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 2010-12-12

00:01:01the_KyleBut I can use mkamsboot to keep my player from booting to the OF when I plug in the cable when the player is powered off?
00:01:31the_KyleIt will create a bootloader that will function in that way?
00:01:42saratogaassuming you are good at ARM asm, yes
00:02:50the_KyleSo it doesn't work automatically since I patched my Rockbox for USB support?
00:02:54pixelmasaratoga: so the ASMv2 USB patch doesn't apply to the bootloaders?
00:03:31saratogaits not even the bootloaders, its the code compiled into rbutil and mkamsboot
00:03:35saratogaand no thats not changed by it
00:04:19pamaurythe amsv2 usb patch applies to the usb drive; as far as I know, no bootloader has a complete usb stack
00:05:02the_KyleSo the bootloader can't pass control to Rockbox if the player is powered off?
00:07:05pamaurywhat do you mean ? the bootloader passes control to rockbox in any case except on targets where rockbox doesn't have a usb stack, in which case is usually loads the OF when usb is connected at power on
00:07:17pamauryThat's the case of amsv2 based targets
00:07:33amee2kmmmh, kugel's threading patch doesn't use target/hosted/sdl/thread-sdl.c anymore, right?
00:07:57 Quit casainho (Remote host closed the connection)
00:08:16pamauryThe amsv2 usb patch is an attempt at having stable usb in rockbox and thus at having (finally) a working usb stack in rockbox
00:08:42amee2ki was almost glad i started to get that one in my head, when i noticed its not being compiled anyway
00:08:54amee2kso now i'm scratching my balls over where the threading went 0.0
00:08:56 Join casainho [0] (~chatzilla@
00:10:37BuschelTheSeven: Cool, I just enabled 16 bit register writes
00:11:04***Saving seen data "./dancer.seen"
00:11:04BuschelLCD_CON |= 0x100 plus some minor tweaks
00:11:16the_Kylepamaury: I have enabled the USB patches, but when my player is powered off and I plug in a cable, it loads the OF instead of passing to Rockbox. If it's powered on before I plug in the cable, it works as expected, although it is a bit slow sometimes, but every mp3 player I have purchased in the past, because they are cheap for the most part, were slow also, so I don't really notice the difference.
00:11:20 Quit kevku (Ping timeout: 272 seconds)
00:12:34the_Kylepamaury: Is it not passing to Rockbox because I haven'tused mkamsboot since I enabled USB?
00:13:14TheSevenBuschel: tweaks in config regs or just software things?
00:13:17pamauryI guess so, I think the bootloader only loads rockbox if it's compiled with ROCKBOX_HAS_USB in config.h and friends (like in my patch)
00:13:42BuschelTheSeven: just in the register write functions in the lcd driver −− nothing else
00:14:21*TheSeven needs to figure out how to do that for DMA then
00:14:34TheSevenit can't handle 32bit, right?
00:14:48Buschelit could, didn't try yet
00:15:00saratogathe_Kyle: maybe i was unclear, it doesn't boot rockbox because no one has written code to do so
00:15:44saratogayou can fix this by implementing the equivilent of the patch I linked above for AMSv2 using your knowledge or ARM asm, or else wait until USB support is stable enough that its enabled by default and we release a new rbutil with support
00:16:56TheSevenBuschel: let me know if it does :)
00:17:51TheSevenwhat's the better option? duplicating the ATA driver for the ipod classic, fixing it for all the other targets, or producing another #ifdef hell?
00:17:51the_Kylesaratoga: So there's no code at all at this point to pass to Rockbox when a cable is plugged in while powered off? I can just be sure it's powered on before I plug in the cable, I was just hoping there was a way to avoid loading the OF in the event I forget to turn the power on e.g. before charging.
00:18:19TheSeveni won't be able to map that to the ipod classic's hardware as-is
00:18:53saratogawhats wrong with the current ATA driver?
00:19:16TheSevenit assumes that the task file is directly mapped on the system bus
00:20:09TheSeven(which is only partially true on the classic)
00:21:00the_KyleNot even if ROCKBOX_HAS_USB is enabled, which according to pamaury should work?
00:21:59saratogayou're welcome to look at that patch and see for yourself if it changes mkamsboot
00:22:45pamaurythe_Kyle: my assumption is the bootloader has the code for it; if like saratoga says, there is no code for it, it won't work
00:23:37 Quit froggyman (Read error: Connection reset by peer)
00:24:49the_KyleThanks for the help. I'll take a look at things, but I'll probably have to wait until USB is more stable, since I don't have enough knowledge of the code, especially the low-level stuff like the bootloader, to be able to patch it.
00:25:02BuschelTheSeven: 32bit does not seem to work
00:25:05*TheSeven wonders if he should introduce macros to read/write ATA task file regs
00:26:11wodzhmm, lcd controller can't handle the speed of DMA transfers at 124 MHz on HD300 :-(
00:26:39TheSevenwodz: shouldn't DMA usually have some kind of flow control?
00:26:51 Join AlexP_ [0] (~alex@rockbox/staff/AlexP)
00:27:05wodzTheSeven: you mean wait states?
00:27:08the_KyleFor now, I'll just be sure the player is powered on before charging it or manipulating files. I just don't like the stuff that the OF puts on the player. It rewrites its idea of what it thinks the directory structure should be and everything.
00:27:22 Join Horschti [0] (~Horschti@xbmc/user/horscht)
00:27:39TheSevenwodz: usually there is a DMAREQ line from the peripheral to the DMA controller that tells it that it's ready to send/receive more data
00:27:57wodznot the case here
00:28:19wodzlcd is connected to lower half of the bus and that's all
00:28:20TheSevenso you're basically doing memory2memory on a peripheral?
00:28:53 Join designate72 [0] (
00:29:32 Quit AlexP (Ping timeout: 264 seconds)
00:29:58 Nick AlexP_ is now known as AlexP (~alex@rockbox/staff/AlexP)
00:30:06wodzHow do I know at which freq cpu is running? Maybe I could use DMA transfer when not boosting?
00:30:35gevaertsthe_Kyle: from what I understand, boot-to-rockbox-with-usb is reasonably easy to change (although I'm not exactly sure where), it's just not exactly safe. The change is basically in the only part of the code that can cause your clip to be bricked (or very nearly so, requiring opening it and shorting pins) if you get it wrong
00:31:09 Quit Horscht (Ping timeout: 265 seconds)
00:31:31TheSevenwodz: there's a global cpu_frequency variable
00:31:59TheSevenbut you would need to make sure to interrupt the DMA transfer if a boosting request comes in while it's already running
00:32:06the_KyleOuch. I don't think I want to do that. I'll just wait, or at least wait for a patch that has been tested and found to be at least somewhat safe.
00:32:42the_KyleThis is my primary player, and a sim isn't going to tell me if it's safe or not.
00:33:22TheSevenwell, basically it should be safe if you know what you're doing, are careful, and double- and triple-check things before flashing :)
00:33:41TheSeven(and possibly ask other people to review)
00:34:00TheSevenbut if you've never dealt with ARM assembly code before, this isn't really the place where you should start :)
00:34:54the_KyleI can get help. You all have been nothing but helpful so far. I just don't know that I know enough about the code to mess with it in that way.
00:43:40wodzha, this seems to work. 45 MHz speedup is from 438 -> 728 fps, boosted fps is unchanged 1098
00:43:43 Quit {phoenix} (Read error: Connection reset by peer)
00:44:22Buschelseems to be a good day for LCD speed ups :)
00:44:39 Quit Horschti (Ping timeout: 265 seconds)
00:45:54wodzam I correct that boost/unboost changes on tick?
00:46:18 Join Horscht [0] (
00:46:19 Quit Horscht (Changing host)
00:46:19 Join Horscht [0] (~Horschti@xbmc/user/horscht)
00:46:41 Join fdinel [0] (
00:46:52wodzI mean the shortest possible interval between freq change is 10ms?
00:47:21 Join Horschti [0] (~Horschti@xbmc/user/horscht)
00:49:40 Quit ender` (Quit: The problem with political jokes is they get elected. -- Henry Cate, VII)
00:50:02BuschelTheSeven: now it seems like I have a reached a maximum speed on my LCD -> 64.5 fps unboosted and 129.5 fps boosted
00:50:12Buschel(for RGB full screen)
00:50:35the_KyleIs recording sampling done on a hardware level, or is it done in Rockbox? I get a lot of aliasing if I set the recording frequency below 44KHz.
00:50:35 Quit leavittx (Ping timeout: 255 seconds)
00:51:02the_Kyle32 seems OK, actually, but below that it starts getting bad.
00:51:28 Quit Horscht (Ping timeout: 276 seconds)
00:55:16CIA-7New commit by wodz (r28799): HD300 - further speedup of lcd_update() by utilizing DMA transfer when unboosted. The gain is 438 -> 728 fps @ 45MHz.
00:57:19CIA-7r28799 build result: All green
00:57:43 Join saratoga_ [0] (9803c22e@gateway/web/freenode/ip.
00:58:05saratoga_ok i think all the libmad algorithmic stuff is working, now the asm part
01:00:31 Quit wodz (Quit: Leaving)
01:02:10*TheSeven desperately tries to make the ipod classic port even compile :/
01:05:28 Join webguest79 [0] (
01:06:14TheSevendo we have any define that allows to determine if we're being included from an asm file?
01:06:39pamaury_ASM ? Or something like that
01:06:54pamauryThat's pretty common to have one, rockbox should be no exception
01:07:45 Quit webguest79 (Client Quit)
01:08:47 Join GeekShadow [0] (~Antoine@
01:08:47 Quit GeekShadow (Changing host)
01:08:47 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
01:09:45TheSevenapparently a handful of files uses "#define ASM", but most don't define anything
01:09:53 Quit casainho (Ping timeout: 245 seconds)
01:10:13pamauryperhaps it's on the as command line ?
01:10:39TheSevenwhy would some files have one then?
01:11:03*TheSeven discovers that the ATA mess is much deeper than he thought
01:11:55pamaurydon't know; perhaps there is none and we should have one
01:12:03 Quit dantje_ (Quit: Ex-Chat)
01:12:27 Quit [Saint] (Quit: I'm only going to Heaven if it feels like Hell, I'm only going to Heaven if it tastes like caramel...)
