This is the mail archive of the gdb@sources.redhat.com mailing list for the GDB project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: gdbserver cross compile


To be honest, I don't remember how I solved this in our local
environment... you could just rebuild that one configure file locally
with 2.5x if that works.  I think it's simple enough to work in both. 
Most other configure files in the source are not.

On Mon, May 05, 2003 at 05:46:16PM -0400, Harris, Jeff wrote:
> I tried setting both cross_compiling and ac_cv_prog_cc_cross environment
> variables to "yes" before running configure.  Neither seemed to stop it from
> running its program and overriding my values.
> 
> In thinking about setting ac_cv_header_stdc to yes, this also won't help
> since the problem program, the one which checks ctype macros, will still be
> run unless I can get cross_compiling to be set to "yes".
> 
> Jeff
> -----Original Message-----
> From: Daniel Jacobowitz [mailto:drow@mvista.com] 
> Sent: Monday, May 05, 2003 5:35 PM
> To: Harris, Jeff
> Cc: gdb@sources.redhat.com
> Subject: Re: gdbserver cross compile
> 
> 
> On Mon, May 05, 2003 at 05:29:48PM -0400, Harris, Jeff wrote:
> > I was having problems with getting gdbserver to correctly detect whether
> it
> > is being built with a cross-compiler or not.  I was trying to
> cross-compile
> > from i386 (RedHat 7.2) to i386(gcc 3.2.3 and glibc 2.3.2).  The configure
> > script for gdbserver looks to have been created with an old version of
> > autoconf (2.13 I believe).  The mechanism it uses to detect a cross
> compiler
> > simply tries to compile a simple program and run it.  If it runs, it
> decides
> > it isn't cross-compiling otherwise it is cross-compiling.  
> > 
> > My problem was that while this program runs with the cross-compiler, the
> > program to detect whether standard C headers are used does not run.  As a
> > result, the STDC_HEADERS #define is not set and ultimately a compile error
> > to occur.
> > 
> > There doesn't seem to be a way to override the configure script's
> detection
> > mechanism.  However, if I rerun autoconf (version 2.53) in the gdbserver
> > directory, the resulting configure script is updated to include better
> > logic.  Basically, it compares the build and host options to configure to
> > determine its cross-compile state.  This mechanism works for me since I
> have
> > different values for these options.
> > 
> > Is an updated version of configure something that would be considered for
> > GDB 5.4?  
> 
> Sadly, no.  You can't use autoconf 2.5x with the current source tree,
> in general; the conversion process is still being talked about.
> 
> It's pretty easy to fake out the 2.13 logic for this case.  I don't
> remember exactly what the magic was, since I can't find any specific
> reference to it in our cross build scripts; but all you should need to
> do is preload one of the configure script's cache variables with
> something that says no, don't run programs.
> 
> -- 
> Daniel Jacobowitz
> MontaVista Software                         Debian GNU/Linux Developer
> 

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]