00:08:07 | | Quit Kitar|st () |
00:12:31 | | Quit kugel (Remote host closed the connection) |
00:13:24 | | Part Boldfilter |
00:14:51 | * | n1s wonders why he gets a rm /path/to/apps/codecs/tta.o at the end of the build if any of the tta files changed |
00:17:32 | | Join Kitar|st [0] (Kitar_st@BSN-182-142-21.dial-up.dsl.siol.net) |
00:17:56 | | Quit dfkt (Ping timeout: 265 seconds) |
00:19:19 | | Join dfkt [0] (dfkt@unaffiliated/dfkt) |
00:21:13 | CIA-5 | New commit by nls (r26004): FS #11272 by Gerhard Zintel fixing a bug when resetting the database in keybox |
00:26:33 | | Quit mirak_ (Quit: Ex-Chat) |
00:26:44 | | Join Buschel [0] (~ab@p54A3F5E9.dip.t-dialin.net) |
00:26:51 | funman | n1s: i was seeing the same thing for wmapro before tta was added |
00:27:11 | n1s | ah |
00:36:33 | | Quit funman (Quit: free(random());) |
00:40:10 | Buschel | Can anyone with a Coldfire target check the effect of FS #11261 (before and after patch) on a mpc file? The test is quite simple: start to play a file and immediately seek several minutes into the file. |
00:42:47 | | Quit efyx (Remote host closed the connection) |
00:47:50 | | Quit bluebro () |
00:48:15 | n1s | Buschel: if you can wait till tomorrow morning, i can do it |
00:48:58 | Buschel | that's fine. I am also waiting for the response by the originator of the fs entry. |
00:57:10 | | Quit dfkt (Quit: -= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.) |
00:58:27 | | Join toffe82_ [0] (~chatzilla@12.169.218.14) |
00:59:00 | | Quit toffe82 (Ping timeout: 265 seconds) |
01:00 |
01:00:05 | | Quit halmi_ (Read error: Connection reset by peer) |
01:03:24 | | Quit n1s (Quit: Lämnar) |
01:04:55 | | Quit powell14ski (Ping timeout: 258 seconds) |
01:05:33 | | Join solexx_ [0] (~jrschulz@e176104040.adsl.alicedsl.de) |
01:07:40 | | Quit Buschel (Ping timeout: 260 seconds) |
01:09:08 | | Quit solexx (Ping timeout: 258 seconds) |
01:11:32 | | Join powell14ski [0] (~powell14s@c-24-9-7-198.hsd1.co.comcast.net) |
01:17:07 | | Quit petur (Quit: Leaving) |
01:26:56 | | Join piotrekm [0] (~piotrek@unaffiliated/piotrekm) |
01:44:07 | | Quit Curtman (Quit: Leaving) |
01:50:02 | *** | Saving seen data "./dancer.seen" |
02:00 |
02:18:17 | Unhelpful | saratoga: any code that's currently in control of the UI will call get_action to get the next action, also passing a context. contexts can inherit other contexts, so what a selected context actually contains in terms of button mappings can be a bit complicated |
02:18:51 | saratoga | i'm just trying to find the functions that link button presses to actual changes of state in the playback engine |
02:20:49 | Unhelpful | right, and that's where things get messy, becaue whatever has control of the UI at present decides how to interpret buttons, by selecting a context which maps button events to action, and then in code that acts on the received action. |
02:21:26 | | Join Rob2222 [0] (~Miranda@p4FDCA211.dip.t-dialin.net) |
02:22:14 | saratoga | yeah i can't follow the code |
02:22:23 | saratoga | do you happen to know what the functions are in playback? |
02:22:29 | saratoga | maybe i can work backwards |
02:22:41 | Unhelpful | it looks like menu.c has a hook for stop actions... apps/gui/wps.c handles a few actions as well |
02:23:56 | | Quit Rob2223 (Ping timeout: 240 seconds) |
02:24:06 | * | Unhelpful is a bit surprised that more playback actions aren't supported in the menu, actually |
02:25:19 | saratoga | gui_wps_show looks promising |
02:26:11 | Unhelpful | what exactly are you looking to do? if you're just going to manipulate playback state from inside code you probably want to look at the audio_* stuff in playback.c |
02:27:51 | | Quit S_a_i_n_t (Ping timeout: 240 seconds) |
02:29:24 | | Join S_a_i_n_t [0] (S_a_i_n_t@203.184.0.75) |
02:44:28 | | Part toffe82_ |
02:49:11 | | Quit liar (Ping timeout: 258 seconds) |
02:49:38 | | Join liar [0] (~liar@clnet-p09-185.ikbnet.co.at) |
02:56:58 | | Join Blue_Dude [0] (~chatzilla@rockbox/developer/Blue-Dude) |
02:57:08 | Blue_Dude | saratoga: What are you trying to do exactly? Maybe I can point you in the right direction... |
02:57:27 | | Quit piotrekm (Quit: piotrekm) |
03:00 |
03:04:07 | saratoga | Blue_Dude: curious about externally driving the playback engine from a script for automated testing |
03:05:21 | | Join ischeria1 [0] (~ischeriad@p5B0A0BE9.dip0.t-ipconnect.de) |
03:05:34 | Blue_Dude | That could be driven by calls to functions in audio.h. Should work for hwcodec and swcodec targets. |
03:06:26 | Blue_Dude | IIRC the plugin API exposes them also. |
03:09:01 | | Quit ischeriad (Ping timeout: 265 seconds) |
03:11:40 | saratoga | Blue_Dude: what is AUDIO_SRC_PLAYBACK? |
03:11:49 | saratoga | can rockbox actually record it's own output? |
03:15:03 | Blue_Dude | Looking for that reference... |
03:18:34 | | Join joeyg [0] (~apoelstra@S010600236999fec1.vs.shawcable.net) |
03:19:26 | Blue_Dude | Oh OK. It distinguishes what source gets output to the hardware: playback, radio, or line in. I have no idea if it means the recording hardware can tap it as its going out. |
03:20:34 | Blue_Dude | MIC IN is another possibility. |
03:21:04 | Blue_Dude | But that one may be recoding only. Depends on hardware. |
03:21:50 | saratoga | hmm ok |
03:21:50 | Blue_Dude | SPDIF IN... |
03:22:00 | saratoga | so i'd have to add another way to dump PCM to disk |
03:22:13 | saratoga | i wonder if kugel's work makes this any easier |
03:22:16 | Blue_Dude | I don't know if we can loopback to the recording hardware. |
03:22:46 | Blue_Dude | Well the sim already does this. Did you mean on the target? |
03:23:08 | saratoga | sim is fine |
03:23:23 | Blue_Dude | ./rockboxui −−debugaudio |
03:23:28 | saratoga | how does that work? |
03:23:55 | Blue_Dude | Dumps the PCM output to disk as it's sent to the PCM driver. |
03:24:15 | Blue_Dude | Makes a .RAW file. Essentially a wav file without the header. |
03:24:36 | saratoga | it does this at 1x speed or as fast as possible? |
03:25:15 | | Join bieber [0] (~bieber@162-78.97-97.tampabay.res.rr.com) |
03:26:21 | Blue_Dude | At 1X speed. If you want to dump it all ASAP, use test_codec. |
03:27:42 | saratoga | yeah but that doesn't use playback, so it doesn't test the actual playback code, only the codecs |
03:27:44 | Blue_Dude | Test_codec makes wav files, too, so you don't have to convert. OTOH, if you're testing the interface, −−debugaudio captures exactly what's being sent out. |
03:28:34 | Blue_Dude | test_codec uses the playback chain as far as dsp. It doesn't actually try to output anything through the playback engine itself. |
03:28:45 | saratoga | yeah i know i've worked a bunch on it |
03:28:54 | Blue_Dude | For that, use the sim. |
03:29:37 | Blue_Dude | Is there a problem with the playback engine? Or are you trying to benchmark it somehow? |
03:29:44 | saratoga | theres a lot of problems with it |
03:29:51 | saratoga | lots and lots of things can crash it |
03:29:54 | Blue_Dude | Well, yeah :) |
03:30:47 | saratoga | even simple things like repeat mode often don't work correctly, and then theres low memory targets |
03:31:27 | Blue_Dude | I don't like how the buffering context switches before the track is played out... |
03:32:00 | saratoga | isn't that for things like crossfade where its needed? |
03:32:59 | Blue_Dude | Nah. It just switches when the buffer is finished with it, not when it's actually played. |
03:33:12 | Blue_Dude | Crud. Got to go... Back in a while... |
03:50:06 | *** | Saving seen data "./dancer.seen" |
03:54:57 | Unhelpful | maybe for debugging purposes the sim could run the PCM output at some multiple of realtime? |
03:58:20 | | Join hebz0rl_ [0] (~hebz0rl@dslb-088-065-216-224.pools.arcor-ip.net) |
03:59:30 | | Quit hebz0rl (Read error: Operation timed out) |
04:00 |
04:10:14 | | Quit adnyxo (Quit: Leaving) |
04:15:16 | | Quit TheSeven (Ping timeout: 276 seconds) |
04:16:48 | | Join saratogalab [0] (~9803c20d@gateway/web/freenode/x-nnrklwjtnepbbkch) |
04:17:17 | Blue_Dude | Unhelpful: That's a possibility. But that would involve essentially artificially overclocking all of Rockbox, wouldn't it? I don't know. |
04:18:14 | saratogalab | i assume you could just change whatever clock SDL is using |
04:18:29 | | Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven) |
04:18:39 | Blue_Dude | Maybe just a modification of sound.c in rockboxui would serve. Instead of waiting for the SDL driver to request more samples, it could just bypass SDL and send output to disk via the debugaudio mechanism already there. |
04:19:13 | saratogalab | is the only limit on playback/buffering/codecs the rate at which the PCM buffer empties? |
04:19:22 | Blue_Dude | But it would have to be patient enough to recognize when it's starving for samples vs just running out of samples at the end of playlist. |
04:20:05 | Blue_Dude | Well, there's the kernal timer as well. |
04:21:43 | | Join pixelma_ [0] (quassel@rockbox/staff/pixelma) |
04:21:43 | | Quit pixelma (Disconnected by services) |
04:21:46 | | Quit amiconn (Disconnected by services) |
04:21:48 | | Join amiconn_ [0] (quassel@rockbox/developer/amiconn) |
04:22:02 | | Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma) |
04:22:08 | | Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn) |
04:22:24 | Blue_Dude | You can cut out the SDL code in sound.c and replace it with a delay function that clocks out samples at 2X, 4X or whatever. At some point you're probably going to starve a thread for data but I have no idea where it will break down first. |
04:23:17 | Blue_Dude | And the UI will be just plain weird. |
04:23:45 | saratogalab | well the idea was to drive it via a script and then compare the PCM output or check for deadlocks |
04:23:50 | saratogalab | the GUI isn't so important |
04:24:02 | saratogalab | detecting deadlocks will be interesting |
04:24:13 | Blue_Dude | I think it would be fun. |
04:26:29 | Blue_Dude | Or you could rewrite playback and buffering and save us all the trouble... |
04:26:46 | saratogalab | only if i can drop all features except playback and seeking |
04:27:47 | Blue_Dude | Sold! |
04:28:05 | saratogalab | eventually it would be really useful to have some simple automated testing built into the build system |
04:28:47 | saratogalab | queue a few tracks and just skip around them, seek, rebuffer, switch codecs, etc |
04:29:03 | saratogalab | that way if someone breaks something its clear |
04:31:26 | Blue_Dude | A script engine is considerably more than "playback and seeking". |
04:32:21 | saratogalab | it could be as simple as a list of commands that are fired into the action code every X ticks |
04:32:35 | saratogalab | i think the scripting is easier then hacking rockbox to do it all |
04:32:42 | Blue_Dude | Hm. |
04:33:29 | Blue_Dude | I wouldn't mind being able to seek across track boundaries. |
04:35:05 | * | Unhelpful sees problmes in changing the playback "speed" if you're trying to debug things that may be due to interactions between threads... :/ |
04:36:20 | saratogalab | threading isn't the same in the sim anyway, so it won't be like on target |
04:37:49 | Unhelpful | i thought we used locks to make the sim threads *seem* cooperative despite their being platform threads? |
