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 2012-12-26

00:12:46 Quit kevku (Ping timeout: 264 seconds)
00:13:46***Saving seen data "./dancer.seen"
00:17:41 Join eckoit [0] (~ryan@
00:18:21 Quit eckoit (Client Quit)
00:22:05 Quit zaphee (Remote host closed the connection)
00:35:06 Quit Wardo (Read error: Connection reset by peer)
00:38:50 Quit melmothX (Quit: bau)
00:46:04 Join thomasjfox [0] (~thomasjfo@rockbox/developer/thomasjfox)
00:53:31 Quit Rower85 (Quit: Hmmm...)
01:01:46 Quit shamus (Read error: Connection reset by peer)
01:02:29 Join shamus [0] (
01:05:19 Quit bertrik (Read error: Connection reset by peer)
01:48:46 Quit shamus (Read error: Connection reset by peer)
01:49:41 Join shamus [0] (
02:05:19 Quit thomasjfox (Quit: Konversation terminated!)
02:07:37 Join saratoga [0] (47e22765@gateway/web/freenode/ip.
02:12:40saratogawith a fresh source checkout i can't build the codec lib test program on ubuntu
02:13:44saratogamake: *** No rule to make target `/rockbox/codecs/pcm_record.h', needed by `/rockbox/codecs/lib/rbcodec/test/warble.o'.
02:13:47***Saving seen data "./dancer.seen"
02:13:50saratogaany ideas?
02:14:23[Saint]If you figure out the dependencies for the ypr0 toolchain for 64bit Ubuntu, gimme a yell?
02:14:45*[Saint] *cannot* build the bastard thing.
02:17:40gevaertssaratoga: which target are you configuring for?
02:17:52saratogagevaerts: i tried a few
02:18:06gevaertsIt works for me for the sdl app one
02:18:24saratogaif they have recording, i get that error, if they do not, i get a conflict on "commit_dcache" with target-hosted.h
02:18:31saratogalet me try that
02:19:03 Join sakax [0] (
02:19:07saratogaah that seems to work
02:19:23saratogadid we always have to select SDL for warble? i thought i didn't used to do that
02:19:35saratogaalthough i guess it makes sense
02:21:25gevaertsNot sure. I seem to remember that not all targets work
02:30:34saratogaare there any sites with higher bitrate opus files handy?
02:32:51 Join solarcloudpng [0] (~solarclou@
02:33:22[Saint]transcode that sheeeeeeet, biatch.
02:34:02*[Saint] refuses to believe saratoga has no lossless files lying around. ;)
02:41:26 Quit ender` (Quit: Sin is an imaginary disease invented to sell you an imaginary cure.)
02:48:04 Quit shamus (Read error: Connection reset by peer)
02:49:04 Join shamus [0] (
02:49:19pamaurythere a many build errors, something about libtlsf
02:52:14[Saint]pamaury: for a particular target, for all, that a response to my ypr0 toolchain comment earlier?
02:56:34pamauryjust a remark, I committed a few things and noticed that are a few reds but they seems irregular, perhaps a buildbot problem
02:57:08pamaurysee by yourself:
02:57:37*[Saint] wonders why 'saint' isn't connected.
02:58:17 Quit shamus (Read error: Connection reset by peer)
02:58:36 Join shamus [0] (
03:02:14 Quit pamaury (Ping timeout: 252 seconds)
03:22:31 Quit sentriz (Quit: Leaving)
03:22:39 Quit nateloaf (Quit: Leaving.)
03:22:56 Quit prof_wolfff (Remote host closed the connection)
03:28:17 Join Belzebub_ [0] (~torrentow@
03:29:11 Quit Belzebub (Ping timeout: 265 seconds)
03:29:21 Quit shamus (Read error: Connection reset by peer)
03:29:37 Join shamus [0] (
03:30:09 Quit solarcloudpng (Ping timeout: 255 seconds)
03:30:15 Quit solarcloud (Ping timeout: 244 seconds)
03:31:59 Join Strife89 [0] (~Strife89@
03:32:34 Join tuxpaula [0] (~solarclou@
03:34:01 Quit brkfstsqd (Ping timeout: 252 seconds)
03:42:46 Quit Strife89 (Ping timeout: 264 seconds)
03:46:16 Join brkfstsqd [0] (
04:05:22 Join TheSphinX_ [0] (
04:07:53 Quit TheSphinX^ (Ping timeout: 252 seconds)
04:13:48***Saving seen data "./dancer.seen"
04:40:03 Quit SuperBrainAK (Quit: pbly going to sleep /_\)
04:58:08 Join amiconn_ [0] (amiconn@rockbox/developer/amiconn)
04:58:08 Quit amiconn (Disconnected by services)
04:58:08 Quit pixelma (Disconnected by services)
04:58:09 Join pixelma_ [0] (pixelma@rockbox/staff/pixelma)
04:58:11 Nick pixelma_ is now known as pixelma (pixelma@rockbox/staff/pixelma)
04:58:13 Nick amiconn_ is now known as amiconn (amiconn@rockbox/developer/amiconn)
05:06:20 Quit TheSeven (Disconnected by services)
05:06:30 Join [7] [0] (~quassel@rockbox/developer/TheSeven)
05:19:42saratogaC_MULC in libopus would probably be a lot faster in armv5 since we could use packed multiply-shift operations instead of loading 16 bits and masking off values
05:20:50*[Saint] thinks you codec optimization guys are nutbags ;)
05:21:05[Saint]Looking into some of the "simple" codecs has often made me want to cry.
05:21:14[Saint]Braver men than I.
05:26:34saratogacodecs are just sequences of for loops operating on arrays
05:26:47saratogathey're easy to optimize once you realize that
05:27:12[Saint]its the "must. squeeze. every. last. bit." that I can't "see".
05:27:19[Saint]I don't have the foresight to do so.
05:28:28saratogayou mean every last cycle?
05:29:03[Saint]Yes, indeed. I realised after pressing enter that using "bit" was rather non-obvious.
