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 2009-08-12

00:00:04 Join linuxstb_ [0] (n=linuxstb@rockbox/developer/linuxstb)
00:00:10 Quit linuxstb (Nick collision from services.)
00:00:14 Nick linuxstb_ is now known as linuxstb (n=linuxstb@rockbox/developer/linuxstb)
00:00:46martian67DarkSpectrum, toor?
00:01:09JdGordon|isnt it in the wiki page?
00:01:19DarkSpectrumonly has the user login
00:01:26bluebrotherDarkSpectrum: check the wiki, IIRC it's documented somewhere there.
00:01:32DarkSpectrumgot it
00:01:43JdGordon|add it to the wiki
00:01:45JdGordon|which was it?
00:04:27 Quit bluebrother ("leaving")
00:10:58 Quit bmbl ("Bye!")
00:12:43 Quit Lynx_ (" Try HydraIRC -> <-")
00:16:53 Quit TruthTaco (Read error: 110 (Connection timed out))
00:17:38 Join TruthTaco [0] (
00:17:55 Quit stripwax (Read error: 104 (Connection reset by peer))
00:19:19 Quit J-23 ("Flying cow pressed ^D on my keyboard.")
00:21:12 Quit pamaury ("Quitte")
00:29:41 Quit linuxstb (Read error: 110 (Connection timed out))
00:34:28 Quit ender` (" Computers work fine as long as users aren't allowed anywhere near them.")
00:36:08 Quit efyx (Remote closed the connection)
00:36:19 Quit captainkwel ("Page closed")
00:46:25 Join bittin_ [0] (
00:55:57 Quit Zagor ("Clint excited")
00:57:12 Quit bertrik (Read error: 113 (No route to host))
00:57:48 Quit GeekShadow ("The cake is a lie !")
00:59:27Unhelpfulgevaerts: if it's 2D, are there going to be "corner" covers, or just covers extending up and down from center as well as to each side?
01:00:21gevaertsI'd just use a cross
01:00:39Unhelpfuli go to album *fairly* often, because when i'm out with my family that's a faster way to find something that i can actually listen to with my daughter
01:01:47Unhelpfulok... a cross with the up/down items displayed flat is fairly easy to draw. with the up/down items tilted will require some new drawing code. with corner covers that are tilted around both x and y axes i don't want to touch (and it will be slow on archos then ;)
01:03:46 Join Strife1989 [0] (
01:03:49Unhelpfuli have no idea how i would actually organize the data in that case, either... i suppose keeping the same all-albums list adding an artist list with a pointer to the first album per artist would do
01:04:12 Quit Strife89 (Nick collision from services.)
01:04:20 Nick Strife1989 is now known as Strife89 (
01:04:52Unhelpfulalso, nobody but you has "artist art"... and you have that as your album covers :P
01:06:25gevaertswell, maybe there's another way, but I really think that a single linear list of all albums must be about the worst possible way to organise your music
01:07:27 Quit jfc (Read error: 104 (Connection reset by peer))
01:07:36Unhelpfulgevaerts: it's pretty, though ;)
01:08:44 Quit dmb (Read error: 104 (Connection reset by peer))
01:11:00Unhelpfulletting the accel go faster doesn't, by itself, strike me as a great idea, since at top speed the animation gets down to about one frame per cover. i think that switching to letters when you hit the accel cap might be nifty...
01:13:04***Saving seen data "./dancer.seen"
01:21:15 Join webguest63 [0] (n=8fa6e22a@gateway/web/cgi-irc/
01:21:53 Quit z35 (Read error: 113 (No route to host))
01:24:05 Quit webguest63 (Client Quit)
01:25:58 Quit JdGordon| ("Miranda IM! Smaller, Faster, Easier.")
01:26:25 Join thegeek [0] (
01:32:49 Quit Thundercloud (Remote closed the connection)
01:41:49 Join JdGordon_ [0] (i=ad7541b0@gateway/web/freenode/x-ebdf8a0f67211f03)
01:43:04 Join brenosabino [0] (n=bb28bc6d@gateway/web/cgi-irc/
01:45:14 Join CaptainKwel [0] (
01:47:41 Quit Strife89 (Read error: 110 (Connection timed out))
01:49:02 Quit faemir ("Leaving")
01:56:29 Quit brenosabino ("CGI:IRC (EOF)")
01:57:39 Part toffe82
02:00:08 Quit jgarvey ("Leaving")
02:00:40 Quit Rondom (Nick collision from services.)
02:00:51 Join Rondom [0] (
02:04:43 Quit pixelma_ (" .")
02:04:44 Quit DarkSpectrum (Read error: 104 (Connection reset by peer))
02:05:25 Quit JdGordon_ (Ping timeout: 180 seconds)
02:44:43 Join advcomp2019_ [0] (n=advcomp2@unaffiliated/advcomp2019)
02:46:12 Quit markun (
02:46:12 Quit panni_ (
02:46:12 Quit Beta2K (
02:46:12 Quit gevaerts (
02:46:12 Quit AndyIL (
02:46:12 Quit Ridayah (
02:46:12 Quit Dhraakellian (
02:46:12 Quit maraz (
02:46:12 Quit linuxguy3 (
02:46:12 Quit blithe_ (
02:48:57 Quit Llorean (
02:48:57 Quit tarbo_ (
02:48:57 Quit krazykit (
02:48:57 Quit preglow (
02:48:57 Quit rphillips (
02:48:57 Quit avacore^ (
02:48:57 Quit advcomp2019 (
02:48:57 Quit dionoea (
02:48:57 Quit fuzzie (
02:48:57 Quit gibbon_ (
02:48:57 Quit pike (
02:49:14kugelUnhelpful: any plans on committing the lcd driver work?
02:49:35NJoinpanni_ [0] (
02:49:35NJoinBeta2K [0] (
02:49:35NJoingevaerts [0] (n=fg@rockbox/developer/gevaerts)
02:49:35NJoinAndyIL [0] (i=AndyI@
02:49:35NJoinRidayah [0] (
02:49:35NJoinlinuxguy3 [0] (
02:49:35NJoinmaraz [0] (
02:49:35NJoinDhraakellian [0] (
02:49:35NJoinmarkun [50] (n=markun@rockbox/developer/markun)
02:49:35NJoinblithe_ [0] (
02:49:51 Join Llorean [0] (n=DarkkOne@rockbox/user/Llorean)
02:49:51NJointarbo_ [0] (n=me@unaffiliated/tarbo)
02:49:51NJoinkrazykit [0] (
02:49:51NJoinpreglow [0] (
02:49:51NJoinpike [0] (n=pike@xbmc/gc/pike)
02:49:51NJoinadvcomp2019 [0] (n=advcomp2@unaffiliated/advcomp2019)
02:49:51NJoinrphillips [0] (n=rphillip@unaffiliated/rphillips)
02:49:51NJoinfuzzie [0] (
02:49:51NJoingibbon_ [0] (
02:49:51NJoinavacore^ [0] (
02:49:51NJoindionoea [0] (n=dionoea@videolan/developer/dionoea)
02:49:59Unhelpfulkugel: markun had remembered objections being raised to the merge at some point... i was thinking i ought to take a call on the mail list for comment before merge.
02:50:17kugelwhat objection?
02:50:29Unhelpfulhe didn't remember that part :)
02:50:53kugelthat's not a valid objection then
02:50:57kugelI can't really imagine any
02:52:45Unhelpfulcoding style comes to mind, or binsize - a choice has to be made between accepting positive deltas on some targets or having C-code includes.
02:52:45 Quit thegeek (Read error: 54 (Connection reset by peer))
02:53:24Unhelpfuli personally think that the latter are OK pretty much any time that they provide benefits - in this case reduced duplication without a binsize hit.
02:54:31 Quit jhulst ("Read error: EOF")
02:54:51 Join jhulst [0] (
03:01:15 Quit preglow (
03:01:15 Quit Llorean (
03:01:15 Quit rphillips (
03:01:15 Quit avacore^ (
03:01:15 Quit advcomp2019 (
03:01:15 Quit dionoea (
03:01:15 Quit fuzzie (
03:01:15 Quit gibbon_ (
03:01:15 Quit pike (
03:01:15 Quit krazykit (
03:01:15 Quit tarbo_ (
03:01:25kugelUnhelpful: what are you talking about?
03:01:32kugelwhere do you want to include c code?
03:01:40Unhelpfulkugel: look at the patch? :P
03:02:25Unhelpfulit replaces the bulk of the LCD bitmap text code in each lcd driver with #include <lcd-bitmap-common.c>
03:02:45NJoinLlorean [0] (n=DarkkOne@rockbox/user/Llorean)
03:02:45NJointarbo_ [0] (n=me@unaffiliated/tarbo)
03:02:45NJoinkrazykit [0] (
03:02:45NJoinpreglow [0] (
03:02:45NJoinpike [0] (n=pike@xbmc/gc/pike)
03:02:45NJoinadvcomp2019 [0] (n=advcomp2@unaffiliated/advcomp2019)
03:02:45NJoinrphillips [0] (n=rphillip@unaffiliated/rphillips)
03:02:45NJoinfuzzie [0] (
03:02:45NJoingibbon_ [0] (
03:02:45NJoinavacore^ [0] (
03:02:45NJoindionoea [0] (n=dionoea@videolan/developer/dionoea)
03:02:48kugelUnhelpful: uh, I didn't see that yet
03:02:52kugelwhy is that?
03:03:23 Quit tarbo_ (SendQ exceeded)
03:03:35Unhelpfulif the functions live in a different file some targets take a binsize and speed hit due to long calls - almost all of the ARM7, and i don't *know* how it works on coldfire but binsize was larger there without the include.
03:04:04 Quit Llorean (Connection timed out)
03:04:21 Join jhulst_ [0] (
03:04:25kugelI think you're a bit too obsessed with that long calls
03:04:52kugelI don't like this including either
03:04:53Unhelpfuland i think that this shouldn't cost speed or size if it doesn't have to. :)
03:04:54 Quit advcomp2019 (Connection timed out)
03:05:00 Quit jhulst (Success)
03:05:00 Nick jhulst_ is now known as jhulst (
03:05:14Unhelpfulall of the LCD remote drivers work this way already, btw. :)
03:05:50Unhelpfulit's not *just* that, either. without the include some static variables have to be non-static, which leads to them needing renames to avoid name clashes.
03:06:34kugelwhich? the default viewport?
03:06:42Unhelpfulno, the current viewport.
03:08:27Unhelpfulon targets with a remote, making this non-static is a minor problem as they have the same name. either the current viewport needs a different name per display, or only the main-LCD current VP can be declared non-static. i don't like conditional qualifiers *much* better than a C-file include.
