03:23arcol1: I have a laptop which I like very much (really good screen). It is equiped with nv17 video card. With newer kernels it freeze during boot like 9 out of 10 times. So starting the computer takes 2-3 hours, if I'm unlucky.
03:25arcol1: I tried everything I could, namely nouveau.noaccel=1, drm_kms_helper.edid_firmware=VGA-1, LVDS-1, nomodeset (which starts the computer), tried with systemd, upstart, also tried kernels 3.19, 4.1
03:26arcol1: when it freeze, the keyboard is frozen too, no ctrl-alt-del, no REISUB sysrq magic, nothing. On the screen the last dmesg is displayed, but the cursor is blinking.
03:28arcol1: I'm really out of ideas, what I could do. I used this laptop for years with ubuntu 9.10, 10.4, and worked like a charm (video playback was smooth). The sole reason of upgrading is because of lack of newer firefox and chrome/chromium. So can not use gmail.
03:29arcol1: the ubuntu 15.04 live cd mostly starts. Like 2 of 3 times. But once installed on harddrive, it mostly does not start (like 1 out of 10 times).
03:30arcol1: when the computer is booting, the screen flickers many times during the boot. Like 8 times or so.
03:30arcol1: when it does not boot, it does not flicker ones.
03:30arcol1: the liveCD sometimes start with an ubuntu logo with green artifacts around it, and when shutdown the whole screen is green background.
03:31arcol1: flicker ones -> flicker once
03:31arcol1: no idea how to debug further
03:32arcol1: maybe I could upgrade the VBIOS. But can not find a site with relevant firmware these days
03:34joi: are there any logs left from failed boots?
03:34joi: by logs I mean dmesg and xorg.log
03:35arcol1: I'll look it, the compromised laptop is in the booted state
03:36joi: if kernel dies because of oops/panic you could catch it with netconsole
03:38arcol1: I don't think Xorg.log is generated when it freezes. It freezes at the beginning of the boot stage. I looked into /var/log/Xorg.0.log and Xorg.0.log.old, and all the two are successful
03:39arcol1: I have /var/log/dmesg, dmesg.0, dmesg.1.gz, and 2,3,4.gz I'll browse through them
03:39joi: can I see dmesg from successful boot?
03:41arcol1: of course
03:41arcol1: I'll upload somewhere
03:41arcol1: I can even give you ssh access to the machine in question
03:41joi: to some stranger over internet? :D
03:42arcol1: its fresh install
03:42arcol1: nothing on it
03:42arcol1: its reinstalled at least 6 times
03:42joi: no, I don't need it :)
03:43arcol1: there is no kernel panic
03:43arcol1: the dmesg is simply stops
03:45arcol1: ^^ this is the dmesg (successful boot), although nouveau.noaccel=1 kernel parameter
03:45arcol1: it is more likely to boot with it
03:46joi: looks good
03:46joi: can you catch dmesg from failed boot with netconsole?
03:51arcol1_nx9105: sorry, wifi trouble
03:51arcol1_nx9105: joi, still here?
03:52joi: so, can you catch dmesg from failed boot with netconsole?
03:52arcol1_nx9105: what do I need for it?
03:53arcol1_nx9105: i have only wifi with that laptop (hp nx9105)
03:53joi: that's a problem
03:54joi: because netconsole relies on network card being enabled very early int the boot process
03:54arcol1_nx9105: so I need wired connection
03:55arcol1_nx9105: I need to lookup a netconsole tutorial. I may be able to do it. I'll grab a router from my neighboor. I have three laptop, so I will be able to fetch the message from one, while I'm in contact with you. If you are here to help me
03:55joi: or use something else - serial console
03:58joi: I might not be here later, but someone else can guide you what to do next
03:59arcol1_nx9105: I think I'll finish within an hour
03:59arcol1_nx9105: I'm reading this ^^
04:00joi: if netconsole won't find anything, I'd recommend finding the last kernel which works reliably and using git bisect to find exact commit which broke nouveau for you
04:01arcol1_nx9105: oh dear
04:01arcol1_nx9105: I think I never used nouveau with it before. I used the "nv" driver with 10.04 and 9.10
04:03arcol1_nx9105: maybe I should try to upgrade the VBIOS first, no?
04:04arcol1_nx9105: i read on the net, it is may be upgradeable
04:04arcol1_nx9105: or does it not help with nouveau?
04:06joi: never heard of anyone doing it and fixing boot with nouveau...
04:09glennk: nv17, thats a crusty old agp card afaik
04:09glennk: could try adding nouveau.agpmode=0 as boot parameter
04:09joi: yeah, but still supported by nouveau
04:24arcol_sam: joi: still here?
04:25arcol_sam: I set up netconsole between two laptop
04:25arcol_sam: I try to put it in grub config, and reboot to see what happens
04:25arcol_sam: now it works (rmmod, modprobe, and I see on the other laptop the output)
04:35arcol_sam: a little help needed
04:35arcol_sam: I loaded the netconsole module, when the machine started. It works, I can see message on the other machine. The source ip address is what ifconfig shows (192.168.1.7)
04:36arcol_sam: the question is, do I need to set the ip address manually in /etc/network/interfaces file, or is the ip is set by the kernel via the parameter?
