Previous day | Jump to hour: 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 | Next day

Seconds: Show Hide | Joins: Show Hide | View raw
Font: Serif Sans-Serif Monospace | Size: Small Medium Large

Click in the nick column to highlight everything a person has said.
The Logo icon identifies that the person is a core developer (has commit access).

#rockbox log for 2010-12-25

00:03:02 Nick sinthete1 is now known as sinthetek (
00:03:10 Quit sinthetek (Changing host)
00:03:10 Join sinthetek [0] (~sinthetek@unaffiliated/sinthetek)
00:09:03 Quit mischasworld (Ping timeout: 240 seconds)
00:40:44***Saving seen data "./dancer.seen"
00:54:37 Quit Keripo (Quit: Leaving.)
01:00:46 Join Keripo [0] (
01:01:17 Quit Keripo (Client Quit)
01:09:12 Join PurlingNayuki [0] (~PurlingNa@
01:09:51 Quit JesusFreak316 (Ping timeout: 240 seconds)
01:20:10saratogawhat do people think about disabling doom on players with <16MB of RAM? until someone tweaks the memory usage somehow its going to crash pretty regularly on those
01:21:06 Quit factor (Quit: Leaving)
01:21:48Tornewhen it's been suggested lots of people complain that they run doom on those players
01:22:17 Join factor [0] (
01:23:59 Join henk [0] (
01:26:10 Quit factor (Read error: Connection reset by peer)
01:26:48 Join Keripo [0] (
01:28:47saratogamaybe this will motivate them to fix it
01:28:48 Join Llorean1 [0] (
01:28:48 Quit Llorean1 (Client Quit)
01:29:24PurlingNayukiWell finally we can boot and use OF on ONDA_VX777!
01:30:29 Quit Llorean (Ping timeout: 276 seconds)
01:42:13 Join factor [0] (
01:46:36 Part PurlingNayuki
01:50:46 Join LambdaCalculus37 [0] (
01:50:46 Quit LambdaCalculus37 (Changing host)
01:50:47 Join LambdaCalculus37 [0] (~rmenes@rockbox/staff/LambdaCalculus37)
01:53:49 Quit factor (Ping timeout: 265 seconds)
02:04:43 Quit Kitar|st (Ping timeout: 260 seconds)
02:07:23 Join DSStrife89 [0] (
02:10:04 Join Kitar|st [0] (
02:15:04 Quit DSStrife89 (Quit: ClIRC v0.05 (homebrew IRC client for Nintendo DS))
02:18:01 Join Strife89 [0] (
02:18:27Rasii am affected by this bug
02:18:43Rasimy build of rockbox is the latest that is available
02:18:53Rasistill this bug is marked as solved for months
02:19:03Rasii am on a sansa e260
02:21:01gevaertsIt might be a different (new) bug with similar symptoms
02:22:29saratogamost of the stuff in that thread looks likes its fixed, so file a new bug report with detailed directions on how someone can reproduce your problem
02:24:20Rasiheh... ok
02:24:28Rasisadly its not reproducable at all :)
02:24:31Rasihappens randomly
02:25:07Rasiup until today i thought my computer does strange things while transfering music, only letting half tracks pass
02:39:12 Join t0rc [0] (~t0rc@unaffiliated/t0rc/x-5233201)
02:39:40CtcpPing from gevaerts!~fg@rockbox/developer/gevaerts
02:40:47***Saving seen data "./dancer.seen"
02:53:08 Quit mortalscan (Ping timeout: 260 seconds)
03:03:12 Join mortalscan [0] (~mortalsca@
03:07:22 Quit LambdaCalculus37 (Quit: This computer has gone to sleep)
03:20:53 Quit t0rc (Quit: Leaving)
03:31:18 Quit GeekShadow (Quit: The cake is a lie !)
03:37:48 Join factor [0] (
03:48:14 Join JesusFreak316 [0] (
03:55:02 Join bertrik [0] (
03:55:03 Quit bertrik (Changing host)
03:55:03 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
03:57:27 Join Horscht [0] (
03:57:28 Quit Horscht (Changing host)
03:57:28 Join Horscht [0] (~Horschti@xbmc/user/horscht)
03:58:17 Quit keanu (Ping timeout: 255 seconds)
04:00:08 Quit Horschti (Ping timeout: 240 seconds)
04:01:42 Quit mortalscan (Ping timeout: 276 seconds)
04:02:10 Quit krazykit (Ping timeout: 240 seconds)
04:04:52 Quit TheSeven (Ping timeout: 264 seconds)
04:07:31 Join krazykit [0] (~krazykit@
04:08:23 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
04:18:57 Join mystica555_ [0] (
04:26:06 Quit simonrvn (Read error: No route to host)
04:37:54 Quit Strife89 (Quit: Merry Christmas, everyone.)
04:38:29 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
04:38:29 Quit pixelma (Disconnected by services)
04:38:31 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
04:39:22 Quit amiconn (Disconnected by services)
04:39:23 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
04:39:40 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
04:40:49***Saving seen data "./dancer.seen"
04:43:03 Join FBI_Guy [0] (
04:45:06 Quit FBI_Guy (Client Quit)
05:02:32 Quit TheSeven (Ping timeout: 255 seconds)
05:03:57 Join LambdaCalculus37 [0] (
05:03:57 Quit LambdaCalculus37 (Changing host)
05:03:57 Join LambdaCalculus37 [0] (~rmenes@rockbox/staff/LambdaCalculus37)
05:06:17 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
05:13:38 Quit madalu (Ping timeout: 260 seconds)
05:54:05 Join Horschti [0] (~Horschti@xbmc/user/horscht)
05:54:12 Quit Keripo (Quit: Leaving.)
