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 2004-10-07

00:00:56 Quit marc77 (Read error: 110 (Connection timed out))
00:00:59amiconnI'll retry with latest cvs code (had a testing directory which is no direct checkout)
00:02:24[IDC]Dragonwhat a comprehensive MMC info you've got
00:02:50amiconnIt's still not everything that is contained in the MMC registers ...
00:03:13[IDC]Dragonthe year has only 4 bits, how short-sighted
00:03:31amiconnDid you notice: You can get info for both internal and external
00:03:56[IDC]DragonI didn't
00:04:21amiconnAh, I remember that you have no working external card :(
00:07:13amiconnArgh, my const'ing causes that dreaded "initialization discards..." warning if PREFER_C_WRITING is defined
00:07:25amiconnHidden bug :/
00:09:30amiconnNo panic with my latest build! :)
00:10:30amiconnWindows chkdsk does find no filesystem error either
00:14:33[IDC]Dragon\o/ Jens \o/
00:15:30[IDC]Dragonthis will make Ondio hacking interesting!
00:19:10amiconnFAT32 still working as well.
00:19:30amiconnOf course I didn't do extensive tests
00:19:55[IDC]Dragonthe test suite doesn't help us here, of course
00:21:22amiconnI could adapt my filesystem stress test plugin (which I originally wrote for checking FAT32 whether it fails when saving recorded data)
00:21:37amiconnThis has to be adapted to the player lcd
00:21:58[IDC]Dragonwhat does it do?
00:23:01amiconnIt writes lots of data (pseudo-random sequence) to disk with a behaviour as close as possible to how the recording code does it
00:23:26amiconn(writing 2 large blocks with a pseudo-random split point)
00:23:45[IDC]Dragonlinear write, not terribly interesting
00:23:47amiconnThen it reads back all data and compares with the pseudo-random sequence
00:24:19[IDC]Dragonhonestly, I don't expect trouble from the unsigned
00:24:30amiconnWell, it's not completely linear, because the FAT is accessed in between
00:24:53[IDC]Dragonif some code only worked because it was signed, it must have been strangely wrong
00:25:14amiconnOkay, perhaps I should simply commit
00:25:46[IDC]Dragonhow much work is it to port your test?
00:26:33amiconnI would have to cut down the output to 2 lines. Shouldn't be that hard, since I use a roll function (restart at top line if bottom one was used last)
00:28:15amiconnBut running it can take a loong time. When I did the test with the recorder (using a little less than 2 GB in total) it ran ~9 hours
00:28:28amiconnOf course I wouldn't test with 2 GB here
00:28:39[IDC]DragonI was about to ask
00:29:45amiconnIt's a nice little stress test anyway. Perhaps I'll run it on the Ondio too, stressing the MMC driver
00:29:54[IDC]DragonI'm building with unsigned ,too
00:30:11[IDC]Dragondon't stress writing too hard
00:30:46amiconnI think it's no problem, since it doesn't write the same sector everytime
00:31:06[IDC]Dragonexcept for the fat
00:31:43amiconnThis is written to more than once, yes, but not that often too.
00:32:24amiconnOf course we'd need plugin support to do the test
00:33:27[IDC]Dragonah, I remember
00:34:21amiconnMy test plugin needs a patch to plugin.[ch]. fsync() is not in the plugin api by default
00:34:45[IDC]DragonOK, forget it
00:35:59amiconnI've still got the .patch file, so it's easy
00:37:22[IDC]DragonI'm starting it on my box now
00:37:37[IDC]Dragon(the build)
00:38:57 Quit PRVSaosn ("( :: NoNameScript 3.81 :: )")
00:39:18 Join ripnetUK [0] (
00:46:26amiconn[IDC]Dragon: Testing on FAT32 with ~100 MB of data...
00:47:18[IDC]DragonI did a manual test session, no problems.
00:47:56[IDC]Dragonrecording, deleting files, bookmarking,...
00:48:30amiconnBoth fat16 and fat32?
00:49:24[IDC]Dragonno, just FAT32
00:49:45[IDC]DragonI'm not worried about a broken FAT16 in cvs
00:50:17[IDC]Dragon(well, no more than we have that now)
00:50:39amiconnYou're right. So when my fat32 test completes without error, I'll commit, then I'll do the fat16 test.
00:51:07[IDC]Dragonbut killing peoples HD file systems would be very embarrassing
00:53:02 Quit plok ("I'm outta here!")
00:53:06***Saving seen data "./dancer.seen"
01:02:42amiconn[IDC]Dragon: You just saved one config bit ;)
01:03:11amiconnANd I have to rebuild the .voice files
01:04:21[IDC]Dragonnot urgently
01:05:25[IDC]DragonOndio can write, hooray!
