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

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

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

#rockbox log for 2009-10-09

00:00:33 Quit SUSaiyan (Read error: 104 (Connection reset by peer))
00:00:38 Join SUSaiyan [0] (
00:02:54 Join Thundercloud [0] (
00:04:18 Join enthdegree [0] (
00:04:54enthdegreeHow good does the guitar tuner work on the c200vI?
00:07:48 Quit evilnick ("Page closed")
00:09:00 Quit domonoky1 (Read error: 104 (Connection reset by peer))
00:15:06 Quit petur ("Zzzz")
00:16:47 Quit n1s ("Lämnar")
00:19:38CIA-43New commit by tomers (r23019): RTL: Fix context regarded as bit-field value, resulted in wrong contexts get ...
00:35:59 Quit LambdaCalculus37 ("Fwump")
00:40:40CIA-43New commit by tomers (r23020): RTL: Add support for DB init message
00:41:27***Saving seen data "./dancer.seen"
00:50:00 Quit GeekShadow ("The cake is a lie !")
00:52:27 Join mt [0] (n=MTee@rockbox/developer/mt)
00:54:39mtlinuxstb: Is ci->id3>offset maintained through resets ?
00:55:07kugelmt: you mean reboots?
00:55:22mtkugel : Yes :)
00:58:46kugelsome track info is stored in nvram.bin
00:59:26kugelinfo from that is loaded once you resume (maybe also when playing a different song)
01:00:27mtkugel: Great. Thank You !
01:01:04 Nick fxb is now known as fxb__ (
01:03:53mttomers: What's the problem with scrolling ? Sorry got busy last time and didn't have time to do anything beyond testing on a sim build.
01:04:18kugelsvn up
01:04:20mtI'm testing now on target, but it seems to me that scrolling is working as expected in the menus ?
01:04:46kugelwell, if you expect it to be inverted...
01:04:56mtkugel: No, I didn't mean there was actually a problem, I was asking because he told me last time to look into that.
01:05:01mtwell, yes.
01:05:16kugelshould be fine for you then
01:05:43tomersmt: Does scrolling in Arabic works? Do you have a long string that scrolls? Are you able to see the entire string (does all of it scrolls?). I am unable to do so...
01:06:23mtOh, That ! I thought you meant the scrollwheel.
01:07:27kugelI thought you mean that too!
01:08:34tomersmt: Wasted time :-)
01:08:50*tomers Going to bed. Good night!
01:08:56 Part tomers
01:09:04mtFortunately didn't have much time to waste ..
01:13:57 Quit bertrik (Read error: 113 (No route to host))
01:35:13 Join JdGordon [0] (n=jonno@rockbox/developer/JdGordon)
01:35:18 Quit Thundercloud (Remote closed the connection)
01:36:52enthdegreeHow good does the guitar tuner work on the c200vI?
01:40:48 Quit enthdegree ("Lost terminal")
01:46:09 Join chandoo [0] (
01:49:40 Join intrados [0] (
01:52:39 Quit JdGordon ("Leaving.")
01:52:53 Part toffe82
02:00:39 Join linuxguy4 [0] (
02:14:00 Quit linuxguy3 (Read error: 113 (No route to host))
02:18:22notlisteningLatest build just killed my e200v2 lots of prefetch aborts
02:21:42 Join pixelma_ [0] (i=quassel@rockbox/staff/pixelma)
02:21:42 Quit pixelma (Nick collision from services.)
02:22:01 Nick pixelma_ is now known as pixelma (i=quassel@rockbox/staff/pixelma)
02:22:27 Quit amiconn (Nick collision from services.)
02:22:30 Join amiconn_ [0] (i=quassel@rockbox/developer/amiconn)
02:22:49 Nick amiconn_ is now known as amiconn (i=quassel@rockbox/developer/amiconn)
02:29:59mtwhat is this line for ? It seems to break bidir scrolling (with certain files) in both RTL and LTR VP's : s->offset = s->width - (current_vp->width - xpos);
02:30:51mtline 324 in firmware/drivers/lcd-bitmap-common.c
02:34:38mtsorry, line 323
02:35:48 Quit kugel (Remote closed the connection)
02:41:30***Saving seen data "./dancer.seen"
02:46:41 Join evilnick [0] (n=evilnick@rockbox/staff/evilnick)
02:47:08 Quit efyx_ (Remote closed the connection)
02:56:40CIA-43New commit by mt (r23021): Handle text scrolling in RTL.
02:59:56 Quit bekabug ()
03:05:42CIA-43New commit by mt (r23022): Fix a bug in a bidirectional text scrolling; s->offset doesn't need to be set when backward scrolling starts. This fixes bidirectional scrolling in ...
03:07:14 Quit Rob2223 ()
03:07:56 Join Rob2222 [0] (
03:21:28 Quit Zarggg ()
03:21:33 Join clefurgey [0] (
03:22:00 Quit clefurgey (Client Quit)
03:25:36 Quit notlistening ("Leaving")
03:26:13 Quit snookums ()
03:41:22 Quit dfkt ("-= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.")
03:54:02 Quit chandoo ("Leaving")
04:01:01 Quit froggyman (Read error: 104 (Connection reset by peer))
04:06:08 Quit TheSeven (Nick collision from services.)
04:06:28 Join The_Seven [0] (
04:06:39 Nick The_Seven is now known as TheSeven (
04:12:49 Join T44 [0] (
04:25:56 Join JdGordon [0] (n=jonno@rockbox/developer/JdGordon)
04:31:07 Quit Topy44 (Read error: 113 (No route to host))
04:38:27 Quit Rondom (Nick collision from services.)
04:38:38 Join Rondom [0] (
04:40:31 Quit JdGordon (Read error: 104 (Connection reset by peer))
04:41:32***Saving seen data "./dancer.seen"
04:44:13 Join JdGordon [0] (n=jonno@rockbox/developer/JdGordon)
04:48:55 Join JdGordon1 [0] (
04:50:30 Quit JdGordon1 (Client Quit)
04:53:53 Quit JdGordon (Read error: 60 (Operation timed out))
04:55:59 Join JdGordon [0] (n=jonno@rockbox/developer/JdGordon)
05:01:19CIA-43New commit by mc2739 (r23023): add Seth Opgenorth to docs/CREDITS ...
05:04:18 Join fyre^OS [0] (
05:04:21 Quit GodEater (Read error: 113 (No route to host))
05:07:27 Join MTughan [0] (n=nnMTugha@synecdoche/developer/
05:07:47MTughanHow can I get the Rockbox version currently installed? All I can get for a version number is an SVN revision.
05:08:11 Join Zarggg [0] (
05:11:03 Quit Zarggg (Read error: 54 (Connection reset by peer))
05:11:11 Join Zarggg_ [0] (
05:11:19mc2739MTughan: the SVN revision is the version
05:11:41MTughanOkay. Any way to correspond that to a major.minor version release as on the site then?
05:12:13MTughanI don't see any revisions in the release notes.
05:13:39 Quit panni_ (Read error: 104 (Connection reset by peer))
05:13:43mc2739not really, the release is basically a branch off of the svn.
05:14:44MTughanHmm... Should I just try to use the Rockbox Utility to upgrade then, and if it needs an upgrade, it will?
05:15:06mc2739If I remember correctly, 3.4 was r22817 maybe
05:15:45MTughanI've got r21534 right now.
05:16:11 Quit JdGordon ("Leaving.")
05:16:22mc2739Yes, you can use it to install the latest release (3.4) or the current build (r23023)
05:17:20mc2739The release is usually more stable, but the current builds may have bug fixes that the release will not have
05:17:36MTughanAs is typical for trunk builds.
05:22:25 Quit fyrestorm (Read error: 110 (Connection timed out))
05:23:25 Quit fyre^OS (Read error: 110 (Connection timed out))
05:34:41MTughanAll right, seems to have upgraded to 3.4 stable fine. Thanks for the help mc2739.
05:36:01 Part MTughan ("I love the smell of napalm in the morning!")
05:36:59 Quit benime ()
05:49:55CIA-43New commit by kkurbjun (r23024): Vertical Stride: Add support for yuv to native output.
05:55:15CIA-43New commit by kkurbjun (r23025): M:Robe 500: Enable Vertical stride for main build.
05:55:54 Quit fdinel ("Miranda IM! Smaller, Faster, Easier.")
06:05:21 Join shai [0] (
06:19:29 Join funman [0] (n=fun@rockbox/developer/funman)
06:29:26 Join gtkspert_ [0] (
06:29:39 Quit Res1 (Read error: 60 (Operation timed out))
06:30:14 Join Res1 [0] (
06:36:46 Nick Zarggg_ is now known as Zarggg (
06:37:44 Quit gtkspert (Read error: 101 (Network is unreachable))
06:41:35***Saving seen data "./dancer.seen"
06:46:02 Join FOAD_ [0] (n=dok@
06:59:30 Quit FOAD (Read error: 110 (Connection timed out))
06:59:31 Nick FOAD_ is now known as FOAD (n=dok@
07:01:54funmanah blackjack manual is out of sync for e200, resume is down, not rec
07:04:00funmanand i can't figure "show high scores" from the source of blackjack.c :/
07:08:43funmanI think the "In menu" section must be removed, the menu is browsed with directional keys + select
07:12:56CIA-43New commit by funman (r23026): manual: e200v2 hasn't the bootloader on disk
07:17:50 Join MTughan [0] (n=nnMTugha@synecdoche/developer/
07:18:35MTughanI'm trying to create a WPS screen with conditional viewports, but it doesn't seem to be working. I preload a viewport with %Vl, give it some stuff, then load it with %Vd later on. But it doesn't show.
07:19:47MTughanOh, there we go... Didn't pay close enough attention to how the order is supposed to go.
07:20:31 Join n1s [0] (n=n1s@rockbox/developer/n1s)
07:20:38funmandoes manual also respect 80 columns lines rule ?
07:26:03 Join DerPapst [0] (
07:26:20 Quit liar|netbook ("Verlassend")
07:29:55 Quit DerPapst (Client Quit)
07:32:01 Quit Tristan (Read error: 104 (Connection reset by peer))
07:32:11 Part MTughan ("I love the smell of napalm in the morning!")
07:32:46pixelmafunman: it tries to but not as strict - often because if you add something an just break the rule because of a few characters, you start thinking "better leave it this way than adjusting those x next lines of the paragraph." and you have to be careful when breaking those lines with a long list inside \opt{} and often that wasn't done at all
07:32:50 Join Tristan [0] (
07:33:50 Join gtkspert [0] (
07:34:09funmani'll do my best when lines are already broken
07:34:39funmanCube manual looks out of sync as well : button rec is documented as max frame rate for e200 but it looks like it is select
07:36:03funmanshould I correct this in the same time I add Fuze mappings ?
07:37:26funmanI'll do that ;)
07:39:05 Quit gtkspert_ (Read error: 101 (Network is unreachable))
07:48:12 Join einhirn [0] (
07:56:28funmanlamp manual isn't very correct about the key to quit: just all keys which are not left or right do quit
07:57:20funmanshould we write "any key" ?
07:59:41 Join flydutch [0] (
08:00:39n1sfunman: "which key is the *any* key?"
08:01:17funmanhum the metronome isn't fully documented on e200 ..
08:02:14funmanlet's downgrade e200v1 to unstable until it's fixed
08:14:27AB3JUI suspect the only time my e200v1 would be unstable is if it was plugged into my FM transmitter/charger for it and I key up 50W on my 2m rig
08:18:11funmanpacbox is out of sync as well ..
08:19:51funmansince first days of 2007 it seems :/
08:27:13 Join Zagor [242] (n=bjorn@rockbox/developer/Zagor)
08:27:31 Join FOAD_ [0] (
08:28:43funmanpixelma: pegbox use 1 line per pad unlike other plugins, is that wanted ?
