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 2009-11-10

00:00:00sansaAMSI just got rockbox running on my ams sansa and it works perfectly, it does everything i want it to and more, good job rockbox team people.
00:01:08 Join antil33t [0] (
00:01:27TheSevenbut even without any mdct it's slower than flac :-P
00:01:30*kugel can't really see how this can happen
00:03:07kugelpixelma: I assume it crashes somewhere in apps/gui/usb_screen.c
00:03:39 Join AEnima1577 [0] (
00:03:47 Quit sansaAMS ("CGI:IRC (EOF)")
00:03:49pixelmakugel: and another thing (already told you but) - changing to the WPS looks really weird when it happens slow (either a very complicated WPS or a slow target) because you still see the list and then the bitmaps and progressbar are drown on top of it, every textual info that does is not the file's meta data (and some dummy things for the meta data), when the metadata gets filled out everything seems to be redrawn and only then looks correctly
00:04:11pixelmadrawn too
00:05:12gevaertspixelma: if you still have the .elf file, objdump can be helpful to find a more accurate location
00:05:12kugelwe can't really help that
00:06:41pixelmaa simple clear screen wouldn't be possible? And it worked quicker not that long ago (at least this is how it felt)
00:07:23pixelmathe list showing through looks really nasty
00:08:03pixelmagevaerts: can you help me using it? I think I've been told once already but forgot
00:08:16Utchybanngevaerts: I have another trivial patch when SECTOR_SIZE!=512 FS #10770.
00:08:40*kugel finds a blank screen more ugly
00:09:09 Quit robin0800 (Read error: 104 (Connection reset by peer))
00:09:14gevaertspixelma: sure
00:09:46 Join robin0800 [0] (
00:10:38 Quit Omlet05 ("( :: NoNameScript 4.22 :: )")
00:10:48 Quit ender` (" It's all fun and games until someone loses an eye. Then it's fun and games you can't see.")
00:11:20 Quit flydutch ("/* empty */")
00:11:48 Quit panni_ (Read error: 113 (No route to host))
00:12:22 Join Tomis [0] (n=Tomis@
00:12:28 Quit Tomis (Remote closed the connection)
00:12:32gevaertspixelma: "sh-elf-objdump −−syms rockbox.elf | sort" should output the list of addresses and symbols, including static functions
00:13:23pixelmakugel: than that? ... don't believe you
00:14:40pixelmait looks nonprofessional
00:15:55pixelmaI forgot in my description before that you still see the statusbar in the list and half of the "loading" splash (in my case)
00:16:00 Quit antil33t1 (Read error: 110 (Connection timed out))
00:16:09JdGordon|kugel: fyi, me and kkurbjun had a bit of a talk yesterday about statusbars still being broken... the logic in viewport.c breaks remote statusbar if main is custom and remote is "top"
00:16:13gevaertsUtchybann: ( 2048 / ( SECTOR_SIZE / 512 )) looks a bit weird I think. Why not just 1048576/SECTOR_SIZE ?
00:16:27 Quit freqmod_qu (Read error: 110 (Connection timed out))
00:16:48gevaertsor 1024*1024/SECTOR_SIZE
00:17:57 Quit HBK (Read error: 131 (Connection reset by peer))
00:18:12gevaertsso in total p->size * SECTOR_SIZE / 1048576
00:18:53TheSevengevaerts: that looks likely to overflow
00:19:03 Join HBK [0] (
00:19:42 Join antil33t1 [0] (
00:19:48 Quit antil33t (Nick collision from services.)
00:19:54 Join panni_ [0] (
00:20:17gevaertsp->size*( SECTOR_SIZE/512)/2048 ?
00:20:46gevaertsnearly the same as the original, so probably not worth it to change...
00:21:04TheSevenyes, that one looks good
00:21:08gevaertsTheSeven: do you know of other places that are still wrong?
00:21:12TheSevenwhere is that 2048 coming from btw?
00:21:18TheSevenno I think that was the last one
00:21:36gevaertsdivide by 2048 converts sectors to megabytes
00:21:37Utchybanngevaerts: it's looks weird, but I try not to overflow.
00:21:52gevaertsUtchybann: I noticed by now :)
00:21:58pixelmagevaerts: the address is between cuesheet_subtrack_changed and open_voicefile (if I may put it this way)
00:22:48JdGordon|because its a NULL reference on wps_gui->data->"something"
00:22:52saratogaTheSeven: the wma 192k codec itself takes 14.28 MHz on PP5024 verses 22.2 on the Nano2g, so most of the slow down is due to the MDCT library
00:23:03JdGordon|which is being called from skin_refresh() which is being called with NULL data
00:23:06JdGordon|which shouldnt be happening
00:23:17*JdGordon| is 99% sure of that
00:23:24saratogae.g. wma itself is 0.6432 as fast as PP, but mdct is just 0.3023 as fast
00:23:24TheSevengevaerts: your solution above should be safe up to 1TB, while still being readable, so I guess it's the way to go
00:23:29TheSevenit may deserve a comment though
00:23:29saratogaso something is going badly wrong in mdct
00:23:56gevaertsTheSeven: is it better than Utchybann's original? p->size / ( 2048 / ( SECTOR_SIZE / 512 ))
00:24:18TheSevenit's better in terms of readability I think
00:24:32TheSevenUtchybann's may work with even bigger drives if they have big sectors
00:25:25gevaertsI think I'll go with the patch from FS then
00:25:43gevaertsan actually tested patch is always good :)
00:25:50TheSevensaratoga: where is mdct_backward coming from?
00:26:15saratogaTheSeven: apps/codecs/lib
00:26:27kugelJdGordon|: I'm fairly sure that it worked with the initial commit
00:26:34TheSevenah, in the codec lib
00:26:49CIA-6New commit by gevaerts (r23596): Also fix Debug->View partitions when SECTOR_SIZE!=512 ...
00:27:24Utchybanngevaerts: thanks.
00:28:45kugelJdGordon|: skin_refresh shouldn't even be called in early usb
00:29:31saratogaTheSeven: could you try benchmarking WMA again with mdct enabled but the mdct code (but not data) taken out of IRAM? wma192k is a fine sample for this
00:29:42*gevaerts thinks that a proper solution would just have used (p->size>>11)*(SECTOR_SIZE>>9) :)
00:29:48saratogajust remove the ICODE sections for mdct2.c and mdct_arm.S
00:29:58saratogaerr ICODE attributes
00:30:13kugel then again, cuesheet_subtrack_changed() shouldn't be called with id3 == NULL (there's logic to prevent that)
00:30:19pixelmaJdGordon|: I thought "what does WPS data have to do with the early USB screen" but since custom status bar WPS tags would be alloed ther too, right?
00:30:34*TheSeven wonders if the asm code for it is actually being used
00:30:41TheSeventhis could be the root cause
00:30:52pixelmaI wonder what it has to do in cuesheet_subtrack_changed at all
00:30:58kugelpixelma: no skin is loaded when early usb kicks in
00:31:20TheSevenhm, it should be, according to SOURCES
00:32:25saratogaTheSeven: I put in a #warning in SOURCES and it looks like it gets compiled
00:33:53TheSeveni already found (and fixed) some places where such things did check for a specific target or soc instead of an architecture
00:34:05TheSevenbut looks like this isn't the case here
00:34:25 Join robin0800_ [0] (
00:35:19kugelsaratoga: any chance TheSeven's iram commit can be changed to check iram size instead of socs/arcs??
00:35:33 Quit robin0800 (Read error: 104 (Connection reset by peer))
00:35:58pixelmakugel: I'd like to help but not today (it's already quite late - and testing on Ondio is always very tedious, especially if it involves flashing Rockbox as I at least need to put 3 files from different places into different folders and unzipping a full build takes ages
00:35:59TheSevenkugel: the problem is that the iram size is only defined in the linker script, not in a header
00:36:12TheSevenwe should really change that, but it's out of my reach right now
00:36:46*pixelma just had the idea of a "make flashzip" or something that would just package those 3 files for easy unzipping to the player
00:37:22kugelpixelma: you can install the main binary only too
00:38:44 Join robin0800 [0] (
00:38:51pixelmano that doesn't help for early USB because I need 3 files there - main binary, the flash file and the plugin for flashing...
00:38:54 Quit midgey ()
00:38:57TheSevenwith mdct moved out of iram we're still at ~300%
00:38:58amiconnkugel: No, because early USB is a flash thing
00:39:09pixelmabecause I need to flash it
00:39:13amiconnSo you need at least the main binary, the .ucl and rockbox:flash.rock
00:39:34kugelpixelma: I mean you don't need to unzip the full build (i.e. including plugins and codecs
00:40:09pixelmamain binary goes into root, rockbox.ucl into .rockbox, rockbox.flash.rock into .rockbox/rocks/viewers
00:40:17kugelthat .rock can be mv'd manually too
00:40:20amiconnFull builds don't include codecs on hwcodec...
00:40:24pixelmaI know, it's still tedious
00:40:43kugelyou could do make simple script
00:40:54amiconnpixelma: I always put the .ucl into the root when doing things like this (in fact I even move it after unzipping a full build)
00:41:10 Quit robin0800 (Remote closed the connection)
00:41:23pixelmaoh, didn't know that was possible
00:41:31amiconnWay easier to flash it this way than switching file view to all (normally have it set to "supported") and dig into the .rockbox folder
00:41:40 Join MG_Man [0] (n=MGMan@
00:41:49amiconnThe .ucl can be anywhere - rockbox_flash.rock is a viewer after all
00:42:18MG_ManHmm, I'm in a strange pickle here
00:42:47MG_ManAfter using rbutil on my h320 it still doesn't boot rockbox
00:43:09pixelmaamiconn: unless you don't have to switch the file view
00:43:39TheSevenhm, test_codec logfile is empty? :-/
00:43:43pixelmaMG_Man: did you update the firmware from within the original firmware?
00:44:07pixelmaI believe there were also reports of people who had to do that twice
00:44:17MG_ManI'll do so again then
00:45:22 Quit robin0800_ (Read error: 60 (Operation timed out))
00:45:38MG_ManSame deal
00:45:45MG_ManI wonder if 1.29E is unsupported or something
00:47:06TheSevensaratoga: dammit
00:47:13MG_Manoh, it seems it isn't
00:48:02kugelJdGordon|: I'm curious about r23575. You changed it to fall back to fullscreen? Why that. the info vp doesn't have any redraw problems
00:48:06TheSevensaratoga: 52.95MHz (96) / 53.33MHz (128) / 54.13MHz (192) / 54.95MHz (256) / 55.73MHz (320)
00:48:12TheSeven*FASTER* than IRAM!
00:48:31TheSevenfull report:
00:48:37JdGordon|kugel: ?
00:49:41TheSevenany ideas? :-/
00:50:00kugelJdGordon|: what do you not understand?
00:50:31JdGordon|which commit was that? im still at work
00:50:47kugellook it up (the commit message I mean)
00:52:19JdGordon|im at work...
00:53:05MG_ManStill refuses to boot rockbox...
00:53:32MG_Mannow with 1.28E
00:53:53kugelJdGordon|: you cant read commit messages at work?
00:53:59gevaertsMG_Man: you did also install rockbox, right, and not just the bootloader?
00:54:11JdGordon|i can if they are linked to me.... but im actually working
00:54:15JdGordon|sort of :p
00:54:52kugelJdGordon|: "Fix the error condition to actually set the fullscreen viewport. Also remove a comment which doesnt really add anything.. the comments 15lines above it explain whats happening"
00:55:54JdGordon|I remember agreeing that the fullscreen viewport made more sense as the error condition
00:56:05kugelwe concluded here ( "JdGordon|yeah, change it to %Vi") that %Vi is the best to pick if intersection fails. I'm not sure why you changed that
00:56:43JdGordon|oh wait...
00:58:22 Quit Randominality ("ChatZilla 0.9.85 [Firefox 3.6b1/20091029171059]")
00:58:33kugelJdGordon|: :?
00:58:52JdGordon|put it back if oyu want
00:58:54JdGordon|my bad
00:59:52kugelI also wondered why you just changed what I did very recently, intentionally and after a small talk about without asking back again
01:00:15JdGordon|serves you right for not being online?
01:00:25JdGordon|and what change?
01:01:11kugelthe change that the commit did
01:03:17TheSevensaratoga: still around?
01:09:40 Quit amiconn (Nick collision from services.)
01:09:42 Join amiconn_ [0] (i=quassel@rockbox/developer/amiconn)
01:10:02 Nick amiconn_ is now known as amiconn (i=quassel@rockbox/developer/amiconn)
01:10:53 Join pixelma_ [0] (i=quassel@rockbox/staff/pixelma)
01:10:53 Quit pixelma (Nick collision from services.)
01:10:54 Quit Rob2222 (Read error: 60 (Operation timed out))
01:11:13 Nick pixelma_ is now known as pixelma (i=quassel@rockbox/staff/pixelma)
01:12:29***Saving seen data "./dancer.seen"
01:12:39 Join Rob2222 [0] (
01:15:02MG_ManEven if it was just the bootloader, it jumps straight to the iriver, and there is a .rockbox folder
01:15:27MG_ManJust to be clear ,the USb is plugged into the DATA port, and the device in USB mode, correct?
01:16:52kugelTheSeven: is only managed by #ifdefs as well, so it seems to me as simple as moving all preceding defines in that file into a separate header
01:17:06MG_ManDone by pressing ON/PLAY twice while plugged in?
01:17:46MG_ManMD5 hashes of both the unmodified firmware and the one on the iriver root is the same, so for whatever reason rbutil didn't do it's thing
01:18:54MG_ManShould I try installing just the bootloader instead of Full Installation?
01:21:32 Join krabador [0] (
01:26:57 Join stripwax [0] (
01:30:32 Quit gevaerts (Nick collision from services.)
01:30:43 Join gevaerts [0] (n=fg@rockbox/developer/gevaerts)
01:34:34 Quit kugel (Remote closed the connection)
01:35:34 Quit mikroflops (Read error: 104 (Connection reset by peer))
01:35:44 Join mikroflops [0] (
01:45:16 Join dfkt_ [0] (i=dfkt@unaffiliated/dfkt)
01:45:36 Quit shai ("Leaving")
01:48:28 Quit JdGordon| ("Miranda IM! Smaller, Faster, Easier.")
01:51:38 Quit Stephen__ ("Leaving")
01:54:46TheSevenhm, executing from DRAM without caches, 1024 iterations of a 1024 instruction loop = 35979264 clocks
01:56:07 Part toffe82
01:56:22TheSeventhe same from IRAM = 18907136 clocks
01:57:27MG_ManDoes running rbutil from a rar instead of a hard drive have anything to do with this or something?
01:58:04MG_ManShould I try a different cache location?
01:58:37LambdaCalculus37Try running rbutil in a folder in your hard drive and not from a rar.
01:59:50 Quit Thundercloud (Remote closed the connection)
02:00:52 Quit dfkt (Nick collision from services.)
02:00:55 Nick dfkt_ is now known as dfkt (i=dfkt@unaffiliated/dfkt)
02:01:30 Quit jgarvey ("Leaving")
02:02:40MG_ManSTILL no effect!
02:03:13MG_ManEven after rbutil supposedly does it's thing, the H300.hex on the root of the iriver still has the same MD5 of the unmodified 1.28EU
02:03:26TheSevenexecuting from DRAM with icache, 1024 iterations of a 1024 instruction loop = 1163264 clocks
02:03:29MG_ManShould I try deleting it off of the player first?
02:03:40TheSeventhe same from IRAM = 1097728 clocks
02:03:58TheSevenlooks pretty much sane, any ideas?
02:05:36kkurbjunJdGordon: I have a new segfault with the wps...
02:06:07kkurbjunif I use the %cs tag it segfaults on the mrobe 500
02:06:34kkurbjunI'm not sure what the reason - offhand I would guess it is because it doesn't have all 4 screens
02:09:21MG_ManThere we go, I delted the .hex off of the player and the size looks about right now - 3,091KB
02:10:42 Join MethoS- [0] (n=clemens@
02:10:53TheSeven1024 iterations of writing 1024 words of IRAM ("str r0, [r3], #4") = 10698752 clocks
02:11:00TheSeven(dcache disabled)
02:11:59TheSevensame with DRAM = 23511040 clocks
02:13:01 Join CaptainKwel [0] (
02:13:37TheSeventhe same with reads instead of writes in IRAM = 19087360 clocks
02:14:07MG_ManWell alright, it works
02:14:11 Quit krabador (Read error: 104 (Connection reset by peer))
02:14:13TheSeventhe same with DRAM = 36143104 clocks
02:14:20MG_ManI guess something on Windows was locking up the file somehow
02:15:32TheSevensaratoga: summary:
02:15:35 Join Rob2223 [0] (
02:16:33TheSevenhas anyone done similar things on other ARMv4 processors?
02:16:58TheSevenI would have expected IRAM to be a little faster, but at least it is twice as fast as dram
02:17:30TheSevenwhy is the tremor imdct a little slower if executed from iram then?
02:17:38TheSeventhe caches seem to be doing their job anyways
02:18:48 Quit Kopfgeldjaeger (Read error: 104 (Connection reset by peer))
02:19:04 Join Kopfgeldjaeger [0] (
02:20:17TheSevenhm, time to go to sleep for me...
02:23:51amiconnTheSeven: I guess this iram is in fact integrated dram. That would explain the slowness
02:24:11amiconnThe PP IRAM is single-cycle SRAM
02:24:25TheSevenamiconn: hm, but why is it slower when running from *external* dram instead of *internal* dram?
02:24:44TheSevenargh, the other way round, of course
02:25:22amiconnSo it's as fast as the cache (and in case of the PP5002, it's in fact faster, because the cache controller has a bug and the iram controller does not)
02:26:22amiconnYour tests look like iram is faster than dram, just not very much
02:26:44TheSevenroughly factor 2
02:27:15TheSevenbut if i put the imdct into iram, it's slower than when it's running from dram
02:27:33amiconnAre caches enabled for iram (I guess they are)
02:28:16amiconnIf so, you might observe cache aliasing. I don't know how likely that is though, as it becomes less likely with increasing associativity of the cache
02:28:21TheSeveni used the same cache setup rockbox uses in my 3rd and 4th test (execution speed with icache)
02:28:21 Quit LambdaCalculus37 ("Leaving")
02:28:58TheSevenwe should have a 256-way set associative cache here
02:29:21amiconnCache aliasing is in some cases very prominent on coldfire (which has a direct mapped instruction cache)
02:29:47amiconnHmm, 256 way...
