Merge branch 'tj-percpu' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/misc into core/percpu

18 files changed, 107 insertions, 212 deletions

diff --git a/arch/x86/include/asm/pda.h b/arch/x86/include/asm/pda.h
deleted file mode 100644
index 5976cd803e9..00000000000
--- a/arch/x86/include/asm/pda.h
+++ /dev/null
@@ -1,45 +0,0 @@
-#ifndef _ASM_X86_PDA_H
-#define _ASM_X86_PDA_H
-
-#ifndef __ASSEMBLY__
-#include <linux/stddef.h>
-#include <linux/types.h>
-#include <linux/cache.h>
-#include <linux/threads.h>
-#include <asm/page.h>
-#include <asm/percpu.h>
-
-/* Per processor datastructure. %gs points to it while the kernel runs */
-struct x8664_pda {
-	unsigned long unused1;
-	unsigned long unused2;
-	unsigned long unused3;
-	unsigned long unused4;
-	int unused5;
-	unsigned int unused6;		/* 36 was cpunumber */
-	unsigned long stack_canary;	/* 40 stack canary value */
-					/* gcc-ABI: this canary MUST be at
-					   offset 40!!! */
-	short in_bootmem;		/* pda lives in bootmem */
-} ____cacheline_aligned_in_smp;
-
-DECLARE_PER_CPU(struct x8664_pda, __pda);
-extern void pda_init(int);
-
-#define cpu_pda(cpu)		(&per_cpu(__pda, cpu))
-
-#define read_pda(field)		percpu_read(__pda.field)
-#define write_pda(field, val)	percpu_write(__pda.field, val)
-#define add_pda(field, val)	percpu_add(__pda.field, val)
-#define sub_pda(field, val)	percpu_sub(__pda.field, val)
-#define or_pda(field, val)	percpu_or(__pda.field, val)
-
-/* This is not atomic against other CPUs -- CPU preemption needs to be off */
-#define test_and_clear_bit_pda(bit, field)				\
-	x86_test_and_clear_bit_percpu(bit, __pda.field)
-
-#endif
-
-#define refresh_stack_canary() write_pda(stack_canary, current->stack_canary)
-
-#endif /* _ASM_X86_PDA_H */
diff --git a/arch/x86/include/asm/percpu.h b/arch/x86/include/asm/percpu.h
index 165d5272ece..ce980db5e59 100644
--- a/arch/x86/include/asm/percpu.h
+++ b/arch/x86/include/asm/percpu.h
@@ -133,12 +133,6 @@ do {							\
 /* We can use this directly for local CPU (faster). */
 DECLARE_PER_CPU(unsigned long, this_cpu_off);
 
-#ifdef CONFIG_X86_64
-extern void load_pda_offset(int cpu);
-#else
-static inline void load_pda_offset(int cpu) { }
-#endif
-
 #endif /* !__ASSEMBLY__ */
 
 #ifdef CONFIG_SMP
diff --git a/arch/x86/include/asm/pgtable_64.h b/arch/x86/include/asm/pgtable_64.h
index ba09289acca..1df9637dfda 100644
--- a/arch/x86/include/asm/pgtable_64.h
+++ b/arch/x86/include/asm/pgtable_64.h
@@ -11,7 +11,6 @@
 #include <asm/processor.h>
 #include <linux/bitops.h>
 #include <linux/threads.h>
-#include <asm/pda.h>
 
 extern pud_t level3_kernel_pgt[512];
 extern pud_t level3_ident_pgt[512];
diff --git a/arch/x86/include/asm/processor.h b/arch/x86/include/asm/processor.h
index f511246fa6c..48676b943b9 100644
--- a/arch/x86/include/asm/processor.h
+++ b/arch/x86/include/asm/processor.h
@@ -379,8 +379,29 @@ union thread_xstate {
 #ifdef CONFIG_X86_64
 DECLARE_PER_CPU(struct orig_ist, orig_ist);
 
