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 2010-12-03

00:01:42 Quit ender` (Quit: If Java had true garbage collection, most programs would delete themselves upon execution. -- Robert Sewell)
00:06:21 Quit n1s (Quit: Lämnar)
00:17:26*pamaury starts to disassemble the second elf file of the ____ section of the fuze+ firmware
00:22:37 Join designate72 [0] (
00:24:05 Join JdGordon| [0] (~jonno@rockbox/developer/JdGordon)
00:24:33 Quit JdGord (Quit: Bye)
00:25:01***Saving seen data "./dancer.seen"
00:28:08 Quit chattr (Remote host closed the connection)
00:32:00preglowsaratoga: yeah, i kinda expected it to be
00:32:08preglowtheir codecs are nice and concise
00:32:52 Join n1s [0] (
00:32:53 Quit n1s (Changing host)
00:32:53 Join n1s [0] (~n1s@rockbox/developer/n1s)
00:34:22 Join chattr [0] (
00:34:54 Join Keripo [0] (
00:35:54pamauryhum, there are breakpoint instructions in the fuze+ code !
00:42:02 Quit BlakeJohnson861 (Ping timeout: 255 seconds)
00:49:28 Quit mt (Ping timeout: 245 seconds)
00:50:11n1shmm, interestingly gcc can decide a function call is unlikely with no branch hints or profiling info at all
00:50:33 Quit dfkt (Quit: -= SysReset 2.53=- Sic gorgiamus allos subjectatos nunc.)
00:50:52CIA-7New commit by kugel (r28728): Android: Make lcd updates synchronous, doesn't make it faster but smoother (no updates are skipped) and guaranteed to be glitch free. ...
00:51:35casainhopamaury: what that means? why is there a Brekpoint instrautcion?
00:51:51casainhopamaury: are you doing debug?
00:52:12pamauryno, I just find it weird that the fimware code has some breakpoint instructions in it
00:52:35casainhowhat firmware is that?
00:52:58CIA-7r28728 build result: 5 errors, 43831 warnings (kugel committed)
00:53:38casainhooriginal firmware?
00:54:03casainhoit's easy to access OF?
00:54:11casainhohow did you got it?
00:54:37casainhohaving break point instructions is very weird!
00:55:07pamauryyou can easily download the firmware and I spent quite a number of hours figuring out the format and encryption scheme with bertrik
00:55:15pamaurythe tool is in svn
00:56:26CIA-7New commit by kugel (r28729): Fix red by moving the difinition up.
00:57:36 Join BlakeJohnson86 [0] (
00:58:01casainhoah, ok! :-)
00:58:25casainhomaybe format/encrytipon schme is worng ;-)
00:59:01CIA-7r28729 build result: 723 errors, 62215 warnings (kugel committed)
00:59:05casainhomsut be worng... I mean, there should no be breakpoint instructions, or the fimrware would not run,, right?
00:59:57pamaurythese are conditional
01:00:11pamauryI guess there are a cheap & low-level panic :)
01:00:21pamauryI assure you the format is right :)
01:00:42kugelperhaps they distributed a debug built, would also explain the zero key
01:01:33pamaurypossibly, that would explain some strange conditions in the code which seriously look like assertions
01:02:10casainhowhat SOC does it uses?
01:02:35pamaurystmp37xx == imx23
01:02:51casainhopamaury: I am working also with imx233!
01:03:02 Quit Keripo (Quit: Leaving.)
01:03:16pamauryyes, I know. We weren't talking about imx23 by chance :)
01:03:47casainhohey, I think guys at Chumby don't know how to decrypt image file!! and you got it????
01:04:31pamauryI documented it on the wiki and I wrote the sbinfo tool in svn with bertrik
01:04:34casainhowhere is the tool on SVN?
01:05:03 Join Keripo [0] (
01:05:11kugelwhat the heck
01:06:11CIA-7New commit by kugel (r28730): That was a bit too far upwards :(
01:06:19casainhopamaury: thanks!!!!!
01:06:47pamaurykugel: you particulary good at generated lots of builds errors
01:07:19casainhopamaury: so, will you create imx23 on Rockbox tree?
01:07:29kugelwell, I checked the sims before the first commit but not a real target. now I checked a real target but not the sims :'(
01:07:36casainhopamaury: do you have any working code?
01:07:36kugelbad luck
01:07:58pamaurywell, first we need to disassmemble the firmware to figure out the pins
01:08:11pamaurybertrik is working on some part and I'm working on another one
01:08:26kugelI suspect we also don't know yet how to inject some code?
01:08:29pamauryIf we get sufficiently enough information, we'll try to port it I guess
01:08:30CIA-7r28730 build result: All green
01:08:57casainhohere is the code I have for imx233, for Propendous board I am using:
01:09:10casainhowe got a lot of help from Chumby devs!
01:09:19pamauryNone of us as a real target. It seems there a reset mode which puts a recovery version of the firmware and you can update the firmware by putting the file at the root in MSC mode
01:10:07pamauryIf we are lucky, we can just write the code, convert it to .sb using elftosb2 and update the firmware just like with the OF
01:10:08casainhothere is a USB mode recovery
01:10:16pamauryon fuze+ ?
01:10:26 Join robin0800 [0] (
01:10:34casainhoon imx233 there is a USB recovery mode....
01:11:12pamauryI think I read somewhere in the doc that the manufacturer can enable/disable boot modes, so we don't really know if it's enabled and how to trigger it
01:11:56casainhoyes, there are fuse bits, that manufacture can burn
01:12:23casainhothere is a windows application to communicate by USB with imx233 and we can see/configure that bits
01:12:31pamauryyeah, and we saw that they did a nice job at keeping the zero key ;)
01:13:02casainhoyeah... that's de default key of imx233... so maybe they are not changing nothing on the fuse bits...
01:13:05 Quit n1s (Quit: Lämnar)
01:13:18pamauryThere might be some difference between imx233 which is a board and imx23 which is on the fuze+
01:13:36casainhohmmm, ok
01:13:48kugelI think they regretted the USB recovery on the e200 (it made more trouble than it helped) and the non-existent usb recovery on the ams series worked pretty well so I assume they deactivated it again
01:14:03 Quit Stummi (Ping timeout: 260 seconds)
01:14:15pamaurywhy more trouble ?
01:14:20casainhohey, there is only imx233!
01:15:04 Join pdrai [0] (DraaK@
01:15:32 Join Stummi [0] (Stummi@rockbox/developer/Stummi)
01:15:53JdGordon|kugel: I've had no rockbox time for ages, you don't want to take over the open() patch do you?
01:15:55pamaurywhen it comes to the chip, I think there is one which is imx23=imx233=stmp37xx, you can call it the way you want
01:17:55kugelJdGordon|: I can have a look...maybe
01:18:12pamaurykugel: actually, someone on the forum and on irc (if iirc) told us he tried the "reset function" (hold on/off for 20 seconds) and he said it acted like a firmware upgrade. So maybe there is a kind of recovery function
01:18:31JdGordon|kugel: I'd also ask about the dynamic screen size one but IIRC you wernt particularly keen on it?
01:18:42 Quit robin0800 (Remote host closed the connection)
01:19:56kugelI'm not overly keen on it because it doesn't really solve any problems because themes and many plugins are still resolution dependend, but I'm be in favor in generally
01:20:58 Quit Keripo (Ping timeout: 240 seconds)
01:21:05 Join Keripo [0] (
01:21:56kugelI thought the other day that we could possibly have themes resolution independent if we let %V (and friends) take % (e.g. x=10% of the width) as parameters and have bmp resize for the theme's images
01:22:50casainhopamaury, so you don't have a Sansa Fuze+?
01:23:20pamaurynot yet ;)
01:23:20JdGordon|kugel: I dont tinhk that would solve it all that much
01:23:38JdGordon|and that is ony a bit scary
01:32:18casainhopamaury: so, what is your motivation?
01:32:48pamaurymake sure the fuze+ arise
01:33:12pamauryI just don't want to buy a fuze+ if I'm not sure we can port it
01:33:24pamauryAnd that wasn't clear until we decrypted the sb format
01:34:51casainhod oyou know how many FLASH chips fuze+ have inside?
