Hello,

I use a custom kernel and nvidia-dkms, but tested this with the regular kernel and nvidia package, and get precisely the same problem(s). As such, I'm going to try anything with DKMS and the custom kernel, so I don't need to be switching back and forth and rebooting all the time. After updating my system, I get a blank screen when I start the X.

Normally (before the update, and in my current state), I see the first few lines of output of startx, the screen turning on and off (because of modesetting), and then my .xinitrc getting executed.

However, with the latest Xorg and nVidia versions, I see the first few lines, and then X appears to freeze. I can switch to another tty and check that it executed my .xinitrc and my WM is running, but if I switch back, I just see a frozen image of whatever I was looking at before switching to the terminal running the X. In fact, if when it starts I do <Meta-Shift-Enter>touch x-works<Ctrl-D><Meta-Shift-Q> it does create a file called x-works and exit normally. A log of this is available as Xorg.18.log (attached).

I'm not very knowledgeable about the X, but it seems to me it is related to this part of the log: [ 5167.048] (II) modeset(1): using drv /dev/dri/card0 [ 5167.048] (EE) Screen 1 deleted because of no matching config section. [ 5167.048] (II) UnloadModule: "modesetting" while in Xorg.17.log (attached), it does its job.

Here's the combinations of up/down-grading I tried after I got the blank screen. The versions I had before the update were 355 and 17 and graphical acceleration did work, but trying now, it doesn't (tried with glxgears). Also, to downgrade xorg-server I also needed to downgrade xf86-input-evdev and xf86-video-intel. I always remembered to recompile the nvidia module with DKMS.

Xorg versions 1.17 1.18 nVidia 355 No accel nVidia complaint about X version, crash versions 358 OK Blank screen

$ lspci |grep 3D 04:00.0 3D controller: NVIDIA Corporation GM108M [GeForce 840M] (rev a2)

I have also mangled with nvidia-xconfig, but it only made things that were working not work, or had no visible effect. My xorg.conf is also attached.

I'm curious as to why the 358-1.17 combo works, and latest doesn't, and why the 355-1.17 combo used to work, and now works but glxgears doesn't work with it. Thank you in advance for any suggestion,

P.S.: Sorry that the attachments are in a tgz, but the original was rejected (over 40K).

João Miguel

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1

flow flow@nym.mixmin.net on Sun, 2015/11/08 16:34:

...

Is it possible to boot archlinux-2015.11.01-dual.iso (and install then) with grub?

For me the boot entry looks like this:

menuentry "Arch Linux ISO image" --class arch --class gnu-linux --class gnu --class os --class tool { insmod iso9660 loopback loop /archlinux/archlinux.iso linux (loop)/arch/boot/x86_64/vmlinuz archisobasedir=arch img_dev=/dev/sdaX img_loop=/archlinux/archlinux.iso archisodevice=/dev/loop0 copytoram=y quiet initrd (loop)/arch/boot/intel-ucode.img (loop)/arch/boot/x86_64/archiso.img }

Thank you for reply. Looks like this entry will boot.

menuentry "Arch" --class arch --class gnu-linux --class gnu --class os --class tool { search --no-floppy --fs-uuid --set=root e80c8dac-7444-4dbd-ad60-3d8e77bf1010 insmod iso9660 loopback loop /img/archlinux-2015.11.01-dual.iso linux (loop)/arch/boot/x86_64/vmlinuz archisobasedir=arch img_dev=/dev/sdb2 img_loop=/img/archlinux-2015.11.01-dual.iso archisodevice=/dev/loop0 copytoram=y quiet #initrd (loop)/arch/boot/intel-ucode.img (loop)/arch/boot/x86_64/archiso.img initrd (loop)/arch/boot/x86_64/archiso.img }

Is is possible to replace img_dev value with an UUID or label? - -- Have a nice day!

~~~ This PGP signature only certifies the sender and date of the message. It implies no approval from the administrators of nym.mixmin.net. Date: Mon Nov 30 23:12:15 2015 GMT From: flow@nym.mixmin.net