01:25:43 Quit JesusFreak316 (Ping timeout: 245 seconds)
01:26:33TheSevennow i'm *just* getting errors about undefined references during linking
01:34:07 Quit dfkt (Quit: -= SysReset 2.53=- Sic gorgiamus allos subjectatos nunc.)
01:39:51Buschelis there any chance that an owner of an iPod nano 2G with an ILI type LCD can verify FS #11807 ?
01:44:10TheSeventhe ilitek one seems to be quite common actually
01:44:22TheSeventype 3 is the one that barely anybody has
01:45:33BuschelI would really like to see a test before I commit. I hate to be forced to fix afterwards...
01:48:32saratogawhat do people think of that automatic resume patch on the tracker?
01:51:53 Quit shai_ (Ping timeout: 260 seconds)
01:52:07CIA-7New commit by theseven (r28800): usb-s3c6400x.[ch], button-clickwheel.c: Move s5l8701-specific parts to where they belong, prepare for s5l8702
01:53:15 Quit bertrik (Quit: :tiuQ)
01:53:54CIA-7r28800 build result: 0 errors, 71 warnings (theseven committed)
01:55:44 Quit ReimuHakurei (Ping timeout: 265 seconds)
01:59:35 Quit Buschel (Ping timeout: 240 seconds)
02:00:01CIA-7New commit by theseven (r28801): Fix yellow, already assign values for various S5L8702 defines
02:00:14 Join Horscht [0] (~Horschti@xbmc/user/horscht)
02:01:44*TheSeven wonders if that meier crossfeed thing will be committed one day
02:01:51CIA-7r28801 build result: 0 errors, 66 warnings (theseven committed)
02:02:00 Join ReimuHakurei [0] (~reimu@
02:02:33 Quit Horschti (Ping timeout: 264 seconds)
02:03:16CIA-7New commit by theseven (r28802): Really fix yellow this time, SVN only committed half of what I wanted...
02:03:52 Quit factor (Ping timeout: 241 seconds)
02:05:24CIA-7r28802 build result: All green
02:06:13 Quit ReimuHakurei (Ping timeout: 241 seconds)
02:08:26 Join timccc [0] (~timccc@
02:11:08***Saving seen data "./dancer.seen"
02:11:27 Join simonrvn [0] (
02:17:35 Quit GeekShadow (Remote host closed the connection)
02:21:12 Quit Judas_PhD (Quit: This is a quitting message)
02:24:29 Quit pamaury (Remote host closed the connection)
02:29:47 Join Horschti [0] (~Horschti@xbmc/user/horscht)
02:33:21 Quit Horscht (Ping timeout: 259 seconds)
02:35:56 Join jfc^3 [0] (
02:38:15 Join Topy [0] (
02:38:15 Quit T44 (Read error: Connection reset by peer)
02:39:36 Quit jfc (Ping timeout: 255 seconds)
02:40:54 Join Horscht [0] (~Horschti@xbmc/user/horscht)
02:43:33 Quit Horschti (Ping timeout: 240 seconds)
02:50:20 Join Horschti [0] (~Horschti@xbmc/user/horscht)
02:54:18 Quit Horscht (Ping timeout: 260 seconds)
02:56:01 Quit linuxguy3 (Ping timeout: 272 seconds)
02:56:59 Quit fdinel (Ping timeout: 245 seconds)
02:57:22 Join linuxguy3 [0] (~timj@
03:00:08 Join Horscht [0] (~Horschti@xbmc/user/horscht)
03:02:00 Quit Keripo (Quit: Leaving.)
03:02:34 Join Keripo [0] (
03:03:03 Quit Horschti (Ping timeout: 260 seconds)
03:03:05 Quit Keripo (Client Quit)
03:04:59saratoga_does eterm in the rockbox dev image not allow copy and paste?
03:05:52CIA-7New commit by moos (r28803): Update the list of targets used on english.lang file.
03:07:46CIA-7New commit by moos (r28804): Merge the translation of too strings for a generic use.
03:08:07CIA-7r28803 build result: All green
03:10:00 Join Keripo [0] (
03:10:33CIA-7r28804 build result: All green
03:11:05 Join Horschti [0] (~Horschti@xbmc/user/horscht)
03:13:08 Quit Horscht (Ping timeout: 240 seconds)
03:16:51 Quit mystica555_ (Read error: Operation timed out)
03:21:27 Join Horscht [0] (~Horschti@xbmc/user/horscht)
03:21:39saratoga_wow our VMware image is really terrible
03:21:51saratoga_i'm just trying it now, i see why people try and use cygwin
03:22:20saratoga_just downloading an ubuntu or debian image from sourceforge and running in current virtualbox for Windows is 100x less annoying
03:24:53 Quit Horschti (Ping timeout: 276 seconds)
03:25:13 Join mortalscan [0] (~mortalsca@
03:26:22saratoga_anyone know how to uninstall the old arm compiler?
03:26:59 Join cjcopi [0] (
03:29:16 Join Horschti [0] (~Horschti@xbmc/user/horscht)
03:31:51CIA-7New commit by moos (r28805): Punctuation consistency, revert part of r27363.
03:33:09 Quit Horscht (Ping timeout: 264 seconds)
03:33:58CIA-7r28805 build result: All green
03:39:22 Join Horscht [0] (
03:39:25 Quit Horscht (Changing host)
03:39:25 Join Horscht [0] (~Horschti@xbmc/user/horscht)
03:41:06 Join TheSphinX^ [0] (
03:42:15 Quit Horschti (Ping timeout: 260 seconds)
03:42:31 Quit TheSphinX^ (Client Quit)
03:43:41 Join S_a_i_n_t [0] (S_a_i_n_t@
03:49:15 Join Horschti [0] (~Horschti@xbmc/user/horscht)
03:49:59 Join factor [0] (
03:53:01 Quit Horscht (Ping timeout: 272 seconds)
03:53:03 Join T44 [0] (
03:55:16 Quit Horschti (Quit: Verlassend)
03:56:47 Quit Topy (Ping timeout: 240 seconds)
04:01:00saratoga_omg does the version of vmware player we tell people to use really not support 2 processors
04:01:39saratoga_make 7zip
04:02:24saratoga_^^^ hmm well it definitely doesn't support 'taking control of the windows keyboard when brought to focus'
04:03:44scorchemeh - use vmware server
04:03:58saratoga_the wiki explicitly states to use this version
04:04:14saratoga_i'll update, test and then change the wiki
04:04:29saratoga_is there any reason we use vmware instead of virtualbox?
04:04:53saratoga_last i checked virtualbox is nice because its open source and will let you use an unlimited number of cores without paying
04:05:04saratoga_although maybe vmware has that now too
04:06:07 Quit saratoga (Ping timeout: 265 seconds)
04:08:28saratoga_heh apt-get fails on our image as well due to missing download servers
04:10:02 Quit BHSPitMonkey (Ping timeout: 272 seconds)
04:11:10***Saving seen data "./dancer.seen"
04:11:49scorcheyeah - it is way out of date - i made an updated one a while back, but didnt have time to properly test it - and people on the forum faled to properly test it enough to put on the wiki as well
04:11:55scorchei will eventually make another
04:12:01 Join DerPapst1 [0] (
04:12:28saratoga_i'm just going to upload the same one with the new compiler
04:13:43saratoga_how the hell do i turn this thing off
04:13:47saratoga_no matter what i do it just reboots
04:14:24 Quit DerPapst (Ping timeout: 245 seconds)
04:14:34scorchethere is a power off option
04:15:27saratoga_even shutdown -h just reboots
04:15:54 Quit Keripo (Read error: Connection reset by peer)
04:18:43 Join Keripo [0] (
04:20:07saratoga_haha vmware wants me to register to update
04:20:09saratoga_i'll get right on that
04:21:52 Join JesusFreak316 [0] (
04:21:57 Quit Keripo (Read error: Connection reset by peer)
04:24:35 Join Keripo [0] (
04:30:15 Quit designate72 (Quit: Leaving)
04:32:02 Quit pixelma (Disconnected by services)
04:32:03 Quit amiconn (Disconnected by services)
04:32:03 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
04:32:04 Join pixelma_ [0] (quassel@rockbox/staff/pixelma)
04:32:07 Nick pixelma_ is now known as pixelma (quassel@rockbox/staff/pixelma)
04:32:23 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
04:34:38 Join froggyman [0] (~seth@unaffiliated/froggyman)
04:35:41 Quit TheSeven (Ping timeout: 250 seconds)
04:36:04 Join ReimuHakurei [0] (
04:38:23 Quit moos (Quit: ChatZilla 0.9.86 [Firefox 3.6.13/20101203075014])
04:39:08 Quit JesusFreak316 (Remote host closed the connection)
04:39:36 Join TheSeven [0] (~TheSeven@rockbox/developer/TheSeven)
04:39:40 Quit efyx (Remote host closed the connection)
04:41:47 Join Barahir [0] (
04:41:54saratoga_does leave temp files around?
04:41:59saratoga_my image got a lot bigger when i ran it
04:42:17 Quit Barahir_ (Read error: Operation timed out)
04:47:37soapdue to dead space no longer being zeroed out?
04:47:43 Quit froggyman (Quit: Bye)
04:47:47saratoga_soap: probably that, how do i zero it?
04:48:11soapcat /dev/zero > zero.txt
04:48:18soapthat should completely fill up the drive
04:48:23soapthen rm zero.txt
04:50:01 Quit the_Kyle (Quit: Leaving.)
04:51:37 Join the_Kyle [0] (~kyle@
04:54:00 Quit elcan (Read error: Connection reset by peer)
04:54:07 Join elcan [0] (
04:56:02 Quit sasquatch (Quit: WeeChat 0.3.2)
04:56:28 Join sasquatch [0] (
04:57:47 Join shai [0] (
05:01:34 Join froggyman [0] (~seth@unaffiliated/froggyman)
05:02:04S_a_i_n_tsaratoga_: I believe (depending on what you actually did, however) it leaves some temp files in the form of tmp/rbdev-build and tmp/rbdev-dl which easily adds up to a few hundred MBs.
05:06:19saratoga_wow got it 20MB smaller the original :)
05:07:10 Quit the_Kyle (Quit: Leaving.)
