[patch 0/6] i386 gdt and percpu cleanups

LKML Archive on lore.kernel.org
help / color / mirror / Atom feed

* [patch 0/6] i386 gdt and percpu cleanups
@ 2007-03-31  2:00 Jeremy Fitzhardinge
  2007-03-31  2:00 ` [patch 1/6] i386: Account for module percpu space separately from kernel percpu Jeremy Fitzhardinge
                   ` (7 more replies)
  0 siblings, 8 replies; 13+ messages in thread
From: Jeremy Fitzhardinge @ 2007-03-31  2:00 UTC (permalink / raw)
  To: Andi Kleen; +Cc: Andrew Morton, virtualization, lkml

Hi Andi,

This is a series of patches based on your latest queue (as of the
other day, at least).

It includes:
 - the most recent patch to compute the appropriate amount of percpu
   space to allocate, using a separate reservation for modules where
   needed.
 - make the percpu sections page-aligned, so that percpu variables can
   be page aligned if needed (which is used by gdt_page)
 - page-align the gdt
 - remove the pda and convert all pda usages into percpu variables
   (percpu variables still use the %fs prefix mechanism the pda used)
 - some improvements to asm-i386/percpu.h to make asm access to percpu
   variables easy
 - define per_cpu_offset in asm-i386/percpu.h, to match asm-generic/

Thanks,
	J

-- 


^ permalink raw reply	[flat|nested] 13+ messages in thread

* [patch 1/6] i386: Account for module percpu space separately from kernel percpu
  2007-03-31  2:00 [patch 0/6] i386 gdt and percpu cleanups Jeremy Fitzhardinge
@ 2007-03-31  2:00 ` Jeremy Fitzhardinge
  2007-03-31  2:00 ` [patch 2/6] Allow percpu variables to be page-aligned Jeremy Fitzhardinge
                   ` (6 subsequent siblings)
  7 siblings, 0 replies; 13+ messages in thread
From: Jeremy Fitzhardinge @ 2007-03-31  2:00 UTC (permalink / raw)
  To: Andi Kleen
  Cc: Andrew Morton, virtualization, lkml, Rusty Russell, Eric W. Biederman

[-- Attachment #1: separate-module-percpu-space.patch --]
[-- Type: text/plain, Size: 4079 bytes --]

Rather than using a single constant PERCPU_ENOUGH_ROOM, compute it as
the sum of kernel_percpu + PERCPU_MODULE_RESERVE.  This is now common
to all architectures; if an architecture wants to set
PERCPU_ENOUGH_ROOM to something special, then it may do so (ia64 is
the only one which does).

Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: Eric W. Biederman <ebiederm@xmission.com>
Cc: Andi Kleen <ak@suse.de>

---
 include/asm-alpha/percpu.h   |   14 --------------
 include/asm-sparc64/percpu.h |   10 ----------
 include/asm-x86_64/percpu.h  |   10 ----------
 include/linux/percpu.h       |    9 ++++++++-
 init/main.c                  |    7 ++-----
 kernel/module.c              |    2 +-
 6 files changed, 11 insertions(+), 41 deletions(-)

===================================================================
--- a/include/asm-alpha/percpu.h
+++ b/include/asm-alpha/percpu.h
@@ -1,19 +1,5 @@
 #ifndef __ALPHA_PERCPU_H
 #define __ALPHA_PERCPU_H
-
-/*
- * Increase the per cpu area for Alpha so that
- * modules using percpu area can load.
- */
-#ifdef CONFIG_MODULES
-# define PERCPU_MODULE_RESERVE 8192
-#else
-# define PERCPU_MODULE_RESERVE 0
-#endif
-
-#define PERCPU_ENOUGH_ROOM \
-	(ALIGN(__per_cpu_end - __per_cpu_start, SMP_CACHE_BYTES) + \
-	 PERCPU_MODULE_RESERVE)
 
 #include <asm-generic/percpu.h>
 
===================================================================
--- a/include/asm-sparc64/percpu.h
+++ b/include/asm-sparc64/percpu.h
@@ -4,16 +4,6 @@
 #include <linux/compiler.h>
 
 #ifdef CONFIG_SMP
-
-#ifdef CONFIG_MODULES
-# define PERCPU_MODULE_RESERVE 8192
-#else
-# define PERCPU_MODULE_RESERVE 0
-#endif
-
-#define PERCPU_ENOUGH_ROOM \
-	(ALIGN(__per_cpu_end - __per_cpu_start, SMP_CACHE_BYTES) + \
-	 PERCPU_MODULE_RESERVE)
 
 extern void setup_per_cpu_areas(void);
 
===================================================================
--- a/include/asm-x86_64/percpu.h
+++ b/include/asm-x86_64/percpu.h
@@ -10,16 +10,6 @@
 #ifdef CONFIG_SMP
 
 #include <asm/pda.h>
-
-#ifdef CONFIG_MODULES
-# define PERCPU_MODULE_RESERVE 8192
-#else
-# define PERCPU_MODULE_RESERVE 0
-#endif
-
-#define PERCPU_ENOUGH_ROOM \
-	(ALIGN(__per_cpu_end - __per_cpu_start, SMP_CACHE_BYTES) + \
-	 PERCPU_MODULE_RESERVE)
 
 #define __per_cpu_offset(cpu) (cpu_pda(cpu)->data_offset)
 #define __my_cpu_offset() read_pda(data_offset)
===================================================================
--- a/include/linux/percpu.h
+++ b/include/linux/percpu.h
@@ -11,8 +11,15 @@
 
 /* Enough to cover all DEFINE_PER_CPUs in kernel, including modules. */
 #ifndef PERCPU_ENOUGH_ROOM
-#define PERCPU_ENOUGH_ROOM 32768
+#ifdef CONFIG_MODULES
+#define PERCPU_MODULE_RESERVE	8192
+#else
+#define PERCPU_MODULE_RESERVE	0
 #endif
+
+#define PERCPU_ENOUGH_ROOM						\
+	(__per_cpu_end - __per_cpu_start + PERCPU_MODULE_RESERVE)
+#endif	/* PERCPU_ENOUGH_ROOM */
 
 /*
  * Must be an lvalue. Since @var must be a simple identifier,
===================================================================
--- a/init/main.c
+++ b/init/main.c
@@ -369,11 +369,8 @@ static void __init setup_per_cpu_areas(v
 	unsigned long nr_possible_cpus = num_possible_cpus();
 
 	/* Copy section for each CPU (we discard the original) */
-	size = ALIGN(__per_cpu_end - __per_cpu_start, SMP_CACHE_BYTES);
-#ifdef CONFIG_MODULES
-	if (size < PERCPU_ENOUGH_ROOM)
-		size = PERCPU_ENOUGH_ROOM;
-#endif
+
+	size = ALIGN(PERCPU_ENOUGH_ROOM, SMP_CACHE_BYTES);
 	ptr = alloc_bootmem(size * nr_possible_cpus);
 
 	for_each_possible_cpu(i) {
===================================================================
--- a/kernel/module.c
+++ b/kernel/module.c
@@ -430,7 +430,7 @@ static int percpu_modinit(void)
 	pcpu_size = kmalloc(sizeof(pcpu_size[0]) * pcpu_num_allocated,
 			    GFP_KERNEL);
 	/* Static in-kernel percpu data (used). */
-	pcpu_size[0] = -ALIGN(__per_cpu_end-__per_cpu_start, SMP_CACHE_BYTES);
+	pcpu_size[0] = -(__per_cpu_end-__per_cpu_start);
 	/* Free room. */
 	pcpu_size[1] = PERCPU_ENOUGH_ROOM + pcpu_size[0];
 	if (pcpu_size[1] < 0) {

-- 


^ permalink raw reply	[flat|nested] 13+ messages in thread

* [patch 2/6] Allow percpu variables to be page-aligned
  2007-03-31  2:00 [patch 0/6] i386 gdt and percpu cleanups Jeremy Fitzhardinge
  2007-03-31  2:00 ` [patch 1/6] i386: Account for module percpu space separately from kernel percpu Jeremy Fitzhardinge
@ 2007-03-31  2:00 ` Jeremy Fitzhardinge
  2007-03-31  2:00 ` [patch 3/6] Page-align the GDT Jeremy Fitzhardinge
                   ` (5 subsequent siblings)
  7 siblings, 0 replies; 13+ messages in thread
From: Jeremy Fitzhardinge @ 2007-03-31  2:00 UTC (permalink / raw)
  To: Andi Kleen
  Cc: Andrew Morton, virtualization, lkml, Rusty Russell, Ingo Molnar

[-- Attachment #1: allow-per-cpu-variables-to-be-page-aligned.patch --]
[-- Type: text/plain, Size: 10146 bytes --]

Let's allow page-alignment in general for per-cpu data (wanted by Xen, and
Ingo suggested KVM as well).

