kernel compiling in woof-ce

Message

stemsee · #91 Post by **stemsee** » Sun 22 Feb 2015, 09:15

There is the original script by Iguleder and then 01micko, which I have altered. I added my own time stamp, there was already $TODAY, I didn't add it to modules.sfs because at the end of the script they are moved to a dir with $timestamp which goes to the minute and takes care of multiple compiles in one day without overwriting. If the kernel source.tar.bz2 or .sfs package is overwritten it doesn't matter, only the kernel version is important.

stemsee · #92 Post by **stemsee** » Tue 24 Feb 2015, 13:58

Has anyone else noticed the strange events at www.kernel.org ???

Having completed the unattended kernel kit - the latest kernels are labelled thusly 3.19 and 4.0-rc1 ... The script can be tricked to accommodate then naming for kernel and aufs. But now it needs an update! Funny that! Sods law!

stemsee · #93 Post by **stemsee** » Tue 24 Feb 2015, 18:24

Here is 4.0-rc1 kernel and modules. But no aufs. And I haven't tested yet because I don't have a full install to test it on. It should be ale to load a rufwuuf style virtual full install in initrd for boot to ram. Oh it is 64bit.

https://mega.nz/#!lBp3EJQR!521FnLuB1Cju ... LjuR3aVejA

stemsee · #94 Post by **stemsee** » Wed 25 Feb 2015, 23:08

here is 3.19 with aufs, 64bit, tested. Verbose boot.

https://mega.nz/#!wUwVSJiI!SssrKaNw-sqX ... nhFhoyYEmY

peebee · #95 Post by **peebee** » Thu 26 Feb 2015, 10:25

Hi Stemsee

I notice when I look inside your kernel-modules.sfs for 3.18.6-pae and 3.19 that they only have the /lib directory whereas the kernel packs produced by 666philb have lots of other directories.

Can I drop your kernels into woof-ce puppies like tahpup and slacko6beta just by renaming the kernel-modules.sfs appropriately - or is there more to do?

Grateful for any guidance you can supply.
Thanks
peebee

stemsee · #96 Post by **stemsee** » Thu 26 Feb 2015, 13:44

Hi peebee

the /etc dir holds DOTconfig-kernel-version
the /boot dir holds system.map
/usr/include for headers
and /lib holds firmware and modules
/usr/sbin /bin for aufs related components.

I compiled them manually... I just dropped them in place, and they are running. I haven't tried compiling with them or running apps which depend specifically on those kernel headers. I can supply the patched sources.

stemsee · #97 Post by **stemsee** » Thu 26 Feb 2015, 17:10

Linux founder, Linus Torvalds, explains the sudden move to 4.0

http://www.zdnet.com/article/linux-kern ... er-to-4-0/

Dry Falls · #98 Post by **Dry Falls** » Thu 05 Mar 2015, 18:56

stemsee wrote:Has anyone else noticed the strange events at www.kernel.org ???

Yes. k3.18.7 identifies itself as 'Unicorn' which tells me kernel.org has partnered with redhat. In Lighthouse, k3.18.7 misses the hardware clock altogether at bootup, and when the desktop appears, it goes immediately into screen-saver mode. Then it starts to get interesting. This may not be your experience, but it's how it plays out on my machine. It's almost like the kernel expects a full installation and gets confused by the puppy layered architecture -- looking in the base file for something in the save file and vice versa.

Another illustrative read is who's writing linux today? from the same source.

df

stemsee · #99 Post by **stemsee** » Thu 05 Mar 2015, 19:14

Unicorn was part of the last kernel I compiled. oops!

real time clock (rtc) run ubuild.sh then at first option enter 888, then at kernel config choose 2 make menuconfig. Then update settings to fix.

stemsee · #100 Post by **stemsee** » Sat 07 Mar 2015, 08:47

kernel 3.18.8-64

stemsee · #101 Post by **stemsee** » Sat 07 Mar 2015, 19:18

sukk-0.9
one bug fix and some small changes.

stemsee · #102 Post by **stemsee** » Sat 07 Mar 2015, 20:10

3.18.8-pae

Dry Falls · #103 Post by **Dry Falls** » Sun 08 Mar 2015, 16:57

stemsee wrote:sukk-0.9
one bug fix and some small changes.

