Package Details: nvidia-340xx 340.108-6

Git Clone URL: https://aur.archlinux.org/nvidia-340xx.git (read-only, click to copy)
Package Base: nvidia-340xx
Description: NVIDIA drivers for linux, 340xx legacy branch
Upstream URL: https://www.nvidia.com/
Licenses: custom
Conflicts: nvidia
Submitter: simonsmh
Maintainer: JerryXiao (graysky)
Last Packager: graysky
Votes: 34
Popularity: 3.13
First Submitted: 2019-06-07 10:29
Last Updated: 2020-03-19 13:20

Dependencies (6)

Required by (6)

Sources (5)

Latest Comments

1 2 3 4 5 6 ... Next › Last »

wurbelgrumpff commented on 2020-03-22 22:41

with upgrading to kernel 5.5.10-arch1-1 dkms also failed due to the gcc version mismatch. With the altered command from @osvcos sudo env IGNORE_CC_MISMATCH=1 dkms install -m nvidia -v 340.108 -k 5.5.10-arch1-1 the upgrade of kernel and nvidia worked fine again.

graysky commented on 2020-03-19 13:20

OK, added in 340.108-6

https://aur.archlinux.org/cgit/aur.git/commit/?h=nvidia-340xx&id=9ae3a28fa19b416e5e8cc972c1da824bcc00893a

vova7890 commented on 2020-03-19 10:34

That simply adds SLAB_USERCOPY flag to kmem_cache_create, that come from some kernel version. You can see that in 390xx kernel module

` nv_memdbg_init();

/* Allocated memory that is used for copying to/from userspace should be
 * tagged as such, on kernels that support this feature.
 */
nvidia_stack_t_cache = NV_KMEM_CACHE_CREATE_USERCOPY(nvidia_stack_cache_name,
                                                     nvidia_stack_t);
if (nvidia_stack_t_cache == NULL)
{
    nv_printf(NV_DBG_ERRORS, "NVRM: stack cache allocation failed!\n");
    rc = -ENOMEM;
    goto failed6;
}

rc = nv_kmem_cache_alloc_stack(&sp);

`

And i had tested this on nvidia 210 gpu, work fine.

graysky commented on 2020-03-19 08:50

I have no hardware myself for testing. Does the error in dmesg also cause breakage or is it just a harmless line in dmesg? Can you post a link or two from your searching? Can you verify there are no ill-effects of that patch?

vova7890 commented on 2020-03-19 01:01

I'm facing this message in dmesg: Bad or missing usercopy whitelist? Kernel memory exposure attempt detected from SLUB object 'nvidia_stack_t'

A little search and I had portback this patch to 340xx. Can you please add this? https://zvova7890.com.ua/s/ed9JJG8KyQKQQBx

osvcos commented on 2020-03-16 20:51

@holyArch As graysky wrote, this is probaly due to the gcc version mismatch. Try with: sudo env IGNORE_CC_MISMATCH=1 dkms install -m nvidia -v 340.108 -k 5.5.9-arch1-2

holyArch commented on 2020-03-16 20:40

@graysky The mirrors I use are all listed under Successfully Syncing Mirrors, so that's not the issue.

Edit: Thank you, osvcos, the command words (after upgrading kernel).

graysky commented on 2020-03-16 19:57

Is your mirror out of sync? You need the new toolkit that was used to build the kernel.

holyArch commented on 2020-03-16 19:01

@graysky

DKMS make.log for nvidia-340.108 for kernel 5.5.9-arch1-2 (x86_64)

gcc-version-check failed:

The compiler used to compile the kernel (gcc 9.2) does not exactly match the current compiler (gcc 9.3). The Linux 2.6 kernel module loader rejects kernel modules built with a version of gcc that does not exactly match that of the compiler used to build the running kernel.

If you know what you are doing and want to override the gcc version check, you can do so by setting the IGNORE_CC_MISMATCH environment variable to "1".

In any other case, set the CC environment variable to the name of the compiler that was used to compile the kernel.

Failed CC version check. Bailing out!

make: [nvidia-modules-common.mk:201: build-sanity-checks] Error 1 make: Entering directory '/var/lib/dkms/nvidia/340.108/build/uvm' cd ./..; make module SYSSRC=/lib/modules/5.5.9-arch1-2/build SYSOUT=/lib/modules/5.5.9-arch1-2/build KBUILD_EXTMOD=./.. make[1]: Entering directory '/var/lib/dkms/nvidia/340.108/build' NVIDIA: calling KBUILD... make[2]: Entering directory '/usr/lib/modules/5.5.9-arch1-2/build' test -e include/generated/autoconf.h -a -e include/config/auto.conf || ( \ echo >&2; \ echo >&2 " ERROR: Kernel configuration is invalid."; \ echo >&2 " include/generated/autoconf.h or include/config/auto.conf are missing.";\ echo >&2 " Run 'make oldconfig && make prepare' on kernel src to fix it."; \ echo >&2 ; \ /bin/false) make -f ./scripts/Makefile.build obj=.. \ single-build= \ need-builtin=1 need-modorder=1 scripts/Makefile.build:42: ../Makefile: No such file or directory make[3]: No rule to make target '../Makefile'. Stop. make[2]: [Makefile:1693: ..] Error 2 make[2]: Leaving directory '/usr/lib/modules/5.5.9-arch1-2/build' NVIDIA: left KBUILD. nvidia.ko failed to build! make[1]: [Makefile:197: nvidia.ko] Error 1 make[1]: Leaving directory '/var/lib/dkms/nvidia/340.108/build' make: *** [Makefile:222: ../Module.symvers] Error 2 make: Leaving directory '/var/lib/dkms/nvidia/340.108/build/uvm'

graysky commented on 2020-03-14 22:23

@holyArch - Builds fine for me in a clean chroot. Not sure what to tell you.https://gist.github.com/graysky2/55b9912ce90a82d7a7714f9b6f0854c2

And please do not flag out of date because you ran into a build error. It's out of date when upstream releases something new.