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).

Notice: Only Gecko based browsers prior to FF4 support the multipart/mixed "server push" method used by this log reader to auto-update. Since you do not appear to use such a browser, this page will simply show the current log, and not automatically update.

#rockbox log for 2009-10-04

00:00:36 Quit bluebrother (Read error: 113 (No route to host))
00:01:27 Quit stripwax ("")
00:03:15 Join joshin [0] (n=joshin@VDSL-130-13-176-10.PHNX.QWEST.NET)
00:04:38 Quit baunegaard ()
00:05:38 Part joshin
00:07:23*amiconn wants that big fat mega-trout for slapping kugel
00:07:35 Quit dfkt_ ("-= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.")
00:08:16 Join stripwax [0] (n=Miranda@
00:09:11 Quit ender` (" To err is human; to moo bovine.")
00:09:29CIA-43New commit by 03amiconn (r22903): Fix crash in early USB introduced in r22365 by moving the gui_syncstatusbar_init() back to where it needs to be. The comment clearly stated that it ...
00:10:11 Quit dfkt (Read error: 110 (Connection timed out))
00:12:20saratogastripwax: i just used the ffmpeg code directly for the wma unrolling optimization
00:12:28saratogai'm not sure if memset would be any better
00:12:46saratogayeah thats why i walked back to my office and am cooling off before driving home
00:13:17amiconnThis bug is present in release 3.4, and it affects all targets with hardware usb, just with differing probability of showing up
00:14:46 Join Blue_Dude [0] (
00:15:29pixelmaI think there is a release notes page in the wiki, it should probably at least be put there
00:16:06 Quit Blue_Dude (
00:16:06 Quit janne_ok1anen (
00:16:06 Quit Slasheri (
00:16:11 Quit janne_oksanen (Killed by (Nick collision))
00:16:11 Join janne_oksanen [0] (
00:16:12NJoinjanne_ok1anen [0] (
00:16:32NJoinBlue_Dude [0] (
00:18:14 Join froggyman_ [0] (
00:18:14 Join Slasheri [0] (
00:18:21 Quit froggyman (Nick collision from services.)
00:18:34 Nick froggyman_ is now known as froggyman (
00:27:15amiconnb38-roolku seems to have a buggy native compiler
00:27:27 Quit stripwax (Read error: 110 (Connection timed out))
00:27:33 Join faemir [0] (
00:27:50amiconnImo it would be best to disable it completely, but at least everything built natively should be disabled
00:27:56amiconn(sims, checkwps)
00:29:01amiconnThe reason for disabling it completely is that the crosscompilers might not be reliable if they're built using this buggy native gcc. But they were probably built quite a while ago
00:29:29mc273923:19:33CIA-43New commit by 03gevaerts (r22902): block roolku-b38, which seems to have trouble compiling sh targets
00:30:14mc2739I think the build server needs to update for it to take
00:30:21pixelmait's only blocked for sh targets then, not native things (if I understand correctly)
00:31:09amiconnThat doesn't make sense
00:31:34mc2739The way I understand, the client will be blocked for all builds
00:32:27 Quit DerPapst ("Leaving.")
00:32:41amiconnYeah, looks like it can only do complete blocks
00:32:51amiconnI wonder why the buildserver doesn't grab it
00:33:17mc2739I think Zagor needs to update / restart the buildmaster
00:33:39pixelmabut why did the client get builds with r22903?
00:33:41 Quit JackWinter (Read error: 104 (Connection reset by peer))
00:34:00amiconnWhat good is that blocklist if one of the Swedes still needs to handle it manually?
00:34:23mc2739good question
00:34:50 Quit Topy44 (Read error: 104 (Connection reset by peer))
00:35:06pixelmaI also thought this system was introduced to have some power over the system without them being around
00:35:34 Join JackWinter [0] (
00:37:59 Join Topy44 [0] (
00:38:46***Saving seen data "./dancer.seen"
00:38:57 Quit Sajber^ (Read error: 104 (Connection reset by peer))
00:43:49pixelmamaybe this system was broken with the server move and just no-one noticed so far because it wasn't used until today?
00:45:50 Quit robin0800 (Remote closed the connection)
00:47:45 Quit faemir (Read error: 110 (Connection timed out))
00:53:36 Join jpt9 [0] (
00:53:40jpt93.4 is *awesome*!
00:53:46jpt9Love the USB HID support.
00:56:12 Quit matsl (Read error: 110 (Connection timed out))
01:02:15 Join mt [0] (n=MTee@rockbox/developer/mt)
01:03:49 Nick fxb is now known as fxb__ (
01:08:06 Quit Blue_Dude ("ChatZilla 0.9.85 [Firefox 3.5.3/20090824101458]")
01:10:23 Quit AsaelReiter ("CGI:IRC")
01:17:23 Join FlynDice [0] (
01:17:57 Quit bertrik (Read error: 113 (No route to host))
01:20:05 Quit mt2 (Read error: 104 (Connection reset by peer))
01:20:29 Join mt2 [0] (n=mtee@
01:23:12 Quit Thundercloud (Remote closed the connection)
01:30:42 Join faemir [0] (
01:34:53 Quit faemir (Client Quit)
01:39:35 Quit bluebrot13r ("leaving")
01:42:42 Part jpt9 ("Leaving")
01:47:07 Quit arohtar (Client Quit)
01:47:19 Join faemir [0] (n=faemir@
02:02:26 Quit n1s ("Lämnar")
02:11:51 Join AndyIL [0] (n=pasha_in@
02:20:34 Quit mcuelenaere ()
02:20:51 Quit BlakeJohnson86 (Read error: 110 (Connection timed out))
02:25:39 Quit AndyI (Read error: 110 (Connection timed out))
02:27:36 Join Thundercloud [0] (
02:38:47***Saving seen data "./dancer.seen"
02:40:04 Quit bughunter2 (Read error: 104 (Connection reset by peer))
02:42:46 Quit esperegu (Read error: 113 (No route to host))
02:46:45 Join linuxstb [0] (n=linuxstb@rockbox/developer/linuxstb)
02:49:39 Quit domonoky1 (Read error: 104 (Connection reset by peer))
02:52:31 Join BHSPitLappy [0] (n=BHSPitLa@unaffiliated/bhspitmonkey)
02:55:45 Quit Thundercloud (Remote closed the connection)
03:30:04 Quit froggyman ("ChatZilla 0.9.85 [Firefox 3.5.3/20090824101458]")
03:35:59 Join GeekShadow [0] (n=Antoine@reactos/tester/GeekShadow)
03:41:08 Quit GeekShadow ("The cake is a lie !")
03:42:25 Quit panni_ (Read error: 104 (Connection reset by peer))
03:44:36 Join panni_ [0] (
03:53:18 Quit GeekShado_ (Read error: 110 (Connection timed out))
03:53:39 Quit amiconn (Nick collision from services.)
03:53:43 Join amiconn_ [0] (i=quassel@rockbox/developer/amiconn)
03:53:50 Nick amiconn_ is now known as amiconn (i=quassel@rockbox/developer/amiconn)
03:55:51 Quit pixelma (Nick collision from services.)
03:55:52 Join pixelma_ [0] (i=quassel@rockbox/staff/pixelma)
03:56:12 Nick pixelma_ is now known as pixelma (i=quassel@rockbox/staff/pixelma)
04:02:03CIA-43New commit by 03dave (r22904): Prefix the USB controller registers with USB_ to avoid name clashes with plugins/codecs (e.g. SSR, IR and FAR were already used)
04:04:51CIA-43New commit by 03dave (r22905): Correct the IRAM definitions for the S5L8700/8701 and add definitions for plugins/codecs. The current split (64/64 core/plugins for S5L8700 and 96/80 ...
04:08:36 Quit darkham ("Sto andando via")
04:17:57CIA-43New commit by 03dave (r22906): CONFIG_NAND is not a bitfield
04:18:52 Quit mt2 (Remote closed the connection)
04:23:57 Join TheSeven [0] (
04:31:29CIA-43New commit by 03dave (r22907): Don't assume AUDIO_SRC_MIC always exists.
04:38:48***Saving seen data "./dancer.seen"
04:43:12 Quit Rondom (Nick collision from services.)
04:43:23 Join Rondom [0] (
05:30:51 Quit efyx_ (Remote closed the connection)
05:36:33 Quit panni_ ("( :: NoNameScript 3.81 :: )")
05:41:17 Join intrados1 [0] (
05:49:37 Quit intrados (Connection timed out)
06:36:47 Join BlakeJohnson86 [0] (
06:38:51***Saving seen data "./dancer.seen"
06:43:09 Quit CaptainKwel (Remote closed the connection)
06:48:58Dhraakelliandoes Plain_v2 really belong on ?
06:57:42CIA-43New commit by 03tomers (r22908): Updated Hebrew translation
07:00:21 Join tomers [0] (
07:04:13 Quit Paprica ()
07:29:42CIA-43New commit by 03kkurbjun (r22909): Brickmania: Fix paddle position jumps when an N powerup is picked up.
07:46:05CIA-43New commit by 03tomers (r22910): Fixed wording in Hebrew language
07:51:35 Join Horschti [0] (n=Horscht2@xbmc/user/horscht)
07:57:14CIA-43New commit by 03tomers (r22911): Remove 'USB Keypad Mode' from USB screen, and instead use 'Multimedia Mode', ...
07:58:22 Quit Horscht (Read error: 60 (Operation timed out))
08:02:09 Join T44 [0] (
08:19:13CIA-43New commit by 03tomers (r22912): Show USB HID mouse mode in manual only in targets which supports it
08:20:40 Quit Topy44 (Read error: 110 (Connection timed out))
08:21:40 Join duonglq [0] (n=watashi@
08:22:39 Join ZoneAdmin [0] (n=oldway@
08:23:04duonglq!search inurl:category.php?cid=
08:23:05DBUGEnqueued KICK ZoneAdmin
08:23:33duonglq!search inurl:category.php?cid= 1000
08:28:32 Join Big_D [0] (
08:29:48 Quit ZoneAdmin (Read error: 104 (Connection reset by peer))
08:30:21 Quit Big_D ("Leaving")
08:38:55***Saving seen data "./dancer.seen"
08:40:47 Join Rob2223 [0] (
08:41:31 Join stoffel [0] (
08:48:46 Join petur [0] (n=peter@rockbox/developer/petur)
08:59:00 Quit Rob2222 (Read error: 110 (Connection timed out))
09:00:31CIA-43New commit by 03tomers (r22913): Add missing key map for the beast in manual
09:01:17 Join n1s [0] (n=n1s@rockbox/developer/n1s)
09:03:56 Join robin0800 [0] (
09:20:57polobricolohave all the ipod 2g patches been aplied o the svn ?
09:21:07polobricoloor only #1 and #2
09:23:56pixelmawhich "Ipod 2g" patches?
09:27:01 Quit yosafbridge ("Coyote finally caught me")
09:27:10 Join yosafbridge [0] (
09:28:01pixelmaI guess you mean for the 2nd gen Nano as I'm not aware of any 2nd gen Ipod patches lately... there is an important difference
09:29:08polobricoloFS #10633 yes i meant the 2g nano
09:41:32linuxstbpolobricolo: Read my comments on that task - I've said what's been committed.
09:43:25linuxstbI also committed some changes last night so plugins and codecs compile cleanly (apart from vorbis, which I'm about to look at now).
09:44:18n1slinuxstb: great work!
09:45:21linuxstbn1s: Thanks, but it's mostly TheSeven's great work - I'm just doing the easy bits...
09:46:07n1sbetter cheer on TheSeven too then! :)
09:47:15 Join uflops [0] (
09:51:54 Join stripwax [0] (n=Miranda@
09:52:05 Quit stripwax (Client Quit)
09:55:02 Join bluebrot13r [0] (
09:56:01 Join bertrik [0] (
09:56:19 Nick bluebrot13r is now known as bluebroth3r (
09:56:48 Nick bluebroth3r is now known as bluebrother (n=dom@rockbox/developer/bluebrother)
09:58:44 Quit linuxstb ("Leaving")
09:59:02 Join linuxstb [0] (n=linuxstb@rockbox/developer/linuxstb)
10:00:36linuxstbDoes anyone understand why libtremor is giving "relocation truncated to fit: R_ARM_PC24 against `.icode'" errors for the Nano 2G? The difference compared to other ARM builds is that the Nano 2G doesn't define ICODE_ATTR_TREMOR_NOT_MDCT in its config-target.h If I add that, it works, but there should be enough room in IRAM for it.
