nalin at redhat.com
Fri Sep 24 23:40:51 UTC 2004
On Sat, Sep 25, 2004 at 01:14:12AM +0200, Thomas Zehetbauer wrote:
> although I suspect it is not recommended practice I do a regulary "rpm
> --freshen *.rpm" from my local rawhide mirror. This recently left me
> with an almost empty /dev directory that did not even allow me to boot
> with init=/bin/bash. I guess the udev update was missing udevstart in
> the %postinstall script.
The trigger in udev should have done that when dev was removed. Did you
have UDEV_INITRD changed from its default to "no"? That'd break your
box at reboot (not starting udev early implies that you can depend on
/dev being there, and it's gone), and then it's time to find a rescue
> Now I am still missing at least my USB tty devices (for Palm)
Historically, I wasn't been able to attach to the device until the
PDA was attempting to sync. When you press the sync button, does udev
create the device nodes? If not, then there's a bug somewhere.
> and have a
> non world-readable /dev/rtc.
That's a "MAKEDEV-and-udev disagree on permissions" thing. There are
probably going to be others -- please file this as a bug against either
MAKEDEV or udev. At some point maybe I should wrestle a udev person for
the right to decide, and take a dive.
More information about the test