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 2008-11-24

00:00:40 Quit bluebrother ("leaving")
00:00:47BigBambiI'd give it a little longer - I've never done it myself so I don't know what to expect
00:01:35 Join frostbyte [0] (n=48237b10@gateway/web/cgi-irc/
00:01:42BigBambiGive me a moment and I'll find my e200 and try it having some problems with my e250
00:02:00BigBambigo ahead
00:02:02frostbyterockbox works great
00:02:11frostbyteeverything works as its supposed to
00:02:14frostbyteexcept my music
00:02:21bluefoxxthere we go
00:02:27frostbytei just purchased a microsd hc 8gb
00:02:29bluefoxxworks fine now :3
00:02:33BigBambifrostbyte: please put sentances on one line
00:02:37BigBambibluefoxx: no problems
00:02:54BigBambifrostbyte: everything except music is a pretty bg except :)
00:03:08 Quit tyfoo ("Carpe diem") after i insert the microsd hc 8gb and it "updates" its database, all my music is a high pitch noise
00:03:36BigBambiRockbox updates or the OF updates?
00:03:50frostbytewhat exactly do i need to do
00:03:56BigBambiWhat updates its database
00:04:16frostbyterockbox build
00:04:28BigBambiHow are you updating the database?
00:04:34BigBambiAnd what version of Rockbox?
00:04:44BigBambiAnd what format music, and was it bought or not?
00:05:00 Join robin0800 [0] (
00:05:00 Quit DerDome ("Leaving.")
00:05:01bluefoxxtips...don't try opening anything and everything in text editor, don't delete database.idx files while playback is going, and if your me, never go to shut down without making sure it turns off...
00:05:10frostbyteversion r19190-081123
00:05:11BigBambiAnd does the music play OK if you go via the filetree and not the database?
00:05:12bertrikI think there's something weird in button-target.h for clip: it #defines BUTTON_VOL_UP but later uses BUTTON_VOLUP (without the extra underscore)
00:05:25frostbytemusic is mp3
00:05:31bluefoxxfirst one gives divide by zero, second one causes a panic, third tends to not shut down but halt with panic: event not found
00:05:33frostbyteit was purchased
00:05:43BigBambiSo it has DRM?
00:05:54BigBambiIn which case Rockbox does not support DRM infected files
00:06:11frostbyteare you talking to me or bluefoxx?
00:06:11BigBambii.e. purchased from where?
00:06:18bluefoxxdrm doesnt playback...go pirate your tunes like everyone else :P [note: i am being SARCASTIC!]
00:06:33BigBambibluefoxx: Even so, please don't
00:07:47frostbytei bought it from best buy
00:07:47n1sbluefoxx: those sound like bug, i think the text editor one is in the tracker already but could you please file the other two, i don't understand your last issue: "never go to shut down without making sure it turns off" what happens when you try to shut down?
00:07:50frostbyteperfectly new
00:08:00BigBambiNo, where did you buy the music from?
00:08:07bluefoxxpanic: event not found
00:08:38bluefoxxand the divide by zero is my own stupidity, i wondered what would happen if i opened matrix.rock in text editor
00:09:03frostbyteok, only some of the songs work. Is there a way to reupdate the database and get ALL the songs to work. Another thing, please put "for frostbyte, or for blue" before your sentences....tnks
00:09:13n1sbluefoxx: you just try to shut down and rockbox panics?
00:09:28BigBambifrostbyte: for blue would be confusing
00:09:29 Join SirFastolfe [0] (n=446c60f3@gateway/web/cgi-irc/
00:09:38BigBambifrostbyte: It would be better to use full nicks
00:09:45frostbytewhats full nicks?
00:09:51bluefoxxn1s; sometimes...
00:10:06BigBambifrostbyte: You asked me too put for blue before a message, nobody is called blue.
00:10:07bluefoxxfrostbyte; type the first part of someones name and hit tab
00:10:09 Quit markun (" ajax IRC Client")
00:10:26BigBambifrostbyte: Anyway, from where did you buy the music that doesn't work
00:10:38n1sbluefoxx: this is on a sansa e200? and what are you doing when you shut down (music playing, database updating, etc)?
00:10:56bluefoxxusually music playback...
00:10:58bertrikok, BUTTON_MAIN never seems to be referenced, so that's why it doesn't give a compilation error
00:11:06bluefoxxthis build is several months old, before viewports
00:11:17*BigBambi slaps bluefoxx
00:11:21bluefoxxi use it cause i modded a theme and am to lazy/busy to recode with viewports
00:11:24frostbytethere "bluefoxx"
00:11:29BigBambiNo reporting bugs on old versions!
00:11:31frostbytenick name for bluefoxx is BLUE
00:11:38n1soh, very likely it has been fixed since then :/
00:11:42BigBambifrostbyte: Do you want help?
00:11:50frostbyteill make it simple
00:12:05bluefoxxfrostbyte; i may have blue on highlight, but htere are others too have the word 'blue' in they're nick...
00:12:12SirFastolfehey I just registered on the rockbox wiki and I was directed to come here to get editing priveleges?
00:12:25BigBambiSirFastolfe: Sure, what is your wiki name
00:12:35amiconnlinuxstb: around?
00:12:44BigBambiSirFastolfe: What are you wanting to edit?
00:13:02SirFastolfesamsung p2, some of the links are out of date and I was gonna update them
00:13:09BigBambiRighto, one mo
00:13:10frostbytehow do i make all my music(some of it is pooched, some of it is bought) to play instead of a high pitched whine, how do i save certain radio stations, and how do i change the size of font in rockbox?
00:13:13frostbytethere ya go
00:13:17 Quit ender` (" All power corrupts, but we need electricity.")
00:14:03BigBambiSirFastolfe: There you go, no spamming now :)
00:14:09SirFastolfety :)
00:14:18 Quit SirFastolfe (Client Quit)
00:14:33BigBambifrostbyte: For the third time, where did you buy/get the music that doesn't work?
00:14:50BigBambifrostbyte: For the radio stations and fonts, read the manual
00:14:57*kugel notices BigBambi always says the same sentence and makes the same smiley after giving someone write permissions for the wiki
00:15:12 Quit n1s ()
00:15:19 Quit bonsaiMonkey (Read error: 113 (No route to host))
00:15:20frostbyteim a genious...i just needed to reinstall rockbox, reformat the mp3 player, and it worked
00:15:32frostbytebtw...can you give me the url for the radio stations and fonts manual
00:15:43BigBambikugel: I don't think that is quite true, I sometimes say Done, no spamming now :)
00:15:44frostbyteafter done, tnks bigBanbi
00:15:54BigBambifrostbyte: It is linked from every page on the Rockbox site
00:16:10kugelBigBambi: yea, now I remember you sometimes say "promise not to spam" too
00:16:17kugelsorry for the confusion ;)
00:16:23BigBambiVariation is the spice of life :)
00:16:42frostbytepeace dude!!!
00:16:45BigBambino probs
00:16:46 Quit frostbyte ("CGI:IRC (EOF)")
00:17:40bluefoxxdoesnt work...screws up and causes a green screen. i have to go, i just missed an appointment...
00:17:44 Quit robin0800 (Remote closed the connection)
00:17:45 Part bluefoxx ("<Rocketbrigand> i read that as "bluefoxx orgasms files"")
00:27:12bertrikwhat happened!?
00:28:34 Join nuonguy [0] (
00:28:35kugelgot jhMikeS's way to work
00:36:41 Quit faemir (Remote closed the connection)
00:39:31 Quit bertrik ("Leaving")
00:41:59 Join BHSPitMonkey [0] (n=stephen@unaffiliated/bhspitmonkey)
00:50:50 Quit culture (Success)
00:55:23kugelmaybe not
00:57:44 Quit MethoS- (Remote closed the connection)
00:58:04 Join MethoS- [0] (
00:59:26 Join jeffdameth [0] (
01:04:28 Quit EspeonEefi (Read error: 60 (Operation timed out))
01:04:53 Join EspeonEefi [0] (i=eefi@STRATTON-TWO-EIGHTY.MIT.EDU)
01:07:46 Quit herrwaldo (Read error: 60 (Operation timed out))
01:11:57 Quit jeffdameth1 (Read error: 113 (No route to host))
01:23:35 Join ajonat [0] (n=ajonat@
01:25:01 Join jeffdameth1 [0] (
01:29:09 Quit jeffdameth (Read error: 110 (Connection timed out))
01:30:04 Quit Nico_P (Remote closed the connection)
01:33:32 Quit Bensawsome (Read error: 54 (Connection reset by peer))
01:34:07 Join Bensawsome [0] (n=Bensawso@unaffiliated/bensawsome)
01:39:37 Quit jhulst (Remote closed the connection)
01:41:15 Quit Thundercloud (Remote closed the connection)
01:43:50 Quit lasser (Read error: 60 (Operation timed out))
01:44:59 Part pixelma
01:45:26 Join pixelma2 [0] (n=marianne@rockbox/staff/pixelma)
01:46:48kugelpondlife, jhMikeS: (for the logs) I put a commit candidate for backlight thread fading up, please have a look
01:47:45 Join fdinel [0] (
01:48:07kugelfdinel: hello my friend :)
01:49:19fdinelkugel: hey mate :) will you be around in about an hour? I'll be back then
01:49:43fdinelgotta bath the kids ;)
01:49:50kugelfdinel: unfortunately no. Any news? Then I of course will read the logs (I'll do anyway, tbh)
01:50:33***Saving seen data "./dancer.seen"
01:51:01 Quit Zambezi (Read error: 110 (Connection timed out))
01:51:16fdinelwell nothing much on the fuze buttons side, but hey, anything is already found, one will need DBOP to use buttons on the fuze IMO... and since I got no fuze, you guys will need to better investigate on this one... I can help with suppositions, but nothing much more sadly... :/
01:51:58fdinelwith what I found, there don't seem to be any directly accessible buttons on the gPIOs
01:52:03kugelfdinel: the said thing is, I don't have a real idea of what DBOP is. I searched google but I didn't find answers
01:52:28fdinelwell maybe not that sad, but we'll need one adventurous :/
01:52:44kugelhave you had a look at the diagnosis mode? If yes, I hope you didn't confuse something with all the other tests it offers
01:52:54fdinelit'll be trials'n'errors
01:53:05kugelI'm open to that, no worries
01:53:13fdinelnope I didn't have a look there
01:53:22kugelI saw the main menu today, and I'll give all I can to be able to actually scroll through it
01:53:29fdinelyou could try with somthing that "enables" a wait or not
01:53:33fdinellike we did at first
01:54:22kugelwell, we have lcd and backlight for debugging, I don't think we need to work with delays
01:54:36fdinelI gotta go, maybe we can talk later or else I'll talk alone so you can read the backlogs ;)
01:54:49fdinelsee ya
01:55:22kugelfdinel: I'll definitely read the logs. I'll be up in ~6-7h, maybe then. see you
01:55:48 Quit kugel ("ChatZilla 0.9.84 [Firefox 3.0.4/2008111318]")
01:57:08 Nick JdGordon|zzz is now known as JdGordon (n=jonno@rockbox/developer/JdGordon)
02:04:36 Join kugel [0] (n=chatzill@unaffiliated/kugel)
02:05:46kugelfdinel: that might be important: the lcd (fuze and e200v2) is using DBOP (see ams3525_dbop_init in lcd-e200v2.c and lcd-fuze.c).
02:05:47 Quit MethoS- (Read error: 131 (Connection reset by peer))
02:05:51 Quit kugel (Client Quit)
02:09:16 Quit perrikwp (" ajax IRC Client")
02:12:13 Quit w23 ("Leaving")
02:16:54Unhelpfuladding a "skip_lines" callback that skips N whole lines ahead in the input file. if N == 0, skips to start of next line, or not at all if at start-of-line. i *think* it seems sane, tbh, it will never be called with 0 :)
02:17:04 Quit gevaerts (Nick collision from services.)
02:17:16 Join gevaerts [0] (n=fg@rockbox/developer/gevaerts)
02:29:47 Quit jeffdameth1 (Connection timed out)
02:31:29 Quit tessarakt ("Client exiting")
02:31:35fdinelkugel: indeed DBOP is also used for LCD on the clip and m200... Maybe I'll have to better look at the OF to ensure that what I saw was really for buttons and wheel...
