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 2013-11-21

00:00:48 Join Scall- [0] (
00:02:01mudlordhi, does the newest git version of rockbox require a bootloader update, when upgrading from 3.13?
00:02:37gevaertsWhich player?
00:02:41 Quit pamaury (Ping timeout: 240 seconds)
00:03:01mudlordsansa clip zip
00:03:07gevaertsThen no
00:03:15 Quit Scall- (Read error: Operation timed out)
00:11:52 Quit bertrik (Remote host closed the connection)
00:12:05 Quit ender` (Quit: The early bird may get the worm, but the second mouse gets the cheese.)
00:16:18 Join Scall| [0] (
00:17:56[Saint]Generally speaking, if you have to update the bootloader, we'll tell you about it.
00:18:20[Saint]Otherwise, there's pretty much no point in updating the bootloader.
00:20:21JdGordon[Saint]: pong!
00:22:45[Saint]JdGordon: what is supposed to happen with the skin engine (wrt: updating) when the LCD turns off?
00:23:25[Saint]I got some fairly different results with battery benching the WPS with hold on vs the main menu with hold on, making me suspect something was still updating in the background.
00:24:00[Saint](fallback WPS with peak meters {during playback} vs. fallback sbs in main menu)
00:25:31JdGordonumm, really straining my memory here
00:28:00[Saint]I had a brief look, but I got lost very quickly, and I didn;t want to assume that whatever the code said should happen was actually happening.
00:40:56*[Saint] seriously wonders if cooltwou is just having a giant laugh at our expense
00:58:14 Quit lebellium (Quit: ChatZilla [Firefox 26.0/20131118212339])
01:05:25 Join treaki_ [0] (
01:07:46 Quit treaki (Ping timeout: 252 seconds)
01:08:03 Join DuperMan [0] (
01:08:19DuperManw00t where' malauy
01:12:49[Saint]Lets try to A - use real, English words and B - stay on topic, please.
01:13:24JdGordon[Saint]: umm, i think depending on the target, some will/should update in the background and most should not
01:13:50JdGordonwhat results are you seeing?
01:14:09[Saint]JdGordon: I thought as much based solely on runtimes, it made a measurable difference.
01:14:18[Saint](around 2 hours less)
01:14:30JdGordonin the wps?
01:14:41JdGordonwhich target?
01:14:58[Saint]Yeah, wps vs. lists.
01:15:01[Saint]iPod Video.
01:15:28JdGordonpeak meters enabled anywhere?
01:15:41[Saint]Yes. fallback wps.
01:16:07[Saint]But I was thinking with hold on and the LCD disabled, that shouldn;t matter. Guess I was wrong.
01:16:32[Saint](hold isn;t necessarily relevant here, I just have my targets set to sleep the LCD immediately on hold)
01:17:56[Saint]I guess updating in the background is to avoid showing stale information (however briefly) when the targets wakes the LCD?
01:19:12JdGordonyeah, peakmeters are drawn (and refresh rate == 25HZ) any time the peak meter is neabled
01:19:17JdGordonregardless of lcd state
01:19:20JdGordonso thats a bug
01:19:42[Saint]Aha. Good to know.
01:19:47JdGordonim not sure the wps is updated even in this case
01:20:32[Saint]Well, that gives me more of a reason to remove the peak meters from the fallbac wps then I guess.
01:20:51[Saint]But, it would be nicer if it just stopped updating when the LCD was powered down.
01:21:15[Saint]Not /quite/ sure how to achieve that, I'll have a look this evening.
01:21:59[Saint]The runtime gave me the impression that something was pegging the CPU, nice to know I was on the right track.
01:22:02JdGordonshould be really easy
01:22:23JdGordonskin_wait_for_action(), in there check the lcd is awake
01:22:51JdGordonit *may* need to be forced to do a full update when the lcd is reactivated (though i have a feeling this is done automatically anyway)
01:23:01[Saint]Does that also need special handling for remotes, or should that "Just Work"?
01:23:25 Quit Strife89 (Ping timeout: 240 seconds)
01:23:55[Saint]And, yes. I believe you're correct, I seem to recall a fullscreen update being forced when waking the LCD.
01:26:32 Nick SuperBrainAk is now known as DormantBrain (~andy@2001:470:8:a61::5f92:59a1)
01:27:32 Quit Scall| (Ping timeout: 272 seconds)
01:27:49 Join Scall [0] (~chat@unaffiliated/scall)
01:29:33JdGordonyou'd need to do some benchmarking
01:29:57JdGordonI'm guessing the sleep(0) is what is actually killing the battery more than the skin drawing
01:30:26JdGordonthat function draws both displays, so you'd need to have it check if either are visible to keep the PM active
01:30:57[Saint]Hmmm. Possibly. But realistically, it makes absolutely zero sense to continue updating when the LCD is powered down.
01:31:09[Saint]And, yes.
01:31:11JdGordonI did a benchmark before the customsbs code and found there was no measurable difference between the skin engine and the regular lcd functions so drawing the menu screen
01:37:36***Saving seen data "./dancer.seen"
01:43:51DuperManI BUILD ZEN STUFSS
01:43:54DuperManHOO RAY
01:44:14DuperManall from other's code o'course
01:50:31Mode"#rockbox +o JdGordon" by ChanServ (ChanServ@services.)
