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

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

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

#rockbox log for 2010-05-19

00:00:18evilnick_BJust to rule out any compiling problems from your machine/setup
00:00:19amiconnI know where the table with the OF values is (which we can't use, at least not for all targets, as we do set up some other values in a better way), but not where the code is that uses the table
00:01:08max242there is already one: FS #11267, pretty much describing what i have, but i think it's title should be rephrased
00:01:13amiconnUnfortunately I forgot to document that when finding it a long time ago :(
00:01:53gevaertsmax242: is this with an official build?
00:03:19 Quit Boldfilter (Quit: Boldfilter)
00:03:59max242gevaerts: issue got started since r25299, and is pretty persistant ...
00:04:16gevaertsThat's not what I asked :)
00:04:24evilnick_BBut is that from an official build?
00:04:32evilnick_BOr from your own compiled version?
00:06:25max242evilnick_B: from the 'current-builds' part of the rockbox website
00:07:14max242i'm running r26133 right now
00:07:17evilnick_BTHAT's what we needed, thanks
00:07:31max242you're welcome
00:08:53 Quit Battousai (Read error: Operation timed out)
00:09:02 Quit jgarvey (Quit: Leaving)
00:09:19max242i also posted a message on this in the forum:
00:10:08max242is bor_ka sometimes on irc, he seems to have also lockup issues which got started since r25299
00:10:35 Join Battousai [0] (~bryan@gentoo/developer/battousai)
00:11:14max242as he originated the flyspray ticket FS #11267
00:12:39 Quit phi1ippe (Quit: leaving)
00:13:04 Quit stripwax (Read error: Connection reset by peer)
00:13:17max242i'm off to bed now, i will pester you guys later ;-)
00:13:59 Quit max242 (Quit: CGI:IRC (EOF))
00:17:52 Quit efyx (Remote host closed the connection)
00:21:24 Quit lpereira (Quit: Leaving.)
00:25:40 Quit notlistening (Ping timeout: 248 seconds)
00:38:28 Quit merbanan (Ping timeout: 260 seconds)
00:39:28 Quit DerPapst (Quit: Leaving.)
00:44:47 Quit ender` (Quit: Anyone who thinks people lack originality should watch them folding roadmaps.)
00:47:17 Part domonoky
01:07:41 Quit pamaury (Quit: Page closed)
01:08:25 Quit solexx_ (Read error: Connection reset by peer)
01:08:27 Join solexx [0] (
01:21:38 Part toffe82
01:24:45 Join n1s [0] (~n1s@rockbox/developer/n1s)
01:30:39ssorgatem_FS #11293 - Catalan language update
01:34:44ssorgatem_mm, I'm unable to apply the diff from FS #11293, all 16 hunks fail to apply
01:35:31ssorgatem_why is it?
01:35:50Torneif every chunk of a diff fails it usually means you have the wrong value for -p
01:38:09ssorgatem_Torne: true
01:38:35ssorgatem_Torne: I suppose it's already time to go to bed
01:44:35 Quit linuxstb (Ping timeout: 276 seconds)
01:44:52 Join Pikidalto [0] (~Piki@unaffiliated/piki)
01:45:24Pikidaltohi, i was just wandering if there were any plans to port to the Philips GoGear SA32XX series
01:45:53n1sPikidalto: we don't make plans for ports
01:46:29n1sand i have not heard that target mentioned as one someone is investigating or working on a port for
01:46:35Pikidalton1s: so basically you just go ahead do the ports if it's possible?
01:47:28Pikidalton1s: yes, i was afraid of that, i had not seen any mention of it on the site
01:47:54n1swell, usually an interested owner of a player starts a port, after the groundwork is done (a way to execute custom code) it's not uncommon that others join in
01:50:02CIA-5New commit by jethead71 (r26154): Gigabeat S: Implement LCD contrast, invert and flip modes. Enhance LCD power management. Include init data but it's not needed yet (identical to ...
01:50:22Pikidaltoi'm no programmer, so i wouldn't know where to start, other than posting information on the player and pics of the inside (my mother is interested in rockbox just for the cool Plugins, but i doubt she'd like having a disassembled GoGear lol) :-(
01:51:50n1sPikidalto: the most important information about the hardware is what kind of processor or System on Chip is in the player, if it is one we already have working drivers for a port is usually substantially easier
01:52:32***Saving seen data "./dancer.seen"
01:52:43Pikidalton1s: i no longer have the packaging or manual, but i could try to find it on google
01:53:07Tornethe packaging or manual are extremely unlikely to say what the processor/soc are
01:53:19n1sthat's not usualy mentioned in the manual, but if you can find pictures of the inside they can be useful
01:53:27Torneand a lot of the google results for that kind of thing for various players point at us.. ;)
01:54:45Pikidaltoany ideas how to get the CPU/System information without popping open the case?
01:55:24n1swell, it can sometimes be found out from a firmware update
01:57:53 Quit evilnick_B (Quit: Page closed)
01:59:20n1shmm, i'd guess it's a STMP3500
02:00:11Pikidaltofor my GoGear? currently trying to do a firmware update
02:00:39Torne*doing* a firmware update won't tell you anything, probably..
02:00:43Tornelooking at it might :)
02:01:11 Nick fxb is now known as fxb__ (
02:01:23Pikidaltotrue, i guess, only probelm is *getting* the firmware file
02:02:00 Quit bieber (Ping timeout: 260 seconds)
02:02:15 Join bieber [0] (
02:02:38n1sPikidalto: the product brief for the gogear sa3285 shows that it supports "SMV" video which seems to be Sigmatel Motion Video which they have in their stmp35xx sdk
02:03:01n1salso it lists support for mp3 and wma only, which is ey common for players based on this SoC too
02:03:24Pikidaltothis one specifically is a 3245/37, but it shouldn't be much different
02:03:40Pikidaltoguess i should read the lingo page :-)
02:04:15n1sSystem on Chip
02:05:30n1swell, the stmp35xx SoC's have a dsp core (no cpu) which AFAIK ther's no free compiler for so a port of rockbox for this is *very* difficult (read not going to happen)
02:08:04n1sjhMikeS: any measurable battery time improvement from that lcd power management improvement?
02:08:15n1seh s/life/time/
02:09:49jhMikeSn1s: haven't had a chace to really test but it's the same thing as the F, so I imagine given that it's not insignifant, it won't hurt anything.
02:10:41n1sjhMikeS: ok, btw, do you know if the pmic can tell the current battery current?
02:10:47 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
02:10:59jhMikeSnoone else seems to actually producing benches lately and I've done my share already. :)
02:11:11n1show much current is sucked from the battery that is
02:11:27n1sjhMikeS: heh, i'll do one tomorrow then!
02:11:37jhMikeSn1s: it *can* if it's wired for it, but that sense is shorted. only the charging sense works.
02:15:08 Join Rob2222 [0] (
02:17:48*jhMikeS better redump the IPU since he thinks he just add a bit set that need not be there.
02:18:48 Quit Rob2223 (Ping timeout: 265 seconds)
02:24:48Pikidaltooh wow, i'm in Live Support for philips, they don't have the specs on their own MP3 Player
02:25:13Pikidaltomore like, it's a closely guarded secret
02:28:31 Join FlynDice [0] (
02:32:21 Quit MagusG (Read error: Connection reset by peer)
02:33:16 Join MagusG [0] (
02:33:16Pikidaltowell i'm out, thanks for the time guys, i'll be back if i find more info on this player
02:33:20 Part Pikidalto ("Once you know what it is you want to be true, instinct is a very useful device for enabling you to know that it is")
02:33:22 Join Xqtftqx [0] (
02:33:25XqtftqxYO WASSUP
02:34:03 Join n00b81 [0] (~185b52cd@gateway/web/freenode/x-zztpchmvaxlwmibh)
02:34:22Xqtftqxlogbot, this message will go down in history
02:34:34XqtftqxWhat i am typing now, will always remain in the rockbox logs
02:35:34 Join kramer3d [0] (~kramer@unaffiliated/kramer3d)
02:36:15XqtftqxWelcome to #rockbox kramer3d! I am the automated welcome bot! Please read the rules before chatting!
02:36:31kramer3dno youre not
02:36:32kramer3di know you
02:36:49XqtftqxOh really
02:37:02kramer3di also know your step brother Mark
02:37:02XqtftqxHow so?
02:37:09XqtftqxI dont have a step brother.
02:39:55 Quit DataGhost (Ping timeout: 240 seconds)
02:41:47 Quit n1s (Quit: Lmnar)
02:42:56CIA-5New commit by jethead71 (r26155): Gigabeat S: Wrong thing messing with IPU_CONF. No fiddling is needed just yet.
02:47:44 Quit cdb (Ping timeout: 260 seconds)
02:51:00 Nick Ypsy is now known as YPSY (
02:59:05kisakoh ... an overlooked detail on the main page of
02:59:51kisaksansa clip and fuzev2 is double listed ... they need to be removed from unusable ports
03:00:30 Quit kramer3d (Quit: Leaving)
03:02:29 Part n00b81
03:10:35CIA-5New commit by mc2739 (r26156): Remove clip and fuzev2 from unusable ports
03:15:03 Join ischeria1 [0] (
03:18:30 Quit ischeriad (Ping timeout: 264 seconds)
03:33:28mc2739B4gder or Zagor: Could you update the web server for a main page change? Thanks
03:52:34***Saving seen data "./dancer.seen"
04:07:25 Quit TheSeven (Ping timeout: 240 seconds)
04:11:40 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
04:15:48 Quit pixelma (Disconnected by services)
04:15:48 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
04:15:52 Quit amiconn (Disconnected by services)
04:15:54 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
04:16:08 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
04:16:13 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
04:22:03 Join CGL [0] (~CGL@
04:24:21 Quit Xqtftqx (Ping timeout: 264 seconds)
04:50:34 Join cdb [0] (
04:54:50 Quit panni_ (Quit: ( :: NoNameScript 3.81 :: ))
05:04:52 Quit BlakeJohnson86 (Quit: Leaving.)
05:09:32 Join BlakeJohnson86 [0] (~bjohnson@2002:1876:a27b:0:227:13ff:fe65:1262)
05:10:45 Quit wincent_balin (Ping timeout: 240 seconds)
05:17:27 Join LambdaCalculus37 [0] (~rmenes@rockbox/staff/LambdaCalculus37)
05:28:12 Quit Horscht (Quit: Verlassend)
05:50:00 Quit Unhelpful (Ping timeout: 240 seconds)
05:52:35***Saving seen data "./dancer.seen"
06:04:33 Quit LambdaCalculus37 (Quit: Fwump)
06:10:11 Join Unhelpful [0] (~quassel@rockbox/developer/Unhelpful)
06:11:58*S_a_i_n_t is plagued by "need to hard-reset to turn the DAP (2 different iPod Nano1Gs) on" today...
06:12:57S_a_i_n_tBetween the two of them, it's happened ~10 times today, and I find it a little too much of a coincidence that they both ended up with "white screen of death" trying to recover from it.
06:14:33S_a_i_n_tRestored, re-installed RB...haven't seen the problem again yet...*yet*.
06:28:54 Quit anewuser (Quit: for SELL 2 by the price of 1 now!)
06:36:39 Join webguest32 [0] (
06:37:22 Quit webguest32 (Client Quit)
06:37:31 Join webguest94 [0] (
06:40:13 Join pupuserb7dc64 [0] (
06:42:53 Join saratoga [0] (~9803c6dd@gateway/web/freenode/x-zzmofwwvlikygwyn)
06:43:03 Quit webguest94 (Quit: CGI:IRC (Ping timeout))
06:44:23 Part pupuserb7dc64
06:51:24 Quit mikroflops (Ping timeout: 240 seconds)
06:53:33 Join mikroflops [0] (
07:07:27 Join stoffel [0] (
07:19:18 Join puetzk [0] (
07:19:47 Quit S_a_i_n_t ()
07:20:06 Join S_a_i_n_t [0] (S_a_i_n_t@
07:23:13puetzkIs there a known issue with the recording gain on AMSv2 Sansas (specifically a Clip+). The microphone is extremely quiet, and if I turn up the gain it starts clipping long before reaching a reasonable volume. I didn't find anything obvious in Flyspray or the wiki (which said recording worked).