03:08:29kugelI'm not really convinced of that including, but this stuff is still better than the massive duplication we have now
03:10:21kugelbringing it up on the is an awesome idea :>
03:10:32UnhelpfulML? :)
03:13:08***Saving seen data "./dancer.seen"
03:17:52kugelUnhelpful: I don't think the name clash is much of a problem
03:18:43 Join moos [0] (i=mustapha@rockbox/staff/moos)
03:18:46Unhelpfulwell, adding prefixes is a rather invasive change that i'd like to stay away from
03:20:06kugelit's a search&replace job, much less controversial
03:20:15Unhelpfula conditional static qualifier is a better change - i'm *assuming* that the existence of a non-static current_vp will not override the static one in the LCD remote driver at link time?
03:21:22kugeland most of the time "current_vp" can be a #define to main_current_vp or remote_current_vp
03:22:23 Join BHSPitMonkey [0] (n=stephen@unaffiliated/bhspitmonkey)
03:22:31 Join tarbo [0] (n=me@unaffiliated/tarbo)
03:22:46Unhelpfuli also don't like the idea of this costing binsize on some targets. on top of that, there would *still* be a C-file include from lcd-remote-bitmap-common.c ;)
03:43:23 Join darkham [0] (n=Soreme@
03:44:03 Join StealthyXIIGer [0] (
03:46:51 Quit kugel (Remote closed the connection)
03:54:22 Join Strife89 [0] (
03:54:54 Quit panni_ ("( :: NoNameScript 3.81 :: )")
03:57:41 Quit parafin (
03:57:41 Quit Tristan (
03:57:41 Quit bubsy (
03:57:41 Quit Erant (
03:57:41 Quit Galois (
03:58:32 Part darkham
04:00:33 Join bubsy_ [0] (i=Bubsy@
04:00:33NJoinparafin [0] (
04:00:33NJoinTristan [0] (
04:00:33NJoinGalois [0] (
04:00:34NJoinErant [0] (
04:06:48 Quit TheSeven (Nick collision from services.)
04:07:03 Join The_Seven [0] (
04:07:07 Nick The_Seven is now known as TheSeven (
04:07:07 Join brenosabino [0] (n=bb28bc6d@gateway/web/cgi-irc/
04:09:12brenosabinohey guys, has anyone got playback working on c200v2?
04:09:32saratogai think the wiki says it works
04:10:00brenosabinoits not working for me in the latest build
04:10:11 Quit timc (Read error: 110 (Connection timed out))
04:12:29saratogatry running a build thats reported to work and see if it works for you
04:12:40saratogathen you'll know if its something wrong with your end or the current build
04:15:43 Join jfc [0] (
04:16:38brenosabinoi dont know which build was working. in the latest releases they fixed something for low ram devices like mine but it didn't help.
04:18:48saratogaI commited that, and it doesn't fix anything, it just disables a feature which doesn't work on those targets
04:20:20brenosabinobtw on the latest release when i compiled it didnt had fm radio or any plugin anymore in the build
04:20:59 Join Adam-g1 [0] (
04:21:28Adam-g1hey all, I need some help with rockbox on my cowon d2 >.> anybody help me please?
04:21:47 Quit StealthyXIIGer (Read error: 110 (Connection timed out))
04:22:12saratogaask a question
04:23:00*JdGordon would just about kill for multifont about now...
04:23:18Adam-g1well.. it seems to work pretty nicely, but when I try to play my mp3's, I can't seem to find my memory card in the file browser
04:23:40Adam-g1does it not support memory cards yet?
04:24:46Adam-g1I can seem to only access the memory that's on the D2 :/
04:29:13CIA-6New commit by 03kkurbjun (r22262): M:Robe 500: Make endpoint requests more flexible.
04:29:23brenosabinoIs there any chance that flac playback works someday on c200v2?
04:29:54brenosabinoI could play my flacs at my nokia phone but it has a damn hiss problem at the 2.5mm port.
04:29:57JdGordonno reason why not... assuming that is a target we are working on
04:30:06saratogait'll probably be fixed eventually since its so similar to the clip
04:31:19 Nick bubsy_ is now known as bubsy (i=Bubsy@
04:33:53 Quit Adam-g1 ()
04:34:10CIA-6New commit by 03kkurbjun (r22263): M:Robe 500: Put the irq stack and fiq stack in iram. Reduce memory for fiq stack since it is currently unused.
04:45:58 Quit Strife89 ("Bed.")
04:51:24 Nick advcomp2019_ is now known as advcomp2019 (n=advcomp2@unaffiliated/advcomp2019)
05:05:13brenosabinoill try build 21693 at my c240v2 now. seems to be the last working build on the clipv1 according to anythingbutipod forums
05:13:12***Saving seen data "./dancer.seen"
05:14:12brenosabinocan i patch the firmware with old bootloader using the newest mkamsboot? Or do I need to compile the old version too?
05:16:04 Quit brenosabino ("CGI:IRC")
05:16:08 Join brenosabino [0] (n=bb28bc6d@gateway/web/cgi-irc/
05:23:20brenosabinostill just plays a few seconds even with a very low bitrate (32bkps) mp3
05:24:13brenosabinotime to go to bed.
05:24:31 Quit brenosabino ("CGI:IRC")
05:33:22 Quit tchan ("WeeChat 0.3.0-rc2")
05:35:24 Join tchan [0] (n=tchan@lunar-linux/developer/tchan)
05:39:13 Quit ehntoo (Read error: 110 (Connection timed out))
05:50:38 Quit Horscht ("Verlassend")
05:51:21 Join Necos [0] (
05:52:43Necosis there anyone using the japanese language pack around?
05:55:24Necosjust trying to figure out why some characters showed up under the japanese sansa firmware and not the rockbox lang pack
05:56:12JdGordonmissing glyphs in the fonr
05:56:42Necosaccording to the svn notes, there are no glyphs missing... just an oversight?
05:57:16saratogawhich font is this?
05:57:17JdGordonit could be
05:58:52 Join DarkSpectrum [0] (
05:58:59saratogathat should support everything in unicode
05:59:18saratogaare your tags originally in unicode?
05:59:35Necosthat's what i thought... i could be wrong
06:01:20Necos浜崎あゆみ <−−−− doubt this shows up properly
06:02:32Necosi thought they were all in UTF-8... i might have screwed something up nonetheless...
06:02:47Necosaudacious and winamp both pick up the fonts, however...
06:03:25Necosalthough i have the auto-detection in audacious set for japanese
06:03:44DarkSpectrumthats awsome, didn't even know i was using a unicode font
06:04:37Necosdid it show up properly?
06:04:58DarkSpectrumfor me it did
06:05:11Necosit didn't even show up for me in urxvt lol
06:05:25DarkSpectrumanyway, on topic, is my build client still showing warnings?
06:07:28DarkSpectrumi see the 15 warnings on build 22260, but i want to verify my last build was submitted without warnings
06:07:49Necoswhat build client?
06:09:27saratogaI don't see any errors on the build table
06:09:47DarkSpectrumtrying to see if mine even submitted anything
06:09:58saratogathe logs say you did a few builds
06:10:42DarkSpectrumis there an easy way i can search the log to see what i've done?
06:10:51Necosi'll have to double-check my fonts and make sure they're in utf-8... i'll bbl
06:11:41saratogatheres a link next to each build in the table showing who built what
06:12:01 Quit Kumool ("Leaving")
06:13:00 Nick fxb is now known as fxb__ (
06:13:24DarkSpectrumyeah but there has to be a faster way then hovering over each one to see who did it
06:14:12saratogatheres a link listing each build that you can search, its on the left hand side next to each build round
06:14:18 Quit fdinel ("Miranda IM! Smaller, Faster, Easier.")
06:34:58 Quit mt (Read error: 104 (Connection reset by peer))
06:35:46 Join mt [0] (n=MTee@rockbox/developer/mt)
06:44:02 Join intrados [0] (
06:59:43 Quit mt (Read error: 113 (No route to host))
07:07:15 Join ehntoo [0] (
07:13:15***Saving seen data "./dancer.seen"
07:23:36 Join timc [0] (
07:24:27 Quit CaptainKwel (Remote closed the connection)
07:30:56 Join stoffel [0] (
07:32:17 Quit ehntoo (Read error: 110 (Connection timed out))
07:34:57 Join FOAD_ [0] (
07:51:51 Quit FOAD (Read error: 110 (Connection timed out))
07:51:52 Nick FOAD_ is now known as FOAD (
07:56:19 Quit bzed (
07:56:19 Quit martian67 (
07:57:21NJoinbzed [0] (
07:57:21NJoinmartian67 [0] (n=martian6@about/linux/regular/martian67)
07:58:35amiconnUnhelpful: Some lcd drivers use C code includes already (the ones where the same pixel format exists for main and remote displays)
07:59:23amiconnIncluding a .c isn't really nice, but okay to avoid duplication. Just never ever put such (non-inline) stuff in a .h
08:01:53Unhelpfulamiconn: that was my feeling, including a .c when there's a reason is not really that awful. and i mentioned the remote LCD includes to kugel as well. :)
08:02:11amiconnUnhelpful: Btw, what happened to the long-call veneer stuff?
08:02:51Unhelpfuli was never able to produce a decently concise source and linker script to trigger the offset problem :/
08:03:39tmztan .xpm is a c file, and it's designed to be included
08:03:43tmzt(just an example)
08:05:01Unhelpfulamiconn: markun had said he thought he remembered you having concerns about unifying LCD code - i *assume* that would be one of the earlier FS #4817 patches. he couldn't remember *what* the problem had been, though, and it's surely possible that it wasn't even you :)
08:11:17 Join flydutch [0] (
08:14:13 Quit krazykit (Remote closed the connection)
08:14:24 Join krazykit [0] (
08:14:38amiconnIf current_vp needs to be non-static, you could use the very same mechanism as the combined main+remote lcd drivers (using LCDFN() to prepend either lcd_ or lcd_remote_). The include solution wouldn't need that, of course, as it can stay static
08:17:42 Join Rob2222 [0] (
08:18:38Unhelpfulanother reason that i favor the include way, but mostly i don't like the idea of this costing a binsize increase on some targets
08:18:38amiconnRegarding the concerns I'm not sure whether it was me. If it was, it was a long time ago. Need to dig in the logs
08:21:14Unhelpfuli tried google, but only found people asking if the task was old and needed closing, and linusn suggesting at one point that it should wait until after VP merge... that's already getting a bit ancient-history ;)
08:21:31 Quit stoffel (Read error: 104 (Connection reset by peer))
08:25:05 Join ender` [0] (
08:28:36 Join Zagor [242] (n=bjorn@rockbox/developer/Zagor)
08:35:49 Quit Rob2223 (Read error: 110 (Connection timed out))
08:38:05 Join petur [50] (n=petur@rockbox/developer/petur)
08:41:54 Quit Stephen__ ("Leaving")
08:59:25 Join bertrik [0] (
09:04:51 Quit n1s (Remote closed the connection)
09:06:10 Join einhirn [0] (
09:11:46 Join linuxstb [0] (n=linuxstb@rockbox/developer/linuxstb)
09:12:53 Quit TheSeven (Read error: 113 (No route to host))
09:13:19***Saving seen data "./dancer.seen"
09:27:47 Join Thundercloud [0] (
09:30:59 Join bmbl [0] (n=Miranda@unaffiliated/bmbl)
09:40:30 Quit DarkSpectrum (Read error: 104 (Connection reset by peer))
09:40:49 Join mt [0] (n=MTee@
09:40:50 Quit shaggy-h ()
09:41:03 Join DarkSpectrum [0] (
09:43:02 Join sbhsu_ [0] (
09:43:53 Quit FlynDice (Remote closed the connection)
09:44:14 Join FlynDice [0] (
09:46:20 Quit Thundercloud (Remote closed the connection)
09:50:02 Join shaggy-h [0] (n=kiwi@
09:52:30 Join gammy [0] (
09:53:17gammyI just blew something related to analogue audio out on my iRiver iHP-120. Does anyone have just the board (or a complete player) for sale?
