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 2009-07-20

00:00:43 Join dmb [0] (n=Dmb@unaffiliated/dmb)
00:09:08 Nick AfterDea1h is now known as AfterDeath (n=icxcnika@freenode/weird-exception/network-troll/afterdeath)
00:09:36 Join ademille [0] (
00:11:43 Quit petur ("Zzzzzz")
00:14:16*gevaerts thinks that maybe the build clients should send an email address along when connecting
00:14:32Zagorthat might be a good idea
00:14:46pixelmahmm... how should that work?
00:14:53 Join mc2739_ [0] (
00:15:06pixelmasilly me, I overlooked the "address"
00:15:10ZagorI've already had to dig up one person's email to ask that he look at his client
00:15:11bertrikhm, we don't really have a well-established way to tell codecs to turn on their analog input for fm radio it seems
00:15:18 Quit mc2739 (Nick collision from services.)
00:15:19 Nick mc2739_ is now known as mc2739 (
00:15:34 Join Nikkado [0] (
00:15:37bertrikfaking it by calling it 'monitoring' or something like that
00:16:08 Join stripwax [0] (
00:17:18mcuelenaereZagor: could you add a VX747 sim to the build table? (there are no Onda sims being built currently)
00:17:18 Join robin0800 [0] (
00:17:59Zagormcuelenaere: you can do it too. just add a line to www/buildserver/builds
00:18:25Zagorif you ask nicely maybe gevaerts will do a benchmark build for it, too :)
00:18:28*gevaerts runs off to find the appropriate score
00:19:07mcuelenaereZagor: ah, but still don't you need to update the build server? (eg do svn up)?
00:19:19Zagorno it does that by itself
00:19:26mcuelenaereah nice :)
00:19:43 Quit scorche (" rawr...that is all...rawr")
00:20:40 Join scorche [50] (n=scorche@rockbox/administrator/scorche)
00:21:13gevaertsmcuelenaere: your magic number (a.k.a. score) is 13810
00:21:26mcuelenaereis that the last column?
00:21:36*mcuelenaere already wondered about those
00:23:26 Quit m67_l3 (Read error: 110 (Connection timed out))
00:23:30CIA-69New commit by mcuelenaere (r21973): Add Onda VX747 simulator build
00:27:09pixelmawhat's up with the bootloader yellows (rasher's cygwin server and pyros-roolku)
00:27:35roolkupyros is cygwin as well
00:27:40gevaertsZagor: another wild and probably stupid idea : maybe the score cutoff should take into account which arches the client can build
00:30:36 Join m67_l3 [0] (
00:31:09Zagorgevaerts: yeah, it makes sense.
00:32:11gevaertsit does make some sense. The question is if we want to complicate things for that, and if we want to encourage selected-arch clients
00:32:49roolkupixelma: maybe it wants an explicit cast?
00:34:41saratogastripwax: if you get a chance, could you profile cook ? I'd help mt figure out where best to optmize it
00:35:27 Quit shotofadds ("Leaving")
00:36:05Zagormc2739: is your net connection crappy? I get ping timeouts and lots of very slow ping responses from you.
00:37:05mc2739Zagor: yes I have been having problems with my provider
00:37:47 Quit Nikkado (Read error: 110 (Connection timed out))
00:38:46mc2739Zagor: if it is causing problems, I can shut down my client until they get it fixed
00:38:58 Quit bmbl ("Bye!")
00:39:19Zagorno it's fine. I just wanted to make sure it's on your side.
00:39:36Zagorsome "unstable" clients is actually good for testing
00:41:21stripwaxsaratoga - where would I find some example files?
00:41:27stripwaxI don't use realaudio at all
00:41:29saratogastripwax: i'll upload mine right now
00:41:48stripwaxI won't be able to profile tonight −− maybe tomorrow evening at the earliest
00:41:55saratogathats fine
00:42:14CIA-69New commit by roolku (r21974): cast explicitly to unsigned int (fix yellow on cygwin)
00:42:34mc2739gevaerts: re: microsd problem on boot - it seems to work correctly on usb connect
00:42:52saratogawhenever you have a chance will be fine, I would just like some idea where the codec uses its time
00:42:59stripwaxamiconn - with your recent mdct_arm.S changes, what did you use to measure speedup (test_codec, or some tight loop, or something else)?
00:43:07saratogain the meantime there are plenty of things to fix
00:43:09gevaertsmc2739: I actually suspected that, but I hadn't tried yet. Thanks for the information!
00:43:16stripwaxsaratoga - thanks
00:43:18saratogastripwax: weren't those done by Unhelpful ?
00:43:28gevaertsso it's really the disk_mount() code that misses it
00:44:50CtcpPing from gevaerts!n=fg@rockbox/developer/gevaerts
00:45:31 Quit ender` (" Kids. You gotta love them. I adore children. A little salt, a squeeze of lemon--perfect. -- Harry Dresden")
00:45:35CIA-69New commit by bertrik (r21975): Update the uda1380 codec driver to allow it to be used without the WSPLL for playback and allow the use of a 256Fs clock signal directly at the SYSCLK ...
00:45:38stripwaxsaratoga - gosh, you're right. I guess I just assumed it was amiconn :)
00:45:43 Nick fxb__ is now known as fxb (
00:46:03saratogahe asked about benchmarking it iirc, and I suggested test_codec, but I'm not sure if thats what he used
00:46:15stripwaxUnhelpful - is that what you used?
00:46:28saratogaas always, if anyone is really interseted in mdct performance, they're welcome to my half finished split radix mdct code!
00:47:30saratogaanyway i have to run, good night
00:48:17 Quit saratoga ("Page closed")
00:49:08 Join funman_ [0] (n=fun@rockbox/developer/funman)
00:49:46 Nick funman_ is now known as funman (n=fun@rockbox/developer/funman)
00:49:47mcuelenaeregevaerts: why were the SD (& MMC) drivers put before the NAND drivers in the multistorage driver?
00:50:09mcuelenaerewouldn't that make SD always disk 0? (even though disk 0 should be non-removable)
00:50:26gevaertsprobably for no good reason at all...
00:50:52roolkuzagor: some oddity with the m:robe 100 SIM - who built it?
00:50:56mcuelenaereit wouldn't help the Onda, but it could solve some problems for the D2 perhaps..
00:51:31gevaertspossibly. Let's talk to shotofadds about it before changing it though
00:51:52gevaertsI'd actually like to have a primary storage define which would always get 0
00:52:36gevaertsthat could possibly still give undesirable results if a target has three or more different drivers, but I think that's not likely to happen soon
00:52:43Zagor"Completed: build mrobe100sim client seattle-jdgordon seconds 47 uplink 0 score 204149"
00:52:59stripwaxunhelpful - if you're able to measure stalls on armv6 and up, would I be right in saying most of mdct_butterfly_16 incurs one cycle stall for each add/rsb combination?
00:53:03 Quit ademille (Read error: 113 (No route to host))
00:53:44Tornestripwax: shouldn't be, if you just mean successive ALU ops
00:53:57 Quit robin0800 (Read error: 110 (Connection timed out))
00:54:11Torneeverything but multiply in the alu is trivial and has result feedback paths, afaik
00:54:12stripwaxTorne - for example, add r9, r9, r5 ; rsb r10, r9, r5, asl
00:54:39stripwaxBecause the rsb op depends on the result of the add, won't that stall?
00:54:41Zagor"Fatal build error: Missing log file. Disabling client." but there is obviously a bug because nobody else was assigned it...
00:54:49mcuelenaeregevaerts: hmm yes, a primary storage define would solve it too
00:54:54Torneit shouldn't, because the ALU feeds results directly back into itself
00:55:08gevaertsI'll see if I can come up with a clean way to do that tomorrow
00:55:09stripwaxAh, neat. Thanks
00:55:25Torneit works differently for multiply but the others should be fine
00:55:37gevaertsThe volume names also still need a bit of work
00:55:48*mcuelenaere wonders where the assumptions are made that drive 0 should always be present, apps/ or firmware/ ?
00:56:01stripwaxThe Tremolo guy has a heavily-annotated mdct by the way (not sure if it's useful but appears to note stalls on xscale) - might come in handy?
00:56:38 Join sirninja [0] (
00:58:02 Quit _lifeless (Read error: 110 (Connection timed out))
00:58:35sirninjaI can't get my database to work. I've put the file database.ignore in all the folders except where I keep my music and I'll start initializing, but it will still say there's 900 files somehow and if I check database info under debug, it says initialized: yes, db ready: no, progress: -1%
00:59:01gevaertsmcuelenaere: I can't find a specific place now, but there's at least the fact that the <microSDxx> things appear as virtual subdirectories in the root. Now if you take the root away...
00:59:42pixelmasirninja: which player do you have?
01:00:03sirninjapixelma: 5th generation ipod video
01:00:05mcuelenaeregevaerts: I've had Rockbox function fine without root
01:00:20***Saving seen data "./dancer.seen"
01:00:32mcuelenaereI was unmounting the wrong disk to invoke the effect though :)
01:00:44 Join _lifeless [0] (n=lifeless@
01:00:44gevaertsthat's cheating :)
01:02:00mcuelenaereRockbox is forcing me to cheat!
01:02:10pixelmasirninja: it sounds like the initialise finishes correctly (disk activity stops). If that's true, how do you reboot your Ipod afterwards?
01:02:35sirninjahold down the play/pause button. is that what I'm supposed to do?
01:03:07 Quit Rob2223 ()
01:03:57 Join Rob2222 [0] (
01:03:58pixelmahmm, yes that's correct. And you don't see a splash "Committing database" on next boot?
01:04:20sirninjano. it just acts as if I never initialized it
01:05:39 Join Nikkado [0] (
01:08:06 Quit m67_l3 (Connection timed out)
01:08:09sirninjanevermind. I just completely reinstalled everything and it appears to be working now
01:09:35sirninjajust curious, is there any way to get out of that usb screen you get when you plug it into the computer?
01:10:32gevaertsyou mean except by unplugging?
01:12:08gevaertsyes and no. You can connect in charge-only mode by holding a specific key (the manual knows which one that is for your player), but you can't go to that mode without unplugging first
01:13:59 Quit bertrik (Read error: 113 (No route to host))
01:14:02gevaertsalso note that on ipod there's still an issue that rockbox doesn't always charge properly from USB, which is why the latest release still reboots to the Apple disk mode for USB
01:19:50 Quit _lifeless (Remote closed the connection)
01:20:06 Join _lifeless [0] (n=lifeless@
01:20:13kkurbjunJdGordon: if you want any help with openocd let me know, I would be happy to try and offer insight on it
01:20:19kkurbjunwhere are you with it now?
01:20:38 Join m67_l3 [0] (
01:24:50 Nick fxb is now known as fxb__ (
01:28:17 Quit Nikkado (Read error: 110 (Connection timed out))
01:31:11 Join markdrago [0] (
01:34:40JdGordonkkurbjun: hey, same as before, except now im pretty sure my openocd install is broken...
01:34:54funmansaratoga: the SD slot pins look accessible enough, I don't know how a "spectrum analyser" looks like however
01:35:35funmansaratoga: I'll bug you until this task reaches the top of your TODO list, I would like to see stable SD access on Sansa AMS ..
01:35:53Zagorooops. /me accidentally disabled flyspray
01:37:03 Quit Thundercloud (Remote closed the connection)
01:37:11JdGordonand nothing of value was lost :p
01:38:03*funman mentions JdGordon only flyspray has the other copy of clipv2 code
01:39:13CIA-69New commit by kkurbjun (r21976): M:Robe 500: Add Rockbox logo for 640x480 and 480x640.
01:41:29*linuxstb expects a big binsize increase...
01:42:05 Join DarkSpectrum- [0] (
01:42:05 Quit DarkSpectrum (Read error: 54 (Connection reset by peer))
01:42:17kkurbjunYeah, I do too :-D
01:42:29linuxstbIs it worth it?
01:42:30JdGordonisnt it still set to 320x240?
01:42:43kkurbjunM:robe has memory to spare - it's a much nicer presentation
01:42:46funmanhow much memory does the m:robe500 has?
01:43:15kkurbjunactually, JdGordon is right, binsize won't increase now - it will when the defines are changed so that it runs in vga mode
01:44:21funmansaratoga: shows which pin is the clock (5th pin where left pin is pin 1, and up left corner is truncated, and pins are visible)
01:44:38kkurbjunit has 64 MB, I guess it's not a ton, but enough for nearly anything after you take out the logo and the like - I was thinking about that and I wonder if there would be a way to store the logo in the plugin buffer or audio buffer so that when you actually do a load you get that memory back
01:45:22JdGordonwe dont really *need* a splash...
01:45:55kkurbjunno, it's not really needed on any target
01:46:21linuxstbMost targets could (do?) display it in the bootloader anyway...
01:46:22funmankkurbjun: I would bother when rockbox will have trouble functioning on this color, 132x80x16 lcd player with 2.32MB of SDRAM
01:47:24kkurbjunlinuxstb: is there a define to disable it in the main build? I thought it is always displayed in the main build
01:47:40linuxstbkkurbjun: Not yet...
01:48:26kkurbjunhmm, I agree it would be nicer in the bootloader that would ave the memory easier than hacking it into the linker script
01:49:20JdGordonthere isnt a gui frontend to openocd is there?
