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).

Notice: Only Gecko based browsers prior to FF4 support the multipart/mixed "server push" method used by this log reader to auto-update. Since you do not appear to use such a browser, this page will simply show the current log, and not automatically update.

#rockbox log for 2013-07-28

00:01:02 Join Guest11127 [0] (~Slayer@c-69-143-178-62.hsd1.va.comcast.net)
00:18:35***Saving seen data "./dancer.seen"
00:18:48 Quit Rower (Quit: soer)
00:19:52 Quit olspookishmagus (Quit: All for nothing)
00:27:22 Quit pamaury (Ping timeout: 240 seconds)
00:35:18 Nick SuperBrainAK is now known as DormantBrain (~andy@shared02.balt01.cd.2g2u.net)
00:52:13 Quit liar (Ping timeout: 240 seconds)
01:00
01:07:43 Quit theunleet (Quit: CGI:IRC)
01:14:20 Join syscrash [0] (~syscrash@poipu/developer/syscrash)
01:15:23 Quit jlbiasini (Remote host closed the connection)
01:31:31 Quit mirak (Quit: Ex-Chat)
01:42:32 Join mirak [0] (~mirak@static-176-182-46-92.ncc.abo.bbox.fr)
02:00
02:09:22 Quit bertrik (Ping timeout: 240 seconds)
02:18:39***Saving seen data "./dancer.seen"
03:00
03:15:35 Quit ender` (Quit: It's not worth doing something unless someone, somewhere, would much rather you weren't doing it.-- Terry Pratchett)
03:40:56 Join Beta2K [0] (~Beta2K@d24-36-136-246.home1.cgocable.net)
03:43:50 Quit lebellium (Quit: ChatZilla 0.9.90.1 [Firefox 23.0/20130722172257])
03:59:53 Join mrtux_ [0] (~colin@75-167-98-77.desm.qwest.net)
04:00
04:01:59 Quit mrtux (Ping timeout: 264 seconds)
04:01:59 Nick mrtux_ is now known as mrtux (~colin@75-167-98-77.desm.qwest.net)
04:18:42***Saving seen data "./dancer.seen"
04:29:46 Quit amiconn (Disconnected by services)
04:29:48 Join amiconn_ [0] (amiconn@rockbox/developer/amiconn)
04:29:52 Nick amiconn_ is now known as amiconn (amiconn@rockbox/developer/amiconn)
04:30:41 Quit pixelma (Disconnected by services)
04:30:42 Join pixelma_ [0] (pixelma@rockbox/staff/pixelma)
04:30:44 Nick pixelma_ is now known as pixelma (pixelma@rockbox/staff/pixelma)
04:41:33 Quit mrtux (Quit: ZNC - http://znc.in)
04:42:16 Join mrtux [0] (~colin@2001:388:f000::26cd)
04:43:14 Quit mrtux (Changing host)
04:43:14 Join mrtux [0] (~colin@unaffiliated/mrtux)
04:48:34 Join Strife89 [0] (~Strife89@adsl-98-80-231-240.mcn.bellsouth.net)
05:00
05:13:29 Join shamus [0] (~shmaus@ip-206-192-195-49.marylandheights.ip.cablemo.net)
05:46:37 Quit [7] (Disconnected by services)
05:46:50 Join TheSeven [0] (~quassel@rockbox/developer/TheSeven)
06:00
06:18:44***Saving seen data "./dancer.seen"
07:00
07:06:37 Quit CaptainKewl (Read error: Connection reset by peer)
07:09:36 Join kevku [0] (~kevku@2001:0:c38c:c38c:3c54:79ba:3d69:bedb)
08:00
08:03:25*[Saint] notes that the sample rate selector is basically unusable.
08:05:02[Saint]I assume there isn't *supposed* to be an audible difference here?
08:07:13[Saint]...that magically fixes itself.
08:07:16[Saint]Hummm.
08:10:35 Quit Strife89 (Ping timeout: 256 seconds)
08:18:48***Saving seen data "./dancer.seen"
08:20:56 Join Rower [0] (husvagn@v-413-alfarv-177.bitnet.nu)
08:27:45 Nick syscrash is now known as jvd (~syscrash@poipu/developer/syscrash)
09:00
09:36:43 Quit bluebrother (Disconnected by services)
09:36:48 Join bluebrother^ [0] (~dom@rockbox/developer/bluebrother)
09:58:57 Quit shamus (Read error: Connection reset by peer)
10:00
10:00:02 Join shamus [0] (~shmaus@ip-206-192-195-49.marylandheights.ip.cablemo.net)
10:15:01 Join bertrik [0] (~quassel@rockbox/developer/bertrik)
10:18:51***Saving seen data "./dancer.seen"
10:36:39 Quit rdn (Remote host closed the connection)
10:41:00coppersaratoga: well shit, the log is empty
10:42:38copperand the battery is full
10:42:50copperI saw it running though
10:42:53copperdunno what happened
10:43:18[Saint]I have seen this happen before now.
10:43:30[Saint]It also REALLY chokes on the insane APE.
10:43:47[Saint]if you listen to it, its just nasty schreeching and noise.
10:43:52copperI removed all ape above c1000, lets see if that fixes it
10:48:45copper[Saint]: test_codec decodes as fast as possible, right?
10:49:06[Saint]there's no way to do otherwise.
10:49:19[Saint]so "yes".
10:50:34[Saint]what is the ultimate question behind that one?
10:50:55copperI was wondering if instead, it would decode at real time, and measure the CPU frequency or something
10:51:02copperin real time*
10:51:39[Saint]No. There's no need. As it is trivial to ascertain realtime requirements from the decoding speed.
10:52:06[Saint]that would just be a slower, less efficient way of doing what it already does.
10:52:12copperlast night when I launched it, it put the Fuze+ face down, maybe the abord key was triggered, or something
10:52:22copperabort
10:52:48[Saint]If so, there should still be logs for everything it completed.
10:53:05[Saint]unless it cancelled before a single test finished.
10:53:07copperdunno what happened then
10:53:29copperright now it's running, and the display is turning on every time it tests a new file, I think
10:58:30coppermeh
10:58:36copperno opus files in the test corpus
11:00
11:13:54 Join jlbiasini [0] (~metaphysi@89.136.132.239)
11:16:23 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
11:18:08coppersaratoga: http://pastebin.com/4knGqLKU
11:18:19 Quit JdGordon (Ping timeout: 246 seconds)
11:23:52[Saint]fuck me that's a beast...
11:24:08[Saint]that thing is foolishly overpowered.
11:26:20copperah?
11:28:26copperoy
11:28:32copperMPC is nearly twice as fast as Vorbis
11:32:50copperOpus 128kbps: 171.61% real time :-/
11:33:15[Saint]101% is essentially irrelevant.
11:33:30[Saint]everything higher than 100% is just a bonus :)
11:36:27copperdoesn't it mean lower battery life?
11:36:32 Join JdGordon [0] (~jonno@CPE-121-217-135-158.lnse2.cht.bigpond.net.au)
11:36:32 Quit JdGordon (Changing host)
11:36:32 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
11:36:52[Saint]It does, sure. But I happen to think that anything past 8H is irrelevant.
11:37:02copper8 hours?
11:37:12copperbecause you charge your DAP every night?
11:37:25[Saint]Does anyone really *need* 50H+ for a DAP when we can chare basically anywhere these days? :)
11:37:35copperchare?
11:37:39copperah
11:37:41[Saint]*charge
11:37:51coppertoo much stuff to charge
11:38:07copperI appreciate the fact that I only need to charge my DAP like once a week
11:38:13pamauryThis charging every day habit is killing me
11:38:45copper[Saint]: also, fewer battery cycles
11:39:01[Saint]I just make sure there's a charger at home, one at work, and one in the car.
11:39:13pamauryI appreciate that my daps have longer battery life so that I don't have to bring a charger for example
11:39:17[Saint]Haven't had a low battery inyears.
11:40:06pamauryAnd when you phone will last 1 hour on battery will you say "that's ok I have a charger in my pocket" ? ;)
11:40:53[Saint]The hilarious thing about that example is that I have been known to charge my DAP from my phone on occasions. :)
11:41:13[Saint]And both from my laptop.
11:42:11[Saint]Fwiw, I'm not saying that battery life isn't a good thing. I'm just saying that there's no real need to eek out every last cycle for performances sake, really.
11:42:25[Saint]There's a point where performance is "Good Enough (TM)"
11:42:57 Quit JdGordon (Ping timeout: 276 seconds)
11:43:31pamaurySo one day on battery for your mostly idle phone is good enough ?
11:43:54[Saint]How do you come to that conclusion?
11:44:12[Saint]I use my phone a LOT, and it does over a full day - so...yes.
11:44:20 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
11:45:09[Saint]As long as it lasts me a "working day", I couldn't care less.
11:45:18copper[Saint]: double the performance is quite significant IMO
11:45:28[Saint]Perhaps I'm a minority. But I see no need for many days worth of battery for devices we use daily.
11:46:02[Saint]More to the point, for devices we use for a few hours at a time.
11:46:37[Saint]A few years ago, possibly this bothered me...now, charging is so absurdly easy, pretty much anywhere, I couldn;t care.
11:46:50copper[Saint]: more charging = less battery life over months or years
11:47:05bertrikthis doesn't seem right: http://en.wikipedia.org/wiki/Opus_%28audio_format%29#Hardware_support
11:47:29copperbertrik: the "hardware" qualification?
11:47:32bertrikyes
11:47:40[Saint]copper: realistically, honestly, how many batteries have you killed or lessened the lifespan of in a NOTICEABLE fashion?
11:47:42copperobviously they mean DAP support
11:48:01copper[Saint]: all of my really old gear
11:48:19[Saint]If you value said gear. Replace the battery.
11:48:21copperespecially phones
11:48:26[Saint]You'd need to do so eventually anyway.
11:48:34copperyes but the later the better
11:48:35copper:)
11:48:57[Saint]I don't really buy into the "more charging == less lifespan" shit, even if it is true. You'd need to be really anal to notice this.
11:49:22copperthen let's just say that better performance doesn't hurt? :)
11:49:40[Saint]No, I never said it did hurt. I just said it wan't necessarily relevant :)
11:50:07[Saint]There is a very clear "Good Enough" line.
11:50:24copperrunning test_codec on opus again, this time with the sampling frequency set to 48kHz
11:50:49copperwhat did you say about that option earlier?
11:51:12[Saint]I switched to 48K and everything sounded *reaaaaaaaallly* nasty.
11:51:14copper06:03:26 UTC * [Saint] notes that the sample rate selector is basically unusable.
11:51:19[Saint]Rebooted, and it magically fixed itself.
11:51:23copperhmm
11:52:15copperbertrik: I think that in that context, "hardware support" usually means anything that doesn't run on a computer
11:52:39copperDAPs, hi-fi units, car stereos, etc…
11:53:07copperthough, I'm not sure how smartphones and tablets should be qualified
11:53:14copperthose are basically computers
11:53:17 Quit dv_ (Read error: Operation timed out)
11:54:16copperI guess the distinction is whether the hardware platform has a software platform that allows people to write software for it
11:54:28 Join dv_ [0] (~quassel@chello080108009040.14.11.vie.surfer.at)
11:54:54copperexplicitely
11:54:57bertrikcopper: hm, yes, maybe if you twist the meaning of "hardware" enough, you could call it hardware support
11:55:14copperthat's what people mean by it anyway
11:55:24copperit's a shortcut
11:55:56copperand that's how people understand it
11:56:22copperis there any hardware left with actual hardware support for codecs?
11:56:36copperdoes anyone optimize MP3 playback using a specialized chip anymore?
11:56:48 Quit JdGordon (Read error: Connection reset by peer)
11:56:54[Saint]outside of here?
11:56:57[Saint]I doubt it.
11:57:00coppereven in here
11:57:10bertrikI think you still have these gumstick type of very cheap mp3/wma players with some kind of hw accelleration
11:57:14[Saint]In here - likely not for many many moons.
11:58:33bertrikthe gumstick types are too limited to run rockbox
12:00
12:00:17 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
12:05:39copperthe difference in performance between FLAC and everything else is huge
12:06:19[Saint]0.01MHz realtime :)
12:08:28copperok, sampling freq makes no difference in performance for Opus
12:08:35copperwhich was to be expected
12:09:03bertriksome nasty re-resampling may be going on for opus though
12:09:27 Join DuperMan [0] (~Duper@46-116-90-92.bb.netvision.net.il)
12:09:55bertrikopus is 48 kHz natively, so it might be that it went from 44.1 kHz source to 48 kHz opus to 44.1 kHz rockbox native to 48 kHz resampled
12:10:09copperI know
12:10:30copperah, no, I misread you
12:10:48copperis sampling freq is set to 48kHz, why would there be some re-resampling
12:10:49copper?
12:12:17bertrikbecause someone has to make sure and write code that unnecessary resampling is skipped
12:12:43copperah, that's not done yet?
12:13:04bertrikit's not clear what the exact status is
12:13:57bertrikso, assume the worst :)
12:15:09copperthe worst would be the DAP exploding in my pocket against my junk
12:15:27 Quit JdGordon (Ping timeout: 264 seconds)
12:18:55***Saving seen data "./dancer.seen"
12:21:14[Saint]Nahhh...nothing there of interest. ;)
12:21:49 Join y4n [0] (~y4n@unaffiliated/y4ndexx)
12:22:19 Join JdGordon [0] (~jonno@CPE-121-218-128-223.lnse3.cht.bigpond.net.au)
12:22:25 Quit JdGordon (Changing host)
12:22:25 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
12:28:31coppercpu frequency of the Fuze+: 64000
12:28:42coppercpu frequency of the iPod Classic: 54000000
12:28:45copperwhat does it mean?
12:30:05[Saint]different units of measure?
12:30:51copperso is it really 64MHz and 54MHz?
12:30:58pamauryfrequencies in the debug screen of the fuze+ are in kHz
12:31:10[Saint]copper: yep.
12:31:13copper64 vs 54 doesn't seem much higher
12:31:27pamaury64MHz is the lowest frequency of the fuze+ though
12:31:28[Saint]that's the *minimum* clock.
12:31:39[Saint]ie. unboosted.
12:31:47copperhmmm
12:31:59copperI want to test the Clip+
12:32:27pamaurywhy are you interested in cou frequencies ?
12:33:01copper09:24:09 UTC <[Saint]> that thing is foolishly overpowered.
12:34:10[Saint]Wow...and you really thought 64MHz was the max clock?
12:34:21[Saint]Its MUCh higher :)
12:35:07copperI don't know anything!
12:37:14copper[Saint]: can you put the Fuze+ into perspective for me?
12:37:26coppercompared to, say, an iPod Classic and a Clip+
12:37:47[Saint]It _should_ eat both for breakfast.
12:38:19[Saint]i.mx233 doesn't mess around.
12:39:05[Saint]its mac clock is somewhere in the range of 700MHz.
12:39:09[Saint]*max
12:39:11 Quit JdGordon (Ping timeout: 256 seconds)
12:39:17[Saint]680 or so, iirc.
12:39:56[Saint]couple that with solid state storage, and the Classic becomes a bit of a dinosaur.
12:41:48[Saint]bah...silly memory. I may be thinking of something else. I want to say ~450MHz this time.
12:41:55DuperManiPod mini with a Compact Flash
12:42:01DuperMangood balance imo
12:42:17DuperManan opinion grealy prejudiced towarsd what I own
12:42:39[Saint]DuperMan: Its a nice, quite underpowered, target.
12:42:51DuperMantrue. plays very nice with flac
12:43:02DuperManbut db things make it insane
12:43:25DuperManthat's more to do with it expecting an hdd though, and controlling a cf in actuality
12:43:34coppertesting both the Classic and the Clip+
12:43:36DuperManwanna buy the original 4gb microdrive though?:D
12:43:49[Saint]Nup. I have a small army of them.
12:44:05[Saint]copper: turns out I "whoops"ed.
12:44:12copperhmm?
12:44:14DuperManhehe. ever buffed the apple logo on the back of the mini? it so shiny
12:44:14[Saint]its 454MHz max clock for the F=
12:44:23[Saint]*+
12:44:46[Saint]which is still pretty nuts.
12:45:02DuperMany b gr... oh. that's the native clock?
12:45:05DuperManoO
12:45:30DuperManfuze plus or clip plus?
12:45:31[Saint]Makes your Mini's 80 look a little dated, huh?
12:45:55DuperMannah. I upgraded from my 30mhz palm IIIx a month ago ;)
12:46:40DuperMan454... ain't that best gutted for a cheaper still r-pi?
12:46:45[Saint]fUZE+, BTW.
12:46:53[Saint]eeeks, kitten-caps.
12:47:07copperI guess they needed a speedy CPU for the OF interface
12:47:19DuperManOF?
12:47:27[Saint]Original Firmware
12:49:08DuperManwasn't it like 128 by 160?
12:49:21DuperManmustv'e had mad effects for such a low res to be demanding:D
12:49:27copper240x320
12:49:52DuperManreminds me of my hx4700 behemoth
12:50:11DuperMan600mhz 600x480 iirc intel armstrong hehe
12:50:17DuperMan(ppc)
12:50:59DuperManboth cf and sd though... also got an hx2700 I should find a new batt for, then tcpmp
12:51:26 Join lebellium [0] (~chatzilla@lns-c10k-ld-02-m-212-194-176-149.dsl.sta.abo.bbox.fr)
12:52:00DuperManTHE POCKET PC / WINDOWS MOBILE >=6.5 LIVES
12:52:24copperFuze+ performance spreadsheet of codecs that are of interest to me: http://outpost.fr/url/28s7
12:52:24 Join JdGordon [0] (~jonno@CPE-121-218-128-145.lnse3.cht.bigpond.net.au)
12:52:26 Quit JdGordon (Changing host)
12:52:26 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
12:53:15DuperManwhat's coloumn b standing for?
12:53:39DuperMantime to decode a frame or frame per x time?
12:54:03copper% realtime
12:54:51DuperManso it could transcode otf had you cared for that
12:55:14copper?
12:55:14DuperManbig stinky feature request: for targets with wifi, dlna servor?
12:55:16DuperMan^^
12:55:32coppertranscoding performance is unrelated to Rockbox decoding performance
12:55:45coppertranscoding performance depends on your computer
12:56:01DuperMantoo dependant on io perf and other bottlenecks?
12:56:10[Saint]The targets we support with wifi all have MUCh better implementations in their own firmware - so...probably not gonna happen.
12:56:11copperand "% real time" means nothing to transcoding "on the fly"
12:56:16DuperManI was thinking more along the lines of making the player be a server
12:56:46DuperMancopper: I fail to see how the two are seperate
12:56:48copperjust because encoding is faster than real time, doesn't mean it's fast enough
12:57:03copperDuperMan: you don't transcode with Rockbox
12:57:08copperyou transcode with your computer
12:57:12DuperManbut... it... does... all other things being equal
12:57:18copperhow is Rockbox performance related to your computer's performance?
12:57:21copperwhat?
12:57:29copperrockbox doesn't transcode anything
12:57:35DuperMancopper, I was being madddddd and suggesting adding a dlna or upnp server in rockbox
12:57:50DuperMannot speaking of actual existing features
12:58:10[Saint][22:57:31] <copper> rockbox doesn't transcode anything <−−- well...technically...
12:58:18[Saint]re: Recording
12:58:20DuperMana dac is a dac
12:58:29DuperManeh [Saint]?
12:58:38copper[Saint]: recording means encoding not transcoding
12:58:45copperalso, doesn't the recorder use WAV?
12:58:48DuperManit transcodes a pcm stream
12:58:50DuperMan><
12:58:58DuperManor repackages for wav
12:59:27[Saint]I'm pretty sure we capture everything as raw PCM and then convert if the target file is anything other than that.
12:59:31copperit doesn't even encoding anything per se
12:59:32[Saint]Not entirely sure, though.
12:59:38copperencode*
12:59:53DuperMan^^^ repackage
13:00
13:00:05DuperMana wav is a container for a raw pcm stream
13:00:09copperI know
13:00:16DuperManImma clarifies
13:00:19copperI don't consider converting raw PCM to WAV "encoding"
13:00:32[Saint]options for recording are PCM, Wav, and Mp3...pretty sure there's no magic involved, so it must be doing *something* to get to mp3 in the end :)
13:00:34DuperManI don't consider it converting :P
13:00:49copperMP3, really?
13:00:54DuperManlol real time selective octet handling
13:00:55DuperMan:D
13:01:07DuperManit only HEARS the bits relevant to cbr mp3
13:01:10DuperMan;P
13:01:22DuperManalso, magic
13:01:27[Saint]copper: really really.
13:01:42copperhmmm
13:01:43[Saint]In which case, we *muct* encode.
13:01:49[Saint]but, this is pure pedantism.
13:01:50DuperMananybody wants a screenless iriver h340?
13:01:51copperI think the Classic results are going to be somewhat surprising
13:02:02DuperManshow 'em
13:02:10coppernot done yet
13:02:27DuperManI have battery dead x5l and lcd dead h340
13:02:28DuperMan:(
13:02:42DuperMansrsly, anybody awesome enough to have use for those?
13:02:52[Saint]So, yeah, sorry for that - it was really assholeish - but, yeah, Rockbox can technically transcode.
13:03:05DuperManit wasn't! it DOES
13:03:07DuperManarrrgh
13:03:49DuperManit wasn't even pedantic, seeing as you did not point out an irellevent to the topic lingual mistake
13:04:12copperI stand corrected about MP3 recording
13:04:20DuperManbut pointed out that the action in question is done under a certain scenario
13:04:51[Saint]I _think_ there's also a WAV-to-MP3 plugin.
13:05:00DuperManbut my original comment wasn't about the actual, but about a trollish feature request
13:05:01DuperMan:P
13:05:15DuperManmeh. it probably isn't hardware accelerated
13:05:23DuperManso must be sluggish mad
13:05:38DuperMantoo many possible cpu and dac archs
13:05:54DuperManor kernel has api/hal?
13:06:21[Saint]Its not /sooo/ trollish, but realistically, the targets (presently) that have wifi are all "Application" targets, meaning we run as an app hosted in the targets OS, rather than replacing said OS, and the OS almost certainly has a better solution for this than we could implement.
13:06:29DuperManlike, for accessing the hw encoding machinary
13:06:55DuperManhmmm... gotta praise the voodoo you do (we do?:D) on making android sound divine though
13:07:14[Saint]do what?
13:07:19[Saint]...remind me of the babe!
13:07:24DuperManbtw, are my ears better served by the WM1811 DAC in my phone or whatever's in the ipod mini?
13:07:26*[Saint] skulks away
13:07:35*DuperMan is confused
13:07:54DuperMan"do what?" guess I placeboed myself
13:08:13[Saint]I thought you were quoting from David Bowie's "Dance the Magic Dance" :)
13:08:48copperhmmm, how do I quit the test_codec once it's done, on the Classic?
13:08:59DuperManrofl :) no sorry
13:09:42[Saint]copper: insane iPod keymap tells me it'll probably be "select+<something>"
13:09:51[Saint]menu+select, probably.
13:16:12copperiPod Classic codec performance: http://pastebin.com/KfP5Za26
13:16:31DuperManthanks!
13:16:44copperhttp://outpost.fr/url/2lC0
13:17:10DuperManit raped your fuse
13:17:28[Saint]errrr...
13:17:29coppernot really
13:17:37[Saint]Are you sure you know what you're looking at?
13:17:38copperperformance figures are very different though
13:17:57copperon the classic, Vorbis is faster than Musepack!!
13:18:06coppermarginally
13:18:25DuperManme? no
13:18:25[Saint]DuperMan: if its not obvious, there's a clear winner here.
13:18:30[Saint]...and it isn;t the Classic.
13:18:35copperand the differences between most codecs are much less pronounced
13:18:56copper[Saint]: er, not really
13:19:09copperLAME and VOrbis are faster on the Classic
13:19:20[Saint]...but, everything else?
13:19:22DuperManwhoops. yeah
13:19:23[Saint]:)
13:19:34copperALAC too
13:19:35DuperManI assumed the codecs would be stacked samely in both tables
13:19:39DuperManxD
13:19:56copperWavPack too
13:20:09 Join Nephiel [0] (~5332f9b8@www.haxx.se)
13:20:11DuperManMY BAD
13:20:43copperAPE too!
13:21:00DuperManI have a clip+ around, but need to do the fake jtag thing to revive it's flash
13:21:02 Quit lebellium (Quit: ChatZilla 0.9.90.1 [Firefox 23.0/20130725195523])
13:21:06DuperMannand*
13:21:22 Join lebellium [0] (~chatzilla@lns-c10k-ld-02-m-212-194-176-149.dsl.sta.abo.bbox.fr)
13:22:12[Saint]copper: well..hmmm...I would say the winner is clear, even though the Classic is (marginally) faster for a few codecs, look at the CPU usage.
13:22:32copper?
13:22:36[Saint]The Classic spends a LOT more CPU doing essentially the same thing.
13:23:03copperaw snap, the log on the Clip+ is empty
13:23:03[Saint]Fuze+ a52@192
13:23:05[Saint]0.06MHz needed for realtime
13:23:15[Saint]Classic?
13:23:16[Saint]18.48MHz needed for realtime
13:23:36[Saint]just a bit of a difference ;)
13:23:39DuperMan0.06. must haz teh hawt dac
13:24:00copperah
13:24:27[Saint]while the realtime numbers are similar, the CPU usage is by far in favor of the F+
13:24:42DuperMana dac is a discrete cpu though
13:25:52copper[Saint]: does that mean the Fuze+ CPU is faster, or the Fuze+ target is more optimized?
13:26:13TheSevendamn, I'm *almost* there
13:26:13DuperMancopper: probably means the dac unit does most of the work
13:26:21copperwhat
13:26:23DuperManwhich should count as 'target be optimized'
13:26:25[Saint]I believe it to be a combination of a much faster CPU, and much faster storage.
13:26:25coppera DAC doesn't decode
13:26:28DuperManand 'target is gooder'
13:26:51DuperMandoesn't it? only pure streams? tell that to my denon
13:26:53coppera DAC converts PCM to analog
13:27:00bertrikThe CPU plays a large part in playback performance, but other stuff like memory access speed/latency can have a lot of impact
13:27:23DuperManyeah, caches too
13:27:36DuperManneeds to be as much like realtime systems as possible really
13:28:24copperre-running the Clip+ test
13:28:35TheSevenumsboot v0.2 *almost* connects... but something chews up the CSW of the SCSI INQUIRY transaction...
13:28:52DuperManbloated linux kernel
13:28:57pamauryon the fuze+ the memory latency is huge, the ddr clock has a huge impact
13:28:58TheSevenfun fact: the very same code and driver works perfectly on a different device (STM32 SoC)
13:30:15pamauryTheSeven: is the amsv2 driver rewrite making some progress ?
13:30:40TheSevenno idea, I'm doing it in the other direction right now
13:31:02TheSevenporting the UMSboot userspace/application layer to the other framework which the USB driver comes from
13:31:30TheSevenjust to get that part out of the way and to have an actually tested code base for porting to rockbox
13:32:16TheSevenbut right now it looks like something breaks the second non-EP0 IN transfer for some reason
13:32:26TheSevenno matter what I try to send, the host always receives 2 bytes: 00 01
13:32:56TheSeventhat's kinda suspicious because those bytes might make sense as a GET STATUS EP0 response. but how do they end up on EP1?
13:34:15pamaurybad buffer configuration ? I think on this chip you can choose the buffer offsets and possibly overlap them or something like that ?
13:34:25copperI added "needed CPU MHz" to the spreadsheets
13:36:07pamauryTheSeven: so if I understand correctly, the same code on two different devices, one works while the other doesn't ?
13:37:01TheSevenyes, mostly the same code
13:37:07TheSevencompletely different SoC, but same USB core
13:37:14TheSevendifferent core config though
13:37:22TheSeventhe ipod is DMA-only, the other one PIO-only
13:37:43TheSevenI just checked: even if I don't even attempt to send a CSW, that garbage CSW arrives at the host
13:37:54TheSevenI also don't ever send a GET STATUS response or any other 2 byte EP0 packet
13:38:03TheSevenso where one earth does that garbage come from?
13:38:23pamauryDMA only ? I thought the chip was always capable for PIO
13:39:41TheSevenI thought that as well... but PIO seems to be broken on the nano2g as well
13:40:07TheSevenI can receive packets using PIO, but I will never ever get a TX fifo empty IRQ, so I can't send
13:40:19pamauryOn the clip+ I tried PIO but it didn't work much better either, it wasn't fully broken but not better dma
13:43:03pamauryat least that means that all the people who tried to write the driver aren't completely dumb, there really is something we don't understand :-/
13:43:45pamaurywhat about reverse engineering the whole nano2g or clip+ driver ? could that help or maybe you already did ?
13:53:54DuperManWHAT ABOUT U DO IT IF IT SEEMS TRIVIAL
13:54:14DuperMansorry reflex
13:55:14[Saint]...are you just here to troll, or...?
14:00
14:01:16DuperManto troll and educate myself, as well as lurk for the off chance somebody with a penchant for the mini comes along wanting to become my mentor
14:02:28[Saint]You need to stick around for ~1Y+ before you start trolling our active developers ;)
14:03:35 Join ender` [0] (krneki@foo.eternallybored.org)
14:03:50pamauryDuperMan: I'm very aware of what reverse engineering work is if you want to know :)
14:04:09DuperManoh lol well I troll codeworkx on the cyanogen places, even cotulla on xda-devs
14:04:12DuperManno cred? xD
14:04:36copperClip+ codec performance: http://pastebin.com/6mx8Lcdv and http://outpost.fr/url/2odq
14:05:22DuperManpretty equivalent to the classic
14:05:35coppernot really
14:05:41copperVorbis is twice slower on the Clip+
14:05:47DuperManoO oh
14:05:58copperit's actually the slowest codec among those that I'm interested in
14:06:09DuperManflac is more throughput = drains batt but easy to actually play right?
14:06:16DuperManunparses to pcm
14:06:41DuperManvorbis is famously complex, but better than mp3 at similar bitrates derp
14:08:34copperthe Fuze+ kills the Classic and the Clip+ in FLAC performance
14:09:29copperpamaury: given the very low demand for CPU processing on the Fuze+, maybe you could put it in a low power state or something?
14:09:37copperis battery usage on the Fuze+ optimized already?
14:10:33copper[Saint]: the difference in "needed CPU MHz" between the Fuze+ and the other too is insane
14:10:37coppertwo*
14:10:48pamauryit's already super optimised: we run at 64MHz most of the time and sleep the cpu whenever there is nothing to do
14:10:53copperok cool
14:11:06*DuperMan eyes his Palm IIIx
14:11:18DuperManalas, no headphone jack
14:11:20pamaurywith the upcoming patches for the touchpad we might reach ~40h of battery life
14:11:33coppernice
14:12:18DuperMansorry for such obscene trolling, but shame the old creative njb3 series didn't see more love
14:12:21copperusually I feel like the Fuze+ takes a long time to charge, do you see the same thing?
14:12:27 Quit tchan (Quit: WeeChat 0.4.1)
14:14:06pamauryhonestly I didn't benchmark charging, I think we are already charging at the same current level as the OF, we don't have the battery spec so that's the safest we can go imo
14:15:09pamauryif you want to benchmark OF vs RB on this matter, maybe we could have figures
14:17:52coppermeh, I can't really be bothered to do that
14:18:59***Saving seen data "./dancer.seen"
14:19:05coppergreat job on the Fuze+ anyway, can't thank you enough for that :)
14:19:18copperI love that thing (with Rockbox).
14:22:10jlbiasinipamaury: ping
14:22:16pamaurythanks :) we have made a long way since the beginning !
14:22:19pamauryjlbiasini: pong
14:22:19TheSevenpamaury: looks like the version of the OTG used in the ipods is a bit older and doesn't have some of the features that my driver relies on for PIO mode
14:22:29TheSevenespecially it doesn't seem to have an IRQ for that fifo empty condition
14:22:46TheSevenhow did you deal with that in PIO mode?
14:22:55TheSevenwere you polling it?
14:23:44jlbiasinipamaury: do I have to declare the do_interupt function in the button_target.h? The compiler seems to want that and I don't know how to turn it around otherwise
14:23:50pamauryI don't remember honestly, I think I relied on IRQ
14:24:01TheSeventhe nano2g driver was initially implemented by reverse engineering it btw, but it nevertheless didn't work properly :)
14:24:17TheSevenwell, there doesn't seem to be an IRQ for this on that core version... maybe there is one on AMS?
14:24:30pamauryjlbiasini: no, the do_interrupt() function is already in button-fuzeplus.c, maybe I made a typo like interrupt vs interupt
14:24:45TheSeventhe core version used in the nano2g seems to be fairly close to the one in the s3c6400x datasheet, which completely lacks the registers and IRQ bits for FIFO management
14:25:14pamauryTheSeven: at some point I implemented PIO, maybe you can look at the history of the the driver file or on flyspray to find the code
14:29:42pamauryTheSeven: FS #11664
14:30:13pamaurysearch for PIO
14:38:22coppermaybe it should be recommended somewhere, to unplug all unnecessary USB devices, before connecting a Rockboxed DAP
14:38:27copperwrt USB problems
14:38:43copperI'm almost certain my USB DAC was the one causing my problems, directly or indirectly
14:40:11DuperManwhich dac? fiio one?
14:40:17copperODAC
14:40:18jlbiasinipamaury: no it wasn't (at least with the actual head, so I added a static void do_interrupt(void); at the beginnig of the button-fuzeplus.c file. Is that the correct way to do?
14:40:22DuperManmy e17 bsods often
14:40:35DuperManbut fidelity's worth it :P
14:41:27pamauryjlbiasini: should be unnecessary: do_interrupt() is in the same file as the caller and before, why do you need to do this ? compile error ?
14:41:39jlbiasiniyes
14:41:51pamaurydid you modify my code ?
14:42:37jlbiasininot that much only adding the report rate modification
14:43:16jlbiasiniis that place where appears the do interrupt function important (I means top/end of the fil for instance?)
14:43:55pamaurycan you pastebin the code ? I don't think the report rate modification is needed by the way. Yes if you call it before it is defined, you need to declare it before: "static void do_interrupt();"
14:45:19jlbiasinithe report rate has to be lower when in very_low_power
14:45:26jlbiasiniat least
14:46:32jlbiasiniwell *has* not but I would like to test it to see if it's improving battery life without being anoying
14:46:52pamauryok, but I doubt it, if you don't touch the device, there will be no report sent
14:47:57jlbiasinipamaury: ok here http://pastebin.com/MVL4uDgQ
14:48:29TheSevenpamaury: looks like you used one TX fifo for all EPs?
14:48:52pamauryI don't remember, that was a long time ago and I probably didn't know what I was doing ^^
14:49:46pamauryjlbiasini: why put do_interrupt at the end ? it should stay within the #ifdef #endif because it's not needed in the bootloader
14:49:58 Quit kevku (Ping timeout: 260 seconds)
14:50:17TheSevenhm... I have two datasheets that are explicitly contradicting each other
14:50:31TheSevens3c6400x says that everything nonperiodic shall be pointed at fifo 0
14:50:33jlbiasiniyeah i just moved it to see if it was changing something to my compilation problem
14:50:51TheSevenstm32f4 says that each EP shall have its own fifo or things might get messed up
14:50:52pamauryTheSeven: I based on the s3c6400x datasheet iirc
14:51:56TheSevenmixing the data in a fifo in device mode just sounds like a very bad idea
14:52:01jlbiasinipamaury; I thing I deleted the touchpad sensitivity by error
14:52:07TheSevenand IIRC is responsible for a whole bunch of USB lockup problems on the classic
14:52:07jlbiasini*think
14:52:20TheSevenso I tend to not trust the s3c6400x datasheet there
14:52:52pamauryok, seems safer to have each EP have its own fifo
14:53:07pamauryif the FIFO large enough to do this, always ?
14:53:27TheSevenyes
14:54:30TheSevenotherwise you may run into situations where the data that the host is waiting for can't be pushed into the fifo because it is full with data for another EP
14:54:32pamaurythat fifo config param has been mysterious to me, it's not very well explained
14:54:47TheSevenit's much better explained in the stm32f4 datasheet
14:55:09pamauryI should read this one ^^
14:55:37TheSevenit seems to be for a newer core revision though that handles PIO differently (and more efficiently)
14:55:57TheSevenI figure that this PIO change might be one of the reasons why things need to be in the same FIFO on the 6400x
14:56:05TheSevenbut for DMA it doesn't make any sense
14:57:07pamauryI found another chip which uses a very similar core: Rockchip Nano-B uses the same as stm32f2
14:57:24TheSevenshould be the same one then, f2/f4 are identical
14:57:27jlbiasinipamaury: this is the error I get http://pastebin.com/NYU01SAQ (after redoing the touchpad_set_sensitivity()
14:58:21pamauryjlbiasini: you need to "static void do_interrupt(void);" at the top of the file if you implement do_interrupt after it is called
14:58:53jlbiasiniahah so should I put it above unstead of declaring it?
14:59:09pamauryyes, like it was done in the original patch
14:59:20jlbiasiniooops sorry
15:00
15:10:52 Join tchan [0] (~tchan@lunar-linux/developer/tchan)
15:20:54jlbiasinipamaury: how long is 60 * HZ? 1 minutes?
15:21:10jlbiasinior is it device dependend?
15:21:34gevaerts1 minute, yes
15:22:20gevaertsI don't believe HZ is different on different devices right now, but it could be, and if everything uses HZ properly, it doesn't matter
15:22:38gevaertsHZ is "ticks per second"
15:22:55gevaertsSo 60 * HZ is "ticks per minute"
15:23:28jlbiasiniyeah that is what I though thanks for confirmation
15:23:42[Saint]copper: Rockboxed devices can do some "Weird Shit (TM)" if left connected to some things.
15:23:45[Saint]...no idea why.
15:23:58[Saint]I don't think anyone knows.
15:24:15TheSevenpamaury: as far as I can tell the s3c6400x variant of the core can only have in flight transfers on one endpoint at a time in PIO mode
15:24:22copperI think it would be a cheap and potentially effective recommendation to make
15:24:28gevaertsNormally (i.e. outside of some very specific low-level drivers) you don't need to know that HZ=100 :)
15:24:31[Saint]Some BIOS freak the fuck out and just hang, some machines refuse to shut down or hibernate.
15:24:34[Saint]Lots of fun things.
15:25:14*gevaerts doesn't like recommendations that are based on vague annecdotes and that can prevent people doing useful things
15:26:27[Saint]Where's the recommendation?
15:26:54copper[Saint]: recommending to unplug all unnecessary USB devices
15:27:04copperprior to connecting a Rockbox device
15:27:23[Saint]Ah. I thought that was directed at me. WHoops.
15:28:15TheSevenindeed. there's no way for the core to identify which endpoint a packet belongs to if multiple are enabled
15:28:18pamaurycopper: what ? I've never heard of other usb devices interfering with a rockbox device, actually that seems pretty impossible given how usb works, except for a hub possibly
15:28:37copperpamaury: my devices almost always crash when my ODAC is connected
15:28:42copperand NEVER crash when it's not
15:28:53TheSevensounds like a buggy driver
15:29:09copperdoesn't matter what the problem is, until it's fixed
15:29:13pamauryyeah, buggy OS, nothing related to rockbox
15:29:30pamaurywe can't warm against all possibly buggy code in the world
15:29:34pamaury*warn
15:29:37coppersigh
15:29:51copperso you'd rather have people's DAPs crash?
15:30:03[Saint]than do weird broken shit? yep.
15:30:12copperI mean, you guys already recommend changing skins
15:30:24copperisn't that anecdotal evidence?
15:30:44copperno-one even knows why skins could be a problem
15:30:55copperexactly*
15:31:20pamaurywhich device are you talking about which crashes with this ODAC thing ?
15:31:28jlbiasinigevaerts: now that I have you here, did you saw the comment of Lorenzo about that touchdev disable on keylock? he was suggesting to let that function and on devices that don't handle touchdev sleep, to have it just filtering the button reporting output...
15:31:29copperClassic, Fuze+, Clip+
15:31:56copperthing is, Rockbox crashes, and no-one knows why
15:32:10[Saint]as I understand it if Rb works fine without this thing plugged, but doesn't when it is, the other thing is at fault - or the host.
15:32:12copperso I mean to do anything that can prevent it from happening
15:32:21copper[Saint]: I'm not pointing fingers
15:32:28copperthat's not the point
15:32:56copperrather than tell users, "yeah, sorry, Rockbox crashes", I'd rather tell them "try to unplug everything else"
15:33:20[Saint]so, you're asking to Rockbox workaround some unknown fault to "fix" a problem that probably isn't theirs?
15:33:26jlbiasinior use the OF usb bootloader
15:33:36copperwhat?
15:34:11copperif, for instance, it's a USB driver bug
15:34:32copperwell, lots of people happen to be using linux with the exact same driver, since it's completely generic
15:34:32gevaertscopper: the theme problem has been seen on several devices, by several people. No, we don't know the exact cause, but in my mind it's gone beyone anecdotal
15:35:22copperI mean to recommend anything that can help prevent a Rockbox device from crashing, that's all
15:35:37gevaertsjlbiasini: no, I haven't. When?
15:35:42DuperManhmmm... db generation creating 4gb files that are just artifacts of the fs malfunctioning
15:35:43*[Saint] recommends not using the device
15:35:53DuperManis it a normal fat32 on 64gb cf ipod thing?
15:35:54[Saint]..it'll definitely not crash then.
15:35:57copper[Saint]: ok cool, I'll stop using Rockbox on all my three DAPs!
15:35:59[Saint]problem solved.
15:36:07copperor did you mean the ODAC?
15:36:31jlbiasinigevaerts: it was here http://www.rockbox.org/mail/archive//rockbox-dev-archive-2013-07/0008.shtml on the mailing list
15:36:36DuperMancopper: to prevent it from crashing don't use it
15:36:49copperthat's just silly
15:36:56 Quit mirak (Quit: Ex-Chat)
15:37:01DuperManthat's the only true solution, for any os
15:37:06DuperManeven jets
15:37:22[Saint]It was my remarkably awesome attempt at absurdium ad reducto.
15:37:46[Saint]reductio ad absurdum, even.
15:37:57DuperManyou have no idea how absurdium ad reducto really is fitting
15:37:58DuperManah
15:38:00DuperManhehe
15:38:07gevaertscopper: if you qualify the recommendation enough, e.g. "Some people have observed issues when certain specific USB devices are also connected, so if you see USB instability or crashes, you might try disconnecting non-essential devices (and if this makes a difference, tell us about this)", I think it might be a good idea
15:38:14 Join Epicanis [0] (~kvirc@sdsl-68-238-63-20.static.ngn.east.myfairpoint.net)
15:38:30DuperManso you weren't silly to the point of vanishing
15:38:46*[Saint] vanishes
15:38:48copperwhat
15:38:51copperI don't speak latin
15:39:03gevaertsDuperMan: what do you mean by "db generation creating 4gb files that are just artifacts of the fs malfunctioning"?
15:39:03DuperMangevaerts: that's oldddddd hat
15:39:31DuperManit was mostly a recommendation for the n00b/moms out there that stick usb things in unpowered ports
15:39:37gevaertsDuperMan: you mean I should encourage the discussion going off-topic? No, sorry.
15:39:47DuperMangevaerts: my ipod mini 1st gen does just that
15:40:17DuperManit lists in the fat a 4gb file that doesn't effect real available space
15:40:24gevaertsRight
15:41:01DuperManso... is that common, insofar as you're aware?
15:41:23gevaertsI've never seen anyone report that specifically, but yes, check the filesystem
15:41:33DuperManI assume I need to compile my own builds and have the build not assume it's talking to an hdd
15:41:39gevaertsNo
15:41:48gevaertsRockbox should work fine with CF cards
15:41:54[Saint]I have only personally seen db generation *really* thrash an already trashed fs.
15:42:23DuperManhmmm... it's been factory reset a few times, but I don't remember formatting it low level
15:42:28[Saint]I have not personally seen it actually cause issues in stable devices.
15:42:29gevaertsIf you're sure you start with a clean filesystem and you end up with a corrupted filesystem, and you didn't have a crash or an unsafe unplug in between, there's a bug
15:42:36DuperManso... "no" to letting itunes format for me?
15:42:46gevaertsYou don't do low-level formats on non-floppy devices typically :)
15:42:57DuperManhehe yeah but it sure is fun
15:42:58DuperMan:)
15:43:07gevaertsWell, you typically *can't*
15:43:32DuperManehrm... disk part is kinda cool like that, but no reconfiguring the controller hhe
15:43:49DuperManI meant a full rather than quick format
15:44:24gevaertsThat shouldn't make any difference at all
15:44:58DuperManso to be sure I'm working with a workable thingy, I should put the cf in a card reader and manually build the partitions etc.? or my pletorous normal formattings shouldv'e sufficed?
15:45:06TheSeven(besides taking longer and wearing out the media)
15:45:11DuperMan^
15:45:33DuperManit's a thing you do when frustrated. I can't be the only one who did...
15:45:36DuperMan:P
15:45:42TheSevenIF things work right, you should be able to format it through rockbox's USB interface, but then again that isn't exactly the most robust thing
15:45:42*[Saint] giggles
15:45:48[Saint]"workable thingy"
15:46:15DuperManwell "ipod" isn't a more respectable term
15:46:17DuperMan:)
15:46:23gevaertsDuperMan: doesn't make *any* difference. If whatever formats the device makes sure the FAT is clean and the root directory is empty, and the partition tables make sense, you're fine
15:47:19DuperManhmmm... wouldn't that make an ipod panic and sadface? not sure if the handover from the bootloader is timed for such a case
15:47:30DuperManlike, I won't have the magic apple folder of ipodstuffs
15:48:06gevaertsWell, I'm assuming you don't let the tools touch the firmware partition
15:48:12DuperManI'll save us all the trouble. the cf card I use is a dolgix:D
15:48:36DuperManI haven't so far, but that keeps on happening. best results are when I set spindown to 0 to disable it
15:49:32DuperManbut it always seem to happen after a while. also; battery use reports are whimsical
15:50:31DuperManif anybody knows, will compiling a build that supports exfat work or will the ipod bootloader won't be able to handoff to the kernel?
15:51:01DuperMan(I mean, format it to exfat after I made sure my rockbox will be able to read it. I know it won't be OF compatible)
15:51:23gevaertsThe ipod bootloader doesn't care about the data partition
15:51:44gevaertsOf course, "compiling a build that supports exfat" implies implementing exfat support first, so good luck :)
15:52:03DuperMananother idea I had is working with the ipodlinux bootloader, but that's such an old fad I'm unable to find data on it
15:52:10DuperManah! xD
15:52:16DuperManhehehe thanks, wish I could
16:00
16:05:18 Join kevku [0] (~kevku@2001:0:c38c:c38c:e6:4b86:3d69:bedb)
16:07:03 Quit akaWolf (Ping timeout: 264 seconds)
16:07:17 Join akaWolf [0] (~akaWolf@188.134.9.161)
16:07:25 Quit akaWolf (Changing host)
16:07:25 Join akaWolf [0] (~akaWolf@unaffiliated/akawolf)
16:07:47 Quit Guest11127 (Read error: Connection reset by peer)
16:19:02***Saving seen data "./dancer.seen"
16:19:55soap!seen linuxstb
16:19:59soap.seen linuxstb
16:20:01soapgah
16:20:21bertrik /msg logbot seen linuxstb
16:20:30soapty
16:20:39jlbiasinigevaerts: so any idea about http://www.rockbox.org/mail/archive//rockbox-dev-archive-2013-07/0008.shtml
16:29:45gevaertsjlbiasini: I suspect he knows that part of the code better than I do, so I'll believe him :)
16:30:04jlbiasiniok
16:30:15 Part jvd
16:31:12jlbiasiniso I will start implementing it then...
16:31:36 Quit pamaury (Read error: Operation timed out)
16:51:29 Quit jlbiasini (Quit: jlbiasini)
17:00
17:02:29 Quit [Saint] (Remote host closed the connection)
17:02:48TheSevenwhat the hell...
17:03:08TheSeventhis OTG decrements the "packets to be sent" counter below zero, wrapping to 1023
17:04:40 Join [Saint] [0] (~saint@rockbox/user/saint)
17:06:36 Join webguest53 [0] (~3c1182e6@www.haxx.se)
17:07:17 Quit webguest53 (Client Quit)
17:17:19TheSevenhm
17:17:31TheSevenapparently these ipods indeed have a buggy old version of this core
17:29:49 Quit thegeek (Read error: Connection reset by peer)
17:31:33 Join thegeek [0] (~thegeek@40.200.16.62.customer.cdi.no)
17:33:43 Quit kevku (Ping timeout: 245 seconds)
17:42:54NephielHi there
17:43:52NephielI was looking at FS #11376 http://www.rockbox.org/tracker/task/11376
17:47:11Nephieldoesn't seem so hard to fix. Am I right in assuming track indexes aren't always in order on playlists?
17:48:06gevaertsWell, it's probably easy for someone who knows that subsystem :)
17:48:20gevaertsWhich could be you, after a few days' work!
17:49:22Nephielwell, that's what I am trying to do ;)
17:50:51[Saint]I'm not actually sure this is anything that "broke" as the tracker suggests.
17:51:09[Saint]I recall seeing this behavior years and years and years ago.
17:51:26gevaertsWell yes, it's clearly old
17:51:45gevaertsBut if I read the report correctly, it's also quite clearly a bug
17:52:01[Saint]Oh, yes, no denying that.
17:52:26Nephielwell, I've been seeing it for quite a while and I found it quite annoying, that's why I wanted to fix it
17:53:03[Saint]that's a good reason. :)
17:53:05Nephieljust built from git, bug still there
17:53:38*gevaerts never actually moved a track to before the current track in his playlist
17:53:38[Saint]The code that handles this probably hasn;t changed for some time.
17:54:01gevaertsIt sounds like something only very few playlist powerusers would do
17:54:13gevaertsWhich is why nobody fixed it yet, probably
17:55:13NephielActually it happens whenever you move any track to any position before the currently playing track
17:55:49Nephielexcept to the first one (that looks like a prepend and works as intended)
17:57:18[Saint]Moving a track to before the currently playing track likely occurs rarely, as gevaerts suggested.
17:57:27[Saint]I can't think of any use case for it personally.
17:57:52[Saint]I want to listen to my tracks, so moving them before the current one seems counterproductive to their playback.
17:58:39gevaertsI can imagine some use cases, but none of them is something I'd personally need
17:58:50gevaertse.g. preparing a playlist for later use
17:59:01gevaerts(while listening, so you get a proper feel)
17:59:27[Saint]I do that on a PC, as its so damn painful on-device.
17:59:41gevaertsWell yes, due to this bug :)
17:59:50NephielI do use it. Suppose I have a track coming up that I don't want to listen to right now, but don't want to remove it from the list as I might want to listen to it later
18:00
18:00:06Nephielso I move it above the currently playing one
18:00:15[Saint]>>|
18:00:22[Saint]that's my solution for that :)
18:00:31gevaerts[Saint]: you can only do that when you reach the track
18:01:10gevaertsi.e. not three tracks in advance
18:01:44gevaertsHmmm
18:01:53Nephiel>>| works but not if you're listening while having a nap
18:02:13gevaertsWhat happens if you move a track the other way, i.e. from before the current track to after? Do things get confused then?
18:02:28[Saint]I imagine so?
18:02:55Nephielno, that works. In fact I think I have it figured out, patch incoming
18:03:49 Join krabador [0] (~krabador_@unaffiliated/krabador)
18:05:50Nephielbut I'd like to know if I can simply look at playlist indexes to check if a track comes before another in the playlist
18:06:45Nephielif not, I'll have to look at display_index for each track I want to check
18:09:56*gevaerts has no idea
18:10:18gevaertsI'd suggest putting a patch on gerrit using your first assumptions, and waiting for a review by someone who knows
18:11:33*[Saint] suggests looking at http://www.rockbox.org/wiki/UsingGit#Preparing_to_make_changes
18:12:25[Saint](thought it doesn't actually state it there, there's also a Real Name policy)
18:13:42[Saint]the git config username should be your realname, and the git config email address the same as the address used in gerrit.
