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 2011-02-19

00:00:03kkurbjunyes, it definitely is on microblaze for sure.
00:00:06Tornethe advantage of actual PIC code is you cna share pages of libraries between processes even if they're mapped at different addresses, and can mmap them directly from disk
00:00:18Torneneither of which are relevant for us
00:00:24Tornesince we're single-process without mmap :)
00:00:41Tornerelocatable code is the opposite of PIC code in most ways :)
00:01:36Tornesince we need to load plugins in full from disk anyway, the effort to then relocate them once they're already fully in ram is comparitively tiny
00:01:51 Join hebz0rl [0] (
00:02:01Torneand then the resulting binaries will be exactly as efficient as if they were fully relocated by the linker, just slightly larger on disk for the relocation table (which can be discarded after load)
00:02:43 Quit hebz0rl (Read error: Connection reset by peer)
00:03:02Torneit would be kinda sucky if we had to build, say, mpegplayer as PIC and lose one or two registers there :)
00:15:02 Quit {-phoenix-} (Remote host closed the connection)
00:17:09CIA-70New commit by kugel (r29328): Fix yellows and checkwps.
00:22:15CIA-70r29328 build result: 905 errors, 124 warnings (kugel committed)
00:25:16kugelfirmware including apps header :\
00:26:03 Join Judas_PhD [0] (
00:28:39pamaurywhere ?
00:29:15CIA-70New commit by kugel (r29329): Fix red caused by incorrect includes.
00:30:56 Join dfkt_ [0] (dfkt@unaffiliated/dfkt)
00:31:29 Join panni__ [0] (
00:31:52 Quit casainho (*.net *.split)
00:31:52 Quit maxsin (*.net *.split)
00:31:53 Quit Jerom1 (*.net *.split)
00:31:53 Quit panni_ (*.net *.split)
00:31:53 Quit Kitar|st (*.net *.split)
00:31:53 Quit dfkt (*.net *.split)
00:31:53 Quit mudd1 (*.net *.split)
00:31:53 Quit JdGordon| (*.net *.split)
00:31:53 Quit Battousai (*.net *.split)
00:31:53 Quit TheSeven (*.net *.split)
00:32:07 Join Battousai [0] (~bryan@gentoo/developer/battousai)
00:32:13 Join mudd1 [0] (
00:32:28pamaurykugel: native cooperative threading is really nice :)
00:32:32 Join casainho [0] (~chatzilla@
00:33:06 Join JdGordon| [0] (
00:33:07 Quit JdGordon| (Changing host)
00:33:07 Join JdGordon| [0] (~jonno@rockbox/developer/JdGordon)
00:33:19 Join maxsin [0] (
00:33:25CIA-70r29329 build result: 656 errors, 0 warnings (kugel committed)
00:33:43 Join Jerom [0] (~jerome@
00:33:53pamaurygetting better ;)
00:34:26kugelmeh, I checked checkwps
00:34:38Strife89Probably not high score material this time
00:34:42Strife89You hope. ;)
00:34:54kugelah, h300 checkwps is fine but others aren
00:35:00 Quit factor_ (Quit: Leaving)
00:35:17 Join factor [0] (~factor@
00:36:22Strife89Score is 25,725 right now. ;)
00:37:00 Quit casainho (Client Quit)
00:37:30 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
00:38:14 Join Kitar|st [0] (
00:38:18 Join vnl [0] (
00:38:20gevaertsStrife89: scores don't add up
00:40:05kugelhmm, why does that error now? or rather how did it not error before?
00:41:26 Quit kevku (Quit: KVIrc 4.0.2 Insomnia
00:41:52 Join BHSPitMonkey [0] (~stephen@unaffiliated/bhspitmonkey)
00:43:15kugelat least I keep introducing new reds instead of overlooking :)
00:49:58kugelo noh, this was overlooking
00:50:21*TheSeven wants FS #11917 to be fixed, but can't make it work :/
00:52:20TheSevenTorne: regarding PIC vs. relocatable: you might want to have a look at how i implemented it in emCORE
00:52:42TheSeveni hacked up elf2flt quite a bit and wrote a loader for that
00:52:57Tornewhat did you need to hack it for?
00:53:00kugelthis error is strange
00:53:05user890104TheSeven: i can provide more information on the issue i reported, if needed
00:53:35 Quit maxsin (Quit: Leaving.)
00:54:02TheSevenTorne: some different requirements I had, like UCL compression instead of gzip, .text/.data/.bss being contiguous, some more header fields/flags, ...
00:54:24TheSevenand most importantly a different kind of interaction with the linker, as the old one was cocking up with eabi objects
00:54:26 Join maxsin [0] (
00:54:53TheSeveni let the linker do the linking now, and elf2emcoreapp just generates the relocation data and header
00:55:39TheSeveni didn't manage to make it relocate non-fully-linked eabi objects, and also unresolved relocations went unnoticed the way it was implemented... kinda strange
00:55:59TheSeveni now just pass −−emit-relocs to ld, and transform the relocation information into the FLT format
00:56:12kugelHAVE_CORELOCK_OBJECT is still defined but for some reason it's not seen in thread.h anymore
00:56:42TheSeven(and commented out half of elf2flt, as it doesn't need to handle all that complex symbol resolving stuff)
00:57:03Tornewell, that sounds useful :)
00:58:00TheSevenalso i reworked the build system, as it was a rather big mess to even compile that thing in its original state and had some nice bugs
00:58:30kugelah, I see it now
00:58:49thomasjfoxkugel: There's a #ifdef CPU_PP before it
00:59:05TheSevenwhat i still don't fully understand is what the matter with that bfd.h file is
00:59:10kugelthe corelock definition is now within #ifdef ASSEMBLER_THREADS
00:59:28TheSevendo i really need to take that from the very same binutils build? and if so, why isn't it copied into some system include dir on installation?
00:59:48TheSevenit's kinda nasty that throws it away after compiling, while elf2flt needs it
01:00:37Torneelf2flt is part fo the extended gcc suite, no?
01:00:46Torneso the "natural" way is to unpack it into the same place
01:00:50Torneand then everything is there :)
01:01:00Tornelibbfd isn't really meant to be used outside gcc
01:01:01 Quit maxsin (Ping timeout: 240 seconds)
01:01:12Tornedoing it that way has never ever worked for me though
01:01:15Tornewith anything :)
01:01:36TheSevendoing it which way?
01:01:43 Part toffe82
01:01:56Tornehaving binutils, gcc, libc, etc all sharing a single source dir
01:02:07Torneso that a single configure/make/make install does the lot
01:02:08TheSevenuser890104: I know what's going on and why it happens, but if I fix it, USB breaks completely for an unknown reason
01:02:34Torneelf2flt is also derived from gcc and friends and thus is able to work the same way
01:02:52Torneat which point it wouldn't need any of bfd's headers to be installed because it would get them from the source tree
01:03:34TheSevenhow do other tools that need to access .elf/.o files without having access to the bfd headers during compilation do it?
