Tiny Core Linux
Tiny Core Base => Raspberry Pi => piCore Test Releases => Topic started by: bmarkus on February 04, 2014, 12:26:00 AM
-
First test release of piCore v5.2 available.
http://tinycorelinux.net/5.x/armv6/release_candidates/
Change log:
* rebuildfstab: do not replace fstab entries for a device that does not have "Added by TC" on the line (thanks to Gerald Clark)
* init: increase the default inode count
* ondemand: don't list extensions under subdirs in onboot maintenance
* busybox updated to 1.22.1 and split suid/nosuid for better security
* ldd: Added quotes for binaries with spaces in their names
* /etc/services: modified to suit rpcbind rather than portmap
* tc-functions: Removed the getpasswd stars to allow backspace to work
* added accelerated arm-mem library to spead-up memcmp, memcpy, memmove & memset
* updated Raspberry Pi firmware
-
Hi bmarkus
I will test the new version soon.
What kernel are you using.
The new 3.12.9 has some interesting development, in the i2s area, also the new firmware has better usb audio handling.
Steen
-
Kernel is not updated, see list of changes in the first post. Goal is to have 5.2 changes from x86 adapted + arm-mem library.
Kernel is changing too often, I do not like to follow it continously with all consequencies in every 1-2 weeks. RedHat/CentOS is on 2.6.32 :) When 3.12.9 would be finished maybe 3.14.0 will be there or, 3.13.3? Why not 3.13.x?
Can you explain a bit new features?
-
Hi Belá
As you know I very interested in the audio part of raspberry and microcore, so therefore my focus might be too narrow.
However, some of the interesting changes are:
Firmware: https://github.com/raspberrypi/linux/issues/320 . this issue describes crackling noise in the HDMI output, but also via an USB_DAC. This is a problem often encountered when using raspberry and an external USB-DAC. However, this was added to the firmware 2 days ago.
Kernel: The I2S connection is very interesting for an audio player build on the raspberry - for the exact reasons mentioned above. By using I2S we don't see the same cracks and pops in the audio as we see using USB-DACs.
The current kernel in microcore supports the HiFiBerry I2S DAC (and also other DACs) with great results. But the new kernel 3.12.9 supports a pure digital I2S HiFiBerry card http://www.hifiberry.com/hbdigi (http://www.hifiberry.com/hbdigi) which will give our raspberry a digital audio output.
https://github.com/raspberrypi/linux/commit/a76cc555cbac2a2136a7a0b4078381704a57d59d (https://github.com/raspberrypi/linux/commit/a76cc555cbac2a2136a7a0b4078381704a57d59d)
By the way I'm not in any way connected with the company selling HiFiBerry audio cards, I just like the cards, and they are at the moment the only available I2S audio cards for our raspberry.
I would like to support the addition of digital out (S/PDIF or Toslink) in the piCorePlayer, therefore I had started building the newer kernel for microcore, but then I read this thread and thought that you might be working on the same.
Steen
-
Hi Steen
I like your player and I delayed 5.1 to make the I2S DAC support working correctly :) Problem is that kernel is changing too fast and also as you remember around 3.12.2---5 there were broken unworking releases. This 3.12.7 is a stable version and I will replace kernel later a bit when more new features accumulated. Of course custom kernels for appliances like your are welcome as well as feedback, comments, etc.
5.2 script base looks OK, I'm using 5.2alpha1. What is happening is a repo cleanup specially to eliminate use of libiconv.tcz and at the same time moving accelared arm-mem function to eglibc-2.18 source tree as a new port to gain performance by default not with so overwriting as it is now.
Kernel 3.13.y is too young, it is not consider at the moment.
-
I remember linux from years ago.....anytime you are on the latest stable tree, you get updates like mad. But that is the price of cutting edge.
The Updates in 5.2 sound good
-
I have a model A setup for testing for another project, so I downloaded this to test (I downloaded the SSH version).......the card loaded fine, but then I realized that the model A does not have a network connection without wireless.
So I mounted the card on a different machine and manually installed the Wifi extensions onto the card.....
Perhaps adding wifi to the ssh image would be helpful.
Sadly the 8192cu Kernel Modules are still broke. I'll just install my 3.12.7 kernel with modified drivers.........there has been not changes to this in the 3.12.9 kernel either.
I'll get everything loaded up and see what happens.
One annoyance that has been there since the last version.. Running filetool.sh -b. There is a linefeed/cr missing at the end of the output.
-
Perhaps adding wifi to the ssh image would be helpful.
It would increase significantly the size of the image and TC's beuaty (one of) is its small size. Also I do not want to offer one more prebuilt image. Maybe, piCore-X with preinstalled SSH and WIFI parts is a good choice. Questin is how to configure WiFi connection at the first time w/o wired net. Either from console or some automatic connection (it was alerady discussed in the forum if I'm right). Third option is configure offline on another machine editing SD card content.
Sadly the 8192cu Kernel Modules are still broke. I'll just install my 3.12.7 kernel with modified drivers.........there has been not changes to this in the 3.12.9 kernel either.
Can we add your module to the repo?
One annoyance that has been there since the last version.. Running filetool.sh -b. There is a linefeed/cr missing at the end of the output.
