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

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

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

#rockbox log for 2008-11-30

00:01:22tical2kI got a Belkin car device that connects to the dock plugin and has some basic (pause/play next and previous) functions on it. They currently dont work with rockbox. I noticed a lot of the remote accessories dont work with rockbox. Is there any development in progress on this?
00:02:35tical2kthanks for the info
00:03:34*gevaerts likes this e280. Rockbox usb is extremely unreliable on it
00:03:59 Quit robin0800 (Read error: 60 (Operation timed out))
00:04:10bertrikdo you see anything weird on the bus?
00:04:33gevaertsI haven't looked with the analyser yet. I'll do that tomorrow
00:06:46 Join beylulbahta [0] (
00:10:17 Quit bmbl ("Woah!")
00:14:44 Quit {phoenix} (Read error: 110 (Connection timed out))
00:20:45 Join jhulst [0] (n=jhulst@unaffiliated/jhulst)
00:21:49tical2klinuxstb, thanks again for the info the patched build works for my device
00:24:39skiyalso folks, just a quick and (probably stupid) question, does rockbox support ogg-vorbis/flac on the "Archos Jukebox Studio 20 20GB"
00:25:10skiyI heard it said somewhere that it didn't support ogg/flac but thought I'd check before buying one off ebay..
00:28:06 Part llo7f
00:29:28 Quit tical2k ("Leaving")
00:33:42*skiy withdraws his best offer from ebay as most of his music is in ogg :\
00:38:14 Quit bertrik ("Leaving")
00:38:33 Part blkhawk
00:39:34kugelskiy: they cannot play anything but mp3, because they have a dedicated chip for that, and not a general purpopse cpu like newer daps
00:39:48kugelit's not a rockbox issue, it's a limitation of the hardware
00:40:00 Part beylulbahta
00:42:26scorche|shthen have a "general purpose cpu"...just one that runs at 12MHz...
00:43:07kugeloh, sorry then. But not powerful enough, so basically not usable?
00:43:18scorche|shnot for decoding ausio
00:43:22*scorche|sh cant type
00:43:42scorche|shwhat else would we run all the other non-decoding processes on?
00:44:32kugelyea, sorry. Of course they must have such a thing.
00:44:59kugelI was basically saying, they don't have such a unit like new daps, as in they don't use it for decoding unlike the newer ones
00:45:45 Nick JdGordon|zzz is now known as JdGordon (n=jonno@rockbox/developer/JdGordon)
00:57:34amiconnlinuxstb: around?
00:58:48kugelJdGordon: did you play around with your v2 a bit?
00:59:51JdGordonmaybe today a bit
01:00:42kugelerr, my fuze hangs if I don't #define HAVE_LCD_ENABLE
01:02:14 Join krazykit` [0] (n=kkit@
01:02:44 Quit krazykit (Read error: 104 (Connection reset by peer))
01:08:13linuxstbamiconn: Yes
01:08:29amiconnDid you ever actually test ape mono playback?
01:08:34amiconnI just get silence....
01:08:59amiconnAlso there are some todo's in decoder.c related to 8 bit streams
01:09:19linuxstbI'm 99% sure I tested mono streams. Have you tried them with demac?
01:11:00 Join domonoky1 [0] (
01:11:19amiconnThe comment is suspicious - it says "do nothing unless it's 8-bit audio". I would think that it does at least need to SCALE()
01:11:46 Quit lasser (Read error: 110 (Connection timed out))
01:12:04amiconnAnd guess what - if I do that, it plays...
01:12:23amiconndemac won't have this issue as SCALE() is a no-op for non-rockbox
01:12:56kugellinuxstb: lcd_enable might have been a wrong guess, it does at least something which isn't as optional as lcd_enable should be
01:14:06 Join gregorovius [0] (
01:15:42linuxstbamiconn: Are you talking about 8-bit or mono? Or was your test 8-bit mono?
01:16:09amiconn16 bit mono
01:16:49linuxstbAh yes, I see now.
01:17:08 Quit ibseo ("quit")
01:17:50linuxstbI guess that comment is based on what the original source does. You're of course right that we need to scale it for Rockbox.
01:18:18amiconnWhat do you think about this solution: (the #if0 part is svn for comparison)
01:18:59amiconnIt uses the same system as is used in some other places to handle 1 or 2 channels without if/else
01:19:28amiconnWithout making the whole loop conditional, I mean
01:19:39amiconnIt's a little slower though
01:20:06 Quit domonoky1 (Read error: 104 (Connection reset by peer))
01:21:00JdGordoninteresting patch just submitted...
01:21:00amiconnOn a related matter: What would be needed to support 8 bit?
01:21:46kugellinuxstb: was that one directed to me (with the original source)?
01:21:59kugelJdGordon: which?
01:22:12JdGordon"Philips GoGear HDD16x0/HDD63x0 patches for power, sound, buttons, lcd"
01:22:29linuxstbkugel: No, I'm just replying to amiconn at the moment...
01:23:15amiconnAlso, couldn't be 8-bit also be pseudo-stereo, i.e. there's another conditional missing?
01:23:53kugellinuxstb: ok, could've been very well an answer to my lcd_enable "question" too
01:25:00linuxstbamiconn: I would have to look at the original Monkey's source - maybe pseudo-stereo isn't used for 8-bit streams. Your patch looks nice, but as you say, it's a little slower, so I would probably prefer more if/else statements.
01:25:10amiconnThe if(decoded1) *decoded1++ = left costs about 0.2 MHz on PP5022
01:25:53amiconnBut these loops are small, so they can be split
01:27:08amiconnentropy.c has similar loops, but then they use largeish static inline functions, so splitting them would inflate the code quite a bit
01:28:05 Quit domonoky (Read error: 110 (Connection timed out))
01:31:30*amiconn has a hard time reading the original monkey's audio source
01:31:33amiconnIt's C++ :(
01:31:36*linuxstb too
01:33:05linuxstbBut doesn't demac already handle 8-bit pseudo-stereo - along with all other bit-depths in the same part of decoder.c?
01:36:10linuxstbIf you look at lines 259 onwards of Prepare.cpp, it seems that it converts 8-bit streams to an unsigned 8-bit value. So for Rockbox, we wouldn't need that, and can just scale (IIUC).
01:37:04linuxstbkugel: What is "optional" about lcd_enable?
01:41:26amiconnlinuxstb: Prepare.cpp is for encoding?
01:41:41linuxstbThere are two classes in there - Prepare and Unprepare
01:41:59linuxstbUnprepare is the decoding part
01:42:16 Quit JdGordon (Remote closed the connection)
01:42:26 Quit jhMikeS (Nick collision from services.)
01:42:28 Join _jhMikeS_ [50] (n=jethead7@rockbox/developer/jhMikeS)
01:42:28kugellinuxstb: iiuc a target should still work just fine when you remove the HAVE_LCD_ENABLE #define in the target config, or am I wrong?
01:42:32 Quit petur ("Zzzzz")
01:42:43 Join JdGordon [0] (n=jonno@rockbox/developer/JdGordon)
01:43:19kugelexcept maybe that it has less battery runtime
01:43:51amiconnlinuxstb: demac.c doesn't seem to handle 8 bit. It checks for 16 bit / 24 bit explicitly.
01:45:02amiconnAfaik 16 bit (and higher) samples are usually signed, while 8 bit samples are usually unsigned. Do you think libdemac should handle that, or demac.c?
01:45:24*amiconn would like to have a 24 bit testfile...
01:46:40 Join toffe82 [0] (
01:49:04linuxstbamiconn: I'm not sure about 8-bit at all... Let me see if I have a 24-bit file somewhere.
01:50:04amiconnMaking an 8 bit testfile from 16 bit should be simple (e.g. using audacity). Making a 24 bit testfile from 16 bit wouldn't make sense (low bits would be all-zero)
01:51:06amiconnSeems like the original code only uses pseudo-stereo for 16 bit.
01:52:18 Quit ender` (" Python: executable pseudocode. Perl: executable line noise.")
01:53:10linuxstbkugel: Yes - assuming the lcd driver is correct...
01:54:05***Saving seen data "./dancer.seen"
01:56:56 Join AndyIL [0] (i=AndyI@
01:57:07kugellinuxstb: And our Fuze's lcd enable does something that's not optional, but needed, as it's not working without.
01:57:20kugellinuxstb: even though, the bootloader works, which doesn't define that
01:57:41amiconnlinuxstb: Btw, the mono predictor is 3x as fast as the stereo predictor on PP, even though it's C vs. asm.
01:58:10kugellinuxstb: I #if 0'd the else part and it works for now.
01:58:44amiconnThe other parts are almost exactly twice as fast, as one would expect
01:58:50kugelI'll investigate that later. Btw, I think I understood the disassembly pretty good. At some points I still stuck though
02:01:14kugellinuxstb: I've for example noticed noticed the OF does 1 lcd_write_reg more in lcd_window_y that we have in our driver, although it doesn't seem to change much
02:01:41linuxstbkugel: I remember simplifying those window functions - check the svn history.
02:06:08linuxstbJdGordon: IMO we should credit people by name in commit messages when committing patches, rather than just referring to the FS number...
02:07:50JdGordonfor the first time sure, but after that there is no need?
02:08:06 Quit AndyI (Read error: 110 (Connection timed out))
02:08:15JdGordononce they are in credits task numbers should be enough...
02:08:22JdGordonbut ill change if im wrong
02:09:54linuxstbI just find it a pain to have to refer to FS tasks to see basic info on a commit like who wrote it...
02:09:57kugellinuxstb: do you mean you simplified that lcd_write_reg away?
02:10:14kugel(I haven't looked the cleanup up yet)
02:10:33 Quit Seed ("cu, Andre")
02:10:36kugelit could also be possible that this command was introduced in 1.01.15 which I was disassembling
02:11:22linuxstbkugel: I don't know which write_reg you're talking about - just that I think I removed some. As I said, check the svn history.
02:12:59 Quit toffe82 (Read error: 110 (Connection timed out))
02:14:49kugellinuxstb: anyways, I have no issues with the window_y/_x functions, lcd_enable seems problematic
02:20:06 Quit gevaerts (Nick collision from services.)
02:20:18 Join gevaerts [0] (n=fg@rockbox/developer/gevaerts)
02:32:31 Quit _jhMikeS_ (Nick collision from services.)
02:32:35 Join jhMikeS [50] (n=jethead7@rockbox/developer/jhMikeS)
02:42:32 Quit DerDome ("Leaving.")
02:42:40 Quit Schmogel ("Miranda IM! Smaller, Faster, Easier.")
02:53:35 Join Darksair [0] (n=user@
02:56:57 Join unstable [0] (i=unstable@tor/regular/sid)
02:57:09unstableWhat is a good not too expensive media player that rockbox can run on?
02:57:21unstableThat I can buy.
02:58:06 Join toffe82 [0] (
03:02:40 Quit herrwaldo ("Konversation terminated!")
03:09:23 Quit toffe82 (Read error: 60 (Operation timed out))
03:12:20 Join mimepp [0] (n=yutao@
03:18:15unstableIs that device supported? I don't see it listed anywhere.
03:18:33kugelIf it's not listed it's not supported
03:27:58Unhelpfulunstable: if you can receive shipments in the US, and flash-player capacity is acceptable, has refurbed e200v1 players you can install rockbox on
03:28:32unstableUnhelpful: e280 is supported?
03:28:45unstableDo they make this anymore, it says sold out everywhere.
03:31:25 Part unstable
03:34:50 Join saratoga [0] (n=98039ee9@gateway/web/cgi-irc/
03:44:14 Join Thundercloud [0] (
03:54:08***Saving seen data "./dancer.seen"
03:55:05 Quit soap ()
03:58:03 Join soap [50] (n=soap@rockbox/staff/soap)
04:03:03 Quit Bensawsome ("The awsome is gone :(")
04:03:38 Quit Thundercloud (Remote closed the connection)
04:10:27 Join Bensawsome [0] (n=Bensawso@unaffiliated/bensawsome)
04:14:51 Join ap0 [0] (
04:19:45 Quit ap0 (Client Quit)
04:21:21 Quit Darksair ("To Arch or Gentoo? That is the question...")
04:22:58 Join LambdaCalculus37 [0] (
04:25:27 Quit BlakeJohnson86 (Remote closed the connection)
04:28:04 Quit agaffney (Read error: 110 (Connection timed out))
04:31:21 Join BlakeJohnson86 [0] (
04:41:08 Join shadearg [0] (
04:49:41Unhelpfulanybody want to take a crack at this resize-on-load m3-remote-on-m5 issue? i'm getting garbled icons etc on the m3 remote in sim, but i can verify that the same vertical interleave code that works on m3 is being called when a bitmap is loaded for the remote on the m5 - except that the bitmaps on the remote are a mess on m5 :/
04:50:27 Join XavierGr [0] (n=xavier@rockbox/staff/XavierGr)
05:00:02 Quit Bensawsome ("The awsome is gone :(")
05:03:26 Quit gregorovius ()
05:07:39 Quit miepchen^schla (Read error: 110 (Connection timed out))
05:09:01Unhelpfulhrm, the direct mono output is broken, somehow.
