Subject: Re: current on macppc
To: emanuel stiebler <emu@ecubics.com>
From: Markus W Kilbinger <kilbi@rad.rwth-aachen.de>
List: current-users
Date: 10/27/2001 14:16:52
>>>>> "emanuel" == emanuel stiebler <emu@ecubics.com> writes:

    >> > Were you using NFS? I was annoyed by similar "No such file or
    >> > directory" errors and gave up using NFS on macppc. Sometimes
    >> > limiting the rsize and wsize to 1024 was a valid workaround,
    >> > but in many cases I could not avoid those errors.
    >> 
    >> Nope. It is a local copy of /usr/src.

    emanuel> Addition to my own reply :-( I use NFS to do a "rsync" of
    emanuel> the cvs which in on one of my servers here. I'll check,
    emanuel> if there are differences in the trees ...

I can confirm severe (pmap, uvm related?) problems with recent
-current on my tibook, especially with nfs. They are of two kinds,
depending on using NEWPMAP or not:

With NEWPMAP the kernel doesn't panic, but disrupts many (not all) nfs
files (fatal if using /usr/*src nfs mounted). Most obvious thing I see
is the effect on 'vers.c' created on the fly in
/sys/arch/macppc/MACHINE/compile: Except the first 'ostype' line most
of the time (not always!) the other lines are simply missing! I've
never seen a disruption within a line.

There is also disruption on object files which I haven't examined.

Without NEWPMAP I don't see these file disruptions, but get kernel
panics very often.

Beside that I tried the NEW_TOOLCHAIN to compile kernels: Only with
'-O' I get working kernels. With standard '-O2' it panics somewhere in
the device attachments.

Markus.