Because larger alignments can use more room, we increase the max per-cpu
memory to 64k rather than 32k: it's getting a little tight.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Acked-by: Ingo Molnar <mingo@elte.hu>
Cc: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
---
 arch/alpha/kernel/vmlinux.lds.S   |    2 +-
 arch/arm/kernel/vmlinux.lds.S     |    2 +-
 arch/cris/arch-v32/vmlinux.lds.S  |    1 +
 arch/frv/kernel/vmlinux.lds.S     |    1 +
 arch/i386/kernel/vmlinux.lds.S    |    2 +-
 arch/m32r/kernel/vmlinux.lds.S    |    2 +-
 arch/mips/kernel/vmlinux.lds.S    |    2 +-
 arch/parisc/kernel/vmlinux.lds.S  |    2 +-
 arch/powerpc/kernel/setup_64.c    |    4 ++--
 arch/powerpc/kernel/vmlinux.lds.S |    6 +-----
 arch/ppc/kernel/vmlinux.lds.S     |    2 +-
 arch/s390/kernel/vmlinux.lds.S    |    2 +-
 arch/sh/kernel/vmlinux.lds.S      |    2 +-
 arch/sh64/kernel/vmlinux.lds.S    |    2 +-
 arch/sparc/kernel/vmlinux.lds.S   |    2 +-
 arch/sparc64/kernel/smp.c         |    6 +++---
 arch/x86_64/kernel/setup64.c      |    4 ++--
 arch/x86_64/kernel/vmlinux.lds.S  |    2 +-
 arch/xtensa/kernel/vmlinux.lds.S  |    2 +-
 init/main.c                       |    4 ++--
 kernel/module.c                   |    8 ++++----
 21 files changed, 29 insertions(+), 31 deletions(-)

===================================================================
--- a/arch/alpha/kernel/vmlinux.lds.S
+++ b/arch/alpha/kernel/vmlinux.lds.S
@@ -69,7 +69,7 @@ SECTIONS
   . = ALIGN(8);
   SECURITY_INIT
 
-  . = ALIGN(64);
+  . = ALIGN(8192);
   __per_cpu_start = .;
   .data.percpu : { *(.data.percpu) }
   __per_cpu_end = .;
===================================================================
--- a/arch/arm/kernel/vmlinux.lds.S
+++ b/arch/arm/kernel/vmlinux.lds.S
@@ -59,7 +59,7 @@ SECTIONS
 			usr/built-in.o(.init.ramfs)
 		__initramfs_end = .;
 #endif
-		. = ALIGN(64);
+		. = ALIGN(4096);
 		__per_cpu_start = .;
 			*(.data.percpu)
 		__per_cpu_end = .;
===================================================================
--- a/arch/cris/arch-v32/vmlinux.lds.S
+++ b/arch/cris/arch-v32/vmlinux.lds.S
@@ -91,6 +91,7 @@ SECTIONS
 	}
 	SECURITY_INIT
 
+	. =  ALIGN (8192);
 	__per_cpu_start = .;
 	.data.percpu  : { *(.data.percpu) }
 	__per_cpu_end = .;
===================================================================
--- a/arch/frv/kernel/vmlinux.lds.S
+++ b/arch/frv/kernel/vmlinux.lds.S
@@ -57,6 +57,7 @@ SECTIONS
   __alt_instructions_end = .;
  .altinstr_replacement : { *(.altinstr_replacement) }
 
+  . = ALIGN(4096);
   __per_cpu_start = .;
   .data.percpu  : { *(.data.percpu) }
   __per_cpu_end = .;
===================================================================
--- a/arch/i386/kernel/vmlinux.lds.S
+++ b/arch/i386/kernel/vmlinux.lds.S
@@ -194,7 +194,7 @@ SECTIONS
 	__initramfs_end = .;
   }
 #endif
-  . = ALIGN(L1_CACHE_BYTES);
+  . = ALIGN(4096);
   .data.percpu  : AT(ADDR(.data.percpu) - LOAD_OFFSET) {
 	__per_cpu_start = .;
 	*(.data.percpu)
===================================================================
--- a/arch/m32r/kernel/vmlinux.lds.S
+++ b/arch/m32r/kernel/vmlinux.lds.S
@@ -110,7 +110,7 @@ SECTIONS
   __initramfs_end = .;
 #endif
 
-  . = ALIGN(32);
+  . = ALIGN(4096);
   __per_cpu_start = .;
   .data.percpu  : { *(.data.percpu) }
   __per_cpu_end = .;
===================================================================
--- a/arch/mips/kernel/vmlinux.lds.S
+++ b/arch/mips/kernel/vmlinux.lds.S
@@ -119,7 +119,7 @@ SECTIONS
   .init.ramfs : { *(.init.ramfs) }
   __initramfs_end = .;
 #endif
-  . = ALIGN(32);
+  . = ALIGN(_PAGE_SIZE);
   __per_cpu_start = .;
   .data.percpu  : { *(.data.percpu) }
   __per_cpu_end = .;
===================================================================
--- a/arch/parisc/kernel/vmlinux.lds.S
+++ b/arch/parisc/kernel/vmlinux.lds.S
@@ -181,7 +181,7 @@ SECTIONS
   .init.ramfs : { *(.init.ramfs) }
   __initramfs_end = .;
 #endif
-  . = ALIGN(32);
+  . = ALIGN(ASM_PAGE_SIZE);
   __per_cpu_start = .;
   .data.percpu  : { *(.data.percpu) }
   __per_cpu_end = .;
===================================================================
--- a/arch/powerpc/kernel/setup_64.c
+++ b/arch/powerpc/kernel/setup_64.c
@@ -583,14 +583,14 @@ void __init setup_per_cpu_areas(void)
 	char *ptr;
 
 	/* Copy section for each CPU (we discard the original) */
-	size = ALIGN(__per_cpu_end - __per_cpu_start, SMP_CACHE_BYTES);
+	size = ALIGN(__per_cpu_end - __per_cpu_start, PAGE_SIZE);
 #ifdef CONFIG_MODULES
 	if (size < PERCPU_ENOUGH_ROOM)
 		size = PERCPU_ENOUGH_ROOM;
 #endif
 
 	for_each_possible_cpu(i) {
-		ptr = alloc_bootmem_node(NODE_DATA(cpu_to_node(i)), size);
+		ptr = alloc_bootmem_pages_node(NODE_DATA(cpu_to_node(i)), size);
 		if (!ptr)
 			panic("Cannot allocate cpu data for CPU %d\n", i);
 
===================================================================
--- a/arch/powerpc/kernel/vmlinux.lds.S
+++ b/arch/powerpc/kernel/vmlinux.lds.S
@@ -139,11 +139,7 @@ SECTIONS
 		__initramfs_end = .;
 	}
 #endif
-#ifdef CONFIG_PPC32
-	. = ALIGN(32);
-#else
-	. = ALIGN(128);
-#endif
+	. = ALIGN(PAGE_SIZE);
 	.data.percpu : {
 		__per_cpu_start = .;
 		*(.data.percpu)
===================================================================
--- a/arch/ppc/kernel/vmlinux.lds.S
+++ b/arch/ppc/kernel/vmlinux.lds.S
@@ -130,7 +130,7 @@ SECTIONS
   __ftr_fixup : { *(__ftr_fixup) }
   __stop___ftr_fixup = .;
 
-  . = ALIGN(32);
+  . = ALIGN(4096);
   __per_cpu_start = .;
   .data.percpu  : { *(.data.percpu) }
   __per_cpu_end = .;
===================================================================
--- a/arch/s390/kernel/vmlinux.lds.S
+++ b/arch/s390/kernel/vmlinux.lds.S
@@ -99,7 +99,7 @@ SECTIONS
   . = ALIGN(2);
   __initramfs_end = .;
 #endif
-  . = ALIGN(256);
+  . = ALIGN(4096);
   __per_cpu_start = .;
   .data.percpu  : { *(.data.percpu) }
   __per_cpu_end = .;
===================================================================
--- a/arch/sh/kernel/vmlinux.lds.S
+++ b/arch/sh/kernel/vmlinux.lds.S
@@ -54,7 +54,7 @@ SECTIONS
   . = ALIGN(PAGE_SIZE);
   .data.page_aligned : { *(.data.page_aligned) }
 
-  . = ALIGN(L1_CACHE_BYTES);
+  . = ALIGN(PAGE_SIZE);
   __per_cpu_start = .;
   .data.percpu : { *(.data.percpu) }
   __per_cpu_end = .;
===================================================================
--- a/arch/sh64/kernel/vmlinux.lds.S
+++ b/arch/sh64/kernel/vmlinux.lds.S
@@ -85,7 +85,7 @@ SECTIONS
   . = ALIGN(PAGE_SIZE);
   .data.page_aligned : C_PHYS(.data.page_aligned) { *(.data.page_aligned) }
 
-  . = ALIGN(L1_CACHE_BYTES);
+  . = ALIGN(PAGE_SIZE);
   __per_cpu_start = .;
   .data.percpu : C_PHYS(.data.percpu) { *(.data.percpu) }
   __per_cpu_end = . ;
===================================================================
--- a/arch/sparc/kernel/vmlinux.lds.S
+++ b/arch/sparc/kernel/vmlinux.lds.S
@@ -65,7 +65,7 @@ SECTIONS
   __initramfs_end = .;
 #endif
 
-  . = ALIGN(32);
+  . = ALIGN(4096);
   __per_cpu_start = .;
   .data.percpu  : { *(.data.percpu) }
   __per_cpu_end = .;
===================================================================
--- a/arch/sparc64/kernel/smp.c
+++ b/arch/sparc64/kernel/smp.c
@@ -1449,11 +1449,11 @@ void __init setup_per_cpu_areas(void)
 	/* Copy section for each CPU (we discard the original) */
 	goal = PERCPU_ENOUGH_ROOM;
 
-	__per_cpu_shift = 0;
-	for (size = 1UL; size < goal; size <<= 1UL)
+	__per_cpu_shift = PAGE_SHIFT;
+	for (size = PAGE_SIZE; size < goal; size <<= 1UL)
 		__per_cpu_shift++;
 
