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 2010-04-03

00:00:02Lloreangevaerts: I don't know how this works. Is there an explicit request for them, or are they given to the host as part of something else, or?
00:00:13gevaertsthey're part of the device descriptor
00:00:24gevaertsyou can't leave them out
00:00:32Lloreanlinuxstb: Well, iPods have never been drm-only content, so his guess that they may want to require someone to buy the "iPod kit" may not be too far off the mark.
00:01:37gevaertsThe only precedent I see is that on some other consumer devices, it's possible to change these numbers in an obscure database called the "registry" or some similar confusing name
00:02:20 Join Strife89 [0] (
00:03:14 Part Llorean
00:03:40 Join Llorean [0] (~DarkkOne@rockbox/user/Llorean)
00:04:27Lloreangevaerts: What sort of options does that usually have?
00:04:29 Join troll [0] (
00:05:23 Quit dfkt (Read error: Connection reset by peer)
00:05:38 Join jennifur [0] (
00:06:16 Quit troll (Client Quit)
00:06:26gevaertsVID/PID and classes to support I think. It's not at all meant for end users to change, but you can do it
00:06:37 Join troll-brasil-c20 [0] (
00:07:05troll-brasil-c20hello put the bios to boot rockbox but it does not start only the original will?
00:08:20gevaertsYou mean the bootloader? Which device is this?
00:09:44*Strife89 guesses at "c200". Perhaps a v2.
00:10:00troll-brasil-c20so the bootloader but the screen stays black but only bind holding>> starts the original sansa
00:10:07 Quit preglow_ (*.net *.split)
00:10:07 Quit blithe_ (*.net *.split)
00:10:07 Quit Farthen (*.net *.split)
00:10:07 Quit stavrob_ (*.net *.split)
00:10:24gevaertsIs rockbox itself installed?
00:11:56troll-brasil-c20botei asend manually put the buttons on and off normally but nothing appears on the screen and
00:12:26 Join preglow [0] (
00:12:39 Join stavrob [0] (
00:13:08 Join Farthen [0] (
00:13:26Strife89troll-brasil-c20: THere are two parts that you need to install Rockbox: the bootloader, and Rockbox itself.
00:13:47Strife89troll-brasil-c20: It sounds like you've only installed the bootloader.
00:14:11Strife89troll-brasil-c20: Did you attempt to install Rockbox manually, or did you use Rockbox Utility.
00:14:17 Quit xiainx (Ping timeout: 276 seconds)
00:15:02troll-brasil-c20Rockbox manually
00:15:44troll-brasil-c20Rockbox manually because it does not support sansac200v2
00:15:56 Join blithe_ [0] (~blithe@
00:15:58Strife89troll-brasil-c20: Okay, then let's try something. Connect your device to the computer.
00:17:44Strife89troll-brasil-c20: Get this zip file:
00:18:24 Quit troll-brasil-c20 (Quit: CGI:IRC (EOF))
00:18:33 Join xiainx [0] (
00:20:07 Join troll-brasil-c20 [0] (
00:20:16 Quit blithe_ (*.net *.split)
00:20:16troll-brasil-c20Download it
00:20:55Strife89troll-brasil-c20: Okay, now open the folder that you downloaded it to.
00:21:39 Join Luca_S [0] (
00:22:10Strife89troll-brasil-c20: Okay, right-click on the file and click "Extract Here".
00:22:26Strife89troll-brasil-c20: You must right-click on the .zip file itself.
00:23:04*Luca_S just started a battery bench on fuzev2, using the lame_128 test file looping
00:23:24Strife89troll-brasil-c20: Okay, it should have made a new folder called "rockbox-sansac200". Open it.
00:23:38Strife89troll-brasil-c20: Inside you will see a folder called ".rockbox".
00:24:00Strife89troll-brasil-c20: Copy ".rockbox" to your Sansa.
00:24:05 Join blithe [0] (~blithe@
00:24:25Strife89Is the folder copied?
00:24:46 Quit phanboy4 (Read error: Connection reset by peer)
00:25:09 Nick froggyman is now known as froggyperson (~me@unaffiliated/froggyman)
00:25:17 Part froggyperson
00:25:45Strife89troll-brasil-c20: Is ".rockbox" in the root of the Sansa?
00:25:59Strife89troll-brasil-c20: Use Safely Remove Hardware to make it safe to unplug your Sansa.
00:26:21Strife89Then, once Windows says it's safe, unplug your Sansa.
00:26:53troll-brasil-c20unplug Sansa
00:27:01Strife89troll-brasil-c20: Now, turn off your Sansa, then turn it back on.
00:27:12Strife89troll-brasil-c20: Don't hold Left when you turn it on.
00:27:35Strife89troll-brasil-c20: If everything has been done correctly, Rockbox should start.
00:28:16troll-brasil-c20the screen is black and light blue on the menu is weak and then off
00:29:21troll-brasil-c20to squeeze a button it lights up but nothing appears on the display
00:30:22*linuxstb wonders if this is why the c200v2 port is classed as "unusable"
00:30:32troll-brasil-c20I have an 8gb memory card for it is beyond the sansa should I put them in?
00:31:12 Quit JohannesSM64 (Quit: WeeChat 0.3.2-dev)
00:31:18Strife89troll-brasil-c20: If you did everything right, and the Sansa is indeed not booting Rockbox, then it is possible that Rockbox is just not ready for your type of Sansa.
00:31:41 Join JohannesSM64 [0] (
00:31:42Strife89troll-brasil-c20: And Rockbox does support 8GB cards.
00:31:45troll-brasil-c20To do this it need not be formatted alright because it has space available
00:33:12Strife89troll-brasil-c20: You should not need to format the Sansa or the memory card.
00:33:48troll-brasil-c20but has evolved considerably since the first time I did once he died now he is already giving dual boot:)
00:34:12Strife89troll-brasil-c20: Indeed, there is progress. :)
00:34:31Strife89troll-brasil-c20: With that said, it seems that Rockbox is not ready to run on your type of Sansa. It may be better to remove it for now.
00:34:51 Join froggyman [0] (~me@unaffiliated/froggyman)
00:35:19Strife89troll-brasil-c20: Would you like to remove Rockbox for now?
00:36:14troll-brasil-c20irei remolver mas vo tar sempre de olho e testando sempre que novas versões forem saindo
00:37:12troll-brasil-c20and delete and do not?
00:37:35troll-brasil-c20but I will remolver vo tar always watching and always testing new versions are coming out
00:38:33Strife89troll-brasil-c20: Are you saying that you'd prefer to keep Rockbox on your Sansa, and just keep testing new versions?
00:40:03troll-brasil-c20so you guys know the Sansa supports all version of bios 3:02:05 indepedent of the country so that the European deiche without fm
00:41:36troll-brasil-c20I'll try to put the Rockbox whenever you update out until a stable version
00:41:52gevaertsthat's probably the best thing to do
00:43:43troll-brasil-c20us here in Brazil we will be happy with that because here come few and nei sansa but has to sell and tamos eye on updates
00:46:24 Quit troll-brasil-c20 (Quit: CGI:IRC (EOF))
00:47:14 Join troll-brasil-c20 [0] (
00:48:17troll-brasil-c20the simulator runs well but for now the unit will not:)
00:50:24troll-brasil-c20Thank you for trying to help and good luck in advance
00:51:48troll-brasil-c20goodbye and see you soon
00:54:05 Quit troll-brasil-c20 (Quit: CGI:IRC (EOF))
00:54:48 Join anewuser [0] (anewuser@unaffiliated/anewuser)
00:55:20 Quit Luca_S (Quit: CGI:IRC (EOF))
00:57:22 Join webguest05 [0] (
00:57:23 Quit blairb (Ping timeout: 240 seconds)
00:57:34 Quit webguest05 (Client Quit)
00:57:43 Quit xiainx (Ping timeout: 240 seconds)
00:57:49 Join blairb [0] (
00:59:35 Part froggyman
00:59:50 Join xiainx [0] (
01:01:32***Saving seen data "./dancer.seen"
01:03:28 Join fml [0] (
01:04:07fmlThe last commit to the manual misses a file, I think (hotkeys.tex).
01:04:48gevaertshm, that might be better on the mailing list so he sees it...
01:05:32fmlMaybe. But I'l leave in a moment. The missed file is referenced in "browsing_and_playing.tex"
01:05:54 Quit fml (Client Quit)
01:06:31pixelmahuh... maybe that is breaking my goban.tex patch and not something wrong I did
01:11:14*gevaerts sent the email
01:11:35 Quit xiainx (Ping timeout: 240 seconds)
01:18:37 Quit stripwax (Quit:
01:19:07 Quit geertvdijk (Ping timeout: 252 seconds)
01:19:21 Quit bertrik (Quit: De groeten)
01:19:24 Join FlynDice [0] (~FlynDice@
01:20:44 Join _deXter [0] (~dexter@
01:30:17 Quit efyx (Remote host closed the connection)
01:30:17 Quit Farthen (Remote host closed the connection)
01:30:53 Quit jd (Read error: Connection reset by peer)
01:31:31 Quit CGL (Remote host closed the connection)
01:31:33 Join jd [0] (
01:31:33 Quit jd (Changing host)
01:31:33 Join jd [0] (~jd@Wikipedia/HellDragon)
01:32:21 Part _deXter
01:32:23soapin all seriousness, gevaerts / Llorean, what is the /harm/ in Rockbox shipping with the manufacturer USB IDs, but making their values accessible for change deep in the debug menu?
01:32:27 Join _deXter [0] (~dexter@
01:32:56gevaertssoap: if that's the best option I can live with it
01:33:58soapWell, I agree with you that the "best" option is for motherhating manufactures not to abuse their consumers and put in stupid locks - but that is something we have no power over.
01:34:58 Join Blue_Dude [0] (
01:35:26Blue_Dudefml: Working on the missing file. Sorry.
01:36:24soapThe three issues I see are (cough) binsize/complexity, lying to RBUtil, and provoking the wrath of the mysterious USB overlords. The third issue being likely moot as it wasn't like this can't be (slightly less easily) changed already.
01:37:10CIA-5New commit by Blue_Dude (r25443): Manual update for hotkeys
01:37:44gevaertsI think that hiding it in the debug menu without any persistence isn't really a service to the people who need it though
01:37:45Blue_DudeForgot to add new manual files to svn. Fixed now.
01:37:52soapFourth issue being the possible breaking of some Apple accessories.
01:38:36 Join linuxguy3 [0] (
01:38:48gevaertsI'd suggest a hidden setting, but we Don't Do Those, right?
01:38:55soapDidn't realize none of the debug entries have persistence.
01:39:03soapWas about to say almost exactly the same thing.
01:40:34soapOn a wholly unrelated subject - am I the only one who got the PM from sixthofmay? If not would someone who has actually been following said thread care to follow up?
01:45:04 Quit JohannesSM64 (Ping timeout: 248 seconds)
01:47:01 Quit FlynDice (Ping timeout: 264 seconds)
01:47:26 Quit Blue_Dude (Quit: ChatZilla 0.9.86 [Firefox 3.6.2/20100316074819])
01:47:47 Join FlynDice [0] (~FlynDice@
01:48:42 Quit FlynDice (Remote host closed the connection)
01:48:50 Join FlynDice [0] (~FlynDice@
01:52:23 Join kugel [0] (~kugel@rockbox/developer/kugel)
01:52:44 Quit FlynDice (Read error: Connection reset by peer)
01:53:06 Join FlynDice [0] (~FlynDice@
01:53:07kugelah great, I can use an irq for the fuzev2 wheel
01:54:21kugelno more polling hack!
01:54:47 Join liar [0] (
01:58:06 Join Hillshum [0] (
01:59:36linuxstbgevaerts: Talking about USB, I don't suppose Rockbox gets any info about the host before it sends the IDs?
01:59:55 Quit FlynDice (Ping timeout: 268 seconds)
02:00:11gevaertsThe device descriptor is the very first thing the host asks for
02:02:56 Join JohannesSM64 [0] (
02:03:45Lloreansoap: I've got nothing at all against hiding it in the debug menu.
02:04:55 Join gnathos [0] (
02:05:16gnathosis theseven here?
02:07:32*gevaerts looks around
02:07:34gevaertsapparently not
02:10:02Strife89!seen TheSeven
02:10:21Strife89Wait, no bot. *facepalm*
02:26:25 Part domonoky
02:29:15 Join xiainx [0] (
02:37:25 Quit liar (Ping timeout: 258 seconds)
02:37:33 Quit mt (Remote host closed the connection)
02:40:06 Quit linuxstb (Ping timeout: 258 seconds)
02:46:43 Quit kugel (Remote host closed the connection)
02:50:47 Join liar [0] (
02:56:20 Join CGL [0] (~CGL@
03:00:36 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
03:01:35***Saving seen data "./dancer.seen"
03:06:55 Quit liar (Ping timeout: 258 seconds)
03:09:31 Join liar [0] (
03:11:41 Quit linuxstb (Ping timeout: 265 seconds)
03:23:14 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
03:23:16 Join jeffp [0] (
03:26:41 Quit Schmogel (Ping timeout: 240 seconds)
03:30:34 Part jeffp
03:31:48 Join FlynDice [0] (~FlynDice@
03:35:14 Quit panni_ (Quit: ( :: NoNameScript 3.81 :: ))
03:36:49 Quit liar (Ping timeout: 258 seconds)
03:37:21 Join liar [0] (
03:37:31 Quit CGL (Ping timeout: 246 seconds)
03:41:02 Quit FlynDice (Read error: Connection reset by peer)
03:41:12 Join FlynDice [0] (~FlynDice@
03:45:46 Quit FlynDice (Read error: Connection reset by peer)
03:49:43 Join FlynDice [0] (~FlynDice@
03:50:08jhMikeS"linuxstb jhMikeS: Do you think you'll be around Rockbox this summer and able to help out with any mpegplayer SoC project? " <−− most likely
03:50:57jhMikeS"n1s jhMikeS: do you have plans for the voltage/freq scaling on the beast+ " < hell yes, and it's a huge PITA to unravel Freescale's junk on the matter :)
03:53:54 Join the_miker [0] (
04:05:42 Quit Barahir (Ping timeout: 260 seconds)
04:05:53gnathosyo anyone seen theseven?
04:07:00 Part gnathos ("Leaving")
04:07:06 Join Barahir [0] (
04:10:17 Join gump [0] (
04:13:39gumpanybody on that is familiar with the ipod nano 2g?
04:14:16S_a_i_n_t_what's your question?
04:14:27gumpi cant get the rockbox bootloader off of it
04:14:44S_a_i_n_t_define please.
04:14:53S_a_i_n_t_what have you tried, what errors are you getting?
04:14:59gumpthe apple firmware functions perfectly but the rockbox bootloader loads every time i start it up and says cannot find rockbox.ipod
04:15:11gumpi have reformatted it many times
04:15:31gumpit says boot to disk mode when i get that error so i do, and when i try to reformat it says the device is read only
04:15:31S_a_i_n_t_have you tried using ipodpatcher to remove the bootloader?
