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 2009-01-18

00:00:00 Join BigBambi [0] (n=Alex@rockbox/staff/BigBambi)
00:01:44 Join solexx_ [0] (
00:07:48 Join tajlero [0] (
00:11:56 Join {phoenix} [0] (
00:13:39 Quit tajlero ("Leaving")
00:14:37 Quit {phoenix} (Read error: 104 (Connection reset by peer))
00:18:34 Quit solexx (Read error: 110 (Connection timed out))
00:21:13 Nick JdGordon|zzz is now known as JdGordon (n=jonno@rockbox/developer/JdGordon)
00:22:27 Quit n17ikh ()
00:28:50 Join tyfoo [0] (
00:35:13 Quit gevaerts (Nick collision from services.)
00:35:24 Join gevaerts [0] (n=fg@rockbox/developer/gevaerts)
00:42:29 Join tajlero [0] (
00:46:09 Join ndundu [0] (n=47a77fb9@gateway/web/cgi-irc/
00:46:34ndunduI have a Sansa E250, and can't get the rockbox firmware, the gui and tthe ocmmand line tell me theres no sansa on the system = (
00:47:28 Join domonoky1 [0] (
00:48:17 Quit tajlero (Remote closed the connection)
00:48:29 Quit japc_ (Read error: 60 (Operation timed out))
00:50:30 Join japc_ [0] (
00:50:51 Quit tessarakt ("Client exiting")
00:52:04 Quit ndundu ("CGI:IRC")
00:52:20 Join tajlero [0] (
00:52:41 Join MethoS- [0] (
00:56:36 Quit MethoS (Read error: 60 (Operation timed out))
00:57:14soapin MSC or MTP mode?
00:57:25soapand, is it an E250 v1 or v2?
00:57:47soapie. what Sansa firmware is on it, nd (oh - he is gone)
00:58:26 Quit tvelocity[a] (Remote closed the connection)
00:58:56 Quit linuxstb (Remote closed the connection)
01:00:09 Quit culture (Connection timed out)
01:00:09 Quit parafin (
01:00:09 Quit freqmod_qu (
01:00:09 Quit suom1 (
01:00:09 Quit balou (
01:05:03 Join parafin [0] (
01:05:08 Join Monobro [0] (
01:05:24NJoinbalou [0] (
01:05:24NJoinfreqmod_qu [0] (i=quassel@2001:700:300:1430:213:d3ff:fee9:5ed0)
01:05:37MonobroQuestion: I have a birkced c250 and I figured I need the e200 eraser tool to fix it. anyone know where to grab it?
01:06:11 Join suom1 [0] (
01:07:49Monobroa sansa c250, naturally
01:08:13Monobroall forum posts on the topic are incomplete, either saying what you need with no links to it, or links to something without instructions
01:15:36 Join linuxstb [0] (n=linuxstb@rockbox/developer/linuxstb)
01:16:27krazykitMonobro, this is on the wiki
01:18:16 Quit tajlero ("Leaving")
01:18:44 Join tajlero [0] (
01:20:15 Quit Schmogel ("Miranda IM! Smaller, Faster, Easier.")
01:21:17Monobrokrazykit, could you show me where? I searched the page
01:21:40krazykitread the Manufacturing Mode section closely
01:22:38krazykitbut try doing recovery mode first, which is an easier process, and see if that doesn't fix it
01:23:38Monobrodon't see the eraser tool or any info on that
01:23:44Monobroi'll keep looking around though, thanks
01:26:04 Quit nplus (Remote closed the connection)
01:26:20 Quit Monobro (Read error: 131 (Connection reset by peer))
01:29:07JdGordondoes memcpy() crash if the src pointer is NULL?
01:31:19linuxstbThat would depend on whether the device has memory at 0x0...
01:36:14 Quit tajlero ("Leaving")
01:37:14 Join tajlero [0] (
01:42:21 Join n17ikh|Lappy [0] (n=n17ikh@
01:46:06 Join Horscht [0] (n=Horscht@xbmc/user/horscht)
01:47:34 Quit Horscht (Read error: 104 (Connection reset by peer))
01:47:41 Join Horscht [0] (n=Horscht@xbmc/user/horscht)
01:49:02amiconnUnhelpful: Imo codecs should also be converted to use a loader-initialized api pointer, for consistency
01:49:30***Saving seen data "./dancer.seen"
01:49:42amiconnThey also use mem wrappers iirc
01:55:10 Quit Slack_ ("Ex-Chat")
02:02:14 Join Slack_ [0] (
02:02:56 Quit tajlero ("Leaving")
02:05:59 Join itcheg [0] (i=62db4767@gateway/web/ajax/
02:16:05 Quit n17ikh|Lappy (Read error: 60 (Operation timed out))
02:16:27 Join n17ikh|Lappy [0] (n=n17ikh@
02:17:37 Quit dfkt ("-= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.")
02:19:11 Quit MethoS- (Remote closed the connection)
02:20:21 Quit n17ikh|Lappy (Client Quit)
02:20:43 Join Aurix_Lexico1 [0] (
02:23:34 Join Aurix_Lexico2 [0] (
02:26:28 Quit flydutch ("/* empty */")
02:33:22 Quit m0f0x (Read error: 104 (Connection reset by peer))
02:37:16 Quit Aurix_Lexico1 (Read error: 110 (Connection timed out))
02:37:27 Quit tyfoo ("Carpe diem")
02:37:36 Quit Aurix_Lexico (Read error: 110 (Connection timed out))
02:40:16 Join m0f0x [0] (
02:40:45 Quit timc`` (Read error: 104 (Connection reset by peer))
02:40:49 Join Aurix_Lexico [0] (
02:40:55 Join timc`` [0] (n=aoeu@
02:41:14 Join Darksair [0] (n=user@
02:41:59 Part domonoky1
02:47:52 Join Rob2223 [0] (
02:51:12 Quit Aurix_Lexico2 (Read error: 110 (Connection timed out))
02:57:14 Join Horschti [0] (n=Horscht@xbmc/user/horscht)
02:57:25 Quit Horschti (Read error: 104 (Connection reset by peer))
03:06:28 Quit Rob2222 (Read error: 110 (Connection timed out))
03:10:22 Quit Horscht ("Verlassend")
03:14:40 Quit film42 (Client Quit)
03:14:46Unhelpfulamiconn: it's on my list, as is moving mem wrappers to the lib. did we ever determine if the cache wrappers serve a legitimate purpose, or if they're just a convenience? i don't see any reason the ones in mpegplayer can't be replaced with rb->func calls, and that will be smaller and faster than calling a function that just calls rb->func
03:16:38 Join n17ikh|Lappy [0] (n=n17ikh@
03:22:26 Join Horscht [0] (n=Horscht@xbmc/user/horscht)
03:26:52 Quit ender` (" Documentation is like sex: when it's good, it's very good, and when it's bad it's still better than nothing.")
