Advent Vega kernel source code now available!

[Guest foxsen]

Eduardo, I have tried your touchscreen patch and it does not fix my problem.

After more investigation, I find a simple way to hit the problem and know more about the status:

1, If i keep press on the screen during the driver loading time, the touchscreen will almost surely dead. (the problem is not reproduced with shuttle test rom). Even if I built the driver as a module, and press on the touchscreen during insmod, it will die.

2, when the touchscreen dies(i.e. lost reponse or behave strange to user), the driver is normally in a state with endless interrupts, instead of no reponse. The "no acknowledge" seems to be not fatal. An example out is:

<6>[ 9698.565179] it7260 4-0046: [0] - X:673, Y:153, P:8, V:1

<6>[ 9698.570509] it7260 4-0046: [1] - X:849, Y:302, P:8, V:1

<6>[ 9698.575827] it7260 4-0046: [2] - X:0, Y:0, P:0, V:0

<4>[ 9698.580844] 7260 irq 0

<6>[ 9698.583946] it7260 4-0046: processed points:

<6>[ 9698.588289] it7260 4-0046: [0] - X:673, Y:152, P:8, V:1

<6>[ 9698.593665] it7260 4-0046: [1] - X:849, Y:302, P:8, V:1

<6>[ 9698.598955] it7260 4-0046: [2] - X:0, Y:0, P:0, V:0

<4>[ 9698.604005] 7260 irq 0

<6>[ 9698.607149] it7260 4-0046: processed points:

<6>[ 9698.611526] it7260 4-0046: [0] - X:673, Y:153, P:8, V:1

<6>[ 9698.616908] it7260 4-0046: [1] - X:848, Y:302, P:8, V:1

<6>[ 9698.622251] it7260 4-0046: [2] - X:0, Y:0, P:0, V:0

<4>[ 9698.627266] 7260 irq 0

<6>[ 9698.630356] it7260 4-0046: processed points:

<6>[ 9698.634709] it7260 4-0046: [0] - X:673, Y:153, P:8, V:1

<6>[ 9698.640072] it7260 4-0046: [1] - X:849, Y:302, P:8, V:1

<6>[ 9698.645378] it7260 4-0046: [2] - X:0, Y:0, P:0, V:0

<4>[ 9698.650393] 7260 irq 0

<6>[ 9698.653580] it7260 4-0046: processed points:

<6>[ 9698.657925] it7260 4-0046: [0] - X:673, Y:152, P:8, V:1

<6>[ 9698.663321] it7260 4-0046: [1] - X:848, Y:302, P:8, V:1

<6>[ 9698.668637] it7260 4-0046: [2] - X:0, Y:0, P:0, V:0

<4>[ 9698.673691] 7260 irq 0

<6>[ 9698.676837] it7260 4-0046: processed points:

<6>[ 9698.681218] it7260 4-0046: [0] - X:673, Y:152, P:8, V:1

<6>[ 9698.686594] it7260 4-0046: [1] - X:849, Y:302, P:8, V:1

<6>[ 9698.691901] it7260 4-0046: [2] - X:0, Y:0, P:0, V:0

<4>[ 9698.696919] 7260 irq 0

<6>[ 9698.700011] it7260 4-0046: processed points:

<6>[ 9698.704363] it7260 4-0046: [0] - X:673, Y:153, P:8, V:1

<6>[ 9698.709727] it7260 4-0046: [1] - X:848, Y:302, P:8, V:1

<6>[ 9698.715051] it7260 4-0046: [2] - X:0, Y:0, P:0, V:0

<4>[ 9698.720018] 7260 irq 0

<6>[ 9698.723243] it7260 4-0046: processed points:

<6>[ 9698.727610] it7260 4-0046: [0] - X:673, Y:152, P:8, V:1

<6>[ 9698.732952] it7260 4-0046: [1] - X:849, Y:302, P:8, V:1

<6>[ 9698.738247] it7260 4-0046: [2] - X:0, Y:0, P:0, V:0

<4>[ 9698.743211] 7260 irq 0

<6>[ 9698.746373] it7260 4-0046: processed points:

