[PATCH] Allow per-cpu variables to be page-aligned

From: Rusty Russell
Date: Wed Mar 21 2007 - 02:11:05 EST

Next message: Eric Dumazet: "Re: [RFC] SLAB : NUMA cache_free_alien() very expensive because ofvirt_to_slab(objp); nodeid = slabp->nodeid;"
Previous message: Andrew Morton: "Re: [patch 13/26] Xen-paravirt_ops: Consistently wrap paravirt opscallsites to make them patchable"
Next in thread: Rusty Russell: "[PATCH 0/4] i386 GDT cleanups"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

[This was part of the GDT cleanups and per-cpu-> pda changes, which I
have revised, but this stands on its own. The only change is catching
the x86-64 per-cpu allocator too].
==
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@xxxxxxxxxxxxxxx>
Acked-by: Ingo Molnar <mingo@xxxxxxx>
---
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 +-
include/linux/percpu.h | 2 +-
init/main.c | 4 ++--
kernel/module.c | 10 +++++-----
22 files changed, 31 insertions(+), 33 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
@@ -195,7 +195,7 @@ 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/include/linux/percpu.h
+++ b/include/linux/percpu.h
@@ -11,7 +11,7 @@

/* Enough to cover all DEFINE_PER_CPUs in kernel, including modules. */
#ifndef PERCPU_ENOUGH_ROOM
-#define PERCPU_ENOUGH_ROOM 32768
+#define PERCPU_ENOUGH_ROOM 65536
#endif

/*
===================================================================
--- a/init/main.c
+++ b/init/main.c
@@ -369,12 +369,12 @@ 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);
+ size = ALIGN(__per_cpu_end - __per_cpu_start, PAGE_SIZE);
#ifdef CONFIG_MODULES
if (size < PERCPU_ENOUGH_ROOM)
size = PERCPU_ENOUGH_ROOM;
#endif
- ptr = alloc_bootmem(size * nr_possible_cpus);
+ 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;
@@ -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] = -ALIGN(__per_cpu_end-__per_cpu_start, PAGE_SIZE);
/* Free room. */
pcpu_size[1] = PERCPU_ENOUGH_ROOM + pcpu_size[0];
if (pcpu_size[1] < 0) {

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Next message: Eric Dumazet: "Re: [RFC] SLAB : NUMA cache_free_alien() very expensive because ofvirt_to_slab(objp); nodeid = slabp->nodeid;"
Previous message: Andrew Morton: "Re: [patch 13/26] Xen-paravirt_ops: Consistently wrap paravirt opscallsites to make them patchable"
Next in thread: Rusty Russell: "[PATCH 0/4] i386 GDT cleanups"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]