * ret.stride);
+	ret.token = ioremap(addr, dcr_c * ret.stride);
+	if (ret.token == NULL)
+		return ret;
+	pr_debug("mapped at 0x%p -> base is 0x%p\n",
+		 ret.token, ret.token - dcr_n * ret.stride);
+	ret.token -= dcr_n * ret.stride;
+	return ret;
+}
+
+void dcr_unmap(dcr_host_t host, unsigned int dcr_n, unsigned int dcr_c)
+{
+	dcr_host_t h = host;
+
+	if (h.token == NULL)
+		return;
+	h.token -= dcr_n * h.stride;
+	iounmap(h.token);
+	h.token = NULL;
+}
+
+#endif /* !defined(CONFIG_PPC_DCR_NATIVE) */
Index: linux-malta-RH5/arch/ppc/Kconfig
===================================================================
--- linux-malta-RH5.orig/arch/ppc/Kconfig	2007-01-10 14:46:07.000000000 +1100
+++ linux-malta-RH5/arch/ppc/Kconfig	2007-01-10 15:18:58.000000000 +1100
@@ -77,9 +77,11 @@ config 6xx
 
 config 40x
 	bool "40x"
+	select PPC_DCR_NATIVE
 
 config 44x
 	bool "44x"
+	select PPC_DCR_NATIVE
 
 config 8xx
 	bool "8xx"
@@ -95,6 +97,15 @@ endchoice
 config PPC_FPU
 	bool
 
+config PPC_DCR_NATIVE
+	bool
+	default n
+
+config PPC_DCR
+	bool
+	depends on PPC_DCR_NATIVE
+	default y
+
 config BOOKE
 	bool
 	depends on E200 || E500
Index: linux-malta-RH5/include/asm-powerpc/dcr-mmio.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-malta-RH5/include/asm-powerpc/dcr-mmio.h	2007-01-10 15:18:58.000000000 +1100
@@ -0,0 +1,51 @@
+/*
+ * (c) Copyright 2006 Benjamin Herrenschmidt, IBM Corp.
+ *                    <benh@kernel.crashing.org>
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+
+#ifndef _ASM_POWERPC_DCR_MMIO_H
+#define _ASM_POWERPC_DCR_MMIO_H
+#ifdef __KERNEL__
+
+#include <asm/io.h>
+
+typedef struct { void __iomem *token; unsigned int stride; } dcr_host_t;
+
+#define DCR_MAP_OK(host)	((host).token != NULL)
+
+extern dcr_host_t dcr_map(struct device_node *dev, unsigned int dcr_n,
+			  unsigned int dcr_c);
+extern void dcr_unmap(dcr_host_t host, unsigned int dcr_n, unsigned int dcr_c);
+
+static inline u32 dcr_read(dcr_host_t host, unsigned int dcr_n)
+{
+	return in_be32(host.token + dcr_n * host.stride);
+}
+
+static inline void dcr_write(dcr_host_t host, unsigned int dcr_n, u32 value)
+{
+	out_be32(host.token + dcr_n * host.stride, value);
+}
+
+extern u64 of_translate_dcr_address(struct device_node *dev,
+				    unsigned int dcr_n,
+				    unsigned int *stride);
+
+#endif /* __KERNEL__ */
+#endif /* _ASM_POWERPC_DCR_MMIO_H */
+
+
Index: linux-malta-RH5/include/asm-powerpc/dcr-native.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-malta-RH5/include/asm-powerpc/dcr-native.h	2007-01-10 15:21:59.000000000 +1100
@@ -0,0 +1,72 @@
+/*
+ * (c) Copyright 2006 Benjamin Herrenschmidt, IBM Corp.
+ *                    <benh@kernel.crashing.org>
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+
+#ifndef _ASM_POWERPC_DCR_NATIVE_H
+#define _ASM_POWERPC_DCR_NATIVE_H
+#ifdef __KERNEL__
+#ifndef __ASSEMBLY__
+
+typedef struct {} dcr_host_t;
+
+#define DCR_MAP_OK(host)	(1)
+
+#define dcr_map(dev, dcr_n, dcr_c)	{}
+#define dcr_unmap(host, dcr_n, dcr_c)	{}
+#define dcr_read(host, dcr_n)		mfdcr(dcr_n)
+#define dcr_write(host, dcr_n, value)	mtdcr(dcr_n, value)
+
+/* Device Control Registers */
+void __mtdcr(int reg, unsigned int val);
+unsigned int __mfdcr(int reg);
+#define mfdcr(rn)						\
+	({unsigned int rval;					\
+	if (__builtin_constant_p(rn))				\
+		asm volatile("mfdcr %0," __stringify(rn)	\
+		              : "=r" (rval));			\
+	else							\
+		rval = __mfdcr(rn);				\
+	rval;})
+
+#define mtdcr(rn, v)						\
+do {								\
+	if (__builtin_constant_p(rn))				\
+		asm volatile("mtdcr " __stringify(rn) ",%0"	\
+			      : : "r" (v)); 			\
+	else							\
+		__mtdcr(rn, v);					\
+} while (0)
+
+/* R/W of indirect DCRs make use of standard naming conventions for DCRs */
+#define mfdcri(base, reg)			\
+({						\
+	mtdcr(base ## _CFGADDR, base ## _ ## reg);	\
+	mfdcr(base ## _CFGDATA);			\
+})
+
+#define mtdcri(base, reg, data)			\
+do {						\
+	mtdcr(base ## _CFGADDR, base ## _ ## reg);	\
+	mtdcr(base ## _CFGDATA, data);		\
+} while (0)
+
+#endif /* __ASSEMBLY__ */
+#endif /* __KERNEL__ */
+#endif /* _ASM_POWERPC_DCR_NATIVE_H */
+
+
Index: linux-malta-RH5/include/asm-powerpc/dcr.h
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-malta-RH5/include/asm-powerpc/dcr.h	2007-01-10 15:21:59.000000000 +1100
@@ -0,0 +1,44 @@
+/*
+ * (c) Copyright 2006 Benjamin Herrenschmidt, IBM Corp.
+ *                    <benh@kernel.crashing.org>
+ *
+ *   This program is free software;  you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY;  without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See
+ *   the GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with this program;  if not, write to the Free Software
+ *   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
+ */
+
+#ifndef _ASM_POWERPC_DCR_H
+#define _ASM_POWERPC_DCR_H
+#ifdef __KERNEL__
+#ifdef CONFIG_PPC_DCR
+
+#ifdef CONFIG_PPC_DCR_NATIVE
+#include <asm/dcr-native.h>
+#else
+#include <asm/dcr-mmio.h>
+#endif
+
+/*
+ * On CONFIG_PPC_MERGE, we have additional helpers to read the DCR
+ * base from the device-tree
+ */
+#ifdef CONFIG_PPC_MERGE
+extern unsigned int dcr_resource_start(struct device_node *np,
+				       unsigned int index);
+extern unsigned int dcr_resource_len(struct device_node *np,
+				     unsigned int index);
+#endif /* CONFIG_PPC_MERGE */
+
+#endif /* CONFIG_PPC_DCR */
+#endif /* __KERNEL__ */
+#endif /* _ASM_POWERPC_DCR_H */
Index: linux-malta-RH5/arch/powerpc/sysdev/dcr.S
===================================================================
--- linux-malta-RH5.orig/arch/powerpc/sysdev/dcr.S	2007-01-10 15:19:14.000000000 +1100
+++ /dev/null	1970-01-01 00:00:00.000000000 +0000
@@ -1,39 +0,0 @@
-/*
- * "Indirect" DCR access
- *
- * Copyright (c) 2004 Eugene Surovegin <ebs@ebshome.net>
- *
- * This program is free software; you can redistribute it and/or modify it
- * under  the terms of  the GNU General Public License as published by the
- * Free Software Foundation;  either version 2 of the License, or (at your
- * option) any later version.
- */
-
-#include <asm/ppc_asm.h>
-#include <asm/processor.h>
-
-#define DCR_ACCESS_PROLOG(table) \
-	rlwinm  r3,r3,4,18,27;   \
-	lis     r5,table@h;      \
-	ori     r5,r5,table@l;   \
-	add     r3,r3,r5;        \
-	mtctr   r3;              \
-	bctr
-
-_GLOBAL(__mfdcr)
-	DCR_ACCESS_PROLOG(__mfdcr_table)
-
-_GLOBAL(__mtdcr)
-	DCR_ACCESS_PROLOG(__mtdcr_table)
-
-__mfdcr_table:
-	mfdcr  r3,0; blr
-__mtdcr_table:
-	mtdcr  0,r4; blr
-
-dcr     = 1
-        .rept   1023
-	mfdcr   r3,dcr; blr
-	mtdcr   dcr,r4; blr
-	dcr     = dcr + 1
-	.endr
Index: linux-malta-RH5/arch/powerpc/kernel/ppc_ksyms.c
===================================================================
--- linux-malta-RH5.orig/arch/powerpc/kernel/ppc_ksyms.c	2007-01-10 14:46:06.000000000 +1100
+++ linux-malta-RH5/arch/powerpc/kernel/ppc_ksyms.c	2007-01-10 15:21:59.000000000 +1100
@@ -225,7 +225,7 @@ EXPORT_SYMBOL(mmu_hash_lock); /* For MOL
 extern long *intercept_table;
 EXPORT_SYMBOL(intercept_table);
 #endif /* CONFIG_PPC_STD_MMU_32 */
-#if defined(CONFIG_40x) || defined(CONFIG_BOOKE)
+#ifdef CONFIG_PPC_DCR_NATIVE
 EXPORT_SYMBOL(__mtdcr);
 EXPORT_SYMBOL(__mfdcr);
 #endif
Index: linux-malta-RH5/include/asm-ppc/reg_booke.h
===================================================================
--- linux-malta-RH5.orig/include/asm-ppc/reg_booke.h	2007-01-10 14:46:47.000000000 +1100
+++ linux-malta-RH5/include/asm-ppc/reg_booke.h	2007-01-10 15:21:59.000000000 +1100
@@ -9,41 +9,9 @@
 #ifndef __ASM_PPC_REG_BOOKE_H__
 #define __ASM_PPC_REG_BOOKE_H__
 