04:39:03 | saratogalab | timing won't be the same though, so if theres threading bugs they won't be the same |
04:39:35 | saratogalab | playback bugs on the clip where nondeterministic on the device, so I assume timing was a factor |
04:40:26 | saratogalab | but pcs are so fast, and we have so many on the build system we could do each test 100 times a day and catch all the serious ones |
04:43:29 | Unhelpful | if we could send test scripts to the build system to run on sims ;) |
04:49:32 | | Quit arbingordon (Read error: Connection reset by peer) |
04:50:06 | | Join arbingordon [0] (~w@unaffiliated/arbingordon) |
04:50:20 | | Quit arbingordon (Read error: Connection reset by peer) |
04:54:16 | | Quit dys (Ping timeout: 276 seconds) |
04:54:19 | | Join Barahir [0] (~jonathan@frnk-590f419b.pool.mediaWays.net) |
04:54:30 | | Quit fdinel (Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org) |
04:54:44 | | Join dys [0] (~andreas@krlh-5f736c6a.pool.mediaWays.net) |
04:58:19 | | Quit Barahir_ (Ping timeout: 276 seconds) |
05:00 |
05:01:29 | | Quit Blue_Dude (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539]) |
05:31:34 | jhMikeS | funman: Don't know. The other drivers should already stop if the cb returns < 0 since that's how it's supposed to work. One thing's for sure, it needs to be sped up. It almost doesn't matter if it starts and stops recording at this point. |
05:50:08 | *** | Saving seen data "./dancer.seen" |
05:50:12 | | Join shai [0] (~Shai@l192-117-110-233.cable.actcom.net.il) |
05:51:49 | | Quit shai (Client Quit) |
05:53:32 | | Join shai [0] (~Shai@l192-117-110-233.cable.actcom.net.il) |
05:56:44 | | Join Forsaken_Boy [0] (~chatzilla@24.139.225.41) |
06:00 |
06:29:38 | | Join BHSPitMonkey [0] (~stephen@unaffiliated/bhspitmonkey) |
06:43:32 | | Quit Strife89 (Ping timeout: 264 seconds) |
06:43:49 | | Join FlynDice_ [0] (~FlynDice@c-24-19-225-90.hsd1.wa.comcast.net) |
06:44:55 | | Quit FlynDice (Ping timeout: 276 seconds) |
06:45:13 | | Join Strife89 [0] (~Strife89@adsl-80-145-33.mcn.bellsouth.net) |
06:50:30 | | Join Strife1989 [0] (~Strife89@adsl-67-49-42.mcn.bellsouth.net) |
06:50:59 | | Quit Strife89 (Disconnected by services) |
06:51:06 | | Nick Strife1989 is now known as Strife89 (~Strife89@adsl-67-49-42.mcn.bellsouth.net) |
06:51:51 | | Nick FlynDice_ is now known as FlynDice (~FlynDice@c-24-19-225-90.hsd1.wa.comcast.net) |
07:00 |
07:14:46 | | Join mikroflops_ [0] (~yogurt@90-227-45-110-no112.tbcn.telia.com) |
07:18:45 | | Quit mikroflops (Ping timeout: 260 seconds) |
07:21:47 | | Quit Horscht (Ping timeout: 246 seconds) |
07:25:21 | | Join Boldfilter [0] (~Boldfilte@adsl-82-77-225.jax.bellsouth.net) |
07:27:09 | | Join Horscht [0] (~Horscht2@xbmc/user/horscht) |
07:29:52 | | Quit sevard (Ping timeout: 258 seconds) |
07:31:20 | | Join animalstyle [0] (~animalsty@c-68-81-240-138.hsd1.pa.comcast.net) |
07:31:49 | | Quit animalstyle (Client Quit) |
07:38:40 | | Join esperegu [0] (~quassel@145.116.15.244) |
07:49:06 | | Quit antil33t () |
07:50:09 | *** | Saving seen data "./dancer.seen" |
07:50:13 | JdGordon | who knows rockbox recording code? |
07:54:21 | | Quit S_a_i_n_t (Ping timeout: 240 seconds) |
07:54:26 | | Join ender` [0] (krneki@foo.eternallybored.org) |
07:54:42 | | Join S_a_i_n_t [0] (S_a_i_n_t@203.184.2.142) |
07:57:58 | | Join sevard [0] (sev@216.164.6.24) |
08:00 |
08:19:14 | | Quit sevard (Ping timeout: 245 seconds) |
08:19:23 | amiconn | JdGordon: I know the hwcodec recording code. Does that help? |
08:20:27 | | Join sevard [0] (sev@216.164.6.24) |
08:21:56 | amiconn | Regarding tta - I think I know why the filter asm doesn't help much. Unlike the filters in libdemac and libffmpegflac, the libtta filter function doesn't loop over multiple samples, but is called once per sample |
08:22:44 | amiconn | The C version is 'static inline', using asm the way ucchan did un-inlines the filter function |
08:23:01 | * | amiconn should try to convert that to inline asm |
08:24:44 | | Quit anewuser (Quit: for SELL 2 by the price of 1 now!) |
08:25:44 | CIA-5 | New commit by jethead71 (r26005): pitch_detector: Use continuous recording, even if the algorithm is too slow for that and record a whole buffer even if the min frequency is higher. ... |
08:26:33 | JdGordon | amiconn: I'm looking for a good way to know if recording has taken the buffer or not? in_recording_screen() doesnt sound good enough... |
08:28:44 | jhMikeS | JdGordon: it takes in the init, using audio_get_buffer |
08:30:06 | jhMikeS | more specifically, pcmrec_init calls audio_get_recording_buffer |
08:31:15 | JdGordon | I found that, but It doesnt expliclty release it? |
08:31:21 | | Join DerPapst [0] (~Alexander@p4FE8F97C.dip.t-dialin.net) |
08:31:50 | JdGordon | and while it has it the buffering thread is still active and will cause big problems if anything calls bufopen() |
08:33:13 | JdGordon | this could also mean that me wanting to use the buffering.c code to store the radio art images is a bad idea? |
08:35:04 | JdGordon | http://www.rockbox.org/tracker/task/11263?getfile=21870 has my hack to work around this.. send an event in pcmrec_init saying get out of the buffer, and one in pcmrec_close() saying the buffer is safe again, but I'm not sure that is the best |
08:35:16 | jhMikeS | no it doesn't "release" it. it just stops using it and playback grabs it back later |
08:35:19 | JdGordon | and that is swcodec only right? |
08:35:29 | jhMikeS | yes |
08:35:35 | JdGordon | yeah, which doesnt sound very safe |
08:36:47 | JdGordon | could we change recording to actually open a handle requesting the whole buffer instead of just stealing it? that way any calls to bufopen() would fail because the buffer appears to be full... |
08:38:06 | jhMikeS | where the heck is rocklib_aux.c ? I don't have it in my tree. is it auto-generated? |
08:38:13 | | Quit DerPapst (Quit: Leaving.) |
08:39:09 | | Join Buschel [0] (~ab@p54A3E7DA.dip.t-dialin.net) |
08:41:07 | * | jhMikeS has no idea how to fix _that_ particular error nor why it happens |
08:41:41 | CIA-5 | New commit by jethead71 (r26006): Fix some sim warnings. |
08:45:29 | | Quit Buschel (Ping timeout: 245 seconds) |
08:45:49 | | Join stoffel [0] (~quassel@p57B4ADA5.dip.t-dialin.net) |
08:50:55 | jhMikeS | JdGordon: so long as recording gets what it needs, which is whatever is setup there |
08:52:28 | jhMikeS | it's audiobuf to audiobufend |
08:55:25 | | Nick aberet is now known as OJ_simpson (~tty0@70-57-133-13.albq.qwest.net) |
08:55:26 | jhMikeS | hmmm...it seems maybe that lua thingy can't properly handle "const unsigned long list[]" |
08:56:44 | | Join flydutch [0] (~flydutch@host138-162-dynamic.14-87-r.retail.telecomitalia.it) |
08:56:51 | jhMikeS | dilemma is: quick fix and change the function or fix the incomplete parser >:[ |
08:57:05 | | Nick OJ_simpson is now known as aberet (~tty0@70-57-133-13.albq.qwest.net) |
08:58:59 | | Join funman [0] (~fun@rockbox/developer/funman) |
09:00 |
09:00:46 | JdGordon | almost the same question... If i make it safer to use the buffer when audio is stopped, is that going to make people more likely to use it as a general purpose buffer (i.e malloc!) |
09:01:38 | * | jhMikeS wonders who "people" are |
09:05:12 | | Quit JdGordon (Ping timeout: 248 seconds) |
09:09:49 | funman | jhMikeS: http://pastie.org/959928 ? |
09:11:51 | jhMikeS | funman: thanks...hehe. I just did the same but I put it as the last one. any reason it should be up there? |
09:12:08 | funman | no i think it's unordered |
09:13:58 | | Join kramer3d [0] (~kramer@unaffiliated/kramer3d) |
09:15:26 | funman | pitch_detector doesn't recognize my voice :o |
09:16:03 | | Join n1s [0] (~n1s@rockbox/developer/n1s) |
09:17:05 | CIA-5 | New commit by jethead71 (r26007): Fix LUA red and yellow. |
09:17:37 | jhMikeS | it used to respond with "hello rafael"? |
09:17:57 | funman | yes, i just asked it to open the door and it refused ;) |
09:19:14 | jhMikeS | it's taking on my personality already :) |
09:20:16 | funman | just blowing air in the mic would display a tune but not now |
09:20:50 | jhMikeS | I did tune my guitar with it which went pretty well. |
09:21:34 | funman | ah it works with higher volume |
09:21:44 | | Join DerPapst [0] (~Alexander@dslb-088-069-146-108.pools.arcor-ip.net) |
09:22:00 | funman | (and with a tune played from the computer - i have no instrument) |
09:23:45 | | Join petur [0] (~petur@rockbox/developer/petur) |
09:24:27 | funman | ranma: can you dump I2SOUT registers from OF when playing mp3 ? |
09:25:44 | funman | ranma: and DMA registers too ? |
09:28:19 | | Quit kramer3d (Quit: Leaving) |
09:31:35 | | Quit Strife89 (Quit: Sleeeeeeeeeeep.) |
09:32:12 | n1s | funman, jhMikeS I guess FS #10635 can be closed now, seems fixed to me at least |
09:37:24 | | Join lpereira [0] (~lucien@did75-8-82-226-27-213.fbx.proxad.net) |
09:40:50 | funman | yep looks good |
09:43:45 | jhMikeS | n1s: the crashing? I suppose. |
09:44:17 | n1s | yeah, it doesn't crash for me anymore anyway, thanks for that btw :) |
09:47:11 | jhMikeS | just doing my job, sir. :) |
09:48:16 | n1s | so was it the fact that it didn't stop playback that made it crash or the nasty starting and stopping of recording? |
09:49:02 | jhMikeS | which targets was it again? I think funman fixed the fact dma wasn't stopping because it ignore the return value or something. |
09:49:21 | jhMikeS | my problem could've been something else with getting versions out of sync |
09:50:10 | n1s | it crashed on my c220 but not on h300 |
09:50:11 | *** | Saving seen data "./dancer.seen" |
09:50:18 | n1s | eh c240 |
09:50:32 | n1s | not the c240 doesn't crash anymore |
09:50:36 | n1s | now* |
09:51:56 | funman | the callback stopped dma, and drivers stopped dma again when the callback returned |
09:52:50 | n1s | double dtop -> crash? |
09:53:00 | n1s | s/dtop/stop/ |
09:54:36 | funman | looking at pcm-pp.c:pcm_rec_dma_stop() it shouldn't .. |
09:56:23 | funman | to get recording working on clipv1 i must set PCM_NUM_CHUNKS to 32 but then i have overflow warnings |
09:58:14 | | Quit saratogalab (Quit: Page closed) |
09:58:48 | * | n1s closed the bug now |
09:58:52 | | Quit Forsaken_Boy (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539]) |
09:59:00 | funman | not with mp3 enc though |
09:59:02 | n1s | only 258 to go |
10:00 |
10:03:03 | funman | is fs#11237 a real bug ? (should we mute playback at lowest volume when we show the muted icon?) |
10:03:24 | jhMikeS | the audio driver is supposed to do that |
10:04:24 | jhMikeS | funman: btw, you get overflow warnings #2? you set the chunk size to 2048? |
10:05:25 | funman | 2 (encoder) sometimes together with 1, i kept the chunk size to 8192 |
10:05:27 | jhMikeS | err, 4096 and make NUM_CHUNKS 512 |
10:05:56 | funman | not possible, 4096*512 == 8192*256 == 2M |
10:05:58 | CIA-5 | New commit by alle (r26008): 22 pixel high Nimbus-like font, containing only digits and the period; intended for use in FM screen for displaying frequency (FS #11271) |
10:06:14 | funman | we need to go with 512kB maximum |
10:06:50 | jhMikeS | funman: I thought the problem was just with the chunk being too big |
10:07:02 | funman | hm it's unrelated |
