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

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

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

Notice: Only Gecko based browsers prior to FF4 support the multipart/mixed "server push" method used by this log reader to auto-update. Since you do not appear to use such a browser, this page will simply show the current log, and not automatically update.

#rockbox log for 2014-08-20

00:07:40 Join Scall [0] (~chat@unaffiliated/scall)
00:10:17 Quit efyx_ (Ping timeout: 255 seconds)
00:10:34 Quit efyx__ (Ping timeout: 244 seconds)
00:12:22***Saving seen data "./dancer.seen"
00:12:40 Quit pystar89 (Ping timeout: 246 seconds)
00:13:25 Join TheGuyWhoIsTheGu [0] (
00:13:28 Quit ender` (Quit: We're not lost. We're locationally challenged.-- John M. Ford)
00:18:02 Join Guest81512 [0] (
00:18:02 Quit Guest50306 (Read error: Connection reset by peer)
00:23:22 Quit TheGuyWhoIsTheGu (Read error: Connection reset by peer)
00:23:24 Quit Guest81512 (Read error: Connection reset by peer)
00:23:59 Join Guest50306 [0] (
00:29:51 Join efyx_ [0] (
00:30:07 Join efyx__ [0] (
00:38:14 Quit lebellium (Quit: ChatZilla [Firefox 32.0/20140814150857])
01:11:54 Join franklin [0] (
01:14:29 Nick SuperBrainAK is now known as DormantBrain (~andy@2001:470:8:a61::5f92:59a1)
01:34:54franklinTheSeven, any news on the timer bug?
01:36:11 Quit efyx__ (Ping timeout: 246 seconds)
01:36:11 Quit efyx_ (Ping timeout: 246 seconds)
01:36:46 Join efyx_ [0] (
01:36:48 Join efyx__ [0] (
01:39:39 Nick DormantBrain is now known as SuperBrainAK (~andy@2001:470:8:a61::5f92:59a1)
01:50:48 Quit shamus (Ping timeout: 244 seconds)
01:51:09 Join shamus [0] (
02:02:29franklinHow can I get the USB HID feature working on ipod classic? defining ENABLE_USB_HID adds the interface, but no functionality
02:03:57[Saint]Are you surprised it wasn't disabled entirely arbitrarily?
02:04:07[Saint]It wasn't enabled for a reason. ;)
02:11:22 Join pystar89 [0] (
02:12:25***Saving seen data "./dancer.seen"
02:12:33franklin[Saint], what was that reason?
02:12:55[Saint] doesn't work, perhaps? Duh.
02:13:03franklinWhy doesn't it work? :)
02:13:13[Saint]Fucked if I know.
02:14:43 Quit bcobco ()
02:15:01[Saint]I...what? Just, no. No.
02:15:14[Saint]Man, the questions you ask sometimes...
02:17:44franklinDo you think gevaerts would know?
02:17:58franklinIt seems like he did a lot of work on the USB stack...
02:18:32[Saint]I think if you were to ask him he'd get very pissed about it...why not try it out?
02:18:55franklinWell, he's away now... "asleep"
02:18:57[Saint](we both know he's not been happy about you singling him out for random arbitrary requests over the past weeks)
02:19:17[Saint]Its not funny. At all.
02:19:35franklinOk then... does anyone else know /anything/ about the USB stack?
02:24:05franklinBut seriously, the only major USB changes have been contributed by him and pamaury
02:26:38franklinAnd TheSeven did some major work on the S5L USB driver, too
02:28:27franklinAnd Rafael Carre made about a million commits for USB, too :)
02:29:49[Saint]Just fix what you can and leave the things that are quite obviously over your head for someone else.
02:35:22franklinwhere do I start? there's no USB code in firmware/targets/arm/s5l8702
02:35:38franklinIn the s5l8700 dir?
02:36:08franklinnever mind... the usb driver in s5l8700 is for nano2g and 6g
02:38:28franklinw00t there's a datasheet!!! :D
02:39:31 Quit Cinos (Ping timeout: 240 seconds)
02:42:05franklinIs the s3c6400x a USB controller or a SoC? Is there one in the Classic?
02:43:09 Join Cinos [0] (
02:44:28[Saint]And you wanted to port the Linux kernel to this thing.
02:44:56franklinI have no tools to take it apart with... broke :)
02:45:01franklin(I am)
02:45:19[Saint]If you think you need to you're quite highly confused.
02:45:30[Saint]Dude, just fix the things you can.
02:45:37[Saint]You're jumping in WAY over your head.
02:45:57franklinAnd I pop right out... buoyancy!
02:46:01[Saint]I have no idea what you're capable of. Apparently neither do you.
02:46:27franklinI've written TONS of low-level code... just very little for ARM
02:48:12franklinIt'd be real nice if half the links on the FreeMyIpod wiki worked :)
02:49:04[Saint]Its a wiki.
02:49:12[Saint]DOn't like it, fix it.
02:50:56franklinWell, anyhow, I am assuming that's the chip in there that's running USB and the coprocessor
02:51:06*franklin pings TheSeven
02:51:11 Join cmhobbs [0] (
02:51:20 Quit cmhobbs (Changing host)
02:51:20 Join cmhobbs [0] (~cmhobbs@fsf/member/cmhobbs)
02:56:04franklinHey cmhobbs
02:57:18franklinTheSeven, is the S3C6400X the coprocessor?
03:00:04 Quit AlexP (Read error: Connection reset by peer)
03:02:46 Quit krabador (Quit: Take the time.)
03:04:56 Join krabador [0] (~krabador@unaffiliated/krabador)
03:05:35 Quit RiD (Ping timeout: 245 seconds)
03:11:17 Join RiD [0] (~RiD@
03:12:18 Quit cmhobbs (Ping timeout: 260 seconds)
03:22:00franklinDoes USB HID work on the nano2g?
03:22:35 Quit ZincAlloy (Quit: Leaving.)
03:23:14franklinHmm... nope
03:23:27franklinWhat does HAVE_USB_HID_MOUSE do?
03:30:07 Quit the-kyle (Remote host closed the connection)
03:32:52 Join the-kyle [0] (
03:40:59 Quit the-kyle (Remote host closed the connection)
03:43:44 Join the-kyle [0] (
03:54:03 Quit krabador (Quit: Take the time.)
