hrm... what is the copyright on github.com/OpenIndiana/openindiana-completions/tree/master ?

anyone know offhand

there's notices (including some joyent ones), but no actual license

probaly public domain

i mean, those files are not really .. sufficiently clever

*that said*, some weirdo lawyers (the same weirdos who think the CDDL was "incompatible" with the GPL) tend to claim that things written as plug-ins for GPL code are automatically GPL'ed as well, and that would probably also cover these bash completions

(cf. wordpress plugins which are required to be GPL'ed. which is why i stopped writing them.)

and if that's the common opinion, that repository actually violates the GPL :p

2023-11-09T04:28:36.282012+00:00 bob tpm: [ID 992232 kern.info] NOTICE: tpm0: TP

M version is 2.0

woo

(though probably shouldn't be a notice)

jbk what is the status with zfs crypto?

[illumos-gate] 15990 SMB: Previous Session Logoff can prevent new connections from being established -- Gordon Ross <gwr⊙rc>

tsoome_: we still have the bug were we ara not compatible with upstream zfs crypto right?

at least last time i checked like a yearish ago

I do think so, yes

[illumos-gate] 16017 various small issues in lmrc -- Hans Rosenfeld <rosenfeld⊙gho>

newer versions of percona look for tcp_info, for now I'm simply removing that code on us as it's only used for some statistics

jperkin yes, even today they are adding more fields to it lwn.net/Articles/944663 , now I'm just trying to match those fields to our fields, I also add it to lx brand, I'll submit it for code review next week

tsoome_: what do you mean?

there was some compatibility issue, wasnt it?

with openzfs, yeah.. i just need to find some time to finish wiring up the property

hrm.. i don't suppose there's a way to override the bootfs zpool property?

updating the bootloader using piadm seems to have corrupted the value and now the system panics because vfs_mountroot is failing

... and of course my mbp seems to have stopped recognizing usb thumb drives (hopefully just a reboot will reset that)

teach me to dare ever upgrade anything :P

.. and it's not recognizing the USB

hrm.. the usb image uncompressed was 205mb.. that seems off

reminds me - have the ZFS implementations in Solaris and illumos always been compatible withe each other? I have an old system that I think is running Solaris 11 from nearly a decade ago. Might be nice to migrate it if possible (or update it if not). Assuming I can find a place for it where I don't mind the noise (and the heat, but winter is here in NY)

(well, nearly a decade)

jbk if you do get the prompt, you can set currdev

what is the format of that? the current value is 'zfs:zones/boot:' -- the last : makes me thing something else should be there

zfs:pool/dataset/.../dataset:

the last colon is denoting the end of the device descriptor

you can always test with 'ls zfs:pool/dataset/.../dataset:'

but that is assuming the pool itself is accessible

the pool seems to be ok

the problem is that the bootfs zpool property appears to be bad

and for whatever reason.. the smartos USB images aren't recognized as bootable on my system

(though they had been in the past)

or at least the value loader is passing to the kernel

loader itself is able to read the files from the pool

zfs_parse_bootfs is failing

what does lsdev -v tell?

ah, zfs_parse_bootfs() as from kernel....

hrm.. stepping through it, it's returning EINVAL on the pool name

hrm.. lsdev -v shows the pool

but 'hole-2' for where i'd expect the l2arc device

... i wonder if when i used piadm to update the boot loader if it clobbered the l2arc device

what does lsdev -v tell for bootfs name?

ok

so the smartos usb image isn't liked by my server, but omnios is

hrm.. so actually it's something with the loader image from the recent smartos

i booted off teh omnios image and was able to import the pool

there's a symlink that points to the directory with all the various loader bits

i changed it to point to the old one and it came up...

something to dig into more later

also might be worth comparing how omnios and smartos are building their usb images

ok, zfs-bootfs env variable (or boot property for kernel) is built as "poolname/objnum", where objnum is object number of the rootfs dataset. so you can check if the objnum does match the dataset... but if you did see EINVAL on pool name, what was the value of bpath in zfs_parse_bootfs() then?

'zones'

without dataset part?

also if you have mdb prompt still, does ::prtconf -v show the same value for zfs-bootfs property?

it is a bit curious because the zfs_bootfs() in loader code does use explicit "%s/%" PRIu64 format string

and in fact, we set both env variable and kernel command line, so boot -B prom_debug=true should also list the value passed with command line in debug output.

(we probably could stop passing command line now with loader;)

aren't we still using some legacy stuff for the handoff between loader and the kernel?

current kernel does process the env module from loader

for few years now

I left the commandline buildup to be present there for case if we boot older kernel, which does not process env module

[illumos-gate] 15759 ipmgmtd: the comparison will always evaluate as 'true' -- Toomas Soome <tsoome⊙mc>

if someone has time this change could use more reviewers code.illumos.org/c/illumos-gate/+/3095/14..15

→ CODE REVIEW 3095: 14525 Would like kmdb module for vmm (NEW) | illumos.org/issues/14525

@neirac --> honestly, now that you have Jordan too, you might wanna send it back to RTI-land if nobody else bites.

danmcd thanks, I'll do that.

(Follow her advice on testing notes too!)