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

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

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

#rockbox log for 2009-11-09

00:00:32TorneUnhelpful: yeah, if the beast gets to the point of thinking the disk has an error, but it can't fix the error because of e.g. not being able to do 512 byte sector writes it will just reboot and reformat in a constant loop until you power off
00:00:45Torne*all* errors are handled, ultimately, by kicking the "nuke disk from orbit" function
00:00:50Torneif less drastic measures fail
00:01:02gevaertsTorne: as far as I know it's only this one ipod disk that doesn't do 512 bytes
00:01:12Tornegevaerts: yeah, but iirc the bootrom also doesn't do lba48
00:01:33Torneso if it tries to repartition a disk with more than 128GiB it may also go wrong
00:01:52gevaertsah indeed. Depends on how it handles that case exactly
00:01:53TheSevenhm, I consider nuking the disk more evil than showing an "empty battery" logo, indeed.
00:02:14UnhelpfulTorne: another argument for rockbootflash - if we can shoehorn charging into rockboot, too, then there's really nothing other than corrupt flash you couldn't recover from rockboot, since it has bootloader usb mode..
00:02:28Tornei'm still trying to figure out the exact details of the config dat stored in the flash
00:02:44Torneyou can disable reformatting in there, but i think that bsame bit also changes other things
00:02:52Torneyou can make it think it's a devboard :)
00:03:12Torneand disabling the reformatting means in that situation it will just go into a plain reboot loop
00:03:15Tornewithout formatting )
00:03:18Torneit still won't let you recover
00:03:44TheSevendoes it have a bootrom DFU or something similar that lets you recover from a bad flash?
00:03:44TorneUnhelpful: well, we'll see i guess ;)
00:03:50TorneTheSeven: pretty sure it doesn't
00:03:57Tornethe available evidence suggests it's configured for nonsecure boot
00:04:00TheSevennot good.
00:04:04Torneand thus runs direct from flash address 0
00:04:12Tornethe chip range does have a secure boot option though
00:04:29Tornebut boot config is done in hardware of course
00:04:29Unhelpfulhrm, is there jtag on the board somewhere, perhaps unsoldered? :D
00:04:40Tornethere's a serial port, unsoldered
00:04:43Tornedunno about jtag
00:04:56Tornethe bootrom prints loads of stuff to said port, as seen on the wiki
00:06:12*Torne checks the partition thing
00:06:50Tornealso i lie, i only have 1.2 and 1.3 updters apparently
00:07:00Tornebut several of the flash dumps i have are neither 1.2 nor 1.3
00:07:10Torneso i must have a 1.0 or 1.1 somewhere :)
00:08:34 Quit bertrik ("De groeten")
00:10:01Torneyeah, it just checks for type=0 or type=ff
00:10:06UnhelpfulTorne: yes, but a jtag is more as we can (probably?) program flash through one. ;)
00:10:07Torneso you could 'hide' another partition from it that way
00:10:23Torneif you made a hacked rockbox that ignored the type and checked for a nonzer start lba, or similar
00:10:45Tornehow do you tell if something does LBA48? :)
00:11:08*gevaerts doesn't know
00:11:12TorneUnhelpful: nwell i've not even opened this one :)
00:11:58UnhelpfulTorne: hrm, i have the rest of the one i *don't* use about. probably even a mobo with broken usb will suffice for looking for jtag. ;)
00:12:01gevaertsAnyway, that's a really easy thing to change in the source, and probably one that we could reasonably even put in svn with appropriate #ifdeffing
00:12:37*Torne updtes wiki re number of paritions
00:13:31Torneyeah it's definately only LBA28 though.
00:13:43TorneBut that shouldn't actually be a problem as long as the bootloader doesn't try to read past 128GiB
00:13:55Torneit might even be ok doing a reformat..
00:14:08Torneas long as there aren't mor than 2^32 sectors
00:14:23Tornesince the FAT/rootdir on the media partition will be at ~150MB in
00:14:32TorneFAT doesn't write any data on format except to near the start
00:14:54gevaertssomeone should try!
00:15:10Tornethat would be awesome
00:15:16gevaertsactually, I think someone did. I can't remember who :\
00:15:17 Quit stripwax ("Miranda IM! Smaller, Faster, Easier.")
00:15:27Tornegevaerts: they may have been using the dodgy ipod disk though
00:15:27 Join stripwax [0] (
00:15:30Torneas that's an easy one to get, no?
00:15:34gevaertsthe thing is, you need to be able to experiment yourself...
00:15:48Torneme? or people in general? :)
00:15:58gevaertswhoever has this big disk
00:16:05Torneah, yeah
00:16:14Unhelpfulyes, somebody tried with a giant disk, and had difficulties. and it seemed that either the hardware or rockboot could not address the disk...
00:16:16Torneideally you'd wanna try it on a beast which has the serial port connected
00:16:26Torneall the 'early' fails in the bootloader have debug messages
00:16:45*gevaerts seems to remember that there's something on the forums
00:16:48Torneall the partitioning/formatting stuff is very verbose
00:17:05Torneit's quiet later though, but it checks a verbosity flag soemwhere
00:17:12Tornewhich is probably also in teh flash config area but i'm not sure which bit
00:17:38Tornethe flash config bit is hard to undertand, i'm gonna coem back to it later when i have more of the code mapped out as the more xrefs to it there are in the DB the easier it is to deduce things about it
00:18:01 Quit dmb (Connection reset by peer)
00:19:19TorneI expect the problem there is that he expanded the partition by hand
00:19:22 Quit bmbl (Read error: 104 (Connection reset by peer))
00:19:34Torneand when he formatted it on his PC he probably violated one of the crazy arbitrary rules
00:19:36gevaertscombined with your findings, I think it shouldn't be too hard to get to work with maybe this 0xff partition hack
00:19:51Tornewell as i said it might work even with just twopartitions
00:20:30Tornei'm not sure about it making the storage partition only 128GiB, I guess i might've missed something in CreatePartition but i would expect it to just be using 32-bit values
00:21:33gevaertsit might be worth it to PM him on the forums and find out if he still has the disk
00:21:46AlexPIt seems that the 4096 sectors weren't an issue?
00:22:00Tornelarger sectors are not innately an issue
00:22:14 Join dmb [0] (n=Dmb@unaffiliated/dmb)
00:22:14Tornebut if the bootloader tries to *write* to a disk which doesn't do the 512 byte sector emulation it will die
00:22:21AlexPah right
00:22:35gevaertsAlexP: from what I understand the vast majority of 4096-sector disks do 512-byte sector emulation
00:22:36Tornealso if it sees a FAT BPB which says the sector size is anything other than 512 it reformats
00:22:54Tornesoo you would have to make sure your format tool didn't say the sector size was 4096 in the FAT
00:23:23AlexPI take it this disk does do the emulation, as he says that the OF bootloader created the 127 GB partition
00:23:25Tornethere are a lot of ways to get the disk layout wrong, basically :)
00:23:37TorneYeah i'm surprised by that but I my have missed somethign in CreatePartition
00:23:42Tornecba to read it all again now, maybe tomorrow
00:23:48AlexPheh :)
00:23:52Tornei would expect it to use whatever ATA IDENTIFY said the capacity was
00:24:01Torneunless ATA IDENTIFY returns >28bit capacities in a new field?
00:24:05Tornei'm not very expert on AT
00:24:09AlexPWell it'd be interesting - if we could hack around it I'd get the disk :)
00:24:40Torneidle question about the beast, unrelated to the bootrom: do we do boosting on it? the debug menu indicates not
00:24:52AlexPNo, it is fixed at half max frequency
00:25:03Torneany particular reason? just curious
00:25:13Tornealso i assume you know it has automatic dfvs in hardwre
00:25:15AlexPDon't know :)
00:25:50AlexPIt's battery life is still pretty crap, so frequency and especially voltage scaling would be nice
00:25:57Tornewould rockbox cope with having the clock freq change dynamically under its feet? :)
00:26:02Tornethe ahrdware can do it entirely on its own
00:26:12Tornebased on measurement of the pipeline load
00:26:51AlexPCould you just enable that and disable boosting in Rockbox?
00:27:17AlexPand let it handle it?
00:27:29Tornewell that's the idea of the hardware
00:27:33Tornei only read he imx31 datasheet..
00:27:40AlexPyeah, I meant w.r.t. Rockbox :)
00:27:43TorneI have no idea
00:27:46Tornethat was what i wa asking :0
00:27:52AlexPOR would it cock up timings or some such
00:27:53 Quit AEnima1577 ("Leaving.")
00:27:58AlexPTorne: Oh, right :)
00:28:01Torneyou are overestimating how much I know about how rockbox works
00:28:12TorneI know far more about ARM in general than the actual details of rockbox here ;)
00:28:58gevaertsjust try it!
00:29:30*AlexP waits for Torne's patch to do just that :)
00:30:01Tornewell i could give it a go i guess, but the datasheet is rather cryptic on that one
00:30:02 Join fyre^OS [0] (
00:30:18 Join hd [0] (
00:31:05AlexPTorne: Beast on standby :P
00:31:05 Quit Unhelpful ("No Ping reply in 90 seconds.")
00:31:14Tornenot now ;)
00:31:15 Join Unhelpful [0] (
00:33:37TheSevenc5000 decoding is approaching it's end by the way
00:33:50gevaertsalready? :)
00:33:52 Quit HellDragon (Operation timed out)
00:34:20Torneit would be good if someone knew about timing and stuff commented ont he subject before anyone sat there poring over the datasheet for too many hours ;)
00:34:22TheSevenc5000 is more than insanely slow - at least compared to other players
00:35:44TheSeven15 seconds to go
00:36:41Tornebut yeah, if someone wants to ahve a go with a bigger disk in the beast, particularly if they are willing to try a serial cable to debug, they should let me know ;)
00:37:41*gevaerts thinks it's probably a good idea to ping cool_walking
00:38:09 Quit martian67 (Nick collision from services.)
00:38:10 Join lol3 [0] (
00:38:12Torneyeah i'll pm them
00:38:46Tornehm, i would email them but the captcha is broken :)
00:39:05TheSevenhm, how big is that c5000 filter buffer?
00:39:25TheSevenafaik the nano2g has 96KB codec IRAM
00:40:19UnhelpfulTorne: well, the tick is driven by a timer interrupt, isn't it? it should cope with frequency changes, i'd think.
00:42:39gevaertsTheSeven: don't forget to update SoundCodecMonkeysAudio
00:42:40 Join kisuke [0] (
00:43:49TheSevengevaerts: the test just finished
00:44:12TorneUnhelpful: assuming that the changes don't disrupt any of the *other* clocks in the system, yes ;)
00:44:14TheSevenmy guess was quite good
00:44:18TheSeven3.76% realtime
00:44:24Tornei really didn't read the details of the DFVS hardware, just noticed that it exists
00:44:26TheSevenwhat the hell is going on there?
00:44:34Torneincidentally we could probably turn the voltage adaption hardware on anyway..
00:44:36 Quit dmb (Connection timed out)
00:44:52Torneit is *supposedly* supposed to be ablet o control the voltage autoamtically absed on the current clock and the environmental conditinos
00:45:08Tornethis is a pretty sweet SoC
00:45:12Unhelpfulalso, there's a small problem if we're only talking about *frequency* changes - amiconn and others have tried adjusting only clock and it saves pretty much nothing.