01:49:22funmanwhat's the point of saving 0.009% of SDRAM here?
01:49:39funmanoops, 0.9%
01:50:00kkurbjunJdGordon: no, you can use gdb with it and I think there are frontends for that
01:50:13kkurbjunbut I've always just used it in a terminal
01:50:33 Join Nikkado [0] (
01:50:36funmanJdGordon: you can follow the 1 month old
01:50:55JdGordonyeah, just saw that
01:51:04linuxstbkkurbjun: Do you have a bootloader on the m:robe 500?
01:51:50kkurbjunlinuxstb: yes, but right now it is not really necessary, it will be once JdGordon starts flashing the device
01:52:05JdGordonif i get openocd going :/
01:52:44kkurbjunthere's 3 parts to the boot process right now, a small svg that uses hooks in the OF to load an image from the Hard drive to a known address, then the rockbox bootloader, and then the main build
01:53:14 Quit m67_l3 (Connection timed out)
01:53:14JdGordonbugger.... cant play now
01:53:18 Quit JdGordon ("Leaving.")
01:53:34kkurbjunthe bootloader could be eliminated, but it doesn't add much to the boot time
01:53:51CIA-69New commit by kkurbjun (r21977): Fix Onda
01:59:14 Quit sirninja ("Leaving")
01:59:47 Nick DarkSpectrum- is now known as DarkSpectrum (
02:00:56 Quit efyx_ (Remote closed the connection)
02:04:04Unhelpfulstripwax: i think amiconn has a microbenchmark for mdct based on the one i put on pastebin a bit ago... if it's not in svn already maybe somebody ought to commit it :)
02:04:37stripwaxsounds good
02:06:03 Quit roolku ()
02:06:44funmansoap: I think rbutil doesn't see new current builds (with the change of build system)
02:07:38DarkSpectrumno it doesnt
02:10:23 Join webguest39 [0] (n=04fed293@gateway/web/cgi-irc/
02:10:43 Join mc2739_ [0] (
02:11:46webguest39s their
02:12:16webguest39Is there a rockbox for sansa version 2?
02:12:39scorchewhich sansa?
02:13:16 Quit Nikkado (Connection timed out)
02:13:17 Quit stripwax (Read error: 104 (Connection reset by peer))
02:14:56funmanwebguest39: it is at the moment in the testing phase, the known problems include file system crashes and corruption (for internal storage and microsd slot). More details at
02:15:55 Join Nikkado [0] (
02:16:40FlynDicefunman: I've tried a patch with the assumption that MCLK = CLK_IDE and it works fine for both internal and microsd on my e280v2. It does not make voltage scaling work for the uSD at 1.05v though. Init timeout still.
02:17:32funmanFlynDice: my main concern is the problems reported by testers
02:18:31FlynDiceDo you think the MCLK = PCLK issue is not a factor with that?
02:20:15funmani didn't think about it, but now i don't think that's a problem - what is the error reported (by SD code) when using a lower voltage already?
02:20:20Unhelpfulstripwax: i'll try to work up the macros i used into something fairly reusable... or should i just remove the things it actually benchmarks from my plugin and add it as bench_framework or something?
02:21:43FlynDicemicroSD init failed: -2
02:22:54 Quit webguest39 ("CGI:IRC (Ping timeout)")
02:23:07funmanFlynDice: 1 second timeout while waiting for the card to power up, so the commands are sent successfully and it is a powering problem
02:24:34FlynDicefunman: yes it seems to be, but I thought perhaps the frequency issue might be the culpret
02:25:10 Quit mc2739 (Read error: 110 (Connection timed out))
02:26:21FlynDicefunman: on another note, is there a reason we don't go to widebus (4 bit data transfer mode) or am I just missing it in the code?
02:27:40funmanFlynDice: i think SET_BUS_WIDTH command with argument |= 2 (10b) selects 4 bits data transfer
02:30:38 Join linuxstb_ [0] (n=linuxstb@rockbox/developer/linuxstb)
02:30:47 Quit linuxstb (Read error: 113 (No route to host))
02:31:07 Nick linuxstb_ is now known as linuxstb (n=linuxstb@rockbox/developer/linuxstb)
02:34:32FlynDicefunman: Not for the card but for the pl180 MCI_CLOCK(i) bit 11
02:35:07funmanhum no there is no reason .. perhaps I had thought "let's see this when SD driver works fine"
02:35:30 Join m67_l3 [0] (
02:37:17FlynDiceI can buy that, I've just found that assuming you've overlooked something is usually wrong......
02:38:16 Quit Nikkado (Connection timed out)
02:38:40 Join roolku [0] (
02:39:17 Quit roolku (Client Quit)
02:40:21 Quit Zagor ("Clint excited")
02:40:30Blobis there anyway in rockbox 3.3 to force the time to be in 12 hour mode instead of 24 hour mode?
02:41:39 Nick Ypsy is now known as YpsyZNC (
02:41:46funmanyes, just check the manual for where the setting is exactly
02:42:22evilnick_homeBlob: Yes, IIRC there is a rough limit for m4b files
02:45:06Blobany clue that that time limit might be?
02:46:11evilnick_homeHave a search on the forums, I do remember other people mentioning it but I don't use audiobooks myself
02:49:23Blobtook me a little bit to get used to rockbox, but I'm starting to really like it, though in the latest SVN build as of yesterday USB support wasn't working very well on the ipod mini first gen 4gb device
02:51:45 Quit mcuelenaere ()
02:55:05DarkSpectrumwhat does it mean when i see "fix red", "fix yellow" on the main page?
02:58:58 Nick mc2739_ is now known as mc2739 (
02:59:24funmanDarkSpectrum: bad commit messages for "fix compilation error(s)" and "fix compilation warning(s)"
02:59:27 Quit funman ("free(random());")
03:00:24***Saving seen data "./dancer.seen"
03:03:54 Quit Mikachu (Read error: 60 (Operation timed out))
03:04:00 Join Mikachu [0] (
03:11:56 Join JdGordon [0] (n=jonno@rockbox/developer/JdGordon)
03:17:19 Join evilnick_home1 [0] (
03:21:58aidywould it be possible to create a wps that displays text as part of the moving part of a progress bar?
03:24:57JdGordonkkurbjun: you still round?
03:32:10 Quit evilnick_home (Read error: 113 (No route to host))
03:33:59 Part wincent ("Kopete 0.12.7 :")
03:48:20 Join Nikkado [0] (
03:52:19 Quit m67_l3 (Read error: 110 (Connection timed out))
03:58:42 Quit dfkt ("-= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.")
04:06:38 Quit TheSeven (Nick collision from services.)
04:06:55 Join The_Seven [0] (
04:06:59 Nick The_Seven is now known as TheSeven (
04:17:28 Join bubsy_ [0] (i=Bubsy@unaffiliated/bubsy)
04:18:29 Quit bubsy (Read error: 60 (Operation timed out))
04:21:39 Nick bubsy_ is now known as bubsy (i=Bubsy@unaffiliated/bubsy)
04:22:01 Quit timc (Read error: 110 (Connection timed out))
04:23:39 Quit TheSeven (Nick collision from services.)
04:23:56 Join The_Seven [0] (
04:24:00 Nick The_Seven is now known as TheSeven (
04:31:26 Quit mc2739 ("ChatZilla 0.9.85 [Firefox 3.0.11/2009060215]")
04:40:41 Quit _lifeless (Read error: 110 (Connection timed out))
04:45:07 Quit Nikkado (Remote closed the connection)
04:47:41 Quit DarkSpectrum (Read error: 104 (Connection reset by peer))
04:53:53kkurbjunJdGordon: I'm here now
04:54:13JdGordonwanna have another crack at getting openocd going?
04:54:27BlobThanks for the work on RockBox all
04:54:42JdGordoni checked the connection in windows with the digilent software and it shows up so the soldering is all good
04:55:21Blobfpga development JdGordon?
04:56:13JdGordonwe need to do some jtag fiddling for one of the in development targets
04:57:55Blobfun stuff
04:58:37kkurbjunsure, do you still have cat's configuration?
04:59:34Blobanything a starter in software programing via c/c++ but experiance with ASM and hardware development can do to help?
05:00:27JdGordonkkurbjun: yeah, but openocd wont load it at all now.. i'm running 0.2.0
05:00:29***Saving seen data "./dancer.seen"
05:01:38kkurbjunJdGordon: you will need to edit the config for the jtag 3 . Blob, there's tons to help out with in rockbox - the jtag stuff is pretty specific and needs some hardware, mrsomeone's todo list is a good place for ideas of things that need/could be done or the gsoc page for projects people havn't taken up yet
05:02:05kkurbjunso you're not using the ubuntu package for openocd?
05:02:14JdGordoni am... apparently they updated
05:03:07kkurbjunwhat message do you get when you start openocd?
05:03:27kkurbjunthe version I have in 9.04 is labeled 1.0 (2008-11-26-09:52)
05:04:23kkurbjunin the config file change the parport_cable to dlc5
05:04:42kkurbjunset reset_config to none too
05:05:58JdGordonthat bit is fine... it looks like it barfs on the jtag_device line
05:08:53kkurbjunwhat does openocd say?
05:09:02kkurbjunis it complaining about the idcode or anything like that?
05:09:17kkurbjunfrom the documentation that should be correct
05:09:22kkurbjunjtag_device 4 0x1 0xf 0xe
05:09:33kkurbjunyou might need to add an id code in at the end
05:09:49kkurbjunoh, nevermind, it has the idcode instruction in it
05:10:01JdGordonthat might be ok actually... the target line is definalty changed though
05:11:09kkurbjunlooks like you need to remove reset_halt
05:11:29 Join Nikkado [0] (
05:11:35kkurbjun(I'm using "info openocd")
05:11:42kkurbjunthe man page did't have much
05:11:59JdGordonim going from
05:13:03kkurbjunI think that is for the latest code that uses tcl to configure everything
05:15:08 Quit Nikkado (SendQ exceeded)
05:15:25JdGordon says its for 0.2.0-in-development...
05:15:29JdGordonthats what im runnig...
05:16:02 Join Nikkado [0] (
05:16:42kkurbjunoh gotcha
05:17:03kkurbjunyeah, the configuration is completely different than the cfg file that cat made
05:17:37kkurbjunI've havn't used a recent checkout with that configuration setup so I don't know how much use I'll be
05:18:20kkurbjunIf you can go back to the older release in ubuntu it would probably be easier, but if you want to learn all about openocd it would probably be worth learning their new setup
05:21:43JdGordoni can only find the source pacakge
05:25:21 Join BHSPitMonkey [0] (n=stephen@unaffiliated/bhspitmonkey)
05:27:48kkurbjunhmm, looks like I got my package from universe
05:29:37 Join m67_l3 [0] (
05:32:25kkurbjunJdGordon: are you 64 or 32 bit?
05:33:39kkurbjunthat's the version that I'm using
05:35:06JdGordonyou sure?
05:35:08 Quit Sajber^ (Read error: 104 (Connection reset by peer))
05:35:18kkurbjunthat's what synaptic is telling me
05:35:19JdGordonremoved and installed that one and its still 0.2.0-in-dev
05:36:21kkurbjunyeah, it matches my version string too
05:36:35kkurbjunrun which openocd
05:36:42kkurbjunwhere is it pointing to?
05:37:05kkurbjunok, openocd should be in /usr/bin/openocd
05:37:07CtcpVersion from Rick!i=rick@unaffiliated/rick
05:37:24kkurbjunif you change you path to grab from /usr/bin first it should work
05:38:14JdGordonah yes
05:38:36 Quit Nikkado (Read error: 110 (Connection timed out))
05:39:24JdGordoncrap... battery is flat again
05:40:05JdGordonError: JTAG communication failure, check connection, JTAG interface, target pow
05:40:30kkurbjunis that powered?
05:41:58JdGordonblue light is on
05:42:24 Quit Horscht ("Verlassend")
05:43:28kkurbjunare you running as sudo?
05:43:57JdGordonit usually helps if the right cable is connected :p
05:44:00JdGordonok, we are going
05:44:07JdGordonInfo: JTAG device found: 0x0792602f (Manufacturer: 0x017, Part: 0x7926, Version: 0x0)
05:44:57kkurbjunnow run telnet localhost 4444
05:45:07kkurbjunin a new teminal
05:45:12JdGordonwe are ine
05:45:49kkurbjunso now you can perform all the flash operations, set hardware breakpoints, etc
05:46:07kkurbjunflashing can take a long time throughthe cable
05:46:27JdGordonwhere is the docs for the commands?
05:46:35JdGordonhow can i read the flash?
05:46:41JdGordoni guess thats a good start :p
05:46:58kkurbjunuse "info openocd" in a terminal.. try flash info first
05:47:06kkurbjunyou might have to halt the target
05:47:15kkurbjunor probe
05:47:25 Join ucchan [0] (
05:48:25kkurbjunyou can type help in the telnet window too and it displays all the commands with a brief description
05:48:50kkurbjunand you can do help flash probe for example for more detailed info
05:49:15JdGordonflash info shows plenty of stuff :)
05:50:20kkurbjunso I think flash info gives you a breakdown of the sectors
05:50:35kkurbjunand tells what their status bits are like whetehr they are protected and the like
05:51:37kkurbjunwhat I did with the gigabeat was find a section of flash that I could erase and re-write without killing the of
05:51:55kkurbjunusing sections that contain images are usually a somewhat safe bet
05:52:16JdGordon8MB of flash?