01:05:34[IDC]Dragonjust dumped the ROMs
01:06:26 Quit _aLF ("Leaving")
01:07:16amiconnRebuilding the .voice files should be easy, but I wonder how out "Klara" will pronounce ".talk" (without a hint, that is)
01:08:11[IDC]DragonIm not sure how to name that entry, in general
01:08:21[IDC]Dragonfeel free to improve
01:09:28amiconnJust found that the german voice: entry is just "Sprachdatei". This should work of course, but it may confuse users
01:09:59[IDC]DragonI think it's fine
01:10:18amiconnI would associate "Sprachdatei" with a .voice file
01:10:26amiconnUnfortunately there is no simple german equivalent to "clip"
01:11:37[IDC]DragonI'll sleep well now
01:11:50[IDC]Dragonand thanks!
01:11:58 Quit [IDC]Dragon ()
01:24:58 Quit mecraw ("Trillian (")
01:53:10 Quit AciD (Remote closed the connection)
01:53:40 Join iRiverMan [0] (
02:02:02 Join ashridah [0] (ashridah@
02:24:13 Quit iRiverMan ("CGI:IRC (EOF)")
02:37:00 Quit amiconn (" Peer didn't reset this connection")
02:53:07***Saving seen data "./dancer.seen"
03:01:23 Quit midk (Read error: 104 (Connection reset by peer))
03:22:50 Join midk [0] (
04:09:05 Join scott666_ [0] (
04:10:29 Quit scott666 (
04:10:29 Quit gromit``` (
04:10:29 Quit Professor (
04:18:24NJoingromit``` [0] (
04:53:11***Saving seen data "./dancer.seen"
05:26:09 Quit ashridah ("gone")
06:20:23 Quit scott666_ ("i'll be back...eventually...")
06:52:21 Join LinusN [0] (
06:53:15***Saving seen data "./dancer.seen"
07:26:34 Join methangas [0] (
07:27:41 Join [IDC]Dragon [0] (
07:30:57[IDC]Dragongood morning
07:31:35[IDC]Dragonwhat a nice one, with FAT16 working :-)
07:32:05 Quit midk ("just STOP it arspy")
07:32:10 Join midk [0] (
07:45:01*[IDC]Dragon has the Ondio back together for the first time, before it was "just" a PCB on the bench
07:45:22[IDC]DragonI stripped the wires off
07:47:09LinusNwhat wires?
07:47:40[IDC]Dragondebug wires ;-)
07:56:31 Join Chronic007 [0] (~Miranda@
08:11:18 Join plok [0] (
08:15:31[IDC]DragonBagder: r u there?
08:17:04*plok is away - Automatically set away. - messages will be saved.
08:30:18[IDC]Dragondunno what he did wrong
08:45:30[IDC]Dragonhey, the OndioFM build fits as rombox!
08:46:53[IDC]Dragonanother 0.3 sec boot time gained
08:47:18 Join amiconn [0] (
08:47:54plokboot time on the iRiver is very long :( (H340 at least)
08:52:11LinusNi heard it takes ages if you use the id3 database
08:53:17***Saving seen data "./dancer.seen"
08:53:35 Join webguest84 [0] (
08:53:47 Quit webguest84 (Client Quit)
08:55:22 Join AciD [0] (
08:55:35plokI don't use the database but it still takes roughly 10 seconds to boot up I'd guess
08:57:31LinusNthe H120 takes quite a while too
09:11:24 Join Zagor [242] (
09:11:48LinusNZagor: welcome
09:12:16Zagormy ribs hurt :)
09:12:33LinusNZagor: i'm having problems deciding how to handle the iriver #ifdefs in for example backlight.c
09:13:01LinusNi'm opting for #ifdef IRIVER_H100
09:14:10Zagoryes, i agree. i think when the conditional is "which circuit board is this", the general player define is suitable
09:14:47Zagorwe can always refine it later if we find the need
09:15:08LinusNwe could have a PLATFORM define with several values
09:15:22LinusN#if PLATFORM == IRIVER_H100
09:15:46Zagorsure, no problem
09:16:06LinusNi'll go for #ifdef IRIVER_H100 for now
09:19:19[IDC]Dragonwe already have PLATFORM_ID, but it's use was discouraged
09:19:39[IDC]Dragonbecause we should check for features instead
09:20:16[IDC]Dragonoh, forget that
09:20:30[IDC]Dragonutter bullshit, I did that locally
09:20:49LinusNyou had me confused for a while...
