On Dec 5, 2008, at 4:00 PM, David Cantrell wrote:
I'm gzipping the patch and attaching it because the list manager keeps
telling me that the size is too large. It is, but, well, that's how
it
came from IBM.
If I don't hear anything from people after a few days, I will take it
that people are ok with this patch going in to rawhide. It's
basically
just a big rewrite to linuxrc.s390. There is also a new helper script
for linuxrc.s390 that scans for devices on the system. And a one line
patch to mk-images to make sure the new helper script is installed.
So, some comments from trying to go through and at least look for
obvious problems/things to fix. Some of these may only be rawhide
applicable and given that the patch was against the RHEL5 codebase,
they may not be (easily) doable there
* A lot of checking for things like a 2.6 kernel. There's _always_
going to be at least a 2.6 kernel and such checks just needlessly
complicate the script, making it that much harder to see what's really
going on
* Are udev and friends getting installed into the s390 initrd right now?
* Some of the stuff in lsznet makes me think that modules should be
getting auto-loaded by udev given the existence of modaliases and
matching against things in /sys/bus/ccw
* Didn't we stop allowing telnetd and switch to only sshd for logging
into the install environment on s390 or am I just imagining things
that I wish we had done? :)
* Manual mknods shouldn't ever need to be present; udev should be
creating device nodes
* modprobe instead of insmod if modules have to be manually loaded so
that at least dependent modules can be auto-loaded. This also lets us
kill the $LO hack
* There is code duplicated between the linuxrc and lsznet (see, eg,
the declaration of $CU) and there's already divergence between the two
* cardtype2cleartext() is the sort of thing which gets out of date and
then becomes critical must fix bugs... if this information is
important, it probably should be something we can query either with
modinfo on a module or having the kernel tell it to us directly
* Aren't we now doing udev rules for persistent device naming rather
than modprobe.conf aliases?
* Why are they disabling ipv6 autoconf?
* Some of the helptext (helptext_nettype() is the one I'm seeing right
now) is concerning. If something isn't supported, then we shouldn't
support it. If the code is there, it's going to be considered supported
* Also, given that these values have changed at least somewhat over
time, if there's going to be help text, there needs to be some
commitment for keeping it updated
* Translators are going to hate a lot of the text. And given that
this is being shown on the line mode terminal, how many translations
can really be displayed?
* Is there a reason if your IPL device is FCP that you wouldn't be
doing a CD install? Shouldn't we default to detecting the CD like we
do on other arches here unless you've passed 'askmethod'
* Checking bash versions is also silly
Okay, now that my eyes are bleeding, I'm going to be done for now.
The mix of new functionality, reformatting and changing the way things
work makes this essentially unreviewable for correctness so I wasn't
really able to address anything there :/
Jeremy
_______________________________________________
Anaconda-devel-list mailing list
Anaconda-devel-list@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/anaconda-devel-list
