x86/uaccess: Zero the 8-byte get_range case on failure on 32-bit

While zeroing the upper 32 bits of an 8-byte getuser on 32-bit x86 was
fixed by commit 8c860ed ("x86/uaccess: Fix missed zeroing of ia32 u64
get_user() range checking") it was broken again in commit 8a2462d
("x86/uaccess: Improve the 8-byte getuser() case").

This is because the register which holds the upper 32 bits (%ecx) is being
cleared _after_ the check_range, so if the range check fails, %ecx is never
cleared.

This can be reproduced with:
./tools/testing/kunit/kunit.py run --arch i386 usercopy

Instead, clear %ecx _before_ check_range in the 8-byte case. This
reintroduces a bit of the ugliness we were trying to avoid by adding
another #ifndef CONFIG_X86_64, but at least keeps check_range from needing
a separate bad_get_user_8 jump.

Fixes: 8a2462d ("x86/uaccess: Improve the 8-byte getuser() case")
Signed-off-by: David Gow <[email protected]>
Signed-off-by: Thomas Gleixner <[email protected]>
Acked-by: Linus Torvalds <[email protected]>
Link: https://lore.kernel.org/all/[email protected]

Author: sulix

arch/x86/lib/getuser.S

@@ -88,12 +88,14 @@ SYM_FUNC_END(__get_user_4)
 EXPORT_SYMBOL(__get_user_4)
 
 SYM_FUNC_START(__get_user_8)
+#ifndef CONFIG_X86_64
+	xor %ecx,%ecx
+#endif
 	check_range size=8
 	ASM_STAC
 #ifdef CONFIG_X86_64
 	UACCESS movq (%_ASM_AX),%rdx
 #else
-	xor %ecx,%ecx
 	UACCESS movl (%_ASM_AX),%edx
 	UACCESS movl 4(%_ASM_AX),%ecx
 #endif