10:07:17 | funman | recording is disabled on clipv1 & pals because of small memory |
10:10:10 | jhMikeS | well, it does need that. however, anything flash based probably can use a different strategy and flush encoded data frequently |
10:10:11 | | Quit Zarggg (Read error: Connection reset by peer) |
10:10:35 | | Join Zarggg [0] (~zarggg@65-78-69-194.c3-0.eas-ubr6.atw-eas.pa.cable.rcn.com) |
10:11:26 | jhMikeS | 2M = 11.89s pcm. I'm sure that could go way down on a flash target as well |
10:13:13 | jhMikeS | probably not < 1s though if you're getting overflows |
10:13:18 | funman | hm as3514 mutes before teh lowest step |
10:15:47 | pixelma | I think volume isn't completely muted on my c200 at the lowest setting (could have changed though, not sure at the moment) |
10:16:23 | funman | pixelma: shouldn't have changed |
10:16:55 | pixelma | that's a v1 in case that wasn't clear |
10:17:15 | funman | yep, they use the same code |
10:17:33 | | Quit aberet (Quit: Leaving.) |
10:19:56 | | Quit liar (Quit: Verlassend) |
10:21:15 | funman | r13447 mentions an 'audio improvement' for as3514, any idea what it is ? |
10:21:40 | | Join kugel [0] (~kugel@rockbox/developer/kugel) |
10:23:11 | * | n1s doesn't like bugs that happen sometimes... |
10:27:23 | S_a_i_n_t | As opposed to bugs that happen all the time? |
10:27:31 | S_a_i_n_t | You like those better? ;) |
10:30:24 | n1s | yes! |
10:30:35 | jhMikeS | bugs that happen all the time are easier to squash |
10:33:09 | | Quit TheSeven (Ping timeout: 268 seconds) |
10:33:22 | n1s | jhMikeS: does the beast charge if you power it down with the charger connected? |
10:35:51 | CIA-5 | New commit by funman (r26009): as3525: change SDRAM settings only in models for which there is a bootloader ... |
10:35:55 | CIA-5 | New commit by funman (r26010): as3514: mute headphones at the lowest volume ... |
10:36:21 | funman | 257 to go :D |
10:36:41 | n1s | \o/ |
10:37:11 | jhMikeS | n1s: nope. it's controlled by software. I was considering a mode where it finishes that and then powers down when complete. |
10:38:21 | n1s | ah, yeah, or just prevent shutdown like some other targets do when charger is connected |
10:39:15 | jhMikeS | HAVE_POWEROFF_WHILE_CHARGING? I think that's the define. It could be removed. but really, it should finish the job and shut off by itself. |
10:39:47 | jhMikeS | for now, removing it sounds reasonable |
10:44:08 | CIA-5 | New commit by jethead71 (r26011): Cache align the pitch detector audiobuffer where needed. Some other misc. changes to try to make sure everything builds. |
10:49:00 | | Quit mt (Ping timeout: 240 seconds) |
10:51:34 | | Join mt [0] (~mtee@rockbox/developer/mt) |
10:52:55 | CIA-5 | New commit by nls (r26012): Disable shutting down with the charger connected on the gigabeat s since it doesn't charge while powered off. Ideally we should implement a mode where ... |
10:55:01 | | Join DataGhost [0] (~dataghost@17-18-ftth.onsnetstudenten.nl) |
10:55:01 | | Quit DataGhost (Changing host) |
10:55:01 | | Join DataGhost [0] (~dataghost@unaffiliated/dataghost) |
11:00 |
11:00:20 | n1s | gah, found another bug |
11:01:46 | n1s | abd it's one of those that seems so unlikely to trigger that it might have been there for years... |
11:02:08 | * | n1s tests other targets |
11:02:17 | jhMikeS | which is what? |
11:02:58 | n1s | playing a flac and then starting a mpc while it's still playing -> whitenoise, any other transition with mpc, flac or other codecs doesn't seem to do it... |
11:05:13 | | Join fml [0] (~chatzilla@p5DD2C34D.dip.t-dialin.net) |
11:05:52 | fml | Hello. How do I activate an .fms file? Just 'play' it in the browser? No matter where it is? |
11:06:23 | kugel | fml: I didn't understand your last commit |
11:06:34 | kugel | what's the point of a font which contains only numbers? |
11:06:36 | Llorean | n1s: I seem to remember someone complaining that seeking in the current version of MPC is pretty bad (just slow) as well. It would be worth checking to see if that was introduced when MPC was updated to handle the new sv? |
11:07:27 | fml | kugel: I thought the commit message did explain it. What do you not understand? |
11:07:28 | n1s | Llorean: yeah, i was going to test Buschel's patch to fix the slowness when i wanted to see how slow it was before |
11:07:56 | jhMikeS | n1s: isn't that the same deal with the ascodec sansa's as the beast? |
11:08:00 | n1s | good news is that his patch fixed the slowness, i'll try to go back in history to see if i can spot anything |
11:08:00 | Llorean | n1s: I meant the flac/mpc crash might be around that time too? It seems like something that might've been noticed. |
11:08:44 | n1s | jhMikeS: eh, what's the same deal? :) |
11:08:46 | Llorean | fml: Is there an indication for users that the font is unusable even as a basic font? |
11:08:49 | kugel | fml: I mean, a font takes several KB in ram no matter of how many glyphs it has got |
11:08:55 | Llorean | I'm not sure we should be including fonts that can't even render the menus in English. |
11:09:16 | funman | i can record on clipv1 and i only get warnings (encoder buffer overflow) when recording PCM at 88.2kHz or 96kHz |
11:09:29 | jhMikeS | n1s: with the charging |
11:10:07 | fml | Llorean: good point! |
11:10:07 | n1s | hmm, the flac->mpc whitenoise thing happens on h300 but not the beast... endianness messup??? (but how can the codecs influence eachother like this) |
11:10:09 | kugel | funman: these freqs don't make sense to use anyway, do they? |
11:10:24 | n1s | jhMikeS: no idea |
11:10:37 | fml | kugel: but if you want to use it −− even if only for the frequency −−you take it into account |
11:10:46 | * | jhMikeS things those are ludicrous frequencies |
11:10:55 | funman | kugel: well you never know .. |
11:11:03 | kugel | fml: no, why? |
11:11:13 | jhMikeS | if you're recording thing for bats or dolphins, sure it makes sense |
11:11:35 | funman | btw we could even go higher than that ! |
11:11:39 | fml | kugel: because you *want* large digits (if you want them) |
11:12:01 | funman | 3MHz on as3525v1 |
11:12:05 | kugel | yes, I understand that, but why limit the font to only numbers? |
11:12:15 | n1s | jhMikeS: i can shut down the c200 when charging from usb so it's probably the same deal |
11:12:19 | fml | Llorean: I might add basic latin letters later |
11:12:51 | fml | kugel: because I needed just that now and had no time to design all other glyphs |
11:13:36 | fml | kugel: am I right that I can play an .fms file and it will be set as FMS? |
11:13:46 | kugel | I would think so |
11:14:21 | kugel | the commitdiff suggests it's possible |
11:14:37 | pixelma | the path to the bitmaps needs to be correct though I guess |
11:16:08 | funman | as3525v2 records microphone on both channels unlike as3525v1 |
11:16:46 | | Join komputes [0] (~komputes@ubuntu/member/komputes) |
11:18:05 | * | jhMikeS wonders if he should IRAM-up the pitch plugin (or more accurately commit that) |
11:19:29 | jhMikeS | seems to get H120 going alot better |
11:19:39 | n1s | sounds like a good idea |
11:19:43 | | Join pamaury [0] (~pamaury@rockbox/developer/pamaury) |
11:21:55 | Llorean | jhMikeS: Plugin? |
11:22:39 | n1s | pitch_detector |
11:25:17 | | Part fml |
11:25:57 | Llorean | Ah |
11:26:55 | | Quit BHSPitMonkey (Read error: Connection reset by peer) |
11:34:57 | | Join dfkt [0] (dfkt@unaffiliated/dfkt) |
11:39:51 | n1s | hmm, it happens on ape -> mpc too |
11:40:06 | pixelma | Unhelpful: maybe you know - is the album art size limited to the display size in a way that e.g. llets me only load square album art to the shortest dimension of the display size (e.g. on a 160x128 display a square album art is only allowed to be 128x128)? |
11:40:41 | n1s | and alac... |
11:41:31 | Unhelpful | pixelma: shouldn't be - loading 160x128 album art is valid, but a square album art will be scaled to 128x128 to maintain aspect ratio - you'd need to have wide album art to actually fill that. |
11:41:40 | n1s | and tta... |
11:41:53 | n1s | so lossless -> mpc == whitenoise ??? |
11:42:11 | funman | happens on sim ? |
11:42:20 | n1s | nope, seems cf only |
11:42:25 | pixelma | Unhelpful: the latter was my question- am I not allowed to load it 160x160 and let a viewport do the clipping? |
11:42:38 | * | n1s needs to drink coffe before testing more |
11:43:12 | Unhelpful | erm, so you'd want to load a 160x160 image unscaled and have it clipped to fit? i don't think that can be done at present |
11:43:31 | Llorean | n1s: And it only happens when playback is interrupted in one by another, and not on "next track" style playlist transitions? |
11:44:48 | pixelma | Unhelpful: well doesn't necessarily need to be a 160x160 image but a square one at least so I wouldn't need two bitmaps |
11:45:03 | | Quit kugel (Ping timeout: 245 seconds) |
11:45:18 | Unhelpful | pixelma: i'm not really quite sure i understand what you *want* it to do? |
11:46:14 | | Quit esperegu (Read error: Connection reset by peer) |
11:46:29 | | Join kugel [0] (~kugel@rockbox/developer/kugel) |
11:47:39 | pixelma | have the album art use the full display (width on a landscape screen) without looking stretched or having to prepare a second set of bitmaps. I wouldn't mind if there are missing parts |
11:48:25 | funman | jhMikeS: pcm_record.c : why not trusting storage_spinup_time() in pcmrec_refresh_watermarks() ? |
11:50:03 | | Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow) |
11:50:13 | *** | Saving seen data "./dancer.seen" |
11:52:34 | jhMikeS | funman: trusting? |
11:53:18 | funman | if (spinup_time == 0) spinup_time = 3.5s |
11:53:38 | | Join Schmogel [0] (~Miranda@p3EE22845.dip0.t-ipconnect.de) |
11:54:08 | jhMikeS | reasonable default if it hasn't been set |
11:54:17 | funman | only ata sets it non 0 |
11:54:41 | funman | perhaps you were only thinking of ata then? |
11:54:42 | jhMikeS | yeah, but iirc, it wasn't always set when it needed to be |
11:54:46 | funman | ok |
11:55:22 | jhMikeS | yes, ata was in fact the primary concern, and petur will kill you if you mess up his tapings :) |
11:55:34 | funman | ^^ |
11:55:49 | funman | well if i change this i will ask testing |
11:56:19 | * | petur loads his gun |
11:56:24 | amiconn | Spinup time is only set properly after the first spinup that happens under control of the main ata driver |
11:56:46 | | Join MethoS- [0] (~clemens@134.102.106.250) |
11:57:05 | amiconn | The 3.5 seconds are in fact ooooold. I already tjought about lowering it for flash targets at some point |
11:57:57 | | Quit Schmogel (Ping timeout: 245 seconds) |
11:59:00 | jhMikeS | do we even need it for flash targets? |
11:59:31 | amiconn | n1s: My guess is that mpc doesn't set the necessary emac mode |
11:59:38 | funman | no |
12:00 |
12:00:08 | amiconn | jhMikeS: Well, we do need some kind of watermark. Setting spinup time to something low is probably easiest |
12:00:30 | | Quit MethoS- (Remote host closed the connection) |
12:02:15 | jhMikeS | amiconn: recording doesn't accept very low numbers, but that's all based on a drive. it probably also needs to know if the volume with the destination directory is flash or not. if flash only, then compile different code. |