02:32:06fdinelkugel: But you said you already saw rockbox running without being able to navigate on the fuze, does that mean you already have a "failsafe" combination to revert to the OF, using for example the hold button or USB? if that's the case, then nothing to worry about bricking your fuze, you just need to configure DBOP and try to read from it using settings from the OF
02:32:38fdinelkugel: And then simply display what you read on the LCD so that you know when a button is pressed is it changes the DBOP readings...
02:33:09fdinelkugel: anyway, hope that help, let's try to talk again "live" so we can try some stuff when you'll be back...
02:33:47 Join Dhraakellian [0] (
02:34:11 Quit jhMikeS (Nick collision from services.)
02:34:17 Join jhMikeS [50] (n=jethead7@rockbox/developer/jhMikeS)
02:35:09 Join mc2739_ [0] (
02:36:07 Quit mc2739 (Nick collision from services.)
02:36:21 Nick mc2739_ is now known as mc2739 (
02:45:36 Quit aarcane ("Leaving")
03:09:38Dhraakelliangiven current development trends, what's the rough ETA of fully functional/usable Rockbox on the Sansa V2 targets? (Weeks, Months, some vague time next year or beyond?)
03:10:11 Join saratoga [0] (n=439f411b@gateway/web/cgi-irc/
03:10:13Dhraakellian(or "no idea. isn't there an FAQ entry about this?")
03:10:34advcomp2019Dhraakellian, there is no etas
03:13:50 Join perrikwp [0] (i=18ac0c41@gateway/web/ajax/
03:15:22saratogai think the FAQ just says not to ask
03:15:56Dhraakelliansaratoga: I was looking for that
03:16:48 Quit saratoga ("CGI:IRC (EOF)")
03:17:19Dhraakellianrephrased: What's the current progress? I see that some things are listed as already working. How are others, like sound, coming along?
03:17:54LloreanDhraakellian: They aren't donel.
03:18:08LloreanSeriously, where they're at is no indication of how long it'll take for them to be finished.
03:18:14LloreanSo it's a binary state, "not done" and "working"
03:20:03 Part toffe82
03:20:19Dhraakellianwhich is why I was asking based on current activity, but I suppose that doesn't take into account unexpected roadblocks, changes in interest, differing expertises in different areas that need to be covered.
03:20:32Dhraakellianquestion withdrawn, then
03:20:46 Nick fxb is now known as fxb__ (
03:27:14 Join Darksair [0] (n=user@
03:30:00jhMikeSkugel: pong (for the logs)
03:31:32 Join Zambezi [0] (
03:40:54 Join Strife89 [0] (n=michael@
03:50:20 Join bluefoxx [0] (
03:50:34***Saving seen data "./dancer.seen"
03:56:40Unhelpfulon the grayscale targets, would LCD_WIDTH bytes on the stack be a bad thing to do, or not? i doubt those targets have tons of memory to spare, but i don't know how much that balances against smaller screen...
03:58:58 Quit Strife89 ("Bed.")
03:59:48LloreanUnhelpful: Well, there's only a couple that are of interest (the H120 and the M5). I'm not sure if the M5 is grayscale or pure mono.
03:59:48 Quit fdinel ("Miranda IM! Smaller, Faster, Easier.")
03:59:57LloreanBut I'm surprised with have AA enabled in <=2bpp targets
04:00:53LloreanAnd the iPods, sorry
04:00:57LloreanWhich all have 32
04:04:49Unhelpfulit's 1/24 as much buffer, per pixel, as the color targets use... but it's still a pretty hefty chunk. it's not hard to switch between a static and an on-stack buffer, i could could both, and ask somebody with hardware to test, after i'm sure the stuff works in sim.
04:05:23LloreanUnhelpful: Even the smallest mem mono target, relative to the smallest mem color target, is 1/8 of the memory.
04:10:51 Join jhulst [0] (n=jhulst@unaffiliated/jhulst)
04:15:14 Quit mc2739 ()
04:15:42 Join blkhawk- [0] (
04:19:57Unhelpfuli think it can be done w/ four bytes on the stack. collect enough to write out one byte worth of pixels, write it to however many lines, then do the next block. and yes, anybody doing AA on grayscale is probably nuts :/
04:28:01 Join saratoga [0] (n=98039ac5@gateway/web/cgi-irc/
04:28:12JdGordonhas anyone got a mpeg on their dap they can test mpegplayer with? check if there is audio if mpegplayer is started whie playback is paused?
04:28:14saratogais there a penalty for doing lots of small reads?
04:28:33JdGordoninstead of 1 big read?
04:28:35 Nick Darksair is now known as Awaysair (n=user@
04:29:18saratogaJdgordon: works fine
04:29:27saratogapaused mp3, started mpegplayer, sound works
04:29:38JdGordonok thanks, which target?
04:29:45saratogaJdGordon: well a lot of 100 bytes reads verses a smaller number of 1kb reads
04:31:17JdGordonif its sequencial I would guess reading SECTOR_SIZE blocks instead of 100 bytes might be better? although I dont know how fat puts files on the actual disk
04:31:36JdGordonbut there probably isnt much difference
04:31:53 Quit blkhawk (Read error: 110 (Connection timed out))
04:32:06saratogasince we're going through all this hassel to efficiently resize anyway, maybe thats worth thinking about
04:32:31saratogathough i guess saving memory is probably more important then saving cpu time
04:32:40 Nick blkhawk- is now known as blkhawk (
04:33:00JdGordonnot if its a local temp variable wasting the memory
04:33:38 Quit Bensawsome ("The awsome is gone :(")
04:34:00saratogatrue, if the sector size is small enough to fit one on the stack, then that'd make sense
04:34:41JdGordonthe main thread is only using 51% on my h300... so plenty of wiggle room
04:35:57saratogaif you read the first half of a sector, and then read the second half in sequence, does it have to seek twice?
04:36:18saratogaor does the disk's own buffer just retransmit the sector
04:36:24saratogaprobably the latter i'd guess
04:36:39JdGordonmaybe... maybe not?
04:39:01saratogainternet says the toshiba drives have a 2MB DRAM buffer that stores recent sectors
04:39:09saratogaso i guess going one sector at a time doesn't really matter
04:39:26saratogaaside maybe from some CPU time spent in the ATA driver
04:40:23*Unhelpful heard "resize" mentioned, and wonders if he needs to be paying attention
04:42:42saratogaUnhelpful: you were talking about the read size before?
04:44:16Unhelpfulsaratoga: only in terms of reducing a static buffer (actually, it's on-stack, now, and 24B wide). the code that fills it, which i didn't change substantially, was already doing 1*<stored pixel size> reads
04:44:56saratogaI was just wondering aloud if theres some advantage to requesting a larger number of pixels at a time
04:45:07saratogaso as to avoid a lot of calls to the ata driver
04:46:49Unhelpfulsaratoga: a good question. there are more than a few places where speed-vs-memory tradeoffs are available - i removed a line of scale buffer, at the cost of two extra multiplies per output pixel in the linear scaler.
04:47:18saratogaUnhelpful: how small have you got the patch?
04:47:36Unhelpfulin terms of bin/memory size?
04:48:07 Join webguest15 [0] (n=dee1e309@gateway/web/cgi-irc/
04:49:21Unhelpfullast build for beast, results (bin/ram size) were: 522124/1632116 for vanilla svn, 527224/1647252 for "old" resize-on-load, 526532/1642292 for "new" resize-on-load
04:51:31 Quit webguest15 (Client Quit)
04:59:31 Join miepchen^schlaf_ [0] (
05:01:13 Nick Awaysair is now known as Darksair (n=user@
05:04:45saratogaso you've saved about 5.5KB so far, not bad
05:04:57saratogaalso, i've committed the first part of the Sansa power savings patch
05:07:19Unhelpfulsaratoga: i think that's about the size of the old-scaler line cache, isn't it? the binary size didn't go down so much, but if i understand correctly, what i'm concerned about is bin+ram, right?
05:10:28 Quit Darksair (Remote closed the connection)
05:11:11saratogaI'm not really sure how the ramsize is calculated, but yes I think so
05:11:18 Join Darksair [0] (n=user@
05:11:27saratogai suppose you could verify this by putting some buffer on the stack and seeing if the ramsize goes down accordingly
05:12:47 Quit bluefoxx (Read error: 104 (Connection reset by peer))
05:13:00 Join bluefoxx [0] (
05:13:00 Quit saratoga ("CGI:IRC (EOF)")
05:14:11 Quit miepchen^schlaf (Read error: 110 (Connection timed out))
05:25:19 Quit scorche (Read error: 104 (Connection reset by peer))
05:25:30 Join thegeek_ [0] (
05:25:43 Join Llorean1 [0] (
05:26:00 Quit Llorean (Nick collision from services.)
05:26:04 Nick Llorean1 is now known as Llorean (
05:26:10 Join scorche [0] (i=Blah@rockbox/administrator/scorche)
05:32:15 Quit thegeek (Read error: 60 (Operation timed out))
05:34:39 Quit reacocard (Read error: 110 (Connection timed out))
05:39:25 Quit Horscht ("")
05:41:11JdGordonim doing a bit of cleanup for the e200r installer and I want to remove the e200r target from configure (its not used)... Should I make the e200 line "e200/e200r" or just leave it as e200 ?
05:42:40JdGordonarg, ignore that... different bootloader :/
05:43:37 Quit bluefoxx ("bye for now, returning later")
05:44:43 Join fdemocracy [0] (i=maha@gateway/gpg-tor/key-0xB9002659)
05:44:50 Join daurnimator [0] (
05:45:54 Quit Darksair ("Everything that has a beginning has an end.")
05:46:39 Join Darksair [0] (n=user@
05:50:41***Saving seen data "./dancer.seen"
05:51:56 Quit daurnimator (Remote closed the connection)
05:59:10 Quit Darksair ("Do you hear that? This is the sound of inevitability. This is the sound of your death, Mr. Anderson.")
05:59:51 Join Darksair [0] (n=user@
06:19:12 Quit Seed ("cu, Andre")
06:20:42 Join |AhIoRoS|_ [0] (n=ahioros@
06:51:10 Quit Darksair ("To Arch or Gentoo? That is the question...")
06:56:27 Join Darksair [0] (n=user@
07:14:32 Quit XavierGr ()
07:28:09 Join stoffel_ [0] (
07:28:18Dhraakellianhow likely is data corruption if one builds rockbox for the e200v1 with USB transfer enabled?
07:29:11JdGordonsome people get it and some dont so it depends....
07:29:15Dhraakellian...and actually uses it to transfer files to one's shiny 8GB microSDHC
07:29:28JdGordonyou dont have a sdhc reader?
07:29:34LloreanJdGordon: For the e200, it's gonna happen eventually, if you use it long enough
07:29:52LloreanSince it's the SD bug, not the USB bug that's the problem, so it doesn't really matter what your PC does, etc.
07:29:55Dhraakellianhave an e200 with a working spring in the microSD slot but not a working latch to hold the card in
07:30:49DhraakellianI'm using a dab of hot glue (at the suggeston of someone in -community) to hold in a 2GB microSD
07:31:28DhraakellianCan't do that with the 8GB card because I then wouldn't be able to easily take it out and use the mobilemate adapter to transfer to/from it
07:33:15JdGordonbit of sticky tape and paper should be less permananet?
07:34:54Dhraakellianthat'd make the side all sticky. the idea was that this would just flake off under the assault of a thumbnail
07:35:02Dhraakellianor something along those lines
07:35:42Dhraakellianso, just out of curiosity, you said the problem is in the microSD driver? Does that mean that USB transfer to/from the internal storage would be safe?
07:35:46Dhraakellianor am I misunderstanding?
07:37:35JdGordonnot microSD... the SD driver handles both the internal storage and the microSD card
07:37:52*JdGordon isnt sure which it effects more though
07:38:24Dhraakellianis it likely to be a problem with the SansaV2 family?
07:38:56Dhraakellian(although the SansaV2 family knows about microSDHC in the OF, so I suppose it wouldn't be so much of an issue)
07:41:44 Join Dieterbe_ [0] (
07:47:41 Quit Dieterbe (Read error: 110 (Connection timed out))
07:50:45***Saving seen data "./dancer.seen"
07:52:17 Join kkurbjun [0] (
07:58:11LloreanJdGordon: Both, really. Copying files to/from internal storage from/to a MicroSD card also triggers the corruption, last I heard.
