Add a clts() wrapper around the 'clts' instruction to <machine/cpufunc.h>

on x86 and use that to implement stop_emulating() in the fpu/npx code.
Reimplement start_emulating() in the non-XEN case by using load_cr0() and
rcr0() instead of the 'lmsw' and 'smsw' instructions.  Intel explicitly
discourages the use of 'lmsw' and 'smsw' on 80386 and later processors in
the description of these instructions in Volume 2 of the ADM.

Reviewed by:	kib
MFC after:	1 month

4 files changed, 31 insertions, 17 deletions

diff --git a/sys/amd64/amd64/fpu.c b/sys/amd64/amd64/fpu.c
index 77e11bc..ca951ad 100644
--- a/sys/amd64/amd64/fpu.c
+++ b/sys/amd64/amd64/fpu.c
@@ -73,10 +73,6 @@ __FBSDID("$FreeBSD$");
 #define	fxrstor(addr)		__asm __volatile("fxrstor %0" : : "m" (*(addr)))
 #define	fxsave(addr)		__asm __volatile("fxsave %0" : "=m" (*(addr)))
 #define	ldmxcsr(csr)		__asm __volatile("ldmxcsr %0" : : "m" (csr))
-#define	start_emulating()	__asm __volatile( \
-				    "smsw %%ax; orb %0,%%al; lmsw %%ax" \
-				    : : "n" (CR0_TS) : "ax")
-#define	stop_emulating()	__asm __volatile("clts")
 
 static __inline void
 xrstor(char *addr, uint64_t mask)
@@ -109,13 +105,14 @@ void	fnstsw(caddr_t addr);
 void	fxsave(caddr_t addr);
 void	fxrstor(caddr_t addr);
 void	ldmxcsr(u_int csr);
-void	start_emulating(void);
-void	stop_emulating(void);
 void	xrstor(char *addr, uint64_t mask);
 void	xsave(char *addr, uint64_t mask);
 
 #endif	/* __GNUCLIKE_ASM && !lint */
 
+#define	start_emulating()	load_cr0(rcr0() | CR0_TS)
+#define	stop_emulating()	clts()
+
 #define GET_FPU_CW(thread) ((thread)->td_pcb->pcb_save->sv_env.en_cw)
 #define GET_FPU_SW(thread) ((thread)->td_pcb->pcb_save->sv_env.en_sw)
 
diff --git a/sys/amd64/include/cpufunc.h b/sys/amd64/include/cpufunc.h
index 829efc1..94d4133 100644
--- a/sys/amd64/include/cpufunc.h
+++ b/sys/amd64/include/cpufunc.h
@@ -107,6 +107,13 @@ clflush(u_long addr)
 }
 
 static __inline void
+clts(void)
+{
+
+	__asm __volatile("clts");
+}
+
+static __inline void
 disable_intr(void)
 {
 	__asm __volatile("cli" : : : "memory");
@@ -702,6 +709,9 @@ intr_restore(register_t rflags)
 int	breakpoint(void);
 u_int	bsfl(u_int mask);
 u_int	bsrl(u_int mask);
+void	clflush(u_long addr);
+void	clts(void);
+void	cpuid_count(u_int ax, u_int cx, u_int *p);
 void	disable_intr(void);
 void	do_cpuid(u_int ax, u_int *p);
 void	enable_intr(void);
diff --git a/sys/i386/include/cpufunc.h b/sys/i386/include/cpufunc.h
index 7116f76..62d268d 100644
--- a/sys/i386/include/cpufunc.h
+++ b/sys/i386/include/cpufunc.h
@@ -97,6 +97,13 @@ clflush(u_long addr)
 }
 
 static __inline void
+clts(void)
+{
+
+	__asm __volatile("clts");
+}
+
+static __inline void
 disable_intr(void)
 {
 #ifdef XEN
@@ -688,6 +695,9 @@ intr_restore(register_t eflags)
 int	breakpoint(void);
 u_int	bsfl(u_int mask);
 u_int	bsrl(u_int mask);
+void	clflush(u_long addr);
+void	clts(void);
+void	cpuid_count(u_int ax, u_int cx, u_int *p);
 void	disable_intr(void);
 void	do_cpuid(u_int ax, u_int *p);
 void	enable_intr(void);
diff --git a/sys/i386/isa/npx.c b/sys/i386/isa/npx.c
index 0da580f..50c812c 100644
--- a/sys/i386/isa/npx.c
+++ b/sys/i386/isa/npx.c
@@ -100,15 +100,6 @@ __FBSDID("$FreeBSD$");
 #define	fxrstor(addr)		__asm __volatile("fxrstor %0" : : "m" (*(addr)))
 #define	fxsave(addr)		__asm __volatile("fxsave %0" : "=m" (*(addr)))
 #endif
-#ifdef XEN
-#define	start_emulating()	(HYPERVISOR_fpu_taskswitch(1))
-#define	stop_emulating()	(HYPERVISOR_fpu_taskswitch(0))
-#else
-#define	start_emulating()	__asm __volatile( \
-				    "smsw %%ax; orb %0,%%al; lmsw %%ax" \
-				    : : "n" (CR0_TS) : "ax")
-#define	stop_emulating()	__asm __volatile("clts")
-#endif
 #else	/* !(__GNUCLIKE_ASM && !lint) */
 
 void	fldcw(u_short cw);
@@ -123,11 +114,17 @@ void	frstor(caddr_t addr);
 void	fxsave(caddr_t addr);
 void	fxrstor(caddr_t addr);
 #endif
-void	start_emulating(void);
-void	stop_emulating(void);
 
 #endif	/* __GNUCLIKE_ASM && !lint */
 
+#ifdef XEN
+#define	start_emulating()	(HYPERVISOR_fpu_taskswitch(1))
+#define	stop_emulating()	(HYPERVISOR_fpu_taskswitch(0))
+#else
+#define	start_emulating()	load_cr0(rcr0() | CR0_TS)
+#define	stop_emulating()	clts()
+#endif
+
 #ifdef CPU_ENABLE_SSE
 #define GET_FPU_CW(thread) \
 	(cpu_fxsr ? \