[RFC 0/3] ARM: thread_info to task_struct

[RFC 0/3] ARM: thread_info to task_struct

[Zubin Mithra]

These are a set of preparatory patches for moving the thread_info into
the task_struct for ARM. The patches that perform the actual move are
not implemented yet. A patch to make a register(such as sp_usr) point to
thread_info will likely be required, eventually.

Zubin Mithra (3):
  arm: factor out current_stack_pointer
  arm: make cpu a percpu variable
  arm: add refcounting for task stacks

 arch/arm/include/asm/cputype.h       |  3 +++
 arch/arm/include/asm/percpu.h        |  2 ++
 arch/arm/include/asm/smp.h           |  3 ++-
 arch/arm/include/asm/stack_pointer.h |  9 +++++++++
 arch/arm/include/asm/thread_info.h   |  6 +-----
 arch/arm/kernel/process.c            | 21 +++++++++++++++------
 arch/arm/kernel/return_address.c     |  1 +
 arch/arm/kernel/setup.c              |  2 +-
 arch/arm/kernel/smp.c                |  5 +++--
 arch/arm/kernel/stacktrace.c         |  8 ++++++++
 arch/arm/kernel/topology.c           |  5 +++++
 arch/arm/kernel/traps.c              |  5 +++++
 arch/arm/kernel/unwind.c             |  1 +
 13 files changed, 56 insertions(+), 15 deletions(-)
 create mode 100644 arch/arm/include/asm/stack_pointer.h

-- 
2.16.2.804.g6dcf76e118-goog

[RFC 1/3] arm: factor out current_stack_pointer

[Zubin Mithra]

Multiple header files that rely on current_stack_pointer do not have the
necessary include <asm/thread_info.h>, and are thus fragile to changes
in the header soup.

Subsequent patches will affect the header soup such that including
<asm/thread_info.h> will result in a circular header include. Factor
current_thread_info into its own header and have all the users include
this header explicitely.

Signed-off-by: Zubin Mithra <zsm@chromium.org>
---
 arch/arm/include/asm/percpu.h        | 2 ++
 arch/arm/include/asm/stack_pointer.h | 9 +++++++++
 arch/arm/include/asm/thread_info.h   | 6 +-----
 arch/arm/kernel/return_address.c     | 1 +
 arch/arm/kernel/stacktrace.c         | 1 +
 arch/arm/kernel/unwind.c             | 1 +
 6 files changed, 15 insertions(+), 5 deletions(-)
 create mode 100644 arch/arm/include/asm/stack_pointer.h

diff --git a/arch/arm/include/asm/percpu.h b/arch/arm/include/asm/percpu.h
index a89b4076cde4..6a19e634d0cf 100644
--- a/arch/arm/include/asm/percpu.h
+++ b/arch/arm/include/asm/percpu.h
@@ -16,6 +16,8 @@
 #ifndef _ASM_ARM_PERCPU_H_
 #define _ASM_ARM_PERCPU_H_
 
+#include <asm/stack_pointer.h>
+
 /*
  * Same as asm-generic/percpu.h, except that we store the per cpu offset
  * in the TPIDRPRW. TPIDRPRW only exists on V6K and V7
diff --git a/arch/arm/include/asm/stack_pointer.h b/arch/arm/include/asm/stack_pointer.h
new file mode 100644
index 000000000000..5a6a8c6d9208
--- /dev/null
+++ b/arch/arm/include/asm/stack_pointer.h
@@ -0,0 +1,9 @@
+#ifndef __ASM_STACK_POINTER_H
+#define __ASM_STACK_POINTER_H
+
+/*
+ * how to get the current stack pointer in C
+ */
+register unsigned long current_stack_pointer asm ("sp");
+
+#endif
diff --git a/arch/arm/include/asm/thread_info.h b/arch/arm/include/asm/thread_info.h
index e71cc35de163..56317024d221 100644
--- a/arch/arm/include/asm/thread_info.h
+++ b/arch/arm/include/asm/thread_info.h
@@ -25,6 +25,7 @@
 struct task_struct;
 
 #include <asm/types.h>
+#include <asm/stack_pointer.h>
 
 typedef unsigned long mm_segment_t;
 
