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

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

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

#rockbox log for 2010-02-01

00:00:08 Join tarbo [0] (~me@unaffiliated/tarbo)
00:00:10JdGordonno, someone elses
00:00:47kugelso the buffer is full with 2 backdrops now?
00:02:45kugelah, the 2 fullscreen mono bitmaps are still there
00:05:10JdGordonI tihnk I need to make the skin buffer size selectable before fms goes in so it doesnt get crazy
00:05:15JdGordon(limited in both directions)
00:05:30JdGordonits either too much reserved, or not enough
00:06:03 Quit tarbo (Ping timeout: 265 seconds)
00:06:05 Nick S_a_i_n_t_ is now known as S_a_i_n_t (S_a_i_n_t@
00:06:23kugelthe increase it got should be sufficient for the fms I imagine
00:06:26LambdaCalculus37Can someone explain to me how to exclude specific plugins from being built on the SA9200? It's only the ones that have no bitmaps for the 128x160 screen.
00:07:13kugelLambdaCalculus37: #ifdef in SOURCES
00:07:40JdGordonkugel: not really, the buffer only increases enough for a backdrop, which is always loaded unless %Xd is there. so you cant have a bmp intensive WPS and sbs and fms all together
00:09:22kugela theme which worked before now needed a buffer increase to still work?
00:09:53JdGordonno, I'm not sure it ever did
00:10:05JdGordonbut I thought that was a silly size anyway
00:10:40kugelI'm thinking something went wrong if it worked before
00:11:00kugelI used that Klean theme once on my e200, although it wasn't v4
00:11:17LambdaCalculus37kugel: I'm a wee bit confused on how to add in plugins I want to exclude, though.
00:14:28kugelui viewport is completely gone now?
00:14:50JdGordonnot if you dont specify a sbs
00:15:23kugelah I found where it's parsed
00:15:35S_a_i_n_tso you can still use ui veiwport if no .sbs is loadd?
00:16:21JdGordonand it is used internally if the inbuilt bar is enabled
00:16:21S_a_i_n_taha...that's where i'm messing up
00:16:46 Join TopyMobile_ [0] (
00:18:50kugelso when is that inbuilt statusbar finally replaced with the sbs version?
00:18:53 Quit stoffel (Remote host closed the connection)
00:19:00 Quit jgarvey (Quit: Leaving)
00:19:36JdGordonprobbaly never :p
00:19:50JdGordonit's no longer getting in the way so much
00:20:09 Join tarbo [0] (~me@unaffiliated/tarbo)
00:21:09 Quit TopyMobile__ (Ping timeout: 272 seconds)
00:21:18pamauryHave there been recent changes to french translation ? The playing screen displays some stranges strings on my device
00:22:18kugelno, but there's a possible memory corruption in buffering.c which may result in garbage strings
00:23:28pamauryNot garbage strings but for example when in a playlist, it display say "11 suivant: 14" but it doesn't make sense to use "suivant:" for say 11 "out of" 14
00:23:39pamaury*to say
00:23:48 Join n17ikh [0] (
00:24:32pamauryHum, the same happens in english: "11 next: 14"
00:24:47pamauryIs it the normal text ?
00:25:06pamauryAlso it display "automatic track change only" above the track title
00:25:09kugeldid you compile yourself?
00:25:19kugeltry a current build
00:25:36pamauryThat's the first time it happens to me
00:26:04pamauryWait, I'll try to recompile without -j 3
00:26:18kugeldid you do make bin?
00:26:21 Quit tarbo (Ping timeout: 265 seconds)
00:26:25JdGordonyour language is out of date
00:26:57pamauryAh yes you're right, for strange reasons it didn't copy the language files o
00:28:02*kugel recommends make install :)
00:28:08 Quit JdGordon (Quit: Leaving.)
00:28:49pamauryI haven't done a make install for days ! ;)
00:30:36 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
00:30:55 Quit JdGordon (Remote host closed the connection)
00:32:50 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
00:34:41 Quit Strife89|VM (Quit: Shutting down the VM.)
00:35:36 Quit perfectdrug (Quit: perfectdrug)
00:40:07pamauryDoesn't make install copy language files ?
00:40:10 Join tarbo [0] (~me@unaffiliated/tarbo)
00:43:13JdGordonmake install only works with the sim
00:43:16JdGordonyou need make zip
00:44:24pamaurybut I don't want to make a zip !
00:44:59pamauryanyway I managed to get a clean build with make install.
00:45:46pamauryHum, I puzzled. When using dircache with logf enabled I get a strange error message which happens from time to time
00:46:33kugelJdGordon: wrong
00:46:58kugelpamaury: make install should work (it does make zip + extract behind the scenes)
00:47:07 Quit tarbo (Ping timeout: 272 seconds)
00:47:16CIA-8New commit by funman (r24428): Fuzev1: estimate current use with battery_bench results
00:47:32 Quit flydutch (Quit: /* empty */)
00:48:11JdGordonok then
00:48:31CIA-8New commit by funman (r24429): Fuzev1: estimate current use with battery_bench results
00:50:16kugelfunman: shouldn't CURRENT_NORMAL and CURRENT_BACKLIGHT be the other way around?
00:50:53kugelor am I confused by the naming?
00:52:36 Quit efyx_ (Read error: Connection reset by peer)
00:53:54 Quit S_a_i_n_t (Ping timeout: 265 seconds)
00:55:09 Join TopyMobile__ [0] (
00:55:45CIA-8New commit by funman (r24430): Modify version strings for 3.5
00:55:49 Quit ender` (Quit: If I had only finished this sentence,)
00:58:34 Join perfectdrug_ [0] (
00:59:01 Quit TopyMobile_ (Ping timeout: 256 seconds)
00:59:56 Join froggyman [0] (
01:01:59 Quit pamaury (Quit: abort();)
01:02:42 Join martian67 [0] (
01:02:42 Quit martian67 (Changing host)
01:02:42 Join martian67 [0] (iTIPeHu2HT@about/linux/regular/martian67)
01:03:23 Join S_a_i_n_t [0] (S_a_i_n_t@
01:05:42 Join S_a_i_n_t_ [0] (S_a_i_n_t@
01:06:07 Part S_a_i_n_t_
01:06:13 Join S_a_i_n_t_ [0] (S_a_i_n_t@
01:07:31 Quit Tomis (Read error: Connection reset by peer)
01:08:24 Join Tomis [0] (~Tomis@
01:09:05 Quit S_a_i_n_t (Ping timeout: 248 seconds)
01:09:54 Nick S_a_i_n_t_ is now known as S_a_i_n_t (S_a_i_n_t@
01:12:20 Quit TopyMobile__ (Quit: TopyMobile__)
01:13:53 Quit Tomis (Ping timeout: 248 seconds)
01:14:55 Quit S_a_i_n_t (Ping timeout: 246 seconds)
01:18:29 Quit DerPapst (Quit: Leaving.)
01:22:08 Join tarbo [0] (
01:22:08 Quit tarbo (Changing host)
01:22:08 Join tarbo [0] (~me@unaffiliated/tarbo)
01:23:01 Quit andrewRB (Read error: Connection reset by peer)
01:28:00 Quit tarbo (Ping timeout: 264 seconds)
01:30:44piotrekmis it possible to use the eabi gcc compiler when compiling the nano2g rockbox, and if so will it improve anything?
01:33:41***Saving seen data "./dancer.seen"
01:34:53saratogaits mostly possible (search the logs from this month)
01:34:58saratogait saves some memory
01:37:33brybotHello, I have a 5th generation iPod that I put Rockbox on over summer. I have not used my iPod in months. Anyway, I remember then, and have the same problem now, that when Rockbox loads, my iPod vibrates. It sounds like the seeker head on the hard drive is moving about rapidly. This however, does not happen when I am using the apple software. Any idea what could be causing this?
01:40:04 Join andrewRB [0] (
01:42:09 Join tarbo [0] (
01:42:09 Quit tarbo (Changing host)
01:42:09 Join tarbo [0] (~me@unaffiliated/tarbo)
01:47:28 Join Llorean1 [0] (
01:47:29 Quit tarbo (Ping timeout: 248 seconds)
01:47:41 Quit Llorean (Killed (NickServ (GHOST command used by Llorean1!
01:47:43 Nick Llorean1 is now known as Llorean (
01:47:46 Quit Llorean (Changing host)
01:47:46 Join Llorean [0] (~DarkkOne@rockbox/user/Llorean)
01:47:56liarTheSeven: ping
01:48:23 Quit grndslm (Read error: Operation timed out)
01:51:04 Quit lyngaas (Read error: Connection reset by peer)
01:53:44 Quit Casainho (Quit: ChatZilla 0.9.86 [Firefox 3.5.7/20100106054534])
01:54:29 Join S_a_i_n_t [0] (S_a_i_n_t@
01:55:49 Quit kugel (Ping timeout: 258 seconds)
01:59:11 Join grndslm [0] (
02:01:04 Quit perfectdrug_ (Quit: perfectdrug_)
02:01:16S_a_i_n_tIf anyone's planning on compiling for the nano2g from current svn...don't bother
02:01:20S_a_i_n_tshe's broken.
02:01:22 Join lyngaas [0] (
02:01:59liarand this: is the fix for it
02:02:13 Join tarbo [0] (~me@unaffiliated/tarbo)
02:02:17S_a_i_n_tWe've been at it all night...
02:02:23S_a_i_n_tand not in *THAT* way :P
02:02:34saratogabuildserver says the 2g built?
02:02:47liarsaratoga: it fails at runtime
02:02:52saratogaah ok
02:02:59liarsaratoga: for nanos with less than 4 banks
02:04:01liarbecause of one variable is initialized too late
02:06:24 Quit Sajber^ (Ping timeout: 265 seconds)
02:09:27 Quit tarbo (Ping timeout: 272 seconds)
02:10:26 Join RusselKubes [0] (
02:12:49RusselKubesHello everyone. I just recently registered, I was wondering if I could get write privileges for the wiki? There was a page about iPod Hard Drives that I want to add info for for my model.
02:14:03TheSevenliar: pong
02:14:25mc2739RusselKubes: <- id that your wiki user name?
02:14:30S_a_i_n_tTheSeven: Hello Stranger... :D
02:15:42mc2739RusselKubes: done - please do not spam
02:15:43liarTheSeven: to keep it short: nand_type[x] isnt initialized when nand_power_up is called the first time.. thats why nand_power_up calls nand_reset for unexisting banks
02:16:11RusselKubesok, thank you very much :)
02:16:24TheSevenyep, I got it
02:17:05*TheSeven wonders why we haven't caught that before - obviously there aren't many nanos with less then 4 banks...
02:21:15 Join JdGordon| [0] (
02:21:15 Quit JdGordon| (Changing host)
02:21:15 Join JdGordon| [0] (~Miranda@rockbox/developer/JdGordon)
02:22:09 Join tarbo [0] (~me@unaffiliated/tarbo)
02:23:46CIA-8New commit by theseven (r24431): Fix iPod Nano 2G bank detection broken in r24414.
02:25:44 Quit RusselKubes (Quit: CGI:IRC)
02:28:09 Quit tarbo (Ping timeout: 265 seconds)
02:31:24 Part froggyman
02:33:44 Join Hawson [0] (
02:36:07 Part brybot
02:36:48CIA-8New commit by unhelpful (r24432): FS #10943, optimized division and clz routines to replace libgcc routines for ARM. Replaces libgcc support functions for unsigned and signed 32-bit ...
