00:02.30 | *** part/#htc-linux s_j_newbury (n=s_j_newb@cpc1-derb5-0-0-cust343.lei3.cable.ntl.com) |
00:03.15 | *** join/#htc-linux cr2 (n=konversa@crpl22.physik.uni-wuppertal.de) |
00:06.24 | Kevin2 | Hi |
00:12.21 | *** join/#htc-linux psokolovsky_ (n=psokolov@82.193.98.2) |
02:14.59 | *** join/#htc-linux elysion (n=elysion@k203c.lanhovi.tontut.fi) |
05:06.12 | *** join/#htc-linux cyrill62 (n=cyrill62@ble59-5-82-233-205-36.fbx.proxad.net) |
05:20.33 | *** join/#htc-linux tudenbart (n=willi@xdsl-81-173-170-65.netcologne.de) |
05:43.09 | *** join/#htc-linux cyrill62 (n=cyrill62@ble59-5-82-233-205-36.fbx.proxad.net) |
06:12.19 | *** join/#htc-linux clustered (n=sddd@218.111.170.212) |
06:13.29 | clustered | hello everyone... |
06:18.05 | clusco | anyone here? |
06:46.37 | Ralith | no. |
06:47.01 | *** join/#htc-linux rmoravcik (n=rmoravci@icm1-orange.orange.sk) |
06:47.04 | clusco | hehe |
06:47.04 | clusco | allo Ralith |
06:47.04 | clusco | allo rmoravcik |
06:47.04 | clusco | emm... |
06:47.05 | rmoravcik | clusco: hi |
06:47.06 | clusco | rmoravcik |
06:47.06 | clusco | i just sucessfully install linux on my htc alpine device.... |
06:47.06 | clusco | n now running opie.... |
06:47.07 | clusco | how to exit from opie n back to wince ? |
06:47.07 | clusco | im still living with thing problem |
06:47.08 | rmoravcik | try run command reboot or use reset button |
06:47.08 | clusco | this problem |
06:47.09 | clusco | run on console? |
06:47.09 | rmoravcik | yes |
06:47.10 | clusco | wait |
06:47.11 | clusco | im booting linux now |
06:47.11 | clusco | htc alpine... the one device without extended rom unlocker.... even there's no WM5 too |
06:47.12 | clusco | too shame... |
06:47.14 | clusco | emmm |
06:47.15 | clusco | nothing happen when i run reboot command |
06:47.15 | rmoravcik | try ask renatostore, he is working on htc alpine |
06:47.16 | clusco | i had ask on xda forum.... no one answering yet |
06:47.19 | clusco | hard reset is the answer |
06:47.19 | clusco | waaaaaa |
06:47.22 | clusco | thanks very much |
06:47.22 | clusco | see yah next time |
06:47.22 | clusco | gtg |
08:13.27 | *** join/#htc-linux Fossi (n=cmewes@195.177.48.115) |
09:12.43 | *** join/#htc-linux pH5 (n=ph5@p5485F398.dip.t-dialin.net) |
09:32.01 | *** join/#htc-linux psokolovsky_ (n=psokolov@82.193.98.2) |
09:37.16 | *** join/#htc-linux psokolovsky__ (n=psokolov@82.193.98.2) |
10:38.21 | *** join/#htc-linux goxboxlive (n=goxboxli@195.159.97.196) |
10:38.28 | goxboxlive | aloha |
10:57.51 | psokolovsky__ | Hi! |
10:58.19 | psokolovsky__ | goxboxlive, common touchscreen driver has arrived. are you ready to test in on universal? ;-) |
11:12.16 | *** join/#htc-linux cyrill62 (n=cyrill62@ble59-5-82-233-205-36.fbx.proxad.net) |
11:13.01 | goxboxlive | psokolovsky__ Of course i will test it. Is is available for the Universal too? Or do cr2 have to rewrite something? |
11:14.07 | psokolovsky__ | goxboxlive, will have to write shim code. it's just early notice, pH5 will test it on magician first, etc. |
11:15.04 | goxboxlive | i c. |
12:19.44 | pH5 | psokolovsky__: so the driver works on the magician (it's even linear :)), but there is something severely wrong with the pressure sensitivity/pen down recognition. |