02:30:23TheSevennope, just looked it up, it's 64-way, but still enough
02:31:24TheSeven4 words per line, 4 banks of 64-way set-associative cache = 4KB, harvard style, both for icache and dcache
02:31:45 Quit Rob2222 (Read error: 113 (No route to host))
02:33:04*amiconn can't find where TheSeven stated the slowdown
02:34:01TheSeven (IMDCT in IRAM) vs. (IMDCT in DRAM) vs. (no IMDCT at all)
02:34:52amiconnHmm, the difference is rather small
02:34:52TheSevenit's not much difference, but it's the wrong way round...
02:35:03amiconnHow far away is iram from dram?
02:35:12 Quit courtc (K-lined)
02:35:13TheSevenyeah, i just thought of that
02:35:16*amiconn suspects longcalls
02:35:56amiconnIf caches are working well, using iram for code is probably a bad idea, as it introduces longcalls
02:36:01*TheSeven doesn't know where they should happen though
02:36:09amiconnUsing it for data wouldn't cause this kind of problem
02:36:43TheSeventhe imdct is all-iram, so there shouldn't be any longcalls in there
02:36:48amiconnIf iram is big enough to put *all* this codec's code there, you could verify that theory
02:37:02TheSevenand the code calling the imdct is probably using longcalls anyways as it is in a different object file
02:38:10TheSevenhm, the imdct calling the butterfly is probably also a longcall, but that one will also always be thre
02:38:21TheSeveni don't see where moving stuff around could introduce new longcalls
02:38:38saratogaisn't a long call just a couple extra cycles?
02:38:41TheSevenactually we observed it with vorbis that gcc *failed* to do longcalls when it would have needed to with static functions
02:39:08TheSevensaratoga: if it's happening when the imdct calls the butterfly, that may accumulate...
02:39:26amiconnTheSeven: I know. That's why I introduced this STATICIRAM macro back then
02:39:37 Quit MethoS- (Remote closed the connection)
02:39:51TheSevenso is it even dynamically deciding what's getting a longcall and what isn't?
02:40:01amiconnIt's a partial fake. If the target is an arm target and uses longcalls, those functions won't be static
02:40:05TheSevenit can't know where the segments are, as this is only definied in the lds
02:40:37amiconnNot in the classic arm abi. In fact the method to decide which functions need longcalls and which don't is rather braindead
02:41:02amiconn(and on top of that, buggy)
02:41:09TheSevenhm, so i can't see how any longcalls could ever arise from moving stuff around? either it's doing them anyways, or it'll fail, right?
02:42:13amiconnThat's why people are working on getting eabi to work.
02:42:34TheSevendoes eabi really fix this? how can it?
02:42:47TheSeventhis looks like a flaw of the split between compiler and linker to me
02:43:27amiconnIt is. eabi fixes this wrong split
02:43:28stripwaxany chance that random access is faster for dram than iram, but sequential faster for iram than dram - or would random reads in iram always outperform dram in the same way as that pastie link suggests for icode ?
02:43:35saratogaTheSeven: can you compute the absolute random access memory latency from those measurements you did with IRAM and DRAM?
02:44:32TheSevenwhat exactly do you mean with "absolute random access memory latency"?
02:44:40amiconnIirc the method in the old abi is that functions are longcalled if they (1) are in another translation unit (C file), or (2) are in the same translation unit, but in another section than .text (with the exception of static functions, which is imo a bug)
02:44:46TheSevenhow do you specify random access?
02:45:20TheSevenamiconn: how do they do it in eabi?
02:45:33 Quit stripwax ("")
02:45:37amiconn(2) probably hits you, since that means if a C file is compiled with ICODE functions, *all* functions within that file will also call each other by longcall, whereas they don't without ICODE
02:46:19amiconnWith eabi, all functions are compiled for shortcall. The linker inserts longcall veneers where necessary
02:47:13*TheSeven doesn't consider this a proper fix either, but at least it isn't quite as ugly as what it's doing right now
02:47:33 Join jgarvey [0] (
02:49:55 Quit jon-kha (Read error: 145 (Connection timed out))
02:49:58TheSevenso what we observed with c5000 is probably just a dramatic increase of the dcache miss rate?
02:50:28saratogaTheSeven: absolute random access == time to load a single word from DRAM/IRAM without caching or prefetch in nanoseconds or clock cycles
02:50:51saratogarandom in that it cannot be anticipated by a cache or prefetcher
02:54:19saratogait seems to me that since the ARM9 core is standardized, the only real difference between the nano2g and other arm cores is the memory
02:58:28 Quit MG_Man ("...GOT AWAY SAFELY!")
02:59:11TheSeven2048 iterations of 512 more-or-less randomly distributed reads: DRAM = 35913728 clocks, IRAM = 18874368
02:59:18TheSevenso not much difference to the sequential one
03:00:22TheSevenso it's around 34 clocks per DRAM access and 18 clocks per IRAM access
03:00:35saratogaat 300MHz?
03:01:08saratogaslow memory would go a long ways towards explaining the difference with PP
03:01:48TheSevenor, to be exact, 191692800Hz
03:02:44saratogathats actually even a little slower then AMS too IIRC
03:03:04TheSevenmy measurements aren't really accurate as there is of course some overhead of the loop and the measuring itself, and i use a timer that only ticks every 16384 CPU clocks, but they should be good enough for a ball park number
03:03:41saratogaprobably accurate within a clock or two
03:05:03TheSevenlet's just for fun do the same with the bootrom and the nor flash
03:06:01TheSevenbootrom speed = IRAM speed
03:07:19TheSevenand NOR is ridiculous: 205144064 clocks = 195 clocks per read
03:07:48saratogaflash is slow :)
03:07:56saratogaTheSeven: what do you clock the AMBA at?
03:08:18TheSevenand the APB at 48MHz
03:08:23TheSevenbut that shouldn't matter
03:09:00TheSevenand the core is running in async bus mode
03:09:25TheSeven(external DRAM is said to be 32MHz, I haven't verified this)
03:10:12nawksiram is in ipod?
03:10:13 Quit AaronM ("Emo Time In My Corner... //_-")
03:10:20 Join AaronM [0] (
03:10:26saratogai think all the ipods have iram
03:10:28nawksoh, onchip ram?
03:10:28saratogamost players do
03:10:58*TheSeven will finally go to sleep now, it's much too late again
03:11:43saratogaisn't async mode slower?
03:12:21saratogaalso if the DRAM is really 32MHz running it in sync with the AHB might be worth trying
03:12:23TheSeveni don't think we can clock it sync the way it is configured, and fastbus will be even slower for sure
03:12:33***Saving seen data "./dancer.seen"
03:12:44saratogais async the option where its an integer multiple of the AHB?
03:12:54TheSevenare there requirements on the AHB to FCLK ratio?
03:13:05TheSevenI think that's sync
03:13:18saratogatheres a mode where they're the same
03:13:21*TheSeven needs to check these things tomorrow
03:13:24saratogaa mode where they're an integer multiple
03:13:26TheSevensame = fastbus iirc
03:13:30saratogaand a mode where theres no fixed relationship
03:13:53TheSevenok, i'm out.
03:13:54saratogaideally you want the integer multiple one since it maximizes bus performance
03:14:00saratogasorry a fixed relationship!
03:14:16saratogaok talk to you later
03:16:00 Quit mikroflops (Read error: 104 (Connection reset by peer))
03:16:17 Join mikroflops [0] (
03:19:20saratogaTheSeven: (for the logs) theres a few registers controlling both DRAM and IRAM access times and clocks, its possible they may not have been initialized optimally, but I have no idea if you're already looked at them
03:24:03 Join FOAD_ [0] (
03:25:42 Quit dfkt ("-= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.")
03:29:45 Quit lyngaas (Read error: 110 (Connection timed out))
03:35:36 Quit hebz0rl ()
03:36:24 Quit FOAD (Read error: 110 (Connection timed out))
03:36:25 Nick FOAD_ is now known as FOAD (
03:37:48 Join lyngaas [0] (
03:44:48 Quit DerPapst ("Leaving.")
03:47:09 Quit AEnima1577 ("Leaving.")
03:52:01 Join jon-kha [0] (
04:02:33 Join mc2739_ [0] (n=mc2739@rockbox/developer/mc2739)
04:10:12 Quit jgarvey ("Leaving")
04:13:02 Quit Rondom (Nick collision from services.)
04:13:13 Join Rondom [0] (
04:16:24 Quit mc2739 (Read error: 110 (Connection timed out))
04:19:25 Nick mc2739_ is now known as mc2739 (n=mc2739@rockbox/developer/mc2739)
04:22:49 Quit linuxguy3 (
04:22:49 Quit kkurbjun (
04:22:49 Quit jvd (
04:22:49 Quit blithe (
04:22:49 Quit droidcore (
04:24:08NJoinblithe [0] (
04:24:18NJoinlinuxguy3 [0] (
04:24:24 Quit nawks (Remote closed the connection)
04:28:26 Join jvd [0] (n=syscrash@
04:28:28NJoindroidcore [0] (
04:32:27 Quit TheSeven (Nick collision from services.)
