October 2020 - openSUSE Kernel - openSUSE Mailing Lists

[opensuse-kernel] Re: [opensuse-factory] Re: No kernel 5.9 in Kernel:stable and Tumbleweed
by Dario Faggioli 23 Oct '20

23 Oct '20

On Thu, 2020-10-22 at 12:48 +0200, Axel Braun wrote: > It would be great if you could announce in which snapshot the 5.9 > kernel will > arrive, so every nivida user can take appropriate action > Yeah, so about this, what's the latest kernel with which nvidia-uvm / CUDA actually works? Asking because with kernel 5.8.14-1, I'm getting this: https://bugzilla.opensuse.org/show_bug.cgi?id=1177995 With 5.8.15-1, the module build is disabled, but I've been told (couldn't test it first hand yet) that if one builds it and tries to load it, the issues above may have gone away? And with 5.9.1, it just won't be there, right? I have an NVIDIA, but I'm not a CUDA user myself, it's just for the sake of giving accurate enough advice to users. :-) Regardsa -- Dario Faggioli, Ph.D http://about.me/dario.faggioli Virtualization Software Engineer SUSE Labs, SUSE https://www.suse.com/ ------------------------------------------------------------------- <<This happens because _I_ choose it to happen!>> (Raistlin Majere)

2 1

[opensuse-kernel] Re: [opensuse-factory] Re: No kernel 5.9 in Kernel:stable and Tumbleweed
by Jiri Slaby 23 Oct '20

23 Oct '20

Hi, On 22. 10. 20, 12:48, Axel Braun wrote: > Am Sonntag, 18. Oktober 2020, 10:07:47 CEST schrieb Jiri Slaby: >> I have just submitted 5.9.1. It will appear in Kernel:stable soon. Once >> 5.8.15 with BleedingTooth fixes is accepted to TW (sr#842044), 5.9.x >> will be submitted to TW as well. > > It would be great if you could announce in which snapshot the 5.9 kernel will > arrive, so every nivida user can take appropriate action 5.9.1 was submitted in: https://build.opensuse.org/request/show/843006 It seems no build of any package is broken. openQA demonstrates some breakage though. So once this is accepted, next snapshot will. I can leave a note here when the sr is accepted. thanks, -- js suse labs -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe(a)opensuse.org To contact the owner, e-mail: opensuse-kernel+owner(a)opensuse.org

1 0

[opensuse-kernel] No kernel 5.9 in Kernel:stable and Tumbleweed
by Jiri Slaby 18 Oct '20

18 Oct '20

Hi, this is just so you know: I intentionally linger with pushing kernel 5.9 into K:s and TW. Nvidia (mainly CUDA – uvm module) has licensing problems with 5.9. It always had, but 5.9 forbids this bad behavior now¹⁾. To cite, Nvidia promises to provide something "early". They will allegedly provide some announcement somewhere too (I don't know the details, don't ask me). Seife might (or might not) know more. If that doesn't happen until 5.9.1²⁾, I won't wait any longer and will proceed with pushing 5.9.1 to K:s and TW. They had enough time to fix their issue. More precisely since 5.9-rc1 which was released on Aug 16 13:04:57 2020 -0700. That is almost _two_ months! In fact, they must have been aware of the problem given they have been using this so-called GPL condom for ages. While I understand there might be CUDA users in TW, I see a little reason why one company would stop rolling our rolling distro. Users relying on uvm module should simply stay ("zypper al" is your friend) with 5.8 until the mess is resolved. If they don't announce anything, we will, so you can keep up. ¹⁾ https://github.com/torvalds/linux/commit/262e6ae708 ²⁾ There is no 5.9 queue in stable-queue git yet. It will take few more days or a week. thanks, -- js suse labs

6 14

[opensuse-kernel] Re: [opensuse-factory] No kernel 5.9 in Kernel:stable and Tumbleweed
by Hans-Peter Jansen 17 Oct '20