12:19.59 | pH5 | in gpe-sketchpad I only draw dotted lines |
12:20.35 | pH5 | and when I just put down the pen at one point, it behaves as if I'd hammer this point with the pen like mad. |
12:21.23 | psokolovsky__ | Hi! |
12:21.41 | psokolovsky__ | pH5, that's close to what happened on h4000 vs hx4700 ;-) |
12:21.50 | psokolovsky__ | pH5, do you supply .gpio? |
12:23.06 | psokolovsky__ | it's optional, and first means to detect if pen was lifted up |
12:23.21 | psokolovsky__ | otherwise, there's adhoc pressure threshold, hardcoded |
12:23.30 | psokolovsky__ | it's of course must be made adjustable |
12:23.34 | psokolovsky__ | via platform_data |
12:24.41 | pH5 | psokolovsky__: I do supply the PEN_IRQ gpio. it is active-low. |
12:25.35 | psokolovsky__ | pH5, irq is a must. .gpio corresponding to it is optional, see above. but it is needed on hx4700 for proper detection (h4000 can live w/o it) |
12:26.01 | psokolovsky__ | pH5, let me commit hx4700 stuff ;-) |
12:28.46 | psokolovsky__ | pH5, committed. So, try passing .gpio as the first step |
12:29.12 | pH5 | psokolovsky__: "I do supply the PEN_IRQ gpio. it is active-low." |
12:29.44 | pH5 | btw, why don't we expose the pressure data to userspace? |
12:29.56 | psokolovsky__ | pH5, ah, ok. Then, go figure what's wrong with it ;-). Maybe it bounces and you shouldn't supply it ;-) |
12:30.05 | psokolovsky__ | pH5, that's on todo ;-) |
12:30.41 | psokolovsky__ | pH5, we should switch to Kevin2's inlude/linux/touchscreen-adc.h as platform data for ads7846 too |
12:31.03 | psokolovsky__ | so, a bit of refactoring yet to do ;-) |
12:32.01 | psokolovsky__ | pH5, I'm glad it at least works a bit OOB. I fixed lots of races and hangs on hx4700 which didn't manifest on h4000 ;-) |
12:39.05 | pH5 | psokolovsky__: what does disable_irq_nosync do compared to disable_irq? |
12:40.00 | psokolovsky__ | pH5, "is safe to call from isr" vs "safe with care" ;-). that change I made when debugging weird dis/en races. |
12:40.29 | psokolovsky__ | pH5, I committed some debug logging for ads7846ts, maybe it'll help a bit |
12:41.38 | pH5 | psokolovsky__: I am trying to understand at what place it thinks I lifted the pen. maybe the gpio really bounces when reading via SSP, but I know for sure the old driver used the gpio to detect the lifted pen. |
12:41.48 | psokolovsky__ | AFAIU, dis_irq() kinda tries to ensure that isr is completed before disabling irq, which is kinda nonsense in our case |
12:42.15 | pH5 | I see. |
12:42.55 | psokolovsky__ | pH5, I added initial logging for just pen down/up detected, feel free to add more to catch exact cause |
12:44.10 | pH5 | psokolovsky__: ok. did you have a look at the probe error handling I added? |
12:45.33 | *** join/#htc-linux goxboxlive (n=goxboxli@9.80-202-160.nextgentel.com) |
12:47.07 | psokolovsky__ | pH5, looks well, sure |
12:48.16 | pH5 | especially the test on request_irq might be good for people compiling both hx4700_ts and the ads driver in the next few weeks :) |
12:50.00 | psokolovsky__ | pH5, yes, you're right, sure. I may just say that I applied attention to err checking/handling in ads7846_ssp - ssp is really capricious thing, something a bit away and it hanged ;-) |