-	ptr = alloc_bootmem(size * NR_CPUS);
+	ptr = alloc_bootmem_pages(size * NR_CPUS);
 
 	__per_cpu_base = ptr - __per_cpu_start;
 
===================================================================
--- a/arch/x86_64/kernel/setup64.c
+++ b/arch/x86_64/kernel/setup64.c
@@ -103,9 +103,9 @@ void __init setup_per_cpu_areas(void)
 		if (!NODE_DATA(cpu_to_node(i))) {
 			printk("cpu with no node %d, num_online_nodes %d\n",
 			       i, num_online_nodes());
-			ptr = alloc_bootmem(size);
+			ptr = alloc_bootmem_pages(size);
 		} else { 
-			ptr = alloc_bootmem_node(NODE_DATA(cpu_to_node(i)), size);
+			ptr = alloc_bootmem_pages_node(NODE_DATA(cpu_to_node(i)), size);
 		}
 		if (!ptr)
 			panic("Cannot allocate cpu data for CPU %d\n", i);
===================================================================
--- a/arch/x86_64/kernel/vmlinux.lds.S
+++ b/arch/x86_64/kernel/vmlinux.lds.S
@@ -194,7 +194,7 @@ SECTIONS
   __initramfs_end = .;
 #endif
 
-    . = ALIGN(CONFIG_X86_L1_CACHE_BYTES);
+  . = ALIGN(4096);
   __per_cpu_start = .;
   .data.percpu  : AT(ADDR(.data.percpu) - LOAD_OFFSET) { *(.data.percpu) }
   __per_cpu_end = .;
===================================================================
--- a/arch/xtensa/kernel/vmlinux.lds.S
+++ b/arch/xtensa/kernel/vmlinux.lds.S
@@ -198,7 +198,7 @@ SECTIONS
   __ftr_fixup : { *(__ftr_fixup) }
   __stop___ftr_fixup = .;
 
-  . = ALIGN(32);
+  . = ALIGN(4096);
   __per_cpu_start = .;
   .data.percpu  : { *(.data.percpu) }
   __per_cpu_end = .;
===================================================================
--- a/init/main.c
+++ b/init/main.c
@@ -370,8 +370,8 @@ static void __init setup_per_cpu_areas(v
 
 	/* Copy section for each CPU (we discard the original) */
 
-	size = ALIGN(PERCPU_ENOUGH_ROOM, SMP_CACHE_BYTES);
-	ptr = alloc_bootmem(size * nr_possible_cpus);
+	size = ALIGN(PERCPU_ENOUGH_ROOM, PAGE_SIZE);
+	ptr = alloc_bootmem_pages(size * nr_possible_cpus);
 
 	for_each_possible_cpu(i) {
 		__per_cpu_offset[i] = ptr - __per_cpu_start;
===================================================================
--- a/kernel/module.c
+++ b/kernel/module.c
@@ -346,10 +346,10 @@ static void *percpu_modalloc(unsigned lo
 	unsigned int i;
 	void *ptr;
 
-	if (align > SMP_CACHE_BYTES) {
-		printk(KERN_WARNING "%s: per-cpu alignment %li > %i\n",
-		       name, align, SMP_CACHE_BYTES);
-		align = SMP_CACHE_BYTES;
+	if (align > PAGE_SIZE) {
+		printk(KERN_WARNING "%s: per-cpu alignment %li > %li\n",
+		       name, align, PAGE_SIZE);
+		align = PAGE_SIZE;
 	}
 
 	ptr = __per_cpu_start;

-- 


^ permalink raw reply	[flat|nested] 13+ messages in thread

* [patch 3/6] Page-align the GDT
  2007-03-31  2:00 [patch 0/6] i386 gdt and percpu cleanups Jeremy Fitzhardinge
  2007-03-31  2:00 ` [patch 1/6] i386: Account for module percpu space separately from kernel percpu Jeremy Fitzhardinge
  2007-03-31  2:00 ` [patch 2/6] Allow percpu variables to be page-aligned Jeremy Fitzhardinge
@ 2007-03-31  2:00 ` Jeremy Fitzhardinge
  2007-03-31  2:00 ` [patch 4/6] Convert PDA into the percpu section Jeremy Fitzhardinge
                   ` (4 subsequent siblings)
  7 siblings, 0 replies; 13+ messages in thread
From: Jeremy Fitzhardinge @ 2007-03-31  2:00 UTC (permalink / raw)
  To: Andi Kleen; +Cc: Andrew Morton, virtualization, lkml, Rusty Russell

[-- Attachment #1: i386-gdt-cleanups-page-align-the-gdt.patch --]
[-- Type: text/plain, Size: 3467 bytes --]

Xen wants a dedicated page for the GDT.  I believe VMI likes it too.
lguest, KVM and native don't care.

Simple transformation to page-aligned "struct gdt_page".

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Acked-by: Jeremy Fitzhardinge <jeremy@xensource.com>

---
 arch/i386/kernel/cpu/common.c |    6 +++---
 arch/i386/kernel/entry.S      |    2 +-
 arch/i386/kernel/head.S       |    2 +-
 arch/i386/kernel/traps.c      |    2 +-
 include/asm-i386/desc.h       |    9 +++++++--
 5 files changed, 13 insertions(+), 8 deletions(-)

===================================================================
--- a/arch/i386/kernel/cpu/common.c
+++ b/arch/i386/kernel/cpu/common.c
@@ -22,7 +22,7 @@
 
 #include "cpu.h"
 
-DEFINE_PER_CPU(struct desc_struct, cpu_gdt[GDT_ENTRIES]) = {
+DEFINE_PER_CPU(struct gdt_page, gdt_page) = { .gdt = {
 	[GDT_ENTRY_KERNEL_CS] = { 0x0000ffff, 0x00cf9a00 },
 	[GDT_ENTRY_KERNEL_DS] = { 0x0000ffff, 0x00cf9200 },
 	[GDT_ENTRY_DEFAULT_USER_CS] = { 0x0000ffff, 0x00cffa00 },
@@ -48,8 +48,8 @@ DEFINE_PER_CPU(struct desc_struct, cpu_g
 
 	[GDT_ENTRY_ESPFIX_SS] = { 0x00000000, 0x00c09200 },
 	[GDT_ENTRY_PDA] = { 0x00000000, 0x00c09200 }, /* set in setup_pda */
-};
-EXPORT_PER_CPU_SYMBOL_GPL(cpu_gdt);
+} };
+EXPORT_PER_CPU_SYMBOL_GPL(gdt_page);
 
 DEFINE_PER_CPU(struct i386_pda, _cpu_pda) = {
 	._pda = &per_cpu___cpu_pda,
===================================================================
--- a/arch/i386/kernel/entry.S
+++ b/arch/i386/kernel/entry.S
@@ -558,7 +558,7 @@ END(syscall_badsys)
 #define FIXUP_ESPFIX_STACK \
 	/* since we are on a wrong stack, we cant make it a C code :( */ \
 	movl %fs:PDA_cpu, %ebx; \
-	PER_CPU(cpu_gdt, %ebx); \
+	PER_CPU(gdt_page, %ebx); \
 	GET_DESC_BASE(GDT_ENTRY_ESPFIX_SS, %ebx, %eax, %ax, %al, %ah); \
 	addl %esp, %eax; \
 	pushl $__KERNEL_DS; \
===================================================================
--- a/arch/i386/kernel/head.S
+++ b/arch/i386/kernel/head.S
@@ -599,7 +599,7 @@ idt_descr:
 	.word 0				# 32 bit align gdt_desc.address
 ENTRY(early_gdt_descr)
 	.word GDT_ENTRIES*8-1
-	.long per_cpu__cpu_gdt		/* Overwritten for secondary CPUs */
+	.long per_cpu__gdt_page		/* Overwritten for secondary CPUs */
 
 /*
  * The boot_gdt_table must mirror the equivalent in setup.S and is
===================================================================
--- a/arch/i386/kernel/traps.c
+++ b/arch/i386/kernel/traps.c
@@ -1037,7 +1037,7 @@ fastcall unsigned long patch_espfix_desc
 fastcall unsigned long patch_espfix_desc(unsigned long uesp,
 					  unsigned long kesp)
 {
-	struct desc_struct *gdt = __get_cpu_var(cpu_gdt);
+	struct desc_struct *gdt = __get_cpu_var(gdt_page).gdt;
 	unsigned long base = (kesp - uesp) & -THREAD_SIZE;
 	unsigned long new_kesp = kesp - base;
 	unsigned long lim_pages = (new_kesp | (THREAD_SIZE - 1)) >> PAGE_SHIFT;
===================================================================
--- a/include/asm-i386/desc.h
+++ b/include/asm-i386/desc.h
@@ -18,10 +18,15 @@ struct Xgt_desc_struct {
 	unsigned short pad;
 } __attribute__ ((packed));
 
-DECLARE_PER_CPU(struct desc_struct, cpu_gdt[GDT_ENTRIES]);
+struct gdt_page
+{
+	struct desc_struct gdt[GDT_ENTRIES];
+} __attribute__((aligned(PAGE_SIZE)));
+DECLARE_PER_CPU(struct gdt_page, gdt_page);
+
 static inline struct desc_struct *get_cpu_gdt_table(unsigned int cpu)
 {
-	return per_cpu(cpu_gdt, cpu);
+	return per_cpu(gdt_page, cpu).gdt;
 }
 
 extern struct Xgt_desc_struct idt_descr;