07:23:45 Join gurenko [0] (
07:25:59gurenkoanyone knows if there are any news about rockbox for sansa fuze v02?
07:26:29puetzkBesides what's in the wiki at SansaAMS?
07:26:40gurenkoanyone knows if there are any news about rockbox for sansa fuze v02?
07:26:47 Quit gurenko (Client Quit)
07:26:55 Join gurenko [0] (
07:27:30gurenkohi! anyone knows if there are any news about rockbox for sansa fuze v02?
07:27:41puetzkyou've asked that three times
07:28:05gurenkobut computer freeze
07:28:54puetzk got updated today, so that's probably pretty good status
07:28:54puetzkI just put it on a Clip+, which is more or less the same hardware
07:29:54gurenkook, thank you
07:30:12gurenkoi'm going to see it
07:31:38puetzkI'm having a little trouble with the recorder (it's very quiet), but so far rockbox is faring better than the OF, which wants to reread tags every time it boots (and seems to hang while doing so). YMMV
07:34:16 Join katsuru [0] (
07:37:16katsuruHello every1 i am trying to install rockbox on my Clip+, i am on ubuntux64 and i cannot execute mkamsboot from the terminal...
07:37:27 Join notlistening [0] (
07:37:47katsurukatsuru@katsuru-desktop:~/Desktop/rbinstall$ sudo ./mkamsboot clppa.bin bootloader-clipplus.sansa pfile.bin
07:37:47katsurusudo: ./mkamsboot: command not found
07:37:53katsuruany help please
07:38:13FlynDicekatsuru: did you make mkamsboot yet?
07:38:13puetzkare you in the directory where you downloaded it?
07:38:21katsuruyeah i am
07:38:26katsururbinstall in desktop
07:38:39puetzkis the file marked executable? (chmod a+x mkamsboot if not)
07:39:07katsurupuetzk, worked
07:39:09katsuruthank you ;)
07:40:19katsurupuetzk, on the website they say i gotta paste the output file on the root of my device... by that they mean on the SDcard right?
07:40:40puetzkNo, the device itself should go to mass storage mode and mount as a filesystem
07:40:54puetzkthe expansion card is separate, and I don't think it will firmware update from that
07:41:16puetzkmine called itself "SANSA CLIPP" as the volume label
07:41:25katsuruso is mine
07:41:46puetzkbut that's the internal flash, not the uSD card
07:41:53puetzk(if you have one in)
07:42:17katsuruohhh damn right... mine doesn't even have a card.... i am sorry man
07:42:27katsurui am just used to have my 2Gb one connected to the PC
07:42:35 Quit gurenko (Quit: CGI:IRC)
07:46:32 Quit notlistening (Ping timeout: 264 seconds)
07:48:00puetzkI just tried in the OF, and recordings have a normal volume there
07:52:37***Saving seen data "./dancer.seen"
07:53:16 Join puetzk_ [0] (
07:54:30puetzk_so the quiet recordings to appear to be something rockbox is setting up differently
07:54:44 Quit puetzk (Disconnected by services)
07:54:44 Nick puetzk_ is now known as puetzk (
08:02:02 Join kramer3d [0] (~kramer@unaffiliated/kramer3d)
08:03:05 Quit stoffel (Remote host closed the connection)
08:06:01 Quit puetzk (Remote host closed the connection)
08:06:08 Join puetzk [0] (
08:06:48 Quit puetzk (Client Quit)
08:19:53 Quit kramer3d (Read error: Connection reset by peer)
08:21:30 Join Szpila [0] (~sszpila@
08:21:30 Join ender` [0] (
08:29:55 Join LinusN [0] (~linus@rockbox/developer/LinusN)
08:33:07 Join mitk [0] (~mitk@
08:38:34 Join simon [0] (
08:39:17simonhiya. I've got an iPod Nano that won't be patched with rockbox because its filesystem partition is apparently empty.
08:40:38linuxstbsimon: A first generation Nano?
08:41:28simonoh darn, it turns out this iPod is called an iPod shuffle and rockbox doesn't even seem to support it. :-)
08:42:04linuxstbNo, it doesn't. It's completely different hardware to the ipods Rockbox does run on.
08:48:52 Join flydutch [0] (
08:59:17 Join gurenko [0] (
09:10:28 Quit gurenko (Quit: CGI:IRC)
09:12:48 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
09:14:16 Join watto [0] (~watto@
09:15:48 Join Zagor [0] (~bjst@rockbox/developer/Zagor)
09:18:32 Join DerPapst [0] (
09:26:48 Quit markun (Ping timeout: 276 seconds)
09:27:18DerPapstTorne: I doubt my iPod would be a good testing candidate. Afaik there are 2 different display types. The motherboard of my original 3G iPod broke and i replaced it with a mobo from a defunct iPod i bought of ebay. However, the replaced mobo is for the other display type and now the contrast is always way to high. If i boot the original firmware the display is barely readable because it's too dark.
09:27:20DerPapstSo i probably would have to change the default contrast settings anyway before compiling to make sure any bootloader messages are readable. And my lowest contrast value would be way below those 3G iPods with a matching display + controller.
09:35:20DerPapstTorne: but i can test the behaviour of the bootloader though. I assume the code is already in svn?
09:35:45 Join petur [0] (~petur@rockbox/developer/petur)
09:38:48DerPapsthmm... i'll think i even install rb + bootloader in OSOS ;)
09:40:17DerPapstwhee... my build is way outdated... 100130 :D
09:43:08 Join funman [0] (~fun@rockbox/developer/funman)
09:45:22kenguestanyone here that worked on the iriver h300 port?
09:46:58B4gderkenguest: it works better if you ask your question instead...
09:47:03 Join markun [0] (~markun@rockbox/developer/markun)
09:52:39***Saving seen data "./dancer.seen"
09:52:54kenguestB4gder: it was going to be more of a statement than a question ;-) I owe at least one person a drink for making using my h340 bearable!
09:53:34B4gderlots of Rockbox devs hang out here, and a bunch of them/us are involved in the h3xx series too
09:53:49funmanbeer for everyone!
09:55:17DerPapsti've got some nice tree conflicts.. where is uisimulator/sdl/button.c now?
09:55:48DerPapstnvm. found it ;)
09:58:35 Join swilde [0] (
10:03:01 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
10:03:40 Quit polobricolo (Ping timeout: 260 seconds)
10:11:04 Quit n17ikh (Ping timeout: 265 seconds)
10:14:45 Join n17ikh [0] (
10:15:07 Join kugel [0] (~kugel@rockbox/developer/kugel)
10:17:38 Join hebz0rl [0] (
10:21:41 Join efyx [0] (
10:21:45funmanthe as3531 datasheet mentions a vectored interrupt controller with 32 vectored interrupts, so i thought it might be a PL192, but the OF uses a register not present in the pl192
10:22:07funmannot that we'll need more than 16 interrupts anyway
10:27:52kugelfunman: saw my scrollwheel patch?
10:28:05funmanyes but didn't test yet
10:28:11pamauryfunman: is rebooting to the OF on the clip+ supposed to work ?
10:28:32funmanpamaury: yes if you have patched with a recent mkamsboot
10:28:33kugelfrom what I can tell it works really well now, played a bit around (in bubbles) on my trip to the uni
10:29:15funmankugel: there is another patch on FS, did you see it ?
10:31:13kugelno, but now. that one should work better but I think it doesn't necessarily detect overflows
10:31:34kugelv doesn't have a reliable value if the overflow occured
10:32:01funmanif(time < last_wheel_post) should be enough to detect it ?
10:33:03kugelit could be > as well, the top bits are cut off but that doesn't always mean that time smaller than last_wheel_post, no?
10:33:40kugelbut maybe it's good enough
10:33:42funmanit works fine (didn't test 23 minutes ;) )
10:33:53CIA-5New commit by pamaury (r26157): as3525v2: add partial usb init code
10:34:07funmanhm test_disk stops when writing the last bits to storage
10:35:22 Quit TheSeven (Ping timeout: 264 seconds)
10:35:39funmanlast commits in it are related to open/creat
10:36:39 Join n1s [0] (~n1s@rockbox/developer/n1s)
10:37:56pamauryhow could my commit add errors on a completely unrelated target ?
10:38:19funman/bin/sh: /home/robert/rb/builds/build-samsungyh920/apps/core_asmdefs.h: No such file or directory
10:38:27pamauryI didn't modify anything !
10:38:29 Join einhirn [0] (
10:38:47funmanperhaps a missing make dependency
10:39:27pamaurybut that would have break at the last build
10:40:23funmankugel: there should be no need to cast open_wrapper in plugin.c no ?
10:40:40kugelpamaury: a lot of for() loops just for logf?
10:41:02pamaurythat is called debugging :)
10:41:21kugelwill those be optimized away in a non-logf build?
10:41:25kugelI hope so
10:41:27pamauryI think so
10:41:41pamauryAnyway, the driver is non functional and there will be removed at the end
10:43:02pamauryI suspect my init code about fifo is wrong (a night revelation) but as there are no explaination in the linux code, that will be hard to notice at this stage
10:44:37 Join wodz [0] (
10:44:40pamauryI will move more things to compile time constants at the next commit, I don't want to pollute the code by reading to registers to handle every hardware configuration whereras it will not change for a specific target
10:49:09funmantest_disk freezes on close()
10:52:47 Quit Strife89 (Ping timeout: 258 seconds)
10:54:50 Join Strife89 [0] (
10:55:17funman(fuzev1, write & verify, closing after writing succeeded)
11:01:00 Join blairb [0] (
11:02:19funmankugel: shouldn't HAVE_PLUGIN_CHECK_OPEN_CLOSE be defined only for sim ?
11:03:14kugeldunno, gevaerts ?
11:03:20kugelor pamaury ?
11:03:25 Part blairb
11:03:48funmananyway the close wrapper looks ok
11:03:50pamaurywhy ?
11:04:14funmanit freezes even with it undefined
11:04:19funmanah no it worked!
11:04:22gevaertsfunman: why? we could disable it for the release, but what's the advantage of sim-only?
11:04:44funmandunno, just thinking logf isn't enabled on non-sim builds
11:05:44wodzhmm I probably know what the problem with keyreading on HD200 is - lm339 has response time at best 300ns which coresponds to 3.33 MHz sampling rate. This means that I have to set adc divider to 6.75 or more (8 is the closest possible) for 45 MHz cpu clock and 18.61 (32 is the closest possible) for 124 MHz. I was too optimistic when selecting divider
11:05:53gevaertsfunman: it doesn't only logf
11:07:24funmanhm with a reduced test size test_disk works fine
11:07:34 Join blairb [0] (
11:11:51 Part blairb
11:12:22 Join blairb [0] (
11:12:28 Part blairb
11:14:38 Join blairb [0] (
11:15:34 Quit blairb (Client Quit)
11:16:54 Join blairb [0] (
11:18:57 Quit blairb (Client Quit)
11:19:18 Join blairb [0] (
11:19:31funman <- a setting on as3543, what is this threshold, how much the current drawn by the battery is reduced when reaching end of charge ?
11:21:35 Quit phanboy_iv (Ping timeout: 240 seconds)
11:22:09ranmaI suppose so, if your charging current is set to 200mA and the threshold to 10%, then you will get the interrupt when the current has dropped to 20mA mode.
11:22:20*n1s batterybenches beast with rev 26153
11:22:34funmanto, or by ?
11:23:17funmanhm test_disk works fine on fuzev2 :'(
11:23:22 Join dfkt [0] (dfkt@unaffiliated/dfkt)
11:23:51TorneDerPapst: ah, right, i guess that does make your contrast probably irrelevant to everyone else ;)
11:27:29funmanclip+ seems to charge at 3.9V/50mA
11:27:59funmanwith 4.2V charging voltage i don't have end of charge itnerrupt
11:29:24CIA-5New commit by wodz (r26158): HD200 - lm339 response time is at best 300 ns so adcclk can not be too high
11:29:55 Quit linuxstb (Ping timeout: 252 seconds)
11:30:27DerPapstTorne: guess so :P
11:31:51gevaertsI'm not sure actually
11:32:04gevaertsDoesn't the OF just guess the wrong one of the same two choices?
