00:05.15 | Kevin2 | cr2: The jornada haret cvs doesn't seem to do anything special with DMA in its linboot code. |
00:06.06 | cr2 | yes, but it's not right ;) i did not boot with this code. |
00:08.48 | cr2 | the kernel was hacked to be relocated above OHCI (at RAM+2MB) and then disabled OHCI and DMA in jornada820.c |
00:08.51 | cr2 | ugly ;) |
00:08.56 | Kevin2 | Oh. Well, adding support for mainline should be pretty easy. Just add a new machine class that sets OEMInfo[0] and PlatformType to the values you cite above in its constructor. Then add in the necessary preHardwareShutdown and hardwareShutdown methods. |
00:09.15 | cr2 | true. |
00:10.07 | Kevin2 | As for the relocate stuff - can't you just add 2MBs to the memphysaddr variable? |
00:10.45 | cr2 | Kevin2: if haret will disable the OHCI, there will be no reason to do nonstandard things. |
00:12.36 | lkcl | cr2: do you think there could be more ASIC3 GPIO pins involved, with sound? or more GPIO pins? |
00:12.39 | lkcl | i mean, there aren't that many ASIC3 GPIO outputs _left_ :) |
00:13.35 | cr2 | lkcl: maybe pxa GPIO |
00:13.37 | Kevin2 | Okay. If you get around to it, email me a patch with your support. You should be able to checkout mainline and build it with egcc. Building egcc should be pretty straight forward also (see the INSTALL file from haret). Until we get cegcc fixed, you can change haret.cpp with the following patch: |
00:13.39 | Kevin2 | +extern "C" void __gccmain(); |
00:13.39 | Kevin2 | + |
00:13.39 | Kevin2 | <PROTECTED> |
00:13.39 | Kevin2 | <PROTECTED> |
00:13.40 | Kevin2 | <PROTECTED> |
00:13.42 | Kevin2 | <PROTECTED> |
00:13.43 | lkcl | D2, D7, D9, D13, C7, C11, C14, C15 |
00:13.44 | Kevin2 | + __gccmain(); |
00:13.50 | Kevin2 | And then just add -e WinMain to LDFLAGS in the makefile. |
00:14.06 | cr2 | Kevin2: ok. |
00:15.35 | cr2 | lkcl: you should add the bitmask field to SableGPIO (like on universal) |
00:15.54 | lkcl | GPIO 15, 20, 21, 57, 77, 99, 109, 114, 116, 119 |
00:15.59 | lkcl | what's that, dude? |
00:17.01 | cr2 | http://wiki.xda-developers.com/index.php?pagename=UniversalGPIO |
00:17.03 | lkcl | oops baby needs _more_ feeding! |
00:17.05 | lkcl | ok. |
00:17.12 | lkcl | will take a look |
00:17.45 | lkcl | what am i looking at? |
00:17.55 | lkcl | oh right. |
00:18.01 | lkcl | the 2nd column |
00:19.23 | lkcl | ok. am gonna try something a bit drastic. |
00:21.46 | lkcl | i'm gonna try em alll! |
00:25.48 | cr2 | Kevin2: one more your advice needed. how can we trace the wince USB host communication endpoint ? i see it as the last stumbling block on the universal. |
00:27.03 | Kevin2 | Heh, I liked your idea of loading a usb driver that snooped the traffic. :-) |
00:27.27 | Kevin2 | The problem with USB is that it is all dma. So, we can't use the normal wi stuff. |
00:27.55 | Kevin2 | However, it should be possible to have the WI code trigger a snoop of the traffic. I'd need to know the ohci spec more before commenting on how feasible this is. |
00:28.26 | cr2 | i have looked at the snoopypro source, it's very intel-x86-specific |
00:28.47 | Kevin2 | On a similar note, earlier today I started coding up some enhancements to wi. I'd like the code to be able to do a "memory poll" on irq, trace, and breakpoint events. |
00:29.21 | cr2 | there is something in the msdn on how you can put your own usb driver to grab data from usb host driver. |
00:29.31 | cr2 | will you do the egpio too ? |
00:30.11 | cr2 | yeah, it's the "memory poll". |
00:30.22 | Kevin2 | This is useful for tracking down certain complex events. For example, we could have a trace event on the serial port do a query of the gpio space. |
00:30.52 | cr2 | yes. |
00:31.07 | Kevin2 | egpio might fall out of that too. So, every irq event, I could poll the egpio registers and report them. |
00:31.34 | cr2 | or check if they have changed. |
00:31.39 | lkcl | ok - how about this: if it's already 'on', then a gpio dump and asic3 dump will show up, on the 'unidentified' IOs, as 'on', right? |
00:31.46 | cr2 | maybe better. |
00:31.50 | lkcl | i know what i'm talking about |
00:32.11 | Kevin2 | This might help with usb - if we knew where in memory the stuff was written to/from, we could have wi poll that memory on a usb trace event. |
00:32.45 | Kevin2 | cr2: Right, each memory area would check for changes and would have a mask of bits to ignore. |
00:32.47 | cr2 | Kevin2: on universal i know it. and the enpoint locations are also known. |
00:33.08 | cr2 | try 'dump pxa27xusb' |
00:33.45 | cr2 | you will see the apache ohci location |
00:34.20 | cr2 | lkcl: i didn't get what you want. |
00:34.34 | lkcl | candidates are: A0, A1, B14, C15, |
00:34.36 | lkcl | that's it. |
00:34.48 | lkcl | that's all the 'unidentified' ASIC3 GPIOs that are 'on'. |
00:35.00 | lkcl | cr2: it's like one of those logic games. |
00:35.23 | lkcl | the ak4641 codec and reset GPIOs are 'on' and we can't tell, in wince, what they are, right? |
00:35.31 | lkcl | oh - unless in wince we can switch them off! |
00:35.42 | Kevin2 | cr2: Cool. But does it always dma out of the areas listed by the dump? (That is, are you sure it doesn't dma to/from all over memory?) |
00:35.43 | lkcl | is there a command in wince to switch off asic3 gpios? |
00:36.11 | lkcl | if i flip those ones A0, A1, B14 and C15 off, and sound doesn't play, i know i've got the right ones. |
00:36.14 | lkcl | right? |
00:36.28 | *** join/#htc-linux hlbot (n=adm@82.67.76.179) |
00:36.37 | lkcl | anyone know a haret command to flip GPIO / ASIC3 gpio on/off? |
00:36.40 | cr2 | Kevin2: the base ohci address is constant. the EP/EDs are in a limited buffer from the data i have seen. |
00:37.07 | cr2 | lkcl: by direct memory writes. |
00:37.18 | lkcl | oh bog i knew you were going to say that cr2 :) |
00:37.51 | cr2 | but you should be very careful ;) |
00:38.02 | lkcl | cr2: to explain: there are only a few 'unidentified' ASIC3 GPIO outputs left. |
00:38.18 | lkcl | there are in fact only 4 that are set to '1' - A0, A1, B14 and C15. |
00:38.37 | cr2 | find the speaker power first. |
00:38.44 | lkcl | my wishful thinking logic tells me that those _have_ to be related to sound. |
00:38.44 | lkcl | A6. |
00:38.47 | lkcl | got that one already. |
00:38.55 | cr2 | by tracing the LCD click. |
00:38.56 | cr2 | ok. |
00:38.59 | lkcl | i know that one - i've seen it go up-and-down enough times. |
00:39.04 | Kevin2 | cr2: If you know where the buffers are, can you just watch reads/writes to those buffers? |
00:39.13 | lkcl | ok. |
00:39.20 | cr2 | all inputs are excluded. |
00:39.38 | cr2 | Kevin2: yes, but i'm not an usb expert too. |
00:40.14 | cr2 | tracing this QCT serial datastream would have been much easier :) |
00:40.36 | cr2 | i mean QCT: serial port. |
00:41.42 | cr2 | lklc: can you put the "sound related" bits back to wiki ? |
00:42.36 | lkcl | que? it's missing? |
00:42.38 | lkcl | oh yes of course. |
00:43.46 | lkcl | done |
00:45.03 | cr2 | hmm. all three are outputs. |
00:46.18 | *** join/#htc-linux hlbot (n=adm@iclem.net) |
00:46.34 | lkcl | yep. |
00:46.50 | lkcl | they are. |
00:48.41 | cr2 | there is a QCT_DIAG key in the wince registry |
00:48.57 | cr2 | lkcl: one of then is the headphone power |
00:49.06 | lkcl | i guessed that much. |
00:49.25 | lkcl | oh - wait! i have an MDA headphone - just arrived today! |
00:50.08 | lkcl | bollocks. it's USB. |
00:50.39 | lkcl | that would be funny if that was a USB audio device |
00:50.58 | lkcl | ok, where was i? |
00:51.05 | cr2 | headphone |
00:51.33 | lkcl | oh yes. going to write to 0xa9200098 the value 0x8041 and then the value 0x8042 |
00:51.40 | lkcl | and play a sound |
00:53.53 | lkcl | huh. nope. sound still going |
00:54.37 | lkcl | oh oops. |
00:54.42 | lkcl | it's 0xa9200018 :) |
00:55.08 | cr2 | phw ? |
00:55.19 | cr2 | you can use physical too. |
00:55.52 | cr2 | there are two serial drivers: qctusb.dll and qct.dll |
00:58.49 | lkcl | huh. why, if i am writing to this address, is A0 and A1 still set?? |
00:59.26 | lkcl | o cos i got the wrong address ha ha |
01:00.05 | lkcl | i got it. |
01:00.35 | lkcl | sound goes off when i flip A1 |
01:00.41 | lkcl | and back _on_ when i switch it back on. |
01:00.51 | lkcl | i'm guessing that's reset |
01:03.10 | lkcl | hmm... A0 does nothing... |
01:03.18 | lkcl | has no effect |
01:05.56 | lkcl | C15 is related to acx. ha ha :) |
01:06.10 | lkcl | i just accidentally powered off acx. oops. |
01:07.27 | lkcl | yep - C15 is acx-related. |
01:10.19 | *** join/#htc-linux hlbot (n=adm@iclem.net) |
01:12.04 | lkcl | it's gotta be one of those gpio pins. |
01:16.10 | psokolovsky | cr2, lkcl: and via what i/f uda1380 gets audio data? |
01:16.25 | lkcl | i2s i presume, off top-of-head |
01:16.54 | lkcl | oh btw i'm _slowly_ working out, psoko, what the audio codec reset/power pins are |
01:17.09 | lkcl | i've identified, i think, reset, as asic3 A1 |
01:17.19 | lkcl | by directly poking the memory address for asic3 (!) |
01:17.21 | psokolovsky | lkcl, what that util you use to scan i2c bus? |
01:17.27 | lkcl | i2cdetect |
01:17.34 | lkcl | from debian |
01:17.35 | psokolovsky | lkcl, cool ;-) don't fry sth tho ;-) |
01:17.39 | psokolovsky | ok, thanks |
01:17.51 | lkcl | hey psoko it's only a device. i can always get another one. |
01:18.07 | lkcl | and i have an htc hermes to play with if i fry it (arrived this morning) |
01:18.10 | psokolovsky | lol ;-) |
01:18.19 | psokolovsky | congrats! |
01:18.34 | lkcl | it's a _really_ nice-looking device, too. small, lightweight |
01:18.46 | lkcl | htc really learned their lessons well with this one, from the universal. |
01:19.03 | lkcl | utms, two cameras, samsung processor (so it has 6 days standby) |
01:21.27 | psokolovsky | nice, hope S3Cxxxx devices will get nice kick soon ;-) |
01:21.50 | cr2 | psokolovsky: uda1380 on the phones has very tricky setup |
01:22.11 | psokolovsky | cr2, well, how it used at all? same i2s? |
