Re: New package: linux5.5-5.5.2 (tested on x86_64).

Re: New package: linux5.5-5.5.2 (tested on x86_64).

[voidlinux-github]

[-- Attachment #1: Type: text/plain, Size: 286 bytes --]

New comment by xtraeme on void-packages repository

https://github.com/void-linux/void-packages/pull/18849#issuecomment-582763752

Comment:
It's restricted to be built only for i686 and x86_64. This way we can push this without introducing build failures due to outdated configuration.

Re: [PR PATCH] [Merged]: New package: linux5.5-5.5.2 (tested on x86_64).

[voidlinux-github]

[-- Attachment #1: Type: text/plain, Size: 981 bytes --]

There's a merged pull request on the void-packages repository

New package: linux5.5-5.5.2 (tested on x86_64).
https://github.com/void-linux/void-packages/pull/18849

Description:
This enables X86_UMIP among other things, also restores
TRANSPARENT_HUGEPAGE_MADVISE=y (which was set to
_ALWAYS again, see [1]).

How to know what is the policy to enable or disable
any option? simple: we follow what upstream defaults to, e.g:
if an option is disabled by default (N), we don't enable
it unless it's explicitely required. For any other options
that can be built as modules, we build them as modules.

So the linux kernel configuration that void uses depends
on what upstream sets by default.

For now this only contains working configurations for
x86 and x86_64. People with working hardware should adapt
this configuration to the target arch appropiately.

Built and tested on x86 and x86_64.

[1] https://blog.nelhage.com/post/transparent-hugepages/

[ci skip]

Re: New package: linux5.5-5.5.2 (tested on x86_64).

[voidlinux-github]

[-- Attachment #1: Type: text/plain, Size: 416 bytes --]

New comment by q66 on void-packages repository

https://github.com/void-linux/void-packages/pull/18849#issuecomment-582869093

Comment:
This introduces a bunch of rather questionable changes to the kernel config, like disabling page posioning, restricting CPUs to 64, disabling KASLR, and disabling a class of broadcom wifi drivers, possibly others. How about not merging things like this without undergoing review?

Re: New package: linux5.5-5.5.2 (tested on x86_64).

[voidlinux-github]

[-- Attachment #1: Type: text/plain, Size: 452 bytes --]

New comment by q66 on void-packages repository

https://github.com/void-linux/void-packages/pull/18849#issuecomment-582869093

Comment:
This introduces a bunch of rather questionable changes to the kernel config, like disabling page posioning (and a bunch of other hardening options), restricting CPUs to 64, disabling KASLR, and disabling a class of broadcom wifi drivers, and others. How about not merging things like this without undergoing review?

Re: New package: linux5.5-5.5.2 (tested on x86_64).

[voidlinux-github]

[-- Attachment #1: Type: text/plain, Size: 197 bytes --]

New comment by xtraeme on void-packages repository

https://github.com/void-linux/void-packages/pull/18849#issuecomment-582872551

Comment:
Feel free to update it, it took me a long while already.

Re: New package: linux5.5-5.5.2 (tested on x86_64).

[voidlinux-github]

[-- Attachment #1: Type: text/plain, Size: 221 bytes --]

New comment by xtraeme on void-packages repository

https://github.com/void-linux/void-packages/pull/18849#issuecomment-582874156

Comment:
CPUs is restricted to 64 because this is the default.

I simply used defaults.

Re: New package: linux5.5-5.5.2 (tested on x86_64).

[voidlinux-github]

[-- Attachment #1: Type: text/plain, Size: 384 bytes --]

New comment by xtraeme on void-packages repository

https://github.com/void-linux/void-packages/pull/18849#issuecomment-582876362

Comment:
You are talking about "without undergoing review", linux5.4 reverted my change to not enable THP by default, as stated in my commit message. Who reviewed this? obviously nobody.

So I don't think this is worse than that was there before.

