Merge branch 'core/percpu' into x86/core

13 files changed, 222 insertions, 30 deletions

diff --git a/arch/x86/include/asm/a.out-core.h b/arch/x86/include/asm/a.out-core.h
index 3c601f8224b..bb70e397aa8 100644
--- a/arch/x86/include/asm/a.out-core.h
+++ b/arch/x86/include/asm/a.out-core.h
@@ -55,7 +55,7 @@ static inline void aout_dump_thread(struct pt_regs *regs, struct user *dump)
 	dump->regs.ds = (u16)regs->ds;
 	dump->regs.es = (u16)regs->es;
 	dump->regs.fs = (u16)regs->fs;
-	savesegment(gs, dump->regs.gs);
+	dump->regs.gs = get_user_gs(regs);
 	dump->regs.orig_ax = regs->orig_ax;
 	dump->regs.ip = regs->ip;
 	dump->regs.cs = (u16)regs->cs;
diff --git a/arch/x86/include/asm/elf.h b/arch/x86/include/asm/elf.h
index f51a3ddde01..83c1bc8d2e8 100644
--- a/arch/x86/include/asm/elf.h
+++ b/arch/x86/include/asm/elf.h
@@ -112,7 +112,7 @@ extern unsigned int vdso_enabled;
  * now struct_user_regs, they are different)
  */
 
-#define ELF_CORE_COPY_REGS(pr_reg, regs)	\
+#define ELF_CORE_COPY_REGS_COMMON(pr_reg, regs)	\
 do {						\
 	pr_reg[0] = regs->bx;			\
 	pr_reg[1] = regs->cx;			\
@@ -124,7 +124,6 @@ do {						\
 	pr_reg[7] = regs->ds & 0xffff;		\
 	pr_reg[8] = regs->es & 0xffff;		\
 	pr_reg[9] = regs->fs & 0xffff;		\
-	savesegment(gs, pr_reg[10]);		\
 	pr_reg[11] = regs->orig_ax;		\
 	pr_reg[12] = regs->ip;			\
 	pr_reg[13] = regs->cs & 0xffff;		\
@@ -133,6 +132,18 @@ do {						\
 	pr_reg[16] = regs->ss & 0xffff;		\
 } while (0);
 
+#define ELF_CORE_COPY_REGS(pr_reg, regs)	\
+do {						\
+	ELF_CORE_COPY_REGS_COMMON(pr_reg, regs);\
+	pr_reg[10] = get_user_gs(regs);		\
+} while (0);
+
+#define ELF_CORE_COPY_KERNEL_REGS(pr_reg, regs)	\
+do {						\
+	ELF_CORE_COPY_REGS_COMMON(pr_reg, regs);\
+	savesegment(gs, pr_reg[10]);		\
+} while (0);
+
 #define ELF_PLATFORM	(utsname()->machine)
 #define set_personality_64bit()	do { } while (0)
 
diff --git a/arch/x86/include/asm/mmu_context.h b/arch/x86/include/asm/mmu_context.h
index 52948df9cd1..f923203dc39 100644
--- a/arch/x86/include/asm/mmu_context.h
+++ b/arch/x86/include/asm/mmu_context.h
@@ -79,7 +79,7 @@ do {						\
 #ifdef CONFIG_X86_32
 #define deactivate_mm(tsk, mm)			\
 do {						\
-	loadsegment(gs, 0);			\
+	lazy_load_gs(0);			\
 } while (0)
 #else
 #define deactivate_mm(tsk, mm)			\
diff --git a/arch/x86/include/asm/percpu.h b/arch/x86/include/asm/percpu.h
index 0b64af4f13a..aee103b26d0 100644
--- a/arch/x86/include/asm/percpu.h
+++ b/arch/x86/include/asm/percpu.h
@@ -34,6 +34,12 @@
 #define PER_CPU_VAR(var)	per_cpu__##var
 #endif	/* SMP */
 
+#ifdef CONFIG_X86_64_SMP
+#define INIT_PER_CPU_VAR(var)  init_per_cpu__##var
+#else
+#define INIT_PER_CPU_VAR(var)  per_cpu__##var
+#endif
+
 #else /* ...!ASSEMBLY */
 
