bug-gmp
[Top][All Lists]
Advanced

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

Re: Problems linking with gmp...


From: Will Galway
Subject: Re: Problems linking with gmp...
Date: Wed, 11 Sep 2002 20:04:06 -0700 (PDT)

Kevin Ryde writes:
 > Will Galway <address@hidden> writes:
 > >
 > >   > cc gmp-test.c -I/home/users/galway/include -L/home/users/galway/lib 
 > > -lgmp -o gmp-test
 > >   Undefined                       first referenced
 > >    symbol                             in file
 > >   __gmp_printf                        gmp-test.o
 > >   ld: fatal: Symbol referencing errors. No output written to gmp-test
 > 
 > If that's the only symbol missing, ie. it found the others, then
 > perhaps you're hitting an older libgmp, pre-dating gmp_printf.  I
 > don't know why that would happen if the -L is right.  Maybe adding -V
 > or -Wl,-V for verbose output would show what's going on.

Well, it took me quite awhile, but I think I now understand what was
going wrong.

There >>is<< an "older libgmp" that is being hit, on /usr/local/lib/,
which has a vintage 3.1 libgmp installed.  However, as you note, I'd
expect my -L to override that.  HOWEVER, there seems to be some issue
as to what architecture I'm compiling for, the default when running cc
is NOT "-xarch=v9".  (I'm not sure what the corresponding gcc
construct is.)

So, when I compile my test program without "-xarch=v9", at link time
ld "correctly" decides to link to the older libgmp, compiled without
"-xarch=v9".  

If, I instead compile using

 cc -xarch=v9 gmp-test.c -I/home/users/galway/include -L/home/users/galway/lib 
-lgmp -o gmp-test

things work great.  So, I guess it was a "simple" problem, but not
quite as simple as I'd first thought!

Thanks!

--Will
( mailto:address@hidden http://www.cecm.sfu.ca/~wfgalway )




reply via email to

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