-#ifndef __ASSEMBLY__
-/* Device Control Registers */
-void __mtdcr(int reg, unsigned int val);
-unsigned int __mfdcr(int reg);
-#define mfdcr(rn)						\
-	({unsigned int rval;					\
-	if (__builtin_constant_p(rn))				\
-		asm volatile("mfdcr %0," __stringify(rn)	\
-		              : "=r" (rval));			\
-	else							\
-		rval = __mfdcr(rn);				\
-	rval;})
-
-#define mtdcr(rn, v)						\
-do {								\
-	if (__builtin_constant_p(rn))				\
-		asm volatile("mtdcr " __stringify(rn) ",%0"	\
-			      : : "r" (v)); 			\
-	else							\
-		__mtdcr(rn, v);					\
-} while (0)
-
-/* R/W of indirect DCRs make use of standard naming conventions for DCRs */
-#define mfdcri(base, reg)			\
-({						\
-	mtdcr(base ## _CFGADDR, base ## _ ## reg);	\
-	mfdcr(base ## _CFGDATA);			\
-})
-
-#define mtdcri(base, reg, data)			\
-do {						\
-	mtdcr(base ## _CFGADDR, base ## _ ## reg);	\
-	mtdcr(base ## _CFGDATA, data);		\
-} while (0)
+#include <asm/dcr.h>
 
+#ifndef __ASSEMBLY__
 /* Performance Monitor Registers */
 #define mfpmr(rn)	({unsigned int rval; \
 			asm volatile("mfpmr %0," __stringify(rn) \

RHBZ#:
------
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=228099

Description:
------------
8 of 12 patches by Ben Herrenschmidt of IBM required for supporting IBM 
customers purchasing Cell blade configurations.  

This patch implements support for DCR based MPIC implementations.  Such 
implementations have the MPIC_USES_DCR flag set and don't use the 
phys_addr argument of mpic_alloc (they require a valid dcr mapping in the 
device mode).

RHEL Version Found:
-------------------
feature enhancement for 5.1.

Upstream Status:
----------------
This has been accepted into the kernel and can be viewed in the 2.6.20.3 
kernel.

Test Status:
------------
This patch was tested by Benjamin Herrenschmidt of IBM.  It has been
ack'd by Olof Johansson and signed off by Paul Mackerras.

Proposed Patch:
----------------
Please review and ACK for RHEL 5.1.
-

Index: linux-2.6/arch/powerpc/Kconfig
===================================================================
--- linux-2.6.orig/arch/powerpc/Kconfig	2007-01-22 15:49:46.000000000 +1100
+++ linux-2.6/arch/powerpc/Kconfig	2007-01-22 15:53:24.000000000 +1100
@@ -440,6 +440,7 @@
 	bool
 	select PPC_CELL
 	select PPC_DCR_MMIO
+	select MPIC
 	default n
 
 config PPC_IBM_CELL_BLADE
Index: linux-2.6/arch/powerpc/platforms/cell/setup.c
===================================================================
--- linux-2.6.orig/arch/powerpc/platforms/cell/setup.c	2007-01-22 14:38:56.000000000 +1100
+++ linux-2.6/arch/powerpc/platforms/cell/setup.c	2007-01-22 17:29:34.000000000 +1100
@@ -50,6 +50,7 @@
 #include <asm/spu.h>
 #include <asm/spu_priv1.h>
 #include <asm/udbg.h>
+#include <asm/mpic.h>
 
 #include "interrupt.h"
 #include "iommu.h"
@@ -88,10 +89,54 @@
 		pci_read_irq_line(dev);
 }
 
+static void cell_mpic_cascade(unsigned int irq, struct irq_desc *desc,
+			      struct pt_regs *regs)
+{
+	struct mpic *mpic = desc->handler_data;
+	unsigned int virq;
+
+	virq = mpic_get_one_irq(mpic, regs);
+	if (virq != NO_IRQ)
+		generic_handle_irq(virq, regs);
+	desc->chip->eoi(irq);
+}
+
+static void __init mpic_init_IRQ(void)
+{
+	struct device_node *dn;
+	struct mpic *mpic;
+	unsigned int virq;
+
+	for (dn = NULL;
+	     (dn = of_find_node_by_name(dn, "interrupt-controller"));) {
+		if (!device_is_compatible(dn, "CBEA,platform-open-pic"))
+			continue;
+
+		/* The MPIC driver will get everything it needs from the
+		 * device-tree, just pass 0 to all arguments
+		 */
+		mpic = mpic_alloc(dn, 0, 0, 0, 0, " MPIC     ");
+		if (mpic == NULL)
+			continue;
+		mpic_init(mpic);
+
+		virq = irq_of_parse_and_map(dn, 0);
+		if (virq == NO_IRQ)
+			continue;
+
+		printk(KERN_INFO "%s : hooking up to IRQ %d\n",
+		       dn->full_name, virq);
+		set_irq_data(virq, mpic);
+		set_irq_chained_handler(virq, cell_mpic_cascade);
+	}
+}
+
+
 static void __init cell_init_irq(void)
 {
 	iic_init_IRQ();
 	spider_init_IRQ();
+	mpic_init_IRQ();
 }
 
 static void __init cell_setup_arch(void)
Index: linux-2.6/arch/powerpc/sysdev/mpic.c
===================================================================
--- linux-2.6.orig/arch/powerpc/sysdev/mpic.c	2007-01-22 14:38:56.000000000 +1100
+++ linux-2.6/arch/powerpc/sysdev/mpic.c	2007-01-22 15:49:46.000000000 +1100
@@ -147,33 +147,51 @@
  */
 
 
-static inline u32 _mpic_read(unsigned int be, volatile u32 __iomem *base,
-			    unsigned int reg)
-{
-	if (be)
-		return in_be32(base + (reg >> 2));
-	else
-		return in_le32(base + (reg >> 2));
+static inline u32 _mpic_read(enum mpic_reg_type type,
+			     struct mpic_reg_bank *rb,
+			     unsigned int reg)
+{
+	switch(type) {
+#ifdef CONFIG_PPC_DCR
+	case mpic_access_dcr:
+		return dcr_read(rb->dhost,
+				rb->dbase + reg + rb->doff);
+#endif
+	case mpic_access_mmio_be:
+		return in_be32(rb->base + (reg >> 2));
+	case mpic_access_mmio_le:
+	default:
+		return in_le32(rb->base + (reg >> 2));
+	}
 }
 
-static inline void _mpic_write(unsigned int be, volatile u32 __iomem *base,
-			      unsigned int reg, u32 value)
+static inline void _mpic_write(enum mpic_reg_type type,
+			       struct mpic_reg_bank *rb,
+ 			       unsigned int reg, u32 value)
 {
-	if (be)
-		out_be32(base + (reg >> 2), value);
-	else
-		out_le32(base + (reg >> 2), value);
+	switch(type) {
+#ifdef CONFIG_PPC_DCR
+	case mpic_access_dcr:
+		return dcr_write(rb->dhost,
+				 rb->dbase + reg + rb->doff, value);
+#endif
+	case mpic_access_mmio_be:
+		return out_be32(rb->base + (reg >> 2), value);
+	case mpic_access_mmio_le:
+	default:
+		return out_le32(rb->base + (reg >> 2), value);
+	}
 }
 
 static inline u32 _mpic_ipi_read(struct mpic *mpic, unsigned int ipi)
 {
-	unsigned int be = (mpic->flags & MPIC_BIG_ENDIAN) != 0;
+	enum mpic_reg_type type = mpic->reg_type;
 	unsigned int offset = MPIC_INFO(GREG_IPI_VECTOR_PRI_0) +
 			      (ipi * MPIC_INFO(GREG_IPI_STRIDE));
 
