6.5.1 kernel attempt withdrawn

Moderator: Forum moderators

Post Reply
ozsouth
Posts: 1569
Joined: Sun Jul 12, 2020 2:38 am
Location: S.E. Australia
Has thanked: 241 times
Been thanked: 704 times

6.5.1 kernel attempt withdrawn

Post by ozsouth »

I attempted a 6.5.1 64bit kernel. Someone got the ubuntu flavour before I withdrew it. I used gcc 11.2.0 which is not compatible with much (basically only s15pup64-22.12) & wouldn't compile under F95CE-4 due to it's gcc of 9.4.0. Also peebee in his kernel, had to revert uconn.c to 6.4.6 version. Too many issues. I suggest the person who downloaded it discard their copy.

User avatar
peebee
Posts: 1636
Joined: Mon Jul 13, 2020 10:54 am
Location: Worcestershire, UK
Has thanked: 157 times
Been thanked: 714 times
Contact:

Re: 6.5.1 kernel attempt withdrawn

Post by peebee »

ozsouth wrote: Mon Sep 04, 2023 9:23 am

Also peebee in his kernel, had to revert uconn.c to 6.4.6 version.

FYI: Very specific issue with Nvidia GT218 graphics with `DMS-59` connector.... There was a change to uconn.c at version 6.4.7 to cure another obscure Nouveau problem which killed booting with the above card.... which I have!!

Builder of LxPups, SPups, UPup32s, VoidPups; LXDE, LXQt, Xfce addons; Chromium, Firefox etc. sfs; & Kernels

User avatar
rockedge
Site Admin
Posts: 6551
Joined: Mon Dec 02, 2019 1:38 am
Location: Connecticut,U.S.A.
Has thanked: 2757 times
Been thanked: 2627 times
Contact:

Re: 6.5.1 kernel attempt withdrawn

Post by rockedge »

After reading the topic I decided to try out building a 6.5.1 kernel with the woof-CE kernel-kit. On a KLV-Airedale-RT running on a QEMU virtual machine I was able to successfully compile 6.5.1 with built in overlayfs and no AUFS. The compile with 4 cores was about 40 minutes using the VM. KLV is using gcc v12+.

It took a small bit of adjusting things and adding a few development packages to do the run.

Screenshot(98).jpg
Screenshot(98).jpg (32.59 KiB) Viewed 1036 times

I am looking forward now to try a compile with AUFS 6.5 patched and enabled.

ozsouth
Posts: 1569
Joined: Sun Jul 12, 2020 2:38 am
Location: S.E. Australia
Has thanked: 241 times
Been thanked: 704 times

Re: 6.5.1 kernel attempt withdrawn

Post by ozsouth »

@rockedge - good! The aufs side wasn't a problem for me - kernel actually worked, but the gcc thing worries me. I usually use 9.3.0 - which would probably have worked out better. Peebee used 13.2 in his 6.5.1. The wide variation means some things (like broadcom wl) won't compile. I may have to just drop it. Also, I still worry about that uconn.c thing even though it is a rare case. Will probably be a revert in a future kernel.

User avatar
rockedge
Site Admin
Posts: 6551
Joined: Mon Dec 02, 2019 1:38 am
Location: Connecticut,U.S.A.
Has thanked: 2757 times
Been thanked: 2627 times
Contact:

Re: 6.5.1 kernel attempt withdrawn

Post by rockedge »

Anyone willing to test out this kernel built on a QEMU virtual machine running KLV-Airedale-RT can find it here -> https://rockedge.org/kernels/data/Kerne ... overlayfs/

First kernel I compiled on a KLV-Airedale using a QEMU VM.

dimkr
Posts: 2424
Joined: Wed Dec 30, 2020 6:14 pm
Has thanked: 53 times
Been thanked: 1202 times

Re: 6.5.1 kernel attempt withdrawn

Post by dimkr »

According to https://tracker.debian.org/pkg/linux-signed-amd64, 6.5.x should land in Debian unstable soon, then https://github.com/puppylinux-woof-CE/w ... kernel.yml will rebuild it for Puppy using the matching GCC version. (Currently, it's building 6.4.x, which works fine for me.)

If you want a reliable source of recent kernels (without aufs!) to test on your hardware before you try to build yourself, you might find this useful. It also builds the Debian testing (currently, 6.4.x), stable (frozen on 6.1.x) and oldstable (frozen on 5.10.x) kernels, if you want something more stable.

User avatar
rockedge
Site Admin
Posts: 6551
Joined: Mon Dec 02, 2019 1:38 am
Location: Connecticut,U.S.A.
Has thanked: 2757 times
Been thanked: 2627 times
Contact:

Re: 6.5.1 kernel attempt withdrawn

Post by rockedge »

As I attempt a 6.5.1 build with aufs6.5 I am getting this error:

Code: Select all

./build.sh: line 333: git_aufs_util_branch: command not found

It is the same build.conf as the successful build with only overlayfs builtin and no aufs except now with AUFS=yes

This seems to be the section of code responsible for setting aufs version:

Code: Select all

# ===============================
if [ "$AUFS" != "no" ] ; then
	if [ ! "$aufsv" ] ; then
		git_aufs_branch ${kernel_version} # sets $aufsv
	fi
	git_aufs_util_branch # sets $aufs_util_branch

	[ "$aufsv" ] || exit_error "You must specify 'aufsv=version' in build.conf"
	log_msg "aufs=$aufsv"
	log_msg "aufs_util=$aufs_util_branch"
fi
ozsouth
Posts: 1569
Joined: Sun Jul 12, 2020 2:38 am
Location: S.E. Australia
Has thanked: 241 times
Been thanked: 704 times

Re: 6.5.1 kernel attempt withdrawn

Post by ozsouth »

I made 6.5.2 (aufs/overlayfs) with my standard 2018-Kit, so didn't try github version (I would have had I failed again). Using gcc 9.3.0 & glibc 2.30, everything worked, so I'm sure 6.5.1 would have worked with that combo.
@rockedge - I haven't seen the build.sh that contains AUFS=yes (or no), but it would appear something is wrong with it. I have a 2018-Kit with aufs commented out (in many places) that I use for overlayfs only. In my standard 2018-Kit, buildu.sh builds an Ubuntu clone puppy kernel (builds.sh builds a slacko one), with aufs/overlayfs (aufs by default; overlayfs also if it is set in DOTconfig).

dimkr
Posts: 2424
Joined: Wed Dec 30, 2020 6:14 pm
Has thanked: 53 times
Been thanked: 1202 times

Re: 6.5.1 kernel attempt withdrawn

Post by dimkr »

6.5.3 is available at https://github.com/puppylinux-woof-CE/w ... #artifacts (for two weeks, until the next build). 6.5.x is finally pushed to Debian unstable and rebuilt in https://github.com/puppylinux-woof-CE/w ... 6812631578.

Post Reply

Return to “Kernels”