Re: target/2878: ARM: cannot use 3.0 to build stage 1 of bootstrap; bad code generated at -O0?

Re: target/2878: ARM: cannot use 3.0 to build stage 1 of bootstrap; bad code generated at -O0?

[nathan]

Synopsis: ARM: cannot use 3.0 to build stage 1 of bootstrap; bad code generated at -O0?

State-Changed-From-To: open->analyzed
State-Changed-By: nathan
State-Changed-When: Fri Jun  8 06:09:04 2001
State-Changed-Why:
    confirmed. This is an arm showstopper

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view&pr=2878&database=gcc

Re: target/2878: ARM: cannot use 3.0 to build stage 1 of bootstrap; bad code generated at -O0?

[mmitchel]

Synopsis: ARM: cannot use 3.0 to build stage 1 of bootstrap; bad code generated at -O0?

State-Changed-From-To: feedback->closed
State-Changed-By: mmitchel
State-Changed-When: Wed Jun 13 10:27:22 2001
State-Changed-Why:
    Fixed.

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view&pr=2878&database=gcc

Re: target/2878: ARM: cannot use 3.0 to build stage 1 of bootstrap; bad code generated at -O0?

[rearnsha]

Synopsis: ARM: cannot use 3.0 to build stage 1 of bootstrap; bad code generated at -O0?

Responsible-Changed-From-To: unassigned->rearnsha
Responsible-Changed-By: rearnsha
Responsible-Changed-When: Tue Jun 12 03:34:50 2001
Responsible-Changed-Why:
    I have a patch for this
State-Changed-From-To: analyzed->feedback
State-Changed-By: rearnsha
State-Changed-When: Tue Jun 12 03:34:50 2001
State-Changed-Why:
    Please try this patch:
    http://gcc.gnu.org/ml/gcc-bugs/2001-06/msg00456.html

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view&pr=2878&database=gcc

Re: target/2878: ARM: cannot use 3.0 to build stage 1 of bootstrap; bad code generated at -O0?

[Philip Blundell]

The following reply was made to PR target/2878; it has been noted by GNATS.

From: Philip Blundell <philb@gnu.org>
To: Rod Stewart <stewart@lab43.org>
Cc: gcc-gnats@gcc.gnu.org, gcc-bugs@gcc.gnu.org
Subject: Re: target/2878: ARM: cannot use 3.0 to build stage 1 of bootstrap; bad code generated at -O0? 
Date: Sun, 27 May 2001 00:33:27 +0100

 >Did you have glibc 2.2 installed?
 
 I just confirmed that a single bootstrap with BOOT_CFLAGS="-O0" fails in the 
 same way during stage 2.  This was on a glibc 2.2.2 system.
 
 p.
 
 

Re: target/2878: ARM: cannot use 3.0 to build stage 1 of bootstrap; bad code generated at -O0?

[Philip Blundell]

The following reply was made to PR target/2878; it has been noted by GNATS.

From: Philip Blundell <philb@gnu.org>
To: Rod Stewart <stewart@lab43.org>
Cc: gcc-gnats@gcc.gnu.org, gcc-bugs@gcc.gnu.org
Subject: Re: target/2878: ARM: cannot use 3.0 to build stage 1 of bootstrap; bad code generated at -O0? 
Date: Wed, 23 May 2001 23:31:40 +0100

 >Did you have glibc 2.2 installed?
 
 Yes; 2.2.2 to be exact.
 
 p.
 
 

Re: target/2878: ARM: cannot use 3.0 to build stage 1 of bootstrap;bad code generated at -O0?

[Rod Stewart]

The following reply was made to PR target/2878; it has been noted by GNATS.

From: Rod Stewart <stewart@lab43.org>
To: Philip Blundell <philb@gnu.org>
Cc: <gcc-gnats@gcc.gnu.org>, <gcc-bugs@gcc.gnu.org>
Subject: Re: target/2878: ARM: cannot use 3.0 to build stage 1 of bootstrap;
 bad code generated at -O0? 
Date: Wed, 23 May 2001 17:07:49 -0400 (EDT)

 On Wed, 23 May 2001, Philip Blundell wrote:
 
 > I configured with ".../configure --disable-checking" on an armv4
 > system.  I didn't use --with-cpu or anything.  I'll try it on a second
 > system later, just in case it was due to some local lossage here.
 
 I configured with my standard build options of:
 [build@deimos gcc]$ ./xgcc -v
 Reading specs from
 /usr/local/lib/gcc-lib/armv4l-unknown-linux-gnu/3.0/specs
 Configured with: /home/build-rpm/BUILD/gcc/configure --with-cpu=strongarm
 --enable-languages=c,c++,objc,f77 --prefix=/usr/local --enable-shared
 --enable-threads --without-cvs armv4l-unknown-linux-gnu
 Thread model: posix
 gcc version 3.0 20010522 (prerelease)
 [build@deimos gcc]$
 
 Did you have glibc 2.2 installed?
 
