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 2013-01-24

00:01:09jlbiasiniOk i'm recompiling but it might take while because I'm also recompiling my kernel
00:01:45jlbiasinithat's a nasty
00:02:27pamaury indeed
00:02:52jlbiasinioh wait I did not #include <sigaction.h>
00:03:14jlbiasinijust asm/signal
00:03:32 Join dynion [0] (
00:03:40pamaurytry sigaction and remove asm/signal
00:05:49 Quit Strife89 (Ping timeout: 245 seconds)
00:08:55 Quit nomada (Ping timeout: 255 seconds)
00:09:17pamauryok, don't mind, if it works with this that's fine, we'll see this tomorrow
00:10:11 Quit dynion (Quit: CGI:IRC)
00:10:13jlbiasinino it doesn't but obvouisly it's just about to find which file to include
00:10:16 Join prof_wolfff [0] (
00:10:31jlbiasiniwell at least it does with the parameter remplace
00:10:51jlbiasiniAnyway I'm going to bed to
00:11:10 Quit jlbiasini (Quit: jlbiasini)
00:12:34 Quit mt_ (Ping timeout: 248 seconds)
00:18:13 Quit Rower (Quit: Hmmm...)
00:18:18 Quit Tim_Elliott (Quit: Leaving)
00:20:15 Join mt [0] (~quassel@
00:20:16 Quit mt (Changing host)
00:20:16 Join mt [0] (~quassel@rockbox/developer/mt)
00:27:52 Quit lebellium (Quit: ChatZilla 0.9.89 [Firefox 19.0/20130116072953])
00:36:50 Join [Saint] [0] (~quassel@
00:37:11 Nick [Saint] is now known as Guest78563 (~quassel@
00:40:04 Quit Edsansaclipplus (Quit: CGI:IRC)
00:42:17[Saint_]I got some totally unexpected results with test_codec and the 10 band eq.
00:42:29[Saint_]I'll put them up on the gerrit page now.
00:43:04[Saint_]long story short, it turns out that when compared to the 5 band eq, the 10 band eq made decoding slightly faster.
00:43:34[Saint_]possibly because it actually needed to boost occasionally? (a classic)...dunno.
00:44:10gevaertsI don't think test_codec output goes through the eq
00:44:58[Saint_]I thought "with dsp" used the eq.
00:45:24gevaertsAh, is there a "with dsp" option?
00:45:33[Saint_]Ok, well...that's entirely disinteresting then ;)
00:45:33*gevaerts isn't sure then
00:46:04[Saint_]there's a few test_codec options, speed test, and speed test with dsp is what I used.
00:46:19[Saint_]you can also restrict boosting iirc.
00:47:10*gevaerts recommends waiting for someone knowledgeable to wander in
00:51:12 Part zaphee
00:52:38 Quit CaptainKewl (Quit: ( :: NoNameScript 4.22 :: ))
00:53:28 Quit Guest38472 (Read error: Connection reset by peer)
00:54:50 Join TheSphinX_ [0] (
00:57:20 Quit TheSphinX^ (Ping timeout: 240 seconds)
01:07:35*[Saint_] is fairly willing to assume that the 'with dsp' option is actually using the eq.
01:07:45[Saint_]the code got me lost, though.
01:08:03 Quit pamaury (Ping timeout: 256 seconds)
01:14:57 Quit funman (Ping timeout: 252 seconds)
01:16:04 Join funman [0] (~fun@rockbox/developer/funman)
01:18:33 Quit ParkerR (Excess Flood)
01:19:51 Join TheSphinX^ [0] (
01:20:39 Quit bluebrother (Disconnected by services)
01:20:42 Join bluebrother^ [0] (~dom@rockbox/developer/bluebrother)
01:21:29 Join ParkerR [0] (ParkerR@unaffiliated/parkerr)
01:21:51 Quit fs-bluebot (Ping timeout: 248 seconds)
01:22:55 Quit TheSphinX_ (Ping timeout: 245 seconds)
01:23:08 Join fs-bluebot [0] (
01:25:23 Join funman_ [0] (~fun@rockbox/developer/funman)
01:25:38 Quit ParkerR (Excess Flood)
01:27:34[Saint_]nicely formatted test_codec results of 10 band eq documented at g386 now:
01:27:36fs-bluebotGerrit review #386 at : 10 Band EQ w/Presets adapted from VLC by Hayden Pearce (changes/86/386/4)
01:29:08 Quit funman (Ping timeout: 276 seconds)
01:29:12[Saint_]the classic is a bit of a beast. flac_8 in ~9MHz, lame_320 and vorbis_500 in ~30MHz
01:30:01JdGordon[Saint_]: are you going to deal with the inline comments?
01:30:45 Quit dfkt (Quit: -= SysReset 2.55=- Sic gorgiamus allos subjectatos nunc.)
01:31:11[Saint_]I didn't quite understand your comment. I'm not reinventing the's exactly the same as it's done in git HEAD, just...more of it.
01:32:05[Saint_]Comments on the task I adapted it from said it would be accepted as is if it was proven to run on all targets.
01:32:22[Saint_]Basically all I did was change the frequency stepping and add presets.
01:32:48[Saint_]and sync an ancient patch to HEAD
01:33:59 Join ParkerR [0] (ParkerR@unaffiliated/parkerr)
01:34:25JdGordonreplacing those magic values with a #define would make it much nicer
01:34:38JdGordoneasier to change how many bands are available
01:34:52JdGordoni.e the low ram targets may only want 5 bands
01:35:24[Saint_]from what I can see, it _should_ run on all targets that have the eq already. there's a 1~2MHz increase or less in CPU usage, and the small screen targets already scroll to accommodatethe gui eq screen so I thought it was pretty much good to go.
01:36:56[Saint_]sorry, that is poorly worded. that cpu usage increase is caused by using the dsp anyway...the increase from the extra bands is very trivial.
01:37:02[Saint_]~1MHz or less.
01:40:22soap1Mhz smells way too low.
01:40:37[Saint_]soap: since you seem to know about these things, do you think I should be using 32, 64, 125, 250, ... or 30, 60, 120, 250, ...
01:41:12soapI'm searching now as I can't believe saratoga or buschel hasn't already profiled this, but for a long time the video just about died on Senab's build with more bands turned on.
01:41:26[Saint_]and, I thought about that...but maybe it causes boosting more often? the classic can do all the codecs I tested fine without boosting in theory but I didn't disable boosting for the test_codec runs I did.
01:41:27soapand that's after MP3 was already below 30Mhz.
01:41:55soapmy memory may be faulty as shit, so don't hold the boat on my account.
01:42:07 Quit prof_wolfff (Ping timeout: 248 seconds)
01:42:28[Saint_]do you have a suggestion about the frequency stepping? 32, 64, ... vs 30, 60, ...?
01:44:11 Quit Raptors (Read error: Connection reset by peer)
01:44:40soapMy headphones / truck can't do 30 and my ears can't hear 16 in music, so put them where you like! ;) The center points will be changeable, no?
01:45:50[Saint_]It confuses me, if ISO 266 is indeed a preferred standard, then why does virtually nothing seem to use it? :-S
01:47:16soapwhat do you mean?
01:48:27soap10-bands on even octaves starting at 30 has long been standard. Isn't that what ISO 266 is asking for?
01:48:42[Saint_]well, according the the patch I was reading from ISO 266 standard is 30, 60, 120, 250
01:48:53 Nick funman_ is now known as funman (~fun@rockbox/developer/funman)
01:49:01[Saint_]but lots of stuff is at 32, 64, 125, 250, ...
01:49:28soapyou see a 15 band hardware EQ it's most likely 2/3rds of an octave, again with the same range
01:50:07soapthe real number is 31.25
01:50:26soapexcuse me - 32
01:50:53soapdifferent people have rounded the text on hardware eqs differently.
01:52:21[Saint_]I don't think we can do decimals, the accepted values is just in Hz, so 32, ~, 16000.
01:52:55[Saint_]I wasn't sure what to do with the rounding, as its clearly closer to 31, but it seems to be common to use 32 and double from there.
01:53:48[Saint_]double from 16 or 32, then round to 125 and double from there.
01:55:46***Saving seen data "./dancer.seen"
01:55:49 Nick [Saint_] is now known as [Saint] (~saint@rockbox/user/saint)
01:56:05 Quit mt (Read error: Connection reset by peer)
01:57:42[Saint]if anyone else wants to do the same/similar test with other targets, I would appreciate it.
01:57:48 Join froggymana [0] (
01:57:48 Quit froggymana (Changing host)
01:57:48 Join froggymana [0] (~me@unaffiliated/froggyman)
01:58:00[Saint]I will test on an iPod Color this evening.