05:29:20saratogaok wanted to make sure you weren't talking about encoders
05:29:30*[Saint] nods
05:29:52saratogafor decoders its actually not so complicated, most loops are fairly simple, and to optimize them you just try to make as many cycles as possible be spent doing a multiplication-add operation
05:30:20saratogato do that you look at the source and destination arrays and try to figure out how to set them up logically so the multiplier is never waiting on work
05:30:25saratogawhen you do that you're pretty much done
05:31:05[Saint]It always amuses me when I see such "It really isn't so complicated..." responses. As I'm sure, to you, it isn't. :)
05:31:06saratogaand usually thats what gcc has the most trouble with since it doesn't know quite as much about the organization of stuff in memory as someone who reads the code
05:31:31[Saint]But then, there's people much smarter than I that barf on the skin language we have, so, at least I understand something well :P
05:31:57 Join TheSphinX^ [0] (
05:32:00saratogawell its a finite series of steps that becomes fairly obvious how it should go once you look at it step by step
05:32:12saratogathe end result is complicated but the steps are small and individually simple
05:32:37saratogaload val1, load val2, multiply val1*val2, save product into location 3
05:33:35[Saint]I suppose I don't think like that because I've never really worked on anything that needed me to eek out every last bit of performance.
05:34:44saratogai guess its a lot more obvious once you know assembly :)
05:34:59 Quit TheSphinX_ (Ping timeout: 260 seconds)
05:36:16[Saint]I have somewhat of a "good enough" attitude to some things. In my mind, 101% realtime on the supported targets is where I'd stop.
05:37:05 Join Rower85 [0] (
05:37:08saratogai just want every format to have good battery life
05:37:17*[Saint] nods
05:40:07derfsaratoga: I can make some higher-bitrate Opus files if you need them.
05:42:11saratogaderf: no i found one
05:44:02saratogaderf: i think i'm going to take a look at the opus fft
05:45:09derfLet me know what I can do to help.
05:45:52saratogado you know much about the celt part of the codec?
05:47:43saratogai profiled a couple test files and they're almost all short-ish block lengths
05:48:06saratogathe N=60 FFT (which i guess is N=120 MDCT) is about 85% of my test sample blocks, is that normal?
05:48:18saratogafor vorbis and wma is the opposite with such short blocks being very rare
05:48:26derfYeah, that is in fact not normal.
05:48:47derfWere all your test samples Fatboy or something?
05:48:58saratogai'm working on a termal so i didn't actually listen to them
05:49:07saratogado you have a more representative sample?
05:54:02derf is 0.64% short blocks.
05:54:49saratogaah much better
05:54:57saratogaso this one is mostly 960 point mdcts
05:55:20saratogai've seen intermediate values as well, but only in some files, how does the encoder decide what block sizes it wants to use?
05:55:59derfSee transient_analysis() in celt/celt_encoder.c
05:58:42derfBasically, it'll switch between 960 and 120 if you use the default frame size (20 ms), based on the analysis done in that function.
06:00:21saratogaderf: are there always just two sizes? i thought i saw 120 length frames as well
06:01:10saratogasorry 120 length FFTs, so 240 length frames
06:03:46derfThere is some stuff in the exp_analysis branch that does real variable-frame-size stuff like Vorbis.
06:04:31derfWhich tries to align the frame boundaries with the transients, instead of just making each whole 20 ms frame short blocks or long blocks.
06:04:51derfI think opusenc will also use a different frame size for the very last frame in the file sometimes.
06:05:09derf(i.e., to reduce the amount of padding required)
06:05:59saratogaso in practice just ignoring the odd sizes for optimization is ok?
06:06:23derfIgnoring 120 and 240 you mean?
06:07:20derfI'm not sure what optimizations you're going to do that wouldn't apply to those if they also apply to 60 and 480.
06:07:43derfBut yes, I'd expect 60 and 480 to be the most important sizes.
06:07:46saratogaderf: picking which butterflies to really work on
06:08:48derf60 is (3,4,5) and 480 is (2,3,4,4,5).
06:08:48saratogaalso maybe splitting the 4 radix thing for N=480 into a special case, since the loops and strides could be hard coded and probably more extensively optimized that way
06:09:03derf120 will be (2,3,4,5) and 240 will be (3,4,4,5).
06:09:08saratogai saw
06:09:29derfDoesn't 480 have two radix-4 loops?
06:09:29saratogareordering the triddle factors for specific window sizes (i think thats what they call them?) may also be worthwhile
06:10:10derfPersonally I thought you could save a lot in the radix-2 step, since it's always done at one end (I forget if it's always first or always last).
06:10:41derfAnd maybe move the muls into the normalization step (I'd have to check if that causes problems for fixed-point, though).
06:11:22derfBecause it's just scaling everything by sqrt(0.5).
06:12:55saratogawhich is the normalization step?
06:13:52***Saving seen data "./dancer.seen"
06:14:07derf#ifndef FIXED_POINT fout[st->bitrev[i]].r *= st->scale; fout[st->bitrev[i]].i *= st->scale;
06:14:11 Quit soap (Ping timeout: 245 seconds)
06:14:20derfI guess that's fft-only, not ifft.
06:15:51saratogaah ok
06:16:01derf(and also float-only)
06:16:04saratogalooking at your file, the 2 butterfly is the smallest contribution
06:16:40derfWell, it is the simplest one to start with.
06:17:08derfIt's one C_MUL, one C_SUB, and one C_ADDTO.
06:17:56saratogaC_MUL looks like a good candidate for optimization on newer arm, since the 16 bit mul instructions should just work on it i think
06:18:24derfIs newer ARM your target here?
06:18:32saratogai want to do both
06:18:42saratogausually i start with armv4 since its much easier
06:18:50saratogathen i try and do armv5 if i'm not sick of it
06:18:56derfFewer instructions to choose from :).
06:19:22saratogathe packed instructions in armv5 are way faster if you can use them, but the whole thinking about alignment thing messes with my head
06:19:45derfI know the Android folks would be interested in NEON opts, but that is probably the least interesting to you guys.
06:19:51derfARMv6 you mean?
06:19:54saratogabut all the recent rockbox targets are at least armv5
06:20:05saratogav6 has some more, but they're not so useful
06:20:18saratogai think they're more for video stuff
06:20:38saratogathe 32x16>>16 ops are all v5
06:21:01derfYeah, I guess v5 got most of the multiplies already.
