RE: Automated arm-linux-gcc and powerpc-linux-gcc build script

[Yves Rutschle <y dot rutschle at indigovision dot com>]

> Even if you know how to build a crossgcc, please look at and tell me
> if there's anything I'm doing wrong.  I think the procedure is
> correct, however, both because I finally understand all of it, and
> because it *doesn't* need newlib.  :^)

So, I've been recompiling compilers all day and here are
my results and comments:
- no comment on the procedure, I think it's good.
- It worked perfectly straigt out of the box on my RedHat 7.something
- It didn't work on Cygwin; gcc-core builds fine, then glibc
crashes somewhere in the build with the following message:

make  -C math subdir_lib
make[2]: Entering directory `/newgcc/glibc-2.2.2/math'
.././scripts/mkinstalldirs /newgcc/build-glibc/math
mkdir /newgcc/build-glibc/math
/cygdrive/p/perl/bin/perl gen-libm-test.pl -u
../sysdeps/arm/libm-test-ulps -o "/newgcc/build-glibc/math/"
Can't open /newgcc/build-glibc/math/libm-test.c: No such file or
directory at gen-libm-test.pl line 430.
make[2]: *** [/newgcc/build-glibc/math/libm-test.stmp] Error 2
make[2]: Leaving directory `/newgcc/glibc-2.2.2/math'
make[1]: *** [math/subdir_lib] Error 2
make[1]: Leaving directory `/newgcc/glibc-2.2.2'
make: *** [all] Error 2

I didn't get a chance to look more into it yet. It might be
related to those infamous CR-LF translation thingy, my copy
of Cygwin is very poorly configured (I don't usually use it
much at all and as a matter of fact I don't understand why
the CR-LF should be a problem. Anyway). Although if it is so,
why should it only crash in the middle of glibc. hmm.

On a side note, I have a couple of comments:
- It could be good to provide an option for those people who
already have a configured linux kernel to pick up the headers
from there rather than unpacking/configuring again.
- I am not sure a "solve all" script is the way to go. gcc-2.95.2
for ARM required a specific patch for example. I think when getting
all the specific cases in, you'll end up with a monster script
that will be unmaintainable. After all, that's what configure is
supposed to do already :-)  On the other hand, it does show the
steps to follow, which is good.
- I think there should be a note somewhere in the script warning
that the process will generate 900 MB of files...
- The script should definitely stop after an error (I think
make returns 0 on success and some error code otherwise.. my
scripting abilities are weak so I don't really know how you'd
test that but I think that would be the best way to go).

Cheers,
Yves



------
Want more information?  See the CrossGCC FAQ, http://www.objsw.com/CrossGCC/
Want to unsubscribe? Send a note to crossgcc-unsubscribe@sourceware.cygnus.com