| 00:01:43 | | Quit n1s () |
| 00:02:16 | | Join crzyboyster [0] (n=4b596e13@gateway/web/cgi-irc/labb.contactor.se/x-a85719e65ad76cda) |
| 00:03:12 | crzyboyster | What does everyone think of this version of cabbiev2 for the gigabeat > http://forums.rockbox.org/index.php?topic=10030.msg114838#msg114838 ? |
| 00:04:22 | Nico_P | I love it!! :p |
| 00:04:43 | * | Llorean is a fan |
| 00:05:04 | pixelma | see logs, not that long ago (though it's now yesterday's log) |
| 00:05:52 | Nico_P | I'd like to mention that the background was slightly hacked so a bit more work might be needed on that respect |
| 00:07:15 | | Join rasher_ [0] (n=rasher@62.79.64.148.adsl.hs.tiscali.dk) |
| 00:07:35 | Nico_P | haah one less bug! :) |
| 00:08:17 | jhMikeS | Who's good at approaching companies around here? I'm afraid I'd just get in a fight with them and they don't want that. :) |
| 00:08:51 | | Quit MethoS- (Remote closed the connection) |
| 00:11:49 | crzyboyster | Which day's log? I couldn't find it being discussed in yesterday's log... |
| 00:12:25 | kugel | saratoga: Are you sure it's not charging at all? I'm refering to your last post in "cpchan branch" |
| 00:12:44 | linuxstb | gevaerts: I'm not happy with the structure (i.e. the #ifdefs, and this should probably be moved to the target-specific parts of the source), but this patch (apply on top of your patch) gives the correct iManufacturer/iProduct/iSerial for ipods (plus VendorId and ProductId in inquiry) - http://www.davechapman.f2s.com/rockbox/ums_serial.diff |
| 00:12:47 | kugel | It does charge, very very slowly though. I was on a train trip recently, started with an empty battery. I put the sansa into a power source and watched a movie with rockbox. I was able to watch 1 hour + listen a few minutes to music after disconnecting the sansa from the power source |
| 00:13:32 | pixelma | crzyboyster: yesterday's log as in 10th of February, not the current.txt (ok, I prefer reading) |
| 00:15:06 | linuxstb | jhMikeS: Bagder seems to have more success than most... |
| 00:15:15 | saratoga | kugel: on my sansa the battery charger status register indicates that the charge circuit is powered down |
| 00:15:24 | saratoga | and we certainly do not enable it |
| 00:15:56 | kugel | saratoga: Ok, but why was my sansa not empty after I disconnected it from the power source? |
| 00:16:00 | saratoga | so unless your bootloader is enabling it, i don't see how it could charge |
| 00:16:38 | kugel | Nico_P: Oh I just noticed your last commit. I had this issue too, but I couldn't reproduce it and I had it very rarely. I didn't think of the next track + VBR combo at all |
| 00:16:39 | saratoga | kugel: perhaps the battery heated up which raised voltage high enough to run rockbox futher |
| 00:16:40 | * | jhMikeS summons Bagder to sweet talk FS into getting the MC13783 RM :) |
| 00:17:00 | pixelma | saratoga: battery charger status register - is this something in a debug screen? |
| 00:17:04 | rasher_ | saratoga: are you saying the sansas don't charge in Rockbox? |
| 00:17:11 | | Quit crzyboyster ("CGI:IRC") |
| 00:17:42 | kugel | saratoga: Hmm, well. So tell me, why is charging disabled? |
| 00:17:59 | kugel | It was enabled if iirc |
| 00:18:26 | kugel | -if |
| 00:18:28 | pixelma | just asking because I don't trust what Rockbox is telling me when USB ("charger") is connected - the reported battery voltage of 4,6V and up can't be right |
| 00:18:51 | | Quit asdrubal (Read error: 110 (Connection timed out)) |
| 00:19:32 | | Quit keanu ("Leaving") |
| 00:19:57 | | Join countrymonkey [0] (n=4b05639a@gateway/web/cgi-irc/labb.contactor.se/x-1f75a7f6572306b2) |
| 00:20:26 | kugel | Hmm |
| 00:20:27 | | Quit countrymonkey (Client Quit) |
| 00:20:44 | XavierGr | Slasheri: for the logs, your bootloader.iriver file that you've sent me works (of course I tested the iriver_flash.rock with current svn and the crc checksum works too) |
| 00:21:43 | gevaerts | linuxstb: I think the ipod stuff should probably come before the flash stuff in identify2inquiry(). Right now flash-based ipods won't get a correct id. |
| 00:21:47 | | Quit rasher (Read error: 110 (Connection timed out)) |
| 00:21:47 | | Nick rasher_ is now known as rasher (n=rasher@rockbox/developer/rasher) |
| 00:22:18 | saratoga | kugel: see http://www.rockbox.org/tracker/task/8363 |
| 00:22:26 | Nico_P | kugel: I found a reliable way to repro so it wasn't *too* hard to track down (still not straightforward though) |
| 00:25:20 | saratoga | bit 0 of 0x22 (the first byte added to debug_menu.c in that patch) is the charger enable bit |
| 00:25:34 | | Quit robin0800 (" HydraIRC -> http://www.hydrairc.com <- Chicks dig it") |
| 00:25:55 | kugel | yea, I see |
| 00:26:01 | | Join NoFullName [0] (n=full@cpe-24-160-175-197.columbus.res.rr.com) |
| 00:26:18 | kugel | saratoga: Ahh, thanks. I remember that task, somehow I've forgotten about it. Does that patch a bad job, or is it reliable (for e200 at least)? If the latter what is the reason for not committing? |
| 00:26:19 | gevaerts | linuxstb: as far as the #ifdefs are concerned, there's a lot of stuff in the usb stack that needs cleanup, so a bit more right now won't hurt |
| 00:26:20 | pixelma | saratoga: do you get more reasonable battery voltage readouts during charging with that patch? |
| 00:26:24 | | Quit Axio (Remote closed the connection) |
| 00:26:34 | NoFullName | Hi there, I got some issue with panic error anyone could help? |
| 00:27:11 | saratoga | pixelma: I haven't tried it yet |
| 00:27:25 | saratoga | i didn't see much point until USB support was in and it seemed like it was a ways off |
| 00:27:41 | saratoga | i think it would work ok for the e200 (not the c200 though) |
| 00:27:43 | jhMikeS | isn't a version of gcc available that compiles for ARM11? |
| 00:28:10 | stripwax | NoFullName - is it reproducible? which device? |
| 00:29:49 | | Join asdrubal [0] (n=abc@cpe-76-190-216-97.neo.res.rr.com) |
| 00:29:49 | NoFullName | stripwax: It's third time that happen on my Sansa E200 series: It read" *PANIC* Udating size on empty dir entry 39" I tried to google for solution, it's very similar to this http://osdir.com/ml/systems.archos.rockbox.sourceforge/2003-12/msg00311.html |
| 00:30:21 | stripwax | NoFullName - did you by any chance try and delete database.* files when you got that? |
| 00:30:41 | saratoga | pixelma: I was thinking of doing a charge cycle hooked up to an ADC with that patch so I could verify that it really did observe correct safety precautions (over heating detection, etc) |
| 00:30:59 | saratoga | that specs are a bit vague about what the hardware will do IMO |
| 00:31:02 | stripwax | NoFullName - in fact, were you doing anything in particular when you got that PANIC or did it just happen 'by itself' when listening to music? |
| 00:31:05 | NoFullName | stripwax: Nope. what's the extension for database file? *? |
| 00:31:27 | BigBambi | *.tcd in /.rockbox |
| 00:31:50 | NoFullName | stripwax: I just turn it on and got that message. |
| 00:31:58 | stripwax | NoFullName - actually I meant 'did doing that cause the problem' rather than 'try doing that' :) |
| 00:32:01 | stripwax | NoFullName - hm. |
| 00:32:19 | | Quit mirak (Read error: 110 (Connection timed out)) |
| 00:32:23 | pixelma | saratoga: I was just talking about the number I can read in the "view battery" debug screen |
| 00:32:24 | stripwax | NoFullName - can you try resetting settings while turning it on? I'm afraid I don't know the Sansa keypress to do that |
| 00:32:52 | saratoga | I guess that should have been directed at kugel |
| 00:32:58 | saratoga | i mixed up your questions |
| 00:33:01 | NoFullName | stripwax: Sansa won't let me turn off the device. I have to either let the battery drain out or take up the battery. |
| 00:33:11 | BigBambi | NoFullName: Hold power |
| 00:33:12 | stripwax | Nico_P - hm, I can't reproduce that playlist not-buffering-the-correct-track problem I mentioned using sim .. |
| 00:33:14 | pixelma | saratoga: heh :) |
| 00:33:34 | BigBambi | NoFullName: For 15 seconds or more - it is a hardware reset |
| 00:33:46 | stripwax | and my ipod is running a battery_bench right now so I can't reproduce on that either :) |
| 00:33:50 | NoFullName | stripwax: lol. amazing. it turn off! |
| 00:33:52 | Nico_P | jhMikeS: what do you mean? |
| 00:34:25 | * | stripwax tries enabling dircache in sim and trying again |
| 00:34:31 | Genesis | bye |
| 00:34:38 | | Quit Genesis (Remote closed the connection) |
| 00:35:17 | amiconn | stripwax: You need multiple targets ;) |
| 00:35:21 | rasher | NoFullName: Not amazing - it's built to do that. Hold record while booting to reset settings |
| 00:35:35 | kugel | saratoga: I don't own such electricity stuff (ADC, not even a device to messure the power consumtions of i.e. my pc), so I unfortunately can't help :( |
| 00:35:48 | stripwax | amiconn - heh. i wonder if I can reproduce the bug on my h120 .. |
| 00:36:03 | NoFullName | rasher: well, I tried that before for 10 seconds and it didn't work, but I guess you need to hold longer than that. :) |
| 00:36:08 | linuxstb | gevaerts: Hmm, flash-based ipods (i.e. the Nano) use the ATA driver, so they should use the ATA identify info. |
| 00:36:17 | Nico_P | amiconn: do you have an idea how much time the ipod will take to arrive in France? |
| 00:36:18 | kugel | Nice to have this cleared now, I really thought it's charging at least a bit |
| 00:36:25 | BigBambi | NoFullName: Yes, 15 seconds or more as I said |
| 00:37:12 | NoFullName | BigBambi: So my solution is try to delete *tcd file in /.rockrock directory? |
| 00:37:17 | BigBambi | eh? |
| 00:37:35 | BigBambi | I was just telling you which files were database files |
| 00:37:46 | | Join countrymonkey [0] (n=4b05639a@gateway/web/cgi-irc/labb.contactor.se/x-e221173d7af4eebc) |
| 00:38:08 | NoFullName | BigBambi: oh ok. I will delete it anyway :) If you don't heard from me it's mostly likely it works. |
| 00:38:09 | BigBambi | NoFullName: I didn't really read about your problem |
| 00:38:13 | countrymonkey | Is rbspeexenc built into rbutil? I'm having problems building .talk clips. |
| 00:38:21 | BigBambi | NoFullName: heh, good luck :) |
| 00:38:32 | NoFullName | thanks everyone! |
| 00:38:42 | | Part NoFullName |
| 00:38:43 | stripwax | NoFullName - try resetting your settings first before deleting the .tcd files .. |
| 00:38:44 | stripwax | oh |
| 00:39:02 | jhMikeS | Nico_P: Right now it's just set to arm9tdmi. |
| 00:39:06 | gevaerts | linuxstb: ok. |
| 00:39:16 | Nico_P | jhMikeS: n1s changed that today |
| 00:40:13 | preglow | countrymonkey: yeah, it's built in |
| 00:40:21 | jhMikeS | I tried to change it but without success |
| 00:40:21 | preglow | at least in any recent version |
| 00:40:49 | countrymonkey | I cannot figure out why I'm having problems building .talk clips! :p |
| 00:41:05 | Nico_P | jhMikeS: maybe I don't really understand what you mean... isn't http://svn.rockbox.org/viewvc.cgi/trunk/tools/configure?r1=16272;r2=16273 the correct change? |
| 00:41:30 | Nico_P | I had to update my cross-compiler to be able to compile with that change |
| 00:41:46 | jhMikeS | I wanted to use the mcrr instruction |
| 00:42:24 | Nico_P | which is? |
| 00:42:42 | linuxstb | gevaerts: But HAVE_FLASH_STORAGE is defined for the Nano, so I'm not sure if the other code in the USB driver works correctly... |
| 00:42:44 | countrymonkey | does rbutil do only the dirrectory you set it to do or does it do all directories under that one also. And can it do a whole drive like voicebox can? |
| 00:43:10 | saratoga | it transfers registers between processors i think |
| 00:43:11 | jhMikeS | guess I failed to discover the correct type |
| 00:43:34 | jhMikeS | 4.0.3 seems to work |
| 00:43:39 | gevaerts | There are a lot of #ifdefs in usb_core.c that could be removed/centralized by using function pointers. That would also open the way to supporting more than one usb function at the same time (provided the hardware supports enough endpoints). Are there objections to doing that ? |
| 00:43:51 | | Quit jhulst (Success) |
| 00:44:00 | stripwax | Nico_P - hm, maybe I can reproduce the bug in sim after all. From 'stopped', I play a track on an album, and check the Buffering debug info - I see track count: 23 (which is correct). Music starts playing, I then go back to fiiles, hold Select on a different album, and Insert Next. Then go back to the Buffering debug info - I see track count is now 19 (which is not the number of tracks on the different album) |
| 00:44:07 | stripwax | Is that correct behaviour? |
| 00:44:35 | Nico_P | could be |
| 00:44:37 | stripwax | I can't reproduce the playing-the-wrong-music behaviour on sim, but I can see that as soon as the current track finishes, then the entire rest of the playlist gets buffered in a one-er |
| 00:44:37 | linuxstb | gevaerts: I don't think there are enough endpoints to do that... |
| 00:45:03 | stripwax | I imagine on-device that that won't work well |
| 00:45:08 | Nico_P | ah, it's not buffered when the playlist changes? |
| 00:45:12 | stripwax | nope |
| 00:45:17 | Nico_P | that's not good |
| 00:45:19 | stripwax | right |
| 00:46:04 | gevaerts | linuxstb: the rest of the code doesn't use HAVE_FLASH_STORAGE to differentiate between types. It either uses the cardinfo stuff or ata. cardinfo gets used if HAVE_HOTSWAP is defined. |
| 00:46:06 | stripwax | It looks like the real and usefl indicators drop back down to almost zero when i Insert Next. prior to that they were almost full |
| 00:46:33 | stripwax | as if only the currently-playing track remains buffered, everything else gets zapped, and then it needs to buffer everything on the track change |
| 00:46:56 | gevaerts | linuxstb: if there are only enough endpoints for a single function we could still make it runtime-configurable. |
| 00:47:20 | linuxstb | gevaerts: Yes, I just noticed that. That doesn't seem logically correct (using HAVE_HOTSWAP), but will work for current targets. |
| 00:47:30 | gevaerts | linuxstb: anyway the first function is not fone yet, so it's still a bit hypothetical |
| 00:47:34 | | Quit obo ("bye") |
| 00:47:49 | Nico_P | stripwax: it makes sense for the existing buffered data to be flushed, but I don't really understand why no new data is then buffered |
| 00:47:58 | jhMikeS | Nico_P: mcrr lets you do cache range operations in one instruction |
| 00:47:58 | stripwax | bleh, it does it for Insert Last also. real and usefl drop to a very small number (compared to how much was buffered before) |
| 00:48:08 | linuxstb | gevaerts: Yes, I think the usb_serial driver works (to some extent), so we want to make it run-time configurable. |
| 00:48:11 | Nico_P | jhMikeS: and it's new to ARM11? |
| 00:48:15 | stripwax | Insert Last should very much *not* flush the existing buffer |
| 00:48:22 | stripwax | unless I'm missing something? |
| 00:48:31 | Nico_P | stripwax: agreed, but the playback code doesn't know the diff |
| 00:48:39 | Nico_P | to it, both are just "new playlist" |
| 00:48:47 | stripwax | oh, ouch. |
| 00:49:00 | | Join fasmaie [0] (n=yohann@c-98-216-170-85.hsd1.ma.comcast.net) |
| 00:49:06 | stripwax | it rebuffers the entire buffer whenever you add anything to the end of the playlist? |
| 00:49:19 | Nico_P | whenever you change the playlist |
| 00:49:34 | gevaerts | linuxstb: the cardinfo stuff is declared in hotswap.h, so following that logic HAVE_HOTSWAP is correct. However once a HD player with SD slot turns up we are in trouble |
| 00:49:42 | linuxstb | Can ipod users (with Linux) check their logs and let tell me what the line "scsi 24:0:0:0: Direct-Access Apple iPod 1.62 PQ: 0 ANSI: 0" looks like? Specifically, do you get "1.62" displayed, or another number? |
| 00:49:54 | linuxstb | gevaerts: Yes, or a flash-based target with no removable cards... |
| 00:49:57 | stripwax | Nico_P that's not ideal |
| 00:50:36 | stripwax | What's the Track Count refer to (it certainly doesn't seem to be either the number of tracks buffered nor the number of tracks in the playlist) |
| 00:51:12 | linuxstb | (that request was for UMS with the OF, not the Rockbox UMS patch) |
| 00:51:15 | Nico_P | indeed not. if that conforts you, it's not new to MoB. FS #8029 describes the same kind of issue. |
| 00:51:24 | stripwax | Nico_P - ta |
| 00:51:37 | Nico_P | stripwax: it's the number of buffered tracks (according to playback.c) |
| 00:51:42 | gevaerts | linuxstb: I think the entire storage infrastructure should be reworked to support these combinations. It's done to some extent (i.e. by having the sansa code have an ata-lookalike interface), but it will have to be more dynamic at some point. |
| 00:51:50 | stripwax | Nico_P - hm, it can't be right |
| 00:52:02 | Nico_P | i.e. the number of tracks for which metadata is loaded |
| 00:52:11 | stripwax | oh, just the metadata? ok then |
| 00:52:27 | Nico_P | audio is handles by buffering.c |
| 00:52:37 | stripwax | was comparing the Track Count to the amount of usefl data |
| 00:52:51 | * | gevaerts is afraid that his last remark might sound like volunteering... |
| 00:53:21 | Nico_P | stripwax: usefl is also relative to where the read position is in the buffer |
| 00:53:29 | | Quit ender` (" To get as fewest unhappy people as possible, always bully the same ones.") |
| 00:53:32 | linuxstb | gevaerts: That's what happens when you start something in Rockbox - it never ends... |
| 00:53:46 | Nico_P | it's basically (readpos - writepos) % wrap |
| 00:54:35 | gevaerts | linuxstb: looks like it |
| 00:54:35 | stripwax | any way to see what metadata has been buffered at any point in time? |
| 00:55:19 | Nico_P | I don't think so, but I'm also not sure I quite understand the question |
| 00:55:44 | stripwax | Nico_P - just wondering if I can get any sense why adding 11 tracks to a 23-track playlist changes the Track Count from 23 to 19 .. |