01:35:05pamaurythe wiki as some photos on the fuze+
01:35:11casainhowhere is stored firmware + user files?
01:35:36pamauryit is probably stored in a nand memory
01:37:33pamaurytime to go to bed, see you
01:39:52casainhosee you
01:41:35 Quit pamaury (Remote host closed the connection)
01:55:38 Join webguest10 [0] (
01:55:45 Quit webguest10 (Client Quit)
02:01:56 Quit kugel (Remote host closed the connection)
02:09:40 Quit casainho (Remote host closed the connection)
02:10:31mc2739bertrik: (logs) - looks like another FM Radio chip in Fuzev2 - see FS #11791
02:12:39 Part designate72 ("Leaving")
02:13:54 Join elcan [0] (
02:16:00 Join ReimuHakurei [0] (~ReimuHaku@
02:25:03***Saving seen data "./dancer.seen"
02:25:13 Quit ReimuHakurei (Quit: > Me... > Quitting IRC... > LOLYEAHRIGHT!)
02:25:21 Quit iq (Remote host closed the connection)
02:26:38 Join LambdaCalculus37 [0] (
02:26:38 Quit LambdaCalculus37 (Changing host)
02:26:38 Join LambdaCalculus37 [0] (~rmenes@rockbox/staff/LambdaCalculus37)
02:29:00 Join iq [0] (~iq@unaffiliated/iq)
02:29:13 Quit LambdaCalculus37 (Client Quit)
02:31:35 Join LambdaCalculus37 [0] (
02:31:35 Quit LambdaCalculus37 (Changing host)
02:31:35 Join LambdaCalculus37 [0] (~rmenes@rockbox/staff/LambdaCalculus37)
02:33:59 Part LambdaCalculus37
02:34:34 Quit DerPapst (Quit: Leaving.)
02:34:46 Join madalu [0] (~user@unaffiliated/madalu)
02:35:22saratogapreglow: I'm still hoping some day we get a GSOC student willing and able to tackle an ffmpeg aac port to rockbox, but it'd be so much work i'm not sure it'll ever happen
02:37:55 Join LambdaCalculus37 [0] (~rmenes@rockbox/staff/LambdaCalculus37)
02:41:38 Quit liar (Ping timeout: 240 seconds)
02:59:46 Quit madalu (Remote host closed the connection)
03:01:12 Join xxcv [0] (
03:03:41 Quit JesusFreak316 (Read error: Connection reset by peer)
03:08:27 Join madalu [0] (~user@unaffiliated/madalu)
03:31:29 Join Llorean [0] (
03:31:31 Quit Llorean (Changing host)
03:31:31 Join Llorean [0] (~DarkkOne@rockbox/user/Llorean)
03:55:57 Join eWill [0] (
03:58:51eWillI want to use RButil to create a voice file −− without the device connected. Possible?
04:00:46eWillnever mind, I think I figured it out.
04:01:07 Quit ehntoo (Remote host closed the connection)
04:04:53eWillstill can't make a voice file for fuze v2 −−- Encoding of C:/Users/<user name>/AppData/Local/Temp/rbvoice//LANG_CREDITS.wav failed
04:04:56 Quit InsDel (Read error: Connection reset by peer)
04:14:24 Quit TheSeven (Ping timeout: 245 seconds)
04:16:42 Quit tchan (Quit: WeeChat 0.3.3-dev)
04:18:39 Join Barahir_ [0] (
04:19:12 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
04:19:26 Join tchan [0] (~tchan@lunar-linux/developer/tchan)
04:22:01 Quit Barahir (Ping timeout: 255 seconds)
04:25:05***Saving seen data "./dancer.seen"
04:43:00 Quit amiconn (Disconnected by services)
04:43:00 Quit pixelma (Disconnected by services)
04:43:00 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
04:43:02 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
04:43:04 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
04:43:20 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
04:50:02 Quit sasquatch (Quit: WeeChat 0.3.2)
04:50:27 Join sasquatch [0] (
04:56:30 Quit Keripo (Quit: Leaving.)
04:57:10 Join Keripo [0] (
04:58:27 Quit ReimuHakurei__ (Quit: Stand by, Ready!)
05:01:41 Quit anewuser ()
05:08:09 Join ReimuHakurei [0] (~reimu@
05:15:20 Join someuser [0] (
05:15:44someuserjoin #slimrat
05:16:13 Quit someuser (Client Quit)
05:25:13 Join froggyman [0] (~seth@unaffiliated/froggyman)
05:30:34 Quit factor (Quit: Leaving)
05:31:06 Join factor [0] (
05:33:00 Quit LambdaCalculus37 (Quit: bed)
05:33:28 Quit Horscht (Quit: Verlassend)
05:48:30 Quit timccc (Ping timeout: 240 seconds)
05:48:30 Quit ReimuHakurei (Read error: Connection reset by peer)
05:53:28 Quit JdGordon (Ping timeout: 255 seconds)
05:54:41 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
05:57:20 Quit factor (Quit: Leaving)
05:57:45 Join factor [0] (
05:59:33 Quit factor (Excess Flood)
06:00:00 Join factor [0] (
06:08:18 Quit Strife89 (Ping timeout: 240 seconds)
06:15:24 Join Jeshikah [0] (
06:20:36 Quit madalu (Ping timeout: 276 seconds)
06:23:10Jeshikaha great audio/video format converter that's free) the videos would absolutely not play under rockbox. i assume that it's an issue of incompatible codecs, so i'm asking that here
06:23:48JdGordon|you might want to try again, sarting at the begining of your question
06:24:09Jeshikahagh, lemme try it in sections:
06:24:31Jeshikahwhat codecs does rockbox support for the mpg files?
06:25:08***Saving seen data "./dancer.seen"
06:29:31Jeshikahtyvm, will try it and see how it works for me
06:38:45Jeshikahyay, it's working now, lol, it seems it was the choice of width/height that was messing it up, lol
06:39:23 Quit Jeshikah (Quit: CGI:IRC 0.5.9 (2006/06/06))
06:42:59 Join Strife89 [0] (
06:46:33 Quit xxcv ()
06:48:41JdGordon|WTF?! "PANIC: No font!"
06:48:45JdGordon|never seen that one before
06:57:04 Quit JdGordon| (Quit: leaving)
06:58:30 Quit Judas_PhD (Quit: This is a quitting message)
06:59:11 Join ReimuHakurei [0] (~reimu@
07:00:48 Join xxcv [0] (
07:06:53 Quit UnclePervyJesus (Ping timeout: 260 seconds)
07:12:30 Join ReimuHakurei_ [0] (~reimu@
07:14:49 Quit ReimuHakurei (Ping timeout: 245 seconds)
07:33:15 Join timccc [0] (~timccc@
07:33:41 Join n1s [0] (
07:33:41 Quit n1s (Changing host)
07:33:41 Join n1s [0] (~n1s@rockbox/developer/n1s)
07:42:55 Quit panni_ (Quit: ( :: NoNameScript 3.81 :: ))
07:42:56n1suser mayaaren396 in the forums seems like a spammer
07:43:48 Quit factor (Read error: Connection reset by peer)
07:46:48 Join einhirn [0] (
07:55:27 Join ReimuHakurei__ [0] (~reimu@
07:55:48 Quit ReimuHakurei_ (Read error: Connection reset by peer)
08:00:38 Join factor [0] (
08:06:36 Join evilnick_ [0] (
08:07:27 Join fyre^OS [0] (
08:09:53 Quit evilnick (Ping timeout: 245 seconds)
08:09:54 Quit fyrestorm (Ping timeout: 260 seconds)
08:10:29 Quit kadoban (Ping timeout: 260 seconds)
08:25:10***Saving seen data "./dancer.seen"
08:40:11 Join Zagor [0] (~bjst@rockbox/developer/Zagor)
08:46:14 Join Unit193 [0] (
08:47:55 Join ender` [0] (
08:49:46 Quit Strife89 (Quit: Bed)
08:50:40 Quit MethoS- (Remote host closed the connection)
08:52:04 Quit advcomp2019 (Ping timeout: 276 seconds)
08:52:20 Quit simonrvn (Ping timeout: 240 seconds)
08:52:27 Part Unit193
08:53:19 Join simonrvn [0] (
08:53:49 Join Judas_PhD [0] (
09:02:48 Quit TheSeven (Ping timeout: 245 seconds)
09:08:27 Quit eWill (Quit: ChatZilla 0.9.86 [Firefox 3.6.12/20101026210630])
09:11:50 Quit antil33t ()
09:14:08 Join Rob2223 [0] (
09:14:14 Join advcomp2019 [0] (
09:14:16 Quit advcomp2019 (Changing host)
09:14:16 Join advcomp2019 [0] (~advcomp20@unaffiliated/advcomp2019)
09:17:41 Quit Rob2222 (Ping timeout: 255 seconds)
09:21:31 Join petur [0] (d408b802@rockbox/developer/petur)
09:24:24 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
09:26:43 Join kevku [0] (~kevku@2001:7d0:0:f000::135d)
09:28:32 Join LinusN [0] (~linus@rockbox/developer/LinusN)
09:30:06 Quit n1s (Quit: Lämnar)
09:43:52AlexPgod damn spammers
09:44:08AlexPFricking loads of them at the moment
09:53:04Stummimaybe there should be a better captcha?
