02:19 mupuf: skeggsb: oh...
07:56 nv7300: hello
07:56 nv7300: i get an black screen when i try to boot every modern linux live system
07:57 nv7300: I have an nvidia 7300 card. The card is normaly working totaly fine
07:58 nv7300: at the moment i had to boot with "nomodeset nouveau.modeset=0". Now VESA is running. Not nouveau
07:59 nv7300: any developer here?
08:00 imirkin: pastebin dmesg when booting without turning off modesetting
08:00 nv7300: i would like to help fixing this error
08:01 nv7300: imirkin: have you an idea why now VESA is running and not nouveau? I just want to disable KMS.
08:01 imirkin: no kms means no nouveau
08:02 imirkin: vesa is the next fallback on the list... after that is probably vga
08:02 nv7300: imirkin: so also no kms with an intel card mean no intel driver and so on?
08:02 imirkin: mmmm... iirc intel still supports some form of UMS
08:03 imirkin: nouveau has the luxury of not having legacy things to worry about
08:03 imirkin: it was only merged in 2.6.34 or so
08:03 imirkin: although there have been talks of nuking UMS in intel too. not sure what the status of that is tbh
08:05 nv7300: imirkin: okay. thanks for the information. Did you have an idea how i can boot an live system on one pc with such a problem to check dmesg? Its have to be live-system and there is no other pc i could ssh into the live-system.
08:05 imirkin: no smart phone?
08:06 nv7300: no wlan
08:06 imirkin: :(
08:06 imirkin: you could.................. hm
08:06 nv7300: no usb-network-host-adapter here
08:06 imirkin: oh, well it should be in the logs
08:06 imirkin: just have to find it... search for 'nouveau ['
08:07 nv7300: after grub i get quickly "no signal" from the monitor
08:07 imirkin: that should find a boot with nouveau in it
08:07 imirkin: and then just expand up/down to the surrounding boots
08:08 imirkin: it's not an onboard chipset of some sort with a DVI connector, right?
08:08 imirkin: that was GeForce 6150SE iirc though
08:08 nv7300: i have to press reset-button on the pc to get an working screen.
08:09 nv7300: its an nvidia 7300 with DVI monitor connected to (1920x1200)
08:10 imirkin: look through your logs for previous boots... they should be in there
08:10 imirkin: like /var/log/messages
08:10 nv7300: you forgot: Live-system. There are no logs after pressing reset-button
08:10 imirkin: oh, it's a livecd? didn't understand that bit.
08:11 nv7300: [16:05] <nv7300> imirkin: okay. thanks for the information. Did you have an idea how i can boot an live system on one pc with such a problem to check dmesg? Its have to be live-system and there is no other pc i could ssh into the live-system.
08:11 imirkin: yeah, i didn't understand what you meant by 'live system' -- never heard it called that before
08:12 nv7300: ah, okay. how did the most people call it?
08:12 imirkin: livecd :)
08:12 imirkin: even though it might be liveusb or whatever else... livecd is still the term most often used
08:12 imirkin: anyways, i gtg
08:12 imirkin: without logs, debugging this will be impossible
08:12 imirkin: (even with logs it may prove very difficult)
08:13 nv7300: maybe i could borrow the gpu-card of the pc
08:13 nv7300: (regular pcie card)
08:13 nv7300: i should get the error in every pc
08:13 imirkin: yeah
08:14 imirkin: i'll be on later... will read scroll back
08:14 nv7300: imirkin: are you an dev? i could provide you an system you could ssh onto later
08:20 nv7300: @all: what kernel command line could i use to set manualy the resolution of the monitor for kms?
08:21 nv7300: simply: video=1280x1024 ?
08:22 nv7300: lets test. rebooting
08:23 robclark: nv7300, http://nouveau.freedesktop.org/wiki/KernelModeSetting/ .. scroll down to bottom...
08:24 robclark: video=<connectorname>:1280x1024 ...
12:41 RSpliet: wow, talking about code collecting mould like that 3 week old slice of pizza in the fridge
12:42 RSpliet: how did this "break HWSQ" code happen :-S
12:44 RSpliet: oh... I bet I make it overflow
12:50 imirkin_: ?