| 00:56:45 | Nico_P | well I guess your 23 tracks would fit in the buffer, and when you added the 11 tracks everything got rebuffered and only a total of 19 tracks could fit |
| 00:57:17 | | Quit kugel ("ChatZilla 0.9.80 [Firefox 2.0.0.12/2008020710]") |
| 00:58:11 | stripwax | Nico_P - that makes sense I suppose. Are you able to reproduce the nonbuffering problem? |
| 00:58:18 | stripwax | of audio, that is |
| 00:58:29 | Nico_P | stripwax: alloc and real can help you too on that. alloc is the spaced allocated, real is the space actually used |
| 00:58:48 | Nico_P | I've just seen the problem occur |
| 00:58:50 | stripwax | alloc stays full, real drops very low |
| 00:59:25 | Nico_P | yeah, the buffering thread just doesn't notice it needs to get working |
| 00:59:30 | | Quit lee-qid (Read error: 104 (Connection reset by peer)) |
| 00:59:53 | stripwax | does stopping playback flush the buffer too? stopping playback doesn't seem to change Track Count, even though alloc drops to zero. |
| 01:00 |
| 01:00:42 | | Quit BHSPitLappy (Read error: 110 (Connection timed out)) |
| 01:00:55 | Nico_P | that's probably because the variables that are used to calculate track_count aren't reset when you stop playback |
| 01:01:01 | * | Nico_P checks |
| 01:02:10 | Nico_P | the debug screen isn't really self-explanatory... it's most useful when you know the underlying code at least a bit |
| 01:02:20 | stripwax | true |
| 01:02:47 | Nico_P | the buffering started just before the track change occured |
| 01:02:53 | Nico_P | which is logical |
| 01:04:44 | stripwax | not sure I follow - so the buffering is ok? |
| 01:05:23 | jhMikeS | Nico_P: http://jhmikes.cleansoap.org/DDI0211J_arm1136_r1p5_trm.pdf |
| 01:05:23 | Nico_P | well at least it's not completely broken. it's just that it doesn't take advantage of a spinning disk when it should |
| 01:06:58 | Nico_P | a nice read :) |
| 01:07:06 | stripwax | - and disk will be spinning at that point, since user changed the playlist |
| 01:07:18 | Nico_P | indeed |
| 01:07:35 | Nico_P | the playback thread should even have told the buffering thread to buffer some data |
| 01:08:13 | *** | Saving seen data "./dancer.seen" |
| 01:08:27 | | Join SacredTerror [0] (n=Miranda@pool-71-190-214-151.nycmny.fios.verizon.net) |
| 01:10:26 | stripwax | Does it wake the thread up? |
| 01:10:53 | countrymonkey | testing |
| 01:10:54 | stripwax | actually I don't know the buffering thread at all, so feel free to ignore that question :) |
| 01:12:20 | Nico_P | hmm in fact in that case there won't be a buffering request (for quite good reason) |
| 01:13:04 | stripwax | why? |
| 01:13:06 | Nico_P | and the disk activity detection code in the buffering thread is disabled |
| 01:14:07 | Nico_P | the func that loads a track from playback.c is audio_load_track. it has a boolean param (start_play) indicating whether that track will be played immediately or not. we only request buffering if start_play == true |
| 01:14:23 | Nico_P | because we don't want to request buffering for every track we add |
| 01:15:04 | stripwax | however playing a single track causes the entire album to be buffered |
| 01:15:38 | gevaerts | linuxstb: I uploaded a new patch with your changes integrated, and a few minor things. Tomorrow I'll start refactoring the usb stuff a bit to make it more flexible. |
| 01:15:44 | linuxstb | amiconn: Do you know if the PP has a serial number embedded, and if so, where it is? |
| 01:16:05 | stripwax | I thought we would request buffering for all the next 'n' tracks until the buffer is full (but I assume that is not the case) |
| 01:16:08 | | Quit countrymonkey ("CGI:IRC (Ping timeout)") |
| 01:16:28 | linuxstb | gevaerts: Do you have any idea where the serial number is coming from that the Sansa OF is generating? |
| 01:17:44 | jhMikeS | linuxstb: I think it's in the ROM somewhere |
| 01:18:00 | Nico_P | gevaerts, linuxstb: maybe it would be good to commit the current patch before starting to refactor |
| 01:18:21 | linuxstb | jhMikeS: I didn't think there was a ROM on the Sansas? |
| 01:18:23 | | Quit einhirn (Read error: 104 (Connection reset by peer)) |
| 01:18:30 | jhMikeS | linuxstb: yes, it's an i2c rom |
| 01:19:03 | Nico_P | stripwax: requesting buffering of a track aborts the current buffering and starts buffering the requested track, so if we request it for every track we add the first one won't have a chance to get buffered, and it's the most important one |