05:52:17kkurbjunor if you can find a sector that is unused it would be better
05:52:22kkurbjunyeah, it's huge
05:52:35JdGordonim dumping the whole lot... should i stop it?
05:53:13kkurbjunstopping commands in openocd from what I reemmber was troublesome, I think occasionally I had to reset the target and restart openocd when I did that
05:53:18JdGordoni have no idea where to though :p
05:53:26kkurbjunthe cable is pretty slow so I would guess it might take a few hours to dump it
05:53:27 Join evilnick_home [0] (
05:53:57JdGordonits done 840KB so yea prtty slow
05:54:08JdGordon4K/s or so
05:54:17kkurbjunthat's not too bad, faster than I expected
05:54:40JdGordonall the flash blocks(?) are "protection state unknown"
05:54:42kkurbjunwriting is definately slow because of the amount of commands you ave to issue when talking to a cfi flash
05:54:52kkurbjunthey are probably unprotected
05:54:58kkurbjunthe gigabeat didn't have any protected
05:55:11kkurbjunI would stay away from those flags becuase some are one time sets
05:55:57JdGordonso how do i find a safe block to play with?
05:56:07Blobi just thought I should let you guys know the usb support in the latest SVN is very unstable for me on the ipod mini 1g device, along with the ipod 4g grayscale device
05:56:46*JdGordon wishses he actually knew what he was doing here :p
05:56:52kkurbjunjdgordon, I have some disassembly of the of from cat that I've gone through which might give some pointers
05:57:45kkurbjunso the general idea is that you change the reset vector at 0x0 to jump to your code instead of in the OF. it would just be a branch instruction - then you place your code at that address
05:58:05kkurbjunyou will want to have something in CRT0.s that tells it whether you want to boot the OF or rockbox
05:58:27kkurbjunif you want the OF you just branch back to the OF location that the reset vector originally pointed to
05:58:50kkurbjunand if you want to run rockbox you keep going through crt0.s
05:59:19JdGordonyou're expecting 0x0 to just have a pointer? or a jump instrcution also?
05:59:28kkurbjunI am guessing there's alot fo stuff that we don't have yet in terms of initialization like the memory, the external interfaces, gpios, LCD, etc
05:59:47JdGordonso i'm going to be learning arm asm :)
05:59:53kkurbjun0x0 always has an instruction, sometimes it's a ld sometimes it's a branch
06:00:14Bloband sometimes it's a nop for interupts with a jmp
06:00:15kkurbjunthe ld would be if you want to set an absolute jump rather than a relative location
06:00:46kkurbjun:), yeah, you will definately be learning arm asm - it's not too bad actually, you can get most stuff with just a quick reference sheet
06:01:17BlobI've used 8051, x86, pic 16f, 18f
06:01:38kkurbjunblob the vector table always has a branch or load instruction first - the fiq is the only one that could differ from that since it's at the end of the vector table
06:01:51kkurbjun(at least with arm)
06:01:56 Quit ucchan (Read error: 104 (Connection reset by peer))
06:02:57 Join ucchan [0] (
06:03:34JdGordonis this flash dump going to be useful? or not for a while yet?
06:03:52Bloball gumps help
06:05:00kkurbjunJdGordon: it would be useful to make sure that the image that openocd makes matches the descrambled firmware image from the updater
06:05:14kkurbjunthat way if something goes wrong you have something to go back to also
06:11:54 Quit evilnick_home1 (Read error: 113 (No route to host))
06:12:44kkurbjunso looking at the OF disassembly it does a ldr PC, =0x30, so it is doing an absolute set
06:13:05kkurbjunso you can either change the address at 0x30 or you can change the initial instruction to a branch
06:14:14JdGordondumped 8388608 byte in 1194.320068s so not too slow
06:14:38JdGordonok easy, so we just need to change that and get some code writen to handle it all :p
06:15:34kkurbjunthe gigabeat crt0.s would be a good place to see what generally needs to be done
06:15:54JdGordonf or s?
06:16:04kkurbjunalso, the linux kernel has a bunch of the low-level initialization for the dm320 too
06:20:47kkurbjunJdGordon: here is the neros linux stuff I was thinking of:
06:21:39 Quit ucchan (Read error: 104 (Connection reset by peer))
06:23:08kkurbjunI guess it's not too bad, but we need to make sure that the registers are setup the way that the OF does and then hopefully things like the LCD don't need additional initialization
06:23:33kkurbjunI've been looking at the OF though and I have not seen anything yet that indicates there is much more to the LCD
06:24:00kkurbjunthat was one of the bigger problems with the F, and that was already mostly figured out by someone else
06:24:09 Join ucchan [0] (
06:24:35 Quit m67_l3 (Remote closed the connection)
06:24:55kkurbjunI think the check you might want ot do in the bootloader for now might be if the charger is inserted or not to decide whether to boot rockbox or the OF
06:26:06kkurbjunonce we have a real, working bootloader then we could make it more sophisticated and check for the touchscren being held, or even the remote depending on how far into the boot process you want to go before checking it
06:27:07JdGordonyeah, one thing at a time though... lets get code runiing from the flash first?
06:27:15kkurbjun:), sounds good
06:27:27JdGordonyou coming to DCW?
06:27:45kkurbjunif you want you could just not even bother with the of for now and jsut work withthe bootloader right at the begining of the flash
06:28:07kkurbjunI have not decided yet, it won't work out to fit around the same time my sister was coming out
06:28:32JdGordonoh well... come anyway :p
06:29:01JdGordonwith jtag going, it should be impossible for me to get it into a bricked state rught?
06:29:15kkurbjunyeah, you should always be able to recover
06:29:46kkurbjunbefore you start flashing, I would look at the disassembly and the dump you have with a hex editor to make sure that they look the same
06:30:24kkurbjunif it matches up for the most part then you have a good dump and the jtag chain is working reliably
06:30:32 Join m67_l3 [0] (
06:31:13kkurbjunI was always able ot recover the gigabeat and it didn't even have a supported flash type - it was /almost/ cfi compliant.
06:31:17 Quit m67_l3 (SendQ exceeded)
06:31:58 Join m67_l3 [0] (
06:32:03kkurbjunand cat's recovered from a bad flash with the jtag setup
06:33:02 Quit ucchan (Read error: 104 (Connection reset by peer))
06:33:33 Join ucchan [0] (
06:38:06 Quit ucchan (Read error: 104 (Connection reset by peer))
06:38:31 Join advcomp2019_ [0] (n=advcomp2@unaffiliated/advcomp2019)
06:39:01 Join Exiled [0] (
06:43:17 Join ucchan [0] (
06:43:53 Join _lifeless [0] (n=lifeless@
06:45:52 Quit ucchan (Read error: 54 (Connection reset by peer))
06:50:02 Join ucchan [0] (
06:53:20 Quit m67_l3 (Read error: 110 (Connection timed out))
06:56:23 Quit advcomp2019 (Read error: 110 (Connection timed out))
06:56:28 Quit Blob (Read error: 110 (Connection timed out))
06:59:07 Quit ucchan (Read error: 104 (Connection reset by peer))
07:00:31***Saving seen data "./dancer.seen"
07:02:22 Quit Exiled ("( :: NoNameScript 4.22 :: )")
07:04:03 Join ucchan [0] (
07:05:23 Join m67_l3 [0] (
07:10:00 Quit r0b- (Read error: 110 (Connection timed out))
07:18:22CIA-69New commit by jdgordon (r21978): rework cuesheet support: ...
07:18:27 Join r0b- [0] (
07:20:31 Join Nikkado [0] (
07:25:04 Nick Nikkado is now known as martian67 (
07:28:18 Quit m67_l3 (Connection timed out)
07:28:36ucchanThere are a lot of patches that I made. (see
07:28:39CIA-69New commit by jdgordon (r21979): woops, remove that sim code which I forgot to remove from the older versions of the patch... ordinarily I would have just said fix red, but that would ...
07:28:55ucchanThese management is serious, and I want to ask for some committing.
07:29:06ucchanFirst of all, could you examine committing about the patch of the text viewer plugin (FS #9853, FS #9855, FS #9892, FS #9893, FS #9898, FS #9902, FS #9990) ?
07:29:17ucchanWhom should I ask though do not exist the mentainer of this plugin ?
07:30:19JdGordonyou've come at a bad time.. most devs are asleep right now...
07:30:24JdGordon(and im about to goto bed..)
07:30:37JdGordonyou should email the dev mailing list and try to get someone to commit them...
07:30:58JdGordonis there any order those need to be commited in?
07:33:56JdGordonfor your codecs patches you probably want to try getting saratoga or linuxstb's attention
07:34:40 Join iwonder [0] (
07:35:30 Join Nikkado [0] (
07:39:57pixelmakkurbjun: you made the new Rockbox logos from the svg, right? Just saying that those don't have the structure in the "box" part, so they are different from the smaller colour ones
07:43:17 Quit martian67 (Connection timed out)
07:43:23iwonderanyone know of a program that finds files that are in one folder (or sub-folder) but not in the other, regardless of the file & folder name given?
07:44:31scorcheiwonder: does this relate to rockbox?
07:45:00iwondercompare advance has it, but alas, it doesn't seem to work reliably
07:45:17 Quit TheSeven (Read error: 113 (No route to host))
07:47:53 Quit ej0rge ("leaving")
07:50:28 Part iwonder
07:50:53 Join m67_l3 [0] (
07:55:22 Nick advcomp2019_ is now known as advcomp2019 (n=advcomp2@unaffiliated/advcomp2019)
07:57:32 Quit Nikkado (Read error: 110 (Connection timed out))
08:05:27 Join Nikkado [0] (
08:06:33 Quit CaptainKwel (Remote closed the connection)
08:09:18 Join homielowe_ [0] (n=homielow@
08:09:18 Quit homielowe (Read error: 104 (Connection reset by peer))
08:13:14 Quit m67_l3 (Read error: 110 (Connection timed out))
08:15:04 Join ender` [0] (
08:20:28 Join m67_l3 [0] (
08:28:20 Quit Nikkado (Read error: 110 (Connection timed out))
08:33:36 Quit ucchan (Read error: 54 (Connection reset by peer))
08:34:20 Join ucchan [0] (
08:37:18 Quit safetydan ("Leaving.")
08:40:25 Join Nikkado [0] (
08:41:47 Quit amiconn (Nick collision from services.)
08:41:49 Join amiconn_ [0] (i=quassel@rockbox/developer/amiconn)
08:41:50 Join pixelma_ [0] (i=quassel@rockbox/staff/pixelma)
08:41:50 Quit pixelma (Nick collision from services.)
08:41:56 Join Rob2223 [0] (
08:42:04 Nick pixelma_ is now known as pixelma (i=quassel@rockbox/staff/pixelma)
08:42:08 Nick amiconn_ is now known as amiconn (i=quassel@rockbox/developer/amiconn)
08:43:37 Quit m67_l3 (Read error: 110 (Connection timed out))
08:44:15 Join flydutch [0] (
08:48:53 Join Rand_Althor [0] (
08:50:17Rand_AlthorWhy does my e200R take longer to boot than my c200?
08:58:11 Join bertrik [0] (n=bertrik@
09:00:02 Quit Rob2222 (Read error: 110 (Connection timed out))
09:00:34***Saving seen data "./dancer.seen"
09:01:01 Quit ucchan (Read error: 104 (Connection reset by peer))
09:02:02 Join ucchan [0] (
09:05:23 Join m67_l3 [0] (
09:08:35Rand_AlthorAnother question: Is it bad for yor DAP to copy new Rockbox builds to it constantly? I'm asking because I heard that every every time a file is written to a FAT partition, the _entire_ FAT (file allocation table) get's rewritten, and the current build for my DAP has 256 files in it.
09:09:37 Quit GodEater (Read error: 104 (Connection reset by peer))
09:10:09 Join GodEater [0] (n=bibble@
09:12:36 Join stripwax [0] (
09:13:32 Quit Nikkado (Read error: 110 (Connection timed out))
09:13:54 Quit Tuplanolla (Read error: 110 (Connection timed out))
09:23:11 Quit bertrik ("De groeten")
09:27:05 Join Thundercloud [0] (i=thunderc@
09:28:25 Quit m67_l3 (Read error: 110 (Connection timed out))
09:30:01 Quit stripwax (Read error: 54 (Connection reset by peer))
09:34:42 Join petur [50] (n=petur@rockbox/developer/petur)
09:35:19 Join m67_l3 [0] (
09:36:58 Part homielowe_
09:40:57 Quit ucchan (Read error: 104 (Connection reset by peer))
09:44:04 Join ucchan [0] (
09:48:12 Quit Thundercloud (Read error: 104 (Connection reset by peer))
09:52:58 Join BradC [0] (n=joust@
09:54:58 Quit ucchan (Read error: 54 (Connection reset by peer))
09:55:29 Join ucchan [0] (
09:58:41 Quit m67_l3 (Read error: 110 (Connection timed out))
10:00:29 Join m67_l3 [0] (
10:04:02 Quit ucchan (Read error: 104 (Connection reset by peer))
10:04:27linuxstbRand_Althor: No, that isn't a problem.