11:32:13funmanah no, with this low voltage i get continuous end of charge
11:36:13 Join Evilnick_ [0] (
11:37:26 Quit blairb (Quit: Ex-Chat)
11:38:47 Quit evilnick (Ping timeout: 240 seconds)
11:39:27wodz3.9V/50mA this is rather unusual - You mean 50mA at CC phase and 3.9V at CV phase?
11:39:38funman4.15V/100mA seems to work
11:39:43funmani don't know what i mean
11:40:03wodzfunman: lol
11:40:13 Join blairb [0] (
11:41:49 Quit jfc (Read error: Connection reset by peer)
11:42:10funmanjust trying different settings for the clips to both charge, and detect end of charge correctly
11:42:24funmanafaiu, the current drawn can be arbitrary
11:42:32funmanjust not too high to not melt the wires
11:42:43markunfunman: didn't you have some problems with charging in the OF? I have them as well now.
11:42:58funmanmarkun: yes on clipv1, but i left it plugged and it eventually took off
11:43:08funmancurrently discharging it completely to try again
11:43:11 Nick fxb__ is now known as fxb (
11:43:18 Join jfc [0] (
11:43:37funmanwodz: the charger voltage should not be too much higher than the battery voltage ?
11:43:52wodzfunman: generally no
11:44:05wodzI men for li-ion and lipo batteries
11:44:40markunfunman: mine issue is probably unrelated then. Plugging in USB while holding 'select' crashed rockbox, so I rebooted into the OF and left it connected at night. The next morning I assumed it would be full, but still showed it was charging and rockbox told me it was at 30%..
11:45:17wodzfunman: typical limit is about 4.3V and because of this most charger circuits sets upper limit to 4.2 to be on the safe side
11:46:39funmanif i set it to 4.15V and not 4.2V then end of charge is detected immediately on clip+ (battery 100%)
11:47:34wodzfunman: when speaking about current drawn You are not quite right - You cannot set it too high because battery will heat up and eventually protection circuit (integrated with every li-ion/lipo) will cut off
11:47:58funmanah thanks
11:48:50wodzand if You set current way too high - smoke and fire may come into action :-)
11:49:18funmani can understand current but voltage is still mysterious to me
11:49:59 Join mikroflops_ [0] (
11:50:17wodzstandard setting for charging li batteries is C/10 which means take capacity in mAh/10 = charge current in mA
11:50:51funmanwe use higher values on sansas (don't know where these charging current come from btw)
11:51:19wodzbut from the other hand good lipo can withstand 3-5C charging
11:52:25funmanif i set charger voltage to 4.15V, is it normal if the battery has a higher voltage?
11:52:40***Saving seen data "./dancer.seen"
11:52:46wodzcan You elaborate?
11:53:21 Quit blairb (Remote host closed the connection)
11:53:23funmani set 'maximum charger voltage' to 4.15V in as3543 registers, and reading battery voltage when charging shows 4.16
11:54:05wodzthat is ok - it comes probably from scalling adc readout
11:54:11 Quit mikroflops (Ping timeout: 276 seconds)
11:54:49funmanhm indeed now the battery voltage is flat, but still charging
11:55:10wodzfunman, thats ok also - You reached CV phase
11:56:18 Join blairb [0] (
12:00:03 Quit mikroflops_ (Ping timeout: 240 seconds)
12:00:29kugelcan I lock a semaphore in thread A and and release it in thread B (with the intention that thread A waits if it tries to lock the semaphore again before B released it)?
12:03:26wodzwhat is default procedure to calibrate percent_to_volt_charge[] ?
12:03:53 Join blairb_ [0] (
12:04:00 Quit blairb_ (Remote host closed the connection)
12:04:14wodzbtw it only makes sense in CC phase of charging
12:04:18 Join Forsaken_Boy [0] (~chatzilla@
12:04:32 Quit blairb (Quit: Ex-Chat)
12:04:32kugelwodz: battery bench
12:04:53wodzbattery bench during charging and no playback?
12:05:03kugelyea, why not?
12:05:10wodzjust asking
12:05:11funmanwodz: that's what i did
12:05:39funmani don't get what's so special with this close(bigfile) on fuze, writing the 300MB prior to closing succeeded
12:06:44kugelmaybe it's related to the lot freezes I see on it recently
12:06:51 Quit bieber (Ping timeout: 260 seconds)
12:06:57funmanindeed but i have a potential fix which i'm testing
12:06:57kugelwhere recently == since 2 month or so
12:07:06 Join bieber [0] (
12:07:36funmangive a lower priority to audio interrupts -> move down the list of vectored interrupts
12:07:46funmanpictureflow/database work fine but i wanted to confirm with test_disk
12:10:28 Join tchan1 [0] (
12:11:51funmanthis guy on the forum found the problems appeared when audio interrupts were added
12:12:08 Quit tchan (Ping timeout: 245 seconds)
12:12:52 Quit powell14ski__ (Ping timeout: 265 seconds)
12:13:19 Quit BlakeJohnson86 (Read error: Operation timed out)
12:15:15wodzis there reliable way to determine mean frequency (% boosted) during playback?
12:15:42funmannope, test_codec is the only precise bench
12:15:51 Join powell14ski__ [0] (
12:16:47 Join BlakeJohnson86 [0] (~bjohnson@2002:1876:a27b:0:227:13ff:fe65:1262)
12:17:02wodzI was thinking about setting second timer with fixed div and read this timer value in system tick interrupt
12:17:52wodzdoes it make sense?
12:18:25funmantick already read boosting and show it in buffering thread
12:18:32 Quit GeekShadow (Ping timeout: 260 seconds)
12:18:45wodzbut I heard several times it is unreliable
12:19:27funmani say that because it just see if it's boosted each tick, if we are boosting between 2 ticks it's not shown
12:20:25wodzso it counts boost when it is in boosted state entering system tick interrupt?
12:24:02 Join mt_ [0] (~mtee@
12:24:09funmanbut if boosting times are typically longer than 1 tick it could work
12:24:51wodzI think I try with timer approach and compare results
12:25:48ssorgatem_I've updated catalan translation and i've posted the diff at FS #11293
12:26:18 Quit mt (Ping timeout: 258 seconds)
12:26:26ssorgatem_I've had to manuallt edit catalan.lang, as wouldn't save new translated strings
12:26:27 Nick mt_ is now known as mt (~mtee@
12:26:33n1swodz: i think the boostin percent works pretty well in the buffering screen, but it's too coarse to measure codec speeds
12:27:31 Join blairb [0] (
12:28:09ssorgatem_Is it the right procedure?
12:28:14 Quit blairb (Client Quit)
12:28:55wodzn1s: OK so if test_codec returns comparable resutls to other CFs, test_mem also and buffering screen shows way higher boost % how to narrow it down?
12:28:59wodzssorgatem_, yes
12:29:53n1swodz: that is weird indeed...
12:30:02funmanssorgatem_: also tell bluebroth3r if you have problems with the translate website
12:30:16ssorgatem_funman: ok
12:30:38ssorgatem_bluebroth3r: ping?
12:30:57wodzn1s: lcd performance is superb with about 1000fps fullscreen updates
12:31:18 Join blairb [0] (
12:31:40 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
12:32:20funmanwodz: did you measure time needed to run set_cpu_frequency ?
12:32:22n1swodz: if test_codec is as fast i'd guess something that isn't running while running test_codec but is running during normal playback sucks the rest of the cpu...
12:33:39wodzfunman: 1) I don't see the way to measure this 2) I can do nothing about this time it is hardware constrain and should be common to all CFs
12:34:27funman1) boost/unboost 1000 times, measure number of ticks
12:34:31n1sare you using the dma for audio?
12:34:44wodzn1s: yes as all other CSs
12:35:36wodzfunman: I don't think this will be reliable
12:36:14 Quit yawny (Ping timeout: 246 seconds)
12:37:21 Join max242 [0] (
12:38:26max242is there an easy way to lower the priority of the audio interrupts?
12:38:43max242i'm interested and willing to test things out tonight
12:39:10max242i wonder if it will solve the pictureflow issues i see since r25299
12:41:15funmanmax242: ^
12:41:22max242i saw ranma is the author of the changes in r25299
12:41:40max242can he give us a clue what's wrong and what we can change in the code
12:41:59max242yes, i'm here for a short while during my lunch break
12:42:40funmani don't think r25299 is wrong in any way
12:43:18max242ok, no problem. Is there any place described how to work with patches? I never did that myself up to now
12:43:59max242this evening, on my personal laptop where i have the rockbox dev. environment installed, i'll give it a try
12:44:29 Join elcan [0] (
12:44:29max242thanks! I have to go back to work now, cheers ;-)
12:45:41 Quit max242 (Quit: CGI:IRC (EOF))
12:56:02kugeljhMikeS: ping
12:57:52kugelgevaerts: I got an answer on the mini2440 ml re x-server start problem
12:58:40wodzWhat is default policy in rockbox about datasheets? Should we keep local copy or relay on some external internet source?
12:59:50kugelit usually depends on whether it publicly available or not
13:00:14 Join lpereira [0] (
13:00:16kugelread: if you signed an NDA or something like that to get it
13:00:31 Quit GeekShadow (Ping timeout: 260 seconds)
13:02:19wodzI have couple of public pdfs
13:05:40 Quit blairb (Quit: Leaving)
13:08:08funmanranma: does test_disk work on c200v2 ?
13:08:14funman(write & verify)
13:13:28 Join evilnick|ipad [0] (
13:14:15 Quit n1s (Ping timeout: 248 seconds)
13:14:57 Quit kugel (Ping timeout: 264 seconds)
13:19:04funmanhm it locks on reading a sector, not on writing
13:19:08 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
13:20:09funmanwith a 30MB test file
13:20:23 Quit Farthen (Remote host closed the connection)
13:21:05 Join Farthen [0] (
13:22:24funman2nd try on a 30MB file: closing succeeds -> test finish and pass
13:23:01pamauryfunman: what is your problem ? close hangs on a very big file ?
13:23:04 Quit Farthen (Remote host closed the connection)
13:23:29 Join Farthen [0] (
13:25:00pamaurydo you have a clue ? Is it writing/reading something ?
13:25:11funmanreading 1 sector i think
13:25:24funmannot sure if the freeze is in fat/file code or in as3525 sd driver
13:26:14 Join wincent_balin [0] (
13:26:30pamaurydoes it happen on other devices ?
13:26:38funmannot on fuzev
13:27:07pamaurythen perhaps it's the sd driver but why would it happen with a big file only ?
13:28:17 Part LinusN
13:29:31 Nick Evilnick_ is now known as evilnick (
13:29:40 Quit evilnick (Changing host)
13:29:40 Join evilnick [0] (~Evilnick@rockbox/staff/evilnick)
13:29:40 Quit Farthen (Remote host closed the connection)
13:31:36funmanit's writing in sectors 0xCxxxx, and after write finished the last SD op i see is a read of 1 sector at 0x5xxxx
13:32:14pamauryit hangs completely the device ? Can't you debug it to see which part hang ?
13:32:28funmanwell no it still works
13:32:54funmani don't know FAT at all so i don't want to look here
13:33:30pamauryI know fat so I can have a look if you are sure there is a problem witht he fat code :) Perhaps the sd read fails and the fat keeps trying or locks
13:34:00funmanyes that's what i think
13:34:04funmanor rather the read is incorrect
13:36:00pamauryincorrect in which sense ?
13:36:09pamauryperhaps the data is corrupted
13:36:20funmanwell now it works
13:36:25 Quit bieber (Ping timeout: 246 seconds)
13:36:38 Join bieber [0] (
13:37:03funmanthe controller should check data crc
13:39:55ranmaSo far I got Create 23 files/s Open 50 files/s Dirscan 6225 files/s Delete 52 files/s Create(512, A): 296 KB/s
13:40:06ranmaWrite seems to hang...
