nvidia-gfxG05
No description set
-
10
derived packages
- Download package
-
Checkout Package
osc -A https://api.opensuse.org checkout X11:Drivers:Video/nvidia-gfxG05 && cd $_
- Create Badge
Refresh
Refresh
Source Files
Filename | Size | Changed |
---|---|---|
README | 0000000410 410 Bytes | |
Xwrapper | 0000001175 1.15 KB | |
_constraints | 0000000109 109 Bytes | |
_service | 0000000342 342 Bytes | |
alternate-install-present | 0000000258 258 Bytes | |
check-build.sh | 0000000168 168 Bytes | |
generate-service-file.sh | 0000001327 1.3 KB | |
kernel-6.10.patch | 0000002615 2.55 KB | |
kernel-6.12.patch | 0000003786 3.7 KB | |
kernel-6.13.patch | 0000002014 1.97 KB | |
kernel-fix-empty-nv_memdbg_add.patch | 0000002291 2.24 KB | |
kmp-filelist | 0000001086 1.06 KB | |
kmp-post.sh | 0000005561 5.43 KB | |
kmp-postun.sh | 0000001076 1.05 KB | |
kmp-pre.sh | 0000000085 85 Bytes | |
kmp-preun.sh | 0000000614 614 Bytes | |
kmp-trigger.sh | 0000005622 5.49 KB | |
kmp-triggerpostun.sh | 0000000320 320 Bytes | |
modprobe.nvidia | 0000000115 115 Bytes | |
modprobe.nvidia.install | 0000001625 1.59 KB | |
my-find-supplements | 0000001205 1.18 KB | |
n_kernel_write.patch | 0000001450 1.42 KB | |
nvidia-gfxG05.changes | 0000076026 74.2 KB | |
nvidia-gfxG05.rpmlintrc | 0000000804 804 Bytes | |
nvidia-gfxG05.spec | 0000013381 13.1 KB | |
nvidia-persistenced.tar.bz2 | 0000001733 1.69 KB | |
nvidia-settings.desktop | 0000000152 152 Bytes | |
objtool-fix.patch | 0000000458 458 Bytes | |
pci_ids-470.256.02 | 0000008776 8.57 KB | |
pci_ids-470.256.02.legacy | 0000001872 1.83 KB | |
pci_ids-470.256.02.new | 0000004362 4.26 KB | |
preamble | 0000001060 1.04 KB | |
sle15-sp7.patch | 0000000600 600 Bytes | |
wait_on_bit_lock-check_better-cflags.patch | 0000000793 793 Bytes | |
x11-video-nvidiaG05.changes | 0000073035 71.3 KB | |
x11-video-nvidiaG05.spec | 0000027122 26.5 KB |
Comments 65
410.73 has been released yesterday https://devtalk.nvidia.com/default/topic/1043354/unix-graphics-announcements-and-news/linux-solaris-and-freebsd-driver-410-73/
also as 410.xx isn't a long-lived branch release I wonder does this mean that openSUSE has changed their nvidia driver update policy? (which would be great)
Thanks for letting us know. 410.xx is t a long-lived branch release. No, we didn't change our nvidia driver update policy.
Hi. I am now this branch G05 because since TW was updated to kernel 4.19 it is the one I can compile. Unfortunately yesterday I wanted to update the G04 rmps I had in the distribution by the G05 and although some were updated I was missing a couple of rpms and I used the facts from here with osc build and although they were updated normally, after restart, the X did not work for a permissions problem ... Will that be a mistake or did I do something wrong when updating? Are you using TW 4.19 with these rpms today? Now I see I downloaded NVIDIA-Linux-x86_64-410.78.run this morning,not the same than yesterday ...410-73 Thanks
I'm afraid you now have a mixture of G04/G05 packages installed. I suggest to uninstall all existing G04/G05 packages on your system (rpm -e $(rpm -qa|grep nvidia)), then reinstall the G05 packages from the repository.
I made a snapper rollback I'm back to G04 rpms. I use kernel 4.18 and today I update only one rpm nvidia-gfxG04-kmp-blablabla I see that 4.19 have now nvidia modules. I reboot and go to 4.19 and all works fine! Well is better I make the change to G05 or continue with G04? I have GeForce GT 730 What's your recommendation? Thanks
Never change a running system. ;-) Depends. GT 730 (0x0F02) is no longer supported by 410.78. Other GT 730 GPU IDs are.
