Van pár feladat aminek a végrehajtását szeretem elkerülni a számítógépemen. Ilyen például a Qt fordítás. Baromi sokáig tart, és végig üvegeli a a gépet. A gentoos "compile 5 hours to run 5 ns faster" filozófia pedig eddig még nem érintette meg a lelkemet.

Beágyazott szuttyok esetén nem mindig adatik meg az a lehetőség, hogy bináris csomagot használjon az ember. Így volt ez most is, amikor egy Friendlyarmos panelre kellett Qt-t varázsolnom. Az openembedded sajnos csak úgy képes Qt-t fordítani, ha az először lefordítja az x86-osat, majd az embeddedet. Ezt el kívántam kerülni, ezért 2 napig reszeltem a bitbake és az mkspec fájlokat, de végül feladtam.
Képtelen volt desktopon futó qmaket összehozni, így végül ezt a leírást használtam:

http://billforums.station51.net/viewtop ... p;p=44#p44

Illetve megspékeltem az opciókat ezekkel:


./configure -embedded arm -xplatform qws/linux-arm-g++ -prefix /usr/local/Qt -little-endian -no-qt3support -opensource -no-accessibility -no-webkit -no-script -no-javascript-jit -no-scripttools -nomake examples -nomake demos -nomake docs -nomake translations -qt-mouse-tslib -DQT_KEYPAD_NAVIGATION -reduce-relocations -fast


A fordítás egy Intel T7400 C2D-n kevesebb mint 20 perc alatt lezajott.

Öröm van a köbön.

I have put a dot to the end of a long story.
About three years ago I have get an Sky networks rebagged Netgear DG834GT from my friend Azbeszt.

First of all I have disassembled it, and realized that it has an unsoldered USB port. I have placed the soldered the missing components, (connector, voltage regulator, and the D+ D- protector resistors), but it did not detected any USB devices. I have talked with other people who owns a BCM63XX based routers, and one of them have done this trick on a Comtrend router:

http://img530.imageshack.us/img530/2545 ... 6v2xp2.gif

He notified us, about he had to pull up a pin to enable the port.
I have populated all of the unpopulated pullup resistors place, but I have not succeded.

As the time went on I have purcashed two other unit.
So I have decided to desolder a CPU one of them.
The desoldering was done in my friend's lab.

According to this pinout:
http://www.f-x.fr/wikini/wakka.php?wiki=Bcm6348PinOut/

Only the D+ D- and the USB_FLT was routed out from the BGA.
I have traced the USB_FLT trace, but I have not found the other end of the stripe. It does not seems to be connected to anywhere with desoldered CPU. With populetated CPU it has ~90KOhm resistance to the ground. Fortunatelly the trace is connected to an stripe in the inner layer with a trough hole via. So very carefully it is possible to solder a wire to the via.

The via is located in the half way between C702 and C423.
To locate it flip your board to face to the soldering side of the PCB. Rotate it to have the LAN connectors closer to you.


This area will be in the upper side of it:


There is a drawing which via is it:


(please note that this is not scale drawing)

Solder a thin Cuz wire to it and connect it to the 3V3. The easiest place to find the 3V3 is the serial connector (J503) third pin.
Connect to it with an 2K resistor.

Power up your device, and do
cat /proc/bus/usb/devices

And you should see your device details there.
Or you can use the lsusb (from usbutils package).

The additional information about the hack's details could be found here:
http://wiki.openwrt.org/oldwiki/openwrt ... ar/dg834gt

Currently it works for me with an r18196 of the trunk. I will try the bleeding edge version asap.

I had some problem with my laptop's microphone input with the Ubunutu 10.04.

I have inserted the following line to the end of the /etc/modprobe.d/alsa-base.conf


options snd-hda-intel model=3stack-660
options snd-hda-intel probe_mask=1


And now I have the Mic, and the Front Mic and the line channel in the alsamixer. The microphone connector, is the MIC in the alsamixer, and the Mic2 in the system's sound preferences.

BTW, I have disassembled it, and found a pinhead connector near the soundcard IC (ALC660), so I think it is possible to add an internal microphone.

Just a small note to myself, because I have forgotten it again.

Connect the RS board to your PC trough it's WAN port.
Give static IP to your PC in the 192.168.1.255 subnet.
(IP 192.168.1.1 mask: 255.0.0.0 gw: 192.168.1.20)

cd to your Openwrt's bin directory. (bin/ar71xx)
Power up your RS (trough POE) while holding down it's reset button.
You can release it when the RF led starts light.
Run tftp in the bin/ar71xx directory:

$tftp
binary
trace
put openwrt-ar71xx-ubnt-rs-squashfs-factory.bin


If the process is successful then wait 5-6 minutes. Once the upgrade is finished you the RS will reboot itself.
You will notice it because the wired connection is downed during the reset.

After the reboot it will wait a few minutes, because it erases jffs2 filesystem, and it is takes a long time.
Don't panic, it will finish it.

Today I wanted to print some documents. Click print, wait nothing happends. I have checked the CUPS command panel, and it said, that the
/usr/lib/cups/filter/pdftopdf failed. Happiness. I have tried to run it, and

mm@lapos:~$ /usr/lib/cups/filter/pdftopdf
/usr/lib/cups/filter/pdftopdf: symbol lookup error: /usr/lib/cups/filter/pdftopdf: undefined symbol: _ZN13GfxColorSpace5parseEP6Object


This happends on my notebook running Ubuntu 9.10 Karmic Koala x86_64.

I remember that there were a CUPS update few days ago. I have forced back the old version of cups, but it did not helped. So I have made a bit workaround with

mm@lapos:~$ sudo mv /usr/lib/cups/filter/pdftopdf /usr/lib/cups/filter/pdftopdf2
mm@lapos:~$ sudo /etc/init.d/cups restart

AAnd happiness.
Now I feel that once I will be angry to myself about this action, but currently I can print, and this is what is matter.