This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [committed[ config.gcc: abstract thread_File stting for allrtemstargets
- From: Joel Sherrill <joel dot sherrill at OARcorp dot com>
- To: Ralf Corsepius <corsepiu at faw dot uni-ulm dot de>
- Cc: Nathanael Nerode <neroden at twcny dot rr dot com>, GCC Patches <gcc-patches at gcc dot gnu dot org>
- Date: Tue, 30 Sep 2003 08:00:57 -0500
- Subject: Re: [committed[ config.gcc: abstract thread_File stting for allrtemstargets
- Organization: OAR Corporation
- References: <20030930095924.GA2720@twcny.rr.com> <1064919501.32444.762.camel@mccallum.corsepiu.faw.uni-ulm.de>
Ralf Corsepius wrote:
On Tue, 2003-09-30 at 11:59, Nathanael Nerode wrote:
use_fixproto=no
^^^^^^^^^^^^^^^
... I am hesitant about this.
But that's not a behavior change. That was just a relocation from a random
clause in the t-rtems file.
Urgh, I see ... STMP_FIXPROTO is hard-coded into t-rtems.
I currently think you have tripped a bug in RTEMS gcc.
Doesn't configuring with --with-newlib suppress it, too?
I am coming in late so excuse my commenting on back messages.
AFAIK there is no one using any C library with RTEMS except
newlib. So making the thread and fixproto RTEMS configurery
common is valid. If it turns out in the future that someone
wants to build with another C library, then all that logic
will be in one place and can be wrapped with a check that
newlib is enabled. It can be done now, if we want to plan
ahead.
Nathaniel noticed a pretty important point -- the *-rtems*
targets are intentionally as similar as possible. So there
may be other opportunities for sharing.
From a user's perspective, the instructions for invoking gcc
to compile end user code follow this pattern:
TARGET-gcc -qrtems -BPREFIX/TARGET/lib/ \
-BPREFIX/TARGET/BOARD/lib/ --specs bsp_specs -c FILE.c
Each board provides a specs extension file which specifies how to
link an application. I would love to see that abstracted into
an RTEMS target independent argument to gcc like --rtems-bsp=BOARD
and the magic be inside gcc rather than visible to the user.
But unfortunately, I don't have a clue where to begin to add that.
Sadly the use of the -B and specs extensions has made it impossible
to find a way to run the gcc testsuite directly on rtems targets.
That's why I usually test the *-elf* cousin.
So long term, any commonality spotted and leveraged improves things.
Ralf
--
Joel Sherrill, Ph.D. Director of Research & Development
joel@OARcorp.com On-Line Applications Research
Ask me about RTEMS: a free RTOS Huntsville AL 35805
Support Available (256) 722-9985