06:35 imirkin: karol: didn't you have a patch to cover some or all of the compat issues?
08:16 kherbst: imirkin: yeah, and I pushed it
08:19 kherbst: imirkin: the question is rather do we want to make an attempt to fix everything we can think of (and add more tests) or just expose compat like AMD does, because I doubt we are worse off
08:31 karolherbst: docmax: .... well, yeah. xfce4 managers your session so it tries to restore whatever config was set beforehand
11:11 karolherbst: ohh, we have a nvammiotracereplay tool
11:11 karolherbst: mhh
11:11 karolherbst: now having an option to bisect with that might be super helpful
11:12 karolherbst: like nvammiotracereplay --replay_percentage=0.5
11:12 karolherbst: or something
11:14 karolherbst: ohh, we have -s
11:14 karolherbst: nice
11:23 rhyskidd: karolherbst: where did you get the various register names for this from? https://github.com/karolherbst/nouveau_tools/blob/master/dbg_falcon.sh
11:24 rhyskidd: i have a slightly expanded/different list of the Falcon registers that can be reached by the icd
11:24 rhyskidd: will send patch
11:34 karolherbst: uhm
11:34 karolherbst: rhyskidd: they should be inside rnndb already, no?
11:35 karolherbst: or do you mean the falcon registers?
11:35 karolherbst: mhh, let me think
11:35 karolherbst: rhyskidd: might be envydis
11:36 karolherbst: https://github.com/envytools/envytools/blob/master/envydis/falcon.c#L62-L76
11:36 karolherbst: mhh
11:36 karolherbst: it is incomplete
11:36 karolherbst: iv2 is there isnce fuc5 or something
11:38 karolherbst: rhyskidd: does this script even works? I don't know if I left it in a broken state or not
11:39 karolherbst: but in longterm I really want to have a proper debugger we could use, also for debugging the gr issues we have on a few boards
12:05 mwk: huh, there is an iv2 now?
12:09 karolherbst: mwk: yeah
12:09 karolherbst: but... it is a weird one
12:09 karolherbst: I think imirkin knows something about it
12:16 karolherbst: mwk: from the release firmwares: https://gist.githubusercontent.com/karolherbst/3b9a42e405813294a0f1efd2dbcc1fba/raw/9185238d7b3ece1d3ef6cc459653739633bc061e/gistfile1.txt
12:17 karolherbst: "bset $flags ie2"
12:17 karolherbst: need to find where iv2 was used again
12:18 karolherbst: ohhw ait, no, that is soemthing else
12:18 mwk: yeah
12:18 mwk: as I remember, the problem was that we found an extra set of interrupt enables, but no interrupt vector
12:19 mwk: and there's a talk by nvidia that strongly implies another kind of interrupts with externally-delivered vector, which would imply lack of iv2
12:20 karolherbst: yeah... maybe that was it
12:21 karolherbst: but I don't really remember anymore. I think I might have played around with that special reg and it did something
12:21 karolherbst: but..
12:21 karolherbst: again, I don't remember
12:27 karolherbst: mwk: mhh, if that "iv2" special reg has no functionality, it could be used as a gpr, or are we limited by the encoding?
12:27 karolherbst: or maybe it doesn something else
12:27 karolherbst: still wondering
12:28 mwk: it's a special register, the only insn that can encode it is a mov to/from sr
12:29 karolherbst: okay
12:30 mwk: assuming it even exists
12:30 mwk:is still not sold on that
12:32 mwk: oh, hm
12:32 mwk: the gk20a driver headers mention it existing
12:33 mwk: but
12:33 mwk: no
12:33 mwk: it's not a register
12:33 mwk: it's a trap cause
12:33 mwk: hw_pwr_gk20a.h has a list of special regs, and only iv0-iv1 are present
12:34 mwk: it also has a list of trap causes, and it has iv0-iv2
12:40 karolherbst: mhhh
12:40 karolherbst: mwk: I remember that I could write to and read from that register at least
17:05 docmax: i need the xfce4 channel
17:05 docmax: the damn xorg.conf
20:17 Pie_Mage: https://ss64.com/bash/fg.html
20:17 Pie_Mage: lol wow
20:17 Pie_Mage: such IRC fail today
20:18 Lyude: freenode is just not a good irc network
22:03 RSpliet: Lyude: quite honestly I'm amazed IRC is still a thing in the first place
22:04 Lyude: RSpliet: yeah
22:04 Lyude: i think we all are :P
22:04 Lyude: even as someone who still admins an irc net
22:04 Lyude: IRC has at least survived the lifetime of two whole AIMs
22:04 RSpliet: Arguably the same is to be said about e-mail
22:05 RSpliet: But no, we reboot IRC and call it Slack :')
22:05 Lyude: and maybe like, 4 slacks?