09:21:06[IDC]Dragonwe have ARCHOS_xxx, which are flags
09:21:11LinusNimho, feature checks are platform specific
09:21:33[IDC]Dragonyes, but you check for the feature,not the platform
09:21:38LinusNbut we should go from ARCHOS_xxx to PLATFORM= instead
09:22:01[IDC]Dragonrecently; ilike multi-valued items, since the # of platforms is growing
09:22:27[IDC]Dragonbecause this is another bunch of flags
09:22:49LinusNmy view of this is:
09:23:06LinusNHAVE_xxx are for general features, like HAVE_BACKLIGHT
09:23:32[IDC]Dragonfor the backlight, I'd say CONFIG_BACKLIGHT = aaa|bbb|ccc for different hookups
09:23:34LinusNPLATFORM_xxx is the platform, eg IRIVER_H100 or ARCHOS_RECORDER
09:24:25LinusN[IDC]Dragon: agreed
09:24:42LinusNbut the CONFIG_BACKLIGHT values may differ between platforms
09:24:52 Join uski [0] (
09:25:07[IDC]Dragonyes, or not exist if no backlight
09:25:14LinusNthen we agree
09:25:17LinusNuski: hi
09:25:41uskihi LinusN :)
09:25:46uskireceived my package ?
09:25:57LinusNuski: I got your stuff
09:26:07 Join Tron|Uni [0] (
09:26:15uskiwhat's up ? long time no see
09:26:40Tron|Unierr, "hi!"
09:27:40LinusNuski: not much, i'm working a lot on the iriver port
09:32:36*Bagder crawls in
09:33:31Bagdernice warning flood yday LinusN ! :-)
09:33:42Bagder1254 ought to be some kind of record!
09:37:19LinusNBagder: :-)
09:39:30 Join pillo [0] (
09:41:01LinusNdoes Dave Hooper pop into this channel?
09:47:38LinusNdoes the H300 have a real time clock?
09:58:05LinusNBagder: u there?
09:58:10 Join amiconn_ [0] (
09:58:23Zagorhe does sometimes
09:58:58LinusNi want to add his scrambling code to our scrambler
09:59:21BagderI'm here
10:00:18LinusNcan we change the config/make system so we can build a target without scrambling?
10:01:17LinusNcurrently it requires a $tool
10:01:48*Bagder looks
10:01:51 Join iRiverMan [0] (
10:03:34BagderLinusN: make the tool named cp for now
10:03:52LinusNcan't do that
10:04:09LinusNit is prefixed with the path to the rockbox tools dir
10:04:40Bagderwant me to fix?
10:05:02*Bagder dives
10:05:37LinusNi'd like to add the iriver scrambling algorithm, but the source is currently unavailabe
10:05:49 Quit ripnetUK ()
10:06:12Zagorthe link is dead?
10:06:20Zagori have it downloaded
10:06:28LinusNDNS doesn't resolve
10:06:38LinusNdo we have his permission btw?
10:07:28 Quit amiconn (Nick collision from services.)
10:07:28 Nick amiconn_ is now known as amiconn (
10:08:12BagderLinusN: try this
10:08:28Bagderor 'try now' rather
10:08:43ZagorLinusN: the license allows it, with a bsd-style banner clause. we can ask him for an exception to the clause.
10:08:54 Quit [IDC]Dragon ()
10:09:43amiconnhi all
10:09:50Bagderhi Jens
10:10:25LinusNBagder: works, but now it UCL-packs.....
10:11:11BagderI didn't change that
10:11:29Bagderthat's a different problem
10:11:33LinusNi know, i didn't get that far before
10:12:05Bagderlemme think a while on how to correct that
10:12:15Bagderunless you have an idea
10:12:17amiconnZagor: Next button issue: If you enter a sub-browser (plugins/ langs etc) from the menu, the first entry gets selected immediately. This seems to be a player-only issue
10:13:27LinusNit shouldn't uclpack targets that don't support flashing
10:13:37LinusNlike the player for instance
10:13:50Bagderthe question is how it is nicest done in the makefile
10:14:19Bagderprobably the target shouldn't be included in the all: line
10:14:32amiconnZagor: I wonder how this happens. I thought the button code does only send *one* button event for button down, then waits until repeat kicks in, then send button|BUTTON_REPEAT events continiuously until release, then sends one single release event.
10:14:42Bagderlike how the ROMTARGET already works
10:15:12BagderI'll work on a fix
10:15:13Zagoramiconn: yes, that's how it works
10:15:55amiconnSo I wonder what causes the sub-browser issue on the player. The button down event gets "eaten" by the menu code entering the sub-browser...
10:17:37Zagoraha. "run" is now play+release, since play+repeat is context menu...
10:17:47Zagorthe puzzle keeps growing :)
10:18:20Zagor(i.e. you see the same problem on recorder if you use PLAY to select the "browse plugins" menu entry)
10:20:32BagderLinusN: ... and update!
10:22:07LinusNBagder: how do i make it not to build plugins?
10:22:24Bagderifdef them in SOURCES
10:22:45LinusNwhat is the plugins="yes" for in configure?
