I'm not sure if this is FreeBSD specifically, but maybe someone here can help. I have a machine with ZFS, and it's been a backup target for many years. I forget or never really knew exactly how all the files came into place (mostly a mix of rsync, and perhaps some zfs send|recv when migrating to new HW). Anyway, many of the older directories are prone to "Invalid argument" in many cases.

In particular, the problem seems to revolve around facls. For example, with some directories, I can `ls -l` it, but I'll get "ls: path/to/file: Invalid argument". Not fetching any acls with plain `ls` works. I cannot `getfacl`: "getfacl: www: Invalid argument", cannot setfacl. But I can `stat`. Any ideas about how to fix these directories?

what's "zfs get aclmode <volume>" set to?

rtprio - passthrough

i wondre if setting it to discard would 'fix' it

can give it a shot. I'm actually in the progress of trying to migrate this data to a new machine, so anything temporary is ok by me

i think discard is the default, maybe it was set from a zfs recv

ah. I was under the impression that only impacts new files, but I might be confusing that with something else.

Interestingly though, everything from the last 4 years or so in the same dataset is fine, and it only affects some of the older directories :shrugs:

rtprio - no go on the "discard" attempt though :-(

youre right, it probably does

there are other acl flags you can twiddle with

hey whats the freebsd developer channel?

nevermind, #freebsd-dev

kevans: actually, it's already been reported here: freebsd/atf #26

Title: execute_with_shell() failed on FreeBSD/MIPS with GCC 4.2.1 · Issue #26 · freebsd/atf · GitHub

26 – uucpd has wrong _PATH_UUCICO bugs.freebsd.org/bugzilla/show_bug.cgi?id=26

the temporary string returned by atf::env::get needs to be saved to a local variable to be used later

oh right, because it otherwise goes out of scope and the c_str() result is invalidated

so is the ATF_SHELL default being substituted in there correctly, or no?

apteryx: bah, I thought someone had pushed our contrib/ changes back into the github repo, but apparently not

br@ (reporter of that issue, even) fixed this locally: cgit.freebsd.org/src/commit/?id=c5072d5c94034

Title: src - FreeBSD source tree

I'll get it pushed to freebsd/atf and see what else might be missing

the diff is actually larger than I expected: 41 files changed, 46 insertions(+), 32 deletions(-)

mostly small things

OK, I had a very small fix (the same without the typo) coming as PR in case that's needed

tested working

apteryx: this is really nitty, but can you move the `shell` declaration/definition up one line, please? I'll merge it and drop the patch in this branch that does the same

oh, bah, I don't think I was given access to this version of the atf repo

kevans: I can do this yes

apteryx: thanks; I'm poking around to try and get access, but we'll get it taken care of asap

is this fine? paste.debian.net/1305702

Title: debian Pastezone

pkgconf question: does it behave the same as pkg-config in cross-compilation contexts?

apteryx: yes, perfect, thank you

pr refreshed

Good morning, guys.

Morning

I wish you fast progress for today, guys. FreeBSD 14 works good so far on my tiny virtual private server, running VMWare ESXi at my providers location.

Hello

Has anyone got successful experience of using net-im/py39-matrix-synapse port?

I had very strange trouble on new operating system

V-T60: successful experience in which way?

after transfer. First the federation didn't work at all, as i understand. It was giving errors about SSL

than i reached such a state that the messages via federation were successfully transmitted

but nothing received back

i just updated some dependencies with pip

Probably that how it would work? matrix-org.github.io/synapse/latest…alling-as-a-python-module-from-pypi

Title: Installation - Synapse

you probably should uninstall the port and install the lib through pip then

Installing as a Python module from PyPI

yes

i did mkdir -p ~/synapse, virtualenv -p python3 ~/synapse/env, source ~/synapse/env/bin/activate, pip install --upgrade pip, pip install --upgrade setuptools

and now pip install matrix-synapse is getting completed

how will i use service synapse start/stop/enable though?

the daemon will be unoperable (it will even get removed with the port)

awooo

howling to the moon, Amy? :P

I can't remove my jail

root@p5kpl-am:/usr/local/jails/containers/classic/dev # ls

fd pts

these two are anyway present

# rm -rf fd/

rm: fd/: Operation not supported

chflags -R noschg fd/

then remove it

# chflags -R noschg dev/

root@p5kpl-am:/usr/local/jails/containers/classic # rm -r dev/

rm: dev/fd: Operation not supported

rm: dev/pts: Operation not supported

rm: dev/: Device busy

isn't that the same from documentation? # chflags -R 0 /usr/local/jails/classic

which didn't work

Okay, well I didn't know in the advance because you didn't explain what you already did. It sounds like you need to umount those directories.

