Re: problems with -current

From: Pete Carah <pete_at_altadena.net>
Date: Tue, 29 Nov 2005 12:55:02 -0800
Maksim Yevmenkin wrote:
> dear hackers,
> 
> after cvsup'ing recent current (as of yesterday evening pst) and
> building/installing new kernel/world i started having problems:
> 
> problem 1: startx simply reboots my laptop. i can see x coming up, then
> after just a few seconds my laptop reboots. no window manager, no
> nothing. my laptop is a dell with some radeon video card. will provide
> more details on request.

On mine (I865 integrated controller) the system hangs on startx; I did
see one panic but couldn't tell what it was since system booted directly
from graphics mode.

> 
> problem 2: when system is booted and filesystem is not clean fsck is
> started and checks all dirty filesystems. then boot process continues as
> usual, however root filesystem is still mounted read-only, i.e.
> 
> --->  /dev/ad0s1a on / (ufs, local, read-only)  <---
> devfs on /dev (devfs, local)
> /dev/ad0s1d on /tmp (ufs, local, soft-updates)
> /dev/ad0s1f on /usr (ufs, local, soft-updates)
> /dev/ad0s1e on /var (ufs, local, soft-updates)
> linprocfs on /usr/compat/linux/proc (linprocfs, local)
> devfs on /var/named/dev (devfs, local)

The RW update happens right if there was no fsck, but fails if there was
one.  This goes back to the pre-mid-1990's thing about doing an
immediate reboot after any fsck...  In addition, there was an API change
that didn't make it into UPDATING: if mount(8) and friends (mount_*) are
recompiled and tried with an old kernel, NOTHING related to mount
options seems to work; the interface for options changed.  (this has
happened slowly; mount_msdos picked up the change last week sometime but
ufs only recently).  Made things so that I couldn't use noauto since it
got presented to the driver which rejected the mount.  Not serious for
usb drives since all it does with msdosfs is not mount anyhow, and not
hang the boot or drop into single-user.  BUT this kind of change is
supposed to go into UPDATING *BEFORE COMMIT*, since it requires sync
between userland and kernel.  Not to speak of apparently not being
finished!!!!

> 
> so, anyone sees this as well or its just me.
Yes.

-- Pete
Received on Tue Nov 29 2005 - 19:58:23 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:38:48 UTC