-	if (mpic->flags & MPIC_BROKEN_IPI)
-		be = !be;
-	return _mpic_read(be, mpic->gregs, offset);
+	if ((mpic->flags & MPIC_BROKEN_IPI) && type == mpic_access_mmio_le)
+		type = mpic_access_mmio_be;
+	return _mpic_read(type, &mpic->gregs, offset);
 }
 
 static inline void _mpic_ipi_write(struct mpic *mpic, unsigned int ipi, u32 value)
@@ -181,7 +199,7 @@
 	unsigned int offset = MPIC_INFO(GREG_IPI_VECTOR_PRI_0) +
 			      (ipi * MPIC_INFO(GREG_IPI_STRIDE));
 
-	_mpic_write(mpic->flags & MPIC_BIG_ENDIAN, mpic->gregs, offset, value);
+	_mpic_write(mpic->reg_type, &mpic->gregs, offset, value);
 }
 
 static inline u32 _mpic_cpu_read(struct mpic *mpic, unsigned int reg)
@@ -190,8 +208,7 @@
 
 	if (mpic->flags & MPIC_PRIMARY)
 		cpu = hard_smp_processor_id();
-	return _mpic_read(mpic->flags & MPIC_BIG_ENDIAN,
-			  mpic->cpuregs[cpu], reg);
+	return _mpic_read(mpic->reg_type, &mpic->cpuregs[cpu], reg);
 }
 
 static inline void _mpic_cpu_write(struct mpic *mpic, unsigned int reg, u32 value)
@@ -201,7 +218,7 @@
 	if (mpic->flags & MPIC_PRIMARY)
 		cpu = hard_smp_processor_id();
 
-	_mpic_write(mpic->flags & MPIC_BIG_ENDIAN, mpic->cpuregs[cpu], reg, value);
+	_mpic_write(mpic->reg_type, &mpic->cpuregs[cpu], reg, value);
 }
 
 static inline u32 _mpic_irq_read(struct mpic *mpic, unsigned int src_no, unsigned int reg)
@@ -209,7 +226,7 @@
 	unsigned int	isu = src_no >> mpic->isu_shift;
 	unsigned int	idx = src_no & mpic->isu_mask;
 
-	return _mpic_read(mpic->flags & MPIC_BIG_ENDIAN, mpic->isus[isu],
+	return _mpic_read(mpic->reg_type, &mpic->isus[isu],
 			  reg + (idx * MPIC_INFO(IRQ_STRIDE)));
 }
 
@@ -219,12 +236,12 @@
 	unsigned int	isu = src_no >> mpic->isu_shift;
 	unsigned int	idx = src_no & mpic->isu_mask;
 
-	_mpic_write(mpic->flags & MPIC_BIG_ENDIAN, mpic->isus[isu],
+	_mpic_write(mpic->reg_type, &mpic->isus[isu],
 		    reg + (idx * MPIC_INFO(IRQ_STRIDE)), value);
 }
 
-#define mpic_read(b,r)		_mpic_read(mpic->flags & MPIC_BIG_ENDIAN,(b),(r))
-#define mpic_write(b,r,v)	_mpic_write(mpic->flags & MPIC_BIG_ENDIAN,(b),(r),(v))
+#define mpic_read(b,r)		_mpic_read(mpic->reg_type,&(b),(r))
+#define mpic_write(b,r,v)	_mpic_write(mpic->reg_type,&(b),(r),(v))
 #define mpic_ipi_read(i)	_mpic_ipi_read(mpic,(i))
 #define mpic_ipi_write(i,v)	_mpic_ipi_write(mpic,(i),(v))
 #define mpic_cpu_read(i)	_mpic_cpu_read(mpic,(i))
@@ -238,6 +255,38 @@
  */
 
 
+static void _mpic_map_mmio(struct mpic *mpic, unsigned long phys_addr,
+			   struct mpic_reg_bank *rb, unsigned int offset,
+			   unsigned int size)
+{
+	rb->base = ioremap(phys_addr + offset, size);
+	BUG_ON(rb->base == NULL);
+}
+
+#ifdef CONFIG_PPC_DCR
+static void _mpic_map_dcr(struct mpic *mpic, struct mpic_reg_bank *rb,
+			  unsigned int offset, unsigned int size)
+{
+	rb->dbase = mpic->dcr_base;
+	rb->doff = offset;
+	rb->dhost = dcr_map(mpic->of_node, rb->dbase + rb->doff, size);
+	BUG_ON(!DCR_MAP_OK(rb->dhost));
+}
+
+static inline void mpic_map(struct mpic *mpic, unsigned long phys_addr,
+			    struct mpic_reg_bank *rb, unsigned int offset,
+			    unsigned int size)
+{
+	if (mpic->flags & MPIC_USES_DCR)
+		_mpic_map_dcr(mpic, rb, offset, size);
+	else
+		_mpic_map_mmio(mpic, phys_addr, rb, offset, size);
+}
+#else /* CONFIG_PPC_DCR */
+#define mpic_map(m,p,b,o,s)	_mpic_map_mmio(m,p,b,o,s)
+#endif /* !CONFIG_PPC_DCR */
+
+
 
 /* Check if we have one of those nice broken MPICs with a flipped endian on
  * reads from IPI registers
@@ -845,7 +894,7 @@
  */
 
 struct mpic * __init mpic_alloc(struct device_node *node,
-				unsigned long phys_addr,
+				phys_addr_t phys_addr,
 				unsigned int flags,
 				unsigned int isu_size,
 				unsigned int irq_count,
@@ -855,6 +904,7 @@
 	u32		reg;
 	const char	*vers;
 	int		i;
+	u64		paddr = phys_addr;
 
 	mpic = alloc_bootmem(sizeof(struct mpic));
 	if (mpic == NULL)
@@ -883,6 +933,7 @@
 	if (flags & MPIC_PRIMARY)
 		mpic->hc_ht_irq.set_affinity = mpic_set_affinity;
 #endif /* CONFIG_MPIC_BROKEN_U3 */
+
 #ifdef CONFIG_SMP
 	mpic->hc_ipi = mpic_ipi_chip;
 	mpic->hc_ipi.typename = name;
@@ -893,15 +944,52 @@
 	mpic->irq_count = irq_count;
 	mpic->num_sources = 0; /* so far */
 
+	/* Check for "big-endian" in device-tree */
+	if (node && get_property(node, "big-endian", NULL) != NULL)
+		mpic->flags |= MPIC_BIG_ENDIAN;
+
+
 #ifdef CONFIG_MPIC_WEIRD
 	mpic->hw_set = mpic_infos[MPIC_GET_REGSET(flags)];
 #endif
 
+	/* default register type */
+	mpic->reg_type = (flags & MPIC_BIG_ENDIAN) ?
+		mpic_access_mmio_be : mpic_access_mmio_le;
+
+	/* If no physical address is passed in, a device-node is mandatory */
+	BUG_ON(paddr == 0 && node == NULL);
+
+	/* If no physical address passed in, check if it's dcr based */
+	if (paddr == 0 && get_property(node, "dcr-reg", NULL) != NULL)
+		mpic->flags |= MPIC_USES_DCR;
+
+#ifdef CONFIG_PPC_DCR
+	if (mpic->flags & MPIC_USES_DCR) {
+		u32 *dbasep;
+		dbasep = (u32 *)get_property(node, "dcr-reg", NULL);
+		BUG_ON(dbasep == NULL);
+		mpic->dcr_base = *dbasep;
+		mpic->reg_type = mpic_access_dcr;
+	}
+#else
+	BUG_ON (mpic->flags & MPIC_USES_DCR);
+#endif /* CONFIG_PPC_DCR */
+
+	/* If the MPIC is not DCR based, and no physical address was passed
+	 * in, try to obtain one
+	 */
+	if (paddr == 0 && !(mpic->flags & MPIC_USES_DCR)) {
+		u32 *reg;
+		reg = (u32 *)get_property(node, "reg", NULL);
+		BUG_ON(reg == NULL);
+		paddr = of_translate_address(node, reg);
+		BUG_ON(paddr == OF_BAD_ADDR);
+	}
+
 	/* Map the global registers */
-	mpic->gregs = ioremap(phys_addr + MPIC_INFO(GREG_BASE), 0x1000);
-	mpic->tmregs = mpic->gregs +
-		       ((MPIC_INFO(TIMER_BASE) - MPIC_INFO(GREG_BASE)) >> 2);
-	BUG_ON(mpic->gregs == NULL);
+	mpic_map(mpic, paddr, &mpic->gregs, MPIC_INFO(GREG_BASE), 0x1000);
+	mpic_map(mpic, paddr, &mpic->tmregs, MPIC_INFO(TIMER_BASE), 0x1000);
 