10:06:10 Join ucchan [0] (
10:07:00 Quit mt (Read error: 113 (No route to host))
10:07:18 Join njr [0] (n=nishant@
10:07:32 Join robin0800 [0] (n=robin080@
10:10:15njrHi... any suggestion about an MP3 player(That runs rockbox) for around $100, 8Gb memory, good sound quality with or with video playbakc
10:10:22 Join webguest92 [0] (n=62757ca1@gateway/web/cgi-irc/
10:13:36linuxstbnjr: Do you want something that Rockbox runs well on (which means devices you can only buy used), or something that Rockbox is still in development for (which means possibly serioud bugs and unimplemented features, but which are easier to buy) ?
10:13:54Mikachuif you can find an 8GB 1gen ipod nano it would fit your description
10:15:49 Quit webguest92 ("CGI:IRC (Ping timeout)")
10:16:18GodEaterdid the 1st gen nano come in 8GB? I thought it only went up to 4 ?
10:16:36linuxstbMikachu didn't claim they existed!
10:17:01 Join webguest72 [0] (n=62757ca1@gateway/web/cgi-irc/
10:17:08scorcheGodEater: it did...
10:17:14 Quit webguest72 (Client Quit)
10:18:04 Join webguest15 [0] (n=62757ca1@gateway/web/cgi-irc/
10:18:34linuxstbscorche: Are you sure? Wikipedia claims otherwise.
10:18:35 Join DarkDefender [0] (
10:18:44njrHmm, linuxstb I would prefer something new... but I would defnitly want audio playback playlists and good sound quatlity upto scratch
10:18:50njrdon't care about video
10:18:55webguest15So... I have an iPod 1g.
10:19:13linuxstbwebguest15: You mean the original 1st generation ipod?
10:19:16 Nick webguest15 is now known as WS|Thermal (n=62757ca1@gateway/web/cgi-irc/
10:19:19njrI don't think the new ipods can have their firmware changed
10:19:25scorchelinuxguy3: then wikipedia is wrong...though i dont see them claiming it went up to 8GB
10:19:26WS|ThermalLinuxstb: Yes, I mean the original 1st generation iPod.
10:19:31scorchelinuxstb: ^^
10:19:45WS|ThermalIts old and clunky and bricky but I'd like to run Rockbox on it.
10:19:58njroh sorry I didn't read the 1st gen bit
10:20:11WS|Thermalunfortunately apparently its hardware identifier is 0xFFFFFFFF which means that it thinks its a 2G or something.
10:20:14GodEaterscorche: that's right. they *don't* claim it went up to 8GB. which is what we're saying.
10:20:36WS|ThermalAnd my wheel doesn't work which makes it useless.
10:20:40Mikachui was pretty sure they existed
10:20:45Mikachubut mine is 2gb
10:20:59WS|ThermalI was wondering if there are still any custom 1st-gen builds around, seeing as there's a patch to get it working.
10:21:13WS|ThermalOr if I simply have to grab it via SVN and compile it with Cygwin.
10:21:22WS|ThermalJust wanted to check if there were builds floating around before I did that.
10:22:00GodEaterWS|Thermal: is there something wrong with the builds at ?
10:22:06linuxstbWS|Thermal: Your best choice is to compile yourself from svn - then you always have the most recent.
10:22:07 Join _zic [0] (
10:22:46WS|ThermalLinuxstb: Is the process for applying a diff patch *then* compiling simple on Windows or should I boot up a Ubuntu LiveCD or something similar?
10:23:07linuxstbWS|Thermal: It's identical in cygwin or Ubuntu.
10:23:08WS|ThermalGodEater: Installed Rockbox 1G/2G and my scrollwheel doesn't work, apparently due to
10:23:30WS|ThermalLinuxstb: Alright, cool. I'm more comfortable with Ubuntu though, I might just end up using that. xD
10:23:34 Quit m67_l3 (Read error: 110 (Connection timed out))
10:23:47WS|ThermalSo it'd be a case of grabbing it from SVN, using the diff utils to apply the patch, compiling
10:24:00WS|Thermalthen copying the binaries to the iPod (if the loader is already setup)?
10:24:22linuxstbWS|Thermal: There are instructions in the wiki for compiling Rockbox - see the "For developers" section in the index.
10:24:37WS|ThermalAlright, thanks so much for the help.
10:24:51WS|ThermalHave a nice night or day, depending on your time zone. =)
10:25:32 Quit BHSPitMonkey ("Ex-Chat")
10:26:34 Quit WS|Thermal ("CGI:IRC (EOF)")
10:32:51 Quit ucchan (Read error: 104 (Connection reset by peer))
10:33:22 Join ucchan [0] (
10:37:59 Quit ucchan (Read error: 104 (Connection reset by peer))
10:39:00 Join ucchan [0] (
10:39:32linuxstbnjr: I couldn't really recommend any of the newer devices yet. Assuming you want a flash-based device, the "version 1" Sansa E200 is popular.
10:40:51 Quit Rand_Althor ("ChatZilla 0.9.85 [Firefox 3.5.1/20090715094852]")
10:43:37 Quit ucchan (Read error: 104 (Connection reset by peer))
10:48:56 Join ucchan [0] (
10:51:44 Join bmbl [0] (n=Miranda@unaffiliated/bmbl)
10:54:44 Join KBH [0] (n=hbk@
10:55:01 Join Rob2222 [0] (
10:55:07 Quit flydutch (
10:55:07 Quit Rob2223 (
10:55:07 Quit amiconn (
10:55:07 Quit r0b- (
10:55:07 Quit JdGordon (
10:55:07 Quit dmb (
10:55:07 Quit J-23 (
10:55:07 Quit SirFunk (
10:55:07 Quit tarbo_ (
10:55:07 Quit obo (
10:55:07 Quit HBK (
10:55:07 Quit sitwon (
10:55:13 Quit sbhsu (
10:55:13 Quit YpsyZNC (
10:55:13 Quit Kohlrabi (
10:55:13 Quit MuscleNerd (
10:55:45 Join PaulJam [0] (
10:55:54NJoinflydutch [0] (
10:55:54NJoinamiconn [0] (i=quassel@rockbox/developer/amiconn)
10:55:54NJoinr0b- [0] (
10:55:54NJoinJdGordon [0] (n=jonno@rockbox/developer/JdGordon)
10:55:54NJoindmb [0] (n=Dmb@unaffiliated/dmb)
10:55:54NJoinJ-23 [0] (
10:55:54NJoinSirFunk [0] (
10:55:54 Join obo [0] (n=obo@rockbox/developer/obo)
10:55:54NJoinYpsyZNC [0] (
10:55:54NJoinHBK [0] (
10:55:54NJoinsitwon [0] (
10:55:54NJoinsbhsu [0] (
10:55:54NJoinKohlrabi [0] (
10:55:54NJoinMuscleNerd [0] (
10:56:26 Join tarbo [0] (n=me@unaffiliated/tarbo)
10:56:39PaulJamhmm, did anyone notice some wps corruption with the current build?
10:58:17 Quit SirFunk (SendQ exceeded)
10:59:21 Quit njr (Read error: 110 (Connection timed out))
11:00:36***Saving seen data "./dancer.seen"
11:03:40 Quit aditya (Read error: 104 (Connection reset by peer))
11:04:09linuxstbPaulJam: What kind of corruption?
11:05:36PaulJamthe backdrop has some lines of random coloured pixels in it. looks like sonething is writing to the wps buffer.
11:07:18 Quit ucchan (Read error: 104 (Connection reset by peer))
11:08:42linuxstbI've just tried a sim, and that works fine. I can't test on target at the moment...
11:09:28PaulJamlinuxstb: this is a screendump from the target (H300) the uisim looks fine here too.
11:10:11linuxstbIt would be useful to try some older daily builds and see if you can find when it started.
11:10:59 Join maruk [0] (
11:12:00PaulJamyes, ill try going back. i think it must have been a recent change. last build i used before updating was r21964 i think.
11:12:51linuxstbMaybe last night's cuesheet changes - they seem the only buffering/wps-related changes recently...
11:13:47 Quit HBK (Connection timed out)
11:14:16 Join mt [0] (n=mt@
11:14:44 Join Grahack [0] (
11:17:55PaulJamyes, daily build r21977 looks fine.
11:20:46 Join aditya [0] (n=aditya@
11:20:55 Quit r0b- (Read error: 110 (Connection timed out))
11:21:03linuxstbPaulJam: Do you have cuesheet support enabled?
11:21:14 Join r0b- [0] (
11:21:38PaulJami did reset setting at boot, so no.
11:30:06 Quit _zic ("Ухожу")
11:32:24 Join dash32 [0] (n=dash32@
11:45:45PaulJamJdGordon (if you read the logs): Please have a look at .
11:48:31 Join Sajber^ [0] (
11:50:33 Join Tuplanolla [0] (n=jani@unaffiliated/tuplanolla)
11:56:38 Join einhirn [0] (
11:57:20 Quit PaulJam (Nick collision from services.)
11:57:27 Join PaulJam_ [0] (
12:06:56 Join mankind_ [0] (
12:11:25 Join AndyI [0] (i=AndyI@
12:11:45 Join mcuelenaere [0] (
12:22:06 Join s1mple_m4n [0] (
12:25:11 Quit AndyIL (Read error: 110 (Connection timed out))
12:25:49 Quit aditya (Read error: 110 (Connection timed out))
12:27:52 Join aditya [0] (n=aditya@
12:31:35 Quit s1mple_m4n ("Ex-Chat")
12:47:41 Join LinusN [0] (n=linus@gateway/web/cgi-irc/
12:48:18 Part LinusN
12:48:20 Join LinusN [0] (n=linus@rockbox/developer/LinusN)
12:58:35 Nick YpsyZNC is now known as Ypsy (
12:59:41 Join TheSeven [0] (
13:00:41***Saving seen data "./dancer.seen"
13:03:14 Join tmzt_ [0] (
13:08:17 Join tmzt__ [0] (n=tmzt@
13:14:38 Quit tmzt (Read error: 110 (Connection timed out))
13:18:42 Quit DarkDefender (Read error: 104 (Connection reset by peer))
13:20:54 Quit tmzt_ (Read error: 110 (Connection timed out))
13:21:13 Join MethoS [0] (n=clemens@
13:28:18 Quit mcuelenaere ()
13:35:07 Nick fxb__ is now known as fxb (
13:36:01 Join paulk [0] (n=paulk@
13:39:17paulkHello !
13:39:35AfterDeathHi !
13:40:28 Join dfkt [0] (i=dfkt@unaffiliated/dfkt)
13:41:31paulkI've just modified the page, adding the informations for fedora users !
13:44:28paulkwell, good bye !
13:44:36 Quit paulk ("Ex-Chat")
13:54:11robin0800All commits since 3.0, last 12 months, last four weeks.... this from the home page, might be better if it says 3.3 not 3.0
14:02:11 Quit MethoS (Remote closed the connection)
14:07:12 Quit mt (Remote closed the connection)
14:09:46 Join mt [0] (n=mt@rockbox/developer/mt)
14:19:58 Quit mt (Remote closed the connection)
14:20:11 Join mcuelenaere [0] (
14:20:25 Join mt [0] (n=mt@rockbox/developer/mt)
14:25:07 Join _zic [0] (
14:37:14 Quit mt (Remote closed the connection)
14:39:08 Join LambdaCalculus37 [0] (i=44a0430d@rockbox/staff/LambdaCalculus37)
14:44:57 Quit mankind_ ("Leaving")
14:49:07 Quit _zic (Remote closed the connection)
14:56:08 Join wark [0] (
14:59:24 Join Lear [0] (i=chatzill@rockbox/developer/lear)
15:00:42***Saving seen data "./dancer.seen"
15:01:58 Part LinusN
15:02:01 Quit gevaerts (Nick collision from services.)
15:02:10 Join gevaerts [0] (n=fg@rockbox/developer/gevaerts)
15:11:21LearHm, I suspect something bad with r21978. I get screen corruption on entering the WPS with r21979, but not with r21977.
15:11:37LearCuesheets aren't enabled, btw.
15:12:01 Join evilnick [0] (i=0c140464@gateway/web/freenode/x-24f058470764c383)
15:13:06linuxstbLear: Yes, there's already a FS task about it...
15:14:32 Join Xerion_ [0] (
15:18:41LearA bit surprised it's only some screen corruption (that I've seen so far at least). Putting struct cuesheet on the stack can't be good. :)
15:23:33CIA-69New commit by gevaerts (r21980): take first_drive into account when handling hotswap. This is not important as long as this is the only driver in use, but it avoids bugs as soon as ...
15:26:20 Join saratoga [0] (i=9803c6dd@gateway/web/freenode/x-e0437418cccb7b90)
15:27:05saratogafunman: I hooked up my microsd card and e200v2 to a scope, and the only clock singal I see is at 90MHz
15:27:19saratogahowever, with the probe attached, rockbox cannot seem to read the card, and sometimes panics
15:27:29saratogaso I'm not 100% certain that its actually giving the clock
15:27:42saratogathe 90MHz signal I see is fairly weak
15:29:12saratogahow hard would it be to make a build that has the SD clock always on even if theres no card?