10:22:52Bagderah, right
10:22:56Bagderthat's how I made it
10:22:58Bagderset it to ""
10:23:13LinusNi have tried not to set it, but it doesn't work
10:23:17amiconnZagor: Ah, yes. I did not get that since I usually use Right to enter the browsers. Strangely, it works for Ondio, which has no separate Play...
10:24:08Zagorthat's because the ondio context menu is broken :) (I just noticed)
10:24:28BagderLinusN: odd, it worked for the ondios before
10:24:32LinusNplugins="" doesn't help
10:24:51LinusNit still tries to build the plugins library
10:25:10Zagorhmm, tricky one
10:26:52LinusNENABLEDPLUGINS isn't used anywhere
10:29:02amiconnLinusN: When plugins were disabled for Ondio, it did also still build the plugin lib. Seems that the Makefile for the plugin lib needs a fix
10:32:48BagderLinusN: ENABLEDPLUGINS is used in plugins/Makefile
10:33:03LinusNBagder: check my commit
10:33:04Bagderuse it in lib/Makefile as well, and I think it should work
10:33:23LinusNi think it should be done in apps/Makefile
10:33:51Bagderright, that's probably nicer
10:34:01Bagderthen you can remove it from plugins/Makefile
10:34:32LinusNdoing it now
10:35:36LinusNok, now the iriver target builds ok
10:36:47Bagdergotta run off
10:40:34*LinusN has a fairly safe lead the wiki statistics :-)
10:48:07Chronic007I was noticing that earlier....thanks for all your hard work Linus
10:53:18***Saving seen data "./dancer.seen"
10:53:51 Join PaulS [0] (
10:55:21Chronic007oh-yes, the good iriver port conversation is about to begin
11:04:46 Join hima [0] (~hima_mhmd@
11:05:27 Join MooMaunder [0] (~me@
11:10:29 Join ashridah [0] (ashridah@
11:12:11 Quit iRiverMan ("CGI:IRC (EOF)")
11:26:17 Part hima
11:48:07LinusNPaulS: u there?
12:01:30LinusNany luck with the jtag?
12:01:37PaulSThanks for the tip about pulling up the power supply line. When I hold the system reset, it doesn't power off.
12:02:09LinusNi have added valuable info for you in the schematics
12:02:29LinusNcd ..
12:03:10PaulSI didn't try JTAG again. I did try the BDM interface again, but BKPT still didn't seem to do anything, so I imagine I'm using the wrong pads.
12:03:37PaulSLinusN: Valuable info? Hmm!
12:03:51LinusNcheck the "connectors" schematic
12:04:16LinusNnow you can trace the correct pads from the debug connector
12:04:50 Quit PaulS ("CGI:IRC (EOF)")
12:08:26 Join PaulS_ [0] (
12:09:29PaulS_LinusN: Thanks a lot. This will take the guesswork out of it. So far I only know I had the CPU reset correct.
12:14:26LinusNi hope it helps
12:15:00PaulS_I'm downloading the naked pics as I speak to see how far off I was.
12:15:13 Nick PaulS_ is now known as PaulS (
12:17:16PaulSCongrats on the blinkenleds, BTW, as well as the (presumably not actually functional) rockbox build.
12:17:45LinusNdoes the H300 have a real time clock?
12:17:46PaulSWere you using the drive access light, or the charging LED?
12:19:43LinusNthe backlight
12:20:01PaulSAh! Those are LEDs too...
12:25:58ashridahLinusN: sneaky :)
12:30:42PaulSIt appears that my guesses about the debug pads seem to agree with the connector schematic.
12:30:44*LinusN sends his second coldfire patch to the binutils project
12:32:56dwihnoLinusN: How many of those patches has been sent without any response?
12:33:42LinusNmy very first patch (the windows resource compiler) still hasn't received a response
12:34:04LinusNbut the first coldfire patch has been accepted
12:34:09 Join R3nTiL [0] (
12:36:26dwihnoIt's about giving stuff back! :)
12:42:10PaulSGood night, folks.
12:42:39 Part PaulS ("Z")
12:52:12 Join HuwSy [0] (
12:53:11HuwSylil help, im looking to flash my recorder10 back to the original firmware, whats the easiest way
12:53:20***Saving seen data "./dancer.seen"
12:56:35LinusNrename the original firmware file to firmware_rec.bin and run the firmware_flash.rock plugin again
12:56:57LinusNassuming you did back up the original firmware when you flashed rockbox the first time
12:57:02HuwSyiv lost the original file be the problem
12:57:17LinusNwhich version was it?
