have 2 virtual and 1 physical hosts running on gcc-14 [primary] built illumos:)

tsoome_ - nice :) - I'm opening a gcc14 lgrp cleanup change now, which is the last one I know of. Looks like we can shortly start using 14 as a shadow. Thanks for all of your work on getting us to here.

yw. we should get those os_tests fixed tho:D

ah and the smatch fix is still waiting in RTI queue - but it is biting only if tools are built with gcc14:)

and 8 epic fails

is a pseudo-terminal a valid uart backend for bhyve ? I'm getting this Unable to initialize backend '/dev/pts/3' for pci uart trying to pass -s 7,uart,/dev/pts/3 to a bhyve vm. According to the message seems so. I'm doing tcp to serial so I created a pseudo-terminal to receive data from the tcp connection and write to that port so I could share with the vm.

man is telling: /dev/xxx Use the host TTY device for serial port I/O.

also it has some examples

illumos could have freebsd-style nmdm (i've tried to make it but my kernel-fu is too low)

tsoome so that means I did not need to use the uart backend?, the section of the manpage you are refering to is -l lpcdev ? I tried -l com4./dev/pts/3 but it still the vm does not boots

is it in zone config?

that is, bhyve in zone?

tsoome yes, I'm using bhyve_extra_opts in smartos to add that line

if so, it will complicate things because the device files are from zone;)

tsoome thanks!, I understood now

but you can use pipe/socket instead and socat

[illumos-gate] 16585 ping ignores ipv6 scope id in target address -- Bill Sommerfeld <sommerfeld⊙ho>

can someone take a look at my RTI for illumos.org/issues/16370 ? (Dan did the code review so he can't approve)

→ BUG 16370: ipclassifier should pay more attention to sin6_scope_id (Pending RTI) | code.illumos.org/c/illumos-gate/+/3355

[illumos-gate] 16370 ipclassifier should pay more attention to sin6_scope_id -- Bill Sommerfeld <sommerfeld⊙ho>

[illumos-gate] 16868 Typos in pthread_rwlock_timedwrlock(3C) -- Andy Fiddaman <illumos⊙fn>

[illumos-gate] 16869 auditreduce: dangling pointer error -- Andy Fiddaman <illumos⊙fn>

that auditreduce one was wild

yeah, if it mattered, that whole program would benefit from a rewrite.

we have had moderately informal chats about a new binary audit format that was less troublesome, too.

it would benefit from being someone who uses audit heavily who did these things, probably, though.

It's been a while, but we used auditing and did quite a bit of analysis of the records

The first thing we did though was use praudit and awk to get the records into a form suitable for ingestion

From that point of view we didn't care what the file format or tools looked like, because all the smarts was done elsewhere

ptribble: you'd presumably have cared about ingestion performance

Yeah, so there were a couple of fixes I put in for praudit - #9106 and #9830

richlowe: my take these days is that the best answer for a lot of things like that is "a thin layer over sqlite3".

The slow part was praudit translating things it stored as numbers into the relevant names

any kind core member to check my RTI of 16866 ?:)

thanks!

[illumos-gate] 16866 smatch: passing argument from incompatible pointer type -- Toomas Soome <tsoome⊙mc>

I hate how much of a wrench that processing with GPUs has thrown into virtualization