05:09:25 Join the_Kyle [0] (~kyle@
05:09:58 Quit froggyman (Quit: Bye)
05:10:46 Quit InsDel (Read error: Connection reset by peer)
05:14:12 Quit mortalscan (Quit: Leaving)
05:18:08 Quit saratoga_ (Ping timeout: 265 seconds)
05:45:04 Join Judas_PhD [0] (
06:10:30 Quit timccc (Remote host closed the connection)
06:11:01 Join timccc [0] (~timccc@
06:11:13***Saving seen data "./dancer.seen"
06:15:17 Quit panni_ (Quit: ( :: NoNameScript 3.81 :: ))
06:35:53 Join froggyman [0] (~seth@
06:35:53 Quit froggyman (Changing host)
06:35:53 Join froggyman [0] (~seth@unaffiliated/froggyman)
06:35:56 Join telliott [0] (
06:36:04 Quit shai (Ping timeout: 245 seconds)
06:38:08 Quit Judas_PhD (Quit: This is a quitting message)
06:40:07 Quit froggyman (Client Quit)
06:43:19 Quit ReimuHakurei (Ping timeout: 272 seconds)
06:47:28 Quit Keripo (Quit: Leaving.)
06:48:25 Quit DerPapst1 (Quit: Leaving.)
06:53:16 Join ReimuHakurei [0] (
06:58:22 Join Topy44 [0] (
07:14:07 Join froggyman [0] (~seth@
07:14:08 Quit froggyman (Changing host)
07:14:08 Join froggyman [0] (~seth@unaffiliated/froggyman)
07:23:02 Quit telliott (Quit: Easy as 3.14159265358979323846...)
07:40:13 Join Keripo [0] (
07:49:34 Quit Primula (Quit: o.O)
08:11:16***Saving seen data "./dancer.seen"
08:56:56 Join kevku [0] (~kevku@2001:7d0:0:f000::135d)
08:58:52S_a_i_n_tAnyone here use the ZXBox plugin?
08:59:30*S_a_i_n_t is trying to figure out if it is broken, or if the images he is trying to load are broken.
09:00:55pixelmasometimes, haven't tried in a while though
09:08:47 Quit Keripo (Ping timeout: 255 seconds)
09:08:54 Join Keripo [0] (
09:10:06 Join Judas_PhD [0] (
09:15:00 Join Buschel [0] (
09:15:52 Quit Buschel (Client Quit)
09:34:04 Join Horscht [0] (~Horschti@xbmc/user/horscht)
09:58:55 Join bmbl [0] (
09:58:55 Quit bmbl (Changing host)
09:58:55 Join bmbl [0] (~bmbl@unaffiliated/bmbl)
09:59:47 Join Buschel [0] (
10:02:25 Join stoffel [0] (
10:03:59BuschelS_a_i_n_t: yesterday you mentioned that you "mostly have LDS176 LCDs" in your nano 2Gs. does this mean you also have a target with an ILI type?
10:07:51 Join bertrik [0] (
10:07:51 Quit bertrik (Changing host)
10:07:51 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
10:08:58 Join leavittx [0] (~leavittx@
10:10:27 Join {phoenix} [0] (
10:11:20***Saving seen data "./dancer.seen"
10:14:13 Quit ReimuHakurei (Ping timeout: 272 seconds)
10:17:12S_a_i_n_tBuschel: IIUC, there's 3 LCD types that the Nano2G of which seems to be quite rare. IIRC, I have 5 type 1, (7) LCD Nano2Gs and 1 type 2, (5) (I *think* that's it...anyway, I think its in the car) LCD Nano2G
10:18:04S_a_i_n_tI'll need to double check which type exactly the other one is, but...I do know I have 2 different LCD types.
10:18:41Buschelwell, it would be great to have the LCD driver patch on other LCD types as well
10:18:41 Join BHSPitMonkey [0] (~stephen@unaffiliated/bhspitmonkey)
10:20:04S_a_i_n_tTheSeven has the "rare" type (is it the LEADIS LCD? my menory is broken today) also if my memory serves.
10:20:09S_a_i_n_tBuschel: ^
10:20:27S_a_i_n_tI *think* he does...rather.
10:20:40Buschelthe LDS type seems to be the common case −− that is what TheSeven and I own
10:23:51S_a_i_n_twith the LDS176 screened Nano2Gs I have, there seems to be a "bright" and a "dark" version...I have several, and I have RB'd several for friends also.
10:24:05 Quit Judas_PhD (Quit: This is a quitting message)
10:24:47S_a_i_n_tAccording to debug, the screens are the same...but some are definitely a lot clearer/brighter than others and it doesn't seem to have anything to do with how old/new the Nano in question is.
10:25:28Buschelis this also visible in Apple OS? or just when those targets were rb'ed
10:28:54S_a_i_n_tIt is visible in the Apple OS also.
10:30:22S_a_i_n_tHmmm...I'm sure (almost positive) that the Nano2G has three possible LCDs but the driver only seems to handle 2. My mistake.
10:31:19Buschelyou wre right. TheSeven also stated there are three types of which 1 is very seldom
10:34:47S_a_i_n_tAh, right...I thought so, but the code only suggests we know how to deal with 2 of them.
10:36:21S_a_i_n_tperhaps the third type is the bright or dim LDS176 type screens I have seen/own.
10:36:53Buschelcould you test the patch on those units?
10:37:19Buschelfor now it was only verified on my "golden device" ;)
10:37:30S_a_i_n_tyep. I can do so in the next hour or so. just need to run out for a bit presently.
10:37:45S_a_i_n_t"golden device"?
10:37:51*S_a_i_n_t lols.
10:38:33BuschelS_a_i_n_t: that would be perfect :)
10:39:27amee2kdoes the SDL target have a fucking quit key
10:39:47amee2ki can't get the profiling info when i kill it with ^C ffs
10:41:14amee2kany hints?
10:42:31S_a_i_n_tI don't believe it has a quit key, fucking or otherwise.
10:54:10 Quit BHSPitMonkey (Remote host closed the connection)
10:56:58 Join JdGordon1 [0] (
10:58:28 Join JdGordon2 [0] (
11:00:04 Quit JdGordon (Ping timeout: 250 seconds)
11:01:04 Quit JdGordon1 (Ping timeout: 240 seconds)
11:01:07 Join JdGord [0] (6fdcf785@gateway/web/freenode/ip.
11:02:29 Quit JdGordon2 (Ping timeout: 240 seconds)
11:05:05 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
11:08:34 Quit JdGord (Ping timeout: 265 seconds)
11:11:17 Join pamaury [0] (
11:11:18 Quit pamaury (Changing host)
11:11:18 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
11:18:38UtchybannBuschel: FS #11807 *PANIC* my nano 2G.
11:19:14BuschelUtchybann: good!
11:19:24Buschelwhat type of LCD do you have?
11:19:38UtchybannLCD is LDS176
11:19:50Buschelhmm, same than mine...
11:20:02Buschelwhat is the exact error dislayed?
11:20:14Utchybannfull panic message is : Unhandled IRQ 1A: INT_SPI
11:20:43UtchybannI get the same error on my 2 nano 2g (both have same LCD type).
11:20:49Buschelis this with version v05?
11:21:11Utchybannmy fault. I download the patch last night.
11:21:38Buschelhmm, v03 is changing much less than v05...
11:24:01Utchybannis it supposed to work with r28805 ?
11:24:57BuschelI did not update since this night, my patch works against r28799
11:25:11Buschelthat's what I use here
11:26:09Utchybannlet me check with v05 and r28799
11:26:53 Join stooo [0] (
11:27:25Buscheldoes it crash instantly or after a while?
11:29:14UtchybannI don't even see the rockbox logo.
11:30:12Buscheldo you also update the bootloader or just rockbox.ipod
11:30:35bertrikCan someone with a fuze v2 test the patch in for me?
11:30:48Utchybannmake zip; unzip -d /media/ipod
11:31:04Utchybannso no bootloader update
11:33:43Utchybannr28799+lcd_v05 seems to work. no panic
11:34:25Buschelmaybe the panic is connected to anything that was done after r28799...
11:34:44Buschelwhat is the speed you reach with test_fps?
11:36:20UtchybannI guess, I have to compile test_fps ...
11:39:07 Join bimbel [0] (~bmbl@unaffiliated/bmbl)
11:39:25 Quit bmbl (Disconnected by services)
11:39:31 Nick bimbel is now known as bmbl (~bmbl@unaffiliated/bmbl)
11:53:55UtchybannMain 1/1: 64.5 fps, 1/4: 258.0; YUV 1/1 34.5, 1/4 137.5 (CPU 48 Mhz)
11:54:39 Join dantje [0] (
11:55:04Buschelexactly my numbers.
11:55:39Utchybann129, 516; 109.5 434.5 (CPU 192Ã)
11:56:15Buschelnearly the same for me (only really minor diffs though)
11:56:45 Join GeekShadow [0] (~Antoine@reactos/tester/GeekShadow)
11:57:35Utchybannnice. I will try r28805 + lcd_v05.
11:58:22Buschelbtw, everything working fine when using lcd sleep?
11:59:18Utchybannlcd sleep ? you mean the total lcd power off after a few seconds ?
11:59:37Buschelyes. does it sleep well and wakes up as it is supposed to do?
12:01:45JdGordonanyone notice the stock android is more happy to kill the rockbox service than the custom roms? (specifically CM)
12:03:11 Join casainho [0] (
12:05:14Utchybannr28805 + lcd_v05 => panic.
12:05:56Buschelcan you track this down to the change which introduces this?
12:07:08UtchybannI'm trying
12:08:44Utchybannhmm, r28805 => panic.
12:09:10Buscheltry r28800 and r28801
12:11:21 Join ender` [0] (
12:11:24***Saving seen data "./dancer.seen"
12:12:02Utchybannfor me r2880[012] should be seen as a single commit.
12:13:56 Quit casainho (Ping timeout: 240 seconds)
12:16:24 Join liar [0] (
12:19:54Utchybannr28800 => panic.
12:20:21Buschelthere we go
12:20:58UtchybannTheSeven: something is wrong in r28800 (PANIC Unhandled IRQ 1A: INT_SPI).
12:21:06 Join Kupop [0] (
12:22:23 Join casainho [0] (
12:22:49casainhodoes anyone knows why Rockbox bootloader takes so much memory? Binary size: 51360
12:22:50casainhoActual size: 51360
12:22:52casainhoRAM usage: 354448
12:23:16casainhowould be possible to make it less than 32kBytes?
12:23:18UtchybannI will revert to r28799+lcd_v05 and report issue if any.
12:23:43bertrikcasainho, maybe the LCD frame buffer?