06:21:14derfv6 is far more useful for video coding.
06:21:27derfWhereas v5 gives you almost nothing useful.
06:21:30saratogaits for 8 bit source values right?
06:21:40derfSo I tend to forget that v5 already got some packed instructions.
06:21:43saratogai looked years ago and lost interest
06:22:26derfNot just 8-bit values, but also the packed adds, packed subtracts, packed saturation, etc.
06:22:35saratogado the android people want NEON? i just assumed audio was so fast it wasn't an issue
06:23:02derfWell, they've spent a lot of time writing NEON for iSAC.
06:23:11derfSo clearly they still care.
06:23:36derfOf course, they stopped asking quite so loudly when I told them to use the fixed-point version instead of the float one.
06:23:50saratogai want to learn neon
06:23:51derfWhich gave them something like a 67% runtime reduction.
06:23:58derfNEON is pretty easy.
06:24:01saratogait looks like it'd be easy
06:24:04derfMakes MMX/SSE look like toys.
06:24:27derfA gekk of a lot easier than ARMv6, too, for that matter.
06:24:38saratogabut i dislike newer ARM processors since they're complicated enough that i can't clearly understand instruction scheduling on them
06:25:07derfI don't really understand beyond an A8.
06:25:16derfBut they gimped the NEON unit in the A9 anyway.
06:25:19saratogayou understand the A8?
06:25:26derfThe A8's not so bad.
06:25:32derfIt's still in-order.
06:25:42saratogai spent like 30 minutes reading the neon table and couldn't understand any of it
06:25:49saratogainstruction timing
06:26:07saratogai don't think i ever figured out if neon Add/mul were pipelined or not
06:26:13derfThey are.
06:26:45saratogaeven for 128 bit mul ops?
06:27:03saratogathat is, can it sustain 4x32 bit muls per cycle
06:28:25 Join nateloaf [0] (
06:29:21derfI think the A9 can issue one every two cycles.
06:29:40derfI don't remember on the A8, and I can't find my copy of the TRM for it.
06:30:01saratogathe arm11 had such a clear cycle timing diagram
06:30:25saratogacomplete with examples and everything
06:30:32derfOh, there it is.
06:33:28 Join |akaWolf| [0] (~akaWolf@
06:34:04saratogaso when they say "16.6.2. Advanced SIMD integer ALU instructions" and "cycles: 1" thats the pipeline stall before you can use the ALU again, not the latency right?
06:34:43derfThe pipeline stages are in the columns after that.
06:35:57derfI.e., VADD Qd,Qn,Qm requires Qn and Qm in stage N2, and returns the result in stage N3.
06:36:41derf(meaning Qd will be available in stage N2 for an instruction that issues the following cycle)
06:38:30derfIt can also dual-issue load/stores (as well as permute instructions, though that's probably not something you care about a lot).
06:38:47derfI.e., one load/store and one arithmetic operation can issues in the same cycle.
06:39:03saratogawhat does the N in N2 refer to ?
06:39:26derfN is for the NEON pipeline.
06:39:41saratogaah yes
06:39:42derf(which is separate from the normal ARM pipeline)
06:39:50saratogai vaguely remember that from the diagram
06:40:34derfIf either of the instructions is multi-cycle, then the last cycle of the first instruction can still be shared with the first cycle of the second instruction.
06:40:55derf(all this dual-issue stuff goes away in the A9, though)
06:43:14derfLooks like 4x32 VMUL is 4 cycles on an A8.
06:43:28saratogabefore you can issue another one?
06:43:47saratogaisn't that just the latency?
06:44:02derfLatency is 6 cycles.
06:44:27derfWell, actually more.
06:44:30saratogaso the entire 4 cycles front he source read to the dest read can't issue another?
06:44:53 Join eckoit [0] (~ryan@
06:44:57derfQdLo is available in N6 of the second cycle (so it would be available for N1 7 cycles after the VMUL starts issuing)
06:45:11saratogawhat does the "Cycles:1" mean then?
06:45:44derfQdHi is available in N6 of the 4th cycle, so it would be available 9 cycles after the VMUL starts issuing.
06:45:58derfsaratoga: Not sure what you mean.
06:46:28derfYou're looking at the A8 TRM, right?
06:46:36derfTable 16-19?
06:46:55saratogait says 1 under cycles, then N2 for load and N6 for write
06:47:08saratogaor rather reg read, reg write
06:47:18derfWaiting for that page to load...
06:47:24derf(internet here is pretty crappy)
06:47:54saratogathats just an HTML version of Table 16-19?
06:48:02saratogaTable 16-19 (no question mark intended)
06:48:23derfWell, your link too me to 16.1.
06:49:01saratogai think the html frames on that page make pasting weird
06:49:08derfRight, so...
06:49:15derfSee under "Register format"?
06:49:31derfThe first part of that table is for the Dd,Dn,Dm form.
06:49:40derfAnd for .8 and .16 data sizes only.
06:49:55saratogai guess what i never understood is what the "cycles" column is supposed to mean
06:49:58derfE.g., 8x8 or 4x16 muls.
06:50:09derfIt's showing what happens on each cycle.
06:50:29 Quit [Saint] (Remote host closed the connection)
06:50:31saratoga"Assuming no data hazards, the instruction takes a minimum of one cycle to execute as indicated by the value in the Cycles column."
06:50:52saratogadoes that just mean decode takes 1 cycle and you can issue another (non multiply) instruction on the next cycle?
06:50:55derfYeah, that is leaving out quite a bit.
06:51:22saratogai guess that would be something like the 'cycles to issue'
06:51:47derfFor a VMUL.[S|U][8|16] Dd,Dn,Dm, that is correct.
06:52:22derfFor a VMUL.[S|U][8|16] Qd,Qn,Qm, it takes 2 cycles before you can issue another arithmetic instruction.
06:52:51derfDouble those counts for the VMUL[S|U]32 versions.
06:52:57saratogabut you can't issue another multiply instruction for 4 (or 5) cycles?
06:53:01derf(i.e., 2 and 4 cycles, respectively)
06:53:21derfNo, you can still issues more multiplies.