Code: Select all

...Extracting the Aufs sources
Extracting the kernel sources
Adding Aufs to the kernel sources
Resetting the minor version number
Reducing the verbosity level
Applying ../patches/*
./ubuild.sh: line 312: ../patches/*: No such file or directory
Error: failed to apply ../patches/* on the kernel sources.
<root> /initrd/mnt/dev_save/Downloads/sukk-0.9/sukk-0.9
bash-4.2#

df

stemsee · #104 Post by **stemsee** » Sun 08 Mar 2015, 19:25

Dry Falls · #105 Post by **Dry Falls** » Thu 12 Mar 2015, 14:44

I
think I'm starting to get the hang of your build system. I merged the
/proc/config.gz from fathouse-retro and JL64 under 3.18.5 for a new
DOTconfig-64, edited build64 and version patch to suit and dropped into
Dist/sources the k3.18,8 package & aufs with Sukk.8. Everything went
well, and the new kernel, modules and sources popped out in about 56
minutes.

Still won't run chromium-sandbox. I thought it might be in the memory
allocation, but not (chrome puts everything you've been doing into the
swap drive and proceeds to fill your memory till it's used to the point
of snapping). Still searching through 6,000 lines to find the
difference (why it works right out of the box with fathouse k3.16.2, but
not any of the newer kernels I've compiled with your builds.

I'd stay with k3.16.2 but there's a lot it won't do or gets wrong, and
that won't do.

Thanks,
df

ps., it's not that I want the fancy browser, but a lot of folks do, and
I'm not going to be the one who tells them they can't! Hell, I've
already done three impossible things this morning.

And in case anyone's interested, here is the customized config for k3.18.8:

Dry Falls · #106 Post by **Dry Falls** » Fri 20 Mar 2015, 00:31

I think I've fixed it, starting with line 209:

Code: Select all

CONFIG_BPF_SYSCALL=y
CONFIG_SHMEM=y
CONFIG_AIO=y
CONFIG_ADVISE_SYSCALLS=y

[/code]

stemsee · #107 Post by **stemsee** » Sun 22 Mar 2015, 15:45

Excellent stuff! So now chrome works properly, right? I have been trying to incorporate popcorn into my own distro, and even trying to upgrade the kernel in Chromecastpup-v3 but chrome always has problems and popcorn depends on chrome (apparently).

So I will change that for all configs and release sukk-1.0 after testing.

stemsee · #108 Post by **stemsee** » Sun 22 Mar 2015, 23:15

sukk-1.0 with DryFalls config edits, small bug fix for aufs download file.

Dry Falls · #109 Post by **Dry Falls** » Mon 23 Mar 2015, 01:28

Thanks Stemsee. This link Diseased Newt might explain why syscall/bpf is the key. syscall has to be enabled. Peebee directed me (thanks, Peebe) to the kernel/modules for the fatdog700 final (k3.18.7) which he used in the upgrade of LxPup64 from 14 to 15. Well, it runs lighthouse tolerably well even though I've yet to get fd700 to boot on this machine since the change from k3.16.2 (although virtual machine will boot it!). That is how I extracted the file to compare to my config from JL64-3.18.8. Unfortunately, their config file won't recompile in your kit because of a check built in to detect changes (as near as I can tell) and terminate the compile if found. Seems kind of proprietary to me, but that's just me. Anyway, while the new compile works well in lighthouse, I couldn't guarantee it will work with other distros. For example, I couldn't get LxPup64-14 to boot with it, so please be for-warned. It may not generalize. I'm going to try it in fatdog and see if it will finally boot to a desktop, but at the moment it's not a big priority.

df

stemsee · #110 Post by **stemsee** » Mon 23 Mar 2015, 08:08

Well, your discovery has solved my chrome and popcorn problems. I have updated Chromecastpup-v3 to 3.19.2-64bit kernel.

NOTE BENE: I had to install the modules with headers to the savefile in order for them to load on boot!! They would not load as an zdrv_*_.sfs !!

So now I will remaster with modules 'in' after deleting old modules.

(old)Puppy Linux Discussion Forum

(old)Puppy Linux Discussion Forum

kernel compiling in woof-ce

DOTconfig

fixed?