03:54:30franklinsaratoga, I broke G#904 into G#919 and G#920 now
03:54:36fs-bluebot3Gerrit review #904 at : 3Improved iPod CHIP8 keymaps and fixed a bug that prevented it from working by Franklin Wei
03:54:36fs-bluebot3Gerrit review #919 at : 3Fixed a chip8 bug by Franklin Wei
03:54:37fs-bluebot3Gerrit review #920 at : 3Improved CHIP8 iPod keymaps by Franklin Wei
03:54:48*franklin slaps fs-bluebot for spamming :)
03:54:49fs-bluebotfranklin: ouch!
03:56:10 Join jhMikeS [0] (
03:56:20 Quit jhMikeS (Changing host)
03:56:20 Join jhMikeS [0] (~jethead71@rockbox/developer/jhMikeS)
03:57:01 Join cmhobbs [0] (~cmhobbs@fsf/member/cmhobbs)
03:57:25cmhobbsfranklin, a late hello :D
03:58:35franklinso, cmhobbs, have you tried plugin-writing yet?
04:02:03cmhobbsunfortunately not
04:02:05cmhobbsi haven't had time
04:02:10cmhobbsi'm on holiday with my family this weekend though
04:02:15cmhobbsi hope to roll up my sleeves and get into it
04:02:27cmhobbsi'll have two devices there because my dad's got one as well
04:02:42franklinsome of the plugins are terrible examples, they probably haven't been updated in years
04:03:05franklinSo I'd humbly say that 2048 is probably the best example of plugin structure in there
04:05:37franklinThere are so many HZ assumptions in the plugins, especially in the DOOM code
04:12:27***Saving seen data "./dancer.seen"
04:17:11 Quit amiconn (Disconnected by services)
04:17:11 Join amiconn_ [0] (amiconn@rockbox/developer/amiconn)
04:17:11 Quit pixelma (Disconnected by services)
04:17:11 Join pixelma_ [0] (pixelma@rockbox/staff/pixelma)
04:17:14 Nick pixelma_ is now known as pixelma (pixelma@rockbox/staff/pixelma)
04:17:14 Nick amiconn_ is now known as amiconn (amiconn@rockbox/developer/amiconn)
04:23:05jhMikeS[Saint]: I just saw the dump you posted, thanks. It looks like a bunch of startups.
04:32:07cmhobbsi'll take a look at it for sure
04:32:36franklinWell, have fun :)
04:37:52JdGordonjhMikeS: ploco posted a patch which fixed it
04:40:10 Quit franklin (Remote host closed the connection)
04:43:33jhMikeSwhere's that
04:44:30fs-bluebot3Gerrit review #894 at : 3RaaA: Android work around the Kitkat ART crashes by Chiwen Chang
04:45:13jhMikeSso it's actually related to that?
04:45:34JdGordondunno, didnt look at the diff, just tried it and it works
04:45:37JdGordonso yeah maybe
04:45:50JdGordonneed someone on an old android to try master
04:46:33jhMikeSthe pc is obviously garbage when it crashes
04:48:13jhMikeSI tried to set that up but it's giving me a headache with the tool crashing when trying to make the apk, no matter the version
04:55:24jhMikeSIf it were directly in the changed code, I would expect every target to die at boot but that's not the case obviously.
05:05:33 Quit jhMikeS (Ping timeout: 240 seconds)
05:11:28 Quit ender| (Read error: Connection reset by peer)
05:11:29 Quit RiD (Quit: A good plan today is better than a perfect plan tomorrow.)
05:11:50 Join ender| [0] (krneki@2a01:260:4094:1:42:42:42:42)
05:14:35 Quit steffengy (Disconnected by services)
05:14:36 Join steffengy1 [0] (
05:54:12 Quit TheSeven (Ping timeout: 250 seconds)
05:55:38 Join TheSeven [0] (~quassel@rockbox/developer/TheSeven)
06:01:13 Join jhMikeS [0] (
06:01:21 Quit jhMikeS (Changing host)
06:01:21 Join jhMikeS [0] (~jethead71@rockbox/developer/jhMikeS)
06:12:29***Saving seen data "./dancer.seen"
06:17:25[Saint]jhMikeS: JdGordon: didn't you see how aggressive that patch is?
06:17:37[Saint]Oops - double highlight.
06:17:41JdGordonI didnt look at the details
06:17:54[Saint]There's no way it'd be accepted
06:18:13[Saint]Also - neither of us are using ART. ;)
06:19:12jhMikeSbut it does point to some other sort of issue
06:19:43[Saint]Let me start my shit up and I'll get you the android build requiremnets
06:21:22jhMikeSif that patch fixes that crash then it's not really in what I did itself but some side effect (a side effect that should not exist in the design itself)
06:22:14jhMikeSsomewhere it ends up loading a messed up jump buffer since that PC value is obvious nonsense
06:24:58[Saint]Android SDK Tools, Android NDK Platform-tools, Android SDK Build-tools 19.1, SDK Platform 19
06:26:02jhMikeSthat's what I got in place but aapt dies with "illegal instruction"
06:27:14 Quit cmhobbs (Ping timeout: 244 seconds)
06:27:46jhMikeSgranted my vm is sort of ancient
06:27:47[Saint]just updating my SDK tools and NDK platform tools now to see if that's it.
06:28:46jhMikeSthe kernel 2.6.26 lenny (hell if I want to upgrad that crap again right now if not strictly necessary)
06:31:39[Saint]hmmmm...actually. I just updated and it spat an error about missing zipalign
06:31:50[Saint]seems like paths may have been juggled around.
06:32:49[Saint]Ahhhh....also need build-tools rev. 20
06:32:51jhMikeSI got me a zipalign
06:33:16jhMikeSI just grabbed revision 19 off android sdk
06:33:28jhMikeStried 20 as well with no difference
06:34:37[Saint]sec - still chasing changed requirements.
06:35:12jhMikeSI was piecing it together it wanted 19 and 19.1.0 for this stuff
06:36:31jhMikeSI got ndk 10 and it only goes to 19
06:37:02[Saint]looks like android moved some shit around. bah.
06:41:42jhMikeSthe automatic build spits out some warnings that aren't shown in the table
06:42:03jhMikeSactually errors
06:43:25[Saint]ok - I can build with current toolset using the versions given above
06:43:53[Saint]just needed to make a simlink for zilalign from build-tools to tools
06:45:35[Saint]That is, for the record:
06:46:20[Saint]Android SDk Tools 23.0.2, Android SDK Platform-tools 20, Android SDK Build-tools 19.1, and SDK Platform 19
06:46:28jhMikeS[Saint]: explain why that patch is "aggressive". I don't know that part of things enough to know what's going on there.