09:54:01AlexPOne that isn't always the same would be a start
09:56:37Stummibtw, on first call the capture is not rendered correct
09:57:09Stummi(and i don't like graphical captchas. Somthing like "how many is three plus four?" is also good)
10:00:18AlexPYeah, I know
10:00:24AlexPBut I can't do anything about this
10:00:46 Join Kitr88 [0] (
10:00:47AlexPscorche: Could you have a look at fixing/changing the forum captcha?
10:01:05AlexPThe current one is a) broken on first try and b) always asks the same thing
10:01:19pamauryI find it strange that the Lyre project has a binary copy of elftosb2 in their svn, is this allowed ?
10:01:34AlexPWe've just had a big increase in spammers, maybe they just worked it out
10:03:00 Quit JdGordon (Ping timeout: 240 seconds)
10:04:08 Quit Kitar|st (Ping timeout: 255 seconds)
10:04:53 Quit Kitr88 (Ping timeout: 240 seconds)
10:05:31 Join efyx [0] (
10:07:46 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
10:07:47 Join Kitar|st [0] (
10:25:11***Saving seen data "./dancer.seen"
10:25:55 Join mt [0] (~mt@
10:32:09AlexPaaargh, more of the bastards
10:43:30 Quit simonrvn (Quit: ZNC -
10:43:56StummiAlexP, do you have direct access to the files for the forum? I don't know the SMC-Forum, but I think shouldn't be a big issue to find out whre the captcha-code is generated and hack this to generate a realy random code. As quick'n'dirty-solution for the actuall spammer-problem
10:47:14 Join simonrvn [0] (
10:47:25Bagderaren't these spammers human?
10:47:43Stummii think that they are bots
10:49:36Stummioh, i thought the forum-files are in svn somewhere under www or so. But it seems that i was wrong
10:53:13Zagoryes, the forum is hosted on a different machine ran by scorche
11:04:00 Join antil33t [0] (
11:04:59 Part timccc
11:27:25 Join hebz0rl [0] (
11:42:35 Quit mt (Ping timeout: 255 seconds)
12:03:50AlexPStummi: No, I don't have any access
12:04:29AlexPBagder: I'm not sure - the captcha is broken (it always gives the same question), so if someone found that out it'd be easy to get the bots to do it
12:04:43AlexPEither way, it'd be nice to have a working one :)
12:12:35 Quit kevku (Ping timeout: 260 seconds)
12:13:49 Join sampattuzzi [0] (~sam@
12:14:29 Quit sampattuzzi (Quit: Ex-Chat)
12:14:51 Join sampattuzzi [0] (~sam@
12:16:17gevaertsAlexP: banjia26
12:16:24 Join kugel [0] (~kugel@rockbox/developer/kugel)
12:16:25AlexPalready done :)
12:16:34gevaertsgood :)
12:20:54sampattuzziHello people of rockbox. I'm trying to debug some USB issues which I reckon are due to rockbox not handleing I/O errors correctly. Could somebody point me to the correct section of the usb stack and possibly relevant documentation for the USB spec?
12:21:46 Join MethoS- [0] (~clemens@
12:25:12***Saving seen data "./dancer.seen"
12:26:19 Join InsDel [0] (
12:27:46gevaertssampattuzzi: You're probably looking for SCSI specs here
12:28:06gevaertsAnd in the rockbox code, firmware/usbstack/usb_storage.c
12:28:38sampattuzzigevaerts, cool, I will take a look at that, see if it makes any sense.
12:28:45gevaertsWhat I would do is start with USB tracing (if you're on linux, wireshark can do that) of both the OF and the rockbox code
12:28:46Torneyou might get more by talking to us about what issues you've got, and why you think there's a problem with our handling
12:29:01gevaertsIf you manage to get such traces, please attach them to your flyspray task :)
12:29:49gevaertsThe main problem is that I've never been able to test these things due to a lack of suitably broken hardware
12:30:15sampattuzzigevaerts, out of interest, I noticed lots of logf() calls in the code but couldn't find a way of enabling them in the documentation.
12:30:47Tornelogf.h sort of explains it ;)
12:31:02gevaertssampattuzzi: enable the "#define LOGF_ENABLE" near the top of the source file, then pick "Advanced" when configure asks, and then enable logf there
12:31:59gevaertsTorne: context is FS #10873
12:32:24sampattuzzigevaerts, thanks, I was just about to post a link.
12:40:17 Quit InsDel (Read error: Connection reset by peer)
12:44:32sampattuzziAh, I could do with that new linux kernel patch. The build is making my system sloow.
12:45:08bzedsampattuzzi: I usually run something like nice ionice -c 3 make to build stuff
12:45:15bzedbetter than nothing ;)
12:46:07sampattuzzibzed, cheers, I'm giving that a go now.
12:51:27 Quit sampattuzzi (Ping timeout: 255 seconds)
12:52:00 Quit xxcv ()
12:52:47Torneah, he's gone
12:52:57Tornei was gonna point out there's a tiny shell script that does exactly the same thing as that patch
12:53:02Tornewithout needing a new kernel :)
12:54:47 Join dfkt [0] (dfkt@unaffiliated/dfkt)
12:59:19 Join jhMikeS [0] (~jethead71@
12:59:20 Quit jhMikeS (Changing host)
12:59:20 Join jhMikeS [0] (~jethead71@rockbox/developer/jhMikeS)
12:59:20 Quit _jhMikeS_ (Disconnected by services)
13:36:40 Quit bluebroth3r (Ping timeout: 265 seconds)
13:36:57 Join DerPapst [0] (
13:38:11 Join bluebrother [0] (
13:38:11 Quit bluebrother (Changing host)
13:38:11 Join bluebrother [0] (~dom@rockbox/developer/bluebrother)
13:43:20 Quit kugel (Ping timeout: 272 seconds)
13:44:22 Join timccc [0] (~timccc@
13:45:26 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
13:46:39 Join mt [0] (~mt@
13:47:07 Join sampattuzzi [0] (~sam@
13:49:40sampattuzzigevaerts, I obviously missed a step, what did you say I had to do in the logf.h file?
13:50:08gevaertsin logf.h, nothing
13:50:24gevaertsIn the source file where you want logging, look for "#define LOGF_ENABLE" and enable it
13:51:53gevaertsAnd then do an (A)dvanced build
13:51:54sampattuzziah, okay, that's what I missed. Excuse my ignorance.
13:52:40sampattuzziI did the second part.
13:54:03 Quit Rob2223 (Read error: Connection reset by peer)
13:54:20 Quit Bawitdaba (Read error: Connection reset by peer)
13:54:34 Join Bawitdaba [0] (
13:54:39 Join Rob2222 [0] (
14:01:55sampattuzziOkay, got myself a trace and uploaded it to the bug. Can't work out what it's doing though.
