[DEV][UPDATE: 2-10-11][Optimus One multitouch fixes]

[kpbotbot]

This thread has gone on a different approach on resolving this, so I'll be setting aside my old post (I wont be deleting them for some reasons). Everything under the hash army (###) in this single post is about getting the Spica stuff on the O1. We moved to focusing on the O1's ts driver itself.

Other note: Fixes are independent. Choose whichever works for you. A universal fix may come in the future though (Tell me if this post still confuses you. I'll edit ASAP)

Multitouch axis inversion fix (First release: Feb 6, 2011): Forum member waleedq came up with a workaround that inverts the inversion. This seems very straightforward, and it works. Please note, however, that this is a workaround and not a fix. As always, do a backup.

Stuff you should know:
- Most of us who tried it (inc. me), lost our OC.
- mik_os' touchscreen lag fix is here. Tested it with Android Status while touching the screen.
- Axis inversion is still a problem. Check video 2.

Videos: http://forum.xda-developers.com/showpost.php?p=11178060&postcount=67
Link to the download link: http://forum.xda-developers.com/showpost.php?p=11166426&postcount=47
waleedq's source: https://github.com/waleedq/ThunderG-Kernel/blob/thunderg_oc/drivers/input/touchscreen/mcs6000_ts.c

Faster readout fix (First release: Feb 8, 2011) : Der_Caveman released for testing a kernel with some delay fix (modified from stock kernel). Apparent goals were to make touchscreen input readouts faster. Also, useless and non-working functions that doesn't work were removed. No CPU spike (somewhere around 30%cpu use when touching screen)

UPD: v3 is based on drellisdee's kernel source, so yes, it's OCable.

Link to download link: http://forum.xda-developers.com/showpost.php?p=11195243&postcount=84
Link to download link(v3): http://forum.xda-developers.com/showpost.php?p=11221504&postcount=126 or a forum upload http://forum.xda-developers.com/showpost.php?p=11230915&postcount=137
Changes and stuff: http://forum.xda-developers.com/showpost.php?p=11206827&postcount=91

Touchscreen optimization + axis inversion workaround(First release: Feb 10, 2011):Der_Caveman and waleedq's work combined by Sippolo.

Link to download link + source + details: http://forum.xda-developers.com/showpost.php?p=11256797&postcount=198
Link to download link + source + details (v1.1): http://forum.xda-developers.com/showpost.php?p=11270721&postcount=237

#########################################################
Okay, so I just recently found out that the Galaxy Spica (was supposed to choose it over the O1 --') -unofficially- supports multitouch.

A link to the driver's was PMed to me by adi4motion:
https://github.com/Gabriel-LG/i5700-multitouch-kernel

Anyway, is it possible that we could implement their method on our drivers? I understand that there might be some hardware differences between the two. I'm just trying my luck

Lastly, for sample videos, check YouTube. I have to say that the performance with those drivers really are impressive (well, I saw the axis invert once)

 

[Der_Caveman]

https://github.com/Gabriel-LG/i5700-multitouch-kernel/tree/master/drivers/input/touchscreen

I didn't find any melfas drivers in there- so I doubt that the Galaxy Spica has a melfas touchscreen controller in it.

 

[kpbotbot]

But it's possible that we could write a driver based on those, right?

 

[ccdreadcc]

only kpbotbot can do tht and i hope he does

 

[N3mesis]

What about O1 brother the LG Chic? Does it have the same issues?

Sent from my LG-P500

 

[kpbotbot]

only kpbotbot can do tht and i hope he does

Sorry, can't code. But I assure you, I'm trying my best not to just leech (And I'm still figuring out how to compile FroYo to learn how ROMs get born)

What about O1 brother the LG Chic? Does it have the same issues?

There may be a guy here in the forum that has the Chic. Maybe he can test out the inversion issues (If you're reading this and you have the Chic, please test if the axis inverts using a tool like touch test xD)

 

[Der_Caveman]

But it's possible that we could write a driver based on those, right?

