kernel compiling in woof-ce

Message

stemsee · #561 Post by **stemsee** » Sun 14 Apr 2019, 01:01

gdrive share packages directory containing kernel 5.0.7 x64

https://drive.google.com/drive/folders/ ... O3-6UjDSyz

s243a · #562 Post by **s243a** » Sun 14 Apr 2019, 02:07

s243a wrote:
stemsee wrote:Hi a243a

How did you get on?

I assumed that you are using woof-ce kernel kit, and not SUKK, so I didn't reply. But now I am curious as to how you got on.

stemsee
My attempted fixes at xz compression didn't work. That said the kernal sources for slacko64.6.9.9.9 should work with xz compression because I have no issue using the exiting compile kernal with xz compression. I know those because I tried it with the tazpup build-kit. I'm pretty sure I used SUKK. I'm temporarily abandoning my attempts at using SUKK for the moment.

Perhaps the solution is to compare the compiled options that are used with slacko64.6.9.9.9 with what SUKK uses.

I wonder if the following had anything to do with my issue:

rufwoof wrote: I had previously attempted to xz extreme compress that initrd, but it wouldn't boot. Then I remembered you have to include the --check=crc32 xz compression parameter when forming a initrd.xz with extreme compression and that boots fine. 12MB xz compressed initrd.xz filesize (alongside a 5MB vmlinuz).

http://murga-linux.com/puppy/viewtopic. ... 03#1022903

stemsee · #563 Post by **stemsee** » Tue 25 Jun 2019, 18:21

I have updated nubuild.sh.

Removed the ability to build.deb packages as the process has changed a lot. Although I think I perceive an easy hack may be possible.

Dependency checks now occur in a loop that checks for internet connection, and prompts for devx and internet.

The script ends by opening SUKK/dist/packages in rox.

stemsee · #564 Post by **stemsee** » Mon 18 Nov 2019, 16:49

I have just compiled kernel 5.3.11-x86_64. It's running nicely. It imported the default hostname (fatdog in this case).

Sharing the compilation dist dir from gdrive. Includes Huge package + md5, vmlinuz, kernel-modules.sfs, kernel sources.sfs.
https://drive.google.com/drive/folders/ ... RVz19zQdai

stemsee · #565 Post by **stemsee** » Mon 18 Nov 2019, 21:44

5.3.11-i686-pae

https://drive.google.com/drive/folders/ ... 7RGjuxdtAB

#566 Post by **rockedge** » Tue 26 Nov 2019, 14:53

working on compiling a patched for PREEMPT RT kernel 5.2.21 or a similar kernel version.

So far some slight errors appear when attempting to boot a Bionic64 with a completed kernel and the system drops down to shell and can not mount /dev/loop2 or the puppy .sfs

LateAdopter · #567 Post by **LateAdopter** » Tue 26 Nov 2019, 18:55

Hello stemsee

Thanks for this 5.3.11 kernel. It comes at an opportune moment. My Athlon II X2 240 motherboard broke during the summer and, now that AMD have released it, I have ordered an Athlon 3000g and an AM4 motherboard to replace it.

I am updating a copy of Bionicpup64 with all of the latest bits, so that they recognise the new cpu/gpu. The 5.3 series kernels have a lot of relevant AMD fixes for it too.

In common with the Fatdog kernels your firmware does not have the AMDGPU firmware for this year's AMD processors: raven2 and picasso. I copied them out of the Debian Buster firmware-amd-graphics package.

The vmlinuz on the google drive is 0 bytes, but the one in the huge file is OK.

Thanks again

#568 Post by **rockedge** » Tue 26 Nov 2019, 21:09

Hello everyone,

Kernel 5.3.13_x86_64 is available here -> http://rockedge.org/kernels/data/kernel ... 64.tar.bz2

more at http://rockedge.org/kernels/

stemsee · #569 Post by **stemsee** » Wed 27 Nov 2019, 13:39

LateAdopter wrote: The vmlinuz on the google drive is 0 bytes, but the one in the huge file is OK.

Thanks again

Thanks for your feedback. I'll take a look.

Regarding firmware, SUKK relies on the two firmware packages made ages ago by 01micko ... I guess he's not keeping them up-to-date. I will try to update SUKK with the option to get the massive firmware package from kernel.org

Rockedge - your last but one post indicates a compile effort which seems to have failed due to the aufs patches not taking...

cheers
stemsee

#570 Post by **rockedge** » Wed 27 Nov 2019, 21:02

yes stemsee that is correct. The aufs builds successfully. The problem is applying the full PREEMT RT patches.

I have although since been able to compile a full PREEMT RT kernel using version 4.19.82_x86_64-rt30 for which the RT patches were successfully applied.
I will make this kernel available for download.

It is trying to apply the RT patches to kernel 5.2.21 that the build stops.

Some will work but the process stops at the beginning of applying the patches. Using only the patches that would load produced a faulty kernel in experimentation.

stemsee · #571 Post by **stemsee** » Wed 27 Nov 2019, 22:00

Rockedge

Maybe you could build a more recent kernel (5.3/4) using the latest 5.2.21 rt patch. Or report the errors from patching.

stemsee

#572 Post by **rockedge** » Thu 28 Nov 2019, 14:56

▶—— Linux Kernel ——◀

Kernel Release: 4.19.82-rt30 SMP PREEMPT RT
Build Date: RT Wed Nov 27 09:24:12 EST 2019
Build GCC: 7.4.0
OS Support: GNU/Linux
Architecture: x86_64
SMP Enabled: Yes

Kernel Command Line:
psubdir=/BB64 pmedia=atahd pfix=fsck net.ifnames=0

This is a fully preemptible kernel and can be applied for use as CNC machine controller, for electronic music production and software synthesizers and audio / video recording and editing or any real time applications one wants a Puppy Linux to do.