-- 


^ permalink raw reply	[flat|nested] 13+ messages in thread

* [patch 4/6] Convert PDA into the percpu section
  2007-03-31  2:00 [patch 0/6] i386 gdt and percpu cleanups Jeremy Fitzhardinge
                   ` (2 preceding siblings ...)
  2007-03-31  2:00 ` [patch 3/6] Page-align the GDT Jeremy Fitzhardinge
@ 2007-03-31  2:00 ` Jeremy Fitzhardinge
  2007-03-31  2:35   ` Andi Kleen
  2007-03-31  2:00 ` [patch 5/6] cleanups to help using per-cpu variables from asm Jeremy Fitzhardinge
                   ` (3 subsequent siblings)
  7 siblings, 1 reply; 13+ messages in thread
From: Jeremy Fitzhardinge @ 2007-03-31  2:00 UTC (permalink / raw)
  To: Andi Kleen; +Cc: Andrew Morton, virtualization, lkml, Rusty Russell

[-- Attachment #1: i386-convert-pda-into-the-percpu-section.patch --]
[-- Type: text/plain, Size: 23094 bytes --]

Currently x86 (similar to x84-64) has a special per-cpu structure
called "i386_pda" which can be easily and efficiently referenced via
the %fs register.  An ELF section is more flexible than a structure,
allowing any piece of code to use this area.  Indeed, such a section
already exists: the per-cpu area.

So this patch:
(1) Removes the PDA and uses per-cpu variables for each current member.
(2) Replaces the __KERNEL_PDA segment with __KERNEL_PERCPU.
(3) Creates a per-cpu mirror of __per_cpu_offset called this_cpu_off, which
    can be used to calculate addresses for this CPU's variables.
(4) Simplifies startup, because %fs doesn't need to be loaded with a
    special segment at early boot; it can be deferred until the first
    percpu area is allocated (or never for UP).

The result is less code and one less x86-specific concept.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Cc: Andi Kleen <ak@suse.de>
---
 arch/i386/kernel/asm-offsets.c |    5 -
 arch/i386/kernel/cpu/common.c  |   17 -----
 arch/i386/kernel/entry.S       |    5 -
 arch/i386/kernel/head.S        |   31 +--------
 arch/i386/kernel/i386_ksyms.c  |    2 
 arch/i386/kernel/irq.c         |    3 
 arch/i386/kernel/process.c     |   12 ++-
 arch/i386/kernel/smpboot.c     |   34 ++++------
 arch/i386/kernel/vmi.c         |    6 -
 arch/i386/kernel/vmlinux.lds.S |    1 
 include/asm-i386/current.h     |    5 -
 include/asm-i386/irq_regs.h    |   12 ++-
 include/asm-i386/pda.h         |   99 ------------------------------
 include/asm-i386/percpu.h      |  132 +++++++++++++++++++++++++++++++++++++---
 include/asm-i386/processor.h   |    2 
 include/asm-i386/segment.h     |    6 -
 include/asm-i386/smp.h         |    4 -
 17 files changed, 179 insertions(+), 197 deletions(-)

===================================================================
--- a/arch/i386/kernel/asm-offsets.c
+++ b/arch/i386/kernel/asm-offsets.c
@@ -15,7 +15,6 @@
 #include <asm/processor.h>
 #include <asm/thread_info.h>
 #include <asm/elf.h>
-#include <asm/pda.h>
 
 #define DEFINE(sym, val) \
         asm volatile("\n->" #sym " %0 " #val : : "i" (val))
@@ -101,10 +100,6 @@ void foo(void)
 
 	OFFSET(crypto_tfm_ctx_offset, crypto_tfm, __crt_ctx);
 
-	BLANK();
- 	OFFSET(PDA_cpu, i386_pda, cpu_number);
-	OFFSET(PDA_pcurrent, i386_pda, pcurrent);
-
 #ifdef CONFIG_PARAVIRT
 	BLANK();
 	OFFSET(PARAVIRT_enabled, paravirt_ops, paravirt_enabled);
===================================================================
--- a/arch/i386/kernel/cpu/common.c
+++ b/arch/i386/kernel/cpu/common.c
@@ -18,7 +18,6 @@
 #include <asm/apic.h>
 #include <mach_apic.h>
 #endif
-#include <asm/pda.h>
 
 #include "cpu.h"
 
@@ -47,12 +46,9 @@ DEFINE_PER_CPU(struct gdt_page, gdt_page
 	[GDT_ENTRY_APMBIOS_BASE+2] = { 0x0000ffff, 0x00409200 }, /* data */
 
 	[GDT_ENTRY_ESPFIX_SS] = { 0x00000000, 0x00c09200 },
-	[GDT_ENTRY_PDA] = { 0x00000000, 0x00c09200 }, /* set in setup_pda */
+	[GDT_ENTRY_PERCPU] = { 0x00000000, 0x00000000 },
 } };
 EXPORT_PER_CPU_SYMBOL_GPL(gdt_page);
-
-DEFINE_PER_CPU(struct i386_pda, _cpu_pda);
-EXPORT_PER_CPU_SYMBOL(_cpu_pda);
 
 static int cachesize_override __cpuinitdata = -1;
 static int disable_x86_fxsr __cpuinitdata;
@@ -627,20 +623,13 @@ void __init early_cpu_init(void)
 #endif
 }
 
-/* Make sure %gs is initialized properly in idle threads */
+/* Make sure %fs is initialized properly in idle threads */
 struct pt_regs * __devinit idle_regs(struct pt_regs *regs)
 {
 	memset(regs, 0, sizeof(struct pt_regs));
-	regs->xfs = __KERNEL_PDA;
+	regs->xfs = __KERNEL_PERCPU;
 	return regs;
 }
-
-/* Initial PDA used by boot CPU */
-struct i386_pda boot_pda = {
-	._pda = &boot_pda,
-	.cpu_number = 0,
-	.pcurrent = &init_task,
-};
 
 /*
  * cpu_init() initializes state that is per-CPU. Some data is already
===================================================================
--- a/arch/i386/kernel/entry.S
+++ b/arch/i386/kernel/entry.S
@@ -132,7 +132,7 @@ 1:
 	movl $(__USER_DS), %edx; \
 	movl %edx, %ds; \
 	movl %edx, %es; \
-	movl $(__KERNEL_PDA), %edx; \
+	movl $(__KERNEL_PERCPU), %edx; \
 	movl %edx, %fs
 
 #define RESTORE_INT_REGS \
@@ -560,7 +560,6 @@ END(syscall_badsys)
 
 #define FIXUP_ESPFIX_STACK \
 	/* since we are on a wrong stack, we cant make it a C code :( */ \
-	movl %fs:PDA_cpu, %ebx; \
 	PER_CPU(gdt_page, %ebx); \
 	GET_DESC_BASE(GDT_ENTRY_ESPFIX_SS, %ebx, %eax, %ax, %al, %ah); \
 	addl %esp, %eax; \
@@ -685,7 +684,7 @@ error_code:
 	pushl %fs
 	CFI_ADJUST_CFA_OFFSET 4
 	/*CFI_REL_OFFSET fs, 0*/
-	movl $(__KERNEL_PDA), %ecx
+	movl $(__KERNEL_PERCPU), %ecx
 	movl %ecx, %fs
 	UNWIND_ESPFIX_STACK
 	popl %ecx
===================================================================
--- a/arch/i386/kernel/head.S
+++ b/arch/i386/kernel/head.S
@@ -317,12 +317,12 @@ 2:	movl %cr0,%eax
 	movl %eax,%cr0
 
 	call check_x87
-	call setup_pda
 	lgdt early_gdt_descr
 	lidt idt_descr
 	ljmp $(__KERNEL_CS),$1f
 1:	movl $(__KERNEL_DS),%eax	# reload all the segment registers
 	movl %eax,%ss			# after changing gdt.
+	movl %eax,%fs			# gets reset once there's real percpu
 
 	movl $(__USER_DS),%eax		# DS/ES contains default USER segment
 	movl %eax,%ds
@@ -331,9 +331,6 @@ 1:	movl $(__KERNEL_DS),%eax	# reload all
 	xorl %eax,%eax			# Clear GS and LDT
 	movl %eax,%gs
 	lldt %ax
-
-	movl $(__KERNEL_PDA),%eax
-	mov  %eax,%fs
 
 	cld			# gcc2 wants the direction flag cleared at all times
 	pushl $0		# fake return address for unwinder
@@ -341,7 +338,11 @@ 1:	movl $(__KERNEL_DS),%eax	# reload all
 	movb ready, %cl
 	movb $1, ready
 	cmpb $0,%cl		# the first CPU calls start_kernel
-	jne initialize_secondary # all other CPUs call initialize_secondary
+	je   1f
+	movl $(__KERNEL_PERCPU), %eax
+	movl %eax,%fs		# set this cpu's percpu
+	jmp initialize_secondary # all other CPUs call initialize_secondary
+1:
 #endif /* CONFIG_SMP */
 	jmp start_kernel
 