Please be more specific about your problems too, we can't read your mind.

oh

devfs on /usr/local/jails/containers/classic/dev (devfs)

when did i mount it?

don't remember at all

I am trying to get my ThinkPad X280 to work with HDMI. It works if I boot with the cable in. But not if I plug it in after boot. Is there any "magic trick"?

i've just spinned up my brand new thick jail environment

then looked inside of the /var/log

ld-elf.so.1: Shared object "libssl.so.30" not found, required by "sendmail"

what does this error mean? do i have to worry about it?

V-T60: do you have /usr/lib/libssl.so.30? this should be in base, so it seems like something is wrong with the install (assuming this is base sendmail, not ports)

i do have on host

# ls -al /usr/lib/libssl.so.30

-r--r--r-- 1 root wheel 692208 Nov 10 10:23 /usr/lib/libssl.so.30

in the jail though?

# cat /usr/lib/libssl.so

/usr/lib/libssl.so /usr/lib/libssl.so.111

only /usr/lib/libssl.so and /usr/lib/libssl.so.111

no /usr/lib/libssl.so.30 though

did you install a different OS version in the jail than the host?

seems that no

# uname -a

FreeBSD matrix-synapse 14.0-RELEASE FreeBSD 14.0-RELEASE #0 releng/14.0-n265380-f9716eee8ab4: Fri Nov 10 05:57:23 UTC 2023 root⊙rnfo:/usr/obj/usr/src/amd64.amd64/sys/GENERIC amd64

root@p5kpl-am:/var/log # uname -a

FreeBSD p5kpl-am 14.0-RELEASE FreeBSD 14.0-RELEASE #0 releng/14.0-n265380-f9716eee8ab4: Fri Nov 10 05:57:23 UTC 2023 root⊙rnfo:/usr/obj/usr/src/amd64.amd64/sys/GENERIC amd64

looks more like totally same version

what does 'file /bin/ls' say in the jail?

/bin/ls: ELF 64-bit LSB pie executable, x86-64, version 1 (FreeBSD), dynamically linked, interpreter /libexec/ld-elf.so.1, for FreeBSD 13.2, FreeBSD-style, stripped

host: # file /bin/ls

/bin/ls: ELF 64-bit LSB pie executable, x86-64, version 1 (FreeBSD), dynamically linked, interpreter /libexec/ld-elf.so.1, for FreeBSD 14.0 (1400097), FreeBSD-style, stripped

did you install jail with bsdinstall, after upgrading host from 13 to 14? in this cases you also need to update files in ... uh, i think it's /usr/freebsd-dist, which is what bsdinstall uses

so the output is different

i did clean install of FreeBSD 14 yesterday in the evening

there was no 13.2

though, i followed this guide docs.freebsd.org/en/books/handbook/jails/#classic-jail

Title: Chapter 17. Jails and Containers | FreeBSD Documentation Portal

odd, but it seems like you ended up with 13 in the jail somehow anyway

ah... did you edit the url in that fetch command to be 14.0 instead of 13.2?

they recommend this way fetch download.freebsd.org/ftp/releases/amd64/amd64/13.2-RELEASE/base.txz -o /usr/local/jails/media/13.2-RELEASE-base.txz

yesterday i did, but today i didn't

(fwiw, i don't understand why this breaks your sendmail, but i assume it's not wha you want)

and it anyway updated later

in the next command it got patched

so you updated the jail to 14 after installing it?

With the files copied, the next thing to do is update to the latest patch level by executing the following command:

 

# freebsd-update -b /usr/local/jails/containers/classic/ fetch install

^ this command somehow took 13.2 and did 14 from it

(instead of classic i just wrote my matrix-synapse, and that's all)

that's odd, fetch + install shouldn't do a release upgrade, just upgrade to latest 13.2-pX

strange, should i substitute to 14 in url?

so something went wrong

i should start over

V-T60: just out of interest can you check 'file /usr/sbin/sendmail' (or it might be /usr/libexec/sendmail/sendmail) to see if that's the 13.x or 14.x version?