05:58:03 Quit Horscht (Ping timeout: 260 seconds)
06:21:49 Quit T44 (Read error: Connection reset by peer)
06:21:58 Quit factor (Ping timeout: 260 seconds)
06:23:00 Join tchan1 [0] (
06:25:27 Quit tchan (Ping timeout: 265 seconds)
06:30:11 Quit LambdaCalculus37 (Quit: This computer has gone to sleep)
06:33:22 Quit S_a_i_n_t (Ping timeout: 240 seconds)
06:33:45 Join LambdaCalculus37 [0] (~rmenes@rockbox/staff/LambdaCalculus37)
06:37:31 Join factor [0] (
06:39:22 Quit JesusFreak316 (Ping timeout: 240 seconds)
06:40:51***Saving seen data "./dancer.seen"
06:42:58 Quit the_Kyle (Ping timeout: 240 seconds)
06:54:07 Join [Saint] [0] (S_a_i_n_t@
07:01:15 Join Horscht [0] (~Horschti@xbmc/user/horscht)
07:01:41 Join Topy44 [0] (
07:04:09 Quit Horschti (Ping timeout: 240 seconds)
07:07:22 Join kadoban [0] (
07:18:30 Join simonrvn [0] (
07:25:40 Quit tchan1 (Quit: WeeChat 0.3.3-dev)
07:28:17 Quit froggyman (Quit: Ex-Chat)
07:28:37 Join tchan [0] (~tchan@lunar-linux/developer/tchan)
07:40:00 Quit LambdaCalculus37 (Quit: This computer has gone to sleep)
07:42:37 Quit tchan (Quit: WeeChat 0.3.4-rc1)
07:44:16 Quit Kitar|st ()
07:46:22 Join tchan [0] (~tchan@lunar-linux/developer/tchan)
07:59:58 Join Horschti [0] (~Horschti@xbmc/user/horscht)
08:03:32 Quit Horscht (Ping timeout: 255 seconds)
08:14:55 Join Llorean [0] (
08:15:02 Quit Llorean (Changing host)
08:15:02 Join Llorean [0] (~DarkkOne@rockbox/user/Llorean)
08:24:41 Join mortalscan [0] (~mortalsca@
08:25:51 Join tttt_ [0] (411dfb31@gateway/web/freenode/ip.
08:27:12 Quit tttt_ (Client Quit)
08:31:54 Join traveler [0] (411dfb31@gateway/web/freenode/ip.
08:33:37travelerI downloaded a newer version of rockbox's source and attempted to import a plugin I had that compiled for an earlier version. When I try to compile I get the message "multiple definition of `__header'". it compiles fine without my added plugins though
08:35:47traveleris there a way to remedy this?
08:40:53***Saving seen data "./dancer.seen"
08:43:23 Quit Galois (Remote host closed the connection)
08:43:33 Join Galois [0] (
08:43:34 Quit Galois (Remote host closed the connection)
08:44:35 Join Galois [0] (
08:47:15 Quit traveler (Quit: Page closed)
08:50:36 Join mischasworld [0] (
08:57:55 Join The_Pwny [0] (
09:06:11 Quit mischasworld (Ping timeout: 240 seconds)
09:08:14 Quit mortalscan (Ping timeout: 260 seconds)
09:48:11 Quit kadoban (Ping timeout: 240 seconds)
10:13:50 Join The_Pwny_ [0] (
10:13:51 Quit The_Pwny_ (Client Quit)
10:14:24 Quit The_Pwny (Ping timeout: 255 seconds)
10:15:03 Join stoffel [0] (
10:37:49 Join T44 [0] (
10:37:49 Quit T44 (Client Quit)
10:40:56***Saving seen data "./dancer.seen"
10:43:51 Join kevku [0] (~kevku@2001:7d0:0:f000::135d)
10:46:18 Join T44 [0] (
10:48:11 Quit Topy44 (Ping timeout: 240 seconds)
10:52:24 Join pamaury [0] (
10:52:24 Quit pamaury (Changing host)
10:52:24 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
10:53:36 Join n1s [0] (
10:53:37 Quit n1s (Changing host)
10:53:37 Join n1s [0] (~n1s@rockbox/developer/n1s)
10:54:29 Quit factor (Ping timeout: 240 seconds)
11:16:55 Quit utanapischti (Ping timeout: 265 seconds)
11:25:54 Quit ved (Ping timeout: 260 seconds)
11:44:25 Quit n1s (Quit: Lämnar)
11:49:42 Join JdGordon| [0] (~jonno@rockbox/developer/JdGordon)
11:54:24 Join Horscht [0] (~Horschti@xbmc/user/horscht)
11:55:06 Quit pamaury (Ping timeout: 265 seconds)
11:57:20 Quit Horschti (Ping timeout: 260 seconds)
12:05:41 Join webguest32 [0] (
12:06:15webguest32is anyone here?
12:07:51Bushmillsonly one webguest32 and 118 bots
12:10:57 Join JdGordon1 [0] (
12:12:02 Quit JdGordon| (Ping timeout: 240 seconds)
12:12:46 Quit webguest32 (Quit: CGI:IRC (EOF))
12:15:45 Quit T44 (Quit: Leaving)
12:15:54 Join Topy44 [0] (
12:18:55 Quit JdGordon1 (Ping timeout: 260 seconds)
12:21:14 Join JdGordon| [0] (
12:21:14 Quit JdGordon| (Changing host)
12:21:14 Join JdGordon| [0] (~jonno@rockbox/developer/JdGordon)
12:29:34 Quit kevku (Read error: Operation timed out)
12:35:15 Join Buschel [0] (
12:36:35 Join kevku [0] (~kevku@2001:7d0:0:f000::135d)
12:36:47 Join JdGord [0] (~jonno@
12:39:19Buschelsoap: i guess you are enjoying x-mas time? ;) if you should proceed with the tests on your nano, could you do me one favor? -> perform test_fps with boosted (80MHz) / unboosted (30MHz) on your nano? the numbers on the wiki are outdated.
12:41:00***Saving seen data "./dancer.seen"
12:50:23 Join thomasjfox [0] (
12:51:21 Join ved [0] (
12:54:35 Quit Buschel (Ping timeout: 260 seconds)
12:57:29 Quit JdGordon| (Remote host closed the connection)
13:04:50 Quit thomasjfox (Ping timeout: 272 seconds)
13:07:24 Join JdGordon| [0] (~jonno@rockbox/developer/JdGordon)
13:07:25 Join thomasjfox [0] (
13:11:58Dreamxtremewhen is 3.8 due?
13:14:01 Join DerPapst [0] (
13:16:13 Join Buschel [0] (
13:18:39 Quit Dreamxtreme (Ping timeout: 265 seconds)
13:19:25 Quit Buschel (Read error: Connection reset by peer)
13:22:48 Join sasquatch [0] (~username@
13:22:55 Join benedikt93 [0] (~benedikt9@unaffiliated/benedikt93)
13:25:34 Quit thomasjfox (Remote host closed the connection)
13:25:41 Join n1s [0] (~n1s@rockbox/developer/n1s)
13:26:33 Join Dreamxtreme [0] (~Dre@
13:30:34 Join kugel [0] (~kugel@rockbox/developer/kugel)
13:31:15 Quit sasquatch (Ping timeout: 260 seconds)
13:32:03 Join sasquatch [0] (
13:32:42 Quit B4gder (Ping timeout: 272 seconds)
13:38:53 Join B4gder [0] (~daniel@rockbox/developer/bagder)
13:41:24 Quit mystica555_ (Ping timeout: 240 seconds)
13:43:39 Quit B4gder (Ping timeout: 250 seconds)
13:44:10 Join JdGordy [0] (~jonno@
13:44:11 Quit JdGord (Read error: Connection reset by peer)
13:46:19 Join B4gder [0] (~daniel@rockbox/developer/bagder)
13:46:41 Quit bluebroth3r (Ping timeout: 265 seconds)
13:48:06 Join wodz [0] (
13:48:18 Join bluebrother [0] (
13:48:18 Quit bluebrother (Changing host)
13:48:18 Join bluebrother [0] (~dom@rockbox/developer/bluebrother)
13:55:44 Quit sasquatch (Quit: ban xmas!)