05:09:29 Quit LambdaCalculus37 ("Ka-chunka")
05:24:24 Join willwolfe [0] (n=d173e841@gateway/web/cgi-irc/
05:27:32 Quit willwolfe (Client Quit)
05:35:33 Quit shadearg ("goodbye.")
05:35:44 Join shadearg [0] (
05:46:24 Join shelby [0] (
05:47:04shelbyi know its kind of sad but i have never been so pissed in my life over how i cant charge my ipod and use it at the same time. i dont want to mount it and transfer files i want to charge it whilst listening to music. how do i go about that?
05:47:29shelbybecause whenever i plug it in it has that picture of a usb dongle and then it restarts in disk mode
05:47:35shelbyand i need it to stay in rockbox
05:47:37shelbyreally badly
05:47:42shelbyplease. any help?
05:47:44saratogai think you hold play or something
05:48:17shelbyi hold play when i plug it in?
05:48:21kugelsaratoga: hey, I've edited the ams sansa forum thread to reflect my current "progress" and problems with the buttons
05:48:50saratogamanual says menu, at least for the video, might want to check for your ipod model
05:48:57saratogakugel: ok i'll look at it now
05:49:53shelbyholding down play DOES NOT WORK
05:50:02shelbyim sorry to be such a prick... but... lol
05:50:04shelbyim a prick.
05:50:07shelbyany other suggestions?
05:50:11kugelsaratoga: also, if you read the logs, funman and me found something new regarding clocking and why r19234 broke the svn bootloader
05:50:14saratogashelby: read manual
05:50:40saratogakugel: i still haven't managed to read anything over dbop, it seems rather complicated
05:50:45shelbywhich one? there seems to be a million billion manuals scattered everywhere across the site and it would be so easy just for you to tell me whether i would waste my time doing that.
05:50:48saratogabut fortunately, GPIO should work fine for that
05:50:56saratogashelby: theres only one manual
05:51:01saratogawell one per player
05:51:11shelbyokay. where does it talk about usb charging and staying on?
05:51:17shelbynevermind. let me look.
05:51:18kugelsaratoga: but it apparently doesn't
05:51:41saratogakugel: theres only 4 DBOP pins which are not also wired to GPIO, and none are used for buttons
05:51:44shelbyi would soooooo rather not have an ipod than deal with this bull :\
05:51:48kugelfeel free to have a look at it. I couldn't get the buttons to read (seperately) with the same code just the pins swapped
05:51:59saratogalet me see what else is on GPIOD
05:52:31kugelsaratoga: the buttons work (the ones found as of now) in the bootloader, so yes, they are on gpio
05:52:48shelbyis it in the general settings section?
05:53:03kugelsaratoga: and GPIO B and C are physically connected to the GPIO ports
05:53:08saratogakugel: it could be that GPIOD is shared with something else, so your'e fighting with another driver for access to them
05:53:33kugelI'm not using GPIOD to work at all
05:53:39shelbyim pretty sure the manual dosent cover this
05:53:46kugelI'm still fighting with the buttons on GPIO B and C
05:53:56kugelto get them work in the main binary
05:54:08shelbyseriously. its not in the manual
05:54:11***Saving seen data "./dancer.seen"
05:54:38shelbysoo... does anyone have a suggestion besides the generic RTFM solution?
05:54:43kugelsaratoga: most of the SD stuff is on GPIOD, that's what I can tell you
05:55:44kugelshelby: I think you need to press a button if you don't want the ipod to reboot/show the usb connected screen.
05:55:55*shelby wonders which button
05:56:02shelbyits only *one* button, correct?
05:56:23shelbywhich i hold down as i am in the main rockbox menu roughly 1 second before i plug in the usb cable to the ipod?
05:56:29kugelshelby: I can't tell which one, since I don't have an ipod, but I'd be really surpised if it's not covered by the manual. Have you downloaded the pdf version? There you can searchin
05:56:41perrikwpshelby: read here
05:57:12saratogakugel: maybe the SD driver is putting the GPIO in some odd mode in the main build but not in bootloader?
05:57:15shelbythhaannk you perrikwp!!
05:57:23*shelby hugs perrikwp
05:57:25perrikwpyour welcome
05:57:47kugelsaratoga: As far as I know, sd driver only uses GPIOD
05:58:01kugeland again, I don't fight with D at all, only B and C
05:58:13saratogasorry didn't mean to say D
05:58:48saratogai noticed while trying to read some of the GPIO pins that I had to explicitly set and reset the direction and DBOP registers carefully otherwise they didn't work
05:58:56saratogaif you're already doing that then i have no idea
05:59:31kugelI explicitly set the DIR before reading them
06:00:00kugeland I really tried the same code for all pins. Some are read properly, but most aren't
06:00:08saratogaand AFSEL too?
06:00:25kugeland I reset after finished reading
06:01:19saratogaperhaps its a timing issue, does the main binary set any clocks not set in the bootloader?
06:01:39kugelno, the binary doesn't set up clocks. It's done in the bootloader
06:01:48 Quit shelby ("leaving")
06:01:52kugelbut I tried putting some delays anyway, without success
06:03:15kugelsaratoga: you focus on finding the rest of the buttons?
06:03:29saratogakugel: that would be the wheel right?
06:03:43kugelplus home and hold
06:04:25saratogai'm still kind of busy but i'll keep playing with it in my spare time
06:04:39 Join Bensawsome [0] (n=Bensawso@unaffiliated/bensawsome)
06:05:24kugelsaratoga: maybe the rest of the buttons can wait until we got the ones we have working?
06:05:58 Quit Bensawsome (Client Quit)
06:08:58 Join massiveH [0] (
06:09:11 Quit amiconn (Nick collision from services.)
06:09:15 Join amiconn_ [50] (n=jens@rockbox/developer/amiconn)
06:10:23 Quit kugel ("ChatZilla 0.9.84 [Firefox 3.0.4/2008111318]")
06:12:43 Quit desowin__ (Read error: 60 (Operation timed out))
06:14:32 Join desowin_ [0] (n=desowin@
06:20:18 Join Darksair [0] (n=user@
06:25:11 Quit pixelma2 (Nick collision from services.)
06:25:19 Join pixelma2_ [0] (n=marianne@rockbox/staff/pixelma)
06:25:21 Nick pixelma2_ is now known as pixelma2 (n=marianne@rockbox/staff/pixelma)
06:25:47 Quit __lifeless (Read error: 110 (Connection timed out))
06:35:49 Quit skipper (Read error: 145 (Connection timed out))
07:02:21 Quit XavierGr ()
07:03:02 Quit massiveH ("Leaving")
07:07:47 Quit saratoga ("CGI:IRC (EOF)")
07:09:50 Quit mimepp ("Leaving.")
07:31:12 Quit reacocard (".")
07:32:30 Join reacocard [0] (n=reacocar@WL-135.CINE.HMC.Edu)
07:37:10 Quit JdGordon (Remote closed the connection)
07:38:37 Join JdGordon [0] (
07:48:45 Quit skiy (Read error: 110 (Connection timed out))
07:50:44 Quit BHSPitLappy ("Ex-Chat")
07:52:19 Join Bensawsome [0] (n=Bensawso@unaffiliated/bensawsome)
07:54:13***Saving seen data "./dancer.seen"
07:57:25 Nick Bensawsome is now known as [i][B][e][n] (n=Bensawso@unaffiliated/bensawsome)
07:59:00 Nick [i][B][e][n] is now known as Bensawsome (n=Bensawso@unaffiliated/bensawsome)
08:30:04 Join agaffney [0] (n=agaffney@gentoo/developer/agaffney)
08:37:53 Quit JdGordon (Read error: 104 (Connection reset by peer))
08:49:25 Join Rob2222 [0] (
08:58:19 Quit jhulst (Read error: 148 (No route to host))
09:07:27 Quit Rob2223 (Read error: 110 (Connection timed out))
09:12:33 Join jhulst_ [0] (n=jhulst@unaffiliated/jhulst)
09:22:43 Quit reacocard (".")
09:22:50 Join reacocard [0] (
09:23:01 Join JdGordon [0] (n=jonno@rockbox/developer/JdGordon)
09:23:29*JdGordon grumbles at the button bar
09:31:06 Quit jhulst_ (Read error: 54 (Connection reset by peer))
09:31:18JdGordonif that could be removed it would make things much easier...
09:32:03 Join jhulst_ [0] (n=jhulst@unaffiliated/jhulst)
09:36:07 Quit jhulst_ (Remote closed the connection)
09:36:22 Join jhulst_ [0] (n=jhulst@unaffiliated/jhulst)
09:37:46*JdGordon gets a menu on his e200v2 (finally!)
09:38:12 Join stoffel_ [0] (
09:47:48 Quit nuonguy ("This computer has gone to sleep")
09:54:18***Saving seen data "./dancer.seen"
10:03:57 Quit beast__ (Read error: 145 (Connection timed out))
10:06:09JdGordonhmm... sticking a mutex around the dbop/gpiob/c accesses doesnt help.... but the problem is definatly there... if button_int() isnt called the lcd seems to work more corectly
10:06:24 Join Seed [0] (
10:08:55 Nick jhulst_ is now known as jhulst (n=jhulst@unaffiliated/jhulst)
10:09:07 Join lasser [0] (
10:16:52 Join skiy [0] (
10:19:27 Join Thundercloud [0] (
10:38:06 Join bertrik [0] (
10:41:34 Join bmbl [0] (n=Miranda@unaffiliated/bmbl)
10:50:23 Quit pixelma2 ("-")
10:51:01 Quit reacocard ("leaving")
10:51:09 Join reacocard [0] (
10:51:31 Join pixelma [0] (n=Marianne@rockbox/staff/pixelma)
10:52:29 Quit Thundercloud (Remote closed the connection)
10:53:53 Join ender` [0] (
10:58:04 Join pROFIL [0] (
11:04:27 Join domonoky [0] (n=Domonoky@rockbox/developer/domonoky)
11:06:14pROFILHello everyone!
11:06:39pROFILAnyone who knows how it going with the Samsung YP-P2 Porting?
11:07:11domonokypROFIL: take a look a the corresponding forum thread..
11:09:30 Quit jhulst (Read error: 113 (No route to host))
11:13:03pROFIL"When you first attach your device in USB boot mode to your computer" What does USB boot mode means?
11:14:20 Join {phoenix} [0] (
11:16:47pROFILOr, how do get my player in USB boot mode?
11:17:52*domonoky doesnt know, remember rockbox isnt ready/working on this samsumg players..
11:18:34*pROFIL gets sad
11:19:31JdGordonready/working or even started on that player
11:21:00 Join JdGordon_ [0] (n=jonno@rockbox/developer/JdGordon)
11:21:50 Quit stoffel_ ("leaving")
11:23:17bertrikI have a battery discharge curve now of the sansa clip. It looks very much like other lithium battery discharge curves.
11:24:17bertrikI'm surprised it took nearly 11 hours to drain the battery from about 4.15V to 3.3V, I'm pretty sure I turned up the cpu frequency to 240 MHz during the test.
11:24:23 Nick amiconn_ is now known as amiconn (n=jens@rockbox/developer/amiconn)
11:24:51 Join petur [50] (n=petur@rockbox/developer/petur)
11:25:05pROFILJdGordon: But you don't know how to even turn it into usb boot mode? Like pressing a button when I connect it to the computer?
11:25:23amiconnbertrik: Well the CPU isn't actually doing much if you're not playing music...
11:26:11bertrikamiconn, yes indeed, but is there some kind of idle instruction that we execute?
11:26:42pixelmacan someone comment on how similar the H100 and H300 keymap is overall (I'm wondering if I should combine the new keymap-*.tex files of these 2 into one, seeing that they share one keymap-*.c file
11:27:26JdGordon_should be identical
11:27:56pixelmaok, will look into it some more
11:28:31JdGordon_pixelma: my bubbles change last week needs a quick fix for the manual.. i was going to try it but got scared in bubble.tex... you just need to remove the first column i tihnk
11:28:57bertrikamiconn, I'll have a closer look at the data sheet of the processor
11:28:58ameyerdoes anyone know how different hardware revision 2 clips are from hardware revision 1 clips?
11:29:31bertrikNobody knows yet, as far as I know
11:33:00ameyerbut I'd imagine that if it's different enough to need different firmware, that's not a good sign for rockbox
11:34:26amiconnpixelma: Core keymap should be identical, but a few plugins use different mappings. I know that jpeg.rock does
11:34:44amiconnbertrik: Assuming it is implemented for AS3525, of course we do
11:36:05pixelmaamiconn: thanks, the plugin button tables in the manual are independent anyways
11:38:54bertrikamiconn, any idea where I can find it in the code?
11:47:12bertrikamiconn, I think I found it: as3525 does something with mcr to wait for interrupt in core_sleep
11:50:49*domonoky wonders if rockbox mutexes need some sort of init.. the sd code just calls mutex_lock and und mutex_unlock, without initialisizing the mutex struct..