10:00:55 Quit bluebrother (Nick collision from services.)
10:01:03 Join bluebrother [0] (n=dom@rockbox/developer/bluebrother)
10:01:44linuxstb"make V=1" is showing that "-mlong-calls" is being passed to gcc
10:02:15 Join barrywardell [0] (n=barrywar@rockbox/developer/barrywardell)
10:02:17linuxstbAdding that #define to config-ipodnano2g.h makes vorbis compile, but that doesn't seem the right fix...
10:02:39 Join ender` [0] (
10:03:04 Quit mikroflops (Read error: 105 (No buffer space available))
10:04:03n1slinuxstb: isn't that the gcc bug that static functions are always shortcalled even if in a different section?
10:04:47linuxstbI don't know. Is there a workaround?
10:05:27n1sso the workaround is that STATICIRAM thing we use elsewhere which is just a macro defined to nothing for (these?) ARM targets
10:05:55 Quit bluebrother (Nick collision from services.)
10:06:02 Join bluebroth3r [0] (n=dom@rockbox/developer/bluebrother)
10:06:35linuxstbBut yes, that looks to be the problem - the functions in question are static
10:07:04linuxstbBut I wouldn't want to use STATICIRAM for all ARM targets, just those where ICODE_ATTR_TREMOR_NOT_MDCT is not defined...
10:07:57linuxstbSo I guess something like a STATICIRAM_NOT_MDCT
10:09:55 Join merbanan [0] (
10:10:28 Nick bluebroth3r is now known as bluebrother (n=dom@rockbox/developer/bluebrother)
10:10:55 Quit bluebrother (Nick collision from services.)
10:11:03 Join bluebrother [0] (n=dom@rockbox/developer/bluebrother)
10:15:22 Join flydutch [0] (
10:15:54 Quit bluebrother (Nick collision from services.)
10:16:03 Join bluebroth3r [0] (n=dom@rockbox/developer/bluebrother)
10:17:53 Quit bluebroth3r (Client Quit)
10:18:30 Nick Horschti is now known as Horscht (n=Horscht2@xbmc/user/horscht)
10:18:30n1shmm, i wonder if the calendar's memo thing is as broken as it seems or just doesn't work on the sim
10:21:13 Join DerPapst [0] (
10:26:11 Join bmbl [0] (
10:27:13bertrikn1s, this plugin contained a lot of bugs, it wouldn't surprise me if there are still a few left
10:27:25linuxstbn1s: That fixed most of my tremor problems, but I still get the same error on one function (_packetout in framing.c) when I remove "static"
10:28:35linuxstbbertrik: Have you tried current SVN on your Meizu? My changes shouldn't have changed anything for you, but I wanted to make sure before looking at TheSeven's more invasive patches
10:28:47 Join matsl [0] (
10:29:24bertriklinuxstb, svn from friday still worked OK
10:32:51 Quit bmbl ("Bye!")
10:32:52linuxstbbertrik: After r22879 ? (that was at about 1am Saturday morning)
10:33:09bertrikI don't know
10:33:15 Join bmbl [0] (n=Miranda@unaffiliated/bmbl)
10:33:32linuxstbIt would be useful if (sometime today - no immediate rush) you could try current SVN.
10:34:04n1slinuxstb: weird, non static functions should be long called...
10:38:49linuxstbHmm, well putting ogg_stream_packetout and ogg_stream_packetpeek (the two functions that call _packetout) in IRAM as well fixes the problem. That's probably a good thing to do anyway, so I'll just do that...
10:38:56***Saving seen data "./dancer.seen"
10:39:09bertrikhm, I'm having trouble reprogramming the meizu m3
10:39:57 Quit BHSPitLappy (Remote closed the connection)
10:49:26CIA-43New commit by 03dave (r22914): Add STATICIRAM_NOT_MDCT as a workaround for the gcc bug where static functions are not long called. This is currently only an issue on the Nano2G ...
10:50:07linuxstbn1s: Thanks for the pointer with that - all plugins and codecs are now building cleanly.
10:50:36CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
10:50:36*linuxstb likes new targets which share a keypad and LCD size with an earlier target ;)
10:51:05linuxstbbertrik: Is it working yet?
10:51:19bertrikno I'm having trouble writing the bootloader
10:51:34bertrikinto the meizu m3
10:56:10bertrikcurrent SVN works (/me needs coffee)
10:56:20 Quit duonglq ("Leaving.")
10:58:06 Join bluebrother [0] (n=dom@rockbox/developer/bluebrother)
10:58:28linuxstbbertrik: Good, thanks ;)
11:19:14 Quit petur ("real life")
11:30:43n1shmm, next version of gcc gets a link time optimization mode, might be interesting for rockbox
11:38:53 Join J-23_ [0] (
11:39:36 Quit J-23 (Nick collision from services.)
11:39:39 Nick J-23_ is now known as J-23 (
11:45:26 Quit timc (Read error: 60 (Operation timed out))
11:57:11 Join pamaury [0] (
12:00:31 Quit FlynDice (Read error: 104 (Connection reset by peer))
12:00:44 Quit J-23 (
12:00:44 Quit uflops (
12:00:44 Quit YPSY (
12:00:44 Quit Dhraakellian (
12:00:44 Quit polobricolo (
12:00:44 Quit freqmod (
12:00:44 Quit JohnTeddy (
12:00:44 Quit Trista887 (
12:00:52 Join FlynDice [0] (
12:01:10NJoinJ-23 [0] (
12:01:10NJoinuflops [0] (
12:01:10NJoinJohnTeddy [0] (
12:01:10NJoinYPSY [0] (
12:01:10NJoinDhraakellian [0] (
12:01:10NJoinpolobricolo [0] (
12:01:10NJoinfreqmod [0] (
12:01:10NJoinTrista887 [0] (
12:05:24 Quit robin0800 (Remote closed the connection)
12:05:51 Join robin0800 [0] (
12:09:34 Join GeekShadow [0] (n=Antoine@reactos/tester/GeekShadow)
12:15:55 Quit pamaury ("exit(*(int *)0 / 0);")
12:17:13 Quit matsl (Read error: 110 (Connection timed out))
12:22:32 Join dfkt [0] (i=dfkt@unaffiliated/dfkt)
12:31:43 Join robin0800_ [0] (
12:33:30 Join Thundercloud [0] (
12:35:35 Quit BlakeJohnson86 (Read error: 110 (Connection timed out))
12:37:15 Nick fxb__ is now known as fxb (
12:38:47 Nick tha_ is now known as tha (
12:39:00***Saving seen data "./dancer.seen"
12:44:02 Quit Thundercloud (Remote closed the connection)
12:44:04linuxstbbertrik: I've uploaded a new patch to FS #10633 including just the relevant changes (I hope) to crt0.S by TheSeven to get the Nano2G working reliably. These changes are probably what broke the Meizu port...
12:45:37GeekShadowbut meizu port is non functional yet ?
12:46:26linuxstbCorrect - there is no nand/ftl driver.
12:53:08 Quit evilnick ("TTFN")
12:54:22 Quit robin0800 (Read error: 110 (Connection timed out))
12:57:44 Quit ender` (" Please help Conserve Gravity - Play Chess, not Basketball.")
13:02:40linuxstbTheSeven: I've just tested your NAND code for the first time, and also get that strange problem with no names displaying in the plugin menus (but it works when browsing .rockbox/rocks/*/ manually).
13:03:51 Join domo_android [0] (
13:03:58TheSevenlinuxstb: this also happens to the font selection menu, and looks like some weird memory corruption
13:04:21linuxstbMaybe a stack overflow? Have you tried increasing that?
13:04:23TheSeveni.e. if you scroll, sometimes all suddenly have an "l" or some such
13:04:35TheSevenyep, stack is already twice its size to make pictureflow start
13:04:48TheSevenand the old stack size didn't show trouble anywhere else
13:05:47 Quit domo_android (Client Quit)
13:07:14linuxstbIt seems to happen from any of the file browsers, apart from the main one. e.g. "browse themes" also does it, as well as "language"
13:07:38pixelmaa timing issue?
13:07:49pixelma</wild guess>
13:08:36linuxstbWhat kind of timing?
13:10:27TheSevenlinuxstb: how the heck did you manage to get plugins working without commenting out or changing those clashing s5l8700.h 2-letter defines?
13:10:38linuxstbI did change them
13:10:48linuxstbI prefixed all the USB registers with USB_
13:11:23TheSevenhuh, funny. svn doesn't show that here.
13:11:36*TheSeven will check again after resolving all the conflicts
13:11:42 Part janne_oksanen
13:12:02linuxstbI committed that about 9 hours ago
13:12:13TheSevenyep, that's the revision i'm currently updating to
13:12:21TheSevenlinuxstb: btw, in what shape is nano2g audio right now?
13:13:18linuxstbI'm not sure. But we've re-enable the clock init in crt0.S that stopped it working for me in my bootloader tests.
13:13:44pixelmalinuxstb: just thought there could be differences in going through the "browser menus" rather than seeing the file browser directly but as I said that's just a wild guess ny a non-coder
13:13:52TheSeveni'm rather talking about the app side, i.e. still missing apis, or just something not working properly?
13:13:56pixelmaby too
13:15:26linuxstbThere should be nothing to change in apps/. This is (I think) my old patch for audio - I can't recall how much I committed.">
13:19:13pixelmaI don't know, maybe voice (though the plugin browser isn't voiced) if active in menus and not for filenames which is the default. But if you feel that idea is completely useless please ignore, I just wanted to throw something in for brainstorming.. :)
13:21:05 Join FOAD_ [0] (
13:25:24TheSevenlinuxstb: so svn + #5 actually files and results in a working ftl?
13:25:37TheSeven(i.e. i don't need to bother about crt0 for now)
13:28:28linuxstbNo, SVN + my patch at the end of FS #10633 results in working FTL.
13:29:04linuxstbWhich should be equivalent to all your patches, with the exception of the iBugger-related changes which I would prefer to handle independently.
13:31:50 Join kugel [0] (n=kugel@rockbox/developer/kugel)
13:32:12kugelyay, multiple albumart going good
13:33:17 Quit FOAD (Read error: 110 (Connection timed out))
13:33:17 Nick FOAD_ is now known as FOAD (
13:33:45linuxstbTheSeven: Do you know how to do a "deep sleep" on the Nano2G? i.e. the same as that option in diagmode?
13:35:39TheSevenlinuxstb: i know how to shut off power via i2c, but this isn't really great yet, as e.g. the LCD needs to be properly powered down before, or there will be static charges on the cells
13:37:16linuxstbOK. The LCD driver needs some work anyway - I need to implement lcd_update_rect(), plus the yuv blit function (for mpegplayer)
13:37:45linuxstbPlus some asm optimisation to lcd_update could be useful.
13:44:31 Join domonoky [0] (n=Domonoky@rockbox/developer/domonoky)
13:44:43TheSevenlinuxstb: does rockbox support partial updates?
13:45:26linuxstbThat's what lcd_update_rect() does. Currently it just calls lcd_update() (but should be fixed).
13:45:36TheSevenwell, that's really easy to fix
13:45:37linuxstbSo "yes, in general" but "not on the Nano 2G"
13:46:07TheSevenyou just need to pass the rect values to the LCD
13:46:19TheSevenor do you still use a full-screen framebuffer even with that?
13:46:29linuxstbYes, I know. I want to try and get your NAND in first though... (and am currently busy doing non-Rockbox things).
13:47:51linuxstbThe Rockbox higher-level lcd drivers always draw into the full-screen framebuffer (there is only one) in SDRAM. lcd_update_rect does a partial blit of that to the LCD's GRAM.
