I had an install of Leap 15.1 running fine until a problem in another “sid” distro seemed to wipe out access to a number of other systems, including my Leap 15.1 install. Yesterday I used the same flash installer that I possibly used in April to run a fresh install, and other than its complaining about the EFI partition being “less than 300MB” the install appeared to go well, including downloading updates during the install.
Other thing that was perhaps “unusual” was that I had to “accept” or reject the package “mesa-dri-nouveau”?? which included some warnings about it “being unstable” . . . which I went forward with. The computer has an Nvidia graphics card, and historically the “mesa” nouveau package was considered “good” . . . .
Anyway on reboot, the Grub window loaded and I selected Leap and it went to the black screen, there was a few lines starting with “[OK] Reached target Basic system” and then the line “3.9643597]fb: switching to nouveaufb from EFI VGA” . . . and then it “freezes.”
Shut it down with the power button, then worked my way back to the Leap “Rescue System” . . . which got to the “rescue login” line, into which I entered the new[old] user name that I had used in the previous install, again entered in the installer user identification lines . . . and the password, which it didn’t accept, tried a few other options, none of them were accepted???
So, question is, is this something relating to the “mesa-dri-nouveau” package, which isn’t seeming to “work”? I tried to get into a TTY to try to run ref/dup . . . but that didn’t work . . . or, is there some “universal” OpenSUSE Rescue login ID & password that has to be used, rather than my own user name and password?
Or, is this a “botched” install, and only a nuke/pave will be the “fix”???
I don’t think this is the NET installer, this was a “4 - 5GB” “installer” . . . from several months back, but then I think it asked me if I wanted to “download updates” via the “web” . . . which I’ve done previously . . . and I said Yes.
But, no, I don’t think I can get to a terminal, I tried several ways, but it “stalls” at the “switching to nouveaufb” line and won’t go past that point. So, OK, it seems like that “nouveau” package is broken, and likely I would have to try another install, and “don’t accept ‘mesa-dri-nouveau’ package” when the time comes; and/or don’t accept the “download latest updates” during the install . . . ???
But, if I don’t accept “nouveau” during the install, how would I then be able to “see” the options for getting the Nvidia items??? Is this something where waiting a week or more might get the nouveau package fixed?
at boot screen press e find line starting linux or linuxefi go to the end of the line (it wraps) add a space and nomodeset.that should use failsafe fallback drivers
The warning at install simply says that nouveau is experimental. The problem happened because you did the update which brought in the broken kernel.Yep not something that is obvious. But the NVIDIA driver should work if you can manage to install it.
Or reinstall and DON’t update until the kernel is fixed
Thanks for the thought, I’ll look at it, but this is “multi-boot” situation, that uses Grub . . . have to scroll down to the “Leap” option and then try the “e” key?
I’ll give it a shot, but, seems like if the one and only kernel I have is a broken one, likely won’t make a difference which video module I’m using??
OK, tried to edit my last post but it wouldn’t let me. So, the “nomodeset” option did work!!! I’m back into my old user profile in Leap 15.1!!
So, question is, do I try to find Nvidia driver options, run a ref/dup to try to pull in a new kernel, or do nothing, use the “nomodeset” method some time down the road when a new kernel might be there?
The working 4.12.14-lp151.27 installation kernel can be installed. An update kernel in the standard update repo is expected over the next few days. It’s been tested and just waiting its turn in processing.
right now, and I ran a “zypper ref/dup” . . . and a few packages showed up, but nothing for a kernel, and I tried to get Yast to “check the system” . . . but I either don’t know how to do that, or . . . “it’s fine” . . . .
I checked “lsmod” and it shows “nouveau” in a bunch of places . . . I searched “nvidia” in Yast and basically it just showed me “nv” . . . which I think is an “old” package that we used to try to use in PPC linux to run our nvidia carded Macs . . . we “wished” to be able to use “nouveau” back then, but couldn’t qualify for it.
So, anyway, right now I’m in the GUI and it seems to be “working”?? Should I retro-install the older kernel you have listed? Or, wait a few days and run the zypper through its paces for the newer kernel??
Well . . . I guess I spoke too soon . . . I put Leap into “suspend” for a few minutes to do some other stuff . . . when I came back I hit the “shift” key and the computer started running . . . something . . . but the display power light didn’t turn “blue” and nothing showed up on the display. I hit the power button quickly and nothing happened . . . so I shut it down.