02:39:44 Nick fxb is now known as fxb__ (
02:42:43 Join tarbo [0] (
02:42:43 Quit tarbo (Changing host)
02:42:43 Join tarbo [0] (~me@unaffiliated/tarbo)
02:48:24 Quit tarbo (Ping timeout: 264 seconds)
02:48:43 Quit toffe82 (Read error: Connection reset by peer)
02:53:29 Join FlynDice [0] (
02:53:44 Quit AndyI (Read error: Connection reset by peer)
02:54:03 Join AndyI [0] (~pasha_int@
02:54:59 Join AndyIL [0] (~pasha_int@
02:55:00 Quit AndyI (Read error: Connection reset by peer)
02:56:27 Join Tomis [0] (~Tomis@
03:02:45 Join tarbo [0] (~me@unaffiliated/tarbo)
03:06:34 Join Strife1989 [0] (
03:06:46 Quit chaos (Ping timeout: 276 seconds)
03:08:05 Join chaos [0] (~chaos@gentoo/user/ch4os)
03:08:59 Quit Strife89 (Ping timeout: 272 seconds)
03:10:16 Quit tarbo (Ping timeout: 240 seconds)
03:16:46 Join stooo1 [0] (
03:19:45 Quit stooo (Ping timeout: 272 seconds)
03:22:44 Join tarbo [0] (
03:22:44 Quit tarbo (Changing host)
03:22:44 Join tarbo [0] (~me@unaffiliated/tarbo)
03:22:51 Join rhodan [0] (
03:26:23 Quit liar (Quit: Verlassend)
03:28:34 Quit tarbo (Ping timeout: 265 seconds)
03:30:08 Quit MethoS- (Read error: Connection reset by peer)
03:30:56piotrekmUnhelpful: in firmware/target/arm/support-arm.S:665 there is a ":" missing i suppose
03:31:24 Join wind [0] (
03:31:25piotrekmi mean it should be "__aeabi_idivmod:"
03:33:42***Saving seen data "./dancer.seen"
03:33:42piotrekmbtw, i get this when compiling with −−eabi, any help? : rockbox/build/firmware/libfirmware.a(pcm-s5l8700.o): In function `fiq_handler': pcm-s5l8700.c:(.icode+0x58): undefined reference to `dma_callback'
03:34:41Unhelpfulpiotrekm: i don't know what that one's about, but thanks for the heads-up on the missing colon.
03:35:30CIA-8New commit by unhelpful (r24433): Remove heaps of trailing whitespace from new file.
03:35:36CIA-8New commit by unhelpful (r24434): Missing colon in support-arm.S for EABI.
03:36:02Unhelpfulurgh, i didn't realize i hadn't pushed that first change already :/
03:36:15S_a_i_n_thow do i delete the bootloader from a nano2g using command-line ipodpatcher?
03:36:55piotrekmS_a_i_n_t: it should ask you what to do
03:37:07piotrekmjust run it without any commands
03:39:00 Join BlastTyrant [0] (
03:39:19S_a_i_n_tthanks...i thought so, but thought it may be different with the osbk
03:39:52S_a_i_n_twanted to ask irst, and not be sorry later.
03:42:41 Quit rhodan (Remote host closed the connection)
03:42:44 Join tarbo [0] (~me@unaffiliated/tarbo)
03:48:23 Quit tarbo (Ping timeout: 265 seconds)
03:51:44 Quit LambdaCalculus37 (Ping timeout: 248 seconds)
03:52:16 Quit ps-auxw (Quit: leaving)
03:53:04 Join ps-auxw [0] (~arneb@2001:470:c807:0:1532:4e5f:2ad3:4123)
03:53:26 Nick Strife1989 is now known as Strife89 (
03:53:42 Quit ps-auxw (Client Quit)
03:54:11 Join Rob2223 [0] (
03:55:28 Join ps-auxw [0] (~arneb@2001:470:c807:0:1532:4e5f:2ad3:4123)
03:56:40 Join LambdaCalculus37 [0] (
03:56:40 Quit LambdaCalculus37 (Changing host)
03:56:40 Join LambdaCalculus37 [0] (~rmenes@rockbox/staff/LambdaCalculus37)
03:57:45 Quit Rob2222 (Ping timeout: 272 seconds)
04:02:43 Join bangfoo [0] (
04:02:47 Join tarbo [0] (
04:02:48 Quit tarbo (Changing host)
04:02:48 Join tarbo [0] (~me@unaffiliated/tarbo)
04:03:45 Quit bangfoo (Client Quit)
04:09:25 Quit tarbo (Ping timeout: 256 seconds)
04:11:21 Join RusselKubes [0] (
04:13:35 Quit piotrekm (Ping timeout: 272 seconds)
04:19:51 Quit thegeek (Read error: Connection reset by peer)
04:22:47 Join tarbo [0] (~me@unaffiliated/tarbo)
04:25:57 Join thegeek [0] (
04:28:28 Quit tarbo (Ping timeout: 246 seconds)
04:28:36 Quit panni_ (Quit: ( :: NoNameScript 3.81 :: ))
04:32:59 Join Blast_Tyrant [0] (
04:33:05 Quit TheSeven (Killed (NickServ (GHOST command used by The_Seven)))
04:33:16 Join The_Seven [0] (~theseven@rockbox/developer/TheSeven)
04:33:27 Nick The_Seven is now known as TheSeven (~theseven@rockbox/developer/TheSeven)
04:36:25 Quit BlastTyrant (Ping timeout: 245 seconds)
04:42:47 Join tarbo [0] (~me@unaffiliated/tarbo)
04:42:54 Quit Blast_Tyrant (Quit: Leaving)
04:44:00 Join BlastTyrant [0] (
04:47:15BlastTyrantHello everyone. After updating from r24254 to r24404 on my Fuze I noticed a problem with the BlackSunrise theme that I made (I get the same results in the 1/31 compiled simulator). I just wanted to find out if this problem is a bug or if I'm missing something obvious:
04:47:43BlastTyrantIt seems that the backdrop specified in the theme cfg shows through the WPS backdrop in non-static areas of the screen. I put up an example of this on Imageshack:
04:48:17JdGordon|gimme 5 min and I'll help
04:48:25 Quit tarbo (Ping timeout: 246 seconds)
04:48:48BlastTyrantsure thing, thanks
04:53:54 Quit LambdaCalculus37 (Quit: Fwump)
04:54:03 Join Barahir [0] (
04:54:30JdGordonBlastTyrant: if the theme online somewhere?
04:55:44BlastTyrantyes, it's up on the Fuze theme page:
04:57:05 Quit Barahir_ (Ping timeout: 240 seconds)
04:57:23JdGordonfuze is the same scren as h300 yeah?
04:58:47 Part andrewRB ("Leaving")
04:58:53JdGordonwhich of your screenshots is what it is supposed to look like?
04:58:59 Join andrewRB [0] (
05:00:27BlastTyrantthis one is what it looked like originally:
05:00:35BlastTyrantand this is what it's doing now:
05:01:18BlastTyrant*it's supossed to look like the first one
05:01:53JdGordonah, the progressbar isnt shown
05:01:59JdGordonis that part of the wps backdrop?
05:02:19JdGordonhmm, ok
05:02:43BlastTyrantand the next artist text usually scrolls and has the same problem with the bottom bar being transparent where it scrolls
05:02:47 Join tarbo [0] (
05:02:48 Quit tarbo (Changing host)
05:02:48 Join tarbo [0] (~me@unaffiliated/tarbo)
05:03:29 Quit chaos (Ping timeout: 240 seconds)
05:04:04JdGordon%we shouldnt go in the .sbs
05:04:10JdGordonnot that it makes any difference
05:04:43JdGordonand comments should match the code :p
05:04:44JdGordon# Disable the Status Bar
05:04:55 Join chaos [0] (~chaos@gentoo/user/ch4os)
05:05:05BlastTyrantyes, I realized that about 40 minutes ago and changed it
05:05:41JdGordonah, so you want it on or off?
05:06:25JdGordonon by the looks of it?
05:06:47BlastTyrantyeah, I thought that I needed it on...
05:08:21 Quit tarbo (Ping timeout: 256 seconds)
05:09:46JdGordonBlastTyrant: try putting the %pb in a viewport
05:10:51FlynDiceJdGordon: Did you ever do a postmortem examination of your clip+ bricking? ie do you have an idea what went wrong there?
05:11:44JdGordonI didnt, but it was assumed that the OF needs some registers which our bootloader mangles
05:12:36FlynDiceYou don't still have the patch you used by any chance?
05:14:34JdGordonI dont, but I tinhk I just added some #if's for the clip+
05:15:32FlynDiceOk thanks, figured I'd at least ask before I try next......
05:16:26JdGordontalk to funman before fiddling
05:18:16JdGordonBlastTyrant: I think I know whats happening. the default viewport isnt cleared when entering the WPS, because if it was the sbs would be overwritten, which also means the wps backdrop isnt being drawn into that area
05:18:26JdGordonso putting it into a new viewport should fix it
05:20:19 Join Blast_Tyrant [0] (
05:22:48 Join tarbo [0] (
05:22:48 Quit tarbo (Changing host)
05:22:48 Join tarbo [0] (~me@unaffiliated/tarbo)
05:23:48 Quit BlastTyrant (Ping timeout: 260 seconds)
05:24:27JdGordonBlast_Tyrant: did you see my replies?
05:26:24Blast_Tyrantyes, YChat had some connection problem and killed my connection, but it should be fine now - I'm trying the new viewport now, but it'll take a few mins
05:29:12 Quit tarbo (Ping timeout: 264 seconds)
05:30:24JdGordonhmm, no, it still clears too much of the area
05:33:46***Saving seen data "./dancer.seen"
05:33:52JdGordoncrap, I tihnk I know whats happening (really this time)
05:34:54JdGordonyeah, this is a real bug
05:34:57Blast_TyrantThe new viewport doesn't seem to fix it, but I'll fiddle with it a litte more
05:35:12JdGordonno, its my fault, you shouldnt need to do anything
05:36:11Blast_Tyrantis this be from r24336 then? the display->clear_display() line?
05:36:21Blast_Tyranti was looking at that earlier
05:37:54JdGordonno, that was all changed again
05:38:37JdGordonwell no. Becuase of orering that is the actual issue
05:38:46JdGordonbut removing that probably wont fix anything
05:40:30 Quit fdinel (Read error: Connection reset by peer)
05:40:36 Quit Strife89 (Quit: Bed.)
05:40:53Unhelpfulinteresting, where are we doing signed division in rockbox before we even hit the splash screen?
05:41:23JdGordonsetting up viewports maybe?
05:42:13UnhelpfulJdGordon: and the other question, i have a rather large test set with operands of all sizes, so how do i have an error i've not caught? :/
05:42:48 Join tarbo [0] (
05:42:48 Quit tarbo (Changing host)
05:42:48 Join tarbo [0] (~me@unaffiliated/tarbo)
05:44:06 Quit JdGordon| (Quit: Miranda IM! Smaller, Faster, Easier.
05:44:42 Join Strife89|Desktop [0] (
05:44:44*Unhelpful had better disable the signed divider on armv5+ until he finds the problem
05:47:21CIA-8New commit by jdgordon (r24435): make sure skins always draw with their backdrop, otherwise the backdrop only changes on a full redraw which usually ends up being in the wrong order
05:47:32JdGordonBlast_Tyrant: that commit fixes your problem
05:47:53 Quit tarbo (Ping timeout: 240 seconds)
05:50:26CIA-8New commit by unhelpful (r24436): Some sort of issue in the signed divider is causing Gigabeat S to abort on startup, disable this routine until it's fixed.