12:50.42 | pH5 | psokolovsky__: right, there certainly is a reason why it took so long until somebody started to do something about it ;) |
12:51.17 | psokolovsky__ | yep ;-) |
13:11.10 | *** part/#htc-linux lama (n=lama@estel.wpia.uw.edu.pl) |
13:19.11 | pH5 | psokolovsky__: I get many many alternating started/finished to sample messages and one spurious ts irq when I lift the pen.. |
13:21.34 | psokolovsky__ | pH5, so gpio apparently bounces pretty heavy ;-I I just wonder, how to handle that. well need to look at magician_ts close how it deals with that... |
13:21.53 | cr2 | psokolovsky__: this driver assumes that the pen_ is on pxa gpio, and does not set the power bits for tsc2046 |
13:22.43 | psokolovsky__ | cr2, yep, that's all on todo. feel free to patch it to. |
13:22.43 | cr2 | and the power bits must be set, otherwise the ts will be nonlinear on some htc devices. |
13:22.55 | psokolovsky__ | cr2, as for power bits, there handling should be mostly ok. |
13:23.00 | cr2 | yes, i know that hx4700 does not need it. |
13:23.13 | psokolovsky__ | cr2, do you mean internal ref voltage after all? |
13:23.30 | cr2 | hmm, i'll need this driver on s3c24xx too. |
13:23.43 | psokolovsky__ | cr2, I saw htcuiversal_ts2 paltform_data magic, we for sure need to merge it in |
13:24.04 | cr2 | i forgot how the name ;) |
13:24.27 | cr2 | yes, htcuniversal_ts2 accounts for these things. |
13:24.29 | psokolovsky__ | cr2, perfect, that's all achievable - just write alternative ADC driver. don't start with that now though, let's first convert to generic ADC API |
13:24.48 | cr2 | ok. |
13:27.10 | pH5 | psokolovsky__: the old hx4700 driver checked the gpio after sampling, in the timer callback. |
13:27.56 | psokolovsky__ | pH5, well, on hx4700, the current CVS version works pretty well ;-) |
13:28.09 | psokolovsky__ | pH5, does magician do the same? |
13:28.57 | pH5 | I just copied the hx4700 driver for the magician and worked around the hx4700 driver's strange alternating readout scheme a bit. |
13:31.31 | psokolovsky__ | pH5, ok, so can you try to swap pen_up vs read_xy order and see if it helps? |
13:32.33 | psokolovsky__ | pH5, and are you 100% sure it's exactly gpio? did you see bouncing in debyg dumps? did you try to dump/comment out threshold checks? |
13:34.56 | pH5 | ah, I thought it ignores the pressure threshold when I supply a gpio? |
13:35.01 | pH5 | I'll check.. |
13:35.06 | Kevin2 | Hi |
13:35.11 | pH5 | hi Kevin2 |
13:35.52 | Kevin2 | Been quite around here the last couple of weeks. I've been busy with work myself. |
13:36.42 | pH5 | "Trying to free already-free IRQ 147" |
13:36.47 | psokolovsky | pH5, no, gpio has priority though. i.e. if gpio tells pen is up, sampling ends. otherwise, it still checks bounding off data and pressure threshold - if any fails, it considers that as singnal to stop sampling too |
13:37.01 | pH5 | heck, why does it do that when I rmmod ads7846ts-ssp.. |
13:37.26 | psokolovsky | bounding check was from original h1910 driver, I'm not sure if it is really needed, but I added it while trying to make both h4000 & hx4700 work |
13:37.53 | pH5 | psokolovsky: aha, so i should look out for sampling errors instead of gpio edges. |