06:53:41derfBut they need to have their arguments available in the indicated pipeline stages.
06:54:01saratogaahhh ok
06:54:26saratogaso the throughput for independent multiply instructions is how many bits per cycle?
06:54:42saratogafor VMUL.[S|U][8|16] Dd,Dn,Dm that is
06:55:06saratogathats vaguely what i remembered
06:55:10derfThe D[n] registers are 64 bits.
06:55:19saratogaoh of course
06:55:26saratogafor int32 is it still 64?
06:56:00derfThat's the 3rd and 4th sections of that table "(.32 normal)"
06:56:05 Join [Saint] [0] (~saint@rockbox/user/saint)
06:56:53derf.32 is the data size... "normal" just means it's not a "widening" or "long" multiply.
06:57:00*[Saint] wonders why a Rockbox'ed N2G would make the boot process of a PC *totally barf*...
06:57:34saratogawait how do i read the cycles column under the 4th section?
06:57:35[Saint]I even cheto see that the BIOS wasn't trying to boot from it, by disabling booting from all removable devices, dice.
06:57:47saratogais it saying that on cycle 1, X happens, then on cycle 2, Y?
06:57:49[Saint]*s/cheto/checked to/
06:57:53saratogaand so on to cycle 4?
06:58:39 Quit [Saint] (Remote host closed the connection)
06:58:43derfsaratoga: Yes, where "X" happens in a specific pipeline stage.
06:59:10derfI.e., each cycle has its own set of pipeline stages associated with it.
06:59:10saratogaok now i understand
07:00:44 Join [Saint] [0] (~saint@rockbox/user/saint)
07:02:28[Saint]Hmmmmm, same result.
07:02:40derfOne last bit that's not in the table: multiply-accumulator forwarding.
07:02:41derfSee the footnote at the very bottom of that page.
07:02:50[Saint]The Ubuntu boot process hates Rockbox?
07:03:08[Saint]Neither my laptop, nor desktop, will boot with a Rockbox device connected.
07:03:33[Saint]I checked an N2g, a mini2G, and a Classic.
07:03:53derf[Saint]: I have that same problem, but it doesn't even get past the BIOS.
07:04:01[Saint]All were "charge only" at the time I power cycled the machine.
07:04:21derfNot sure if it's the Rockbox device specifically. I just unplugged all my USB devices to get past it.
07:04:24[Saint]I didn't want them mounted, as there's some data I care about on the latter two.
07:04:43[Saint]derf: well, I have several other USB devices attached.
07:04:51[Saint]I can only reproduce this with Rockbox.
07:05:06[Saint]Ohhhh...interesting test. Lets try the Apple OF.
07:05:18derfIt didn't occur to me to narrow it down to a specific one.
07:05:27 Quit [Saint] (Remote host closed the connection)
07:07:24 Join [Saint] [0] (~saint@rockbox/user/saint)
07:07:30derfsaratoga: Also, this may be of some help:
07:07:48[Saint] seems it's doing it.
07:07:55*[Saint] a word
07:07:56derfThat's the summary of "WTF do these instructions actually do?" that I wanted but could not find in ARM's own documentation.
07:08:02[Saint] seems it's Rockbox doing it.
07:08:03derfSo I just wrote my own.
07:08:31[Saint]Under the Apple OF, the machines boot fine.
07:08:42[Saint]What could Rockbox possibly be messing up here?
07:08:57[Saint]Or, interfering with.
07:09:24*[Saint] notes that he has disabled booting from removable devices in the BIOS of both machines he is testing this on.
07:10:05[Saint]I suspected a few times that a connected Rb device messed with booting, but never looked into it more than "Huh? ...that was weird".
07:11:11[Saint]I can get no meaningful debug info from either machine.
07:22:52 Quit eckoit (Quit: eckoit)
07:36:25 Quit shamus (Read error: Connection reset by peer)
07:37:18 Join shamus [0] (
08:05:17 Quit saratoga (Quit: Page closed)
08:12:44[Saint]Well, I can consistently reproduce this reboot weirdness with a connected Rockbox device.
08:13:13[Saint]No idea why this happens, but, yeah.
08:13:54***Saving seen data "./dancer.seen"
08:14:36ParkerR_Umm could it just possibly be that the device pulls too much power upon insertion?
08:15:07ParkerR_Ohhh, I thought this was in another channel. I am so sorry
08:19:54 Quit brkfstsqd (Remote host closed the connection)
08:23:02 Join soap [0] (~soap@rockbox/staff/soap)
08:23:38 Quit shamus (Read error: Connection reset by peer)
08:23:49 Join shamus [0] (
08:30:50 Join mortalis [0] (
08:42:15 Join kevku [0] (x@2001:470:dcc6:0:69:69:69:69)
09:06:06 Join stoffel [0] (
09:08:24[Saint]I suspected it might be the case, but I checked anyway, and it doesn't happen with a 'charge only' cable.
09:08:54[Saint](actually a standard Apple cable with the data lines cut)
09:09:23[Saint]Nor does it happen with a 'real' cable when booted up under the Apple OF.
09:09:34*[Saint] shrugs
09:19:39 Quit DonAman (Ping timeout: 260 seconds)
09:27:49 Join DonAman [0] (
09:28:12 Quit DonAman (Changing host)
09:28:12 Join DonAman [0] (~DocHollid@unaffiliated/phifedoc)
09:48:41 Join TheSphinX_ [0] (
09:52:12 Quit TheSphinX^ (Ping timeout: 244 seconds)
09:53:19 Join melmothX [0] (~melmoth@unaffiliated/melmothx)
09:59:14 Quit shamus (Read error: Connection reset by peer)
09:59:25 Join shamus [0] (
10:01:04 Quit stoffel (Remote host closed the connection)
10:13:55***Saving seen data "./dancer.seen"
10:24:10 Quit melmothX (Ping timeout: 255 seconds)
10:24:37 Quit [Saint] (Quit: Quitty Quit Quit McQuitson & Co. Ltd.)