15:29:26linuxstbLear: Where is struct cuesheet on the stack? I only noticed pointers to cuesheets.
15:30:49 Quit Xerion (Read error: 110 (Connection timed out))
15:30:49 Nick Xerion_ is now known as Xerion (
15:31:23LearI mentioned it on FS task, but check audio_finish_load_track in playback.c.
15:31:39 Join DarkDefender [0] (
15:31:41LearGonna do a quick test without it.
15:31:45linuxstbThe patch was a FS task?
15:32:18LearNo, the one about the bug. The one you referred to...
15:33:59linuxstbBah, declarations mixed with code...
15:34:48LearAh, see it now.
15:35:04linuxstbThat's why I missed it...
15:36:30linuxstbBut doesn't the location of the definition mean that space will only be reserved on the stack for it if that block of code is run?
15:37:19LearYou're not the first one to believe that... Stack space for all autos is reserved on function entry.
15:37:45LearWouldn't help those with cuesheets anyway.
15:39:56LearAnyway, I removed the if statement containing that struct, and that fixed the problem for me.
15:43:45linuxstbbufalloc seems an oddly named function - it allocates a buffer _and_ copies data into it...
15:45:10LearAnother "bad" thing: there's also buffer_alloc, which deals with another buffer... :)
15:47:24CIA-69New commit by gevaerts (r21981): Make sd_present() and mmc_present() look only at the actual presence of a card, not at initialisation state
15:47:38*gevaerts forgot to mention the bug number :(
15:48:27 Join roolku [0] (
15:49:48roolkuzagor/bagder: there seems to be a hickup with the build system: the summary is missing:
15:49:51 Quit TheSeven (Read error: 104 (Connection reset by peer))
15:50:07gevaertspixelma: could you test if the ondio mmc card is properly recognised on boot sometime?
15:51:30 Join TheSeven [0] (
15:51:42 Join mankind_ [0] (n=mankind@
15:55:37linuxstbDoes anyone else miss the date/time of the commit from the build table?
15:56:35soapHow rude would it be to put the buildclient in a while loop?
15:56:45gevaertsis it still needed?
15:57:00linuxstbAlso, is the name at the top of the screen being a link to the zip (where available) a new feature?
15:57:23soapgevaerts, was that directed at me or linuxstb?
15:57:25gevaertssoap: I think that if you put a sleep 60 or so in it as well, there should be no issue
15:57:33gevaertssoap: at you
15:57:35linuxstbsoap: My message was directed to anyone...
15:57:45linuxstbsoap: But yours wasn't...
15:57:47*linuxstb gets his coat
15:58:26soapIs it still needed? That I don't know. I haven't awaken to find it stopped for a few days, but I'm about to leave town and be w/o even ssh access for weeks.
15:58:40 Join Lynx_ [0] (n=Lynx@
15:59:09 Quit mankind_ (Remote closed the connection)
16:00:10 Quit Grahack ("Leaving.")
16:03:31linuxstbLear: Hmm, and (unless grep is failing me), that if() statement is the only code that's using bufalloc()...
16:04:26linuxstb(and nothing was using it before that commit)
16:09:06 Join Ubuntuxer [0] (
16:10:27Learlinuxstb: Seems like it, yes.
16:14:17gevaertssoap: I'd say put it in a loop, with a sleep to avoid overloading things. We (at least Zagor and Bagder) can ban clients now...
16:15:14soapgevaerts, put it in with a 60 second sleep like you suggested. Maybe I'll go 300 or so to be nicer.
16:15:40soapI don't want to be banned!
16:15:44Tornegevaerts: i'm pretty sure I've fixed the PP5020 ATA DMA patch, I'm wondering how much more testing it would need :)
16:16:39soapwith or without alignment?
16:16:58Tornei've just changed it to only DMA for things which happen to be cacheline aligned
16:17:54TorneI don't think we can actually correct for the misalignment in a way which doesn't suck
16:18:24Torneif the address to copy the first sector to is not aligned then nor will any of the other sectors be, so you can't just do either end with PIO and use DMA for th emiddle, the request has to be for whole sectors
16:19:00TorneUSB seems to get a perforance improvement with this, at least; the usb buffers are already aligned
16:19:08 Join jgarvey [0] (
16:19:28Torneand there's probably some *fairly* easy tweaking to buffering.c to make the majority of audio buffering requests aligned as well
16:20:03Tornebut exactly what the nicest way to do that is i'm not sure, and it's kinda orthogonal to getting DMA to work on pp5020
16:20:38 Join BryanJacobs [0] (
16:20:53gevaertsSo what's next? A set of test builds for all PP ATA targets?
16:21:09Torneit's only pp5020
16:21:17Tornewell, 502x
16:21:27Tornebut yes, that would seem like a good idea
16:21:46*gevaerts tends to forget about those three ipods :)
16:21:47Tornei've tested it on my player as much as i can, with just the ata patch + alignment fix (i.e. no buffering changes)
16:22:01Tornethe stats patch show sit does DMA for about 10% of requests
16:22:12Tornei.e. most things are not in fact sufficiently aligned, but some are by fluke
16:22:34linuxstbHow big is the cacheline?
16:22:39Torne16 bytes, i believe
16:22:48Torneit's what the stuff on says
16:23:00Torneand making 16 byte alignment required makes all my problems go away
16:24:02gevaertsTorne: you're running with UDMA1?
16:24:11Torneer, yes
16:24:12 Quit mcuelenaere ()
16:24:16Tornewhatever the v0.7 patch you posted does
16:24:24Tornei have no changes to the dma code other than my one liner for alignment
16:24:43Torneyeah, udma1, says the ipod
16:25:12gevaertsok. UDMA4 is a lot faster again, but I'm not sure how safe it is
16:25:21gevaertsat least according to
16:25:34Tornewell again that's somethhing we can look at later, i think
16:25:48Tornedreamlayers seemed pretty sure that modes higher than 1 were not ok at unboosted cpu speed
16:26:11gevaertsindeed. Let's leave higher levels for later...
16:26:24Torneimx31 gets to use scatter/gather to work around this
16:26:29Tornewhich is Not Fair :)
16:26:57TorneHm, actually i do notice one thing looking in th eimx31 version, though, which is that cacheline interference should only happen on reads, and writes should only need to be 4-byte aligned.
16:27:17linuxstbgevaerts: Do you know what device those benchmarks are done on?
16:27:20Tornemy patch requires it for both, which isn't strictly necessary
16:27:29gevaertslinuxstb: Horscht's ipod video 80GB
16:27:45Tornei gave horsct a link to my patch yesterday and he said he'd have a go but not heard back from him yet
16:28:23gevaertsTorne: well, there won't be many unaligned writes anyway...
16:28:30Tornetrue, perhaps none
16:28:41 Join Grahack [0] (
16:28:42Tornebut as it should be safe, it might be nicer to allow it
16:28:45*gevaerts doesn't see many real usecases for non-USB writes
16:28:49Torneyou still need 4-byte alignment to use DMA at all, though
16:30:18Torneso i guess a test build would be a good plan probably
16:30:48Tornethen if htat's ok we can probably commit this? it won't be as big a performance improvement as it *could* be but it shouldn't make anything *slower*, and it should help for usb
16:31:20Tornethen i guess we need a sensible way of getting buffering.c to put audio at nice alignments :)
16:33:56BryanJacobsTorne: not too hard to do
16:34:03TorneBryanJacobs: yah, we discussed it before
16:34:27BryanJacobsI think the conclusion we came to was to make add_handle (+ shrink_handle) place things at the right alignments?
16:34:32Tornea nice way would probably do it only for targets that wanted it to, though
16:34:42Torneyes, but that's not sufficient
16:34:52Torneyou need the handle struct to end at an aligned place
16:34:57Tornebut you also need to tweak the file offset
16:35:11Tornebecause if the file's offset is, say, 37, then the disk reads will never be aligned :)
16:35:16BryanJacobswell, or you can just set h->data to be a little after the end of the handle struct
16:35:26BryanJacobsoh, it has to be aligned at the SOURCE too?
16:35:38Tornebut it transfers whole sectors
16:35:51 Quit wark ("Leaving... for now. But I'll be back.")
16:35:55Torneso the alignment of where the sector has to go needs to be 16.
16:35:56BryanJacobsthen just do h->len += foo; h->offset -= foo
16:36:09BryanJacobsand increment the ridx a bit after the first read
16:36:11Torneyes, there's already a patch that does that on the FS# for this
16:36:18FlynDicesaratoga: RE 90MHz SD signal: Was it right at 90 or down around 82 by any chance. CLK_IDE is set at 90MHz but actually runs at 82.4.
16:36:19Tornejust it does it for 4-byte alignment
16:36:28BryanJacobsok then you're all set
16:36:31Torneit just needs changing ot 16, and for h->data to be 16-byte aligned.
16:36:45Tornebut still it should probably only be doing this for targets wher eit's needed
16:36:59Torneand should maybe be able to deal with more than one alignment :)
16:37:13 Quit BryanJacobs ("Java user signed off")
16:37:25 Join BryanJacobs [0] (
16:37:33TorneBryanJacobs: also I looked at your diagram
16:37:37BryanJacobsif you use a #define for the alignment level that should work
16:37:38Torneand it's very helpful :)
16:37:44BryanJacobsthanks :-)
16:37:51Tornei agree that your strategy seems reasonable
16:37:55BryanJacobsI've been thinking about making a call flowchart
16:37:57Tornei had previously misunderstood what you meant slightly
16:38:05Torneyou are in fact talking about what i was thinking of :)
16:38:21Torne(but it's not like it's up to me, so hey)
16:38:32BryanJacobswell, it's certainly not up to me either
16:38:52BryanJacobsI think I'm going to implement the "conservative" strategy first - not allowing things to be placed anywhere but at widx
16:39:23BryanJacobsoh, anyway, if you have a #define ALIGN_BYTES, setting it to 1 will be OK for targets that don't require alignment
16:39:30BryanJacobsit won't do any harm there
16:39:38BryanJacobsand the compiler should optimize it out
16:40:02Tornesomething like that, yes
16:44:38saratogafunman: I did better measurements on the SD card during a file transfer
16:44:42saratogaits 62.4MHz
16:44:46saratogain rockbox
16:44:56saratogai suppose I could try the OF
16:48:55Tornegevaerts: ok, i posted a tidier patch that just does it for reads, and also removes the TODO comment about handling cacheline interference :)
16:50:58*gevaerts makes some builds
16:51:15*Torne also updates his own build to current svn
16:51:34saratogaFlynDice: sorry my previous number was a bit off, i have to actually do an SD transfer to get the clock high
16:52:01saratogathe 90MHz signal I saw was just noise in the background, its always there but but 100x weaker then the 62.4MHz signal during a transfer
16:52:45saratogathe OF looks stranger, I had to do the measurement with the USB cable plugged in, so its not exactly the same as rockbox's test, but I see a lot of different clocks on the line
16:53:25saratoga62.4MHz was not one of them
16:55:33saratoga16.35MHz seems the most common peak in the OF's clock, but it also sometimes does 2 and (about) 3x as much
16:58:21 Nick tmzt__ is now known as tmzt (n=tmzt@
16:59:37 Join n1s [0] (n=n1s@rockbox/developer/n1s)
17:00:33saratoga48Mhz also seems common in the OF
17:00:44***Saving seen data "./dancer.seen"
17:00:47FlynDicesaratoga: Seems to point to MCLK = PCLK and not the MCLK = CLK_IDE we were speculating about...
17:02:51Tornegevaerts: Hm, actually there is one other issue: FS #9721
17:04:20gevaertsTorne: maybe I should just add that to the test builds?
17:04:35TorneBut for actual commit that's not ideal
17:04:39Tornebecause it panics on write errors :)
17:04:46Torneshake player -> panic is not ver ynice
17:04:54Tornefor the test build that should be fine though
17:05:04Tornewe want to know if write errors are showing up :)
17:05:11Torneeek, i'm being forcibly rebooted
17:05:13Tornebrb :)
17:05:35gevaertsTorne: still better than silently failing
17:05:49 Join timc [0] (n=aoeu@
17:06:13 Quit TheSeven (Read error: 110 (Connection timed out))
17:10:04 Join jhulst [0] (
17:12:13saratogaFlynDice: I'll leave it setup on the analyzer if you want more measurements
17:12:54 Join funman [0] (n=fun@rockbox/developer/funman)
17:13:46saratogaFlynDice: looking online, 62 MHz seems too fast for SD
17:13:50saratogaperhaps this is the problem?
17:13:51linuxstbJdGordon: Seen ?
17:14:03funmansaratoga: where do you see 62MHz is too fast?
17:14:17saratogafunamn: wikipedia says 50MHz for SD
17:14:21JdGordonyes, working on a quick fix now... only have about 15min before i have to go though
17:14:42JdGordonserves me right for testing on hwcodec insetad of my usual swcodec targets :p
17:16:01saratogaif we're running cards a little too fast for spec, that might explain why some devices have glitches and some do not
17:16:52linuxstbJdGordon: It didn't even compile on swcodec! ;)
17:17:10JdGordon? it compiled fine?