01:03:50Tornethey stay as far away from libbfd as they possibly can, generally
01:03:51TheSeveni can't think of an example of such a tool right now, but i found one some weeks ago
01:03:54Tornebecaue it's an insane nightmare world
01:04:11Tornepeople tend to use elflib or whatver it's called
01:04:21Tornebecause its API isn't so terrifyingly and unnecessarily generic
01:04:50Torneit's slightly frightening that elf2flt uses bfd :)
01:05:21TheSevenwell, at least I made it work if I manage to steal bfd.h and libbfd.h quick enough during binutils compilation
01:05:41Torneright. i have no idea how the "right" way is
01:05:44pamauryperhaps elf2flt plan to work on something else than elf ;)
01:05:51Torneother than to suspect that they might expect you to do it with a unified sourcetree
01:06:13Tornethe docs for all that are atrocious
01:06:13TheSevenwell, their configure wants a −−with-binutils-build-dir=/whatever
01:06:28Torneright, but toplevel configure will probably provide that if it's unified
01:06:45Tornesince with a unified build everything gets built without anything needing to be installed
01:06:53Torneand so the components need to know where each other are
01:07:40Tornethe instructions i found for this process involve steps like "untar x over the top of y, and, er, assume that whichever archive was released most recently is okay to overwrite files from the other one"
01:07:54Tornesince *all* these things have copies of all the toplevel stuff
01:08:05Torneand it does seem you are expected to just let htem overwrite each other :)9
01:08:33TheSevenis llvm better regarding that?
01:08:45Tornenever tried
01:09:02Tornenobody actually seems to build toolchains this way
01:09:11CIA-70New commit by kugel (r29330): Cleanup preprocessor around corelock usage and move its definition outside #ifdef ASSEMBLER_THREADS
01:09:11Torneeven though all the actual docs for the relevant packages tend to imply that's how it's done
01:09:25TheSevenbut gcc is getting on my nerves because of other things as well
01:09:27pamauryTheSeven: llvm is really nice to build
01:09:45pamaury(apart that it takes gigabytes of disk :-/)
01:09:51TheSevenalso, the rockbox toolchain produces way bigger code than for example yagarto. i have no explanation why, but it's a major problem for me
01:11:00thomasjfoxkugel: Why does android use a stack size of 0x1000 again? Native/SDL threads use 0x400. I'm just wondering if I need to patch that up for maemo, too.
01:11:07TheSeveni have that note exploit thing, where i need to fit a NAND driver + FTL + FAT32 code into ~3.7KB of space
01:11:55kugelthomasjfox: sdl threads don't use DEFAULT_STACK_SIZE
01:11:59TheSeventhis worked just fine with yagarto (yay, 3 bytes free, when using UCL-compressed thumb code), and with the toolchain compiled by it's like 300 bytes too big
01:12:17 Nick panni__ is now known as panni_ (
01:13:02 Quit MethoS- (Remote host closed the connection)
01:13:27kugelthomasjfox: it shouldn't need changes from maemo, sigaltstack uses bigger stacks (MINSIGSTKSZ+0x3000)
01:13:47thomasjfoxkugel: I want to try ASSEMBLER_THREAD again ;)
01:13:50TheSevenand also i've seen gcc save registers on the stack that it doesn't even use (it seems to always save r4, even though it would only need to save lr in some trivial wrapper)
01:14:06pamaurywhat is yagarto ?
01:14:16thomasjfoxkugel: Maybe the crashes on database init were related to the recent uninitialized vars fixes on metadata parsing
01:14:25kugel0x1000 works for android but I don't know for maemo
01:14:29CIA-70r29330 build result: All green
01:14:31TheSevenpamaury: the new name of the gnuarm-win32 toolchain
01:14:44thomasjfoxkugel: Congratulations!
01:14:59pamauryfinally :)
01:15:00TheSevenso basically also an eabi gcc, just a slightly different version, and probably different patches/compile-time options
01:15:03 Join t0rc [0] (~t0rc@unaffiliated/t0rc/x-5233201)
01:15:10 Join MethoS- [0] (~clemens@
01:16:40 Quit pamaury (Remote host closed the connection)
01:18:50thomasjfoxkugel: Database rebuilt worked with ASSEMBLER_THREADS
01:19:16kugelI suspected it was a compiler bug
01:19:28thomasjfoxkugel: I have to do a full .deb package rebuild, it segfaults on playback (codec load)
01:19:52thomasjfoxThough that segfault is probably just a mismatch of defines during compile time
01:21:52kugelTheSeven: can yagarto build rockbox?
01:22:58kugelfwiw, I don't think yagarto makes smaller code, but rather the newer gcc version
01:31:11thomasjfoxkugel: SIGALTSTACK and SDL_THREADS are working here. Testing ASSEMBLER_THREADS with full rebuild next
01:32:09CIA-70New commit by thomasjfox (r29331): N900: Prevent stuck up/down keys on systems with shared up/down cursor mapping
01:32:46TheSevenkugel: no idea
01:33:27TheSevenkugel: the yagarto version I use has gcc 4.5.0
01:33:52TheSeven(we're using 4.4.4)
01:37:08CIA-70r29331 build result: All green
01:40:09 Quit vnl (Ping timeout: 240 seconds)
01:40:17***Saving seen data "./dancer.seen"
01:41:15 Quit dfkt_ (Read error: Connection reset by peer)
01:42:35 Quit bertrik (Quit: :tiuQ)
01:48:45 Quit Judas_PhD (Ping timeout: 264 seconds)
01:50:18CIA-70New commit by thomasjfox (r29332): Use SIGALTSTACK in maemo build
01:51:37thomasjfoxkugel: ASSEMBLER_THREADS still crashes on playback. Compiler bug ;)
01:52:07kugeltried upping the stack size?
01:52:22thomasjfox0x1000 like android
01:52:30thomasjfoxok, one more try with 0x10000
01:53:21 Quit mudd1 (Ping timeout: 250 seconds)
01:54:30CIA-70r29332 build result: All green
01:55:05 Quit Jerom (Quit: Leaving.)
01:57:36 Quit sideral (Quit: Leaving.)
01:58:29thomasjfoxkugel: It crashes on various strange places. If I comment them out, it continues and crashes elsewhere. Atleast I can hear the playback :)
01:59:35kugelalright, so stick with sigaltstack I'd say
02:01:06thomasjfoxexactly :)
02:01:18 Quit liar (Quit: Leaving)
02:02:01 Join Judas_PhD [0] (
02:02:06thomasjfoxkugel: I'm currently testing windowed mode again. The window manager cuts away ~20-30 pixel at the bottom by inserting a status bar at the top.
02:02:41thomasjfoxkugel: Unfortunately it doesn't add this offset to the touchscreen surface, so the current cabbiev2 port is not usable
02:03:12thomasjfoxkugel: What should we do here? Add a close/switch window button? Custom theme?
02:03:50kugelperhaps you can modify the lcd driver to translate the offset so that it cuts off at the top?
02:05:45thomasjfoxkugel: looks like I was wrong, "touching" the items in the main menu is exactly the touched position
02:06:08thomasjfoxkugel: So it's just the WPS that sucks for now because all the playback controls are at the bottom
02:06:38kugelcan you hide the statusbar at runtime?
02:06:59kugeli.e. show it in the main menu, hide in the wps?
