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

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

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

#rockbox log for 2009-02-17

00:00:14 Quit n1s ()
00:00:46gevaertspyro_maniac: I know, it should (within BUFFER_SIZE), but I've never had that work properly...
00:03:34pyro_maniacgevaerts: so this serial connection is untested?
00:04:13gevaertsNot untested as such, just not used very much in this sort of circumstances
00:05:28pyro_maniacbut there is a chance to get this work in this way, right?
00:05:32LambdaCalculus37bluebrother: All right... patch applied... here goes!
00:05:37*LambdaCalculus37 makes
00:06:36mcuelenaerecan any DAP-with-touchscreen-user (Cowon D2?) test FS #9918 and give feedback please?
00:08:43pyro_maniacgevaerts: have you any ideas? I don't have a device where it works, so I can't test the "right" way
00:08:50 Quit amiconn (Nick collision from services.)
00:08:51 Join amiconn_ [50] (n=jens@rockbox/developer/amiconn)
00:08:51 Join pixelma_ [50] (n=pixelma@rockbox/staff/pixelma)
00:08:51 Quit pixelma (Nick collision from services.)
00:09:04mcuelenaeregevaerts: shouldn't USB driver handling be something like this?
00:09:05 Nick pixelma_ is now known as pixelma (n=pixelma@rockbox/staff/pixelma)
00:09:11 Nick amiconn_ is now known as amiconn (n=jens@rockbox/developer/amiconn)
00:11:05gevaertsmcuelenaere: that's an option. It's not really necessary though
00:11:30 Join sarixe [0] (
00:11:46mcuelenaereso you think USB serial should currently still need changes to usb.c and usb_core.h?
00:11:58mcuelenaereor a not-at-compiler-time implementation?
00:12:10gevaertspyro_maniac: maybe try looking for the code that dumps logf to disk
00:12:28gevaertsmcuelenaere: what do you mean?
00:12:56mcuelenaereI mean a way to let the user decide what USB driver gets loaded at runtime
00:13:02 Quit casainho ("ChatZilla 0.9.84 [Firefox 3.0.6/2009020911]")
00:13:22mcuelenaereperhaps something like plugins and codecs get loaded
00:13:31gevaertsThat will need a lot more changes
00:14:02mcuelenaeretrue, and currently it isn't really needed as there are only 2 drivers
00:14:39*mcuelenaere thinks this could get added on the Mr. Someone's TODO list
00:14:44gevaertsOne basically. Serial is not at all needed for end users, so it requiring compile-time options is not an issue
00:15:17pyro_maniacgevaerts: can tell me, where this happens?
00:15:23gevaertspyro_maniac: no
00:15:38mcuelenaereyes, but if (at some time) an MTP driver gets added, choosing which driver gets loaded would be nice
00:15:43gevaertsI'd have to search the same way as you...
00:16:39gevaertsSure, but that's not for this week
00:16:53Unhelpfulit would mean moving the USB driver out of the core, and into an API. i think that at *most* it would be sensible to allow a plugin to hook and replace the core USB handler, and just have one USB handler in core.
00:17:03pyro_maniacgevaerts: ok I will search for that tomorrow. I have to go now. thanks for help again
00:17:47Unhelpfuland actually, that wouldn't really be useful for *some* cases (i'm thinking gdb over usb serial ;)
00:17:49gevaertsUnhelpful: sure. usb_core.c will always be in the core.
00:18:05*LambdaCalculus37 is starting to fix some English here and there in the manual
00:18:20 Quit pyro_maniac ("Leaving.")
00:18:25BigBambiLambdaCalculus37: Mind your spelling :)
00:18:31mcuelenaereUnhelpful: that looks like another way to do this, but it wouldn't allow running plugins and some USB drivers at the same time
00:18:33LambdaCalculus37BigBambi: I will. :)
00:19:04 Quit flydutch ("/* empty */")
00:19:06BigBambiLambdaCalculus37: I want to see lots of "u"s and no "z"s :)
00:19:15Unhelpfulmcuelenaere: no, it certainly wouldn't. the only alternate USB usage i can think of where i would care about that is debugging, though, and we don't have that yet :/
00:19:38mcuelenaeredoesn't GDB over USB require an USB driver which doesn't use interrupts?
00:20:05gevaertsI'm not sure how well gdb would work with usb serial. The current usb implementation needs interrupts, threads, ...
00:20:11LambdaCalculus37BigBambi: Yes sir. :)
00:20:20BigBambihehe :)
00:20:39*bluebrother would consider an Ipod acting as soundcard fun
00:20:45Unhelpfulheh, that would be a lot of "fixing" ;)
00:21:14mcuelenaerebluebrother: yes, and controlling the FM chip from a PC would mean you could use it as a FM receiver :)
00:21:30mcuelenaereI mean that + routing audio to the PC
00:24:08 Quit undertakingyou (Remote closed the connection)
00:25:45 Join jfc^2 [0] (
00:26:40 Join undertakingyou [0] (
00:27:01 Quit jfc^3 (Read error: 60 (Operation timed out))
00:28:08 Quit bluebrother ("leaving")
00:28:13moosUnhelpful: it seems that you forgot the attachments for fs#9916
00:28:25 Part undertakingyou it does! :/
00:29:31***Saving seen data "./dancer.seen"
00:29:52 Quit Hillshum ("Leaving")
00:29:57Unhelpfulfixed, enjoy some appalling use of empty for loops, among other things.
00:30:44 Join cool_walking_ [0] (i=cb3b81c3@gateway/web/ajax/
00:32:42 Quit parafin (Read error: 60 (Operation timed out))
00:33:55LambdaCalculus37bluebrother: (for the logs) Another brick wall:
00:36:44 Join parafin [0] (
00:37:55 Nick Beaver`univ is now known as Beaver`alszik (
00:40:18 Quit cool_walking_ (" ajax IRC Client")
00:40:53*LambdaCalculus37 will start doing some more manual work later
00:41:09 Quit {-phoenix-} ("Konversation terminated!")
00:41:14LambdaCalculus37I'm going to go take a break. See everyone later.
00:41:40 Quit LambdaCalculus37 ("Fwump")
00:43:20 Quit mcuelenaere ()
00:44:59 Nick JdGordon|zzz is now known as JdGordon (n=jonno@rockbox/developer/JdGordon)
00:48:58 Quit MethoS- (Remote closed the connection)
00:56:59 Quit Thundercloud (Remote closed the connection)
00:58:36 Join cool_walking_ [0] (i=cb3b81c3@gateway/web/ajax/
00:58:53 Quit perrikwp (" ajax IRC Client")
00:59:03JdGordonPaulJam: hey, can you send me the wps which partially works?
01:02:20 Join evilnick [0] (i=0c140464@gateway/web/ajax/
01:03:01 Join mellery_ [0] (
01:04:45JdGordonPaulJam: ta, I tried finding the cause of these lockups yesterday and unfortunatly adding splash()'s got rid of the problem :p
01:05:33UnhelpfulJdGordon: logf, maybe?
01:07:02JdGordonyeah, maybe
01:11:00 Quit gevaerts (Nick collision from services.)
01:11:10jhMikeSWhy does the front page still state that rockbox doesn't "run" on the gigabeat S series? I'd sure say it runs on it even if not "supported".
01:11:12 Join gevaerts [0] (n=fg@rockbox/developer/gevaerts)
01:11:26 Quit SoapWork (Remote closed the connection)
01:13:03JdGordonas proof that its out of date and that we really have a working 4g nano port !
01:13:16mooshi jhMikeS still those installation issues (license problem if you recall...)
01:13:49Unhelpful#9919 is the initial support of buffer_alloc in pictureflow
01:13:59jhMikeSdoesn't mean it doesn't run or is it to discourage traffic?
01:14:01 Join midijunkie [0] (
01:14:40toffe82moos: we know have an official source for the nk.bin
01:15:08jhMikeSJdGordon: And I don't know enough about nano support to know if you're being sarcastic or not ;)
01:15:32moosone time Bagder made it, but it was reverted, linuxstb wanted so until we work out installation procedure support and the license issue...
01:15:37 Quit mellery_ ("Ex-Chat")
01:16:40moosjhMikeS: to discourage traffic of people floading with OF recovery weirdness
01:17:15JdGordonjhMikeS: I was joking :) we had a guiy in here ysetrday asking why rbutil couldnt find his ipod "color" which turned out to be a g4 nano
01:17:49jhMikeSJdGordon: I see. Well, if it has a color screen then it must be a "color" iPod?
01:19:27 Quit obo ("Bye")
01:23:15 Join perrikwp [0] (i=18ac0c41@gateway/web/ajax/
01:23:29scorchethen again, there is the video nano...
01:23:36*scorche stabs
01:24:57JdGordonamiconn: you're not still up are you?
01:26:19 Nick jfc^2 is now known as jfc (
01:31:39 Quit moos ("Time to say moo !")
01:33:11 Quit jhMikeS (Nick collision from services.)
