00:12phoenixz: imirkin: Well, #systemd says its not their issue, but we did get to the point where I ctrl-alt-backspace killed X, relogin and tadaaaaah, 3 monitors!
00:13phoenixz: imirkin: BUT, now I have extreme screen and input freezes
00:13phoenixz: Everything just freezes for 5-7 seconds every 20 seconds or so.. I have dstat running and on complete system freezes it will show missed ticks, but it doesn't do so now, so I can only imagine its X hanging..
00:14phoenixz: Also the entire thing is VERY VERY slow, like 10 FPS or soemthing
00:14imirkin: pastebin xorg log
00:14phoenixz: yeah, glxgears shows 15FPS
00:15phoenixz: will do. Please be patient, these freezes makes it a bit hard to get anyting done now :)
00:15phoenixz: But at least I got three monitors.. I'm arleady happy! thopugh I still need to figure out what the exact issue is, I don't want my boot up to require login, kill X, relogin..
00:17phoenixz: imirkin: btw: systemd suggested that perhaps it was plymouth, starting X *before* some acl's were applied, causing X not having access to the cards. Subsequently, if I kill X and restart it, it will finc the ACLs applied, so then it DOES work.. still gotta check that.
00:18nyef: phoenixz: If that's the case, try disabling the X on boot thing, and kick it off manually. Not an absolute test, but if it works it's indicative of at least *something* in terms of a missed dependency.
00:19imirkin: could be. bug again well outside the "nouveau's problem" category
00:19nyef: Alternately, compare the X logs from the first and second starts.
00:20phoenixz: http://pastebin.com/wnCjFJS2 note: contains the first startup that fails, and then I ctrl-alt-backspace kill X and it restarts again, all in this log
00:20nyef: Advantages to a linearized startup sequence: It's far, far less vulnerable to people being sloppy about declared dependencies.
00:21phoenixz: imirkin: correct, this but appears to be outside the scope of nouveau, though my current 15FPS might still be within.. I hope we could fix that.. Then the startup issue, as I'll call it from here, should not be too hard anymore to fix now
00:22phoenixz: nyef: yeah, I'll try to disable plymouth at startup.. a nosplash as boot parameter should do IIRC.
00:22imirkin: phoenixz: pastebin the output of glxinfo
00:23dboyan: whoa, I saw patches that changes disk_cache API on the list, maybe I should update my previous patch after its landing
00:24phoenixz: imirkin: I know that this asus geforce 210 is nowhere near top of the line (again, I need a quiet station, no small noisy ventilators) but even those card should be able to do better than the (at full screen) 8FPS glxgears @1920x1080, right?
00:24phoenixz: imirkin: pasting
00:26phoenixz: imirkin: argh.. sorry for being so slow, but I cant do much faster right now..
00:26phoenixz: imirkin: http://pastebin.com/QzetdLXL
00:26imirkin: i'm in no rush
00:27imirkin: phoenixz: as root, cat /sys/kernel/debug/dri/0/pstate and pastebin that
00:27imirkin: (oh wait, crap... you're on kernel 4.4 aren't you... you'll need to boot with nouveau.pstate=1 then)
00:28phoenixz: imirkin: /sys/kernel/debug/dri/0/ doesn't have the file "pstate", I only see bufs clients DVI-I-1/ gem_names HDMI-A-1/ name vbios.rom VGA-1/ vm vma
00:29imirkin: yeah. you're on an old kernel =/
00:29imirkin: you need to boot with nouveau.pstate=1 to enable the pstate file, and it'll be in sysfs
00:29imirkin: rather than debugfs
00:30imirkin: my guess is those gpu's are clocked super-low due to being fanless
00:30imirkin: however they should support reclocking, so we can make use of that
00:31phoenixz: I"m currently running kernel 4.4.0-59-generic, though I see 4.8.0-32 available.. Would it help if I upgrade to that version?
00:31phoenixz: imirkin: nouveau.pstate=1 can be added in grub @ boot time with the edit option?
00:32imirkin: 4.8 would have the new logic
00:32imirkin: it was changed in 4.5
00:33phoenixz: imirkin: I'll install the newer kernel first
00:34phoenixz: then reboot and also try the nosplash, to see if it fixes the startup issue with no acces to card1 and card2
00:35phoenixz: Done.. rebooting..
00:42Lyude: Is there any kind of documentation on the custom language piglit uses for testing shaders?
00:44imirkin: sure - it's in tests/shaders/shader_runner.c
00:45Lyude: starting to get this bit by bit
00:45imirkin: look at some simple ones ... and copy
01:19Horizon_Brave: Hello folks
01:21Horizon_Brave: so...two questions... If a device like a wireless card or graphics card uses firmware...and firmware by design is 'burned' into ROM on the card, why is it that I"m reading around and there's many references to software drivers on the OS side, needing to "load" firmware to the device...