 -Rms
 

Re: target/2878: ARM: cannot use 3.0 to build stage 1 of bootstrap; bad code generated at -O0?

[Philip Blundell]

The following reply was made to PR target/2878; it has been noted by GNATS.

From: Philip Blundell <philb@gnu.org>
To: Rod Stewart <stewart@lab43.org>
Cc: gcc-gnats@gcc.gnu.org, gcc-prs@gcc.gnu.org, gcc-bugs@gcc.gnu.org
Subject: Re: target/2878: ARM: cannot use 3.0 to build stage 1 of bootstrap; bad code generated at -O0? 
Date: Wed, 23 May 2001 19:28:48 +0100

 >Are you certain this is still a problem?
 
 It certainly still happened at the time I reported it; I guess it's possible 
 it's been fixed in the last couple of days, though it doesn't seem that likely.
 
 I configured with ".../configure --disable-checking" on an armv4 system.  I 
 didn't use --with-cpu or anything.  I'll try it on a second system later, just 
 in case it was due to some local lossage here.
 
 p.
 
 

Re: target/2878: ARM: cannot use 3.0 to build stage 1 of bootstrap;bad code generated at -O0?

[Rod Stewart]

The following reply was made to PR target/2878; it has been noted by GNATS.

From: Rod Stewart <stewart@lab43.org>
To: <philb@gnu.org>
Cc: <gcc-gnats@gcc.gnu.org>, <gcc-prs@gcc.gnu.org>, <gcc-bugs@gcc.gnu.org>
Subject: Re: target/2878: ARM: cannot use 3.0 to build stage 1 of bootstrap;
 bad code generated at -O0?
Date: Wed, 23 May 2001 13:56:44 -0400 (EDT)

 On 20 May 2001 philb@gnu.org wrote:
 
 > >Synopsis:  ARM: cannot use 3.0 to build stage 1 of bootstrap; bad
 > code generated at -O0?
 
 > >Description:
 > Bootstrapping the 3.0 branch on arm-linux, using a compiler previously
 > built from the same sources to build stage 1, eventually fails here:
 
 Phil,
 
 Are you certain this is still a problem?  I'm currently using gcc version
 3.0 20010522 to build itself.  It is now using the stage2 compiler.  This
 is with the following tools:
 	gcc-ss-3.0-0_20010523
 	binutils-2.10.1-1_nw2
 	glibc-2.1.3-22_nw3
 
 I did install the new gcc in /usr/local/bin, with that directory in my
 path first.  I guess it could have gotten confused and used my 2.95
 compiler in /usr/bin.  (I'll do another build once this one finishes with
 2.95 removed from the machine).
 
 Any ideas on how I can reproduce this?
 Thanks,
 -Rms
 

target/2878: ARM: cannot use 3.0 to build stage 1 of bootstrap; bad code generated at -O0?

[philb]

>Number:         2878
>Category:       target
>Synopsis:       ARM: cannot use 3.0 to build stage 1 of bootstrap; bad code generated at -O0?
>Confidential:   no
>Severity:       serious
>Priority:       high
>Responsible:    unassigned
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Sun May 20 03:36:01 PDT 2001
>Closed-Date:
>Last-Modified:
>Originator:     philb@gnu.org
>Release:        unknown-1.0
>Organization:
>Environment:

>Description:
Bootstrapping the 3.0 branch on arm-linux, using a compiler previously built from the same sources to build stage 1, eventually fails here:

./xgcc -B./ -B/usr/local/armv4l-unknown-linux-gnu/bin/ -isystem /usr/local/armv4l-unknown-linux-gnu/include -O2   -DIN_GCC    -W -Wall -Wwrite-strings -Wstrict-prototypes -Wmissing-prototypes -isystem ./include  -fomit-frame-pointer -fPIC -g0 -DHAVE_GTHR_DEFAULT -DIN_LIBGCC2 -D__GCC_FLOAT_NOT_NEEDED  -I. -I. -I/home/pb/egcs/3.0/gcc/gcc -I/home/pb/egcs/3.0/gcc/gcc/. -I/home/pb/egcs/3.0/gcc/gcc/config -I/home/pb/egcs/3.0/gcc/gcc/../include  -DL_bb -c /home/pb/egcs/3.0/gcc/gcc/libgcc2.c -o libgcc/./_bb.o
In file included from /home/pb/egcs/3.0/gcc/gcc/libgcc2.c:1298:
/home/pb/egcs/3.0/gcc/gcc/gcov-io.h: In function `__store_long':
/home/pb/egcs/3.0/gcc/gcc/gcov-io.h:67: Internal compiler error in canon_hash, at cse.c:2463
Please submit a full bug report, with preprocessed source if appropriate.
See <URL: http://www.gnu.org/software/gcc/bugs.html > for instructions.
make[3]: *** [libgcc/./_bb.o] Error 1

Since it works to use GCC 2.95 for building stage 1, it seems likely that GCC 3.0 is miscompiling itself at -O0.
>How-To-Repeat:

>Fix:

>Release-Note:
>Audit-Trail:
>Unformatted: