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 2008-02-20

00:00:04amiconnBoost/unboost takes 250..500us on PP5022/5024, so it shouldn't be done too often
00:00:27hachiso I'm confused then
00:00:33hachiwith a known issue
00:00:35*amiconn wonders why his linux box doesn't automount the ipod
00:00:39hachinobody rolled back the changes that caused it?
00:00:47 Quit JdGordon|w (" ajax IRC Client")
00:00:55gevaertsamiconn: a 128k transfer at full speed takes 100 ms, so in comparison boost/unboost is lightning fast
00:01:05hachiI mean, that seems like the perfect candidate for 'hey... maybe I shouldn't have committed this code'
00:01:25 Join JdGordon|w [0] (i=836b0049@gateway/web/ajax/
00:01:52n1shachi: we are in active development, sometimes things will remain broken until someone figures out the correct fix
00:02:43rasherThe correct fix not always being "revert"
00:03:11hachiwell, I'm a bit stuck because I don't have any decent clue on how to build toolchains for myself... I'll give it a shot
00:04:06n1shachi: you could try out some build from the unsupported builds section in the forum
00:04:59stripwaxn1s - is that suggestion because those builds are old (and unsupported..), or because some of the unsupported builds incorporate fixes for the problem?
00:05:36amiconngevaerts: Mini G2 mounts fine under Linux (debian-amd64). Full speed of course, and there are some differences in dmesg output
00:05:48amiconn(comared to when using apple diskmode)
00:05:49n1sstripwax: at least the ryan temp one is from r 156xx so it's old enough
00:05:50rasherDomonoky, bluebrother (for the logs): You might want to tag/branch rbspeex when tagging/branching rbutil releases
00:06:32jhMikeSwhy worry about running boosted if the device is powered externally?
00:06:49gevaertsamiconn: some differences are expected. Can you paste the dmesg output somewhere ?
00:07:02 Join fasmaie [0] (
00:07:07amiconnThe H10 gets quite warm when running constantly boosted, even when not charging in parallel
00:07:10hachin1s: I'd kinda be more interested in learning how to build toolchains... I've got a nokia tablet that's arm based, and a neo 1973... these things come with a built toolchain so you can just 'go'
00:07:23hachinobody talks about how to build these things properly
00:07:28hachiyou're just supposed to know it somehow
00:07:47stripwaxhachi - you can read the wiki page on rockbox which tells you how to use the prebuilt toolchains
00:07:55n1shachi: well, then I recommend a trip to the wiki :)
00:07:56amiconngevaerts: First is apple diskmode, second is rockbox usb
00:08:04jhMikeSnothing I have gets very warm running boosted and the core sleeps more anyway
00:08:06hachiI was just up on the wiki, I don't see any section for it
00:08:17n1scheck the CrossCompiler page
00:08:21stripwaxhachi - keep looking, there's a VAST amount of developer info in the wiki..
00:09:20stripwaxstart at "DocsIndex" and then For Developers..
00:09:33gevaertsSpeed results including boosted at
00:09:51stripwaxbtw not necessarily necessary to actually build your own toolchain - ought to be able to use cygwin or vmware as appropriate
00:10:24rashergevaerts: nice speeds
00:10:29amiconngevaerts: How do you test speed?
00:10:50gevaertsamiconn: dd if=/dev/sdc of=/tmp/testfile bs=128k count=100
00:10:57 Quit einhirn (Read error: 104 (Connection reset by peer))
00:11:05gevaertsamiconn: and then back (so I don't destroy data)
00:11:57hachithat's funny... I figured the front page of the wiki would be the index :\
00:12:23amiconneh? That outputs several hundred MB/s here...
00:12:30scorche|windexes are usually at the end of books...not the beginning..
00:12:49amiconnHmm, no, only if I use the partition
00:13:06scorche|wthen again, it isnt a *true* index
00:13:12stripwaxhachi - no, the front page states "Some pages of interest". wiki is searchable, by the way..
00:13:26amiconn620kB/s without boosting and with whatever blocksize the patch defaults to
00:14:06gevaertsI defaults to 64k
00:14:31amiconnHaha, and the write test fails immediately
00:14:41amiconn...due to sector 0 protection in the ata driver
00:14:55amiconn*PANIC* Writing on sector 0
00:15:35gevaertsamiconn: maybe add skip=1 (or 2) to the dd line ?
00:16:08*gevaerts values his ability to ruin a partition table easily
00:16:58*amiconn will try to skip the firmware partition, so he doesn't need to restore in case of (not too hard) failure
00:17:13gevaertsamiconn: or seek... I never can remember which one. Also probably best try to use matching skip and seek for read and write
00:17:22 Quit alienbiker99 (Read error: 104 (Connection reset by peer))
00:18:09 Join robin0800 [0] (
00:19:41gevaertsamiconn: about the dmesg output, the differences that I see are the disk model and version (we get them from ata identify), Mode sense where I'm still not sure which pages we actually need, so it outputs none, and ANSI:, which specifies the SCSI spec version we follow (4=spc-2), Apple denies having read the spec, we don't
00:19:42JdGordon|wskip is read seek is write iirc
00:21:10 Join barrywardell [0] (n=barrywar@
00:21:15robin0800the current build is 16354 this appears to be one commit back from the front page however the latest commit is 16357 whats going on?
00:22:10amiconnHmm, writing doesn't seem to work...
00:22:45gevaertsWhat happens ?
00:22:47 Quit MethoS- ("Konversation terminated!")
00:23:16 Quit roxfan (Connection timed out)
00:23:45amiconndd finishes after ~50ms, and claims 231MB/s write speed
00:23:59amiconnEmulated disk led icon stays on
00:24:13*gevaerts wants a disk like that
00:24:22amiconnThere seems to be no further problem
00:24:41amiconnFor read accesses, the disk led icon properly follows activity
00:24:45gevaertsDoes the disk have a write cache ?
00:25:14amiconnThat would certainly not cache 13MB
00:25:21gevaertsForget that. USB speed should still slow it down
00:25:44*amiconn didn't use a repeat count either, but transferred a larger block instead
00:25:47 Join cool_walking_ [0] (
00:27:14 Join cheetah65 [0] (
00:27:16gevaertsMaybe add conv=fdatasync or oflag=direct ?
00:27:36*amiconn now tried to write an actual file
00:28:14pixelmarobin0800: the front page lists all commits, including the ones for the manual and rbutil which don't trigger a build round (no code changes for the targets). But I can't see what r16355 and 16356 should have been, even svn log doesn't tell me...
00:29:38rasherpixelma: domonoky tagging a rbutil release
00:29:39amiconnFilesystem is still okay (checked with apple diskmode and windows chkdsk)
00:30:05gevaertsThat's good news at least
00:30:40amiconnWindows mounts it fine via rockbox usb as well
00:32:08cheetah65i had a rockbox related question, whenever i play a song with album art enabled, it hangs up for a second or two. ive tried clearing settings, the whole 9 yards. any ideas?
00:32:17pixelmarasher: thanks, I just don't know what that involves but I would have expected that this also shows up somewhere in the history
00:32:21cheetah65nothing on forums
00:32:30scorche|wpixelma: it doesnt track branches
00:33:29pixelmaah hmm... vague idea now :)
00:33:33amiconn...and writing works there too.
00:33:54*amiconn is now playing an album on his mini which he copied from windows using rockbox usb mode :)
00:34:24stripwaxcheetah65 - what device do you have, which build of rockbox are you running?
00:34:46cheetah65h10 20gb, latest daily build
00:34:53scorche|wpixelma: in other words, it will only monitor "trunk" and not "branches", "songdbj", "tags" or "www"
00:34:56gevaertsHow does boost work ? Do I just call cpu_boost(true) when I need speed, and cpu_boost(false) when I don't ?
00:35:02amiconngevaerts: Does the patch use 64 byte packets, or still 16?
00:35:09gevaertsamiconn: 64
00:35:43stripwaxgevaerts - oh, what was the fix for the corruption when using >=32 byte packets?
00:35:55*stripwax missed that
00:35:58jhMikeSgevaerts: you can use trigger_cpu_boost/cancel_cpu_boost which don't accumulate counts and are thread local
00:35:58amiconncpu_boost(true) increments a counter, cpu_boost(false) decrements it. Rockbox will boost when the counter becomes > 0 and unboost if it becomes == 0
00:36:21 Quit kugel ("ChatZilla 0.9.81 [Firefox]")
00:36:23amiconnSo multiple threads can boost/unboost as needed, and we don't loose track
00:36:32gevaertsstripwax: good question. I guess it was a caching problem, which got fixed somewhere on the way, but I don't know exactly when
00:36:46markunamiconn: do you think it's likely that this 380HN chip is the UDA1380HN?
00:36:48rasherThis boosting system always appeared elegantly simple to me.
00:37:09pixelmascorche|w: yes, understood now. Thanks :)
00:38:07stripwaxcheetah65 - did this problem start occurring for you recently, or has it always happened for you? and by 'hangs up', do you mean that the audio doesn't come out more-or-less immediately −− or that it *does* start playing audio but the screen doesn't refresh until later?
00:38:11gevaertsamiconn: The problem will be to unboost exactly once if I only boost on actual I/O. If some I/O fails the usb_storage code might not get called again properly. I could add a flag somewhere to keep track of this, but that somehow seems wrong
00:38:46cheetah65stripwax: it started occuring after AA was implemented
00:38:53jhMikeSgevaerts: the trigger/cancel functions track for you
00:39:06cheetah65im going to try a fresh install right now
00:39:14stripwaxcheetah65 - are you album art bitmaps particularly large?
00:39:37 Quit ender` (" I was in the grocery store. I saw a sign that said "pet supplies." So I did. Then I went outside and saw a sign that said "")
00:39:37amiconnmarkun: For sure.
00:39:39cheetah65i have a few different sizes
00:40:02Nico_PjhMikeS: aliask said he had gotten the SPI driver to work
00:40:14stripwaxActually I meant large in filesize rather than necessarily dimension. e.g. if these are all 24 or 32-bit bmps, try reducing colour depth and saving as e.g. 256 color bitmaps
00:40:17jhMikeSgevaerts: iow calling trigger/cancel repeatedly is ok as calling the opposite function of the previous call does the change
00:40:38gevaertsjhMikeS: Thanks. That sounds easier
00:41:31***Saving seen data "./dancer.seen"
00:42:16gevaertsNow the question is if this heat problem when boosted will not come back if you copy lots of files...
00:42:17 Quit conando (Read error: 110 (Connection timed out))
00:42:18n1smarkun: might be interesting
00:42:24jhMikeSNico_P: I had it working awhile ago but I want to have a new driver model for interrupt wakeup style (not spin/yield). we want this processor asleep as much as possible and to let the peripherals do the work since whatever it gets done it gets done quickly and it can spend most time doing absolutely nothing
00:43:00cheetah65my h10 is refusing to connect right now, i'll come back at a later time
00:43:22jhMikeSI guess the rockbox USB stack works that way though I have a new kernel object optimized for that purpose.
00:43:41Nico_PjhMikeS: I agree completely. IIUC aliask didn't manage to make it interrupt based
00:43:51 Part cheetah65
00:43:55 Join ol_schoola [0] (
00:44:25*amiconn thinks that the rockbox usb stack might even be faster than the H10's built-in diskmode
00:44:37jhMikeSI did the int style drivers for meg-fx i2c and ata to work it out on a running target
00:44:44*gevaerts thinks amiconn should try
00:44:44amiconnThe latter annoys me everytime I want to upgrade the build...
00:45:07 Quit spiorf ("Read error: 110 (Connection timed out)")
00:46:28 Quit fasmaie (Read error: 113 (No route to host))
00:47:45 Quit n1s ()
00:47:47rasherDomonoky/bluebrother (for the logs again): Any chance of a working make install target for rbutil?
00:47:55pixelmagevaerts: could it be that it is even faster than using USB with the c200 firmware (I mean including the wait for the database refresh)? ;)
00:48:04gevaertsIt would be useful to know if boost is really needed everywhere, or just on sansa.
00:48:08rasherDomonoky/bluebrother: And an option to avoid all the static linking
00:48:23 Quit homielowe (Remote closed the connection)
00:48:33amiconngevaerts: Given what transfer speed I get for reading, boosting should be useful on all PP502x targets
00:49:28gevaertspixelma: On OF I get 4.5MB/s reading or writing. That's still 5x as fast, so I guess it depends on how many files you have for the database refresh to look at.