04:37arcol_sam: email@example.com/eth0,firstname.lastname@example.org/ loglevel=7
04:37arcol_sam: ^^ this is the kernel parameter
04:37arcol_sam: (in grub: /etc/defaults/grub)
04:42joi: arcol_sam: I don't understadn the question... what's the problem?
04:44arcol_sam: 1. I started the computer in question. 2. I loaded the module with correct source IP address (looked at ifconfig output) --> WORKS
04:45arcol_sam: But if I give the same parameters to kernel command line -> does not work
04:45arcol_sam: I suspect I need to set up the ip address manually, because once it is started it is set by networkmanager I assume
04:45joi: you need to load netconsole before nouveau
04:45arcol_sam: which is not available at boot time
04:46joi: so you could blacklist nouveau on boot, load netconsole and manually load nouveau
04:46arcol_sam: ahh, maybe that is the problem, that there is not netconsole loaded in boot time
05:07arcol_sam: half success
05:07arcol_sam: at 146 sec, I can start seeing dmesg on the remote computer (nc)
05:08arcol_sam: the problem must be with network configuration
05:12arcol_sam: ok, was a typo in /etc/network/interfaces file (ifacte instead of iface)
05:19arcol_sam: joi: I think I configured netconsole, the best I could. The problem, it kicks in after 30 second of boot, but the freezing happens before when it happens
05:25joi: then boot with nouveau.modeset=0, after boot rmmod nouveau && modprobe (or insmod, don't remember which one) nouveau modeset=1
05:29arcol_sam: then I should boot without X, no?
05:33arcol_sam: googling ... (how to disable X.org booting with ubuntu and upstart)
05:46arcol_sam: I can not rmmod nouveau
05:46arcol_sam: Module nouveau is in use
05:49arcol_sam: I'll blacklist nouveau completely
05:49arcol_sam: and restart
05:50arcol_sam: the machine freeze even with modeset=0
05:53joi: nouveau.modeset=0 in the kernel commandline I presume?
05:55arcol_sam: at try 4, it booted into the text mode
05:55arcol_sam: here is the /proc/cmdline
05:56arcol_sam: BOOT_IMAGE=/boot/vmlinuz...-generic root=UUID=4ede...77 ro text nouveau.modeset=0 loglevel=7
05:56arcol_sam: (I typed it from the screen)
05:56arcol_sam: and I'm in the console (no X started)
05:57joi: so it seems... it's not nouveau
05:57arcol_sam: dmesg -n 8, and rmmod netconsole && modprobe netconsole, and the logs are appearing on the other computer
05:57arcol_sam: (tried with rmmod lp && modprobe lp)
05:58arcol_sam: so I try rmmod nouveau && modprobe nouveau modeset=1
05:58arcol_sam: it was successful
05:58arcol_sam: now I have smaller fonts
05:58arcol_sam: (on the screen)
05:59arcol_sam: and full of log on the other computer
05:59arcol_sam: maybe I could blacklist nouveau, and retry?
06:00joi: so you are chasing something else, nouveau.modeset=0 completely disables nouveau
06:00joi: it loads, but does nothing
06:01arcol_sam: I try blacklisting nouveau (it then use vesa, no?), and try to boot
06:01arcol_sam: into text mode
06:01arcol_sam: see if it hangs or not
06:02joi: wait a sec
06:04joi: when you said "at try 4, it booted into the text mode" it ment "I screwed up kernel command line 3x and it booted on 4th" or "I booted 4x with the same command line and it worked only on 4th try"?
06:04arcol_sam: the second
06:05arcol_sam: I booted 4 times, and the 4th was successful. The other 3 just hung
06:06joi: please pastebin output of lsmod
06:06arcol_sam: the problem with netconsole, that it does nothing, until I rmmod netconsole && modprobe netconsole
06:06arcol_sam: wait a sec, I'll grab a pendrive
06:10arcol_sam: ^^ lsmod after nouveau reloaded (1. booted into textmode with nouveau.modeset=0 2. rmmod nouveau && modprobe nouveau modeset=1 )
06:10arcol_sam: here is the dmesg: http://pastebin.com/hj9jTDH6
06:14arcol_sam: maybe it is the Bluetooth?
