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 2005-05-23

00:00:01Bagderrather naive
00:00:38preglowand stupid, considering how easy it would have been for them to add an usb check before they access the disk
00:00:55preglowthey've got about two seconds worth of splash screen that could be used for checking quite thoroughly, as a matter of fact
00:01:12LinusNthis is interesting btw:
00:02:20preglowi'd say
00:03:38amiconnLinusN: The file size is odd. My viewer.rock is 19060 bytes...
00:03:59LinusNthat depends on the compiler version
00:04:04 Join rasher [0] (
00:04:07crashd /win 22
00:04:22rashersomeone was saying that the snow.rock wasn't updated in the daily builds either
00:04:32amiconnLinusN: Yeps... but it shouldn't deviate that much within the same line of gcc (3.3.x)
00:04:42preglowrasher: big on log-reading, aren't you ;)
00:05:14rashernah, just read it once in a while
00:05:36LinusNBagder: is the zip creation script updated for the new dir structure?
00:05:55preglowafter you fixed IRiverBoot for me yesterday without even being on the channel, i've started attributing you with paranormal powers
00:05:57Bagderat least it is supposed to
00:06:32preglowworks here
00:06:37preglowi've got working snow and all
00:06:41preglowand i did 'make zip'
00:06:50amiconnThe filke dates in the dailies match the day they're supposed to be built
00:08:05*rasher tries using a daily build
00:09:00Bagdercheck the daily source package
00:10:08LinusNBagder: the daily build zips contain the wrong .rock's
00:10:28BagderI think I know why!
00:10:38Bagdermake clean doesn't remove the old ones!
00:10:44LinusNthey are taken from build-recorder8m/*.rock instead of build-recorder8m/apps/plugins
00:11:04LinusNthat shouldn't matter
00:11:08Bagderyes it does
00:11:13Bagderit finds the old ones too
00:11:35LinusNand copies the over the correct ones
00:11:43Bagderbut make clean is lame
00:11:47Bagder'rm -rf *' is better
00:11:56LinusNmuch better :-)
00:11:59LinusNin the root
00:12:04preglowthat's not funny
00:12:13 Join matsl [0] (
00:12:18BagderLinusN: well, its running as Zagor's users ;-)
00:12:47preglowi usual try to do it as root
00:12:49preglowmuch cool
00:13:05LinusNBagder: can you remove the old files?
00:13:06preglowdamn, what's up with this premature entering habit of mine
00:13:34BagderLinusN: you mean the old dailies?
00:13:36 Quit matsl (Remote closed the connection)
00:13:48 Quit ashridah ("Leaving")
00:17:03LinusNno, the old rocks
00:17:18Bagderthe script will actually do rm -rf * first now
00:17:21Bagderinstead of make clean
00:17:34Bagderthen run configure
00:17:36Bagderthen make
00:17:39Bagderthen make package
00:20:13rasherWhy wasn't this a red build:
00:20:38Bagderlame parser? ;-)
00:20:54rasherFair enough
00:22:55Bagderand I'm too tired atm to do anything about it ;-)
00:24:08rasherI can understand this
00:25:53rasherI found a thread on MisticRiver mentioning someone who had his h320 display replaced by a repair shop (not iRiver related)
00:27:21preglowand i thought the iriverlinux thread was all that was of value on misticriver :P
00:28:30Bagderthat is interesting
00:29:29rasherHe wasn't mentioned by name though, so I fired of a private-message to him
00:30:32rasherI guess there's a thread somewhere about it
00:39:50 Quit Harpy (Read error: 60 (Operation timed out))
01:08:54rashertotally failing to find this thread though
01:12:31 Join cYmen_ [0] (
01:12:50***Saving seen data "./dancer.seen"
01:13:43 Quit cYmen (Read error: 104 (Connection reset by peer))
01:17:36 Part MoosK1maro
01:22:40*preglow goes w00t at the newest mailing list post
01:26:21rasherOkay, I found it
01:26:38rasherBut the only info in that thread is the name of iRiver's repair shop
01:27:12rasherpreglow: looks like you have yourself a companion
01:27:40preglowi surely hope so
01:27:47preglowi wish i had more info on tremor to give him
01:27:56preglowonly opt i can find quickly is windowing
01:28:14preglowbut there should be a dot product somewhere, if i recall the vorbis specs correctly
01:30:54rasher"CMS Peripherals" is the name of the repair shop they use in the UK, it seems
01:31:00 Join textchimp [0] (
01:31:23preglowhow does this help us?
01:31:37rasherI wonder if it'd be possible to ask them which screen is in the h3x0
01:32:43rasherI just don't like to think that this searching through that forum was useless
01:33:52preglowsure, mail them ;)
01:38:22 Quit DMJC (Remote closed the connection)
01:46:42 Join ]RowaN[ [0] (
01:47:43]RowaN[ive just reformatted my irivier h120... when i try to copy some files back onto it it says "error: the path is too deep" .. not sure what it means, any ideas guys?
01:48:58LinusN]RowaN[: the total file name, including path, must not exceed 256 characters
01:49:14LinusNit's a FAT32 limit
01:49:51]RowaN[hmm, it never mattered before, stupid internal iriver format
01:50:09]RowaN[ahah i just tried again, with the file it wouldnt let, and its done it with no probs now
01:52:02preglowsomeone put fat32 to sleep already
01:53:54LinusNi think fat32 is pretty ok
01:53:58]RowaN[hmm windows is telling me "crc error" when i try to copy mp3s to my iriver now... i dropped it the other day you see.. i think parts of the hd r fukked
01:54:06]RowaN[but scandisk says its all ok oddly
01:54:15LinusNit might be the usb
01:54:25]RowaN[how do you mean?
01:54:35preglowthen again, there aren't any other windows compatible file systems with specs out there
01:54:55]RowaN[ATA error: -71
01:54:57]RowaN[says rockbox
01:55:29]RowaN[wont turn off now... lets find a pin and press reset =]
01:55:31 Join Bryan` [0] (
01:55:36 Part Bryan`
01:55:48textchimpso, uh, what can i do to help get sound working on the iriver port? my coding skills are mediocre but subject to improvement through stubbornness
01:55:51LinusN]RowaN[: how is your battery?
01:56:01]RowaN[and plugged in
01:56:25]RowaN[says *panic* ata: -71 now
01:56:29LinusNtextchimp: there's not much you can do right now, i'm afraid
01:56:32preglowtextchimp: unless you're leet at optimizing codecs, not much, i guess
01:56:44textchimpthat's what i figgered
01:57:15LinusN]RowaN[: weird
01:57:36LinusN]RowaN[: have you seen the "check hdd connection" message with the iriver firmware?
01:58:05]RowaN[ive just reset,, gonna try again copying files
01:58:36]RowaN[ah its getting a bit further now
01:59:12]RowaN[people... let this be a lesson... HOLD ON TO YOUR HD PLAYERS DONT DROP THEM =p
01:59:34textchimpi've sure dropped mine a few times
01:59:45textchimpdid you have yours in its little protective leather jacket?
01:59:49LinusN]RowaN[: have you dropped it?
01:59:50]RowaN[nope =[
02:00:02]RowaN[yeah last week it fell out of my inside pocket onto concrete!!
02:00:15]RowaN[bah again...
02:00:16LinusNopen it up and reseat the hard drive
02:00:21]RowaN[ata error: -1 press ON to dubug
02:00:23textchimpgotta put it in that thing, it's probably saved mine a few times
02:00:42]RowaN[reseat the hd? theres no where for it to move
02:01:30]RowaN[yup =]
02:01:39preglowi'm too tired to read straight
02:02:03LinusN]RowaN[: it doesn't take much to cause a bad hdd connection
02:02:08preglowthe connection might be loose
02:02:13LinusNit's definitely worth a try
02:02:18preglowit's really simple to jerk it back in place
02:02:22LinusNit has worked for ithers
02:02:31textchimpwhat tools do you need to open it up anyway?