<6>[ 9698.750714] it7260 4-0046: [0] - X:673, Y:153, P:8, V:1

<6>[ 9698.756016] it7260 4-0046: [1] - X:849, Y:302, P:8, V:1

<6>[ 9698.761376] it7260 4-0046: [2] - X:0, Y:0, P:0, V:0

<4>[ 9698.766336] 7260 irq 0

<6>[ 9698.769482] it7260 4-0046: processed points:

<6>[ 9698.773857] it7260 4-0046: [0] - X:673, Y:153, P:8, V:1

<6>[ 9698.779171] it7260 4-0046: [1] - X:849, Y:302, P:8, V:1

<6>[ 9698.784577] it7260 4-0046: [2] - X:0, Y:0, P:0, V:0

<4>[ 9698.789546] 7260 irq 0

<6>[ 9698.792722] it7260 4-0046: processed points:

<6>[ 9698.797141] it7260 4-0046: [0] - X:673, Y:153, P:8, V:1

<6>[ 9698.802447] it7260 4-0046: [1] - X:849, Y:302, P:8, V:1

<6>[ 9698.807811] it7260 4-0046: [2] - X:0, Y:0, P:0, V:0

<4>[ 9698.812772] 7260 irq 0

<6>[ 9698.815864] it7260 4-0046: processed points:

<6>[ 9698.820275] it7260 4-0046: [0] - X:673, Y:153, P:8, V:1

<6>[ 9698.825578] it7260 4-0046: [1] - X:849, Y:302, P:8, V:1

<6>[ 9698.830937] it7260 4-0046: [2] - X:0, Y:0, P:0, V:0

<4>[ 9698.835915] 7260 irq 0

<6>[ 9698.839062] it7260 4-0046: processed points:

<6>[ 9698.843510] it7260 4-0046: [0] - X:673, Y:152, P:8, V:1

<6>[ 9698.848823] it7260 4-0046: [1] - X:849, Y:302, P:8, V:1

<6>[ 9698.854240] it7260 4-0046: [2] - X:0, Y:0, P:0, V:0

<4>[ 9698.859188] 7260 irq 0

<6>[ 9698.862280] it7260 4-0046: processed points:

<6>[ 9698.866699] it7260 4-0046: [0] - X:673, Y:153, P:8, V:1

<6>[ 9698.872011] it7260 4-0046: [1] - X:848, Y:302, P:8, V:1

<6>[ 9698.877379] it7260 4-0046: [2] - X:0, Y:0, P:0, V:0

<4>[ 9698.882339] 7260 irq 0

<6>[ 9698.885419] it7260 4-0046: processed points:

<6>[ 9698.889839] it7260 4-0046: [0] - X:673, Y:153, P:8, V:1

<6>[ 9698.895163] it7260 4-0046: [1] - X:849, Y:302, P:8, V:1

<6>[ 9698.900479] it7260 4-0046: [2] - X:0, Y:0, P:0, V:0

<4>[ 9698.905534] 7260 irq 0

<6>[ 9698.908683] it7260 4-0046: processed points:

<6>[ 9698.913074] it7260 4-0046: [0] - X:673, Y:152, P:8, V:1

<6>[ 9698.918441] it7260 4-0046: [1] - X:849, Y:302, P:8, V:1

<6>[ 9698.923750] it7260 4-0046: [2] - X:0, Y:0, P:0, V:0

<4>[ 9698.928767] 7260 irq 0

<6>[ 9698.931857] it7260 4-0046: processed points:

<6>[ 9698.936198] it7260 4-0046: [0] - X:673, Y:153, P:8, V:1

<6>[ 9698.941572] it7260 4-0046: [1] - X:848, Y:302, P:8, V:1

<6>[ 9698.946863] it7260 4-0046: [2] - X:0, Y:0, P:0, V:0

<4>[ 9698.951906] 7260 irq 0

<6>[ 9698.955049] it7260 4-0046: processed points:

<6>[ 9698.959410] it7260 4-0046: [0] - X:673, Y:153, P:8, V:1

<6>[ 9698.964803] it7260 4-0046: [1] - X:849, Y:302, P:8, V:1

<6>[ 9698.970131] it7260 4-0046: [2] - X:0, Y:0, P:0, V:0

<4>[ 9698.975166] 7260 irq 0

<6>[ 9698.978258] it7260 4-0046: processed points:

<6>[ 9698.982611] it7260 4-0046: [0] - X:673, Y:153, P:8, V:1

<6>[ 9698.987976] it7260 4-0046: [1] - X:848, Y:302, P:8, V:1

<6>[ 9698.993281] it7260 4-0046: [2] - X:0, Y:0, P:0, V:0

<4>[ 9698.998301] 7260 irq 0

<6>[ 9699.001379] it7260 4-0046: processed points:

<6>[ 9699.005752] it7260 4-0046: [0] - X:673, Y:152, P:8, V:1

<6>[ 9699.011142] it7260 4-0046: [1] - X:849, Y:302, P:8, V:1

<6>[ 9699.016475] it7260 4-0046: [2] - X:0, Y:0, P:0, V:0

<4>[ 9699.021525] 7260 irq 0

<6>[ 9699.024692] it7260 4-0046: processed points:

<6>[ 9699.029040] it7260 4-0046: [0] - X:673, Y:152, P:8, V:1

<6>[ 9699.034420] it7260 4-0046: [1] - X:848, Y:302, P:8, V:1

<6>[ 9699.039711] it7260 4-0046: [2] - X:0, Y:0, P:0, V:0

<4>[ 9699.044668] 7260 irq 0

<6>[ 9699.047829] it7260 4-0046: processed points:

<6>[ 9699.052182] it7260 4-0046: [0] - X:673, Y:152, P:8, V:1

<6>[ 9699.057549] it7260 4-0046: [1] - X:848, Y:302, P:8, V:1

<6>[ 9699.062874] it7260 4-0046: [2] - X:0, Y:0, P:0, V:0

<4>[ 9699.067843] 7260 irq 0

<6>[ 9699.071055] it7260 4-0046: processed points:

<6>[ 9699.075429] it7260 4-0046: [0] - X:673, Y:152, P:8, V:1

<6>[ 9699.080757] it7260 4-0046: [1] - X:849, Y:302, P:8, V:1

<6>[ 9699.086139] it7260 4-0046: [2] - X:0, Y:0, P:0, V:0

<4>[ 9699.091085] 7260 irq 0

It looks like a firmware bug, some internal queue overflow? But even so it is strange suspend/resume does not recover it, for the driver will reset it7260 during resume

I have made clear the reason: the driver try to recalibrate the touchscreen during init, but recalibrate will drive the touchscreen mad if pressed.

[Guest ejtagle]

I have made clear the reason: the driver try to recalibrate the touchscreen during init, but recalibrate will drive the touchscreen mad if pressed.

That is absolutely true... But the original 2.2 rom also did it... I have thought several times on removing the calibration performed at startup.... There are arguments for both keeping and removing it...

[Guest ejtagle]

Confirmed

And what about voice search ? ... With the latest audio_hw.c posted, it should also work... as all voice recording apps ... :o

[Guest Scanno]

And what about voice search ? ... With the latest audio_hw.c posted, it should also work... as all voice recording apps ... :o

I also have a full time job..:-) .. for me it is now 15:40 so still at work. Did not have time yet to build the new audio_hw.c

Edit:

Regarding extending the data partition... were you thinking abut aufs ?

Edited July 24, 2012 by Scanno

[Guest ejtagle]

I also have a full time job..:-) .. for me it is now 15:40 so still at work. Did not have time yet to build the new audio_hw.c

Edit:

Regarding extending the data partition... were you thinking abut aufs ?

Don't worry, i also have a full time job .. :) .. Yes, aufs was in mind... There are several other ways to do it... For example, we could have a readonly data partition (or just mount it as readonly), and overlay a data partition stored on the SD card... If present, otherwise, remount as rw... Really don't know, there are endless ways to do what we need ;)

[Guest foxsen]

That is absolutely true... But the original 2.2 rom also did it... I have thought several times on removing the calibration performed at startup.... There are arguments for both keeping and removing it...

Current I choose to remove it from the driver init, but wrote a small problem to detect some hardware key combination, so if the user thinks the touchscreen needs recalibration, he can use the hardware buttons to achieve it. I also use key combinations to rmmod/insmod if things goes really bad. Will test this on tens of machines. Later we can implement some gui to run at first boot, let the user choose to recalibrate, and warn them about no touch.

What do you think about this way?

[Guest Scanno]

Current I choose to remove it from the driver init, but wrote a small problem to detect some hardware key combination, so if the user thinks the touchscreen needs recalibration, he can use the hardware buttons to achieve it. I also use key combinations to rmmod/insmod if things goes really bad. Will test this on tens of machines. Later we can implement some gui to run at first boot, let the user choose to recalibrate, and warn them about no touch.