 	/* Reset */
 	if (flags & MPIC_WANTS_RESET) {
@@ -926,17 +1014,16 @@
 
 	/* Map the per-CPU registers */
 	for (i = 0; i < mpic->num_cpus; i++) {
-		mpic->cpuregs[i] = ioremap(phys_addr + MPIC_INFO(CPU_BASE) +
-					   i * MPIC_INFO(CPU_STRIDE), 0x1000);
-		BUG_ON(mpic->cpuregs[i] == NULL);
+		mpic_map(mpic, paddr, &mpic->cpuregs[i],
+			 MPIC_INFO(CPU_BASE) + i * MPIC_INFO(CPU_STRIDE),
+			 0x1000);
 	}
 
 	/* Initialize main ISU if none provided */
 	if (mpic->isu_size == 0) {
 		mpic->isu_size = mpic->num_sources;
-		mpic->isus[0] = ioremap(phys_addr + MPIC_INFO(IRQ_BASE),
-					MPIC_INFO(IRQ_STRIDE) * mpic->isu_size);
-		BUG_ON(mpic->isus[0] == NULL);
+		mpic_map(mpic, paddr, &mpic->isus[0],
+			 MPIC_INFO(IRQ_BASE), MPIC_INFO(IRQ_STRIDE) * mpic->isu_size);
 	}
 	mpic->isu_shift = 1 + __ilog2(mpic->isu_size - 1);
 	mpic->isu_mask = (1 << mpic->isu_shift) - 1;
@@ -956,10 +1043,11 @@
 		vers = "<unknown>";
 		break;
 	}
-	printk(KERN_INFO "mpic: Setting up MPIC \"%s\" version %s at %lx, max %d CPUs\n",
-	       name, vers, phys_addr, mpic->num_cpus);
-	printk(KERN_INFO "mpic: ISU size: %d, shift: %d, mask: %x\n", mpic->isu_size,
-	       mpic->isu_shift, mpic->isu_mask);
+	printk(KERN_INFO "mpic: Setting up MPIC \"%s\" version %s at %llx,"
+	       " max %d CPUs\n",
+	       name, vers, (unsigned long long)paddr, mpic->num_cpus);
+	printk(KERN_INFO "mpic: ISU size: %d, shift: %d, mask: %x\n",
+	       mpic->isu_size, mpic->isu_shift, mpic->isu_mask);
 
 	mpic->next = mpics;
 	mpics = mpic;
@@ -973,14 +1061,14 @@
 }
 
 void __init mpic_assign_isu(struct mpic *mpic, unsigned int isu_num,
-			    unsigned long phys_addr)
+			    phys_addr_t paddr)
 {
 	unsigned int isu_first = isu_num * mpic->isu_size;
 
 	BUG_ON(isu_num >= MPIC_MAX_ISU);
 
-	mpic->isus[isu_num] = ioremap(phys_addr,
-				      MPIC_INFO(IRQ_STRIDE) * mpic->isu_size);
+	mpic_map(mpic, paddr, &mpic->isus[isu_num], 0,
+		 MPIC_INFO(IRQ_STRIDE) * mpic->isu_size);
 	if ((isu_first + mpic->isu_size) > mpic->num_sources)
 		mpic->num_sources = isu_first + mpic->isu_size;
 }
Index: linux-2.6/include/asm-powerpc/mpic.h
===================================================================
--- linux-2.6.orig/include/asm-powerpc/mpic.h	2007-01-22 14:38:59.000000000 +1100
+++ linux-2.6/include/asm-powerpc/mpic.h	2007-01-22 15:49:46.000000000 +1100
@@ -3,6 +3,7 @@
 #ifdef __KERNEL__
 
 #include <linux/irq.h>
+#include <asm/dcr.h>
 
 /*
  * Global registers
@@ -225,6 +226,23 @@
 #endif /* CONFIG_MPIC_BROKEN_U3 */
 
 
+enum mpic_reg_type {
+	mpic_access_mmio_le,
+	mpic_access_mmio_be,
+#ifdef CONFIG_PPC_DCR
+	mpic_access_dcr
+#endif
+};
+
+struct mpic_reg_bank {
+	u32 __iomem	*base;
+#ifdef CONFIG_PPC_DCR
+	dcr_host_t	dhost;
+	unsigned int	dbase;
+	unsigned int	doff;
+#endif /* CONFIG_PPC_DCR */
+};
+
 /* The instance data of a given MPIC */
 struct mpic
 {
@@ -264,11 +282,18 @@
 	spinlock_t		fixup_lock;
 #endif
 
+	/* Register access method */
+	enum mpic_reg_type	reg_type;
+
 	/* The various ioremap'ed bases */
-	volatile u32 __iomem	*gregs;
-	volatile u32 __iomem	*tmregs;
-	volatile u32 __iomem	*cpuregs[MPIC_MAX_CPUS];
-	volatile u32 __iomem	*isus[MPIC_MAX_ISU];
+	struct mpic_reg_bank	gregs;
+	struct mpic_reg_bank	tmregs;
+	struct mpic_reg_bank	cpuregs[MPIC_MAX_CPUS];
+	struct mpic_reg_bank	isus[MPIC_MAX_ISU];
+
+#ifdef CONFIG_PPC_DCR
+	unsigned int		dcr_base;
+#endif
 
 #ifdef CONFIG_MPIC_WEIRD
 	/* Pointer to HW info array */
@@ -305,6 +330,8 @@
 #define MPIC_SPV_EOI			0x00000020
 /* No passthrough disable */
 #define MPIC_NO_PTHROU_DIS		0x00000040
+/* DCR based MPIC */
+#define MPIC_USES_DCR			0x00000080
 
 /* MPIC HW modification ID */
 #define MPIC_REGSET_MASK		0xf0000000
@@ -337,7 +364,7 @@
  * that is senses[0] correspond to linux irq "irq_offset".
  */
 extern struct mpic *mpic_alloc(struct device_node *node,
-			       unsigned long phys_addr,
+			       phys_addr_t phys_addr,
 			       unsigned int flags,
 			       unsigned int isu_size,
 			       unsigned int irq_count,
@@ -350,7 +377,7 @@
  * @phys_addr:	physical address of the ISU
  */
 extern void mpic_assign_isu(struct mpic *mpic, unsigned int isu_num,
-			    unsigned long phys_addr);
+			    phys_addr_t phys_addr);
 
 /* Set default sense codes
  *

RHBZ#:
------
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=228099

Description:
------------
9 of 12 patches by Ben Herrenschmidt of IBM required for supporting IBM 
customers purchasing Cell blade configurations. 

    [POWERPC] Souped-up of_platform_device support

This patch first splits of_device.c and of_platform.c, the later 
containing the bits relative to of_platform_device's. On the "breaks" side 
of things, drivers using of_platform_device(s) need to include 
asm/of_platform.h now and of_(un)register_driver is now 
of_(un)register_platform_driver.

In addition to a few utility functions to locate of_platform_device(s), 
the main new addition is of_platform_bus_probe() which allows the  
platform code to trigger an automatic creation of of_platform_devices for a 
whole tree of devices.

The function acts based on the type of the various "parent" devices 
encountered from a provided root, using either a default known list of bus 
types that can be "probed" or a passed-in list. It will only register 
devices on busses matching that list, which mean that typically, it will 
not register PCI devices, as expected (since they will be picked up by 
the PCI layer).

This will be used by Cell platforms using 4xx-type IOs in the Axon 
bridge and can be used by any embedded-type device as well.

RHEL Version Found:
-------------------
feature enhancement for 5.1.

Upstream Status:
----------------
This has been accepted into the kernel and can be viewed in the 2.6.20.3 
kernel.

Test Status:
------------
This patch was tested by Benjamin Herrenschmidt of IBM.  It has been
ack'd by Olof Johansson and signed off by Paul Mackerras.