12:02:27 | AlexP | Why are we adding fonts with only numbers? |
12:02:55 | AlexP | If it is wanted for the FM screen I would have thought including it in a theme bundle would make more sense |
12:03:31 | Llorean | AlexP: I basically agree. |
12:04:16 | jhMikeS | pcmrec would probably be simpler if it had a buffering thread *and* a control thread. right now it weird stuff so it can be responsive to input. |
12:04:19 | Llorean | If a font is to be visible in "browse fonts" it should be able to handle the menus in the default language at a bare minimum. |
12:04:23 | AlexP | yes |
12:04:39 | AlexP | perhaps a quick mail to the dev list is in order |
12:04:42 | | Join watto [0] (~watto@193.203.81.165) |
12:05:06 | Llorean | AlexP: While numerical fonts make sense (for a variety of purposes) there needs to be a situation in place to separate them first (as well as, possibly, a way to load them setting aside much less RAM) |
12:06:23 | Llorean | Though there should also probably be a deifnition of what they would include both as a minimum and maximum (digits, colon, comma, period, as a possible example) |
12:07:15 | jhMikeS | pcmrec could have much of the core functionality taken from the UI and run by the pcmrec thread instead, making the screen a tack-on piece rather than required. |
12:07:23 | AlexP | I agree they make sense, in fact I have done a couple myself for exactly this - I have 40 and 50 point ones for big screen FM |
12:07:36 | AlexP | I just wouldn't add them to the font pack |
12:08:44 | Llorean | Definitely not in the font pack we have rbutil download for users. |
12:08:54 | funman | jhMikeS: what's about 'not accepting low numbers' ? i'm playing with very low numbers here |
12:09:21 | funman | 23 encoder chunks |
12:09:52 | jhMikeS | funman: see how the code rejects ridiculous spinup times (for a hard disk, starting line 762)? |
12:11:37 | * | kugel apparently needs another bigish commit :( |
12:12:32 | | Join tscherny [0] (~stefan@f053014038.adsl.alicedsl.de) |
12:15:30 | | Join Jaykay [0] (~chatzilla@p5DC578C3.dip.t-dialin.net) |
12:18:48 | kugel | would anyone object to renaming our math.h? |
12:18:53 | funman | i don't get what PANIC_SECONDS is |
12:19:24 | funman | why is it below FLUSH_SECONDS ? |
12:21:59 | gevaerts | kugel: why? |
12:22:51 | kugel | the sim uses a little trick to get the system's math.h for code in uisimulator/. that trick doesn't work with the target tree and so it includes our math.h |
12:22:52 | funman | ah it just boosts priority when approaching the high watermark |
12:23:04 | | Quit stoffel (Ping timeout: 246 seconds) |
12:23:18 | kugel | but our math.h doesn't declare any functions, but only constants which are not even part of iso c |
12:25:02 | jhMikeS | funman: The panic time is if things are getting too full and it must boost thread priority over other things like UI to get the data out |
12:25:42 | gevaerts | kugel: traditional unix math.h has the functions, so I think that our math.h really is a stripped down "standard" math.h |
12:26:14 | funman | jhMikeS: it does 'panic' even before the buffer is ready for flushing though |
12:26:34 | kugel | gevaerts: ours hasn't got a single function |
12:26:43 | gevaerts | so? |
12:26:53 | gevaerts | kugel: isn't the proper solution to move it to libc/include? |
12:27:05 | kugel | some code in uisimulator to be moved into the target tree wants the functions |
12:27:14 | jhMikeS | funman: well, with ATA, if you got under PANIC_SECONDS remaining until full, it panics. if you change buffer sizes, those values have to refect it |
12:27:48 | kugel | well, the sim doesn't use that mechanism yet so it includes libc/include as well |
12:28:00 | funman | hm aren't the watermarks in available recorded chunks, rather than in available free chunks? |
12:28:07 | gevaerts | ok, then *that*'s the real problem I think |
12:29:19 | funman | hm i must have things backwards |
12:29:19 | kugel | actually I asked before committing whether the sim should already use the mechanism to exclude libc/ but you were opposed to that |
12:29:19 | jhMikeS | funman: they're convered to that from remaining time to recorded chunks |
12:29:33 | n1s | Llorean: nope, happens on regular transitions too |
12:29:33 | jhMikeS | or are they...(been a long time) |
12:29:40 | n1s | amiconn: ah, that sounds plausible |
12:30:22 | jhMikeS | funman: yeah, enc_num_chunks - <some calculated value>. |
12:30:36 | funman | http://pastie.org/960074 |
12:30:45 | gevaerts | kugel: I wasn't opposed to switching, I just thought it a bad idea to do too much at the same time. It seems that now might be a good time :) |
12:31:04 | kugel | ah, now I understand, makes sense :) |
12:31:57 | jhMikeS | funman: about the only ata target it regularly panics is x5 recording WavPack at 88kHz. |
12:33:05 | | Join bmbl [0] (~Miranda@vpn224-113.fh-bielefeld.de) |
12:33:05 | | Quit bmbl (Changing host) |
12:33:05 | | Join bmbl [0] (~Miranda@unaffiliated/bmbl) |
12:33:05 | funman | i have put the spinup time adjustement under #if STORAGE_ATA, the rest isn't modified |
12:33:05 | funman | yeah i'm testing wavpack@96kHz for worst case |
12:34:02 | n1s | amiconn: indeed, mpc doesn't set up the emac mode |
12:34:07 | jhMikeS | "define LOW_SECONDS .1" <−− ?? |
12:34:27 | funman | jhMikeS: the whole buffer is 1.23s @ 96kHz |
12:37:00 | * | jhMikeS was just commenting on the floaty thing (these could be in 1/10s form or something) |
12:37:27 | funman | we convert to ticks with *HZ so it's ok |
12:37:31 | tscherny | hello everybody, i would like to edit the wiki |
12:37:42 | funman | i don't know if it's legal C but it works in gcc so.. |
12:37:44 | AlexP | What is your wiki name? |
12:37:50 | funman | (as long as HZ is builtin constant of course) |
12:37:53 | tscherny | StefanWanja |
12:38:09 | | Join stoffel [0] (~quassel@p57B4ADA5.dip.t-dialin.net) |
12:38:11 | | Quit archivator (Ping timeout: 240 seconds) |
12:38:52 | jhMikeS | funman: yeah, but does it up-convert to floats or turn HZ*flow into an int first? |
12:39:01 | jhMikeS | HZ*float |
12:39:15 | funman | latter |
12:39:31 | funman | let me check again |
12:39:58 | jhMikeS | line 737, 4*.1 will give 0, no? |
12:40:12 | | Join liar [0] (~liar@clnet-p09-185.ikbnet.co.at) |
12:40:21 | funman | hm good catch |
12:40:32 | AlexP | tscherny: OK, should be fine |
12:40:47 | funman | ah it's why i used .25 first :o) |
12:41:52 | jhMikeS | could do PREC*CONST*x / (PREC*y), for other PREC can be 1, for this PREC can be 10 |
12:42:36 | funman | or make PREC 1000 and use milliseconds? |
12:43:01 | funman | (CONST/1000) would be integer |
12:43:13 | funman | oops, and it would be the 0 integer :o) |
12:43:18 | jhMikeS | if you want, but I was just trying to avoid the extra instruction for targets than can use plain seconds |
12:46:25 | funman | 4*.25 could an integer but i don't know how to write that cleanly |
12:46:47 | n1s | amiconn: setting the emac mode fixed it, thanks |
12:48:02 | funman | 4*1/4 |
12:48:28 | CIA-5 | New commit by nls (r26013): Restor setting of EMAC mode on coldfire lost in r25056 fixes whitenoise when transitioning from a codec using a different EMAC mode. |
12:48:31 | funman | #define LOW_SECONDS 1/4 -> low_watermark = (sample_rate*4*LOW_SECONDS + .... |
12:48:44 | jhMikeS | 4/4 |
12:48:55 | funman | ? |
12:49:03 | jhMikeS | :_) |
12:49:04 | n1s | hmm, cook is not setting the emac mode either... |
12:49:24 | funman | jhMikeS: oh, was that a bool ? ;) |
12:49:26 | n1s | mt: where do you think such setup should go in the cook codec? |
12:49:27 | * | jhMikeS just want to be careful not to get overflows in any of that too |
12:51:01 | jhMikeS | typedef bool oney |
12:51:43 | funman | http://pastie.org/960074 |
12:52:25 | funman | i could add a comment which says fractions must be integer fractions of 4 |
12:53:07 | | Quit bmbl (Quit: Bye!) |
12:54:18 | | Quit kugel (Ping timeout: 264 seconds) |
12:56:27 | | Quit komputes (Ping timeout: 260 seconds) |
12:56:50 | funman | it removes spinup adjustements for existing flash targets though, should that require testing? |
12:56:55 | jhMikeS | that would change all of mathematics |
12:57:12 | funman | ? |
12:57:26 | funman | the spinup time change? |
12:58:00 | | Quit liar (Quit: Verlassend) |
12:58:24 | jhMikeS | if fractions, generally speaking, must be :) Oh, you meant these paticular ones. sorry. :) |
12:58:30 | funman | yep |
12:58:59 | funman | i added "fractions must be integer fractions of 4 because they are evaluated with X*4*XXX_SECONDS, that way we avoid float calculation" for that particular case (MEM <= 2) |
12:59:17 | funman | ok for commit? i tested on fuze, spinup_time change didn't cause bad effects as expected.. |
12:59:43 | | Join swilde [0] (~wilde@aktaia.intevation.org) |
13:00 |
13:00:19 | | Join kugel [0] (~kugel@rockbox/developer/kugel) |
13:00:53 | jhMikeS | probably should mention that the order of the constants can't be rearranged either |
13:00:59 | funman | the only targets with flash & recording are sansas, ondiofm and iaudio7 |
13:03:03 | tscherny | thanks |
13:03:25 | funman | pixelma: could you test http://pastie.org/960098 on ondiofm please? |
13:04:03 | pixelma | not for the next 6 hours at least |
13:04:18 | funman | should I put it on flyspray ? |
13:05:00 | funman | or commit and fix bugs if they happen |
13:05:03 | | Nick fxb__ is now known as fxb (~felixbrun@h1252615.stratoserver.net) |
13:05:24 | jhMikeS | guess it's fine to commit on the grounds that it doesn't change code for the ata targets. |
13:05:40 | funman | let's do that |
13:07:59 | * | n1s gets the trusty old straightened paperclip |
13:08:19 | CIA-5 | New commit by funman (r26014): Enable recording on clipv1/m200v4/c200v2 ... |
13:08:58 | CIA-5 | New commit by uchida (r26015): fix: The second playback time is not correct in playing TTA musics continuously. |
13:09:01 | CIA-5 | New commit by nls (r26016): Set up EMAC mode in cook, fixing whitenois when transitioning from a codec using different emac mode.' |
13:12:36 | jhMikeS | n1s: the old "forgot to set macsr" trick? |
13:13:16 | n1s | jhMikeS: yep, or in mpc, "removed the call setting macsr" :) |
13:13:46 | funman | ranma: how's c200v2 FM, we need to remove this annoying red entry on SansaAMS ! |
13:18:05 | tscherny | hey guys, does anyone of you have the sources for the sigmatel STMP36XX platform? |
13:18:27 | funman | jhMikeS: i can copy the mono channel into the muted one, but i don't see it on the peakmeter since it has to happen at the end of the DMA transfer :/ |
13:18:46 | | Quit hebz0rl_ (Quit: Ex-Chat) |
13:18:47 | funman | tscherny: nothing on the wiki ? |
13:19:00 | tscherny | well there are links, but the go to sigmatel.com |
13:19:04 | tscherny | which is not there anymore |
13:20:04 | jhMikeS | funman: for e200, I just did the copy from the ISR if the source was MIC, but it's not DMA so it's done during each FIFO emptying. |
13:20:28 | n1s | sigh, playing the a52/ac3 file in the test_files after anything else hardlocks, but not only on cf |
13:20:50 | funman | yes i've seen that, but now i removed ISR based copying ;) |
13:21:38 | jhMikeS | funman: where is it now? |
13:21:51 | funman | well now it uses dma |
13:21:59 | jhMikeS | e200 v1? |
