The latest driver update broke my hithero stable X11 setup with multiple { adapters | (separate) screens } . Reproducible: Always Steps to Reproduce: 1. upgrade from nvidia-drivers-355.11-r2:0/355 to -358.09:0/358 2. reboot 3. start X with {twm|fluxbox|xfce4} Actual Results: :: Only the first screen (attached to Intel Xeon) gets available, :: The other three (attached to NVidia ) stay blank. Expected Results: 4. mask latest :0/358 driver 5. re-emerge -355.11-r2:0/355 6. reboot 7. start X with { twm | fluxbox | xfce4 } :: All four screens are available again. Switch to :0/358 ==> FAILURE again. Switch to :0/355 ==> WORKING again. lspci | grep VGA 00:02.0 VGA compatible controller: Intel Corporation Xeon E3-1200 v3 Processor Integrated Graphics Controller (rev 06) 01:00.0 VGA compatible controller: NVIDIA Corporation GK104 [GeForce GTX 660 Ti] (rev a1) Xeon (integrated) : Intel HD Graphics P4600, 350-1250MHz NVidia ( pci ) : PNY GTX 660 Ti OC2 equery list x11-base/* [IP-] [ ] x11-base/xorg-drivers-1.17:0 [IP-] [ ] x11-base/xorg-server-1.17.2-r2:0/1.17.2 [IP-] [ ] x11-base/xorg-x11-7.4-r2:0 # equery list -p nvidia-drivers ... [IP-] [ ] x11-drivers/nvidia-drivers-355.11-r2:0/355 [-P-] [M ] x11-drivers/nvidia-drivers-358.09:0/358
Created attachment 414612 [details] emerge --info nvidia-drivers > ei_nv
Created attachment 414614 [details] Xorg.0.log from WORKING :0/355 Notabene: In my case, ... (EE) NVIDIA(1): Failed to initialize the GLX module is no problem at all but accepted and o.k. because my primary screen is attached to the Intel HD Graphics.
Created attachment 414616 [details] Xorg.0.log__ERROR__0-358 $ grep WW Xorg.0.log* ... (WW) NVIDIA(1): Unable to get display device for DPI computation. ... (WW) NVIDIA(2): Unable to get display device for DPI computation. ... (WW) NVIDIA(3): Unable to get display device for DPI computation.
Comment on attachment 414612 [details] emerge --info nvidia-drivers > ei_nv That info is for the wrong version
Created attachment 414618 [details] etc_X11_xorg.conf.d.tar Multiple HW && Multiple separate X Screens ( intel | nvidia | (nouveau) )
1) This is a public beta of Nvidia's proprietary code. Unless it's Gentoo's configuration that is causing issues, reporting problems here makes little sense. 2) You should report bugs upstream. 3) Is USE=kms?
(In reply to Jeroen Roovers from comment #4) > Comment on attachment 414612 [details] > emerge --info nvidia-drivers > ei_nv > > That info is for the wrong version Jeroen, The only diff is drivers version - I would have to re-do the whole hazzle again; If you _really_ _need_ it , give me a note and I will supply it later.
(In reply to Manfred Knick from comment #7) > (In reply to Jeroen Roovers from comment #4) > > Comment on attachment 414612 [details] > > emerge --info nvidia-drivers > ei_nv > > > > That info is for the wrong version > > Jeroen, > > The only diff is drivers version - > I would have to re-do the whole hazzle again; > If you _really_ _need_ it , > give me a note and I will supply it later. emerge -vp =x11-drivers/nvidia-drivers-358.09 If you find properly reporting bugs is a hassle, then why do you bother at all? :)
(In reply to Jeroen Roovers from comment #6) > 1) This is a public beta of Nvidia's proprietary code. > Unless it's Gentoo's configuration that is causing issues, I have to apologize: Overlooked a typo in my package.keywords which unintendely unmasked (~) -358.09 . > reporting problems here makes little sense. ? Saving other colleagues's time ? > 2) You should report bugs upstream. Yes - planned for later. > 3) Is USE=kms? No.
