llvm dev - Feb 2012 - [LLVMdev] difference in function prologue generated with clang and gcc

Hello Anton,
Thanks for the reply. I have not specified optimization level explicitly
during compilation. For GCC default is O0 ie., no optimization. Do you mean
that clang uses other optimization level other than O0 ?
Could you please clarify ?

On Thu, Feb 16, 2012 at 6:52 PM, Anton Korobeynikov <anton at
korobeynikov.info
> wrote:
> Hello
>
> > The prologue length in .debug_line is 157 for clang generated one,
> whereas
> > it is 34 for gcc generated one. I am curious about the results of
making
> > prologue generated by clang look similar with one generated by gcc.
> > Could anyone let me know why this difference exists and if it is for
good
> > /better purposes than for gcc. ?
> 1. This is not function prologue length. It's the header (aka
> 'prologue') length of .debug_line section
> 2. The length of function prologue is 3 instructions in case of clang
> (12 bytes) and 2 instruction (8 bytes) in case of gcc
> 3. Comparison of code size of unoptimized code does not make any sense.
>
> Hope this makes the stuff clear.
>
> --
> With best regards, Anton Korobeynikov
> Faculty of Mathematics and Mechanics, Saint Petersburg State University
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20120217/ac6e3b63/attachment.html>

> Thanks for the reply. I have not specified optimization level explicitly
> during compilation. For GCC default is O0 ie., no optimization. Do you mean
> that clang uses other optimization level other than O0 ?
No, here it's still -O0. However it makes no sense to compare the size
of unoptimized code.

-- 
With best regards, Anton Korobeynikov
Faculty of Mathematics and Mechanics, Saint Petersburg State University

If a reason is needed: It makes no sense, because it's not something the
maintainers care about. Small size is not -O0 is about.

You might as well compare how compressible the clang output is compared
to g++. "zip compresses clang output better than gcc, but 7z compresses
gcc output better than clang."

Arnt

Hello

I am trying to run following dejaGNU testcases on Clang compiled output.

/* Inlined inline function must have abstract DIE  */
/* { dg-do compile } */
/* { dg-options "-O2 -gdwarf-2 -dA -fpreprocessed" } */
/* { dg-final { scan-assembler "3.*DW_AT_inline" } } */
#1 "test.h"
inline int t()
{
}
int q()
{
  t();
}

The testcase fails because, DW_AT_inline is not present in assembly output.
Could anyone let me know why DW_AT_inline is not emitted ?
I am aware that Clang uses C99 mode by default.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20120224/209abd96/attachment.html>