12:23:44 Join TheLemonMan [0] (
12:23:53BuschelUtchybann: could you try to only keep the clickwheel driver at pre-r28800?
12:24:07casainhobertrik: like having a bootloader that doesn't write/use LCD?
12:24:42bertrikyes, something like that. But can't you confirm first by looking into the map file?
12:25:18casainhobertrik: hmmm, I will try to look and see if I understand..
12:25:43 Quit S_a_i_n_t (Quit: I'm only going to Heaven if it feels like Hell, I'm only going to Heaven if it tastes like caramel...)
12:26:27bertrikAnyone with a sansa fuze v2 online here now?
12:27:22casainhobertrik: can you please look?
12:30:34 Join Rob2222 [0] (
12:32:03bertrikcasainho, see line 1740/1741, using about 150k for the LCD framebuffer (I guess 320*240 16 bit?)
12:32:23bertrikI also saw some kind of codepage table taking 64kB
12:33:31bertrikIt will probably be hard to reduce it to less than 32 kB
12:33:33BuschelUtchybann: looks like clickwheel driver is the problem. the ISR for the wheel driver was renamed without a match in system-s5l8700.c
12:33:48BuschelUtchybann: recompiling now
12:34:55casainhobertrik: ok. However my LCD now is 1bit, 64X48...
12:35:21casainhobertrik: ok, I may need to think on a 1st and 2nd bootloader... Thanks!
12:35:33UtchybannBuschel: sorry. I have to go now.
12:35:40Buschelsee you
12:36:04bertrikBuschel, for some reason the interrupt handlers of many rockbox targets are not declared and it's up to the linker to connect them, defaulting to some unhandler using "weak linking" IIRC
12:39:06Buschelyes, I will fix this issue in a few minute (hopefully)
12:41:55CIA-7New commit by Buschel (r28806): Fix bug introduced with r28800 (missing interrupt handler).
12:42:00Buschelgot it
12:44:41CIA-7r28806 build result: All green
12:48:39 Join bertrik_ [0] (
12:48:39 Quit bertrik_ (Changing host)
12:48:39 Join bertrik_ [0] (~bertrik@rockbox/developer/bertrik)
12:48:47 Quit bertrik (Read error: No route to host)
12:50:21 Nick bertrik_ is now known as bertrik (~bertrik@rockbox/developer/bertrik)
13:16:39 Join teru [0] (
13:17:29 Join InsDel [0] (
13:36:44 Quit Rob2222 (Ping timeout: 240 seconds)
13:39:37 Join moos [0] (moos@rockbox/staff/moos)
13:39:45 Quit bluebroth3r (Ping timeout: 240 seconds)
13:40:38CIA-7New commit by theseven (r28807): iPod Nano 2G: Correct clickwheel interrupt handler name, this time consistently.
13:41:07 Join dfkt [0] (dfkt@unaffiliated/dfkt)
13:41:10moosbertrik: hi, still no found someone with a Fuzev2?
13:41:32*JdGordon has a fuzev2 i tinhk
13:41:38JdGordondo we need one?
13:41:43 Join bluebrother [0] (
13:41:43 Quit bluebrother (Changing host)
13:41:43 Join bluebrother [0] (~dom@rockbox/developer/bluebrother)
13:42:12moosHello, JdGordon: it seems bertrik need testing for
13:42:43CIA-7r28807 build result: All green
13:46:12JdGordonok, I'm charging both fuzzes (one is v1 one is v2, dunno which is which)... I'll try to remember to test tomorow
13:46:19JdGordonwhich probably means nagging me :D
13:46:43JdGordondo the fuze OF's trickle charge if the batt is completly flat?
13:47:25JdGordonah no, hub was overloaded...
13:48:14moosit have to charge yes
13:48:30JdGordoncan you do a build for me with the patch please?
13:49:07moosof course just few instants
13:50:06 Join efyx [0] (
13:50:08JdGordonisnt that for the clip+?
13:50:13JdGordonor fuzev2 also?
13:52:04moosreadinf the logs, bertrik seems to need testing for fuve v2 too
13:52:13moosmaybe an AMSv2 thing
13:53:58bertrikJdGordon, I think the hardware indeed charges with a low current when the battery voltage is very low
13:54:22JdGordonchanged usb ports and its going now, I tihnk the huyb was just overloaded
13:54:44JdGordon2xfuzes, ipod, phone, sd reader, external hard disk, keyboard dongle
13:55:29bertrikI could just commit and see who complains :)
13:55:44moosbuilding... :)
13:56:17JdGordonbertrik: I wont be able to test till tomorow morning, but yeah if oyu commit and it breaks I'm sure we'll know about it pretty quicjky
13:57:33moosif you have few minutes, I could search mine and test, if nedded
13:58:07bertrikmoos, very much appreciated if you could do that
13:58:34moosno problem, I will.
13:58:56bertrikhave you seen any problem accessing the uSD card on your fuze v2, suddenly several people seem to have problems with it
13:59:27moosbertrik: nope, never encountered any...
14:03:47 Quit antil33t (Read error: Connection reset by peer)
14:03:56 Join antil33t [0] (
14:11:26***Saving seen data "./dancer.seen"
14:13:50 Quit slooopy (Quit: Verlassend)
14:16:18 Quit Kupop (Ping timeout: 264 seconds)
14:26:37 Quit casainho (Read error: Operation timed out)
14:27:24moosbertrik: all seems to working weel
14:27:32bertrikthanks for testing
14:27:40moostried with 2 micro sdhc cards class2
14:28:01 Join Kitr88 [0] (
14:28:06moosdid some copy/past, playing some sonds without problems
14:28:09bertrikI'm trying to make the sd driver for AMSv2 more similar to the AMSv1 sd driver (porting some bugfixes)
14:28:27moosnice! good luck on this
14:28:47moosdon't hesitate to ask again for tests :)
14:30:32bertrikthe fuze v2 seemed more "sensitive" to small changes, for example when we changed the main clock from 240 MHz to 248 MHz, fuze v2 sd card communication failed for some people
14:31:05 Quit Kitar|st (Ping timeout: 276 seconds)
14:31:45 Join MethoS- [0] (~clemens@
14:32:21 Quit Kitr88 (Ping timeout: 255 seconds)
14:33:15moosohoh, strange thing. (btw I enabled adjustable frequencies here, and no problems for me neither so far)
14:33:33moosthere are maybe few hardware differences somwhere
14:35:55bertrikyeah, there is a newer hardware variant where the sd connections actually seem to make sense, e.g. not sharing the button light with some sd card control line :P
14:37:43 Join Kitar|st [0] (Kitarist@
14:38:19 Join notlistening [0] (
14:41:20 Quit MethoS- (Remote host closed the connection)
14:43:31*Buschel could adapt iPod video's asm yuv blit to nano 2g
14:44:59moosbertrik: silly them, like always :)
14:47:30 Quit antil33t (Read error: Connection reset by peer)
14:47:39 Join antil33t [0] (
14:49:45 Quit notlistening (Remote host closed the connection)
14:53:42 Quit teru (Quit: Quit)
14:57:19BuschelYUV is now nearly twice as fast as in svn :)
14:57:57mooswoot! :)
14:58:00moosgood job
14:58:28Buschelstill some room for tweaking left
15:01:41 Quit antil33t (Read error: Connection reset by peer)
15:01:49 Join antil33t [0] (
15:04:10CIA-7New commit by bertrik (r28808): AMSv2: only switch sd cards to high speed mode for for v2 sd cards, just like is done for AMSv1
15:06:22CIA-7r28808 build result: All green
15:15:38bertrikhm, the RCA sent as part of the APP_CMD in the AMSv1 driver is always 0, while it is equal to the RCA of the selected card in the AMSv2 driver
15:16:42bertrikoh, I think I confused myself
15:22:51 Join n1s [0] (~n1s@rockbox/developer/n1s)
15:27:08 Quit stoffel (Ping timeout: 240 seconds)
15:29:54 Quit InsDel (Read error: Connection reset by peer)
15:30:28 Nick Stummi_ is now known as Stummi (Stummi@rockbox/developer/Stummi)
15:31:50TheSevendoes anybody have some information about the audio codec used in the ipod classic?
15:32:03TheSevenapparently it's a cirrus one, but i don't know anything more than that
15:32:11TheSevendoes rockbox support any cirrus codec?
15:32:31TheSevenare there publicly-available cirrus codec datasheets?
15:32:38TheSevendo they maybe have an identification register?
15:33:06n1sno, don't think we have any other targets with cirrus codecs
15:33:59bertrikdo we have pictures of the cirrus chip?
15:45:20pamauryomg, the lcd code of the fuze+ bootloader is crazy
15:45:41 Join hebz0rl [0] (
15:45:56pamauryeither is supports a large number of lcd types, either it is horribly bloated
15:47:49 Join DerPapst [0] (
15:49:01TheSevenbertrik: not of the die
15:49:06TheSevenand the package is apple-branded
15:53:36 Quit antil33t (Read error: Connection reset by peer)
15:53:44 Join antil33t [0] (
16:01:38CIA-7New commit by Buschel (r28809): FS #11708 - Major speedup of iPod nano 2G. Part 1: Loop unrolling and reduction of FIFO register polling. +50% for RGB, +34% for YUV.
16:03:32CIA-7New commit by gevaerts (r28810): Add MikMod plugin, ported by Jason Yu, with some minor work by Craig Mann and William Peters (FS #8806)
16:03:46CIA-7r28809 build result: All green
16:04:59 Join shai [0] (
16:05:58CIA-7r28810 build result: All green
16:08:47 Join designate72 [0] (
16:10:48CIA-7New commit by Buschel (r28811): FS #11807 - Major speedup of iPod nano 2G. Part 2: Use 16 bit data width and simplify write commands. Gives another +27% for YUV.
16:11:28***Saving seen data "./dancer.seen"
16:12:42CIA-7r28811 build result: All green
16:16:13bertriknice fix, got faster *and* smaller :)
16:16:21gevaertsMust be a bug :)
16:16:56CIA-7New commit by Buschel (r28812): FS #11807 - Major speedup of iPod nano 2G. Part 3: Unify different write commands. No change in speed.
16:18:33 Join stooo1 [0] (
16:18:36 Quit stooo (Read error: Connection reset by peer)
16:18:52CIA-7r28812 build result: All green
16:22:12 Nick YPSY is now known as Ypsy (
16:23:22CIA-7New commit by Buschel (r28813): FS #11807 - Major speedup of iPod nano 2G. Part 4: Introduce asm for yuv blitting. Overall speedup of part1-4 is +50% for RGB and +93% for YUV.