01:58:02 Join mt [0] (~quassel@
01:58:02 Quit mt (Changing host)
01:58:02 Join mt [0] (~quassel@rockbox/developer/mt)
01:58:13 Quit froggymana (Remote host closed the connection)
01:58:49 Join Raptors [0] (
02:00:17[Saint]The way I did it was three runs of test_codec for each case: speed test HEAD w/oDSP, speed test HEAD w/DSP, speed test 10 Band EQ patch w/DSP; average the results.
02:00:20soapthe ipod color should answer the PP question once and for all
02:01:14[Saint]One run is probably enough, I did three and averaged just to try to get a clearer picture and hopefully eliminate the possibility of a 'miracle run'.
02:02:32[Saint]I don't have any AMS players I can get to easily, so if someone got bored and had a Clip/Fuze/etc laying around, go nuts.
02:03:51[Saint]I've been trying out the presets today, and although they're kinda kludged they're really not too bad.
02:06:11 Quit ParkerR (Excess Flood)
02:06:20[Saint]One thing this made me think of, as well, is that Rockbox really needs a "first run script". Something capable of inclusion into a build that runs once after boot and then wipes itself out. SO that one can easily remove depricated files.
02:07:09[Saint]For example, were this committed, there would be several 5 band EQ presets leftover in the users eqs folder that would be useless.
02:08:56[Saint]could probably do it with LUA and the rockboot feature or whatever it's called.
02:09:28 Quit ender| (Read error: Operation timed out)
02:09:45soapthe worst that could happen is they load an old .cfg and only the first 5 bands get set, right? You didn't change the .cfg format did you?
02:11:39[Saint]I just thought it might be nice to have a way to guarantee we could wipe out depricated files without user intervention.
02:12:09soapwhat if the user created their own files i the /.rockbox/eqs?
02:12:14soapwould you wipe those too?
02:12:24soapor just the rockbox-issued ones?
02:12:41[Saint]Only known files, ones that came with a previous binary.
02:12:47[Saint]SHouldn't ever wipe user files.
02:12:49soapown files IN /.rockbox/eqs
02:13:04soapso just ship new preset replacements for all
02:13:24 Join TheSphinX_ [0] (~briehl@
02:13:27soaponly use 5 bands of the possible 10 of the ones you don't feel up to reworking. define the other five bands as off.
02:14:02[Saint]Ah, right, yes. good point.
02:14:09[Saint]I can do that, genius.
02:14:39[Saint]Another is caused by me renaming Default to Flat to be more obvious, but I can undo that I guess.
02:14:59 Join ParkerR [0] (ParkerR@unaffiliated/parkerr)
02:15:01 Quit TheSphinX^ (Read error: Operation timed out)
02:15:18Torneor just ship two opies
02:15:21Tornethey rae tiny
02:15:23Tornenobody cares
02:15:49[Saint]have Default (the previous, 5 band eq), and Flat (the new, 10 band eq)?
02:16:02Torneno, i mean, just ship two copies of exactly the same flat 10 band preset
02:16:06Tornewith both names
02:16:14Tornebecause it's a tiny textfile and who cares
02:16:29soapuser confusion. "I had selected default and found rockbox so warm and balanced. I then tried flat and all the subtle foot tapping goodness went away and the soundstage collapsed"
02:17:02[Saint]audiophile trap for patrick82 wannabes
02:17:28soapthen ship one called "Apple.cfg" and another "Cowon.cfg" and make them both flat.
02:17:34soapif you /really/ want to troll.
02:18:58[Saint]But, anyway, ok...yeah, I'll convert the existing 5 band presets to 10 band and and just leave the new frequency steps with 0 values.
02:19:37[Saint], I don't really like that idea.
02:19:46[Saint]maybe I should try and guestimate the gaps.
02:21:00[Saint]it would be nice if there was a way to wipe them out...
02:22:44[Saint]yeah, I might try and fill the blanks instead of leaving the values blank.
02:25:47 Join ender| [0] (whatever@2a01:260:4094:1:42:42:42:42)
02:28:56[Saint]Ok, I'll do that when I get back home from my pending missions out and about, then as far as I'm concerned it should be able to be committed unless anyone comes up with a target that it completely destroys runtime/realtime playback on.
02:30:24[Saint]I'll bench the iPod Color and a Nano2G as well, and maybe a Fuze if I can find it. I could probably also do RaaA but I'm not sure if that really matters.
02:33:05[Saint]Do you think it might help if I provided test binaries and put them on the forum?
02:35:27Tornei have found generally that if i do that i get maybe one or two people testing half a dozen common players
02:35:31Tornewhich is better than nothing
02:35:43Torneit depends how much you have scripted making the test builds :)
02:35:58Torneif you have a sinlge command to do it and can jut leave it for a couple hours and get a dir of zips then it's worth it ;)
02:36:04Tornei think our release script will do it in fact no?
02:41:04saratoga[Saint]: something is wrong with those benchmarks
02:41:16saratogaits probably about 1-2 MHz per EQ band on ARMv5
02:42:06[Saint]Well, I can only go by what test_codec gave test_codec actually using the EQ with the "speed test with DSP" option?
02:42:29[Saint]As near as I can figure it is.
02:42:58saratogait should
02:43:07saratogaare all 10 bands actually enabled though?
02:43:18[Saint]The other thing I thought of is that perhaps I should do the test with boosting disabled.
02:43:52[Saint]I thought perhaps this actually caused more boosting during decoding and thats why there was a very trivial difference.
02:44:25saratogathe codec results are done at constant clock speed
02:44:36[Saint]but, anyway, yeah...that's the average of three runs on test_codec.
02:44:47[Saint]besides doing them again, not sure what else I can do.
02:44:51saratogabasically what those results probably indicate is that the EQ isn't enabled in test_codec for some reason
02:45:01[Saint]I'll be testing other targets tonight as well.
02:45:33[Saint]well the with/without DSP runs are clearly different...
02:45:48[Saint]There just doesn't seem to be much difference in adding 5 more bands.
02:46:10[Saint]But there is a clear difference with speed test with and without DSP.
02:46:32[Saint]+~2MHz on the Classic
02:49:22saratogathe difference is due to the normal DSP code
02:49:27saratogathe EQ isn't being called though
02:49:51saratogalooking at the code i don't see how it wouldn't be called
02:50:01saratogaare you sure you actually have all 10 bands configured in settings?
02:51:00Torneyou know i keep getting tempted to make a little thing for our website thta just asks you a series of questions to determine if your ipod is supported or not
02:51:04Tornewell not just ipod, player
02:51:41Tornepeople do not read our list, and even when they do there are things not on the list (like early wip ports), and then they constantly get their model wrong
02:51:50saratoga"Is it listed on the front page" "No?
02:51:54saratoga"Go away"
02:51:58Tornesaratoga: right but that doesn't help if they don't know what it is
02:52:01Tornewhich lots of people don't
02:52:08Tornee.g. all the ipodvideo/classic confused folks
02:52:44Torneare there still ports on TargetStatus that aren't listed on the front page at all?
02:52:49Torneor did someone fix that
02:53:11Torneyeah, there are
02:53:12Tornee.g. xfi
02:53:28[Saint][14:50:03] <saratoga> are you sure you actually have all 10 bands configured in settings? <−− quite
02:54:41[Saint]perhaps I should do the tests with a very aggressive and unrealistic amount of gain/precut on each band?
02:55:15[Saint]At the moment I'm using a quite sane "Pop" preset.
02:55:48[Saint]But, iiuc, it shouldn't matter even if I have "Flat/Default enabled.
02:56:47[Saint]By my understanding it shouldn't matter if the EQ is on, but all values are set flat, it should still consume additional CPU cycles...correct?
02:57:12Tornei would hope not
02:57:18Torneiwould hope we skip processing for bands that are on 0db
02:58:10[Saint]I got lost quite quickly in the code.
02:58:14[Saint]quite quickly.
02:58:31Tornewell set every band to 1db an dsee :)
02:59:58[Saint]I'll re-test when I get back from my missions, I'll do the same test, but this time with very aggressive gain/precut, basically creating an artificial 'worst case scenario'.
03:00:09saratogajust looking at the code its about 5 multiplies, loads and stores per sample, and you do 88200 samples per second
03:00:10[Saint]See if that makes a considerable difference.