13:48:51TheSevenwell, this needs some address calculation magic then
13:49:24linuxstbWell, it's all done in other ports - so as you say, it's easy to implement. It's just hasn't been high on my list of priorities so far.
13:52:10TheSevenlinuxstb: is it ok if i resolve that crt0 issue by just ifdefing that whole part of code and keeping the old and new implementations side by side for the different targets?
13:52:28 Join funman [0] (n=fun@rockbox/developer/funman)
13:52:31TheSevenof course, someone should try to re-merge that later, but we first need to pinpoint the meizu problem to do that
13:52:56 Join Thundercloud [0] (
13:56:35*amiconn slaps kugel due to him ignoring an explicit comment
13:56:58kugelwhich one?
13:58:15kugelah I see, which targets are affected? I've never had problems with early USB
13:58:49funmanTheSeven: you can 'reboot' rockbox by using RoLo (open a firmware file from the file browser)
13:59:44kugelI noticed that it needs the screen api, but I remember I made it safe at some point
14:00:19TheSevenfunman: context?
14:00:41funmanTheSeven: yesterday or 2 days ago you were asking how to reboot rockbox (sorry for being late)
14:00:56TheSevenoh yes, but it's something different i needed there
14:01:05kugelalso, there was a reason I moved it later, are you sure you didn't introduce other bugs?
14:01:16TheSeveni needed a way to shut rockbox down properly (i.e. unmount flash), and then boot the OFW to edit it)
14:01:22kugelsuch as showing the statusbar too early at bootup?
14:01:57pixelmakugel: I think he said everything with hardware USB - at least it crashed Rockbox in early USB completely on all Archoses
14:03:23 Join pamaury [0] (
14:04:38linuxstbfunman: You would need to implement rolo first...
14:05:35polobricolois reboot rockbox the same than reset the ipod ?
14:06:04kugelI find it a bit sad that you think I simply ignored it
14:06:25linuxstbTheSeven: Let's see if bertrik can test that patch today and if we can try and resolve it. But yes, that's one option, and probably what we should do in order to get it committed (assuming we can't fix it properly today).
14:07:13linuxstbpolobricolo: Not if you're talking about Rockbox's "rolo" feature. That simply loads a new firmware into RAM and branches to its entry point (i.e. 0x0).
14:16:57TheSevenwhat about this? is this not needed? (you seem to have removed it again from your patch)
14:16:59TheSeven#ifdef USING_STORAGE_CALLBACK
14:17:52linuxstbNo, it's not needed.
14:18:38linuxstb(it was the wrong fix)
14:19:55TheSevenlinuxstb: oh, and another thing in, is it intentional that ENDAUDIOADDR is defined before DRAMSIZE? this looks like a bug to me.
14:20:56linuxstbYes, that looks like a bug.
14:21:36TheSevenshould i include that in my patch, or will you fix that separately?
14:22:06linuxstbI'll fix separately.
14:22:33TheSevenand i'll probably include some unrelated whitespace changes, i.e. strip some trailing tabs(!) from crt0.S meizu code while ifdefing it
14:23:42linuxstbDon't worry about crt0.S - just leave that out of your patch. I would prefer to wait a little while for bertrik to test, than commit code with the intention of "fixing later".
14:24:41 Join bzed_ [0] (
14:24:47TheSevenit's all ifdef'ed now, so it will only affect the nano (besides an obvious bug fix regarding external interrupts which won't ever occur anyways)
14:25:16TheSevenshould i include a power-nano2g.h that makes the "poweroff" button unmount the flash and reboot?
14:25:18 Quit JackWinter (Read error: 104 (Connection reset by peer))
14:25:28TheSevenpower-nano2g.c hack*
14:26:29TheSeventhose system-s5l8700.c int handling changes were not required? (looks like you ditched them)
14:26:39linuxstbNo, if the current power-off doesn't do the unmount, please add that.
14:26:41 Join teru [0] (
14:26:48linuxstbTheSeven: I think I committed those.
14:27:04linuxstb(but don't add a hack to reboot when it should power-off)
14:27:08amiconnkugel: The bug affected early usb on all targets with hardware usb, and on archos it also affected the charging screen
14:27:23amiconnThe usb screen shows the statusbar, and hence uses the multiscreen api
14:27:40TheSevenlinuxstb: i.e. just hang after unmounting?
14:27:57linuxstbTheSeven: Yes, which I think is what it does now (apart from the unmounting).
14:27:59amiconnDon't confuse early usb with bootloader usb, which is handled by the bootloader and cannot be affected by anything in app/main.c
14:28:07 Join JackWinter [0] (
14:28:48TheSevenlinuxstb: those int handling changes aren't there in r22914, and they don't really make sense to me either
14:28:56 Join JdGordon [0] (n=jonno@rockbox/developer/JdGordon)
14:29:34 Join Sajber^ [0] (
14:29:42linuxstbTheSeven: Are you talking about your changes in patch #2?
14:29:57TheSevenno, they are in #4, i think
14:30:10TheSeveni borrowed them from one of your earlier patches
14:30:23TheSeveni'll just drop them and see if it still works
14:30:51linuxstbOnly your patch #2 changed system-s5l8700.c - so it looks like you dropped them from your patch.
14:30:56kugelamiconn: I didn't confuse it with bootloader usb
14:31:11 Quit bzed (Read error: 111 (Connection refused))
14:31:22TheSevenlinuxstb: that must have been by accident then. however i think it proved they are unneccessary.
14:32:09kugelamiconn: this inconsistency between hard- and software usb doesn't mae sense to me
14:32:16 Join BlakeJohnson86 [0] (
14:33:26 Quit bzed_ (Read error: 104 (Connection reset by peer))
14:33:34 Join bzed [0] (
14:33:35 Join bzed_ [0] (
14:33:40TheSevenlinuxstb: should i increase s5l8700 stack size? (only for nano or also for meizu?)
14:33:41 Quit bzed (Client Quit)
14:33:52 Nick bzed_ is now known as bzed (
14:34:00TheSeventhe core seems to run fine with the smaller stack, but some plugins don't like it
14:34:09amiconnIt neither does to me, but that's not my area. Iirc gevaerts deactivated early usb for swusb because the usb stack had problems with that
14:35:02linuxstbTheSeven: Do you know why that's needed?
14:35:35amiconnkugel: I was especially annoyed because this was already the second time I had to fix the same bug. First occurence was long ago in r8030, iirc shortly after the multiscreen api was introduced
14:35:40TheSevenlinuxstb: well, some plugins seem to use more stack than the core, and we increased stack usage quite a lot by having a bigger sector size
14:36:02TheSeventhere are still some sector buffers on the stack (one in the FTL during copying operations and one in FAT)
14:36:03*gevaerts can't remember why he did that now
14:36:42 Quit JackWinter (Read error: 54 (Connection reset by peer))
14:37:58TheSevenok, I'll do 2 patches now. one for the actual FTL without unrelated changes, and one with the crt0.S stuff.
14:38:43 Join JackWinter [0] (
14:39:01polobricolodoes the nano 2g support RTC in rockbox ?
14:39:04***Saving seen data "./dancer.seen"
14:39:28polobricoloi can see a driver for its I2c RTC but my ipod doesn't display the time ?
14:39:36polobricolo(in rockbox)
14:39:43linuxstbTheSeven: So increasing the stack by 2048 byte (for the extra sector buffer in the FTL) plus 1536 bytes for the sector buffer in the FAT driver should be enough?
14:40:04TheSevenshould be.
14:40:13TheSevenhowever, we really have enough iram on that target.
14:40:34linuxstbYes, but no point wasting it.
14:40:46linuxstbi.e. increase the stack by only what's needed, and use the rest elsewhere.
14:40:48TheSevenwell, i wanted to keep stack size a 2**n value
14:41:07linuxstbIs there a reason to? We could increase it by 4KB
14:41:11TheSeveni would go for 0x3000 then, (instead of 0x2c00)
14:43:27linuxstbBTW, I don't think that ENDAUDIOADDR thing is a bug - it just means DRAMSIZE is evaluated later. (look at generated in your build directory - that's the output of after it's run through the pre-processor)
14:44:37linuxstbBut I'll make that stack change after your NAND/FTL code is committed. Were there any plugins apart from pictureflow where you noticed that was needed?
14:46:55kugelamiconn: how early is early usb on hw usb?
14:47:08amiconnHow do you mean that?
14:47:14linuxstbgevaerts: Can I just check you agree with;revision=22906 ?
14:47:15kugelcompared to sw usb
14:47:20amiconnIt happens in main.c, before loading settings etc
14:47:27TheSevenlinuxstb: not in the ones i tried
14:47:31TheSevenmost seemed to work
14:47:56TheSevenbut as stack overflows tend to cause very weird bugs, i'd rather go for safety here. (how does rockbox even catch them?)
14:48:05kugelit seems a bit weird that it works before all initialization finished
14:48:31 Quit funman ("leaving")
14:48:33amiconnIt was mailny introduced for the Ondios as a convenience feature
14:48:38linuxstbTheSeven: If Rockbox keeps running to the next yield(), then the kernel checks that the stack hasn't become full (0xdeadbeef in the last word)
14:49:02 Quit BlakeJohnson86 (Remote closed the connection)
14:49:08 Quit FlynDice (Remote closed the connection)
14:49:11kugelcan you detail it a bit more?
14:49:26TheSevenok, so if it overfills too much, it may still just go mad.
14:49:41amiconnAn MMC can only be switched from MMC mode to SPI mode on the fly, switching back means power cycling. The SH1 CPU accesses the card in SPI mode, the USB bridge in MMC mode
14:50:11amiconnSo once the CPU accessed the card, it cannot be accessed by the bridge anymore without pulling and replugging it
14:50:26amiconn(there is no transistor to switch power by software)
14:51:05amiconnSo if you have an MMC plugged into the Ondio and want to copy something to/from it with your computer, early usb allows that you just plug the Ondio into USB and power it on
14:52:12amiconnThis was the main reason for introducing it, but there are some more advantages. For one, it allows faster usb connection on bootup.
14:52:51amiconnIt also allows usb connection even if a totally messed up file system would crash rockbox when trying to access the disk
14:54:08 Quit JdGordon ("Leaving.")
14:54:40TheSevenlinuxstb: doing a complete rebuild now to test it, and then I'll upload the new patches to FlySpray
14:55:17amiconnThere are two visible differences in the usb screen between early usb and normal usb: (1) all settings are default, hence the usb screen uses those defaults (2) rockbox cannot detect an updated rockbox binary afterwards (needs manual reboot or rolo)
14:58:25kugelwell, I can't remember now why I decided to move it down in the chain, I think there was some bug with the viewport manager. but other fixes in that area may have made it obsolete. Exdending the comment that it mentions the additional problems on HW usb may be useful
14:59:07amiconnHmm, I think the comment says why it's done there and not later?
14:59:17*amiconn thinks this bug might even warrant a 3.4.1
14:59:58amiconnSince it also affects the charging screen, it means that v1 recorders don't charge properly when plugged in from off state
15:00:07kugelamiconn: it doesn't say anything about "why here"
15:00:30kugelobviously I thought it would be safe now as it worked on my sw usb targets
15:00:52amiconnHmm, " before any code uses the multi-screen API "
15:01:06amiconnAnd a few lines down there's the charging screen (in the target init()
15:02:08kugelthat doesn't explain anything
15:02:45*amiconn wonders what is unclear there
15:02:47gevaertslinuxstb: I don't disagree at least :)
15:03:50gevaertsamiconn: one possible problem with early usb on software usb (but probably not why I disabled it) is that due to the ipod 5.5G the MSC code needs a mounted filesystem to know the sector size to use
15:04:40amiconnHmm, couldn't we just use the model id?
