dreamport

Measuring temperature using OpenWRT

2012-02-12T22:19:52Z

See my x86 solution for measuring temperature here!

I’m using USB Temper devices again, but this time they are a newer revision, so the previous app with the source code I modified doesn’t work anymore. I downloaded this source code from here (look at the comments) and modified that so I can define from which USB bus and dev I want the temperature readings from. I had to cross compile this for the router (TP-Link MR3220) running OpenWRT. You can download my modified pcsensor source from here.

First I needed the OpenWRT toolchain, which you can get by compiling the OpenWRT buildroot. Here’s a good tutorial how you can compile buildroot. There are also some notes in the OpenWRT wiki about cross compilation.

For the pcsensor app to work, I needed libusb 0.1. I downloaded the source code of v0.1.12, extracted it and created a script called doconfig in the libusb source code directory:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16

#!/bin/bash
TOOLCHAIN_PATH=/home/nonoo/Work/openwrt/openwrt/trunk/staging_dir/toolchain-mips_r2_gcc-4.5-linaro_uClibc-0.9.32
export CC=$TOOLCHAIN_PATH/bin/mips-openwrt-linux-gcc
export LD=$TOOLCHAIN_PATH/bin/mips-openwrt-linux-ld
export NM=$TOOLCHAIN_PATH/bin/mips-openwrt-linux-nm
export CXX=$TOOLCHAIN_PATH/bin/mips-openwrt-linux-c++
export AR=$TOOLCHAIN_PATH/bin/mips-openwrt-linux-ar
export OBJDUMP=$TOOLCHAIN_PATH/bin/mips-openwrt-linux-objdump
export RANLIB=$TOOLCHAIN_PATH/bin/mips-openwrt-linux-ranlib
export STRINGS=$TOOLCHAIN_PATH/bin/mips-openwrt-linux-strings
export STRIP=$TOOLCHAIN_PATH/bin/mips-openwrt-linux-strip
export AS=$TOOLCHAIN_PATH/bin/mips-openwrt-linux-as
export CFLAGS=-I$TOOLCHAIN_PATH/include
export LDFLAGS=-L$TOOLCHAIN_PATH/lib
export STAGING_DIR=$TOOLCHAIN_PATH
./configure --host mips-openwrt-linux --prefix=/home/nonoo/Work/openwrt/opt

After running it, I created a domake file also:

1
2
3
4

#!/bin/bash
TOOLCHAIN_PATH=/home/nonoo/Work/openwrt/openwrt/trunk/staging_dir/toolchain-mips_r2_gcc-4.5-linaro_uClibc-0.9.32
export STAGING_DIR=$TOOLCHAIN_PATH
make $*

I ran domake; domake install and then I got the libusb libs under /home/nonoo/Work/openwrt/opt/lib. The STAGING_DIR variable you see above is needed for the toolchain compiler. I already added these cross compiling stuff to pcsensor’s Makefile. If you want to compile it on x86, just use my older Temper app’s Makefile.

libusb is needed on the router as well, so install it with opkg install -d pendrive libusb and run ldconfig.

Using a flash drive with OpenWRT

2012-02-12T22:20:17Z

On the weekend I’ve been hacking a TP-Link MR3220 for my friend Penya. The router has a USB port and I wanted to use a USB flash drive to expand the storage space available on the router. I had two flash drives, one noname 1GB and a Sandisk Cruzer 8GB. I had problems with the noname drive, random disconnects during data writing etc. so I used the Sandisk one.

First I formatted the drive to ext4 under Ubuntu, then plugged it in the router. After logging in with SSH, I installed the needed USB storage modules: opkg update; opkg install kmod-usb-storage kmod-fs-ext4. After that I was able to mount the drive under /mnt/pendrive1 (I had to create that directory to mount to): mount /dev/sda1 /mnt/pendrive1

If it says that there’s no /dev/sda1 device, look at dmesg to find out the newly attached flash drive’s device name. I wanted automount, so I installed the needed packages: opkg install blkid block-mount e2fsprogs

I edited /etc/config/fstab:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20

config global automount
option from_fstab 1
option anon_mount 1

config global autoswap
option from_fstab 1
option anon_swap 0

config mount
option target /mnt/pendrive1
# option device /dev/sda1
option uuid "88e22364-e0c5-42ca-b4e1-d5e9fa9d21ab"
option fstype ext4
option options rw,sync
option enabled 1
option enabled_fsck 1

config swap
option device /dev/sda2
option enabled 0

Note that I set the UUID instead of the device name. It’s useful if the drive comes up under a different device name. UUID uniquely identifies it. To find out your drive’s UUID, run blkid. If you have a swap partition on your drive and want to use it as swap space, you can set it here.

