gcc-2.95 build FAILED! - RedHat Linux/Alpha 5.2

gcc-2.95 build FAILED! - RedHat Linux/Alpha 5.2

[Alexander L. Belikoff]

Yep - justs failed to build on:

- Alpha dp264 (dual CPU)
- RedHat 5.2
- egcs 1.1.2, binutils 2.9.1.0.15-1

This is pretty painful and surprising, considering that egcs-19990712
built just fine. Anybody cares to fix that or just "stick to RedHat
6.0, you pathetic looser!" ;-)

We are slowly migrating on our Alphas from RedHat 4.2 to RedHat 5.2.
6.0 is out of question - there are too many complaints about general
stability and speed. On RedHat 5.2 the situation with the compilers is
the following:

1. Egcs 1.0.3 (supplied by RedHat) - gives ICE on certain optimization
   flags (I reported it quite a while ago).

2. Egcs 1.1.2 (built manually) - has the infamous memcpy bug.

3. Gcc 2.95 - fails to build.

This doesn't look very optimistic, does it? ;-) Anyway, here goes the
make log:


if [ x"no" = xyes ]; then \
  /bb/home/abel/eb/gcc/xgcc -B/bb/home/abel/eb/gcc/ -B/usr/local/gcc-2.95/alphaev6-unknown-linux-gnu/bin/ -c -g -O2 -fvtable-thunks -D_GNU_SOURCE -fno-implicit-templates  -mieee -I../../../../gcc-2.95/libstdc++ -I../../../../gcc-2.95/libstdc++/stl -I../libio -I../../../../gcc-2.95/libstdc++/../libio -nostdinc++ -D_IO_MTSAFE_IO  -DF \
    `for N in MAIN ADDCC ADDCF ADDFC SUBCC SUBCF SUBFC MULCC MULCF MULFC DIVCC DIVCF DIVFC PLUS MINUS EQCC EQCF EQFC NECC NECF NEFC ABS ARG POLAR CONJ NORM COS COSH EXP LOG POWCC POWCF POWCI POWFC SIN SINH SQRT; do echo " -D${N}"; done` \
    ../../../../gcc-2.95/libstdc++/cinst.cc -o pic/fcomplex.o; \
else true ; fi
/bb/home/abel/eb/gcc/xgcc -B/bb/home/abel/eb/gcc/ -B/usr/local/gcc-2.95/alphaev6-unknown-linux-gnu/bin/ -c -g -O2 -fvtable-thunks -D_GNU_SOURCE -fno-implicit-templates  -mieee -I../../../../gcc-2.95/libstdc++ -I../../../../gcc-2.95/libstdc++/stl -I../libio -I../../../../gcc-2.95/libstdc++/../libio -nostdinc++ -D_IO_MTSAFE_IO -DF `for N in MAIN ADDCC ADDCF ADDFC SUBCC SUBCF SUBFC MULCC MULCF MULFC DIVCC DIVCF DIVFC PLUS MINUS EQCC EQCF EQFC NECC NECF NEFC ABS ARG POLAR CONJ NORM COS COSH EXP LOG POWCC POWCF POWCI POWFC SIN SINH SQRT; do echo " -D${N}"; done` \
  ../../../../gcc-2.95/libstdc++/cinst.cc -o fcomplex.o
/tmp/cc2jzCeM.s: Assembler messages:
/tmp/cc2jzCeM.s:3711: Error: unknown opcode `sqrttsu'
/tmp/cc2jzCeM.s:3781: Error: unknown opcode `sqrttsu'
make[4]: *** [bigstmp-complx] Error 1
make[4]: Leaving directory `/bb/home/abel/eb/alphaev6-unknown-linux-gnu/ieee/libstdc++'
make[3]: *** [multi-do] Error 1
make[3]: Leaving directory `/bb/home/abel/eb/alphaev6-unknown-linux-gnu/libstdc++'
make[2]: *** [multi-all] Error 2
make[2]: Leaving directory `/bb/home/abel/eb/alphaev6-unknown-linux-gnu/libstdc++'
make[1]: *** [all-target-libstdc++] Error 2
make[1]: Leaving directory `/bb/home/abel/eb'
make: *** [bootstrap] Error 2


PS. In case you are going to suggest upgrading binutils, could you
please also explain what exactly was fine with the latter package 3
weeks ago, that got horribly wrong now?..