@@ -362,23 +363,6 @@ check_x87:
 	ALIGN
 1:	movb $1,X86_HARD_MATH
 	.byte 0xDB,0xE4		/* fsetpm for 287, ignored by 387 */
-	ret
-
-/*
- * Point the GDT at this CPU's PDA.  On boot this will be
- * cpu_gdt_table and boot_pda; for secondary CPUs, these will be
- * that CPU's GDT and PDA.
- */
-ENTRY(setup_pda)
-	/* get the PDA pointer */
-	movl start_pda, %eax
-
-	/* slot the PDA address into the GDT */
-	mov early_gdt_descr+2, %ecx
-	mov %ax, (__KERNEL_PDA+0+2)(%ecx)		/* base & 0x0000ffff */
-	shr $16, %eax
-	mov %al, (__KERNEL_PDA+4+0)(%ecx)		/* base & 0x00ff0000 */
-	mov %ah, (__KERNEL_PDA+4+3)(%ecx)		/* base & 0xff000000 */
 	ret
 
 /*
@@ -553,9 +537,6 @@ ENTRY(empty_zero_page)
  * This starts the data section.
  */
 .data
-ENTRY(start_pda)
-	.long boot_pda
-
 ENTRY(stack_start)
 	.long init_thread_union+THREAD_SIZE
 	.long __BOOT_DS
===================================================================
--- a/arch/i386/kernel/i386_ksyms.c
+++ b/arch/i386/kernel/i386_ksyms.c
@@ -28,5 +28,3 @@ EXPORT_SYMBOL(__read_lock_failed);
 #endif
 
 EXPORT_SYMBOL(csum_partial);
-
-EXPORT_SYMBOL(_proxy_pda);
===================================================================
--- a/arch/i386/kernel/irq.c
+++ b/arch/i386/kernel/irq.c
@@ -23,6 +23,9 @@
 
 DEFINE_PER_CPU(irq_cpustat_t, irq_stat) ____cacheline_internodealigned_in_smp;
 EXPORT_PER_CPU_SYMBOL(irq_stat);
+
+DEFINE_PER_CPU(struct pt_regs *, irq_regs);
+EXPORT_PER_CPU_SYMBOL(irq_regs);
 
 /*
  * 'what should we do if we get a hw irq event on an illegal vector'.
===================================================================
--- a/arch/i386/kernel/process.c
+++ b/arch/i386/kernel/process.c
@@ -39,6 +39,7 @@
 #include <linux/random.h>
 #include <linux/personality.h>
 #include <linux/tick.h>
+#include <linux/percpu.h>
 
 #include <asm/uaccess.h>
 #include <asm/pgtable.h>
@@ -57,7 +58,6 @@
 
 #include <asm/tlbflush.h>
 #include <asm/cpu.h>
-#include <asm/pda.h>
 
 asmlinkage void ret_from_fork(void) __asm__("ret_from_fork");
 
@@ -65,6 +65,12 @@ static int hlt_counter;
 
 unsigned long boot_option_idle_override = 0;
 EXPORT_SYMBOL(boot_option_idle_override);
+
+DEFINE_PER_CPU(struct task_struct *, current_task) = &init_task;
+EXPORT_PER_CPU_SYMBOL(current_task);
+
+DEFINE_PER_CPU(int, cpu_number);
+EXPORT_PER_CPU_SYMBOL(cpu_number);
 
 /*
  * Return saved PC of a blocked thread.
@@ -343,7 +349,7 @@ int kernel_thread(int (*fn)(void *), voi
 
 	regs.xds = __USER_DS;
 	regs.xes = __USER_DS;
-	regs.xfs = __KERNEL_PDA;
+	regs.xfs = __KERNEL_PERCPU;
 	regs.orig_eax = -1;
 	regs.eip = (unsigned long) kernel_thread_helper;
 	regs.xcs = __KERNEL_CS | get_kernel_rpl();
@@ -712,7 +718,7 @@ struct task_struct fastcall * __switch_t
 	if (prev->gs | next->gs)
 		loadsegment(gs, next->gs);
 
-	write_pda(pcurrent, next_p);
+	x86_write_percpu(current_task, next_p);
 
 	return prev_p;
 }
===================================================================
--- a/arch/i386/kernel/smpboot.c
+++ b/arch/i386/kernel/smpboot.c
@@ -53,7 +53,6 @@
 #include <asm/desc.h>
 #include <asm/arch_hooks.h>
 #include <asm/nmi.h>
-#include <asm/pda.h>
 
 #include <mach_apic.h>
 #include <mach_wakecpu.h>
@@ -98,6 +97,9 @@ EXPORT_SYMBOL(x86_cpu_to_apicid);
 EXPORT_SYMBOL(x86_cpu_to_apicid);
 
 u8 apicid_2_node[MAX_APICID];
+
+DEFINE_PER_CPU(unsigned long, this_cpu_off);
+EXPORT_PER_CPU_SYMBOL(this_cpu_off);
 
 /*
  * Trampoline 80x86 program as an array.
@@ -456,7 +458,6 @@ extern struct {
 	void * esp;
 	unsigned short ss;
 } stack_start;
-extern struct i386_pda *start_pda;
 
 #ifdef CONFIG_NUMA
 
@@ -784,20 +785,17 @@ static inline struct task_struct * alloc
 /* Initialize the CPU's GDT.  This is either the boot CPU doing itself
    (still using the master per-cpu area), or a CPU doing it for a
    secondary which will soon come up. */
-static __cpuinit void init_gdt(int cpu, struct task_struct *idle)
+static __cpuinit void init_gdt(int cpu)
 {
 	struct desc_struct *gdt = get_cpu_gdt_table(cpu);
-	struct i386_pda *pda = &per_cpu(_cpu_pda, cpu);
-
-	pack_descriptor((u32 *)&gdt[GDT_ENTRY_PDA].a,
-			(u32 *)&gdt[GDT_ENTRY_PDA].b,
-			(unsigned long)pda, sizeof(*pda) - 1,
-			0x80 | DESCTYPE_S | 0x2, 0); /* present read-write data segment */
-
-	memset(pda, 0, sizeof(*pda));
-	pda->_pda = pda;
-	pda->cpu_number = cpu;
-	pda->pcurrent = idle;
+
+	pack_descriptor((u32 *)&gdt[GDT_ENTRY_PERCPU].a,
+			(u32 *)&gdt[GDT_ENTRY_PERCPU].b,
+			__per_cpu_offset[cpu], 0xFFFFF,
+			0x80 | DESCTYPE_S | 0x2, 0x8);
+
+	per_cpu(this_cpu_off, cpu) = __per_cpu_offset[cpu];
+	per_cpu(cpu_number, cpu) = cpu;
 }
 
 /* Defined in head.S */
@@ -824,9 +822,9 @@ static int __cpuinit do_boot_cpu(int api
 	if (IS_ERR(idle))
 		panic("failed fork for CPU %d", cpu);
 
-	init_gdt(cpu, idle);
+	init_gdt(cpu);
+ 	per_cpu(current_task, cpu) = idle;
 	early_gdt_descr.address = (unsigned long)get_cpu_gdt_table(cpu);
-	start_pda = cpu_pda(cpu);
 
 	idle->thread.eip = (unsigned long) start_secondary;
 	/* start_eip had better be page-aligned! */
@@ -1188,14 +1186,14 @@ static inline void switch_to_new_gdt(voi
 	gdt_descr.address = (long)get_cpu_gdt_table(smp_processor_id());
 	gdt_descr.size = GDT_SIZE - 1;
 	load_gdt(&gdt_descr);
-	asm volatile ("mov %0, %%fs" : : "r" (__KERNEL_PDA) : "memory");
+	asm("mov %0, %%fs" : : "r" (__KERNEL_PERCPU) : "memory");
 }
 
 void __init native_smp_prepare_boot_cpu(void)
 {
 	unsigned int cpu = smp_processor_id();
 
-	init_gdt(cpu, current);
+	init_gdt(cpu);
 	switch_to_new_gdt();
 
 	cpu_set(cpu, cpu_online_map);
===================================================================
--- a/arch/i386/kernel/vmi.c
+++ b/arch/i386/kernel/vmi.c
@@ -524,8 +524,6 @@ static void vmi_pmd_clear(pmd_t *pmd)
 #endif
 
 #ifdef CONFIG_SMP
-extern void setup_pda(void);
-
 static void __devinit
 vmi_startup_ipi_hook(int phys_apicid, unsigned long start_eip,
 		     unsigned long start_esp)
@@ -550,12 +548,10 @@ vmi_startup_ipi_hook(int phys_apicid, un
 
 	ap.ds = __USER_DS;
 	ap.es = __USER_DS;
-	ap.fs = __KERNEL_PDA;
+	ap.fs = __KERNEL_PERCPU;
 	ap.gs = 0;
 
 	ap.eflags = 0;
-
-	setup_pda();
 
 #ifdef CONFIG_X86_PAE
 	/* efer should match BSP efer. */
===================================================================
--- a/arch/i386/kernel/vmlinux.lds.S
+++ b/arch/i386/kernel/vmlinux.lds.S
@@ -26,7 +26,6 @@ OUTPUT_ARCH(i386)
 OUTPUT_ARCH(i386)
 ENTRY(phys_startup_32)
 jiffies = jiffies_64;
-_proxy_pda = 0;
 
 PHDRS {
 	text PT_LOAD FLAGS(5);	/* R_E */
===================================================================
--- a/include/asm-i386/current.h
+++ b/include/asm-i386/current.h
@@ -1,14 +1,15 @@
 #ifndef _I386_CURRENT_H
 #define _I386_CURRENT_H
 
-#include <asm/pda.h>
 #include <linux/compiler.h>
+#include <asm/percpu.h>
 
 struct task_struct;
 
+DECLARE_PER_CPU(struct task_struct *, current_task);
 static __always_inline struct task_struct *get_current(void)
 {
-	return read_pda(pcurrent);
+	return x86_read_percpu(current_task);
 }
  
 #define current get_current()
===================================================================
--- a/include/asm-i386/irq_regs.h
+++ b/include/asm-i386/irq_regs.h
@@ -1,25 +1,27 @@
 /*
  * Per-cpu current frame pointer - the location of the last exception frame on
- * the stack, stored in the PDA.
+ * the stack, stored in the per-cpu area.
  *
  * Jeremy Fitzhardinge <jeremy@goop.org>
  */
 #ifndef _ASM_I386_IRQ_REGS_H
 #define _ASM_I386_IRQ_REGS_H
 
-#include <asm/pda.h>
+#include <asm/percpu.h>
+
+DECLARE_PER_CPU(struct pt_regs *, irq_regs);
 
 static inline struct pt_regs *get_irq_regs(void)
 {
-	return read_pda(irq_regs);
+	return x86_read_percpu(irq_regs);
 }
 
 static inline struct pt_regs *set_irq_regs(struct pt_regs *new_regs)
 {
 	struct pt_regs *old_regs;
 
-	old_regs = read_pda(irq_regs);
-	write_pda(irq_regs, new_regs);
+	old_regs = get_irq_regs();
+	x86_write_percpu(irq_regs, new_regs);
 
 	return old_regs;
 }
===================================================================
--- a/include/asm-i386/pda.h
+++ /dev/null
@@ -1,99 +0,0 @@
-/*
-   Per-processor Data Areas
-   Jeremy Fitzhardinge <jeremy@goop.org> 2006
-   Based on asm-x86_64/pda.h by Andi Kleen.
- */
-#ifndef _I386_PDA_H
-#define _I386_PDA_H
-
-#include <linux/stddef.h>
-#include <linux/types.h>
-#include <asm/percpu.h>
-
-struct i386_pda
-{
-	struct i386_pda *_pda;		/* pointer to self */
-
-	int cpu_number;
-	struct task_struct *pcurrent;	/* current process */
-	struct pt_regs *irq_regs;
-};
-
-DECLARE_PER_CPU(struct i386_pda, _cpu_pda);
-#define cpu_pda(i)	(&per_cpu(_cpu_pda, (i)))
-#define pda_offset(field) offsetof(struct i386_pda, field)
-
-extern void __bad_pda_field(void);
-
-/* This variable is never instantiated.  It is only used as a stand-in
-   for the real per-cpu PDA memory, so that gcc can understand what
-   memory operations the inline asms() below are performing.  This
-   eliminates the need to make the asms volatile or have memory
-   clobbers, so gcc can readily analyse them. */
-extern struct i386_pda _proxy_pda;
-
-#define pda_to_op(op,field,val)						\
-	do {								\
-		typedef typeof(_proxy_pda.field) T__;			\
-		if (0) { T__ tmp__; tmp__ = (val); }			\
-		switch (sizeof(_proxy_pda.field)) {			\
-		case 1:							\
-			asm(op "b %1,%%fs:%c2"				\
-			    : "+m" (_proxy_pda.field)			\
-			    :"ri" ((T__)val),				\
-			     "i"(pda_offset(field)));			\
-			break;						\
-		case 2:							\
-			asm(op "w %1,%%fs:%c2"				\
-			    : "+m" (_proxy_pda.field)			\
-			    :"ri" ((T__)val),				\
-			     "i"(pda_offset(field)));			\
-			break;						\
-		case 4:							\
-			asm(op "l %1,%%fs:%c2"				\
-			    : "+m" (_proxy_pda.field)			\
-			    :"ri" ((T__)val),				\
-			     "i"(pda_offset(field)));			\
-			break;						\
-		default: __bad_pda_field();				\
-		}							\
-	} while (0)
-
-#define pda_from_op(op,field)						\
-	({								\
-		typeof(_proxy_pda.field) ret__;				\
-		switch (sizeof(_proxy_pda.field)) {			\
-		case 1:							\
-			asm(op "b %%fs:%c1,%0"				\
-			    : "=r" (ret__)				\
-			    : "i" (pda_offset(field)),			\
-			      "m" (_proxy_pda.field));			\
-			break;						\
-		case 2:							\
-			asm(op "w %%fs:%c1,%0"				\
-			    : "=r" (ret__)				\
-			    : "i" (pda_offset(field)),			\
-			      "m" (_proxy_pda.field));			\
-			break;						\
-		case 4:							\
-			asm(op "l %%fs:%c1,%0"				\
-			    : "=r" (ret__)				\
-			    : "i" (pda_offset(field)),			\
-			      "m" (_proxy_pda.field));			\
-			break;						\
-		default: __bad_pda_field();				\
-		}							\
-		ret__; })
-
-/* Return a pointer to a pda field */
-#define pda_addr(field)							\
-	((typeof(_proxy_pda.field) *)((unsigned char *)read_pda(_pda) + \
-				      pda_offset(field)))
-
-#define read_pda(field) pda_from_op("mov",field)
-#define write_pda(field,val) pda_to_op("mov",field,val)
-#define add_pda(field,val) pda_to_op("add",field,val)
-#define sub_pda(field,val) pda_to_op("sub",field,val)
-#define or_pda(field,val) pda_to_op("or",field,val)
-
-#endif	/* _I386_PDA_H */
===================================================================
--- a/include/asm-i386/percpu.h
+++ b/include/asm-i386/percpu.h
@@ -1,9 +1,30 @@
 #ifndef __ARCH_I386_PERCPU__
 #define __ARCH_I386_PERCPU__
 
-#ifndef __ASSEMBLY__
-#include <asm-generic/percpu.h>
-#else
+#ifdef __ASSEMBLY__
+
+/*
+ * PER_CPU finds an address of a per-cpu variable.
+ *
+ * Args:
+ *    var - variable name
+ *    reg - 32bit register
+ *
+ * The resulting address is stored in the "reg" argument.
+ *
+ * Example:
+ *    PER_CPU(cpu_gdt_descr, %ebx)
+ */
+#ifdef CONFIG_SMP
+#define PER_CPU(var, reg)			\
+	movl %fs:per_cpu__this_cpu_off, reg;		\
+	addl $per_cpu__##var, reg
+#else /* ! SMP */
+#define PER_CPU(var, reg) \
+	movl $per_cpu__##var, reg;
+#endif	/* SMP */
+
+#else /* ...!ASSEMBLY */
 
 /*
  * PER_CPU finds an address of a per-cpu variable.
@@ -18,14 +39,107 @@
  *    PER_CPU(cpu_gdt_descr, %ebx)
  */
 #ifdef CONFIG_SMP
-#define PER_CPU(var, cpu) \
-	movl __per_cpu_offset(,cpu,4), cpu;	\
-	addl $per_cpu__##var, cpu;
-#else /* ! SMP */
-#define PER_CPU(var, cpu) \
-	movl $per_cpu__##var, cpu;
+/* Same as generic implementation except for optimized local access. */
+#define __GENERIC_PER_CPU
+
+/* This is used for other cpus to find our section. */
+extern unsigned long __per_cpu_offset[];
+
+/* Separate out the type, so (int[3], foo) works. */
+#define DECLARE_PER_CPU(type, name) extern __typeof__(type) per_cpu__##name
+#define DEFINE_PER_CPU(type, name) \
+    __attribute__((__section__(".data.percpu"))) __typeof__(type) per_cpu__##name
+
+/* We can use this directly for local CPU (faster). */
+DECLARE_PER_CPU(unsigned long, this_cpu_off);
+
+/* var is in discarded region: offset to particular copy we want */
+#define per_cpu(var, cpu) (*({				\
+	extern int simple_indentifier_##var(void);	\
+	RELOC_HIDE(&per_cpu__##var, __per_cpu_offset[cpu]); }))
+
+#define __raw_get_cpu_var(var) (*({					\
+	extern int simple_indentifier_##var(void);			\
+	RELOC_HIDE(&per_cpu__##var, x86_read_percpu(this_cpu_off));	\
+}))
+
+#define __get_cpu_var(var) __raw_get_cpu_var(var)
+
+/* A macro to avoid #include hell... */
+#define percpu_modcopy(pcpudst, src, size)			\
+do {								\
+	unsigned int __i;					\
+	for_each_possible_cpu(__i)				\
+		memcpy((pcpudst)+__per_cpu_offset[__i],		\
+		       (src), (size));				\
+} while (0)
+
+#define EXPORT_PER_CPU_SYMBOL(var) EXPORT_SYMBOL(per_cpu__##var)
+#define EXPORT_PER_CPU_SYMBOL_GPL(var) EXPORT_SYMBOL_GPL(per_cpu__##var)
+
+/* fs segment starts at (positive) offset == __per_cpu_offset[cpu] */
+#define __percpu_seg "%%fs:"
+#else  /* !SMP */
+#include <asm-generic/percpu.h>
+#define __percpu_seg ""
 #endif	/* SMP */
 
