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).

Notice: Only Gecko based browsers prior to FF4 support the multipart/mixed "server push" method used by this log reader to auto-update. Since you do not appear to use such a browser, this page will simply show the current log, and not automatically update.

#rockbox log for 2005-02-12

00:01:50amiconnBagder: The MEMORYSIZE selection isn't included in the simulator makefiles :(
00:02:46Bagderwant me to fix?
00:02:57amiconnYes please, if possible.
00:03:11amiconnI need this for making the size in buffer.c variable
00:05:30amiconnI guess I need to reconfigure?
00:10:29HClwhatcha working on?
00:11:39amiconnHCl: Making the simulator mp3 buffer size variable, for building rockboy...
00:12:03amiconn(well, not only)
00:20:50 Quit Stryke` ("Friends don't let friends listen to Anti-Flag")
00:20:54amiconnBagder: Rockbox info -> Buffer: 1.676 MB :)
00:21:21amiconnTrying the same with iRiver sim...
00:23:51HClok :)
00:26:03Bagdertime for sleep here, have fun!
00:26:06amiconnBagder: I get a strange effect:
00:26:36amiconnI added -DMEM=${MEMORYSIZE} to the DEFINES in both X11 and Win32 sim makefile
00:26:49amiconnWhile this works for X11, it doesn't for Win32
00:27:02amiconnI get a parse error in plugin.c line 501
00:28:08amiconnHmm, it's not the Win32 sim, but the iRiver sims
00:28:25amiconnIs MEMORYSIZE unset in the iRiver makefiles?
00:28:51Bagderexport MEMORYSIZE=32
00:28:59amiconnAh, I'm silly....
00:29:04amiconnDidn't reconfigure...
00:29:16amiconnNite Bagder
00:31:44amiconnFunny: Buffer: -9.-283MB
00:31:56linuxstbFLAC decoding is working in the simulator, but I think it's crashing on the target.
00:34:01HClhow big are flac files in general?
00:34:05HCllike, the music files
00:34:12gromit`<−−−−−−−−- like this −−−−−−−−−−−−−−>
00:34:37sneakumsabout 50% of the wav, i found in a brief test
00:35:20preglowahh, i notice how terribly unfamiliar i am with inline asm in gcc
00:35:31HClsneakums: ok.
00:36:17linuxstbIn reality about 60% of a WAV. But obviously, they sound identical.
00:36:20preglowwould anyone happen to know how to format a mac instruction in inline asm? :P
00:37:31XShocKabout FLAC files. wouldn't it be not really reliable to play them in player? i mean batteries will be dead very quickly
00:37:51XShocKi mean discharged, not dead
00:38:05preglowweell, you can pump quite a few hours out of it
00:38:06sneakumsit'd use more battery, but i don't see a reliability problem as such
00:40:21linuxstbBut FLAC may even be lighter on a CPU than MP3 or OGG - we don't know that yet.
00:40:49linuxstbBut I agree, it's probably not the most efficient format.
00:41:40lImbusdepends on what is wanted, I think. battery life is the price to pay for that quality
00:42:25amiconnWhee, Buffer: 31.679 MB
00:42:37lImbusand I doubt the iRiver analog-hardware is of such high quality that one can hear the difference between any of the other codecs (with decend bitrate of course) and FLAC
00:42:45lImbusamiconn: congratz :-)
00:43:21linuxstblImbus: Don't forget the digital output...
00:43:31lImbusof course
00:43:38amiconn@iRiver owners: What does Info->Rockbox info currently tell you about the buffer size?
00:43:42lImbusthat's why I added the word "analog"
00:44:06linuxstblImbis: sorry :-).
00:44:28lImbuslinuxstb: the iRiver after all is a portable music player. I don't know of any digital-input headphones :-9
00:45:36lImbusand battery life does not matter if you are stationary with a high-quality digital equipment
00:46:11amiconnXShocK: It will soon be fixed in cvs :)
00:46:46amiconn(It's a simple overflow btw, calculating the "true" MB from the # of bytes)
00:46:47preglowi'd guess the decoding itself is pretty light compared to the mdct based formats
00:47:02linuxstbI agree that I probably couldn't tell the difference between a well-encoded MP3 or OGG and FLAC on the iRiver, but I just prefer to keep my collection in one format, and not go to the effort of evaluating lossy codecs. At least with FLAC I can be 100% I am getting the best out of the hardware.
00:47:04pregloweverything is time domain in flac
00:47:37lImbuslinuxstb: Full ACK.
00:47:49lImbusthis is the advantage of having portable flac.
00:48:08lImbusotherwise the compression ratio would kill any argument as well
00:49:51linuxstbamiconn: Buffer: -10, -298MB
00:52:50amiconnlinuxstb: I couldn't fit my entire music collection on the largest capacity iRiver in FLAC format, while I can do so in MP3. That's a kiler argument for me.
00:53:24DMJC-Lall my cds are ripped as ogg at 6.5 quality
00:53:25***Saving seen data "./dancer.seen"
00:53:25 Join bagawk [0] (~Lee@bagawk.user)
00:53:31DMJC-LI'm not redoing em, nuff said
00:53:54linuxstbThat's going to be the great thing about Rockbox on the iRiver - everyone will be happy :-)
00:54:22amiconnI have pretty much everything as MP3 (guess why!), and some OGGs
00:56:03HCli have everything as mp3
00:56:12HClbut i want rockbox to be as compatible as possible
00:56:58 Quit Hohoman ("( :: NoNameScript 3.81 :: )")
00:57:24linuxstbI think I've said it before, but the "big three" codecs in my mine are MP3 (and MP2 for my digital radio recordings), OGG and FLAC. Unless someone else gets there first, I'm going to be working on those.
00:57:34linuxstb*my mind
00:58:17amiconnXShocK, linuxstb: Fix for buffer display committed.
00:58:40 Quit jyp ("poof!")
00:58:54lImbusamiconn: I can't fit my music-collection in mp3 on any avaiable hdd-hard disk :-|
00:59:02lImbusbagawk: Hi lee
00:59:11bagawklImbus, hey
00:59:13lImbusintrested in pictures of the cf-hdd-adapter ?
00:59:20XShocKok. rebuilding
00:59:30amiconnlImbus: Really? I have an 80 GB disk in my archos...
00:59:38XShocKisn't the CF already ata compatible?
00:59:53lImbusbagawk: I did not find it, so I could not scan it, but I still have a picture (photo) of it
01:00:15sneakumsmy vorbis collection is 6200 files at q2, fits nicely in my iriver
01:00:31bagawklImbus, OK I might go out and get parts to make one, if i get school work done (which I have a lot fo...)
01:00:32lImbusamiconn: I got about 160 GB. I admit, there a part of it I virtually never listen to.
01:00:57bagawkMy HD is 160gb
01:01:10sneakumsmy biggest disk is 80GB, and it has 40G free
01:01:12bagawkWhat format/quality?
01:01:44amiconnToo bad that anything about 128 GB (137 GB in "manufacturer gigabytes") won't work in either mobile player...
01:01:53linuxstbamiconn: info->rockbox info now showing a buffersize of 30.664MB (I have a 1MB plugin area configured).
01:02:07lImbusquality is mp3 vbr min 192 max 360 or what the real max is
01:02:20amiconnMP3 max is 320 kbps
01:02:42lImbusbagawk: it's not worth the work of getting the parts and soldering the hdd-cf-adpater yourself. I got one cheapo at ebay (about 10 euro)
01:02:51 Quit bagawk ("Leaving")
01:03:15lImbusug. I hope he saw that one before going out and spending that money
01:03:37XShocKamiconn: now it shows Buffer: 21.632MB
01:03:52amiconnAs it should be...
01:04:30lImbusamiconn: I still cannot build a Gmini or iRiver Sim, not even on a completely new installed BlueChip-cygwin
01:04:58amiconnHmm, strange
01:05:02lImbusjust for info as you all had some problems these days getting sim building
01:05:11lImbusthe problem existed before, and still does
01:05:35amiconnI only had problems building the iriver sim. I fixed that issue.
01:06:01amiconnIf you update to current cvs, you need to reconfigure all your sim build directories
01:06:08lImbusit sounds very simple: while compiling apps/recoder/widgets.c, the include file limits.h cannot be found, thus the symbol MAX_INT is not found and blablabla
01:06:43lImbusI installed a new cygwin (BlueChip style), cvs checkout, build tools, configured new build directory, build, daaang
01:07:07amiconn*when* did you checkout exactly?
01:07:25amiconnThere was a problem (jyp caused it, jyp fixed it) for a while
01:07:28lImbusmhmm. it was either yesterday in the late evening or today
01:08:31CtcpIgnored 2 channel CTCP requests in 1 minute and 2 seconds at the last flood
01:08:31*lImbus is doing cvs update and tools-build and reconfigure and all that on his existing build environment
01:08:48lImbus(win32 sim that is)
01:09:50XShocKwin32 sim does not build. error on buffer.c:23 :)
01:10:26lImbusI'm tempted to delete the include instruction and replace the MAX_INT by (unsigned) -1
01:10:37amiconnXShocK: You need to reconfigure
01:11:44lImbusuh. even more strange. I think it's worth a multiline-paste. BRACE !
01:11:50XShocKamiconn: just reconfigured, the smae error.
01:11:52amiconnlImbus: Works just fine here. Checkout again, reconfigure and retry
01:12:07lImbusCC ../../apps/settings.c
01:12:07lImbus../../apps/settings.c:23:20: limits.h: No such file or directory
01:12:07lImbus../../apps/settings.c: In function `set_int':
01:12:07DBUGEnqueued KICK lImbus
01:12:07lImbus../../apps/settings.c:1338: error: `INT_MAX' undeclared (first use in this funct
01:12:08***Alert Mode level 1
01:12:08lImbus../../apps/settings.c:1338: error: (Each undeclared identifier is reported only
01:12:12lImbus../../apps/settings.c:1338: error: for each function it appears in.)
01:12:14lImbusmake[1]: *** [/home/guest/GminiSim/settings.o] Error 1
01:12:16lImbusmake[1]: Leaving directory `/home/guest/uisimulator/win32'
01:12:18lImbusmake: *** [sim] Error 2
01:12:38lImbusI SWEAR it was on widget.c before.
01:13:24lImbusamiconn: I really updated and reconfigured (I created a new build directory GminiSim7th
01:14:30amiconnNow that is strange. I have no idea what's going on then. I'm using pretty much up to date cygwin and it works...
01:14:52amiconnOfficial builds are fine as well....
01:15:32XShocKhmm. where MEM is declared?
01:15:50lImbusofficial sim builds ? going to download it...
01:15:51amiconnIt's declared in the Makefile generated by configure (now)
01:16:01amiconnThat's why you need to reconfigure
01:16:24amiconn...and before rebuilding, probably need to 'make clean'
01:16:43 Join jyp [0] (
01:16:51 Join LinusN [0] (
01:17:08HClhey linus
01:17:18preglowLinusN: i've got a bug for you
01:17:28LinusNnow nice, just what i needed
01:17:38amiconnhi LinusN
01:17:45LinusNhi amiconn
01:18:43XShocKhi Linus
01:19:09lImbushi Linus
01:19:40preglowi hate to interrupt the greet list, but the bug is as follows: plug in usb, start rockbox, now stop rockbox, and behold, it never shuts off
01:19:50HClit doesn't shut off
01:19:51preglowinsert 'plug out usb' some place in there
01:19:52HCltill you remove usb
01:20:00preglowit doesn't shut off then either
01:20:04HClit did for me o.o
01:20:08amiconnOne needs to be careful with <tab> completion these days... There are now 3 nicks starting with 'li'
01:20:25LinusNpreglow: will have a look
01:20:33XShocKamiconn: sorry, just deleted everything and checked out again, everything worked
01:20:42preglowLinusN: no worries, i don't think this is something people do often
01:21:05lImbusXShocK: If I could only pretend this.
01:21:27lImbusamiconn: are there official sim builds to download, or did I misunderstand you ?
01:21:29 Quit jyp ("poof!")
01:22:09***Alert Mode OFF
01:22:22amiconnlImbus: Afaik you can't download them (and Win32 builds are only done for player and recorder), but the automated builds are doing them, so one can see whether something is broken
01:22:24LinusNok, i found the bug
01:22:40amiconnLinusN: That was fast :)
01:22:49LinusNi knew where to look
01:23:02LinusNand it isn't very easy to fix...
01:23:26preglowi apologize :/
01:23:31amiconnLinusN: If I would go getting a H3xx, I would surely run into problems. I don't have a wiggler...