04:15:53gumpno. ive used itunes and the hp reformat tool
04:16:09gumpit will only reformat when i start it with the hold switch on
04:16:13gumpotherwise it says it is read only
04:16:54S_a_i_n_t_Try ipodpatcher
04:17:29gumpalright. i'll have to boot to windows, im in linux right now. brb
04:17:44 Quit gump (Quit: Leaving)
04:22:54 Quit FlynDice (Read error: Connection reset by peer)
04:27:31 Join gump [0] (
04:28:31S_a_i_n_t_gump: You left too quickly, there *is* a linux binary for ipodpatcher
04:28:48S_a_i_n_t_but you can get the all of them here
04:28:51gumplol yes i noticed that just now
04:29:31gumpwow. it worked.
04:29:39gumpive been trying to fix this for weeks
04:29:54gumpgonna try and restore it now
04:29:58S_a_i_n_t_And *that's* how you uninstall the bootloader ;)
04:30:16S_a_i_n_t_No worries man.
04:30:30S_a_i_n_t_iTunes restore should work fine now.
04:30:51gumpso what do you think the problem was with rockbox?
04:30:55gumpshould i try installing it again?
04:31:13S_a_i_n_t_What happened exactly?
04:31:44 Join Rob2223 [0] (
04:31:45gumpi installed it and it worked fine. it was processing all the tags, then i restarted it and the bootloader said "could not load rockbox.ipod switch to disk mode"
04:31:57gumpand from then on i couldnt format it or anything
04:32:22S_a_i_n_t_Hmmmmmm, seems a little odd.
04:32:23gumpit didnt even show up as an ipod when in disk mode
04:32:33S_a_i_n_t_My advice would be to put Rockbox back on it.
04:32:52S_a_i_n_t_I have 2 RB'd Nano2g's...and they both work fine.
04:33:11S_a_i_n_t_But it is still an "unstable" build, so it has a few hiccups now & then.
04:33:35gumpwell i will try it then
04:35:27 Quit Rob2222 (Ping timeout: 265 seconds)
04:36:04S_a_i_n_t_If you need to ask a question during install, just ask...
04:36:10S_a_i_n_t_Better safe than sorry ;)
04:48:36S_a_i_n_t_Wooo! "Symmetry" for iPod Nano 1/2G is *finally* finished! (I *think*...)
04:52:19S_a_i_n_t_The newest version is on the Theme Site...
04:55:14 Quit gump (Quit: Leaving)
05:00:58 Join Boldfilter [0] (
05:01:39***Saving seen data "./dancer.seen"
05:04:12 Quit Barahir (Ping timeout: 264 seconds)
05:04:55 Join n1s [0] (~n1s@rockbox/developer/n1s)
05:06:19 Quit _deXter (Quit: Lost terminal)
05:06:52 Quit GeekShadow (Quit: The cake is a lie !)
05:10:35 Join FlynDice [0] (~FlynDice@
05:14:56 Quit FlynDice (Read error: Connection reset by peer)
05:17:16 Join FlynDice [0] (~FlynDice@
05:24:05 Join krazykit` [0] (
05:26:25 Part jennifur
05:28:08 Quit krazykit (Ping timeout: 276 seconds)
05:30:43 Join mandred [0] (
05:33:04 Join funman [0] (~fun@rockbox/developer/funman)
05:37:09 Join jennifur [0] (
05:37:29CIA-5New commit by nls (r25444): 'floor 0' files requiríng too much memory should no longer crash since the longjmp stuff
05:39:33 Quit FlynDice (Read error: Connection reset by peer)
05:40:16 Join FlynDice [0] (~FlynDice@
05:43:08FlynDicefunman: How's your clip+ working now? I had problems reading uSD all day today and finally have a chance to look now.
05:43:35funmanFlynDice: uSD has some problems, also reported on ABI forums; likely due to changed CPU speed
05:43:57funmanalthough it's random: sometime insert uSD => panic, sometime i can play a bit from it
05:44:19jennifurI've been getting a panic when I play from the uSD
05:44:29funmanI didn't test extensively since i'm still running a battery bench (now at 13hours, if it goes beyond 15h30 i'll switch the CPU normal freq to 24MHz
05:44:29jennifurI've the fuzeV2 though.
05:44:36FlynDiceI was going to try some delays after setting divs, OF seems to do that ~ 40 nops
05:45:14jennifurIf there is anything I can help test out, let me know.
05:45:31funmanFlynDice: i suggest wait until my bench is finished, so you dont' have to do the same thing 2 times
05:45:38 Quit JohannesSM64 (Max SendQ exceeded)
05:45:43funman120 -> 60/240MHz and then 60/240 -> 24/240
05:45:57FlynDiceI've got 25 hours while Im in Atlanta to split between rockbox and getting the new laptop cleaned up
05:46:27FlynDiceI'll just see if it helps for now...
05:48:22FlynDiceThats amazing that it will function at 24 mhz.
05:49:46 Quit anewuser ()
05:49:51funmanFlynDice: pclk (still at 60) can be different from fclk, but I booted with much lower freqs (down to 12MHz i think)
05:49:57funman(pclk freqs)
05:51:59funmanI'm much more impressed by the impressive battery life
05:53:17funmanIf we can dump the as3514 (well as3517) registers when the OF is running it could point us to the "suck the battery juice" bit in AMSv1
05:56:37funmanbtw kugel sent me a fuzev2 and i'll buy a clipv2 today so I'll have the 3 amsv2 targets to compare
05:58:48 Quit FlynDice (Ping timeout: 264 seconds)
05:59:10 Join FlynDice [0] (~FlynDice@
05:59:39 Join Falco98 [0] (
06:00:18Falco98Torne: u around?
06:01:54 Quit FlynDice (Read error: Connection reset by peer)
06:02:16 Join FlynDice [0] (~FlynDice@
06:02:29Falco98hm. Can anyone tell me where I can download the 3.4 release? Everything I'm trying leads to dead-ends. It's not listed on the old releases archive page.
06:02:50Falco98I'm trying to debug something, as I think 3.5 introduced a bug to the 5.5g ipod
06:05:20 Quit n1s (Ping timeout: 265 seconds)
06:05:51funmanFalco98: , but did you try a current build to see if the bug was still there?
06:06:29Falco98i've just gone thru the current svn as well as 3.5.1, i had been running an unmodified version of 3.5.0, all three have the same issue
06:06:40Falco98that's why i'm after 3.4, i don't remember having the problem prior to upgrading
06:06:52 Join tomers [0] (
06:07:08Falco98thx for the link... i hadn't found taht by searching any of the main pages
06:07:38S_a_i_n_t_Falco98: What *is* the issue you're having?
06:08:00 Join n1s [0] (~n1s@rockbox/developer/n1s)
06:08:28Falco98Plugging in USB from a powered-off state gets me only as far as the graphic of a USB cable with the words "multimedia mode" below it; no data connection is established with my PC though.
06:09:15Falco98it used to show this graphic only for a moment before resetting into the "do not disconnect" screen (which seems to be from the original firmware, not sure).
06:09:51Falco98in any case, I don't remember having this problem before upgrading (though I could be wrong, it could be a wonky ipod, hence why i'm checking)
06:10:37S_a_i_n_t_Have you tried diasbling USB HID?
06:10:58funmanand what OS is your PC running?
06:11:28Falco98saint, can you specify what HID is?
06:11:39Falco98is that something i would disable on the PC end or on the ipod end?
06:12:00S_a_i_n_t_settings/general settings/USB HID
06:12:04S_a_i_n_t_*I think*
06:12:08Falco98what is it?
06:12:12S_a_i_n_t_Oh, and its on the ipod.
06:12:17S_a_i_n_t_in rockbox settings
06:12:34Falco98but like.. i mean, what's it do
06:12:41S_a_i_n_t_It has something to do with how the ipod enumerates to the host.
06:12:52tomersFalco98: I makes your DAP, when connected to USB, to not only be identified (enumerated) as a mass storage device (a.k.a disk-on-key...)
06:13:06tomersbut also as a keyboard, and a mouse
06:13:07Falco98well.. when this happens, it doesn't appear to get ANY data connection to the PC. at all.
06:13:26S_a_i_n_t_Find it, check if its diasbled or not, if its not, diasble it and see if you can transer data/get a connection.
06:13:28Falco98other times, like when it works "right", it has absolutely no issues.
06:13:42Falco98well i just downgraded to 3.4 though, so i will explore that first :-P
06:14:19tomersthen, when the DAP is connected to the computer, the player's buttons, which were once unusable, are now mapped to send keyboard strokes or mouse movements to the PC the player is connected to. e.g. pressing Vol+ on the player is like pressing Vol+ on a multimedia keyboard
06:14:50Falco98oh... is that why, when i installed a daily build a few minutes ago, my PC popped up with a "new hardware found" wizard?
06:14:58Falco98(and the ipod's status screen said "keypad mode")
06:14:58S_a_i_n_t_I haven't heard of this on a 5.5G iPod, but the Nano2G has some pretty big issues with USB HID on for some people.
06:15:08S_a_i_n_t_yep, that'll be it.
06:15:28tomerssome old operatin systems (OSX of some older version, which I can't remember) have bug in USB handling, which prevent the player from being attached to the computer. You must disable USB HID if you want to connect your player to a computer that runs on of these older operating systems
06:15:31Falco98where would one get the drivers to satiate that NHFW though?
06:15:35Falco98i had to cancel out of it
06:16:06tomersYou don't need any 3rd party drivers
06:16:11S_a_i_n_t_On my Nano1Gs, HID works fine...On the Nano2Gs I have, I can't turn HID on at all and establish a connection to the PC
06:16:20tomersThe OS should use its default keyboard and mouse drivers
06:16:38Falco98i might try it again at some point but it definitely refused to install automatically
06:16:43Falco98but i also didn't try very hard
06:16:54tomersMaybe you should refresh the driver, i.e. delete the player in Device manager (Windows) and then disconnect and reconnect the player
06:17:02Falco98like i didn't try selecting "automatically install" or anything, i just cancelled
06:17:33S_a_i_n_t_tomers: I also get HID issues in XP, with the Nano2G. I just *can't* have HID turned on and still connect to the PC
06:17:46tomersAFAIK Nano2G is unstable, so think doesn't always work for it
06:18:19Falco98sigh.. i hate putting so many shutdown/startup cycles on my poor ipod doing all this testing :-P
06:18:23S_a_i_n_t_Yes, it is unstable...I *think* HID is disabled in the 2g Nano now anyway.
06:18:27S_a_i_n_t_For that reason.
06:18:41tomersUSB HID requires use of USB Interrupt transfers, which require that functionality to be implemented in the USB driver in Rockbox. I'm not sure this is implemented well for the nano 2g
06:19:55Falco98crap.. 3.4 didn't solve my issue either :(
06:20:03Falco98let me try disabling HID and see if that makes any difference
06:20:56tomersFalco98: Look at firmware/export/config.h:880 - USB_HAS_INTERRUPT is commented out −− seems to be broken
06:21:24tomersSo I wonder how you have the USB HID option at all - it shouldn't be shown as it is disabled
06:21:30tomersunless you uncommented that line
06:21:31Falco98i'm not finding it
06:21:41 Quit FlynDice (Read error: Connection reset by peer)
06:21:50 Join FlynDice [0] (~FlynDice@
06:21:56Falco98but still, when i plug in USB from a poweroff state, it comes up and only gets as far as that shiny picture of a USB plug
06:22:17Falco98it never resets into the DND screen unless i unplug the usb cable (lol)
06:22:32S_a_i_n_t_DND screen?
06:22:38Falco98"do not disconnect"
06:22:47Falco98the screen at which it makes a USB connection with the PC
06:22:58Falco98i don't know the official name for it sry :-P
06:23:51tomersFalco98: Can you please answer these questions: How do you have HID in 2g Nano? It is disabled by default, since it is broken as of now. Did you change anything in the code?
06:23:51 Quit FlynDice (Read error: Connection reset by peer)
06:23:54S_a_i_n_t_Isn't it *supposed* to go to the 'Rockbox USB Screen' on cable connect?
06:24:00 Join FlynDice [0] (~FlynDice@
06:24:17S_a_i_n_t_Or is it still supposed to go into disk mode for the release builds?
06:24:20Falco98tomers: I'm not sure I can answer that, since i'm running a 5.5g
06:24:52S_a_i_n_t_tomers: you're going off on a tangent man...
06:24:58Falco98saint: i've always experienced it to go into disk mode. the "usb cable" screen doesn't give me any connectivity.
06:25:07tomersFalco98: sorry, I guess i got confused
06:25:12S_a_i_n_t_I was just giving back-story on HID issues, using the Nano2g as an example.
06:25:43Falco98no prob, i'm not exactly being uncomplicated about this
06:26:12S_a_i_n_t_Falco98: can yout not find the HID settings?
06:26:34Falco98i'm in 3.4 now - it doesn't seem to have an HID setting (as far as i can see?)
06:27:26Falco98or, "yes" is my answer :-P
06:27:32S_a_i_n_t_If it does, it'll be in "settings/General Settings/"
06:27:37Falco98as in "yes, i cannot find them"
06:27:53Falco98i see nothing naming "USB" at all in general settings.
06:28:19S_a_i_n_t_Hmmmm, I thought that 3.4 had this as an available setting.
06:28:25S_a_i_n_t_Must be wrong.
06:28:58S_a_i_n_t_Besides the point however, in a release build it *should* be going into disk mode on cable connect.
06:30:40S_a_i_n_t_I can't think of a reason why it wouldn't to be honest, it may be a stupid question, but it is *definitely* a release build, and not a daily build...right?
06:30:49Falco98i'm running 3.4 atm
06:30:51Falco98so yes
06:30:58tomersFalco98, S_a_i_n_t_: I added the usb_hid setting in r23322
06:31:14Falco98when i plug usb in when it's already powered on, it gets to the disk mode screen with no problem
06:31:33Falco98only from powered-off does it hang at the USB icon screen
06:31:33tomersr23322 | tomers | 2009-10-23 15:29:19 +0200 (Fri, 23 Oct 2009)
06:31:45Falco98i'm starting to think, though, that this may just be my ipod being weird
06:31:50 Quit FlynDice (Ping timeout: 260 seconds)
06:32:04S_a_i_n_t_It doesn't seem like a hardware fault.
06:32:05Falco98which is a little frustrating
06:33:14Falco98oh, get this... when i installed torne's revision of the svn build, when i plugged in USB from an OFF state, i got the USB icon plus the song title from the WPS screen stayed on screen, and continued scrolling, in the middle of nowhere
06:33:18S_a_i_n_t_I'm just not sure if its "desired behaviour" that the iPod should go into Rockbox's USB screen from connecting when the power is off or not.
06:33:21Falco98i wanted to take a picture of it
06:33:43S_a_i_n_t_Weird ;)
06:33:46Falco98well i wouldn't mind it if it actually got a data connection with the USB screen
06:34:40 Quit Boldfilter (Quit: Boldfilter)
06:34:48S_a_i_n_t_My suggestion would be to go back to 3.5/3.5.1 (if you have the buffering issue) and see if disabling USB HID gives you connectivity in the Rockbox USB screen.
06:38:01 Join TillW [0] (
06:38:01 Join Boldfilter [0] (
06:39:17Falco98i think i'll use the modified version of 3.5.1 posted here though,
06:39:50S_a_i_n_t_the startup issue fix?
