6 files changed, 162 insertions, 40 deletions

diff --git a/arch/x86/include/asm/cpufeature.h b/arch/x86/include/asm/cpufeature.h
index 630e623f61e0..dca9c545f44e 100644
--- a/arch/x86/include/asm/cpufeature.h
+++ b/arch/x86/include/asm/cpufeature.h
@@ -176,6 +176,7 @@
 
 #if defined(__KERNEL__) && !defined(__ASSEMBLY__)
 
+#include <asm/asm.h>
 #include <linux/bitops.h>
 
 extern const char * const x86_cap_flags[NCAPINTS*32];
@@ -284,6 +285,62 @@ extern const char * const x86_power_flags[32];
 
 #endif /* CONFIG_X86_64 */
 
+/*
+ * Static testing of CPU features.  Used the same as boot_cpu_has().
+ * These are only valid after alternatives have run, but will statically
+ * patch the target code for additional performance.
+ *
+ */
+static __always_inline __pure bool __static_cpu_has(u8 bit)
+{
+#if __GNUC__ > 4 || (__GNUC__ == 4 && __GNUC_MINOR__ >= 5)
+                asm goto("1: jmp %l[t_no]\n"
+                         "2:\n"
+                         ".section .altinstructions,\"a\"\n"
+                         _ASM_ALIGN "\n"
+                         _ASM_PTR "1b\n"
+                         _ASM_PTR "0\n"         /* no replacement */
+                         " .byte %P0\n"         /* feature bit */
+                         " .byte 2b - 1b\n"     /* source len */
+                         " .byte 0\n"           /* replacement len */
+                         " .byte 0xff + 0 - (2b-1b)\n"  /* padding */
+                         ".previous\n"
+                         : : "i" (bit) : : t_no);
+                return true;
+        t_no:
+                return false;
+#else
+                u8 flag;
+                /* Open-coded due to __stringify() in ALTERNATIVE() */
+                asm volatile("1: movb $0,%0\n"
+                             "2:\n"
+                             ".section .altinstructions,\"a\"\n"
+                             _ASM_ALIGN "\n"
+                             _ASM_PTR "1b\n"
+                             _ASM_PTR "3f\n"
+                             " .byte %P1\n"             /* feature bit */
+                             " .byte 2b - 1b\n"         /* source len */
+                             " .byte 4f - 3f\n"         /* replacement len */
+                             " .byte 0xff + (4f-3f) - (2b-1b)\n" /* padding */
+                             ".previous\n"
+                             ".section .altinstr_replacement,\"ax\"\n"
+                             "3: movb $1,%0\n"
+                             "4:\n"
+                             ".previous\n"
+                             : "=qm" (flag) : "i" (bit));
+                return flag;
+#endif
+}
+
+#define static_cpu_has(bit)                                     \
+(                                                               \
+        __builtin_constant_p(boot_cpu_has(bit)) ?               \
+                boot_cpu_has(bit) :                             \
+        (__builtin_constant_p(bit) && !((bit) & ~0xff)) ?       \
+                __static_cpu_has(bit) :                         \
+                boot_cpu_has(bit)                               \
+)
+
 #endif /* defined(__KERNEL__) && !defined(__ASSEMBLY__) */
 
 #endif /* _ASM_X86_CPUFEATURE_H */
diff --git a/arch/x86/include/asm/i387.h b/arch/x86/include/asm/i387.h
index da2930924501..c991b3a7b904 100644
--- a/arch/x86/include/asm/i387.h
+++ b/arch/x86/include/asm/i387.h
@@ -16,7 +16,9 @@
 #include <linux/kernel_stat.h>
 #include <linux/regset.h>
 #include <linux/hardirq.h>
+#include <linux/slab.h>
 #include <asm/asm.h>
+#include <asm/cpufeature.h>
 #include <asm/processor.h>
 #include <asm/sigcontext.h>
 #include <asm/user.h>
@@ -56,6 +58,11 @@ extern int restore_i387_xstate_ia32(void __user *buf);
 
 #define X87_FSW_ES (1 << 7)     /* Exception Summary */
 
+static __always_inline __pure bool use_xsave(void)
+{
+        return static_cpu_has(X86_FEATURE_XSAVE);
+}
+
 #ifdef CONFIG_X86_64
 