What do you think about this way?

Personally I do not think it is such a good idea. From a normal user perspective the current way is the most 'monkey proof'. People will press the powerbutton until the tablet goes off when there is no response. If you are assigning a hardware key to calibrate the screen, the chances that it goes wrong are greater. Besides asking a user if he/she wants to calibrate during boot time? Please no... I do not want to answer a question each time I boot/reboot my tablet.

Everything needs to be as monkey proof as possible....

[Guest ejtagle]

Current I choose to remove it from the driver init, but wrote a small problem to detect some hardware key combination, so if the user thinks the touchscreen needs recalibration, he can use the hardware buttons to achieve it. I also use key combinations to rmmod/insmod if things goes really bad. Will test this on tens of machines. Later we can implement some gui to run at first boot, let the user choose to recalibrate, and warn them about no touch.

What do you think about this way?

If i remember it, the 2.2 shuttle rom did a recalibration each time the back button was pressed... Don't know if it is such a good idea, remember each recalibration rewrites the nand flash touchscreen controller... The other thing that could be done is to enable automatic selfcalibration.. Never tried it , but supposedly, it self adapts...

[Guest brucelee666]

And what about voice search ? ... With the latest audio_hw.c posted, it should also work... as all voice recording apps ... :o

Eduardo,

Ok performed same test as last audio_hw so soundhound/skype/voice recorder and all 3 worked the same with new version.

Now onto voice search, it now no longer crashes and seems to pick up the word but does nothing else have attached a log for you to look at see mic3 it has displayed text after some time but nothing like the word I spoke (this could just be highlighting another issue with my rom as I have tried it again and it did recognize the word test but then errored out).

Also attached a mic3 which is using voice search in youtube app, this does not seem to pick up or search for anything, and attached a mic4 this is using voice for keyboard entry this picks up mic and is nearly accurate (on all three tests the one word I spoke was test keyboard entered text)

log3_mic666gv.txt

log4_mic666yt.txt

log5_mic666key.txt

[Guest ejtagle]

Eduardo,

Ok performed same test as last audio_hw so soundhound/skype/voice recorder and all 3 worked the same with new version.

Now onto voice search, it now no longer crashes and seems to pick up the word but does nothing else have attached a log for you to look at see mic3 it has displayed text after some time but nothing like the word I spoke (this could just be highlighting another issue with my rom as I have tried it again and it did recognize the word test but then errored out).

Also attached a mic3 which is using voice search in youtube app, this does not seem to pick up or search for anything, and attached a mic4 this is using voice for keyboard entry this picks up mic and is nearly accurate (on all three tests the one word I spoke was test keyboard entered text)

I see no errors in the log... Maybe is just a misconfiguration of the recognizer, for example, the language to recognize... Or the location ...

[Guest Scanno]

I see no errors in the log... Maybe is just a misconfiguration of the recognizer, for example, the language to recognize... Or the location ...

Compiled with the latest audio_hw.c and looks like it is working. One strange thing and that is i have to wait a few seconds before i can speak. Looks a bit like tto much gain / static. You can see that the level indicator (the circel around the mic symbol) is almost at maximum and slowly goes down. When it is down it works pretty good.

Edited July 24, 2012 by Scanno

[Guest ejtagle]

Compiled with the latest audio_hw.c and looks like it is working. One strange thing and that is i have to wait a few seconds before i can speak. Looks a bit like tto much gain / static. You can see that the level indicator (the circel around the mic symbol) is almost at maximum and slowly goes down. When it is down it works pretty good.

Maybe, when the recording starts, there is a click that is fooling the voice recognition... I will investigate that :)

[Guest foxsen]

Personally I do not think it is such a good idea. From a normal user perspective the current way is the most 'monkey proof'. People will press the powerbutton until the tablet goes off when there is no response. If you are assigning a hardware key to calibrate the screen, the chances that it goes wrong are greater. Besides asking a user if he/she wants to calibrate during boot time? Please no... I do not want to answer a question each time I boot/reboot my tablet.

Everything needs to be as monkey proof as possible....

I agree that answer question each boot is bad idea. I mean for the first time boot only. Later on if it is needed user can use hardwaare key to recal.