01:50:53Kick(#rockbox DuperMan :#rockbox bye!) by JdGordon!cb1380e2@rockbox/developer/JdGordon
01:50:54 Join DuperMan [0] (
01:50:54 Quit fs-bluebot (Read error: Connection reset by peer)
01:51:17 Part JdGordon ("DuperMan bye!")
01:51:22 Join JdGordon [0] (cb1380e2@rockbox/developer/JdGordon)
02:04:15 Join Strife89 [0] (
02:10:05 Join cmhobbs [0] (~cmhobbs@fsf/member/cmhobbs)
02:14:18 Quit pixelma (Remote host closed the connection)
02:14:18 Quit amiconn (Remote host closed the connection)
02:15:38 Join pixelma [0] (pixelma@rockbox/staff/pixelma)
02:15:38 Join amiconn [0] (amiconn@rockbox/developer/amiconn)
02:38:54mudlordforgive me for this: but I thought floats were illegal
02:39:05mudlordI was wondering as I was wanting to do some graphics stuff using Pi
02:39:21mudlordfor example: a C64 style twister demo
02:44:02 Quit foolsh (Remote host closed the connection)
02:46:10 Join Zarggg [0] (
02:56:52 Quit cmhobbs (Ping timeout: 248 seconds)
03:08:48 Join amayer [0] (
03:10:10 Part mudlord
03:11:02 Quit funman (Ping timeout: 272 seconds)
03:12:07 Join funman [0] (
03:28:39saratogamudlord: libwmavoice is disabled
03:29:07saratogamt ported the decoder when he did wma pro, but didn't finish porting it to fixed point
03:29:46 Quit Scall (Ping timeout: 272 seconds)
03:30:52[Saint]He be gooooooone, yo.
03:37:40***Saving seen data "./dancer.seen"
03:39:00 Quit SrRaven (Ping timeout: 248 seconds)
03:42:10 Join SrRaven [0] (
03:44:48 Join Scall [0] (~chat@unaffiliated/scall)
03:51:14 Quit Scall (Ping timeout: 245 seconds)
03:53:46 Quit amayer (Quit: Leaving)
04:33:41 Quit pixelma (Disconnected by services)
04:33:42 Join pixelma_ [0] (pixelma@rockbox/staff/pixelma)
04:33:44 Nick pixelma_ is now known as pixelma (pixelma@rockbox/staff/pixelma)
04:33:44 Quit amiconn (Disconnected by services)
04:33:44 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
04:33:49 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
04:39:52 Join Scall- [0] (
04:44:07 Quit Scall- (Ping timeout: 252 seconds)
04:52:49 Nick DormantBrain is now known as SuperBrainAk (~andy@2001:470:8:a61::5f92:59a1)
05:08:13 Quit [7] (Disconnected by services)
05:08:26 Join TheSeven [0] (~quassel@rockbox/developer/TheSeven)
05:37:44***Saving seen data "./dancer.seen"
05:49:22 Join Scall [0] (~chat@unaffiliated/scall)
05:53:48 Quit Scall (Ping timeout: 246 seconds)
06:11:55 Join kugel [0] (
06:11:55 Quit kugel (Changing host)
06:11:55 Join kugel [0] (~kugel@rockbox/developer/kugel)
06:13:20 Quit kiwicam (Remote host closed the connection)
06:17:13 Quit kugel (Ping timeout: 245 seconds)
06:22:45 Join kiwicam [0] (~quassel@
06:27:58 Nick x56 is now known as everybody (
06:28:06 Nick everybody is now known as x56 (
06:39:52 Join Scall [0] (~chat@unaffiliated/scall)
06:44:43 Quit Scall (Ping timeout: 245 seconds)
07:00:10 Quit Strife89 (Ping timeout: 246 seconds)
07:16:37 Quit akaWolf (Ping timeout: 248 seconds)
07:17:55 Join akaWolf [0] (~akaWolf@unaffiliated/akawolf)
07:37:47***Saving seen data "./dancer.seen"
07:52:46 Join kevku [0] (~kevku@2001:470:27:773:0:feed:c0f:fee)
07:55:14 Join mortalis [0] (~kvirc@
08:22:50 Join wodz [0] (
08:32:55 Join wodz_ [0] (
08:32:55 Quit wodz (Read error: Connection reset by peer)
08:34:16 Quit funman (Changing host)
08:34:16 Join funman [0] (~fun@rockbox/developer/funman)
08:35:32 Join LinusN [0] (
08:44:33 Join ender` [0] (
08:49:54 Join bertrik [0] (~quassel@rockbox/developer/bertrik)
09:07:40 Quit kevku (Ping timeout: 245 seconds)
09:21:06 Join lebellium [0] (
09:21:16 Quit bertrik (Remote host closed the connection)
09:23:56 Join petur [0] (5bb7304d@rockbox/developer/petur)
09:23:59 Join Zagor [242] (~bjst@rockbox/developer/Zagor)
09:37:50***Saving seen data "./dancer.seen"
09:39:11 Nick pookie_ is now known as olspookishmagus (
09:53:52 Quit amiconn (Remote host closed the connection)
09:53:52 Quit pixelma (Read error: Connection reset by peer)
09:55:07 Join pixelma [0] (pixelma@rockbox/staff/pixelma)
09:55:07 Join amiconn [0] (amiconn@rockbox/developer/amiconn)
10:02:45 Join kevku [0] (~kevku@2a01:d0:ffff:34a::8:3)
10:14:55 Join maruk [0] (
10:22:28 Nick SuperBrainAk is now known as DormantBrain (~andy@2001:470:8:a61::5f92:59a1)
11:31:23 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
11:32:15copperpamaury: did you get my battery benchmark?
