x86/paravirt, 64-bit: don't restore user rsp within sysret

There's no need to combine restoring the user rsp within the sysret
pvop, so split it out.  This makes the pvop's semantics closer to the
machine instruction.

Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citirx.com>
Cc: xen-devel <xen-devel@lists.xensource.com>
Cc: Stephen Tweedie <sct@redhat.com>
Cc: Eduardo Habkost <ehabkost@redhat.com>
Cc: Mark McLoughlin <markmc@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

arch/x86/kernel/asm-offsets_64.c

@@ -62,7 +62,7 @@ int main(void)
 	OFFSET(PV_IRQ_irq_disable, pv_irq_ops, irq_disable);
 	OFFSET(PV_IRQ_irq_enable, pv_irq_ops, irq_enable);
 	OFFSET(PV_CPU_iret, pv_cpu_ops, iret);
-	OFFSET(PV_CPU_usersp_sysret, pv_cpu_ops, usersp_sysret);
+	OFFSET(PV_CPU_usergs_sysret, pv_cpu_ops, usergs_sysret);
 	OFFSET(PV_CPU_swapgs, pv_cpu_ops, swapgs);
 	OFFSET(PV_MMU_read_cr2, pv_mmu_ops, read_cr2);
 #endif

arch/x86/kernel/entry_64.S

@@ -59,8 +59,7 @@
 #endif	
 
 #ifdef CONFIG_PARAVIRT
-ENTRY(native_usersp_sysret)
-	movq	%gs:pda_oldrsp,%rsp
+ENTRY(native_usergs_sysret)
 	swapgs
 	sysretq
 #endif /* CONFIG_PARAVIRT */
@@ -275,7 +274,8 @@ sysret_check:
 	CFI_REGISTER	rip,rcx
 	RESTORE_ARGS 0,-ARG_SKIP,1
 	/*CFI_REGISTER	rflags,r11*/
-	USERSP_SYSRET
+	movq	%gs:pda_oldrsp, %rsp
+	USERGS_SYSRET
 
 	CFI_RESTORE_STATE
 	/* Handle reschedules */

arch/x86/kernel/paravirt.c

@@ -141,7 +141,7 @@ unsigned paravirt_patch_default(u8 type, u16 clobbers, void *insnbuf,
 		ret = paravirt_patch_nop();
 	else if (type == PARAVIRT_PATCH(pv_cpu_ops.iret) ||
 		 type == PARAVIRT_PATCH(pv_cpu_ops.irq_enable_sysexit) ||
-		 type == PARAVIRT_PATCH(pv_cpu_ops.usersp_sysret))
+		 type == PARAVIRT_PATCH(pv_cpu_ops.usergs_sysret))
 		/* If operation requires a jmp, then jmp */
 		ret = paravirt_patch_jmp(insnbuf, opfunc, addr, len);
 	else
@@ -193,7 +193,7 @@ static void native_flush_tlb_single(unsigned long addr)
 /* These are in entry.S */
 extern void native_iret(void);
 extern void native_irq_enable_sysexit(void);
-extern void native_usersp_sysret(void);
+extern void native_usergs_sysret(void);
 
 static int __init print_banner(void)
 {
@@ -332,7 +332,7 @@ struct pv_cpu_ops pv_cpu_ops = {
 #ifdef CONFIG_X86_32
 	.irq_enable_sysexit = native_irq_enable_sysexit,
 #else
-	.usersp_sysret = native_usersp_sysret,
+	.usergs_sysret = native_usergs_sysret,
 #endif
 	.iret = native_iret,
 	.swapgs = native_swapgs,

arch/x86/kernel/paravirt_patch_64.c

@@ -15,7 +15,7 @@ DEF_NATIVE(pv_cpu_ops, clts, "clts");
 DEF_NATIVE(pv_cpu_ops, wbinvd, "wbinvd");
 
 /* the three commands give us more control to how to return from a syscall */
-DEF_NATIVE(pv_cpu_ops, usersp_sysret, "movq %gs:" __stringify(pda_oldrsp) ", %rsp; swapgs; sysretq;");
+DEF_NATIVE(pv_cpu_ops, usergs_sysret, "swapgs; sysretq;");
 DEF_NATIVE(pv_cpu_ops, swapgs, "swapgs");
 
 unsigned native_patch(u8 type, u16 clobbers, void *ibuf,
@@ -35,7 +35,7 @@ unsigned native_patch(u8 type, u16 clobbers, void *ibuf,
 		PATCH_SITE(pv_irq_ops, irq_enable);
 		PATCH_SITE(pv_irq_ops, irq_disable);
 		PATCH_SITE(pv_cpu_ops, iret);
-		PATCH_SITE(pv_cpu_ops, usersp_sysret);
+		PATCH_SITE(pv_cpu_ops, usergs_sysret);
 		PATCH_SITE(pv_cpu_ops, swapgs);
 		PATCH_SITE(pv_mmu_ops, read_cr2);
 		PATCH_SITE(pv_mmu_ops, read_cr3);

include/asm-x86/irqflags.h

@@ -112,8 +112,7 @@ static inline unsigned long __raw_local_irq_save(void)
 
 #ifdef CONFIG_X86_64
 #define INTERRUPT_RETURN	iretq
-#define USERSP_SYSRET					\
-			movq	%gs:pda_oldrsp, %rsp;	\
+#define USERGS_SYSRET					\
 			swapgs;				\
 			sysretq;
 #else

include/asm-x86/paravirt.h

@@ -143,7 +143,7 @@ struct pv_cpu_ops {
 
 	/* These three are jmp to, not actually called. */
 	void (*irq_enable_sysexit)(void);
-	void (*usersp_sysret)(void);
+	void (*usergs_sysret)(void);
 	void (*iret)(void);
 
 	void (*swapgs)(void);
@@ -1505,10 +1505,10 @@ static inline unsigned long __raw_local_irq_save(void)
 	movq %rax, %rcx;				\
 	xorq %rax, %rax;
 
-#define USERSP_SYSRET							\
-	PARA_SITE(PARA_PATCH(pv_cpu_ops, PV_CPU_usersp_sysret),		\
+#define USERGS_SYSRET							\
+	PARA_SITE(PARA_PATCH(pv_cpu_ops, PV_CPU_usergs_sysret),		\
 		  CLBR_NONE,						\
-		  jmp PARA_INDIRECT(pv_cpu_ops+PV_CPU_usersp_sysret))
+		  jmp PARA_INDIRECT(pv_cpu_ops+PV_CPU_usergs_sysret))
 #endif
 
 #endif /* __ASSEMBLY__ */