03:39:18 Quit timc`` (Remote closed the connection)
03:49:32***Saving seen data "./dancer.seen"
03:54:52 Join film42 [0] (
03:55:50 Join timc`` [0] (n=aoeu@
03:59:10 Quit goffa (Read error: 104 (Connection reset by peer))
04:05:06 Join Willwolfe [0] (
04:05:43 Quit Slack_ (Operation timed out)
04:06:19 Join Slack_ [0] (
04:08:10 Quit Slack_ (SendQ exceeded)
04:08:36 Quit film42 (Client Quit)
04:08:56 Join Slack_ [0] (
04:12:15 Quit PaulJam (".")
04:14:41 Join Makuseru [0] (
04:19:14 Join Barahir_ [0] (
04:26:02 Join Rob2222 [0] (
04:31:08 Quit Barahir (Read error: 110 (Connection timed out))
04:33:20 Quit Rob2223 (Read error: 60 (Operation timed out))
04:33:33 Join Rob2223 [0] (
04:37:51 Join tajlero [0] (
04:41:50 Quit itcheg (" ajax IRC Client")
04:44:20 Join kugel [0] (n=chatzill@unaffiliated/kugel)
04:44:52 Quit miepchen^schlaf (Read error: 60 (Operation timed out))
04:47:03tmztkugel: what version of svn should be used to build mkamsboot for a fuze to ensure that OF is bootable?
04:48:14 Quit Rob2222 (Read error: 110 (Connection timed out))
04:50:06 Join blkhawk- [0] (
04:55:20 Quit tajlero (Remote closed the connection)
05:01:23 Quit Aurix_Lexico ("Leaving.")
05:01:57 Join Aurix_Lexico [0] (
05:02:07 Join tajlero [0] (
05:05:58 Quit timc`` (Connection timed out)
05:06:40 Join timc`` [0] (n=aoeu@
05:06:45 Quit blkhawk (Read error: 113 (No route to host))
05:07:04 Nick blkhawk- is now known as blkhawk (
05:14:22 Quit Willwolfe ("ChatZilla 0.9.84 [Firefox 3.0.1/2008070208]")
05:19:02 Quit Aurix_Lexico (Read error: 110 (Connection timed out))
05:25:51 Join jhulst [0] (n=jhulst@unaffiliated/jhulst)
05:26:35 Quit XavierGr ()
05:49:11 Quit m0f0x ()
05:49:36***Saving seen data "./dancer.seen"
05:54:02 Join m0f0x [0] (
06:03:38 Join Aurix_Lexico [0] (
06:07:05 Quit timc`` (Read error: 145 (Connection timed out))
06:12:56 Quit sputnik (Read error: 110 (Connection timed out))
06:17:01 Quit Makuseru (Read error: 104 (Connection reset by peer))
06:21:49 Join pixelma_ [0] (n=pixelma@rockbox/staff/pixelma)
06:22:09 Quit amiconn (Nick collision from services.)
06:22:11 Join amiconn_ [50] (n=jens@rockbox/developer/amiconn)
06:23:54 Quit pixelma (Read error: 110 (Connection timed out))
06:26:42 Quit krazykit ("Connection reset by beer")
06:26:45 Join timc`` [0] (n=aoeu@
06:31:41 Quit tajlero ("Leaving")
06:34:31JdGordonanyone around who uses an ipod and an accessory which relies on the track change event? (I'm not actually sure we support any like this though)
06:41:41 Join sajes [0] (n=sajes@
06:41:57 Part sajes ("Wrong channel.")
06:46:57 Join _lifeless [0] (n=lifeless@
06:48:11 Join film42 [0] (
07:05:01 Quit __lifeless (Read error: 110 (Connection timed out))
07:20:28 Quit kugel (Read error: 110 (Connection timed out))
07:27:54 Quit Aurix_Lexico (Remote closed the connection)
07:35:40 Join Aurix_Lexico [0] (
07:47:34 Quit jhulst (Read error: 113 (No route to host))
07:49:38***Saving seen data "./dancer.seen"
08:02:16 Join fyrestorm [0] (
08:03:43 Quit timc`` (Read error: 104 (Connection reset by peer))
08:03:58 Join timc`` [0] (n=aoeu@
08:18:56 Quit film42 (Client Quit)
08:26:52 Join Rob2222 [0] (
08:30:48 Join japc__ [0] (
08:31:49 Join XavierGr [0] (n=xavier@rockbox/staff/XavierGr)
08:32:51 Join film42 [0] (
08:32:57 Quit Aurix_Lexico (Read error: 60 (Operation timed out))
08:42:33 Join Makuseru [0] (
08:42:45 Quit japc_ (Read error: 110 (Connection timed out))
08:45:27 Quit Rob2223 (Read error: 110 (Connection timed out))
08:45:53 Quit daurnimator (Read error: 60 (Operation timed out))
08:46:06 Join daurnimator [0] (
08:49:35 Quit XavierGr ()
08:51:55 Join Prot [0] (
08:52:29ProtQuestion: I'm getting 'permission for disc access denied!' from the rockbox utility with c250 sansa on vista/windows7. any times on how to get past it?
08:52:42Protapparently 1.07 doesn't have this bug, but the complete installation causes http 404
08:53:07Protany times? i guess I meant any tips
09:04:23 Join XavierGr [0] (n=xavier@rockbox/staff/XavierGr)
09:19:31 Quit Makuseru (Read error: 104 (Connection reset by peer))
09:23:15 Join Jaykay [0] (
09:34:29 Join Makuseru [0] (
09:46:25 Join flydutch [0] (
09:46:26 Nick amiconn_ is now known as amiconn (n=jens@rockbox/developer/amiconn)
09:49:02 Quit Makuseru (Read error: 104 (Connection reset by peer))
09:49:41***Saving seen data "./dancer.seen"
09:51:03 Quit n17ikh|Lappy ()
09:54:42JdGordonpixelma_: hey, any chance you could test 9795 on hwcodec today/tonight please?
10:10:00 Join bmbl [0] (n=Miranda@unaffiliated/bmbl)
10:14:42 Join Jaykay_ [0] (
10:19:40 Join Jaykay__ [0] (
10:20:49 Quit Jaykay__ (Client Quit)
10:33:40 Quit Jaykay (Read error: 110 (Connection timed out))
10:34:06 Quit Darksair ("People who are zhuangbility want to show their niubility but only reflect their shability.")
10:35:49 Quit film42 (Remote closed the connection)
10:36:35 Quit Jaykay_ (Read error: 110 (Connection timed out))
10:53:58 Join {phoenix} [0] (
11:04:37 Quit Horscht (Read error: 104 (Connection reset by peer))
11:06:14 Join n17ikh|Lappy [0] (n=n17ikh@
11:06:46 Quit n17ikh|Lappy (Client Quit)
11:10:03 Join ender` [0] (
11:15:19 Join Horscht [0] (n=Horscht@xbmc/user/horscht)
11:16:01 Quit BigBambi (Remote closed the connection)
11:17:28 Join tvelocity [0] (
11:31:23amiconnUnhelpful: Afaics the optimised scaling neglects proper rounding
11:34:31 Join n17ikh|Lappy [0] (n=n17ikh@
11:37:25 Join Horschti [0] (n=Horscht@xbmc/user/horscht)
11:42:04 Join Valerian [0] (n=chatzill@
11:43:04 Quit Horscht (Read error: 113 (No route to host))
11:43:10 Quit nuonguy ("This computer has gone to sleep")
11:45:20 Join MethoS [0] (
11:46:59 Quit Valerian (Client Quit)
11:48:59 Join miepchen^schlaf [0] (
11:49:44***Saving seen data "./dancer.seen"
11:53:29 Quit BHSPitLappy (Remote closed the connection)
11:57:37 Join os2user [0] (n=chatzill@
12:11:55 Quit XavierGr (Read error: 104 (Connection reset by peer))
12:19:29 Quit miepchen^schlaf ()
12:20:25 Join XavierGr [0] (n=xavier@rockbox/staff/XavierGr)
12:34:16 Join fragilematter [0] (n=barbu_do@
12:36:43jhMikeSUnhelpful: Cache wrappers hides the fact that some CPUs cache functions are not inlines. Really, I was considering removing inlining for all of them. They aren't used much and are fairly slow. Also, align_buffer out of the core...not sure. It seemed appropriate for it to be there.