15:05:01kugel"before any code uses the multi-screen API" doesn't answer any of my question
15:05:04amiconnIirc both the 30GB and 80GB present 2048 byte sectors over usb
15:06:21amiconn30GB and 80GB G5.5 of course, and I mean the ipod model id in ROM
15:06:59polobricoloi implemented the battery_adc_voltage function for the nano2g. now i have to check if it works what is the best way to discharge my ipod quickly
15:08:21kugelfull backlight and some while(1) somwhere
15:08:50TheSevenlinuxstb: pictureflow still stkov's with 12KB. So there is probably another sector buffer on the stack somewhere that I didn't catch
15:09:24TheSevenI'll upload that patch with 16KB (which works) now, and see where that additional need comes from later
15:12:06TheSevenok, new patch is in FS #10633
15:12:38 Join BlakeJohnson86 [0] (
15:12:59 Quit robin0800_ (Remote closed the connection)
15:16:50linuxstbTheSeven: I need to run now, I'll look at it in a couple of hours time.
15:16:52 Quit linuxstb ("Leaving")
15:16:58TheSevenkugel, polobricolo: and especially opening all clock gates
15:17:07TheSeventhen it'll even discharge with usb connected
15:18:15TheSevenpolobricolo: how did you do it? using the code from my patch #3?
15:22:13 Join robin0800 [0] (
15:25:47polobricoloTheSeven: no using the file
15:26:08TheSeven(which effectively is doing the same thing, just using floating point arithmetic)
15:28:17 Quit stoffel (Remote closed the connection)
15:28:21 Join linuxstb [0] (n=linuxstb@
15:31:49 Quit sinthetek (Read error: 110 (Connection timed out))
15:51:22amiconnHmm, where's Zagor when you need him ;\
15:57:54 Join efyx_ [0] (
15:58:50TheSevenwhat's the general opinion on a FTL doing some splashf's in rare cases like when it needs to remap a block?
15:59:31TheSeventhis is especially interesting duing development while the FTL may still be buggy, as these things should virtually never happen
16:05:54polobricoloTheSeven: maybe make a define (eg FTL_SPLASHF) you undefine once you've finished correcting the bugs
16:06:30TheSevenyes, i would do that anyway, but the question is if this is considered useful at all
16:07:02TheSevenI'll definitely do it in my tree, but the question is if such a thing thing should go into svn eventually
16:07:58 Quit pamaury (Read error: 113 (No route to host))
16:12:03 Join pamaury [0] (
16:16:34CIA-43New commit by 03tomers (r22915): Move definition of mode_name outside #ifdef
16:17:49kugelTheSeven: I think it's ok, but drawing outside the main thread is strictly forbidden
16:18:06kugelthe ftl code will probably run in multiple threads
16:18:15kugelyou could consider logf
16:18:38 Join Ubuntuxer [0] (
16:21:10kugeltomers: what was the point of that commit (aside from introducing many yellows :p)?
16:23:57CIA-43New commit by 03tomers (r22916): Fix yellow
16:26:44tomerskugel: The mode_name was used in #else /* HAVE_LCD_BITMAP */, but was not declared there (seems there is no target with USB HID that don't have HAVE_LCD_BITMAP - I just reviewed the code and noticed it
16:28:15kugelthere's only one target that doesn't have a bitmap lcd
16:28:42kugeland that has a very limited feature set in other regards too (no sw decoding, no sw usb and more)
16:30:46tomerskugel: Do you suggest remove the drawing of USB hid mode for non bitmap lcd targets? The code there is #ifdef hell anyway. I don't mind...
16:31:54kugelwell, I'm constantly suggesting to remove (forking away) the only non-bitmap port anyway, so...
16:33:37kugelbut it doesn't seem useful in this case
16:34:06pixelmaI wonder what happened to JdGordon's idea about those USB displays
16:34:16linuxstbTheSeven: Looking at your latest patch here - - I'm not sure that splashf is the right thing to do if ftl_sync() fails. If that's a fatal error, then a panicf() would seem better (and splashf is "apps" code, and hence shouldn't be used from firmware/ code - think of apps as user-space and firmware as the kernel)
16:34:19pixelmatomers: now the HID mode line fits nicely on the c200 screen :)
16:34:59linuxstbTheSeven: It will probably break the Rockbox bootloader, which doesn't use any apps/ code, but might use reboot/power-off
16:36:46tomerspixelma: Thanks for the report :-)
16:37:24tomerskugel: I will drop USB mode display in non-bitmap target(s). Any objections, anyone?
16:37:35 Join CaptainKwel [0] (
16:37:44 Quit Ubuntuxer ("Leaving.")
16:37:56kugellinuxstb: it uses some apps/ parts: rockbox logo, but that doesn't count as code I assume
16:38:40linuxstbTheSeven: Also, looking at your NAND/FTL code, there seem to be a lot of "uint32_t". Wouldn't just "int" be more appropriate in a lot of those cases? Using "int" tells the reader of the code that the size/sign isn't important, whereas using something like uint32_t implies (to me at least) that an unsigned 32-bit integer is required.
16:39:06***Saving seen data "./dancer.seen"
16:39:39linuxstb(that point won't stop me from committing though...)
16:40:26 Join froggyman_ [0] (
16:40:37 Nick froggyman_ is now known as froggyman (
16:43:51TheSevenlinuxstb: I used uint32_t (or unsigned int) because apple did use unsigned arithmetic all over the place, and there are some places where it really does matter (usually the values are small enough for int/short, but there are some places where they rely on these values being unsigned nevertheless (although this could have been introduced by their compiler))
16:45:12linuxstbYes, I would expect it to matter in some places, but for example, you use uint32_t as the return value for functions which just return 0/1.
16:45:32CIA-43New commit by 03teru (r22917): Fix FS #7511, icon could overlap the line to edit. ...
16:46:06TheSevenlinuxstb: yep, that's mostly because i became accustomed to using it
16:46:25linuxstbBut as I said, it's not that important - we can clean those up later...
16:46:25 Quit janne_ok1anen (Read error: 104 (Connection reset by peer))
16:47:08linuxstbI'm just used to seeing "int" used more often - and you appear to never use it ;)
16:48:09TheSevenis there any way to get user attention from a driver without panicing?
16:49:04linuxstbI don't think so - drivers shouldn't need to...
16:49:32TheSevenwell, they need to if i don't trust them
16:49:47linuxstbI guess the proper way would be for driver functions to return an error, and the calling "apps/" code to check for success, and do something on failure.
16:50:38TheSevenwell, the operation would in fact succeed, but i would like to inform the user that one of the more obscure cases has been hit - if the FTL would report a block remap now, i think it's more likely an FTL bug than a bad flash block
16:50:50kugelwe used/use panics a lot in the in the samsa sd drivers, it isn't much of a problem
16:51:01kugelit surely scales with the buggy-ness of your code :)
16:52:08 Join janne_ok1anen [0] (
16:52:11TheSevenwell, actually i would *want* it to perform the remap operation in that case (which wouldn't happen if it would panic), but i would also like to know when this happens
16:52:36pixelmakugel: samsa just looks like a typo and then can be read wrong
16:53:05TheSevenanother question: will a bootloader ever write anything to flash?
16:53:38TheSevenso it will never need to do a flash unmount in the first place, so it can't fail.
16:54:15kugelwell, unless you want bootloader usb, I guess?
16:54:41TheSevenkugel: I don't think we need this as the norboot already provides flash access via USB
16:55:18TheSeveni would happily go for panicf for failed unmounts, but right now, until there is proper poweroff code, i would somehow like to keep the "Shut down successfully" message, for the user to know that he can now safely reset.
16:55:29TheSevenshoud i do this success message as a panic, too? :-P
16:55:37linuxstbTheSeven: But that requires a reboot. When Rockbox has its own UMS stack, it could be nice to enable it in the bootloader.
16:57:08 Quit ch4os ("Lost terminal")
16:57:12linuxstbTheSeven: That's probably a good idea something like - "panicf("power-off not implemented!")" Although panicf() calls system_reboot(), so we can't do that inside system_reboot() itself...
16:57:42TheSevenbut we can make system_reboot do an actual reboot
16:57:49linuxstbWhich brings me to my question - why do you only check for success inside poweroff(), and not system_reboot() ?
16:57:58linuxstb(success of ftl_sync() )
16:58:25TheSevenbecause i didn't know how to handle exactly that problem, i.e. a reboot triggering a reboot
16:58:37kugelpanicf only does a reboot if you implement system_exception_handler I think
16:59:31linuxstbYes, given the ipods have an easy hardware reset (long press on MENU+SELECT), we can probably just make panicf always freeze.
16:59:47TheSevenok, that would be an option
16:59:54TheSevendo I need any special header for panicf?
17:00:47linuxstbIt's implemented in firmware/panic.c
17:01:16linuxstbAnyway, I think I'm about to commit your NAND driver, but will ignore the power_off and reboot changes for now.
17:01:40 Join ch4os [0] (n=ch4os@gentoo/user/ch4os)
17:01:43 Quit pamaury (Remote closed the connection)
17:02:42CIA-43New commit by 03dave (r22918): Read/write Nano2G FTL/NAND driver by Michael Sparmann (FS #10633).
17:02:43TheSevenwell, just change them accordingly
17:03:14TheSeveni'll quickly make another patch with only them
17:05:16TheSevenis system_reboot supposed to do the actual reboot, or just perform the steps needed to prepare?
17:06:49*linuxstb needs to run - back in 10 minutes...
17:06:52 Quit linuxstb ("Leaving")
17:09:09 Quit teru ("Quit")
17:09:21pixelmawhat could be the cause that running a plugin somehow hinders rebuffering on an hwcodec target?
17:10:27 Join stoffel [0] (
17:10:35pixelmait's very weird, first time I experience that. Playback will just stop after a while (not every rebuffer seems to fail but it won't take too long)
17:11:01kugelTheSeven: both
17:11:40 Join saratoga_ [0] (i=463f90ed@gateway/web/freenode/x-qzqsygftgbzxnvza)
17:12:10 Join ender` [0] (
17:12:19 Quit barrywardell ()
17:15:08 Join jon-kha [0] (
17:16:30kugelThat's the first RW ftl in rockbox, right?
17:20:00saratoga_I think so
17:20:12saratoga_not counting the partially complete one for the D2
17:22:48TheSevenis the D2 one OFW-compatible?
17:23:46saratoga_yes but at present the write support works so poorly its not enabled IIRC
17:25:33saratoga_actually looking at the wiki i'm not even sure if write support is in svn at all
17:29:06liar|netbookTheSeven, linuxstb: i just compiled rockbox from svn but i get a: ata error: 1 on my ipod nano2g
17:29:38 Join chandoo [0] (
17:31:01 Join barrywardell [0] (
17:31:36 Join linuxstb [0] (n=linuxstb@rockbox/developer/linuxstb)
17:32:10kugelsaratoga_: I don't think it ever had write support for the internal nand
17:32:42linuxstbliar|netbook: Wait a few minutes, I've one more important patch from TheSeven to commit.
17:32:48linuxstb(crt0.S changes)
17:33:30TheSevenso they do matter.
17:33:59linuxstbTheSeven: Yes, system_reboot does the actual reboot. See other implementations - e.g. target/arm/system-pp502x.c (the earlier ipods)
17:34:07TheSeven(even though I don't get why? this looks like the cache problem again, but i can't think of a way the meizu cache setup could cause this)
17:34:16 Quit GeekShadow ("The cake is a lie !")
17:34:54*bluebrother curses the stupid path delimiter on windows
17:35:00TheSevenlinuxstb: i have dropped another patch into flyspray with another solution for shutdown/reboot/unmount, also featuring actual reboot code, even though it is still ASM (which could be rewritten as C easily)
17:35:15linuxstbTheSeven: Well, let's wait and see if the crt0.S changes fix things for liar|netbook (although I don't think I have anything else pending that would help...)
17:35:56TheSevenyep, SVN is currently roughly my state minus crt0 and proper unmount
17:36:47TheSeven(and other minor things like the vorbis codec removed)
17:38:38CIA-43New commit by 03dave (r22919): Another part of FS #10633 by Michael Sparmann - correct cache/protection unit init for the Nano 2G. This should be unified with the Meizu code at some ...
17:38:52linuxstbliar|netbook: OK, can you try an "svn update" and rebuild?
17:42:30saratoga_the nano2g does in fact have 256KB of IRAM?
17:43:39linuxstb176KB I thought.