04:32:47 Join The_Seven [0] (n=theseven@rockbox/developer/TheSeven)
04:32:59 Nick The_Seven is now known as TheSeven (n=theseven@rockbox/developer/TheSeven)
04:35:24 Quit JdGordon ("Leaving.")
04:37:55 Join JdGordon| [0] (
04:40:09 Join Blue_Dude [0] (n=chatzill@rockbox/developer/Blue-Dude)
04:46:13CIA-6New commit by blue_dude (r23597): Crossfade: carved out crossfade related code with lots of HAVE_CORSSFADE conditionals, eliminated fade buffer on low memory targets
04:47:08JdGordon|thats only in the commit message right? :)
04:47:45Blue_DudeI hope so. :)
04:47:55NJoinkkurbjun [0] (n=kkurbjun@rockbox/developer/kkurbjun)
04:48:12Blue_DudeThis should be interesting.
04:53:15JdGordon|where is HAVE_CROSSFADE defined?
04:53:26JdGordon|or intentionally not done that commit?
04:53:31 Quit panni_ (Read error: 113 (No route to host))
04:53:39Blue_DudeMy commits always take longer than advertised. The build engine doesn't like me.
04:54:00Blue_DudeIt's defined in the config-*.h files I think. It's target dependent.
04:54:12kkurbjunJdGordon|: I am logging a bunch of bugs in a flyspray task associated with the custom statusbar and the wps in general
04:54:27kkurbjunI am getting a variety of things form graphic glitches to segfaults
04:54:37JdGordon|oh goody :)
04:54:43JdGordon|Blue_Dude: that was there already?
04:54:55Blue_Dudeskin tokens? I am amazed.
04:55:09JdGordon|of course
04:55:15kkurbjunkugel does not seem to be setup with a flyspray account or at least not a dev account - should I assign this to you for now?
04:55:22Blue_DudeThe define? Yes. I just expanded its scope.
04:55:51JdGordon|kkurbjun: na, if he title is obvious its enough
04:56:01JdGordon|I'll have a look later if i get my work done
04:56:07JdGordon|Blue_Dude: ok, cool
04:59:05Blue_DudeOh, frak, I know nothing about WPS tokens. Is there any way I can not reference that global setting and have it remain transparent to the user? Should the skin engine just ignore WPS crossfade tokens if it's not built for crossfade?
04:59:43Blue_Dudeapps/gui/skin_engine/skin_tokens.c line 770
05:00:04JdGordon|when is crossfade not enabled?
05:00:59 Quit AaronM ("Emo Time In My Corner... //_-")
05:01:24kkurbjunJdGordon|: I logged the bugs I have come across so far in FS #10771 : - I jsut came across those in a couple of minutes of testing so I'll add to that if I come across more
05:05:02 Join AaronM [0] (
05:05:48JdGordon|Blue_Dude: have the token return NULL if its not crossfade build
05:06:19JdGordon|that will make the token always mean "crossfade is off" on non crossfade builds
05:11:54 Join Addison__ [0] (
05:12:36***Saving seen data "./dancer.seen"
05:13:40 Quit Addison__ (Read error: 104 (Connection reset by peer))
05:13:55 Join Addison__ [0] (
05:15:13saratogaJdGordon|: crossfade takes a fair amount of memory, so the clip cannot use it
05:15:49saratogam200 as well I think
05:15:59JdGordon|and c200 apparently
05:16:11JdGordon|its also swcodec only?
05:16:26Blue_Dudeback now. OK, I'll try that.
05:17:18JdGordon|that will at least make it compile :)
05:18:18saratogaJdGordon|: i thought hwcodec had it too
05:18:22Blue_DudeIt should. I'll leave appropriate functionality to whoever does those things.
05:20:07saratogawhats next for playback?
05:21:42CIA-6New commit by blue_dude (r23598): HAVE_CROSSFADE defs: fix yellow and red
05:22:35Blue_Dudesaratoga: gonna keep working through pcmbuf until it's factored out. Then I'm going to work on playback itself.
05:23:37saratogaBlue_Dude: have you seen fs#10605? its not directly related to what you're doing, but its a bug fix for buffering.c, so you might be interested
05:27:19Blue_DudeHm. How do you mark a task to watch in FS?
05:27:57saratogatheres a watch button on the upper right IIRC
05:28:57Blue_DudeI found it.
05:29:36Blue_DudeI've marked it for later study. I'm not sure yet whether I'm going to get into the buffering engine unless I have to.
05:30:07 Quit addisonj12 (Read error: 110 (Connection timed out))
05:30:34 Join addisonj12 [0] (
05:31:07Blue_DudeOoh, look at that delta. Almost 1000 bytes of code, not including the fact that it saves another 8192 bytes of dynamically allocated buffer space. I can live with that.
05:34:55 Quit Blue_Dude ("ChatZilla 0.9.85 [Firefox 3.5.5/20091102152451]")
05:34:57 Quit JdGordon| ("Miranda IM! Smaller, Faster, Easier.")
05:35:10 Join JdGordon [0] (n=jonno@rockbox/developer/JdGordon)
05:37:02 Quit Addison__ (Read error: 110 (Connection timed out))
05:37:29 Join Addison__ [0] (
05:37:49 Join david_ [0] (
05:39:32JdGordonwell fuck.. of course early usb is going to segfault.... none of the gui stuff is ready yet
05:42:23 Quit addisonj12 (Read error: 60 (Operation timed out))
05:44:09 Join uflops [0] (
05:44:10 Quit mikroflops (Read error: 104 (Connection reset by peer))
05:46:54CIA-6New commit by jdgordon (r23599): hopefully fix "early usb"
05:47:41 Quit Horscht ("Verlassend")
05:48:23kkurbjunJdGordon: there's not an option to load different backdrops depending on what screne you are on is there?
05:48:35JdGordonthere is WPS and regular
05:48:52JdGordoncurrently there is no other option
05:48:52JdGordonhow much more would you like?
05:48:54kkurbjun:), I'm trying to use all the memory with the wps ;)
05:48:55JdGordonbearing in mind the HUGE amount of ram needed
05:49:37JdGordonIt would be very easy to load the bmp into the wps buffer and use it from there instead of the seperate buffers.. but really.. how many do ou actually want?
05:49:47kkurbjunwell, the problem is that I want to shade the area of the backdrop behind the ui viewport
05:50:02kkurbjunbut that looks bad if you go into a plugin that keeps the backdrop
05:50:13kkurbjunor if you go into the quickscreen or something similar
05:51:24kkurbjunand I can't do it with an image in the custom statusbar cause that will be overwritten
05:51:35JdGordonwell... each skin could have its own backdrop... it gets fun where both the sbs and wps specify one...
05:52:11kkurbjunoh, interesting, I didn't think about trying that
05:52:22kkurbjunboth the sbs and the wps would be specifying one
05:52:36JdGordonI mean.. thats a posibility... but thats not hoocked up yet
05:53:31kkurbjunit would be interesting if there was a way to load backdrop "patches" for the different screens
05:53:46kkurbjunthat would be alot more conservative with the space
05:53:53JdGordonor just give priorities... so the sbs would only show if no other skin had one set
05:55:04kkurbjunI was thinking you could have images that are the "patches" to the backdrop. When you go into a screen those patch buffers would be swapped with the standard backdrop. When you leave the screen they would be swapped back
05:55:19JdGordonthat sounds messy :)
05:55:19kkurbjunthat might be a bit complicated though
05:55:39*JdGordon is doing rec skins now... no sidetracking!
05:55:41LloreanHow much memory were you planning to dedicate to storing multiple backdrops, exactly?
05:56:29JdGordonALL OF IT!
05:56:34kkurbjunI was thinking 20 MB would be reasonable
05:56:48LloreanYou could give each viewport a mask color to blend with the backdrop, so you could "fade" portions of the backdrop (either lighten or darken depending on the theme choice)
05:57:07LloreanSo that it gives small viewports a graphical effect without using more images.
05:57:30 Part david_ ("Ex-Chat")
05:59:10kkurbjunLlorean, that could work too - although it might be difficult to recover the original since you would potentially loosing information with the fade
05:59:44kkurbjunand the effect I'm hoping to do is a curved edge that is darkened which would complicate an algorithm
06:02:52kkurbjunIf a viewport could use an image as a backdrop that might work or at least get close to the functionality
06:03:28kkurbjunI think that would be fairly doable but it would take alot of changes to the lcd drivers
06:04:38LloreanAnd an awful lot of RAM
06:05:38kkurbjunLlorean: well, it would be a skin/theme dependent thing if you did it without the reserved backdrops
06:05:51LloreanHow do you mean?
06:06:06LloreanUnless you're planning on dynamically allocating memory for the images, it has to be reserved.
06:06:15Lloreanin which case changing themes will require stopping playback, at the very least, if not a reboot.
06:06:37kkurbjunI thought jdgordons work dynamically allocated most of the images
06:09:02LloreanDoesn't it allocate them within a finite buffer still?
06:14:42LloreanSo you'd need to extend the buffer (skin buffer or wps buffer, what's it called these days?) if you wanted to allow quite a few more images to be used
06:14:46kkurbjunyeah, I was looking at that - currently I have room to spare int he buffer
06:15:00LloreanIt might be better to just say "themes require a restart now" and plan to allow them to use a lot more memory, at the user's request.