09:58:32 Quit sbhsu (Read error: 110 (Connection timed out))
10:01:18 Join LinusN [0] (n=linus@rockbox/developer/LinusN)
10:02:20 Join J-23 [0] (
10:12:05 Quit kachna (Read error: 113 (No route to host))
10:13:42 Quit BHSPitMonkey ("Ex-Chat")
10:16:46 Join efyx [0] (
10:21:15 Quit martian67 (
10:21:15 Quit bzed (
10:21:56NJoinbzed [0] (
10:21:56NJoinmartian67 [0] (n=martian6@about/linux/regular/martian67)
10:22:08 Quit mortti (Read error: 60 (Operation timed out))
10:22:11 Join mortti [0] (
10:22:44 Quit Zagor ("reboot time")
10:27:25 Join pamaury [0] (
10:27:54 Join Zagor [242] (n=bjorn@rockbox/developer/Zagor)
10:33:40 Join GeekShadow [0] (n=Antoine@reactos/tester/GeekShadow)
10:42:52pamauryhello, I tried to build a simulator version of rockbox following the wiki instructions and it fails at the "make fullinstall" step: "Installing a full setup in your '' dir" "ERROR: No PREFIX given"
10:43:02pamauryCan someone explain me why ?
10:52:14gammypamaury: Presumably because no [installation] prefix was given. Iknow nothing of this simulator business, but that's my assumption. Perhaps you could rectify this by specifying "−−prefix=<prefix>" during the configure.
10:53:34gammypamaury: you could also check the Makefile to see what PREFIX relates too. Perhaps you can set an environment variable for it, eg "mkdir dir; PREFIX=./dir/ make fullinstall"
10:53:42gammy"relates _to_" even
10:54:12pamaurythat's what I was trying to do but it doesn't work if I do only for make fullinstall so I'm trying it with configure. Wait and see :)
10:54:47pamauryIt works !
11:03:55 Join mt_ [0] (n=MTee@
11:07:06 Quit martian67 (
11:07:06 Quit bzed (
11:08:01NJoinbzed [0] (
11:08:01NJoinmartian67 [0] (n=martian6@about/linux/regular/martian67)
11:13:24***Saving seen data "./dancer.seen"
11:13:33 Quit freqmod (Read error: 60 (Operation timed out))
11:15:26 Join pyro_maniac [0] (
11:17:44 Quit martian67 (
11:17:44 Quit bzed (
11:18:43NJoinbzed [0] (
11:18:43NJoinmartian67 [0] (n=martian6@about/linux/regular/martian67)
11:21:20 Join n1s [0] (n=n1s@rockbox/developer/n1s)
11:25:23 Join raphi [0] (
11:25:51 Quit mt (Read error: 110 (Connection timed out))
11:26:17 Quit advcomp2019 (Read error: 54 (Connection reset by peer))
11:26:40 Join advcomp2019 [0] (n=advcomp2@unaffiliated/advcomp2019)
11:26:51 Nick mt_ is now known as mt (n=MTee@
12:01:54 Quit linuxstb ("Leaving")
12:02:03 Quit DarkSpectrum (Read error: 104 (Connection reset by peer))
12:02:10 Join linuxstb [0] (n=linuxstb@rockbox/developer/linuxstb)
12:02:22 Join freqmod [0] (
12:02:25 Join DarkSpectrum [0] (
12:07:59 Join funman [0] (n=fun@rockbox/developer/funman)
12:11:52funmani got an answer from the marketing product manager at austriamicrosystems
12:11:57funman"I think the specification for our device (AS353x series) might not give you additional information about the chip you are searching for."
12:14:07bertrikwhat a totally useless answer....
12:15:24funmanperhaps he means the chip in sansa clipv2/fuzev2 is not an as353x?
12:17:30bertrikI have no idea. Well, it's nice at least to get an response at all.
12:19:48pamauryfunman: I submit a patch on FS: the previous bugfix of logf contained a bug !
12:21:46hatseflatsis there by the way anyone thinking / working on alternative filesystems that can be used in rockbox? like ext or others?
12:23:02hatseflatsa patent ridden ratsnest like fat would be nice to avoid completely
12:23:53funmanhatseflats: rockbox isn't likely to be sued as it is nonprofit
12:24:15funmanand the fat patent used against tomtom could be worked around in the linux kernel
12:24:18pamauryfurthermore fat is really simple to implement and works flawlessly on windows and linux
12:24:49funmanso, no there isn't anyone thinking on alternative filesystems, fat just is enough and other filesystems are not wanted
12:24:50pamaurywhich is not the case of ext2/ext3. Except if we have a working MTP implementation
12:25:54funmanI have the austriamicrosystems guy's cellphone but international calls are expensive
12:25:55hatseflatsflawlessly would be such that I don't have to hack my kernel so I don't get the wrong device total after formatting
12:26:27funmanhatseflats: if you see a bug in rockbox fat driver, you can send a patch/bug report
12:26:42pamauryis there a bug in the fat driver ?
12:26:45hatseflatsfunman: sure, but I'm not quite decided on who's fault it is :)
12:28:14linuxstbhatseflats: There's resistence amongst Rockbox devs to the idea of supporting alternative filesystems - it's even on the "no-do" list. But personally I would use ext2/3 if Rockbox supported it... It just needs to be implemented in a very clean way, so targets short on RAM can easily disable it, and so it also doesn't add unneeded complexity (read: increase in code size) to the existing FAT driver.
12:31:54funmanone way would be to have probing code and then load the needed filesystem code
12:32:17funmanoh .. you can't load code from the storage if you don't have a filesystem driver yet
12:32:51linuxstbAnd what about the bootloader?
12:33:24linuxstbBut I think that's a problem we don't need to solve until someone actually implements a new filesystem. Which to be honest I doubt anyone ever will...
12:34:06bertrikmaybe a flash file system ...
12:34:46funmanthat also means losing compatibility with the OF at some point
12:34:47 Join DarkSpectrum- [0] (
12:34:47 Join gtkspert_ [0] (
12:34:55linuxstbbertrik: You mean instead of FAT32 on top of a FTL?
12:35:39linuxstbfunman: Yes, but anyone who wants to use an alternative FS would expect that.
12:35:47linuxstb(unless we're talking HFS+ on ipods)
12:35:49bertrikyes, just dreaming for now though
12:36:43linuxstbWon't that require support for the same filesystem on the user's PC?
12:38:45Tornelinuxstb: MTP would be a better bet there.
12:38:45bertrikeither that, or have MTP support in rockbox
12:39:01TorneIt's all but impossible to do flash filesystems over USB
12:39:13Tornethey're incompatible with standard block devices, and thus also with MSC
12:39:46bertrikTorne, ah right, didn't think o fthat
12:39:47*linuxstb wonders if enough devs are interested in using ext2 (or at least, are not against an implementation) to propose it as a Google SoC project next year...
12:40:23Tornebertrik: MTP would have to come first before a non-block-device FS :)
12:40:24linuxstbThat would seem a step backwards though - one of the main features of Rockbox has always been plain MSC access.
12:40:28bertrikI see no practical advantage in ext2
12:41:12TorneMSC is shit, though :L)
12:41:29TorneFAT as a transport protocol needs to be shuffled off this earth as soon as is possible :)
12:41:39 Quit mt (Read error: 104 (Connection reset by peer))
12:41:47Tornethough there are alternatives other than MTP as a replacement, admittedly
12:41:58Tornethe object storage extensions to SCSI are quite delicious. :)
12:43:12Torneplain MSC support is only an advantage because most of the alternatives are poorly supported or just plain awful at present
12:43:15Tornethis won't always be true
12:47:39hatseflatsbertrik: a flash optimized fs is generally a bad idea (tm) nowadays because of hardware wearlevelling, so if it was implemented, it should be covered in red tape and exclemation points
12:47:52hatseflatsso that people only use it if and when it's a good idea
12:48:16funmannot all rockbox targets have hardware wearlevelling
12:48:23linuxstbhatseflats: Presumably it would only be enabled on targets where there is no hardware wear-levelling. Lots of our newer flash-based targets are like that.
12:48:25hatseflatsnot all, exactly.
12:48:53bertrikI expect more potential targets to use flash instead of sdcard or harddisk in the future
12:48:58funmansandisk use SD, probably because they don't have to pay royalties
12:49:23hatseflatsimo it should be up to the user to use a fs, and rockbox should silently take it like a sm slave, but I'm having ifs and doubts with software wearlevelling fs's
12:49:26hatseflatso hwell
12:49:28hatseflats* oh well
12:49:32Tornehatseflats: er, why?
12:49:36 Quit gtkspert (Read error: 101 (Network is unreachable))
12:49:41bertrikit's a non-problem IMO
12:49:44TorneIf you have bare NAND flash as many devices do then you have to do it in software
12:50:02Torneand doing it in the FS is *better*
12:50:37Torneat least in theory. you have more information.
12:50:44Tornespecific implementations may or may not be better
12:51:28bertrikIf we would have a flash file system, we would not have to force it on non-flash block-based devices too
12:51:46TorneUhm, flash filesystems can't be used on block based devices
12:51:51Tornewithout emulation layers in between
12:51:54Tornethe opposite of an FTL, almost :)
12:51:56bertrikWe could just make it some kind of compile-time option, specific for each target
12:52:03Tornethey are completely independant concepts
12:52:40Torneblock device flash emulatoin is only used for stuff like testing flash filesystems on loopback devices :)
12:53:18 Quit DarkSpectrum (Read error: 110 (Connection timed out))
12:54:50Tornehatseflats: i think you are misunderstanding how flash FSes work, anyway. real flash FSes *cannot be used* on block devices unless you insert an emulation layer. Anything with hardware wear levelling has a block device interface.
