public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
* [Bug middle-end/15043] New: MIPS 64-bit ABIs handle SFmode args inefficiently - endianness dependent
@ 2004-04-21 13:54 nigel at mips dot com
  2004-04-21 14:02 ` [Bug target/15043] " pinskia at gcc dot gnu dot org
                   ` (2 more replies)
  0 siblings, 3 replies; 4+ messages in thread
From: nigel at mips dot com @ 2004-04-21 13:54 UTC (permalink / raw)
  To: gcc-bugs

Compiler version: gcc version 3.4.0 20040416 (prerelease)

Take the following trivial code fragment:

  float pass_float (float a)
  { return a; }

Compile with flags -S -O2 -mips64 -mabi=n32 -EL and this generates the optimal code:

	j	$31
	mov.s	$f0,$f12

However change to -EB and you get the following:

	addiu	$sp,$sp,-16
	swc1	$f12,0($sp)
	lwc1	$f0,0($sp)
	j	$31
	addiu	$sp,$sp,16

This appears to be related to the way assign_parms() handles cases where
arguments are smaller than the word size, and when BLOCK_REG_PADDING is also
defined (see line 4474 in function.c). Is it really necessary to create a stack
slot for these cases? 

Note that if you select a different 64-bit ABI such as EABI then the problem
reverses, and big-endian generates the good code, little-endian bad, because
FUNCTION_ARG_PADDING returns a different value for these ABIs.

-- 
           Summary: MIPS 64-bit ABIs handle SFmode args inefficiently -
                    endianness dependent
           Product: gcc
           Version: 3.4.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: middle-end
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: nigel at mips dot com
                CC: gcc-bugs at gcc dot gnu dot org
 GCC build triplet: i686-pc-linux-gnu
  GCC host triplet: i686-pc-linux-gnu
GCC target triplet: mipsisa32-unknown-elf


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15043


^ permalink raw reply	[flat|nested] 4+ messages in thread

* [Bug target/15043] MIPS 64-bit ABIs handle SFmode args inefficiently - endianness dependent
  2004-04-21 13:54 [Bug middle-end/15043] New: MIPS 64-bit ABIs handle SFmode args inefficiently - endianness dependent nigel at mips dot com
@ 2004-04-21 14:02 ` pinskia at gcc dot gnu dot org
  2004-04-21 15:58 ` rsandifo at redhat dot com
  2004-07-29  4:26 ` pinskia at gcc dot gnu dot org
  2 siblings, 0 replies; 4+ messages in thread
From: pinskia at gcc dot gnu dot org @ 2004-04-21 14:02 UTC (permalink / raw)
  To: gcc-bugs



-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|middle-end                  |target
           Keywords|                            |pessimizes-code


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15043


^ permalink raw reply	[flat|nested] 4+ messages in thread

* [Bug target/15043] MIPS 64-bit ABIs handle SFmode args inefficiently - endianness dependent
  2004-04-21 13:54 [Bug middle-end/15043] New: MIPS 64-bit ABIs handle SFmode args inefficiently - endianness dependent nigel at mips dot com
  2004-04-21 14:02 ` [Bug target/15043] " pinskia at gcc dot gnu dot org
@ 2004-04-21 15:58 ` rsandifo at redhat dot com
  2004-07-29  4:26 ` pinskia at gcc dot gnu dot org
  2 siblings, 0 replies; 4+ messages in thread
From: rsandifo at redhat dot com @ 2004-04-21 15:58 UTC (permalink / raw)
  To: gcc-bugs


------- Additional Comments From rsandifo at redhat dot com  2004-04-21 15:36 -------
Subject: Re:  MIPS 64-bit ABIs handle SFmode args
 inefficiently - endianness dependent

"pinskia at gcc dot gnu dot org" <gcc-bugzilla@gcc.gnu.org> writes:
>           Component|middle-end                  |target

Are you sure?  FWIW, I think Nigel's original classification is correct.

If you don't like "middle-end", maybe optimization would be better?

Richard


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15043


^ permalink raw reply	[flat|nested] 4+ messages in thread

* [Bug target/15043] MIPS 64-bit ABIs handle SFmode args inefficiently - endianness dependent
  2004-04-21 13:54 [Bug middle-end/15043] New: MIPS 64-bit ABIs handle SFmode args inefficiently - endianness dependent nigel at mips dot com
  2004-04-21 14:02 ` [Bug target/15043] " pinskia at gcc dot gnu dot org
  2004-04-21 15:58 ` rsandifo at redhat dot com
@ 2004-07-29  4:26 ` pinskia at gcc dot gnu dot org
  2 siblings, 0 replies; 4+ messages in thread
From: pinskia at gcc dot gnu dot org @ 2004-07-29  4:26 UTC (permalink / raw)
  To: gcc-bugs


------- Additional Comments From pinskia at gcc dot gnu dot org  2004-07-29 04:26 -------


*** This bug has been marked as a duplicate of 16815 ***

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|                            |DUPLICATE


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15043


^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2004-07-29  4:26 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2004-04-21 13:54 [Bug middle-end/15043] New: MIPS 64-bit ABIs handle SFmode args inefficiently - endianness dependent nigel at mips dot com
2004-04-21 14:02 ` [Bug target/15043] " pinskia at gcc dot gnu dot org
2004-04-21 15:58 ` rsandifo at redhat dot com
2004-07-29  4:26 ` pinskia at gcc dot gnu dot org

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).