00:45:14AlexPTorne: Anything for battery life would be good :)
00:45:30Unhelpfulit's very, *very* sweet in terms of idle efficiency ;)
00:45:34TorneOh yes
00:45:37Torneit will be
00:45:55Tornebut the voltage scaling should help
00:47:02Tornei've employed my housemate as a tester, btw :)
00:47:12Tornehe's using the beast on a daily basis now :)
00:47:17AlexPcool :)
00:47:25Torneso i can keep using the ipod for DMA testing ;)
00:47:32AlexPI'm also very happy to test any and everything :)
00:47:51Torneyah, well, actual development wise i have rather a list to be getting on with tbh :)
00:47:59Tornewhether i get to any of this stuff is debateable
00:48:01 Quit fyrestorm (Read error: 110 (Connection timed out))
00:48:06amiconnTheSeven: For armv4, insane filter buffer is 34816 bytes. For armv5 and higher as well as for coldfire, half of that
00:48:14Tornethe ipodvideo is my actual device, remeer, the beast is just a loaner :)
00:48:18amiconnSee apps/codecs/demac/libdemac/decoder.c
00:48:58 Quit reid05 (Success)
00:49:00 Join reid05 [0] (
00:52:16 Quit niekie (Remote closed the connection)
00:53:46UnhelpfulTorne: turning that on would just be flipping some bits, though, right?
00:56:55TorneUnhelpful: Er, probably
00:57:02Tornethere was some stuff about it generating interrupts and so on
00:57:10Tornego read the imx31 datasheet
00:57:21Tornei onlyt gave it a quick flick through
00:57:21Unhelpfulah... but maybe we don't need to see those?
00:57:35Torneyes, it might be sufficient to not even notice :)
00:58:05Unhelpfulno easier way than to try ;)
00:59:52 Quit n17ikh (Read error: 54 (Connection reset by peer))
01:02:20*TheSeven would like to kill someone for that:
01:03:44 Nick drf|away is now known as drf|laptop (
01:04:47drf|laptopSo guys
01:04:50linuxstbTheSeven: "svn blame" will tell you who... ;)
01:04:58*linuxstb hopes it isn't him...
01:05:04drf|laptopOn my Toshiba Gigabeat S, I keep getting "Dir Buffer is Full"
01:05:09drf|laptopDoes that mean I have too many files?
01:05:26TheSevenlet me have a look...
01:06:55 Quit ender` (" C++ is a modern language where your parent can't touch your privates but your friends can!")
01:07:16Tornedrf|laptop: it means you have too many files in one directory, I believe; there is a setting to change the limit
01:07:30drf|laptopOh. So I can just turn the limit of?
01:07:33TheSevenintroduced by jethead71 in r15977
01:07:35linuxstbTheSeven: Where is that from? I can't find it with grep...
01:07:42TorneThe memory used is statically allocated, you can't remove the limit
01:07:43drf|laptopI just put all my songs in the main directory, to make it easier since I don't use Database view
01:07:47 Quit mikroflops (Read error: 110 (Connection timed out))
01:07:52Tornethe more you set it to, the more memory it will use up
01:08:00 Join n17ikh [0] (
01:08:01Tornebut the beast has a lot of memory, so it's probably not a big deal.
01:08:01drf|laptopmemory as in disk space?
01:08:02linuxstbTheSeven: Ah...
01:08:06TorneNo, the player's RAM
01:08:11 Quit pixelma (Nick collision from services.)
01:08:13 Join pixelma_ [0] (i=quassel@rockbox/staff/pixelma)
01:08:18 Join uflops [0] (
01:08:28 Quit amiconn (Nick collision from services.)
01:08:32 Join amiconn_ [0] (i=quassel@rockbox/developer/amiconn)
01:08:32 Nick pixelma_ is now known as pixelma (i=quassel@rockbox/staff/pixelma)
01:08:38 Nick amiconn_ is now known as amiconn (i=quassel@rockbox/developer/amiconn)
01:09:04drf|laptopWait, where's the setting at? It's not in General Settings - File View, at least I don't see it there
01:09:06linuxstbTheSeven: Is that a problem though?
01:09:19TheSevenit may be a performance hit and is just the wrong way round
01:11:16AlexPdrf|laptop: RTFM
01:11:58***Saving seen data "./dancer.seen"
01:12:05 Join AndyIL [0] (n=pasha_in@
01:12:37TheSevenis there some properly encoded test mpeg file lying around somewhere?
01:13:29AlexPInconsistent wiki page capitilisation is annoying
01:13:47linuxstbTheSeven: Did you find the problem with mpegplayer on the Nano2G? (I haven't kept up with the logs the last few days)
01:14:22drf|laptopWoah. I set the limit to 10000 (the highest), still says the dir buffer is full
01:14:24TheSevenlinuxstb: I just decided to have a try hunting it
01:14:25drf|laptopI have that many songs :/
01:14:38TheSevenbut i bet it was introduced as a side effect of r23234
01:17:14 Quit robin0800 (Remote closed the connection)
01:18:17TheSevenlinuxstb: i think i just spotted it
01:22:52 Quit AndyI (Read error: 110 (Connection timed out))
01:24:03Tornedrf|laptop: do you actually have more than 10,000 files in one directory?
01:24:19Torneif so i would, er, suggest not doing that, srsly
01:24:26Tornemake directories called A B C D E etc :)
01:24:32drf|laptopNah, I figured it out
01:24:37drf|laptopI just had to restart the player >_>
01:24:39drf|laptopI have about 5000
01:24:44drf|laptopall in the root directory
01:24:54TorneYes, al these limits are set because it determines how much RAM it sets aside at boot time
01:25:00Tornebtw, i hope you have the directory cache turned on
01:25:08Torneotherwise you will get some seriously poor performance out of that ;)
01:25:19TorneLarge directories take a loooong time to index
01:25:28Tornethe dircache avoids this if it's enabled
01:28:39 Quit flydutch ("/* empty */")
01:30:47 Quit TheSeven ("ChatZilla 0.9.85 [Firefox 3.5.4/20091016092926]")
01:32:50 Nick drf|laptop is now known as drf|away (
01:33:41 Quit hebz0rl ("Verlassend")
01:38:12CIA-6New commit by kkurbjun (r23572): M:Robe 500 - Enable ASM optimized ATA interface improving disk access speed.
01:38:21 Join TheSeven [0] (n=theseven@rockbox/developer/TheSeven)
01:39:39 Join perfectdrug [0] (
01:39:56TheSevenmpegplayer first plays garbage audio with racing video, then video stalls and audio plays normally for a while, and then, after some minutes, they're finally in sync again and play fine
01:40:08perfectdrughi, any X5 user around?
01:41:16 Quit stripwax ("")
01:42:28TheSevenand audio is a little clicky
01:43:17perfectdrugI just filed a bugreport and would like if someone could confir it:
01:43:44 Quit Thundercloud (Remote closed the connection)
01:44:53 Quit perfectdrug ("CGI:IRC")
01:46:18amiconnBah, gone
01:48:11 Join Blue_Dude [0] (n=chatzill@rockbox/developer/Blue-Dude)
01:48:43Blue_DudeBTW, Torne, you're the man. I just looked up directory cache and learned something. I never knew I needed that.
01:49:20 Join robin0800 [0] (
01:49:24Tornepeople in general: why is the dircache not defaulted to on, on targets which have a rotating disk and a reasonable amount of RAM? It's ahrd to believe that anyone would *really* actually want it off on ipodvideo, for example
01:49:59Tornei mean ok if you have really long files and you want the maximum possible playback life, I guess, but that's hardly usual :)
01:50:08Blue_DudeI have a flash based player and it helped there too.
01:50:18Torneyes, it can help on flash devices :)
01:50:23Tornebut it's less clear cut
01:50:25Blue_DudeLots of files in some directories, up to 1000.
01:51:03Blue_DudeIt meant 1/2 second or more of lag. Just enough to be annoying.
01:51:28Blue_DudeJust wanted to let you know I appreciate the heads up.
01:53:52 Quit Blue_Dude ("ChatZilla 0.9.85 [Firefox 3.5.5/20091102152451]")
01:55:26 Quit reid05 (Read error: 110 (Connection timed out))
01:56:07 Join reid05 [0] (
01:57:02 Join z35 [0] (
02:06:30 Join perfectdrug [0] (
02:06:57 Join BHSPitLappy [0] (n=BHSPitLa@unaffiliated/bhspitmonkey)
02:08:17perfectdrugamiconn: I'm sorry back again
02:17:59 Quit MethoS- (Remote closed the connection)
02:29:39kkurbjunJdGordon are you around?
02:30:16*TheSeven gives up on mpegplayer
02:31:19kkurbjunI'm trying to mess with this custom status bar stuff and it just rockbox with a segfault in the simulator if I enable the custom statusbar - I just took the and loaded it - immediately the sim crashed with a seg fault
02:31:46 Quit robin0800 (Remote closed the connection)
02:34:56kkurbjunactually, without even loading a custom statusbar, and changing the statusbar option from Top to Custom it segfaults
02:38:47 Join froggyman| [0] (
02:39:08kkurbjunit looks like it may be related to a target having a remote LCD - I can crash it on the H300 and MR500 by just enabling the setting
02:39:26kkurbjunthe gigabeat F sim does not crash when I turn it on
02:40:46 Join seani [0] (n=seani@
02:44:35 Nick uflops is now known as mikroflops (
02:55:33 Nick fxb is now known as fxb__ (
02:57:35 Join toffe82_ [0] (
02:58:58 Quit kisuke (Read error: 60 (Operation timed out))
03:02:51 Quit froggyman| ("CGI:IRC")
03:03:28 Join froggyman| [0] (
03:04:03 Join saratoga_ [0] (i=463f90ed@gateway/web/freenode/x-ltrxoycjrxarxycw)
03:04:32saratoga_TheSeven: whats with the slow decode time on the nano2g? you have IRAM right?
03:05:25saratoga_WMA is only slightly faster on the Nano2G then on PP inspite of PP being clocked at just 80MHz and being ARM7TDMI
03:11:16 Join grant__ [0] (i=d83bf1dc@gateway/web/freenode/x-odyriallsaedethy)
03:12:00***Saving seen data "./dancer.seen"
03:12:33 Quit toffe82 (Read error: 110 (Connection timed out))
03:13:10grant__ive got a bug in the ipod 2nd gen version of rockbox
03:13:16grant__i cant select my files
03:13:50grant__i have had this bug before and it was fixed but now ive got it again
03:13:54grant__any help?
03:18:44 Join Strife89 [0] (
03:19:23 Join beta_ [0] (
03:19:31 Nick beta_ is now known as beta2k (
03:19:35 Quit grant__ ("Page closed")
03:24:31 Quit perfectdrug ("CGI:IRC")
03:25:28 Join AEnima1577 [0] (
03:26:30 Join grant__ [0] (i=d83bf1dc@gateway/web/freenode/x-bptpseacokicaykc)
03:28:25 Quit grant__ (Client Quit)
03:50:32 Quit saratoga_ ("Page closed")
03:54:17 Join evilwombat [0] (
03:57:43 Join kisuke [0] (
03:58:49 Quit froggyman| ("CGI:IRC")
04:00:29 Part evilwombat ("^D")
04:07:36 Part kisuke
04:07:47 Join kisuke [0] (
04:07:59 Quit Rondom (Nick collision from services.)