11:51:35bertrikdomonoky, I too noticed that mutexes apparently work without being initialised ...
11:53:21domonokybertrik: i wonder if they dont really work right, and cause our reboot problem.. at least i can put a panic before the mutex_lock in the sd_transfer_sector, checking for reentry, and it gets called. if i put the check after the mutex, i only see the reboot.
11:54:14bertrikwell, fix it and try again ! :)
11:54:22***Saving seen data "./dancer.seen"
11:54:24 Join MethoS [0] (
11:56:02jhMikeSdomonky: no, mutexes will not work properly if not initialized.
11:56:14 Quit JdGordon_ ("Miranda IM! Smaller, Faster, Easier.")
11:56:58jhMikeSthey should also never be reinitialized if there's any possibility that a thread could be accessing it at the time
11:58:58*domonoky sees a mutex_init n other code, which is missing in the sd-code...
12:01:05pROFILYey, I got my P2 in usb boot mode :D
12:01:14 Join stripwax [0] (
12:01:29 Quit stripwax (Client Quit)
12:01:59domonokylooks like this fixes the reboot, while starting playback.... and now i see other problems.. :-)
12:02:03*jhMikeS emphasises the importance of initializing all kernel objects since "0" for all members isn't the case :)
12:02:17 Join stripwax [0] (
12:02:51amiconnjhMikeS: What would be the best way of communication between 2 threads if there's not just a binary flag to pass?
12:03:13amiconnCan a semaphore indicate multiple states, or do I need a queue?
12:03:50jhMikeSamiconn: semaphore is counted with a min and max. waiting decreases the count, and if the count is < 0, it blocks. release increases the count.
12:04:04jhMikeS*counted with a max rather
12:04:05domonokyhm, starting playback now gives me again this lovely "not enough space in ringbuffer". But i also get a reboot, when the backlight should go off... probably more init problems ? :-)
12:04:13amiconnAh, hmm.
12:04:17daurnameyer: we have rockbox on the clip?
12:04:32*amiconn will probably use a semaphore and use a separate struct for communication
12:05:08jhMikeSamiconn: the wakeup object is really a sort of simple one-message queue with a stick flag (much less memory)
12:06:10domonokydaurn: people are working on clip and other ams-sansa targets, but its not ready.
12:06:20amiconnThis is for the libdemac dual-core split. The CPU thread will do the entropy decoding, then pass the data to the COP, which will (in the first implementation) do the filtering, prediction, and decorrelation
12:06:56amiconnIn a second step, it should allow the split to depend on the compression level, for better balancing the load
12:07:42amiconn-c1000 and -c2000 should then work as described, while -c3000 and up should only do the filtering on COP, then pass the data back to the CPU, which will do prediction and decorrelation
12:07:57jhMikeSamiconn: with a counted sem use can use it to make a blocking queue of samples. SPC uses two semaphores with a max of 2 each to block the codec if empty or the synth if full.
12:09:08amiconnWith a little more squeezing, this 2nd step will make even -c3000 usable on PP (maybe not on 5020, but on 5002 and 5022)
12:09:37bertrikdaurn, yes, a lot of stuff already works: buttons, display, rtc, battery readout, flash read/write, it boots and allows you to browse the menus, even plugins work, sound doesn't quite work yet
12:09:45jhMikeSis there squeezing possible to have -c5000 useable on imx31?
12:13:26amiconnAtm I have no idea for that
12:13:50amiconnCould the disk access use dma?
12:15:11*amiconn cannot imagine how ide dma works, be it on PC or an embedded platform...
12:15:18jhMikeSusing the SDMA controller is rather involved. I'm pretty sure we have to actually write code for that. it has no simple DMA controller. then using the ATA DMA modes is possible.
12:15:50bertrikdomonoky, I added the mutex_init in as3525-codec.c but didn't spot the missing mutex_init in ata_sd_as3525.c at the time
12:16:38jhMikeSimx31 still needs the ATA FIFO serviced by the SDMA controller when using disk DMA
12:16:40domonokywow ! i got a codec loading on m200v4 !! :-)
12:17:24bertriknice! any sound too ?
12:17:57domonokyi still have to remove some panicf to get to the sound.. :-)
12:20:29domonokyhm, not sound from this.. but that may be caused ba other not-implemented functions.. :-)
12:20:32 Join MethoS- [0] (
12:21:24domonokybut the codecs load, and the pcm_play_dma_start is called... which is much better, then just the reboot we got yesterday.. :-)
12:21:58amiconnI mean, I know how normal dma works. But for ide dma, you set a special dma transfer mode in the disk (as opposed to pio). So I think the disk itself does the dma somehow. But how can it do that (access the host's memory)?
12:22:26amiconnIf this isn't the case, I wonder why the special dma modes even exist...
12:23:42domonokyamiconn: wouldnt this "special dma mode" just enable the dma_request and dma_request_clear lines in the disk-controller ?
12:24:35domonokyand via this request-lines, it tells the dma-controller when to transfer data to the hosts memory... at leats thats how it works on as3525..
12:25:58amiconnBut iiuc you could also use the disk's pio mode in conjunction with a dma controller, couldn't you?
12:27:09amiconn(at least if iordy is disabled)
12:27:38domonokyonly if the enabling of the dma_request lines is seperate from dma/pio mode setting... but i am not sure..
12:28:16amiconnI mean plain transfer, without checking an 'ready' lines
12:28:32amiconnPIO is pure timed transfer if iordy isn't used
12:29:14 Quit MethoS (Read error: 104 (Connection reset by peer))
12:29:20 Join MethoS-- [0] (
12:29:32*domonoky doesnt know. But the dmac has to somehow know when to transfer data in/out of the disc fifo..
12:30:27 Join herrwaldo [0] (
12:31:31amiconndomonoky: It would just transfer as fast as the external data port allows (which then needs some sort of waitstate setting). That's what we do for lcd write on H300
12:32:25amiconnBut I think it became a little clearer now... thanks
12:34:04 Join denes [0] (
12:35:28 Join tyfoo [0] (
12:38:09jhMikeSamiconn: there's the mode for the drive of course and then some setup in the ATA controller itself.
12:38:53*domonoky commited the muxtex_init fix..
12:40:15jhMikeSPIO should be available alongside DMA for sending commands
12:43:24domonokywow, someone got sound on the Phillips GoGear HDD16x0..
12:44:26amiconnThe PP IDE controller supports dma, even dma.
12:44:35*amiconn thinks we should try to find out how to use it
12:45:18amiconnjhMikeS: On a different matter, do you have an idea why my local c200 build shows that weird power-up behaviour?
12:45:59 Quit MethoS- (Read error: 110 (Connection timed out))
12:46:33amiconnI suspect some sort of race condition in init code that is triggered by my local changes. There are no changes which actually affect target specific code for c200 though; the lowest level stuff is some unfinished lcd-16bit.c rework that certainly changes its speed
12:46:41 Quit denes_ (Read error: 113 (No route to host))
12:47:17amiconnMy 14 other targets show no problem, just c200 does...
12:48:47jhMikeSamiconn: What is the behavior exactly?
12:49:27amiconnWhen I switch it on, the rockbox bootloader loads rockbox, rockbox shows the splash screen for a fraction of a second, then the c200 powers down again
12:50:24amiconnI *can* make it boot after several tries when applying some tricks: Pull the microsd card, hold RIGHT during boot, and start pressing POWER again when the splash screen appears
12:50:37amiconnOnce booted, it behaves normally...
12:50:42jhMikeSonly your local build?
12:51:13jhMikeSgot a patch? maybe it would show something.
12:51:19amiconnIn fact only a local build from my main tree. My secondary tree (plain SVN) works, as does the official current build
12:52:19jhMikeSthe first sleeping is done right around the splash screen
12:53:24 Quit pROFIL ()
12:53:34jhMikeSperhaps waiting for an interrupt triggers different behavior with some other register set differently?
12:57:24 Join moos [0] (
12:57:26 Join MethoS [0] (
12:57:55 Quit jhMikeS (Nick collision from services.)
12:58:01 Join jhMikeS [50] (n=jethead7@rockbox/developer/jhMikeS)
13:00:58amiconnjhMikeS:"> <== patch with all my changes which apply to c200
13:02:05amiconnThe lcd-16bit code is unfinished and doesn't work properly with a background image (image won't be visible most of the time), but shouldn't cause any non-visual problems
13:04:12 Join jaykay [0] (n=579e6bdc@gateway/web/cgi-irc/
13:07:32 Quit jaykay (Client Quit)
13:07:36 Join Jaykay_ [0] (
13:07:38 Nick Jaykay_ is now known as Jaykay (
13:09:44 Join Schmogel [0] (
13:13:00bertrikI notice that struct text_message is often (but not always) made static within functions. I don't see why it should be.
13:13:29bertrikIt's not a very big struct, just two members: a pointer and an int
13:14:16 Quit MethoS-- (Read error: 110 (Connection timed out))
13:15:18bertrikit's a const structure, referring only to other consts, BTW
13:15:31Jaykayive got a little problem with vmware..... i cant type < when i press this button, nothing happens. can anybody help me?
13:18:31bertrikJaykay, if it has nothing to do with rockbox, then we can't help you
13:19:05*gevaerts guesses that this is about the vmware image used to compile rockbox
13:19:20Jaykayit has something to do with rockbox cause i try to compile rockbox with the rockbox-image for vmware player.....
13:19:32pixelmaI guess the available vmware image has a different keyboard (probably US/UK) set
13:19:47jhMikeSamiconn: I don't see how codec_load_buf could have any part at that stage. Is there a change that current_vp could be getting switched behind the scenes?
13:20:08Jaykayi already changed it to german, with xf86cfg and kbdconfig....
13:20:09*domonoky guesses that the vmware does not have the correct keyboard setting for Jaykay, but thats a bit off-topic here.
13:21:02domonokyif all fails there is always: "ALT+60" :-)
13:21:12gevaertsMaybe you also need to set some vmware setting?
13:21:15*gevaerts doesn't know
13:21:39gevaertsDoes that vmware image run X?
13:22:52 Join Jaykay_ [0] (
13:23:17Jaykay_..... so if i shouldnt ask here for this problem, where can i do this?
13:24:33 Quit jhMikeS (Nick collision from services.)
13:24:39 Join jhMikeS [50] (n=jethead7@rockbox/developer/jhMikeS)
13:26:59gevaertsJaykay_: one problem is that not many rockbox developers actually use the vmware image
13:27:31Jaykay_what do they use?
13:28:20gevaertsA lot of people run linux natively, and some use cygwin
13:30:18bertrikAny chance it would work better in virtualbox?
13:33:46Jaykay_ill try virtualbox, then cygwin^^
13:34:54 Join MethoS- [0] (
13:38:07 Join kugel [0] (n=chatzill@unaffiliated/kugel)
13:39:48 Quit Jaykay (Read error: 110 (Connection timed out))
13:41:35gevaertsamiconn: running at 60MHz definitely makes things worse
13:54:26***Saving seen data "./dancer.seen"
13:56:10 Quit MethoS (Read error: 110 (Connection timed out))
13:57:23pixelmaJdGordon: re. bubbles.tex - yes, deleting the "in menu" part in the button table works. The only question is whether we want to explain the now menu items but I think they are quite self-explanatory. (would have a patch for the deletion ready for commit if you want to)
13:58:25 Quit MethoS- (Read error: 113 (No route to host))
13:58:39 Join MethoS [0] (
14:03:33 Join funman [0] (
14:05:11Jaykay_can anyone tell me how to use the .vmdk-image in virtualbox or show me a guide?
14:06:25 Join bertrik_ [0] (
14:06:31 Quit bertrik (Read error: 110 (Connection timed out))
14:07:07linuxstbJaykay_: No, but going back to your original problem, have you tried all the other keys to see where <> appear? You should be able to fix it by correcting the keymap in Linux, but I've no idea what Linux is in that vmware image...
14:07:55pixelmawe should ask scorche, he wanted to update the image anyways ;)
14:08:34*pixelma runs away
14:10:27domonokyfunman: did you notice, that i found the cause for the reboot on ams-sansas ? :-)
14:11:58 Quit Horscht ("We don't make mistakes, we just have happy little accidents")
14:12:59Jaykay_so now i cant even start the image correctly, now its like a DOS-windows....
14:13:16pixelmalinuxstb: I believe it's some debian flavour, a bit older but not sure , that info helps you any
14:13:43Jaykay_sorry i didnt get it
14:14:16gevaertsJaykay_: I'm not sure if X is expected to work without reconfiguration if you run the vmware image in virtualbox
14:15:41funmandomonoky: yes, thanks a lot!
14:16:12Jaykay_now its starting as expected.... gevaerts:i didnt get it too, sorry ^^
14:16:15funmanI remember bertrik_ also found this problem in i2c code
14:16:34funmanwhat did you have to change to get codec loading ?
14:16:40 Nick bertrik_ is now known as bertrik (
14:17:11domonokyfunman: enough buffer space (~ 816k) and the sd_mutex fix.