Thanks in advance,


-- 
Alexander L. Belikoff
Bloomberg L.P. / BFM Financial Research Ltd.
abel@vallinor4.com, abel@bfr.co.il

Re: gcc-2.95 build FAILED! - RedHat Linux/Alpha 5.2

[Jeffrey A Law]

  In message < m33dy3b73j.fsf@vermis.bfr.co.il >you write:
  > /tmp/cc2jzCeM.s:3711: Error: unknown opcode `sqrttsu'
  > /tmp/cc2jzCeM.s:3781: Error: unknown opcode `sqrttsu'
Your assembler is out of date.  Get a newer one.

jeff

Re: gcc-2.95 build FAILED! - RedHat Linux/Alpha 5.2

[Alexander L. Belikoff]

Jeffrey A Law <law@cygnus.com> writes:

>   In message < m33dy3b73j.fsf@vermis.bfr.co.il >you write:
>   > /tmp/cc2jzCeM.s:3711: Error: unknown opcode `sqrttsu'
>   > /tmp/cc2jzCeM.s:3781: Error: unknown opcode `sqrttsu'
> Your assembler is out of date.  Get a newer one.
> 
> jeff

But it was good enough three weeks ago, when gcc was in (apparent)
deep freeze! Is there any reason for losing backward compatibility in
such an abrupt way? Binutils is a major package we are definitely not
going to upgrade it unless such an upgrade is critical - we are
talking about production systems... This basically rules out
*previous* (read, "stable") release of a major Linux distribution...
:-(

-- 
Alexander L. Belikoff
Bloomberg L.P. / BFM Financial Research Ltd.
abel@vallinor4.com, abel@bfr.co.il

Re: gcc-2.95 build FAILED! - RedHat Linux/Alpha 5.2

[Jeffrey A Law]

  In message < m3wvvf9ptb.fsf@vermis.bfr.co.il >you write:
  > But it was good enough three weeks ago, when gcc was in (apparent)
  > deep freeze! Is there any reason for losing backward compatibility in
  > such an abrupt way? Binutils is a major package we are definitely not
  > going to upgrade it unless such an upgrade is critical - we are
  > talking about production systems... This basically rules out
  > *previous* (read, "stable") release of a major Linux distribution...
The assembler is broken, plain and simple.  Get a newer one.  There may be
a way to disable those instructions, but I wouldn't know what it is since
I do not normally work on alphas,

jeff

Re: gcc-2.95 build FAILED! - RedHat Linux/Alpha 5.2

[Jeffrey A Law]

  In message < m3wvvf9ptb.fsf@vermis.bfr.co.il >you write:
  > But it was good enough three weeks ago, when gcc was in (apparent)
  > deep freeze! Is there any reason for losing backward compatibility in
  > such an abrupt way? Binutils is a major package we are definitely not
  > going to upgrade it unless such an upgrade is critical - we are
  > talking about production systems... This basically rules out
  > *previous* (read, "stable") release of a major Linux distribution...
The assembler is broken, plain and simple.  Get a newer one.  There may be
a way to disable those instructions, but I wouldn't know what it is since
I do not normally work on alphas,

jeff

Re: gcc-2.95 build FAILED! - RedHat Linux/Alpha 5.2

[Alexandre Oliva]

On Aug  1, 1999, abel@bfr.co.il (Alexander L. Belikoff) wrote:

> Jeffrey A Law <law@cygnus.com> writes:
>> In message < m33dy3b73j.fsf@vermis.bfr.co.il >you write:
>> > /tmp/cc2jzCeM.s:3711: Error: unknown opcode `sqrttsu'
>> > /tmp/cc2jzCeM.s:3781: Error: unknown opcode `sqrttsu'

>> Your assembler is out of date.  Get a newer one.