| 01:19:09 | linuxstb | Could someone dump it and compare with the USB serial ID? |
| 01:19:59 | Nico_P | stripwax: but once the buffering thread starts buffering data, it'll go on util the buffer is full. it just needs a starter basically |
| 01:20:15 | Nico_P | current starters are an explicit request or a low buffer |
| 01:20:42 | stripwax | Nico_P - wouldn't the starter always be the currently-playing track? or is that the problem, there's no way to 'add' tracks to the buffer if something is already in there (because rebuffering would blow it away)? |
| 01:21:43 | stripwax | i.e. Insert Next, Insert Last, etc would just rebuffer everything starting from the currently playing track |
| 01:21:52 | Nico_P | stripwax: the problem here is the code at playback.c:1855 (in audio_load_track) |
| 01:22:27 | Nico_P | playlist events will trigger a rebuffer (i.e. readding all the tracks), but will never ask the buffering thread to buffer data |
| 01:22:55 | Nico_P | I think I have an idea though |
| 01:23:16 | stripwax | Nico_P - I have to head off now but will be reading logs with interest when I return :) |
| 01:23:23 | stripwax | gnight |
| 01:23:26 | Nico_P | gnight |
| 01:23:29 | | Quit stripwax ("Miranda IM! Smaller, Faster, Easier. http://miranda-im.org") |
| 01:23:48 | * | jhMikeS wonders why he calculated 534MHz when the S processor is supposed to be 532MHz. |
| 01:25:09 | gevaerts | Nico_P: I don't mind comitting the current patch, but I'm not sure it's really needed. The refactoring I want to do is not terribly heavy, mostly moving stuff to different files and trying to get rid of some #ifdefs. |
| 01:26:12 | pixelma | Nico_P: logf.txt looks identical with a skipping and a normal folder (just playback.c enabled logf) |
| 01:26:18 | gevaerts | Nico_P: but I wouldn't enable the usb storage stuff by default yet. I think it needs more testing my people who know what they are getting into |
| 01:27:14 | Nico_P | gevaerts: it's your call, but it would be better not to commit a huge patch |
| 01:27:41 | linuxstb | The speed is also an issue IMO - on ipods at least (where the OF disk mode is very convenient), I would still prefer to use it to the Rockbox UMS code. But we could always commit, but not enable. |
| 01:28:53 | gevaerts | linuxstb: I think enabling it now in official builds is madness, so as far as I'm concerned you don't need to worry about that. |
| 01:30:00 | linuxstb | gevaerts: BTW, when you copy a new rockbox.mi4 file, does Rockbox detect it, and prompt you to load it? |
| 01:30:04 | Nico_P | pixelma: strange |
| 01:30:06 | gevaerts | linuxstb: yes |
| 01:30:12 | linuxstb | And it works? |
| 01:30:12 | pixelma | Nico_P: should I try enabling buffering.c too? Or playlist? |
| 01:30:50 | Nico_P | pixelma: maybe in the playlist code, yes. but I don't even know whether there are logf messages in there |
| 01:31:33 | gevaerts | Nico_P: I'll try to get a patch tomorror night that keeps it disabled with an easy-to-use #define to enable it. Hopefully by then I'll also have the sd card working in windows. |
| 01:31:36 | | Join Triple_7 [0] (n=18d97411@gateway/web/cgi-irc/labb.contactor.se/x-4ac5c8e8e1465039) |
| 01:31:47 | gevaerts | Nico_P: we can decide then whether to commit or not. |
| 01:31:48 | Triple_7 | Hello |
| 01:32:17 | gevaerts | linuxstb: looks like it anyway. |
| 01:33:12 | linuxstb | gevaerts: OK. I haven't managed to get it to work on my ipod - with some #ifdef changes, it now does the detection, but it refuses to load it. |
| 01:33:25 | | Quit Triple_7 (Client Quit) |
| 01:34:12 | * | gevaerts is going to sleep a bit earlier than on previous days today... tomorrow is a working day |
| 01:36:02 | * | amiconn wonders whether this code is faster than ipod G5/nano diskmode, or H10 UMS mode |
| 01:37:12 | pixelma | Nico_P: doesn't look like there is |
| 01:37:23 | * | gevaerts gets 300 to 450 kbps |
| 01:37:29 | | Join cool_walking_ [0] (n=Miranda@203-59-129-195.perm.iinet.net.au) |
| 01:38:23 | amiconn | gevaerts: It's not the kbps - both G5/nano diskmode and H10 mus mode insert rather long pauses between files |
| 01:38:54 | gevaerts | amiconn: between files ? How do they do that ? |
| 01:39:29 | amiconn | If you're copying a few large files, it's fast-ish (even a bit faster than fullspeed), but with many small files like when unzipping a rockbox build, it takes ages |
| 01:39:48 | * | gevaerts is considering if we can call our usb-resets a feature |
| 01:39:55 | amiconn | I don't know why that happens, but it's a fact, and annyoing |