GT 730 (0x0F02) is the video_card I have... Well I'll see you on G04 drivers Bye
FYI: there's 410.93 now: https://devtalk.nvidia.com/default/topic/1045858 -> seems this is needed for the upcoming 4.20 kernel
Release highlights since 410.78:
Thanks! On my TODO list now ...
The build is disabled, 418.43 wasn't built. Is it expected?
The build has never been enabled due to legal reasons. Check README file for instructions how to build packages manually.
I'll rephrase: Do you have any control on whatever is available on download.nvidia.com/opensuse/tumbleweed where the "vendor" is obs://build.suse.de/Proprietary:X11:Drivers?
Updated packages including 418.43 will be available soon on download.nvidia.com/opensuse/tumbleweed. Everything needed for this has been triggered.
I'm looking forward. Will be support upcoming short branch releases without further delay?
There are no plans to package also Short -lived Branches
v418.56 has been released (long-lived) https://devtalk.nvidia.com/default/topic/1048767/b/t/post/5322490/#5322490 PS. it's too bad that opensuse doesn't officially provide, short-lived, beta and/or vulkan beta drivers as it those would be useful for gamers especially now with proton that benefits from latest (vulkan) driver, and many don't like or know how to make an unofficial build...and I don't think it should be expected either
Updating to 418.56 now on my TODO list.
new driver release (v430.14) https://devtalk.nvidia.com/default/topic/1051837/announcements-and-news/linux-solaris-and-freebsd-driver-430-14-long-lived-branch-release-/ nothing major/interesting in this release...sadly
package update done. Should be available shortly.
https://download.nvidia.com/opensuse disappeared it seems
Indeed a refresh of the nvidia repositories no longer work for me.
Problem retrieving files from 'nvidia-leap-15.1'. Timeout exceeded when accessing 'https://download.nvidia.com/opensuse/leap/15.1/media.1/media'. Please see the above error message for a hint. Skipping repository 'nvidia-leap-15.1' because of the above error.
Hi Stefan, the latest change concerning /usr/lib/tmpfiles.d/nvidia-logind-acl-trick.conf broke all G0{3,4,5} builds with similar errors like:
[ 534s] RPMLINT report: [ 534s] =============== [ 536s] 0 packages and 0 specfiles checked; 0 errors, 29 warnings. [ 536s] Traceback (most recent call last): [ 536s] File "rpmlint.py", line 378, in <module> [ 536s] File "rpmlint.py", line 166, in main [ 536s] File "rpmlint.py", line 224, in runChecks [ 536s] File "TmpFilesCheck.py", line 50, in check [ 536s] FileNotFoundError: [Errno 2] No such file or directory: '/tmp/rpmlint.nvidia-gfxG05-kmp-default-440.64_k5.5.7_1-30.1.x86_64.rpm._p8r5pl0/usr/lib/tmpfiles.d/nvidia-logind-acl-trick.conf'
package for Leap 15.2 is broken: ksym(default:set_page_dirty_lock) = f9bf7120 benötigt von nvidia-gfxG04-kmp-default-390.132_k5.3.18_lp152.10-lp152.14.1.x86_64 wird nirgends zur Verfügung gestellt
Will you update it?
Stephan
WTF? Are the ksyms created again? I recently fixed this. They should not be created at all... Will look into this tomorrow...
Ok. Packages on the server simply have not been updated yet ... should hopefully happen this week ... the fix in RPM changelog will be:
Thu May 14 11:06:38 UTC 2020 - Stefan Dirsch sndirsch@suse.com
See here: https://forums.opensuse.org/showthread.php/540365-nvidia-(closed-source)-driver-is-missing-dependencies
Hi Stefan,
just noted, that the kernel modules aren't generated properly, if kernel-syms are missing.
Cannot reproduce this. Installing nvidia-gfxG05-kmp-default builds kernel modules even without kernel-syms package installed.
I just got hit by a disk space issue after updating the nvidia package when re-creating the initrd image, likely due to /tmp being a tmpfs now and filled with the nvidia driver build files. Can the driver use a physical partition? /var/tmp?
my tmp folder is set to use 2GB.
I don't understand. Could you please explain in detail? "Sources" for the nvidia glue layer are located below /usr/src/linux-obj and that's also where everything for the driver is being built.
Interesting, then I don't know what filled my /tmp partition. the initramfs creation failed. I thought the nvidia driver maybe left files in /tmp that could cause the issue I saw: https://paste.opensuse.org/19864203
No, it just triggers the recreation of initram, but it's not the culprit of the full disk.