08:30:34CIA-43New commit by funman (r23027): calculator: use home as clear button for sansa fuze
08:30:38CIA-43New commit by funman (r23028): cube manual: correct mistakes ...
08:30:41CIA-43New commit by funman (r23029): metronome manual : document e200 buttons
08:30:45CIA-43New commit by funman (r23030): pacbox manual : fix e200 buttons
08:35:14 Join Rob2223 [0] (
08:35:16 Join TheSeven|Mobile [0] (n=TheSeven@
08:35:44pixelmafunman: time to dig up my plan (and do it) to make the plugin button tables in the manual adhere to one style but it's a lot of work, I guess I have to do it one by one...
08:35:52pixelmaor at least in smaller chunks
08:36:14 Quit FOAD (Read error: 145 (Connection timed out))
08:36:14 Nick FOAD_ is now known as FOAD (
08:36:20funmanallright I'll just follow the style of each file for now
08:36:38 Quit Lss (Read error: 131 (Connection reset by peer))
08:37:05n1shmm, many plugins have preprocessor conditions for inclusion both in SOURCES and in the plugin's .c file and often they are not the same, is one place preferred over the other? I like doing it only in SOURCES more
08:37:24pixelmasome are very hard to read (I think it was the goban.tex one that gave me the biggest headache
08:38:18 Join JdGordon [0] (i=7bf38c1f@gateway/web/freenode/x-gkfqivnsnnvfjkkw)
08:39:28pixelman1s: I think that has historic reasons, SOURCES didn't always exist and maybe later other authors just copied it. I'd also like it in one place in SOURCES better
08:41:36***Saving seen data "./dancer.seen"
08:47:36 Join batman52 [0] (
08:48:41n1spixelma: thanks, let's see if i have the patience to clean that up :)
08:51:17 Quit TheSeven (Read error: 113 (No route to host))
08:52:22CIA-43New commit by nls (r23031): Tiny simplification
08:53:04 Quit Rob2222 (Read error: 110 (Connection timed out))
08:55:18 Join Thundercloud [0] (
09:03:04 Join maruk [0] (
09:03:53funmanpixelma: should "In menu" section be removed for snake2 now it uses a standard menu ?
09:05:11funman+ blackjack
09:06:54funman+ snake
09:14:00 Join JackWinter2 [0] (
09:17:03 Quit JackWinter (Read error: 110 (Connection timed out))
09:19:23 Join JackWinter3 [0] (
09:25:43 Join esperegu [0] (n=quassel@
09:28:12 Quit JdGordon (Ping timeout: 180 seconds)
09:31:19 Quit JackWinter2 (Read error: 110 (Connection timed out))
09:34:10CIA-43New commit by funman (r23032): rockboy: remove useless assignement for fuze
09:34:15CIA-43New commit by funman (r23033): Sansa Fuze : add plugin keymaps to the manual ...
09:35:09 Join petur [0] (n=peter@rockbox/developer/petur)
09:37:27 Quit einhirn (Read error: 110 (Connection timed out))
09:38:15 Join einhirn [0] (
09:38:42funmanThere's no "Preview" button in FosWiki ?
09:39:26funmanthe editor is not exactly WYSIWYG
09:40:18linuxstbfunman: There's still a preview with "edit raw" I think.
09:40:18Zagorfunman: you can switch to the old text editor, it has a preview
09:43:23 Nick SIGSEGV2 is now known as pjm0616 (n=user@
09:46:07 Quit Thundercloud (Remote closed the connection)
09:47:57funmanIf I read well the SansaAMS wiki (just edited by myself :p ), we only need to finish properly the manual for Fuzev1 & e200v2 before marking them as stable
09:48:20funmanI'll ask the other developers on the ML
09:49:07linuxstbfunman: So "Rockbox runs well on these players." ?
09:49:25funmanon fuze yes
09:49:50linuxstbThere was also a suggestion that "Stable" means at least one official release - so those would need to wait until 3.5. But I think that was just a suggestion, nothing that's been agreed.
09:53:05*linuxstb hopes we can get the Nano2G up to that state before 3.5
09:55:14 Quit faemir ("Leaving")
10:02:02funmanshould I still use the bootloader_ams_pp branch for building new mkamsboot binaries ? Then I would need to backport all commits made since the branch was created
10:02:12funmanor I could build it from trunk and tag rbutil/mkamsboot/
10:05:08 Join GodEater [0] (
10:07:19linuxstbfunman: I never liked that branch, so would suggest just using trunk.
10:07:44linuxstbIn any case, that branch was just for a single release, which has been made.
10:08:06funmani was just thinking the name should have been versioned
10:09:03linuxstbWas there a single version number, or did the different components all get their own?
10:09:24*linuxstb guesses the latter, in which case, maybe it could be renamed to bootloader_ams_pp_july2009
10:11:49*petur plans to revert r21034 (recording keymapping) and then look for a better solution from there...
10:14:07funmanbuilding a universal binary of mkamsboot on osx is a bit tricky because we need to build libucl 2 times
10:18:21linuxstbfunman: Looks like the libucl Makefile supports an "OUT" variable. I think you would need to use that to build two versions in subdirs within mkamsboot
10:18:57linuxstbfunman: In fact, it seems to support building two Mac versions already.
10:19:04 Join DerPapst [0] (
10:19:33funmanoh right
10:19:40linuxstbLooks like rbutil uses it.
10:19:47linuxstb(presumably for the mkamsboot functionality).
10:24:58funman"make libmkamsboot-universal" doesn't work for me : "lipo: archive with no architecture specification: libmkamsbooti386.a (can't determine architecture for it)"
10:26:31linuxstbWhat about the individual ones - libucli386.a and libuclppc.a ?
10:27:11funmanhum i'm mistaken i wanted to know how mkamsboot' Makefile build libucl, but I think libucl is linked to rbutilqt, not to libmkamsboot
10:27:36*linuxstb was wondering what libmkamsboot was
10:27:57funmanwhat rbutilqt (and mkamsboot, which is just a frontend) link to
10:39:30 Join liar|netbook [0] (
10:41:37***Saving seen data "./dancer.seen"
10:47:45 Quit Bob_C (Read error: 113 (No route to host))
10:49:42CIA-43New commit by funman (r23034): mkamsboot: build and package (dmg) an universal OSX binary
10:50:48funmanZagor: can I send you the released mkamsboot binaries for you to put them on the download server ?
10:54:14 Join Bob_C [0] (
10:58:08funmanwould there be an advantage if i only copy rbutil/mkamsboot to tags/ ? a checkout of this tag wouldn't build since it misses libucl, but that would perhaps save space on the server
10:58:42linuxstbNo, no space is used for copies IIUC
10:59:04linuxstbOr rather, very little space I would expect.
11:00:19CIA-43New commit by funman (r23035): Tag release v1.1 of mkamsboot
11:01:54funmanwe'll need a 1.2 for clip+ if it's compatible
11:02:40linuxstbDoes current rbutil support the AMS Sansas?
11:02:47linuxstb(I mean the current release)
11:02:50tmzthas anything been done on clip+?
11:03:05funmanno, for a firmware update is not yet available
11:03:22funmanthe chip printings are similar if not identical to clipv2 however
11:30:39amiconnZagor: What's your opinion regarding the charging sceen / early usb crash in release 3.4? Should we have a 3.4.1?
11:30:54amiconnJust note that the fix is not yet committed to that branch
11:33:45CIA-43New commit by amiconn (r23036): Fix crash in early USB introduced in r22365 by moving the gui_syncstatusbar_init() back to where it needs to be. The comment clearly stated that it ...
11:33:52amiconnNow it is
11:34:47 Quit n1s (Read error: 110 (Connection timed out))
11:36:30linuxstbamiconn: Would the ipod video lcd_blit_yuv() be a sensible one to merge with the Nano2G (and possibly the Color/1st gen Nano) ?
11:37:38amiconnProbably not
11:38:39amiconnThe ipod video implementation lacks the dithering option, because of its very tight speed requirements
11:38:41*linuxstb sees a target/arm/lcd-as-memframe.S which could be complemented by one for the "not a memory-mapped framebuffer) versions
11:40:11amiconnThe non-memory-mapped versions are all slightly different, due to the integrated lcd controller access
11:40:31amiconnI don't see how we could unify that without getting into an ifdef hell
11:41:06tmztamiconn: what do those non-memory-mapped lcd controllers look like?
11:41:14tmztsomething you have to dma manually?
11:41:18linuxstbamiconn: Or do you think just copy/paste? i.e. what I already did, but with a better copy?
11:41:37 Quit JackWinter3 (Remote closed the connection)
11:42:34amiconntmzt: Most don't (and probably can't) use dma. Just write to a port, often preceded by polling a flag before (on arm targets)
11:43:00tmztone port?
11:43:13tmztI can look at an example
11:43:49amiconnNot always the same port
11:44:10 Join never [0] (
11:44:24amiconnlinuxstb: Also the Video's lcd hookup (bcm) is quite different to all other PP targets
11:44:39 Nick never is now known as nowt (
11:44:57CIA-43New commit by mc2739 (r23037): as3514 RTC driver: correct calculation, fixes FS #10625
11:45:00 Join robin0800 [0] (
11:45:08amiconnlinuxstb: The other arm implementation *are* mostly copy and paste, apart from those not yet using assembler routines
11:46:44*amiconn remembers needing quite a few changes when porting the asm implementation to c200
11:48:01linuxstbAlthough there seems to be a bigger problem on the Nano2G - both RGB and YUV updates are 54.0fps, so something is holding them back.
11:48:06nowtI've noticed in 3.4 on the iRiver H300 that the rec button now triggers the playlist view, which looks like it displays only filenames (that or track # - name). Is there any way to change what information it displays in this view?
11:48:34amiconnlinuxstb: That just means the CPU always has to wait for the lcd controller
11:49:00tmztamiconn: do you have a good example for that? the non-memory-mapped lcd
11:49:15amiconnSo assembler wouldn't give a speed advantage, but iirc the C implementation doesn't implement dithering
11:49:45 Join JackWinter [0] (
11:50:22linuxstbamiconn: But is that something that might be fixable?
11:50:25 Quit z35 ("Leaving")
11:52:06amiconntmzt: All targets apart from Sansa e200 (v1), Sansa View, Gigabeat F and S, m:robe 500, zen vision and Cowon D2
11:52:18tmztokay, thanks
11:52:58nowtI'd like it to display more than just the filename, perhaps artist and album instead for instance.
11:53:05amiconnThat doesn't mean the lcd controller doesn't use memory mapped i/o, it means that it doesn't use a memory mapped frame buffer
11:53:11nowtAm I out of luck?
11:54:16linuxstbnowt: I think so, yes.
11:54:34amiconnlcd-as-memframe.S just writes into this framebuffer. Other implementations need to write the data to the lcd controller's port in the correct order, obeying the access protocol
11:55:37nowtBummer, oh well I might make a feature request out of the question
11:56:22gevaertsnowt: you mean a feature idea :)
11:56:33linuxstbnowt: It's been a wanted feature for a long time. I think there may even be a patch on Flyspray to try and implement it.
11:57:16nowtgevaerts: I suppose I do :)
11:57:43nowtlinuxstb: I'll check it out, thanks
11:58:32tmztI can't believe that's fast enough to update the screen, as in, I had no idea it was done that way for anything but old monochrome lcds
12:02:50 Quit funman ("free(random());")
12:04:28 Join FOAD_ [0] (
12:08:10 Part nowt
12:16:15 Quit DerPapst ("Leaving.")