01:22Horizon_Brave: shouldn't the burned in firmware already be there? I see it in the debian community a lot...where they have firmware that's freely available from their repos...and then other non-free firmware
01:22Horizon_Brave: why does the device need to have anything loaded though, if the firmware should be there from the manufactuerer?
01:23mjg59: It's not burned in
01:25Lyude: it also gets changed every now and then
01:25Lyude: that's the big reason it's not just burned in
01:26imirkin: and it's cheaper not to have a rom on board
01:29gnarface: Horizon_Brave: traditionally it had to be burned onto an EEPROM or something like that. it's been decades since that was considered a technical limitation.
01:29Horizon_Brave: so...if that's the case... and the firmware can just be downloaded and simply installed by the OS driver...how is it that some firmware is proprietary? or "non-free"? If we have access to it, how is it not possible to just open it up and see the code?
01:30gnarface: Horizon_Brave: now, MUCH of the hardware out there can dynamically load a new copy of the ROM after it boots up, with no need to actually re-flash itself (so long as it can always find the firmware on disk where it expects to)
01:31Horizon_Brave: how can nivida do their "closed source" only policy if the firmware for their devices is just another piece of software?
01:31gnarface: Horizon_Brave: and no, you can't "just open it up to see the code" if it's a binary blob. you can reverse engineer it to some degree, but it's not like legos where you can take it apart into the same components it was assembled from.
01:32gnarface: Horizon_Brave: it's more like soup. you can taste it and study it in a lab to see what was put into it, and maybe with some time and effort have a very good idea of the ingredients, but you're not gonna find a physical copy of the recipe used to make the soup anywhere in the actual soup.
01:33Horizon_Brave: gnarface: okay...so the devices can load and "update" their own firmware given that they know the location of where you downloaded it to... aaand right of course, so the companies write the firmware, build it out so it's binary only, and then toss it to us, and say "here use this to get all the features of XZY card etc.."
01:33Horizon_Brave: gnarface: Nice...I'll mentally modify that analogy and use chili instead so it appeals more to me :P
17:48phoenixz: imirkin: I'm back! So yesterday you told me to try and add nouveau.pstate=1 so I could check.. something in /sys, and / or install a new kernel... I installed a new kernel and my entire installation went full kamikaze on me.. Since reinstallation is a pure horror I spent about 4 hours trying to repair it, then had to give up, make a backup of the terrabyte of info I had on the drive, reinstall and finally I am here
17:48phoenixz: Good news though, after reinstall, somehow, miraculously, my 3 monitors are working!
17:49phoenixz: imirkin: Still though, I'm left with one issue, namely the slow video performance.. if / when you have some time, could you perhaps help me out with that?
17:50imirkin_: phoenixz: so ... did you boot with that parameter and/or new kernel?
17:51phoenixz: imirkin: nope, not yet.. I just restarted, and checking.. I'm on 4.4.0-53-generic atm
17:51imirkin_: ok. well boot with nouveau.pstate=1
17:51phoenixz: imirkin: Basically, since the reinstall requires me to already fight with boot-repair and such.. took me 4 hours to install this horror show, but it just booted, yay!
17:51phoenixz: imirkin: okay, going to do that right now
17:52phoenixz: If I don't come back then.. well, I guess it suicided again.. fingers crossed...
17:53Horizon_Brave: Hi everyone
17:54imirkin_: hi dr nick!
17:58phoenixz: Yes! I'm back!
17:58phoenixz: So reboots are also working, that's good..
17:58phoenixz: imirkin_: I added the kernel parameter
17:58Horizon_Brave: imirkin_: lol I think his line is more "Hi Everybody!"
17:58imirkin_: phoenixz: ok. great. now pastebin the output of cat /sys/kernel/debug/dri/0/pstate
17:59imirkin_: phoenixz: er sorry
17:59imirkin_: phoenixz: cat /sys/class/drm/card0/device/pstate
17:59phoenixz: Horizon_Brave: If you're referring to me, I'm just so so happy that my computer is working at all.. systemd and UEFI are bloody disasters
17:59imirkin_: phoenixz: no, to what i said a second before you joined
17:59imirkin_: Horizon_Brave: yes, it is... but ... close enough
17:59Horizon_Brave: phoenixz: lol..no..it's a Simpsons reference
17:59phoenixz: Horizon_Brave: Ah, never mind then
17:59Horizon_Brave: how dare you! systemd isn't bad
18:00Horizon_Brave: UEFI...is.. well it's well intentioned..but causes a lot of headache..for me atleast
18:00Horizon_Brave: what's the issue?