| 01:41:14 | gevaerts | amiconn: is this on windows or linux ? It might be related to the 'removable' bit in the scsi inquiry data |
| 01:41:22 | amiconn | windows |
| 01:42:35 | amiconn | But it doesn't happen with other ipod's diskmode (e.g. my mini G2), or when booting into the OF and using its usb mode |
| 01:43:28 | amiconn | ipod diskmode resides in flash rom, and is also called 'emergency diskmode'. On all ipods except G5 and Nano, it's as fast as the OF's usb mode |
| 01:46:31 | linuxstb | gevaerts: I've just compared the scsi inquiry data on my ipod, and there are two differences - 1) "version" is 0x00 [no conformance claimed] in Rockbox, 0x03 [SFC] in the OF; 2) Resp_data_format is 0 in Rockbox, 3 in the OF. Plus the OF includes a "Unit serial number", which I'm guessing is the disk's ID? |
| 01:46:54 | | Join barrywardell [0] (n=barrywar@89-125-27-182.dhcp-ripwave.irishbroadband.ie) |
| 01:47:15 | gevaerts | amiconn: I would say it's a problem with the device_removable bit. Maybe there's a setting in the device manager to fix that |
| 01:48:03 | linuxstb | gevaerts: Plus one more - "RMB" is 1 in RB, 0 in the OF (I guess that's the device_removable bit you're talking about)... |
| 01:48:38 | linuxstb | Sorry, reverse that - 1 in OF, 0 in RB... |
| 01:49:08 | amiconn | Why does rockbox not set the removable bit? |
| 01:49:12 | linuxstb | In fact, reverse it back... |
| 01:49:24 | * | linuxstb should go to sleep |
| 01:50:05 | pixelma | Nico_P: or do I have to let playback run till it stops at the end of playlist? I thought let it playing until it skips and stop would be enough |
| 01:50:36 | Nico_P | latter should be fine, yes |
| 01:51:23 | pixelma | then there is no difference... |
| 01:52:20 | gevaerts | amiconn: what does it mean to be removable ? In SCSI terms I guess it means something like a cd or a floppy, so we should certainly set it for the sd card slot. But should we set the removable bit just because you can unplug the cable (I can also unplug my scsi disks, but they are not marked as removable...) ? I think it's not entirely clear |
| 01:52:55 | rasher | gevaerts: Could it perhaps be intended originally for stuff like hot-swappable disks? |
| 01:53:10 | gevaerts | linuxstb: version should be 3. If not, it looks like you have something overwriting the inquiry data before it gets used. |
| 01:53:17 | rasher | In which case the dap should certainly set it as well |
| 01:53:30 | Nico_P | pixelma: maybe it's the codec that for some reason asks for the next track |
| 01:53:52 | soap | PS - Slashdot story on"Best Open Source License for Hardware" is eerie in its timing. |
| 01:54:08 | linuxstb | gevaerts: It is 3 - I got very confused over which output was which... Here it is - http://www.pastebin.ca/899294 |
| 01:54:14 | rasher | soap: surprisingly light on details though |
| 01:54:32 | scorche|sh | soap: the slashdot slime is gaining sentience |
| 01:55:21 | gevaerts | rasher: The thing is that windows uses that bit to decide whether writes should be synchronous, so setting it means being a lot slower (but users don't have to choose "safe remove") |
| 01:56:03 | | Quit fasmaie (Read error: 113 (No route to host)) |
| 01:56:31 | | Join perrikwp [0] (i=4aa794a0@gateway/web/ajax/mibbit.com/x-928dae5b54895e32) |
| 01:56:49 | rasher | gevaerts: Ah, so I take it the OF (of Sansa at least) don't set it? |
| 01:57:12 | gevaerts | linuxstb: I had that problem earlier today. inquiry got overwritten by the create_thread in usb_core.c (it was ok ust before the call and had zeroes just after and at the first line of the thread itself) |
| 01:57:57 | pixelma | Nico_P: repeated the test with enabled logf in buffering.c too, in a skipping folder I get: "bunch of bufclose", then "releasing all tracks", then another bunch of "bufclose", then "clearing tracks:3/8,0" "Codec finished". The line below shows the path to the folder, I can paste the complete output if you want but wanted to make a test run with a working folder too before. |
| 01:58:12 | gevaerts | rasher: I think the sansa sets it, but on a flash device the effect is a lot different than on a disk, since you don't have huge seek times |
| 01:58:30 | jhMikeS | gevaerts: say more. |
| 01:58:34 | Nico_P | pixelma: ok, thanks :) |
| 01:58:46 | | Quit SacredTerror ("Miranda IM! Smaller, Faster, Easier. http://miranda-im.org") |
| 01:59:35 | saratoga | the AS chip on th epp5024 actually has 128 bits of memory for storing serial numbers or encryption keys |