01:33:17 Join jhMikeS [50] (n=jethead7@rockbox/developer/jhMikeS)
01:41:16JdGordonUnhelpful: if that global_context is a global var anyway, and presumably only going to be one per plugin... why the need to pass it into evey call?
01:43:12 Quit jhMikeS (Nick collision from services.)
01:43:18 Join jhMikeS [50] (n=jethead7@rockbox/developer/jhMikeS)
01:48:37 Quit Taylor ("Leaving")
01:51:39 Quit gregzx ("ChatZilla 0.9.84 [Firefox 3.0.6/2009011913]")
01:59:46 Part piotrek2234
02:01:20 Quit dandin1 ()
02:03:52 Quit dfkt ("-= SysReset 2.53=- Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn.")
02:05:43 Quit perrikwp (" ajax IRC Client")
02:07:53 Quit JdGordon (Remote closed the connection)
02:12:19 Part toffe82
02:12:26 Join JdGordon [0] (
02:14:00 Quit evilnick (" ajax IRC Client")
02:18:24 Quit PaulJam (".")
02:19:28 Quit midijunkie ("?(???~•~)?")
02:24:12 Quit JdGordon (Read error: 104 (Connection reset by peer))
02:28:47 Nick fxb is now known as fxb__ (
02:29:36***Saving seen data "./dancer.seen"
02:30:22 Part gartral1
02:35:23 Join JdGordon [0] (
02:44:26 Quit goffa (Read error: 60 (Operation timed out))
02:45:08 Join goffa [0] (n=goffa@
02:50:53 Quit sarixe ("Leaving")
03:01:29UnhelpfulJdGordon: the idea was supporting more than one allocator. i'm not sure it's a good idea, though, still. it wouldn't be hard to rework without it.
03:02:44Unhelpfuli don't think running multiple allocators is of practical value... one allocator with multiple memory segments might be sensible, though (say, the remaining plugin buffer + stolen audio buffer)
03:06:29Unhelpfuli should figure out what causes the data abort on my e200, first, though. it flashed slides very briefly first, at least :)
03:07:14 Join blkhawk- [0] (
03:08:30 Join noobtaculus [0] (n=40e63f20@gateway/web/cgi-irc/
03:11:26 Quit noobtaculus (Client Quit)
03:11:29 Join noobtaculus [0] (n=40e63f20@gateway/web/cgi-irc/
03:11:42 Quit noobtaculus (Client Quit)
03:14:30 Join BHSPitMonkey [0] (n=stephen@unaffiliated/bhspitmonkey)
03:15:33Unhelpfulmaybe a macro that defines an appropriate global_context, possibly an array of contexts if we want to support multiple segments later?
03:18:34Unhelpfulwhat, exactly, does data abort on ARM mean? i know unaligned access can be one cause...
03:19:13 Join rocko [0] (
03:19:58 Join kugel [0] (n=kugel@rockbox/developer/kugel)
03:20:54 Quit blkhawk (Read error: 110 (Connection timed out))
03:21:12 Nick blkhawk- is now known as blkhawk (
03:23:22kugelany target tree expert here?
03:23:36kugelit seems to not include the correct debug-target.h here
03:24:40 Quit pixelma (Nick collision from services.)
03:24:41 Join pixelma_ [50] (n=pixelma@rockbox/staff/pixelma)
03:24:57 Nick pixelma_ is now known as pixelma (n=pixelma@rockbox/staff/pixelma)
03:25:37 Join Horschti [0] (n=Horscht@xbmc/user/horscht)
03:42:02jhMikeSkugel: is there a debug-target.h at a lower level? I think it picks the deepest one but I'm not 100% sure on that.
03:43:07 Quit Horscht (Read error: 110 (Connection timed out))
03:43:15kugeljhMikeS: I was told that it picks highler level ones first after r19866
03:44:11kugellower* I mean
03:44:20kugeland make (deps) seems to pick the correct include (the correct one showed grepping make.dep), but it still included the higher one
03:44:28jhMikeSyeah, or else gigabeat S would choose PP files
03:45:11kugelwell, it didn't work
03:46:13kugelmaybe it's using lower ones, but only if there's no such file in the same dir as the .c?
03:48:18 Quit Rob2223 (Read error: 104 (Connection reset by peer))
03:48:27 Join Rob2222 [0] (
03:48:43jhMikeSmaybe move the .c file then? All the targets I've got seem to have the .c in the same dir as the xxx-target.h files.
03:48:54 Quit pixelma (Nick collision from services.)
03:48:55 Join pixelma_ [50] (n=pixelma@rockbox/staff/pixelma)
03:49:04kugelthe .c is mostly common over the ams sansas
03:49:11 Nick pixelma_ is now known as pixelma (n=pixelma@rockbox/staff/pixelma)
03:49:28kugelit's not #ifdef enough (yet) to fork it imo
03:52:05UnhelpfulHM. does core bufalloc catch attempts to free invalid handles?
03:53:59jhMikeScore buffalloc has no free
03:54:32jhMikeSnvm, I was thinking buffer_alloc :p
03:56:23jhMikeSit does look like it checks it's list of valid handles using find_handle
03:56:29Unhelpfulit's "bufclose" actually, seems to work for freeing non-file buffers as well.
03:56:56Unhelpfulit looks like pictureflow depended on that behavior :)
03:59:34 Join sarixe [0] (
03:59:46Unhelpfulshould i fix it in pictureflow, or in buffer_alloc? seems to me that trying to free some value that is not an alloc result is the behavior that is wrong, and should be fixed
04:03:42kugelUnhelpful: I'd wait how turns out
04:04:03UnhelpfuljhMikeS: what would mpegplayer need to use buffer_alloc? besides it not being horribly experimental code...
04:04:30jhMikeSUnhelpful: Perhaps. Personally I'm the type that believes in going the more robust route.
04:04:36kugelunless it's a bug of course
04:05:14Unhelpfulkugel: the trouble with that is that the malloc API can not support compaction without virtual memory, since the caller is just handed a pointer to the memory.
04:05:45jhMikeSUnhelpful: libmpeg2 dynamically allocates its frambuffers and preview mode does as well. It would also allow allocating overlays and alternate framebuffers for the IPU.
04:05:51UnhelpfuljhMikeS: well, what exactly would be needed for optional multicore safety? there are some extra alignment requirements, i know, but i'm sure that's not it....
04:06:58jhMikeS1) Shared data structures in noncacheable memory 2) A synchronization scheme that doesn't just involve globally locking the allocator would be best (maximize concurrency).
04:07:21kugelUnhelpful: alone the description sounds like it would have very little memory waste:
04:07:53kugelI haven't looked at it deeply, but maybe it does some more than the standard malloc to keep waste low
04:08:10kugel"- and it doesn't fragment the memory as much. "
04:09:57Unhelpfulthat's nice to have, also, but i think that malloc API is more useful for "things we don't want to port away from malloc" than as a general solution to things that need dynamic allocation. you can do your best to *manage* fragmentation, and certainly allocating best-fit instead of first-fit is a way to do that, but i'm talking about being able to collect fragmented pieces into one contiguous segment again.
04:10:23 Quit rocko ("Leaving")
04:10:58 Join rocko [0] (
04:11:14kugelanyway, I'm off. night
04:11:17 Join Barahir [0] (
04:11:18 Quit kugel ("ChatZilla 0.9.84-rdmsoft [XULRunner]")
04:13:43 Quit AndyIL (Read error: 60 (Operation timed out))
04:16:08 Join AndyI [0] (i=AndyI@
04:17:26 Join ipod5580gb [0] (
04:18:22ipod5580gbCan anyone help me with getting TuneFM to work with my ipod. It does not power up when connected when running RockBox 3.1
04:22:21 Join Llorean [0] (
04:26:36 Quit Barahir_ (Read error: 110 (Connection timed out))
04:29:19 Quit Beaver`alszik (Read error: 54 (Connection reset by peer))
04:29:24 Join Bavlito [0] (
04:29:40***Saving seen data "./dancer.seen"
04:35:53UnhelpfuljhMikeS: actually, you know, that might be a decent use for the multiple-allocators support, have each core "own" a chunk of memory and a buffer_context
04:36:41Unhelpfulalthough, that's no good for allocating things both cores might share
04:36:47jhMikeSUnhelpful: It's basically like that now though the first allocator allocates the buffer for the second one.
04:37:15ipod5580gbCan anyone give me a bit of help with a FM Transmitter?
04:37:41 Quit miepchen^schlaf (Read error: 110 (Connection timed out))
04:37:45Unhelpfulhrm... buffer_alloc would be *quite* unsafe to use in that fashion, *unless* that allocation was the first one made.
04:39:32Unhelpfulit would be wasteful, anyway, an allocation has a 12B overhead, you could just divide the buffer in half, and pass each half to a separate buf_init_ctx call with its own struct buffer_context
04:39:36 Quit Bavlito (Read error: 104 (Connection reset by peer))
04:42:35 Join Beaver`alszik [0] (
04:44:13 Quit HBK ()
04:44:35 Join HBK [0] (
04:44:35jhMikeSas it is now, the libmpeg2 buffer is the first buffer allocated
04:47:10 Join saratoga [0] (n=9803c6dd@gateway/web/cgi-irc/
04:47:39jhMikeSwhatever the case, things are limited because it wants whatever is left from all that for the disk buffer
04:49:21jhMikeSand it uses whatever remains in the libmpeg2 buffer to make thumbnails and fails that if enough isn't there
04:49:26Unhelpfulhrm... i *could* add some calls to allow moving or resizing the allocation buffer. moving it would require an extra pointer stored in context, since the buffer doesn't actually know where buffer start is, now.