08:00:30Dhraakellianhuh. So the 'cheat' of using the OF to transfer to internal storage and then cut&pasting to the microSDHC is just as vulnerable?
08:03:19 Quit miepchen^schlaf_ (Client Quit)
08:08:24JdGordonanyone got a e200 where the radio seems to disappear from the menu?
08:08:41 Join kugel_ [0] (n=chatzill@
08:08:45 Nick kugel_ is now known as kugel (n=chatzill@
08:09:05pixelma2Llorean: the M5's display is greyscale (160x128), similar to the H100's but physically smaller/smaller pixeels. Maybe you confused it with the monochrome red on black M:Robe100
08:09:08LloreanJdGordon: They seem just uncommon enough that people who have the problem only ever show up once.
08:09:30pixelma2JdGordon: happens on my c200 occasionally but very very unpredictable
08:09:32Lloreanpixelma2: I just wasn't sure if it was mono or grayscale.
08:10:16JdGordonpixelma2: :/ very very unpredictable sounds like fun... ok i might get you to test a patch later
08:11:26pixelma2yes it is fun, and I don't know when I could give you results....
08:14:25DhraakellianJdGordon: I had the problem on my first e260 (which got RMA'd due to the headphone jack), but I haven't really noticed one way or the other with the replacement
08:15:13DhraakellianI'll take a look when this current transfer is done
08:15:42*Dhraakellian will even do it without headphones
08:18:13 Join reacocard [0] (n=reacocar@WL-135.CINE.HMC.Edu)
08:20:18 Join Rob2223 [0] (
08:21:17Dhraakellianit found the radio
08:21:28Dhraakellianin the basement with no headphone cable attached
08:26:39 Join Bagderr [241] (n=daniel@rockbox/developer/bagder)
08:27:03 Nick Bagderr is now known as B4gder (n=daniel@rockbox/developer/bagder)
08:28:41 Quit BigBambi (Read error: 113 (No route to host))
08:32:00 Join lasser [0] (
08:35:06kugelamiconn assumes ram timing to be the reason for the sd bug. Maybe that's also the reason for the missing radio?
08:38:20 Quit Rob2222 (Read error: 110 (Connection timed out))
08:38:45Dhraakellianso would it have anything whatsoever to do with reception quality on bootup?
08:41:08 Quit Zambezi (Read error: 110 (Connection timed out))
08:41:08 Join Zagor [0] (n=bjorn@rockbox/developer/Zagor)
08:44:24 Join ender` [0] (
08:45:55 Quit JdGordon (Remote closed the connection)
08:47:35 Join JdGordon [0] (
08:47:36amiconnkugel: Btw, I never said that RAM timing might be the cause of the SD bug. That's most likely a missing SD controller timing setting. I suspect RAM timing to be the cause of the USB problems on PP.
08:48:08 Join daurnimator [0] (n=fake@unaffiliated/daurnimator)
08:48:56kugeloh, my bad
08:49:12*kugel confused that
09:09:14 Join petur [50] (n=petur@rockbox/developer/petur)
09:17:41linuxstbamiconn: You pinged last night?
09:20:10*linuxstb sees the test codec bug-fix and slaps forehead
09:30:26 Quit Darksair ("People who are zhuangbility want to show their niubility but only reflect their shability.")
09:31:02 Join Darksair [0] (n=user@
09:33:32 Join Nico_P [50] (n=nicolas@rockbox/developer/NicoP)
09:39:04 Join Thundercloud [0] (
09:43:04gevaertskugel: I was the one who speculated (without any real information) that RAM timings might be involved in SD as well
09:43:29 Join LinusN [0] (n=linus@rockbox/developer/LinusN)
09:43:56 Join einhirn [0] (
09:47:45kugelgevaerts: ah ok
09:47:59kugelso I wasn't completely off ;)
09:48:50kugellinuxstb: I managed to boot into the main menu yesterday. The color was slightly green'ish, the driver might need calibration or something like that
09:49:32kugelLinusN: ping
09:50:46***Saving seen data "./dancer.seen"
09:53:05kugelLinusN: I heard you played with backlight fading on h300 some time ago, so I might get your attention for FS #680
09:53:09kugelFS #6800
09:56:01 Quit jhulst (Read error: 113 (No route to host))
09:58:20LinusNhmm, what's the purpose of USE_BACKLIGHT_THREAD_FADING?
10:00:27kugelit enables the fading algorithm. The backlight works without as well (no fading of course), so I introduced the #define. The backlight code is different for targets which don't use the fading, i.e. I need some #ifdef'ing in backlight.c
10:00:54linuxstbkugel: So the colour is correct in the bootloader, but wrong in main Rockbox?
10:01:51kugellinuxstb: I can hardly say when I only see the rockbox logo and little peaces of small text in the bootloader. But yes, it looks correct
10:03:11linuxstbkugel: You should be able to easily test - e.g. draw some rectangles of different colours in the bootloader. If that's the case, then you need to look for what different lcd code is run in the main build, compared to the bootloader.
10:03:51kugelthe colors aren't really wrong in the main rockbox, they have a slight green-tinted
10:04:11kugels/have a/are
10:04:18linuxstbHaving a green tint sounds wrong to me...
10:05:04LinusNkugel: ok, i see now, you can either have hardware fading, software fading, or no fading (as in the sims)
10:06:12LinusNand remove the #if 0's
10:07:06kugeli left #if 0's? not on purpose, sorry. And yea, I thought of that renaming too. I didn't introduce the name for that symbol if that counts as excuse
10:16:48 Join Zambezi [0] (
10:22:27kugelLinusN: are you going to try and test the patch?
10:23:05LinusNwhat targets do we need further testing on?
10:27:32 Quit Thundercloud (Remote closed the connection)
10:30:21kugelLinusN: well, I've made some changes. It would be nice if a test would happen on all affected targets.
10:31:01LinusNok, i don't have many targets with me now, but i can test on x5 at least
10:31:37kugelx5 and h300 should behave the same, as well as e200 and c200, d2 at least very similar to x5/h300
10:31:48kugelI tested e200 myself
10:34:10*petur could test h300 and maybe H10 tonight
10:34:45petur(don't know if H10 has brightness control)
10:36:56 Quit BHSPitMonkey (Remote closed the connection)
10:38:10*linuxstb wonders what the HAVE_BACKLIGHT_SET_FADING define in config-gigabeat.h means - it doesn't appear to be used anywhere.
10:38:55kugellinuxstb: no idea
10:45:12 Quit JdGordon (Read error: 104 (Connection reset by peer))
10:46:11LinusN}' token
10:46:11LinusNmake: *** [/cygdrive/c/linus/rockbox/x5/firmware/backlight.o] Error 1
10:46:59Mode"#rockbox +o logbot " by ChanServ (ChanServ@services.)
10:47:52kugeleh, forgot a ; :S
10:48:24 Join JdGordon [0] (n=jonno@rockbox/developer/JdGordon)
10:48:56kugelLinusN: where are #if 0's?
10:50:23LinusNin backlight.c, around backlight_fading_state = NOT_FADING;
10:51:01kugelin the latest version?
10:51:38kugelah lol! I didn't upload v14
10:54:08 Join at0m|c [0] (
10:54:39 Part at0m|c ("more has been done with less")
10:56:15kugelLinusN: now :/
10:56:36 Join culture [0] (
10:56:40JdGordonanyone have ideas about how to neaten up the radio code? its currently split up (mostly) into tuner_* for the actual driver and radio_* for apps, but those radio_ functions should maybe be moved into firmware (or at least out of recorder/radio.c)
10:59:14 Quit ajonat ()
10:59:30preglowwell, if it makes sense for them to be in firmware/
11:00:00preglowi don't really know how big a fan i am of the whole firmware/ apps/ thing anymore...
11:02:21 Join _lifeless [0] (n=lifeless@
11:03:56JdGordonwhy not? its good to keep drivers seperate from the rest
11:09:49 Quit kugel (Remote closed the connection)
11:09:54 Quit JdGordon ()
11:10:33 Join JdGordon [0] (n=jonno@rockbox/developer/JdGordon)
11:12:32 Quit Darksair ("Emacs = ESC-Meta-Alt-Ctrl-Shift")
11:14:43JdGordonLinusN: any objections to merging s1a0903x01.c and fmradio.c?
11:16:30JdGordonor move fmradio.c to something more meaningful in the sh target folder? (I want to free up fmradio.c)
11:16:45LinusNi have no objections
11:17:02LinusNmove it to the target tree
11:18:22 Quit JdGordon (Remote closed the connection)
11:19:23 Join JdGordon [0] (n=jonno@rockbox/developer/JdGordon)
11:20:48 Nick fxb__ is now known as fxb (
11:25:09JdGordonLinusN: and rename it to what?
11:25:56amiconnI'd suggest firmware/target/sh/archos/fmradio_serial.c
11:26:48JdGordonthat works
11:31:36 Quit JdGordon (Remote closed the connection)
11:32:08 Join JdGordon [0] (n=jonno@rockbox/developer/JdGordon)
11:35:02 Quit culture (Connection timed out)
11:40:14 Join robin0800 [0] (
11:40:46Zagorwow. the win32 corss-mingw simbuild compiled without a single complaint
11:41:29JdGordonbut does it work?
11:41:39Zagornow how do I tell wine to look in /usr/local/mingw32-sdl/bin for the SDL.dll?
11:42:05gevaertsCopy it to the directory that has the .exe
11:42:17Zagorgevaerts: ah, clever
11:42:45Zagoryup, works
11:44:30Zagorplugins too
11:45:41Zagorand mp3 decoding
11:46:34rasherZagor: Probably doesn't work on Windows until you strip the binaries though
11:46:39 Join Darksair [0] (n=user@
11:46:49Zagoroh, is wine more forgiving about that?
11:47:03rasherI think it might be
11:47:17B4gderwasn't that a plain size issue?
11:47:51rasherI never fully understood it, but stripping fixes it
11:48:15ZagorI have terminal server access to a windows box, I'll give it a try there
11:48:20 Quit n17ikh|Lappy ()
11:50:48***Saving seen data "./dancer.seen"
11:52:30 Join kugel [0] (n=chatzill@unaffiliated/kugel)
11:52:56Zagorbrickmania works on native windows too
11:53:48rasherMaybe things have changed since I first tried
11:55:14 Nick fdemocracy is now known as maha (i=maha@gateway/gpg-tor/key-0xB9002659)
11:55:32Zagormp3 crashes, but that is probably related to the error printed when starting: "Unable to open audio: wanveOutOpen(): There is no driver installed on your system."
11:56:57rasherThat certainly sounds like an SDL issue
11:57:08Zagorhaha, the fire plugin over terminal server strains the network a bit...
11:57:21Zagorumm, how do I exit that?
11:58:19Zagorbtw why don't we display the background by default?
11:58:46kugelI wouldn't like it if it did
11:58:51rasherI think it would make sense to reverse the logic
11:58:56JdGordonbecause the bg is pointless and annoying after a while
11:59:20rasherDepends how you're using the sim, I guess
11:59:45rasherI noticed simulator builds on misticriver also defaulted to showing the background
12:00:09ZagorI think it would be very nice to greet first-time sim users with the background. I can agree it's not terribly useful in the long run.
12:01:18Zagorperhaps it's time we renamed the "archos" dir...
12:03:34rasherIt really doesn't make any sort of sense. Something like "disk" (or a better, generic, term) would be more useful
12:04:47JdGordon"target" ?
12:07:06 Quit nuonguy ("This computer has gone to sleep")
12:07:50 Join __lifeless [0] (n=lifeless@
12:08:58kugelreally, how often has this topic already been brought up?
12:09:09kugelthis isn't gonna change ever anyway
12:09:16 Quit _lifeless (Read error: 113 (No route to host))
12:10:12Zagorkugel: haha
12:10:26JdGordonit will change if/when the sim gets multivolume support
12:10:50kugelwhy? archos, archos_ archos__ etc :)
12:11:02Zagoror if/when mr someone is sufficiently annoyed by it
12:12:18kugelboth relatively unlikely to happen soon afaics
12:12:33JdGordonpatches are always welcome :p
12:12:52*kugel points to gevaerts
12:13:35JdGordonspeaking of which... Zagor, is there any way to force users to say which version in bug reports instead of havingthe only choice as "current build" ?