04:08:07 Part kisuke
04:08:09 Join Rondom [0] (
04:08:26 Join kisuke [0] (
04:09:38kisukeok does the installer need to run as root
04:13:10 Quit AEnima1577 ("Leaving.")
04:15:22 Part kisuke
04:15:37mc2739kisuke: it depends on the target - some need root and others do not
04:18:48 Quit AaronM ("Emo Time In My Corner... //_-")
04:26:07 Quit BlakeJohnson861 ("Leaving.")
04:26:30 Join BlakeJohnson86 [0] (
04:27:36 Join Llorean [0] (
04:34:14 Quit TheSeven (Nick collision from services.)
04:34:33 Join The_Seven [0] (n=theseven@rockbox/developer/TheSeven)
04:34:45 Nick The_Seven is now known as TheSeven (n=theseven@rockbox/developer/TheSeven)
04:41:29 Quit hd (Client Quit)
04:45:31 Join mike265 [0] (
04:46:26 Join webguest44 [0] (
04:46:41 Part mike265
04:46:41 Quit webguest44 (Client Quit)
04:48:41LambdaCalculus37There were people here!
04:49:09 Join keastes [0] (
04:50:13 Quit Riku (Read error: 131 (Connection reset by peer))
04:51:31keastesim running ubuntu 9.10 and the and the auto installer will not ID it any suhgestions?
04:51:56keastesiPod video*
04:52:34mc2739keastes: did you try running as root?
04:54:06keastesit is running as root
04:55:02 Join HellDragon [0] (
04:55:30mc2739keastes: have you tried to select the ipod manually?
04:56:07keastesyes, but do i need the firmware artition or data partition?
04:56:22 Join antil33t [0] (
04:57:35mc2739keastes: I don't have an ipod, so I am not sure, but I think you need to select the data partition
05:03:48 Join robin0800 [0] (
05:06:24keastesok, data partion selected, rockbox says error will robinson i honna go fail epicly ok ;)
05:06:42 Quit antil33t1 (Read error: 113 (No route to host))
05:06:46keastesdont ask ive been up waaaaaaaaaay to long
05:10:50keastesterminal says the error was a short read , any help?
05:12:02***Saving seen data "./dancer.seen"
05:14:22 Quit J-23 (Read error: 104 (Connection reset by peer))
05:14:57 Join J-23 [0] (
05:15:10 Nick J-23 is now known as Guest73665 (
05:16:39 Quit LambdaCalculus37 ("Fwump")
05:22:43 Join VashTS [0] (
05:22:52 Part VashTS
05:26:03 Join AaronM [0] (
05:27:17 Join midgey [0] (n=tjross@rockbox/developer/midgey)
05:29:31 Join antil33t1 [0] (i=antil33t@
05:29:38 Quit antil33t (Nick collision from services.)
05:37:20 Part toffe82_
05:46:21CIA-6New commit by jdgordon (r23573): Add a bunch of markers into the tokens enum to make figuring out which token isnt in the bugout output easier. ("FIXME code(80)" is alot worse than ...
05:46:33 Quit Horscht ("Verlassend")
05:47:24kkurbjunjdgordon did you see the message I wrote earlier?
05:47:50JdGordonI havnt looked at any logs since friday morning
05:47:58kkurbjunthe custom status bar appears to crash targets with a remote lcd
05:48:13JdGordonblame kugel :)
05:48:17kkurbjunif I enable the custom statusbar on either the H300 or the Mr500 it segfaults
05:48:25JdGordonin the sim?
05:48:36JdGordonok, ill have a look in a few min
05:48:56JdGordonhave you got a sbs I can try with?
05:48:59kkurbjunall you have to do is set the menu from top to custom and it should crash
05:49:17kkurbjunI also tried loading the sbs that you have for the standard statusbar in svn
05:49:20kkurbjunthat crashes it too
05:50:21JdGordonI tihnk it just crashed on me for the same reason
05:50:26JdGordonh100 sim
05:50:57kkurbjunyeah, I think it may relate to having a remote - it didn't crash on the gigabeat F sim
05:50:58JdGordonyay, null reference
05:51:42JdGordonits trying to update the bar before its even loaded :/
05:51:59JdGordonbecause the load order is fucked
05:51:59kkurbjunI have a question on the %Vi tag by the way - does it have an effect on the lists similar to the cfg setting?
05:52:09kkurbjun:), nice
05:52:32JdGordonit does.. that was the big argument... read the comments at the top (?) of viewport.c for an explanation
05:52:58kkurbjunI looked through the dev mailing list you and kugel had - ahh, I'll take a look at that - I wasn't quite sure the purpose fo the two
05:53:18JdGordonif both exist, the list is put in the overlapping rectangle
05:53:54kkurbjunahh, interesting.. why would the user want to use the cfg setting though
05:54:08*JdGordon is not having this argument again :)
05:54:16JdGordonI always thought that jhaving both was pointless
05:54:30kkurbjunhaha.. ok
05:55:00*JdGordon wonders why the hell the remote statusbar is ignoring the remote statusbar sertting!
05:55:09kkurbjunI kind of understood what kugel was saying, but I don't see why a user would want to set a list indipendent of the skin
05:57:16 Join AEnima1577 [0] (
06:00:20kkurbjunI guess kugels arguement is that there should still be a setting for themes that do not have an sbs associated with them.. at least that's how I am interpreting that now - I don't see how supporting a .cfg option is useful now that there is a better alternative - at least that's my impression
06:01:57JdGordonit does mean that if you dont want a sbs, you can still get the ui viewport where you want it
06:02:03kkurbjunI would rather the setting be done only with the %Vi tag rather than worrying about the cfg setting unless there is an underlying purpose for it other than allowing people to arbitrarily specify a window for lists independent of the theme.
06:02:30kkurbjunsorry, not theme, but skin
06:02:56JdGordonmy view is that skins shouldnt be seperate from themes, and thinking that mixing and matching would work is silly
06:03:08JdGordonbut thats just me, and I relaly dont want to have this argument again :)
06:03:35CIA-6New commit by jdgordon (r23574): fix red.. markers need to be outside #ifdefs
06:03:37kkurbjunahh, I ddin't realize there was a distinction between themes and skins - I was just trying to use the proper terminology
06:03:55JdGordontheme is the .cfg, skins are the .wps or .sbs etc
06:04:05JdGordontheme is made up of skins and other options
06:04:25kkurbjunI would agree that they should be packaged as one
06:05:24JdGordonin fact we force that on the theme site
06:06:04kkurbjunI can't think of a "theme" that I would want to use independent of the associated skin
06:06:27kkurbjunwhen I created that one theme/skin I set the theme settings to be appropriate for the skin
06:06:55kkurbjunI don't know how else a themer/skinner would be expected to use/create the files
06:10:20kkurbjunJdGordon: what happens if the ui setting in the cfg and the %vi do not overlap?
06:10:35kkurbjunin terms of the lists?
06:10:36JdGordonthen you get the full screen
06:11:15kkurbjunahhh.. interesting
06:11:46JdGordonor at least, thats what it did at the begining of the week :)... it might have been changed
06:12:14kkurbjunintuitively I would not have expected that, but I guess it's better than not showing anything - that seems like it would just break the sbs
06:12:44JdGordonyeah, there is no right answer there
06:13:02JdGordoneek... that condition is undefined right now apparentrly...
06:13:08JdGordonI thought kugel fixed that
06:13:11*JdGordon fixes
06:13:56kkurbjunso, I'm looking at the mailing list thread, it seems like you feel this should be a cfg setting, not a %Vi setting?
06:14:21JdGordonthats where I started
06:14:38JdGordon%Vi is going to be needed for the rec skins :)
06:15:35kkurbjunSo is there some hidden/underlying need for the cfg setting?
06:15:49CIA-6New commit by jdgordon (r23575): Fix the error condition to actually set the fullscreen viewport. Also remove a comment which doesnt really add anything.. the comments 15lines above ...
06:16:54JdGordonit comes down to this (I think)... 1) user doesnt have sbs but wants to position the lists, 2) user has a sbs with %vi but wants a smaller area (!)
06:17:25 Quit Strife89 ("Bed.")
06:17:36JdGordongetting rid of the setting makes more sense to me now than getting rid of Vi.. but i dont care anymore.. I got the consistancy I wanted
06:18:39kkurbjunYeah, in case one the user could just create a sbs with the %vi tag since (I believe) the intent is to completely remove the old builtin statusbar
06:19:28kkurbjuncase 2 seems like a pointless usecase
06:20:54 Join shai [0] (n=Shai@
06:21:55kkurbjunalso case 2 seems like it could cause some nasty side effects if you loaded a theme/skin that sets a UI viewport and then load another theme/skin that does not account for the cfg setting
06:22:38kkurbjunI mean the first theme/skin sets a ui viewport with the cfg setting
06:22:55JdGordoncorrect again :)
06:23:25JdGordonI'm hoping one way or another there is a riot when the current bar is forcibly removed and these start annoying people
06:24:41JdGordonsetting either of the statusbar settings to custom without a sbs loaded is not a very clever idea :)
06:25:26kkurbjunyeah, I read the conversation between you and kugel before, but I didn't have a context for the conversation since I never tried to dive into the details of using an sbs - then I tried to understand it and got confused :-D
06:26:06JdGordonalso, the outcome happened in IRC not the ml... so its probably hard to follow
06:26:06kkurbjunafter understanding it it does seem like a bit of a supportability timebomb
06:26:29JdGordonfeel free to start the ml thread again :)
06:27:14 Quit TopyMobile_ (Remote closed the connection)
06:28:15kkurbjunso, in terms of underlying code is the cfg setting and the %vi setting functionally equivalent, or are there exceptions?
06:30:14JdGordonthey are now equivilant, before (at the time of the argument) they wernt.. sort of
06:39:48 Join ShapeShifter499 [0] (
06:41:59JdGordonkkurbjun: I dont know if this changes anything, but %Vi should only actually mean anything if that skin is "active"... that means nothing now, but my plan is that the rec screen uses %Vi to specify were its internal list gets displayed (so ignoring the user setting entirely)
06:43:20CIA-6New commit by blue_dude (r23576): Playback: removed duplicate functions
06:45:06CIA-6New commit by blue_dude (r23577): pcmbuf: refactoring, renamed confusing variables, moved some code around, but still no functional changes
06:45:07kkurbjunJdGordon: how will that work with a custom sbs and a recording screen skin?
06:45:32JdGordongood question :) the rec one will have full priority
06:45:37JdGordoni tihnk :p
06:46:10CIA-6New commit by blue_dude (r23578): Removed unneeded includes
06:46:38kkurbjunso will the sbs still show on the recording screen?
06:46:48JdGordonif the user wants it
06:47:54JdGordonalso, what does %we/wd actually mean now?
06:48:05JdGordonthey used to mean enable/disable the statusbar... but now?
06:48:17JdGordonnow being the time when the bar is removed entirely?
06:48:18kkurbjunas far as I would know htat has not changed
06:48:51kkurbjunI guess it would mean enable if a statsubar is available
06:51:04kkurbjunI guess I didn't think about multiple skins case.. I could see users wanting to mix a recording screen skin with a wps skin - the reason being that someone who primarily uses the player for listening might make a really soild wps theme, but neglects the recording screen, or vice versa.. is it going to be possible for a user to mix a wps skin and a recording skin?