 #include <linux/stringify.h>
@@ -45,6 +51,22 @@
 #define __percpu_arg(x)		"%" #x
 #endif
 
+/*
+ * Initialized pointers to per-cpu variables needed for the boot
+ * processor need to use these macros to get the proper address
+ * offset from __per_cpu_load on SMP.
+ *
+ * There also must be an entry in vmlinux_64.lds.S
+ */
+#define DECLARE_INIT_PER_CPU(var) \
+       extern typeof(per_cpu_var(var)) init_per_cpu_var(var)
+
+#ifdef CONFIG_X86_64_SMP
+#define init_per_cpu_var(var)  init_per_cpu__##var
+#else
+#define init_per_cpu_var(var)  per_cpu_var(var)
+#endif
+
 /* For arch-specific code, we can use direct single-insn ops (they
  * don't give an lvalue though). */
 extern void __bad_percpu_size(void);
diff --git a/arch/x86/include/asm/processor.h b/arch/x86/include/asm/processor.h
index a6643f68fbb..a0133838b67 100644
--- a/arch/x86/include/asm/processor.h
+++ b/arch/x86/include/asm/processor.h
@@ -393,8 +393,14 @@ union irq_stack_union {
 };
 
 DECLARE_PER_CPU(union irq_stack_union, irq_stack_union);
+DECLARE_INIT_PER_CPU(irq_stack_union);
+
 DECLARE_PER_CPU(char *, irq_stack_ptr);
+#else	/* X86_64 */
+#ifdef CONFIG_CC_STACKPROTECTOR
+DECLARE_PER_CPU(unsigned long, stack_canary);
 #endif
+#endif	/* X86_64 */
 
 extern void print_cpu_info(struct cpuinfo_x86 *);
 extern unsigned int xstate_size;
diff --git a/arch/x86/include/asm/ptrace.h b/arch/x86/include/asm/ptrace.h
index 6d34d954c22..e304b66abee 100644
--- a/arch/x86/include/asm/ptrace.h
+++ b/arch/x86/include/asm/ptrace.h
@@ -28,7 +28,7 @@ struct pt_regs {
 	int  xds;
 	int  xes;
 	int  xfs;
-	/* int  gs; */
+	int  xgs;
 	long orig_eax;
 	long eip;
 	int  xcs;
@@ -50,7 +50,7 @@ struct pt_regs {
 	unsigned long ds;
 	unsigned long es;
 	unsigned long fs;
-	/* int  gs; */
+	unsigned long gs;
 	unsigned long orig_ax;
 	unsigned long ip;
 	unsigned long cs;
diff --git a/arch/x86/include/asm/segment.h b/arch/x86/include/asm/segment.h
index 1dc1b51ac62..14e0ed86a6f 100644
--- a/arch/x86/include/asm/segment.h
+++ b/arch/x86/include/asm/segment.h
@@ -61,7 +61,7 @@
  *
  *  26 - ESPFIX small SS
  *  27 - per-cpu			[ offset to per-cpu data area ]
- *  28 - unused
+ *  28 - stack_canary-20		[ for stack protector ]
  *  29 - unused
  *  30 - unused
  *  31 - TSS for double fault handler
@@ -95,6 +95,13 @@
 #define __KERNEL_PERCPU 0
 #endif
 
+#define GDT_ENTRY_STACK_CANARY		(GDT_ENTRY_KERNEL_BASE + 16)
+#ifdef CONFIG_CC_STACKPROTECTOR
+#define __KERNEL_STACK_CANARY		(GDT_ENTRY_STACK_CANARY * 8)
+#else
+#define __KERNEL_STACK_CANARY		0
+#endif
+
 #define GDT_ENTRY_DOUBLEFAULT_TSS	31
 