01:23:43preglowLinusN: so this is an archos problem as well?
01:24:00LinusNpreglow: not really, archos handles it a little differently
01:24:31lImbusamiconn: absolutely. I suppose it has to do with the resting environment. cygwin, env$, microsoft-devstudio installed on the same machine...
01:25:12preglowLinusN: on a positive note, it seems binutils support the mac/emac extensions
01:25:20LinusNyes it does
01:25:36amiconnpreglow: No problem on archos
01:25:58preglowi am yet to discover their syntaxes, but i'll probably arrive at the solution given a day or three of intense study
01:26:10LinusNpreglow: my first patch to the binutils project concerned the emac
01:26:28preglowthat's right, i think i remember reading about that
01:26:31LinusNpreglow: there are example source codes on the motorola site
01:26:40preglowthen that's where i'll turn my attention, heh
01:28:09preglowi don't seem to have to do anything by myself today ;) thanks
01:30:28LinusNhmm, the bug i found wasn't a bug after all... :-)
01:31:09preglowahh, it's a feature? :P
01:31:54LinusNbut i think i found "your" bug
01:32:56LinusNfix committed
01:34:45 Quit mecraw ()
01:38:57LinusNtime to sleep, nit guys
01:39:11 Part LinusN
01:44:06 Join etienne [0] (
01:44:58etienneanybody have an idea why my power adapter overheats? (fm recorder)
01:45:17etiennei also managed to fry my roomate's power adapter for his archos
01:48:20lImbusthere are not much of those techies in here atm. are you able to measure the power drain ?
01:48:52etiennenope, i have no idea how to do that
01:49:09lImbusdo you own a multimeter ?
01:49:19etienneyes i do
01:49:45etiennehow do i do it
01:49:46lImbusdid you ever do anything by setting it to the DC-A -mode
01:50:38etienneno...i tried looking through the menus for some settings
01:50:43amiconnHehe, snow on the player...
01:50:46etiennebut i could only find the battery capacity setting
01:50:55lImbusamiconn: crazy guy :-)
01:51:05etienneamiconn....i won't mind using it for snowboarding tomorrow actually :)
01:51:58lImbusetienne: the fmrecorder has charging control in hardware, I don't think there is a lot to see and configure in rockbox
01:52:51etienneno quick fix i guess then
01:53:08etiennewhat will draining measurement tell me?
01:53:23amiconnetienne: No offense, but check the following: (1) Do you have the correct adapter? What parameters does it tell you? (2) Is it for the correct mains voltage?
01:54:02etienneno offense taken.
01:54:27etienneThe thing is, my roomate have the exact same model and because my adapter (9v 700mA) overheated i borrowed his
01:54:41etienneand ended up buring it out
01:54:54amiconnThe fm/v2 adapter should be 6 V / 700 mA ...
01:54:55etienneso he borrowed mine...and it didn't overheat and work fine for him
01:54:58lImbusetienne: the measurment would tell you the current that flows
01:55:18etienneamiconn, the one that burnt out was 6v/700mA
01:55:33etiennebut apparently you can go up to 12v....10 v being desired
01:55:50amiconnNope, not for the v2/fm
01:55:51lImbusnononoooo. that's not v2
01:55:52amiconnThe 9V adapters are for the V1 and the player
01:56:09etiennei see
01:56:16etienneso the 9v is too high for it?
01:56:18lImbusdamn. I just thought half an hour ago that we should clarify this in the battery and charging faq
01:56:31lImbusyes, it is
01:56:52lImbusin my humble opinion, this should still not fry the charger
01:57:10etienneand it doesn't charge by USb either does it (looks on the screen like it does...but doesn't tend to keep a charge long"
01:57:24lImbusit "should" fry the archos, as far as I understood electronics
01:57:54etiennei'm glad the adapter fried and not the archos then....but it makes me wonder why the 6v adapter died
01:58:20DMJCsafety mechanism blew perhaps?
01:58:36amiconnCharging via usb will take rather long, because the current is lower. And the doesn't hold-charge-long problem may be the battery contacts
01:59:05amiconnThis is *not* the same problem as the v1/player battery contacts.
01:59:22etiennei just tried opening it and it looks like i need to solder to get to the battery (the rockbox website only have dissasembly photos for different models)
01:59:52amiconnI don't know there very details, as I don't own a fm or v2
02:00:17etiennei hate getting haused by ebay sellers *sigh*
02:00:19lImbusme neither
02:00:43amiconnetienne: eBay? Okay, I ask again: What mains voltage the 6V charger is for?
02:00:58lImbusI thought about it for a moment
02:01:04amiconnPlus, what is the actual mains voltage where you live?
02:01:05etienneit is for 110
02:01:20etienneit worked fine on my room mates archos, which is exactly the same as mine
02:02:02amiconnHmm, strange. As you told he has a 9V charger, I thought he would have a v1 recorder or a player
02:02:04etienneactually, both worked for him, and neither for me
02:02:24etienneno, i got the 9v charger with mine.....baught it over ebay
02:02:46etiennemy charger (9v) is a lot bigger than his(the real archos 6v charger)
02:03:02etiennemine is a 9v 600mA
02:03:11lImbusoh, so you bought a archos device with a universal power supply
02:03:34amiconnetienne: The 9V 600mA charger is the one for rec. v1/ player
02:03:51lImbusamiconn: should it overheat ?
02:03:56etienneits not universal, no settings on it and such and only one plug that fits the thing perfectly....won't be surprised if it came with the archos
02:04:25lImbusand why died the other one, if it was the original delivered with another fmrecorder ?
02:04:36 Join iSheep [0] (
02:04:48amiconnlImbus: No it shouldn't. But as the 6V charger overheated as well, I suspect a problem with the box. Perhaps a short circuit in the socket.
02:05:15amiconnPlus the box came with the wrong charger...
02:05:31etienneare all fm recorders V2?
02:06:00etiennebecause this one is definitely fm recorder, doesn't say much else "Jukebox FM Recorder"
02:06:05amiconnThe fm and v2 recorders are the same hw design (LiIon battery), only the v2 doesn't have a radio
02:06:06lImbusall recorders having a radio are version2 recorders, yes.
02:06:29amiconnThe v1 recorders and player have AA size NiMH batteries
02:07:42iSheepAnyone familiar with the FM Recorder and issues with the headphone jack?
02:07:47 Quit preglow ("and thusly")
02:08:22lImbusetienne: see
02:09:45etiennei see what you mean about the charger spec
02:09:56CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
02:09:56*lImbus nods
02:10:59etiennei wonder if having dead batterys can cause the charger to overheat?
02:11:22etienneit just never charges enough/constant drain.....i'm speculating here :)
02:12:31etienneanyways....thanks for the help guys
02:12:42etienneand ladies...possibly
02:12:52etienneat least i now know not to use the 9v one
02:13:07 Quit etienne ()
02:14:08amiconnEeeek! CPUAdrErr in snow.rock
02:14:19iSheepI'm having a problem with no audio on my Jukebox. Coincidentally, the LCD screen recently died, so I'm in bad shape. I think the audio is not working, because I think I'm blindly getting a song playing when Rockbox starts up and would (presumably) ask if I want to resume playback
02:16:40lImbusmhmm. that sounds hard to check, iSheep
02:16:59iSheepYeah... not a good combo :(
02:17:29iSheepBut I'm almost positive a song would be playing, because if I press the right button, the hard drive light flickers as if it were switching songs
02:18:45iSheepFor a little bit after the screen died, the audio was still working. I could play a song blindly. But now, for some reason, I can't seem to do that
02:20:05lImbusmhmm. having a broken screen should be no big problem with rockbox, as you're still having voice-feedback like many blind users. but NOT having audio renders a music player somehwat userless
02:20:25lImbusit's still a battery powered external hard drive though
02:21:10iSheepright. well, what I'm trying to decide is if I should buy a new LCD screen so I can see what's going on
02:21:10iSheepI don't understand how the audio could just go out like that
02:21:29lImbusif it's just the connector that's broken, that should be fixeable cheaper
02:21:36 Quit iSheep (Read error: 104 (Connection reset by peer))
02:24:42 Join iSheep [0] (
02:24:50iSheepSorry about that
02:24:56DMJChe didn't even see it
02:25:07iSheepInternet woes. :) What did I miss?
02:25:20DMJCnothing worth seeing
02:25:32iSheepoh ok
02:26:13iSheepwell, anyone with any ideas on where I should go from here? I could go buy a new screen, but I would hate to do that if I'll never have sound anyway. or is there a way to fix the sound?
02:26:16lImbuswb iSheep :-)
02:27:07lImbuswell, if you feel handy to open the box (or a friend of you), do it, and look what's wrong with the audio connector
02:27:21lImbusthese wore out from time to times
02:27:41 Quit edx (Read error: 110 (Connection timed out))
02:27:53iSheepyeah, actually I have the thing opened and totally taken apart, really
02:28:01iSheepI just don't know what to look for, because things seem to be okay
02:28:16lImbusfirst you should try if you can't fix it (at least some rumble and crackle) by turning and shifting/pushing the headphone-plug softly
02:28:51iSheepI tried that while I *think* I was listening to a song. I didn't hear anything at all in the headphones
02:28:52lImbusit may be very very tiny cracks in the soldering joint of the headphone jack
02:29:31iSheepwould that be near where the green and yellow wires connect?
02:29:40lImbusplease have a look at the wiki (or the mailinglist-archive), iirc, there is a article about such a problem
02:30:13lImbusI dunno, I never saw a fm recorder
02:30:28iSheepyeah, I'll check it out. should I just search the wiki?
02:30:42lImbustry to, as well as the mailinglistarchive.
02:30:56lImbusi have to go to bed now anyways
02:31:16iSheepall right. well, thanks for the lead. I appreciate it
02:31:18lImbusiirc, somebody documented his replacment work of a damaged socket
02:31:34lImbusthat's all included in rockbox :-) you're welcome, anytime
02:31:45iSheephave a good one
02:32:04 Quit linuxstb (Read error: 60 (Operation timed out))
02:32:20lImbusgood luck while searching. start with the docsindex maybe
02:32:36iSheepok, will do
02:32:45*lImbus is detaching from computer-related stuff
02:32:49lImbuswill go to bed now
02:32:57lImbusbood nite #rockbox
02:32:57iSheepAll righty. See ya later
02:33:20lImbusyeah, see ya. hope to see that I was right :-)
02:33:37iSheepsame here!
02:33:56 Quit lImbus ("initiating bed-mount sequence")
02:47:00 Quit _aLF ("Leaving")
02:53:29***Saving seen data "./dancer.seen"
03:03:41 Quit amiconn (" while(night) amiconn.sleep();"")
03:03:46 Quit iSheep (Read error: 54 (Connection reset by peer))
03:25:47 Join ashridah [0] (
03:32:02 Quit XShocK (Read error: 104 (Connection reset by peer))
03:33:37 Join XShocK [0] (
03:43:42 Quit jpburton5150 (Read error: 110 (Connection timed out))
03:47:32 Part CrunchyWhiteMeat
03:50:09 Join coob [0] (
04:00:19 Quit Cassandra_ (Read error: 60 (Operation timed out))
04:06:11 Join QT_ [0] (as@area51.users.madwifi)
04:08:34 Join edx [0] (
04:09:01 Join Cassandra_ [0] (~christi@
04:09:59 Quit cYmen ("leaving")
04:16:01 Quit QT (Read error: 110 (Connection timed out))
04:53:33***Saving seen data "./dancer.seen"
05:20:34 Join jpburton5150 [0] (
06:41:23 Join midk [0] (
06:42:01 Quit midk (Client Quit)
06:49:55 Quit jpburton5150 (Read error: 60 (Operation timed out))
06:51:02 Part hubble
06:53:36***Saving seen data "./dancer.seen"
06:59:29 Quit Taxi|3 (Read error: 110 (Connection timed out))
07:16:38 Join webguest52 [0] (
07:17:54webguest52anyone willing to answer a qestion about flashing a daily build on recorder?