@@ -75,11 +76,6 @@ struct thread_info {
 	.addr_limit	= KERNEL_DS,					\
 }
 
-/*
- * how to get the current stack pointer in C
- */
-register unsigned long current_stack_pointer asm ("sp");
-
 /*
  * how to get the thread information struct from C
  */
diff --git a/arch/arm/kernel/return_address.c b/arch/arm/kernel/return_address.c
index 36ed35073289..d76e64250816 100644
--- a/arch/arm/kernel/return_address.c
+++ b/arch/arm/kernel/return_address.c
@@ -14,6 +14,7 @@
 #if defined(CONFIG_FRAME_POINTER) && !defined(CONFIG_ARM_UNWIND)
 #include <linux/sched.h>
 
+#include <asm/stack_pointer.h>
 #include <asm/stacktrace.h>
 
 struct return_address_data {
diff --git a/arch/arm/kernel/stacktrace.c b/arch/arm/kernel/stacktrace.c
index a56e7c856ab5..d519b8e0797f 100644
--- a/arch/arm/kernel/stacktrace.c
+++ b/arch/arm/kernel/stacktrace.c
@@ -4,6 +4,7 @@
 #include <linux/stacktrace.h>
 
 #include <asm/sections.h>
+#include <asm/stack_pointer.h>
 #include <asm/stacktrace.h>
 #include <asm/traps.h>
 
diff --git a/arch/arm/kernel/unwind.c b/arch/arm/kernel/unwind.c
index 0bee233fef9a..860f8ac187e0 100644
--- a/arch/arm/kernel/unwind.c
+++ b/arch/arm/kernel/unwind.c
@@ -45,6 +45,7 @@
 #include <linux/spinlock.h>
 #include <linux/list.h>
 
+#include <asm/stack_pointer.h>
 #include <asm/stacktrace.h>
 #include <asm/traps.h>
 #include <asm/unwind.h>
-- 
2.16.2.804.g6dcf76e118-goog

[RFC 2/3] arm: make cpu a percpu variable

[Zubin Mithra]

Without CONFIG_THREAD_INFO_IN_TASK, core code maintains thread_info::cpu
and arch specific code can use this to build raw_smp_processor_id().
With CONFIG_THREAD_INFO_IN_TASK, core code maintains task_struct::cpu,
and arch specific code cannot access this due to header file circular
dependency.

Instead, we can maintain a percpu variable containing the cpu number.

This also means that cpu numbers obtained using smp_processor_id cannot
be used to set_my_cpu_offset. Use task_cpu(current) instead to get the
cpu in those cases.

Earlier raw_smp_processor_id() was current_thread_info()->cpu :-
	mov	r3, sp
	bic     r3, r3, #8128
	bic     r3, r3, #63
	ldr     r0, [r3, #16]

Now it is *raw_cpu_ptr(&cpu_number) :-
	movw	r3, #57344
	movt	r3, #32917
	mrc	15, 0, r0, cr13, cr0, {4}
	ldr	r0, [r3, r0]

Signed-off-by: Zubin Mithra <zsm@chromium.org>
---
 arch/arm/include/asm/cputype.h | 3 +++
 arch/arm/include/asm/smp.h     | 3 ++-
 arch/arm/kernel/setup.c        | 2 +-
 arch/arm/kernel/smp.c          | 5 +++--
 arch/arm/kernel/topology.c     | 5 +++++
 5 files changed, 14 insertions(+), 4 deletions(-)

diff --git a/arch/arm/include/asm/cputype.h b/arch/arm/include/asm/cputype.h
index cb546425da8a..10148c7af12f 100644
--- a/arch/arm/include/asm/cputype.h
+++ b/arch/arm/include/asm/cputype.h
@@ -2,6 +2,9 @@
 #ifndef __ASM_ARM_CPUTYPE_H
 #define __ASM_ARM_CPUTYPE_H
 
+#include <asm/percpu.h>
+DECLARE_PER_CPU(int, cpu_number);
+
 #include <linux/stringify.h>
 #include <linux/kernel.h>
 
diff --git a/arch/arm/include/asm/smp.h b/arch/arm/include/asm/smp.h
index 709a55989cb0..7e7b198174d6 100644
--- a/arch/arm/include/asm/smp.h
+++ b/arch/arm/include/asm/smp.h
@@ -18,7 +18,8 @@
 # error "<asm/smp.h> included in non-SMP build"
 #endif
 