18:14:04NephielI was about to post it to Flyspray
18:14:31gevaertsWe do prefer gerrit, but if it's a simple patch I suspect it can go on flyspray
18:14:57gevaertsWell, I do prefer bugfixes on flyspray over no bugfixes at all, anyway :)
18:15:57[Saint]Shhhhuuuuuush!
18:16:01[Saint]He's fixing bugs.
18:16:07[Saint]We want him to come back! :P
18:18:53 Quit krabador (Quit: Sto andando via)
18:19:04***Saving seen data "./dancer.seen"
18:29:05NephielSubmitted http://www.rockbox.org/tracker/task/12887
18:30:23NephielIt doesn't allow me to add FS #11376 as related task but it's linked in the description
18:30:39gevaertsWhy didn't you just add it to FS #11376?
18:31:41gevaertsAnd especially, why does flyspray want to add lots of empty lines to everything it outputs? While it doesn't *really* break patches, it does break rss feeds...
18:31:58gevaertsIt didn't use to do that until quite recently
18:32:04[Saint]this seems like a new thing.
18:32:06*gevaerts looks for a Swede to look at this
18:32:14[Saint]aha, glad I'm not the only that noticed.
18:32:15Nephieloops, you're right, I should have added it to 11376
18:33:06[Saint]twice now I've clicked a flyspray patch link and sat there "waiting for the page to load"
18:33:14[Saint]...before realizing I needed to scroll.
18:33:43gevaerts[Saint]: well, for me it mainly breaks my new tasks and edited tasks "smart bookmarks"
18:34:43NephielI had the blank space issue too with patches, but screenshots are worse, they don't load at all
18:38:08gevaertsBagder: do you have any idea why just about anything in php on the server (flyspray, your blog, ...) adds loads of empty lines at the start of all output?
18:38:27gevaertsThat's fine for html, but it breaks anything els
18:38:28gevaertse
18:59:10 Join SovonHalder [0] (~RED@115.187.62.29)
18:59:11Bagdergevaerts: hm, no I don't but I figure I should look into it
18:59:51gevaertsBagder: you could also try to find Zagor of course :)
19:00
19:04:56Bagderthat's also a good idea! =)
19:06:00SovonHalderI posted a thread around 2 hours ago on the forum under 'Support and General Use>Hareware'....haven't received any replies yet..PLEASE help me..I am desperate.. The thread is this : Support and General Use
19:06:06SovonHalderhttp://forums.rockbox.org/index.php/topic,43310.0.html
19:06:25[Saint]2 hours is a VERY short time.
19:06:32copperSovonHalder: if all else fails, I suggest reformatting the iPod's drive with emCORE, and then refrain from renaming the drive
19:07:38copperdoes anyone know how Rockbox would react after changing the drive label on the PC?
19:07:56SovonHalderreally? all this is happening just because RENAME'd my drive name ?
19:08:32copperI don't know, but you yourself said that's when the problems started happening
19:08:45SovonHalderApparently yes..
19:08:55SovonHalderbut I hate to believe such thing
19:09:00[Saint]Rockbox shouldn't give a flying fudge about the volume label.
19:10:31copperany chance that Windows would be doing something else during that operation, that would explain it?
19:10:47copperor maybe it's really a Windows problem altogether?
19:10:58TheSevenSovonHalder: looks like there's file system corruption, possibly caused by not properly ejecting the ipod or by hardware malfunction
19:11:24TheSeventhe battery behavior that you reported is somewhat normal. rockbox only has a crude estimate of the battery state
19:11:25SovonHalderI have itunes installed..should I uninstall it ?
19:11:35SovonHalderNormal ?
19:11:40SovonHalderHoly jeez
19:11:57TheSevenso it can happen that it says 95% while it's actually 100%
19:12:06TheSevenand it won't even start charging before it drops below 85-90%-ish
19:12:08[Saint]Depends how you want to manage your music.
19:12:11SovonHalderfirst two days it worked fine...then suddenly it just like that!!!??
19:12:15[Saint]iTune sis no longer required.
19:12:38TheSeventhe charging is hardware controlled, so it will behave the same as it did with the apple firmware. it's just that our battery meter might be a bit off at times
19:12:46[Saint]I rather suspect that at some point you didn't properly eject the device.
19:12:59[Saint]This can, and often will, trash the filesystem.
19:13:21TheSevenespecially in combination with the rather flaky USB support of the classic
19:13:37SovonHalderSaint: I think I might have done that
19:14:15SovonHalderwhen the Drive disappeared, i couldn't find an option for safe removal...so I just ejected..
19:14:32copperSovonHalder: do you have a copy of your files on your PC?
19:14:38SovonHalder:)
19:14:40SovonHalderof course
19:15:11copperthen long press Menu + Select until you get to the emcore menu
19:15:22copperdownscroll until "Tools"
19:15:32copper"reformat data partition"
19:15:41copperthat should clean up the mess
19:15:42[Saint]Argh! No!
19:15:47copperwhy not?
19:15:50[Saint]menu+select == bad!
19:15:54copperwhat?
19:16:12TheSevencopper: shut it down properly if you can still do so (i.e. if it isn't completely locked up)
19:16:19[Saint]You should only do this if you can't avoid it.
19:16:33TheSevenmenu+select resets physically stress the hard disk drive
19:16:34coppersorry, I meant while off
19:16:34[Saint]Encouraging it as a shutdown/reset is bad behavior.
19:16:53[Saint]...reset the player, while off?
19:16:55copperhow do you avoid emcore going directly to rockbox?
19:16:58[Saint]why would that be needed?
19:17:01copperlong press play?
19:17:05[Saint]any button.
19:17:08TheSevenif you have fastboot enabled? just press and hold any button during boot
19:17:13copperok, sorry
19:17:27SovonHalderformatting data partition would recover system files? if they are corrupted?
19:17:35[Saint]Hell no. :)
19:17:47TheSevenSovonHalder: it will basically revert your ipod to the state that it was in immediately after installing emcore
19:17:52copperjust unpack the rockbox build zip onto the formatted drive
19:18:00[Saint]copper is possibly unwisely skipping that route.
19:18:11[Saint]You can of course attempt recovery of the filesystem.
19:18:26TheSevencan be tricky if USB is too flaky
19:18:36[Saint]see: attempt :)
19:19:12[Saint]I always like to try first, beets a format and shifting many GB of data around.
19:19:17SovonHalderSaint: so you were telling about system files corruption due to improper unsafe removal of my drive.. then I don't need to reformat data partition right ?
19:19:20[Saint]*beats too
19:20:10SovonHalderJust tell em what should I do now..
19:20:55TheSevenyou can try to repair the file system, but these reconnects might be preventing you from doing that
19:20:56[Saint]No. You don't *need* to - but, recovery may not work, if you do have your files backed up and the time isn't an issue, then by all means, just format it.
19:21:27[Saint]Its not something you need to do, necessarily, but it may be faster to do so in the end.
19:21:58[Saint]It just happens to be a pet peeve of mine when recovery of the filesystem isn;t even a first suggested option. :)
19:22:29SovonHalderDo I need need iTunes uninstalled ?
19:22:44SovonHaldersome apple services caused issues last time
19:22:52SovonHalderwhen I was installing
19:23:01[Saint]With installation. We're past that now.
19:23:01SovonHalderrockbox for the first time
19:23:25[Saint]If iTunes is how you manage your music, feel free to leave it there.
19:23:35SovonHalderI hate iTunes...
19:23:47SovonHalderthat's why i installed rockbox..
19:24:09SovonHalderI haven't played a single track in last 3 years in iTunes..so I'll uninstall it.. next ?
19:24:18SovonHalderhow do I format ?
19:24:47 Join kevku [0] (~kevku@2001:470:27:773:0:feed:c0f:fee)
19:25:01TheSevenSovonHalder: within the emcore boot menu, scroll to the right, there's a "tools" icon
19:25:13TheSevenin the tools menu, there's a "reformat data partition" option
19:25:39[Saint]it'll give some form of "are you sure?" message - tell it yes.
19:26:14SovonHalderreformatting dara parttion..
19:26:25SovonHalderIt's done..next ?
19:26:47TheSevennow just plug it in and check if it works again
19:27:07TheSeven(boot rockbox if you haven't done that already)
19:27:14TheSeventhen extract the rockbox zip file to reinstall that
19:27:16[Saint]TheSeven: no new build?
19:27:32[Saint]ah - yes: here - http://build.rockbox.org/data/rockbox-ipod6g.zip
19:27:42SovonHalderno rockbox directory−−−−installation complete
19:28:00TheSevenwell, be aware that new (as in younger than a year) builds are likely to show more USB trouble than older ones for some reason
19:28:00[Saint]this is expected, you just wiped it out :)
19:28:08[Saint]extract the above linked .zip to the root of the device.
19:28:30SovonHalderw8
19:28:35TheSevenand then just shut it down and boot it back up, then you're running a current rockbox version
19:28:54[Saint]RbUtil could also take over at this point if you wanted.
19:29:29[Saint]RoLo should fire, no need for shutdown.
19:30:00SovonHalderit took 12 seconds to show the drive
19:30:02SovonHalderis this normal ?
19:32:15SovonHalderhow do I shut it down?
19:32:22SovonHalderlong press the play button ?
19:32:23[Saint]hold play.
19:32:27SovonHalderokay
19:33:51SovonHalderVersion: 49bcf35-130724
19:34:57SovonHalderCan I use this theme?
19:34:58SovonHalderhttp://themes.rockbox.org/index.php?themeid=1883&target=ipodvideo
19:35:06[Saint]yes.
19:35:15SovonHalderI mean do themes create problems ?
19:35:18[Saint]you can use any of the themes from the iPod Video section.
19:35:26[Saint]they can, yes.
19:36:12SovonHalderreally? I mean are those problems often? If so, I can hang on to the default theme..
19:36:23SovonHalderIy's just the theme is simple
19:36:55[Saint]Themes are user submitted - there is no expectation of quality with them.
19:37:27SovonHalderBy quality did you mean 'stabily' or 'compatibility' ?
19:38:19[Saint]We can only guarantee the code parses, not that it is correct, or functions as intended.
19:39:21SovonHalderI'm sorry for bothering you all.. my bad.. but I have to transfer over 60GBs of music to my drive now. That's not something one can do whenever he wants..so any words on how do I prevent happening such issues in furute ?