02:07:18thomasjfoxnot that I know of
02:07:28thomasjfoxThat would be the difference between windowed/fullscreen mode
02:07:34thomasjfoxI could make this switchable
02:08:34thomasjfoxThough the windowed mode would be preferred as the user is familiar with the controls already
02:12:36 Quit factor (Quit: Leaving)
02:13:43CIA-70New commit by kugel (r29333): Android: Rewrite lcd subsystem to use the SurfaceView API. ...
02:18:01CIA-70r29333 build result: All green
02:26:48CIA-70New commit by kugel (r29334): Correct setServiceActivity implementation
02:30:54CIA-70r29334 build result: All green
02:31:46CIA-70New commit by kugel (r29335): Android: Re-attach the media button receiver when entering Rockbox.
02:34:54 Quit ender` (Quit: I think I remember an episode of MacGyver where he overthrew a violent dictator with a rubber band, 2 bottle caps, and some navel lint.)
02:36:09CIA-70r29335 build result: 4 errors, 0 warnings (kugel committed)
02:37:44kugelstrange error
02:38:02kugelBagder: ^ ?
02:45:12CIA-70New commit by thomasjfox (r29336): Add missing cabbiev2 800x480 backdrop
02:48:53 Quit MethoS- (Remote host closed the connection)
02:49:05CIA-70r29336 build result: 4 errors, 0 warnings (thomasjfox committed)
02:50:13CIA-70New commit by gevaerts (r29337): Add storebror-daniel to blockedclients
02:59:20 Join mystica555 [0] (
03:13:32 Join milk [0] (
03:15:59kugelgevaerts: aha, the error actually says what to do
03:19:10CIA-70New commit by kugel (r29338): Explicitely add -lpthread to the linker command for the new sigaltstack threads.
03:19:30kugelgevaerts: storebrow-daniel should work now
03:20:11*S_a_i_n_t wonders why the Android widgets seem to *hate* |<< and >>|, and why line and box type widgets give dissimilar results.
03:20:14CIA-70New commit by gevaerts (r29339): Remove storebror-daniel from blockedclients again
03:21:01S_a_i_n_tand also why the .wps/.sbs doesn't recognise the correct playback state when the state is changed by the widget and not the app.
03:21:17gevaertskugel: I'm not actually sure if doing this sort of threading changes is a good idea during a freeze
03:21:19S_a_i_n_tparticularly, "stop".
03:21:47kugelgevaerts: it doesn't affect any released target, so
03:22:02gevaertskugel: it seems to affect simulators of released targets
03:22:12kugelbut yea, I initially wanted to wait until the freeze is over, but for some reason I completely forgot about it now
03:23:24CIA-70r29338 build result: All green
03:25:57kugelI don't mind if it's (temporarily) reverted if you feel unhappy with it
03:27:08gevaertsHow sure are you that this doesn't affect any of the simulator builds like e.g. cygwin?
03:27:47S_a_i_n_tcygwin builds are alread broken afaik.
03:27:51S_a_i_n_t(for sims)
03:28:10gevaertsor mingw
03:28:10S_a_i_n_tah, no...wait...I'm thinking SDL.
03:28:13kugelgevaerts: I test compiled and run under mingw
03:29:01*gevaerts thinks AlexP should decide on this one
03:38:55*S_a_i_n_t also wonders if knowledgeable pople in the area of SDL have seen the SDL_WaitEvent error faults reported with disktidy lately.
03:40:18***Saving seen data "./dancer.seen"
03:43:58 Quit kugel (Quit: leaving)
04:06:25LloreanDo we build simulators of releases?
04:10:59S_a_i_n_tseems so.">
04:11:34S_a_i_n_t(linked from the daily sims page)
04:12:05LloreanSo how is this patch affecting simulator builds? Does it change their threading too?
04:14:37 Join L-Strife89 [0] (~Strife89@
04:18:16 Join maxsin [0] (
04:22:53 Quit elcan (Ping timeout: 240 seconds)
04:24:08 Quit thomasjfox (Remote host closed the connection)
04:24:12 Join elcan [0] (
04:24:34 Quit L-Strife89 (Ping timeout: 260 seconds)
04:32:43 Join thomasjfox [0] (~thomasjfo@rockbox/developer/thomasjfox)
04:33:39thomasjfoxS_a_i_n_t: valgrind output of disktidy crash using H320 sim / SDL threads:
04:34:09 Quit Judas_PhD (Quit: This is a quitting message)
04:38:05 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
04:38:05 Quit amiconn (Disconnected by services)
04:38:12 Quit thomasjfox (Remote host closed the connection)
04:38:21 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
04:38:37 Quit pixelma (Disconnected by services)
04:38:39 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
04:38:41 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
04:47:19 Join Judas_PhD [0] (
04:52:51 Join Horschti [0] (
04:53:04 Join Xerion_ [0] (
04:55:37 Join Barahir_ [0] (
04:57:45 Join kkit`sh [0] (
04:58:59 Quit Barahir (Ping timeout: 260 seconds)
04:59:56 Join Kitr88 [0] (
05:00:38 Join T44 [0] (~Topy44@
05:00:38 Quit Kitar|st (Ping timeout: 255 seconds)
05:00:39 Quit Xerion (Ping timeout: 255 seconds)
05:00:40 Quit Horscht (Ping timeout: 255 seconds)
05:00:40 Quit Topy44 (Ping timeout: 255 seconds)
05:00:40 Quit jfc (Ping timeout: 255 seconds)
05:00:40 Quit kkit|sh (Ping timeout: 255 seconds)
05:00:41 Join jfc [0] (
05:00:41 Quit TheSeven (Ping timeout: 264 seconds)
05:00:42 Quit Horschti (Changing host)
05:00:42 Join Horschti [0] (~Horscht@xbmc/user/horscht)
05:00:44 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
05:00:46 Nick Xerion_ is now known as Xerion (
05:32:31 Quit Rob2222 (Ping timeout: 240 seconds)
05:34:08 Join Rob2223 [0] (
05:34:11 Join Zarggg [0] (
05:36:57 Quit Slasheri (Ping timeout: 276 seconds)
05:39:05 Join Slasheri_ [0] (
05:40:19***Saving seen data "./dancer.seen"
05:45:49 Join L-Strife89 [0] (~Strife89@
05:46:54 Quit maxsin (Quit: Leaving.)
05:59:19 Join Keripo [0] (
06:00:27 Quit Keripo (Client Quit)
06:19:58 Join Horscht [0] (
06:19:58 Quit Horscht (Changing host)
06:19:58 Join Horscht [0] (~Horscht@xbmc/user/horscht)
06:20:58 Quit BHSPitMonkey (Ping timeout: 240 seconds)
06:22:59 Quit Horschti (Ping timeout: 260 seconds)
06:34:12 Join BHSPitMonkey [0] (~stephen@unaffiliated/bhspitmonkey)
06:41:23 Quit Judas_PhD (Quit: This is a quitting message)
06:42:05 Quit L-Strife89 (Ping timeout: 276 seconds)
06:44:51 Nick S_a_i_n_t is now known as [Saint] (S_a_i_n_t@
06:45:39 Join Judas_PhD [0] (
06:47:34 Quit Judas_PhD (Client Quit)
06:50:31 Join Judas_PhD [0] (
07:09:24 Quit Zarggg (Read error: Connection reset by peer)
07:10:24 Join Zarggg [0] (
07:37:29 Quit panni_ (Quit: ( :: NoNameScript 3.81 :: ))
07:40:20***Saving seen data "./dancer.seen"
07:45:40 Quit Zarggg (Quit: Zarggg)
07:54:25 Quit t0rc (Remote host closed the connection)
07:59:07 Join mudd1 [0] (
08:01:28 Quit [Saint] (Quit: I'm only going to Heaven if it feels like Hell, I'm only going to Heaven if it tastes like caramel...)