12:57:52 Join [IDC]Dragon [0] (
12:58:03 Quit plok (Ping timeout: 14400 seconds)
12:58:30LinusNyou can find the original firmware here:
12:58:52HuwSythank you
12:59:09LinusNit's the "internal_rom_20000...." file
12:59:31LinusNrename it to firmware_rec.bin and put it in the root of your archos, then run firmware_flash.rock
12:59:39LinusNand pray....
12:59:57[IDC]Dragonhe's exaggerating...
13:00:09LinusNseriously, i don't know if 1.28 works on the rec10, but i assume that it does
13:00:46LinusNthe only prob would be the hw mask, but i assume that firmware_flash.rock preserves it
13:00:58[IDC]Dragonyes, it does
13:02:39LinusNthan all is fine.
13:06:46HuwSyhum... its not letting me load up into rockbox to even attempt a flash. i think its overly flat
13:07:10HuwSyoh well ill hopefully get it sorted later, thanx for the help
13:08:59 Quit HuwSy ()
13:11:32[IDC]Dragoneverybody wants a flashback...
13:12:09[IDC]Dragonamiconn: do you read?
13:31:18 Quit R3nTiL ()
13:34:15amiconn[IDC]Dragon: Now I'm here
13:40:30[IDC]DragonI have the flash plugin working now
13:41:50[IDC]Dragonso you could flash yours tonight
13:42:08[IDC]Dragoneven RomBox works for me :-)
13:45:33amiconnSo it should work on the SP for sure
13:46:01amiconnDid you already look into the RoLo issues with the archos fw?
13:47:13[IDC]Dragonno, not yet
13:47:52[IDC]DragonI actually forgot about that issue
13:48:32amiconnThat reminds me that we wanted to add a table of issues to the wiki topic, together with status
13:48:55[IDC]Dragonyes please
14:06:52 Part Chronic007
14:40:44[IDC]DragonBagder: reading?
14:49:51 Join zoevi [0] (
14:53:24***Saving seen data "./dancer.seen"
14:58:10Bagderreading now
14:58:23 Join webguest46 [0] (
14:59:51 Quit zoevi ("CGI:IRC")
15:01:42BagderLinusN: time to add iriver to the cvs build table?
15:01:51[IDC]DragonBagder: how about enabling the Ondio daily build?
15:02:09Bagderme fix!
15:02:33[IDC]Dragonthe download, I mean, with the pretty picture and stuff
15:02:49Bagderthe whole package
15:03:16[IDC]Dragondid Jens send you a picture?
15:03:31Bagderah, no
15:03:33amiconnNo I didn't. Have to close the Ondio first.
15:03:34Bagderonly your pics
15:04:05[IDC]Dragonthose are my FM pictures
15:04:28 Join elinenbe [0] (~elinenbe_@
15:04:56[IDC]Dragonthen use it for both, for the time being
15:05:26elinenbeLinusN: a huge congrats! Great work... everything starts somewhere!
15:06:01amiconn[IDC]Dragon: How did you take the pictures? With the scanner?
15:06:26[IDC]Dragonamiconn: digicam. mine was also not closed for the pic, just placed
15:07:01[IDC]Dragonscanner is even better, if you have one of those thick flatbeds
15:07:42[IDC]DragonI have a thin one with a LED bar, has no depth focus at all
15:08:07Bagderreload daily build page
15:08:41[IDC]Dragonthat was quick
15:08:43LinusNi think adding iriver to the daily builds page is a bit premature
15:08:52LinusNelinenbe: thanks
15:09:04BagderLinusN: it would help us not break it when we modify things
15:09:12Bagderand vice versa
15:09:22LinusNit still needs a patched version of the compiler
15:09:37LinusNand it only compiles crt0.S and backlight.c :-)
15:09:46Bagderatm, yes
15:10:04Bagderoh well, I won't push it
15:10:09elinenbeLinusN: if I have an iriver, can I flash the rockbox firmware?
15:10:11LinusNwhat the heck, add it then
15:10:25elinenbejust to see the backlight flash?
15:10:29BagderLinusN: do we have the patch compiler on the host?
15:10:40elinenbebut, how would I flash it back to the iriver firmware?
15:10:41LinusNelinenbe: yes, but you won't be able to restore the original firmware
15:10:54amiconn[IDC]Dragon: My digicam doesn't make such nice pics as yours. First it's only 2 Mpixel, second it has no macro feature, closest range is ~60 cm
15:10:57LinusNso your player will be a paperweight
15:11:14amiconnI'll try to do it with the scanner @work tomorrow
15:11:17ashridahelinenbe: you'd be fairly screwed without being able to take it apart and plug in a bdm to restore the original firmware
15:11:24LinusNBagder: i sent my second patch to the binutils project today
15:12:06Bagderyou will be the coldfire binutils king! ;-)
15:12:23BagderLinusN: is the gcc/binutils installed/working on labb?