12:51 RSpliet: imirkin_: sorry, I'm complaining out loud aren't I :p
12:51 RSpliet: struct { u8 data[512]; u8 size; } c;
12:52 RSpliet: ^ guess what happens if size exceeds (or equals) 512 and you go writing there
12:53 imirkin_: don't do that? :)
12:54 RSpliet: imirkin_: easier said than done :p
12:54 imirkin_: also size is _very_ unlikely to exceed or equal 512
12:54 imirkin_: perhaps you meant 255?
12:55 RSpliet: that's a fair point
12:56 RSpliet: interestingly enough, size is supposed to hold a value of up to 512. Let's make that u16 :-)
12:57 imirkin_: no u9? :)
12:58 RSpliet: I can't imagine an arch that would benefit from that :p
12:58 glennk: fpga?
12:58 imirkin_: iirc pdp's had 18-bit words
12:58 RSpliet: sorry, let me rephrase that
12:58 glennk: or the video encode/decode blocks on various gpus
12:59 RSpliet: I can't imagine a sensible and well-used arch that will ever have an NVIDIA GPU attached to benefit from that
13:00 glennk: having size after data looks cache unfriendly
13:00 RSpliet: and overflow unfriendly :D
13:00 glennk: well, before is underflow unfriendly, but thats not quite as often
13:01 imirkin_: well, pdp's were certainly well-used, and, at the time, sensible. so if i can figure out how to attach a PCI bus to one, i'm golden
13:01 buhman: imirkin_: dewit
13:01 glennk: i've seen pdp-10 done on a fpga
13:02 imirkin_: pdp-10 had 36-bit words. looks like i'd need a pdp-9 with 18-bit words :)
13:03 glennk: pdp-11 also had that "awesome" pdp-endian word layout
13:06 buhman: imirkin_: can you make me a kepler using this manufacturing process: https://upload.wikimedia.org/wikipedia/commons/0/0f/KL10-backplane.jpg ?
13:07 buhman: pcb and integrated circuits not allowed; needs to be 100% wire wrap
13:07 glennk: i don't think there's enough copper mined annually for the wire needed for such a project
13:08 buhman: heh
14:19 zolo: somebody have archived conversation from 4.3.2015?
14:20 zolo: at this channel
14:22 imirkin_: see topic
14:27 zolo: sorry :D
14:31 zolo: but i have problem. All conversation i had here is not in the log.
14:31 zolo: it was 015-03-04
14:31 zolo: *2015-03-04
14:31 tobijk: inspect the channelname :>
14:33 zolo: im dumb...
14:33 zolo: but i cant find logs of this room, where is it?
14:33 tobijk: channel=nouveau
14:34 zolo: i found it... thanks for help bye :)
15:33 susanne: I'm running a 3.19 kernel with a GeForce 9300M GS.I got acpi_video0 and nv_backlight, but only acpi_video0 changes the brightness.How can I disable nv_backlight?
15:34 susanne: (or get it working)
15:36 susanne: brb reboot
15:52 imirkin_: susanne: boot with video.use_native_backlight=0
15:52 susanne: imirkin_: did that
15:52 susanne: tried every possibility of modinfo -p video params
15:53 susanne: (sorry that I didn't mention that)
15:53 imirkin_: what's your full kernel cmdline?
15:53 susanne: (although I set it in modprobe.conf)
15:53 imirkin_: is it a module?
15:54 susanne: yes it is
15:54 imirkin_: and it gets loaded when the modprobe.conf is available?