06:46:46[Saint]its coring out the entire battery metering system
06:46:59[Saint]granted the host can do this itself..but
06:47:22[Saint]meaning, we'll never be able to display the battery ststus
06:48:16 Join ploco [0] (dce9b7f9@gateway/web/freenode/ip.
06:48:58[Saint]Oh - and, for the record, openjdk7 rdk/jre
06:49:13[Saint](and a 64 bit multiarch system ofc)
06:49:28jhMikeSwait, if that patch looks for android-5, why is my configured build wanting android-19?
06:49:31ploco[Saint]: have a look again, nothing is disabled in that patch now. battery works
06:49:50[Saint]oh, my mistake. I'm stuck in the past I guess.
06:49:54jhMikeSI mean, changes it from android-5
06:50:55[Saint]we most certainly need platform 19
06:51:07[Saint]mail went out about that yonks ago.
06:51:31[Saint]just seems like we're looking for zipalign in the wrong place
06:51:52[Saint]easily fixed, and easily worked around with a symlink
06:52:28jhMikeStools/configure still has much older stuff named
06:53:21[Saint]that may be setting the minimum platform API
06:53:36[Saint]ie. the minimum version we'll run on
06:53:42[Saint](he says without loking at all)
06:54:39[Saint]anyhoo - I can build with current tools with a slight workaround - so I'm not sure what's up with your system.
06:54:51[Saint]but my SDK/NDK are *much* much newer.
06:57:47[Saint]and my kernel is orders of magnitude newer as well, but I sincerely doubt that's it.
06:57:49jhMikeSI pulled the latest stuff
06:58:19[Saint]you must've been fighting pretty hard to keep that kernel version whilst still keeping the rest of the system up-to-date.
06:58:19jhMikeSI'll change explicitly to 19 to make sure it even builds
06:58:23[Saint](assuming it is)
06:58:57jhMikeSit ain't. it builds rockbox, that's that's used for and hasn't had any issues until trying this
07:24:18 Quit ploco (Quit: Page closed)
07:36:14 Join kugel [0] (~kugel@rockbox/developer/kugel)
07:36:45 Join xaioscn [0] (
07:42:59xaioscnHi All, I'm seeing a bit of weird behavior on my Sansa Clip Zip 4GB running 3.13 of Rockbox. I have music on both the Internal Storage and on a 32gb SDHC Class 4 San Disk microSD card. When I select Database->Artists->All Tracks with Shuffle on it seems to find all 2566 Songs on both the Internal and External Storage, however when it goes to play them it will skip over any songs on the External
07:43:00xaioscnStorage to the next avaliable song on the Internal Storage. I am able to select any song (or even folder) on the External Storage and play it without issue. I seem to recall this working in the past, but I did recently load up a fair number of new songs (several hundred in fact in multiple folders on the SD card). This MP3 player does live in the car constantly plugged in so I am not ruling
07:43:01xaioscnout hardware issues due to the extreme heat and cold it has been exposed to in the past (I use the auto-on and auto-off features frankly #1 killer feature of Rockbox). Any ideas? I'd also settle for the ability to have it generate a playlist soley off of the external storage if that's a work around. Thanks.
07:46:43 Join saratoga_ [0] (123e1c18@gateway/web/freenode/ip.
07:46:51saratoga_xaioscn: could try rebuilding the database
07:50:58xaioscnsaratoga_ Thank you that did the trick! Should I force it to rebuild the database everytime I load up new songs?
07:51:56saratoga_shouldn't be necessary unless something goes wrong
07:53:15xaioscnAwesome, I rarely load new songs (and it looks like it did find about 1,000 new ones) so I'm good to rock and roll! Thanks for all your hardwork on the project, it really makes these little Clip Zip's shine they work great in the car on shuffle with that auto on/off feature and big SD cards.
08:05:54 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
08:12:32***Saving seen data "./dancer.seen"
08:21:09 Join LinusN [0] (
08:21:12 Join ender` [0] (
08:35:53kugel[Saint]: the ART patch isnt agressive
08:36:19[Saint]it /was/ - we've already established that.
08:37:32 Quit pamaury (Ping timeout: 240 seconds)
08:41:26 Quit saratoga_ (Quit: Page closed)
08:43:47jhMikeSI'm still having stupid problem. Is it possible to try the sigaltstack threads and newer sdk in root.make alone? (just to root.make part of the patch)
08:52:08 Join bertrik [0] (~quassel@rockbox/developer/bertrik)
09:00:20kugeljhMikeS: just #define HAVE_SIGALTSTACK_THREADS in $builddir/autoconf.h
09:01:29jhMikeSI would except I can't get an apk to make. make it seems bombs out with 'illegal instruction"
09:02:10kugelyou can install sdk platform 19 and sdk tools 19.1.0 from the sdk manager (run $sdkdir/tools/android)
09:03:04kugelif it dies in zipalign you could comment try to skip that build step, zipaligning is optional
09:03:05jhMikeSI've already got all that
09:03:23 Quit bertrik (Remote host closed the connection)
09:03:55jhMikeSsays "make: *** [/home/user/rockbox/builds/android/bin/resources.ap_] Illegal instruction"
09:05:27kugelthat combination worked on my box, but it is amd64 (yours seems to be x86)
09:05:37TheSevenoh my goodness
09:05:45jhMikeSit's sort of not a box
09:06:50 Join Zagor [242] (~bjst@rockbox/developer/Zagor)
09:14:12 Join petur [0] (5bb7304d@rockbox/developer/petur)
09:14:34 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
09:22:11 Quit scorche` (Ping timeout: 250 seconds)
09:24:11 Join scorche [0] (~scorche@rockbox/administrator/scorche)
09:27:20*jhMikeS thinks its weird that r3 is treated non-volatile
09:27:41jhMikeSthe abi is differnt
09:37:22 Quit scorche (Ping timeout: 244 seconds)
09:38:26jhMikeSalot of pure asm is not going to work properly unless it stacks r3 as well
09:41:13jhMikeSanother thing: hosted targets should never use our errno.h, never ever. it has to match the host, not our own defs
09:51:04*[Saint] thought he made it clear a 64bit machine was required
09:51:11[Saint]Sorry if I didn't...
09:51:43[Saint]Needs a 64bit multiarch system.
09:54:54*jhMikeS wasn't clear
09:55:37[Saint]I thought *I* wasn't clear.
09:55:58[Saint]...then who was phone?
09:56:02jhMikeSfunny though, why? It's gets through everything else
09:56:43[Saint]I honestly have no idea.
09:58:04jhMikeSexactly what is required to be 64bit?