22:05 Pie_Mage: slack < IRC
22:05 Lyude: yeah slack is pretty not great
22:06 Lyude: i'm very much not interested in messengers that don't have some sort of open platform (unless slack does??)
22:06 HdkR: I personally enjoy when slack goes down and takes literally everything with it
22:06 Lyude: because they will be gone in like 3 years anyway
22:06 Lyude: so says the cycle of instant messengers
22:06 HdkR: "Oops, our connection system died. Guess everyone worldwide cant log in."
22:07 RSpliet: Lyude: shit... MSN messenger is dead? :-OOO
22:07 Lyude: it's occurring to me now that all things considered, it's not terribly surprising IRC is still here
22:07 Lyude: RSpliet: yep! fun fact: it only actually died like
22:07 Lyude: 3 years ago?
22:07 loonycyborg: I know there are some alternatives
22:07 Lyude: it lived on for years after MS said it discontinued because they never shut down the servers, they just disabled the official clients
22:07 loonycyborg: like martix and rocketchat and mattermost
22:08 Pie_Mage: i find it strange that we haven't solved the 'send strings to other people over the wire' thing yet
22:08 Lyude: we have
22:08 Lyude: people keep getting distracted by stickers!!
22:08 Lyude: (really though)
22:08 RSpliet: We just haven't solved the "I don't want your strings on my wires" problem...
22:09 Pie_Mage: how can you express yourself on the internet in 2018 without emojis?!
22:09 HdkR: 🤣
22:11 Lyude: 🅱🅱🅱🅱🅱🅱
22:12 HdkR: Ruining IRC with this poison
22:12 HdkR: :)
22:20 Pie_Mage:suspects it's a coalition between slack and discord
22:37 nyef: Slack? Bah. How do I configure it so that a message that bumps a notification counter sounds different from one that doesn't?
22:37 nyef: Thus far, I have yet to find a way.
22:54 karolherbst: RSpliet: well all IRc alternatives are just worse
22:55 karolherbst: Pie_Mage: please no, both Slack and discord are evil and their software stack is electrum based, which you should never use
22:56 karolherbst: or how that chromium thing was called
22:56 karolherbst: electron
22:56 karolherbst: at least electron has this nice list of software you should never use
23:03 RSpliet: karolherbst: That's probably because the OSS community isn't trying :-P
23:03 karolherbst: oh well
23:04 karolherbst: then you could just improve IRC itself anyway
23:04 karolherbst: there isn't anything really what slack or discord or so add to IRC
23:11 RSpliet: Nah, I think what might make IRC better is manageability. It's all a tad... hacky. With chanservs, nickservs etc ;-)
23:12 karolherbst: why is that hacky?
23:13 karolherbst: doesn't really make a difference if you do seperate API calls or just use the only thing IRC really needs
23:13 karolherbst: latter just leads to simplier design and implementations
23:14 karolherbst: the annoying part is merely that every server is different and those services have different APIs
23:14 RSpliet: There's quite some divergence in channel- and user modes beteen like freenode and quakenet. I'm quite convinced there's a common ground that can be standardised ;-)
23:14 karolherbst: sure
23:15 karolherbst: but having to msg services to do something isn't a hack
23:16 RSpliet: It's not particularly clean design. And divergence prevents useful user-friendly interfaces to control these kind of settings.
23:16 karolherbst: why isn't it a clean design?
23:16 karolherbst: instead of PRIVMSG you could also have dedicated commands
23:16 karolherbst: but
23:17 karolherbst: those are basically the same thing
23:17 karolherbst: sure you could just come up with some IRC API and specify like every single command and everything
23:17 karolherbst: and just not use PRIVMSG but CHANSERV
23:17 karolherbst: doesn't change the implementation
23:19 karolherbst: and IRC is actually specified inside a specification
23:20 karolherbst: the problem is that the services are different on every server
23:20 karolherbst: not that we send PRIVMSG to them
23:20 karolherbst: in the end it kind of gives the servers freedom to do what they want though
23:21 karolherbst: but yeah, maybe some common ground would be nice