Google Summer of Code: Improving RAIDframe parity handling

A NetBSD system, in order to tolerate disk failures, can use the software RAID driver raid(4). Currently, if that system is shut down uncleanly (e.g., loses power or crashes), then when it comes back up it will have to check the entire RAID set's redundancy information. This process can take many hours, during which it imposes a substantial load on the system. It is also a distinct disadvantage to using NetBSD in server applications, and the inclusion of a journaling filesystem in NetBSD 5 makes it all the more prominent.

The goal of my Summer of Code project is to shorten that check from hours to minutes.

[Read More] [0 comments]

 

Google Summer of Code Generic file system mounting project

Progress report of the generic file system mounting GSoC project. This project aims to remove the struct <fs>_args opaque container (the data argument in the mount(2) system call) so as to get a generic way to mount a file system whatever its type is.

[Read More] [0 comments]

 

XML tools update

My Google Summer of Code student, Nhat Minh Le, is working on a suite of simple, efficient, stream-oriented tools for processing XML on UNIX systems. Nhat Minh is making good progress on xmlgrep, a grep-alike program that understands XML syntax.

Read about Nhat Minh's progress on his blog.

Keep reading for my explanation of the niche where Nhat Minh's tools fit.

[Read More] [1 comment]

 

Google Summer of Code zfs-port project

Progress report of zfs-port Google Summer of Code project. This project aims to continue effort to port zfs file system to NetBSD. ZFS is a well known filesystem developed by Sun Microsystems, inc. for their Solaris operating system.

[Read More] [0 comments]

 

netpgp

NetBSD-current has recently had a new addition - that of netpgp, a BSD-licensed library that can perform digital signature signing and verification, and also encryption and decryption of files. An email explaining netpgp was sent to the tech-security mailing list, but I thought I'd give a short summary of that email here...

[Read More] [0 comments]

 

Impromptu bugathon

Yesterday, prompted by the threat of reaching 5000 active bug reports several developers began fixing and closing.

As of this writing the count of active PRs has dropped from 4976 to 4893 -- that's 83 resolved -- and quite a few more have been placed in submitter feedback awaiting confirmation that they are in fact fixed.

This is a positive step, and a pretty big number for a single day that wasn't a previously scheduled bugathon. Unfortunately, to clear through the backlog we'd have to do this much every single day for two full months.

Despite all the cleanup work over the past year and a half, the database is still full of reports of already-fixed problems. These not only inflate the total counts; they get in the way and make it harder to find real problems to work on. Finding and closing them needs to be a group effort.

If you're a developer: when you fix a problem please include the PR number in the commit message so it gets pulled into the database; also, please close PRs that you fix, or leave a comment explaining what still needs to be done (e.g. pullups, real fixes vs. quick hacks, etc.)

If you're a user: if you have filed a bug report and it's been fixed, but the report is still open, please follow up to let us know it can be closed. And also, note that GNATS is particular about followups: they need to be sent to gnats-bugs AT netbsd.org (not gnats-admin) and the Subject: line needs to begin with "Re:" and the category and bug number in the form "port-eniac/12345". Other followups are likely to be dropped as spam, or may end up opening a new PR instead. In any event, if you send a followup and you don't get a copy back from GNATS, it probably got lost. If you can't figure out what's wrong, post to the netbsd-bugs mailing list.

And finally, if you are tired of the Gnats search interface you may find these lists helpful... but be aware that they often don't update very rapidly as there are manual steps involved.

[0 comments]

 

Recent Xen-related Changes to NetBSD-current

With all the emphasis on NetBSD-5.0 over the last few months (and, yes, it's looking good now, there may be a third release candidate, and the final bits are being polished off now), NetBSD-current has taken a back seat. Whilst this is quite understandable, there have been some really neat features added recently. Manuel Bouyer has done even more work on NetBSD/Xen, and has added support for an i386 domain0 kernel with PAE support - this requires a i386PAE or x86_64 hypervisor (available in pkgsrc).

He has also added a PCI pass-through support for Xen3. The dom0 kernel gets a pciback PCI driver, to which the device specified in the pciback.hide boot parameter will attach. DomU kernels get a xpci device, to which pci buses will attach.

The virtualisation benefits of NetBSD are growing even more, with the recent sighting of a jym-xensuspend branch in the repository. This branch contains the code required to support the Xen save/restore/migrate facilities. It affects domU frontend drivers (xbd, xennet, xencons, hypervisor), autoconf(9) machinery, as well as MD code (mostly pmap(9)).

I have to thank both Manuel Bouyer and Jen-Yves Migeon for doing so much great work with NetBSD/Xen - formidable

[0 comments]

 

Supporting "high memory" on NetBSD/sgimips

I purchased an SGI O2 some years back, and over time I have been finding little bits of time to improve support for the hardware on NetBSD. One of the things that has always bothered me is that NetBSD could not use more than 256MB of RAM, so I set out to fix it.

[Read More] [3 comments]

 

NetBSD LVM stripe target available

Thanks to the hard work of Reinoud Zandijk, NetBSD now has stripe target for device-mapper driver available. With updated device-mapper driver, user can create striped Logical Volume.

[Read More] [0 comments]

 

WAPBL fixes committed to NetBSD-current

Andrew Doran has committed some fixes which address a number of problems with the logging feature of the fast file system. Logging provides for very fast updates of file system control data, and removes the need to check file systems after a system outage. The full details can be found in Andrew's commit message.

Fixing these aspects of logging is one of the last remaining tasks to be done before the release of 5.0, and so these changes by Andrew bring that event a bit closer - the next stage is for them to be pulled up to the NetBSD-5 branch within the source code repository.

We're very excited to finally ship logging for FFS. For many applications it enables better performance and stronger atomicity guarantees. Additionally, avoiding file system checks is of real importance for servers and embedded systems, where delays in restarting the system cannot be tolerated.

[0 comments]

 

Soft dependencies removed

Removal of the soft dependencies feature of the fast file system (FFS) was today completed by Andrew Doran. It was announced in a mail and discussed several times at the NetBSD mailing lists.

Soft dependencies will continue to be available in NetBSD 5.0. From 6.0 onwards the preferred method of maintaining integrity and achieving optimal performance with fast file systems is logging (WAPBL).

The file system format is upwards compatible: existing systems will continue to operate with no configuration changes, and migration to logging will be a simple matter of changing /etc/fstab.

[0 comments]

 

LVM2 tools update

LVM2 tools and libdevmapper library in the base system have been updated to version 2.02.44 for lvm2 tools and 1.02.31 for libdevmapper lib. LVM2 tools are part of NetBSD logical volume management subsystem.

[Read More] [0 comments]