12:55:21TorneIt's only possible to make such a mistake on an OS which has this kind of emulation support for testing, if you go and force the emulataion layer to get loaded and used (as people have been known to do with the mtdblock device in linux)
12:56:29bertrikTorne, file systems using simple block-devices have a very simple interface between them, essentially just read block X and write block X. Is there a similar interface between flash file systems and flash memories?
12:56:48TorneYes, but it's much less simple
12:57:08Torneand it's not necessarily the same for all types of flash either
12:58:17Torne <- this is the standard linux mtd nand interface
12:58:29hatseflatsah, directly speaking with the nand fs's
12:58:35Torneyou'll notice that api has dozens of functoins
12:58:35hatseflatsdon't know them, you are right.
12:58:42Tornerather than the handful that a block device has
12:59:26Tornehatseflats: there's a difference between an FS that's aware it may be running on an SSD and knows how to tweak its access pattern/etc and maybe issue ATA TRIM and similar
12:59:44Torneand an FS like UBIFS or JFFS which is constructed entirely around erase blocks and linear write block ordering and so on
13:02:19tmztisn't there a standard ftl though?
13:02:27tmztfor these Samsung devices?
13:02:28Tornetmzt: there are loads and loads of standard FTLs
13:02:36Tornedozens, i would expect
13:02:45Tornemost of them only work on specific kinds of flash
13:03:15TorneSamsung have several FTLs that they license for their flash parts
13:03:18tmztokay, so those devices with that ftl support a block layer, but a raw mtd filesystem like ubifs or jffs or yaffs2 or (I think) logfs doesn't need the ftl
13:03:32Tornetmzt: an mtd filesystem can't even *use* the ftl.
13:03:38Torneit's not tha tit doens't need it, it literally can't work :)
13:04:27Torneubifs is also a special case as ubifs uses a *different* interface, ubi, to talk to the flash, which is a higher level interface than mtd but still distinct from a block devie
13:04:52tmztunstructured/unordered block interface?
13:05:06Torneunordered, yes
13:05:33TorneUBI does wear levelling and bad block replacement, presenting an interface which is a series of unordered logical erase blocks
13:05:38tmztwhat is this about? is rockbox considering adopting one of those?
13:06:19tmztor is this about refactoring the ata stuff, since it won't support flash that doesn't resembled a disk device
13:06:20Tornepeople were suggesting that one day we could use a flash fs instead of implementing the OFW's FTL, perhaps.
13:06:38Torneon devices whose storage is raw flash
13:06:51tmztbut it would have to be able to share space with OF, or is this removing OF entirely?
13:10:41TorneThat would be removing the OF entirely.
13:10:53TorneFTLs generally don't support flash partitioning.
13:11:07 Quit funman ("free(random());")
13:13:25***Saving seen data "./dancer.seen"
13:14:56 Join ehntoo [0] (
13:18:55bertrikit seems the Samsung FTL uses a NAND API with a few functions like init, probe, read, write, erase and sync
13:24:48pamauryI have a question about sprintf and friends: the current implementation relies on a more general function called "format" which is nice because its uses a function instead of a buffer to write. Wouldn't it make sense to make it available through a header rather than hiding it ?
13:28:19Tornebertrik: that's not sufficient for all possible situations; e.g. when there is hardware to calculate and check ECCs.
13:28:48Torneand it depends if you want to address the OOB data seperately or not, and so on
13:39:40 Quit gammy ("fuck it")
13:48:06 Join AndyI [0] (i=AndyI@
13:49:45pamauryTorne: do you what HAVE_REMOTE_LCD stands for ? In logf there is a strange displayremote which looks similar to logfdisp but called every time logf is called
13:49:59Torneit means there is a remote with an LCD on it
13:50:07Tornelogf prints to this LCD if it exists.
13:58:54 Part pike
13:59:56 Quit AndyIL (Read error: 110 (Connection timed out))
14:02:27n1slinuxstb: what would be the practical benefit of ext2?
14:04:48 Quit ehntoo (Read error: 110 (Connection timed out))
14:12:35 Part pyro_maniac ("Leaving.")
14:15:09 Join fml [0] (n=4fd3c9b1@gateway/web/cgi-irc/
14:15:50 Join funman [0] (n=fun@rockbox/developer/funman)
14:17:25fmlHello. I have a weird problem with the H120 simulator. After setting all settings to their default values, if I clear the backdrop image, the screen of the main unit is drawed black. Only the icons are visible. Does anybody see this too?
14:18:28fmlThe remote is drawed correctly. I use it to again go to the settings and reset them (so that the backdrop image is set again) −− that "repairs" the main screen.
14:22:07 Join kugel [0] (n=kugel@rockbox/developer/kugel)
14:23:29 Join DarkSpectrum [0] (
14:29:46 Quit linuxstb (Read error: 60 (Operation timed out))
14:38:28 Join lasser [0] (
14:39:45 Join LambdaCalculus37 [0] (i=44a0430d@rockbox/staff/LambdaCalculus37)
14:40:12 Quit lasser (Client Quit)
14:42:17 Quit DarkSpectrum- (Read error: 110 (Connection timed out))
14:50:07 Join linuxstb [0] (n=linuxstb@rockbox/developer/linuxstb)
14:51:02 Quit n1s ("Lmnar")
15:02:18fmlkugel: Ok, I commented in that task.
15:04:17CIA-6New commit by 03kugel (r22264): Fix FS #10505 - "Background changes to inverted when cleared" as well as a problem that lets the viewport parser reject correct WPSes, both introduced ...
15:06:05 Join stoffel [0] (
15:08:42 Quit antil33t (Read error: 104 (Connection reset by peer))
15:08:56 Join antil33t [0] (n=Mudkips@
15:10:05CIA-6New commit by 03gevaerts (r22265): blacklist jdgordon-qnap. It misses "nice"
15:13:29***Saving seen data "./dancer.seen"
15:14:52 Join fdinel [0] (
15:14:53kugelfml: the viewport_load_config is probably a left over from before the viewport'ified lists were committed (the patch contained costumizability)
15:15:23kugelNULL instead of VP_ERROR would work too
15:15:28fmlkugel: I wonder why we don't get compiler warnings on this
15:15:48kugelit's just a declaration, that never throws warnings
15:16:18fmlkugel: I'd say it the other way around: VP_ERROR works too (by a pure chance)
15:16:47kugelfeel free to change it
15:17:20kugelmy customlist patch uses VP_ERROR for other functions too, that's why it's there.
15:17:35kugelbut I don't really care, NULL is technically more correct in this case
15:18:10 Join evilnick [0] (i=0c140464@gateway/web/freenode/x-27321350bd977513)
15:19:05fmlkugel: as of now, I'd say it's absolutely the same _technically_. But not semantically :-) Try to change the value of VP_ERROR and you'll see :-) I'll change soon.
15:20:15kugelfml: try to change NULL and you'll get the same :)
15:32:18 Quit AB3JU (
15:32:56NJoinAB3JU [0] (
15:37:11CIA-6New commit by 03alle (r22266): Remove unneeded symbols and improve the comment to the VP parsing function
15:44:54CIA-6New commit by 03alle (r22267): Change the function name in strnatsort so that the code doesn't contradict itself
15:49:32CIA-6New commit by 03alle (r22268): Remove dead code
15:51:03kugelmaybe I shouldn't have used #if 0, but #ifndef ROCKBOX ...
15:54:51 Quit funman ("free(random());")
15:58:01fmlkugel: I thought about defining or not defining IGNORE_LEADING_SPACES but then decided against it.
16:00:15 Part LinusN
16:19:14 Quit linuxstb (Read error: 110 (Connection timed out))
16:31:11 Quit DarkSpectrum (Read error: 54 (Connection reset by peer))
16:31:33 Join DarkSpectrum [0] (
16:32:24 Quit evilnick ("Page closed")
16:38:29CIA-6New commit by 03kugel (r22269): Make kbd_input() show a cancel splash to indicate user abort better and for better consistency all over the place. Change checking for its return ...
16:47:16 Join gtkspert [0] (
16:50:34kugelWhy doesn't the "may be used initialized" warning come up in the sims?
16:50:46*fml spots a typo in rockboymenu.c (wrong result checking)
16:51:01fmlrockboy/menu.c ^^
16:51:15CIA-6New commit by 03kugel (r22270): Both of this isn't needed anymore as it's done at the end of the function.
16:51:19CIA-6New commit by 03kugel (r22271): Fix yellows.
16:53:34CIA-6New commit by 03zagor (r22272): Jdgordon got nice
16:53:50 Join toffe82 [0] (n=chatzill@
16:54:46kugelfml: indeed
16:55:12JdGordonZagor: I wont be reneabled untill the next commit?
16:55:35JdGordonoh, its in the middle of a round
16:55:59kugelyou should be now, there's another round :)
16:56:34Zagorthe block file is only loaded every 10 minutes
16:57:16 Join jgarvey [0] (
17:00:19CIA-6New commit by 03kugel (r22273): Fix mistake at checking the return in rockboy. Thanks Al Le for spotting.
17:01:33kugelI guess I was too early then
17:01:48 Quit gtkspert_ (Read error: 101 (Network is unreachable))
17:03:50 Quit TruthTaco (Read error: 60 (Operation timed out))
17:10:28 Join TruthTaco [0] (
17:12:36 Quit Zagor ("Don't panic")
17:13:33***Saving seen data "./dancer.seen"
17:20:46 Quit bmbl ("Bye!")
17:21:12gevaertsZagor: (logs)
17:22:05 Join Blue_Dude [0] (
17:25:47 Quit fml ("CGI:IRC (EOF)")
17:26:06 Join bmbl [0] (n=Miranda@unaffiliated/bmbl)
17:34:45CIA-6New commit by 03kugel (r22274): Protect viewport.h against multiple inclusion.
17:39:06 Join Strife89 [0] (
17:41:29 Join evilnick [0] (i=0c140464@gateway/web/freenode/x-6dda82f8cf9334a2)
17:41:56 Join funman [0] (n=fun@rockbox/developer/funman)
17:46:19funmanFlynDice: i'll check if AMS OF uses widebus mode for SD cards
17:47:01funmanobo: how's going view's work?
17:50:15FlynDicefunman: I can get widebus to work for browsing files and loading album art but playback will not work for more than 3-4 secs, eerily similar to the spot we were in with the mmu before you found the CCU setting...
17:50:43funmanwhat happens after those 3-4 seconds?
17:51:12FlynDiceplayer is functional besides playback it seems
17:51:58funmananything useful in buffering thread debug menu perhaps?