08:06:39 Join [Saint] [0] (S_a_i_n_t@
08:11:22 Join L-Strife89 [0] (~Strife89@
08:11:39 Join factor [0] (~factor@
08:34:15 Quit L-Strife89 (Ping timeout: 260 seconds)
08:45:05 Join esperegu [0] (~quassel@
09:07:09 Quit BHSPitMonkey (Quit: Ex-Chat)
09:36:28 Quit factor (Read error: Connection reset by peer)
09:38:59 Join fkhodkov [0] (
09:40:03 Join fkhodkov_ [0] (
09:40:11 Quit fkhodkov_ (Remote host closed the connection)
09:40:24***Saving seen data "./dancer.seen"
09:40:36 Quit fkhodkov (Client Quit)
09:41:10 Join fkhodkov [0] (
09:52:50[Saint]can anyone enlighten me as to what I'm doing wrong here?
09:53:36[Saint]afaik I'm setting the path correctly, but configure seems to disagree.
09:54:11 Join factor [0] (~factor@
09:59:03 Quit sasquatch (Quit: WeeChat 0.3.2)
09:59:28 Join sasquatch [0] (
09:59:39 Join einhirn [0] (
10:05:39 Join kugel [0] (~kugel@rockbox/developer/kugel)
10:06:07 Quit einhirn (Quit: Miranda IM! Smaller, Faster, Easier.
10:07:12 Join kugel_ [0] (~kugel@rockbox/developer/kugel)
10:07:53 Quit kugel (Remote host closed the connection)
10:08:00 Nick kugel_ is now known as kugel (~kugel@rockbox/developer/kugel)
10:10:01 Join einhirn [0] (
10:14:15 Quit einhirn (Client Quit)
10:14:43kugel[Saint]: no idea...
10:18:20 Quit parafin (Ping timeout: 276 seconds)
10:18:50[Saint] my ubuntu VM has decided to not accept my password at boot.
10:19:04[Saint]this is fun.
10:22:12 Join kevku [0] (~kevku@2001:7d0:0:f9af:feed:feed:feed:feed)
10:33:28 Quit Unhelpful (Ping timeout: 250 seconds)
10:33:59 Join Unhelpful [0] (~quassel@rockbox/developer/Unhelpful)
10:34:41JdGordon|Buschel: my mini is still going strong (album on repeat for the last 25 odd hours), if i remember i'll put svn without the fix on it tonight and see if it actually helped at all
10:35:56JdGordon|Zagor: If you mst move it albumart.c is far better than the skin engine
10:36:27JdGordon|kugel: does RaaA now require 2.3?
10:37:18 Join stoffel [0] (
10:45:43 Quit Dreamxtreme (Ping timeout: 276 seconds)
10:46:19 Join vnl [0] (
10:48:46 Quit Judas_PhD (Quit: This is a quitting message)
10:49:21AlexPgevaerts, kugel: I have to be guided by you guys on technical aspects - things like cygwin sim builds are a bit of a difficult case as they get broken from time to time with nobody noticing anyway as so few people use them.
10:50:03AlexPgevaerts, kugel: We are due to branch tomorrow, maybe it would be best to revert the change in the branch if nobody has tested it? (To reduce noise in svn)
10:50:31 Join Dreamxtreme [0] (~Dre@
10:51:53kugelrasher has windows simulators on his site, but that doesn't count as release does it? anyway, I tested a mingw build and cygwin should also work
10:52:12kugelyou can also supply −−sdl-threads to configure to have the old behavoir
10:55:35AlexPkugel: well, as I say I trust you guys - I would have preferred to not put it in, but seeing as it is I'll leave it to you to decide whether it is worth taking it out again
10:56:32kugelas I said, I actually planned to wait but forgot about it. I
10:56:38kugelI'm not opposed to reverting
10:57:21kugelalthough it'd be a slight pita due to the post commits :)
10:57:36AlexPyes, I guessed that :)
10:59:34 Join jaykay1 [0] (
11:00:19 Quit milk (Read error: Connection reset by peer)
11:02:20 Quit Unhelpful (Ping timeout: 250 seconds)
11:03:52 Join TheLemonMan [0] (~lem0n@
11:04:23 Join Unhelpful [0] (~quassel@rockbox/developer/Unhelpful)
11:10:09kugelAlexP: should revert it
11:10:41kugel(with git cherry-pick it was only a very minor pita :) )
11:12:09 Join thomasjfox [0] (~thomasjfo@rockbox/developer/thomasjfox)
11:15:11 Join casainho [0] (
11:20:01 Join Judas_PhD [0] (
11:20:53AlexPkugel: Well seeing as you have taken the time to find it, maybe for completeness it'd be nice to do that. The other option is to wait until branching and do it in the branch, but it could then be a bugger if anyone else touches the code before then
11:21:33kugeland then revert the revert after branching? :)
11:22:33 Join webguest74 [0] (
11:23:10thomasjfoxThe new thread code is playing fine next to me ;)
11:23:39kugelalright, I take it as that we officicially care about simulators for releases as well?
11:23:39AlexPkugel: Well as I say, I give way to you on whether it is necessary :)
11:23:44thomasjfoxMaybe just default to SDL threads for compatibility reasons in 3.8?
11:23:45 Quit webguest74 (Client Quit)
11:24:01AlexPkugel: This is it, I'm not sure - we don't officially release any
11:24:19AlexPI don't think we've really thought about it before
11:24:51*AlexP is in two minds :)
11:24:53kugelyou and gevaerts decide I guess
11:25:04*AlexP waits for gevaerts :)
11:25:15 Join bertrik [0] (
11:25:15 Quit bertrik (Changing host)
11:25:15 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
11:25:42thomasjfoxIIRC gevaerts said to wait for AlexP. Deadlock?
11:26:03AlexPhe did, but I meant to discuss it :)
11:26:51thomasjfoxkugel: What about sticking to SDL threads as default for the release? That shouldn't change behavior?
11:27:19kugelwhat release? we don't release simulators :)
11:30:44thomasjfoxkugel: I was just thinking how we can get back the old behavior without reverting your lovely new code
11:31:20AlexPI think in this case we can probably leave it
11:31:34AlexPWe don't release sims, and there is the flag if needs be
11:31:43AlexPBut I'll conflab with gevaerts
11:35:25thomasjfoxlinuxstb: You were right about the play/pause button. The first thing my brother asked me when I showed him rockbox was "so, where is the pause button?"
11:36:59 Join ender` [0] (
11:39:08thomasjfoxkugel: Just ran the sigaltstack stuff in valgrind. Any idea about this?
11:39:52thomasjfoxkugel: Ah, forget it
11:40:05thomasjfoxkugel: valgrind can't detect the stack change
11:40:15 Join stripwax [0] (
11:40:28***Saving seen data "./dancer.seen"
11:40:38 Join parafin [0] (parafin@2001:470:1f0b:81::1)
11:41:53thomasjfoxkugel: Hmm, still unhappy with "−−max-stackframe=34334767696". I'll try a SDL threads build.