11:35:36pamaurycan you post the link again ? I didn't download the file
11:36:23copper17:36:53 UTC <copper> pamaury: fuze+ battery benchmark, LAME V0, -20 dB, earbuds attached:
11:37:53***Saving seen data "./dancer.seen"
11:37:58copperweren't you guys talking about 40h battery life for the Fuze+, a while ago?
11:38:10pamauryI reached 35h some while ago
11:38:25wodz_pamaury: The device doesn't respond to the repeated HWSTUB_INFO_TARGET request.
11:38:26pamaurysomeone reported 43h but that was without headphones so I'm not sure it counts
11:38:45pamauryI'm going to battery bench a few older revisions to see if there was a regression
11:39:01wodz_pamaury: I mean rk27xx hwstub
11:39:22pamaurywodz_: but it answers to different requests ? (target, version, ...)
11:39:48copperI'm current benching my Classic with a year old build, and at 46% right now, it's been running for 16 hours, versus 12 hours for yesterday's benchmark (with the latest build)
11:40:18coppereither the calibration is different, or there is indeed a battery life regression. We'll see when the benchmark completes.
11:41:56wodz_pamaury: I tried only this one so don't know
11:42:52wodz_pamaury: When hwstub_shell starts it sends HWSTUB_INFO_TARGET and get correct response.
11:43:15wodz_but sending it again after initialization doesn't work
11:43:31wodz_so the device part is locked for some reason
11:44:57pamaurythat's puzzling because the shell sends 4 requests on init: version, target, layout and features
11:45:11wodz_and all this requests are handled correctly
11:45:56pamaurydid you check with wireshark ? Maybe an invalid requests which get STALLED for whatever reason and the core is locked ?
11:46:21wodz_pamaury: the repeated request looks exactly the same
11:48:49wodz_pamaury: I vaguely remember the core being odd in handling some parts of the usb protocol. You made some changes to the usb core in rockbox to accomodate. Maybe this is related.
11:49:34pamauryI need to read your code and the code of the whole driver to remember
11:50:45pamauryhum, why is there no code in your usb_drv_init() ?
11:52:42wodz_in rb driver it configures only semaphores
11:53:29pamauryyeah but rockbox you control the whole state, here you should at leat soft disconnect and reconnect no ?
11:53:38pamaurydepends on how you run it of course
11:53:43pamaurybut in doubt
11:55:23 Join lorenzo92 [0] (~chatzilla@
11:56:25wodz_pamaury: Hmm you might be right. The core hard reset and soft reconnect is handled in usb connect interrupt BUT now I recall it didn't fire properly.
11:57:20pamaurywodz_: do you recall why we need "while (TX0BUF & TXFULL) /* TX0FULL flag */;" in ctr_write() ?
11:58:10lorenzo92pamaury: found the audio issue, it was the gate pin is not active high but active low ;)
11:58:29wodz_pamaury: not really
11:58:31pamaurylorenzo92: good :) easy fix
11:59:12lorenzo92pamaury: WTF I hear a horrible noise when touching the touch pad xD I think there are still some electrical issues haha
11:59:39lorenzo92only when FM radio is enable, so I think the mux is either not correct or somthing similar
11:59:43wodz_TX0FULL: Transmit Buffer status Indicate one packet has been written and available to be transmitted in the data registers, will be cleared by the UDC 2.0 for a successful IN transaction when receive ACK from UDC 2.0.
11:59:50wodz_quoting doc
12:03:42lorenzo92pamaury: and I actually don't hear any FM, chip is initialized and I see registers, but I'm unsure it is working hum
12:04:09pamaurylorenzo92: then how can you say sound working ?!
12:04:35lorenzo92pamaury: the trick is the keyclick :D
12:04:46lorenzo92awful but great while debugging
12:05:23lorenzo92pamaury: now I have a look at the schematics and see if tuner needs the internal clock or has an external one..
12:08:15pamaurywodz_: I'm puzzled by ctr_read()
12:08:30pamauryshouldn't the buf address by incremented *after* being written to the register ?
12:09:01wodz_let me see
12:09:21pamauryand no, you don't use ctr_read() for the first read
12:09:24 Join Scall- [0] (
12:09:24pamaury*ah no
12:09:43pamauryany reason for this ?
12:09:53lorenzo92pamaury: now I understand, I missed the XTAL configuration, let me try again
12:11:11pamaurywodz_: I seem to recall in hwstub the usb_drv_recv() function is supposed to be blocking
12:11:50wodz_pamaury: ctr_read() is called in irq. It is just helper function.
12:11:59wodz_pamaury: oh, that might be the case then
12:12:11pamaurywodz_: why don't you use it in usb_drv_recv() too ?
12:13:06pamauryanyway that doesn't example the bug, these transfers only send things, not receive
12:13:12wodz_Because of the sequence of buffer increment. in usb_drv_recv() you need to prime transfer in isr you need to point to the next block
12:13:36pamauryyou could increment buffer pointer after the write to the register ?!