05:54:02JdGordonHAHA, I just noticed the forum description for the "User interface" sub forum. nothing but 3 NODO's and heated debate topics :p
05:59:36 Join ball [0] (
06:02:46 Join tarbo [0] (
06:02:46 Quit tarbo (Changing host)
06:02:46 Join tarbo [0] (~me@unaffiliated/tarbo)
06:08:51 Quit tarbo (Ping timeout: 272 seconds)
06:22:45 Join tarbo [0] (
06:22:45 Quit tarbo (Changing host)
06:22:45 Join tarbo [0] (~me@unaffiliated/tarbo)
06:26:17 Quit CaptainKewl (Quit: ( :: NoNameScript 4.22 :: ))
06:28:44 Quit tarbo (Ping timeout: 252 seconds)
06:41:11 Join BHSPitMonkey [0] (~stephen@unaffiliated/bhspitmonkey)
06:42:52 Join tarbo [0] (~me@unaffiliated/tarbo)
06:45:52 Join S_a_i_n_t_ [0] (S_a_i_n_t@
06:46:49 Quit S_a_i_n_t (Read error: No route to host)
06:46:50 Quit S_a_i_n_t_ (Read error: Connection reset by peer)
06:47:53 Join S_a_i_n_t [0] (S_a_i_n_t@
06:50:12 Quit tarbo (Ping timeout: 264 seconds)
06:57:05 Quit S_a_i_n_t (Ping timeout: 240 seconds)
07:04:00 Join n1s [0] (~n1s@
07:04:00 Quit n1s (Changing host)
07:04:00 Join n1s [0] (~n1s@rockbox/developer/n1s)
07:05:53 Join tarbo [0] (
07:05:53 Quit tarbo (Changing host)
07:05:53 Join tarbo [0] (~me@unaffiliated/tarbo)
07:12:17 Quit tarbo (Ping timeout: 240 seconds)
07:15:04 Join shai [0] (
07:15:20 Quit saratoga (Quit: Page closed)
07:30:34 Join JdGordon| [0] (
07:30:34 Quit JdGordon| (Changing host)
07:30:34 Join JdGordon| [0] (~Miranda@rockbox/developer/JdGordon)
07:33:47***Saving seen data "./dancer.seen"
07:33:51 Quit n1s (Quit: Lämnar)
07:46:37 Quit gevaerts (Killed (NickServ (GHOST command used by gevaerts_!
07:46:46 Join gevaerts_ [0] (~fg@rockbox/developer/gevaerts)
07:50:44 Join tarbo [0] (~me@unaffiliated/tarbo)
07:51:59 Quit Zarggg (Read error: Connection reset by peer)
07:52:06 Join Zarggg_ [0] (
07:57:28 Join Horscht [0] (
07:57:28 Quit Horscht (Changing host)
07:57:28 Join Horscht [0] (~Horscht2@xbmc/user/horscht)
07:59:18 Quit kadoban (Read error: Connection reset by peer)
08:00:24 Quit tarbo (Ping timeout: 264 seconds)
08:00:59 Join kadoban [0] (
08:07:07 Quit wind (Quit: CGI:IRC (EOF))
08:11:30 Join tarbo [0] (
08:11:30 Quit tarbo (Changing host)
08:11:30 Join tarbo [0] (~me@unaffiliated/tarbo)
08:18:11 Quit tarbo (Ping timeout: 252 seconds)
08:23:38 Quit RusselKubes (Quit: Ex-Chat)
08:23:48 Quit Horscht (Quit: Verlassend)
08:28:55 Quit ball (Quit: Lost terminal)
08:31:09 Join Zagor [0] (
08:31:09 Quit Zagor (Changing host)
08:31:09 Join Zagor [0] (~bjst@rockbox/developer/Zagor)
08:31:29 Join tarbo [0] (
08:31:29 Quit tarbo (Changing host)
08:31:29 Join tarbo [0] (~me@unaffiliated/tarbo)
08:31:53 Quit TheSeven (Ping timeout: 240 seconds)
08:34:48andrewRBI assume this isn't meant to be spat out by PHP at the top of the themes site? "Warning: zip_read() expects parameter 1 to be resource, integer given in /home/themes/private/themesite.class.php on line 156"
08:37:54 Join ender` [0] (
08:38:09 Quit tarbo (Ping timeout: 248 seconds)
08:39:57 Quit arbingordon (Read error: Connection reset by peer)
08:42:07 Quit thegeek (Read error: Connection reset by peer)
08:42:10 Join flydutch [0] (
08:44:36 Nick stooo1 is now known as stooo (
08:44:51 Join PaulJam [0] (
08:47:13 Quit AndyIL (Ping timeout: 248 seconds)
08:51:30 Join tarbo [0] (
08:51:30 Quit tarbo (Changing host)
08:51:30 Join tarbo [0] (~me@unaffiliated/tarbo)
08:53:06 Join bmbl [0] (~Miranda@unaffiliated/bmbl)
08:54:04 Join AndyI [0] (~pasha_int@
08:58:20 Quit tarbo (Ping timeout: 252 seconds)
09:01:02 Join petur [0] (~petur@rockbox/developer/petur)
09:02:01 Quit andrewRB (Quit: Leaving)
09:03:57 Nick fxb__ is now known as fxb (
09:09:40 Join RusselKubes [0] (
09:17:47 Quit elcan (Read error: Connection reset by peer)
09:19:30CIA-8New commit by pixelma (r24437): Update German translation including minor fixes - be more accurate about 'shuffle' (not 'random') and fix some occurences of wrong spaces.
09:20:47 Join Bagder [0] (
09:20:47 Quit Bagder (Changing host)
09:20:47 Join Bagder [0] (~daniel@rockbox/developer/bagder)
09:23:41CIA-8New commit by pixelma (r24438): Update German translation including minor fixes - be more accurate about 'shuffle' (not 'random') and fix some occurences of wrong spaces.
09:23:53 Join elcan [0] (
09:24:03*pixelma nags about the branch info once more but has to leave
09:24:04 Quit JdGordon| (Quit: Miranda IM! Smaller, Faster, Easier.
09:24:49 Quit RusselKubes (Quit: Ex-Chat)
09:33:50***Saving seen data "./dancer.seen"
09:40:23 Quit phanboy4 (Ping timeout: 272 seconds)
10:14:37 Nick fxb is now known as fxb__ (
10:16:14 Nick gevaerts_ is now known as gevaerts (~fg@rockbox/developer/gevaerts)
10:19:11 Quit elcan (Ping timeout: 252 seconds)
10:22:48 Join elcan [0] (
10:36:26 Join kugel [0] (kugel@
10:36:26 Quit kugel (Changing host)
10:36:26 Join kugel [0] (kugel@rockbox/developer/kugel)
10:37:18 Quit kugel (Read error: Connection reset by peer)
10:37:28 Join DerPapst [0] (
10:37:34 Join kugel [0] (kugel@
10:37:34 Quit kugel (Changing host)
10:37:34 Join kugel [0] (kugel@rockbox/developer/kugel)
10:42:36 Quit kugel (Ping timeout: 265 seconds)
10:43:49 Quit BHSPitMonkey (Quit: Ex-Chat)
10:47:31 Quit shai (Read error: Connection reset by peer)
10:47:41 Join shai [0] (
10:47:53 Quit shai (Client Quit)
10:51:21 Join S_a_i_n_t [0] (~st.lasciv@
11:00:25 Join LinusN [0] (~linus@gateway/web/cgi-irc/
11:00:26 Quit LinusN (Changing host)
11:00:26 Join LinusN [0] (~linus@rockbox/developer/LinusN)
11:06:31 Join kugel [0] (kugel@
11:06:32 Quit kugel (Changing host)
11:06:32 Join kugel [0] (kugel@rockbox/developer/kugel)
11:09:33 Join einhirn [0] (
11:14:49 Quit bluebrother (Ping timeout: 246 seconds)
11:16:35 Join bluebrother [0] (
11:16:35 Quit bluebrother (Changing host)
11:16:35 Join bluebrother [0] (~dom@rockbox/developer/bluebrother)
11:32:35 Join Suit_Of_Sables [0] (
11:33:06Suit_Of_SablesI wish I still had my iPod 5.5G T_T
11:33:35Suit_Of_Sablesdamn encoded firmware
11:33:52***Saving seen data "./dancer.seen"
11:37:11 Quit FlynDice (Remote host closed the connection)
11:45:32 Join Frampis [0] (
11:57:54 Join Horscht [0] (
11:57:54 Quit Horscht (Changing host)
11:57:54 Join Horscht [0] (~Horscht2@xbmc/user/horscht)
11:59:16 Quit DerPapst (Read error: Connection timed out)
12:01:58 Join DerPapst [0] (
12:02:20 Join MethoS- [0] (~clemens@
12:05:03Frampisis there a way to customise the splash screen and the USB icon that's displayed when the player is connected to a computer?
12:07:25S_a_i_n_tyes, but you have to edit the rockbox source code.
12:07:33S_a_i_n_twhat target do you have?
12:07:37FrampisI was afraid of that :P
12:07:56S_a_i_n_tI have a "dock connector" USB icon i made for ipod.
12:08:43 Join wanttoknow [0] (
12:10:25FrampisI was thinkin I'd crop an approppriate sized version of this:
12:10:45Frampisbut since the graphics are hardcoded I'm not going to bother
12:11:07S_a_i_n_tthat's a lazy way of looking at it :P
12:11:15FrampisI can't code at all
12:12:48S_a_i_n_tneither could I until I learnt how :D
12:14:59S_a_i_n_tThat image would be pretty hard to pull off, but not impossible.
12:16:25S_a_i_n_tFrampis: this is my "USB Connected" logo at the moment the magenta shows as transparent on the DAP screen.
12:22:00FrampisI'd probably just make it fil the whole screen, I'm too lazy to photoshop it
12:22:06 Join dfkt [0] (dfkt@unaffiliated/dfkt)
12:22:21Frampisbut I'm not going to do it anyways because it's hardcoded :P
12:29:14 Quit Suit_Of_Sables (Quit: Suit_Of_Sables)
13:02:41 Join _zic [0] (
13:06:38 Quit Blast_Tyrant (Ping timeout: 265 seconds)
13:17:49pixelmaif someone is bored - I think the target names table on needs updating
13:21:39Bagdercan Zagor's script be used?
13:24:36pixelmano idea. But while you are here: could you change the CIA bot settings so it also announces the branch? I saw it in other channels and I think it would be good to know while we are committing to the release, www, and the mdct (?) branch
13:24:53AlexPThat is a god idea
13:25:20pixelma(and trunk of course)
13:26:35AlexPer, good idea :)
13:33:53***Saving seen data "./dancer.seen"
13:35:12 Join David [0] (
13:35:37 Nick David is now known as Guest89113 (
13:36:19Guest89113Upgrading New Rockbox version issue
13:36:49pixelmamore info needed
13:37:06BagderCIA is probably Zagor's business, I haven't fiddled with that
13:38:43 Quit bzed (Read error: Connection reset by peer)
13:40:01 Join Sajber^ [0] (
13:41:37Guest89113Upgrading my Toshiba F60 to the new Rockbox from the old rockbox version the first upgrade didn't seem to work so I started again but thought I should uninstall the rockbox AND bootloader first to start fresh. Used the Rockbox utility to do this and all seemed fine. Tried to reinstall everything but now Rockbox Utility cannot recognise the Toshiba. When turninf off the Toshiba text on the unit states 'Rockbox boot loader, Version 3.0, loading firmware,
13:44:43 Quit Zarggg_ (Read error: Connection reset by peer)
13:45:08 Join bzed [0] (
13:45:52 Join Zarggg [0] (
13:55:43 Quit Guest89113 (Quit: CGI:IRC (EOF))
13:57:07 Nick evilnick is now known as S_A_N_T (~evilnick@rockbox/staff/evilnick)
13:57:23 Nick S_A_N_T is now known as evilnick (~evilnick@rockbox/staff/evilnick)
14:02:25 Quit evilnick (Quit: Leaving)
14:07:23 Quit MethoS- (Remote host closed the connection)
14:13:58 Join maruk [0] (
14:27:18 Quit _zic (Quit: Ex-Chat)
14:37:17 Join casainho [0] (
14:40:40 Join phroggyman [0] (
14:44:54 Quit petur (Read error: Connection reset by peer)
14:45:14 Join petur [0] (~petur@rockbox/developer/petur)
14:46:23 Quit petur (Read error: Connection reset by peer)
14:46:59 Join petur [0] (~petur@rockbox/developer/petur)
15:04:50 Part LinusN
15:07:29 Join Blast_Tyrant [0] (
15:09:10 Join liar [0] (
15:23:01 Join funman [0] (~fun@rockbox/developer/funman)
15:23:49funmanFlynDice: the patch JdGordon used didn't restore GPIOA_DIR before branching to OF, the rest looked correct so I suppose it was the problem
15:33:55***Saving seen data "./dancer.seen"
15:40:40 Quit PaulJam (Ping timeout: 240 seconds)
15:46:13TorneHmm. If I skip ahead tracks while in the debug buffering screen, the useful data count doesn't go down
15:46:25Tornei would expect it to drop by one track's worth..