06:40:10 Join saratoga_lab [0] (~9803c20d@gateway/web/freenode/x-uzqqrbupkqksjwwy)
06:40:14 Join phanboy4 [0] (
06:40:20Falco98i thought startup was right
06:40:34Falco98the issue was they intermittently refuse to start up without being hard reset
06:40:43S_a_i_n_t_Well, they both are kinda
06:40:44Falco98the original solution introduced some errors
06:40:56Falco98i figured my current issue may have been introduced with the original workaround
06:41:03Falco98as well as the Scion ipod dock issue
06:41:04S_a_i_n_t_yes, however this fix seems *very* prominsing.
06:41:39Falco98*crosses fingers*
06:41:49 Quit arbingordon (Ping timeout: 240 seconds)
06:41:52Falco98though i already tried this fix and my startup/usb issue seemed to remain :-P
06:42:06CIA-5New commit by funman (r25445): as3525v2: disable interrupts when writing to the PMU registers ...
06:43:54S_a_i_n_t_This fix for the startup/shutdown issue is *completely* different to the initial one from the same thread.
06:44:24S_a_i_n_t_And it seems to be working very well for those who are testing it currently, though I only know of two people doing so.
06:44:39S_a_i_n_t_3 now, counting yourself.
06:45:12Falco98well other than the scion issue (and possibly this USB poweron issue) i'm not sure i had any problems with the original workaround
06:45:20Falco98i never had the random-poweron issue as far as i know
06:45:40Falco98but plugging it into the scion dock from a powered-off state got me a "connection error" message
06:47:25S_a_i_n_t_I was right, USB HID *is* in Settings/General settings...just not in 3.4 ;)
06:47:28Falco98btw i still don't see USB HID in settings
06:47:36Falco98i'm running 3.5.1 again
06:47:58Falco98just confirmed thru system info
06:48:41S_a_i_n_t_Something is messed up then...its in the manual for your device.
06:49:25S_a_i_n_t_I'm just gonna check if its even defined anymore in current SVN
06:50:58S_a_i_n_t_Its definitly defined for the iPod video...
06:51:34 Quit mandred (Quit: leaving)
06:53:42S_a_i_n_t_ipodvideo.h/line 205 "#define HAVE_USB_HID_MOUSE"
06:53:48Falco98is there some residual difference between letting it idle poweroff versus long-holding play?
06:53:57S_a_i_n_t_So, it should be there.
06:54:08S_a_i_n_t_And, no...I don;t think so.
06:54:25S_a_i_n_t_It should be the same between idle poweroff/manual shutdown.
06:54:34Falco98i'll tell you in a minute what happens when i plug this one in
06:56:51Falco98grumble, it still hangs on the "multimedia mode" USB plug screen
06:57:17S_a_i_n_t_And There's no setting for HID in Settings/General Settings?
06:57:25S_a_i_n_t_it should be down the bpottom of the list.
06:57:33S_a_i_n_t_bottom even.
06:57:39Falco98well, i'll check again...
06:58:57Falco98the last entry on the bottom is "Voice"
06:59:19Falco98tell me if i'm on the wrong settings screen or something :-P
07:00:22S_a_i_n_t_you *should* have the setting "Hotkey" as the last in the list in that menu...
07:00:26Falco98yeah, that's about what mine looks like..
07:00:27S_a_i_n_t_This is weird.
07:00:43Falco98mine appears like that picture.. different theme of course
07:00:49Falco98but same choices as far as i can see
07:01:03 Join froggyman [0] (~me@unaffiliated/froggyman)
07:01:05 Join FlynDice [0] (~FlynDice@
07:01:10Falco98actually "playback settings" isn't in the list either
07:01:27S_a_i_n_t_Right, Can you try deleting the .rockbox folder from the device completely?
07:01:40S_a_i_n_t_and then start from fresh.
07:01:42Falco98i guess
07:01:43***Saving seen data "./dancer.seen"
07:01:51Falco98i'm not crazy about losing my configuration settings
07:01:57Falco98i guess i'll reserve the .cfg file
07:02:10S_a_i_n_t_yeah, just back up the .cfg
07:03:30Falco98and what file can i keep if i don't wanna lose my running playlist? just playlist control?
07:04:07S_a_i_n_t_Not sure about that one sorry.
07:05:20 Quit phanboy4 (Read error: No route to host)
07:05:45 Join phanboy4 [0] (
07:07:07 Join Barahir [0] (
07:10:05 Join wind [0] (
07:12:49Falco98k, just did a completely virgin reinstall of 3.5.1 (with the new poweron fix) and still no USB HID in settings/general settings
07:14:10S_a_i_n_t_Hmmmm, I'm fairly convinced it should be there. I *could* be wrong, but if I'm wrong, then the manual is too.
07:14:40Falco98fwiw it's not in my simulator either
07:14:41 Quit Boldfilter (Quit: Boldfilter)
07:14:41 Quit linuxstb (Ping timeout: 258 seconds)
07:14:53Falco98and the version i have built is at least newer than 3.4, though maybe not as new as 3.5
07:16:12S_a_i_n_t_Torne would probably be the man to speak to here, I'm pretty much out of ideas :\
07:16:23Falco98i can check the svn if you like
07:17:22S_a_i_n_t_Current builds handle the USB screen differently IIRC
07:17:41S_a_i_n_t_they will *always* go to the "rockbox USB screen" on cable insert
07:18:06S_a_i_n_t_whereas, release builds AFAIK will go to diskmode of cable insert.
07:20:59Falco98that's what i seem to have seen
07:21:24Falco98though iirc the svn build i tried, when plugging usb from poweroff, still didn't get data connectivity
07:21:55 Join CGL [0] (~CGL@
07:24:02Falco983.5.1 just went to disk mode correctly after plugin
07:24:07Falco98that's cool
07:24:22Falco98i seem to have completely fracked my settings tho :-P
07:24:38S_a_i_n_t_didn't you save your .cfg?
07:25:13 Part jennifur
07:25:46Falco98i don't think i properly transferred the auxiliary folders like fonts, icons, etc
07:26:00Falco98and the playlist control file i guess is touchy enough it refuses to be transferred
07:26:07Falco98but i can live with that one
07:26:40saratoga_labrun scandisk
07:27:58 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
07:28:17 Quit FlynDice (Ping timeout: 252 seconds)
07:28:24 Join kramer3d [0] (~kramer@unaffiliated/kramer3d)
07:28:32Falco98saratoga_lab: can i use the windows gui one?
07:28:59funmanyep, scandisk is the gui
07:29:29Falco98well, i consider scandisk to be an old dos command-line deal
07:29:45Falco98the win gui is called "error checking" or something
07:29:53Falco98it came up clean, as far as i can tell
07:30:19 Quit Horscht (Ping timeout: 276 seconds)
07:32:13 Join BHSPitMonkey [0] (~stephen@unaffiliated/bhspitmonkey)
07:32:46 Join Horscht [0] (~Horscht2@xbmc/user/horscht)
07:37:55Falco98still no USB HID... i think that may only be in the SVN
07:38:44Falco98cold-plugging again, just to test....
07:38:48Falco98*holding breath*
07:38:59Falco98it worked
07:39:14Falco98so my ipod *isn't* broken :-P
07:42:55 Quit saratoga_lab (Quit: Page closed)
07:42:56CIA-5New commit by funman (r25446): as3525v2: acknowledge USB connection/deconnection in SD thread ...
07:43:03 Join saratoga_lab [0] (~9803c20d@gateway/web/freenode/x-ukkmtnoxaykabrtf)
07:43:20Falco98what's that commit, funman?
07:43:41funmanFalco98: ?
07:43:43Falco98[or, any relation to what we've been talking about here?]
07:43:49Falco98ah k
07:44:06funmanas3525v2 is the chip of a series of Sandisk players
07:45:46CIA-5New commit by funman (r25447): Clipv2: reboot to OF when USB is inserted
07:47:19funmanstill can't get fuzev2 FM to work :(
07:47:53Falco98i just updated my ipod to the svn build, i now have an entry for "hotkey" at the bottom of settings/general
07:47:56funmanOF FM code is a bit different between AMSv1 and v2 so I have to do some work ^^
07:48:39Falco98ohh.. USB HID is under the "system" submenu
07:48:41Falco98d'oh seems things have been moved around a bit.
07:49:16Falco98there's a big submenu now
07:49:27saratoga_labtheres also a manual with these things in it . . .
07:49:32Falco98the whole "settings / system" folder has been moved it seems
07:49:35S_a_i_n_t_*Now* try disabling HID, and see if you get connectivity with the rockbox usb screen ;)
07:49:47Falco98actually i'm gonna try it as-is first
07:49:59Falco98when i tried my virgin 3.5.1 install it started working again
07:50:05S_a_i_n_t_saratoga_lab: the manual is what led us astray here.
07:50:09Falco98(that was 3.5.1 with torne's fix)
07:50:15saratoga_labhow so?
07:50:19Falco98yeah, the manual is wrong :-P
07:50:21S_a_i_n_t_Its incorrect.
07:50:23Falco98or, not up-to-date
07:50:59saratoga_lababout what?
07:51:04S_a_i_n_t_It seems as though Settings/General settings has had a shuffle around, and the manual hasn't.
07:51:25Falco98my HID devices just installed
07:52:07Falco98yeah, like settings / system is now under settings / general
07:53:42saratoga_labisn't that what the manual says?
07:54:13Falco98now that you say it that way...
07:57:10Falco98i kinda wish they'd make the USB mode screen acknowledge when the ipod's been unmounted from the system
07:57:29saratoga_labi'm not sure thats even possible
07:57:32Falco98at least so we know the disk is unmounted... it still seems to work as a mouse
07:57:45Falco98maybe not, but in the OF disk mode it knows when it's been unmounted
07:57:52Falco98the message changes to "ok to disconnect"
07:58:01funmanit knows when it's been ejected, not unmounted
07:58:12Falco98s/unmounted/ejected/ then
07:58:29Falco98i thought ejecting a disk was another word for unmounting it
07:59:05funmanon linux at least, if i unmount the device is still present and can be remounted, if i eject it disappears and i have to disconnect/reconnect to see it again
07:59:16Falco98ah k
08:05:40Falco98so i guess it should be noted that anyone experiencing any weird poweron / usb plug-in issues should try a virgin install... my issues seem to have been cleared up by doing so.
08:06:40S_a_i_n_t_Awesome, I kinda had a feeling there was some orphaned files in there somewhere
08:06:51S_a_i_n_t_or files left over from a previous revision.
08:07:14Falco98also i figure 3.5.1 may not have the hid setting because it will not ever work as a hid
08:07:23S_a_i_n_t_How exactly were/are you transferring ".rockbox" to your DAP?
08:07:48Falco98windows (file) explorer
08:08:10Falco98generally dragging the new .rockbox folder into the root dir of the player and selecting "overwrite all"
08:08:22Falco98which is pretty much word-for-word what the manual says to do to upgrade
08:08:27S_a_i_n_t_Yes, but I mean are you using "Extract To..." or copying the files somewhere, then moving them to the device
08:08:46S_a_i_n_t_The manual states "Extract To"
08:08:48Falco98my default zip program is winrar, which extracts to a temp directory itself before moving the files to the target
08:08:55S_a_i_n_t_there is less room for error this way.
08:09:20*S_a_i_n_t_ suggests using 7zip
08:09:26S_a_i_n_t_(its WAY better) ;)
08:09:30Falco98maybe, but with winrar i expect it's a lot more error-proof... i can try 7zip too
08:10:02 Quit BHSPitMonkey (Quit: Ex-Chat)
08:10:45Falco98also there should be a warning for people newly upgrading to a HID-enabled version that the first time they plug into USB they will be greeted with New Hardware Wizard screens
08:11:04Falco98i notice that you can just pick "no" and then "automatically" and the driver installations are straightforward
08:11:15Falco98but someone not expecting this may think it's an error
08:11:47S_a_i_n_t_It *should* handle the installation itself
08:12:01Falco98it did not go automatically
08:12:15Falco98if you're referring to the windows-side driver installations
08:12:32S_a_i_n_t_I've never had to click yes/no to install anything, just "found new hardware, Your hardware is installed and ready to use"
08:12:33Falco98i had to manually* install a mouse and keyboard driver
08:12:50Falco98(manually, as in, check "automatically" and then "next")
08:12:56S_a_i_n_t_Weird, I'm on XP pro, and it does it automatically for me.
08:13:28Falco98XP pro here as well
08:13:41S_a_i_n_t_Found new hardware (HID compliant *whatever*), your hardware is installed and ready to use"...that simple.
08:13:55S_a_i_n_t_Weird yours goes differently.
08:14:11Falco98i'll try on my laptops later and get back to you on how they act
08:14:17Falco98one is XP home and the other is XP pro
08:14:34Falco98and on my living room pc i have a hacked version of XP pro, it may act yet differently :-P
08:16:08S_a_i_n_t_Perhaps I already had the driver installed, shouldn't *need* to install a driver IIUC, it should already be there, windows just needs to find it.
08:16:44Falco98it's a little inaccurate of me to say "install" a driver
08:16:47Falco98that's just what it feels like
08:16:53Falco98i have to direct it, let's say
08:17:01saratoga_labmost likely what happened was that you had a corrupted files system, and then fixed it with check disk
08:17:07Falco98luckily it was painless after hitting "go"
08:17:10S_a_i_n_t_Ahhh, but still, I've never even had to do that much.
08:17:10saratoga_labthat allowed you to update rockbox properly
08:17:35saratoga_labdoes anyone have a D2?
08:17:52saratoga_labi'm trying to make sense of the ClipV2 test_codec results and want to see what the D2 gives
08:18:28 Quit mc2739 (Ping timeout: 240 seconds)
08:18:50 Quit S_a_i_n_t_ (Quit: PC needs a restart...Dagnabbit!)
08:19:35funmanmrobe500 & zenvision (and lyreproto) have the same CPU
08:19:58saratoga_labthe D2 as well
08:20:04saratoga_labah i see what you mean
08:20:06saratoga_labyes any will do
08:20:48saratoga_labon my clipv2, MP3, WMA and Vorbis all come in within 1MHz of each other at 192k
08:21:18Falco98for some reason my dynamic playlist keeps getting hosed after i manually shutdown, even though afaik it shouldn't be
08:21:55funmansaratoga_lab: test_codec might be a good tool to measure DRAM/IRAM speed
08:22:13saratoga_labadd a test ram option?
08:22:35funmanno i mean modifying .lds to use IRAM or not
08:22:59funmannot sure if it would be noticeable, will try later today
08:25:11saratoga_labah good idea
08:25:30saratoga_labi wonder if theres any parallelism available if you alternate accesses between the two memory blocks
08:25:40saratoga_lablike put code in one, data in the other
08:26:25 Join S_a_i_n_t [0] (S_a_i_n_t@
08:27:02funmanhm all the ape test files are too big for 8MB of ram
08:27:49saratoga_lablooking at the fuze and nano2g benchmarks, I wonder if the difference between PP and AMSv2 is that the ability to do fast loads without using LDM helps a lot in the mp3 filterbank
08:28:03saratoga_labactually i'm sure it helps a lot but i'm surprised it would make that much of a difference
08:28:47saratoga_labthat doesn't explain why the fuzev1 is so amazingly slow compared to the amsv2 though
08:29:03saratoga_labthe better multiplier shouldn't make such a huge difference on its own
08:30:16funmanbut PP has the COP so that makes a lot of difference?