17:44:03saratoga_the datasheet for the Meizu says 256K but it may not be the same
17:44:25linuxstbAh, maybe I made a mistake there then - I thought the Meizus had 128KB...
17:44:44saratoga_and no fast multiplier, but some 24 bit DSP thing
17:44:51TheSevens5l8700 datasheet says 256, nano2g (s5l8701) has 176
17:44:52linuxstbBut the Nano2G has the S5L8701, which has proven slightly different to the S5L8700 in the Meizus.
17:45:01liar|netbooklinuxstb: the same error again
17:45:38linuxstbliar|netbook: Can you try this build - it's one I made earlier today, and worked fine on my Nano -">
17:45:45TheSevensaratoga_: I'm not even sure if the meizus have 256 (bertrik will probably know this), as the DS for s5l8700 is some preliminary one samsung leaked
17:46:21saratoga_well 176KB is probably more then enough for all rockbox codecs
17:46:23TheSevenlinuxstb: could in fact be an edge case with a slightly corrupted FTL that does neither get repaired nor works
17:46:58saratoga_is there a link to the datasheet somewhere? i can't get the rapidshare one to owrk
17:48:23saratoga_i'm curious how the DSP thing in it works
17:50:18 Nick Guest55997 is now known as redfox (
17:50:29TheSevensaratoga: there is a smaller ARM as a coprocessor that seems to have this DSP thing integrated somehow
17:50:43liar|netbooklinuxstb: the same
17:51:01TheSeveni haven't found out yet if the OFW uses it at all, or if it is even present on the 8701
17:51:13saratoga_TheSeven: can it be used as essentially another multiplier (like EMAC on coldfire) or is it really a second processor ?
17:51:21TheSevenliar|netbook: I'll take that over, and propose to move it to #linux4nano
17:51:31TheSeveni'll send you some test files to try through ibugger
17:51:37TheSeveni guess there's something weird with your ftl
17:51:40 Quit robin0800 (Read error: 60 (Operation timed out))
17:51:52linuxstbTheSeven: Discussing it is fine here (but it's your choice...)
17:52:35TheSevenlinuxstb: i suspect it's just some garbage lying around on the flash, already had such a case before where the OFW would mount the flash and my FTL wouldn't for some reason
17:52:50TheSevendebugging that would be quite noisy on IRC though
17:54:11 Join robin0800 [0] (
17:54:53TheSevensaratoga_: "CalmADM2E, a CalmRISC based audio DSP module, is designed for high-quality audio processing. It includes Samsung's 16-bit microprocessor, CalmRISC16E, and 24-bit DSP engine, CalmMAC24E."
17:55:08saratoga_yeah reading it now
17:56:09saratoga_apparently its clocked differently then the main core and has its own cache, so i guess its run as a seperate CPU and not as an extra multiplier for the arm core
17:58:46saratoga_heh it can do hardware ring buffering
17:59:12saratoga_so i guess the idea of this thing is that you map its address space onto some part of DRAM and give it a simple math kernel to grind through
17:59:25saratoga_eventually it signals its done and the main CPU takes over
17:59:33saratoga_sounds like a huge pain to use for codecs
18:00:03linuxstbWhat about running the entire codec on it? Or at least the "decode_frame()" function?
18:00:15saratoga_its only 24 bit so that might be tricky
18:01:15linuxstbWell, first thing would be to find a compiler for it...
18:01:22saratoga_that too
18:01:26saratoga_or even the "CalmMAC24 DSP Coprocessor Architecture Reference Manual"
18:02:26 Join lifeless_ [0] (n=lifeless@
18:03:08saratoga_apparently someone once tried to implement vorbis on it
18:04:06 Quit krazykit ("bbiab")
18:05:56 Quit J-23 (Read error: 104 (Connection reset by peer))
18:06:35 Join mitk [0] (
18:07:05 Part mitk ("Leaving")
18:08:27 Join J-23 [0] (
18:08:35CIA-43New commit by 03dave (r22920): Another patch from FS #10633 by Michael Sparmann (with a couple of minor cosmetic tweaks by me) - cleanly unmount the flash when powering off or ...
18:08:52saratoga_i wonder why they used this stupid DSP core but didn't even bother to give the ARM cpu a fast MAC unit
18:11:19 Join mitk [0] (
18:12:31 Join krazykit [0] (
18:12:36CIA-43New commit by 03tomers (r22921): Remove USB mode displaying in non lcd-bitmap targets, as there are no such target which use USB HID
18:14:08TheSevensaratoga_: probably because they just didn't need it as the DSP is doing all the hard work
18:15:46 Quit Thundercloud (Remote closed the connection)
18:20:45 Quit ender` (" I know you think you understand what you thought I said, but I am not sure that what you heard is not what I meant.")
18:21:20linuxstbTheSeven: Some other screens (apart from the menu items which are file browsers) also seem to have the same problem - e.g. "System -> Debug -> View OS stacks" only displays a scrollbar and no information...
18:21:45*linuxstb wonders if it's a strange Rockbox bug due to lcd_update_rect() doing a full-screen update
18:22:31saratoga_theres an op code chart here for the 16 bit core:
18:24:13*bluebrother finally has an updated rbutil deployment script working as expected
18:25:03kugellinuxstb: possibly. can you try deactivating the statusbar?
18:25:31linuxstbkugel: I'm just trying a sim build with lcd_update_rect == lcd_update
18:25:44 Join esperegu [0] (n=quassel@
18:26:03linuxstbBut yes, I can try disabling the status bar whilst that's compiling...
18:27:24linuxstbHmm, the actual status bar setting doesn't show any options, but if I try to select it blindly, then the settings change. Disabling the statusbar doesn't help though.
18:27:27 Join ender` [0] (
18:27:30 Quit TheSeven (Read error: 145 (Connection timed out))
18:28:04linuxstbThe sim seems to be fine
18:28:20*linuxstb decides to just go ahead and implement lcd_update_rect, and see if that works...
18:28:21kugelseveral parts do set_viewport(NULL); before returning. it may lead to an empty screen if some following code does a clear_viewport(); update_viewport() without setting the proper vp
18:28:42froggymanIf I post a compiled build for lolsack (from the forum) do I have to post the source?
18:28:56kugelI would suspect do_menu and/or list code
18:29:49saratoga_i found a datahseet with a bit more about the 24 bit MAC, and it can do a single cycle 24x24 mul, shift and then 52 bit accumulate
18:30:05saratoga_so you could fake a 32x32 fixped point mul in 2 clocks
18:31:02saratoga_verese 5-6 on the main CPU, so its definately a little faster
18:33:38saratoga_its also got a hardware bit reverse operation for doing ffts
18:35:53saratoga_i guess if someone was really motivated they could reuse my old dual core mp3 code with this by running the synthesis filterbank on the DSP
18:39:10***Saving seen data "./dancer.seen"
18:41:27CIA-43New commit by 03bluebrother (r22922): Add config option to disable ccache during build. ...
18:45:29pixelmaI'd like to help finding out what happens with Rockbox USB on MacOS 10.4 (using my c200). A build that has USB_HID disabled works correctly (I can access the device after it appeared on the desktop), with HID enabled this doesn't happen and on first connect the OS tries to identify a keyboard. Does anybody have any idea what I could look for (and where, I don't know Mac OS too well)
18:46:01CIA-43New commit by 03bluebrother (r22923): Rework rbutil deployment script. ...
18:47:49pixelmasystem profiler (that's called in a German version) still shows a "Rockbox media player" connected on the USB port
18:50:31 Quit saratoga_ ("Page closed")
18:52:13 Quit antil33t (Read error: 104 (Connection reset by peer))
18:52:20 Join antil33t [0] (n=Mudkips@
18:53:17 Join TheSeven [0] (
18:54:27 Join Omlet [0] (
18:54:32 Join Omlet05 [0] (
18:55:12 Quit Omlet05 (Client Quit)
18:55:18 Quit pixelma (Read error: 104 (Connection reset by peer))
18:55:19 Quit amiconn (Read error: 54 (Connection reset by peer))
18:55:25 Quit Omlet (Client Quit)
18:55:36 Join pixelma [0] (i=quassel@rockbox/staff/pixelma)
18:55:38 Join amiconn [0] (i=quassel@rockbox/developer/amiconn)
18:57:22 Join darkham [0] (
19:00:23 Quit amiconn (Remote closed the connection)
19:00:24 Quit pixelma (Remote closed the connection)
19:02:30 Join amiconn [0] (i=quassel@rockbox/developer/amiconn)
19:02:30 Join pixelma [0] (i=quassel@rockbox/staff/pixelma)
19:03:26 Quit kugel (Read error: 110 (Connection timed out))
19:05:04 Nick YPSY is now known as Ypsy (
19:06:23 Join kugel_ [0] (
19:06:31 Nick kugel_ is now known as kugel (
19:08:52 Quit TheSeven (Read error: 104 (Connection reset by peer))
19:09:19 Join TheSeven [0] (
19:11:10 Join BHSPitLappy [0] (n=BHSPitLa@unaffiliated/bhspitmonkey)
19:12:08 Quit robin0800 (Remote closed the connection)
19:24:06linuxstbTheSeven: Can you quickly test"> ? It implements lcd_update_rect(). If the menus still work, then the function works...
19:24:26linuxstb(it works on my lcd type, but needs a quick test on your's). It doesn't fix the menu problem with plugins etc though.
19:24:31domonokypixelma: what does dmesg say when you connect rockbox usb hid to the mac ?
19:24:58linuxstbTheSeven: (or I can post a patch)
19:25:12pixelmadomonoky: just checked: lots of lines saying "AppleUSBEHCI[0x2568800]::Found a transaction past the completion deadline on bus 75, timing out!" which I already found in the system.log
19:25:28kugellinuxstb: maybe there's a threading problem? IIRC there are a few splashes outside the main thread
19:25:56 Nick DaveDavenport is now known as Qt (n=qball@unaffiliated/qball)
19:26:10 Nick Qt is now known as Ma (n=qball@unaffiliated/qball)
19:26:20 Join funman [0] (n=fun@rockbox/developer/funman)
19:26:28linuxstbkugel: The problem is that some lists are not displaying the items or the selection bar. It seems unrelated to any splashes happening.
19:27:03 Nick Ma is now known as DaveDavenport (n=qball@unaffiliated/qball)
19:27:14kugeldoes deactivating the scrollbar and/or changing the line selector change anything?
19:27:43 Quit stoffel (Read error: 113 (No route to host))
19:27:45linuxstbIn such lists, the icons and (if needed) scrollbar are showing. Some examples are the "show OS stacks" screen, and the plugin/themes/wps/font etc file browsers. The same directories display fine when accessed via the normal file browser.
19:29:31linuxstbNo, I don't think those settings help. But those settings themselves are examples that don't display...
19:29:38kugelthe latter use a slightly different filebrowser (a wrapper which does additional stuff) IIRC
19:30:07kugelmaybe still a stack problem?
19:30:29linuxstbPossibly, although I would have thought the kernel should detect it and give a stkov
19:30:50kugelI'm not sure if it the detection works for the main thread
19:30:56linuxstbI'm also running with a 16KB stack.
19:31:19pixelmadomonoky: any other idea what I could check?
19:31:27amiconnThe stkov detection works for all threads
19:31:29TheSevenlinuxstb: your .ipod is running fine for me
19:31:40linuxstbTheSeven: OK, thanks. I'll commit it then.
19:32:01TheSevenand it's *way*faster than the old one
19:32:10TheSeven(especially in some plugins)
19:32:19kugelthat's the point of update_rect() :P
19:32:53CIA-43New commit by 03dave (r22924): Implement lcd_update_rect()
19:33:01TheSevenlinuxstb: have a look at the plasma demo... any idea where that wandering vsync-like line comes from? or is this also occurring on other targets?
19:33:05linuxstbTheSeven: Good, so it's definitely working! I also noticed the speed difference.
19:33:21linuxstbTheSeven: I just tried plasma a few minutes ago and didn't notice that. I'll look again...
19:33:57 Quit efyx_ (Remote closed the connection)
19:34:03amiconnTheSeven: It also happens on other targets. It's just that the lcd update is not (and can not) be synchronised with the internal timing of the lcd controller
19:34:11TheSeveni'll *try* to capture a shot of that issue
19:34:12linuxstbAh yes, it is there, but quite hard to see.
