[PATCH 3.2 093/104] x86/retpoline: Avoid retpolines for built-in __init functions

From: Ben Hutchings
Date: Sun Mar 11 2018 - 23:49:12 EST


3.2.101-rc1 review patch. If anyone has any objections, please let me know.

------------------

From: David Woodhouse <dwmw@xxxxxxxxxxxx>

commit 66f793099a636862a71c59d4a6ba91387b155e0c upstream.

There's no point in building init code with retpolines, since it runs before
any potentially hostile userspace does. And before the retpoline is actually
ALTERNATIVEd into place, for much of it.

Signed-off-by: David Woodhouse <dwmw@xxxxxxxxxxxx>
Signed-off-by: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
Cc: karahmed@xxxxxxxxx
Cc: peterz@xxxxxxxxxxxxx
Cc: bp@xxxxxxxxx
Link: https://lkml.kernel.org/r/1517484441-1420-2-git-send-email-dwmw@xxxxxxxxxxxx
[bwh: Backported to 3.2: adjust context]
Signed-off-by: Ben Hutchings <ben@xxxxxxxxxxxxxxx>
---
include/linux/init.h | 9 ++++++++-
1 file changed, 8 insertions(+), 1 deletion(-)

--- a/include/linux/init.h
+++ b/include/linux/init.h
@@ -3,6 +3,13 @@

#include <linux/compiler.h>

+/* Built-in __init functions needn't be compiled with retpoline */
+#if defined(RETPOLINE) && !defined(MODULE)
+#define __noretpoline __attribute__((indirect_branch("keep")))
+#else
+#define __noretpoline
+#endif
+
/* These macros are used to mark some functions or
* initialized data (doesn't apply to uninitialized data)
* as `initialization' functions. The kernel can take this
@@ -40,7 +47,7 @@

/* These are for everybody (although not all archs will actually
discard it in modules) */
-#define __init __section(.init.text) __cold notrace
+#define __init __section(.init.text) __cold notrace __noretpoline
#define __initdata __section(.init.data)
#define __initconst __section(.init.rodata)
#define __exitdata __section(.exit.data)