Proposed Patch:
----------------
Please review and ACK for RHEL 5.1.
-

Index: linux-malta-RH5/arch/powerpc/kernel/Makefile
===================================================================
--- linux-malta-RH5.orig/arch/powerpc/kernel/Makefile	2007-01-10 16:31:15.000000000 +1100
+++ linux-malta-RH5/arch/powerpc/kernel/Makefile	2007-01-10 16:31:23.000000000 +1100
@@ -23,7 +23,7 @@ obj-$(CONFIG_PPC64)		+= setup_64.o binfm
 obj-$(CONFIG_PPC64)		+= vdso64/
 obj-$(CONFIG_ALTIVEC)		+= vecemu.o vector.o
 obj-$(CONFIG_PPC_970_NAP)	+= idle_power4.o
-obj-$(CONFIG_PPC_OF)		+= of_device.o prom_parse.o
+obj-$(CONFIG_PPC_OF)		+= of_device.o of_platform.o prom_parse.o
 procfs-$(CONFIG_PPC64)		:= proc_ppc64.o
 obj-$(CONFIG_PROC_FS)		+= $(procfs-y)
 rtaspci-$(CONFIG_PPC64)		:= rtas_pci.o
Index: linux-malta-RH5/arch/powerpc/kernel/dma_64.c
===================================================================
--- linux-malta-RH5.orig/arch/powerpc/kernel/dma_64.c	2007-01-10 16:32:24.000000000 +1100
+++ linux-malta-RH5/arch/powerpc/kernel/dma_64.c	2007-01-10 16:32:53.000000000 +1100
@@ -28,6 +28,9 @@ static struct dma_mapping_ops *get_dma_o
 	if (dev->bus == &ibmebus_bus_type)
 		return &ibmebus_dma_ops;
 #endif
+	if (dev->bus == &of_platform_bus_type)
+		return &of_platform_dma_ops;
+
 	return NULL;
 }
 
Index: linux-malta-RH5/arch/powerpc/kernel/of_device.c
===================================================================
--- linux-malta-RH5.orig/arch/powerpc/kernel/of_device.c	2007-01-10 16:12:39.000000000 +1100
+++ linux-malta-RH5/arch/powerpc/kernel/of_device.c	2007-01-10 16:41:31.000000000 +1100
@@ -9,30 +9,27 @@
 #include <asm/of_device.h>
 
 /**
- * of_match_device - Tell if an of_device structure has a matching
- * of_match structure
+ * of_match_node - Tell if an device_node has a matching of_match structure
  * @ids: array of of device match structures to search in
- * @dev: the of device structure to match against
+ * @node: the of device structure to match against
  *
- * Used by a driver to check whether an of_device present in the
- * system is in its list of supported devices.
+ * Low level utility function used by device matching.
  */
-const struct of_device_id *of_match_device(const struct of_device_id *matches,
-					const struct of_device *dev)
+const struct of_device_id *of_match_node(const struct of_device_id *matches,
+					 const struct device_node *node)
 {
-	if (!dev->node)
-		return NULL;
 	while (matches->name[0] || matches->type[0] || matches->compatible[0]) {
 		int match = 1;
 		if (matches->name[0])
-			match &= dev->node->name
-				&& !strcmp(matches->name, dev->node->name);
+			match &= node->name
+				&& !strcmp(matches->name, node->name);
 		if (matches->type[0])
-			match &= dev->node->type
-				&& !strcmp(matches->type, dev->node->type);
+			match &= node->type
+				&& !strcmp(matches->type, node->type);
 		if (matches->compatible[0])
-			match &= device_is_compatible(dev->node,
-				matches->compatible);
+			match &= device_is_compatible(
+					      (struct device_node *)node,
+					      (char *)matches->compatible);
 		if (match)
 			return matches;
 		matches++;
@@ -40,16 +37,21 @@ const struct of_device_id *of_match_devi
 	return NULL;
 }
 
-static int of_platform_bus_match(struct device *dev, struct device_driver *drv)
+/**
+ * of_match_device - Tell if an of_device structure has a matching
+ * of_match structure
+ * @ids: array of of device match structures to search in
+ * @dev: the of device structure to match against
+ *
+ * Used by a driver to check whether an of_device present in the
+ * system is in its list of supported devices.
+ */
+const struct of_device_id *of_match_device(const struct of_device_id *matches,
+					   const struct of_device *dev)
 {
-	struct of_device * of_dev = to_of_device(dev);
-	struct of_platform_driver * of_drv = to_of_platform_driver(drv);
-	const struct of_device_id * matches = of_drv->match_table;
-
-	if (!matches)
-		return 0;
-
-	return of_match_device(matches, of_dev) != NULL;
+	if (!dev->node)
+		return NULL;
+	return of_match_node(matches, dev->node);
 }
 
 struct of_device *of_dev_get(struct of_device *dev)
@@ -71,96 +73,8 @@ void of_dev_put(struct of_device *dev)
 		put_device(&dev->dev);
 }
 
-
-static int of_device_probe(struct device *dev)
-{
-	int error = -ENODEV;
-	struct of_platform_driver *drv;
-	struct of_device *of_dev;
-	const struct of_device_id *match;
-
-	drv = to_of_platform_driver(dev->driver);
-	of_dev = to_of_device(dev);
-
-	if (!drv->probe)
-		return error;
-
-	of_dev_get(of_dev);
-
-	match = of_match_device(drv->match_table, of_dev);
-	if (match)
-		error = drv->probe(of_dev, match);
-	if (error)
-		of_dev_put(of_dev);
-
-	return error;
-}
-
-static int of_device_remove(struct device *dev)
-{
-	struct of_device * of_dev = to_of_device(dev);
-	struct of_platform_driver * drv = to_of_platform_driver(dev->driver);
-
-	if (dev->driver && drv->remove)
-		drv->remove(of_dev);
-	return 0;
-}
-
-static int of_device_suspend(struct device *dev, pm_message_t state)
-{
-	struct of_device * of_dev = to_of_device(dev);
-	struct of_platform_driver * drv = to_of_platform_driver(dev->driver);
-	int error = 0;
-
-	if (dev->driver && drv->suspend)
-		error = drv->suspend(of_dev, state);
-	return error;
-}
-
-static int of_device_resume(struct device * dev)
-{
-	struct of_device * of_dev = to_of_device(dev);
-	struct of_platform_driver * drv = to_of_platform_driver(dev->driver);
-	int error = 0;
-
-	if (dev->driver && drv->resume)
-		error = drv->resume(of_dev);
-	return error;
-}
-
-struct bus_type of_platform_bus_type = {
-       .name	= "of_platform",
-       .match	= of_platform_bus_match,
-       .probe	= of_device_probe,
-       .remove	= of_device_remove,
-       .suspend	= of_device_suspend,
-       .resume	= of_device_resume,
-};
-
-static int __init of_bus_driver_init(void)
-{
-	return bus_register(&of_platform_bus_type);
-}
-
-postcore_initcall(of_bus_driver_init);
-
-int of_register_driver(struct of_platform_driver *drv)
-{
-	/* initialize common driver fields */
-	drv->driver.name = drv->name;
-	drv->driver.bus = &of_platform_bus_type;
-
-	/* register with core */
-	return driver_register(&drv->driver);
-}
-
-void of_unregister_driver(struct of_platform_driver *drv)
-{
-	driver_unregister(&drv->driver);
-}
-
-
-static ssize_t dev_show_devspec(struct device *dev, struct device_attribute *attr, char *buf)
+static ssize_t dev_show_devspec(struct device *dev,
+				struct device_attribute *attr, char *buf)
 {
 	struct of_device *ofdev;
 
@@ -232,41 +146,8 @@ void of_device_unregister(struct of_devi
 	device_unregister(&ofdev->dev);
 }
 
