andyf: dilos is steel active, but have no updates on site :) but has updated apt repo :)

vab: I have a T5120 that's a perfect candidate for OI testing. I thought SPARC support being killed in upstream illumos meant OI was also killing it, though

KungFuJesus: THanks to the effort of ptribble and klaus SParc is being kept alive in a fork. If you send me your mail, I'll gladly put you in contact with Klaus

awesome, my T4 is also EOL with Solaris upstream

I also have a sunfire v240 with a GPU that might make a viable X11 based testbed

I _think_ it's just the crappiest available common ATI rage card, so it's probably relegated to very unaccelerated fb only drivers, but...hey, it's something

Solaris still supports T4, it was T1-T3 that were EOL'ed

KungFuJesus: most of the sparc stuff is being done by Peter Tribble

you might want to look at tribblix/talk to ptribble when he's on IRC about it

alanc: Hmm, for some reason I remember the upgrade path past 11.3 being impossible

that's when T1, T2, & T3 were dropped

KungFuJesus Klaus has a V240 where he does a lot of dev work. Even gcc 11 is working. :-)

hmm, I swear pkg upgrade was failing for me for some reason

vab: does that mean he fixed the system call issues? I imagine Peter would be interested in that

in any case it would be nice to have an illumos distribution on those guys given that the support cycle is probably on the tail end, anyway

I might finally write some VIS3 accelerated routines for zlib-ng, hah

The V240 might also make for a good testbed for big endian fixes in mesa, too. That way I can decipher if it's just a nouveau issue or a core mesa problem

of course llvmpipe gets me a lot of that too, but it doesn't always exercise the same paths

pkg upgrade could fail for reasons other than CPU support

wait, like what?

I mean, obviously it could, but I thought if you were on the beaten path that should just always work

for y'all

like installing third party packages that depend on stuff we obsoleted, or playing around with version-lock facets or freezes

oh, ah.

or if you're not using the support repo, the 11.3 -> 11.4 upgrade will fail because the upgrade requires things that are only in 11.3 support repo be installed first

I thought you meant in the general scheme of things.

though that last line seems like just something you shouldn't have done

but I guess that depends on whether you convinced customers to bother doing upgrades, yet.

it's only been 15 years

many people argued for the "bunny hop" bits to be placed in the /release repo too but they were unsuccessful

(it's a new version of pkg that understands some new attributes used in the 11.4 packages, and some ON changes I forget)

it's definitely something I can't imagine shawn doing willingly

richlowe Sorry, been out of the loop for a while... what system call issues are there?

Currently the problem is that the T5440 I gave him will happily run with three CPU and memory boards.

Once he puts the last in, it fails.

→ BUG 2757: 32bit sparc direct system calls should not rely on compiler mannerisms (In Progress)

module /platform/SUNW,T5440/kernel/sparcv9/unix: text at [0x1000000,

0x10e14e5] data at 0x1800000

module /platform/SUNW,T5440/kernel/sparcv9/genunix: text at [0x10e1500,

0x13051bf] data at 0x1936ac0

module /platform/SUNW,T5440/kernel/misc/sparcv9/platmod: text at

[0x13051c0, 0x13051cf] data at 0x199ed60

module /platform/SUNW,T5440/kernel/cpu/sparcv9/sun4v: text at

[0x13051e0, 0x1307c4f] data at 0x199f300

OpenIndiana Hipster 2023.04 Version illumos-69019bc162 64-bit

Ethernet address = 0:14:4f:a0:81:1a

ERROR: Last Trap: Fast Data Access MMU Miss

I'll ask him. He's not usually in irc/libera.

[illumos-gate] 15254 %ymm registers not restored after signal handler -- Robert Mustacchi <rm⊙fo>

rmustacc: congrats, and thank you.

Thanks. Next we'll have to do the stack improvements, but this should help at least folks and reduce a bunch of gnarly corruption.

Woo!

I feel like there's a way to wait for an SMF service, manually, and I have forgotten what it is.

does that ring any bells for anyone?

the equivalent of pwait(1), but for a service FMRI to come online

like svcprop -w or whatever it is?

There's a syncronous start...

jlevon: that seems to be for properties only

nahamu: I am just wanting to wait for the system to calm down after booting, for eg, by waiting for multi-user-server to come online

richlowe: yeah but you watch for the state like svcadm -s does no?

remember ssh, console-login, etc, start before that.

jlevon: clever!

You may be able to just drop a svcadm enable -s as well?

hello, installed the latest OpenIndiana and stuck without internet. dhcpagent spams "select_best: no valid OFFER/BOOTP reply". meanwhile I check logs on my router and it sends offer response to pc, also I checked 67 and 68 ports on pc using snoop, and I can see, that some information(including interface hostname that I assigned) comes back to the

OpenIndiana pc. somehow it can't get IP address. Trying to fix this few hours in a row, any help is appreciated

any other warnings? github.com/illumos/illumos-gate/blo…-inet/sbin/dhcpagent/request.c#L455 is the code used to "score" an offer.

look at the options found in the OFFER

it will reject offers missing a lease time or a server id.

the message you quoted comes from select_best() in the same file; I suspect it's hitting the "compute_points_v4 returns -1" path.

so, I checked and server definitely provides with server id, and regarding lease time, it sends address-time (equals 600). when you mentioned time, I have wrong location/time set on pc, maybe it has something to do with offer expiration. let me try to fix time

ps: sever has lease time in settings, so I assume it sends it as well

best to actually look at what options appear on the wire in the OFFER message. a 600s lease is on the very short side. snoop -v will decode options.