| 01:59:48 | saratoga | so the sansas could get it from there, though i really doubt it |
| 02:00 |
| 02:00:06 | jhMikeS | saratoga: the fuse array might be the place for it |
| 02:00:09 | | Join karashata [0] (n=Kimi@bas3-kitchener06-1096622343.dsl.bell.ca) |
| 02:00:12 | | Part karashata |
| 02:00:17 | | Join karashata [0] (n=Kimi@bas3-kitchener06-1096622343.dsl.bell.ca) |
| 02:00:18 | pixelma | Nico_P: does that already give you an idea? |
| 02:00:19 | * | jhMikeS forgot about that |
| 02:00:43 | Nico_P | pixelma: not really, no |
| 02:00:59 | pixelma | ok |
| 02:01:08 | Nico_P | well actually "codec finished" is very weird |
| 02:01:14 | | Join crzyboyster [0] (n=4b596e13@gateway/web/cgi-irc/labb.contactor.se/x-7a56872bd0798a62) |
| 02:01:43 | | Join gtkspert [0] (n=gtkspert@203-206-96-109.dyn.iinet.net.au) |
| 02:02:15 | crzyboyster | Nico_P: What font does the new gigabeat cabbiev2 use? helvR10 or helvr12? And can I have the actual wps? |
| 02:02:27 | | Quit barrywardell () |
| 02:02:46 | Nico_P | helvR12. sure you can have it |
| 02:03:36 | crzyboyster | Did you make it by modifying the wps in SVN right now? |
| 02:03:43 | gevaerts | jhMikeS: about the seek times or about the thread issue (I guess the thread) ? I noticed that one of the static structs (_inquiry) in usb_storage got half-zeroed, and it seems to happen in this create_thread call. It doesn't happen any more in the latest version of the patch because I re-initialize _inquiry everytime I need it (for unrelated reasons). It happens in v5 of the patch, but in that one I moved the declarations around to make t
|
| 02:03:46 | Nico_P | crzyboyster: just a sec though, I need to finish something |
| 02:03:58 | Nico_P | crzyboyster: I made it some time ago by changing the one in SVN |
| 02:04:48 | * | gevaerts thought he mean it when he said he was going to sleep earlier... |
| 02:04:59 | gevaerts | *meant |
| 02:05:19 | Nico_P | interesting. audio_flush_and_reload_tracks() gets called twice when I use playlist > insert |
| 02:05:19 | | Quit saratoga ("CGI:IRC (EOF)") |
| 02:05:48 | jhMikeS | gevaerts: it wasn't the thread doing it when it started to run? |
| 02:06:34 | gevaerts | jhMikeS: no. I checked the values at the first line of the thread. |
| 02:07:11 | pixelma | Nico_P: should I add it the two files to the tracker entry once I hae them both (shouldn't be too long) or do you prefer something else? |
| 02:07:46 | crzyboyster | Nico_P: Post it up onto the forum thread and I will rename and upload to patch tracker. And what files were modified? |
| 02:07:58 | jhMikeS | hmmm...perhaps a stack init problem though I'd expect 0xdeadbeef to be seen then |
| 02:08:31 | gevaerts | jhMikeS: yes, that's what I thought as well. It was zeroes though |
| 02:09:00 | Nico_P | stripwax: I have a patch that improves things (has a couple drawbacks though): http://pastebin.ca/899310 |
| 02:09:09 | | Quit XavierGr () |
| 02:09:34 | jhMikeS | gevearts: anything using UNCACHED_ADDR must also be cache aligned and padded |
| 02:09:36 | Nico_P | pixelma: yes, add them to the tracker. I probably won't have time to look this evening. it's already very late and I need to go to bed |
| 02:09:48 | pixelma | me too |
| 02:10:30 | Nico_P | crzyboyster: no need to add it to the tracker. if it's agreed on I'll commit |
| 02:10:58 | jhMikeS | gevearts: that's an out of the blue guess since it's sounds like that sort of thing to me |
| 02:11:06 | | Join DerPapst [0] (n=DerPapst@p5B23C586.dip.t-dialin.net) |
| 02:11:17 | gevaerts | jhMikeS: cache aligned means 32 bytes ? |
| 02:11:21 | * | DerPapst is still reading though the logs |
| 02:11:32 | DerPapst | linuxstb: 1.62 iirc |
| 02:12:27 | | Join psycho_maniac [0] (i=psycho_m@ppp192.hk.centurytel.net) |
| 02:12:57 | crzyboyster | Are all the files named properly? And I have no problem against Nico_P's version (though I don't own a gigabeat)... |
| 02:13:58 | linuxstb | DerPapst: That's a 5.5g? |
| 02:14:00 | | Quit gtkspert_ (Read error: 101 (Network is unreachable)) |
| 02:14:16 | DerPapst | yes |
| 02:14:27 | DerPapst | don't know about my 3G though |
| 02:14:30 | linuxstb | DerPapst: Thanks. So it's the same values for a Nano, ipod Color, 5g and 5.5g |
| 02:14:43 | gevaerts | jhMikeS: I think I'll drop those UNCACHED_ADDR pointers from anything that's not performance critical, and just memcpy the data |
icon identifies that the person is a core developer (has commit access).