11:42:22 Quit stripwax (Client Quit)
11:42:44 Quit parafin (Client Quit)
11:42:50kugelthomasjfox: I never used valgrind, what does that mean?
11:42:51 Join parafin [0] (
11:43:05 Join stripwax [0] (
11:43:07thomasjfoxkugel: Do you know how valgrind works?
11:44:06thomasjfoxkugel: It intercepts OS calls like malloc() and others. So it can check for buffer overflows/underflows and use of uninitialized variables at runtime
11:44:13thomasjfoxIn fact, it checks all memory accesses
11:44:32thomasjfoxIt examines how a program behaves at runtime
11:44:57thomasjfoxThat's how I found out the uninitialized vars issue Buschel recently fixed
11:45:20thomasjfoxIt's easy as running "valgrind ./rockbox" with the SDL app build or a sim build
11:46:11 Join robin0800 [0] (
11:51:08 Join pamaury [0] (
11:51:09 Quit pamaury (Changing host)
11:51:09 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
11:51:38 Join Kitar|st [0] (
11:53:16kugelthomasjfox: I get Thread creation failed. "Retryingmake_context(): Operation not permitted" in valgrind
11:53:32kugelthe first " should be before Thread
11:53:35thomasjfoxkugel: Which valgrind version?
11:54:36thomasjfoxshould be fine
11:54:49thomasjfoxlet me take a close look at the output
11:55:30 Quit Kitr88 (Ping timeout: 260 seconds)
11:55:50thomasjfoxStill get the invalid write though as mentioned on pastie
11:55:57thomasjfoxCould be a false positive
11:56:14 Quit Kitar|st (Ping timeout: 250 seconds)
11:56:21 Join Stummi [0] (~Stummi@rockbox/developer/Stummi)
12:00:30 Quit TheLemonMan (Quit: free(me))
12:01:15 Join Kitar|st [0] (
12:01:27thomasjfoxkugel: I don't think sigaltstack is compatible with valgrind at the moment, it doesn't get past the startup screen
12:02:25 Join Keripo [0] (
12:02:29 Quit Keripo (Client Quit)
12:02:30kugel"Thread creation failed. Retrying" comes from valgrind, "make_context(): Operation not permitted" comes from rockbox
12:02:55kugelso one of those library calls is not allowed within valgrind and valgrind detects that we try to make threads
12:03:21kugelah no wait, the "Thread creation failed. Retrying" also comes from rockbox
12:03:35thomasjfoxI don't get any of those error messages, it just hangs
12:04:42thomasjfoxkugel: ah, now I get it too with a sim build
12:04:49thomasjfoxkugel: Doesn't happen with the SDL app build
12:05:51kugelkugel: yea, the default app build stubs DEBUGF()
12:05:55kugelthomasjfox: ^
12:05:59 Join inox [0] (
12:07:25kugelthomasjfox: sigaltstack() fails within valgrind
12:08:24stripwaxinox - hello
12:08:47 Quit inox (Client Quit)
12:09:28 Join liar [0] (
12:10:03 Join webguest71 [0] (
12:10:48stripwaxwebguest71 - hello
12:15:31 Quit webguest71 (Quit: CGI:IRC (Ping timeout))
12:15:42 Join webguest90 [0] (
12:16:59 Quit webguest90 (Client Quit)
12:18:06AlexPkugel: See ? :)
12:21:46thomasjfoxkugel: do you have a standalone / development version of the ucontext stuff?
12:22:00kugelyes, somewhere
12:22:03thomasjfoxkugel: The example found here works fine in valgrind
12:22:24thomasjfoxkugel: The example under "figure 3"
12:23:55thomasjfoxkugel: oh, the example in "figure 4" doesn't. I'll ask the valgrind poeple about it
12:24:05kugelthomasjfox: figure 4 uses makecontext, we don't
12:25:06kugelI compiled valgrind, how do I execute the test suits?
12:25:14 Part jaykay1
12:26:50 Quit Judas_PhD (Quit: This is a quitting message)
12:27:09thomasjfoxkugel: my rpm spec file first does "make check"
12:27:28thomasjfoxkugel: and then "make regtest"
12:27:47kugelalright, trying that. it has a sigaltstack test
12:29:41kugelstack_switch: valgrind -q ./stack_switch
12:29:41kugel*** stack_switch failed (stderr) ***
12:31:12kugelbut the sigaltstack test run fine
12:31:30thomasjfoxkugel: is stack_switch your implementation?
12:33:42 Join esperegu_ [0] (~quassel@
12:34:30thomasjfoxkugel: are you on a 64bit or 32bit system?
12:34:35 Join GeekShadow [0] (~Antoine@
12:34:36 Quit GeekShadow (Changing host)
12:34:36 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
12:34:44 Quit GeekShadow (Remote host closed the connection)
12:34:54thomasjfoxme too. maybe that's the issue?
12:35:03 Quit esperegu (Ping timeout: 276 seconds)
12:35:38thomasjfoxanyway, I didn't notice anything unusal with your sigaltstack code, playing for hours now. So it's fine :)
12:41:44kugelhm, I can't run the compiled valgrind
12:41:54 Join dfkt [0] (dfkt@unaffiliated/dfkt)
12:41:57 Join TheLemonMan [0] (~lem0n@
12:43:30thomasjfoxkugel: Stupid question: Did you run "make install"? It looks for some components at runtime
12:43:47kugelI installed into <src-dir>/bin
12:45:00kugelaha, strace shows it looks in /usr/local/lib
12:45:00thomasjfoxkugel: Running the 32bit version of valgrind against the 32bit version of "figure 4" works. So it's a 64bit system issue
12:45:30thomasjfoxkugel: In fact, how does rockbox cope with 64bit??
12:45:42thomasjfoxkugel: Shouldn't we force gcc to 32bit?
12:45:53kugelwe have 64bit sims for years without problems
12:48:17kugelthomasjfox: does the 32bit valgrind also work with a 32bit sim?
12:48:38thomasjfoxis there an easy way to compile a 32bit sim?
12:49:17kugeldon't know
12:49:34kugelI assume specifying the 32bit cross compiler gcc should work
12:50:24thomasjfoxI added the -m32 and −−target to simcc. Lets see how that goes
12:50:39kugelconfigure sees $CROSS_COMPILE
12:53:28kugelthomasjfox: valgrind's testcase is slightly different
12:54:02 Join MethoS- [0] (~clemens@
12:55:17kugelthomasjfox: I sent you my standalone code
12:55:39thomasjfoxkugel: checking
12:56:21 Quit casainho (Ping timeout: 246 seconds)
12:58:40kugelmodifying the test case to use a static stack, instead of a mmap'd one also works
13:00:24 Join maxsin [0] (~maxsin@188-115-150-221.broadband.tenet.odessa.UA)
13:02:43thomasjfoxkugel: Sent you the output
13:03:14kugelwell, it's a different error
13:04:27kugelthe manpage says [EPERM] An attempt was made to modify an active stack.
13:06:56thomasjfoxkugel: Sent you another output with "−−track-origins=yes".