14:02:55gevaertsA USB-level trace would be much more interesting I think, especially if you have both rockbox and the OF
14:03:48sampattuzziIn what sense?
14:04:52sampattuzziA trace from the linux side then?
14:05:05gevaertsyes, e.g. using wieshark
14:06:59 Join n1s [0] (~n1s@rockbox/developer/n1s)
14:08:43 Join jfc^2 [0] (
14:09:20 Quit scorche (Disconnected by services)
14:09:27 Join scorche` [0] (~scorche@rockbox/administrator/scorche)
14:09:37 Join tchan1 [0] (
14:10:59 Quit jfc^3 (Read error: Operation timed out)
14:12:49 Quit tchan (Ping timeout: 245 seconds)
14:14:05amee2khm... did anyone else ever have the idea of porting rockbox to PC style platforms?
14:14:45amee2ki recently got my hands on two really old laptops... a thinkpad that i modded into a serial terminal using linux
14:14:49amee2kand now i've got this old
14:15:31amee2kArmada 1700 here and kinda thought it would make a nice jukebox with rockbox on it
14:16:05amee2ktheres nothing on the wiki at least that suggests there is an x86 port
14:16:36tmztamee2k: there's a simulator
14:16:55gevaertsYou should be able to do something based on the SDL application port
14:17:06amee2kits like a P2 200 in it
14:17:07tmztwhich is a native build using libsdl, assumming you have a framebufer
14:17:25amee2kthe windows 98 on it takes ages to load
14:17:46amee2ki could try install linux on it and see if the simulator works
14:17:54amee2kbut a native port would be like
14:17:58amee2kinfinitely superior
14:18:32gevaertsWell, feel free to work on it
14:18:40n1sit would also take a lot of work whereas compiling the sdl app takes a few minutes tops
14:19:03amee2kgevaerts: yeah... if i ever get the hang of this i might just do it
14:19:36amee2kwhat i'd find kinda ugly about just using the simulator is that you'd end up with the emulation screen on an X display
14:19:41amee2kinstead of a real display
14:20:12amee2kand i'm not sure if i can get any useful performance out of a slow processor like that
14:20:15gevaertsYou can use SDL on a framebuffer
14:20:43*amee2k ponders
14:22:34amee2kso i'd basically package an embedded linux as bootloader lol
14:25:13***Saving seen data "./dancer.seen"
14:27:57 Quit antil33t (Read error: Connection reset by peer)
14:28:07 Join antil33t [0] (
14:28:52 Join designate72 [0] (
14:30:54 Quit sampattuzzi (Ping timeout: 255 seconds)
14:44:57 Join TheLemonMan [0] (~lem0n@
14:45:55 Join Strife89 [0] (
15:07:28 Join sampattuzzi [0] (~sam@
15:07:51 Join komputes [0] (~komputes@ubuntu/member/komputes)
15:07:56amee2kmmmh, i'll first see if i can fix the ipod and play around with the plugin api. if i can't even get that in my thick head, not much point in trying a port >_<
15:10:30Torneporting to native x86 is going to involve a *reasonable* amount of specialist x86 knowledge
15:10:37Torneto implement, say, context switching
15:11:14Tornei don't want to stop you but you may find it easier and more useful to just get the SDL version on an absolutely minimal linux distro
15:11:20gevaertsThe advantage is that such knowledge is reasonably widespread and things are well documented
15:11:26TorneWell, true
15:11:50n1sbut you'd still need to write drivers and stuff
15:11:52Tornea tiny linux distro consisting of basically just a kernel and a copy of rockbox with libsdl compiled to use the framebuffer, though
15:12:03Torneis going to boot in a comparable time to a native rockbox port
15:12:09Tornebut you have to do *way* less to get it running
15:12:30amee2kyeah, what gevaerts said. i don't need to reverse engineer the platform first
15:12:36Torneso, yeah. if you really want to do it, go for it; but x86 is *really weird*
15:12:50Torneeven though the docs exist, you may be underestimating just how weird it is
15:12:55amee2kalso, i understand it that rockbox assumes a plain memory model so no need to deal with segmentation or paging
15:12:57gevaertsOf course you *can* reverse engineer it, and only look at documentation when you really get stuck :)
15:13:10Torneamee2k: you can't avoid dealing with segmentation on x86, at least in a minimal way
15:13:23Torneand you would probably want to write a pagetable anyway
15:13:36amee2kTorne: well, yeah. you set up three segments that span the entire available memory and leave it at that
15:13:49amee2ki did some real mode os deving when i was in school, maybe 6 or 7 years ago
15:14:07n1smake it a payload for coreboot and flash to the bios chip!
15:14:08amee2kit was kinda cool, but i didn't get too much into PM stiff
15:14:10Torneyou don't need to do anything *complicated* with the MMU but it's probably nicer to have it enabled and make a fixed virtual memory map
15:14:25Tornewe do that on the ARM platforms we support that have an MMU
15:14:27gevaertsn1s: does coreboot support such old laptops? :)
15:14:41amee2khmmm thats an interresting idea actually
15:14:51amee2kto deal with awkward physical memory mappings
15:15:04n1sgevaerts: it hardly supports anything so you will still have all the fun porting to do :)
15:15:27amee2kiirc, handling the memory stuff kinda was what got me stuck with protected mode stuff back then
15:15:54amee2kwell, i was about to put a multiboot header on the kernel thingy and use grub to load it
15:16:01Tornebut anyway; the advantages I can imagine of running rockbox natively on x86 versus running it under linux is virtually zero
15:16:16amee2kthat'd dump me right into protected mode with at least a half decent memory layout provided
15:16:16Tornefor a sufficiently small linux.
15:17:18amee2kthese days small and linux is sort of a contradiction unless you really want to start a distro from scratch
15:17:28Torneer, why?
15:17:36amee2kfor the serial terminal mod on the thinkpad i tried to fit linux on a 128M flash card
15:17:39amee2kforget it :P
15:17:51Tornei have several working linux systems that fit in 4MB
15:17:58Tornemaybe one or two that fit in 2MB
15:18:14amee2kyeah, but support for them is awkward, and they're usually specialized
15:18:23Torneyou would only need a kernel, busybox to handle boot/init, and rockbox and the libraries it uses (SDL)
15:18:28TorneI'd be surprised if that came to more than 16MB
15:18:29amee2kso you'd effeectively end up forking the distro and modifying it for your stiff
15:18:32Torneyou could make it a ramdisk image, even
15:18:56Torneno, this is not using some specialised system
15:18:58Bagderamee2k: for small embedded things you don't need to complicate things and use a "distro"
15:19:01amee2kyeah... thats kinda what i meant by making a distro from scratch
15:19:13TorneYou don't hvae to make it from scratch
15:19:16TorneOE builds it for you
15:19:24amee2k"OE" ?
15:19:24sampattuzzigevaerts, I'm looking at the wireshark traces right now and they appear completely different. OF shows only protocol USB and info URB_BULK whereas rockbox has a lot of USBMS and varied info.
15:19:42amee2kdoes that work?
15:19:45Torneof course
15:19:49Tornewhy wouldn't it work?
15:19:51Bagderbuildroot should also be decent
15:19:55amee2ki looked into emdebian but its a pile of wank
15:19:58sampattuzzigevaerts, would you like me send them to you in some format?
15:20:13gevaertssampattuzzi: can you attach them to the bug report?
15:20:16amee2kthey don't even have a precompiled 486 kernel lol
15:20:20Torneno it isn't.
15:20:27Torneand why would you have precompiled kernels for embedded systems?
15:20:29Tornethat would be mad
15:20:33sampattuzzigevaerts, which would be the best file format, they are very long.
15:20:36Torneanyway. this is kinda offtopic.
15:20:48Tornefeel free to try and port to native x86 if you want
15:20:55sampattuzzigevaerts, flat text would seem a little inaccessible.