13:22:18 | funman | sorry |
13:22:33 | funman | i've seen how it is done on e200v1 but i removed ISR copying from e200v2 (and others) code |
13:22:51 | funman | on clipv2/clip+/fuzev2 both microphone channels work though |
13:23:07 | funman | so the trick is only needed on clipv1 & friends |
13:24:37 | CIA-5 | New commit by uchida (r26017): updates the README document. |
13:26:27 | | Join ucchan [0] (~ucchan@softbank126102048044.bbtec.net) |
13:32:51 | | Join dfkt_ [0] (dfkt@unaffiliated/dfkt) |
13:33:13 | amiconn | funman: Doesn't the ADC have a setting to do the copying, i.e. route the same input to both right & left channel adc? |
13:33:32 | funman | not that i've seen |
13:33:45 | amiconn | Hmm, odd |
13:33:46 | funman | and afaik jhMikeS added this code after not finding such a setting |
13:33:57 | funman | well you have the datasheet? |
13:35:09 | | Quit pamaury (Ping timeout: 258 seconds) |
13:36:05 | | Join pamaury [0] (~c2c7a50a@rockbox/developer/pamaury) |
13:36:21 | amiconn | Depends. Do you mean AS3514? |
13:36:29 | funman | i don't know if as3514 one is public but the as3525 one is and it has the same data |
13:36:52 | | Quit dfkt (Ping timeout: 252 seconds) |
13:37:49 | | Join dfkt [0] (dfkt@unaffiliated/dfkt) |
13:40:44 | jhMikeS | on as3514, I don't recall seeing any sort of thing to tie the left and right together |
13:42:02 | funman | hm should wavpack playback be realtime ? |
13:42:13 | | Quit dfkt_ (Ping timeout: 265 seconds) |
13:42:17 | n1s | yeah |
13:42:22 | | Quit dfkt (Ping timeout: 252 seconds) |
13:43:00 | | Join dfkt [0] (~dfkt@unaffiliated/dfkt) |
13:44:36 | funman | samplerate is too low after running pitch_detector |
13:45:11 | amiconn | hmm |
13:46:01 | * | amiconn doesn't like those ascodecs because they're rather crippled |
13:47:49 | funman | codecs don't use pcm_set_frequency() ? |
13:48:22 | jhMikeS | hmmm...the changer is supposed to reset it. should probably add that in. |
13:48:41 | funman | i see all plugins reset to HW_SAMPR_DEFAULT at exit (but not pitch_detector) |
13:49:02 | amiconn | Codecs cannot do that. They're decoding asynchronously |
13:49:04 | funman | perhaps best to remove the reset from plugins and put it at codec loading? |
13:49:44 | amiconn | The pcm playback engine is what would need to change pcm frequency, if we ever sample rate changes during playback |
13:49:51 | jhMikeS | funman: yeah...woops. shoulda noticed that |
13:49:56 | amiconn | if we ever *support* |
13:50:16 | *** | Saving seen data "./dancer.seen" |
13:50:29 | funman | asm() foo : if i don't use asm volatile () but specify "memory" as clobber, shouldn't that prevent the asm() from being optimized away? |
13:50:30 | n1s | how hard would it be to support different sample rates? |
13:50:38 | jhMikeS | n1s: sucky |
13:50:52 | funman | i give it pointers as argument, the pointers don't change but the content does |
13:51:00 | n1s | ah so "not worth it" |
13:51:02 | amiconn | Well, for pcm integration on hwcodec we would need that. |
13:51:43 | | Join komputes [0] (~komputes@ubuntu/member/komputes) |
13:51:43 | amiconn | sample rate change cannot be gapless, of course |
13:51:49 | funman | http://pastie.org/960130 < channel copy for as3525 : the peak meter of the 2nd channel is a bit off but recording works |
13:52:03 | jhMikeS | n1s: well, the DSP needs reconfiguring, there would have to be silence to conceal hardware glitching in the switch |
13:52:29 | funman | hm i guess i could keep track of pcm_rec_dma_get_peak_buffer() calls and modify samples already written |
13:52:52 | n1s | ah, so it's the switching that's sucky, not the play-at-different-SR? |
13:53:49 | | Join adnyxo [0] (~aaron@adsl-065-013-002-216.sip.asm.bellsouth.net) |
13:54:00 | jhMikeS | n1s: has to switch at *exactly* the right moment too, so instructions on the buffer itself are needed |
13:54:52 | n1s | right, i can imagine it becomming quite complicated... |
13:55:04 | jhMikeS | no crossfade either |
13:55:18 | n1s | indeed |
13:56:02 | jhMikeS | of course the change means voice would have a break in it if it's playing at the time |
13:56:05 | mt | n1s: Just got back .. I guess anywhere before initializing the decoder ? |
13:56:28 | n1s | mt: heh, already committed it, found an init function that seemed the logical place |
13:56:47 | mt | :) alright |
13:56:53 | n1s | jhMikeS: ah, yes that too |
13:57:14 | jhMikeS | e200v1's mic + backlight on is way too noisy for tuner not to get confuser. :\ |
13:57:22 | jhMikeS | *confused |
13:57:25 | amiconn | On hwcodec it won't be that bad, because there is only one possible source at any time anyway |
13:57:26 | * | n1s is getting convinced that switching samplerate brings too much complication compared to the gains |
13:57:39 | amiconn | Switching mp3 -> pcm will take ~1 second though |
13:59:10 | | Quit komputes (Ping timeout: 240 seconds) |
14:00 |
14:03:46 | pixelma | Unhelpful: so what I wanted is not possible at the moment? Found that out by trying and wondered if it was by design and if there was a way around that (except preparing album art bmps at different sizes) |
14:04:39 | Unhelpful | right now, no, it doesn't work like that. it should scale to the largest size that fits the viewport while maintaining aspect ratio |
14:04:43 | jhMikeS | personally, I'd rather an option to run it at say, 48K all the time instead |
14:05:39 | amiconn | Well, on hwcodec there is no other choice than switching rate (or even codec) if we want to integrate pcm playback |
14:07:31 | | Quit ucchan (Quit: Leaving...) |
14:08:36 | jhMikeS | amiconn: I can imagine. All the concern is in the DSP and PCM output anyway, which isn't HWCODEC at all. |
14:09:44 | amiconn | No, but the pcm packets come from the playback thread afaiu, and integrating pcm for hwcodec means playback engine unification, which in turn means the engine has to learn to tag the packets it outputs |
14:10:09 | amiconn | And once it's doing that for hwcodec, it could be extended to swcodec |
14:12:05 | jhMikeS | it does "tag" on the pcmbuf descriptors to get track transitions right. it calls a callback |
14:13:57 | amiconn | It already does that? |
14:17:13 | jhMikeS | yeah, if one is specified for the just-finished buffer in order to let the WPS know when to change over |
14:18:05 | amiconn | Ah, hmm, so that mechanism just needs to be extended |
14:18:28 | pixelma | Unhelpful: but the principle works in general - in my c200 SBS I load the album art at the same size as in the WPS and then put it in a viewport to crop it to a strip (as a statusbar kind of thing) to not get a "jumpy" look and don't get too small. I just hit the limit of smallest screen dimension which I find a pity |
14:19:32 | amiconn | For hwcodec we'd need a flag to indicate whether it's mpeg audio or pcm, and if it is pcm, whether it's 8 or 16 bit, little endian or big endian (only for 16 bit, obviously), mono or stereo, and which of the 9 possible sample rates |
14:19:54 | | Join Luca_S [0] (~5d3fc54b@giant.haxx.se) |
14:21:24 | amiconn | Whenever those flags change compared to the previous buffer, the mas needs to be reconfigured |
14:25:08 | funman | peakmeter is absolutely identical for both channels on sansa e200v1/c200v1 ? |
14:25:19 | jhMikeS | on swcodec it's implement in the pcmbuf. I suppose it would need to be elsewhere for hwcodec. |
14:25:44 | kugel | gevaerts: ok, so also exclude the .c files or only the headers for now? |
14:26:40 | gevaerts | kugel: I'd keep headers and c files together, so everything |
14:27:51 | kugel | I'm fine with that, but that means the sim cannot detect anymore when these get broken, but I think that's very unlikely to happen |
14:29:10 | | Quit DataGhost (Ping timeout: 240 seconds) |
14:29:45 | pixelma | funman: looked to me this way when recording with my c200 through the microphone a while ago |
14:33:02 | funman | thanks |
14:33:46 | tscherny | how can I contact a wiki user? |
14:33:54 | tscherny | i can't see his email |
14:34:05 | Torne | you don't, if they haven't provided any contact details |
14:35:09 | CIA-5 | New commit by jethead71 (r26018): Give pitch_detector the IRAMming it deserves. |
14:37:06 | CIA-5 | New commit by kugel (r26019): Include host system's header files and don't compile our c library replacements in the sim. ... |
14:43:46 | CIA-5 | New commit by kugel (r26020): Move math.h to firmware/libc/include/ and fix slight incompatibilities between our and the host's math.h |
14:44:59 | | Join anewuser [0] (anewuser@unaffiliated/anewuser) |
14:47:26 | kugel | hm |
14:48:02 | * | jhMikeS wonders if this autocorrelation stuff can be done in less than n^2 time :\ |
14:48:19 | | Join Forsaken_Boy [0] (~chatzilla@24.139.225.41) |
14:48:49 | CIA-5 | New commit by alle (r26021): Revert r26008. Fonts in the font pack should at least be able to display English menus. This font can maybe included in a theme bundle. |
14:52:18 | CIA-5 | New commit by kugel (r26022): Fix wrong #ifdefs. |
14:57:26 | amiconn | kugel: Imo the sim *should* use our code wherever possible |
14:57:41 | amiconn | For apps/ and firmware/ code that is, of course |
14:58:18 | | Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven) |
14:59:00 | kugel | amiconn: but with raaa that's not really possible everywhere |
14:59:49 | amiconn | Yes, understood. But then raaa and the sim should be handled differently |
14:59:58 | CIA-5 | New commit by funman (r26023): as3525v1: duplicate the right channel into the left channel when recording microphone |
15:00 |
15:01:02 | funman | now the only problem is silent right headphone when doing passthrough |
15:01:05 | kugel | amiconn: that doesn't make sense for code which never changes in reality |
15:01:28 | kugel | I understand that principle but it's really not needed for this c library code |
15:02:21 | amiconn | Well, the sim started out as a pure ui simulator, but nowadays it's actually more like a testbed for everything that doesn't absolutely need the hardware for testing |
15:02:48 | amiconn | Hence it should use as much of our code as possible, so that we can debug/ optimize/ whatever that code |
15:02:55 | kugel | this code doesn't need testing anymore |
15:03:37 | amiconn | Other code might also behave differently depending on the c lib stuff in use |
15:05:41 | kugel | then the code is wrong |
15:07:13 | pamaury | amiconn: I don't see your point, the c lib is the c lib, it has a well define semantics that doesn't depend on the implementation |
15:07:21 | | Part tscherny |
15:07:30 | pamaury | Furthermore, the host libc is better optimized for the platform |
15:10:55 | kugel | right |
15:13:36 | funman | why do we stop max volume at 6dB on sansas? we could make it even higher |
15:13:51 | funman | s/sansas/as3514/ |
15:16:01 | funman | the volume steps and limits seem to be the same on as3543 but some people complain it isn't as high as the OF |
15:16:31 | funman | someone wants to breaks her ears and compare maximum volume of OF & rockbox on clipv1 & clipv2 ? =) |
15:17:14 | | Join evilnick_B [0] (~0c140464@rockbox/staff/evilnick) |