-DECLARE_PER_CPU(char[IRQ_STACK_SIZE], irq_stack);
+union irq_stack_union {
+	char irq_stack[IRQ_STACK_SIZE];
+	/*
+	 * GCC hardcodes the stack canary as %gs:40.  Since the
+	 * irq_stack is the object at %gs:0, we reserve the bottom
+	 * 48 bytes of the irq stack for the canary.
+	 */
+	struct {
+		char gs_base[40];
+		unsigned long stack_canary;
+	};
+};
+
+DECLARE_PER_CPU(union irq_stack_union, irq_stack_union);
 DECLARE_PER_CPU(char *, irq_stack_ptr);
+
+static inline void load_gs_base(int cpu)
+{
+	/* Memory clobbers used to order pda/percpu accesses */
+	mb();
+	wrmsrl(MSR_GS_BASE, (unsigned long)per_cpu(irq_stack_union.gs_base, cpu));
+	mb();
+}
 #endif
 
 extern void print_cpu_info(struct cpuinfo_x86 *);
diff --git a/arch/x86/include/asm/smp.h b/arch/x86/include/asm/smp.h
index 68636e767a9..45ef8a1b9d7 100644
--- a/arch/x86/include/asm/smp.h
+++ b/arch/x86/include/asm/smp.h
@@ -15,7 +15,6 @@
 #  include <asm/io_apic.h>
 # endif
 #endif
-#include <asm/pda.h>
 #include <asm/thread_info.h>
 #include <asm/cpumask.h>
 
diff --git a/arch/x86/include/asm/stackprotector.h b/arch/x86/include/asm/stackprotector.h
index c7f0d10bae7..36a700acaf2 100644
--- a/arch/x86/include/asm/stackprotector.h
+++ b/arch/x86/include/asm/stackprotector.h
@@ -2,7 +2,7 @@
 #define _ASM_STACKPROTECTOR_H 1
 
 #include <asm/tsc.h>
-#include <asm/pda.h>
+#include <asm/processor.h>
 
 /*
  * Initialize the stackprotector canary value.
@@ -16,13 +16,12 @@ static __always_inline void boot_init_stack_canary(void)
 	u64 tsc;
 
 	/*
-	 * If we're the non-boot CPU, nothing set the PDA stack
-	 * canary up for us - and if we are the boot CPU we have
-	 * a 0 stack canary. This is a good place for updating
-	 * it, as we wont ever return from this function (so the
-	 * invalid canaries already on the stack wont ever
-	 * trigger).
-	 *
+	 * Build time only check to make sure the stack_canary is at
+	 * offset 40 in the pda; this is a gcc ABI requirement
+	 */
+	BUILD_BUG_ON(offsetof(union irq_stack_union, stack_canary) != 40);
+
+	/*
 	 * We both use the random pool and the current TSC as a source
 	 * of randomness. The TSC only matters for very early init,
 	 * there it already has some randomness on most systems. Later
@@ -33,7 +32,7 @@ static __always_inline void boot_init_stack_canary(void)
 	canary += tsc + (tsc << 32UL);
 
 	current->stack_canary = canary;
-	write_pda(stack_canary, canary);
+	percpu_write(irq_stack_union.stack_canary, canary);
 }
 
 #endif
diff --git a/arch/x86/include/asm/system.h b/arch/x86/include/asm/system.h
index 8cadfe9b119..52eb748a68a 100644
--- a/arch/x86/include/asm/system.h
+++ b/arch/x86/include/asm/system.h
@@ -86,17 +86,28 @@ do {									\
 	, "rcx", "rbx", "rdx", "r8", "r9", "r10", "r11", \
 	  "r12", "r13", "r14", "r15"
 