15:21:04Tornebut I seriously think there will be no real benefit over using linux/sdl
15:21:11gevaertssampattuzzi: native wireshark saves, compressed
15:21:15amee2kwell, i'll play around with it on the ipod and see how it goes
15:21:18Torneif you don't know how to make a sufficiently small linux then that's a different problem, unrelated to rockbox :)
15:21:27Tornei can show you, if you ask some other time ;)
15:22:16amee2kthat would be cool so i might just take you up on that sometime
15:23:59 Join evilnick_B [0] (0c140464@rockbox/staff/evilnick)
15:25:13Tornethe interesting and rockbox-relevant part, then, would be getting the RaaA-on-SDL done nicely
15:25:31Torne(rather than running the simulator)
15:25:51Torneyou could develop/debug/test that on a full linux for convenience
15:26:05amee2khehe yeah
15:26:20n1sthe sdl app already works although it's not much different from a touch sim
15:26:33amee2ki've been thinking about a convenient way to update the firmware without the device acting as usb device
15:26:40Tornen1s: yeah, i meant making it "nice"
15:27:01Tornei.e. keyboard shortcuts, proper filesystem scanning, etc
15:27:07Tornesome of the same issues we have on android, i guess
15:27:08amee2kwell, it should work in full screen mode, obviously
15:27:22amee2kand the configuration menu would probably need some way to configure the keymap
15:27:58Tornefor a PC booting off an intenral hard disk updating the firmware is easy.. just copy it over
15:28:04amee2kthough that can stay in a header file for the first try
15:28:06n1si think RaaA as it is would work *better* for a dedicated system's only UI than it works as a regular desktop app
15:28:13Tornen1s: yes, undoubtedly
15:28:29amee2kTorne: yeah, but you need to get the hard drive out and on a usb adapter
15:28:36amee2kevery time you update it
15:28:54amee2kuntil you got it working enough so you can upload updates via USB
15:29:28n1samee2k: can't you run a shell on your busybox thing to copy stuff via usb?
15:30:09Torneyou wouldn' tneed to do anything to make usb work if it was running the linux kernel
15:30:12amee2ki'm not sure if you can make a USB host controller act as a device end too
15:30:12Torneyou just plug in a usb stick
15:30:16Torneyou can't.
15:30:35amee2kyeah, i kinda expected it won't work
15:31:13amee2ki could make a USB to serial bridge with an avr controller, but the transfer rate would drag ass
15:31:35Torneyah. or if you used linux, then USB host would Just Work and you could use any usb storage device.
15:31:35n1salthough for early phase debugging you probably want to run it in a VM
15:31:39amee2kmost uarts top out at something like 115.2 kbaud
15:31:44Tornealso, you could have networking
15:31:47amee2kand thats a tough call for the AVR too
15:32:02Torneyou could just have a button that ran a script in the host OS which downloaded and installed hte latest build from the internet :)
15:32:08amee2krockbox does support a network stack?
15:32:23Torneno, but it wouldn't need to
15:32:32Tornelinux has one.
15:32:37Torneand busybox implements wget and friends
15:32:38amee2ki see
15:32:48Torneso you just exit rockbox, run a shell script that downloads a new build and installs it, then run rockbox again
15:32:55Torneyou can automate that esaily :)
15:33:54amee2khmmm i'm kinda starting to like the idea of layering it on a linux kernel
15:34:18Tornethere are lots of advantages. the only real advantage to doing it natively is it *might* boot slightly faster
15:34:32Tornebut we're likely talking a difference measured in a single digit number of seconds here, if that
15:34:54amee2kfunctionality and using existing device drivers are certainly big + points
15:35:32amee2kespecially using the linux VFS layer. i kinda understand rb natively only mounts a single file system at a time
15:35:41Torneno, we support multiple mounts
15:35:52Tornesee devices with both internal storage and an SD slot
15:35:54Tornelike the fuze
15:36:08amee2kthe simulator intercepts the raw file system calls and redirects them to appropriate C library calls, right?
15:36:20Torneyou wouldn' tbe using the simulator
15:36:26Torneyou'd be using the SDL app port
15:36:34Tornethey're similar, but not the same thing
15:37:22n1sso you would use linux filesystem and harddrive functions
15:37:51Torneindeed. rather than simulating a single device filesystem, the port would just have access to the entire host's FS
15:38:00Torneso yeah, it would see whatever was mounted on the linux VFS.
15:38:26Tornethings like mounting USB devices and so on can just be don eby the host, with busybox udev scripts to automount inserted storage devices, etc
15:39:27Torneanyway, if you wanted to do something like this a good start would be experimenting with the sdl port on a regular linux system
15:39:42Tornejust fire it up and try using it and see what works well and what doesn't :)
15:39:59Tornecutting the host linux down to the bare minimum can be done seperately later
15:43:02Tornewell, you might want to try specifically getting it to run fullscreen on the framebuffer console with SDL.
15:43:04 Quit TheLemonMan (Quit: free(me))
15:43:05Tornerather than in a window under X
15:43:19Tornebut that can (and should) still be done on a regular linux distro, just don't run X.
15:43:33amee2ki've got my projects desk full of ipod and battery charger parts right now
15:43:36Torneyou can debug with a real debugger when stuff goes wrong, there, which helps a lot ;)
15:43:43gevaertssampattuzzi: thanks. I'll have a closer look later today
15:43:57amee2kwhich will hopefully improve when the fucking flash card for the pod finally arrives >:|
15:44:10amee2kthen i'll install debian on the lappy
15:44:17sampattuzzigevaerts, it all looks very confusing but I would love to know more about how it's working.
15:44:50amee2ki'm not too fond of the idea of killing X on my desktop
15:45:01amee2ktoo much crap running that i don't want to kill along with it
15:46:03 Nick Strife89 is now known as Desk-Strife89 (
15:46:25gevaertssampattuzzi: the annoying thing is that wireshark doesn't seem to have nice decoding for UMS yet
15:46:45Torneamee2k: well you don't have to kill it, just switch to a tty ;)
15:46:58Torneas long as your console is a framebuffer console
15:47:23gevaertshm, it should
15:50:59pamaurybertrik: any progress on fuze+ ?
15:51:52 Join domonoky [0] (~Domonoky@rockbox/developer/domonoky)
15:51:55 Quit antil33t (Read error: Connection reset by peer)
15:52:04 Join antil33t [0] (
15:55:47 Join ReimuHakurei_ [0] (~reimu@
15:56:25pamauryI'm seeing some pin control registers accessed in the second elf, I don't know yet what is the use
15:57:39 Quit ReimuHakurei__ (Read error: Connection reset by peer)
15:57:50 Quit MagusG (Ping timeout: 276 seconds)
16:03:55pamauryI think the second elf file basically deals with clocks and ram
16:06:39bertrikok, the chumby command file was structured with several parts each initing something: power, clocks, sdram
16:09:32 Join robin0800 [0] (
16:11:38gevaertssampattuzzi: if you open the apple trace, then "Mark All Packets" (Edit menu), then unmark the four last GET CONFIGURATION lines (nrs 47,48,49,50), save as, select "marked packets only", then save, and open the new file, you'll get *much* better decoding
16:16:08gevaertssampattuzzi: anyway, that OF trace will be *very* helpful in understanding what we're supposed to do
16:16:46sampattuzzigevaerts, brilliant! I'm glad we're getting somewhere.
16:22:22 Join liar [0] (
16:25:00saratogaAlexP: ban chong11chong
16:25:17***Saving seen data "./dancer.seen"
16:33:45 Quit sampattuzzi (Ping timeout: 255 seconds)
16:35:30 Join panni_ [0] (
16:36:28 Quit petur (Quit: Page closed)
16:38:34 Join jgarvey [0] (
16:39:46 Join kevku [0] (~kevku@2001:7d0:0:f000::135d)
16:42:38 Join sampattuzzi [0] (~sam@
16:47:17 Join toffe82 [0] (
16:47:39pamauryomg :o So much initialisation code for DRAM !