17 Oct '20

Am Samstag, 17. Oktober 2020, 00:30:38 CEST schrieb Patrick Shanahan: > * Jan Engelhardt <jengelh(a)inai.de> [10-16-20 17:50]: > > On Friday 2020-10-16 21:42, Stefan Dirsch wrote: > > >JFYI, SUSE were in contact with NVIDIA since the kernel patch was > > >proposed on the kernel mailing list. Unfortunately SUSE cannot share > > >discussions[...] . So remaining silent when such topics pop up does not > > >necessarily mean SUSE wouldn't care > > > > Readers do not necessarily see it the same way. > > There is a sizable difference between no reply and > > "we are in contact". While terse, it isn't zero-length, > > and that works in SUSE's favor. Please consider these > > four words for future events :) > > yes! > > Many of us, <users>, believe that NVidia products provide better > performance with graphics applications such as photo editing which may or > may not be factual, but comes from history. I can remember when anything > other than NVidia was a real pain. Patrick, you will find everything (fullly working) NVIDIA in my repos, including a couple of custom kernel builds, that reflect my home:frispete:kernel{,:HEAD} builds as well. Just ask, if you have questions... Cheers, Pete -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe(a)opensuse.org To contact the owner, e-mail: opensuse-kernel+owner(a)opensuse.org

1 1

Re: [opensuse-factory] Re: [opensuse-kernel] No kernel 5.9 in Kernel:stable and Tumbleweed
by Thomas Zimmermann 16 Oct '20

16 Oct '20

Hi On Fri, 16 Oct 2020 05:38:17 -0400 Neal Gompa <ngompa13(a)gmail.com> wrote: > On Fri, Oct 16, 2020 at 5:30 AM Hans-Peter Jansen <hpj(a)urpla.net> wrote: > > > > Am Freitag, 16. Oktober 2020, 01:54:00 CEST schrieb Michal Kubecek: > > > On Thu, Oct 15, 2020 at 02:38:19PM +0200, Hans-Peter Jansen wrote: > > > > Self-crippling comes to my mind. > > > > > > Yes, "self-crippling" is probably the most fitting term for users who > > > decided to ignore the long existing (20 years?) licensing problem and > > > keep buying NVidia hardware because they believe it's their right to > > > demand someone to "just make it work somehow" (i.e. do some dark magic > > > to mask the incompatibility) so that they can keep pretending the > > > problem does not exist. > > > > > > Even more fitting for those who do it on a rolling distribution with > > > latest kernel like Tumbleweed. > > > > Michal, I share your sediments from deep inside my open source heart, > > but.. > > > > All those kids, that came to me this year to buy/build a PC (I do this > > from time to time for pedagogical reasons) have a list in their hands > > with must have equipment, and guess what's *always* on position Nr. 1: > > > > NVIDIA GTX 20{70,80,90} > > > > depending on their budget and granted, they never want to discuss the > > manufacturer part of the story. About 30% get an openSUSE TW installation > > on top of the Win10 basement, and about 10% finally use TW primarily > > (because their favorite game is Minecraft, which is mostly fine with a > > decent GPU, using extensive texture packs and a huge view distance, > > *even* on Linux!). Those guys love Krita and Blender as well. > > > > I always ask myself, why is AMD not able to catch up in this segment. My > > humble guess is, they're *too* successful. Fun fact: almost all PCs I've > > talked about are equipped with their CPUs. AMD is messing up the GPU > > market, *because* their GPUs are so successful in special markets (crypto > > currency and such). Attracting the youth is *far* more harder. > > > > Attracting people is something we as openSUSE have to catch up on as well. > > Things like GPU acceleration for all kinds of workloads (UI, GFX, KI) > > should work out of the box for GPUs with open sourced drivers, but > > contrasts a lot with the X vs. Wayland mess, HiDPI issues, HDR absence, > > tearing on simple video playback, ... > > > > There are two big problems with AMD right now for *most* people: > > * No CUDA support > * No NVENC/NVDEC support > > The former is basically a requirement for anything remarkably close to > prosumer, nevermind the scientific world which has been trapped into > CUDA as well. The content creator and gaming groups *need* NVENC/NVDEC > because that's what all the tools support. Even Open Broadcaster > Software only supports NVENC for GPU-accelerated streaming. > > Somehow, for over a decade, there still is no CUDA reimplementation > for Free platforms. Maybe I missed something, but it seems like nobody > ever tried. Consequently, NVIDIA Is dominant because that API is just > not usable without an NVIDIA card. There's virtually no adoption of > OpenCL, which means that for other GPUs to succeed, CUDA needs to work > on those GPUs. But there's no reimplementation of CUDA, so we're back > to where we started. There are efforts to create something like CUDA, but free. Don't hold your breath, though. https://www.youtube.com/watch?v=ZTq8wKnVUZ8 https://www.khronos.org/sycl/ Best reards Thomas > > This is how open source loses. > -- Thomas Zimmermann Graphics Driver Developer SUSE Software Solutions Germany GmbH Maxfeldstr. 5, 90409 Nürnberg, Germany (HRB 36809, AG Nürnberg) Geschäftsführer: Felix Imendörffer -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe(a)opensuse.org To contact the owner, e-mail: opensuse-kernel+owner(a)opensuse.org