01:22.17 | cr2 | the input is I2S in ssp mode, and the output is via ssp |
01:22.35 | cr2 | don't ask me how it ia all routed :) |
01:22.53 | psokolovsky | oh my... |
01:23.02 | cr2 | there is also the bluetooth in SSP mode, and the BB :) |
01:23.50 | cr2 | use 'wi' to trace i2c |
01:24.16 | cr2 | set trace 0xXXXX1688 |
01:24.30 | cr2 | XXXXdepends on your mapping. |
01:24.44 | lkcl | to clear a GPIO pin i just set the right GPCR bit(s) right? |
01:25.03 | cr2 | psokolovsky: you should post your wince static memory mapping. |
01:25.12 | cr2 | lkcl: yes. |
01:25.14 | lkcl | ok |
01:25.20 | lkcl | here goes GPIO 3 :) |
01:25.38 | psokolovsky | cr2, where? ;-) |
01:25.54 | cr2 | hmm. it seems i have found it. |
01:25.54 | cr2 | HTC3GVT.dll |
01:26.15 | cr2 | psokolovsky: you have wince2003 ? |
01:26.21 | psokolovsky | yep |
01:26.33 | cr2 | heh. i have deleted it. |
01:27.01 | psokolovsky | I can post haret' dump mmu on wiki if it's useful |
01:27.26 | cr2 | it's too much. |
01:27.31 | cr2 | i need nk.exe |
01:27.51 | cr2 | do you know where your bootloader ends ? |
01:28.00 | cr2 | probably at 0x40000 |
01:28.07 | cr2 | as usual. |
01:28.16 | cr2 | can you pastebin the |
01:28.25 | cr2 | pd 0x40000 0x2000 |
01:28.25 | cr2 | ? |
01:28.33 | cr2 | sorry. |
01:28.42 | cr2 | pd 0xa0040000 0x2000 |
01:37.53 | psokolovsky | Ok, so hereby I declare that I failed to solve hx4700 sound issue, or even proceed sufficiently to find the cause. |
01:38.14 | psokolovsky | Suspects are haret boot, pxa27x, and of course, ak4641 codec |
01:39.05 | psokolovsky | There're lots of wierd stuff there, however |
01:39.33 | psokolovsky | look for example, that machine chims for uda are not like for ak |
01:39.38 | psokolovsky | how that could happen? |
01:40.02 | psokolovsky | but converting ak to uda's format lock up the machine |
01:40.06 | Kevin2 | psokolovsky: Did you try with gnu-haret? I'm not sure what haret could do to mess up sound. |
01:40.17 | psokolovsky | Kevin2, yes, tried gnu-haret |
01:40.26 | Kevin2 | Do you have a pxa27x cpu? |
01:40.58 | psokolovsky | Kevin2, I'm not sure to what, but that discrepancy with my test env and what other people reported as working |
01:41.08 | psokolovsky | Kevin2, the istuation is: |
01:41.20 | psokolovsky | pxa255+uda1380 works, |
01:41.29 | psokolovsky | pax270+ak4641 - doesn't |
01:41.55 | psokolovsky | so, we could try cris-cross testing, |
01:42.23 | psokolovsky | plus try to find out is theer pxa27x device with i2s sound working with haret boot at all |
01:43.42 | lkcl | huh. ok. none of the GPIOs i CLRd resulted in audio going off. |
01:43.42 | lkcl | odd. |
01:45.56 | lkcl | oops. i'm looking at a dump pxa27xgpio and they're different from what's on SableGPIO. oops. |
01:47.39 | cr2 | hmm. pxa270+uda1380 does not work for pH5 |
01:48.03 | cr2 | lkcl: mux_cli |
01:48.06 | cr2 | AT+CBST=134,1,0 |
01:48.17 | cr2 | http://doc.trolltech.com/qtopia2.1/html/phonelibrary.html |
01:48.28 | cr2 | i have it on the universal |
01:49.11 | cr2 | http://doc.trolltech.com/qtopia2.1/html/phonelibrary.html#10 |
01:49.15 | lkcl | good man |
01:50.40 | lkcl | but the best bit is a document - harald will be pleased |
01:52.19 | cr2 | it may make sense to list the used/supported AV codecs. |
01:52.52 | lkcl | gimme a sec... |
01:53.03 | Speedy2 | Hey guys. |
01:54.26 | psokolovsky | cr2, doesn't work? so, let's suspect pxa27x+haret |
01:55.27 | psokolovsky | cr2, I actually did bunch of dumb things, like that ak-to-uda conversion, or unapplaying this patch: http://kernel.org/git/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=99532559dc7a8e686b2cef14c780a7ad5dbd4a31 |
01:55.36 | psokolovsky | or adding /proc/dma after all ;-) |
01:56.13 | psokolovsky | well, it's all the same conclusions - nothing pulls DMA |
01:56.20 | cr2 | isn't /proc/dma and ISA thing ? |
01:56.52 | psokolovsky | cr2, who care ;-). I found a recent patch on l-a-k and adapted it for pxa |
01:57.13 | psokolovsky | so, besides i2s, dma is used for udc, as could have been expected |
01:57.17 | psokolovsky | nothing else |
01:57.31 | cr2 | :) |
01:57.42 | psokolovsky | you know, it can be anything, including me booting from nfs and thus having UDC always on... |
01:57.43 | cr2 | i didn't use UDC for a long time. |
01:57.45 | psokolovsky | it's weird |
01:57.56 | cr2 | FIR may use DMA. |
01:58.00 | psokolovsky | dod you use wifi under linux? ;-) |
01:58.02 | cr2 | and ??UART. |
01:58.13 | cr2 | i use wifi. acx100 |
01:58.23 | psokolovsky | well, both are likely on deman, of course, and none active for me |
01:58.36 | cr2 | acx100 does not use DMA. |
01:58.49 | cr2 | was a surprise for me. |
01:58.50 | lkcl | ok. gpio 10, 111 and 112 i'm testing next, which are unknown GPIOs that were marked as inputs on SableGPIO but are actually outputs |
01:58.58 | psokolovsky | so well, if you confirm magacian doens't work, that's at least one constraint |
01:59.18 | psokolovsky | I'm going to wait for bd2 to test sound on native linux boot |
01:59.22 | cr2 | lkcl: the list was copied from beetles ;) by you. |
01:59.28 | lkcl | oh. was it? ha ha |
01:59.36 | lkcl | that's very funny |
01:59.38 | cr2 | psokolovsky: it can be the uda1380 setup. |
01:59.43 | cr2 | on magician |
01:59.55 | psokolovsky | well, maybe |
01:59.56 | lkcl | so it's probably complete bullshit |
02:00.05 | cr2 | as i have said, the phones using uda1380 use ssp+i2s for the data. |
02:00.18 | psokolovsky | ah, damn... |
02:02.18 | lkcl | ah _ha_ i think i found another acx/wifi gpio :) |
02:02.22 | lkcl | ha ha |
02:02.59 | lkcl | GPIO 111 |
02:03.24 | cr2 | lkcl: do you have the bitmask now ? |
02:03.34 | lkcl | no - but i'm using your page :) |
02:03.41 | cr2 | ok. |
02:03.53 | lkcl | ok. GPIO 111 is acx related: my wifi just dropped out when i changed it. |
02:03.57 | lkcl | xxxxing hooray. |
02:05.58 | lkcl | that's worth trying out. |
02:10.57 | lkcl | uhhh... i think it works |
02:11.10 | lkcl | about to try ifconfig wlan0 up... |
02:11.22 | lkcl | it WORKS! |
02:11.27 | lkcl | ah you little fucker |
02:11.29 | lkcl | it works. |
02:11.55 | lkcl | get_mask 0x00000000, set_mask 0x00000000 - after update |
02:11.55 | lkcl | sta_list_add: sta=00:14:A5:54:34:CC |
02:11.55 | lkcl | sta_list_add: sta=00:12:17:24:A3:FC |
02:11.55 | lkcl | acx_i_timer: adev->status=1 (SCANNING) |
02:11.55 | lkcl | continuing scan (1 sec) |
02:11.56 | lkcl | /lib/modules/2.6.18-hh0/kernel # iwlist wlan0 scan |
02:12.00 | lkcl | iwlist wlan0 scan |
02:12.02 | lkcl | wlan0 Scan completed : |
02:12.04 | lkcl | <PROTECTED> |
02:12.06 | lkcl | <PROTECTED> |
02:12.08 | lkcl | <PROTECTED> |
02:12.10 | lkcl | <PROTECTED> |
02:12.12 | lkcl | <PROTECTED> |
02:12.14 | lkcl | <PROTECTED> |
02:12.16 | lkcl | <PROTECTED> |
02:13.55 | cr2 | conrgatulations :D |
02:14.06 | cr2 | i'm going to bed. |
02:14.11 | cr2 | good night. |
02:14.43 | lkcl | ok dude! |
02:18.30 | Kevin2 | I just booted 2.6.18-hh0 on my apache. |
02:23.02 | lkcl | Kevin2: is that.. like... good, then? :) |
02:28.43 | psokolovsky | lkcl, damn, congrats!!! ;-) |
02:28.50 | lkcl | he he wheeeeeeeee |
02:29.00 | lkcl | now i want that damn SOUND gpio damnit |
02:29.49 | psokolovsky | and feel really sick from those ak4641 issues ;-( cannot sleep, need to hack something out too ;D |
02:30.00 | lkcl | psoko ok |
02:30.01 | lkcl | ok. |
02:30.12 | lkcl | well, here's what i'm trying - and it resulted in me finding _one_ of them. |
02:30.14 | lkcl | ok. |
02:30.27 | lkcl | do you have a complete page listing all of the _known_ GPIOs? |
02:30.42 | lkcl | for the hx4700? |
02:30.49 | lkcl | oh - yours is a DMA problem, isn't it |
02:31.14 | psokolovsky | yep, it was known to work |
02:31.19 | lkcl | arse. |
02:31.28 | lkcl | ok then i probably have that to look forward to ha ha |
02:31.56 | psokolovsky | yep, have fun! ;-) |
02:32.33 | lkcl | btw psoko do you know any good docs on writing tty line discipline drivers? |
02:33.10 | lkcl | i am considering writing one which puts 0x2 at the front, and writes a checksum at the end |
02:37.50 | Kevin2 | psokolovsky: Is there anything sound related I can test on my apache? |
02:38.53 | psokolovsky | Kevin2, sure, please try your luck with that ak4641 thing ;-) |
02:39.09 | psokolovsky | I see apache is also pxa27x+ak, right? |
02:40.38 | Kevin2 | Yes. |
02:41.10 | lkcl | Kevin2: i managed to find one of the IO pins for the ak4641 by directly writing to asic3 memory. |
02:41.14 | Kevin2 | psokolovsky: Is there code I should start looking at? The apache doesn't have anything defined for sound yet. |
02:41.33 | lkcl | it would be very helpful to have that as a haret command, for both asic3 and pxa gpio |
02:41.38 | lkcl | but not essential. |
02:42.07 | Kevin2 | I think I have found all the gpio pins necessary. I just don't know how to wire up the kernel to utilize it all. |
02:42.12 | lkcl | basically what i did was go through all of the as-yet-unidentified GPIO outputs that were set to '1' and pull them low. |
02:42.14 | lkcl | you have? |
02:42.28 | lkcl | ok, then copy sound/arm/htcuniversal_sound.c |
02:42.43 | lkcl | to htcapache_sound.c |
02:42.49 | lkcl | have you got: |
02:42.53 | psokolovsky | Kevin2, h4700/htcuniversal/sth else too - tehy all the same code |
02:43.20 | lkcl | codec power-up; codec reset; speaker; headphone |
02:43.23 | lkcl | that's 4 |
02:43.40 | lkcl | you don't really need the headphone but if you've got it it will avoid you having to comment out code |
02:44.01 | lkcl | psoko: htc sable, and probably, very likely, the htc apache, too. |
02:44.49 | Kevin2 | lkcl: Yes, I've identified those 4 pins on my apache. (I don't have an asic3 - they're on egpio here.) |
02:44.54 | lkcl | great. |
02:44.55 | lkcl | ok. |
02:45.09 | lkcl | then you're all set! |
02:45.19 | lkcl | copy the htcuniversal_sound.c |
02:45.47 | lkcl | check sound/arm/Kconfig, look for HTCUNIVERSAL and copy the lines, s/UNIVERSAL/APACHE |
02:45.54 | lkcl | do the same for sound/arm/Makefile. |