12:19:55 Quit FOAD (Read error: 110 (Connection timed out))
12:19:55 Nick FOAD_ is now known as FOAD (
12:21:26 Quit niekie (Read error: 113 (No route to host))
12:22:22 Join niekie [0] (
12:24:50 Join gtkspert_ [0] (
12:26:04 Quit Zarggg (Read error: 110 (Connection timed out))
12:36:37 Quit JackWinter (Remote closed the connection)
12:38:40 Quit gtkspert (Read error: 101 (Network is unreachable))
12:40:13 Join JackWinter [0] (
12:40:19 Quit liar|netbook (Read error: 110 (Connection timed out))
12:41:40***Saving seen data "./dancer.seen"
12:45:46 Quit niekie (Remote closed the connection)
12:47:48 Join niekie [0] (
12:54:40 Quit JackWinter (Remote closed the connection)
12:58:30 Join JackWinter [0] (
13:11:23 Quit FlynDice (Remote closed the connection)
13:29:27 Quit JackWinter (Remote closed the connection)
13:30:58*petur struggles to understand the recording_screen tex file wrt keymapping :/
13:32:19 Join JackWinter [0] (
13:40:28 Nick YPSY is now known as Ypsy (n=ypsy@
13:41:53YpsyHowdy, is there some option to turn on cover art? I've got jpg's in every folder and the original Sansa Fuze firmware displays them correctly. Any ideas?
13:44:06AB3JUYpsy: use a WPS that displays cover art
13:44:54YpsyThe preview in the RButility shows cover art on the one im using (electricbarsofcolor) and I've also tried other themes
13:45:16gevaertsare the files named correctly?
13:45:33 Join Topy [0] (
13:45:49YpsyHm no just randomly a single .jpg in the folder
13:46:05*gevaerts refers to the manual then :)
13:46:23Ypsy:P Ive searched the wiki for cover art but didnt find anything
13:48:01mc2739Ypsy: have you looked here?
13:48:19YpsyAh, guess thats the correct keyword to look for *g* ty
13:58:48 Quit esperegu (Read error: 104 (Connection reset by peer))
13:59:40YpsyCoolio now its working :) thx
14:04:26 Quit antil33t (Read error: 104 (Connection reset by peer))
14:04:32 Join antil33t [0] (n=Mudkips@
14:06:03 Quit T44 (Read error: 110 (Connection timed out))
14:06:28 Join darkham [0] (
14:07:24YpsyIs there a way to change the button layout on the Fuze?
14:10:41mc2739Ypsy: yes, you can checkout the source, make your changes and then compile a custom build
14:11:40YpsyIs it also possible to make the whole database sorting thing more like on the ipod or similar?
14:12:10YpsyWith the "Show all" etc.
14:12:55Torneis there a reason why drivers/pcf5060[56].c are seperate?
14:13:08Tornethe ipod (which is the one that uses pcf50605) actually has an 06 anyway :)
14:13:32Torneat least, it looks like it does (05 doesn't have adc's, but the ipod uses the pcf for adc in exact accordance with the 06 datasheet)
14:14:59 Join kugel [0] (
14:15:39gevaertsYpsy: I personally don't use the database and I've never used the ipod OF, so I have no idea what you mean, but maybe can help a bit
14:16:15kugellinuxstb: "I never liked that branch, so would suggest just using trunk." nice one..
14:17:16YpsyNice gevaerts that's exactly what I was looking for :)
14:21:38*GodEater thinks we ought to mention that virtually none of us in this channel use the database, and so any questions he has about it need to go into the forums.
14:22:33*kugel uses the database
14:23:04GodEateryay - well volunteered kugel
14:28:12 Join tomers [0] (
14:29:50Ypsy:P why aren't you using them?
14:30:02Ypsyand why is my doom crashing while loading map2 (data abort)? :(
14:30:24kugelon the fuze?
14:30:40kugelknown problem, likely not enough ram
14:30:46Ypsyah okay
14:30:51CIA-43New commit by kugel (r23038): Backport r23036 to the bootloader branch.
14:32:56 Join notlistening [0] (
14:33:11CIA-43New commit by kugel (r23039): Remove PP bootloader v6 tag and mkamsboot 1.1 tags (going to be retagged).
14:34:33notlisteningAny signs of Sansa AMS issues? I am getting a lot of prfetch errors on my player after updating last night
14:34:34GodEaterwhy aren't I using the database? It offers nothing to me that my filesystem doesn't already.
14:34:52CIA-43New commit by kugel (r23040): Tag v6 of the PP bootloaders
14:35:46Zagorwhat is the status of D2 nand write support?
14:36:22CIA-43New commit by kugel (r23041): Tag v1.1 mkamsboot
14:38:06CIA-43New commit by kugel (r23042): Tag v1 of the Sansa AMS bootloaders
14:38:24GodEatergood god kugel. Commitfest or what ?
14:38:55GodEaterI'm just impressed with your industriousness :)
14:39:01kugellooks like I introduced incosistent naming of the tags :/
14:39:01notlisteningsmall bits make better commits ;)
14:39:26gevaertskugel: do it again then! :)
14:40:08*kugel isn't sure if he can svn mv multple tags at once
14:41:42***Saving seen data "./dancer.seen"
14:42:10CIA-43New commit by kugel (r23043): Rename pp bootloader tag for better naming consistency with the other tags ("_" instead of "-")
14:42:20 Nick Ypsy is now known as YPSY (n=ypsy@
14:44:45CIA-43New commit by kugel (r23044): Rename mkamsboot v1.1 tag for better naming consistency with the other tags ("_" instead of "-")
14:45:03kugelshould be all for now :)
14:45:58*kugel wonders whether someone can build mac & windows binaries of mkamsboot 1.1
14:46:57 Join teru [0] (
14:48:24kugeloh no....
14:48:32kugelI was so sure I removed that RC bits
14:49:10 Quit maruk ("Leaving.")
14:49:23*kugel slaps himself very hard
14:54:54linuxstbkugel: ?
14:57:35GodEater23044 hasn't even made it into the build page yet
14:57:44GodEaterso we won't get to see the red for a bit ;)
14:58:28 Join domonoky [0] (n=Domonoky@rockbox/developer/domonoky)
14:58:42CIA-43New commit by kugel (r23045): Remove all RC bits from the ams bootloaders and mkamsboot. Also fix a compilation warning (I was 100% sure I did that already).
15:00:37notlisteningFrom a few months ago like june time is there a need to rebuild the AMS bootloader?
15:02:50linuxstbkugel: So you're using the old bootloader branch for mkamsboot v1.1 ?
15:03:05kugelyes of course
15:03:14kugelthat one is tested and works
15:03:19linuxstbYou can't keep a branch open forever...
15:03:27 Quit darkham (Remote closed the connection)
15:03:28gevaertsGodEater: not trunk
15:03:31kugelI won't
15:03:44kugelthe branch can be removed soon really
15:03:46linuxstbBoth funman and I thought trunk was better.
15:04:03linuxstbAnd then you come along, and without any discussion do something else...
15:04:16kugelYou didn't also
15:04:39gevaertskugel: it can't. I agree that it can be "closed", but you have to leave it there
15:04:54kugelI really don't like the attitude of of just making a release based of experimental code, when we have a known-to-work branch
15:05:05kugelgevaerts: huh?
15:05:27gevaertskugel: you're doing releases based on that branch. That means you have to keep it around
15:05:31linuxstbkugel: But mkamsboot is standalone.
15:05:37kugelgevaerts: that's why it's tagged
15:05:45gevaertshm, yes
15:05:48kugeltag, then remove branch
15:05:56*gevaerts asks to be ignored
15:06:33linuxstbkugel: It's true that the original decision was just funman and me, but you can't just come along and revert commits without discussion.
15:07:00kugellinuxstb: you just said "i don't like that branch, so use trunk", funman just followed it. I wished you have waited for the people that made the branch for a reason
15:07:22kugelthere was no so called "discussion"
15:07:57linuxstbSo you're arguing that it's OK to come and revert things without discussion if there was no discussion originally?
15:08:32gevaertswhat was reverted?
15:08:41kugelno, but I think it's not a good idea to use trunk for bootloaders if we have a nice working branch
15:08:42linuxstbtagging mkamsboot 1.1 from trunk (IIUC).
15:09:17linuxstbkugel: But mkamsboot is not "bootloaders" in the same way as the bootloaders that use firmware/ code.
15:10:24kugelbut mkamsboot is just as well known to work in the branch
15:10:57linuxstbYes, but with the inconvenience of committing things twice. I would argue that we should forget that branch and just commit mkamsboot changes in a single place - trunk.
15:10:57kugeland mkamsboot being buggy is way more dangerous than buggy bootloaders
15:11:10linuxstbSo the code is different in trunk and the branch?
15:11:32kugelI don't know, but I don't think it matters
15:11:42kugelwe can make 1.2 off the branch if you want
15:11:49kugelonce it's tested
15:11:50 Join Dege [0] (n=bla@
15:11:55linuxstbOf course it matters - funman was developing mkamsboot, and chose the trunk as the version he was working on, testing, and tagging...
15:11:57kugeloff trunk*
15:13:51kugelmkamsboot is a *very* critical tool. the version that we *certainly* know it works should be used
15:15:15linuxstbSo what are the differences that made you need to use the branch?
15:16:02kugelit says 1.1, not rXXXXX, I don't know of other changes. but I know it runs well
15:17:38linuxstbOK, that should have been fixed.
15:19:55kugelI need to re-tag now anyway
15:20:42CIA-43New commit by kugel (r23046): Remove mkamsboot v1.1 and sansa ams bootloaders v1 tags again, since they still say RC
15:21:19*gevaerts proposes to not re-tag until everyone agrees about what should be done
15:21:52kugelok (I almost hit enter :p)
15:23:18linuxstbkugel: Have you also backported the required changes to tools/ucl/ to build on OSX? They were made in August in trunk.
15:24:27kugelhrm, I don't think so
15:24:42*kugel can't remember such a change at all
15:25:24linuxstbThat's why it's good to use trunk, not a 3-month old branch...
15:26:26kugelwe should test again then
15:26:55linuxstbWell, or at least ask funman what testing he's done...
15:27:31linuxstbmkamsboot is easy to test rigorously - just compare generated binaries with earlier versions.
15:27:54tmztdoes ucl result in identical binaries each time?
15:27:55kugelthe current installation uses mkamsboot from the branch, I think people have used it on their macs
15:28:18linuxstbYes, but the changes were to build a universal binary.
15:28:19kugel(at least Lambda on his clip)
15:28:37linuxstbI'm sure it would have compiled fine on a Mac before, but just for a single arch.
15:29:39 Quit r00s (
15:29:39 Quit pcc1 (
15:29:47kugelok, that's actually a reason to use trunk (much better than "I don't like the branch")
15:30:40kugelbut we still want it to say "mkamsboot v1.1" not "mkamsboot rXXXXX", right?
15:31:32linuxstbBut that's the reason I didn't like the branch in the first place - creating two places things need to be committed to.
15:32:14gevaertsbranches are useful, but they should be short-lived
15:33:00*gevaerts thinks that the proper way to handle this is branch now, change strings, test, and tag, all done within one day
15:33:05linuxstbkugel: Yes, it should be v1.1
15:33:45kugelgevaerts: the bootloaders don't need to be build from trunk, there was no such important change
15:34:07 Quit linuxguy4 (Remote closed the connection)
15:34:19 Join linuxguy3 [0] (
15:36:09gevaertskugel: in that case handle bootloaders and mkamsboot differently
15:36:29*linuxstb would simply add #ifndef APPVERSION around the definition of APPVERSION in the Makefile, and build with "make APPVERSION=v1.1 mkamsboot"
15:36:39gevaertsthat's another way
15:39:07 Join maruk [0] (
15:39:31CIA-43New commit by kugel (r23047): Remove PP bootloader v6 tag, I tagged from the wrong directory
15:40:03CIA-43New commit by kugel (r23048): Tag v6 of the PortalPlayer bootloaders
15:40:44CIA-43New commit by kugel (r23049): Tag v1 of the AMS Sansa bootloaders
15:40:50*kugel really failed today
15:42:47kkurbjunkugel, you rang?