18:00phoenixz: Horizon_Brave: oh crap, don't ban me please... Last time I said that on #ubuntu I got banned for it.. Well, I got a warning, then I said, "okay, I'll just say its great then" and I got banned..
18:01phoenixz: imirkin_: http://pastebin.com/87TmhycH
18:01Horizon_Brave: phoenixz: lol what? what dick would ban someone for an opinion...
18:01phoenixz: Horizon_Brave: Let
18:02imirkin_: phoenixz: ok. now, this might hang your box, but try this, as root: echo 0f > /sys/class/drm/card0/device/pstate
18:02phoenixz: Horizon_Brave: Lets just say that I've seen that more and more in loads of channels in the last 2 years.. You cannot critizise anymore, because then apparently you attack people..
18:02phoenixz: imirkin_: Will do.. IF hang, I'll be back in a few
18:02Horizon_Brave: phoenixz: sadness..
18:03phoenixz: imirkin_: still here..
18:03imirkin_: phoenixz: ok, cat the file again - what does the AC line say?
18:04phoenixz: imirkin_: 0f: core 589 MHz shader 1402 MHz memory 600 MHz AC DC *
18:04phoenixz: AC DC yeah :)
18:04imirkin_: phoenixz: sorry, the line that starts with AC:
18:04Horizon_Brave: can't get any better than that
18:05Horizon_Brave: you're using Ubuntu ?
18:05phoenixz: Horizon_Brave: problem with UEFI and systemd though.. UEFI basically effs up install, linux mint 18.1 installer is quite borked as well, but with UEFI its just plain impossible.. When I got the computer first day in, it took me 4 days and.. I dunno, over like 30 reinstalls to get an actually booting system.. Then systemd has been a nightmare as well
18:05phoenixz: Horizon_Brave: mint
18:06phoenixz: imirkin_: AC: core 589 MHz shader 1401 MHz memory 597 MHz
18:06imirkin_: ok great
18:06imirkin_: did things speed up perchance?
18:07Horizon_Brave: is that the speed/output from your onboard graphics card?
18:07Horizon_Brave: or do you have a dedicated card?
18:07imirkin_: his cards are so dedicated that he has three of them
18:07phoenixz: imirkin_: Nope.. Still having around 5FPS here, and if I try to move a window around, it just hangs for 4 seconds
18:08imirkin_: phoenixz: hmmm... can you repeat the same exercise for card1 and card2?
18:08phoenixz: Horizon_Brave: 3 ASUS nvidia 210 cards.. nothing high end, but I want my system quieeeetttt
18:08Horizon_Brave: man, that's loyalty
18:08phoenixz: Horizon_Brave: how so?
18:08phoenixz: imirkin_: done, no effect, still
18:09Horizon_Brave: lol, well that's three different fans all going at once..
18:09imirkin_: hm, i wonder if the cards come up in a lower pcie lane width mode. we didn't have pcie lane width setting back in 4.4...
18:09imirkin_: Horizon_Brave: fanless
18:09Horizon_Brave: phoenixz: lol again it was a joke to what imirkin said...
18:09phoenixz: imirkin_: also, checked output with cat, and it looks the same
18:09phoenixz: Horizon_Brave: no fans at all, thats why its quiet :D
18:10imirkin_: phoenixz: as root, can you run lspci -vvnn -d ::300
18:10imirkin_: and pastebin the results
18:10phoenixz: imirkin_: http://pastebin.com/jDybRzzb
18:11imirkin_: phoenixz: also pastebin output of 'xrandr'
18:11phoenixz: I know these cards are low end, but I should be able to get better than 5FPS@1920x1080 from them, right?
18:11phoenixz: imirkin_: http://pastebin.com/2ecLxB2A
18:12imirkin_: ok, so, my GUESS is that it's due to the lower pcie speed/lanewidth
18:12imirkin_: LnkSta: Speed 2.5GT/s, Width x8
18:12imirkin_: whereas the cards should be capable of x16. dunno about 5GT/s, depends if your mobo supports PCIev2
18:12imirkin_: actually i guess it does, so 5GT/s should be possible
18:13Horizon_Brave: umm...I think any card should do better than 5fps lol...
18:13imirkin_: oh, ouch - LnkSta: Speed 2.5GT/s, Width x4
18:13imirkin_: for that last one.
18:13phoenixz: Its an ASUS Maximus VIII board, if htat helps?
18:13imirkin_: they're actually probably x8 GPUs to begin with, and that third PCIe port is probably only wired for x4
18:14imirkin_: however with a newer kernel version you should be able to get 5GT/s instead of 2.5GT/s
18:14imirkin_: that said, i dunno how big of a diff it'll end up making.
