<azonenberg>
the boards are not currently attached to the VM but i can hook them up if needed
<whitequark>
not logged in; doing some preparation localluy
<azonenberg>
If you want a HiL test case to run, try "./bin/HIL_PGA tests/greenpak4/PGA.txt " from build/
kmehall has quit [Remote host closed the connection]
scrts has quit [Ping timeout: 250 seconds]
scrts has joined ##openfpga
<azonenberg>
whitequark: so i now have both boards plugged into the VM
<azonenberg>
one with a 46620 and one with a 46140
<azonenberg>
i have a socketed 46621 sitting around next to them
<whitequark>
ack
<azonenberg>
I'm going to bang up a time-sharing tool
<azonenberg>
so we can both run tests w/o stepping on each other
DocScrutinizer05 has quit [Disconnected by services]
DocScrutinizer05 has joined ##openfpga
scrts has quit [Ping timeout: 250 seconds]
<whitequark>
azonenberg: can you create a GH API token with the repo:status scope
<whitequark>
this is for integration with GH commit view as well as PRs maybe
<azonenberg>
whitequark: how do i do that again?
<whitequark>
azonenberg: then configure GH to use a "webhook" that posts events to the (publicly routable HTTP endpoint)/change_hook/github?project=openfpga
<rqou>
you technically can automate the dns challenge
<rqou>
using rndc
<whitequark>
rqou: you are technically welcome to write the scripts for that
<rqou>
i might at some point actually
<azonenberg>
whitequark: anyway firewall and DNS should be configured
<azonenberg>
At some point down the road i plan to move this VM to a different host but if you have everything in ansible it should be easy to clone over, right?
<azonenberg>
(as in when i get a dedicated VM server in the garage)
<whitequark>
azonenberg: trivial. wipe and reimage
<azonenberg>
but thats a ways out anyway
<whitequark>
should take about ten minutes to set up
<azonenberg>
Perfect
<whitequark>
with no user interaction :D
<azonenberg>
I don't like using the vmware on my desktop for hosting public services
<rqou>
lol
<azonenberg>
so as soon as i get a chance i'll rig up a kvm box in the garage
<rqou>
so it's not even on esxi or whatever?
<azonenberg>
i'll just gamble that in the meantime, nobody has a buildbot 0day + vmware escape that they're willing to burn on me
<azonenberg>
rqou: vmware workstation :p
<azonenberg>
this is literally for hosting greenpak devkits for unit testing
<rqou>
wow, that's um...
<rqou>
why not xen or kvm?
<azonenberg>
b/c better usb support
<rqou>
my usb is working fine with kvm
<rqou>
i can use my mouse+keyboard :P
<azonenberg>
i've had lots of fancy embedded stuff (cameras, jtag dongles, etc) not work in virtualbox or any other hypervisor
<rqou>
yeah that makes sense
<azonenberg>
Anyway, if possible i want to move things around in the immediate future so that, as a minimum, the web side of the dashboard is hosted on a dedicated vm box somewhere "not here"
<azonenberg>
the build node i dont care as much about being local
* azonenberg
wonders how hard it would be to set up a push system where the dashboard is rendered to static HTML then scp'd to a VPS or something
<azonenberg>
So clients never talk straight to the VM
<whitequark>
azonenberg: put the thing on an rpi and stop caring whether it's compromised or not?
<whitequark>
seems less work
<azonenberg>
you know
<azonenberg>
i do have a beaglebone floating around
<azonenberg>
Let's do initial test configuration on this VM
<azonenberg>
running debian
<azonenberg>
then clone the image over to the beaglebone once we have the testing complete
<rqou>
waiting for the "doesn't work on ARM" bugs :P
<azonenberg>
rqou: hey, gotta test somehow ;)\
<rqou>
want me to spin up a PPC(be) too? :P
<azonenberg>
lol one new isa at a time
<azonenberg>
whitequark: anyway, if we get this working today
<azonenberg>
tomorrow or sunday we can clone the image over to the bone
<rqou>
might actually trigger endianness bugs with a ppcbe
<whitequark>
Detail: Could not connect to openfpga-dashboard.antikernel.net
<azonenberg>
i'll have to find it, hook up an ethernet cable, etc
<whitequark>
says certbot
<azonenberg>
whitequark: can you reach it on http?