14:06:40 Quit [Saint] (Quit: Happy Generic Holiday Period from Hayden and Karmin... 0/)
14:08:04 Join [Saint] [0] (S_a_i_n_t@
14:14:02wodzJdGordy: would you find 10minutes to close FS #11831 ?
14:16:52JdGordyI thought I did?
14:16:57JdGordyThat's the tx one?
14:19:11wodznop - %Rr, %Rh, %Rn, %Rs not documented
14:19:57JdGordyI don't do documentation ;)
14:20:00 Quit n1s (Quit: Lämnar)
14:20:49wodzI noticed this :-)
14:21:36 Join n1s [0] (
14:21:36 Quit n1s (Changing host)
14:21:36 Join n1s [0] (~n1s@rockbox/developer/n1s)
14:23:15wodzJdGordy: what do you think about this:
14:26:58JdGordyWhat about it?
14:27:21JdGordyI'm actually not sure tbe whole thing loaded... ill lokk when I get too a pc
14:27:29JdGordyI'm my phone currenttly
14:28:07wodzJdGordy: nothing urgent - this is modification of cabbiev2.fms based on your suggestions
14:29:48 Quit kugel (Ping timeout: 240 seconds)
14:41:01***Saving seen data "./dancer.seen"
14:42:02 Quit JdGordon| (Remote host closed the connection)
14:42:19 Join JdGordon| [0] (
14:42:19 Quit JdGordon| (Changing host)
14:42:19 Join JdGordon| [0] (~jonno@rockbox/developer/JdGordon)
14:43:27 Quit [Saint] (Quit: Won't you yake me to Funky-Town? (I've heard there's less earthquakes...))
14:57:28 Quit Guest52307 (Read error: Operation timed out)
14:59:29 Join factor [0] (
14:59:38 Quit JdGordon| (Read error: Connection reset by peer)
15:00:04 Join JdGordon| [0] (
15:00:05 Quit JdGordon| (Changing host)
15:00:05 Join JdGordon| [0] (~jonno@rockbox/developer/JdGordon)
15:19:48 Quit wodz (Ping timeout: 240 seconds)
15:34:13 Quit benedikt93 (Quit: Bye ;))
15:44:06 Join mortalscan [0] (~mortalsca@
15:47:59 Join kugel [0] (~kugel@rockbox/developer/kugel)
15:48:19 Join bmbl [0] (
15:48:20 Quit bmbl (Changing host)
15:48:20 Join bmbl [0] (~bmbl@unaffiliated/bmbl)
15:49:13kugel \o/
15:52:26kugelhm, he hasn't pushed his work yet
15:55:11 Join Feisar [0] (
15:55:36 Nick Feisar is now known as Guest61093 (
16:06:15 Quit JdGordon| (Read error: Connection reset by peer)
16:06:41 Join JdGordon| [0] (
16:06:42 Quit JdGordon| (Changing host)
16:06:42 Join JdGordon| [0] (~jonno@rockbox/developer/JdGordon)
16:18:24 Join sasquatch [0] (~username@
16:29:50 Quit JdGordy (Ping timeout: 240 seconds)
16:33:17 Quit JdGordon| (Ping timeout: 276 seconds)
16:41:02***Saving seen data "./dancer.seen"
16:43:25kugelFS #11834 makes updates a lot faster indeed
16:48:57 Join LambdaCalculus37 [0] (
16:48:58 Quit LambdaCalculus37 (Changing host)
16:48:58 Join LambdaCalculus37 [0] (~rmenes@rockbox/staff/LambdaCalculus37)
16:52:47 Quit evilnick (Ping timeout: 265 seconds)
16:58:06 Quit LambdaCalculus37 (Quit: This computer has gone to sleep)
17:04:44 Join LambdaCalculus37 [0] (~rmenes@rockbox/staff/LambdaCalculus37)
17:11:01 Quit kugel (Ping timeout: 240 seconds)
17:22:40 Join kugel [0] (~kugel@
17:22:41 Quit kugel (Changing host)
17:22:41 Join kugel [0] (~kugel@rockbox/developer/kugel)
17:29:17 Join thomasjfox [0] (
17:31:22 Quit kugel (Ping timeout: 240 seconds)
17:32:47 Join froggyman [0] (~seth@
17:32:47 Quit froggyman (Changing host)
17:32:47 Join froggyman [0] (~seth@unaffiliated/froggyman)
17:43:56 Quit TheSeven (Ping timeout: 260 seconds)
17:47:51 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
17:50:29 Join kadoban [0] (
17:51:04 Join kugel [0] (~kugel@
17:51:05 Quit kugel (Changing host)
17:51:05 Join kugel [0] (~kugel@rockbox/developer/kugel)
17:52:33thomasjfoxHi kugel
17:52:57thomasjfoxThe maemo port is now living here: git://
17:54:31thomasjfoxIt already contains an own application target and a hosted/target/maemo subdir
17:55:05 Join fyrestorm [0] (
17:56:09kugelnice, will check it out later
17:58:03thomasjfoxI'll send an email to rockbox-dev as not everybody follows IRC ;)
18:00:23kugelactually, everyone follows irc and many don't follow the ml :-)
18:01:38gevaertskugel: I doubt that
18:01:48gevaertsIRC+ML should catch everyone though
18:06:45 Join ppawel [0] (
18:16:53 Quit stoffel (Read error: Operation timed out)
18:17:54 Join evilnick [0] (
18:21:02 Quit kugel (Ping timeout: 240 seconds)
18:21:50 Join kugel [0] (~kugel@
18:21:51 Quit kugel (Changing host)
18:21:51 Join kugel [0] (~kugel@rockbox/developer/kugel)
18:22:39 Join JesusFreak316 [0] (
18:29:01 Quit shai (Ping timeout: 240 seconds)
18:29:20 Join Strife89TX [0] (~cstrife89@
18:31:05 Quit LambdaCalculus37 (Quit: This computer has gone to sleep)
18:32:04 Join shai [0] (
18:33:15Strife89TXDid the Clip+ gain USB support before or after 3.7.1? (Wiki says it has support)
18:34:25pixelmaafaik SVN doesn't have USB support for the Clip+ yet but there is a patch in the tracker. It seems to be troublesome on Windows still though, IIRC
18:35:10 Join mystica555_ [0] (
18:36:08gevaertsIt's a bit more complicated than that
18:36:51gevaertsThere is USB support (disabled) in svn, but it's not stable. The patch on the tracker works around some of the instabilities (by retrying things that should have worked in the first place)