In the O1 a melfas MCS6000 touchscreen controller reads the capacitive touchscreen. The controller itself got a proprietary firmware in it (don't know if its from melfas or LG)- you can find that somewhere in the system folder (melfas.bin(?)). Then there is the LG kernel driver which "interacts" with the melfas controller ic- it also has the ability to flash the melfas controller.
So if there is a problem in the MCS6000 firmware- we're most likely screwed, because lack of datasheets.

Furthermore I want to know how android gets the multitouch data. Maybe the android interface itself is not that multitouch friendly?

The easiest way would have been that each finger represents a "mouse event", but I doubt that this way was chosen...

 

[kpbotbot]

I see. Is it possible to decompile that .bin somehow? I know, that question might sound stupid xD

I too did a search for melfas related stuff including source codes and the like. The only stuff I found was about Samsung phones having the same touch screen.

Let's just hope (for now that we cant do anything about it) that LG -at least- releases an improved version of their drivers for their 2.3 update

 

[kpbotbot]

Update: The Chic has the same problem :/
http://forum.xda-developers.com/showpost.php?p=11014035&postcount=5

 

[Der_Caveman]

Maybe the axis inversion is somehow also android related.
Multitouch is something new for linux.

 

[kite86]

Maybe the axis inversion is somehow also android related.
Multitouch is something new for linux.

I think is just the crappy touch controller that's installed on our device; It can't spot more than 2 touches, and has got axis inversion. So I don't think there's so much to do with drivers

 

[Der_Caveman]

I think is just the crappy touch controller that's installed on our device; It can't spot more than 2 touches, and has got axis inversion. So I don't think there's so much to do with drivers

LG only implemented "two fingers"- one can read that in LG's sources.
Again: axis inversion could also be android related (if android only requests coordinates without "finger assignments" bugs are preassigned).

 

[kpbotbot]

I don't really care whether it would accept more than 2 presses That would be awesome for people doing extreme gaming with the phone, but with normal use it's awkward to use 3 fingers.

I bet the axis inversion issue is very much fixable. We just don't know how yet. But considering the fact that the Spica has less of this as a problem, we might fix it a bit, or completely fix it.

This is just my opinion though. I saw some guy explain multitouch with an illustration.

 

[Der_Caveman]

Wow, I have found this:

http://www.mjmwired.net/kernel/Documentation/input/multi-touch-protocol.txt

Just had to open my eyes

Let's see...

 

[Der_Caveman]

I bet the axis inversion issue is very much fixable. We just don't know how yet. But considering the fact that the Spica has less of this as a problem, we might fix it a bit, or completely fix it.

You are 100% sure that the Spica does not have "the axis inversion bug"?

 

[Kel_Sceptic]

You are 100% sure that the Spica does not have "the axis inversion bug"?

I'm 100% sure.

 

[kpbotbot]

I talked to a guy who used to own the phone. He said the drivers perform well.

I didn't say that there really was no inversion. It happens less (inverted once on the video I watched)

 

[Der_Caveman]

I talked to a guy who used to own the phone. He said the drivers perform well.

I didn't say that there really was no inversion. It happens less (inverted once on the video I watched)

Ok, then I would say there is hope.

 

[Kel_Sceptic]

Ok, then I would say there is hope.

I sure do prey that someone will take this project and fix the driver or write it from scrath. As i can see the code is not that long

our driver
https://github.com/mik9/ThunderG-Kernel/blob/master/drivers/input/touchscreen/mcs6000_ts.c

spica's driver
https://github.com/Gabriel-LG/i5700...4461c4a3d4e2be0b5f4b1a94b451604c6541f1#diff-0

 

[kpbotbot]

I have 2 months of doing absolutely nothing at home (stopped this sem. doctor said i have to rest. and yes, I have a life xD), so I might be able to understand a bit of the code. It really isn't that complex, but my lack of programming knowledge hinders me from modifying the code.

However, I sure do hope that someone figures this out