16:50:42 Join bertrik_ [0] (~bertrik@rockbox/developer/bertrik)
16:50:43 Quit bertrik (Read error: Connection reset by peer)
16:57:36 Quit sampattuzzi (Ping timeout: 255 seconds)
16:59:36 Quit mortalscan (Remote host closed the connection)
16:59:52 Join mortalscan [0] (~mortalsca@
17:00:05 Nick bertrik_ is now known as bertrik (~bertrik@rockbox/developer/bertrik)
17:01:02 Part LinusN
17:04:52 Quit TheSeven (Ping timeout: 240 seconds)
17:11:18 Join sampattuzzi [0] (
17:11:30 Join kugel [0] (~kugel@rockbox/developer/kugel)
17:12:57 Quit {phoenix} (Remote host closed the connection)
17:16:51 Nick tchan1 is now known as tchan (
17:17:00 Quit tchan (Changing host)
17:17:00 Join tchan [0] (~tchan@lunar-linux/developer/tchan)
17:22:51 Quit einhirn (Quit: Miranda IM! Smaller, Faster, Easier.
17:23:50AlexPsaratoga: done
17:26:25gevaertssampattuzzi: ping
17:27:39 Part Zagor
17:30:26 Quit robin0800 (Remote host closed the connection)
17:31:06 Quit sampattuzzi (Ping timeout: 260 seconds)
17:33:15 Join TheLemonMan [0] (~lem0n@
17:35:14 Join Topy [0] (
17:38:09 Join sampattuzzi [0] (
17:38:46sampattuzzigevaerts, I'm here now, what did you want to say?
17:39:01 Quit T44 (Ping timeout: 245 seconds)
17:39:11gevaertssampattuzzi: I posted a quick patch. Could you try it?
17:39:16gevaertsAlthough I suspect it's not enough
17:47:12sampattuzzigevaerts, okay well it's not mounted so I will do a trace for you.
17:47:49 Quit riotz (Read error: Operation timed out)
17:52:45pamaurywtf, the fuze+ firmware seems to have several dram settings !
17:52:53 Join riotz_ [0] (riot33r@gateway/shell/
17:53:58sampattuzzigevaerts, uploaded
17:56:50n1spamaury: they could use different ram chips in the same model, other players do this
17:57:25pamaurythere are 4 different settings, it's hard to tell the difference, some really low-level settings
17:57:37pamauryI don't know how they choose yet
17:58:40n1sif it's runtime selected somehow, we should probably do that too or risk some players not working
17:59:23n1sor jus use the most lax of the sets of settings, as that should work for all chips and efficiency is not too important at this stage
18:02:17pamaurywell, perhaps we can just reuse the same init code
18:02:27pamauryafter all, the sb format is quite flexible
18:02:37pamaurywe can just replace the relevant part of it :)
18:03:08pamauryof course, we need to check what the init code does, especially for virtual memory, interrupts and so one
18:07:52 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
18:14:25pamauryhum, there are some tables in the imx23 document, I'll have to check if the values match
18:17:41pamauryoops, there is a unknown register !
18:25:18***Saving seen data "./dancer.seen"
18:25:34 Join Strife89TX [0] (~cstrife89@
18:32:56scorche`OK....tonight i will investigate the compatibility of multiple mods we have done on the forums with new version of SMF to evaluate an upgrade
18:33:56scorche`depending on the results of that evaluation, i might or might not take the forums offline tomorrow and perform the upgrade
18:34:32scorche`(note: this is also dependent on how many 2500+ word final papers i need to work on this weekend)
18:34:36 Nick scorche` is now known as scorche (~scorche@rockbox/administrator/scorche)
18:36:04scorcheat the same time, i will look at upgrading a few other services on the server (might as well), so the themesite might be affected in some manner as well
18:37:53 Quit ReimuHakurei_ (Ping timeout: 240 seconds)
18:40:19TheLemonManSamsung put ways more encryption in themes than in fw and bootloader .-.
18:41:44Strife89TXPerhaps artwork is more important to them than a commercial firmware?
18:41:44alexbobPheaven forbid people personalize their players
18:42:23n1smaybe they never heard of the chain of trust
18:42:30Strife89TXArt could be stolen for other works, I guess.
18:43:19 Quit Guinness (Read error: Connection reset by peer)
18:44:31n1syes but if you can break into the firmware that decrypts the art, no amount of encryption on the art is effective as you can either decrypt it the same way as the firmware does or dump the decrypted data
18:45:50 Join domonoky1 [0] (
18:46:45gevaertsUMS error handling is a mess
18:46:57 Quit domonoky (Ping timeout: 245 seconds)
18:52:23 Quit DerPapst (Quit: Leaving.)
18:52:25pamauryerror handling in rockbox is a mess
18:52:50n1strue that
18:52:52gevaertsMaybe, but usually it's not because someone did weird things in a spec
18:53:40pamauryah, you mean the protocol error handing :)
18:53:52*gevaerts claims that a bad sector is *not* a protocol error and therefore an endpoint HALT is *not* an appropriate response
18:54:13tmztit's a sense error
18:54:57gevaertsWell no, it returns an error code in the sense data :)
18:55:08tmzterr, yeah
18:55:14tmztbut not in the transport
18:55:16gevaertsBut in USB it also involves stalling the endpoint, which messes everything up
18:56:09pamauryyeah, ums is really different from other usb protocols
18:56:24gevaertsMaybe there's an alternative solution
18:56:38pamauryto what ?
18:56:57gevaertsMy guess is that the stall is there because you send less data than expected, which would otherwise confuse the host
18:57:38gevaertsBut according to my reading of the UMS spec, that stall is optional
18:57:52pamauryI don't have the spec in mind
18:58:29gevaertsSo in theory, we could just send enough data (all zeroes or something, or just random junk) instead
18:58:47gevaertsThe host should know that this data is junk because we still return an error code
18:59:08gevaertsThe problem I have right now is that after a stall the CSW doesn't always get through
18:59:16gevaertsIt sometimes gets eaten by the stall itself
19:00:35gevaertsAnd that's not at all trivial to fix
19:01:54sampattuzzigevaerts, how does tho OF deal with it?
19:02:29gevaertsBy having a different usbstack design :)
19:03:50sampattuzziokay, I didn't realize they could be that different.
19:05:20saratogascorche: can we get that comments/blog software for the front page while you're at it ;)
19:06:46pamaurybertrik: perhaps we should setup a wiki page for our firmware analysis ?
19:07:16 Join DerPapst [0] (
19:10:52AlexPscorche: ta
19:11:13 Quit sampattuzzi (Read error: Connection reset by peer)
19:14:02TheSevengevaerts: my disk is really nice. it has: 1. sectors that can't be written and can't be read, 2. sectors that can be written but can't be read, 3. sectors that can both be written and read successfully, but don't return the data that was written :)
19:14:16TheSevenbut I haven't yet spotted read-only blocks
19:14:29gevaertsTheSeven: yes, I want that disk :)
19:15:04pamauryTheSeven: nice disk :)
19:15:17TheSevenalso, there are both consecutive and scattered bad sectors
19:15:31TheSeventhe scattered ones have almost identical distance in between them
19:15:55 Join sampattuzzi [0] (
19:16:09gevaertssampattuzzi: could you try the patch I just uploaded?
19:16:36 Quit TheLemonMan (Quit: free(me))
19:16:51sampattuzzigevaerts, certainly
19:17:36 Join Horscht [0] (~Horschti@xbmc/user/horscht)
19:18:33gevaertssampattuzzi: oh, wait. That patch has some bits in there you don't want...
19:18:48 Join krabador [0] (
19:18:52gevaertsThere are two lines that check for sector 10000. Remove those
19:19:02gevaerts(they're for local simulation of bad sectors)
19:20:33 Quit DerPapst (Read error: Connection reset by peer)
19:20:36sampattuzzi47 and 58?
19:21:03 Join DerPapst [0] (
19:21:16gevaertsIn the patch, yes, but renove them from usb_storage.c after patching
19:21:24gevaertsOtherwise I suspect the patch won't apply
19:27:21sampattuzzistill no luck
19:29:43sampattuzzigevaerts, new trace for you.
19:30:31gevaertssampattuzzi: that looks a bit short
19:30:39gevaertsDid you capture the correct bus?
19:30:59sampattuzzilet me just check that.
19:32:31sampattuzzigevaerts, yep, it only produces 21 new packets when I plug it in. Then that's it.