15:43:11gevaertskugel: if you had done this earlier, at least you could have grabbed 23000 or 23032 :)
15:43:50kugelgevaerts: you'd have accused me for cheating then :p
15:44:01gevaertstrue :)
15:44:02kugelkkurbjun: re: r22934; sleep() yields
15:45:06kkurbjunkugel, thanks for the heads up
15:46:38kugelkkurbjun: it's now likely to yield twice. not sure if it matters
15:47:05 Join darkham [0] (
15:47:31kugelkkurbjun: I adapted some fixed point math for chopper, it runs much smoother but I somehow can't make the speed to match with SVN
15:48:19kkurbjunyeah, it doesn't matter for brickmania as it appears to get through each frame within 30 ms even with two yields
15:48:51kkurbjuncool, I saw that question you had on the math for adding a number twice versus shifting it left once
15:49:48kkurbjundo you have a snapshot of the current progress?
15:50:10kkurbjunI would be interested to take a look
15:50:33 Join liar|netbook [0] (n=liar@
15:51:24kugelkkurbjun: is a patch OK too?
15:51:57kkurbjunyeah, patch is fine
15:53:50kugel actually builds
15:54:44kugelit seems to work best with CHOPPER_SPEED_FACTOR == 3, probably because the original initial speed is 3
15:55:32kugelbut I seem to get different results after just recompiling :/
15:56:09NJoinr00s [0] (
15:56:52NJoinpcc1 [0] (
16:00:16 Join intrados1 [0] (
16:04:30CIA-43New commit by kkurbjun (r23050): M:Robe 500: Make the pre-compiled images build with vertical strides.
16:10:30 Quit notlistening (Remote closed the connection)
16:10:39kkurbjunkugel, I havn't had a chance to try it yet, but one thing that I found to help keep the math sane was to use fixed point numbers in everything but firmware calls. That way I didn't have to keep track of whether I should be using an INT3/4 - the rule for me was to never use it unless it was a call to draw the image or the like.
16:11:35 Quit GodEater ("leaving")
16:11:42kugelthe code looks heavily burrowed (camelCase everywhere), so I try to avoid changing it too much
16:11:46 Join GodEater [0] (
16:11:59 Quit intrados (Read error: 110 (Connection timed out))
16:12:09kugelbut yea, that might actually result in less changing
16:16:49 Join JackWinter2 [0] (
16:18:02kugelkkurbjun: how do you do % with fixedpoint math?
16:20:16kugelall that literal numbers are making it messy...
16:20:56 Quit Zagor ("Don't panic")
16:22:08 Quit intrados1 (Read error: 110 (Connection timed out))
16:24:06 Part domonoky
16:27:37 Join ender` [0] (
16:28:01 Join toffe82 [0] (n=chatzill@
16:30:25 Join Strife89 [0] (
16:30:33 Quit JackWinter (Connection timed out)
16:32:22 Quit darkham (Remote closed the connection)
16:39:34 Join jgarvey [0] (
16:40:05kkurbjunhmm, mod?, I'm not sure offhand, I think doom has a mod implementation
16:40:08kkurbjunlet me check
16:40:37kkurbjunkugel, doom has it in m_fixed.h
16:41:44***Saving seen data "./dancer.seen"
16:42:13 Join froggyman [0] (
16:42:42 Nick froggyman is now known as Guest30105 (
16:47:55 Quit flydutch (Read error: 60 (Operation timed out))
16:50:24 Join funman [0] (n=fun@rockbox/developer/funman)
16:51:28 Join panni_ [0] (
16:53:19 Join MethoS- [0] (n=clemens@
16:54:18*kugel just lost his progress
16:54:35funmankugel: i'm going to answer about your commits on mailing list
16:55:18 Nick YPSY is now known as Ypsy (n=ypsy@
16:55:42kugelfunman: ok
16:56:34 Join brett_ [0] (
16:56:41 Nick brett_ is now known as brettnem (
16:56:49brettnemHello all
16:57:43brettnemhey, I'm having some trouble with my Sansa e250 in OS X. I have rockbox installed (had to manually install from a newer mac using leopard). Tiger doesn't ever recognize the e250.. any ideas?
16:58:37brettnemthe original sansa firmware (1.2.x) didn't work on tiger either.. I even tried to downgrade to 1.1.x as recommended. player said it was upgrading, it rebooted and still in 1.2. .That was before I stuck rockbox on.
16:58:43brettnemany ideas please? :)
17:00:27 Quit JackWinter2 (Read error: 54 (Connection reset by peer))
17:01:41 Join intrados1 [0] (
17:02:52 Join midgey [0] (n=tjross@rockbox/developer/midgey)
17:02:58 Join JackWinter [0] (
17:05:23 Nick Guest30105 is now known as froggyman (
17:06:15 Quit brettnem (Read error: 104 (Connection reset by peer))
17:07:03 Quit robin0800 (Read error: 110 (Connection timed out))
17:09:55 Quit JackWinter (Read error: 104 (Connection reset by peer))
17:10:35 Nick Ypsy is now known as YPSY (n=ypsy@
17:11:10 Nick YPSY is now known as Ypsy (n=ypsy@
17:11:13 Join robin0800 [0] (
17:11:42 Quit teru ("Quit")
17:12:08 Join JackWinter [0] (
17:12:43 Join brett_ [0] (
17:12:43 Nick brett_ is now known as Guest79036 (
17:12:47 Nick Guest79036 is now known as brettnem (
17:13:01brettnemsorry, lost connection.. any ideas?
17:13:28 Join TheSeven [0] (
17:13:38 Join Horscht [0] (n=Horscht2@xbmc/user/horscht)
17:15:02brettnemI get errors like this in my syslog:
17:15:02brettnemOct 9 10:11:45 brett-imac-g5 kernel[0]: USBF: 506.542 AppleUSBEHCI[0x2724800]::Found a transaction past the completion deadline on bus 75, timing out!
17:15:03brettnemOct 9 10:11:56 brett-imac-g5 kernel[0]: USBF: 517.543 AppleUSBEHCI[0x2724800]::Found a transaction past the completion deadline on bus 75, timing out!
17:15:03DBUGEnqueued KICK brettnem
17:15:03brettnemOct 9 10:12:24 brett-imac-g5 kernel[0]: USBF: 545.543 AppleUSBEHCI[0x2724800]::Found a transaction which hasn't moved in 5 seconds on bus 75, timing out!
17:15:03brettnemOct 9 10:12:30 brett-imac-g5 kernel[0]: USBF: 551.543 AppleUSBEHCI[0x2724800]::Found a transaction which hasn't moved in 5 seconds on bus 75, timing out!
17:16:21linuxstbbrettnem: I think there's a known problem with OS X and Rockbox's HID USB feature. I think the solution is to disable that feature in the settings. (I'm not sure how - I know pixelma has been investigating)
17:16:28linuxstbIs "Tiger" 10.4 ?
17:16:41linuxstbThen yes, I think that's the issue.
17:16:49brettnemI've looked all over the settings menus and I don't see *any* USB options.. am I missing something?
17:17:29brettnemmaybe some setting somewhere hiding advanced options? <shrug> ?
17:17:34 Quit TheSeven|Mobile (Read error: 104 (Connection reset by peer))
17:17:38linuxstbAh, maybe it isn't a user-setting, but rather something that can be disabled if you compile rockbox yourself.
17:18:30brettnemhum.. I can handle compiling stuff in linux.. but I've never done anything for a device like this.. any info on doing that?
17:19:02linuxstbThere are many wiki pages about it (maybe too many...). Try
17:19:25 Quit intrados1 (Connection timed out)
17:19:33linuxstbOr perhaps here -
17:19:37linuxstb(that should also apply to OS X)
17:19:43brettnemwhat is HID and do I lose something by disabling it?
17:20:08linuxstbIt makes the buttons on your Sansa appear as buttons on your computer.
17:20:11Torneyou lose a feature that doesn't work on your OS
17:20:40brettnemoh, I don't need that anyway. :)
17:20:54CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
17:20:54*linuxstb doubts anyone _needs_ that feature ;)
17:21:06brettnemok.. so I'll just look for disabling HID... you think that's it then?
17:22:17linuxstbYes. Or at least, I know that HID is causing problems on OS X 10.4
17:23:41brettnemok, thanks.. I'll give it a shot
17:25:39 Quit froggyman ("CGI:IRC (EOF)")
17:36:45amiconngevaerts: Would it be possible to detect OSX 10.4 enumeration failure on the device side (i.e. in rockbox)?
17:36:55midgeyspeaking of OS X, compiling is broken on 10.6
17:37:16midgeyif anyone is around with a 10.5 install, I have a patch that fixes building the simulators on 10.6
17:37:30midgeyno idea what is causing cross-compiled builds to fail
17:41:29 Join CaptainKewl [0] (
17:42:07 Join bmbl [0] (n=Miranda@unaffiliated/bmbl)
17:42:35 Quit funman ("free(random());")
17:47:01shaitomers, ping
17:47:35 Join pamaury [0] (
17:49:29 Quit CaptainKewl (Remote closed the connection)
17:49:35 Join Grahack [0] (
17:50:44tomersshai: pong
17:51:10tomersshai: Hi! Is the iPod bug (mouse mode stucks) solved now?
17:51:28 Join intrados1 [0] (
17:51:48 Join DerPapst [0] (
17:52:32 Quit intrados1 (SendQ exceeded)
17:54:43 Quit Unhelpful ("No Ping reply in 90 seconds.")
17:54:53 Join Unhelpful [0] (
17:55:57TheSevendoes anybody have an idea what could be causing these lockups if you select a song while playing another one?
18:00:22 Join froggywoman [0] (
18:00:36 Join chandoo [0] (
18:04:52 Join Zoxc [0] (
18:06:26kugelTheSeven: missing yields?
18:06:52kugelour sd drivers do yield() a lot during transfers (every few ticks or even less)
18:08:07gevaertsamiconn: I doubt it
18:08:19midgeyTheSeven: do the lockups occur if the player is still buffering or does it happen regardless of buffering state?
18:08:21gevaertsMaking HID configurable is possble of course
18:08:25pixelmalinuxstb, Torne: the HID feature works in 10.4 - at least if the USB prober is running, but the disk(s) won't mount even then. Without USB prober I noticed some reactions on button input but only once in felt 100 tries and then buttons "stick".
18:09:11Tornethat's weirder than i heard then :)
18:10:55pixelmaand turning HID support off is currently no user setting, the only way currently is to use a build that disables it (it's a simple define to comment out if I remember correctly)
18:11:18pixelmaTorne: it is very much
18:13:15TheSevenkugel, midgey: Playback works perfectly fine, even skipping forward/back, the only thing that fails is opening another file from the file or database browser, if another file was played before (even if it was stopped in the meantime). and yes, the drivers yield.
18:14:14kugelTheSeven: well, that still sounds like the ata/ftl thread steals runtime for buffering, playback and audio thread
18:14:45kugelskipping tracks doesn't mean flash transfer
18:15:05TheSevenskipping backwards beyond the point where you started playback should mean flash transfers
18:15:14TheSevenand even playing more than half an hour works fine
18:15:57linuxstbTheSeven: Maybe it's another stack overflow. The buffering thread might need its stack increasing.
18:16:37kugel I see now yield() in ftl-nano2g
18:16:48TheSevenkugel: not the ftl, but the nand driver
18:17:03midgeydo those blocking waits in the ftl and nand need a yield?