15:46:47Tornetrack count goes down, but nothing else changes
15:48:32Horschti see what you are trying to do here. I noticed the same
15:49:09Torneoh, wait, it does the same if i skip *not* in the debug screen
15:49:13Tornethat seems.. wrong :)
15:49:21Torneif usefl goes down while playing then it should also go down while skipping
15:49:36Torneand yes, you can probably guess what i'm trying to do :)
15:49:48Tornenot actually *your* bug, though, just testing it with DMA
15:49:54Horschtbut then again, the pause also appears when I did not skip any songs.
15:50:11Horschtand there i though i was the center of the universe
15:51:03Torneno, some people have reported the same happening with DMA
15:52:04 Join panni_ [0] (
15:52:52Horschti think it might be a CPU priority issue, though. As increasing the anti-skip buffer did not solve my issue (I assume the people reporting the DMA issue are actualy experiencing the same issue as I)
15:54:36Tornepossibly, but they are also messing around with crazy combinations of patches
15:54:44Tornee.g. not boosting during buffering, or using the gui boost
15:55:43 Quit phroggyman (Quit: CGI:IRC (EOF))
15:57:17Horschti stoped using custom builds a while ago, now I only use current builds
15:57:49Torneincidentally you *might* actually find that DMA makes your problem go away :)
15:58:09Tornebecause buffering with DMA means the buffering thread can yield while the DMAs are inprogress
15:58:10 Join casainho_ [0] (~chatzilla@
15:58:42Tornei can't reproduce your problem, btw
15:58:45Tornei just tried it
15:58:56Tornehuge album art theme, and equaliser running
15:58:59Tornewith just 5s antiskip
15:59:04Torneand it works fine over rebuffer
15:59:08TorneI have DMA on, though :)
15:59:16Horschtwhat theme were you using?
15:59:36Horschthm... weird indeed...
15:59:47Tornenot weird, if it's just cpu starvation..
15:59:57Torneas i said, ata dma reduces cpu usage :)
16:00:11Horschtwhat's the size of your
16:00:19Torneno idea
16:00:26Tornewhatever picard downloaded for me from amazon
16:00:28Tornehuge, i expect
16:00:31Horschtmine are 240x240
16:00:35 Quit casainho (Ping timeout: 256 seconds)
16:00:36 Nick casainho_ is now known as casainho (~chatzilla@
16:02:07Torne500x500 apparently
16:02:11Torneat least for the album i was using
16:02:12Unhelpfulis it a bmp? jpeg costs more to decode but might actually be cheaper if you're scaling down by a large factor, because of the 1/8 scale-on-decode
16:02:25Horschti myself use bmp
16:02:39Tornei use whatever picard happens to download automatically, so hey :)
16:02:40Unhelpfulbmp will be more i/o of course ;)
16:03:03Horschti use 240x240 bmp because that's what my wps uses
16:03:31Tornealso, i normally have antiskip set to a minute :)
16:03:36Tornei reduced it to 5s for this though
16:03:43Horschtbut it'll still skip on rebuffer. If I use cabbie (default theme), no skipping.
16:03:53Horschtand cabbie uses what? 100x100?
16:04:05Tornei think so.
16:04:57*Unhelpful wonders if the jpeg loader is yielding on unscaled loads... or the bmp loader for that matter
16:05:25Torneanyway it definately sounds like a cpu issue
16:05:33 Join MethoS- [0] (~clemens@
16:06:16Unhelpfulthe scaler yields. the bmp loader does not, and it does not call the scaler if you're loading a file of the desired size... or at least that's how it looks
16:06:22Torneaha :)
16:06:39Tornewhere's the code?
16:06:51Horschtyou and your programming jibberish :p
16:07:55TorneHorscht: Try resizing one of your album arts to be bigger
16:08:01Torneand see if that makes the problem go away :)
16:08:18Tornethat would pretty much prove it
16:08:50Horschtone? Wouldn't I need to resize it all?
16:09:12*Torne was assuming you'd test it by just playing one album
16:09:48Horschthm... I usualy just random play all my library... but for that purpose i think i can make an exception...
16:10:05Tornei assume it does it frequently/every time?
16:10:26Horschtevery time
16:10:39Torneright. then any old single album that's bigger than the buffer will do ;)
16:10:46Unhelpfulthe scaler yields per line read.... erm, processed?
16:10:59TorneUnhelpful: yah, someting like that
16:11:22Tornei guess "large bitmap which is coincidentally the right size" is probably the worst latency case for AA loading
16:11:23Horschtnow... where did I put my ipod?
16:11:24UnhelpfulTorne: i wrote it.... but i just double-checked. ;)
16:12:55 Quit wanttoknow (Quit: CGI:IRC (EOF))
16:14:10Horschtdamnit, i wish my ipod had a phone built in... I could ring it and find it that way
16:15:18UnhelpfulHorscht: they make an ipod like that, you know. ;)
16:15:29Horschtbut I can't run rockbox on it
16:18:44 Join robin0800 [0] (
16:20:56 Quit Sajber^ (Read error: Connection reset by peer)
16:24:13Horschtwell, at least it also happens with my insanely oversized folder.jpg coverart
16:24:24Torneoh, it does?
16:24:35Tornethat might still be it, i guess
16:24:47Tornedepends if the jpeg loader is yielding *enough*
16:24:56Horschtyes, but that is insanely oversized (1800x1800)
16:25:15 Quit bmbl (Ping timeout: 246 seconds)
16:25:17TorneSo even with the "free" scale down the rest of the work might be too much to do in one go without yielding more often than it does
16:25:47Tornetrying a 250x250 bmp or similar might be very interesting though still
16:25:59HorschtI am gonna try a moderately oversized jpg and then bmp, though
16:29:41 Quit casainho (Read error: Connection reset by peer)
16:32:31Horschtjpeg, 300x300 = same
16:32:35Horschtnow for bmp
16:33:25Torneyou mean it still skips?
16:33:59Horschtstill skips
16:38:42Horschtweird... 300x300 bmp did not
16:38:57TorneOk, that's an interesting data point :)
16:40:31 Quit panni_ (Read error: Connection reset by peer)
16:40:52 Quit funman (Quit: free(random());)
16:42:43Tornethe resizer yields between each row of the bitmap
16:43:16Torne(not sure if that's input row or output row but either way that's 250+ yields)
16:43:19Tornethat probably helps :)
16:43:30Torneit's interesting that jpeg doesn't fix it though
16:46:49 Join toffe82 [0] (~chatzilla@
16:49:33 Join fyrestorm [0] (
16:52:44 Nick fxb__ is now known as fxb (
16:55:07UnhelpfulTorne: the scale_h_* functions are called per input row
16:55:14Unhelpfulmaybe move the yield from there to the loaders?
16:55:39Tornemaybe, but if jpeg also doesn't work..
16:56:30*Torne can't reproduce this problem at all with jpeg album art
16:56:49Tornei can't easily get any bitmaps at the moment :)
16:57:03Torneand my build has full DMA enabled which might well mask the problem anyway
16:57:09 Join anewuser [0] (anewuser@unaffiliated/anewuser)
16:57:48TorneHorscht: I'm probably going to be committing the buffering cacheline alignment changes soon, which will mean that almost all buffering reads will be done with DMA on ipod
16:57:55Tornethis might well make your problem go away :)
17:00:52Tornebut it would be nice to work out what it is anyway..
17:02:45Torneanyone have an opinion on whether the buffering cacheline alignment stuff should be ifdef'ed or not? :)
17:06:41 Quit kugel (Ping timeout: 258 seconds)
17:11:59 Quit krazykit (Ping timeout: 256 seconds)
17:13:15 Join krazykit [0] (~kkit@
17:15:12 Join jobec [0] (
17:17:09 Nick YPSY is now known as Ypsy (
17:19:37jobecevening. i ran a couple of battery tests on clipv1 (of and rb). if these results seem relevant, would it be possible to get a write access on wiki?
17:20:03Torneanyone who demonstrates they're a human being by asking for it ca nhave write access, yes :)
17:21:31 Quit krazykit (Ping timeout: 260 seconds)
17:22:33 Quit MethoS- (Remote host closed the connection)
17:25:52 Join MethoS- [0] (~clemens@
17:29:04AlexPjobec: What is your wiki name?
17:30:43AlexPjobec: OK, done
17:31:25AlexPno problem
17:33:58***Saving seen data "./dancer.seen"
17:39:56 Quit Bagder (Quit: It is time to say moo)
17:41:06CIA-8New commit by b0hoon (r24439): Packard Bell Vibe 500: add the simulator
17:41:06 Join bmbl [0] (~Miranda@
17:41:07 Quit bmbl (Changing host)
17:41:07 Join bmbl [0] (~Miranda@unaffiliated/bmbl)
17:42:39 Join Res1 [0] (
17:42:49 Join krazykit [0] (~kkit@
17:48:02Tornegevaerts/people in general: do we like code that's fractionally smaller or code with less ifdefs? :)
17:48:39Tornekugel expressed the desire to have the buffering alignment stuff without the ifdefs; this technically makes the code slightly bigger on targets which don't care..
17:51:24 Join TheSeven [0] (~theseven@rockbox/developer/TheSeven)
17:56:34gevaertsTorne: what does "slightly" mean?
17:57:41Tornesome number of bytes? :)
17:57:58Torneif the compiler were superhumanly powerful it would optimise almost the whole lot away when the storage align mask was 0
17:58:19Tornenot very big
17:58:43 Quit petur (Quit: work->home)
17:59:18 Join guest [0] (
18:00:09Tornei dunno, i've not compiled it to compare ;)
18:00:26Tornethe patch is tiny, and even if it failed ot optimise it out it'd only be a few dozen instructions at worst
18:00:39Torne(and it makes buffer handles one word longer)
18:00:45GodEaterI think that's acceptable
18:00:55gevaertsI'd do it without ifdefs then
18:01:14Torneit should work for other targets which would like stuff to be aligned properly, also :)
18:01:26Torneeven if they don't actually require it, it might be faster? depends how their caches work
18:02:49*Torne ponders doing horrible things with the gcc return address intrinsic and logf to track down commonly used unaligned buffers :)
18:04:01guesthello, i want to add an option "shut down" to home screen of current theme. CAN ANYONE PLS HELP!
18:06:16guesti have been asking this question here for last three days.. is the question too dumb?
18:06:55pixelmaI at least don't understand what you want to do (and haven't seen this question before)
18:07:42guestthanks for replying.. in a few themes, I have seen an option "shut down" at the home (the page where files, database, all options are there)
18:07:58topikthe "home screen" being the main menu, or the "while playing screen" ?
18:08:09guestmy current theme doesnt have that, and theme owner doesnt want to add it.. so i thought i would myself add that for me
18:08:21topikcan you list one that has it?