14:17:25gevaertsConnecting my e200 in rockbox, mounting it, and unmounting it, gives a total of 30 errors (flagged as CRC errors, but in reality they are caused by the packet being incomplete). The same test running the OF gives no errors. No hubs were involved, and linux gave no errors in dmesg (i.e. all were corrected at the first retry)
14:18:24domonokywav codecs seems to work, so that it calls pcm_play_dma_start(). mp3 gives some undefined instruction error, maybe the codec buffer is too small for it..
14:19:20*pixelma wonders which buttons are mapped to what keys on a keyboard in a Clip sim
14:20:05pixelmaI only found the arrow keys, resp. 2, 4, 6, 8 working so far
14:21:09 Nick Jaykay_ is now known as Jaykay (
14:21:27funmanhome = insert, select = space, page up/down = volume +/-, p = power
14:21:52pixelmawhy so different to the other sims?
14:22:22funmani didn't find any rule
14:22:38 Join stoffel_ [0] (
14:22:49funmanexcept using numeric keypad a lot, and i don't have a numeric keypad on my laptop
14:23:37pixelmathe others mainly use the num pad (plus u = USB connected, h = toggle hold switch, j = remote hold switch)
14:23:46funmanperhaps additional cases for numeric keypad can be added for other buttons than up/down/left/right
14:24:05*domonoky finds this whole keymap thing pretty confusing. You never really know which contexts are used, and there are no checks against dublicate button-definitions (but they then wont work)... :-/
14:24:13pixelmaand F5 = screenshot
14:26:07pixelmafunman: using the num pad (including the operands) has the advantage that you can "explain" the mapping in the sim background
14:26:43bertrikfunman, did you experiment with charging on the clip?
14:27:07funmanpixelma: feel free to add numpad definitions to the keys but please leave also the current one for people without numpad
14:27:26funmanbertrik: Yes, I just noticed it charged (did no timing/curve)
14:27:52bertrikfunman, I made a discharge curve of the clip with battery bench last night
14:28:15pixelmafunman: currently on a laptop without num pad too, but was unable to find out the mapping
14:28:20bertrikI'll post it on the SansaRuntime wiki
14:28:41funmanpixelma: see uisimulator/sdl/button.c (search for CLIP_PAD)
14:30:01*gevaerts actually hopes that if we ever find a real solution for these USB issues, we can go back to running USB at 30MHz
14:30:25bertrikhmm, on second thought I'll post it on the SansaClip page, as it's not really a runtime curve
14:33:48JdGordonpixelma: the manual is your domain... im happy for you to commit the fix, unless you dont have time and would rather i did?
14:36:00JdGordonfunman: any suggestions for the e200v2? I can get as far as the menu if i dont try reading the buttons ever... I've tried using a mutex around the dbop gpios but either get a white lcd or nothing... (this is with the cpu freq stuff #if0'ed out
14:36:40pixelmaI could do it myself, just not in the case an explanation should be added (only have about half an hour left now)
14:36:46*domonoky detects that the theme doesnt work with ~660k audio-buffer, so there are more places where we allocate buffers from the audio-buffer without really checking the size.
14:37:00funmankugel: JdGordon: domonoky can you give a try at the patch I put on FS #9588 ?
14:37:14JdGordonpixelma: yeah, I dont think an explanation is needed.. it should be fairly self-explanatory
14:37:49pixelmaok, could also be done later if someone feels it's neede
14:38:01funmanJdGordon: I think driver problems should be handled after we use a correct cpu clock (see the comments in FS #9588)
14:38:10funmanJdGordon: did you try using interrupts on gpio to read the buttons ?
14:38:14JdGordonok, trying it as soon as my net decides to play nice
14:38:38JdGordonno, I only just got a lcd working again.. if i get a chance ill tyr tomorow
14:40:13JdGordonfunman: does that patch need a new bootloader?
14:40:20funmanyes it only changes the bootloader
14:40:48funmanthe clock/sdram init procedure is not redone in the main binary
14:40:51 Join faemir [0] (
14:42:24JdGordonwhite lcd in bootloader
14:42:36JdGordonhmm... probably bootloader
14:42:54funmancan you change the dbop clock frequency (with cgu_dbop) until you fix that ?
14:43:05funmanhave a look at what kugel used for the fuze
14:43:14 Join Horscht [0] (n=Horscht@xbmc/user/horscht)
14:43:31 Quit moos (Read error: 104 (Connection reset by peer))
14:44:36Jaykayafter solving the first i have a second problem: when i type "svn co svn://" it says "unknown hostname any idea?
14:44:41 Join n1s [0] (n=nils@rockbox/developer/n1s)
14:45:17domonokyJaykay: make sure you have working inet access in the VM.
14:46:38funmanwhat about having a svn checkout in the VM, so only 'svn up' would be needed?
14:46:41Jaykayi didnt change anything in the vmware settings..... and it worked an hour ago. how can i ckeck the conetction?
14:47:00funmanJaykay: piing
14:47:10JdGordonfunman: (half asleep).. where is the cgu_dbop?
14:47:34funmanJdGordon: lcd-e200v2.c ams3525_dbop_init()
14:48:04Jaykayconnect: network is unreachable. ok i admit internet is not warking^^ how can i solve this?
14:48:36funmanhum well .. the peripheral clock doesn't change with my diff so perhaps the dbop clock is based on plla ? :/ I'm lost
14:49:28 Join MegafEee [0] (n=Linux@unaffiliated/megaf)
14:50:09gevaertsIf people want to have a look at my USB traces, they are at gevaerts/usb">
14:50:12 Join mirak [0] (
14:50:16funmanJdGordon: or perhaps you only need to adjust delays (kugel changed them too)
14:50:33JdGordonchanging it to CGU_DBOP = (1<<3) | (3-1); got the lcd in the bootloader
14:51:09domonokyfunman: your patch gives a garbled display in the bootloader on m200v4, but the main binary is working fine...
14:51:29funmanJdGordon: if you don't see too slow lcd operation it's good; but you might want to tweak the delays as well (for faster operation). Please concatenate your changes to my patch and post it on the report.
14:51:43funmanWe'll use incremental diff and when this works on all models we will commit
14:51:44JdGordonand in the main binary :)
14:51:53funmandomonoky: garbled how?
14:52:08JdGordonno buttons so checking for lcd updates is a bit hard
14:52:18domonokyi see a nearly black area, instead of the logo.
14:52:26funmanJdGordon: if it doesn't take 30minutes to draw a screen it's ok ^^
14:52:48funmandomonoky: same story, tweak the dbop clock and the delays
14:52:49JdGordonits noticable, maybe .25s?
14:53:14funmanJdGordon: try lowering the time in lcd_delay() / and highering the dbop clock frequency
14:53:38*JdGordon gets frustarted at the backlight turning off
14:53:53domonokyhm.. looks more like random lcd contents.. if i shutdown and reboot directly afterwards i see the old content :-)
14:53:57funmandomonoky: the difference is that set_cpu_frequency() isn't called in the bootloader I think
14:54:22kugelJdGordon, funman: I have a patch which makes the display less flickering.
14:54:58*JdGordon reckons the lcd speed isnt a big issue and buttons are more important :)
14:55:07JdGordonim not getting noticable flickering
14:55:25kugelhm, I do
14:56:27 Nick fxb__ is now known as fxb (
14:56:43JdGordonno flickering at all, even when I replaced update_rect with regular update
14:57:28kugelJdGordon: could be fuze specific, they use a slightly different driver
14:58:02kugelJdGordon: Ah, I see. What my patch does is already implemented in lcd-e200v2.c
14:58:36JdGordonyep... trying to read buttons breaks everything....
14:59:11kugelJdGordon: I managed to read the down and power button, but not the others
14:59:45JdGordonthats one up on me... did you change anything?
14:59:51 Quit amiconn (" switching horsts ;)")
15:01:05 Join amiconn [50] (n=jens@rockbox/developer/amiconn)
15:02:00kugelJdGordon: I changed much, reading the buttons seperately for example
15:02:59domonokykugel: did you try, if you can read other buttons, if you move the check up ? (ie timing, as BUTTON_POWER and BUTTON_DOWN are the first two buttons checked.)
15:03:10JdGordonfunman: we cant change the lcd delays at the different cpu speeds? no need to try to find a speed which works for all if its too hard
15:03:12kugelAlso, reading down and power together works, but when I add e.g. reading select (with the same code), no button at all works anymore
15:04:13funmanJdGordon: I don't know how hard it is, just that the Clip needed no modification. I don't know if the OF does cpu frequency change
15:04:18kugelJdGordon: I reduced the delay to x*=256 which works fine for me
15:04:53kugelI'm also unsure if we need delay at all. I think the 1.01.15 fuze firmware doesn't have such a delay function anymore
15:04:56JdGordonok, ill try that
15:05:19domonokyfunman: your patch works fine if i change the clkdiv to (3-1) in the m200v4 lcd driver
15:05:29kugelremoving the delay in _display_on in the fuze driver didn't harm
15:07:37JdGordonx *= 512; seems to work
15:08:05 Quit mirak ("Ex-Chat")
15:08:15JdGordonas does commenting out the line ocmpletly
15:08:21JdGordonill try it at full speed
15:08:43funmandomonoky: did you try at minimal & maximal cpu freq ?
15:08:50 Join AndyI [0] (i=AndyI@
15:08:56kugelJdGordon: yea, as I said, the delay is probably not needed at all (maybe only if dbop isn't clocked too fast)
15:09:11kugelfunman: I'm gonna try your patch later
15:09:15kugelneed to go now
15:09:20domonokynope, i did just a short test.. does cpu boosting already work ?
15:09:23 Part pixelma
15:09:25funmanyes it does
15:09:48 Join pixelma2 [0] (n=marianne@rockbox/staff/pixelma)
15:09:50funmanjust change the arg to set_cpu_frequency() in system-as3525.c
15:10:04JdGordonand move it outside the #ifdef BOOTLOADER block
15:10:08funmandocumented values were based on 240MHz base clock, and now it's 248 but that's not a big deal
15:10:14funmanhm ..
15:10:16 Join franky [0] (
15:10:32JdGordonyep, assuming its runnig at MAX its fine with 0 delay
15:10:41JdGordonalthough a full update is noticable
15:10:42funmanthanks JdGordon for spotting this :/
15:11:32*funman wonder what set_cpu_frequency() is defined to in BOOTLOADER
15:12:39JdGordonits not probably
15:12:55JdGordonthe bootloaders dont have HAVE_ADJUSTABLE_CPU_FREQ do they?
15:13:07funmanit's not conditional in config-clip.h at least
15:13:48bertrikgevaerts, you see partial packets, then nothing for some time and then more packets? I'm asking this because I can imagine that problems can also occur when a new packets pre-empts an transmission of an earlier packet still in progress.
15:14:23domonokyat least in the main-binary it works with all frequencys (as long as the debug menu cpu-frequency works already) which should be enough.. :-)
15:14:47funmanI'll post an updated patch, domonoky please concatenate your diff with it and then we'll wait for JdGordon and kugel
15:14:51gevaertsbertrik: bertrik I don't see any indication of pre-empting
15:15:01funmanoh JdGordon you pasted the diff already
15:15:37gevaertsbertrik: you can download the traces at gevaerts/usb"> and have a look yourself. Maybe I missed something obvious
15:15:39JdGordonyeah, I'm happy with the lcd performance as it is (maybe even get rid of the delay)... it can be fiddled with later
15:16:09 Quit bertrik ("switching horses")
15:16:09domonokyfunman: my change is target\arm\lcd-ssd1815.c, Line 105, s/4/3 :-) very small patch.
15:19:06funmancool, now we just need the Fuze fix
15:19:46funmanameyer: you asked about Clipv2, do you own one?
15:19:54*gevaerts is annoyed. The D2 usb enumerates correctly if you're watching the logf display, but it hangs if you don'
15:21:40*JdGordon wants to commit the time/date screen patch, but the button bar gets in the way if its enabled :/
15:23:14 Join bertrik [0] (
15:30:56bertrikgevaerts, it seems the data from the very first READ is already truncated, right?
15:31:27gevaertsbertrik: Yes. That could be a coincidence though
15:34:05n1sbertrik: static consts are often cheaper in binsize than creating constants on the stack
15:35:01 Quit stoffel_ (Read error: 113 (No route to host))
15:36:22funmandomonoky: I'm now failing inside codec_load_ram() ..
15:37:18 Join MethoS- [0] (
15:37:39domonokyfunman: that works for me. do you know where it fails there ?
15:37:59funmansoon ^^
15:38:17domonokyand how big is your codec buffer, and is it in IRAM ? ( i didnt apply your iram patch)..
15:38:34bertrikn1s, that surprises me a bit
15:38:51funman256kb in iram, yes
15:39:10bertrikand we seem to be making it static in only about half the cases
15:39:33*domonoky has 265kb in dram..