18:17:31kugelTheSeven: timeout seems to be HZ, way too high
18:17:33TheSevenmidgey: all blocking waits already have a yield
18:17:35*TheSeven still doesn't see the difference between skipping backwards and selecting the *same* track from the file menu
18:18:31TheSevenkugel: I originally went for HZ/10, but had some issues (which were probably caused by somthing else and are fixed now)
18:18:45amiconngevaerts: Are the sub-functions of a composite device ordered in some way?
18:18:47kugelit should be more like HZ/50
18:18:59TheSeventhat timeout wasn't ever hit (besides during bank detection) though
18:19:00pixelmabrettnem: the orignial firmware works for me though (with a c200) but it also fails sometimes to automount, trying to connect again often helps - and of course I don't get access to the microSDHC, a card without HC would probably work because I see the "drive" but couldn't test
18:19:40*kugel would just give it a try
18:19:48kugelsamsa sd driver yields every 5 ticks
18:19:52gevaertsamiconn: they're ordered as in struct usb_class_driver in usb_core.c
18:19:58TheSevenkugel: yes, i'll try it, but i don't expect a difference
18:20:39TheSevenand it's yielding all the time wile waiting for a timeout anyways
18:20:42brettnempixelma: I've tried to connect *several* times without luck
18:21:17Bob_CDoes our USB stack provide an Interface Association Descriptor?
18:21:32brettnemweird thing is, I tried to downgrade the sansa firmware to 1.1.x per recommendations to work with Tiger and although the e250 said it updated the firmware, there was no change.
18:22:15gevaertsBob_C: no
18:22:53pixelmawith or without a card plugged? I remember some weirdness when connecting with an inserted card even on Windows. And maybe my situation is a bit different as it is a c200 running a quite old firmware (the one it came with)
18:23:12pixelmabrettnem: ^
18:23:26kugelTheSeven: the code is like a mess
18:23:33brettnempixelma, there was no card
18:23:37amiconngevaerts: Hmm, does that mean MSD is already reported first?
18:23:43TheSevenkugel: I know :-)
18:23:52Bob_Cgevaerts: We probably should, shouldn't we?
18:23:56pixelmabrettnem: ok, I'm out of ideas then
18:24:12TheSeventhat's the result of programming it without understanding the meaning of everything in the beginning
18:24:23pixelmaexcept trying a Rockbox build without HID support, of course :)
18:24:58gevaertsamiconn: yes
18:25:08brettnempixelma: yeah, I'll give that a try.. I'm not sure if I'm up to the build process at the moment :)
18:25:14midgeypixelma: are you on 10.4 right now?
18:25:26gevaertsBob_C: how would that help? Both MSC and HID are single-interface classes
18:25:26 Quit krazykit (Read error: 113 (No route to host))
18:25:28brettnempixelma: BTW, it works fine on Leopard
18:25:51brettnembut my media isn't on my leopard machine. ;)
18:26:09pixelmayeah, I guess we would have heard a lot more complaints if it wouldn't
18:26:44Bob_CIAD is the official way to implement composite devices isn't it? It may work without one, but it leads to problems like we have with MacOs
18:27:17*amiconn should try a hid enabled build on his Amiga
18:27:22kugelTheSeven: IIUC it constantly yields, and cancels operation after 1 second?
18:27:23gevaertsBob_C: the link you showed a few days ago says that IAD isn't actually used at all on OSX
18:27:30TheSevenkugel: yes
18:27:49kugelmaybe it yields to often then ;)
18:28:01TheSevenwell, can't hurt, can it?
18:28:07brettnempixelma: any chance an older build may work better?
18:28:28kugelTheSeven: probably not, if you're using dma
18:28:39TheSevencan anyone tell me while we're waiting for the disk to spin down at both lines 756 and 767 of powermgmt.c?
18:28:50pixelmabrettnem: probably the once before HID support - but after Rockbox USB
18:29:09linuxstbTheSeven: s/while/why/ ?
18:29:28*TheSeven wonders why he's typing such crap again
18:30:07pixelmamidgey: do you have an idea what to test? I wouldn't want to set up a build environment though, it's a work computer
18:30:09linuxstbbrettnem: Rockbox 3.3 could be better for you.
18:30:12Bob_CWhat I meant was that if use the spec it should work correctly on compliant stacks, and not lead to random behaviour depending on vagaries of the host
18:30:17 Quit maruk ("Leaving.")
18:30:20linuxstbbrettnem: (it has USB, but not HID)
18:30:28kugelTheSeven: isn't there a nand thread?
18:30:31 Join Lss [0] (
18:30:38 Join Lss__ [0] (
18:30:38midgeypixelma: ahh, no i need a computer with a build env (specifically building the sim)
18:30:50 Quit Lss (Client Quit)
18:30:51 Quit Lss__ (Read error: 104 (Connection reset by peer))
18:30:53TheSevenno, it's all done from the calling thread, as we need to block that anyways until it's done
18:30:55 Join n1s [0] (n=n1s@rockbox/developer/n1s)
18:31:03 Join Lss [0] (
18:31:11midgeybasically, apple changed the way the section attribute works in 10.6 and so the sim will fail to link
18:31:22pixelmamidgey: if you give brettnem some time (and a helping hand...) ;)
18:31:37midgeyi have a fix for snow leopard, idk if leopard or tiger will work with it
18:31:57kugelTheSeven: that seems a bit strange. using a thread, you can block out all other core threads. without the mutex will not block the thread whoever calls it first
18:31:59brettnemI can try 3.3
18:32:08*TheSeven is tempted to move the audio hw shutdown above saving the glyph cache, and removing one of those wait loops
18:32:10midgeychecking __ENVIRONMENT_MAC_OS_X_VERSION_MIN_REQUIRED__ is no fun.. so i'd like to avoid the special casing
18:32:15gevaertsBob_C: as far as I know using an IAD isn't needed if you only have single-interface functions
18:32:16brettnemdo I just copy the build to my devices .rockbox dir, like I originally did?
18:32:44kugelbut you may just ignore me, I have only little knowledge in that area
18:33:05TheSevenkugel: shouldn't that mutex just ensure that only one thread is in there at a time?
18:33:26kugelyea, that's what I said
18:33:37kugelor meant to say :)
18:33:52TheSevenwell, that's what I intended - why should I need an extra thread then?
18:34:17brettnembtw I don't see any 3.x version in the old releases dir
18:34:23kugeldo you have implemented the virtual led already?
18:34:28amiconnmidgey: Afaik the sim builds don't use section attributes, so I'm curious what/where the problem is
18:34:46TheSevenof course one could thereby unblock the calling thread while doing some cleanup operations, but that's performance improvements which will come later
18:34:55linuxstbbrettnem: They are in duh.. :)
18:35:14Bob_Cgevaerts: well if it works no need to change it I guess
18:35:30TheSevenkugel: I don't think so, whatever that is.
18:35:32midgeyamiconn: firmware/export/config.h, DATA_ATTR is special cased for the sim
18:35:34linuxstbbrettnem: Not your fault - the website needs updating...
18:35:58kugelTheSeven: it's a small icon at the very right of the statusbar, indicating flash accesses
18:37:35kugelit's possible that a nand thread would be needed for that
18:38:00kugellooking at the sd drivers, the extra thread is only used for handling microsd insertion/extracting and the virtual led
18:38:18brettnemso question.. while I'm trying to get 3.3 on my sansa.. what video formats are supported? I thought just about all vlc formats were.. thought I read that somewhere..
18:38:38 Quit Dege (Read error: 110 (Connection timed out))
18:38:39amiconnmidgey: I wonder what that's good for...
18:39:02linuxstbbrettnem: No, just MPEG-1/MPEG-2 video with mpeg audio (stored in an MPEG-2 program stream). The manual and/or the PluginMpegplayer wiki page should describe it more.
18:39:18midgeyamiconn: not sure, might be some left over from the old sim
18:39:30amiconnI don't thinks so
18:39:42brettnemhmm.. just copied .rockbox over.. rebooted and it's still 3.4.. grumble
18:40:25kugelnot even for virtual led it seems
18:40:36amiconnAh yes. This is for plugins and codecs to force the api pointer into the data section, so it doesn't get cleared by bss zeroing
18:41:07midgeyah, i suppose that makes sense
18:41:19*amiconn wonders why apple does silly things like the data section not working like on any other platform
18:41:47***Saving seen data "./dancer.seen"
18:43:12 Join bertrik [0] (
18:44:18brettnemoh here we go.. yeah I did dsomething stupid
18:45:39 Join intrados1 [0] (
18:47:37brettnemboo.. it didn't work.. with 3.3 I get this in the syslog:
18:47:38brettnemUSBF: 6214.718 AppleUSBEHCI[0x2724800]::Found a transaction past the completion deadline on bus 75, timing out!
18:48:15brettnemalso get one of these:
18:48:16brettnemUSBF: 6225.728 AppleUSBEHCI[0x2724800]::DoIOTransfer - error 0xe000404f queueing request
18:50:43 Quit froggywoman ("CGI:IRC")
18:52:05 Nick Ypsy is now known as YPSY (n=ypsy@
18:56:09*TheSeven wonders how to best implement the FTL shutdown
18:56:22TheSevenintroduce a storage_shutdown() function?
18:56:52 Quit Strife89 (Read error: 60 (Operation timed out))
18:57:15 Join Strife89 [0] (
18:59:05kugelTheSeven: nand_enable()? maybe?
18:59:32kugelthat could also be called when rockbox is running, to save battery
18:59:52TheSevenif anything, storage_sleep or storage_sleepnow, whatever they're supposed to do
19:00:08linuxstbTheSeven: How much work is the nand shutdown/mount? On hard-disk targets, Rockbox spins the disk down after a period of inactivity, and I was wondering if that mechanism would make sense on the Nano2G.
19:00:32 Join intrados2 [0] (
19:00:35linuxstb(e.g. to protect against a user doing a hard MENU+SELECT reset)
19:00:56TheSevenlinuxstb: unmounting the ftl unneccessarily introduces quite some additonal wear
19:01:10linuxstbOK, then that's a bad idea.
19:01:39TheSevenand after a hard reset, the norboot will clean it up, so booting will take ~20 seconds longer, but it will be sane afterwards
19:02:25TheSevenin fact, not unmounting it seems to yield more stable results than unmounting it properly on shutdown, but that additional boot time is just nasty
19:02:59TheSevendepending on how much was written before, unmounting it may take from zero to 3 seconds time
19:03:21 Join Strife1989 [0] (
19:05:18amiconnkugel: Hmm, there's a statusbar bug that seems to affect all rtc sims
19:05:39amiconnTime display sometimes jumps to the left, overlaying the volume and play status icons
19:06:02*amiconn also noticed a bug in ipod G1/G2 wheel simulation
19:06:48TheSevenamiconn: I just saw that clock jumps left bug on my (hardware) nano2g!
19:06:50amiconnEach cursor button press moves by two steps in lists, so it's impossible to get to every other entry :(
19:07:30amiconnTheSeven: So it's not just sims... I didn't update my targets yet
19:08:40linuxstbamiconn, kugel: I've seen that as well on the Nano2G. All default settings. Also, I don't know if this is a new statusbar bug, or a problem with the time/date screen, but if I change the time, and go back to the time screen, the time is correct in the main part of the display,but wrong in the status bar.
19:08:51 Quit intrados1 (Connection timed out)
19:09:02 Join Thundercloud [0] (
19:09:04 Quit Strife89 (Read error: 60 (Operation timed out))
19:09:40amiconnlinuxstb: Seen it in c200 and e200 sim so far (the two rtc-enabled sims I tried so far)
19:10:19 Join saratoga [0] (i=98039f25@gateway/web/freenode/session)
19:10:36saratogaTheSeven: does playback crash if you go from an MP3 track to an Ogg track in the same playlist?
