[LLVMdev] Program compiled with Clang -pg and -O crashes with SEGFAULT

[Qiao Yang]

Hi,

I am trying to compile a simple program with Clang 3.3  on Linux and used -pg and -O2 option. The program would crash with segfault. Interestingly if I compile it with -pg option only it works. Do you have any idea why it crashes? And any workaround?

$ cat myprog.c 
int main() {
    return 0;
}

$ clang -v -pg -O2 myprog.c 
clang version 3.3 (tags/RELEASE_33/final)
Target: x86_64-pc-linux-gnu
Thread model: posix
 "/usr/bin/clang" -cc1 -triple x86_64-pc-linux-gnu -emit-obj -disable-free -disable-llvm-verifier -main-file-name myprog.c -mrelocation-model static -fmath-errno -masm-verbose -mconstructor-aliases -munwind-tables -fuse-init-array -target-cpu x86-64 -target-linker-version 2.22 -momit-leaf-frame-pointer -v -resource-dir /usr/bin/../lib/clang/3.3 -internal-isystem /usr/local/include -internal-isystem /usr/bin/../lib/clang/3.3/include -internal-externc-isystem /usr/include/x86_64-linux-gnu -internal-externc-isystem /include -internal-externc-isystem /usr/include -O2 -fdebug-compilation-dir /home/vagrant/work/c++ -ferror-limit 19 -fmessage-length 204 -pg -mstackrealign -fobjc-runtime=gcc -fobjc-default-synthesize-properties -fdiagnostics-show-option -fcolor-diagnostics -backend-option -vectorize-loops -o /tmp/myprog-oJBSKs.o -x c myprog.c
clang -cc1 version 3.3 based upon LLVM 3.3 default target x86_64-pc-linux-gnu
ignoring nonexistent directory "/include"
#include "..." search starts here:
#include <...> search starts here:
 /usr/local/include
 /usr/bin/../lib/clang/3.3/include
 /usr/include/x86_64-linux-gnu
 /usr/include
End of search list.
 "/usr/bin/ld" -z relro --hash-style=gnu --build-id --eh-frame-hdr -m elf_x86_64 -dynamic-linker /lib64/ld-linux-x86-64.so.2 -o a.out /usr/lib/gcc/x86_64-linux-gnu/4.8/../../../x86_64-linux-gnu/gcrt1.o /usr/lib/gcc/x86_64-linux-gnu/4.8/../../../x86_64-linux-gnu/crti.o /usr/lib/gcc/x86_64-linux-gnu/4.8/crtbegin.o -L/usr/lib/gcc/x86_64-linux-gnu/4.8 -L/usr/lib/gcc/x86_64-linux-gnu/4.8/../../../x86_64-linux-gnu -L/lib/x86_64-linux-gnu -L/lib/../lib64 -L/usr/lib/x86_64-linux-gnu -L/usr/lib/gcc/x86_64-linux-gnu/4.8/../../.. -L/lib -L/usr/lib /tmp/myprog-oJBSKs.o -lgcc --as-needed -lgcc_s --no-as-needed -lc -lgcc --as-needed -lgcc_s --no-as-needed /usr/lib/gcc/x86_64-linux-gnu/4.8/crtend.o /usr/lib/gcc/x86_64-linux-gnu/4.8/../../../x86_64-linux-gnu/crtn.o

Running this program it would give me stack trace:
Program received signal SIGSEGV, Segmentation fault.
mcount () at ../sysdeps/x86_64/_mcount.S:46
46	../sysdeps/x86_64/_mcount.S: No such file or directory.
(gdb) bt
#0  mcount () at ../sysdeps/x86_64/_mcount.S:46
#1  0x00007ffff7dd6568 in ?? () from /lib/x86_64-linux-gnu/libc.so.6
#2  0x0000000000000000 in ?? ()

If I compare the assembly code with and without optimization, the only difference seems to be that the optimized version removed some register preservation ops before calling mcount().

-pg version, which works:
----------------------------------------
main:                                   # @main
	.cfi_startproc
# BB#0:
	pushq	%rbp
.Ltmp2:
	.cfi_def_cfa_offset 16
.Ltmp3:
	.cfi_offset %rbp, -16
	movq	%rsp, %rbp
.Ltmp4:
	.cfi_def_cfa_register %rbp
	subq	$16, %rsp
	callq	mcount
	movl	$0, %eax
	movl	$0, -4(%rbp)
	addq	$16, %rsp
	popq	%rbp
	ret

-pg -O2 version, which crashes:
--------------------------------------------
main:                                   # @main
	.cfi_startproc
# BB#0:
	pushq	%rax
.Ltmp1:
	.cfi_def_cfa_offset 16
	callq	mcount
	xorl	%eax, %eax
	popq	%rdx
	ret

For those who are familiar with assembly code, does it ring some bell?

--Qiao