13:40:13funmanranma: not the speed, the write & verify
13:41:22 Join kugel [0] (~kugel@rockbox/developer/kugel)
13:41:36 Quit kugel (Remote host closed the connection)
13:42:31 Quit GeekShadow (Ping timeout: 248 seconds)
13:43:13funmani'm building with applied
13:44:00 Join kugel [0] (~kugel@rockbox/developer/kugel)
13:44:34funmangoddamn it works
13:45:50wodzCharge monitoring is *SO* overcomplicated on MPIO. I think I fuck this at all and relay only on AC plug detection
13:46:04 Join LinusN [0] (~linus@rockbox/developer/LinusN)
13:46:09*funman is ok with the fucking part
13:48:45wodzMPIO uses two pins OUT and IN to detect charging state. First it sets OUT high and reads IN. If IN=low device is charging in CC regime, else device reached C/10 condition. Than it sets OUT to HiZ and reads IN. If IN=low device is charging in CV regime, else timeout has occured
13:49:48 Join Jaykay [0] (
13:50:28ranmafunman: Test passed.
13:50:46funmanranma: thanks
13:50:53funmancan you test database rebuild & picture flow cache?
13:51:38 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
13:51:52funmanthese locked with svn (i had ~2GB of mp3 split in albums, all with tags)
13:52:10*petur looks slightly past LinusN at his h120
13:52:41***Saving seen data "./dancer.seen"
13:53:43 Join esperegu [0] (~quassel@
13:53:53LinusNpetur: the parallel port card is in the mail...
13:56:26funmansame test without my lcd debug code at entry & exit of sd_transfer_sectors() -> lock (file system formatted prior to installing rockbox)
13:59:19 Quit B4gder (Quit: It is time to say moo)
14:00:21 Join n1s [0] (~n1s@rockbox/developer/n1s)
14:02:02mc2739Zagor: Could you update the web server for a main page change? Thanks
14:03:26Zagormc2739: done
14:04:13 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
14:09:58 Join adnyxo [0] (
14:10:16ranmafunman: I ran "update now" from the menu, but since that runs in the background, how do I know if it's working?
14:10:22*ranma hasn't really used the Database...
14:10:57funmanif you delete .rockbox/database_* it will be renew totally
14:11:10funman(and block until it's done)
14:11:23pamauryranma: you have database status in debug menu
14:11:45S_a_i_n_tranma: I always wondered that myself, I'm not sure there is a cue for when "update now" completes :/
14:11:52CIA-5New commit by funman (r26159): sd-as3525: give timeout in HZ units, not in ticks ...
14:11:57CIA-5New commit by funman (r26160): as3525: reorder vectored interrupts ...
14:12:02CIA-5New commit by funman (r26161): sd-as3525: avoid division when calculating current bank, we only deal with 1 or 2 banks
14:14:44 Quit TheSeven (Ping timeout: 276 seconds)
14:16:09funmanstill locks on fuze :o
14:19:07ranmaOk, it says 'Initialized: Yes' and 'Progress: -1% (398 entries)', so I guess that worked too.
14:19:18kugelranma, S_a_i_n_t: I always look at the disk indicator of the statusbar
14:20:38 Join katsuru_ [0] (
14:24:22 Quit katsuru (Ping timeout: 258 seconds)
14:28:10 Join Schmogel [0] (
14:31:35 Join merbzt [0] (~benlar@
14:42:24jhMikeSkugel: sir?
14:42:44kugeljhMikeS: saw my question about semaphores earlier today?
14:43:05jhMikeSkugel: nope, haven't been paying attention here
14:43:29kugeljhMikeS: can I lock a semaphore in thread A and and release it in thread B (with the intention that thread A waits if it tries to lock the semaphore again before B released it)?
14:43:50 Join Farthen [0] (
14:43:56jhMikeSthat's fine. sems aren't owned like mutexes. really, they're just counted.
14:44:21kugelok, there goes my suspicion why my code doesn't work :\
14:44:25jhMikeSif it didn't work, dual core mp3 and spc wouldn't work very well
14:45:14jhMikeSare you talking about rockbox object? can't necessarily speak for other implementation, though they wouldn't be semantically correct i suppose
14:45:25kugelyea I do
14:46:30kugel is my code, I would like to make lcd updates asynchronous (because they're slow) on as3525(v2)
14:47:30kugelhere it is:
14:48:33gevaertskugel: asynchronous lcd might be useful on RaaA too I think. I still suspect SDL screen updates for the stuttering on my n900
14:50:37kugelgevaerts: I think it could easily done for all lcd drivers too, but most are fast enough so that it doesn't matter
14:51:01jhMikeSkugel: wouldn't that cause artifacts if something starts drawing while the update is happening?
14:51:23kugeljhMikeS: that's what the semaphore is supposed to prevent
14:51:49jhMikeSso then you slow it down elsewhere while it waits anyway? :)
14:52:29*jhMikeS 'll look at the pastie before saying more
14:53:41gevaertsThat's when you add double buffering I guess
14:54:07kugeljhMikeS: no, the main benefit would be letting the other threads run while the lcd thread waits for the fifo to empty before it feeds new data
14:54:37kugelI think it wouldn't slow down anything, the worst that can happen is nothing is improved
14:55:00jhMikeSkugel: yea, that's why I'm looking. I'm not too familiar with that hardware.
14:57:02jhMikeSthe patch looks like it's off from the source tree I have. was there a commit that changed dbop?
14:57:17kugelit's against current svn
14:58:46*jhMikeS is a few revisions behind
14:59:51kugelfirmware/target/arm/as3525/dbop-as3525.c exists for a long time now :)
15:00:28jhMikeSyeah, bit I see no queues, sems, mutexes or any such thing while they're clearly already there in the patch
15:00:48kugelthe patch adds them
15:02:42 Quit antil33t (Read error: Connection reset by peer)
15:02:48 Join antil33t [0] (
15:03:59funmanhm i think i detect end of charge OK on clip+ now
15:04:06*jhMikeS notices not all have a "+" next to them.
15:04:40kugeljhMikeS: are you looking at the first or second paste?
15:04:57kugelignore the first one
15:05:23jhMikeSah...hahaha...I got on the first one, not the second. :P
15:10:43funmanon as3543 no need to clear the interrupt on plugging, perhaps they have a better detection
15:11:39 Join liar [0] (
15:11:50 Quit liar (Remote host closed the connection)
15:12:28jhMikeSnow I'm getting failed hunks :\
15:12:41 Join panni_ [0] (
15:13:18kugeljhMikeS: you need -p1
15:13:36 Part Zagor
15:13:46kugeloh uh, the SOURCES and system-as3525.c might fail
15:14:40jhMikeSremove it?
15:14:50jhMikeSit is system
15:15:14jhMikeShmmm...looks important
15:15:20kugelI think you can ignore the hunk if you don't want to run the code
15:15:44jhMikeSthat should be fine
15:23:23***Started Dancer V4.16
15:23:23***Connected to on port 6667
15:23:23***Logfile for #rockbox started
15:23:24Mode"logbot :+i" by logbot
15:23:25CtcpVersion from frigg!~frigg@freenode/utility-bot/frigg
15:23:25***Server message 501: 'logbot :Unknown MODE flag'
15:23:25 Join logbot [0] (
15:23:25 Join komputes [0] (~komputes@ubuntu/member/komputes)
15:23:25 Join panni_ [0] (
15:23:25 Join antil33t [0] (
15:23:25 Join Farthen [0] (
15:23:25 Join merbzt [0] (~benlar@
15:23:25 Join Schmogel [0] (
15:23:25 Join katsuru_ [0] (
15:23:25 Join adnyxo [0] (
15:23:25 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
15:23:25 Join n1s [0] (~n1s@rockbox/developer/n1s)
15:23:25 Join esperegu [0] (~quassel@
15:23:25 Join Jaykay [0] (
15:23:25 Join kugel [0] (~kugel@rockbox/developer/kugel)
15:23:25 Join bieber [0] (
15:23:25 Join wincent_balin [0] (
15:23:25 Join evilnick|ipad [0] (
15:23:25 Join lpereira [0] (
15:23:25 Join elcan [0] (
15:23:25 Join mt [0] (~mtee@
15:23:25 Join BlakeJohnson86 [0] (~bjohnson@2002:1876:a27b:0:227:13ff:fe65:1262)
15:23:25 Join powell14ski__ [0] (
15:23:25 Join tchan1 [0] (
15:23:25 Join Forsaken_Boy [0] (~chatzilla@
15:23:25 Join jfc [0] (
15:23:25 Join evilnick [0] (~Evilnick@rockbox/staff/evilnick)
15:23:25 Join dfkt [0] (dfkt@unaffiliated/dfkt)
15:23:25 Join Strife89 [0] (
15:23:25 Join wodz [0] (
15:23:25 Join einhirn [0] (
15:23:25 Join efyx [0] (
15:23:25 Join hebz0rl [0] (
15:23:25 Join n17ikh [0] (
15:23:25 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
15:23:25 Join swilde [0] (
15:23:25 Join markun [0] (~markun@rockbox/developer/markun)
15:23:25 Join funman [0] (~fun@rockbox/developer/funman)
15:23:25 Join petur [0] (~petur@rockbox/developer/petur)
15:23:25 Join DerPapst [0] (
15:23:25 Join watto [0] (~watto@
15:23:25 Join flydutch [0] (
15:23:25 Join simon [0] (
15:23:25 Join mitk [0] (~mitk@
15:23:25 Join ender` [0] (
15:23:25 Join Szpila [0] (~sszpila@
15:23:25 Join S_a_i_n_t [0] (S_a_i_n_t@
15:23:25 Join saratoga [0] (~9803c6dd@gateway/web/freenode/x-zzmofwwvlikygwyn)
15:23:25 Join Unhelpful [0] (~quassel@rockbox/developer/Unhelpful)
15:23:25 Join cdb [0] (
15:23:25 Join CGL [0] (~CGL@
15:23:25 Join amiconn [0] (quassel@rockbox/developer/amiconn)
15:23:25 Join pixelma [0] (quassel@rockbox/staff/pixelma)
15:23:25 Join ischeria1 [0] (
15:23:25 Join MagusG [0] (
15:23:25 Join FlynDice [0] (
15:23:25 Join Rob2222 [0] (
15:23:25 Join solexx [0] (
15:23:25 Join Battousai [0] (~bryan@gentoo/developer/battousai)
15:23:25 Join nimak [0] (
15:23:25 Join AlexP [0] (~ap@rockbox/staff/AlexP)
15:23:25 Join Kitar|st [0] (Kitar_st@
15:23:25 Join bluebroth3r [0] (~dom@rockbox/developer/bluebrother)
15:23:25 Join shai [0] (
15:23:25 Join ved [0] (
15:23:25 Join togetic [0] (~togetic@unaffiliated/ibuffy)
15:23:25 Join YPSY [0] (
15:23:25 Join Kohlrabi [0] (
15:23:25 Join yosafbridge [0] (
15:23:25 Join Tuplis [0] (
15:23:25 Join kenguest [0] (~radagast@
15:23:25 Join rvvs89 [0] (
15:23:25 Join krazykit [0] (~kkit@
15:23:25 Join linuxguy3 [0] (
15:23:25 Join Xerion [0] (
15:23:25 Join Guest55923 [0] (
15:23:25 Join MuscleNerd [0] (
15:23:25 Join scorche [0] (~scorche@rockbox/administrator/scorche)
15:23:25 Join Zambezi [0] (Zulu@unaffiliated/zambezi)
15:23:25 Join crwl [0] (
15:23:25 Join advcomp2019__ [0] (~advcomp20@unaffiliated/advcomp2019)
15:23:25 Join gevaerts [0] (~fg@rockbox/developer/gevaerts)
15:23:25 Join sinthetek [0] (~sinthetek@unaffiliated/sinthetek)
15:23:25 Join Barahir_ [0] (
15:23:25 Join ssorgatem_ [0] (
15:23:25 Join jordan` [0] (
15:23:25 Join Llorean [0] (~DarkkOne@rockbox/user/Llorean)
15:23:25 Join Curtman [0] (
15:23:25 Join slck [0] (Venci@Slackware.SlackPix.Com)
15:23:25 Join topik [0] (
15:23:25 Join @ChanServ [0] (ChanServ@services.)