After finishing with the config I ran the following commands:

1
2
3
4

rm -f /etc/fstab
ln -s /tmp/fstab /etc/fstab
/etc/init.d/fstab enable
/etc/init.d/fstab start

If the last command freezes for some reason, run rm /var/lock/fstab.lck before it. If all goes well, now you have your flash drive automounting if it’s plugged in.

To install packages on the drive, you have to add it to the opkg destination list. Edit /etc/opkg.conf:

1
2
3
4
5
6

src/gz snapshots http://downloads.openwrt.org/snapshots/trunk/ar71xx/packages
dest root /
dest ram /tmp
dest pendrive /mnt/pendrive1
lists_dir ext /var/opkg-lists
option overlay_root /overlay

(Notice the dest pendrive line, I added only that to the file.)

After that you can install packages using the drive this way: opkg install -d pendrive packagename.

It’s usefule to add the bin and sbin directories on the flash drive to the PATH, so edit it in /etc/profile:

1	export PATH=/bin:/sbin:/usr/bin:/usr/sbin:/mnt/pendrive1/bin:/mnt/pendrive1/sbin:/mnt/pendrive1/usr/bin:/mnt/pendrive1/usr/sbin

You’ll need to configure ldconfig for the dynamic libraries on the drive to work, if you don’t have ldconfig yet, install it: opkg install -d pendrive ldconfig

Add these lines to ld.so.conf:

1 2	/mnt/pendrive1/lib /mnt/pendrive1/usr/lib

And then run ldconfig every time you install a lib.

Installing Midnight Commander on OpenWRT

It’s nice to have mc on the router, so let’s install it:

1
2
3
4
5

opkg install -d pendrive mc librpc
ln -s /mnt/pendrive1/etc/mc /etc/mc
touch /etc/mc/sfs.ini
mkdir -p /usr/lib/mc/extfs.d
touch /root/.mc/cedit/Syntax

If mc says error opening terminal: xterm, then make sure you have these lines in /etc/profile:

1 2	export TERMINFO=/mnt/pendrive1/usr/share/terminfo export TERM=linux

Using a webcam on OpenWRT

Install the needed kernel modules (for my webcam, kmod-video-uvc) and the capture application (in this case fswebcam) with opkg install -d pendrive kernel kmod-video-uvc fswebcam. Create the module loader shell script /etc/init.d/webcam:

1
2
3
4
5

#!/bin/sh
insmod /mnt/pendrive1/lib/modules/2.6.39.4/i2c-dev.ko
insmod /mnt/pendrive1/lib/modules/2.6.39.4/videodev.ko
insmod /mnt/pendrive1/lib/modules/2.6.39.4/v4l2-common.ko
insmod /mnt/pendrive1/lib/modules/2.6.39.4/uvcvideo.ko

If you want it to autostart at boot, run chmod +x /etc/init.d/webcam. To capture an image, run: fswebcam –jpeg 100 –save c.jpeg -r 640×480 -v

GLX X11/Xorg problems

2012-02-12T18:00:59Z

X won’t start up with compiz? Xorg log says it’s stuck at initializing extension GLX? glxinfo segfaults in libc? You have an Nvidia card and tried everything to bring life to compiz?