02:02:42textchimpit has a weirdass screw doesn't it?
02:02:43preglowa small torx screwdriver
02:02:44amiconnMrf. Either gcc is braindead, or I don't understand wut's going on :-/
02:02:44LinusNa tiny torx screwdriver
02:02:45]RowaN[a few days ago i opened it up.. hd connection looked fine.. just incase i pulled the hd away then reslotted it
02:02:52preglowamiconn: like what?
02:03:07LinusNamiconn: i'm betting on #2
02:03:09amiconnI'm trying to get rockbox to work (on archos) with -O2 / -Os
02:03:11]RowaN[textchimp: same set of tools that opens my t68 mobile fone =]
02:03:30amiconnIt always crashes with IllInstr at 0x000000F0
02:03:36amiconnThat's the internal rom
02:03:42preglowi really don't think that's gcc's error, it happens with both sh and 68k
02:03:48amiconnI checked the disassembly
02:03:50preglowi've never known gcc to generate wrong code with Os
02:04:16amiconnThe code it produces for the IRQ3 interrupt handler is looking completely wrong
02:05:14amiconnNo 'rte' at the end, but instead a 'jmp'
02:06:03amiconnadditionally, it seems the if/else processing therein is missing as well
02:06:19preglow]RowaN[: at least the drive they use in h1x0 is supposed to be able to take a bit of a bang before it's damaged
02:08:18preglowLinusN: should i commit the imdct opt i've got going even though it's still a wee bit slower than the old one, btw? i won't be able to look at it again for quite some time
02:08:44]RowaN[when i dropped it, i "lost" about 100 mp3s.. they just became unreadable from the hd
02:08:56preglow]RowaN[: bad news...
02:09:27]RowaN[if it wasnt for rockbox i woudlnt have even been able to turn my iriver back on
02:09:37]RowaN[coz original firmware hangs on boot when files r fucked
02:09:39preglowoh well, assuming the hd is broken, they're easily replacable
02:09:50]RowaN[cool.. where from, how much, how big =] sells 30 and 60 gig hds now
02:10:16LinusNpreglow: yes i think you should commit it
02:10:37preglowLinusN: ok, i'll see about it tomorrow
02:11:00preglowbut i think i'll call it a day for now
02:11:01preglowlater, all
02:11:15LinusNsleep tight preglow
02:11:31preglowtight, well and not too long
02:11:35 Quit preglow ("leaving")
02:12:47LinusNi should get some sleep too
02:12:50LinusNnite all
02:13:06 Part LinusN
02:27:09 Quit rasher ("CGI:IRC")
02:27:40amiconnGcc clearly generates wrong code with both -Os and -O2
02:27:55amiconn(for the IRQ3 interrupt handler)
02:31:52 Join silencer_ [0] (
02:31:52 Quit silencer (Read error: 54 (Connection reset by peer))
02:38:43 Join silencer [0] (
02:52:38 Quit cYmen_ (Read error: 113 (No route to host))
02:55:54 Quit silencer_ (Read error: 110 (Connection timed out))
03:05:23 Quit ]RowaN[ ()
03:12:52***Saving seen data "./dancer.seen"
03:13:16 Join amiconn_ [0] (
03:21:48 Quit amiconn (Read error: 60 (Operation timed out))
03:21:49 Nick amiconn_ is now known as amiconn (
04:00:23 Join BTKDaImMaikata [0] (~Bager@
04:00:24 Quit Bager (Read error: 104 (Connection reset by peer))
04:18:33 Join silencer_ [0] (
04:34:38 Quit silencer (Read error: 110 (Connection timed out))
05:04:24 Join courtc [0] (
05:08:20 Join ehntoo [0] (
05:12:54***Saving seen data "./dancer.seen"
05:29:11 Quit ehntoo ("Leaving")
05:45:25 Quit courtc (Read error: 110 (Connection timed out))
05:45:36 Join courtc [0] (
07:12:58***Saving seen data "./dancer.seen"
07:20:34 Quit Aditya ("Download Gaim:")
07:51:34 Quit asdsd____ (Read error: 104 (Connection reset by peer))
07:53:55 Quit Sucka ("a bird in the bush is worth two in your house")
08:01:49 Join LinusN [0] (
08:26:40 Join ashridah [0] (
08:46:58 Quit edx (Read error: 110 (Connection timed out))
09:13:02***Saving seen data "./dancer.seen"
09:15:02 Nick Lynx_awy is now known as Lynx_ (HydraIRC@
09:17:47 Join Harpy [0] (
09:20:38 Join DMJC [0] (
09:40:43 Join Lost-ash [0] (
09:40:46 Quit ashridah (Nick collision from services.)
09:41:01 Nick Lost-ash is now known as ashridah (
10:03:43 Join preglow [0] (
10:09:07 Quit Stryke` (Read error: 104 (Connection reset by peer))
10:09:40 Join Stryke` [0] (~Chairman8@resnet-241-86.resnet.UMBC.EDU)
10:12:30LinusNi wonder if we can find a suitable connector that can fit into the 6-pin connector of the remote control
10:12:57ashridahLinusN: for testing, or plugging in alternate objects?
10:13:16LinusNrs232 debug output in my case
10:13:35LinusNalternate objects is always nice, of course
10:14:28preglowLinusN: oooh, that sounds nice
10:16:14LinusNi have the code, and i also have a connector
10:16:26LinusNfrom a dead remote control
10:16:55LinusNbut maybe others would like to use it
10:22:31 Join bobTHC [0] (
10:23:17bobTHChi folks !
10:25:10amiconnhi everyone
10:26:53 Join edx [0] (
10:28:05RickLinusN: that's what *I* was thinking
10:28:12RickI mean, for doing custo stuff
10:28:22Rickwould be nifty
10:28:51Rickalthough when I talked about it myself here ages ago I didn't get a very positive response
10:36:42amiconnLinusN: I got rockbox to work when compiled with -O2 or -Os. However, gcc clearly generates wrong code for IRQ3(), and it needs a special workaround to get this fixed
10:36:54LinusNhow nice
10:37:14amiconnCompiling with -O3 doesn't work; it fails to assemble the compiled system.c
10:39:02amiconnI tested with recorder v1. -Os decreases size ~6 KB, -O2 increases size ~9 KB (gcc 3.3.5, size with -O ~184 KB)
10:42:22amiconnThe workaround is adding an empty asm block at the end of the function: asm volatile ("");
10:45:27amiconnThere's another change I did, but don't know if it strictly necessary. The bit setting macros (and_b / or_b / xor_b) aren't defined volatile. Iiuc they should be; so I added that
10:45:53amiconnLinusN: What do you think?
10:51:29 Join B4gder [0] (
10:51:48B4gdergood day
10:52:32amiconnpreglow: I found this trick from googling; obviously this is an architecture independent gcc bug. The article I found was for netbsd on ppc (!)
10:53:28LinusNamiconn: well, if it works i'm all for it, then the rombox builds could be -Os
10:54:04LinusNbut then we *might* encounter some problems, since some bugs might be related to the optimizing
10:54:16LinusNand we'll have an even harder time reproducing them
10:55:05amiconnThis would require compiling everything twice. Currently the rombox builds are only linked differently
10:55:14LinusNof course
10:55:34amiconnThe only target where it would actually allow to make rombox fit (again) is the recorder v2
10:55:55 Join [IDC]Dragon [0] (
10:56:09amiconnFor the others (fmrecorder and ondiofm) even -Os isn't sufficient
10:56:10LinusNif you ask me, rombox isn't worth the effort, but that's me
10:56:40[IDC]Dragonhi there
10:56:41amiconnRombox is really nice to have, but I'd rather go for [IDC]Dragon's solution
10:56:45B4gderI still for making rombox a special build option
10:56:50B4gderI am
10:56:50amiconn(bootbox instead of archos fw)
10:57:07LinusNamiconn: agreed
10:57:12[IDC]Dragonsmall&fast code is always nice...