Hello Stefan, do you know why is the driver shipping /usr/lib64/libnvidia-egl-wayland.so.1.1.5 while the newest is /usr/lib64/libnvidia-egl-wayland.so.1.1.6? Either way, changes are minor and I can't do much with it since I am on a laptop with optimus.
That's the version that comes with 460.56. Not sure where you found version 1.1.6.
https://github.com/NVIDIA/egl-wayland/tree/1.1.6
Well. The packages are just repackaging NVIDIA's shell script archive. I even didn't know, there would be sources available for this library. I assume NVDIA will update this component once it is well tested.
Tumbleweed user here. On my kepler based laptop the nvidia-gfxG05 is not working as intended (as it was some months ago). Using prime-select offload the nvidia modules will be loaded but something (or better something with the nvidia modules) is still missing. I can tell that on my debian sid box everything is running smoothly thanks to some patches applied by the debian maintainer. Here's the link: https://sources.debian.org/src/nvidia-graphics-drivers-tesla-470/470.141.03-3/debian/module/debian/patches/ This is the changelog: " * Backport get_task_ioprio changes from 510.85.02, acpi changes from 510.85.02 and 515.65.01, drm_frambuffer.h changes from 515.76 to fix kernel module build for Linux 6.0. (Closes: #1021974) " The patches will also fix the offload (besides the building errors with kernel 6.0). Thanks
Thanks. I've just replaced the current kernel-6.0.patch with the Debian patches 0020-0023. Could you build the RPMs yourself and test?
I’m not very practical tbh. Building rpms is something that I’ve never tried to do in the past. I will try as soon as I have some time. I can only say that if applied correctly these patches will certainly solve all the issues. Thanks for being fast
Ok. Unfortunately I can't give you prebuilt RPMs for legal reasons. I'll test ASAP myself on Kepler.
I’ll try to follow this guide (https://openbuildservice.org/help/manuals/obs-user-guide/art.obs.bg.html) or something more straightforward…
@neuromante The current G05 packages from nvidia server work for me. But this is just a single Kepler card (K600), i.e. no Optimus system. No Intel GPU on the bus. I'll try now the new G05 packages with the new patches applied.
Thanks Stefan. Let me know how does it work.
@neuromante I don't see any regressions with the new patches applied. But I only tested glxgears, glxinfo, vulkaninfo and Return to Monkey Island, which needs
VK_ICD_FILENAMES=NONE %command%
as lunch option. But this is no regression, since the previous version with the old Kernel 6.0 patch needs it as well. This env. variable forces the game to switch to OpenGL backend instead of Vulkan backend.Ok. I must admit that I'm not a great tester by my side...I even patched the driver under Arch as well and I see no issues so far...
I ended up building the package (was missing some steps and understanding of how it works) but it seems to be “linked” to kernel 6.0.6 (which I believe is in factory repos). I guess the built modules will be copied to /lib/modules/kernel…6.0.6 and not to 6.0.5…I’ll take a look closer tomorrow.
Could be that kernel-devel, kernel-default-devel has version 6.0.6, so modules have been installed to 6..0.6 modules directory, but kernel-default is still 6.0.6, so update also kernel-default package. And boot this new kernel.
Tested the new patched package this morning (kernel 6.0.6) and (at least) for me it's working pretty good. Offload...performances...I tested some games with Steam and Lutris (using my nvidia card) and see no regressions or issues.
Thanks for the feedback! With the next driver update the new patches will be in place.
Hi, I see new drivers named "nvidia-gfxG05-kmp-default-470.141.03_k6.0.8_1-58.10.x86_64" when I do zypper dup. They are on the https://download.nvidia.com/opensuse/tumbleweed/x86_64/ repo as well. Are they the new drivers with the patches applied you were talking about?
Yes, the patches are now included in G05.
Thanks.
Basically you need to install osc package, checkout the package source (osc co ...) and build it (osc build ...) (check also the README in the package sources) and install the package. KMP Package will be enough.
Ok, yeah basically I think I got the procedure and It's not that hard but I'm hesitating 'cause the system will download a lot of packages from Factory and it will ask to install them all...Now...not clear why it needs to install packages I've already installed but if I can avoid to mix packages from different repos (I know that Factory is more or less Tumbleweed...) I will do. Maybe something I still have to understand. Thanks
Are you talking about "osc build ..."? This creates a chroot system independantly of your running system. Don't be afraid of that. ;-)
Ah ok thanks (good to know)
with an additional "osc service dr" before "osc build" in order to fetch the tarballs once..