10:25:53 Join melmothX [0] (~melmoth@unaffiliated/melmothx)
11:02:44 Nick Belzebub_ is now known as Belzebub (~torrentow@
11:03:01 Quit Belzebub (Changing host)
11:03:01 Join Belzebub [0] (~torrentow@unaffiliated/blown-engine)
11:18:17 Join ender` [0] (
11:25:05 Quit funman (Ping timeout: 244 seconds)
11:27:03 Join funman [0] (~fun@rockbox/developer/funman)
11:27:52 Join [Saint] [0] (~saint@rockbox/user/saint)
11:33:14 Quit bluebrother (Disconnected by services)
11:33:19 Join bluebrother^ [0] (~dom@rockbox/developer/bluebrother)
11:34:03[Saint]scorche`: or any other forum type dude, this "person" is quite highly suspicious -,33874.msg216204.html#msg216204
11:34:06 Join funman_ [0] (~fun@rockbox/developer/funman)
11:34:27[Saint]necro-posting on a thread from March, with the exact wording of a section of the release notes.
11:35:11[Saint](user has a single post, a verbatim section of the 3.11.2 release notes)
11:35:49 Join brianmilw [0] (
11:36:00 Quit fs-bluebot (Ping timeout: 256 seconds)
11:36:25[Saint]Errrr, sorry. Google lied to my eyes.
11:36:48[Saint]It's a verbatim section of gevaerts' release email
11:37:07 Quit funman (Ping timeout: 252 seconds)
11:37:21 Join fs-bluebot [0] (
11:37:47[Saint]No ICQ address though ;)
11:48:12 Join lorenzo92 [0] (
11:48:58brianmilwquick and simple question, I have been searching and my eyes and brain hurt. I share a ipod with the wife; older ipod video 5 and need a bootloader with a selection screen on boot instead of using key combinations. I plan to dual boot between the stock and rockbox kernel. what would be the quickest approach, idea bootloader? I keep finding outdated information from 2007 and am half asleep. Can I simply navigate to a
11:48:58brianmilwsetting in rockbox or modify a config file for the currently loaded rockbox bootloader installed from the rbutil.
11:50:09brianmilwwould the ipl bootloaded modified for just the apple and rockbox kernel be my best solution?
11:53:53brianmilwanyone please? i am multitasking with 4 different things right now, a quick lead in the right direction is all i need, i can research from there.
12:01:34*brianmilw had too many beers
12:01:40 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
12:02:14 Quit brianmilw (Quit: sleep)
12:06:35 Quit nateloaf (Quit: Leaving.)
12:13:58***Saving seen data "./dancer.seen"
12:27:43 Join ack [0] (
12:33:10pamaurybluebrother^: ping
12:33:13 Quit tuxpaula (Ping timeout: 256 seconds)
12:35:21 Join Topy44|2 [0] (
12:35:21 Quit Topy44 (Read error: Connection reset by peer)
12:36:20pamaurybluebrother^: I have some questions regarding bootloader install in RBUtil. First, what is the meaning of the different options in Capability (IsFile ? IsRaw ?). Second, it seems that CanCheckVersion allows RButil to compare versions but this seems unimplemented, am I right ?
12:49:45 Quit ender` (Quit: You know you created god in your image when he hates the same people you do.)
12:58:12 Join ender` [0] (
13:15:03[Saint]I seem to recall looking at that at a point.
13:15:44[Saint]If I'm not mistaken, I think it is/was/could be to allow RbUtil to serve release client builds.
13:28:20 Quit lorenzo92 (Remote host closed the connection)
13:57:46 Quit [7] (Remote host closed the connection)
14:06:25 Quit shamus (Read error: Connection reset by peer)
14:06:50 Join shamus [0] (
14:08:39 Join karma [0] (
14:13:59***Saving seen data "./dancer.seen"
14:17:40 Join amayer_ [0] (
14:18:58gevaerts[Saint]: HID, maybe?
14:20:34gevaertsFully? Charging mode normally sets up a HID device
14:21:17gevaertsOn the nano2g it is, IIRC
14:21:19[Saint]Well, the N2G at least has HIS disabled I think, doesn't it? My usual config.cfg I dump onto my players disables it anyway.
14:21:31 Join Wardo [0] (
14:21:41gevaertsCan you check using lsusb -v -d <ids>?
14:21:42*[Saint] word failed. What happened there?
14:24:07[Saint]I have an N2G beside me now, it doesn't show at all in lsusb
14:24:33gevaertsdoes dmesg complain about it?
14:24:35 Quit sakax (Ping timeout: 245 seconds)
14:24:40[Saint]doing the whole "press a button to disable enumeration" thing, to stop it panicing and dying.
14:24:51[Saint]No, no complaints from dmesg
14:25:33[Saint]I'm at a total loss, but, it *is* reproducible here. ANd only Rockbox seems to trigger it.
14:25:41gevaertsOK, then I don't know what's going on
14:26:12[Saint]In both machines the BIOS is a little lacking, and the error logging therein isn't catching any failures.
14:26:55[Saint]it just...hangs. It doesn't even get to POST
14:37:10 Join sakax [0] (
14:56:28 Quit mortalis (Quit: Leaving)
15:01:49 Quit pamaury (Ping timeout: 265 seconds)
15:02:45 Quit karma (Quit: CGI:IRC (EOF))
15:09:36 Join bertrik [0] (
15:09:36 Quit bertrik (Changing host)
15:09:36 Join bertrik [0] (~quassel@rockbox/developer/bertrik)
15:11:11 Quit nosa-j (Ping timeout: 252 seconds)
15:11:49 Join nosa-j [0] (~m00k@
15:21:07 Join jlbiasini [0] (~metaphysi@
15:40:18 Quit bertrik (Remote host closed the connection)
15:46:50 Quit jlbiasini (Remote host closed the connection)
15:59:31 Quit shamus (Ping timeout: 250 seconds)
16:02:53 Quit Wardo (Ping timeout: 256 seconds)
16:12:05 Join shamus [0] (
16:12:18 Join y4n [0] (~y4n@unaffiliated/y4ndexx)
16:14:01***Saving seen data "./dancer.seen"
16:16:51 Quit shamus (Ping timeout: 250 seconds)
16:17:15 Join shamus [0] (
16:29:08 Join XavierGr [0] (XavierGr@rockbox/staff/XavierGr)
16:30:25 Quit melmothX (Read error: Connection reset by peer)
16:30:27 Join pamaury [0] (
16:30:27 Quit pamaury (Changing host)
16:30:27 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
16:50:53 Join eckoit [0] (~ryan@
16:55:50 Quit shamus (Read error: Connection reset by peer)
16:56:08 Join shamus [0] (
17:12:51 Join melmothX [0] (~melmoth@unaffiliated/melmothx)
17:15:56[Saint]Hmmmm, even with USB disabled, the N2G is locking up occasionally when connected.