As a “test” I just let Grub list show up and I arrow-ed down to Leap and hit return . . . same issue as before, dmesg shows up, but freezes at the “switching to nouveaufb . . .” . . . line.
So I know how to get passed that point, but I’ll probably wait for a few days to hopefully catch the new kernel . . . see if we can get to a new plateau at that point . . . .
Thanks again for the follow up details . . . no particular rush on it, I could likely wait, or I could install the “testing” kernel; but over in ubuntu-land they have an “Additional Drivers” app that searches for proprietary drivers that might work and it’s a “click” away . . . does OpenSUSE have something like that?? Yast seemed to only find the retro “nv” option . . . is this something where I’d have to go to Nvidia’s website and find the, what .rpm?? package for a driver and then use GDebi?? or something like that to install the package??
Thanks kindly . . . appreciate that; this nvidia card is relatively new, so I don’t have deep background on how to use linux and proprietary drivers for nvidia cards. I’ll check into it in a day or so . . . .
Thanks for the post and the links to the further details . . . . This is a desktop . . . I’ll probably try to install it as per the suggestion to use Yast . . . have to kick it down the road a few days in any event . . . possibly new kernel will be in the tubes by that point . . . and then a little more time to fiddle with getting Nvidia drivers set up.
So I logged in using the “nomodeset” adjustment . . . and then I ran zypper and it showed a new kernel available . . . so I ran that, and rebooted back into a working GUI . . . w/o using “nomodeset” . . . .
I followed the wiki to find the steps to “add nvidia” to the repos using yast . . . and then added the drivers auto-selected in yast for nvidia . . . and rebooted into the working GUI . . . along with the video card spinning up . . . with more enthusiasm.
Then I was looking at “SJLPHI’s” thread which included the “add CUDA to the repos” . . . but that thread linked to Nvidia’s website and offer a few “rpm” options, “local” or “network” or “runfile” . . . ?? I looked back in Yast, but it wasn’t listing any “CUDA” options?? And, then, I was looking for something like the Ubuntu app “GDebi package installer” . . . which will install I believe, “.deb” files that are manually downloaded . . . but I couldn’t find or figure that one out . . . .
Is there a link to a wiki that would show how OpenSUSE handles that kind of thing? Or, is this something that Yast would find in the “community services” list? I don’t recall whether CUDA was showing up there when I checked “nvidia” . . . ?? I have a CUDA addition in one of my OSX partitions, it does seem to speed things up, but . . . it’s not a deal breaker . . . I’m just trying to bring Leap 15.1 up to operational levels . . . .
Hello n_s,
First, if you don’t need GPU programming, CUDA isn’t necessary, so in my set of instructions you can skip steps 3 and 4 an do the following:
Install LEAP 15.0 fresh, without nouveau activated (otherwise the laptop freezes and cannot boot)
Do not upgrade as it often installs nouveau somehow.
This rpm package will add the necessary cuda repositories and flag which nvidia you need. To do this, You can use the rpm installatino method. Go to the directory where you downloaded the rpm package and type in
Thanks very much for the detailed reply . . . I did get “nvidia” installed, but I don’t recall seeing that the “x11-G05xxx” list item was checked in Yast, it was something like only 4 or maybe 5 “nvidia” items . . . the top ones seemed to be “390” . . . and I think that is what my other linux systems are using, this is a '12 Mac Pro . . . I think the card is a GTX 780 3 GB??? i.e., it’s not the “newest” cutting edge card, but it got me into OSX 10.14 for somewhat of a “reasonable” price to have the card “flashed” for Mac.
I don’t use the card for 3-D stuff . . . but I do use CUDA to run distributed.net blocks . . . but I do that in OSX to make sure that the temperature controls are “factory” . . . .
Anyway, I’ll again read through your posts and see if I can get it together . . . I also don’t think Yast pulled in the “Codecs” packages???
nvidiaG05 is mostly the latest packages for nvidia. It’s almost arbitrary what is needed/installed, it depends on your hardware and repository driver version. You may have installed G01,G02,G03 or G04 for which, it’s perfectly fine as long as everything is working. GTX780 is relatively old, and Yast probably chose G03 or G04 package, this is fine.
Now that you have nvidia installed, don’t forget to eliminate “nomodeset” in boot option.
If you feel like installing cuda. Just be extremely careful because if installing nvidia package installed a different version of nvidia drivier than what cuda wants by default, it will un-install the nvidia driver and the kernel that it is compiled with.
Just make sure that when you install cuda, and it wants to remove a kernel, it is also installing another one.