Thanks I’ll try asap (since this issue is my only gripe with tumbleweed)
Hi Stefan here we are again :)
Build error with 6.2.0: CONFTEST: ib_peer_memory_symbols CC [M] /usr/src/kernel-modules/nvidia-470.161.03-default/nvidia/nv.o CC [M] /usr/src/kernel-modules/nvidia-470.161.03-default/nvidia/nv-pci.o CC [M] /usr/src/kernel-modules/nvidia-470.161.03-default/nvidia/nv-acpi.o CC [M] /usr/src/kernel-modules/nvidia-470.161.03-default/nvidia/nv-cray.o /usr/src/kernel-modules/nvidia-470.161.03-default/nvidia/nv-acpi.c:84:19: error: initialization of ‘void ()(struct acpi_device *)’ from incompatible pointer type ‘int ()(struct acpi_device , int)’ [-Werror=incompatible-pointer-types] 84 | .remove = nv_acpi_remove_two_args, | ^~~~~~~~~~~~~~~~~~~~~~~ /usr/src/kernel-modules/nvidia-470.161.03-default/nvidia/nv-acpi.c:84:19: note: (near initialization for ‘nv_acpi_driver_template.ops.remove’) CC [M] /usr/src/kernel-modules/nvidia-470.161.03-default/nvidia/nv-dma.o CC [M] /usr/src/kernel-modules/nvidia-470.161.03-default/nvidia/nv-i2c.o cc1: some warnings being treated as errors make[3]: ** [/usr/src/linux-6.2.0-1/scripts/Makefile.build:253: /usr/src/kernel-modules/nvidia-470.161.03-default/nvidia/nv-acpi.o] Error 1 make[3]: *** Attesa per i processi non terminati.... /usr/src/kernel-modules/nvidia-470.161.03-default/nvidia/nv-dma.c:963: warning: "IMPORT_SGT_STUBS_NEEDED" redefined 963 | #define IMPORT_SGT_STUBS_NEEDED 0 | /usr/src/kernel-modules/nvidia-470.161.03-default/nvidia/nv-dma.c:957: note: this is the location of the previous definition 957 | #define IMPORT_SGT_STUBS_NEEDED 1 | make[2]: *** [/usr/src/linux-6.2.0-1/Makefile:2036: /usr/src/kernel-modules/nvidia-470.161.03-default] Error 2 make[1]: *** [Makefile:242: __sub-make] Error 2 make[1]: Leaving directory '/usr/src/linux-6.2.0-1' make: *** [Makefile:80: modules] Error 2 /
Patches are already available. See in Archlinux AUR: https://aur.archlinux.org/packages/nvidia-470xx-dkms or Rpmfusion for Fedora 38: https://koji.rpmfusion.org/koji/buildinfo?buildID=24952
Thanks again for your work
Mmmhhh trying to build locally the package but I have:
[ 2s] now finalizing build dir... [ 2s] ... running 01-add_abuild_user_to_trusted_group [ 2s] ... running 02-set_timezone_to_utc [ 2s] ... running 03-set-permissions-secure [ 2s] ... running 11-hack_uname_version_to_kernel_version [ 2s] su: cannot open session: Module is unknown [ 2s] Error: TOPDIR empty [ 2s] [ 2s] tumbleweed failed "build nvidia-gfxG05.spec" at Fri Mar 3 11:43:57 UTC 2023. [ 2s]
Following README for the instructions
@neuromante Hmm. Build works for me. Against which build target are you trying to build?
I type:
osc build -k ./packages openSUSE_Factory x86_64 nvidia-gfxG05.spec
afterocs build dr
@neuromante I can build it that way and it creates a RPM.
[...]
[ 180s] shannon finished "build nvidia-gfxG05.spec" at Fri Mar 3 12:32:19 UTC 2023. [ 180s]
/tmp/build-root/openSUSE_Factory-x86_64/home/abuild/rpmbuild/SRPMS/nvidia-gfxG05-470.161.03-0.nosrc.rpm
/tmp/build-root/openSUSE_Factory-x86_64/home/abuild/rpmbuild/RPMS/x86_64/nvidia-gfxG05-kmp-default-470.161.03_k6.2.1_1-0.x86_64.rpm
Hi! New kernel, new/same issue :) A (working) patch to build against kernel 6.3.1 is available here: https://aur.archlinux.org/cgit/aur.git/tree/kernel-6.3.patch?h=nvidia-470xx-utils Thanks again
https://bugzilla.suse.com/show_bug.cgi?id=1211115