18:14imirkin_: the fact that it's so slow is definitely weird
18:14imirkin_: but you're using all kinds of modern fanciness that i've shied away from, and i have no idea how it behaves, since i never use it
18:15phoenixz: imirkin_: modern fancyness?
18:15phoenixz: like systemd? :)
18:15imirkin_: like GL-based things that run on your screen for no reason
18:15Horizon_Brave: does this operate by using the slowest speed of the three? meaning even though the first two cards are wider in bit width, the last card is the maximum that the other two can use?
18:15imirkin_: like a compositor, etc
18:16imirkin_: totally unnecessary window effects
18:16imirkin_: who knows
18:16phoenixz: imirkin_: well, problem is that I tried installing newer kernels with my previous install and it killed it
18:16phoenixz: imirkin_: correct, using a compositor
18:17phoenixz: imirkin_: I could do without that if needed, obviously, but that is just the nice little touch you like to have when you have this much hardware..
18:17phoenixz: imirkin_: Would it help if I only use 2 cards, then?
18:17imirkin_: phoenixz: worth trying
18:18imirkin_: could be that third one in the x4 slot gumming up the works. i really don't know.
18:18imirkin_: i do know that people have successfully created 20-screen video walls
18:18phoenixz: imirkin_: These nvidia 210 cards have HDMI, DVI and VGA.. They should be able to run dual monitor, right?
18:19imirkin_: out of 5x dual-gpu dual-output (so 4 per slot) nvidia cards
18:19imirkin_: yes. they can support 2 heads.
18:19imirkin_: and you can get DVI <-> HDMI adapters
18:19phoenixz: imirkin_: yeah, I've seen those, and I'm always wondering about that when I look at the trouble I have with my 3 multi monitors..
18:19imirkin_: (i think... definitely HDMI -> DVI. the other way around is a little more questionable in my mind)
18:20phoenixz: imirkin_: I take it I'll have to physically remove that 3rd card then, right?
18:20Horizon_Brave: lol hey that was my idea
18:20imirkin_: phoenixz: no, just don't use the screen on it.
18:20phoenixz: and pray that my machine will boot again, with all the trouble I've had
18:20imirkin_: phoenixz: you can disable it by doing xrandr --output HDMI-2-3 --off
18:20phoenixz: Lets see if I have a store nearby that sells these converters, I don't have one lying around
18:21phoenixz: imirkin_: Could I try that right now?
18:21phoenixz: imirkin_: I mean, if I disable it, should I then see an increase in speed?
18:21imirkin_: that's my theory.
18:22phoenixz: Holy crap, it does appear to go quite a bit faster.. Still some hiccups, but this looks a lot better already..
18:22phoenixz: lemme try glxgeard
18:23Horizon_Brave: phoenixz: are you using the nvidia drivers for this?
18:23Horizon_Brave: or the nouveau?
18:24imirkin_: nouveau. i don't do free support for nvidia's blob driver.
18:24imirkin_: (or paid support for it, for that matter)
18:24Horizon_Brave: stick it to the man!
18:24phoenixz: imirkin_: Don't touch that stuff with a 30 foot pole?
18:24imirkin_: phoenixz: i do when tracing it :)
18:24nyef: ... "with a 10 foot snide remark"?
18:24Horizon_Brave: jeez...their driver is a binary blog too? AND their firmware? lol, do they release anything to you guys??
18:25phoenixz: imirkin_: lol.. I take it you're a nouveau driver dev?
18:25imirkin_: likely guess.
18:26phoenixz: Horizon_Brave: something something linus something fuck you nvidia
18:27phoenixz: Yeah, you have my sympathy.. I really don't want to think about having to deal with all that..
18:27phoenixz: heh, weird..
18:27phoenixz: I have 3 screens
18:27phoenixz: the middle one is off now
18:27imirkin_: coz the middle one is plugged into the third gpu
18:27phoenixz: and the mouse pointer can only stay in the right one.. I have quassel in the left one, but the mouse pointer can't get there anymore..
18:29phoenixz: errr.. How do I re-enable that screen again? :) --on isn't an option, apparently
18:30phoenixz: imirkin_: yeah, its off for the obvious reason, but the weird thing is that the mouse cursor is now limited to the right screen, I can't get it to the (for the rest correctly functioning) left screen
18:30imirkin_: phoenixz: xrandr --output HDMI-1-2 --right-of HDMI-1 --auto
18:30imirkin_: er, that will make your other screen next to the first one
18:30imirkin_: if you want it back, do
18:30imirkin_: phoenixz: xrandr --output HDMI-2-3 --right-of HDMI-1 --auto
18:37phoenixz: imirkin_: looks like I have to train myself to work a bit more with xrandr I see...
18:38phoenixz: imirkin_: well thanks a lot for the help so far! Going to look for a DMI converter I guess!