08:30:19saratoga_labbah why do we even include the ape c5000 test track all it does is make running all the test tracks take 2x as long
08:30:27saratoga_labyes i'm comparing to PP without COP
08:30:48funmanhmm COP is only used for mp3, right?
08:30:59saratoga_labi added the COP optimization after i decided it was going to be way more work then i wanted to write a faster mp3 filterbank :)
08:31:06saratoga_labMP3 and SPC only i think
08:33:38 Join Boldfilter [0] (
08:38:35 Quit Falco98 (Quit: ChatZilla 0.9.86 [Firefox 3.6.2/20100316074819])
08:45:32saratoga_labhuh on the gigabeat f, MP3 really performs little different then on PP even though its arm9
08:46:06saratoga_labwhich seems to suggest that its not the improved load/store unit making the difference
08:46:41saratoga_labunless the F is somehow deficient in some other way
08:48:38funmanif i reduce CODEC_SIZE from 1MB to 768kB all codecs still build (not sure if the ones using malloc run though)
08:50:20amiconniirc the problem is the vorbis codec and very long files
08:51:08saratoga_labamiconn: AAC and very long files
08:51:17amiconnVorbis uses malloc, and the amount of memory it needs depends on track length
08:51:19 Join stoffel [0] (
08:51:29 Part Boldfilter
08:51:30saratoga_labvorbis has another problem with floor 0 files, but its unrelated to length
08:51:41saratoga_laband probably solved now, at least for buffer sizes much above 500KB
08:51:51funmanhm with my diff files still play but test_codec only shows "0 of 0" and doesn't answer
08:52:03saratoga_labAAC is the only reason we can't have 512KB codec memory on all targets now
08:52:36saratoga_labfunman: on the amsv2?
08:52:52funmanyeah, i might have missed something when putting the whole codec in iram though
08:52:54 Quit froggyman (Quit: umm num num num....sleep)
08:52:57saratoga_labon v1 it always runs with 320KB of codec memory so I don't know why 768k would be a problem
08:53:20funmanonly for Clip/c200/m200 which have 2MB of memory
08:53:53saratoga_labso the IRAM defines actually do something on the Fuzev1?
08:54:17funmaniram attributes work and some code/data is put in IRAM
08:54:41funmanon Clipv1 we put the whole codec in iram already so we disable enable iram attributes only for the core
08:54:57funmaniram is split in two: 1 part for core functions, the rest for the full codec
08:55:07saratoga_labhow much for the codec?
08:55:45funman0x48000 = 288kB
08:56:14saratoga_labthats pretty impressive that basically everything seems to work with such a small codec memory
08:56:24saratoga_labwe really need to fix aac so we can reclaim all that lost buffer space
08:59:35 Quit linuxstb (Ping timeout: 252 seconds)
09:00:01funmanhm my fuzev1 died when doing the folder speed test
09:00:43funmanand the problem with fuzev2 is metadata i think: get_metadata() returns 0 length
09:01:06 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
09:01:45***Saving seen data "./dancer.seen"
09:10:38funmanhm can't tell what's wrong: boosting only after reading meta has no effect, and it's read correctly when I look at the WPS
09:11:06 Quit tomers (Ping timeout: 268 seconds)
09:21:21 Join bmbl [0] (~Miranda@unaffiliated/bmbl)
09:25:16 Join Luca_S [0] (
09:26:05Luca_Sfuzev2 battery benching: just passed 9hrs, still running
09:26:20funmanon Clip+ 60MHz -> 24MHz gives at least 1 more hour (16h30 and counting)
09:28:37funmansandisk advertises the Fuze battery life as 24 hours
09:28:55funman"approximate value for 4 minutes songs at 128kbps"
09:29:33Luca_Sit's consistent with my experience - about 3 workdays of 8hrs each
09:29:35funmanalright Clip+ just stopped at 16h30
09:29:42pixelmasoap: sixthofmay the one with the big SSD in the X5? I took part in that thread but didn't receive a PM
09:29:58saratoga_labupdated the codec page with gigabeat f results (arm920t)
09:30:00Luca_SI noticed that when OF said battery 100%, RB said battery ~95%. when we'll have write support a real battery bench will be necessary, isn'it?
09:31:10saratoga_labvorbis and wma require more Mhz to decode, but MP3 requires the same, consistent with the amsv1 results
09:32:27 Join flydutch [0] (
09:32:29saratoga_labthe gigabeat F and Fuzev1 perform almost identically
09:34:14funmanLuca_S: probably, i'm a bit lost with battery code
09:35:54funmanhm 24MHz is too slow on Fuzev2
09:38:24funmanwheel is much less responsive
09:38:48Luca_Skugel said earlier that he found a way to get wheel data using irq instead of polling
09:38:56Luca_Smaybe it would help in lowering the clock?
09:40:34saratoga_labmake it clip only and then let kugel figure out if its worth having on the fuze
09:40:47saratoga_labthe bigger screen may make 24Mhz a little slow on the fuze
09:41:23saratoga_labshouldn't be a problem on the clip though
09:42:04Luca_Sthis clock change affects playback pitch?
09:42:40funmane200v1 is at 24MHz/30MHz though and it has the same screen size
09:42:57saratoga_labthe e200v1 runs at 30Mhz while playing
09:43:05saratoga_lab24 is just the idle clock which is basically never used
09:43:21saratoga_labfwiw I think its fine at 24 mhz, only minor slowness in menus
09:43:38saratoga_labbut thats a matter of disagreement on PP
09:47:34 Quit liar (Read error: Connection reset by peer)
09:48:34 Join Infa [0] (
09:48:45 Quit Infa (Client Quit)
09:48:46amiconnIdle (default) clock is used in the radio screen. The PP sansas do have a radio
09:49:02CIA-5New commit by funman (r25448): Clipv2/Clip+: lower DEFAULT/NORMAL frequency from 60MHz to 24MHz ...
09:50:58funman100% of clipv1 users never experience playback crashes
09:51:17funmanstatistics on a sample of 3 users
09:51:44saratoga_labbefore i fixed runtime estimation i put roughly 200 hours on mine without a crash
09:52:18funmani think we should move it to stable for next release
09:54:54 Quit phanboy4 (Ping timeout: 258 seconds)
09:56:45 Quit CGL (Quit: Saliendo)
09:57:42LloreanI haven't had playback crashes since the fix for them on the ClipV1, but I also only used it very heavily for about two weeks after the chance.
10:13:48 Join liar [0] (
10:16:42 Quit temp1234 (Quit: Page closed)
10:17:08 Quit Horscht (Ping timeout: 258 seconds)
10:18:03 Join Horscht [0] (~Horscht2@xbmc/user/horscht)
10:19:00 Join timaeus [0] (~8c718113@gateway/web/freenode/x-wcmdlqqszqkhlipo)
10:22:30 Join Horschti [0] (~Horscht2@xbmc/user/horscht)
10:22:51 Join webguest24 [0] (
10:26:56 Quit Horscht (Ping timeout: 268 seconds)
10:31:14 Join ender` [0] (
10:32:30 Join stripwax [0] (
10:39:32 Quit wind (Quit: CGI:IRC (EOF))
10:40:06 Quit shai (Quit: Leaving)
10:40:19 Join Kitar|st [0] (
10:42:59 Quit Kitr88 (Ping timeout: 276 seconds)
10:43:09 Quit Horschti (Quit: Verlassend)
10:44:41 Quit Kitar|st (Ping timeout: 240 seconds)
10:45:46 Join Farthen [0] (
10:48:15 Quit stoffel (Remote host closed the connection)
10:48:58 Quit webguest24 (Quit: CGI:IRC (EOF))
10:49:19 Join Kitar|st [0] (
10:57:17 Quit kramer3d (Ping timeout: 276 seconds)
10:58:03 Join kramer3d [0] (~kramer@unaffiliated/kramer3d)
10:59:12 Quit bluebrother (Disconnected by services)
10:59:13 Join bluebroth3r [0] (~dom@rockbox/developer/bluebrother)
11:00:34 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
11:01:47***Saving seen data "./dancer.seen"
11:03:47JdGordonmy CF ipod mini2g got 18hr sitting in the menu (with the inbuilt bar being used) which is curious
11:03:55JdGordonit got 18hr sitting in the WPS last week
11:04:01JdGordonsame revision and album
11:04:25JdGordonim going to do one more bench sunday night in a screen without the skin engine running at all
11:06:03saratoga_labJdGordon: don't we disable updating the WPS if the screen is off?
11:06:13JdGordonnot on the mini2g
11:06:21JdGordonits screenis perfectly visible with the backlight off
11:06:25saratoga_labah ok
11:06:35 Join Horscht [0] (~Horscht2@xbmc/user/horscht)
11:06:56JdGordonI was more interested in the time difference because it used to add about an hour of batt life if you stayed in the menu instead of the wps
11:07:07JdGordonthis is before we disabled updating when the lcd is off
11:07:13JdGordonso should be even more now
11:12:48 Join Strife1989 [0] (
11:14:36 Quit Strife89 (Ping timeout: 260 seconds)
11:16:36 Quit kramer3d (Quit: Leaving)
11:17:05 Join catfish [0] (
11:19:45 Join TheSeven [0] (~theseven@rockbox/developer/TheSeven)
11:20:45TheSevenhey, i have a crazy idea :-)
11:20:48TheSevenwho says that rockbox can only do sound?
11:20:49TheSevenlet's make it do *light*!
11:20:54TheSevenever thought about a DAP driving some DMX lights?
11:22:00JdGordonare you drunk? :)
11:23:40saratoga_labtheres not all that many output pins available on most players, so an arm dev board might work better
11:24:02TheSeventhere's the UART on the ipod dock connector...
11:24:32saratoga_labif you're hooking up another CPU to the dock, why not just do everything on the other cpu?
11:24:34TheSevendriving DMX using that should be piece of cake :-)
11:24:42TheSevenno, we don
11:24:50*TheSeven swears at his '
11:24:57saratoga_labyour device talks over serial already?
11:24:58TheSevenno, we don't need an external CPU
11:25:12TheSevenDMX is basically symmetric RS485 at 250kbaud
11:25:30saratoga_labah then thats easy enough
11:27:04TheSevenmaking it drive moving lights would of course be insane - you need proper means of manual control to do that - but driving an audio-synced chase on some pars for a smaller party should work just fine :-)
11:29:04pixelmaAlexP: do I remember correctly that it was you who had to additionally install the xcolor package for the tex environment?
11:30:02AlexPpixelma: No, it is those on Debian/Ubuntu
11:30:38AlexP(I'm on Arch)
11:31:04AlexPBut it was me who added the use of xcolor and added the note to the wiki about maybe needing it
11:31:11AlexPOther distros might as well of course
11:31:23AlexPIt depends how modularly tex is packaged
11:31:36pixelmaah, so you're saying it is missing for the people on Debian/Ubuntu? Is it really called xcolor or part of some texlive-dev or so?
11:32:10AlexPThe package is "latex-xcolor"
11:32:40AlexPFor Debian/Ubuntu anyway
11:33:09pixelmaok, thanks.. found it just now on ManualHowto even though I looked at that page before :\
11:33:24 Join merbanan [0] (
11:34:25AlexPIt is a standard package, but whether you need to explicity install it for your distro depends on how they packaged it
11:35:25pixelmait wasn't needed on cygwin btw... but that doesn't have a tex4ht package and I couldn't figure out how to install it manually (quite some time ago)
11:35:26AlexPYeah, I think we checked cygwin at the time (when the dependency was added)
11:35:26 Quit saratoga_lab (Quit: Page closed)
11:37:09stripwaxWhile playing a playlist, I did an Insert Last from file browser to append something else to my playlist, but when I looked at my Buffering Status in debug, I saw the buffer had been reduced to just the current playing track, and cpu was stuck boosted. It stayed like that until the current playing track finished playing (and the buffer drained), and at that point is refilled the entire buffer normally. But is that expected behaviour?
11:37:17 Join Horschti [0] (~Horscht2@xbmc/user/horscht)
11:38:39JdGordoncpu was stuck boosted? or boosted because of the codec?
11:38:39stripwaxcodec is vorbis, so about 10% boost. after the Insert Last, the freq was stuck at 80Mhz
11:38:47 Quit Horscht (Ping timeout: 258 seconds)
11:39:31funmanjust got the Clipv2
11:39:48catfishhi, i have checked the buyers guide as well as the FAQ but I couldnt find any player that is available shops
11:40:29stripwaxI've occasionally seen this behaviour both with and without fs#11141
11:41:51stripwaxBut it doesn't seem to be guaranteed reproducible
11:41:52 Join banan_ [0] (
11:41:52catfishI'm looking for something like ipod nano - any recommentations?
11:41:52linuxstbA used ipod nano? (1st or 2nd generation)
11:41:52stripwaxcatfish - ipod nano? via ebay etc?
11:41:52 Quit banan_ (Client Quit)
11:42:44catfishi checked ibay also - but there is only one very used one available
11:45:04catfishare you working on the newer nano ones?
11:45:04linuxstbThe Sansa e200s or fuze (v1) are similar to the Nano. i.e. flash-based. But they also have FM radio and a microSD card slot.
11:45:04catfishwill have a look on ebay for the fuze
11:45:04 Quit Luca_S (Quit: CGI:IRC (Ping timeout))
11:45:44 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
11:45:48CIA-5New commit by funman (r25449): Clipv2: enable USB stack (and use correct product ID)
11:45:55linuxstbNo-one is working on the newer Nanos at the moment. I would expect Rockbox to support them at some point in time, but even if someone started working on them today, it's likely to be at least 6 months until they're working well.
11:46:31funmanlooks like CPU frequency change broke recording on all as3525v2
11:46:46catfishare you working on any other ones which arent EOL?
11:47:04linuxstbThe newer versions of the Sansa Clip
11:47:17linuxstbAnd I think the newer version of the Fuze.
11:47:33linuxstbBut I don't know how close they are to becoming usable.
11:49:31 Quit stripwax (Quit:
11:51:07 Quit liar (Quit: Verlassend)
11:53:24bertrikA still-playing dynamic playlist is saved on shutdown, but a finished dynamic playlist is not, is that correct?
11:53:39S_a_i_n_tI *think* so...
11:53:54 Join Lixun [0] (~8984fa0b@gateway/web/freenode/x-lmahdzizjcoanayr)
11:54:16 Quit Lixun (Client Quit)
11:55:30catfishare there any difference between the 4GB verison of the fuze, or are they all v1?
11:55:31catfishi guess 8GB is above v1
11:58:25bertrikS_a_i_n_t, I guess that complicates the recent mailing list discussion about rewinding the playlist after completion a bit
11:58:42 Join liar [0] (
11:59:26S_a_i_n_tI wouldn't know sorry, I've alway *meant* to sign up to the mailing list, but I'm pretty much always here so I figure I don't miss much.
12:00:15S_a_i_n_tAll I know is that I'm pretty sure a dynamic playlist gets trashed on shutdown if its finished playing.