18:37:24gevaertsSo the patch doesn't actually *fix* the issues, which I understand is mainly why it's still a patch
18:39:33 Join Keripo [0] (
18:41:06***Saving seen data "./dancer.seen"
18:52:36 Quit n1s (Quit: Lämnar)
19:01:15 Join wodz [0] (
19:09:17 Join WonTu [0] (
19:09:31 Part WonTu
19:13:11 Quit kugel (Ping timeout: 260 seconds)
19:14:47 Join LambdaCalculus37 [0] (
19:14:47 Quit LambdaCalculus37 (Changing host)
19:14:47 Join LambdaCalculus37 [0] (~rmenes@rockbox/staff/LambdaCalculus37)
19:18:37 Quit Strife89TX (Ping timeout: 240 seconds)
19:19:57 Quit Guest61093 (Read error: Operation timed out)
19:23:25 Quit JesusFreak316 (Ping timeout: 240 seconds)
19:28:08 Join kugel [0] (~kugel@rockbox/developer/kugel)
19:28:22kugelthomasjfox: I am looking at your patch right now
19:28:45 Quit wodz (Ping timeout: 240 seconds)
19:30:17kugelmaemo_tell_rockbox_to_stop_audio() is hacky like hell :) perhaps you can do it like the android port does? see telephony-android.c
19:31:00kugelbtw, we dislike //-style comments
19:31:37thomasjfoxThe comments will be changed next
19:33:31thomasjfoxI just pushed another change which will be good for upstream
19:33:47thomasjfoxCan you take a look if this is a valid thing to do (include stdio.h)?
19:35:24kugelyea, should be right
19:35:54thomasjfoxok, I'll open a FS ticket
19:36:19thomasjfoxNow it also compiles for old maemo v4 devices
19:36:30thomasjfoxUnfortunately without arm assembler optimizations
19:36:58kugelno need for an FS ticket
19:37:25thomasjfoxWhat's the best for such small changes?
19:38:13thomasjfoxDo you know the ldmpc arm assembler instruction? Seems not available on old mameo devices
19:38:36kugelFS is generally the best, but for small changes and if you can poke a developer earler, pastebin or a git pull is ok
19:38:41thomasjfoxOr at least not available with some special compiler flag
19:38:49kugelldmpc is a macro IIRC
19:39:27kugelIIRC funman added it to enable thumb compilation
19:40:44 Quit kadoban (Ping timeout: 250 seconds)
19:40:53Torneyes, it's to allow the same code to be built with arm or thumb, even on ARMv4T where loading to PC doesn't set the T bit
19:41:10TorneARMv4T you have to pop into another register then bx to it to do a proper interworking return
19:41:29thomasjfoxOk, found the macro. Thanks
19:41:38Torneis it not defined already?
19:41:48Torneit jst needs to do the regular ldm on anything maemo runs on
19:41:53TorneARMv4T is old
19:42:27thomasjfoxI think the macro expansion is somehow not working
19:42:36thomasjfoxI'll check the #defines first
19:43:01Torneit may not be defined for ARMv6
19:43:14Torneor ARMv7
19:43:16kugelthomasjfox: what toolchain is used for maemo?
19:43:16Tornewhichev er your device is
19:43:40CIA-7New commit by kugel (r28892): Add stdio.h include for SEEK_SET define to various files. ...
19:44:09thomasjfoxkugel: To be honest, I didn't look under the hood yet. The complete toolchain is called "scratchbox".
19:44:22kugelah ok
19:45:01thomasjfoxCompilation for the N900/ARMv7 works. The old N8xx uses an ARMv6 and fails.
19:45:47kugelI think ldmpc is an assembler macro. perhaps the maemo4 scratchbox assember doesn't support it?
19:45:48CIA-7r28892 build result: All green
19:46:11Tornethey all use GNU as, so macros should be fine
19:46:13thomasjfoxEither that or the __ASSEMBLER__ define is not set
19:46:24gevaertskugel: binutils and gcc 4.2.1 for maemo5
19:47:35thomasjfoxbinutils 2.16.91 20051213 for maemo4 (failing)
19:51:44 Quit TheSeven (Read error: Connection reset by peer)
19:51:48 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
19:57:39gevaerts"warning: remote HEAD refers to nonexistent ref, unable to checkout"
19:58:57 Quit LambdaCalculus37 (Ping timeout: 265 seconds)
19:59:59thomasjfoxgevaerts: Was that for me? ;)
20:00:13gevaertsthomasjfox: I get that when cloning your repository, yes
20:00:21thomasjfoxLet me have a look
20:00:59thomasjfoxOk, I'll fix it
20:01:07kugelgevaerts: you can add it as a remote to an existing clone
20:01:58gevaertskugel: yes, but that assumes an existing clone
20:02:08kugeli thought you have one
20:02:27gevaertsNot anywhere near my scratchbox setup
20:04:08thomasjfoxThe problem is that my local HEAD points to origin/master
20:04:22thomasjfoxI have to find out how to fix HEAD on
20:05:59thomasjfoxbtw: Like is suspected, maemo v4 doesn't define __ASSEMBLER__
20:06:30*kugel remembers something about __ASSEMBLER__ being broken
20:07:23thomasjfoxThe linux kernel uses in some places __ASSEMBLY__. I'm still looking up the difference
20:09:55thomasjfoxgevaerts: It should work anyway: git checkout -b maemo-port origin/maemo-port
20:11:15kugelapps/codecs/demac/libdemac/predictor-cf.S has some work around
20:12:27kugelbut it's an .S file so it knows it's only touched by as
20:14:18kugelthomasjfox: if binutils is from 2005, then gcc probably also, then it's probably <4.0 which apparently has the __ASSEMBLER__ bug
20:15:25saratogais the N900 Meego or whatever they call it now?
