[Bug optimization/10409] [ARM] "__attribute__((const))" apparently not working

[Bug optimization/10409] [ARM] "__attribute__((const))" apparently not working

[dhazeghi@yahoo.com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10409


dhazeghi@yahoo.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |WAITING
  GCC build triplet|                            |i686-pc-cygwin
   GCC host triplet|                            |i686-pc-cygwin
 GCC target triplet|                            |arm-elf


------- Additional Comments From dhazeghi@yahoo.com  2003-06-04 00:22 -------
Hello,

with gcc 3.2.3 and 3.3 branch, I see the seem behavior that you do. With gcc mainline, I only get 
one call to pure_func. Would it be possible for you to check whether this works for you too with 
gcc mainline (cvs), or to confirm whether the following assembly is correct:
        .file   "test.c"
        .text
        .align  2
        .global test
        .type   test, %function
test:
        @ args = 0, pretend = 0, frame = 0
        @ frame_needed = 1, uses_anonymous_args = 0
        mov     ip, sp
        stmfd   sp!, {fp, ip, lr, pc}
        sub     fp, ip, #4
        bl      pure_func
        mov     r0, r0, asl #1
        ldmea   fp, {fp, sp, pc}
        .size   test, .-test
        .ident  "GCC: (GNU) 3.4 20030603 (experimental)"

Thanks,

Dara



------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug optimization/10409] [ARM] "__attribute__((const))" apparently not working

[gonz@ratloop.com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10409


gonz@ratloop.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   GCC host triplet|i686-pc-cygwin              |i686-pc-ddd


------- Additional Comments From gonz@ratloop.com  2003-06-04 06:04 -------
Thanks Dara,

The assembly code you pasted looks correct, however, it is pushing
a lot of unneccessary registers.  Was this compiled with "-O2"?

I would definitely like to test the latest 3.4 sources, but it's
going to take me a few weeks to allocate the time.  If 3.4 fixes
the problem, then we could close this bug.

-Pete




------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug optimization/10409] [ARM] "__attribute__((const))" apparently not working

[dhazeghi@yahoo.com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10409


dhazeghi@yahoo.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|WAITING                     |NEW
     Ever Confirmed|                            |1
   Last reconfirmed|0000-00-00 00:00:00         |2003-06-04 06:42:46
               date|                            |


------- Additional Comments From dhazeghi@yahoo.com  2003-06-04 06:42 -------
Yes, this was compiled with -O2. -O3 doesn't change it either. Actually, when I try it with -O0, the 
problem returns. Furthermore, if I try the testcase on powerpc-darwin, the problem's there at all 
optimization levels. Looks like this one's still a bug (though not only in your specific case)...

Dara



------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug optimization/10409] [ARM] "__attribute__((const))" apparently not working

[gonz@ratloop.com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10409



------- Additional Comments From gonz@ratloop.com  2003-06-04 07:43 -------
I would expect the problem to occur with -O0, since folding pure functions is a 
common subexpression optimization, not a semantics issue.  

Regarding my previous comments about the assembly langauge, I should note that 
under GCC 3.2.3, this code:

int test(int x) {
  return pure_func(x) << 1;
}

...produces this output:

	.size	_Z8testorigi,.Lfe1-_Z8testorigi
	.align	2
	.global	_Z4testi
	.type	_Z4testi,function
_Z4testi:
	@ Function supports interworking.
	@ args = 0, pretend = 0, frame = 0
	@ frame_needed = 1, uses_anonymous_args = 0
	mov	ip, sp
	stmfd	sp!, {fp, ip, lr, pc}
	sub	fp, ip, #4
	bl	_Z9pure_funci
	mov	r0, r0, asl #1
	ldmea	fp, {fp, sp, lr}
	bx	lr

...which is nearly identical to the code you pasted, and therefore GCC 3.4 is 
doing the right thing.  If "-fomit-frame-pointer" is failing to eliminate those 
extra pushes, it is a totally unrelated bug.

Therefore, it seems your most significant finding is that powerpc-darwin is 
still has the problem.

-Pete




------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug optimization/10409] "__attribute__((const))" apparently not working

[rearnsha@gcc.gnu.org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10409


rearnsha@gcc.gnu.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|normal                      |enhancement
           Keywords|                            |pessimizes-code
            Summary|[ARM]                       |"__attribute__((const))"
                   |"__attribute__((const))"    |apparently not working
                   |apparently not working      |


------- Additional Comments From rearnsha@gcc.gnu.org  2003-06-05 13:59 -------
It is not incorrect to fail to remove duplicate calls to "const" functions.  So
at best this is an enhancement request.  As noted elsewhere, this is a
shortcoming in our CSE.



------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug optimization/10409] "__attribute__((const))" apparently not working

[pinskia@physics.uc.edu]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10409


pinskia@physics.uc.edu changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED


------- Additional Comments From pinskia@physics.uc.edu  2003-06-08 16:23 -------
This is fixed on the mainline (20030608) for powerpc-darwin and powerpc-aix (which 
was filled as a different bug) and arm-elf and ix86-pc-linux-gnu.