06:15:25LloreanThat way people can use as many backdrops (or partial backdrops in the form of images) as they'd like, at their own expense.
06:16:53kkurbjunyeah, and it would save a bunch of space for minimal themes that do not use images
06:17:40kkurbjunif the backdrops could be included in that space entirely it would potentially save a couple of megs on the mr500 for someone that does not use backdrops
06:18:25 Join mikroflops [0] (
06:18:54kkurbjunbackdrops or images in the theme that is
06:19:34LloreanIt'd also give theme designers a lot more freedom
06:19:43LloreanNo more "if you're one byte over, suddenly something doesn't work'
06:20:23kkurbjunyeah, right now I'm a bit constricted in terms of what I can do with this theme when you consider the custom statusbar also
06:21:21 Join addisonj12 [0] (
06:21:31JdGordonremember that the required amount of ram is not known untill after its successfully loaded
06:22:57LloreanThat's solvable though
06:23:27kkurbjunJdGordon: what about a hard limit of something like 1/4 of the total available buffer space
06:23:50 Join shai [0] (
06:23:52JdGordonyou cant reclaim unused ram
06:23:57JdGordonLlorean: how is that solvable?
06:24:34kkurbjunJdGordon: what about using the metadata on buffer stuff for the skin?
06:25:08LloreanJdGordon: At the very worst case, you read through everything once, discarding it as you go the first time but counting
06:25:22JdGordonMoB blocks are possibly moved around... thats very dangerous because pointers are used throughout the skin buffer stuff... if offsets were used instead you could compact the buffer.. but thats alot of work to fix
06:25:46JdGordonthats not really an ideal solution
06:26:11kkurbjunhow about malloc :-D
06:26:11LloreanYes, but it may still be better than forcing a permanent limit on theme size for each target
06:27:15JdGordonit could remember how much it used last time and use slightly more than that this time if its loading the same theme
06:28:21LloreanSo just reload again and again until there's enough allocated?
06:28:25Lloreanisn't that worse than just loading twice?
06:30:55 Quit mikroflops (Remote closed the connection)
06:31:41JdGordonno, you would have some known upper limit which is used if its not known
06:31:50 Join mikroflops [0] (
06:31:53JdGordonor a config value in the theme cfg
06:33:54kkurbjuna value in the theme cfg seems reasonable
06:37:48*JdGordon just about has rec skins working
06:38:55 Quit uflops (Read error: 110 (Connection timed out))
06:39:28JdGordonexcept now its the boring part of hooking up the tokens to actually be useful
06:39:52 Quit Addison__ (Read error: 110 (Connection timed out))
06:42:28 Join uflops [0] (
06:43:49 Quit daggett ("Ex-Chat")
06:45:23 Join BHSPitMonkey [0] (n=stephen@unaffiliated/bhspitmonkey)
06:50:10 Quit uflops (Remote closed the connection)
06:50:43 Join uflops [0] (n=yogurt@
06:53:20 Quit mikroflops (Read error: 60 (Operation timed out))
06:57:55LloreanJdGordon: Then you don't get the benefit of being able to use a smaller buffer for smaller skins.
07:01:17 Join AEnima1577 [0] (
07:01:34 Join mikroflops [0] (
07:06:27 Quit mikroflops (Remote closed the connection)
07:07:21 Join mikroflops [0] (
07:10:03 Quit uflops (Read error: 145 (Connection timed out))
07:12:01JdGordonLlorean: sure you do... if a smaller theme is loaded the setting will change
07:12:40***Saving seen data "./dancer.seen"
07:13:30 Quit mikroflops (Remote closed the connection)
07:14:07 Join mikroflops [0] (
07:16:14 Quit addisonj12 (Read error: 104 (Connection reset by peer))
07:16:26 Join addisonj12 [0] (
07:18:36pixelmahwcodec does not have crossfade because it can not decode two files at the same time
07:21:17 Quit mikroflops (Read error: 104 (Connection reset by peer))
07:22:12 Join mikroflops [0] (
07:27:43 Quit phanboy4 (Read error: 54 (Connection reset by peer))
07:30:51 Join phanboy4 [0] (
07:32:59 Quit CaptainKwel (Read error: 110 (Connection timed out))
07:33:44 Join bluebrother [0] (
07:34:19 Quit liar (Read error: 113 (No route to host))
07:37:33pixelmaJdGordon: unfortunately the early usb fix doesn't work, thanks for looking into it though
07:40:33 Quit phanboy4 (Read error: 54 (Connection reset by peer))
07:43:50 Join phanboy4 [0] (
07:47:28 Join stoffel [0] (
07:50:04 Quit bluebroth3r (Read error: 110 (Connection timed out))
07:59:10pixelmathe reported error address is almost the same as before
08:05:26 Join einhirn [0] (
08:07:49 Quit mikroflops (Read error: 110 (Connection timed out))
08:10:01 Join Rob2222 [0] (
08:13:39 Quit einhirn (Read error: 104 (Connection reset by peer))
08:16:37JdGordonah crud
08:16:45JdGordonI guess it was all wishful thinking though :p
08:17:49JdGordonpixelma: does the recv1 have early usb?
08:20:23amiconnAll archoses have early usb
08:21:27 Join Zagor [242] (n=bjorn@rockbox/developer/Zagor)
08:21:49amiconnIn fact early usb is built at least for all hardware usb targets, just that for most of them you won't see it, because you'd have to plug usb exactly at the right time
08:22:29JdGordonZagor: hey, can you add kugel to the list of devs in flyspray please? (and maybe give some more people permission to add devs also? )
08:22:33amiconnToo early, and you end up in bootloader usb, or some OF loader usb.
08:22:51JdGordonI guess i flash to flash it to get that mode though?
08:23:03amiconnEarly usb becomes interesting if rockbox is flashed, and the bootloader doesn't handle usb
08:23:28JdGordons/first flash/have
08:23:47amiconnYes, probably. Iirc the archos firmware will catch usb otherwise
08:24:20amiconnMy archoses were all flashed ages ago
08:24:25ZagorJdGordon: kugel is already in the dev group
08:25:28Zagoreh no, wrong dev group...
08:25:59JdGordonlost of people are missing from the "assign to" group
08:26:00 Join ender` [0] (
08:26:34JdGordonamiconn: ok, how do I flash the recv1?
08:27:04amiconnThere is a reason why I rewrote the respective chapter in the manual...
08:27:38JdGordonmanuals arnt for reading :)
08:28:27 Quit Rob2223 (Read error: 110 (Connection timed out))
08:29:06JdGordon"Use when flashing Rockbox (see section ??. In the ideal case, you’ll need this tool only once. For safety reasons you may wish to deletefirmware_flash.rock from /.rockbox/rocks once flashing is complet" :/
08:29:06 Join linuxstb [0] (n=linuxstb@rockbox/developer/linuxstb)
08:29:11JdGordonyou want me to fix this or not?
08:30:53amiconnOh, didn't see that reference
08:31:22amiconnAlso, deleting firmware_flash.rock probably isn't that helpful, as it will come back whenever you update rockbox
08:32:19 Join AndyIL [0] (n=pasha_in@
08:33:09JdGordonit will work with the charger plugged in right?
08:33:13JdGordonmy batteries are pretty shot
08:34:25amiconnThe recorder does not work at all with just the charger. It needs the batteries in sufficiently working condition
08:35:21amiconnBut if it boots properly and rockbox doesn't show the low bat warning, it should be possible to flash it as well. firmware_flash checks for low bat and refuses to flash in that case
08:35:40JdGordonright, all flashed and good
08:35:54JdGordonearly usb seems to work fine
08:36:59JdGordonoh.. bah.. I need to rebuild a flash image from svn?
08:37:28amiconnYou need to flash a current .ucl
08:38:05amiconnRight after flashing you'll have rockbox 3.2
08:38:31*JdGordon shouldnt be doing anything involving flashing in his current sleepy state
08:38:37JdGordonyes, i have 3.2 now
08:39:38amiconnRead 12.5.4
08:39:39JdGordonneed to rolo to current build to use the plugin :p
08:40:22amiconn[08:39:39] <amiconn> Read 12.5.4
08:41:03JdGordonyay! ok, panicf
08:41:29amiconnNot panicf, but UIE (unless you've hit a different bug)
08:43:13JdGordongcc will turn "int old_bars = early_usb ? 0 : viewportmanager_get_statusbar();" into what I need to happen right? i.e that function wont be called?
08:43:29JdGordonits not doing some magic and storing its return value regardless of early_usb?
08:43:40 Quit AndyI (Read error: 110 (Connection timed out))
08:47:21*JdGordon is king!