04:54:52 Quit sarixe ("Leaving")
04:55:58 Join blkhawk- [0] (
04:56:30Unhelpfulhow does this sound, for a possible concurrency scheme... an allocator per-core, with handles valid on that core, but the buffers each allocator uses are stored back-to-back in the available space. *if* an allocation fails, even after compacting the existing allocations in that buffer together, we take a global lock, and move the join between the two buffers such that the allocation can succeed, and leave roughly the same amount of free
04:56:30Unhelpful space in each buffer.
04:58:01Unhelpfulallocations and fetches from both buffers would have to be blocked for the duration, but it should not be a very frequent event
04:59:31 Join sarixe [0] (
05:00:22Unhelpfulthe big issue i see is that this could still lead to a thread on one core checking ot the pointer for its handle, and having the pointer invalidated without calling yield :/
05:02:57jhMikeSthere's also multithreaded codecs that might need to share things
05:05:33Unhelpfulas in multicore? or just threaded? threads are quite safe, as long as they understand that yielding invalidates the pointers associated with handles
05:06:36 Join perrikwp [0] (i=18ac0c41@gateway/web/ajax/
05:08:20jhMikeSwell, like mpa.c. mpegplayer make no such assumptions, it's too complex to guarantee what code "yields" or not and so it written as if preemptive.
05:08:22ipod5580gbAnyone around that can help me?
05:10:56UnhelpfuljhMikeS: well, i think the way i'd handle that is to have any function "check out" its handles again around calls to other functions... but, yeah, this is not an allocator for all purposes. i don't think we can really have one, without using virtual memory, that can suit *every* use.
05:13:24 Quit blkhawk (Read error: 113 (No route to host))
05:13:56 Nick blkhawk- is now known as blkhawk (
05:17:25saratogaipod5580gb: looking at the logs, I'd guess you need to turn on the accessory power supply, or that your device doesn't work with rockbox
05:18:01jhMikeSUnhelpful: it's the approach that's needed really that stuff. playback.c had problems when one target's version of a function blocked or yielded and other didn't. in many cases there really is no difference in treatment between the two types of multitasking at all except in very simple code that has complete control.
05:19:00Unhelpfuli favored pointer-checkout speed over pretty much everything else, though. that's a table lookup, and inlined ;)
05:19:02 Join gartral [0] (
05:19:24gartralthe buyers guide page in the wiki is outdated
05:19:57gartralparticularly, it says that the sansas can only record via microphone
05:20:00Unhelpfulthe handle table never actually moves, so it would also be quite possible, instead of using handles at all, to pass the address of the table entry back
05:20:40Unhelpfuli'm not sure if gcc could be made to understand that the contents of that entry are volatile, while the pointer to it is not?
05:21:36jhMikeSsure, declare the needed members as volatile
05:22:37 Join tvelocity[away] [0] (
05:23:40Unhelpfulit's a union, not a struct... would that work?
05:24:21jhMikeSI think so. If you look at the volatile member it should be treated as such.
05:26:50Unhelpfulit needn't *actually* be a union, in the handle table... it was mostly a matter of wanting to be able to treat the entire buffer as a collection of the same thing, whether it's pointers to allocated blocks, integers indicating their length/status, etc.
05:28:34 Quit tvelocity[a] (Connection timed out)
05:29:46Unhelpfulbut, if adding, maybe, a volatile void *data to union buffer_data would make it easier to use in more places, now would be the time, while it's only being used experimentally in one place :)
05:30:45jhMikeSvoid * volatile data ? i have my volatile in the wrong place? my C is still a little rusty where qualifiers are concerned. :)
05:32:28jhMikeSthat would point to volatile data, not a volatile pointer to data
05:33:35 Quit thegeek (Read error: 104 (Connection reset by peer))
05:33:46 Join thegeek [0] (
05:34:10jhMikeSI've been through the ringer on that stuff so that volatile thing is pretty down pat :)
05:35:19Unhelpfulah, i get it. would that maybe be better than using the integer handles? everything that refers to the int handles can be replaced with passing the handle address directly, quite easily. the ints are never actually used inside the allocator, they're just converted to/from addresses when handles are passed in from or out to the caller
05:37:08 Join tvelocity[a] [0] (
05:37:29 Quit Horschti ("Verlassend")
05:38:15jhMikeSso long as an old pointer can't alias a new one but then one it's freed it shouldn't be passed back. a little different from threads using slot pointer publicly.
05:39:26Unhelpfulhrm... yeah, slots in the handle table *can* be replaced after they've been freed.
05:40:56jhMikeSthe scheduler encodes a handle with a slot index in the low byte and a version number in the high byte as a guard, if the id in the slot doesn't match, it's considered removed but not treated as an error.
05:42:53 Quit Llorean ("Leaving.")
05:45:56 Join Llorean [0] (
05:48:38jhMikeSit's 0xvvnn. 0x0000 is reserved as invalid and 0x01nn is the first handle for any slot, version incremented when it exits and recycles every 255 threads in a slot
05:49:25 Quit JdGordon (Read error: 104 (Connection reset by peer))
05:49:50 Join dude187 [0] (
05:52:01Unhelpfulhrm. i'm not sure we could do that with allocator handles, they're really *just* pointers to the actual data.
05:52:17 Quit tvelocity[away] (Read error: 110 (Connection timed out))
05:52:45Unhelpfulthey're set to NULL on free, but of course, if we want to guard against the handle being reallocated, that doesn't help
05:53:48 Quit Aurix_Lexico (Remote closed the connection)
05:54:36 Part gartral
05:55:57soapsaratoga, the first HA/RB cross-post?
05:56:04jhMikeSit's implemented that way to guard against a thread_wait call referring to a newly created thread using a pointer to one it expects to be exiting. it's a problem a programmer can't write away properly himself.
05:56:17soapI wasn't as nice on HA as you were on RB
05:56:22 Quit perrikwp (" ajax IRC Client")
05:56:45 Join perrikwp [0] (i=18ac0c41@gateway/web/ajax/
05:57:04Unhelpfulis there a similar problem we expect to see with a memory allocator? i don't think you're crazy, i'm just not sure what we're trying to prevent :/
05:58:23ipod5580gbJust wondering if anyone can help me with getting a FM Transmitter to work.
05:58:53jhMikeSI'm just discussion a solution for something under the most difficult conditions where resources allocated to a thread must no longer be in use before a plugin or codec exits.
06:00:05jhMikeSipod5580gb: I'm not aware of any such thing being officially supported.
06:00:14Unhelpfulhrm. but this isn't a system-wide allocator, it's intended to be used by a plugin or codec on memory that it already owns.
06:01:15 Join JdGordon [0] (n=jonno@rockbox/developer/JdGordon)
06:01:59soapjhMikeS, prelim accessory support is in SVN
06:02:29soapipod5580gb, which FM transmitter, which ipod (?) and what have you tried?
06:03:09ipod5580gbWell it is TUNEFM from Belkin, Ipod5.5 80gb, with rockbox 3.1
06:03:33soaphave you turned on the accessory power supply?
06:03:43ipod5580gbI have
06:03:52ipod5580gbAnd tried the car dock even.
06:03:56ipod5580gbIt doesn't power up
06:04:06jhMikeSsoap: I had guessed that much but ipod5580gb has been poking for quite some time for something to be said. So I gave the best answer I could about the issue.
06:04:39soapsorry for jumping in late and assuming, jhMikeS
06:04:48saratogalooks like it doesn't work then
06:05:15soapipod5580gb, when you say "it" doesn't power up - you mean the FM transmitter or the iPod?
06:05:25ipod5580gbShit, guess I will look at maybe returning it or something
06:05:29ipod5580gbYa, I mean, the tramisster doesn't turn on
06:05:35ipod5580gbno numbers appear or anything
06:05:53*jhMikeS forgives soap ;-)
06:06:02soapThere is a wiki page IpodAccessories and a thread in the Unsupported Builds forum. Within those two resources is the sum of knowledge so far on what works and what doesn't.
06:06:34soapWhen I get off my lazy ass I'll merge the two sources into one, but until that time you can read the two - even contribute to the wiki! - and learn all Rockbox knows.
06:06:53ipod5580gbYa done my best to look around, but guess either I live with rebooting in the apple firmware or return it.
06:07:32soapIt appears to be one of those models which has no display of its own and uploads BMPs to the iPod to show what it is doing.
06:07:52jhMikeSsoap: Not without the initiation ceremonies! One must vow to protect the rb brotherhood's secrets to the death. :p
06:07:53soapIf I had to guess, I'd bet none of the accessories of that type will function properly/fully.