15:19:14 | ThomasAH | funman: at ±0dB the clip+ is quite fine when connected to aux-in of our kitchen radio ... but in bed -45 to -40dB are more suitable. Today I found fixed.cfg in the manual, this helps alot to not break my ears when going to bed after listening in the kitchen :) |
15:19:19 | amiconn | pamaury: Our c lib isn't a full one, it is cut down (on purpose) |
15:19:26 | pixelma | I only know that the c200 is noticably quieter than my other targets. It's hard to compare with the OF as that one has only 15 volume steps or so and I remember finding the smallest too loud already |
15:19:55 | mt | aaand .. sound ! |
15:20:04 | amiconn | And the sim doesn't need to be optimized, it should rather be as close to the target it simulates as possible. Raaa is of course different and should use the host libs |
15:20:06 | mt | (although there's still too much nois) |
15:20:08 | | Join komputes [0] (~komputes@ubuntu/member/komputes) |
15:20:10 | mt | *noise |
15:20:17 | n1s | mt: yay! |
15:20:22 | mt | :) |
15:21:02 | pamaury | amiconn: I know this but having fiddle with it, I know that most of the code is a direct copy of [insert here the naked the glibc-lib we use] thus the majority of the code was not written by rockbox |
15:21:42 | CIA-5 | New commit by kugel (r26024): Fix remaining reds/yellows. |
15:21:47 | amiconn | Our *printf() implementation is definitely different |
15:22:25 | | Quit rhodan (Read error: Connection reset by peer) |
15:22:29 | funman | pixelma: indeed we can go much lower than the OF, but i guess we must set a limit for max volume to not become deaf (but higher than european limits though :p ) |
15:22:43 | pamaury | it's just simpler, it doesn't support all modifiers and then ? |
15:22:48 | n1s | amiconn: we already used host snprintf on linux, because our snprintf would break alsa |
15:22:58 | | Join rhodan [0] (~Quassel@2001:1608:12:2::38) |
15:24:00 | pixelma | funman: if I compare line level to other targets that was - I remember having to turn up the volume on a car sterea when using the c200 compared to using a different target at line level |
15:24:10 | funman | n1s: what's the link between snprintf & alsa ? |
15:24:34 | funman | pixelma: c200 has line out on the dock connector ? |
15:24:48 | n1s | funman: dunno, but i know sound wasn't working with our snprintf |
15:25:21 | kugel | mt just experienced that the other day |
15:25:33 | funman | audiohw_set_lineout_vol() use less steps and doesn't amplify with the DAC |
15:26:33 | pixelma | funman: no (well could be but I don't have dock), I use and compared headphone out |
15:26:44 | kugel | mt: your math.h problem should be fixed now |
15:27:23 | funman | ah ok, the as3514.c 'lineout' doesn't count there: it's the FM volume |
15:29:22 | | Join JohannesSM64 [0] (~johannes@cm-84.215.116.196.getinternet.no) |
15:37:44 | | Quit joeyg (Quit: lions and tigers and bears, oh my!) |
15:41:36 | CIA-5 | New commit by funman (r26025): builds: make clipv2/clip+/fuzev2 unstable |
15:41:56 | | Join liar [0] (~liar@clnet-p09-185.ikbnet.co.at) |
15:44:59 | CIA-5 | New commit by funman (r26026): front page: put clipv2/clip+ and fuzev2 in unstable category |
15:45:24 | | Join bmbl [0] (~Miranda@unaffiliated/bmbl) |
15:45:39 | funman | is there a special post commit hook on www/ svn? |
15:48:05 | mt | kugel: You experienced noise in a codec ? |
15:48:17 | kugel | no |
15:48:20 | mt | kugel: And thanks for fixing those math.h problems :) |
15:48:40 | mt | kugel: Ah sorry, misread the line |
15:49:05 | | Join Blue_Dude [0] (~chatzilla@rockbox/developer/Blue-Dude) |
15:50:20 | *** | Saving seen data "./dancer.seen" |
15:52:29 | AlexP | funman: No, a Swede needs to update it |
15:53:10 | funman | that's what i thought, but my svn ci still hasn't returned ;) |
15:53:30 | kugel | gevaerts: I think I'm gonna submit the pth thread work after the target tree stuff |
15:53:55 | funman | no swedes around :o |
15:54:08 | kugel | funman: your svn ci? |
15:54:16 | kugel | what about git? |
15:54:16 | funman | yes the svn command |
15:54:28 | funman | i have a separate svn checkout for www/ |
15:54:47 | kugel | traitor! ;) |
15:54:55 | funman | didn't bother with git |
15:55:36 | * | funman wonders where he will put his second git tattoo |
15:56:23 | | Quit funman (Quit: free(random());) |
16:00 |
16:02:33 | | Join towolf [0] (~towolf@f051154085.adsl.alicedsl.de) |
16:04:47 | towolf | salve, kudos for the great work on sansa lately. i'm wondering now, is there a way for permanently delete the MUSIC, PORT, AUDIBLE, etc, folders that the OF recreates at each boot? |
16:04:58 | towolf | if not, can i hide them? |
16:16:16 | | Join joeyg [0] (~apoelstra@S010600236999fec1.vs.shawcable.net) |
16:18:18 | | Quit Forsaken_Boy (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539]) |
16:18:41 | | Quit mt (Ping timeout: 240 seconds) |
16:23:15 | | Join Strife89 [0] (~Strife89@adsl-67-49-42.mcn.bellsouth.net) |
16:26:18 | pixelma | towolf: aren't those hidden? What's your "show files" setting set to? |
16:31:29 | | Quit JohannesSM64 (Ping timeout: 264 seconds) |
16:34:42 | | Join towolf_ [0] (~towolf@e176237096.adsl.alicedsl.de) |
16:34:51 | towolf_ | pixelma, not sure. got to check. what should it be set to? |
16:36:37 | pixelma | if you don't want to see hidden files, it should be "supported" not "all" |
16:37:11 | towolf_ | pixelma, thanks. |
16:37:17 | FlynDice | towolf_: I don't think there's a way to permanently delete them. You can delete them in rockbox I think but as you've found out the OF recreates them when it boots up. |
16:38:10 | | Quit towolf (Ping timeout: 276 seconds) |
16:39:04 | towolf_ | FlynDice, but there are default audio clips inside. where does it store them? on the sd, in a hidden partiton? there doesn't seem to be one. |
16:40:30 | | Quit anewuser (Quit: for SELL 2 by the price of 1 now!) |
16:41:50 | FlynDice | they are in ##MUSIC# which is visible if you have show files set to all, the directory is visible but the files are not. You can see them if you connect in MTP mode, in linux at least. |
16:41:58 | pixelma | if it's similar to the c200v1 the default tracks were uploaded in MTP mode and I could find them in some system/mtpcontent subfolder with a different file extension too (which is why the Rockbox database wouldn't be able to find them). I don't know about the newer Sansas though |
16:44:30 | | Join JohannesSM64 [0] (~johannes@cm-84.215.75.42.getinternet.no) |
16:47:54 | | Quit mc2739 (Ping timeout: 260 seconds) |
16:48:05 | kugel | gevaerts: almost finished |
16:50:14 | | Quit JohannesSM64 (Ping timeout: 248 seconds) |
16:59:29 | | Join S_a_i_n_t_ [0] (S_a_i_n_t@203.184.0.71) |
17:00 |
17:00:59 | | Quit S_a_i_n_t (Ping timeout: 246 seconds) |
17:03:00 | | Join Buschel [0] (~ab@p54A3BB80.dip.t-dialin.net) |
17:03:19 | | Quit towolf_ (Remote host closed the connection) |
17:05:21 | | Join JohannesSM64 [0] (~johannes@cm-84.215.75.42.getinternet.no) |
17:06:04 | | Join toffe82 [0] (~chatzilla@12.169.218.14) |
17:09:40 | | Quit Buschel (Ping timeout: 240 seconds) |
17:10:48 | | Quit petur (Quit: beertime) |
17:11:24 | | Join esperegu [0] (~quassel@145.116.15.244) |
17:16:42 | | Quit komputes (Ping timeout: 246 seconds) |
17:20:18 | | Quit Llorean (Read error: Connection reset by peer) |
17:20:48 | | Quit Luca_S (Quit: CGI:IRC) |
17:21:00 | | Join komputes [0] (~komputes@ubuntu/member/komputes) |
17:25:06 | | Quit lpereira (Quit: Leaving.) |
17:36:13 | kugel | new patch at FS #11234 |
17:40:59 | | Join Llorean [0] (~DarkkOne@rockbox/user/Llorean) |
17:43:55 | | Quit liar (Quit: Verlassend) |
17:45:28 | | Join Schmogel [0] (~Miranda@p3EE2230E.dip0.t-ipconnect.de) |
17:45:42 | | Join funman [0] (~fun@rockbox/developer/funman) |
17:46:55 | | Join mt [0] (~mtee@rockbox/developer/mt) |
17:48:01 | | Quit Blue_Dude (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539]) |
17:50:22 | *** | Saving seen data "./dancer.seen" |
17:59:56 | | Quit komputes (Ping timeout: 264 seconds) |
18:00 |
18:02:09 | | Join arbingordon [0] (~w@unaffiliated/arbingordon) |
18:09:43 | | Quit ender` (Quit: The reason they call it the American Dream is because you have to be asleep to believe it. -- George Carlin) |
18:15:10 | | Quit TheSeven (Ping timeout: 240 seconds) |
18:29:23 | | Quit jfc^3 (Quit: I wonder what this button over here does..) |
18:29:46 | | Join jfc^2 [0] (~john@dpc6682208002.direcpc.com) |
18:31:10 | | Quit S_a_i_n_t_ (Ping timeout: 260 seconds) |
18:36:30 | funman | http://pastie.org/960542 -> which one looks better ? |
18:36:45 | kugel | gevaerts: may I get your attention on RaaA for a second? I finished the target tree work (the first step mentioned in the mail) |
18:37:01 | funman | they produce the same code |
18:37:51 | | Quit toffe82 (Read error: Connection reset by peer) |
18:38:21 | gevaerts | kugel: yes, I saw that |
18:39:05 | pamaury | Anyone here who has some flash knowledge (I mean low level like device driver) ? |
18:39:45 | Torne | what kind of flash? |
18:40:37 | pamaury | That's the problem, I don't know but I have the code that deals with it (well the assembly one ^^). I wonder if there is some kind of uniformity among flash chip commands |
18:40:49 | pamaury | I only have a datasheet that *might* apply |
18:41:02 | mt | funman: I like the second one more |
18:41:16 | Torne | there is a reasonable level of consistency, yes |
18:41:30 | Torne | what's it from? |
18:41:45 | pamaury | the code ? |
18:42:50 | kugel | funman: I like neither :) |
18:43:15 | funman | kugel: ah, what would you do then? |
18:43:17 | kugel | but the second shows more obviously that we cannot make gcc do what we want |
18:43:29 | kugel | the first is just strange ;) |
18:44:00 | Torne | pamaury: well, the chip, but yeah, whichever ;) |
18:44:50 | | Quit joeyg (Quit: lions and tigers and bears, oh my!) |
18:45:46 | | Join fml [0] (~d9e3d2bb@giant.haxx.se) |
18:45:52 | mt | funman: I like this more than the previous two http://pastie.org/960552 :) |
18:46:28 | | Join preglow [0] (thomj@tvilling2.pvv.ntnu.no) |
18:46:32 | pamaury | Torne: I don't which chip it is, that's from a Creative Muvo V100 but the flash chip is under the lcd screen and it's nearly impossible to desolder it so I can't read the model. I have good reasons to think it might be samsung one. The datasheet that could match is "samsung k9w8G08U1M" but there a command which doesn't match in the code, it sends $85 $10 but also $85 $11 at some point and only the first is ok :( |
18:46:44 | mt | (just some indentation nothing important) |
18:46:57 | * | fml invites everybody to look at and comment on FS #11273 (Format FM frequency depending on the regional settings) |
18:48:02 | CIA-5 | New commit by funman (r26027): as3525: add some comments in the microphone channel copy loop ... |
18:48:51 | | Join CGL [0] (~CGL@190.207.156.230) |
18:50:55 | pamaury | fml: I don't know that part of the code so the patch is a little bit hard to follow but on the whole I'm not against getting rid of this extra 0 :) |
18:51:54 | | Quit swilde (Remote host closed the connection) |
18:52:21 | funman | someone with soldering skills is coming to devcon ? |
18:52:27 | funman | skills, and hardware |
18:52:57 | | Join zak1111 [0] (~zak@adsl-71-147-38-80.dsl.emhril.sbcglobal.net) |
18:53:32 | gevaerts | funman: hardware shouldn't be a problem. Skills probably depend on what you need |