02:46.27 | lkcl | when you're done on that, do a make menuconfig and you'll find your new option you added to Kconfig under sound alsa somewhere |
02:46.32 | lkcl | go for it, dude. |
02:46.41 | lkcl | we need more people to test this ak4641 |
03:03.50 | Kevin2 | psokolovsky: What is the difference between SET_HX4700_GPIO and _N variant? |
03:04.22 | psokolovsky | one is for "normal" another for negated gpios, apparently |
03:04.48 | lkcl | _N i've always seen used to do with interrupts |
03:04.51 | lkcl | oh. |
03:04.54 | lkcl | no, wrong. |
03:04.58 | lkcl | sorry - ignore me |
03:05.30 | Kevin2 | Okay. Thanks, that is what I thought. |
03:36.06 | Kevin2 | Okay. I get: PXA2xx i2Sound: HTCApache Audio activated |
03:36.13 | Kevin2 | Now for the big question -- how do I test this thing? |
03:38.24 | Kevin2 | psokolovsky: How did you test to see if sound works or not? |
03:39.09 | psokolovsky | Kevin2, the high-level test is to boot fam0.8.4/gpe, go to to Sound setting, and make should that test button does chirp ;-) |
03:39.31 | psokolovsky | that's porbably most accessible test unless you build images yourself |
03:40.00 | Kevin2 | Okay. I need to get fam/gpe installed then. I'm just using a stripped down boot environment right now. |
03:40.05 | psokolovsky | but it uses OSS emu layer of ALSA, so may add some artifacts (but it should work and does work) |
03:40.20 | psokolovsky | low-level test is to use alsa utils package |
03:40.25 | Kevin2 | Can you direct me to some pointers for installing/running a distribution? |
03:40.27 | psokolovsky | http://www.handhelds.org/moin/moin.cgi/AlsaHandheld |
03:40.57 | psokolovsky | Kevin2, you probably can d/l install tarball for close device (hx4700 apaprently), and boot it with your kernel |
03:41.03 | psokolovsky | most stuff should work ;-) |
03:41.15 | psokolovsky | http://familiar.handhelds.org/releases/v0.8.4/install/download.html |
03:41.50 | psokolovsky | and boot via NFS or from SD card (I usually do the 1st, but again, that may give some artifacts) |
03:42.04 | psokolovsky | (well, any type of boot gives some of course ;-) ) |
03:42.34 | psokolovsky | or maybe hx4700 is wrong choice if you have qvga screen |
03:42.56 | psokolovsky | all in all, I booted h2200's distro on h4000 at stoneage times ;-) |
03:43.27 | psokolovsky | you cna be sure that there's no iwmmx optmization or something there anyway |
03:44.33 | Kevin2 | I have standard 240x320 screen. |
03:45.01 | psokolovsky | h2200 may be a good choice then, it works on other devices ;-) |
03:45.18 | Kevin2 | But those images look like they want to change the rom. I need something I can just expand to a filesystem. Then I need to run make modules_install from my kernel. |
03:46.56 | psokolovsky | Kevin2, of course they provide tarball in addition to jffs2 image |
03:46.59 | psokolovsky | 1sec |
03:47.39 | psokolovsky | Fish here: http://familiar.handhelds.org/releases/v0.8.4/install/files/h2200/ |
03:49.56 | Kevin2 | Cool. In my last compile I pulled in the corgi bl stuff - my screen just blanked instead of the normal ghost to white stuff. |
03:51.17 | psokolovsky | you mean it worked as it should or made LCD always-black? |
03:52.22 | Kevin2 | No - it worked like it should - I was idle for several minutes and it went to black instead of fading to white. |
03:53.03 | psokolovsky | so, nice! that's how it should become once we started to use common stuff ;-) |
03:53.33 | psokolovsky | I really don't know what issues cr2 has on universal, but that's likely to gpio sequence, not corgi-bl per se |
03:54.06 | psokolovsky | Kevin2, well, and now some less cheerful news - haret 0.4.3 doeesn't boot on h3900 |
03:54.14 | Kevin2 | I still need to code up the lcd enable/disable sequence. I'm just doing a cken on / off in the stuff I cloned. |
03:54.54 | Kevin2 | psokolovsky: Try the latest haret (either by building from cvs or svn). |
03:54.56 | psokolovsky | Kevin2, I see. h4000 needs some cleanup too |
03:55.21 | psokolovsky | Kevin2, ok, will try. just fyi |
03:55.21 | Kevin2 | Create a file "earlyharetlog.txt" -- when this file is present, it should give early diagnostic messages in haretlog.txt. |
03:56.00 | Kevin2 | The latest binary cr2 and I worked on is in svn as haret.exe-gccmain |
03:56.08 | Kevin2 | svn://www.shadowmite.com/home/svn/ppc6700/haret |
03:58.06 | psokolovsky | ok |
03:58.09 | Kevin2 | psokolovsky: Does haret start, but not boot linux -- or does it not start at all? |
03:58.23 | psokolovsky | will try to build myself after all probably, but tomorrow then |
03:58.53 | psokolovsky | it hangs with logo still displayed after therrmometer is up and it should run kernel |
03:59.56 | Kevin2 | Okay. haret-0.4.3 is the original linux boot code. I've never had much luck with it. I'd be interested to hear if the new boot code works for you (haret >= 0.4.4). |
04:00.38 | psokolovsky | ok |
04:12.43 | psokolovsky | well, at least fixed suspend/resume code for h3900 with native wince bootloader ;-) |
04:13.23 | lkcl | argh A0 and A1 _are_ sound-related. |
04:14.07 | Kevin2 | psokolovsky: Would you recommend testing with gpe or opie images? |
04:14.09 | lkcl | just going to try them... *sigh* |
04:14.25 | lkcl | like i thought i'd tried them a dozen times before... |
04:14.29 | psokolovsky | Kevin2, I use gpe ;-) |
04:14.45 | lkcl | eeuw, psoko |
04:16.06 | psokolovsky | we talk about common testing settings here ;-) you don't think ak suddenly will work with opie? ;-) |
04:22.35 | lkcl | ok i thought i'd best read the datasheet on the ak4641, like. |
04:22.58 | lkcl | it requires a HIGH TO LOW transition on one of the pins, to perform the reset of internal registers. |
04:22.59 | lkcl | oops. |
04:23.14 | lkcl | and there's also a delay required |
04:26.19 | lkcl | i might be getting somewhere... |
04:28.20 | psokolovsky | hope you'll discover what was itching us... |
04:28.26 | Kevin2 | Cool! TS worked out of the box on that gpe image. |
04:28.48 | lkcl | yaay well done Kevin2 |
04:29.46 | psokolovsky | Kevin2, that's nice ;-) soon buttons will work too (in OE.dev at least). of course, if you guys support common buttonmap idea. |
04:31.03 | lkcl | cool. reboot actually works. |
04:31.54 | lkcl | saves me pulling the battery out |
04:32.27 | lkcl | wish i'd known that earlier |
04:32.42 | lkcl | i've taken the battery out about... 200 times so far... |
04:33.02 | lkcl | i can't believe wireless works. |
04:33.32 | psokolovsky | lkcl, so it really works? what's the speed? |
04:33.36 | lkcl | yep. |
04:33.44 | lkcl | 802.11b i think |
04:33.48 | lkcl | it's acx100 |
04:34.02 | Kevin2 | Do you know what the default ip gpe uses? |
04:34.11 | Kevin2 | (for usb link) |
04:34.18 | psokolovsky | yep, what actual throughput you get? you transferred some 100mb file, right? ;-) |
04:34.28 | lkcl | psoko: don't be daft :) |
04:34.36 | psokolovsky | Kevin2, better, pass ip= on kernel command line |
04:34.41 | lkcl | it's listed in /etc/network/interfaces Kevin2 |
04:34.44 | psokolovsky | no default afaik, need to configure |
04:41.17 | Kevin2 | psokolovsky: The "test" button is grayed out. I think it happened because gpe started before I could load the htcapache sound module. |
04:41.31 | Kevin2 | Is there a way I can tell it to restart the sound stuff? |
04:41.37 | psokolovsky | Kevin2, (un)tick som checkboxes pls ;-) |
04:42.00 | psokolovsky | Kevin2, ah, you load modules? |
04:42.17 | psokolovsky | well, /etc/init.d/gpe-dm stop / start |
04:42.53 | psokolovsky | it still should pick up in real time, I guess, |
04:43.03 | psokolovsky | so, that's that "automatic volume" |
04:43.13 | psokolovsky | who'd think you couldn't tets sound with it, lol |
04:43.25 | psokolovsky | wince is not that braindead, after all ;-) |
04:43.44 | Kevin2 | Oh, you're right - if I unset "automatic volume" than I can test. |
04:44.32 | Kevin2 | Well, I don't hear anything. But I'm not getting any useful diagnostics either. |
04:45.14 | psokolovsky | Kevin2, see dmesg |
04:45.53 | Kevin2 | Oooh - the brightness meter works out of the box. |
04:45.59 | psokolovsky | ;-) |
04:46.03 | lkcl | Kevin2 - what device is this? |
04:46.25 | lkcl | o yes htc apache. |
04:46.31 | Kevin2 | lkcl: HTC apache (aka xv6700 aka ppc6700) |
04:47.09 | Kevin2 | How do I modify volume? |
04:47.32 | lkcl | oo, you poor person you: CDMA.... |
04:47.52 | psokolovsky | Kevin2, slider in the same applet |
04:48.09 | psokolovsky | Kevin2, anything in dmesg? |
04:48.39 | lkcl | are you seriously telling me that the htc apache supports usb host??? |
04:49.04 | lkcl | like... i could like... plug a usb keyboard into it?? |
04:49.23 | Kevin2 | psokolovsky: Unfortunately, I have my TS stuff sending debug info to dmesg -- so I can't see anything useful. I'll need to recompile. |
04:49.43 | Kevin2 | lkcl: I can network between my desktop and phone via usb. |
04:49.54 | psokolovsky | I see. or try via ssh? |
04:50.06 | lkcl | that's usb client networking, not usb host. |
04:50.16 | Kevin2 | Yeah, I'm ssh'd into my phone now. |
04:50.25 | lkcl | usb gadget (g_ether) |
04:51.39 | psokolovsky | Kevin2, try that cat /dev/random >/dev/dsp then |
04:52.21 | Kevin2 | I don't see anything from dmesg. |
04:53.28 | psokolovsky | too bad |
04:53.31 | lkcl | i'm nearly done for the night |
04:53.35 | lkcl | starting to sink... |
04:53.52 | psokolovsky | but I didn't realy do myself that cat, so cannot say |
04:54.04 | psokolovsky | ALSA sound/core/pcm_native.c:1510: playback drain error (DMA or IRQ trouble?) |
04:54.08 | Kevin2 | I'm going to recompile without ts debug messages - that'll help. |
04:54.12 | psokolovsky | is what both cr2 and me get |
04:56.22 | Kevin2 | Is there an equivalent file for /etc/rc.local? I need to load a keyboard map file for apache, but don't know where to add it in the startup scripts. |
05:01.07 | lkcl | ok i get UU in the i2cdetect box instead of XX now. |
05:02.03 | lkcl | i get timeouts on the i2c bus... |