06:07:59ipod5580gbActually it does have a display, let me find a picture
06:08:34soapgreat, another bloody series of products sold with the same name yet different models.
06:09:05soapPM me all the details you can dig on the item, model name, number, every little thing you can think of, and I'll add it to the wiki when I get a chance.
06:09:16ipod5580gbI know, sort of pissing me off as well.
06:09:36soapis what I had been looking at.
06:11:14ipod5580gbIs there a good FM trasmitter you would recommended that work with RockBox?
06:11:28soapread the IpodAccessories wiki page and please also pm me (on the forums is ideal) all the info it asks for as well, including the debug info. Or fill out the wiki yourself if you want.
06:11:39soapI've never used an iPod accessory.
06:12:37 Quit ipod5580gb (Read error: 60 (Operation timed out))
06:12:56 Join ipod55800 [0] (
06:17:37ipod55800Damn, lost connection.
06:17:43ipod55800Is there a good FM trasmitter you would recommended that work with RockBox?
06:21:32 Quit saratoga ("CGI:IRC (EOF)")
06:22:42soapas I said - I've never owned one, though perhaps i should buy one.
06:23:24UnhelpfuljhMikeS: i don't think this allocator can do anything like that, without changing what is stored in the handle table. that's certainly possible, we could use a struct that also contains a version counter, instead of buffer_data. it would double the cost of the handle itself, a 1/3 gain in the overhead per allocation, which is not a big deal if most allocations are large.
06:23:25soapWith buschel and dreamlayer's insane attempts to turn the iPod 5G into a perpetual motion machine I should probably watch for breaking of accessory support. ;)
06:23:37 Join lymeca [0] (
06:25:16Unhelpfulit'd break the idea of handing the actual address of the handle table entry back to the caller, but that's broken anyway if we want some memory manager to *always* be able to check on a handle when something wants to use it.
06:26:29ipod55800Thanks, I appreciate your help. Shitty it doesn't work.
06:26:45 Quit rocko ("Leaving")
06:29:41***Saving seen data "./dancer.seen"
06:33:21 Quit jhMikeS (Nick collision from services.)
06:33:27 Join jhMikeS [50] (n=jethead7@rockbox/developer/jhMikeS)
06:54:49*jhMikeS wouldn't mind more thorough commenting in that patch so he'd know what's supposed to be going on :)
06:58:03 Quit pixelma ("No Ping reply in 30 seconds.")
07:01:14 Join pixelma [50] (n=pixelma@rockbox/staff/pixelma)
07:02:32 Quit ipod55800 ()
07:08:02 Quit HBK (Read error: 104 (Connection reset by peer))
07:08:27 Join HBK [0] (
07:09:45 Quit CaptainKwel (Read error: 110 (Connection timed out))
07:11:06jhMikeSUnhelpful: You treat it like an array of union buffer_data and allocations grow from the left while the handle array grows from the right?
07:13:08 Quit perrikwp (" ajax IRC Client")
07:25:25 Quit lymeca (Connection timed out)
07:26:10 Join n1s [0] (n=nils@rockbox/developer/n1s)
07:58:16 Quit cool_walking_ (" ajax IRC Client")
07:58:49 Join perrikwp [0] (i=18ac0c41@gateway/web/ajax/
08:24:30 Quit BHSPitMonkey ("Ex-Chat")
08:25:28 Join Rob2223 [0] (
08:29:42***Saving seen data "./dancer.seen"
08:33:30 Quit nuonguy (Connection reset by peer)
08:40:34*JdGordon is starting to get annoyed with the simulated backlight
08:40:39 Join ender` [0] (
08:42:15LloreanSo set the backlight to always on. :-P
08:42:35JdGordonthats generally annoying
08:43:08 Quit Rob2222 (Read error: 110 (Connection timed out))
08:44:30*JdGordon wonders why the hell he has a completyl blank WPS screen with the real default wps
08:45:35JdGordonthats on the e200... on the h300 it just locks up :(
09:00:08 Quit z35 ("Leaving")
09:04:58 Join petur [50] (n=petur@rockbox/developer/petur)
09:11:42 Quit perrikwp (" ajax IRC Client")
09:14:10amiconnUnhelpful: I don't think that you can make an allocator that moves blocks on compaction multi-core safe
09:15:36amiconnEven if you guard the handles so that one core doesn't change a handle while the other accesses it, there's still the possibility that one core starts a compaction while the other is accessing a block of data
09:16:26amiconnSo this mechanism can only work for a single core with cooperative threading. The swcodec bufalloc has the same limitation
09:27:14 Join rocko [0] (
09:30:55JdGordonwell this is wierd.. the remote lcd seems to be working but the main lcd is borked
09:36:12 Join bmbl [0] (n=Miranda@unaffiliated/bmbl)
09:38:08 Quit GodEater_ (" ajax IRC Client")
09:45:23 Join B4gder [241] (n=daniel@rockbox/developer/bagder)
09:46:05 Join LinusN [0] (n=linus@rockbox/developer/LinusN)
09:48:55 Join GodEater_ [0] (i=c2cbc962@rockbox/staff/GodEater)
09:49:05 Join Thundercloud [0] (
09:55:18 Quit daurnimator (Read error: 60 (Operation timed out))
09:55:31 Join daurnimator [0] (
09:58:37 Quit GodEater_ (" ajax IRC Client")
09:58:53 Join GodEater_ [0] (i=c2cbc962@rockbox/staff/GodEater)
10:11:50 Join pyro_maniac [0] (
10:14:00 Quit bmbl (Read error: 104 (Connection reset by peer))
10:15:48 Join piotrek2234 [0] (
10:18:18 Quit kachna|lappy (Read error: 113 (No route to host))
10:22:07*JdGordon curses at his code
10:23:31 Quit piotrek2234 (Read error: 104 (Connection reset by peer))
10:29:44***Saving seen data "./dancer.seen"
10:32:07 Quit Thundercloud (Remote closed the connection)
10:35:08 Join midijunkie [0] (
10:37:15 Quit linuxstb (Read error: 113 (No route to host))
10:48:02 Join _lifeless [0] (n=lifeless@
10:48:37 Join miepchen^schlaf [0] (
10:58:04 Nick Beaver`alszik is now known as Beaver`univ (
11:04:10 Quit midijunkie (Read error: 104 (Connection reset by peer))
11:05:33 Quit __lifeless (Read error: 110 (Connection timed out))
11:05:58 Join __lifeless [0] (n=lifeless@
11:06:16 Quit _lifeless (Read error: 110 (Connection timed out))
11:07:43 Join akur [0] (
11:07:45 Part akur
11:13:54 Part rob_
11:29:34 Join perrikwp [0] (i=18ac0c41@gateway/web/ajax/
11:30:01 Join bmbl [0] (n=Miranda@unaffiliated/bmbl)
11:31:25 Join JdGordon_ [0] (
11:32:50 Quit daurnimator (Read error: 60 (Operation timed out))
11:33:03 Join daurnimator [0] (
11:41:26 Join robin0800 [0] (
11:44:26 Join kachna|lappy [0] (
11:56:25 Quit JdGordon_ (Read error: 110 (Connection timed out))
12:11:22 Join kugel [0] (n=kugel@rockbox/developer/kugel)
12:29:47***Saving seen data "./dancer.seen"
12:42:26 Quit kugel (Read error: 104 (Connection reset by peer))
12:43:02 Quit rocko ("Leaving")
12:49:53 Join linuxstb [0] (n=linuxstb@rockbox/developer/linuxstb)
12:52:53 Quit idshark (Remote closed the connection)
13:05:24 Quit freqmod_qu (Remote closed the connection)
13:17:45 Quit miepchen^schlaf ()
13:23:55 Join bandan [0] (
13:45:20 Quit jhMikeS (Nick collision from services.)
13:45:26 Join jhMikeS [50] (n=jethead7@rockbox/developer/jhMikeS)
13:52:29 Part bandan
13:58:28 Join zimba42 [0] (
14:02:18zimba42I happily booted rockbox on my e200v2-4GB with working buttons. Does someone need any kind of testing on that target?
14:03:31 Quit kachna|lappy (Read error: 110 (Connection timed out))
14:03:40zimba42Is the hardware on that target slower than the e200v1 one? (the demos seem to run slower)
14:06:53 Join dfkt [0] (i=dfkt@unaffiliated/dfkt)
14:07:30Bagderit is faster, although possibly not taken advantage of yet
14:08:36zimba42Good :-) ..anyway, great to have the v2 supported at all.
14:09:15 Nick fxb__ is now known as fxb (
14:12:43 Quit parafin (Remote closed the connection)
14:13:15 Join parafin [0] (
14:17:48 Quit __lifeless (Remote closed the connection)
14:18:05 Join __lifeless [0] (n=lifeless@
14:18:35 Join tyfoo [0] (
14:26:36rvvs89Woah v2 support, when did this happen?