13:38.22 | psokolovsky | Hi Kevin2! Hope soon you ADC API stuff will shine ;-) there're few issues to consider though... (not now ;-) ) |
13:38.26 | pH5 | maybe. I need to get that irq problem right though. rebooting between rmmod/insmod because of segfaults in the irq subsystem is not much fun. |
13:38.40 | psokolovsky | I guess |
13:40.03 | Kevin2 | I'm a bit confused by all the complexity in the debounce stuff for ts. I wonder why a simpler algorithm wont work. |
13:42.33 | psokolovsky | pH5, committed small fix for free_irq |
13:42.37 | psokolovsky | Kevin2, the algorithm is simple actually ;-) different devices require different params though ;-) |
13:47.11 | Kevin2 | I'm off to work.. |
13:57.16 | pH5 | psokolovsky: much better :) the problem is that I get err=1 for many y measurements. |
13:57.34 | pH5 | rc time constant too big. |
13:58.15 | psokolovsky | pH5, oops. I really should have pointed to that earlier, but I really read too much about gpio flipping ;-) |
13:58.25 | psokolovsky | pH5, d = 8 ? |
13:58.56 | pH5 | shouldn't sensing errors be handled differently from the "pressure too low" case? |
13:59.07 | pH5 | I certainly have to increase c or d. |
13:59.41 | pH5 | is there a certain reason why d=8 makes sense or was this chosen arbitrarily? |
13:59.47 | psokolovsky | pH5, apparently, they must. |
14:00.03 | pH5 | I know that my wince driver measures 12 times for x and 14 times for y direction. |
14:00.15 | psokolovsky | pH5, that was in original h1910's driver ;-). of course, that should be adjustable via platform_data |
14:00.36 | psokolovsky | 8 just gives perfect linearity/debouncing on both h4000 & hx4700 |
14:01.16 | psokolovsky | pH5, let's have all those params adjustable. of course, first just find what values work for you, we can add paltform_data later |
14:01.57 | pH5 | psokolovsky: I should put some printks in the putget loop and make R(t) graphs to find the best cutoff point. |
14:02.43 | psokolovsky | pH5, how complex ;-). bump value, load drawpad-kind app, and just draw straight lines ;-) |
14:03.11 | pH5 | psokolovsky: nah, too simple ;) |
14:07.45 | pH5 | Still, if a single x or y measurement fails, it should be ignored, but not necessarily stop the sampling. |
14:09.44 | psokolovsky | pH5, probably. just such errors usually happen when pen is lifted, and extra detection that sampling should be over... |
14:10.35 | pH5 | with d=8,c=14 all x errors are vanished, but I still get the occasional y error |
14:15.15 | psokolovsky | but as I told, I re-added them semi-blind trying to make it work, as we collect more data, we should elaborate it |
14:15.31 | psokolovsky | pH5, I suggest bumping d a bit |
14:29.43 | pH5 | psokolovsky: if I increase d, most of the measurements end early and it becomes nonlinear |
14:34.02 | pH5 | example y measurement in the middle of the screen: 2201 2147 2147 2152 2153 2151 2394 2167 2162 2160 2159 2159 2384 2157 |
14:34.04 | pH5 | huh? |
14:34.39 | pH5 | this is forced to measure 14 times. and I can regularly see the 7th and 13th value to jump out. |
14:49.06 | *** join/#htc-linux goxboxlive (n=goxboxli@9.80-202-160.nextgentel.com) |