20:17:02kugelthomasjfox: perhaps it's possible to use newer versions (or even the maemo5 scratchbox) for maemo4
20:17:35 Join Feisar [0] (
20:17:37thomasjfoxkugel: The maemo5 scratchbox lets me switch into maemo4 mode
20:17:52thomasjfoxSo I'm able to test compilation at least somehow
20:17:58kugelah good
20:18:01 Nick Feisar is now known as Guest27198 (
20:18:20 Join JesusFreak316 [0] (
20:23:08 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
20:25:11thomasjfoxgevaerts: git repo is fixed now. I didn't find a way to update .git/HEAD on a remote server. Until I found the "default branch" setting on :)
20:25:34gevaertsthomasjfox: I seem to have working branch now
20:30:12thomasjfoxkugel: Do you know the reason for the __ASSEMBLER__ bug?
20:30:31thomasjfox"touch test.S; cpp -dM test.S" shows __ASSEMBLER__
20:30:59thomasjfoxMaybe it only happens if the config.h file gets included. I'll try that next
20:31:02kugelbut it seems as 2.16 and/or gcc 3.4.x have it
20:36:45 Join krabador [0] (
20:37:12*gevaerts thinks that cpu usage still looks high compared to what he heard about android
20:38:18*thomasjfox thinks it's really time to use native ARM thread code instead of the SDL thread stuff
20:38:48thomasjfoxThe SDL thread code is -very- expensive as most threads are busy-waiting on the main lock
20:39:12gevaertsyes, probably worth a try
20:39:20thomasjfoxThough the current CPU usage is only ~8% above the native Nokia player
20:39:24thomasjfoxWould be nice to beat it :)
20:39:32kugelyou should be able to just copy thread-android-arm.c and kernel-sdl.c from my ucontext_thread branch (and change thread.c to include it)
20:39:39thomasjfoxOh, try to switch of the display / change to another application
20:39:40gevaertsAndroid people report things like 6% CPU usage during playback
20:39:47thomasjfoxThat turns of the SDL output
20:40:03thomasjfoxkugel: I'll try that later on
20:40:20gevaertsogg doesn't work apparently
20:40:40thomasjfoxDid'nt try ogg yet
20:41:10***Saving seen data "./dancer.seen"
20:42:08thomasjfoxCan I easily turn on the complete compiler invocation line when I build rockbox?
20:42:41kugelmake V=1
20:43:11thomasjfoxhehe, tried with VERBOSE=1 ;)
20:44:20gevaertsit's apparently still possible to trigger small audio glitches by turning the LCD on and off if rockbox is in the background
20:44:31gevaertsMaybe getting the CPU usage lower would help there
20:44:47*gevaerts also doesn't understand why pulseaudio needs so much CPU
20:44:56thomasjfoxgevaerts: Did it properly detect a N900 during configure?
20:45:08gevaertsOh, well, it still runs a lot better than it used to :)
20:45:12thomasjfoxok, so it's using the gstreamer backend
20:45:48gevaertsthomasjfox: I don't get audio glitches any more with rockbox in the foreground, so it's definitely a lot better
20:46:24thomasjfoxImproving the buffer size should help here: Less glitches and lower pulseaudio CPU usage
20:46:40gevaertsYes, but that comes at a cost too
20:46:55thomasjfoxLet's see how this is going to work out with native ARM threads
20:47:40thomasjfoxRunning ~12 threads is also expensive. Also I don't think the SDL thread code sets thread priorities (does it?)
20:48:22gevaertsI think it doesn't
20:48:49gevaertsAnd if threads busy-wait (or nearly so), that definitely won't help
20:49:13gevaertsSo yes, different threading code is a good first thing to try next
20:49:43thomasjfoxbtw: The maemo port will make a pandora port quite easy
20:50:05thomasjfoxBoth of them share almost the same "requirements"
20:50:13gevaertsA pandora app-port :)
20:50:22thomasjfoxNow we just need to get a Pandora...
20:50:22gevaertsSome people also want a pandora native port
20:52:19thomasjfoxArgh, the __ASSEMBLER__ gets killed by "-std=gnu99"
20:54:07 Quit shai (Ping timeout: 240 seconds)
20:55:07 Join shai [0] (
20:56:58thomasjfoxToo bad funman is not here:
20:57:20thomasjfoxWould have saved me the debugging :)
20:57:53kugelthomasjfox: look what I said at 20:11 :)
20:58:22thomasjfoxYes, exactly. It just didn't came to my mind that we can add this to all the .S files and be done with it
20:58:58kugelwhich was my next sentence :P
20:59:22thomasjfox122 .S files...
20:59:53kugelperhaps makefiles can solve this for s
21:00:24thomasjfoxthere's the -Wa option for gcc
21:00:39thomasjfoxPass additional arguments to the assembler
21:00:56thomasjfoxlike −−defsym
21:01:03kugelthe assember is after the preprocessor
21:01:32 Join wodz [0] (
21:01:49*gevaerts decides to leave this merging of thread code to people who know what they're doing
21:02:02kugelperhaps you can add something to the %.S rule in tools/root.make (line 344+)
21:03:46wodzis archos player USB cable crossed or straight trough?
21:05:05wodzhmm crossed doesn't make much sense in USB I guess
21:07:39wodzwhat does bitclr32(&CCU_IO, 1<<12); in Fuze v2 fm radio code?
21:09:24kugelgevaerts: *could* work
21:11:11gevaertshm, how do you undo local modifications again with git?
21:12:11wodzgit reset −−hard (if not commited)
21:13:17 Quit anewuser ()
21:13:30wodzthis will trash all not commited modifications
21:13:50gevaertskugel: only 56 undefined references :)
21:13:56wodzyou can also undo per file basis by git checkout −− /some/file.c
21:14:08*gevaerts decides to try a make clean
21:15:31kugelah well it probably needs some more changes
21:16:58gevaertshm, does this need your android bits as well?
21:18:40kugelnext try
21:18:45 Join dantje [0] (
21:21:32gevaertshm, where is PLATFORM set to PLATFORM_MAEMO?
21:22:42 Quit thomasjfox (Ping timeout: 240 seconds)
21:24:03kugelanother try:
21:29:53gevaertskugel: thread.c was missing from SOURCES for maemo
21:30:36gevaertsNow it's still missing stackbegin and stackend though
21:32:38CIA-7New commit by alle (r28893): Minor fixes in the disktidy description in the manual
21:35:50 Quit ppawel (Quit: Leaving)
21:35:51kugelgevaerts: in the last one is an error in SOURCES
21:35:57gevaertsyes, I noticed
21:37:22gevaertsThere's also a problem in firmware/target/hosted/maemo/pcm-gstreamer.c which uses the SDL mutex. I removed the locking there for now...
21:38:13kugelit's not needed anyway, queue_post is interrupt safe
21:38:25gevaertsIt should be
21:41:53 Join thomasjfox [0] (
21:42:47gevaertsOK, it builds, but it gets stuck on the splash screen
21:43:14thomasjfoxMy box just freezed while I was away
21:43:22thomasjfoxgevaerts: Does it consume CPU?