19:10:39*TheSeven wonders if there are any docs on the synopsis of storage_sleep and storage_sleepnow
19:10:46 Quit lifeless_ (Read error: 110 (Connection timed out))
19:10:58TheSevensaratoga: didn't try this yet. need to get an ogg track :-)
19:11:09saratogawell any non-MP3 format
19:11:11linuxstbamiconn: I'm talking about on-target (Nano2G)
19:11:30 Part toffe82
19:11:32saratogaif so it may be due to lack of icache cleaning between tracks
19:12:48amiconnlinuxstb: Yeah. I don't remember this happening earlier though. Maybe the bug hides in the new rtl code?
19:12:56kugelI saw that too
19:13:11linuxstbamiconn: Ah yes, it could be.
19:13:17kugelIt seems to happen since the addition of RTL
19:13:19TheSevensaratoga: but then it wouldn't crash if i try only mp3 files?
19:13:41saratogaTheSeven: becuase you'll only get a crash if the system needs to reload the .codec file
19:13:46Zoxcdoes rockbox have fancy menus with smooth animations yet?
19:14:00linuxstbZoxc: No.
19:14:10*amiconn would strongly object menu animations
19:14:14TheSevenwell, even if it reloads it, it shouldn't hurt as long as it's overwritten with itself
19:14:22saratogathe problem on many targets is that only ARM's dcache is coherent with memory, so icache needs to be cleaned after any self modifying code runs (such as a codec loader)
19:14:35kugelAm I the statusbar guy now? :P
19:14:45amiconnkugel: yeps
19:14:49TheSevenah, self-modifying... didn't think of that :-/
19:14:53*linuxstb just followed amiconn's lead ;)
19:14:56saratogathough this may not be your issue, it has been a problem on most arm targets so far since everyone forgets about it :)
19:15:23saratogaif its the case you just need to double check that you're linking the correct cache flush functions for your arm target
19:15:29pixelmacould the clock thing in the statusbar also be related to n1s' RTC driver cleanup?
19:15:38saratogawhich should be the same as AMS/GigabeatF anyway
19:15:42brettnempixelma: if 3.3 didn't provide any additional help, will recompiling wihtout HID help still?
19:15:45TheSevensaratoga: In fact I don't think we're linking any
19:15:48amiconnsaratoga: Both the plugin and codec loader take care of this and flush icache
19:15:48*kugel didn't really touch statusbar code in SVN
19:16:05saratogaamiconn: they do but only if the cache flush functions are implemented
19:16:16saratogaotherwise they ifdef to NOPs :)
19:16:25amiconnOf course
19:16:31kugelone could try if unsetting the RTL in the sb viewports help
19:17:23amiconnkugel: Moving around in lists makes the time display jump between its correct place and the wrong one
19:17:52*linuxstb pings tomers and points to the last few minutes of logs...
19:17:54amiconnI wonder how the statusbar viewport can get influenced this way
19:18:11pixelmabrettnem: hmm... not sure, it did work for me on my c200 when I tried some time ago
19:18:36brettnemis HID not is 3.3?
19:19:00kugelamiconn: no idea, but the time jumps to the left, as if something tries to draw it RTL'ish
19:19:17TheSevenIt looks like I need to add a new storage function
19:19:18kugelthe lcd driver puts_* adjusts text if the RTL flag is set
19:19:18amiconnYes, and even for ltr languages...
19:19:29TheSevenwould you go for storage_shutdown or storage_sync?
19:20:01brettnempixelma: Could the native e200 firmware have *anything* to do with this problem? I know for a fact that the 1.2.x sansa firmware is incompatible with 10.4
19:20:09TheSevenkugel: sounds good
19:20:58TheSevenbrettnem: sounds rather like 10.4 being incompatible to the 1.2.x firmware?
19:21:47brettnemperhaps... it's in a bunch of forums that downgrading to 1.1.x firmware makes it work on 10.4.. but is that firmware even used once rockbox is loaded?
19:21:51midgeywhen i was on 10.4, usb was flaky at best
19:22:07midgeybrettnem: the OF should have no affect
19:22:10brettnemugh.. I have a G5 PPC.. I can't really upgrade
19:22:46midgeydoes 1.2.x fail since it tries to autodetect if the host support MTP?
19:22:49amiconnOSX 10.5 still supports ppc afaik
19:23:08midgeyyep, 10.5 runs on ppc
19:23:08brettnemI don't know why 1.2.x fails
19:23:23amiconnOnly 10.6 dropped ppc support
19:23:41brettnemmaybe I'll try to up to 10.5
19:24:36midgeybrettnem: as an added bonus, finder in 10.5 won't lock up if a network drive disappears but that's getting off topic
19:25:02 Join intrados3 [0] (
19:25:08 Quit intrados2 (Read error: 60 (Operation timed out))
19:25:17 Join notlistening [0] (
19:25:35brettnemhmm.. I'm not sure I can even still get 10.5.. I think apple is really putting us PPC users to the fire
19:26:00kugelamiconn: I got a fix
19:26:27kugelI told tomers to do something with the flag in viewport_set_defaults(), but it should've been viewport_set_fullscreen()
19:27:50gevaertsbrettnem: HID was in 3.3
19:28:03brettnemso maybe go back to 3.2?
19:28:46gevaertsas an interim solution, yes. We need to solve it properly though
19:28:57amiconnbrettnem: You could make an own build without HID support
19:29:04linuxstbgevaerts: It was? The changelog says it was new in 3.4 -
19:29:28brettnemyeah.. I'm considering it....
19:30:25 Quit tomers (Read error: 148 (No route to host))
19:30:26gevaertslinuxstb: HID was enabled in 3.3, but the sending of HID events wasn't
19:30:39gevaertsso problems coming from using a composite device will be there
19:30:44linuxstbgevaerts: Ah, a subtle difference ;)
19:32:06linuxstbgevaerts: Do you know if there's a FS task for this? We should probably add it to the 3.4 (and 3.3?) known bugs list.
19:32:36midgeyanyone have any ideas why arm-elf-ar would send SIGABRT when using the r switch?
19:32:43gevaertsI don't think there's a task
19:33:09*linuxstb goes to search, and then add one if not found
19:35:28pixelmasorry, I didn't open a task even though I knew about it :\
19:36:28 Quit MethoS- (Remote closed the connection)
19:36:54 Join MethoS- [0] (n=clemens@
19:39:17*linuxstb adds the devil's FS task - FS #10666
19:39:30linuxstbpixelma: No problem.
19:40:06linuxstbWhich released targets does this affect?
19:40:27gevaertsfor releases, all PP except ipods
19:41:32linuxstbSo Sansa C200, E200, E200R, iriver H10 and m:robe 100 ?
19:41:32gevaertswould it help if I add (or more accurately, extend) an #ifdef that allows for easy disabling of HID during MSC connections at build time?
19:41:42gevaertsthat looks correct, yes
19:42:48 Quit TheSeven (Read error: 104 (Connection reset by peer))
19:43:27 Join TheSeven [0] (
19:44:20*linuxstb adds to the release notes
19:44:45TheSevenany hints on where to do the storage_flush in shutdown_hw?
19:44:48TheSevendepending on where i put it, it will either not fulfill it's purpose or cause spindown-up-down in some cases
19:45:35linuxstbTheSeven: Shouldn't it just be in the same place as storage_spindown() ?
19:46:24TheSeventhe spindown is only done at safe battery levels for some reason, but well, that make sense for an FTL commit, too
19:47:11linuxstbTheSeven: I'm not sure about the name storage_flush() though. Isn't it more like an unmount? To me, "flush" sounds like it will write pending things to flash, but then carry on working. Is that accurate?
19:47:30TheSevenit's something in between
19:48:26TheSevenit will in fact only commit pending things, and then carry on working, but one should not call it unneccessarily (e.g. only on shutdown and after writing *extremely* important things)
19:48:42TheSevenso it's a flush function used to unmount it :-)
19:48:46linuxstbDo we have an "unmount/eject" ? Maybe that could be useful if Rockbox ever supports external drives with USB host.
19:48:54n1spixelma: i don't think a bug in my rtc cleanup would cause this statusbar weirdness since it only touched the low level drivers so should only cause wrong time to be read/set
19:49:32linuxstbTheSeven: But I guess the name isn't that important - a comment next to the call to it would be very helpful to future readers I think though.
19:49:49linuxstb(as this is a concept specific to the Nano2G so far)
19:50:04TheSevenwe should also add such comments for other things in there...
19:50:15TheSeven(e.g. what is the difference between spindown/sleep/sleepnow?)
19:50:30*linuxstb would like to know that too
19:50:35*bertrik too
19:50:37pixelman1s: guess you are right
19:50:44notlisteninganyone have scrolling problems on their e200?
19:51:27bertriknotlistening, you're talking about an e200v1, right? I could check on my e200
19:51:37gevaertsnotlistening: what sort of scrolling?
19:52:08brettnempixelma: where's the IFDEF for HID? I'm going to try compiling myself
19:52:38 Quit robin0800 (Remote closed the connection)
19:52:38notlisteningSorry bad question, a v2 and scrolling menu items with the scroll wheel
19:53:38gevaertsbrettnem: can you handle patches?
19:54:29 Quit chandoo ("Leaving")
19:54:33notlisteningmoving single items is now very hard and even just lightly touching the wheel seems to make it go hay wire
19:54:45 Join robin0800 [0] (
19:55:48CIA-43New commit by midgey34 (r23051): Fix compilation of simulators on Mac OS X 10.6. Tested on 10.5.8 and 10.6.1.
19:57:54brettnemgevaerts: sure
19:58:09gevaertsok, I'll provide one then :)
19:58:13brettnemmore so if the commit cleanly, but yeah, I can probably handle it. :)
19:59:36*TheSeven thinks powering off the backlight before shutting down the LCD is a good idea
19:59:37gevaertsbrettnem: gevaerts/nohid.patch">
20:00:25bertrikTheSeven, you're getting the blooming effect :) ?
20:00:42gevaertshm, wait
20:00:45brettnemeasy enough.. thanks for the patch
20:00:47TheSevenI'm getting a brief white flash while powering down
20:01:11gevaertsbrettnem: wait a bit, I'm not entirely sure
20:01:18 Join esperegu [0] (
20:01:25linuxstbTheSeven: That's not a panic? It worked nicely on my Nano when I tried it last night.
20:01:29brettnemok, I haven't applied yet.. just let me know. thanks for your help
20:01:59gevaertsthat's more solid
20:02:01TheSevenlinuxstb: current svn is still shutting down the lcd in power_off()
20:02:40brettnemok, that makes more sense to me.. who knows what happens if the def is undef
20:04:07brettnemsweet.. building..
20:04:17brettnemhopefully it's using the right compiler. :)
20:05:53kugelnotlistening: yes that's something I need to fix
20:07:02TheSevengah, backlight_off() fades the backlight
20:07:45bertrikhow many "clicks" are there for one full rotation of the e200v2 and fuze scrollwheel? (and how many scroll-wheel pin changes per click?)
20:08:20kugelbertrik: for the latter, it's 2/4 for e200v2/fuze
20:08:21notlisteningno worries kugel just though it was me
20:08:53kugelTheSeven: nano2g has the fading already?
20:09:42notlisteningdid you make it more sensitive as i do not even have to scroll it now?
20:10:59kugelwell, i changed it to not lose acceleration due to a missed read. while that works well on my fuze, it seems to decrease accuracy for slow wheel movement
20:12:55*bluebrother wonders if it makes sense to lupdate the rbutil translations every now and then even if not translation was updated.