17:16:23[Saint]it's a full "playback stalls, hard reset is the only out" lockup.
17:21:43 Join SuperBrainAK [0] (
17:25:16 Join jlbiasini [0] (~metaphysi@
17:25:45jlbiasinipamaury: ping?
17:25:54pamauryjlbiasini: pong
17:26:03pamauryhow are you ? it has been a long time
17:26:15jlbiasinijep a long time !!
17:27:42jlbiasiniI'm still very busy I live in romania now I work as a volunteer for an NGO on their CRM software... I was wondering if you would push my few patches
17:28:34pamaurysure, can you recall me which ones, I've been so busy recently, just coding a few hours here and there
17:28:46jlbiasini g#352
17:28:49fs-bluebotGerrit review #352 at : [FUZE+] Rockblox plugin slight keymaps update by Jean-Louis Biasini (changes/52/352/3)
17:29:06jlbiasini g#353
17:29:08fs-bluebotGerrit review #353 at : [Fuze+] PictureFlow: complete keys mapping by Jean-Louis Biasini (changes/53/353/1)
17:29:22jlbiasini g#354
17:29:23fs-bluebotGerrit review #354 at : [Fuze+] imageviewer: update/complete keymaps by Jean-Louis Biasini (changes/54/354/2)
17:29:54jlbiasinithose are pretty safe it's only keymaps stuff
17:31:01pamaurydone ! thanks
17:31:10jlbiasiniactually I figured out that there is still one game plugin that would need the same treatment as rockblox but I can't recall wich one...
17:31:12pamauryif you have some time, you could test the new frequency scaling code
17:31:30jlbiasiniJep I just installed it on my device
17:31:46[Saint]Hum, just checking on a stashed I need to comment the entire block of USB defines, or, is disabling HAVE_ROCKB)X_USB enough to wipe out everything depending on it?
17:32:25jlbiasinipamaury: a propos does that suppose some bootloader recompilation?
17:33:51jlbiasiniI still have a very strange behaviour on with my laptop: if the device is connected when I power it up it blocks the booting process
17:34:13pamauryyeah, lots of people have reported this and not only with the fuze+
17:34:21[Saint]I'm assuming, for the sake of cleanliness, that I should remove/comment the entire block of USB defines, but for my purposes, disabling only HAVE_ROCKBOX_USB "just works".
17:34:32pamaurywhen I'm back home, maybe I should have a look at this with the usb analyzer
17:34:40pamauryperhaps the stack doesn't answer to some request
17:34:46[Saint]jlbiasini: interestingly, I made a long rant about this earlier :)
17:34:52jlbiasiniAh ok! I thought I was alone with it...
17:35:06[Saint]it happens even with USB disabled (for me at least), so...I have no idea.
17:35:59gevaerts[Saint]: HAVE_ROCKBOX_USB?
17:36:18[Saint]shit, sorry s/HAVE/USE/
17:36:18jlbiasinithe funniest is that if one disconnect and reconnect the device a few second later, the booting process goes on and then get stuck again...
17:36:41[Saint]jlbiasini: ah, I didn't experiment with that.
17:36:47gevaertsRight. IIRC, disabling USE_ROCKBOX_USB does *not* disable the USB stack
17:37:07[Saint]Does you machine even get to POST? <−− jlbiasini
17:37:16[Saint]My laptop/desktop don't.
17:37:36[Saint]gevaerts: so, the "clean" version is to disable the entire block of USB defines, I assume?
17:38:10gevaertsYou want to remove HAVE_USBSTACK
17:38:57[Saint]but then, aren't a fe other defines irrelevant?
17:39:00gevaertsNot sure if you need to disable more than that. You shouldn't have to, I think, unless there's some missing #ifdef HAVE_USBSTACK somewhere, which would cause the build to fail
17:39:42jlbiasini[Saint]: I don't know what you are talking about :D!
17:39:51[Saint]there's USB_OTG defines, HID, ENDPOINTS, product/vendor ID...
17:39:57jlbiasiniwhat is POST
17:40:01[Saint]jlbiasini: nevermind ;)
17:40:54gevaertsUSE_ROCKBOX_USB says to use our usb storage driver for storage access. CONFIG_USBOTG is IMO misnamed and says which chip is used. HAVE_USBSTACK controls whether the entire USB stack is built at all
17:41:16pamauryjlbiasini: early stage of bios boot
17:41:22gevaertsjlbiasini: "power on self test", i.e. early BIOS boot
17:41:30[Saint]Aha, USBOTG is indeed misleadingly named. I wondered why it was there at all, but, now I know.
17:43:14jlbiasiniWell I guess it does reach bios early stage as it display ASUS logo it's before the possibility to get into bios but I guess it's already bios
17:43:36gevaertsIt's BIOS from power on to the bootloader
17:44:31jlbiasiniwell it's definivily before the grub anyway
17:45:54[Saint]Mine black screens immediately, with no disk access.
17:46:01jlbiasini[Saint]: woaw do you mean that you can't even power on your laptop with a rockbox device connected?
17:46:25[Saint]It powers on, both machines just sit at a black screen, though.
17:46:50[Saint]Normally I'd be at my login in ~3 seconds or so.
17:47:00jlbiasiniWell I suppose it depend when the bios is displaying a logo
17:47:26[Saint]Pffffft! real BIOS' don't display vendor logos ;)
17:47:48[Saint]That's a second of boot time wasted! :D
17:48:00jlbiasini3 seconds from power to login? WOAw do you only have a graphic server!!!