10:57:24amiconnLinusN: I meant what's your opinion about adding 'volatile' to the bit macros
10:57:40LinusNamiconn: go ahead
10:57:41[IDC]Dragoncurrently, our copile flags are "lame"
10:57:53amiconnhi [IDC]Dragon
10:57:59[IDC]Dragonhi Jens
10:58:12amiconnI finally solved the 0308 mystery...
10:58:23[IDC]Dragonyes, saw that
10:58:50amiconn...and found a mistake in your logic table. I also added the pinout & table for 0308
10:58:53[IDC]Dragon(I had a hard time to blame the rebuild for it)
10:59:16[IDC]Dragonsorry for the logic error
10:59:59amiconnThe circuit is indeed very similar, only it uses an AC32 instead of AC08 and 2 gate inputs are swapped (pin 12 & 13)
11:00:01[IDC]Dragondoes Trevor know already?
11:00:38amiconnI mailed him, but he didn't respond yet. Maybe he's away, mentioned some conference...
11:01:01[IDC]Dragonseems to be his facourite waste of time ;-)
11:01:13amiconnI also experimented with the mas pll
11:01:30[IDC]Dragonblowing up poor 3507 chips...
11:01:47amiconnNope; the fm interference thing you mentioned
11:02:01amiconnThere s indeed some interference
11:03:01amiconnIt seems I understand the mas pll now. It is indeed running at 24.576 MHz (for48/32/24/16) and 22.5792 MHz (for 44.1/22.05)
11:03:23amiconn...the 4th harmonics being ~98.3 MHz and 90.3 MHz
11:03:48amiconn..and the interference is audible with the samsung tuner (!)
11:05:06amiconnI implemented a hack to avoid the interference, which works
11:05:16[IDC]Dragonsamsung doesn't surprise me
11:05:38amiconnThe philips seems to be immune agains this, so no special handling necessary
11:06:17[IDC]Dragona slight "detune" of the MAS should be sufficient
11:06:25[IDC]Dragonwhat did you do?
11:07:17amiconnYes. I use the pitch function to 'pitch away' the mas by a maximum of 0.3 %
11:07:55amiconnAs is the solution is hacky because it accesses app data (global_settings) from firmware code
11:09:01[IDC]Dragonwhy global_settings?
11:09:15amiconnTo find the current rec_frequency setting
11:09:43[IDC]Dragonthe FM screen could inform the mpeg code
11:09:56amiconnI have another idea
11:10:06[IDC]Dragonleaving it up to there to do countermeasures
11:10:23RickLinusN: How would you "probe" to see if the remote is plugged in? (iriver)
11:10:32amiconnmpeg_set_recording_options contains a variable which could be made static, and an access function added
11:10:35preglowyou got fm interference from the mas clock?
11:10:58amiconn[IDC]Dragon: I added the countermeasure to tuner_samsung.c
11:11:02amiconnpreglow: yup
11:13:06***Saving seen data "./dancer.seen"
11:14:06LinusNRick: i think we could try to change the direction of one of the inputs and see if the port status changes
11:14:16LinusNor something like that
11:14:20Rickah the detection for ide power controllability in the player
11:23:00amiconn[IDC]Dragon: I also tried to exceed the pll offset range for 3507. That doesn't work; it seems the pll snaps back to the default
11:29:54[IDC]DragonLinusN: did you ever try/consider pitching the 3507?
11:42:03amiconnIt is possible to pitch, but only in the small range I mentioned saturday
11:43:51ashridahah, good, i'm glad someone's starting to look at tremor optimisations again
11:48:15preglowi'm glad someone's looking at optimisations at all
11:48:23preglowhope he has the will to start working on it for real :P
12:02:32 Join MoosCamaro [0] (
12:02:42MoosCamaromorning all
12:04:15preglowgood day
12:05:45preglowdamn, my battery was _completely_ discharged, the player didn't even respond when i pressed play
12:05:51preglowwas a bit nervous for a second there
12:09:47preglowold imdct_l is 300% at 112kbps, new is 290% :/
12:14:06preglowLinusN: we probably should use something like the .cachetest section scheme we do now
12:14:26preglowi do now, at least, i'll strip it before commiting
12:45:13preglowwhy doesn't my player ever turn itself off?
12:46:06preglowidle poweroff is set to 10 minutes, but it never turns off
12:46:59LinusNis the charger connected?
12:47:26preglowyes, but it behaves the same without the charger connected
12:47:33preglowthat's why i have to charge now, ehh
12:47:47LinusNmaybe the charger detection fails
12:47:55ashridahLinusN: yeah, i've noticed this too
12:48:14*LinusN checks the code
12:52:56HClit actually did crash for me
12:52:58HClwhen it ran out of battery
12:53:01HClthe other day
12:53:03HCland it hadn't shut down
12:53:18ashridahLinusN: hrm. i just set it down to 1m and it shutoff.
12:53:47preglowyeah, i think rasher mentioned it only happens sometimes
12:54:01preglowis it conditional in any way? the last thing i usually do before i don't touch it anymore, is run a plugin
12:54:02ashridahi think i've seen it happen if i've just used usb-connect mode
12:54:09preglowperheps something doesn't get reset
12:54:19preglowthat can also be it
12:55:51ashridahokay, just turned it back on, had it set to 1m, and it didn't shut off after 1m this time around
12:56:00LinusNthere are lots of conditions for the poweroff
12:56:11ashridahactually, it seems to have taken longer than a minute, since it just shut off
12:56:22ashridahat at least over 2 minutes
12:56:28LinusNdisk activity, user interaction, music playback
12:57:06LinusNthe 1m timeout starts counting after the disk has spun down
12:57:28ashridahwhich is set to 5s here
12:58:04LinusNi guess some debugging is in order here
12:58:31 Join asdsd____ [0] (
12:58:39 Part asdsd____
13:00:04ashridaha 1m shut down took almost 2 minutes from hard drive spindown to shutdown.
13:00:29ashridah(on a fresh powerup)
13:00:35LinusNoh, there's another thing
13:01:49LinusNthe thread that performs the auto poweroff runs only once a minute
13:02:15ashridahso it can be anywhere from 1m to 1m59s or so?
13:02:16LinusNso it can be one minute off
13:05:11LinusNthe logic is in firmware/powermgmt.c - handle_auto_poweroff()
13:13:07***Saving seen data "./dancer.seen"
13:14:23Zagornow that we have many contributors, perhaps someone feels like taking on the powermgmt rewrite?
13:14:51B4gderthe Zagor lives!!! :-O
13:15:10*B4gder curses yacc
13:15:31Zagorbagder: Why do you say the zagor lives?
13:16:15B4gdersorry, I must've typed Mx- doctor
13:16:22B4gderM-x even
13:16:30ZagorIs it because of your plans that you say sorry you must've typed mx doctor?
13:16:43*Zagor loves doctor
13:16:59B4gderM-x psychoanalyze-pinhead !
13:18:17ZagorFUN is never having to say you're SUSHI!!
13:21:04B4gderI'm building qt/e on my powerpc setup... and it fails on the FIRST file it compiles :-(
13:21:30HClmmm qt/e
13:21:35HCli have that on my pda
13:22:39B4gderI've built and used it before, it is neat
13:23:38 Quit textchimp (Read error: 54 (Connection reset by peer))
13:27:47 Join tucoz [0] (
13:28:18tucozHi, for you swedes. An archos jukebox exploded when charging:
13:29:24ashridahwas the battery's integrity compromised by something?
13:29:46LinusNand of course he didn't have a backup...fool!
