x86/paravirt: add sysret/sysexit pvops for returning to 32-bit compatibility userspace

In a 64-bit system, we need separate sysret/sysexit operations to
return to a 32-bit userspace.

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/ia32/ia32entry.S

@@ -61,6 +61,19 @@
 	CFI_UNDEFINED	r15
 	.endm
 
+#ifdef CONFIG_PARAVIRT
+ENTRY(native_usergs_sysret32)
+	swapgs
+	sysretl
+ENDPROC(native_usergs_sysret32)
+
+ENTRY(native_irq_enable_sysexit)
+	swapgs
+	sti
+	sysexit
+ENDPROC(native_irq_enable_sysexit)
+#endif
+
 /*
  * 32bit SYSENTER instruction entry.
  *
@@ -151,10 +164,7 @@ sysenter_do_call:
 	CFI_ADJUST_CFA_OFFSET -8
 	CFI_REGISTER rsp,rcx
 	TRACE_IRQS_ON
-	swapgs
+	ENABLE_INTERRUPTS_SYSEXIT32
-	sti		/* sti only takes effect after the next instruction */
-	/* sysexit */
-	.byte	0xf, 0x35
 
 sysenter_tracesys:
 	CFI_RESTORE_STATE
@@ -254,8 +264,7 @@ cstar_do_call:
 	TRACE_IRQS_ON
 	movl RSP-ARGOFFSET(%rsp),%esp
 	CFI_RESTORE rsp
-	swapgs
+	USERGS_SYSRET32
-	sysretl
 	
 cstar_tracesys:	
 	CFI_RESTORE_STATE

arch/x86/kernel/asm-offsets_64.c

@@ -62,7 +62,9 @@ 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_usergs_sysret, pv_cpu_ops, usergs_sysret);
+	OFFSET(PV_CPU_usergs_sysret32, pv_cpu_ops, usergs_sysret32);
+	OFFSET(PV_CPU_usergs_sysret64, pv_cpu_ops, usergs_sysret64);
+	OFFSET(PV_CPU_irq_enable_sysexit, pv_cpu_ops, irq_enable_sysexit);
 	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

@@ -59,7 +59,7 @@
 #endif	
 
 #ifdef CONFIG_PARAVIRT
-ENTRY(native_usergs_sysret)
+ENTRY(native_usergs_sysret64)
 	swapgs
 	sysretq
 #endif /* CONFIG_PARAVIRT */
@@ -275,7 +275,7 @@ sysret_check:
 	RESTORE_ARGS 0,-ARG_SKIP,1
 	/*CFI_REGISTER	rflags,r11*/
 	movq	%gs:pda_oldrsp, %rsp
-	USERGS_SYSRET
+	USERGS_SYSRET64
 
 	CFI_RESTORE_STATE
 	/* Handle reschedules */

arch/x86/kernel/paravirt.c

@@ -141,7 +141,8 @@ 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.usergs_sysret))
+		 type == PARAVIRT_PATCH(pv_cpu_ops.usergs_sysret32) ||
+		 type == PARAVIRT_PATCH(pv_cpu_ops.usergs_sysret64))
 		/* If operation requires a jmp, then jmp */
 		ret = paravirt_patch_jmp(insnbuf, opfunc, addr, len);
 	else
@@ -193,7 +194,8 @@ 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_usergs_sysret(void);
+extern void native_usergs_sysret32(void);
+extern void native_usergs_sysret64(void);
 
 static int __init print_banner(void)
 {
@@ -329,10 +331,10 @@ struct pv_cpu_ops pv_cpu_ops = {
 	.write_idt_entry = native_write_idt_entry,
 	.load_sp0 = native_load_sp0,
 
-#ifdef CONFIG_X86_32
 	.irq_enable_sysexit = native_irq_enable_sysexit,
-#else
+#ifdef CONFIG_X86_64
-	.usergs_sysret = native_usergs_sysret,
+	.usergs_sysret32 = native_usergs_sysret32,
+	.usergs_sysret64 = native_usergs_sysret64,
 #endif
 	.iret = native_iret,
 	.swapgs = native_swapgs,

arch/x86/kernel/paravirt_patch_64.c

@@ -14,8 +14,9 @@ DEF_NATIVE(pv_mmu_ops, flush_tlb_single, "invlpg (%rdi)");
 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, irq_enable_sysexit, "swapgs; sti; sysexit");
-DEF_NATIVE(pv_cpu_ops, usergs_sysret, "swapgs; sysretq;");
+DEF_NATIVE(pv_cpu_ops, usergs_sysret64, "swapgs; sysretq");
+DEF_NATIVE(pv_cpu_ops, usergs_sysret32, "swapgs; sysretl");
 DEF_NATIVE(pv_cpu_ops, swapgs, "swapgs");
 
 unsigned native_patch(u8 type, u16 clobbers, void *ibuf,
@@ -35,7 +36,9 @@ 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, usergs_sysret);
+		PATCH_SITE(pv_cpu_ops, irq_enable_sysexit);
+		PATCH_SITE(pv_cpu_ops, usergs_sysret32);
+		PATCH_SITE(pv_cpu_ops, usergs_sysret64);
 		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