08:48:01CIA-6New commit by jdgordon (r23600): really fix early usb this time
08:52:30JdGordonthats not completly perfect... no bar is displayed at all... which means no clock or battery meter
08:52:44JdGordondunno how much time I can really be bothered fixing that with a bandaid though
08:53:23amiconnThat also means no disk access indicator on Ondio - bad
08:54:00JdGordonwait for the computer to safely unmount the disk and it shouldnt be a problme
08:55:26amiconnThe indicator is very helpful in usb mode, even if not *absolutely* necessary
08:55:38amiconnThe recorder has a LED, so no problem there
08:56:14amiconnBattery meter in turn is rather important on recorder and Player in usb mode
08:56:48*JdGordon hopes johnson & johnson doesnt sue us for using the word band-aid in the source code :p
08:56:51amiconnThis is because sustained disk access in usb mode will drain the batteries even with the charger connected
08:57:38*amiconn thinks the status bar is quite important in the usb screen, hence there should be a proper fix
08:59:42 Join Teyphas [0] (
09:00:44TeyphasHello guys! Can you help me :D I loaded ILoader and Rockbox. Rockbox was succsefull to run.. but i can't open games and just says can't open file bla bla bla...any solution?
09:02:09ZagorTeyphas: did you unpack the whole zip?
09:02:43 Quit AndyIL ()
09:03:20Teyphasi downloaded .rockbox putted on my ipod nano 2g. Can see them on removable disk directory .rockbox\rocks\games
09:04:39JdGordongrr.. the damn battery animation isnt animating.. and i dont think the time is updating
09:06:39JdGordonoh FFS
09:07:16 Join LinusN [0] (n=linus@rockbox/developer/LinusN)
09:08:02 Quit phanboy4 ("Leaving")
09:09:42 Quit saratoga (
09:10:50 Join petur [50] (n=petur@rockbox/developer/petur)
09:11:00NJoinsaratoga [0] (i=9803c6dd@gateway/web/freenode/x-ktupxllmowxfxavh)
09:12:43***Saving seen data "./dancer.seen"
09:12:44 Join maruk [0] (n=papier@
09:13:17JdGordon is a partial fix.... if early usb is really that important... go back to pre custom sbs then
09:14:56pixelmathat's a weird advice
09:17:39 Quit Zarggg (Read error: 110 (Connection timed out))
09:22:26 Quit Teyphas ("CGI:IRC (EOF)")
09:24:14 Quit Aratsu (Read error: 110 (Connection timed out))
09:24:47 Join Thundercloud [0] (
09:26:32 Join flydutch [0] (
09:35:01 Quit AEnima1577 (Client Quit)
09:37:58 Join DerPapst [0] (
09:45:00 Join B4gder [0] (n=dast@
09:47:23 Quit Thundercloud (Remote closed the connection)
10:01:19 Join liar [0] (
10:10:42 Join Tomis [0] (n=Tomis@
10:12:43 Quit reid05 (Read error: 110 (Connection timed out))
10:13:07 Join reid05 [0] (
10:42:40 Quit liar (Client Quit)
10:45:12 Quit daurn| (Remote closed the connection)
11:09:23 Quit DerPapst ("Leaving.")
11:12:44***Saving seen data "./dancer.seen"
11:14:56 Join einhirn [0] (
11:28:09 Quit BHSPitMonkey ("Ex-Chat")
11:34:24seaniI'm getting "codec failure" message playing anything in a UISimulator C200 build (r23593M-091110). I've reconfigured, "make clean"''ed tried "known good" mp3's etc - any pointers?
11:36:24gevaertsseani: this is a clean simulator build, i.e. you didn't copy .rockbox from a target build?
11:37:02B4gderand remember to 'make install'
11:39:17seanigevaerts: No, I started with a totally fresh co / subdirectory - but I think I missed "make (full)install" as B4gder suggests. Tsk. Ta!
11:40:30seaniYes that was it - oaf. Thanks gents.
11:47:48 Quit stoffel (Remote closed the connection)
11:55:37 Join DerPapst [0] (
12:08:44 Join liar [0] (
12:18:45 Join Teyphas [0] (
12:21:47TeyphasHello again, people. So..Anyone can help me with problem (games and apps isn't working) :( I only get blackjack running. I really like Rockbox, but i can't open files in Demos, Games and apps. Sorry if'm kinda noobish, i just installed it today :D Thanks for attention
12:22:45rasherB4gder: is there a good reason make doesn't do install for the simulator?
12:25:05 Quit DerPapst ("Leaving.")
12:28:05 Join Strath [0] (
12:28:22 Quit Strath (Client Quit)
12:41:15 Join dfkt [0] (i=dfkt@unaffiliated/dfkt)
12:47:10 Nick fxb__ is now known as fxb (
12:48:06 Quit Teyphas ("CGI:IRC (EOF)")
12:50:22 Join Grahack [0] (n=chri@
12:54:57B4gderrasher: I believe it is to enable us to install partly broken builds etc
12:54:59 Join bubsy [0] (n=bubsy@
13:12:47***Saving seen data "./dancer.seen"
13:13:41 Join stoffel [0] (n=quassel@
13:23:24 Quit Grahack (Read error: 145 (Connection timed out))
13:23:25 Nick YPSY is now known as Ypsy (
13:27:18 Join Bob_C [0] (
13:32:19 Quit einhirn (Read error: 104 (Connection reset by peer))
13:33:03 Join einhirn [0] (
13:37:50 Join teru [0] (
13:42:13 Join Bob_C__ [0] (
13:42:32 Quit Bob_C_ (Read error: 110 (Connection timed out))
13:49:12 Join mikroflops [0] (
13:53:58 Quit stoffel (Remote closed the connection)
13:56:51 Quit Bob_C (Read error: 110 (Connection timed out))
14:03:33 Join Cal_ [0] (
14:05:09Cal_hey xp won't allow me to name my rockbox folder .rockbox how do i fix this?
14:07:28 Join Blue_Dude [0] (n=chatzill@rockbox/developer/Blue-Dude)
14:07:44ZagorCal_: use the command line
14:09:09 Join Dgby714 [0] (
14:09:11Cal_wtf ms that worked...
14:10:24Cal_in the error on boot my ipod it said rockbox.ipod missing, leave it change it to .rockbox or .rockbox.ipod?
14:11:02Cal_lol upon booting...
14:11:23Zagor.rockbox is the directory in which rockbox.ipod is located (among many other files)
14:11:51Cal_right thank you uber rockbox master
14:16:29pixelmaCal_: how did you put your rockbox build onto your Ipod? Just unzipping to the root should give you the correct folder structure, no renaming or moving needed
14:18:06 Join kugel [0] (n=kugel@rockbox/developer/kugel)
14:18:36Cal_unzipped with izarch, xp doesnt allow dots at the begining of files aperntly, c prompt work to ren
14:18:55B4gderxp does allow it, must be a stupid unzip program
14:19:57Cal_dunno why either izarch or xp did not perserve the dot upon file creation when unzipped, however when i explored the archive the dot and paths where there
14:23:13 Join ch4os [0] (n=ch4os@gentoo/user/ch4os)
14:24:17Cal_i have to manually boot in disc mode to load with files and see dir, is that normal for the ipod ver of rockbox, its not a big deal.
14:24:25 Quit chaos (Read error: 60 (Operation timed out))
14:24:31 Nick ch4os is now known as chaos (n=ch4os@gentoo/user/ch4os)
14:26:18Cal_id really have to say rockbox is the only reason to own a ipod, imao.
14:27:57 Part B4gder
14:28:59 Join Zarggg [0] (
14:29:04Cal_thankyou again bye
14:29:15 Quit Cal_ ("ChatZilla 0.9.85 [Firefox 3.0.15/2009101601]")
14:29:33*kugel wonders why JdGordon tries to fix usb screen instead of the viewportmanager
14:30:01kugelI think the sb handling there is buggy if it returns bogus values instead of 0/false everywhere
14:30:37 Join TheSeven|Mobile [0] (n=TheSeven@
14:31:16 Part TheSeven|Mobile
14:32:16 Join TheSeven|Mobile [0] (n=TheSeven@rockbox/developer/TheSeven)
14:36:39liarlinuxstb: ping
14:40:53CIA-6New commit by teru (r23601): usb screen: show logo at center of ui viewport and try to fit logo and title to the viewoprt to avoid possible redraw issue with sbs.
14:42:03 Join Bob_C [0] (
14:44:41 Join Topy44 [0] (
14:52:03 Quit Bob_C__ (Read error: 110 (Connection timed out))
14:55:34 Quit chaos ("Lost terminal")
15:05:44 Join phanboy4 [0] (
15:10:44kkurbjunkugel: I filed a FS bug that has a bunch of bugs recorded with the custom statusbar
15:12:50***Saving seen data "./dancer.seen"
15:13:58linuxstbliar: ?
15:15:53kugelkkurbjun: bug 4 doesn't seem related to sbs
15:16:20kugelthe list touchscreen handling reacts on outside areas too, for example to make tapping the statusbar to get to main menu possible
15:16:57kugelI think that's questionable, but the tapping the sb (classic one) is a nice short cut
15:18:18liarlinuxstb: i just installed the rockbox bootloader on a nano2g of a friend of mine and its unable to boot the apple os(readfw error -8)
15:20:22 Join MethoS- [0] (n=clemens@
15:25:12linuxstbliar: That would seem to indicate that there isn't any apple firmware there. Did you install with ipodpatcher? Was the ipod unmodified before you ran ipodpatcher?
15:26:25liarlinuxstb: its unmodified and i installed with ipodpatcher yep
15:27:38linuxstbliar: Then I've no idea... Did you cleanly eject after installing?
15:28:26linuxstbCan you run "ipodpatcher −−list -v" and pastebin the output?