07:19:27 Join pcngss [0] (
07:20:45 Join Shulberry [0] (
07:21:39 Quit pcngss (Client Quit)
07:21:53 Quit webguest52 ("CGI:IRC (EOF)")
08:40:04 Join methangas [0] (
08:48:37 Quit Shulberry (Read error: 60 (Operation timed out))
08:53:38***Saving seen data "./dancer.seen"
08:53:56 Join midk [0] (
08:56:56 Quit XShocK (Read error: 104 (Connection reset by peer))
08:56:58 Join midk_ [0] (
09:00:52 Join sneakums_ [0] (
09:03:11 Quit midk (Read error: 104 (Connection reset by peer))
09:05:28 Quit sneakums (Read error: 60 (Operation timed out))
09:22:02 Join Shulberry [0] (
09:34:46 Join Lynx_ [0] (HydraIRC@
09:45:40 Quit midk_ ("Leaving")
10:02:05 Join fizze [0] (
10:17:11 Quit Lynx_ (" HydraIRC -> <- \o/")
10:20:54 Join Lynx_ [0] (HydraIRC@
10:34:50 Quit fizze (Read error: 110 (Connection timed out))
10:43:01 Join ze__ [0] (
10:53:40***Saving seen data "./dancer.seen"
10:57:32 Quit ze (Read error: 110 (Connection timed out))
10:57:32 Nick ze__ is now known as ze (
11:03:01 Join GnagelRam [0] (
11:04:07 Join linuxstb [0] (
11:12:50 Join Patr3ck [0] (
11:15:00rasherWhat does "Can't find entry point" mean? (getting it in the recorderv2-sim
11:15:29Lynx_rasher: I'll tell you when you are older ;))
11:15:44*rasher beats up Lynx_
11:16:13Lynx_and rightly so ;)
11:37:21 Part GnagelRam
11:38:12rasherah, got it
11:38:24rasherhad my preprocessor directives messed up
11:49:31rasherbmp2rb is so not getting it
11:54:20 Join ze__ [0] (
11:58:43Lynx_still working on the bumping logo?
12:00:21rasherI made a smaller version that will fit on the recorder with room to spare
12:00:37Lynx_cool, then i can try it too, when it's done
12:00:51rasherwell it worked on the recorder
12:00:54rasherjust didn't move sideways :)
12:01:15rasherbut I have messed up .. something
12:01:30rasherso now the bmps are totally corrupted
12:04:14linuxstbMorning all. My iRiver FLAC decoder is working perfectly in the SIM (converting a FLAC file to raw PCM file), but on the iRiver itself it is crashing in a read() command - on about the 3rd or 4th read(), the hard drive LED stays on and the read() function never returns...
12:06:48rasherHurray! And also, aww!
12:08:10 Quit ze (Read error: 110 (Connection timed out))
12:08:10 Nick ze__ is now known as ze (
12:11:43 Quit DMJC (
12:11:43 Quit Ka (
12:12:09NJoinDMJC [0] (
12:12:09NJoinKa [0] (
12:16:16*rasher sighs
12:20:16 Quit coob (Read error: 60 (Operation timed out))
12:30:55 Join R3nTiL [0] (~zorroz@
12:32:48rasherHurray, I have a bouncing logo on the recorderv2-sim
12:35:40 Join muesli- [0] (
12:36:23linuxstbNice one. Still no joy with my FLAC decoder - but it's now displaying some nice stats on the screen whilst it's decoding (up until it crashes).
12:37:05 Nick sneakums_ is now known as sneakums (
12:37:45 Join R3nTiL_ [0] (~zorroz@
12:38:06 Quit R3nTiL (Read error: 54 (Connection reset by peer))
12:43:15 Quit Patr3ck ()
12:43:47rasherhrm, can't bring up the menu on the ondiofm-sim
12:43:55 Join jyp [0] (
12:51:01HClwhy does it crash?
12:51:36HCllinuxstb: check what memory address that read is writing to?
12:52:00*HCl yawns
12:52:13rasherdo I have to seed rb->rand() ?
12:52:36 Join amiconn [0] (
12:53:26rasherHi there
12:53:41***Saving seen data "./dancer.seen"
12:53:48amiconnrasher: On the Ondio, bringing up the menu requires a long press of the MODE button
12:54:05linuxstbHCl: Yes, I've done that, and it looks fine.
12:54:12amiconnThat's because there are so few buttons...
12:54:27linuxstbI'm now trying to read the whole file into the MP3 buffer first, and then set the decoder going on that.
12:56:30amiconnrasher: If your bouncing logo works on on the recorder, it should also work on the Ondio. The lcd is identical.
12:56:44rasheryeah, I'm still failing to bring up the menu though
12:56:56amiconnX11 or Win32 sim?
12:57:42rasherAs a side product, I have a 91x32 rockbox logo
12:57:47rasherif that's of any use anywhere
12:58:08*amiconn is building an Ondio X11 sim
12:58:17 Quit DMJC (
12:58:17 Quit Ka (
12:58:28rasherit doesn't give me a "received ev xxxxx" on the console
12:58:33rasherso it looks right
12:58:46NJoinDMJC [0] (
12:58:46NJoinKa [0] (
12:58:58amiconnPerhaps the X11 sim doesn't handle the button repeat correctly.
12:59:01 Join cYmen [0] (
12:59:03rashercould be
12:59:05rasheroh well
13:00:13linuxstbHurray! It seems to be working. Slowly...
13:00:20*HCl grumbles at hp
13:00:22rasherat 11mhz :)
13:00:25HCldon't you hate it when like
13:00:28amiconnHmm, it doesn't work. It does definitely work in the Win32 sim though
13:00:31HClyou have a serious problem with a laptop
13:00:36HCland they just treat you like an idiot
13:00:39HClat an helpdesk
13:01:16rasherI don't think I've ever bothered with manufacturer helpdesks
13:01:30linuxstbIt's currently taking about 11 seconds to decode 1 second of audio....
13:02:35*HCl blinks at his installer as it says amd confidential o.o
13:02:46linuxstbWhen is Linux going to inject steroids into our iRivers? :-)
13:02:48rasheramiconn: do you know if rand() always returns the same first time on the sims?
13:03:23amiconnrasher: I don't know
13:04:24rasherokay, it looks like the logo always starts in the same direction
13:04:29rasherbut not on my iRiver
13:04:41rasherAlright,"> if you want a bouncing logo :)
13:09:31amiconnDo you use the X11 sim on windows/ cygwin, or on linux?
13:10:00amiconnHmm. Could you check something for me?
13:11:09amiconnOpen uisimulator/x11/screenhack.c, change lines 270 and 287 to read '#if 1' instead of '#if 0' and recompile the sim.
13:12:04amiconnThen start the sim and hold down the INS key. The interesting question is: Do you also get keyrelease events between each 2 keypress events?
13:12:45rasherI do
13:13:20amiconnHmm. SO the release/repeat handling in the x11 sim is broken, and the Ondio sim can't call the menu because of this.
13:15:30amiconnYou are the first to notice this...
13:16:01amiconnI usually use the Win32 sim, because it's more convenient on windows. Button repeat is working there.
13:16:02rasherMaybe it's caused by versions of libraries?
13:16:38amiconnI don't know. The X11 sim is rather old code, and I'm not much of an X11 expert.
13:16:45 Quit muesli- (Read error: 113 (No route to host))
13:17:04amiconnHowever, I get the same behaviour with cygwin X11 on windows.
13:18:49amiconnHmm. It looks like this situation should be handled by the code...
13:19:02 Join bsec [0] (
13:24:42rasherdon't look at me :)
13:25:32 Join webguest84 [0] (
13:27:21 Quit webguest84 (Client Quit)
13:30:30 Join Hohoman [0] (
13:30:53amiconnrasher: I found the actual problem.
13:31:17rasherOh boy
13:31:19rasherand a fix?
13:31:30amiconnThe repeat handling works *almost* correct in the X11 sim in that it properly detects repeats
13:32:15amiconnHowever, the target button driver is supposed to send keydown, and when repeat kicks in continue with sending keyrepeat
13:32:31amiconnIt should only send keyup at the very end of the sequence.
13:32:42rasherMakes sense
13:32:52amiconnThe x11 driver sends keydown, keyup, the keyrepeat....
13:33:09rasheroh ah
13:33:31amiconnThis cannot work with the event loops in rockbox code, as they need to tell the difference between short & long press
13:34:12 Join Quelsaruk [0] (~kvirc@
13:34:17amiconnThey do this by looking at the next event after a keydown. If this is keyup, it was a short press, if it is keyrepeat, it is a long press
13:34:47amiconnThe x11 driver breaks this by sending keyup before the repeat, hence rockbox always counts this as a short press
13:35:20 Join Sucka [0] (
13:35:27amiconnhi Quelsaruk
13:35:44amiconnrasher: Fixing this might be difficult, but I'll look into it.
13:36:52rasherdon't rush it for my sake
13:43:47 Join funkymonkey [0] (
13:43:47 Quit methangas (Read error: 104 (Connection reset by peer))
13:44:43rasherAnother Dane!
13:46:12sneakumsdown with denmark.
13:46:53DMJC can someone fix rockblox on iriver/
13:47:15rasherit works afaics
13:47:26DMJCit's tiny
13:47:31DMJCand comes across not down
13:47:51rasherI thought you meant there was a bug
13:48:09DMJCthe bug is, it's tiny and the iriver display is huge?
13:49:10sneakumssounds like a good first project
13:49:22rashersneakums: Get on it
13:49:26sneakumsno thanks.
13:49:33DMJCmy first project was fixing snake
13:49:56sneakumsthen you're the perfect person to fix rockblox!
13:50:04DMJCer.. no
13:50:21linuxstbSomeone who knows about these things needs to fix the battery status on the iRiver. But the TODO list for the iRiver is getting smaller each day...
13:50:37sneakumsit's scary how fast it's shrinking
13:51:22linuxstbI was amazed during the "flashing party" last Sunday at how much "just worked" (I know that's unfair to Linus and others!).
13:51:26DMJCstill waiting on the audio stuff
13:52:17*rasher ponders trying to change snake2
13:52:17linuxstbI think Linus wants to make the rest of Rockbox stable before tacking the audio. Which is good, because there is no new audio systerm written yet for him to use.
13:52:43DMJCthe rest of it seems stable enough
13:52:51sneakumsa reasonable approach
13:54:08linuxstbDMJC: Yes, it is. I think the main thing left (outside the audio) is a function to change the CPU speed.
13:54:09rasherDoes anyone know if lcd_clear_display() is expensive?
13:54:32amiconnNot that expensive
13:54:39rashercompared to clearing a rectangle the size of the logo
13:54:50amiconnIt's a simple memset() for the whole framebuffer
13:54:57DMJChas rockboy been uploaded into cvs?
13:55:00rasherwon't bother then
13:55:25amiconnrasher: Clearing a rectangle is way more expensive with the current implementation
13:55:34rasherHeh, alright
13:55:47amiconn...because it clears each pixel individually. Sic!
13:56:43HClDMJC: not yet.. *looks at amiconn* :x
13:56:50HCli think it should work in sim first?
13:56:56HClbefore its allowed to be added to cvs?
13:57:24DMJCrockboy worked on my iriver
13:57:40HClyes, it works on irivers, but not in the simulators yet
13:57:41DMJCit's just that I want to build it against my fbuild
13:58:08linuxstbHCl: How big does PLUGIN_RAMSIZE need to be for your rockboy now?
13:58:08HCl*stares at his cat walking very determined into his room*
13:58:14HClhold on, let me check what he's doing >.>
13:58:19HCllinuxstb: 700k should be enough
13:58:21HClfor now
13:58:55*R3nTiL_ slaps DMJC-L around a bit with a large trout
13:59:23 Join webguest85 [0] (
13:59:25linuxstbHCl: Can you decrease that by moving any buffers to the mp3_buffer ? People have been taking about a 512K plugin size on the iRiver.
13:59:33amiconnIf I manage to get it running in the sim, I'll try to make it run on the archos as well. Should work, and probably even a bit faster than it currently does on the iRiver.
13:59:41linuxstb^talking about (linuxstb can't type today)
13:59:50HCllinuxstb: i did my best, i moved most buffers to the mp3 buffer by making a malloc implementation
14:00:08amiconnOf course this wouldn't be able to handle all roms, because of the ram size, and the display needs to be shrinked to 1/2
14:00:24 Quit R3nTiL_ ()
14:00:36HCl :p
14:01:15amiconnI also have some ideas how to speed up lcd refresh, but I'll have to look in what order gnuboy spits out the scanlines
14:01:45amiconnThe badness is that gnuboy produces scanlines, but the bytes in the rockbox framebuffer are aligned vertically
14:01:47 Quit webguest85 (Client Quit)
14:01:58rasheryou'll have to do pretty evil screen-chopping though
14:02:33amiconnSo you do one lcd_update_rect() for each scanline, total of 8 per framebuffer "byte line"
14:03:08amiconnIf gnuboy always writes sequentially, it's sufficient to lcd_update_rect() in the last scanline of each block
14:03:30HCli set it to the first, at the moment..
