SliTaz Forum
You are not logged in.
Pages: 1
So I download v5 iso and install in virtualbox. Cant get X to work. No screen found. No matter
Tazpkg recharge -> Tazpkg upgrade
I install Slitaz-toolchain, bc, ncurses-dev, perl, and linux-source.
I issue: /usr/bin/get-linux-source
and it hangs for as long as I let it. Then it dawns on me that Slitaz does not ship with working internet so I download wget with Tazpkg -gi wget.
Reissue command. /usr/bin/get-linux-source
It completes!
I cd to /usr/src/linux-3.16.55
Guess what I see. Only Slitaz patches.
http://doc.slitaz.org/en:guides:kernel
That is the only thing that shows up for building kernel for documentation.
Does anyone have a working method on how to rebuild kernel for Slitaz 5? I need to do this as many items come with acpi atlas button controller including some models of eeepc by Asus. My hardware in particular has this needed functionality as there is a directional pad on the front of my old hardware tablet computer: webdt366. Slitaz has never included this as a kernel module or as built in so I am forced to rebuild. Only tools are not working.
Is there anything I am not doing correctly? Is there better documentation to refer to?
I went in to the 3.16.55 directory and manually tried to wget the correct kernel from kernel.org however when I went to decompress it the tar error indicated that there were too many levels of symlinks. Move it out one directory and it decompresses fine but I can see from here that unless your integrations get included I will not be able to make a .tazpkg out of my efforts.
So please team Slitaz. Please do enlighten on how a user in need of redoing the kernel can accomplish this on v5?
Offline
atlas_btns.ko.xz is now owned by linux-input-misc-3.15.65.tazpkg
CONFIG_INPUT_ATLAS_BTNS: x86 Atlas button interface
General information
The Linux kernel configuration item CONFIG_INPUT_ATLAS_BTNS:
prompt: x86 Atlas button interface
type: tristate
depends on: CONFIG_X86 && CONFIG_ACPI
defined in drivers/input/misc/Kconfig
found in Linux kernels: 2.6.21–2.6.39, 3.0–3.19, 4.0–4.20, 5.0–5.6, 5.7-rc+HEAD
modules built: atlas_btns
lkddb acpi "ASIM0000" : CONFIG_INPUT CONFIG_INPUT_ATLAS_BTNS CONFIG_INPUT_MISC : drivers/input/misc/atlas_btns.c # in 2.6.23–2.6.39, 3.0–3.19, 4.0–4.20, 5.0–5.6, 5.7-rc+HEAD
lkddb module atlas_btns CONFIG_INPUT_ATLAS_BTNS : drivers/input/misc/Kconfig : "x86 Atlas button interface" # in 2.6.21–2.6.39, 3.0–3.19, 4.0–4.20, 5.0–5.6, 5.7-rc+HEAD
Offline
I understand logic of staying mainline as updates do not require a rebuild to initiate.
That being said it would still be nice to build my own kernel. I have a 44 pin ata header with a proprietary dom or disk on module. Manufacturer grounded pin 44 so if I buy generic drive it fails. Buy it from them for 10x pricing ind it works. Once known its not hard to work around however on Slitaz V5:
The end result is if I build the kernel and add in /ATA/CS5535 as built in my hdd gets recognized as /dev/hda which is contextually acurrate.
If I use slitaz 3.16.55 out of cooking I have to add wireless, bluetooth, drm and a whole bunch else including input-misc. My HDD gets seen as /dev/sda on rolling and this is not 100 percent accurate. AS soon as I go to save with writefs gzip /dev/sda1 I get superblock error that goes in to aborting journal as journal block not found.
I have done this three times now to get same results. I make iso in tazlito with additions. Put it to usb using lili loader version 2. Boot hardware and check lsmod and a few other things. In each case I get failed login command on slim and see that tux has kmap and lang errors. In fact I chown tux:tux and get user / group error. Follow past posts for same problems to no solution. So I install and figure I can fix errors once reboot. I go to do my first save and locale/keymap fails and each time I get corrupted rootfs.gz. Cant boot to saves. Save just finished and all commands result in errors that revolve around corrupted filesystem.
Any chance someone has way of building kernel they can offer that works in slitaz 5? I feel that with proper /dev/hda association I have better chance. Want to try with cs5535 built in. Feel sda getting rare disk on module incorrect through old geode system on a chip interface. This chip was just installed and was running a perfect operating puppy linux. This resulted the same on the previous chip as well. Chip = hdd. Would like to use slitaz 4 and slitaz 5 but kernel building not working. After installing true wget and running get-linux-source I see /usr/src/lnux-3.16.55 showing only slitaz patch directory contents. How does one get kernel to download too?
Offline
Turns out that cacerts was the missing glue needed for the scripts to complete. In make menuconfig now. Hoping this goes better than V4 that fails on make tazpkg.
Offline
This is how far I was able to get. Error near the end of kernel building. See attached picture.
Offline
This forum... Attach image first. The type...
Offline
size rules on image uploads ........
[attachment=48961,2987]
Offline
At first I attempted to set processor type to geode error as above happened.
Tried to set back to neutral pentium pro, sock settings, above error happened again. Issued make bzImage && make modules only after make clean
This is on a slitaz-v5 iso in virtualbox vm on imac intel 4 core with 32gb ram. 12 gb reserved for live image running as live image. I cant get gui on slitaz-v5, No screens found so I run as live. More than enough ram.