12:00:28S_a_i_n_tBut I may however, be quite wrong about this ;)
12:02:09linuxstbcatfish: This page (and the pages linked from it) may help you with the newer Sansa devices -
12:02:58 Join JdGordon1 [0] (
12:03:14 Quit JdGordon1 (Client Quit)
12:03:37funmancatfish: you can know the fuze version only by powering it and looking for the firmware version in the menu (it starts by v1 or v2)
12:04:47catfishfunman, i don't own one - I'm just figuring out which on I'll order
12:04:48 Join JdGordon1 [0] (
12:05:21 Quit JdGordon (Quit: Leaving.)
12:05:29funmanthat's what I mean: you can't order a Fuze "v1" or "v2" unless the seller looked in the menu, and told you the version (without lying)
12:06:28bertrikS_a_i_n_t, the discussion was basically about a patch that allows the playlist to be rewinded after completion. If a completed playlist is thrashed, it makes things like that a little harder.
12:07:16 Join kugel [0] (~kugel@rockbox/developer/kugel)
12:07:41kugelfunman: 24MHz can't be the problem, the UI doesn't need more than a few MHz - if at all
12:07:58S_a_i_n_tbertrik: wouldn't "repeat all" do the same thing...?
12:08:00Unhelpfulfunman: certain sellers have carried refurb e200 players that they knew to be v1. don't know if the same has happened to fuze yet, maybe not as it's not supported and those sellers were using the v1 label as a feature (rockbox ready)?
12:08:12kugelthe slowness must be caused by something else; maybe memory clock or dbop clock?
12:08:38UnhelpfulS_a_i_n_t: i think we mean rewind/skip back, not "start over"
12:08:39funmanafaiu only the CPU clock should be changed there
12:08:40bertrikS_a_i_n_t, no, that rewinds the playlist AND continues playing
12:09:16kugelI doubt it
12:09:24funmankugel: i thought perhaps using interrupt based scrollwheel reading would give a better clue
12:09:24S_a_i_n_tahhhh, i getcha now.
12:09:43S_a_i_n_tHmmmm. Pass *shrugs*
12:09:50kugelthe UI is sluggish at 60MHz too, that freq works just fine on the as3525v1
12:10:17funmanUnhelpful: ah there is a website which sells rockboxed e200v2/fuzev1 , targetting blind users.
12:10:35kugelfunman: I tried it tonight, it doesn't change the sluggishness
12:10:50Unhelpfulfunman: i was more thinking of sellers like froobi... and i thought fuzev1 was not officially supported yet?
12:10:57kugelI think the whole UI is slow not just the scrollwheel
12:11:13funmandid you try test_fps ?
12:11:23funmanUnhelpful: it is since 3.5
12:11:47Unhelpfulfunman: shows what attention i pay lately :/
12:13:32kugelfunman: 9fps :\
12:13:48kugel(at 24MHz)
12:14:29*bertrik has a look at the fuze v2 display driver
12:14:48funmanclipv2 uses dbop too, i'll try to see what it gets
12:14:54kugelfunman: almost 100fps at 240MHz (which is also not overly fast)
12:15:14funmanare fps linear with CPU freq?
12:15:20linuxstbkugel: That's the fuzev2?
12:15:31kugelthe fuzev1 gets 100fps at 62MHz already
12:15:49kugellinuxstb: yes
12:17:30kugelfunman: what is the pclk at ?
12:18:22bertrikI guess pclk cannot ever be higher than the processor clock
12:19:41bertrikThe theoretical maximum display fps of the fuze would be about 38 fps at 24 MHz DBOP clock
12:21:37kugelthat would be plenty
12:23:14 Quit liar (Ping timeout: 258 seconds)
12:25:10 Join Luca_s [0] (
12:25:24 Nick Luca_s is now known as Luca_S (
12:26:23kugelfunman: does the datasheet of the newer ams chip describe the new cgu_dbop bits?
12:34:02 Quit FOAD (Quit: I'll be back)
12:37:10 Join pamaury [0] (~c2c7a50a@rockbox/developer/pamaury)
12:40:00kugelbertrik: we reach the maximum on fuzev1 relatively easily; not at all on fuzev2 though
12:41:42JdGordon1kugel: what backdrop issues are there with the skin engine?
12:41:50JdGordon1and whats the story with dynamic load sizeing?
12:43:05kugelchanging from the main menu to the wps (or back) is always a bit glitchy: flickering and the backdrop doesn't change instantly; also in the wps there are left overs from scrolling lines that appear before the album art show
12:43:52JdGordon1is it on flyspray?
12:44:50S_a_i_n_tIf i use "clear backdrop", when i try to set a new backdrop using the context menu, I gt "backdrop failed" 9 tims out of ten.
12:48:47 Quit Zarggg (Ping timeout: 276 seconds)
12:57:25JdGordon1interesting.... if you play a new dirplay playlist while paused it doesnt unpause?
12:57:49 Join FOAD [0] (
13:01:13funmankugel: I only have the as3531 datasheet, the SoC is completely different, the only useful source of information is the overall description and address of *some* as3543 registers, and this doesn't include the descriptions of individual bits of these registers
13:01:50kugelthe linux patch doesn't have cgu_* anymore it seems
13:01:51***Saving seen data "./dancer.seen"
13:01:57funmanon as353x ?
13:02:08kugelnot cgu_dbop at least
13:02:41kugelmaybe the new bits are some clock source selector
13:02:53funmanno indeed, the only similarities between as353x and the Clipv2/+ and Fuzev2 are: the as3543 i2c audio/pmu chip, the SD controller and perhaps the USB controller
13:03:20funmani'll send you the datasheet
13:03:37kugelwell if it's not helpful I don't need it :)
13:03:58kugelbut sure, I can take a look
13:05:10 Join efyx [0] (
13:05:44pamauryfunman: Well I'll start reading the usb code of as353x, can I create a page on wiki ? Is there any problem in publishing those information (not the code) ?
13:06:14funmanwell: yes you can, no there is no problems
13:08:46kugelthe backlight_on failure in the wps seems to be related to fading. without fading it works
13:09:29funmaneven with r25445 ?
13:20:29 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
13:20:38kugelfunman: I assume either the dbop or the ram is too slow at 24 (and even at 60) MHz, or both
13:23:46funmantesting ram should be easy, just loop for N ticks reading or writing some array
13:24:39kugelheh, the debug screen shows audio can be managed at 24MHz just fine, but if I increase the update rate by holding select it cannot manage it anymore :)
13:25:08 Join domonoky [0] (~Domonoky@rockbox/developer/domonoky)
13:25:35 Join nik1 [0] (
13:25:54funmanhm the codec should boost when appropriate so it would show display update slowness
13:26:07funmando you have an e200v1? (which runs at 30MHz)
13:26:20kugelno problem there
13:27:10kugel24MHz is also fine on it
13:27:33funmaneven with holding select?
13:29:20kugelwell, it boosts sometimes instead of never but the pcm never goes empty
13:29:50 Quit merbanan (Ping timeout: 260 seconds)
13:30:40 Quit steve|m (Ping timeout: 276 seconds)
13:31:46kugelhm, the fuzev2 doesn't seem to boost when the pcm empties
13:32:10 Quit funman (Quit: free(random());)
13:32:38 Join Schmogel [0] (
13:34:10kugelthat's strange
13:36:00kugelthe lcd performance scales lineary with the cpu clock
13:36:15kugelthat shouldn't happen :\ the dbop should limit
13:36:59gevaertsAlexP: try FS #11167
13:37:34gevaertsI don't think it's the font loading that's hurting you on the beast
13:38:08AlexPOK, will do (after some lunch :))
13:39:08JdGordon1font loading could be better though...
13:39:22JdGordon1AlexP: pixelma: so what are we going to do about fm skinning?
13:40:03AlexPJdGordon1: It is fine for me on both beast and H100 (with a bigger buffer) - wasn't the argument that the bufer resizing stuff should go in first?
13:40:20JdGordon1yes, but there has been no news on that either
13:41:42AlexPWell it is up to you - the fms stuff seems great, so I guess the options are either up the grey/mono buffer and stick it in, or try and fix the buffer resizing stuff
13:42:32JdGordon1how much bigger is needed?
13:43:07 Part nik1
13:44:03AlexPquite a bit, for me it is the use of multifont that pushes it up
13:44:12AlexPI think I needed to change it to 8x or so
13:44:28AlexPgevaerts: I guess dircaqche makes even more sense given I have a 120 GB in the S
13:44:39AlexPadd "disk" somewhere
13:46:09gevaertsAlexP: You mean to blame, or to enable?
13:46:18AlexPto blame
13:46:44gevaertsyes, unless of course you despise flac and use wav instead
13:46:56AlexPeven worse, mp3/ogg
13:47:01AlexPand it is nearly full
13:47:32AlexPanyway, I'll try the patch shortly
13:47:39gevaertsOf course on the beast you don't really notice a minute more or less given the speed of the OF boot part
13:48:14AlexPI have single boot
13:48:18AlexPSO I do notice :)
13:51:40gevaertsAnyone around with a reasonably large and full flash based player?
13:56:47 Join liar [0] (
13:57:20pamauryI have a clip+ with less than an hundred MiB. I also have my e200 which is not full but I can put all my music on it to make it full :)
13:58:01JdGordon1gevaerts: CF count?
13:59:02JdGordon1for the io priorities, wouldnt it make more sense to tie it in with the thread priotiry instead of the somwhat hacky way of the patch?
13:59:34gevaertsCF would count I guess. I just want to see if there's some effect
13:59:41JdGordon1with some way to raise/loawer the io prio?
13:59:59pamauryWhat is CF ?
14:00:03gevaertsJdGordon1: how would you actually schedule IO?
14:00:06JdGordon1compact flash
14:00:21pamaurygevaerts: what do you want to test ?
14:00:33JdGordon1umm, same way as cpu?
14:00:50 Quit blairb (Quit: Leaving)
14:01:21gevaertspamaury: I want to see if FS #11167 makes a difference on flash, i.e. enable dircache, see how long the boot takes, see how fast (subjectively) playback starts, apply the patch, see if it changes
14:02:41JdGordon1adding a splash would be much better than subjective timing
14:02:48gevaertsJdGordon1: that means replacing the locks in block drivers by queues. Do we really want that?
14:02:49 Quit antil33t (Read error: Connection reset by peer)
14:02:57 Join antil33t [0] (
14:03:12gevaertsYes, or a bootchart line where playback actually starts, and setting the start screen to WPS
14:03:56JdGordon1~5s from press to music with dircache on (svn)
14:04:12 Quit TheSeven (Quit: ChatZilla 0.9.86 [Firefox 3.6.2/20100316074819])
14:04:33JdGordon1absolutly no change with dricache off
14:05:19pamaurybootchart is in svn now ?
14:06:00JdGordon1i tihnk so, those numbers are subjective, not bootchart
14:06:22JdGordon1AlexP: multifont isnt useable on ggrey/mono now right?
14:08:13JdGordon1I wonder if we should add 20K to every target to fit two skin fonts in, that shold then make fms fit also
14:08:32JdGordon1and also encourage dynamic sizing to get finished
14:08:50gevaertsJdGordon1: maybe a stupid question, but is the 10K per font fixed? What if you load a smaller font?
14:08:59JdGordon1yes, 10k per
14:09:32JdGordon1we went the technically stupider but workable option
14:09:47pamauryOn e200, without dircache it boots in around 2sec, with fresh dircache build it boots in around 3sec (subjective). I'll now test with your patch but I'm not sure I'll see the difference without bootchart
14:11:41JdGordon1that sort of time difference is statistically negligable anyway
14:12:45pamauryNot really, there is clear difference between them for me. 1sec difference on a 2sec time is not negligeable !
14:12:59gevaertspamaury: was that the foreground scan?
14:13:21pamaurygevaerts: with your patch, it seems to boot in 2sec with and without dircache
14:13:41pamauryAll are foreground scan
14:14:16pamauryBut that's all subjective as I said
14:14:28gevaertshm, I'd assume that the foreground scan shouldn't change
14:15:04gevaertsJdGordon1: where should I add a log line to see when playback actually starts?
14:16:07JdGordon1umm... I don't tihnk thats so simple
14:16:13JdGordon1logf isnt threadsafe right?
14:16:17pamauryWell, I guess I need some logging because it's too hard to notice the difference. I can also try to have thousands of little instead of big ones
14:16:35gevaertssure it is
14:16:55gevaertsWell, actually, I'll add a bootchart line, but that's the same
14:17:28pamauryhuh, I'm not so sure logf is thread-safe, it will mix up lines
14:17:33JdGordon1ah, audio_play() blocks (supposedly), so root_menu.c before gui_wps_show() would do
14:18:29JdGordon1audio_play() calls queue_send() on the playback thread, so if that blocks then thats the best place
14:18:49gevaertsIsn't that the point where playback starts buffering?
14:19:40gevaertsi.e. on my F60 10 seconds before I actually hear sound
14:20:20JdGordon1I'm too tired to work out if audio_play_start() actually starts audio or the buffering mechnism
14:20:33JdGordon1but thats what the playback thread calls after audio_play()
14:20:41JdGordon1comments suggest it does start playback
14:20:54JdGordon1Although... I see the wps before hearing music, so maybe not
14:21:25gevaertsyes, that's the problem
14:23:38JdGordon1another thing.. we should maybe split starting audio from starting the PS on boot
14:23:55JdGordon1rewrite the whole damn apps/ folder imo!
14:24:49gevaertsGet busy! ;)
14:24:52amiconnlogf() is threadsafe, since we have cooperative threading
14:25:06amiconnIt's not interrupt safe, but that doesn't matter here
14:25:36 Part domonoky
14:25:45*JdGordon1 goned
14:27:09 Quit Strife1989 (Quit: See ya.)
14:28:46gevaertsamiconn: do you have opinions on FS #11167? I'd want to commit fairly soon, but I'm not sure if and how the mechanism should be generalised to tagcache and other background scanners, so I'd like more opinions
14:34:39pamaurygevaerts: implement I/O priority on a per thread basis ? For example, each thread as a "I/O delta" parameter and the storage refuses a read/write if the delta to last I/O is greater than the parameter
14:34:52pamaury* refuse=delay
14:38:44gevaertsIt should obviously be thread based I think, but maybe two classes are enough? Background and immediate
14:40:03kugelanother parameter for create_thread?
14:40:20gevaertsor an extra function to set it
14:40:32kugelor maybe a single struct to pass?
14:40:57*gevaerts doesn't care much. He wants to get the actual mechanism right, the way to set the numbers is not very important
14:41:19bertrikgevaerts, I haven't really thought this through myself, but having simply two categories background and immediate sounds good
14:42:03gevaertsIf a background IO thread sees that an immediate thread has done IO recently, it waits until this is no longer the case, with possibly a maximum waiting time after which it can run anyway
14:43:36*kugel tends to agree with pamaury
14:43:45gevaertsThe exact same mechanism might work with more than two levels
14:44:26*gevaerts doesn't really fully understand pamaury's proposal
14:45:27bertrikI think the mechanism should not be too intrusive, like refusing I/O
14:45:45gevaertsoh, definitely not
14:47:14pamauryMy proposition is the following: each thread has internal two storage parameters: io_delta and last_io_tick. On each read/write, the storage_code compute current_tick-last_io_tick and if the result if < io_delta, then the read/write is delayed (for example it yields/sleep)
14:47:34pamauryBy adjusting io_delta, you can "immediate" and "background" threads like you said
14:48:23 Join TheSeven|Mobile [0] (~theseven@rockbox/developer/TheSeven)
14:48:27gevaertspamaury: doesn't that also slow down IO if there's no other IO going on?