1 0

[opensuse-kernel] Info for arm64 kernel config for 5.10
by Guillaume Gardet 09 Oct '20

09 Oct '20

Hi, When kernel 5.10-rcX will be out and arm64 config will be updated, could you make sure the new CONFIG_CRYPTO_USER_API_ENABLE_OBSOLETE option remains _disabled_ please? The only known user would have been iwd/ell which are new enough already to not require it anymore thanks to the following commits: * https://git.kernel.org/pub/scm/network/wireless/iwd.git/commit/?id=1db8a85a… * https://git.kernel.org/pub/scm/libs/ell/ell.git/commit/?id=53482ce421b727c2 * https://git.kernel.org/pub/scm/libs/ell/ell.git/commit/?id=7f6a137809d42f6b Cheers, Guillaume IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you. -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe(a)opensuse.org To contact the owner, e-mail: opensuse-kernel+owner(a)opensuse.org

1 0

[opensuse-kernel] Dropping pointless bug references from kernel changelog - Re: [opensuse-factory] New Tumbleweed snapshot 20201004 released!
by Stefan Brüns 08 Oct '20

08 Oct '20

On Montag, 5. Oktober 2020 23:00:48 CEST Dominique Leuenberger wrote: The kernel changelog contains pointless, repeated bug references to a long closed bug ("Continuous stable update tracker for Tumbleweed"). > ==== kernel-source ==== > Version update (5.8.10 -> 5.8.12) > Subpackages: kernel-default kernel-docs > > - hv_netvsc: Switch the data path at the right time during > hibernation (git-fixes). > - commit 79e03c2 > - Linux 5.8.12 (bsc#1012628). > - net/mlx5e: Fix endianness when calculating pedit mask first bit > (bsc#1012628). > - net/mlx5e: Use synchronize_rcu to sync with NAPI (bsc#1012628). > - net/mlx5e: Use RCU to protect rq->xdp_prog (bsc#1012628). > - Revert "netns: don't disable BHs when locking "nsid_lock"" > (bsc#1012628). Can this noise please be removed? Adding these references just to "comply" with some policy is (--redacted--). Kind regards, Stefan -- Stefan Brüns / Bergstraße 21 / 52062 Aachen home: +49 241 53809034 mobile: +49 151 50412019

2 1

[opensuse-kernel] hibernate issue on Tumbleweed because of missing module in initrd
by Dario Faggioli 02 Oct '20

02 Oct '20

Hey, So, a TW user reported in an user group (before anyone ask, yes, I'm trying to get him to open a bug! :-) ) that he's not able to hibernate on Tumbleweed. Apparently, this is known: https://www.reddit.com/r/openSUSE/comments/hjec9g/resume_from_hibernation_i… There's a bug open about hibernating, but I am not really sure it is the same think: https://bugzilla.opensuse.org/show_bug.cgi?id=1127763 Anyhow, the user also said that adding the `resume` module to initrd, as described here, things work: https://www.reddit.com/r/openSUSE/comments/hjec9g/resume_from_hibernation_i… I've checked, and we do put that module inside initrd on Leap. Is there a particular reason not to have it in TW as well? Regards -- Dario Faggioli, Ph.D http://about.me/dario.faggioli Virtualization Software Engineer SUSE Labs, SUSE https://www.suse.com/ ------------------------------------------------------------------- <<This happens because _I_ choose it to happen!>> (Raistlin Majere)

4 3