15:12:23elinenbeLinusN: I wouldn't put it up on the daily builds page yet then... that is just asking for trouble.
15:12:40Bagderno daily builds, no
15:12:45Zagoryou will be cold fire binutils helpdesk ;)
15:12:46Bagderbut the cvs build table
15:12:49LinusNelinenbe: only the build status, not any downloads
15:12:55 Quit gromit``` (Read error: 110 (Connection timed out))
15:13:17[IDC]Dragonamiconn: we only need the medium and small size pic, 2 MP from 60 cm should be enough for that
15:13:25LinusNBagder: no m68k compiler on labb, only on my computer
15:13:32BagderZagor: check the daily build log careful tomorrow, the ondios should get built now
15:13:49ashridahwhat does BDM stand for anyway?
15:13:59ashridahbreakout debugging module ?
15:14:00BagderLinusN: aha, care you install it on labb so that we can get it cvs-built?
15:14:01LinusNashridah: Background Debug Mode
15:14:07ashridahthat was my second guess
15:14:12LinusNBagder: ok
15:14:22[IDC]Dragonamiconn: the big one wast just my raw data, before scaling it to about the size I found on the build page
15:16:18amiconn[IDC]Dragon: Another problem with digicam is to get the lighting right, and the neutral gray background
15:22:15[IDC]Dragonamconn: I went on the balcony, placed it on the white backside of a calendar
15:22:31[IDC]Dragons/amconn/amiconn (sorry)
15:22:57amiconn[IDC]Dragon: Use TAB completion ;)
15:23:05[IDC]Dragonthe white balance was still way off, but that can be corrected
15:23:28[IDC]Dragontab doesn't work if there's more people with A
15:23:29amiconnThe balcony approach does only work if you are at home when there is daylight
15:24:20[IDC]Dragonjust make a picture with any light, I'll try to fix it
15:24:40[IDC]Dragondiffuse light preferred
15:24:56 Join gromit` [0] (
15:26:17amiconn[IDC]Dragon: Does your Ondio really have such a strong blue colour? Mine does look more violet...
15:28:44[IDC]Dragonit's a bit more violet than it appears
15:32:38 Quit uski (Read error: 110 (Connection timed out))
15:43:04LinusNBagder: /usr/local/m68k-gcc
15:43:18LinusNpatched with my latest addition
15:47:27[IDC]DragonBagder: I guess tomorrow there will automagically appear something for the download link?
15:47:39Bagderif everything works
15:48:39Bagdeririver h100 added to cvs build table
15:49:19[IDC]DragonI don't see it, mabe after the next commit?
15:50:02BagderI just added it to the script that builds stuff after each commit
15:51:53LinusNand you added the compiler to the PATH?
15:52:12Bagdertime to add the ondios to ?
15:52:39[IDC]Dragonwhy not?
15:53:00Bagderdon't look at me! ;-)
15:53:05LinusN[IDC]Dragon: your job
15:53:16[IDC]Dragonok, not now (no cvs)
15:53:28BagderI'm off
15:56:02[IDC]Dragonwho did the mp3ClipGen.vbs script?
15:56:14[IDC]DragonI fail to remember
15:59:37*[IDC]Dragon googles like crazy
16:07:41amiconn[IDC]Dragon: Search the ml
16:11:23LinusNtime to go
16:11:25LinusNbye all
16:11:40 Part LinusN
16:13:09[IDC]DragonI tried. Brian Wolven is a candidate, but I haven't found a proof
16:16:49[IDC]Dragonhe is, now I found it
16:17:19[IDC]Dragonor? no, this was playlist scripts
16:19:33*amiconn is looking into the script
16:20:29amiconnHmm, no hint in there :(
16:20:30[IDC]Dragonseems to be Brian:
16:23:03amiconnIt does look like that, yes
16:23:22[IDC]Dragonthe voice UI is olter than I thought
16:23:56[IDC]Dragonand v2.2 even more
16:38:28 Join mecraw [0] (~lmarlow@
16:53:26***Saving seen data "./dancer.seen"
17:16:59 Quit webguest46 ("CGI:IRC (EOF)")
17:19:54 Join ripnetUK [0] (
17:20:16*ripnetUK builds a 68000 cross compiler
17:20:49ripnetUKIm trying to cook a m68k cross compiler - do I need to use the cvs to get LinusN's patches in, or does that only apply if im using gdb?
17:27:29Zagori'm not sure what his latest patch was
17:35:48ripnetUKthe onbe im talking about was assembling the 68k equivelent of sse instructions
17:36:10ripnetUKi guess its a bit academic until we have working code anyway :)
17:36:21Zagorthe EMAC, yes. for that you need his patch.