14:48:35kugelif a immediate thread calls a io func it could set a timeout (current_tick+X), if a backgorund thread calls a io func it sleeps while TIME_BEFORE(current_tick, timeout)
14:48:43kugelsounds simple enough to me
14:48:44pamauryyes, you'll have to implement something like that also
14:50:45pamauryI just want to make you aware of the fact that a more general machanism makes sense if you want to generalize your patch to tagcache and others.
14:51:27gevaertsIf you do this timeout based, I don't see why the delta is stil useful
14:51:46*gevaerts decides to put his ideas in code
14:52:12bertrikhow about some kind of priority-based mutex for disk-access, where the waiting thread with the highest IO priority is granted access to the disk
14:52:54pamauryyou also have to let the background thread a little bit of disk access !
14:53:20Luca_Suh oh.. audacity is starting to behave in a strange way during the battery bench
14:53:22gevaertsok, so the delta would be a maximum wait time, not a general wait time
14:53:48bertrikpamaury, I don't see why, it just has to wait until the higher priority tasks are done
14:53:52kugela mutex would remove the need for the storage drivers to implement a mutex, wouldn't it?
14:54:23gevaertsbertrik: will that prevent thrashing?
14:54:32gevaerts(the mutexes I mean)
14:55:17 Join krabador [0] (
14:57:16bertrikgevaerts, it might not, if a high priority IO task is done and releases the mutex, it's probably immediately taken by one of the waiting lower IO tasks
14:57:55pamaurygevaerts: yes you're right, the delta would be a sort of maximum wait time, it's better this way
14:58:14bertrikDidn't linux have some kind of very simple mechanism where it simply waited for a few ms before switching I/O tasks, assuming that after a read was done, the same task would most probably do another read?
14:58:47gevaertsI suspect one priority number will work. If you're priority 0, you won't wait. If you're priority 10, you're lower than 5, and you wait for up to 10 units
14:59:12gevaertsbertrik: I think the thing I'm working on is equivalent to that
14:59:22pamauryI think Linux has a full blown I/O scheduler
14:59:37pamauryWell several :)
14:59:47pamauryThe default one is CFQ or something like that iirc
15:00:02gevaertsThey also have lots of caching and queues
15:00:29pamauryLinux is complicated ;)
15:01:52***Saving seen data "./dancer.seen"
15:02:26bertrikJust to be clear, the problem is that on disk-based players there are two threads fighting for disk access thereby causing a lot of seeking, right?
15:03:16gevaertsthat's the main problem, yes.
15:03:42pamauryAnd one of these thread is a background thread
15:03:48gevaertsThere's also the fact that even without seeking buffering might be delayed by dircache or similar, which I'd like to avoid if possible
15:06:36kugelhm, without boosting, mem access seems to be more than 10 slower; apparently it scales with the cpu clock as well
15:13:31*bertrik looks forward to gevaerts implementation
15:15:48kugelindeed, around 10 times slower with 1/10 freq
15:16:15kugelshould I commit the test_mem plugin?
15:16:42bertrikmaybe you can implement the delay and the priority in just a simple number per disk, the "current IO priority level". Threads with IO priority higher than that get immediate access and raise the disk IO-priority level to their level, lower threads are blocked. The per-disk IO-priority level decreases by one for each tick, so that eventually the lower level blocked threads get access.
15:17:48bertrikkugel, sounds like a useful thing to have, is there a patch of it somewhere?
15:19:57AlexPgevaerts: Any specific requirements to test this? Just time with and without, or do I need to enable/disable stuff in between?
15:20:33gevaertsAlexP: just see if it helps you. I'm now basically redoing the thing anyway :)
15:20:50AlexPOK :)
15:22:04gevaertsbertrik: I considered doing that. I decided not to change the priority though, I'm not sure we should do that
15:23:21kugel700kB per tick on my fuzev2
15:23:25gevaertshm, this was meant to make booting faster, not hugely slower...
15:24:45 Quit pamaury (Quit: Page closed)
15:25:16kugel7MB per tick when boosted
15:25:18 Quit liar (Ping timeout: 258 seconds)
15:25:32kugelno wait, my math is wrong
15:25:42 Join mc2739 [0] (~mc2739@rockbox/developer/mc2739)
15:26:05bertrikkugel, ok, looks good to me. Maybe buf should be made volatile to make absolutely sure that indeed memory is accessed (optimisation may conclude that x is never used). Also put your own name in the header if you wrote it. And maybe you can have the plugin calculate kB/s or Mb/s
15:26:21kugel70k per tick unboosted, betwee 750k and 1M boosted
15:27:37 Quit mc2739 (Client Quit)
15:27:51 Join liar [0] (
15:28:05 Join mc2739 [0] (~mc2739@rockbox/developer/mc2739)
15:28:57 Join panni_ [0] (
15:29:08kugelheh, volatile buf and x makes it way slower ;)
15:29:21kugelbut now read is slower than x
15:29:47kugelI eventually wanted to make it discard x, because if x is on the stack each read is a write too
15:31:08AlexPgevaerts: Until main menu shows: Current build (no dircache) 6.5s with dircache 9.6 s with patch + dircache 6.5 s
15:31:30Luca_Si'm no expert, but i'm not sure about this testing method. if the cpu is unboosted, all the cpu operations will be affected, not just loads and stores; the increments and the comparisons too. are they negligible?
15:31:36gevaertsAlexP: from poweron?
15:31:41AlexPgevaerts: This is with all other settings reset
15:31:48AlexPgevaerts: From seeing the RB bootloader
15:31:52gevaertsOK, so with the (big) default font
15:32:23gevaertsAlexP: how about until you can actually hear music?
15:32:26AlexPyeah, 15-adobe-helvetica
15:32:28bertrikkugel, can't you use some kind of standard functions instead of your own loops, like memset and memcpy?
15:32:31AlexPgevaerts: I'll try
15:32:39 Quit TheSeven|Mobile (Ping timeout: 240 seconds)
15:32:49gevaertswith dircache. Without, it's not going to make a difference
15:34:05 Join pamaury [0] (~c2c7a50a@rockbox/developer/pamaury)
15:37:06kugelbertrik: should I? that only works for writes anyway, and currently (going by the disassembly) gcc is doing exactly what I want
15:37:26AlexPgevaerts: With start screen set to resume without patch 16.4 s with patch 8.2 s until I hear music
15:37:48AlexPQuite a good improvement :)
15:38:22kugelLuca_S: yes, basically
15:38:56kugelthere's 3 million memory access but only a handful other instructions
15:39:51kugelmaybe the loop counter increment isn't that negible but that's impossible to get around
15:40:58gevaertsAlexP: yes, I think it's not too bad :)
15:41:43AlexPgevaerts: So hurry up and get it in :P
15:41:47 Quit krabador (Ping timeout: 258 seconds)
15:45:31gevaertsnew patch ready
15:45:41gevaerts(and briefly tested on gigabeat f)
15:45:57AlexPWould you like me to try?
15:48:17gevaertssure. I'm especially interested in reviews, but testing is always welcome
15:50:22bertrikgevaerts, silly question maybe but what is that IF_MD2 in the patch?
15:51:24gevaertsbertrik: it makes the argument disappear on single-drive systems
15:51:39gevaertsThis and other related macros are defined in mv.h
15:52:16*kugel totally hates IF_MD*
15:52:51gevaertskugel: I see your point, but this patch isn't the place to remove them
15:53:05bertrikyeah, feels like a giant hack (not attacking this particular patch though)
15:53:13*gevaerts points to . Have a look at the time for the latest build
15:53:28kugelgevaerts: that's correct, unfortunately :)
15:53:40bertrikludicrously fast
15:56:00kugelso, it must be the dbop that is so slow
15:56:07kugele200v1 memory is *way* slower
15:56:26AlexPgevaerts: Also 8.2 here :)
16:01:12bertrikgevaerts, if there are two threads with equal io priority, the current one will keep access until the timeout, right?
16:01:23kugelroolku pimped his clients?
16:02:18gevaertsbertrik: with equal io, things shouldn't change compared to svn. Whoever calls the function first will get in first, there will be no extra sleeping
16:03:05gevaertshm, I think I can make it work for multiple drives pretty easily
16:04:52CIA-5New commit by kugel (r25450): Memory benchmark plugin. Tests memory speeds and prints the kB/s.
16:06:31CIA-5New commit by alle (r25451): Properly set the keywords for the recently added file
16:06:41 Join roolku [0] (
16:08:00 Quit flydutch (Quit: /* empty */)
16:08:03roolkuyeah, 1:45 min. :)
16:08:03kugelroolku: your new clients own :)
16:09:14gevaertsroolku: what sort of machines are those?
16:09:55roolkudual xeons with i7 core
16:10:35kugel2x 4 cores?
16:11:21roolkukugel: and hyperthreading, so they are reporting 16 cores all together :)
16:11:26kugelbertrik: it appears the dbop is slowing things down. the memory is slow, but not slow enough to make the UI suck that much
16:11:36roolkuour new cluster
16:11:39kugelroolku: insane :)
16:12:09bertrikkugel, but dbop is not used during the test, or is it?
16:12:11roolkunot sure how long I can use them for rockbox though...
16:12:59kugelbertrik: not at all
16:14:25CIA-5New commit by kugel (r25452): Move delta calculation up so that only the loop time counts.
16:15:38kugelbertrik: strangely though the mem is on v2 slower when unboosted, but about the same when boosted
16:17:01bertrikkugel, I don't know the clocking scheme of the v2 ams sansa, I read that there's no longer a selection of fastbus, async or sync modes between cpu and the rest of the SoC.
16:17:16JdGordon1gevaerts: something to think about... we dont want background threads waiting for the forground thread if they are accessing different disks (not volumes though!)
16:17:29gevaertsJdGordon1: yes, see latest patch :)
16:17:41JdGordon1:) just making sure
16:18:09gevaertsI couldn't do volumes anyway, that's a different layer
16:18:17JdGordon1oh right
16:20:26kugelbertrik: the "clock modes" section in the arm922t reference manual is replaced by a "tightly-coupled memory interface" section in the arm926ejs one
16:20:50CIA-5New commit by alle (r25453): Small cosmetic changes to the hotkey description
16:23:21kugeldo we need the speculative noncacheable instruction fetches feature of armv5?
16:23:34JdGordon1(this might be target dependant but...) Is writing a in ram BMP to the lcd "expensive"?
16:23:54JdGordon1I assume compared to decoding it would be negligable?
16:24:03kugelour armv4 which don't have that don't have that problem
16:24:26kugelJdGordon1: well, it's a memcpy, you can use my just committed plugin to see how expensive it is
16:24:42JdGordon1that would invalidate tomorows bench
16:25:12CIA-5New commit by alle (r25454): Restore wrongly deleted text
16:25:41JdGordon1I'm doing a series of benchmarks with the same svn version and album, but in different screens
16:26:07 Join Barroom [0] (
16:26:30 Join FlynDice [0] (~FlynDice@
16:26:40 Quit FlynDice (Read error: Connection reset by peer)
16:27:13S_a_i_n_twhat is/are "playlist viewer indices" please?
16:27:16 Join FlynDice [0] (~FlynDice@
16:27:52 Join anewuser [0] (anewuser@unaffiliated/anewuser)
16:28:26S_a_i_n_tits a config file option, but i don't know what it does.
16:28:54JdGordon1I think its to show/hide the index numbers in the playlist viewer
16:29:01JdGordon1the real viewier, not the one on the wps
16:29:08S_a_i_n_tthe manual doesn't say (before i get an RTFM)
16:29:28S_a_i_n_thmmmmm...seems plausible, thanks.
16:31:41 Quit FlynDice (Ping timeout: 245 seconds)
16:32:46 Join FlynDice [0] (~FlynDice@
16:40:51gevaertsJdGordon1: I seem to remember a pretty significant difference between the old rockbox_default and the menu, back in early 2008. That one has/had peakmeters
16:41:25JdGordon1this is why I',m doing the tests
16:41:54JdGordon1assuming there is any difference, it doesnt show any difference if the enginge is running at all
16:42:15JdGordon1tomorows bench will be in the hw debug screen which does nothing while it waits for a keypress (no skin updating at all)
16:42:33 Quit FlynDice (Ping timeout: 260 seconds)
16:45:10JdGordon1ctrl+alt+e apparently
16:45:14JdGordon1maybe thats windows only thoguh
16:45:31 Join Horscht [0] (~Horscht2@xbmc/user/horscht)
16:46:06 Quit Barroom (Quit: CGI:IRC (EOF))
16:46:28roolkuI think the build server has problems coping. it is not reporting n02 when it is clearly doing builds and uploading was n09 and n15 in the previous round
16:48:36 Quit Horschti (Ping timeout: 246 seconds)
16:53:39 Quit xiainx (Ping timeout: 240 seconds)
16:54:13 Join xiainx [0] (xiainx@wpa062009.Wireless.McGill.CA)
16:55:06 Join CaptainKewl [0] (
16:56:01 Join FlynDice [0] (~FlynDice@
16:57:15 Quit FlynDice (Read error: Connection reset by peer)
16:57:30 Join FlynDice [0] (~FlynDice@
16:58:15 Quit FlynDice (Read error: Connection reset by peer)
17:01:15S_a_i_n_tDammit, this sucks. Howcome you can't turn icons off with a theme .cfg?
17:01:37S_a_i_n_tI have "show icons: no" in my theme .cfg, but it still shows icons :\
17:01:53***Saving seen data "./dancer.seen"
17:02:09 Quit avar (Ping timeout: 240 seconds)
17:02:13 Join angelwolf71885 [0] (
17:04:14 Join FlynDice [0] (~FlynDice@
17:04:20 Quit JdGordon1 (Ping timeout: 246 seconds)
17:06:42 Join Blue_Dude [0] (~chatzilla@
17:06:42 Quit Luca_S (Quit: CGI:IRC (EOF))
17:06:58S_a_i_n_tMan, that *really* sucks...A theme should be able to turn the icons off if so desired.
17:07:11kugelyou're doing it wrong, that's it
17:07:20S_a_i_n_tI'm pretty damn sure you used to be able to do that with a theme.cfg
17:07:22kugeli.e. RTFM ;)
17:07:35Blue_DudeI want to throw an idea out there. I'd like to make the pitchscreen settings (pitch and speed) permanent between sessions like most other audio settings. Any objections or problems with this approach?
17:08:00kugelI like the idea to store it in the bookmarks way better
17:08:18S_a_i_n_tHow am I "doing it wrong"?
17:08:52kugelwell, compare what you wrote with what's written at the page I linked
17:08:56gevaertsI also think that bookmarks are the better place
17:08:58Blue_DudeWell, I'm working on a patch to store version info in bookmarks so any changes to the bookmark format will be backwards compatible. Still, I'd like to keep pitch and speed information.
17:09:01S_a_i_n_tshow icons: no seems fine according to the link you just posted.
17:09:19kugelbut "no" is not a valid value...
17:09:41Blue_DudeSounds like that's going to be shot down pretty decisively...
17:09:55 Quit FlynDice (Quit: Gotta go fly!)