15:54 imirkin_: (it's in your initrd of whatever)
15:56 susanne: it's not build into my init but get's loaded by it (so it should set the value) /sys/module/video/parameters/use_native_backlight is 0 also
15:59 susanne: anyway the last cmdline I tried was this BOOT_IMAGE=/boot/vmlinuz-linux root=UUID=eb1c98b5-e02e-49cc-bcc6-b39fdf391feb rw quiet acpi_osi=Linux acpi_backlight=vendor
15:59 susanne: so maybe it doesn't work right now because I set it to use acpi :P
16:00 susanne: (tried to many different things really :/ rebooted 1 time too much)
16:02 susanne: I think I will reboot now setting video.use_native_backlight=0 and not the acpi settings
16:07 imirkin_: ah, you want to drop acpi_backlight=vendor
16:07 imirkin_: that forces nouveau's which you don't want
16:07 imirkin_: also i'd recommend dropping acpi_osi=Linux -- that's more likely to cause issues than not
16:07 susanne_: imirkin_: yup did it and I think I solved it
16:08 susanne_: ok it doesn't really solve the problem with nv_backlight but it solved my underlying problem I wanted to solve
16:08 imirkin_: right, it won't make nv_backlight work
16:08 imirkin_: the way these things are all hooked up is very variable
16:08 imirkin_: linux tries its best :)
16:08 susanne_: yes but xfce4 now finally can control the backlight after I did a pkexec xfpm-power-backlight-helper --set-brightness-switch 1
16:10 susanne_: anyway I'm sick of rebooting, but I guess it will work now :) thx imirkin_
16:10 imirkin_: enjoy :)
16:11 susanne_: yeah I will be the happiest person the day I can use pstates ;)
16:11 imirkin_: G98?
16:11 susanne_: yeah
16:11 imirkin_: the one i have boots to its highest pstate
16:12 imirkin_: mobile one probably boots to middle or low though
16:13 susanne_: imirkin_: right but I enabled it a few times in my own kernel and it kinda worked
16:13 imirkin_: yeah, we killed it in 3.13
16:13 susanne_: depending on what values I was allowing to be set
16:13 imirkin_: RSpliet was going to look at restoring it...
16:13 imirkin_: (as was i)
16:16 susanne_: yeah I was focused on the card ones when this still was my main laptop, but now that I got a intel/ati hybrid card I don't really follow the development of nouveau anymore (sadly because it always made me happy)
16:17 susanne_: even the stupid phoronix guy always talks shit about the project.anyway thank you guys for your work, really appreciate it
16:17 imirkin_: haha, we don't pay too much attention to him :)
16:25 susanne_: anyway I'm sure to look back into this channel from time to time hopefully I can get the power consumption for this laptop down
16:26 susanne_: gn8
16:55 RSpliet: imirkin_: damn, so close :D
16:57 imirkin_: are you sending the same hwsq down as the blob?
16:58 RSpliet: well, I fixed some of the MR values
16:59 imirkin_: you looked at the traces i sent you too right?
16:59 RSpliet: got gnome shell started, it's just... there's no mouse cursor and slightly too much corruption to be useful :D
16:59 RSpliet: not yet
16:59 imirkin_: i guess it makes sense to just make it work on yours first :)
16:59 RSpliet: that's my approach
17:00 RSpliet: there's loads of useful vbios->reg mappings I already figured out that we didn't have before
17:00 imirkin_: awesome :)
17:01 RSpliet: and I picked the most annoying card in the world, so generalisation should be a breeze
17:01 RSpliet: (famous last words?)
17:01 imirkin_: hehehe
17:01 RSpliet: untrue, too... I guess a GDDR5 card would be even more painful... not sure if they exist though. I guess there should be some NVA0 with GDDR5 out in the wild
17:02 imirkin_: no GDDR5's
17:02 imirkin_: first and only one is that stupid GT 240
17:02 RSpliet: wow
17:02 imirkin_: maybe there were some in the 300 series that were also nva3's
17:02 RSpliet: I guess they made it just for the learning experience :D
17:02 RSpliet: yeah, or some quadro's
17:03 imirkin_: well, it was a quick "easy" way of upping perf
17:03 imirkin_: while waiting for the GF100
17:03 RSpliet: well, sure, but why do that with a silly mid-end card
17:04 RSpliet: (and I reckon adjusting a memory controller to support GDDR5 is not as easy as you make it sound :p)
17:04 imirkin_: just add quotes :)
17:05 RSpliet: anyway, ftr, this NVA0 has 7 memory partitions
17:05 imirkin_: i think the one i sent you a trace for had 8
17:05 RSpliet: I don't expect a lot of differences
17:06 imirkin_: you never do!