03:00:43saratogaso figure 2 cycles per op times ~10 ops times 100k samples = 2MHz per band
03:02:05[Saint]but, anyway, yeah...I'm procrastinating. I need to get on with my missions. Leave a comment on the tracker if you test on other targets before I get a chance to, or leave a message in the logs for me. o/
03:03:10saratoga(for armv5, figure about double that on armv4 due to half speed multiplier and load unit)
03:07:38 Quit Guest78563 (Ping timeout: 276 seconds)
03:38:58 Quit user890104 (Read error: Operation timed out)
03:39:24 Join Xerion_ [0] (
03:39:27 Join user890104 [0] (Venci@unaffiliated/user890104)
03:40:00 Quit Xerion (Read error: Connection reset by peer)
03:40:01 Nick Xerion_ is now known as Xerion (
03:55:47***Saving seen data "./dancer.seen"
03:57:11saratoga" Assume a band is disabled if the gain is zero "
03:57:18saratogalooks like they're skipped if there is no gain
04:04:27 Join SuperBrainAK [0] (
04:08:01 Join TheSphinX^ [0] (
04:11:47 Quit TheSphinX_ (Ping timeout: 245 seconds)
04:12:55 Join [Saint_] [0] (~quassel@
04:13:19 Nick [Saint_] is now known as Guest62018 (~quassel@
04:27:13 Join amiconn_ [0] (amiconn@rockbox/developer/amiconn)
04:27:14 Quit amiconn (Disconnected by services)
04:27:16 Nick amiconn_ is now known as amiconn (amiconn@rockbox/developer/amiconn)
04:27:19 Join akaWolf [0] (~akaWolf@unaffiliated/akawolf)
04:27:45 Join pixelma_ [0] (pixelma@rockbox/staff/pixelma)
04:27:45 Quit pixelma (Disconnected by services)
04:27:47 Nick pixelma_ is now known as pixelma (pixelma@rockbox/staff/pixelma)
05:17:53 Join thegeek_ [0] (
05:18:55 Quit TheSeven (Disconnected by services)
05:19:05 Join [7] [0] (~quassel@rockbox/developer/TheSeven)
05:20:50 Quit thegeek (Ping timeout: 240 seconds)
05:22:43 Join Rower [0] (
05:26:37 Quit jhMikeS (Ping timeout: 256 seconds)
05:32:05 Join TheSphinX_ [0] (
05:34:27 Quit TheSphinX^ (Ping timeout: 248 seconds)
05:55:51***Saving seen data "./dancer.seen"
05:56:36 Join TheSphinX^ [0] (
05:56:47 Quit [Saint] (Remote host closed the connection)
06:00:11 Quit TheSphinX_ (Ping timeout: 264 seconds)
06:03:49 Join [Saint] [0] (~saint@rockbox/user/saint)
06:12:57 Quit saratoga (Ping timeout: 245 seconds)
07:10:48 Quit SuperBrainAK (Quit: pbly going to sleep /_\)
07:12:28 Quit Buglouse (Read error: Operation timed out)
07:14:33 Join webguest59 [0] (
07:17:04 Join scorche [0] (~scorche@rockbox/administrator/scorche)
07:19:25 Quit scorche` (Ping timeout: 256 seconds)
07:21:17 Quit webguest59 (Quit: CGI:IRC (EOF))
07:26:09 Join Roboturner913 [0] (6378e8b5@gateway/web/freenode/ip.
07:27:29Roboturner913Who knows stuff about the Fuze+?
07:28:12Roboturner913Specifically, I am trying to figure out a nasty screen flicker....
07:29:38 Join Buglouse [0] (~Buglouse@unaffiliated/Buglouse)
07:36:18 Quit Roboturner913 (Quit: Page closed)
07:39:49 Quit Buglouse (Ping timeout: 255 seconds)
07:40:29 Join enriched [0] (~quassel@
07:53:07 Join jhMikeS [0] (~jethead71@
07:53:07 Quit jhMikeS (Changing host)
07:53:07 Join jhMikeS [0] (~jethead71@rockbox/developer/jhMikeS)
07:55:53***Saving seen data "./dancer.seen"
07:58:41 Join melmothX [0] (~melmoth@unaffiliated/melmothx)
07:59:50 Quit jhMikeS (Ping timeout: 240 seconds)
08:00:56 Join jhMikeS [0] (~jethead71@rockbox/developer/jhMikeS)
08:01:54 Join Buglouse [0] (~Buglouse@unaffiliated/Buglouse)
08:19:58 Join ender` [0] (
08:22:27 Quit Buglouse (Ping timeout: 248 seconds)
08:33:25 Join Zagor [0] (
08:33:25 Quit Zagor (Changing host)
08:33:25 Join Zagor [242] (~bjst@rockbox/developer/Zagor)
08:37:43 Join Buglouse [0] (~Buglouse@unaffiliated/Buglouse)
09:00:32 Join bertrik [0] (~quassel@rockbox/developer/bertrik)
09:01:57 Quit Buglouse (Ping timeout: 246 seconds)
09:05:51 Join einhirn [0] (
09:10:59 Join petur [0] (~petur@rockbox/developer/petur)
09:18:03 Join kevku [0] (~kevku@2a01:d0:ffff:34a::8:3)
09:20:23 Join LinusN [0] (
09:26:53 Quit bertrik (Ping timeout: 276 seconds)
09:38:28 Join Buglouse [0] (~Buglouse@unaffiliated/Buglouse)
09:44:23 Join pystar89 [0] (
09:52:35 Quit Buglouse (Ping timeout: 276 seconds)
09:55:54***Saving seen data "./dancer.seen"
10:00:48 Quit Guest62018 (Ping timeout: 248 seconds)
10:02:53 Join jlbiasini [0] (~metaphysi@
10:25:30 Join Buglouse [0] (~Buglouse@unaffiliated/Buglouse)
10:27:41jlbiasinipamaury: kugel: I found the missing include it's <signal.h> and not <asm/signal.h>
10:42:05 Quit Buglouse (Ping timeout: 252 seconds)
10:46:11 Join wodz [0] (
10:46:49 Join bebna [0] (~a.fasold@
10:52:27kugeljlbiasini: pamaury said the thread-unix.c includes it already
10:54:10jlbiasinikugel: no it doesn't
10:56:15 Join Buglouse [0] (~Buglouse@unaffiliated/Buglouse)
10:57:18jlbiasinikugel so here we go
11:02:32wodzhow it is possible that other sims (as well as our buildfarm) does not signal problem?
11:16:03kugeljlbiasini: it's includedd a few lines above
11:16:04jlbiasinino idea or perhaps I'm missing a dev package on my debian?
11:16:23jlbiasinistupid me
11:16:25kugelthe third include is <signal.h>
11:16:46kugelyou just added it a second time. but perhaps the order is important
11:16:48jlbiasiniHow is it possible that it doesn't work if not double???
11:16:59jlbiasiniah right !
11:17:17jlbiasiniI will try
11:17:49kugelsigcation needs _POSIX_C_SOURCE >= 1 || _XOPEN_SOURCE || _POSIX_SOURCE. I guess one of the header after define one of them
11:24:58wodzWhat about pushing g#339 and g#340? The only reason this is not merged yet is that dreamlayers is not active anymore. The patch is widely tested and closes quite serious problem on many targets advertised as stable.
11:25:01fs-bluebotGerrit review #339 at : Fix FS #12391 : Memory corruption on PP502x after commit_discard_idcache() by Boris Gjenero (changes/39/339/1)
11:25:21JdGordonso push it
11:27:02jlbiasinikugel: ok I don't understand anything now it's working
11:27:15jlbiasiniI reinstall the git tree
11:27:55jlbiasiniI didn't had any unstashed change reported and everything up to date... quite a mystery...
11:36:11 Join [Saint_] [0] (~quassel@
11:36:35 Nick [Saint_] is now known as Guest25974 (~quassel@
11:48:34 Quit efyx (Ping timeout: 248 seconds)
11:50:46 Quit TheSphinX^ (Read error: Operation timed out)
11:53:13[Saint]new results at g386
11:53:36 Quit fs-bluebot (Read error: Connection reset by peer)
11:53:59[Saint]I must've accidentally reset the EQ to 0 during the initial testing.
11:54:00wodzhmm, magically bootloaders can't find commit_discard_dcache() when linking
11:54:14[Saint]New results show a clear increase in CPU usage.
11:54:27[Saint]But, not as much as I had been lead to believe it would be.
11:54:59[Saint]Except for flac...which got crippled by it.
11:55:54 Join TheSphinX^ [0] (~briehl@
11:55:56***Saving seen data "./dancer.seen"
11:56:37[Saint]lame 320 and vorbis 500 saw a ~5MHz gain for realtime decoding (~30MHz vs ~36MHz), but flac 8 /almost/ doubled in CPU usage from ~9MHz to ~15MHz.