(In reply to Jeroen Roovers from comment #8) > emerge -vp =x11-drivers/nvidia-drivers-358.09 [ebuild U ~] x11-drivers/nvidia-drivers-358.09:0/358::gentoo [355.11-r2:0/355::gentoo] USE="X acpi -gtk2 -gtk3 -kms% -multilib -pax_kernel -tools -uvm" 0 KiB
(In reply to Jeroen Roovers from comment #8) > If you find properly reporting bugs is a hassle, then why do you bother at > all? :) Beg your pardon ?!
Comment on attachment 414616 [details] Xorg.0.log__ERROR__0-358 TYPO: corfrection: ... "Xorg.0.log from ERROR :0/358"
Could you try with USE=kms. I'm still in two minds about the modesetting driver being optional at all here.
Preliminary *Cross-Reference* to upstream : http://nvidia-submit.custhelp.com/app/account/questions/detail/i_id/861541
Created attachment 414632 [details] Xorg.0.log WORKING :0/358 with use-flag kms enabled
Created attachment 414634 [details] emerge --info nvidia-drivers for WORKING :0/358 with use-flag kms enabled
(In reply to Jeroen Roovers from comment #13) > Could you try with USE=kms. I'm still in two minds about the modesetting > driver being optional at all here. With USE=kms enabled, I get all four screens again. At first sight, loading applications seems to work on all of them. I'm using XFCE4 4.12 right how. # emerge -p -v x11-drivers/nvidia-drivers ... [ebuild R ~] x11-drivers/nvidia-drivers-358.09:0/358::gentoo USE="X acpi kms -gtk2 -gtk3 -multilib -pax_kernel -tools -uvm" # equery list x11-drivers/nvidia-drivers ... [IP-] [ ] x11-drivers/nvidia-drivers-358.09:0/358
(In reply to Jeroen Roovers from comment #13) > ... I'm still in two minds about the modesetting > driver being optional at all here. @ Jeroen : Another observation from the erroneous situation before which might further substantiate the need for +kms : The mouse was (visible and) blocked correctly (no way out LEFT | DOWN) on the first monitor connected to the Intel, but seemingly could move out to the three black monitors ( RIGHT | ABOVE ) connected to the NVidia devices, naturally giving no visible cursor icon there.
@ Jeroen : I would not mind marking this Bug as "RESOLVED : WORKSFORME". But perhaps we wait some days in order to avoid DUPLICATEs ? It's up to you. Thanks. Manfred
/usr/portage/x11-drivers/nvidia-drivers/nvidia-drivers-358.09.ebuild : IUSE="acpi multilib kernel_FreeBSD kernel_linux kms pax_kernel +tools gtk2 gtk3 +X uvm" *Proposal* : Change "kms" to "+kms" for -358.09-r1 . My current settings: (quite minimalistic) # equery uses nvidia-drivers ... + + X : Install the X.org driver, OpenGL libraries, XvMC libraries, and VDPAU libraries + + acpi : Add support for Advanced Configuration and Power Interface - - gtk2 : Install nvidia-settings with support for GTK+ 2 - - gtk3 : Install nvidia-settings with support for GTK+ 3 + + kms : Enable support for kernel mode setting (KMS) - - multilib : On 64bit systems, if you want to be able to compile 32bit and 64bit binaries - - pax_kernel : PaX patches from the PaX project - - tools : Install additional tools such as nvidia-settings - - uvm : Install the Unified Memory kernel module (nvidia-uvm) for sharing memory between CPU and GPU in CUDA programs
That's good news. Thanks for looking into this. I guess I'll switch to IUSE=+kms next and then watch the fallout from that change.
If somebody now finds a bug with USE=kms, then that should go to a new bug report.
(In reply to Jeroen Roovers from comment #21) > ... I guess I'll switch to > IUSE=+kms next and then watch the fallout from that change. (In reply to Jeroen Roovers from comment #22) > If somebody now finds a bug with USE=kms, > then that should go to a new bug report. Perfect. Thanks for your suggestion in #c13. Updated *upstream* : ... http://nvidia-submit.custhelp.com/app/account/questions/detail/i_id/861541