On Sat, 2021-05-08 19:00:03 +0000, David Holland <dholland-bugs%netbsd.org@localhost> wrote:
> On Sat, May 08, 2021 at 02:10:01PM +0000, jbglaw%lug-owl.de@localhost wrote:
> > 32,nbmakefs: Writing inode 152 (work/./bin/cat), bytes 528384 + 4096: No space left on device
>
> This means makefs thinks bin/cat is at least 528384 bytes long and
> likely longer. I assume it's crunchgen output, so being that size in
> the first place is not ridiculous, but ... I wonder if crunchgen
> broke.
>
> You say you're doing toolchain tests but didn't say what you were
> testing...
Toolchain testing was my starting point: I'm trying to regularly build
a good number of cross-gcc configurations. With that setup running, I
moved over to give further projects a run. NetBSD came to mind, so I
did. Basically, fetching NetBSD trunk from the Github mirror (though
that's a bit awkward as a plain "git pull" won't work too well) and
then run:
./build.sh -U -O "${OBJ_DIR}" -m "${BUILD_MACHINE}" -a "${BUILD_ARCH}" tools
./build.sh -U -u -O "${OBJ_DIR}" -m "${BUILD_MACHINE}" -a "${BUILD_ARCH}" release
./build.sh -U -u -O "${OBJ_DIR}" -m "${BUILD_MACHINE}" -a "${BUILD_ARCH}" iso-image
That's with $CC / $CXX being set/exported to Debian's "gcc-snapshot"
binaries right now, could also use plain system gcc or clang.
I'll re-test and monitor HDD consumption, but with > 100 GB free I
rather guess that it's not my disk filling up, but a disk image. As
I'm new into cross-building NetBSD, I might have done things quite
wrong, but nothing obviously IMHO, so I decided to drop a bug report.
MfG, JBG
--
Attachment:
signature.asc
Description: PGP signature