Downloads ->
http://rockedge.org/kernels/
http://rockedge.org/kernels/data/kernel ... 64.tar.bz2

**

stemsee · #573 Post by **stemsee** » Mon 02 Dec 2019, 16:59

5.3.14 x86_64
https://drive.google.com/open?id=1ZA1Md ... PV30ZnCcXa

EDIT: This and the previous kernel, were compiled on an NTFS partition!

festus · #574 Post by **festus** » Fri 13 Dec 2019, 17:39

rockedge wrote:▶—— Linux Kernel ——◀

Kernel Release: 4.19.82-rt30 SMP PREEMPT RT...

Hello, rockedge,

I have been using your "Kernel Release: 4.19.82-rt30 SMP PREEMPT RT" in bionicpup64 since 29nov2019 with excellent results and great satisfaction.

I've compiled several pkgs, manipulate/record sound files, office work, and general daily tasks.

And it idles at 0 to 1% CPU Usage.

IMO, this is a kernel that should not be overlooked...

Thank you for taking the time to compile this rt kernel.

bliss,
festus

stemsee · #575 Post by **stemsee** » Tue 24 Dec 2019, 13:39

Hi rockedge

I'm using your rt kernel with fatdog64. After fine tuning my system for realtime audio processing I get a recording latency of 1.3ms which is plenty good enough for duplex recording with live instruments; this is on a core i3 3217U with 4gb ram.

The only thing is that you configure the kernel with CONFIG_HZ_300=y it would probably give better performance/timer resolution if CONFIG_HZ_1000=y instead!

I tried to compile a newer rt kernel but it failed and I didn't troubleshoot.

Very nice anyway.
stemsee

Dry Falls · #576 Post by **Dry Falls** » Mon 30 Dec 2019, 22:34

Aufs 5.4 is up and running. Latest nubuild.sh zipped through the k5.4.6. Posting from it now.

#577 Post by **rockedge** » Tue 31 Dec 2019, 00:30

Hey Dry Falls!

Thanks for the info.

#578 Post by **rockedge** » Tue 31 Dec 2019, 13:33

I attempted to build a RT kernel 5.4.5 using the 5.4.5-rt3 patches but half way through applying the RT patches there is an error and the build stops.

This the same problem so far with any attempt to compile a RT kernel with any kernel version 5+

I have not tried out nubuild.sh but I will today.

peterw · #579 Post by **peterw** » Fri 03 Jan 2020, 18:53

I don't know if this is the best place to request additions to the Puppy DOTconfig but I will and hopefully someone will point me in the right direction if I am wrong.

Basically, Puppy is one of the best Distros to use on a Chromebook, especially, as they near EOL after 6.5 years from the Chromebook type release (and many are in that sate). One of the issues is that some of the sound chips are not supported in Puppy. This is currently being discussed here: http://www.murga-linux.com/puppy/viewtopic.php?t=117617.

To try and find a solution, I recompiled the kernel to get it to recognise the max98090 chip and the relevant part of my DOTcongfig file is shown below. However, looking at it I see that I have not caught all types of Chromebooks as there are two other "is not set" values also shown.

# CONFIG_SND_SST_ATOM_HIFI2_PLATFORM_PCI is not set
CONFIG_SND_SST_ATOM_HIFI2_PLATFORM_ACPI=m
# CONFIG_SND_SOC_INTEL_SKYLAKE is not set
CONFIG_SND_SOC_ACPI_INTEL_MATCH=m
CONFIG_SND_SOC_INTEL_MACH=y
CONFIG_SND_SOC_INTEL_HASWELL_MACH=m
CONFIG_SND_SOC_INTEL_BDW_RT5677_MACH=m
CONFIG_SND_SOC_INTEL_BROADWELL_MACH=m
CONFIG_SND_SOC_INTEL_BYTCR_RT5640_MACH=m
CONFIG_SND_SOC_INTEL_BYTCR_RT5651_MACH=m
CONFIG_SND_SOC_INTEL_CHT_BSW_RT5672_MACH=m
CONFIG_SND_SOC_INTEL_CHT_BSW_RT5645_MACH=m
CONFIG_SND_SOC_INTEL_CHT_BSW_MAX98090_TI_MACH=m
CONFIG_SND_SOC_INTEL_CHT_BSW_NAU8824_MACH=m
CONFIG_SND_SOC_INTEL_BYT_CHT_DA7213_MACH=m
CONFIG_SND_SOC_INTEL_BYT_CHT_ES8316_MACH=m
CONFIG_SND_SOC_INTEL_BYT_CHT_NOCODEC_MACH=m

Perhaps the experts who know much more than I and compile more advanced kernels could add those features in future compilations

stemsee · #580 Post by **stemsee** » Fri 03 Jan 2020, 19:14

Hello Peterw

These options pertain to the snd-soc/ucm sound card that comes with atom processor chips.

I submitted a kernel DOTconfig for atom cpu to woof-ce > kernel-kit. it should have everything enabled.

stemsee

(old)Puppy Linux Discussion Forum

(old)Puppy Linux Discussion Forum

kernel compiling in woof-ce

Aufs 5.4

Additions to DOTconfig