05:15.38 | Kevin2 | Hrmm the only thing I get is: |
05:15.41 | Kevin2 | PXA2xx i2Sound: HTCApache Audio activated |
05:22.07 | lkcl | ok that's good. i think. |
05:23.23 | lkcl | PM: Adding info for i2c:0-0057 |
05:23.26 | lkcl | that's err... new |
05:24.01 | lkcl | where the hell did _that_ one come from??? |
05:24.05 | lkcl | 0057??? |
05:24.10 | lkcl | o gawd i bet that's the camera. |
05:25.27 | Kevin2 | Hrmm. I didn't compile in OSS support. If gpe wants oss, then it wont work, of course. |
05:25.50 | lkcl | you don't need OSS support you need OSS _emulation_ |
05:25.55 | lkcl | it's in a different area |
05:26.13 | lkcl | go into ALSA config. don't enable OSS modules |
05:26.25 | Kevin2 | Well, yeah, but I don't have either. |
05:26.31 | lkcl | i think something goes wrong, here, that crashes the i2c |
05:26.34 | lkcl | ?? |
05:27.43 | lkcl | somehow |
05:36.33 | lkcl | last boot... |
05:36.46 | lkcl | last try at sound today |
05:38.33 | Kevin2 | Okay. When I click test, the log shows it activates the rear speaker. The rear speaker stays on for about 10 seconds, and then it gets disabled. |
05:38.43 | Kevin2 | I don't see any error messages. I don't hear any sound either. |
05:39.16 | Kevin2 | (Well, I can hear a faint buzz that's from the powered up speaker, but no sound effects.) |
05:40.17 | lkcl | well, good luck - i am very out. |
05:41.21 | lkcl | load on i2cdetect (download the lm-sensors_2.10. from http://ftp.uk.debian.org and libsensors3_2.10 as well) |
05:41.57 | lkcl | extract them both with dpkg-deb -X lmsensors....deb i2c |
05:42.08 | lkcl | and dpkg-deb -X libsensors....deb i2c |
05:42.25 | lkcl | then tar -cvzf i2c.tgz i2c/ |
05:42.34 | lkcl | copy that to your apache |
05:42.42 | lkcl | and then run /usr/sbin/i2cdetect -0 |
05:42.52 | lkcl | http://www.lm-sensors.org/wiki/i2cToolsDocumentation |
05:43.10 | lkcl | if you haven't got debian, GO GET IT! :) kubuntu would do just as well |
05:43.10 | Kevin2 | We've done that in the past. There is a wiki on xda-dev with the info. |
05:43.17 | lkcl | o cool. |
05:43.42 | lkcl | if you have the initialisation correct, the ak4641 will show up on the i2cdetect scan. |
05:44.19 | lkcl | make sure that your htcapache.c file (or whatever does the initialisation of the asic3) _doesn't_ have the asic3 GPIO pins that you're pulling up _already_ set |
05:44.33 | lkcl | cos when you boot from wince, the ak4641 remains initialised, then. |
05:45.00 | lkcl | it's the only way that i accidentally managed to detect the ak4641 on the i2c bus. |
05:45.43 | Kevin2 | ls -l /sys/bus/i2c/devices/ |
05:45.45 | lkcl | when i set A0 and A1 to 0 in the asic3 initialsation (line 133 of htcsable.c) then the ak4641 disappeared from the scan |
05:45.50 | Kevin2 | lrwxrwxrwx 1 root root 0 Nov 23 00:45 0-0012 -> ../../../devices/platform/pxa2xx-i2c.0/i2c-0/0-0012 |
05:46.04 | Kevin2 | cat /sys/bus/i2c/devices/0-0012/name |
05:46.04 | Kevin2 | ak4641-i2c at 0-0012 |
05:46.11 | Kevin2 | So, I think it is binding on the i2c side okay. |
05:46.57 | lkcl | not necessarily. |
05:47.01 | lkcl | detect is not the same as initialised. |
05:47.15 | lkcl | i have the same thing as you |
05:47.37 | lkcl | but, now that i have the name, my entire i2c bus is crap. |
05:47.49 | lkcl | it doesn't help that the modules are stuffed and cannot be unloaded properly |
05:47.51 | lkcl | rmmod'd |
05:47.55 | Kevin2 | Okay. I definitely clear the pins at startup. (My printks also confirm this.) |
05:48.21 | lkcl | do an lsmod |
05:48.49 | lkcl | how many things are listed as using i2c_pxa? |
05:49.04 | lkcl | is the Used by count equal to 1, but no modules listed? |
05:49.19 | Kevin2 | I have i2c builtin. |
05:50.06 | lkcl | ahh :) |
05:50.31 | lkcl | my i2c bus is crashed. |
05:50.55 | Kevin2 | It is possible I may have power/reset reversed.. I was really only guessing which was which. |
05:52.44 | lkcl | he he |
05:52.51 | lkcl | yep - you'll soon find out :) |
05:53.19 | Kevin2 | Good night. |
05:53.38 | lkcl | night dude |
07:03.25 | *** join/#htc-linux WizMaui (n=WizMaui@62.112.90.179) |
07:04.44 | *** join/#htc-linux rob_w (n=rob_w@p549B9B81.dip0.t-ipconnect.de) |
07:36.17 | *** join/#htc-linux rob_w (n=rob_w@p549BCB98.dip0.t-ipconnect.de) |
08:03.25 | *** join/#htc-linux Fossi (n=cmewes@195.177.48.115) |
08:11.28 | *** join/#htc-linux TripleDES (n=sergio@84.77.161.251) |
08:24.26 | *** join/#htc-linux rob_w|mis (n=rob_w@p549BBF0F.dip0.t-ipconnect.de) |
08:50.26 | *** join/#htc-linux goxboxlive (n=goxboxli@9.80-202-160.nextgentel.com) |
08:53.29 | goxboxlive | good morning |
08:56.00 | Speedy2 | Hi |
09:09.49 | *** join/#htc-linux goxboxlive (n=goxboxli@9.80-202-160.nextgentel.com) |
09:53.44 | *** join/#htc-linux TripleDE1 (n=sergio@84.77.163.149) |
09:58.46 | goxboxlive | psokolovsky: You asked yesterday what failure message we got on our universal when sound driver is loaded during boot: Here is an output: http://pastebin.ca/255767 |
10:18.46 | *** join/#htc-linux psokolovsky_ (n=psokolov@ip.85.202.124.214.dyn.sub-9.broadband.voliacable.com) |
10:39.42 | *** join/#htc-linux Timmeh (n=pope@mail.ccww.co.uk) |
10:49.05 | *** join/#htc-linux TripleDES (n=sergio@84.77.155.74) |
10:56.17 | goxboxlive | psokolovsky_ ': I saw that you stayed up late last night. Are you awake? |
11:06.16 | goxboxlive | Kevin2 / lklc /psokolovsky_ : I have made the lmsensors and som other developer stuff. You can can download them from here: http://gnulinux.biz/files//Universal/developer-stuff/ |
11:06.56 | goxboxlive | s/lklc/lkcl/ |
11:09.00 | *** join/#htc-linux g3gg0_ (n=g3gg0@ppp-82-135-15-150.dynamic.mnet-online.de) |
11:10.00 | cr2 | goxboxlive: i have found an interesting file yesterday. |
11:10.47 | goxboxlive | what kind of file? |
11:11.39 | cr2 | htc3gvt.dll |
11:11.47 | cr2 | in \windows |
11:11.47 | goxboxlive | :-) |
11:11.59 | goxboxlive | the let's rip apart |
11:12.15 | cr2 | it's not a xip file, it can be just copied. |
11:12.22 | goxboxlive | cool |
11:12.46 | cr2 | it calls AT+CBST=134,1,0 |
11:12.47 | goxboxlive | cr2: I saw that Kevin2 had managed to shut of lcd |
11:12.53 | goxboxlive | ok |
11:13.14 | cr2 | he does not have any gpio code at all. he just have switched the clock off. |
11:13.21 | goxboxlive | ok |
11:13.31 | goxboxlive | He is also using the corgi |
11:14.02 | cr2 | http://doc.trolltech.com/qtopia2.2/html/phonelibrary.html#10 |
11:14.47 | goxboxlive | Yes i have seen it |
11:14.55 | cr2 | goxboxlive: look at the corgi-bl.c code, and you'll see a light at the end of the tunnel ;) |
11:15.08 | goxboxlive | SHall i try to run it? |
11:15.54 | cr2 | if you'll run 'strings' and 'strings -el' on htc3gvt.dll you'll get many interesting things. |
11:16.04 | cr2 | forget about the backlight now. |
11:17.17 | cr2 | VT_Fake324M_CommDump.bin |
11:17.18 | cr2 | \VT_Comm_Tx_ST.txt |
11:17.18 | cr2 | \VT_Comm_Rx_ST.txt |
11:17.18 | cr2 | %010lu - %010lu |
11:17.18 | cr2 | \VT_Comm_Tx.Bin |
11:17.18 | cr2 | \VT_Comm_Rx.Bin |
11:17.29 | cr2 | these are the most interesting. |
11:18.18 | cr2 | it seems to me that some logging can be enabled. |
11:19.05 | cr2 | we only need to find out how. |
11:19.07 | goxboxlive | well there where no light in that tunnel :-) |
11:19.29 | goxboxlive | ok |
11:20.27 | cr2 | and we can also list all the codecs. and compare with ffmpeg & co. |
11:21.05 | goxboxlive | ok |
11:28.10 | goxboxlive | cr2: I did a speaker test: take a look: http://pastebin.ca/255812 why isent it permitted? |
12:04.56 | *** join/#htc-linux skodde (n=skodde@unaffiliated/skodde) |
12:08.46 | lkcl | allo darlins |
12:11.08 | *** join/#htc-linux alex-begin (n=345rt@84.232.212.62) |
12:11.48 | goxboxlive | hi lkcl |
12:31.14 | cr2 | goxboxlive: because the developers did not find a better error. |
12:31.34 | cr2 | the problem is the dma stall. |
12:31.58 | goxboxlive | :-) |
12:31.58 | goxboxlive | so it has nothing to do with the audio in root group |
12:32.13 | goxboxlive | i c |
12:49.35 | *** join/#htc-linux alex-begin (n=345rt@84.232.212.62) |
13:08.02 | *** join/#htc-linux Timmeh (n=pope@mail.ccww.co.uk) |
13:09.41 | Timmeh | hi cr2 thanks for replying to my post on xda-dev (tim_programmer) |
13:10.55 | Timmeh | i thought it would be alot of work...especially for me 'the newb' |
13:11.11 | Timmeh | but hey we all gotta learn somewhere :-) |
13:13.41 | cr2 | hi Timmeh |
13:13.48 | cr2 | goxboxlive: http://www.voip-info.org/wiki/view/Asterisk+H324M |
13:16.35 | Timmeh | quick question youve got source on sourceforge and on HH.org.. are these the same or is one old etc |
13:16.35 | cr2 | h263 and amr are implemented afaik. |
13:16.51 | cr2 | sf.net is old, but working .12 kernel |
13:17.01 | Timmeh | ok great |
13:17.13 | cr2 | the real work is done on blueangel .18-hh0 in hh.org |
13:18.11 | Timmeh | cool will have to get that later (when i'm not supposed to be working ;-) ) |
13:18.26 | Timmeh | i was using the sf version |
13:18.32 | Timmeh | thought it seemed a bit slim |
13:18.42 | goxboxlive | i c |
13:43.05 | psokolovsky_ | Hi! |
13:44.15 | psokolovsky_ | goxboxlive: fsck, what can I say! I spent whole damn night yesterday hunting within DMA and I2S area, and now you say me that it can be I2C-related? funny ;-) |
13:46.43 | *** join/#htc-linux goxboxlive (n=goxboxli@9.80-202-160.nextgentel.com) |
13:47.33 | cr2 | goxboxlive: do you include kphone/pi in your opie image ? |
13:47.53 | goxboxlive | no i havent done that. |
13:47.59 | goxboxlive | Should i? |
13:48.04 | cr2 | i think yes. |
13:48.15 | goxboxlive | i will then chief :-) |
13:53.08 | cr2 | goxboxlive: we also need to removed tslib dependency on x11 and gtk. if at the end we will include asterisk in the image, we don't need any unnecessary stuff in flash. |
13:54.57 | goxboxlive | cr2: Well i have tried that before, but it ends up with no working touchscreen. |
13:58.46 | goxboxlive | cr2: I have seen that hx4700 is trying to remove the tslib, i tried also, but it didnt build. |