00:49:54 Join ompaul [0] (n=ompaul@gnewsense/friend/ompaul)
00:50:16 Quit Nico_P (Remote closed the connection)
00:50:17jhMikeSgevaerts: I wondered about just buffering a bunch of data into the audio buffer that will be available and leaking it to disk as fast as possible. I think retailos does that.
00:50:26gevaertsamiconn: OK. I'll try adding that. I probably won't finish it today though (I don't like posting a new patch without having copied at least a few hundred megabytes with it both ways, which still takes time)
00:51:13 Quit JdGordon|w (" ajax IRC Client")
00:52:40jhMikeSare transfers done in parallel with disk read/write?
00:52:42gevaertsjhMikeS: That's certainly possible, but if we want to do that cleanly we have to de deferred SCSI error reporting (as well as letting the OS know that we do write caching). That adds some more complexity again.
00:52:48pixelmagevaerts: could imagine, the data transfer seem to be quite fast in the OF but then all the waiting... btw. is the patch safe to test on WinXP now?
00:53:17gevaertsjhMikeS: no. Everything is still purely sequential
00:53:58gevaertspixelma: it should be safe. Plugging and unplugging the sd card while connected will probably not work correctly, but everything else should be fine.
00:54:00jhMikeSso no new transfer is initiated until the previous data is fully read or written?
00:54:01amiconngevaerts: Similar differences in dmesg output for H10 (OF vs. rockbox) as for mini
00:55:16 Quit waldo (Remote closed the connection)
00:55:46 Quit robin0800 (" HydraIRC -> <- IRC with a difference")
00:55:49amiconnH10 is a bit slower on reading (as I expected - that disk is quite crappy)
00:56:23gevaertsjhMikeS: exactly. I would like to add a buffer so we can do that in parallel, but it probably won't be worth it on high speed (if/when we get that working), and it needs some of the same complexity as a big write cache (if we want to do it properly)
00:57:23 Join dandin1 [0] (
00:58:10gevaertsjhMikeS: we could of course just tell the OS that everything is fine, but I don't think that's a good idea. Sooner or later someone will loose data because of it.
01:00:57 Quit amiconn (Nick collision from services.)
01:01:04 Join amiconn [50] (n=jens@rockbox/developer/amiconn)
01:01:14*amiconn unzips a rockbox build from winodws to h10 using rockbox usb mode
01:01:54*gevaerts can't remember the last time he used the OF for that
01:02:12LloreanDoes it work well enough, other than being slow, to go into SVN?
01:03:20jhMikeSgevearts: have a header for each block and you can give it instructions to execute and return code field to store for each one and only involve the thread occasionally
01:03:43 Join courtc [0] (
01:04:46gevaertsLlorean: I can get it to crash more easily than the version currently in svn (v7). On the other hand I think it requires doing things a "normal" user doesn't do (run lsusb -v in a loop), and this also crashes the one in svn after a while (but it takes longer)
01:05:57gevaertsLlorean: Once the boosting code is added it will be about twice as fast as the one in svn, and I haven't heard anything about lost data recently.
01:06:32 Quit stripwax (Read error: 104 (Connection reset by peer))
01:07:28amiconngevaerts: It seems to work properly on H10 (small) as well
01:07:30gevaertsjhMikeS: I guess it would be something like that. I vote we stabilize a bit first though.
01:07:43amiconnThat's interesting as H10 is PP5020, not PP5022
01:07:45LloreanI'd just like to see a relatively stable, normal-user-usable USB mode, even if it's slow. So that we can get past the "why does it ask for drivers" stage of questions.
01:08:18amiconnUnzipped a whole build and did chkdsk afterwards
01:08:27gevaertsamiconn: great. So it has now been tested on nearly all pp502x devices (I'm not sure if we've had all types of ipod).
01:08:52amiconnIt's slower than OF diskmode though (without boosting at least)
01:08:58 Quit ompaul (Client Quit)
01:10:01jhMikeSgevaerts: of course it should be stable first. concurrency and some read/write behind will max it out ultimately (/me always wears the concurrency hat :)
01:12:08gevaertsLlorean: Give me a bit more time. I'd really like to try to get it a bit more solid.
01:12:10 Quit Isolinear (Read error: 110 (Connection timed out))
01:12:23Lloreangevaerts: Oh, I didn't mean to rush you or anything.
01:12:43LloreanJust stating that in my opinion, even if it's slower than disk mode, it'd be nice to have for people once it's solid enough. :)
01:15:24gevaertsjhMikeS: I agree we do need concurrency. The nice thing about this is that I can keep tweaking it and cleaning it up without some boss telling me that the budget for this project is now all used up, so as far as I am concerned it will get faster, better, and cleaner.
01:16:19amiconnOkay. Boosting makes rockbox usb almost as fast as OF on H10
01:17:02amiconnUnzipping a rockbox build from windows: OF 2:43 min, Rockbox boosted: 3:14 min, uboosted: 5:03 min
01:19:28*amiconn observed no data corruption so far
01:19:37gevaertsI think I'm going to experiment a bit tomorrow with using 20kb USB transfers (the maximum for one transfer descriptor), and 64k or 128k buffer for disk IO. That should bring a lot of simplicity back in the driver, while not being too expensive in speed terms
01:20:19*jhMikeS tells gevaerts to get to work and to stop lying around or you'll work through your lunchbreak :p
01:20:50scorche|wgevaerts: now that you mention the budget...
01:21:01*gevaerts thinks he will ignore jhMikeS's advice and go to sleep soon
01:21:17gevaertsscorche|w: don't tell me we ran out...
01:22:37scorche|wgevaerts: well we only had 0 to spend and we already spent 0...
01:22:59gevaertsscorche|w: so we used it all up already :(
01:23:05 Join jhulst [0] (n=jhulst@unaffiliated/jhulst)
01:23:21gevaertsOn the other hand, we still have everything left that we started out with
01:23:57rasherI guess it's time to have a cup of no tea, then
01:24:40 Quit dandin1 ()
01:25:26*gevaerts doesn't drink his cup of no tea
01:26:15 Join alienbiker99 [0] (
01:26:16preglowthen you won't spill it, at least
01:26:17*ze doesn't not drink a non cup of no tea
01:26:46amiconnmarkun: It's interesting that the meizu seems to use arm in big endian mode
01:28:40 Quit MajorC ()
01:31:58 Nick fxb is now known as fxb__ (
01:32:09 Join csc` [0] (n=csc@archlinux/user/csc)
01:33:09gevaertsOn my sansa with 64k buffers, it seems that boosting just for reading and writing is about 2% slower on reads and 3.5% slower for writes than manual boosting for the entire test.
01:34:13 Join Bradmont [0] (
01:35:25Bradmontcan anyone tell me how to uninstall themes & modules? I can't figure it out.
01:35:50Bradmonter, plugins
01:36:02LloreanWhy would you want to remove plugins, they'll just be back with the next update.
01:36:33LloreanAs for themes, just delete the files you installed. Each theme may have different files, so you'll have to go look, but generally you're looking for a .cfg file, a .wps file, and a bunch of bitmaps in a folder with the same name as the .wps file
01:37:04Bradmontwow, I feel stupid...
01:37:15BradmontI didn't notice the .rockbox hidden directory on my player...
01:38:14 Join moos [0] (
01:38:32*gevaerts adds a package manager to his list of useful additions to rockbox which already contained mkfs, fsck and fdisk
01:38:48 Quit Bradmont (Client Quit)
01:40:20 Join jurrie [0] (
01:41:36 Quit gevaerts ("Sleeping time")
01:46:52 Part toffe82
01:49:57 Join Soap_ [50] (n=Soap@rockbox/staff/soap)
01:53:10 Join axionix_ [0] (
01:58:33 Quit Thundercloud (Remote closed the connection)
02:00:07 Quit lee-qid (Read error: 110 (Connection timed out))
02:09:16 Quit axionix (Read error: 110 (Connection timed out))
02:11:46 Quit jgarvey ("Leaving")
02:12:09 Quit ol_schoola ()
02:29:28 Quit scorche|w ("CGI:IRC (EOF)")
02:35:52 Join goffa [0] (n=goffa@
02:41:32***Saving seen data "./dancer.seen"
02:43:18 Quit cool_walking_ (Read error: 113 (No route to host))
02:43:57 Join cool_walking_ [0] (
02:45:36 Join Lambduh [0] (
02:48:51 Quit barrywardell ()
02:50:12 Join TiMiD[FD] [0] (n=TiMiD[FD@
03:00:46 Quit Soap_ (Remote closed the connection)
03:00:53 Join hannoman [0] (
03:08:13 Part pixelma
03:08:26 Join Soap_ [50] (n=Soap@rockbox/staff/soap)
03:09:13 Join Rob222241 [0] (
03:15:13 Quit csc` ("Powering Off")
03:17:14 Join csc` [0] (n=csc@archlinux/user/csc)
03:18:25 Quit csc` (Client Quit)
03:23:31 Join csc` [0] (n=csc@archlinux/user/csc)
03:24:44 Quit Rob2222 (Read error: 113 (No route to host))
03:25:43 Quit csc` (Client Quit)
03:26:27 Join ol_schoola [0] (
03:30:38 Join csc` [0] (n=csc@archlinux/user/csc)
03:31:27 Quit hannoman ("Miranda IM! Smaller, Faster, Easier.")
03:35:01 Join crazy_bus [0] (
03:35:13 Join jcollie [0] (
03:35:28 Quit EspeonEefi ("さよなら")
03:38:14crazy_busthere's some things bugging me in the wiki. Is it possible to activate my wiki account so I can edit?
03:40:07 Join DerDom1 [0] (
03:40:09 Quit DerDome (Nick collision from services.)
03:40:17 Nick DerDom1 is now known as DerDome (
03:46:16Lloreancrazy_bus: Out of curiosity, what's bugging yah?
03:46:23 Quit DerDome ("Leaving.")
03:49:28crazy_bushere's one of the pages it says (only mic for now) but fm recording feature has being implemented in rockbox
03:50:35LloreanI was just curious, "bugging me" made me wonder if it was subjective content somewhere
03:50:47crazy_busthe spectrum player also doesn't mention working on my player but it does
03:56:13 Nick neskaya is now known as Neskaya (i=featherd@wiktionary/neskaya)
03:56:32 Quit Kernel (Remote closed the connection)
03:58:06 Quit Rob222241 (Read error: 104 (Connection reset by peer))
03:58:38 Join Rob2222 [0] (
03:58:49 Quit jcollie (Read error: 104 (Connection reset by peer))
04:04:32 Join Rob222241 [0] (
04:13:45 Join jcollie [0] (
04:19:06 Quit defukt (Read error: 110 (Connection timed out))
04:19:23 Quit Lambduh (Remote closed the connection)
04:19:32 Quit csc` ("Powering Off")
04:22:38 Quit Rob2222 (Read error: 113 (No route to host))
04:41:34***Saving seen data "./dancer.seen"
04:41:42 Quit miepchen^schlaf (Read error: 110 (Connection timed out))
04:41:59 Join miepchen^schlaf [0] (
04:57:04 Quit spr0k3t_ (Remote closed the connection)
05:06:23 Quit jhulst (Read error: 110 (Connection timed out))
05:12:02 Quit BHSPitMonkey_ ("Ex-Chat")
05:25:57 Quit SirFunk (Remote closed the connection)
05:35:41 Quit Horscht ("Snak 5.3.3 Unregistered copy. Evaluation period is over. Program will now quit. Thanks for using Snak.")
05:45:09 Quit jcollie (Read error: 113 (No route to host))
05:48:55 Quit TiMiD[FD] ("leaving")
05:51:22 Quit nplus (Remote closed the connection)
05:52:03 Join Lone_Wanderer [0] (
05:54:23 Join jhulst [0] (n=jhulst@unaffiliated/jhulst)
05:55:45 Join BHSPitMonkey_ [0] (n=stephen@
05:57:32 Join defukt [0] (n=WTFOMGBB@
06:06:25 Quit perrikwp (" ajax IRC Client")
06:10:50 Quit crazy_bus (Remote closed the connection)
06:11:39 Join [1]cool_walking_ [0] (
06:26:49 Join TiMiD[FD] [0] (n=TiMiD[FD@
06:27:01 Part Lone_Wanderer
06:27:14TiMiD[FD]I would like to commit
06:27:47TiMiD[FD]any objections ?