19:33:46gevaertssampattuzzi: dmesg should tell you which bus number is used.
19:34:28gevaertse.g. "usb 2-1: new high speed USB device using ehci_hcd and address 33" means bus 2
19:38:01sampattuzzigevaerts, that should be better now.
19:40:39 Join stoffel [0] (
19:42:05gevaertssampattuzzi: can you also provide dmesg output for that one? I don't see anything unexpected in the trace
19:43:47gevaertsThe read starting at packet 561 fails, which gets notified by 566, after which the host retrieves sense data (567/570) that indicate the error
19:44:01gevaertsAfter that the host does a Test Unit Ready and then goes on reading somewhere else
19:44:04 Quit kugel (Remote host closed the connection)
19:45:12*gevaerts will look more after dinner
19:45:45 Quit krazykit (Ping timeout: 245 seconds)
19:45:53sampattuzzigevaerts, done, but enjoy you dinner first.
19:47:53 Join krazykit [0] (
19:57:09 Quit sampattuzzi (Quit: Ex-Chat)
19:58:49scorche|shsaratoga: i am still working on finals so the fact that i hope to find any time to work on it this weekend is a miracle in and of itself ;)
20:02:08saratogayes I figured so
20:06:21 Quit Strife89TX (Quit: Vamoose!)
20:06:34 Quit DerPapst (Read error: Connection reset by peer)
20:07:06 Join DerPapst [0] (
20:13:28 Join InsDel [0] (
20:25:19***Saving seen data "./dancer.seen"
20:26:45 Quit factor (Read error: Connection reset by peer)
20:27:03 Join factor [0] (
20:33:44*gevaerts is confused
20:35:37 Join {phoenix} [0] (
20:38:04 Join bmbl [0] (~bmbl@unaffiliated/bmbl)
20:38:25 Quit designate72 (Quit: Leaving)
20:49:04 Join designate72 [0] (
20:55:43 Quit Llorean (Read error: Connection reset by peer)
20:56:57 Join Llorean [0] (
21:11:05saratogan1s: I don't think make install makes a zip
21:13:05 Quit bmbl (Quit: Verlassend)
21:15:08saratogai think it assembles a local .rockbox folder, then does a bunch of file copies into the simdisk folder without bothering to zip it first
21:19:27 Join t0rc [0] (~t0rc@unaffiliated/t0rc/x-5233201)
21:29:07 Quit jgarvey (Quit: Leaving)
21:41:59CIA-7New commit by pamaury (r28731): sbinfo: fix crazy condition to avoid empty elf file generation (it is reversed)
21:42:14AlexPzagor: Did you ever sign up for the Rockbox gmail calendar?
21:42:58pamaurywhat is the rockbox gmail calendar ?
21:43:35AlexPpamaury: It will be to do things like have release schedules etc.
21:43:46AlexPI posted to the dev list about it, let me find the link
21:43:59*Dreamxtreme looks
21:44:11*Dreamxtreme really should sign up for the mailing lists
21:44:24pamauryreally ? I can't remember seeing such a mail...
21:44:31CIA-7r28731 build result: All green
21:44:36pamaurybut I want to join :)
21:45:20AlexPpamaury: I've just tried to get, but it is taken :(
21:45:39AlexPAnyone any preference for what I should get?
21:46:31pamauryrockbox-dev ?
21:47:30AlexPI'll try that
21:48:36AlexPThat isn't available either
21:48:44AlexPwell, isn't
21:49:36pamauryrockbox-committers ?
21:49:46pamauryrockbox-rule-the-world ? :)
21:50:58AlexPhehe :)
21:51:57pamaurybertrik: I put some firmware information on this wiki page:
21:52:41AlexPOK, it is rockbox.calendar for now
21:59:19*pamaury starts the analysis of the third elf file of the ___ section of the fuze+ firmware
22:00:21 Quit komputes (Read error: Connection reset by peer)
22:01:46Rob2222are there any news regarding the "rockbox recognize some SD cards only sometimes in sansas" problem?
22:02:10 Quit Llorean (Changing host)
22:02:10 Join Llorean [0] (~DarkkOne@rockbox/user/Llorean)
22:08:18 Join csh_ [0] (~chatzilla@
22:10:40csh_hi again. i thought that since there is no fpu in mp3 players, can't we use fixed-point numbers instead of float and double? there are fixed-point arithmetic libraries for c and c++, and these libraries can do calculations of sin, cos, sqrt, etc.
22:12:53pamauryWho has access to it ?
22:14:14AlexPpamaury: gevaerts and I (and the Swedes in time) have access to edit, everyone has access to view
22:14:27AlexPpamaury: That is gevaerts and I as "release managers"
22:14:36pamauryhow do I access the calendar ?
22:14:48AlexPYou can add it to your gmail calendar
22:15:42AlexPYeah :)
22:15:57AlexPWe figured it is better to have to go via the release managers to change stuff
22:16:09saratogacsh_: yes, all rockbox codecs are in fixed point precision
22:16:14AlexPBut this way, everyone can see the schedule in advance
22:17:26csh_so, if i convert doubles and floats to fixed points, i can get enough performance from libmodplug, can't i?
22:18:04saratogacsh_: correct, that is the way codecs are ported to rockbox
22:18:58csh_excellent, on free times i will work on it. thanks a lot saratoga
22:19:04saratogano problem
22:19:14saratogatake a look at our existing codecs (or mod decoder)
22:19:21saratogait should give you an idea how fixed precision works
22:19:27saratogait is quite different then floating point
22:19:56csh_ok, firstly i will look at it. thanks
22:20:16 Quit csh_ (Quit: ChatZilla 0.9.86 [Firefox 3.6.12/20101026210630])
22:21:51 Quit t0rc (Quit: Give someone code, help them with one project. Teach someone to code, help them rule the world.)
22:22:06pamauryit seems there is some code duplication in the fuze+ init code
22:24:05saratogais there a patch somewhere for the code from last summer's TTS project?
22:25:22***Saving seen data "./dancer.seen"
22:27:25AlexPWas there any code?
22:29:30saratogawiki says he converted it to fixed point
22:31:19n1si think the memory alloc's that it (festival?) did turned out to be too unsuitable for conversion to static allocation so he basically gave up
22:31:30 Join ReimuHakurei [0] (~reimu@
22:31:34n1ssaratoga: make install did definitely make a zip at some point
22:31:49n1smaybe it's changed but then i don't see why it should be so slow
22:32:46pamauryany arm expert here ?
22:32:57*n1s points to Torne
22:33:08pamaurywhat does this mean: MCR p15, 0, R0,c7,c10, 4
22:33:30pamauryI'm not really familliar with the arm internals, appart from basic instructions
22:33:31AlexPBagder: about?
22:34:01AlexPPM :)
22:34:13Tornepamaury: that's a DSB
22:34:33Torneeverything on c7 is cache control
22:34:44pamaurywhat does DSB mean ?
22:34:50Tornec10, 4 is data sync barrier, or on older arms drain write buffer
22:35:30Torne(on older arms draining the write buffer was sufficient for a DSB)
22:36:05pamauryok thanks, I think I need to find a good doc on this or read the arm manual
22:36:34Torneit's all in teh ARM ARM
22:36:41Tornejust.. read an old version, if you have an old chip
22:36:44Torneit's less confusing that way
22:37:13*Torne recommends DDI0100E for ARM9 and DDI0100I for ARM11
22:37:28Tornethe ARMv7 arm is way harder to follow
22:37:39pamauryit's an arm9
22:37:51Tornerev E is easy to find on the internet, google for that number and pdf
22:38:07 Quit stoffel (Remote host closed the connection)
22:38:07Torneand that version is a mere 811 pages, not too uch to take in ;)
22:39:59Torneanyway, you need DSB when you're writing something to ram that you want to be visible outside the core proper, unless the memory is unbuffered
22:40:03n1sor maybe i'm wrong, the install target in root.make does call buildzip with "zip -r0" as an arg but i now realized that someone made make install do target builds too now, right?
22:40:28Torneit's just a writecombining buffer basically, but it's written back lazily so unless you DSB then other bus masters, or the ARM's own MMU, won't see the writes
22:40:34pamauryok thanks
22:40:34Tornewell, aren't guaranteed to see the writes.