19:39:42SovonHalder*future
19:39:52[Saint]Always safely remove.
19:39:53lebellium[Saint]: although what you say is correct, it seems that you are frightening him :) I would simply say "they may create problems but most of them don't and if there is any problem with a theme you can always switch back to the default cabbiev2"
19:40:14copperSovonHalder: I haven't had problems with that theme
19:40:44SovonHalderCooper: then I'll stick with you..
19:40:54[Saint]lebellium: I'm just trying very hard not to say "Yes, it will work as advertised", because I can't make that claim :)
19:41:38[Saint]I never said *don't* try it...
19:41:38SovonHalderplease tell me do I prevent happening such issues in future ?
19:42:03copper17:39:53 UTC <[Saint]> Always safely remove.
19:42:05[Saint]The only thing you can do, really, is always safely remove the device.
19:42:06SovonHalderrather than safe removal
19:42:11[Saint]that's it.
19:42:12SovonHalderokay..
19:42:29SovonHaldercan I rename the drive label /
19:42:31SovonHalder?
19:42:48[Saint]In theory, yes.
19:45:12SovonHalderI just connected it to my PC & it took 30 seconds to sho me drive.. is this also normal ?
19:45:52TheSeventhat can be normal
19:45:52copperWindows can be sucky that way
19:46:00TheSevenwindows sometimes reads the whole FAT while mounting
19:46:25TheSevenwhich is ~320MB on a 160GB ipod
19:46:35TheSevenso 30 seconds sounds about right
19:49:02SovonHalderI just renamed. ..nothing wrong happened.. went A OK..
19:49:31*Nephiel just commited a fix for FS #11376 to gerrit: http://gerrit.rockbox.org/r/#/c/530/
19:54:46copperwell fuck.
19:55:06coppermy Fuze+ just crashed while connected to my laptop, after hours of copying files
19:55:13coppernothing else connected
19:55:32SovonHalder:(
19:55:36 Join Strife89 [0] (~Strife89@adsl-98-80-224-47.mcn.bellsouth.net)
19:56:25SovonHalderI am not a jinx..am i? :P
19:56:46coppernow fsck.vfat is "Reclaiming unconnected clusters." and it's taking 100% of one CPU core and I have no idea when it will complete
19:56:48[Saint]No. It just blew a prior assumption away, that's all.
19:57:00copper[Saint]: it doesn't
19:57:15copperthe ODAC being connected is certainly a trigger
19:57:25copperthat doesn't mean it is the only cause of Rockbox crashing
19:57:29[Saint]you said your devices *never* crashed without the ODAC
19:57:45copperyeah
19:58:14[Saint][01:28:40] <copper> pamaury: my devices almost always crash when my ODAC is connected
19:58:14[Saint][01:28:44] <copper> and NEVER crash when it's not
19:58:17copperstill doesn't erase the fact that they almost always crash with the ODAC connected
19:59:05SovonHalderumm...what is the full form of ODAC ?
19:59:13coppergeez, what's the point of running fsck with -v when there's no progress
19:59:20copperSovonHalder: what?
19:59:33SovonHaldernevermind..sorry
20:00
20:03:23NephielSovonHalder: I think it's a mix of Objective Desktop Amplifier and Digital-Analog Converter
20:05:56SovonHalderwhen I click on safely remove 'Apple iPo...' in system tray..it shows 'Safe to Romove Hardware' but the USB logo on iPod classic screen doesn't disappear.. then awhen I remove the cable there showed a tiny warning for 12 seconds that somewhat says 'Error Accessing Playlist control file' or something like that...what is this issue ? I haven't copied a single track yet
20:07:15[Saint]The USB image is still there. because its still connected. :)
20:07:39copperugh, copying to my 64 GB microSDXC card through the adapter and my laptop's card reader is twice as fast as copying via the Fuze+
20:07:44[Saint]It doesn't mean "I'm connected, and mounted", it merely means "I see I'm connected via USB"
20:12:55 Join Guest11127 [0] (Slayer@c-69-143-178-62.hsd1.va.comcast.net)
20:16:16 Join krabador [0] (~krabador_@unaffiliated/krabador)
20:19:07***No seen item changed, no save performed.
20:28:23TheSevenis there anybody here for whom UMSboot fails to mount properly on nano2g?
20:28:35TheSevenI need testers for the new one
20:29:29SovonHalderWell, how long the USB Image stays? It's about 3 minutes+ since I saw 'Safe to Remove' banner.
20:30:20[Saint]Is it still connected via USB?
20:30:45SovonHalderyes
20:30:52SovonHalderI haven't unplugged it yet
20:30:57[Saint]...did you read what I wrote above?
20:31:03[Saint]It doesn't mean "I'm connected, and mounted", it merely means "I see I'm connected via USB"
20:31:15SovonHalderOo..sorry..
20:31:20SovonHalderI missed tat line
20:31:23[Saint]the device will give no visual indicator it is safe to eject.
20:31:46SovonHalderwhat about that 'Error Accessing Playlist control file' then ?
20:32:16[Saint]I believe that is expected for a first boot.
20:34:50 Quit krabador (Ping timeout: 260 seconds)
20:39:10SovonHalderI disconnected.. turned of my ipod..turned on my ipod..connected via usb−−-again shows 'Error Accessing Playlist control file'
20:39:31SovonHalderI safely removed..connected again... still shows 'Error Accessing Playlist control file'
20:40:09TheSevenSovonHalder: how old is that ipod?
20:40:47SovonHalderit's iPod Classic 7G 160G.. I bought it around 4 months ago Brand Newq
20:40:51SovonHalder*New
20:44:31TheSevenfully working (on my device and ubuntu precise) new umsboot source: https://mega.co.nz/#!4Iw3USyD!EK2O-Q-jdumd6xYPeb1FVwfyh0x-tR0ljUvA2AfGU7M
20:44:38TheSevenno lcd driver yet, but the rest works just fine
20:44:59TheSevenSovonHalder: OK, so I hope we can rule out HW damage
20:45:14TheSevenI've seen numerous hard drives fail on the 80GB and 120GB series
20:46:17SovonHalderthere are about Eight threads on this particular isue on rock box forum saying "" 'Error Accessing Playlist control file' on my bla bla bla""
20:46:52SovonHaldereveryone says it's not that important
20:47:00SovonHalderis that right ?
20:47:42 Join krabador [0] (~krabador_@unaffiliated/krabador)
20:49:46 Join lorenzo92 [0] (~chatzilla@host171-106-dynamic.17-79-r.retail.telecomitalia.it)
20:54:05 Join fs-bluebot [0] (~fs-bluebo@g225252157.adsl.alicedsl.de)
20:57:10lorenzo92kugel: did you read about usb suppport on hosted platforms?
20:57:24lorenzo92(if you had time :) )
21:00
21:05:06 Join thegeek_ [0] (~thegeek@40.200.16.62.customer.cdi.no)
21:05:41 Quit thegeek (Read error: Connection reset by peer)
21:05:42 Join rdn [0] (~oop@cpe-69-204-124-212.buffalo.res.rr.com)
21:06:18 Quit Nephiel (Ping timeout: 246 seconds)
21:10:57saratogacopper: i guess the clockspeed is reported wrong in plugins?
21:13:28saratogawhat is the boosted clock speed on the Fuze+?
21:14:37[Saint]~450MHz
21:15:14[Saint]maybe plugins expect the clockspeed in Hz?
21:15:30[Saint]the F+ (weirdly) reports in KHz
21:18:03copperno clue
21:18:20*[Saint] thinks he may have solved that riddle.
21:18:39[Saint]That would account for the really low realtime requirement
21:18:48saratogathe fuze+ memory is really slow
21:19:02saratogacopper: feel like doing one more benchmark?
21:19:11saratogaBagder: around?
21:19:18[Saint]saratoga: you think its KHz vs Hz?
21:19:23saratogayeah looks like it
21:19:29saratogathe numbers are off by a factor of 1000
21:19:34[Saint]the debug screen in the F+ reports in KHz
21:19:43[Saint]..no idea why.
21:19:45saratogathe define is probably wrong on the fuze+
21:19:54TheSevenSovonHalder: it is in most cases an indicator of rockbox not being able to access the file system at all (or write it), which typically IS rather important
21:20:06kugellorenzo92: no, where?
21:20:24copperwhat?
21:20:36copperwhat numbers are wrong?
21:20:43saratogacan you run the benchmark one more time with the boost option disabled?
21:20:48[Saint]CPU realtime
21:21:01copperuh?
21:21:04saratogathis will mean that the CPU clock is much lower relative to the memory (or equivalently that the memory seems much faster)
21:21:12saratogai believe there is an option in test_codec to do this
21:21:16[Saint]No idea why such foolishly low numbers didn;t arouse suspicion in me earlier.
21:21:29[Saint]now I look at it, its so improbable I should've caught it.
21:21:30coppercpu real time is within range of the other targets
21:21:48[Saint]0.01MHz?
21:21:57copperah
21:21:59TheSevenless than one clock per sample, funny :)
21:22:04copperithought you meant %
21:22:22[Saint]I *really* should've caught that earlier.
21:22:26copperyes
21:22:29copperI blame you
21:22:36[Saint]that's fair. :)
21:23:13coppersaratoga: I'll ping you in about 40 minutes
21:23:27saratogayeah the numbers are fine, but they're just in units of GHz not MHz :)
21:23:33saratogaok
21:23:38saratogaalso i have Opus files
21:23:48saratogahttp://web.mit.edu/mgg6/www/opus/
21:23:55copperso, * 1000?
21:24:00saratogai've been trying to get Bagder to upload these for months
21:24:16saratogayeah, but just do 454/realtime*100
21:24:27copperwait, what?
21:24:42saratogaMHz = 454/%realtime*100
21:25:10coppermmmkay?
21:25:20saratogai can fix that later
21:25:32saratogai have a script which formats them for the wiki
21:25:35saratogai can have it do the math itself
21:25:59saratogaanyway since the fuze+ never boosts, the boosted test results are probably misleading
21:26:14saratogamost likely when the CPU is unboosted the memory latency is far lower, and therefore much less cycles are needed
21:26:40[Saint]you can force an unboosted test_codec run, no?