06:28:02 Quit keanu (Read error: 104 (Connection reset by peer))
06:29:41 Quit cool_walking_ (Read error: 113 (No route to host))
06:29:42 Nick [1]cool_walking_ is now known as cool_walking_ (
06:41:35***Saving seen data "./dancer.seen"
06:43:41 Join jcollie [0] (
06:45:06 Quit DaCapn (Remote closed the connection)
06:45:21 Join DaCapn [0] (
07:02:07 Quit DaCapn (Remote closed the connection)
07:10:27midkayBagder: wake up! (again!) i have something for you.
07:19:14 Join gtkspert [0] (
07:19:22 Join MethoS- [0] (
07:19:41 Join MethoS_mobile [0] (
07:23:30 Quit MethoS- (Remote closed the connection)
07:25:42 Join [1]cool_walking_ [0] (
07:25:42 Quit cool_walking_ (Read error: 104 (Connection reset by peer))
07:25:50 Nick [1]cool_walking_ is now known as cool_walking_ (
07:44:59 Quit MethoS_mobile (Read error: 110 (Connection timed out))
07:54:38 Join conando [0] (
07:55:23 Quit cool_walking_ (" HydraIRC -> it doesn't let you change this message easily.")
07:55:52 Nick JdGordon is now known as JdGordon|zzzz (n=jonno@rockbox/developer/JdGordon)
07:55:56 Nick JdGordon|zzzz is now known as JdGordon|zzz (
08:02:49 Join LinusN [0] (n=linus@rockbox/developer/LinusN)
08:12:06 Join mrkiko [0] (
08:12:19mrkikoHi all!
08:13:02mrkikoAs I asked yesterday - can someone tell me how to access the pitch screen without the remote controller on a iRiver H3x? Yeasterday I lost my Internet connection and probably the LinusN answer regarding this.
08:18:00Lloreanmrkiko: As a note, there are logs. If someone already answered it, and you know they have, you should just read the log.
08:23:22LinusNmrkiko: hold Navi to pop up the WPS context menu, then select "Pitch" from the menu
08:25:59 Quit BigBambi (Remote closed the connection)
08:30:40 Join Rob2222 [0] (
08:33:04*amiconn partially disagrees with soap ...
08:35:03LloreanLinusN: Re the sleep timer thing, I absolutely don't think it should always be persistent. It'd be frustrating for my device to turn off 15 minutes into a 4 hour car trip, and I have to either pull over or fiddle with the menus while driving to unset the sleep timer.
08:35:25LloreanBut the solution I offered was more as what I thought might be easier for him to accomplish, than for what I thought would be a suitable SVN solution to the problem
08:36:03LinusNLlorean: come on, do you really think i want the sleep timer to start *counting*?
08:36:32LloreanWell if you have to restart it counting manually, why does it need to be persistent?
08:37:16LloreanIt seems there's not a real benefit to saving it if it doesn't start counting, at least to me.
08:37:17LinusNbecause it can be handy to have a default value when you set it
08:38:27LloreanIt might save a few seconds, but if there was a way to start it counting, I think that'd be more useful. And I *think* that's what he wanted, since he said 90% of the time he needs the timer set
08:38:29LinusNthe benefit is to make it easier to start it, since you won't have to edit the counter starting from 0
08:38:57LinusNand you often use the same value, like 1 or 2 hours
08:39:17LinusNso you only have to press Select to set it
08:39:30LinusNi'd say it's much more convenient than today
08:39:37LloreanWell, I can certainly agree that saving the position isn't bad. But there should be a way to have it start counting from every boot too, I think.
08:39:52LloreanAnd I *think* but am not sure that's what he was requesting.
08:40:52*amiconn thinks that the current behaviour is perfectly fine (not remembering the sleep timer at all)
08:41:32LinusNi think saving the last entered value would be very convenient for those who use the sleep timer often
08:41:36***Saving seen data "./dancer.seen"
08:41:40LinusNbut not start counting
08:41:59LloreanI don't see why you couldn't have a "start timer from boot: yes/no" option to go with it.
08:43:59LinusNi can go with that, but i'm not sure it would be equally convenient, since you may easily forget to turn it off when you don't want a countdown
08:44:14LinusNand then you will be annoyed and stop using that option
08:44:34amiconnWhat would be the actual benefit of those extra things? I'd expect that if someone uses the sleep timer, the set time will often differ from previous invocation
08:44:45LinusNamiconn: why would it?
08:44:48LloreanLinusN: For people who mainly use the sleep timer.
08:44:52GodEateron the contrary, I think it'd be always the same
08:45:09LinusNamiconn: the sleep timer is a delay, not an absolute time
08:45:15LloreanThe last dozen or so times i've used my gigabeat, it's on the dock and I've been in bed. having it default to a timer of a couple hours would've been lovely. Turn it on, auto-resume, and not worry about it.
08:45:15amiconnI know
08:46:05amiconnThe only reason for me to (rarely) use the sleep timer is when I want the thing to switch off after a certain, well defined time. And that time will be different each time I use that feature
08:46:22GodEaterI think a lot of people actually use it when going to sleep
08:46:25LloreanSee, for me, it's "I don't want the noise to last past my first REM cycle"
08:46:37amiconnI never actually used it as a _sleep_ timer though; for that case I just used end-of-album + idle poweroff
08:46:59 Quit andrew_ (Read error: 110 (Connection timed out))
08:47:45LinusNamiconn: when you use it as a sleep timer, i believe you are likely to use the same value every time
08:48:14*GodEater agrees
08:48:20*Llorean thirds the motion.
08:48:43 Quit Rob222241 (Read error: 110 (Connection timed out))
08:49:37LinusNstill, what would be the problem with changing from the last set value instead of from 0
08:50:23GodEaterthat's the other point I was going to make too ;)
08:57:05amiconnOptionitis (especially with Llorean's idea)
08:59:12 Quit jhMikeS (Nick collision from services.)
08:59:17 Join jhMikeS [50] (n=jethead7@rockbox/developer/jhMikeS)
09:05:38 Join petur [50] (n=petur@rockbox/developer/petur)
09:07:09LinusNamiconn: a can agree with optionitis regarding the "start timer on boot" option
09:07:33LinusNbut i think remembering a value is merely a convenience thing
09:12:59GodEaterit doesn't change the number of items in the menu
09:13:30 Quit |sanchez| (Read error: 104 (Connection reset by peer))
09:16:07 Quit jhulst ("Konversation terminated!")
09:17:23 Join CaptainSquid [0] (
09:23:57mrkikoLlorean: regarding the "sleep timer", it would be relatively siple to add a setting determining if it should be saved or not
09:24:41LinusNnow *that* would be optionitis :-)
09:24:44mrkikoLlorean: so that if we save it, we start counting; else we not.
09:25:12LinusNi think just remembering the last value is enough
09:26:53mrkikoLinusN: ... it's a copromise solution...
09:27:16GodEaterno, just remembering the last value is the compromise
09:28:35 Quit Llorean (Read error: 110 (Connection timed out))
09:29:08LinusNi think remembering the last value is the most sensible thing to do
09:29:32GodEaterit adds no more options, and is a pretty intuitive thing to do I think
09:29:36LinusNit makes it much easier to start the sleep timer for those who actually use it as a sleep timer
09:30:21 Join aliask [0] (n=chatzill@rockbox/developer/aliask)
09:33:17 Join stripwax [0] (
09:36:13 Quit stripwax (Client Quit)
09:40:57 Join Nico_P [50] (n=nicolas@rockbox/developer/NicoP)
09:40:58 Join MethoS_mobile [0] (
09:43:51 Join homielowe [0] (
09:44:38 Quit MethoS_mobile (Remote closed the connection)
09:44:55mrkikoI think you're right. Aniway I think...
09:45:23mrkikoI think that a rockbox newbie will not expect this. This may not be a problem, but I don't understand how it could be considered intuitive.
09:45:53mrkikoWhy for example, should the sleep timer behave this way and the idle power off timer not? Knowing the rockbox logic a little I understand this, but I don't think it to be intuivie...
09:47:38GodEaterI do
09:49:31mrkikoI agree aniway
09:54:13 Quit z35 (Read error: 113 (No route to host))
09:54:49 Quit TiMiD[FD] ("leaving")
09:56:01Bagdera good read on foss copyrights, licenses, organizations etc
09:57:32Bagdersomewhat US-centric (no surprisingly)
10:08:44 Quit Nico_P (Remote closed the connection)
10:09:17 Quit jhMikeS (Nick collision from services.)
10:09:23 Join jhMikeS [50] (n=jethead7@rockbox/developer/jhMikeS)
10:41:38***Saving seen data "./dancer.seen"
10:42:21 Join hamdiya [0] (n=c3e5ecd6@gateway/web/cgi-irc/
10:43:04 Quit mrkiko (Remote closed the connection)
10:44:41 Join Mathiasdm [0] (
10:45:16 Quit hamdiya (Client Quit)
10:45:46 Join hamdiya [0] (n=c3e5ecd6@gateway/web/cgi-irc/
10:50:46 Quit hamdiya ("CGI:IRC (Ping timeout)")
11:04:18 Quit jhMikeS (Nick collision from services.)
11:04:24 Join jhMikeS [50] (n=jethead7@rockbox/developer/jhMikeS)
11:11:55 Join pixelma [50] (i=pixelma@rockbox/staff/pixelma)
11:19:58GodEaterhas anyone heard from linuxstb this week ?
11:23:46peturlinuxstb (n=dave@rockbox/developer/linuxstb) signed off 12 hours and 45 minutes ago (Read error: 104 (Connection reset by peer))
11:23:46 Join Arathis [0] (
11:24:35GodEateryeah, but I don't think he actually uttered a word yesterday :)
11:35:05 Join Thundercloud [0] (
11:38:54 Join stewball [0] (n=WTFOMGBB@
11:52:26 Quit defukt (Read error: 110 (Connection timed out))
11:52:49Soap_"* amiconn partially disagrees with soap ..." ?
11:54:12Soap_can I assume it is my comments in the user mailing list regarding the H10?
11:54:28GodEaterI wondered what he meant by that too
12:03:45amiconnsoap: yes
12:05:25Soap_Just trying to warn him "I knew H320, I served with H320, and you H10 are no H320"
12:08:28Soap_Though I forgot to comment on the H10's FM.
12:08:35*petur mostly agrees with Soap_
12:09:17peturthe H10 buttonlayout (and sliderpad) suck big time
12:21:55 Join homielowe_ [0] (
12:26:02 Quit homielowe_ (Client Quit)
12:26:45 Join homielowe_ [0] (
12:27:27 Quit homielowe_ (Client Quit)
12:29:02 Quit homielowe (Read error: 104 (Connection reset by peer))
12:41:42***Saving seen data "./dancer.seen"
12:50:37 Quit Arathis ("Bye, bye")
13:03:34markunpetur: how does the slider work currently?
13:04:08markunup, down and center?
13:06:47 Join mf0102 [0] (n=michi@
13:16:56 Quit jhMikeS (Nick collision from services.)
13:17:02 Join jhMikeS [50] (n=jethead7@rockbox/developer/jhMikeS)
13:17:48peturup/down only
13:18:44peturI have some real slider code but I don't like the performance
13:18:56peturthe touchpad really sucks
13:20:48markunpetur: do you have any ideas on how to use the slider? I still have my code to get x and y positions from the Gigabeat cross.
13:21:24peturyou mean how I created a slider interface?
13:22:20peturbasicly you track movement, but because of the nature of the touchpad, it needs filtering and position checks
13:22:42peturI think I posted some code in FS - let me get it
13:23:44petur8409 - my first attempt is at the bottom but it is not really ok
13:25:08markunWhat I meant was: what do you do with filtered position info? Translate it back to actions?
13:25:49peturyes... for me it was just scroll up/down
13:27:41 Nick fxb__ is now known as fxb (
13:43:23***SPY: Authentication failed for jhMikeS
13:44:01 Join DerDome [0] (
13:44:23markuncan I tell objdump that the start position of a file is for example 0x8000000 ?