<rqou>
you can fix endianness bugs by switching to ppc64le :P
<whitequark>
nope
<azonenberg>
ok, let's see...
<whitequark>
but it does listen
<whitequark>
oh hm
<whitequark>
oh nvm that was a wrong bindv6only setting
<azonenberg>
i just fixed a typoed ip address in the ipv6 firewall
<openfpga-github>
[openfpga-buildbot-config] whitequark pushed 1 new commit to master: https://git.io/vXINl
<openfpga-github>
openfpga-buildbot-config/master 39fabec whitequark: Add basic Ansible and Buildbot configuration.
<openfpga-github>
[openfpga] azonenberg pushed 1 new commit to master: https://git.io/vXIN0
<openfpga-github>
openfpga/master 85ea6f5 Andrew Zonenberg: gp4prog/gpdevboard: Test cases and gp4prog can now request exclusive hardware locks to avoid conflicts
<azonenberg>
whitequark: also can you change the hostname to "openfpga-buildbot" or "openfpga-build"
<azonenberg>
in keeping with the trend for the github bot?
<openfpga-github>
[openfpga-buildbot-config] whitequark pushed 1 new commit to master: https://git.io/vXINj
<openfpga-github>
openfpga-buildbot-config/master 26f8b2b whitequark: nginx: use proxy_buffering off for streaming logs.
<openfpga-github>
[openfpga-buildbot-config] whitequark force-pushed master from 26f8b2b to 3dd7354: https://git.io/vXIAJ
<openfpga-github>
openfpga-buildbot-config/master 3dd7354 whitequark: nginx: use proxy_buffering off for streaming logs.
massi has joined ##openfpga
<cr1901_modern>
azonenberg: i like having a 4k display that i can have build output side by side with the source code <-- tbh, I'm prob going to use a tiled window manager for this at some point when I go back *nix. If necessary, make a daemon or plugin that can paste a command I want to run on keypress
<whitequark>
[ 8562.405369] Out of memory: Kill process 17204 (clang) score 752 or sacrifice child
<whitequark>
[ 8562.405388] Killed process 17204 (clang) total-vm:817984kB, anon-rss:738300kB, file-rss:20396kB
<azonenberg>
did i not give the vm enough ram?
<whitequark>
nope
<azonenberg>
shut it down
<azonenberg>
and i'll add more
<whitequark>
just kill it
bb-openfpga has quit [Quit: buildmaster reconfigured: bot disconnecting]
<azonenberg>
Done, restarting
bb-openfpga has joined ##openfpga
<openfpga-github>
[openfpga-buildbot-config] whitequark pushed 1 new commit to master: https://git.io/vXIAm
<openfpga-github>
openfpga-buildbot-config/master 551705e whitequark: system: set hostname.
bb-openfpga has quit [Client Quit]
bb-openfpga has joined ##openfpga
<whitequark>
as a bonus it won't rebuild everything from scratch every time
<whitequark>
with buildbot
<azonenberg>
vs travis?
<whitequark>
yeah
mifune has joined ##openfpga
<azonenberg>
cr1901_modern: its not much work to just click another window, or alt-tab
<azonenberg>
then up-arrow enter
<cr1901_modern>
azonenberg: If I have to move my mouse from one window to another, me overshooting the window or needing to click in a specific position is enough to break my concentration
<rqou>
so I'm actually installing debian ppc right now for lulz, and amazingly it seems to work just fine
<cr1901_modern>
And alt-tab isn't my favorite, since it'll cycle through ALL windows
<whitequark>
pfff lol azonenberg
<whitequark>
why didn't travis catch this i wonder
<azonenberg>
cr1901_modern: and looking at build output isnt?