13:30:38tucozthe article suggests that the battery might have had a small damage :)
13:32:56amiconnI don't understand the whole article, but my guess is that it was a fm or v2 recorder.
13:32:56amiconnThere must be a reason why I don't exactly like LiIons...
13:32:56 Join cYmen [0] (
13:33:49Zagorit was a gmini 400
13:34:42preglowarticle says it was a archos jukebox
13:34:45ashridahan aussie kid here managed to make his ipod burst into flames by a) running his ipod through the was, then b) opening it up and piercing the battery with a screwdriver (tool!)
13:34:56B4gderwatch the picture
13:35:25Zagorpreglow: this is a tabloid. they don't care about fact accuracy.
13:35:25amiconnReminds me of [IDC]Dragon's fm recorder...
13:35:27ashridahi have an ultra-old liion battery that's currently going through a stage of expanding
13:35:47tucozBagder, didn't the article say that the player exploded ;)
13:35:55preglowB4gder: archos replaces old jukesboxes with gminis
13:36:37ashridahahaha. the media here tried to claim that the kid's ipod actually did explode. all it really did was pop, burn his bed, and give him a nice dose of nasty smoke+fumes
13:36:37Zagorgood point
13:36:49ashridahso all the headlines were 'ipods go boom!'
13:36:54HClheh o.o
13:37:04ashridahnah. Australia
13:37:11HClokay o.o
13:37:15B4gderyou see, all that moden music is bad for you ;-)
13:37:27Zagormodem music?
13:37:43B4gderbe nice, I can't type ;-)
13:37:44ashridahyou read the article, and they all say "well, not exploded, it went pop, and scorched his bed, but he had to be treated for smoke inhalation! that's bad, right?! oh, and ambos are heros"
13:37:56tucozwe should be protected from it. Let's suggest a new law
13:38:47ashridahyeah, we all clearly need to revert to the stone age
13:38:51 Join sox [0] (
13:38:53ashridaheverything was far safer then
13:39:15ashridahnone of this radiation crap, hardly anyone died from cancer (the average age being about 30 or so)
13:39:36soxhey all
13:39:58soxwhat's this error Im getting on a new box im setting up trying to compile the rockbox tools dir: ../firmware/export/config.h:23:22: autoconf.h: No such file or directory
13:40:29tucozI get this error as well
13:41:03HClmeans you haven't configured properly.
13:41:11soxconfigured what
13:41:15ashridahit gets created by ../tools/configure
13:42:03tucozbut when I run ../tools/configere, make, I get an error that conf-something is missing.
13:42:18HClwhat error?
13:42:20tucozand I then run make in tools
13:42:39tucozand get the autoconf.h error
13:42:58ashridahconvbdf and friends aren;'t being compiled automatically if they've been cleaned atm
13:43:20soxso you have to run make convbdf?
13:43:24ashridahtucoz: after you've run 'make' in the tools directory, does the firmware build successfully?
13:43:45B4gderif there's a autoconf warning when building tools, something is wrong
13:43:50ashridah(nevermind the autoconf.h error in /tools/
13:43:51 Quit lostlogic (Connection reset by peer)
13:43:55 Join lostlogic [0] (
13:43:58tucozHCL,make[1]: /home/martin/src/rockbox/rockbox-devel/tools/convbdf: Command not found
13:44:03ashridahB4gder: well, building the firmware by itself doesn't even go into tools
13:44:09B4gderI know
13:44:20B4gderI removed that :-)
13:44:52B4gdersince I'm on a quest to lower the amount of sub-makes within makefiles
13:45:11ashridahso when is tools supposed to get built? :)
13:45:23tucozashridah, yes everythings working after the tools/make
13:45:38B4gderI'd prefer a separate 'make tools' I think
13:45:39HCli say tools should get build when someone tries to configure and the tools haven't been built.
13:46:38B4gderyes, sounds fair
13:46:56ashridahand nothing in the tools directory should depend on autoconf :)
13:47:49B4gderright, that's a bug
13:49:26tucozso the autoconf error in tools/make is not because I'm doing something wrong then?
13:49:28ashridahkill it, KILL IT!
13:49:36B4gdertucoz: no
13:49:42B4gderI don't think so
13:50:11tucozB4agder: good :)
13:50:13 Quit BTKDaImMaikata (Read error: 104 (Connection reset by peer))
13:50:41 Join Bager [0] (~Bager@
13:53:03tucozhmm, should mkboot.c be built when running make in the tools directory?
13:53:17tucozit doesn'n for me
13:54:01ashridahtucoz: because it comes after the one that's attemptign to use autoconf i'd say
13:54:15tucozah, ok
13:54:41ashridahrun make mkboot in tools
13:54:56tucozyes I have used to do that
13:55:48tucozIt's after this that make bails out: In file included from ../firmware/drivers/lcd-player-charset.c:19:
13:57:23preglowis there any point in doing .aligns before asm subroutines all the time? i would have thought the linker handled this
14:00:18 Join TCK- [0] (
14:02:49LinusNpreglow: no, i don't think so
14:04:00 Part tucoz
14:04:03amiconnThe linker does only handle this for the whole section
14:04:20 Quit TCK (Read error: 60 (Operation timed out))
14:05:20B4gderthe autoconf.h in tools dir problem should be fixed now
14:06:30preglowwell, coldfire will toss an exception if it's fed code that isn't aligned
14:06:39preglowat least easy to discover
14:07:59pregloware CONFIG_CPU and the rest of those fellows available in makefiles?
14:08:30amiconnOn SH, code is required to be word aligned. However, there are instructions for which it makes sense to long align them avoid bus contention between instruction fetch and data access
14:09:18preglowi was thinking of adding -DASO_IMDCT to MADFLAGS for coldfire, what's the best way of doing that from the makefile then?
14:09:30B4gdertucoz: you build from cvs?
14:09:46LinusNpreglow: config-h100.h
14:10:02 Quit Seed (Nick collision from services.)
14:10:09 Join Seed [0] (
14:10:24preglowLinusN: ahh, goodie
14:10:26soxpreglow: getting close to committing the optimized mad stuff....? :-) we're getting closer sound!
14:10:59B4gderlinus had a great idea we discussed yday: having the remote lcd for debug logging
14:11:00LinusNpreglow: you can add it to the global CFLAGS too
14:12:11preglowLinusN: adding ASO_IMDCT to config-h100 feels cleaner
14:12:21B4gderwith an 8 pixels font, we get 8 lines of text and we can have like 500 or 1000 lines scrollback
14:12:39preglowsox: nothing to say hurray about, it will actually get slower with this update
14:13:02preglowsox: libmad should already be ready for sound
14:13:12preglowB4gder: hey, that _is_ a good idea
14:13:28B4gderwe could switch it on/off in the debug menu
14:13:40HClyea, that sounds great o.o
14:14:13*HCl works on writing tiny test programs for his and markuns programming language
14:14:25HClany suggestions? i got fibonacci, faculty, sum, interest, power..
14:14:37HCl :P
14:14:51HClit only has variables, if then else, while, and repeat until
14:14:51HCl :P
14:15:07 Quit sox ("Snak 4.13 IRC For Mac -")
14:15:47LinusNB4gder: we could have a general debug log in memory
14:16:07B4gderand display it when you enable the remote-lcd for it
14:16:10LinusNand we can choose to watch it live with the remote, or afterwards on the main lcd
14:16:18LinusNor both
14:16:24B4gderthis is just getting better!
14:16:34LinusNor save it on disk
14:16:36B4gderwe should add an lcd window to the sim
14:16:37HCl :P
14:16:42B4gderremote lcd even
14:17:00B4gderI guess that "we" means me ;-)
14:17:15LinusNDBGLOGF() or maybe just LOGF()?