 /*
diff --git a/arch/x86/include/asm/stackprotector.h b/arch/x86/include/asm/stackprotector.h
index 36a700acaf2..c2d742c6e15 100644
--- a/arch/x86/include/asm/stackprotector.h
+++ b/arch/x86/include/asm/stackprotector.h
@@ -1,8 +1,54 @@
+/*
+ * GCC stack protector support.
+ *
+ * Stack protector works by putting predefined pattern at the start of
+ * the stack frame and verifying that it hasn't been overwritten when
+ * returning from the function.  The pattern is called stack canary
+ * and unfortunately gcc requires it to be at a fixed offset from %gs.
+ * On x86_64, the offset is 40 bytes and on x86_32 20 bytes.  x86_64
+ * and x86_32 use segment registers differently and thus handles this
+ * requirement differently.
+ *
+ * On x86_64, %gs is shared by percpu area and stack canary.  All
+ * percpu symbols are zero based and %gs points to the base of percpu
+ * area.  The first occupant of the percpu area is always
+ * irq_stack_union which contains stack_canary at offset 40.  Userland
+ * %gs is always saved and restored on kernel entry and exit using
+ * swapgs, so stack protector doesn't add any complexity there.
+ *
+ * On x86_32, it's slightly more complicated.  As in x86_64, %gs is
+ * used for userland TLS.  Unfortunately, some processors are much
+ * slower at loading segment registers with different value when
+ * entering and leaving the kernel, so the kernel uses %fs for percpu
+ * area and manages %gs lazily so that %gs is switched only when
+ * necessary, usually during task switch.
+ *
+ * As gcc requires the stack canary at %gs:20, %gs can't be managed
+ * lazily if stack protector is enabled, so the kernel saves and
+ * restores userland %gs on kernel entry and exit.  This behavior is
+ * controlled by CONFIG_X86_32_LAZY_GS and accessors are defined in
+ * system.h to hide the details.
+ */
+
 #ifndef _ASM_STACKPROTECTOR_H
 #define _ASM_STACKPROTECTOR_H 1
 
+#ifdef CONFIG_CC_STACKPROTECTOR
+
 #include <asm/tsc.h>
 #include <asm/processor.h>
+#include <asm/percpu.h>
+#include <asm/system.h>
+#include <asm/desc.h>
+#include <linux/random.h>
+
+/*
+ * 24 byte read-only segment initializer for stack canary.  Linker
+ * can't handle the address bit shifting.  Address will be set in
+ * head_32 for boot CPU and setup_per_cpu_areas() for others.
+ */
+#define GDT_STACK_CANARY_INIT						\
+	[GDT_ENTRY_STACK_CANARY] = { { { 0x00000018, 0x00409000 } } },
 
 /*
  * Initialize the stackprotector canary value.
@@ -15,12 +61,9 @@ static __always_inline void boot_init_stack_canary(void)
 	u64 canary;
 	u64 tsc;
 
-	/*
-	 * Build time only check to make sure the stack_canary is at
-	 * offset 40 in the pda; this is a gcc ABI requirement
-	 */
+#ifdef CONFIG_X86_64
 	BUILD_BUG_ON(offsetof(union irq_stack_union, stack_canary) != 40);
-
+#endif
 	/*
 	 * We both use the random pool and the current TSC as a source
 	 * of randomness. The TSC only matters for very early init,
@@ -32,7 +75,50 @@ static __always_inline void boot_init_stack_canary(void)
 	canary += tsc + (tsc << 32UL);
 
 	current->stack_canary = canary;
+#ifdef CONFIG_X86_64
 	percpu_write(irq_stack_union.stack_canary, canary);
+#else
+	percpu_write(stack_canary, canary);
+#endif
 }
 