15:32:58 Join Grahack [0] (
15:33:28 Quit teru ("Quit")
15:33:56 Quit phanboy4 ("Leaving")
15:36:15 Quit kugel (Read error: 110 (Connection timed out))
15:42:11liarlinuxstb: no i cant.. partition table is corrupted and i dont have itunes to restore it atm
15:44:20linuxstbThat would probably explain why the apple firmware can't be loaded...
15:45:19liaryeah, but rockbox worked before and now i have to restore the ipod...
16:03:40 Join phanboy4 [0] (
16:04:00 Quit Grahack ("Leaving.")
16:04:29 Join ch4os [0] (n=ch4os@gentoo/user/ch4os)
16:06:11 Quit goffa ("leaving")
16:06:39 Join goffa [0] (n=goffa@
16:07:36liarlinuxstb: i restored it. i think i only have an old version of the bootloader(which i got from you some weeks ago) and i dont know how to encrypt the one i build from svn
16:08:07CIA-6New commit by mc2739 (r23602): Correct target id for Samsung YH*
16:12:15linuxstbliar: There's no need to build from SVN - you can use the latest official release. But if you want to encrypt your own, you need to do that on an actual Nano2G - browse to the file in the file browser, open up the context menu (long press on the centre button), select "open with..." and then "crypt_firmware". That will create a ".ipodx" file
16:22:11liarahh the official ipodpatcher includes a bootloader :-) i build it myself...
16:23:02 Join DerPapst [0] (
16:24:46liarthat one is still not booting the OF, but nvm he doesnt need it
16:26:06 Quit phanboy4 (Read error: 104 (Connection reset by peer))
16:28:03 Join toffe82 [0] (n=chatzill@
16:29:27 Join phanboy4 [0] (
16:30:00 Join Horscht [0] (n=Horscht2@xbmc/user/horscht)
16:30:23 Join robin0800 [0] (
16:30:41 Quit robin0800 (Remote closed the connection)
16:31:00 Join robin0800 [0] (
16:36:28 Quit TheSeven|Mobile (Read error: 145 (Connection timed out))
16:45:17 Join evilnick_B [0] (i=0c140464@rockbox/staff/evilnick)
16:45:36 Join kugel [0] (n=kugel@rockbox/developer/kugel)
16:51:39 Quit liar (Client Quit)
16:56:59 Join kugel_ [0] (
16:57:19 Quit kugel_ (Read error: 104 (Connection reset by peer))
16:58:53 Nick ch4os is now known as chaos (n=ch4os@gentoo/user/ch4os)
16:59:58 Join kkurbjunW [0] (n=karlk@
17:01:39 Quit Zagor ("Don't panic")
17:08:25 Join Grahack [0] (
17:09:46 Quit antil33t1 (Read error: 104 (Connection reset by peer))
17:09:54 Join antil33t [0] (
17:12:52***Saving seen data "./dancer.seen"
17:20:16 Quit Tomis (Read error: 110 (Connection timed out))
17:20:32 Join BdN3504 [0] (
17:22:10 Join TheSeven|Mobile [0] (n=TheSeven@
17:34:41 Quit thegeek (Read error: 54 (Connection reset by peer))
17:34:53 Quit TheSeven|Mobile (Read error: 60 (Operation timed out))
17:35:36 Join Omlet [0] (i=omlet05@
17:45:15 Join jgarvey [0] (
17:46:57 Join kugel_ [0] (
17:50:40 Quit chaos (Read error: 110 (Connection timed out))
17:51:47JdGordonkugel (very quickly) the problem is the load/init order... viewportmanager isnt even initialised when early usb happens
17:51:55JdGordonso fixing it is going to do much good
17:53:37 Quit kugel (Read error: 60 (Operation timed out))
17:54:28 Join n1s [0] (n=n1s@rockbox/developer/n1s)
17:55:30 Join AEnima1577 [0] (
17:55:44 Quit AEnima1577 (Client Quit)
17:56:54 Join domonoky [0] (n=Domonoky@rockbox/developer/domonoky)
17:57:52kugel_JdGordon: it shouldn't even be a problem
17:58:38 Join Tomis [0] (n=Tomis@
17:59:30kugel_loaded_ok should be false from the beginning, and global_settings.statusbar should be != STATUSBAR_CUSTOM from the beginning (unless that's the new default which I didn't intend). That means it should be impossible to get down to skin_* functions even if the area is uninitialized
18:02:06shaiYou guys know why my HDD icon at the top right corner keeps showing as if my HDD is really working? I mean.. it could be true... but why is it always on?
18:02:56BdN3504hey kugel,
18:03:00 Join stripwax [0] (
18:03:12BdN3504how do i make a theme automatically load an sbs file?
18:03:38BdN3504is it sbs: sbsfile in the theme.cfg?
18:03:49kugel_plus "statusbar: custom"
18:04:11BdN3504ah ok
18:05:25BdN3504nice :) i think i'll unleash an old theme again.
18:05:27shaiIf I connect it to the USB, it stops but then continues after I remove it from the USB
18:05:32shaiAlso does this after a reboot...
18:05:36 Quit BdN3504 ("CGI:IRC")
18:05:38shaifor a prolonged period of time...
18:05:47 Join thegeek [0] (
18:06:04JdGordonkugel_: settings arnt initialised at all yet (iirc)... so the value of global_settings.statusbar is random
18:07:12kugel_JdGordon: settings_reset() is called very early. that revers everything to its default
18:07:20JdGordonnot before early usb
18:07:37JdGordonoh bloody hell
18:07:40JdGordonhiding up there
18:09:23kugel_and even then, it would still be a bug if the statusbar tries to draw without loaded_ok being true
18:10:12mc2739shai: what device do you have?
18:10:23shaiThere... it stopped after 10min
18:10:34*JdGordon is hell bent of removing the inbuilt bar before the weekend... that will fix some of this
18:10:34shaiiPod 30G
18:10:41shaimc2739, iPod 30G
18:11:02shaiI never noticed this behavior before...
18:11:13mc2739shai: do you have directory cache turned on?
18:11:22shaimc2739, I'll check
18:12:14shaiI don't recall where it is... so I'll just go over it all...
18:12:31shaiNo, its off
18:12:48 Quit kugel_ ("exit(0);")
18:12:49mc2739turning it on should help
18:13:07gevaertsshai: do you use the database, and is it set to scan for updates? If so, you really need the directory cache I think
18:13:11shaiIts after every reboot... it seems...
18:13:19 Quit petur ("*plop*")
18:13:25 Join liar [0] (n=liar@
18:13:41shaigevaerts, I do use it. Yes, its set to scan for updates...
18:13:46shaiI've now enabled dir. cache
18:14:03shaiI rebooted the iPod again .. and its using the HDD...
18:14:18shaiI'm assuming that this much usage will now get reduced after I've enabled dir. cache?
18:15:36Tornewhen dircache is on it will spend maybe up to a minute in the background populating the dircache
18:15:45Tornethen it will quickly go and check which bits might need a db update
18:15:45 Join kugel [0] (n=kugel@rockbox/developer/kugel)
18:16:02Tornethe db update bit should take a *lot* less time with the dircache on, unless you really have changed a lot of files since last time ;)
18:16:13Tornebut the dircache is not saved on disk, so it does take a *little* time to create it at boot.
18:16:29 Quit kkurbjunW (Remote closed the connection)
18:17:47Unhelpfulkugel: have you had any progress on eabi+fuze?
18:17:57 Join kkurbjunW [0] (n=karlk@
18:18:58 Join daggett [0] (
18:19:13kugelUnhelpful: not at all, haven't had any time
18:22:05shaigevaerts, after every reboot... it turns dir. cache back to Off state.
18:22:59gevaertsshai: with a clean reboot, i.e. not a hard reset?
18:23:33shaiHard reset? I click on the Select and Menu buttons and it does a reboot after like 2-3 seconds...
18:23:42gevaertsthat's a hard reset. Don't do that
18:23:47shaiOh ...
18:23:52shaiThen what is a softone?
18:24:00shaiTurning it Off and On again?
18:24:15Tornehold play till it turns off, then press select to turn back on
18:24:26shaiI just have
18:24:58shaiBut sometimes.. I have to have a hard-reset ...
18:25:16shaiI've seen it get stuck at times... which is the only way I was able to get it to reboot...
18:25:40shaiWhat should I do then? Is this the only setting I need to "worry" about after a hard-reset ?
18:25:43gevaertswell yes, if that happens (it shouldn't of course, but...) a hard reset is the proper solution
18:26:11TorneSettings don't change back because of hte hard reset
18:26:12gevaertsSetting changes are saved on shutdown, so any changes are lost with a hard reset
18:26:18Torneit's just that you lose changes..
18:26:21Torneoh there we go
18:26:42*shai does a hard-reset to check it
18:27:06shaiIts all good guys :)
18:27:33shaiSo Off saves my settings to when I need to use the Hard-Reset in case or "no other way"....
18:27:49shai10x :)
18:28:03shaiProblem solved.
18:29:12 Quit kkurbjun (
18:31:11NJoinkkurbjun [0] (n=kkurbjun@rockbox/developer/kkurbjun)
18:31:56Torneyes. we expect that you shut down at least occasiionaly :)
18:32:03Torneso settings saving is left until then
18:36:25 Quit Tomis (Connection timed out)
18:37:36 Quit maruk ("Leaving.")