18:38imirkin_: phoenixz: there's another tool, arandr, which has been described as more user-friendly. i've never used it.
18:38imirkin_: (usually i find things described as "user-friendly" to be extremely difficult to use)
18:39phoenixz: imirkin_: yeah, userfriendly is usually "nice for those who don't know" but an utter horror for the rest
20:44phoenixz: oh my gawd, nvidia, please stop making video cards with DVI! Nobody wants that crap!!
20:44phoenixz: Found a cable, only took like 2-3 hours or somehting
20:44phoenixz: Nobody sells those dvi > hdmi cables anymore.. Anyway, connected and testing..
20:46abff: I'm here to lurk, I had an issue last night I suspect was caused by nouveau but I'm not ready to point the finger yet.
20:46imirkin_: abff: when in doubt, it's nouveau's fault
20:47imirkin_: that's the bright side. the downside is that it's unlikely that there will be a solution available.
20:50abff: imirkin_ so I've been told, but it'll be a lot of googling and trudging through kern.log before I'm certain
20:50abff: at anyrate I'm here to listen
20:50imirkin_: care to describe the issue to see if it sounds familiar?
20:52abff: Last night I was watching some video on kodi, chatting on IRC. When I closed kodi my fan started going wild so I loaded top and saw a kworker using 100% cpu. Tried closing every window in xfce4 to no avail
20:52imirkin_: what gpu do you have?
20:52abff: when I went to emulate a new terminal to kill -9, xfce froze, I panicked and did a hard shut down. If I would have stayed calm I could have done something more useful
20:52abff: uhh its a geforce something or other one sec
20:52imirkin_: did you see a BUG printed due to a spinlock being re-acquired?
20:53imirkin_: abff: lspci -nn -d 10de::300
20:53abff: NVIDIA Corporation GF119M [GeForce GT 520M]
20:53imirkin_: hmmm ok.
20:53abff: I have a log one sec
20:53imirkin_: prolly not what i had in mind.
20:54abff: NMI watchdog: BUG: soft lockup - CPU#1 stuck for 23s!
20:54nyef: Hunh. Another GF119?
20:54abff: It's happening on all CPUs
20:54abff: it happened**
20:54imirkin_: nyef: mobile
20:54imirkin_: abff: pastebin the trace
20:55abff: toshiba p755
20:55nyef: ... And now I'm thinking of "Subgenius Police Usenet Tactical Unit Mobile" for some reason.
20:58abff: game over man, game over
20:59abff: It must be exhausting trying to write a driver for the number of cards nvidia makes. . .
21:00imirkin_: looks unrelated to nouveau.
21:00abff: This is why I didn't want to point fingers, may I ask what led you to that conclusion?
21:01abff: So I have a point of contention to work off of while trying to determine the actual cause
21:01imirkin_: no functions with nouveau/nvkm in the name
21:01imirkin_: are all the traces the same?
21:02imirkin_: i think there are additional lines which could be useful
21:02imirkin_: like the process which triggered this
21:02abff: I know! it was kworker shinnanigans
21:02abff: and I'm not yet smart enough to know what it was doing
21:03imirkin_: a kworker doing sys_open? i don't think
21:03imirkin_: that that happens
21:05abff: Here I'll pastebin the whole grep of the log
21:05abff: if you think you can help but there is a section that has nouveau all over it
21:06abff: @ 04:47:41 I believe is when the problem started
21:06abff: I probably didn't grep far enough into the future but I didn't let it sit for too long before rebooting
21:07imirkin_: ok, so looks like it died coming up from runtime suspend
21:07imirkin_: and probably left some lock locked
21:07imirkin_: which in turn killed the other stuff like "top"
21:07imirkin_: Mar 16 04:48:32 kara kernel: [224850.946895] BUG: unable to handle kernel paging request at ffff969f218abffc
21:08imirkin_: (in evo_wait)
21:08abff: I see SMP there too
21:08imirkin_: that just means the kernel was built with CONFIG_SMP
21:08imirkin_: (i think)
21:08abff: lol I was just about to stop samba forever
21:08abff: all it's doing is trivial sharing with my wii
21:09imirkin_: what's the connection between SMP and samba?
21:09imirkin_: (SMP != smbd)
21:09abff: the connection is I forgot it was smb
21:09imirkin_: SMP = symmetric multiprocessing
21:09imirkin_: aka multi-cpu
21:10imirkin_: [as opposed to things like SIMD and whatnot]
21:10nyef: Symmetric Multi-Processing, Server Message-Block, whatever, right? (-:
21:11nyef: ... Surface-Mount Device, Simple Message Queue...
