NetBSD-Bugs archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: port-sun2/56158: sun2 m68000: super-block backups (for fsck -b #) at: 32,nbmakefs: Writing inode 152 (work/./bin/cat), bytes 528384 + 4096: No space left on device



Hi Matthew,

On Sat, 2021-05-08 21:20:02 +0000, matthew green <mrg%eterna.com.au@localhost> wrote:
>  guess here -- do you by chance have HAVE_GCC=10 set in the
>  environment some where?  i see that with GCC 10 and sun2.

My initial GCC in use is (nearly) GCC "master" (for the ogged build,
it was:
	/usr/lib/gcc-snapshot/bin/gcc --version
	gcc (Debian 20210320-1) 11.0.1 20210320 (experimental) [master revision 3279a9a5a9a:6526c452d22:5f256a70a05fcfc5a1caf56678ceb12b4f87f781]
), but I did not (at least: not intentionally) set HAVE_GCC=10
anywhere. To be sure, I'll add an `export` call just before the
./build.sh invocations to be sure.

For this build, I had export'ed CC=gcc-snapshot (as above) and called
build.sh on an amd64 Linux host to cross-build a sun2 release.

>  sun2 is still using GCC 9 because the ramdisk overflows like
>  this and the kernel is already too large with the currently
>  sized ramdisk to be loaded.  (the kernel has a limited size,
>  and it has grown some with GCC 10, and also, the ramdisk has
>  overflowed it's allocation, wanting the kernel to be even
>  larger to accept the ramdisk.  i have no real solution as
>  the INSTALL kernel and the ramdisk are both already extremely
>  limited..

Sounds reasonable. So your guess is that, contrary to what a sun2
build should have used (GCC 9), by setting HAVE_GCC=10 a wrong gcc
would have been build in `./build.sh tools`?  I'm quite sure not
having set HAVE_GCC, but we'll see---will keep you updated!

MfG, JBG

-- 

Attachment: signature.asc
Description: PGP signature



Home | Main Index | Thread Index | Old Index