Will check
-
It would increase significantly the size of the image and TC's beuaty (one of) is its small size. Also I do not want to offer one more prebuilt image. Maybe, piCore-X with preinstalled SSH and WIFI parts is a good choice. Questin is how to configure WiFi connection at the first time w/o wired net. Either from console or some automatic connection (it was alerady discussed in the forum if I'm right). Third option is configure offline on another machine editing SD card content.
On a Model A Rpi, setting up Wifi has to be done from console. Model B's worked fine with just SSH. The struggle was getting all of the wifi dependencies and extensions. There are 6 or 7 different extensions and MD5 files that had to be grabbed manually with wget.
I've had problems with module loading compatibility by just replacing a single driver....so I now just replace the whole kernel and rebuild all of my module extensions that I need. I'm using the linaro cc toolchain provided on the RPi github. The only modification to the 8192cu driver is to edit the makefile and change the option CONFIG_POWER_SAVE to 'n'. before compiling. There is something really messed up in the driver, as if it goes into power save, the driver quits responding to arp requests even after wakeup. I've tried to disable the power savings through the /etc/modprobe.conf but that is not working either.
Can you compile busybox without the password strength checking? I have to keep my own version of busybox around, just to set the password that I want to use.
-
Can you compile busybox without the password strength checking? I have to keep my own version of busybox around, just to set the password that I want to use.
No, it is the standard behavior of Linux for users, except root. As root you can create a weak password for a user, there is just a warning.
-
The only modification to the 8192cu driver is to edit the makefile and change the option CONFIG_POWER_SAVE to 'n'. before compiling. There is something really messed up in the driver, as if it goes into power save, the driver quits responding to arp requests even after wakeup. I've tried to disable the power savings through the /etc/modprobe.conf but that is not working either.
I will update the kernel module with this change.
-
Can you compile busybox without the password strength checking? I have to keep my own version of busybox around, just to set the password that I want to use.
No, it is the standard behavior of Linux for users, except root. As root you can create a weak password for a user, there is just a warning.
Didn't know root could do that..... thanks for the tip.........(Deleting busybox source :)
-
Driver updated in repo, please try
-
Driver updated in repo, please try
Will do..
I was digging down into the driver on github. The top level directory for rtl8192cu shows no recent modification, but digging down the directory tree, there was a patch made to the driver 9 days ago (Jan-28) I will test the latest 3.12.9 kernel too.
-
Driver updated in repo, please try
http://distro.ibiblio.org/tinycorelinux/5.x/armv6/tcz/wireless-3.12.7-piCore+.tcz.md5.txt
is a zero length file. I downloaded manually and generated a md5sum for the file.
The tcz looks intact, but when I try to load the module, I get an invalid module format error. That is what happens when I try to match modules that I compile with kernels that I didn't compile.
-
Driver is not fixed in 3.12.9 either.
Currently running the 5.2 alpha core /w 3.12.9 kernel.
-
Driver updated in repo, please try
http://distro.ibiblio.org/tinycorelinux/5.x/armv6/tcz/wireless-3.12.7-piCore+.tcz.md5.txt
is a zero length file. I downloaded manually and generated a md5sum for the file.
The tcz looks intact, but when I try to load the module, I get an invalid module format error. That is what happens when I try to match modules that I compile with kernels that I didn't compile.
Fixed, please retry
-
Fixed, please retry
I'll give it a try tonight..... What do you think the problem with the module was?
-
I'll give it a try tonight..... What do you think the problem with the module was?
Our kernel version is 3.12.7-piCore+ but if you are compiling kernel in a clean source, it will be 3.12.7-piCore without trailing + which indicates 'dirty' or modified source which is the case of git cloned RPi kernel.
-
I'll give it a try tonight..... What do you think the problem with the module was?
Our kernel version is 3.12.7-piCore+ but if you are compiling kernel in a clean source, it will be 3.12.7-piCore without trailing + which indicates 'dirty' or modified source which is the case of git cloned RPi kernel.
Hi bmarkus, how do you solve this naming problem?
If I define the name in the config file before compiling, is that the correct way to solve it?
Steen
-
Our kernel version is 3.12.7-piCore+ but if you are compiling kernel in a clean source, it will be 3.12.7-piCore without trailing + which indicates 'dirty' or modified source which is the case of git cloned RPi kernel.
Hi bmarkus, how do you solve this naming problem?
If I define the name in the config file before compiling, is that the correct way to solve it?
Steen
I'm using the git cloned version. There is a script in the source tree which evaluates wether + is needed or not. It my help:
http://stackoverflow.com/questions/19333918/dont-add-to-linux-kernel-version
Note exactly for this case, but can be useful:
http://tjworld.net/wiki/Linux/Kernel/Build/CustomiseVersionString
-
I understand the '+' or no '+'. But how would the driver that you recompiled not matched your version of the kernel?
-
By accident built a wrong tree. Late evening.
-
By accident built a wrong tree. Late evening.
Ok, I completely understand.
-
when running tce-update, since I had extensions that were not on the repo, there were errors, but the error log /tmp/upgrade_errors.lst is empty.
The wireless kernel drivers updated appropriately.
rtl8192cu Wifi driver is fine now.