12:14:13JdGordonand regardless of that, "Version 3.0" should be added to that list
12:14:49Zagoradding "Version 3.0" and "Daily build (write which)" as options seems reasonable
12:16:57kugelJdGordon: btw, re status of ams sansas: I booted quite successfully into rockbox with the patch funman pastebin'ed
12:17:26kugelwhich he calls "that enables reliable read and write support with dma"
12:18:13kugelJdGordon: I'd really like to know if the buttons on the e200v2 work (the scrollwheel isn't supposed to work iiuc)
12:18:38JdGordonok, my tree is mangled atm, can you dcc me a build?
12:18:44kugelmy target is the fuze in case you don't remember
12:19:01JdGordonyou can build a e200 though....
12:19:24kugelsure, what's dcc? I'd just compiled one for you
12:19:44JdGordonsend through irc
12:19:45kugelDistributed Checksum Clearinghouse isn't it I guess
12:20:27kugeldo you have svn bootloader at least?
12:20:49JdGordonfrom a while ago... how recent is needed?
12:21:00kugelnot sure
12:21:07kugelwe'll see
12:31:58kugelg2g, will read the logs
12:32:03kugelJdGordon: ^
12:32:17 Quit kugel ("ChatZilla 0.9.84 [Firefox 3.0.4/2008111318]")
12:32:30 Join TiMiDo [0] (
12:32:32TiMiDoanyone in here?
12:32:47 Join pondlife [50] (n=Steve@rockbox/developer/pondlife)
12:32:51*scorche wishes TiMiDo would stop asking that each time he enters
12:33:14TiMiDoi have a question I'm trying to installed.
12:34:10pondlifeJdGordon, Zagor: How about if Flyspray bug reports had a compulsory "revision" field that was blank by default - to force the user to enter the r number..?
12:34:18TiMiDorockbox on my ipod But i was getting this error Starting bootloader installation Searching for ipods No ipods found.
12:34:32linuxstbTiMiDo: Which ipod do you have?
12:34:34TiMiDoand i actually point it to /media/IPOD and is not working though? any ideas
12:34:43TiMiDolinuxstb, nano 4th generation 8gb
12:34:53linuxstbRockbox won't work on that.
12:35:00TiMiDooh for real.
12:35:05TiMiDoi though it did since it's here.
12:35:21scorche"it's here"?
12:35:25TiMiDolinuxstb, do you know any GUI applications to put songs on the ipod besires gtkpod?
12:35:40scorcheTiMiDo: not really on-topic..
12:36:18JdGordonkugel: nope... white screen... I got to the menu (somehow magically) with svn yesterday though... but after a few atempts and buttons didnt work
12:37:17TiMiDolinuxstb, would. it be supported it later on?
12:37:39linuxstbTiMiDo: Impossible to say, but no-one is currently working on it.
12:38:04TiMiDooh ic
12:39:01 Quit Darksair (Read error: 104 (Connection reset by peer))
12:39:12TiMiDoi though they we're lol
12:39:19TiMiDosince the ipodlinux project is deac
12:39:42scorchewhat does that have to do with getting rockbox on your ipod?
12:39:50TiMiDonothing really
12:40:22 Join Darksair [0] (n=user@
12:40:31 Join bmbl [0] (n=Miranda@unaffiliated/bmbl)
12:42:01JdGordonpondlife: yeah, forcing them to write the rev number would be best, but I wonder if flyspray allows that?
12:42:28pondlifeA compulsory, numeric field.... no idea
12:49:35JdGordonanyone interested in the radio code, comments on FS #9573 would be appreciated
12:51:37 Quit fyre^OS ("( :: NoNameScript 4.22 :: )")
12:51:43 Join moos [0] (
13:04:01Zagorpondlife: it would be great. but it would require major flyspray patching, including the database schema. no fun. :(
13:04:14pondlifeOK, just wondered
13:04:42*Llorean suspects a lot of people would just put in "3.0" or similar anyway
13:11:23 Nick Darksair is now known as Awaysair (n=user@
13:11:41 Join tyfoo [0] (
13:14:42 Part maha
13:15:22 Join domonoky [0] (
13:21:38 Join kugel [0] (n=chatzill@unaffiliated/kugel)
13:22:48kugelJdGordon: white screen? weird. But interesting that the buttons didn't work for you
13:24:13 Quit kugel (Remote closed the connection)
13:25:05 Quit __lifeless (Read error: 113 (No route to host))
13:26:01 Quit bmbl ("Woah!")
13:26:27 Join kugel [0] (n=chatzill@unaffiliated/kugel)
13:29:12 Join mofux [0] (
13:30:29 Quit stoffel_ ("Lost terminal")
13:32:17 Quit kugel (Nick collision from services.)
13:32:20 Join kugel_ [0] (n=chatzill@
13:32:27 Nick kugel_ is now known as kugel (n=chatzill@
13:34:08 Join omgun [0] (n=79b45a2d@gateway/web/cgi-irc/
13:34:44kugelthe todays log page is extremly slow, isn't it?
13:36:12 Nick Awaysair is now known as Darksair (n=user@
13:37:48J-23is it needed to reformat an e200v2 before trying to run Rockbox on it?
13:38:18J-23as I see before doing it only bootloader works, and after formatting player Rockbox booted properly last time I tried
13:39:16domonokyJ-23: i think the problem is still the access over 1 GB...
13:39:40*J-23 patches.
13:40:55kugeldomonoky: you noticed that JdGordon said the buttons aren't working on the e200v2?
13:40:59Zagorkugel: slow?
13:41:21kugelZagor: took some minutes to see the full today's log
13:41:30domonokykugel: no, and i have a m200v4 and my buttons are working, so i dont care :-)
13:41:40Zagorkugel: which browser?
13:41:57 Join funman [0] (
13:41:59kugelff3.0.4, ubuntu 32bit
13:42:39Zagorodd. it takes a couple of seconds for me.
13:42:47kugeldomonoky: please some more brotherliness for the other ams sansas :(
13:43:19kugelZagor: yea, now it was faster again again
13:43:40JdGordondomonoky: haha
13:43:42kugelbut it wasn't the first time I need to wait a couple of minutes for the full log
13:43:55omgunsamsung T9 rockbox is developing?
13:45:55kugelfunman: do you at least care a little that e200v2 apparently don't work?
13:46:11funmanthe i2sout hardware in as3525 supports both 16-bits and 18-bits samples. I wonder why this extra 2 bits precision ?
13:46:32kugelthe buttons
13:46:45funmankugel: I find it a bit sad that nobody dedicates time to the e200v2, but I can't help on this.
13:47:17funmanI just can say last time michael chicoine tested the buttons they were working (except the one which weren't found yet)
13:47:19domonokythe problem is, that people working on the ams sansas, dont have a e200v2...
13:47:42JdGordonyeah yeah, I'm here... just really cant find motivation for it atm :p
13:48:00*JdGordon also doesnt have the datasheets if someone wants to pas them on
13:48:10funmanJdGordon: you can ask B4gder for it
13:48:44funmansee r19035 / FS #9532 for e200v2 buttons (committed less than 3 weeks ago)
13:49:12 Join fograven [0] (
13:49:39*JdGordon did that commit :p
13:50:33 Join Nibbl [0] (
13:50:49***Saving seen data "./dancer.seen"
13:51:01kugelwell, I guess you the guy who did the driver based it on the SansaV2HardwareMapping page
13:51:03funmanby the way I'm not sure restoring the gpio direction registers is needed at all, they should be set in a button_device_init() just like for the Clip. After checking in the wiki page that these pins are not used for something else than buttons
13:51:23fogravenwill rockbox work for Ipod Nano 3G
13:51:40kugelfograven: nope
13:51:56Lloreanfograven: Out of curiosity, is there something we could do to make you trust the list of supported players more?
13:53:19fogravenis there some firmware that will
13:53:33funmanexcept the official Apple firmware, no
13:54:58funmandomonoky, JdGordon, kugel: i'm curious if you tested the DMA code pasted in the forum in the bootloader; to confirm that it equally fails on all models and not just the Clip ?
13:55:25kugelfunman: yep, checksum fails
13:55:26funmanwe could use the current sd_read_sectors() function in the bootloader until Someone fixes it
13:56:24*JdGordon hasnt tested anything for about 3 weeks (apart from the build kugel sent me )
13:56:49*domonoky tests now..
13:57:01 Part fograven
13:57:29kugelwell, the initialization of the controller in the main binary is entirely independent of the bootloader (i.e. it doesn't care how the bootloader leaves sd) that should work
13:57:58kugelJdGordon: I can't really tell why you get a white screen though
13:58:17funmankugel: that *works* for sure, but imo it's not very clean
13:58:44JdGordonwhats not clean? re-iniitlaizing in the main build?
13:58:53kugelfunman: of course not
13:59:08kugelJdGordon: wrongly initialize in the bootloader
13:59:18funmanJdGordon: using 2 different sd_read_sectors() : one for the bootloader and one for the main build
13:59:23 Join Bensawsome [0] (n=Bensawso@unaffiliated/bensawsome)
13:59:25funmankugel: nothing suggests a wrong initialization
13:59:48kugelfunman: is it right?
14:00:04funmankugel: ?
14:00:35kugelnevermind, lets not be picky
14:01:19*domonoky gets a "file not found" with dma in the bootloader...
14:02:18funmandomonoky: then the (supposedly same) bug occurs earlier in the m200v4 than in Clip and Fuze
14:02:51funmanat least your FAT32 partition was detected, else an ATA error would be detected
14:03:14JdGordoncan someone direct link me the dma patch?
14:04:13funman : 1 link away
14:04:42domonokyhm, it more looks like my fat32 partition is now screwed up, the OF hangs at "rebuilding library"...
14:04:50 Quit kugel (Remote closed the connection)
14:05:18 Join kugel [0] (n=chatzill@
14:06:07funmandomonoky: in the bootloader sd_write_sectors() is not active, so I'm not sure where the problem comes from ..
14:06:55*domonoky boots liniux, to check the disc.. (damn mtp)
14:08:09funmanisn't there a key combination to use msc like on the Clip?
14:09:39 Quit kugel (Client Quit)
14:10:09funmanand a msc/mtp/auto setting ?
14:10:15JdGordonarg... no adc-target.h
14:10:20 Join _lifeless [0] (n=lifeless@
14:10:26 Quit _lifeless (Read error: 54 (Connection reset by peer))
14:10:50 Part TiMiDo ("Leaving")
14:11:11domonokyfunman: i dont think there is a keycombo... but on linux it should default to msc
14:12:47JdGordonfunman: domonoky: can you send me adc-target.h?
14:13:33JdGordonhmm.... why is it looking in the e200v2 folder for it?
14:13:49funmanJdGordon: just look in target/arm/as3525 (r19076)
14:14:48funmanJdGordon: look at TARGET_INC in the Makefile
14:14:56JdGordonyeah, I see it there... I wonder why the Makefile doesnt...
14:15:25*domonoky it als goes into mtp on linux, is there some way to disable mtp on linux ?
14:15:58 Join schNibbler [0] (
14:16:06funmandomonoky: in the settings menu?
14:16:27JdGordonbad checksum with the dma bootloader
14:17:17funmanthe behaviour is consistent on all models then, in the bootloader transfers start fine, and then are corrupted.
14:17:49funmanIf I check MCI_STATUS , everything appears to be fine; but I should enable interrupts on the 'problematic' bits to be sure
14:23:11funmansame result: no overrun/timeout
14:23:39 Join LambdaCalculus37 [0] (i=44a04303@gateway/web/ajax/
14:24:17funmanThe problem could come from SDRAM and not SD/DMA
14:25:34 Join n1s [0] (n=nils@rockbox/developer/n1s)
14:27:30 Quit Nibbl (Read error: 113 (No route to host))
14:28:12*domonoky found the keycombo for msc on m200v4... its select while connecting the usb :-)
14:29:05LambdaCalculus37domonoky: The m200v4 doesn't have selectable MSC like the older models?
14:29:40domonokyLambdaCalculus37: it does have a otpion in the menu, but if can not get into the of... :-)
14:30:28LambdaCalculus37Ahh. :)
14:31:01funmanI think this rebuilding database after reboot is a good way to brick devices. A more clever design would be to leave the choice
14:32:23domonokyfunman: yes, many mp3player makers do such silly thing...
