bug-gmp
[Top][All Lists]
Advanced

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

Re: [Bug-gmp] Re: SCO OpenServer and SCO UnixWare fix to config.guess an


From: Kevin Ryde
Subject: Re: [Bug-gmp] Re: SCO OpenServer and SCO UnixWare fix to config.guess and make bug
Date: 30 Jul 2000 07:57:50 +1000
User-agent: Gnus/5.0802 (Gnus v5.8.2) Emacs/20.5

Boyd Lynn Gerber <address@hidden> writes:
>
> Here is the UnixWare message.  I though I had included it, but I did not.
> 
> gcc -c -DHAVE_CONFIG_H -I. -I. -I.. -I.. -DOPERATION_copyi -g -O
> -fomit-frame-po
> inter -mcpu=pentium -march=pentium tmp-copyi.s  -fPIC -DPIC -o
> .libs/copyi.lo
> UX:as: ERROR: tmp-copyi.s:64:maximum fill size for .align out of
> range: 0x90
> make[2]: *** [copyi.lo] Error 1
> make[2]: Leaving directory `/usr/zenez/build/uw7/gmp-3.0.1/mpn'
> make[1]: *** [all-recursive] Error 1
> make[1]: Leaving directory `/usr/zenez/build/uw7/gmp-3.0.1'
> make: *** [all-recursive-am] Error 2

The "0x90" is meant to specify a nop fill, for the benefit of old
versions of gas.  You can get rid of it by deleting the following
three lines in mpn/x86/x86-defs.m4

        define(`ALIGN',
        m4_assert_numargs(1)
        defn(`ALIGN')`,0x90')

You might want to check that the assembler does fill with 0x90.  When
configured for i386, mpn/diveby3.o for instance should have a nop at
offset 0x1f.  It'll crash very quickly if this is not so.

The next version of GMP will have an autoconf test to see whether 0x90
is needed/allowed.  Let us know if there's some variant syntax we need
to cope with.


reply via email to

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