# file /usr/libexec/sendmail/sendmail

/usr/libexec/sendmail/sendmail: setgid ELF 64-bit LSB pie executable, x86-64, version 1 (FreeBSD), dynamically linked, interpreter /libexec/ld-elf.so.1, for FreeBSD 14.0 (1400097), FreeBSD-style, stripped

14.0 i guess

both in host and thick jail

that's odd, it seems like jail ended up half-upgraded somehow. i would suggest deleting it and start with 14.0 when creating the jail, but i don't know how it ended up like that

immutable files

(premium footgun)

babz: but wouldn't it still have installed the new /usr/libssl.so.30 then? that didn't exist in 13 so it couldn't be immutable

/usr/lib/libssl i mean

hm... isn't there some weird edge case / bug where freebsd-update can install updates from the host into the jail? maybe sendmail was updated on the host... i can't remember where i saw that now

ah i was thinking of bugs.freebsd.org/bugzilla/show_bug.cgi?id=235318 but that's something else, probably not related

Title: 235318 – when updating jails using -b, freebsd-update is reusing host's /var/db/freebsd-update without indication of doing so, unless explicit -d as well

I fixed the issue with mentioned earlier wya

way

jbo: i can't assign sublime-music PR back to you anymore (since it was assigned away from me by triage) but based on vvd@'s comment i hope this is ready for merging now

postgres:\ :lang=en_US.UTF-8:\ :setenv=LC_COLLATE=C:\ :tc=default:

i will be creating such: createdb --encoding=UTF8 --locale=C --template=template0 --owner=synapse_user synapse

what do i have to substitute in my system for such command?

i know that it required me to set up the lang to C

V-T60: i think what you have is correct for UTF-8 encoding / C collation. run initdb and try to start postgres, if the locale is wrong, it will produce an error

remember it from before

okay

--locale is equal to collation?

--locale sets all the various --lc-* options including --lc-collate

i think --lc-collate is the only one that has to match at both initdb and runtime though because it affects how indexes are created

can sysvmsg=new sysvsem=new sysvshm=new harm anything to me in /etc/jail.conf?

running bootstrap script ... 2024-01-30 17:29:47.402 EET [46210] FATAL: could not create shared memory segment: Function not implemented

that's what i have now

won't that break another features?

V-T60: that should be fine, i think it's only not the default because it's a fairly new feature

used to be you had to let the jail access the global shm which is obviously much worse

debdrup: finally got my pc migrated, running my first buildworld on it, can't get cpu temp > 86˚C so far

dch: you're doing well, i can't get my CPU to go above ~62C in buildworld or poudriere. but i think this is because the AMD X3D CPUs just run cooler

lw do you have that fancy AIO cooler stuff? I am using humble fans for the moment

it's not in a case yet, just sitting on my desk.

dch: no, just a (large) fan

I'd move to an AIO thing if it turns out that the pc is noisy overnight, but atm it seems pretty quiet

maybe i build my own case out of wood, all I really need is power buttons and a few more USB slots on the front

lw, sorry mate, I had some terrible weeks. won't be able to jump on this before weekend

jbo: no rush at all, just wanted to mention it

lw, lol, I just saw that mark was not happy about me assigning the PR to you - I didn't know that :s

jbo: i found it a bit odd but i guess it sort of makes sense if it's policy? ... but like, i could have done something about it: i could have assigned it back to you :-P

i guess maybe the idea is that you personally don't "own" the bug and perhaps someone else could have picked it up in the mean time? although that would be odd/annoying

(i mean, annoying for you mostly since you put all that effort into this then someone comes and take the bug out from under you)

they are welcome to do that - plenty of work for everybody :)

lw, if you want to help me out: I wanted to do this myself (because it's a minor change and easy to be handled by the committer) but you could address the requested change here: reviews.freebsd.org/D43477

Title: ⚙ D43477 multimedia/mediaelch: New port

ah, i didn't know about that syntax, i can fix that

please be sure not to change _anything_ else in the patch if possible

jbo: do you want me to put my branch on github or something (or even make a phab review for it) rather than submitting patches in bugzilla? i don't mind either way, just wondering if this might be easier

lw, nono, bugs reports via bugzilla are the way to go.

ok

lw, typically I would have changed this on-the-fly without you knowing. I just was totally buried the last two weeks

if you can wait a few more days you can also leave it as is - makes not much more work for me

just in case you're bored :p

i have nothing particular to do right now so it's np

at least i have a proper git worktree setup now + a custom testport workflow which makes all this much easier

no more messing with poudriere, just 'sudo testport' and it does testport, bulk, and sets up a pkg repository for the branch

jup, your PRs will certainly get through faster if you properly testported it and everything

note that committers don't do runtime tests. that is on the maintainer's plate. we only do build tests.