17:10:13 Join FlynDice [0] (~FlynDice@
17:11:19gevaertsBlue_Dude: I'd find it hugely annoying if speed is kept at some nonstandard value if I pick up a player and start playing some music. Much more so than picking it up to play a new audiobook/podcast and finding that I still need to speed it up a bit
17:11:27angelwolf71885so the only way to disable icons from the status bar on the mail menu is by theming them into the arrangement you want them in?
17:11:32Blue_DudeThere's already a patch out there to store pitchscreen information in bookmarks but it breaks backwards compatibility. I'm working on a mod to fix that.
17:12:26Blue_Dudegevaerts: I hear you, but I play mostly stuff that need to be sped up and I have to do it every time. I guess the bookmark mod would help a lot though.
17:12:59S_a_i_n_tangelwolf71885: Wait...what are you trying to do?
17:13:30angelwolf71885actuialy its more of a question the stat bar where the battery and clock is
17:13:36kugelangelwolf71885: you can disable the statusbar?
17:13:47angelwolf71885not disable it
17:14:07angelwolf71885re range the icons on it and disable unneeded ones
17:14:13gevaertsangelwolf71885: you're asking if you can customise the status bar without customising it?
17:14:26S_a_i_n_tseems like it ;)
17:14:28kugelyou need a base skin (so yes, themeing) if you want to keep it but also want prettier icons
17:14:50angelwolf71885ah ok thanks thats the thing i was wondering
17:15:19angelwolf71885i was just curios mostly how to make the time center and the battery all the way to the right
17:15:35 Quit FlynDice (Ping timeout: 246 seconds)
17:15:44Blue_DudeOK, thanks for the reality check. I'll concentrate on bookmarks then.
17:16:13S_a_i_n_tOpening up "" will probably give you a clue
17:16:30angelwolf71885ah ok thanks :)
17:16:35S_a_i_n_tthat and the CustomWPS page in the wiki
17:17:13angelwolf71885i went through the WPS info on the wiki but it was all Greek to me lol
17:17:47S_a_i_n_twell, if thats the case I'm not sure how much luck you'll have with the .sbs then
17:17:54S_a_i_n_tBut, its not that hard to learn
17:18:21S_a_i_n_tTrial & error works best as a teaching tool in cases such as this.
17:18:43angelwolf71885well probbably not much more luck xD thats kinda why i was asking if re arangeing the stat bar was from themes lol
17:19:45angelwolf71885yah figureing it out is going to take playing with it and recording the results thanks for the clues :)
17:20:55 Quit Blue_Dude (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
17:26:50 Join avar [0] (avar@wikipedia/avar)
17:28:52 Join Zarggg [0] (
17:30:13 Join CGL [0] (~CGL@
17:30:21 Join phanboy4 [0] (
17:30:30 Quit Adubb (Read error: Connection reset by peer)
17:35:04 Join Adubb [0] (~aldubuc@
17:36:19kugelcan I get the value of compile time constants during compilation?
17:39:32 Quit CaptainKewl (Quit: ( :: NoNameScript 4.22 :: ))
17:40:24kugelI assume that we - instead of using a divider for clocking down - clock down pclk directly which also slows down all peripherals including dbop
17:40:58kugel(I don't know whether something else is actually possible)
17:41:32kugelbut even at 60MHz DBOP is too slow, something is wrong since on fuzev1 DBOP at 62MHz works just fine
17:45:25 Quit Xerion (Ping timeout: 245 seconds)
17:58:26 Join jennifur [0] (
18:03:17 Join funman [0] (~fun@rockbox/developer/funman)
18:04:05 Quit pamaury (Ping timeout: 252 seconds)
18:04:38 Quit phanboy4 (Read error: Connection reset by peer)
18:04:44 Join pamaury [0] (~c2c7a50a@rockbox/developer/pamaury)
18:14:46funmankugel: <- here is the code I used to verify that the peripheral clock (at least the one used by the timer) is independant from the CPU clock
18:16:52kugeldoesn't the timer have its own frequency (1.2MHz?)
18:17:35*funman slaps himself repeatedly
18:17:37funmanyeah ..
18:19:15 Quit angelwolf71885 (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
18:21:11kugelbut anyway, the lcd performance just sucks even at 60MHz the UI is really sluggish, we must be missing something
18:22:08kugelin the meanwhile I test whether the button reads disturb the lcd too much (I deactivated the button read routines at ran test_fps via autorock) but they dont
18:22:53kugeldo you think the pixel swapping makes it slower? I almost can't believe that
18:23:13funmanhm that could be
18:23:30kugelit shouldn't make it 90% slower though?
18:24:59 Quit Hillshum (Remote host closed the connection)
18:27:26 Quit kadoban (Read error: Connection reset by peer)
18:27:47bertrikI guess there's still some divider somewhere that makes it run within range at 240 MHz and severely low at 24 MHz
18:27:48funmanmemory speed is much lower on fuzev2 than on fuzev1 indeed
18:29:00kugelwe should turn off the speculative instruction caching, we don't need that IMO
18:31:05 Part jennifur
18:31:40kugelthe tcm stuff is mystic to me too
18:32:11*funman notes to read this stuff later
18:33:03 Join cousteau [0] (
18:35:56funmanidea: try writing different values to CGU_PERI[1:0] now that we can change CPU freq
18:36:20 Join jennifur [0] (
18:37:47kugelbtw, is there a reason for 24MHz?
18:37:55 Quit pixelma (Disconnected by services)
18:37:56 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
18:38:14funman"low, and around e200v1 frequency"
18:38:16 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
18:39:07funmanI noticed weird noises in the headphones while listening to Clipv2: the noise seems to appear each time the CPU is boosted, but doesn't go away when I force it to be always boosted. I'll need to investigate further
18:39:28 Quit amiconn (Disconnected by services)
18:39:28 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
18:39:40 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
18:39:59funmanI'm sorry to have brought so many problems, but I think it helps us understanding better. If you feel it creates too much problems reverting should be as easy as commenting out ADJUSTABLE_CPU_FREQ in config/*.h
18:40:24*kugel is curious about TIMERIF_EN (or the timer module APB IF in general)
18:40:45kugelfunman: no need to be sorry, it shows we're not there yet ;)
18:43:03cousteaujust a quick question: this firmware isn't intended for low-end mp3+video players like this right?
18:43:26cousteaujust more advanced things like iPods and the like?
18:43:53gevaertscousteau: the more advanced ones are mor likely to have hardware powerful enough to run rockbox, yes
18:44:29gevaertsbut it's impossible to see what hardware is inside on a picture like that :)
18:44:33 Quit pixelma (Ping timeout: 265 seconds)
18:44:40 Quit amiconn (Ping timeout: 264 seconds)
18:44:51cousteauok, let me see if I can find the specs...
18:45:02Farthenthe display doesn't look very good :-) I don
18:45:21Farthen't think that you'd have much fun with it
18:45:41gevaertsFarthen: have you ever seen a c200 display?
18:45:51*cousteau is considering building his own mp3 player
18:46:00gevaertsOr the display on the iaudio M3? ;)
18:46:21Farthenno, only experience with some creative devices and ipods so far
18:47:08cousteauI really don't know why these devices are "video"-enabled
18:47:11Farthencousteau: it would be much more expensive than buying an "advanced" DAP
18:47:40cousteaui.e. a weird video format called MTV which specs are very fuzzy
18:47:47Farthenand it would be pretty clumsy
18:47:59funmantest_fps shows comparable results for clipv1 & v2 when boosted (although there's an error when calculating CPU load), but Clipv2 is much slower (4x to 8x) when unboosted
18:48:00cousteau(well, it simply has no specs)
18:48:41gevaertscousteau: ah, that points to it being one of those Chinese 8-bit or 16-bit CPUs with a DSP. If so, you won't ever run rockbox on it
18:49:45cousteauyes, typical Sigmatel thing, I think
18:50:52cousteauit plays songs, radio, and has hundreds of features I'll never use
18:51:31cousteausuch as video, images, lyrics, games...
18:52:25cousteau"e-books" (i.e. displays plain txt files)...
18:54:00 Join FlynDice [0] (~FlynDice@
18:56:04 Quit antil33t (Read error: Connection reset by peer)
18:56:10 Join antil33t [0] (
18:56:31FlynDicefunman: re try writing different values to CGU_PERI[1:0] : I have done this and when I go check the register value on the debug page b1:0 are always cleared.
18:56:50 Join Strife89|PalmTX [0] (~cstrife89@
18:57:18funmanthe bits could be write only though
18:57:55 Join mischasworld [0] (
18:57:56 Part cousteau
18:57:56*FlynDice has crappy hotel connection and intermittant freezes with new laptop ubuntu setup so apologizes for being here then not in advance... ;-)
18:58:27 Part jennifur
18:58:53*funman propose to send the rockbox ninjas at the said hotel, to get the connection fixed
18:59:06 Quit kugel (Remote host closed the connection)
18:59:26 Quit Strife89|PalmTX (Client Quit)
18:59:37*FlynDice assumes lotus position and waits patiently
19:01:13FlynDiceI can't think of a good reason that those bits would be write only
19:01:14*funman acknowledges that CGU_PERI represents Buddha, and walks without doubts
19:01:35funmanyep :/
19:01:55***Saving seen data "./dancer.seen"
19:04:22 Quit n1s (Quit: Lmnar)
19:05:44 Join froggyman [0] (~me@unaffiliated/froggyman)
19:08:48 Join amiconn [0] (quassel@rockbox/developer/amiconn)
19:09:08 Join pixelma [0] (quassel@rockbox/staff/pixelma)
19:10:47CIA-5New commit by funman (r25455): Clipv2: fix grey lcd output ...
19:11:19*funman feels like it was not very important at this point ..
19:13:20 Quit amiconn (Ping timeout: 245 seconds)
19:13:45 Quit pixelma (Ping timeout: 258 seconds)
19:14:39 Quit funman (Quit: free(random());)
19:15:15 Quit linuxstb (Ping timeout: 248 seconds)
19:15:25 Quit FlynDice (Ping timeout: 245 seconds)
19:15:40 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
19:21:32 Join arbingordon [0] (~w@unaffiliated/arbingordon)
19:22:15 Quit linuxstb (Ping timeout: 265 seconds)
19:22:42 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
19:24:08 Join JohannesSM64 [0] (
19:25:45 Join moos_ [0] (
19:25:50 Join dockimble [0] (~dockimble@
19:27:44gevaertsWhat's the typical worst-case seek time on a slowish disk?
19:27:56 Quit moos (Ping timeout: 258 seconds)
19:27:56 Nick moos_ is now known as moos (
19:30:44 Join amiconn [0] (quassel@rockbox/developer/amiconn)
19:30:48 Join FlynDice [0] (~FlynDice@
19:31:50 Quit FlynDice (Read error: Connection reset by peer)
19:32:07 Join FlynDice [0] (~FlynDice@
19:33:21 Quit FlynDice (Client Quit)
19:33:40 Join FlynDice [0] (~FlynDice@
19:35:29 Quit FlynDice (Read error: Connection reset by peer)
19:35:42 Join FlynDice [0] (~FlynDice@
19:36:16 Quit amiconn (Ping timeout: 265 seconds)
19:38:16 Quit FlynDice (Client Quit)
19:38:31 Join FlynDice [0] (~FlynDice@
19:45:14 Join TheSeven [0] (~theseven@rockbox/developer/TheSeven)
19:47:55 Quit FlynDice (Ping timeout: 245 seconds)
19:48:09 Quit tchan (Quit: WeeChat
19:51:31 Quit xiainx (Ping timeout: 264 seconds)
19:53:15 Quit JohannesSM64 (Quit: WeeChat 0.3.2-dev)
19:57:45 Quit xavieran (Ping timeout: 276 seconds)
19:58:24 Quit Schmogel (Quit: Miranda IM! Smaller, Faster, Easier.
20:00:22 Join JohannesSM64 [0] (
20:01:47 Join tchan [0] (~tchan@lunar-linux/developer/tchan)
20:05:46 Quit roolku ()
20:09:27 Join CaptainKewl [0] (
20:10:21 Join xavieran [0] (
20:13:05 Join fml [0] (
20:15:28fmlHello. I wonder whether we could/should push the "?" into the localized string in onplay.c:1331. This would also eliminate the second call (appending "?").
20:16:56 Join Zagor [0] (~bjst@rockbox/developer/Zagor)
20:16:59fmlAlso, I have a somewhat strange effect when I assign the hotkey function: after I confirm that I do want to change the assignment, the confirmation is displayed both in a "pop up" and as a printed text at the top of the screen.
20:17:08 Join Luca_S [0] (
20:17:49CIA-5New commit by zagor (r25456): Moved old news items to history.t
20:18:23gevaertsfml: I'm not *entirely* sure, but it looks as if the code makes two messages, one with the ? and one without
20:18:36Luca_SFuzeV2 battery bench ended. lame_128.mp3 played in loop for 17 hrs 50 minutes :)
20:18:50gevaertsIf so, I think the code is wrong. You can't just assume that you can make a question by appending "?" in any language
20:19:23Luca_Sthe OF still boots after RB shutdown however
20:19:25Luca_Sit even plays
20:19:58bertrikbut it does show nearly empty battery,right?
20:20:00Luca_Salthough the battery indicator is nearly empty
20:20:12fmlgevaerts: where are the two messages? I only see one.
20:20:30gevaertsfml: strcpy(line2, line1)
20:20:44Luca_SRB battery reading says 3.348
20:20:57gevaertsIt copies the string from line1 to line2, and appends "?" to line2
20:21:34Luca_Sanything else to check before recharging it?
20:22:41bertrikI don't know, would be fun perhaps to do a battery bench while charging
20:22:41fmlgevaerts: ah, and both are used! What do the params of gui_syncyesno mean?
20:22:57gevaertsfml: that's GUI code! I don't know that! :)
20:23:26Luca_Sbertrik: I suppose you are talking about a real battery bench with filesystem write support?
20:23:48bertrikoh yeah, I forgot about that
20:24:55gevaertsLuca_S: well, that does point to what you could do next :)
20:25:29Luca_Si'll ask for write support to the easter bunny :D
20:25:41Luca_Sthat's all I can do :D
20:26:05Luca_Sgotta go now, see you tomorrow
20:27:01fmlgevaerts: I would push the ? into the localized string and just use "yes" and "no" as the possible answers. The current code repeats the action text as the yes-answer.
20:28:30gevaertsfml: maybe. I haven't looked at the actual screen
20:32:25 Join FlynDice [0] (~FlynDice@
20:32:40fmlgevaerts: I will do it, both because of efficiency and better localizability
20:34:13gevaertsZagor: did you see roolku's comment at ?
20:35:01gevaertsAlso, 1:42 isn't too bad :)
20:37:27ZagorI don't see what he is referring to
20:37:51 Quit dockimble (Quit: WeeChat 0.2.1)
20:39:03gevaertsI don't really know either
20:39:07 Quit FlynDice (Ping timeout: 260 seconds)
20:39:12 Join kugel [0] (~kugel@rockbox/developer/kugel)
20:44:27*kugel wonders what funman meant with "*funman acknowledges that CGU_PERI represents Buddha, and walks without doubts"
20:45:10fmlHrm... May I change the recently introduced hotkey entries in the lang files? I woudn't like to deprecate them and introduce new, they're very fresh.