15:23:25 Join ThomasAH [0] (
15:23:25 Join fxb [0] (
15:23:25 Join dionoea [0] (~dionoea@videolan/developer/dionoea)
15:23:25 Join Hadaka [0] (
15:23:25 Join Kamyk_ [0] (
15:23:25 Join Utchybann [0] (
15:23:25 Join daurnimator [0] (daurnimato@freenode/staff/daurnimator)
15:23:25 Join stavrob [0] (
15:23:25 Join sharp [0] (
15:23:25 Join blithe [0] (~blithe@
15:23:25 Join CIA-5 [0] (cia@
15:23:25 Join maraz [0] (
15:23:25 Join aevin [0] (eivindsy@unaffiliated/aevin)
15:23:25 Join tmzt [0] (
15:23:25 Join simabeis [0] (
15:23:25 Join avacore [0] (
15:23:25 Join kadoban [0] (
15:23:25 Join bug2000 [0] (~bug@unaffiliated/bug2000)
15:23:25 Join FOAD [0] (~dok@
15:23:25 Join Guest951 [0] (
15:23:25 Join pjm0616 [0] (~user@
15:23:25 Join scorche|sh [0] (~scorche@rockbox/administrator/scorche)
15:23:25 Join cYmen [0] (
15:23:25 Join wombat23 [0] (
15:23:25 Join bzed [0] (
15:23:25 Join jae [0] (
15:23:25 Join grndslm [0] (
15:23:25 Join jhMikeS [0] (~jethead71@rockbox/developer/jhMikeS)
15:23:25 Join ps-auxw [0] (~arneb@2001:470:c807:0:1532:4e5f:2ad3:4123)
15:23:25 Join Torne [0] (torne@rockbox/developer/Torne)
15:23:25 Join soap [0] (~soap@rockbox/staff/soap)
15:23:25 Join Topy44 [0] (
15:23:25 Join whydoubt [0] (
15:23:25 Join lostlogic [0] (~lostlogic@rockbox/developer/lostlogic)
15:23:25 Join parafin [0] (
15:23:25 Join xavieran [0] (
15:23:25 Join rhodan [0] (~Quassel@2001:1608:12:2::38)
15:23:25 Join sevard [0] (sev@
15:23:25 Join mc2739 [0] (~mc2739@rockbox/developer/mc2739)
15:23:25 Join kisak [0] (
15:23:25 Join rasher [0] (~rasher@rockbox/developer/rasher)
15:23:25 Join GodEater [0] (~bibble@rockbox/staff/GodEater)
15:23:25 Join Bagder [0] (~daniel@rockbox/developer/bagder)
15:23:25 Join Zarggg [0] (
15:23:26 Join Beta2K [0] (
15:23:26 Join bluefoxx [0] (
15:23:26 Join Galois [0] (
15:23:26 Join ranma [0] (
15:23:26 Join detaos [0] (
15:23:26 Join BeFalou [0] (~mamutoi@unaffiliated/befalou)
15:23:26 Join shaggy-h [0] (
15:23:26 Join preglow [0] (
15:23:30wodzConnection refused: Can't connect to host ''
15:23:41jhMikeSkugel: wouldn't one thread lock the sem, leave it locked and then the thread would lock the sem again?
15:23:53 Join teru [0] (
15:24:02funmantest_disk works fine if i disable sd_enable()
15:24:17kugeljhMikeS: it should wait when it tries to lock it a second time
15:24:20jhMikeSnvm, thought I saw another lock in there :(
15:24:20CIA-5New commit by wodz (r26162): Add charging/discharging indication to battery debug menu if CONFIG_CHARGING >= CHARGING_MONITOR
15:24:22funmanor add a delay before disabling it at exit of transfer
15:22:46wodzaa ok, temporary problem
15:23:17jhMikeSkugel: oh, no I did spot something :) dbop_write_data. what is supposed to happen?
15:24:20kugelin svn it does the transfer, after the patch it should poke the lcd thread to start the transfer
15:25:47 Join Locke_Fireclaw [0] (
15:26:00 Part Locke_Fireclaw
15:26:10jhMikeSdbop_write_data_ex (called by thread) calls dbop_write_data which calls semaphore_wait. what if another thread locked it already? dbop_write_data gets called externally?
15:26:30kugelah yes, I fixed that one already without success
15:27:52jhMikeSso, you get an int for the TXFIFO empty?
15:28:05kugelalmost empty, yes
15:28:28jhMikeSI'd fill the data in the INT handler, not on-thread to be honest.
15:29:47jhMikeSwhen starting, prefill and then keep filling from the ISR
15:29:48kugelthat could work but it's not the cause of the problem is it?
15:30:02jhMikeSnot sure, and it would sure be easier to see
15:30:20jhMikeSyou said you fixed the deadlock possibility?
15:30:43kugelwhen I boot rockbox I get a white screen (and apparently a panic because a button press reboots immediately)
15:31:23kugelwodz: svn is slow here too
15:31:58wodzwhat is the frequency of battery graph update in debug->battery?
15:32:15 Join jgarvey [0] (
15:32:16jhMikeSkugel: perhaps the correct int enables aren't being observed?
15:32:39jhMikeSit looks like it could be 100% interrupt based to keep the fifos going, no thread, no anything else really.
15:33:05wodzfrom 3rd screen I judge it should be 1 per minute but in reality it is more like 1/few minutes
15:33:11 Join LinusN [0] (~linus@rockbox/developer/LinusN)
15:35:57kugeljhMikeS: I still need the semaphore to prevent artifacts. I guess I can also release it from the isr?
15:36:47jhMikeSkugel: why?
15:36:48kugelhm, it might be tricky if thread_switch() is called from an isr
15:36:57jhMikeSkugel: you can't do taht
15:37:13jhMikeSthat would effectively cause preemption
15:37:39kugelI would like the thread which wants the update to continue before the update finishes (hence asynchronous)
15:38:33jhMikeSthe ISR would be the worker, not a thread
15:39:23kugelbut I cannot block out the calling thread from re-entering this way, can I?
15:39:40kugelwell, busy polling some variable would work but that's not really nice
15:41:16 Quit wodz (Quit: Leaving)
15:43:53jhMikeSkugel: yes, use the wakeup. wake it from the ISR when finished. wait on it at the beginnine. first thread in does the wait and it will block other threads. next wakeup lets the next thread go in.
15:44:37 Join evilnick_B [0] (~0c140464@rockbox/staff/evilnick)
15:45:00kugelbut what if there's no update running? there'd be no isr to signal it
15:46:33funmanhmm what is the purpose of storage_enable() in firmware/usb.c::usb_slave_mode() ?
15:47:22 Quit panni_ (Ping timeout: 246 seconds)
15:49:25pamauryfunman: I think it's because in slave mode the storage is hardware controlled (ie us-ata bridge)
15:49:40jhMikeSkugel: I don't understand? when the ISR exhausts the data, it signals it. it becomes a "not busy" signal. if you run out of data in a prefill, signal it from the thread instead.
15:49:48pamauryso usb.c disable software control and then enable it on unplug
15:49:58pamaurybut that's juste a guess
15:50:08funmanthat makes sense
15:50:17funmanit's the only place where it's used
15:50:31funmanon as3525 we have it but it's internal to this file
15:50:35 Quit mitk (Quit: Leaving)
15:50:36kugeljhMikeS: you said in wait for the signal in the beginning. if there's no update currently and no isr running then it would cause a deadlock wouldn't it?
15:51:01jhMikeSkugel: no, because it would remain signaled after the last transfer
15:51:31kugelah so you can signal first, then wait and the wait would be a no-op?
15:51:31pamauryfunman: that's true, all other places use the more specialized {ata,mmc,}_enable()
15:51:39jhMikeSkugel: set it signaled right after creation too
15:51:56pamaurybut how can it be internal ? When there are multi drive enable, it is define in storage.c
15:52:46jhMikeSkugel: you wait upon entering the update function, if TBOP is busy, it blocks out the thread until TBOP is done with whatever it was doing
15:53:14jhMikeSif no thread was there, it just stay signaled until something comes along and starts a new update
15:53:16kugelok, I didn't know you can signal before the wait
15:54:11jhMikeSyes, it's basically a single-count semaphore. if I just added some irq disabling in the sem code, sems could be wakeups, even multicount ones.
15:59:37funmantoo much differences between as3514/3515 and as3543 :/
16:00:05funmansome registers were added, some removed, some moved, some modified
16:00:24funmansometimes the bits are the same but not at the same place (they are shifted)
16:01:14funmanthe adc channel we read for battery is charger out, but on as3514 this one is charger in
16:01:43 Quit rvvs89 (Changing host)
16:01:43 Join rvvs89 [0] (ivo@pdpc/supporter/base/rvvs89)
16:03:40funmansomeone with a 100% charged fuzev2 want to test a patch? mine just takes ages to charge (the OF still shows the battery icon moving)
16:04:38*jhMikeS was told input fifo int weren't available and that's why the scrollwheel needs polling but the DS says otherwise
16:05:24*kugel doesn't see the correlation
16:05:41*jhMikeS was just commenting on something from long ago
16:05:55funmannevermind it just finished
16:06:10CIA-5New commit by funman (r26163): as3525v2: fix end of charge detection
16:06:15CIA-5New commit by funman (r26164): sd-as3525: wait 100µs before disabling SD clocks ...
16:06:21kugelyou cannot interrupt on bits on DBOP_DIN, which is unrelated from DBOP_DOUT
16:07:03jhMikeSkugel: ah, I guess just on pop error then
16:09:06jhMikeSanyway, looks like you should mask 'almost empty' when out of data and enable 'empty' which does signal, then remask it
16:11:46jhMikeSit looks like "almost empty" is bit 9, not bit 7 (which is 'almost full') ??
16:12:23kugelbut well, it still worked
16:12:33funmanUnhelpful: any idea with the asmdefs.h failure on roolku builds ?
16:12:45kugelI enable that when it's completely full so almost full works too
16:13:07 Quit pamaury (Quit: Quitte)
16:13:16jhMikeSkugel: Two datasheets show different bits
16:13:23kugelas I said the wakeup/WFI part works
16:13:55jhMikeS1.1 shows one thing, 1.12 shows another
16:14:30kugeljhMikeS: eh. the irq enable bit is bit 7, the status bit is bit 9
16:15:13jhMikeSoh, crap, I think had 3524 open, not 3524...what's going on? I read it like five time. :\
16:15:15kugel(DS version 1.13)
16:17:22*jhMikeS wonder just what the hell he was looking at
16:17:57*funman dunno about 3524
16:17:57kugelfunman: the preprocessor condition for INT_GPIOA in the ivt looks wrong to me
16:18:08kugeljhMikeS: you need 3525
16:18:19jhMikeSfrack, was looking at status reg by mistake (doing too much here)
16:18:24funmankugel: why?
16:18:47kugellook at the conditions in INT_GPIOA()
16:19:19funmanthe 1st one should be HOTSWAP i think
16:19:30kugelalso HOTSWAP makes more sense than MULTIDRIVE
16:19:42 Part Szpila
16:19:46funmanMULTIDRIVE also works with ramdisks
16:20:22jhMikeSkugel: this looks like a really easy thing that needs only a wakeup and the proper ISR. basically it's like handling audio or the serial ports.
16:20:31 Quit GeekShadow (Read error: Connection reset by peer)
16:20:39kugelyes, you convinced me long ago :P
16:21:01funmankugel: ?