06:51:23kkurbjunI guess an sbs is really a third skin correct?
06:51:46JdGordonsbs is hopefully going to be the only one that can be mixed with other skins
06:51:51JdGordonone at a time ever
06:52:01JdGordonrec and wps doesnt make sense together
06:52:31kkurbjunyeah, they would show one at a time, but are effectively loaded together though right?
06:52:51JdGordonI guess it in theory would be possible to create a single skin which has *everything* in it, but it would be painfully hard to modify
06:53:00JdGordonyes, all are loaded together
06:53:27JdGordonand ys, in theory any amount can be displayed on the screen at any time, but right now only the sbs can be updated in the background
06:54:17kkurbjunthe reason I'm asking is that with the sbs functionality which includes things like album art (in my understanding) it could make for some strange displays/behavior when you have the sbs and the recording screen showing at the same time
06:55:00JdGordonwhich is why it doesnt really make sense to split skins from themes
06:55:15JdGordonsbs A cant be expected to work with wps B and rec C and fm D
06:56:12 Quit StealthyXIIGer (Read error: 60 (Operation timed out))
06:56:14kkurbjunwhat would happen if album art is enabled in the sbs while you are in the recording screen?
06:57:05kkurbjunthe reason I ask is if those conditions are not automatically handled it could make it difficult to effectively use the sbs while music is playing when considering the recording screen
06:57:32JdGordonit wont crash... thats the main thing
06:57:48kkurbjunI mean if you are listening to music you might want the album art to show - if you are recording you definately would not want anythign showing
06:58:02kkurbjunanything related to albumart that is
06:58:04JdGordonyou'd have to use the playback mode conditional
06:58:04CIA-6New commit by blue_dude (r23579): pcmbuf: fix yellow. More renaming.
06:58:09JdGordonor the current screen conditional
06:58:24JdGordonwhatever happens it will be a mess
06:59:22kkurbjungotcha, so most of the burden will be on the theme author, but they will be able to account for the current screen in the sbs
07:00:16kkurbjunif they can account for the current screen in the sbs and use that to lay things out differently I don't necessarily see a problem
07:00:19JdGordon%cs is the tag i tihnk
07:01:36JdGordonanother thing which I'd love to be able to figure out is doing dynamic viewport placement... i.e I want the ui viewport to be different in regular lists and in context menus
07:01:36kkurbjunyeah, I can't think of a really clean way to support a custom statusbar independent of the theme/wps skin/fm skin/recording skin offhand
07:02:27kkurbjunoh that would be interesting
07:02:52kkurbjun:), I think you should integrate lua and use that for the skinning ;)
07:02:57 Quit keastes (Read error: 60 (Operation timed out))
07:03:02JdGordonactually... hmm.. that could possibly done using conditional viewports alsop
07:03:12JdGordonhow hard would that be?
07:03:55*JdGordon thinks out loud
07:04:42JdGordonall it would need is another tag which hold the viewport label of the current UI viewport from the skin, that of course couldnt change except when switching between screens... but it could be done
07:04:58JdGordonthen fix up %cs to get more fine grained values
07:05:46kkurbjunso you would be able to specify something like %Via or something similar?
07:06:38JdGordonI tihnk it would actually be pretty simple to do
07:07:49 Join StealthyXIIGer [0] (
07:08:31 Quit liar (Client Quit)
07:10:21 Nick drf|away is now known as drf|laptop (
07:12:04***Saving seen data "./dancer.seen"
07:12:45JdGordonarg.. stop distracting me! I really wanted to get the rec skininng going tonight :p
07:13:03JdGordonarg... damn segfault needs fixing though :)
07:13:47kkurbjun:), sorry
07:16:14 Quit robin0800 (Remote closed the connection)
07:16:46*JdGordon is apparently a moron :)
07:17:57CIA-6New commit by jdgordon (r23580): fix segfault on remote targets where no sbs is loaded for the remote
07:18:28JdGordonthat should fix your crash
07:19:38*pixelma wonders if there is anything in the same area which would explain why early USB is broken *again* at least on Archos
07:20:48kkurbjunjdgordon - it doesn't segfault anymore, but now the main statusbar setting appears to have an effect on the remote statusbar
07:20:59pixelmait was broken in r3.4, now broken again in SVN (different error address)
07:21:06kkurbjunand I keep getting weird effects with the statusbar set to top/bottom now
07:21:50pixelmaerr... forgot that it was fixed at some point between 3.4 and now
07:22:25JdGordonkkurbjun: what sort of wierd effects?
07:22:30JdGordonis that with that change?
07:22:52kkurbjunJdGordon: it doesn't seem to be paying attention to the statusbar with the lists anymore
07:23:30pixelmaJdGordon: and for what it's worth - I don't see a filetype array full on my Ondio (not until yesterday's r23566)
07:23:31kkurbjunJdGordon: I am not sure, I thought I was mostly up-to-date till the commits later in the day
07:23:45 Join keastes [0] (
07:24:03JdGordonpixelma: whas a filtype added yesterday?
07:24:46kkurbjunJdGordon: when I set the remote statusbar in the mr500 sim it shows the statusbar for a second and then it disappears
07:25:23JdGordonok, slightly wrong fix
07:25:43pixelmaJdGordon: no, but you claimed last week to have that problem on your Recorder and since no-on complained "no-one would use hwcodec" anymore. I'm just saying that I don't see it, so can't complain about this one
07:25:56kkurbjunif I load a custom statusbar it seems to heed the %Vi setting, but the builtin statusbar does not work properly
07:27:06CIA-6New commit by jdgordon (r23581): try that previous commit again...
07:27:15JdGordonnow its fixed!
07:27:36JdGordon... and note there way no question mark on the end of that!
07:31:07kkurbjunJdGordon: that seems to fix the standard status bar with the main screen and the remote, but if I set the main screen to custom, and the remote to top the remote does not display the statusbar
07:31:26JdGordonoh come on....!
07:32:18JdGordonwow, that is fucked!
07:32:28kkurbjunthe builtin statusbar works properly for both if I use top top/bottom/off settings, but if I mix in custom with the remote or with the main screen it causes some strange behavior
07:33:25 Quit bluebrother (Nick collision from services.)
07:33:27 Join bluebroth3r [0] (n=dom@rockbox/developer/bluebrother)
07:35:40drf|laptopHmm, so guys... there's not a way to have one font for menus and one font for playback, is there?
07:36:33JdGordonyes there is.... there is no way to have 2 fonts which are not the system font though
07:37:12kkurbjunJdGordon: I think I found another bug, but it's unrelated to the crashing - line scrolling does not appear to work with the %Vi setting in the lists
07:37:14drf|laptopwell I like having a bigger font but it totally messes up the theme
07:38:04kkurbjunthat does seem to work actually
07:38:14JdGordonyeah, works here OK
07:38:59JdGordonpixelma: will there be a jihad out on me if the inbuilt bar is removed without fixing the rec bar first?
07:39:42pixelmadoes someone else see lines that are scrolling when entering the USB screen stay there and keep on scrolling?
07:39:52drf|laptopAlbum Art Extracter for RockBox <−− Will that work for iTunes tracks?
07:39:57 Quit HBK (Read error: 148 (No route to host))
07:40:10pixelmaJdGordon: I guess so...
07:40:18JdGordondamn :p
07:40:57 Join Heather [0] (
07:42:29HeatherHello. I have a Sansa e250 I just updated to the latest rockbox version. I don't think it uninstalled all the way before the install I just did. How would I go about wiping it all out so I can redo the install to be sure?
07:45:59JdGordonwhats the actual problem?
07:46:13Heathernone, really, I guess.
07:47:37 Quit reid05 (Read error: 104 (Connection reset by peer))
07:49:50 Quit antil33t1 (Read error: 110 (Connection timed out))
07:53:24CIA-6New commit by blue_dude (r23582): pcmbuf: eliminate add_chunk as a separate function
07:54:02 Quit StealthyXIIGer (Read error: 104 (Connection reset by peer))
07:54:18 Join StealthyXIIGer [0] (
07:54:38JdGordonanyone know of any tokens which arnt valid on remotes?
07:54:43JdGordonbut are valid on the main lcd
07:55:48 Quit jfc^3 (Read error: 104 (Connection reset by peer))
07:56:02 Nick drf|laptop is now known as drf|away (
07:56:08JdGordoncrap... the viewport tag is different between grey and b/w :(
07:56:11 Join jfc^3 [0] (
07:58:44pixelmaand colour
07:59:00 Join antil33t [0] (
07:59:37JdGordonpulled off the colours and it still doesnt like the %V lines
08:00:31 Join antil33t1 [0] (
08:00:39 Quit antil33t (Nick collision from services.)
08:01:40JdGordonERR: Limits exceeded !!!
08:02:37pixelmaJdGordon: tokens not valid on remotes but on the main screen is a bit dependent on the remote/main target combination. E.g. backdrop tag and album art are invalid on the Iriver remote (monochrome) but not on their main targets
08:03:03 Quit StealthyXIIGer (Read error: 60 (Operation timed out))
08:07:27pixelmaon the Iaudio remote both are allowed (even though %C doesn't work correctly yet)
08:07:54JdGordonok, thanks
08:08:22HeatherI can't get my sansa e250 with the latest rockbox to connect in multimedia mode to my linux install
08:08:26CIA-6New commit by jdgordon (r23583): redo these bitmaps as 1bit bmp's so they show correclty on all displays
08:10:26 Join antil33t [0] (
08:10:40CIA-6New commit by jdgordon (r23584): very initial recording sbs support... also remote sbs using the same skin code
08:10:54 Join stoffel [0] (
08:16:12 Join Zagor [242] (n=bjorn@rockbox/developer/Zagor)
08:16:50 Quit BHSPitLappy (Remote closed the connection)
08:19:55Heatherhas anyone gotten the latest rockbox on the sansa e200 working on Ubuntu?
08:22:40 Quit antil33t1 (Read error: 110 (Connection timed out))
08:23:48 Join AndyI [0] (n=pasha_in@
08:29:31 Join ender` [0] (
08:31:26advcomp2019_Heather, what is wrong
08:31:57HeatherI can't get my sansa e250 to connect to my ubuntu PC while using USB HID
08:35:03 Quit midgey ()
08:35:32 Join flydutch [0] (
08:42:03 Quit AndyIL (Read error: 110 (Connection timed out))
08:43:48advcomp2019_what version of ubuntu because there was a bug about that
08:44:01 Nick advcomp2019_ is now known as advcomp2019 (n=advcomp2@unaffiliated/advcomp2019)
08:47:50Heatherooc 9.04
08:47:54Heathernix the 'ooc'
08:51:27 Nick Guest73665 is now known as J-23 (
08:51:46 Quit Heather ("Bye!")