-#define raw_smp_processor_id() (current_thread_info()->cpu)
+#include <asm/cputype.h>
+#define raw_smp_processor_id() (*raw_cpu_ptr(&cpu_number))
 
 struct seq_file;
 
diff --git a/arch/arm/kernel/setup.c b/arch/arm/kernel/setup.c
index fc40a2b40595..6323b4d80306 100644
--- a/arch/arm/kernel/setup.c
+++ b/arch/arm/kernel/setup.c
@@ -521,7 +521,7 @@ static void __init elf_hwcap_fixup(void)
 void notrace cpu_init(void)
 {
 #ifndef CONFIG_CPU_V7M
-	unsigned int cpu = smp_processor_id();
+	unsigned int cpu = task_cpu(current);
 	struct stack *stk = &stacks[cpu];
 
 	if (cpu >= NR_CPUS) {
diff --git a/arch/arm/kernel/smp.c b/arch/arm/kernel/smp.c
index 2da087926ebe..9584a52eb156 100644
--- a/arch/arm/kernel/smp.c
+++ b/arch/arm/kernel/smp.c
@@ -372,7 +372,8 @@ asmlinkage void secondary_start_kernel(void)
 	 * All kernel threads share the same mm context; grab a
 	 * reference and switch to it.
 	 */
-	cpu = smp_processor_id();
+	cpu = task_cpu(current);
+	set_my_cpu_offset(per_cpu_offset(cpu));
 	mmgrab(mm);
 	current->active_mm = mm;
 	cpumask_set_cpu(cpu, mm_cpumask(mm));
@@ -436,7 +437,7 @@ void __init smp_cpus_done(unsigned int max_cpus)
 
 void __init smp_prepare_boot_cpu(void)
 {
-	set_my_cpu_offset(per_cpu_offset(smp_processor_id()));
+	set_my_cpu_offset(per_cpu_offset(task_cpu(current)));
 }
 
 void __init smp_prepare_cpus(unsigned int max_cpus)
diff --git a/arch/arm/kernel/topology.c b/arch/arm/kernel/topology.c
index 24ac3cab411d..34b2136deab6 100644
--- a/arch/arm/kernel/topology.c
+++ b/arch/arm/kernel/topology.c
@@ -30,6 +30,9 @@
 #include <asm/cputype.h>
 #include <asm/topology.h>
 
+DEFINE_PER_CPU_READ_MOSTLY(int, cpu_number);
+EXPORT_PER_CPU_SYMBOL(cpu_number);
+
 /*
  * cpu capacity scale management
  */
@@ -310,6 +313,8 @@ void __init init_cpu_topology(void)
 	for_each_possible_cpu(cpu) {
 		struct cputopo_arm *cpu_topo = &(cpu_topology[cpu]);
 
+		per_cpu(cpu_number, cpu) = cpu;
+
 		cpu_topo->thread_id = -1;
 		cpu_topo->core_id =  -1;
 		cpu_topo->socket_id = -1;
-- 
2.16.2.804.g6dcf76e118-goog

[RFC 3/3] arm: add refcounting for task stacks

[Zubin Mithra]

When CONFIG_THREAD_INFO_IN_TASK is enabled, thread stacks may be freed
before a stack is destroyed. This patch adds in recounting to ensure
freed stacks are not used. If CONFIG_THREAD_INFO_IN_TASK is not used, no
refcounting is performed.

Signed-off-by: Zubin Mithra <zsm@chromium.org>
---
 arch/arm/kernel/process.c    | 21 +++++++++++++++------
 arch/arm/kernel/stacktrace.c |  7 +++++++
 arch/arm/kernel/traps.c      |  5 +++++
 3 files changed, 27 insertions(+), 6 deletions(-)

