Previous day | Jump to hour: 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 | Next day

Seconds: Show Hide | Joins: Show Hide | View raw
Font: Serif Sans-Serif Monospace | Size: Small Medium Large

Click in the nick column to highlight everything a person has said.
The Logo icon identifies that the person is a core developer (has commit access).

#rockbox log for 2012-01-06

00:00:07 Join Topy [0] (
00:00:17 Quit T44 (Read error: Connection reset by peer)
00:00:17CIA-88New commit by jethead71 (r31589): Coldfire: Fix alpha bitmap drawing from messing up the emac status register. ...
00:00:29max131i can build the bootloader myself, i also do run mkamsboot to patch the OF, but i don't find on the wiki anymore a description of how to turn the bootloader.bin into a bootloader.sansa file
00:01:05funmanmax131: 'make' does that already
00:01:10max131otherwise i am willing to give a couple of tries to determine at which point the USB issues started
00:02:01saratoga_max131: you don't need to change the bootloader to test this, just rockbox builds
00:02:14CIA-88r31589 build result: All green
00:04:26 Part jlbiasini
00:04:38max131well, i am definitely sure it worked with the r31083 and now it doesn't anymore
00:04:57max131and the only change i did was a change of bootloader
00:05:46saratoga_most of the early USB builds worked intermittently
00:06:15max131ok, to make it a bit more complete: i had r31083 patched with as3525v2-enable-usb.diff
00:06:58max131pity it does not work at all anymore on my Sony VAIO
00:07:00saratoga_max131: FWIW, r31083 doesn't have USB support
00:07:05saratoga_oh a patched build
00:07:11saratoga_yeah that will probably not work
00:07:11 Quit megal0maniac (Quit: CGI:IRC (EOF))
00:07:28[7]funman: I'm pretty much stuck
00:07:57[7]for some odd reason the OTG dislikes having multiple IN transactions at the same time
00:08:23max131well, what is confusing then is that on the MajorChanges wiki page, it indicates 'USB enabled on Sansa AMSv2'
00:08:31[7]judging from the linux kernel source this problem is not unknown, but they circumvent it by not using DMA at all, which I think is not an option for us
00:08:50saratoga_max131: yes, USB is now enabled, but it wasn't back in early December
00:09:02[7]the workaround would be to add mutexes (or some more trivial guard) to ensure that there is only ever one IN transfer at a time
00:09:16[7]this does of course increase the risk of deadlocks if the host doesn't behave properly
00:09:31[7]and it still didn't solve the HID trouble... no idea what's going on with that
00:09:58max131what does it actually mean the 'USB is now enabled', if it does not work in Windows 7?
00:10:23saratoga_max131: it means that the USB driver is compiled and enabled on the device
00:10:45max131ah ok, i understand now
00:11:12saratoga_FWIW aside from you, most people report that it works on Windows 7
00:11:43funmanmax131: it's enabled so we get reports when it doesn't work
00:11:47saratoga_funman, [7]: do either of you know anything about the state of the USB audio driver and AMS?
00:11:56funman[7]: why isn't PIO not an option ?
00:12:15funmansaratoga_: audio driver is pamaury's work and not committed afaik, gevaerts might know
00:12:25max131well, on another laptop running Windows 7 it is working, but it takes about 20 seconds or so before i see the USB driver letters popping up
00:12:35[7]I'd think PIO is even slower than having only one IN transaction at a time
00:12:58funmanFS #11644 might have benches
00:12:59fs-bluebot Playlist isn't loaded after reboot if it had ended (patches, new)
00:13:09funmanno, not this one :)
00:13:25[7]max131: that's normal
00:13:28funmanFS #11664
00:13:29fs-bluebot Workaround for usb random failures on amsv2. (patches, closed)
00:14:08[7]windows reads the whole FAT while mounting the volume, and that can take a while, especially if storage or the usb driver is rather slow or possibly buggy
00:14:28[7]it also depends on the cluster size the volume is formatted with
00:14:35funmanno there are no benchmarks
00:14:36[7]the smaller the clusters, the longer it takes, almost linearly
00:14:54funman[7]: doesn't the usb storage driver have a cache?
00:15:05[7]on the host or device side?
00:15:47max131the other strange thing i do notice is that on this particular laptop the USB-letters just show up for a second when extracting the USB cable
00:15:56[7]and windows issues single-sector reads while reading the fat
00:16:06[7]so you'll get a storage_read_sectors_md() call for every fat sector
00:16:53[7]not particularly efficient :)
00:20:42 Join megal0maniac [0] (
00:22:10 Join amithkk [0] (u4289@gateway/web/
00:22:10 Quit amithkk (Changing host)
00:22:10 Join amithkk [0] (u4289@2buntu/writers/amithkk)
00:25:41max131ok guys, i'm gonna try this on a couple of different computers in the forthcoming days, i'll keep you posted. bye
00:25:46 Join r4y [0] (
00:26:28 Quit max131 (Quit: CGI:IRC)
00:30:16r4yHello, I have an RCA mp3 player model th1814wm-a which will no longer mount, but I can play music from. I clicked safe removal and waited like I was supposed to so I don't know what caused this problem. I am running Ubuntu 10.04 LTS.
00:30:44r4yI didn't find rca on your site.
00:32:31megal0maniacWhich version of Rockbox are you running?
00:33:14megal0maniacOh wait. Are you running Rockbox at all?
00:33:25r4yI was told about Rockbox and haven't tried it yet
00:35:00 Quit KL1P (Quit: CGI:IRC (EOF))
00:35:32saratoga_r4y: rockbox doesn't run on your player
00:36:17megal0maniacSomeone with a little more Linux experience may be able to help you out, but honestly, I don't think that this is the best place to find an answer to your problem.
00:36:59r4yhmm, well TY for trying to help me.
00:37:54r4yI guess I will have to upgrade the firmware using Windows and if that doesn't work return it.
00:38:04r4ySee you.
00:38:07 Part r4y
00:43:45megal0maniacOn my Fuze+, occasionally the music will start playing faster, but with no pitch change, and there is "garbage" audio. Logic tells me that it's dropping samples. Usually after opening and closing a plugin. Restart usually fixes the problem, but what technical information should I gather when experiencing this behaviour?
00:44:04megal0maniacI see that the buffer is quite full, but it's also quite full when operating correctly
00:44:42saratoga_i think theres a flyspray task about that
00:45:07megal0maniacHmmm... I didn't find one, but I'll check again...
00:46:29funmanbluebrother^: lipo: mkamsboot-ppc and mkamsboot-i386 have the same architectures (ppc) and can't be in the same fat output file
00:48:17megal0maniacI haven't found anything. But back to my question, what debug info should I look for when this happens?
00:49:33 Quit nick_p (Quit: Leaving)
00:50:58 Quit ender` (Quit: The process takes just over six seconds, and would sound like a coffee machine passing an aluminum baseball bat if there were any atmosphere to carry the sound. -- Howard Tyler: Schlock Mercenary)
00:52:12megal0maniacIs there some kind of changelog between builds? I can't seem to reproduce the problem on the latest build.
00:52:32saratoga_megal0maniac: yes, see the svn logs on the front page
00:52:56megal0maniacThanks. Sorry, I'm kinda new at this
00:58:04bluebrother^funman: huh? lipo shouldn't be involved anymore at all.
00:58:34bluebrother^and that "make RBARCH=i386" looks like either something went wrong or you're using an outdated Makefile
00:59:10***Saving seen data "./dancer.seen"
00:59:11bluebrother^though I have to admit that I forgot trying to build the dmg
00:59:40funmanlibtools.make ?
01:01:52bluebrother^ah, that's a dependency on mkamsboot-dmg
01:03:18bluebrother^ok, that's still broken :/
01:15:44 Join domonoky1 [0] (
01:16:48 Quit domonoky (Ping timeout: 240 seconds)
01:19:26 Quit domonoky1 (Read error: Connection reset by peer)
01:19:34 Quit megal0maniac (Quit: CGI:IRC (Ping timeout))
01:24:22 Quit dfkt (Quit: -= SysReset 2.55=- Sic gorgiamus allos subjectatos nunc.)
01:30:16 Join dfkt [0] (dfkt@unaffiliated/dfkt)
01:33:19funmanJdGordon: ping
01:33:42funmanis it possible to extend the menuitems in debug_menu.c ?
01:33:55funmanI would like to have either entries with a NULL function
01:34:14funmaneither a pointer to another set of entries that would be displayed together in this menu
01:34:30JdGordonthat should be doable
01:34:33funmane.g. extended_menu which could be NULL or pont to target-specific additional entries
01:35:29JdGordona submenu would be easier i tihnk
01:35:42JdGordonbut what you want is probably doable
01:35:48funmangood idea
01:36:04funmanwe could but at least dbg_ports amd dbg_hw_info in that submenu
01:36:13funmanthe goal is move all the target code back into firmware/target
01:36:30JdGordonall the drawing code needs to stay in apps though
01:36:35JdGordonwell, should stay in apps
01:36:44funmanyeah sure
01:37:10funmanfirst candidate is the only user of debug-target.h
01:37:10 Quit lebellium (Quit: ChatZilla 0.9.88 [Firefox 10.0/20111228055358])
01:39:30 Quit [Saint] (Ping timeout: 255 seconds)
01:41:05funmanwhat does run_debug_screen do ?
