32bit pae kernels - 2 made - 5.4.158 & 4.19.217

[ozsouth]

WARNING:-
OLD KERNELS & FDRVS SHOULD NOT BE USED WITH WOOF-CE PUPPIES CREATED AFTER 2022.
STRUCTURAL CHANGES WILL LIKELY CAUSE CORRUPTION.

I've made two 32bit pae kernels, with firmware enclosed, via bionicpup32-8.0, which has gcc 7.5.0.
Both briefly tested, OK in bionicpup32-8.0. Use at own risk.
Initial source release was November 2021 - I have had more success compiling kernels from then.

1. 5.4.158 32bit pae

Once downloaded, expand in an empty folder with tar -jxvf, & rename kernel-modules.sfs-5.4.158-oz-pae to zdrv ... (same as one to be replaced) & rename vmlinuz-5.4.158-oz-pae to vmlinuz & then substitute for originals.

Kernel here: https://archive.org/download/Puppy_Linu ... ae.tar.bz2
Sources here: https://archive.org/download/Puppy_Linu ... oz-pae.sfs
Headers here: https://archive.org/download/Puppy_Linu ... e-i686.sfs

2. 4.19.217 32bit pae

Once downloaded, expand in an empty folder with tar -jxvf, & rename kernel-modules.sfs-4.19.217-oz-pae to zdrv ... (same as one to be replaced) & rename vmlinuz-4.19.217-oz-pae to vmlinuz & then substitute for originals.

Kernel here: https://archive.org/download/Puppy_Linu ... ae.tar.bz2
Sources here: https://archive.org/download/Puppy_Linu ... oz-pae.sfs
Headers here: https://archive.org/download/Puppy_Linu ... e-i686.sfs

I apologise if any ads shown at Mediafire links are inappropriate.

[amethyst]

Looks like some sublime work from you here ozsouth. I've downloaded the 5.4 kernel and it boots like a house on fire. This is the first new kernel after 4.4 that does not have some serious hangtime at bootup on my old laptop. Using it with Xenial7.5. Thanks.
BTW - I see at bootup an error message reporting "wrong EFI loader signature". Just reporting it as a matter of interest, does not seem to have any ill effects.

[ozsouth]

@amethyst - Think that's a bootloader issue, but once booted, hopefully OK.

[rockedge]

I also see the error (or is it info?) when booting KLV-Airedale and other WeeDog32's as well. Doesn't seem to effect anything in KLV or WDL but I should track it down to what it really means.

I have lots of success compiling and using 5.4+ kernels as well and the 4.19+ series before that. Around 5.14+ it just became a hit or miss for some reason. I have a few 32 bit full real time kernels including some non-PAE versions.

[mikewalsh]

@amethyst :-

Nic, that "wrong EFI loader signature" is nowt to worry about.

With this new HP rig, I HAVE to use 4-series or later kernels in order to get functional audio.....especially with Pups that originally came with 3-series kernels. Reason? Quite simply, the audio chip in this machine didn't even exist at the time the 3-series kernels would have been 'current.....so there's no way that any 3-series kernel could possibly have drivers for it. They wouldn't even know about it.

So; 32-bit Tahrpup, Xenial, and a couple of Slackos that I run - Micko's original 5.6.0 and Sailor's sublime re-spin on 5.7.0 (5.7.1) - which all came with 3-series kernels, these have all been updated to the 4.1.30 "huge" kernel Ally has listed over at archive.org. And all work perfectly.

Every single one of my Puppies exhibits that same 'error message' at boot (or, as Erik points out, it could simply be 'info'). It seems to be endemic to the 4-series & later kernels; perhaps because Puppies don't use the same GRUB2 setup employed by every mainstream distro out there.....all of which also employ a (seemingly) far more complicated method for upgrading/re-registering new kernels in their systems.

And this applies to every single 64-bit Pup I run, too.....all of which use 4- or 5-series kernels.

I'm convinced that Puppy's relatively 'simple' modular kernel replacement method throws the thing for a loop.....because it is expecting to be subjected to a far more complex set of installation steps! But in terms of operational functionality & stability, that's all it is.....JUST an error message. It doesn't prevent the thing from functioning as intended.

I'm just reporting my own experiences, that's all; I don't for one moment think it has anything to do with Oz's build procedure. I hope that sets your mind at ease a bit.

Mike.

[rockedge]

I don't for one moment think it has anything to do with Oz's build procedure. I hope that sets your mind at ease a bit.

I totally agree. Every kernel I've built upwards from version 5.0 until now shows the same message.

I'll try to see what it takes to get an EFI signature and what to do with one.