17:17:27funmanthe OF mentions a desired clock frequency of 20MHz for µSD and 65MHz for internal storage
17:17:35linuxstbJdGordon: I meant hwcodec... And it was the sims... So just ignore me...
17:17:46JdGordonI'll do my best :D
17:17:58funmanthe pl180 doc mentions 20MHz maximum for MMC and 25MHz maximum for SD
17:18:54saratogafunman: most of the energy on the clock for the OF was at 16MHz
17:19:14funmansaratoga: we should work on a build with lower clock frequencies, so you can verify if the measured CLK line is lowered accordingly (and confirm that mclk == pclk)
17:19:35saratogaok i'll leave the instrument hooked up, just give me a patch and I will tell you the clock
17:19:48saratogaalso if you could leave the clock energized all the time in the patch it would make testing much easier
17:19:50funmanthe requirement is "pclk >= 3/8 mclk", and "pclk >= 3/8 pclk" is valid
17:20:28saratogathe probe tends to cause panics if i use it will transfering so measuring the exact frequency is hard if the bus switches off during an error
17:21:57saratogabah ucchan logged in last night while no one was around
17:23:02JdGordonlinuxstb: fix is up... i have to get to work now, so if oyuw aant to test it commit if it works
17:23:24 Join polobricolo [0] (i=424b37e6@gateway/web/freenode/x-94e2b69595a590d8)
17:24:24funmanthe 25 or 50MHz is given by TRAN_SPEED in the csd (we want to check linux code for pl180)
17:25:30 Part Ubuntuxer
17:26:05funmansaratoga: for keeping clock always on, making sd_enable() return immediately (no effect) should be enough
17:26:15Tornegevaerts: yes, but retrying a couple of times at least would be nice :)
17:26:28Tornegevaerts: and/or switching to read only mode
17:26:40saratogafunman: so any idea why we use such a fast clock for SD?
17:26:55gevaertsTorne: maybe whenever we get a read or write error, we can shuffle the playlist? :)
17:26:56funmansaratoga: faster is better
17:27:03JdGordonwe could just about get rid of the cuesheet option if it wasnt for hwcodec :
17:27:09Tornegevaerts: hehe
17:27:30funmanI didn't know about the maximal clock rate and the storage appeared to run just fine until recently ..
17:27:35JdGordoni.e always having it enabled and using the audio buffer completly... stupid hwcodec needs the static buffer though...
17:27:44saratogabut you know how to change it?
17:27:54funmansaratoga: yes, i'm doing it now
17:28:50 Join readabil1ty [0] (n=chad@
17:32:33PaulJam_JdGordon: a quick test of the patch shows no issues (no WPS corruption and no crashes), but i don't use cuesheets, so i have no idea if this (still) works correctly.
17:33:15JdGordonthanks.. it *should* work, but yeah its had no testing...
17:33:22JdGordonliterally put together in 5 min
17:36:00 Join TheSeven [0] (
17:36:12LearIt looks ok, whatever that's worth... Would prefer a strlcpy over strcpy though. :)
17:36:34PaulJam_meh, i just noticed that "statusbar: on" in a theme file now removes the statusbar. maybe _someone_ should update all the themes on the themes page accordingly...
17:37:11 Quit funman (Read error: 113 (No route to host))
17:37:42 Join funman [0] (n=fun@rockbox/developer/funman)
17:38:14 Quit polobricolo ("Page closed")
17:40:13 Join captainkwel [0] (i=2669ecc2@gateway/web/freenode/x-fea1ba28429d7e98)
17:40:25*JdGordon has no idea why there is the cuesheeet.audio_filename...
17:42:10JdGordonits unused...
17:42:23JdGordonfuck, gotta run to work... ill commit this on the way in
17:42:24 Quit Byan (Read error: 110 (Connection timed out))
17:43:21 Quit readability (Read error: 110 (Connection timed out))
17:43:54gevaertshm, maybe I'm being stupid here. Making test builds for ATA patches based on a revision with WPS known corruption may not be too smart...
17:44:21*gevaerts will make new builds once FS #10453 is fixed
17:47:19JdGordonits worse than just corruption... its almost certainly an undetected stck overflow
17:48:45Tornegevaerts: ooh crap
17:48:48Tornei hadn't noticed that either
17:48:52*Torne cancels the build he's doing :0
17:49:47funmansaratoga: with a clock of 15.5MHz (62/4), the internal storage can't be initialised
17:49:53*BryanJacobs chuckles
17:50:31CIA-69New commit by jdgordon (r21982): fix FS #10453 - the cuesheet changes broke playback on swcodec, rework it so it uses the audio buffer instead of a temp buffer from the start (which ...
17:50:49JdGordonnow work time
17:50:51funmanDATA CRC FAIL
17:51:08saratogafunman: do any delays need to be initialized?
17:51:14saratogaerry adjusted
17:51:45funmannot sure, bank selection is near the end of initialization
17:59:11funmangevaerts: the "no partition found. insert USB cable" dialog in main.c/init() should reboot immediately when ROCKBOX_USE_USB isn't defined?
18:02:34funmansaratoga: so it looks like changing SD clocks needs a bit more work
18:02:54funmansaratoga: I think you pointed out a very interesting problem
18:03:38 Join JdGordon| [0] (i=483e3c25@gateway/web/freenode/x-da20dd5d8183dcf1)
18:05:03 Quit petur ("work->home")
18:06:54 Join muesli [0] (
18:07:29muesliPaulJam_ stupid question, when will r21982 be released? ;)
18:08:01PaulJam_muesli: it should be available on the current builds page.
18:10:38muesliah, i love you!
18:11:06PaulJam_i didn't fix it.
18:11:12mueslijust modded my h120 to a cf-player and now even old wps work fine :D
18:12:15muesliluv ya anyway and thanks to the unknown fixer :D
18:12:52*JdGordon| glares at muesli
18:13:19 Join ej0rge [0] (
18:13:39muesliJdGordon *cuddle*
18:14:31mueslibtw concerning saving energy for a cf-modded player without hdd: which disk spindown time shall be fine? ->254s ?
18:17:42PaulJam_if it has an effect on powerconsumption at all on a cf modded player then i would assume the setting should be at the lowest setting.
18:19:41muesliyeah, of course...
18:20:32 Join n00b81 [0] (n=n00b81@unaffiliated/n00b81)
18:20:33mueslisame question regarding dircache. i guess disabling since its not neccessary anymore shall save some juice?
18:22:44PaulJam_i'm not entirely sure, but it could be that some parts of the database are faster with dircache on (if you use that feature).
18:23:52muesliok...i guess theres not that much change to notice. with an upgraded battery you'll get up to 28h of playback. and since the hdd was replaced it shall be even more
18:26:07muesliah..anti-skip buffer was cruical to battery life afaik. i guess the smallest value shall do it again?
18:28:00BryanJacobsthat one you definitely want to be as small as possible
18:28:32BryanJacobswell, actually depends on how often you skip tracks
18:28:55 Quit Zambezi (Read error: 104 (Connection reset by peer))
18:29:03LearErh, no, it isn't related to that kind of skips...
18:29:08amiconnUnhelpful: I didn't use a special benchmark for mpegplayer idct and yuv blit speed testing. I just played an .mpg and measured total playtime (fps display is too imprecise)
18:29:16BryanJacobsyes it is, because you want the CF to be inactive as much as possible
18:29:37BryanJacobsif you skip forward often you'll spend longer with the CF/HDD active if the antiskip buffer is larger
18:30:07Mikachuhow is the anti-skip buffer different from the regular buffer? (i read the manual entry but it didn't help)
18:30:16BryanJacobsthe antiskip buffer is decoded PCM
18:30:22Mikachuis it the low water mark?
18:30:31BryanJacobsoh, no, I'm wrong
18:30:32funmanamiconn: so you didn't play it realtime?
18:30:43LearLow water mark for file buffering, not pcm.
18:30:43BryanJacobsthe antiskip buffer is the low-water mark, Mikachu is right
18:30:57BryanJacobsthe PCM buffer is independent
18:31:04BryanJacobsspeaking of which, what determines the PCM buffer size?
18:31:07muesliBryanJacobs if that vlaue was too small the h320 was skipping while scroling
18:31:41*BryanJacobs goes to look at where the "antiskip buffer size" figures into buffering.c
18:31:59PaulJam_BryanJacobs: i think it is fixed size (only crossfade infulences its size)
18:32:04funmanBryanJacobs: see pcmbuf_get_next_required_pcmbuf_size (depends on crossfading and memory size)
18:32:07JdGordon|for some reason im thinking its not even used?
18:32:18LearE.g., if the anti-skip buffer is 1 minute, file buffering starts when there is 1 minute worth of data left in the file buffer.
18:32:38BryanJacobsLear: that would be the low water mark
18:33:22 Quit maruk ("Leaving.")
18:33:30LearYes, but that's what the anti-skip buffer adjusts.
18:33:42BryanJacobsfunman: it's not there...
18:34:08BryanJacobsoh, you mean for the PCM buffer size
18:34:09BryanJacobsgot it.
18:34:39Mikachudoesn't the pcm buffer just always try to stay full?
18:35:15BryanJacobsdefine "try"?
18:35:25BryanJacobsthe codec gets invoked when there are enough data ready for it
18:35:42Mikachu"it doesn't wait for it to empty to any particular ratio before it starts decoding"
18:35:56Mikachufor various values of "it" ;)
18:36:05BryanJacobsI believe that is correct - there's no cost to starting up the codec to read from memory to memory
18:36:06LearYes, the codec thread works whenever it can. If the pcm buffer goes low, the CPU is boosted.
18:36:34Mikachuwell, i guess it waits for one decoded block to fit or something
18:37:01BryanJacobsthe CPU also gets boosted if buffering falls behind
18:37:41JdGordon|whats the smallest reasonable size the PCM buffer could be? it doesnt really need to be 512K on flash targets
18:38:02Mikachuwhy is the pcm buffer affected by disk type? don't you mean the file buffer?
18:38:04BryanJacobsit's based on time, not size
18:38:08funmanJdGordon|: it is dependant on CPU speed
18:38:13JdGordon|hmm.. although you'd boost more if it was smaller wouldnt you?
18:38:23BryanJacobsonly if the codec can't keep up
18:38:39BryanJacobsif the codec unboosted can decode at greater than realtime that won't cause any boosts
18:38:45JdGordon|in that case... i mean how large can we make it so flash targets dont need to boost?
18:38:52JdGordon|seem as reading from disk is fast
18:39:02Mikachui did at one point change the plugin buffer to 28MB on my nano with no issues (pcm buffer same size, file buffer smaller)
18:39:48funmani think you need to measure variations in decoding latency for all the codecs
18:39:50Mikachu(i am not using that change anymore)
18:40:18funmanthe pcm buffer needs to be big enough to avoid underruns
18:40:54JdGordon|20MB PCM, 0MB file buffer should work goodly...
18:41:09JdGordon|10MB file buffer even
18:42:29JdGordon|maybe not
18:42:40funmanwhat would be the point ?
18:43:04JdGordon|less boosting
18:43:45funmanwhich boosting case would be removed?
18:43:49JdGordon|assuming boosted cpu needs mre power than flash access
18:43:50Learless frequent boosting only
18:43:56PaulJam_wouldn't that be worse for the battery since it would stay boosted all the time while it fills those 10 MB PCM buffer? even when the fileformat wouldnÄt require boosting usually?
18:44:23JdGordon|back in 10...
18:44:51funmanPaulJam_: the CPU isn't required to be boosted until the PCM buffer fills
18:45:06LearStill, could be interesting to see if it makes a difference...
18:46:19 Join bertrik [0] (
18:46:43PaulJam_one thing with a large PCM buffer would also be the latency when changing DSP settings.
18:47:24 Quit JdGordon| (Ping timeout: 180 seconds)
18:47:34r0b-is there any way to make mpegplayer play other mpeg types like mpeg4
18:48:09linuxstbbertrik: Hi. Do you have your Meizu audio hack patch somewhere? I'm curious to try it out... (timers are killing me...)
18:48:10bertrikno, the plugin can't handle mpeg4
18:48:11PaulJam_r0b-: there is, but i don't think you'd like it.
18:48:22r0b-writing code?
18:48:32linuxstbYes, writing an mpeg4player...
18:48:39bertriklinuxstb, I'll try to make a clean patch ...
18:51:21 Join JdGordon| [0] (n=Miranda@nat/microsoft/x-24d3cc7f6caae7e7)
18:52:08JdGordon|even if the codec cant decode at realtime, with a large enough PCM buffer you wouldnt have to boost very often (if at all)...
18:52:33Mikachuat some point you have to fill your large buffer
18:52:39JdGordon|you would have to hit the disk more often though, but on flash targets thats not an issue because they are fast.. I dont know if the power tradeoff there makes sense
18:52:47JdGordon|Mikachu: sure, but not boosted
18:53:07funmanJdGordon|: codecs do control boosting individually
18:53:47Mikachuthe size of the buffer doesn't affect how much you need to boost on average
18:54:37dzMikachu: except when seeking outside of the buffer, in which case larger is worse
18:55:06Mikachubut it's harder to seek outside the buffer then
18:55:08dz(or skipping tracks, which I do on an all too frequent basis)
19:00:22 Join Gigawatts [0] (n=chatzill@
19:00:47***Saving seen data "./dancer.seen"
19:03:10 Part Gigawatts
19:07:33n00b81Anyone here have a 6G willing to do some tests?