09:58:44jhMikeSand the fact I can download 32-bit sdk/ndk
10:00:44[Saint]Maybe I'm confusing an unrelated issue - someone with a lot more of a clue should likely confirm or deny now I'm doubting myself.
10:01:08[Saint]It just occurred to me that I may be confusing opinion and fact.
10:01:56jhMikeSI wouldn't expect the tools to be doing anything if they were only 64-bit arch but they run
10:01:58[Saint]kugel is probably the one with the absolute oversight in this area.
10:02:37jhMikeSso, I'm just examining a objdump and some things stike me
10:02:43kugelthe sdk and ndk should work on 32bit machines
10:04:25kugeljhMikeS: about errno, the host errno isnt multiplexed across our threads
10:04:41jhMikeSI did it already
10:05:20jhMikeSadded errno save and restore to the context
10:05:46kugelbut for our code to use this errno it needs our header, no?
10:06:18jhMikeSjust the function redirect
10:06:41 Join lebellium [0] (
10:07:03kugelthe function redirect is in our header?
10:07:38jhMikeS#define errno (*__errno())
10:08:43jhMikeSthread.c needs to see the real one of course, not ours
10:09:38kugelthe above define is in our header
10:10:43jhMikeSthat should mask the host one everwhere else
10:10:51jhMikeSbut don't use the values, use the system ones
10:11:05kugelyou said "hosted targets should never use our errno.h"
10:11:39jhMikeSI should have clarified what I meant
10:12:11jhMikeSuse the hosts values but our errno #define
10:12:27jhMikeSunless using host threads
10:12:28kugelhow do you do that? you get both with include errno.h
10:12:34***Saving seen data "./dancer.seen"
10:12:56jhMikeSI'm not sure it's right at this time
10:13:50kugel(the alternative would be to actually set errno at context switch, it has overhead unfortunately)
10:14:18jhMikeSwell, you'd better or else it not going to be right
10:17:51jhMikeSit's just a "maybe" if it isn't considered context and it would mix up with the hosts own functions' setting of it
10:20:58kugelwithin the single host thread (with N rb threads) there is only a single errno, we can do whatever we want. only library functions ever set it (we don't call these from our context switch routines), and only in case of error (it's never set back to 0 by the host)
10:27:18jhMikeSsure, that's where the issue starts, if one of ours sets it, we context swich, then another sets it and switches back, now it's probably wrong
10:28:35kugelis that with setting errno on CS or without?
10:28:51kugel(without things will be fishy if the __errno() override isn't seen)
10:30:21jhMikeSbut, I want host functions' setting of it to be able to blend with our functions that set it.
10:30:42jhMikeSso, nvm, code should see the hosts errno, not ours
10:31:20jhMikeSI thought it out earlier when doing it then had to rethink it again :)
10:32:53jhMikeSand CS just saves and restores the host value for each "thread"
10:32:56kugelagain, the host errno isn't multiplexed
10:33:12kugelso we'd have to do it on CS
10:33:17*kugel is too slow
10:33:18jhMikeSI put that in
10:35:27 Join AlexP [0] (~alex@rockbox/staff/AlexP)
10:35:35jhMikeSI see "__errno@plt". maybe they gave it a similar name?
10:35:51jhMikeSplt == platform I'm guessing
10:36:23jhMikeSso perhaps it's fine
10:36:25funmanProcedure Linkage Table?
10:36:43jhMikeShaha, ok
10:37:26jhMikeSI see filesystem functions there too
10:37:32jhMikeSthat's a call to android code then?
10:37:32 Join kuldeepdhaka [0] (~kuldeepdh@unaffiliated/kuldeepdhaka)
10:38:12jhMikeSsection .plt
10:38:55jhMikeSI'm also wondering when r3 has to be stacked and when not, some functions do it, some don't
10:39:15jhMikeSeven if r3 is clobbered
11:00:50 Nick SuperBrainAK is now known as DormantBrain (~andy@2001:470:8:a61::5f92:59a1)
11:04:36 Join scorche [0] (~scorche@rockbox/administrator/scorche)
11:16:23 Join scorche` [0] (
11:16:33 Quit scorche (Disconnected by services)
11:16:33 Quit scorche` (Changing host)
11:16:33 Join scorche` [0] (~scorche@rockbox/administrator/scorche)
11:26:37jhMikeSthe problem has to do with the compiler moving stuff across the save/load, where before it just happened to be ok
11:28:19*jhMikeS 's been waiting for the this day to come for years
11:29:07JdGordonjhMikeS: so you fixed all the android crashes? :)
11:29:56jhMikeSI didn't fix it yet. I see the issue though and what changes in the assembly output that messes it up.
11:30:54jhMikeSthe context save/restore really isn't portable. I was just working on a patch that swaps it one function call in one spot which would fix it.
11:31:03jhMikeSthat's why that patch alleviates the problem
11:33:55jhMikeSit save r5 before r5 is loaded with the thread pointer, then loads r5 with load_context, then assumes it still point to the thread, but it doesn't
11:37:34kugeljhMikeS: which android crash is this? the one under ART?
11:37:37jhMikeSit means the basic design that's been there forever really has to be corrected or else this stuff will pop up here and there
11:38:07jhMikeSkugel: the ART patch correct it but due to different context saving / restoring in switch_thread
11:38:51kugelyea, so there is a problem with our custom CS? I remember under maemo it was also problematic
11:39:09kugelbut when I look at it it seems safe for EABI
11:39:50jhMikeSyes there is. the compiler doesn't know registers were clobbered
11:40:58kugelIIUC because the CS is a function call it has to stack r0-r3, all other registers need to be stored in the context struct, or am I wrong?
11:41:34jhMikeSI have frankly been surprised for awhile it's worked at all
11:42:45kugelenlighten me if/when you know what'S wrong with it exactly, because I cant see it
11:43:31jhMikeSno, it saves a reg, then the compiler loads that reg with something else, loads the new context, and the compiler uses the reg after that but the load clobbered what it (rightfully) assumes it still contains
11:44:16kugelit loads the regs just before the return of switch_thread(), should be safe?
11:45:04jhMikeSyou don't know what order it will place things, really
11:46:36jhMikeSr5 gets clobbered but then it loads errno after that. I suppose moving errno load to before the load might patch it up for now.
11:47:34kugelwhere is errno loaded?