21:26:53saratogayeah i think its in the test_codec plugin
21:27:00saratogaIIRC i added it years ago
21:27:10saratogaor at least i think i committed that
21:27:14saratogaif its not there then my bad :)
21:27:28 Join Nephiel [0] (~5332f9b8@www.haxx.se)
21:28:08saratogalooking at these results though our MDCT lib does really, really badly on modern targets
21:28:15saratogawe definitely over-optimized it for PP
21:29:11saratogastripwax came up with all these neat math that makes the trig tables small enough to fit into IRAM, but they end up having really irregular access patterns which probably ruins caching on devices without IRAM
21:29:13 Quit SovonHalder ()
21:29:13copperdid you see my Classic and Clip+ results?
21:29:18saratogayeah i saw them above
21:29:29copperok
21:29:49saratogado you have the build numbers?
21:29:54saratogafor the wiki
21:31:23copper49bcf35
21:31:53saratogathanks
21:32:16saratogadamn it forgot that my perl script was converted to ruby at some point
21:32:20saratogahave to figure out how to run that
21:34:06 Quit krabador (Ping timeout: 260 seconds)
21:34:10 Nick DormantBrain is now known as SuperBrainAK (~andy@shared02.balt01.cd.2g2u.net)
21:37:08lorenzo92kugel: well there is something bad happening when connecting the usb cable
21:37:37lorenzo92kugel: i.e. app crashes, for example when doing a screendump or something else that is not a stub in usb_enable
21:38:07lorenzo92now I have setup the serial, I suspect a concurrency issue or something related (memory issue)
21:40:26saratogacopper: http://www.rockbox.org/wiki/CodecPerformanceComparison#Freescale_i.MX21_40454MHz_ARM926EJ_45S_44_Fuze_43_41
21:41:41coppernice
21:43:19 Join krabador [0] (~krabador_@unaffiliated/krabador)
21:44:00lorenzo92kugel, pamaury: I confirm, segmentation fault when inserting usb cable and right after doing screendump
21:44:12lorenzo92on hosted target ypr0 at least...
21:51:05saratogahmm our ruby script cannot handle comparing some of the old test codec logs to the newer ones
22:00
22:00:01saratogaaccording to these tests, vorbis has gotten slower in recent years
22:00:03saratogai wonder why
22:00:08saratogaon AMSv2 at least
22:00:22lorenzo92kugel: funny, with debug symbols + logf usb screendump doesn't crash when inserting cable, but when removing it \o/
22:02:21lorenzo92also funny that gdb doesn't work now pff
22:06:32 Quit Nephiel (Quit: CGI:IRC)
22:07:29coppersaratoga: what do you want me to benchmark?
22:07:53saratogacan you grab those Opus files and add them to your folder?
22:08:07saratogaand then repeat the test, but this time set the boost to off so it tests at 64 MHz?
22:08:16saratogai think there should be an option in test_codec itself to do this
22:10:18coppersaratoga: I think you forgot wavpack x4 on the wiki
22:12:22saratogathe ruby scirpt ignores it for some reason
22:14:37kugellorenzo92: probably too small stack
22:15:04lorenzo92kugel: hum and what value do you suggest?
22:15:29kugelwhich thread does the screendump run on?
22:16:00lorenzo92"usb"
22:16:33lorenzo92but i'm not sure it's due to small stack
22:16:35kugeltry 1MB or so
22:16:39lorenzo92ok
22:17:13saratogakugel: did you write the test codec ruby script?
22:17:36kugelyea
22:19:01 Quit Raptors (Read error: Connection reset by peer)
22:19:09***Saving seen data "./dancer.seen"
22:19:12saratogaany idea why it can't parse "wv_normx4.wv" but works with "wv_fastx3.wv"
22:20:24lorenzo92kugel: interesting, with exactly 1 mb it quits with "Killed"...seems like out of memory
22:21:28 Quit y4n (Quit: Assumption is the mother of all fuckups)
22:23:01lorenzo92kugel: 512k, doesn't get killed at startup, but also crashes
22:24:54kugelhm. can you get the backtrace from gdb?
22:28:17coppersaratoga: running new tests now with opus files included, on the Fuze+, Clip+ and iPod Classic
22:28:28saratogathanks
22:28:30copperI selected boosting=no on the FUze+ and the Classic
22:28:34saratoganice
22:28:36copperI don't see that option on the Clip+
22:28:37saratogathose will be interesting
22:28:42saratogayeah the clip+ has boosting disabled
22:28:53saratogait was a bit unstable and we never figured out why
22:29:17lorenzo92kugel: i have gdb running atm, what should i do?
22:29:26kugellorenzo92: I dont think concurrency is it, thats generally not a problem with out cooperative thread architecture
22:29:30copperugh, it's a lot slower
22:29:36copperthis is going to take ages
22:29:57kugellorenzo92: get a backtrace when it crashes
22:31:13lorenzo92kugel: http://pastie.org/private/6fk6heneazyjqcb1511bg
22:31:38kugelthat's without debug symbols right?
22:31:51kugelyou can look up the addresses in the .map file
22:32:27lorenzo92yep no symbols
22:33:08kugelyou can also try to use arm-*-addr2line to map the addresses to source file lines, but that needs a non-stripped .elf (there should be one in the build dir)
22:34:13lorenzo92yes there is the elf, i'll give a try, previous addresses don't tell anything interesting
22:34:59 Join stripwax [0] (~Miranda@rockbox/developer/stripwax)
22:35:49lorenzo92kugel: crtstuff.c ?
22:36:43kugelwhat command did you run?
22:36:59lorenzo92arm-ypr0-linux-gnueabi-addr2line -e rockbox.elf -a 0x00050624
22:37:32lorenzo92ahh
22:37:34lorenzo92wait
22:37:42lorenzo92with -f i also get usb_thread
22:37:43lorenzo92;)
22:38:30kugelwhere is crtstuff.c?
22:38:31lorenzo92kugel: and also queue_wait
22:38:57lorenzo92kugel: http://pastie.org/private/sjzhpkl8dhnaog77aujlw
22:39:57kugelah it seems addr2line can't properly map to the source files
22:40:18kugelperhaps it gives a more useful file if run from the top-level folder?
22:42:35lorenzo92nope, but it might be fine, no? actually queue_wait is used in usb_thread
22:43:43kugelyea but you have to look up the disassembly to find the actual line
22:44:11kugelI dont know how queue_wait can possibly crash.
22:44:36 Join Zarggg [0] (~zarggg@24.229.140.62.res-cmts.sm.ptd.net)
22:46:26 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
22:48:44 Join tertu [0] (~tertu@65-128-181-81.mpls.qwest.net)
22:55:22lorenzo92kugel: okay, so it seems that the crash happens right in the first lines, address points to a load instruction
22:58:40coppersaratoga: if the Clip+ isn't "boosted", why is it so much faster than the other two?
22:59:05copperor is it boosted and what didn't work was, erm, "unboosting" it?
23:00
23:00:02kugellorenzo92: perhaps called with a NULL pointer?
23:00:22lorenzo92i was thinking the safe, i'll place a temporary panic in case, let's see
23:00:26lorenzo92*same
23:04:29coppersaratoga: Clip+ codec performance log: http://pastebin.com/yHV8c2Gy
23:05:14 Quit Strife89 (Ping timeout: 264 seconds)
23:05:18copperthe other two are maybe halfway through
23:14:17lorenzo92kugel: now stops at switch_thread()
23:19:45coppersaratoga: opus 128k on the Classic, unboosted: 99.55% real time :-/
23:24:46lorenzo92kugel: unfortunately i can't really figure out what's going on atm :(
23:25:27lorenzo92i only hope it's not a fault that gets undiscovered in native ports....
23:28:28ender`are there any known problems with USB on Clip Zip in current git version? Windows either reports an error, or doesn't detect the player at all when running rockbox for me
23:32:42saratogayeah it doesn't seem to be all that stable
23:32:54saratogacopper: that means it'll very rarely boost, which is not too bad
23:33:24saratogacopper: are you doing this tests with or without the DSP effects (hopefully without)
23:33:26ender`ok. is this only a problem in git, or is it the same in stable, too?
23:33:37copperwithout
23:33:47copperwell
23:34:08kugellorenzo92: can you show how you increased the stack to 512k?
23:34:20copperall DSPs are disabled in the normal Rockbox menus, and I didn't select anything else
23:34:38kugelif it crashes in different places it really looks like a stack overflow
23:34:43lorenzo92kugel: well now I was running on default stack size...anyway just 512*1024
23:34:52copperhmm
23:35:20lorenzo92kugel: wops is a long
23:35:23lorenzo92^^
23:35:40kugelso it was 2MB?
23:35:44saratogacopper: isn't there an option test_codec that asks if you want to benchmark with DSP too?
23:35:45lorenzo92yep
23:36:04lorenzo92i forgot the /sizeof(long)
23:36:05copperyes but I didn't select that
23:36:27copperI ran test_codec folder
23:37:14lorenzo92kugel: but then the question remains the same (not only the song), why the hell if i place a sleep or printf around i may be lucky and it don't get a crash?!
23:38:02kugeldont know
23:38:43kugellorenzo92: are you sure screendump is called from the ui thread?
23:38:47kugelusb thread*
23:39:17kugelhosted doesn't have screendump so far, for native it's on the usb thread but for sims it's on a separate one
23:39:31kugelnow I don't know how you implemented it
23:39:48lorenzo92yeah, i've just double checked...it's an inline function that calls screen_dump
23:40:06lorenzo92well actually screen dump works perfectly, there was just a path problem i solved in another patch
23:40:50 Join Raptors [0] (~whoneedsa@216-58-33-203.cpe.distributel.net)
23:43:10lorenzo92kugel: the point is that even if I fill something in usb_enable (to say a system() call or similar), it crashes ...
23:43:54kugelmight be OOM, if rockbox claims all memory there's little left for system()
23:44:08kugelremember that systems spawns several processes (at least 2)
23:44:45lorenzo92pretty sure (i tested it long time ago) that issuing a system() through a menu button perfectly works
23:45:50lorenzo92or it may be that usb thread is heavy and memory badly managed, who knows...
23:46:44kugelcan you do without system()? what are you trying to execute?
23:47:14 Quit Raptors (Read error: Connection reset by peer)
23:47:30 Join Raptors [0] (~whoneedsa@216-58-33-203.cpe.distributel.net)
23:47:45lorenzo92forgetting about system(), there is still screendump...anyways I have to call a safe-mode-like script for usb mode
23:49:05lorenzo92kugel: idea! what if trying to lower memsize a little?
23:49:27kugeltry it, you can adjust it in configure
23:49:52lorenzo92indeed, compiling
23:53:16lorenzo92kugel: nope, nothing
23:55:25 Join jrw_ [0] (~601cd28a@www.haxx.se)
23:57:51 Quit jrw_ (Client Quit)

Previous day | Next day