17:51:59FlynDicethe cpu frq is locked at 248 and the lcd disk light is on steady
17:52:50funmani don't see where i could have left a deadlock which wouldn't trigger a panic
17:53:04FlynDiceyes cpu is steady boosted and I2SO is off
17:53:08funmani need to look at that file carefully again : i want to add retry on send_cmd errors
17:53:28funmanwhat's the state of bufferign thread? empty?
17:53:56FlynDicewe seem to get an ok transfer speed without widebus though
17:54:34funmanwe hadn't set up the controller for wide bus so that looks normal
17:54:53funmani mean, i can understand we see the same performance, not explain it :)
17:54:55kugelfunman: does FlynDice fix your non-working uSD?
17:55:26funmankugel: no, as written in the flyspray entry
17:55:56funmani listened to music stored on this µSD this afternoon while skating without any crashes though
17:56:14kugelfunman: yea, but your first comment was "Thanks for the patch, it works fine with my 2 µSD cards", hence I was confused
17:56:38funmanthe 2nd comment starts with "correction"
17:57:19funmani have no way to reproduce an error in data transfer from/to this card
17:57:44funmani formatted it since, i'll try again test_disk on it in the future
18:04:52CIA-6New commit by 03blue_dude (r22275): New committer!
18:05:12kugelBlue_Dude: Welcome! \o/
18:05:20kugelFlynDice: did you try test disk?
18:06:09FlynDicekugel: not yet I've been trying to get widebus working
18:07:35pamauryfunman: did you look at the logf fix I posted on FS recently (FS #10515) ?
18:09:12 Quit petur ("work->home")
18:10:09CIA-6New commit by 03funman (r22276): Fix a wrong memcpy in logf() introduced in r22253 ...
18:10:23*LambdaCalculus37 sees a new committer :)
18:10:36funmanpamaury: no i hadn't looked at it :)
18:10:44pamaurythanks :)
18:12:01pamauryI have a question about usb_drv_* things ? Is there any usb expert here ?
18:12:04FlynDicekugel: One other thought I had re widebus. Do you think the GPIOD being dual use with MMC/SD could be an issue? I'm not all that familiar and you worked with that for the buttonlight on disk access issue
18:12:41kugelcould be, build without HAVE_BUTTON_LIGHT to be sure
18:14:23kugelfunman: you could probably deactivate it also when messing with the problematic microsd, I don't really knno how this buttonlight business affects our driver
18:15:08funmankugel: did you see how the OF deactivate button light while SD transfers?
18:15:45kugelI haven't really looked at the of, except for the SD_MCI_POWER thing
18:15:51gevaertsBlue_Dude: congratulations!
18:16:08*gevaerts sees pamaury's question and tries to hide
18:17:04pamaurygevaerts: ah I though you weren't there :)
18:17:05kugelpamaury: gevaerts! gevaerts! gevaerts!
18:17:25gevaertsI'm not! I don't exist!
18:17:49CIA-6New commit by 03gevaerts (r22277): block qnap-jdgordon again: it doesn't have perl in the expected place
18:18:03pamauryI was wondering: what happen if someone call usb_drv_send_nonblocking while another call has been issued and the transaction is not finished
18:18:18gevaertson the same endpoint? Don't do that
18:18:28funmangevaerts: what about a simple script which checks basic requirements? nice, perl in some place, etc?
18:18:39kugeldid JdGordon actually any successful build before attaching as client?
18:18:39pamauryyes on the same endpoint
18:19:09pamaurywell I was sure I would get that answer... That's related to the usbserial driver in fact
18:19:17gevaertspamaury: I don't think any of the drivers will handle that properly. ARC certainly doesn't.
18:19:53pamaurygevaerts: so weird things could arise if someone calls usbserial_send two times on a row
18:19:58kugelthey aren't thread safe?
18:20:07gevaertskugel: nothing to do with threads
18:20:14pamaurythe same thread
18:20:45pamauryexample: usbserial_send('a',1);usbserial_send('b',1); The first transaction won't be finished when the second will happened
18:20:58*kugel read "usb_drv_send_nonblocking while another call has been issued and the transaction is not finished" as "call it, while a previous call didn't return"
18:21:19gevaertspamaury: usb_serial_send() just puts things in a ringbuffer
18:21:23gevaertskugel: *nonblocking
18:22:28pamauryhum, so there shouldn't be any problem calling it lots of time on a row ?
18:23:18gevaertspamaury: there shouldn't be, no. You can overflow the ringbuffer of course, but that should only eat data, not crash. That doesn't mean it's guaranteed bug-free of course
18:23:26 Quit jordan`` (Read error: 110 (Connection timed out))
18:24:22pamauryok I will test more, because I'm working on a new implementation of logf and for practical reasons it's simpler to call usbserial_send for each character rather than one or two times at the end
18:24:33pamauryAnd I get really weird results with that
18:24:45 Join JdGordon| [0] (n=Miranda@nat/microsoft/x-e6e62c1c0350c22e)
18:25:25gevaertspamaury: please try to avoid that. It should work well, but I don't really like the idea
18:25:35pamaurywhy ?
18:25:47pamaurybecause it's subsubsubsuboptimal ?
18:25:55gevaertsit could be I think, yes
18:26:53pamauryok,I will avoid that but still I'm wondering why I get different results
18:26:53gevaertsalthough that will depend on the case. If you log a line every now and then, I expect two transactions instead of one. If you log lots in a row, it won't be too bad, but you'll overflow the ringbuffer anyway
18:27:22pamaurywhat's the size of the ringbuffer ?
18:27:40gevaertsdefault 512 bytes
18:27:49gevaertsyou can change the define though
18:28:05pamauryshould be large enough normally
18:28:23gevaertsI expect so, yes
18:29:34gevaertsif we want to use it for other things than logging (ppp?), I guess just usb_serial_send() isn't a good API. I'm not sure how likely that is to happen
18:31:11pamaurythat's for logging
18:32:39gevaertsI know. I can just imagine some usecases actually wanting to know when their actual bytes go out
18:35:14 Join _Shaid [0] (
18:36:01pamauryyes that's true
18:36:58pamauryAlso, is the usbserial implementation really working ? I have lots of problem with my sansa (or perhaps my logf is buggy related to usbserial)
18:37:22gevaertsit's not always very reliable. I haven't used it recently
18:38:07pamaurywhy ? I is a usb problem or really the usbserial ?
18:38:28gevaertsnot sure
18:38:53bertriklast time I looked at usbserial for another project (this is a few years ago), there were also some peculiarities in the windows xp drivers
18:38:57 Quit Shaid (Read error: 104 (Connection reset by peer))
18:38:57 Nick _Shaid is now known as Shaid (
18:39:07CIA-6New commit by 03gevaerts (r22278): JdGordon promises that qnap-jdgordon now works fine
18:39:33*JdGordon| promises nothing
18:39:43pamauryI guess for Windows, a simple interface like this won't suffice, it problem requires the whole CDC-ACM things
18:39:44gevaertsyes, our current implementation does not work with xp, due to it being pretty incomplete
18:39:54funmanJdGordon| doesn't, but JdGordon does!
18:40:15*JdGordon| wouldnt trust that guy....
18:41:02bertrikgevaerts, what is missing then?
18:41:49bertrikIIRC, windows just sends some of those line control commands (DTR) when opening a USB serial port and not much more
18:41:49gevaertsbertrik: I'm not exactly sure, but IIRC there needs to be an interrupt endpoint that does some things, and some control things
18:42:16gevaertsit may not use the other things, but it could very well check if the endpoint is there
18:42:27JdGordon|scorche: all our flights suck... assuming the 3 of us goto the airport together you will be stuffing around for 2 hours and me 1... +any extra time for gev to make his flight
18:42:41JdGordon|WTF? gev<tab> in that sentance chanegd it to 24/7!
18:42:53*gevaerts denies being 24/7
18:43:00JdGordon|sorry, wrong window
18:49:57 Join mt [0] (n=MTee@rockbox/developer/mt)
18:49:58 Quit funman ("Lost terminal")
18:50:25 Join funman [0] (n=fun@rockbox/developer/funman)
18:54:55kugelwhere is gui_statusbar_set_screen() defined? it's called but I can't find it
18:55:16JdGordon|I have a feeling thats #defined in screen_access.h
18:55:30JdGordon|or used to be
18:56:07kugelah got it
18:57:09*kugel wonders whether "/* Must be done before any code uses the multi-screen APi */" before gui_syncstatusbar_init() is still valid
18:57:30JdGordon|probbaly not
18:59:26*kugel tries
18:59:30pamaurygevaerts: there is a really strange thing with usbserial: if send a burst of usbserial_send (each of size >= 16 in this case) then only the first seems to be catched by usbserial on linux and in fact on the first seems to be actually sent.
18:59:53pamauryBut if I introduce a sleep(HZ/10) between each call, then everything works !
19:00:24gevaertspamaury: that seems to point to the ringbuffer not working properly
19:00:53pamauryok, I'll investigate that
19:03:57pamaurygevaerts: I also have a question about usbserial: what does this line of code means (in subserial_init_connection) ?
19:04:02pamaury usb_drv_recv(ep_out, receive_buffer, sizeof receive_buffer);
19:04:38pamaurythere also is a comment:
19:04:42gevaertspamaury: tell the driver to receive some data
19:04:50pamauryprime rx endpoint
19:04:55pamaurywhat does that mean ?
19:05:32kugelstupid setttings_apply() at boot
19:05:34gevaertshm, that comment should probably be changed. That's a bit specific to the ARC driver. "priming an endpoint" means telling the hardware to expect data
19:06:32pamauryWithout this it would accept data from the host ?
19:07:28pamauryBut the drivers shouldn't be aware of that no ?
19:07:56gevaertswhy not? The driver decides where to put the data...
19:08:10pamaurywell, that's true :)
19:09:02 Join captainkwel [0] (i=2669ecc2@gateway/web/freenode/x-aceee27dadef84bc)
19:11:14 Join sajes [0] (n=sajes@
19:11:40sajesOi. Just rockboxed my samsa. :D.
19:13:37***Saving seen data "./dancer.seen"
19:17:07 Join AsaelReiter [0] (n=d44c6c11@gateway/web/cgi-irc/
19:20:28FlynDicefunman: re: [08:46] <funman> FlynDice: i'll check if AMS OF uses widebus mode for SD cards, I would be interested if you find a SET_CLR_CARD_DETECT(ACMD42) associated with this also. It controls a pull-up resistor on DAT3.
19:21:41funmanFlynDice: i have some troubles to start disassembling (can't find a correct blue color for my terminal)
19:22:22 Join merbanan [0] (
19:22:37 Join domonoky [0] (n=Domonoky@rockbox/developer/domonoky)
19:25:52pamaurygevaerts: is there any way to sniff the usb packets on linux ?