12:37:10 Join Makuseru [0] (
12:38:14 Quit Makuseru (Read error: 104 (Connection reset by peer))
12:43:52 Quit MethoS (Read error: 104 (Connection reset by peer))
12:46:20 Join MethoS [0] (
12:54:11 Join MethoS- [0] (
12:55:08 Join tim__ [0] (n=aoeu@
12:55:09 Join t0mas [0] (n=tomas@rockbox/developer/t0mas)
12:56:10 Quit timc`` (Read error: 54 (Connection reset by peer))
12:59:30 Join MethoS-- [0] (
13:01:25 Quit MethoS (Read error: 145 (Connection timed out))
13:05:25 Join domonoky [0] (
13:13:12 Join Buschel [0] (
13:16:44 Quit MethoS- (Read error: 110 (Connection timed out))
13:17:20 Join PaulJam [0] (
13:18:38 Join Darksair [0] (n=user@
13:19:48 Nick pixelma_ is now known as pixelma (n=pixelma@rockbox/staff/pixelma)
13:20:31 Join MethoS [0] (
13:21:39 Nick Barahir_ is now known as Barahir (
13:33:49 Quit XavierGr (Read error: 104 (Connection reset by peer))
13:37:05 Quit MethoS-- (Read error: 110 (Connection timed out))
13:41:39 Nick fxb__ is now known as fxb (
13:42:55 Join gregzx [0] (
13:43:59 Join MethoS- [0] (
13:48:20 Quit MethoS (Read error: 60 (Operation timed out))
13:49:46***Saving seen data "./dancer.seen"
13:51:42 Join MethoS-- [0] (
13:52:27 Join culture [0] (
13:57:08 Join MethoS [0] (
13:57:31 Join archivator [0] (i=foobar@
14:05:46 Quit scorche|sh (Remote closed the connection)
14:05:49 Join scorche|sh [0] (
14:06:19 Quit blithe_ (Remote closed the connection)
14:06:30 Join blithe [0] (
14:06:50 Quit MethoS- (Read error: 110 (Connection timed out))
14:08:11 Join dfkt [0] (i=dfkt@unaffiliated/dfkt)
14:09:23 Nick JdGordon is now known as JdGordon|zzz (n=jonno@rockbox/developer/JdGordon)
14:10:44 Quit MethoS-- (Read error: 110 (Connection timed out))
14:15:27 Join itcheg [0] (i=62db4767@gateway/web/ajax/
14:18:14 Join tessarakt [0] (
14:20:02 Quit pixelma (" - Chat comfortably. Anywhere.")
14:20:51 Quit jhMikeS (Nick collision from services.)
14:20:57 Join jhMikeS [50] (n=jethead7@rockbox/developer/jhMikeS)
14:21:39 Join MethoS- [0] (
14:24:05 Quit parafin ("So long and thanks for all the fish")
14:24:07 Join parafin [0] (
14:25:02 Join pixelma [0] (n=pixelma@rockbox/staff/pixelma)
14:29:02 Quit MethoS (Read error: 145 (Connection timed out))
14:30:39 Quit JdGordon|zzz (Read error: 110 (Connection timed out))
14:33:28 Part Buschel
14:38:00 Join Schmogel [0] (
14:38:13 Quit Schmogel (Client Quit)
14:41:07 Join miepchen^schlaf [0] (
14:49:29 Quit mrkiko ("Lost terminal")
14:49:45 Nick n17ikh|Lappy is now known as n17ikh (n=n17ikh@
14:55:29 Join mrkiko [0] (
14:57:45 Quit Horschti ("Verlassend")
14:59:05 Join ibseo [0] (
15:01:03 Quit fragilematter ("Time to eat!")
15:04:31 Join gartral [0] (n=Gartral@
15:05:29gartraldi the H10 5 and 20 gig DAPs have different sized screens?
15:10:26mrkikoAre some rockbox-supported players again in the market?
15:14:52Unhelpfulamiconn: you mean for pictureflow? i can change those divisions to round to nearest. if you mean that it no longer dithers, as the core scaler does, that's actually deliberate. because the renderer is using point sampling (any better would probably be terribly slow), dithered output suffers from aliasing.
15:15:58amiconnI mean the addition, which is now missing
15:17:28 Quit HellDragon (Remote closed the connection)
15:17:55 Join robin0800 [0] (
15:20:52 Quit MethoS- (Read error: 104 (Connection reset by peer))
15:22:05 Join Seed [0] (
15:24:01CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
15:24:01*amiconn thinks that the old solution might even have been faster
15:24:15 Join MethoS- [0] (
15:24:20amiconnThe new one uses 64*64bit multiplies...
15:24:42 Join gregzx_ [0] (
15:24:55amiconnUnhelpful: What are the limits for ctx->divisor and ctx->round?
15:33:05Unhelpfulamiconn: if i read the asm correctly, <32> * <64> >> 32 is output as a umull, discarding one of the output registers
15:33:14Unhelpfuldivisor and round are 32-bit
15:34:02 Quit gregzx (Nick collision from services.)
15:34:05 Quit robin0800 (Remote closed the connection)
15:34:07 Nick gregzx_ is now known as gregzx (
15:34:08amiconnarm is not the only architecture this will run on
15:35:34amiconnAnd I know they are 32 bit - I was asking what range they actually use
15:40:13Unhelpfulamiconn: the scaler is already quite non-optimal on other architectures already, as that's the same math the scaler itself is using per-pixel.
15:42:28 Join kushal_12_27_200 [0] (n=kushal@
15:42:59 Join krazykit [0] (
15:43:39*Unhelpful can't figure out what that compiles to on sh-1 at all
15:44:03*amiconn is more concerned about coldfire, actually
15:45:11Unhelpfulthe other version already needed a 64-bit multiply, and was doing a 64-bit divide as well. sh-1 and coldfire lack a 32*32->64 multiply?
15:46:43Unhelpfulactually, it looks good to me on coldfire... if mulu is what i think it is?