21:43:51gevaertsSo that's good :)
21:43:56thomasjfoxDo you have gdb on your n900?
21:44:18gevaertsnot right now. Apparently I removed it
21:44:57thomasjfoxapt-get install gdb ;)
21:45:28thomasjfoxThen you can see where it's hanging
21:45:35gevaertspthread_cond_wait() from SDL_CondWait()
21:45:41 Quit fyrestorm (Quit: Ur skills' fireproof like a wooden panel -- U got feds talking leet on your IRC channel!)
21:46:17gevaertsfrom wait_for_interrupt() in kernel_sdl.c
21:46:30*kugel looks
21:48:09kugelthe sdl app works with it
21:48:21kugel(in my ucontext_thread branch)
21:49:02thomasjfoxI'll give it a go in a second. Where's the URL? (my box crashed...)
21:49:47kugelhm, no idea
21:50:24 Join benedikt93 [0] (~benedikt9@unaffiliated/benedikt93)
21:50:25 Join fml2 [0] (
21:51:26fml2kugel: hello. What is the correct way to get the rocks/app dir on the sim? In disktidy, PLUGIN_DIR is used which is not correct on the sim.
21:51:42fml2kugel: is there a plugin API function for that?
21:51:46kugelPLUGIN_DIR is correct on the sim
21:52:20fml2kugel: ah, sorry, PLUGIN_APPS_DIR is used in the code
21:52:33kugelstill correct
21:52:48kugelthe I/O layer has the necessary magic to translate to real paths in the filesystem
21:53:20fml2kugel: for me, it's defined as /.rockbox/..... How can I bring it to the correct value?
21:53:32kugelas I said, it is correct
21:54:21fml2kugel: so it's tweaked in the low level funcs?
21:55:21kugelgevaerts: i don't know why it should freeze there
21:55:44kugelgevaerts: it looks like the corresponding SDL_CondSignal() calls is missing but I see it there (in sim_exit_irq_handler())
21:56:47 Join Bas [0] (~Bas_82@
21:56:54fml2kugel: it doesn't seem to work for me
21:56:57BasHello veeryone
21:57:04BasI mean everyone
21:58:26thomasjfoxkugel: I've fixed the __ASSEMBLER__ issue in the Makefiles. Works fine with gcc 3.x and 4.2.1.
21:58:37 Quit froggyman (Ping timeout: 240 seconds)
22:00:21 Join Buschel [0] (
22:00:37 Quit mystica555_ (Ping timeout: 240 seconds)
22:00:53Buschelkugel: did you see FS #11832 ?
22:01:09CIA-7New commit by alle (r28894): Extract config file saving code to a function; Do not write the NUL character to the text config file
22:01:28kugelBuschel: yes, now :)
22:01:50Buschelkugel: can reproduce this as well (using r28893)
22:02:44kugelthat's strange, the lstat call is within an ifdef which shouldn't be true on cygwin
22:03:04CIA-7r28894 build result: All green
22:03:23kugelah, /me slaps forehead
22:03:49kugelBuschel: try s/ifdef/if/ in uisimulator/common/io.c please
22:04:10kugel(i.e. #ifdef HAVE_LSTAT => #if HAVE_LSTAT)
22:04:25*Buschel tries
22:04:39kugelbut that doesn't fix localtime_R
22:08:45kugel has a working implementation
22:11:42CIA-7New commit by alle (r28895): Disktidy: introduce a local variable to avoid repeated use of the indexed expression
22:12:32 Quit krabador (Quit: Sto andando via)
22:13:24thomasjfoxgevaerts: I'm also running the new thread code now. I've seen "Updating database" once before it stalled.
22:13:27CIA-7r28895 build result: All green
22:13:54thomasjfoxkugel: Does the native arm thread code mix with SDL_Thread?
22:13:58gevaertsthomasjfox: so a race contidion or something like that?
22:14:09kugelthomasjfox: the assembler fix looks ok to me, but I would prefer it wrapped into a ASMFLAGS var
22:14:24 Quit Guest27198 (Ping timeout: 276 seconds)
22:14:33thomasjfoxkugel: Good idea
22:14:52thomasjfoxgevaerts: Didn't mange to reproduce it yet
22:15:02kugelthomasjfox: yes it mixes OK, it needs some tweaks here and there though. it is working in my ucontext_thread branch where I also use the rockbox thread implementation
22:18:43kugel(in the sdl app I mean=)
22:20:03thomasjfoxThe maemo "osso" thread is running, I just added debug output to the "display on/off callback"
22:20:13thomasjfoxIt's stalling in wait_for_interrupt(), too
22:20:36kugelthe osso thread shouldn't be calling wait_for_interrupt
22:20:46thomasjfoxno it's not
22:20:59thomasjfoxJust wanted to make sure that at least this thread is running
22:21:59thomasjfoxAnother thread is at sim_enter_irq_handler()
22:22:29kugelit appears the SDL_CondSignal in sim_exit_irq_handler() isn't working properly
22:23:28thomasjfoxis that function meant to be thread safe?
22:23:52thomasjfox"−−handlers_pending" is certainly not :)
22:25:02kugeli suspect interrupts are disabled
22:25:15kugelwhat's the value of interrupt_level?
22:25:48kugelthomasjfox: it's protected by a mutex so yes
22:25:58thomasjfoxOk, just found that out
22:26:08*thomasjfox is looking for interrupt_level
22:26:50kugelah yes
22:27:07kugelso interrupts are disabled, the handlers wait in enter_irq_handler, and the main thread waits for an interrupt :)
22:27:11kugelthat can't work
22:27:15Buschelkugel: with "#ifdef" in io.c i get another error message: make: *** No rule to make target `/cygdrive/c/development/rockbox_test/build-pcsim-nano2g/sys/statvfs.h', needed by `/cygdrive/c/development/rockbox_test/build-pcsim-nano2g/uisimulator/common/io.o'.
22:27:42 Join Greek_o_nikos [0] (
22:27:55kugelBuschel: you changed the wrong one I think
22:28:11kugel#ifdef HAVE_LSTAT should be changed to #if, the #if HAVE_STATVFS is correct
22:28:26 Part Greek_o_nikos ("Leaving.")
22:28:34kugelthomasjfox: the question is now why wait_for_interrupt() is called with interrupts disabled. that shouldn't happen
22:29:13Buschelsilly me, will try again
22:29:27thomasjfoxI'll have a look at the backtrace again
22:30:21thomasjfoxSo we can detect this error in wait_for_interrupt() and output a message?