14:01.45 | goxboxlive | cr2: Are we able to use this: # People should consider using udev's /dev/input/touchscreen0 symlink |
14:01.45 | goxboxlive | # instead of detect-stylus |
14:02.19 | cr2 | you need to remove the gtk code from detect-stylus, and the dependencies. imho, it should be a separate package. |
14:03.00 | cr2 | if irdadump is a separate package, then gtk-detect-stylus which has heavy dependencies should be anyway. |
14:07.02 | goxboxlive | cr2: This is the depedencies of detect-stylus: DEPENDS = "virtual/libx11 xcursor-transparent-theme xrdb" |
14:13.38 | psokolovsky_ | goxboxlive: |
14:13.38 | psokolovsky_ | <psokolovsky_> Hi! |
14:13.39 | psokolovsky_ | <psokolovsky_> goxboxlive: fsck, what can I say! I spent whole damn night yesterday hunting within DMA and I2S area, and now you say me that it can be I2C-related? funny ;-) |
14:13.45 | lkcl | cr2: you're right about the A0 and A1 being ak4641 power and reset |
14:13.54 | lkcl | however, i2c is fscked |
14:14.41 | lkcl | psoko, gobox: after initialising the ak4641, the entire i2c gives timeouts on devices. |
14:14.46 | lkcl | names can be detected - once. |
14:14.46 | cr2 | goxboxlive: well, remove them. if the mainline oe is reluctant to remove this code, we will have our own version. we are talking about it for half of the year ;) |
14:15.11 | lkcl | brb darlings |
14:16.09 | goxboxlive | cr2: Are we able to make a symlink to /dev/input/touchscreen0 . If we are we might get rid of detect-stylus |
14:16.58 | psokolovsky_ | goxboxlive, can you tell me if you do anything special to get verbose i2c logging? |
14:17.15 | cr2 | there is nothing wrong in the detect-stylus itself, it's just the gtk code that has nothing to do there. |
14:17.26 | goxboxlive | cr2: ok |
14:17.46 | goxboxlive | psokolovsky_ : I have enabled debug in the kernel .config |
14:17.52 | psokolovsky_ | goxboxlive, cr2: but getting rid of detect-stylus is direction in which OE.dev moves anyway |
14:18.06 | psokolovsky_ | goxboxlive, didn't patch anything, right? |
14:18.32 | cr2 | pskokolovsky_: ok. |
14:18.48 | goxboxlive | no |
14:18.58 | psokolovsky_ | ok |
14:19.19 | goxboxlive | psokolovsky i know, have have seen that the hx4700 is getting rid of it |
14:19.27 | psokolovsky_ | I'll still leave it for bd2 to try it on flashed hx4700 first ;-) |
14:19.35 | psokolovsky_ | yeah |
14:19.53 | psokolovsky_ | and will need to do that consistently in the end ;) |
14:21.32 | *** join/#htc-linux JTRipper (n=jtripper@p54B1D552.dip.t-dialin.net) |
14:24.50 | goxboxlive | psokolovsky_ To get rid of detect-stylus, is there anything in udev i have to add? |
14:28.49 | psokolovsky_ | goxboxlive, well, udev rule should be already there |
14:28.54 | goxboxlive | good |
14:41.20 | Kevin2 | Hi |
14:54.49 | Kevin2 | Well, no luck for me with sound. It's definitely toggling the speaker / headphone gpios in the right way ( and detecting headphones), but I hear no sound. |
14:55.03 | Kevin2 | I don't get any useful diagnostic messages at all. |
14:57.10 | *** join/#htc-linux babeloued (n=Fabrice@faim1.itesoft.com) |
14:58.59 | lkcl | psoko: there's a stack of i2c support options |
14:59.01 | lkcl | # CONFIG_I2C_DEBUG_CORE is not set |
14:59.01 | lkcl | # CONFIG_I2C_DEBUG_ALGO is not set |
14:59.01 | lkcl | # CONFIG_I2C_DEBUG_BUS is not set |
14:59.01 | lkcl | # CONFIG_I2C_DEBUG_CHIP is not set |
14:59.28 | lkcl | allo Kevin2 |
14:59.35 | lkcl | hmm. |
14:59.56 | lkcl | Kevin2: can i ask you a favour: could you possibly reply on kd@hh and explain where you're at, at the moment? |
15:00.23 | lkcl | and definitely run i2cdetect once the snd-htcapache_audio module is loaded |
15:01.05 | Kevin2 | lkcl: Where I am with respect to what? |
15:07.36 | psokolovsky | Kevin2, Hi! |
15:07.46 | *** join/#htc-linux RoEn_PC (n=roen@p54A67710.dip.t-dialin.net) |
15:08.02 | psokolovsky | Kevin2, you likely don't have alsa logging enabled, that's why you didn't get anything in dmesg |
15:10.11 | cr2 | btw, do we know something about the wifi eeprom structure from the acx code ? can we read the eeprom ? this will be a good and simple i2c test. |
15:10.29 | goxboxlive | cr2: i have build without detect-stylus, but the TS doesnt work. Was it /dev/input0 that was the TS? |
15:10.59 | cr2 | the wince i2c code on the universal always checks that the asic3 gpioa0 is high. but i have never seen anything that can drop it low. |
15:11.57 | cr2 | goxboxlive: detect-stylus is a program that helps to find out that the touchscreen is at /dev/input/event0. you can hardcode it in /etc/init.d/opie |
15:12.25 | cr2 | the keyboard and all the buttons are hardcoded in the c++ code as /dev/input/event1 |
15:12.54 | cr2 | that makes me some trouble to split the gpio-keys, asic3-keys and the pxa keyboard ;-) |
15:13.37 | cr2 | and i don't know what is the right solution. |
15:14.35 | cr2 | so it stays as is, with buggy suspend/resume behaviour now. |
15:14.43 | psokolovsky | cr2, I bet, both opie and gpe actually scan devices, and/or use some mux interface |
15:15.01 | psokolovsky | all in all, both opie & gpe work OOB with split inputs |
15:15.22 | psokolovsky | (that was my concern too, if you read my original RFC to suppott shared input device) |
15:15.29 | cr2 | psokolovsky: i'm pretty sure that odevice_htc.cpp has /dev/input/event1 hardcoded. |
15:15.50 | psokolovsky | and I'm pretty sure that opie does work still ;-) |
15:16.09 | lkcl | Kevin2: getting sound |
15:16.10 | psokolovsky | ah, odevice_htc.cpp? |
15:16.19 | cr2 | yes, because we load/find the modules in the right order. |
15:16.20 | psokolovsky | then don't do that! ;-) |
15:16.31 | lkcl | cr2: good point. |
15:16.37 | cr2 | and don't use the usb/BT mouse/keyboard/whatever. |
15:16.44 | psokolovsky | do like rest of people do ;-) |
15:17.17 | cr2 | goxboxlive has BT mouse and keyboard, and has tested them with universal :) |
15:17.17 | lkcl | well if i knew how to initialise it, i would run the i2s 'command' reading thing on it |
15:17.34 | lkcl | what's the EEPROM used for, again? |
15:17.54 | cr2 | lkcl: to store the radio type, ethernet address and such things. |
15:17.58 | lkcl | ok. |
15:18.04 | cr2 | look in the acx code. |
15:18.33 | cr2 | you can dump it from /proc |
15:18.46 | cr2 | don't remember how it is implemented. |
15:19.09 | cr2 | the tiacx must have its own i2c. |
15:19.33 | cr2 | that's why i was surprised that we see it on the pxa i2c bus. |
15:19.55 | cr2 | or pretend to see. |
15:23.43 | Kevin2 | Nov 23 10:23:36 h2200 user.debug kernel: ALSA sound/core/pcm_native.c:1510: playback drain error (DMA or IRQ trouble?) |
15:26.01 | Kevin2 | I don't see anything that would indicate an i2c error. I've compiled in all the i2c debug stuff, and it seems okay. (I see reads/writes on the i2c bus with no indication of a problem.) |
15:31.59 | cr2 | Kevin2: what other dma devices do we have ? |
15:32.11 | cr2 | is it possible to force FFUART to use dma ? |
15:35.35 | *** join/#htc-linux tim_ (n=pope@mail.ccww.co.uk) |
15:37.09 | goxboxlive | hmmm, touchscreen is working when i do cat /dev/input/event0 but it doesnt work when using Opie |
15:37.33 | goxboxlive | export QWS_MOUSE_PROTO=:TPanel:/dev/input/touchscreen0 |
15:39.41 | psokolovsky | atetntion please, ak4641 is speaking! |
15:39.44 | psokolovsky | ------------------ |
15:39.45 | psokolovsky | <bd2_> HAHA! |
15:39.45 | psokolovsky | <bd2_> sound is working! :-) |
15:39.45 | psokolovsky | <bd2_> I've soft rebooted hx4700 after dma error, insmoded snd modules, and speaker test gives white noise |
15:39.46 | psokolovsky | <psokolovsky> bd2_, DON'T TOUCH IT! |
15:39.48 | psokolovsky | <psokolovsky> let people hear that wonder first! |
15:39.50 | psokolovsky | ----------- |
15:40.13 | cr2 | soft rebooted ? what's that ? |
15:41.27 | psokolovsky | he has Linux flashed |
15:41.37 | psokolovsky | so I guess, "reboot" |
15:41.57 | cr2 | but that means buggy dma setup ? |
15:42.16 | psokolovsky | maybe, pls join #handhelds.org |
15:42.54 | cr2 | hmm. need to go. will be back later. |
15:43.03 | psokolovsky | ok |
16:00.27 | *** part/#htc-linux babeloued (n=Fabrice@faim1.itesoft.com) |
16:23.14 | *** join/#htc-linux babeloued (n=Fabrice@faim1.itesoft.com) |
16:56.52 | *** join/#htc-linux jeanseb (n=jeanseb@88.164.32.155) |
16:59.00 | jeanseb | salut |
17:05.12 | psokolovsky | Kevin2, how exactly you deal with dma in haret, in 2 words? |
17:06.59 | *** part/#htc-linux babeloued (n=Fabrice@faim1.itesoft.com) |
17:11.48 | *** join/#htc-linux pH5 (n=ph5@p5485F917.dip.t-dialin.net) |
17:13.47 | cr2 | hi again |
17:14.09 | cr2 | pH5: did you find the OWM in AIC3 ? not asic3. |
17:16.55 | pH5 | hey cr2, yes I did. |
17:17.57 | cr2 | do you have the code ? |
17:19.36 | pH5 | cr2: yup, but now I use mreimer's ds1wm driver instead. |
17:20.55 | cr2 | ok, but how do you deal with it ? |
17:21.45 | pH5 | http://en.pastebin.ca/255994 |
17:22.31 | pH5 | I didn't have time to write the battery driver yet, so I'm missing the connection of w1_ds2760 to the battery classdev |
17:24.00 | cr2 | hmm. i suspect the same in blueangel. it uses +2, +4, +6 and +8 registers. |
17:25.35 | pH5 | the blueangel aic2 has the ds1wm too? |
17:26.17 | pH5 | then we should write a common aic2/aic3 driver that can be connected to ds1wm and a led driver for both ba and magician |
17:26.46 | cr2 | the battery uses something there. the +0xA offset has the leds. |
17:26.59 | cr2 | aic2 ? |
17:27.18 | pH5 | isn't it labeled AIC2 in the blueangel? |
17:27.29 | cr2 | i'll look. |
17:27.51 | cr2 | there is OWM irq, but i have not yet found it. |
17:28.03 | *** join/#htc-linux pleemans (n=peter@d51A5E421.access.telenet.be) |
17:28.06 | pH5 | the led driver sure is similar. |
17:28.25 | cr2 | the current code uses tsc2200, and afaik there are some problems. and it's not clear where is the second battery. |
17:29.54 | cr2 | yeah, looks like aic2. |
17:30.26 | pH5 | does it have gpios? |
17:30.41 | pH5 | (besides what the led driver uses) |