11:49:12*kugel looks at old code
11:50:00jhMikeSin thread.c in thread_store/load_context
11:50:02kugelyes, i think load_context() must be last
11:50:58kugelreordering that might help, the compiler shouldnt be able to reorder itself becuase load_context() is asm volatile
11:51:15kugelhowever, I think the ART crashes are older than your reworks
11:53:02jhMikeSI remember reading somewhere it's not guaranteed it won't move things and it doesn't know certain regs are changing
11:54:09kugelit can only reorder memory accesses, but not function calls and IIRC not asm volatile blocks
11:54:46kugel(i.e. a asm volatile block should be treated as a function call w.r.t to reodering optimizations)
11:55:38jhMikeSif that's so, then r4+ are safe to assume don't get clobbered if not listed
11:55:59jhMikeSchanging the order appears to make it look good though
11:56:34jhMikeSso, I guess I can just push that one right now
11:58:48jhMikeSthe rework I've got also gets rid of the start pointer and just sets it all up to directly jump it to startup code
11:59:08kugeljhMikeS: i think the registers still need to be listed
11:59:40jhMikeSthey aren't and never were
11:59:53jhMikeSnothing to to do with anything I ever did
12:00:43kugelright, that might be a bug, and may have worked because it's just before the end of switch_thread() where gcc doesnt do anything to the regs anymore
12:01:30jhMikeSI'm just putting it back that way for now :)
12:01:37kugelhm no, wait, listing them would make gcc stash the regs (and restore afterwards) right?
12:02:24jhMikeSI think that's correct. There's a big note "only!" for r0 which I take it somebody went through the trouble.
12:03:20kugelyes for clobber. maybe r4-r14 should be in hte list of output operands
12:03:43kugelwhich is empty there
12:05:17jhMikeSA single point of swapping it also makes it easier to not save or restore if the thread doesn't change
12:07:09kugelwin32 fiber threads to true swapping, store context is a nop there
12:08:39fs-bluebotBuild Server message: 3New build round started. Revision 5fb3702, 253 builds, 30 clients.
12:09:10*jhMikeS saw that while digging around
12:11:34*jhMikeS hates what he sees ldm, then pop back to back
12:12:35***Saving seen data "./dancer.seen"
12:16:37fs-bluebotBuild Server message: 3Build round completed after 479 seconds.
12:18:46jhMikeSI wonder if you could do nothing in store/load but just tell it you did
12:34:18jhMikeSheh, that does indeed work with the exception of sp
12:34:58jhMikeSno redundant loads or stores, the function call takes care of it
13:03:40 Join Guest81512 [0] (
13:06:02 Quit Guest50306 (Ping timeout: 240 seconds)
13:13:32 Quit kuldeepdhaka (Ping timeout: 246 seconds)
14:12:39***Saving seen data "./dancer.seen"
14:39:39 Join ZincAlloy [0] (
14:58:31 Quit olspookishmagus (Ping timeout: 245 seconds)
14:59:49 Join olspookishmagus [0] (
15:00:13 Nick olspookishmagus is now known as Guest48784 (
15:02:30 Join RiD [0] (~RiD@
15:06:48 Join amayer [0] (
15:56:47 Quit Zagor (Quit: Clint excited)
16:04:36 Part LinusN
16:12:40***Saving seen data "./dancer.seen"
16:25:43 Join kuldeepdhaka [0] (~kuldeepdh@unaffiliated/kuldeepdhaka)
16:36:47 Quit lebellium (Quit: ChatZilla [Firefox 32.0/20140818191513])
16:40:31 Quit kugel (Ping timeout: 244 seconds)
16:44:41 Join franklin [0] (
16:49:45 Join kuldeepdhaka_ [0] (~kuldeepdh@unaffiliated/kuldeepdhaka)
16:50:44franklinAnyone know why the USB HID feature doesn't work on iPod Classic?
16:52:54 Quit kuldeepdhaka (Ping timeout: 255 seconds)
17:06:22franklingevaerts, you wrote a lot of the USB stack, didn't you?
17:06:30gevaertsI did
17:06:56pamaurywhat is the usb core on the ipod classic ?
17:08:17*franklin wonders if it would be possible to play back PCM audio through the piezo speaker
17:09:52franklinI tried compiling with ENABLE_USB_HID, it adds all the features you would expect (remote_control), but it doesn't work :(
17:10:17gevaertsThat might be why it's disabled
17:10:34franklinhaha I know... why doesn't it work?
17:10:58 Join TheGuyWhoIsTheGu [0] (
17:11:42pamauryisn't the s3c6400 a synopsys core ? with which we have some many problems ?
17:12:00TheGuyWhoIsTheGuThe Rockbox Android Daily Build doesn't work after installing it, it stops working, like ot. Force closes.
17:12:32franklinpamaury, the datasheet has "Samsung" written on it :)
17:13:23franklinMaybe it's from Samsung :P
17:13:46gevaertsThe soc is from samsung, yes
17:13:50pamauryit doesn't mean anything, USB cores usually are IPs sold by very few companies
17:13:51franklinThere is a datasheet, though
17:13:58pamauryjust checked, it's a synopsys core
17:14:00franklinAh I see
17:14:07franklinJust samsung-branded
17:14:51gevaertsTheGuyWhoIsTheGu: the android port seems to be in a fairly bad shape these days, mostly involving newer android versions
17:16:18franklinSo, no one is working on supporting it?
17:16:24TheGuyWhoIsTheGuOh ok, well ill have to wait for a functional build
17:16:44franklinTheGuyWhoIsTheGu, expect it between now and never :(
17:17:03franklinThere's not that much work being done on it :(
17:17:05pamaurymany tried, no one succeeded, TheSeven did a complete rewrite for emcore and someone attemped a rockbox port, I don't know what is the state of this
17:17:31franklinSo TheSeven would be the best to ask?
17:17:37TheGuyWhoIsTheGuHahaha well ok, no one has an older build??
17:17:47pamauryfranklin: yes
17:17:51franklinOf course there are archived builds :)
17:18:12franklinBut apparently, "All Quassel clients vanished from the face of the earth..."
17:18:27TheGuyWhoIsTheGuSo no links :(
17:19:13franklinSo how would I dump the PCM data to the piezo?
17:19:35franklin(I am 100% serious)
17:19:56gevaertsTheGuyWhoIsTheGu: I'm not at all convinced that older builds would help you
17:20:13franklinThe problem is the newer Android version?
17:20:41gevaertsLikely, yes
17:20:49TheGuyWhoIsTheGuI remember that I had a working build of Rockbox in my android phone with 4.4.4
17:20:51gevaertsWell, sitll a bug in rockbox
17:21:07franklinSo how would I dump PCM data to the piezo?
17:22:37franklinIt /is/ possible:
17:32:47franklinAnd there's an expired patent for something like it:
17:37:15franklinSo, how do I do it in Rockbox?