 /* Ignore delayed exceptions from user space */
@@ -91,15 +98,15 @@ static inline int fxrstor_checking(struct i387_fxsave_struct *fx)
    values. The kernel data segment can be sometimes 0 and sometimes
    new user value. Both should be ok.
    Use the PDA as safe address because it should be already in L1. */
-static inline void clear_fpu_state(struct task_struct *tsk)
+static inline void fpu_clear(struct fpu *fpu)
 {
-        struct xsave_struct *xstate = &tsk->thread.xstate->xsave;
-        struct i387_fxsave_struct *fx = &tsk->thread.xstate->fxsave;
+        struct xsave_struct *xstate = &fpu->state->xsave;
+        struct i387_fxsave_struct *fx = &fpu->state->fxsave;
 
         /*
          * xsave header may indicate the init state of the FP.
          */
-        if ((task_thread_info(tsk)->status & TS_XSAVE) &&
+        if (use_xsave() &&
             !(xstate->xsave_hdr.xstate_bv & XSTATE_FP))
                 return;
 
@@ -111,6 +118,11 @@ static inline void clear_fpu_state(struct task_struct *tsk)
                           X86_FEATURE_FXSAVE_LEAK);
 }
 
+static inline void clear_fpu_state(struct task_struct *tsk)
+{
+        fpu_clear(&tsk->thread.fpu);
+}
+
 static inline int fxsave_user(struct i387_fxsave_struct __user *fx)
 {
         int err;
@@ -135,7 +147,7 @@ static inline int fxsave_user(struct i387_fxsave_struct __user *fx)
         return err;
 }
 
-static inline void fxsave(struct task_struct *tsk)
+static inline void fpu_fxsave(struct fpu *fpu)
 {
         /* Using "rex64; fxsave %0" is broken because, if the memory operand
            uses any extended registers for addressing, a second REX prefix
@@ -145,42 +157,45 @@ static inline void fxsave(struct task_struct *tsk)
         /* Using "fxsaveq %0" would be the ideal choice, but is only supported
            starting with gas 2.16. */
         __asm__ __volatile__("fxsaveq %0"
-                             : "=m" (tsk->thread.xstate->fxsave));
+                             : "=m" (fpu->state->fxsave));
 #elif 0
         /* Using, as a workaround, the properly prefixed form below isn't
            accepted by any binutils version so far released, complaining that
            the same type of prefix is used twice if an extended register is
            needed for addressing (fix submitted to mainline 2005-11-21). */
         __asm__ __volatile__("rex64/fxsave %0"
-                             : "=m" (tsk->thread.xstate->fxsave));
+                             : "=m" (fpu->state->fxsave));
 #else
         /* This, however, we can work around by forcing the compiler to select
            an addressing mode that doesn't require extended registers. */
         __asm__ __volatile__("rex64/fxsave (%1)"
-                             : "=m" (tsk->thread.xstate->fxsave)
-                             : "cdaSDb" (&tsk->thread.xstate->fxsave));
+                             : "=m" (fpu->state->fxsave)
+                             : "cdaSDb" (&fpu->state->fxsave));
 #endif
 }
 
-static inline void __save_init_fpu(struct task_struct *tsk)
+static inline void fpu_save_init(struct fpu *fpu)
 {
-        if (task_thread_info(tsk)->status & TS_XSAVE)
-                xsave(tsk);
+        if (use_xsave())
+                fpu_xsave(fpu);
         else
-                fxsave(tsk);
+                fpu_fxsave(fpu);
+
+        fpu_clear(fpu);
+}
 
-        clear_fpu_state(tsk);
+static inline void __save_init_fpu(struct task_struct *tsk)
+{
+        fpu_save_init(&tsk->thread.fpu);
         task_thread_info(tsk)->status &= ~TS_USEDFPU;
 }
 
 #else  /* CONFIG_X86_32 */
 
 #ifdef CONFIG_MATH_EMULATION
-extern void finit_task(struct task_struct *tsk);
+extern void finit_soft_fpu(struct i387_soft_struct *soft);
 #else
-static inline void finit_task(struct task_struct *tsk)
-{
-}
+static inline void finit_soft_fpu(struct i387_soft_struct *soft) {}
 #endif
 
 static inline void tolerant_fwait(void)
@@ -216,13 +231,13 @@ static inline int fxrstor_checking(struct i387_fxsave_struct *fx)
 /*
  * These must be called with preempt disabled
  */
-static inline void __save_init_fpu(struct task_struct *tsk)
+static inline void fpu_save_init(struct fpu *fpu)
 {
-        if (task_thread_info(tsk)->status & TS_XSAVE) {
-                struct xsave_struct *xstate = &tsk->thread.xstate->xsave;
-                struct i387_fxsave_struct *fx = &tsk->thread.xstate->fxsave;
+        if (use_xsave()) {
+                struct xsave_struct *xstate = &fpu->state->xsave;
+                struct i387_fxsave_struct *fx = &fpu->state->fxsave;
 