13:46:40 Quit jcollie ("Ex-Chat")
13:48:50markunnever mind, I found it
13:49:22 Quit Mathiasdm ("Yuuw!")
13:59:25 Join Rondom [0] (
14:11:04 Join nplus [0] (n=nplus@141.25.Globcom.Net)
14:16:17 Join gevaerts [0] (
14:16:46 Join Nico_P [50] (n=nicolas@rockbox/developer/NicoP)
14:18:33*gevaerts ended up coding some more last night, and testing during today's lunch break
14:19:22peturfreak :p
14:19:42gevaertsI uploaded a new patch, performance numbers at
14:20:38gevaertsI now get 970 kB/s writes and 950 kB/s reads. I think it will be nearly impossible to do better on full speed
14:21:14gevaertsAt the same time, the transfer buffer got smaller
14:22:09markungevaerts: how much did you get with the OF again?
14:22:41gevaertsmarkun: 4.5 MB/s, but they cheat by using high speed.
14:23:16markunis it difficult to add high speed support?
14:23:17GodEaterheh - I like that that's considered cheating ;)
14:23:27GodEatermarkun: yes - it doesnt' work !
14:23:39 Join BitTorment__ [0] (
14:23:46*GodEater wonders if markun has been asleep the last couple of weeks
14:24:11GodEaterlucky you
14:27:28gevaertsI might steal some performance back to go to 2x16k buffers (current patch uses 2x32k). That costs about 2% but allows a lot more simplicity in the USB driver, which might improve stability.
14:28:16markunGodEater: I'll read the patch tracker comments
14:29:59 Join EspeonEefi [0] (i=espeonee@STRATTON-FOUR-O-EIGHT.MIT.EDU)
14:32:33gevaertsmarkun: high speed support will most probably require reverse engineering the OF. As far as I understand the hardware it should already work, but it still doesn't for unknown reasons.
14:37:00*amiconn wonders whether LinusN might have found something USB related in his appe OF RE endeavour
14:37:32LinusNi have only disassembled the bootloader so far
14:37:42LinusNwhich uses full speed iirc
14:38:30amiconnIt uses high speed, even though it's fairly slow on G5
14:38:33GodEaterI didn't know the bootloader had any usb support at all
14:38:44amiconnBut ~2MB/s can't be fullspeed
14:38:47 Quit BitTorment_ (Read error: 110 (Connection timed out))
14:38:48LinusNthen i should perhaps work a little more on that
14:38:52GodEateroh - we're talking disk mode ?
14:39:24amiconnOn Mini G2, diskmode is as fast as OF - maybe that would be a better choice for disassembly
14:39:48LinusNnah, all i want to know is how to do high speed
14:40:12LinusNdoes the mini have the same pp chip?
14:41:22gevaertsI think the missing bits might well be some DMA setup issue outside of the USB controller address range.
14:41:34gevaertsSo it might be hard to find
14:41:43***Saving seen data "./dancer.seen"
14:42:50amiconnThe mini ROM code seems to be designed to be run both on Mini G1 and G2, i.e. PP5020 and PP5022
14:43:29LinusNamiconn: care to send me a bin?
14:45:45jhMikeSgevaerts: that usb_core_init call really has to get out of usb_detect and be done on-thread only
14:46:34 Join nicktastique [0] (n=nick@unaffiliated/nicktastic)
14:46:58 Quit aliask ("ChatZilla 0.9.81 [Firefox]")
14:47:04 Join LambdaCalculus37 [0] (n=44a0430f@gateway/web/cgi-irc/
14:47:42 Join kugel [0] (i=kugel@unaffiliated/kugel)
14:48:25LambdaCalculus37Morning, all!
14:48:59GodEatergood afternoon
14:49:32jhMikeSgood aeon
14:52:05gevaertsjhMikeS: I don't disagree. I also would like to get the USB connect screen triggering out of there, and only show it when really needed (i.e. when the host is actively accessing the disk). I guess we mainly want the thread creation gone. The rest doesn't do much (just some variable setting).
14:53:27jhMikeStrue. just using the existing usb thread itself would be better because I can name at least one plugin where connecting usb will fail to create the thread anyway (uses all the slots up)
14:56:14amiconnThe USB screen should be shown during the whole duration of the connection
14:56:30GodEaternot if you're only charging
14:57:00amiconnYes, but that's something different, as it's not an MSD connection
14:58:07gevaertsamiconn: the use case I have in mind is that you plug the player in, copy some files, choose "eject" or "safe remove hardware", after which Windows (at least) doesn't access the device at all any more. I can imagine someone wanting to play some music then while the battery is still charging.
15:00:16gevaertsamiconn: I have to test this some more, but I think we get a nice SCSI command to tell us that sort of stuff (either vSTART_STOP_UNIT or SCSI_ALLOW_MEDIUM_REMOVAL)
15:03:23preglowgevaerts: that'll work as you intend also on linux, i think
15:05:33gevaertspreglow: it should. We just have to test on some more exotic/outdated systems as well (MacOS 9, Windows98,...)
15:06:20jhMikeShow about having usb_enable just take a parameter that says "for_what_reason" USB_POWERED, etc. etc.
15:08:32amiconnusb_enable is actually ums_enable as I see it
15:08:38 Join Ramla_ [0] (
15:08:47 Quit Ramla (Read error: 104 (Connection reset by peer))
15:09:56gevaertsCurrently it's mainly that, yes. If I get serial working at the same time as ums it will be more.
15:10:54gevaertsI think we have to differentiate between "plugged in", "usb protocol running", and "mass storage active". Only the last one requires disk lock-out.
15:10:58*jhMikeS just needs the queasy feeling to go away of seeing certain kernel calls done in interrupt context that will trash it :)
15:11:26*gevaerts only wants queue_post in interrupt context
15:13:28jhMikeSthat's about the only one intended for it (+ a few misc other queue functions)
15:15:03*gevaerts points out that he only inherited that part of the code, and that the current code already runs a lot less in interrupt context than it used to.
15:16:09gevaertsI guess most of this code was written to quickly get things working, and when that didn't happen it wasn't cleaned up.
15:16:50gevaertsNow that it actually works, it's worth spending time to make it work cleanly
15:22:10jhMikeSI'm guessing a number of FS bugs have the context issue at the root (unit responsive after plugging, etc.)
15:24:23*gevaerts volunteers to try and clean this up
15:25:13 Join udoprog [0] (
15:25:53udoproganyone knows when a stable release for rockbox will be available for ipod nano gen 1?
15:26:11udoprogthe current build screwed with my ipod, so i'd wather wait : P
15:26:45preglowno idea at all
15:26:51preglowwe haven't even planned a release
15:27:03preglowand what do you mean by "screwed" with your ipod?
15:27:46*LinusN has an interrupt driven serial port driver on his ipod video
15:27:54udoprogi got strange behaviour on boot, hanging and such, also weird when adding audio-files, like detexting the same file 4 times
15:29:30udoprogi might try with a later build if i get in the mood/get some spare time
15:29:37*gevaerts likes interrupt driven drivers as long as the "interrupt" part doesn't do too much
15:33:54 Join MethoS- [0] (
15:35:14*amiconn thinks that a driver can't be driven, it should drive itself ;)
15:36:21*jhMikeS gets the "four files" thing on sansa but thought it was from a phantom folder created by trying to rename the SD card
15:37:04preglowLinusN: \o/
15:37:15preglowLinusN: what was wrong with the interrupt setup?
15:37:35LinusNmy own incompetence :-)
15:38:55preglowoh, the old forehead slapping routine
15:39:07preglowforehead: no offence
15:39:46jhMikeS forehead to sore head?
15:39:55LinusNsomething like that
15:41:29 Join postglow [0] (n=54bd792b@gateway/web/cgi-irc/
15:46:04*preglow greets the enigma
15:48:09 Join SSnake [0] (
15:49:17nplusIf I have understood well, the database browser uses ID3, but the playlist viewer uses filenames - am I right?
15:50:20 Quit postglow ("CGI:IRC (EOF)")
15:50:49preglownot much of a talker, is he
15:51:22preglownplus: the database is the only thing that deals with id3 as a means of selecting music
15:53:23nplusso, preglow, how can I make playlist_viewer and file_browser show cyrillic filenames?
15:54:03preglowthey should do so by default if you've got the right font, afaik,
15:55:54nplusyes, but they don't
15:56:07nplusI'll try taking a pic
15:56:12preglowwell, do you have the right font?
15:57:09 Quit Rondom (Remote closed the connection)
15:59:18 Join Horscht [0] (n=Horscht@xbmc/user/horscht)
16:00:04 Join Falafel [0] (
16:01:20nplushmm, it's hard taking a picture of sansa's display, but I have tried unifont and 6+12x13 - they seem to show proper tags in the database
16:02:40nplusthe same is if I use helvR12
16:03:16preglowwell, weird
16:03:26preglowi'm not an expert on this at all, though, so i don't know
16:19:51markunnplus: do you use linux?
16:20:23markunmy guess is that your player is not mounted with the correct codepage then
16:20:44nplusactually it isn't a bug
16:21:15markunif you use a UTF-8 locale and you mount the drive as latin1 it will still look correctly in linux after you transfer them, but not in rockbox or on other computers.
16:21:20nplusi tried a build without patches and there wasn't any problem
16:21:48 Join barrywardell [0] (
16:21:51markundo you know which patch causes it?
16:22:05nplusno, I'll try to figure it out
16:22:18markunok, good luck
16:22:23markunstrange problem
16:26:22markunnplus: which features do you miss in the official build?
16:28:25markunLinusN: do you now going to implement the ipod accessory protocol?
16:28:38markunare you...
16:28:49nplusviewports, bmp resize, backlight behavior for sansa, casegame, video in main menu, shutdown in menu...
16:29:03nplusmainly those
16:29:22LinusNmarkun: i have it implemented
16:29:44LinusNmarkun: well, only the mode 2 commands (the simple remote buttons)
16:36:31 Join scorche|w [0] (n=42c007b2@rockbox/administrator/scorche)
16:38:16amiconnLinusN: I also have a simple remote for my mini to test...
16:39:18LinusNthat's good
16:39:21amiconn...and as ipl has a serial driver for pp5002, adapting your work for pp5002 shouldn't be too difficult either
16:39:35*amiconn should also have a 2nd Gen remote somewhere
16:39:57LinusNi wouldn't be surprised if my code worked on pp5002 unaltered
16:40:54 Quit kugel (Read error: 104 (Connection reset by peer))
16:41:20amiconnI wonder why the ipl guys didn't manage to get serial working on pp502x
16:41:43 Join z35 [0] (
16:41:45***Saving seen data "./dancer.seen"
16:42:22 Join toffe82 [0] (
16:42:29amiconnI don't think the actual remote code will work unaltered though. Iirc the protocol is slightly different between the early remotes and the later ones
16:42:44 Join JamPS [0] (
16:42:59 Quit DerDome (Remote closed the connection)
16:43:06JamPSnice, the sansa usb support is broken again :P
16:43:29 Join DerDome [0] (
16:43:50 Join dandin1 [0] (
16:44:40JamPSNow it goes black and jams
16:44:59 Quit DerDome (Remote closed the connection)
16:45:32 Join DerDome [0] (
16:45:50GodEaterit's not actually changed in the builds distributed on the website...
16:46:10JamPSwhat do you mean?
16:46:26GodEaterI mean the USB code has not changed for the sansa since before xmas
16:46:30JamPSI now have build r16354
16:46:41JamPSyou wish
16:47:00JamPSsomething has changed again
16:47:17LinusNGodEater: barry wardell made a usb related commit yesterday
16:47:26GodEaterah did he ?
16:47:29GodEatermust have missed that
16:47:37GodEaterI thought gevaerts was the only one working on it
16:47:50*GodEater apologises for being misinformed
16:48:07LinusNr16354 to be exact
16:48:20JamPSbah I should probably just stick with a firmware that works
16:48:26LinusN"FS #8588: disable USB device again when USB is disconnected on PortalPlayer targets. Previously, the USB device was left on after USB disconnection causing extra battery drain. This alleviates this problem somewhat, although it seems like battery drain may still be higher after a USB disconnection, so we're still missing something."