+/* For arch-specific code, we can use direct single-insn ops (they
+ * don't give an lvalue though). */
+extern void __bad_percpu_size(void);
+
+#define percpu_to_op(op,var,val)				\
+	do {							\
+		typedef typeof(var) T__;			\
+		if (0) { T__ tmp__; tmp__ = (val); }		\
+		switch (sizeof(var)) {				\
+		case 1:						\
+			asm(op "b %1,"__percpu_seg"%0"		\
+			    : "+m" (var)			\
+			    :"ri" ((T__)val));			\
+			break;					\
+		case 2:						\
+			asm(op "w %1,"__percpu_seg"%0"		\
+			    : "+m" (var)			\
+			    :"ri" ((T__)val));			\
+			break;					\
+		case 4:						\
+			asm(op "l %1,"__percpu_seg"%0"		\
+			    : "+m" (var)			\
+			    :"ri" ((T__)val));			\
+			break;					\
+		default: __bad_percpu_size();			\
+		}						\
+	} while (0)
+
+#define percpu_from_op(op,var)					\
+	({							\
+		typeof(var) ret__;				\
+		switch (sizeof(var)) {				\
+		case 1:						\
+			asm(op "b "__percpu_seg"%1,%0"		\
+			    : "=r" (ret__)			\
+			    : "m" (var));			\
+			break;					\
+		case 2:						\
+			asm(op "w "__percpu_seg"%1,%0"		\
+			    : "=r" (ret__)			\
+			    : "m" (var));			\
+			break;					\
+		case 4:						\
+			asm(op "l "__percpu_seg"%1,%0"		\
+			    : "=r" (ret__)			\
+			    : "m" (var));			\
+			break;					\
+		default: __bad_percpu_size();			\
+		}						\
+		ret__; })
+
+#define x86_read_percpu(var) percpu_from_op("mov", per_cpu__##var)
+#define x86_write_percpu(var,val) percpu_to_op("mov", per_cpu__##var, val)
+#define x86_add_percpu(var,val) percpu_to_op("add", per_cpu__##var, val)
+#define x86_sub_percpu(var,val) percpu_to_op("sub", per_cpu__##var, val)
+#define x86_or_percpu(var,val) percpu_to_op("or", per_cpu__##var, val)
 #endif /* !__ASSEMBLY__ */
 
 #endif /* __ARCH_I386_PERCPU__ */
===================================================================
--- a/include/asm-i386/processor.h
+++ b/include/asm-i386/processor.h
@@ -425,7 +425,7 @@ struct thread_struct {
 	.vm86_info = NULL,						\
 	.sysenter_cs = __KERNEL_CS,					\
 	.io_bitmap_ptr = NULL,						\
-	.fs = __KERNEL_PDA,						\
+	.fs = __KERNEL_PERCPU,						\
 }
 
 /*
===================================================================
--- a/include/asm-i386/segment.h
+++ b/include/asm-i386/segment.h
@@ -39,7 +39,7 @@
  *  25 - APM BIOS support 
  *
  *  26 - ESPFIX small SS
- *  27 - PDA				[ per-cpu private data area ]
+ *  27 - per-cpu			[ offset to per-cpu data area ]
  *  28 - unused
  *  29 - unused
  *  30 - unused
@@ -74,8 +74,8 @@
 #define GDT_ENTRY_ESPFIX_SS		(GDT_ENTRY_KERNEL_BASE + 14)
 #define __ESPFIX_SS (GDT_ENTRY_ESPFIX_SS * 8)
 
-#define GDT_ENTRY_PDA			(GDT_ENTRY_KERNEL_BASE + 15)
-#define __KERNEL_PDA (GDT_ENTRY_PDA * 8)
+#define GDT_ENTRY_PERCPU			(GDT_ENTRY_KERNEL_BASE + 15)
+#define __KERNEL_PERCPU (GDT_ENTRY_PERCPU * 8)
 
 #define GDT_ENTRY_DOUBLEFAULT_TSS	31
 
===================================================================
--- a/include/asm-i386/smp.h
+++ b/include/asm-i386/smp.h
@@ -8,7 +8,6 @@
 #include <linux/kernel.h>
 #include <linux/threads.h>
 #include <linux/cpumask.h>
-#include <asm/pda.h>
 #endif
 
 #if defined(CONFIG_X86_LOCAL_APIC) && !defined(__ASSEMBLY__)
@@ -112,7 +111,8 @@ do { } while (0)
  * from the initial startup. We map APIC_BASE very early in page_setup(),
  * so this is correct in the x86 case.
  */
-#define raw_smp_processor_id() (read_pda(cpu_number))
+DECLARE_PER_CPU(int, cpu_number);
+#define raw_smp_processor_id() (x86_read_percpu(cpu_number))
 
 extern cpumask_t cpu_callout_map;
 extern cpumask_t cpu_callin_map;

-- 


^ permalink raw reply	[flat|nested] 13+ messages in thread

* Re: [patch 4/6] Convert PDA into the percpu section
  2007-03-31  2:00 ` [patch 4/6] Convert PDA into the percpu section Jeremy Fitzhardinge
@ 2007-03-31  2:35   ` Andi Kleen
  2007-03-31  2:54     ` Rusty Russell
  2007-03-31  3:00     ` Jeremy Fitzhardinge
  0 siblings, 2 replies; 13+ messages in thread
From: Andi Kleen @ 2007-03-31  2:35 UTC (permalink / raw)
  To: Jeremy Fitzhardinge; +Cc: Andrew Morton, virtualization, lkml, Rusty Russell

On Saturday 31 March 2007 04:00, Jeremy Fitzhardinge wrote:
> Currently x86 (similar to x84-64) has a special per-cpu structure
> called "i386_pda" which can be easily and efficiently referenced via
> the %fs register.  An ELF section is more flexible than a structure,
> allowing any piece of code to use this area.  Indeed, such a section
> already exists: the per-cpu area.

Hmm, I'm a little reluctant. This moves i386 more away from x86-64
again. If we ever merge them it would mean more work. Do you really need it?



-Andi

^ permalink raw reply	[flat|nested] 13+ messages in thread

* Re: [patch 4/6] Convert PDA into the percpu section
  2007-03-31  2:35   ` Andi Kleen
@ 2007-03-31  2:54     ` Rusty Russell
  2007-03-31  3:00     ` Jeremy Fitzhardinge
  1 sibling, 0 replies; 13+ messages in thread
From: Rusty Russell @ 2007-03-31  2:54 UTC (permalink / raw)
  To: Andi Kleen; +Cc: Jeremy Fitzhardinge, Andrew Morton, virtualization, lkml

On Sat, 2007-03-31 at 04:35 +0200, Andi Kleen wrote:
> On Saturday 31 March 2007 04:00, Jeremy Fitzhardinge wrote:
> > Currently x86 (similar to x84-64) has a special per-cpu structure
> > called "i386_pda" which can be easily and efficiently referenced via
> > the %fs register.  An ELF section is more flexible than a structure,
> > allowing any piece of code to use this area.  Indeed, such a section
> > already exists: the per-cpu area.
> 
> Hmm, I'm a little reluctant. This moves i386 more away from x86-64
> again. If we ever merge them it would mean more work. Do you really need it?

Well, I think the merge should go the other way in this case: this
really does simplify things.

The only thing stopping x86-64 from doing the same as i386 is the
stack-protector stuff.  And that can be fixed (unfortunately requires a
gcc patch to change the %gs:40 to %gs:__gcc_stack_protector_offset and
emit a weak absolute symbol __gss_stack_protector_offset = 40).

I shall prepare a patch for that next week; I've been busy Kleening up
lguest 8)

Rusty.



^ permalink raw reply	[flat|nested] 13+ messages in thread

* Re: [patch 4/6] Convert PDA into the percpu section
  2007-03-31  2:35   ` Andi Kleen
  2007-03-31  2:54     ` Rusty Russell
@ 2007-03-31  3:00     ` Jeremy Fitzhardinge
  1 sibling, 0 replies; 13+ messages in thread
From: Jeremy Fitzhardinge @ 2007-03-31  3:00 UTC (permalink / raw)
  To: Andi Kleen
  Cc: Jeremy Fitzhardinge, virtualization, Andrew Morton, lkml, Rusty Russell

Andi Kleen wrote:
> On Saturday 31 March 2007 04:00, Jeremy Fitzhardinge wrote:
>   
>> Currently x86 (similar to x84-64) has a special per-cpu structure
>> called "i386_pda" which can be easily and efficiently referenced via
>> the %fs register.  An ELF section is more flexible than a structure,
>> allowing any piece of code to use this area.  Indeed, such a section
>> already exists: the per-cpu area.
>>     
>
> Hmm, I'm a little reluctant. This moves i386 more away from x86-64
> again. If we ever merge them it would mean more work. Do you really need it?

It cleans things up a fair bit:

   1. At initialization, it doesn't require %fs to be loaded before
      being able to use per-cpu variables, since you can use percpu with
      %fs set to a plain 0-based 4G segment; you can defer
      initialization until SMP bringup (which is never on a UP kernel).
      PDA requires %fs to be specially set up to point to an initial
      PDA, which includes setting up a gdt entry, generally before C
      code is run.  For paravirtualized boot, this setup needs to be
      replicated by each hypervisor startup sequence; without the PDA,
      it becomes a non-issue (especially since hypervisors typically
      start up with %fs as a flat segment anyway).  Overall, both UP and
      SMP boot is simpler and less fragile.
   2. Adding things to the pda requires changing <asm/pda.h>, which
      often means including extra headers to allow added definitions. 
      Since pda.h is used to implement things like "current" and
      "smp_processor_id", it gets included everywhere.  Any header
      included in <asm/pda.h> effectively gets included everywhere in
      the kernel.  Also, it turns pda.h into a concentrated nest of
      patch conflicts.  percpu requires no central modifications to add
      a new percpu variable.
   3. There's no disadvantage to using a percpu at all, especially if
      you can use the x86_*_percpu functions which allow direct access
      to the variable via %fs.  If one construct will do, why have two? 
      Removing the pda removes quite a bit of unnecessary code.
   4. I think, ultimately, it would be better to migrate x86_64 away
      from using the pda to all percpu too, though this has some tricky
      bits for now.