17:42:56 Quit ashridah ("sleep")
17:43:15 Part Zagor
18:14:23 Join marc77 [0] (
18:19:32 Join bagawk [0] (
18:20:24 Part bagawk
18:23:14 Join uski [0] (
18:25:57 Quit [IDC]Dragon ("CGI:IRC")
18:40:50 Part pillo
18:50:51 Join _aLF [0] (
18:52:09 Join Tim[RIP] [0] (
18:52:18 Part Tim[RIP]
18:53:27***Saving seen data "./dancer.seen"
18:57:33 Quit marc77 (" HydraIRC -> <- IRC has never been so cool")
18:59:38 Quit ripnetUK ()
19:05:59amiconnBagder: The channel's topic is a bit outdated...
19:09:08 Quit AciD (Read error: 104 (Connection reset by peer))
19:26:35 Join AciD [0] (
19:30:44 Join [IDC]Dragon [0] (
19:30:56[IDC]Dragonhi again!
19:31:22[IDC]Dragonamiconn: in case you'd like to flash:
19:31:54[IDC]Dragonplus a cvs up-to-date build for the plugins, of course
19:36:43uski[IDC]Dragon, rockbox works on the Ondio ?
19:36:56*uski should try to keep up to date with the latest news :)
19:37:00 Quit AciD (Connection reset by peer)
19:37:36 Join AciD [0] (
19:37:42uskiif i buy an Ondio, when will I be able to run rockbox on it ?
19:38:05[IDC]Dragonuski: now
19:38:42[IDC]Dragonor rather: as soon as you'll get it delivered ;-)
19:38:44uskiyou have support for the NAND flash ?
19:39:00[IDC]Dragonit's an internal MMC
19:39:06uskioh ok
19:39:09uskiso i can be changed ?
19:39:30uskior is it a bulk MMC chip, without the connector ?
19:39:42[IDC]Dragonno, it's a chip that does MMC protocol
19:40:06[IDC]Dragonbut you can use an external MMC, too
19:40:44uskiinteresting.. :
19:40:59[IDC]Dragonor, you unsolder the chip and hook upo an MMC internally, there's enough space
19:41:13uskieven more interesting
19:41:29[IDC]Dragonwould be cool: 1 GB internal, plus n*1GB external
19:41:35uskiexactly :))
19:41:41uskibut a bit expensive today
19:41:50uskibut maybe in a few months prices will be lower
19:42:08[IDC]DragonI'm away now
19:42:12uskioh, by the way, is there a native US english speaker here ? i need some help for my english homeworks ;) only 5 minutes..
19:42:16uskisee you [IDC]Dragon !
19:51:56 Quit [IDC]Dragon ()
20:03:19 Join ripnetUK [0] (
20:12:02 Quit AciD (Read error: 104 (Connection reset by peer))
20:21:01 Quit elinenbe (" Like's GUI? Then try HydraIRC -> <-")
20:29:23 Join AciD [0] (
20:53:31***Saving seen data "./dancer.seen"
21:22:15 Quit ripnetUK ()
21:45:08 Join scott666_ [0] (
22:20:00 Join [IDC]Dragon [0] (
22:20:17amiconnhi again Jörg
22:20:30[IDC]Dragonhi there!
22:20:39[IDC]Dragonflashed? ;-)
22:20:57uski[IDC]Dragon, the flashing guy.
22:20:59amiconn[IDC]Dragon: Nope, not yet
22:21:23amiconn[IDC]Dragon: Some questions beforehand:
22:22:07amiconn(1) Do you also get the strange RoLo behaviour with the archos fw on your FM? (just for comparison)
22:22:19[IDC]DragonI haven't tried
22:22:28amiconn(2) Do you believe it's safe for the SP? You obviously couldn't test...
22:23:04uski(what's the SP?)
22:23:07 Quit methangas (" HydraIRC -> <- irc client ownage!")
22:23:34amiconnuski: There are 2 different Ondio models, see the wiki article and/or archos site
22:23:37[IDC]DragonI think it's safe. the bootloader is identical, I put the original plus Rockbox into the image (one at least should work), and the bootloader has minimon as the last resort
22:24:38[IDC]DragonI can try the SP image, if it comforts you
22:24:55amiconn[IDC]Dragon: That's why I'm asking whether you get the strange rolo behaviour too. If it is so, and archos fw works for you from flash, flashing should be safe for me too
22:25:02[IDC]DragonMAS and tuner shoudn't matter at boot time
22:25:24amiconn(because then the strange behaviour is caused by some init rockbox does)
22:25:34[IDC]Dragonthe rolo'ed version is different from the real ROM one
22:25:58amiconn[IDC]Dragon: If you try the SP image on fm, chances are that it hangs on boot with both firmwares
22:26:20[IDC]DragonI can uart-boot
22:26:50[IDC]Dragonin fact, I have to, because the plugin won't let me use the wrong one
22:26:55amiconn...because the MAS init tries to start the decoding application, but the MAS memory cells to do so have different addresses. So the init waits forever
22:28:04[IDC]Dragonok, so it may hang. was just a suggestion to ease you.