+#ifdef CONFIG_CC_STACKPROTECTOR
+#define __switch_canary							  \
+	"movq %P[task_canary](%%rsi),%%r8\n\t"				  \
+	"movq %%r8,%%gs:%P[gs_canary]\n\t"
+#define __switch_canary_param						  \
+	, [task_canary] "i" (offsetof(struct task_struct, stack_canary))  \
+	, [gs_canary] "i" (offsetof(union irq_stack_union, stack_canary))
+#else	/* CC_STACKPROTECTOR */
+#define __switch_canary
+#define __switch_canary_param
+#endif	/* CC_STACKPROTECTOR */
+
 /* Save restore flags to clear handle leaking NT */
 #define switch_to(prev, next, last) \
-	asm volatile(SAVE_CONTEXT						    \
+	asm volatile(SAVE_CONTEXT					  \
 	     "movq %%rsp,%P[threadrsp](%[prev])\n\t" /* save RSP */	  \
 	     "movq %P[threadrsp](%[next]),%%rsp\n\t" /* restore RSP */	  \
 	     "call __switch_to\n\t"					  \
 	     ".globl thread_return\n"					  \
 	     "thread_return:\n\t"					  \
 	     "movq "__percpu_arg([current_task])",%%rsi\n\t"		  \
-	     "movq %P[task_canary](%%rsi),%%r8\n\t"			  \
-	     "movq %%r8,%%gs:%P[pda_canary]\n\t"			  \
+	     __switch_canary						  \
 	     "movq %P[thread_info](%%rsi),%%r8\n\t"			  \
 	     LOCK_PREFIX "btr  %[tif_fork],%P[ti_flags](%%r8)\n\t"	  \
 	     "movq %%rax,%%rdi\n\t" 					  \
@@ -108,9 +119,8 @@ do {									\
 	       [ti_flags] "i" (offsetof(struct thread_info, flags)),	  \
 	       [tif_fork] "i" (TIF_FORK),			  	  \
 	       [thread_info] "i" (offsetof(struct task_struct, stack)),   \
-	       [task_canary] "i" (offsetof(struct task_struct, stack_canary)),\
-	       [current_task] "m" (per_cpu_var(current_task)),		  \
-	       [pda_canary] "i" (offsetof(struct x8664_pda, stack_canary))\
+	       [current_task] "m" (per_cpu_var(current_task))		  \
+	       __switch_canary_param					  \
 	     : "memory", "cc" __EXTRA_CLOBBER)
 #endif
 
diff --git a/arch/x86/kernel/asm-offsets_64.c b/arch/x86/kernel/asm-offsets_64.c
index 64c834a39aa..8793ab33e2c 100644
--- a/arch/x86/kernel/asm-offsets_64.c
+++ b/arch/x86/kernel/asm-offsets_64.c
@@ -11,7 +11,6 @@
 #include <linux/hardirq.h>
 #include <linux/suspend.h>
 #include <linux/kbuild.h>
-#include <asm/pda.h>
 #include <asm/processor.h>
 #include <asm/segment.h>
 #include <asm/thread_info.h>
@@ -48,10 +47,6 @@ int main(void)
 #endif
 	BLANK();
 #undef ENTRY
-#define ENTRY(entry) DEFINE(pda_ ## entry, offsetof(struct x8664_pda, entry))
-	DEFINE(pda_size, sizeof(struct x8664_pda));
-	BLANK();
-#undef ENTRY
 #ifdef CONFIG_PARAVIRT
 	BLANK();
 	OFFSET(PARAVIRT_enabled, pv_info, paravirt_enabled);
diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c
index 7976a6a0f65..3887fcf6e51 100644
--- a/arch/x86/kernel/cpu/common.c
+++ b/arch/x86/kernel/cpu/common.c
@@ -30,7 +30,6 @@
 #include <asm/genapic.h>
 #endif
 
-#include <asm/pda.h>
 #include <asm/pgtable.h>
 #include <asm/processor.h>
 #include <asm/desc.h>
@@ -881,12 +880,13 @@ __setup("clearcpuid=", setup_disablecpuid);
 #ifdef CONFIG_X86_64
 struct desc_ptr idt_descr = { 256 * 16 - 1, (unsigned long) idt_table };
 
-DEFINE_PER_CPU_PAGE_ALIGNED(char[IRQ_STACK_SIZE], irq_stack);
+DEFINE_PER_CPU_FIRST(union irq_stack_union,
+		     irq_stack_union) __aligned(PAGE_SIZE);
 #ifdef CONFIG_SMP
 DEFINE_PER_CPU(char *, irq_stack_ptr);	/* will be set during per cpu init */
 #else
 DEFINE_PER_CPU(char *, irq_stack_ptr) =
-	per_cpu_var(irq_stack) + IRQ_STACK_SIZE - 64;
+	per_cpu_var(irq_stack_union.irq_stack) + IRQ_STACK_SIZE - 64;
 #endif
 
 DEFINE_PER_CPU(unsigned long, kernel_stack) =
@@ -895,15 +895,6 @@ EXPORT_PER_CPU_SYMBOL(kernel_stack);
 
 DEFINE_PER_CPU(unsigned int, irq_count) = -1;
 