01:41:24funmangives a shortcut to a specific entry of the menu?
02:01:13 Join [Saint] [0] (~Saint]@unaffiliated/saint/x-8516940)
02:09:39JdGordonfunman: sounds about right
02:10:36funmanJdGordon: you want to work on this? or give me some pointers?
02:11:03JdGordonbusy at work, so not at the moment :)
02:11:23funmanno hurry i guess
02:11:38funmanthough i'm tempted to mov e that prototype in debug_menu.c
02:12:09funmandreamlayers: didn't Wmissing-prototypes warn about debug_menu.c ?
02:13:57 Join CaptainKewl [0] (
02:14:05 Join The_Partybros [0] (
02:14:50The_PartybrosYo Yo Yo Whats Happin Mahfakaz
02:16:34 Quit The_Partybros ()
02:22:24 Quit robin0800 (Quit: Leaving)
02:25:19Mirso i googled my clip issue today
02:25:29Mirended up finding the log file for this channel
02:25:33Mirit was amusing
02:38:22 Quit dfkt (Quit: -= SysReset 2.55=- Sic gorgiamus allos subjectatos nunc.)
02:39:33 Quit anewuser_ ()
02:39:49 Join anewuser [0] (~anewuser@
02:39:49 Quit anewuser (Changing host)
02:39:49 Join anewuser [0] (~anewuser@unaffiliated/anewuser)
02:41:00 Quit saratoga_ (Quit: Page closed)
02:42:36 Quit amiconn (Quit: - Chat comfortably. Anywhere.)
02:42:46 Join amiconn [0] (quassel@rockbox/developer/amiconn)
02:48:29 Quit TheLemonMan (Quit: WeeChat 0.3.6)
02:59:12***Saving seen data "./dancer.seen"
03:17:23 Quit factor (Ping timeout: 240 seconds)
04:16:35 Quit pixelma (Disconnected by services)
04:16:37 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
04:16:39 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
04:17:56 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
04:17:56 Quit amiconn (Disconnected by services)
04:18:18 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
04:24:58 Join nick_p [0] (
04:28:35 Quit dys (Ping timeout: 252 seconds)
04:29:04 Quit [7] (Disconnected by services)
04:29:15 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
04:29:29 Join dys [0] (
04:37:24 Join nosa [0] (
04:39:23 Quit nosa-j (Ping timeout: 240 seconds)
04:39:24 Nick nosa is now known as nosa-j (
04:59:15***Saving seen data "./dancer.seen"
05:18:25 Join icarusfactor [0] (~factor@
05:26:18 Nick icarusfactor is now known as factor (~factor@
05:40:07 Quit Zarggg (Quit: Rebooting client...)
05:51:15 Join Rob2222 [0] (
05:54:55 Quit Rob2223 (Ping timeout: 240 seconds)
06:10:55 Join Scr0mple [0] (~Simon@
06:13:08 Quit Scromple (Ping timeout: 240 seconds)
06:14:14 Join Scromple [0] (~Simon@
06:16:33 Quit Scr0mple (Ping timeout: 252 seconds)
06:25:49 Quit CaptainKewl (Quit: ( :: NoNameScript 4.22 :: ))
06:28:26 Quit nick_p (Quit: Leaving)
06:53:18 Quit factor (Ping timeout: 276 seconds)
06:59:18***Saving seen data "./dancer.seen"
07:01:44 Quit Keripo (Quit: Leaving.)
07:14:31 Join curtism [0] (
07:14:52curtismdoes rockbox have an option to display time left of a track on the WPS?
07:16:09funmani remember seeing it so very likely.
07:24:36curtismwould it be a configuration option or something in the WPS theme? because i've looked around in the settings and didn't see anything
07:30:52funmandepends on the theme
07:44:21JdGordonits entirely up to the theme
07:44:24JdGordonmost show it
07:45:11JdGordonfunman: buzz me next week about the debug menu thing, i should have time late sunday night (my time) or monday
07:47:12funmanJdGordon: okay thanks!
08:03:38 Join factor [0] (~factor@
08:03:44 Quit factor (Read error: Connection reset by peer)
08:08:23 Join factor [0] (~factor@
08:25:05 Join bertrik [0] (
08:25:06 Quit bertrik (Changing host)
08:25:06 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
08:32:59*[Saint] wonders when bluebrother^ is going to port rbutil to android :)
08:33:17 Quit Scromple (Read error: Connection reset by peer)
08:33:29[Saint]I mean, didn't you hear? PCs are dead :p
08:34:27[Saint]Seriously though...rbutil on an Android tablet/device_with_USB_host would be pretty cool.
08:35:54 Join nick_p [0] (
08:38:24 Join megal0maniac [0] (
08:38:37[Saint]On a kinda related hard/easy would it be to push notifications of updates to RaaAoA?
08:39:56[Saint]Roughly: "Get notified of update, option to 'dismiss' or 'update now'"
08:40:53[Saint]Derp. That's the Market's job isn't, nevermind me.
08:42:06 Quit megal0maniac (Client Quit)
08:47:51JdGordon[Saint]: yep.. well the way it should be done is market updates for the java bits, and have that manually download the on request/automatically so the apk isnt 5MB
08:47:59JdGordonand yes, rbutil to get themes would be ncie
08:48:06JdGordoniirc there is a tracker task to do it
08:49:46[Saint]Not just that...I'd like to be able to plug <DAP> into <android_device_with_USB_host> and do a full installation.
08:51:11JdGordonah, you were asking for android support in rbutil then, which is not really what you said :)
08:51:52[Saint]"port rbutil to android" isn't that?
08:52:19JdGordonthat means "make rbutil run on android"....
08:52:36JdGordonanyway, unless it had adb integrated the best it could do is push the apk to the sdcard
08:52:46[Saint]Yes...still not seeing a problem.
08:53:39JdGordonnot saying there is a problem, just pointless
08:53:47JdGordonthe apk should be smarter and on the market though
08:56:13[Saint]I don't think its necessarily *pointless*'d probably be a lot of work, but I for one would enjoy being able to do full installs and updates, grab themes, make voice/talk clips for my DAPs from my android devices.
08:58:33 Join ender` [0] (
08:59:19***Saving seen data "./dancer.seen"
09:06:03[Saint]rasher: ping?
09:06:31[Saint]rasher: in case you're not aware...your android builds haven't updated in 4 days.
09:06:50[Saint]I suspect you need to update your NDK
09:06:55 Quit bertrik (Ping timeout: 240 seconds)
09:12:17[Saint]rasher: unless I'm greatly mistaken, r31543 broke your automated builds. Though updating the SDK (I said NDK above, in error, apologies) should kick things back to life again.
09:17:55 Join einhirn [0] (
09:21:33 Join User006 [0] (
09:22:11User006Hello after installing rockbox on an ipod video 30 gig is it safe to delete the folders: iPod_Control / Calendars / Contacts etc.
09:22:22User006basicaly delete every folder but .rockbox?
09:36:52 Quit User006 ()
09:38:28 Quit mystica555_ (Ping timeout: 240 seconds)
09:51:04 Join Newguy1 [0] (
09:51:16 Join pamaury [0] (
09:51:17 Quit pamaury (Changing host)
09:51:17 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
10:09:01 Join mystica555_ [0] (
10:09:02 Join p3tur [0] (~petur@rockbox/developer/petur)
10:12:21 Join r4y [0] (
10:15:20r4yI want to buy a cheap 4 to 8 Gigabite mp3 player that is the most compatible with rockbox. I don't want it for video. I have bought from tigerdirect and newegg before. I have only owned an rca mp3 player which will not currently mount that I want to return
10:15:53r4yAlso I want it to be compatible with Linux.
10:17:33r4yI didn't see this link before:
10:18:13[Saint]Have a look at the main page, supported targets are lusted there.
10:18:39[Saint]...just pick something you like.
10:19:31nick_pr4y: Sandisk Sansa Clip+ is good, plus one of the only ones in production
10:19:39[Saint]One of the SanDisk players would probably suit you.
10:20:19r4ySorry, your right, TY
10:20:22[Saint]A ClipV1 or V2, Clip+, or a FuzeV1 or V2
10:21:15[Saint]All of which are cheap and available second-hand or refurbished
10:21:16r4yTY for the link. $50 for 20 GB, wow
10:23:29[Saint]You can just get a 2GB Clip or Fuze, and buy a ~2-4GB (up to 32GB if you want...) micro sdcard
10:24:12bluebrother^[Saint]: I don't see that going to happen (the Rockbox Utility on Android thing)
10:24:35 Join Tuggers [0] (
10:24:41CIA-88New commit by bluebrother (r31590): Remove OS X special universal binary handling. ...
10:25:16[Saint]bluebrother^: I suspected it would be a lot of work... I can dream, though :)
10:25:24r4yLooks like Sandisk Sansa Clip+ is my best bet. Thank you for the help.
10:26:02[Saint]If you don't mind a monochrome screen, its a nice player.
10:26:20r4yI am glad I came here to ask. TY for having an irc channel for rockbox.
10:26:39CIA-88r31590 build result: All green
10:27:28[Saint]Happy to help those that come in here with a great attitude and an interest in Rockbox.
10:28:01r4yI was recently told about it a few weeks ago
10:28:40bluebrother^I want to make Rockbox Utility use lame.dll / directly as next change. Should simplify setup for TTS on Archos devices.