16:48:56scorche|wJamPS: rockbox isnt released for those devices are a tester of a firmware that is in development...
16:49:02JamPSMy sansa just goes black when I plug in the wire
16:49:17JamPSI have to reset
16:49:33JamPSBut why those commits affect my sansa?
16:49:50LinusNbecause sansa has a portalplayer cpu
16:50:12JamPSah ok
16:50:53 Quit Siku (Nick collision from services.)
16:50:59 Join Siku [0] (
16:51:03LinusNi suggest we revert that commit if it causes this kind of problem
16:51:04JamPSBut usb has never worked with rockbox, it just rebooted to original firmware.
16:51:16pixelmathe automatic reboot from a running Rockbox into the OF doesn't work in every build it's been like that for a while - and it changes even when completely unrelated code changes
16:51:27JamPSnice :P
16:51:46LinusNpixelma: aha
16:52:06GodEaterJamPS: correct, but gevaerts is most of the way to fixing "usb not working in rockbox"
16:52:12pixelmabut it's reliable if you plug USB when the player is turned off
16:52:19JamPSyea I know
16:52:44scorche|wthen do that?
16:53:16JamPSoh btw windows says that this device can perform faster now when it freezes :P
16:55:56LinusNwow, my ipod froze, and menu+select has no effect...
16:56:14Horschthold it long enough
16:56:17Horschthad it once, too
16:56:18GodEaterLinusN: hold switch to on, then off first ;)
16:56:30GodEateras per apple's instructions...
16:57:37LinusNno go
17:00:10 Quit toffe82 (Read error: 104 (Connection reset by peer))
17:00:12 Join toffe82_ [0] (
17:00:21 Nick toffe82_ is now known as toffe82 (
17:00:38LinusNand i thought that the menu+reset was somehow hardwired
17:01:07JamPSdid you press it like 20 seconds?
17:01:09 Quit freqmod__nx (Read error: 104 (Connection reset by peer))
17:01:28JamPSmight work...
17:02:24LinusNi tried over a minute
17:02:34preglowLinusN: you pressed them DEAD simultaneous, yes?
17:02:44preglowi've yet to see an ipod that can be reset that way, and i don't think i'm alone
17:02:51preglowcan/can't :/
17:03:09LinusNreally odd, it has always worked before
17:04:52JamPSI opened my sansa about 5 times before I knew that it could be reset by pressing power 15secs :P
17:07:17dionoeaLinusN: maybe your batteries just ran out
17:07:17 Quit toffe82 (Read error: 104 (Connection reset by peer))
17:07:57LinusNdionoea: no, it's stuck on the rockbox splash screen
17:08:13 Quit CaptainSquid (Remote closed the connection)
17:08:51GodEaterhold to on, then to off, then menu+select has always worked for me too :(
17:08:58Horschtfor me, as well
17:09:10 Join toffe82_ [0] (
17:09:56LambdaCalculus37Me as well.
17:09:59GodEaterI've only seen occasions where menu+select doesn't work if I've not done the hold trick first
17:11:47LinusNremoving the battery helped
17:12:50LinusNreally strange
17:14:33LambdaCalculus37Indeed. I never saw that kind of behavior on my iPod before.
17:14:40LinusNme neither
17:14:41 Join Llorean [0] (
17:14:49 Quit Llorean (Client Quit)
17:14:50 Join freqmod_nx [0] (
17:15:12LambdaCalculus37Usually for me, if the hold trick doesn't work the first time, a second time will get it to work.
17:18:24 Join desowin [0] (n=desowin@atheme/developer/desowin)
17:21:23LinusNthe recipe was simple: set up an irq and don't handle it
17:24:20 Join jgarvey [0] (
17:24:24barrywardellLinusN: I don't think my commit should have any negative effect
17:25:02LinusNbarrywardell: it could just be the odd code alignment issue pixelma mentioned
17:26:57barrywardellIf it is causing problems, I'm fine with it being reverted, but I didn't notice and new problems with it
17:27:19 Quit toffe82_ (Read error: 110 (Connection timed out))
17:32:53JamPSgood luck to you guys making rockbox better :)
17:37:00LinusNthanks, we need it :-)
17:41:36 Join Domonoky [0] (n=Domonoky@
17:43:06 Join Arathis [0] (
17:46:18markunDoes anyone know what this "V24000" chip could be?
17:46:46markunI've read claims that it's an FM radio receiver chip by Sanyo
17:46:58LambdaCalculus37markun: I'll Google it for you.
17:47:29 Join DaCapn [0] (
17:48:10LambdaCalculus37markun: Confirmed. Here's the info:
17:48:13LambdaCalculus37Check the first post.
17:53:08 Quit XavierGr (Nick collision from services.)
17:53:16markunLambdaCalculus37: yes, but who wrote that. I would prefer to see it on a sanyo site.
17:53:19 Join XavierGr [0] (n=xavier@rockbox/staff/XavierGr)
17:53:42LambdaCalculus37Back to Google I go.
17:53:45LambdaCalculus37markun: Good point. :)
17:54:32 Join tvelocity_ [0] (
17:55:43markuntoo bad that I couldn't find a datasheet for the CPU or this FM radio chip
17:56:39LambdaCalculus37markun: I'm coming up empty handed.
17:56:52LambdaCalculus37No listing on Sanyo's site that I could find.
17:57:22markunme neither
17:58:03 Join Axio [0] (
17:59:17 Nick JdGordon|zzz is now known as JdGordon (n=jonno@rockbox/developer/JdGordon)
17:59:49 Join agm3nt [0] (
18:02:59 Quit gevaerts ("work->home")
18:03:17 Join ender` [0] (
18:09:21 Join n1s [0] (
18:11:17 Join herrwaldo [0] (
18:13:12 Quit tvelocity (Read error: 110 (Connection timed out))
18:16:52 Quit SSnake (Read error: 110 (Connection timed out))
18:19:58 Join hannesd [0] (
18:22:10 Quit petur ("work->home")
18:23:05 Join BigBambi [0] (n=Alex@rockbox/staff/BigBambi)
18:25:20 Join japc [0] (
18:25:33jhMikeSmarkun: I think the docs used for the sansa FM chip will do
18:26:11 Quit LambdaCalculus37 ("CGI:IRC (EOF)")
18:27:16jhMikeSthe one's used were for LV24000,1,2. the sansa FM chip is also simple marked "V24020"
18:29:19 Quit japc (Remote closed the connection)
18:29:53 Join gevaerts [0] (
18:35:24 Quit XavierGr (Nick collision from services.)
18:35:35 Join XavierGr [0] (n=xavier@rockbox/staff/XavierGr)
18:38:00 Join japc [0] (
18:41:47***Saving seen data "./dancer.seen"
18:42:07 Quit ch4os_ ("Lost terminal")
18:44:02 Quit japc (Remote closed the connection)
18:46:28 Join ch4os_ [0] (n=ch4os@unaffiliated/ch4os/x-059673)
18:47:36gevaertsI just changed the usb driver to only use one transfer descriptor, and not support chaining. This limits transfer sizes to 16k (actually 20k if we really want). It seems to be much more stable. I have now been running badblocks -w with a lsusb -v loop for 5 minutes without any problem. The svn version stops bulk transfers after a few seconds of this, and usully recovers after a bus reset. Recent patches usually don't recover. This one do
18:47:50gevaerts...doesn't even stop.
18:47:57*gevaerts should really type shorter lines
18:48:56 Join perrikwp [0] (i=98216e29@gateway/web/ajax/
18:49:09LinusNgevaerts: do you think it's time for a new commit?
18:49:31gevaertsThat means the version I am running now is probably the most stable one yet. The "cost" is bein limited to 16k transfers, which means a 2% slowdown.
18:49:55gevaertsLinusN: I think so. I
18:50:59rasherStill full speed?
18:51:33gevaertsI'll upload a new patch sometime tonight. After this I'm going to tackle jhMikeS' remarks about threads and interrupt context.
18:52:03gevaertsrasher: yes, but at least twice as fast as two days ago.
18:53:24gevaertsI think we should still leave it disabled for now, and once the threading and interrupt cleanups are done it should be ready for general consumption
18:54:42 Quit Horscht ("I am root. If you see me laughing, you better have a backup")
18:55:40gevaertsAs far as I can see it doesn't even slow down significantly under this lsusb load. I stil get over 900 kB/s write speed
18:55:55LinusNdamn nice
18:58:26 Join JdGordon|w [0] (i=836b0049@gateway/web/ajax/
18:58:44 Quit scorche|w ("CGI:IRC (EOF)")
18:59:53 Join petur [50] (n=petur@rockbox/developer/petur)
19:02:21*gevaerts is now going to try if all this solved the serial/storage coexistence issue by accident
19:02:51 Join Horscht [0] (n=Horscht@xbmc/user/horscht)
19:02:51*LinusN would LOVE usb-serial logf()
19:07:16 Join Phill [0] (
19:07:21 Quit barrywardell (Remote closed the connection)
19:07:40 Quit tedrock (Read error: 104 (Connection reset by peer))
19:09:19PhillLinusN: I have usb-serial logf working-ish (although obviously not in combination with mass storage). It works okay as long as there are not too many logfs in quick succession.
19:09:30gevaertsIt works !
19:10:24LinusNgevaerts: wohooo!
19:11:08*gevaerts now has cu session open that echoes everything back with case changed, while copying files
19:11:22LinusNhow cute
19:12:19rashergevaerts: Brilliant
19:12:27 Quit dandin1 ()
19:12:28*preglow has been waiting for this for a long time
19:12:59gevaertsPhill: can you post a patch somewhere ? It might be useful to integrate it in the "big" usb patch.
19:13:01rasherNow to hpe that the sansa dock talks over serial
19:13:21rasheroh wait.. that's.. different serial
19:13:28rashersilly me
19:13:49LinusNrasher: let's hope that my pp502x serial code works for the sansa too
19:14:01LinusNtime to get me a sansa :-)
19:14:34LinusNwhat's with the sansa dock btw?
19:14:39*rasher has access to a sansa dock
19:14:44LinusNdoes it have a remote or what?
19:14:54rasherAnd sound output
19:14:59 Join pondlife [50] (n=Steve@rockbox/developer/pondlife)
19:15:01 Join LambdaCalculus37 [0] (n=44a0430f@gateway/web/cgi-irc/
19:15:42Phillgevaerts: I'll try to find some time later this evening to boot into linux and get the patch. Hopefully I can tidy it up a bit first too.
19:16:25gevaertsPhill: OK. I'll upload my new patch in a few minutes. We can always merge later
19:16:27rasherLinusN: costs about as much as the e200 itself, too
19:16:32LinusNrasher: wow
19:16:44rasherDepending on where you buy them, of course
19:16:48LinusNgevaerts: still don't want svn access?
19:17:12 Join ompaul [0] (n=ompaul@gnewsense/friend/ompaul)
19:18:47gevaertsLinusN: Maybe it's time to start considering it.
19:19:07preglowi'd say
19:20:13*pondlife pops in to congratulate LinusN and gevaerts
19:20:25 Part pondlife
19:21:58 Join japc [0] (
19:23:26 Quit EspeonEefi ("さよなら")
19:24:40 Quit japc (Client Quit)
19:24:53 Part LinusN
19:24:57gevaertsPatch is uploaded. Serial is enabled, so Windows might ask for a driver.
19:24:59 Join japc [0] (
19:25:01*gevaerts is going to eat something now
19:25:51BigBambiCongrats :)
19:27:45 Quit perrikwp (" ajax IRC Client")
19:28:53 Quit japc (Remote closed the connection)
19:31:29 Join MethoS-- [0] (
19:41:05amiconngevaerts: With ums_v22, I get an immediate data abort on mini G2 when plugging USB
19:43:39amiconnThe crash happens in set_serial_descriptor
19:45:42 Join beheadingpain [0] (n=not@
19:46:01 Join remotechief_311 [0] (
19:46:46beheadingpainhi i've been eating kfc chicken wings 3 weeks straight is that bad for my body? how long can i continue?
19:47:21LambdaCalculus37Uh, what!?
19:47:49 Join gregzx [0] (
19:47:58krazykitbeheadingpain, this is an on-topic channel. please take non-rockbox chatter elsewhere
19:48:01LambdaCalculus37What does KFC have to do with Rockbox?