14:26:44 Join CaptainKewl [0] (
14:26:46rvvs89Never mind I'll check the web site
14:27:14BigBambiIt didn't yet
14:27:29 Join miepchen^schlaf [0] (
14:27:32BigBambiThey are being worked on, and it is going well, but they are not yet supported
14:29:07 Join Rob2222 [0] (
14:29:50***Saving seen data "./dancer.seen"
14:30:03 Join LambdaCalculus37 [0] (i=44a04303@rockbox/staff/LambdaCalculus37)
14:34:33zimba42it's not officially supported.. but works
14:34:49 Join grdxyxy [0] (n=eric@
14:35:26 Quit grdxyxy (Remote closed the connection)
14:35:48 Join grdxyxy [0] (n=eric@
14:37:19 Quit gromit` (Read error: 60 (Operation timed out))
14:44:58 Join gromit` [0] (
14:47:23 Quit Rob2223 (Read error: 110 (Connection timed out))
14:56:11 Part LinusN
15:00:34 Join midijunkie [0] (
15:05:54 Part grdxyxy
15:10:55 Quit jordoex (Read error: 110 (Connection timed out))
15:12:52 Nick JdGordon is now known as JdGordon|zzz (n=jonno@rockbox/developer/JdGordon)
15:14:07 Quit avis (Remote closed the connection)
15:22:44 Quit n1s ()
15:23:53 Join piotrek2234 [0] (
15:25:48Unhelpfulamiconn: i know it will still not be multicore-safe, but what about the idea of passing the handle's address back, rather than its index, and adding a member to the handle that's marked volatile, for use in accessing the data... would gcc do a decent job of knowing when that pointer-to-a-pointer needs to be reloaded from memory, or should i just stick with the int handles and leave the caller to make sure that the pointer returned is us
15:25:48Unhelpfuled safely?
15:28:31UnhelpfuljhMikeS: yes, allocations grow up from the bottom of the buffer, handles down from the top. the handle deallocation routine only worries about shrinking the handle table by one, if it's called for the last handle, which is sufficient for making sure that the allocator doesn't extend the handle table when an allocation fails. i should probably remove even that check, since it can extend it by one handle at the most.
15:29:10 Part piotrek2234
15:29:18Unhelpfuli'll work on comments now, actually. if i say what/why it does what it does, that only makes it easier for others to say what i'm doing wrong ;)
15:41:38 Join jaykay [0] (
15:50:40 Quit CaptainKewl (Read error: 110 (Connection timed out))
15:54:56 Quit bmbl (Read error: 110 (Connection timed out))
15:57:31 Join z35 [0] (
16:03:41 Join rocko [0] (
16:08:30 Join PaulJam [0] (
16:13:06 Join MethoS- [0] (
16:14:49 Quit rocko ("Leaving")
16:19:14 Quit zimba42 (Read error: 113 (No route to host))
16:24:26 Quit rwong (Remote closed the connection)
16:24:36 Join rwong [0] (
16:24:54 Join toffe82 [0] (
16:26:19 Quit B4gder (Read error: 60 (Operation timed out))
16:27:19 Quit sarixe ("Leaving")
16:29:15 Join dfkt_dt [0] (i=dfkt@unaffiliated/dfkt)
16:29:46 Quit robin0800 (Remote closed the connection)
16:29:54***Saving seen data "./dancer.seen"
16:35:25 Quit dfkt (Nick collision from services.)
16:35:31 Nick dfkt_dt is now known as dfkt (i=dfkt@unaffiliated/dfkt)
16:39:18 Join evilnick [0] (i=0c140464@gateway/web/ajax/
16:39:20 Join sarixe [0] (
16:39:21 Join CaptainKewl [0] (i=2669ecc2@gateway/web/ajax/
16:40:04evilnickHi All; are there any settings that need changing to ensure that the Beast charges via USB?
16:41:18 Join bmbl [0] (n=Miranda@unaffiliated/bmbl)
16:43:26 Quit miepchen^schlaf ()
16:44:01 Join B4gder [241] (n=daniel@rockbox/developer/bagder)
16:47:10LambdaCalculus37evilnick: Heyo, let me check out for you.
16:47:27 Join MethoS-- [0] (
16:48:50LambdaCalculus37evilnick: Nope, doesn't seem to be anything you need to change.
16:50:49LambdaCalculus37n1s: Ping
16:50:55LambdaCalculus37Oh crap, forgot.
16:51:17LambdaCalculus37n1s: (for the logs) I need to know if you can help me check out if DFU mode exists on the 4th gen nano.
16:57:21 Nick Beaver`univ is now known as Beaver (
16:57:52 Quit jaykay (Read error: 110 (Connection timed out))
17:00:04 Join fdinel [0] (
17:02:56 Quit MethoS- (Read error: 110 (Connection timed out))
17:05:52 Join webguest98 [0] (n=5657aa5c@gateway/web/cgi-irc/
17:07:58webguest98hello I have a problem with my rockbox. when my player is on the usb recharging than my sansa e280va works normally. but when I disconnect my player then he turns off and don't start up again. I can don't start up him by my self. can someone help me?
17:10:10 Quit webguest98 (Client Quit)
17:10:13 Join webguest15 [0] (n=5657aa5c@gateway/web/cgi-irc/
17:11:19 Quit webguest15 (Client Quit)
17:20:58 Join {-phoenix-} [0] (
17:26:51Unhelpfulevilnick: nope, it should work "out of the box". you'll need to have enough charge for it to start, though, to be able to charge from USB.
17:28:02PaulJampetur: have you made any progress with the H300 bootloader issues?
17:28:38peturnope... Linus was looking into it but he has no time right now...
17:29:27 Join robin0800 [0] (
17:30:25evilnickUnhelpful: So is it expected behaviour for the beast to turn off due to lack of battery power, when used as an external HDD? Maybe if there are constant transfers then the beast can't get enough power from USB alone?
17:31:48Unhelpfulerm, possibly? you should probably use the AC if you're doing that.
17:36:46evilnickCool - makes sense.
17:37:27 Quit fdinel (Read error: 110 (Connection timed out))
17:43:33 Join gregzx [0] (
17:46:29 Part TMM ("Ex-Chat")
18:06:16 Quit perrikwp (" ajax IRC Client")
18:15:49 Quit petur ("work->home")
18:21:00 Join rocko [0] (
18:23:28jhMikeSevilnick: The beast isn't going to necessarily maintain charge very well on USB alone with heavy HDD use.
18:24:35evilnickjhMikeS: I'm glad that I have the charger with me today then ;)
18:24:41 Join domonoky [0] (n=Domonoky@rockbox/developer/domonoky)
18:28:42 Join bertrik [0] (
18:29:57***Saving seen data "./dancer.seen"
18:31:33 Join lymeca [0] (
18:42:00 Join jaykay [0] (
18:45:51jaykayjhmikes: with FS #9910 the runtime was ten minutes less
18:46:00jaykayi guess i did something wrong...
18:50:49 Join flydutch [0] (
18:54:08 Quit tyfoo (Read error: 104 (Connection reset by peer))
18:54:30 Quit B4gder ("It is time to say moo")
18:56:22 Join jordoex [0] (
19:00:44jhMikeSjaykay: 10 minutes? I'd think normal variability could cause more than that (initial charge level, different HD activity, etc.)
19:03:01 Quit pyro_maniac ("Leaving.")
19:03:13 Join saratoga [0] (n=9803c6dd@gateway/web/cgi-irc/
19:03:57 Quit robin0800 (Read error: 104 (Connection reset by peer))
19:04:06jaykayok, i took the time from 90%to zero to avoid a different charge level... now your patch gives 10min more
19:04:22jaykayand its on the e200, so without a hd
19:06:28saratogawhat format?
19:07:20jaykayogg VBR 256kbps
19:08:18 Join Lear [0] (n=chatzill@rockbox/developer/lear)
19:08:57jhMikeSI might expect some minutes more at most, but still it could get swamped in variability. Consistently less over many trials I would be surprised (but it would take many).
19:10:24jaykayshould o do some more benches?
19:11:42jhMikeSIf it's not much trouble, thanks. At least we get 1-2 MHz back even if it doesn't extend the batt by much (or at all). I mean I could just turn mine on and and let it go too. :)
19:12:22jhMikeSpardon I meant "%" back.
19:18:20jhMikeSIn fact, I've just started one (but I have no previous data of my own).
19:18:35jhMikeSjaykay: What was the total time?
19:19:23jaykayjhmikes: with patch: 19:18, without: 19:29
19:20:27jaykayi also had no previous data, only a bench with r19896, and with that revision you patch didnt patch
19:20:44 Quit bertrik (Read error: 60 (Operation timed out))
19:22:07jhMikeSThat's from just cutting it at 90%? I though you said it was 10min more if you just gave the total time from the battery bench.
19:23:02 Join bertrik [0] (
19:25:47jhMikeSon the more recent ones, 90% appears to be a good 2hrs into the run
19:26:34jaykayjhmikes: no without cutting: your patch gives ten minutes less. with cutting: the patch gives 10 minutes more
19:26:46 Join idshark [0] (
19:27:12 Join einhirn [0] (
19:28:00 Join Thundercloud [0] (
19:28:42 Quit idshark (Remote closed the connection)
19:29:16jhMikeScan you post both of those?