diff --git a/arch/arm/kernel/process.c b/arch/arm/kernel/process.c
index 1523cb18b109..d3f87d3049da 100644
--- a/arch/arm/kernel/process.c
+++ b/arch/arm/kernel/process.c
@@ -297,7 +297,7 @@ EXPORT_SYMBOL(dump_fpu);
 unsigned long get_wchan(struct task_struct *p)
 {
 	struct stackframe frame;
-	unsigned long stack_page;
+	unsigned long stack_page, ret = 0;
 	int count = 0;
 	if (!p || p == current || p->state == TASK_RUNNING)
 		return 0;
@@ -306,16 +306,25 @@ unsigned long get_wchan(struct task_struct *p)
 	frame.sp = thread_saved_sp(p);
 	frame.lr = 0;			/* recovered from the stack */
 	frame.pc = thread_saved_pc(p);
-	stack_page = (unsigned long)task_stack_page(p);
+
+	stack_page = (unsigned long)try_get_task_stack(p);
+	if (!stack_page)
+		return 0;
+
 	do {
 		if (frame.sp < stack_page ||
 		    frame.sp >= stack_page + THREAD_SIZE ||
 		    unwind_frame(&frame) < 0)
-			return 0;
-		if (!in_sched_functions(frame.pc))
-			return frame.pc;
+			goto out;
+		if (!in_sched_functions(frame.pc)) {
+			ret = frame.pc;
+			goto out;
+		}
 	} while (count ++ < 16);
-	return 0;
+
+out:
+	put_task_stack(p);
+	return ret;
 }
 
 unsigned long arch_randomize_brk(struct mm_struct *mm)
diff --git a/arch/arm/kernel/stacktrace.c b/arch/arm/kernel/stacktrace.c
index d519b8e0797f..ff20d5941724 100644
--- a/arch/arm/kernel/stacktrace.c
+++ b/arch/arm/kernel/stacktrace.c
@@ -1,6 +1,7 @@
 #include <linux/export.h>
 #include <linux/sched.h>
 #include <linux/sched/debug.h>
+#include <linux/sched/task_stack.h>
 #include <linux/stacktrace.h>
 
 #include <asm/sections.h>
@@ -105,6 +106,9 @@ static noinline void __save_stack_trace(struct task_struct *tsk,
 	struct stack_trace_data data;
 	struct stackframe frame;
 
+	if (!try_get_task_stack(tsk))
+		return;
+
 	data.trace = trace;
 	data.skip = trace->skip;
 	data.no_sched_functions = nosched;
@@ -118,6 +122,7 @@ static noinline void __save_stack_trace(struct task_struct *tsk,
 		 */
 		if (trace->nr_entries < trace->max_entries)
 			trace->entries[trace->nr_entries++] = ULONG_MAX;
+		put_task_stack(tsk);
 		return;
 #else
 		frame.fp = thread_saved_fp(tsk);
@@ -137,6 +142,8 @@ static noinline void __save_stack_trace(struct task_struct *tsk,
 	walk_stackframe(&frame, save_trace, &data);
 	if (trace->nr_entries < trace->max_entries)
 		trace->entries[trace->nr_entries++] = ULONG_MAX;
+
+	put_task_stack(tsk);
 }
 
 void save_stack_trace_regs(struct pt_regs *regs, struct stack_trace *trace)
diff --git a/arch/arm/kernel/traps.c b/arch/arm/kernel/traps.c
index 5e3633c24e63..3eb923033591 100644
--- a/arch/arm/kernel/traps.c
+++ b/arch/arm/kernel/traps.c
@@ -216,6 +216,9 @@ static void dump_backtrace(struct pt_regs *regs, struct task_struct *tsk)
 	if (!tsk)
 		tsk = current;
 
+	if (!try_get_task_stack(tsk))
+		return;
+
 	if (regs) {
 		fp = frame_pointer(regs);
 		mode = processor_mode(regs);
@@ -239,6 +242,8 @@ static void dump_backtrace(struct pt_regs *regs, struct task_struct *tsk)
 
 	if (ok)
 		c_backtrace(fp, mode);
+
+	put_task_stack(tsk);
 }
 #endif
 
-- 
2.16.2.804.g6dcf76e118-goog

[RFC 1/3] arm: factor out current_stack_pointer

[Mark Rutland]

On Thu, Mar 15, 2018 at 05:08:57PM +0000, Zubin Mithra wrote:
> Multiple header files that rely on current_stack_pointer do not have the
> necessary include <asm/thread_info.h>, and are thus fragile to changes
> in the header soup.
> 
> Subsequent patches will affect the header soup such that including
> <asm/thread_info.h> will result in a circular header include. Factor
> current_thread_info into its own header and have all the users include
> this header explicitely.

Nit: s/explicitely/explicitly/

> Signed-off-by: Zubin Mithra <zsm@chromium.org>

I can't remember precisely what the circular include dependency was that
necessitated this for arm64, but regardless this looks reasonable to me.
FWIW:

Acked-by: Mark Rutland <mark.rutland@arm.com>

... though we'll need the rest of the THREAD_INFO_IN_TASK patches for
this to be a worthwhile change.

