[Bug optimization/12928] New: [tree-ssa] Bad assembly code produced when compiling glibc

[Bug optimization/12928] New: [tree-ssa] Bad assembly code produced when compiling glibc

[bbooth at redhat dot 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=12928

           Summary: [tree-ssa] Bad assembly code produced when compiling
                    glibc
           Product: gcc
           Version: tree-ssa
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: optimization
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: bbooth at redhat dot com
                CC: gcc-bugs at gcc dot gnu dot org
  GCC host triplet: i686-pc-linux-gnu

GCC version:
    gcc (GCC) 3.5-tree-ssa 20031106 (merged 20031026)

GCC configure options:
    --prefix=/notnfs/bbooth/tree-ssa/i686/inst
    --srcdir=/notnfs/bbooth/tree-ssa/gcc
    --disable-multilib

Command line & compiler/assembler output:

/notnfs/bbooth/tree-ssa/i686/inst/libexec/gcc/i686-pc-linux-gnu/3.5-tree-ssa/cc1
-fpreprocessed sigsuspend.i -quiet -dumpbase sigsuspend.c -mtune=pentiumpro
-auxbase-strip /home/bbooth/tree-ssa/build/i686/GLIBC/signal/sigsuspend.o -O2
-Wall -Winline -Wstrict-prototypes -Wwrite-strings -version -ftime-report -o
sigsuspend.s
GNU C version 3.5-tree-ssa 20031106 (merged 20031026) (i686-pc-linux-gnu)
	compiled by GNU C version 3.5-tree-ssa 20031106 (merged 20031026).
GGC heuristics: --param ggc-min-expand=30 --param ggc-min-heapsize=4096

Execution times (seconds)
 lexical analysis      :   0.01 (14%) usr   0.00 ( 0%) sys   0.01 (11%) wall
 parser                :   0.02 (29%) usr   0.00 ( 0%) sys   0.02 (22%) wall
 tree gimplify         :   0.01 (14%) usr   0.00 ( 0%) sys   0.01 (11%) wall
 bypass jumps          :   0.01 (14%) usr   0.00 ( 0%) sys   0.01 (11%) wall
 symout                :   0.01 (14%) usr   0.00 ( 0%) sys   0.01 (11%) wall
 TOTAL                 :   0.07             0.02             0.09
 as -V -Qy -o /home/bbooth/tree-ssa/build/i686/GLIBC/signal/sigsuspend.o
sigsuspend.s
GNU assembler version 2.14.90.0.4 (i386-redhat-linux) using BFD version
2.14.90.0.4 20030523
sigsuspend.s: Assembler messages:
sigsuspend.s:83: Error: non-constant expression in ".if" statement
sigsuspend.s:84: Error: non-constant expression in ".if" statement
sigsuspend.s:87: Error: non-constant expression in ".if" statement

==============

.i file to come.

[Bug optimization/12928] [tree-ssa] Bad assembly code produced when compiling glibc

[bbooth at redhat dot 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=12928



------- Additional Comments From bbooth at redhat dot com  2003-11-06 20:18 -------
Created an attachment (id=5074)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=5074&action=view)
Associated .i file

[Bug optimization/12928] [tree-ssa] Bad assembly code produced when compiling glibc

[bbooth at redhat dot 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=12928



------- Additional Comments From bbooth at redhat dot com  2003-11-06 20:18 -------
Created an attachment (id=5075)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=5075&action=view)
The assembly code causing the errors.

[Bug optimization/12928] [tree-ssa] Bad assembly code produced when compiling glibc

[pinskia at gcc dot gnu dot 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=12928


pinskia at gcc dot gnu dot org changed:

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


------- Additional Comments From pinskia at gcc dot gnu dot org  2003-11-06 20:28 -------
This is not a gcc problem from the source:
     ".if 1 - \\name\n\t"
Report it to glibc.

[Bug optimization/12928] [tree-ssa] Bad assembly code produced when compiling glibc

[dnovillo at redhat dot 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=12928



------- Additional Comments From dnovillo at redhat dot com  2003-11-06 20:38 -------
Subject: Re:  [tree-ssa] Bad assembly code produced
	when compiling glibc

On Thu, 2003-11-06 at 15:28, pinskia at gcc dot gnu dot org wrote:

> This is not a gcc problem from the source:
>      ".if 1 - \\name\n\t"
> Report it to glibc.
>
Hmm, we don't get this problem with gcc 3.2.  Is this something we've
deprecated since then?


Diego.

[Bug optimization/12928] [tree-ssa] Bad assembly code produced when compiling glibc

[pinskia at gcc dot gnu dot 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=12928


pinskia at gcc dot gnu dot org changed:

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


------- Additional Comments From pinskia at gcc dot gnu dot org  2003-11-06 20:48 -------
I take that back, I needed to look into the asm a little more:
	bpushl .L__X'$0, $0
	bmovl .L__X'$0, $0
	movl $72, %eax
	int $0x80
	bpopl .L__X'$0, $0

%k2 == $0?

[Bug optimization/12928] [tree-ssa] Bad assembly code produced when compiling glibc

[pinskia at gcc dot gnu dot 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=12928


pinskia at gcc dot gnu dot org changed:

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


------- Additional Comments From pinskia at gcc dot gnu dot org  2003-11-06 21:05 -------
Still valid as "C" constraint, I think glibc wanted "c" here instead (and maybe "d" instead of "D"):
Specifies constant that can be easily constructed in SSE register without loading it from memory. 
>From the source:
"aCD" (0)

[Bug optimization/12928] [tree-ssa] Bad assembly code produced when compiling glibc

[bbooth at redhat dot 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=12928


bbooth at redhat dot com changed:

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


------- Additional Comments From bbooth at redhat dot com  2003-11-07 19:45 -------
What about this case...