17:37:25TheGuyWhoIsTheGuWhere could I upload a working build of Rockbox for android 4.4.4 because I found it. I have the APK
17:37:54franklinWhere'd you get it?
17:38:43TheGuyWhoIsTheGuI had it on my phone and a backup of my sd card on my PC
17:39:53franklinSome file-sharing site, maybe? Or you could ask the RB website admins to host it if you like
17:40:13TheGuyWhoIsTheGuOk I'll ask
17:40:31copperdropbox, google drive
17:40:53franklinmega will taken down soon enough...
17:42:00 Quit petur (Ping timeout: 246 seconds)
17:42:18gevaertsRockbox won't host it for you
17:42:40 Quit Guest48784 (Ping timeout: 272 seconds)
17:42:43franklinWill they link to it?
17:43:46 Join olspookishmagus [0] (
17:43:48TheGuyWhoIsTheGuWhere do I ask?
17:44:04franklinI bet gevaerts knows :P
17:44:10 Nick olspookishmagus is now known as Guest76939 (
17:44:29gevaertsFirst of all, where did you get your build from? Is it from unmodified sources?
17:44:58gevaertsBecause if it isn't, you need to provide the source as well
17:45:25TheGuyWhoIsTheGuFrom Rasher's website
17:45:27franklinugh GPL issues...
17:45:33gevaertsOK, that's unmodified then
17:45:42gevaertsfranklin: this isn't an unfortunate side effect...
17:45:46TheGuyWhoIsTheGuIt's an old revision
17:46:08franklinGPL is great sometimes... and just annoying other times
17:46:22gevaertsfranklin: yes, and this is one of the times where it's great
17:46:31gevaertsYou could post in the "unsupported builds" section on the forums
17:46:39franklinGreat for the people in the future... but not for us :)
17:47:01 Join TheGuyWhoIsTheG2 [0] (b8a34d2d@gateway/web/freenode/ip.
17:48:24TheGuyWhoIsTheG2I'm the one with the working Android build, but I'm connected on my computer. So do I post it in the forums with a link to Google Drive?
17:49:18gevaertsThat should work
17:49:36gevaertsAdd the revision in the post
17:50:09franklinI think you could attach it, if the forum allows .apks
17:50:14franklin(I don't think it does)
17:54:03franklinWow MEGA is great
17:58:59 Quit TheGuyWhoIsTheGu (Read error: Connection reset by peer)
18:02:21TheGuyWhoIsTheG2The link to the post
18:02:43 Quit pamaury (Ping timeout: 250 seconds)
18:05:12TheGuyWhoIsTheG2I heard that the iPod Classic is a very fast device for Rockbox, is that true and is there any suggestions for other devices?
18:05:33franklinYes it is fast, and is rather complete
18:05:45franklinTrue, there are tons of small to medium-sized bugs
18:06:22TheGuyWhoIsTheG2So if I had to build a device today, an iPod Classic would be a great option?
18:06:25franklin(like the user timer not working, which is why DOOM is broken :( as well as the difficult install process)
18:06:30franklinbuild a device?
18:07:14franklinIf you want a complete rockbox port, get a sansa or an earlier ipod
18:07:35franklinbut classic looks quite pleasing and runs rockbox quite nicely other than for a few bugs
18:07:47franklin(though install is a bit tedious)
18:07:55TheGuyWhoIsTheG2I had an iPod video and it was slow while using the eq that's why I'm asking
18:08:30frankliniPod classic sounds perfect
18:08:39franklineven with 10-band EQ
18:08:55*gevaerts finds that hard to believe
18:08:59franklinit does :)
18:09:16gevaertsIf you use ten bands, you're almost certainly doing something wrong, in which case it *can't* sound perfect
18:09:32franklinI count 10 bands
18:10:04TheGuyWhoIsTheG2When I used a lot of bands the playback and the device was slow and buffering occured
18:11:39gevaertsIn practice, the ipod video is more or less the slowest device you can get
18:12:04franklinI have a nice 216 MHz CPU in mine
18:12:12gevaertsYou don't
18:12:30gevaertsIt's the same CPU as other old ipods and sansas, but it has a much bigger screen than those, and that does make a difference
18:12:30TheGuyWhoIsTheG2Wow, for the first Rockbox device to have, I took the worse one
18:12:39franklinThe classic?
18:12:42***Saving seen data "./dancer.seen"
18:12:54gevaertsfranklin: video
18:12:55franklinAh you're thinking of the video
18:13:14franklinMy /classic/ has a nice 216MHz CPU
18:13:21gevaertsTheGuyWhoIsTheG2: it's not a bad device. It's just a bit underpowered if you want lots of DSP or CPU-intensive codecs
18:13:43*gevaerts would like to point out that MHz is not a useful unit of CPU performance
18:13:55franklingevaerts, it is in some cases
18:14:11TheGuyWhoIsTheG2Oh ok, well maybe I should have learn a little bit more about EQs and how they work
18:14:18franklinFor example, a 4MHz CPU is almost always going to slower than a 16MHz one
18:14:27 Quit Kohlrabi (Quit: - Chat comfortably. Anywhere.)
18:14:35TheGuyWhoIsTheG2Do you guys have any plans for an iOS app, I have an old iPhone 3G taking more and more dust, it's jailbroken but all the tweaks need a newer version :(
18:14:36 Join Kohlrabi [0] (
18:14:41franklinUnless that 4MHz is a CISC and the 16MHz is some crazy VLIW RISC
18:14:47gevaertsThe 4 MHz one could be a pipelined 32 bit thing, while the 16MHz one could be an 8 bit slow design
18:14:54gevaertsAlso, RAM matters too
18:15:15gevaertsTheGuyWhoIsTheG2: I'm not aware of any real efforts in that direction
18:15:30franklinTheGuyWhoIsTheG2, $99 for a dev membership
18:15:45gevaertsSome people have talked about it over the years, but I doubt if anyone ever even tried to get started
18:15:50franklinIf you want to cough that up and start working on it, go for it
18:16:02TheGuyWhoIsTheG2franklin: that's the problem
18:16:27TheGuyWhoIsTheG2but do you need a dev membership to make a jailbreak app?
18:16:53gevaertsYou don't even need one for a non-jailbreak device. You only need one if you want other people to be able to run it
18:17:00franklingevaerts: for comparing similar CPUs (say intel core i3s), clock freq is pretty good
18:17:01gevaertsSo you don't need one to get started
18:17:06franklin(for comparing)
18:17:21franklingevaerts, you need a Mac
18:17:23gevaertsfranklin: yes, but in the rockbox context, there's a *lot* of variation
18:17:27franklinOr a hackintosh
18:17:37franklingevaerts, not much... most ARM
18:17:50franklinSome MIPS/SH/coldfire
18:19:25TheGuyWhoIsTheG2ok I see why there no real effort for iOS devices
18:19:37gevaertsfranklin: within the ARM devices, there can be a *huge* difference in performance/MHz. Up to almost 9x in some cases...