08:51:47 Join Rob2222 [0] (
08:53:13 Quit stoffel (Read error: 113 (No route to host))
08:57:52CIA-6New commit by jdgordon (r23585): fix the icons so the correct ones are shown
08:57:52 Quit Rob2223 (Read error: 60 (Operation timed out))
09:01:44 Join petur [50] (n=petur@rockbox/developer/petur)
09:03:47 Join maruk [0] (
09:05:59 Join niekie [0] (
09:07:48 Quit ShapeShifter499 ("ChatZilla 0.9.85 [Firefox 3.5.4/20091028153816]")
09:11:40 Join Thundercloud [0] (
09:12:08***Saving seen data "./dancer.seen"
09:19:14 Join B4gder [0] (n=dast@
09:23:57 Join MethoS- [0] (n=clemens@
09:29:46 Quit petur ("*plop*")
09:41:09 Join petur [50] (n=petur@rockbox/developer/petur)
09:48:39 Quit Zagor ("Don't panic")
09:51:48 Join Zagor [242] (n=bjorn@rockbox/developer/Zagor)
09:58:06 Quit Thundercloud (Remote closed the connection)
09:59:31 Quit phanboy4 (Read error: 60 (Operation timed out))
10:05:03 Quit MethoS- (Remote closed the connection)
10:06:46 Join liar [0] (
10:09:01 Quit Zagor (Remote closed the connection)
10:16:34 Join daurn| [0] (n=daurnima@freenode/staff/daurnimator)
10:17:22 Quit daurn (Read error: 60 (Operation timed out))
10:20:24 Join DerPapst [0] (
10:30:45 Join Zagor [242] (n=bjorn@rockbox/developer/Zagor)
10:33:26 Join einhirn [0] (
10:50:57 Join |DaMaGeD| [0] (n=dam@
10:53:56 Quit |DaMaGeD| (Client Quit)
11:02:44 Quit BHSPitMonkey ("Ex-Chat")
11:12:11***Saving seen data "./dancer.seen"
11:41:08TheSevensaratoga: I'm also wondering about some of these decoding times
11:41:19TheSevenIIRC 96KB of IRAM are reserved for codecs
11:41:30TheSevenI may try to further increase that
11:41:58 Join stoffel [0] (
11:42:39gevaertsTheSeven: I don't know if IRAM allocation is based on IRAM size or manual. Maybe it just doesn't use as much as it could?
11:47:04TheSevencould also be possible, I just caught a bug of that kind in mpegplayer yesterday...
11:47:57TheSevenwhat's confusing me though is the extreme difference in decoding time for c4000 and c5000, which is very different to most ARMv4 targets, but indeed also a little similar to PP
11:49:34 Join ender [0] (
11:51:32TheSevenhm, demac seems to have a max. memory usage of 149156 bytes, according to comments and file sizes
11:51:47TheSevenI could possibly manage to fit that completely into IRAM
11:52:38TheSevenis codec code always in iram?
11:53:19gevaertsI don't think so
11:56:58 Quit ender (Read error: 104 (Connection reset by peer))
12:00:56 Part linuxstb ("Leaving")
12:03:51 Join LinusN [0] (n=linus@rockbox/developer/LinusN)
12:07:25 Quit ender` (Read error: 110 (Connection timed out))
12:14:16TheSevengevaerts: the insane filter buffer should already be in iram, and i also moved the code to iram now, but it didn't make any difference
12:17:05 Join at0m [0] (
12:41:56 Join MethoS- [0] (n=clemens@
12:41:59 Join antil33t1 [0] (
12:42:06 Quit antil33t (Nick collision from services.)
12:55:10 Join Tomis [0] (n=Tomis@
12:55:37 Join ender` [0] (
12:58:10 Quit liar (Read error: 110 (Connection timed out))
12:58:19TheSevenhm, ape is the only codec that needs more than 48KB of iram
12:59:56TheSeveni could give them up to 128K, but what should i put there?
13:09:22 Join reid05 [0] (
13:12:12***Saving seen data "./dancer.seen"
13:26:43seaniAfternoon. Could I ask for feedback on an approach to menu options at: ? Ta!
13:34:50 Join dfkt [0] (i=dfkt@unaffiliated/dfkt)
13:35:37 Quit MethoS- (Remote closed the connection)
13:39:45pixelmaI already dislike the "Recent bookmarks" menu item behing switchable (and off by default IIRC)
13:39:55pixelmajust my very first impression
13:41:10pixelmabeing too
13:45:00seanipixelma: This is for support reasons, I guess?
13:53:31*gevaerts thinks that we should go through the NoDo list (and other less official "never!" items) at every devcon
13:53:59gevaertsthe world changes, the developer population changes, so I don't think any feature should be out forever on principle
13:54:08*B4gder agrees
13:54:28gevaertsI also don't think irc or mailing list discussions are likely to be a good way to discuss them
13:54:55 Join teru [0] (
13:55:05 Part LinusN
13:56:33Unhelpfulmailing list discussions are the epitome of rational debate! ;)
14:18:07 Quit drf|away ("MSI laptops FTW!")
14:18:08 Quit keastes (Read error: 104 (Connection reset by peer))
14:34:00 Quit n17ikh (Read error: 60 (Operation timed out))
14:38:37 Join n17ikh [0] (
14:40:07TheSevenrb->pcm_play_data(get_more, NULL, 0);
14:42:40TheSevennow this is really ugly.
14:43:18 Join antil33t [0] (
14:45:27ZagorTheSeven: which part? the api, or the use of it?
14:45:47TheSeventhe use of it, and the corresponding hack needed in the api
14:46:03TheSeventhat's why mpegplayer is failing with the new s5l870x pcm driver
14:46:24TheSevendoesn't a zero-size audio packet usually mean "end of stream"?
14:47:59ZagorI don't see how it is so bad. the api has dual data transfer mechanisms: either send a packet directly, or have the core call a function to get it. if you use a function, there's no point having to spend extra code just to send the first packet in a non-standard manner
14:56:29Zagorand checking parameters 'start' and 'size' for non-zero values is hardly a "hack", is it?
14:56:52 Quit antil33t1 (Read error: 110 (Connection timed out))
14:59:12TheSevenhm, it works now, but audio quality is very poor
15:00:15 Nick fxb__ is now known as fxb (
15:01:32 Join jgarvey [0] (
15:02:36 Part B4gder
15:04:16 Join Horscht [0] (n=Horscht2@xbmc/user/horscht)
15:06:10 Join B4gder [0] (n=dast@
15:08:54TheSevenhow big are the pcm chunks mpegplayer throws at the pcm?
15:09:40 Join hebz0rl [0] (
15:10:46 Join evilnick_ [0] (i=0c140464@rockbox/staff/evilnick)
15:11:17ZagorI don't know
15:12:16***Saving seen data "./dancer.seen"
15:18:08 Join midgey [0] (
15:25:29 Quit DerPapst ("Leaving.")
15:26:24CIA-6New commit by teru (r23586): keyboard: some random cleanup.
15:30:20 Join jus__ [0] (
15:56:33 Join Omlet05 [0] (i=omlet05@
15:56:46 Quit AEnima1577 ("Leaving.")
15:58:15 Join Lss [0] (n=Lss@
15:58:32 Quit jus__ (Read error: 113 (No route to host))
16:02:05saratogaTheSeven: most codecs only use 48k of IRAM because thats all the original SWCODEC targets had, but some have since been changed to use 64 or 80KB of IRAM (Vorbis, WMA)
16:02:21saratogausing more then that provides relatively little benefit on most codecs
16:02:30TheSevensaratoga: they don't for S5L870x, although I can't see why
16:02:41saratogayou've checked the .map files?
16:03:04TheSeveni just set the codec iram size in the linker script to 48k and it still links fine
16:03:18 Quit ender` (" The better I get to know men, the more I find myself loving dogs. -- Charles de Gaulle")
16:03:40TheSevenor is there a way to tell the linker to put something only into iram if it fits?
16:03:41saratogaoh i misunderstood
16:03:55saratogayes the checks are actually for CPU model
16:04:15TheSevenhm, i couldn't find them in vorbis and wma though
16:04:17saratogabut the difference here will be small, and it doesn't explain why mp3 for instance is so slow
16:04:42TheSevenhm, mp3 wasn't that awfully slow, there are other things that are much worse
16:04:51 Join Blue_Dude [0] (n=chatzill@rockbox/developer/Blue-Dude)
16:04:54TheSevenwma and ape-c5000 are weird
16:05:06saratogawmadec.h has the check for it
16:05:24TheSevenwhy is c5000 that much slower than c4000 even though the insane buffer *is* in iram?
16:05:44Blue_DudeTheSeven: apparently the max audio frame size in mpegplayer is 1152 samples...
16:06:04TheSevenyes, i've dumped some things at runtime, and most of the packets have that max size
16:06:36TheSevenso i split them into 2560+2048 byte chunks in the driver
16:06:50TheSevenand i can't see why this would be too slow
16:06:56TheSevenbut it's clicking like crazy
16:07:34Blue_Dudempegplayer bypasses pcmbuf, doesn't it?
16:07:36 Join antil33t1 [0] (
16:07:43 Quit antil33t (Nick collision from services.)
16:07:59TheSevenor rather it has its own copy of it
16:08:10saratoga#if (CONFIG_CPU == PP5022) || (CONFIG_CPU == PP5024)
16:08:23saratogafrom wmadec.h for allocating IRAM
16:08:29Blue_DudeI suppose you could jack up the packet size, but might cause sync problems, and still not fix yours.
16:08:33saratogaare you sure the clocks and timer are correct?
16:08:35TheSevenyes, i found it now, but what shall we do about it?
16:08:42saratogaperhaps its just not measured correctly?
16:09:16saratogaand is your IRAM actually implemented as SRAM cells and not internal DRAM like on AMS?
16:09:30TheSevenafaik yes
16:09:54saratogaone simple thing to check would be to time random memory access to it and dram and verify that its actually faster
16:10:05TheSeventhe s5l8720 even has 192K of SRAM and 32M of internal DRAM :-)
16:10:06saratogawe tried this on AMS and to our surprise IRAM and DRAM were the same
16:10:28saratogait turns out the internal IRAM and internal DRAM were more or less the same
16:11:16saratogaits odd that you get such great runtime with such slow codecs
16:11:20TheSevensaratoga: but this doesn't really explain the speed drop between c4000 and c5000
16:11:39saratogaape is like the only codec i haven't read through so i can't really help you there
16:12:25TheSevensome of the codecs seem to be pretty much fine, while others are very slow... hmm. dcache?
16:12:29saratogahuh flac is still very fast for you
16:12:48saratogawhile wma is very slow
16:13:45TheSevenI'll re-check wma with more iram usage in a minute
16:15:38saratogai have to go but i'll try and look into this more
16:15:52 Quit teru ("Quit")
16:19:56 Join jus__ [0] (
16:26:28seaniIf I want to call the talk_* functions what do I need to do in ../tools/configure? I've selected (A)dvanced and (V)oice, but I still get compilation errors.
16:31:20 Join freddyb [0] (n=fred@
16:34:22 Join toffe82 [0] (n=chatzill@
16:35:17jus__is possible to compile rockbox for cowon X5 with an hdd bigger than 137go?like for Archos?
16:36:25TorneI don't see why not, LBA48 should work on anything, really. But I've not used an X5..
16:44:35 Join gb_master [0] (i=82c03212@gateway/web/freenode/x-ovkeyhriaxwcyqig)
16:46:11 Join ender` [0] (
16:48:09jus__sorry, where can i download the sources?
16:49:30jus__with svn
16:49:58B4gderit's not very hard to find
16:50:15CIA-6New commit by theseven (r23587): Fix mpegplayer on S5L870x
16:51:40 Join bmbl [0] (n=Miranda@unaffiliated/bmbl)
16:55:18 Nick YPSY is now known as Ypsy (
17:01:16freddybIs anyone interested in trying out my keyboard patch for scroll wheel models? FS #10763. There's a gif animation on the bottom of the page if you just want to see it w/o patching.