21:11phoenixz: imirkin_: Okay.. Connected DVI > HDMI, HDMI came up, DVI not yet, but had that weird mouse pointer issue again (with the DVI in the middle).. So rebooted, and now the entire computer didn't come up at all! So I removed the 3rd card, reboot, and now it booted okay, got into desktop, but only the 1st card (only HDMI connected) shows monitor, the 2nd card doesnt.
21:12phoenixz: 2nd monitor IS primary display though for some reason, so I have to fight a bit to get system settings window in my visible monitor. Now I see there 3 monitors, yay!! 2 appear enabled, one disabled (note: only one gives image, still).. I try enabling 3rd, nohting happens.
21:12imirkin_: nyef: Surface-Mount Technology - everyone's favorite. esp those 0402's
21:12phoenixz: imirkin_: xrandr does show HDMI and DVI being connected, but doesn't show monitors: http://pastebin.com/QxQ0x7ws
21:12imirkin_: phoenixz: i'd highly recommend flipping which card is the primary
21:12phoenixz: imirkin_: correction, xrandr does show the monitors connected, but still, they don't get a signal
21:13phoenixz: imirkin_: okay.. How would I do htat? I take it with xrandr?
21:13imirkin_: the 2 hdmi monitors should have signal
21:13imirkin_: phoenixz: no... it'd have to be in the bios
21:13phoenixz: I really really need to dive into the documentation of that command..
21:13imirkin_: phoenixz: or just plug your DVI thing into the "other" GPU
21:13phoenixz: Ah, okay..
21:13phoenixz: imirkin_: I think that would be easier.. replugging
21:14glennk: imirkin_, don't sneeze around 0402s
21:14abff: imirkin_ thanks for taking a look at the log, before I afk what is that paging request thing about
21:14Tomin: I think you forgot Simultaneous Multi-Threading
21:14imirkin_: phoenixz: ok, so now you can do xrandr --output HDMI-1-2 --off; xrandr --output DVI-I-1 --right-of HDMI-1 --auto; xrandr --output HDMI-1-2 --right-of DVI-I-1 --auto;
21:15imirkin_: glennk: esp when they're not soldered on ;)
21:15imirkin_: abff: well, looks like it messed something up. it's not supposed to get those errors.
21:16phoenixz: uh oh... I can no longer drag windows, I think kwin is borking here.. trying xrandr
21:16phoenixz: yeah, gotta reboot I think.. brb!
21:17abff: thanks again imirkin_ :)
21:18nyef: I was more playing off the visual similarity between b, d, p, and q. And, yeah, anything smaller than 0805 is too small for me to work with at this point. /-:
21:19imirkin_: anything that's surface-mount is too small for me :)
21:25nyef: I can manage 1206, and I'm hoping that I can handle 0805, but 0603 might as well be dust or sand grains as far as that goes.
21:35kattana: so far virgl is doing better than on an amd
21:35kattana: running a linux guest with full compositing and everything working.
21:35kattana: xorg log reports even vdpau working
21:36imirkin_: that has nothing to do with vdpau working
21:37kattana: I would say yes, the vm is passing vdpau acceleration to virgl/nvidia on host properly.
21:38kattana: virgl on a amd compositing is being drawn correctly, got artifacts, missing borders etc.
21:48phoenixz: imirkin: Okay, that wasn't a good idea.. If I connect the DVI in the 1st card, system wont start.. I do get to GRUB, after that, pure darkness.. I tried to type the password and see if I get any reaction from the system, but nop..
21:48phoenixz: So its on the second card again, but still, only the first monitor is showing image, the other two show nothing
21:49phoenixz: But at least now I can execute xrandr again.. lemme try that
21:50phoenixz: imirkin: okay.. did so, and.. nothing
22:03phoenixz: imirkin: yeah, I figured, np, when you have time!
22:03phoenixz: Meanwhile.. The devices are called a bit differently, I think, so I changed your line to: xrandr --output HDMI-1 --auto; xrandr --output DVI-I-1-2 --right-of HDMI-1 --auto; xrandr --output HDMI-1-2 --right-of DVI-I-1-2 --auto;
22:04phoenixz: But that still doesn't work.. I checked xorg log
22:04phoenixz: xorg log now shows "Cannot do multiple crtcs without X server dirty tracking 2 interface" and (EE) NOUVEAU(G0): failed to set mode: No space left on device..
22:04phoenixz: that first message (according to google results) appears to point back at a nouveau bug
22:05phoenixz: oh god, I'm going to have to use a different kernel, am I not? :) Last time I tried that, the system went kamikaze.. I'll see what I can do there
22:07phoenixz: https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-nouveau/+bug/1597437 mentions https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-nouveau/+bug/1597437/comments/5 ?
22:08kattana: phoenixz: what's your issue? multiple monitors?