18:08:27Tornealso, hm, where shohuld the definition of the storage alignment go? dreamlayers has it in config.h (the top level one, not target specific)
18:08:33guestyes, main menu.. pardon me for my less knowledge
18:08:33guestyes, 1 min
18:08:43Torneit's not really a target specific parameter, it's determined by the chip type/DMA config/etc
18:08:57Torneis config.h the right place for that kind of thing?
18:10:07topikseems it requires the patch from JdGordon's link, not just a theme
18:10:44guest@JdGordon, thanks for link, i will try to apply patch on my ipod's rockbox
18:10:59guestwhich i am sure wont be an easy task!
18:11:02 Quit guest (Quit: CGI:IRC)
18:11:05Torneyes, it is not anything to do with the theme, it's just that patch. The theme author took the screenshot on a build with that patch :)
18:11:47 Join guest [0] (
18:12:01TheSeven[18:10]<Torne>yes, it is not anything to do with the theme, it's just that patch. The theme author took the screenshot on a build with that patch :)
18:12:30guestthanks Torne..
18:12:45guestlet me try that out.. you guys here for some more time? till then i would try appyling patch
18:13:13Horschtyou have to compile from source
18:13:43guestooops :( so i would be required to hv some maven/svn client at my laptop, right?
18:13:52guestlet me find out instructions to do so somewhere on the rockbox site
18:14:47mc2739guest: and
18:14:53guestthanks mc2739
18:14:57 Join Omlet [0] (omlet05@
18:16:46CIA-8New commit by torne (r24440): Align addresses in the buffering code to STORAGE_ALIGN_MASK if the target has one. ...
18:17:25*Torne thinks he could've worded the commit comment for that more awkwardly if he tried. Maybe. :)
18:17:41*Torne also expects this will ruin the "All 158 builds are OK" record :)
18:20:21Torneooh, it worked
18:20:45Tornebinsize increase is up to 208 bytes
18:20:52Tornebut that's mostly only on the targets that actually need it
18:21:08Tornethe ones that don't are <150 bytes
18:21:47TheSevenhm, how much space is still left on the ondio rombox?
18:22:13Tornedunno, but this doesn't affect hwcodec
18:25:12TheSevenwell, it did. but probably rather because of that gcc randomness
18:28:33 Join bertrik [0] (
18:29:16 Join tarbo [0] (~me@unaffiliated/tarbo)
18:30:40bertrikthere's more than 50% difference between a battery runtime benchmark on my clip v1 (1GB) vs the one done by JouniPaulus (clip v1, 2 GB)
18:31:55bertrikHe ran the benchmark with 256 kbps mp3 and I did it with 160 kbps ogg, could this explain the huge difference?
18:31:58jobeci noticed that myself, too. but still the experiences without taking time have been in the same order
18:32:56 Quit maruk (Quit: Leaving.)
18:33:09 Join Kitr88 [0] (
18:33:57jobeci can run it again after recharging for verification
18:34:05 Join JdGordon1 [0] (
18:34:38bertrikjobec, I believe you even without verification
18:34:49 Quit Kitar|st (Ping timeout: 265 seconds)
18:35:49 Quit tarbo (Ping timeout: 252 seconds)
18:36:10amiconnThe buffer align thing for dma might be useful on coldfire as well
18:36:26amiconnThere it would be sufficient to aligh stuff to even address
18:36:37 Join kugel [0] (~kugel@rockbox/developer/kugel)
18:37:00UnhelpfulTorne: the image readers will probably be doing lots of reads to unaligned buffers - iirc the layout when scaling is bmp struct, output bmp buffer, decode and read buffers, scaler buffer
18:37:11amiconnRight now we don't do dma, but we could. It wouldn't be faster, but it would free cpu cycles
18:37:30Tornebuffering is already aligned to four without setting STORAGE_ALIGN_MASK
18:37:57TorneUnhelpful: and yes, that's the kind of thing i'm going to investigate
18:37:59bertrikjobec, just makes me wonder: is it the codec? or the battery itself? or the fact that mine is 1GB and yours is 2GB?
18:38:06TorneUnhelpful: but it's only relevant when the buffer is bigger than a sector
18:38:08amiconnDMA is only possible for even alignment though, since the coldfire dama engine cannot split words
18:38:32bertrikI do remember the 1 GB clips being a bit different than the 2 GB and bigger clips
18:38:58 Nick fxb is now known as fxb__ (
18:39:07Torneamiconn: well, you could take advantage of the stuff i'm doing then :)
18:39:15JdGordon1can someone commit 10945?
18:39:21amiconnbertrik: MP3 is significantly more demanding than vorbis on ARM. That combined with the higher bitrate...
18:39:33UnhelpfulTorne: ah. that might be a non-issue, then... bmp uses an on-stack read buffer that is sized, at maximum, to hold a screen row of 32-bit values. jpeg uses a *much* smaller read buffer, 256B i think, though we could easily make it sector-sized and aligned if it would speed things up.
18:39:49TorneUnhelpful: if it's less than a sector it reads into the cache in file.c and then copies from there
18:40:01TorneUnhelpful: obviously the cache in file.c needs to be aligned :)
18:40:03Tornewhich i've not done yet
18:40:04bertrikamiconn, but would that account for more than a 50% runtime difference?
18:40:24 Join Bullet` [0] (
18:40:37amiconnDepends on how much power the cpu draws compared to the other components
18:40:50UnhelpfulTorne: sector == 512B? if we're still speaking of a reasonable size we could round the buffers for both up to sector sizes...
18:42:40bertrikI'll start a clip v1 battery benchmark playing mp3 tonight then
18:42:47Unhelpfuland jpeg should probably yield per row-passed-to-scaler when scaling, and per MB-row when not, i would think... if not scaling the reader itself is surely where the heavy lifting is happening
18:43:11 Quit DerPapst (Quit: Leaving.)
18:47:13kugelbertrik: the default clock on samsas is high enough to do ogg/mp3 without boosting at all, so the difference should be subtle (if any)
18:47:43TorneUnhelpful: that might *also* be worth doing, yes, but that's kinda a seperate thing
18:48:08Torneand it's not just sector *sized* reads, it's sector aligned reads :)
18:48:10kugelTorne: worth hacking a test into test_disk if sotrage_align_mask > ?
18:48:35Tornekugel: I have an expanded crazy test_disk that messes around with raw sector accesses, i might commit that at some point
18:48:51Tornebut yes, it probably should also use storage_align_mask instead of just assuming that 4-byte-aligned is enough
18:48:51kugelyea, why not
18:49:05kugel(to both :) )
18:49:09Tornewell the "why not" would be "because the write speed part of the test overwrites bits of your disk"
18:49:11 Join tarbo [0] (~me@unaffiliated/tarbo)
18:49:30TorneI assumed that my freshly formatted ipod had nothing useful outside of the first 1mil sectors so I just arbitrarily told it to write over the rest :)
18:49:32amiconnkugel: That's not true if the core has an efficient sleep implementation
18:49:44UnhelpfulTorne: that can be done quite easily for the jpeg reader - it does disk reads to fill a buffer, which is processed via a getc-type interface
18:50:00TheSevendoes anyone have an idea what could have caused that massive ramsize fluctuation in r24417 on nano2g?
18:50:13TorneUnhelpful: I could *also* instrument the file code to find out where we are not doing zero copy IO
18:50:23kugelamiconn: any idea if we have that? I think we use the generic sleep (which all arm except pp do?)
18:50:30Tornewe'd need people to run these builds and give their logf results back though
18:50:39Tornesince just me doing it is not going to reflect all use cases :)
18:51:03TorneUnhelpful: anyway i'm not going to do this, like, now. :)
18:51:29*Torne waits to see if anyone discovers their player burning to the ground after the DMA stuff first :)
18:51:40Torneespecially the pp502x devices that nobody actually tried ;)
18:52:08bertrikkugel, amiconn: I don't believe the dramatic effect of the codec yet, but we'll see what happens when I run another runtime benchmark
18:52:09JdGordon1pp502x is my daily DAP
18:52:17amiconnTheSeven: You mean G4 greyscale?
18:52:27TorneJdGordon1: which, though?
18:52:37Tornei'm pretty sure it works on most/all the ipods
18:52:47Tornebut afaik nobody ever tested the 9708 patches on non-ipods
18:53:02kugelwell, I tested it quickly on my samsung
18:53:09TheSevenamiconn: don't get what you mean?
18:53:10JdGordon1nag me this evening and I can test on a e200
18:53:20Tornewell, it's enabled for all pp502x, and it should now be being used for 95%+ of buffering reads
18:53:25Torneso if there's a problem i'm sure we will hear about it :)
18:53:33amiconn[18:50:02] <TheSeven> does anyone have an idea what could have caused that massive ramsize fluctuation in r24417 on nano2g?
18:54:04amiconnThe table shows such fluctuation for ipod G4 greyscale, not nano G2
18:54:17TheSevenoops, yes
18:54:29TheSevenwell, anyways, what could cause such a thing?
18:54:41kugelinvestigate :)
18:55:13kugelthe build system isn't totally reliable yet so I tend to blame that
18:55:13*Torne votes "wizards did it", since no code changed
18:55:19 Quit tarbo (Ping timeout: 260 seconds)
18:55:23amiconnThat build was built on saratoga's machine. Maybe there's something special about the compiler?
18:55:32amiconnIt's the standard version though...
18:55:46 Join saratoga [0] (~9803c6dd@gateway/web/freenode/session)
18:55:46 Quit saratoga (Changing host)
18:55:46 Join saratoga [0] (~9803c6dd@gateway/web/freenode/x-jvqiuerqlgaexlou)
18:55:55saratogaGSOC2010 has been announced
18:56:15saratogawe should probably make an effort to promote it, gather ideas for prospective projects, etc
18:56:23 Join DerPapst [0] (
18:56:41*kugel wants to participate this time
18:56:48 Quit kaniini (Ping timeout: 248 seconds)
18:56:49JdGordon1are we going to bother?
18:57:15saratogai've heard from a few current SVN members about applying, so I hope so
18:57:27JdGordon1we havn't really had much luck with gsoc, and it is a PITA
18:57:33saratogasure we have
18:57:39 Join webguest50 [0] (
18:57:40saratogaWMA, Cook, ATRAC3 codecs
18:57:47kugelare you saying 2009 wasn't a success?
18:58:06topiktrick someone into the rockbox-as-an-app thingy
18:58:14saratogathat would be kugel
18:58:33kugelthat's my fallback at least :p
18:59:06 Join evilnick [0] (~0c140464@rockbox/staff/evilnick)
18:59:10saratogaalso MOB and HID were GSOC projects
19:00:43saratogamaybe this year we'll get lucky and get someone interested in more MPEG codec work too
19:00:48saratogaMP3 and AAC could use cleaning up
19:01:04JdGordon1so we get one good one per year and 2 or 3 mostly flops
19:01:08kugelis cleaning up gsoc worth?
19:01:18saratogayes I think so
19:01:29 Join funman [0] (~fun@rockbox/developer/funman)
19:01:30saratogai'd like to see the ffmpeg MP3 decoder merged into libmad
19:01:40saratogaand the ffmpeg aac decoder replacing most of libfaad
19:02:01saratogabetter mp3 is by far the most widely used codec, its also one of the least efficient in rockbox
19:02:14JdGordon1embedded AA!
19:02:21JdGordon1and lyrics support
19:02:26funmansaratoga: isn't it the oldest codec designed ?
19:02:33kugelraap would be interesting since it (imo) involves cleaning up sim stuff (=refactoring into the target tree) and touchscreen improvements
19:02:38saratogayou mean oldest format or oldest in rockbox?