+static inline void setup_stack_canary_segment(int cpu)
+{
+#ifdef CONFIG_X86_32
+	unsigned long canary = (unsigned long)&per_cpu(stack_canary, cpu) - 20;
+	struct desc_struct *gdt_table = get_cpu_gdt_table(cpu);
+	struct desc_struct desc;
+
+	desc = gdt_table[GDT_ENTRY_STACK_CANARY];
+	desc.base0 = canary & 0xffff;
+	desc.base1 = (canary >> 16) & 0xff;
+	desc.base2 = (canary >> 24) & 0xff;
+	write_gdt_entry(gdt_table, GDT_ENTRY_STACK_CANARY, &desc, DESCTYPE_S);
+#endif
+}
+
+static inline void load_stack_canary_segment(void)
+{
+#ifdef CONFIG_X86_32
+	asm("mov %0, %%gs" : : "r" (__KERNEL_STACK_CANARY) : "memory");
+#endif
+}
+
+#else	/* CC_STACKPROTECTOR */
+
+#define GDT_STACK_CANARY_INIT
+
+/* dummy boot_init_stack_canary() is defined in linux/stackprotector.h */
+
+static inline void setup_stack_canary_segment(int cpu)
+{ }
+
+static inline void load_stack_canary_segment(void)
+{
+#ifdef CONFIG_X86_32
+	asm volatile ("mov %0, %%gs" : : "r" (0));
 #endif
+}
+
+#endif	/* CC_STACKPROTECTOR */
+#endif	/* _ASM_STACKPROTECTOR_H */
diff --git a/arch/x86/include/asm/syscalls.h b/arch/x86/include/asm/syscalls.h
index c0b0bda754e..68b1be10cfa 100644
--- a/arch/x86/include/asm/syscalls.h
+++ b/arch/x86/include/asm/syscalls.h
@@ -29,21 +29,21 @@ asmlinkage int sys_get_thread_area(struct user_desc __user *);
 /* X86_32 only */
 #ifdef CONFIG_X86_32
 /* kernel/process_32.c */
-asmlinkage int sys_fork(struct pt_regs);
-asmlinkage int sys_clone(struct pt_regs);
-asmlinkage int sys_vfork(struct pt_regs);
-asmlinkage int sys_execve(struct pt_regs);
+int sys_fork(struct pt_regs *);
+int sys_clone(struct pt_regs *);
+int sys_vfork(struct pt_regs *);
+int sys_execve(struct pt_regs *);
 
 /* kernel/signal_32.c */
 asmlinkage int sys_sigsuspend(int, int, old_sigset_t);
 asmlinkage int sys_sigaction(int, const struct old_sigaction __user *,
 			     struct old_sigaction __user *);
-asmlinkage int sys_sigaltstack(unsigned long);
-asmlinkage unsigned long sys_sigreturn(unsigned long);
-asmlinkage int sys_rt_sigreturn(unsigned long);
+int sys_sigaltstack(struct pt_regs *);
+unsigned long sys_sigreturn(struct pt_regs *);
+long sys_rt_sigreturn(struct pt_regs *);
 
 /* kernel/ioport.c */
-asmlinkage long sys_iopl(unsigned long);
+long sys_iopl(struct pt_regs *);
 
 /* kernel/sys_i386_32.c */
 asmlinkage long sys_mmap2(unsigned long, unsigned long, unsigned long,
@@ -59,8 +59,8 @@ struct oldold_utsname;
 asmlinkage int sys_olduname(struct oldold_utsname __user *);
 
 /* kernel/vm86_32.c */
-asmlinkage int sys_vm86old(struct pt_regs);
-asmlinkage int sys_vm86(struct pt_regs);
+int sys_vm86old(struct pt_regs *);
+int sys_vm86(struct pt_regs *);
 
 #else /* CONFIG_X86_32 */
 
diff --git a/arch/x86/include/asm/system.h b/arch/x86/include/asm/system.h
index c22383743f3..c00bfdbdd45 100644
--- a/arch/x86/include/asm/system.h
+++ b/arch/x86/include/asm/system.h
@@ -23,6 +23,20 @@ struct task_struct *__switch_to(struct task_struct *prev,
 