Thanks,
Mark.

> ---
>  arch/arm/include/asm/percpu.h        | 2 ++
>  arch/arm/include/asm/stack_pointer.h | 9 +++++++++
>  arch/arm/include/asm/thread_info.h   | 6 +-----
>  arch/arm/kernel/return_address.c     | 1 +
>  arch/arm/kernel/stacktrace.c         | 1 +
>  arch/arm/kernel/unwind.c             | 1 +
>  6 files changed, 15 insertions(+), 5 deletions(-)
>  create mode 100644 arch/arm/include/asm/stack_pointer.h
> 
> diff --git a/arch/arm/include/asm/percpu.h b/arch/arm/include/asm/percpu.h
> index a89b4076cde4..6a19e634d0cf 100644
> --- a/arch/arm/include/asm/percpu.h
> +++ b/arch/arm/include/asm/percpu.h
> @@ -16,6 +16,8 @@
>  #ifndef _ASM_ARM_PERCPU_H_
>  #define _ASM_ARM_PERCPU_H_
>  
> +#include <asm/stack_pointer.h>
> +
>  /*
>   * Same as asm-generic/percpu.h, except that we store the per cpu offset
>   * in the TPIDRPRW. TPIDRPRW only exists on V6K and V7
> diff --git a/arch/arm/include/asm/stack_pointer.h b/arch/arm/include/asm/stack_pointer.h
> new file mode 100644
> index 000000000000..5a6a8c6d9208
> --- /dev/null
> +++ b/arch/arm/include/asm/stack_pointer.h
> @@ -0,0 +1,9 @@
> +#ifndef __ASM_STACK_POINTER_H
> +#define __ASM_STACK_POINTER_H
> +
> +/*
> + * how to get the current stack pointer in C
> + */
> +register unsigned long current_stack_pointer asm ("sp");
> +
> +#endif
> diff --git a/arch/arm/include/asm/thread_info.h b/arch/arm/include/asm/thread_info.h
> index e71cc35de163..56317024d221 100644
> --- a/arch/arm/include/asm/thread_info.h
> +++ b/arch/arm/include/asm/thread_info.h
> @@ -25,6 +25,7 @@
>  struct task_struct;
>  
>  #include <asm/types.h>
> +#include <asm/stack_pointer.h>
>  
>  typedef unsigned long mm_segment_t;
>  
> @@ -75,11 +76,6 @@ struct thread_info {
>  	.addr_limit	= KERNEL_DS,					\
>  }
>  
> -/*
> - * how to get the current stack pointer in C
> - */
> -register unsigned long current_stack_pointer asm ("sp");
> -
>  /*
>   * how to get the thread information struct from C
>   */
> diff --git a/arch/arm/kernel/return_address.c b/arch/arm/kernel/return_address.c
> index 36ed35073289..d76e64250816 100644
> --- a/arch/arm/kernel/return_address.c
> +++ b/arch/arm/kernel/return_address.c
> @@ -14,6 +14,7 @@
>  #if defined(CONFIG_FRAME_POINTER) && !defined(CONFIG_ARM_UNWIND)
>  #include <linux/sched.h>
>  
> +#include <asm/stack_pointer.h>
>  #include <asm/stacktrace.h>
>  
>  struct return_address_data {
> diff --git a/arch/arm/kernel/stacktrace.c b/arch/arm/kernel/stacktrace.c
> index a56e7c856ab5..d519b8e0797f 100644
> --- a/arch/arm/kernel/stacktrace.c
> +++ b/arch/arm/kernel/stacktrace.c
> @@ -4,6 +4,7 @@
>  #include <linux/stacktrace.h>
>  
>  #include <asm/sections.h>
> +#include <asm/stack_pointer.h>
>  #include <asm/stacktrace.h>
>  #include <asm/traps.h>
>  
> diff --git a/arch/arm/kernel/unwind.c b/arch/arm/kernel/unwind.c
> index 0bee233fef9a..860f8ac187e0 100644
> --- a/arch/arm/kernel/unwind.c
> +++ b/arch/arm/kernel/unwind.c
> @@ -45,6 +45,7 @@
>  #include <linux/spinlock.h>
>  #include <linux/list.h>
>  
> +#include <asm/stack_pointer.h>
>  #include <asm/stacktrace.h>
>  #include <asm/traps.h>
>  #include <asm/unwind.h>
> -- 
> 2.16.2.804.g6dcf76e118-goog
> 