15:47:08amiconnThere is mulu.w and mulu.l
15:47:36 Join Nico_P [50] (n=nicolas@rockbox/developer/NicoP)
15:47:46*amiconn would like to know the value range, which might offer further optimisation possibilities
15:48:23amiconnUnhelpful: Btw, there's now a settings bug in PF
15:48:35amiconnNo matter how often I disable "Show FPS", it is reenabled after a plugin restart
15:49:48***Saving seen data "./dancer.seen"
15:51:45Unhelpfulthe value is approximately (the last bit is zeroed, to avoid a 64-bit division) equal to (1<<32) / sf, where sf is the product of the input dimensions when scaling down, and the product of the output dimensions when scaling up
15:52:48Unhelpfuli'll have to look into the settings bug. i could kick myself over that, as i thought that i'd made sure TYPE_BOOL worked properly. :/
15:53:09 Join how [0] (n=kushal@
15:55:22 Join fragilematter [0] (n=barbu_do@
15:55:26Unhelpfulthis seems to say it's 32*32->64?
15:55:58amiconnNo, it's 32*32->32
15:56:17 Join BigBambi [0] (n=Alex@rockbox/staff/BigBambi)
15:56:19amiconn(mulu.l that is. mulu.w is 16*16->32)
15:57:01Unhelpfulthe asm output looks like the second form for mulu.l given there
15:57:16 Quit kushal_12_27_200 (Nick collision from services.)
15:57:21 Nick how is now known as kushal_12_27_200 (n=kushal@
15:57:52amiconnYes, the 3rd form doesn't exist for coldfire
15:57:58Unhelpfuli put just the multiply-shift in a function, it compiles to some moves and this: mulu.l 12(%a6),%d0:%d3
15:58:24Unhelpfulahhh... i must not be using the right arguments to gcc, i'll see which we actually use
15:58:35amiconnThis is a bad source of information; coldfire is a risc implementation of m68k and some addressing modes, forms, and even instructions don't exist on coldfire
15:59:27*amiconn recommends the coldfire programming reference manual
15:59:40amiconnIt's in our wiki iirc
16:00:27 Quit culture (Read error: 54 (Connection reset by peer))
16:01:14 Join culture [0] (
16:01:17Unhelpfuli'll have to check that later... it might be best, at least on some architectures, to go back to having a multiply in each scaler, if it's 16*16->32 instead of calling a 64-bit multiply function.
16:02:06Unhelpfuli'm pretty sure this means that pictureflow.c is already full of code that will work horribly on non-arm, by the way
16:02:46amiconnThat's why I was asking for the actual range used; I think it might be possible to avoid dealing with 64 bit ints, and avoid the division at the same time
16:03:36 Join robin0800 [0] (
16:14:15 Join Aurix_Lexico [0] (
16:28:39 Join tyfoo [0] (
16:28:52 Quit kushal_12_27_200 ("Leaving")
16:32:46 Join stripwax [0] (
16:38:54 Join axionix_ [0] (
16:40:45 Join arohtar [0] (
16:49:30 Quit stripwax ("Miranda IM! Smaller, Faster, Easier.")
16:52:16 Quit axionix (Read error: 110 (Connection timed out))
16:56:03 Join efyx [0] (
16:57:46 Quit faemir (Read error: 110 (Connection timed out))
16:58:01 Quit tyfoo (Read error: 60 (Operation timed out))
17:08:57 Join axionix [0] (
17:11:14Unhelpfulamiconn: if we shift values right by 8 places, it can use an unsigned 32x32->32... do we have *those* everywhere?
17:12:05amiconnNo, SH1 only has 16*16->32
17:12:26amiconnBut even then, 32*32->32 is better than anything that involves 64 bit ints
17:13:49amiconnOn SH1, 32*32->32 is a relatively simple libgcc routine (in asm), while the 64 bit multiplication routines are rather crappy
17:22:34amiconnUnhelpful: I know why your TYPE_BOOL doesn't work properly
17:22:52 Quit axionix_ (Read error: 110 (Connection timed out))
17:22:54amiconnYou cannot cast between int and bool and expect it to magically work
17:23:38amiconnThey're different, incompatibel types. Depending on architecture and endianess it *may* work, but it is pure coincidence and never guaranteed
17:24:54*amiconn is looking into the issue
17:26:17amiconnHehe, "%s: 10%d\n" is a funny format string...
17:29:51Unhelpfulamiconn: so, it needs boolvar ? 1 : 0 to convert to int, and intvar ? true : false to convert to bool, then?
17:29:55Unhelpfulthat's a quick fix
17:30:09amiconnI already have a fix, just need to test
17:30:23amiconnIt will write "on" or "off" for booleans like the core config
17:30:49 Join fredddy [0] (
17:30:52Unhelpfulamiconn: the trouble with that is that many things are using ints for booleans now
17:30:58 Nick fxb is now known as fxb__ (
17:31:48amiconnI actually think the bug was the wrong format string
17:32:13amiconnBut anyway, if we have a bool type, using "on"/ "off"is better imho
17:32:58amiconnExplicit and implicit casts between int and bool do work; what's doesn't work is casting pointers and then expect the read/written value to be correct
17:34:25Unhelpfulthe problems with on/off are: 1) it would be nice to convert the things that use ints to bools, when they get ported away from oldmenuapi, and 2) mpegplayer at least expects to be able to call set-one and get-one functions that use an int
17:35:18amiconnThe settings are broken anyway now
17:35:25Unhelpfulpictureflow is the only thing using bools for on/off settings. others are abusing ints.
17:35:35 Join fdinel [0] (
17:37:44amiconnUnhelpful: My pictureflow.cfg: zoom: 100 / show fps: 101 / resize: 101
17:39:00Unhelpfulright... that's because of the typo. fixing the typo should fix the bug, and still allow things that use ints now to use bool and core set_bool to work with their old config files.
17:39:40*amiconn doesn't understabd
17:39:46 Quit agaffney (Read error: 104 (Connection reset by peer))
17:39:52 Quit fragilematter ("Out for today. Exam tomorrow...")
17:40:02Unhelpfulamiconn: many plugins use 0/1 and TYPE_INT for on/off settings.
17:40:13 Quit itcheg (" ajax IRC Client")
17:40:32amiconnI know, because TYPE_BOOL is new
17:40:32Unhelpfuland expect to be able to write or read just one of these settings...
17:40:43 Join agaffney [0] (n=agaffney@gentoo/developer/agaffney)
17:40:50amiconnI don't see the problem
17:41:07Unhelpfulright, but if TYPE_BOOL expects to find "on" or "off" in config, won't it break old configs for these plugins?
17:41:45amiconnIt will, once, when the plugin is converted to use TYPE_BOOL
17:42:21amiconnMy solution just compares the string with "on", everything else than "on" means false
17:42:40Unhelpfulalso, mpegplayer uses configfile_update_entry and configfile_get_value, which expect settings to be fixed-size - but perhaps those just need to go away, and it should write the whole config?
17:43:02Unhelpful_get_value already has to read up to the line it's looking for, anyway.
17:43:10amiconnFixed size? Who added that??
17:43:22amiconnImo this is just wrong and needs to be removed
17:43:25Unhelpfulamiconn: that's what all the %10d is.
17:43:47amiconnBack when I checked configfile.[ch] last, this wasn't there
17:44:15amiconnThe config files are meant to be plain assci, one setting per line, with no other formatting requirements
17:44:40Unhelpfulcould it perhaps recognize "1" or "on" as true, anything else as false? that would make existing configs continue to work.