11:57:36wodzah right it doesn't build quite big part of system-pp502x.c for bootloader
11:58:30wodz15 is still not much - below boost threshold
12:01:51 Join efyx [0] (~efyx@
12:04:35 Quit Buglouse (Ping timeout: 245 seconds)
12:06:09wodzHmm, I think I'll revert last commit. I don't have time now to dive into code how this should be arranged. For bootloaders cache_init() is not called so I think using cache functions is not safe there. Comments are a bit scary.
12:07:14kugel[Saint]: cpu increase should be constant, regardless of the codec
12:08:04gevaerts[Saint]: are you forcing boost during testcodec runs?
12:08:41gevaertsI'd consider doing that to reduce variables
12:09:05wodztest_codec boost by itself
12:09:41wodzyou have to *force unboost* to test unboosted performance
12:09:56[Saint]Aha, right. Ok. Good to know.
12:11:52 Quit jlbiasini (Quit: jlbiasini)
12:12:17kugelgevaerts: testcodec always boosts
12:12:33gevaertskugel: [Saint] told me it doesn't :)
12:12:41*gevaerts learns never to believe [Saint]!
12:12:44kugelhe's wrong then :)
12:13:58gevaerts[Saint]: by the way, could you run a set of APE benchmarks on the classic one of these days?
12:14:17 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
12:17:50[Saint]APE crashes and burns.
12:17:56[Saint]And I don't know why.
12:18:03[Saint]So does wmapro
12:18:11[Saint]All I get is high pitched noise.
12:18:31gevaertsI meant in test_codec
12:18:37[Saint]It either outputs a terrible noise, or skips the file completely.
12:18:43gevaertsStill, if it's buggy, not worth trying :)
12:19:02[Saint]I don't know what the problem with it is...
12:19:17[Saint]I thought it was my test_files library became corrupt...but, nope.
12:19:56[Saint]I first noticed it when I saw test_codec claiming 16000000% realtime for APE
12:22:58wodzI don't understand it. HAVE_ATA_DMA for PP was disabled in 55fab77 so before it had to build cleanly. Now enabling it again spits errors for bootloaders. In the meantime I don't see any significant changes.
12:23:10wodzany clues?
12:30:36pamaurywodz: missing include when in bootloader ?
12:30:56pamaurypreviously it did not use commit_discard perhaps ?
12:32:40wodzit did previously (aka when HAVE_ATA_DMA) cpucache_invalidate() but since then names have changed
12:33:46 Join redhot [0] (~kvirc@
12:34:00wodzbut it should not compile cpucache_invalidate() at all then
12:47:05 Quit Poodlemastah (Ping timeout: 245 seconds)
12:48:50wodzOk, in ancient version cpucache_invalidate() was empty stub in bootloader build. I wonder how it was supposed to work with dma transfers.
12:49:38 Part redhot ("Once you know what it is you want to be true, instinct is a very useful device for enabling you to know that it is")
12:49:52 Join Poodlemastah [0] (
12:49:52 Quit mt (Read error: Connection reset by peer)
12:50:44 Quit petur (Quit: *plop*)
12:51:04 Join mt [0] (~quassel@rockbox/developer/mt)
12:51:36wodzis it possible to define HAVE_ATA_DMA only for non bootloader builds?
12:52:18 Quit mt (Read error: Connection reset by peer)
12:53:27pamaurywodz: Probably, depends on where it is defined
12:53:32 Join mt [0] (~quassel@rockbox/developer/mt)
12:53:56wodzHAVE_ATA_DMA is defined in firmware/export/config/xxx.h files
12:55:39wodzyea it works but looks ugly
12:57:01 Join mt_ [0] (~quassel@
12:57:02 Quit mt (Read error: Connection reset by peer)
12:57:05wodzoh, we stub out some lcd functions this way so this is not new technique :-)
12:58:00 Quit Poodlemastah (Read error: Connection reset by peer)
12:58:00 Quit mt_ (Read error: Connection reset by peer)
12:58:40 Join mt [0] (~quassel@
12:58:41 Quit mt (Changing host)
12:58:41 Join mt [0] (~quassel@rockbox/developer/mt)
12:59:36pamaurywe already do that for many things
13:00:19wodzpamaury: sure in many places but not so much in config/xxx.h
13:00:42pamaurynearly all config files do it !
13:01:13kugelwodz: sure it's possible
13:01:23 Join Poodlemastah [0] (
13:01:31kugeljust define it within #ifndef BOOTLOADER
13:04:47 Quit Guest25974 (Ping timeout: 248 seconds)
13:07:15 Join Buglouse [0] (~Buglouse@unaffiliated/Buglouse)
13:19:34 Quit Buglouse (Ping timeout: 255 seconds)
13:31:50 Join jlbiasini [0] (~metaphysi@
13:35:09 Join mortalis [0] (
13:39:10pamauryjlbiasini: I don't understand your gerrit patch ?!
13:39:56jlbiasininothing to understand it's abandonned
13:40:18jlbiasiniYou did experienced the same errors as I was yesteday didn't you ?
13:40:52jlbiasiniI reinstalled my tree and now I can't reproduce it anymore
13:42:15jlbiasinistange because I didn't had any thing unstaged. In fact I had this signal.h include so I didn't so there was already one, but kugel was suspecting perhaps the order ofthe include matters
13:42:59jlbiasinibut after it worked once, it was always working even after reseting the file... mystery...
13:44:14wodzjust like our 'lib/libtlsf.a: could not read symbols: Archive has no index; run ranlib to add one' random error :-)
13:44:54wodzI'd love to hit it locally and inspect intermediate files to understand whats going on
13:47:16pamauryis it always the same bot ?
13:51:31pamauryare we building with make -j ?
13:51:55wodzwith -jn where n is num of cores AFAIK
13:54:00pamaurysometimes it can yield very strange results
13:55:07Zagorpamaury: -j can give strange results if you don't take care of your dependencies. we take rather good care of ours.
13:55:21wodzapparently not good enough
13:55:58***Saving seen data "./dancer.seen"
13:56:05 Join Buglouse [0] (~Buglouse@unaffiliated/Buglouse)
13:56:06pamauryI can tell you for sure that even in Makefiles with perfect dependencies I've hit some weird errors, usually only happen once, of course
13:56:23Zagorwodz: that's entirely possible, of course. bugs happen.
13:56:40Zagorpamaury: you talk like -j is black magic. it's just parallell compilation.
13:56:40pamauryI'm not blaming make though, that could be a bug in the compilers
13:56:44kugelwhat makes you think it's a make -j error? just because it's strange?
13:57:11pamaurybecause you get it only with make -j and not with make ?
13:57:36kugelhow do you know?
13:58:39wodzWith limited number of builds on our farm this days we could switch to -j1 for a week or so and see if strange problems go away.
13:59:22pamaurySight, I'm not saying "this is make's fault", I said (please read) 1) i'm not blaming make 2) i've got errors only with make -j 3) which I don't have with make
13:59:22pamauryThat's all, conclude what ever you want
13:59:57pamauryCorrelation is not implication, as one says
14:00:19Zagorlibtlsf certainly look suspicious
14:01:22wodzZagor: In which way?
14:02:32Zagorwodz: in that it's very often failing there.
14:03:27wodzactually this is the only static lib we produce which seems to trigger this error
14:04:28wodzcould it be that codecs for example do not depend on libtlsf.a ?
14:04:43ZagorI'll take a peek
14:05:00 Join amee2woof [0] (
14:05:17wodzin plugins IMO we explicitly compile tlsf.c file
14:07:23 Quit Buglouse (Ping timeout: 276 seconds)
14:07:27wodzok, mikmod and pdbox do use libtlsf but do not compile a copy and I recall seeing failures there
14:08:00 Quit __jae__ (Read error: Operation timed out)
14:09:21 Join __jae__ [0] (
14:10:22 Join amayer_ [0] (
14:12:57Zagoraac codec for instance does not specify dependency on libtlsf.a, so it is linked before libtlsf is completed
14:15:24wodzmaybe make libplugin.a depended on libflsf.a
14:16:27Zagorbut only a few codecs use tlsf, right?
14:16:43wodzI mean libcodec
14:17:06Zagorthen adding it as a depencency for all would affect build performance. all codecs would have to wait for tlsf to complete before linking.
14:17:32Zagorwe specify in codecs.make which codecs use which libs. it feels to me like tlsf should be specified there.