22:30:29kugelthomasjfox: that would break on a real target also, on some targets wait_for_interrupt() is really a wfi instruction that sleeps the cpu until an extern interrupt comes in
22:30:41thomasjfox(for the sdl-app case)
22:30:51 Join Horschti [0] (~Horschti@xbmc/user/horscht)
22:30:55kugelyou could check for interrupt_level != 0, then panic :)
22:31:02thomasjfoxor in the maemo implementation
22:31:56CIA-7New commit by alle (r28896): disktidy: add a comment to the function
22:32:19kugelnot sure if it helps, but the CondSignal() should probably placed before the MutexUnlock()
22:32:21 Quit wodz (Ping timeout: 250 seconds)
22:32:32thomasjfoxBacktrace thread 1: scroll_thread() -> sleep() -> switch_thread() -> wait_for_interrupt()
22:33:07thomasjfoxBacktrace thread 2: tick_timer() -> sim_enter_irq_handler()
22:33:42kugeland tick_timer() gets stuck?
22:33:44CIA-7r28896 build result: All green
22:33:45thomasjfoxBacktrace thread 3: gui_message_loop() -> sim_enter_irq_handler()
22:33:48 Quit Horscht (Ping timeout: 240 seconds)
22:35:09 Join Buschel_ [0] (
22:35:11thomasjfoxlooks like. If I let the execution continue and interrupt again, the backtrace is still at the same position
22:35:29 Quit Buschel (Ping timeout: 265 seconds)
22:35:36 Nick Buschel_ is now known as Buschel (
22:35:56kugelwell, try to return in wait_for_interrupt if interrupt_level != 0, but I maintain that shouldn't happen in the first place
22:36:36kugelthe scheduler explicitely enables interrupts before calling core_sleep()
22:37:04kugelah no, it doesn't
22:37:31kugelthere we got it
22:37:41kugeladd enable_irq() to core_sleep()
22:38:25 Join GeekShad0w [0] (~Antoine@
22:38:27kugel(my ucontext_thread implementation has it, but I forgout about it
22:38:46CIA-7New commit by alle (r28897): disktidy: fix the situation where specifying that a dir should be deleted could lead to removing a file with that name
22:40:30CIA-7r28897 build result: All green
22:40:57thomasjfoxkugel: Something seems wrong with the way we include the file in thread.c: make didn't detect I changed the file
22:41:02thomasjfoxDoing make clean now
22:41:11***Saving seen data "./dancer.seen"
22:41:14gevaertskugel: seems to work
22:41:19kugelthomasjfox: you need make dep for #include changes
22:41:33kugelwe build dependency information as a seperate step
22:41:43gevaertsAlthough it seems not to help CPU usage much
22:43:24thomasjfoxI'll do an oprofile run
22:44:00gevaertsStill 20% on 250MHz for mp3
22:44:05 Join [Saint] [0] (S_a_i_n_t@
22:44:22 Join froggyman [0] (~seth@unaffiliated/froggyman)
22:44:25kugelgevaerts: that's what I also get on my phone
22:44:37gevaertsNot overly bad then :)
22:44:51*gevaerts tries to find how to stop playback
22:45:25thomasjfoxTab in the middle of the screen
22:45:37thomasjfoxUpper-left corner gets you in the menu
22:46:00kugelah yes, this horrible grid mode :)
22:46:18thomasjfoxpowertop outputs 47,4% in C2 state
22:46:33thomasjfoxI'll just retry this with SDL threads
22:46:49gevaertsIdle in the main menu is about 2.5% (at 250MHz), which drops to 0.5 to 1.5% with LCD off
22:46:58Buschelkugel: now −− as you said −− _localtime_r is the problem left...
22:47:04kugelgevaerts: that's not too bad
22:47:53kugelBuschel: the link I posted has an implementation
22:49:07kugelgevaerts: that's what the 1GHz phones on android also have, only mine seems to go even lower
22:49:09thomasjfoxOld code is only 29% in C2 state. So the new thread code gave a boost of ~17%. Yeah!
22:49:27gevaertskugel: the problem is that during playback, pulseaudio uses as much or more CPU as rockbox
22:49:56thomasjfoxgevaerts: I could try to do an alsa backend without gstreamer
22:50:22kugeli seem to have upto 30% cpu (at 246MHz) during decoding
22:50:32thomasjfoxI'm just not sure if this will break "libplayback" as it depends on the pulseaudio stream name
22:50:33gevaertsthomasjfox: that might work, but it doesn't seem very clean
22:50:53kugelwhat's bad about alsa?
22:51:01gevaertsAs such, nothing
22:51:12kugelI made an half-working alsa pcm implementation as gsoc preperation
22:51:16gevaertsBut the goal is to integrate this in the environment
22:51:37gevaertsAnd then direct alsa may break some assumptions
22:51:51kugelbut alsa is part of the environment, isn't it?
22:51:55thomasjfoxYes, giving it some more thought, it will break audio playback if the silent profile is active
22:52:21kugelalsa will go through pulseaudio too I think
22:52:26thomasjfoxLet see if it still deadlocks on incoming calls with the new thread code
22:52:41gevaertsright, probably
22:52:56gevaertsI know the "old" sdl code went through pulseaudio, with similar CPU usage
22:53:03kugelat least that's what I found when I played with it (IIRC)
22:54:26CIA-7New commit by alle (r28898): disktidy: add a debug message if a file cannot be deleted
22:56:24CIA-7r28898 build result: All green
22:56:27thomasjfoxkugel: Do you think it's safe to keep the new thread code in there?
22:56:44 Quit benedikt93 (Quit: Bye ;))
22:57:03Buschelkugel: seems to compile now. what is the right way for #ifdef'ing localtime_r() ?
22:57:03kugelmy patch was rather proof of concept, taking it as is would add even more hacks
22:57:44kugelBuschel: #ifdefing the call or implementation? for the implementation #ifdef WIN32 I'd say
22:58:52 Join anewuser [0] (anewuser@unaffiliated/anewuser)
22:59:42kugelthomasjfox: feel free to take it though, you need do to a lot of cleanup anyway before we can push it into rockbox svn :)
23:04:05thomasjfoxkugel: Can you think of a better way to do the maemo_tell_rockbox_to_stop_audio() implementation?
23:04:10thomasjfoxThe signal can arrive at any time
23:04:25kugelsee telephony-android.c
23:04:29thomasjfoxYou said I should look at the android implementation
23:04:56kugelqueue_* is interrupt safe so you don't need to care of where the call comes from
23:04:59thomasjfoxIs queue_broadcast() "thread safe"?