I installed slitaz-toolchain, perl5, bc, ncurses-dev, then linux-source, install bc and cacerts and run /usr/bin/get/linux-source and make menuconfig
here I search out all “geode” with “/“ as well as / for “cs5535” and make sure ata is built in. most else I leave as modules. penmount touchscreen, alsa cs5535 within pci, etc.
When done my kernel (on slitaz v3) identifies hdd as /dev/hda and recognizes all hardware but has hard time with atlas_btns. Depending on when I modprobe in init depends in /sysfs/ inclusion or /ev/input/event (evdev = better and working). I think this is dependent on when acpid starts it should hand of to hal —listener but on your os it gets included as sysfs /devices/virtual... not going to work...
On v4 I have similar errors as v5 but have yet to install cacerts package.
On v5 I cant complete kernel build due to slitaz script errors that I cant figure out. Been on this OS for 20 years. Can someone from slitaz dev team please offer help on what this error is? On V5 do you have documentation on what is needed for kernel building? I may be on old docs....
Most slitaz scripts offer no verbosity. While I respect what your work is capable of I feel that I am in am in-between non existant or lacking documentation that does not exist as well as lack of script verbosity that alludes to true error. May I ask for help here? What can I provide that can help diagnose error with your tools?
Kernels are mostly agnostic. If I need to build in chroot let me know. Technically I shouldn’t. Unless you changed gcc for the builds.....
If I run v5 stock with all sorts of slitaz kernel module additions and on cooking kernel with this hardware I cant get slim to make an xserver. It shows the login screen but fails to log in. Claims it cant log in or execute login command. I I install the /dev/sda gets corrupted as its really an old ata 44 pic device. Cant reboot after saving o writefs command. All fs is corrupted. This no login command was also due to no tux:tux user/group avail. On past installs to ext3 partition I see oddities as well. I would cd to /home/tux and see /skel. Nothing else. I would manually copy contents of skel out and delete skel but you have scripts that build this os that often fall on their face. You should figure out monitoring scripts as well or make your scripts check what they have done to validate before moving on.
Do you include all minimum packages within talito-flavors? I mean recipes? Not sure I see all deps for jwm in jwm flavor is what I am suggesting.
I see on V5 that tazlito takes more time to make iso image. Good idea. Check md5 on installs. Process/check is safer than blast and hope.
Agan, your tools fail with no indication as to why and no documentation is searchable for my problem. Developers please hint as to what is not working correctly.
Offline
I comented out the trap line as it was sending invalid signal specification. Kernel built, modules built, make tazpkg failed with various mkefile errors.
Traps are things that devs put in to make sure their scripts are running as expected. In this case the trap looks to be the problem. Tried on inel and geode for processor type.
I figured out how to build kernel on v4 Sitaz. Edit the /usr/bin/get-linux-source script to not use busybox and prefer wget. Install wget and cacerts from slitaz v5 and now script comopletes and I can build kernel and tazpkg. On V5 there are script errors still.
Offline
Hi, First at all sorry for my english
My two cent:
Not build on virtualbox, is recommended make only on real installation
skip get-linux-source and wget source direct from kernel.org (https://mirrors.edge.kernel.org/pub/linux/kernel/)
unpack the source (tar -xvf or similar)
zcat /proc/config.gz > .config (config of actual kernel)
make menuconfig and load the .config created
edit on menuconfig the option requested for your system
make and make install create vmlinux on folder boot (new kernel to load from grub)
make modules_install create the folder and module of new kernel on folder /lib/module/
enjoy
Offline
Here is an old guide to compile SliTaz kernel
https://tuxonomy.wordpress.com/2010/04/02/kernel-2-6-33-1-on-slitaz/
Offline
I have two versions of target hardware. One has 256mb of ram and a 500mb hdd. The other has 500mb of ram and a 4gb hdd. I am not sure that either will be suitable build hosts. I have made kernels in v3 and in v4 by changing the get-linux-source code to reflect wget instead of busybox and script works. This is not where I have problems.
I have problems making tazpkg as authors of scripts use hooks that are not working for a cross compile. I am not truly cross compiling I am making custom kernel to geode specific hardware. When tazpkg runs it fails each time at make, showing errors that do not tell anything.
Old kernel build guides do not cover scenario. They need to be updated. This is why I cal to developers to help. Their tools are not working.
I appreciate Gibor but I do not think my hardware will allow me to build kernel in traditional method as ram will only expand to 500mb. This is why I try in VM.
Next question will be why tazito feels in needs to rewrite and download new kernel when I manually put it in to /packages folder? Does anyone have a way to tell tazlto not to download what it mis identifies as dependencies? I want bluetooth but I build in to my kernel.. Tazlito downloads linux-bluetooth and modules fail to load as they are not build for my geode specific kernel. Unregognized module format. I have pata hardware and sata/cs5535 causes corrupted saves and will not work. I cant use stock kernel and add slitz-package/modules. Need to build my own.
How can I convert tazlito on v4 to ask for dependency solving instead of assuming. Or, can I install tazlito from version 3 in to version 4? Tazlito in V3 does not have this behavior.
Offline
Pages: 1
[ Generated in 0.018 seconds, 7 queries executed - Memory usage: 1.56 MiB (Peak: 1.77 MiB) ]