[PATCH] strub: avoid lto inlining

[Alexandre Oliva <oliva@adacore.com>]

The strub builtins are not suited for cross-unit inlining, they should
only be inlined by the builtin expanders, if at all.  While testing on
sparc64, it occurred to me that, if libgcc was built with LTO enabled,
lto1 might inline them, and that would likely break things.  So, make
sure they're clearly marked as not inlinable.

Regstrapped on x86_64-linux-gnu, also testing on sparc-solaris2.11.3.
Ok to install?


for  libgcc/ChangeLog

	* strub.c (ATTRIBUTE_NOINLINE): New.
	(ATTRIBUTE_STRUB_CALLABLE): Add it.
	(__strub_dummy_force_no_leaf): Drop it.
---
 libgcc/strub.c |    8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/libgcc/strub.c b/libgcc/strub.c
index b0f990d9deebb..5062554d0e1e6 100644
--- a/libgcc/strub.c
+++ b/libgcc/strub.c
@@ -36,7 +36,12 @@ see the files COPYING3 and COPYING.RUNTIME respectively.  If not, see
 # define TOPS <
 #endif
 
-#define ATTRIBUTE_STRUB_CALLABLE __attribute__ ((__strub__ ("callable")))
+/* Make sure these builtins won't be inlined, even with LTO.  */
+#define ATTRIBUTE_NOINLINE \
+  __attribute__ ((__noinline__, __noclone__))
+
+#define ATTRIBUTE_STRUB_CALLABLE \
+  __attribute__ ((__strub__ ("callable"))) ATTRIBUTE_NOINLINE
 
 /* Enter a stack scrubbing context, initializing the watermark to the caller's
    stack address.  */
@@ -72,7 +77,6 @@ __strub_update (void **watermark)
 /* Dummy function, called to force the caller to not be a leaf function, so
    that it can't use the red zone.  */
 static void ATTRIBUTE_STRUB_CALLABLE
-__attribute__ ((__noinline__, __noipa__))
 __strub_dummy_force_no_leaf (void)
 {
 }

-- 
Alexandre Oliva, happy hacker            https://FSFLA.org/blogs/lxo/
   Free Software Activist                   GNU Toolchain Engineer
More tolerance and less prejudice are key for inclusion and diversity
Excluding neuro-others for not behaving ""normal"" is *not* inclusive