Key combination can be some kind of monkey proof: for example, press both volume+/- for 5 seconds, or press volume+/- plus a swith of software rotation lock. In fact I am using these for reload/recal.

[Guest foxsen]

If i remember it, the 2.2 shuttle rom did a recalibration each time the back button was pressed... Don't know if it is such a good idea, remember each recalibration rewrites the nand flash touchscreen controller... The other thing that could be done is to enable automatic selfcalibration.. Never tried it , but supposedly, it self adapts...

Yes, at least some are. But for first time boot only. For the rewrite consideration, current driver recalibrate for each boot already. I think the recalibrate result should be usable for a long time since it will be save somewhere in the controller. CMD 0x13(recalibrate) is meant for mass production usage. If the environment does not change a lot, it should has no need to recalibrate.

[Guest ejtagle]

Yes, at least some are. But for first time boot only. For the rewrite consideration, current driver recalibrate for each boot already. I think the recalibrate result should be usable for a long time since it will be save somewhere in the controller. CMD 0x13(recalibrate) is meant for mass production usage. If the environment does not change a lot, it should has no need to recalibrate.

Well, i don't think so.. The purpose of the recalibration app is to send that command. If the calibration is valid depending only on environment, it should not be required... :S

[Guest ejtagle]

Talking about Voice Search, the problem is that when you tap the screen to start it, the mic captures the tap and outputs a signal that Voice Search confuses as the start of the voice :wacko: . Even worse, as the tap is a high powered impulse, the AGC (Automatic ganancy control) reduces gain to the point that the normal voice is not recognized as such. Essentially, there is no easy fix for this problem.

What i did was to add a muting delay from the start of the audio capture process to the moment actual non muted data is dispatched to Voice Search. That delay is used to hopefully discard the screen tap noise. The delay is 1/4 of a second, should be enough and should not interfere with any recording activity. There is not too much else we could try.

Attached the modified audio_hw,c. Hope it makes Voice search work a little better... :)

audio-24jul-11pm.rar

[Guest ejtagle]

And an attempt to fix the audio codec... Sorry, i am nearly falling asleep... I think i've found the bug, but had no time to test it... Please, post dmesg and last_dmesg if it does not work. I promise tomorrow i will test it :)

rel-15r7-v4.rar

[Guest Scanno]

And an attempt to fix the audio codec... Sorry, i am nearly falling asleep... I think i've found the bug, but had no time to test it... Please, post dmesg and last_dmesg if it does not work. I promise tomorrow i will test it :)

Sorry Eduardo, does not work.. The audio driver does not get recognized now...

Just copied over the new files and rebuild the kernel.. Have to go to work now..

dmesg0725.txt

Edited July 25, 2012 by Scanno

[Guest foxsen]

Well, i don't think so.. The purpose of the recalibration app is to send that command. If the calibration is valid depending only on environment, it should not be required... :S

I have contacted with ITE support, he confirms my guess. He tells me that 7260 has two use mode:

1, like shuttle, depend on power on firmware recalibrate and 0x13

2, continously recalibrate during use, firmware set a interval and do recalibrate frequently

0x13 command will disable firmware's power-on-calibrate(no way to reenable). So once we use 0x13, we need to use it to recalibrate later.

0x13 calculate the capacity in current environment as a base for calculating touches(so we can understand that why press-on-recalibrate will make the controller repeatly report touches, it should be too low a threshold), in theory we don't need to recalibrate each boot because the environment(hot/cold, wetness etc.) normally won't change a lot. Ofcourse we can provide a tool for user to recalibrate explicitly in case things go wrong.

Regards

Edited July 25, 2012 by foxsen

[Guest ejtagle]

I have contacted with ITE support, he confirms my guess. He tells me that 7260 has two use mode:

1, like shuttle, depend on power on firmware recalibrate and 0x13

2, continously recalibrate during use, firmware set a interval and do recalibrate frequently

0x13 command will disable firmware's power-on-calibrate(no way to reenable). So once we use 0x13, we need to use it to recalibrate later.

0x13 calculate the capacity in current environment as a base for calculating touches(so we can understand that why press-on-recalibrate will make the controller repeatly report touches, it should be too low a threshold), in theory we don't need to recalibrate each boot because the environment(hot/cold, wetness etc.) normally won't change a lot. Ofcourse we can provide a tool for user to recalibrate explicitly in case things go wrong.

