Linux 4.15-rc6 Released To Ring In 2018

[phoronix]

Phoronix: Linux 4.15-rc6 Released To Ring In 2018

Linus Torvalds has released the sixth weekly release candidate of the upcoming Linux 4.15 kernel...

http://www.phoronix.com/scan.php?page=news_item&px=Linux-4.15-rc6-Released

[tildearrow]

Typos:

Linux 4.15-rc6 is mostly comprised of x76 updates that are arriving

[Michael]

Typos:

Double negation mixup, cleared up now, as it should be here in 3 weeks.

[Guest]

I have a simple solution for this problem: Buy AMD hardware instead.

Nvidia will NEVER again get any good from the Linux Community. this time is over

We (linux community, me personally and about two other individuals plus one canonical employee) typically provide patches at nvidia dev forums hours/days after they break in linus-master tree.

Nvidia only supports kernels _after_ they've been released on their monthly schedule per branch (bi-weekly releases of beta/stable drivers), typically this means that newest kernels in distributions are supported by latest drivers, in few rare cases when they weren't, distro maintainers used our (adapted) patches or delayed driver package releases until the issue was resolved.

End user only hits these problems if they are running master tree or latest rc-X releases though (and chances are that these are fixed by patches provided at said forum by the time anyone notices).

AMD users have _some_ advantage here because open-source drive is mainlined, but on the other hand they don't have release-day support for newest hardware (Nvidia typically adds support during each of their driver releases).

I am personally compiling git master kernel daily and fixing/releasing patches as breakages appear, hence why I can safely say that you have no fucking idea what you are talking about.

Linux tpruzina 4.15.0-rc6-tpruzina-master-dirty #1407 SMP PREEMPT Mon Jan 1 04:30:02 CET 2018 x86_64 Intel(R) Core(TM) i7-7700K CPU @ 4.20GHz GenuineIntel GNU/Linux

4.14.10 and 4.15-RC6 still breaks nvidia. According to kernel devs "it's not their problem." I guess that means we all have debug and patch video drivers ourselves. Ubuntu's graphics-drivers PPA is compltely useless to me right now. Heck, the idiot-proofed Software & Updates app doesn't even report the 384.90 install error. It acts like everything went perfectly fine.

Would like a quote for that "it's no their problem" part. I'm not saying that they should or even can support every RC, unless their driver gets mainlined it is literally impossible to do. As far as patching your drivers yourself, that is the price you pay for running bleeding edge kernels. Oh and as stated above, you don't even how to fix the code yourself, there are people who are already doing it and providing patches as needed. You can definitely run RC-6 with our patches and it takes almost no effort to do so (no idea about 4.14.0, I never run anything but git-master kernel but I know that there are some people who backport these patches for older/stable kernels as well, I'm just not one of them).

I'm sorry that your distribution (or PPA) doesn't provide bleeding edge drivers and kernels hours after they've been released, they probably have life too and/or want to actually test drivers before exposing thousands of users to possible regressions.

[eydee]

I have a simple solution for this problem: Buy AMD hardware instead.

Nvidia will NEVER again get any good from the Linux Community. this time is over

Surely everyone should run and grab some low-end cards for 600+ € just because there are some software issues. First get AMD to make cards, and at an acceptable price.

[bridgman]

Surely everyone should run and grab some low-end cards for 600+ € just because there are some software issues. First get AMD to make cards, and at an acceptable price.

???

https://www.newegg.com/Product/ProductList.aspx?Submit=ENE&DEPA=0&Order=BESTMATCH&Description=rx550&N=-1&isNodeId=1

There are much cheaper cards but Lexa is pretty new and maybe the best perf/$$. Basically a Cape Verde-sized Polaris.

[ryszardzonk]

4.14.10 and 4.15-RC6 still breaks nvidia.

In software bugs happen quite often. Existence of bugs as such is not disqualifying. Problematic may be only what gets done with them and how fast they get fixed. 4.14.10 has been released on 29-12 and 4.15-RC6 just yesterday which means there have not been all that much time to fix an error especially your post is from 03:17 AM next day... You expected someone to fix a problem in the middle of the night for You? How so if I bet you probably didn't even report it to the bugzilla anywhere.
Besides bugs to get fixed need to be reproducible which in my case for example a reported bug in Gentoo went unfixed for over a year hence no one was able to reproduce it. When I actually sit down to look at it it turned out it was caused by out of the tree package which was improperly installed. And whether You like it or not this is just an example of many similar cases which obviously are hard to catch by software devs.

[starshipeleven]

4.14.10 and 4.15-RC6 still breaks nvidia. According to kernel devs "it's not their problem."

For those not in the loop, this is a case of NVIDIA driver using GPL-driver-only interfaces when it should not, and this breaks when an option for newer AMD processors is enabled.

The answer to the NVIDIA dev reporting this on the mailing list was that "using GPL-driver-only interfaces" was begrudgingly tolerated, but that any breakage caused in the NVIDIA driver is their own problem, so it's on them to work around the issue or fix it in their own driver. Patches to fix that are not accepted upstream.

[Serafean]

4.14.10 and 4.15-RC6 still breaks nvidia. According to kernel devs "it's not their problem." I guess that means we all have debug and patch video drivers ourselves. Ubuntu's graphics-drivers PPA is compltely useless to me right now. Heck, the idiot-proofed Software & Updates app doesn't even report the 384.90 install error. It acts like everything went perfectly fine.

You clearly don't understand how linux is developed : In-tree drivers get fixed/tested for regressions immediately. For out of tree drivers; only the developer of the driver cares. Hence breakage will occur.
ie: get your driver into mainline, or no one will care about it.

[jabl]

It'll be interesting to see how much of a hit the kaiser/kpti causes..