17:01:45 Quit Zagor ("Don't panic")
17:04:32freddybI posted a patch for Fuze(v1) recording on FS #10371. This is an interrupt driven record patch to make due until DMA gets included.
17:12:20***Saving seen data "./dancer.seen"
17:14:50 Join AEnima1577 [0] (
17:20:56 Join StealthyXIIGer [0] (
17:27:50 Quit gb_master ("Page closed")
17:31:59 Quit rvvs89 ("Lost terminal")
17:32:41 Join HBK [0] (
17:33:57 Join rvvs89 [0] (
17:43:20 Join GeekShadow [0] (n=Antoine@reactos/tester/GeekShadow)
17:46:02 Join Rand_Althor [0] (
17:46:26Rand_AlthorDidn't there used to be a iink to the last 12 months of commits?
17:47:50B4gderdid it?
17:48:11B4gderuhm, yes it did
17:48:35saratogamore of those pages would be nice
17:48:48Rand_AlthorDoes that page no longer exist?
17:49:24saratogabut is gone
17:49:51saratoga would be nice
17:49:52B4gderand no since 3.4
17:50:01Rand_AlthorGood enough for what I need. Thank you.
17:50:08 Quit Rand_Althor (Client Quit)
17:52:20 Part B4gder
17:53:00 Quit AEnima1577 ("Leaving.")
17:55:06 Quit toffe82 (Read error: 104 (Connection reset by peer))
18:02:55 Join toffe82 [0] (n=chatzill@
18:04:06 Join AEnima1577 [0] (
18:04:39jus__I compiled a X5 rockbox with the two constantes for archos big disk, it seams to work system menu says DD1:absente (no present)
18:06:32Blue_DudeAnybody here use crossfade much?
18:07:02Blue_DudeI've been working on pcmbuf, and I want to make sure the cleanups haven't changed behavior.
18:07:44 Quit petur ("work->home")
18:07:53 Join n1s [0] (n=n1s@rockbox/developer/n1s)
18:10:27 Quit StealthyXIIGer (Client Quit)
18:11:55CIA-6New commit by blue_dude (r23588): pcmbuf: more refactoring
18:12:13evilnick_Blue_Dude: I'll check when I get home, if there's no report beforehand.
18:13:25Blue_DudeThanks. I'll be checking the logs for problem reports.
18:13:51 Join stripwax [0] (
18:14:15 Quit stripwax (Client Quit)
18:16:25evilnick_Blue_Dude: Silly question, but is it possible to make the crossfade settings less reliant on restarting playback? There are some settings that (AFAIK) shouldn't affect the currently-buffered audio data, but you are still prompted to restart playback
18:17:33Blue_DudeI haven't started delving into the crossfade functions themselves yet, just the track change code leading up to it. What settings are you referring to?
18:19:55evilnick_I'll confirm when I have chance to check, but IIRC: Enable crossfade, then as a song is playing, change the fade in/out duration and you have to restart playback
18:24:30Blue_DudeThe problem with fade out delay/duration is that the pcm buffer size is dependent on those values. So if you change them, the pcm buffer has to change size, which means stopped playback.
18:25:33Blue_DudeIt might also stop playback when you change fade in values, but if so I don't know why.
18:26:28*gitster` is getting sick of carrying FS #10680 and FS #10681 locally in his tree forever, and wishes somebody upstream to commit and close them... kugel? somebody?
18:26:49evilnick_I'll have a play around when I'm able to, but I recall it changing for EITHER fade in or fade out; and as fade-in wouldn't be needed until the next song this seemed suboptimal to me.
18:27:01Blue_DudeIt seems to me that if the pcm buffer must change size, then playback should stop, change buf size, rebuffer, then resume without your input. I'll have to check.
18:27:29Blue_DudeThere's all kinds of goofy stuff going on in playback. This might be one of them.
18:27:56 Join dfkt_ [0] (i=dfkt@unaffiliated/dfkt)
18:28:11evilnick_As a crossfade user, it's sort of annoying, but then I usually stick to one bunch of settings, so only change them if I've reset my settings.
18:28:20gevaertsgitster`: I'd guess JdGordon is the other obvious candidate. I prefer not to do it myself because I really have no clue about the wps code
18:28:21 Quit jus__ ("Ex-Chat")
18:29:28 Quit bmbl ("Bye!")
18:30:31 Join daggett [0] (
18:31:00Blue_Dudeevilnick_: if you're using the track skip setting, does it take a few seconds to start fading in the next track or is it very prompt? It seemed to take too long to start the next track to me. That's one of the areas I'd like to know if there's been an adverse change.
18:31:04*gitster` thanks gevaerts for a pointer.
18:33:28evilnick_Blue_Dude: (From memory) I always thought that the track skip was affected by the crossfade settings unnecessarily - but I'd have to test it out before stating that as fact.
18:33:28 Join domonoky [0] (n=Domonoky@rockbox/developer/domonoky)
18:33:43JdGordonassuming they both work... they look fine, someone commit them
18:34:07evilnick_i.e. If I had the crossfade set to only fade on shuffle, then a manual trackskip should happen immediately, but it didn't seem to.
18:34:08*JdGordon is doing his usual "late for work as its 9.30am already and im still in irc" routine
18:34:16Blue_Dudeevilnick_: ok. I think there's a way to speed up the process, but I'm avoiding making real changes until after cleaning up the code.
18:34:23gevaertsJdGordon: ok
18:34:50JdGordongitster`: CustomWPS on the wiki needs to be updated
18:34:58 Quit GeekShadow (No route to host)
18:35:04*gevaerts will commit
18:35:14Blue_Dudeevilnick_: That's why I'd like to know if there are any changes to the user, because I haven't made any intentionally.
18:36:05 Quit maruk ("Leaving.")
18:40:14 Quit swilde ("ERC Version 5.3 (IRC client for Emacs)")
18:43:48CIA-6New commit by gevaerts (r23589): Allow %?Sp<...> to be used as a conditional ...
18:45:58 Quit dfkt (Read error: 110 (Connection timed out))
18:47:52 Join Aratsu [0] (
18:51:37CIA-6New commit by gevaerts (r23590): Don't define pitch_speed_enum() on MAS3507D. Fixes yellow
18:55:38 Join polobricolo [0] (
18:56:46 Join bertrik [0] (
18:57:19CIA-6New commit by gevaerts (r23591): Implement %Ss for playback speed ...
18:58:13Unhelpfulsaratoga: -O3 has the *worst* performance on quite a few codecs :/
18:58:22gevaertsgitster`: I did have to fix FS #10681 a bit to compile on HWCODEC :)
18:59:01 Nick dfkt_ is now known as dfkt (i=dfkt@unaffiliated/dfkt)
18:59:08n1sUnhelpful: heh
18:59:11gevaertsWPS_TOKEN_SOUND_SPEED depends on SWCODEC, not just on !MAS3507D
19:00:43n1sUnhelpful: that's consistent with what i've seen, O2 is usually faster than O1 but O3 is only rarely faster than O2 (coldfire and pp)
19:02:58n1shmm, should test gcc 4.4 for coldfire
19:05:17n1s3.4.6 is very old
19:06:21n1sand the various improvements to the coldfire gcc *should* mean it generates better code
19:06:41 Join BdN3504 [0] (
19:09:50n1snow the other parts of the compiler likely counteracts that with some overzealous code size sacrafices for tiny speed improvements
19:12:02 Join panni_ [0] (
19:12:22***Saving seen data "./dancer.seen"
19:12:22CIA-6New commit by blue_dude (r23592): pcmbuf: refactoring
19:12:59 Join stripwax [0] (
19:13:38 Quit stripwax (Client Quit)
19:14:10 Join JdGordon| [0] (n=Miranda@nat/microsoft/x-ajvppqrlaerzfsah)
19:17:00 Join DerPapst [0] (
19:18:24TheSevensaratoga, gevaerts:
19:18:38TheSeven(after IRAM definition fixes)
19:19:14TheSevennot that much difference :-/
19:24:43 Quit flydutch ("/* empty */")
19:27:03 Join Biont [0] (
19:28:43 Join phanboy4 [0] (
19:29:06BiontDoes anyone know when the simulators get updated next? I need a CowonD2 simulator for windows but I'm too stupid to build one myself.
19:30:31Biont(It might work on second try, but I thought it might be better to just wait instead)
19:32:00 Join webguest80 [0] (
19:32:36 Quit webguest80 (Client Quit)
19:33:53bertrikI am not aware of some regular schedule for simulator releases
19:37:59 Quit stoffel (Read error: 54 (Connection reset by peer))
19:40:52BiontI would have been surprised if there was a schedule :) What are the usual intervals?
19:41:53pixelmairregular I guess ;)
19:42:58 Join kisuke [0] (i=459030e2@gateway/web/freenode/x-ozcjjmhzlizfvmwx)
19:44:44 Join Nz17 [0] (
19:44:55kisukerunning the auto installer on ubuntu 9.10 i get a short read error when tring to install to an ipod video (5th GEN), how can i not get this
19:45:24BiontI guess noone knows, then. No problem, I just thought I might be lucky
19:46:30 Join Thundercloud [0] (
19:46:43 Quit AEnima1577 ("Leaving.")
19:48:49Unhelpfuln1s: are there call-locality issues on coldfire as on arm? would there be the possibility of a large space savings?
19:49:12 Quit kisuke ("Page closed")
19:54:50JdGordon|gitster`: are you in CREDITS already?
19:55:06gevaertsJdGordon|: I checked :)
19:55:11JdGordon|ok :)
19:57:33 Quit einhirn ("Miranda IM! Smaller, Faster, Easier.")
19:58:41 Join AEnima1577 [0] (
20:00:01 Join Grahack [0] (
20:01:44Nz17Are there any known players that Rockbox runs on that use AA or AAA batteries and have 1GB or more capacity?
20:02:21pixelmaok, it's definitely the custom status bar commit that breaks early USB again on Archos (r23257 works, r23258 doesn't). Since the crash happens in cuesheet marker drawing (if I read the map correctly), I wonder if it is somehow related to the problem with the progressbar width which lets pure %pb fail too
20:03:00 Quit polobricolo ()
20:03:28 Join polobricolo [0] (
20:04:17 Quit polobricolo (Client Quit)
20:05:05JdGordon|more likely is it tries to draw the bar before its loaded.. or decided there isnt one to load
20:05:41pixelmaNz17: the Ondios with a 1GB+ MMC (up to 4GB available), but those only play mp3 (mp2) and wav with a plugin. Other than that I believe the Sansa m200s also use AAA batteries but it's a port in early stages and I don't know how usable it is (I think it hasn't come far yet)
20:06:00pixelmaOndios are probably also hard to get now
20:06:04 Join polobricolo [0] (
20:06:23Nz17@Pixelma: LOL, ya they are from the looks of eBay.
20:06:47Nz17Perhaps I should get a m200s and use it as-is until the port matures?
20:07:25*pixelma looks around for domonoky
20:07:34*domonoky hides
20:09:45domonokythe m200v4 port is still really in a early state. it has the same playback problems as clip because of only 2MB ram, a display bug, and more...
20:09:48pixelmaJdGordon|: your fix today in the morning wouldn't have an influence on this, would it?