> But it was good enough three weeks ago, when gcc was in (apparent)
> deep freeze!

Was it on the very same host (i.e., was config.guess' output the
same)?  Maybe you get this opcode for alphaev56, but not for alphaev5
(this is just intended as an example)

I could build gcc 2.95 on a RedHat 6.0/alpha (an AlphaStation 200),
with binutils 2.9.1.0.24 (the one that ships with RH6.0), if this
helps.

-- 
Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il
oliva@{dcc.unicamp.br,guarana.{org,com}} aoliva@{acm.org,computer.org}
oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org}
** I may forward mail about projects to mailing lists; please use them

Re: gcc-2.95 build FAILED! - RedHat Linux/Alpha 5.2

[Richard Henderson]

On Mon, Aug 02, 1999 at 11:07:18PM -0300, Alexandre Oliva wrote:
> Was it on the very same host (i.e., was config.guess' output the
> same)?  Maybe you get this opcode for alphaev56, but not for alphaev5
> (this is just intended as an example)

The sqrt insn appears as of ev6.


r~

Re: gcc-2.95 build FAILED! - RedHat Linux/Alpha 5.2

[Richard Henderson]

On Mon, Aug 02, 1999 at 11:07:18PM -0300, Alexandre Oliva wrote:
> Was it on the very same host (i.e., was config.guess' output the
> same)?  Maybe you get this opcode for alphaev56, but not for alphaev5
> (this is just intended as an example)

The sqrt insn appears as of ev6.


r~

Re: gcc-2.95 build FAILED! - RedHat Linux/Alpha 5.2

[Alexander L. Belikoff]

Alexandre Oliva <oliva@dcc.unicamp.br> writes:

> On Aug  1, 1999, abel@bfr.co.il (Alexander L. Belikoff) wrote:
> 
> > Jeffrey A Law <law@cygnus.com> writes:
> >> In message < m33dy3b73j.fsf@vermis.bfr.co.il >you write:
> >> > /tmp/cc2jzCeM.s:3711: Error: unknown opcode `sqrttsu'
> >> > /tmp/cc2jzCeM.s:3781: Error: unknown opcode `sqrttsu'
> 
> >> Your assembler is out of date.  Get a newer one.
> 
> > But it was good enough three weeks ago, when gcc was in (apparent)
> > deep freeze!
> 
> Was it on the very same host (i.e., was config.guess' output the
> same)?  Maybe you get this opcode for alphaev56, but not for alphaev5
> (this is just intended as an example)

/usr/local/egcs.new/bin/gcc -v
Reading specs from /usr/local/egcs.new/lib/gcc-lib/alphaev6-unknown-linux-gnu/2.95/specs
gcc version 2.95 19990712 (prerelease)

-- 
Alexander L. Belikoff
Bloomberg L.P. / BFM Financial Research Ltd.
abel@vallinor4.com, abel@bfr.co.il

Re: gcc-2.95 build FAILED! - RedHat Linux/Alpha 5.2

[Alexander L. Belikoff]

Alexandre Oliva <oliva@dcc.unicamp.br> writes:

> On Aug  1, 1999, abel@bfr.co.il (Alexander L. Belikoff) wrote:
> 
> > Jeffrey A Law <law@cygnus.com> writes:
> >> In message < m33dy3b73j.fsf@vermis.bfr.co.il >you write:
> >> > /tmp/cc2jzCeM.s:3711: Error: unknown opcode `sqrttsu'
> >> > /tmp/cc2jzCeM.s:3781: Error: unknown opcode `sqrttsu'
> 
> >> Your assembler is out of date.  Get a newer one.
> 
> > But it was good enough three weeks ago, when gcc was in (apparent)
> > deep freeze!
> 
> Was it on the very same host (i.e., was config.guess' output the
> same)?  Maybe you get this opcode for alphaev56, but not for alphaev5
> (this is just intended as an example)