16:25:18CIA-7r28813 build result: All green
16:31:00CIA-7New commit by bertrik (r28814): AMSv2: handle sd card ACMDs similar to how it's done for AMSv1
16:32:40 Quit DerPapst (Read error: Connection reset by peer)
16:32:51CIA-7r28814 build result: All green
16:33:14 Join DerPapst [0] (
16:41:24 Join jpt9 [0] (~jpt9@unaffiliated/jpt9)
16:41:33jpt9Help! How do I get out of Doom on my Sansa Clip v2?
16:42:03jpt9Never mind.
16:42:47 Join ReimuHakurei [0] (
16:46:30CIA-7New commit by Buschel (r28815): iPod nano 2G does use less current since the latest optimizations.
16:48:26 Quit ReimuHakurei (Ping timeout: 240 seconds)
16:48:28CIA-7r28815 build result: All green
16:49:22jpt9My god... it does temporally dithered PictureFlow...
16:49:26jpt9This is insane.
16:59:21 Join Kupop [0] (
17:02:52 Quit T44 (Quit: Leaving)
17:12:34TheSevenBuschel: someone should probably do some real current measurements
17:12:48TheSeveni know for sure that the adc readings are way off, i'd suspect by ~7mA
17:13:23BuschelTheSeven: the numbers do pretty well match the achievable runtime 400 mAh / 17 mA => 23.5h
17:13:35BuschelTheSeven: I had it running for 24+ h
17:13:46TheSevenyeah, but your ipod probably doesn't have anything near 400mAh any more
17:14:08 Join kugel [0] (~kugel@rockbox/developer/kugel)
17:15:09Buschelwell, we either correc the default capacity or use "real" mA's for the current. right now both match perfectly −− maybe because both have the same amount of deviation from reality :)
17:16:25Buschelbtw, we really should check if we introduce the measured/averaged current consumption for the runtime estimation. this should work quite well for non-HDD targets
17:16:51BuschelFS #10890
17:16:56 Quit JdGordon (Ping timeout: 265 seconds)
17:17:33 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
17:17:34Buschelfor now it is only available for nano 2G and Video 5G...
17:20:02kugelGodEater: does it still not find the compiler for you?
17:24:45kugelgevaerts: could/should we use mikmod's .mod detection for the metadata parser?
17:25:36gevaertskugel: I don't know. I think we want the metadata parser to agree with the codec
17:26:22 Quit Llorean (Quit: Leaving.)
17:27:31 Join ReimuHakurei [0] (
17:29:34kugelgevaerts: i.e.?
17:30:33amee2khey kugel :)
17:31:23CIA-7New commit by nls (r28816): FS #11802 by Li Jie fixing a typo in the doom buttonmap for SA9200.
17:31:34gevaertskugel: hm, I'm not sure *at all* now :)
17:31:59amee2ki've been pondering today... if it saps cpu time on end, it'll have to spend the time somewhere. so i swept what little i know about profiling out form under the carpet.... and failed miserably >_<
17:32:39 Quit efyx (Remote host closed the connection)
17:33:04TheSevennls: backporting candidate?
17:33:14CIA-7r28816 build result: All green
17:33:24amee2ki've added -pg to the GCCOPTS variable and recompiled (freshly configured in a new build dir). but i can't coerce it to produce the file with the profiling data
17:33:41TheSevenBuschel: assuming a constant inner resistance of the battery and constant current reading offset, the offset would be 7mA
17:33:49n1sTheSeven: no, that player is classed as "unusable" so it's not part of the release
17:34:10gevaertskugel: I haven't actually looked at the code. What I'm mainly concerned about (maybe "concerned" is a bit strong) is that we don't need (or want) the metadata parser to have code to support mod variants the codec doesn't support
17:34:11TheSeven(i'm currently using a compensated-by-7mA current reading to compensate for the inner resistance)
17:34:19TheSevenn1s: oh, right
17:34:34amee2ksince i couldn't find out what keybinding is Quit (and noone else here seems to know) i tried attaching gdb and called exit() from inside the program
17:34:43amee2kbut not even that got it to dump the profiling data
17:34:44kugelgevaerts: what does the codec support?
17:35:04*kugel thinks the codec should perhaps be dropped
17:35:41gevaertskugel: I don't know exactly
17:35:41kugelamee2k: IIRC currently the only exit way is the X button on the window :)
17:36:03amee2khehe. which doesn't exist when i run it full screen >_>
17:36:09BuschelTheSeven: hmm, this would mean my (internally measured) 18mA - 7mA = 11mA real current. 11mA * 24h (real runtime) => 264 mAh battery capacity. sounds reasonable, I always thought that nano's had 300mAh and not 400mAh batteries...
17:36:25gevaertsDropping the codec would imply not supporting mod files in regular playlists any more for people who only have "simple" mod files
17:36:28kugelamee2k: should be easy enough to add an exit combo, though
17:37:12kugelgevaerts: we could make the playlist code call the plugin :)
17:37:19kugeldidn't we want something like that anyway?
17:37:29kugel(for movies?)
17:37:36gevaertsSure we could, but that's going to be nontrivial
17:37:36amee2kmmmh, i think i found where the key presses are evaluated. no idea how to properly trigger the rockbox shutdown though
17:38:08amee2ki tried tracing how the ipod center button + play/pause shutdown works, but couldn't find it yet
17:38:15gevaertsamee2k, kugel: you could add a shutdown menu item on android and call exit from there?
17:38:38gevaertsamee2k: center button + play/pause isn't shutdown on ipod
17:38:45gevaertsIt's emergency hardware reset
17:38:46n1skugel: assuming that its is aplugin only because it needs more memory than a regular codec wouldn't it be easier to let codecs steal the buffer than hacking up the playlist code?
17:39:24TheSevenisn't the problem that the buffer is in use while a codec is running?
17:39:31gevaertsn1s: hack up the playlist code, or hack up the buffering code? Yes, nice choice there :)
17:39:38TheSeven(does stealing the audio buffer from a plugin neccessarily stop playback?)
17:39:48n1sTheSeven: yes
17:39:50TheSeven(stealing only parts of it, not the entire buffer of course)
17:39:58 Quit ze_ (Quit: Ex-Chat)
17:40:01n1sit's an all or nothing thing
17:40:03gevaertsYou can't steal parts of it with the current code
17:41:02*TheSeven proposes some kind of steal-from-audiobuffer-malloc thing and grabs his pitchfork just in case... ;)
17:41:17gevaertskugel: why doesn't sys_poweroff() just call exit() in the app?
17:41:29 Quit bluebrother (*.net *.split)
17:41:29 Quit bmbl (*.net *.split)
17:41:29 Quit sasquatch (*.net *.split)
17:41:29 Quit elcan (*.net *.split)
17:41:29 Quit simonrvn (*.net *.split)
17:41:29 Quit _jhMikeS_ (*.net *.split)
17:41:29 Quit yosafbridge (*.net *.split)
17:41:29 Quit pjm0616 (*.net *.split)
17:41:30 Quit crwl (*.net *.split)
17:41:30 Quit zu (*.net *.split)
17:41:30 Quit literal (*.net *.split)
17:41:30n1si think it would be nicer (not nice, but nicer) to let some codecs steal the buffer which would trigger a full rebuffer when they exit than playing some files in plugins
17:41:54TheSevenn1s: and where does the codec get its data input from then?
17:42:08kugelgevaerts: which app?
17:42:23gevaertskugel: well, any, really
17:42:36gevaertsOr the sim too actually
17:42:55kugelsys_poweroff is never called IIRC
17:43:09 Join bluebrother [0] (~dom@rockbox/developer/bluebrother)
17:43:09 Join bmbl [0] (~bmbl@unaffiliated/bmbl)
17:43:09 Join sasquatch [0] (
17:43:09 Join elcan [0] (
17:43:09 Join simonrvn [0] (
17:43:09 Join _jhMikeS_ [0] (~jethead71@rockbox/developer/jhMikeS)
17:43:09 Join yosafbridge [0] (
17:43:09 Join pjm0616 [0] (~user@
17:43:09 Join crwl [0] (
17:43:09 Join zu [0] (
17:43:09 Join literal [0] (
17:43:21n1sTheSeven: right, it needs to be a little cleverer than just taking the buffer
17:44:44TheSevenisn't the root problem that it needs to read the file non-linearly?
17:45:38TheSevencould some interface to peek ahead in the audio buffer data without consuming it solve it?
17:45:57gevaertsIt also may need to keep more data in RAM
17:46:42TheSevenhow much? could that be handled by stealing the plugin buffer?
17:47:00 Quit bluebrother (*.net *.split)
17:47:00 Quit bmbl (*.net *.split)
17:47:00 Quit sasquatch (*.net *.split)
17:47:00 Quit elcan (*.net *.split)
17:47:00 Quit simonrvn (*.net *.split)
17:47:01 Quit _jhMikeS_ (*.net *.split)
17:47:01 Quit yosafbridge (*.net *.split)
17:47:01 Quit pjm0616 (*.net *.split)
17:47:01 Quit crwl (*.net *.split)
17:47:01 Quit zu (*.net *.split)
17:47:01 Quit literal (*.net *.split)
17:47:13 Join efyx [0] (
17:47:38 Join bluebrother [0] (~dom@rockbox/developer/bluebrother)
17:47:38 Join bmbl [0] (~bmbl@unaffiliated/bmbl)
17:47:38 Join sasquatch [0] (
17:47:38 Join elcan [0] (
17:47:38 Join simonrvn [0] (
17:47:38 Join _jhMikeS_ [0] (~jethead71@rockbox/developer/jhMikeS)
17:47:38 Join yosafbridge [0] (
17:47:38 Join pjm0616 [0] (~user@
17:47:38 Join crwl [0] (
17:47:38 Join zu [0] (
17:47:38 Join literal [0] (
17:47:53gevaertsI don't think that's enough in general. Also, that's not at all a clean thing to do right now
17:48:12TheSeveni don't think there is a clean option right now
17:48:30TheSevenso we either need to do major restructuring or go for a dirty solution
17:49:15amee2kkugel: it doesn't seem to be effective either. at least not when i attach gdb to call it
17:49:29TheSeventhe cleanest way to do such things would probably relocateable codecs/plugins allocated from the audio buffer
17:49:34kugelamee2k: what did yo do?