10:28:56r4yI have this rca th1814 which will not mount on Windows and Linux it seems
10:29:14r4yIt still plays music and can hold 4 GB
10:29:15bluebrother^seems there isn't _any_ template / default for encoding voice clips for them :o
10:30:37 Quit GodEater_ (Ping timeout: 258 seconds)
10:31:18r4yWindows said this th1814 mp3 player is set to mtp mode and the easyrip software said it needs to be set to msc mode, but I don't know how to change it, but I would rather get an mp3 player that is more compatible with Linux
10:32:36Tuggersi'm interested in doing google summer of code this year, specifically the audio codec optimization
10:32:42TuggersI have asm experience (MIPS, 6502) but not ARM experience
10:33:11Tuggersdo you know any good guides to learn ARM NEON?
10:34:34Tuggersalso how do you test the rockbox code? do I need to buy an audio player to test code on?
10:35:18[Saint]We do have simulators, but a physical target is best.
10:35:39gevaertsFor optimisation work you'd need a real target, yes.
10:36:04gevaertsI'd like to recommend not spending money on gsoc thins at this point in time yet though
10:37:37gevaertsWe don't know at all yet if we're going to be in (or even if there will be a gsoc 2012, to be complete), so while studying and preparing at this time is fine (knowing things is never useless), but it's *much* too early to really count on things
10:38:19*gevaerts can't answer the ARM questions. Maybe saratoga?
10:38:29gevaertshe's on US time though
10:38:56[Saint]As for NEON, that would only apply to RaaA would it not?
10:41:13Tuggersyeah now that I think about it I bet there's not a lot of music players with Cortex based cpus in them
10:41:17 Join stoffel [0] (
10:41:51Tuggersso let me modify that, any good guides for ARM v4/v5/v6?
10:43:46r4yOK, so then will this one do?:
10:45:00TuggersI read a guide for GBA programming (uses an ARM7TDMI/ARMv4), but I assume the ARMv5/v6 are fairly different in how they're set up
10:48:37nick_pr4y: yes looks good, not in US, but price seems about the same in UK - black 8gb is usually the best deal
10:49:14r4yTY for helping me out. This is going to be great.
10:49:19nick_pr4y: hang on - refurbished? sure you could get it new for that
10:49:35gevaertsTuggers: I'd recommend hanging around for a while. We don't have *that* many specialists...
10:50:27[Saint]r4y: nick_p: this is probably *-community talk now.
10:51:13r4yI am sorry, TY for guiding me.
10:51:36Tuggersgevaerts: OK I can idle
10:52:41[Saint]r4y: No worries, you weren't to know. #rockbox-community is the "stuff not related to Rockbox/development specifically" channel.
10:53:05r4yOK, that's neat
10:54:47r4yI don't have much more to say. Wouldn't refurbished be cheaper. I noticed nick_p isn't there. I am about leave. Sorry for post here after what you just told me.
10:57:28r4yIt's crazy that models come and go so fast
10:57:56[Saint]Depending on the supplier, a refurb is likely cheaper yes. Its probably also cheaper to get, say, a 2GB player and a 6GB sdcard than it is to get an 8GB player as well.
10:59:06[Saint]Have a look around, good deals pop up from time to time.
10:59:24***Saving seen data "./dancer.seen"
11:00:51r4yI would prefer more room not just because I can store more, but becuase more room means more space around the files. From what I read Linux puts space between files where as Windows does not which is why Windows needs to be defragged evenually. But I am guessing that that is way over simplified compared to the true reality of it.
11:02:00r4yI mean tranfering files it would make it easier I would theorize but I know very very little about this
11:04:29 Join y4n [0] (y4n@unaffiliated/y4ndexx)
11:07:57r4yI mean very little.
11:12:59r4yThank you all, sorry for posting almost useless info. Take care.
11:13:35r4yThank you Saint and Nick_p
11:14:00r4ybye :) ,I didn't know this was the devs channel
11:14:33 Part r4y
11:56:58 Join n1s [0] (
11:57:00 Quit n1s (Changing host)
11:57:00 Join n1s [0] (~n1s@rockbox/developer/n1s)
12:00:02 Join liar [0] (
12:02:42 Quit stoffel (Ping timeout: 260 seconds)
12:16:53 Quit n1s (Read error: Connection timed out)
12:17:40 Join n1s [0] (~n1s@rockbox/developer/n1s)
12:27:37 Join stoffel [0] (
12:29:12 Join TomColler [0] (
12:44:28 Join TheLemonMan [0] (
12:45:48 Join prof_wolfff [0] (
12:47:20prof_wolfffhi all, i am using Rockbox on an ipod classic 80Gb for about 3 moths, its great!!!, TY to all people making it possible
12:48:16prof_wolfffthere was a few issues on this port, so downloaded the source code and made some patches mainly based on nano2g port
12:49:29prof_wolfffnow i am very happy with my ipod, the running time is doubled and it seems more stable (normal use, no exhaustive testing)
12:50:46prof_wolfffi think maybe other people could use and test it on other classic models
12:51:43prof_wolfffbut i dont know how to post the patches, just resgister and post?
12:52:32bluebrother^post them to the tracker. Or even better, discuss them first. Either here or on the mailing list
12:56:01 Quit nick_p (Quit: Leaving)
12:56:30prof_wolfff one patch fixed the L-R channels that were inverted on my ipod
12:58:10 Part TomColler
12:58:33prof_wolfffported some code from nano2g source and now the framerate is also doubled tweaking the PHTIME register like nano2g does
12:59:25***Saving seen data "./dancer.seen"
12:59:27[Saint]Define "fixed", does it just swap the channels...or is it configurable?
13:00:11[Saint]With the stock Apple phones, I believe the channels are correct.
13:00:18prof_wolfffit configures the I2S register to invert the clock signal
13:00:23bluebrother^hmm, make voice tries to run the cross compiler?
13:00:35 Join rarog [0] (
13:00:43pixelmabluebrother^: for the features, no?
13:01:34[Saint]prof_wolfff: yes...but, is it linked to a menu setting? (The "right" way to do this IMO)
13:01:44rarogpamaury: hello, I'm currently looking into the problem of windows showing only one drive (usually internal) and I have some test results from a diagnostic tool. is there an open bug for this?
13:01:51prof_wolfffalso cpu scaling is working on my ipod, and lcd sleep for lcd_display type 1
13:01:58bluebrother^pixelma: not sure. I thought features.txt uses the hosted cpp instead.
13:02:07bluebrother^well, building the sh compiler for now ... let's see
13:02:14pixelmaprof_wolfff: TheSeven knows most about the newer Ipods, maybe try to get his attention (consider him pinged now ;) )
13:02:27rarogor should I just add it to FS #12458
13:02:28fs-bluebot [Fuze+] The SD card is not reported in usb mode (bugs, new)
13:02:40[Saint]He's on holiday I think
13:02:58prof_wolfffi think the i2s setup was wrong, CPU was sending one format and the CODEC expecting other
13:02:59[Saint]TheSeven I mean.
13:03:06bluebrother^yay. binutils doesn't build on this machine :(
13:03:47TheSevenprof_wolfff: sounds like you volunteer to maintain these ports? :)
13:03:49pixelmaand for what it's worth, I don't agree with [Saint]'s view on the right fix
13:04:19TheSevenseriously, not many people are digging that deep to get stuff fixed...
13:04:46prof_wolfffi own a classic and i can do some programing / testing
13:05:08TheSevenand you apparently have no fear of reading datasheets and poking at HW registers :)
13:05:17TheSevenwhat exactly did you change?
13:05:33TheSevendid clock scaling work out of the box when you enabled it?
13:05:54TheSevenhow much did that impact battery runtime?
13:06:11prof_wolfffI2STXCON = 0xb100859 instead 0xb100059
13:06:58TheSevenso that's swapping LRCK polarity on the SoC sideß
13:07:03prof_wolfffcpu scaling was very unstable but getting the pmu polling out of the ISR seems to work
13:07:22TheSevenhuh, which ISR?
13:09:07prof_wolfffthe holdswitch routine was polling the pmu (using i2c), running from timer interrupt
13:09:26TheSeventhat explains a lot of trouble
13:09:31TheSeveni2c isn't irq-safe at all
13:09:52TheSevenso the hold switch detection is run from a tick task!?
13:10:25prof_wolfffyes, i get out and put it on the powermgmt thread and now moving the clickwheel dont break
13:10:48*TheSeven falls off his chair
13:11:00TheSeventhis might well be the bug i was hunting for for *months*
13:11:16prof_wolfffon other ipods holdwitch can be detected reading an GPIO but on classic port it is done reading the pmu
13:11:17TheSevenit's always the most trivial things...
13:11:35TheSevenyeah, i didn't find a gpio that reflected the hold switch state
13:11:47*bluebrother^ helps TheSeven getting up again
13:11:50TheSevenand it seems like at least diagmode reads it from a pmu gpio as well
13:12:13pamauryrarog: can't remember, I'm not sure, in doubt open one and we'll merge or redirect if it's a duplicate
13:12:16prof_wolfffi also search for the GPIO port, but cant find it
13:12:38pamauryrarog: ah indeed, FS #1245 is fine
13:12:38fs-bluebot Resume playlist in the same way as if you rebooted (feature, closed)
13:12:52pamauryoops, FS #12458
13:12:52fs-bluebot [Fuze+] The SD card is not reported in usb mode (bugs, new)
13:13:25TheSevenhm, looks like the same might happen with charger_inserted() as well
13:13:36TheSeventhat's run from the USB tick task and is probably doing I2C as well
13:13:48rarogpamaury: ok, as far as I see, almost everything is set identical by OF and Rockbox, but yet some values are different and I'm not sure if they are relevant to the problem.