/usr/local/egcs.new/bin/gcc -v
Reading specs from /usr/local/egcs.new/lib/gcc-lib/alphaev6-unknown-linux-gnu/2.95/specs
gcc version 2.95 19990712 (prerelease)

-- 
Alexander L. Belikoff
Bloomberg L.P. / BFM Financial Research Ltd.
abel@vallinor4.com, abel@bfr.co.il

Re: gcc-2.95 build FAILED! - RedHat Linux/Alpha 5.2

[Alexandre Oliva]

On Aug  1, 1999, abel@bfr.co.il (Alexander L. Belikoff) wrote:

> Jeffrey A Law <law@cygnus.com> writes:
>> In message < m33dy3b73j.fsf@vermis.bfr.co.il >you write:
>> > /tmp/cc2jzCeM.s:3711: Error: unknown opcode `sqrttsu'
>> > /tmp/cc2jzCeM.s:3781: Error: unknown opcode `sqrttsu'

>> Your assembler is out of date.  Get a newer one.

> But it was good enough three weeks ago, when gcc was in (apparent)
> deep freeze!

Was it on the very same host (i.e., was config.guess' output the
same)?  Maybe you get this opcode for alphaev56, but not for alphaev5
(this is just intended as an example)

I could build gcc 2.95 on a RedHat 6.0/alpha (an AlphaStation 200),
with binutils 2.9.1.0.24 (the one that ships with RH6.0), if this
helps.

-- 
Alexandre Oliva http://www.dcc.unicamp.br/~oliva IC-Unicamp, Bra[sz]il
oliva@{dcc.unicamp.br,guarana.{org,com}} aoliva@{acm.org,computer.org}
oliva@{gnu.org,kaffe.org,{egcs,sourceware}.cygnus.com,samba.org}
** I may forward mail about projects to mailing lists; please use them

Re: gcc-2.95 build FAILED! - RedHat Linux/Alpha 5.2

[Alexander L. Belikoff]

Jeffrey A Law <law@cygnus.com> writes:

>   In message < m33dy3b73j.fsf@vermis.bfr.co.il >you write:
>   > /tmp/cc2jzCeM.s:3711: Error: unknown opcode `sqrttsu'
>   > /tmp/cc2jzCeM.s:3781: Error: unknown opcode `sqrttsu'
> Your assembler is out of date.  Get a newer one.
> 
> jeff

But it was good enough three weeks ago, when gcc was in (apparent)
deep freeze! Is there any reason for losing backward compatibility in
such an abrupt way? Binutils is a major package we are definitely not
going to upgrade it unless such an upgrade is critical - we are
talking about production systems... This basically rules out
*previous* (read, "stable") release of a major Linux distribution...
:-(

-- 
Alexander L. Belikoff
Bloomberg L.P. / BFM Financial Research Ltd.
abel@vallinor4.com, abel@bfr.co.il

Re: gcc-2.95 build FAILED! - RedHat Linux/Alpha 5.2

[Jeffrey A Law]

  In message < m33dy3b73j.fsf@vermis.bfr.co.il >you write:
  > /tmp/cc2jzCeM.s:3711: Error: unknown opcode `sqrttsu'
  > /tmp/cc2jzCeM.s:3781: Error: unknown opcode `sqrttsu'
Your assembler is out of date.  Get a newer one.

jeff

gcc-2.95 build FAILED! - RedHat Linux/Alpha 5.2

[Alexander L. Belikoff]

Yep - justs failed to build on:

- Alpha dp264 (dual CPU)
- RedHat 5.2
- egcs 1.1.2, binutils 2.9.1.0.15-1

This is pretty painful and surprising, considering that egcs-19990712
built just fine. Anybody cares to fix that or just "stick to RedHat
6.0, you pathetic looser!" ;-)