-void __cpuinit pda_init(int cpu)
-{
-	/* Setup up data that may be needed in __get_free_pages early */
-	loadsegment(fs, 0);
-	loadsegment(gs, 0);
-
-	load_pda_offset(cpu);
-}
-
 static DEFINE_PER_CPU_PAGE_ALIGNED(char, exception_stacks
 	[(N_EXCEPTION_STACKS - 1) * EXCEPTION_STKSZ + DEBUG_STKSZ])
 	__aligned(PAGE_SIZE);
@@ -967,9 +958,9 @@ void __cpuinit cpu_init(void)
 	struct task_struct *me;
 	int i;
 
-	/* CPU 0 is initialised in head64.c */
-	if (cpu != 0)
-		pda_init(cpu);
+	loadsegment(fs, 0);
+	loadsegment(gs, 0);
+	load_gs_base(cpu);
 
 #ifdef CONFIG_NUMA
 	if (cpu != 0 && percpu_read(node_number) == 0 &&
diff --git a/arch/x86/kernel/head64.c b/arch/x86/kernel/head64.c
index af67d3227ea..f5b27224769 100644
--- a/arch/x86/kernel/head64.c
+++ b/arch/x86/kernel/head64.c
@@ -91,8 +91,6 @@ void __init x86_64_start_kernel(char * real_mode_data)
 	if (console_loglevel == 10)
 		early_printk("Kernel alive\n");
 
-	pda_init(0);
-
 	x86_64_start_reservations(real_mode_data);
 }
 
diff --git a/arch/x86/kernel/head_64.S b/arch/x86/kernel/head_64.S
index c8ace880661..a0a2b5ca9b7 100644
--- a/arch/x86/kernel/head_64.S
+++ b/arch/x86/kernel/head_64.S
@@ -207,19 +207,15 @@ ENTRY(secondary_startup_64)
 
 #ifdef CONFIG_SMP
 	/*
-	 * early_gdt_base should point to the gdt_page in static percpu init
-	 * data area.  Computing this requires two symbols - __per_cpu_load
-	 * and per_cpu__gdt_page.  As linker can't do no such relocation, do
-	 * it by hand.  As early_gdt_descr is manipulated by C code for
-	 * secondary CPUs, this should be done only once for the boot CPU
-	 * when early_gdt_descr_base contains zero.
+	 * Fix up static pointers that need __per_cpu_load added.  The assembler
+	 * is unable to do this directly.  This is only needed for the boot cpu.
+	 * These values are set up with the correct base addresses by C code for
+	 * secondary cpus.
 	 */
-	movq	early_gdt_descr_base(%rip), %rax
-	testq	%rax, %rax
-	jnz	1f
-	movq	$__per_cpu_load, %rax
-	addq	$per_cpu__gdt_page, %rax
-	movq	%rax, early_gdt_descr_base(%rip)
+	movq	initial_gs(%rip), %rax
+	cmpl	$0, per_cpu__cpu_number(%rax)
+	jne	1f
+	addq	%rax, early_gdt_descr_base(%rip)
 1:
 #endif
 	/*
@@ -246,13 +242,10 @@ ENTRY(secondary_startup_64)
 
 	/* Set up %gs.
 	 *
-	 * On SMP, %gs should point to the per-cpu area.  For initial
-	 * boot, make %gs point to the init data section.  For a
-	 * secondary CPU,initial_gs should be set to its pda address
-	 * before the CPU runs this code.
-	 *
-	 * On UP, initial_gs points to PER_CPU_VAR(__pda) and doesn't
-	 * change.
+	 * The base of %gs always points to the bottom of the irqstack
+	 * union.  If the stack protector canary is enabled, it is
+	 * located at %gs:40.  Note that, on SMP, the boot cpu uses
+	 * init data section till per cpu areas are set up.
 	 */
 	movl	$MSR_GS_BASE,%ecx
 	movq	initial_gs(%rip),%rax
@@ -285,7 +278,7 @@ ENTRY(secondary_startup_64)
 #ifdef CONFIG_SMP
 	.quad	__per_cpu_load
 #else
-	.quad	PER_CPU_VAR(__pda)
+	.quad	PER_CPU_VAR(irq_stack_union)
 #endif
 	__FINITDATA
 
@@ -431,12 +424,8 @@ NEXT_PAGE(level2_spare_pgt)
 	.globl early_gdt_descr
 early_gdt_descr:
 	.word	GDT_ENTRIES*8-1
-#ifdef CONFIG_SMP
 early_gdt_descr_base:
-	.quad   0x0000000000000000
-#else
 	.quad	per_cpu__gdt_page
-#endif
 