13:14:46pamauryprobably not but reported everything you can. If you hide the first drive, the sd card is 100% reliable ?
13:14:46prof_wolfffon my patch charger inserted is reading GPIO, so i think it is not dangerous, it also detect external adapter
13:15:04*TheSeven is impressed
13:15:11TheSevenhow on earth did you find that bug?
13:15:44prof_wolfffTY, i am more impressed with all the Rockbox work
13:15:47*TheSeven still isn't sure what it has to do with clock scaling, but it would explain a lot of random lockups at least
13:16:36prof_wolfffwhen lowering cpu the i2c transfer takes more time, so the probability of collisions increases, i think that is the reason
13:17:54rarogyes, then the sd card shows up.
13:18:17rarogand I can read and write the data.
13:19:04 Join robin0800 [0] (
13:19:49rarogI updated the bug with the report.
13:19:59TheSevenhm, if everything works right, the I2C transfer time doesn't change much, it's more bound to the I2C clock which shouldn't change.
13:20:35TheSevenif it does though, that could explain some more things
13:20:46rarogrockbox seems to be sending hub/port data "Hub 4, Port 6", while OF doesn't do this at all. rockbox also doesn't set any serial number, but this should be irrelevant I think.
13:20:49prof_wolfffto catch the bug takes me days, i begin loking at the clickwheel functions, it conduces me to i2c section
13:21:02TheSevenmaybe an IRQ gets in between the two clkcon writes?
13:21:15TheSevenwe might want to do an IRQ lockout in that function
13:21:31gevaertsrarog: hub and port data?
13:22:02prof_wolfffthere is a mutex on the i2c_read/write, using it into an interrupt maybe it can get looked forever
13:23:24raroggevaerts: yes... I'm not completely sure, what this means, but this is reported by the diag tool, I used.
13:23:43funmanclassic works => new usb-s3c6400x target?! \o/
13:24:11gevaertsrarog: that's data that's not transferred over USB at all
13:24:56rarogbut it's reported.
13:25:09prof_wolffffor cpu scaling i am dividing by 2 everything (bus and cpu clocks), then the i2c transfers takes me double time (40us @216MHz vs 80us @108MHz), but that clock can be indevidually doubled when unboosting
13:25:16gevaertsnot by rockbox
13:25:32rarogbut they are only shown if I connect to usb with rockbox.
13:25:49rarogthey are not reported by diag tool if I connect usb with OF.
13:26:46rarogpamaury: ok, I have another idea, why this might fail while looking into windows registry. I have different keys for rockbox ##?#USBSTOR#Disk&Ven_Rockbox&Prod_Internal_Storage&Rev_0.00#6&2e474e3a&0#{53f56307-b6bf-11d0-94f2-00a0c91efb8b} and ##?#USBSTOR#Disk&Ven_Rockbox&Prod_SD_Card_Slot&Rev_0.00#6&2e474e3a&2#{53f56307-b6bf-11d0-94f2-00a0c91efb8b} which was created only when I connected with...
13:26:48rarog...internal drive hidden.
13:26:49gevaertsOne possible reason could be that the diag tool (or the OS) uses hub and port as a unique identifier if there's no serial number
13:26:51prof_wolfffi also lowered the cpu voltage to 1.1V and it looks stable
13:26:52 Quit robin0800 (Read error: Connection reset by peer)
13:27:49pamauryrarog: ok, try the following: plug your device, open the device manager and uninstall it. Then unplug and replug
13:27:59rarogWith OF there seem to be two keys: ##?#USBSTOR#Disk&Ven_SanDisk&Prod_Sansa_Fuze+&Rev_0100#000000010000120B0002DCDA908B920B&0#{53f56307-b6bf-11d0-94f2-00a0c91efb8b} and ##?#USBSTOR#Disk&Ven_SanDisk&Prod_Sansa_Fuze+&Rev_0100#000000010000120B0002DCDA908B920B&1#{53f56307-b6bf-11d0-94f2-00a0c91efb8b}
13:28:24rarogusingsame name but with two enumerations...
13:28:30rarogI can try this.
13:28:45pamaurygevaerts: do we report the serial number ? What needs to be implemented for this to work ?
13:29:23gevaertspamaury: you need an implementation of set_serial_descriptor() for your platform in usb_core.c
13:29:54bluebrother^hmm. Why does Rockbox Utility use a Quality value of 8 for rbspeex while make voice uses 4?
13:30:00pamaurythere is not generic mechanism for this ? SD cards all have a serial number of something no ?
13:30:31gevaertsThat would work for internal SD, yes
13:31:01raroguninstalling and reinstalling just brings back one device.
13:31:04bluebrother^seems there is quite a bit of work to do :/
13:31:14gevaertsWe have PP ipod, AS3514 (i.e. e200v1 and c200v1, maybe some more), and ATA
13:31:16rarogdepending on the hiding setting the internal storage or the sd card.
13:31:27gevaertsAdding more semi-generic methods like sd should work
13:32:03pamaurybut that's more a problem of windows trying not to reprobe by caching the device configuration.
13:32:08pamauryrarog: so does it work now ?
13:32:30rarogno, still only one device at time with rockbox, not both drives.
13:33:00pamauryeven after uninstalling and windows redoing the installation ?
13:33:10n1sgevaerts: doesn't the as3514 have a unique serial or something too?
13:33:24gevaertsn1s: yes, that's why I mentioned it :)
13:33:30n1saha :)
13:34:35 Join robin0800 [0] (
13:34:41pamauryI'll try to see if the imx233 has such a mechanism, or I'll compare see what the OF does
13:35:17rarogwhat I could try would be deleting also all settings from CurrentControlSet in the registry to have again just really clean settings.
13:35:25TheSevenprof_wolfff: you should really submit that to flyspray... this looks like it may well be a major improvent for that port :)
13:35:36rarogand then copying the registry settings.
13:36:41prof_wolfffi think so, i will try todo that this weekend, i am getting 22h30m running time while playing flac
13:37:01TheSevenhm, this is getting close to what apples reaches :)
13:37:54pamauryapparently the imx233 has support for a unique id in hardware but it's not clear where it is stored
13:38:34TheSeventhe fun fact is that apple doesn't do clock/voltage scaling at all
13:39:44prof_wolfffi dont know that!, i was thinging they do scaling, how can it be explained?
13:40:13TheSevenheh, no idea
13:40:36pamauryrarog: does the OF report a serial number ?
13:40:44prof_wolffflcd sleep saves a lot of power on my ipod, but scaling much more
13:41:02rarog000000010000120B0002DCDA908B920B is the serial number reported by it.
13:42:05pamauryrarog: do you have windows or a virtual machine running windows ?
13:42:14rarogReal windows.
13:43:25pamaurydo you know the BitBurner tool from Freescale ?
13:43:26prof_wolfffi am very nervous with the volumen when listeing to music, to save power i include an option on the settings menu to not to awake lcd (it its sleeping) when touching the screenwheel
13:44:38rarogNot yet.
13:45:20 Join TomColler [0] (
13:45:43pamauryok, do you have a Freescale account to download it ? It's there:, it's in the IMX_OTP_TOOLS
13:46:01pamauryotherwise I can send it to you by mail. It allows to read the OTP bits
13:46:52rarogNo freescale account, so mail would be better.
13:46:55[Saint]prof_wolfff: all ipods with LCD sleep could do with that, IMO.
13:47:05[Saint]That sounds like a nice idea.
13:47:44[Saint]"Don't wake on volume change" I mean, in case it was non-obvious.
13:48:20rarogI attached the created registry settings, though I don't know if they are show only the symptoms or also give clues to the original problem.
13:48:22prof_wolfffall hardware that HAVE_LCD_SLEEP and HAVE_SCROOLWHEEL support the option (ipods and sansas)
13:48:58[Saint]Ah, right. Forgot about the sansas. Thanks.
13:49:18 Part TomColler
13:49:27[Saint]Yes, very nice idea.
13:49:55prof_wolfffand for me it saves lots of power :)
13:51:18[Saint]I could be wrong...but I'd be willing to say that has a high chance of being committed.
13:51:36[Saint]I would certainly like to see it go in.
13:51:58pamauryprof_wolfff: that looks interesting. Actually, not waking up the screen on volume update with any device that has physical volume keys sounds interesting
13:52:20*[Saint] nods.
13:52:40prof_wolfffmy current patches are based on an 29xxx version i will try to update them to the las release and post it as soon as i can
13:52:46[Saint]One of those "why did no one think of that before?" kinda things :)
13:55:28prof_wolfff:) well.. i was a pain for me to touch the clickwheel, the patch was very simple so i do it, latter i added the option to the menu
13:56:14funmanpamaury: physical volume keys? you mean as opposed to scrollwheel?
