This will matter most to you if your connection to the Internet is poor: fetch(1) now will time out on data transfers too.
Welcome Sergey Zigachev, new DragonFly committer who has already committed an amdgpu fix.
If you pay attention to your daily security run emails on your DragonFly system, you may see entries like this:
+nlookup() at nlookup+0x623 0xffffffff806eff13 +kern_rmdir() at kern_rmdir+0x25 0xffffffff80706db5 +sys_rmdir() at sys_rmdir+0x4c 0xffffffff80706f0c +syscall2() at syscall2+0x11e 0xffffffff80bd9f9e
I see it on the machine where I run this Digest, as the caching mechanism adds and deletes files rapidly. Matthew Dillon has placed it behind a sysctl, so your messages log will be a little less noisy by default.
makefs(8) now supports HAMMER2 on DragonFly, so you can create HAMMER2 file system images, same as a CD image or a DOS disk.
mlockall(2) in DragonFly has been revamped for compatibility with other implementations. This should have no obvious end user impact, other than a bit easier to port stuff. I want to mention it to note the work done.
You can now set a description for a network interface on DragonFly. Don’t use ETH0, please.
If you are trying to use both NAT and IPv6 with pf on DragonFly, there was a bug (seen here with FreeBSD) with :0 where it would use link-local addresses. It’s now fixed.
If you are using ‘set skip on …’ in your pf config, it used to match any interface that matched the specified type. It now only matches members of that named group. That may change behavior of your pf rules; check the commit to see what to look for.
There’s a new sysctl(8) setting, sysctl.debug, which shows you which sysctl nodes are being requested. I am entertained by the pseudo-recursive style of my explanation.
There’s some bugfixes for HAMMER2 and the kernel that will probably mean a point release soon.
I think I know what Aaron Li might want to work on for DragonFly…
(I am only guessing; I have not asked.)
It’s apparently possible to get a panic by yanking a HAMMER2 disk out of your system, which is only likely when using a USB thumb drive, formatting it to HAMMER2, and not bothering to unmount it. Anyway, that poorly-described-by-me problem is fixed.
There’s an odd bug in ipfw that is now fixed in DragonFly 6.2/6.3. If you are using ipfw and adding networks and hosts in a specific order, the netmask will be set wrong.
There’s also a problem with the overnight bulkfree cleanup in Hammer that’s had various attempts to fix it over time – it’s now really truly fixed. It mattered only if you had an extremely large number of inodes – 100000000 or so,
Matthew Dillon wrote up an explanation for both.
I use date(1) just rarely enough that I can never remember the right arguments to create a human-readable result. Now, there’s an -I arg to date(1) that uses a word instead of a format string to get ISO8601 output.
I realize my title is a little bit buffalo buffalo buffalo, but it makes sense: getopt(3) now has a double colon option to indicate an optional argument. I link to it because I like seeing the length of the trip to DragonFly. It started as a GNU option, then showed up in NetBSD, brought to FreeBSD, and now I’m posting about it.
It may be because I am a nerd but I enjoy reading detailed explanations of bugfixes like this one for HAMMER2. This fix is present in the 6.2 release, of course.
I tagged DragonFly 6.2, and I’m planning for release later this week. Release notes and ISO/IMGs to come with the release, as usual.
You didn’t need it or use it, but the name itself has a certain symbolism.
Tomohiro Kusumi has removed the old GNU implementation of ext2fs from DragonFly and added the non-GPL-encumbered FreeBSD implementation of ext2fs.