12:13:52pamauryah no, you cannot control the receive size
12:13:59pamauryI forgot this core is dumb
12:14:24wodz_I can't remember the details but I was unable to find sane solution other then this
12:15:02lorenzo92dammit /* save binsize by only detecting presence for targets where it may be absent */
12:15:13lorenzo92I was sure that the chip was on in this way ^^
12:19:03pamaurywodz_: I don't see anything else wrong
12:19:04lorenzo92pamaury: got the problem. I confused the power pin vs the pin that tells me if radio is present or not ^^
12:19:39pamauryI'll try with the usb analyser tonight
13:02:30lorenzo92pamaury: fm radio works !! :)
13:03:02lorenzo92there is a bug in the tea chip driver, because it does not turn on radio before probing it, anyone here to explain why?
13:03:17wodz_lorenzo92: which target? Z5?
13:03:38lorenzo92wodz_: not only, all target using the chip
13:03:56pamaurylorenzo92: are you using HEAD ?
13:04:04pamaurybecause I fixed such a bug recently
13:04:14pamauryAnd I'm pretty sure it was in a tea driver
13:04:30lorenzo92I synced yesterday
13:04:42 Join ikeboy [0] (
13:05:11lorenzo92pamaury: HUGE bug in audio volume, at some point lowering it I got full output, you can imagine xD
13:05:23pamaurysee commit 0463bf4cb252a7719feec4d7fede0e960ed876ed
13:05:29pamauryit was tea5760uk driver
13:05:40lorenzo92I'm using the other variant
13:06:06pamauryit's a bug then
13:06:19lorenzo92exactly the same, I'll do a gerrit for it now
13:06:27lorenzo92or you do it?
13:07:17pamaurypush something to gerrit, I'll review it
13:16:22lorenzo92 g#670
13:17:09lorenzo92pamaury: I don't remember the command to yell a gerrit haha, anyways here you are
13:17:12ikeboydealfisher has a 4GB clip and 8GB microsd card for 26.99 if anyones interested here:
13:17:32ikeboyin dollars
13:17:55copperDoes anyone here NOT own a Clip+ already? ;-)
13:18:56ikeboyI don't
13:22:21 Quit ikeboy (Quit: ikeboy)
13:36:12 Join pretty_function [0] (~sigBART@
13:37:54***Saving seen data "./dancer.seen"
13:39:54 Join OrIWillChecklogs [0] (~random-na@
13:40:12 Quit OrIWillChecklogs (Client Quit)
13:41:05 Join OrIWillChecklogs [0] (~random-na@
13:41:41 Quit OrIWillChecklogs (Read error: Connection reset by peer)
13:58:36 Join krabador [0] (~krabador_@unaffiliated/krabador)
14:24:25pamaurylebellium: " g#670" should print the details but the bot is dead apparently
14:25:05lebelliumyeah that's what we concluded with lorenzo92 :)
14:26:20pamauryoops I reviewed but did not publish
14:27:09 Join cmhobbs [0] (~cmhobbs@fsf/member/cmhobbs)
14:31:20 Quit wodz_ (Quit: Leaving)
14:32:39lorenzo92pamaury: thanks :)
14:37:09 Quit krabador (Quit: Sto andando via)
14:42:29 Quit mortalis (Ping timeout: 245 seconds)
14:42:36 Join dfkt [0] (OxO29A@unaffiliated/dfkt)
14:52:34 Join amayer [0] (
15:00:37 Quit cmhobbs (Ping timeout: 248 seconds)
15:08:30 Join Strife89 [0] (
15:18:43 Quit kevku (Ping timeout: 260 seconds)
15:31:40 Join treaki__ [0] (
15:33:27 Quit treaki_ (Ping timeout: 264 seconds)
15:37:58***Saving seen data "./dancer.seen"
15:39:36 Quit Scall- (Quit: Bye bye)
15:41:46 Join hashbang [0] (
15:41:51hashbanghiya there
15:45:24 Join Scall [0] (~chat@unaffiliated/scall)
15:45:47pamauryhashbang: if you have a question, please ask and if someone know he/she will answer
15:46:38hashbangI've just upgraded from an old version of RB to 3.12. Will 3.13 work on the h3xx, or does the 10 band advanced EQ require a faster CPU?
15:48:03gevaerts3.13 will work just fine
15:48:26gevaertsI don't know if the h300 can handle ten bands, but then you don't have to use them all
15:48:51gevaertsIn fact, usually if you need more than four or five, you're probably misunderstanding how a parametric EQ works
15:49:24hashbanggevaerts: I'd probably only use the presets - but they seem to use all the EQ bands
15:49:41hashbanggevaerts: or am I mis-reading the preset files?
15:50:10coppergevaerts: can you expand on that?
15:56:02gevaertshashbang: you aren't. Those presets (i.e. the ones in 3.13) were a mistake, and current builds don't include them any more.
15:56:29gevaertsSo if I were you, I'd save the old (3.12) ones somewhere and put them back after installing 3.13
15:57:08gevaertscopper: see e.g.;a=commit;h=d475dd36a3702231fc76ef4dfdf771acbc730423
15:57:20hashbanggevaerts: aha, that makes sense. I understand the extra bands only use CPU if they're actually enabled, so 3.13 with 3.12 presets will use same CPU as 3.12 by itself.
15:58:19coppergevaerts: I don't understand what's wrong with using 10 bands with different values?