18:39:33 Quit kkurbjun (
18:40:48n1sTorne: is it really saved only on shutdown, not on the next disk access?
18:45:35 Join uflops [0] (
18:45:42 Quit mikroflops (Read error: 104 (Connection reset by peer))
18:46:04 Join bertrik [0] (
18:46:42Tornen1s: hm, it might be saved at other points
18:47:05Tornebut in general you shouldn't count on it being saved unless you *did* shut down ;)
18:47:48 Join bmbl [0] (n=Miranda@unaffiliated/bmbl)
18:48:11 Join JdGordon| [0] (
18:53:05 Join kkurbjun [0] (
18:58:29 Part kkurbjunW
19:03:56 Quit phanboy4 (Read error: 60 (Operation timed out))
19:04:59LloreanDon't we usually refer to the iPod generations as "G2" or "G3" rather than "2G" and "3G" so as to avoid confusion with capacities (especially in regard to the nanos?)
19:05:26*domonoky thinks we use "2nd gen" :-)
19:06:04 Quit Zarggg ()
19:06:11LloreanWell, the page in the wiki is currently 2G
19:06:30n1si personally use XG for generations
19:06:35LloreanAnd someone in the forums said they had a Nano 2G making it somewhat unclear (since in the past 2G has been people referring to capacity)
19:07:23LloreanAh, no, I see we said "use 1g/1G/etc for generation" in the forum posting stuff
19:07:26LloreanNevermind me.
19:07:28UnhelpfulXgen perhaps?
19:08:07n1scapacity should be in GB, no confusion, yay :)
19:08:37 Quit uflops (Read error: 104 (Connection reset by peer))
19:09:04 Join mikroflops [0] (
19:09:11Lloreann1s: Except many people can't be bothered to type out the even single extra letter
19:10:39n1strue, but we should simply stop helping such users
19:10:51n1shmm, s/such // maybe ;)
19:12:56***Saving seen data "./dancer.seen"
19:13:01LloreanA three strikes rule for Rockbox support?
19:13:32 Quit Blue_Dude ("ChatZilla 0.9.85 [Firefox 3.5.5/20091102152451]")
19:22:04 Quit JdGordon| ("Miranda IM! Smaller, Faster, Easier.")
19:30:19 Join JdGordon| [0] (n=Miranda@nat/microsoft/x-vlwjblsptnykuedp)
19:32:46 Join freqmod_qu [0] (
19:35:12amiconnkugel: Thing is, the built-in status bar *should* be drawn in early usb
19:35:34amiconnOf course disk access is forbidden at that point, hence it has to be the built-in one
19:37:58 Join chaos [0] (n=chaos@gentoo/user/ch4os)
19:38:39 Join Zarggg [0] (
19:42:11 Join phanboy4 [0] (
19:44:23 Quit einhirn ("Miranda IM! Smaller, Faster, Easier.")
19:47:24kugelamiconn: of course
19:47:48kugelamiconn: the crash was IIUC because something caused the skin engine to step in
19:49:01*amiconn doesn't seem to understand
19:49:20amiconnIsn't the skin engine responsible for drawing the statusbar now?
19:49:31kugelnot the built-in one
19:50:00kugelclassic and custom sb are to different beasts, but the api is the same
19:50:26amiconnHmm, why is that?
19:50:59kugelbecause its another huge job to fully replace the the classic sb
19:52:03JdGordon|not a huge job
19:52:16 Quit robin0800 (Read error: 60 (Operation timed out))
19:52:34kugelsignificant enough to be a separate task imo
19:53:46 Part LinusN
19:54:32 Quit MethoS- (Remote closed the connection)
19:58:47JdGordon|seperate to what?
20:00:02*kugel got the sim to repro the problem
20:02:49kugelpixelma: ping
20:05:34kugelcan you try if fixes early usb (apply to r23598 please)?
20:07:12kugeli.e. before JdGordon's commits
20:08:42pixelmawill test but it'll take a bit (and I also need something for dinner)
20:13:25kugelno problem. maybe amiconn could test too?
20:14:31 Join uflops__ [0] (
20:15:51 Quit mikroflops (Read error: 110 (Connection timed out))
20:16:33 Join hillshum [0] (
20:17:36 Join domonoky1 [0] (
20:22:54 Quit stripwax ("")
20:23:10 Join mikroflops [0] (
20:24:18 Quit domonoky (Read error: 145 (Connection timed out))
20:26:12 Join Thundercloud [0] (
20:35:42pixelmakugel: yes, that fixes early USB on my Ondio
20:38:26pixelmahaven't tried JdGordon|'s fixes, except the very first one though and can't comment on which way is better
20:40:50 Quit flydutch ("/* empty */")
20:46:34 Quit bmbl ("Bye!")
20:54:15 Quit uflops__ (Read error: 110 (Connection timed out))
20:54:27 Quit phanboy4 ("Leaving")
21:10:47 Quit mikroflops (Read error: 104 (Connection reset by peer))
21:12:58***Saving seen data "./dancer.seen"
21:13:26CIA-6New commit by kugel (r23603): Revert r23599 and r23600 and fix the early usb problem at the root. The the viewportmanager and custom statusbar was initialized too late. ...
21:15:02 Join mikroflops [0] (
21:33:23 Join Strife89 [0] (
21:39:28 Quit Thundercloud (Read error: 54 (Connection reset by peer))
21:42:20 Join panni_ [0] (
21:44:53 Join pamaury [0] (
21:49:34 Join AndyI [0] (n=pasha_in@
21:56:02 Join funman [0] (n=fun@rockbox/developer/funman)
22:03:23 Join Stephen__ [0] (
22:09:14 Join stoffel [0] (
22:11:12 Join polobricolo [0] (
22:16:31 Quit HellDragon (Read error: 131 (Connection reset by peer))
22:16:50 Join HellDragon [0] (
22:23:15 Join Thundercloud [0] (
22:33:34 Join mt [0] (n=mtee@
22:34:58 Join hebz0rl [0] (
22:35:42 Quit hillshum (Remote closed the connection)
22:37:46 Quit kugel (Read error: 54 (Connection reset by peer))
22:39:12 Quit mikroflops (Read error: 110 (Connection timed out))
22:40:15 Join rjg [0] (
22:40:15 Quit pamaury ("exit(*(int *)0 / 0);")
22:41:32 Quit stoffel (Remote closed the connection)
22:43:36 Quit funman ("free(random());")
22:44:07rjgevening, just thought I'd double check this isn't likely to be a theme-specific issue before declaring it as a bug: - if it's not clear, what happens is that if a song is playing and the screen turns off and you leave it a while, when you hit a button to turn it on again, a buffer of the theme's updates proceeds to display as fast as it can
22:46:46rjg(it also doesn't update the currently playing song unless you come out of the now playing screen and back in, or let the screen turn off and you put it back on again)
22:46:54rjgit's happening in the past couple of builds
22:47:04rjgtarget is iriver H300
22:50:11 Join phanboy4 [0] (
22:55:58 Quit jgarvey ("Leaving")
22:57:35 Quit Bagder ("*plopp*")
23:01:03 Join GeekShadow [0] (n=Antoine@reactos/tester/GeekShadow)
23:13:02***Saving seen data "./dancer.seen"
23:15:01 Join petur [0] (
23:16:17 Join mikroflops [0] (
23:18:15 Quit n1s ("Lmnar")
23:22:05 Quit petur ("Zzzz *burp*")
23:23:55 Join BBBradley [0] (
23:25:32BBBradleyWhat is the best way to leave the WPS and stop audio? I've used "audio_stop()" but wondered is there another way?
23:26:25 Nick Ypsy is now known as YPSY (
23:28:08AaronMlook in the source of the stop button 'holding play for a little while'
23:28:34JdGordon|BBBradley: if you want to stop audio then that is fine
23:29:29BBBradleyI want to stop the audio and exit the wps at the same time
23:29:51Tornestopping audio always exits the wps
23:30:05BBBradleyTorne thanks
23:30:20Torneit's not the while stopped screen ;)
23:30:36gevaertsTorne: that one is called the "menu" I think
23:30:41BBBradleyI'm still fighting an issue where I exit the wps using "audio_stop()" and my menu won't spin until I press the menu key. It's annoying.
23:31:13evilnick_B(for the logs) bluebrother: The changes to crossfade are a big improvement, and I haven't seen/heard any issues with them. Tested on a gigabeat F.
23:33:49JdGordon|wrong blue* :)
23:35:39evilnick_BOops! Blue_Dude: See above ^^
23:35:41 Join killan_ [0] (
23:35:41 Quit GeekShadow (Read error: 131 (Connection reset by peer))
23:35:41 Quit killan (Read error: 131 (Connection reset by peer))
23:36:04 Join GeekShadow [0] (n=Antoine@reactos/tester/GeekShadow)
23:39:37BBBradleyAaronM what file is that in?
23:40:17AaronMidk, i dont code
23:40:37AaronMi just know the stop button in rockbox is preety much what your looking for
23:43:34 Quit bertrik ("De groeten")
23:44:21 Quit hebz0rl ()
23:48:37 Quit dfkt (Read error: 110 (Connection timed out))
23:56:16 Join panni__ [0] (
23:56:57 Quit panni_ (Read error: 113 (No route to host))

Previous day | Next day