14:17:44wodzcould be
14:17:54Zagorthough I'm a bit confused because I don't even see tlsf specified there at all
14:18:24wodzIt quite new move to have it compiled as separate lib
14:18:52wodzthe dependency bit has been omitted apparently
14:22:06Zagorbut where is it even linked? that's supposed to be handled in that file too
14:23:39Zagorahh, libtlsf.make adds it to EXTRA_LIBS
14:24:02wodzhmm, so it should propagate down to all codecs
14:25:26ZagorI think that's a bad idea. all codecs should not be linked with tlsf
14:25:49Zagorthose using it should list it in codecs.make
14:26:27wodzthats another matter, but if libtlsf.a is added to EXTRA_LIBS why it fails?
14:27:11wodzcurrently the very first codec should compile and link libtlsf.a
14:27:27Zagoryou're thinking serially. don't. :-)
14:27:34Zagorthey are all build at the same time
14:27:44 Join Buglouse [0] (~Buglouse@unaffiliated/Buglouse)
14:28:05pamaurythat's nasty, it starts building the codec before having the dependency ?!
14:28:50Zagorpamaury: yes, since we haven't said the lib is a dependency
14:28:52wodzah, you mean one codec started producing libtlsf.a but didn't write file completely while the second tried to link with this incomplete .a ?
14:29:01Zagorwodz: yes
14:30:04wodzI don't see how individual dependency could solve this
14:31:10 Quit wodz (Quit: Leaving)
14:43:44 Quit einhirn (Ping timeout: 245 seconds)
14:53:38 Quit AlexP (Ping timeout: 248 seconds)
14:55:43 Join AlexP [0] (~alex@rockbox/staff/AlexP)
14:55:45 Join nateloaf [0] (
14:57:39 Quit jlbiasini (Ping timeout: 256 seconds)
14:58:37 Join jlbiasini [0] (~metaphysi@
15:00:42 Join einhirn [0] (
15:08:01 Quit einhirn (Ping timeout: 255 seconds)
15:19:50 Join n1s [0] (
15:19:50 Quit n1s (Changing host)
15:19:50 Join n1s [0] (~n1s@rockbox/developer/n1s)
15:33:48copperHmmm, the Clip Zip port is still marked "unstable" on the homepage
15:35:00 Quit pamaury (Ping timeout: 245 seconds)
15:36:32Zagoras a matter of fact, *only* vorbis uses tlsf
15:37:03 Join einhirn [0] (
15:37:17 Quit AlexP (Remote host closed the connection)
15:37:33 Join AlexP [0] (~alex@rockbox/staff/AlexP)
15:37:35ZagorI can't remembed the purpose of EXTRA_LIBS. the name is too generic.
15:38:51Zagorand when I remove libtlsf from it, only libsetjmp remains
15:39:18 Quit mortalis (Quit: Leaving)
15:45:55 Quit Poodlemastah (Quit: ZNC -
15:51:22 Join eckoit [0] (~ryan@
15:56:00***Saving seen data "./dancer.seen"
15:56:43 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
16:07:37Zagorthat should fix the tlsf dependency problems
16:07:49 Join dfkt [0] (dfkt@unaffiliated/dfkt)
16:09:47 Quit kevku (Ping timeout: 264 seconds)
16:12:36Zagor...except it didn't. hmm.
16:25:02 Part LinusN
16:26:01 Join Poodlemastah [0] (
16:26:13 Quit Poodlemastah (Client Quit)
16:31:45 Quit jlbiasini (Quit: jlbiasini)
16:48:48 Join lebellium [0] (
17:07:07 Join PurlingNayuki [0] (~PurlingNa@
17:08:34PurlingNayukiThere're still several major bugs with RaaA... but really works fine with me.
17:12:01kugelZagor: some plugins too, no?
17:12:30 Quit PurlingNayuki (Quit: Leaving.)
17:12:41Zagorkugel: yes, I fixed both codecs and plugins
17:12:52Zagorthough the tlsf dependency issue remains. not sure why.
17:13:38funmancopper: unstable means incomplete manual and maybe broken graphics for a couple of games in this case
17:14:04copperthere was a lot of talk about promoting it to stable, I thought it was a done deal
17:14:34funmanno idea sorry
17:15:54gevaertscopper: IIRC (but it's a bit vague) that this discussion was after 3.12, and promoting to stable at random times between releases has some awkward side effects
17:16:12 Join y4n [0] (~y4n@unaffiliated/y4ndexx)
17:16:43 Quit the-kyle (Remote host closed the connection)
17:17:07 Join WalkGood [0] (~4@unaffiliated/walkgood)
17:18:02 Join kevku [0] (~kevku@2001:470:27:773:0:feed:c0f:fee)
17:21:44 Join Ward [0] (
17:22:08 Nick Ward is now known as Guest54714 (
17:39:27 Quit Zagor (Quit: Clint excited)
17:40:55 Join the-kyle [0] (
17:42:18 Quit the-kyle (Remote host closed the connection)
17:43:33 Quit bebna (Quit: Leaving.)
17:43:55 Join dynion [0] (
17:44:16dynionhey guys! think i might have found a bug... or it's just a human error...]
17:44:52dynionthe video 5th gen goes into ipod after every reboot
17:45:01dynioninstead of rockbox
17:45:26 Join the-kyle [0] (
17:46:05dynion[saint] ?
17:46:10Torneyou are probably leaving the hold switch on
17:46:12Torneread the manual
17:46:59dynioneven without hold it doesnt go into rockbox
17:47:27Torneor you are holding menu to turn it on, which also dualboots the original firmware
17:47:44dynionor is the rockboc installation gone after one time
17:47:47dynionah... ok:)
17:47:56Tornepress the select button one time to turn it on normally
17:48:52dynionwell first i have to get it out completely
17:48:56dynioninstead of the sleep mode
17:49:27dyniongod this is a frustrating os... now i know why i hated apple
17:50:09dynionshould i re-rockbox it before doing it again? cause i can't get it offline properly
17:51:08Tornethe manual explains how dualbooting works
17:51:17dynionoh you didnt mean the apple manual
17:51:19Torneto ge tback out of the apple OS you have to hard reset
17:51:27dynionyou guys made your own... awesome:D
17:53:34 Join Bug2000 [0] (~bug2000@unaffiliated/bug2000)
17:56:01***Saving seen data "./dancer.seen"
17:57:21dynionThanks torne!
17:57:29dynionHuman error it is:D
17:58:08dynionso from now on... if i want to start it just use the big round button:D sounds idiot-proof enough
17:59:20 Quit dynion (Quit: CGI:IRC (EOF))
18:00:17 Join pretty_function [0] (~sigBART@
18:00:45 Quit pretty_function (Client Quit)
18:26:38 Join thomasjfox [0] (~thomasjfo@rockbox/developer/thomasjfox)
18:26:48 Quit preglow (Ping timeout: 240 seconds)
18:27:34 Join preglow [0] (
18:27:35 Quit preglow (Changing host)
18:27:35 Join preglow [0] (thomj@rockbox/developer/preglow)
18:27:40 Quit aevin_ (Ping timeout: 264 seconds)
18:27:57 Join aevin [0] (eivindsy@unaffiliated/aevin)
18:30:46 Quit einhirn (Quit: Miranda IM! Smaller, Faster, Easier.
18:39:20 Join lorenzo92 [0] (~chatzilla@
18:45:12 Join TheSphinX_ [0] (
18:46:26 Quit melmothX (Remote host closed the connection)
18:46:55 Join melmothX [0] (~melmoth@unaffiliated/melmothx)
18:47:33 Quit TheSphinX^ (Ping timeout: 252 seconds)
18:48:42 Join TheSphinX^ [0] (
18:49:02*lorenzo92 implementing backlight controller for R1 in rb
18:52:13 Quit TheSphinX_ (Ping timeout: 260 seconds)
18:54:12 Join TheSphinX_ [0] (
18:57:04 Quit Bug2000 (Read error: Connection reset by peer)
18:57:19 Quit TheSphinX^ (Ping timeout: 248 seconds)
18:58:13 Quit melmothX (Quit: #)
18:59:43 Join TheSphinX^ [0] (
19:00:24 Quit TheSphinX_ (Read error: Operation timed out)
19:06:00 Join pamaury_ [0] (~quassel@rockbox/developer/pamaury)
19:06:37 Quit pamaury (Ping timeout: 255 seconds)
19:07:02 Quit lorenzo92 (Ping timeout: 276 seconds)
19:07:59 Quit Rower (Quit: sover)
19:08:09 Join Strife89 [0] (
19:13:10 Part eckoit
19:14:48 Join mt_ [0] (~quassel@
19:14:48 Quit mt (Read error: Connection reset by peer)
19:27:06[Saint][03:33:49] <copper> Hmmm, the Clip Zip port is still marked "unstable" on the homepage <−− you expect otherwise?