15:40:09n1sbertrik: in quite some cases the text messages can't be static consts as they depend on arguments to a function or similar
15:40:31amiconnJdGordon: The button bar cannot get in the way by definition, 'cause it's a desired feature
15:40:57 Quit MethoS (Read error: 60 (Operation timed out))
15:41:00JdGordonits an anoying waste of screen...
15:41:52funmandomonoky: some problem with addresses, codecbuf is in iram, and hdr->*_addr are in dram
15:42:37domonokyso its surely the iram patch making you problems..
15:42:48funmanof course ^^
15:45:58 Join nplus [0] (n=nplus@141.25.Globcom.Net) puts codecs in DRAM
15:47:41bertrikgevaerts, you've probably noticed this, but it seems the number of missing bytes in big bulk transfers is always 1 + 32*N
15:49:13 Join raphi [0] (
15:50:03funmandomonoky: ok codec_main() of mpa.c is running \o/
15:50:51raphihi guys, does somebody know a good page for self-repairing iPods? actually I would like to see the internal design etc.. THX
15:53:04funmandomonoky: hmm I believe codecs overwrite already busy iram
15:53:57funmans/I believe//
15:54:27***Saving seen data "./dancer.seen"
15:54:44funmanand I don't know how the plugin (and codec) linker script can know which size of the IRAM is already filled by the main binary
15:55:06 Join fml [0] (n=4fd3ccae@gateway/web/cgi-irc/
15:55:15gevaertsbertrik: I hadn't actually. Interesting
15:55:23n1sfunman: usually iram is split into 2 sections, one used by the core and one for plugins/codecs
15:56:18fmlpixelma2: why is the key map file (.tex) for irivers named h1xx_h300? Why not h1xx_h3xx?
15:56:31funmann1s: Ok I see what you mean, IRAMORIG is shifted in
15:57:23 Part franky
15:59:14*gevaerts slaps several foreheads
16:06:07*bertrik wonders whose foreheads were slapped exactly
16:06:55gevaertsI found why USB didn't work properly on the D2. A missing #define...
16:09:43n1sJdGordon: any plans to work on FS #9260?
16:10:01gevaertsNot entirely there yet. It stops after 700MB
16:11:56fmlgevaerts: integer overflow?
16:12:08 Quit MegafEee (Remote closed the connection)
16:12:26 Join MegafEee [0] (n=Linux@unaffiliated/megaf)
16:13:15bertrikgevaerts, doing a blocking write seems to use a wakeup_wait to wait for completion, but usb_drv_wait seems to look directly in the qh structure, is this correct?
16:14:39bertrik(lines 589, 717 in usb-drv-arc.c)
16:16:16gevaertsbertrik: I'm not sure if it is correct, but nothing seems to ever call usb_drv_wait()
16:16:53bertrikah right
16:17:19gevaertsIt probably needs to go. Nothing like unused functions taking up binsize
16:18:25*gevaerts should probably trace a beast connection, to make sure that it really works properly
16:25:54 Join webguest81 [0] (n=4dfd9ab7@gateway/web/cgi-irc/
16:26:36 Quit webguest81 (Client Quit)
16:28:32 Join ap0 [0] (
16:28:34 Quit ap0 (Remote closed the connection)
16:29:06 Join bertrik_ [0] (
16:29:11 Quit bertrik ("Leaving")
16:29:33 Nick bertrik_ is now known as bertrik (
16:34:43 Quit fml ("CGI:IRC")
16:41:48 Join BigBambi_ [0] (n=Alex@rockbox/staff/BigBambi)
16:41:58 Join ender1 [0] (
16:43:29 Join agaffney_ [0] (n=agaffney@gentoo/developer/agaffney)
16:43:43 Quit agaffney (Nick collision from services.)
16:43:47 Nick agaffney_ is now known as agaffney (n=agaffney@gentoo/developer/agaffney)
16:44:48 Quit nplus (
16:44:48 Quit Jaykay (
16:44:48 Quit ender` (
16:44:48 Quit skiy (
16:44:48 Quit Seed (
16:44:48 Quit BigBambi (
16:44:48 Quit Bagder (
16:44:48 Quit l7 (
16:45:46NJoinl7 [0] (
16:47:55NJoinSeed [0] (
16:48:05funmandomonoky: iram part of codecs is not copied ..
16:48:46domonokyso thats why we get undefined instructions with mpa ? sounds likely..
16:49:19funmanlook for USE_IRAM in export/config.h and add the as3525 case (I have linking problems but that may be local to my diffs)
16:50:16funmanld: rockbox.elf: Not enough room for program headers, try linking with -N
16:50:31funmanld: rockbox.elf: final linked failed: Bad value
16:51:04 Join Horschti [0] (
16:51:48 Quit Horscht (Nick collision from services.)
16:54:29*domonoky gets the same error...
16:55:00funmanI understand that rockbox is too big, but I remember another message (region DRAM is full)
16:55:35funmanlinking with -N does work, but then "region PLUGIN_IRAM is full"
16:55:59funmanlet's not define USE_PLUGIN_IRAM
16:56:24*funman is a bit lost
16:57:19 Quit Llorean (Read error: 104 (Connection reset by peer))
16:57:48 Join Llorean [0] (
16:58:40NJoinskiy [0] (
16:59:03funmanafter all if codec buffer is already in iram, no need to bother copying icode/idata etc..
16:59:07bertrikgevaerts, one other thing about USB: when a bulk transfer goes wrong, it seems that there is always a problem with the first transaction (but possible with other transactions in the transfer too)
17:00:22 Nick soap is now known as soap|sh (n=soap@rockbox/staff/soap)
17:00:41domonokyfunman: putting the codec buffer into iram is new for rockbox, so it might not work correctly, but i get the same errors with codec buffer in dram.
17:00:58 Nick soap|sh is now known as soap (n=soap@rockbox/staff/soap)
17:01:23funmandomonoky: I think you need to modify apps/plugins/
17:01:48funmannow codec seems to load but has no effect
17:01:53funman(no reboots at least ^^)
17:02:03NJoinBagder [241] (n=daniel@rockbox/developer/bagder)
17:02:10NJoinnplus [0] (n=nplus@141.25.Globcom.Net)
17:04:05kugelfunman: I'm going to test the patch now. What's the plck now?
17:04:36funmannot modified, it's not relative to the pll : it's frequency is derived from the 24MHz crystal
17:05:13domonokyfunman: does it call pcm_play_dma_start() ? For me the wav codec does it, but only once.
17:05:39domonokyfunman: is the iram really at 0 ?
17:06:48funmanthere is other aliases however, but it's just more convenient to use 0
17:07:15 Join Horscht86 [0] (
17:07:35 Nick Horscht86 is now known as Horscht (
17:07:36gevaertsbertrik: I'm not sure. The transfer starting at event #115730 has a crc errors and some missed replies, but the first few transactions are OK
17:08:35linuxstbdomonoky: I assume you're not overwriting the interrupt vectors at 0x0?
17:08:56funmandomonoky: not sure .. it's called sometimes but the WPS overwrites my lcd_puts()
17:09:03 Join jhulst [0] (n=jhulst@unaffiliated/jhulst)
17:09:07funmanhm the timer is now at 0:01 !!
17:11:01funmandomonoky: do you want my diff for rockbox app and plugins/codecs linker scripts?
17:11:18domonokyfunman: would be nice.
17:11:19funmanit's not very clear to read, but it seems to work ^^
17:12:35kugelfunman: not sure why JdGordon was happy with the lcd performance. His clock is too slow on the fuze
17:12:47funmanwhat it does is use 0 -> 0x8000 as iram for the main build, and 0x10000 -> 0x30000 as a codec buffer. It also declares iramsize as 0 so codecs think they have no iram available
17:13:01funmankugel: perhaps because the hardware of fuze and e200v2 are different
17:13:22funmanhe mentioned slowness but not critical (i.e. it can wait for a later fix)
17:13:29kugelfunman: iirc it's the same controller
17:13:44funmanwhy aren't the 2 drivers merged then??
17:14:08kugelfunman: because the fuze is flipped, so there's some difference
17:14:16kugelbut yea, one could try merging
17:14:21 Part raphi
17:14:22funmanif it's minor differences, #defines could help
17:14:40funmanso we would avoid duplicating the work, or fixing one and forgetting the other
17:14:58*gevaerts is browsing the d2 filesystem from his PC while the d2 is running rockbox :)
17:15:26funmangevaerts: you dd the filesystem when running the OF, and browse that image after having rebooted the d2 to rockbox ?
17:15:36kugelfunman: I'll have a look at that. Anyway, I tested to remove the lcd_delay, and it works flawlessly. Remove it altogether with this patch or seperately
17:15:50funmankugel: just put it in please
17:16:32gevaertsfunman: The OF has its own formatter built in, so no need to do all that :)
17:16:41funmantake the patch on FS #9588 , add what you need to get a correctly working fuze lcd driver, paste the new patch.
17:17:25funmanif removing the delays results in correctly working display, then remove them
17:17:43funmankugel: make sure you test with != cpu clock speeds (change the argument of set_cpu_frequency() )
17:18:42funmandomonoky: pcm_play_dma_start() is only called once, when starting / or each time I pause and unpause
17:19:40domonokyfunman: same for me... something is not working correctly.. maybe because it calls pcm_bytes_waiting() and we return always 0 ? :-)
17:19:56kugelfunman: huh?
17:19:57*funman returns the amount of bytes not sent through the dma channel
17:20:20funmankugel: in the patch I posted : + set_cpu_frequency(CPUFREQ_DEFAULT);
17:20:36funmanreplace DEFAULT with MAX and see if your lcd driver still works
17:21:08linuxstbkugel: What makes you think it's the same controller? The register layout is different...
17:21:21domonokyfunman: so thats not why it only requests one pcm transfer... :-/
17:21:31kugelwill do
17:22:07funmanperhaps my 'chain' of transfers isn't working
17:23:36kugellinuxstb: the driver looks quite similar. I haven't looked into it detailed so I might be wrong
17:23:50linuxstbYes, it's definitely similar, but not the same.
17:23:53bertrikgevaerts, ah yes, you're right
17:24:02funmanby the way I think we should create 2 linker scripts : 1 for targets with 2M of RAM and 1 for targets with >2M (with the codec buffer in SDRAM)
17:24:13domonokybut i would think playback should call pcm_play_dma_start multiple times, even when the dma transver doesnt work.
17:24:17kugelfunman: works with max
17:24:25 Quit Horschti (Read error: 110 (Connection timed out))
17:24:40funmankugel: then post an updated patch please, and I'll commit it
17:25:23kugelbut I think when we want to support videos, we might need to have to raise pclk, since I think the lcd is too slow for animations (and I already used the lowest divider now)
17:25:35kugelbut yea, that can surely be fixed later
17:26:31 Quit MegafEee ("KVIrc 4.0.0 Insomnia")
17:27:46n1sfast lcd updates are good for the general UI, not only videos
17:29:40kugeln1s: Sure, but I think for videos it needs to be a bit faster than for UI, and it's not so slow
17:30:39kugelfunman: done!
17:31:39funmankugel: thanks
17:32:00kugelyou're welcome :)
17:33:00n1skugel: faster updates are always good to avoid the ui feeling sluggish, especially when playing music and not boosting
17:33:51funmanIMO we should mesure battery life reduction with higher pclk to find a good compromise
17:38:48domonokyquestion for people with knowledge over rockboxs pcm system: should pcm_play_dma_start() only play the requested size, or should it somehow use the "pcm_callback_for_more" ??
17:39:31kugeln1s: yes, but we aren't at the point we can enjoy the UI yet. As I said, that can be fixed later, for now it's only important that it works
17:39:45linuxstbdomonoky: The DMA interrupt handle will call the get_more() callback when the first DMA transfer is complete.
17:39:52domonokyat least the dma handler in pcm-coldfire looks like it uses the pcm_callback_for_more.. so we probably miss that..
17:40:39n1skugel: yes, sure getting it working ok is more important than optimization i just disagree with your statement that fast lcd is more important for video that other things :)
17:41:39kugeln1s: I just meant, for UI 12fps would be already sufficient (imho), which isn't the case for videos
17:41:54domonokylinuxstb: it gets a bit more complicated, as the dmac on as3525 only can transfer 2^11bit at once.. so we need more then 1 transfer for one "chunk"... but we really miss the callback at moment :-)
17:42:45n1skugel if you only get 12 fps that means that if something else is using 50% of the cpu you get 6 fps and if you are not boosting you would get something like 2fps
17:43:48*funman hears something that looks like voice
17:44:06*n1s wonders what a voice looks like...
17:44:14kugelwave forms
17:44:23domonokywow.. now with the get-more callback ?
17:44:26funmanway too fast voice, then painful sound
17:45:07funmanwow now something that looked like AC/DC guitar ..
17:45:27domonokywould you like to share our testpcm code ? it think your attempt fits now better then my LLI code.. :-)
17:45:30funmanbut my code is probably very broken ..