14:03:36HCli think
14:03:46DMJCwhere is the gnuboy src kept?
14:03:52HClno, i just set it to the scanline
14:03:55DMJCI mean the rockboy src heh
14:04:12HClthere's a pretty recent one on my ftp server, but hold on
14:04:19HCli'll make a more recent zip
14:04:45amiconnrasher: I don't need more chopping than on iRiver if I leave out every 2nd scanline & column
14:06:00amiconnIf this really works, I could go a bit further...
14:06:53amiconnSince it makes no sense to set away 1/3 of all ram (on the archos) as plugin ram, I could make a special build, which does only gameboy emulation.
14:07:37amiconnSince there is RoLo on the archos (starting another firmware file from within the current one), this would be simple to use...
14:08:47amiconnIt could even 're-RoLo' standard rockbox on exit
14:09:21rasherthat'd be a nice way to do it
14:09:50amiconnYou know that I must be crazy...
14:10:55 Join elinenbe [0] (
14:11:17elinenbehello... what's this with the gamboy emulator I see on the misticriver forum? is that for real?
14:11:29*HCl grins
14:11:31HClyea, i made it :3
14:11:37HClits still rather slow
14:11:44HClbut i'm hoping to get it running properly
14:11:48elinenbeHCl: what is it based off of? gnuboy?
14:11:59elinenbeHCl: fabulous work!
14:12:10HClthanks :p
14:12:21elinenbeyou know I would always joke about a gamboy emulator on the archos, but wow! you really made it happen!
14:12:28elinenbedo you have any other screenshots?
14:12:37*rasher watches jyp go insane with the "long policy" fixes
14:12:40HClum, there are 4 on rockbox's site..
14:13:05HCli managed to get a decent screen with pokemon yellow
14:13:07elinenbeI am looking now...
14:13:10HClbut since its still black and white
14:13:20elinenbeis the screen large enough (in pixels)?
14:13:21HClmost color roms mess up since they need at least 4 bit grayscale
14:13:34HClits 16 rows too small
14:13:46HClbut that needn't be a problem
14:13:51elinenbewhy is that?
14:13:53HClseveral solutions have been suggested for that...
14:14:21rasherBagder: shouldn't the codecs be removed from IriverPort now that they're on a seperate page?
14:14:21elinenbe2 buttons on the side for up/down scrolling?
14:14:26HClswitching between dropping top 16 and bottom 16 pixels with the hold button.. dropping 8 at the top and 8 at the bottom.. dropping a line every 9 lines
14:14:37elinenbeHCl: I am so impressed −− amazing work!
14:14:42HClthe biggest problem is not being able to press two buttons at the same time
14:14:56elinenbeHCl: is it a plugin? how does it currently work?
14:15:05HCllast version is built as a viewer
14:15:12HClyou can just select gameboy roms in the browse thing
14:15:54elinenbewow! wow! wow!
14:16:12elinenbeoh... one more thing. what is the speed (percentage of real-time)?
14:16:34HClat the moment it does about one screen update per second...
14:16:46HClit needs to be at least 30 times faster to be actually playable
14:17:24elinenbeeh... hehe
14:17:33elinenbeso, you need some nice optimizations!
14:17:45HClout of sheer curiousity
14:17:57HClwhere's the thread/stuff on rockboy on misticriver?
14:18:46elinenbethe link was from
14:19:29HClah :)
14:20:15rasherit's linked from the wiki somewhere
14:20:22rasherthat and another thread
14:21:35DMJCsomeone rigged up an 80gb ihp-140
14:21:43DMJCthey used a 3.5" hdd
14:21:49DMJCbut the thing worked
14:21:55linuxstbrasher: I've been thinking about moving the codec info from iRiverBoot to the codec page - I'll do it now.
14:22:16rasherlinuxstb: Bagder did it an hour ago
14:22:23rasherbut didn't remove it fromt he iriverport page
14:22:27rashersee the recent changes
14:23:04linuxstbrasher: I should keep up to date...
14:23:04rasherah, I think he removed it now
14:23:18rasherit wasn't an hour ago
14:23:22Suckajust thought i'd mention, on the rockbox site in the 'features' section it says iriver cant display txt files natively, but it can
14:23:25rasherthe GMT times keep confusing me
14:23:32bsecis e the super mario rom playable ? (ultra slow)
14:23:39Suckaincase anyone cares ;D
14:25:46HClthat means the 1.8" drives will work
14:25:47bsecHCl: what are the buttons for controlling ?
14:26:08bsecthe play button?
14:26:09DMJChaha someone is creating new parts for their ihp-120 so they can get a 140's hdd into it
14:26:23HClrec = start play = A stop = B
14:26:29HClmode = select
14:26:32HCljoystick button = quit
14:26:41bsecah thx
14:26:41rasherDMJC: that is just WRONG
14:26:41 Join preglow [0] (
14:28:23bsecwoho its starts after 3 minutes =)
14:28:41HClwhat rom did you load?
14:29:11HClloads fairly ok here
14:29:24preglowafter three minutes?
14:29:29preglowisn't that a bit much? heh
14:29:33HClit is
14:29:50HClthe first screenoutput is after 14 emu cycles
14:30:03*HCl yawns and goes to fetch his iriver
14:31:14bsecyeah first screenoutput is at 14
14:31:41preglowmy iriver does the tremendous task of multiplying five by four, left shifting the result by one and printing the result right now
14:31:49*HCl times it
14:32:00bsecgame start 56, strange
14:32:06bsecit really toke 3 minutes before
14:32:16HClsomething is wrong with it.
14:32:22HClit crashed on loading rom
14:32:27preglowusb support in rockbox really bloody rocked
14:32:29HClmaybe my last memory optimalization causes trouble
14:32:35preglowit's five times faster than on the old os
14:32:41bsecim using your latest dev sources
14:33:02rasherpreglow: transfer times? Or just response to plugin/out?
14:33:12HClyea, it seems to be malfunctioning
14:33:14preglowrasher: the last
14:33:16HClhold on
14:33:22rasherpreglow: yeah, it's very snappy
14:33:30preglowrasher: when you plugged out with iriver firmware, it had to rescan the entire disk, rockbox doesn't
14:33:43preglowit's wonderful for development
14:34:14rasherYes, I love it
14:34:33bsecand finally u can create directorys, delete files etc on the player
14:34:40bsecand see all files
14:34:42amiconnpreglow: Scanning the whole disk on boot/ usb unplug is really silly design, imho
14:34:48preglowamiconn: i agree
14:35:10rasherIt's beyond silly.
14:35:11preglowamiconn: especially in irivers case, since they turn on the disk when you browse files anyway
14:35:17rasherIt's probably beyond fucking stupid as well.
14:36:19*HCl yawns
14:36:29*rasher jumps
14:37:03HClthat was one memory optimalization that messed up
14:37:39HCltakes less than 30 secs to start now
14:37:56HClbsec: reget it from my ftp
14:39:58DMJCwhy is the emulation so slow?
14:40:29preglowinterpreter core, 10mhz cpu with bogged down ram
14:40:32preglowthat's probably why
14:41:26bsecthe real cpu power is something with 140mhz?
14:42:26bsecshould be enough for gb emulation
14:42:33preglowbut we'll see
14:42:55linuxstbWhat percentage speed-up do people think we will get with a full-speed Rockbox?
14:43:23preglowwell, it should scale linearly
14:43:24HClat least 11 times as fast..
14:43:25preglowperhaps even less
14:43:30preglowdepends on the ram
14:43:50preglowbut yes, 11 times as fast isn't a bad estimate
14:44:01preglowbut we'll just have to wait and see on this one, might be several other factors
14:44:28 Quit jyp ("poof!")
14:44:34HCli'm eager to start on dynarec
14:44:45HClbut i'll refrain myself till the iriver actually runs at full speed
14:45:06linuxstbAt the current speed of my FLAC decoder, I'm looking for about a 15 times as fast increase from Linus...
14:45:49linuxstbBut I haven't looked at optimising it in any way yet, so it's looking good.
14:46:20 Quit bsec ("quork!")
14:50:23Lynx_linuxstb: can flac be more processing intensive to decode than ogg?
14:50:57preglowrockbox doesn't care whether a file is binary or not, no?
14:51:02linuxstbLynx_: I have no idea. There is more (compressed) data to process, but I'm guessing the actual decoding for FLAC is simpler.
14:51:37linuxstbI'm about finished with FLAC, and will move on to libmad and then Tremor (ogg), so I can do some speed comparisons then.
14:52:15Lynx_linuxstb: libmad is kinda important ;)
14:52:40linuxstbYes, but I'm already familiar with libmad, so that should be quick to get running on the iRiver.
14:53:44***Saving seen data "./dancer.seen"
14:55:32preglowit's very easy to build, at least
14:56:19linuxstbpreglow: it depends if all the libc functions it uses are implemented, and how it manages memory.
14:56:59preglowlinuxstb: a couple of mallocs, requires almost no libc, i've built it myself
14:57:09preglowmy iriver locked up in usb mode, woop
14:57:41linuxstbpreglow: should be very quick to implement a test decoder then.
14:57:50preglowlinuxstb: indeed
14:59:35 Quit gromit` (Remote closed the connection)
15:01:07preglowis there some kind of limit on how big my internal arrays can be in a plugin?
15:01:41 Join gromit` [0] (~gromit`
15:02:16ashridahi'd imagine that declaring huge amounts ofthings on the stack wouldn't be a good idea
15:02:39linuxstbpreglow: If you are using the standard firmware, the plugin size is 32KB maximum - code and data.
15:03:02linuxstbThis needs to be increased in CVS for the iRiver - HCl needs about 700K for rockboy.
15:03:10preglowwell, i'm below that, but my plugin quite obviously crashes rockbox
15:04:47preglowashridah: i happen to to use a 8kb array on the stack, yes :P
15:05:03preglowashridah: so i have to declare these things as static globals?
15:05:23linuxstbpreglow: I think I saw somewhere that the stack size is 2K, but I could be wrong.
15:07:40preglowyup, that was it
15:07:44preglowafter making it static it works fine
15:07:47preglowit's just for testing anyway
15:08:12preglowwee, my h120 is making a sawtooth
15:08:32preglowlike a sound
15:08:51sneakumssweet as
15:09:05HClwith hdd or what?
15:09:14sneakumswith the audio hardware, i assume
15:09:41preglowi'm working on making emac macros
15:09:45preglowso i'm diskwriting right now
15:10:16preglowi don't think linus is done mapping out the audio hardware yet
15:10:22sneakumswhat's emac?
15:10:31preglowextended multiply accumulate unit
15:10:38preglowgood for digital signal processing
15:10:52preglowcan be put to good use in more or less all of the codecs
15:11:05preglowand in stuff like equalizers and spectrum analyzers
15:11:13preglowwhich i suspect people will start asking for sooner or later
15:14:16 Join jyp [0] (
15:19:12linuxstbpreglow: based on current performance, FLAC is going to need a lot of optimising before it can be used at the same time as equalizers, cross-fading etc
15:19:27linuxstbSo we will need an emac expert on the team...
15:20:59preglowyes, i'm experimenting with it now to find out how it works
15:21:21preglowbut what makes you say this, btw? have you done any tests?
15:22:11linuxstbYes, I have a FLAC "viewer" running on the iRiver - it currently reads the .flac file into RAM, and then decodes it. Current speed is about 8% of real-time.
15:22:32rasherare you writing to disk?
15:22:43linuxstbrasher: No, just discarding the decoded samples.
15:22:51preglowthen bad news for us
15:23:11preglowthat comes as a surprise, really, i'd have thought flac was really fast
15:23:23linuxstbBut I have made zero effort at any optimisation.
15:23:38sneakumsand is that with the iriver running at 11mhz?
15:23:52rasherwhen you say current speed, is that speed *after* having read the file?
15:24:14linuxstbrasher: Yes. It reads all the file into RAM, then closes the file, then starts decoding.
15:25:25preglowi hope the ram is severely bogged down
15:25:30preglowthat would explain things
15:25:36linuxstbpreglow: So do it.
15:26:00preglowi know it's running with far more wait states than required, but not how many
15:26:20linuxstbAlso, the 8% is with a 16-bit/44.1KHz file - 48KHz or higher files will obviously be slightly slower.
15:26:41preglowbut still, we've got something to work on now
15:26:42preglowgood job
15:28:43linuxstbAlso, memory usage for FLAC is quite reasonable - the .rock itself is about 69000 bytes, and it needs about 140K of malloc'd RAM to decode a file.