14:49.06 | *** join/#htc-linux cyrill62 (n=cyrill62@ble59-5-82-233-205-36.fbx.proxad.net) [NETSPLIT VICTIM] |
14:49.06 | *** join/#htc-linux cr2 (n=konversa@crpl22.physik.uni-wuppertal.de) [NETSPLIT VICTIM] |
14:49.06 | *** join/#htc-linux Ralith (n=ralith@soggy202.drizzle.com) [NETSPLIT VICTIM] |
14:49.06 | *** join/#htc-linux Tjikkun (n=tjikkun@82-204-54-115.dsl.bbeyond.nl) [NETSPLIT VICTIM] |
14:49.06 | *** join/#htc-linux gw280 (n=george@vm.gwright.org.uk) [NETSPLIT VICTIM] |
14:49.07 | *** join/#htc-linux toi (n=pleemans@d5152D3B4.access.telenet.be) |
14:49.07 | *** join/#htc-linux pof (n=pof@62.57.1.173) [NETSPLIT VICTIM] |
14:49.07 | *** join/#htc-linux anYc (i=mario@hadince17.hadiko.uni-karlsruhe.de) [NETSPLIT VICTIM] |
14:49.07 | *** join/#htc-linux _scatterp (i=scatterp@packet.filtered.ca) [NETSPLIT VICTIM] |
14:49.07 | *** join/#htc-linux awelux (n=awelux@lvps87-230-8-217.dedicated.hosteurope.de) [NETSPLIT VICTIM] |
14:51.35 | *** join/#htc-linux psokolovsky (n=psokolov@82.193.98.2) [NETSPLIT VICTIM] |
14:51.36 | *** join/#htc-linux pH5 (n=ph5@p5485F398.dip.t-dialin.net) |
14:51.36 | *** join/#htc-linux hlbot (n=adm@iclem.net) [NETSPLIT VICTIM] |
14:51.36 | *** join/#htc-linux lilliput (n=Touriste@82-47-146-8.cable.ubr11.brad.blueyonder.co.uk) [NETSPLIT VICTIM] |
14:51.36 | *** join/#htc-linux TeringTuby (n=maarten@189-66-dsl.ipact.nl) |
14:51.37 | *** join/#htc-linux drno (n=drno@eki.student.utwente.nl) [NETSPLIT VICTIM] |
14:51.37 | *** join/#htc-linux Kmarc (i=kari@markos.biz) [NETSPLIT VICTIM] |
14:51.40 | psokolovsky | pH5, then use c=6? ;-) |
14:51.44 | psokolovsky | or maybe that simple converge algorithm indeed wouldn't work for everything and averaging is required |
14:51.44 | psokolovsky | but first thing to try is of course just discard too big delta values... |
14:51.45 | pH5 | I wonder what those spikes are. They are less pronounced at the upper end and at different positions at the lower end of the screen. |
14:51.46 | psokolovsky | (though proper impl of that would require averaging in the first place ;-) ) |
14:51.48 | psokolovsky | pH5, LCD synchro? zauruses full of it |
14:51.48 | psokolovsky | with d=200 I was able to see them on hx4700 too |
14:51.48 | pH5 | psokolovsky: ah right |
14:51.51 | *** join/#htc-linux RoEn_PC (n=roen@p54A65AD3.dip.t-dialin.net) |
16:05.17 | *** join/#htc-linux rmoravcik (n=rmoravci@icm1-orange.orange.sk) |
16:08.08 | *** join/#htc-linux rmoravcik (n=rmoravci@icm1-orange.orange.sk) |
16:15.41 | *** join/#htc-linux WizMaui (n=WizMaui@62.112.90.231) |
16:17.37 | *** join/#htc-linux Ralith (n=ralith@66.150.12.86) |
16:37.39 | *** join/#htc-linux Fossi (n=cmewes@195.177.48.115) |
16:47.17 | *** join/#htc-linux rmoravci1 (n=rmoravci@icm1-orange.orange.sk) |
16:49.59 | *** join/#htc-linux psokolovsky_ (n=psokolov@82.193.98.2) |
17:27.32 | *** join/#htc-linux pleemans (n=peter@d51A5E76A.access.telenet.be) |
18:17.46 | *** join/#htc-linux dothebart (n=willi@xdsl-213-196-243-15.netcologne.de) |
20:44.30 | *** join/#htc-linux goxboxlive (n=goxboxli@9.80-202-160.nextgentel.com) |
20:47.53 | *** join/#htc-linux Hawk|- (n=Hawk@p57A51E04.dip0.t-ipconnect.de) |
22:40.10 | *** join/#htc-linux cyrill62 (n=cyrill62@arf62-1-82-237-248-201.fbx.proxad.net) |