 ENTRY(phys_base)
 	/* This must match the first entry in level2_kernel_pgt */
diff --git a/arch/x86/kernel/process_64.c b/arch/x86/kernel/process_64.c
index aa89eabf09e..c422eebb0c5 100644
--- a/arch/x86/kernel/process_64.c
+++ b/arch/x86/kernel/process_64.c
@@ -47,7 +47,6 @@
 #include <asm/processor.h>
 #include <asm/i387.h>
 #include <asm/mmu_context.h>
-#include <asm/pda.h>
 #include <asm/prctl.h>
 #include <asm/desc.h>
 #include <asm/proto.h>
@@ -638,13 +637,6 @@ __switch_to(struct task_struct *prev_p, struct task_struct *next_p)
 	percpu_write(kernel_stack,
 		  (unsigned long)task_stack_page(next_p) +
 		  THREAD_SIZE - KERNEL_STACK_OFFSET);
-#ifdef CONFIG_CC_STACKPROTECTOR
-	/*
-	 * Build time only check to make sure the stack_canary is at
-	 * offset 40 in the pda; this is a gcc ABI requirement
-	 */
-	BUILD_BUG_ON(offsetof(struct x8664_pda, stack_canary) != 40);
-#endif
 
 	/*
 	 * Now maybe reload the debug registers and handle I/O bitmaps
diff --git a/arch/x86/kernel/setup_percpu.c b/arch/x86/kernel/setup_percpu.c
index efbafbbff58..90b8e154bb5 100644
--- a/arch/x86/kernel/setup_percpu.c
+++ b/arch/x86/kernel/setup_percpu.c
@@ -77,30 +77,6 @@ static void __init setup_node_to_cpumask_map(void);
 static inline void setup_node_to_cpumask_map(void) { }
 #endif
 
-/*
- * Define load_pda_offset() and per-cpu __pda for x86_64.
- * load_pda_offset() is responsible for loading the offset of pda into
- * %gs.
- *
- * On SMP, pda offset also duals as percpu base address and thus it
- * should be at the start of per-cpu area.  To achieve this, it's
- * preallocated in vmlinux_64.lds.S directly instead of using
- * DEFINE_PER_CPU().
- */
-#ifdef CONFIG_X86_64
-void __cpuinit load_pda_offset(int cpu)
-{
-	/* Memory clobbers used to order pda/percpu accesses */
-	mb();
-	wrmsrl(MSR_GS_BASE, cpu_pda(cpu));
-	mb();
-}
-#ifndef CONFIG_SMP
-DEFINE_PER_CPU(struct x8664_pda, __pda);
-#endif
-EXPORT_PER_CPU_SYMBOL(__pda);
-#endif /* CONFIG_SMP && CONFIG_X86_64 */
-
 #ifdef CONFIG_X86_64
 
 /* correctly size the local cpu masks */
@@ -207,15 +183,13 @@ void __init setup_per_cpu_areas(void)
 		per_cpu(cpu_number, cpu) = cpu;
 #ifdef CONFIG_X86_64
 		per_cpu(irq_stack_ptr, cpu) =
-			(char *)per_cpu(irq_stack, cpu) + IRQ_STACK_SIZE - 64;
+			per_cpu(irq_stack_union.irq_stack, cpu) + IRQ_STACK_SIZE - 64;
 		/*
-		 * CPU0 modified pda in the init data area, reload pda
-		 * offset for CPU0 and clear the area for others.
+		 * Up to this point, CPU0 has been using .data.init
+		 * area.  Reload %gs offset for CPU0.
 		 */
 		if (cpu == 0)
-			load_pda_offset(0);
-		else
-			memset(cpu_pda(cpu), 0, sizeof(*cpu_pda(cpu)));
+			load_gs_base(cpu);
 #endif
 