 #ifdef CONFIG_X86_32
 
+#ifdef CONFIG_CC_STACKPROTECTOR
+#define __switch_canary							\
+	"movl %P[task_canary](%[next]), %%ebx\n\t"			\
+	"movl %%ebx, "__percpu_arg([stack_canary])"\n\t"
+#define __switch_canary_oparam						\
+	, [stack_canary] "=m" (per_cpu_var(stack_canary))
+#define __switch_canary_iparam						\
+	, [task_canary] "i" (offsetof(struct task_struct, stack_canary))
+#else	/* CC_STACKPROTECTOR */
+#define __switch_canary
+#define __switch_canary_oparam
+#define __switch_canary_iparam
+#endif	/* CC_STACKPROTECTOR */
+
 /*
  * Saving eflags is important. It switches not only IOPL between tasks,
  * it also protects other tasks from NT leaking through sysenter etc.
@@ -44,6 +58,7 @@ do {									\
 		     "movl %[next_sp],%%esp\n\t"	/* restore ESP   */ \
 		     "movl $1f,%[prev_ip]\n\t"	/* save    EIP   */	\
 		     "pushl %[next_ip]\n\t"	/* restore EIP   */	\
+		     __switch_canary					\
 		     "jmp __switch_to\n"	/* regparm call  */	\
 		     "1:\t"						\
 		     "popl %%ebp\n\t"		/* restore EBP   */	\
@@ -58,6 +73,8 @@ do {									\
 		       "=b" (ebx), "=c" (ecx), "=d" (edx),		\
 		       "=S" (esi), "=D" (edi)				\
 		       							\
+		       __switch_canary_oparam				\
+									\
 		       /* input parameters: */				\
 		     : [next_sp]  "m" (next->thread.sp),		\
 		       [next_ip]  "m" (next->thread.ip),		\
@@ -66,6 +83,8 @@ do {									\
 		       [prev]     "a" (prev),				\
 		       [next]     "d" (next)				\
 									\
+		       __switch_canary_iparam				\
+									\
 		     : /* reloaded segment registers */			\
 			"memory");					\
 } while (0)
@@ -182,6 +201,25 @@ extern void native_load_gs_index(unsigned);
 #define savesegment(seg, value)				\
 	asm("mov %%" #seg ",%0":"=r" (value) : : "memory")
 
+/*
+ * x86_32 user gs accessors.
+ */
+#ifdef CONFIG_X86_32
+#ifdef CONFIG_X86_32_LAZY_GS
+#define get_user_gs(regs)	(u16)({unsigned long v; savesegment(gs, v); v;})
+#define set_user_gs(regs, v)	loadsegment(gs, (unsigned long)(v))
+#define task_user_gs(tsk)	((tsk)->thread.gs)
+#define lazy_save_gs(v)		savesegment(gs, (v))
+#define lazy_load_gs(v)		loadsegment(gs, (v))
+#else	/* X86_32_LAZY_GS */
+#define get_user_gs(regs)	(u16)((regs)->gs)
+#define set_user_gs(regs, v)	do { (regs)->gs = (v); } while (0)
+#define task_user_gs(tsk)	(task_pt_regs(tsk)->gs)
+#define lazy_save_gs(v)		do { } while (0)
+#define lazy_load_gs(v)		do { } while (0)
+#endif	/* X86_32_LAZY_GS */
+#endif	/* X86_32 */
+
 static inline unsigned long get_limit(unsigned long segment)
 {
 	unsigned long __limit;
diff --git a/arch/x86/include/asm/traps.h b/arch/x86/include/asm/traps.h
index cf3bb053da0..0d5342515b8 100644
--- a/arch/x86/include/asm/traps.h
+++ b/arch/x86/include/asm/traps.h
@@ -41,7 +41,7 @@ dotraplinkage void do_int3(struct pt_regs *, long);
 dotraplinkage void do_overflow(struct pt_regs *, long);
 dotraplinkage void do_bounds(struct pt_regs *, long);
 dotraplinkage void do_invalid_op(struct pt_regs *, long);
-dotraplinkage void do_device_not_available(struct pt_regs);
+dotraplinkage void do_device_not_available(struct pt_regs *, long);
 dotraplinkage void do_coprocessor_segment_overrun(struct pt_regs *, long);
 dotraplinkage void do_invalid_TSS(struct pt_regs *, long);
 dotraplinkage void do_segment_not_present(struct pt_regs *, long);
diff --git a/arch/x86/include/asm/uaccess.h b/arch/x86/include/asm/uaccess.h
index 0ec6de4bcb0..b685ece89d5 100644
--- a/arch/x86/include/asm/uaccess.h
+++ b/arch/x86/include/asm/uaccess.h
@@ -186,7 +186,7 @@ extern int __get_user_bad(void);
 