19:07:51n00b81*6G classic.
19:09:22LambdaCalculus37n00b81: I don't have one to use. What kind of tests?
19:09:34*LambdaCalculus37 should reiterate...
19:09:41LambdaCalculus37I don't have one with me right now.
19:09:43TheSeventrying if code gets executed, e.g. the tweeter tweets
19:10:59bertriklinuxstb, see
19:11:13bertrikexperimental USB stuff is mixed up a bit in it though
19:11:40bertrikwavclip.c is the audio sample, stubs.c contains some stubs to play pcm without pcm.c
19:12:00bertrikand the code to set up pcm is in bootloader/meizu_m3.c
19:13:21linuxstbbertrik: Thanks.
19:14:10gevaertsCan anyone have a look at and shout if I'm unclear?
19:15:55Tornegevaerts: seems reasonable, but the lack of a . on "the main result is usb transfers should be faster" seems like you trailed off without saying something else
19:15:59Tornedunno whether you really did or not :)
19:16:30gevaertsTorne: unless you know of other expected results, I'll just add a period :)
19:16:46Tornewell, buffering migh tbe a little faster (and thusly use a little less power)
19:16:58gevaertsyes, but is it noticeable?
19:17:00Tornebut it's not likely to be a human noticable improvement
19:17:16Torneit might show up on a battery bench. i've not actually tried.
19:17:28Torneas i was considering the degree of improvement uninteresting at thsi stage
19:17:48Tornesince it's only using DMA by chance :)
19:17:53 Join barrywardell [0] (
19:17:54bertriklinuxstb, you'll probably have to clean up some hacks in firmware/SOURCES too
19:18:06gevaertsThat sentence is actually mainly there to make people interested in testing :)
19:18:29Tornegevaerts: Yes. That's probably sensible :)
19:18:47gevaertsThere is a risk after all, even if we expect it to be very small
19:18:56TorneThat'll do I think
19:18:58Torneand yes, of course
19:18:58 Join Zambezi [0] (
19:19:40 Join mc2739 [0] (
19:21:38 Quit mc2739 (Client Quit)
19:21:57 Join mc2739 [0] (
19:22:41mc2739gevaerts: Afftected players are : irived
19:26:04*gevaerts looks in vain!
19:26:43*mc2739 cleans glasses
19:26:59 Join Horscht [0] (n=Horscht2@xbmc/user/horscht)
19:27:35*gevaerts just hopes that mc2739 won't look at the last modification timestamp
19:30:37CIA-69New commit by dave (r21983): Add quotes around the name of any missing tools (patch, make etc) to help clarify the error message
19:30:38roolkugevaerts: players are still "afftected" ;)
19:31:09linuxstbOops, one too many files....
19:31:16gevaertsroolku: never!
19:33:25CIA-69New commit by dave (r21984): Oops: revert r21983 for checkwps.h - it was an unrelated change committed in error
19:36:33 Join shotofadds [0] (n=rob@rockbox/developer/shotofadds)
19:37:03shotofaddsgevaerts: did you test r21981 on your D2? the SD card no longer works on bootup
19:37:20shotofaddsI don't really understand that change
19:37:44gibbon_seems like my build host is finally working :)
19:39:42*bertrik starts his build *client* :P
19:40:09JdGordon|mc2739: cuesheet is working fine for me on my ipod video.... are you doing something wierd for it to not show the title?
19:41:12gevaertsshotofadds: no, only on sansa. The problem was that on sansa, the card_info thing is only initialised during read or write, not before, so using the "initialized" flag made things not work
19:41:28shotofaddsah. any chance you could fix it?
19:41:47mc2739JdGordon|: it shows title encoded in the id3 of the track. The cuesheet has different titles for different sections of the track - these titles are not shown now.
19:42:04JdGordon|these titles are shown for me...
19:42:56gevaertsshotofadds: I really didn't mean to get deep into drivers when doing this storage_* abstraction :\
19:43:13gevaertsAlso, I don't see why this would fail on the D2...
19:43:39 Quit einhirn ("Miranda IM! Smaller, Faster, Easier.")
19:44:41mc2739here is my cuesheet
19:45:27mc2739although it has not changed and did work before
19:46:32 Quit n00b81 ("Leaving")
19:47:26gevaertsshotofadds: it's basically doing what it did before FS #9545. disk.c used to just call card_detect()
19:47:36JdGordon|mc2739: does any of the info work?
19:48:09shotofaddsgevaerts: ok, I'll have a look to see what the TCC driver is doing differently. I thought it was functionally identical to the PP driver :/
19:48:09JdGordon|my test .cue doesnt have the FILE line, but that shuoldnt change anything
19:48:52mc2739I get the hash marks, but the album and artist are the same as in the tags
19:49:52JdGordon|is this the first song in the playlist?
19:51:10 Join tomcat_ha [0] (
19:51:21mc2739I played that file from a directory with two other files. It is the second file in the directory.
19:51:39JdGordon|what target?
19:52:12mc2739e200v1 and v2
19:53:07JdGordon|do the marks come up instantly?
19:54:31JdGordon|ok, well i'll have a look tonight, I have no idea why the marks are shoing but not the tags...
19:54:41mc2739I just changed the title and artist data and tested - it is reading the changed data, but when I skip to the second mark, the data does not change to that data - it stays on the track 1 data
19:56:07JdGordon|just to be sure... this is with the very latest build?
19:58:42*pixelma wonders what time is reported by the build client
20:01:40 Join SirFunk [0] (
20:01:47mc2739JdGordon|: Ok, sorry, maybe a false alarm. I just cleared settings and it looks like its working now
20:02:08*JdGordon| takes a sledgehammer to mc2739 :D
20:02:16JdGordon|ok, close the task then if it is actually working
20:02:18 Join readability [0] (n=chad@
20:02:33*mc2739 deserves that
20:03:12JdGordon|would anyone complain if the cuesheet optoin was removed from swcodec? so it is always enabled? I think I can get its ram usage to 0 if the playlist tracks dont have a .cue
20:03:24JdGordon|hwcodec would keep the option though
20:04:22mc2739is the cuesheet cached somewhere that survives reboots?
20:04:51PaulJam_Llorean: concerning your comment in FS #10457, the manual says in chapter 7.13. "Also the information found in the cuesheet file will replace the information from the ID3 tags."
20:05:45shotofaddsgevaerts: I don't know how to fix this, and I hadn't intended to spend the evening debugging :/
20:10:12JdGordon|mc2739: no
20:11:34 Join petur [50] (n=petur@rockbox/developer/petur)
20:12:04 Quit funman ("strstr(0,0)")
20:12:49JdGordon|Llorean: yeah, the cuesheet tags do overwrite the id3 tags for performer and track title
20:17:30 Quit readabil1ty (Read error: 110 (Connection timed out))
20:21:32 Join mc2739_ [0] (
20:22:24 Quit mc2739 (Nick collision from services.)
20:22:26 Nick mc2739_ is now known as mc2739 (
20:22:31 Quit r0b- (Read error: 110 (Connection timed out))
20:31:47 Join fdinel [0] (
20:31:54 Quit mc2739 ("ChatZilla 0.9.85 [Firefox 3.0.11/2009060215]")
20:32:48 Quit tomcat_ha ("If visible identification is not possible, the pathologist may be able to take fingerprints from the body until decay sets in)
20:33:26 Quit KBH (Read error: 104 (Connection reset by peer))
20:33:40 Join HBK [0] (
20:34:51 Quit muesli ()
20:35:43 Join gb_master [0] (
20:37:50 Part gb_master
20:41:04 Quit FlynDice (Remote closed the connection)
20:41:50 Join tomcat_ha [0] (
20:42:33gevaertsshotofadds: if it worked properly on d2 before r21981, maybe we can just revert that one for sd-tcc780x.c?
20:45:08*pixelma is going to test r21984 on the Ondio
20:45:40shotofaddsgevaerts: that would work in the meantime, yes. but right now I have a problem
20:46:14shotofaddswhile trying to spot differences between the PP and TCC drivers I copied in a couple of lines which were obvisouly wrong, and now my card won't power up at all
20:46:24shotofaddseven with a PC card reader :/
20:46:37shotofadds16Gb SDHC
20:46:47shotofaddsouch even more
20:46:55gevaertsindeed so :(
20:47:06 Join Thundercloud [0] (
20:47:14 Join Rand_Althor [0] (
20:47:26shotofaddsI only bought this one a few days ago to replace one I trashed a few weeks ago. Not happy...
20:47:55pixelmagevaerts: is roloing the same as booting with the new build (so I don't have to flash it and can easily return)?
20:48:59Rand_Althorgevaerts: Is there any way to use the c200erase tool on an external 4GB MicroSDHC card?
20:49:24gevaertspixelma: it should be the same, yes
20:49:36gevaertsRand_Althor: why would you want to do that?
20:49:43shotofaddsis there any kind of card status register I can access before the card has power up fully? or is OCR all I have to go on>
20:49:43pixelmaalso, I consider testing with the smaller card first after reading shotofadds' story
20:50:11shotofaddspixelma: unfortunately the driver was working fine this morning :(
20:50:28pixelmasad to hear that
20:50:52shotofaddsso i had a false sense of security.. but anyway my stupidity is off topic
20:51:01JdGordon|na :)
20:51:01*gevaerts thinks that this sounds like a case for the world famous Rockbox Fund
20:51:19Rand_Althorgevaerts: seem to have altered some important bit on mine, so now the computers' BIOS sees it as a hard disk, and my DAP can't see it.
20:51:45shotofaddsgevaerts: does it cover stupid people?
20:52:53shotofaddsI dont understand what would have caused this card not to power up any more. OCR reports 0xFF8000 as it always did, but the 'powered up' bit never gets set, even after waiting 1sec+
20:52:58gevaertsshotofadds: well, Zagor said it covered me when I managed to make early USB code read from SD and write to internal flash on my c200..
20:52:59LloreanJdGordon|: Someone should update the wiki page then, I wasn't aware WPS tags were even listed in the manual
20:53:16shotofaddsgevaerts: that's rather an impressive one, too ;-)
20:53:33gevaertsshotofadds: well, I did manage to recover it after a week or so :)
20:53:52*shotofadds will try a warranty replacement first
20:54:48LloreanJdGordon|: Why are they even the same tag?
20:55:06gevaertsRand_Althor: c200erase uses the standard rockbox SD driver to access the flash, so if your card doesn't show up over USB when plugged into a sansa running rockbox, adapting it won't help anyway
20:55:10LloreanShouldn't there be a cuesheet conditional? If there's cuesheet values, test the conditional to decide if they're used, so that people can keep using the ID3 values?
20:55:14JdGordon|because the cuesheet tags are literally copied onto the id3 tags...
20:55:20LloreanMuch like we do with the folder structure - let people choose rather than defaulting.
20:56:19JdGordon|adding cue tags could be done.. but a massive PITA
20:56:44LloreanWell, at the very least, the CustomWPS page is wildly inaccurate. Whoever added support for that feature should've updated it.
20:56:55 Quit Lear ("ChatZilla 0.9.85 [Firefox 3.5.1/20090715094852]")
20:57:23JdGordon|if i remember, ill fix it tonight... need to check the code to be sure which tags get spoofed
20:59:18bertrikmarkun, any luck with the database scan idea yet?
21:00:47JdGordon|Llorean: but yeah, not spoofing them would be better, and adding some cue specific tags would be a nice adition...
21:00:49***Saving seen data "./dancer.seen"
21:01:03 Quit LambdaCalculus37 (
21:01:10JdGordon|I'd like to see [1/3]/5 sort of thing for the playlist position
21:01:16 Join efyx_ [0] (
21:03:12 Quit flydutch ("/* empty */")
21:04:24 Quit Rand_Althor ("ChatZilla 0.9.85 [Firefox 3.5.1/20090715094852]")
21:04:46 Quit saratoga ("Page closed")
21:07:01pixelmagevaerts: card present on boot and hotswap seems to be working correctly. I just notice one oddity which is that the new build crashes with an old one's resume data, I admit that there are a few revisions between them though (r21323)
21:07:50pixelmaresume works with a resume point created with the new build (and even the old one accepts that)
21:08:11JdGordon|did someone fiddle with the ordering in settings_list.c and not bump the version number there?
21:09:11gevaertspixelma: I don't think that can really be caused by the storage changes. Maybe the cuesheet changes?
21:09:40JdGordon|na, shouldnt have
21:09:43 Quit barrywardell (Remote closed the connection)
21:10:10pixelmasounds like a candidate though ;)
21:10:40JdGordon|lies and slandar!
21:11:10pixelmaI could give you an address but that's with a backlight enabled build
21:11:44JdGordon|if you have the address then check your file
21:12:39pixelmaI would but I seem to always forgot how it goes
21:15:25JdGordon|open it and do a search for the first 6 digits or so
21:17:49pixelma" 0x0000000009011150 _cuesheet_subtrack_changed" seems pretty close to I09: CPUAdrEr at 09011168
21:18:36JdGordon|latest build?