 		DBG("PERCPU: cpu %4d %p\n", cpu, ptr);
diff --git a/arch/x86/kernel/traps.c b/arch/x86/kernel/traps.c
index 98c2d055284..ed5aee5f3fc 100644
--- a/arch/x86/kernel/traps.c
+++ b/arch/x86/kernel/traps.c
@@ -59,7 +59,6 @@
 #ifdef CONFIG_X86_64
 #include <asm/pgalloc.h>
 #include <asm/proto.h>
-#include <asm/pda.h>
 #else
 #include <asm/processor-flags.h>
 #include <asm/arch_hooks.h>
diff --git a/arch/x86/kernel/vmlinux_64.lds.S b/arch/x86/kernel/vmlinux_64.lds.S
index a09abb8fb97..c9740996430 100644
--- a/arch/x86/kernel/vmlinux_64.lds.S
+++ b/arch/x86/kernel/vmlinux_64.lds.S
@@ -220,8 +220,7 @@ SECTIONS
    * so that it can be accessed as a percpu variable.
    */
   . = ALIGN(PAGE_SIZE);
-  PERCPU_VADDR_PREALLOC(0, :percpu, pda_size)
-  per_cpu____pda = __per_cpu_start;
+  PERCPU_VADDR(0, :percpu)
 #else
   PERCPU(PAGE_SIZE)
 #endif
@@ -262,3 +261,8 @@ SECTIONS
  */
 ASSERT((_end - _text <= KERNEL_IMAGE_SIZE),
 	"kernel image bigger than KERNEL_IMAGE_SIZE")
+
+#ifdef CONFIG_SMP
+ASSERT((per_cpu__irq_stack_union == 0),
+        "irq_stack_union is not at start of per-cpu area");
+#endif
diff --git a/arch/x86/xen/enlighten.c b/arch/x86/xen/enlighten.c
index 75b94139e1f..bef941f6145 100644
--- a/arch/x86/xen/enlighten.c
+++ b/arch/x86/xen/enlighten.c
@@ -1645,7 +1645,6 @@ asmlinkage void __init xen_start_kernel(void)
 #ifdef CONFIG_X86_64
 	/* Disable until direct per-cpu data access. */
 	have_vcpu_info_placement = 0;
-	pda_init(0);
 #endif
 
 	xen_smp_init();
diff --git a/include/asm-generic/vmlinux.lds.h b/include/asm-generic/vmlinux.lds.h
index aa6b9b1b30b..53e21f36a80 100644
--- a/include/asm-generic/vmlinux.lds.h
+++ b/include/asm-generic/vmlinux.lds.h
@@ -430,22 +430,10 @@
   	*(.initcall7.init)						\
   	*(.initcall7s.init)
 
-#define PERCPU_PROLOG(vaddr)						\
-	VMLINUX_SYMBOL(__per_cpu_load) = .;				\
-	.data.percpu vaddr : AT(VMLINUX_SYMBOL(__per_cpu_load)		\
-				- LOAD_OFFSET) {			\
-		VMLINUX_SYMBOL(__per_cpu_start) = .;
-
-#define PERCPU_EPILOG(phdr)						\
-		VMLINUX_SYMBOL(__per_cpu_end) = .;			\
-	} phdr								\
-	. = VMLINUX_SYMBOL(__per_cpu_load) + SIZEOF(.data.percpu);
-
 /**
- * PERCPU_VADDR_PREALLOC - define output section for percpu area with prealloc
+ * PERCPU_VADDR - define output section for percpu area
  * @vaddr: explicit base address (optional)
  * @phdr: destination PHDR (optional)
- * @prealloc: the size of prealloc area
  *
  * Macro which expands to output section for percpu area.  If @vaddr
  * is not blank, it specifies explicit base address and all percpu
@@ -457,39 +445,23 @@
  * section in the linker script will go there too.  @phdr should have
  * a leading colon.
  *
- * If @prealloc is non-zero, the specified number of bytes will be
- * reserved at the start of percpu area.  As the prealloc area is
- * likely to break alignment, this macro puts areas in increasing
- * alignment order.
- *
  * This macro defines three symbols, __per_cpu_load, __per_cpu_start
  * and __per_cpu_end.  The first one is the vaddr of loaded percpu
  * init data.  __per_cpu_start equals @vaddr and __per_cpu_end is the
  * end offset.
  */
-#define PERCPU_VADDR_PREALLOC(vaddr, segment, prealloc)			\
-	PERCPU_PROLOG(vaddr)						\
-		. += prealloc;						\
-		*(.data.percpu)						\
-		*(.data.percpu.shared_aligned)				\
-		*(.data.percpu.page_aligned)				\
-	PERCPU_EPILOG(segment)
-
-/**
- * PERCPU_VADDR - define output section for percpu area
- * @vaddr: explicit base address (optional)
- * @phdr: destination PHDR (optional)
- *
- * Macro which expands to output section for percpu area.  Mostly
- * identical to PERCPU_VADDR_PREALLOC(@vaddr, @phdr, 0) other than
- * using slighly different layout.
- */
 #define PERCPU_VADDR(vaddr, phdr)					\
-	PERCPU_PROLOG(vaddr)						\
+	VMLINUX_SYMBOL(__per_cpu_load) = .;				\
+	.data.percpu vaddr : AT(VMLINUX_SYMBOL(__per_cpu_load)		\
+				- LOAD_OFFSET) {			\
+		VMLINUX_SYMBOL(__per_cpu_start) = .;			\
+		*(.data.percpu.first)					\
 		*(.data.percpu.page_aligned)				\
 		*(.data.percpu)						\
 		*(.data.percpu.shared_aligned)				\
-	PERCPU_EPILOG(phdr)
+		VMLINUX_SYMBOL(__per_cpu_end) = .;			\
+	} phdr								\
+	. = VMLINUX_SYMBOL(__per_cpu_load) + SIZEOF(.data.percpu);
 