17:47:11funmanfull diff (with linker changes)
17:47:30*domonoky sees a few gentlemen mails in the not so far future :-)
17:47:31 Quit jhulst (Remote closed the connection)
17:47:39funmanhmm I notice a wrong cgu_audio divider
17:48:32funmandomonoky: which frequency for i2so_mclk do we want to use?
17:48:55domonokyit has to fit 128x the sample frequency...
17:49:04domonokyso ~5,6Mhz..
17:49:17funmanand don't forget the PLL just changed :)
17:50:15funmanwow I definitely hear real sound, but it looks like pcm transfers are repeated
17:54:29***Saving seen data "./dancer.seen"
17:54:31funmandomonoky: I notice dma_size should be >> 4 after it's read from the callback (I store the number of 4*32bits transfers)
17:55:48domonokywhat m_clock divider do you use now ?
17:56:19funman248MHz / 44
18:01:08funmanperhaps pcm_set_frequency() would be worth writing
18:07:44funmanmy source is WAV@44.1KHz but I still have problems finding the correct frequency :(
18:08:19*domonoky gets "noises" with this code, but no recognicable sound..
18:09:59funmantry with lower clock (divider == 100)
18:13:49*domonoky now gets short burst of recognicable sound, with bursts of noice in between..
18:14:10funmangetting close :)
18:16:02funmanI too get frequent noise (pops, cracks) but the major part is recognizable sound
18:16:20funmanlooks like it's too slow, but I know the rythm of this song and in fact it's too quick
18:16:47funmanperhaps we are transferring too much at a time ..
18:17:00bertrikdma_start_addr should probably be a unsigned char * instead of a void *
18:17:16funmancosmetics :)
18:23:00funmanI get something better with 1*32bits transfers ..
18:23:59funmanlots of cracks and weird sound, but it's definitely there ..
18:24:59funmanwhat are the requirements for the 'gentlemen' e-mail?
18:25:50domonokysvn-ready pcm code ? :-)
18:25:53gevaertsA mail client?
18:26:16domonokyand of course, dont forget the ladys.. :-)
18:26:33funmandomonoky: perhaps I misread the OF and it used 4*8bits transfers
18:27:22funmanmp3 doesn't decode however, perhaps it's too slow to fill the pcm buffer
18:27:34gevaertsfunman: the first one boasts about playing four seconds
18:28:17funmanmp3 is 36.81% realtime :-)
18:28:59domonokywow DMA_S4 is much better.. still some noises, but really recogniceable..
18:29:06funmanand the cpu should be maxed (248MHz)
18:29:08domonokyäh DMA_S1
18:29:18funmandomonoky: I used S1 and clock divider 44 (~5.6MHz)
18:31:09funmanperhaps a low pclk is too slow for SD transfers (or we use a too high clock divider in SD code)
18:32:42funmanwith PCLK @ 62MHz instead of 24MHz I see 168% realtime decoding speed for a mp3
18:33:27domonokyso maybe we should higher pclk again.. :-)
18:33:37*funman looks somewhere else
18:33:43funmanI want to verify SD code first
18:34:30funmanah we use maximal speed already
18:34:40funmanso I suppose yes .. we want to do that :o
18:35:35 Join SoapWork [0] (n=42c07542@gateway/web/cgi-irc/
18:36:27funmandma_start_addr should be increased _after_ dma_enable_channel()
18:36:59funmanno more cracks/pops ^^
18:38:38 Quit Darksair ("Everything that has a beginning has an end.")
18:39:44 Join guyjones [0] (
18:39:54funmanI have deadlocks in mp3 codec, but wav is fine ..
18:40:13bertrikhow about we try to follow the clock recommendations on page 112 of the datasheet?
18:40:46 Join Thundercloud [0] (
18:41:48funmandepends which precision we can get by not modifying the PLL
18:42:08funmanif needed there is another pll, we could use it to keep fclk independant of pclk
18:48:59*domonoky now gets no noises anymore, but it shutsdown shortly after start of playing.. but there may be other changes in my code which causes this.
18:49:17funmandomonoky: are you playing wav ?
18:50:32funmanAfter these emotions, I'm going to take some rest ;)
18:50:37funmanthanks a lot
18:50:58 Quit funman ("leaving")
18:59:00 Join miepchen^schlaf [0] (
19:01:07 Join leachim6 [0] (
19:02:03leachim6when I use the Apple OS on my iPod 5g (original) all of my albums have art
19:02:20leachim6is there a way I can automatically get album art for all my albums in rockbox
19:02:29leachim6without doing hundreds of albums manuall
19:03:32bertrikdomonoky, what did you change compared with funman's patch?
19:05:00domonokybertrik: i only changes what funman said, changed the m-clock divider, move the dm_start_addr increasing.. but there are of course much more changes in my tree from experimenting...
19:05:51domonokybertrik: are trying to repoduce our sound experiment ?
19:06:12bertrikdomonoky, yeah playing with it a bit
19:06:33domonokyyou need to make sure, you have enough audio buffer... and try wav files.
19:07:58*domonoky has now 1.14Mb buffer, now that the codec is in IRAM.. :-)
19:08:31leachim6so...can anyone relate to my situation
19:08:48domonokybertrik: you get any sound ?
19:09:00leachim6my bad
19:09:04leachim6"How to Use Embedded iTunes Cover Art On An iPod "
19:09:05leachim6from the wiki
19:09:09leachim6Reading is fun!
19:09:24bertrikdomonoky, I get sound from an mp3, but it repeats a lot and has noise
19:09:31domonokyleachim6: take a look at:
19:10:02leachim6I know...that's where I found this from
19:10:29domonokybertrik: hurray ! repeating is probably from too slow encoder, and the noises from the incorrect dma_start_addr
19:12:12domonokyand perhaps the dma_size, funman said it should be >>4 after the get_more callback.
19:12:59domonokyanother change: DMA_S1 instead of DMA_S4
19:13:59bertrikI'm not familiar at all yet with the DMA controller
19:15:00domonokyperhaps we should just wait for a cleand up patch from funman :-)
19:16:07 Quit goffa_ (Read error: 60 (Operation timed out))
19:16:48leachim6does anyone here use this on linux ->
19:17:58 Join goffa [0] (n=goffa@
19:19:27bertrikdomonoky, what do you use as a wav file? some 16-bit stereo wav file? and is there an easy command line tool to create a wav from an mp3?
19:20:44 Join BHSPitLappy [0] (n=BHSPitLa@unaffiliated/bhspitmonkey)
19:20:53domonokyyes,just a small 44khz/16bit wav file...
19:25:02bertrikdomonoky, it goes way too fast here
19:25:55domonokythen change the mclock divider in CGU_AUDIO ... its not correct anymore. (different pll frequency).
19:30:17 Join Nico_P [50] (n=nicolas@rockbox/developer/NicoP)
19:31:34 Join gromit` [0] (
19:33:41 Join goffa_ [0] (n=goffa@
19:35:13 Join nuonguy [0] (
19:38:35 Quit gromit`` (Read error: 110 (Connection timed out))
19:39:14 Join funman [0] (
19:41:11funmanI noticed a problem with DMA_S1 (1*32bits) : I miss part of the sound
19:41:32funmanI listen to AC/DC (low frequency instruments and high frequency voices) and the voices are hardly noticeable
19:42:43 Quit m0f0x ()
19:42:47 Join gromit`` [0] (
19:44:31 Join Loules [0] (
19:46:58*funman wonders if we should use 16 bits wide transfers
19:48:02 Quit goffa (Read error: 110 (Connection timed out))
19:49:30funman2 16-bit audio samples (one per channel) : both values are written to the 32 bits wide I2SO_DATA register
19:49:41funman(at the same time)
19:50:02 Quit gromit` (Read error: 110 (Connection timed out))
19:51:33domonokythat how the datasheet says it should be for 16bit samples.
19:54:30funmanso I think we are correct to use 32bits transfers
19:54:32***Saving seen data "./dancer.seen"
19:58:09 Quit ender1 (" The most important lesson we can learn from history is the inability of humans to learn lessons from history -- Aldous Huxl")
19:59:22domonokycould you post a patch again, with all your recent changes ?, then i will try on a fresh svn again, to see if to shutdowns are gone..
20:02:24 Join robin0800 [0] (
20:04:34kugelfunman, domonoky: great job of you. I love you :)
20:05:49bertrikwe only hear the difference between left and right or something like that :P ?
20:06:23 Join ender` [0] (
20:07:44 Join jhulst [0] (n=jhulst@unaffiliated/jhulst)
20:07:45funmanat the condition to be very focused, because making the difference between 2 noises is hard :P
20:08:52 Join karashata [0] (n=karashat@
20:09:54 Join MethoS-- [0] (
20:10:21Bagderfunman: congratulations and well done!
20:10:49 Quit goffa_ (
20:10:49 Quit BHSPitLappy (
20:10:49 Quit Seed (
20:10:49 Quit l7 (
20:10:49 Quit MethoS- (
20:10:49 Quit soap (
20:10:49 Quit perrikwp (
20:11:28funmanI was not alone - especially if considering the usefulness/time given ratio :-)
20:11:42NJoinl7 [0] (
20:11:46 Join goffa [0] (n=goffa@
20:12:14kugeldomonoky, funman: Really, great job. I'm looking forward to hear something from my fuze
20:14:46NJoinSeed [0] (
20:15:27gevaertskugel: you could try adding a playlist control file (maybe stolen from another player) and setting the right settings for auto-resume
20:15:54domonokyfunman: if i only apply your patch to svn, i get a dma error channel 0 ...
20:16:13funmanweird Oo
20:16:36domonokychannel 0 is sd, correct ?
20:16:46funmandid you update the bootloader as well ? (cgu_peri change)
20:17:09funmanperhaps it's not needed if we can modify the divider in cgu_ide to get fast transfers
20:17:33domonokyyes, and the dma error comes, when i start to play something.
20:18:01funmanwhat if you look at MCI_STATUS(0) ?
20:20:28funmanjhMikeS: do you remember something about high frequencies not being played on iaudio x5 ?
20:23:16bertrikfunman, maybe it's a property of the codec, like the HPCM
20:23:53funmanthat's what I was looking for, but you seem to know that part better than us ;)
20:24:19bertrik=headphone common mode, a kind of virtual ground not being enabled. That would also explain hearing L minus R instead of stereo
20:24:40domonokyfunman: MCI_STATUS is 020A000 when i get the dma error. now lets check the datasheet :-)
20:25:29NJoinBHSPitLappy [0] (n=BHSPitLa@unaffiliated/bhspitmonkey)
20:26:22NJoinsoap [50] (n=soap@rockbox/staff/soap)
20:29:02funmandomonoky: perhaps stupid: did you change config-m200v4.h ?
20:30:58domonokyhm, nope. perhaps i should adjust the codec size :-)
20:31:17 Join mofux [0] (
20:31:22 Join XavierGr [0] (n=xavier@rockbox/staff/XavierGr)
20:31:27*kugel is curious about the CPU_FREQ #define in config-fuze.g
20:31:30kugel.h that is
20:32:16funmanjust grep for it, it's used in timers or something
20:32:35kugelit doesn't fit out value though
20:33:15kugeluhm, why is my binsize 1,3MB?
20:33:38kugelwith mass features disabled
20:33:58funmanvim :)
20:33:59 Join shotofadds [0] (n=rob@rockbox/developer/shotofadds)
20:34:36 Join gromit` [0] (
20:35:04kugelBinary size: 1236636
20:35:05kugelActual size: 1236628
20:35:07kugelRAM usage: 806540868
20:35:17kugelthere IS something wrong
20:35:32funmandid you change your linker script ?
20:37:01bertrikon the c200 there is a trick to shorten database refresh by making a directory read-only. Do we know what file the ams sansa update on database refresh?
20:37:16*bertrik hears true stereo
20:37:20kugelfunman: svn version is the same
20:37:34gevaertsshotofadds: maybe you saw it already, but I got USB to work on the D2
20:37:40domonokyfunman: now with the codec size of the clip, instead of the dma error, it shutsdown again.. :-/
20:37:40funmankugel: I think the problem can be seen in the map file
20:37:46 Part Loules
20:37:55funmandomonoky: sorry I have no clue here..
20:38:14*domonoky will investigate..
20:38:22bertrikit was indeed the HPCM_off setting, it should not be set for the clip
20:38:37bertrik(in as3514.c)
20:39:08kugelI'm not so familiar with maps file, maybe you spot something obvious
20:39:23funmansorry I don't want to read that big file
20:39:58kugelfunman: memory configuration is a line 265, it looks weird to me
20:40:05funmanand I got the same RAM usage on the clip
20:40:16shotofaddsgevaerts: yes, I saw and I'm just trying it now. good work!
20:40:26kugelIRAM 0x00000000 0x00050000 - DRAM 0x30000000 0x00680000 - *default* 0x00000000 0xffffffff
20:40:48funmanbertrik: wow, now I hear the all the frequencies !
20:41:06kugelfunman, Bagder: what do you think? Maybe it's time to put the real targets on the build table as well?