15:58:43gevaertscopper: nothing's *wrong* with doing that, it's just probably a waste of CPU
15:59:08gevaertsWith a parametric EQ you can adjust the width of a band
15:59:41copperI don't see the relation between the width of 5 bands, and using 10 bands with refined tuning
16:00:50gevaertscopper: most people who claimed they wanted more than five bands really meant they wanted to be able to copy the config from some non-parametric system with fixed bands
16:01:04copperthat would be my case, yes
16:01:19gevaertsAnd while you *can* do that now, you
16:01:33copperI guess I don't understand what "parametric" and "band width" means
16:02:03gevaerts're really better off studying a bit and making a simpler config, if you care about battery life (and therefotre CPU usage)
16:02:36*gevaerts isn't a specialist either
16:03:19coppergah, this needs a converter
16:04:07gevaertscopper: look at e.g. (only one side for now, let's ignore the stereo bit). That's a non-parametric EQ, so you need lots and lots of sliders. For the adjustment shown on that photo, a parametric EQ probably only needs one band
16:04:31gevaertsAnd yes, we could really use a graphical representation of what the eq is set to
16:04:35copperyes because that's single curve
16:06:03copperwith few data points
16:06:08hashbang # the EQ presets on gerrit that need some work?
16:06:21copperparameters, whatever it's called
16:07:26gevaertshashbang: I believe that that's the one that went into 3.13 and was removed again later
16:07:59gevaertsOr at least, a related one
16:08:01*gevaerts isn't sure
16:08:21gevaertsAh, yes. A related one
16:09:50*gevaerts won't go into the issue of whether the entire concept of "genre"-specific EQ presets makes any sense at all :)
16:10:22hashbanggevaerts: probably ought to be based on output device (e.g. headphone model) + bass/treble
16:10:29copperthat would be like 2 parametric bands, right?
16:10:39hashbanggevaerts: so use EQ to get flat response, then use bass and treble to taste
16:10:59gevaertshashbang: that's what I'd say, yes. Not everyone agrees though
16:11:27gevaertscopper: I think so, yes
16:11:35pamaurycopper: what is the audio player on this capture ?
16:11:35copperkay, I'll try to figure it out
16:12:25pamauryah yeah, heard about it some time ago, some people told me that it is very good, is it the case ?
16:12:40copperlove it
16:14:01copperpamaury: I particularly like internal and conditional tags
16:14:14copperand regex search capabilities
16:14:37 Quit pretty_function (Remote host closed the connection)
16:15:04 Join pretty_function [0] (~sigBART@
16:17:15 Quit Strife89 (Ping timeout: 264 seconds)
16:17:28 Quit Zagor (Quit: Clint excited)
16:19:39 Quit pretty_function (Ping timeout: 264 seconds)
16:28:14hashbanggevaerts: thanks for the clarification and the tip to restore the 3.12 eq presets. :-)
16:28:59 Join kevku [0] (~kevku@2001:470:27:773:0:feed:c0f:fee)
17:00:17 Quit Zarggg (Quit: Zarggg)
17:00:54pamaurycopper: is your Fuze+ available for another benchmark ?
17:02:21pamaurycan you do a benchmark for a particular revision ?
17:04:37pamauryok, let me have a look, currently I'm benchmarking a commit as of january
17:07:26pamaurycopper: this one c4f2a46e0dfee336ce7016cd62608097f15367b8
17:07:56copperthat commit sounds rather familiar :P
17:08:18coppersame codec and volume?
17:08:30copperLAME V0, -20 dB?
17:10:02copperok, will let you know
17:10:14pamauryyeah I thought I would spare you the 0.1dB volume bug ^^
17:15:37 Join krabador [0] (~krabador_@unaffiliated/krabador)
17:16:39 Join Strife89 [0] (~Strife89@2602:306:250e:659:d559:5f8e:7e2:3b71)
17:18:23 Quit mc2739 (Ping timeout: 246 seconds)
17:18:39 Join chrisjj [0] (561bb732@gateway/web/freenode/ip.
17:20:31 Join mc2739 [0] (~mc2739@rockbox/developer/mc2739)
17:25:52copperI hope you'll find something: those battery benchmarks are slow
17:26:25copper(and possibly taxing for the battery?)
17:26:43 Join Narod [0] (
17:27:34saratogahashbang: the problem with the old presets is that they don't make a lot of sense for how our EQ works, and they are very inefficient
17:28:47 Quit Narod (Client Quit)
17:29:14 Join Narod [0] (
17:30:15pamaurycopper: I don't think a benchmark or two will damage your battery
17:30:23pamauryjust don't leave it in a very discharged state
17:30:46saratogayeah we should be cutting off the battery before it really gets that low
17:32:04pamauryit shut down at 3.3V
17:35:54 Quit petur (Quit: Page closed)
17:37:59***Saving seen data "./dancer.seen"
17:57:48 Quit hashbang (Quit: Leaving.)
18:03:41 Join pretty_function [0] (~sigBART@
18:07:46 Join fs-bluebot [0] (
18:07:46fs-bluebotBuild Server message: New build round started. Revision adc5033, 243 builds, 36 clients.
18:08:40bluebrotheronly that that round has been finished long ago ...
18:08:46*bluebrother needs to fix the bot
18:10:52lorenzo92...otherwise everyone feels like don't remembering how to ^^
18:11:17bluebrothernever thought that bot would live that long :)
18:12:15 Quit krabador (Quit: Sto andando via)
18:17:00lorenzo92pamaury: refactor done! i'm going to gerritize it soon
18:21:39 Join bertrik [0] (
18:21:40 Quit bertrik (Changing host)
18:21:40 Join bertrik [0] (~quassel@rockbox/developer/bertrik)
18:31:39 Part maruk
18:37:09lorenzo92pamaury: gerritized ;)
18:37:44 Join rela [0] (~x@pdpc/supporter/active/rela)
18:38:07 Part LinusN
18:39:26pamaurylorenzo92: why does your patch reports a diff to powermgmt-imx233.c with empty diff ?