17:45:45amiconnThat would be possible
17:45:52linuxstbamiconn: Hmm, seemed that sneaked in with an mpegplayer commit...
17:45:56Unhelpfulor things that convert could bump their config versions
17:46:01amiconnStill, this silliness with constant widths needs to be fixed first
17:46:28Unhelpfulbetter they reset to defaults than read in broken values
17:47:42 Quit agaffney (Read error: 104 (Connection reset by peer))
17:47:50 Join agaffney [0] (n=agaffney@gentoo/developer/agaffney)
17:48:19Unhelpfuli'll let you go ahead if you have a working "on"/"off" version. we can just reset defaults when loading old config files, using the file version. that's better than making pre-TYPE_BOOL settings always false on read
17:48:57amiconnThe problem with on/off is that it breaks constant width, so that needs to be fixed first
17:49:12amiconnI was totally unaware of that
17:49:15rasherlinuxstb: Specifically here:
17:49:50***Saving seen data "./dancer.seen"
17:55:19 Join Horscht [0] (n=Horscht@xbmc/user/horscht)
17:56:44 Quit ibseo ("quit")
17:57:00 Quit Horscht (Client Quit)
17:57:26 Join Horscht [0] (n=Horscht@xbmc/user/horscht)
17:59:36amiconnUnhelpful: For the mpegplayer resume points constant width is obviously needed because they can be updated. But since there are also arbitrary many, they aren't part of the config array
18:00:16amiconnSo I think they should get their own type (e.g. TYPE_DYN_INT), and the number should be zero-padded instead of space-padded
18:00:51 Join tvelocity[a] [0] (
18:01:11 Quit tvelocity (Connection timed out)
18:03:07 Quit Horscht ("Verlassend")
18:03:35 Join Horscht [0] (n=Horscht@xbmc/user/horscht)
18:04:26amiconnHmm, or the resume points could get their own file
18:06:22Unhelpfulthat seems sensible, anyway - they're not "configuration"
18:07:10Unhelpfulif you're looking to do something more complex, then, should i commit fix for the printf typo for now, to at least make it work until then?
18:07:13 Join EleminoP [0] (
18:07:16 Part EleminoP
18:09:03 Join kugel [0] (n=chatzill@unaffiliated/kugel)
18:11:14Unhelpful*a* fix
18:11:39amiconnCheck the front page...
18:12:55Unhelpfuli see that. thanks :)
18:14:54 Quit Darksair (Remote closed the connection)
18:15:10 Join BXCracer [0] (
18:37:26 Join itcheg [0] (i=62db4767@gateway/web/ajax/
18:38:08 Join tyfoo [0] (
18:50:28 Quit arohtar (Client Quit)
18:50:41 Join faemir [0] (
18:52:13 Join z35 [0] (
18:58:27 Quit Nico_P (Remote closed the connection)
19:02:26 Quit robin0800 (Read error: 60 (Operation timed out))
19:02:39 Join petur [50] (n=petur@rockbox/developer/petur)
19:03:32 Join nuonguy [0] (
19:05:19 Join Dauron [0] (
19:07:38 Quit BXCracer (Read error: 110 (Connection timed out))
19:11:18 Join ToHellWithGA [0] (
19:12:48ToHellWithGAi read that the chips in the newer ipods are nasty and make it more or less impossible to get rockbox on such a device
19:13:02ToHellWithGAwas that ever the case with previous generations of ipods?
19:13:39 Quit itcheg (" ajax IRC Client")
19:14:11ToHellWithGAi'm interested in the history of this project so i can compare it to the advancement of specialized chips
19:15:04 Join saratoga [0] (n=9803c6dd@gateway/web/cgi-irc/
19:15:21saratogaits not the chips, its that the firmware is encrypted
19:15:39kugelToHellWithGA: the chips aren't the problem. the encryption is, which makes it impossible to run custom code
19:15:54kugellol saratoga, you're blazing fast
19:15:56ToHellWithGAso y'all would be up against brute forcing encryption
19:16:02ToHellWithGAwhich is illegal, right?
19:16:47saratogaits not illegal just impossible
19:17:06ToHellWithGAi wonder what Apple feels it must protect. the reasons i purchased my iPod were the sleek look and the availability of Rockbox
19:17:14ToHellWithGAreally impossible?
19:17:29ToHellWithGAa kershmillion bit key or something as awful
19:17:40kadobanbruteforching decently implemented cryptography is basically impossible, yes
19:17:48 Quit gregzx (Read error: 104 (Connection reset by peer))
19:18:02ToHellWithGAi should hope they're far from decent then
19:18:09ToHellWithGAthanks for explaining that to me
19:18:13faemirYou know, I wouldn't even bother. I compared the sound of my 5.5 gen ipod with my friend's cowon iaudio7 and there is such a big different it's not even funny :(
19:18:39saratogawe're talking about the 6g, the 5.5 already runs rockbox
19:19:21ToHellWithGAfaemir: so long as the storage, ease of use, and aesthetics are on the same level I would be fine with switching
19:19:25faemirYes, but the 6g doesn't sound better than the 5.5g :P
19:19:44faemir(my point is I won't even bother getting an ipod next time due to sound quality)
19:19:54faemirlosing rockbox will be fail though :(
19:19:56ToHellWithGAperhaps as data density increases i'll be able to swap the hdd from my ipod
19:20:08ToHellWithGAif its controller can handle larger disks, that is
19:20:18faemirI need a new battery, my ipod one is being dodgy now...
19:20:31faemirToHellWithGA, try the cowon s9 ;)
19:20:38ToHellWithGAmy girlfriend's ipod mini will turn off fully at random
19:20:43 Join danpeterson [0] (
19:20:52rasherThis is going off topic fast
19:20:53krazykitplease try to stay on topic, as this IS a rockbox channel
19:20:56faemiryeh, mine randomly goes down to 0% and I have to leave for 5 mins to go back to normal
19:20:59faemirsorry :(
19:21:01ToHellWithGAsorry y'all
19:21:16faemiroh epic 3.1 is released
19:21:22 Part ToHellWithGA ("Oh no!")
19:21:51kugelsaratoga: any success on the compiler side? I read the logs and you seem to have some problems?
19:22:08kugelI'm really curious that a binutils release doesn't work for arm
19:22:55 Quit krazykit ("Connection reset by beer")
19:23:02saratogakugel: i think its just the gcc version ubuntu uses that can't compile it for arm
19:23:12saratogabut theres a simple patch for it
19:23:23 Quit _Auron_ (Read error: 110 (Connection timed out))
19:23:24 Nick Dauron is now known as _Auron_ (
19:23:25kugelah cool
19:24:14saratogabasically the compiler i got doesn't work and i'm not sure why
19:24:29saratogai need to try a different binutils version to rule out the possibility that the snapshot build i grabbed was broken
19:27:00kugelsaratoga: I think the two patches here can be committed
19:32:41saratogakugel: does the button light not work now on the AMS models?
19:33:20saratogadoes the fuze have a button light?
19:35:45kugelthe buttonlight doesn't work in svn
19:35:49saratogaperhaps I broke mine when swapping the wheel
19:36:01saratogashould it be on when in the OF USB mode?