17:50:05TheSevenat least for high-mem targets
17:50:10amee2kattach gdb to the process, then "call sys_poweroff()"
17:50:24kugelthat does nothing
17:50:55kugelbutton-sdl.c is where you'd add a exit combo
17:51:00gevaertsTheSeven: how does that solve the problem?
17:51:21amee2kmmm just wanted to see if it works at all
17:51:44amee2kalso, lots of stuff in apps/ seems to have calls to it
17:51:53TheSevenplugins/codecs could allocate as many space as they need and we wouldn't waste memory for things that aren't loaded at that time
17:52:14TheSevenmultiple plugins at the same time (pebbles?) would also be possible
17:52:50 Join fyrestorm [0] (
17:52:54gevaertsOh, sure it would allow lots of things, but I don't see how it handles the need for codecs to use more memory on its own
17:53:24TheSevenwhat i'm thinking of is roughly the following:
17:53:27kugelamee2k: should map F6 to exit
17:54:21n1swe would need a way to alloc chunks of buffer without stopping playback
17:54:22TheSevenwe have core .text/.bss at static addresses, maybe some buffers for e.g. dircache, and the rest is used as general-purpose memory
17:55:07TheSevenin that general-purpose memory codecs and plugins (probably with their .text/.bss sections separated) could be loaded, and audio buffer data would float around them
17:55:39TheSevenas long as it's only four non-audio chunks in that buffer, i don't think fragmentation will be a concern on >=8MB targets
17:56:26TheSevenand the mod codec could calculate how much memory it needs and size its .bss chunk accordingly
17:56:59TheSevenyeah, I know, no MMU, WhyNoMalloc etc., but I can't think of a better way right now.
17:57:07gevaertsTheSeven: yes, but you'd still need dynamic resizing of the audio buffer, which on its own (without relocation) would solve the problem too
17:57:45gevaertskugel: sys_poweroff() is called by the sim if you hold the shutdown button
17:58:10gevaertsSo I don't really know why that one doesn't just exit()
17:59:00gevaertsWhat RaaA would do with that is its own problem of course, but possibly a Player-like Shutdown item in the root menu might be a good idea
17:59:32TheSevengevaerts: hmm, that might be true, but it won't allow e.g. the whole codec, plugin and audio buffer space to be used by a huge codec
18:00:11gevaertsTheSeven: I'd argue that that's not desirable anyway :)
18:00:40amee2kkugel: system-sdl.c, likes 68..70 ... wtf is that stub?!
18:00:44kugelgevaerts: the sdl exit code does a lot of cleanup before the exit() call
18:00:59n1syes all the different schemes come down to being able to resize the audio buffer on the fly
18:01:13kugelamee2k: yes it is :)
18:01:33amee2kwas that what i called from gdb?
18:02:19gevaertsTaking away the *entire* audio buffer will probably introduce lots of fun issues, codecs habe the entire codec buffer today, and the plugin buffer is used by some non-plugin things already (the playlist viewer?)
18:02:51 Quit stooo1 (Ping timeout: 265 seconds)
18:05:00jpt9Hey. I'm on Vista Ultimate, running the latest version of Rockbox Utility. I have a Sansa Clip v2, and I'm trying to generate a voice file for it...
18:05:17jpt9Rockbox Utility keeps crashing around 72% or so through the encoding process.
18:05:32jpt9Any ideas?
18:06:16jpt9Talkfiles seem to work fine, though.
18:06:53 Join panni_ [0] (
18:11:29***Saving seen data "./dancer.seen"
18:12:21 Quit ReimuHakurei (Ping timeout: 240 seconds)
18:15:21 Quit moos (Ping timeout: 276 seconds)
18:18:22CIA-7New commit by Buschel (r28817): Tab police.
18:20:50CIA-7r28817 build result: All green
18:21:32 Quit Topy44 (Ping timeout: 245 seconds)
18:22:35n1sBuschel: please don't do that, it makes merging upstream changes a lot more annoying
18:23:36 Join stoffel [0] (
18:23:42Buschelhmm, thought this would be no with the right tools... will keep my fingers off this in future
18:24:50Buschelwanted to type "no problem", of course
18:24:57n1salso the indentation is messed up in some places see codebook.c:91-93
18:25:40n1ssvn merge cant ignore whitespace differences afaiu so it creates conflicts all over the place
18:29:05Buscheljust revert, if you are expecting issues with the next upstream merge
18:34:01n1si'll do it in a bit then
18:36:29 Quit designate72 (Quit: Leaving)
18:45:25 Join designate72 [0] (
18:50:07TheSevenwhat instruction is best for getting the two least significant bits of a register into some condition bits?
18:54:04 Join casainho [0] (~chatzilla@
18:57:57 Join fdinel [0] (
18:59:09CIA-7New commit by nls (r28818): Revert tab police as it makes merging upstream changes more annoying and messed up indentation in some places.
18:59:59 Part jpt9
19:01:09CIA-7r28818 build result: All green
19:01:29kugelTheSeven: tst?
19:02:39TheSevenit looks like MOV Ry, Rx,LSL#31 might be the best option
19:02:46 Join saratoga [0] (600afc5f@gateway/web/freenode/ip.
19:03:14TheSevener, MOVS of course
19:04:47saratogadoes anyone know why the AMSv1 players briefly get confused and try to enter USB mode when plugging in an AC charger?
19:05:17saratogait seems they detect any power on the dock connector as USB, try to connect, realize there is no USB device, and then disconnect
19:05:25 Join ReimuHakurei [0] (~reimu@
19:05:30saratogathe e200v1 however doesn't do this
19:05:58 Join Llorean [0] (~DarkkOne@rockbox/user/Llorean)
19:06:04n1ssaratoga: my fuze does that when i connect a regular usb connection while rb is running
19:06:22 Join robin0800 [0] (
19:06:23n1si can only connect successfully if i insert the cable while the fuze is off
19:06:24saratogainstead of going into USB mode?
19:07:04saratogaon the fuzev1?
19:07:22 Quit ReimuHakurei (Read error: Connection reset by peer)
19:07:25saratogamight be worth fileing a bug report on that if you haven't
19:08:03saratogaso we know theres a dev with usb problems on the fuze and not just the occasional user
19:08:21n1seh, now it seems to work but it used to do that...
19:08:56n1si should probably try to bisect it to see if any recent change fixed it
19:12:58 Join BHSPitMonkey [0] (~stephen@unaffiliated/bhspitmonkey)
19:16:33n1si won't do that today though
19:22:46 Quit Llorean (Quit: Leaving.)
19:24:38 Quit efyx (Remote host closed the connection)
19:28:09amee2khow can i run make in a way so it shows the full commands that it runs?
19:28:20amee2kand not just "CC <file>"
19:29:29kugelamee2k: make V=1
19:32:27amee2kthanks for your quit patch, it works (except for a typo on line 35 ;) but i'm still not getting any profiling info >_<
19:32:48kugeli didn't even test compile it :p
19:33:09amee2ki kinda figured that from the typo :)
19:33:34amee2kbut now i'm beginning to doubt that i'm compiling with the proper profiling options here
19:34:08amee2kalso, if it works then who cares if it was tested >_>
19:38:57 Join mortalscan [0] (~mortalsca@
19:46:07 Quit shai (Ping timeout: 245 seconds)
19:46:19mc2739saratoga: regarding amsv1 USB, thats how ranma set it up. It tries to make a data connection and then times out to charge mode. I find it rather annoying, but do not know enough to fix it.
19:47:09saratogais there a GPIO or ADC pin we can read to check for a charger?
19:47:16saratogaIIRC thats how PP does it
19:48:36gevaertsIn what way does it "try to connect"?
19:49:12gevaertsWe don't do explicit charger detection on any target
19:50:23saratogagevaerts: playback stops, you get the USB screen, then it disconnects and you get the WPS again
19:51:25gevaertsThat's weird
19:51:44*gevaerts looks at code
19:52:06saratogawhere would this be in the code?
19:52:28mc2739 line 675
19:52:42gevaertsusb.c is where everything comes together
19:52:47*gevaerts looks there now
19:53:00gevaertsIt's been a while since I looked at this though
19:55:45mc2739saratoga: I think on amsv1 dbop_din bit 9 can be used for dock/charger
19:56:21 Quit Strife89 (Ping timeout: 240 seconds)
19:56:22saratogamc2739: doesn't that depend on the charging device to short some pin on the dock though?
19:57:06 Join funkyjive [0] (
19:57:41gevaertsOK, the PP code sends USB_POWERED or USB_UNPOWERED on connect (or disconnect), and USB_INSERTED when it detects some activity on the bus
19:57:42 Join Strife89 [0] (
19:57:45saratogaPP seems to have some code for determining if theres a USB controller, but i don't understand it
19:57:46mc2739saratoga: that might be correct, I have confirmed that with dock connections, but I do not have a charge only device
19:57:59gevaertsAMS sends USB_INSERTED directly
19:58:10saratogahow easy is that to fix?
19:58:29funkyjiveI think I'd like to report a bug, but I'd like to run it by you guys to make sure it is not "works as designed"
19:58:44saratogamc2739: as I understand it, theres no analog 'usb connected' wire to sense, you have to actually ask the wire to do something and see if a controller responds
19:59:01funkyjivewhen you have an album with a large number of tracks, you can see that rockbox sorts the track numbers alphabetically and not numerically as it should ...
19:59:20 Join shai [0] (
19:59:52gevaertsfunkyjive: you mean in the file browser?
20:00:43funkyjiveWhen I go Database->Album->Album Name
20:01:05funkyjivegevaerts: thanks for your tip the other day, btw. you are totally the man.
20:01:27funkyjiveI think the problem occurs when you have over 99 tracks ....
20:01:49gevaertsright. I don't know about the database really
20:02:45 Join Llorean [0] (~DarkkOne@rockbox/user/Llorean)
20:03:11funkyjivewhen I look at this particular album, it goes 01,02 ... 08, 09,10,100,101 .... 109,11,110,111,112 .. 119, 12
20:03:46n1sthat is very weird naiming
20:03:47funkyjiveI think I am going to try to side-step it by seeing if I can tag the track numbers to make sure they are 0 padded ....
20:05:22 Join edboyer93 [0] (
20:05:29 Quit BHSPitMonkey (Remote host closed the connection)
20:07:51funkyjiveI do know that it orders the tracks correctly on my wifes Sansa View. So to me it feels like a bug ....