13:56:21funmanor touchpads/screens
13:57:33pamaurysome device don't have a physical volume keys. For example the not-yet-rockboed-creative zen xfi2, only a touchscreen
13:57:46prof_wolfffit can be done but not with this patch, the work is done on the clickwheel section, but later it can be moved to the button one
14:00:26TheSevenprof_wolfff: if we figure that out what apple is doing differently, the clock scaling advantage will be the bonus runtime :)
14:00:37TheSeventhat's the 50+ hours part :)
14:01:56prof_wolfffyes, cant imagine what is the 'trick' but it is a very interesting thing
14:02:34TheSevenpossibly some still unknown sleep states of some components of the SoC
14:03:42TheSevenmeasurements show that the OF (sitting in the main menu) consumes less power than emCORE with zero CPU load (the CPU just waking up for a couple of microseconds every ~10 seconds to check for work)
14:05:55prof_wolfffthats is amazing, the datasheet mentioned something about an sleep state at 32KHz intended for displaying the clock screen
14:11:05 Quit rarog (Quit: ChatZilla 0.9.88 [SeaMonkey 2.5/20111121045514])
14:25:08 Quit stoffel (Ping timeout: 240 seconds)
14:30:48 Join TomColler [0] (
14:38:20 Join dfkt [0] (dfkt@unaffiliated/dfkt)
14:59:27***Saving seen data "./dancer.seen"
15:09:02 Part TomColler
15:10:21TheSevenprof_wolfff: yeah, but you can't possibly feed i2s data at those clock speeds
15:10:50TheSevenwhile audio is playing you'll have to keep the AXI/AHB/APB/whatever clocks at sane speeds
15:11:12topik[Saint]: sounds a bit like FS #9305
15:11:13fs-bluebot Context sensitive backlight on key press (patches, unconfirmed)
15:11:17TheSevenand AFAICT apple keeps the AXI clock at >200MHz all the time
15:12:02*TheSeven wonders if they possibly shut down the DDR3 memory controller and play audio from an IRAM buffer
15:27:13 Quit robin0800 (Quit: Leaving)
15:29:19 Quit dv_ (Ping timeout: 240 seconds)
15:30:43 Join dv_ [0] (
15:40:02prof_wolfffTheSeven: maybe menu browsing can be handled using only interrupts (clickwheel, pmu...) to awake the CPU for the time necesary to process the event
15:42:13prof_wolfffbut the importan thing is while playing music, its not probable apple is disabling DDR while they dont do cpu scaling, the IRAM seems too small and the HDD spinup/down will probably eat the battery
15:45:17prof_wolfffone thin i can imagine at this moment is using IRQ driven DMA and some dedicate HW to do the work while the CPU is sleeping and the buses are at full speed, the support only a few codecs, maybe the s5l7002 have some kind of DSP (7000 does) and the audio playback is done by hardware, awakening the cpu only on keypress???
15:47:10 Join jlbiasini [0] (
15:58:29jlbiasinibagder: and anyone with acces to server. Now we have Rockbox Utility handling mkimxboot. pamaury already uploaded a tagged bootloader file that is ready to go on the server. It is here:
16:08:02 Join domonoky [0] (~Domonoky@rockbox/developer/domonoky)
16:11:36prof_wolfffTheSeven: sorry, you talk about disabling ram *controller* not the ram itself, i think its a great idea, i wonder if the ram can be put on some 'sleep' state while the controller is disabled
16:26:34 Quit p3tur (Quit: Bye)
16:56:29 Part [Saint] ("I know its a sin to kiss and swallow.")
16:59:30***Saving seen data "./dancer.seen"
17:15:06 Join kadath [0] (
17:16:14pamauryTheLemonMan: if you are interested, I've written a program that generates a nice PDF with the differences between the stmp3700 and stmp3780 registers
17:16:55TheLemonManpamaury: you mean imx23 ?
17:17:13pamauryyeah, stmp3780=imx233
17:18:00TheLemonManoh cool, send it :)
17:18:16 Quit einhirn (Quit: Miranda IM! Smaller, Faster, Easier.
17:19:10TheLemonManthe thread switcher goes nuts here, i suspect its a mmu problem but im not sure about how to debug it
17:20:28CIA-88New commit by jethead71 (r31591): Coldfire: lcd_update call got dropped from dbg_hw_info. Replace.
17:22:26CIA-88r31591 build result: All green
17:23:52 Quit amithkk (Remote host closed the connection)
17:25:48 Join benedikt93 [0] (~benedikt9@unaffiliated/benedikt93)
17:26:01 Join asper [0] (5d2b231f@gateway/web/freenode/ip.
17:26:06asperhi guys
17:26:08asperheppy 2012
17:26:27asperI would like to know if someone has the HEX file for the REMOCLONE hardware
17:26:44asperI am able only to find schematics in the rockbox homepage
17:28:28 Quit liar (Ping timeout: 252 seconds)
17:29:42pamauryTheLemonMan: it's not perfect, here is the result:
17:31:21pamauryI can't remember who are the others working on an stmp3700, there is possibly low_light but I've not seen him in a while
17:33:00TheLemonManmy bsp code is not yet ready for merging but i can upload it to my github to share the efforts
17:34:06pamauryyes that would be nice, although I would like to reuse as much as the existing code as possible of course
17:35:01TheLemonManwell my code is just the imx233 folder with the address/bitfields adjusted
17:35:09TheLemonManand the gpmi module i'm slowly working on
17:36:19pamauryI noticed there are a few nasty differences though, for example the power subsystem is quite different when it comes to charging it seems
17:37:50saratogapamaury: now that USB is working on AMSv2, how hard would it be to adapt the USB audio driver to it?
17:37:50TheLemonManwell thats a pretty far point in the roadmap heh
17:38:11pamauryand the stmp3700 uses the apbh/x dma while the imx233 has a dedicated dma for a few peripherals like lcdif
17:39:55pamaurysaratoga: I already tried the conversion with the old driver but it didn't work for some reason, in its current state it requires quite a rewrite of the driver plus I don't really know how isochronous transfers work on the AMSv2
17:40:16saratogahmm that doesn't sound promising
17:40:37pamauryIt's not hard though, the first step would be to convert the driver to the new api
17:40:44saratogai'm curious after seeing some tests of various external sound cards showing most of them losing to the clip+
17:43:03pamauryI'll see what I can do but someone who worked on the driver recently would probably do the work quickier (at least for the conversion part). Perhaps funman is motivated :)
17:43:28saratogaif i ever get free time again, whats a good way to learn about USB?
17:44:42pamauryreading the code, read usb in a nutshell, read the spec, try to write/change a driver (like the serial one), look at usb drivers (like linux ones)
17:45:16pamauryunfortunately that would also require to sync my changes with current HEAD, I don't think that's difficult, it just requires more time
17:47:52 Quit asper (Quit: Page closed)
17:51:03pamauryI'll see what I can do, I understand this feature might have gained some interest with the ams players
17:51:16pamauryand with the fuze+ also
17:53:33saratogapreviously it was fairly pointless on the available software USB targets
17:54:02saratogathe newer ones are actually pretty respectable vs. < $200 external USB devices
17:57:29 Quit pamaury (Remote host closed the connection)
17:57:44TheSevenprof_wolfff: yes, the ram can be put into self-refresh mode
17:58:15TheSevenand it is very well possible to play audio from IRAM and keep the DDR controller disabled most of the time
17:58:40TheSevenand wake up, say once a second, to let the codec decode some more audio from ddr (mp3) to iram (pcm)
17:59:04TheSeveni doubt that they have dedicated decoding hardware, decoding audio doesn't really eat much CPU on these modern ARM cores
18:00:46saratogafor that to be worthwhile, you would have to keep the CPU clocked fairly high or else have very little cost to going into and leaving self refresh mode
18:00:56bluebrother^is voice not working in the sim for HWCODEC targets?
18:01:07*bluebrother^ guesses not
18:04:15 Join amithkk [0] (u4289@2buntu/writers/amithkk)
18:07:31jlbiasinihello all! in something like: "pluginlib_getaction(HZ/2, plugin_contexts, ARRAYLEN(plugin_contexts))" HZ/2 is the prequancy of testing button, isn't it?
18:11:56jlbiasinibluebrother^: could you commit my patch on the rockbox manual? It is incomplete (only half of it has been writtenand readed, until end of plugins game) but would like to see what I could improve in plugin before going and writing stuff that I would rewrite otherwise... And I would like to avoid correcting my patch if other patch are done in the meantime....
18:14:23bluebrother^jlbiasini: you're talking about FS #12492?