18:40:13pamaurylorenzo92: you declare your target as having HOLD but you didn't put any value in the table for hold
18:40:55pamauryand the {0, IMX233_BUTTON_LRADC_END} is incorrect, it should be the default value of lradc when there is no button (probably 3300 or something)
18:41:16pamauryand hold is probably 0 so add {0, IMX233_BUTTON_LRADC_HOLD} at the beginning of the table
18:42:18pamauryalso why do you declare fmradio i2c as harware but set sw defines ? is it software or hardware ?!
18:45:41 Join foolsh [0] (
18:47:06lorenzo92pamaury: for the diff, no idea, I'll look into it
18:47:18lorenzo92for the hold button hum, my hold button is gpio based
18:47:24lorenzo92not lradc based
18:47:40pamauryah sorry, I misread
18:47:50pamaurymy comment still apply to the last line of the table though
18:48:29lorenzo92pamaury: for radio, indeed, I placed the pins only for testing :)
18:53:55 Quit linuxguy3 (Ping timeout: 245 seconds)
19:27:41coppernote to self: it's easier to set positive gain in the EQ and compensate with adequate precut
19:38:02***Saving seen data "./dancer.seen"
19:45:21coppergevaerts: alright, I think I managed to convert my desktop EQ with just the low and high shelf filters :)
19:45:56copperit's a rather simple V shaped EQ anyway
19:46:01 Quit Narod ()
19:49:29 Quit pretty_function (Remote host closed the connection)
19:49:56 Join pretty_function [0] (~sigBART@
19:54:29 Quit pretty_function (Ping timeout: 246 seconds)
20:04:10copperyup, sounds like my desktop \o/
20:29:08 Quit pamaury (Ping timeout: 248 seconds)
20:37:05 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
20:49:58 Join einhirn [0] (
20:50:03 Quit einhirn (Client Quit)
21:03:01 Quit Strife89 (Ping timeout: 240 seconds)
21:09:14 Join ikeboy [0] (
21:16:13 Join Strife89 [0] (~Strife89@2602:306:250e:659:f9bb:d536:26f4:b5ca)
21:21:51 Quit krnlyng (Ping timeout: 246 seconds)
21:23:59 Join shai [0] (
21:24:27 Quit shai (Read error: Connection reset by peer)
21:33:23 Quit rela (Read error: Connection reset by peer)
21:38:06***Saving seen data "./dancer.seen"
21:40:34 Quit gelraen (Read error: Connection reset by peer)
21:40:42 Join gelraen [0] (
22:04:36 Quit fs-bluebot (Ping timeout: 248 seconds)
22:05:15 Quit bluebrother (Ping timeout: 246 seconds)
22:06:25 Quit ikeboy (Quit: ikeboy)
22:07:13 Join bluebrother [0] (~dom@rockbox/developer/bluebrother)
22:10:09 Quit lorenzo92 (Remote host closed the connection)
22:11:15 Nick DormantBrain is now known as SuperBrainAk (~andy@2001:470:8:a61::5f92:59a1)
22:11:37 Join wodz [0] (
22:29:27wodzpamaury: ping
22:30:29 Join kugel [0] (~kugel@rockbox/developer/kugel)
22:33:42 Join krnlyng [0] (~liar@
22:39:33CtcpPing from gevaerts!~fg@rockbox/developer/gevaerts
22:43:25 Quit Strife89 (Ping timeout: 240 seconds)
22:49:34 Quit mc2739 (Quit: leaving)
22:52:44pamaurywodz: pong
22:54:03wodzpamaury: What I found interesting is that according to wireshark the last request send to device before things go wrong is SET_FEATURE
22:54:52wodzthis set feature seems to ask to enter test_mode
22:57:17pamauryhum, very strange
22:57:27pamauryare you sure it is sent the device ?
22:57:43pamaury*to the
22:58:22pamauryhave you tried to run the code with valgrind ? maybe there is a memory corruption happening
22:59:39wodzpamaury: what is the meaning of 'Protocol' field in wireshark?
22:59:49wodzI mean USB vs USBHUB
23:01:12pamauryusbhub means the requests was sent to a hub
23:01:46pamauryusually set_features is used on hub to reset devices for example
23:02:02wodzok so this is SET_FEATURE is irrelevant
23:02:17pamauryprobably except if it sent to this device specifically
23:02:18wodzanyway the device simply stops responding
23:03:08pamaurywhat is the lates requests sent ? get_version ?
23:03:28pamaurysorry get target ?
23:04:16wodzsucessfull or failing?
23:05:47pamauryI'm asking this because the HWSTUB_INFO_TARGET is special: its buffer stage is 64-bytes exactly, that's one packet
23:05:57pamauryso maybe the driver is getting confused in some way
23:06:21pamaurymaybe the core doesn't notify the completion of the ZLP packet for example
23:06:35pamauryand the code gets stuck
23:06:59 Quit bertrik (Remote host closed the connection)
23:07:31wodzhow to change request size so it will be 63 for example?