Re: New package: linux5.5-5.5.2 (tested on x86_64).

[voidlinux-github]

[-- Attachment #1: Type: text/plain, Size: 246 bytes --]

New comment by xtraeme on void-packages repository

https://github.com/void-linux/void-packages/pull/18849#issuecomment-582877545

Comment:
Note that I've used oldconfig and looking at each option one by one.

It took me literally a few hours.

Re: New package: linux5.5-5.5.2 (tested on x86_64).

[voidlinux-github]

[-- Attachment #1: Type: text/plain, Size: 475 bytes --]

New comment by q66 on void-packages repository

https://github.com/void-linux/void-packages/pull/18849#issuecomment-582877652

Comment:
The existing config was fine as it was already - just take the config from previous version, run oldconfig on it in the new tree, update options that are new/have changed, and possibly change the things you explicitly wanted to change later. There is no need to go through creation of a new config and is likely to do more harm than good.

Re: New package: linux5.5-5.5.2 (tested on x86_64).

[voidlinux-github]

[-- Attachment #1: Type: text/plain, Size: 166 bytes --]

New comment by xtraeme on void-packages repository

https://github.com/void-linux/void-packages/pull/18849#issuecomment-582877798

Comment:
@q66 this is what I did. 

Re: New package: linux5.5-5.5.2 (tested on x86_64).

[voidlinux-github]

[-- Attachment #1: Type: text/plain, Size: 657 bytes --]

New comment by q66 on void-packages repository

https://github.com/void-linux/void-packages/pull/18849#issuecomment-582877652

Comment:
The existing config was fine as it was already - just take the config from previous version, run oldconfig on it in the new tree, update options that are new/have changed, and possibly change the things you explicitly wanted to change later. There is no need to go through creation of a new config and is likely to do more harm than good.

Maybe then it also wouldn't have taken you a few hours...

Yes, it is worse than it was before; these things are listed are just a few that I immediately saw, and there's more.

Re: New package: linux5.5-5.5.2 (tested on x86_64).

[voidlinux-github]

[-- Attachment #1: Type: text/plain, Size: 211 bytes --]

New comment by xtraeme on void-packages repository

https://github.com/void-linux/void-packages/pull/18849#issuecomment-582878432

Comment:
If you don't trust me, try yourself with the existing linux5.4 config.

Re: New package: linux5.5-5.5.2 (tested on x86_64).

[voidlinux-github]

[-- Attachment #1: Type: text/plain, Size: 730 bytes --]

New comment by q66 on void-packages repository

https://github.com/void-linux/void-packages/pull/18849#issuecomment-582885898

Comment:
"You are talking about "without undergoing review", linux5.4 reverted my change to not enable THP by default, as stated in my commit message. Who reviewed this? obviously nobody."

previous mistakes should not justify making them again. Besides, I only call out things I notice problems with, in this case I went to diff the new config against the old one to be able to see what changed so I could use that for the ppc configs.

I don't have an x86_64 machine at hand to test that out, but I had another user on IRC confirm that oldconfig *does not* ask for those things that were changed.

Re: New package: linux5.5-5.5.2 (tested on x86_64).

[voidlinux-github]

[-- Attachment #1: Type: text/plain, Size: 404 bytes --]

New comment by q66 on void-packages repository

https://github.com/void-linux/void-packages/pull/18849#issuecomment-582887278

Comment:
Anyway, my main point here is that these kinds of things just should not happen, as there should be no rush with introducing any new packages into the repository, so we can take our time; 5.4 is a longterm tree and 5.5 does not introduce anything extremely important.

Re: New package: linux5.5-5.5.2 (tested on x86_64).

[voidlinux-github]

[-- Attachment #1: Type: text/plain, Size: 186 bytes --]

New comment by xtraeme on void-packages repository

https://github.com/void-linux/void-packages/pull/18849#issuecomment-582762995

Comment:
@q66 @pullmoll please update the !x86 configs