 
 #ifdef CONFIG_X86_32
-#define __put_user_asm_u64(x, addr, err)				\
+#define __put_user_asm_u64(x, addr, err, errret)			\
 	asm volatile("1:	movl %%eax,0(%2)\n"			\
 		     "2:	movl %%edx,4(%2)\n"			\
 		     "3:\n"						\
@@ -197,7 +197,7 @@ extern int __get_user_bad(void);
 		     _ASM_EXTABLE(1b, 4b)				\
 		     _ASM_EXTABLE(2b, 4b)				\
 		     : "=r" (err)					\
-		     : "A" (x), "r" (addr), "i" (-EFAULT), "0" (err))
+		     : "A" (x), "r" (addr), "i" (errret), "0" (err))
 
 #define __put_user_asm_ex_u64(x, addr)					\
 	asm volatile("1:	movl %%eax,0(%1)\n"			\
@@ -211,8 +211,8 @@ extern int __get_user_bad(void);
 	asm volatile("call __put_user_8" : "=a" (__ret_pu)	\
 		     : "A" ((typeof(*(ptr)))(x)), "c" (ptr) : "ebx")
 #else
-#define __put_user_asm_u64(x, ptr, retval) \
-	__put_user_asm(x, ptr, retval, "q", "", "Zr", -EFAULT)
+#define __put_user_asm_u64(x, ptr, retval, errret) \
+	__put_user_asm(x, ptr, retval, "q", "", "Zr", errret)
 #define __put_user_asm_ex_u64(x, addr)	\
 	__put_user_asm_ex(x, addr, "q", "", "Zr")
 #define __put_user_x8(x, ptr, __ret_pu) __put_user_x(8, x, ptr, __ret_pu)
@@ -289,7 +289,8 @@ do {									\
 		__put_user_asm(x, ptr, retval, "l", "k", "ir", errret);	\
 		break;							\
 	case 8:								\
-		__put_user_asm_u64((__typeof__(*ptr))(x), ptr, retval);	\
+		__put_user_asm_u64((__typeof__(*ptr))(x), ptr, retval,	\
+				   errret);				\
 		break;							\
 	default:							\
 		__put_user_bad();					\
@@ -525,8 +526,6 @@ struct __large_struct { unsigned long buf[100]; };
  */
 #define get_user_try		uaccess_try
 #define get_user_catch(err)	uaccess_catch(err)
-#define put_user_try		uaccess_try
-#define put_user_catch(err)	uaccess_catch(err)
 
 #define get_user_ex(x, ptr)	do {					\
 	unsigned long __gue_val;					\
@@ -534,9 +533,29 @@ struct __large_struct { unsigned long buf[100]; };
 	(x) = (__force __typeof__(*(ptr)))__gue_val;			\
 } while (0)
 
+#ifdef CONFIG_X86_WP_WORKS_OK
+
+#define put_user_try		uaccess_try
+#define put_user_catch(err)	uaccess_catch(err)
+
 #define put_user_ex(x, ptr)						\
 	__put_user_size_ex((__typeof__(*(ptr)))(x), (ptr), sizeof(*(ptr)))
 
+#else /* !CONFIG_X86_WP_WORKS_OK */
+
+#define put_user_try		do {		\
+	int __uaccess_err = 0;
+
+#define put_user_catch(err)			\
+	(err) |= __uaccess_err;			\
+} while (0)
+
+#define put_user_ex(x, ptr)	do {		\
+	__uaccess_err |= __put_user(x, ptr);	\
+} while (0)
+
+#endif /* CONFIG_X86_WP_WORKS_OK */
+
 /*
  * movsl can be slow when source and dest are not both 8-byte aligned
  */
diff --git a/arch/x86/include/asm/uv/uv.h b/arch/x86/include/asm/uv/uv.h
index 8ac1d7e312f..8242bf96581 100644
--- a/arch/x86/include/asm/uv/uv.h
+++ b/arch/x86/include/asm/uv/uv.h
@@ -3,6 +3,9 @@
 
 enum uv_system_type {UV_NONE, UV_LEGACY_APIC, UV_X2APIC, UV_NON_UNIQUE_APIC};
 
+struct cpumask;
+struct mm_struct;
+
 #ifdef CONFIG_X86_UV
 
 extern enum uv_system_type get_uv_system_type(void);