19:27:28copper<copper> there was a lot of talk about promoting it to stable, I thought it was a done deal
19:27:51 Join xela [0] (
19:28:36xelahey all. i managed to get a 240gb toshiba to work in a ipod video. problem now is replay "stutters" a couple of times when playing a song.
19:30:09 Join bertrik [0] (~quassel@rockbox/developer/bertrik)
19:32:15 Join lorenzo92 [0] (~chatzilla@
19:32:17xeladoes this ring a bell for somebody?
19:34:02 Quit lorenzo92 (Client Quit)
19:34:22 Join lorenzo92 [0] (~chatzilla@
19:34:28 Join lebellium_ [0] (
19:35:49 Quit lebellium (Ping timeout: 245 seconds)
19:36:15[Saint]copper: "stable" has less to do with the stability of the port, and more to do with meeting a set of requirements: complete manual, all applicable plugins, rbutil support, etc.
19:36:36copperyeah I know, I thought those requirements were met
19:36:40[Saint]Apparently it's lacking in at least one of those areas, which doesn't entirely surprise me.
19:36:44[Saint]It's a very new port.
19:37:57 Join lebellium [0] (
19:38:03[Saint]Changing the port classifications to something a little more obvious has been discussed quite a few times, but no one ever seems to come up with anything better than unusable, unstable, and stable.
19:38:32[Saint]but it is often amusing, and confusing, that unusable and unstable rarely actually mean that literally.
19:38:34 Quit lebellium (Read error: Connection reset by peer)
19:39:02 Join lebellium [0] (
19:39:34 Quit lebellium_ (Ping timeout: 245 seconds)
19:43:56 Quit lebellium (Read error: Network is unreachable)
19:44:27 Join lebellium [0] (
19:44:49 Quit zoktar (Quit: -)
19:46:15 Quit akaWolf (Ping timeout: 245 seconds)
19:53:17 Join zoktar [0] (~zoktar@unaffiliated/zoktar)
19:53:58 Quit lebellium (Read error: Connection reset by peer)
19:54:29 Join lebellium [0] (
19:56:03***Saving seen data "./dancer.seen"
19:57:36 Quit xela (Ping timeout: 256 seconds)
20:01:46 Join xela [0] (
20:05:14n1sxela: does it stutter only when rebuffering?
20:09:08 Quit amayer_ (Quit: amayer_)
20:11:44 Join amayer_ [0] (
20:12:09 Quit WalkGood (Ping timeout: 256 seconds)
20:17:17 Join nomada [0] (
20:18:02 Quit Strife89 (Quit: ------->)
20:19:40xelanls: when does rebuffering take place
20:20:41 Join prof_wolfff [0] (
20:21:19n1sjust before the buffer is empty or when skipping out of what's buffered, the disk will spin up then and if it takes too long for some reason that will cause a dropout. You can also see the buffer state in the "buffering thread" screen in the debug menu
20:27:47 Join Strife89 [0] (
20:29:50xelanls hard to see it happen
20:31:41[Saint]It's quite easy to see it happen, actually.
20:32:34[Saint]System - Debug - View buffering thread
20:32:39[Saint]xela: ^
20:32:49[Saint]You can see the buffer empty and fill in realtime.
20:32:59[Saint](with playback, of course)
20:33:07xelai know but to start a replay and then do it in time
20:33:36[Saint]You don't need to, it will refill the buffer again fairly quickly.
20:33:47[Saint]It isn't as it it only happenes once.
20:34:24n1sand if you use lossy music it should take a half hour or more for it to empty if you playlist is long enough
20:34:53[Saint]s/lossy/really hideously lossy/
20:35:45n1sit could be the 64MB model!
20:37:12[Saint]Oh...guess you're right, actually. A rough guess with my classic suggests it will take about ~30 mins to deplete the buffer with lame 320
20:37:29[Saint](64MB ~59MB available buffer)
20:38:38[Saint]Allowing playback control in the view buffering thread screen might be a good idea.
20:38:53[Saint]The you could actually cause a rebuffer and watch it happen.
20:39:19[Saint]afaik the keys (on at least a few targets) are free to allow play/pause and seeking.
20:41:12n1si thought you could skip from that screen
20:41:17 Join TheSphinX_ [0] (
20:41:44xelai will try
20:42:09xelait stutter doesnt happen when the playback continues from track to track
20:42:15xelaonly when i start the playback
20:42:18xelaand not always
20:43:56[Saint]n1s: apparently not on the Classic?
20:44:11[Saint]Unless for <some_awesome_reason> the keys are no longer << and >>
20:44:16xelanls/Saint: skipping doesnt work there
20:44:29[Saint]I seem to recall it used to, as n1s said.
20:44:32[Saint]It likely broke.
20:44:49 Quit TheSphinX^ (Ping timeout: 276 seconds)
20:46:10n1si can do it on my c200 with the down/menu button but it has a rather old build
20:46:16*lorenzo92 backlight and charger detection working on R1!
20:46:45xelanls seems like i can skip with the wheel
20:48:09n1sxela: skip untill the number on the left of the slash after "usfl" drops to 0 and it should rebuffer
20:49:53xelatried it
20:50:13xelanls it takes a little while until playback starts because the disk spins up
20:50:18xelabut it doesnt stutter
20:50:24xelaso the effect is different
20:50:46n1shmm, yeah a small dropout is expected when skipping like this
20:52:49xelaand not every song stutters
20:53:42n1sxela: in the buffer thread screen, is pcm almost full most of the time?
20:58:08xelanls: yes
20:58:37n1shmm, i'm out of ideas then
20:59:12xelai am not quite sure whether or not it happens only with vbr files
20:59:55n1swhat codec?
21:00:34n1sxela: and is this with a very recent build?
21:01:29xela(Buffer: 59.3MB)
21:01:35 Quit dv_ (Read error: Connection reset by peer)
21:02:41 Join dv_ [0] (
21:03:27n1sperhaps it's worth it to test with a current build, there were changes to enable ata dma for the ipod video among others today so that might help but it's just a wild guess
21:03:51xelanls: and to describe the stuttering: the playback starts for about 1-2 seconds, then a gap of maybe 0.5-1 second and then conitnues. sometimes up to three "stutters"
21:04:36xelanls: it used to work like a charm before i replaced the hd (but was an older build on the old hd). so my guess would have been a different timing/whatnot of the new hd
21:05:37n1sso only when starting playback?
21:06:15[Saint]Hum, yes...seems skipping int eh buffering thread view is broken.
21:07:12lorenzo92kugel: pff we will need to find a clean way to select the right io pin to start the radio, I mean, R0 and R1 have the same chip but different RST pin ^^
21:07:13n1sperhaps the new disk is slower and rockbox doesn't wait enough before starting but this would surprise me, might be worth trying current build still
21:07:39xela[Saint] I am able to skip using the wheel. but behaviour seems to be slightly different (since it doesnt stutter then)
21:08:00xelanls are there nightlies somewhere?
21:08:47[Saint]<< and >> are irrlevant, and skipping is done with the scrollwheel...
21:08:48n1sxela: apparently we call the current builds dev builds now, they are the very latest rev
21:08:54[Saint]how, ...obvious. :-S
21:09:12 Join SuperBrainAK [0] (
21:09:20n1s[Saint]: yeah it's on various unexpected buttons on all targets it seems :)
21:09:27pixelma[Saint]: well, this menu is for people who know what they do... it even says "keep out" ;)
21:10:02[Saint]skipping tracks with the volume control seems /very/ non-obvious to me ;)
21:10:17[Saint]Especially considering the fact that the << and >> buttons are free.
21:10:33xela[Saint] << is used to leave the view
21:10:44[Saint]Apparently I'll be looking at key maps today, favourite thing.
21:10:52n1s[Saint]: you have git access, don't you? :P
21:10:59[Saint]xela: menu+select is more consistent, imo.
21:11:10[Saint]n1s: Not direct, no.
21:11:39 Quit y4n (Quit: PÆNTS ØLF!)
21:11:43pixelmaI've been told that it was put on non-obvios buttons in this menu on purpose (long time ago though)
21:11:48[Saint]Although this isn't a plugin, the keymap could be greatly improved by allowing full playback control and using menu+select to exit the screen.
21:11:52n1sah, well left leaves the view on most targets it seems so perhaps that's why as this feature was probably added in later
21:11:56pixelmanon-obvious too
21:12:37[Saint]I say treat it like a plugin on the devices we can.