19:48:07moosspam !
19:48:17*LambdaCalculus37 pulls the Anti-Spam Cord
19:48:19 Join japc [0] (
19:48:45 Quit MethoS- (Read error: 110 (Connection timed out))
19:50:28 Quit remotechief_311 (Remote closed the connection)
19:55:31 Quit DaCapn (Read error: 104 (Connection reset by peer))
19:56:04amiconngevaerts: Btw, those hex[] arrays are actually const, so they should be declared const. That'll save some init code gcc otherwise adds
19:56:55amiconnAnd I found the culprit - struct usb_string_iSerial is unaligned. Is there a reason why all those structs are declared packed?
19:58:30gevaertsamiconn: yes. They go over the wire as-is, so if they are not packed it won't work. Th other alternative would be to copy them all to a temporary buffer.
19:59:20preglowalmost preferable, with arm...
19:59:34gevaertsamiconn: usb_string_iSerial should probably also be cache aligned to work properly.
20:00:56beheadingpaini have high cholesterol how long till i'll die if i don't lower it and keep eating chicken wings and sit on my leg all day?
20:01:10gevaertspreglow: only seven of the fields actually ever need to be changed, and three of those are chars. Otherwise I would agree
20:01:23amiconnOkay, understood. However, declaring them packed also makes them unaligned (another gcc glitch? Afaiu they shouldn't be unaligned unless also explicitly declared unaligned)
20:01:28DBUGSent KICK beheadingpain to server
20:01:29Kick(#rockbox beheadingpain :Rotten persons outside, please) by logbot!n=bjst@gateway/web/cgi-irc/
20:01:51*petur is amazed by the default kick message :)
20:02:29amiconnWould it work to declare alignment in usb_ch9.h together with the 'packed' attribute?
20:02:51preglowamiconn: of the entire struct or members? packed structs should definitely break member alignment if necessary
20:02:53amiconnThen all incarnations would be sutably aligned in memory...
20:03:05amiconnThe entire struct is obviously unaligned
20:03:19amiconnHave a look at e.g. struct usb_string_descriptor
20:04:05amiconnIt consists of 2 8-bit-members, then an array of 16 bit elements. So the whole struct needs to be 16 bit aligned in memory in order to not crash arm
20:04:26 Part at0m|c ("Clear Channel Event")
20:04:32amiconnBut by default it isn't.
20:06:04preglowweird, i wouldn't expect packedness to affect alignment of the entire thing at all...
20:07:29gevaertsamiconn: struct usb_string_iSerial is the only struct usb_string_descriptor that needs to be modified, and it isn't part of a bigger struct. The config_data struct needs to be packed, but we could assemble it by hand in a buffer, or declare four nearly-identical copies of it.
20:08:12amiconnThis is just by chance
20:08:47amiconnAny in-ram struct usb_string_decsriptor has a 50% chance to be misaligned to access its 16 bit members
20:09:58gevaertsamiconn: true, but they are (or should be) const, so the gcc initialisation code needs to write them. After that the rockbox code treats them as an opaque buffer to send over the wire if the host asks for them
20:10:06amiconnImho we have 3 options. (1) Explicitly align every such struct (as you did for sansa, but not for ipod). (2) declare alignment along with packedness in usb_ch9.h
20:10:25preglowwhich sounds good
20:10:38preglowthe latter should be the best, if gcc has it work...
20:10:40amiconn(3) use 'unaligned' pointers to access the 16 bit members
20:10:58amiconnThis last option causes bigger and slower code though
20:11:23 Join spiorf [0] (n=spiorf@
20:11:46gevaertsThese members are only accessed very rarely, and only in very few places, so that shouldn't be a big problem.
20:12:38amiconnNo, but if proper alignment can be ensured (like in this case, it should be preferred, imho
20:13:29 Join bluebrother [0] (n=dom@rockbox/staff/bluebrother)
20:14:44*bluebrother notices the addition of serial port registers for pp5020
20:16:01amiconngevaerts: Seems to work nicely
20:16:59amiconnI just changed the attributes for struct usb_string_descriptor in usb_ch9.h to read:
20:17:01amiconn__attribute__ ((packed,aligned(2)));
20:17:39 Quit moos (Connection reset by peer)
20:17:50gevaertsI'll do the same change here.
20:17:51 Join moos [0] (
20:17:55 Join Rob222241 [0] (
20:18:26amiconnThen you can remove the explicit attribute for usb_string_iSerial on sansa in usb_core.c
20:19:40amiconnIs there a reason why this is declared different for sansa than for the rest?
20:20:09gevaertsamiconn: I probably forgot to change both after fixing it for sansa.
20:20:32amiconnI mean in general, not just the attribute
20:23:40gevaertsOh ok. The OF for sansa and ipod have different length serial number strings, and we started out trying to get exactly the same. I guess we could just take the largest one and leave the extra zeroes, since we decided to not follow the OF exactly here.
20:24:12amiconnHmm, ok
20:24:25amiconnI also notice that there is no serial number (yet) for H10
20:24:45amiconnIt probably needs figuring out where the OF gets this number from?
20:25:55*amiconn also wonders whether the code should be checked for endian issues
20:26:16amiconnRight now it's only used on little endian targets, but that might change
20:26:58gevaertsActually it doesn't really work properly on c200 either (i.e. I don't get the same as the OF). For some reason e200 and c200 don't get their serial from the same place
20:30:21gevaertsI think those same usb descriptors will be the hardest to get endian-clean. SCSI should be mostly there (SCSI is big endian...), except for one or two magic numbers.
20:30:41 Quit spiorf (Remote closed the connection)
20:31:19 Quit Phill (Read error: 110 (Connection timed out))
20:32:02gevaertsBy the way, does anyone have a script somewhere that builds all targets in one go ?
20:33:39bluebrothergevaerts: I did something very basic for building all manuals ... it's mainly a "for i in $foo; do make -C $i; done"
20:34:36*amiconn never built *all* targets so far in years of rockbox hacking
20:35:25 Quit Rob2222 (Read error: 113 (No route to host))
20:35:59*gevaerts might not build all targets, but checking if everything still compiles for should-be-unaffected targets sounds useful
20:36:15amiconnI rather select the target(s) that might be affected by a change, and if it's a generic change, a selection of as few targets as possible that covers the whole range of properties which might be affectede
20:37:31gevaertsThat's probably a good way to do it, once you know which those targets are
20:37:32 Join Munkie [0] (
20:37:36*gevaerts is still a newbie
20:37:55pixelmabluebrother: does that still work? ;)
20:38:25MunkieIs the sever up? Trying to install themes and its saying it cant find any?
20:38:27bluebrotherpixelma: just started testing ;-)
20:38:44bluebrotherand in fact I build the manual-pdf target. I think that should still work ...
20:38:53pixelmayou are aware that you need "make manual" now?
20:39:04bluebrotheryes, I am.
20:39:35bluebrotherI already thought about removing the manual configuration option and simply use the (N)ormal build instead
20:39:51scorche| did logbot know to kick him?...did someone do that manually?
20:40:27bluebrotherwhich reminds me, I wanted to fix something regarding output names for the manual. They're a bit inconsistent.
20:40:49 Quit ch4os_ (Read error: 104 (Connection reset by peer))
20:40:57peturscorche|sh: logbot can be manipulated ;)
20:41:01MunkieWhy would it be giving me and error? the fonts work and the games, but not the themes
20:41:25scorche|shpetur: of course, but by whom?
20:41:38*petur smiles
20:41:51***Saving seen data "./dancer.seen"
20:41:52pixelmascorche: check the logs (at least the current.txt has it) :)
20:42:24peturscorche|sh: it really is a nice way for devs to kick somebody when no ops are around
20:42:39pixelmaMunkie: do you mean installing themes with rbutil for an Ipod Video?
20:42:56*amiconn wonders what makes the H10 microdrive so crappy compared to the one used in the ipod mini
20:44:01peturonly the microdrive?
20:44:16***Server message 505: 'logbot :Private messages from unregistered users are currently blocked due to spam problems, but you can always message a staffer. Please register! ( )'
20:44:16Munkieyup i do
20:44:18amiconnIt's dead slow
20:44:44*LambdaCalculus37 wonders if a CF mod would do wonders for an H10
20:44:45Munkieive got dead slow internet, so it makes no difference to me
20:44:48amiconnRockbox unzipping times: Mini G2, diskmode: 16 sec. Mini G2, rockbox: 35 sec. H10, OF: 2:42. H10, rockbox: 308:
20:45:04*petur notices rasher and JdGordon are interested in the powers of logbot :)
20:45:05*scorche|sh shall have to investigate logbot a bit further
20:45:26bluebrotherMunkie: the themes are on a different server.
20:45:27amiconnSpinup time is also looong for a microdrive: almost 3 sec instead of ~500ms
20:45:42bluebrotherscorche|sh: which reminds me ... how's the official themes thingy going on?
20:45:59peturamiconn: is the pinout compatible with CF?
20:46:07*gevaerts likes the disk in the H10. It makes the rockbox usb stack look good
20:46:18JdGordon|wpetur: would be nice to know how to use him when I get back and everyone is out when they are needed...
20:46:20scorche|shbluebrother: it has been ready for ages...i am just waiting for viewports in WPS to be committed
20:46:50pixelmaMunkie: the Ipod Video themes were taken down from the server rbutil searches because of bandwidth issues. I'm not sure if they are available through a mirror (scorche|w?), if not then have a look at the WpsGallery in the Rockbox wiki.
20:46:54scorche|shJdGordon|w: well, let me find out more, maybe do a bit of tuning, and then we shall see
20:46:58bluebrotherpetur: the microdrive in the mini has the same pinning as a CF card, so I guess the H10 does too.
20:47:12scorche|shpixelma: soap has them mirrored
20:47:23LambdaCalculus37I guess the only way to truly find out is to perform the mod and see what happens.
20:47:23amiconnpetur: No. It has a so-called F-FLEX interface
20:47:38bluebrothermaybe we should remove the "install all" button in the themes window.
20:47:44gevaertsamiconn: I would say that the main problem probably is huge seek times
20:47:46pixelmascorche|w: do you have a link for Munkie then?
20:47:49bluebrotherI wasn't a fan of that button in the first place anyway.
20:48:04amiconnIt's a Seagate ST660211FX
20:48:55scorche|shpixelma: and scorche|w is gone now that i can ssh through work's vpn :)
20:49:41amiconnThe mini has a Hitachi HMS360604D5CF00
20:49:42amiconnThat one is CF type II
20:50:24LambdaCalculus37And the Seagate drive in the H10 isn't CF Type II, if I am to understand this correctly?
20:50:48bluebrotherso it's type I? Or something completely different?
20:51:01amiconnCompletely different
20:51:01pixelmascorche|sh: see, if you are here as scorche + scorche|* then you're making it hard for tab completion (and I didn't check back) :)
20:51:01*bluebrother thought all microdrives were CF II
20:51:33amiconnMicrodrive with CF connector are always CF II. The only difference between type I and II is thickness
20:51:38amiconn3.3mm vs. 5mm
20:51:45peturdesowin: is there a mac version of qtscrobbler? If not, I assume it is possible?
20:52:03desowinpetur: I heard of people having success compiling it on mac
20:52:10amiconnCF == compact flash, FX == F-FLEX
20:52:22bluebrotheryep, but up to my knowledge all microdrives were 5mm. Seems my knowledge was wrong.
20:52:23amiconnANd the one in the H10 is ..FX
20:52:35amiconnNo, your knowledge is correct
20:52:42desowinpetur: don't own nor have access to mac myself though
20:52:45amiconnBut not all microdrives have CF connectors
20:53:20*bluebrother learned something new today :)
20:53:33 Join qwedsa_ [0] (
20:54:21*LambdaCalculus37 has learned the Tech Paper "Microdrive Technical Differences"!
20:54:52*gevaerts remembers Sinclair microdrives
20:55:05 Join tessarakt [0] (
20:58:35MunkieAre there any rockbox programming tutorials anywhere?
20:59:04bluebrotherthere is quite some information in the wiki. But no tutorials.
20:59:49Munkieok cool cool. Would c# work?
20:59:58BigBambiNo, C
21:00:06BigBambiand bits of assembly
21:00:16LambdaCalculus37Plain, classic C.