19:30:00gevaertsthere's wireshark+usbmon these days
19:31:19 Join Horscht [0] (n=Horscht2@xbmc/user/horscht)
19:32:25 Join n1s [0] (n=n1s@rockbox/developer/n1s)
19:40:09 Quit kugel (Read error: 110 (Connection timed out))
19:40:33 Join kugel [0] (n=kugel@rockbox/developer/kugel)
19:42:08 Join StealthyXIIGer [0] (
19:48:54Blue_DudeI'm going to commit a couple of plugin changes: FS #10502 and FS #10504. I won't be able to close the FS items yet though.
19:49:08LambdaCalculus37Blue_Dude: What are those plugin changes, exactly?
19:49:58Blue_Dude10502 adds DSP benchmarking to test_codec. 10504 makes use of dsp_output_count and dsp_input_count in the mpegplayer audio thread.
19:50:35saratogaFS #10502 could probably be closed
19:51:33Blue_DudeYeah. I'd like to commit the change first though.
20:00:11 Join Strife1989 [0] (
20:00:17 Quit Strife89 (Nick collision from services.)
20:00:31 Nick Strife1989 is now known as Strife89 (
20:01:46 Quit pamaury ("Quitte")
20:01:59 Join pamaury [0] (
20:05:29Blue_DudeHm. Having trouble using patch at the command line. Says can't find files and suggests using −−strip.
20:06:02saratogause -p0 or maybe higher int
20:06:05funmanBlue_Dude: perhaps you need to tweak -p level
20:06:07saratogaif it was made with git
20:06:21funman-pN will strip the N first leading directories from the patch
20:06:56Blue_DudeI used svn diff to generate the patch. Now I need to apply it to another branch.
20:07:33funmanyou would use patch -p0, provided you are in the root of rockbox checkout and generated the diff from the root of a rockbox checkout
20:07:39saratogatry incrementing the p level until it takes
20:07:48 Quit alexbobp (Connection timed out)
20:07:56pamaurygevaerts: does it require a custom kernel build or any kernel will do ?
20:08:17Blue_Dudeok, it liked -p0
20:08:33saratogamost SVN diffs will need p0
20:08:38gevaertspamaury: depends. Some distributions have usbmon by default, but presumably some don't
20:08:41 Join jordan` [0] (i=gromit@
20:08:47funmanBlue_Dude: when looking at the patch, pay attention to the line "diff −−git a/firmware/logf.c b/firmware/logf.c
20:08:59*pamaury believes ubuntu don't :(
20:09:06bertrikin short, use -p1 for patches made by funman :)
20:09:10funman"patch" will attempt to patch "b/firmware/logf.c" and "patch -p1" will patch firmware/logf.c
20:09:29funmanthe patchlevel could be different if you ran svn diff while working inside firmware/ directory
20:09:41funmanwell, you'd have to cd firmware/ first .
20:09:48gevaertspamaury: "grep USB_MON /boot/config-*" will tell you
20:09:52Blue_DudeThe weird thing is that I ran the patch from the root and tried to apply it at the root.
20:10:05Blue_Dude−−ran the diff −−
20:10:26pamaurygevarts: seems to support it: CONFIG_USB_MON=y
20:10:45pamaurygevaerts: but I can't modprobe usbmon
20:11:04gevaertspamaury: "y" means it's built-in, no modprobe needed
20:11:07 Join stephen__ [0] (
20:11:14 Quit AsaelReiter ("CGI:IRC (EOF)")
20:11:21pamaurygevaerts: ah :)
20:11:53pamaurybut when I run wireshark it doesn't show any usb interface
20:11:55funmanBlue_Dude: patch manpage says if you don't specify -p it will strip ALL leading directories
20:12:28CIA-6New commit by 03blue_dude (r22279): Adds DSP testing and WAV writing to test_codec
20:12:34gevaertspamaury: as root?
20:12:52pamaurygevaerts: ah that could explain the thing
20:13:22kugelBlue_Dude: please try to reference to a FS# if there's any related one
20:13:39 Quit StealthyXIIGer (Connection timed out)
20:13:51Blue_Dudekugel: Yeah you're right. Forgot. That was FS #10502.
20:13:53pamaurygevaerts: I don't have any usb interface, I have ethernet interfaces only
20:14:07funmanBlue_Dude: an advantage of doing that, is that direct links to the FS are generated in the irclogs
20:14:53gevaertspamaury: maybe you need to mount the debugs thing? I don't know...
20:15:08*gevaerts hasn't used wireshark for usb in a very long time
20:15:13kugelfunman: and on the front page :)
20:16:03pamaurygevaerts: I have found out, it requires another obscure mount command line with usbfs
20:17:19Blue_Dudefunman: Happily climbing the learning curve...
20:17:42Blue_DudeFS# 10502 is OK to close now, too.
20:18:57 Join alexbobp [0] (
20:19:46kugelBlue_Dude: your wish is my command :>
20:19:58 Quit einhirn ("Miranda IM! Smaller, Faster, Easier.")
20:21:23mtBlue_Dude: A bit late, but congratulations ! :)
20:22:58 Quit saratoga ("Page closed")
20:23:59 Quit funman ("free(random());")
20:25:29 Quit Strife89 ("Bad weather, better unplug stuff.")
20:28:34CIA-6New commit by 03blue_dude (r22280): FS #10504: Make mpegplayer audio thread use correct sample count
20:29:22 Join Thundercloud [0] (
20:29:54pamaurygevaerts: is there a way to workaround the capture size which is quite limited... ?
20:30:12gevaertsI don't know. I haven't ever used it for real
20:30:30gevaertsI've just played around with it a bit
20:36:30 Quit timc (Remote closed the connection)
20:38:42n1sBlue_Dude: congratz. and ask Zagor for dev access on FS
20:39:14Blue_Dudenls: thanks. I'm keeping an eye out for him. Will probably just shoot him an email.
20:42:17CIA-6New commit by 03blue_dude (r22281): FS #10512: Bookmarking does not behave correctly (fixes r22192)
20:45:23*moos joins the congratulations and add bravo for the green ;)
20:45:40 Quit n1s ("Lmnar")
20:46:40bertrikI have a question about list.c
20:47:01bertrikWe have three of those, a common list.c, a list.c for charcell and a list.c for bitmap targets, right?
20:47:53bertrikthe common list.c now seems to contain some list functions that are completely enclosed in #ifdef HAVE_LCD_BITMAP, can't we move those to the list.c for bitmap targets?
20:50:24bertrikfor example function gui_list_get_item_offset is implemented in the common list.c but is never used there, it's only used in bitmap/list.c
20:51:17gevaertssounds like a clear thing to move
20:52:42bertrikok, I'll carefully go ahead with that
20:53:16 Quit flydutch ("/* empty */")
20:55:23bertrikoh, that would be too easy of course, it refers to some local variables
20:55:45gevaertsof course :)
20:57:21jordan`oops sorry
21:00:00kugelbertrik: the other bitmap is only for drawing actually
21:00:22kugelthe other list.c*
21:00:34CIA-6New commit by 03blue_dude (r22282): FS #10446: Bug defense in dsp.c, minor tweaks and comments
21:01:39Blue_DudeAnd that clears my backlog... Whew.
21:02:31kugelbertrik: I'd rather rename instead of breaking this separation
21:02:47bertrikI don't get it
21:03:09kugelthe list.c in the subfolders do the drawing, the normal list.c is the logic code
21:03:14bertrikI'd just like to move some of the bitmap specific code from apps/gui/list.c to apps/gui/bitmap/list.c
21:04:00bertrikthe charcell/list.c only does list_draw indeed
21:04:09kugelyes, but that would mix logic and drawing code, I wouldn't want that, hence renaming the list.c's in the subfolders
21:04:14bertrikbut the bitmap/list.c does a lot more
21:05:43 Quit fdinel (Read error: 104 (Connection reset by peer))
21:06:09kugelwell, the touchscreen stuff is a bit more it seems
21:07:29bertrikwhen something is only used by the list_draw function from apps/gui/bitmap/list.c, wouldn't it make sense to actually move it to apps/gui/bitmap/list.c ?
21:07:50bertrikhow about I make a patch so you can have a look at it?
21:08:32kugelwhat is only used by list_draw?
21:10:33bertrikFunction gui_list_get_item_offset. Functions gui_synclist_scroll_right/left are referred from apps/gui/list.c but only #ifdef HAVE_LCD_BITMAP
21:12:43kugelthe former seems fine to move, the other two not really
21:12:54 Join n1s [0] (n=n1s@rockbox/developer/n1s)
21:13:38***Saving seen data "./dancer.seen"
21:14:10 Join bluebrother [0] (n=dom@rockbox/developer/bluebrother)
21:14:49CIA-6New commit by 03rob (r22283): D2: Re-enable SD(HC) driver as there have been no further reports of damaged cards.
21:15:34kugelmaybe a subfolder list/ with list-engine.c, list-draw-[bitmap|charcell].c, simplelist.c would be a good idea
21:17:55bertrikheh, it started as a simple fix (just remove some extern declarations from a C file) and before you know I'm creating entire new source files ... :P
21:21:47 Join fdinel [0] (
21:26:08CIA-6New commit by 03rob (r22284): TCC: Implement ECC error correction for sectors read from NAND. Tested on D2 (78x, MLC) and M200 (77x, SLC).
21:33:04 Quit efyx (Remote closed the connection)
21:34:54pamaurygevaerts: I have some new info about the usbserial problem
21:34:56JdGordon|kugel: FOLDERS FOR EVERYONE.... ahahhehhehehee
21:35:00 Quit amiconn (Nick collision from services.)
21:35:02 Join amiconn_ [0] (i=quassel@rockbox/developer/amiconn)
21:35:13 Quit pixelma (Nick collision from services.)
21:35:14 Join pixelma_ [0] (i=quassel@rockbox/staff/pixelma)
21:35:22 Nick amiconn_ is now known as amiconn (i=quassel@rockbox/developer/amiconn)
21:35:31 Nick pixelma_ is now known as pixelma (i=quassel@rockbox/staff/pixelma)
21:35:32bluebrothera folder for three separate files? Sounds like overkill to me
21:36:00kugeloverkill? does a folder cost anything? 4 files, btw
21:36:23bluebrotherit doesn't cost anything, but it does cost time to navigate.
21:36:35bluebrotherso yes, it does cost something.
21:36:49kugelalso, right now in apps/gui/ there are 2 folders with 1 file each
21:37:16kugelbluebrother: sorry, that sounds absurt to me
21:38:00bluebrotherdoesn't sound like a good thing to me either. Either other files in those folders have been deleted, or files that are planned to be put there haven't been put there yet. Or someone thinking that a deep deep folder structure makes it easier to find things
21:38:13gevaertspamaury: what sort of info?
21:38:41 Join robin0800 [0] (
21:38:48kugelbluebrother: no, the guy who just cried "FOLDERS FOR EVERYONE" added those :)
21:38:54bluebrotherkugel: feel free to disagree with me, but splitting things up too much can be a slowdown as much as not splitting things up when necessary.
