Alex Hornung’s I/O scheduler is now in DragonFly; all reports I’ve seen from users say it makes interactivity much better. It’s not on by default; read his very detailed post and followup for details.
‘dylanr’ has built 2 interesting films using Gource to visualize DragonFly development; he’s mentioned them in comments here, but I want to make sure people see them.
Update: see the full multi-year film of DragonFly commits linked in the comments by dylanr; thanks for doing this!
OpenSSL (which recently hit 1.0, though that’s not in DragonFly yet) has been patched to cover a recent security issue, thanks to Peter Avalos.
Jan Lentfer’s updated BIND to 9.5.2-P3, too.
Venkatesh Srinivas’s new sysctl, “debug.panic” is available for those who want to panic their machine on purpose, but don’t have direct access to the keyboard.
A problem found by Jan Lentfer and fixed by Matthew Dillon means that you can get a good performance boost if you’re running bleeding-edge DragonFly from the last month or so. Or, you can just wait a week for the 2.6 release.
DragonFly 2.6 has been branched, and should be released next week. Check the tag message for a list of the many, many commits.
YONETANI Tomokazu has eliminated cvsup, replacing it with net/csup from pkgsrc. The README notes that the pkgsrc package devel/cvsync is another alternative if you need to retrieve the repository and not just the checked out files..
Matthew Dillon has implemented what he calls “REDO” records in Hammer, which reduce the amount of time taken flushing data to disk. It’ll be in the 2.6 release, but it isn’t on by default.
Jordan Gordeev’s work on 64-bit vkernels has also been brought in, so virtual systems are now available for x86_64 users.
Matthew Dillon’s added some tools for building system snapshots, a previously custom process. Look at the README for details.
Alex Hornung has committed his initial work on Linux support, which is over 6,500 lines so far. (Thanks, Alex!) He’s continuing to work on it, though going by his commit message, Java, Opera, Tomcat, etc. are supported so far. The only major item missing at this point is Flash. There are other followups, such as this note about chrooting into the Linux subsystem.
“Device initiated power management” via AHCI is now possible, thanks to Johannes Hofmann. If I understand it correctly, it lets the computer handle power reductions automatically, which is more efficient than setting by hand.
Michael Neumann has added his port of the e1000 driver from FreeBSD, though he doesn’t recommend using it yet. He’s looking for testers who have this hardware.
OpenSSL version 0.9.8m has been imported by Peter Avalos; this version contains a bugfix for a security issue.
There’s several projects in the works:
- Rui Paulo is working on an update of wireless device support in DragonFly; Matthew Dillon added libbsdxml to DragonFly as part of the prep work.
- Joe Talbott is working on unionfs; Nicolas Thery supplied some links, while Matthew Dillon wants deduplication.
Michael Neumann has fixed the ability to stream Hammer data between 32 and 64 bit systems. However, this is a change to 64-bit systems that requires them to match; make sure that you are not mixing 64-bit systems built before and after this commit on the 21st.
I can’t find the commit message in the mail archive, so I’ll quote it here:
Antonio Huete Jimenez notes that some programs have been enabled in the x86_64 build; if you’re running bleeding edge 2.5, please try them and see if they work in 64-bit.
Michael Neumann has ported igb(4) and em(4), and he needs people with the corresponding hardware to test it. Those are network cards, if you aren’t familiar with those short names.
This one’s a few days old, but I’m still trying to catch up with all the events lately: swapcache now has two flags to control whether just meta-data or all data is cached for any given set of directories; caching everything is only worthwhile if the swapcache device can keep up with the resulting traffic.
This requires a rebuild of world, if you’re running 2.5 bleeding edge.
Naoya Sugioka’s tmpfs port is now ready to go. It’s still considered experimental, but it’s worth trying. tmpfs(5) is different because it keeps data in RAM once, and pages out only when needed. This best-case scenario is an improvement to mount_mfs(8) and md(4), its predecessors.
It’s running now on pkgbox64 and already seems to be speeding up the bulk build process.