risk of moving to a threaded sshd close to the release date. Add
mention of a potential new problem raised on freebsd-current
regarding ssh authentication sometimes failing with an ssh socket
write error.
Remove 5.0 documentation task: most of what's left is covered by
other task items, and we've hit a number of the big name issues
(devfs, mdmfs, etc).
Change the status of the truss item: there are potential fixes, but
they were deemed too risky this close to the release because they
changed parts of the fork and exec code paths.
Add a comment to the iir entry appealing for hardware for developers.
Remove the RC1 perl package issue: that's now resolved.
the only two left to convert are highly obscure and never used, according
to reliable reports. Reinstate the item about removing performance degarding
debug options still it still partially applies. Remove the item about
acpi debug messages since it was done a while ago.
UFS2 using an 8k boot block, so it's not an intractable task. Note
that we're currently looking at further testing before a merge, so
if it happens, it will be after RC2. (Note: this probably obligates
us to do a light-weight RC3).
Scott Long has committed the removal of performance impacting debugging
features. Remove this task.
The mac_lomac merge is basically complete, all that remains is the
pending documentation commit, update.
documentation if needed.
Remove gdb import task: it won't get gdb working on sparc64 using
stock gdb source code, that will require using the port either way.
As such, there's no motivation to bring in the new gdb yet.
to have much to go on right now, so other than encouraging developers
to pick up and chase these issues, it's not clear what to do with them.
If you use smbus, please test it; if you have access to a crash dump
or crash information involving inpcb, please post about it.
Move mly, iir driver issues to the RC2 list from the RC1 list.
I'd like to get these resolved by RC2, but it's not happening
before we cut RC1 starting later today.
status to patches circulating. Unless I missed the commit of the
patches, which is also possible.
Update MAC Framework userland merge task: recursive file system
labeling is now supported. A few other minor tweaks remain, if
time happens.
Update the mac_lomac task: it's pretty much complete, but we need
to merge the sample policy file for src/share (commit request
pending approval of re@) and commit the documentation. We also
need to scrub the old LOMAC from the tree (currently not built).
Commit request also pending on that.
Add a task to make sysinstall know how to speak UFS2 for partition
labeling and newfsing. I own the task; am currently circulating
patches.
submitted adduser and rmuser scripts (committed by scottl), so we're
now mostly done. All that remain are mmroff and afmtodit. Chown
task to Mark Murray since he's a guilty-looking party.
Thanks Mike, Scott.
item.
Patches were developed and sent to the one-true-awk author, Brian
Kernighan, and he was kind enough to integrate them fast, and has
already published a new version with these fixes included.
Robert Watson on behalf of re@ says that the import should be
coordinated with the FreeBSD one-true-awk maintainer, obrien@.
to be removed/modified so that the dependency no longer exists. Most
of these have been caught over the last year, but several have not,
including:
/usr/bin/mmroff
/usr/bin/afmtodit
/usr/sbin/adduser
/usr/sbin/rmuser
Add a TODO item that these must be dealt with before 5.0-RELEASE.
Submitted by: kris
a failure of the timeout() updating the system display. Otherwise,
the system seems fully function, syscons simply neglects to update
screen output. Assign ownership to scottl@ since he's tracking the
symptoms and can reproduce, but any fixes appreciated. Categorize
as "must fix" for 5.0-RC1.
Submitted by: scottl
was on 4.x-STABLE, which is to say "not perfectly", but adequately
for the purposes of the TODO item. Thanks to Tim Robbins for grabbing
this one by the horns.
Any long-term fixes for smbfs stability issues relating to forceable
unmount and dropped connections also welcome.
due to a misunderstanding that import of unpatched 5.3 would support
FreeBSD/sparc64 out of the box, but since apparently it can't, it's
now a desired feature. sparc64 users will continue to have to rely
on the gdb53 port.
Submitted by: obrien
extrapolate that we have a general problem deleting non-FreeBSD slices.
Submitted by: Christian Brueffer <chris@unixpages.org>
I thought a patch was floating around for this, but seem to have lost
it -- could someone please send it to me?
prior to the release, and populate with two items:
- Potential problems with if_ie, if_le, and if_wl from Sam Leffler's
API changes. He'd like to see immediate testing and feedback regarding
these drivers.
- Additional testing for the vlan code, which was changed in a variety
of ways late in the development cycle.
Additional submissions welcome.
information on the specific cards that are seeing failures. I'd like
to see more testing of the mly and iir drivers with other configurations
to see if it's a general driver problem. For both of these configurations,
4.7-RELEASE appears to work just fine.
(1) obrien is reporting problems using smbfs to access Win2k shares, and
there have been other less specific reports of problems with smbfs.
Testing of smbfs in various environments, and fixing of bugs, is
much encouraged.
(2) NETNCP and newfs appear still not to be reconnected to the build from
the original KSE integration work, and do not build. Fixing these
is complicated by an apparent lack of Netware installs easily
available to developers, meaning that while syntactic fixes might
be made, it's difficult to know if they are right.
Developers are encouraged to explore these problems and see if we can't
work out what's going on: smbfs and nwfs are both very important to
the success of FreeBSD 5.0.
been resolved. Some still remain regarding OpenSSH/OpenPAM.
Add a task item relating to the lastlog permission denied warning from
OpenSSH, and set ownership to DES.