 /**
  * PERCPU - define output section for percpu area, simple version
diff --git a/include/linux/percpu.h b/include/linux/percpu.h
index 9f2a3751873..0e24202b5a4 100644
--- a/include/linux/percpu.h
+++ b/include/linux/percpu.h
@@ -9,34 +9,39 @@
 #include <asm/percpu.h>
 
 #ifdef CONFIG_SMP
-#define DEFINE_PER_CPU(type, name)					\
-	__attribute__((__section__(".data.percpu")))			\
-	PER_CPU_ATTRIBUTES __typeof__(type) per_cpu__##name
+#define PER_CPU_BASE_SECTION ".data.percpu"
 
 #ifdef MODULE
-#define SHARED_ALIGNED_SECTION ".data.percpu"
+#define PER_CPU_SHARED_ALIGNED_SECTION ""
 #else
-#define SHARED_ALIGNED_SECTION ".data.percpu.shared_aligned"
+#define PER_CPU_SHARED_ALIGNED_SECTION ".shared_aligned"
 #endif
+#define PER_CPU_FIRST_SECTION ".first"
 
-#define DEFINE_PER_CPU_SHARED_ALIGNED(type, name)			\
-	__attribute__((__section__(SHARED_ALIGNED_SECTION)))		\
-	PER_CPU_ATTRIBUTES __typeof__(type) per_cpu__##name		\
-	____cacheline_aligned_in_smp
+#else
+
+#define PER_CPU_BASE_SECTION ".data"
+#define PER_CPU_SHARED_ALIGNED_SECTION ""
+#define PER_CPU_FIRST_SECTION ""
+
+#endif
 
-#define DEFINE_PER_CPU_PAGE_ALIGNED(type, name)			\
-	__attribute__((__section__(".data.percpu.page_aligned")))	\
+#define DEFINE_PER_CPU_SECTION(type, name, section)			\
+	__attribute__((__section__(PER_CPU_BASE_SECTION section)))	\
 	PER_CPU_ATTRIBUTES __typeof__(type) per_cpu__##name
-#else
+
 #define DEFINE_PER_CPU(type, name)					\
-	PER_CPU_ATTRIBUTES __typeof__(type) per_cpu__##name
+	DEFINE_PER_CPU_SECTION(type, name, "")
 
-#define DEFINE_PER_CPU_SHARED_ALIGNED(type, name)		      \
-	DEFINE_PER_CPU(type, name)
+#define DEFINE_PER_CPU_SHARED_ALIGNED(type, name)			\
+	DEFINE_PER_CPU_SECTION(type, name, PER_CPU_SHARED_ALIGNED_SECTION) \
+	____cacheline_aligned_in_smp
 
-#define DEFINE_PER_CPU_PAGE_ALIGNED(type, name)		      \
-	DEFINE_PER_CPU(type, name)
-#endif
+#define DEFINE_PER_CPU_PAGE_ALIGNED(type, name)				\
+	DEFINE_PER_CPU_SECTION(type, name, ".page_aligned")
+
+#define DEFINE_PER_CPU_FIRST(type, name)				\
+	DEFINE_PER_CPU_SECTION(type, name, PER_CPU_FIRST_SECTION)
 
 #define EXPORT_PER_CPU_SYMBOL(var) EXPORT_SYMBOL(per_cpu__##var)
 #define EXPORT_PER_CPU_SYMBOL_GPL(var) EXPORT_SYMBOL_GPL(per_cpu__##var)