14:17:24B4gderI vote for plain LOGF()
14:17:37LinusNgets my vote too
14:17:58B4gderor possibly even logf()
14:18:11LinusNthen we can just have a log viewer in the debug menu
14:18:12amiconnImho this should be disableable with a #define
14:19:34amiconnIt could also show the DEBUGF() messages
14:19:37preglowi agree with all that's been said
14:19:45LinusNamiconn: not sure about that
14:19:48amiconn...having 2 flavours of debug builds
14:20:21amiconnOne would log messages to remote lcd and memory, and another one using a gdb stub
14:21:20LinusNi suggest the remote lcd option is only a way of watching the memory log live
14:23:54B4gderyes, sounds good
14:25:41 Join tucoz [0] (
14:26:23tucozBagder: the tools/make is working great now. And yes, I'm building from cvs
14:26:57*tucoz away
14:33:05 Join textchimp [0] (
14:47:08 Quit lostlogic ("Going to the moon")
14:59:06B4gderAnd remember: X Windows is to graphics hacking as roman numerals are to the square root of pi (quote from sim/x11/screenhack.c)
14:59:44amiconnB4gder: true...
15:00:25preglowi LOVE how linux sound playback SKIPS when i plug the h120 in
15:00:38*LinusN runs off
15:00:45preglowgo, linus, go
15:00:49 Part LinusN
15:03:31amiconnpreglow: It does? Hmm, windows doesn't do that
15:04:55 Join webguest53 [0] (
15:05:53 Part webguest53
15:06:28 Part tucoz
15:06:56preglowamiconn: i know
15:07:13HCldoesn't that have to do with your soundcard?
15:07:18preglowamiconn: i also suspect only this particular linux machine does that...
15:07:24preglowHCl: why should it?
15:08:21preglowdoes rockbox do crc checking on mpeg frames on archoses?
15:08:45HClsome soundcards handle computer load better than others
15:08:54 Join ahezhara [0] (
15:09:08preglowthe sound card shouldn't care
15:09:50preglowusb insert doesn't add load anyway
15:09:58preglowat least it shouldn't
15:10:02HClon linux, apparently it does
15:10:05HClotherwise it wouldn't skip
15:10:17preglowi really don't think load is the problem here
15:10:23Zagorpreglow: you run xmms?
15:10:33preglowlinux screwing up badly on insertion is the problem
15:10:38preglowZagor: no, vlc right now
15:10:46preglowZagor: i've yet to find a music player i'm satisfied with...
15:12:35 Quit ahezhara (Client Quit)
15:13:09***Saving seen data "./dancer.seen"
15:13:19ashridahpreglow: i don't have that kind of problem here and my buffers are decidedly short (about 50ms or so, which, admittedly, is large), are you using alsa or oss?
15:13:47ashridah(also, if you've got a bunch of hotplug scripts, things can go nuts spawning processes when stuff gets plugged in, iirc
15:13:56ashridahone of the reasons people keep going "screw this, lets rewrite hotplug in c)
15:14:05B4gderthe remote lcd code doesn't check for simulator
15:14:17*B4gder digs in
15:14:23preglowno idea, really, whatever vlc uses
15:14:25preglowprobably oss..
15:14:39preglowhotplug isn't c?
15:16:11 Join ahez [0] (
15:16:32ashridahpreglow: nah, many implementations have it built with a bunch of scripts
15:16:44ashridahthere have been various pushes to rewrite it
15:16:52 Join Aditya [0] (
15:17:04preglowyeah, my system is just a bunch of sh scripts
15:18:40 Quit ahez (Client Quit)
15:20:03preglowbut yeah, does rockbox do crc checking on mpeg frames on archoses?
15:20:31B4gderrick: you here?
15:20:32 Join Lost-ash [0] (
15:20:56 Quit ashridah (Nick collision from services.)
15:21:02 Nick Lost-ash is now known as ashridah (
15:21:15amiconnpreglow: ?? On archos, all mpeg encoding/decoding is done by the mas
15:22:54preglowyeah, i know
15:23:07preglowjust wondering if you've got any error handling on incorrect crc in mpeg frames
15:23:12preglowdisabling it in libmad gives a tiny boost
15:23:43B4gdercan't you enable/disable this in the mas?
15:23:46preglowpersonally i don't see any point in performaning crc checking on a portable, if i can't hear an error in an mp3 file, i don't care about it anyway
15:23:57preglowperformaning, that's a new onre
15:24:07amiconnB4gder: For encoding, the mas allows to select whether to generate crc or not
15:24:21B4gderah, ok. must've been that
15:24:22textchimpare you guys talking about codec optimizing for the iriver? damn that's exciting, this is GROUND ZERO
15:24:48preglowdon't know if i'd call it exciting...
15:24:58amiconnWhen decoding, the mas mutes the output when it encouters crc errors
15:25:05amiconnIirc this is not configurable
15:25:30preglowright now i'm battling with iram, i've just got too little
15:26:01 Quit [IDC]Dragon ("CGI:IRC")
15:26:09textchimpthis is open source firmware development IN ACTION
15:26:54preglow88 bytes too much :/
15:28:04B4gderdo we really need a lcd_remote_getstringsize ?
15:28:12B4gderto me that is just odd
15:28:50amiconnB4gder: It's not odd.
15:28:51B4gderwhat does it do?
15:29:07amiconnMain & remote lcd may have different fonts in effect
15:29:09Rickit gets the width/height of a string in the current font on the remote?
15:29:11 Join Aison [0] (
15:29:26amiconnThe actual code is shared
15:29:58B4gderthat is odd to me too ;-)
15:30:08B4gderI thought we were going for "true" multifont support
15:30:22B4gderand then you'd use stringsize for a given font
15:30:36B4gdernot assuming a particular one for a particular lcd
15:31:19amiconnNot *assuming* but using the currently selected one
15:31:35B4gderso we can select one for the remote ?
15:31:52amiconnWe have two fonts, FONT_SYSFIXED and FONT_UI
15:32:07amiconnYou can set both main and remote lcd to use either one
15:32:07B4gder"setting" a font is weird if a screen uses multiple ones
15:32:22amiconnIt isn't weird
15:32:31B4gderno, possibly not
15:32:59amiconn...much like it makes sense to set draw modes and colours, draw some objects, set other parameters etc
15:33:31amiconn...instead of giving all these parameters with each and every call
15:34:03B4gderI'm not talking about every call here
15:34:21B4gderbut I see your point
15:35:16B4gderrick: you tried the new remote lcd size in the code?
15:36:33preglowbah, to hell with this
15:39:06RickB4gder: nope
15:39:14RickB4gder: it was fixed?
15:39:29B4gderthe size might be 128x64
15:39:36B4gderwhich explains some oddities ;-)
15:40:02Rickthe size is 65x132
15:40:08Rickor something like that
15:40:13*Rick can't remember the height
15:40:22B4gderthe iriver product site says 128x64
15:40:34Rickhow odd
15:41:05RickI wonder why the lcd specs say 65x132 (again, whatever the height is ;P)
15:41:20B4gderwhat lcd specs is that?
15:41:30Ricklemme find it
15:42:18Rickyeah, 65x132
15:42:38Rick(the iriver remote has the lcd on its side and upsidedown)
15:42:46amiconnThe lcd itself is 64x128, 99% sure
15:42:56Rickthat's odd
15:42:58amiconnThe controller can handle max. 65x132
15:43:13amiconnIt's the same as with the archos recorders and Ondios
15:43:24Rickit doesn't mention anything about the 64x128
15:43:36amiconnThe lcd itself is 112x64 there, the controller can handle a maximum of 65x132
15:43:44amiconnSeems to be a common size...