-struct of_device* of_platform_device_create(struct device_node *np,
-					    const char *bus_id,
-					    struct device *parent)
-{
-	struct of_device *dev;
-
-	dev = kmalloc(sizeof(*dev), GFP_KERNEL);
-	if (!dev)
-		return NULL;
-	memset(dev, 0, sizeof(*dev));
-
-	dev->node = of_node_get(np);
-	dev->dma_mask = 0xffffffffUL;
-	dev->dev.dma_mask = &dev->dma_mask;
-	dev->dev.parent = parent;
-	dev->dev.bus = &of_platform_bus_type;
-	dev->dev.release = of_release_dev;
-
-	strlcpy(dev->dev.bus_id, bus_id, BUS_ID_SIZE);
-
-	if (of_device_register(dev) != 0) {
-		kfree(dev);
-		return NULL;
-	}
-
-	return dev;
-}
-
 EXPORT_SYMBOL(of_match_device);
-EXPORT_SYMBOL(of_platform_bus_type);
-EXPORT_SYMBOL(of_register_driver);
-EXPORT_SYMBOL(of_unregister_driver);
-EXPORT_SYMBOL(of_device_register);
 EXPORT_SYMBOL(of_device_unregister);
 EXPORT_SYMBOL(of_dev_get);
 EXPORT_SYMBOL(of_dev_put);
-EXPORT_SYMBOL(of_platform_device_create);
 EXPORT_SYMBOL(of_release_dev);
Index: linux-malta-RH5/arch/powerpc/kernel/of_platform.c
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ linux-malta-RH5/arch/powerpc/kernel/of_platform.c	2007-01-10 16:40:33.000000000 +1100
@@ -0,0 +1,391 @@
+/*
+ *    Copyright (C) 2006 Benjamin Herrenschmidt, IBM Corp.
+ *			 <benh@kernel.crashing.org>
+ *    and		 Arnd Bergmann, IBM Corp.
+ *
+ *  This program is free software; you can redistribute it and/or
+ *  modify it under the terms of the GNU General Public License
+ *  as published by the Free Software Foundation; either version
+ *  2 of the License, or (at your option) any later version.
+ *
+ */
+
+#undef DEBUG
+
+#include <linux/string.h>
+#include <linux/kernel.h>
+#include <linux/init.h>
+#include <linux/module.h>
+#include <linux/mod_devicetable.h>
+#include <linux/slab.h>
+#include <linux/pci.h>
+
+#include <asm/errno.h>
+#include <asm/dcr.h>
+#include <asm/of_device.h>
+#include <asm/topology.h>
+#include <asm/pci-bridge.h>
+#include <asm/ppc-pci.h>
+#include <asm/atomic.h>
+
+
+/*
+ * The list of OF IDs below is used for matching bus types in the
+ * system whose devices are to be exposed as of_platform_devices.
+ *
+ * This is the default list valid for most platforms. This file provides
+ * functions who can take an explicit list if necessary though
+ *
+ * The search is always performed recursively looking for children of
+ * the provided device_node and recursively if such a children matches
+ * a bus type in the list
+ */
+
+static struct of_device_id of_default_bus_ids[] = {
+	{ .type = "soc", },
+	{ .compatible = "soc", },
+	{ .type = "spider", },
+	{ .type = "axon", },
+	{ .type = "plb5", },
+	{ .type = "plb4", },
+	{ .type = "opb", },
+	{ .type = "ebc", },
+	{},
+};
+
+static atomic_t bus_no_reg_magic;
+
+/*
+ *
+ * OF platform device type definition & base infrastructure
+ *
+ */
+
+static int of_platform_bus_match(struct device *dev, struct device_driver *drv)
+{
+	struct of_device * of_dev = to_of_device(dev);
+	struct of_platform_driver * of_drv = to_of_platform_driver(drv);
+	const struct of_device_id * matches = of_drv->match_table;
+
+	if (!matches)
+		return 0;
+
+	return of_match_device(matches, of_dev) != NULL;
+}
+
+static int of_platform_device_probe(struct device *dev)
+{
+	int error = -ENODEV;
+	struct of_platform_driver *drv;
+	struct of_device *of_dev;
+	const struct of_device_id *match;
+
+	drv = to_of_platform_driver(dev->driver);
+	of_dev = to_of_device(dev);
+
+	if (!drv->probe)
+		return error;
+
+	of_dev_get(of_dev);
+
+	match = of_match_device(drv->match_table, of_dev);
+	if (match)
+		error = drv->probe(of_dev, match);
+	if (error)
+		of_dev_put(of_dev);
+
+	return error;
+}
+
+static int of_platform_device_remove(struct device *dev)
+{
+	struct of_device * of_dev = to_of_device(dev);
+	struct of_platform_driver * drv = to_of_platform_driver(dev->driver);
+
+	if (dev->driver && drv->remove)
+		drv->remove(of_dev);
+	return 0;
+}
+
+static int of_platform_device_suspend(struct device *dev, pm_message_t state)
+{
+	struct of_device * of_dev = to_of_device(dev);
+	struct of_platform_driver * drv = to_of_platform_driver(dev->driver);
+	int error = 0;
+
+	if (dev->driver && drv->suspend)
+		error = drv->suspend(of_dev, state);
+	return error;
+}
+
+static int of_platform_device_resume(struct device * dev)
+{
+	struct of_device * of_dev = to_of_device(dev);
+	struct of_platform_driver * drv = to_of_platform_driver(dev->driver);
+	int error = 0;
+
+	if (dev->driver && drv->resume)
+		error = drv->resume(of_dev);
+	return error;
+}
+
+struct bus_type of_platform_bus_type = {
+       .name	= "of_platform",
+       .match	= of_platform_bus_match,
+       .probe	= of_platform_device_probe,
+       .remove	= of_platform_device_remove,
+       .suspend	= of_platform_device_suspend,
+       .resume	= of_platform_device_resume,
+};
+EXPORT_SYMBOL(of_platform_bus_type);
+
+static int __init of_bus_driver_init(void)
+{
+	return bus_register(&of_platform_bus_type);
+}
+
+postcore_initcall(of_bus_driver_init);
+
+int of_register_platform_driver(struct of_platform_driver *drv)
+{
+	/* initialize common driver fields */
+	drv->driver.name = drv->name;
+	drv->driver.bus = &of_platform_bus_type;
+
+	/* register with core */
+	return driver_register(&drv->driver);
+}
+EXPORT_SYMBOL(of_register_platform_driver);
+
+void of_unregister_platform_driver(struct of_platform_driver *drv)
+{
+	driver_unregister(&drv->driver);
+}
+EXPORT_SYMBOL(of_unregister_platform_driver);
+
+static void of_platform_make_bus_id(struct of_device *dev)
+{
+	struct device_node *node = dev->node;
+	char *name = dev->dev.bus_id;
+	u32 *reg;
+	u64 addr;
+	int magic;
+
+	/*
+	 * If it's a DCR based device, use 'd' for native DCRs
+	 * and 'D' for MMIO DCRs.
+	 */
+#ifdef CONFIG_PPC_DCR
+	reg = (u32 *)get_property(node, "dcr-reg", NULL);
+	if (reg) {
+#ifdef CONFIG_PPC_DCR_NATIVE
+		snprintf(name, BUS_ID_SIZE, "d%x.%s",
+			 *reg, node->name);
+#else /* CONFIG_PPC_DCR_NATIVE */
+		addr = of_translate_dcr_address(node, *reg, NULL);
+		if (addr != OF_BAD_ADDR) {
+			snprintf(name, BUS_ID_SIZE,
+				 "D%llx.%s", (unsigned long long)addr,
+				 node->name);
+			return;
+		}
+#endif /* !CONFIG_PPC_DCR_NATIVE */
+	}
+#endif /* CONFIG_PPC_DCR */
+
+	/*
+	 * For MMIO, get the physical address
+	 */
+	reg = (u32 *)get_property(node, "reg", NULL);
+	if (reg) {
+		addr = of_translate_address(node, reg);
+		if (addr != OF_BAD_ADDR) {
+			snprintf(name, BUS_ID_SIZE,
+				 "%llx.%s", (unsigned long long)addr,
+				 node->name);
+			return;
+		}
+	}
+
+	/*
+	 * No BusID, use the node name and add a globally incremented
+	 * counter (and pray...)
+	 */
+	magic = atomic_add_return(1, &bus_no_reg_magic);
+	snprintf(name, BUS_ID_SIZE, "%s.%d", node->name, magic - 1);
+}
+
+struct of_device* of_platform_device_create(struct device_node *np,
+					    const char *bus_id,
+					    struct device *parent)
+{
+	struct of_device *dev;
+
+	dev = kmalloc(sizeof(*dev), GFP_KERNEL);
+	if (!dev)
+		return NULL;
+	memset(dev, 0, sizeof(*dev));
+
+	dev->node = of_node_get(np);
+	dev->dma_mask = 0xffffffffUL;
+	dev->dev.dma_mask = &dev->dma_mask;
+	dev->dev.parent = parent;
+	dev->dev.bus = &of_platform_bus_type;
+	dev->dev.release = of_release_dev;
+
+	/* We do not fill the DMA ops for platform devices by default.
+	 * This is currently the responsibility of the platform code
+	 * to do such, possibly using a device notifier
+	 */
+
+	if (bus_id)
+		strlcpy(dev->dev.bus_id, bus_id, BUS_ID_SIZE);
+	else
+		of_platform_make_bus_id(dev);
+
+	if (of_device_register(dev) != 0) {
+		kfree(dev);
+		return NULL;
+	}
+
+	return dev;
+}
+EXPORT_SYMBOL(of_platform_device_create);
+
+
+
+/**
+ * of_platform_bus_create - Create an OF device for a bus node and all its
+ * children. Optionally recursively instanciate matching busses.
+ * @bus: device node of the bus to instanciate
+ * @matches: match table, NULL to use the default, OF_NO_DEEP_PROBE to
+ * disallow recursive creation of child busses
+ */
+static int of_platform_bus_create(struct device_node *bus,
+				  struct of_device_id *matches,
+				  struct device *parent)
+{
+	struct device_node *child;
+	struct of_device *dev;
+	int rc = 0;
+
+	for (child = NULL; (child = of_get_next_child(bus, child)); ) {
+		pr_debug("   create child: %s\n", child->full_name);
+		dev = of_platform_device_create(child, NULL, parent);
+		if (dev == NULL)
+			rc = -ENOMEM;
+		else if (!of_match_node(matches, child))
+			continue;
+		if (rc == 0) {
+			pr_debug("   and sub busses\n");
+			rc = of_platform_bus_create(child, matches, &dev->dev);
+		} if (rc) {
+			of_node_put(child);
+			break;
+		}
+	}
+	return rc;
+}
+
+/**
+ * of_platform_bus_probe - Probe the device-tree for platform busses
+ * @root: parent of the first level to probe or NULL for the root of the tree
+ * @matches: match table, NULL to use the default
+ * @parent: parent to hook devices from, NULL for toplevel
+ *
+ * Note that children of the provided root are not instanciated as devices
+ * unless the specified root itself matches the bus list and is not NULL.
+ */
+
+int of_platform_bus_probe(struct device_node *root,
+			  struct of_device_id *matches,
+			  struct device *parent)
+{
+	struct device_node *child;
+	struct of_device *dev;
+	int rc = 0;
+
+	if (matches == NULL)
+		matches = of_default_bus_ids;
+	if (matches == OF_NO_DEEP_PROBE)
+		return -EINVAL;
+	if (root == NULL)
+		root = of_find_node_by_path("/");
+	else
+		of_node_get(root);
+
+	pr_debug("of_platform_bus_probe()\n");
+	pr_debug(" starting at: %s\n", root->full_name);
+
+	/* Do a self check of bus type, if there's a match, create
+	 * children
+	 */
+	if (of_match_node(matches, root)) {
+		pr_debug(" root match, create all sub devices\n");
+		dev = of_platform_device_create(root, NULL, parent);
+		if (dev == NULL) {
+			rc = -ENOMEM;
+			goto bail;
+		}
+		pr_debug(" create all sub busses\n");
+		rc = of_platform_bus_create(root, matches, &dev->dev);
+		goto bail;
+	}
+	for (child = NULL; (child = of_get_next_child(root, child)); ) {
+		if (!of_match_node(matches, child))
+			continue;
+
+		pr_debug("  match: %s\n", child->full_name);
+		dev = of_platform_device_create(child, NULL, parent);
+		if (dev == NULL)
+			rc = -ENOMEM;
+		else
+			rc = of_platform_bus_create(child, matches, &dev->dev);
+		if (rc) {
+			of_node_put(child);
+			break;
+		}
+	}
+ bail:
+	of_node_put(root);
+	return rc;
+}
+EXPORT_SYMBOL(of_platform_bus_probe);
+
+static int of_dev_node_match(struct device *dev, void *data)
+{
+	return to_of_device(dev)->node == data;
+}
+
+struct of_device *of_find_device_by_node(struct device_node *np)
+{
+	struct device *dev;
+
+	dev = bus_find_device(&of_platform_bus_type,
+			      NULL, np, of_dev_node_match);
+	if (dev)
+		return to_of_device(dev);
+	return NULL;
+}
+EXPORT_SYMBOL(of_find_device_by_node);
+
+static int of_dev_phandle_match(struct device *dev, void *data)
+{
+	phandle *ph = data;
+	return to_of_device(dev)->node->linux_phandle == *ph;
+}
+
+struct of_device *of_find_device_by_phandle(phandle ph)
+{
+	struct device *dev;
+
+	dev = bus_find_device(&of_platform_bus_type,
+			      NULL, &ph, of_dev_phandle_match);
+	if (dev)
+		return to_of_device(dev);
+	return NULL;
+}
+EXPORT_SYMBOL(of_find_device_by_phandle);
+
+struct dma_mapping_ops of_platform_dma_ops;
+EXPORT_SYMBOL(of_platform_dma_ops);
Index: linux-malta-RH5/include/asm-powerpc/of_device.h
===================================================================
--- linux-malta-RH5.orig/include/asm-powerpc/of_device.h	2007-01-10 16:12:53.000000000 +1100
+++ linux-malta-RH5/include/asm-powerpc/of_device.h	2007-01-10 16:38:35.000000000 +1100
@@ -4,6 +4,7 @@
 