16:21:02 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
16:21:12funmanhmm no
16:21:26funmani guess it should just be ifdef SANSA_FUZEV2, not SCROLLWHEEL + AS3525v2
16:22:14kugelmaybe even s/CONFIG_CPU == AS3525v2/defined(INCREASED_SCROLLWHEEL_POLLING)/
16:22:22kugelerr, !defined()
16:22:31jhMikeSkugel: only problem is other processor modes that might touch the display and need to write something (I suppose a mode check and simply busy write in that case)
16:22:41funmankugel: well it's not related
16:22:48kugelit is
16:23:14funmanhow we read the wheel on fuzev2 isn't configurable
16:23:16kugelINCREASED_SCROLLWHEEL_POLLING and scrollwheel interrupt are the exact opposite (and mutually exclusive)
16:24:45kugelit's actually "#define INCREASED_SCROLLWHEEL_POLLING (defined(HAVE_SCROLLWHEEL) && (CONFIG_CPU == AS3525))"
16:25:26 Quit swilde (Quit: ERC Version 5.3 (IRC client for Emacs))
16:25:34funmanbutton_gpioa_isr() won't work on fuzev1/e200v2, and INCREASED_SCROLLWHEEL_POLLING won't work on fuzev2
16:26:13 Nick CGL is now known as [CGL] (~CGL@
16:26:22kugelthat's not right
16:26:34kugelwe did INCREASED_SCROLLWHEEL_POLLING earlier on the fuzev2
16:26:49funmanok but now it won't work
16:27:10kugelyou cannot activate it by simply defining it, but that's the reason it's not in config.h, it still describes the wheel mechanismk
16:27:21 Part LinusN
16:27:48jhMikeSwhat is read strobe?
16:29:14kugeljhMikeS: bertrik may be able to explain
16:29:22kugelhe figured out the dbop din reading
16:30:26jhMikeSkugel: can the pop fifo be read by DMA?
16:30:43CIA-5New commit by funman (r26165): better preprocessor conditions for enabling GPIOA interrupt ...
16:31:13funmankugel: i did (defined(SANSA_FUZEV2) && !defined(INCREASED_SCROLLWHEEL_POLLING))
16:31:19kugelwhy that?
16:31:22 Join panni_ [0] (
16:31:34kugelI don't think defined(SANSA_FUZEV2) is needed
16:32:17funmanit is needed: if you build for fuzev1 you don't need the gpioa interrupt regardless how scrollwheel code works
16:33:14kugelit should be "#if defined(HAVE_HOTSWAP) || INCREASED_SCROLLWHEEL_POLLING
16:33:29kugelINCREASED_SCROLLWHEEL_POLLING is always defined, but only true for fuzev2
16:34:00funmanaah it's not a HAVE_*
16:34:19kugeland defined(SANSA_FUZEV2) is still unneeded
16:34:38funmannope i don't agree:
16:34:50*jhMikeS wonders about using increased polling rates *only* if the wheel is turning, otherwise, cut back and the user won't notice
16:35:00kugelfunman: why?
16:35:07funmanif you build for fuzev1 you don't need the gpioa interrupt regardless how scrollwheel code works
16:35:26kugelyou need it for hotswap
16:35:32kugeland INCREASED_SCROLLWHEEL_POLLING is 0 for fuzev1
16:35:36funmanthat's why we have the HOTSWAP condition
16:35:46funmanbutton_gpioa_isr() only exist on fuzev2
16:35:53kugelyou don't get it if you don't define HOTSWAP
16:36:14funmanyes i will, because INCREASED_SCROLLWHEEL_POLLING will be true
16:36:39kugelah, I mean !INCREASED_SCROLLWHEEL_POLLING of course
16:37:13funman#if defined(HAVE_HOTSWAP) || \ (defined(SANSA_FUZEV2) && !INCREASED_SCROLLWHEEL_POLLING)
16:37:21funmanthis one is good ?
16:37:37funman(i checked that it builds this time :s )
16:37:47kugelit will work for sure, but I'm thinking the defined(SANSA_FUZEV2) is redundant
16:38:21funmanredundant with !INCREASED_SCROLLWHEEL_POLLING ?
16:38:45kugelah wait, it's 0 for non-scrollwheel targets as well
16:38:48funmanone is configurable not the other
16:39:18*jhMikeS wonders at what these companies will come up with just to look clever and innovative, then totally screw up the implementation
16:39:21 Quit n1s (Ping timeout: 264 seconds)
16:40:03funmanbuild fuzev1 with HOTSWAP disabled and INCREASED_SCROLLWHEEL_POLLING disabled -> build breaks
16:40:20kugelso, #if defined(HAVE_HOTSWAP) || \ defined(HAVE_SCROLLWHEEL) && !INCREASED_SCROLLWHEEL_POLLING) should be most correct
16:40:39kugelit sure does
16:40:46funmanthen the condition is incorrect
16:40:47kugel"you cannot activate it by simply defining it, but that's the reason it's not in config.h, it still describes the wheel mechanism"
16:41:02kugelconfigurability doesn't really matter here
16:41:06funmanok then
16:41:30funmani'll keep FUZEV2 because it's more explicit
16:41:54kugelit really doesn't matter because we probably won't get more targets anytime soon
16:41:58CIA-5New commit by funman (r26166): fix previous commit
16:42:04kugelfor as3525*
16:42:06funmansansafuzev3 !!
16:42:49funmanwith dual HD screen & 3d wheel
16:43:45 Join bmbl [0] (~Miranda@unaffiliated/bmbl)
16:45:56jhMikeSfunman: 3d wheel? is that a sphere?
16:46:30funmanyeah i guess
16:46:34*funman greps for HAVE_SCROLLSPHERE
16:46:52kugelHAVE_BALLS, rather
16:47:23funmanah damned i have waited for the clipv1 to charge but i forgot to run battery bench
16:48:02 Quit bieber (Ping timeout: 276 seconds)
16:48:16 Join bieber [0] (
16:49:23funmani'm discharging clipv1/v2/+/fuzev2 to make charging graphs
16:49:50funmanafter that i think there are issues with recording, perhaps they aer fixed with interrupt priority reordering
16:50:24funmanalso with FM, i have different tuning in station mode if i go to a certain station from the previous one or the next one
16:51:19funmanperhaps the rare µSD problems are fixed too
16:52:03*funman is not sure if the amsv2 will be stable for 3.6
16:52:09kugelfunman: did your cpu_set_frequency patch work out well?
16:52:17funmani was expecting you to test it ;)
16:52:48kugelI can make me a build I guess
16:52:54funmani leave that for the end, when all the other features work ok (not sure if i will look at usb code though, i think pamaury & ranma are way more experts than me)
16:53:09funmani am not sure how multiple stores are scheduled
16:55:29funmani forgot also: there are hiss on clip+/v2 (very big on v2)
16:55:35funmanvery loud*
16:56:07kugeljhMikeS: the isr needs to be interruptable
16:56:15 Quit funman (Quit: free(random());)
16:57:07kugelI don't know if that works automagically (due to the VIC) or if I need to enable interrupts in the isr explicitely
16:58:46Unhelpfulfunman: looks like a deps issue? asmdefs is generated, and the deps script is *supposed* to fix up deps on core_asmdefs.h to give a full path.
16:58:47jhMikeSkugel: you need to program special prologue/epilogue for that that drops to SVN mode and enables interrupts. what about aborts and such?
17:00:35*Unhelpful wonders if there's a way to look at the generated deps file for the failed build
17:00:44 Join Szpila [0] (
17:02:32jhMikeSjust enabling IRQ in the ISR will corrupt SPSR if not stacked and not dropping back to SVC will corrupt LR in any nested calls because it won't be banked and the ISR doesn't necessarily restore that
17:05:05 Quit Farthen (Remote host closed the connection)
17:05:28 Join Farthen [0] (
17:08:58kugeljhMikeS: meh :(
17:09:27kugelthe lcd updates are so slow they would lock out the tick for too long I fear
17:10:56jhMikeSkugel: I was referring to other exceptions that display data like prefetch abort and such. I think filling a FIFO will be done quickly enough.
17:13:28kugelhm, maybe
17:14:09kugelit ~5MB/s, the fifo is 512kB big
17:15:46kugelbut: the button driver (from tick task) waits for the fifo to empty before reading the actual buttons so in irq would be triggered in there. will the re-fill isr executed directly after the tick tasks or or will it be dropped?
17:16:01 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
17:17:00 Join n1s [0] (~n1s@rockbox/developer/n1s)
17:23:16jhMikeSkugel: the pending interrupt should stick until serviced and cleared
17:23:27***Saving seen data "./dancer.seen"
17:24:04 Nick fxb is now known as fxb__ (
17:24:24*jhMikeS did a mini battery bench on the S and it seems to pull ahead a little of where the 8.5 hr bench was going
17:24:53jhMikeSat least it show it's not worse off by now :D
17:27:30 Quit lpereira (Quit: Leaving.)
17:27:49n1sjhMikeS: ah, i'm running one now with r 26153 (just before your lcd power commit) to get a baseline, it's just gone on 6h10m now
17:29:09jhMikeSn1s: I hope I've not set the shutoff too high. I don't have alot of data on more recent revisions where to set it exactly.
17:29:34n1si'll post my logs when i finnish if you are interested
17:30:02jhMikeSn1s: great! I look forward.
17:30:18n1saaand, just there it died
17:30:37jhMikeSdied or shut off?
17:31:10n1sdunno, i heard a HD click and the music stopped, the display wasn't on at the time
17:31:31jhMikeSI suppose the bench log will tell
17:31:43n1sthis was with ~200kbps vorbis
17:31:56 Join MethoS- [0] (~clemens@
17:32:56n1sbootloader reported 3.648V but as soon as it started RB it shut down (this time properly at least)
17:33:32n1smaybe the bootloader should refuse to start RB if it will just shut down
17:34:37jhMikeSif it could mange to start RB, I'm guessing it's a bit on the high side, maybe
17:35:22jhMikeSon mine, 3.6-ish meant it had about 15min left on the 8.5hr test
17:36:14n1sthe log says it did a proper poweroff, at 3.722V
17:36:28 Quit linuxstb (Ping timeout: 258 seconds)
17:39:00jhMikeS3.772 ?? that's way too soon. it's not supposed to unless some HD activity faked it.
17:39:13n1sI'll charge it up and run it with latest svn unless it gets too late
17:39:40jhMikeSthat's between the 4th and 5th points on the discharge curve :
17:40:02n1sit's quite possible this battery is dying
17:40:23n1snot 3.772, 3.722
17:40:25jhMikeSI think I'm gonna have to compensate the voltage reading when the disk is spinning, it makes battery reading difficult
17:40:36n1sgtg now
17:43:07 Part watto
17:43:54CIA-5New commit by ranma (r26167): Handshake still grinds to a halt at some point, but it works better now.
17:44:52 Join toffe82 [0] (~chatzilla@
17:47:58CIA-5New commit by teru (r26168): skin_parser.c: fix possibile overflow in parse_setting_and_lang(). simplify comparison of string in parameter in parse_touchregion().
17:56:28CIA-5New commit by teru (r26169): sudoku: remove commented out code.
17:58:01 Join webguest86 [0] (
18:01:11 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
18:01:43CIA-5New commit by alle (r26170): Fix another not nice apostrophe (FS #11296)
18:02:59 Quit petur (Quit: collecting toys at the postoffice :))
18:03:33 Quit webguest86 (Quit: CGI:IRC (EOF))
18:04:02 Quit teru (Quit: Quit)
18:05:57 Join webguest57 [0] (
18:06:44CIA-5New commit by alle (r26171): Use nice 'times' glyphs for describing dimensions (FS #11295)
18:07:28 Quit GeekShadow (Ping timeout: 276 seconds)
18:08:37 Quit webguest57 (Client Quit)
18:12:06 Quit linuxstb (Ping timeout: 258 seconds)
18:12:44 Quit antil33t (Read error: Connection reset by peer)
18:12:50 Join antil33t [0] (
18:14:55 Quit kugel (Read error: Connection reset by peer)
18:16:48 Quit n1s (Ping timeout: 246 seconds)
18:32:47 Join phanboy_iv [0] (
18:34:21CIA-5New commit by alle (r26172): Show another way to learn about convbdf
18:39:23 Join Boldfilter [0] (
18:51:02 Join domonoky [0] (~Domonoky@rockbox/developer/domonoky)
18:52:45 Join Peter93 [0] (
18:54:08 Quit hebz0rl (Read error: Connection reset by peer)
18:58:15 Join polobricolo [0] (
18:59:51 Join fml [0] (
19:00:01polobricoloHi, i'm porting rockbox to a stm32 (arm cortex) device. To what should i set ARM_ARCH ? CORTEX ? C ?