17:31.30 | pH5 | the magician has a huge list of gpio registers that are setup in "InitPASIC3GPIO" or something like it in nk.exe. |
17:32.07 | pH5 | But the only gpios that I found are changed during operation so far are the led gpios and the one to turn on/off the W1. |
17:32.42 | pH5 | (I have no idea whether that's just a gpio connected to the EN pin of the DS1WM or perhaps some kind of cken bit) |
17:32.55 | *** join/#htc-linux BabelOued (n=Fabrice@82.238.28.28) |
17:33.07 | cr2 | i need to check the code in more detail. |
17:33.16 | *** join/#htc-linux JTRipper86 (n=jtripper@p54B1F54C.dip.t-dialin.net) |
17:33.29 | cr2 | it has 2 areas. this "battery" one, and the leds. |
17:34.01 | cr2 | there is something similar also in the alpine cpld. sounds logical after all. |
17:36.00 | psokolovsky | Kevin2, cegcc SVN HEAD broken? |
17:37.23 | psokolovsky | nevermind, read INSTALL ;-) |
17:46.21 | pH5 | cr2: http://wiki.xda-developers.com/index.php?pagename=MagicianAIC3 |
17:48.58 | cr2 | pH5: 4 8 c 10 |
17:49.32 | cr2 | yes :) i have .bus_shift=! |
17:49.36 | cr2 | 1 |
17:50.06 | lkcl | Kevin2: thanks for that message to hh |
17:51.05 | lkcl | OK! someone has sound on the hx4700 - cool! |
17:51.29 | lkcl | ok here's a possible hint/exploration. |
17:51.44 | lkcl | when we do rev-eng we copy the gpio and asic3 params from wince. |
17:51.57 | cr2 | pH5: i don't have (use) aic gpio, it seems. |
17:52.01 | lkcl | _but_ - of course - those initialisation parameters are from _running_ - post-initialisation of wince. |
17:52.24 | pH5 | cr2: this register is used in the led driver at least |
17:52.27 | lkcl | and the A0 and A1 are a classic example of that in action: in htcsable.c initialisation of asic3, A0 and A1 are set 'high'. |
17:52.34 | cr2 | lkcl: it's pxa dma here ? |
17:52.40 | cr2 | pH5: on blueangel ? |
17:52.56 | lkcl | mmmm... there's a difference, isn't there... |
17:53.04 | pH5 | cr2: yes, I remember i wrote the magician led driver and then looked at the ba one and found it mostly similar... |
17:53.21 | cr2 | lkcl: that's why i have the 'Init' field in Foo_asic3. it is empty for sable. |
17:53.35 | cr2 | pH5: platform_driver :) |
17:53.35 | lkcl | ack |
17:54.31 | cr2 | so the current BA battery monitor code is wrong. |
17:54.44 | cr2 | maybe it looks for the backup battery ? |
17:55.07 | cr2 | pH5: on hx4700 it's the ts chip looking after the second battery ? |
17:59.18 | *** join/#htc-linux WizMaui_ (n=WizMaui@62.112.90.179) |
18:04.02 | BabelOued | hi |
18:05.14 | cr2 | hi BabelOued. the battery code need some serious reshuffle. but it's taken from himalaya and is not quite right. |
18:06.36 | BabelOued | i see your message |
18:07.14 | BabelOued | the two register from tsc2200 are AUX1 and AUX2 |
18:07.28 | BabelOued | but value seem strange |
18:08.44 | BabelOued | you've any idea ? |
18:10.17 | cr2 | i think on hx4700 the ts chip monitors the backup battery. |
18:11.37 | psokolovsky | cr2, pretty true should be. but I wonder if on h4000 it monitors both if them? so far, second voltage readings were stable, surprise ;-) |
18:12.30 | cr2 | and the main battery is monitored by OWM on this small CPLD. |
18:12.53 | cr2 | psokolovsky: do you have some asics on h4000 ? |
18:13.20 | psokolovsky | cr2, asic3 ;-P |
18:13.43 | cr2 | btw, what is more interesting, that the asic3 OWM on blueangel is not used. |
18:13.59 | cr2 | psokolovsky: is the OWM on on your asic3 ? |
18:14.05 | psokolovsky | cr2, same appears to be for h4000 |
18:14.17 | psokolovsky | dunno (ashamed) |
18:14.26 | pH5 | cr2: are all those asic3s the same chip, or are there different revisions? |
18:14.29 | cr2 | do you have motherboard pictures ? |
18:14.35 | psokolovsky | but as i told, that driver which I posted gave onlt timeouts for me... |
18:14.56 | psokolovsky | pH5, all the same, guaranteed. or we should think so. |
18:15.18 | psokolovsky | cr2, me? (motherboard pix) |
18:15.26 | cr2 | pH5: must be a different revision. older asic3 was .bus_shift=2, newer =1, your AIC3 also has different .bus_shift from ba. |
18:15.34 | cr2 | psokolovsky: yes. |
18:16.15 | psokolovsky | cr2, yep, see h4000 wiki |
18:16.20 | cr2 | ok. |
18:16.30 | psokolovsky | cr2, and yeah, there're whole 6 pins on the battery ;-D |
18:16.48 | lkcl | i like the htc hermes so much that even though i've not got a sim card in it i'm thinking of taking it along just for show :) |
18:16.55 | cr2 | psokolovsky: imho, it is on all li batteries. |
18:17.21 | pH5 | lkcl: where did you get it and for how much? |
18:17.24 | psokolovsky | cr2, it's proved to be not in h1910 (also asic3) |
18:17.35 | cr2 | lkcl: we need dump functions for haret. |
18:17.45 | cr2 | psokolovsky: i mean the contacts :) |
18:17.59 | psokolovsky | cr2, and h1910 maintainer assured me that he disassembled driver and knows that, and he thinks h4000 shoudl be the same too. |
18:18.08 | psokolovsky | cr2, h1910 has 4 contacts |
18:18.17 | psokolovsky | 5 is most common AFAIK |
18:18.27 | psokolovsky | IIRC those numbers correctly |
18:18.35 | psokolovsky | but the fact they are different |
18:18.57 | cr2 | jornada820 has the li cells with a microcontroller, which reports something over RS232 1200bps (!) |
18:18.59 | cr2 | ok. |
18:19.08 | cr2 | saves them money. |
18:19.24 | psokolovsky | lol ;-) |
18:20.02 | cr2 | i wonder where the current ba battery code comes from ;) |
18:23.09 | cr2 | psokolovsky: h3600 is strongarm ? |
18:24.01 | psokolovsky | cr2, yeah! |
18:24.45 | cr2 | i have some problems with the cpu-sa1100.c there, but i'll post about it later. |
18:25.14 | cr2 | my values for 191MHz CPU speed are different. |
18:25.30 | cr2 | with the current values the jornada820 crashes. |
18:26.45 | cr2 | psokolovsky: btw, i think the h4350 spi flags are taken from ASIC2, but it looks to me that the register layout on ASIC3 is different. |
18:27.36 | psokolovsky | cr2, hm, possibly. but there's only 1 bit diff with BA |
18:28.24 | cr2 | ok. |
18:28.35 | cr2 | 2408w on h4000. good. |
18:29.07 | cr2 | 6101 is BT ? |
18:29.56 | psokolovsky | yeah, it's that pre-6150 thing |
18:30.37 | psokolovsky | same as BA per HHX |
18:33.54 | cr2 | yes. |
18:35.18 | cr2 | hmm. it seems the SDG bootldr makes something to DMA so it is configured properly. sounds like a kernel bug to me. |
18:37.29 | psokolovsky | cr2, for me too ;-) poor leenouks kernel... ;D |
18:39.49 | *** join/#htc-linux goxboxlive (n=goxboxli@9.80-202-160.nextgentel.com) |
18:46.40 | pH5 | hm.. I can't find anything about DMA in the sdg bootldr |
18:50.25 | psokolovsky | pH5, I'd say, good reset would clear all that stuff ;-I |
18:50.27 | cr2 | pH5: "numbered" gpio settings ? and the full hardware cpu reset. |
18:51.26 | psokolovsky | Well, gentlemen, I found that pxa27x dma setup in haret is a bit less comprehensive than even for pa25x. so I ported that. |
18:51.33 | psokolovsky | results are the same ;-( |
18:53.55 | psokolovsky | anyone know anything about that "Flow through |
18:53.55 | psokolovsky | " vs " |
18:53.55 | psokolovsky | fly by" crap? |
18:54.57 | lkcl | pH5: bought it on ebay. $EUR 450. it's brand new: the guy who sold it to me had bought two of them |
18:55.30 | lkcl | cr2: what kind of dump functions, dude? |
18:55.43 | cr2 | lkcl: dump pxa27x* |
18:55.59 | cr2 | dump sc324xx* |
18:56.08 | lkcl | oh right - of course. ha ha |
18:56.20 | cr2 | then porting will be very easy. |
18:57.02 | cr2 | i don't want to know how much time was spent on rx3715 to get the register data with raw haret register access ;) |
18:57.27 | cr2 | look at mach-rx3715.c |
18:57.58 | pH5 | cr2: what is the quickest way to remove the RIL stuff from xanadux haret? |
18:58.12 | cr2 | and there will be more samsung based phones coming. |
18:58.26 | cr2 | pH5: comment out the testril() function. |
18:58.52 | lkcl | yep. it's a great processor. |
18:59.00 | pH5 | cr2: thanks |
19:00.11 | pH5 | lkcl: that sounds like a good deal. still too expensive for me. but it's good to know that linux will be up and running when I can afford one ;) |
19:00.22 | lkcl | ha ha very funny pH5 |
19:01.01 | cr2 | lkcl: that's not very challenging phone to port to. |
19:01.14 | cr2 | ak4641 driver will be completely unusable. |
19:01.55 | cr2 | i'd like to get my dirty fingers on the pseudorange data from the GL :) |
19:03.07 | lkcl | :) |
19:05.10 | lkcl | huh. that's odd. the i2c bus doesn't even _work_ - massive amounts of timeouts - and i haven't even loaded the ak4641 module. |
19:06.03 | cr2 | do you have any working devices on the bus ? |
19:14.44 | cr2 | pH5: unsigned int __iomem *iobase; |
19:15.31 | cr2 | people writing such code should be fried alive ;) |
19:16.43 | lkcl | dratdratdrATDRAT |
19:17.40 | lkcl | ok, leaving the ak4641 device up-and-running / initialised by setting A0 and A1 high in htcsable.c i can then do a detect on the i2c bus, and find the ak4641, and it's up-and-running. |
19:18.02 | lkcl | i can even do an i2cdump 0 0x12 and scan it |
19:20.20 | psokolovsky | well, so yes, there appeas to be no magic bit to reset DMA subsystem in PXA ;-( |
19:20.33 | cr2 | lkcl: A0 is some kind of "i2c enable" on the universal. maybe it's the same on sable ? btw, where is your RTC chip ? |
19:20.49 | lkcl | cr2: oh - ah. argh. right |
19:21.06 | lkcl | oh gawd that would do it, wouldn't it... |
19:21.22 | lkcl | cr2: don't know. |
19:21.31 | lkcl | *sigh*. i think it's time for me to put the camera back... |
19:21.44 | lkcl | ok am gonna do some disassembly... |
19:22.00 | lkcl | i need two known devices on the bus to find out if you're right, cr2. |
19:22.20 | lkcl | AH - yes. yep. that makes sense, cr2. a lot of sense. |
19:22.24 | cr2 | but i2cdetect does not show it ? |
19:22.56 | cr2 | it is strange that you have an i2c RTC chip, and it is not connected. |
19:23.02 | lkcl | no - because it's sticking out the back!!! |
19:23.13 | cr2 | but it may be some evil plot. it can be connected to the BB. |