20:10:39 Join n00b81 [0] (n=n00b81@unaffiliated/n00b81)
20:10:45domonokyother m200 players are even less developed. (there are 4 distinct hardware versions of these, but all look the same)
20:11:04 Join veeloc [0] (
20:11:07pixelmawhy would it want to draw a progress bar though? There is none in the statusbar (and I don't use a custom one there) and cuesheet support is turned off too
20:11:13pixelmaJdGordon|: ^
20:13:08polobricoloOn arm, In firmware/export/thread.h line 104 and firmware/thread.c lines 806-813 why are r0-r3 and r12 not saved ? i can see r4 is sometimes used and the comments says "using r4 because it is non-volatile" does that mean gcc saves some vars in r0-r3 so they are not safe for asm code?
20:14:14Nz17Hmm, I really don't want to settle for a player that doesn't use AA or AAA batteries since they store so much energy and recharge quickly. Does anyone have any player recommendations even if they don't run Rockbox?
20:14:45polobricolo*lines 212-214
20:14:52 Join Strife89 [0] (n=michael@
20:15:22JdGordon|pixelma: email or flyspray it... or annoy kugel when he shows up.... I'll have a look tonight maybe
20:15:29 Join domonoky1 [0] (
20:15:52Unhelpfulpolobricolo: rockbox is cooperatively threaded, so thread changes only happen when the thread being changed from calls yield. since it arrives at the scheduler via a function call, caller-saved registers don't need to be saved - the thread that is giving up control has already done so.
20:15:55JdGordon|if its really trying to draw the pb or cuesheet then something is really wacked
20:16:36Unhelpfulalso, wait longer next time, or check the logs. i answered this on a few hours after the last time you asked. ;)
20:18:38polobricoloUnhelpful: ok i though it was an irq or something like that that made it switch. that means if a thread never yields, rockbox will never switch ?
20:19:23domonoky1jup, a thread without yields is a bug :-)
20:19:36 Join robin0800 [0] (
20:20:42 Quit domonoky (Read error: 104 (Connection reset by peer))
20:23:37Unhelpfulpolobricolo: that's exactly what cooperative multitasking means - code has to cooperate. ;)
20:23:53polobricolook thanks
20:24:49bertrikpolobricolo, it could also be a sleep or a wakeup_wait
20:25:56Unhelpfulor anything that calls those - don't event queue waits, andall of the button-checking things, do so?
20:29:55Nz17Thanks for the help, all.
20:29:58 Quit Nz17 ("Leaving")
20:33:59 Join petur [0] (
20:34:03CIA-6New commit by blue_dude (r23593): pcmbuf: simplify crossfade logic
20:35:08 Quit polobricolo (Read error: 104 (Connection reset by peer))
20:37:10 Part n00b81 ("Leaving")
20:38:03 Join sansaAMS [0] (
20:40:25sansaAMSdoes "Until the USB driver is fully operational, file transfer support is provided by booting the Original Firmware when the USB cable is inserted." mean that the device willauotmatically switch to the original firmware when in usb mode or do i have to manually add the original firmware in order to boot it up?
20:40:54domonoky1sansaAMS: it should switch automagically
20:41:15sansaAMSif it does i should add it to my sansa right now
20:41:47evilnick_sansaAMS: It will automatically reboot into the OF
20:42:02sansaAMScause i've been waiting months to get rockbox on this thing, it would really solve alot of the issues i have
20:42:12 Join liar [0] (n=liar@
20:43:03sansaAMSso upon removal it switching back to rock? if so i'd say it is working fine enough, i don't use recording anyway>
20:45:40domonoky1just try it. you can always uninstall it, if you dont like it..
20:48:49sansaAMSyeah im downloading it now. a lot of these mp3 producers should learn to make their stuff more flexibile in the first place its just good for business, apss are dominating the multi-media area.
20:52:14 Quit robin0800 (Remote closed the connection)
20:52:53 Join JohannesSM64 [0] (
20:53:47JohannesSM64with the 64-bit linux binary, it's not seeing any network connection, though my network is working perfectly. common issue?
20:55:26JohannesSM64erm, now it's working :S
20:55:54JohannesSM64it still complained about non-working connection, but the download worked
21:00:45 Quit sansaAMS ("CGI:IRC (EOF)")
21:01:14CIA-6New commit by theseven (r23594): Make the codecs use more IRAM on S5L870x, as we have plenty of it.
21:06:18BdN3504i got an ipod 5.5g with a dead hard drive. i replaced the hard drive toshiba mk8010gah
21:06:34BdN3504i have plugged the ipod to my windows computer
21:07:00BdN3504a device is recognized, but i only have the option to format it
21:07:16BdN3504when i click on format device, the formatting screen pops up
21:07:38BdN3504but the funny thing is, the capacity of the device is listed with 37,2 gb
21:07:55TheSevenhm, partitioned in a weird way?
21:07:59evilnick_BdN3504: Use Swissknife
21:08:19evilnick_It could be that Windows limits it to 40GB as it's a Primary FAT32 partition
21:08:59BdN3504i have only blank spaces in the spots where i am normally able to select the filesystem
21:09:23Unhelpfulha ha windows, you so crazy :/
21:09:39BdN3504i got the rockbox vmware running in the backgorund
21:09:46BdN3504can i fix it with debian?
21:10:16evilnick_BdN3504: You'll need to use a disk partitioning utility that is not built-in to Windows to partition/format it correctly.
21:12:23***Saving seen data "./dancer.seen"
21:13:10 Join stoffel [0] (
21:14:18Unhelpfulif vmware permits access to whole-disk devices, then you can probably format it from vmware. otherwise, there are a number of 3rd-party windows format utilities available for fixing this specific problem.
21:15:01Torneif you're in windows anyway then "just use itunes" is an option
21:15:16BdN3504cool, thanks. i thought i'd bought a cheapo chinese fake
21:15:34Torneitunes will just do all the stuff right for you ;0
21:15:41BdN3504i'd rather refine from using itunes
21:16:57BdN3504hm, i get two consecutive read/write errors when i select the device in swiss knife, then it shows up normall
21:17:37BdN3504or not... i can't do anything with it.
21:17:52BdN3504i guess i'll use itunes... thanks torne.
21:18:51TheSevenif you get i/o errors, the driver is probably bad
21:19:13evilnick_Yeah, iTunes will probably sort it out, although it is rather awful (imho)
21:19:14Torneor it's doing unaligned sector accesses on one of the drives that doesn't support it
21:22:09BdN3504this sounds horrible... i hope it's going to work. f*cking itunes is 89 megs
21:22:26TheSevenhm, was the mk8010gah that nasty drive?
21:23:33Tornei dunno, but it's one reason
21:24:30 Join jus__ [0] (
21:27:03n1sUnhelpful: i don't *think* so but not very up to date on my coldfire asm and wasn't that good at it when i was :)
21:27:59 Quit BdN3504 ("CGI:IRC")
21:28:28n1sone decent space save with newer gcc is that it knows the correct cost of mult on the emac equipped cpus so it actually uses mult instructions instead of larger and sometimes even slower shit and add
21:28:47n1seh s/shit/shiFt/
21:30:36nawksthere are a plethora of gcc options that affect speed vs. size optimizations
21:31:43 Join stripwax [0] (
21:32:50n1snawks: yes, but if the compiler thinks a mult costs 14 (iirc) cycles instead of 2 (iirc) it will use shift and add sequences in a lot of places as it thinks that will be faster
21:33:44 Quit hebz0rl ("Verlassend")
21:35:11 Join hebz0rl [0] (
21:37:09 Quit jus__ (Remote closed the connection)
21:38:35 Quit Thundercloud (Remote closed the connection)
21:40:26 Join Juice^ [0] (
21:40:34 Quit Juice^ (Remote closed the connection)
21:41:19 Join Stephen__ [0] (
21:41:21 Join balug_ [0] (
21:45:45 Quit AEnima1577 ("Leaving.")
21:46:40 Join bmbl [0] (n=Miranda@unaffiliated/bmbl)
21:51:14 Join AEnima1577 [0] (
21:54:15 Quit stoffel (Remote closed the connection)
21:58:54 Join robin0800 [0] (
21:59:33 Join Tomis2 [0] (n=Tomis@
22:02:12 Quit stripwax ("")
22:04:39 Quit Tomis (Read error: 104 (Connection reset by peer))
22:04:39 Nick Tomis2 is now known as Tomis (n=Tomis@
22:04:44 Join addisonj12 [0] (
22:05:22addisonj12hey, trying to install rockbox onto an old sana 260 from win 7, problem is, windows 7 does not give the player a drive letter
22:05:26addisonj12anyone know how to change that?
22:05:44n1saddisonj12: is the player in msc mode?
22:05:45JdGordon|get it out of MTP mode
22:06:03n1saka follow the installation instructions ;)
22:07:49 Join liar_ [0] (n=liar@
22:08:23 Quit liar (Nick collision from services.)
22:08:35 Nick liar_ is now known as liar (n=liar@
22:09:19*addisonj12 feels silly
22:09:20 Join grant__ [0] (i=d83bf1dc@gateway/web/freenode/x-hkxhgdxufavzyzgo)
22:09:26Stephen__how come theres no icon on themes for ondavx747 ?
22:10:42grant__hey liar, the latest version of rockbox hasnt been working for me and TheSeven told me to ask you if you have a version that is working
22:12:08TheSevenyou might want to try theseven/releases/rockbox/">
22:12:12Strife89Unfortunate nickname. :(
22:14:41 Quit Blue_Dude ("ChatZilla 0.9.85 [Firefox 3.5.5/20091102152451]")
22:16:23UtchybannTheSeven: Hi. Your mpegplayer fix works on my nano too. Thanks.
22:16:59 Quit Biont ("ChatZilla 0.9.85 [Firefox 3.5.5/20091102152451]")
22:17:12TheSevenUtchybann: yeah, that was quite an ugly one
22:17:37TheSeveni needed to dynamically decide how i split the dma transfers in order to make it work for both pcmbuf-based audio and mpegplayer
22:18:36 Join stripwax [0] (
22:19:52 Quit Grahack ("Leaving.")
22:21:52 Quit AaronM (Read error: 104 (Connection reset by peer))
22:22:12 Join AaronM [0] (
22:23:13grant__yah, the fix2 works but the mpeg playing still doesnt work
22:23:56 Part veeloc
22:24:49Utchybanngrant__: I guess fix2 is not supposed to fix mpeg playback.
22:24:59 Join Thundercloud [0] (
22:25:38Utchybanngrant__: mpeg playback is fixed in r23587.
22:27:19 Quit grant__ (
22:27:19 Quit evilnick_ (
22:27:19 Quit saratoga (
22:27:23Utchybanngevaerts, TheSeven: can you look at FS #10656. Now that nano2g port is very usable, I guess we can apply cosmetic fixes.
22:28:08NJoingrant__ [0] (i=d83bf1dc@gateway/web/freenode/x-hkxhgdxufavzyzgo)
22:28:08NJoinevilnick_ [0] (i=0c140464@rockbox/staff/evilnick)
22:28:27UtchybannI have tested it on my mini2g and size are correctly reported.
22:28:50TheSevendoes it catch all of the incorrect sizes?
22:29:05Unhelpfuln1s: i sometimes wish i could better inform gcc about possible ranges of operands. there's really no way to tell it "you can use a 16-bit multiply instruction" that might not cause it to add a sign-extend instruction. :/
22:29:30UtchybannTheSeven: where are incorrect sizes reported ?