-                xsave(tsk);
+                fpu_xsave(fpu);
 
                 /*
                  * xsave header may indicate the init state of the FP.
@@ -246,8 +261,8 @@ static inline void __save_init_fpu(struct task_struct *tsk)
                 "fxsave %[fx]\n"
                 "bt $7,%[fsw] ; jnc 1f ; fnclex\n1:",
                 X86_FEATURE_FXSR,
-                [fx] "m" (tsk->thread.xstate->fxsave),
-                [fsw] "m" (tsk->thread.xstate->fxsave.swd) : "memory");
+                [fx] "m" (fpu->state->fxsave),
+                [fsw] "m" (fpu->state->fxsave.swd) : "memory");
 clear_state:
         /* AMD K7/K8 CPUs don't save/restore FDP/FIP/FOP unless an exception
            is pending.  Clear the x87 state here by setting it to fixed
@@ -259,17 +274,34 @@ clear_state:
                 X86_FEATURE_FXSAVE_LEAK,
                 [addr] "m" (safe_address));
 end:
+        ;
+}
+
+static inline void __save_init_fpu(struct task_struct *tsk)
+{
+        fpu_save_init(&tsk->thread.fpu);
         task_thread_info(tsk)->status &= ~TS_USEDFPU;
 }
 
+
 #endif  /* CONFIG_X86_64 */
 
-static inline int restore_fpu_checking(struct task_struct *tsk)
+static inline int fpu_fxrstor_checking(struct fpu *fpu)
 {
-        if (task_thread_info(tsk)->status & TS_XSAVE)
-                return xrstor_checking(&tsk->thread.xstate->xsave);
+        return fxrstor_checking(&fpu->state->fxsave);
+}
+
+static inline int fpu_restore_checking(struct fpu *fpu)
+{
+        if (use_xsave())
+                return fpu_xrstor_checking(fpu);
         else
-                return fxrstor_checking(&tsk->thread.xstate->fxsave);
+                return fpu_fxrstor_checking(fpu);
+}
+
+static inline int restore_fpu_checking(struct task_struct *tsk)
+{
+        return fpu_restore_checking(&tsk->thread.fpu);
 }
 
 /*
@@ -397,30 +429,59 @@ static inline void clear_fpu(struct task_struct *tsk)
 static inline unsigned short get_fpu_cwd(struct task_struct *tsk)
 {
         if (cpu_has_fxsr) {
-                return tsk->thread.xstate->fxsave.cwd;
+                return tsk->thread.fpu.state->fxsave.cwd;
         } else {
-                return (unsigned short)tsk->thread.xstate->fsave.cwd;
+                return (unsigned short)tsk->thread.fpu.state->fsave.cwd;
         }
 }
 
 static inline unsigned short get_fpu_swd(struct task_struct *tsk)
 {
         if (cpu_has_fxsr) {
-                return tsk->thread.xstate->fxsave.swd;
+                return tsk->thread.fpu.state->fxsave.swd;
         } else {
-                return (unsigned short)tsk->thread.xstate->fsave.swd;
+                return (unsigned short)tsk->thread.fpu.state->fsave.swd;
         }
 }
 
 static inline unsigned short get_fpu_mxcsr(struct task_struct *tsk)
 {
         if (cpu_has_xmm) {
-                return tsk->thread.xstate->fxsave.mxcsr;
+                return tsk->thread.fpu.state->fxsave.mxcsr;
         } else {
                 return MXCSR_DEFAULT;
         }
 }
 