21:12:52[Saint]all scrollwheel targets should be able to have full playback control in this screen.
21:13:18n1si think there are even debug screens that can't be left at all on some targets :)
21:18:04xelanls: checked with latest dev build stuttering still there
21:20:46 Join eckoit [0] (~ryan@
21:22:25xela[Saint]: menu+select is ok, especiall if there is a hint in the view
21:22:38[Saint]There wouldn't be.
21:22:51[Saint]But you'll notice that many plugins on iPod exit this way.
21:22:59[Saint]its jsut that this screen isn't a plugin.
21:23:11[Saint]We can treat it as such, though.
21:28:40xelanls: i wonder how else i could debug/hunt down the stutter thing
21:29:09xelanls: is there something like an error log like if some ata stuff timed out of such?
21:29:41Marexanyone can help me with a STMP3780/MX233 junker ?
21:30:03Marexis it possible to dump the memory map of that device somehow ? pinmux and DRAM config would be probably emough
21:30:28MarexI want to give u-boot a stab on these, since we now have u-boot ported to mx233 as well
21:31:44 Join wodz [0] (
21:32:13wodzMarex: what you mean by 'dump memory map of the device'?
21:33:39MarexI mean dump the registers of the DRAM controller and the PINMUX controller
21:34:11wodzyou mean the values or what?
21:34:34lorenzo92Marex: ask pamaury ;) there is also the hwtool for that :)
21:34:44Marexnow that's the nick Im looking for ;-)
21:34:55Marexpamaury_: poke ?
21:34:58Marexlorenzo92: thanks
21:35:16wodzanyway there is the thing called hwemul in utils/imxtools
21:35:22lorenzo92Marex: you're welcome
21:35:42pamaury_Marex: yes ?
21:35:53 Nick pamaury_ is now known as pamaury (~quassel@rockbox/developer/pamaury)
21:36:07Marexpamaury: can I dump the DRAM controller and PINMUX controller register values somehow ?
21:36:28Marexpamaury: from the running firmware that is, I'd like to give u-boot a stab at this XFi3 device now that we have it ported to mx233
21:36:39Marexit'd give me a portable hacking rig
21:37:29pamauryI don't quite understand for what you want to use u-boot but hey. You could write a plugin to dump all the values but I'm not sure of the use of it
21:37:37Marexpamaury: the thing is, we have our own HW init altogether, so I need to dump the values from the running firmware, to be able to process them and put them into our HW init code
21:38:11Marexpamaury: actually −− is the bootloader in that device locked with some key? Or is the CPU pristine (with no signing key) ?
21:39:50 Join einhirn [0] (
21:39:52pamauryok, so for the dram value: we keep the OF init code (original firmware) does the low-level init of dram and then we only change them to change the frequency (I took them from linux). For the pinmux, well, it's the same thing: the OF init does the init pinmux cfg and we change it. On the X-Fi3 the device is locked with key "0"
21:39:53Marexpamaury: the plan here is I want to turn this device into portable hacking rig to be able to test at least some Linux drivers if I have no EVK around ...
21:40:14Marexpamaury: ah ok .. key 0 is stock key, right ?
21:40:25Marexpamaury: I ordered also fuze+, that one is locked or not ?
21:40:46pamaurysame thing, locked with key only containing zeroes
21:40:51Marexah, super :)
21:41:06 Join anxt [0] (~user@
21:41:14 Join einhirn_ [0] (
21:41:15Marexpamaury: ok, so is it possible somehow to dump the registers from the running firmware then ?
21:41:30Marexpamaury: I dont care if I have to reflash the device or whatever ;-)
21:41:56pamauryif you can run rockbox on it that's trivial yes
21:42:21pamaurythere is a simpler way however
21:42:23Marexpamaury: I didnt manage to compile one from latest git with emdebian 4.7.2-4 toolchain
21:42:34Marexpamaury: oh, please educate me :)
21:43:32pamaurywe use our own toolchain. I wrote a tool which allows one to poke at the registers remotely (from the PC). If you first run the OF init code and then that code, you can read the DRAM and PINMUX registers easily
21:43:51Marexpamaury: hm .. you mean the USB loader ?
21:44:14pamauryyes, you put the device in recovery mode send code by usb
21:44:16Marexfunny thing −− I wrote one as well by reverse engineering the USB protocol between MFGtool and MX28 :/
21:44:31 Quit einhirn (Ping timeout: 256 seconds)
21:44:33Marexwere I to know you had the same thing going, I'd just stick with yours :)
21:44:51pamauryI think the mx28 and mx233 use the same protocol
21:45:02[Saint]Nothing wrong with occasionally reinventing the wheel ;)
21:45:03Marexpamaury: I'm able to jumpstart mx233 chip as well, yep
21:45:05 Part eckoit
21:45:40Marexpamaury: ok ... so I need the init code (.sb ? bootstream?), load it into the device via USB ?
21:45:41pamauryanyway, this piece of code is in our repository, in utils/imxtools/hwemul. It would be better if you manage to compile it yourself. Otherwise I can send you a built one
21:45:52Marexlemme try
21:46:15pamauryyou might have to build the other directories under utils/imxtools/ before
21:47:02Marexok, hwemul_tool ready
21:47:37pamauryyou need to build the code in dev/, did you ?
21:48:19Marexah dang
21:48:25Marex/usr/lib/gcc/arm-linux-gnueabi/4.7/../../../../arm-linux-gnueabi/bin/ld: error: no memory region specified for loadable section `'
21:48:48Marexthis is my LD generating −−build-id ... it's some new stuff in recent binutils
21:49:02pamauryhum, ok, you'll need to modify the linker script file to drop it, you know how do to this ?
21:49:27pamauryin your case that's
21:49:49Marexyea ... I fought this before
21:50:04Marexis there any way to actually configure the cross toolchain prefix ?
21:50:36MarexI might just use tools/configure and use a different toolchain I have sitting around (might just be fastest)
21:50:55pamauryfor the main binary of rockbox I think so, for this little tool no, you'll have to edit the Makefile: the first lines
21:50:59MarexI have this arm-linux-gnueabi- toolchain here which has linker that doesn't generate this junk
21:51:13Marexpamaury: ok then, hang on a bit
21:51:53Marex../lib/hwemul.h:26:24: fatal error: hwemul_soc.h: No such file or directory
21:51:58Marexis this intended ? ;-)
21:52:06MarexI think you missed some file when commiting this stuff
21:52:49pamauryyou need to generate a register map
21:53:11pamauryI'm in the process of changing this but hey, you'll have to do it, let me remember how do to this ^^
21:53:54pamauryyou need to compile the lib/ before
21:54:06pamauryand before this the imxtools/regtools/ ^^
21:54:56Marexok, compiled, hwemul_tool ready
21:55:31pamauryso you have successfully compiled hwemul_tool and the dev/ dir ?
21:56:04***Saving seen data "./dancer.seen"
21:57:16xelanls, [Saint] thanks a lot so far, maybe more another time
21:58:03Marexpamaury: yes
21:58:08MarexReal key: 467cc254f81be8e78d765a2e63339fc9
21:58:10Marexis this right ?
21:58:28Marexah ... it's AES'd 0x00 key
21:58:50pamaurythe real key is random ^^ If you want i'll explain you the firmware format...later :)
21:58:58Marexthere's a document on that
21:59:01pamauryso first, we'll do a little test
21:59:06 Quit xela (Quit: Verlassend)
21:59:23pamauryMarex: yeah, at the time I first wrote the tools there was none ^^
21:59:53pamauryso, in dev/ you should have, can you send it to the device in recovery mode ?
21:59:54Marexheh :)
22:00:00Marexpamaury: let's try
22:00:04pamauryour loader is imxtools/sbloader
22:00:27pamaury(sbloader 0
22:00:35anxthi, i just installed rockbox via the gui tool, I have some dir ##MUSIC## and ##PORT# is this same to rm -rf ?
22:00:52pamaurythe device should appear as fee1:dead in lsusb in case of success
22:02:06MarexI have to hold vol-up, right ?
22:02:20pamauryon which device ?
22:02:52pamauryit's power
22:03:21pamaurya reliable way to trigger it: plug usb, hold power, reset the device
22:05:36Marexok, loaded
22:05:54pamaurydoes it appaear as feel1:dead in lsusb ?
22:06:06pamaurycool :)
22:06:10pamaurynow run hwemul_tool
22:06:51pamaurynow you have a kind of shell
22:07:02pamaury(type help)
22:07:16pamauryyou can read by address (read 0x10) or by name if you load a soc description
22:07:28Marexoh, so read32 is what I want to do now
22:07:30Marexnice :)
22:08:06Marexthe STMP3780 is about the same thing as MX23, right ?
