On Wed, Jan 10, 2001 at 12:28:26AM +1030, Brett Lymn wrote:
> 
> Folks,
>         I am tracking the -current version of NetBSD and I suspect
> that something in the kernel has changed (probably UBC integration)
> that has upset coda.  I get these messages:
> 
> Jan  9 23:44:23 siren /netbsd: uvn_flush: size not set vp 0xd417f514
> Jan  9 23:44:23 siren /netbsd: uvn_flush VSIZENOTSET: tag 18 type VREG, usecount 0, writecount 0, refcount 0, flags (VXLOCK)

This could be serious.

> Jan  9 23:44:23 siren /netbsd:  coda_vop_error: Vnode operation vop_print called, but not defined.

This one is probably innocent.

> Jan  9 23:45:00 siren /netbsd: coda_nc_find: name ., new cred = 0xc08f4f00, cred = 0xc09e1680
> Jan  9 23:45:00 siren /netbsd: nref 182, nuid 0, ngid 0 // oref 252, ocred 100, ogid 100
> Jan  9 23:45:00 siren /netbsd: ref 182  uid 0
> Jan  9 23:45:00 siren /netbsd:  group 0: (0)
> Jan  9 23:45:00 siren /netbsd: 
> Jan  9 23:45:00 siren /netbsd: ref 252  uid 100
> Jan  9 23:45:00 siren /netbsd:  group 0: (100)
> Jan  9 23:45:00 siren /netbsd:  group 1: (0)
> Jan  9 23:45:00 siren /netbsd: 

These looks like `innocent' debug messages, except if ref is a reference
count in which case 182 is way to high.

> Things _seem_ to be working ok except when I do a "make install" as
> root which tries to access some weird file names, viz:
> 
> [root_at_siren] cd /coda/working/basesrc/lib/libcurses/
> [root_at_siren] make install
> install ===> PSD.doc
> install -r   -c -o root  -g wheel -m 444 Makefile
> /usr/share/doc/psd/19.curses/Makefile
> install: /usr/share/doc/psd/19.curses/inst.21188a: Bad address
> *** Error code 1

Ok, there is definitely something wrong. I will try to keep in mind to
take a look at the VFS changes that went into netbsd current.

Jan