Skip to contentSkip to conversation

Assume we're on cpu 0 in early boot

PATCH In review Michael Ellerman 3 participants 6 messages Jun 23, 2006 – Jun 28, 2006 PowerPC

lore
Share & tools
lore
#0 lore raw

PATCH 2/8 Assume we're on cpu 0 in early boot

There's a small period early in boot where we don't know which cpu we're running on. That's ok, except that it means we have no paca, or more correctly that our paca pointer points somewhere random.

So that we can safely call things like smp_processor_id(), we need a paca, so just assume we're on cpu 0. No code should _write_ to the paca before we've set the correct one up.

---

 arch/powerpc/kernel/head_64.S |   10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)
arch/powerpc/kernel/head_64.S +7 -3
Index: to-merge/arch/powerpc/kernel/head_64.S
===================================================================
@@ -1583,9 +1583,6 @@ _GLOBAL(__start_initialization_multiplat
 	/* Setup some critical 970 SPRs before switching MMU off */
 	bl	.__970_cpu_preinit
 
-	/* cpu # */
-	li	r24,0
-
 	/* Switch off MMU if not already */
 	LOAD_REG_IMMEDIATE(r4, .__after_prom_start - KERNELBASE)
 	add	r4,r4,r30
@@ -1908,6 +1905,13 @@ _STATIC(start_here_multiplatform)
 	bl	.__save_cpu_setup
 	sync
 
+	/* Assume we're on cpu for now, we don't actually know yet.
+	 * The early setup code should not write to any paca fields until
+	 * after we've setup the correct paca. See early_setup() */
+	li	r24,0
+	li	r3,0
+	bl	.setup_paca
+
 	/* Do very early kernel initializations, including initial hash table,
 	 * stab and slb setup before we turn on relocation.	*/
 
#1 lore raw

arch/powerpc/kernel/head_64.S _STATIC(start_here_multiplatform) · line 1908

Quoted 1 line · Michael Ellerman
arch/powerpc/kernel/head_64.S +1 -0
+	/* Assume we're on cpu for now, we don't actually know yet.

oops :( "Assume we're on cpu _0_ for now"
-JX

#2 lore raw

PATCH 1/1 Assume we're on cpu 0 in early boot

There's a small period early in boot where we don't know which cpu we're running on. That's ok, except that it means we have no paca, or more correctly that our paca pointer points somewhere random.

So that we can safely call things like smp_processor_id(), we need a paca, so just assume we're on cpu 0. No code should _write_ to the paca before we've set the correct one up.

---

 arch/powerpc/kernel/head_64.S |   10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)
arch/powerpc/kernel/head_64.S +7 -3
Index: to-merge/arch/powerpc/kernel/head_64.S
===================================================================
@@ -1583,9 +1583,6 @@ _GLOBAL(__start_initialization_multiplat
 	/* Setup some critical 970 SPRs before switching MMU off */
 	bl	.__970_cpu_preinit
 
-	/* cpu # */
-	li	r24,0
-
 	/* Switch off MMU if not already */
 	LOAD_REG_IMMEDIATE(r4, .__after_prom_start - KERNELBASE)
 	add	r4,r4,r30
@@ -1908,6 +1905,13 @@ _STATIC(start_here_multiplatform)
 	bl	.__save_cpu_setup
 	sync
 
+	/* Assume we're on cpu 0 for now, we don't actually know yet.
+	 * The early setup code should not write to any paca fields until
+	 * after we've setup the correct paca. See early_setup() */
+	li	r24,0
+	li	r3,0
+	bl	.setup_paca
+
 	/* Do very early kernel initializations, including initial hash table,
 	 * stab and slb setup before we turn on relocation.	*/
 
#3 lore raw
Quoted 7 lines · Michael Ellerman

There's a small period early in boot where we don't know which cpu we're running on. That's ok, except that it means we have no paca, or more correctly that our paca pointer points somewhere random.

So that we can safely call things like smp_processor_id(), we need a paca, so just assume we're on cpu 0. No code should _write_ to the paca before we've set the correct one up.

OK, but why not just do a setup_paca(0) at the start of early_setup(), in C code? You can also remove the redundant paca setting in start_here_common() in head_64.S.

Paul.

#4 lore raw
Quoted 6 lines · Paul Mackerras
Quoted 4 lines · Michael Ellerman

There's a small period early in boot where we don't know which cpu we're running on. That's ok, except that it means we have no paca, or more correctly that our paca pointer points somewhere random.

Quoted 7 lines
Quoted 3 lines

So that we can safely call things like smp_processor_id(), we need a paca, so just assume we're on cpu 0. No code should _write_ to the paca before we've set the correct one up.

OK, but why not just do a setup_paca(0) at the start of early_setup(), in C code? You can also remove the redundant paca setting in start_here_common() in head_64.S.

Yep, duh. New patch RSN.

cheers

Signature · 8 lines
Michael Ellerman
IBM OzLabs

wwweb: http://michael.ellerman.id.au
phone: +61 2 6212 1183 (tie line 70 21183)

We do not inherit the earth from our ancestors,
we borrow it from our children. - S.M.A.R.T Person
#5 lore raw

PATCH 1/1 Assume we're on cpu 0 in early boot

In reply to Paul Mackerras

There's a small period early in boot where we don't know which cpu we're running on. That's ok, except that it means we have no paca, or more correctly that our paca pointer points somewhere random.

So that we can safely call things like smp_processor_id(), we need a paca, so just assume we're on cpu 0. No code should _write_ to the paca before we've set the correct one up.

We setup the proper paca after we've scanned the flat device tree in early_setup(), so there's no need to do it again in start_here_common.

---

 arch/powerpc/kernel/head_64.S  |   11 -----------
 arch/powerpc/kernel/setup_64.c |    3 +++
 2 files changed, 3 insertions(+), 11 deletions(-)
arch/powerpc/kernel/head_64.S +0 -11
Index: to-merge/arch/powerpc/kernel/head_64.S
===================================================================
@@ -1602,9 +1602,6 @@ _GLOBAL(__start_initialization_multiplat
 	/* Setup some critical 970 SPRs before switching MMU off */
 	bl	.__970_cpu_preinit
 
-	/* cpu # */
-	li	r24,0
-
 	/* Switch off MMU if not already */
 	LOAD_REG_IMMEDIATE(r4, .__after_prom_start - KERNELBASE)
 	add	r4,r4,r30
@@ -1962,14 +1959,6 @@ _STATIC(start_here_common)
 	li	r3,0
 	bl	.do_cpu_ftr_fixups
 
-	LOAD_REG_IMMEDIATE(r26, boot_cpuid)
-	lwz	r26,0(r26)
-
-	LOAD_REG_IMMEDIATE(r24, paca)	/* Get base vaddr of paca array  */
-	mulli	r13,r26,PACA_SIZE	/* Calculate vaddr of right paca */
-	add	r13,r13,r24		/* for this processor.		 */
-	mtspr	SPRN_SPRG3,r13
-
 	/* ptr to current */
 	LOAD_REG_IMMEDIATE(r4, init_task)
 	std	r4,PACACURRENT(r13)
arch/powerpc/kernel/setup_64.c +3 -0
Index: to-merge/arch/powerpc/kernel/setup_64.c
===================================================================
@@ -177,6 +177,9 @@ void __init setup_paca(int cpu)
 
 void __init early_setup(unsigned long dt_ptr)
 {
+	/* Assume we're on cpu 0 for now. Don't write to the paca yet! */
+	setup_paca(0);
+
 	/* Enable early debugging if any specified (see udbg.h) */
 	udbg_early_init();