22:08:33pamauryto do this do:
22:08:35pamaurysoc imx233
22:08:35pamauryread HW_PINCTRL_CTRL
22:08:40pamaurySTMP3780=MX233, same chip
22:09:14Marexpamaury: I'm ok with memory addresses
22:09:54pamaurynow I should warn you: the tool runs out of iram so currently the dram is not initialised
22:09:58Marexpamaury: is there not some kind of a difference between STMP3780 and MX233 ? I mean, the bootrom recovery code identifies as 066f:3780 on mx23 too, but I recall the bootrom on STMP3780 is older
22:10:21Marexpamaury: you mean SRAM ? ( stuff at 0x0 )
22:10:33pamaurythere are several versions of the bootrom mostly to fix bugs
22:10:41Marexpamaury: ok, how do I go about initing it with the original bootloader ?
22:10:44Marexpamaury: heh :)
22:11:01pamauryfirst download a firmware upgrade for the zen x-fi3 from creative
22:11:19Marexpamaury: FYI have you ever seen the bootrom sources ?
22:11:54pamaurybut i've disassembled it :)
22:12:19Marexdisassembly might just have been way easier to read
22:12:23pamaurydid you ?
22:12:52Marexpamaury: no
22:13:11pamauryit has a terrible architecture anyway, and has bugs
22:13:29Marexmx28 is good in that it's industrial design
22:13:40Marexand the bugs there were mostly ironed out by FSL guys
22:13:55Marexall right, firmware update downloaded
22:14:09pamauryyou'll need to run cabextract on it
22:14:10 Quit thomasjfox (Ping timeout: 245 seconds)
22:14:28Marexah good, it's this
22:14:50pamauryyou should get a file called it in, right ? :)
22:15:21Marexso zenfwupdater is some sbloader with a pretty face ?
22:15:34pamaurynow, run imxtools/sbtoelf -z -o ZENXFI3
22:15:38Marexall right, I have
22:16:25Marexok, got it
22:16:47pamaurywe have some zen* code but it's for older zens, this one is completely different. You should get files called ZENXFI3*. You want to keep the ZENXFI3.____.* ones, remove the others
22:17:11Marexok, I have it in /tmp anyway
22:17:27Marex14 megs of junk, nice
22:17:33pamauryyou should copy them in hwemul/dev/
22:17:58Marexok ?
22:18:00pamaurylet's assume you put them in hwemul/dev/tmp/
22:18:12pamaurynow edit hwemul.db (in hwemul/dev)
22:18:42pamauryin the sources section, add an entry for all the ZENXFI3.____. files except the last one (number 3 iirc)
22:19:00pamauryfor example: boot0 = "tmp/ZENXFI3.____.0.elf";
22:19:47Marex#4 is the last one
22:20:17Marexnow load them and jump them one after another ?
22:20:33pamaurycall them
22:20:48pamaurynot jump :)
22:20:51pamauryso load boot0; call boot0; load boot1; call boot1;
22:21:01 Join sakax [0] (
22:21:40pamauryand then you should be able to "make" again and get a brand new file which init the dram
22:22:08Marexyea ... you should add hwemul.db into Makefile dep list
22:22:25Marextransfer error at send step 30
22:22:27pamauryah, I was unsure it was, i'll do it then ^^
22:22:35pamauryhum, try again ?
22:23:02pamaurythen maybe I made an error, you should only load the first 3 ___.x.elf files instead of 4,
22:23:09pamaurylet me check
22:23:13amayer_[Saint]: if i hold menu on a classic while plugging into pc and the headphones output constant buzz and player freezes would that be related to g#379
22:23:43pamauryyeah, only load the 0, 1 and 2, not the 3
22:23:54Marexgot it
22:24:01amayer_bluebrother^: is fsbluebot down?
22:24:29pamauryMarex: and then run everything again: go into recovery mode, run sbloader, hwemul_tool and you're done !
22:24:59Marexyep, done ... and RAM seems to be online
22:25:06Marexat least reading it doesn't crash the chip
22:25:17Marexreadline support would be nice ;-)
22:25:19pamauryreading never crashes
22:25:39Marexpamaury: really ? if you read RAM which is dead, it'll jump into DABT handler
22:25:41pamauryyeah I know, that's really just a dev tool, I didn't put too much effort in it
22:25:58Marexno prob, I can relate to that at times
22:26:01pamauryMarex: my experience is that only a few memory ranges trigger an abort
22:26:05pamauryvery strange
22:26:17Marexpamaury: we use that in U-Boot to detect the DRAM size at runtime ;-)
22:26:36pamauryhum, then the dram is part of this region :)
22:26:56pamauryare you a uboot dev or just hacking uboot ?
22:27:28sakaxhello guys - how is the nano2g usb thingy going? :)
22:29:06wodzsakax: stalled
22:33:33wodzso on arm forth plugin takes 92560 bytes in mem + stacks (~8k I think)
22:38:55 Quit pamaury (Ping timeout: 256 seconds)
22:39:29*lorenzo92 nice 58v battery, seems I'm wrong with a variable type :D
22:52:04lorenzo92where do I configure battery ADC reading precision? right now I have 5mV but R1 fuel gauge is more precise 1.25...
22:52:30 Quit n1s (Quit: Ex-Chat)
22:52:39 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
22:54:47pamaurywodz: is that the size of the entire forth kernel ?
22:55:45wodzyes but this number is when compiled with -Os
22:55:57wodz-Os build on x86 doesn't run correctly
22:58:05wodzI also implemented just a few hooks to our pluginapi just to see if it is working
22:59:35 Join [Saint_] [0] (~quassel@
22:59:36wodz133352 is for arm with -O0 (-O0 works as simulator plugin)
22:59:54wodztomorrow I'll test on actual target
22:59:59 Nick [Saint_] is now known as Guest33140 (~quassel@
23:00:25[Saint]amayer_: I highly doubt it, but that patch is really quite would't surprise me if it caused any number of errors.
23:00:30 Quit lorenzo92 (Quit: ChatZilla 0.9.89 [Firefox 18.0.1/20130117041235])
23:01:35 Quit amayer_ (Quit: amayer_)
23:02:23[Saint]amayer_: you might want to try (if you haven't already) verifying it only happens with the patch applied, and if so, comment on the gerrit thread, and if so, filing a bug report.
23:02:46[Saint]errr...bum, I messed that up.
23:03:38[Saint]Check if it happens without the patch applied; if it does - file a bug report; if it doesn't - mention your findings in the gerrit thread for the task.
23:04:00[Saint]Aaaaaaaaaaaaannnnnnnd, he's gone.
23:04:04[Saint]Double bums.
23:04:57[Saint]Can someone please assist me with a simple pass/fail test regarding EQ presets?
23:05:05[Saint]Test procedure:
23:05:30[Saint]Verify EQ is enabled; apply an EQ preset, check if EQ is still enabled.
23:05:39[Saint]Expected result:
23:05:44[Saint]EQ remains enabled
23:12:27wodz g#388
23:15:49 Quit kevku (Ping timeout: 245 seconds)
23:16:24[Saint]wodz: nice :)
23:16:46[Saint](unfortunately, bluebot is on holiday though? :))
23:19:33pamaurysomeone on the fuze+ forums reported 29 hours of playback with eq and mp3 at 320Kbps, I didn't know that was *that* power consuming
23:33:12 Quit einhirn_ (Quit: Miranda IM! Smaller, Faster, Easier.
23:34:41 Quit Strife89 (Quit: Heading home.)
23:35:26 Join TheSphinX^ [0] (
23:38:36 Quit TheSphinX_ (Ping timeout: 252 seconds)
23:38:49 Quit wodz (Quit: Leaving)
23:42:31kugelpamaury: with rockbox?
23:42:36kugelis 29h good or bad?
23:43:03pamaurybetter than the OF (~20h) and worse than mine with basic mp3 (~39h)
23:43:17pamaury(~35 sorry)
23:43:32pamauryyes with rockbox
23:44:56 Join Tim_Elliott [0] (~Tim@
23:45:58 Quit Tim_Elliott (Client Quit)
23:46:19kugelthe OF is really this bad?
23:47:06 Quit SuperBrainAK (Ping timeout: 256 seconds)
23:48:30 Join TheSphin- [0] (
23:51:07 Quit TheSphinX^ (Ping timeout: 252 seconds)
23:56:07***Saving seen data "./dancer.seen"
23:57:52 Quit prof_wolfff (Ping timeout: 256 seconds)

Previous day | Next day