19:00:20fmlHrm... How can I revert a commit? svn up -r <older_release>, and then commit?
19:00:42polobricoloalso would rockbox work even though its a thumb2 only arm ?
19:00:53gevaertsfml: no
19:01:12fmlgevaerts: how then? r26171 should be reverted
19:01:22polobricolofml: i think you use merge (not sure)
19:01:33 Quit esperegu (Remote host closed the connection)
19:02:06gevaertsfml: in this case I'd probably do "svn diff -c 26171 | patch -p0 -R", but svn merge would be cleaner
19:03:17gevaertsI *think* svn merge -r 26171:26170 should do it
19:03:54gevaertsOr svn merge -c -26171
19:04:35Tornepolobricolo: Currently if we build Rockbox for thumb it doens't work
19:04:47Tornepolobricolo: so I would guess that building for a Cortex-M is also going to have problems
19:04:51Tornebut who knows.
19:05:11polobricoloOh my arm is a Cortex-M3
19:05:15 Join DerPapst1 [0] (
19:05:17TorneDont' make it CORTEX, though, that covers Cortex-A as well :)
19:05:42polobricoloI'll define it CM3
19:05:46TorneCortex-M3 architecture is ARMv7-M
19:05:54polobricolowhat's the problem with thumb ? threads ?
19:05:55Torne(Cortex-A series is ARMv7-A)
19:06:03fmlgevaerts: all those commands do not work (for me)
19:06:21TorneM3 is a kind of processor, not an arch, so I wouldn't suggest that..
19:06:27Torneit's a v7-M, srsly.
19:06:42gevaertsfml: try "svn merge -c -26171 ."
19:07:07 Quit DerPapst (Ping timeout: 276 seconds)
19:07:19domonokythose arm version numbers are really strange..
19:07:25polobricolooh i just realised that the family s different from archtecture version (eg ARM8 is an ARMv4)
19:07:41 Quit flydutch (Quit: /* empty */)
19:08:02fmlgevaerts: that seemed to work. Thanks!
19:09:58domonokypolobricolo: ARM_ARCH is used like: "ARM_ARCH >= 5" so using CM3 would probably not work :-)
19:10:18CIA-5New commit by alle (r26173): Revert r26171. Special macro for the image base name is needed. Will be committed soon.
19:10:59*domonoky thinks ARM_ARCH should be 7 for a Cortex cpu
19:12:02polobricoloi've just put it to 7
19:14:18polobricolowhat is ATTRIBUTE_PRINTF ?
19:14:18 Quit Schmogel (Quit: Miranda IM! Smaller, Faster, Easier.
19:14:34 Quit DerPapst1 (Quit: Leaving.)
19:14:44 Quit fml (Ping timeout: 240 seconds)
19:16:22*domonoky thinks that is the special attribute from gcc for variable parameters..
19:17:47 Join wincent [0] (
19:19:00 Join Horscht [0] (~Horscht2@xbmc/user/horscht)
19:19:03 Quit wincent_balin (Ping timeout: 268 seconds)
19:20:44 Quit evilnick (Ping timeout: 240 seconds)
19:22:05 Join DataGhost [0] (
19:22:05 Quit DataGhost (Changing host)
19:22:05 Join DataGhost [0] (~dataghost@unaffiliated/dataghost)
19:23:31***Saving seen data "./dancer.seen"
19:23:36 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
19:24:40 Join GeekShado_ [0] (
19:25:00 Join stripwax [0] (
19:25:16 Quit stripwax (Client Quit)
19:28:34 Quit GeekShadow (Ping timeout: 276 seconds)
19:28:46 Join liar [0] (
19:29:20 Join notlistening [0] (
19:29:45 Quit Peter93 (Ping timeout: 260 seconds)
19:30:41CIA-5New commit by funman (r26174): as3514.c: clean up ...
19:33:31 Join esperegu [0] (~quassel@
19:33:36 Join arbingordon [0] (~w@unaffiliated/arbingordon)
19:33:39 Join _arbingordon [0] (
19:34:14CIA-5New commit by kugel (r26175): Revert r26060. A better fix was found. The acceleration should be fixed once for all now.
19:35:26 Quit arbingordon (Client Quit)
19:35:26 Quit _arbingordon (Client Quit)
19:37:06 Join arbingordon [0] (~w@unaffiliated/arbingordon)
19:37:09 Join _arbingordon [0] (
19:38:41 Quit _arbingordon (Client Quit)
19:38:41 Quit arbingordon (Client Quit)
19:40:15 Quit evilnick|ipad (Remote host closed the connection)
19:41:21 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
19:43:10Tornedomonoky: 7 for cortex-m is maybe not okay though
19:43:37Tornedomonoky: there are instructions that don't exist in v7-m
19:44:03Torneit's not really part of the same architecture series at all, just related
19:50:14 Quit linuxstb (Ping timeout: 258 seconds)
19:51:26 Join baptiste [0] (
19:52:30 Join DerPapst [0] (
19:56:49CIA-5New commit by funman (r26176): as3525: don't use IRAM for usb, and avoid usb storage using uncached addresses behind our back ...
19:58:16 Join arbingordon [0] (~q@unaffiliated/arbingordon)
19:58:16 Join arbingordon_ [0] (
19:58:58 Quit arbingordon_ (Client Quit)
19:58:59 Quit arbingordon (Client Quit)
19:59:03 Quit Kitar|st (Ping timeout: 258 seconds)
20:03:33 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
20:03:51 Join Kitar|st [0] (Kitar_st@
20:04:34 Join funman [0] (~fun@rockbox/developer/funman)
20:04:55funmansorry i hadn't seen the size of ALLOCATE_BUFFER_SIZE :/
20:05:31 Join linuxstb [0] (~linuxstb@rockbox/developer/linuxstb)
20:08:22 Join petur [0] (~petur@rockbox/developer/petur)
20:08:36CIA-5New commit by funman (r26177): revert part of r26176 to avoid large static allocation ...
20:09:03 Join panni__ [0] (
20:11:45 Quit panni_ (Ping timeout: 260 seconds)
20:12:45 Join perfectdrug [0] (
20:12:45 Quit perfectdrug (Client Quit)
20:13:08CIA-5New commit by funman (r26178): as3525: hide our UNCACHED_ADDR from usb_storage.c
20:13:14 Quit Forsaken_Boy (Ping timeout: 258 seconds)
20:15:40funmanpixelma: on c200/e200, FM has the same volume when recording or not?
20:15:48 Join krabador [0] (
20:16:02funmanon clip the volume is louder when not recording
20:19:56funmanhm not only, when using passthrough the sound is different
20:21:15pixelmaI never recorded from radio on any swcodec target, would need to check
20:25:03funmanno need to record, just entering the record menu from FM playback should activate passthrough
20:27:02 Quit baptiste (Quit: ChatZilla 0.9.86 [Firefox 3.5.7/20100106054534])
20:28:33pixelmait's quieter on the c200 too
20:30:07pixelmaand also seems to sound differently (less heights I think) not entirely sure if this is only in my mind though because it's quieter
20:31:48 Join ssorgatem__ [0] (~ssorgatem@
20:33:10 Nick YPSY is now known as Ypsy (
20:33:32 Join ssorgatem [0] (
20:34:14 Quit ssorgatem_ (Read error: Connection reset by peer)
20:35:54*polobricolo just discovered ARMv7 is completely different from other arms and hates it.
20:36:11TorneARMv7-M is completely different, yes
20:36:24TorneARMv7-A is not
20:37:32 Quit ssorgatem__ (Ping timeout: 264 seconds)
20:39:16polobricoloCPSR doesn't even exist. APSR replaces it but it has a different format :(
20:41:15polobricoloThis is going to be REALLY hard to port to rockbox (if i ever manage to port it)
20:42:00Torneyeah, for the most part i would say you probably *don't* want to treat it as being an ARM device
20:42:17TorneARM_ARCH == 7 will cause major problems because lots of ARMv7-A instructions don't exist on M3
20:42:24bluebroth3rssorgatem: pong
20:42:51bluebroth3rssorgatem: for problems with the translation page ping rasher. He created that page, I just pointed to it :)
20:42:53polobricoloProbably. I think i'm going to create a new ARMv7-M cpu ....
20:42:58kisakI imagine it is excruciating to adapt hardware to fit the software
20:43:03polobricoloor add lots of defines
20:43:50funmangevaerts: any idea why you made TCC targets use a static buffer for USB ? r18703 commit log just mentions reorganisation
20:44:18gevaertsdid I?
20:44:26funmanyou did!
20:45:25amiconnpixelma, funman: No surprise that fm recording sounds different on e200/c200 - it can't record at 44.1kHz sample rate, only at 22.05kHz
20:45:46funmanamiconn: just speaking of passthrough, not the actual recording
20:46:08amiconnYes - digital passthrough *is* using the adc
20:46:32funmanadc -> dac ?
20:46:40funmani thought the analog signal was sent to the headphones
20:46:50gevaertshm, strange
20:46:55amiconnThe analog signal is sent to the headphone when not recording
20:47:08funmanah true
20:47:16polobricolobut at same time ARMv7-M supports hardware multithreading (which might be better than soft)
20:47:34funmangevaerts: i am still trying to understand what this UNCACHED_ADDR does here
20:48:01gevaertsfunman: ah, there I disclaim all knowledge. It's magic
20:48:13*amiconn still doesn't understand why AMS built a shitty audio codec like this
20:48:46gevaertsfunman: if I understand it correctly, UNCACHED_ADDR aliases the address to a range where the CPU doesn't use the cache at all
20:48:56funmanthat's right
20:49:13funmanbut there's also a side effect on non-as3525 targets i think: it returns a physical address
20:49:59gevaertsyes, possibly. You obviously need an address that the controller driver can handle
20:50:27funmanbut other usb code doesn't give uncached (or physical) addresses to usb_drv_ functions
20:50:37funmanand that's something that should go into the controller code
20:50:48 Quit ssorgatem (Read error: Connection reset by peer)
20:51:11 Join ssorgatem [0] (
20:51:49gevaertswhich code doesn't?
20:51:59*gevaerts points to USB_DEVBSS_ATTR
20:52:03funmanamiconn: ok for fs#11189 ?
20:52:22funmangevaerts: usb_hid or usb_serial
20:52:30funmanor usb_core even
20:52:41gevaertsthey do
20:53:29pixelmaok, that's why it sounds worse... sure
20:53:42funmanthey use USB_DEVBSS_ATTR, but not UNCACHED_ADDR macro
20:53:58TorneUNCACHED_ADDR returning a physical address sounds kinda weird
20:54:21Tornei'd expect it to return an uncached virtual alias if you have an MMU...
20:54:33gevaertsyes, that's what I understand it to do
20:55:01Torneif you need a physical address "uncached" doesn't make any sense
20:55:01gevaertswell, it's physical too on PP, but we don't use an MMU there
20:55:19Tornewell, not on arches with an MMU
20:55:21Torneon PP it's different ;)
20:56:28gevaertsfunman: USB_DEVBSS_ATTR is uncached by careful choice of other properties I think
20:57:37funmanwhy is that needed?
20:58:32gevaertsit allows not having to flush things
20:59:31gevaertsIf you define USB_DEVBSS_ATTR as something cached, and your controller driver handles that, fine. I predict slower MSC then though
20:59:53gevaertsWhy is it a problem?
