Simon ‘corecode’ Schubert has committed Noah Yan’s work on AMD64 support, making it possible to cross-compile a 64-bit world (not kernel, not yet) on a 32-bit system.
Matthew Dillon has changed the way the installer CDs are built, and mentioned an idea I support wholeheartedly: creating DragonFly CDs/DVDs that come with common software already installed. His changes don’t make that happen, but they do make the possibility easier.
Peter Avalos has updated libarchive to 2.2.6, though it doesn’t look like it changes much for us.
CARP has been added to DragonFly. For those unfamiliar with it, it’s a protocol for having an IP address fail over to a new system without (significant) interruption, similar to IETF/Cisco’s VRRP.
Matthew Dillon’s added ‘part 1/many‘ of the work needed for supporting file systems in userland.
Matthew Dillon found a mbuf problem, and fixed it. It’s severe enough that it will cause 1.10.1 to be brought out very soon; it’s led to some other changes.
Something that is a very good idea: When running 1.10 or later, it will be possible to cd to /usr and update DragonFly source or pkgsrc files using only a make command.
Matthew Dillon has updated libarchive, including a fix for a recent security issue.
Joe Talbott has added support for tying virtual CPUs (in virtual kernels, naturally) to real CPUs, so that a multiprocessor vkernel will actually use multiple processors.
A big thank you goes out to Peter Avalos, who brought in a large quantity of updates to the ahc(4) driver, originally from FreeBSD. Check this month’s commits to see his name a whole bunch of times. (Someone correct me if I have that man page link wrong.)
Matthew Dillon has added a page to the DragonFly website listing the PGP key for the DragonFly security officer.
Credit goes to Hasso Tepper for recently doing a lot of cleanup in DragonFly’s USB code. This is one of those contributions that should be recognized, though it’s hard to have any one part to link to here. Here’s one commit of many.
It’s a buzzword frenzy! Matthew Dillon has committed work performed by he and Joe Talbott on making virtual kernels able to emulate up to 31 CPUs at a time. This has the side effect of making it almost possible to hot-swap CPUs, easily. Testers wanted.
If you want to experiment, Matthew Dillon has added a little program that tests syslink connectivity. Syslink is the protocol that will be used for communication between DragonFly systems in a cluster. Don’t get too excited – it’s just a test program.
Simon ‘corecode’ Schubert has added his patch to allow separate volume control for each application. Also, Hasso Tepper has produced a patch for sound that includes changes taken from FreeBSD 6, which improves device support.
Matthew Dillon asks that 1.9 users test using USB memory devices; he’s recently committed a large number of fixes related to physically removing mounted USB drives. Also, automatically mounting reconnected drives is a small, easy project enabled by this recent work. (See linked article for details.)
Matthew Dillon, while investigating a separate problem, ended up improving the separation between CPUs in a multiprocessor system. The Big Giant Lock is still there, but it’s a move in the right direction.
Matthew Dillon was finally able to reproduce the problems some people with older ATA chipsets would have with the new ATA code; he made some subsequent fixes (working late) along with Simon ‘corecode’ Schubert, and it seems the ATA problems are fixed.
Matthew Dillon has changed virtual kernels so that they can be internally restarted or shut down, just like a real system. OMG.
If you are running bleeding edge DragonFly, make sure your next buildworld/buildkernel is a full one. Matthew Dillon has made changes to ccdconfig/vnconfig that require it.