21:00:23 Join bertrik [0] (
21:00:33BigBambiMunkie: You have to remember these are limited, embedded devices...
21:00:58Munkieok. i must learn that. Rockbox is so amazing. Is there any way to get sleep mode on the ipod video? that will make it perfect
21:01:08bluebrotherc#? /me coughs.
21:01:17Horschtsleep mode, Munkie?
21:01:37bluebrotherI guess he means the "off" mode that Apple uses.
21:02:15bluebrotherand hasn't amiconn figured out how to run the cpu at 11MHz?
21:02:22gevaertsbertrik: have a look at gevaerts/ums/">
21:02:35 Join Llorean [0] (
21:02:38Munkieyes, thats what i mean, so it can load up quick, instead of having to reboot everytime
21:02:48*petur read today that assembly is a no longer used skill since 2000... idiots...
21:03:03gevaertsNot entirely there yet, but close.
21:03:18amiconnbluebrother: huh?
21:03:49 Join indigo_ [0] (
21:03:53bluebrothererr ... yep, meant that.
21:04:04amiconnYes, 32kHz is the sleep (or as I call it, suspend) clock
21:04:13gevaertsbertrik: I added some #ifdefs to usb_storage.c to disable access to the internal flash, and only show the sd card. USBCV MSC tests are a lot safer then
21:04:26amiconnBut that's not the only necessary condition for suspend mode
21:05:09amiconnWe'd also need to find out how to shut down various other components (lcd, audio codec...), and put the ram into low-power self refresh mode
21:05:13Munkiewhat does it usually run at?
21:05:14indigo_hi, after creating a compiler to build rockbox via crossdev (gentoo), I get the following error message while executing "make"
21:05:22amiconnI know how to do this on a 1st and 2nd Gen *only*
21:05:31bertrikgevaerts: ok, very good idea
21:05:32indigo_/home/indigo/rockbox/firmware/target/arm/system-arm.h: In function `queue_enable_queue_send':
21:05:36indigo_/home/indigo/rockbox/firmware/target/arm/system-arm.h:119: warning: asm operand 3 probably doesn't match constraints
21:05:38indigo_/home/indigo/rockbox/firmware/target/arm/system-arm.h:119: warning: asm operand 3 probably doesn't match constraints
21:05:41indigo_/home/indigo/rockbox/firmware/target/arm/system-arm.h:119: error: impossible constraint in `asm'
21:05:44indigo_/home/indigo/rockbox/firmware/target/arm/system-arm.h:119: error: impossible constraint in `asm'
21:05:47indigo_make[1]: *** [/home/indigo/rockbox/build-dir/firmware/kernel.o] Error 1
21:06:00Horschti second amiconn's request
21:06:00amiconnindigo_: Using crossdev isn't recommended.
21:06:01scorche|shindigo_: please use a pastebin next time
21:06:16indigo_scorche|sh: ok, sorry
21:06:53bluebrotherindigo_: what version of arm-elf-gcc do you have?
21:07:03indigo_amiconn: but on it's recommended for gentoo users
21:07:23n1sindigo_: you should probably try using our script to build your crosscompiler
21:07:26bluebrotherthat's too old.
21:07:27indigo_as 4.* seems not to work properly atm
21:07:43n1s4.0.3 is recomended for arm
21:07:44bluebrotheras you also see on that page 4.0 is _required_ for arm targets.
21:07:47indigo_as sayd somewhere in the wiki
21:08:11Lloreanindigo_: 3.4.6 is for Coldfire targets, 4.0.3 is for ARM
21:09:02*amiconn seems to remember that people using crossdev had all sorts of problems
21:09:08amiconnAm I remembering wrong?
21:09:09scorche|shindigo_: where did you see 4.*?
21:09:32 Quit qwedsa (Read error: 110 (Connection timed out))
21:09:36rasheramiconn: I don't think so. Frankly I think that section should be removed
21:09:47indigo_scorche|sh: GCC 4.0.x seems to work fine when building Rockbox for Coldfire platforms. For the time being, it is not the recommended or preferred version to use.
21:10:00 Join defukt [0] (n=WTFOMGBB@
21:10:07n1skeyword there is _coldfire_
21:10:08rasherindigo_: Codeword being _Coldfire platforms_
21:10:09indigo_that's what the wiki says
21:10:19*gevaerts gives everyone a beer
21:10:30*n1s cheers :)
21:10:37peturnice, thanks!
21:10:48scorche|shgevaerts: you have more than a few beers that should be going to you atm :)
21:11:20gevaertsscorche|sh: I can't code after too many of them
21:11:44JdGordon|wsure you can!
21:11:58scorche|shgevaerts: stick with the ballmer peak
21:12:02*gevaerts likes code that still makes sense the next day
21:12:04bluebrothergevaerts: try virtual beer for now ;-)
21:12:31bluebrotherand real beer later ...
21:13:01Lloreanindigo_: The problem is, "Coldfire" is for the iRiver H100 and H300, and the iAudio X5 and M5. Not any other players. It's an entirely different compiler, which you seem to have confused.
21:14:13 Join Domonoky_ [0] (n=Domonoky@
21:15:27 Quit Munkie ()
21:16:29 Quit stewball (Read error: 110 (Connection timed out))
21:16:32indigo_ok, now I try using the crossdev-script in order to get the building process to work. Thanx for the bunch of information.
21:24:05 Join EspeonEefi [0] (i=espeonee@STRATTON-SEVEN-FOURTY-THREE.MIT.EDU)
21:24:42 Quit Jon-Kha (Remote closed the connection)
21:26:41 Join Jon-Kha [0] (
21:27:24Domonoky_bluebrother: did you notice that your rbutil linux build wasnt really static ? (libusb... )
21:29:37bluebrotherbut I guess libusb is common enough to assume it present.
21:29:51rasherbluebrother: except it might be 64-bit
21:30:01rasherAlso, any chance of getting an install target in the makefile? And an option to avoid the static linking
21:30:44bluebrotherrasher: dang, good point.
21:31:06rasheramiconn ran into that yesterday
21:31:10bluebrotherrbutil will only get linked statically if you use -config static when running qmake
21:31:44bluebrothermy development builds are usually dynamic builds.
21:32:03 Quit Domonoky (Read error: 110 (Connection timed out))
21:32:18bluebrotherI can look into adding an install target. Not sure what exactly is needed though (as the makefile is generated by qmake)
21:32:33*bluebrother puts a note on the todo-list
21:33:04rasherI guess "manually" copying isn't much of a problem (I was having a go at debian packaging last night)
21:33:36bluebrotheryep. Of course it would be nice to use the usual "make install"
21:33:49LloreanIs it usual that RButil downloads Rockbox much more slowly than downloading it yourself, or am I just unlucky?
21:35:15bluebrotherI haven't done speed tests, but for the older versions (when I had a good internet connection, not this stupid thing I use right now) I hadn't noticed a huge difference.
21:35:53LloreanI've been downloading for several minutes now, and I'm on a fairly fast connection here. I don't remember it taking this long in the past, but then, usually I just compile my own builds, so it's been a while
21:37:24preglowi don't notice much of a difference
21:43:35LloreanWhat needs to be true for rbutil to create a voice for me?
21:45:59LloreanVoicefile generation gave me a 5k file. =/
21:47:18 Join Phill [0] (
21:49:15bluebrotherLlorean: have you checked the tts settings? Especially the language when using sapi?
21:49:50LloreanI've set it to english
21:49:58LloreanAnd it says I have Microsoft Anna available in voices.
21:50:14bluebrotherhmm. Domonoky_ is the expert for voice.
21:50:25LloreanDo I need rbspeexenc.exe separately, or is the encoder built in to rbutil?
21:51:03Phillgevaerts: i've upload my logf patch to . I'm not sure it's too helpful, as it's probably done in a very horrible way, but it's a starting point! Unfortunately I haven't got time to work on it at the moment.
21:52:22 Join DaCapn [0] (
21:52:31LloreanAnd it's creating 5k vioce files. :(
21:52:34LloreanSo they're empty
21:52:53*preglow never tried with rbutil
21:52:59LloreanI've never tried with it before ether
21:53:06LloreanBut I want to show someone else how to do it on Windows.
21:53:14LloreanAre there known issues with Vista?
21:53:22preglowi would doubt it
21:53:27 Quit MethoS-- (Read error: 110 (Connection timed out))
21:53:29preglownot in windows now, or i'd try
21:54:11LloreanI don't have any special accessibility features installed, but it found the Anna voice on its own, and I went and confirmed my computer at least seemed to be able to speak in Control Panel
21:54:28LloreanI've disabled UAC entirely, so I doubt very much that's getting in the way.
21:54:51n1sLlorean: might be worth a shot trying with admin rights
21:55:19hachiI like the part where I hit the 'skip back' button in rockbox
21:55:23hachiand it keeps going forward
21:55:28LloreanI'm on an admin account, so everything is run with admin rights.
21:55:53amiconnIsn't Anna a German voice?
21:56:02Lloreanamiconn: It says Anna (english)
21:56:13n1sMaybe she's multilingual :)
21:56:17amiconnAha, don't know that one
21:56:25LloreanOr rather "Microsoft Anna - English (United States)"
21:56:29pixelmaweren't there reports of exactly the same issue in the forums? Thought that was fixed though...
21:56:52LloreanYeah, I seem to recall reports of that, but I couldn't remember if it was forums, or list.
21:58:44LloreanWhen I try to generate .talk files, encoding fails
21:58:47LloreanIs that a clue?
21:59:12bluebrothermaybe. Does it tell any details?
21:59:19 Quit LambdaCalculus37 ("CGI:IRC 0.5.9 (2006/06/06)")
21:59:44LloreanEncoding of /path/to/ failed is all
21:59:49gevaertsFor some reason this thing won't work anymore in Windows without an sd card plugged in...
22:00:13 Quit bertrik ("bye")
22:00:50 Quit gevaerts (Read error: 110 (Connection timed out))
22:02:08Lloreanbluebrother: Any idea where it keeps temporary files? I'm not seeing any signs of it even creating the .wav files
22:03:27preglowdoes it create wav files?
22:03:34preglowoh, right, yes it does
22:03:39Phillgevaerts: on my e200 with ums_v22 the drives no longer show in windows, although the devices are listed in device manager
22:04:18Lloreanpreglow: It ought to, I thought. :)
22:04:55preglowyeah, and the code i wrote takes a wav file, so it pretty much has to
22:05:08preglowunless that got changed without me noticing
22:05:09LloreanI can only assume then that the speech isn't working
22:05:18LloreanDo I have to have speech actually enabled somewhere in windows?
22:05:18 Quit Phill ()
22:05:22 Quit mf0102 ("Verlassend")
22:14:12 Quit Arathis ("Bye, bye")
22:14:20 Join stripwax [0] (
22:15:09preglowamiconn: i assume there are no better ways to load an address on coldfire than use move.l?
22:16:26 Quit hannesd (No route to host)
22:18:03 Quit hcs ("Leaving.")
22:20:46 Join hamdiya [0] (n=c3e5ecd6@gateway/web/cgi-irc/
22:21:02 Join gevaerts [0] (
22:21:27 Quit hamdiya (Client Quit)
22:22:29 Join hamdiya [0] (i=c3e5ecd8@gateway/web/ajax/
22:22:32 Quit ol_schoola ()
22:22:37 Quit BitTorment__ ("Do directly to guantanamo bay. Do not pass go. Do not collect your human rights.")
22:23:17hamdiyaneed some help please with my first wps
22:23:35hamdiyai've got alot of the graphics ready but need some help with the coding
22:24:06hamdiyatrying to make a custom progress bar and ive used $xl to preload a few images of the progress bar
22:24:17hamdiyahowever, it does not move
22:24:24hamdiyathis is the code i;ve written
22:24:39 Quit bluebrother (Read error: 110 (Connection timed out))
22:24:50 Quit nicktastique ("Leaving")
22:25:18 Join BitTorment [0] (
22:26:13*Llorean finally tries CabbieV2 on his gigabeat.
22:26:24LloreanI have to say I really like how the current version in SVN looks.
22:26:57hamdiyaanyone please
22:27:31scorche|shhamdiya: have a bit of may even be better off posting this question in the forums
22:27:39pixelmayou wrote "$xl" is that a typo in here (should be "%xl") - and do you want something like a real bar? Because then you could use the %P tag and not several single pictures...