Regards

There is a selfadapting bit in one of the commands i saw.. We could try it... ;) .. Look at it7260_recalibrate_cap_sensor, there i have commented the command that sets to autotuning the touchscreen controller...

Edited July 25, 2012 by ejtagle

[Guest ejtagle]

Sorry Eduardo, does not work.. The audio driver does not get recognized now...

Just copied over the new files and rebuild the kernel.. Have to go to work now..

Those are pretty good news!! (believe it or not :) ... By looking at your dmesg, now i know what is missing...

<7>[ 7.634816] tegra-snd-alc5624 tegra-snd-alc5624: Shuttle sound card probe

<7>[ 7.642710] tegra-snd-alc5624 tegra-snd-alc5624: Shuttle MCLK initialized

<6>[ 7.649918] alc5624 0-0018: ALC5624 probe called

<3>[ 7.668508] alc5624 0-0018: unknown or wrong codec - Expected 10ec, got ec10 [3]

<3>[ 7.676027] alc5624 0-0018: asoc: failed to probe CODEC alc5624.0-0018: -19

<3>[ 7.683280] asoc: failed to instantiate card tegra-alc5624: -19

<3>[ 7.689378] tegra-snd-alc5624 tegra-snd-alc5624: snd_soc_register_card failed (-19)

<6>[ 7.697451] ALSA device list:

<6>[ 7.701648] No soundcards found.

Some of the changes i did to the probing routine are causing word reads to be byteswapped... But the codec is answering as it should... So, attached a corrected version that should work :) (lety's hope!)

rel-15r7-25jul-11am.rar

[Guest ejtagle]

Random note: Why not using squashfs with lzma support ? - Our kernel already supports that sqaushfs format, and compression ratio improves considerably... : http://www.squashfs-lzma.org/ Loaak at compression benchmarks... Quite a big difference :) .. The only thing needed is to enable it in the kernel configuration, and then copress using that format...

[Guest Scanno]

Random note: Why not using squashfs with lzma support ? - Our kernel already supports that sqaushfs format, and compression ratio improves considerably... : http://www.squashfs-lzma.org/ Loaak at compression benchmarks... Quite a big difference :) .. The only thing needed is to enable it in the kernel configuration, and then copress using that format...

Will have a look at that.

[Guest Cass67]

Random note: Why not using squashfs with lzma support ? - Our kernel already supports that sqaushfs format, and compression ratio improves considerably... : http://www.squashfs-lzma.org/ Loaak at compression benchmarks... Quite a big difference :) .. The only thing needed is to enable it in the kernel configuration, and then copress using that format...

When i was squashing stuff for the builds i was putting out i found that the benefits reported by using lzma was not enough to bother with the hassle of having to compile it into the kernel and having to compile the squashfs binary with this support. I made that choice to allow those who wanted to make a change to the squashfs container to do so with the minimum of fuss as i was unsure of the support of lzma in other os types other than Linux. and to be honest in real world i didnt notice too much space reclaim if any or performance gain over default gzip compression. The first couple of alphas i put out used xz im sure...

Edited July 25, 2012 by Cass67

[Guest brucelee666]

I posted this a few days ago about how I had used xz compression for squashfs in jb and to give a comparison, this is with all gapps and some removals of non-essential apps/livewallpapers:-

app start = 89.1mb - default sfs = 63.9mb (66977792 bytes) - xz sfs = 61.3mb (64245760 bytes)

fonts start = 6.2mb - default sfs = 3.2mb (3338240 bytes) - xz sfs = 2.7mb (2789376 bytes)

framework start = 17.8mb - default sfs = 13.5mb (14143488 bytes) - xz sfs = 13.1mb (13758464 bytes)

lib start = 77.3mb - default sfs 36.9mb = (38649856 bytes) - xz sfs = 30mb (31506432 bytes)

usr start = 8.5mb - default sfs = 2.7mb (285491 bytes) - xz sfs = 2.0mb (2056192 bytes)

120.1mb default sfs total - 107.3mb xz sfs total = 12.8mb saving roughly

edit - just to add the xz method had no filter, you can add an arm filter which squeezes a few extra bytes of compression but I have not tried to boot this sfs

Edited July 25, 2012 by brucelee666