15:44:08amiconnThe iriver homepage says 64x128
15:44:28*Rick shrugs
15:44:29amiconnI'll check myself as soon as my iriver arrives
15:44:31Rickif it works, it works ;)
15:46:03 Join kurzhaarrocker [0] (
15:46:55amiconnhi kurzhaarrocker
15:47:17*kurzhaarrocker hasn't implemented triggerd paused recording yet :)
15:48:55kurzhaarrockerwow, it seems there's a whole bunch of Bagder clones around
15:48:56amiconnwas about to ask that ;)
15:49:48B4gderwell, Bager is not me, and Bagder is my client at home
15:49:52B4gderthis is me at work
15:50:07B4gderdoes logo.rock move the remote logo too now?
15:51:47kurzhaarrockerV2.4 is nearly half a year old by now...
15:51:54*B4gder has a simulator window with a remote part now
15:52:16RickB4gder: oh, nifty.
15:52:29B4gdershould make it easier to write stuff for the remote
15:52:48B4gderapps I mean
15:53:41amiconnB4gder: X11, Win32, or both?
15:53:52B4gderx11 only so far, but I'm not done yet
15:54:12B4gderI think I've fixed the dreaded sim dep problems too
15:54:17amiconnWin32 will need a modified background image as well
15:54:32amiconnDoes the x11 sim employ a second window?
15:54:52B4gderno the remote LCD simply appears below the regular one in the same window
15:55:14amiconnHmm. That'll look a bit odd, imho
15:55:28B4gderI'd find it annoying to have two windows popping up
15:55:55amiconnHmm. I'm not sure. Do you use some kind of frames and/or spacing?
15:56:13B4gderthere are lines around the LCD areas
15:58:12 Join oxygen77 [0] (
15:58:12amiconnDoes the simulator allow to "unplug" the remote?
15:58:37 Quit kurzhaarrocker (Remote closed the connection)
15:59:28B4gderdon't expect everything at once ;-)
16:02:45amiconnThe win32 sim needs some work too. There's a bug report... maybe I should look at it
16:05:36 Quit oxygen77 ("CGI:IRC (EOF)")
16:12:46 Quit ashridah ("Leaving")
16:18:23 Quit thegeek_ (Read error: 110 (Connection timed out))
16:20:02B4gderoh. action in the openneo camp
16:25:53preglowthe source code tree is essentially identical to rockbox
16:27:03 Join thegeek [0] (
16:42:47B4gderthey wanted their own race
16:45:41B4gderwe helped them get started, they took the code and ran
16:46:02B4gderalthough we wanted to support theiri player too
16:50:22B4gdergoing home
16:50:24 Quit B4gder ("CGI:IRC")
16:50:24preglowtoo bad
17:13:11***Saving seen data "./dancer.seen"
17:19:18 Join webguest36 [0] (
17:22:43 Join Christopher^ [0] (
17:23:12 Part Christopher^
17:29:25 Quit webguest36 ("CGI:IRC (EOF)")
17:32:38 Join rasher [0] (
17:35:15rasherAny objections against putting cpu_boost calls in jpeg.rock and mandelbrot.rock ?
17:35:37rasher(not that they run without the grayscale patch yet)
17:38:32preglowwell, do they perform badly without it?
17:38:46rasherjpeg is sortof sluggish
17:38:54rasherand I seem to recall mandelbrot being it as well
17:39:04rasherdepending on the size of the jpeg of course
17:39:18rasherand it's only for the duration of the calculations of course
17:39:28rasherI cpu_boost(false) afterwards
17:42:36rasherHaving a grayscale.iriver is sweet
17:43:05 Quit Zagor (Remote closed the connection)
17:43:19rasherand the matching versions of jpeg, mandelbrot and rockboy
17:44:19rasherAh, I think last I tried adding cpu_boost to those two were when the iriver was still running at 11mhz
17:44:32rasherdoesn't make much of a difference now
17:46:52preglowi wouldn't expect it to
17:46:57preglow49mhz is pretty much for those applications
17:47:41rasherfor a 3000x4000 jpeg, there's no real difference
17:48:01preglowsomeone finish the grayscale patch, please :-)
17:48:12*rasher points at amiconn
17:48:31preglowhe hasn't even got a h1x0
17:48:44rasherHe will
17:54:10preglowwell, i certainly wouldn't complain .P
17:54:37rasherhe's already planned it all
17:57:11rasherI like the idea of a debug buffer
17:57:53amiconnMeh, I want my irivier *now*
17:58:09rasherWhen's it expected?
17:59:50amiconnI have no real idea, but I hope it'll be here towards the end of this week
18:01:30rasherthe "view battery" debug options seems a bit strange on iriver
18:01:37rasheror else it just reports oddly when charging
18:04:55amiconnrasher: It seems the seller already got the money, since he already rated me, so the iriver should be on its way... :)
18:04:59 Quit Bagder ("Off to search for that connect-resetting peer guy!")
18:05:13rasheror else he took the money and ran
18:05:30rasher(probably not)
18:05:36rasherI'd hope not
18:05:48rasherAnd it's not like he's from Nigeria or anything
18:05:51 Join Bagder [0] (
18:06:52 Join webguest09 [0] (
18:07:02webguest09what is the login to access to cvs
18:07:12webguest09i have an authentification error
18:07:31rasherAnd an empty password
18:08:25 Quit webguest09 (Client Quit)
18:14:59preglowpeople on ebay try avoid doing that, i'd imagine
18:14:59rasherhrm, you have to boot with the remote to use it?
18:15:07preglowthink so
18:15:17HCluse what?
18:15:19rasherAh, sad.
18:15:23rasherWell to draw to it
18:15:43*Bagder sees two rockbox-logos bounce in his simulator
18:16:03rasherMy two cents regarding that: I'd want them to be two windows
18:16:15Bagderthen feel free to make it do that
18:16:26BagderI don't ;-)
18:16:44rasherYou win.
18:17:13Bagderthis is a first shot at least, we can always improve
18:17:13*rasher gets shocked by an usb cable
18:17:28rasherthat's a first
18:19:47rasherwouldn't it be awkward in the windows sims?
18:20:10BagderI've not enabled this in the win sim
18:20:14Bagderonly made them build fine
18:20:39BagderI'd rather let someone else do the win magic
18:20:50rasherwhich reminds me, didn't I do the logo remote thing stupidly
18:20:58BagderI've corrected that
18:21:07Bagderit flies in my version ;-)
18:22:01rasherI should just check for HAVE_REMOTE_LCD, I was checking for IRIVER_H100 && !defined(SIMULATOR) wasn't I?
18:22:30rasherI was going to fix that
18:22:31BagderI also make it use the defines from the config file
18:22:37Bagderfor width and height
18:22:40rasheryes, it was a horrible mess
18:22:44HCl :)
18:23:03rasherDidn't really intend to commit it at first, just test if the remote would click
18:23:54rasherplease commit this fix
18:24:18Bagderjust did
18:24:31Bagderwith the rest of the sim improvemtns
18:27:04preglowwhy should they be two windows?
18:27:06preglowwhy does it matter?
18:27:23rasherit doesn't much
18:27:30rasherI'd just expect it to be two independent windows
18:27:36rasherall this snow!
18:27:37HClas long as the screens are properly seperated, there shouldn't be a problem.
18:28:05rasherbut for the windows sims, which use images of the device, I think it'd look corny to have the remote tacked on
18:28:11rasherbut then, I don't have to usse it, so
18:28:11HCli guess
18:28:15HCl :P
18:28:17HClsame :P
18:29:24 Join LinusN [0] (
18:29:33preglowand i don't care how corny it looks
18:29:43LinusNpersonally, i don't give a sh*t how the simulator looks, as long as it works
18:29:49preglowwhy, then we agree
18:29:58LinusNit is a tool, not a piece of art
18:30:21Bagderit _is_ art, but art doesn't have to look nice ;-)
18:30:24LinusNi'd throw away those silly bitmaps any day, but people seem to want them
18:30:36rasherI agree on that
18:31:35LinusNjust my 0.02, gotta run
18:31:37 Part LinusN
18:39:51 Quit I-R-Addict (Read error: 110 (Connection timed out))
18:51:01 Join Bluechip [0] (
18:52:04Bluechiphi, i have a quick question ...does "time split recording" miss anything or not? That is, will there be three seconds of music missing as it stops and restarts? The manual contradicts itself on this point :(
18:54:20Bluechip"">no audio is lost at the split point. The
18:54:21Bluechipbreak between recordings is only the time required to stop and restart the
18:54:21Bluechiprecording, on the order of 2-4 seconds."