23:05:03thomasjfoxah ok
23:05:23kugelyou should maybe add sim_enter_irq_handler() though to play nice with the other handlers
23:05:34thomasjfoxI added the dirty hack in there as it was stalling on incoming calls
23:05:40kugel(this isn't done on android as it doesn't do the interupt handler simulation
23:06:49kugelthe incoming call handler on android can also come at any time
23:07:33kugelit even has a means to enable playback again when the call is hung up, which I quite enjoy
23:08:14 Quit shai (Ping timeout: 240 seconds)
23:09:10thomasjfoxmaemo doesn't notify us on hang up unfortunately
23:09:45thomasjfoxThough I noticed it calls a certain callback after the call ;)
23:09:53 Join shai [0] (
23:10:06Buschelkugel: FS #11832 updated with a patch -> can you make a short review?
23:10:41 Join Feisar [0] (
23:11:07 Nick Feisar is now known as Guest85915 (
23:11:16kugelBuschel: looks good to me, maybe make it static? otherwise gcc will both inline and emit the body
23:12:19kugelbtw, I never knew but a lot people seem to prefer memcpy over simple assignments when copying structs. is there any reason for it?
23:12:23gevaertsthomasjfox: one workaround for the lack of native theme might be to rotate the display and steal the android 480x800 theme
23:13:20thomasjfoxwouldn't that suck with the keyboard?
23:13:38Buschelkugel: done. shall I submit?
23:13:46kugelgo for it
23:13:48gevaertsa bit, yes :)
23:14:22kugelnot having cabbie sucks more doesn't it? :)
23:14:23thomasjfoxI'm sure once there's the first .deb package available, someone will create a nice looking theme
23:15:01kugelthomasjfox: I'm not sure about that. the grid mode generally seems to chase people away
23:15:27gevaertskugel: structure assignment won't work with old pre-C99 compilers (possibly pre-C89, can't remember), and people keep old habits. Moreover, a structure assignment will probably be implemented as memcpy() by the compiler anyway
23:15:49B4gderstruct assignment works fine in C89
23:15:51thomasjfoxkugel: :o)
23:16:11gevaertsok, pre-C89 then :)
23:16:53 Quit bmbl (Quit: Verlassend)
23:17:04CIA-7New commit by Buschel (r28899): Make sim build compilable under Cygwin. Fixes FS #11832.
23:17:20gevaertsStill, quite recent!
23:18:15*B4gder admits not learning C until after 1989
23:18:30gevaertsB4gder: me too, but what did the compilers understand?
23:18:44CIA-7r28899 build result: All green
23:19:01B4gderyeah, lots of K&R those days
23:22:52CIA-7New commit by alle (r28900): disktidy: Fix deleting of directories specified by a pattern (FS #11827)
23:23:43kugelgevaerts: a quick test shows that gcc appears to inline memcpy in the assignment case
23:24:06kugelit possibly also doesn't copy padding bytes
23:24:38gevaertsI'd expect not copying padding bytes would be quite expensive
23:24:45CIA-7r28900 build result: All green
23:25:13kugelgevaerts: ah no, I forgot my machine is 64 bit :)
23:25:44kugelarm-gcc also inlines memcpy
23:26:08kugelso I'd think assignment is faster but also bigger
23:26:30B4gderyeah, and the inline version is often (much) slower...
23:26:53B4gderzagor uses to post a link to the gcc bug entry about it...
23:27:14B4gderat least slower compared to glibc I believe
23:27:41gevaertsAh, wait, that's memcmp, not memcpu
23:29:57 Quit fml2 (Quit: ChatZilla 0.9.86 [Firefox 3.6.13/20101206122310])
23:32:34 Join Judas_PhD [0] (
23:34:24kugelthomasjfox: audio_stop() isn't broken, it calls queue_send() which is the blocking version of queue_post() and isn't interrupt safe
23:34:57thomasjfoxkugel: ok
23:35:03thomasjfoxIt still stalls after incoming calls
23:35:17thomasjfoxThe audio gets properly stopped, pcm_dma_pause(1) is called
23:35:36thomasjfoxI remember it being stuck in and endless loop
23:35:42thomasjfoxLet me find the location...
23:35:56thomasjfoxGoogling brought up similar results on the forum IIRC
23:36:07kugelit could be a bug in the pcm part
23:36:37thomasjfoxit was something like: while (no free space in the dma buffer) wait_for_it()
23:37:34 Quit kevku (Ping timeout: 272 seconds)
23:38:15amiconngevaerts: Btw, the N900 on-target toolchain has a bug in 'gas' that makes it crash on certain neon instructions. I didn't check yet whether scratchbox has the same problem; I assume it's okay since thomasjfox' patch already defines ARMv7 for N900?
23:38:50*amiconn noticed it when testing the neon asm building the standalone 'demac' utility with hacked-in #define for armv7
23:39:19thomasjfoxamiconn: Did you define the -mfpu stuff?
23:39:58thomasjfoxGCCOPTS="$GCCOPTS -mfloat-abi=softfp -mfpu=neon" to be precise
23:41:02amiconnI didn't change gcc options at all; the problem I ran into was related to inline asm
23:42:30thomasjfoxYes, you need supply these options to gcc
23:42:42thomasjfoxOtherwise it fails on the neon instructions
23:43:00amiconnIt doesn't fail on all neon instructions, only certain ones
23:43:20amiconnIirc the problem is fixed in later binutils, it's just that the maemo binary package is horribly outdated
23:44:37thomasjfoxkugel: The deadlock was in codec_pcmbuf_insert_callback(). It doesn't leave the first while loop.
23:45:20kugelI've never experienced a deadlock due to an incoming call
23:45:26thomasjfoxamiconn: So as it compiles fine now, we're lucky :)
23:45:26 Quit Judas_PhD (Quit: This is a quitting message)
23:45:51amiconnYeah, looks like scratchbox doesn't have that problem
23:45:53thomasjfoxkugel: So I guess there must be something wrong current pcm-gstreamer backend
23:46:26kugelthat's my suspicion yes
23:47:08 Quit GodEater (Read error: Operation timed out)
23:47:56 Join Judas_PhD [0] (
23:48:11kugelyou said it calls pcm_play_dma_pause? it's supposed to stop, not pause
23:48:41kugelpause/stop seems to be the same thing in your code though
23:49:07thomasjfoxthough it should still call pcm_play_dma_stop()
23:49:23thomasjfoxI just add enter/leave printfs to all pcm_* functions
23:49:32thomasjfoxSo I can see if gstreamer blocks somewhere
23:55:23thomasjfoxkugel: Like you suspected: It hangs in feed_data() and doesn't even leave after the pcm_play_dma_stop call()

Previous day | Next day