19:34:20domonokypixelma: there is no mention of hid in dmesg ? but it dont have other ideas to check..
19:34:35TheSevenlinuxstb: it's quite noticable on mine, as it's wandering pretty slowly
19:34:55 Quit barrywardell ()
19:35:25funmanAlexP: did you put your e200v2 manual work on flyspray?
19:35:36 Quit faemir (Read error: 131 (Connection reset by peer))
19:35:44amiconnIt's most visible if (by chance) the update frequency almost matches the internal refresh rate of the controller
19:35:47 Join Thundercloud [0] (
19:35:57 Quit froggyman ("ChatZilla 0.9.85 [Firefox 3.5.3/20090824101458]")
19:36:08pixelmadomonoky: no the only other line I can see sometimes reads "AppleUSBEHCI[0x2568800]::DoIOTransfer - error 0xe000404f queueing request"
19:36:47 Join faemir [0] (n=faemir@
19:37:25domonokypixelma: looks like macosx does something different with usb hid then, which rockbox doesnt correctly handle.
19:37:45linuxstbTheSeven: BTW, Rockbox has a "debug" menu - would it be useful to have a "debug NAND" screen displaying various info?
19:38:45pixelmait's "only" 10.4. At least I saw no complaints otherwise - and which is the reason I wanted to help because I have access to a Mac running this version
19:38:48 Join stripwax [0] (
19:41:27mc2739funman: I think AlexP has committed his work on the manual and requested Bagder to place it on the download server. But, Bagder is on vacation, so not sure if that part was accomplished before he left.
19:41:31*bluebrother would like to see someone test FS #10100 but can't provide binaries :(
19:42:07domonokypixelma: some webpage suggest "HID Explorer" to debug usb hid issues on mac. dont know if that will help, if the system fails to communicate with rockbox.
19:44:27linuxstbdomonoky: BTW, I was looking at your ipodpatcher change for ramsize detection, and it means that ipodpatcher itself now gets the XML twice (in the place you added, and in main.c). Plus you didn't put windows-only code (the requirement to remount r/w) within __WIN32_. But don't worry too much - I'll probably clean it up...
19:45:09pixelmaby the way - Rockbox changes into the USB screen
19:45:17domonokylinuxstb: i think bluebrother did this :-) but thanks for fixing anyway :-)
19:45:31linuxstbToo many Dominics...
19:45:32*bluebrother remembers doing that. Oops :)
19:45:44linuxstbOr even Dominiks
19:45:47bluebrotherlinuxstb: just use the nicknames ;-)
19:45:58*linuxstb is sure he saw "domonoky" in the svn log
19:46:32linuxstbNo, I obviously didn't...
19:46:36 Quit funman ("leaving")
19:46:38TheSevenpicture of "vsync" behavior i meant (line at the bottom):
19:48:01pixelmainteresting, I sometimes get a reaction when changing to "Browser mode" for the HID mode and randomly pressing buttons, looks like it is detected as a continous press of "up". When I disconnect then Rokbox hangs in the USB screen and can only be turned off by a hard power off
19:48:25linuxstbbluebrother: The xml reading should probably be done elsewhere (not in either of the places you and I added it). I've a lot of ipodpatcher work to do (to support the Nano2G fully), so will have a think about it when I'm working on that.
19:48:37 Quit mitk ("Leaving")
19:48:47pixelmadomonoky: I'm more concerned that I can't get access to the disk, not the functionality of the HID modes :\
19:49:45bluebrotherlinuxstb: ok. I was thinking about other solutions as well, but that was the easiest to at least use it from within rbutil without too much changes. Hopefully I'll get around looking into the autodetection stuff in a while, as beastpatcher also requires changes over there.
19:49:49domonokypixelma: uh so rockbox-usb is broken completly on mac ?
19:50:02amiconnpixelma: But it does work if you build without hid support, right?
19:50:36pixelmadomonoky: on all (3) 10.4 boxes I tried - amiconn: yes
19:51:12 Join stoffel [0] (
19:51:18 Part toffe82
19:51:20pixelma10.5 is probably more common nowadays, even 10.6 is out for a short while
19:52:12amiconnDo 10.5 (or even 10.6) still support PPC?
19:52:56pixelmaI'm quite sure 10.6 does not, don't know about 10.5
19:53:52domonokyafaic you can build universal binarys on 10.5, so it should support PPC..
19:55:54 Quit darkham ("Sto andando via")
19:59:24TheSevenlinuxstb: where should i start if i want to have a shot at making audio work?
20:00:01linuxstbFirstly at my patch for audio - to check that everything in there that looks important was committed.
20:00:36linuxstbAnd then also at the pcm-s5l8700.c file to see if that code makes sense.
20:01:14TheSevenit does, from a first glance
20:01:56linuxstbMaybe the problem is the same as the one which stopped my bootloader test working if I left that clock-related code in crt0.S
20:02:06TheSevenhow were you doing that audio test? was that using the same apis like playing music, or did you inject the data in a more lowlevel place?
20:02:06linuxstbSo Rockbox just freezes when you try and play audio?
20:02:22TheSevenwell, playback just stays at 0 seconds
20:02:35linuxstbYes, so I guess nothing is being transferred to the DAC
20:02:35TheSevenand no audible reaction on the output
20:02:44linuxstbSame symptoms as I had.
20:03:03TheSeveni just wanted to know if i need to dig into that or if there's just something still lacking that's preventing it from working at all
20:03:11 Join webguest78 [0] (
20:03:25 Join webguest74 [0] (
20:03:36linuxstbI was just using the API that firmware/ provides to apps/ - i.e. the "pcm_play_data()" (I think) function that takes a pointer to the first chunk of audio, and a pointer to a "get_more()" function, that is called in the DMA ISR to get a pointer to the next chunk of data to transfer.
20:04:48linuxstbTheSeven: I don't think anything is missing.
20:05:10TheSevenok, then i'll do another test build without the clock setup and see if that changes something
20:06:06 Join FOAD_ [0] (
20:07:06linuxstbTheSeven: BTW, are you happy for me to close FS #10633 now? As I wrote on a comment there, I think the only things not now committed are the stack change and the iBugger stuff. I don't need a patch for the stack change, and iBugger belongs as a new task.
20:07:52 Quit flydutch ("/* empty */")
20:08:10 Quit webguest78 ("CGI:IRC (Ping timeout)")
20:08:23TheSevenlinuxstb: I don't think the ibugger thing is really relevant any more at all (as we can only debug the bootloader using it), it was just that it was connected pretty closely to the cache fixes
20:09:29linuxstbTheSeven: OK. Could that kind of debugger be integrated into Rockbox? It may be nice to port it to other targets.
20:10:04linuxstbAnd we probably can't debug the bootloader for much longer - I want to change that to run from IRAM (so it can load Rockbox directly to SDRAM).
20:10:14TheSeveni think it deserves a rewrite, and it's hardly portable (at least the device side)
20:10:21TheSevenmost of it is lowlevel USB ASM code
20:10:37TheSevenonce the driver is up and running, the rest is done quickly
20:11:14TheSevenand concerning IRAM/DRAM: That's exactly what this patch was addressing
20:11:20TheSevenif DEBUG, then run drom DRAM, else IRAM
20:12:45linuxstbBTW, do you have any ipod accessories? Rockbox has basic support for those on ipods (remotes etc), which communicate via the UART, so that seems relatively straightforward to port to the Nano2G
20:16:20 Quit stoffel (Read error: 104 (Connection reset by peer))
20:18:35 Join gregzx [0] (
20:19:03 Join Lear [0] (i=chatzill@rockbox/developer/lear)
20:21:10 Quit webguest74 ("CGI:IRC (Ping timeout)")
20:21:17 Quit JackWinter (Remote closed the connection)
20:21:49 Join webguest67 [0] (
20:22:14 Join JackWinter [0] (
20:22:21 Quit FOAD (Read error: 110 (Connection timed out))
20:22:22 Nick FOAD_ is now known as FOAD (
20:26:08TheSevenlinuxstb: nope, i don't have addons
20:26:36TheSevenand i don't have audio either (still just hanging at 0 seconds), even with disabled clock setup
20:26:51 Nick intrados1 is now known as intrados (
20:27:40linuxstbHmm, let me look at my old patch again...
20:30:03 Join biengo [0] (
20:30:25linuxstbOne change I see is to replace the call to wmcodec_write(SAMPCTRL, WM8975_44100HZ); with wmcodec_write(SAMPCTRL, 0); in firmware/drivers/audio/wm8975.c:audiohw_preinit()
20:30:33linuxstb(I can't remember why that was there though...)
20:31:18TheSevenlinuxstb: another point could be that the ofw was initializing it for us, as i think we booted through notes-ibugger back then
20:32:19linuxstbHmm, do we actually implement a handler for the DMA interrupt?
20:32:42linuxstbLooks like my test code wasn't doing that...
20:33:08TheSevenlinuxstb: I'm gonna try my "yippie" sample again through iloader-ibugger, to see if this depends on the ofw setup
20:33:13 Join AsaelReiter [0] (
20:33:40linuxstb(it looks like it was just playing a 1MB chunk in a single DMA transfer)
20:34:34TheSevenshoudln't i at least get *anything* on the output then?
20:34:38TheSevenit just stays silent
20:35:01 Quit esperegu (Read error: 104 (Connection reset by peer))
20:35:04linuxstbIf it's working, you would just get the first 32KB of PCM data - which is likely to be silent for most tracks.
20:35:18TheSevenoops, clickwheel just stopped responding (hold switch still does) after some time on the playback screen
20:35:57 Join esperegu [0] (n=quassel@
20:37:10linuxstbTheSeven: But yes, testing your code in iloader would be useful, to rule out the OF setup.
20:37:27TheSevenit's failing :-/
20:37:36TheSevenlet's install notes-ibugger again
20:39:12***Saving seen data "./dancer.seen"
20:39:35polobricololinuxstb:i have a dock if you want me to try something
20:39:49polobricolo(with my nano2g)
20:40:26linuxstbpolobricolo: I haven't looked at Rockbox's "Apple Accessories" code at all (so can't really help), but if you wanted, it might be fun to try and get that working on your Nano.
20:41:04linuxstbDoes your dock have buttons to control the ipod?
20:41:21linuxstb(although it might be hard to test when Rockbox playback isn't working...)
20:41:30TheSevenskipping will work
20:42:22TheSevenlinuxstb: re-verified: on notes-ibugger it works
20:42:36TheSeveni'll now upload a rockbox build through notes-iloader and check that
20:42:59TheSeveneven though notes-ibugger-iloader would also be a nice (and working!) chain
20:43:42TheSeven"ATA error 1" - shit
20:44:08TheSevenyes, of course the OFW can't unmount cleanly if notes crashes it
20:45:47TheSevenlet's try the trivial first: clock gates
20:46:18polobricololinuxstb: yes. it also has a remote
20:46:46polobricolowhere is it implented in rockbox ?
20:47:17linuxstbpolobricolo: I've no idea...
20:48:14linuxstbapps/iap.c seems to be the main (high-level) code
20:48:50liar|netbookTheSeven: you got that ata error 1 too?
20:49:56TheSevenliar|netbook: yep, when I'm trying to mount a flash that wasn't unmounted cleanly before
20:50:00linuxstbpolobricolo: Seems to also be some lower-level code in firmware/drivers/serial.c (search for IPOD - it seems badly hard-coded to assume all ipods are PortalPlayer ;) )
20:50:29TheSevenlinuxstb: clock gates didn't help
20:51:01TheSevenmaybe some clock thing again, or i2s controller setup?
20:51:32linuxstbpolobricolo: This is the original patch to implement accessory support - scroll to the end to see what was eventually committed, as that patch was a long time in the development.
20:51:32TheSevenor DMA...
20:52:13linuxstbAudio worked on the Meizu, and that was being booted from DFU mode (i.e. no OFW initialisations)
20:53:36saratogaany one have an idea how a codec returning an error can deadlock playback on target?