19:23.22 | lkcl | i've seen two other devices on the bus - by mistake. |
19:23.26 | lkcl | 0x50 and 0x57. |
19:23.39 | cr2 | there is a small watch battery soldered in the motherboard on the universal. |
19:23.51 | cr2 | hmm. |
19:24.00 | cr2 | what was the chip name ? |
19:24.13 | lkcl | 1sec... |
19:24.31 | lkcl | rtc: R8564 EF096A - right next to ASIC3 drivers/i2c/chips/rtc8564.c |
19:24.40 | cr2 | ok. |
19:24.41 | lkcl | brb am doing some surgery |
19:24.46 | *** join/#htc-linux FossiFoo (n=Fossi@e176105224.adsl.alicedsl.de) |
19:25.43 | cr2 | btw, this RTC driver is not in the hh.org CVS. |
19:26.12 | cr2 | rtc8564.c is 0x51 |
19:27.04 | cr2 | the same as 2408w ? |
19:27.45 | cr2 | maybe, you should trace the cam first. |
19:32.31 | psokolovsky | cr2, what's that 2408w you talk about? |
19:39.24 | cr2 | wifi eeprom |
19:41.08 | lkcl | ok let's see if i broke it |
19:41.11 | lkcl | haha |
19:48.06 | cr2 | does it work ? |
19:48.53 | lkcl | yehh |
19:48.59 | lkcl | no smoke or anything. |
19:49.04 | lkcl | gotta get ready soon am going out |
19:49.12 | goxboxlive | what? |
19:49.23 | lkcl | still getting UU in the 0x12 column |
19:49.27 | goxboxlive | is working? |
19:49.38 | lkcl | cr2, can you do a .lst for me again on wavedev? |
19:50.07 | lkcl | i'm getting exasperated with this |
19:57.43 | cr2 | lkcl: done |
19:57.50 | cr2 | what about the camera ? |
19:59.32 | goxboxlive | cr2: Even though i removed the depends in the detect-stylus the opie-image is still big. And especiallly after i added kphone/pi |
19:59.43 | goxboxlive | 23.3 MB, but thats with all the developer stuff. |
20:00.55 | goxboxlive | Yes what about the camera, lets make a driver. I can start. I choose the nake, htcuniversal_cam . Now the ball is your, you do the rest :-) |
20:00.55 | cr2 | the developer stuff ? |
20:01.10 | goxboxlive | lmsensors, hcidump and so on |
20:01.20 | cr2 | ok. |
20:01.37 | cr2 | can asterisk be compiled too ? |
20:02.07 | goxboxlive | if it is in the oe i can. I doint knopw what it is, but i can take a look for it. hold on |
20:03.10 | goxboxlive | i am building it now |
20:05.02 | goxboxlive | it depends on libvorbis, i'll upload that to |
20:05.10 | cr2 | ok. |
20:11.10 | goxboxlive | cr2: It's uploaded: asterisk where 7MB . |
20:12.00 | cr2 | ok. we may reuse it later. |
20:12.50 | goxboxlive | I am trying it now. Is it a gui program?? |
20:21.19 | goxboxlive | well have no idea what it is |
20:24.01 | goxboxlive | cr2: About the the backlight. It reminds me of how it was right before you made it work in the .12. It behave the same. |
20:24.04 | goxboxlive | Like in Opie |
20:25.11 | goxboxlive | IThe backlight is dimming on/off(weak) in a cyclus i have in the opie-light-and-power. |
20:30.35 | cr2 | good idea, i'll look at corgi-bl.c now. |
20:34.55 | goxboxlive | good |
20:40.19 | psokolovsky | So well, this ak4641 thing si not conquerrable. |
20:40.41 | psokolovsky | Last thing I did, is read thru pxa27x man, |
20:40.55 | psokolovsky | figureing out how to "reset" DMA registers, |
20:41.06 | psokolovsky | and added that code to kernel DMA init |
20:41.10 | psokolovsky | still the same... |
20:42.29 | goxboxlive | this is strange, why did it work before? |
20:46.04 | cr2 | psokolovsky: there was some dma priority patch recently ? can you revert it ? |
20:46.09 | psokolovsky | goxboxlive, when "before"? |
20:46.29 | goxboxlive | psokolovsky , in the .12 xanadux kernel. |
20:46.39 | psokolovsky | cr2, in may. i tried that. doesn't change anything accept chan# used |
20:47.22 | psokolovsky | goxboxlive, do you have binary? can you boot it right now? either way, I had a look at that code - many things have changed since then |
20:47.52 | goxboxlive | psokolovsky: yes i bootet it a few days ago and tested the sound. |
20:48.08 | psokolovsky | goxboxlive, but if you will eb able to make that code run *now*, from source eventually, we'd have a thread to get us out of this crap |
20:48.21 | psokolovsky | goxboxlive, booted with haret? |
20:48.27 | goxboxlive | yes with haret |
20:48.59 | psokolovsky | goxboxlive, and sound worked? |
20:49.25 | goxboxlive | psokolovsky: It is cr2 who has made the driver (or rewritten it) so i guess you bether talk to him about it. |
20:49.40 | goxboxlive | psokolovsky: Sound worked as it should |
20:50.05 | psokolovsky | goxboxlive, well, we'd need recent first-hand witness first, you quilify ;-) |
20:50.09 | Speedy2 | re |
20:50.31 | psokolovsky | goxboxlive, worked without any tricks like dancing on one leg while booting it? ;-) |
20:50.32 | goxboxlive | Even the mic worked, though it was recording kind of messy. |
20:51.01 | goxboxlive | :-) yes, right out of the kernel compilation. Cr2: What did you do with the .12 sounddriver so it worked? |
20:51.50 | psokolovsky | goxboxlive, so, you built the kernel you booted? and it was couple of days ago? |
20:52.41 | goxboxlive | psokolovsky: I build the kernel/opie-image for about a half years ago, but nothing has changes in the .12 xanadux source three since. |
20:53.17 | goxboxlive | here is the woeking images: http://gnulinux.biz/files//Universal/sd-mmc/2.6.12/ |
20:55.19 | psokolovsky | goxboxlive, thanks, but I'd need exact tag/date of the source checkout this was built from |
20:56.23 | goxboxlive | pskolovsky: If you want, ican just recompile from cvs and try to boot it again to see if it still work? |
20:57.03 | psokolovsky | goxboxlive, actually, that would be preferred. would need fresh today's case, 100% reproducible |
20:57.47 | psokolovsky | goxboxlive, you see, xanadux appears to lack hx4700 support, so I'd need to port to it, and I want to be 100% sure it actually works now |
20:58.01 | goxboxlive | ok i give it a try |
20:59.05 | goxboxlive | ok, i am updateing the cvs now. Just remember that .12 is not gcc-4.x compatibel |
20:59.06 | goxboxlive | use 3.4 or 3.1 |
21:03.09 | *** join/#htc-linux lamikr|lap (n=chatzill@aragorn.kortex.jyu.fi) |
21:03.25 | psokolovsky | goxboxlive, yes, I have 3.4 for local builds (non-OE) |
21:03.27 | psokolovsky | thanks |
21:05.28 | goxboxlive | kernel and modules successfully buildt, aplyijng them now and trying to boot :-) |
21:11.46 | psokolovsky | ok, how it goes? |
21:13.28 | goxboxlive | it's booting |
21:13.42 | goxboxlive | do you want to see some outputs? |
21:14.04 | psokolovsky | so far, I want to hear sound ;-) |
21:17.11 | goxboxlive | well sounds works |
21:17.24 | goxboxlive | even tapping on the opie screen is clicking |
21:17.39 | goxboxlive | snd_pxa_i2sound_ngn_free_rate |
21:17.39 | goxboxlive | i2c_adapter i2c-0: master_send: writing 2 bytes. |
21:17.39 | goxboxlive | snd_pxa_i2sound_ngn_free_rate |
21:18.04 | goxboxlive | is there other ways i can prove it? |
21:18.32 | psokolovsky | goxboxlive, so, it's CVS HEAD as of today, right? |
21:18.37 | goxboxlive | yes |
21:19.06 | psokolovsky | goxboxlive, and you don't changed CPU before running this, no? ;-)) |
21:19.23 | goxboxlive | :-) |
21:19.45 | goxboxlive | i havnet done a thinng, just compiled and aplied to an earlier opie-image |
21:19.53 | goxboxlive | cvs -z3 -d:pserver:anonymous@cvs.sourceforge.net:/cvsroot/xanadux co -ko -r K2-6-12-hh2-xda0-un0 linux-2.6-xda |
21:20.12 | psokolovsky | ok, so I will check out sound drivers from it and next time I'll be in mood, will try to forward port it... |
21:20.15 | psokolovsky | ok, thanks |
21:20.37 | psokolovsky | goxboxlive, well, stop, you used HEAD or -r K2-6-12-hh2-xda0-un0 ? |
21:21.35 | goxboxlive | I am using this: co -ko -r K2-6-12-hh2-xda0-un0 linux-2.6-xda but i guess cr2 is allways aplying to HEAD. |
21:24.24 | psokolovsky | goxboxlive, well, I need *exact* version which is known to work. And -r K2-6-12-hh2-xda0-un0 is even better actually. |
21:24.56 | psokolovsky | So, thanks for checking it once again for me! We'll see, how far I'll be able to go with this... |
21:25.21 | goxboxlive | well i am using this one. The only fiffer is that i use my own username instead of anonymous cvs -z3 -d:pserver:anonymous@cvs.sourceforge.net:/cvsroot/xanadux co -ko -r K2-6-12-hh2-xda0-un0 linux-2.6-xda |
21:26.26 | psokolovsky | np, that doesn't affect code ;-) |
21:26.33 | goxboxlive | i know |
21:27.12 | goxboxlive | even alsamixer works, not with Opie though. |
21:27.54 | goxboxlive | psokolovsky: xanadux has hx4700 support |
21:28.34 | goxboxlive | it might be outdated, but there is a hx4700 directory there |
21:29.10 | psokolovsky | well, I didn't see sound support exactly, but I'll check |
21:29.56 | goxboxlive | http://xanadux.cvs.sourceforge.net/xanadux/linux-2.6-xda/arch/arm/mach-pxa/hx4700/ |
21:40.32 | *** join/#htc-linux FossiFoo (n=Fossi@e176105224.adsl.alicedsl.de) |
22:23.17 | goxboxlive | When i run this speaker test it runs the whole thing, but when i try again it hangs. I have to reload the htcuniversal sounddriver again |
22:23.19 | goxboxlive | root@htcuniversal:/etc/modutils$ speaker-test -D hw:0 -c2 -t sine |
22:23.19 | goxboxlive | speaker-test 1.0.13 |
22:23.20 | goxboxlive | Playback device is hw:0 |
22:23.20 | goxboxlive | Stream parameters are 48000Hz, S16_LE, 2 channels |
22:23.20 | goxboxlive | Sine wave rate is 440.0000Hz |
22:23.21 | goxboxlive | Rate set to 48000Hz (requested 48000Hz) |
22:23.23 | goxboxlive | Buffer size range from 256 to 16384 |
22:23.25 | goxboxlive | Period size range from 128 to 8192 |
22:23.27 | goxboxlive | Using max buffer size 16384 |
22:23.29 | goxboxlive | Periods = 4 |
22:23.33 | goxboxlive | was set period_size = 4096 |
22:23.36 | goxboxlive | was set buffer_size = 16384 |
22:23.37 | goxboxlive | <PROTECTED> |
22:23.39 | goxboxlive | <PROTECTED> |
22:23.41 | goxboxlive | Time per period = 5.800739 |
22:23.43 | goxboxlive | <PROTECTED> |
22:23.45 | goxboxlive | <PROTECTED> |
22:23.47 | goxboxlive | Time per period = 5.998818 |
22:23.49 | *** part/#htc-linux RoEn_PC (n=roen@p54A67710.dip.t-dialin.net) |
22:23.50 | goxboxlive | <PROTECTED> |