18:14:24fs-bluebot add fuze+ manual (patches, unconfirmed)
18:15:02jlbiasinithe file that are in the will not get further edited
18:15:27bluebrother^I was about to leave in a minute, will give it a look in a couple of hours
18:15:39jlbiasiniexcept if I rewwrite all plugin with PLA but we are talking about in 10 year! :D
18:22:20 Join lorenzo92 [0] (~chatzilla@
18:22:48 Quit factor (Ping timeout: 240 seconds)
18:26:54 Quit lorenzo92 (Ping timeout: 240 seconds)
18:28:16 Quit tchan (Read error: Connection reset by peer)
18:29:30prof_wolfffTheSeven: yeah!, it sounds very well, and while the CPU is decoding the data the DMA can update the LCD when needed, really it should first update the framebuffer and then launch the DMA, and/or use the scrolling capabilities on LCD controller for some regions of the screen
18:30:19 Join tchan [0] (~tchan@lunar-linux/developer/tchan)
18:31:07 Join liar [0] (
18:33:33prof_wolfffthe LCD DMA is not working very well for me, after changing the PHTIME register y get 50 fps aprox using the DMA chain and about 80 fps disabling DMA and writing on the DATA register like the nano2g does
18:37:23 Join lorenzo92 [0] (~chatzilla@
18:39:40prof_wolfffi want to try to use DMA but not chaining all the framebuffer data, instead i was planning to enable the LCD FIFO interrupt and launch a small DMA transfer everytime the FIFO is half empty, i dont know if it will work
18:44:58TheSeventhat won't be better
18:45:47*TheSeven wonders how the LCD driver currently works
18:45:49 Join stoffel [0] (
18:46:46 Join factor [0] (~factor@
18:47:32TheSevenprof_wolfff: DMA is slower because the data needs to be transferred twice
18:48:28TheSeventhat's a design issue within rockbox - if the UI layer would support double buffering natively this would be faster than PIO
18:49:08TheSevencurrently the LCD driver has to keep a copy of the current screen contents around because they might change while the DMA transfer is running
18:50:11TheSevenit could possibly be more efficient to just block the calling thread until the DMA transfer completes and let DMA push the data directly do the LCD, but this will only work for full-width updates (otherwise you would need a much more complex DMA descriptor chain)
18:51:24 Nick kugel is now known as kugelp (~kugel@rockbox/developer/kugel)
18:51:37kugelplorenzo92: Hey
18:51:46lorenzo92kugelp: hey ;)
18:52:15kugelphave you played with micrsosd yet?
18:52:30lorenzo92kugelp: I was away :)
18:52:35 Join pamaury [0] (
18:52:35 Quit pamaury (Changing host)
18:52:35 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
18:52:53lorenzo92kugelp: but now I wanted to try to use gpio module for keypad reading hehe .)
18:53:45TheLemonManpamaury: what i had to do for running w/o the mmu ?
18:54:32prof_wolfffyes, i tried that, i made a chain composed by lines on framebuffer memory (no memory copy and thread blocked), it works but the transfer time was the same, then i measured the time that is spent in the copy from framebuffer to RAM and it was a surprise for me that this time is about 1/10 of the DMA transfer (dont remember the ratio exacly but i annoted it somewhere)
18:57:10kugelpthere's a mount system call so should be simple
18:58:41 Quit anewuser (Read error: Connection reset by peer)
18:58:51kugelpcould additionally bind mount into the internal memory for convenience
18:59:31***Saving seen data "./dancer.seen"
18:59:40kugelpsince I expect everyone sets the start file browser setting to the internal memory
18:59:40 Quit liar (Remote host closed the connection)
18:59:58lorenzo92kugelp: yes, should be easy...where do you think is better to place this mounter? button stuff?
19:00:08lorenzo92well maybe sd mount routine in system.c
19:00:35lorenzo92kugelp: a question for multiple keypresses, then. How to "combine" keys?
19:00:40kugelpdon't know really :-)
19:01:27prof_wolfffTheSeven: i supposed the problem could be a misconfiguration on DMA controller or something similar... i discover rockbox a few monts ago and i really dont know very much about the ipod hardware
19:01:29kugelpneeds a separate bit for each button
19:01:55kugelpso you need to replace the enum
19:02:31kugelpotoh imo it works surprisingly well without. I don't miss combos yet
19:02:40jlbiasinihello i'm playing with PLA and lamp but in "if(!(button & (BUTTON_REL|BUTTON_REPEAT)) && !IS_SYSEVENT(button))" what is this !IS_SYSEVENT(button) fonction (obviously some test about those button not being mapped) but is there a way to have PLA button also detected?
19:03:40jlbiasiniI din't find much in the code about this IS_SYSEVENT
19:05:01lorenzo92kugelp: uhm I guess you're right. I implemented some buttons, and *seems* that just returning the specific key does the trick, don't know if it's possible. Need to check with a gameboy rom hah
19:06:08kugelpjlbiasini: system events are communicated the same way as button presses
19:06:37kugelpwith this macro you can tell if its a system event or button press
19:06:44 Join Brijesh__ [0] (Brijesh@
19:07:00jlbiasinilorenzo92: usually you put a | between the key to combine
19:07:13lorenzo92uhm okay thanks ...
19:07:26kugelpthat's bit-or :-)
19:07:48jlbiasinion keymaps or means and!!
19:09:08lorenzo92kugelp: at least using gpio buttons are working , just need to enable multikeystroke ;9
19:09:37 Nick kugelp is now known as kugel (~kugel@rockbox/developer/kugel)
19:10:15prof_wolffftheSeven: so i left that issue, my final tought was that the max DMA transfer speed is limited by the LCD parallel port controller speed but for some reason the DMA cannot reach that reltivelly low speed
19:13:35TheSevenwell, the CPU has to do a lot of copying before the DMA transfer even starts, which can't happen in parallel with the previous transfer
19:13:58TheSevenso the DMA will actually only be active for like 60-70% of the time (estimated)
19:14:12TheSevenwhich matches the 50 vs. 80 fps you're getting
19:15:01pamauryTheLemonMan: basically disable the mmu in crt0.S
19:15:24pamauryand tweak some macros that give the uncached/buffered address
19:15:26 Quit lorenzo92 (Ping timeout: 252 seconds)
19:16:03TheLemonManpamaury: ive set all the address in the cpu config.h to 0x40000000 and commented out the mmu_init call
19:16:16 Nick kugel is now known as kugelp (~kugel@rockbox/developer/kugel)
19:16:24TheLemonManbut hangs at boot
19:19:43pamauryyou need to disable cahing also
19:20:00pamauryand buffering
19:21:22TheLemonMancrt0 already does so, memory_init re-enables them
19:21:35prof_wolffftheSeven: yes, but i measured all the phases and i get for a whole frame (320x240): memcpy(dbl_buf, frambuffer) about 1900 usecs and the DMA transfer time about 22000 usecs, they were a very surprising values for me, maybe the measurement is not correct for some reason, i will look again when i can
19:22:26TheSevenyou need to include the cache flush (somewhere inside the DMA function) into the memcpy measurement, not the DMA one
19:24:05TheSevenalso there may be ticks and other IRQs interfering with these measurements
19:24:09prof_wolfffi dont remember but believe the clean_dcache() was relativelly fast, must give a look to it again...
19:24:14Newguy1I deleted all my music that was put on through iTunes and added it just by folders and now rockbox shows the "searching database" for a second of 2 every time i click on an artist and i only have 3 albums
19:24:24TheSevenit probably takes a while for the rockbox kernel to wake the thread again after the DMA IRQ signals the wakeup
19:24:55Newguy1how do I make it not have to "load" every time i go into a cd
19:25:05TheSeventhis might happen only during the next tick, which can delay it by up to 10 milliseconds, or even more if other threads are ready for execution
19:25:42 Quit amithkk (Write error: Broken pipe)
19:29:33 Join lebellium [0] (
19:30:39Newguy1Also is there any way to make my albums show in database but not under Files>
19:31:20saratogayou can set the view settings to now show music IIRC
19:32:07Newguy1okay ill check that out, and another thing i added just 3 albums to test it out
19:32:23Newguy1and all of them have Folder.jpg in the folder of the album
19:32:31Newguy1and one of them doesnt show the album art on the iPod
19:32:34Newguy1any reason why?
19:32:34 Quit stoffel (Ping timeout: 252 seconds)
19:32:53Newguy1Only thing i can see different is the 3rd one is 600x600 and the other 2 are 500x500
19:33:56prof_wolfffTheSeven: opps i didnt account the awake time, but i continue thinkinig something strange is happening on DMA, when i lowered CPU (108Mhz) these times just doubled while the tick interval is the same (100Hz)
19:34:22TheSevendid you only lower the CPU or also AXI/AHB clocks?
19:34:30TheSeventhe CPU clock shouldn't affect DMA at all
19:34:45prof_wolfffi lower CPU and buses
19:35:11Newguy1is there any way i can make my ipod video battery last longer, it only lasts about an hour
19:35:20prof_wolfffbut modified the timer for using an external source not affected when change the cpu/bus speeds
19:39:36 Join bluebrother [0] (
19:39:36 Quit bluebrother (Changing host)
19:39:36 Join bluebrother [0] (~dom@rockbox/developer/bluebrother)
19:42:34 Quit fs-bluebot (Ping timeout: 240 seconds)
19:43:06 Quit bluebrother^ (Ping timeout: 255 seconds)
19:44:08 Join fs-bluebot [0] (
19:45:18 Join amithkk [0] (u4289@2buntu/writers/amithkk)
19:50:46kugelpTheSeven: the kernel can wake threads before the next tick
19:51:10TheSevenwhen does it do that?
19:51:27TheSevenin this case it's a wakeup triggered inside an IRQ handler, probably while the CPU was sleeping
19:51:35kugelpwhen an irq comes in
19:51:44 Quit factor (Read error: Connection reset by peer)
19:51:48TheSevenNewguy1: replace the battery
19:52:23kugelpthe kernel enters core_sleep() which puts the kernel into wfi state
19:52:32TheSevenif you want to squeeze the last bit out of it, reduce backlight brightness and time, volume, enable lcd sleep, disable accessory power and lineout, ...