19:37:18kugelor wait
19:38:02 Join gregzx [0] (
19:38:05kugelyes, it's on here
19:39:38saratogathe fuze has easily the least friendly internal construction of any mp3 player
19:40:13 Quit _Auron_ (Read error: 54 (Connection reset by peer))
19:41:24saratogakugel: fading looks good on my fuze
19:44:09 Join _Auron_ [0] (
19:45:42kugelsaratoga: yea, I thought so too :)
19:47:06 Nick fxb__ is now known as fxb (
19:47:16 Join BHSPitLappy [0] (n=BHSPitLa@unaffiliated/bhspitmonkey)
19:47:26saratogakugel: did you get anywhere with the MMU?
19:47:58 Quit danpeterson ("ChatZilla 0.9.84 [Firefox 3.0.5/2008120122]")
19:48:10kugelsaratoga: not yet, I need to have some chat with someone who knows how it works on F/X
19:49:04kugelI didn't have time the past few days, but the progress I had was to get the bootloader booting, it hangs at loading the main though
19:49:06 Join saratoga2 [0] (n=9803c6dd@gateway/web/cgi-irc/
19:49:47 Join itcheg [0] (i=62db4767@gateway/web/ajax/
19:49:51***Saving seen data "./dancer.seen"
19:51:19saratoga2kugel: ok just curious
19:51:34saratoga2if the F is doing things differently then the AMS code, then the AMS code is probably wrong
19:51:41saratoga2maybe this will fix some of our playback issues
19:52:44 Quit saratoga ("CGI:IRC (Ping timeout)")
19:53:21kugelsaratoga2: exactly
19:54:03saratoga2at very least its probably going to break half our drivers because they implicitly assume that the CPU is slow as hell :)
19:54:27kugelI copied the mmu setup from F/X-specific crt0.S into the one that AMS uses
19:55:40kugelI don't think the cpu will be faster, but accessing RAM will be
19:55:52 Join TheSkunkMan [0] (
19:56:43saratoga2kugel: without the MMU theres no icache right?
19:56:52kugelit should be in fact faster, with the icache
19:57:02saratoga2a lot faster
19:57:07saratoga2assuming its really off
19:57:16kugelI think it is
19:57:21kugelthe ds indicates
19:57:46 Join nibbler [0] (
20:01:53 Quit itcheg (" ajax IRC Client")
20:02:28kugelsaratoga2: crt0.S is also executed when loading the main binary?
20:03:04kugelamiconn said I may miss to flush & invalidate the caches before executing it, but the crt0.S does that iiuc
20:04:23saratoga2kugel: I'm not sure, the details of how rockbox works are beyond me
20:04:52saratoga2at least at the firmware level
20:04:55kugelhehe, same here, but I'm learning slowly
20:06:06 Join PaulJam_ [0] (
20:06:23 Quit PaulJam (Read error: 145 (Connection timed out))
20:06:30saratoga2kugel: did you ask kkurbjun about the F?
20:06:58saratoga2i'll ping him
20:07:11 Join itcheg [0] (i=62db4767@gateway/web/ajax/
20:09:56kugelah, he seems to be the man to ask
20:11:10 Join goffa [0] (n=goffa@
20:13:52 Join karashata [0] (n=karashat@
20:13:57 Join tusjen [0] (
20:19:27 Join bertrik [0] (
20:20:41 Join axionix_ [0] (
20:21:22kugeljhMikeS: ping
20:23:55 Quit os2user (Remote closed the connection)
20:26:00 Quit itcheg (" ajax IRC Client")
20:27:23 Join troy_ [0] (n=toppy@
20:28:16 Quit Slack_ (SendQ exceeded)
20:29:18 Join Slack_ [0] (
20:34:40 Quit axionix (Read error: 110 (Connection timed out))
20:35:09 Quit troy_ (Read error: 104 (Connection reset by peer))
20:39:45 Join {-phoenix-} [0] (
20:40:42 Part tusjen
20:43:29 Quit Slack_ (SendQ exceeded)
20:46:17TheSkunkManhow does someone typically discover usb bootloader mode on a device? do they just try different combinations of buttons as the player is turning on?
20:46:58 Quit fredddy (Remote closed the connection)
20:49:30LloreanTheSkunkMan: Sometimes they're just explained by the device manufacturer for use when something goes wrong
20:50:21 Quit FOAD (Remote closed the connection)
20:51:44 Join Slack_ [0] (
20:53:04 Quit {phoenix} (Read error: 110 (Connection timed out))
20:53:19 Join FOAD [0] (
20:55:44 Quit saratoga2 ("CGI:IRC (EOF)")
21:00:40 Join JdGordon|zzz [0] (n=jonno@rockbox/developer/JdGordon)
21:01:11 Nick JdGordon|zzz is now known as JdGordon (n=jonno@rockbox/developer/JdGordon)
21:01:40 Join jhulst [0] (n=jhulst@unaffiliated/jhulst)
21:03:34 Quit Slack_ (SendQ exceeded)
21:04:14 Join Slack_ [0] (
21:04:15 Join pisicalanai [0] (n=toma@
21:04:43 Part pisicalanai
21:05:25jhMikeSkugel: pong
21:07:15kugeljhMikeS: hey, I've found out that the cache+mmu setup is exactly the same on the AMS sansas as on F/X
21:07:43kugelbut copy and paste wasn't enough, so I miss something, maybe you can help me?
21:07:57jhMikeSOf they use a similar v4 core I would be surprised.
21:08:01kugel(bootloader works, but main build doesn't execute)
21:08:40jhMikeSdoing memmapping can be a chore but IRAM can be helpful there and I think there's plenty, no?
21:09:59jhMikeSif the main build starts executing from a virtual address but performs remapping itself, that could hang it
21:10:28kugelwe have icache and dcache (8KB each) and 320KB IRAM
21:11:18kugelour bootloader is in IRAM, but the main build is in SDRAM (of course), could that cause problems?
21:11:38jhMikeSit has no mcrr instruction or do you need the rather manual cache range functions as on F/X?
21:11:53kugelthe latter
21:12:22jhMikeSDoes the main binary start executing at 0 or the physical address?
21:12:23 Join Willwolfe [0] (
21:12:37kugelagain, it's exactly the same. Only cache sizes differ between arm922T and arm920T
21:12:50 Join saratoga [0] (n=9803c6dd@gateway/web/cgi-irc/
21:12:54kugelI'm not exactly sure
21:13:29kugelI think it's using the physical adress
21:13:40jhMikeSI haven't really looked at the F/X code too closely. It should probably hop to physical addresses before doing anything with the MMU.
21:14:07saratogasince we start with the MMU disabled, I would expect physical address
21:14:31jhMikeSbut what about the start vector?
21:15:25kugelloadaddress = 0x30000000;
21:15:42kugeli.e. physical address
21:16:09jhMikeSyes, vectors are also relative branches by default and should be relocatable
21:16:31saratogainterrupt vectors?
21:16:48WillwolfeIs there a reason that a build on the clip should behave differently/install differently on reinstalls?