22:15imirkin_: phoenixz: i remember that bug
22:15imirkin_: phoenixz: you need a newer xf86-video-nouveau
22:15imirkin_: it messed up when compiling against a too-new xorg
22:15phoenixz: kattana: multipole monitors indeed
22:15imirkin_: iirc it was fixed in xf86-video-nouveau 1.0.13
22:16phoenixz: imirkin: yeah, thats what I"m reading here.. Now that again, oh oh, wil it ever stop! :)
22:16phoenixz: I'm really having a good stroke of bad luck here.. Let me check if I can find a PPA with never versions
22:17imirkin_: [or nouveau compiled against an older xorg...]
22:19kattana: phoenixz: works in nouveau, but the commands gotta be specific to your set up though.
22:19kattana: xrandr is the tool.
22:20phoenixz: kattana: yeah, we've been at this for a while :) Problem right now is a "bug" in my nouveau driver.. well, not really, more like a different version issue, I suppose.. Trying to update my driver now with newer version from some PPA somewhere
22:21imirkin_: kattana: he's having additional issues as a result of doing gpu screen offloading or whatever it's called
22:21kattana: phoenixz: can I reccomend gentoo?
22:21kattana: gpu screen offloading? what does that do? does it improve games performance?
22:23kattana: phoenixz: what's the purpose of screen offloading+
22:23imirkin_: no, but it lets GPU 1 draw to screen connected to GPU 2
22:23phoenixz: imirkin: quick question, do you happen to know from what version this was resolved? or if not, where could I find this info? like a changelog
22:23imirkin_: <imirkin_> iirc it was fixed in xf86-video-nouveau 1.0.13
22:23phoenixz: Thats the version I'm looking at in a PPA right now
22:24phoenixz: imirkin: 1:1.0.13+git20170208.1516d35b-0ubuntu0ricotz~yakkety
22:26Lyude: hey imirkin_ I thought I added everything I'd need to make mesa support NV_fill_rectangle but I can't seem to get the extension to come up in glxinfo, can you see anything obvious I might be missing? https://github.com/Lyude/mesa/tree/wip/NV_fill_rectangle
22:27imirkin_: Lyude: something that exposes the cap
22:27kattana: interesting arrangement
22:27Lyude: imirkin_: https://github.com/Lyude/mesa/commit/e5ebe06fb1009fa8272583785d6c974a31847b25#diff-d2148248b5e530db60229aa9a7d81e40 ?
22:27imirkin_: oh oops, missed that.
22:29imirkin_: Lyude: my guess is it's an issue with how you're executing stuff?
22:29imirkin_: if you're using LIBGL_DRIVERS_PATH, chances are you're doing it wrong
22:30imirkin_: Lyude: i recommend ./configure --prefix=somewhere; make install, and then LD_LIBRARY_PATH=somewhere/lib
22:30Lyude: my push script for this just throws mesa onto the machine and make installs it over the mesa install
22:30Lyude: so it should just be replacing the system mesa
22:30imirkin_: pastebin glxinfo
22:30Lyude: imirkin_: https://paste.fedoraproject.org/paste/a8plB5c3WZ9Qqij64vWxOF5M1UNdIGYhyRLivL9gydE=/
22:30imirkin_: OpenGL renderer string: Gallium 0.4 on llvmpipe (LLVM 3.9, 128 bits)
22:30imirkin_: OpenGL version string: 2.1 Mesa 13.0.4
22:31imirkin_: that might be a problem.
22:31Lyude: oh wow, just noticed that
22:31Lyude: wasn't doing that before, huh
22:33imirkin_: sounds like you accidentally the whole thing
22:37phoenixz: Upgrade done.. xserver-xorg-video-nouveau 1:1.0.13+git20170208.1516d35b-0 amd64 X.Org X server -- Nouveau display driver
22:38phoenixz: Going to reboot, crossing fingers...
22:42KsChoice: ... well, rebooted, and now I see only one card again :')
22:44phoenixz: ... well, rebooted, and now I see only one card again :')
22:44phoenixz: imirkin: I know you're busy, but if you have some time later.. Again I'm back @ /dev/dri/card1: failed to set DRM interface version 1.4: Permission denied
22:47KsChoice: I already tried booting with nosplash @ grub time, but that just gave me a black screen again
22:48KsChoice: I already tried booting with nosplash @ grub time, but that just gave me a black screen again
22:48phoenixz: I already tried booting with nosplash @ grub time, but that just gave me a black screen again
23:04phoenixz: I'll see if I can revert this update, because it's taking me back to square 1
23:09phoenixz: Meh, that too is off the table, that will get me in dependancy limbo
23:19Lyude: imirkin_: got mesa to show the extension as supported finally :)
23:28Lyude: How would I go about listing all of the tests for piglit?