18:19:53gevaertsHave a look at the table on
18:20:43gevaertsGigabeat S plays -c5000 APE at 112% realtime with a 528MHz CPU, PP502x devices play it at 2% realtime with 80MHz
18:21:00gevaertsSo that's 55x the performance with 6.6x the clock cycles
18:21:15franklinWhy? cache? ram?
18:21:21gevaertsCache is one part
18:21:33gevaertsRAM speed is another
18:22:37gevaertsThe PP502x is, but that's not a factor here
18:22:48gevaertsThe APE decoder doesn't use that
18:23:19franklinShould it?
18:23:29TheGuyWhoIsTheG2Have you guys heard about ViPER4Android
18:23:42gevaertsI don't think it makes much of a difference. APE is stupid anyway :)
18:24:42gevaertsYou're paying *huge* CPU prices (which translate directly to battery runtime) for a filesize gain of maybe a few percents over flac
18:25:11franklin2% realtime :O
18:27:52franklingevaerts, would ipod classic be any better on APE?
18:28:20gevaertsI'd expect it to be fairly similar to nano2g
18:28:33franklinah so about 2% realtime :)
18:28:56franklinwith c5000
18:29:29franklinwow slow
18:29:41gevaertsAPE c5000 is 5% smaller than flac at -8 for the rockbox test track
18:30:20*franklin is considering a Gigabeat S for his next DAP
18:30:43gevaertsIf you think a classic is a pain to install, you haven't tried those :)
18:30:57franklinIt's a breeze to me
18:31:10franklinDownload, chmod +x, exec
18:31:12gevaertsThey have a habit of reformatting the data partition for no reason if you look at them in an odd way
18:31:26franklinlook in them?
18:32:26gevaertsIf they don't like the partition layout or the firmware partition contents or something like that, they helpfully clear everything and tell you you need to restore
18:33:06gevaertsAlso, some of them don't like dual boot while others dislike single boot
18:33:12jhMikeSgevaerts: mine never does that unless I trash it
18:33:24gevaertsYes, some of them are friendly :)
18:33:31jhMikeSit's caused by the microsoft firmware
18:33:31gevaertsYou just won't know in advance
18:33:37 Join rela [0] (~x@pdpc/supporter/active/rela)
18:33:44franklinwell, so classic is the best ATM?
18:33:51jhMikeSit would do it if dual booting but never otherwise
18:33:53 Join kugel [0] (~kugel@rockbox/developer/kugel)
18:33:59franklinjhMikeS, you mentioned dumping PCM to the piezo
18:34:03franklincould it be done?
18:34:46jhMikeSfranklin: probably gonna need some fancy ISR and a to modulate it in a weird way if all it does is beep otherwise
18:35:12franklinAnd it'll never get in HEAD, right?
18:35:25jhMikeSgevaerts: it's possible to revert the S to an older flash and not have the single boot issue
18:35:43franklinso, don't get a Gigabeat S, right?
18:35:57franklinIf I were to buy a new DAP now, what should I buy then?
18:35:58jhMikeSI have one
18:36:02gevaertsjhMikeS: a lot of things are possible. Do you enjoy walking people through all the steps? :)
18:36:30gevaertsAlso, all you gain is being able to playe APE -c5000 :)
18:36:32jhMikeSgevaerts: no, but there is a one-clicky solution that's just like upgrading
18:36:49jhMikeSfranklin: if you like vgm or ape Gigabeat S is your friend
18:36:52franklingevaerts, does it run DOOM? :)
18:37:06jhMikeSS does run doom indeed, it's almost a joke for it
18:37:24gevaertsYou don't need anywhere near that CPU power for doom
18:37:28jhMikeSwhat challenges it? video, vgm and ape
18:37:39jhMikeSvideo barely
18:37:59*franklin is considering building a cluster computer with Gigabeat S' now
18:38:39jhMikeSgevaerts: we just need a gloomier, doomier doom
18:39:05jhMikeSmaybe port quake or something :)
18:39:15gevaertsGo ahead :)
18:39:25franklinWow sansa clip is surprising powerful
18:39:33franklinBut it has the worst screen ever
18:39:55jhMikeSfranklin: don't look at it, just listen!
18:40:10franklinI mean, seriously? Who decided that it would be a good idea to make a screen that was half yellow, half blue?
18:40:13jhMikeSgevaerts: nah...too easy probably
18:40:18franklinWith a split in the middle!?
18:40:33franklinjhMikeS, while you're at it, port a C compiler, too :)
18:40:39jhMikeSfranklin: must be the guy who designed space invaders
18:41:00jhMikeSfranklin: wtf would you need that for?
18:41:04franklinhaha game title, etc goes in top, game goes in bottom
18:41:16franklinjhMikeS, to build itself
18:41:29franklinand to say that I have a self-hosting platform on my ipod
18:41:42jhMikeSport an emulator to run it on first
18:42:47jhMikeSok, 'nuff silliness; this is serious
18:43:00franklinOk... seriously, why isn't 2048 a link on
18:43:19franklinAll the other plugins are
18:43:55 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
18:44:58jhMikeSwhere the actual link to it/
18:46:18 Quit TheGuyWhoIsTheG2 (Quit: Page closed)
18:46:57franklinjhMikeS, the plugin index is generated automatically
18:50:03jhMikeSI gather that
18:50:14franklinjust sayin' :)
18:50:51jhMikeSit's the only one with digits
18:51:21franklinchip8 has digits, too
18:51:59jhMikeSoops, skimmed past that
18:54:05franklinany idea why?
18:57:20jhMikeSnot yet
18:57:44jhMikeSI'm looking for some key difference in the chip8 vs. 2048 and am failing to see it
18:57:59jhMikeSyou have %TOC% while it doesn't but would that matter?
19:01:44jhMikeSchip8 does have a "chip8" though at the top, not _only_ numbers
19:02:16franklinyeah, maybe that's it... try Plugin2048 as the title
19:03:13jhMikeSme or you?