22:28:06amiconnIf possible, just try RoLo into archos fw
22:28:59[IDC]DragonI'll try, but fail to see why this is comforting or not
22:29:36amiconnBtw: I just found a way to speed up mmc writing by 50%, while still polling. Unfortunately this doesn't work for reading, have to implement DMA for speedup
22:29:56[IDC]Dragonwhoo, how?
22:30:15amiconnSending back-to-back by properly switching the SCI mode
22:30:45amiconnI did already perform a quick test, the 50% are real-world, no estimation
22:31:13[IDC]Dragonbut you have a poll loop, with a certain reaction time?
22:31:20uskiooo interesting: it seems to be possible to add a tuner to the SP model, simply by soldering some components
22:31:43amiconnYes, but for sending, I can use the double-buffering of the SCI if it is set to send-only
22:32:25amiconnSo the transfer-loop housekeeping is done in the background, while the last byte is still transferring
22:33:50amiconnJust optimizing a bit further (remembering the last status to avoid switching it every time)
22:34:27amiconnMaybe I don't need DMA for writing at all
22:34:48[IDC]Dragonif you do back to back already, then no
22:36:22amiconnI'm now pretty close to the theoretical max speed for writing. Theory says < ~350 kByte/s, real world is now ~330 kByte/s
22:36:43[IDC]Dragonfor readig, the bitswap loop may poll the DMA pointer, running behind it, for minimum latency.
22:36:58[IDC]Dragon(sorry if you already said that)
22:37:06amiconnAh, nice idea.
22:37:31amiconnMy idea so far was to follow the block count, swapping each block as soon as it is completed
22:38:09[IDC]Dragonfor multiples, yes, also possible
22:38:32[IDC]Dragonbut gives a 1 sector swap latency at the end
22:38:48amiconnYour approach might deliver the lowest latency, but the block-wise approach may even allow to yield() once
22:39:01amiconnSo it's not blocking the whole box
22:40:32[IDC]Dragonrolo'ing Archos s/w hangs at about 60% of its progress bar
22:40:59amiconnOoops. This is different from the SP.
22:41:05[IDC]Dragonhow about one yield() and then running behind the DMA?
22:41:36amiconnI could also yield every time the bitswap has catched up
22:42:12[IDC]Dragonbut that may happen close before the end
22:42:36amiconnOr even better, yield every time the bitswap has caught up, but only if > n bytes to go
22:43:57amiconnProblem: The DMA is chunked into block size anyway, with some necessary polling inbetween
22:44:12[IDC]DragonI think if we do it at the start, we're more sure to reach it again at the end, then stick as close as possible
22:44:25[IDC]Dragondo the uncertain part first
22:45:12[IDC]Dragonwhat block size do you have in mind?
22:45:35amiconnYes, but a complete transfer might run for several seconds. Just yielding once would not help much
22:45:35amiconnThe block size is fixed, 512 bytes
22:45:35[IDC]Dragonah, Ok, I was thinking in sectors
22:46:01[IDC]Dragon(confused again)
22:46:25[IDC]Dragonif you do multi-block, it can be quite long?
22:46:32[IDC]Dragonin one go
22:46:49[IDC]Dragonwith no further command
22:46:59amiconnEach data block comes in with a start byte in front, which has to be polled for (worst case: 10 ms with the internal flash), the 512 bytes of data, then 2 bytes crc
22:47:51amiconnMulti-block transfer is unlimited with MMC, you send a start command, and the card starts transferring block after block until you send a stop command (reading) or a stop token (writing)
22:48:00[IDC]Dragonok. Then we can poll for start, set the DMA, yield, run behind bitswapping.
22:48:15[IDC]Dragonsequence repeats for each sector
22:49:12uskihave a nice technical talk ;)
22:49:39amiconnFor sector-wise, I think its better to use this sequence: poll for start, set DMA, bitswap previous block, yield, wait for end of DMA
22:50:38amiconnIf the thread itself should do the run-behind bitswap continuously, the poll-for-next would have to be done in the ISR
22:53:33***Saving seen data "./dancer.seen"
22:54:26[IDC]DragonI have to leave now
22:54:27 Quit uski ("Leaving")
22:54:38[IDC]Dragon(like uski)
22:54:52 Quit [IDC]Dragon ()
22:54:55amiconnnite [IDC]Dragon
23:20:14 Quit AciD (Read error: 104 (Connection reset by peer))
23:20:45 Join AciD [0] (
23:44:13 Quit gromit` ("Client exiting")

Previous day | Next day