Boot into recovery mode, reinstall the proprietary Nvidia driver, delete /etc/ld.so.nohwcap and suddenly everything starts to work again after hours of frustrated X torture. Of course I’ve found this forum post after I found out the same solution… :)

Esprimo V5535 Ubuntu video driver install (SiS671)

2011-12-24T17:48:58Z

- Download the driver source from here
- Extract it, run ./configure and install all needed *-dev packages
- Install x11proto-gl-dev and mesa-common-dev
- Run make, if it stops with an error, open the corresponding .c file and uncomment the line which gave an error (for example it tries to include mibank.h, but it’s no longer available and even it doesn’t needed for compilation). There are about 3-4 lines to uncomment.
- After the process finished, look for src/.libs/sis671_drv.so. Copy it to /usr/lib/xorg/modules/drivers.
- Run Xorg :1 -configure, move xorg.conf.new from your home dir to /etc/X11, and replace fbdev and vesa to sis671 in the lines starting with Device
- Execute sudo service lightdm restart and now the resolution should be 1280×800
- To fix the console: add blacklist vga16fb to /etc/modprobe.d/blacklist-framebuffer.conf
- If you still have problems, check /var/log/Xorg.0.log

No drives in Ubuntu installer

2011-12-24T15:43:09Z

I tried to install Ubuntu to a desk computer but there were no disks/partitions listed in the installer. Switching to a console and running fdisk -l listed the disk and I was able to partition it using fdisk, formatted to ext4 but it still wasn’t listed in the installer.

The solution: I used this disk in a motherboard RAID setup earlier, I just had to enable the RAID controller in the BIOS, enter the controller’s setup and remove RAID information from the disk. After that, I was able to partition the disk and continue Ubuntu install.

Remote controlled duck

2011-12-21T20:41:49Z

I got this from my friends as a birthday present. A little slow because of the foam, but light speed fast on clear international waters :)

Forwarding sound on LAN

2011-12-17T06:34:42Z

Here’s how to simply forward sound from one computer to another in Ubuntu.

First, install paprefs:

1	sudo apt-get install paprefs

Start it by executing paprefs in a terminal. If every checkbox is grayed out, you have to make a symlink to the PulseAudio modules directory as paprefs searches for the modules in the wrong place:

1
2
3

sudo ln -s /usr/lib/pulse-0.99 /usr/lib/pulse-0.99.3
# or for a newer PulseAudio version:
ln -s /usr/lib/pulse-1.0 /usr/lib/pulse-1.0.0

(Replace the version numbers with your version numbers.)

So after you successfully started paprefs, make the following settings on the server machine:

If you have multiple soundcards, paprefs has this great setting:

On the client side, check this checkbox:

Now restart PulseAudio on both computers with:

1	pulseaudio -k

And then press Alt+F2, enter pulseaudio. Now you should be able to select the server’s sound card(s) on the client side in the Sound settings dialog (click on the speaker icon on the panel, and select Sound Settings… from the menu):

Note: there’s another method for sound forwarding using multicast RTP, see this guide.

Volume control shell script

2011-12-16T20:29:29Z

Here’s a simple shell script I wrote for controlling PulseAudio volume from shell under Linux. It takes two parameters: up and down and sets volume for all sinks.

1
2
3
4
5
6
7
8
9
10
11
12

#!/bin/bash
OLDIFS=$IFS
IFS=$'\n'
for L in `pacmd dump | grep set-sink-volume`; do
if [ $1 = "up" ]; then
pacmd set-sink-volume `echo $L | awk '{print $2}'` `echo $L | awk '{print $3+5000}'`
fi
if [ $1 = "down" ]; then
pacmd set-sink-volume `echo $L | awk '{print $2}'` `echo $L | awk '{print $3-5000}'`
fi
done
IFS=$OLDIFS

My bday party 2011

2011-12-05T14:57:22Z

]]>

32bit OpenGL with wine

2011-12-02T21:37:35Z

Do you have problems running OpenGL apps under Wine with an Nvidia card and a 64bit Linux? Wine says X11DRV_WineGL_InitOpenglInfo couldn’t initialize OpenGL?

Here’s the fix.

- apt-get install ia32-libs, if it’s already installed and you still have problems, move on to the next step
- move everything from /usr/lib/i386-linux-gnu to /usr/lib32 EXCEPT /usr/lib/i386-linux-gnu/glib2.0.
- extract the official 64bit Nvidia driver: ./NVIDIA-Linux-x86-290.10.run -x
- copy libGL.so.290.10 (or newer) and the tls directory from Nvidia driver’s directory called 32 to /usr/lib32
- create symlinks libGL.so.1 and libGL.so in /usr/lib32, they should both point to libGL.so.290.10
- run sudo ldconfig

That’s it. Now wine should run OpenGL apps without problems.