15:30:06rasherSounds very good
15:30:26preglowi'm quite optimistic with regard to cpu usage
15:30:33preglowi think we'll be able to cut that severely
15:31:04HClhow does the audio interface work anyways?
15:31:08HClmaybe i can add sound
15:32:14linuxstbI haven't investigated whether these can be changed, but libFLAC uses a lot of 64-bit integers internally.
15:32:28linuxstbWhich could be another cause of the lack of speed.
15:32:46rasherlinuxstb: did you see the neuros-forum?
15:33:05rasherA FLAC developer is working on removing the need for 64 bit
15:33:22linuxstbrasher: Yes. Last thing I read there was that the developer was going to investigage the 64-bit integers, but I don't think he has yet.
15:33:44rasherSame here
15:33:56linuxstbI'll need to keep an eye on the FLAC CVS.
15:35:19preglowlinuxstb: that's quite nasty, the coldfire doesn't support those directly
15:35:52preglowbut that is an area where the emac will come in handy, the accumulators are 48 bits
15:36:08preglowwhich _might_ be enough
15:36:09linuxstbI'm impressed I'm getting 8% with all these problems :-)
15:36:40preglowhrmf, anyone really handy with preprocessor macros?
15:38:19 Part amiconn
15:40:27jypIs there an emac programming guide or the like somewhere ?
15:41:27rasherah, I do have to seed rand()
15:41:39rasherwith srand(), amazingly :)
15:41:57rasherlooking more random
15:42:57preglowjyp: i haven't found anything for the emac unit apart from the motorola programming guide
15:43:12preglowjyp: there's a paper on the mac unit, but that's a pretty much scaled down version of the emac unit
15:43:13jypYou have an url ?
15:43:20preglowsure, gimme a sec
15:43:48jypI want to see how similar it is to the CalmMAC which is in gminis
15:43:55 Join elinenbe_ [0] (
15:44:19preglowthe reference manuals would be your thing
15:45:04preglowthat's got an opcode reference
15:48:32jypWe have CalmMAC2424 on the gmini
15:48:49jypso, multiplication operands are 24 bits wide
15:49:03jypwhereas EMAC can use either 16 or 32
15:49:37preglowand they've also got some additional load magic
15:49:48preglowmaintaining compatability will be more or less impossible
15:50:07jypThe 'load magic' is probably similar
15:50:28preglowi went through the fixed point macros of libmad and concluded that it will be impossible to ease the coldfire mac instructions efficiently in
15:50:45preglowsince gcc doesn't now of the new accumulator registers
15:52:11jypYup, to do this you'll need a coldfire aware gcc
15:52:55 Quit elinenbe_ (" HydraIRC -> <- Try something fresh")
15:53:04jypIs there a compiler for emac ?
15:53:13jyp(even non-gcc)
15:54:16preglowbut not freely
15:54:22 Quit ashridah ("sleep")
15:54:56preglowit wouldn't help us anyway
15:55:04jypwhy ?
15:55:24preglowahh, if there was one freely available, yes
15:55:32preglowbut i'm willing to bet money that's not the case
15:55:47 Quit elinenbe (Read error: 104 (Connection reset by peer))
15:57:54jypWhat about this ?
15:58:10 Join Tino_ [0] (
16:00:28jypmm, not what I thought it is
16:02:35preglowbut yes, the macros i'm making now will be pretty high level
16:02:49preglowwhich is really pretty necessary, since gcc doesn't know of the acc registers
16:03:07HClwhats emac?
16:03:09preglowhmm, can preprocessor macros have default arguments? ;P
16:03:14HClthe coldfire?
16:03:20preglowdsp unit
16:04:56jypIs anything of mad & the like in cvs yet ?
16:05:33preglowit's not really needed yet, mad compiles fine out of the box
16:06:29jypOn an unrelated note, is the random number generator performance-critical ?
16:08:55HClyou'd want anything as optimized as possible
16:13:35preglowgcc inline asm is really friggin wicked
16:13:35 Join jpburton5150 [0] (
16:14:38preglowmm what?
16:14:45HClhow is it wicked?
16:14:50HClwicked as in evil or wicked as in cool?
16:14:54preglowcool, heh
16:15:01HClok :p
16:15:05preglowas in it really ties the asm statements in with the c codes
16:15:09preglowshares registers and everything
16:16:52HClhow many registers does the coldfire have?
16:17:14pregloweight data register, eight address registers
16:17:21preglowwhere the eight address register is the stack pointer
16:17:37preglowthey can pretty much be used interchangably, with address registers carrying data
16:17:51preglowin our case it also has four 48 bit wide accumulator registers
16:18:24preglowremarkably nice assemby language for this cpu
16:22:14jpburton5150has anyone succesfully compiled on Cygwin?
16:25:17 Part Nibbler ("Leaving")
16:27:30 Join pill [0] (
16:29:03 Join muesli- [0] (
16:29:51muesli-g'day mates
16:30:30muesli-nice nick hcl, are you sour like an acid? ;)
16:34:14 Quit Tino_ ("no reason")
16:34:59HCl :P
16:35:04HClnaw, i'm rather cheery
16:35:39muesli-yummy :-D
16:36:26 Join HELPME [0] (
16:37:25rasherHELPME HELPYOU
16:37:45HELPMEi have Rockbox installed on my H140, if think Rockbox messed up the disk.
16:37:48HELPMENow I canīt load the iRiver firmware!!!
16:38:21HELPMEUSB connection still works with Rockbox though
16:38:33HELPMEand I can still write to disk
16:39:05linuxstbHELPME: What happens if you hold the record button when you turn on your H140?
16:39:23HELPMEbut original firmware stalls at "Read File" (red led off)
16:39:47HELPMElinuxstb: it launches iRiver firmware
16:40:40HELPMEI tried running checkdisk in Windows (when booted from Rockbox)
16:40:48HELPMEit still does not work :(
16:41:23preglowHELPME: read file system?
16:41:32preglowHELPME: what's the status on the battery?
16:41:38HELPMEiRiver firmware stills hangs at startup when loading "read file system"
16:41:50preglowHELPME: have you done a full battery charge?
16:42:08HELPMEpreglow: battery should be ok
16:42:24preglowHELPME: yes, should, but is it? we have had another one of these, and the battery was the problem
16:42:32linuxstbI think Rockbox uses less power than the iRiver firmware, so that could be problem.
16:42:32HELPMEi can't really check it because Rockbox still don't show it correctly
16:42:48preglowHELPME: well, then charge it for an hours time and try again
16:43:16preglowHELPME: the hardware charges itself, so doesn't depend on whether you run rockbox or original firmware
16:43:32HELPMEiRiver firmware normally just says "low battery" or does not turn on
16:43:45HELPMEi don't think that is the problem
16:43:55preglowHELPME: yes, yes, but as i said, we have had another case resembling yours, and then the battery was a problem, so please try
16:44:11preglowit's worth a short anyway
16:44:17preglowshot, even
16:44:33HELPMEright now I am backing up my data
16:44:36preglowjust let the adapter charge for a quarter or something, then try again
16:44:55HELPMEI will charge and try to launch iRiver firmware again later
16:46:17preglowrockbox messing up the disk really shouldn't happen, the fat driver should be the same as the one on archos, and that's gotten a lot of testing
16:46:24preglowplus, i've _really_ abused my player, and nothing has happened yet
16:46:32preglowbut time will tell
16:46:42preglowi think i'll go for a walk in the snow
16:47:08jpburton5150i _still_ cannot get rockbox to compile on cygwin
16:47:19jpburton5150it say stuff about the unrecognized architecture
16:47:30jpburton5150but ive got m68k-elf-gcc set up
16:47:37preglowjpburton5150: i tried as well, but gave up
16:47:54jpburton5150haha... so you cant compile either then?
16:48:34preglowi've got a linux box
16:48:35preglowi use that
16:48:44preglowi'm not particularily fond of cygwin anyway
16:49:14jpburton5150in a bit i might reinstall a linux partition
16:49:19jpburton5150any recommended ones?
16:49:28rasherubuntu or debian
16:49:32rasheris my preferred choice
16:49:36jpburton5150k yeah
16:49:53preglowbut, yes, bbl
16:50:08preglowubuntu is nice
16:50:23 Join Tang [0] (
16:50:32TangHi :)
16:53:13 Quit jyp ("poof!")
16:53:48***Saving seen data "./dancer.seen"
16:55:59 Join Schnueff [0] (~mah@
16:57:38 Join webguest33 [0] (
17:04:16 Quit webguest33 ("CGI:IRC (EOF)")
17:06:35 Quit muesli- (Read error: 113 (No route to host))
17:08:45 Nick Lynx_ is now known as Lynx_awy (HydraIRC@
17:12:01 Quit HELPME ("CGI:IRC")
17:12:04 Join HELPME [0] (
17:18:12rasherHELPME: any luck?
17:18:53 Join Luke1 [0] (
17:19:26Luke1would someone be able to take a look at this eBay auction for a 20 GB FM Recorder and give their opinion?
17:19:33Luke1It says:
17:19:33Luke13 DAY SERVICE
17:19:44Luke1I wonder what that means is wrong if it hangs like that
17:21:16*rasher has no idea
17:21:50HELPMENO! :(
17:21:58Luke1All right. I have an extra 20 GB hard drive, so if that needed replacing it seems like that wouldn't be too bad
17:22:10Luke1but I have no clue what causes a unit to hang like that
17:22:15HELPMEI just feel like I bricked my iRiver :(
17:22:31HELPMEbetter not delete rockbox.iriver :)
17:22:54rasherHELPME: it boots and can write to disk, I doubt it's broken beyond repair
17:23:08HELPMEhope not
17:23:09linuxstbHELPME: If I was you, I would just leave it charging and wait for LinusN to appear.
17:23:11rasherI'm guessing you'll have to wait for someone more in-the-know to come around
17:24:01HELPMEhow does your iRiver
17:24:13HELPMEoriginal firmware behave?
17:24:19rasherMine works as usual
17:24:24linuxstbNever used it...
17:24:27rasherh120 though
17:24:58linuxstbI suppose I could test it for you now - I have a H140.
17:25:08HELPMEmine turns off the backlight under "READ FILE SYSTEM" after the harddisk aktivity stops
17:25:10rasherHELPME: what did you say the error was?
17:25:38rasheroh, it stops in the "read file system" stage?
17:26:11HELPMEit never pass "READ FILE SYSTEM" - harddisk stops reading (and red led) and the backlight is turned off
17:26:23HELPMEI waited 20-25 min.
17:26:35rasherThis shouldn't happen, but you could try running scandisk on it, or even formatting it I guess (after backing up data :)
17:26:51rasherdid you do anything in rockbox before this happened?
17:27:08HELPMEmount /mnt/iriver
17:27:17HELPMEuploaded rockbox
17:27:26HELPMEumount /mnt/iriver
17:27:46HELPMEi tried scandisk in Windows
17:27:59HELPMEdoes linus have a scandisk
17:28:06rashernot for vfat no
17:28:07HELPMElinux :)
17:28:25rasheroh there is?
17:28:39HELPMEmaybe I should try that
17:28:45rasherISTR someone else having fat corruption problems
17:28:50rasheralso using linux
17:28:58rasherand someone mentioning a vfat bug
17:29:04rasherbut I'm not too sure about the details
17:29:48HELPMEafter USB connection, Rockbox release the connection in 1 sec. or less
17:30:00HELPMEiRiver firmware takes up to a minute
17:30:08rasheryes, it's very snappy :)
17:30:19rasherthat's because iRiver reads the entire disk contents
17:30:24rasherfor some reason
17:30:45HELPMEI donīt think so, the complete FAT maybe
17:31:01sneakumsthe fat is just a map of used and free clusters, it shoudlnt' ahev tor ead that at al
17:31:07sneakumsit seems to read the entire directory tree
17:31:18HELPMEon the FAT, right?
17:31:38sneakumsno, the fat is just a map of used and free clusters.
17:31:43sneakumsthe directories are separate
17:32:41HELPMEare you sure - that seems pretty stupid to me
17:33:40sneakumsas far as i can tell, that is what the iriver firmware does
17:34:13HELPMEhow can you tell - are you an FAT32 expert?
17:35:32sneakumsno, but i know the difference between a file allocation table and a directory tree, and can make a reasonable guess as to which one would need to be read when
17:42:09HELPMEhere is some FAT32 info:
17:42:30HELPMEnotice the "Directory Table"
17:42:44HELPMEwhich is part of FAT32
17:42:56sneakumsbut it isn't the FAT
17:43:36sneakumsFAT the file system was so named because it uses a file allocation table to track the used and free clusters
17:43:51sneakumswhich i think is the source of the confusion here
17:52:27HELPMEdamn, dosfsck - FAT32 not supported.