23:08:08pamauryin hwstub_protocol.h, edit the usb_resp_info_target_t structure
23:08:14pamaurybe sure to recompile both the stub and the tool
23:08:20pamauryusing make clean
23:08:29pamaury(my makefile doesn't track dependencies correctly)
23:12:41wodzyeah that is the problem
23:13:00wodzI reduced name[] to 40 and now it doesn't hang
23:13:11wodzpamaury: ^
23:14:05pamauryok, now we know what is the problem
23:14:42pamaurydoes the core notify completion for ZLP ?
23:14:48wodzhow to print value of some mem cell? print(%x, DEV.read32(address)) ?
23:14:59[Saint]copper: I see you, like myself, got rapidly educated wrt: parametric equalizers :)
23:15:18pamauryfor formatting, print(string.format("%x",DEV.read32(addr)))
23:16:45wodzok reading mem works as well
23:17:05pamaurywe need to fix this bug
23:17:09wodznow I need to understand what is the problem
23:17:18[Saint]Ms [Saint] and her workmates insist on using g#401 - but it really is for people with more CPU than sense whom can't be bothered or don't care to learn about the merits of a parametric EQ.
23:17:21 Join cmhobbs [0] (~cmhobbs@fsf/member/cmhobbs)
23:18:13[Saint]...and that's not necessarily a bad thing, I just didn't particularly care to push for those EQ presets because it allows people to do very stupid things on targets that almost certainly can't handle it.
23:18:18pamaurywodz: my theory is that you don't get an interrupt when the ZLP completes, but that's just a theory
23:18:26[Saint]ANd then they'll complain and say its my/our fault.
23:18:29 Nick SuperBrainAk is now known as DormantBrain (~andy@2001:470:8:a61::5f92:59a1)
23:18:32pamauryperhaps because of dma
23:19:09[Saint]Did fs-bluebot die?
23:19:09wodzpamaury: With 64bytes payload we should issue ZLP, righ?
23:21:21copper[Saint]: it's just what usually happens with me: I went from not caring and using dumb means of doing something, to caring and actually thinking of a smarter way of doing the same thing :)
23:21:42[Saint]Pretty much the same with me.
23:21:45coppersometimes I simply forget to use my brain
23:22:27wodzpamaury: But we send ZLP in other occasions sucessfully, don't we?
23:23:01 Quit saratoga (Quit: Page closed)
23:23:27[Saint]I really should have a look at converting those presets to using as little bands as possible, you could assist if you'd like (nudge nudge... :))?
23:23:42pamaurywodz: maybe not, it's quite rare to send a multiple of the packet size on a control endpoint
23:23:58copper[Saint]: you know, I thought about that a couple hours ago
23:24:23pamaurythe Rockchip USB code you sent me suggest there is no difference for the ZLP, so it should trigger the IN0 interrupt normally
23:24:29copperit's a somewhat lengthy process to RMAA settings though
23:24:57[Saint]copper: if you put you mind to it, it should be fairly obvious where those were stole...errr, borrowed from. :)
23:25:06copperI know, I remember :)
23:25:43[Saint]Perhaps you could brief me on the process you used? I have had some difficulty with this myself.
23:26:02wodzpamaury: no I mean do we force ZLP in code somewhere so we are sure the driver is able to produce such thing
23:26:38[Saint]copper: if you have the time, I would very much appreciate a rough outline of how you achieved this.
23:26:41pamaurywodz: may I suggest you try to ack in rockbox ? modify our usbstack so that for example a string descriptor is exactly 64-bytes and see what happens
23:27:46pamaurywodz: yes, in receive controls but your driver has not receive any such control yet
23:27:53copper[Saint]: identify peaks, their center frequency and their amplitude; tinker with either the Q setting or the center frequency until you get the curve that you want, as verified by running RMAA with the EQ on.
23:28:06wodzas this driver is cop/paste and stripped from anything but EP0 I don't see the point
23:28:25copperI noticed that in order to get the curve that I wanted, I had to change the center frequency by as much as 2 kHz in the mids
23:28:31pamaurywodz: the point is to check if that work in rockbox !
23:28:41pamauryand in rockbox you have logf...
23:29:00gevaertscopper: probably in part because of the non-linear frequency scale used on your other eq
23:29:24coppergevaerts: my desktop player EQ is actually faulty
23:29:44copperI get 8 dB variations with 6 dB settings
23:29:47 Quit amayer (Quit: Leaving)
23:30:08copperwith super weird looking curves
23:30:30copperwhereas the Rockbox EQ performs as expected
23:30:39copper(again, as verified with RMAA)
23:32:10coppercreating Rockbox EQ presets to copy existing (iTunes or whatever) presets would be a lengthy process
23:32:31[Saint]Ah, another thing I wanted to ask the various sound-understanding-type-guys here - is there supposed to be a recipe for calculating the amount of replaygain pre-amplification to use?
23:32:34copperand one would wonder, whether those existing presets have merit to begin with
23:33:05[Saint]I averaged the total gain applied and divided by two, which seems ok, but I basically pulled it out of my ass.
23:33:09copper[Saint]: negative pre-amp should match your highest positive replaygain value
23:33:28copperin order to avoid clipping
23:33:56copperi.e. if you have an album with +3 dB replaygain, you should set the pre-amp to -3 dB
23:34:12copperthat way, your music collection will never clip
23:34:15[Saint]~90% of my replaygain values are negative.