13:07:03thomasjfoxkugel: Looks bogus to me
13:07:17 Join DerPapst [0] (
13:07:56kugelI don't know what these messages mean
13:10:45 Quit maxsin (Ping timeout: 260 seconds)
13:12:29kugelthomasjfox: can it possibly have something to do with the multithread environment?
13:14:32thomasjfoxkugel: No idea. I just tried something "stupid": I did zero initialize all the structures in mk_ctx. Now valgrind is happy. huh?
13:14:58thomasjfoxkugel: Let me try that with 64bit code again
13:22:18 Join benedikt93 [0] (~benedikt9@unaffiliated/benedikt93)
13:24:59kugelhaha, "Memcheck: the 'impossible' happened:
13:30:55 Quit jfc (Ping timeout: 240 seconds)
13:33:17 Quit stripwax (Quit:
13:35:00 Join hebz0rl [0] (
13:35:19kugel[Saint]: can you try building a sim on cygwin?
13:35:25thomasjfoxkugel: I'd say valgrind is currently not compatible with sigaltstack
13:35:28kugel(FS #11952)
13:35:35 Join stripwax [0] (
13:35:42kugelthomasjfox: this and 64bit :)
13:36:01kugelI can get valgrind to segfault
13:36:19thomasjfoxkugel: run valgrind on valgrind :o)
13:37:49kugelthat doesn't seem to work
13:40:30***Saving seen data "./dancer.seen"
13:42:28 Join halmi [0] (~netbook@
13:43:04 Join bluefoxx_ [0] (
13:45:26 Join halmi_ [0] (~netbook@
13:46:24 Quit bluefoxx (Ping timeout: 246 seconds)
13:47:29 Quit halmi (Ping timeout: 240 seconds)
13:50:59 Quit user890104 ()
13:51:43pixelmasoap, AlexP, scorche: please ban dfhjkh54 in the forums
13:52:09 Quit hebz0rl (Quit: Leaving)
13:53:44AlexPpixelma: done
13:57:41 Join maxsin [0] (~maxsin@188-115-150-221.broadband.tenet.odessa.UA)
14:07:49 Join DerPapst1 [0] (
14:08:32 Join simabeis_ [0] (
14:08:39 Join alexbobP_ [0] (
14:08:41 Join tomj_ [0] (~thomasjfo@rockbox/developer/thomasjfox)
14:08:49 Join zu_ [0] (
14:08:59 Quit [Saint] (Disconnected by services)
14:09:01 Join S_a_i_n_t [0] (S_a_i_n_t@
14:09:37 Join TheLemonMan_ [0] (~lem0n@
14:10:35 Join esperegu [0] (~quassel@
14:11:13 Join knitt1 [0] (~knittl@unaffiliated/knittl)
14:13:52 Quit DerPapst (Ping timeout: 246 seconds)
14:13:52 Quit zu (Ping timeout: 246 seconds)
14:13:52 Quit alexbobP (Ping timeout: 246 seconds)
14:13:52 Quit simabeis (Ping timeout: 246 seconds)
14:13:53 Quit TheLemonMan (Ping timeout: 246 seconds)
14:13:53 Quit esperegu_ (Ping timeout: 246 seconds)
14:13:53 Quit thomasjfox (Ping timeout: 246 seconds)
14:13:53 Quit ved (Ping timeout: 246 seconds)
14:13:53 Quit knittl (Ping timeout: 246 seconds)
14:14:16 Join ved [0] (
14:14:32 Quit halmi_ (Ping timeout: 240 seconds)
14:18:28 Join dfkt_ [0] (dfkt@unaffiliated/dfkt)
14:22:07 Quit dfkt (Ping timeout: 250 seconds)
14:23:33 Join halmi [0] (~netbook@
14:23:44 Quit tomj_ (Remote host closed the connection)
14:24:00 Join thomasjfox_ [0] (~thomasjfo@rockbox/developer/thomasjfox)
14:24:06 Nick thomasjfox_ is now known as thomasjfox (~thomasjfo@rockbox/developer/thomasjfox)
14:39:08 Quit thomasjfox (Remote host closed the connection)
14:54:31 Quit halmi (Quit: halmi)
14:55:29 Join greyback [0] (
15:03:49 Join Topy44 [0] (~Topy44@
15:04:07 Join afk [0] (~Dre@
15:04:31 Quit T44 (Ping timeout: 240 seconds)
15:06:19 Quit Dreamxtreme (Ping timeout: 250 seconds)
15:11:33 Quit Kitar|st ()
15:14:03gevaertsAlexP, kugel: I don't really agree that we don't release simulators
15:14:26gevaertsWe don't release simulator *builds*, but we do release source that is expected to build working simulators
15:14:55AlexPtrue enough
15:15:01gevaertsSo if simulators have a serious bug, I'd definitely see that as a reason for a point release
15:15:41AlexPAnyway, seeing as cygwin simulators don't currently build with thread errors, I'd say revert it.
15:20:04kugelAlexP: I'm quite certain Buschel just didn't make reconf
15:20:19AlexPah, OK
15:20:28AlexPwell, someone on cygwin needs to test :)
15:20:30pixelmaerrmm, just added a comment to the task that I successfully build an r29339 Gigabeat S sim with my cygwin 1.7 machine here
15:20:33kugelsomeone on cygwin just needs to confirm
15:20:38AlexPpixelma: OK, thanks
15:20:40kugelhehe :)
15:20:48AlexPscore 1 for kugel's intuition :)
15:21:09pixelmathought he would get a notification mail...
15:26:54stripwaxi just built a sim build on cygwin, seems to work fine.
15:27:17stripwaxalthough I've not been following the convo so maybe something more subtle than 'it just works for me' ?
15:28:17stripwaxalso @ 29339
15:29:35 Quit afk (Quit: Ex-Chat)
15:30:03 Join Dreamxtreme [0] (~Dre@
15:38:32 Quit stripwax (Quit:
15:38:58 Quit dfkt_ (Read error: Connection reset by peer)
15:38:59 Join dfkt [0] (dfkt@unaffiliated/dfkt)
15:39:06jhMikeSkugel: the sim wait_for_interrupt is certainly curious. I see no mutex lock anywhere before waiting on the condition. ??
15:39:14 Join GeekShadow [0] (~Antoine@
15:39:14 Quit GeekShadow (Changing host)
15:39:14 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
15:40:32kugeljhMikeS: in tick_start
15:40:34***Saving seen data "./dancer.seen"
15:41:24jhMikeSwouldn't that be by the main thread?
15:41:49jhMikeSof course maybe this doesn't work like I remember :)
15:41:53kugelyes, the same thread that calls wait_for_interrupt()
15:41:56gevaertsAlexP, kugel: I'm in favour of reverting, but I don't really care much if that's today in trunk or tomorrow in the branch
15:42:06AlexPMy thoughts too
15:42:24AlexPtomorrow is probably cleaner if nothing else touches it before then
15:44:39jhMikeSkugel: wait, but what about around the SDL_CondSignal?
15:46:28kugeljhMikeS: why should it grab the lock before signalling?