 #include <linux/device.h>
 #include <linux/mod_devicetable.h>
+#include <linux/dma-mapping.h>
 #include <asm/prom.h>
 
 /*
@@ -12,6 +13,7 @@
  * mechanism
  */
 extern struct bus_type of_platform_bus_type;
+extern struct dma_mapping_ops of_platform_dma_ops;
 
 /*
  * The of_device is a kind of "base class" that is a superset of
@@ -26,6 +28,8 @@ struct of_device
 };
 #define	to_of_device(d) container_of(d, struct of_device, dev)
 
+extern const struct of_device_id *of_match_node(
+	const struct of_device_id *matches, const struct device_node *node);
 extern const struct of_device_id *of_match_device(
 	const struct of_device_id *matches, const struct of_device *dev);
 
@@ -53,8 +57,10 @@ struct of_platform_driver
 };
 #define	to_of_platform_driver(drv) container_of(drv,struct of_platform_driver, driver)
 
-extern int of_register_driver(struct of_platform_driver *drv);
-extern void of_unregister_driver(struct of_platform_driver *drv);
+extern int of_register_platform_driver(struct of_platform_driver *drv);
+#define of_register_driver of_register_platform_driver
+extern void of_unregister_platform_driver(struct of_platform_driver *drv);
+#define of_unregister_driver of_unregister_platform_driver
 extern int of_device_register(struct of_device *ofdev);
 extern void of_device_unregister(struct of_device *ofdev);
 extern struct of_device *of_platform_device_create(struct device_node *np,
@@ -62,5 +68,15 @@ extern struct of_device *of_platform_dev
 						   struct device *parent);
 extern void of_release_dev(struct device *dev);
 
+/* pseudo "matches" value to not do deep probe */
+#define OF_NO_DEEP_PROBE ((struct of_device_id *)-1)
+
+extern int of_platform_bus_probe(struct device_node *root,
+				 struct of_device_id *matches,
+				 struct device *parent);
+
+extern struct of_device *of_find_device_by_node(struct device_node *np);
+extern struct of_device *of_find_device_by_phandle(phandle ph);
+
 #endif /* __KERNEL__ */
 #endif /* _ASM_POWERPC_OF_DEVICE_H */

RHBZ#:
------
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=228099

Description:
------------
10 of 12 patches by Ben Herrenschmidt of IBM required for supporting IBM 
customers purchasing Cell blade configurations. 

This patch makes the Cell DMA code work on both the Spider and the Axon 
south bridges by turning cell_dma_valid into a variable instead of a 
constant.  This will be updated when we have full iommu support.

RHEL Version Found:
-------------------
feature enhancement for 5.1.

Upstream Status:
----------------
This has been accepted into the kernel and can be viewed in the 2.6.20.3 
kernel.

Test Status:
------------
This patch was tested by Benjamin Herrenschmidt of IBM.  It has been
ack'd by Olof Johansson and signed off by Paul Mackerras.

Proposed Patch:
----------------
Please review and ACK for RHEL 5.1.
-

Index: linux-malta-RH5/arch/powerpc/platforms/cell/iommu.c
===================================================================
--- linux-malta-RH5.orig/arch/powerpc/platforms/cell/iommu.c	2007-01-10 14:46:07.000000000 +1100
+++ linux-malta-RH5/arch/powerpc/platforms/cell/iommu.c	2007-01-11 17:04:07.000000000 +1100
@@ -46,6 +46,8 @@
 
 #include "iommu.h"
 
+static dma_addr_t cell_dma_valid = SPIDER_DMA_VALID;
+
 static inline unsigned long 
 get_iopt_entry(unsigned long real_address, unsigned long ioid,
 			 unsigned long prot)
@@ -427,7 +429,7 @@ static void *cell_alloc_coherent(struct 
 	ret = (void *)__get_free_pages(flag, get_order(size));
 	if (ret != NULL) {
 		memset(ret, 0, size);
-		*dma_handle = virt_to_abs(ret) | CELL_DMA_VALID;
+		*dma_handle = virt_to_abs(ret) | cell_dma_valid;
 	}
 	return ret;
 }
@@ -441,7 +443,7 @@ static void cell_free_coherent(struct de
 static dma_addr_t cell_map_single(struct device *hwdev, void *ptr,
 		size_t size, enum dma_data_direction direction)
 {
-	return virt_to_abs(ptr) | CELL_DMA_VALID;
+	return virt_to_abs(ptr) | cell_dma_valid;
 }
 
 static void cell_unmap_single(struct device *hwdev, dma_addr_t dma_addr,
@@ -456,7 +458,7 @@ static int cell_map_sg(struct device *hw
 
 	for (i = 0; i < nents; i++, sg++) {
 		sg->dma_address = (page_to_phys(sg->page) + sg->offset)
-					| CELL_DMA_VALID;
+					| cell_dma_valid;
 		sg->dma_length = sg->length;
 	}
 