14:38:15 Join Schmogel [0] (
14:43:16 Quit omgun ("CGI:IRC (EOF)")
14:49:06 Join TheSphinX^ [0] (
14:49:38 Nick JdGordon is now known as JdGordon|zzz (n=jonno@rockbox/developer/JdGordon)
14:50:46 Join dany_21a_ [0] (
14:51:08 Quit amiconn (Nick collision from services.)
14:51:08 Quit pixelma2 (Nick collision from services.)
14:51:13 Join amiconn_ [50] (n=jens@rockbox/developer/amiconn)
14:51:17 Join pixelma2_ [0] (n=marianne@rockbox/staff/pixelma)
14:51:19 Nick pixelma2_ is now known as pixelma2 (n=marianne@rockbox/staff/pixelma)
14:53:08 Quit dany_21a_ (Remote closed the connection)
14:53:25 Join dany_21a_ [0] (
14:58:32 Join omgun [0] (n=79b45a2d@gateway/web/cgi-irc/
15:00:12 Quit pondlife ("Leaving.")
15:11:25 Quit omgun ("CGI:IRC (EOF)")
15:23:08funmanthe OF uses 4*32bits DMA data transfers for I2SOUT, but the I2SOUT_DATA register is documented as being 32 bits wide, not 4*32 bits .. ?
15:23:14 Join domokin [0] (
15:23:21domokino hi
15:23:35 Quit domokin (Client Quit)
15:24:34LambdaCalculus37funman: Need to refresh my memory... I know that buttons aren't working on the Fuze currently, but what about on the Clip? Once my Clip arrives, I'm going to dig into the code and start testing.
15:24:50linuxstbfunman: What do you mean by "the OF uses 4*32bits DMA data transfers" ?
15:26:03domonokyfunman: it probably also has a fifo before the DAC..
15:27:20domonokyfunman: it even says at page 56, that it has a 128bit fifo.
15:30:43 Join Acky [0] (
15:32:42funmanLambdaCalculus37: all buttons work on the clip
15:33:26LambdaCalculus37funman: Cool. :)
15:33:50funmanlinuxstb: the OF transfers 4 32bits words, then rewind to the start address for the I2S register, transfer another 4*32bits; ..
15:35:19funmandomonoky: I saw this description, but I2SOUT_DATA is documented as 32 bits (page 68); and also as "two 32 bit wide data registers". I'm confused.
15:35:58 Quit Schmogel ("Miranda IM! Smaller, Faster, Easier.")
15:38:37 Quit krazykit (Read error: 104 (Connection reset by peer))
15:39:13 Join krazykit [0] (
15:39:21domonokyfunman: i think this 32bit register sits infront of the 4*32bit fifo. so you can write 4* 32bit to this register, and then you have to wait for the fifo to empty, i think.... and they speak about "two 32bit data registers", because there is also the loopback thing, i think...
15:39:39 Quit JdGordon|zzz (Remote closed the connection)
15:39:45J-23funman: Rockbox with your patch on e200v2 doesn't work, bootloader says "bad checksum"
15:39:51funmanJ-23: thanks
15:40:15J-23I formatted my player before trying.
15:40:35funmanhopefully Someone will find what's wrong because I can't
15:41:37 Join linuxstb_ [0] (n=linuxstb@rockbox/developer/linuxstb)
15:41:47 Quit linuxstb (Nick collision from services.)
15:41:51 Nick linuxstb_ is now known as linuxstb (n=linuxstb@rockbox/developer/linuxstb)
15:41:53 Quit Acksaw (Read error: 60 (Operation timed out))
15:44:00J-23what algorithm are firmware checksums calculated with?
15:44:42funmanJ-23: see bootloader/common.c line 190
15:44:58linuxstbJ-23: Just a simple sum.
15:45:49 Quit schNibbler (Read error: 113 (No route to host))
15:48:31J-23hm, when I disable checksumming, it returns "Data abort at 30000010"
15:48:52LloreanJ-23: The checksum's there for a good reason.
15:49:22J-23yes, I understand
15:50:02LloreanSo why are you trying to bypass it? If it's failing, it's a pretty good indicator that something is wrong and the binary won't work when loaded.
15:50:52***Saving seen data "./dancer.seen"
15:51:43dany_21a_hi funman, J-23: i also just tried the DMA patch from funman, also CRC-fail
15:51:57dany_21a_didnt kugel got it working on his fuze?
15:52:46domonokydany_21a_: dma works fine the main binary, but fails in the bootloader... so old bootloader and new dma-enabled build is fine :-)
15:53:02dany_21a_domonoky: ah..thx fill try that
15:54:17funmandany_21a_: did you notice atomikpunk thinks the buttons on the Fuze are read via DBOP ?
15:56:23J-23is "old" bootloader version that's in SVN?
15:56:24dany_21a_hey - cool work funman, just seen the main-menu again since about half a year (on a real device), ahh... feels good :)
15:56:29dany_21a_J-23: yes
15:57:03funmanweird, I notice the OF addresses the embedded 320kB TRAM from an alias, while it could use 0 like the major part of the OF
15:57:24dany_21a_funman: buttons/DBOP what would that mean, how could they be attached to the SOC... with additional logic, or just on the DBOP lines but same as GPIO?
15:57:55funmanDarksair: don't ask me .. :)
15:58:06funmanoops, i meant dany_21a_ . sorry
15:58:19 Quit Zagor ("Client exiting")
15:59:05J-23SVN bootloader and Rockbox with DMA patch shows terrible gray screen to me
15:59:11dany_21a_me too... just seen that the buttons have one common signal (ground or vcc) and the other goes straight to the zif-connector
16:00:02funmanJ-23: you might try to change the value I modified in firmware/target/arm/as3525/kernel-as3525.c
16:04:05J-23/home/Moarc/rockbox/apps/action.c:26:18: error: lang.h: No such file or directory
16:04:07funmanJ-23: you might want to not use patches also ;)
16:04:30funmanI only pasted it to request help from other developers
16:05:40 Quit robin0800 (Remote closed the connection)
16:06:03 Nick amiconn_ is now known as amiconn (n=jens@rockbox/developer/amiconn)
16:07:22*J-23 resigns
16:09:33 Join {phoenix} [0] (
16:19:06 Quit {phoenix} (Remote closed the connection)
16:23:47 Join saratoga [0] (n=98039ac5@gateway/web/cgi-irc/
16:25:27saratogais there an easy way to get a list of people with SVN access?
16:28:02funmandocs/COMMITTERS ?
16:28:26 Join mc2739 [0] (n=mc2739@
16:28:27*domonoky now knows why he gets a "file not found" in the bootloader... BOOTDIR wasnt set correctly in the config-m200v4.h :-)
16:31:18*domonoky detects that he always gets the same wrong checksum.
16:31:41funmandomonoky: what if you keep the m200 off for ~10s ?
16:31:56mc2739funman: I have not been able to get e200v2 buttons to work except in bootloader
16:32:55domonokyfunman: looks like it stays the same.. will test with a longer off-period... (removing batterys).
16:33:51mc2739but, I can confirm that sd write seems to function - config.cfg and nvram.bin were written to my .rockbox directory
16:34:39funmanmc2739: did you try to use the interrupts on gpio pins to monitor button status ?
16:35:13mc2739funman: no, I have not tried that
16:36:11linuxstbmc2739: Regarding your lcd fix, have you tested that lcd_update_rect() works correctly with your change? e.g. doing some tests to update different areas of the screen? I would expect it to not work correctly. The code in the test_viewports plugin might be useful for that.
16:36:30domonokyfunman: booting in the OF changes the wrong checksum.. a little bit, but it is still similar.
16:36:48mc2739linuxstb:the only test done so far is with the main menu
16:37:06 Join thegeek [0] (
16:37:18funmandomonoky: ok, similar to what I experience
16:38:07dany_21a_if the usb-power changes (on/off) on my fuze, the LCD-light goes on - in what function is that monitored? couldnt find anything
16:39:00funmanusing dma for i2sout is more complex since the data flow must be handled manually.
16:40:38 Quit thegeek (Client Quit)
16:43:22 Join captainkwel [0] (i=2669ecc2@gateway/web/ajax/
16:46:27 Join toffe82 [0] (
16:49:41saratogaoh then i probably shouldn't have committed Toni's patch then
16:49:48saratogadidn't realize he had SVN
16:49:53 Join nplus [0] (n=nplus@141.25.Globcom.Net)
16:50:21mc2739linuxstb: there may be some other deeper problem, originally, the display was rotated and mirrored (fixed in r18913), then the display only showed the first 176 rows and then overwrote the top with the remaining 44 (fixed in r18914)
16:52:55linuxstbmc2739: It just needs someone to refer to the datasheet and check/test the lcd_update_rect() function is setting up the transfer correctly. lcd_update_rect() isn't used in the bootloader, which is why it's just being seen now.
16:53:14 Join mofux_ [0] (
16:54:52mc2739linuxstb: I'll do some testing when I get home from work
16:55:20 Quit thegeek_ (Read error: 110 (Connection timed out))
16:56:44 Quit funman ("leaving")
16:59:03 Part linuxstb ("Leaving")
17:02:16 Join linuxstb [0] (n=linuxstb@rockbox/developer/linuxstb)
17:07:58 Join Seed [0] (
17:09:12 Quit saratoga ("CGI:IRC (EOF)")
17:09:17 Join XavierGr [0] (n=xavier@rockbox/staff/XavierGr)
17:11:19 Quit mofux (Connection timed out)
17:11:37 Join katZeeee [0] (
17:12:46 Part katZeeee
17:13:17 Part B4gder
17:19:53 Join jeffdameth [0] (
17:21:20 Quit mc2739 ()
17:23:32 Join japc [0] (n=japc@
17:24:04 Join {phoenix} [0] (
17:31:08 Quit toffe82 (Read error: 60 (Operation timed out))
17:32:21 Join kugel [0] (n=chatzill@unaffiliated/kugel)
17:33:11 Join bmbl [0] (n=Miranda@unaffiliated/bmbl)
17:34:35 Join toffe82 [0] (
17:40:21 Join mcuelenaere [0] (
17:41:09 Join HBK- [0] (
17:44:30 Join herrwaldo [0] (
17:50:53***Saving seen data "./dancer.seen"
17:58:44 Quit HBK (Read error: 110 (Connection timed out))
18:03:33 Join Lynx_ [0] (
18:06:27 Quit petur ("work->home")
18:10:04 Join Self-Perfection [0] (
18:13:30 Nick Self-Perfection is now known as self_ (
18:13:41 Nick self_ is now known as Self-Perfection (
18:14:05 Join Strife89 [0] (n=michael@
18:14:06 Nick Self-Perfection is now known as foo_1546 (
18:14:26 Nick foo_1546 is now known as Self-Perfection (
18:14:52 Quit moos ("Rockbox rules the DAP world")
18:22:53 Quit TheSphinX^ (Remote closed the connection)
18:25:11 Join mcuelenaere_ [0] (n=mcuelena@rockbox/developer/mcuelenaere)
18:28:53 Join stoffel_ [0] (
18:34:16 Quit Darksair ("(define zero (lambda (f) (lambda (x) x)))")
18:35:01 Join fyrestorm [0] (
18:36:38 Quit mcuelenaere (Read error: 110 (Connection timed out))
18:37:48kugelgevaerts: mkfs.vfat reports 3982336 for my fuze. That*s multiplied with 512 (SECTOR_SIZE) only half of my 4GB
18:38:50kugeldoes that mean the sector size is 1024? and if yes, what would that mean for rockbox which defines SECTOR_SIZE 512
18:39:50domonokykugel: and you are sure mkfs.vfat reports the size in sectors, and not in bytes ?
18:39:59 Nick Self-Perfection is now known as self-perfection (
18:40:08kugelalso, I see many TODO's of the kind "BLOCK_SIZE != SECTOR_SIZE?"
18:40:42kugel3982336 sounds pretty much not like 4billion bytes, does it?
18:41:29kugelI get pretty much exactly my disk space when I multiply with 1024
18:41:57kugelalso, I once tried to format my fuze with 512 as sector size, which didn't work well
18:42:14kugel(at least for the OF)
18:42:45 Join ibseo [0] (
18:43:43 Join bertrik [0] (
18:44:30domonokykugel: but it looks like ~4millon kbyte =4gb - the space for the OF... and i would think, if we get the sector size wrong, we would not be able to read anything on the fuzes SD..
