Subject: Re: build.sh -m vax fail, miniroot full
To: Manuel Bouyer <bouyer@antioche.eu.org>
From: Luke Mewburn <lukem@netbsd.org>
List: current-users
Date: 11/19/2002 09:39:26
On Mon, Nov 18, 2002 at 10:39:26PM +0100, Manuel Bouyer wrote:
  | Hi,
  | A current 'build.sh -m vax' fails with:
  | /home/vax/obj/home/sources/src/tools/tools.NetBSD-1.6I-alpha/bin/nbmakefs -t ffs -B le -s 7m -F work.spec  -N /home/sources/src/etc -o bsize=4096,fsize=512  -o 
  | optimization=space,minfree=0,nsectors=1,ntracks=128  -o density=5k miniroot.fs.tmp work  && mv -f miniroot.fs.tmp miniroot.fs
  | Calculated size of `miniroot.fs.tmp': 7340032 bytes, 1457 inodes
  | miniroot.fs.tmp:        14336 sectors in 112 cylinders of 128 tracks, 1 sectors
  | 	7.0MB in 1 cyl groups (153 c/g, 9.56MB/g, 1408 i/g)
  | super-block backups (for fsck -b #) at:
  | 32,nbmakefs: Image file `miniroot.fs.tmp' has 1406 free inodes; 1457 are required.
  | nbmakefs: Image file `miniroot.fs.tmp' not created.
  | 
  | *** Error code 1
  | 
  | Stop.
  | nbmake: stopped in /home/sources/src/distrib/miniroot
  | 
  | Any idea how to fix this ?
  | I didn't find where the makefs parameters comes from ...

It's usually a matter of tweaking stuff under distrib/vax and looking
at the shared stuff under distrib/common.   Tuning the makefs parameters
can be a bit tricky to get "right" when the target FS is small and full.

Did you compile your stuff with a different optimisation level (the
usual culprit for the install file system running out of space?)

Also, IIRC, alpha->vax cross builds didn't generate a working release,
which is why for NetBSD 1.6 we migrated the vax build back off Matt's
CS20 (alpha) to the dual i386 build box.  (This is probably unrelated
to your specific problem, but may bite you anyway).