20:13:24brettnemgevaerts: Am I correct in assuming that the devices original (manufacturer) firmware has nothing to do with this issue? Because the original firmware does NOT work with Tiger.
20:13:47gevaertsbrettnem: that's right, yes
20:16:35notlisteningkugel, well just to highlight that when using it with speech it is quite hard.
20:16:37CtcpPing from gevaerts!n=fg@rockbox/developer/gevaerts
20:17:18kugelnot only with speech :)
20:17:27 Quit esperegu (Remote closed the connection)
20:17:34brettnemgevaerts: Whoooot! I see my e250 in Finder!
20:17:43 Join esperegu [0] (
20:18:00notlistening;) ok
20:25:11TheSevenkugel: it has hardware-driven fading
20:27:35kugelTheSeven: no way to control or monitor it?
20:28:16TheSevenwe can control it's speed, but we can't set absolute brightness levels (yet)
20:28:30TheSeventhere seem to be absolutely no docs on that chip
20:28:46TheSevenand as apple doesn't use it, we can't use it, even though I'm quite sure the chip supports it
20:32:15CIA-43New commit by bluebrother (r23052): Update german rbutil translation.
20:33:05kugelTheSeven: so only 1 backlight brightness level?
20:34:31TheSeventhere is a power-off/power-on reg and a fade time reg
20:35:34*bertrik is still confused about the backlight control method in the samsung yp-s3 and one of the meizus
20:35:35brettnemok.. so new question
20:35:47brettnemI've got a mpeg-4 video and it won't play.. how do I diagnose?
20:36:04bertrikit seems to set the backlight brightness by a bunch of quick pulses in succession, but PWM'ing also seems to work
20:36:15bertrik(it = the original firmware)
20:37:40brettnemmaybe I need a recommendation for encoding on os x?
20:38:41linuxstbbrettnem: ffmpeg (and its various front-ends) work well for Rockbox video encoding. IIRC, there's something called ffmpegx
20:38:53brettnemsweet.. thanks
20:39:05brettnemsorry, I just found that on the wiki btw
20:39:06linuxstbThe PluginMpegplayer wiki page should have the info you need.
20:39:13brettnemyeah, just landed there..
20:39:24*linuxstb will leave brettnem to the wiki then
20:39:37CIA-43New commit by kugel (r23053): Fix a few problems with RTL, statusbar and custom ui viewport. ...
20:40:53TheSevenis it ok if i make lcd_shutdown power off the backlight?
20:41:02TheSevenseems to be the least-invasive fix for that white flash
20:41:49***Saving seen data "./dancer.seen"
20:42:48kugelthat's probably ok, although other targets don't
20:42:57kugeldon't *do this
20:46:45kugelTheSeven: what's the exact problem? I'm not sure whether other targets need to turn of the lcd explicitly on shut down, but the current code works for them
20:48:57midgeyi think we intentionally don't turn off the lcd on ipods
20:49:21midgeythe "flash" is a clear way for the user to know the device actually shut down
20:50:57 Join Zarggg [0] (
20:51:18brettnemI'm confused.. here's a painful noob question.. MPEG4 or MPEG2?
20:54:32brettnemok.. we'll try that out
20:54:49CIA-43New commit by theseven (r23054): Reduce iPod Nano 2G NAND timeouts to a sane value (20ms)
20:55:29TheSevenkugel: the flash will happen long before it actually shuts down, so that's not a proper indicator
20:55:53n1swow HZ / 1 :)
20:56:00kugeldid the timeout changing actually help now?
20:56:25TheSevendidn't play around too much, but rockbox can at least still access the flash :-)
20:56:47TheSevenn1s: in fact that was a debugging leftover that went unnoticed
20:57:11kugeldoesn't it cancel operation after 20ms now? I'm sure data transfers take longer?
20:57:23amiconnTheSeven: lcd should not fiddle with backlight, as that's the backlight thread's task
20:58:10 Join faemir [0] (n=faemir@
20:58:35TheSevenamiconn: there are 2 alternatives: having the lcd do weird flashing when powering off, or adding another function to the backlight api, which will expose a nano2g-specific issue to other targets
20:59:18TheSevensee FS #10668
20:59:52 Quit pamaury ("exit(*(int *)0 / 0);")
21:00:42TheSevenlinuxstb: what's your opintion on that flyspray?
21:02:32shaiHi :) How do I start the iPod firmware (ie. dual boot) when I have Rockbox installed?
21:02:51TheSevenwhich ipod?
21:02:52linuxstbTheSeven: If you defined storage_flush() to be "(void)0", do you still need NEED_STORAGE_FLUSH ?
21:03:32TheSeveni intended the first to allow others to call it after writing very important things, and the second to tell it that it should be done on shutdown
21:04:37amiconnThe flashing happens on 1st Gen Nano and 5th Gen ipids as well. Basically on all colour lcds which are normally-white
21:05:03 Join merbanan [0] (
21:05:19TheSevenwhy don't you just power off the backlight before? (in shutdown_hw)
21:05:58linuxstbTheSeven: I'm not sure I understand why you need both - i.e. either use a #ifdef and only define (and call) storage_flush if that is true, or define it as "(void)0;" for other targets, and always call it.
21:07:18TheSeveni intended that for the case that a flush may be useful after writing important data, but should not be regularly done on shutdown, i.e. the function is implemented, but that define not set, expecting that a future target may want to do such a thing
21:07:32TheSevenbut if you feel that's unneccessary I'm happy to drop it
21:08:58 Quit n1s (
21:08:58 Quit brettnem (
21:08:58 Quit ender` (
21:08:58 Quit Tristan (
21:08:58 Quit ch4os (
21:08:58 Quit tha (
21:08:58 Quit pjm0616 (
21:08:58 Quit advcomp2019 (
21:08:58 Quit GodEater (
21:08:58 Quit Topy (
21:08:58 Quit mt (
21:08:58 Quit elcan (
21:08:58 Quit Xerion (
21:08:58 Quit BlakeJohnson86 (
21:08:58 Quit togetic (
21:08:58 Quit lyngaas (
21:08:58 Quit Utchybann (
21:08:59 Quit Kohlrabi (
21:08:59 Quit shodanX (
21:08:59bluebrothershai: the IpodFAQ wiki page answers exactly this question
21:08:59amiconnkugel: Red!
21:09:16linuxstbTheSeven: So it means "NEEDS_STORAGE_FLUSH_ON_SHUTDOWN" ?
21:09:27TheSevenmore or less, yes
21:10:38NJoinn1s [0] (n=n1s@rockbox/developer/n1s)
21:10:38NJoinbrettnem [0] (
21:10:38NJoinender` [0] (
21:10:38 Join GodEater [0] (n=bibble@rockbox/staff/GodEater)
21:10:38NJoinTopy [0] (
21:10:38NJoinTristan [0] (
21:10:38NJoinmt [0] (n=MTee@rockbox/developer/mt)
21:10:38NJoinelcan [0] (
21:10:38NJoinXerion [0] (
21:10:38NJoinBlakeJohnson86 [0] (
21:10:38NJointogetic [0] (n=togetic@unaffiliated/ibuffy)
21:10:38NJoinlyngaas [0] (
21:10:38NJoinUtchybann [0] (
21:10:38NJoinadvcomp2019 [0] (n=advcomp2@unaffiliated/advcomp2019)
21:10:38NJoinch4os [0] (n=ch4os@gentoo/user/ch4os)
21:10:38NJointha [0] (
21:10:38NJoinpjm0616 [0] (n=user@
21:10:38NJoinKohlrabi [0] (
21:10:39NJoinshodanX [0] (
21:10:46linuxstbTheSeven: I'm just looking at the storage code now - I'm not sure what is the "right thing"...
21:11:17kugelof course, a player red ;)
21:12:10 Quit Kohlrabi (
21:12:10 Quit togetic (
21:12:10 Quit Utchybann (
21:12:10 Quit shodanX (
21:12:10 Quit Topy (
21:12:10 Quit elcan (
21:12:10 Quit BlakeJohnson86 (
21:12:10 Quit lyngaas (
21:12:10 Quit Xerion (
21:12:10 Quit mt (
21:12:10 Quit GodEater (
21:12:10 Join Kohlrabi_ [0] (
21:12:10 Join GodEater_ [0] (
21:12:21 Join Utchybann [0] (n=lolo@
21:12:23 Part brettnem
21:12:31NJoinTopy [0] (
21:12:37pixelmakugel: you should have learned by now... :P
21:12:37 Join Xerion [0] (i=xerion@
21:12:39NJointogetic [0] (n=togetic@unaffiliated/ibuffy)
21:13:06 Join mt [0] (n=MTee@
21:13:18 Quit Thundercloud (Remote closed the connection)
21:14:06 Join Dege [0] (i=Dege@
21:16:13linuxstbTheSeven: I guess using the #ifdef is safer. A comment next to that define in config-ipodnano2g.h would be nice though.
21:16:13 Quit bluebrother (Nick collision from services.)
21:16:16 Join bluebroth3r [0] (n=dom@rockbox/developer/bluebrother)
21:16:35CIA-43New commit by peter (r23055): Correct a recording screen usability issue created in r21034. Pause used to start too, then was set to only start and now only pauses. Rec used to ...
21:16:36TheSevenok, I'll add this
21:16:59NJoinshodanX [0] (
21:17:24CIA-43New commit by kugel (r23056): Fix player red and remove an obsolete function call.
21:17:44 Join esperegu_ [0] (
21:17:52 Join AndyI [0] (n=pasha_in@
21:18:08peturbugger... next time I should check we're all green before committing :/
21:18:57 Quit esperegu (Read error: 104 (Connection reset by peer))
21:19:50NJoinelcan [0] (
21:22:16NJoinlyngaas [0] (
21:25:26TheSevenlinuxstb: are the other changes OK in your opinion?
21:26:09 Quit Galois (Read error: 110 (Connection timed out))
21:26:43 Join froggyman [0] (
21:29:39 Join efyx_ [0] (
21:31:53 Part Grahack
21:33:08 Join chandoo [0] (
21:33:32 Join Xerion_ [0] (
21:34:42 Quit notlistening ("Leaving")
21:34:47NJoinBlakeJohnson86 [0] (
21:34:48 Quit Xerion (Read error: 145 (Connection timed out))
21:34:48 Nick Xerion_ is now known as Xerion (
21:35:12linuxstbTheSeven: I'm not sure the backlight power-off in lcd_shutdown is right, but I'm not sure where else it could go.
21:36:27 Join tomers [0] (
21:36:32 Join Hillshum [0] (
21:37:13bluebroth3ris there a shortcut for background color in foswiki?
21:37:25 Nick bluebroth3r is now known as bluebrother (n=dom@rockbox/developer/bluebrother)
21:37:44*Hillshum got compiling rbutil
21:37:55 Quit AndyIL (Read error: 113 (No route to host))
21:40:20bluebrotherHillshum: have you built that tree before?
21:40:31HillshumDon't think so
21:41:14bluebrotherlooks like you're missing the libspeex header though it should use our own copy then.
21:41:38bluebrotheror is libspeex installed in a non-standard path on your system?
21:42:44bluebrotheroh, and are you building a static binary?
21:43:20bluebrotherhmm, scrap the last question. Building a static binary will always use our own speex
21:43:31shaiCan I remove the entire playlist under the Playlist sub-menu?
21:44:05shaiIt currently seems to hold files I removed while I was in the iPod software... and they error out cuz they don't exist.
21:44:12Hillshumbluebrother: libspeex-dev is installed
21:44:38bluebrotherHillshum: interesting. Have you tried a clean build?