18:54:46rasherseems clear enough to me
18:54:55rasherI think
18:55:01Bluechipso no time is lost, or 2-4 seconds is lost?
18:55:14rasherNo time lost, but restarting takes 2-4 seconds
18:55:40rasherBut then, I've never used it
18:56:03Bluechipso I need to deduce: what does "restart" mean ...hmmm
18:56:23Bluechiprestart recording
18:56:27Bluechiprestart disk
18:56:31Bluechipmanual restart
18:56:33Bluechipauto restart
18:56:34BagderI don't think there's any audio lost
18:56:36rasherIf I understand right, it means creating the new file etc
18:56:38BluechipIM CONFUSED - lol
18:56:57Bluechipi suppse it can buffer during that time
18:57:16rasherthat's my guessing
18:57:39Bluechipcheers rasher
18:57:55rasherBut let me re-iterate: I've not used recording ever
18:58:09Bagderme neither
18:58:17Bluechipoh yeah man, no holding you to anything, just showing gratitude for taking the time to assist
18:58:40Bluechipit record quite well (someties)
18:58:52rasherbagder: the sim seems to be working fine
18:59:18Bagderthe build table was good to me as well
18:59:20Bluechiptoday it spewed out a garbage file - which is why i want to split tonights recording - it spits out some crap and then everything after that is lost
19:07:22 Part Bluechip
19:10:50 Quit bobTHC ("Smoke Weed Every Day !")
19:13:13***Saving seen data "./dancer.seen"
19:29:22 Quit R486 (Read error: 110 (Connection timed out))
19:34:54 Quit Seed (Nick collision from services.)
19:34:59 Join Seedy [0] (
19:40:16 Nick Lynx_ is now known as Lynx_awy (HydraIRC@
20:04:31 Join Aison` [0] (
20:05:14 Quit Aison` (Read error: 54 (Connection reset by peer))
20:21:22 Join I-R-Addict [0] (
20:25:53 Join webguest93 [0] (
20:26:56 Part webguest93
20:34:45 Join LinusN [0] (
20:35:06LinusNfor the record (no pun intended), the time split doesn't lose a single frame
20:35:34amiconn...and it probably doesn't help to avoid garbage files
20:36:28amiconn... _because_ it doesn't lose frames. It doesn't restart the mas
20:37:18LinusNit just continues dumping garbage in a fresh new file :-)
20:38:17amiconnMy guess is still that the mas "runs away" somehow with a certain combination of high quality settings
20:39:26amiconnYou voted for committing the alternative recording routines, right?
20:39:35 Quit TCK- (Read error: 104 (Connection reset by peer))
20:39:39LinusNyes i did
20:39:47preglowi hate computers
20:39:57amiconnLinusN: Okay, then I'll do that
20:41:11LinusNpreglow: sure you do
20:43:08 Join TCK [0] (
20:45:36preglowthere's not enough cache left for me to commit imdct, btw
20:45:48preglowi lack 88 bytes, or so
20:46:50LinusNthen change the file
20:49:36 Join pbvas [0] (
20:50:01 Quit thegeek (Read error: 104 (Connection reset by peer))
20:50:10preglowsure, but how much to steal?
20:50:52preglowsounds good
20:51:15preglowi'll whip together some food, then do it
20:54:55*HCl bangs his head against the wall over computer cluebies.
20:55:03HCl"how do i make a folder?"
20:57:14amiconnLinusN: I have a design question / suggestion.
20:57:29LinusNamiconn: shoot
20:57:50amiconnCurrently the menu functions (and some others) have a return value that indicates whether they return from usb mode or not
20:58:17LinusNi'd like an enum or bitmask for that
20:58:24amiconnIf they return true, the whole menu 'pile' returns, and the dir is reloaded
20:58:44amiconn(used to jump back to the root, but now a reload of the current dir is tried first)
20:59:22amiconnThis mechanism is 'misused' by the bookmark menu to leave the menu after successfully selecting a bookmark
20:59:49Bagderit was an enum... once upon the time
21:00:00amiconnHowever, since this return value was originally meant to indicate usb connection, it has an unwanted side effect
21:00:34amiconnWhen calling the bookmark menu while in the browser, all is fine. Bookmark is started and wps displayed
21:01:37amiconnHowever, if you call the bookmark menu from wps and select a bookmark, the playback is still started properly and the menu is left, but the wps is left too (!)
21:01:51amiconn...because it 'thinks' there was an usb connection
21:02:36amiconnYou can still re-enter the wps by pressing On (or Mode on ondio), but this is rather irritating
21:03:15LinusNhow is it done in the fm preset menu?
21:03:18amiconnImho the return value should indeed be an enum of some kind, indicating the reason to leave
21:03:31LinusNi agree 100%
21:03:35Bagderme too
21:03:47LinusNi have wanted to do that for a long time
21:03:56LinusNi actually did once, but i never committed it
21:04:04Bagderthat's the reason I made it so initially, but Zagor didn't agree back then
21:04:07*preglow gets his resetting screwdriver
21:04:55preglowLinusN: anything else i need to take care of after having altered the iram size?
21:05:21LinusNchange the iram size in firmware/
21:05:35amiconnLinusN: Iiuc from a quick glance, the fm menu uses a menu callback handler to fake a MENU_EXIT keypress
21:05:51LinusNah yes
21:06:00*LinusN runs away for a few mins
21:06:41pbvasHi preglow, LinusN: I'm playing around with the Tremor optimzations...
21:06:59HCltremor optimizations
21:07:02pbvasnot very sucess at the moment.. stil 28% of real time...
21:07:22pbvasTremor is the Ogg vorbis interger decoder
21:07:49pbvaspreglow, I'm getting a weird error msg from gcc regarding the IDATA_ATTR
21:07:51BagderHCl: don't you read your mail properly? ;-)
21:08:00HCli do
21:08:02HClbut its a lot.
21:08:02preglowpbvas: really? what does it say?
21:08:08HCloh. that.
21:08:16pbvaserror: bitrev causes a section type conflict
21:08:29preglowwhat have you tried putting in iram?
21:08:55pbvasI thought i only had to declared with the IDATA_ATTR tag...
21:09:08HClBagder: i'm not really a fan of email :)
21:09:27preglowpbvas: in theory, that should be enough, but what variables have you tried putting in iram? and are you sure they aren't too big?
21:09:34preglowpbvas: currently, only 32kb of iram is usable
21:09:37BagderHCl: then you miss a lot
21:09:50HClif you say so :)
21:13:15***Saving seen data "./dancer.seen"
21:16:38LinusNpbvas: "section type conflict" means that you didn't declare it with IDATA_ATTR in all places
21:16:52amiconnLinusN: Different thing. The context save/restore functions in thread.c rely on inlining. Is there a reason why they are functions in the first place?
21:17:04amiconnMacros would guarantee inlining, and TCC730 actually uses macros.
21:17:05LinusNif you have an "extern" declaration in a header file, it must be IDATA_ATTR as well
21:17:33amiconnIf there is a reason for these to be actual functions, we could use __attribute__ ((always_inline))
21:18:24LinusNamiconn: i don't think they need to be functions, but always_inline sounds nice to me
21:18:35preglowLinusN: the array in question is static
21:22:16amiconnLinusN: The always_inline was added in gcc 3.1, so it should be okay. Should I add it for iriver as well?