20:08:06 Quit Llorean (Quit: Leaving.)
20:09:39gevaertssaratoga: it may be reasonably easy to change, but I'm not sure
20:11:19gevaertsOne difference is that PP uses a different model. While most drivers still use the usb.c tick task to poll for status changes, the ARC driver sends events on status change
20:11:27funkyjiveit looks like a bug for this has been opened: 11695
20:11:30***Saving seen data "./dancer.seen"
20:12:06 Join ReimuHakurei [0] (~reimu@
20:13:01gevaertssaratoga: if you want to have a go, firmware/target/arm/as3525/usb-drv-as3525.c around line 758 is probably where you want to start. If you get the USB_DEV_INTR_USB_RESET interrupt, usb_detect() in usb-as3525.c should start returning USB_INSERTED instead of USB_POWERED (it currently always says INSERTED)
20:13:25gevaertsThis should of course be cleared when you disconnect
20:18:19*gevaerts test-compiles a quick patch
20:20:44gevaertssaratoga, n1s, mc2739:
20:20:55gevaertsIt compiles, but it's totally untested
20:21:22n1sgevaerts: i'll test later
20:22:07gevaertsMy guess is that all USB drivers apart from ARC have this problem
20:22:50gevaertsAnd I think it could be handled in a way that the driver doesn't actually have to know about this
20:34:53mc2739gevaerts: e200v2 will not enter USB mode with your patch - charging mode only
20:36:20gevaertshm, that's not what I wanted to achieve...
20:38:43gevaertsmc2739: in usb.c, move the #ifdef USB_DETECT_BY_DRV on line 247 down one case, so the USB_POWERED case is handled
20:40:22gevaertsThat would give you
20:40:58gevaertsIt also wouldn't be a fully ready-for-commit fix any more unfortunately. I'm not sure if that case can be enabled for all targets
20:41:07 Quit Mataniko (Ping timeout: 245 seconds)
20:41:30gevaertsAnyone with a nano2g around? Doesit suffer from the same problem?
20:41:45n1sfuze is the same (only charges with patch
20:43:24mc2739gevaerts: with that change I not enter USB mode - testing dock connection next
20:44:35mc2739sorry - remove not
20:44:47gevaertsok :)
20:44:57n1swith the change usb works fine here
20:45:13mc2739dock connection = charging only
20:45:43gevaertswithout any interruption?
20:45:55gevaertsPlayback continues as expected?
20:46:23gevaertsSo all we need is a clean way to do this
20:46:38gevaertsPreferably one that also gets the other targets
20:46:52 Quit Buschel (Ping timeout: 240 seconds)
20:46:55 Join wodz [0] (
20:47:37gevaertsusb.c is a mess :\
20:48:24saratogathanks for looking into this
20:48:27 Quit saratoga (Quit: Page closed)
20:50:15gevaertsah, right...
20:51:10wodzI am totally puzzled why usb works in bootloader and not in rb itself on HD300. I compared gpios and there is no relevant difference between bootloader and rb. Bridge enumerates correctly but can't access disk??? as I get 'Device offlined - not ready after error recovery'
20:51:31wodzanyone have any clue?
20:51:39gevaertswodz: ata driver issues?
20:52:03wodzgevaerts: at what level?
20:52:41gevaertsNo idea really... It sounds as if the hard drive somehow isn't in the proper state
20:52:49TheSevenBuschel (for the logs): With some improvements, I'm getting 206fps quarter and 51fps full RGB in emBIOS with DMA
20:55:13TheSevenwithout DMA, it's a whopping 0.01% slower
20:55:29gevaertsn1s, mc2739:
20:55:36gevaertsThat one should be commit-ready
20:57:26n1sok, testing, i get one hunk failing in usb-as3525.c for some reason
20:57:52gevaertsdid you revert my previous patch?
20:58:03n1syes, it happened with that too
20:58:50wodzwhat does freeze_lock() do?
20:58:54n1spatch says "patch unexpectedly ends in middle of line" and the last hunk fails but i applied it manually
20:59:08gevaertsRight. That's a pastebin issue
20:59:27gevaertsI've seen that too
21:00:30pamaurywodz: it locks the player and freezes the user, in punishment for misuse of rockbox ;)
21:01:15wodzpamaury: that is expaining everything :-)
21:02:44n1sgevaerts: works fine here
21:03:45CIA-7New commit by gevaerts (r28819): Move AMSv1 USB to the USB_DETECT_BY_DRV model, so connecting to a dumb charger works without interrupting playback
21:03:48 Nick balintx is now known as _balintx (
21:04:00gevaertsA similar thing should probably be done for other drivers
21:04:24mc2739gevaerts: USB and dock connection work fine here, too
21:04:41TheSevenwhat exactly does one need to do for this?
21:05:41CIA-7r28819 build result: 5 errors, 0 warnings (gevaerts committed)
21:06:09gevaertsTheSeven: usb_detect() should return USB_POWERED, USB_INSERTED, USB_EXTRACTED or (optionally, less important) USB_UNPOWERED, instead of just USB_INSERTED or USB_EXTRACTED.
21:06:16 Quit user890104 ()
21:06:38TheSevenunpowered being bus activity but no power?
21:06:40gevaertsThe best way to distinguish between those is to use the bus reset interrupt (which you should already handle)
21:06:51 Join Llorean [0] (~DarkkOne@rockbox/user/Llorean)
21:07:14 Quit _balintx (Remote host closed the connection)
21:07:42*gevaerts dislikes that red
21:08:02 Join balintx [0] (
21:12:02gevaertsI'm puzzled. Why does this red happen *only* for m200v4?
21:14:40mc2739gevaerts: could it be that m200v4 does not have HAVE_USB_POWER defined
21:14:53gevaertsah, right
21:15:55gevaertsI actually think we should remove those #ifdefs from the enum
21:16:12gevaertsThey won't be used, but an unused enum value also doesn't really hurt
21:16:44 Quit bmbl (Quit: Verlassend)
21:17:38 Join domonoky [0] (~Domonoky@rockbox/developer/domonoky)
21:20:17CIA-7New commit by gevaerts (r28820): Only return USB_POWERED if USB_DETECT_BY_DRV is actually defined, which might not be the case for e.g. bootloaders (also fix red)
21:20:47 Quit GeekShadow (Quit: The cake is a lie !)
21:22:32CIA-7r28820 build result: All green
21:22:42TheSevenBuschel: exact same FPS values in rockbox with SVN head
21:24:23 Quit fdinel (Quit: Miranda IM! Smaller, Faster, Easier.
21:24:53 Quit n1s (Quit: Lmnar)
21:33:25*TheSeven wonders how to fix those dozens of compilation errors that are still remaining for the ipod classic
21:33:57TheSevenbtw, at which stage should I first commit it?
21:34:01 Quit Kupop (Ping timeout: 255 seconds)
21:34:43gevaertsWhen it compiles I'd say
21:35:42 Join Buschel [0] (
21:35:53 Join saratoga [0] (9803c57f@gateway/web/freenode/ip.
21:36:30*TheSeven wonders if this is a good idea:
21:36:56 Quit funkyjive (Quit: CGI:IRC (EOF))
21:37:06 Join Kupop [0] (
21:37:18TheSevenright now those values are duplicated across a dozen files
21:37:45TheSevenand I need to wrap the task file accesses into a macro, as one can't access them that easily on the classic
21:40:29BuschelTheSeven: good to see rockbox headrev reaches the ultimate DMA emBIOS transfer rates as well
21:40:56BuschelTheSeven: seems like the datarate from S5L to LCD is the limiting factor
21:42:27TheSevenBuschel: do you get the same numbers?
21:42:33BuschelTheSeven: I tried to raise the clock of the IF between S5L and LCD (lowest 3 bits in LCD_CON), but HCLK/4 is the maximum working one −− even when unboosted
21:43:04BuschelTheSeven: no, mine are much higher -> full RGB 129.5, quarter RGB 516.0
21:43:11Buschel(for boosted)
21:43:36TheSevenhm, so what could be affecting it?
21:44:33Buscheleither some different setting in S5L's LCD IF or some setting of the LCD
21:44:36TheSevenBuschel: do you know the bus width of the LCD interface?
21:44:49Buschelshould be 8 bit
21:44:59TheSevenif it's 8 bits, the maximum at 24Mhz should be like ~500fps, if it's serial it should be ~60fps
21:45:10TheSeven(for full RGB)
21:46:30Buschelhow do get those numbers?
21:47:44saratogaweird ubuntu requires texinfo to run
21:48:12kugeldoesn't building gcc require it?
21:51:52 Quit factor (Ping timeout: 240 seconds)
21:55:14BuschelTheSeven: from the preliminary S5L spec the IF should be serial, 3 pin-SPI MPU
21:55:26 Join factor [0] (
21:55:53TheSevenBuschel: doesn't that mean that your fps numbers can't be right?
21:56:52BuschelUtchybann, measured my numbers on his nano as well
21:57:02Buschelthere seem to be differences somewhere...
21:57:38Buscheloops, didn't want to add that "," in my former post
21:57:57wodzTheSeven: look at firmware/target/coldfire/ata-target.h and compare with your pastie
21:58:47TheSevenwodz: that's why i have that #ifndef hell
21:59:05TheSevenIIRC all other targets had identical values, which are needlessly duplicated
21:59:16 Join BHSPitMonkey [0] (
21:59:17TheSeven(and I would have to duplicate them once again)
21:59:23 Quit BHSPitMonkey (Changing host)
21:59:23 Join BHSPitMonkey [0] (~stephen@unaffiliated/bhspitmonkey)
22:00:11 Quit Llorean (Quit: Leaving.)
22:01:28 Join Keripo1 [0] (
22:02:17wodzwell, I don't like this massive ifdefs but I have no cleaner proposition
22:04:02 Quit Keripo (Ping timeout: 265 seconds)
22:04:51TheSevenone would need an "if this is not defined, define it as xyz" preprocessor command :)
22:05:06TheSevencan preprocessor macros expand to preprocessor directives? :)
22:05:43pamauryI don't think so
22:06:07 Quit factor (Read error: Operation timed out)
22:06:14 Quit stoffel (Remote host closed the connection)
22:07:01pamaurywhat do you want to do ?