19:52:55TheSevendoes it attempt a thread switch after waking up from one of those?
19:52:55kugelpif its interrupted it checks the threads again
19:53:11TheSevenbut if it was previously running a thread it doesn't?
19:53:15kugelpit does that for all interrupts, not only the tick
19:53:50TheSeven(otherwise that would defeat the cooperative multitasking concept)
19:53:51kugelpin fact it doesn't distinguish which source woke up the CPU
19:55:06prof_wolffftheSeven: i noticed there is a new ipod classic release based on rockbox r31516 and emcore r859, i am using an old release so i need to update it, what do you think it is better: to post the patches applied to this release or to use the last trunk release?
19:55:13kugelpif its running a thread its not in core_sleep, it checks the other once it yields
19:55:28TheSevenprof_wolfff: if possible to the trunk
19:55:31 Join bertrik [0] (
19:55:31 Quit bertrik (Changing host)
19:55:31 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
19:58:43 Join lorenzo92 [0] (
19:58:49 Join factor [0] (~factor@
19:58:56 Quit factor (Read error: Connection reset by peer)
19:59:32TheLemonManso, i disabled the mmu and i still get a random swi error originated from find_first_bit_set call in thread_switch
20:00:08 Join factor [0] (~factor@
20:00:09TheLemonManand theres no such swi call there
20:00:54kugelpdoesn't swi use the same fault handler as undef insn and prefetch abort now?
20:02:13TheLemonManit still uses UIE
20:09:37jlbiasinikugelp: PLA button seems to be detected as system event, is there a way I could take them out?
20:10:03TheLemonManinteresting fact, without the mmu enabled i finally get a successful ecc8 irq even if after i ack it the screen wraps around and the player hangs
20:10:14TheLemonManseriously, what the fuck is going on ?
20:11:07pamaurywhy do you disable the mmu ?
20:11:55kugelpjlbiasini: pla should take care of system events
20:12:11kugelpI don't think you need to do anything
20:15:03TheLemonManpamaury: ive tried to put all the blame onto the mmu as those problems look like memory corruption to me
20:15:54pamaurythe mmu setup is really simple, it's unlikely that it will cause any problem. More likely you gave the address a virtual address for example
20:17:33TheLemonManthat doesnt explain the missing irq or the swi error in thread switching
20:18:04jlbiasinikugelp: well I could replace the line by a map to PLA_QUIT for example...
20:18:41jlbiasiniif I take the line out there no more way to quit lamp
20:19:25kugelpwhich line?
20:19:34CIA-88New commit by bluebrother (r31592): Split up encoders sources. ...
20:20:15 Join lovasoa [0] (~olojkine@
20:20:16kugelpremove only the sysevent part, pla of course still delivers the quit button
20:21:00jlbiasiniI think I already tried that wait I will redo
20:21:25lovasoaWhy isn't the screen updated after a splashf times out? This causes a disgusting half-drawn message in the WPS for me...
20:21:29CIA-88r31592 build result: All green
20:26:22 Join tchan1 [0] (
20:27:00CIA-88New commit by bluebrother (r31593): Fix "invalid settings" dialog appearing twice. ...
20:28:46CIA-88r31593 build result: All green
20:29:03 Quit tchan (Ping timeout: 276 seconds)
20:29:30CIA-88New commit by bluebrother (r31594): MSVC: fix error in README and update librbspeex build folder.
20:30:25 Join Zarggg [0] (
20:30:40jlbiasinikugelp: ah yes now I remember the problem is then that lamp now quit on every long click or release (BUTTON_REL|BUTTON_REPEAT) because in "if(!(button & (BUTTON_REL|BUTTON_REPEAT)) && !IS_SYSEVENT(button))", !IS_SYSEVENT(button) prevent quit when BUTTON_REL|BUTTON_REPEAT are button call
20:31:23CIA-88r31594 build result: All green
20:31:46bluebrotherjlbiasini: your manual patch changes the blackjack plugin. Is that intentional?
20:33:02jlbiasinioups wait I have to lool at it
20:34:04bluebrotherI'm planning to commit the plugin changes separately anyway so no hurry
20:34:31bluebrotherurgh −− nopt will not work with more than one option
20:35:01jlbiasinibluebrother: yes and no: this is because I was doing the plugin remapping in the same time for pamaury
20:35:12jlbiasiniSo It get in the wrong patch
20:35:54bluebrotherjust see the diffstat in the tracker −− first file
20:36:01jlbiasiniI tested the compilation, it should work and display ok (I read all those modification)
20:36:27bluebrotherand why are you removing that mpiohd200 section from installation.tex?
20:37:56kugelpjlbiasini: show your diff please
20:38:00jlbiasiniI should have writed down those modif... now I don't remember I know I corrected a few error but I might have add a few also
20:38:15kugelpyou the entire line is wrong when using pla
20:38:39bluebrotherthose changes are really quite convoluted :/
20:39:53bluebrotherplease try to make changes more incremental. If you're not using it try git (with git-svn), it makes it really easy keeping such changes separate
20:40:04jlbiasinibluebrother: this mpiohd200 is clearly wrong from me sorry :(
20:42:48jlbiasinibluebrother: convoluted?
20:43:22bluebrotherthere are a lot of loosely related changes.
20:43:44bluebrotherwhy is the target named "fuzeplus" and not "sansafuzeplus" (as configure *should* do)?
20:44:43bluebrotherhmm, seems it hasn't been changed for e200 and friends as well.
20:44:55bluebrotherlooks like something that is still missing from the target rename
20:45:42bluebrotherso let's keep this theme link list as-is for now :/
20:53:15 Join fml [0] (
20:53:35 Nick kugelp is now known as kugel (~kugel@rockbox/developer/kugel)
20:53:53kugeljlbiasini: okay. first pla doesnt handle sys events, I was wrong
20:54:05kugelso you still need the call to default_event_handler
20:54:28fmlbluebrother: Hey! I've seen two typos in the file encoderbase.h: "setttingsList" and "Chlid class should commit the from SettingsList to permanent storage" ("from" should be dropped)
20:55:08kugeljlbiasini: I really don't know what that line is for, I suggest removing it entirely, including the quit=true below
20:55:31jlbiasiniok let's try
20:55:51kugeland you shouldn't put the extra call to button_get_w_tmo there
20:55:55kugelI wonder why you put that
20:56:46jlbiasinibecause PLA doesn't map all key in all situation. On some device I let the normal keys (see the define above)
20:58:32kugeldont do that
20:58:58kugelwhy do you add special handling for some targets?
20:59:10kugelbecause of touchscreen?
20:59:32***Saving seen data "./dancer.seen"
20:59:39kugeltouchscreens are handled by using the grid mode currently. this is far from ideal but they don't need special handling
20:59:50jlbiasinino because they had special keymaps that I didn't want to mess with
21:00:55jlbiasiniTouchscreen are ok, but the only way to improve PLA support will be to have some time old device using the old method
21:01:07jlbiasinibecause there are to much case else
21:01:16bluebrotherfml: thanks, seems nobody noticed that for years ;-)
21:01:24 Join saratoga_ [0] (9803c31c@gateway/web/freenode/ip.
21:01:46kugelokay. but see how metronome does it if you really want special casing
21:01:49jlbiasinithrough about 3 - 5
21:01:55saratoga_Tuggers: i can talk to you a bit about the summer of code, although this year i'm not going to be mentoring due to other committments
21:02:13kugelyou add a another context, not call button_get
21:02:42kugelthe pla call will eat the key press so you'd need to press it twice then anyway
21:02:56bluebrotherheh, there's even one more: setttings :)
21:03:16jlbiasinikugel: even cutting out the part you told me there is still no way to quit and no repeat modifier
21:04:01kugelyou need to add a PLA_QUIT case too
21:04:59kugele.g. case PLA_QUIT: case PLA_SELECT: quit = true;
21:05:46kugel(now I see, that line makes lamp quit on any button that's not changing the colors)
21:08:02jlbiasiniyeah that is then what I thought, what's wrong with having PLA and classig mehod?
21:09:01jlbiasinithis not really clean of course
21:09:32jlbiasiniwell I will have a look to those thrn thanks
21:11:15 Part jlbiasini
21:13:55bluebrotherhmm, the manual still talks about separate builds for the ipod video. That's definitely outdated.
21:15:57Newguy1if i change my music folder to .music will it still show in the database but not in the file list?
21:16:01Newguy1thats what i want
21:16:44bluebrotheryou can also set the hidden flag
21:16:53 Join jlbiasini [0] (
21:16:57 Quit fml (Quit: ChatZilla 0.9.88 [Firefox 9.0.1/20111220165912])
21:16:57 Part jlbiasini
21:17:09Newguy1i just want my "music" to show up in the
21:17:10bluebrotherproperties dialog in Windows
21:17:27Newguy1and when i go to File> not to show the folders
21:17:47Newguy1oh okay i see what you mean
21:17:53Newguy1just make the folder hidden in windows
21:17:54bluebrotheror mtools
21:18:15bluebrotheror whatever tool that can set the hidden flag on a FAT filesystem :)
21:18:23Newguy1yeah that should work because the original itunes folder was hidden and it showed in the database
21:18:49Newguy1any idea why my one album isnt showing the album art?