06:14arcol_sam: I never used it, but in dmesg it is inserted early on
06:14joi: I'm out of ideas, it doesn't look like a nouveau bug; your only option left is finding the last kernel which does not hang and use git bisect to find who broke it
06:15arcol_sam: maybe I could disable all the modules I'm not using
06:15arcol_sam: like bluetooth, firewire, parallel port
06:15joi: could be anything
06:15arcol_sam: just keep the minimum which is needed for the system
06:15arcol_sam: i don't use that either
06:17arcol_sam: [ 13.028631] netpoll: netconsole: device eth0 not up yet, forcing it [ 13.028729] 8139too 0000:02:01.0 eth0: link up, 100Mbps, full-duplex, lpa 0x45E1 [ 13.028825] netpoll: netconsole: carrier detect appears untrustworthy, waiting 4 seconds
06:17arcol_sam: [ 13.028631] netpoll: netconsole: device eth0 not up yet, forcing it
06:18arcol_sam: [ 13.028729] 8139too 0000:02:01.0 eth0: link up, 100Mbps, full-duplex, lpa 0x45E1
06:18arcol_sam: [ 13.028825] netpoll: netconsole: carrier detect appears untrustworthy, waiting 4 seconds
06:18arcol_sam: I don't see on the remote computer anything (before I log in and manually rmmod && modpobe netconsole)
06:19arcol_sam: a) maybe it does not use the eth0 interface
06:19arcol_sam: b) kernel loglevel is not in debug mode (somehow ignores loglevel=7 parameter in kernel boot cmdline)
06:29arcol_sam: I disabled everything I could
06:29arcol_sam: blacklist lp, parport-pc, parport, ppdev
06:29arcol_sam: bluetooth, bnep, btbcm, btusb, rfcomm, btintel
06:30arcol_sam: pcmcia, pcmcia_rsrc, yentasocket, pcmcia_core
06:30arcol_sam: 8139too, firewire_ohci, firewire_core, 8139cp, mii, crc_itu_t
06:30arcol_sam: finger crossed, and I reboot
06:30arcol_sam: stay tuned
06:43arcol_sam: 2 out of 2 successful boot
06:45arcol_sam: dunno why it stuck at "Waiting for network configuration..." there is absolutely no typo in /etc/network/interfaces
06:45arcol_sam: this systemd driving me crazy. :-\
06:45arcol_sam: or upstart, or whatever. Gimme back good'o init system
06:47arcol_sam: seems like it is solved. Thank you joi
06:48arcol_sam: also insisting it is not nouveau fault. I would give up otherwise
06:48arcol_sam: 4 out of 4 succesful boot
06:53joi: arcol_sam: so, which module hangs your system?
06:55arcol_sam: dunno, I blacklisted all the above :)
06:55arcol_sam: no need for bluetooth, firewire, paralel, joystick, pcmcia
06:55arcol_sam: I suspect the bluetooth
06:56arcol_sam: I'm still cleaning up (get rid of netconsole, textmode, etc)
06:56arcol_sam: are you interested?
06:57arcol_sam: I can reenable bluetooth, and see if it is the weak link
06:57arcol_sam: 6 successfull boot in a row
06:59arcol_sam: reenabled bluetooth
07:01arcol_sam: seems like bluetooth is fine
07:01arcol_sam: then pcmcia, parport, firewire left
07:01arcol_sam: or the combination of them
07:02arcol_sam: that is timeconsuming with no outcome. I think I call it a day
07:02joi: you don't need to reboot, just modprobe them one by one
07:16arcol_sam: joi: I need to reboot, because it is non-deterministic
07:17arcol_sam: It is *likely* to fail, but it does not fail all the time
07:17arcol_sam: looks like it is not the bluetooth
07:17arcol_sam: but I don't see the point. Even I find the relevant module, it has no use of others. It is a pretty much ancient laptop
07:17arcol_sam: use of others -> use for others
09:55Quintus_q: I’ve got a question regarding the nouveau kernel module. I’ve transformed an old laptop with a defunct display into a NAS machine, and the nouveau kernel module spams my log with this message a thousand times and more: nouveau E[ DRM] DDC responded, but no EDID for LVDS-1
09:55Quintus_q: How do I disable logging this?
09:56Quintus_q: To be clear: The machine is only accessed by SSH. There’s no point in using the display anyway, so I want to get rid of that message.
10:02imirkin: Quintus_q: just disable nouveau entirely... nouveau.modeset=0
10:03Quintus_q: imirkin: Ok, I’ll try that. Thank you :-)
10:03yann-kaelig: I upgrade to libdrm 2.4.61 and the freeze disappear
10:03yann-kaelig: for nouveau driver
10:03imirkin: yann-kaelig: yeah, that makes sense. 2.4.60 was broken for nouveau.
10:05yann-kaelig: imirkin: ok, so I can report this to funtoo OS.
13:58yann-kaelig: if I open a video file with right click -> open with mpv player, my OS freeze
14:11Karlton: yann-kaelig: so your still freezing with libdrm 2.4.61?
14:12Karlton: or is that unrelated
14:13yann-kaelig: Karlton: two things, I'm freezing if I start reading a file with right click on it an open with mpv player, but I don't try with command line and put the right option for video output, and something wrong with the flashplayer with mozilla, some distortion started. I removed the flahplayer, try again with youtube, no problem
14:18yann-kaelig: What i'm going to do is to compile a normal kernel with nouveau, to see if it's not a problem with the FullRT patch
14:21Karlton: yann-kaelig: I never had problems with mpv before, but I know lots of people have problems with flash player regardless of what video driver they use
14:23Karlton: mpv can play youtube videos now aswell if have ffmpeg compiled with network support and have youtube-dl installed
14:47RSpliet: pmoreau: mind giving my kernel tree another spin?
14:48RSpliet: I made some changes to timing calc
14:48RSpliet: should be a perfect match now... I think