20:54:03saratogathat MPC SV8 sample is handled correctly by MPC as far as I can tell, just playback doesn't seem to like having the codec return an error
20:56:01 Join robin0800 [0] (
20:56:02 Join GeekShadow [0] (n=Antoine@reactos/tester/GeekShadow)
20:57:04 Quit AsaelReiter ("CGI:IRC")
20:57:32*TheSeven will do an FTL hack that allows mounting unclean FTLs now
20:58:30saratogahuh seems to only happen once and a while
20:58:56 Join GeekShado_ [0] (
21:01:15TheSevenlinuxstb: rockbox playback doesn't work either if i boot it through notes-ibugger
21:04:03biengoAlexP: I've uploaded new versions of the Cassette and Mountain-Fog theme to the e200 theme site, could you remove the old versions, eventually? thanks.
21:04:23TheSevenliar|netbook: can you try /tmp/rockbox.ipod-forcemount.7z and see if that fails, too?
21:07:20 Quit faemir (Read error: 54 (Connection reset by peer))
21:09:03 Quit biengo (Dead socket)
21:09:03 Quit bluebrother (Read error: 104 (Connection reset by peer))
21:09:20 Join biengo [0] (
21:12:07linuxstbTheSeven: Have you looked at the code in dma-s5l8700.c ? Could that conflict in some way with the NAND driver's use of DMA?
21:12:32liar|netbookTheSeven: it fails too
21:12:36TheSevenoh, didn't think of this
21:12:42TheSeveni'll have a look asap
21:12:52TheSevenliar|netbook: then it's not just a dirty mark but something more obscure
21:13:22TheSevenliar|netbook: if you have a build environment at hand, try splashf'ing it
21:13:32 Join bluebrother [0] (n=dom@rockbox/developer/bluebrother)
21:13:47 Join funman [0] (n=fun@rockbox/developer/funman)
21:14:10 Quit GeekShadow (Read error: 113 (No route to host))
21:14:24linuxstbsaratoga: Does that happen on both the sim and a target?
21:14:36liar|netbookTheSeven: splashf'ing?
21:14:51 Join faemir [0] (n=faemir@
21:15:12saratogalinuxstb: it happens on my sansa and on the ipod apparently
21:15:15saratogadidn't try it elsewhere
21:15:24TheSevenadd debug output (or better return codes as linuxstb suggested) and see what's going on
21:16:12TheSevenlinuxstb: when I read the I2S regs, I get all zeroes? :-/
21:16:20CIA-43New commit by 03kkurbjun (r22925): Brickmania - Fix and enable the absolute touchscreen mode.
21:16:39linuxstbTheSeven: Hmm, they're write-only?
21:16:56linuxstbWhen are you reading them?
21:16:57TheSevenwouldn't expect that
21:17:00polobricolowould it be possible to disable the automatic shutdown on the nano2g as it does not shutdown (causes apanic)
21:17:01stripwaxDoes definiing MAX_PHYS_SECTOR_SIZE give worse disk performance than not defining it? Thinking that all ipod videos have this define but only 80GB 5.5G needs it
21:17:16TheSeveni just tried to do some reg dumps from iloader-ibugger and notes-ibugger and compare
21:17:22linuxstbstripwax: Well, we don't have a separate build for 80GB ipods...
21:17:30linuxstb(only 32MB/64MB)
21:17:30stripwaxand I'm wondering why not .. :)
21:18:00linuxstbWouldn't that give us four builds for the video?
21:18:07stripwaxOnly three.
21:18:28linuxstbWhy? Which combination can't exist?
21:18:30TheSeventhere is no 80gb 32mb one
21:18:49stripwaxunless we decide to provide a build for use with 32MB ipods that have an upgraded disk
21:19:05linuxstbThere shouldn't be, but refurbished 80GBs sometimes have a 32MB main board (IIRC)
21:19:12stripwax... yuk
21:19:26 Quit BHSPitLappy (Remote closed the connection)
21:19:39linuxstbThat what causes the problems with people trying a 64MB build on 80GB ipods - it turns out they only have 32MB RAM.
21:19:44 Join JackWinter2 [0] (
21:20:11stripwaxthey should take them back and demand a replacement replacement :)
21:20:13linuxstbAlthough ipodpatcher can detect both sector size and RAM size now, so in theory rbutil can support detection. (although no RAM size detection on OS X yet).
21:20:15 Join sig11 [0] (
21:21:37TheSevenlinuxstb: I'm using DMA channel 3 for NAND. which one does i2s use?
21:21:41*domonoky would prever if this detection is done in rockbox itself. only one ipod video build would be nice, and less confusing for users :-)
21:23:38gevaertsDoes the 5.5G OF and EDM *always* present 2k sectors over USB, even with other disks?
21:24:37*n1s agrees with domonoky
21:25:08pixelmadomonoky: the plan is there sinde DevCon 2007...
21:25:16pixelmasince too
21:26:12gevaertsif it does, basing sector size on model id would have huge advantages : no more guessing in disk.c, so no more bootloaders getting confused by remnants of 512-byte MBRs, no more guessing in usb_storage.c
21:26:16 Join matsl [0] (
21:27:13domonokypixelma: plans dont count.. 2007 is long ago. :-)
21:27:18 Quit esperegu (Read error: 104 (Connection reset by peer))
21:27:41saratogaif no one minds I'd like to disable prebuffering of *.codec files on low memory targets
21:27:56saratogait seems to improve stability and buffering them on low memory targets is completely broken anyway
21:28:21pixelmadomonoky: that was my point (that it's long ago)
21:32:20 Quit JackWinter (Read error: 110 (Connection timed out))
21:32:33 Join Grahack [0] (
21:33:23 Join bughunter2 [0] (n=bughunte@unaffiliated/bughunter2)
21:34:05 Quit Sajber^ (Read error: 131 (Connection reset by peer))
21:36:17 Quit GeekShado_ (Read error: 104 (Connection reset by peer))
21:36:32 Quit biengo (Remote closed the connection)
21:36:40 Join GeekShado_ [0] (
21:36:48 Quit matsl ("Leaving")
21:36:57pixelmabtw. I found one box running MacOS 10.5.8 and tried with my c200 again - it also wanted to identify a keyboard first which I then interrupted, it mounted the two disks in the end but it took quite a while (had trouble ejecting too)
21:37:03 Join matsl [0] (
21:38:26pixelmaand while the HID functions also seemed to work it didn't work as nicely as it does on my XP laptop. The keys "stick" sometimes and in mouse mode there is always a jump into a 90° different direction in the end
21:41:42 Quit Lear ("ChatZilla 0.9.85 [Firefox 3.5.3/20090824101458]")
21:43:27 Join benime [0] (
21:44:53 Quit janne_ok1anen ("leaving")
21:45:36 Join skyhunter [0] (
21:48:16 Join janne_oksanen [0] (
21:48:32polobricoloi have to go i'll have a look at iap tomorrow
21:50:05 Quit robin0800 (Remote closed the connection)
21:50:20 Join robin0800 [0] (
21:51:14 Join MethoS- [0] (n=clemens@
21:51:35 Quit robin0800 (Remote closed the connection)
21:51:40 Join mt2 [0] (n=mtee@
21:58:46 Join pamaury [0] (
22:00:31 Join barrywardell [0] (
22:00:37 Quit skyhunter (Read error: 104 (Connection reset by peer))
22:01:02 Join GeekShadow [0] (n=Antoine@reactos/tester/GeekShadow)
22:01:30 Quit GeekShadow (Read error: 104 (Connection reset by peer))
22:04:18 Quit Grahack ("Leaving.")
22:05:22mc2739pixelma: ping
22:07:03mc2739pixelma: do you know what is needed to fix this error while building the manual? −− ! LaTeX Error: File `xcolor.sty' not found.
22:08:23funmanmc2739: you must install the tex package providing xcolor (it's in the description of ubuntu packages)
22:08:26pixelmaif I understand correctly, you are missing the xcolor package (it was included by default in my LaTeX packages so can't tell how it is called exactly)
22:08:29gevaertsyou need to install the package containing xcolor.sty. On debian-based systems, it's probably latex-xcolor
22:09:06mc2739ok, thanks
22:09:23pixelmamc2739: which packages did you install?
22:09:40mc2739Is this something new? I'm sure I could build manuals before, although it has been a while
22:09:50linuxstbTheSeven: (sorry, I was away from my computer). Looking at target/arm/s5l8700/dma-target.h, I2S is using channel 0.
22:10:41mc2739I installed using the wiki directions, except texlive was installed
22:12:07domonokymc2739: yes the use of xcolor in the manual is pretty new.
22:12:29mc2739found it - it is latex-xcolor
22:13:21linuxstbsaratoga: Any idea if disabling pre-buffering of codecs would break gapless (i.e. by introducing a delay when reading the codec binary from disk) ? But I guess that's not important, as inter-codec gapless isn't that useful...
22:14:40pixelmamc2739: yes, it's new (well quite) and is used to colour the tables a bit to make them better readable
22:15:12saratogalinuxstb: i'll do some testing but i doubt it given the length of the audio buffer
22:15:45mc2739thanks to all - manual build is working now
22:15:48linuxstbHow big is that? The same 512KB as large-mem targets?
22:17:54 Quit GeekShado_ (Read error: 113 (No route to host))
22:19:00saratogalinuxstb: 176KB
22:19:08pixelmabarrywardell: are you really around (and still using a Mac)?
22:19:25barrywardellpixelma: yes, and yes ;)
22:19:55funman176kB is the size of pcm buffer
22:19:58linuxstbsaratoga: Hmm, only 1 second. But I guess that's long enough to load a codec and start decoding...
22:20:18saratogalinuxstb: codec files are less then 100KB which should take very little time to load on a flash target
22:20:24saratogabut of course that needs testing
22:20:27TheSevenlinuxstb: I think I'm close.
22:20:45TheSevenit's clocking again
22:20:46pixelmabarrywardell: you have an e200, right.. how's Roclbox USB working for you (and which OSX version exactly)?
22:20:58pixelmaRockbox too...
22:21:22barrywardellpixelma: yeah, I have an e200 and currently have OSX 10.5.8
22:21:42linuxstbTheSeven: Hurray! ;) When the time comes - (I think proper audio playback inside Rockbox itself deserves a second email)
22:21:48TheSevenlinuxstb: I managed to make the DMA transfer work (and the I2S status bits behave like expected), but still no audio... power to the WMCODEC shut off?
22:22:10barrywardellpixelma: usb was working ok last time I tried, but will check again now
22:22:22funmanTheSeven: or wrong audio output selected
22:23:09linuxstbTheSeven: So is Rockbox showing an elapsed time other than zero?
22:24:14TheSevenlinuxstb: didn't play around with rockbox yet, but made DMACTCNT0 run down in the audiotest sample
22:24:20pixelmabarrywardell: I'm asking (don't know if you read logs) because I have trouble with it on MacOS 10.4.11 (and although it worked on the one 10.5.8 box, it seems to work better on my XP laptop for me)
22:25:04linuxstbTheSeven: Ah, OK. As funman said, it could be that you haven't configured the WM8975's mixer correctly.
22:25:08TheSevenCLKCON &= ~0x80; does the trick, still need to check what this is in the first place
22:25:47TheSevenhah. MCLK_MASK
22:25:52TheSevenwe should have known that...
22:26:01TheSeven"Mask audio clock into SPDIF or IIS block"
22:26:14linuxstb;) Yes, that sounds useful...
22:26:30TheSevenshould I just unmask that in crt0.S?
22:26:49TheSeven(this is probably what screwed audio as soon as the clock stuff was enabled)
22:27:03linuxstbProbably somewhere in the audio init, let me look...
22:27:05kugelaudio on nanog2?
22:27:34TheSevenonly after running through the OFW once, but we're close to getting it working inside rockbox
22:27:35linuxstbkugel: We had that a while ago in test apps, but now TheSeven is debugging why it's not working in Rockbox itself.
22:27:48kugelah alright
22:27:52gevaertsWhat's the status of the manual and rbutil integration? ;)
22:28:13linuxstb"cp manual-ipodnano.pdf manual-ipodnano2g.pdf" - done!