23:32Lyude: ah nvm, think I figured out what I needed
23:42Horizon_Brave: annd i'm back
23:46phoenixz: imirkin_: Is there perhaps a posibility that all this mess would work better with wayland? I'm reaching, I know..
23:46Horizon_Brave: phoenixz: lol holy crap, are you guys still troubleshooting the same graphics card problem from earlier?
23:47phoenixz: Horizon_Brave: yeah.. Not only that, I'm back to square one again :( This all started with a /dev/dri/card1: failed to set DRM interface version 1.4: Permission denied error.. LAst thing I had like an hour ago was that there was this incompatibility issue with nouveau and X, so I upgraded both from a PPA et voila, now I have this error again from yesterday..
23:47phoenixz: NO idea how to solve it
23:47imirkin_: phoenixz: i've never used wayland, nor do i have plans on doing so anytime soon
23:48phoenixz: Yesterday I basically "solved" it by reinstalling my computer.. Since each reinstall is like 4 hours on itself, I'm really not looking forward to that.. Plus, that would just bring me back to the point of that Xorg / nouveau compatibility issue
23:48imirkin_: phoenixz: if you don't mind my commenting on your approach... you appear to be banging randomly on the keyboard until something good happens. i've found it's best to try to understand the problem and take steps towards resolving it.
23:48phoenixz: Horizon_Brave: so if you have any ideas or suggestions, I"m all ear :(
23:49Horizon_Brave: imirkin_: isn't X going to eventually be deprecated so you'll be forced to join the Wayland homgemony? lol
23:50Horizon_Brave: phoenixz: are you back to using all three of your 280's? or do you still have 1 of them disabled?
23:51phoenixz: imirkin_: Well not entirely.. ITs more that every solution to the problem gets me in another problem.. I had the permission denied issue, we tried solving that but required new kernel. Installed new kernel, system died.. I reinstalled everything, sytem works, but slow. So I added that pstate=1, we found its the 3 card issue. I remove one card, add DVI cable, now I have the incompatibility issue. Now I upgrade xorg and nouveau and now I am back
23:51phoenixz: to that access denied issue again
23:51phoenixz: Its just, every step I take towards what looks like a solution, gets me into another problem
23:52phoenixz: AFAIK, if I can resolve the access denied issue, all should work okay..
23:53phoenixz: Well, that is, unless I run into yet another issue again :)
23:54Horizon_Brave: access denied? is your sudeors file looking okay? polkit issue?
23:55imirkin_: phoenixz: these are all completely unrelated issues, but you're treating them as one
23:55gnarface: Horizon_Brave: funny thing about deprecation. it's a fictitious state requiring basically unanimous global consent. just having a competing project's maintainers call another project deprecated doesn't actually make it so. (see for reference: sendmail, oss, snort... the list goes on)
23:55imirkin_: Horizon_Brave: perhaps some day, but my belief is that that day is at least 10 years out, but hopefully more.
23:56Horizon_Brave: gnarface: heh, true! but if the project owner himself admits that "Okay...Wayland is just all around better for everyone, don't use X anymore" wouldn't that be valid?
23:56imirkin_: Horizon_Brave: there's not a single official project owner of anything
23:56gnarface: Horizon_Brave: maybe, if he wasn't the forked leader of a forked project that had already been forked at least once before...
23:57gnarface: Horizon_Brave: fact of the matter is nobody is going anywhere without Nvidia's permission, period.
23:57gnarface: Horizon_Brave: (despite that it angers me to have to admit it)
23:58Horizon_Brave: does nVidia have it's teeth in the shaping of the development or adoption of Wayland and Xorg?
23:58phoenixz: imirkin_: well.. I know they are unrelated, its just that they pop up, and I'll have to resolve them somehow.. I mean, the driver incompatibility, required an upgrade, so I dod that.. which got me here, now I have the access denied issue, which I think again is plymouth.. I know its not related, but its here, and it needs to be resolved somehow so..
23:58gnarface: Horizon_Brave: because they control user loyalty on hardware purchases
23:58gnarface: Horizon_Brave: (and because they won't support Wayland)
23:59Horizon_Brave: but imirkin_ you don't support or like wayland either right?
23:59phoenixz: IMHO, I cannot wait for wayland to come, I have a lot of hopes that it will resolve a lot of the issues X is having currently.. 10 years ago I built a 5 user machine, 5 GFX cards, 5 usb keyboards, mice, and audio cards.. Took a huge amount of configuring and messing with X and drivers, but we did it (IIRC we used the official nvidia driver back then).. But looking at the current state of affairs with multimonitors, I feel things have gotten worse
23:59phoenixz: instead of better..