19:29:24jaykayjhmikes: i deeply apologize, with cutting it gives exactly the same time
19:29:43jaykayyes, where? in the tracker?
19:30:01jhMikeSpastebin even
19:30:03 Join Horscht [0] (n=Horscht@xbmc/user/horscht)
19:34:58jaykaywithout patch:
19:35:06jaykaywith patch:
19:35:19 Quit linuxstb (Read error: 113 (No route to host))
19:38:58jhMikeSjaykay: thanks
19:39:49 Quit midijunkie ("?(???~•~)?")
19:48:11 Join efyx_ [0] (
19:48:26 Join bluebrother [0] (n=dom@rockbox/developer/bluebrother)
19:48:57jhMikeShrm, I can't really see any real difference. The unpatched one started off with a little higher charge it appears.
19:51:08 Join perrikwp [0] (i=98213ca1@gateway/web/ajax/
19:52:38jaykayjhmikes: and the first data from the patched one is after 41 seconds :) so the start chrage doesnt make the difference
19:53:00jaykayif youre not totally convinced, cut 10% off
19:53:55 Join mcuelenaere [0] (n=mcuelena@rockbox/developer/mcuelenaere)
19:54:15mcuelenaereany Cowon D2 user in here with some free time available to test a patch?
19:55:44jhMikeSI'm looking at the initial dip, the rise at the start (presumably after backlight off).
20:02:33 Join miepchen^schlaf [0] (
20:06:21 Join shotofadds [0] (n=rob@rockbox/developer/shotofadds)
20:06:57 Quit rocko ("Leaving")
20:07:03shotofaddshi mcuelenaere, I saw your touchscreen patch earlier and will give it a try later (i'm just off for some dinner now...) btw. it's nice to be back ;-)
20:07:18mcuelenaeregood to see you back :)
20:07:32 Join MethoS- [0] (
20:11:54 Join tyfoo [0] (
20:13:32jhMikeSjaykay: 70149-5474=64675 (17:57:55), 69483-5209=64274 (17:51:14) (patch) counting from 90% (where I see 90% clearly start)
20:15:00jaykaywheres the problem (if there is any)?
20:15:53 Join dan__ [0] (n=dan@dynamic-oit-lifenet-b-3.Princeton.EDU)
20:16:00 Quit Thundercloud (Remote closed the connection)
20:17:00 Quit FlynDice (Remote closed the connection)
20:17:27 Join linuxstb [0] (n=linuxstb@rockbox/developer/linuxstb)
20:19:22 Quit MethoS-- (Read error: 110 (Connection timed out))
20:19:40jhMikeSjaykay: No problem. I appreciate your taking the time mostly.
20:20:20 Join pyro_maniac [0] (
20:21:19jaykayill do two more benches (with and without patch) and see whether they confirm the first results
20:21:55 Join FlynDice [0] (
20:21:55jaykayis it possible that you patch shortens battery runtime?
20:22:54 Join Shav3 [0] (n=gabriele@
20:23:09Shav3how to install a theme ?
20:23:28 Quit perrikwp (" ajax IRC Client")
20:25:15jaykayshav3: search the manual or the wiki....
20:30:00***Saving seen data "./dancer.seen"
20:33:16 Quit Shav3 ()
20:44:42bluebrotherLambdaCalculus37: that pastebin looks weird ... like some internal Qt issue :(
20:44:58bluebrothermaybe domonoky can create a build based on the patch?
20:45:37 Join perrikwp [0] (i=98213ca1@gateway/web/ajax/
20:45:56*domonoky logs into jdgordons mac.. where is this patch ?
20:46:18LambdaCalculus37bluebrother: I installed Qt 4.4.0 from source.
20:46:21bluebrotherdomonoky: should be the latest one −− updated shortly ago
20:46:30 Quit pyro_maniac (Read error: 104 (Connection reset by peer))
20:46:41 Join pyro_maniac [0] (
20:51:06domonokyit complains about bootloaderinstallbase.cpp:223, "ent" doesnt exist
20:53:16bluebrotherdomonoky: patch updated
20:53:25bluebrotherseems I uploaded the old file instead :(
20:54:00*bluebrother feels stupid and goes standing in the corner
20:54:33domonokyare there more changes then this fix ?
20:55:10bluebrotherI increased the timeout delay
20:55:38gevaertsmcuelenaere: I guess you don't have a PP device?
20:55:54mcuelenaeregevaerts: why?
20:56:01gevaertsFS #9923
20:56:35mcuelenaereoh, you want me to test it with the usb tracer?
20:57:07gevaertsDepends. If it really seems to improve things, the tracer is the only way to be sure if the issue is really gone
20:57:08mcuelenaerehmm even if I would have a PP device, I don't have access to the tracer during the week (it's at home currently)
20:58:03 Join yziquel_ [0] (n=55da41ba@gateway/web/cgi-irc/
20:58:09gevaertsIt's not that urgent. It's just that you're the current tracer-holder. Maybe I should send you my c250...
20:58:19mcuelenaereheh :)
20:58:30mcuelenaereor I should send the tracer back
20:58:46mcuelenaere(even though I should still use it to finally get USB fully working on the Onda)
20:58:58gevaertsIf you're done with it, sure, but if it's still useful for you I think you'd better keep it
20:59:20mcuelenaereyes it's useful, I just need to spend some time with it..
21:00:28 Join dasvolk [0] (
21:00:39 Quit killan ("( :: NoNameScript 4.22 :: )")
21:01:03dasvolkis anyone doing work on a port for the 6th/7th gen ipods?
21:01:24gevaertsLet's wait for a bit of normal testing of this patch first. If it seems to really improve things, we can discuss how to do the tracer tests
21:01:38bluebrotherdasvolk: no, not actively. See the forum thread about those Ipods −− that's the progress
21:01:45 Quit einhirn (Read error: 104 (Connection reset by peer))
21:01:59bluebrotherso far nobody has been able running custom code on those Ipod in general, so there is no progress
21:02:38dasvolkdamn, that sucks.. I was hoping to put it on my girlfriend's ipod
21:02:45yziquel_hi. i've got a problem on my m:robe box. It seems that it went out of batteries, and now the Rockbox boot loader fails to find /System/OF.mi4 and then fails to find /System/OF.bin. I just cannot get it to connect to my laptop to see what's on the drive. How should I proceed to recover the system?
21:03:10dasvolkI was able to revive a mini with it, the battery tolerances on theapple firmware are damn lies
21:03:20 Join killan [0] (
21:04:16domonokyLambdaCalculus37: try this file">
21:05:26 Quit lymeca (Read error: 110 (Connection timed out))
21:13:19 Quit bluebrother (Nick collision from services.)
21:13:20LambdaCalculus37domonoky: I'll download it and try it out later, when I get off work.
21:13:23 Join bluebroth3r [0] (n=dom@rockbox/developer/bluebrother)
21:13:40*domonoky hopes it works.
21:13:41 Part dasvolk
21:14:23 Nick bluebroth3r is now known as bluebrother (n=dom@rockbox/developer/bluebrother)
21:17:06yziquel_hello. is there any documentation on how the m:robe boots? From when in the boot process is it possible to access the m:robe's filesystem?
21:17:10LambdaCalculus37domonoky: Don't worry, I'll let you know if it does or not. :)
21:17:46bluebrotheryziquel_: it works like the other mi4 devices −− it loads its firmware from a file in mi4 format.
21:18:06bluebrotheryou can plug in usb, then reset the player. It should go to usb mode.
21:19:09 Join roman__ [0] (
21:19:11 Quit perrikwp (" ajax IRC Client")
21:19:44 Join perrikwp [0] (i=98213ca1@gateway/web/ajax/
21:20:33yziquel_bluebrother: loading the .mi4 file is exactly what my player is unable to do. The screen keeps flashing in a loop complaining about that. thanks for the tip on how to get in usb mode...
21:21:06 Join buk_ [0] (
21:21:12linuxstbbluebrother: So it loads the file directly, rather than "flashing" it to a firmware partition (like the Sansas) ?
21:21:31 Join lymeca [0] (
21:22:00bluebrotherlinuxstb: yes. All mi4 bootloader installs only place (and move) a file
21:22:08bluebrotherwell, except sansa ;-)
21:23:07linuxstbThe Sansa's official firmware upgrade is the same - put an mi4 file on the filesystem, and it will get copied to the firmware partition. We use sansapatcher in order to keep a copy of the OF in the firmware partition as well (for easy dual-booting)
21:23:11 Quit Horscht ("Verlassend")
21:23:41 Nick {-phoenix-} is now known as {phoenix} (
21:24:08bluebrotherwell, m:robe 100 f.e. simply has a file in the filesystem (which is hidden by default). We can simply move the OF mi4 file and place ours instead
21:24:13saratogathe sansa stuff seems to be sandisk's work, the update works exactly the same on AMS players
21:24:21saratogaso presumably they're not using the PP code for it
21:25:42linuxstbNo, but Sansa's method is sensible - keeping the firmware off the main filesystem.