22:28:11hamdiyai actually want a custom bar
22:28:23hamdiyathe %P tag is not suitable for my theme
22:28:44hamdiyadid i wirte $xl?
22:29:09pixelmain your post here, yes. I have to admit that I never used %px though
22:29:45hamdiyaohh that
22:30:05hamdiya%px returns the percentage played
22:30:13hamdiyai assumed i should be using this
22:30:29hamdiyait returns a digit from 0-99(100?)
22:30:45hamdiyais the conditional statment corrent by the way?
22:30:48*Domonoky_ hears about voice generation problems with rbutil, which means finally someone tries it.. :-)
22:31:07JdGordon|winteresting logic
22:31:22preglowbut valid
22:31:40gevaertsI can't figure out how to make Windows handle this removable slot correctly. It tries reading a partition table, and gets an error back. This is new in the latest patch (it used to just send back random garbage).
22:32:12LloreanDomonoky_: Yes, I tried it. Windows Vista, nothing special installed. It recognized the one voice I seem to have, but created a 5k file. Talk clip generation doesn't seem to output wavs (I don't seem them anywhere), or at the very least, it errors when trying to encode the wavs.
22:32:13n1shamdiya: looks correct to me, maybe you should try with a sim using the −−debugwps switch
22:32:23pixelmahamdiya: yes, I read that but never used it myself so all I can say I'd probably try the same
22:32:26gevaertsI can work around this by only letting the host know about the slot if there is a card present, and find a real solution later. Any thoughts on that ?
22:32:42hamdiyathe −−debugwps switch
22:32:45hamdiyanoob here
22:32:49hamdiyahow do i do that?
22:33:02n1shamdiya: do you have a simulator?
22:33:08JdGordon|wgevaerts: you could do what my phone does and use a folder for the sd card instead of allowing it to be a seperate drive?
22:33:22hamdiyaimagine making a wps withouth one :O
22:33:24Lloreangevaerts: I agree with JdGordon|w
22:33:33JdGordon|wnot really nice though
22:33:34LloreanSince we present the SD card as a folder in Rockbox, why don't we do that too?
22:33:41gevaertsJdGordon|w: no. UMS exports block devices, not filesystems.
22:33:56LloreanBasically, present windows with the exact same file tree as we present the user. (Or at least, similar, since the folder name may be invalid)
22:34:03n1shamdiya: start it with −−debugwps from a command line
22:34:04LloreanAh, nevermind then. =/
22:34:27Domonoky_Llorean: rbutil puts those temporary files in the temp folder (for voice generation)
22:34:52LloreanDomonoky_: Ah, well when doing .talk generation, the path it mentioned in the error was on the device, when looking for the .wav
22:34:58hamdiyak, lemm try
22:35:01Domonoky_and it deletes it again... for talk generation you can tell it, to keep the wav files..
22:35:13LloreanWhere is the temp folder?
22:35:27n1shamdiya: if you are using a lot of images it might be failing because the buffer is full
22:35:36Domonoky_for talk generation the wav files will be on the device... this is different from voice..
22:35:54LloreanDomonoky_: I tried both talk and vioce.
22:35:57JdGordon|wgevaerts: in that case.. yeah, just ignore the slot if no card is present.. or maybe even ignore the slot completly for the time being
22:36:04preglowJdGordon|w: i bet your phone doesn't use ums
22:36:09LloreanDomonoky_: With talk, the .wav never showed up, and it errored when it tried to encode the .wav that wasn't present.
22:36:15 Join BoBKA [0] (n=58c42d6c@gateway/web/cgi-irc/
22:36:16LloreanWith voice, it reported no errors, but I got a 5k voice file
22:36:23JdGordon|wpreglow: pretty safe bet there...
22:36:26hamdiyathe image is quite small
22:36:33Domonoky_Llorean: then its probably the sapi voice
22:36:49hamdiyaonly loaded it 8 times
22:37:08LloreanDomonoky_: Know of anything I can do? I can make windows speak, so the voice is definitely present, but do I need to do something to enable it to be used by RButil or something?
22:37:22hamdiyais there some kind of log for the −−debugwps?
22:37:29Domonoky_Llorean: would be nice if you could try the sapi_voice.vbs script alone, (i can give you the commands to try)
22:37:32BoBKAwere a can get new plguins for rockbox player e200 series
22:38:21LloreanDomonoky_: Well, I need to leave soon. But I can PM you my email address if you'd like to send me what you want tested, and I'll email you back what results I get, or just try to catch you on IRC again soon. Will that work?
22:39:33Domonoky_just try: "cscript sapi_voice.vbs /language:english"
22:39:46n1shamdiya: are you on windows?
22:39:56LloreanWhere do I get sapi_voice.vbs?
22:40:09Domonoky_then enter: SPEAK<tab>test.wav<tab> testing<enter>
22:40:09n1shmm, dunno if that works on windows :-/
22:40:26Domonoky_Llorean: tools folder of rockbox source.. :-)
22:41:09 Quit BoBKA (Client Quit)
22:41:53***Saving seen data "./dancer.seen"
22:42:01LloreanI got speech there
22:42:03n1shamdiya: I could check it out if you could post a zip somewhere
22:42:14pixelmahamdiya: since I don't see the complete wps (and if you want to show, don't paste it here but use a pastebin), are the images at a different position and does the WPS look correctly otherwise?
22:42:32Domonoky_Llorean: it generated a test.wav ?
22:42:38 Join ch4os_ [0] (n=ch4os@unaffiliated/ch4os/x-059673)
22:42:38scorche|shhamdiya: also, try to not use the enter key as much...put your thoughts on one line please ;)/w 3
22:42:45LloreanDomonoky_: 46k, plays fine.
22:43:13Domonoky_hm.. try cscript sapi_voice.vbs /language:english /listvoices
22:43:24hamdiyathe bg image loads good
22:43:29Domonoky_this will give you a list of aviable voices...
22:43:36hamdiyaand so do the song details
22:43:37LloreanDomonoky_: Doesn't list any\
22:43:59hamdiyais it correct to do this
22:44:00LloreanMicrosoft Anna - English (United States) is the only one
22:44:28Lloreanhamdiya: If you need to make a multi-line paste, please use pastebin
22:44:33LloreanAs per the channel guidelines
22:44:50Domonoky_then try: cscript sapi_voice.vbs /language:english /voice:"Microsoft Anna - English (United States)"
22:44:51hamdiyai must go
22:44:53hamdiyasry guys
22:45:04hamdiyaill come back later
22:45:08hamdiyahope to catch uz
22:45:13 Quit hamdiya (" ajax IRC Client")
22:45:30n1sdo we support loading the same bitmap with multiple ids and positions?
22:45:40Domonoky_and then the SPEAk command again... (ie create a file with a specific voice)
22:46:13LloreanDomonoky_: Says it finds no matching voice
22:46:43Domonoky_so, there is the problem, thank you for testing, i will investigate it..
22:46:57gevaertsOK, that works.
22:47:10pixelmaand I even reminded him...
22:47:15LloreanDomonoky_: If I shorten it to simply "Microsoft Anna" it doesn't complain
22:47:57Domonoky_nice hint.. :-)
22:48:05LloreanAnd it creates a playable wav file
22:48:25Domonoky_jup, if it finds the voice, all should work..
22:48:34LloreanWell, I must go. Good luck. :)
22:49:32pixelman1s: sure (but it will load it x times, it does not know that it's the same image and then it needs a different ID if it should be at a different position)
22:54:49n1spixelma: tried a similar wps like the one he posted in a sim and it works...
23:00:37 Join asdrubal [0] (n=abc@unaffiliated/asdrubal)
23:01:35 Part agm3nt
23:02:22Domonoky_amiconn: do you know how to best solve the problem with sapi_voice.vbs listing other voicenames then accepting as voice selection ? ( ie. getDescription returns not the same as needed as "name: " value in the select string...
23:04:29Domonoky_we could either loop over all voices and comparing the description while selecting the voice, (ie not using the selectstring for voice selection) or change the listvoices code, to give the real voice names..
23:04:55Domonoky_(i dont know how todo the second..) :-)
23:07:27amiconnShould be rather simple
23:07:40amiconnBtw, I'm currently fiddling with sapi_voice.vbs
23:07:50amiconnFound a sapi4 bug already
23:08:45 Quit JdGordon|w (" ajax IRC Client")
23:09:25 Quit desowin ()
23:09:26*petur reads his mails and sends congratulations to gevaerts
23:09:34Domonoky_so i will leave this to you, and go sleeping :-)
23:09:44 Quit Domonoky_ (Read error: 104 (Connection reset by peer))
23:10:45amiconnToo slow... I just committed a fix :\
23:14:09Soap_what is the best way to remove the Rockbox dev environment? I don't need arm-elf, sh-elf, and m68k-elf on this tiny machine anymore.
23:14:14 Quit stripwax ("Miranda IM! Smaller, Faster, Easier.")
23:15:49gevaertsI am testing if my patch still gives a working system if USE_ROCKBOX_USB is not defined, and I get an interesting result : when I plug in my c250, the backlight immediately turns off and all blue leds turn on, so it looks exactly like manufacturer mode. After this, I have to power it off the hard way. Any hints ?
23:16:33preglowamiconn: i assume there are no better ways to load an address on coldfire than use move.l?
23:16:58amiconnDepends on whether the address is relative to something
23:17:22amiconnSometimes you can use pc relative lea.l
23:17:44*gevaerts thanks petur and gives him a beer
23:18:36preglowamiconn: is there a way to construct a lea.l sequence from a label?
23:19:06pixelmagevaerts: if you hold the sansa close to a lamp in this case, do you see the Rockbox USB screen? If it's that then it's /probably/ the usual hang when it's trying to reboot into OF that's there in some builds...
23:20:41pixelmae.g. I have that just now with my self-compiled build without the USB patch (built on r16350)
23:23:30gevaertspixelma: it is indeed showing the USB screen.
23:24:49 Quit Axio ()
23:30:54 Join perrikwp [0] (i=98216e29@gateway/web/ajax/
23:32:44 Quit ompaul (Client Quit)
23:34:13 Quit ch4os_ ("Lost terminal")
23:38:21 Join corevette [0] (
23:38:57corevetteis there an audio scrobbler for for rockbox
23:40:11peturyou enable it in the settings, and get QTScrobbler to upload
23:40:41 Join ol_schoola [0] (
23:42:31pixelmaread more about it here
23:42:52gevaertsI uploaded ums_v23. I think it is a clear improvement on current svn, so I'd like to commit it. It shouldn't influence anything else. Any thoughts ?
23:43:05*gevaerts would like his first commit not to have to be reverted immediately
23:43:25peturhehe but maybe you can earn some points ;)
23:43:37amiconnIs it enabled or disabled by default?
23:43:50gevaertsamiconn: still disabled.
23:43:52amiconnI noticed that the code spits several warnings when enabled
23:44:27gevaertsI'll check
23:46:24 Quit n1s ()
23:46:53 Quit jgarvey ("Leaving")
23:47:12*petur is seriously stuck in this CF business... two changes that contradict the datasheet (powermanagement and soft reset)
23:48:49preglowgevaerts: if it's an improvement and doesn't affect anything else, i'd say go ahead
23:48:59preglowgevaerts: just make sure to not commit that change that is in your patch :)
23:49:02gevaertsamiconn: I fixed the warnings here.
23:49:27preglowgevaerts: no biggie if there are warnings anyway, as long as you're around after the first build to check out if it went ok
23:49:57gevaertspreglow: I noticed that as well. I try to always run diffstat on it first, but I forgot this time...
23:50:40 Quit alienbiker99 ("( :: NoNameScript 4.02 :: )")
23:51:35preglowgevaerts: just tested on nano, works peachy :)
23:52:21gevaertsOK. Let's go...
23:52:58*preglow does the first commit dance
23:53:18 Quit DataGhost (Nick collision from services.)
23:53:26 Join DataGhost [0] (i=dataghos@unaffiliated/dataghost)
23:53:51*petur guesses some script will miss the login <-> name translation
23:56:38BigBambiPetur was right
23:58:45 Quit Soap_ (Remote closed the connection)

Previous day | Next day