21:00:29funmani expected the controller code to handle that - but i admit i don't know much about usb
21:06:49 Join bluebrother [0] (
21:06:50 Quit bluebrother (Changing host)
21:06:50 Join bluebrother [0] (~dom@rockbox/developer/bluebrother)
21:07:28funmanUSB_DEVBSS_ATTR is only uncached for gigabeats
21:08:02funmanfor other targets it's either in iram, either aligend on cacheline size
21:08:56 Join Kitr88 [0] (~Kitar_st@
21:09:04 Quit Kitar|st (Read error: Connection reset by peer)
21:09:57funmandoes it use DMA ?
21:10:18 Join kaniini [0] (
21:10:22 Quit kaniini (Read error: Connection reset by peer)
21:10:31 Quit bluebroth3r (Ping timeout: 264 seconds)
21:10:34funmanoops i had typed something about usb-drv-arc.c but i deleted it :P
21:10:38 Nick fxb__ is now known as fxb (
21:10:48funmanit = ARC driver
21:11:03gevaertsit does, yes
21:12:01 Join Luca_S [0] (
21:12:17 Join Kitar|st [0] (Kitar_st@
21:12:39funmanusb_drv_recv() is given both iram addresses and uncached addresses, perhaps both correspond to physical addresses
21:13:23gevaertsthey need to be physical addresses, yes
21:13:31 Quit Kitr88 (Ping timeout: 264 seconds)
21:13:33funmanPP caches work for iram too ?
21:13:43 Join hp_sebastian [0] (
21:16:51 Nick fxb is now known as fxb__ (
21:16:55 Nick fxb__ is now known as fxb (
21:17:14funmankugel: my patch for set_cpu_frequency() didn't crash so far, we'll see in the long run
21:17:38 Nick fxb is now known as fxb__ (
21:17:41 Nick fxb__ is now known as fxb (
21:17:57 Join Peter93 [0] (
21:18:00 Quit sharp (Remote host closed the connection)
21:18:06 Nick fxb is now known as fxb__ (
21:18:10 Nick fxb__ is now known as fxb (
21:22:20 Quit einhirn (Quit: Miranda IM! Smaller, Faster, Easier.
21:22:23 Quit DataGhost (Ping timeout: 240 seconds)
21:22:30 Part Szpila
21:22:51 Join Szpila [0] (
21:23:33***Saving seen data "./dancer.seen"
21:25:08jhMikeSfunman: why not just change the memory mapping so that the uncached address is the physical one? Is there some hardware reason?
21:25:13funmanhm i don't see manuals for fuzev2 and the samsungs, i swear they would be here
21:25:23funmanjhMikeS: no specific reason, it was just done that way
21:25:53funmanwell it was easier to have rockbox built with physical address because the MMU is not setup in the bootloader
21:27:11 Quit GeekShado_ (Ping timeout: 240 seconds)
21:27:48funmani think crt0.S has position dependant code
21:27:53jhMikeSany reason? imx sets it up in the bl, and the bl executes at 32MB *after* remap, but starts at a different location. firmware is copied down at 0 then.
21:28:15funmanreason was that mmu/caching code didn't work back then ;)
21:30:33jhMikeSah, that's a poor reason now....erm, right? :)
21:30:33funmanbut if UNCACHED_ADDR means UNCACHED_PHYSICAL_ADDR it should be renamed
21:30:33 Quit wincent (Ping timeout: 264 seconds)
21:30:33funmanso far it's used in greylib and mpegplayer for PP, in usb_storage.c, and in target code
21:30:33DBUGEnqueued KICK funman
21:30:33funmani think greylib & mpegplayer really means 'uncached', to protect from the other core
21:30:33jhMikeSthe problems might go further than that since if uncached=physical, cache handling doesn't have to be done when working with device addreses
21:31:03funmanyou mean hardware registers?
21:31:50jhMikeSno, I mean things such as using memcpy and then sending that buffer to a device. if uncached=physical, it's coherent already
21:32:20jhMikeSpp and imx avoid alot of flush_cache stuff that way
21:32:51 Quit Guest55923 (Ping timeout: 240 seconds)
21:33:10 Join feisar_ [0] (
21:33:32funmanwell you can write to the uncached address & give the physical address to the hardware
21:33:42jhMikeSthe other way has hazards, like the processor could write back from the cached address even after copy to the uncached but virtual one
21:33:49 Quit bluefoxx (Ping timeout: 252 seconds)
21:35:26funmanin sd-as3525.c there is 2 buffers: one is sent to the DMA controller, the other one is the uncached alias and is the one used for memcpy
21:35:39amiconnUNCACHED_ADDR in greylib is for running the isr on the cop
21:35:56funmanamiconn: ping for fs#11189
21:37:32CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
21:37:32*funman likes
21:40:03*jhMikeS thinks it should be changed to uncached=physical, then all targets can share the same code without condition
21:40:17 Quit [CGL] (Remote host closed the connection)
21:41:30 Quit Strife89 (Quit: Reboot.)
21:48:19 Join Strife89 [0] (
21:48:19 Part Szpila
21:50:27funmanthere's only this line in usb_storage.c
21:51:53funmani think it would be doable, but we would need to change both bootloader & build at the same time?
21:51:53funmanunless we make crt0 PIC
21:51:53funmanat least until it sets up MMU
21:57:52funmanreplacing ldr in the vectors by b, and bl by mov lr/bx would work
21:58:18funmanfor old bootloader+new rockbox.sansa
22:04:21 Join yawny [0] (
22:06:12 Quit elcan (Ping timeout: 240 seconds)
22:10:04 Join n1s [0] (~n1s@rockbox/developer/n1s)
22:11:12 Quit bieber (Ping timeout: 260 seconds)
22:13:03 Quit bmbl (Quit: Bye!)
22:17:28mc2739funman: Fuze manual build gets this error: ! LaTeX Error: File `platform/sansafuzev2.tex' not found.
22:17:50mc2739funman: same for the Samsungs also
22:18:31 Join Schmogel [0] (
22:18:44 Quit notlistening (Read error: Connection reset by peer)
22:19:30jhMikeSfunman: you should look at the imx code. it does some footwork to start at 8a0000000, setup the MMU and run at 82000000 and use ints in the bl and fw. perhaps you'll improve upon the scheme and I'll steal that back. :)
22:19:49funmanjhMikeS: i'm ok with current situation
22:22:15*jhMikeS dislikes it wouldn't want to work around it (and a clip is arriving in the mail)
22:22:53funmancool! which one you you get?
22:24:04jhMikeSexcuse me, not clip, fuse. type? cheap! woot sucker special, some 4gb thing with an SD slot. (thx saratoga iirc)
22:28:14 Join stripwax [0] (
22:30:27funmanah i just see that raw m2v streams don't seek in mpegplayer
22:30:50funman(and they don't have sound, but it's normal, right? ;)
22:31:44n1syeah, they dont' *have* sound :)
22:31:46jhMikeSm2v is just video
22:33:03jhMikeSthere's no times in a raw stream either. either you have to parse the whole stream, adding frame durations or perhaps it could seek by percent.
22:37:41 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
22:41:52CIA-5New commit by funman (r26179): sd-as3525v2: masked interrupts and block size won't change, and DMA won't be disabled, so we can just set these in the controller init once for all
22:42:24saratogafunman: i looked at that firmware file from abi that broke that guys clipv2, and its identical to the one i just made using our tools and the current bootloader
22:43:07funmani expected no problem with the file
22:43:33funmanFlynDice clip+ broke while doign nothing too
22:43:46saratogaso a bug in the OF's update code maybe?
22:45:02funmanyep or battery problem when updating
22:45:35funmangrr i have no idea why we have noise when doing SD ops
22:45:42 Quit Jaykay (Quit: ChatZilla 0.9.86 [Firefox 3.6.3/20100401080539])
22:46:19jhMikeSmake sure it's charged before updating huh? I've been reluctant because iiuc there's no recovery. what does that mean exactly for making bootloaders?
22:46:49funmanjhMikeS: nothing special, all the dangerous code is in mkamsboot
22:47:07saratogathe bootloader actually runs after the code that does the bootloading :)
22:47:24funmanyeah, rockbox bootloader is stage2 ;)
22:47:29 Quit krabador (Quit: Sto andando via)
22:49:13jhMikeSit's actually flashing it?
22:49:22funmanno just written to SD
22:49:42saratogathe only memory on the device is a couple KB of actual ROM masked into the chip and the NAND
22:49:54saratogaso everything you can write has to go into the nand
22:52:12dfkt<funman> yep or battery problem when updating <−− i tried updating a clipv1 with low battery once - the stock firmware complained about that and didn't want to flash
22:52:59saratogacould just be a bad sector on his NAND that happened to get written to
22:53:15funmanshouldn't SD stuff handle that?
22:53:23funmanhe didn't say if he saw 'firmware upgrading'
22:53:37saratogawell it should, but theres no way to predict if a sector is going to fail until you try
22:53:58saratogathough you'd think the OF would read back what it wrote to see if it succeeded
22:56:23ssorgatemwho takes care of language updates?
22:57:19jhMikeSsaratoga: so the OF is completely responsible for the upgrading and the 2KB ROM has no code to get at the NAND?
22:57:39saratogajhMikeS: AFAIK yes
22:58:04saratogai think on the e200v2 which had a recovery mode of sorts that gave direct flash access, you could fix it just by dd'ing a bin file to the flash
23:01:29 Quit dfkt (Quit: -= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.)
23:01:32 Join Forsaken_Boy [0] (~chatzilla@
23:04:08 Quit Schmogel (Ping timeout: 248 seconds)
23:05:02Luca_Sthere's also the "water damage" mode - somewhere in the forums it was reported that after water damage (probably shorting something) a clip device showed a small (4Mb) usb mass storage device
23:05:33Luca_Sbut it has been already dismissed as not useful as it seems read only
23:06:15funmanSTORAGE_WANTS_ALIGN work for all storage transfers now ?
23:10:29funmannope, codecs crash :/
23:13:49 Quit Xerion (Quit: )
23:20:56 Part domonoky
23:21:29 Quit Peter93 (Quit: Miranda IM! Smaller, Faster, Easier.
23:21:55 Join GeekShado_ [0] (
23:22:33 Quit stripwax (Quit:
23:23:35***Saving seen data "./dancer.seen"
23:28:27funmanhmm it seems the OF too makes that noise
23:28:46 Join wodz [0] (
23:31:01wodzAm I correct that battery graph in debug->battery should "move" once per minute?
23:31:44bertrikyes, I think it's once a minute
23:32:29 Quit efyx (Remote host closed the connection)
23:32:36wodzHmm that is strange. On MPIO it inserts new value once per several minutes
23:34:28bertrikI'd check the source code to be sure of the interval. jhMikeS is the one who did the latest rework of the power management stuff AFAIK
23:36:31 Join CGL [0] (~CGL@
23:36:39bertrikIt's a 60 second interval indeed. Maybe your HZ timer is running slow?
23:38:55 Quit GeekShado_ (Quit: The cake is a lie !)
23:42:26 Quit esperegu (Remote host closed the connection)
23:43:51 Quit n1s (Quit: Lmnar)
23:45:46 Quit bertrik (Remote host closed the connection)
23:50:14 Join max242 [0] (
23:50:45 Quit max242 (Client Quit)
23:51:18 Join max242 [0] (
23:52:43max242do all you fuze v1 owners wait till 'refreshing your media' message disappears after changing the content on the uSD?
23:53:29max242i typically abort the whole OF update process by pressing the 'power on' button for more then 10 seconds.
23:53:42max242but could that corrupt the filesystem?
23:53:58funmanif you umount/eject it should be ok i guess
23:54:51 Quit petur (Quit: Zzzzz)
23:54:58max242hmm, 'cause i want to check if the picture flow freeze issue is gone now, but have troubles running the latest build
23:55:21max242i guess the filesystem on my fuze got corrupted
23:55:53max242the fuze simply refuses to boot with r26179
23:58:03max242hmm, i think i really have to format my 8G fuze, but i first want to move all of my music to a safe place ... that's gonna take a while
23:58:28 Join stripwax [0] (
23:58:31funmanworks fine here
23:58:39max242so testing out pictureflow will not be for tonight anymore
23:58:51funmanperhaps simpler to just repair the filesystem ?

Previous day | Next day