x86: don't use REP_GOOD or ERMS for user memory clearing

The modern target to use is FSRS (Fast Short REP STOS), and the other
cases should only be used for bigger areas (ie mainly things like page
clearing).

Note! This changes the conditional for the inlining from FSRM ("fast
short rep movs") to FSRS ("fast short rep stos").

We'll have a separate fixup for AMD microarchitectures that have a good
'rep stosb' yet do not set the new Intel-specific FSRS bit (because FSRM
was there first).

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Author: torvalds

arch/x86/include/asm/uaccess_64.h

@@ -81,10 +81,6 @@ __copy_from_user_flushcache(void *dst, const void __user *src, unsigned size)
 
 __must_check unsigned long
 clear_user_original(void __user *addr, unsigned long len);
-__must_check unsigned long
-clear_user_rep_good(void __user *addr, unsigned long len);
-__must_check unsigned long
-clear_user_erms(void __user *addr, unsigned long len);
 
 static __always_inline __must_check unsigned long __clear_user(void __user *addr, unsigned long size)
 {
@@ -97,16 +93,12 @@ static __always_inline __must_check unsigned long __clear_user(void __user *addr
 	 */
 	asm volatile(
 		"1:\n\t"
-		ALTERNATIVE_3("rep stosb",
-			      "call clear_user_erms",	  ALT_NOT(X86_FEATURE_FSRM),
-			      "call clear_user_rep_good", ALT_NOT(X86_FEATURE_ERMS),
-			      "call clear_user_original", ALT_NOT(X86_FEATURE_REP_GOOD))
+		ALTERNATIVE("rep stosb",
+			    "call clear_user_original", ALT_NOT(X86_FEATURE_FSRS))
 		"2:\n"
 	       _ASM_EXTABLE_UA(1b, 2b)
 	       : "+c" (size), "+D" (addr), ASM_CALL_CONSTRAINT
-	       : "a" (0)
-		/* rep_good clobbers %rdx */
-	       : "rdx");
+	       : "a" (0));
 
 	clac();
 

arch/x86/lib/clear_page_64.S

@@ -113,78 +113,3 @@ SYM_FUNC_START(clear_user_original)
         _ASM_EXTABLE_UA(.Lbytes, .Lbytes_exception)
 SYM_FUNC_END(clear_user_original)
 EXPORT_SYMBOL(clear_user_original)
-
-/*
- * Alternative clear user-space when CPU feature X86_FEATURE_REP_GOOD is
- * present.
- * Input:
- * rdi destination
- * rcx count
- *
- * Output:
- * rcx: uncleared bytes or 0 if successful.
- */
-SYM_FUNC_START(clear_user_rep_good)
-	# call the original thing for less than a cacheline
-	cmp $64, %rcx
-	jb clear_user_original
-
-.Lprep:
-	# copy lower 32-bits for rest bytes
-	mov %ecx, %edx
-	shr $3, %rcx
-	jz .Lrep_good_rest_bytes
-
-.Lrep_good_qwords:
-	rep stosq
-
-.Lrep_good_rest_bytes:
-	and $7, %edx
-	jz .Lrep_good_exit
-
-.Lrep_good_bytes:
-	mov %edx, %ecx
-	rep stosb
-
-.Lrep_good_exit:
-	# see .Lexit comment above
-	xor %eax, %eax
-	RET
-
-.Lrep_good_qwords_exception:
-	# convert remaining qwords back into bytes to return to caller
-	shl $3, %rcx
-	and $7, %edx
-	add %rdx, %rcx
-	jmp .Lrep_good_exit
-
-	_ASM_EXTABLE_UA(.Lrep_good_qwords, .Lrep_good_qwords_exception)
-	_ASM_EXTABLE_UA(.Lrep_good_bytes, .Lrep_good_exit)
-SYM_FUNC_END(clear_user_rep_good)
-EXPORT_SYMBOL(clear_user_rep_good)
-
-/*
- * Alternative clear user-space when CPU feature X86_FEATURE_ERMS is present.
- * Input:
- * rdi destination
- * rcx count
- *
- * Output:
- * rcx: uncleared bytes or 0 if successful.
- *
- */
-SYM_FUNC_START(clear_user_erms)
-	# call the original thing for less than a cacheline
-	cmp $64, %rcx
-	jb clear_user_original
-
-.Lerms_bytes:
-	rep stosb
-
-.Lerms_exit:
-	xorl %eax,%eax
-	RET
-
-	_ASM_EXTABLE_UA(.Lerms_bytes, .Lerms_exit)
-SYM_FUNC_END(clear_user_erms)
-EXPORT_SYMBOL(clear_user_erms)

tools/objtool/check.c

@@ -1284,8 +1284,6 @@ static const char *uaccess_safe_builtin[] = {
 	"copy_mc_fragile_handle_tail",
 	"copy_mc_enhanced_fast_string",
 	"ftrace_likely_update", /* CONFIG_TRACE_BRANCH_PROFILING */
-	"clear_user_erms",
-	"clear_user_rep_good",
 	"clear_user_original",
 	NULL
 };