23:34:21copperyup, that's normal
23:34:40copperbut positive gain values exist
23:34:48copperlike, with classical music, or albums from the 80s
23:35:14copperyou can also set the preamp value to match different reference levels
23:35:17copperRG is 89dB
23:35:25copperthe new ITU stuff is 84dB
23:35:35copperthat's what Opus uses, for instance
23:35:56copperi.e. you have to add 5dB to Opus gain in order to match the Replaygain reference level
23:36:35[Saint]The problem I had was it just being ridiculously quiet. I averaged the gain applied accross all tracks and divided by two (which came out at 6dB after a tiny amount of rounding) and applied that as positive pre-amp and it seems ok I guess.
23:36:53[Saint]But I'm never really comfortable with just pulling things out of the sky like that.
23:37:19copperer, your average Replaygain was -12 dB?
23:37:49copperlots of contemporary releases
23:38:03[Saint]Lots of...lots of everything. :)
23:38:08copperwell, you went for the "will work in most cases" way
23:38:10***Saving seen data "./dancer.seen"
23:38:37copperwhich is different from the "will work with absolutely anything" way ;-)
23:39:00[Saint]...which would be?
23:39:05copperI don't need to add gain because all my headphones and IEMs are sensitive enough
23:39:32copper[Saint]: which would be, take the highest Replaygain value you've got in your music collection, and apply preamp gain that matches it
23:39:45[Saint]My IEMs are plenty sensitive - I just like my music _loud_.
23:39:51copperlike I said: if your max gain is +3dB, set preamp to -3dB
23:40:04copperI like it loud too :)
23:40:24[Saint]Your definition of loud and mine might differ greatly, though.
23:40:49[Saint]My volume with IEMs in usually sits around -6dB.
23:40:54copperwell, I've been accused of being completely deaf a lot of times ;)
23:41:12copperanyway, you got the idea
23:41:23 Join petur [0] (~petur@rockbox/developer/petur)
23:41:30wodzpamaury: Yeah, it seems ZLP does not generate ACK interrupt
23:41:43[Saint]No, no. Past tense is fine. I understand now. Thank you.
23:42:18copperbasically, the objective is to avoid applying positive gain to a file that might already have peaks that reach 0 dBFS
23:42:34copperwhich can very well happen with Replaygain
23:42:44pamaurywodz: I'm surprised, but hey, might be just another quirk
23:42:46copperwith high dynamic range content
23:43:24copperReplaygain will try to make it sound as loud as the usual, modern, compressed recordings, but when the material has super high dynamic range, the peaks will then exceed 0 dBFS
23:43:33coppere.g. with classical music
23:44:06copperand then, if you don't apply negative preamp gain, the material will clip
23:44:30copperbut given today's music collections for most people, that's a fringe case
23:44:59[Saint]unless you're gevaerts
23:45:20copperI guess that's why the newer loudness standards use a lower reference level (84 dB vs. 89 dB): it leaves more headroom
23:46:05copperbut then, you'd need DAPs with more max voltage / power to compensate
23:46:10 Nick rocketni1e is now known as rocketnine (~tee@
23:46:59copperattenuating by 6 dB for replaygain and an additional 6 dB for EQing is already pushing it
23:47:25copperadd 6dB more for modern recordings, and you're already attenuating your source by a full 18 dB
23:48:53copperadd 5 dB to match the newer ITU reference level, and that makes a grand total of 23 dB
23:49:14copper(which is quite ridiculous)
23:50:07copperadd to that the French / EU volume cap, and your iPod becomes unusable ;P
23:50:43copper(a volume cap that is completely arbitrary and does not take into account the various impedance and sensitivity values of various ear buds / IEMs / headphones)
23:51:14[Saint]None of my material clips. Primarily, I just wanted to know if I was "Doing It Right ™", and I guess its pleasing to know that I wasn't totally "Doing It Wrong ™" but given this new information I will make some changes.
23:51:58copperyour material could clip, but for such a small amount of contiguous samples, that it might well be inaudible
23:52:22 Join GLC- [0] (
23:52:29GLC-good day
23:53:37GLC-i recently got my iPod 5G back after a few years.. Apple has probably released firmware updates to it
23:54:03[Saint]I sincerely doubt it.
23:54:12GLC-but I really dislike iTunes, so I was thinking of Rockboxing it
23:54:38GLC-it was like 2006 when i last time upped the firm
23:55:05GLC-and the person who had it hasn't even synced music to it, she listened mine
23:55:43wodzpamaury: seems like the workaround works
23:55:44GLC-does RockBox work with WMP12? I have Win 7 pro x64
23:55:52copper[Saint]: the thing with peak values is, you can very well have very marginal peaks that are a full 3 dB (or more) above 99% of the rest of the content
23:55:56pamaurywodz: this usb core is really fucked up ^^
23:56:18wodzpamaury: OF handles ZLP case specially too
23:56:19GLC-could I sync Rockboxed Ipod 5G with WMP12?
23:56:21copperin that case, clipping on those peaks might not matter
23:56:33[Saint]GLC-: It should. For all intents and purposes the device should simply appear as a removable mass storage device.
23:57:09[Saint]Most people prefer to simply drag and drop their audio with Rock box though.
23:57:31GLC-[Saint], ok, that's good to hear
23:57:54[Saint]The choices you make for getting audio to the device really depend on whether or not you are you going to be continuing to use the original firmware or not.
23:58:27GLC-drag and drop? so there's no folder structure in the Ipod then? or does Rockbox auto-create it? but how does then WMP12 could sync the music to the rockbox system?
23:58:44copperthe file structure is what you make it
23:58:56copperthe iPod is just like a normal USB flash drive

Previous day | Next day