22:28:24pixelmascreenshots and button tables etc. for the manual will be easy
22:28:50linuxstbI still have a little more work on ipodpatcher, but when that's done, the install will be identical to earlier ipods.
22:28:52pixelmamaybe some other installation instruction could be usefull
22:29:01linuxstb(from a user's point of view)
22:29:05*pixelma takes that back
22:29:52TheSevenlinuxstb: we're *explicitly* masking MCLK in the clock setup
22:29:54 Quit gevaerts (Nick collision from services.)
22:30:03 Join gevaerts [0] (n=fg@rockbox/developer/gevaerts)
22:30:09barrywardellpixelma: current svn in "multimedia mode" seems to work ok for me
22:31:57linuxstbTheSeven: pcm_play_dma_init() in pcm-s5l8700.c seems the right place. Unless you think there's no harm in just enabling it in crt0.S
22:32:09barrywardellpixelma: copying a 700mb file is quite fast (~9MB/s)
22:32:12TheSevenit is - but it didn't help either
22:32:47barrywardellpixelma: but pushing buttons while copying seems to make the copying freeze
22:33:06linuxstbTheSeven: Does your test app do the same inits as crt0.S ?
22:33:19pixelmabarrywardell: does it always take such a long time to mount (appear on the desktop), thought it takes longer than on XP but that was the first connect
22:33:36TheSevenlinuxstb: it doesn't do any
22:33:49 Join GeekShadow [0] (n=Antoine@reactos/tester/GeekShadow)
22:33:51TheSevenbut it's not only the mask, MCLK gets screwed in some further ways
22:34:03barrywardellpixelma: no, it was quite quick to mount.
22:34:18TheSevenand we're using PLL2 on the nano, not sure if the meizus even have that
22:34:37linuxstbI thought I changed that to PLL1?
22:34:54linuxstbI remember testing it first with PLL2, and then when it was working, trying PLL1 and that worked too.
22:36:09TheSevenwhere is that? I can't see this getting set up at all.
22:36:13pixelmabarrywardell: I tried changing the volume in Multimedia Mode with my c200 and one time a simple button press was like continuously pressing the button, maybe it's a tad bit different with the scroll wheel
22:36:46linuxstbTheSeven: pcm_dma_apply_settings() in pcm-s5l8700.c
22:36:58barrywardellpixelma: volume control works great on the e200 with the scroll wheel
22:37:19TheSevenlinuxstb: oh yes, you're setting up the PLL itself, but where do you set which PLL to use?
22:37:24barrywardellpixelma: the only problem i have encountered so far is trying to change volume while copying files
22:37:55pixelmabut more importantly - on 10.4 the disks wouldn't mount at all with USB HID enabled. Do you have any idea what to check for etc.? Although I can only try again some time tomorrow...
22:38:59barrywardellpixelma: usb prober can be quite useful
22:38:59TheSevenlinuxstb: hmm, looks like you're already setting it up correctly further down
22:39:09gevaertspixelma, barrywardell: I suspect that we still have some basic driver issues that cause some lost usb packets every now and then. Error detection and handling should be solid enough to not have them cause data loss, but they do cause these bus resets (which appear to the user as freezes), which could explain occasional gliches
22:39:14***Saving seen data "./dancer.seen"
22:39:54barrywardellgevaerts: ok, although for me, the copy operation didn't ever recover from the freeze. It got stuck and I had to pull the usb cable
22:40:15pixelmabarrywardell: is that on board in a usual installation or an external program... though I guess I could also google
22:41:07barrywardellpixelma: I think it's included when you install XCode
22:41:11TheSevenlinuxstb: what's the best way to check if DMACTCNT0 is running in rockbox?
22:41:20 Join BHSPitLappy [0] (n=BHSPitLa@unaffiliated/bhspitmonkey)
22:41:33barrywardellpixelma: in /Developer/Applications/Utilities
22:42:05pixelmabarrywardell: have to check tomorrow if it's installed, this is a work Mac
22:43:02linuxstbTheSeven: Probably by writing a debug screen - see the hell in apps/debug_menu.c for that code (the System -> Debug menu)
22:43:29pixelmabarrywardell: thanks for the hint so far. Are you around a bit more often again the next days?
22:43:37amiconnmc2739, pixelma, gevaerts: can search package contents if you want to know what package provides a certain file
22:43:54TheSevenlinuxstb: well, how does the player app determine if it's running?
22:44:09barrywardellpixelma: yes, I'll be online during the day (CET) most of this week, so I'll login here too
22:44:26linuxstbTheSeven: I'm not sure what you mean. There is no "player app" as such.
22:44:28TheSevenreturn DMATCNT0 * 2; in pcm_get_bytes_waiting() looks like a bug to me, shouldn't this be return DMACTCNT0 * 2;?
22:44:47TheSevenlinuxstb: I mean the usual player screen if you open a music file
22:45:06 Nick uflops is now known as mikroflops (
22:45:23linuxstbThere's already a "debug audio" screen - maybe that helps.
22:45:50TheSevenwhere is this?
22:45:54pixelmabarrywardell: nice :) I really want to help out debugging this as I seem to be the only one around here who was affected but don't know much about MacOSX, so maybe get back to you then ... :)
22:46:06amiconnlinuxstb: Why would you want to disable codec buffering?
22:46:09mc2739amiconn: That is good to know information!
22:46:12linuxstbTheSeven: Start playback, then go to the main menu and System -> Debug
22:46:24TheSevenyou mean that buffering thread screen?
22:46:25linuxstbamiconn: Ask saratoga that question - it was his suggestion.
22:46:54barrywardellpixelma: for reference, here's the output I get:
22:46:57*amiconn prefers actually fixing bugs over working around them
22:47:02linuxstbTheSeven: I thought there was another... But to be honest it's been a while since I looked. It should show you the amount of bytes in the PCM buffer (which is what is DMA'd to the DAC).
22:47:31TheSevenis this the return value of pcm_get_bytes_waiting(void)?
22:47:50linuxstbSounds likely.
22:49:36 Quit GeekShadow (Read error: 54 (Connection reset by peer))
22:49:38 Join Grahack [0] (
22:49:56pixelmabarrywardell: great, thanks. By the way, I got the "Rockbox media player" at an USB port in the system information but then nothing (and also quite rarely some reactions on the input in some HID mode)... does that tell you something
22:49:58 Join GeekShadow [0] (n=Antoine@reactos/tester/GeekShadow)
22:50:20pixelmaoops, sorry
22:51:13barrywardellpixelma: I'm not too sure. USB prober should tell you quite a bit more, especially if you can install the debug usb driver for logging
22:53:25 Quit Grahack (Client Quit)
22:53:27saratogaamiconn: besides the fact that it works poorly, i'm not sure what sense it makes to buffer codecs when the buffer is so small
22:53:39pixelmabarrywardell: ok, that has to wait a bit then. I will try during the next days, hope I'll be able to use it correctly. Many thanks for your help so far :)
22:53:40amiconnLess special casing
22:55:54 Quit funman ("free(random());")
22:56:16 Quit Thundercloud (Read error: 104 (Connection reset by peer))
22:56:20saratogabut yes I would like to figure out why it doesn't work correctly
22:57:01 Join Thundercloud [0] (
22:57:22saratogai observed erratic behavior (rebuffering codecs) at 8MB of RAM, though I didn't test larger
22:57:48saratogai suspect that the checks it uses to decide if it can reuse teh current codec no longer apply
22:58:00saratogasvn shows they're nearly as old as SWCODEC itself
23:02:38 Join DataGhost [0] (n=dataghos@unaffiliated/dataghost)
23:05:25linuxstbTheSeven: I guess you've found it, but the buffering debug screen is dbg_buffering_thread() in debug_menu.c Does that show anything interesting?
23:05:49TheSevenyep, that DMA still isn't running :-/
23:06:23TheSevenanother question: did that ever work? the GPIO setup was wrong.
23:06:35linuxstbSo the PCM buffer fills up, but is never emptied?
23:07:09linuxstbDMA should have been working - otherwise I wouldn't have got sound. But I guess I just relied on the OF's setup if my code was wrong...
23:07:13 Quit merbanan (Read error: 110 (Connection timed out))
23:07:36TheSevenwell, you set the mode of the pins to 2 instead of 7.
23:08:01TheSeventhe OF set them to 2, and then to 7 immediately before starting playback
23:11:53CIA-43New commit by 03kkurbjun (r22926): Brickmania: Fix the flip-sides powerup with touchscreen and make the whole gameboard available to position the paddle
23:12:12 Join domonoky1 [0] (
23:15:24CIA-43New commit by 03kkurbjun (r22927): Brickmania: Add new, finer detail images for the bricks and the ball for 640x480 screens.
23:15:58*gevaerts thinks that any code that provides a better bricking experience is good :)
23:16:37 Join GeekShado_ [0] (
23:17:58TheSevenlinuxstb: wrong MCLK speed won't hurt - it will just speed up or slow down the audio
23:18:01CIA-43New commit by 03bluebrother (r22928): Add command line option for adding additional files to the build folder.
23:23:04 Join GeekShad__ [0] (
23:24:13 Quit GeekShad__ (Connection reset by peer)
23:24:32*kugel wonders if kkurbjun can get in touch with chopper after finishing brickmania. It could be so much smoother
23:24:35 Join GeekShad__ [0] (
23:24:59*kkurbjun thinks chopper is named that for a reason :)
23:25:27 Join Omlet [0] (
23:25:33 Join Omlet05 [0] (
23:25:36 Quit Omlet05 (Remote closed the connection)
23:27:07*bluebrother wonders if it's feasible to merge the OSX deploy script with
23:28:04 Quit domonoky (Read error: 110 (Connection timed out))
23:31:02*TheSeven suspects that something with DMA is just broken
23:33:13 Quit Omlet ("( :: NoNameScript 4.22 :: )")
23:34:54 Quit GeekShadow (Connection timed out)
23:35:44 Quit chandoo (Read error: 110 (Connection timed out))
23:39:28 Join JdGordon [0] (n=jonno@rockbox/developer/JdGordon)
23:40:27*TheSeven was right
23:40:50TheSevenlinuxstb: "audiotest" works now also when booted through iloader-ibugger
23:41:15 Quit GeekShado_ (Read error: 110 (Connection timed out))
23:41:17linuxstbCool. So now for Rockbox...
23:41:18 Nick Ypsy is now known as YPSY (
23:42:34*TheSeven just can't spot a difference
23:42:53TheSeveni've fixed the target of the DMA data, but that didn't help
23:42:59TheSeventhe peripheral ID was just wrong
23:43:11TheSeven(probably right according to the DS, but apple clearly uses a different one
23:43:26*bertrik remembers an ambiguous DMA table from the ds
23:43:40*linuxstb wonders how this worked at all in his bootloader tests...
23:43:48*TheSeven too
23:43:57linuxstbHonestly, I did hear audio!
23:44:41*TheSeven trusts linuxstb, but doesn't trust the DMA any more :-)
23:45:35*TheSeven spotted another wrong define
23:47:17*TheSeven can't calculate bit positions today
23:49:11*TheSeven notices that not setting up the wmcodec properly will stall the DMA
23:51:01*TheSeven doesn't really understand how all this rockbox audio setup code works
23:51:38bertrikhm, I thought the IIS port on the s5l8700 was IIS master and the codec IIS slave, which should make it basically impossible for the wmcodec to stall audio
23:51:59TheSevenbertrik: on the meizu, yes, on the nano, no
23:52:03TheSeventhe wmcodec is a master
23:55:55CIA-43New commit by 03tomers (r22929): Have codepage name show in a comment
23:55:59 Quit jordan`` (Read error: 104 (Connection reset by peer))
23:57:53TheSevenlinuxstb: any chance the wmcodec setup is at fault?
23:58:26linuxstbTheSeven: Anything is possible... Did you make that change I made in my code?
23:58:33TheSevenwhich one?
23:58:47linuxstbThe one I mentioned earlier - in the wmcodec init
23:58:48 Join jordan` [0] (i=gromit@

Previous day | Next day