[00:46] <Sarvatt> jcristau: all the nvidia drivers in sid except 96.xx work with it, but yeah fglrx is at least a month behind
[00:47] <Sarvatt> not sure how the blobs work in debian though, in ubuntu someone has to add the new abi and reupload them but the drivers actually do work
[06:53] <jcristau> Sarvatt: all drivers in sid look like they already have xorg-video-abi-15 in their list of alternative dependencies (except 96xx)
[06:53] <jcristau> but 96xx already doesn't support 1.14
[06:54] <jcristau> s/all/& nvidia/
[16:40] <jcristau> Prf_Jakob: is there a plan for a vmware ddx release some time soon?
[16:40] <Prf_Jakob> jcristau: there is a plan yes
[16:40] <Prf_Jakob> jcristau: just need QE to sign off.
[16:40] <jcristau> ok, thanks
[17:23] <alkisg> Hi, is vesafb.ko missing in Trusty? Should udev-fallback-graphics get updated to `modprobe uvesa`?
[17:23] <alkisg> $ grep vesa /etc/init/udev-fallback-graphics.conf 
[17:23] <alkisg>         modprobe -q -b vesafb
[17:23] <alkisg> $ ls /lib/modules/*/kernel/drivers/video/*vesa*
[17:23] <alkisg>  /lib/modules/3.13.0-5-generic/kernel/drivers/video/uvesafb.ko
[17:43] <alkisg> Ah, and v86d is not installed either... from its description and the output of `dmesg`, I'm guessing it's needed...
[17:55] <alkisg> Anyway, I think that the fallback graphics logic has been broken since 12.04
[17:56] <mlankhorst> uvesa is not meant to be used
[18:03] <alkisg> In the 10.04 days, when the "appropriate driver" failed to initialize xorg, a switch to xvesa happened, and a dialog was displayed about starting a session with fallback graphics etc
[18:03] <alkisg> This doesn't work in 12.04, 14.04 etc, xorg just restarts continuously...
[18:14] <tjaalton> it never used vesafb
[18:15] <mlankhorst> indeed
[18:17] <alkisg> Then I don't know what that "udev-fallback-graphics.conf" does, but the symptoms that I see are still there, i.e. no fallback graphics probably since we started using KMS.
[18:18] <alkisg> And someone should remove that udev-fallback-graphics service since vesafb isn't being shipped, it's just producing annoying messages in syslog...
[18:22] <tjaalton> on what hw??
[18:22] <tjaalton> -?
[18:23] <tjaalton> if kms is used but it doesn't work, there's no way to fallback to anything aiui
[18:43] <alkisg> Aaah ok that might be the reason then
[18:44] <alkisg> I've had reports from many schools here, I don't remember the graphics cards but I can note them down when needed, I think I remember some older sis ones...
[18:48] <tjaalton> they don't have kms
[18:50] <alkisg> Then maybe there's no fallback graphics logic anymore :)
[18:50] <tjaalton> sure is
[18:50] <tjaalton> if the xserver crashes
[18:50] <tjaalton> it'll get loaded with a fallback conf
[18:51] <alkisg> Hmm maybe we somehow accidentaly block that in LTSP, I haven't seen it since 10.04
[18:52] <tjaalton> i mean if the xserver crashes when starting with default (no) conf
[18:52] <alkisg> In LTSP by default we don't use a xorg.conf. But we don't use lightdm etc either, we have our own DM called ldm, and X is launched by that
[18:53] <alkisg> We have about 500 schools here, I haven't seen the fallback graphics triggered in years...
[18:53] <alkisg> Do you happen to remember offhand what upstart job starts the fallback graphics?
[18:53] <alkisg> Or is that xorg?
[18:55] <tjaalton> failsafe-x.conf
[18:56] <alkisg> Thank you, reading...
[18:57] <tjaalton> i'd say it's a good thing you never saw it..
[18:58] <alkisg> I did see black screens that we had to work around with XSERVER=vesa though :)
[18:58] <alkisg> Lots of them
[18:58] <tjaalton> crappy sis driver then
[18:58] <tjaalton> it can't detect if you get no output
[18:58] <alkisg> It's not only sis... we have a lot of old cards, a lot of xorg crashes etc
[18:58] <tjaalton> or crappy drivers in general
[18:59] <alkisg> The bad thing is that they worked fine up until 10.04, and they started having serious issues since 12.04
[18:59] <alkisg> And they're very very slow in gtk too *now*, maybe because of the switch to cairo instead of gdk...
[18:59] <alkisg> Like, needing 0.5 sec to draw one menu item in gnome-panel...
[19:00] <tjaalton> no exa support then
[19:00] <alkisg> 10 times slower than 10.04, at least
[19:00] <tjaalton> which 12.04?