19:03:18franklinyou, I need to go now
19:03:22franklincatch you soon
19:03:23jhMikeSoh really
19:04:07 Quit franklin (Remote host closed the connection)
19:05:01 Quit rela (Read error: Connection reset by peer)
19:06:01 Quit kuldeepdhaka_ (Ping timeout: 245 seconds)
19:07:33saratogaeven the iPod video should be fast enough unless you're doing something stupid like turning on a million EQ bands
19:11:29coppersaratoga: like 10 of them?
19:11:48saratogadid we enable 10 bands on the PP players?
19:12:17copperon the iPod Video, they're there
19:12:56saratogaprobably not the best idea
19:13:10copperhence the guy's inquiry into faster devices
19:13:17saratogai always meant to look into moving them to the coprocessor as well rather than the main CPU, but it seemed pointless with only 5 bands
19:13:37copperit's unclear whether he's using it right though
19:13:50copperpossibly using it as a non-parametric EQ
19:15:16jhMikeSoh no, multicore DSP now? :)
19:15:52copperiPods have a co-processor?
19:16:23jhMikeSall portal player do
19:17:20 Join Rower [0] (
19:19:15 Join kuldeepdhaka_ [0] (~kuldeepdh@unaffiliated/kuldeepdhaka)
19:26:25 Join lleeloo [0] (~lleeloo@
19:26:41 Quit lleeloo (Client Quit)
19:29:18 Join lleeloo [0] (~lleeloo@
19:32:36 Part ZincAlloy
19:46:22 Join bluebrother^ [0] (~dom@rockbox/developer/bluebrother)
19:46:33 Quit bluebrother (Disconnected by services)
19:48:39 Quit fs-bluebot (Ping timeout: 260 seconds)
19:50:41 Join krabador_ [0] (~krabador_@
19:52:32 Join fs-bluebot [0] (
19:53:08 Nick krabador_ is now known as krabador (~krabador_@
19:53:28 Quit krabador (Changing host)
19:53:28 Join krabador [0] (~krabador_@unaffiliated/krabador)
20:05:45 Join rela [0] (~x@pdpc/supporter/active/rela)
20:12:44***Saving seen data "./dancer.seen"
20:21:29 Join bertrik [0] (~quassel@rockbox/developer/bertrik)
20:29:12 Join lebellium [0] (
20:33:23 Quit michaelni (Read error: Connection timed out)
20:40:37 Join michaelni [0] (
20:41:43 Quit michaelni (Max SendQ exceeded)
20:44:07 Join michaelni [0] (
20:47:16 Quit michaelni (Client Quit)
20:47:48 Join franklin [0] (980c1337@gateway/web/freenode/ip.
20:50:32franklinsaratoga: have you looked at G#919?
20:50:34fs-bluebot3Gerrit review #919 at : 3Fixed a CHIP8 bug, was part of G#904 by Franklin Wei
20:54:42 Join michaelni [0] (
21:02:06 Quit rela (Read error: Connection reset by peer)
21:07:26 Quit jhMikeS (Ping timeout: 246 seconds)
21:23:56 Join ploco [0] (dce9b7f9@gateway/web/freenode/ip.
21:28:38 Join robertd [0] (ba593ff9@gateway/web/freenode/ip.
21:29:08 Quit robertd (Client Quit)
21:29:17plocojhMikeS: Good news! your thread patch works! and no longer crash in scrolling text. (but doesn't fix ART problem, those are global reference mismatch)
21:29:31 Join robertd [0] (ba593ff9@gateway/web/freenode/ip.
21:31:18franklinploco: thread patch for every target?
21:34:43plocothat affect the host targets the most
21:36:42*franklin needs to go
21:36:47 Quit franklin (Quit: Page closed)
21:43:48plocohmm..take some of my words back. Kitkat ART still panic at power thread, so not only the global reference, must use SIGALTSTACK_THREADS. this sucks.....
21:48:40 Quit kuldeepdhaka_ (Ping timeout: 260 seconds)
21:50:16 Join n1s [0] (~n1s@rockbox/developer/n1s)
21:55:15 Quit ploco (Quit: Page closed)
22:07:22 Join kuldeepdhaka [0] (~kuldeepdh@unaffiliated/kuldeepdhaka)
22:12:48***Saving seen data "./dancer.seen"
22:16:07kugelploco (logs): SIGALTSTACK threads are really lightweight too, it's not a problem
22:16:37kugelactually i think android is the only HOSTED target that doesn't use them
22:18:12 Quit bertrik (Quit: reboot)
22:21:22 Join bertrik [0] (~quassel@rockbox/developer/bertrik)
22:23:33 Join franklin [0] (980c3a61@gateway/web/freenode/ip.
22:47:19 Quit n1s (Quit: Ex-Chat)
23:02:19 Quit lleeloo (Ping timeout: 244 seconds)
23:07:40 Quit amayer (Quit: Leaving)
23:07:45 Quit kugel (Ping timeout: 264 seconds)
23:20:52franklinTheSeven: why doesn't USB HID work on ipod classic?
23:48:19TheSevenfranklin: in git master or with my patches?
23:49:26 Quit lebellium (Quit: ChatZilla [Firefox 32.0/20140818191513])
23:49:33franklingit master
23:49:46TheSevener, it's a mystery if USB works there at all ;)
23:49:49TheSeventhat driver is utterly broken
23:50:31franklindo your patches make HID work?
23:51:51franklinwhat patches? gerrit ones?
23:52:09TheSevenand no, my patches don't make hid work
23:52:18TheSeventhey make everything except for hid work at least ;)
23:52:23TheSeven(and some other minor nitpicks)
23:52:36franklinhow can I make HID work? :)
23:52:46TheSevenby figuring out why it doesn't work in the first place
23:53:09TheSevenfrom what I can tell the HID data packets arrive at the PC, but it ignores them for some reason
23:53:19TheSevenmaybe compare the data packets with a device where it works
23:53:29franklinhow do I intercept USB packets?
23:53:44TheSevenwith e.g. wireshark on linux
23:53:53franklinwireshark works with usb?
23:54:02TheSevenyes, but only on linux
23:54:10TheSevenand if you're messing with USB, be sure to apply g#844
23:54:12fs-bluebot3Gerrit review #844 at : 3usb-designware: New USB driver for Synopsys DesignWare USB OTG core. by Michael Sparmann
23:54:13franklinwell, that's what I use :)
23:54:18TheSeven(and, on ipod nano 2g, g#843 as well)
23:54:19fs-bluebot3Gerrit review #843 at : 3Fix cache coherency on ARM940T (and other ARMv4T cores). by Michael Sparmann
23:54:28franklin(linux, not that patch yet)
23:56:30franklinTheSeven: so what interface?
23:58:43franklinnever mind

Previous day | Next day