21:19:05bluebrotherfile broken?
21:19:11bluebrotherwrong type in file?
21:19:17bluebrotherfile too big?
21:19:20bluebrotherprogressive jpg?
21:19:30Newguy1is 600x600 too big
21:19:31bluebrotherbad mood?
21:19:34Newguy1the 500x500 ones work
21:19:44bluebrotherwhat player?
21:19:53Newguy1ipod video
21:19:56saratoga_i think files have to be enormous before they'll break the decoder
21:20:06saratoga_many megapixels
21:20:06Newguy1i just put 3 albums on to test out rockbox
21:20:27bluebrotherah right, we're doing resizing on load, right?
21:20:35bluebrotherwell, then that shouldn't be much of a problem.
21:20:41bluebrotherperhaps the file isn't a jpg?
21:20:45Newguy1it is
21:20:46bluebrotheror it's a progressive jpg?
21:20:55bluebrotherhave you checked the file?
21:20:55Newguy1im looking at it right now
21:20:59bluebrotheri.e. the magic?
21:21:02Newguy1i can send you it lol
21:21:10*bluebrother rofls
21:21:58saratoga_can you view the jpeg with rockbox?
21:22:19Newguy1let me check
21:23:04Newguy1no it said unsupported file hmmm
21:23:06Newguy1plugin error
21:23:45 Nick tchan1 is now known as tchan (
21:23:51 Quit tchan (Changing host)
21:23:51 Join tchan [0] (~tchan@lunar-linux/developer/tchan)
21:24:00bluebrotherI'm suspecting the file to be a png
21:24:03saratoga_probably corrupt or not actually a jpeg then
21:24:12bluebrotherwhat do the first bytes in the file look like?
21:24:21Newguy1well on my computer its 100% .jpg
21:24:44Newguy1loading 220
21:25:04bluebrotherloading 220?
21:25:04Newguy1odd how the hell do i make it a jpg lol
21:25:30bluebrotheryour image viewer might ignore the file extension. The file extension is not a way to tell the file type
21:25:42bluebrothereven though lot of systems to it that way
21:25:59*bluebrother had some png files with jpg extension some time back
21:26:11Newguy1ah i see
21:26:26bluebrotheropen the file in a text editor and check the first bytes. Does it say JFIF among them?
21:26:33bluebrother0000000: ffd8 ffe0 0010 4a46 4946 0001 0101 0048 ......JFIF.....H
21:26:39Newguy1okay one sec
21:26:43bluebrotherthat's the first 16 bytes of a random jpg file
21:27:09Newguy1it says JFIF
21:27:25bluebrotherthen it's likely to be progressive jpg. Rockbox does not support those.
21:27:38saratoga_doesn't the viewer spit out a different error for progressive?
21:27:47saratoga_IIRC it'll actually tell you that it can't view progressive jpegs
21:27:56bluebrotherno idea.
21:28:11Newguy1sar it only shows the error for a second
21:28:26Newguy1unsupported -4
21:28:28Newguy1it says
21:29:10bluebrotherdoes the fuze+ have more than one partition exposed to the user?
21:29:20Newguy1i dont know what that means
21:30:25Newguy1why does my database have to do a search every time i go into Artists
21:30:33Newguy1i only have 3 albums...
21:30:53bluebrotherreturn (-4); /* other DCT model than baseline not implemented */
21:31:17saratoga_i didn't even know jpeg had multiple dcts
21:31:39bluebrotherme neither. Just grepped the code :)
21:32:45Newguy1oh well im more worried about solving this database search issue lol when i had 7 gigs of music added through itunes i didnt notice it ever say "searching database" when i went into Artists, let alone every time i go into Artists
21:33:31bluebrotherload the database to RAM.
21:33:58*bluebrother looks around for some lolcats
21:34:08bluebrotheror maybe a roflcopter?
21:34:10Newguy1okay i did that now ill restart and see it it changes
21:36:29bluebrotherhmpf, jlbiasini left :(
21:36:39bluebrotherthe manual stuff is still talking about "OF"
21:37:46*bluebrother sighs
21:38:00bluebrotherany manual guy around?
21:38:39*bluebrother is tempted to drop that patch for now
21:42:26Newguy1okay everything looking good load to RAM seems to make it load stuff from my database faster but takes a second or 2 to load album art on track changes.
21:42:41Newguy1so did the regular ipod OS if i recall too
21:45:38CIA-88New commit by bluebrother (r31595): Fix typo in comment. ...
21:46:55Newguy1oh blueb do you know if its possibly to show the album year next to the album in the database?
21:47:26CIA-88r31595 build result: All green
21:48:34saratoga_sure, the database stuff can be customized, check the wiki
21:50:50Newguy1hmmm i just booted it up and put it on hold to boot to the original fw how do i get it to boot up to rockbox again lol
21:52:15Newguy1its not working
21:53:20Newguy1or how do i reboot
21:54:20saratoga_i don't know what the button is on the newer ipods, maybe check the apple docs
21:54:37Newguy1oh i did it its center button and menu
21:55:14lorenzo92kugel: I sent lebellium a build with my gpio buttons patch, he likes that very much :D
21:56:54lorenzo92kugel: now rockboy is much more usable hehe
21:57:07 Quit Zarggg (Quit: Rebooting client...)
21:59:21kugellorenzo92: heh, probably. I never use that
22:00:14lebelliumkugel: if you had tried mario tennis, you would use it \o/
22:02:01 Join Che_NOIA [0] (
22:05:19lorenzo92kugel: I noticed an interesting fact. After SDL removal, I cannot see anymore the digital vu-meter! Only the analog one works....
22:07:32lorenzo92kugel: I mean the I go to see its code!
22:10:45lorenzo92also minimeters are stopped uhm
22:10:58 Part Che_NOIA
22:12:37kugellorenzo92: looks good, but don't use //-style comments
22:12:50lorenzo92yeah ok
22:14:30 Quit y4n (Quit: 6,000,000 ways to die — choose one.)
22:17:56 Quit ender` (Quit: Revised log levels proposal: "fyi," "wtf," and "omg.")
22:20:57kugellorenzo92: I don't suppose you can read multiple pins at once?
22:21:13lorenzo92no we cannot
22:28:23kugellorenzo92: hm okay
22:29:13 Quit adnap (Ping timeout: 260 seconds)
22:29:59kugellorenzo92: just thought we could reduce the syscall overhead
22:30:16lorenzo92kugel: would be
22:30:32kugeldon't know that's why I asked :)
22:34:45lorenzo92kugel: for the patch? I'll clean it a bit and then post it to flyspray?
22:36:43 Join adnap [0] (
22:43:01lorenzo92kugel: ah! btw, we could use directly the gpio also for the "charger sense"...or at least open the minivet device only one not everytime
22:43:55bluebrothernice, encoding the voice clips using libmp3lame at least feels notably faster than using lame.
22:45:10bluebrothernow to see if the created voicefile actually works :)
22:51:36 Join Zarggg [0] (
22:59:35***Saving seen data "./dancer.seen"
23:13:45funmanTheSeven: should I try to store endpoint information by direction too?
23:13:52 Quit Brijesh__ (Ping timeout: 252 seconds)
23:13:56funmanTheSeven: for nano2g / classic; so we could merge a bit more code
23:14:23TheSevenhm, I think the ideal solution would be a driver which autodetects most things
23:14:36 Quit lorenzo92 (Quit: ChatZilla 0.9.88 [Firefox 9.0.1/20111228084940])
23:14:42TheSeveni.e. number of available endpoints and their directions, fifo sizes, ...
23:15:38funmanyeah but those are 2 different things :)
23:16:32TheSevenyeah, but I think that this requires an overhaul of the endpoint information structure anyway, so we should take this into consideration if we need to touch that area anywaay
23:17:03funmani'll cook some patch
23:20:21 Quit TheLemonMan (Quit: WeeChat 0.3.6)
23:21:06pixelmabluebrother: yes, target rename is still unfinished in the manual - unfortunately fixing this is one of the tasks that become longer and more boring the closer you look at it :\
23:22:17pixelmayou'll find target opts in more and more places and if you want to do it right, you need to keep an eye on if there would be a more suitable feature opt etc.
23:22:46 Join TheLemonMan [0] (
23:24:54 Quit TheLemonMan (Client Quit)
23:26:22bluebrotherpixelma: well, as a first measure one could just do a batch replacement.
23:26:43bluebrotherIMO much more things in the manual should be handled via features.
23:26:58bluebrotherbut I have to admit that I haven't looked at the manual in quite some time
23:27:27bluebrothertrying to use libmp3lame for voicefiles on the Archos players right now. Expecting quite a bit of a speed boost (and less hassle to set it up :)
23:30:36 Join TheLemonMan [0] (
23:33:31bluebrothernice, loading libmp3lame works pretty much out of the box on Windows. Encoding fails for some reason though.
23:39:25kugelbluebrother: that's not what "works out of the box" means to me :)
23:49:23bluebrotherwell, it does load the DLL and I can read the version via a function. So loading the DLL works. Using ... well, partly. It could be very well a but not related to the DLL as well
23:51:50*bluebrother curses the number of typecasts this stuff requires
23:52:35 Quit Zarggg (Quit: Rebooting client...)
23:56:05 Join Zarggg [0] (

Previous day | Next day