Peter Avalos has updated OpenSSL to version 0.9.8h, which fixes “two moderate security flaws“. The original diffs came from Andras Voroskoi.
Hasso Tepper has committed Dashu Huang’s “RFC3542 support” Summer of Code project.
In addition to committing acpi_cpu(4), Hasso Tepper has also enabled the powering-down of unused PCI devices. His post to users@ explains the details.
Matthias Schmidt has added a release info page for 2.2. This next release won’t be out for a few months, yet, but if you’re adding something new for that release, write it there so we don’t have to remember it all within 24 hours in January.
Remember rconfig(8)? Matthew Dillon’s added an example that will format a disk with UFS /boot and Hammer /everythingelse.
Matthew Dillon’s made more changes to the boot process, allowing the boot code to boot directly from a /boot partition. I’m abusing the English language with that last sentence.)Â This allows having a UFS /boot and a Hammer /everythingelse
More chunks of the DragonFly Summer of Code projects are getting committed – recently, it’s Louisa Luciani’s LiveDVD work and some of Max Lindner’s work on dma(8). (more DMA work forthcoming)
dhcp-3 has been removed from the base install of DragonFly. Instead, the install CD will come with the pkgsrc version. Matthias Schmidt and Andras Voroskoi ported over the OpenBSD version of dhclient.
Matthew Dillon has committed a significant amount of work from Jordan Gordeev’s Summer of Code project, for AMD64 support. (It is very close to being able to completely boot an AMD64 kernel) As he says in the commit message, the code is the product of many folks, but with much credit to Jordan Gordeev for getting the work to this point. As far as I know, Jordan will continue working on this past the Summer of Code, which makes it a double success.
Matthew Dillon’s been committing parts (example link) of Jordan Gordeev’s Summer of Code project for AMD64 support. It’s not done yet, but it should be by end-of-year.
Sepherosa Ziehau’s work on parallelizing DragonFly networking can be tried out (for those running bleeding edge code) by setting the sysctl kern.intr_mpsafe to 1.
Aggelos Economopoulos is looking for feedback for his NetMP (meaning giant lock removal from the network stack) work.
In a similar vein, Sepherosa Ziehau has committed the first stage of the first step of his parallelization of ipfw(4).
(Thanks to Sascha Wildner for the man page correction)
GPT partitioning is now supported, though Matthew Dillon’s post about it warns that it is very experimental. He also lists some interesting potential projects to go with it.
Matthew Dillon’s committed some initial support for streaming mirroring. With this, two disks can be synchronized over a network link of any speed or reliability – it can be restarted and immediately begin where it left off, and the amount of bandwidth used can be controlled. This sounds neat.
Sepherosa Ziehau’s committed his ETHER_INPUT2 networking upgrades, which moves networking a bit closer to getting out from under the Giant Lock.
Versions of DragonFly later than 2.0 now have bsdcpio, a BSD-licensed version of cpio, as the default version of cpio, instead of the GNU-licensed one. Thanks to Peter Avalos for adding it.
Matthew Dillon made some last-minute changes to Hammer mirroring;Â it’s made the options a lot simpler.
Matthew Dillon posted a July 16th Hammer update where he details causing a lot of write activity on a USB-connected, Hammer-formatted hard drive, and then yanking the USB connector out. Apparently, doing that 50 times over didn’t even faze Hammer. (Of course, be careful trying that with power.) He’s been committing a lot for Hammer, along with Sascha Wildner and Thoman Nikolajsen. A side benefit is that the Hammer work has exposed some issues in CAM.
Bonus link: Matthew Dillon talks about ‘purposeful destabilization‘, and man pages for hammer(8) and mount_hammer(8) are now available online.
Matthew Dillon’s latest commit of mirroring for Hammer has some details on how it works, for the curious.