15:46:57jhMikeSI believe it's supposed to by design
15:47:45kugelI understood it differently
15:48:00kugeland it wouldn't work if nothing is waiting on the condition
15:49:22jhMikeSsemaphore is probably a simpler solution
15:50:52kugelI've done this with a sem for android and it's tricky
15:53:08 Join bluebrother [0] (
15:53:09 Quit bluebrother (Changing host)
15:53:09 Join bluebrother [0] (~dom@rockbox/developer/bluebrother)
15:55:19 Quit bluebroth3r (Ping timeout: 264 seconds)
15:59:30kugeljhMikeS: you can lock if you desperately want to avoid races, but it's not critical in this case
16:00:16kugelso I rather chose to make wait_for_interrupt a bit cheaper
16:02:25jhMikeSI was looking at the SDL source code. It's really not a race any more than not disabling IRQ on a real device while executing WFI, actually quite the same thing.
16:04:02pamaurydo arm core have something like a performance counter or similar ? (arm 920t precisely)
16:04:03jhMikeSlooks safe for the object anyway if not using it for data protection
16:04:52kugelyea, we use it exclusively to not hog the cpuö
16:08:21 Join Buschel [0] (
16:09:15Buschelfwiw, I did not do "make reconf" :/
16:09:19*Buschel does right now
16:20:37 Join Kitar|st [0] (
16:20:37 Quit Kitar|st (Client Quit)
16:23:04 Join Kitar|st [0] (
16:23:04 Quit Kitar|st (Client Quit)
16:23:19 Join Kitar|st [0] (
16:28:44 Join Judas_PhD [0] (
16:30:43 Quit Judas_PhD (Client Quit)
16:30:43kugelBuschel: what are your results?
16:31:10Buschellooks good, build finished in this very second
16:31:22Buschelso, my apologies. I will close the report
16:31:30kugelno problem :)
16:36:24gevaertsAlexP: RHCC wants banning
16:41:59CIA-70New commit by Buschel (r29340): FS #11950: Update czech language file.
16:46:01CIA-70r29340 build result: All green
16:49:22jhMikeSBuschel: I have a patch on your FS. I expect testing or I'm taking my chances with it (and interpreting lack of response as implicit trust in my abilities). :)
16:53:46 Quit Buschel (Ping timeout: 276 seconds)
16:56:35 Join casainho [0] (
17:20:23 Quit TheLemonMan_ (Quit: free(me))
17:39:29 Join esperegu_ [0] (~quassel@
17:40:12 Quit esperegu (Ping timeout: 240 seconds)
17:40:38***Saving seen data "./dancer.seen"
17:42:48CIA-70New commit by kugel (r29341): Android: Fix some race conditions and crashes on startup.
17:44:38jhMikeSkugel: I was getting segfaults in mpegplayer but not before the change. Will that take care of that?
17:45:08kugeljhMikeS: mpegplayer crashes?
17:46:32jhMikeSI reverted to r29326 and it played ok, r29340 was segfaulting when loading movie
17:47:11CIA-70r29341 build result: All green
17:47:27jhMikeSsdl sim on debian btw
17:49:20kugeljhMikeS: I'm assuming stack overflow
17:49:31kugelthey don't use DEFAULT_STACK_SIZE
17:49:54jhMikeSthat always had phony stacks
17:50:11jhMikeSor is that different now?
17:51:52kugelthe stack is used now, and on linux you need a minimum stack size (MINSTKSZ)
17:54:28*jhMikeS will gdb next to check it out then
17:54:37kugeljhMikeS: helps
18:00:44jhMikeSwell, surely 4096+DEFAULT_STACK_SIZE is bigger than 4096 :)
18:03:19 Join Keripo [0] (
18:04:29 Quit Keripo (Client Quit)
18:04:33ender`but what if default stack size is negative? :)
18:08:31jhMikeSthen we're indeed in trouble :)
18:10:12jhMikeShmmm...not enough stack to even get past the splash in gdb for the fun of it
18:12:18CIA-70New commit by kugel (r29342): Android: Fix compilation
18:12:23 Quit MethoS- (Remote host closed the connection)
18:13:28 Join MethoS- [0] (~clemens@
18:14:59 Join jfc [0] (
18:16:36CIA-70r29342 build result: All green
18:20:57 Join dfkt|x [0] (
18:20:58 Quit dfkt|x (Changing host)
18:20:58 Join dfkt|x [0] (~dfkt@unaffiliated/dfkt)
18:23:33 Nick alexbobP_ is now known as alexbobp (
18:24:03 Quit dfkt|x (Remote host closed the connection)
18:37:24 Quit robin0800 (Quit: Leaving)
18:37:30 Join robin0800 [0] (
18:38:19 Join robin0800_ [0] (
18:38:26 Quit robin0800 (Client Quit)
18:40:32 Quit literal (Read error: Connection reset by peer)
18:42:11 Join literal [0] (
18:45:23 Part vnl
18:47:00pixelmaS_a_i_n_t: if I want the same touch "button" do different things on long and short presses - do I really need to define two touch areas in the same spot, one with the long press action and the other with the short press one?
18:47:58pixelmadoesn't it clash somehow? I mean, e.g., play/pause on short press and stop on long, or seeking and skipping
18:48:16kugelworks fine on android
18:48:23 Quit esperegu_ (Read error: Operation timed out)
18:50:44pixelmaI find it a bit inconvenient and misleading. Couldn't it be an additional parameter in the touch area definition (coordinates, short press action, & long press action)?
18:50:52 Join esperegu [0] (~quassel@
18:59:00 Join leavittx [0] (~lev@
19:13:55 Quit leavittx (Ping timeout: 240 seconds)
19:16:46 Join Buschel [0] (
19:18:15 Join leavittx [0] (~lev@
19:19:25pixelmafor anyone involved in the "random skip to next song" discussion last weekend (I think it was) - I also get this sometimes on my c200 running 3.7 especially in playlists I created in small steps, inserting a few songs one by one or a small folder here and there), it sometimes skips to a next track in the middle of a song, if I then skip back everything is fine. Unfortunately I haven't found a way to reliably reproduce though
19:32:37 Nick knitt1 is now known as knittl (~knittl@unaffiliated/knittl)
19:33:15 Part greyback ("Ex-Chat")
19:40:41***Saving seen data "./dancer.seen"
19:43:39 Join panni_ [0] (
19:47:00 Join Horschti [0] (~Horscht@xbmc/user/horscht)
19:50:19 Quit Horscht (Ping timeout: 240 seconds)
19:59:25*TheSeven just read the MBR of a CE-ATA ipod drive \o/
19:59:53gevaertsNow play music from it!
20:00:04TheSevenwell, it's not *that* far away
20:00:40TheSevenbut there seem to be some timing peculiarities :/
20:02:12jhMikeSThe MBR is in some odd meter like 7/8?
20:03:18 Join einhirn [0] (
20:03:22 Quit einhirn (Client Quit)
20:06:54*TheSeven didn't get that joke
20:07:50TheSeventhere's exactly one upside of CE-ATA: everything is DMA LBA48 :)
20:09:05jhMikeSgevaerts seemed to be suggesting to play music from the MBR :P
20:09:55jhMikeSanyway, why's a build with thumb whining about intptr_t but a non-thumb build makes fine :\
20:10:36TheSevensounds like yet another ifdef fuckup?