Certainly, not having this patch at this stage will require me to rework
quite a few of the later patches.  I was going to put off sending out
this patch until later, but reworking everything to work both with pda
and percpu was so fragile and tricky-bug-prone that I decided to push it
early and save myself a lot of work.

    J

^ permalink raw reply	[flat|nested] 13+ messages in thread

* [patch 5/6] cleanups to help using per-cpu variables from asm
  2007-03-31  2:00 [patch 0/6] i386 gdt and percpu cleanups Jeremy Fitzhardinge
                   ` (3 preceding siblings ...)
  2007-03-31  2:00 ` [patch 4/6] Convert PDA into the percpu section Jeremy Fitzhardinge
@ 2007-03-31  2:00 ` Jeremy Fitzhardinge
  2007-03-31  2:00 ` [patch 6/6] Define per_cpu_offset Jeremy Fitzhardinge
                   ` (2 subsequent siblings)
  7 siblings, 0 replies; 13+ messages in thread
From: Jeremy Fitzhardinge @ 2007-03-31  2:00 UTC (permalink / raw)
  To: Andi Kleen; +Cc: Andrew Morton, virtualization, lkml, Rusty Russell

[-- Attachment #1: i386-cleanups-to-help-using-per-cpu-variables-from-asm.patch --]
[-- Type: text/plain, Size: 1209 bytes --]

This patch does a few small cleanups:
 - use PER_CPU_NAME to generate the names of per-cpu variables
 - use lea to add the per_cpu offset in PER_CPU(), because it doesn't
   affect condition flags
 - add PER_CPU_VAR which allows direct access to pre-cpu variables
   with the %fs: prefix on SMP.

Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: Andi Kleen <ak@suse.de>

---
 include/asm-i386/percpu.h |   12 +++++++-----
 1 file changed, 7 insertions(+), 5 deletions(-)

===================================================================
--- a/include/asm-i386/percpu.h
+++ b/include/asm-i386/percpu.h
@@ -16,12 +16,14 @@
  *    PER_CPU(cpu_gdt_descr, %ebx)
  */
 #ifdef CONFIG_SMP
+#define PER_CPU(var, reg)				\
+	movl %fs:per_cpu__##this_cpu_off, reg;		\
+	lea per_cpu__##var(reg), reg
+#define PER_CPU_VAR(var)	%fs:per_cpu__##var
+#else /* ! SMP */
 #define PER_CPU(var, reg)			\
-	movl %fs:per_cpu__this_cpu_off, reg;		\
-	addl $per_cpu__##var, reg
-#else /* ! SMP */
-#define PER_CPU(var, reg) \
-	movl $per_cpu__##var, reg;
+	movl $per_cpu__##var, reg
+#define PER_CPU_VAR(var)	per_cpu__##var
 #endif	/* SMP */
 
 #else /* ...!ASSEMBLY */

-- 


^ permalink raw reply	[flat|nested] 13+ messages in thread

* [patch 6/6] Define per_cpu_offset
  2007-03-31  2:00 [patch 0/6] i386 gdt and percpu cleanups Jeremy Fitzhardinge
                   ` (4 preceding siblings ...)
  2007-03-31  2:00 ` [patch 5/6] cleanups to help using per-cpu variables from asm Jeremy Fitzhardinge
@ 2007-03-31  2:00 ` Jeremy Fitzhardinge
  2007-03-31  2:56 ` [patch 0/6] i386 gdt and percpu cleanups Rusty Russell
  2007-03-31  3:16 ` Jeremy Fitzhardinge
  7 siblings, 0 replies; 13+ messages in thread
From: Jeremy Fitzhardinge @ 2007-03-31  2:00 UTC (permalink / raw)
  To: Andi Kleen; +Cc: Andrew Morton, virtualization, lkml, Rusty Russell

[-- Attachment #1: percpu-define-per_cpu_offset.patch --]
[-- Type: text/plain, Size: 745 bytes --]

Define per_cpu_offset in asm-i386/percpu.h when SMP defined, like
asm-generic/percpu.h does for UP.

Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: Andi Kleen <ak@suse.de>

---
 include/asm-i386/percpu.h |    2 ++
 1 file changed, 2 insertions(+)

===================================================================
--- a/include/asm-i386/percpu.h
+++ b/include/asm-i386/percpu.h
@@ -34,6 +34,8 @@
 
 /* This is used for other cpus to find our section. */
 extern unsigned long __per_cpu_offset[];
+
+#define per_cpu_offset(x) (__per_cpu_offset[x])
 
 /* Separate out the type, so (int[3], foo) works. */
 #define DECLARE_PER_CPU(type, name) extern __typeof__(type) per_cpu__##name

-- 


^ permalink raw reply	[flat|nested] 13+ messages in thread

* Re: [patch 0/6] i386 gdt and percpu cleanups
  2007-03-31  2:00 [patch 0/6] i386 gdt and percpu cleanups Jeremy Fitzhardinge
                   ` (5 preceding siblings ...)
  2007-03-31  2:00 ` [patch 6/6] Define per_cpu_offset Jeremy Fitzhardinge
@ 2007-03-31  2:56 ` Rusty Russell
  2007-03-31  3:05   ` Jeremy Fitzhardinge
  2007-03-31  3:16 ` Jeremy Fitzhardinge
  7 siblings, 1 reply; 13+ messages in thread
From: Rusty Russell @ 2007-03-31  2:56 UTC (permalink / raw)
  To: Jeremy Fitzhardinge; +Cc: Andi Kleen, virtualization, Andrew Morton, lkml

>  - some improvements to asm-i386/percpu.h to make asm access to percpu
>    variables easy

One nitpick: I'd really like PER_CPU() renamed to PER_CPU_ADDR().
That's a separate patch, but I think would be far clearer.

Thanks,
Rusty.



^ permalink raw reply	[flat|nested] 13+ messages in thread

* Re: [patch 0/6] i386 gdt and percpu cleanups
  2007-03-31  2:56 ` [patch 0/6] i386 gdt and percpu cleanups Rusty Russell
@ 2007-03-31  3:05   ` Jeremy Fitzhardinge
  0 siblings, 0 replies; 13+ messages in thread
From: Jeremy Fitzhardinge @ 2007-03-31  3:05 UTC (permalink / raw)
  To: Rusty Russell; +Cc: Andi Kleen, virtualization, Andrew Morton, lkml

Rusty Russell wrote:
> One nitpick: I'd really like PER_CPU() renamed to PER_CPU_ADDR().
> That's a separate patch, but I think would be far clearer.
>   

Seems pretty simple, given that it has precisely one use.

    J

^ permalink raw reply	[flat|nested] 13+ messages in thread

* Re: [patch 0/6] i386 gdt and percpu cleanups
  2007-03-31  2:00 [patch 0/6] i386 gdt and percpu cleanups Jeremy Fitzhardinge
                   ` (6 preceding siblings ...)
  2007-03-31  2:56 ` [patch 0/6] i386 gdt and percpu cleanups Rusty Russell
@ 2007-03-31  3:16 ` Jeremy Fitzhardinge
  7 siblings, 0 replies; 13+ messages in thread
From: Jeremy Fitzhardinge @ 2007-03-31  3:16 UTC (permalink / raw)
  To: Jeremy Fitzhardinge
  Cc: Andi Kleen, virtualization, Andrew Morton, lkml, Rusty Russell

Jeremy Fitzhardinge wrote:
> This is a series of patches based on your latest queue (as of the
> other day, at least).
>   

BTW, the From: line attributions got dropped from a few of these.  These:

    Allow percpu variables to be page-aligned
    Page-align the GDT

should be From: Rusty.  He did most of the work on the others, but I
changed them enough that he shouldn't be saddled with the blame when
things break ;)

    J

^ permalink raw reply	[flat|nested] 13+ messages in thread

end of thread, other threads:[~2007-03-31  4:55 UTC | newest]

Thread overview: 13+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2007-03-31  2:00 [patch 0/6] i386 gdt and percpu cleanups Jeremy Fitzhardinge
2007-03-31  2:00 ` [patch 1/6] i386: Account for module percpu space separately from kernel percpu Jeremy Fitzhardinge
2007-03-31  2:00 ` [patch 2/6] Allow percpu variables to be page-aligned Jeremy Fitzhardinge
2007-03-31  2:00 ` [patch 3/6] Page-align the GDT Jeremy Fitzhardinge
2007-03-31  2:00 ` [patch 4/6] Convert PDA into the percpu section Jeremy Fitzhardinge
2007-03-31  2:35   ` Andi Kleen
2007-03-31  2:54     ` Rusty Russell
2007-03-31  3:00     ` Jeremy Fitzhardinge
2007-03-31  2:00 ` [patch 5/6] cleanups to help using per-cpu variables from asm Jeremy Fitzhardinge
2007-03-31  2:00 ` [patch 6/6] Define per_cpu_offset Jeremy Fitzhardinge
2007-03-31  2:56 ` [patch 0/6] i386 gdt and percpu cleanups Rusty Russell
2007-03-31  3:05   ` Jeremy Fitzhardinge
2007-03-31  3:16 ` Jeremy Fitzhardinge

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).