17:07 RSpliet: the timings are nearly the same
17:07 RSpliet: the memory parameters look remarkably similar from the top of my head
17:09 RSpliet: is that your card, or Martins?
17:10 imirkin_: martin's
17:10 imirkin_: all of those are his iirc
17:10 imirkin_: the traces come from reator
17:10 imirkin_: i wouldn't have had the pateience to do it on my own box
17:11 RSpliet: 8 partitions it is :-)
17:12 RSpliet: nothing noteworthy... I think if mine works, his should work too
17:12 RSpliet: heck, testing is the easiest way to find out :p
17:12 imirkin_: if :)
17:12 imirkin_: mupuf's gone for a while
17:12 RSpliet: is he?
17:13 imirkin_: yeah, for a week or something
17:13 RSpliet: chasing girls I bet
17:13 RSpliet: go him! ;-)
17:14 RSpliet: hope he can have some fun with it when he gets back
17:14 RSpliet: anyway, bed
17:14 RSpliet: night
19:01 tobijk: mh, we have a build breakage in nv50_ir_from_tgsi right now? or is it just me :/
19:04 imirkin: patch on mailing list
19:04 imirkin: thanks for reminding me... i said i'd push it out when i got home
19:04 imirkin: but then i forgot :)
19:05 tobijk: ah allright :)
19:10 tobijk: imirkin: fwiw, you can add my reviewed by :>
19:11 imirkin: too late... i'm already pushing
19:11 tobijk: no problem
19:22 imirkin: tobijk: all pushed
19:22 tobijk: allready got it, thx
19:59 sontek: Hey, I'm on Fedora 21 and it seems to have issues with the nouveau driver. Wondering if I can get some help here... When I first boot it says "init failed, -16"
20:00 sontek: I was able to set nomodeset in the grub conf and boot (not sure what it does) but it allows it to boot at least
20:01 phantomcircuit: im running debian wheezy on an msi GT60 which has a K3100M in an optimus configuration
20:02 phantomcircuit: when i boot there's a complaint about a table lookup
20:02 phantomcircuit: any ideas?
20:02 tobijk: pastebin dmesg?
20:03 phantomcircuit: i can take a picture in a minute
20:04 imirkin: sontek: pastebin dmesg
20:05 imirkin: phantomcircuit: those errors tend to be fairly harmless... iirc they're warnings too
20:05 imirkin: sontek: also 'lspci -d 10de: -nn'
20:05 phantomcircuit: heh well this is apparently fatal
20:05 sontek: imirkin: Working on it, for whatever reason once nomodeset is in the grub line so I can boot I have to login, wait for it to go to sleep, and then wake it up before I can use it
20:05 sontek: the first login doesn't redraw after login or something. Its weird
20:06 imirkin: sontek: the lspci should tell me which gpu you have...
20:12 sontek: 01:00.0 VGA compatible controller [0300]: NVIDIA Corporation GK106GLM [Quadro K2100M] [10de:11fc] (rev a1)
20:13 sontek: http://paste.ofcode.org/yeR3M3AJUHkfjpyzNiUbMH
20:13 sontek: imirkin: that is the dmesg ^
20:17 imirkin: sontek: that dmesg doesn't appear to contain the error you were referring to
20:17 imirkin: is this the HUB_INIT timeout error?
20:17 sontek: imirkin: sorry, it doesn't boot with the error I was mentioning
20:17 sontek: imirkin: I could take a picture of the screen with the error
20:18 sontek: Right now gnome-shell is taking up 400% of my CPU so something whacky is going on
20:20 imirkin: sontek: picture would work... i just want the lines above that "error -16" line
20:29 sontek: imirkin: http://oi59.tinypic.com/dx1opf.jpg
20:29 sontek: imirkin: You were right about HUB_INIT
20:30 imirkin: yeah.... it's an issue with some GK106's
20:30 imirkin: no known solution unfortunately... they boot up mega-off, we try to turn them on, but apparently we're missing some bit of it
20:30 imirkin: we were given instructions on how to do it by nvidia, but apparently that's insufficient, or we're messing up
20:31 sontek: Its weird because I had ubuntu 14.04 on here and didn't have any issue, it wasn't until I put on Fedora that I had issues
20:31 sontek: I wonder what version of nouveau ubuntu was using
20:31 imirkin: hmmm... perhaps doing the turn-back-on thing is breaking some boards?