21:17:27 Join Davekong [0] (n=Davekong@
21:17:51jhMikeSsaratoga: the default vectors at 0x00000000
21:18:02 Quit archivator ("<>")
21:18:09kugelIRAM is mapped to that address
21:18:35kugeljhMikeS: what I'm wondering too is: what is TTB_SIZE?
21:18:55jhMikeSTTB_SIZE = size of remapping table
21:19:07saratogajhMikeS: since we don't need address translation, what exactly does turning the MMU on do?
21:19:16saratogai just assumed we could leave it off and keep using physical memory
21:19:35jhMikeSthe firmware image is to run from 0x30000000?
21:20:01kugelsaratoga: that's what we're doing right now. but it seems we cannot use caches without mmu
21:20:09kugeljhMikeS: yes
21:20:28jhMikeSDoesn't the core expect the vectors to be at 0x00000000 to handle abort mode et al?
21:21:46jhMikeSI imagine you can remap with the same address, only problem is then obtaining a physical address translation for devices to use
21:22:19 Quit nuonguy ("This computer has gone to sleep")
21:22:23jhMikeSThe way the memory is mapped might not be compatible then (though I'm not sure).
21:23:14kugeljhMikeS: I think F/X runs from 0x30000000 too
21:23:41kugelunless I got the code wrong
21:23:50jhMikeSDRAMORIG is still 0x00000000
21:24:18jhMikeSIt starts off at 0x30000000, remaps phys to virt and then I believe jumps to the virt address and continues executing.
21:25:15kugelso that could be the first problem
21:25:27kugel this is what I have now
21:25:40kugelshould we do that too?
21:26:09jhMikeSSee crt0.S line 420 for s3c2440
21:26:54kugelah that maps 0x300.. to 0x0?
21:27:09kugelI have copied that part too, so I (accidentely) did that too
21:27:30kugelwhich, of course, can't work
21:27:32jhMikeSIt makes sure write buffering is on for non-cached sections as well (for framebuffer, _huge_ speed benefit).
21:29:29WillwolfeTried a Windows filecheck/defrag on the clip drive. Still paniced/rebooted after a couple seconds. Would play MP3. Crash on FM. Ran disktidy plugin. Things are OK again.
21:29:36kugeldo we have to pay special attention on IRAM or cache sizes?
21:30:11kugelthe iram is unused on fuze and e200v2, but used for codecs on the 2MB ones, if that matters
21:30:46*gevaerts suspects that Willwolfe is seeing the 1GB limit issue
21:31:02saratogajhMikeS: is memory paged or do the virtual addresses just directly map to physical addresses in a linear fashion?
21:31:14kugelgevaerts: on the clip?
21:31:20gevaertskugel: yes
21:31:38kugelI didn't know the clip suffers from that too
21:31:39WillwolfeI thought it did not apply to the clip?
21:31:45jhMikeSsaratoga: It's done as flat on F/X.
21:31:47gevaertsIt does (I have it)
21:32:01kugelmy clip doesnt
21:32:12jhMikeSsaratoga: It can be done non-flat. Gigabeat S does so.
21:32:23saratogais it done nonflat?
21:32:26gevaertsAt least I think I have it. I haven't seen any issues since I reformatted it to 950MB or so
21:32:26saratogaoh sorry
21:32:30saratogamissed your previous line
21:32:44saratogakugel: how big is your clip?
21:32:59saratogai think the limit is 2GB for the 2GB model
21:33:01kugelmaybe the 4GB suffers from that
21:33:02saratogaand 1GB for the regular
21:33:08gevaertsMine is 2GB as well
21:33:29*gevaerts doesn't disagree
21:33:32saratogai believe the 2GB can access the whole disk, at least thats what funman implied
21:33:33WillwolfeMine is a 2 GB model
21:33:33kugelbut, my clip has had a lot of problems with the filesystem
21:33:41kugelI needed to reformat a zillion times
21:34:04saratogaalthough i suppose its possible theres clips with different physical arrangements of Flash banks out there
21:35:31WillwolfeI havn't actually reformatted as such. But I can get a completely different behavior with simply reapplying a build.
21:35:47 Quit Horscht ("Verlassend")
21:36:03 Join Horscht [0] (n=Horscht@xbmc/user/horscht)
21:36:23kugeljhMikeS: the thing is, this is all new to me, and I have hardly any idea how this memory mapping and stuff works
21:36:27saratogaif you've got it formatted wrong that will happen
21:36:36kugelso I could need some assistance
21:40:51 Join film42 [0] (
21:41:44 Quit tim__ (Read error: 54 (Connection reset by peer))
21:41:58 Join tim__ [0] (n=aoeu@
21:42:19jhMikeSkugel: It makes table of address translations SRC->DST. There's various formats for the entries described in the docs. Granularity here is 1MB but it can be done down to 4KB.
21:43:37WillwolfeFAT, 848 MB used, 1.04 GB free on my clip
21:48:13kugelI'm reading the ds a bit and spot: You must only enable the DCache when the MMU is enabled.
21:48:32kugeland before that ICache can be enabled without MMU
21:49:46 Join Jaykay [0] (
21:49:53***Saving seen data "./dancer.seen"
21:50:30jhMikeSyes, remapping is possible, then caches
21:52:01jhMikeSI imagine it won't know what to cache and not without having the MMU telling it what is cacheable from the flags in the table entries
21:55:08 Quit {-phoenix-} ()
21:55:19 Join {-phoenix-} [0] (
21:55:50jhMikeSI suppose flush_icache/invalidate_icache are also the same as for s3c? Those will also need implementation
21:56:30kugeljhMikeS: yep
21:56:53 Quit gromit`` (Read error: 110 (Connection timed out))
21:57:18kugeljhMikeS: I thought the ones in mmu-arm.c can be used
21:59:51kugelF/X use those too
22:00:05jhMikeSkugel: see /firmware/target/arm/s3c2440/gigabeat-fx/system-target.h
22:00:20kugeloh that's only dcache
22:00:50jhMikeSIt shouldn't be, it should include both
22:01:40 Join gromit` [0] (
22:01:46jhMikeSNope, it's the icache invalidate
22:02:05jhMikeSaka. "Invalidate entire instruction cache"
22:02:40jhMikeSIt might be missing "Flush prefetch buffer"
22:03:50jhMikeSoops, my's got clean_dcache and I missed that :p
22:04:37jhMikeS'mcr p15, 0, %0, c7, c14, 0' (clean and invalidate entire dcache) isn't valid there?
22:04:50WillwolfeSaratoga: How would you recommend reformatting a clip?
22:06:01kugeljhMikeS: I think it is
22:06:02 Quit Jaykay ("ChatZilla 0.9.84 [Firefox 3.0.5/2008120122]")
22:07:01 Nick JdGordon is now known as JdGordon|afk (n=jonno@rockbox/developer/JdGordon)
22:09:52WillwolfeSaratoga: Meaning should I format using a windows utility or the OF format?
22:13:36kugeljhMikeS: Invalidate ICache and DCache "MCR p15,0,Rd,c7,c7,0"
22:14:04kugelClean and Invalidate DCache "MCR p15,0,Rd,c7,c14,1"
22:16:28 Quit bmbl ("Woah!")