19:02:45funmanoldest format
19:02:47topikreplacement mp3 based on ffmpeg would be more glorious than polishing the current one
19:03:01funmanafaik mp3 was the "first" codec on archos :P
19:03:09amiconnsaratoga: Not entirely true. It's less efficient than vorbis *on arm*. On coldfire it's the other way round
19:03:16webguest50I have bought I toslink cable and it planing to record from Spotify, but I don't understand presplit gap. And which record config is the best for recording from Spotify?
19:03:22saratogaon x86 the ffmpeg mp3 decoder is only marginally faster then libmad, so i tend to think its worth merging
19:03:47saratogaamiconn: yes but thats not due to algorithmic efficiency, but due to brute force EMAC power in the QMFs
19:04:06saratogai would prefer to have an algorithmically we designed decoder as it will benefit all targets
19:04:27saratogaand anyway tremor would destroy libmad on coldfire were it to make reasonable use of EMAC
19:04:30 Join pamaury [0] (
19:05:08saratoga"i would prefer to have an algorithmically we designed decoder as it will benefit all targets" -> "i would prefer to have an efficient decoder, as that will benefit all targets"
19:05:22saratogasorry trying to do two things at once
19:05:34amiconnSo why didn't someone interested in vorbis do it?
19:05:46saratogano one has coldfire targets anymore
19:06:03kugelI could also have a step on porting rockbox to that yh-j70
19:06:17saratogai need to get a CF target actually
19:06:47saratogaamiconn: that reminds me, if you get a chance would you benchmark the new IMDCT library on coldfire (preferably with WMA but vorbis will work too)
19:06:49*amiconn has several, but isn't really interested in vorbis
19:07:12 Quit JdGordon1 (Ping timeout: 265 seconds)
19:07:24saratogathe new library should be pretty easy to optimize for CF, but will take some ASM as GCC fails quite badly on ARM, and so probably CF too
19:07:39amiconnGCC knows nothing about coldfire emac
19:08:01saratogathe biggest issue is actually just the butterflies, not EMAC
19:08:08saratogagcc completely screws them up
19:08:41saratogathe new lib is pretty close to the lowest known multiply count for a fourier transform, which is why PP gains so much from it
19:09:14 Join tarbo [0] (~me@unaffiliated/tarbo)
19:09:21saratogathe hand optimized ASM is > 2x faster then the gcc generated code for the pure load/butterfly/store sections
19:09:34amiconnI don't understand much of the i(m)dct business, but at least for 2d, a straight, "brute force" idct using emac is more efficient than a butterfly implementation on coldfire afaiu
19:10:00saratogasince gcc thinks arm like it when you load operands one at a time, store them, and then load them again for the next butterfly
19:10:43saratogaamiconn: an imdct is basically just an N point 1D FFT with N (or maybe 2N I forget) multiplies before/after
19:11:23saratogathe new algorithm is nice in that it has much lower add/mul/load/store counts then the old (although addressing is somewhat uglier)
19:11:30 Join JdGordon| [0] (~Miranda@nat/microsoft/session)
19:11:30 Quit JdGordon| (Changing host)
19:11:30 Join JdGordon| [0] (~Miranda@nat/microsoft/x-dfsvfwyxlfljkbqv)
19:12:00amiconnI.e. coldfire doesn't profit from lowering the multiply count, it profits most from balacing and fusing multiplies and adds
19:12:25saratogayeah that makes sense
19:12:40guesthey, how can i get 3.4 released code using svn?
19:12:55saratogabut i think the real issue with all the transform codecs on CF is just that gcc does a terrible job at the bufferflys, and they're all in c
19:13:07kugelguest: checkout the v3_4 tag
19:13:08 Quit webguest50 (Quit: CGI:IRC)
19:13:17saratogathe lack of EMAC is probably the second biggest issue
19:13:23kugelI suggest to wait for 3.5 though which is due tomorrow
19:13:27amiconnA mac.l is single cycle (albeit with 3 cycles latency), and a mac.l with parallel load is 2 cycles. The loading part of the latter is more efficient than pre-loading operands with movem.l (as that needs n+2 cycles for n regs)
19:14:12guestohh, is it only tomorrow! great. but the code wud be frozen by now, right? ;) can i get link to that pls..will download n apply shutdown patch
19:14:19amiconnAn ordinary muls.l / mulu.l in turn needs 3 cycles...
19:14:42saratogaalso interesting, on PP WMA 192k (with the awful multiplier) is faster then MP3 192k on CF (using all EMAC)
19:14:52funmanguest: just checkout branches/v3_5 instead of trnk/
19:15:43guestyou mean i shoudl type in: "svn co svn:// rockbox"
19:15:54saratogain spite of WMA using a lot of DRAM while CF can fit mp3 almost entirely into IRAM
19:16:06kugelsvn co svn:// rockbox
19:16:31 Quit tarbo (Ping timeout: 252 seconds)
19:16:40amiconnLooks like we need to rethink iram allocation on PP5020 in almost all codecs
19:16:41guestokay, 1 more question. to apply that patch, i can survive by downloading only the rockbox leg.. not the complete tree, right
19:16:58amiconnThen it should be possible to get it much closer to PP5022 speed
19:17:16kugelthe full tree
19:17:18 Join domonoky [0] (
19:17:18 Quit domonoky (Changing host)
19:17:18 Join domonoky [0] (~Domonoky@rockbox/developer/domonoky)
19:17:48guestbcoz i dont need code for rockbox utility...
19:17:53amiconnArrays which are frequently accessed (in loops) and are small enough to fit completely in the cache should go into dram
19:18:07gevaertskugel: I'd say that whether or not cleanup is OK for gsoc depends on who does it. I don't think I'd agree with someone new doing that sort of thing
19:18:19amiconnIt seems that PP5020 has the inverse effect of PP5002 (although to a lesser degree)
19:18:25pamauryCould someone try a manipulation on his device ? The requirement are: dircache available. Ability to do a custom build.
19:18:58amiconnWhile on PP5002 the cache is broken (one waitstate for every access), iram seems to be slightly slower (read: broken) on PP5020
19:19:25 Quit flydutch (Quit: /* empty */)
19:20:37amiconnOnly large arrays should go into iram (since accessing iram is of course still faster than cache misses)
19:20:52funmankugel: do you still want to revert matsch commit in branch ?
19:21:28kugelI would also like to revert it in trunk, since it's in fact broken for all other targets
19:22:25pamaurypamaury: nobody here to test something ?
19:22:25funmanare you sure of the fix ? there's not much time to test it now ;.
19:22:48 Join _zic [0] (
19:25:30kugelfunman: do you mean me?
19:25:59kugelI've reverted it locally once and it didn't show the brokenness
19:26:37funmaniirc you didn't revert it completely but just moved some code
19:29:13kugelno, I reverted it when bisecting the exact cause
19:29:29kugelwhat you mean is when I tried to fix it without completely reverting
19:30:01 Join tarbo [0] (~me@unaffiliated/tarbo)
19:30:23*pixelma tries summoning Zagor
19:31:53 Quit einhirn (Quit: Miranda IM! Smaller, Faster, Easier.
19:34:00***Saving seen data "./dancer.seen"
19:37:09pamaurycould someone try a manipulatioin on his device ? It requires a small change in one file + a recompilation of the firmware only + rebuild of tagcache database. Should take less than 5 minutes
19:37:19 Quit tarbo (Ping timeout: 260 seconds)
19:37:30saratogaamiconn: isn't IRAM cached on PP5020?
19:38:25saratogaor perhaps that is the difference, IRAM is cached on PP5022 and not on PP5020
19:38:40gevaertspamaury: I can have a go
19:38:58amiconnsaratoga: Nope, and it wouldn't make sense
19:39:17pamaurygevaerts: Ok. So first, you have to add #define LOGF_ENABLE in firmware/common/dircache.c (juste before the #include "logf.h") and recompile
19:39:20guest@kugel : its taking too long to download.. i know thats written on the webpage.. but cant we include this patch in to the upcoming build as well or its left for a reason
19:39:21amiconnIRAM is single-cycle sram on PP, as is the cache
19:39:31amiconnIt even can't be cached iirc
19:39:57saratogaamiconn: IRAM is cached on PP5024
19:40:10amiconnIt's not
19:40:10saratogathere is a performance advantage, see
19:40:25saratoga"Combining both cache and iram seems to be the fastest combination (it probably can do something in parallel on the HW level)."
19:41:03 Quit robin0800 (Remote host closed the connection)
19:41:12amiconnThat doesn't mean iram is cached
19:41:16kugelguest: maybe you can leave out bootloader/, rbutil/, manual/ and uisimulator/ but you aren't going to save much download (everything is compressed before transfering anyway) and you need to checkout the other subfolder speparately
19:41:26saratogaamiconn: ah you are right
19:41:28saratogajust misread it now
19:41:36amiconnIt cannot be cached. The cache controller only handles addresses below 0x40000000, and iram is fixed at that address
19:42:01kugelwell if it's as fast as the cache, it would be even a bad idea to cache it
19:42:39saratogakugel: presumably the IRAM is only 32 bits wide, so it could not service both COP and CPU at the same time for instance
19:42:42amiconnWell, unless it doesn't work as designed...
19:42:52JdGordon|kugel: what build are you running?
19:42:59kugelrc right now
19:43:02guest@kugel yes, me downloading complete package.. just that even compiling would take similiar (enormous) time, unless its fast..
19:43:17 Join Buschel [0] (
19:43:28guesti have a old generation laptop :( to add to all this
19:43:31amiconnIt seems that getting it right took them at least 3 tries (public tries, there may have been even more internal ones)
19:43:36kugelyou don't save compile time by only downloading half of the source. it would just fail
19:43:43saratogathe PP5003
19:43:58JdGordon|kugel: can you get a svn going and see if you see any background wierdness? apparently it was broken for a few days, I fixed it but I dont know if it will leave artefacts on the screen in dead areas
19:44:17saratogait apparently had improved cache over the pp5020
19:44:37 Join Tomis2 [0] (~Tomis@
19:44:49saratogahonestly though after all this AMS fun i am looking back at PP more kindly, they certainly could have done worse (by making the IRAM slower then DRAM for instance)
19:44:49 Quit Omlet (Quit: ( :: NoNameScript 4.22 :: ))
19:45:02kugelfunman: nothing changed in buffering.c since then so reverting should be safe
19:45:11amiconnThe 5003 was in interim version between PP5002 and PP5020 iiuc. It had the cache issue fixed; I don't know whether it has the iram issue we observed on PP5020
19:45:17amiconnPP5022 has that fixed as well
19:45:19guest@kugel i agree, i hope we can compile only 1 target.. i.e. the player and that too only for ipod..
19:45:19 Join kaniini [0] (
19:45:30kugelsaratoga: if we had docs we would love all PP probably
19:45:48JdGordon|B4gder: for the release builds, do you want to build them? or just copy them from someone? I can do a full set in 15min or so
19:45:51saratogai remember those posts from the ecos guy who had the docs
19:45:55JdGordon|assuming we are ready to release later today?
19:45:58saratogahe said something like "everything in the docs was a lie"
19:46:16kugelit has many gpio, fast caches and fast and many iram, a very useful user timer
19:46:18*amiconn wonders what happened to the port involving the new PP thing
19:46:28JdGordon|the sansa view?
19:46:35saratogaObo vanished
19:46:35amiconnSomething 6000-ish
19:46:56 Quit Tomis (Ping timeout: 248 seconds)
19:46:56 Nick Tomis2 is now known as Tomis (~Tomis@
19:48:07saratogai like the Nano2G chip, fast IRAM, faster ARM core, and somehow still an order of magnitude more power efficient then PP/AMS
19:48:15saratogashame sandisk didn't use it instead
19:48:17gevaertspamaury: ok, new build installed. What am I looking for?