yeah, i use mediaelch all the time so i'll definitely notice if it's broken

jbo: i was testporting it before, it just took so long to commit, update poudriere branch, testport, install, test... now one command does it all

\

/

does FreeBSD have a Video4Linux compatible API for using video en/decoding hardware on GPUs?

or can DRI do this directly

hm, multimedia/ffmpeg has a V4L option, so there seems to be video4linux stuff on freebsd, but no idea what it does

See v4l2 in ports this is how the webcam driver works for example

can’t help much more than that sorry

Anyone knows if I can make if_bridge support more than 1Gbps with a bhyve VM? the connection I'm interested in is between the Host and the Guest.

antranigv, dont know but why not try?

orbifx I tried, the most I can squeez is 1.8Gbps, I'd like to get much, much more.

antranigv: that seems… bad

what's your config look like? how are you measuring this?

have you tried alternatives like vswitch?

antranigv: Have seen this: forums.freebsd.org/threads/bhyve-vm…penbsd-3x-slower-than-freebsd.92085

Title: bhyve - Bhyve VMs network speed test: OpenBSD 3x slower than FreeBSD? | The FreeBSD Forums

Yours similar results?

meena wait, openvswitch works on FreeBSD?

hrmm… freshports.org/net/openvswitch Last update has been a while

Title: FreshPorts -- net/openvswitch: Production quality, multilayer virtual switch

that's not even the latest 2.x release

tercaL very much so. but I haven't booted a FreeBSD guest yet.

meena I can try compiling manually, that's fine.

meena do you have a setup guide for OpenVSwitch?

antranigv: nope, I've only ever used it indirectly, so, no

welp, it begins: unxz -c Downloads/FreeBSD-14.0-STABLE-amd64-20240125-3f2c2dda2631-266523-memstick.img.xz | dd of=/dev/sdc bs=1M

still not sure if I will end up going with FreeBSD on this ThinkPad long-term, but I've gotta at least see how things are working in order to make an informed decision

That is always the best way, to try it out and see what happens

which thinkpad are you setting it up on?

it's a first-gen E14

one with an i3-10110U

mainly concerned about the wireless chip, which uses the apparently very much beta-quality rtw88 driver

wifi has always been pain for me on fbsd

tm512 as a last resort, you can use a WiFI dongle OR a Linux VM with bridging or something

but yes, the only way for me to use FreeBSD is with Ethernet.

I'm also a little bit concerned about sleep states, which on FreeBSD has driven me to OpenBSD in the past

this was like FreeBSD 9.x though

antranigv: I'm aware of wifibox, which I might end up going for. bit annoying, but it sounds like it works fine

tm512 I've been pretty lucky with Sleep states! all my FreeBSD support never had any issue.

FreeBSD laptops*

I suspect I won't have to rely on wifibox permanently, since this wifi chip does have a driver in the kernel already, just sounds like it needs more time in the oven

for some reason this laptop kinda smells like mushrooms. really curious how it was stored

hoping it airs out sooner rather than later

nice, rtw88 loads, it also appears like the bluetooth radio on this realtek chip attaches as ubt0

bluetooth radio connects separately from the wifi (USB vs PCIe) so even if I have to resort to passing through the wifi to a Linux VM, seems like FreeBSD will still have unrestricted access to bluetooth

make: "/usr/ports/Mk/bsd.port.mk" line 1204: UNAME_r (14.0-RELEASE-p3) and OSVERSION (1302001) do not agree on major version number.

I might be solving this issue so ignore the question for now

ok problem is persisting

does anyone here get some segfaults with these? paste.mozilla.org/preuMHCB paste.mozilla.org/MJB1rc32

Title: Mozilla Community Pastebin/preuMHCB (C)

Users can set CRYPTOGRAPHY_OPENSSL_NO_LEGACY environment variable

how do I set the environment variable on cryptography open ssl?

looking through dmesg from the live USB, and noticed a couple things that I'm a bit concerned about. only the touchpad (not the trackpoint) seems to be detected, and for the touchpad it uses the psm driver which it says is giant locked and might be removed in FreeBSD 15

doesn't the giant lock basically remove all concurrency in the kernel while it's active?

the keyboard seems to be giant locked as well