21:38:54 Join Strife89 [0] (
21:39:02bluebrotherthen blame that guy :)
21:39:38kugelI disagree in this case at least
21:39:42pamaurygevaerts: On my computer with my target, with my program to read usbserial packets, then usbserial seems not to send any usb packet when the size becomes too importants (in my exemple 142 characters)
21:40:08bluebrotherthing is, the question of splitting things in a folder or not is a tightrope walk. One needs to consider carefully if the added folder is a benefit or does make things more dungeon-like.
21:40:42kugelI think there are things you need to consider more carefully
21:40:59pamaurygevearts: I've tried lots of combinaisons and reached the conclusion that its fails only when the size left to transfer in the ringbuffer is too big (but lower than the ringbuffer size still)
21:41:14kugelwell structured code is important imo, even if it means to browser 1 additional subfolder at times
21:41:18gevaertspamaury: any clue whether it's usb_serial.c or the driver?
21:41:18pamauryI will try to determine the exact value
21:41:55pamaurygevearts: to me, it's the driver because I've put logf things everywhere and usbserial forwards directly the request and the buffer to the driver
21:41:56bluebrotherkugel: did I say *anything* about the importance of other things?
21:42:11bluebrotheror the relation to the importance of other decisions?
21:42:19*bluebrother shakes head
21:42:21kugelNo you didn't
21:42:34bluebrotherso what's the point in that?
21:42:40pamaurybut I don't understand why it would fail with usbserial whereas mass-storage works and involves transfers greater than that I beleive
21:42:57kugelbluebrother: you're leaving the topic
21:43:28gevaertspamaury: one special thing about storage is that it only sends either small packets (commands), or full 512-byte packets. Never anything in between
21:43:33bluebrotheryou consider other things more important? Fine, but the decision under discussion is still a tightrope walk. Doesn't change simply because you consider other things more important.
21:43:40bluebrotherkugel: no I don't.
21:44:59pamaurygevearts: that's strange then. I've though about a weird bug being a combinaison of rockbox+kernel+libusb but it also fails with linux usbserial module iirc
21:45:06kugeltightrope walk is excessivly exaggerated
21:45:25pamaurygevearts: the strangest thing is that *it doesn't send any packet*
21:46:52gevaertspamaury: did you verify that it properly gets to sendout()?
21:47:02bluebrotherwell, then consider it exaggerated. Still it's a decision that isn't easy in quite some cases. So why not call it tightrope walk?
21:47:17pamauryyes: it sendout the same buffer and buffer size. I will investigate more about this issue
21:47:44kugeloh dear...
21:48:04 Join Strangelove [0] (n=4fb52117@gateway/web/cgi-irc/
21:48:12bluebrotherI've seen enough cases where deep folder structures holding only a little amount of files got confusing, even to the people who created it. So at least *I* consider it an important question. If you don't, fine.
21:48:52kugel3 isn't deep, and not even deeper as right now
21:49:44Strangelovei have a question.. how do u sync files into the iPod when using Rockbox? i've went through the FAQ and
21:49:57Strangelovemanual and didn't see anything regarding that...
21:50:00bluebrotherkugel: this doesn't change my opinion that folder structure decisions aren't easy. It might be unproblematic in this case, but I wasn't referring to this case anymore. What on earth is the point?
21:50:07bluebrotherStrangelove: any way you like.
21:50:31bluebrotheryou can use Windows Explorer (in case you're using Windows), iTunes, or any other software that syncs to a player.
21:50:34kugelStrangelove: you can drag&drop or use any syncing tool (in disk mode I think)
21:50:48Strangelovei HATE itunes..
21:50:48bluebrotherRockbox does not need any special folder structure or database file so it's completely up to you.
21:50:49kugelor rockbox usb that is
21:51:13Strangelovei had a Zen vision:M that worked great and was soooo easy to organize
21:51:35Strangeloveand now i got an iPod Classic 120gb and it only uses that horrible itunes software..
21:53:11Strangelovethat leads to just one more question.. i didn't understand with all the generations talks, whetever Rockbox supports the new iPod Classic 120gb?
21:54:08stephen__no it doesn't support the classic yet
21:54:32pamaurygevearts: seems to fail when size>=97
21:54:57Strangelove:-( well i have patience, i'll wait for it, it looks like it's worth the wait..
21:55:08 Join aaron424 [0] (
21:55:11Strangelovethanks allot guys
21:55:59Strangeloveta ta
21:56:01 Quit Strangelove ("CGI:IRC")
21:56:02pamaurygevearts: here is what wireshark say: at size=96 works like a charm but at size=97: cannot send after transport endpoint shutdown
21:56:59 Quit fdinel ("Miranda IM! Smaller, Faster, Easier.")
21:57:20 Quit robin0800 (Remote closed the connection)
21:58:27 Quit LambdaCalculus37 ()
21:58:48gevaertspamaury: weird... Anyway, that should mean that if sendout() adds "buffer_transitlength = MIN(buffer_transitlength,96);
21:59:04 Join heavysidecar [0] (n=heavysid@
21:59:07gevaerts" it should work. Maybe that's worth a try?
21:59:39gevaertsIt's not a real solution of course, but it's certainly a useful indication
22:01:35pamauryI'll do just one more test to be sure it's 96
22:03:23bertrikbluebrother, the intention of the list subdirectory was to make file naming easier and less confusing
22:03:53bertrikcurrently we have already have two subdirs containing only one list.c file each
22:05:09 Join stripwax [0] (
22:06:45 Join robin0800 [0] (
22:06:45pamaurygevearts: here's my logf:
22:07:05pamaurygevearts: variables refers to the ones in usbserial
22:07:16kugelJdGordon|: customlist looks good imo
22:08:21 Join kachna [0] (
22:08:25JdGordon|good... how?
22:08:59kugelcommittable very soon
22:10:53gevaertspamaury: ok, so it's definitely not related to ringbuffer wraparound or anything like that. Could you try something like
22:12:15bluebrotherbertrik: it might make sense in this case, especially if it consolidates folders. I still stand by my opinion that creating a good folder structure is non-trivial ;-)
22:13:51 Quit stephen__ ("Leaving")
22:13:51JdGordon|kugel: whats missing?
22:14:06pamaurygevearts: you trick works which is not very surprising
22:14:14kugelJdGordon|: no idea
22:14:36pamaurygeveaerts: I have also tried to do a usb_serial_send(buffer,512); but it doesn't seem to work (will confirm in a minute)
22:14:53 Quit merbanan (Remote closed the connection)
22:15:02 Join Zagor [242] (
22:15:23kugelthere isn't missing much since a long time, I just kept being undecided on the api and stuff
22:15:34JdGordon|which api?
22:15:38gevaertspamaury: not surprising indeed, but still good to know
22:16:01 Join fg56lx [0] (
22:16:14pamaurygevearts: the usb_serial_send with 512 chars also fails, that's really strange. I will check with ubserial module
22:16:24kugelJdGordon|: activating/deactiving for example, or overridding the ui viewport
22:16:34Blue_DudeGot to go, but I just posted an optimized and speeded up version of the limiter DSP process I've been tinkering with at FS #10199. It is as fast as I can make it, and it doesn't take a hit on system performance when it isn't active. I've been running it on target for a few weeks and it *does* work.
22:17:12JdGordon|kugel: we were going to work on that post commit though wernt we?
22:17:29JdGordon|this is the viewport.c mess?
22:17:40kugelbut it's good now I think
22:17:47pamaurygevearts: no it doesn't work. Will would it fail with usbserial and not with mass storage ?
22:17:54kugelor rather, acceptable :>
22:18:07 Quit evilnick ("Page closed")
22:18:19JdGordon|do you wanna put some test builds in the forums?
22:18:45kugeldo you think it's needed?
22:18:47*JdGordon| tries to put some time aside tonight to give the patch a good looking over
22:18:52gevaertspamaury: maybe related to alignment? What happens if you change the aligned(32) to aligned(512) for send_buffer?
22:19:03JdGordon|kugel: no, not really... others might disagree thoguh...
22:20:08amiconnZagor: There is a gap which can't be seen in the SVN backlogs on the website
22:20:32 Join Llorean [0] (
22:21:06Zagoramiconn: 22264-22266?
22:21:19 Join efyx [0] (
22:21:26amiconnToday's log (already the extended one on recent.shtml goes back to r22267, and the next older one ("last 4 week") only goes up to r22263
22:22:13 Quit raphi ("leaving...")
22:22:30 Join aaron424_ [0] (
22:23:18 Part Blue_Dude
22:23:57pamaurygevearts: no it doesn't seem to work
22:24:17pamaurygevearts: is the usb storage buffer 512 bytes aligned ?
22:24:18 Quit aaron424 ("ChatZilla 0.9.85 [Firefox 3.0.13/2009073022]")
22:24:28 Nick aaron424_ is now known as aaron424 (
22:24:38kugelZagor: what I also noticed is that since-12-month doesn't go 12 month back
22:24:52gevaertspamaury: no, also only 32 bytes.
22:25:03*gevaerts misremembered that for a moment
22:25:38kugeloh wait, it does now
22:26:27Zagorthe archive files are not created as often as the front-page list
22:27:03bertrikI got the shuffled list stuff compiling at last, it doesn't make much binsize difference (as expected)
22:30:02gevaertspamaury: I think I'm going to commit the packet size limiting. It doesn't fix the real underlying problem, but at least then usb_serial will actually be usable
22:30:50gevaertsI'll limit it to 32 though. That 96 sounds weird to me, and I suspect that it might be just a random multiple of 32, depending on where exactly in memory the buffer ends up
22:31:24pamauryyes, limit it to 32, we don't understand what this 96 is. Perhaps it's even related to the host
22:31:31 Quit fg56lx (Read error: 104 (Connection reset by peer))
22:31:35 Join thegeek [0] (
22:32:25pamaurygevearts: your previous fix wasn't sufficient, you need two MIN instead of one
22:34:44pamaurygevearts: is the usb_drv_arc doing packet splitting or is this done in hardware or even never done ?
22:35:05CIA-6New commit by 03gevaerts (r22285): Limit usb_serial packets to 32 bytes. It's unclear why this is needed, but usb serial packets larger than 96 bytes seem to never be sent. ...
22:36:05gevaertspamaury: that's done in hardware... Hm, maybe the hardware isn't set up right...
22:36:46gevaertspamaury: is the fix I just committed still wrong?
22:36:48bertrikgevaerts, could it be that the usb serial host driver doesn't expect large packets?
22:36:50pamaurydon't know. I get strange results with wireshark: sometimes it's a "transport endpoint shutdown" and sometimes its "protocal error"
22:36:57pamaurygevearts: no it seems ok
22:39:39gevaertsok. I'll leave it like this for now then
22:40:01gevaertsbertrik: we set up a maxpacketsize of 512. AFAIK the host has to accept that...