17:49:10jlbiasinii'm on debian + xfce + ssd and I still have 10 second to boot...
17:49:52*[Saint] decides to time his laptop.
17:50:23*gevaerts thinks this is off-topic
17:50:31[Saint]Hum, assuming my reactions are anywhere near accurate...I was a bit off. ~8 seconds.
17:50:34[Saint]and, yes.
17:50:37jlbiasini[Saint]: Actually firm logo were implemented for Rockbox debugging usable :D
17:51:31*[Saint] prods jlbiasini into #rockbox-community
17:53:42jlbiasinijep, sorry
17:54:44 Join prof_wolfff [0] (
17:55:38gevaerts[Saint]: I've just pushed a fix to make nano2g (and probably others) build without *just* HAVE_USBSTACK
17:55:52 Quit jlbiasini (Quit: jlbiasini)
17:56:21gevaertsI have no idea if it will use 500mA for charging though
17:57:58[Saint]I'm not sure I understand the association between *CHARGING_ENABLE and *USB_STACK
17:58:40gevaertsHAVE_USB_CHARGING_ENABLE is the one that makes the USB stack tell the charging code if it's allowed to use 100mA or 500mA
17:59:13gevaertsNo, I think I just broke h300 :\
18:00:13gevaertsHAVE_USB_CHARGING_ENABLE is the one that makes the USB code tell the charging code if it's allowed to use 100mA or 500mA. That code can either be the usb stack or something else for HW usb...
18:01:05 Join jlbiasini [0] (~metaphysi@
18:01:31[Saint]I *do* want to do it/want it done cleanly. But ultimately, the purpose is a quick'n'dirty workaround for a panic on USB insertion.
18:01:42[Saint]I don't want to break anything just for that.
18:02:29[Saint]I think USB not working is fine, panicing in insert, notsomuch. Is leaving the USb stack enabled problematic, or simply wasteful?
18:03:01gevaertsI don't know. I don't have a nano2g :)
18:06:54gevaertsIIRC last time someone checked rebooting to the EDM failed too
18:07:46gevaertsIf leaving HAVE_USBSTACK enabled works, it's the better option, because it allows the 500mA charging thing to work
18:07:48[Saint]That was me, and yeah, your reboot on insert patch failed for me.
18:13:29 Join thomasjfox [0] (~thomasjfo@rockbox/developer/thomasjfox)
18:14:05***Saving seen data "./dancer.seen"
18:16:10 Join TheSeven [0] (~quassel@rockbox/developer/TheSeven)
18:17:31 Quit amayer_ (Read error: Operation timed out)
18:22:59 Join amayer_ [0] (
18:23:03 Quit amayer_ (Remote host closed the connection)
18:23:25 Join amayer_ [0] (
18:25:27 Join melmothX_ [0] (~melmoth@
18:25:38 Quit melmothX (Ping timeout: 260 seconds)
18:26:13 Quit melmothX_ (Client Quit)
18:26:36 Join melmothX [0] (~melmoth@unaffiliated/melmothx)
18:46:13 Quit jlbiasini (Remote host closed the connection)
18:48:46 Join z180 [0] (
18:52:37 Join bertrik [0] (~quassel@rockbox/developer/bertrik)
18:52:51 Quit shamus (Ping timeout: 250 seconds)
18:56:02 Join nateloaf [0] (
18:56:58 Join shamus [0] (
19:04:33 Quit shamus (Ping timeout: 250 seconds)
19:08:40z180do most DAPs have a DSP?
19:17:24 Quit nomada (Remote host closed the connection)
19:22:57 Join shamus [0] (
19:30:52 Join pretty_function [0] (~sigBART@
19:34:33 Join nomada [0] (
19:39:45 Quit z180 (Quit: ChatZilla 0.9.89 [Firefox 3.6.28/20120306064154])
19:42:17 Quit nomada (Remote host closed the connection)
19:45:42 Join nomada [0] (
19:57:48 Join lorenzo92 [0] (
20:03:35lorenzo92pamaury: lcd update is pretty fast, I was wrong before, well I can still see that it is updating from startup (from garbage colors) but I think it is normal
20:03:50lorenzo92pamaury: next step SDRAM, right? I have the timings values
20:04:23pamaurydid you dump clkctrl values to see the cpu speed ? SDRAM can be configured by the bootloader using the sdram command
20:04:31 Quit pretty_function (Remote host closed the connection)
20:04:35pamaurylorenzo92: ^
20:04:57 Quit jhMikeS (Ping timeout: 260 seconds)
20:05:14lorenzo92no I need to do the cpu thing, for timings I read the bootloader log and there are these values (ECRAMXXXX)
20:05:17lorenzo92or similar
20:06:39 Join Strife89-N810 [0] (~user@
20:06:41pamauryI think it's much simpler to let the bootloader setup the sdram
20:06:56pamauryat least for a start
20:11:50lorenzo92pamaury: uu I need to redefine clock control registers, some are different for 36xx
20:14:00 Quit shamus (Read error: Connection reset by peer)
20:14:07***Saving seen data "./dancer.seen"
20:14:23 Join shamus [0] (
20:16:26 Quit nomada (Remote host closed the connection)
20:21:14 Join nomada [0] (
20:22:36lorenzo92pamaury: CPU clock divider is 20
20:34:44 Join user__ [0] (~user@
20:34:57 Nick user__ is now known as Strife89 (~user@
20:35:07 Quit Strife89-N810 (Ping timeout: 250 seconds)
20:35:21 Nick Strife89 is now known as Strife89-N819 (~user@
20:35:25 Nick Strife89-N819 is now known as Strife89-N810 (~user@
20:38:24 Quit nomada (Remote host closed the connection)
20:39:11 Join Strife89 [0] (~Strife89@
20:39:37 Join pretty_function [0] (~sigBART@
20:43:44 Quit y4n (Quit: only amiga makes it possible)
20:47:05 Join nomada [0] (
20:48:22 Part Strife89 ("Leaving")
20:51:12 Quit pretty_function (Remote host closed the connection)
20:58:34 Quit bertrik (Remote host closed the connection)
21:00:50 Nick scorche` is now known as scorche (~scorche@rockbox/administrator/scorche)
21:01:54 Join Strife89 [0] (~Strife89@
21:03:50 Join CookieJar [0] (
21:04:09 Quit eckoit (Quit: eckoit)
21:04:35 Join AConfusedMan [0] ( anyone here?