22:07:07 Join factor [0] (
22:07:29wodzTheSeven: maybe just add comment on top of this stating that if you need something else for your target define this in ata-target.h in target tree and point to colfire dir for reference
22:08:30 Quit froggyman (Remote host closed the connection)
22:08:33TheSevenmaybe one huge #ifndef TARGET_DEFINES_ATA_CONSTANTS around it, instead of individual ifndefs?
22:11:07gevaertsThat may look nicer, but I think it's more error prone
22:11:10wodzlooks hacky for me
22:11:31***Saving seen data "./dancer.seen"
22:12:18BuschelTheSeven: it might be that the S5L spec has an error in the LCD_CON description as well. maybe the LCD IF is not set to HCLK/4, but to HCLK/2. this would perfectly match the measurements 48MHz serial IF = ~6 MB/s = ~130 fps
22:12:26*gevaerts doesn't think that a long list of #ifndef XX #define XX is that bad
22:12:57 Join yahya69 [0] (
22:13:04BuschelTheSeven: still does not explain why your LCD is slower...
22:13:33kugelI'd protect the #defines by a single ifdef using one of them (i.e. not an extra TARGET_DEFINES_ATA_CONSTANTS)
22:13:58gevaertskugel: but what if a target only needs to override one or two?
22:14:50kugelhow likely is that? you'd need to duplicate all then
22:16:33 Quit Kupop (Ping timeout: 240 seconds)
22:18:17 Quit mc2739 (Quit: leaving)
22:18:47 Join mc2739 [0] (~mc2739@rockbox/developer/mc2739)
22:19:12 Part yahya69
22:24:12 Quit soap (Quit: soap)
22:24:30 Quit Keripo1 (Quit: Leaving.)
22:25:07 Nick Ypsy is now known as YPSY (
22:26:46 Quit liar (Ping timeout: 240 seconds)
22:28:07 Join Keripo [0] (
22:29:52 Join liar [0] (
22:37:22BuschelTheSeven: what is your LCD_PHTIME set to?
22:38:17 Join yahya69 [0] (
22:40:35*TheSeven needs to look up the address
22:41:22 Join soap [0] (~soap@rockbox/staff/soap)
22:41:32saratogaif I want to computer the numerical value of a pointer, whats the right cast?
22:41:35saratoga(int) gives me a warning
22:42:08TheSevenBuschel: 0xa1f9602
22:42:18 Join InsDel [0] (
22:42:32TheSeven(if i'm assuming correctly that it's at 0x3c600020)
22:43:28TheSevenah, right, ahb, damn...
22:43:38TheSeventhat would be 0x88
22:44:10 Join MethoS- [0] (~clemens@
22:44:28 Join mystica555_ [0] (
22:44:44Buschelhmm, what happens if you set it to 0x44 or 0x22?
22:44:59 Quit InsDel (Client Quit)
22:46:00TheSeven128 (full) / 512 (quarter) fps at 0x22 \o/
22:46:46Buschelthere we go
22:46:58Buscheldoes 0x11 work?
22:47:07TheSeventhat isn't even allowed according to the datasheet
22:47:16BuschelI know :)
22:47:26TheSevenbut seems to work
22:47:39TheSeven171/682 fps with DMA
22:47:51 Join InsDel [0] (
22:47:56TheSevenlet's try without
22:48:07TheSeven(we might be hitting DMA bandwidth limits :) )
22:48:11 Quit dantje (Quit: Ex-Chat)
22:49:23TheSevenhm, no, the other way round
22:49:34TheSeven350/87 fps
22:49:56wodzhehe, this recalls me comment in some engineering software - "here we define mass for all elements - let's try without"
22:49:57Buschel? 350 fps full, 87 quarter ?
22:50:31kugelwe have too much lcd_puts_* functions
22:50:38kugelI feel bad in adding even more
22:51:09TheSevenBuschel: no, i reversed the order for some reason
22:51:39Buschelahh, ok. maybe this is caused by transmission errors / repeats?
22:51:44TheSevenwhether the framebuffer is in DRAM or IRAM doesn't seem to make a difference
22:52:14Buschelyes, the limiting factor is still the LCD IF
22:52:50TheSevenBuschel: it's probably caused by us needing to check the fifo all the time (i'm writing using a stmcsfd r12, {r0-r3} instruction)
22:53:08TheSevenwhen using DMA, there are hardware signals for this :)
22:53:25TheSevenbtw, this is with burst size 4, data size 16 bits
22:54:05TheSeveni wonder why there is no factor 2 speedup for the 0x22 vs. 0x11 thing
22:54:08*TheSeven goes for 0x00 :)
22:54:24gevaertsInfinite speedup? :)
22:54:45TheSeven195fps full screen, 774fps quarter :)
22:54:59TheSevennow that's insane :)
22:55:00JdGordonbertrik: do you still need testing on the fuze?
22:55:34wodzwhat about graphic glitches?
22:55:43 Join Rob2222 [0] (
22:55:49BuschelTheSeven: that's fast :)
22:55:54TheSeven1291s for a quarter update, 5107s for full screen
22:56:02TheSeven(averaged across 1000 iterations)
22:56:25*TheSeven tries 10000
22:56:49Buschelusing DMA? what when using rockbox svn and just changing LCD_PHTIME?
22:58:40*TheSeven would need to recompile rockbox to test that :)
22:59:03Buscheldidn't you have a fast machine? ;)
22:59:19 Quit Zarggg (Quit: Zarggg)
22:59:24TheSeveni'm compiling rockbox on a 2.0GHz singlecore P4 with DDR1 RAM
22:59:33TheSeventakes several minutes for a complete rebuild of one target
22:59:54TheSevenBuschel: you said the LCD IF would be the bottleneck? :)
23:00:26TheSeven257 fps full screen, 1014 fps quarter from IRAM
23:00:42BuschelI still think so. 195 fps is ~9 MB/s. RAM is much faster.
23:00:58TheSevenBuschel: SDRAM latencies...
23:00:58Buscheloops ;)
23:01:56*TheSeven thinks those frame rates are pretty much ridiculous, considering the LCD's internal panel scanning will probably much slower
23:02:29Buschelhigh rates reduce the CPU load for WPS'es
23:03:01TheSevenapparently scheduling overhead starts to get a significant impact :)
23:07:45saratogacrap just realized i'd have to unroll this code another 2x to make it optimal on ARM11 because of the 64 bit aligned loads thing
23:08:16saratogasomehow i didn't realize that if I made each variable half as big i'd have 2x the possible alignments :)
23:08:19TheSeven16% cpu load even though I use DMA
23:09:24TheSevenaccessing the debugger to read the CPU load made it drop a whole FPS :)
23:10:37TheSevenoh, wait
23:10:57TheSeven16% CPU load by the thread issuing the LCD updates, 84% by the scheduler!
23:11:08TheSeven(waiting for DMA seems to yield instead of sleep)
23:13:00 Quit TheLemonMan (Quit: free(me))
23:15:05wodzgevaerts: you where right - this must be rockbox ata driver which screw up usb bridge on HD300. If I cut off power to the drive and enable it again before enabling USB bridge - it works but panics after cable unplug
23:16:54 Join JdGord [0] (~jonno@
23:16:54TheSevenBuschel: making it sleep properly clips the FPS rates to 99fps for DMA because of the scheduler's 100Hz tick (no adaptive wakeups yet), but reduces CPU load to <1% :)
23:17:55Buschelshouldn't this also reduce current consumption?
23:19:02 Quit {phoenix} (Remote host closed the connection)
23:19:22TheSevenBuschel: probably
23:19:30TheSevenbut it won't be terribly much
23:20:24TheSevennow we'll need to figure out what the minimum allowed values for that register are for the individual LCD types
23:21:17TheSevenwhat does apple set it to?
23:22:31Buschelshould be 0x22 on my nano
23:26:52wodzidentify_info.bin is in little endian or big endian?
23:34:10TheSevensomeone with an ilitek lcd around?
23:34:19TheSeventhey seem to be set to 0x88 indeed
23:37:03 Join thegeek_ [0] (
23:37:27 Join Sudos_ [0] (
23:38:32 Join tchan1 [0] (
23:39:30 Quit hebz0rl (Quit: Leaving)
23:40:02 Join pamaury_ [0] (
23:40:16 Join mortalscan_ [0] (~mortalsca@
23:40:16 Join ack [0] (
23:41:01 Join factor_ [0] (
23:41:21 Join Utchy [0] (
23:41:31saratogaif the arm manual lists a 3 cycle result latency, does that mean using it on the next op incurs 2 stall cycles or 3 stall cycles?
23:41:41saratogae.g. if something is fully pipelined is the latency 0 or 1
23:42:03 Join liar_ [0] (
23:42:47 Quit Sudos (Ping timeout: 240 seconds)
23:42:48 Quit mortalscan (Ping timeout: 240 seconds)
23:42:48 Quit ack` (Ping timeout: 240 seconds)
23:42:48 Quit factor (Ping timeout: 240 seconds)
23:42:48 Quit pamaury (Ping timeout: 240 seconds)
23:42:48 Quit tchan (Ping timeout: 240 seconds)
23:42:48 Quit thegeek (Ping timeout: 240 seconds)
23:42:49 Quit liar (Ping timeout: 240 seconds)
23:42:49 Quit yahya69 (Ping timeout: 240 seconds)
23:42:49 Quit guymann (Ping timeout: 240 seconds)
23:42:49 Quit Utchybann (Ping timeout: 240 seconds)
23:42:49 Quit leavittx (Ping timeout: 240 seconds)
23:42:50 Quit Guinness` (Ping timeout: 240 seconds)
23:42:50 Join leavittx [0] (~leavittx@
23:42:50 Join Guinness [0] (
23:43:21 Quit robin0800 (Read error: Connection reset by peer)
23:43:27*Buschel is puzzled by the logic of LCD_PHTIME and LCD_CON registers
23:43:28saratogaah its 2 cycles of stall
23:43:51BuschelTheSeven: the spec again contradicts what is measured...
23:46:16Buschelwill get some sleep now, see you
23:46:20 Quit Buschel (Quit: ChatZilla 0.9.86 [Firefox 3.6.13/20101203075014])
23:46:21 Quit pamaury_ (Read error: Connection reset by peer)
23:46:25 Join guymann [0] (~charles@
23:51:45 Join JdGordon| [0] (~jonno@rockbox/developer/JdGordon)
23:52:55 Quit JdGord (Quit: Bye)
23:55:41 Quit the_Kyle (Ping timeout: 265 seconds)

Previous day | Next day