19:48:58kugelsaratoga: ams was probably very cheap
19:49:04saratogayeah probably
19:49:17saratogaplus the integrated DRAM
19:49:20kugeleven the s3c2440 is the better soc, even though it's a lot older
19:49:50saratogabut anyway, what i was getting before was that I bet mp3 on CF could be a lot faster even than it is if we improved the design of the decoder
19:49:54*amiconn likes coldfire more than this arm stuff
19:50:01pamaurygevaerts: now, you have to do the following. (**) First go into database and start to play a playlist. Then stop it. (**) Go into settings->system->database->Intialise Now. Go into debug menu and wait until the databse build is finished. (**) Go back into database and play ANOTHER playlist. (**)
19:50:18pamaurygevaerts: (**) means "go to debug menu and copy the content of logf buffer"
19:50:23kugeland in fact, the ams choice didn't have an effect for the users (except the improved sound quality), the OF feels the same
19:50:25amiconnIdeally they should have chosen a SoC with proper cache though
19:50:33saratogawhats the advantage of CF over ARMv5E where you have the single cycle MAC?
19:50:42 Join jgarvey [0] (
19:51:30amiconnColdfire emac is more flexible, and it has more registers available
19:51:47*gevaerts enables dircache and starts again
19:52:05pamaurygevaerts: I looking for a dircache error: "fd access error"
19:52:10 Join tarbo [0] (~me@unaffiliated/tarbo)
19:52:12amiconnColdfire also has a division instruction
19:52:19saratogahow many registers does CF have?
19:52:55saratogai always though ARM9E was basically perfect for audio, fast loads/stores, single cycle DSP instructions, short pipeline, available with fast IRAM from some vendors
19:52:57amiconnIt has 8 address and 8 data registers. The pc is separate, so that makes one more reg than arm
19:53:16amiconnThe emac has 4 accumulators in addition to that
19:53:18saratogaplus gcc can target the MAC unlike CF
19:53:23Buschelanything speaking against lowering the minimum "disk_spindown" to 1 second? during the standard buffering process the HDD-spwindown should be as low as reasonable. I also would like to change default spindown from 5 to 2 seconds. Are there use cases that need 5 seconds spindown delay?
19:54:00kugelsaratoga: short pipeline? didn't v5 have a increased pipeline again?
19:54:09amiconnThe accumulators are 48 bit, and it can handle fixed point directly
19:54:38kugel3 on arm7, 5 on arm9tdmi, 9(?) on arm9e ?
19:55:13saratogastill 5 IIRC
19:55:43saratogaamiconn: whats the advantage over mulal on ARM? isn't it still one cycle for the mul, and one for the shift?
19:56:14amiconnBuschel: Spinning down too often is bad for the hdd. And when buffering, rockbox does quick spindown anyway
19:56:45saratogamultiply long accumulate
19:56:58 Join Grahack [0] (
19:56:59amiconnYou men smlal/umlal?
19:57:07saratogaoh maybe
19:57:24 Join moos [0] (
19:57:35 Quit moos (Changing host)
19:57:35 Join moos [0] (moos@rockbox/staff/moos)
19:57:40Buschelamiconn: ok, didn't know that there is a fast shutdown while buffering
19:57:50amiconnBuschel: 0.5 sec
19:57:56pixelmaI'd also guess that cases of spinning down and up quickly is more power hungry than keeping the disk spinning for a little longer
19:58:10Buschelamiconn: perfect :)
19:58:34saratogaIIRC arm can do 32x32+32 = 64 then shift off the top 32 bits in two clocks on arm9e, but i could be mistaken
19:58:47gevaertspamaury: no fd access errors seen
19:59:19 Quit tarbo (Ping timeout: 245 seconds)
19:59:30kugelsaratoga: some armv6 have 9-stage pipeline, maybe I confused it
19:59:31gevaertspamaury: full logf of the session at
19:59:45amiconnsaratoga: The 32x32->64 take away two precious general purpose registers though (of which arm already has effectively one less than cf)
19:59:49pamaurygevaerts: thanks, I'll look at the log
19:59:51saratogawhich is basically all i ever wanted for codecs, well that and a load multiple operation that can stride by a variable amount :)
20:00:03 Join efyx_ [0] (
20:00:18*amiconn would like 32 or even 64 general registers
20:00:19 Join mitk [0] (
20:00:26saratogayeah MIPS would be nice
20:00:41saratogathe amount of work making butterflies fit in registers was really, really annoying compared to MIPS
20:01:01amiconnWell, we have some mips targets. Unfortunately they're all touchscreen crap^h^h^h^hstuff
20:01:12saratogaactually i would also like a hardware complex mul instruction with single cycle latency :)
20:01:46saratogabut i don't think we'll see 6 operand instructions any time soon on arm
20:02:14pamaurygevaerts: hum, are you at svn HEAD ?
20:02:18amiconnWell, coldfire emac with parallel load are 5-operand instructions...
20:02:31saratogaso it does a load and a MAC in one op?
20:02:36mitkHi. Can an dev review and maybe commit FS #10945?
20:03:08amiconnYes (but not in a single cycle)
20:03:36saratogaCF seems neat, i need to get one eventually
20:03:36mitkOne cyccle review, second cycle commit? :)
20:03:38gevaertspamaury: should be, yes. I did run svn up before this
20:04:05saratogamitk: unfortunately i don't know much about playlists, maybe get JdGordon to look at it?
20:04:22kugelor stripwax
20:04:33pamaurygevaerts: That's strange, I would exact same log but with the error. Did you run the playlist like I told you ? Before AND after the database init ?
20:04:44JdGordon|mitk: hi, yeah I saw the patch and plan on commmiting tonight unless somenoe else can do it
20:04:55 Join shaggy-h [0] (
20:05:19 Join kugel_ [0] (
20:05:38*pamaury doesn't believe gevaerts , the log says the contrary
20:05:45*gevaerts is at r24440
20:06:42mitkJdGordon|: Is it worth to commit to the branch also?
20:06:51pamaurygevaerts: could you please retry ? The logf should display a line like "bind: 1//.rockbox/.playlist_control" between each dircache release at least. Here it's not the case
20:07:09 Quit kugel (Killed (NickServ (GHOST command used by kugel_!
20:07:14 Nick kugel_ is now known as kugel (
20:07:16 Quit kugel (Changing host)
20:07:16 Join kugel [0] (~kugel@rockbox/developer/kugel)
20:07:26JdGordon|mitk: maybe, it depends when we release,
20:07:43 Join ajb` [0] (
20:08:00mitkUnderstand. Bye.
20:08:04 Quit mitk (Quit: CGI:IRC)
20:08:14ajb`Hi, I've fixed a little bugget in the UI simulator code that broke bookmarks (and anything else that does a rename).
20:08:39ajb`If a comitter could look at and commit that would be ace thanks :-)
20:09:31kugelajb`: heh, I found that too but I always thought it was a sim limitation
20:10:09kugelI would change strncpy to strlcpy before committing though
20:10:30ajb`Do you want me to submit a new one?
20:11:32ajb`Hmm, where do we define our strlcpy?
20:11:37kugelit doesn't really matter I assume, but we don't use strlcpy in rockbox
20:12:07kugelajb`: firmware/common/strlcpy.c
20:12:22 Quit Buschel ()
20:14:27ajb`All the rockbox includes in the simulator seem to come from fireware/export/, should I create such an export for strlcpy?
20:14:49ajb`Or just include "firmware/include/strings.h" directrly
20:15:42 Join tarbo [0] (~me@unaffiliated/tarbo)
20:16:47 Quit jgarvey (Read error: Connection reset by peer)
20:18:25kugelajb`: string.h is included
20:19:13 Join perfectdrug_ [0] (
20:21:56ajb`kugel: but not strlcpy according to my compiler....
20:22:52kugelit probably includes the system's string.h first
20:23:11 Quit tarbo (Ping timeout: 264 seconds)
20:24:11kugeltry changing <string.h> to "string.h"
20:26:54 Join Xerion [0] (
20:27:11 Quit GodEater (Read error: Operation timed out)
20:27:25 Quit JdGordon| (Changing host)
20:27:25 Join JdGordon| [0] (~Miranda@rockbox/developer/JdGordon)
20:29:28 Quit shaggy-h (Ping timeout: 240 seconds)
20:32:35 Quit amiconn (Killed (NickServ (GHOST command used by amiconn_)))
20:32:36 Quit pixelma (Killed (NickServ (GHOST command used by pixelma_)))
20:32:36 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
20:32:39 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
20:32:48 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
20:32:54 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
20:33:04 Quit kugel (Quit: exit(0);)
20:33:56 Join Farthen [0] (
20:34:08pamaurygevaerts: when you did, did you select custom playlist or did you play, say, an album directly from the database menu ?
20:35:27 Join AdB3 [0] (
20:36:07 Join jgarvey [0] (
20:37:01AdB3could someone build the bleeding edge Windows Rockbox Utility for me???
20:37:24AdB3The custom versions for the D2 are missing
20:37:50 Join tarbo [0] (~me@unaffiliated/tarbo)
20:37:58 Quit funman (Quit: leaving)
20:38:29 Join shaggy-h [0] (
20:38:37 Quit AdB3 (Client Quit)
20:44:19 Quit tarbo (Ping timeout: 245 seconds)
20:44:33 Join GodEater [0] (
20:44:34 Quit GodEater (Changing host)
20:44:34 Join GodEater [0] (~bibble@rockbox/staff/GodEater)
20:48:09 Join kugel [0] (~kugel@rockbox/developer/kugel)
20:49:17 Quit saratoga (Quit: Page closed)
20:50:13 Part Llorean
20:51:41kugelit appears the buffering "regression" was there before
20:52:31ajb`kugel: I am including "string.h", something else odd must be happening
20:54:44 Quit guest (Quit: CGI:IRC)
20:55:11kugelfine, I won't revert r23680 as it appears it didn't cause the bug
20:55:23kugelbut we're going to have a sad "known issue" this time
20:58:28JdGordon|how rare is this?
20:58:41JdGordon|it does sound like a showstopped if its common
20:59:51kugelprobably not a show stopper
21:00:19kugelI'm apparently the only one who even found it, even though it's 100% reproducable
21:00:35JdGordon|whats the repro? I'll try it here
21:00:41JdGordon|is it target dependant/
21:01:05 Join tarbo [0] (~me@unaffiliated/tarbo)
21:01:40kugelturn on "skip to outro" and skip to the last 5s of the song and skip back again
21:01:59kugelit may work without skip to outro, but I have that always on
21:02:19kugelfunman reproduced it on his samsung (with 32MB ram) too
21:02:28 Quit S_a_i_n_t (Ping timeout: 272 seconds)
21:03:13JdGordon|and what happens if its hit?
21:03:29kugeldata abort, string corruption, freeze
21:03:33kugelstrange things
21:03:50JdGordon|not here
21:04:59ajb`Suspects he needs to tweak the include directives to include firmware/include
21:05:49kugelmaybe just keep strncpy as it seems to just use the host OS functions anyway
21:06:17ajb`kugel: In that case the patch is ready to go as currently is in Flypsray
21:07:18ajb`Next question, if I want to ensure a bookmark is created when the USB is plugged in where would I do it?
21:07:52 Quit tarbo (Ping timeout: 252 seconds)
21:08:03 Quit kugel (Remote host closed the connection)
21:13:00 Join Stephen__ [0] (
21:15:08 Quit Grahack (Quit: Tu m'as vu ?)
21:18:35 Join Casainho [0] (~chatzilla@
21:18:58bluebrotherdomonoky: with the release coming up, how should we handle the D2? AFAICS the bootloader installation places the firmware file to the wrong drive
21:19:58domonokyis there a big warning to the user, that he has to move the file ?
21:20:33bluebrothernone that I noticed when installing it on an usb drive
21:21:02domonokythen we should perhaps disable d2 in rbutil for this release.