@@ -487,7 +489,16 @@ void cell_init_iommu(void)
 {
 	int setup_bus = 0;
 
-	if (of_find_node_by_path("/mambo")) {
+	/* If we have an Axon bridge, clear the DMA valid mask and don't
+	 * try to mess around with the iommu (keep it off). This will work
+	 * for malta limited amount of RAM on a RHEL kernel. Upstream is
+	 * smarter but we don't need to backport that now.
+	 */
+	if (of_find_node_by_name(NULL, "axon")) {
+		cell_dma_valid = 0;
+		ppc_md.iommu_dev_setup = iommu_dev_setup_null;
+		ppc_md.iommu_bus_setup = iommu_bus_setup_null;
+	} else if (of_find_node_by_path("/mambo")) {
 		pr_info("Not using iommu on systemsim\n");
 	} else {
 
@@ -509,4 +520,5 @@ void cell_init_iommu(void)
 	}
 
 	pci_dma_ops = cell_iommu_ops;
+	of_platform_dma_ops = cell_iommu_ops;
 }
Index: linux-malta-RH5/arch/powerpc/platforms/cell/iommu.h
===================================================================
--- linux-malta-RH5.orig/arch/powerpc/platforms/cell/iommu.h	2007-01-10 14:46:07.000000000 +1100
+++ linux-malta-RH5/arch/powerpc/platforms/cell/iommu.h	2007-01-10 16:55:47.000000000 +1100
@@ -53,9 +53,11 @@ enum {
 	IOC_ST_ORIGIN     = 0x918,
 	IOC_CONF	  = 0x930,
 
-	/* The high bit needs to be set on every DMA address,
-	   only 2GB are addressable */
-	CELL_DMA_VALID	  = 0x80000000,
+	/* The high bit needs to be set on every DMA address when using
+	 * a spider bridge and only 2GB are addressable with the current
+	 * iommu code.
+	 */
+	SPIDER_DMA_VALID  = 0x80000000,
 	CELL_DMA_MASK	  = 0x7fffffff,
 };
 

RHBZ#:
------
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=228099

Description:
------------
11 of 12 patches by Ben Herrenschmidt of IBM required for supporting IBM
customers purchasing Cell blade configurations.

Hook up of_platform_bus_probe with the cell platform in order to publish 
the non-PCI devices in the device-tree of cell blades as 
of_platform_device(s).

RHEL Version Found:
-------------------
feature enhancement for 5.1.

Upstream Status:
----------------
This has been accepted into the kernel and can be viewed in the 2.6.20.3 
kernel.

Test Status:
------------
This patch was tested by Benjamin Herrenschmidt of IBM.  It has been
ack'd by Olof Johansson and signed off by Paul Mackerras.

Proposed Patch:
----------------
Please review and ACK for RHEL 5.1.
-

Index: linux-malta-RH5/arch/powerpc/platforms/cell/setup.c
===================================================================
--- linux-malta-RH5.orig/arch/powerpc/platforms/cell/setup.c	2007-01-11 17:05:40.000000000 +1100
+++ linux-malta-RH5/arch/powerpc/platforms/cell/setup.c	2007-01-11 17:06:05.000000000 +1100
@@ -51,6 +51,7 @@
 #include <asm/spu_priv1.h>
 #include <asm/udbg.h>
 #include <asm/mpic.h>
+#include <asm/of_device.h>
 
 #include "interrupt.h"
 #include "iommu.h"
@@ -89,6 +90,14 @@ static void __init cell_pcibios_fixup(vo
 		pci_read_irq_line(dev);
 }
 
+static int __init cell_publish_devices(void)
+{
+	if (machine_is(cell))
+		of_platform_bus_probe(NULL, NULL, NULL);
+	return 0;
+}
+device_initcall(cell_publish_devices);
+
 static void cell_mpic_cascade(unsigned int irq, struct irq_desc *desc,
 			      struct pt_regs *regs)
 {

The following should have been attached to this patch:
Original patch was:

commit 4c9d2800be5dfabf26acdeb401cbabe9edc1dcf2
Author: Benjamin Herrenschmidt <benh@kernel.crashing.org>

    [POWERPC] Generic OF platform driver for PCI host bridges.

    When enabled in Kconfig, it will pick up any of_platform_device
    matching it's match list (currently type "pci", "pcix", "pcie",
    or "ht" and setup a PHB for it.

    Platform must provide a ppc_md.pci_setup_phb() for it to work
    (for doing the necessary initialisations specific to a given PHB
    like setting up the config space ops).

    It's currently only available on 64 bits as the 32 bits PCI code
    can't quite cope with it in it's current form. I will fix that
    later.

    Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
    Signed-off-by: Paul Mackerras <paulus@samba.org>

This instead does a cell-only solution with the same code, but less
invasive, to reduce the risk of breaking other platforms.


Index: linux-malta-RH5/arch/powerpc/kernel/pci_64.c
===================================================================
--- linux-malta-RH5.orig/arch/powerpc/kernel/pci_64.c	2007-01-11
18:06:36.000000000 +1100
+++ linux-malta-RH5/arch/powerpc/kernel/pci_64.c	2007-01-11
18:07:09.000000000 +1100
@@ -1003,10 +1003,7 @@ void __devinit pci_process_bridge_OF_ran
 		pci_space = ranges[0];
 		pci_addr = ((unsigned long)ranges[1] << 32) | ranges[2];
 
-		cpu_phys_addr = ranges[3];
-		if (na >= 2)
-			cpu_phys_addr = (cpu_phys_addr << 32) | ranges[4];
-
+		cpu_phys_addr = of_translate_address(dev, &ranges[3]);
 		size = ((unsigned long)ranges[na+3] << 32) | ranges[na+4];
 		ranges += np;
 		if (size == 0)
Index: linux-malta-RH5/arch/powerpc/platforms/cell/setup.c
===================================================================
--- linux-malta-RH5.orig/arch/powerpc/platforms/cell/setup.c	2007-01-11
17:12:42.000000000 +1100
+++ linux-malta-RH5/arch/powerpc/platforms/cell/setup.c	2007-01-11
18:08:34.000000000 +1100
@@ -110,6 +110,61 @@ static void cell_mpic_cascade(unsigned i
 	desc->chip->eoi(irq);
 }
 
+static void __init cell_add_phb(struct device_node *node, int index)
+{
+	struct pci_controller *phb;
+	struct resource r;
+
+	phb = pcibios_alloc_controller(node);
+	if (!phb)
+		return;
+	setup_phb(node, phb);
+	if (of_address_to_resource(node, 0, &r) == 0)
+		phb->buid = r.start;
+	pci_process_bridge_OF_ranges(phb, node, 0);
+	pci_setup_phb_io(phb, index == 0);
+}
+
+static void __init cell_find_and_init_phbs(void)
+{
+	struct device_node *axon, *plb5, *plb4, *np;
+	int index = 0;
+
+	/* Old blades, use generic code */
+	axon = of_find_node_by_name(NULL, "axon");
+	if (axon == NULL) {
+		find_and_init_phbs();
+		return;
+	}
+
+	/* New blades, manually instanciate bridges for now as
+	 * RHEL5 doesn't have the infrastructure to do it from
+	 * of_platform
+	 */
+	for (; axon; axon = of_find_node_by_name(axon, "axon")) {
+		for (plb5 = NULL; !!(plb5 = of_get_next_child(axon, plb5));)
+			if (strcmp(plb5->name, "plb5") == 0)
+				break;
+		if (plb5 == NULL)
+			continue;
+		for (np = NULL; !!(np = of_get_next_child(plb5, np));)
+			if (strcmp(np->name, "pcie") == 0)
+				cell_add_phb(np, index++);
+		for (plb4 = NULL; !!(plb4 = of_get_next_child(plb5, plb4));)
+			if (strcmp(plb4->name, "plb4") == 0)
+				break;
+		of_node_put(plb5);
+		if (plb4 == NULL)
+			continue;
+		for (np = NULL; !!(np = of_get_next_child(plb4, np));)
+			if (strcmp(np->name, "pcix") == 0)
+				cell_add_phb(np, index++);
+		of_node_put(plb4);
+	}
+
+	pci_devs_phb_init();
+}
+
 static void __init mpic_init_IRQ(void)
 {
 	struct device_node *dn;
@@ -174,7 +229,7 @@ static void __init cell_setup_arch(void)
 
 	/* Find and initialize PCI host bridges */
 	init_pci_config_tokens();
-	find_and_init_phbs();
+	cell_find_and_init_phbs();
 	cbe_pervasive_init();
 #ifdef CONFIG_DUMMY_CONSOLE
 	conswitchp = &dummy_con;