[illumos-gate] 16349 SHA2Update length is now of type size_t, not unsigned int -- Bill Sommerfeld <sommerfeld⊙ho>

I'm trying to connect to wifi on Tribblix. I've run '# dladm cteate-secobj -c wpa mykey' followed by '# dladm connect-wifi -k mykey -e <my ESSID> wpi0

Still not connected

I have to admit to never having used wifi on Tribblix, or indeed any illumos distribution

Simply due to not having any appropriate hardware, rather than for any other reason

But I'll assume that you've installed the wifi overlay, to get the drivers and the wpa service

hello all, a question regarding to ZFS, now ZFS on illumos/smartos does not support dRAID, but openzfs supports that, is there any reason to not support? or had plan to migrate from Linux/FreeBSD into illumos kernel, is there anyone know the plan or status? I belive dRAID is very useful case when there is a JBOD with 60*22T disk, to replace failured disk is much quick. a big thanks

tozhu: short answer: Nobody's done the work yet - if someone stepped forward to do the port (and it was done competently) I'm certain it would be accepted.

Hey is this project still maintained

yes

Wait so you guys aram maintaining joynet?

Are*

do you mean joyent?

Yeah sorry iOS only has one irc client and it's garbage. I was just curious because I saw Firefox esr and many other things are still rocking older versions

sommerfeld: If you're talking about Triton (SDC) and/or SmartOS, that's now maintained by MNX, but yes, it's very much alive.

illumos is the kernel and userland base used by SmartOS as well as other operating systems (OpenIndiana, OmniOS, Helios, etc)

tubatom32: openindiana is shipping firefox 123.0 right now.

No I was more talking about the netbsd tree

pkgsrc?

yes I still maintain that, pkgsrc itself contains recent versions of firefox but it does not build without patches and I focus more on server software as the vast majority of our users don't run desktop software

Ahhh ok no I prefer running off the joynet pkgsrc tree. I've been using it for a while and just wanted to make it hasn't been discontinued. Thank youuu for everything you do :)

you're welcome! and yeh as others have said joyent have nothing to with open source these days, the binary package repository is still maintained, now by MNX (who I work for), and of course the majority of pkgsrc work is done by the netbsd volunteers.

tsoome_: does loader always use the EFI serialio protocol for talking to the serial ports when booted under EFI, or does it try to access ports 3f8/2f8/... directly?

it is currently using ports directly

aha!

i see efiserialio.c is in loader, is there a way to make it use that?

I'm trying to get illumos booting in azure as a gen2 vm... and it looks a lot like I'm hitting this: cgit.freebsd.org/src/commit/?id=927…8dd98cb902160093e0dc0bac002d6b43858

i turned off the countdown timer, so it should be sitting at the loader prompt (and the serial log suggests as much), but the serial console is otherwise not working

it has been in my queue for long time to fix it, but I think you could just replace comconsole by efiserialio, but there is a trap with hyperv

hyperv does not like the port speed to be changed, it will hung the vm.

illumos gen1 support works, but your boot disk _has_ to be IDE (HV imposed requirement), and our IDE driver is PIO (aka sloooooooooooooooooooow)

and unfortunately there is none to care about IDE:D

it also means it absolutely cannot handle any sort of serious i/o loads

really bad when you're assumption is that your boot disk is an SSD :)

hm, the hyperv hung was probably just about ARM64, however

i do see in that patch that it appears to check if it's hyper-v and just doesn't set anything

yep.

the efiserialio is waiting for few prerequisites to get done....

stupid vmware and caps lock bug

is that it?

na, its just biting me every now and then

if you can have your own build, efiserialio can just be drop in replacement, but to integrate it, a bit of work is needed.

i guess i'm trying to understand what the downsides would be for that on systems using EFI

or what the problems would be...

biggest issue is problem about identifying ports.

but otherwise, why it has been delayed is about some other patches to clean up the early startup in loader main. there has always been something more urgent to do:D

what's the issue with identifying ports?

on a vaguely related note, I've been seeing poor serial console response to loader on bhyve (host & guest both illumos); seems like most input gets lost but if I keep typing the same key over and over it gets through. not a problem once illumos is booted. any thoughts on where to look first?

jbk serialio does give you an array of port handles, but no names attached.

jbk so you have no option to query the name other than dig with device path protocol to see if you have acpi device path etc.

sommerfeld efi boot?

tsoome_: how do I tell which one the guest is using?

if os is running, prtconf -v -- does it have biosdev-0x82 etc (bios disk data), or efi-version property

prtconf -v | grep bios only prints a line containing "smbios-address"; prtconf -v | grep efi shows an efi-version

so its efi boot.

so would the efiserialio thing be likely to work better for bhyve, too?

tsoome_: is 16344 (fenix?) something on your radar with all your other cleanup?

the ed2k variants of uefi implementations eat up serial keystrokes (from serialio), when you have both "keyboard" and serial ports listed in ConIn. I did work around in my own instance of efiserialio by checking if this port is used by ConIn and did use efi_readkey() instead.

a bit hack, but seems to do...

richlowe indirectly because more recent compilers hate us for not having prototypes:)

richlowe but I hope some more people will join for those cleanups:D

well, should I look into that, or do you have it already?

no, i have not

ok

so feel free to pick it up

tsoome_: it looks like the efiserialio.c that's there uses the ACPI UID of the device as the index (0 -> ttya, 1 -> ttyb, ...) am I missing something, or is that just broken?

wait, does EFI _tell you_ where smbios is?

that, at least, makes me happy.

the canonical way to find one is... uninspiring.

it should work, the trick question is, does UID start from 0 or from 1:)

richlowe yes:)

jbk I can look on it more tomorrow, its almost midnight here and I have early start ...