[RFC 2/3] arm: make cpu a percpu variable

[Mark Rutland]

On Thu, Mar 15, 2018 at 05:08:58PM +0000, Zubin Mithra wrote:
> Without CONFIG_THREAD_INFO_IN_TASK, core code maintains thread_info::cpu
> and arch specific code can use this to build raw_smp_processor_id().
> With CONFIG_THREAD_INFO_IN_TASK, core code maintains task_struct::cpu,
> and arch specific code cannot access this due to header file circular
> dependency.
> 
> Instead, we can maintain a percpu variable containing the cpu number.
> 
> This also means that cpu numbers obtained using smp_processor_id cannot
> be used to set_my_cpu_offset. Use task_cpu(current) instead to get the
> cpu in those cases.
> 
> Earlier raw_smp_processor_id() was current_thread_info()->cpu :-
> 	mov	r3, sp
> 	bic     r3, r3, #8128
> 	bic     r3, r3, #63
> 	ldr     r0, [r3, #16]
> 
> Now it is *raw_cpu_ptr(&cpu_number) :-
> 	movw	r3, #57344
> 	movt	r3, #32917
> 	mrc	15, 0, r0, cr13, cr0, {4}
> 	ldr	r0, [r3, r0]

It's probably worth noting that with the thread_info moved off of the
stack, were we to use that to get at the cpu number, this would require
a similar sequence to the raw_cpu_ptr() code here.

This might make more sense after a patch using a register for the
thread_info (which would also allow for having separate IRQ stacks).

> 
> Signed-off-by: Zubin Mithra <zsm@chromium.org>
> ---
>  arch/arm/include/asm/cputype.h | 3 +++
>  arch/arm/include/asm/smp.h     | 3 ++-
>  arch/arm/kernel/setup.c        | 2 +-
>  arch/arm/kernel/smp.c          | 5 +++--
>  arch/arm/kernel/topology.c     | 5 +++++
>  5 files changed, 14 insertions(+), 4 deletions(-)
> 
> diff --git a/arch/arm/include/asm/cputype.h b/arch/arm/include/asm/cputype.h
> index cb546425da8a..10148c7af12f 100644
> --- a/arch/arm/include/asm/cputype.h
> +++ b/arch/arm/include/asm/cputype.h
> @@ -2,6 +2,9 @@
>  #ifndef __ASM_ARM_CPUTYPE_H
>  #define __ASM_ARM_CPUTYPE_H
>  
> +#include <asm/percpu.h>
> +DECLARE_PER_CPU(int, cpu_number);
> +

Why isn't this defined in <asm/smp.h>, as we do on arm64?

> --- a/arch/arm/include/asm/smp.h
> +++ b/arch/arm/include/asm/smp.h
> @@ -18,7 +18,8 @@
>  # error "<asm/smp.h> included in non-SMP build"
>  #endif
>  
> -#define raw_smp_processor_id() (current_thread_info()->cpu)
> +#include <asm/cputype.h>
> +#define raw_smp_processor_id() (*raw_cpu_ptr(&cpu_number))

It might be worth dragging the comment along from arm64, explaining why
we must use *raw_cpu_ptr(), and not this_cpu_read(), etc.

> --- a/arch/arm/kernel/topology.c
> +++ b/arch/arm/kernel/topology.c
> @@ -30,6 +30,9 @@
>  #include <asm/cputype.h>
>  #include <asm/topology.h>
>  
> +DEFINE_PER_CPU_READ_MOSTLY(int, cpu_number);
> +EXPORT_PER_CPU_SYMBOL(cpu_number);

Why not in smp.c, as with arm64?

> +
>  /*
>   * cpu capacity scale management
>   */
> @@ -310,6 +313,8 @@ void __init init_cpu_topology(void)
>  	for_each_possible_cpu(cpu) {
>  		struct cputopo_arm *cpu_topo = &(cpu_topology[cpu]);
>  
> +		per_cpu(cpu_number, cpu) = cpu;

Likewise, why isn't this in smp_prepare_cpus() in smp.c?

This isn't toplogy related, so it would be better placed there.

Thanks,
Mark.