18:53:38 | Torne | pamaury: are you running code on this? |
18:53:54 | amiconn | funman: Depends on the required skill level |
18:55:17 | funman | solder 8 wires on the 8 pads at the top, left to the lcd cable http://www.rockbox.org/wiki/pub/Main/SansaClip/sansa_clipplus_3.jpg (on the left corner of the white square) |
18:55:21 | pamaury | Torne: yes and no :) Theorically I could because I broke the encryption scheme (well it's so trivial that it took me an hour) and I an assembler and a compiler (more than 15 years old :() and I know how to upload code. I haven't tried to upload custom yet because I have no way to have feedback of any sort yet |
18:55:31 | Torne | pamaury: ah, right |
18:55:54 | Torne | i was gonna say, NAND chips generally identify themselvs if you ask :0 |
18:56:18 | pamaury | The architecture is so weird in fact, it's a stmp3505 running a dsp56k core, a 24-bit arch... |
18:56:45 | | Join liar [0] (~liar@clnet-p09-185.ikbnet.co.at) |
18:57:02 | Torne | does it look roughly like the ONFI spec? :) |
18:57:09 | pamaury | Torne: the chip as a jtag equivalent called once but I don't have hardware for it, and openocd does not support it, I yet don't know what the lcd is not I know how do program it and it does not even have a led ! |
18:57:35 | * | pamaury have a look at onfi |
18:57:54 | Torne | not all NAND chips follow it exactly |
18:58:02 | Torne | but it was defined based ona reasonable crosssection of common practise |
18:58:09 | Torne | so generally it's close enough |
18:58:10 | pamaury | Torne: spec is backward compatible or not ? (which version I pick ?) |
18:58:17 | Torne | just look at the latest one |
18:58:28 | amiconn | funman: Shouldn't be hard, provided my estimated pin spacing is approximately correct |
18:58:36 | pamaury | I think I found, the second command might have to do which two plane nand... |
18:59:40 | funman | amiconn: ranma did it already but i don't know with which hardware (magnifying glass?) |
19:00 |
19:00:02 | pamaury | Torne: from what I saw until now, the onfi spec might match, it defines the 85 11 commands I'm looking for, I need to have a look more closely |
19:00:16 | Torne | well, generally it should match |
19:00:19 | amiconn | At this spacing, a soldering iron with sufficiently fine tip |
19:00:38 | amiconn | Magnifying glass is helpful but probably not required |
19:01:32 | amiconn | It looks like 1.5mm or more (judging by the 6-pin SMDs which are probably 1.27mm |
19:08:14 | pamaury | Torne: what is Copy-Back "interleaved" ? The ONFI spec is not so easy to read |
19:08:25 | | Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven) |
19:08:48 | Torne | pamaury: described in the next chapter, no? |
19:08:55 | | Quit TheSeven (Remote host closed the connection) |
19:08:56 | Torne | all the interleaved access stuff is together |
19:09:35 | pamaury | ah true, I didn't saw it |
19:10:58 | * | pamaury doesn't want to read the whole in order to understand the commands but fears he'll have to |
19:11:05 | saratoga | jhMikeS: autocorrelation can be done in n log n time FWIW |
19:11:40 | saratoga | since its basically a weird kind of convolution, and convolutions can be implemented as FFTs |
19:14:48 | funman | is it useful for pitch detection? |
19:15:03 | Torne | pamaury: the alternative is to grab the datasheet for random other NAND flash parts |
19:15:18 | Torne | pamaury: since they all mostly follow the same interface spec the datasheets are kinda.. overlap-y |
19:15:57 | pamaury | anyway I don't need to understand everything, I just need to have a good idea of what each function do give it a name :) |
19:16:14 | pamaury | And I suspect there some unused pieces of code in this software |
19:17:15 | | Quit fml (Quit: CGI:IRC) |
19:17:48 | saratoga | i don't know how pitch detection works so I'm not sure |
19:20:13 | saratoga | but you could use the fft in the codec lib to compute the autocorrelation via the Wiener–Khinchin theorem, basically FFT the sequence, multiply it by its own complex conjugate, then inverse FFT it |
19:20:27 | | Join shodan45 [0] (~Administr@66.0.111.2) |
19:21:04 | saratoga | on PP and running it on a 44.1khz stereo stream would take about 20 MHz IIRC |
19:22:06 | funman | saratoga: we don't really need to run the as3525v2 at 240MHz, what worst case codec/profile should I try? |
19:22:39 | funman | is there anything but APE that the PP can't decode @80MHz ? |
19:23:38 | funman | SoundCodecs is all green |
19:23:46 | saratoga | AAC-HE |
19:23:52 | shodan45 | my sansa v1 w/ rockbox 3.5 seems to run out of battery when RB says it's around 45% - anyone else see this? |
19:24:03 | saratoga | needs 145MHz on e200v1 |
19:24:13 | saratoga | (but could be easily improved) |
19:24:25 | * | pamaury grumbles about a flash command that is not in the onfi spec :( |
19:24:42 | saratoga | though AAC-He is probably a lot faster on ARM9E |
19:25:00 | funman | 64kaache.m4a ? |
19:25:27 | saratoga | yeah |
19:25:29 | saratoga | wow only 99.27MHz |
19:25:36 | saratoga | arm9e rocks |
19:26:40 | saratoga | heh verses 150.92MHz on the fuzev1 |
19:27:03 | | Part zak1111 ("Leaving") |
19:27:05 | | Quit powell14ski (Read error: Connection reset by peer) |
19:27:18 | | Join powell14ski [0] (~powell14s@c-24-9-7-198.hsd1.co.comcast.net) |
19:27:18 | | Join S_a_i_n_t [0] (S_a_i_n_t@203.184.0.240) |
19:27:25 | | Join mikroflops [0] (~yogurt@90-227-45-110-no112.tbcn.telia.com) |
19:27:40 | funman | impressive |
19:28:21 | funman | i guess we could clock them at 120MHz then |
19:28:31 | saratoga | what does that do to APE? |
19:29:09 | | Join kramer3d [0] (~kramer@unaffiliated/kramer3d) |
19:29:16 | funman | i don't think i ever tried |
19:29:24 | funman | can ape files be cut with dd ? |
19:29:31 | kugel | don't we want max. performance for some demanding plugins? |
19:29:46 | amiconn | Uhm, no variable clock on as3525v2? |
19:29:50 | funman | i think we don't care about plugins, more about battery |
19:29:57 | kugel | i do care |
19:30:13 | saratoga | the fuze (and more so clip) have amazing CPU to screen size ratios |
19:30:24 | funman | amiconn: buggy -> disabled |
19:30:31 | funman | kugel: the same plugins run on e200v1 |
19:30:44 | kugel | yes, but worse |
19:30:48 | * | amiconn thinks this needs fixing, then |
19:31:03 | kugel | e.g. since I play bubbles on the fuze I find it unplayable on the e200v1 |
19:31:07 | funman | amiconn: easy to say |
19:31:33 | funman | kugel: what are teh most demanding plugins? |
19:31:39 | | Join Buschel [0] (~ab@p54A3F1BB.dip.t-dialin.net) |
19:31:41 | | Quit mikroflops_ (Ping timeout: 260 seconds) |
19:32:25 | kugel | I guess doom, rockbox, mpegplayer |
19:32:30 | kugel | maybe pictureflow too |
19:32:43 | | Quit Strife89 (Quit: Leaving.) |
19:32:48 | kugel | rockboy* |
19:32:55 | amiconn | Also pacbox |
19:32:59 | amiconn | and zxbox |
19:33:03 | saratoga | rockboy is probably the only one that actually needs higher clocks on the fuzev2 |
19:33:16 | funman | saratoga: needs 160MHz on clip+ |
19:33:24 | kugel | I'm not a fan of trying to hide the freq switching problem by clocking down the max. clock |
19:33:53 | saratoga | we should just add a max clock speed option to the plugin lib |
19:33:58 | pamaury | Torne: the flash seems to have quite a few commands that are not in onfi, do you know a good reference for that where I can find the model or the datasheet for example ? |
19:34:06 | saratoga | and let plugins request higher boost if they want it |
19:34:13 | kugel | and I think that once that works the advantage from going from 240MHz to 120MHz is entirely vanished |
19:34:16 | funman | it's not hiding i think, this doesn't work at all and we want correct battery life |
19:34:43 | | Join huelk [0] (~huelk@dtmd-4db79145.pool.mediaWays.net) |
19:34:44 | funman | i'm not even sure the OF changes its cpu clock |
19:34:44 | | Quit huelk (Client Quit) |
19:34:53 | kugel | why should there be no way to make it work? the OF can do it, can't it? |
19:35:06 | kugel | what's the OFs default clock? |
19:35:11 | funman | saratoga: (64kaache.m4a) i believe because pclk is lower on clip |
19:35:14 | amiconn | funman: I didn't say it's easy... but it would be The Right Thing to do, imo |
19:35:19 | funman | kugel: dunno, dunno |
19:35:20 | | Quit kramer3d (Ping timeout: 240 seconds) |
19:35:43 | saratoga | so the clip+ is slower then the fuzev2? |
19:35:45 | amiconn | It took quite a while until it worked on pretty much every new architecture |
19:35:47 | funman | amiconn: i don't even know if it's doable, i can't explain what happens when we try |
19:35:57 | | Join huelk [0] (~huelk@dtmd-4db79145.pool.mediaWays.net) |
19:36:09 | funman | saratoga: pclk is 24MHz, and 40MHz on the fuze because of bigger color display |
19:36:24 | saratoga | hmm that sucks |
19:36:32 | funman | it's ok when boosting though |
19:36:57 | funman | so yeah just let things that way |
19:37:40 | * | amiconn remembers hunting the PP5002 crashes related to cpu sleep |
19:38:04 | tmzt | anybody still working on zune seen this: http://zunedevwiki.org/wiki/development/firmware_dump |
19:38:12 | amiconn | Oh, and also unlocking higher clocks than 66MHz |
19:38:22 | funman | if i can plug a jtag adaptor on clip+ i can look in OF the conditions where it changes the clock (if it does it only at boot, and if not what are the input & output clocks of the function) |
19:38:22 | amiconn | (something ipl still doesn't do) |
19:38:24 | saratoga | the "work at >66Mhz bit" |
19:39:17 | saratoga | funman: seems I was wrong, my clipv2 gets 161.69MHz on aache |
19:39:50 | funman | ah i thought you tried on fuze |
19:40:08 | saratoga | this reminds me, does anyone know why the D2 doesn't seem to use IRAM? |
19:40:08 | funman | anyway the amsv2 work pretty well already |
19:40:35 | funman | when FlynDice fixes microSDs, pamaury write the USB driver, and bertrik fixes the charging bug, everything will be perfect O:-) |
19:40:46 | saratoga | that result confuses me though, i can't imagine how anything could be slower then on the fuzev1 |
19:41:40 | funman | saratoga: try http://www.pastie.org/960614 |
19:41:58 | funman | dunno if 72MHz would be too much |
19:42:39 | saratoga | will try on the clipv2 |
19:42:56 | saratoga | thats just the DRAM and not the IRAM? |
19:43:19 | funman | that's DRAM, PCLK, and 'NORMAL' cpufreq when boosting is activated |
19:43:42 | | Quit Xerion (Quit: ) |
19:43:52 | funman | i guess iram just uses pclk too |
19:44:16 | shodan45 | can someone confirm my battery problem? or is my battery dieing? |
19:44:32 | funman | shodan45: what does the OF say about the battery ? |
19:46:32 | saratoga | funman: its a lot faster |
19:46:51 | saratoga | 114.8MHz needed for aache instead of 161 from before |
19:47:22 | shodan45 | funman, seems about the same right now |
19:47:58 | FlynDice | funman: I thought SD was working fine now, is there something I missed? |
19:48:08 | shodan45 | guess that means it's the battery huh? ;( |
19:48:12 | funman | FlynDice: i have some problems with µSD |
19:48:38 | FlynDice | still? I wasn't aware. What's happening? |
19:49:07 | funman | sometimes works fine, sometimes can't initialize, some times just lock the screen |