21:21:07bluebrotherthere is also no notice to the user that the mountpoint selected has to be the sd card
21:21:37bluebrotherso while it's basically working it's nothing I consider suitable for the average user
21:21:49bluebrotheryeah, I was thinking about that too.
21:22:11domonokyjup, i think its not good to provide a "easy install tool" when it does wrong things for a target.
21:22:24 Join samlii [0] (
21:23:11 Quit Farthen (Ping timeout: 264 seconds)
21:23:21samliihow does Rockbox store what file and location is the current resume point?
21:23:27bluebrotherD2 users can still use svn binaries anyway
21:24:02 Join Farthen [0] (
21:24:02 Join tarbo [0] (~me@unaffiliated/tarbo)
21:24:48JdGordon|samlii: /.rockbox/.playlist_control
21:26:52 Quit efyx_ (Remote host closed the connection)
21:27:37*ajb` is confused as to why the USB code in wps is tucked into ffwd_rev() rather than the main loop
21:28:02 Join phanboy4 [0] (
21:29:39JdGordon|that is (presumably) so usb while its doing a ffwd or rewind works
21:29:48JdGordon|that is a secondary button loop
21:29:59JdGordon|the regular loop in wps.c handles usb
21:30:24CIA-8New commit by bluebrother (r24441): Give the user a hint on where find the required bootloader file.
21:30:59 Quit tarbo (Ping timeout: 264 seconds)
21:31:22ajb`JdGordon|: Ahh yes, I see.
21:32:17 Join efyx_ [0] (
21:34:04***Saving seen data "./dancer.seen"
21:34:08 Quit stooo (Ping timeout: 272 seconds)
21:36:02 Nick Ypsy is now known as YPSY (
21:36:06 Quit jgarvey (Ping timeout: 252 seconds)
21:36:10 Nick YPSY is now known as Ypsy (
21:36:34 Join fml [0] (
21:38:16fmlI have a small aestethic issue with the manual. Look at Do we need periods after the names of the menu items?
21:39:15 Join froggyman [0] (
21:39:21Unhelpfulsaratoga: xMULL is 3 cycles on arm9e, and the top word is not available on the next cycle. also as amiconn pointed out the low word will clobber a register. if you can keep the top bits of your constants clear (or wanted to use a signed mul in the first place) armv6 has a signed-only, top-word-only multiply
21:39:52amiconnARMv6 is a bit better here
21:40:48amiconnIt also has this 16bit simd stuff
21:42:11Unhelpfulif one of the values only needs 16-bit precision, SMULWy is the best bet. it's available from arm9e up, it's one cycle to execute, one additional cycle result latency, and it gives you signed 32x16->48>>16
21:42:23bluebrotherfml: the idea was to end each description item with a full stop. This looks good in the pdf manual (imo). No idea why the html puts a line break there.
21:42:34 Join FlynDice [0] (
21:42:58Unhelpfulamiconn: yes, the armv6 versions of the jpeg idct kill the competition, especially the small sizes.
21:43:32amiconnBut the typical armv6 is overpowered for a dap
21:45:29bluebrotherfml: seems the line break is caused by the use of the <dt> tag in the html output. Might be worthwhile to check how to remove that.
21:45:44fmlbluebrother: id doesn't look good in the PDF manual IMO. I must have missed the discussion. A full stop marks a sentence. But those are not seentences.
21:46:35 Join tarbo [0] (~me@unaffiliated/tarbo)
21:46:38fmlbluebrother: I've stumbled upon it in the PDF
21:47:36Unhelpfuli'm starting to think that maybe i should make the jpeg coeff/idct buffer 32-bit values on armv4... although it only really makes a difference when i can load rows, loading columns from the buffer is always going to be 3c/load :/
21:47:53ajb`I'm assuming the simulator should behave as the real target when "U" is pressed for USB insertion? i.e. sound stop playing
21:48:36bluebrotherfml: well, IMO a : looks even worse, and that is whats been used more often years ago. But some separator is needed.
21:53:42 Quit tarbo (Ping timeout: 252 seconds)
21:54:05TheSevenwhat about a colon as a separator?
21:54:10bluebrotherfml: see −− this was added by me :o over 3 years ago
21:54:10TheSevenor just a linebreak?
21:55:04 Join jgarvey [0] (
21:55:52linuxstbfml: Those periods also look odd to me. If there should be any punctuation, shouldn't it be a colon?
21:56:44bluebrotherwell, a colon looks worse IMO. But I'm not actively working on the manual anymore, so if the active contributors decide that a colon would be better that's fine.
21:57:37ajb`suspects the queue broadcast isn't working properly in simulator mode, the threads only see the USB connection on the second toggle.
21:59:53 Quit yosafbridge (Quit: Coyote finally caught me)
22:00:01*TheSeven thinks the way it's being rendered in the html version, with the full stop removed, would look best, also for the pdf (so just using a linefeed as the separator)
22:00:03amiconnUnhelpful: The 3cycles only apply to arm7. arm9 is faster, even if it's a v4
22:01:42 Join yosafbridge [0] (
22:02:00 Quit yosafbridge (Client Quit)
22:03:01fmlIMO it would look best without punctuation. No period, no colon. It's already marked clearly via formatting.
22:03:26TheSevenfml: that's what I'm thinking (in the HTML view, haven't looked at the PDF)
22:03:49 Join yosafbridge [0] (
22:04:25fmlTheSeven: I think it would look good in both
22:05:16 Join petur [0] (~petur@rockbox/developer/petur)
22:09:01Unhelpfulamiconn: ah, i see that. is there any way to differentiate between arm7 and arm9 cores with armv4?
22:09:11 Join tarbo [0] (~me@unaffiliated/tarbo)
22:09:27Torneanything with cp15 tells you exactly what it is; other than that it can be awkward
22:10:07ajb`How do recipients of queue_broadcast messages know who to send responses too?
22:11:32Unhelpfulalso, you'd mentioned that the multiplies killed any real chance of a swar efficiency gain on armv4... might removing the store between the add/sub part and the multiplies change that? it seems that unpacking the packed 16-bit values for the multiply could be rolled together with adding or subtracting them at fairly low cost
22:11:38gevaertspamaury: I think the first one was a single album
22:12:42gevaertspamaury: if you know exactly what I should get, why do you need a tester? ;)
22:13:27pamaurygevaerts: because I need to know if it's easily reproducible on another device.
22:13:55gevaertsdo you need another run?
22:14:11amiconnUnhelpful: #ifdef CPU_ARM7TDMI
22:14:14pamauryIf you have time, I'll be very pleased that you try another time
22:15:25gevaertsok. So start playing from the database, stop, re-init the database, and play another thing? What sort of playlist should I make?
22:15:55pamauryI don't know. On my device I played an album.
22:16:01 Quit JdGordon| (Quit: Miranda IM! Smaller, Faster, Easier.
22:16:04 Quit tarbo (Ping timeout: 252 seconds)
22:16:24 Quit phanboy4 (Ping timeout: 240 seconds)
22:16:41pamauryI'm not sure the type of the playlist is important but you should try an album also to be sure to have the same protocol
22:17:05AlexPfml: I'm not sure what I prefer
22:17:14gevaertsshould I skip through it to touch all tracks, or can I just stop soon?
22:17:18AlexPAlso, catalog should be catalogue in UK English
22:18:30pamaurygevaerts: no just play and stop. Thinking about it, to stop it I just pushed the power button. Not sure it's important also. No need to skip tracks
22:19:16pamaurygevaerts: I'm still not sure about the reason of this "bug" but I have a theory
22:22:36pamauryah ! I'm happy with it
22:22:44*pamaury is happy to see a bug :)
22:23:02gevaertsgood. Can I go back to my jigsaw puzzle now? :)
22:23:35fmlAlexP: "Catalog" in the manual is correct as it's what I see on the player :-) I.e. it should be corrected in both places.
22:24:16pamaurygevaerts: yes
22:24:20pamaurygood luck
22:26:47 Quit ajb` (Ping timeout: 264 seconds)
22:27:27fmlI there's no punctuation after the item then we get a readable sentence since the sequel is in the "does" from. E.g. "Delete Deletes the currently selected file."
22:30:46fmlAnd "Open with" should be capitalized (like on target)
22:31:43 Join tarbo [0] (~me@unaffiliated/tarbo)
22:33:56 Quit Zagor (Quit: Clint excited)
22:38:04 Quit tarbo (Ping timeout: 240 seconds)
22:47:50 Quit fml (Quit: CGI:IRC)
22:48:59 Quit _zic (Ping timeout: 264 seconds)
22:49:39 Join JdGordon_ [0] (~836b004a@rockbox/developer/JdGordon)
22:54:15 Join tarbo [0] (~me@unaffiliated/tarbo)
22:59:19 Quit Bullet` (Quit: Bye les gens =))
23:01:07pamaurySlasheri: you was the one who wrote dircache isn't it ?
23:01:32 Quit tarbo (Ping timeout: 252 seconds)
23:05:10CIA-8New commit by bluebrother (r24442): OSX: Add CFBundleName to display a nicer name in the menu bar. Replace deprecated CFBundleGetInfoString.
23:10:30 Join robin0800 [0] (
23:12:17 Quit JdGordon_ (Ping timeout: 248 seconds)
23:14:46 Join tarbo [0] (~me@unaffiliated/tarbo)
23:15:27 Join einhirn [0] (
23:18:31 Join JdGordon_ [0] (~836b004d@gateway/web/freenode/x-wfeiprbmsybudvgw)
23:18:52 Nick fxb__ is now known as fxb (
23:18:56B4gderpamaury: yes, he wrote it
23:20:29pamaurydid he also wrote tagcache
23:21:08moosslt, yeah he also did
23:21:28moosand he did wrote the software playback enginne at start
23:22:22 Quit tarbo (Ping timeout: 252 seconds)
23:23:30moospamaury: I already say you this yesterday...already forgot? :)
23:24:04pamauryI was unsure, and it was a way to check if he was here or not ;)
23:25:31pamauryI need to talk with him about another dircache bug
23:25:40mooshehe, simply ping him :)
23:26:11moosSlasheri Slasheri Slasheri! :)
23:26:39moosalso know as Flasheri by few here...
23:27:55pamauryDid someone look at "FS #10954 - UI Simulator fails to rename files correctly". The patch seems simple enough to commit it, no ?
23:28:49 Quit bmbl (Quit: Bye!)
23:30:58 Quit evilnick (Quit: Page closed)
23:31:52JdGordon_pamaury: yep, commit it if you want
23:32:06 Join PaulJam [0] (
23:33:10pamauryJdGordon_: did you test it or you trust the fix ?
23:33:45JdGordon_neither :p but any simple fix is good
23:33:56JdGordon_im at work so cant look at it more than which file it touches
23:34:07***Saving seen data "./dancer.seen"
23:34:54mooshaha JdGordon faithful to yourself
23:35:24pamauryJdGordon_: anyway, if rename doesn't work with the simulator, that patch can't be worse
23:36:27 Join tarbo [0] (~me@unaffiliated/tarbo)
23:43:27 Quit tarbo (Ping timeout: 258 seconds)
23:45:10 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
23:45:29CIA-8New commit by pamaury (r24443): FS #10954: fix rename under simulator by Alex Bennee
23:46:09 Quit samlii (Quit: leaving)
23:46:57amiconnMeh, cache aliasing :\
23:51:13 Quit JdGordon_ (Ping timeout: 248 seconds)
23:52:03 Join JdGordon_ [0] (~836b004a@gateway/web/freenode/x-hunylorkuhbvgdsr)
23:54:18 Join akur [0] (
23:55:15 Quit GeekShadow (Quit: The cake is a lie !)
23:55:24 Quit jgarvey (Quit: Leaving)
23:57:53 Join tarbo [0] (~me@unaffiliated/tarbo)

Previous day | Next day