20:45:47kugelwhy deprecate?
20:45:56fmlI want to introduce a separate string for the question and for the confirmation
20:46:31fmlkugel: because they already exist and I want to chnage them (also the names)
20:46:41 Join saratoga [0] (
20:46:56kugelnot sure why they need to be separate but can't you change the one and add the other one?
20:47:05saratogaunrolling the test_mem loops a little might give more accurate numbers
20:47:15saratogae.g. doing 4 loads per loop iteration instead of just 1
20:47:20 Join tomers [0] (
20:47:52kugelsaratoga: I thought that too but the array index needs to be incremented anyway? or do you mean the branch?
20:48:14fmlkugel: because now the question is just constructed from the confirmation by adding a ?. This works for English but won't for e.g. German.
20:48:14saratogakugel: the branch
20:48:15kugelhm maybe not on arm
20:48:40saratoga{x = buf[j]; x = buf[j+1]; x = buf[j+2]; x = buf[j+3];}
20:48:46kugelfml: works for my German :)
20:49:01gevaertsWell, it definitely won't work in general
20:49:21saratogathe branch on some targets will be relatively slow, even compared to memory reads
20:49:26kugelprobably, but I still don't understand why that one needs to be incremented?
20:49:42saratogawhich needs?
20:52:05kugeloops, I meant deprecated w.r.t. to fml's issue
20:52:14saratogaoh ok
20:52:33 Join Bagder [0] (~daniel@rockbox/developer/bagder)
20:52:43saratogai was just thinking on a target like PP, the loop is 2-3 cycles and the add one cycle, while a load to IRAM is just 2 cycles
20:52:44gevaertsfml: I guess not deprecating is fine. There probably aren't translations yet anyway
20:53:24fmlgevaerts: I think so too. I'll upload a patch soon.
20:53:32 Join dockimble [0] (~dockimble@
20:53:39 Quit catfish (Read error: No route to host)
20:54:07 Join catfish [0] (
20:55:34 Quit pamaury (Quit: Page closed)
20:57:26Luca_SI just upgraded to the latest svn
20:57:52Luca_Sthe brightness issue is gone, thank you ;)
20:59:04 Join jennifur [0] (
20:59:21Luca_Sand plugging usb briefly displays the usb logo, then reboots
20:59:46 Quit catfish (Ping timeout: 248 seconds)
21:00:05 Join pamaury [0] (~c2c7a50a@rockbox/developer/pamaury)
21:00:14kugelsaratoga: with unrolling I get an add between each ldr/str
21:01:58***Saving seen data "./dancer.seen"
21:06:25saratogakugel: shouldn't it just be an indexed load?
21:06:26kugelbut it looks like they're free
21:06:31saratogalike LDR r0,[r1,#8]; LDR r0,[r1,#12], etc
21:07:08kugelI thought that too but gcc does "strr2, [r6, r3, lsl #2]"
21:08:32 Quit Luca_S (Quit: CGI:IRC)
21:08:38saratogamaybe its worth trying ASM here
21:08:47saratogawe could also try running it with multiple loads and single loads
21:08:57saratogai'll take a look when i get into work
21:09:13kugelI can't asm for the other arch's though :(
21:09:33saratogayeah that'll have to stay c for now
21:10:21saratogathis plugin could be really helpful for understanding the werid performance we see on ARM targets
21:10:36saratogawhich reminds me, does anyone have a D2 or Mrobe500 that can run test codec on it?
21:13:17kugelsaratoga: we could also test the iram with it
21:15:26saratogakugel: nonsequential accesses would also be very interesting
21:15:34 Quit CGL (Quit: Saliendo)
21:15:44saratogastep 32 bytes at a time so we don't hit the same cacheline
21:16:18 Quit mischasworld (Ping timeout: 265 seconds)
21:18:48kugelwe possibly could get the add away with casting to char array then casting back
21:20:15kugelbut they seem free anyway if I get that right
21:20:45kugelunless we have 1c ldr/str
21:20:54 Quit tomers (Ping timeout: 264 seconds)
21:23:53 Join tomers [0] (
21:33:48fmlgevaerts: patch is up, FS #11171
21:34:39 Join xiainx [0] (
21:35:09fmlkugel: you know the gui code. Could you also review the patch (FS #11171)?
21:38:45kugelmaybe add a to in the lang strings?
21:40:55fmlkugel: you mean "Hotkey assigned to ..."? I think it would be wrong. The function is assigned to the hotkey, not the hotkey to function.
21:41:36 Join kugel_ [0] (
21:41:52 Quit kugel (Disconnected by services)
21:41:53 Quit saratoga (Quit: CGI:IRC (EOF))
21:41:56 Nick kugel_ is now known as kugel (
21:41:56LloreanTo some users at least, since it's one key and multiple functions, it feels like you're picking which function to assign the key to.
21:42:00 Quit kugel (Changing host)
21:42:00 Join kugel [0] (~kugel@rockbox/developer/kugel)
21:42:12kugelfml: the last two lines of the patch look wrong
21:43:18 Quit tomers (Quit: ChatZilla 0.9.86 [Firefox 3.5.8/20100214235838])
21:43:48kugelwhy a splash and not the yes/no messages of the yesno screen?
21:46:19 Join catfish [0] (
21:48:00fmlBecause yesno message is a question. The splash is a confirmation
21:51:05kugeldo we have a confirmation after the yesno screen anywhere else?
21:52:22kugelfml: the yesno screen shows a confirmation message itself
21:54:38fmlkugel: ah, so the yes parameter is the confirmation message?
21:54:43 Quit catfish (Ping timeout: 260 seconds)
21:54:53fmlI thought it's the string to show for "yes"
21:55:44kugelit has a confirmation and a cancel message
21:57:29 Join roolku [0] (
21:59:46 Join FlynDice [0] (~FlynDice@
22:00:55 Quit Bagder (Quit: It is time to say moo)
22:01:57roolkuZagor gevaerts: in 25453 n02 didn't get any jobs, same happened to n09,n15 in 25452 and n09 in 25451 even though they have been online same as the rest, all back in 25454, 25455
22:02:06 Quit anewuser ()
22:02:54Zagorroolku: hmm, strange
22:04:01roolkun04 is offline for real
22:04:15fmlkugel: what happens if one of the parameters is set to NULL? No confirmation?
22:04:46Zagorroolku: aha, "Fatal build error: Missing log file. Blocking n02-roolku."
22:04:56Zagorthat indeed looks like a bug
22:05:48 Join phanboy4 [0] (
22:09:22 Join TMM [0] (~hp@pdpc/supporter/professional/TMM)
22:09:31roolkuZagor: looking at rashers charts, it seems the new servers starting fairly late (large gap before the green bar) - any idea what could be causing this?
22:10:40 Quit TMM (Remote host closed the connection)
22:10:54 Join TMM [0] (~hp@pdpc/supporter/professional/TMM)
22:11:05Zagorroolku: which client are you looking at?
22:11:22 Quit TMM (Read error: Connection reset by peer)
22:11:51roolkuany n* server building a real target (i.e. green bar)
22:12:36Zagorhmm? they all look fine to me. perhaps it's a html/css issue?
22:12:57 Join TMM [0] (~hp@pdpc/supporter/professional/TMM)
22:13:09Zagoroh I do see what you mean in the non-debug view
22:13:18 Quit TMM (Read error: Connection reset by peer)
22:13:49 Join TMM [0] (
22:13:51 Quit TMM (Changing host)
22:13:51 Join TMM [0] (~hp@pdpc/supporter/professional/TMM)
22:13:59 Quit TMM (Read error: Connection reset by peer)
22:14:22kugelfml: I assume so
22:14:33 Quit FlynDice (Read error: Connection reset by peer)
22:14:33kugelsvn gives a confirmation message but no cancel message
22:14:46 Join FlynDice [0] (~FlynDice@
22:15:29Zagorroolku: that is a display issue with rasher's script. the log file shows they all started the same second
22:16:03kugelthey're too fast for the scripts!
22:16:36roolkuokay, fair enough. It seemed strange that it effected the new servers to a greater extend
22:17:22roolkuthey don't seem to like building checkwps though - takes them nearly as long as a proper build :)
22:17:49Zagoryes, that is rather strange
22:18:13ZagorI wonder what is taking so long
22:18:22 Quit FlynDice (Client Quit)
22:18:58 Join catfish [0] (
22:19:31roolkuI had to cheat and add them in small groups otherwise they would only get bootloaders and checkwps and the score would stay low.
22:19:44fmlkugel: new version of the patch is up (FS #11171)
22:20:15 Join FlynDice [0] (~FlynDice@
22:20:52 Quit moos (Read error: Connection reset by peer)
22:20:57 Quit catfish (Client Quit)
22:21:10Zagorroolku: it would most likely have worked out anyway since those machines really rip up the speculative builds
22:21:19 Join pixelma [0] (quassel@rockbox/staff/pixelma)
22:21:29 Join amiconn [0] (quassel@rockbox/developer/amiconn)
22:21:36Zagorbut it is indeed unfortunate that the benchmark builds perform so poorly. and not just for your clients.
22:21:57kugelalso, r25453 had a strange binsize change for the ipod4ggrey
22:22:12 Join moos [0] (moos@rockbox/staff/moos)
22:22:29kugelthose seem happen from time to time
22:23:09Zagoryes, there is some uploading bug that sometimes gives files the wrong names
22:23:17roolkuhave a look at n15 and n18 in"> they just didn't get any proper builds and couldn't improve their score - I had to take out machines with high scores to give thenm a chance :)
22:23:30Zagorand that causes client block (log file not found) and wrong binsize diffs
22:23:40 Part jennifur
22:23:44roolkuZagor: good find
22:24:08kugelgevaerts: why HAVE_IO_PRIORITY ? it seems like a good thing to do on any target
22:24:09Zagorunfortunately I haven't found the cause of it :-(
22:24:34 Quit FlynDice (Ping timeout: 245 seconds)
22:24:55 Join FlynDice [0] (~FlynDice@
22:26:55 Quit pixelma (Ping timeout: 260 seconds)
22:27:20 Quit amiconn (Ping timeout: 265 seconds)
22:31:40 Quit Zagor (Remote host closed the connection)
22:35:26 Join scott666 [0] (
22:36:32fmlI'm about to commit FS #11171 I'd like to do it quickly because it changes lang entries.
22:38:23 Quit FlynDice (Remote host closed the connection)
22:39:44scott666hi, I just got installed a 240GB HD in my ipod video and have been having problems with the boot loader. I installed cygwin and recompiled the source making the changes mentioned in this link but it still does not work. is there someone that can help me?
22:46:12 Join pixelma [0] (quassel@rockbox/staff/pixelma)
22:46:37 Join amiconn [0] (quassel@rockbox/developer/amiconn)
22:47:14 Join geertvdijk [0] (
22:48:13 Join shai [0] (
22:52:31kugelscrollwheel irq isn't going to work well I think
22:53:02kugelI think we can't reliably tell the time difference between 2 interrupts
22:54:37kugelthe timer may wrap, and its maximum value seems to be 15k
22:55:06 Join blairb [0] (
22:56:23 Quit the_miker ()
22:59:37CIA-5New commit by alle (r25457): Make hotkey strings (question and confirmation) better localizable (FS #11171)
22:59:58 Join Strife1989 [0] (
23:02:01***Saving seen data "./dancer.seen"
23:10:48 Join jennifur [0] (
23:15:48 Join saratoga [0] (~9803c6dd@gateway/web/freenode/x-paffiephyotzxixs)
23:20:04gevaertskugel: maybe, maybe not. I'm not sure if it's a win on flash, and it definitely only makes sense if there is a background thread that does IO. If you do a no-database no-dircache build, the priority code is useless
23:20:36kugelisn't it less than 300 bytes?
23:20:58amiconnBtw, it's not so sure that different disks don't affect each other.
23:21:23amiconnIt depends on the controller - it's quite possible that parallel access is impossible
23:21:49saratogai keep getting "internal compiler errors" trying to do simple ASM statements
23:21:54saratogaanyone else seen these
23:22:00 Quit efyx (Remote host closed the connection)
23:22:02amiconnThat is true for ide disks on a single channel (master/slave), as well as e.g. the Ondio (but that does neither have dircache nor database in ram)
23:22:28kugelthat's the case on the as3525v2, I don't know if parallel accesses are implemented on as3525 (two controllers) but probably not because of the limited dma capabilities
23:22:44 Join r2k000 [0] (
23:23:04r2k000Hi All
23:23:04kugelbut the driver will handle it in case parallel access is impossible so I wouldn't worry
23:23:36 Quit dockimble (Quit: WeeChat
23:23:45gevaertsamiconn: true, so we might want to use single numbers for multi-drive single-controller flash systems. FOr (hypothetical) two-disk ide systems I think we still want to separate them to avoid excessive seeking
23:24:14gevaertsBut worst case is that it behaves like it does without the patch I think
23:24:36*amiconn thinks the Elio is very special regarding this (doesn't run rockbox yet though)
23:24:37saratogadoes this look like it shouldn't parse (ignoring that it obviously doesn't do anything useful):
23:25:34gevaertsamiconn: isn't that going to be the same as the D2, i.e. two different controllers?
23:25:57 Quit liar (Ping timeout: 258 seconds)
23:26:20amiconnThe two controllers are controlling storage with different seek behaviour (hdd vs. flash)
23:26:32 Quit phanboy4 (Read error: Connection reset by peer)
23:26:51 Quit pamaury (Quit: Page closed)
23:28:20 Part jennifur
23:28:21CIA-5New commit by alle (r25458): No need to call the same function twice
23:28:34gevaertsI'm still not entirely sure if the relevant numbers have to be different for flash and disk
23:29:21 Quit JohannesSM64 (Excess Flood)
23:30:35 Join JohannesSM64 [0] (
23:30:46gevaertsSTORAGE_MINIMUM_IDLE_TIME should be bigger than the maximum amount of time between two requests from a high priority thread that are considered to be consecutive. I've set that to HZ/10, which might be a bit too high, but I suspect that seek times of 20ms or so won't be that uncommon, so at least on disk targets it can't be much lower
23:33:16gevaertsSTORAGE_DELAY_UNIT should be smaller than STORAGE_MINIMUM_IDLE_TIME, since that's the time you wait before trying again. It's also used (but that can be tweaked) to determine the maximum waiting time since the system will not wait more than <priority> times
23:33:51gevaertsHZ/20 means (with current priority definitions) a maximum wait of about 1.5 seconds
23:34:41gevaertsI think we can keep both numbers for now, possibly reducing IO_PRIORITY_BACKGROUND a bit (to 8 maybe?)
23:36:50 Quit r2k000 (Quit: r2k000)
23:37:58 Join anewuser [0] (anewuser@unaffiliated/anewuser)
23:38:23gevaertsAnyway, the thing seems to work well for me. Any objections to committing it?
23:40:33 Join dockimble [0] (~dockimble@
23:43:29 Join CGL [0] (~CGL@
23:44:36 Quit fml (Quit: CGI:IRC 0.5.9 (2006/06/06))
23:47:34 Join angelwolf71885 [0] (
23:55:28 Quit linuxstb (Read error: Operation timed out)
23:58:41 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)

Previous day | Next day