bug-gmp
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: bugs installing gmp-4.0


From: Torbjorn Granlund
Subject: Re: bugs installing gmp-4.0
Date: 11 Dec 2001 21:19:20 +0100
User-agent: Gnus/5.0807 (Gnus v5.8.7) Emacs/20.7

Nathalie Revol <address@hidden> writes:

  On an INTEL PENTIUM III,
  
  uname -a
  SunOS butanol 5.8 Generic_108529-10 i86pc i386 i86pc
  
  configure used flags for pentium-pro whereas it is a pentium 3,
  here are some extracts of the configure messages:

That is not necessarily a sign of any error.  Pentium Pro
and Pentium III are the same thing from the compiler's point
of view.

What compiler version do you use.  Please use `gcc -v' to
find out.

I think I see what is happening here.

What happens here is that the assembler apparently doesn't
understand the standard form of the CMOV instructions added
with the introduction of Pentium Pro.  Sun thought they'd
invent "better" names for these instructions.

The bug could be said to be in gcc really; it should perhaps
use the non-standard mnemonics required by the Sun
assembler.

For GMP, we should work around this bug by avoiding the
-march=pentiumpro option when no sensible assembler can be
found.

You should perhaps get the latest binutils release from
<ftp://ftp.gnu.org/gnu/binutils/> and install that on your
system.  That will give you a better working assembler.

Kevin, could you add a configure test for this?

  I also installed it on my Sun Ultrasparc, and there is a problem with sed
  at the end of the compilation. (The compiler does not matter, it works
  the same with cc:
  cc -V
  cc: WorkShop Compilers 5.0 98/12/15 C 5.0
  )
  Anyway, gmp-4.0 is installed.

This is a bug/limitation of the system sed.
Please install <ftp://ftp.gnu.org/gnu/sed/>

--
Torbjörn



reply via email to

[Prev in Thread] Current Thread [Next in Thread]