22:17:06 Join timc`` [0] (n=aoeu@
22:18:11 Join tajlero [0] (
22:18:16jhMikeSkugel: either I made an error or they changed it for ARM1136. I really don't remember now. hehe.
22:19:08kugeljhMikeS: the F/X implementation looks correct to me
22:20:11kugeljhMikeS: oh and 'mcr p15, 0, %0, c7, c14, 0' is valid, but only to c&i a single entry, not the whole
22:20:22kugelmaybe they changed it
22:20:26jhMikeSthat just flushes the dcache to ram without invalidating it
22:20:44 Quit Willwolfe ("ChatZilla 0.9.84 [Firefox 3.0.1/2008070208]")
22:20:54kugelClean and Invalidate DCache entry (using index)
22:22:06 Quit faemir (Read error: 104 (Connection reset by peer))
22:22:16kugel(from the ds)
22:23:10kugeljhMikeS: so, it looks alright to use mmu-arm.c and the invalidate_icache() from F/X
22:23:11jhMikeSclean_dcache won't invalidate
22:23:32kugelthat's correct
22:24:05kugelah, you mean invalidate_icache() should also do invalidate_dcache() (not only clean_dcache())?
22:24:19 Join Thundercloud [0] (
22:24:23jhMikeSthat's what's expected of it
22:24:39 Quit karashata ("G'bye everyone!")
22:25:17 Join yhuang [0] (
22:25:43 Quit t0mas ("bye bye")
22:26:07jhMikeSthose really should be renamed to invalidate/flush_cache. I think the naming is a holdover from coldfire which only has an icache.
22:27:15 Part Davekong ("Leaving")
22:27:59 Join Rondom [0] (
22:28:29jhMikeSUnhelpful was having frustrations about the cache function wrappers. Really none of those functions need to be inlined and then the wrappers can go away.
22:29:32jhMikeSPP has ones that are too large for inlining and so they are not there.
22:34:23 Quit tim__ (Connection timed out)
22:35:57gartrali have a few files on my DAP that cause it to abort at 000090D0 (0), the files are ultra high quality MP3 ABR of 300 and at 48 kHz sample rate, the DAP is a Sansa E250 v1, running latest bootloader and official build
22:47:35kugelsaratoga: ok, the icache is probably enabled
22:47:53kugelI notice a big difference if I don't enable it
22:49:35gartralive uploaded one of the songs too a drop, since its "protected" by the dmca, i cant post the link in the open...
22:50:54kugelgartral: most data aborts are related to unproper tags from my experience
22:52:40kugelUnhelpful: test_boost is broken
22:52:45gartralkugel: does a completly untagged file break?
22:53:07kugelI don't think so
22:56:54 Quit Rondom ("Ex-Chat")
22:57:01gartraland the weird part is that the files will play for about 30s and then RB aborts at the address i gave you
22:59:44 Quit petur (Remote closed the connection)
23:00:36 Join krazykit [0] (
23:02:37 Join itcheg [0] (i=62db4767@gateway/web/ajax/
23:03:22linuxstbgartral: Have you tried the files in the UI simulator? Do you have any similar files that work?
23:03:44 Join dberg918 [0] (
23:06:12 Join ibseco [0] (
23:10:02 Join pixelma_ [0] (n=pixelma@rockbox/staff/pixelma)
23:10:08 Quit amiconn (Nick collision from services.)
23:10:10 Join amiconn_ [50] (n=jens@rockbox/developer/amiconn)
23:10:40 Quit pixelma (Read error: 110 (Connection timed out))
23:11:30 Nick amiconn_ is now known as amiconn (n=jens@rockbox/developer/amiconn)
23:12:24amiconnkugel: Doing flush & invalidate in crt0.S of the main binary is too late. If the bootloader enables caches, it needs to flush & invalidate them after loading the main binary but *before* calling it
23:12:32amiconnThe alternative would be to not enable caches in the bootloader, but only in the main binary. That would remove the need for flush & invalidate in the bootloader. It still needs that in the main binary for a working rolo implementation, and also for loading plugins, codecs etc
23:12:36amiconn(actually with the exception of codec on the lowmem as3525 targets, as codecs reside completely in iram on those)
23:13:52kugelamiconn: I tried invalidate_icache() right before kernel_entry() in the bootloader, without success
23:14:30kugel(invalidate_icache() also flushes&invalidates dcache)
23:14:33amiconnYou need to flush the data cache first, then invalidate the icache, and then call the main binary
23:17:12kugelwell, that didn't work
23:17:20kugelbut I think the problem is with the mapping
23:17:33kugelI didn't get that right yet
23:17:33 Join tvelocity [0] (
23:19:08 Quit {-phoenix-} (Read error: 104 (Connection reset by peer))
23:21:08gartrali can update my sims build with any, as long as its built for the target the sim is enulating, right?
23:21:20 Join nuonguy [0] (
23:21:54amiconnThe simulator is a simulator (sic!), not an emulator
23:22:22amiconnPlugins, codecs etc need to be built for the simulator
23:23:53jhMikeSThe simulator emulates a simulator or simulates and emulator?
23:24:05 Quit tvelocity[a] (Read error: 110 (Connection timed out))
23:24:36 Quit tajlero ("Leaving")
23:24:39gartralohh.. 1) i put emulator when i meant simulator (and didnt make the connection till now) and 2) i need too do a sim build for the e200 then, so one moment
23:28:02jhMikeSgevaerts:"> I assumed the c200 situation was the same as e200. This also cleans some junk up from the arc driver between IMX31 and others before it got too cluttered.
23:31:07gevaertsc200 is indeed the same as e200
23:31:21 Quit dberg918 ("Ex-Chat")
23:32:49 Quit nibbler (Read error: 110 (Connection timed out))
23:39:49gevaertsjhMikeS: I think that looks OK. The only small issue I see is that the USB_PLUGGED vs USB_INSERTED naming could be confusing
23:40:16jhMikeSAny suggestion for renaming?
23:40:37gevaertsnot yet. I'm thinking...
23:41:02jhMikeSI wanted to preserve the INSERTED/EXCTRACTED event to function equivalently.
23:41:08gevaertsUSB_PLUGGED means "cable inserted" and USB_INSERTED means "cable inserted and USB signalling seen", right?
23:41:50jhMikeSbasically, charger insertion prepares it to see a bus reset. if it gets that it goes further with full enabling.
23:41:50linuxstbDoes "cable inserted" mean that the cable is attached to anything at the other end?
23:42:18linuxstbSo it's "power" and "data"?
23:42:26jhMikeSit means the charger is present which is indistinguishable from USB power from an electrical standpoint
23:42:56linuxstbIsn't there already a USB_POWERED define being used somewhere?
23:42:59jhMikeSI even checked the PCB on that one
23:43:16jhMikeSlinuxstb: only as a state, not a message event
23:43:52linuxstbWhy not add EVT or similar to those event defines?
23:45:01jhMikeSI could use USB powered as an event and it would conflict with nothing at all.
23:45:30gevaertsI guess that would be better
23:46:30gevaertsAnyway I think using bus reset detection is the right way to go with this
23:49:56***Saving seen data "./dancer.seen"
23:58:42 Quit gregzx (Connection timed out)

Previous day | Next day