20:11:01jhMikeSquestion of the century, when there's no obvious #ifdef around
20:13:20 Quit stoffel (Ping timeout: 276 seconds)
20:17:02jhMikeSseems to be cause by leaving out system.h in the includes in some files
20:23:38*jhMikeS wtf's over thumb binsize deltas
20:25:11 Join stripwax [0] (
20:27:09TheSevenjhMikeS: like a factor of 1.5?
20:31:30jhMikeS1.04 this time actually
20:33:44TheSevenhm, integrating ce-ata support into the rockbox ata driver isn't going to be easy
20:33:53TheSevenespecially if we want it to be efficient
20:34:32*TheSeven ponders forking the driver
20:41:07 Quit GeekShadow (Ping timeout: 240 seconds)
20:41:18 Join mystica555_ [0] (
20:47:11 Join saratoga_ [0] (9803c6dd@gateway/web/freenode/ip.
20:47:11 Quit saratoga_ (Client Quit)
20:47:25saratogajhMikeS: does that patch you posted build for you?
20:47:47saratogaor do I need another patch first to use it?
20:48:16 Join TheLemonMan [0] (~lem0n@
20:48:56 Quit factor (Ping timeout: 240 seconds)
20:50:43 Quit stripwax (Quit:
20:52:57 Join wodz [0] (
20:53:46wodzstupid question - how to simulate memcpy result in IDA? I mean copy some portion of memory into other address
20:54:41jhMikeSsaratoga: what's the issue? there's also an update coming
20:55:00saratogajhMikeS: it can't find intptr_t in codec_thread.h
20:55:06saratogawhich header includes that?
20:55:38jhMikeSfor a thumb build? I just added system.h to codec_thread.c and playback.c
20:56:11jhMikeSinttypes.h I think is the main one though
20:56:37saratogai'll try including that and see if it builds
20:57:50jhMikeSI'll have a new patch in a few anyway since there was a codec change snafu
20:58:29saratogaok cool
20:59:19saratogai skimmed that patch, but really didn't understand how half that stuff worked to begin with
21:01:13 Quit Buschel (Ping timeout: 240 seconds)
21:01:24 Join Jerom [0] (~jerome@
21:01:46*jhMikeS thinks he understands 1/3 to 1/2 of it−− whatever he rewrote himself :)
21:02:21jhMikeSthe rest seems designed by some genetic algorithm
21:08:04kugelare you trying to build in thumb?
21:12:15jhMikeSthe clip does build in thumb
21:12:23saratogait does?
21:12:33saratogashould i apply both those patches?
21:12:51jhMikeSyeah, for the full effect :)
21:13:18jhMikeSthe two I *just* posted of course
21:15:25jhMikeSwithout the metadata wait change, repeatedly seleting the track (so it sounds like a skipping cd) in the playlist viewer can lock.
21:18:17saratogamashing the buttons on the wps screen as fast as I can doesn't dead lock, pretty sure that used to at some point
21:18:23 Join Judas_PhD [0] (
21:19:37saratogai also don't seem to be able to get teh WPS info and the currently decoding track to get out of sync
21:22:17jhMikeSthat's that point, it locks the codec activity to the audio thread for changovers instead of it all racing around uncontrolled
21:25:13 Quit robin0800_ (Quit: Leaving)
21:29:51saratogagood job then
21:30:24saratogai don't seem to be able to get it to mess up now on the clip+
21:36:06saratogawoah just managed to get it to skip to a different folder
21:36:10saratogathats a little weird
21:36:54kugelwhat patch/FS# are you guys talking about?
21:37:10saratogaFS #11863 - Playback / buffer freezes
21:37:45saratogais rockbox expect to skip folders if skip a few times, then hold the forward button down (as if to seek)?
21:37:53saratoganot sure if this is a bug or a feature
21:38:19jhMikeSsaratoga: I've hit folder skip by accident too
21:38:29saratogaah so its supposed to do that?
21:38:31jhMikeSshort then long = folder skip
21:38:52saratogayeah ok
21:38:55saratogaso no issues then
21:39:02TheSevenwodz: there are multiple ways
21:39:32TheSevenif you have the contents that you want to have in that memory area in an external file anyway, you may just use "load additional binary file"
21:39:47TheSevenyou can also rewrite memcpy using Dword() and PatchDword() in IDC
21:39:50jhMikeSsaratoga: great! now I need tracks that pass metadata muster but fail decoding since I'm not confident about the auto change on error part
21:39:57TheSevenneither of that will copy any markup though, just the plain data
21:40:42***Saving seen data "./dancer.seen"
21:54:41 Join FBI_Guy [0] (
22:04:28 Quit kugel (Quit: leaving)
22:07:17 Join GeekShadow [0] (~Antoine@
22:07:18 Quit GeekShadow (Changing host)
22:07:18 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
22:11:07 Quit leavittx (Ping timeout: 240 seconds)
22:18:16 Quit benedikt93 (Quit: Bye ;))
22:26:08TheSeventhat nasty hard drive doesn't even comply to the ce-ata standard!
22:28:03 Join kugel [0] (~kugel@
22:28:04 Quit kugel (Changing host)
22:28:05 Join kugel [0] (~kugel@rockbox/developer/kugel)
22:29:11 Quit saratoga (Quit: Page closed)
22:32:39TheSevendamn, there doesn't seem to be a way to determine if the drive is safe to power off :/
22:32:49TheSevenit keeps BSY asserted while it's spun down
22:33:06 Join factor [0] (~factor@
22:33:31 Quit Jerom (Quit: Leaving.)
22:35:13TheSevenand it asserts RDY all the time
22:35:48jhMikeSpin misconfig of some sort? (random guessing)
22:56:44 Quit wodz (Quit: Leaving)
22:57:09 Quit FBI_Guy (Read error: Connection reset by peer)
23:05:32TheSevenjhMikeS: no, it turned out I get an IRQ when it's ready (or actually about half a second after that)
23:07:00TheSevenok, my emcore ce-ata driver seems to work. but integrating that into rockbox isn't going to be easy :/
23:07:31 Join milk [0] (
23:08:15 Quit Judas_PhD (Quit: This is a quitting message)
23:10:42TheSevenany ideas?
23:10:58TheSevenif i try to integrate it straight forward, we'll end up with an ifdef hell in ata.c
23:11:39TheSevenimplementing it as a separate driver isn't really good either, as they're mutually exclusive
23:21:51jhMikeSno idea really since I'm rather clueless on the necessities atm (should probably see emcore one)
23:23:11TheSevensearch for "if (ceata)"
23:24:59 Join user890104 [0] (
23:27:40 Join Judas_PhD [0] (
23:31:59 Quit mystica555_ (Read error: Connection reset by peer)
23:34:10 Quit Llorean (Read error: Connection reset by peer)
23:35:01jhMikeSfunction pointer table concealed by the storage wrappers?
23:36:47jhMikeSor some sort of thing to plug the other thing in on the fly if that's what you need *shrugs*
23:40:46***Saving seen data "./dancer.seen"
23:50:17 Join einhirn [0] (
23:52:09 Quit Stummi (Quit: Bye!)
23:52:10 Join einhirn_ [0] (
23:54:48 Quit einhirn (Ping timeout: 260 seconds)

Previous day | Next day