17:52:49HELPMEnewer versions must exists
17:52:57sneakumsthe one in Debian is version 2.10
17:53:03sneakumsat least
17:53:14sneakumsthere is a package called dosfstools that has a program called dosfsck
17:53:24sneakumsand the version of the dosfstools package is 2.10
17:53:47sneakumsi've never actually tried it
17:53:47HELPMEI have redhat 9, it's is not installed
17:54:03HELPMEa link?
17:54:04sneakumsworst case, you can download the source an dcompile it yourself
17:54:48 Quit Luke1 ()
17:54:49sneakumshere's the source:
17:55:16HELPMEthanks, i'll try it
17:56:42 Join jyp [0] (
18:02:01preglowHELPME: i'm guessing a format should fix it
18:02:08preglowHELPME: is rockbox able to access the disk?
18:02:54HELPMEyes, can access it
18:03:06HELPMEbut I donīt think a format is a good idea
18:03:35HELPMEthen I might might launch Rockbox (and of course iRiver firmware)
18:04:05 Quit jyp ("poof!")
18:04:09HELPMEI am running dosfsck 2.8 now
18:04:16thegeekif you format the drive
18:04:32preglowHELPME: sure, it would be a last resort
18:04:41HELPMEI will loser rockbox.iriver
18:04:46preglowyes, i know
18:04:47thegeekbut hey
18:04:49HELPMEI still have Rockbox
18:04:52thegeekare you running the latest kernel?
18:05:02HELPMEyes, last kernel
18:05:04thegeekthere was a kernel bug that fucked things with iriver
18:05:08sneakumsafter you format, just copy it over again
18:05:14preglowthegeek: what kernel version?
18:05:21thegeeklet me see
18:05:35thegeek2.6.5 ?
18:05:48HELPMEdosfsck found something
18:05:54 Join jyp [0] (
18:06:34HELPMEdosfsck: Reclaimed 847 unused clusters
18:08:21HELPMEdosfsck: Free cluster summary wrong (211974 vs. 212821)
18:08:30HELPMEI corrected it
18:08:35preglowwell, does it do any good?
18:08:53HELPMEstill testing
18:10:03HELPMEI ran dosfsck again and got the same error :(
18:10:28HELPMEu think so?
18:10:38preglowno, i was being sarcastic
18:11:11preglowbut i don't think the free cluster summary should be very decisive
18:11:13thegeekjust format it
18:11:17rasherI'd just format
18:11:28thegeekand copy over rockbox if it does not work
18:11:41preglowyou can't copy over rockbox after having formatted it
18:11:43preglowthat is
18:11:49preglowformat it, copy rockbox, then restart it
18:11:54rasherHELPME: can you bring up iRiver in usb-mode?
18:12:03thegeekobviously he can
18:12:06HELPMEand hope my computer don't crash :(
18:12:11rasherthegeek: obviously?
18:12:12thegeeksince he is running dosfsck
18:12:14HELPMErasher: yes
18:12:20rasherhe could be using rockbox usb mode
18:12:43rasherso there's no way you can lose usb mode
18:12:52preglowahh, i thought he was using usb mode in rockbox
18:13:01thegeekI thought usb-mode was hardware
18:13:11thegeekbut yeah
18:13:15preglowwell, it is, but i needs software to activate it
18:13:16HELPMErasher: no, I can't bring up iRiver firmware USB mode
18:13:27thegeekwell that changes things a bit;)
18:13:30rasherit does
18:13:32preglownot really
18:13:37preglowyou can format it, put in rockbox, and then restart it
18:13:53 Quit jpburton5150 (Read error: 54 (Connection reset by peer))
18:14:13preglownothing says you have to restart the player after having formated it
18:14:37HELPMEbut what if Rockbox USB is messing up my drive
18:14:38preglowHELPME: this was a h140?
18:14:45HELPMEpreglow: yes
18:14:48preglowHELPME: it can't, usb mode is hardware
18:15:08preglowHELPME: the only thing rockbox does is stop accessing the hard drive so that the computer can do it itself
18:15:59HELPMEa FORMAT through Rockbox might not be failure free
18:16:06preglowwhat makes you say that?
18:16:28preglowthere is no 'through rockbox', all the accesses from the computer goes through hardware
18:16:32preglowrockbox doesn't enter into it
18:16:39preglowit just sits dumbly
18:16:59preglowthe usb->ata is done by a bridge chip
18:17:22HELPMEI like to hear Linus point on this before I do anything stupid
18:17:31preglowas would i ;)
18:20:03 Join ripnetUK [0] (
18:20:15 Quit ripnetUK (Client Quit)
18:23:52 Join coob [0] (
18:25:03 Quit pill (Read error: 104 (Connection reset by peer))
18:29:10 Join hubble [0] (
18:29:35hubbleyes, finally got my first RXAK on the I2C interface! =)
18:29:42 Join webguest87 [0] (
18:31:53rasherhubble: What does this mean?
18:32:39 Join pill [0] (
18:33:32hubblerasher: not much, but it's a start getting the i2c driver to the audio chip working
18:34:17rasherAny reason why I should not modify the pluginindex page to include iRiver info?
18:34:46Quelsarukpluginidex page?
18:34:58rasherand the subpages I guess
18:35:14rasheror should that only be done when the port is ready?
18:35:33rasherAt any rate, that page would look so much nicer as a table
18:36:19rasheroh, it's magic
18:36:48Quelsarukthat page must be changed
18:36:57Quelsarukas archos player can use Snow right now too
18:37:03Quelsarukand maybe more plugins
18:37:08rasherwell the page itself is dynamic
18:37:49rasherso you'd just have to edit the PluginSnow page
18:37:50rasherand so on
18:38:24preglowhubble: do you know how many components are connected by i2c? i know of the codec and the eeprom
18:38:30rasherBut I'd like to try and change the page to be a table
18:38:55rasherguess I'll try and do it - and I'll reverse it if it turns out to be rubbish
18:39:26hubblepreglow: atleast the codec and fm chip
18:40:29preglowthe eeprom is i2c as well, but i don't think rockbox will use that
18:40:56hubblepreglow: ok.. going to see if i get an ack from the fm chip soon now
18:41:21preglowhubble: that'd be really cool
18:41:32preglowstuff really did accelerate after having a bootloader
18:42:20hubblebut audio support is pretty far away.. we need a i2s driver aswell.. maybe should update to todo wiki about that =)
18:42:32HClthat reminds me
18:45:23HClhow do i feed sound to rockbox?
18:45:37hubbleHCl: via i2s
18:45:58hubbleHCl: which can be driving using DMA transfers
18:46:04hubblediving = driven
18:46:25linuxstbHCl: The "audio API" is still to be decided. How would you like it to work?
18:47:07HCllinuxstb: well, hold on a sec..
18:48:05HClwell, i haven't really looked at it yet
18:48:11HClbut pretty much, something that'll play pcm data
18:48:20HCltill it reaches the end of it
18:49:09HClWhenever realtime sound output is operational,
18:49:09HClpcm_submit is responsible for timing, and should not return until it
18:49:09HClhas successfully processed all the data in its input buffer (pcm.buf).
18:49:15HClfrom the gnuboy hacking documentation
18:49:44 Join webguest28 [0] (
18:49:56linuxstbHCl: Yes, I read that as well. I guess that the audio API needs both blocking and non-blocking audio functions then.
18:50:36linuxstbThe audio system will also need to know the sample-rate, number of channels etc of the PCM data.
18:50:55linuxstbBut it's all straightfoward stuff.
18:51:49 Join iriver [0] (
18:52:05 Quit webguest28 (Client Quit)
18:52:17iriverI'm interested in the audio API too
18:53:36iriverHave you looked at the MCF5249 user manual chapter 17?
18:53:46HCli can't say i have
18:53:48 Join Stryke` [0] (~Chairman8@resnet-241-86.resnet.UMBC.EDU)
18:53:52***Saving seen data "./dancer.seen"
18:53:56linuxstbI think I browsed it a couple of months ago...
18:55:20linuxstbAs the Wiki page says, there are probably going to be three separate subsystems - the "audio-playback subsystem", the "codec subsystem" and the low-level hardware drivers.
18:55:30iriverFigure 17-1 (Audio Interface Block Diagram) is quite nice
18:55:51linuxstbPersonally, I'm concentrating on the codecs at the moment. I'll leave the hardware to Linus and the audio-playback to people who are more familiar with Rockbox.
18:56:19iriverWe need some queue struture which can signal when they are almost empty
18:56:32*HCl agrees with linuxstb
19:00:50 Quit pike (Read error: 60 (Operation timed out))
19:01:42iriverwho are the audio-playback people of Rockbox?
19:01:48iriverLinus :)
19:03:14 Part webguest87
19:03:36Tanglol you don't let him some time for real life? lol
19:03:43HCl :P
19:04:09iriverit is great that hubble worked on the i2c
19:04:24iriveris working on.. that is :)
19:04:36 Quit hubble ()
19:04:49linuxstbiriver: I think I'm the only person actively working anything to do with the software parts of audio playback at the moment.
19:04:50HClwhats rxak?
19:05:19coobare you guys still planning on using MAD for mp3 playback?
19:05:24preglowcoob: yes
19:05:34iriveri am looking on the i2s and mcf5249 low-level audio driver API
19:05:35preglowcoob: it's a great decoder, and i don't see any other libs as better alternatives
19:05:40linuxstbcoob: yes - I'm working on getting a test decoder working at the moment.
19:06:30preglowHCl: i2c signal sent from transmitting part, i think
19:06:41coobah right - I have no idea of the comparison between CPU power of the iPod's dual ARM's and the coldfire stuff, but I'm not sure how much optimisation of MAD you'll have to do
19:07:12preglowcoob: i don't know how much we HAVE to do, but i'm having a look at the dsp parts of the coldfire as we speak
19:07:31coobthere's dsp parts? lucky bastards :)
19:07:43linuxstbAs HCl (I think) said earlier today, we want to optimise everything as much as possible - the more efficient the playback, the more we can do at the same time (e.g. games).
19:07:46preglownot much to speak of, there is a mac unit
19:07:57preglowwith 48 bit wide registers
19:07:59preglowmore would be nice
19:08:00HClnot only that, the more efficient, the less battery we use
19:08:06HCland the more runtime we get
19:08:13linuxstbHCl: Absolutely.
19:08:17Tangand it will let more battery power too
19:08:31*rasher has an argument with TWiki
19:08:34Tang(oups i was too late)
19:09:35TangABout the audio codec
19:09:55iriverI was thinking about a general decoder function interface
19:09:56iriverwe could use function pointers for replacing the codec functionality
19:10:04Tangan HA memeber was wondering
19:10:22Tangif the seamless playback was lnked with the decoder libbrary
19:10:26Tang or not
19:10:47Tangsince mp123 (used by foobar) is seamless
19:10:47iriverlinuxstb: what function interface would you suggest?
19:10:54Tangnut not MAD
19:14:36preglowit will be seamless
19:15:16preglowthe decoder library itself doesn't matter, we'll have to write the code that ties different tracks together
19:15:26preglowand we'll make it gapless
19:15:31linuxstbStarting at the lowest level, I think there's going to be a PCM buffer - the low-level audio driver will run in it's own thread, constantly trying to empty that buffer, and the audio system will be doing it's best to make sure it's never empty.
19:15:52linuxstbIf the audio system succeeds, then playback will be gapless.
19:16:07HClsounds great.
19:16:16sneakumsand the winning will be large
19:16:20HClespecially because i'd be able to link gnuboy directly to that buffer xD
19:16:41linuxstbOnly possibly flaw in playback could be when the sample-rate changes (e.g. from 44.1KHz to 48KHz).
19:16:59HClyou need to be able to set the samplerate of the pcm buffer of rockbox
19:17:41linuxstbI think we need more detail than that - i.e. the PCM buffer will contain blocks of data, and each block can be a different samplerate.
19:18:01iriverfunction pointer info:
19:18:37preglowlinuxstb: i really don't think a gap at sample rate change matters, stuff that is supposed to be gapless won't have different sample rates anyway
19:18:54linuxstbpreglow: Agreed.