+static bool fpu_allocated(struct fpu *fpu)
+{
+        return fpu->state != NULL;
+}
+
+static inline int fpu_alloc(struct fpu *fpu)
+{
+        if (fpu_allocated(fpu))
+                return 0;
+        fpu->state = kmem_cache_alloc(task_xstate_cachep, GFP_KERNEL);
+        if (!fpu->state)
+                return -ENOMEM;
+        WARN_ON((unsigned long)fpu->state & 15);
+        return 0;
+}
+
+static inline void fpu_free(struct fpu *fpu)
+{
+        if (fpu->state) {
+                kmem_cache_free(task_xstate_cachep, fpu->state);
+                fpu->state = NULL;
+        }
+}
+
+static inline void fpu_copy(struct fpu *dst, struct fpu *src)
+{
+        memcpy(dst->state, src->state, xstate_size);
+}
+
 #endif /* __ASSEMBLY__ */
 
 #define PSHUFB_XMM5_XMM0 .byte 0x66, 0x0f, 0x38, 0x00, 0xc5
diff --git a/arch/x86/include/asm/processor.h b/arch/x86/include/asm/processor.h
index c71a12d960d4..5a51379dcbe4 100644
--- a/arch/x86/include/asm/processor.h
+++ b/arch/x86/include/asm/processor.h
@@ -376,6 +376,10 @@ union thread_xstate {
         struct xsave_struct             xsave;
 };
 
+struct fpu {
+        union thread_xstate *state;
+};
+
 #ifdef CONFIG_X86_64
 DECLARE_PER_CPU(struct orig_ist, orig_ist);
 
@@ -453,7 +457,7 @@ struct thread_struct {
         unsigned long           trap_no;
         unsigned long           error_code;
         /* floating point and extended processor state */
-        union thread_xstate     *xstate;
+        struct fpu              fpu;
 #ifdef CONFIG_X86_32
         /* Virtual 86 mode info */
         struct vm86_struct __user *vm86_info;
diff --git a/arch/x86/include/asm/thread_info.h b/arch/x86/include/asm/thread_info.h
index d017ed5502e2..d4092fac226b 100644
--- a/arch/x86/include/asm/thread_info.h
+++ b/arch/x86/include/asm/thread_info.h
@@ -242,7 +242,6 @@ static inline struct thread_info *current_thread_info(void)
 #define TS_POLLING              0x0004  /* true if in idle loop
                                            and not sleeping */
 #define TS_RESTORE_SIGMASK      0x0008  /* restore signal mask in do_signal() */
-#define TS_XSAVE                0x0010  /* Use xsave/xrstor */
 
 #define tsk_is_polling(t) (task_thread_info(t)->status & TS_POLLING)
 
diff --git a/arch/x86/include/asm/traps.h b/arch/x86/include/asm/traps.h
index 4da91ad69e0d..f66cda56781d 100644
--- a/arch/x86/include/asm/traps.h
+++ b/arch/x86/include/asm/traps.h
@@ -79,7 +79,7 @@ static inline int get_si_code(unsigned long condition)
 
 extern int panic_on_unrecovered_nmi;
 
-void math_error(void __user *);
+void math_error(struct pt_regs *, int, int);
 void math_emulate(struct math_emu_info *);
 #ifndef CONFIG_X86_32
 asmlinkage void smp_thermal_interrupt(void);
diff --git a/arch/x86/include/asm/xsave.h b/arch/x86/include/asm/xsave.h
index ddc04ccad03b..2c4390cae228 100644
--- a/arch/x86/include/asm/xsave.h
+++ b/arch/x86/include/asm/xsave.h
@@ -37,8 +37,9 @@ extern int check_for_xstate(struct i387_fxsave_struct __user *buf,
                             void __user *fpstate,
                             struct _fpx_sw_bytes *sw);
 
-static inline int xrstor_checking(struct xsave_struct *fx)
+static inline int fpu_xrstor_checking(struct fpu *fpu)
 {
+        struct xsave_struct *fx = &fpu->state->xsave;
         int err;
 
         asm volatile("1: .byte " REX_PREFIX "0x0f,0xae,0x2f\n\t"
@@ -110,12 +111,12 @@ static inline void xrstor_state(struct xsave_struct *fx, u64 mask)
                      :   "memory");
 }
 
-static inline void xsave(struct task_struct *tsk)
+static inline void fpu_xsave(struct fpu *fpu)
 {
         /* This, however, we can work around by forcing the compiler to select
            an addressing mode that doesn't require extended registers. */
         __asm__ __volatile__(".byte " REX_PREFIX "0x0f,0xae,0x27"
-                             : : "D" (&(tsk->thread.xstate->xsave)),
+                             : : "D" (&(fpu->state->xsave)),
                                  "a" (-1), "d"(-1) : "memory");
 }
 #endif