18:45:30kugelmkfs.vfat certainly says sector
18:47:50gevaertsWhat does dmesg say?
18:48:33kugelhm, that one reports 512byte sectors
18:49:16*kugel always read blocks at dmesg, but it says in fact sectors
18:49:25 Join beast_ [0] (
18:50:37 Quit pixelma2 ("-")
18:50:50 Join pixelma [50] (i=pixelma@rockbox/staff/pixelma)
18:54:39beast_i am running rockbox 3.0 on a 60GB video ipod and playback of mpg's that look fine on my pc are choppy on the mpegplayer. Is there a specific fps rate I need to set when converting? Or am I on the wrong path?
18:57:54advcomp2019beast_, did you look at the wiki page for that
18:58:32 Quit EspeonEefi ("さよăȘら")
18:59:50 Join Rondom [0] (
19:00:00 Join perrikwp|class [0] (i=982139a9@gateway/web/ajax/
19:00:59 Join nuonguy [0] (
19:01:04bertrikhmm, CONTEXT_SETTINGS is hardly used in rockbox
19:01:38beast_advcomp2019: yes, that's where I got the correct LCD size from, but there is no fps rate for 1:1 video which is what I am trying to do
19:03:41advcomp2019o ok.. i do not that number too
19:03:52GodEater_LambdaCalculus37: do you know if anyone else tried to access the DFU on a 3G at all ?
19:04:27pixelmabeast_: on the Ipod Video you are probably better off watching videos in the original firmware
19:04:44LambdaCalculus37GodEater_: No, I don't think anyone else has.
19:04:55LambdaCalculus37You may be the first. ;)
19:04:55GodEater_LambdaCalculus37: woo - trailblazing is fun :)
19:05:12GodEater_pixelma: I don't think there's any "probably" about it
19:05:59beast_pixelma: thanks for letting me know. looking forward to the video support being as good as the audio :)
19:06:09GodEater_long wait
19:06:14kugelthe e200v2 apparently buttons work for the fuze
19:06:20pixelmamaybe some entered DFU mode on 3rd gen Nanos by putting the device into the washing machine (remember the stories of 2nd gen Nanos)
19:06:41kugeldomonoky, bertrik: ^
19:06:51bertrikkugel, really?
19:07:21kugelI copied button-e200v2.c into button-fuze.c, and removed the #if 0 in the bootloader
19:07:54kugelI had the strong assumption that it's the same, since the USB is on the same GPIO and lcd is nearly the same (so I thought no DBOP for fuze)
19:08:14bertriknice, so all this time we were overlooking the obvious ... :)
19:08:22pixelmaGodEater: just thought if someone really refuses to use iTunes (as I assume you need to put videos on the Ipod), Rockbox would be the only way to watch videos...
19:08:49kugelbertrik: well, yea
19:09:15*kugel feels great now :p
19:09:26 Join funman [0] (
19:09:39GodEater_pixelma: there's always itunes alternatives though
19:09:39funmanhail to kugel !
19:09:45GodEater_gtkpod etc.
19:09:52kugelfunman: did you read that?
19:10:09funmankugel: so you got all the buttons found on e200, work on your Fuze, with the same settings ?
19:10:20pixelmaGodEater: aha, no Ipod here and that's getting a bit off-topic
19:10:23funmankugel: I can read your mind before your thoughts reach your keyboard !
19:10:33*funman gives his $ to logbot
19:10:40kugelliar, you just read logs before joining ;)
19:12:18 Join culture [0] (
19:12:40funmanvery weird, how come we didn't find them before? perhaps some delay tweaking (that would explain why e200 buttons stopped responding)
19:12:58kugelfunman: the directional buttons, center and power, all work
19:13:02kugelall are even the same
19:13:45funmanquick tell fdinel so he can fix the DMA driver !
19:13:47kugelfunman: maybe.
19:13:55kugelfunman: definitely :P
19:14:06*kugel fears he's still searching in DBOP
19:14:16kugeleven though I told him last night that this is VERY unlikely
19:14:17funmanby that time he would be at work
19:15:37amiconnlinuxstb: Btw, APE 3.97 is quite a bit faster to decode than 3.99 (but the compression levels are a bit less effective, and there's no -c5000)
19:16:18kugelfunman: is there any specific reason you didn't remove the disable_irc() in the bootloader?
19:16:22kugelin your patch that is
19:16:35funmanwe need to use the dma interrupt
19:16:45funmanhm .. they are enabled somewhere else
19:16:52funmanah I understand
19:17:06 Join mcuelenaere [0] (n=mcuelena@rockbox/developer/mcuelenaere)
19:17:15kugelshouldn't disable_irc() then at least be before storage_init()?
19:17:15funmanthey are disabled when executing the new firmware, because if they happen before we copy the new vectors; that's a path to bad things (tm)
19:17:34funmanno, since they must be enabled for storage to work at all
19:20:59 Join Thundercloud [0] (
19:21:34GodEater_well it certainly won't enter DFU mode just by rebooting and doing the same keypress as the 2G :(
19:22:08LambdaCalculus37GodEater_: :(
19:22:25LambdaCalculus37And since there's no hardware partition to trash... hmmm...
19:22:28*GodEater_ tries to think of non-destructive things to try
19:22:29*LambdaCalculus37 has a notion
19:22:48 Quit mcuelenaere_ (Read error: 60 (Operation timed out))
19:22:52*kugel always assumed the hardware mapping of fuze and e200v2 are the same at all
19:22:53LambdaCalculus37GodEater_: But just out of curiousity, what *would* happen if you trashed the entire storage partition?
19:23:32GodEater_no idea
19:23:41GodEater_I guess it's worth a try
19:23:47LambdaCalculus37GodEater_: Be a trailblazer. ;)
19:24:02LambdaCalculus37Try dd'ing /dev/urandom to the whole partition.
19:25:09*GodEater_ cranks his dd handle
19:26:54 Join faemir [0] (
19:27:50kugelfunman: I try different gpios out now, and sometimes the text isn't displayed when I press the home button
19:29:34 Quit Bensawsome ("The awsome is gone :(")
19:29:59funman'the text'?
19:30:25 Join BigBambi [0] (n=Alex@rockbox/staff/BigBambi)
19:31:15kugelfunman: the text which the bootloader shows. loading firmware, model, etc
19:31:17 Join miepchen^schlaf [0] (
19:31:39funmanwhich buttons are missing except the scrollwheel?
19:31:48*bertrik plays with clip keymaps
19:31:58 Quit einhirn ("Miranda IM! Smaller, Faster, Easier.")
19:31:59funmanbertrik: cool !
19:32:13kugelfunman: home and hold
19:32:28 Quit perrikwp|class (" ajax IRC Client")
19:32:41funmankugel: now you might want to find some code to put in mkamsboot for dualboot. What if you reuse the e200v2 case ?
19:33:13kugelfunman: sure, why not. so that both left and usb boots into OF?
19:33:59 Join Zagor [242] (
19:37:07bertrikfunman, just fixing some keymap basics, BTW did you notice the battery level makes sense now?
19:37:19kugelhaha, then I can finally boot into the of without having a pc handy :p
19:37:26 Join Horscht [0] (n=Horscht@xbmc/user/horscht)
19:39:00funmanbertrik: yes I noticed, thanks for doing that
19:40:19 Quit faemir (Remote closed the connection)
19:40:38funmankugel: ?
19:41:28 Join faemir [0] (
19:42:06kugelfunman: yep, exactly this should work
19:42:29funmankugel: ok, can you tell me again your name ? (I don't trust your whois information)
19:42:46kugelthat's what I was about to do, after some tries to find the home/hold button
19:42:52kugelThomas Martitz
19:43:36kugelI'll update the hardware mapping later
19:43:40kugelthe wiki page I mean
19:43:55 Join toffe82_ [0] (
19:44:22GodEater_well mkfs.vfat is as unreliable as ever
19:46:34kugelfunman: there's no afsel on gpiod?
19:46:42kugelor did someone just forget to write it down
19:46:51 Quit Rondom (Nick collision from services.)
19:46:54Zagoramiconn: do you really need to define LIKELY in demac? isn't the define in system.h pulled in?
19:47:07kugellol nevermind typo
19:47:09 Join Rondom [0] (
19:47:22amiconnZagor: Check where I defined it. libdemac can be built standalone...
19:47:36 Join Photoguy [0] (n=Adams@
19:47:44funmankugel: can you check if gpio*dir and gpio*afsel can be modified only once ? (in a button_device_init() for example)
19:48:40amiconnZagor: That's in fact what I used for profiling. I compiled the standalone decoder with -fprofile-arcs, then used it to decode some ape files, then used gcov on the output
19:49:20Zagoryeah standalone is good for debugging and optimising
19:49:32 Quit japc (Read error: 60 (Operation timed out))
19:49:40GodEater_LambdaCalculus37: did you ever do a detailed lsusb on the 2G ?
19:49:48bertrikhow much improvement did we get from (UN)LIKELY so far?
19:50:54funmanLIKELY(appreciable percentage)
19:50:55LambdaCalculus37GodEater_: Yes, I have one here:
19:50:56***Saving seen data "./dancer.seen"
19:51:56Zagorbertrik: amiconns commit message says "Biggest effect on coldfire (-c1000: +8%, -c2000: +5%), but ARM also profits a bit (less than 1% on ARM7TDMI, around 1% on ARM1136)."
19:53:13kugelfunman: i thought they have to be set back for lcd to work
19:53:16 Join mofux [0] (
19:53:20 Join shotofadds [0] (n=rob@rockbox/developer/shotofadds)
19:53:30pixelmaZagor: can you explain to me what problem your pegbox commit last friday fixes?
19:53:37LambdaCalculus37GodEater_: What is the 3rd gen nano giving you?
19:53:56funmankugel: I don't know, but I suppose not all the pins are used in lcd, and gpio*_dir shouldn't matter
19:54:28bertrikZagor, ok thanks
19:55:01GodEater_LambdaCalculus37: couldn't tell you at the moment, it's hooked to Mrs GodEater's laptop
19:55:07GodEater_it only just occurred to me to try it
19:55:18Zagorpixelma: when compiling dependencies, no bitmap exists. so BMPWIDTH_pegbox_pieces is undefined. this means the compiler would fall through to the #error line and hence an #error output every dependency generation.
19:55:25GodEater_the iphone presents about 3 possible configurations when you hook it up over usb
19:55:36 Join markun [50] (n=markun@rockbox/developer/markun)
19:55:38GodEater_so I wondered if there was more than one of the other ipods
19:55:49 Quit mofux_ (Read error: 60 (Operation timed out))
19:55:51Zagorpixelma: the old makefiles simply redirected all errors to /dev/null but I prefer to keep them visible
19:56:09 Join mofux_ [0] (
19:56:09GodEater_I'll get a proper listing in a moment
19:56:16LambdaCalculus37GodEater_: The 2nd gen nano gives me two possible combinations (that I found so far), depending on if it's hooked up via OF or DFU mode.
19:56:27pixelmaZagor: but there are other plugins which get their variables from bitmap sizes (e.g. brickmania)
19:56:42LambdaCalculus37Check the DeviceDetection wiki page; I posted the PID and VID for both modes up there.
19:59:19amiconnThere's further potential in predictor-arm.S on ARMv6 - right now it stalls the pipeline quite a few times
19:59:51amiconnReordering instructions will help (and it won't have a negative effect on earlier arm versions)
20:01:04 Quit toffe82 (Read error: 110 (Connection timed out))
20:03:23 Part LinusN
20:05:15 Quit |AhIoRoS|_ (Client Quit)
20:06:26 Quit funman ("leaving")
20:07:31kugeljhMikeS: ping. 1 last talk about backlight fade please.
20:09:39GodEater_LambdaCalculus37: interesting - the 3G presents two configurations by default
20:10:05LambdaCalculus37GodEater_: Very interesting. What have you found so far?
