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]

Re: compile problem


On 25 Oct 2001, Frans J King wrote:

> I'm not sure if this is the right place to be posting this byt here
> goes.

This is the right place AFAICT.

> I can't compile gdb version 5.0 upwards for arm-elf. The command to
> configure I use is ./configure --host=i686-pc-gnu --target=arm-elf
> --prefix=/usr/local

Three things followed by a lot of conjecture...

1) Don't use 5.0 for arm. It's really broken. Get something newer.
2) If you're configuring for building on the same machine you're running
   on, don't specify "--host", it's not needed. I have had strange
   experiences with configure when I specified "--host and --target" but
   not "--build". So specify ALL three or just "--target" to be safe.
   I'm not saying this is the cause of your problem, or that this is
   necesssarily a problem. It is just my experience. YMMV.
3) On what host system are you trying to build (os, vendor)?

> Once I begin the make, it fails at pty_termios.c in the expect
> directory.
>
> pty_termios.c:174 conflicting types for 'slave_name'
> pty_termios.c:135 previous declaration of 'slave_name'
> pty_termios.c: In function 'exp_getptymaster':
> pty_termios.c:377: incompatible types in assignment

This is in expect, of all places. Your conflict is happening because both
HAVE_OPENPTY and one (or more) of HAVE__GETPTY, HAVE_PTC_PTS, or HAVE_PTMX
is defined by configure. Sounds like configure is messed up about
something.

HAVE_OPENPTY will get defined when openpty () exists, but only for
non-linux hosts. If you're using linux, then #2 above is a solution to
your problem. You'll know you've got this in your build, too, because
"ac_cv_func_openpty" will be set to "yes" in config.cache.

HAVE_PTC_PTS is defined when "AIX new-style pty allocation" exists, i.e.,
/dev/ptc exists but /dev/pts does not. I doubt this is your problem.

HAVE_PTMX is defined when "SVR4 style pty allocation" exists, i.e.,
/dev/ptmx exists, and sysVr4 ptys are determined to be broken. Unlikely to
be your problem.

My guess: the openpty thing.

Keith



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