If you upgrade pkg on your system, it may start erroring out. This is because the default config will confuse the newer version. To fix this, you can copy over a working config and the problem will go away. I expect this may only be a problem until the next release.
MAP_VPAGETABLE has been removed in DragonFly because of conflicts with recent pmap work. This has the unfortunate effect of breaking vkernel(7), but vkernels can be resurrected with changes to use hardware virtualization support.
Note that running DragonFly as a VM is unaffected; that’ll still work just fine. This breakage is DragonFly-vkernel-on-DragonFly specific.
Aaron LI has ported timeout(1) to DragonFly. It’s a way to run a command with a time limit, and I’m happy to say it is a cross-BSD item, coming from NetBSD by way of FreeBSD.
I’m doing a catch-up post here to note all the smaller updates, some cross-BSD, that have gone into DragonFly in the last week or two: openresolv 3.12.0, dhcpcd 9.4.0, tzdata2020f, 802.11 channel definitions, stdbuf(1) and libstdbuf(3), sockaddr_snprintf(), and getaddrinfo(1).
You can’t tell directly from the commit message, but committing to DragonFly may trigger a reminder to MFC, based on commit message content. This is thanks to Aaron LI. It’s little, but this sort of automation is a good idea.
Aaron LI’s added a pw-update.sh script to DragonFly, for use in automating group and user changes, especially as – someday – part of a binary upgrade.
So, if you find yourself in possession of an ADM-3A terminal, and want to attach it to a DragonFly machine, here is the /etc/ttys config (viewed on the ADM-3A itself of course) and the front switch settings that worked for me.
Remember, ^h deletes.
Thanks to liweitianux, the mirrors page on the DragonFly site has been updated. Check again to see if there’s a mirror near you, if you haven’t looked recently.
If you remember HAMMER1’s ability to create a volume that spanned multiple local disks, that capability’s been introduced to HAMMER2. Look at the commit message to see how it works so far.
Note that this is not multi-master replication.
DRM in DragonFly has been updated to match Linux 4.15.18, along with recognizing some new hardware.
This won’t affect your day-to-day operation of DragonFly, but it’s interesting: apparently, uptime was always (now minus boot time). If you reset the clock on the machine, however, it would no longer be accurate. Now it is accurate, for a number of utilties.
I realized I never followed up on the call for testing: re(4) driver updates from Sepherosa Ziehau were indeed tested and found good, so the driver has been updated.
One last thing sneaking in for the week: There’s an update for libressl in DragonFly that fixes CVE-2020-1971. It’s there for 5.8 or -current.
The next release of DragonFly will be 6.0, mostly because 5.10 is an annoying version number rather than any significant version changes. We’re due to release by the biannual calendar schedule – but there’s a DRI bug that needs to be fixed; I plan to tag as soon as that’s done.
If you have a Realtek network card supported by the re(4) driver, Sepherosa Ziehau has a new driver for you to test.
Because of this commit that makes some changes to lib/stdio, you might get more reinstalls than you expect on your next pkg upgrade because of the __DragonFly_version change. This only applies to -current (5.9) users.
(I might be wrong)
If you want to build a kernel with no options, stripped down, here you go. I don’t know how useful it would be…
Matthew Dillon has made significant changes to the callout API in DragonFly. Interesting to look at, but I think no changes from a user point of view.
I’ve seen this multiple times over the years: if ifconfig suddenly stops working, especially after an upgrade. your kernel and world are out of sync. Rebuild and make sure you get both updated.
DragonFly has a new version of libressl, noting cause it has a newer TLS1.3 implementation – something that may be necessary for you.