22:40:53Torneyou basically do it whenever you do a cache clean
22:41:10pamauryactually there are a bunch of these things in the code at the beginning so I need to figure out by myself :)
22:41:11Torneexcept it also applies to memory that's mapped uncached but buffered
22:41:29TorneWell, it's probably using hte MMU
22:41:40 Quit designate72 (Remote host closed the connection)
22:41:48pamauryI haven't seen any mmu code so far but that might change
22:42:34Tornewhich core is it specifically?
22:42:40Tornei.e. is it one with an MMU or an MPU?
22:43:19Torneright, so that's an MMU
22:43:26Torneso it will use it
22:43:32Tornebecause otherwise you can't set up caching/buffering.
22:43:45TorneARMs that have an MMU don't have the protection unit's cache/buffer range registers
22:43:56Torneyou have to write pagetables to configure which memory is cached or bufferd.
22:44:28pamaurysounds sensible, then I'll have to find the code to see how the mmu is configured
22:44:43Tornelook for it setting cp15 register c2
22:44:47Tornethat's the base address of the pagetables
22:45:09Tornealso bit 0 in c1 turns the MMU on
22:45:16Torneone or both of those should be easy to spot :)
22:46:45Torneit's probably very near where you are :)
22:47:05Torneyou have to DSB after writing pagetables
22:47:23Tornethe MMU isn't hooked up to the write buffer so it doesn't see pending writes ;)
22:47:52n1ssaratoga: you are correct, the make install just sends in a zip command for fun but never uses it
22:48:36pamauryhum, I see lots of accesses to c7, c10
22:48:39pamaurybut none to c1
22:48:49pamauryor I missed one in the previous elf files
22:49:09 Join kadoban [0] (
22:53:56Torneis there a seperate bootloader/
22:54:00Torneit's possible the MMU init was done there
22:54:08Torneand this code just relies on it already being in a known config
22:54:45 Quit efyx (Remote host closed the connection)
22:54:50 Join TheLemonMan [0] (
22:55:00pamaurywe don't really know. There is probably a bootloader loading the .sb file; but the .sb file also contains some init code
22:55:39pamauryperhaps there is something in the imx23 doc :)
22:55:47TheLemonManthe .sb is bootstrapped by the onboard rom
22:56:45bertrikI was wondering if maybe the bootloader was in the i2c eeprom on board the fuze+
22:56:49pamauryActually, I don't think the rom does more than reading the sb file and executing the instructions, so I must have missed it
23:00:10TheLemonMansearching for addresses in that doc is just awful :|
23:00:35pamauryTheSeven: I find it quite easy on the contrary
23:00:41saratogaamsv2 is ARM926EJ-S so you should be able to reuse it's MMU setup
23:00:48pamauryyou have a memory map at the end and a register list
23:01:15pamauryTheSeven: you are looking at the fuze+ firmware ?
23:01:19TheLemonManyeah, but i need to get the register name and search it in the rest of document to know what bit does what
23:01:39pamauryhow do you think people do ? :D
23:02:33 Quit ReimuHakurei (Read error: Connection reset by peer)
23:02:48 Join ReimuHakurei [0] (~reimu@
23:03:07pamauryat which part of the firmware are you looking at ?
23:03:07TheLemonManit's pretty tiring :P and im not a hw geek too so i just reverse stuff w/o knowing what it does
23:03:43TheLemonManoh, im not looking at fuze firmware, im reversing my samsung player bootloader
23:09:30pamaurywhat would be the use of SSP at early boot stage ?
23:15:12tmztpower supply?
23:15:45bertrikor maybe communicate with the touchpad, turn off a led or something?
23:16:11 Join T44 [0] (
23:16:31pamaurystill can't tell, I haven't disassemble enough things, I just see that it does some intricate things to setup ssp :)
23:18:31saratogawhat was the reason we still have directory cache off by default?
23:18:39 Join ReimuHakurei_ [0] (~reimu@
23:18:41 Quit ReimuHakurei (Read error: Connection reset by peer)
23:19:03 Quit Topy (Ping timeout: 245 seconds)
23:19:06AlexPsaratoga: Something to do with it failing with large amounts of files?
23:19:11AlexPI can't remember
23:19:18AlexPI asked not long ago
23:19:49saratogahow large is large?
23:20:07bertrikpamaury, I'm seeing some difference in commands sent in sd-as3525.c and sd-as3525v2.c, do you know more about this?
23:20:31AlexPsaratoga: I can't remember
23:20:34pamaurynope, I haven't touch as3525 except for the usb core
23:20:40AlexPHVSC might have been involved
23:20:52bertrikfor example, for as3525, we try high speed mode only for v2 cards, while for as3525v2 it's done regardless of v1/v2
23:21:22tmztpamaury: ssp usually involves dma, especially for sound
23:21:30tmztpamaury: so it might just set it up early
23:21:46pamaurythat's at booting stage, I would be surprised
23:21:53pamauryfuck, that's fuze+ code doesn't make anysense, it's writing a reserved bit !
23:23:56 Quit ReimuHakurei_ (Read error: Connection reset by peer)
23:24:01 Join ReimuHakurei_ [0] (~reimu@
23:24:59Tornein which register?
23:25:10Tornevarious SoCs tend to extend ARM's CP15 stuff slightly
23:25:14saratogaAlexP: seems we agreed to enable it by default on the mailing list 6 months ago
23:25:43pamauryit's a clock register, not arm related as far as can tell
23:26:08AlexPsaratoga: I'm all for it; I just vaguely remember some counter reason last time I asked
23:26:14AlexPLt me check
23:26:23pamaurybut that's a shame, it set/clear a reserved bit from the start routine and I just can't tell what it's doing
23:26:47 Quit Keripo (Quit: Leaving.)
23:27:00TheLemonManwhat FM tuner the fuze+ has ?
23:27:32pamaurythe SansaFuzePlus page as a photo which says Si4703
23:29:40TheLemonManmine has a SI4703 on the I2C bus
23:30:25TheLemonManuhu its the same :D
23:31:00TheSevenany idea why ata transfer speed would scale almost linearly with the transfer size?
23:32:02 Quit ReimuHakurei_ (Read error: Connection reset by peer)
23:33:40 Quit evilnick_B (Quit: Page closed)
23:34:54AlexPsaratoga: From my log searching, various people just said that as far as they know dirchache fails if you have too many small files
23:35:02AlexPBut I think this was all IIRC
23:35:20AlexPI *think* I tested with the HVSC with no issues
23:35:33AlexPBut someone should check then enable IMO
23:35:46gevaertsI think I unpacked HVSC three times on my clip
23:38:16AlexPgevaerts: With success, or did dircche fail?
23:38:39gevaertsI *think* it worked
23:38:45gevaertshm, this may not have been the clip
23:38:53*gevaerts can't remember any details
23:38:58AlexPYeah, I also think it did when I tried with one copy
23:40:50pamauryagain writing to a reserved bit, there is something strange there
23:40:51gevaertsIt must be in the logs somewhere
23:40:51AlexPThis is it, nobody has any concrete don't enable it, just a few "I think it might not have IIRC" type things
23:40:51AlexPI say we enable it personally
23:43:39Torneyeah, llet's just do it
23:43:40Torneif it breaks something horribly we'll, er, fix it.
23:43:40*gevaerts finds things
23:43:40AlexPTorne: Aye
23:43:40gevaertsOK, apparently when I unpacked four copies of HVSC, dircache didn't work
23:43:48gevaerts(on ipod video)
23:43:57gevaertsTwo worked
23:44:44n1sgevaerts: does it fail gracefully?
23:44:44gevaertsn1s: well, I had to check if it was enabled
23:44:48gevaertsSo apparently, yes
23:45:38AlexPhow many files is 4 copies?
23:45:47n1swell, that sounds good
23:46:02AlexPyeah, 4 copies = stupid-amount-of-files
23:46:14AlexPand if it also failed gracefully, then just do it
23:47:49 Join marines [0] (

Previous day | Next day