21:25:52*linuxstb realises Rockbox doesn't do that ;)
21:26:35saratogagevaerts: how do you check for bus resets with PP USB on linux? the wiki doesn't mention it
21:26:46gevaertssaratoga: dmesg
21:29:00 Join Zoxc [0] (
21:29:16 Quit Zoxc (Client Quit)
21:29:52gevaertsIt looks promising anyway. I've just copied 2GB back and forth to my c250 with two hubs in between without any resets, whereas before it usually doesn't even enumerate without resets in that setup
21:30:48 Quit __lifeless (Remote closed the connection)
21:31:04 Join __lifeless [0] (n=lifeless@
21:33:07 Join Aurix_Lexico [0] (
21:36:24 Quit lymeca (SendQ exceeded)
21:36:36 Join bs66_1 [0] (
21:38:16 Join Horscht [0] (n=Horscht@xbmc/user/horscht)
21:40:37linuxstbDoes anyone know why FS #9923 is just for the PP5022? Are people testing it on other PP chips as well?
21:41:25saratogai think dreamlayers only has a 5G, so thats probably why
21:42:23bertrikwhat does that patch do anyway? it seems to 'clock' in some values
21:43:44 Join apaininth [0] (
21:47:14 Quit apaininth (Client Quit)
21:47:53 Join apaininth [0] (
21:48:20 Quit shadearg (Read error: 60 (Operation timed out))
21:48:47gevaertsWe've looked at similar bits before (from some OF), but without much (or indeed any) result. I guess the "this initialization must be done after USB is reset" bit is important
21:48:55 Join pyro_mania1 [0] (
21:49:19 Quit pyro_maniac (Read error: 104 (Connection reset by peer))
21:49:42 Join petur [50] (n=petur@rockbox/developer/petur)
21:49:54 Join shadearg [0] (
21:49:57 Quit apaininth (Read error: 104 (Connection reset by peer))
21:50:14 Quit Horscht ("Verlassend")
21:52:10mcuelenaeregevaerts: for using USB serial on a non-bootloader build, do I only have to compile with logf() and USB_SERIAL enabled in usb_core.h?
21:52:53gevaertsmcuelenaere: basically, yes. It won't be enabled before you select the right option in the debug menu though
21:53:38mcuelenaereis there some kind of priority management in usb_core?
21:53:53mcuelenaerebecause debug_menu only seems to enable the USB serial driver, and not disable mass storage?
21:54:19gevaertsah yes. That's because the ARC core can handle both at the same time
21:54:32gevaertsYou'll have to disable the other I guess
21:54:39mcuelenaereoh ok
21:55:12 Join Horscht [0] (n=Horscht@xbmc/user/horscht)
21:55:16mcuelenaere(just out of interested, is there any predicted behaviour when both drivers would be enabled?)
21:55:46gevaertsI *think* the one with the lowest index in the driver table will win
21:55:46 Join midijunkie [0] (
21:56:11gevaertsAnd that's mass storage
21:56:40 Quit Aurix_Lexico (Read error: 110 (Connection timed out))
21:56:44 Quit LambdaCalculus37 (" ajax IRC Client")
21:57:19yziquel_bluebrother: wow... just had a look at the .rockbox folder.
21:57:52mcuelenaeregevaerts: hmm shouldn't adding 'usb_core_enable_driver(USB_DRIVER_MASS_STORAGE, !serial_enabled);' to logf_usb_serial() do the trick?
21:57:53 Join Aurix_Lexico [0] (
21:58:06bluebrotheryziquel_: looks like a corrupt filesystem.
21:58:18yziquel_can a battery getting flat cause such a corruption?
21:58:22gevaertsI've now copied 4.5GB back and forth to my ipod video over those two hubs, without a single reset. This ipod usually doesn't enumerate at all over that setup
21:58:56yziquel_bluebrother: so can it be the mass storage device partitions that are in a bad shape?
21:59:04linuxstbgevaerts: Do you only have PP5022 devices?
21:59:05gevaertsmcuelenaere: for your case, yes, but being able to use serial to debug mass storage was *very* useful :)
21:59:20mcuelenaerewell it doesn't seem to work here
21:59:24bluebrotherthat doesn't necessarily mean the partitions being messed up. It's the filesystem.
21:59:41bluebrotherof course it could also the hard disk dying and causing weird filesystem issues.
21:59:58 Quit bs66_ (Read error: 110 (Connection timed out))
22:00:01bluebrotherbut with a messed up partiton table you won't be able to simply mount the filesystem at all, so ...
22:00:06gevaertslinuxstb: no. I have two 5020s (mr100 and h10/5)
22:00:16mcuelenaereWindows first gives me some message about 'composed USB drives' (samengestelde USB apparaten), then something about a mass storage driver and then Rockbox crashes
22:00:29linuxstbgevaerts: Did you have the same reset issues on those?
22:00:37gevaertsmcuelenaere: that bit obviously needs work then...
22:00:52rasherIs this patch expected to help for e200?
22:01:14gevaertslinuxstb: IIRC they are a lot less sensitive to it, but I should retest
22:01:16gevaertsrasher: yes
22:01:20linuxstbrasher: The patch needs changing - it's currently #if CONFIG_CPU==PP5022 (e200 is PP5024)
22:01:30rasherlinuxstb: ah
22:01:31yziquel_bluebrother: how can i have a more in depth look at that issue?
22:01:44gevaertslinuxstb: isn't the PP5024 considered a PP5022 in the source?
22:01:51linuxstbNot in config-e200.h
22:02:28bluebrotheryziquel_: first you should chkdsk / fsck.vfat the drive
22:02:54gevaertshm, true
22:03:12gevaertsI guess it needs a bit of testing first
22:03:34saratogaoh i didn't change that for the sansa
22:04:16linuxstbProbably "#ifdef CPU_PP502x" is easiest...
22:04:49rasherIf appropriate
22:05:03gevaertsOnly if it also applies to PP5020
22:05:10linuxstbOf course...
22:06:21gevaertsI think I'll test it on e200 and mr100 now. That should clear things up a lot
22:06:39 Quit Lear ("ChatZilla 0.9.84 [Firefox 3.1b3pre/20090213033101]")
22:06:58 Join Thundercloud [0] (
22:07:52*rasher copies a few gigabytes
22:08:54rasherNo resets yet.. Pretty sure I used to get them with the front port of my desktop
22:09:41gevaertsdid you fix the patch to also do PP5024?
22:10:25 Join midgey|work [0] (i=8dd3ea49@gateway/web/ajax/
22:11:05jhMikeSit's looking promising about enabling the stack by default now?
22:11:32rasherAs far as I know, the signal quality issues were the last thing holding it back?
22:11:50midgey|workit'd be a heck of a 3.2 feature
22:12:33gevaertsI want to commit this patch tonight, but I think it would be wise to postpone enabling the stack by default until we've had a look with the tracer
22:12:52jhMikeSthat will be a major relief, H10 is quick to reboot and test on but e200 is just a pain (database rebuilt supression hardly ever works).
22:12:58 Join n1s [0] (n=n1s@rockbox/developer/n1s)
22:13:22yziquel_bluebrother: thanks a lot.
22:13:28jhMikeSMy troubles aren't very severe though. I have to copy alot of data to get any resets on anything I plug, if I get any.
22:13:28bertrikjhMikeS, oh it doesn't work for you?
22:13:28 Quit yziquel_ ("CGI:IRC")
22:14:53jhMikeSbertrik: I haven't tried the patch per what I said above. I'm not a good candidate. Maybe I should try H10 just to make sure it keeps working there since it's a PP5020
22:15:00bertrikjhMikeS, IIRC some of the old OF settings are cleared only when upgrading the OF through sansapatcher, and this settings clearing is required in some cases for e200 database update suppression to work.
22:15:04evilnickjhMikeS: Of course, if you choose formats that the OF doesn't recognise then the rebuild is fairly minimal
22:15:32 Quit rwong (
22:15:35 Quit Slasheri (
22:15:35 Quit jon-kha (
22:15:40rasherHow do I check I'm connected at hi-speed?
22:15:51Bagderdmesg! ;-)
22:15:58gevaertsrasher: linux? If so, dmesg
22:16:05*rasher feels sheepish
22:16:10saratogathe database never updates for me, its nice
22:17:03jhMikeSbertrik: I used sansa.fmt and completely redid the e260. It worked a few times (but not 100%) and then stopped altogether.
22:17:11*pixelma quite reliably gets resets with the c200
22:17:30midgey|worki'm in the same boat as jhMikeS - datebase refreshes happen fairly often
22:17:38pixelmaso have a nice test setup here
22:18:26jhMikeSI had another firmware problem to fix where it would just hang for awhile at times. That problem never came back after blowing it out.
22:18:28midgey|workprobably has to do being in michigan....
22:18:41bertrikthere's some code in sansapatcher to clear the OF settings which should make the database refresh suppression work , but it's only called in case of an OF upgrade if I remember correctly
22:18:52jhMikeSprobably, all that water + electronics
22:19:33 Join faemir [0] (
22:20:09jhMikeSthat means even if you use sansa.fmt and redo from scratch, it still won't be called?