19:49:25 | funman | didn't look in details, i have plenty of space on the internal storage ;) |
19:49:32 | FlynDice | all targets? |
19:49:36 | funman | yeah |
19:49:43 | funman | not the clipv2 though |
19:49:57 | FlynDice | ... |
19:50:07 | funman | ;D |
19:50:23 | *** | Saving seen data "./dancer.seen" |
19:51:46 | saratoga | funman: so this patch increases the DRAM/IRAM speed when boosted? |
19:52:53 | funman | saratoga: nope boosting is disabled in SVN and anyway pclk/dram/iram speeds are fixed (we change the divider when changing fclk==CPU speed) |
19:53:18 | saratoga | ah ok |
19:53:34 | funman | on AMSv1 pclk and fclk are divided from the same clock (PLL) but on AMSv2 only fclk is divided from this clock, and pclk is divided from fclk |
19:53:57 | saratoga | so this increases it from 24 to 40MHz? |
19:54:21 | funman | 48 |
19:54:30 | funman | we can do only integer fractions of 240MHz (cpu speed) |
19:54:59 | funman | though we could also do 5/8 6/8 or 7/8 of the PLL (240MHz) for the CPU speed |
19:55:06 | | Join stripwax [0] (~Miranda@87-194-34-169.bethere.co.uk) |
19:55:22 | | Quit stripwax (Client Quit) |
19:56:27 | funman | FlynDice: if number of blocks is correctly detected i think we should put an error in sd_transfer*() if we try to transfer too high blocks (after the card capacity) |
19:56:49 | | Quit JohannesSM64 (Ping timeout: 240 seconds) |
19:56:56 | funman | on AMSv1 we don't know the capacity for internal storage but i guess if ranma gets usb working we will quickly find |
19:57:22 | FlynDice | funman: You mean more than the card capacity? |
19:57:48 | funman | if the start block is higher than the capacity |
19:58:20 | FlynDice | Ok, I get that |
19:58:45 | funman | it seems like number of blocks is set correctly for Sansas PP although they use bank switching too |
19:59:29 | FlynDice | how do they figure it out? |
19:59:37 | funman | CSD |
19:59:58 | funman | on amsv1 the CSD reports 979.75MB even on my fuze4gb |
20:00 |
20:02:09 | | Quit S_a_i_n_t () |
20:02:30 | | Join Zagor_ [0] (~bjst@81-235-141-52-no63.tbcn.telia.com) |
20:02:50 | FlynDice | bbl |
20:03:17 | | Quit Llorean (Quit: Leaving.) |
20:07:26 | | Join ender` [0] (krneki@foo.eternallybored.org) |
20:09:10 | | Join S_a_i_n_t [0] (S_a_i_n_t@203.184.0.222) |
20:14:06 | | Quit ender` (Quit: To get as fewest unhappy people as possible, always bully the same ones.) |
20:15:53 | funman | FlynDice: http://pastie.org/960663 < asking the csd again after bank switching gives correct number on fuze |
20:17:39 | | Join Xerion [0] (~xerion@82-170-197-160.ip.telfort.nl) |
20:18:26 | funman | somebody with a 8GB sansa amsv1 ? |
20:18:50 | | Join DerPapst1 [0] (~Alexander@p5099d40e.dip0.t-ipconnect.de) |
20:19:05 | | Quit DerPapst (Ping timeout: 246 seconds) |
20:20:01 | | Join Llorean [0] (~DarkkOne@rockbox/user/Llorean) |
20:20:46 | | Part Llorean |
20:20:49 | funman | (works on clipv1 2GB too) |
20:20:56 | | Join JohannesSM64 [0] (~johannes@cm-84.215.116.196.getinternet.no) |
20:30:08 | | Quit kugel (Remote host closed the connection) |
20:33:06 | | Join ender` [0] (krneki@foo.eternallybored.org) |
20:33:06 | | Join B4gder [0] (~daniel@rockbox/developer/bagder) |
20:35:01 | | Join MethoS- [0] (~clemens@134.102.106.250) |
20:39:31 | | Quit Schmogel (Read error: Connection reset by peer) |
20:45:10 | | Join toffe82 [0] (~chatzilla@12.169.218.14) |
21:00 |
21:00:21 | | Join Strife89DS [0] (~nds@207.144.56.176) |
21:02:33 | | Join phanboy4 [0] (~benji@168.12.157.150) |
21:04:17 | | Join Schmogel [0] (~Miranda@p3EE2230E.dip0.t-ipconnect.de) |
21:08:39 | funman | 1 |
21:10:25 | | Quit stoffel (Remote host closed the connection) |
21:12:01 | | Part watto |
21:17:14 | | Quit DerPapst1 (Quit: Leaving.) |
21:21:38 | | Quit pamaury (Ping timeout: 252 seconds) |
21:27:24 | | Quit Buschel (Ping timeout: 268 seconds) |
21:37:12 | CIA-5 | New commit by funman (r26028): as3525v2: check if SD transfers fit with the card capacity |
21:37:18 | | Quit S_a_i_n_t (Read error: Connection reset by peer) |
21:39:17 | | Join S_a_i_n_t [0] (S_a_i_n_t@203.184.1.233) |
21:47:10 | | Nick jfc^2 is now known as 16SAAKYE6 (~john@dpc6682208002.direcpc.com) |
21:47:21 | | Nick 16SAAKYE6 is now known as jfc (~john@dpc6682208002.direcpc.com) |
21:49:55 | | Quit mt (Read error: Connection reset by peer) |
21:50:25 | *** | Saving seen data "./dancer.seen" |
21:50:29 | | Join mt [0] (~mtee@rockbox/developer/mt) |
21:55:17 | | Quit esperegu (Remote host closed the connection) |
21:57:38 | | Join bluebrother [0] (~dom@f053153142.adsl.alicedsl.de) |
21:57:38 | | Quit bluebrother (Changing host) |
21:57:38 | | Join bluebrother [0] (~dom@rockbox/developer/bluebrother) |
21:57:59 | | Join guysoft22 [0] (guy@bzq-79-180-53-22.red.bezeqint.net) |
21:58:10 | | Join DerPapst [0] (~Alexander@p4FE8F7D0.dip.t-dialin.net) |
21:58:12 | guysoft22 | hi all, where do i get mktccboot to install rockbox on a D2? |
21:58:21 | * | guysoft22 is using debian |
21:59:15 | funman | i think you must build it from svn |
21:59:51 | | Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven) |
22:00 |
22:00:33 | | Quit bluebroth3r (Ping timeout: 260 seconds) |
22:03:29 | guysoft22 | funman, ok, pulling it now.. |
22:03:45 | guysoft22 | guy@2[rockbox]$ svn co svn://svn.rockbox.org/rockbox/trunk rockbox svn: Can't connect to host 'svn.rockbox.org': Connection timed out :( |
22:03:57 | | Quit liar (Read error: Connection reset by peer) |
22:04:13 | guysoft22 | ah workign now |
22:04:14 | guysoft22 | strange |
22:04:17 | | Join petur [0] (~petur@rockbox/developer/petur) |
22:04:50 | | Quit Schmogel (Ping timeout: 265 seconds) |
22:04:57 | | Quit phanboy4 (Ping timeout: 258 seconds) |
22:06:28 | | Quit Strife89DS (Ping timeout: 240 seconds) |
22:09:28 | | Quit JohannesSM64 (Quit: WeeChat 0.3.3-dev) |
22:13:19 | Unhelpful | pixelma: do the WPS docs still specify clipping and aspect ratio controls for the AA tags? i don't think they've ever been supported, but maybe they should be. ;) |
22:26:33 | | Quit merbanan (Ping timeout: 276 seconds) |
22:28:09 | | Join bertrik [0] (~bertrik@rockbox/developer/bertrik) |
22:29:51 | | Join merbanan [0] (~banan@c-b21ce02b-74736162.cust.telenor.se) |
22:31:01 | guysoft22 | yay. got rockbock working on my D2. but doom seems to return an error |
22:31:16 | | Quit MethoS- (Remote host closed the connection) |
22:36:56 | | Quit Jaykay (Ping timeout: 252 seconds) |
22:42:17 | | Join stripwax [0] (~Miranda@87-194-34-169.bethere.co.uk) |
22:42:31 | | Quit Zagor_ (Quit: Leaving) |
22:44:41 | guysoft22 | um, where can i see a map of the 3x3 grid that emulates buttons? i am not sure how it is split |
22:47:00 | stripwax | n1s - cute changes to the tremor asm by the way - thanks! |
22:47:03 | | Nick fxb is now known as fxb__ (~felixbrun@h1252615.stratoserver.net) |
22:47:26 | pixelma | guysoft22: do you want to know what each "part" does in the different screens? |
23:00 |
23:01:20 | | Quit merbanan (Ping timeout: 260 seconds) |
23:01:23 | | Quit evilnick_B (Quit: Page closed) |
23:04:14 | | Quit rhodan () |
23:04:24 | | Join rhodan [0] (~Quassel@2001:1608:12:2::38) |
23:05:21 | kisak | in http://www.rockbox.org/wiki/TargetStatus -> New Platfoems under development -> Install with Rbutil, is the difference between "Yes" and "Yes (not yet released)" whether or not a platform is included when the most current version was published? |
23:06:44 | funman | yes, if it's 'not yet released' that means the code has been written but there is no binary available |
23:07:51 | kisak | so I should change the Sansa AMS v2 devices line to "Yes (not yet released)" |
23:14:18 | funman | yes |
23:15:12 | kisak | I was just running through today's changes and made a couple small wiki changes |
23:24:03 | | Quit CGL (Quit: Soy spammero http://wiki.n00b2hack.com.ve ---- \m/ d(>.<)b \m/) |
23:24:41 | | Join Buschel [0] (~ab@p54A3ED31.dip.t-dialin.net) |
23:26:08 | | Join merbanan [0] (~banan@c-94-255-217-84.cust.bredband2.com) |
23:28:05 | | Quit bmbl (Quit: Bye!) |
23:28:57 | guysoft22 | pixelma, ok, figured it out, and installed doom :) |
23:30:25 | pixelma | you might want to have a look at our manual anyways ;) |
23:30:47 | guysoft22 | pixelma, i will.. |
23:30:53 | guysoft22 | pixelma, the database is not working still :( |
23:31:11 | FlynDice | funman: 0x00F4F000 blocks reported for internal disk e280v2, 0x00F36000 for the 8GB uSD |
23:32:35 | funman | FlynDice: looks alright |
23:33:16 | FlynDice | seems reasonable at least... |
23:33:17 | | Join fdinel [0] (~Miranda@modemcable235.127-131-66.mc.videotron.ca) |
23:33:42 | funman | strange that we have to do STBY -> TRAN (to do select_bank(-1) -> STBY |
23:33:53 | | Quit B4gder (Quit: It is time to say moo) |
23:36:09 | Buschel | n1s: well done with your coldfire_set_macsr-findings! |
23:36:20 | FlynDice | I think the card needs to be in STBY to issue CMD9( send_CSD) perhaps looking now |
23:36:36 | funman | yes that's right |
23:36:40 | funman | you know this by heart :) |
23:36:47 | Buschel | n1s: I forgot to re-include this when exchanging the codec |
23:38:55 | FlynDice | That one got me way too many times... yes cmd 9 need to be given while the card is in STBY state |
23:39:37 | FlynDice | seems I'm way slower than you young guys though... |
23:40:42 | | Join Blue_Dude [0] (~chatzilla@rockbox/developer/Blue-Dude) |
23:41:03 | Blue_Dude | OK I'm stuck. I just tried a clean build of the e200 sim in Cygwin. Several times. And I keep getting this: http://pastie.org/961005 |
23:41:32 | CIA-5 | New commit by funman (r26029): as3525v1: detect correctly internal storage capacity ... |
23:42:14 | funman | FlynDice: i have less kids to slow me down otoh ;) |
23:42:42 | gevaerts | Blue_Dude: hm, that looks as if kugel's changes broke the sim on cygwin... |
23:42:54 | Blue_Dude | Yeah. :( |
23:43:02 | FlynDice | the kids aren't nearly as taxing as the "domestic godess".... ;) |
23:43:15 | funman | :D |
23:43:19 | Blue_Dude | I thought it was something I did or something broke. I tried to reinstall Cygwin components and no joy. |
23:45:56 | | Quit TheSeven (Remote host closed the connection) |
23:47:48 | funman | FlynDice: why did you put the active_card bit in amsv2, does it harm to change B5 at each send_cmd() call ? |
23:49:36 | FlynDice | That came from the OF disassembly but I think I tried it with that code disabled and both cards worked ok. Just never went back to actually test. |
23:50:11 | funman | seems to work (looking at fuzev2 button led) |
23:50:26 | gevaerts | Blue_Dude: I can reproduce it using mingw here |
23:50:27 | *** | Saving seen data "./dancer.seen" |
23:50:31 | | Quit sevard (Read error: Connection reset by peer) |
23:50:37 | Blue_Dude | Really? Oh good. |
23:50:43 | Blue_Dude | That's a start. |
23:50:49 | gevaerts | That doesn't mean I can easily fix it :) |
23:50:58 | | Join sevard [0] (sev@216.164.6.24) |
23:51:07 | Blue_Dude | Well, it's better than a lone cry in the wilderness. |
23:56:56 | | Quit shodan45 (Remote host closed the connection) |
23:57:12 | dfkt | funman, i'm doing some RMAA tests on the clip+, with louder volume levels. will post the results in a few minutes. |