21:26:22CoCoLUSgeneral question
21:26:57CoCoLUSwhat happens when you call an inline function recursively?
21:27:27preglowit doesn't get inlined
21:27:42CoCoLUSthat easy?
21:27:58preglowLinusN: this is weird, the line that gets a section type conflict is a variable declaration, and it's static...
21:28:27amiconnpreglow: is it initialised or bss?
21:28:34preglowamiconn: initialised
21:29:41amiconnCoCoLUS: Yes, that easy. The top level might be an inlined instance, but all recursive calls are made to the non-inlined version
21:29:42preglowi do this all the time in libmad
21:29:47preglowand it works just fine
21:31:10 Quit pbvas ("Leaving")
21:32:45 Join pbvas [0] (
21:32:47CoCoLUSdoes rockbox have a "max size" restrictions of some kind? some space in which the image must fit?
21:32:56LinusNi can put bitrev in idata without problems
21:33:11LinusNCoCoLUS: which platform?
21:33:36preglowwe'll see about bitrev when pbvas changes are put in cvs some day
21:33:39preglowunless he fixed it
21:33:41preglowfixes <-
21:33:57CoCoLUSnothing specifiy... i just wondered why inline functions aren't used more often, i thought they only increased the final filesize...
21:34:20pbvasLinusN, i'll have to get it more optimized before i commit anything ...
21:34:20amiconnLinusN: Got my q?
21:34:42LinusNamiconn: i think we can do it for all targets
21:35:31amiconnHmm. One small issue. I would declare the functions always_inline for all targets, but then the build process might produce warning(s) for gmini
21:35:32preglowCoCoLUS: i don't think inlining helps much for the rockbox platform
21:35:47amiconn...because the functions are declared static, but not defined
21:36:24amiconn...unless I convert the macros blindly
21:43:09preglowit still bloody hangs
21:44:08 Join matsl [0] (
21:46:37preglowi changed IRAMSIZE in and
21:47:12 Quit courtc (Read error: 110 (Connection timed out))
21:47:20LinusNand also the start address of iram in
21:47:28 Join thegeek [0] (
21:47:41 Join courtc [0] (
21:48:13preglowno, didn't do that
21:48:49preglowstill can't see why i'd need to do that, wouldn't that eat into the stack?
21:49:28LinusNhow else are you going to give more space to the plugins?
21:49:56LinusNyou take from the app and give to the plugins
21:49:59preglowwell, stack seems to be 32kb of iram, plus 32kb of iram used for plugins thus far, which makes 65kb out of 96
21:50:28CoCoLUSlinus, it's likely your least favourite question right now, but any progress on mp3 playback / sound arch?
21:50:48 Quit pbvas ("Leaving")
21:51:07preglowit's much more fun when it's a surprise, you're spoiling it for yourself! allocates 0x10000000-0x1000ffff and the plugins 0x10010000-0x10017fff
21:52:29LinusNso to give more iram to the plugins, you need to lower IRAMSIZE in
21:53:09LinusNand then lower IRAMORIG and increase IRAMSIZE accordingly in
21:53:54preglowyep, got it
21:54:35 Join muesli- [0] (
21:55:13preglowbut i'm very sceptical as to the wisdom in commiting something that's both slower and requires more iram
21:55:27preglowit's more elegant, at least ;)
21:55:31LinusNpreglow: i kind of agree
21:55:43LinusNi think you should be the judge
21:55:43 Quit textchimp (Read error: 54 (Connection reset by peer))
21:55:49preglowthen i'll be keeping it
21:55:58preglowthis is too much of a bother for no gains
21:56:05preglowi don't think anyone but me is going to tune it anyway
21:56:08LinusNpreglow: put it in the wiki, for reference
21:56:10preglowso might as well keep it private until then
21:56:18LinusNor in the patch tracker
21:56:40LinusNi'd like to play with it to see why it's slower
22:01:54 Quit TCK (Read error: 110 (Connection timed out))
22:05:13preglowbut the special section for heavily cache dependent functions is still a good idea
22:08:30 Join tvelocity [0] (
22:13:28amiconn2 new warnings for gmini, as expected :(
22:14:02rasheram I reading things wrong, or is there no straight-forward way to set a pixel to a specific "colour"?
22:24:21 Join Musicmad [0] (
22:24:51amiconnBagder: The build system does something I didn't expect
22:25:05amiconnIt uses the source dir(s) as working dir for gcc when you're compiling with -save-temps for analysing a problem, it clutters the source tree
22:27:11MusicmadI'm trying to get the # Devel - m68k-elf-binutils (latest version is 2.15)
22:27:11Musicmad# Devel - m68k-elf-gcc (latest version is 3.3.5) stuff from the mirror but it says "incomplete download" any ideas?
22:27:41amiconn2.15 is too old anyways
22:27:58amiconnYou'll need 2.16 (or a not too old cvs checkout)
22:28:16Musicmadhm ok - where do I get that?
22:28:58amiconn lists all you need
22:30:04amiconnUh oh! I tried building rockbox for recorder v1 with -O3 (I found the problem why it didn't work yesterday)
22:30:21amiconnThe binary actually becomes too large!
22:30:47Musicmadthanks - amiconn, I just wanted to change 2 characters in the source :)
22:31:33amiconnWhat do you want to do? I thought you want to install the dev environment?
22:32:38Musicmadyes that's what I wanted to do. Seems a bit difficult though, so it might not be worth it tonight.
22:32:54Musicmadbut thanks anyway
22:35:45preglowamiconn: i don't think O3 opts will help rockbox much
22:38:42amiconnYes, I think you're right. It's an experiment, and it just 'feels good' if rockbox compiled with various -O is working correctly
22:40:23amiconn-O, -Os and -O2 are working now. Can't test -O3 on recorder v1
22:40:38amiconn(working on archos, that is)
22:41:23preglowO3 is O2 plus all opts that increase code size, i think
22:41:30preglowmore inlining
22:50:15amiconnThe -O3 build crashes.... after some stuttering playback
22:51:21preglowwell, while you're on the mission to make rockbox work with all gcc options, make it work with gcc4 :>
22:52:05amiconnIt's impossible on archos... due to that silly decision of the gcc team
22:52:29preglowwhich is?
22:52:45amiconn...not to allow (weak) aliases to functions not defined in the same .c
22:53:12amiconn...which also doesn't allow aliases to functions defined in an asm block within the same .c
22:53:36preglowis there no way to write around that? i think i remember you saying the coldfire port did it in a better way once
22:54:16amiconnIt's an architecture problem
22:54:48amiconnIiuc coldfire saves both exception type and return address on the stack, so they're easily accessible
22:55:10amiconnSH1 only saves the return address, so we need a way to get the exception type
22:55:43amiconnThis is done by having a number of tiny functions, one per exception, which all call the same main function
22:56:13amiconnThese functions are defined within an asm block, and then aliased
22:58:48 Quit matsl (Remote closed the connection)
23:06:05amiconnWrong code in interrupt routine with -O3 ...
23:06:23amiconnIt seems gcc doesn't like interrupt routines
23:13:19***Saving seen data "./dancer.seen"
23:13:37 Part LinusN
23:31:45 Quit Aison ("( :: NoNameScript 3.72 :: )")
23:37:11 Join ashridah [0] (
23:38:53 Quit muesli- ("ich will Kühe!!!")
23:43:44Bagderamiconn: tried the new makefile setup for the sim?
23:44:00Bagderit seems to fix deps and links things in the proper order now
23:44:19amiconnNot yet
23:44:38amiconnI'm still experimenting with different -O
23:44:47Bagderno hurry
23:45:02amiconnCurrently building gcc 3.4.4 for SH
23:51:48 Join webguest51 [0] (
23:52:05 Join Sucka [0] (
23:57:05 Quit thegeek (Read error: 104 (Connection reset by peer))

Previous day | Next day