22:20:40 Quit toffe82 (Remote closed the connection)
22:21:05bertrikindeed. Have a look at sansapatcher.c line 917 and further where it clears the OF "nvparams" area.
22:21:31 Join LinusN [0] (n=linus@rockbox/developer/LinusN)
22:21:33bertrikmaybe that part should have its own sansapatcher command
22:23:18 Quit n1s ("Lämnar")
22:23:37 Join n1s [0] (n=n1s@rockbox/developer/n1s)
22:24:03*mcuelenaere doesn't understand why USB doesn't work in normal Rockbox while it works in the bootloader..
22:30:03***Saving seen data "./dancer.seen"
22:30:43saratogahmm my sansa only connects in full speed mode
22:30:46saratogai wonder why
22:32:35rasherOkay, 2.0GB just transferred to e280 main memory without any resets. 900MB transferred to msd wihtout resets.
22:33:12gevaertsSame thing here. For some reason my e200 is a lot slower than both c200 and ipod video though
22:33:24saratogadid it connect in highspeed mode?
22:33:31gevaertsit did, yes
22:33:33rasherIt seems slow here as well
22:33:45rasherAlthough I've no numbers
22:34:04rasherJust seems like it took longer than it "should"
22:34:07NJoinrwong [0] (
22:35:22 Join Neovanglist [0] (i=Neovangl@
22:36:17 Quit roman__ (Read error: 110 (Connection timed out))
22:36:34 Quit bertrik ("Leaving")
22:37:14 Quit mcuelenaere ()
22:38:03 Join toffe82 [0] (
22:38:04 Quit bluebrother ("bbl")
22:39:06buk_mcuelenaere: to have usb serial working I had to remove "usb_core_enable_driver(USB_DRIVER_MASS_STORAGE, true);" in usb.c (about line 300)
22:39:22 Quit pyro_mania1 (Read error: 113 (No route to host))
22:39:35 Join pyro_maniac [0] (
22:39:43saratogais there some way i can try and force highspeed mode?
22:40:02 Part CaptainKewl
22:40:03Unhelpfulhrm, my e200 hung when i tried to use serial logf, maybe that was my problem?
22:40:44gevaertssaratoga: you're not using -DUSB_NO_HIGH_SPEED, right?
22:40:44buk_I forget to mention I used it on Cowon D2...
22:41:00saratoga"export EXTRA_DEFINES=-DUSE_ROCKBOX_USB"
22:41:21gevaertsshould be ok
22:41:51 Quit toffe82 (Remote closed the connection)
22:42:15*rasher checks md5sums
22:43:34 Quit perrikwp (" ajax IRC Client")
22:43:47rasherAlthough transfer errors didn't really occur did they?
22:52:31 Join jon-kha [0] (
22:53:09 Join Slasheri [0] (
22:54:18 Join toffe82 [0] (
22:55:53 Quit faemir ("Lost terminal")
22:56:54gevaertsrasher: it seems that e200 not slower than without this patch
22:57:10rashergevaerts: Well that's good
22:57:41rashermd5sums checked out
22:57:56rasherNow to try without the patch to see if I did in fact get resets previously
22:58:02rasherIt's been a while since I used Rockbox USB
22:58:07 Join gartral [0] (
22:59:28BigBambiWould anyone like to comment on (manual install chapter rework). I'd like to get it in, but it changes the layout quite a lot, so would appreciate opinions :)
22:59:51gartralon my e250 under the debug menu, in hardware info, it says PP version: PP5022C.. whats the"C" for?
23:00:06gevaertsBigBambi: unless it's been changed a lot since last time, go for it
23:00:25BigBambigevaerts: I haven't changed it at all
23:00:32BigBambiso OK :)
23:03:29 Quit jaykay (Read error: 110 (Connection timed out))
23:05:59 Quit n1s ("Lämnar")
23:06:14 Join n1s [0] (n=n1s@rockbox/developer/n1s)
23:06:55 Quit n1s (Client Quit)
23:09:49 Join akur [0] (
23:10:19 Part akur
23:11:25 Join domonoky1 [0] (
23:13:58 Quit tchan ("WeeChat 0.2.7-dev")
23:14:14 Quit shotofadds ("Leaving")
23:17:29 Quit ender` (" In every computer error, there are at least two human errors, including the error of blaming it on the computer.")
23:17:46 Join n1s [0] (n=n1s@rockbox/developer/n1s)
23:19:40 Part pyro_maniac
23:19:57 Quit n1s (Client Quit)
23:20:02 Quit domonoky1 (Read error: 104 (Connection reset by peer))
23:20:13 Join n1s [0] (n=n1s@rockbox/developer/n1s)
23:20:40 Quit n1s (Client Quit)
23:20:54 Join n1s [0] (n=n1s@rockbox/developer/n1s)
23:21:04 Join LambdaCalculus37 [0] (i=41585845@rockbox/staff/LambdaCalculus37)
23:21:35 Quit petur ("Zzzzz")
23:22:45rasherGetting a million errors with Rockbox USB without FS #9923 (e280, hi-speed usb, front port usb)
23:22:54rasherSo I think it's safe to say that the patch is an improvement
23:24:07*gevaerts commits
23:25:01rasherMaybe it'd be a good idea to provide some test-builds for people to try?
23:25:14gevaertsOnly for 5022/5024 though. It doesn't solve anything on 5020
23:25:22*midgey|work does the USB danc
23:25:31midgey|workand by danc i mean dance
23:25:34rashergevaerts: Did you enable USB also?
23:25:57gevaertsrasher: we could do that to make the "I want to help by testing" crowd happy, but the only tests I want to see are hardware analyser tests
23:26:19gevaertsrasher: no. That will have to wait until we've seen actual traces with zero errors in them
23:26:37rashergevaerts: Still, it might help identifying targets/areas of interest
23:27:27midgey|workwhat ever happened to those USB conformance tests?
23:28:10gevaertsI guess I should re-run those one of these days
23:30:22gevaertsrasher: you mean things like specific targets that behave differently than expected?
23:31:46gevaertsIn terms of supported targets we're talking about ipod mini 2G, ipod nano, ipod video, c200 and e200. I have all of those except the nano
23:32:48 Quit n17ikh ()
23:33:26gevaertsIf someone wants to make test builds, feel free, but I don't really see the point
23:34:42*rasher won't bother, then
23:34:46n1sgevaerts: has that same init been tested on pp5020?
23:34:53gevaertsn1s: yes
23:35:08gevaertsIt doesn't help there
23:35:08n1sand no improvement?
23:36:39gevaertsIt may be a bit better, or a bit worse, but for this issue I basically don't think it's worth counting resets. If you have bad packets on the wire it's buggy, you don't even need to see bus resets for that
23:38:20LloreanHow long is PP5022 testing likely to take.
23:38:33LloreanSpecifically, it'd be really neat to roll it out to users "soon" since it'd be quite amazing to see it in 3.2
23:39:10gevaertsLlorean: once we bring the players and the tracer together, an hour or so.
23:39:19gevaertsThis will be in 3.2
23:39:30 Quit domonoky (Read error: 110 (Connection timed out))
23:39:39rasherFamous last words.
23:39:48gevaertsexcept if it doesn't actually fix the bug of course, but I think that's unlikely
23:40:54LambdaCalculus37domonoky: Ping!
23:41:01LambdaCalculus37Crud! :(
23:41:54n1sLambdaCalculus37: i'd be happy to help looking for a DFU mode on the nano 4g, any ideas to try?
23:42:34LambdaCalculus37n1s: Try doing a restore with iTunes, and interrupt the process (i.e. disconnect it from USB in the middle of the firmware transfer).
23:42:42LambdaCalculus37That should ruin the firmware.
23:43:04 Part LinusN
23:44:01n1sLambdaCalculus37: ok, will check that out, not today though
23:44:50LambdaCalculus37n1s: No prob; let me know when you do.
23:45:40LambdaCalculus37Once you get the firmware trashed, try to reboot it. When it boots, press and hold BACK+PLAY and see what happens.
23:46:38n1sI had another more complicated idea, which was snooping on the usb traffic when dooing a firmware restore with itunes.
23:47:30LambdaCalculus37n1s: Yes, that'll be good, too. But I want to see if there's still a DFU mode in later nanos.
23:48:11n1sdid you figure out anything fun to do with it on the 2g nano?
23:48:42LambdaCalculus37n1s: I tried sending stuff to it with meizu_dfu, but no dice.
23:50:26 Quit n1s ("Lämnar")
23:50:51 Quit flydutch ("/* empty */")
23:51:15Lloreangevaerts: Excellent
23:53:54 Quit LambdaCalculus37 (" ajax IRC Client")
23:53:58 Quit midgey|work (" ajax IRC Client")
23:54:04 Quit dude187 (Read error: 113 (No route to host))
23:55:24 Join LambdaCalculus37 [0] (i=41585845@rockbox/staff/LambdaCalculus37)

Previous day | Next day