x86: move stac/clac from user copy routines into callers

This is preparatory work for inlining the 'rep movs' case, but also a
cleanup.  The __copy_user_nocache() function was mis-used by the rdma
code to do uncached kernel copies that don't actually want user copies
at all, and as a result doesn't want the stac/clac either.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

arch/x86/include/asm/uaccess_64.h

@@ -27,6 +27,7 @@ copy_user_generic(void *to, const void *from, unsigned len)
 {
 	unsigned ret;
 
+	stac();
 	/*
 	 * If CPU has FSRM feature, use 'rep movs'.
 	 * Otherwise, use copy_user_generic_unrolled.
@@ -38,6 +39,7 @@ copy_user_generic(void *to, const void *from, unsigned len)
 				     "=d" (len)),
 			 "1" (to), "2" (from), "3" (len)
 			 : "memory", "rcx", "r8", "r9", "r10", "r11");
+	clac();
 	return ret;
 }
 
@@ -64,8 +66,12 @@ static inline int
 __copy_from_user_inatomic_nocache(void *dst, const void __user *src,
 				  unsigned size)
 {
+	long ret;
 	kasan_check_write(dst, size);
-	return __copy_user_nocache(dst, src, size, 0);
+	stac();
+	ret = __copy_user_nocache(dst, src, size, 0);
+	clac();
+	return ret;
 }
 
 static inline int

arch/x86/lib/copy_user_64.S

@@ -51,7 +51,6 @@
  * eax uncopied bytes or 0 if successful.
  */
 SYM_FUNC_START(copy_user_generic_unrolled)
-	ASM_STAC
 	cmpl $8,%edx
 	jb .Lcopy_user_short_string_bytes
 	ALIGN_DESTINATION
@@ -123,15 +122,12 @@ EXPORT_SYMBOL(copy_user_generic_unrolled)
  * eax uncopied bytes or 0 if successful.
  */
 SYM_FUNC_START(copy_user_fast_string)
-	ASM_STAC
 	movl %edx,%ecx
 1:	rep movsb
 	xorl %eax,%eax
-	ASM_CLAC
 	RET
 
 12:	movl %ecx,%eax		/* ecx is zerorest also */
-	ASM_CLAC
 	RET
 
 	_ASM_EXTABLE_CPY(1b, 12b)
@@ -160,12 +156,10 @@ SYM_CODE_START_LOCAL(.Lcopy_user_handle_tail)
 	movl %edx,%ecx
 1:	rep movsb
 2:	mov %ecx,%eax
-	ASM_CLAC
 	RET
 
 3:
 	movl %edx,%eax
-	ASM_CLAC
 	RET
 
 	_ASM_EXTABLE_CPY(1b, 2b)
@@ -209,7 +203,6 @@ SYM_CODE_START_LOCAL(copy_user_short_string)
 	decl %ecx
 	jnz 21b
 23:	xor %eax,%eax
-	ASM_CLAC
 	RET
 
 40:	leal (%rdx,%rcx,8),%edx
@@ -233,8 +226,6 @@ SYM_CODE_END(copy_user_short_string)
  *  - Require 4-byte alignment when size is 4 bytes.
  */
 SYM_FUNC_START(__copy_user_nocache)
-	ASM_STAC
-
 	/* If size is less than 8 bytes, go to 4-byte copy */
 	cmpl $8,%edx
 	jb .L_4b_nocache_copy_entry
@@ -327,7 +318,6 @@ SYM_FUNC_START(__copy_user_nocache)
 	/* Finished copying; fence the prior stores */
 .L_finish_copy:
 	xorl %eax,%eax
-	ASM_CLAC
 	sfence
 	RET
 

arch/x86/lib/usercopy_64.c

@@ -45,7 +45,11 @@ EXPORT_SYMBOL_GPL(arch_wb_cache_pmem);
 long __copy_user_flushcache(void *dst, const void __user *src, unsigned size)
 {
 	unsigned long flushed, dest = (unsigned long) dst;
-	long rc = __copy_user_nocache(dst, src, size, 0);
+	long rc;
+
+	stac();
+	rc = __copy_user_nocache(dst, src, size, 0);
+	clac();
 
 	/*
 	 * __copy_user_nocache() uses non-temporal stores for the bulk

tools/objtool/check.c

@@ -1285,6 +1285,9 @@ static const char *uaccess_safe_builtin[] = {
 	"copy_mc_enhanced_fast_string",
 	"ftrace_likely_update", /* CONFIG_TRACE_BRANCH_PROFILING */
 	"clear_user_original",
+	"copy_user_generic_unrolled",
+	"copy_user_fast_string",
+	"__copy_user_nocache",
 	NULL
 };