21:19:18JdGordon|arg, I had that issue on my ajbr (yes amazingly i did some testing on it!) and thought I'd fixed it
21:20:55pixelmaI don't have cuesheet support enabled
21:22:54JdGordon|its working again now though?
21:23:36JdGordon|you are still getting that error?
21:23:48JdGordon|it didnt vanish with a new resume point?
21:24:18gevaertsshotofadds: sd on boot seems to be working for me...
21:24:32pixelmaJdGordon|: I only get the error with an old resume point in the new build, vanishes with a new resume point, yes
21:25:10gevaertsi.e. if I boot with SD inserted, I get the sd as root and the nand in <microSD1>. If I boot without SD inserted, I get nand as root, and SD in <microSD1> as soon as I insert it
21:25:12JdGordon|ok, not too woried about it then... curious why its getting there though
21:25:34shotofaddsgevaerts: I know. I just tried a couple of other cards and both worked with latest SVN. I think this card must have been dead before I updated.
21:25:40shotofaddsand that's rather more worrying
21:26:28gevaertsindeed. Eating cards is much worse than not booting every now and then
21:29:59shotofaddsalthough I actually haven't used Rockbox with this card today
21:30:27shotofaddsI copied some music to it this morning to listen to at work, then used the OF to listen to the radio for a few hours
21:31:21 Join derekja [0] (
21:33:11*shotofadds will add some warnings to the wiki/forum and wait for any reports of dead cards. :/
21:41:32gevaertshm, I added bits of yellow apparently
21:45:45 Part derekja
21:46:40CIA-69New commit by gevaerts (r21985): sd_first_drive is only used when HAVE_HOTSWAP is defined, so add appropriate #ifdefs
21:49:52 Join mc2739 [0] (
21:51:19 Join derekja [0] (
21:53:03gevaertsWe need more fast build clients...
21:53:05 Join Ubuntuxer [0] (
21:53:50derekjaI installed rockbox on a 60 gb ipod. At first I installed the 60gb version, however when I started having crashes, I uninstalled that one and installed the 30gb version. It now crashes less frequently, but still does. Why would this be?
21:55:11gevaertsYou didn't have crashes at first with the 64mb build?
21:55:38derekjagevaerts: I did have crashes with it.
21:55:55gevaertsright from the start, or only after a while?
21:56:05derekjagevaerts: right from the start
21:56:44gevaertsok. the 32mb build sounds like the right thing to try then, but then that one shouldn't crash...
21:56:44 Join Rand_Althor [0] (n=chatzill@
21:57:06derekjagevaerts: I tried that one, less crashes, but still some.
21:57:38Rand_Althorafter that it boots OF
21:57:55 Quit efyx_ (Client Quit)
21:58:44gevaertsderekja: I'd say that the most probable cause is a filesystem that's not entirely clean. I'd run chkdsk or fsck.vfat on it
21:58:57derekjagevaerts: thanks
22:01:25pixelmaJdGordon|: after flashing the new build, it crashes every time if I want to start playback somehow (same address), either starting a new playlist (tried to resume the old one first and then even deleted the playlist control file)
22:02:19pixelmaI don't know why that hasn't happened with the rolo'ed build
22:02:48JdGordon|I guess this shuold affect al hwcodec targets?
22:02:48gevaertsan initialisation issue?
22:04:26Rand_AlthorMy c200 can see my microSD cars, but my e200R cannot.
22:04:53pixelmaJdGordon|: I would guess so but can only test this one
22:05:10JdGordon|which one is this one?
22:06:21JdGordon|ok, can you stick it on FS? ill try to look tonight
22:06:52JdGordon|I saw this in my testing as got it working on my ajbr so i dunno whats happening
22:08:33mc2739JdGordon|: My cuesheet anomalies are back. It only happens when I switch to the widecabbie theme.
22:08:53*JdGordon| bangs head on keyboard
22:09:40mc2739don;t do that, you'll break the keyboard :)
22:10:33pixelmacuesheet support enabled/disabled doesn't make a difference and I also disabled voice to find out if that's something related (as I saw voice functions near in the map file)
22:10:53Rand_AlthorCorrection: On my e200R RB can't see the external card, but the OF can...?
22:11:23mc2739There must be something odd in that theme. All the other stock themes work fine.
22:11:31gevaertsRand_Althor: which revision are you running?
22:12:12JdGordon|amiconn: what does CPuAdrEr actually mean?
22:12:52gevaertsRand_Althor: does it see the card if you remove it and then put it back in?
22:13:06Rand_Althorwhile RB is running?
22:14:05gevaertsok. Then just upgrade to the latest build. That bug was solved in r21981
22:14:31Rand_Althorwhew. thankls
22:14:51 Join Zagor [242] (n=bjst@rockbox/developer/Zagor)
22:15:00Rand_Althor(sorry all the typos (smashed finger)
22:15:19JdGordon|pixelma : if you can be bothered... have a look in apps/cuesheet.c in function cuesheet_subtrack_changed() and just make sure it does if (cue) (or similar) around all the logic code?
22:15:31 Join B4gder [241] (n=daniel@rockbox/developer/bagder)
22:16:18pixelmait looks like widecabbie also draws into the main (full screen) viewport even though it's using more viewports
22:18:01ZagorI've started experimenting with using score/sec as client speed measurement, which lets me experiment with using a best-fit algorithm instead of the current model. it could theoretically save quite a bit of time.
22:18:13*pixelma always thought that this is potentially troublesome as it means overlapping viewports
22:18:19ej0rgegevaerts: thanks
22:20:56gevaertsZagor: sounds interesting! I still think (without much proof) that the current system can make things slower if there are too many slow clients
22:21:46Zagorgevaerts: yes it could. a best-fit could eliminate most of the speculative building
22:21:52 Quit BryanJacobs ("Java user signed off")
22:22:00*mc2739 puts out his hand for more speed
22:22:13pixelmaJdGordon|: there is a longish "if (cue && ...)" around most parts of that function, only "struct cuesheet *cue = id3->cuesheet;" ouside of it
22:23:04*gevaerts looks at the stats for the latest build round and tells Zagor that he won't be happy with any build that takes longer than three minutes ;)
22:24:39 Part Grahack
22:26:14Zagorgevaerts: we need more horsepower for three minute rounds. our workload is 1177729 points, and we work at a rate of ~4500-4700 points/second
22:27:23gevaertsZagor: does that include thrown away horsepower?
22:27:58Zagorno, that is the raw unwasted speed. (average of last 25 builds for 38 clients)
22:28:19Zagorexcluding upload times too, of course
22:29:27Zagorwith those numbers I hope to get steady ~270 second builds
22:29:34gevaertsAnyway, that looks like equivalent to 45 2.33GHz cores. Not bad...
22:30:04 Quit Rand_Althor ("ChatZilla 0.9.85 [Firefox 3.5.1/20090715094852]")
22:30:27Zagoryeah it's quite a "park"
22:30:43bertrikthis trying-to-reuse-the-tcc-usb-driver thing is confusing
22:30:50soapthis should be calibrated, not to 2.33GHz core2 cores, but to Pentium 66s.
22:32:10bertriksome of the current s5l8700 register defines don't match with the samsung datasheet, but seem to match the OF, not sure how they match with the tcc-usb code, or with the tcc-usb docs
22:33:22bertrikI would be happy to first just get some kind of indication from the host address it has seen *anything* (like a USB device connect event)
22:35:47JdGordon|pixelma: ok thats what I thought.. thats correct
22:38:37 Quit lyngaas ("Leaving.")
22:39:40Zagorgevaerts: with the 41-client setup from r21985 the round could be finished in 229 seconds. still a long way to 180...
22:41:01gevaertsZagor: indeed. I was misled by the 7655 cpu seconds (56%) for cancelled builds. Those are probably mostly slow CPU seconds...
22:41:25Zagoryes they are
22:41:48Zagoronce the slow clients finish their bootloaders, they are pretty much out of the game
22:42:25gevaertsyes. That's where better packing comes in. Have some of them do bootloaders, while others do full builds they can just manage...
22:42:29Zagorin contrast, the best-fit has monster-gevaerts doing playersim and satellite-marianne doing sa9200sim
22:43:31gevaertshm, does monster have sdl?
22:43:47 Join lyngaas [0] (
22:43:55*AlexP thinks the gevaerts after the monster- means that gevaerts might know that :)
22:43:56Zagorah well, it's a test matching only. i haven't included the arch limits in it yet.
22:44:29*Dhraakellian wonders how hard it would be to make Sudoku save possible values
22:44:57gevaertsinstalling sdl now :)
22:45:17linuxstbDhraakellian: You mean the scratchpad thing? IIRC, there's a patch to do that on flyspray.
22:45:44Dhraakellianscratchpad... that's a word for it
22:46:10linuxstbbertrik: markun told me he created s5l8700.h from the datasheet, so that's strange if they don't match...
22:46:20pixelmathat patch is quite old, don't know if it still applies cleanly
22:46:49JdGordon|it probably does
22:46:54JdGordon|plugins barely change
22:47:52pixelmawell, removal of the oldpluginapi caused a change in sudoku, and I also think the markers came after that patch
22:50:05bertriklinuxstb, well it does match the OF code better than the documentation does, so it was probaby intentional or maybe I don't have the proper datasheet
22:50:47 Quit n1s ("Lmnar")
22:51:07linuxstbbertrik: Which registers are you talking about?
22:51:46pixelmaI wonders whether I should also give my build client a user name with my nick so any mention would ping me here too
22:51:57bertrikof the USB PHY, for example URSTCON
22:52:17pixelmaor... set another highlight
22:52:52ZagorI ask new client owners to use their irc nicks as username, so I guess we should do it too
22:52:55linuxstbbertrik: That matches my version of the datasheet - 0x3c400008
22:53:17pixelmaok, I'll change that then
22:53:25gevaertsindeed. Those bjst clients are just obscure :)
22:54:37bertrikok URSTCON is as expected, but UCLKCON is not
22:54:47pixelmaof course, I start off with 0 points now again
22:55:18pixelmanot that it had a noticeable amount more before ;)
22:56:15Zagorpixelma: I'll fix that
22:58:58 Quit Ubuntuxer ("Leaving.")
23:00:50***Saving seen data "./dancer.seen"
23:00:58*B4gder is now officially planning to do a rockbox-related talk on FSCONS 2009
23:02:01pixelmagevaerts: already fixed but r21977 gives me a "No partition found. Insert cable and fix it." when I rolo it - with ot without MMC
23:02:28pixelmaI never saw that error before
23:02:50gevaertshm, good thing I changed that...
23:05:41 Join stripwax [0] (
23:06:29 Join ademille [0] (
23:07:53 Part ademille
23:13:52 Quit mc2739 ("ChatZilla 0.9.85 [Firefox 3.0.11/2009060215]")
23:15:14 Quit dash32 (Read error: 110 (Connection timed out))
23:15:45 Quit jgarvey ("Leaving")
23:15:55 Join dash32 [0] (
23:15:59 Join barrywardell [0] (
23:16:20 Join r0b- [0] (
23:18:31 Quit bmbl ("Bye!")
23:20:54 Join n00b81 [0] (n=n00b81@unaffiliated/n00b81)
23:23:13 Join jhulst_ [0] (
23:23:54 Quit jhulst_ (Remote closed the connection)
23:23:57 Join Stephen__ [0] (
23:28:54pixelmaI "still" have the CPUAdrEr with r21977 but updated ata_mmc
23:29:27 Join faemir [0] (n=daniel@
23:29:36pixelmaeven deleted .playlist-control
23:30:03JdGordon|comment out that whole cuesheet block
23:30:12CIA-69New commit by dave (r21986): Nano2G Backlight driver improvements - the previous version didn't work reliably.
23:30:55pixelmaof course the address is slightly different and seems to be in _gui_wps_update. JdGordon|: r21977 was from before your cuesheet commit
23:36:56pixelmaZagor: "Server message: Build round completed after 0 seconds." is it just me, or is there something broken in general?
23:37:11Zagorit broke the server. fixing now...
23:38:04Zagorheh, still broken...
23:41:08r0b-what are cue sheets good for in rockbox?
23:41:37AlexPRockbox just has support for cue sheets
23:41:42 Join wincent [0] (
23:41:44AlexPThey aren't a Rockbox pecific thing
23:41:46pixelmaJdGordon|: I start wondering whether it's something different but which has to do with WPS drawing (maybe progressbar drawing) so that it just hits in cuesheet niw
23:41:51pixelmanow too
23:42:31 Join safetydan [0] (n=deverton@rockbox/developer/safetydan)
23:43:39r0b-well why support them in rockbox?
23:43:55gevaertsbecause people want to use them?
23:43:57AlexPWhy support MP3?
23:44:24AlexPIt is a bit of a silly question to be honest
23:45:28 Quit Lynx_ (" HydraIRC -> <-")
23:45:33 Quit petur ("battery empty so Zzzz now")
23:45:49 Quit faemir (Nick collision from services.)
23:49:11 Join faemir [0] (n=daniel@
23:57:53 Quit bertrik ("De groeten")
23:58:16stripwaxsaratoga - (for the logs) - dunno how reliable this is, but here's your cook profile.

Previous day | Next day