19:19:29preglowit's not worth the pain of having to keep track of the position in the buffer at which we'll have to change sample rate
19:20:13iriverThere will will need to be small PCM buffer internally in the MCF5249
19:20:35linuxstbpreglow: Not sure if I agree. I think it depends how big the PCM buffer is going to be. We don't want to have to wait for it to empty before starting to decode the next track, just because it is using a different samplerate.
19:20:38preglowthe area around the sample rate change probably wont be guaranteed to be up to par by the phillips codec either
19:20:51iriverIt will be transfered directly via DMA to the I2S bus out to the UDA1380
19:21:02*rasher updates PluginIndex ( )
19:21:17rasherIt's slightly less dynamic now, but much easier to read
19:21:37preglowrasher: looks great
19:21:53preglowrasher: that mess really did need to be tableified
19:21:54rasherShould I add iRiver and Gmini to the table?
19:22:00preglowrasher: i don't see why not
19:22:06rasherOkay, I'll go ahead
19:22:37preglowthe information is already there, might as well complete it
19:22:47iriverDMA is performed on the internal audio bus of the MCF5249
19:23:08preglowlinuxstb: i don't think we'll end up with a very large buffer
19:23:20preglowwe might end up with several, as a matter of fact
19:23:27preglowsample rate can be a buffer tag, then
19:23:51iriverthe large buffer will be for undecoded codec data (raw mp3, ogg, flac etc.)
19:23:52preglowif the buffers are small enough, it's more than adequate to only do that check when we change buffers
19:23:57linuxstbYes, "current track" and "next track" PCM buffers would work well.
19:24:31preglowlinuxstb: no, i was thinking a bit more granular than that, several buffers no matter what state you're in
19:24:35preglowthis is the way most audio apis do it
19:25:06linuxstbSo why would we need more than one PCM buffer?
19:25:34sneakumsthe difference between a single buffer with tagged blocks and multiple tagegd buffers is not so obvious to me
19:25:36preglowit's good to have another buffer to work on when you send it to the buffer you've just finished to the codec
19:26:28preglowthis is the way it's done on most audio apis on computers, but there might of course be good reason to not do it this way on a portable
19:26:46linuxstbsneakums: The blocks can be of varying size, the PCM buffers will (I assume) be a constant size.
19:27:04iriverI think a small PCM transition buffer is needed to ensure gapless/crossfading
19:27:17rasherWhat's the name of a joystick-press on the iRiver?
19:27:37preglowlike you have one buffer, fill it, send it to the dac, and instead of waiting until the das is done with it, you can start on a new buffer straight away
19:27:45preglowthe dac
19:27:50preglowi still type like a pig
19:28:24linuxstbpreglow: are you talking about tiny PCM buffers inside the low-level audio driver?
19:28:44preglowi'm talking about what the audio codecs will pass to the low-level audio driver
19:30:10linuxstbI'm thinking that from the point of view of the high-level codecs and audio API, then will just have a PCM buffer to fill with data. They may themselves need additional PCM buffers, and the audio driver may need additional buffers, but there will be one main "transfer PCM buffer".
19:31:15preglowthe audio driver should not need additional buffers
19:31:33preglowit should just dma the memory it gets straight to the codec chip if it can take more data
19:31:42preglowbarring the need for conversion of the data, of course
19:31:55preglowbut we should try avoiding that
19:32:42linuxstbSo starting from the bottom, do we agree that there is a single PCM buffer the audio driver reads from, or do you multiple buffers?
19:32:54linuxstb^do you want multiple buffers?
19:33:34preglowit's the same thing, really, it's just that with one buffer, a fast decoder will constantly have to watch that it doesn't write into an area of the buffer that hasn't yet been sent to the codec chip
19:33:56preglowbut of course, it can all be wrapped in book keeping functions
19:34:14preglowi don't think i'm very seriously against having one buffer, this is just the way i'm used to doing it
19:34:37preglowif the guys who now the hardware better is biased against one of the methods, that's what we'll picj
19:35:23linuxstbBut it should be (mostly) independent of the hardware - we need to abstract the API higher than that.
19:35:52rasherThe keys for the chessclock plugin on iRiver are crazy
19:35:57preglowyep, it will be hardware independent no matter how you do it
19:36:13preglowrasher: then port it to use better keys ;)
19:36:19sneakumsthe idea of threads passing buffers back and forth like a bucket brigade makes a pretty picture in my head, i will say that
19:37:03irivercodec independent is also important - we need a common interface
19:37:06linuxstbsneakums - yes, we need to try and keep it as simple as possible.
19:37:11rasherpreglow: I'll have a look at that later, for now, I be wikihacking
19:37:23preglowsneakums: yes, i like it as well, and the decoder has to call a function to pass it a free buffer, that's where things will block if you've filled all the buffers you can and have to wait before you can decode more
19:37:58linuxstbiriver - yes, that's sort of what I'm looking at in detail at the moment. I'm implementing various decoders using different libraries and then I'll suggest a sensible common API.
19:38:28preglowi think we'll need to codec layers as well, one for streaming codecs, and one for non-streaming ones
19:38:38preglowtwo codec layers, damn, i have to start reading shit before i press enter
19:38:57 Join mecraw [0] (
19:39:11iriveri don't quite follow
19:39:31preglowmp3, ogg = streaming. xm, sid, mod = non-streaming
19:39:50preglowonce the latter is loaded, rockbox doesn't have to think about disk accesses until the track is done
19:40:28preglowwith the former, rockbox has to constantly watch if the audio buffer is nearing its end, and load more data if it's needed
19:41:34rashercould anybody explain this to me: #define CUBE_Z_INC (BUTTON_ON | BUTTON_UP)
19:43:14rasherah, got it
19:43:53iriverpreglow: yeah, there is a lot of things to think about. Didn't think about sid, mod
19:45:15iriverto me sid and mod will also be streaming once they have been through the synth
19:46:48linuxstbNo, I think the difference is that with a streaming codec, you read (and then discard) X bytes of the input stream, and then that generates Y bytes of PCM data.
19:47:09linuxstbWith a SID/MOD/etc file, the input file isn't a stream.
19:47:28rasherCould we honestly claim that metronome is working on iRiver?
19:48:18preglowgotta go the shop, brb
19:48:21rasherI guess I'll put it down as supported
19:54:30rasherI don't get the tap mode..
19:55:14rasherwould perhaps make more sense if I had sound
20:02:02preglowwhat's tape mode?
20:02:09preglowi can't start it
20:02:12preglowtap mode..
20:02:41iriverI guess it is able to tell the tempo you are tapping
20:02:49rasheror should be, at least :)
20:02:52preglowyes, sure, but i don't get how to start it
20:02:57iriveri can't get it to work neither
20:03:13rasherI just went by the keyapping in the source
20:04:22HClwhats metronome again?
20:04:37rasherA thing that goes "tick, tick, tick" at a specified interval
20:04:42preglowthe thing on a piano that goes tick tock
20:04:48Tangpreglowit will be seamless
20:05:05preglowTang: yes, what about it?
20:05:19Tangno i wanted to thak you for the answer
20:05:28preglowno problem
20:05:33Tangi just failde my copy/cut
20:05:37preglowthere's no point in having gaps in the sound
20:05:56TangThanks i'll transmit this on HA
20:07:38Tangyes Hydrogenaudio
20:07:54Tangwith Linus agreement
20:08:25preglowlinux himself has said that rockbox will use the extra lame header data, actually
20:08:26Tangi tried to relay the latestrockbox iRiverport progress
20:08:42preglowthis is getting confusing
20:08:44Tangi see
20:08:49preglowlinusN, actually :P
20:08:55Tangops sorry
20:09:15Tangas i said i was relaying the latest progress at HA
20:09:30Tangi hope this will motivate some skilled coders
20:09:38Tangto make tehir hands dirty
20:09:42Tangin the codec part
20:09:58Tang(especially for MPC playback)
20:11:01preglowmore help is appreciated, of course
20:11:41Tangi've jsut put the info
20:11:55Tangi'll see if there is some reaction
20:12:07Tangand if not I'll try to contact some guys directly
20:13:08Tang(especially Peter pawlowski foobar2k devlopper, who released the integer libmusepack library)
20:13:12 Join xen` [0] (
20:13:49preglowyes, that's what we'll use
20:15:01linuxstbTang: Something no-one has really considered is real-time encoders for Rockbox. Recording will come first as WAV-only, and we will then need to think about a lossy encoder that can work in real-time on the iRiver. Maybe the HA people have some suggestions for that.
20:15:24preglowthat will most certainly require custom code
20:15:43TangOkay linuxstb i'll ask at HA place
20:16:13Tangbut indeed i'm not aware of anything "builtin" for lossless real time encoding
20:16:26Tang(integer one)
20:17:07linuxstbI'm thinking about lossy, not lossless. For lossless, we have WAV.
20:17:15preglowand with a bit of luck, flac
20:17:48linuxstbA flac encoder would be nice, buit the space savings don't make it a high priority IMO.
20:18:09linuxstbI would prefer 100% stable and reliable WAV recordings.
20:18:11 Join XShocK [0] (
20:18:37preglownothing is a high priority right now, heh
20:18:39preglowwe need audio
20:19:04preglowbut yes, a realtime lossy codec would rock
20:19:06linuxstbAlso, battery power is an issue with field-recording - so WAV would hopefully allow the CPU to slow right down to the minimum.
20:19:11TangAh sorry i read too quick
20:19:15rashernote to self: change keymappings for stopwatch (and see about making it use the entire screen)
20:19:19Tangi understood lossless
20:19:25preglowlinuxstb: too bad the hard drive is what sucks most power :/
20:19:34Tangindeed i believed there where some libs for encoding
20:19:42Tang'ill ask for both if you want
20:19:50preglowTang: there are, but they are floating point and too slow for us
20:20:00Tangokay i undertsand
20:20:02preglowTang: mostly, at least, i don't know much about it
20:20:06Tangi'm goona ask
20:20:11linuxstbMost codecs are optimised for fast decoding and slow encoding. Not good for the iRiver.
20:20:32Tanghum for lossless you talk i guess
20:20:49linuxstbTang: No, both lossy and lossless codecs are like that.
20:20:54Tangah okay
20:20:58Tangi'm gonna ask
20:21:09Tangno fear( (lol)
20:21:29Tang(just joking, HA people aren't cruel ;))
20:35:01 Quit Schnueff ("leaving")
20:39:29rasher looking much better now
20:41:05Quelsaruklooking better
20:41:22rasherAnd with up-to-date iRiver info
20:41:29rasher(I hope)
20:41:43Quelsarukrasher: can you add player to the snow and cube plugins?
20:41:51rasheroh right
20:44:11rasherI don't know the keynames though :\
20:44:57Quelsarukthen, don't do it
20:45:12rasheryou do it then :)
20:45:27TangIndeed great job for pluginindex rasher!
20:45:27Quelsarukno way
20:46:02Tangi 've created the asking thread for lossless integer encoder under open source licensing
20:46:26Tangand asked the same for lossy in the HA thread dedicated to the Rbx iRiverport
20:46:39Tangi'm gonna put the link if you want
20:47:34HELPMEToo all linux users: which device /dev/ do you have associated to your iRiver USB mount in /etc/fstab?
20:48:03linuxstbHELPME: I think it depends on the other hardware in your PC - for me, it's /dev/sdb1
20:48:15QuelsarukHELPME: i have my archos in /dev/sda1
20:48:18TangThe lossless thread:
20:48:53TangThe same question for lossy encoder in the rbx iRiverport at HA:
20:49:12TangFeel free to tell me if you see something to be corrected
20:49:15Tangi'll edit
20:51:47linuxstbTang: They seem the right questions to me, thanks.
20:52:01TangOkay linuxstb
20:52:07Tangdon't thank me
20:52:19Tangit's me to thank all you ;)
20:53:38 Join R3nTiL_ [0] (~zorroz@
20:53:54***Saving seen data "./dancer.seen"
20:54:14HELPMElinustb & Quesaruk: thanks, I use /dev/sda1
20:54:34HELPMEso that is not the reason why my H140 disk is seemed up
20:57:48 Quit Sucka ("( :: NoNameScript 3.81 :: )")
20:59:27QuelsarukHELPME: do you still have that problem with bad sectors?
21:00:19HELPMEi will just test
21:01:26preglowwhat, doesn't c guarantee use of arithmetic shifts when you shift signed numbers?
21:01:49QuelsarukHELPME: is your drive under warranty? did you make something that can void that warranty? if not, you could ask the company to repair it
21:02:41HELPMEI install Rockbox bootloader :)
21:02:55HELPMEi don't think iRiver will be happy about it!
21:03:19sneakumswell, they already have your money.