@@ -112,9 +112,17 @@ static inline unsigned long __raw_local_irq_save(void)
 
 #ifdef CONFIG_X86_64
 #define INTERRUPT_RETURN	iretq
-#define USERGS_SYSRET					\
+#define USERGS_SYSRET64				\
-			swapgs;				\
+	swapgs;					\
-			sysretq;
+	sysretq;
+#define USERGS_SYSRET32				\
+	swapgs;					\
+	sysretl
+#define ENABLE_INTERRUPTS_SYSEXIT32		\
+	swapgs;					\
+	sti;					\
+	sysexit
+
 #else
 #define INTERRUPT_RETURN		iret
 #define ENABLE_INTERRUPTS_SYSEXIT	sti; sysexit

include/asm-x86/paravirt.h

@@ -141,9 +141,32 @@ struct pv_cpu_ops {
 	u64 (*read_pmc)(int counter);
 	unsigned long long (*read_tscp)(unsigned int *aux);
 
-	/* These three are jmp to, not actually called. */
+	/*
+	 * Atomically enable interrupts and return to userspace.  This
+	 * is only ever used to return to 32-bit processes; in a
+	 * 64-bit kernel, it's used for 32-on-64 compat processes, but
+	 * never native 64-bit processes.  (Jump, not call.)
+	 */
 	void (*irq_enable_sysexit)(void);
-	void (*usergs_sysret)(void);
+
+	/*
+	 * Switch to usermode gs and return to 64-bit usermode using
+	 * sysret.  Only used in 64-bit kernels to return to 64-bit
+	 * processes.  Usermode register state, including %rsp, must
+	 * already be restored.
+	 */
+	void (*usergs_sysret64)(void);
+
+	/*
+	 * Switch to usermode gs and return to 32-bit usermode using
+	 * sysret.  Used to return to 32-on-64 compat processes.
+	 * Other usermode register state, including %esp, must already
+	 * be restored.
+	 */
+	void (*usergs_sysret32)(void);
+
+	/* Normal iret.  Jump to this with the standard iret stack
+	   frame set up. */
 	void (*iret)(void);
 
 	void (*swapgs)(void);
@@ -1481,18 +1504,24 @@ static inline unsigned long __raw_local_irq_save(void)
 		  call PARA_INDIRECT(pv_irq_ops+PV_IRQ_irq_enable);	\
 		  PV_RESTORE_REGS;)
 
-#define ENABLE_INTERRUPTS_SYSEXIT					\
+#define USERGS_SYSRET32							\
-	PARA_SITE(PARA_PATCH(pv_cpu_ops, PV_CPU_irq_enable_sysexit),	\
+	PARA_SITE(PARA_PATCH(pv_cpu_ops, PV_CPU_usergs_sysret32),	\
 		  CLBR_NONE,						\
-		  jmp PARA_INDIRECT(pv_cpu_ops+PV_CPU_irq_enable_sysexit))
+		  jmp PARA_INDIRECT(pv_cpu_ops+PV_CPU_usergs_sysret32))
-
 
 #ifdef CONFIG_X86_32
 #define GET_CR0_INTO_EAX				\
 	push %ecx; push %edx;				\
 	call PARA_INDIRECT(pv_cpu_ops+PV_CPU_read_cr0);	\
 	pop %edx; pop %ecx
-#else
+
+#define ENABLE_INTERRUPTS_SYSEXIT					\
+	PARA_SITE(PARA_PATCH(pv_cpu_ops, PV_CPU_irq_enable_sysexit),	\
+		  CLBR_NONE,						\
+		  jmp PARA_INDIRECT(pv_cpu_ops+PV_CPU_irq_enable_sysexit))
+
+
+#else	/* !CONFIG_X86_32 */
 #define SWAPGS								\
 	PARA_SITE(PARA_PATCH(pv_cpu_ops, PV_CPU_swapgs), CLBR_NONE,	\
 		  PV_SAVE_REGS;						\
@@ -1505,11 +1534,16 @@ static inline unsigned long __raw_local_irq_save(void)
 	movq %rax, %rcx;				\
 	xorq %rax, %rax;
 
-#define USERGS_SYSRET							\
+#define USERGS_SYSRET64							\
-	PARA_SITE(PARA_PATCH(pv_cpu_ops, PV_CPU_usergs_sysret),		\
+	PARA_SITE(PARA_PATCH(pv_cpu_ops, PV_CPU_usergs_sysret64),	\
 		  CLBR_NONE,						\
-		  jmp PARA_INDIRECT(pv_cpu_ops+PV_CPU_usergs_sysret))
+		  jmp PARA_INDIRECT(pv_cpu_ops+PV_CPU_usergs_sysret64))
-#endif
+
+#define ENABLE_INTERRUPTS_SYSEXIT32					\
+	PARA_SITE(PARA_PATCH(pv_cpu_ops, PV_CPU_irq_enable_sysexit),	\
+		  CLBR_NONE,						\
+		  jmp PARA_INDIRECT(pv_cpu_ops+PV_CPU_irq_enable_sysexit))
+#endif	/* CONFIG_X86_32 */
 
 #endif /* __ASSEMBLY__ */
 #endif /* CONFIG_PARAVIRT */