[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <tip-3d1e236022cc1426b0834565995ddee2ca231cee@git.kernel.org>
Date: Sat, 25 Feb 2017 02:22:34 -0800
From: tip-bot for Josh Poimboeuf <tipbot@...or.com>
To: linux-tip-commits@...r.kernel.org
Cc: tglx@...utronix.de, peterz@...radead.org, fengguang.wu@...el.com,
jpoimboe@...hat.com, torvalds@...ux-foundation.org,
mingo@...nel.org, linux-kernel@...r.kernel.org, hpa@...or.com
Subject: [tip:core/urgent] objtool: Prevent GCC from merging
annotate_unreachable()
Commit-ID: 3d1e236022cc1426b0834565995ddee2ca231cee
Gitweb: http://git.kernel.org/tip/3d1e236022cc1426b0834565995ddee2ca231cee
Author: Josh Poimboeuf <jpoimboe@...hat.com>
AuthorDate: Fri, 24 Feb 2017 22:31:02 -0600
Committer: Ingo Molnar <mingo@...nel.org>
CommitDate: Sat, 25 Feb 2017 10:11:23 +0100
objtool: Prevent GCC from merging annotate_unreachable()
0-day bot reported some new objtool warnings which were caused by the
new annotate_unreachable() macro:
fs/afs/flock.o: warning: objtool: afs_do_unlk()+0x0: duplicate frame pointer save
fs/afs/flock.o: warning: objtool: afs_do_unlk()+0x0: frame pointer state mismatch
fs/btrfs/delayed-inode.o: warning: objtool: btrfs_delete_delayed_dir_index()+0x0: duplicate frame pointer save
fs/btrfs/delayed-inode.o: warning: objtool: btrfs_delete_delayed_dir_index()+0x0: frame pointer state mismatch
fs/dlm/lock.o: warning: objtool: _grant_lock()+0x0: duplicate frame pointer save
fs/dlm/lock.o: warning: objtool: _grant_lock()+0x0: frame pointer state mismatch
fs/ocfs2/alloc.o: warning: objtool: ocfs2_mv_path()+0x0: duplicate frame pointer save
fs/ocfs2/alloc.o: warning: objtool: ocfs2_mv_path()+0x0: frame pointer state mismatch
It turns out that, for older versions of GCC, if a function has multiple
BUG() incantations, GCC will sometimes merge the corresponding
annotate_unreachable() inline asm statements into a single block. That
has the undesirable effect of removing one of the entries in the
__unreachable section, confusing objtool greatly.
A workaround for this issue is to ensure that each instance of the
inline asm statement uses a different label, so that GCC sees the
statements are unique and leaves them alone. The inline asm ‘%=’ token
could be used for that, but unfortunately older versions of GCC don't
support it. So I implemented a poor man's version of it with the
__LINE__ macro.
Reported-by: kbuild test robot <fengguang.wu@...el.com>
Signed-off-by: Josh Poimboeuf <jpoimboe@...hat.com>
Cc: Linus Torvalds <torvalds@...ux-foundation.org>
Cc: Peter Zijlstra <peterz@...radead.org>
Cc: Thomas Gleixner <tglx@...utronix.de>
Fixes: d1091c7fa3d5 ("objtool: Improve detection of BUG() and other dead ends")
Link: http://lkml.kernel.org/r/0c14b00baf9f68d1b0221ddb6c88b925181c8be8.1487997036.git.jpoimboe@redhat.com
Signed-off-by: Ingo Molnar <mingo@...nel.org>
---
include/linux/compiler-gcc.h | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/include/linux/compiler-gcc.h b/include/linux/compiler-gcc.h
index 8ea159f..de47134 100644
--- a/include/linux/compiler-gcc.h
+++ b/include/linux/compiler-gcc.h
@@ -197,10 +197,10 @@
#ifdef CONFIG_STACK_VALIDATION
#define annotate_unreachable() ({ \
- asm("1:\t\n" \
+ asm("%c0:\t\n" \
".pushsection __unreachable, \"a\"\t\n" \
- ".long 1b\t\n" \
- ".popsection\t\n"); \
+ ".long %c0b\t\n" \
+ ".popsection\t\n" : : "i" (__LINE__)); \
})
#else
#define annotate_unreachable()
Powered by blists - more mailing lists