20:31 imirkin: (coz we don't detect that it's off)
20:31 imirkin: if you're up for compiling your own kernel, i could provide a kernel edit
20:32 sontek: Yeah, I'm up for trying it, right now the laptop is running pretty janking because gnome-shell is running at 400% CPU, I'm guessing because of all the weird video stuff
20:32 imirkin: isn't this an optimus setup?
20:33 imirkin: you should just use your intel gpu
20:33 sontek: Yeah, the laptop supports both video cards
20:34 sontek: I don't know if going straight intel will work since I need the docking station to work
20:34 sontek: I should re-install ubuntu 14.04 to see what they had running, maybe they disabled nouveau by default or something
20:35 imirkin: anyways, in engine/device/nve0.c, look for the 0xe6 case, and replace device->oclass[NVDEV_SUBDEV_PWR ] = gk104_pwr_oclass with nvd0_pwr_oclass
20:35 sontek: Whats the best way to disable nouveau so that it will use intel by default?
20:36 imirkin: nouveau.modeset=0
20:37 sontek: I'll try the kernel mod once the laptop is stablize
20:38 sontek: hah. laptop runs perfectly with nouveau.modeset=0, so that was the cause of the gnomeshell 400% cpu as well
20:40 imirkin: no
20:40 imirkin: the cause is that you had nomodeset
20:40 imirkin: which _also_ turned off intel
20:41 sontek: Ahh
20:41 imirkin: and you were running gnome-shell on a probably very-high-resolution screen
20:41 imirkin: and using a software 3d renderer to do that
20:41 sontek: yeah, 2880x<something>
21:19 sontek: imirkin: ok, I think I've got the Fedora kernel source all pulled down and ready
21:20 sontek: device->oclass[NVDEV_SUBDEV_PWR ] = nvd0_pwr_oclass;
21:20 sontek: so you just want it to look like that?
21:22 imirkin: yeah, for the 0xe6 case
21:22 imirkin: (you have a GK106 right?)
21:23 sontek: Yeah GK106GLM
21:27 imirkin: that will disable the workaround we put in which made some cards work
21:27 imirkin: perhaps it's also causing some cards to break :)
21:40 sontek: Its compiling so I'll know soon enough
21:40 sontek: This card does work in Ubuntu 14.04, so something between whatever kernel version they use and the version Fedora 21 uses broke
21:41 imirkin: well, this fix is semi-recent
21:41 imirkin: so perhaps that did it... there's also some suggestion that there's some sort of timing issue
21:41 imirkin: although it has been rather inconclusive... messing with rcu grace periods and whatnot
21:42 sontek: Well, if someone who knows this stuff and lives in the bay area they can totally hack on my laptop to figure it out
21:43 imirkin: no, but i have a friend in the bay area with precisely this problem :)
21:43 imirkin: perhaps geography is a contributing factor?
21:45 sontek: Yeah, must be the different amount of UV we get here in California
21:59 sontek: Kernels take forever even with 8 cores =)
21:59 imirkin: esp if you do distro builds
22:00 imirkin: i mean who knows, maybe you will get an aic79xx that you'll have to boot off of with the kernel you build
22:01 sontek: haha
22:01 imirkin: (and i didn't even get into the ISA NCR controllers...)
22:01 sontek: I followed http://fedoraproject.org/wiki/Building_a_custom_kernel so hopefully I'll get what I already have with just this patch
22:01 imirkin: Documentation/scsi/ChangeLog.1992-1997: * g_NCR5380.c:
22:01 imirkin: heh