20:10:18GodEater_one is the expected "Mass Storage"
20:10:23GodEater_the other is something else
20:10:25 Quit mofux (Connection timed out)
20:10:25GodEater_let me paste the log
20:12:48 Join Bensawsome [0] (n=Bensawso@unaffiliated/bensawsome)
20:13:07*GodEater_ thinks the log would mean more to someone like gevaerts
20:15:34ZagorI see three different configurations: Mass storage, Audio and HID
20:16:25Zagorbleh, I keep mixing up the usb terminology
20:16:39Zagoryou're right there are only two configurations
20:17:00Zagoraudio and hid are both interfaces in the "iPod USB Interface" configuration
20:17:23GodEater_I wonder what the hid one is used for
20:17:43 Quit mofux_ (Read error: 110 (Connection timed out))
20:17:47ZagorI would guess to allow external hosts (cars etc) to change the display
20:17:57 Quit miepchen^schlaf ()
20:18:05GodEater_ah possibly
20:18:10rasherSo you're speculating it's using standard USB for this?
20:18:21 Join mofux [0] (
20:19:02Zagorrasher: well it could be. we have to snoop some messages to find out.
20:19:22*GodEater_ will try and get a usbsnoop done later
20:19:23Zagoror find some code that uses it :)
20:22:08 Join miepchen^schlaf [0] (
20:22:24bertrikUSB devices usually have only one configuration. I think windows only supports one (have to look that up).
20:23:27Zagorbertrik: I don't think the second configuration is intended for "normal" computer use
20:23:51kugelhm, so the fuze and e200v2 button driver are the same for now. does it make sense to already merge them? They're far away from finished
20:23:57*kugel wouldn't think so
20:24:14Zagorkugel: not until we know more imho
20:25:29 Quit stoffel_ (Read error: 113 (No route to host))
20:25:59GodEater_bertrik: are you sure ?
20:26:09GodEater_as I said, the iphone has multiple configurations too
20:28:15 Quit captainkwel (" ajax IRC Client")
20:28:59 Join captainkewl [0] (i=2669ecc2@gateway/web/ajax/
20:29:52bertrikI think I read that on a microsoft page but I can't find the link anymore
20:32:30Zagorduh, the HID class is of course for the accessory protocol
20:32:41Zagorhid interface...
20:32:49GodEater_Zagor: "of course"?
20:32:57GodEater_why is it not present on older iPod's then ?
20:33:52Zagorsorry, I did not mean that as "I'm sure of it". rather like "it makes sense".
20:35:34*gevaerts isn't convinced
20:35:49Zagor indicates there are two protocols: 1-wire (over usb) and 2-wire (not usb)
20:35:59gevaertsI'd expect the ipod to be master on the accessory protocol bus
20:36:52Zagorgevaerts: with HID it doesn't really matter who is master or slave. keyboards are HID slaves for instance.
20:37:09gevaertsExactly. Aren't most accessories slaves?
20:37:38Zagoriap is not only for accessories such as itrip. think docks, stereos or cars. they are rather masters than slaves.
20:38:41gevaertsIt depends on your point of view I guess
20:38:51amiconnGodEater: The older ipods run the accessory stuff via serial
20:38:59 Part dany_21a_
20:39:00 Join petur [50] (n=petur@rockbox/developer/petur)
20:40:07 Join dany_21a_ [0] (
20:44:03 Join akhilleus [0] (n=joe@
20:45:50 Quit reacocard (Read error: 110 (Connection timed out))
20:47:12 Join n17ikh|Lappy [0] (
20:47:21kugeldany_21a_: great news for the fuze (I love that rhyme)
20:49:53 Join stoffel_ [0] (
20:50:10 Quit AndyIL (Read error: 60 (Operation timed out))
20:50:54 Join mofux_ [0] (
20:52:12 Join tessarakt [0] (
20:55:08 Join dabujo [0] (
20:57:58dany_21a_hi kugel... you mean the keymapping?
20:58:07dany_21a_yeah - great work
20:58:32dany_21a_so you can already navigate the menu on your fuze?
20:58:44 Join AndyI [0] (i=AndyI@
20:59:30kugelhaven't tried
20:59:44kugelgoing by e200v2 users, no it shouldn't work
21:00:11dany_21a_oh...okay, i thought it already works on other devices...
21:01:16kugelthey apparently only work in the bootloader
21:01:50 Quit nuonguy ("This computer has gone to sleep")
21:02:53 Quit Photoguy (Read error: 104 (Connection reset by peer))
21:03:08dany_21a_did you read my post at the forum? (status bar fickering)
21:04:10kugelwill try when I intend to boot into main next time ;)
21:05:01dany_21a_even the clock already works... me=amazed
21:05:38kugeldany_21a_: to get the buttons working simply copy the button_int function from button-e200v2.c to button-fuze.c
21:05:59kugelmaybe someone should commit that?
21:06:22Unhelpfulbertrik: i have effective area and linear scalers for color. i'm not considering those to be worthwhile on gray or mono targets, but there are people who want *some* form of scaling there if we have it anywhere else, and i think nearest-neighbor can be done absurdly cheap
21:06:49Unhelpfulas in, cheap enough to call the scaler even if we're not scaling, and remove the unscaled-output code for <16bpp
21:07:00 Quit mofux_ (Remote closed the connection)
21:07:02 Quit mofux (Read error: 110 (Connection timed out))
21:09:08amiconnjhMikeS: What do I need to do if I want a beast build running at 532MHz?
21:09:47kugelOk, is waiting for a commit
21:11:38Unhelpfuli need to test it, but i *think* i see how to formulate a bresenham's-like algorithm that lets me skip many input lines quickly, and that will work for any "slope". and having scaling for the cost of one or two extra adds or compares is probably cheap enough to consider always calling the scaler.
21:13:03bertrikI think speed is not really an issue for scaling
21:14:43Unhelpfulbertrik: the "smooth" scalers for color do a 3-6 multplies per input pixel and per output pixel, as well as copying everything into a buffer with wider sizes and back. i don't really think we should be calling them for unscaled output.
21:16:32n1sUnhelpful: how long does it take to scale a, say 500*500 bmp to 100*100 on one of our targets that don't have overkill cpu's? (pp for example)
21:16:35Unhelpfulthe nearest-neighbor i'm talking about for gray/bw will be doing a couple of divides on init, and then just add/sub/cmp, so it might make more sense to call it unconditionally than to have a separate wad of code that does the same thing without ever skipping pixels.
21:17:11Unhelpfuln1s: only color i've tested on is beast. maybe i should throw this code onto one of my sansas :D
21:17:12n1sdivides are crazy expensive on armv4
21:18:16amiconndivides are crazy expensive on any arm
21:18:26Unhelpfuln1s: yes, but i could always add a special case to skip the divide if in==out
21:18:48amiconnarmv4 is worse with libgcc, not so much with the optimised assembly that's now in the codeclib
21:18:48Unhelpfulstill much cheaper than the code duplication, i would think.
21:19:12amiconnThe arm division routines do their own optimisation
21:19:26Unhelpfulamiconn: right now i'm using a straight /, which i supposed means libgcc's routine.
21:21:02Unhelpfulthere are two on scaler init, to calculate inverses used for division-by-multiplication inside the loop
21:23:42 Quit Rondom (Nick collision from services.)
21:24:00 Join Rondom [0] (
21:24:19ZagorI'm removing "draft version" from the manuals. Yell fast if you disagree...
21:24:20amiconnEven a couple of divisions per line would be okay. One per pixel would probably be bad
21:25:11gevaertsZagor: what's wrong with just removing that for release versions?
21:25:25Unhelpfulamiconn: i don't need a couple per line, though. i just need to divide by a constant, either the input or output dimension, based on scale direction. easier to store its inverse once.
21:25:29 Quit self-perfection (Remote closed the connection)
21:25:36Zagorgevaerts: because it's very annoying when you're trying to read the manual.
21:25:55LambdaCalculus37Zagor: Go for it! :)
21:26:02gevaertsreally? I never even notice it's there.
21:26:14gevaertsAnyway, I don't care much either way
21:26:18Unhelpfulif i could get logf to not cause data aborts on my beast, i'd add some real timing code, and see how much time is actually being spent in the scaler. i count five mississippis while loading a 1600x1200 wallpaper into sliding_puzzle.
21:27:28n1sUnhelpful: you can count ticks and splash
21:28:16Unhelpfuln1s: i suppose i could... splash is synchronous, i'd imagine? i don't want it to hose whatever wants to draw after it.
21:29:31n1sas long as it's called from the gui thread you're fine, but it usually works ok from other threads (ok enough for debugging purposes)
21:30:13amiconnsplash() is synchronous, but doesn't delete, so it stays until overdrawn
21:30:48amiconnAnd afaik sliding_puzzle creates no extra thread, so it runs in the main (== gui) thread
21:32:40Unhelpfulit's the easiest way i have to load a scaled bitmap on demand. i suppose rockpaint could be hacked to use resize-on-load easily enough
21:35:02captainkewl... simdisk?
21:35:39Unhelpfulok, gave up counting missisipis :/
21:35:52Unhelpfulsansa is taking... a while. i should try one of the smaller samples.
21:36:16LambdaCalculus37captainkewl: Calling the sim "drive" simdisk makes more sense in the long run, IMHO.
21:36:36Zagorcaptainkewl: I was going to call it just 'disk' first, but then I realized how confusing conversations that would spawn. better have a distinctive name that is not easily confused with something else.
21:37:21captainkewlsure, makes more sense than "archos". :)
21:37:57n1sdevice_root ?
21:38:10n1sor player_root mayeb?
21:38:46Zagorn1s: I didn't want it complex or technical. just a simple name.
21:39:30gevaerts"My music"?
21:39:48LloreanMaybe just "sim_files"
21:40:31Zagorfeel free to change it if it bothers you :-)
21:40:41*Llorean kinda likes it being "archos"
21:41:00LloreanIf we were ever going to change it, I'd recommend leaving support for archos indefinitely and just adding another option.
21:41:36ZagorLlorean: why bother? nostalgia?
21:41:46*Unhelpful would *not* recommend people put art >2x screen width on a sansa with this patch ;)
21:42:05Unhelpfulthe shakespeare test image is about that size, and takes ~3s to load-and-scale
21:42:16LloreanZagor: Exactly. There's no real harm in it, and it's about the only real nod you can keep around to the first wave of players that won't ever be harmful if it's just left like that.
21:42:55LloreanUnhelpful: 3 seconds isn't really that bad since it happens during buffering.
21:43:25LloreanUnhelpful: Do you ensure the CPU is boosted during scaling?
21:43:53UnhelpfulLlorean: ... i have no clue how? that might be a big part of the difference, we don't do speed scaling on beast yet, correct?
21:44:05ZagorLlorean: *this* is nostalgia: :-)
21:44:41LloreanUnhelpful: No, we don't on the beast as far as I know.
21:44:59Unhelpfuli should probably also do the math to pin down a range on possible scalefactors, and then have the load fail if we step outside that range
21:45:02n1sUnhelpful: rb->cpu_boost(true); will boost the cpu, and no, no scaling on the beast
21:45:39n1sbut it runs at 264MHz and the sansa is at 80MHz when boosted :)
21:45:46Unhelpfulminimums are down from 2x2, or up from 2x2...
21:46:09Unhelpfulmaximums are something completely ridiculous to imagine storing in a bitmap...
21:47:50n1syay, my midi testfile is now playing without sjipping on the c200 :)
21:47:58 Quit stoffel_ ("leaving")
21:48:03Zagorn1s: nice!
21:48:09n1seh s/j/k/
21:49:21Unhelpfulnapkin math says down-from/up-to ~16mpixels in width/height
21:50:58***Saving seen data "./dancer.seen"
21:51:07*linuxstb hands n1s a harder midi file ;)
21:51:33Zagorkugel: committed
21:51:43*amiconn wonders whether we should try floating point libmad on the beast, and then introduce some vfp code...
21:51:54n1slinuxstb: it is still a bit off since all voices are not enabled yet so some files miss a couple of instruments still
21:52:28Unhelpfulbeast has iwmmxt, doesn't it?
21:54:01 Join funman [0] (
21:54:33funmanI have a little problem with FlySpray : instead of timestamps I see the string "bagder" when I'm connected.
21:54:48Unhelpfuln1s: the scaler isn't called from the plugin context itself - the whole point of this work is a scaler that *might* be considered allowable in-core. looks like i include system.h and call cpu_boost(true)?
21:55:16gevaertsfunman: he's everywhere!
21:55:25n1sUnhelpful: ah, yes if you are not in a plugin that's right
21:55:38Unhelpfuli'm assuming i need to unboost it myself, as well? and what about making sure i don't unboost it when something else already had it boosted?