21:03:53Quelsarukcan't you take out the bootloader?
21:03:59preglowhe can't flash it
21:04:01linuxstbHELPME: Have you reformatted yet, or are you waiting for Linus's advice?
21:04:11HELPMEwaiting for linus
21:04:17preglowi don't think he'll be on tonight
21:04:24preglowleast, doesn't seem like it
21:04:50HELPMEQuelraruk: read today's irc log
21:04:56linuxstbSo there's still one option yet - as others have said, there shouldn't be a problem to reformat and then copy rockbox.iriver back. But there's no harm in waiting.
21:05:58preglowrockbox.iriver and .rockbox dir, yes
21:06:18preglowrockbox with no .rockbox dir just fails miserably
21:06:44preglowdoes anyone know how long a 68k 'int' is?
21:06:58sneakumsi'd be surprised if it weren't four bytes.
21:07:02preglowyes, so would i
21:07:08preglowbut that would certainly explain this
21:07:25HELPMEI fixed the clusters errors previously, but now I get them again "Free cluster summary wrong (221742 vs. really 221739)"
21:08:44HELPMEI tried deleting a lot of my data from my H140, but when I tried to shutdown Rockbox it hung showing "Shutting Down"
21:08:57HELPMEhad to reset
21:10:10HELPMEI have also experienced that Rockbox USB also crashes Windows when inserted
21:10:21preglowsizeof(int) = 4, yes
21:10:32linuxstbpreglow: I've just done a check, and "int" is 4 bytes, "short" is 2 bytes and "long long" is 8 bytes.
21:10:42linuxstbBeat me to it...
21:10:43preglowyes, i did so as well
21:11:06preglowHELPME: again... rockbox has nothing to do with the usb
21:11:23linuxstbBut I've noticed that the Rockbox snprintf doesn't support floats or "long long".
21:11:32preglowit just passes control to the usb gateway chip
21:11:40Tanghi HELPME, sorry for your iHP
21:11:44preglowlinuxstb: floats i understand, long longs you should never use
21:11:56rasherIs there any reason the stopwatch shouldn't be keeping more than 10 laptimes?
21:11:59Tangdon't knew the whole story
21:12:03preglowi've noticed it doesn't understand %i
21:12:11Tangis it linked with Rbx or something else?
21:12:39linuxstbpreglow: I was just using them for debugging output.
21:12:48linuxstbWell, trying to use them...
21:13:03preglowlinuxstb: yes, but if you debug something by printing long longs, you're obviously using them :PP
21:13:37linuxstbI was trying to use a long long to calculate the % decoding speed - to prevent overflows.
21:13:51linuxstbBut I changed the formula instead.
21:14:01preglowthat's more qualified use
21:14:48linuxstbBut as I said, libFLAC is full of 64-bit integers, so that's an area to investiage.
21:16:48linuxstbI'm eagerly waiting for Linus to work his magic.
21:17:00linuxstb... and get the iRiver running at full speed.
21:17:58 Quit R3nTiL_ ()
21:18:39*rasher has a look at solitaire.c
21:19:08rasherthat sure is a lot of keybindings
21:21:21izzylinuxstb: have you measured if flac requires more cpu time than mp3 to decode? with average compression on both
21:21:52linuxstbizzy: I'm in the process of getting libmad running on the iRiver as we speak. When I finish that, I can do some speed tests.
21:22:27izzyok, that's good :)
21:22:31izzyhow about on x86?
21:22:37linuxstbWhat would you consider "average compression" for MP3? 128kbps?
21:22:59linuxstbizzy: No, I don't think that will be that useful.
21:23:29izzymaybe more, 168kbps perhaps
21:23:49izzyI was just wondering because I haven't tried flac even on pc :)
21:24:16izzyI use ogg for my own encodings
21:24:54izzycoldfire manual says "MP3 decode requires less than 20MHz CPU bandwidth"
21:26:05linuxstbWell, on my 2.8GHz Xeon, a 4 minute FLAC file takes about 2 seconds to decode.
21:26:34rasherUsing linux? Time it
21:27:09linuxstbrasher: I did. "user 0m1.950s"
21:27:20rasherYou sounded so vague :)
21:27:21linuxstbJust trying OGG now...
21:28:02linuxstb2.594s for a 192kbps OGG
21:29:08sneakums3.3 secs on a 1.5GHz G4 for a 4.5 min ~85kbps vorbis
21:29:19sneakumswith ogg123
21:29:24sneakumsnow there's a useless datapoint for you
21:29:37Tangogg require more processing than mp3
21:29:42Tangit's well known
21:30:16linuxstbJust trying MP3 now - lame "−−alt-preset extreme" gave me 193.4kbps (similar bitrate to the ogg file)
21:30:28linuxstb^sorry, −−alt-preset standard
21:31:29rasherhey, putting "cube" in high-speed mode and limiting the rotation to one axis looks sortof cook
21:31:57linuxstb"madplay -o wav:" gave me 2.834s. So that compares with 2.594 for OGG and 1.950 for FLAC in my very non scientific tests.
21:34:47linuxstbAnd a 192kbps MP2 file was 2.275s - so unsurprisingly MP2 is simpler than MP3 and OGG.
21:35:08linuxstbLet's see how they compare on the iRiver.
21:35:11izzyI would also guess that encoding of flac is easier task than lossy formats
21:35:31linuxstbizzy: Don't make me do all that again :-)
21:35:44izzyI didn't mean that ;)
21:36:31linuxstbBut yes, I think the flac encoder is the quickest of all of them.
21:36:38Tanghum not sure encoding in flac 'll be easier than in lossy format
21:36:57rasherCouldn't the 64bit ints be messing things up?
21:37:15rasheror was that in the encoding?
21:37:25rashernevermind me, I'm babbling
21:37:34linuxstbOK :-)
21:37:56linuxstbWhat did you mean by "messing things up"? Do you mean slowing it down, or causing other problems?
21:38:11rasherslowing it down
21:39:06linuxstbYes, I'm sure it is.
21:39:18izzyflac is still in integers only?
21:39:44sneakumscomputers should all be 64-bit by now anyway
21:39:51linuxstbThe decoding is, but the encoding isn't.
21:39:51sneakumsit's the future, dagnabbit!
21:40:09izzylinuxstb: ok
21:41:02izzyhas anyone tried to compile ogg/tremor yet?
21:41:23linuxstbNo, but it's next on my list after libmad.
21:42:12sneakumsfirst you're flac'y, then go mad, and finally you just sit there and tremor.
21:42:29linuxstbI'll be waving at the end of it.
21:51:23preglowsneakums: hahah
21:54:37 Quit Ka (Read error: 110 (Connection timed out))
21:54:54 Join Ka [0] (
22:12:14HCl :P
22:12:39rashermmm solitaire
22:20:03HELPMEPROBLEM FIXED - My H140 finally started with the iRiver firmware... I don't know what was wrong
22:20:35HELPMEI ran dosfsck several times
22:21:00HELPMEtried starting the iriver firmware 20-30 times with no luck
22:21:12HELPMEstrange :)
22:21:35HELPMEgotta go
22:21:43 Quit HELPME ("CGI:IRC")
22:36:28 Quit Stryke` ("Friends don't let friends listen to Anti-Flag")
22:51:36 Quit DMJC-L ("Leaving")
22:53:57***Saving seen data "./dancer.seen"
22:54:07 Join muesli- [0] (
22:54:35muesli-g'mornin ladiez :D
22:54:39TangHi Muesli!
22:54:46Tangyou're the same from MR?
22:54:49muesli-hi Tang!
22:54:55muesli-yepp, i am ;)
22:55:04muesli-bonjour mon amour ;)
22:55:05TangOkay and so do I for tang ;)
22:55:28muesli-dont blame me for my lousy french ;)
22:55:29Tangwe are on a very serious IRC channel mr Museli :o
22:55:40Tang(you're not English?)
22:55:59muesli-no, i am not...i am german
22:56:29TangAH okay
22:56:36Tangmy english too won't be perfect
22:56:43Tang(cause I'm french)
22:57:13muesli-see query ;)
22:57:21Tang(gutten tag and ich li bedich is all i know in Deutch, don't care the spelling)
22:57:43muesli-see query ;)
22:58:25Tang"see query"?
22:58:49muesli-which irc-client or you using?
22:59:20Tanghum i'm using chatzillia
22:59:22muesli-it will probably be mirc if you're using windows
22:59:30Tangin fact
22:59:39Tang(a plugin for firefox)
22:59:50muesli-a query is a seperate windows for private chats
22:59:57Tangah oaky
23:00:04Tangi see it
23:01:11 Quit DMJC ("Leaving")
23:03:02 Join DMJC [0] (
23:04:43 Join Diway [0] (
23:04:45 Quit DMJC (Client Quit)
23:04:48Diwayhi !
23:05:41TangThis isI've got some answer from amorim on HA board!
23:05:57Tanghe advice wavpack for lossless but also lossy encoding
23:06:15Tang(wavpack is the only one lossy encoder with integer mat)
23:08:02preglowlossless and lossy?
23:09:09preglowi see
23:10:27Tangi send the links:
23:11:11Tangyes indeed wavpack is an hybrid codec:
23:11:19Tangthere is lossy and lossless mode
23:11:32 Quit iriver ("CGI:IRC (Ping timeout)")
23:13:06linuxstbTang: Thanks for that - wavpack looks promising, for both lossy and lossless real-time compression. It's definitely one worth testing on the iRiver itself.
23:13:17muesli-brb in 45mins
23:13:37linuxstbDo you know if wavpack is generally considered a good codec?
23:14:39Tangokay i'm very happy to help :)
23:19:05TangHum i've seen a mistake (a lack in fact)
23:19:16Tangin the "firmware feature comparison"
23:19:21Tangfor iHP
23:19:36Tang" Mid-playlist resume ? for iHP"
23:20:03TangIn fct there is mid playlist resume with original firmware since v1.60 (or 1.40 maybe)
23:20:36rasherWhere do I put patches? The SF patch tracker or do I wait for someone here to get it?
23:21:01Tangalso for:
23:21:18Tang" Supports bad path prefixes in playlists Yes Yes ?"
23:21:24Tangthe answer is "YES" ;)
23:21:48Tang" File Delete & Rename −− No"
23:21:53Tangthe good answer is:
23:22:02Tang"file delete yes"
23:22:06Tangrename NO
23:23:41Tang Backlight-on when charging option −−> the answer is YES in fact: you can select light always on from external power source (AC or external battery pack)
23:24:28Tanghum there is quite many mistake i'll correct it and sen an email
23:24:35Tangseems a better ssolution
23:25:28jypI got the 220 lcd emulated ...
23:25:28 Join amiconn [0] (
23:26:34jypon to writing a rockbox driver for it :)
23:28:08amiconnrasher: I added the player to the snow, mosaique and cube plugin description in the wiki
23:28:47HClwhat about putting sokoban on player?
23:29:08amiconnThat'd be really hard...
23:29:14rasheramiconn: Good, good
23:29:39*amiconn has to investigate a strange problem on Ondio
23:31:09amiconnEither the manufacturer's formatting of the internal flash is messed up, or the rockbox fat16 support has a bug .. :-/
23:31:23rashersniff, fat16 :(
23:31:54amiconnfat32 wouldn't really make sense on a 128 MB flash chip, imho
23:32:27sneakumsmaybe if you want lots of small files
23:32:33sneakumslots and lots
23:32:34rasherI don't.
23:32:37amiconnAfter all, we have to support it (we support fat32 as well), since the archos firmware can only boot off a fat16 partition
23:35:54amiconnsneakums: Why would you want lots of small files on a music player?
23:36:04rasheramiconn: sid support! :)
23:36:09sneakumsi assume you wouldn't
23:36:16rasher(yes I know, not on that device)
23:36:40sneakumsbut fat16 has that 2^16 cluster limit
23:36:45rasherWhere do I put patches for a) solitaire keybindings for iriver, b) logo plugin ?
23:37:21amiconnrasher: Umm, the standard place is the sf patch tracker
23:37:39rashernot sure where it was appreciated
23:37:41 Quit muesli- (Read error: 113 (No route to host))
23:40:33 Join DMJC [0] (
23:50:19rasherpatches added
23:56:17 Join amiconn_ [0] (
23:56:30 Quit amiconn (Nick collision from services.)
23:56:30 Nick amiconn_ is now known as amiconn (
23:57:14amiconnGrmpf. Seems my sector count calculation for MMC was not quite correct :(
23:57:31amiconnI should read the datasheets more thoroughly

Previous day | Next day