22:41:53pamaurynearly any computer host will accept 512 I think. This is perhaps not the case if the host is a portable device
22:42:13gevaertsit has to, or it shouldn't claim to speak usb
22:42:25pamauryit's in the spec ?
22:42:46gevaertsallowed packetsizes are in the spec, yes
22:43:39bertrikwhich host driver is used for the usb serial port?
22:45:44gevaertsI use the generic linux usbserial driver
22:45:58 Join petur [0] (
22:46:37pamauryIt's not even related to usbserial driver. I've tried using a custom program using libusb.
22:46:37bertrikI thought usbserial was for "weird" usb serial devices and cdc-acm for normal cdc compliant devices
22:46:37 Join andrewbeveridge [0] (
22:47:12pamauryyes but cdc-acm requires a cdc-acm compatible device :)
22:47:22gevaertsbertrik: ah, maybe. Well, we don't have all features that a cdc compliant device needs, so that's ok :)
22:48:15andrewbeveridgehi, could anybody explain why the "ACTION_WPS" action codes do not work in any context other than the WPS itself?
22:49:48pixelmabecause they were made to work in the wps context only? If you combine contexts you can easily get "overlapping" and unwanted effects
22:50:40n1syeah, it's, like the whole point of the context sensitive actions :)
22:50:48andrewbeveridgeso for example, a player with a dedicated "play/pause" button. if one is not on the wps, the button is useless?
22:51:17n1sthe button can hava a different action in another context
22:51:39pixelmae.g. resume
22:51:44andrewbeveridgeso there is no way to control music playback while on another screen?
22:52:27 Nick andrewbeveridge is now known as andrewRB (
22:53:02n1sandrewbeveridge: depending on the target... some targets can use their stop button and play/pause/resume in some other screens than wps
22:53:07pixelmasome players that have dedicated volume buttons (like the Gigabeat F and the sansa c200) give you the possibility to control the volume while in lists (menu, browser)
22:54:06 Quit Zagor ("Clint excited")
22:54:07n1sbut this has nothing to do with the actions, WPS* actions are for the wps, an action performed in a different context while it might do the same thing is not a wps action
22:54:41pixelmathat's true
22:54:56kugelman, something is wrong
22:55:14andrewRBok, well the target I am working on is the Cowon D2, so could you explain what I should be looking at if not the target specific keymap please?
22:55:14kugelthe whole recompiles from time to time :(
22:55:53 Quit amiconn (Remote closed the connection)
22:55:54 Quit pixelma (Remote closed the connection)
22:56:01andrewRB(it has dedicated volume buttons, that is exactly what I would like to achieve)
22:56:29 Join AsaelReiter [0] (n=d44c6c11@gateway/web/cgi-irc/
22:57:42CIA-6New commit by 03gevaerts (r22286): fix typo
22:57:50andrewRBah, i see. for example the c200 - i see "ACTION_LIST_VOLUP"
22:57:51pamaurygevearts: did someone manage to raise a gcc internal error ?
22:58:04gevaertspamaury: I suspect things like bad RAM
22:58:28n1sandrewRB: there are other targets that have dedicated volume buttons too, like the gigabeats for example, they use them for ACTION_LIST_VOLUP (and down) in the list
22:58:37gevaertsif the same build client does that a few times more, I guess it's blacklist+talk to owner time
22:58:38*n1s slow
22:58:50AsaelReiterHow can I check the CPU usage of a plugin?
22:59:15pamaurygevearts: would it be possible that you or someone else try to see if this 96 limit for usbserial maintains for other target and with other hosts ?
22:59:29gevaertspamaury: I can, but not today
22:59:53pamaurygevearts: Ok, I will try myself on a another computer
23:04:04 Part heavysidecar ("Ex-Chat")
23:04:08andrewRBn1s: thanks, I now see what I want in the gigabeat-s keymap. Still not sure which file I should be looking at though =/
23:04:26 Part Llorean
23:06:07 Join pixelma [0] (i=quassel@rockbox/staff/pixelma)
23:06:11 Join amiconn [0] (i=quassel@rockbox/developer/amiconn)
23:06:12JdGordon|andrewRB: what are you tring to do?
23:06:34pamaurygevearts: I have the same problem on another computer
23:07:22andrewRBthe D2 has three hardware buttons; volume up/down, and play/pause. I want these to control music playback, whatever screen I am on, since the touch screen is used for navigation.
23:08:08JdGordon|you need to add some new actions, and then in every screen you want them to work you need to add handlign for thme
23:08:15andrewRBI thought I could simply edit "keymap-cowond2.c" but I was wrong
23:08:38JdGordon|the other option is add the handling to default_event_handler(), but thats not very nice
23:09:26kugelomg again!
23:09:59andrewRBmaybe a better question for me to ask was, is there anything I can read to help me learn my way around the rockbox source? Simply reading the code, I'm going round in circles.
23:10:11JdGordon|there isnt...
23:11:32n1sandrewRB: the only easy place to add these actions is in the list context as other targets already have that functionality, other contexts will be a little more work
23:12:02andrewRBok, thanks.
23:12:13 Join ehntoo [0] (
23:12:18 Quit stoffel (Remote closed the connection)
23:13:40***Saving seen data "./dancer.seen"
23:13:52*scorche prods petur and 2 other missing people into -gsoc
23:17:11andrewRBwould there be any reason for me not to add "ACTION_STD_VOLUP"?
23:18:07JdGordon|well... ideally every ACTION_STD_ must be handled in every action handler... and that each target should have it
23:18:14JdGordon|putting it in the list context makes more sense
23:18:27pamaurygevearts: Are there any specifications available for the usb drivers ? (more precisely for sansa e200 −−> usb_drv_arc ?)
23:19:05andrewRBJdGordon|: the list context includes the menu and file browser, right?
23:19:21JdGordon|anything that has that obvious list look
23:19:29andrewRBhaha. great, thanks
23:19:38JdGordon|i.e 90% of rockbox's ui
23:21:21stripwaxAsaelReiter - which plugin? simple answer is probably 'not easily', but by using profiling you might be able to compare cpu time to clock time
23:21:36gevaertspamaury: yes. The iMX31 reference manual (MCIMX31RM.pdf) from``Reference Manuals&fromTrng=false&fromPSP=true&showAllCategories=false&fpsp=1&SelectedAsset=Documentation&&assetLocked=false&leftNavCode=1&assetLockedForNavigation=true&code=i.MX31&isResult=false&c
23:22:20AsaelReiterstripwax: brickmania, with some changes that I wrote
23:22:49gevaertspamaury: is going to work better. That's the SoC of the gigabeat S, but the USB controller is the same as the one in the PP502x chips
23:24:07 Join dfkt [0] (i=dfkt@unaffiliated/dfkt)
23:26:41stripwaxAsaelReiter - if it's possible to isolate your code change, and profile that in a standalone environment, then you could try that.
23:26:48 Join linuxstb [0] (n=linuxstb@rockbox/developer/linuxstb)
23:28:20AsaelReiterI think it's not possible.
23:30:14AsaelReiter(The most important change is simplify some colossal loops)
23:31:25stripwaxIf the code is simplified, and the algorithmic complexity doesn't increase, and the instruction count goes down, then it sounds like a good change, even if the benefit can't easily be quantified
23:31:34pamaurygevearts: it there a document for usb programming in the link you gave me or is it only an electrical sheet ?
23:32:06 Quit stripwax ("")
23:32:10gevaertspamaury: MCIMX31RM.pdf has the full programming interface
23:33:08 Quit AsaelReiter ("CGI:IRC")
23:33:19 Quit HellDragon (Client Quit)
23:33:27 Join AsaelReiter [0] (n=d44c6c11@gateway/web/cgi-irc/
23:35:13pamaurygevearts: perhaps I'm stupid but I can't find it. There are 146 pages and I can't find the full programming interface except board regaisters
23:37:42gevaertspamaury: you mean 146 pages of USB docs, or 146 pages total? If the latter, you've got the wrong file
23:38:18pamaurythe latter, but that's the MCIMX31.pfg
23:38:24pamauryI'll try you're first link
23:38:38gevaertspamaury: you need MCIMX31RM
23:39:08gevaertsyou can try google for MCIMX31RM.pdf I guess
23:39:23pamauryyes that's what I mean, I downloaded the MCIMX31RM.pdf and it's only 146 pages long
23:39:44n1sthe whole RM is thousands of pages
23:40:32amiconn2364 pages, ~14MB
23:41:25pamaurydefinitely I downloaded the wrong one but I can't find it.
23:41:41gevaertshm, there seems to be something wrong there
23:42:10 Join safetydan [0] (n=deverton@rockbox/developer/safetydan)
23:42:59n1samiconn: there's a newer rev with even more pages, not sure if the additions are very interesting though
23:46:16pamauryare you sure the link you gave me is still good ?
23:46:37 Quit krazykit ("Lost terminal")
23:46:43gevaertsit looks like the version they have now isn't complete
23:46:47n1sthis is the real RM
23:47:13n1shmm, or maybe not
23:47:19 Join krazykit [0] (
23:47:30n1ssays rev 2.4 but only 3MB
23:47:34pamauryit's the same: 146 pages lonk
23:48:57 Quit AsaelReiter ("CGI:IRC")
23:49:10bluebrotherlooks like freescale replaced the file on their server
23:49:51domonokygoogle found it in the rockbox irc logs :-)
23:50:00pamaurythanks :)
23:52:01 Quit petur ("Zzzzzz")
23:52:02gevaertsI think the other ARM-based imx SoCs have the same USB core
23:52:48 Join AsaelReiter [0] (n=d44c6c11@gateway/web/cgi-irc/
23:52:49pamaurydoes this include the sansa ?
23:53:53gevaertssansa is PortalPlayer 5022. We don't have a datasheet for that at all, but the USB core is clearly the same (otherwise the driver wouldn't work)
23:53:56 Quit bmbl ("Bye!")
23:55:07pamauryOn the whole, have you got access to lots of specifications ?
23:56:10domonokygevaerts: this ofcourse only applys to the current supported sansas. :-)
23:56:27gevaertsof course :)
23:57:05 Quit AsaelReiter (Client Quit)
23:57:30gevaertspamaury: it depends. Some manufacturers like freescale have freely available specs, others like portalplayer (now nvidia) have none at all. There's everything in between those two
23:58:30pamauryAnd won't portalplayer give you the spec (I read somewhere that they contacted you no ?)
23:58:49linuxstbSandisk contacted us, not PortalPlayer.
23:58:59domonokyand then there is the very seldom case where a manufacturer provides us with normally not public specs (i think only AMS did that) :-)

Previous day | Next day