21:45:04Hillshumof rbutil? I thought this was clean
21:45:22 Quit togetic (Read error: 104 (Connection reset by peer))
21:45:39 Join togetic [0] (n=togetic@unaffiliated/ibuffy)
21:46:02bluebrotherHillshum: ok, so this is a fresh checkout, correct? Then this shouldn't happen :o
21:47:24bluebrotherdoes building rbspeex work? I.e. make in tools/rbspeex
21:48:04 Nick Strife1989 is now known as Strife89 (
21:48:16HillshumNo, same error
21:48:45bluebrotherhmm. What does "pkg-config −−libs speex" return and where is located? In /usr/lib?
21:49:51HillshumIt returns '-lspeex', don't know what that's supposed to mean
21:50:34TheSevenHillshum: that looks sane
21:50:37bluebrotherthat's the compiler option used for linking against libspeex.
21:50:54HillshumWhich I assume rbutil uses
21:51:28bluebrotheryes, and rbspeex too. That's the reason why building fails for you.
21:51:59bluebrotherthough I don't get why it fails if you have libspeex installed. What's the output of "pkg-config −−modversion speex"?
21:52:53bluebrotherhmm. My system is 1.2beta2, so I'd expect your version to definitely work.
21:53:36 Join GeekShadow [0] (n=Antoine@reactos/tester/GeekShadow)
21:54:13bluebrotherdo you have a file /usr/lib/
21:55:17*bluebrother is puzzled
21:55:17TheSevenwhat do you think, should I commit FS #10668 or wait for further input?
21:55:30HillshumTheSeven: Yes
21:55:57*TheSeven is now puzzled, too
21:56:03TheSevenwhat did that "yes" refer to?
21:57:09pixelmalet gevaerts explain it to you (maybe in the other channel though ;) )
21:57:12HillshumTheSeven: You should either commit it or wait for further input
21:59:02linuxstbTheSeven: Given that you've now posted a patch, perhaps you should give it a day or so. Otherwise there would have been no point in posting the patch.
21:59:40TheSevenI somehow think it can't be worse than what's currently in SVN, but on the other hand it touches core stuff
21:59:53TheSeven(even though I can't see how that could hurt any other target)
22:02:23 Quit shai ("Leaving")
22:02:30linuxstbI guess that's the main point, so maybe just go ahead...
22:03:41TheSevenwell, that "maybe" is the issue ;-)
22:03:54 Join Thundercloud [0] (
22:04:13HillshumShould I reinstall libspeex?
22:04:41 Quit merbanan (Read error: 145 (Connection timed out))
22:05:00linuxstbTheSeven: This is probably one of those commits that no-one will shout about until it's committed, and even then, I doubt anyone will...
22:06:51*TheSeven just proved there is indeed codec icache cleaning trouble
22:07:06TheSevenan mp3=>ogg transition results in an undefined instruction exception
22:08:20 Quit tomers (Remote closed the connection)
22:09:21amiconnTheSeven: Is there a reason why you ifdef the call to storage_flush(), but not its definition?
22:09:50amiconnThis will waste a bit of binsize on targets not needing it
22:10:01kugelTheSeven: see mmu-arm.h
22:10:24TheSevenamiconn: I discussed that above. Brief: It may be defined (and thus callable after i.e. very important writes) even though we don't want to call it on shutdown
22:10:53TheSeventhis is not a HAVE define, but a NEEDS (on shutdown) one
22:11:06TheSevenan additional HAVE would be an idea, too
22:11:08kugelrockbox doesn't have very important writes
22:11:28amiconnYes, but this whole storage_flush business only makes sense if we're caching something one way or another
22:12:18TheSevenso what would you propose?
22:12:35 Quit JackWinter (Read error: 131 (Connection reset by peer))
22:12:40amiconnI agree that it makes sense on "naked" flash targets, but certainly not on ata targets or card standard based flash targets
22:12:52TheSevenfor which it is just defined as void(0)
22:13:17amiconnThe individual functions are, but not storage_flush() itself
22:13:49TheSevenoh, right
22:13:49 Join krazykit [0] (
22:13:54 Quit esperegu_ (Read error: 104 (Connection reset by peer))
22:14:09TheSevenso what about renaming that to HAVE_STORAGE_FLUSH and ifdefing storage_flush itself?
22:14:10amiconnIt essentially collapses to storage_flush() { return 0; }, but that's still >0 bytes
22:14:58*amiconn thinks that would be better
22:19:49*amiconn wonders how flash memory cards handle wear levelling efficiently without prior notice regarding shutdowns
22:20:14TheSevenby doing the cleanup on the next boot more efficiently than whimory
22:20:45TheSeveni.e. not committing everything on that boot, but only reconstructing their state
22:21:03TheSevenon the iphone, whimory is doing that, too, but the nano seems to use an older version
22:21:15 Join JackWinter [0] (
22:23:03linuxstbThis sounds like a reason why it would be desirable to replace the OF bootloader in NOR...
22:23:11 Nick Kohlrabi_ is now known as Kohlrabi (
22:23:57TheSevenwhich *will* mean terribly increasing the risk of bricking an ipod through FTL damage, as there will be no norboot to clean it up
22:24:10TheSevenbut it also means that we could drop that FTL altogether
22:24:26linuxstbWell, we would obviously take steps in our replacement NOR code to prevent that.
22:24:40TheSevenif you take steps, you'll need to stay with whimory
22:25:14*linuxstb isn't saying it's easy to replace the NOR code, just desirable... ;)
22:25:34TheSevenI wouldn't do it before we have at least a way to enter DFU without opening the iPod
22:25:59TheSeventhere seems to be such a way, but I haven't worked it out yet
22:26:13TheSevenbut i have caused it to enter DFU accidentally 3 times now
22:26:20TheSeven2 times from the OFW, 1 time from ibugger
22:26:48TheSeventhis probably is related to a GPIO, but we don't know what's driving that one#+
22:26:57 Join froggywoman [0] (
22:29:39linuxstbTheSeven: I still can't get my "crypt_firmware" plugging behaving properly... Should I be padding the image in some way?
22:30:00TheSevenyes, to at least 64 bytes I think
22:30:09TheSeven(because of that HMAC-SHA1 or whatever it is)
22:30:18linuxstbYou mean to a multiple of 64 bytes?
22:30:36 Quit froggywoman (Client Quit)
22:31:54 Quit Sajber^ (Read error: 104 (Connection reset by peer))
22:36:11CIA-43New commit by theseven (r23057): Core changes to allow storage drivers to do cleanup on shutdown, and iPod Nano 2G shutdown code rework (FS #10668)
22:37:48*TheSeven will now fix icache cleaning
22:39:36 Quit einhirn ("Miranda IM! Smaller, Faster, Easier.")
22:40:03TheSevenshould i fork mmu-arm.c, use ifdef S5L870X or add a new define for CPUs without TTB?
22:41:50***Saving seen data "./dancer.seen"
22:42:06linuxstbmmu-arm.S ?
22:42:40*TheSeven is currently analyzing what's different in that IMX31L block
22:42:54*linuxstb was about to ask how the S5L870X code would differ to that
22:43:04 Join einhirn [0] (
22:43:10TheSeventhe code below looks far more optimized
22:43:39*TheSeven will just try to use that and see what happens
22:45:11TheSevenis there any define i need to set to make code actually use them?
22:46:09 Quit bmbl ("Bye!")
22:47:06TheSevendata/rockbox-trunk/firmware/target/arm/mmu-arm.S:44: Error: selected processor does not support `mcrrhi p15,0,r1,r0,c14'
22:47:11midgey -rf *
22:47:50 Quit Hillshum (Read error: 110 (Connection timed out))
22:48:04TheSevenlinuxstb: so that's why i didn't use it the last time
22:48:12TheSevenwhat the heck does that additional R mean?
22:52:41amiconnMoves two regs at once, see arm reference manual
22:53:07amiconnThat's only supported on ARMv5TE or higher
22:59:31TheSeven[22:45]<TheSeven>is there any define i need to set to make code actually use them?
23:00:32kugelmc2739: ping
23:01:32linuxstbTheSeven: See firmware/export/system.h
23:02:09kugelTheSeven: you need to include mmu-arm.h
23:02:20kugelin your system-target.h file
23:02:36kugelthat will define the defines linuxstb mentioned for you
23:03:29kugelsaratoga: ping
23:03:38*kugel needs a e200v2 user
23:03:53 Join Galois [0] (
23:06:22*TheSeven still gets an undefined instruction exception
23:07:09kugeldid you do a complete recompile already? :P
23:07:35TheSevenno, but that's my next step (after checking that this define is really set)
23:10:11 Quit midgey ()
23:10:39 Quit petur (Read error: 104 (Connection reset by peer))
23:11:06 Join petur [0] (
23:12:46 Quit jgarvey ("Leaving")
23:17:29TheSevenkugel: indeed.
23:17:45 Quit einhirn (Read error: 104 (Connection reset by peer))
23:17:47TheSevenat least part of the problem got fixed by recompiling everything
23:19:55TheSevenchanging codecs works now, so the cache part of the issue is probably fixed
23:20:20TheSevenchoosing a different track from the file browser still freezes though
23:21:52 Quit petur ("Zzzzz")
23:24:05*kugel wonders which dependency problem makes certain thinks only work after a complete recompile
23:24:41TheSevenkugel: the next time I suspect such a thing, I'll backup, rebuild and diff, to find that out
23:25:19amiconnThere seems to be a scrolling text bug which I suspect a certain svn revision
23:26:05amiconnIf the scroll step size is >1, it seems to scroll a few pixels beyond the end of the text, showing the left part of the leftmost letter at the far right
23:26:55amiconnmt ^^
23:28:27TheSevenany ideas what else could be causing these lockups?
23:31:45 Join StealthyXIIGer [0] (
23:39:13 Quit chandoo (Read error: 110 (Connection timed out))
23:41:59CIA-43New commit by theseven (r23058): Added S5L870X cache coherency support
23:42:57*TheSeven is wondering where to poke around next
23:44:09amiconnThe freeze?
23:44:33amiconnThat doesn't happen on other targets, so the problem most probably resides in nano2g specific code
23:46:05TheSevenor s5l870x specific code, as none of the other ports has reached that stage yet.
23:46:42TheSevenin fact, there is no nano2g specific audio code
23:46:49TheSeventhat's just wmcodec + s5l870x dma
23:47:21amiconnI don't think that the problem hides in the audio code
23:47:33TheSevenbut where else?
23:47:45amiconnI'd suspect a nand related deadlock
23:47:53TheSevenin fact, the following comments in pcm-s5l8700.c look pretty suspicious:
23:47:55TheSeven - pcm_play_dma_pause is untested, not sure if implemented the right way
23:47:56TheSeven - pcm_play_dma_stop is untested, not sure if implemented the right way
23:47:57TheSeven - pcm_play_dma_get_peak_buffer is not implemented
23:49:15*TheSeven suspects pcm_play_dma_stop
23:49:51amiconnDoes pause work properly?
23:50:48amiconnHmm, and if you stop playback and the start something else it does not freeze?
23:51:16TheSevenit does freeze
23:51:31TheSeventhe first file selection works, the second freezes
23:51:43amiconnThe missing peak implementation means the peakmeter and the plugins relying on peak data (vu_meter and oscilloscope) don't work, but that's not a reason for freezing
23:51:49 Join chandoo [0] (n=chandoo@
23:52:14amiconnThen that makes pcm_play_dma_stop suspicious
23:52:34gevaertsit seems to make TheSeven suspicious as well :)
23:53:43TheSevengah, just hit that wmcodec freeze again while testing
23:54:53TheSevenand what i just saw rather points into the direction of storage
23:55:30TheSeveni managed to escape the now playing screen and look at the buffers in that "frozen" state, they are all empty, and as soon as I try to do anything that accesses the flash, it locks up the main thread
23:58:34TheSevenok, it's even more complicated

Previous day | Next day