21:55:46Zagorfunman: wow!
21:55:59amiconnUnhelpful: You don't need to take care of that.
21:56:06n1sUnhelpful: yes, you need to unboost, same function with a false arg
21:56:30amiconncpu_boost is cumulative. It won't actually unboost until all user let go
21:56:36LambdaCalculus37funman: No mushrooms or snakes? ;)
21:56:42funmanno no !
21:56:50amiconnlinuxstb: Btw, your 120% realtime figure seems to be rather precise - my H180 now just even plays the -c3000 test file without skipping
21:57:25funmanam I the only one to see this ?
21:57:35amiconnThe 115% before r19199 weren't sufficient
21:57:55Zagorfunman: I see dates anyway
21:58:03LambdaCalculus37funman: I saw it too.
21:58:18Unhelpfulheh, i better make sure it always gets released, then :/
21:58:54 Quit LambdaCalculus37 (" ajax IRC Client")
21:59:14 Join BigBambi_ [0] (n=Alex@rockbox/staff/BigBambi)
22:00:38 Quit dabujo ("( :: NoNameScript 4.2 :: )")
22:01:56funmanas a proof lambda and me aren't on drugs :
22:03:20Bagderthat's... weird
22:03:27n1sfunman: i got that once, i think i reset some of my FS settings and it went away
22:04:43n1scheck the "Detailed date format" fiel especially
22:04:46funmanah indeed, "Detailed date format" == "bagder"
22:05:05Unhelpfuli'm not seeing a big change with boost?
22:05:16n1smine has my nick for some reason but the display works anyway...
22:05:39amiconngevaerts: thanks
22:06:05n1sUnhelpful: the difference should be pretty large if it was not boosted before
22:06:22n1sunless you are not cpu limited...
22:09:52 Quit ibseo ("quit")
22:15:06n1samiconn: how intrusive would using floats in libmad for the beast be?
22:15:55amiconnI don't know. Afaik libmad can be built for floating point using some #defines
22:17:01 Quit BigBambi (Read error: 113 (No route to host))
22:18:21 Quit {phoenix} (Remote closed the connection)
22:19:39 Quit nplus (Remote closed the connection)
22:21:06 Quit HBK- ()
22:22:24n1si think it would be worth testing but not if it requires a lot of work to get an idea of possible gains
22:23:33 Quit Rondom ("good night")
22:23:40 Join HBK [0] (
22:30:03 Join stoffel_ [0] (
22:31:56 Join thegeek [0] (
22:32:45*n1s wants a tester with a pp target that can do 22kHz output
22:32:55Unhelpfulthe huge image definitely scales faster w/ boost on. the debug splash can't be right, it claims that was only 200ms... :/l
22:33:56n1sUnhelpful: are you counting ticks?
22:34:36Unhelpfulyes, but i'm outputing ticks * 1000 / HZ
22:35:04Unhelpfulunless it's been long enough for that to overflow, which i guess all depends on HZ.
22:35:12n1sHZ ==100
22:36:19Unhelpfuli see the problem ;)
22:46:04 Join japc [0] (
22:46:23 Quit domonoky (Read error: 104 (Connection reset by peer))
22:46:51 Quit stoffel_ ("leaving")
22:51:25Unhelpfulall i should need is long tick1 = current_tick; {stuff} long tick2 = current_tick; - right?
22:51:54Unhelpfulbecause i'm seeing a couple thousand ticks pass in what i count off as 2s :/
22:52:23 Quit petur ("Zzzzz")
22:52:52funmantry counting slower? :)
22:53:01gevaertsNo, faster :)
22:54:08Unhelpfulsomehow don't think i'm counting *that* slowly :P
22:55:34Unhelpfulanyway, 2x downscale on sansa is on the order of 1-2s. the 1600x1200 sample takes 10s or so.
22:56:10Unhelpfuli don't think people will be using piles of bitmap album art at much larger than screen sizes on a flash player - they already complain about having to use bitmap :P
22:57:23 Quit advcomp2019 (Read error: 54 (Connection reset by peer))
22:57:58 Join advcomp2019 [0] (n=advcomp2@unaffiliated/advcomp2019)
23:04:10 Join MethoS- [0] (
23:06:44 Join fml [0] (n=4fd3c629@gateway/web/cgi-irc/
23:07:52fmlHello. Has anybody performed the runtime test with sansa e200? Does the commit of r19194 bring the improvement? How much?
23:12:08Unhelpfuli believe there's a page of battery_bench results on the wiki
23:14:59 Quit martian67 (Remote closed the connection)
23:14:59bertrikfml, I don't know, I'm curious too
23:16:19 Join martian67 [0] (i=lol3izer@about/linux/regular/martian67)
23:18:29lasserfml: my e280 (r19195) has run for 7:13 h now and I cancelled the test right now, the battery stands at 62%. Tomorrow evening I will have the results for r19173.
23:19:31lassertomorrow evening means in about 20 hours...
23:20:17fmllasser: why r19173 and not r19194? I.e. why without the patch in question?
23:21:17n1sUnhelpful: int start_tick; {stuff}; splashf(HZ, "%d", current_tick - start_tick); is what i do usually and it's reliable
23:22:04Unhelpfuli think it's probably fast enough, if people are putting reasonable-size covers on their players.
23:22:07 Quit Strife89 (Read error: 60 (Operation timed out))
23:23:02n1sunless your start_tick is an automatic var and the function is executed multiple times before displaying the result
23:23:18lasserfml: because I have the sources of r19173 and don't know how to get r19193.
23:23:35n1slasser: svn up
23:24:20shotofaddsgevaerts: did you ever try to TCC77x USB code on your D2? I just get a freeze after the USB logo is displayed
23:24:33lassern1s: I don't know anything about svn, sorry.
23:24:43n1slasser: or rather svn up -r19193
23:24:53n1slasser: are you using a tarball?
23:24:56fmllasser: just install the current RB build
23:25:40Unhelpfuln1s: nope, i'm timing in resize_on_load, around the calls to the actual scalers
23:26:32n1sUnhelpful: then i agree, weird :)
23:26:55bertrikthe UsingSVN wiki says to add svn:keywords to new files. Many files also have svn:eol-style
23:27:11n1sUnhelpful: btw you should have some yields somewhere if it can take that long
23:27:47lassern1s: I use to download the rockbox.7z sources from current builds and compile them with applied patches...
23:28:36n1slasser: if you want a specific revision svn is the way to go (and much easier to update)
23:28:56Unhelpfulit would be easiest to do per-line... i would assume the delay a yield adds depends on how much ohter things are loading the CPU?
23:29:00n1sbertrik: yes svn:eol-style should be set to native
23:29:13fmllasser: but in this case you don't have to compile anything. Just install the most current build, anything after r19194 (we're now past that) will do.
23:29:37bertrikn1s, ok I'll fix that for a couple of files
23:29:50n1sUnhelpful: yield lets other threads run, if you starve them for too long music will stop ui will be unresponsive etc
23:30:22 Quit funman ("leaving")
23:30:27Unhelpfulright, so a per-pixel-count yield would be ideal, but per-line would probably be reasonable?
23:30:41 Join MegafEee [0] (n=Linux@unaffiliated/megaf)
23:31:15 Quit fml ("CGI:IRC 0.5.9 (2006/06/06)")
23:31:17Unhelpfulit looks like only the area scaler would really need to - the bilinear scaler is much faster, presumable because the per-output-pixel work is roughly the same, but the per-input-pixel work is much larger.
23:31:22n1smore than say 100 times per second is probably overkill
23:31:31lassern1s: so, if I want to get the results of r19193 then I should have a look at svn... I have to go and find some time anywhere...
23:31:54n1slasser: yes
23:31:57kugelZagor: thanks for the commit :)
23:32:06kugelAnd thanks for the ROCKBOX_DIR commit.
23:32:11Zagorkugel: you're welcome
23:32:33Unhelpfulhrm... does current_tick update while something's banging on the CPU? i would imagine it's driven by a timer interrupt?
23:32:48Unhelpfuli could just check the tick at convenient places, and yield if needed
23:32:55kugelNow, I think the root dir should also be usable as arg when you configure to build a bootloader
23:34:01Unhelpfulheh, the add/sub/cmp only coordinate conversion works in python. time to make it work in nearest-neighbor scaler :D
23:35:42lasserfml: If I look at the changelog it ssems to me that r19173 must do the job, no code changes for e200 since then. Do you agree?
23:36:14kugelZagor: would be best to use the method he used, but instead of #define ROCKBOX_DIR to #undef BOOTDIR and then #define BOOTDIR
23:36:25 Join Strife89 [0] (n=michael@
23:36:55Zagorkugel: bootdir and rockboxdir are not always the same. some targets put the firmware file in root.
23:37:55Unhelpfuli don't suppose an arm guru could say if comparing to a value is much slower/faster than a sign check?
23:37:56 Quit Lynx_ (Read error: 104 (Connection reset by peer))
23:38:22 Join Lynx_ [0] (
23:40:10kugelyea, I mean for normal builds define ROCKBOX_DIR and for bootloader redefine BOOTDIR
23:41:13kugelI didn't claim they're the same
23:41:18 Quit faemir (Remote closed the connection)
23:41:56kugelZagor: btw: regardless of BOOTDIR, load_firmware at least will always look in the root if the file isn't found in BOOTDIR
23:42:19 Quit lasser ("ChatZilla 0.9.84 [Iceweasel 3.0.3/2008092816]")
23:42:36 Quit Strife89 ("Dinner.")
23:42:41Zagorkugel: it is the opposite that is a problem. if the zip installs the firmware in a dir where the (factory) bootloader doesn't expect it
23:42:44 Join MethoS-- [0] (
23:43:52Zagorkugel: overall I'm sceptical about encouraging such things in the bootloader
23:44:04Zagorthe rolo use case is a lot more valuable
23:44:24kugelZagor: that doesn't have anything to with configuring the bootdir, when you're talking about the factory bootloader
23:45:07Zagorkugel: you're right... :-)
23:45:24kugelthats the problem of the patch you committed actually. and for that reason (to make it kinda usefull) the bootdir should also be configurable imho
23:45:58Zagorhuh? what is the problem?
23:46:44 Quit bertrik ("Leaving")
23:47:55kugelI install my rockbox with changed root dir, but it won't load, since the standard bootloade looks in .rockbox
23:48:18kugelunless I have a normal build installed at the same time
23:48:34Zagorkugel: the rockbox_dir change is explicitly designed to NOT be used for booting. it's designed to be used for rolo testing.
23:48:52Zagorit even says so in configure −−help
23:49:04kugeland what's the specific reason to disallow such a option for bootloaders?
23:49:41Zagorwhat would be the reason to add it? I can't see a use for it, other than vanity changes.
23:49:51 Join saratoga [0] (n=9803c6dd@gateway/web/cgi-irc/
23:50:42saratogafinally managed to piece together a fuze from spare parts
23:50:52saratogathey are not easy to reassemble
23:51:02***Saving seen data "./dancer.seen"
23:51:30kugelZagor: either I edit my config-*.h or do it like building a normal build in a consistent manner
23:52:02kugelthe possibility not not boot from .rockbox is there. It's just the parameter for configure to be implemented
23:52:10kugelto not*
23:52:21Zagorkugel: yes, but what point is there to change this value?
23:52:45Zagoradding things "because we can" is not reason enough
23:53:44*amiconn managed to speed up the ape predictor on armv6 by almost 20%, just by shuffling instructions around
23:53:55amiconnOverall speedup is ~5% for -c1000
23:54:45gevaertsshotofadds: I think I got as far as that as well
23:54:59kugelZagor: I think that's a reason. The change is rather trivial and it doesn't cost any ram or binsize, it's more consistent and it's just a faster way to access a existing possibility
23:55:54kugelanyway, I see no harm in that. But if you're opposed to that, well
23:56:03 Quit MethoS- (Read error: 110 (Connection timed out))
23:56:54 Quit captainkewl (" ajax IRC Client")
23:56:59 Part akhilleus
23:58:03 Part dany_21a_
23:58:07Zagorkugel: everything has a cost. in this case it adds complexity to tools/configure plus it adds another possible error case when people come asking why their ipod won't boot. what gain does it bring that balances this?

Previous day | Next day