We are slowly migrating on our Alphas from RedHat 4.2 to RedHat 5.2.
6.0 is out of question - there are too many complaints about general
stability and speed. On RedHat 5.2 the situation with the compilers is
the following:

1. Egcs 1.0.3 (supplied by RedHat) - gives ICE on certain optimization
   flags (I reported it quite a while ago).

2. Egcs 1.1.2 (built manually) - has the infamous memcpy bug.

3. Gcc 2.95 - fails to build.

This doesn't look very optimistic, does it? ;-) Anyway, here goes the
make log:


if [ x"no" = xyes ]; then \
  /bb/home/abel/eb/gcc/xgcc -B/bb/home/abel/eb/gcc/ -B/usr/local/gcc-2.95/alphaev6-unknown-linux-gnu/bin/ -c -g -O2 -fvtable-thunks -D_GNU_SOURCE -fno-implicit-templates  -mieee -I../../../../gcc-2.95/libstdc++ -I../../../../gcc-2.95/libstdc++/stl -I../libio -I../../../../gcc-2.95/libstdc++/../libio -nostdinc++ -D_IO_MTSAFE_IO  -DF \
    `for N in MAIN ADDCC ADDCF ADDFC SUBCC SUBCF SUBFC MULCC MULCF MULFC DIVCC DIVCF DIVFC PLUS MINUS EQCC EQCF EQFC NECC NECF NEFC ABS ARG POLAR CONJ NORM COS COSH EXP LOG POWCC POWCF POWCI POWFC SIN SINH SQRT; do echo " -D${N}"; done` \
    ../../../../gcc-2.95/libstdc++/cinst.cc -o pic/fcomplex.o; \
else true ; fi
/bb/home/abel/eb/gcc/xgcc -B/bb/home/abel/eb/gcc/ -B/usr/local/gcc-2.95/alphaev6-unknown-linux-gnu/bin/ -c -g -O2 -fvtable-thunks -D_GNU_SOURCE -fno-implicit-templates  -mieee -I../../../../gcc-2.95/libstdc++ -I../../../../gcc-2.95/libstdc++/stl -I../libio -I../../../../gcc-2.95/libstdc++/../libio -nostdinc++ -D_IO_MTSAFE_IO -DF `for N in MAIN ADDCC ADDCF ADDFC SUBCC SUBCF SUBFC MULCC MULCF MULFC DIVCC DIVCF DIVFC PLUS MINUS EQCC EQCF EQFC NECC NECF NEFC ABS ARG POLAR CONJ NORM COS COSH EXP LOG POWCC POWCF POWCI POWFC SIN SINH SQRT; do echo " -D${N}"; done` \
  ../../../../gcc-2.95/libstdc++/cinst.cc -o fcomplex.o
/tmp/cc2jzCeM.s: Assembler messages:
/tmp/cc2jzCeM.s:3711: Error: unknown opcode `sqrttsu'
/tmp/cc2jzCeM.s:3781: Error: unknown opcode `sqrttsu'
make[4]: *** [bigstmp-complx] Error 1
make[4]: Leaving directory `/bb/home/abel/eb/alphaev6-unknown-linux-gnu/ieee/libstdc++'
make[3]: *** [multi-do] Error 1
make[3]: Leaving directory `/bb/home/abel/eb/alphaev6-unknown-linux-gnu/libstdc++'
make[2]: *** [multi-all] Error 2
make[2]: Leaving directory `/bb/home/abel/eb/alphaev6-unknown-linux-gnu/libstdc++'
make[1]: *** [all-target-libstdc++] Error 2
make[1]: Leaving directory `/bb/home/abel/eb'
make: *** [bootstrap] Error 2


PS. In case you are going to suggest upgrading binutils, could you
please also explain what exactly was fine with the latter package 3
weeks ago, that got horribly wrong now?..

Thanks in advance,


-- 
Alexander L. Belikoff
Bloomberg L.P. / BFM Financial Research Ltd.
abel@vallinor4.com, abel@bfr.co.il