20:41:34funmanI prefer to wait when sound will be committed
20:42:02 Quit gromit`` (Read error: 110 (Connection timed out))
20:42:12gevaertsshotofadds: basically you were only missing one define. It needs both HAVE_USBSTACK and USE_ROCKBOX_USB. Maybe USE_ROCKBOX_USB needs to be inverted, as the normal case should be to use rockbox for usb
20:42:42*funman feels safe in new york city with rockbox on the Sansa Clip
20:44:13Nico_Pfunman: congratulations!
20:44:49kugelwhy's my binsize so huge?
20:44:56kugelit's definitely wrong
20:45:08funmanNico_P: thanks, but don't forget i'm not alone ;)
20:45:37 Quit jhulst (Remote closed the connection)
20:45:38 Join Horschti [0] (
20:45:47Nico_Pof course :)
20:46:22 Quit Horscht (Nick collision from services.)
20:47:03 Join Jaykay [0] (
20:47:07linuxstbkugel: If the build itself is working, then the problem will likely be the script
20:47:23funmannow we must clean up and finish the PCM code, then find why the system locks/shutdowns (no more reboots though)
20:47:24bertrikwow, even mp3 seemed to work for a while ! :)
20:47:30jhMikeSfunman: high frequencies not on x5?
20:47:45kugellinuxstb: the build itself is too big (not just what in info is written). And no, I can't say it's working
20:48:00 Join zicho [0] (
20:48:13funmanjhMikeS: in fact bertrik fixed the problem since : something with hpcm setting
20:48:22linuxstbkugel: Then I would look at the linker script.
20:48:24zichoif i get I/O-errors while trying to edit my .rockbox-folder, is there anyway i can fix it, or do i need to format completely?
20:48:24kugellinuxstb: the difference between rockbox.elf and rockbox.bin should be big?
20:48:31funman8 bytes
20:48:40funmanoh oops no. it shouldn't be that big
20:48:48funman(but bigger than 8 bytes)
20:49:05kugelmy rockbox.elf is 798k, rockbox.bin 1,3M
20:49:19kugelfunman: rockbox.bin should be even smaller, right?
20:49:32funmanthat means some problems in memory segments, because rockbox.bin should be smaller (only code, elf headers removed)
20:49:44funmanhm for me it's a bit bigger
20:49:45krazykit`zicho, you at least need to fsck/chkdisk that filesystem.
20:49:47linuxstbzicho: It depends what's causing those errors... What device are you running Rockbox on? Have you tried a disk scan (e.g. chkdsk on windows, fsck.vfat on Linux)
20:49:51funmanperhaps the stacks are objcopied
20:50:16zicholinuxstb, im running on an F60
20:50:25zichoalso im running ubuntu
20:50:34kugelfunman: can you take a look at the linker script?
20:50:35 Join einhirn [0] (
20:50:41kugelI have no idea of those
20:50:45bertrikmaybe implement pcm_play_lock/unlock, should be simple
20:50:46funmankugel: no
20:50:53shotofaddsgevaerts: IIRC I copied those USB #defines from vitja's config-iaudio7.h, so that one is probably wrong too.
20:50:59*bertrik tries
20:51:26funmankugel: if ram usage is heavy, it means you have heavy things in there, period.
20:51:47funmanif you want to find where/what exactly, look at
20:51:51kugelfunman: ram usage is more then the fuze has ram
20:52:03shotofaddsgevaerts: I can browse the root directtory in Windows ;-)
20:52:04kugelyour ram usage is wrong as well, but you don't seem to care
20:52:40funmanexact I don't care because I don't see how this is a problem I can't report to later
20:52:43zicholinuxstb, how do i fsck.vfat on my F60, then?
20:53:07gevaertsshotofadds: great :) So when will the FTL Work properly? ;)
20:53:12funmanram usage is 768MB on the Clip - ridiculously high, and until it's not proven that this is the source of the bugs we encounter, I'll focus on other stuff
20:53:27kugelfunman: also, I'm more worried about my binsize
20:53:35shotofaddsgevaerts: but it only works if I replace the "not supported" panic in usb_drv_send_nonblocking with a return statement.. and I get *lots* of random pauses
20:53:36funmanlooks normal to me
20:53:41shotofaddsgevaerts: haha!
20:53:43kugel1,3M is not normal
20:53:44shotofaddsfunny man
20:53:54*funman decoded ogg & flac
20:54:02kugelit should be <500k, since I have many features disabled
20:54:43funmanlook at individual .o files and check their size
20:54:46gevaertsshotofadds: weird. I didn't notice issues on linux...
20:54:56gevaertsAt least not that sort of issues
20:55:17*amiconn returns, and sees that he needs to congratulate funman and domonoky
20:55:37krazykit`zicho, sudo fsck.vfat /dev/whatever_it_is, once it's unmounted
20:57:30shotofaddsgevaerts: I think I remember reading that Windows attempts to write to sector 0 on connection, or some such craziness. In linux Rockbox just freezes :/
20:57:38kugelI have .o's for ingenic_jz47x
20:58:05gevaertsshotofadds: I've actually seen windows do that.
20:58:14kugeloops, i wasn't in the build dir
20:58:22zichokrazykit`, fsck.vfat didnt show any errors.
20:58:33shotofaddsgevaerts: though I don't know if that's the reason I always get that panic
20:58:38funmanogg/flac/wav sounds OK, but mp3 sounds weird (small cuts). Luckily only during a short time (before it crashes)
20:59:40shotofaddsgevaerts: how difficult would it be to expose a second drive in UMS (eg. for physical NAND access)? I guess this is what you did with the RAM disk stuff..
21:00:37zichokrazykit`, if i format everything on my player, will it be possible to reinstall rockbox?
21:01:15krazykit`zicho, you might have to use the recovery mode (taking it apart) if you reformat. i'm not quite sure how it works.
21:01:39zichowhat does the recovery mode do?
21:02:01 Join einhirn_ [0] (
21:02:22zichoisnt there just some way to replace the corrupt .rockbox with a functional one?
21:02:51kugel"export LD=/usr/local/arm-elf/bin/arm-elf-ar" doesn't sound right, does it?
21:02:59funmannot really
21:03:01kugelI'd expect arm-elf-ld
21:03:11funmanwell ar is used to make static libraries
21:03:12kugelthat's in my makefile
21:03:12gevaertsshotofadds: There are two ways you can do that. One would be to change the nand driver to do multivolume things (i.e. add the appropriate IF_MV2(int drive,) stuff, see the mmc or sd drivers for examples), the other is to use the multidriver patch (FS #9545). The patch doesn't fully work right now for the d2 though, but I'm working on it
21:03:37kugelexport AR=/usr/local/arm-elf/bin/arm-elf-ar is the next line, so it's not ignored
21:03:57gevaertsamiconn: I'm planning to do a trace on the beast tonight or tomorrow, to see if it really doesn't have the same issues as portalplayer
21:04:20krazykit`zicho, you could try to just delete the corrupt .rockbox dir and unzip a new build
21:04:31amiconnI couldn't notice any, the beast is working fine even through my "crazy" hub
21:04:33shotofaddsgevaerts: does your d2 exit the USB screen when you disconnect the cable? mine doesn't
21:04:36zichokrazykit`, if i do that, i get I/O-error
21:04:53gevaertsshotofadds: it doesn't. That's one of the remaining issues
21:05:02amiconn(that's how I always connect it, as my standard mini-b cable is plugged into it)
21:06:02gevaertsamiconn: I know it's unlikely to show anything, but it's just a few minutes work anyway, so I'll do it just in case
21:06:22krazykit`not that it's the best option, but it's an option.
21:06:39zichoi'll look into it
21:06:46bertrikhmm, not so sure what pcm_play_lock is supposed to do exactly...
21:07:11funmanbertrik: domonoky should we open a FS patch to keep our work in sync ?
21:07:24 Quit karashata ("G'bye everyone!")
21:07:33gevaertsshotofadds: sometimes my d2 doesn't see the usb connection (or the charger connection more precisely). Have you ever seen that?
21:07:51bertrikfunman, yes probably
21:07:52 Join karashata [0] (n=karashat@
21:08:07shotofaddsgevaerts: yes, it just happened to me, but I've seen this with the OF too.
21:08:29shotofaddsI can get Rockbox to show up in lsusb, but cannot dd from it. any hints?
21:08:32bertrikfunman, do we stick to DMA_S1 or do you have plans to use bigger transfers later?
21:08:36funmanbertrik: I let you do since you'll be more advanced than I
21:08:47funmanbertrik: I think we must stick to it since I2SO_DATA is only 32 bits
21:08:51gevaertsWhat does dmesg say?
21:09:13bertrikok, then I'll reduce the * 4 *1 to just * 4 :)
21:10:19shotofaddsgevaerts: I was already pasting it ;)
21:10:37funmanbertrik: perhaps it needs to be /4 in dma_enable_channel (to have less calculations)
21:11:23bertrikok, I'll work on that and create an FS patch, ok?
21:11:47 Quit miepchen^schlaf ()
21:13:09funmanby the way, is (int)/(2^x) always assembled as >>x ?
21:13:17gevaertsshotofadds: are you by any chance looking at a some debug screen while trying this?
21:13:27 Join miepchen^schlaf [0] (
21:14:51zichowhat are the necessary files needed for rockbox to boot?
21:15:24linuxstbzicho: Rockbox itself needs everything in /.rockbox/. The Gigabeat's bootloader (Toshiba's code) requires the files inside /GBSYSTEM/
21:15:51linuxstbSo you need to keep both - and don't disconnect from your PC whilst reformatting, before you copy those files back.
21:16:21zichoso if i keep gbsystem and a clean .rockbox, its safe to format?
21:16:55shotofaddsgevaerts: hmm, yes on that occasion I was viewing the logf screen as I saw you mention earlier (although nothing was being output as I had not enabled it in the .c files). Try this one: (same behaviour, but USB plugged in from the menu)
21:18:39kugellinuxstb: the .o files aren't bigger. I think the main binary contains stuff twice or something
21:19:20 Quit einhirn (Read error: 110 (Connection timed out))
21:19:26linuxstbzicho: I think so... You can see some more info here -
21:20:05linuxstbkugel: As I said, I would blame the linker script - it's likely filling up the middle of your binary with zeros. But I'm not an expert on linker scripts, so can't help further.
21:20:08shotofaddsgevaerts: config-m200.h doesn't define USE_ROCKBOX_USB either, and that appeared to work most of the time. unless something has changed semi-recently...
21:20:13gevaertsshotofadds: when you connect "something" (I think in usb_core.c) calls usb_request_exclusive_ata() which sends messages to all queues telling things to back off the disk. As long as not all threads have replied you get the Not Ready replies
21:20:29zicholinuxstb, well, its worth a shot. it doesnt work anyway.
21:20:30 Quit einhirn_ (Read error: 110 (Connection timed out))
21:21:19zichobut isnt there a way to get around I/O-errors?
21:21:37gevaertsshotofadds: it may depend on whether system_reboot() is implemented or not
21:21:38kugellinuxstb: And I don't have even a little idea of linker scripts
21:22:14kugelthat's why I suggest we put the ams sansas on the build table, so if someone with an idea stumbles upon that
21:22:28kugelI think it's a good moment anyway now, as we have sound
21:23:32 Quit advcomp2019 (Read error: 104 (Connection reset by peer))
21:24:49 Join draft [0] (
21:25:36drafti have iPod Mini 4GB (2nd revision) and i think it has some bad sectors on it's HD.. is there any possibility to format the player so it marks the bad sectors?
21:25:58drafti have the latest rockbox installed
21:26:10 Quit zicho ("Ex-Chat")
21:27:07 Join advcomp2019 [0] (n=advcomp2@unaffiliated/advcomp2019)
21:28:14Jaykaydraft: use checkdisk of windows.....
21:28:25shotofaddsgevaerts: well, that makes sense but I can't think what would make the difference between my build and yours. I'll try a plain SVN in a while...
21:28:25Jaykayfrom windows
21:28:57draftJaykay: should i format it?
21:29:00draftonce again?
21:29:47gevaertsdraft: one thing that's fairly popular with ipod minis these days is replacing the hard drive with a compactflash card.
21:30:03draftgevaerts: how much do they cost?
21:30:11drafti'll format the whole thing
21:30:15amiconnDepends on what size you want
21:30:16draftand reinstall rockbox
21:30:18Jaykayjust run chckdsk x: /r , x is the drive letter
21:30:24kugelcan I have more verbose make output?
21:31:11funmanmake V=1
21:31:24gevaertsdraft: depends on capacity. a 32GB card will cost around 60 euro these days I think. if you're happy with 4GB I can't imagine it being more than 10 or so
21:31:25zequick look i'm seeing 8g cf for $20-$30, 16gb for $30-$50 (bottom-end prices, they both go far higher)
21:31:42zeoh sorry, forgot US$ is useless
21:32:29drafti'm using
21:32:56gevaertsamiconn: as predicted the beast has no errors at all
21:34:09*gevaerts wonders why the usb_enable(true) at usb.c:260 is there
21:34:25 Join esthar [0] (
21:34:55gevaerts"make sure the usb controller is operating before we reset the player"?