21:05:08 Nick Strife89-N810 is now known as N810-Strife89 (~user@
21:05:11 Quit CookieJar (Client Quit)
21:05:28[Saint]AConfusedMan: multiple someones.
21:05:39[Saint]Don't ask-to-ask. Just ask.
21:06:03AConfusedManMultiple p[Alright, thanks.
21:06:21AConfusedManBasicall, so I want to rockbox my ipod classic 6g with the beta release.
21:06:39AConfusedManI coudn't find changelogs or release notes anywhere.
21:07:09[Saint]not even on, seems odd.
21:07:13[Saint]You looked, right?
21:07:25AConfusedManLet me find the link.
21:08:14AConfusedManWell...Guess I suck at looking for things.
21:08:44[Saint]What OS does your host system run?
21:08:57AConfusedManStock Apple firmware,
21:09:19[Saint]Not the iPod, your host machine. The machine you'll be installing Rockbox from.
21:09:20AConfusedManLet me find the version.
21:09:27AConfusedManOh; Win 7.
21:09:36 Join Fuzebug [0] (324bd132@gateway/web/freenode/ip.
21:09:45[Saint]then, here:
21:10:02[Saint]Rockbox cannot be installed without the emCORE project, so, start there.
21:10:15[Saint]those instructions will guide you in every step you need to take.
21:10:18AConfusedManI read it's pretty unstable; where do I submit bug reports and such?
21:11:11AConfusedManThanks so much; I've been looking for these things.
21:11:24[Saint]The emCORE link is an interactive step-by-step, sing out if you get stuck.
21:11:57AConfusedManGot it. I'll keep you posted as I go installing.
21:13:45 Quit Rower85 (Quit: Hmmm...)
21:13:56FuzebugCan someone Please fix the keymaps for the text editor on the Fuze +, There isn't a key mapped to the context menu for copying or pasting lines.
21:15:02 Quit |akaWolf| (Ping timeout: 260 seconds)
21:16:01[Saint]kindly open a ticket in the tracker after verifying none exist already.
21:20:55FuzebugThanks I'll check
21:26:05 Quit AConfusedMan (Quit: CGI:IRC (Ping timeout))
21:34:02 Quit Fuzebug (Quit: Page closed)
21:35:35 Quit [Saint] (Remote host closed the connection)
21:37:45 Join [Saint] [0] (~saint@rockbox/user/saint)
21:38:47 Join eckoit [0] (~ryan@
21:39:37 Quit shamus (Read error: Connection reset by peer)
21:39:55 Join shamus [0] (
21:43:22 Quit lorenzo92 (Remote host closed the connection)
21:46:17 Join fuzetester [0] (324bd132@gateway/web/freenode/ip.
21:47:20fuzetesterI created a bug report for the fuze + text editor at:
21:48:44 Quit simonnn (Remote host closed the connection)
21:48:44[Saint]Thanking you.
21:50:22 Quit scorche (Disconnected by services)
21:50:25 Join scorche` [0] (~scorche@rockbox/administrator/scorche)
21:52:17 Join bertrik [0] (~quassel@rockbox/developer/bertrik)
21:52:41 Quit melmothX (Quit: bau)
21:53:28 Quit shamus (Read error: Connection reset by peer)
21:58:09 Join shamus [0] (
22:11:54 Quit nomada (Remote host closed the connection)
22:14:11***Saving seen data "./dancer.seen"
22:14:21 Quit thomasjfox (Ping timeout: 250 seconds)
22:14:38 Quit fuzetester (Quit: Page closed)
22:16:14 Join nomada [0] (
22:17:47 Join jhMikeS [0] (~jethead71@
22:17:47 Quit jhMikeS (Changing host)
22:17:47 Join jhMikeS [0] (~jethead71@rockbox/developer/jhMikeS)
22:20:16 Quit nomada (Remote host closed the connection)
22:23:36 Join nomada [0] (
22:24:53 Quit nomada (Remote host closed the connection)
22:27:05 Quit jhMikeS (Ping timeout: 256 seconds)
22:28:43 Join nomada [0] (
22:29:05 Quit eckoit (Ping timeout: 250 seconds)
22:33:02 Quit nomada (Remote host closed the connection)
22:37:01 Join nomada [0] (
22:57:43 Join jhMikeS [0] (~jethead71@rockbox/developer/jhMikeS)
23:07:49 Join thomasjfox [0] (~thomasjfo@rockbox/developer/thomasjfox)
23:09:45 Join Scromple [0] (~Simon@
23:12:19[Saint]Who's "the wiki guy"?
23:12:27[Saint]Is that you too, scorche`?
23:13:27 Quit DonAman (Quit: The Sleep of Reason brings forth monsters)
23:13:28*[Saint] wonders if wiki member phone numbers can be truncated/redacted/whatever the word is just as email addresses are, for non-members.
23:13:51 Join eckoit [0] (~ryan@
23:14:01[Saint]It occured to me the other day that, yes, I put it there...but, when I did so, I did so assuming it wouldn't be public.
23:14:11[Saint]...but I don't actually care that it is, at all.
23:15:38[Saint]email addresses appear as <address>, or something similar, it'd be kinda cool if the same smarts were applied to the wiki for phone numbers but I couldn't fingure out where I needed to look to do that.
23:15:51[Saint]Or if I even have access to whatever trunk/repo that lies in.
23:16:19[Saint]*figure, too
23:30:49 Quit nomada (Remote host closed the connection)
23:31:23 Quit eckoit (Quit: eckoit)
23:35:38 Join nomada [0] (
23:36:21 Quit amayer_ (Ping timeout: 246 seconds)
23:36:27 Quit nomada (Remote host closed the connection)
23:43:00 Join nomada [0] (
23:46:22 Quit kevku (Ping timeout: 264 seconds)
23:47:28 Join eckoit [0] (~ryan@
23:56:47 Quit shamus (Read error: Connection reset by peer)

Previous day | Next day