GCC version:
    gcc (GCC) 3.5-tree-ssa 20031107 (merged 20031026)

GCC configure options:
    --prefix=/notnfs/bbooth/tree-ssa/x86_64/inst
    --srcdir=/notnfs/bbooth/tree-ssa/gcc
    --disable-multilib

Command line & compiler/assembler output:

/notnfs/bbooth/tree-ssa/x86_64/inst/libexec/gcc/x86_64-unknown-linux-gnu/3.5-tree-ssa/cc1
-fpreprocessed l10nflist.i -quiet -dumpbase l10nflist.c -mtune=k8 -auxbase-strip
/home/bbooth/tree-ssa/build/x86_64/GLIBC/intl/l10nflist.o -O2 -Wall -Winline
-Wstrict-prototypes -Wwrite-strings -std=gnu99 -version -ftime-report -o l10nflist.s
GNU C version 3.5-tree-ssa 20031107 (merged 20031026) (x86_64-unknown-linux-gnu)
	compiled by GNU C version 3.5-tree-ssa 20031107 (merged 20031026).
GGC heuristics: --param ggc-min-expand=30 --param ggc-min-heapsize=4096

Execution times (seconds)
 garbage collection    :   0.02 (12%) usr   0.00 ( 0%) sys   0.02 (11%) wall
 callgraph construction:   0.00 ( 0%) usr   0.00 ( 0%) sys   0.01 ( 6%) wall
 life analysis         :   0.00 ( 0%) usr   0.00 ( 0%) sys   0.01 ( 6%) wall
 life info update      :   0.01 ( 6%) usr   0.00 ( 0%) sys   0.00 ( 0%) wall
 alias analysis        :   0.00 ( 0%) usr   0.00 ( 0%) sys   0.02 (11%) wall
 register scan         :   0.01 ( 6%) usr   0.00 ( 0%) sys   0.00 ( 0%) wall
 preprocessing         :   0.01 ( 6%) usr   0.00 ( 0%) sys   0.02 (11%) wall
 lexical analysis      :   0.03 (18%) usr   0.00 ( 0%) sys   0.01 ( 6%) wall
 parser                :   0.02 (12%) usr   0.00 ( 0%) sys   0.03 (17%) wall
 tree CFG construction :   0.01 ( 6%) usr   0.00 ( 0%) sys   0.00 ( 0%) wall
 tree CFG cleanup      :   0.00 ( 0%) usr   0.00 ( 0%) sys   0.01 ( 6%) wall
 tree thread jumps     :   0.01 ( 6%) usr   0.00 ( 0%) sys   0.00 ( 0%) wall
 expand                :   0.00 ( 0%) usr   0.00 ( 0%) sys   0.01 ( 6%) wall
 CSE                   :   0.01 ( 6%) usr   0.00 ( 0%) sys   0.01 ( 6%) wall
 loop analysis         :   0.01 ( 6%) usr   0.00 ( 0%) sys   0.00 ( 0%) wall
 flow analysis         :   0.00 ( 0%) usr   0.00 ( 0%) sys   0.01 ( 6%) wall
 combiner              :   0.01 ( 6%) usr   0.00 ( 0%) sys   0.01 ( 6%) wall
 if-conversion         :   0.01 ( 6%) usr   0.00 ( 0%) sys   0.00 ( 0%) wall
 scheduling 2          :   0.01 ( 6%) usr   0.00 ( 0%) sys   0.00 ( 0%) wall
 TOTAL                 :   0.17             0.01             0.18
 as -V -Qy -o /home/bbooth/tree-ssa/build/x86_64/GLIBC/intl/l10nflist.o l10nflist.s
GNU assembler version 2.14.90.0.4 (x86_64-redhat-linux) using BFD version
2.14.90.0.4 20030523
l10nflist.s: Assembler messages:
l10nflist.s:406: Error: Incorrect register `%rax' used with `l' suffix

[Bug optimization/12928] [tree-ssa] Bad assembly code produced when compiling glibc

[bbooth at redhat dot 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=12928


bbooth at redhat dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   GCC host triplet|i686-pc-linux-gnu           |x86_64-unknown-linux-gnu

[Bug optimization/12928] [tree-ssa] Bad assembly code produced when compiling glibc

[pinskia at gcc dot gnu dot 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=12928


pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|normal                      |critical
   GCC host triplet|x86_64-unknown-linux-gnu    |
 GCC target triplet|                            |x86_64-unknown-linux-gnu
           Keywords|                            |wrong-code
   Target Milestone|---                         |tree-ssa


------- Additional Comments From pinskia at gcc dot gnu dot org  2003-11-07 19:53 -------
Either a machine description problem or a miscompiling something generated from the *.md files.

[Bug optimization/12928] [tree-ssa] Bad assembly code produced when compiling glibc

[rth at gcc dot gnu dot org]

------- Additional Comments From rth at gcc dot gnu dot org  2004-01-16 20:23 -------
Fixed on mainline by
  http://gcc.gnu.org/ml/gcc-patches/2004-01/msg01612.html

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12928