21:34:56kugelHmm, arm-elf-objcopy makes it so big
21:35:25kugelthe stripped elf is 600K, arm-elf-objcopy -O binary creates a 1,3M .bin
21:36:09amiconnCheck the .map
21:36:38amiconnThere's most probably some section linked into a different memory area than it should
21:36:53kugelamiconn: I don't understand the .map file really. would be nice if you could have a look
21:37:14 Join webguest67 [0] (n=5ce11809@gateway/web/cgi-irc/
21:38:35 Quit skiy (Remote closed the connection)
21:39:16amiconnThe problem is obviously that .iram (.icode...) goes after .bss, so that the whole bss ends up in the binfiöe
21:39:33 Quit webguest67 (Client Quit)
21:39:36 Quit jhMikeS (Nick collision from services.)
21:39:42amiconnThe .lds needs fixing
21:39:42 Join jhMikeS [50] (n=jethead7@rockbox/developer/jhMikeS)
21:41:29bertrikfunman, the FS patch for ams sansa playback is at
21:41:29kugelamiconn: if I know how..
21:41:49bertrikactually didn't change much except for the dma size / 4
21:43:14funmanbertrik: saratoga just told me we have to use unsigned types for /4 to be a >>2; but perhaps it's better to directly use a shift
21:43:58bertrikand size_t is signed, right?
21:44:11funmanno, ssize_t is
21:44:59amiconnfunman: Indeed using >>2 is better. If you do /4 with signed types, gcc does a shift and some postprocessing to ensure the round-towards-zero rule of the division
21:45:05bertrikI think it's not a big deal, just the difference between lsr or asr in assembly I think
21:45:12bertrikamiconn, ah ok
21:46:00amiconnE.g. (-1)/2 yields 0, while (-1)>>1 yields -1
21:47:14amiconnbertrik: It's not. For signed types, gcc does an asr followed by some extra instructions to correct the error I just illustrated
21:47:58funmanhm I thought ogg decoding was more stable than mp3 but I was proved wrong
21:50:29 Quit karashata ("G'bye everyone!")
21:52:36kugelamiconn: can you give me a hint?
21:54:00 Join m0f0x [0] (
21:54:11kugelamiconn: I swapped some positions and now I get a reasonable binsize
21:54:34kugelBut of course I have no idea whether I've done it correctly
21:54:35***Saving seen data "./dancer.seen"
21:55:25kugelram usage is still insane
21:56:15amiconnfunman: Btw, there's some optimisation potential in the Clip's greylib code
21:56:32kugeldoes it look good?
21:56:59amiconnIn the asm, I mean. Since we know that only bit 0..3 and 12..15 are connected, we don't need to duplicate the low byte if we set the correct bits from the beginning
21:57:31amiconnThe output data is composed bitwise anyway
21:57:37funmanto save 1 instruction?
21:57:58amiconnYea, why not? It's a trivial optimisation
21:58:55funmanof course
21:59:32amiconn1 instruction per 8 pixels, that makes 1K instructions per frame, or about 80000 instructions per second
22:00:29drafti get this kinda error message when copying from WinXP to iPod (rockbox) draft/error.png">
22:01:12*gevaerts doesn't understand what he's doing wrong with his multidriver code
22:04:48 Quit stripwax ("Miranda IM! Smaller, Faster, Easier.")
22:06:59 Join jhulst [0] (n=jhulst@unaffiliated/jhulst)
22:07:09draftgevaerts: is it hard to change iPod Mini's drive?
22:07:13draftto compact flash?
22:07:19draftis there any tutorial?
22:07:24gevaertsdraft: it's not very hard
22:08:08draftgevaerts: any tutorials?
22:08:09gevaertsdraft: have a look at
22:08:35shotofaddsgevaerts: I get the same behaviour with SVN + #if 1 (r19276)..
22:08:46kugelI always forget to select patch instead of bug
22:09:05kugeland the category when adding a task
22:09:10gevaertsfunman: Compactflash cards have the same connector as the mini hard drive. The only thing you need to do is add some cardboard or similar to increase the thickness a bit (CF cards are thinner)
22:09:18kugel please have a look
22:09:25gevaertsdraft: that last one was for you...
22:09:40draftis this ok card?
22:09:43gevaertsshotofadds: weird. It works fine here
22:10:32gevaertsdraft: people have used A-Data cards for this before, although they usually use bigger ones
22:10:42*bertrik forgets to read beyond 1 line of IRC talk ;)
22:11:35shotofaddsgevaerts: it can't be stopping all threads, since the backlight timeout (and wakeup) still works in USB mode..
22:11:42 Quit draft ("Lost terminal")
22:12:54kugelshotofadds: still happy with fading on the d2? :)
22:13:05gevaertsshotofadds: it doesn't actually stop them, but the threads (or actually queue handlers) have to reply with a SYS_USB_CONNECTED_ACK message. They then can't read the filesystem until they get a SYS_USB_DISCONNECTED message
22:13:21shotofaddsah, ok. shows how much I know ;)
22:13:33shotofaddskugel: yeah, it's about the only thing that works reliably!
22:14:06*gevaerts likes the automatic "reset settings on boot" feature on the D2 ;)
22:16:42domonokyfunman: strange, if i enable the i2sout_mclock via CGU_AUDIO, i get those shutdowns, withouth m_clock it doesnt shutdown, but ofcourse no sound ..:-/
22:17:26gevaertsshotofadds: can you try gevaerts/"> That one works here
22:17:38funmancan you see how much time the dma callback is called before shutdown?
22:17:47funmanperhaps not aligned source memory ?
22:18:15domonokyit seems to shutdown even before pcm_play_dma_start() is called :-/
22:19:34 Quit amiconn ("swutch...")
22:19:37funmanwell something is definitely wrong, I see reboots, data aborts, undef instructions, deadlocks without messages..
22:19:44 Join amiconn [50] (n=jens@rockbox/developer/amiconn)
22:20:15funmanyou're designed to debug that because you can reproduce it :-)
22:20:57bertrikdomonoky, I enabled battery readout some time about, maybe this has some kind of effect on your m200 ?
22:21:47domonokybertrik: clean svn works.. this shutdowns have something todo with the new code..
22:22:31funmanI see a undef instruction while playing wav
22:22:31Unhelpfulhrm, "actual" size is much, much, larger than "binary" size with recorder... is this a compressed image thing, and should i be using "actual" size?
22:22:48domonokyand battery readout seems to work on m200v4, just not calibrated.
22:23:03funmanthe instruction pointer points just one instruction after the last word of wav.codec (and the alst part is the bss and "abs" so it shouldn't have been executed)
22:24:05kugelUnhelpful: which target? Some targets use the mi4 file format, which needs 512byte padding
22:24:28gevaertskugel: recorder
22:24:51amiconnUnhelpful: Yes, recorder firmware is compressed
22:25:16kugeloh, I read binsize is larger than actual size
22:25:45funmanabs = absolute symbols (defined in crt0.S and linker script it seems)
22:26:34amiconnThis is necessary to get around the archos loader limit (for recorders that can't be flashed, the user doesn't want to flash, or of course even for first-time installation)
22:28:38kugel\o/ I broke the bootloader fuze buttons
22:28:39Unhelpfuli'm still trying to reduce the new-bmp-reader footprint. it's going... ok :/
22:29:35amiconnIf you want to compare firmware files sizes without checking the .txt, build for fmrecorder, recorder v2 or one of the Ondios.
22:30:27amiconnThose firmware files are uncompressed, as they don't need to be. Archos loader limit is 400KB on those compared to 200KB on recorder v1 (and player)
22:30:38shotofaddsgevaerts: that version has the same behaviour (not ready in linux, panic in windows xp) :-(
22:31:07kugelbertrik: while you're at editing, can you have a look regarding the binsize? the .bss needs to be after .iram
22:31:13Unhelpfulamiconn: hrm, but the RAM size from the .txt is significant as well, isn't it?
22:31:21gevaertsshotofadds: really weird. Have you recently reformatted the filesystem?
22:31:37shotofaddsno, not for a long time
22:32:03 Quit pixelma2 ("-")
22:32:07gevaertsIt sometimes makes a difference here (this is on an 8GB D2)
22:32:24 Join pixelma [50] (i=pixelma@rockbox/staff/pixelma)
22:32:40 Quit {phoenix} (Read error: 104 (Connection reset by peer))
22:32:53*pixelma is late to the party but congratulates anyways :)
22:33:11kugelwho's the button thread expert here?
22:33:49kugelI tracked down the fuze's button problems to button_init() now
22:34:22kugelthat one isn't called in the bootloader where the buttons work. With the button thread in the main binary the buttons don't work anymore
22:34:38kugelif I add button_init in the bootloader, they don't work there anymore
22:36:30 Quit funman ("leaving")
22:38:08 Quit bmbl ("Woah!")
22:40:06shotofaddsgevaerts: I can give formatting a try, but not right now. Out of time tonight I'm afraid :/
22:41:03gevaertsshotofadds: ok. I'm not going to do anything anymore today either anyway
22:44:10kugelMaybe the tick task is too fast, so that our gpio gets confused
22:44:35kugelor at least the dbop afsel switching
22:45:01 Join stripwax [0] (
22:47:00shotofaddsgevaerts: seems I can use dd quite reliably in Windows, so it'll be useful even with the current problems. I'll play some more tomorrow!
22:47:32*gevaerts doesn't understand at all...
22:48:43*shotofadds is rather confused too about why dd would work but explorer hangs.. but thats windows for you. night all
22:49:53 Quit shotofadds ("Leaving")
22:51:22 Quit Bensawsome ("The awsome is gone :(")
22:51:50 Join Bensawsome [0] (n=Bensawso@unaffiliated/bensawsome)
22:55:45kugeldomonoky: the button thread is causing problems on my fuze
22:56:21amiconnButton thread? There should be no such thing...
22:57:00kugelwell, tick task
22:57:06amiconnThe button driver uses a tick task, and an isr on some targets
22:58:07kugelwhich targets use an isr?
22:58:15 Join _lifeless [0] (n=lifeless@
22:58:18kugelI think the tick task isn't going to work for the fuze
22:58:33amiconnTick tasks obviously run in isr context too, so you need to be careful
22:58:41amiconnWhy not?
22:59:10amiconnThe tick task is used on all targets, in order to have a stable time base (for button repeat etc)
22:59:45kugeldon't know. using it causes the read problems. the bootloader doesn't use the tisk task and the buttons run fine
22:59:47*domonoky hears "girls,girls,girls" on his m200v4 !! :-)
23:00:00kugelthen it has to work, I suppose
23:00:06domonokysvn-bootloader + patched build was the key ! ..
23:00:15amiconnSome targets use (part of) the button reading in an isr, avoiding polling this way. For a wheel, using an isr is practically mandatory in order not to miss steps
23:00:46kugelthe wheel isn't found yet, so I don't mess with it
23:01:08kugelI'm just trying to get the 4 directional buttons, select and power to work togetjer
23:02:52kugelamiconn: I see, I need to pay more attention on interupts.
23:03:07kugelI never worked with interrupts, I doubt I can do it without help/hints
23:04:41 Quit petur ("Zzzz")
23:05:28 Quit domonoky (Read error: 104 (Connection reset by peer))
23:11:18 Join safetydan [0] (n=deverton@rockbox/developer/safetydan)
23:13:21Unhelpfulhrm. down to 296B gain on recorderv2
23:15:52 Quit linuxstb (Remote closed the connection)
23:18:42 Quit tyfoo ("Carpe diem")
23:18:58 Join linuxstb [0] (n=linuxstb@rockbox/developer/linuxstb)
23:26:48kugelamiconn: Can you help me a bit with the isr? I enabled interrupts for my GPIO ports
23:26:59kugelin button_init_device
23:27:23kugelNow I need to request an interrupt in button_read_device, correct?
23:28:20 Quit linuxstb (Read error: 104 (Connection reset by peer))
23:30:07 Quit Jaykay (Read error: 110 (Connection timed out))
23:31:25 Quit lasser (Read error: 110 (Connection timed out))
23:34:09 Quit stripwax (Read error: 104 (Connection reset by peer))
23:35:45 Join rotator [0] (n=e@rockbox/developer/rotator)
23:39:25 Join gregorovius [0] (
23:40:25 Quit n1s ()
23:48:40kugelJdGordon: ping
23:48:48 Quit EspeonEefi ("さよなら")
23:49:19 Quit gregorovius (Remote closed the connection)
23:50:05 Join linuxstb [0] (n=linuxstb@rockbox/developer/linuxstb)
23:50:16kugelsaratoga, JdGordon: (for the logs) We need to mess with interrupts to get the button work. I fount out, that the buttons aren't detectable using the button tick task without interrupts (i.e. using the tick task in the bootloader breaks the buttons there as well)
23:50:18 Part robin0800 ("Konversation terminated!")
23:54:39***Saving seen data "./dancer.seen"
23:56:43 Quit bertrik (Remote closed the connection)

Previous day | Next day