<azonenberg>
(enoug hto break concentation)
<cr1901_modern>
azonenberg: Less so :P. I mean the text cursor is prob still at the same place I left it if I don't have to move my mouse, and that's PROB where I want it when the build fails
massi_ has joined ##openfpga
<rqou>
anyways, my imac g3 seems to have this really fun problem where the firmware can't seem to boot from cd if the hdd gets replaced with a cf card
<rqou>
fortunately, the cd drive works after booting from the hdd/cf
bb-openfpga has quit [Quit: buildmaster reconfigured: bot disconnecting]
<rqou>
so as a hack I can unplug the cf card, plug it into a usb reader, install to it, and then plug the cf card back in the ide adapter
<rqou>
the only problem is that my cf card reader is unbelievably slow
<whitequark>
azonenberg: wtf is with the VM?
<whitequark>
doesn't respond by ssh, 502's
<whitequark>
PING lucifer.dmz.bainbridge.antikernel.net (10.2.8.8) 56(84) bytes of data.
<whitequark>
what
<whitequark>
oh nvm
<whitequark>
I borked my ssh config
<whitequark>
and my buildbot config
<azonenberg>
lol
<azonenberg>
do you need me to get in locally?
<whitequark>
nono
<whitequark>
unborked
<rqou>
anyways, in case anyone wants an anti-recommendation, don't buy the i(DOT)Connect all in one USB card reader :p
<qu1j0t3>
rqou: as a powerpc Mac user, I salute you.
<qu1j0t3>
rqou: fight the good fight
<rqou>
except this mac is barely fast enough to even run a gui
<rqou>
it's the 333mhz original imac :P
<rqou>
i doubt vlc will actually be able to play anything even if it could launch
<qu1j0t3>
should be plenty fast for a gui
<rqou>
not the bloat that is firefox :P
<qu1j0t3>
iMac G3, eh. nice
<qu1j0t3>
i have an iMac G4, and a few other G4's and G5s
<qu1j0t3>
rqou: hm, yeah. I run Tenfourfox on a PB G4 1.7GHz
<qu1j0t3>
rqou: that's manageable, but yeah 333 MHz won't cope with modern firefox i agree
<rqou>
a big limitation is that there's no altivec on this machine
<qu1j0t3>
yes.
<qu1j0t3>
Tenfourfox claims to make use of Altivec
<rqou>
hmm i should try putting osx 10.4 on this machine next :P
<rvense>
i grabbed an imac g4 last week.. very pretty
<rvense>
the modem in them is the same connector as in the powermac g4s, but the chips on them are for usb and not serial
<rvense>
so now i'm wondering if the connector has both usb and serial or they just reused it with a different pin layout in the later models..
<qu1j0t3>
rvense: Yeah. they are nice.
<qu1j0t3>
rvense: Does it have wifi ?
<rqou>
my imac g3 is actually really special
<rqou>
it's a rev d 333mhz in bondi blue
<rqou>
which doesn't officially exist
<rqou>
my father got it from his friend who at the time was a tester at apple
<rvense>
qu1j0t3: no, but it's placed about one foot from a switch so i'll manage..
<rvense>
should have an ap extreme card somewhere though
<qu1j0t3>
they're super cheap on ebay too
<rvense>
imagine they'll be worth something in a few years if kept in running order
<rvense>
find it really funny how they have a user-accessible laptop-style RAM slot at the bottom, and a full-sized one on top of the motherboard that's meant to be hidden
<qu1j0t3>
right.
<rqou>
hmm as i suspected vlc/qt is trying to run altivec instructions without probing for support first
<qu1j0t3>
ha!
<rqou>
somebody probably messed up the compiler flags