[19:00] <alkisg> I've seen that with sis and older nvidia cards (pre-geforce), I haven't yet pinpointed it exactly
[19:00] <alkisg> We use gnome-fallback with metacity
[19:01] <tjaalton> .{1,2,3}
[19:01] <alkisg> 3
[19:01] <alkisg> QAh
[19:01] <alkisg> Ah
[19:01] <alkisg> I've tried with 3.2 kernel and xorg, and with the raring one
[19:01] <tjaalton> .3 has the raring stack
[19:01] <alkisg> sis crashes xorg with the saucy one
[19:01] <tjaalton> and kernel
[19:01] <alkisg> *raring
[19:02] <alkisg> But the issue was there in both of the kernels/xorg combinations...
[19:02] <tjaalton> so it was already 1.13 that dropped exa
[19:02] <tjaalton> err
[19:02] <tjaalton> not exa
[19:02] <tjaalton> the other one
[19:02] <alkisg> uxa?
[19:02] <tjaalton> no
[19:02] <alkisg> OK I got what you mean, I also don't remember it :)
[19:03] <alkisg> I tried gtkperf and I saw very big variations in some of its tests
[19:03] <alkisg> Like, two PCs having similar times everywhere, but in 2 tests, the first pc would need 50 seconds and the other one 5 seconds...
[19:03] <alkisg> *xaa
[19:04] <tjaalton> yup
[19:04] <alkisg> And I thought it was because of some functions using software instead of hw acceleration, in some drivers...
[19:05] <alkisg> (and those software implementations being very bad... :-/)
[19:07] <tjaalton> probably that too
[19:07] <tjaalton> in other parts
[19:08] <alkisg> Would it be possible to run a very old xorg in a recent ubuntu?
[19:09] <alkisg> E.g. 1.7.6 from Lucid, to Trusty?
[19:13] <alkisg> start on stopped lightdm EXIT_STATUS=[!0] or stopped gdm EXIT_STATUS=[!0]
[19:13] <alkisg> ...that's why it never gets triggered in LTSP, since we're using LDM...
[19:14] <alkisg> I'll check if we can emit a "stopped lightdm" signal even if we don't have lightdm installed...
[19:14] <tjaalton> hehe
[19:14] <tjaalton> it might be possible to forward-port, yes
[19:15] <tjaalton> and rebuild the drivers
[19:20] <alkisg> brb, my i915 hangs every now and then in trusty... :-/
[19:21] <tjaalton> that's progres..
[19:21] <tjaalton> +s
[22:46] <bryceh_> Sarvatt, done any displayport troubleshooting recently?  does this Xorg.0.log jog any ideas?
[22:46] <bryceh_> http://paste.ubuntu.com/6840625/
[22:47] <bryceh_> user has a usb-to-hdmi adapter to run an external monitor
[22:47] <bryceh_> adapter is http://www.amazon.com/Sabrent-1920x1080-1600x1200-DisplayLink-USB-HDMI/dp/B005UKG4KU/ref=cm_cr_pr_product_top
[22:47] <bryceh_> er s/displayport/displaylink/ 
[22:48] <bryceh_> it seems the kernel driver is not exposing the EDID from the monitor
[22:49] <bryceh_> dmesg & xrandr:  http://paste.ubuntu.com/6840629/  http://paste.ubuntu.com/6840630/
[22:49] <bryceh_> the monitor lights up with VESA modes okay
[22:50] <bryceh_> we tried guessing a modeline to force it to 1920x1080, and it seems to work, but screen stays black
 DVI-1-0 connected 1920x1080+1920+0 (0xfa) normal (normal left inverted right x axis y axis) 0mm x 0mm
         Identifier: 0xf3
         Timestamp:  1331890
         Subpixel:   unknown
         Gamma:      1.0:1.0:1.0
         Brightness: 1.0
         Clones:
         CRTC:       4
         CRTCs:      4
         Transform:  1.000000 0.000000 0.000000
[22:51] <bryceh_> I also had him try connecting to other CRTCs but he just gets errors for anything other than 4
[23:04] <RAOF> bryceh_: Can you get the kernel to fake a 1920x1080 EDID for it?
[23:05] <RAOF> That would be something like drm_kms_helper.edid_firmware=DVI-1-0:edid/1920x1080.bin
[23:08] <bryceh_> RAOF, I can give that a try, but shouldn't a forced modeline work around that?
[23:09] <bryceh_> I'm going to have him connect the monitor to a known good setup to collect the correct edid and modeline
[23:09] <RAOF> bryceh_: Depends; My LCD display with broken EDID hardware doesn't get detected as *connected* unless I force an edid.
[23:09] <RAOF> So a modeline doesn't help for my display, but faking an edid does.
[23:09] <bryceh_> hmm, ok
[23:10] <bryceh_> one other weird thing is that he assures me this setup had been working up until recently, without need for modeline forcing