22:23.51 | goxboxlive | <PROTECTED> |
22:23.53 | goxboxlive | Time per period = 5.998805 |
22:23.56 | goxboxlive | <PROTECTED> |
22:23.57 | goxboxlive | <PROTECTED> |
22:23.59 | goxboxlive | Time per period = 5.998805 |
22:24.01 | goxboxlive | <PROTECTED> |
22:24.05 | goxboxlive | <PROTECTED> |
22:24.07 | goxboxlive | Time per period = 5.998825 |
22:24.09 | goxboxlive | <PROTECTED> |
22:24.11 | goxboxlive | <PROTECTED> |
22:24.13 | goxboxlive | Time per period = 5.998867 |
22:24.15 | goxboxlive | <PROTECTED> |
22:24.17 | goxboxlive | <PROTECTED> |
22:24.19 | goxboxlive | But still no sound, juss a buzz and some ticking. |
22:27.29 | Speedy2 | Is it possible the channel(s) are muted? |
22:28.01 | goxboxlive | hmm |
22:28.26 | goxboxlive | well, when i amusing alsmixer the sound is on, buthow do i check if they are? |
22:28.28 | Speedy2 | If I mute a channel on my equipment, I can hear slight ticking when the channel is activated. |
22:28.40 | Speedy2 | goxboxlive: It's possible something within the initalization of that sound chip is off |
22:28.44 | goxboxlive | If it is that easy i throw my self down in the toilett |
22:29.01 | Speedy2 | Well, I'm not saying that's what it is. But that is a possibility. |
22:29.13 | Speedy2 | What's the model chip? |
22:29.18 | Speedy2 | ak something? |
22:29.28 | goxboxlive | hold on |
22:30.06 | goxboxlive | ak4641 |
22:31.56 | Speedy2 | It has stereo out and mono out |
22:32.49 | goxboxlive | yes, the monois the phone speaker and the stereo is the main pda speaker |
22:33.47 | Speedy2 | There is a "softmute" function on this unit |
22:34.33 | goxboxlive | ok, and how do i activate/deactivate it. |
22:35.13 | *** join/#htc-linux doepiedoe (n=doepiedo@84-53-96-51.fiber.unet.nl) |
22:35.33 | cr2 | goxboxlive: does the lcd suspend work correctly in .12 ? |
22:36.16 | goxboxlive | cr2: Yes it shuts all off when the machine is not used for a while |
22:36.27 | cr2 | ok. |
22:37.52 | Speedy2 | Look at page 27 of the spec |
22:37.52 | Speedy2 | one sec |
22:38.00 | Speedy2 | http://www.handhelds.org/platforms/hp/ipaq-hx47xx/ak4641.pdf |
22:38.27 | Speedy2 | It's 0x06 in the register control |
22:40.51 | cr2 | psokolovsky: can we revive the .12 driver ? |
22:41.32 | psokolovsky | cr2, well, I'm going to try that if noone else beat me on that |
22:41.57 | psokolovsky | cr2, but it's not like "driver", it's the whole infrastructure, so it won't be easy... |
22:42.15 | cr2 | i don't see any other option. i'm already running the .12 touchscreen driver. |
22:42.17 | cr2 | hmm. |
22:46.22 | cr2 | psokolovsky: you should be beaten with a big thick stick :) |
22:46.43 | cr2 | asic3_set_gpio_out_b ... 1<<GPIOD_FL_PWR_ON |
22:46.56 | psokolovsky | cr2, for what? ;-) |
22:47.11 | psokolovsky | cr2, did I make a typo? |
22:47.22 | cr2 | i want more error-resistant asic3 functions. fuck that. |
22:47.33 | *** join/#htc-linux goxboxlive (n=goxboxli@9.80-202-160.nextgentel.com) |
22:47.34 | cr2 | gpio_b for D gpio. |
22:47.56 | psokolovsky | cr2, we should have thru-numbering, period |
22:48.01 | psokolovsky | cr2, what file? |
22:48.02 | cr2 | agreed. |
22:48.16 | cr2 | htcuniversal_bl.c |
22:49.41 | psokolovsky | cr2, aa-a-ah! ;-E poor guys. who'd know it turns out a "bear's service", speaking russian ;-) |
22:49.53 | cr2 | ;) |
22:50.04 | psokolovsky | sorry!! |
22:50.13 | cr2 | i think the api is too lowlevel. |
22:50.26 | cr2 | i have made such mistakes too. |
22:50.29 | psokolovsky | yep |
22:56.23 | cr2 | corgi-bl is still not doing the right thing. the proper pxa-pwm driver should handle the ALT gpio settings on resume. |
23:00.06 | psokolovsky | cr2, are they reset for you or what? |
23:00.36 | cr2 | goxboxlive: can you compile the newest kernel ? |
23:01.06 | cr2 | i think each driver should handle its own resources itself, and not rely on some other code. |
23:01.23 | goxboxlive | yes, but do you know why my laptop is getting so damn slow after i have compiled a kernel. It's getting so slow that i have to reboot it. |
23:01.50 | cr2 | dmesg ? |
23:02.02 | psokolovsky | cr2, well, cpu registers across suspend should preserve CPU core code at all. so, I still don't understand what issue you have |
23:02.29 | psokolovsky | cr2, Kevin2 yesterday had only good things to say after trying it on apache |
23:02.48 | psokolovsky | so I still wonder is there some bug... |
23:03.22 | cr2 | well, the code is so simple there, nothing can be wrong :) |
23:03.40 | cr2 | it's more about working code vs. "the right thing" |
23:04.08 | psokolovsky | yep, exactly. and for Kevin2 it out if the box worked right, better than old code (for me and other ports too) |
23:04.25 | cr2 | i have problems with it being used on blueangel |
23:04.52 | cr2 | the pwm setup there is much more complicated. |
23:05.01 | psokolovsky | cr2, well, just one thing - only today I talked with one of palm guys, and tehy have exactly pxapwm_bl driver, which they'd like to mainline/offer for use |
23:05.27 | psokolovsky | but it's the same idea as crorgi_bl, as far as I can tell |
23:06.39 | cr2 | _bl is more of some addon. the pwm is handhled now by the device-specific driver itself. |
23:06.54 | cr2 | corgi_bl is calling set_intensity. that's it. |
23:07.26 | cr2 | and some zaurus-specific stuff, irrelevant on other devices. |
23:07.39 | psokolovsky | cr2, exactly, and that's what nice in it. I can't get what you have such that you can't do in that set_intensity() |
23:09.12 | cr2 | i can, but then the purpose on corgi_bl is not really clear to me, if i do almost everything in the device-specific driver myself. |
23:09.20 | goxboxlive | cr2: You made it, the lcd swith off now when closing lid :-) |
23:09.56 | cr2 | goxboxlive: great. but probably it should suspend on closing lid ? |
23:10.11 | goxboxlive | Now, that's a OPie setting |
23:10.16 | cr2 | ok. |
23:10.24 | psokolovsky | cr2, it's generic BL driver for PDA, that's its purposed. and it does handle some useful/important, but boring thing, saving you from that |
23:10.29 | goxboxlive | great work |
23:10.41 | cr2 | maybe you want to run sendmail/apache on universal :) |
23:11.17 | cr2 | psokolovsky: then it should be cleaned from zaurus-specific stuff. |
23:12.22 | cr2 | and be located in some other directory. |
23:12.33 | psokolovsky | cr2, yes, that was suggested to its maintainer, and in his q. so let's bother with cleaning up our stuff, and let him do his ;-) |
23:13.09 | psokolovsky | cr2, brr, but it's in drivers/video/backlight, what's wrong with that? |
23:14.10 | cr2 | really ? |
23:14.42 | psokolovsky | yep, where it should be? |
23:14.54 | cr2 | ah. ok. |
23:15.46 | cr2 | because all other corgi_* files are in arch/arm/mach-pxa |
23:16.09 | cr2 | which is, hmm, strange. |
23:17.14 | psokolovsky | cr2, they are savage people, they don't have drivers/soc/ in mainline ;-) |
23:17.28 | goxboxlive | Ok, back to the sound. In .12 the kernel bootup doesnt do this: It has only reset once, not twice as here. |
23:17.30 | goxboxlive | <5>snd_htcuniversal_audio_set_codec_reset: 1 |
23:17.31 | goxboxlive | <5>snd_htcuniversal_audio_set_codec_power: 1 |
23:17.31 | goxboxlive | <5>snd_htcuniversal_audio_set_codec_reset: 0 |
23:19.27 | goxboxlive | And man, there where a lot of debug message now during boot |
23:19.37 | cr2 | goxboxlive: we don't have reset. |
23:21.01 | goxboxlive | But why doesn it do this then? Why just we dont remove the reset stuff. The more taht there is in the code, the more confused we get. right? |
23:23.57 | cr2 | i don't know how to call the reset. at least it can't be observed in wince. |
23:24.10 | cr2 | and it works without it. |
23:24.37 | *** join/#htc-linux WizMaui (n=WizMaui@62.112.90.179) |
23:29.51 | goxboxlive | ok |
23:37.53 | Speedy2 | goxboxlive: Any luck? |
23:38.02 | Speedy2 | Can you write some code that sends the data directly over the I2S link? |
23:38.06 | Speedy2 | i.e. bypass alsa. |
23:38.37 | goxboxlive | no |
23:38.41 | Speedy2 | Why not? |
23:40.39 | goxboxlive | Speddy2 I am not a programer, so i dont have the skills to do that. But explain for cr2, maybe he has some good answers. ( i just test the codes, i am kamikaze user) :-) |
23:41.13 | Speedy2 | Ok. Well cr2, my idea is to write a small program that sends the sine wave data directly to the ak CODEC via it's I2S link, bypassing ALSA and everything else |
23:41.21 | goxboxlive | As you can see, i have a lot of typos, it wouldhave been a code that never had worked :-) |
23:41.22 | Speedy2 | That will allow one to figure out if the chip is initalized correctly, etc. |
23:41.58 | Speedy2 | One can just use a sinewave table and keep looping |
23:42.15 | Speedy2 | If I had hardware, I would personally do this |
23:44.38 | psokolovsky | Speedy2, complexity of that is much higher than diagnoze and fix problem in existing driver which is known to work under some conditions |
23:45.17 | psokolovsky | Speedy2, so, unless someone iis 2s/i2c/sound codec wizard, it's just waste of time |
23:45.43 | psokolovsky | Speedy2, on the other hand, *Linux* should provide such ability in easy way. alas, it doesn't |
23:49.28 | cr2 | psokolovsky: Kevin2 has spi.c tool in his svn. i2s is very similar. |
23:50.50 | psokolovsky | cr2, I bet it is ;-) but we have driver which works for otehr people and doesn't work for us, and we can't find what's wrong. do you think that writing something anew will be easier? |
23:51.03 | *** join/#htc-linux WizMaui (n=WizMaui@62.112.90.179) |
23:51.06 | cr2 | no. |
23:52.06 | psokolovsky | I of course would be happy to just do cat /dev/random >/dev/i2s ;-) |
23:55.30 | Speedy2 | I didn't realize the driver was known to work in other conditions. |
23:58.06 | psokolovsky | Speedy2, yes, bd2, hx4700 guy, got it running on 2.6.18 with flash boot. |
23:58.19 | Speedy2 | On an HTC or the hx4700? |
23:58.47 | psokolovsky | hx4700 |
23:59.16 | Speedy2 | Bizarre that it's not working here |
23:59.25 | Speedy2 | Let me check the data sheet |