Thats step one I guess. I eventuelly also need to get things into the illumos buildsystem and not hayashis

but modifying hayashis system first is a good idea

I am having trouble with ipv6 again, at least it's not a kernel panic this time :)

I can't recall the IPv6 panic... did we fix it (if so, cool, if not do you have a vmdump.N file you can share?)?

As for now, what's happening? (Pardon latency & jitter 'cause it's a release-week-Wed at month's beginning so it's a chock-full-of-meetings day).

oh, the ipv6 panic was fixed within the day and I almost learned the kernel debugger in the process but it's been months so I've mostly forgotten :)

I can't get ping to work to an ipv6 address and it looks like my routing table is the problem

wgetpoaste

oops]

ifconfig -a

sigh]

rge0:3 has my currently valid ipv6 addr

I'm betting it will work if i reboot and don't have yesterday's addresses still in the kernel

rge0:3 has a different /64 prefix so it should work, but I can imagine source-address selection takes it rge0:2 first.

Try this:

ifconfig rge0:2 down

That'll take that IPv6 offline.

It'll save you a reboot until next reboot.

If you snoop rge0 for IPv6 pings, you can confirm a bad source address selection on those default-route pings (if you haven't already down-ed rge0:2).

[illumos-gate] 14928 pcieadm rc event collector association decoding -- Robert Mustacchi <rm⊙fo>

thanks, I had already started a reboot and that fixed it :) next time I know to try taking the extra interface down first though, tyvm

it looks like this could turn into an ongoing problem since my ipv6 changes on a near daily basis when i get a new ipv4

i might have to write a script

I do build Illumos on that machine but I also serve some personal web content there for clients to look at.

So, I'm actually working on getting both of my domains to show up on my dyndns provider with both ipv4 and ipv6 addresses

[illumos-gate] 14951 fix ::sysregs under mdb-bhyve on AMD -- Patrick Mooney <pmooney⊙pc>

[illumos-gate] 14901 remove remaining a.out support from sgs -- Richard Lowe <richlowe⊙rn>

Ok, now I have a gnu Makefile question. Does anybody know if... (full message at libera.ems.host/_matrix/media/r0/do…badd3e5607df27f2cb379d09af66851e7f3)

modulo missing endif ofcourse

checkmake says yes gnu make says `make: Fatal error in reader: ../Makefile.master, line 38: Unexpected end of line seen`

I believe you just have 'ifeq' and have to check it against an empty string

but I don't know why I think that, what you want to do works for me

I don't know what you're doing -- or why you're doing it -- but if this is for ARM, it's almost certainly easier to port our make to your host machine than to port our build, properly, to gmake.

I know andyjs and some other people had some magic that ports our build improperly to gmake, too, which might help you in the very short term

but I believe joerg took sun make and did... joerg things to it, which should get you a long way.

richlowe: it's arm. So the situation right now is Hyashi has ported our whole build to gmake. And I want to build ontop of OI. I got our cross toolchain compiled and want to use that to compile Hyashis build system. It is rebased on latest illumos-gate as of 4 days ago but all gmake.

I figured I could get away by switching some paths to host tools which linux needs to compile and then use OI to build the existing build. but for whatever reason gmake on OI does not like that line.

richlowe: i think somehow make is being invoked but I haven't found out how, there is a variable $(MAKE) but that shows gmake in the line before...

ah found where make gets called

it's a hardcoded make in the same file as $(MAKE), well that solves the problem :)

now it's happy :)