22:30:37 Quit freddyb (Remote closed the connection)
22:30:43TheSevenno idea, but there were multiple places I think
22:30:48TheSevenjust flick through those info menus
22:31:01nawksn1s: i guess i don't know what emac means in this case. i see that some arms have EthernetMAC built-in
22:31:17stripwaxenhanced multiply and accumulate
22:31:38UtchybannTheSeven: it fixes System->Rockbox Info and Debug Menu->View disk Info.
22:32:29grant__yah the fix2 works for the fiel browsing but not for the mpeg playing
22:32:29 Part grant__
22:32:29 Join grant__ [0] (i=d83bf1dc@gateway/web/freenode/x-hkxhgdxufavzyzgo)
22:32:29 Quit grant__ ("Page closed")
22:32:29 Join grant__ [0] (i=d83bf1dc@gateway/web/freenode/session)
22:32:55*Unhelpful has never actually even *seen* a compiler-emitted emac instruction, though he's only looked at very little coldfire asm, and most of that hand-coded.
22:33:10UtchybannTheSeven: Debug->Partition Info is still wrong. I will catch this one.
22:33:14 Join stripwax_ [0] (
22:33:38Unhelpfulsaratoga: i never saw your benchmark results. whenever they're ready, let me know (assuming i haven't just missed them).
22:33:55 Quit stripwax_ (Client Quit)
22:34:51grant__so the fix2 does fix the file browing but not mpeg playing
22:35:06n1sUnhelpful: it doesn't use the emac instructions but coldfires with emac have much faster regular multiplies too (they use the emac unit internally iiuc)
22:35:51n1sand gcc 3.4 doesn't know our coldfire models so wwe use as close a match as we can but that has ni emac
22:36:37 Quit grant__ ("Page closed")
22:37:12Unhelpfulnawks: the EMAC is a subunit of some coldfire CPUs (i think *all* that rockbox runs on) which has its own set of registers which can can be used as output of special multiply/accumulate instructions. if i recall correctly a mac instruction takes a single cycle, and obviously if you're doing something like a*b + c*d the ability to do the addition with the same instruction is a huge benefit as well.
22:37:53nawksvery helpful Unhelpful :)
22:38:40Unhelpfuln1s: ah. the call locality issue i refer to with some ARM models has to do with the two ways one can branch on ARM, and the memory layout on some of our targets. if the branch target is very close, a function call, aside from register saving and parameter loading, is a single branch-and-link instruction.
22:39:29Unhelpfultoo far away (>+/- 32MB, i believe) and you have to load an address from memory, copy the pc to lr explicitly, and use a branch-to-register instruction to jump to the address
22:40:08 Join Tomis2 [0] (n=Tomis@
22:40:13Torneno, just two instructions when you're ARm only
22:40:20Tornemov pc, lr; ldr pc, =target
22:40:30Tornesince you can load PC directly.
22:40:37Torneyou only need three isntructions for ARMv4T
22:40:44Unhelpfuli wouldn't be surprised if there exists a similar situation on cf, i know that most instructions have rather limited range for immediates.
22:41:35UnhelpfulTorne: hrm, and then only if you *might* be calling a thumb function? the thing is, when i've thrown gcc simple test cases, it seems to always do things the hard way.
22:42:02TorneYeah, that's for interworking
22:42:21TorneARMv5 and later obey the lsb for determining thumb on pc loads, so the above works
22:42:29Tornebut ARMv4T requires that you bx
22:42:56Torneadd lr, pc, #4; ldr ip, =target; bx ip
22:42:59UnhelpfulTorne: that would explain, i've mostly been testing on e200 :)
22:43:11Tornewe don't use thb though do we?
22:43:31Torneif we're using ARM only you should have the flag to disable interworking :)
22:43:42Unhelpfuli don't *think* we do...
22:44:02Tornebut yah, arm is quite happy loading directly to PC
22:44:30TheSevenhave we ever tried to do?
22:44:36TheSevenwould it make sense?
22:44:50Tornethere's an FS somewhere trying it but it breaks the code, iirc
22:44:56Torneand yes, it would make sense perhaps
22:45:04Tornethumb is a complicated tradeoff
22:45:23Unhelpfuland in any case, on EABI gcc *always* emits bl <label> or blx <label>, and then if label is too far at link time it reroutes the call to a stub address with a load-to-pc followed by address :)
22:45:42Tornethat being the delicious part
22:45:52Torneproper working calls
22:46:24Unhelpfulggc-4.4.1 on beast (which has no long calls) still manages a lovely 1.5KB or so binsize reduction. on e200 its 63KB :)
22:46:28TheSevenwhat are things like glue7 good for if it doesn't do that?
22:46:59Tornethere's an epic number of fixups gcc can do for ARM weirdness
22:47:11Tornefor things like rewriting branches to accomodate armv4t interworking
22:47:15Tornemany of them only work on EABI
22:47:42 Quit Tomis (Read error: 110 (Connection timed out))
22:47:42 Nick Tomis2 is now known as Tomis (n=Tomis@
22:48:06 Join kugel [0] (n=kugel@rockbox/developer/kugel)
22:49:21nawkshas anybody bought a refurb e260/270/280 from compusa recently?
22:49:33nawksi wonder if they are likely to be v1 or v2
22:52:13krazykitnawks, it's pretty impossible to tell unless someone has physical access to the functioning device.
22:53:27nawksdo v1 280s exist?
22:53:42nawksthat is the 8G model
22:55:36 Quit stripwax (Read error: 110 (Connection timed out))
22:55:51 Join LambdaCalculus37 [0] (n=LambdaCa@rockbox/staff/LambdaCalculus37)
22:58:03 Quit bubsy ()
22:58:05 Quit kugel (Remote closed the connection)
22:58:42amiconnUnhelpful: cf has support for 32 bit immediates... no problem there
22:58:42 Quit AEnima1577 ("Leaving.")
22:59:37nawkswhat are some devices with coldfire?
22:59:40amiconnOf course those aren't overly efficient when used back-to-back - too many instructions using extension words in sequence will starve the pipeline
23:00:04amiconnIriver H1x0 series, H300 series; iAudio M3, M5 and X5
23:00:20 Quit JohannesSM64 ("Leaving")
23:02:39 Quit robin0800 (Remote closed the connection)
23:03:52 Quit balug_ ("Ex-Chat")
23:04:35 Join AEnima1577 [0] (
23:04:39 Quit bmbl ("Bye!")
23:04:58 Join FuzeV2 [0] (
23:05:11FuzeV2hello everybody
23:05:29FuzeV2I have a little question
23:06:20FuzeV2I have seen that a recovery mode has been found for ams sansas by shorting some pins
23:06:38FuzeV2has anybody tried doing the same for fuze v2s?
23:06:44 Join Tomis2 [0] (n=Tomis@
23:09:57 Join saratoga [0] (i=9803c6dd@gateway/web/freenode/x-ktupxllmowxfxavh)
23:10:34saratogaUnhelpful: there was almost no difference except for Wavpack and I fixed that bug in SVN already
23:12:26***Saving seen data "./dancer.seen"
23:12:29saratogacompares O2 and default
23:12:47 Quit midgey ()
23:14:57TheSevensaratoga: did you see my paste? it doesn't seem like using more iram changes much
23:15:01Unhelpfulsaratoga: ah, ok. in that case flac (with O2) can actually beat gcc-4.0.3 by a little bit on eabi. the other codecs are mostly 0-4% slower or faster
23:16:52Unhelpfuli would say that with correct performance tuning codecs should not be a blocker for a move to gcc-4.4.1 eabi on arm... but i should probably see results on more targets first.
23:17:23 Nick Ypsy is now known as YPSY (
23:18:26 Join flydutch [0] (
23:19:23 Join Tomis2_ [0] (n=Tomis@
23:21:54 Join Tomis2__ [0] (n=Tomis@
23:22:46 Quit Tomis (Read error: 110 (Connection timed out))
23:23:13FuzeV2aren't there any pcb scans for clipv2?
23:25:19Dhraakelliandoes Rockbox no longer support BMP albumart?
23:25:33 Quit Tomis2 (Connection timed out)
23:25:35 Quit petur ("Zzzz")
23:25:40nawkshi Dhraakellian
23:25:52 Join kugel [0] (n=kugel@rockbox/developer/kugel)
23:26:10pixelmaDhraakellian: it still does
23:26:31Dhraakellianhmm... I must be doing something wrong then
23:26:32 Quit domonoky1 (Read error: 54 (Connection reset by peer))
23:26:54 Quit n1s ("Lämnar")
23:27:18 Join Randominality [0] (
23:27:50 Quit evilnick_ ("Page closed")
23:27:58Dhraakellianokay, cabbieV2 shows it, but a couple others I tried don't
23:29:27 Quit bertrik ("De groeten")
23:30:20 Quit FuzeV2 ("CGI:IRC (EOF)")
23:30:48 Quit Strife89 ("Goin home.")
23:32:32 Quit Tomis2__ (Read error: 104 (Connection reset by peer))
23:32:33saratogaTheSeven: thats expected, theres very little advantage to using more IRAM in codecs
23:33:12*TheSeven still wonders where the huge difference between flac/wma and c4000/c5000 is coming from
23:34:42 Join togetic [0] (n=togetic@unaffiliated/ibuffy)
23:35:14pixelmakugel: early USB is broken again on Archos since the initial custom status bar commit
23:37:38pixelma(different address to the one that happened before, e.g. in 3.4 and which was fixed)
23:39:22 Quit FlynDice (Remote closed the connection)
23:40:01pixelmaI get an I09: CPUAdrEr at <slightly varying address between builds>, if I read the map correctly it happens in some cuesheet related functions
23:41:02 Quit AlexP (Remote closed the connection)
23:42:01 Join AlexP [0] (n=alex@rockbox/staff/AlexP)
23:42:05 Join jus__ [0] (
23:42:19 Quit jus__ ("Ex-Chat")
23:42:38 Quit Tomis2_ (Connection timed out)
23:43:26 Join FlynDice [0] (
23:46:34 Join robin0800 [0] (
23:46:39saratogaTheSeven: have time to do a simple test?
23:47:17saratogacomment out mdct_backward in wmadeci.c like 1286
23:47:25saratogarerun test codec
23:47:39saratogai've got numbers for PP with and without that function
23:54:01CIA-6New commit by gevaerts (r23595): Fix Free and Disk size display when SECTOR_SIZE!=512 ...
23:54:17TheSevengevaerts: what exactly is this fixing?
23:54:19kugelpixelma: I saw that, but I don't know what I can do about it. As I understand, early USB is before any fs is mounted, and thus before settings are loaded. it should show the classic statusbar there
23:54:36TheSevenespecially regarding [22:32]<Utchybann>TheSeven: Debug->Partition Info is still wrong. I will catch this one.
23:54:42 Quit AEnima1577 ("Leaving.")
23:55:03TheSevenwhat about just grepping for 512 and having a look where else they are still floating around?
23:55:05kugelTheSeven, saratoga: the iram usage should really be dependant on iram size, not per cpu
23:55:11gevaertsTheSeven: System -> Rockbiox Info
23:55:17pixelmakugel: well it doesn't get this far - and it broke with r23258 (again)
23:55:28gevaertsand grepping for 512 won't find these. They usually divide by 2
23:55:37UtchybannTheSeven: done and tested. I'm about to update my patch.
23:56:14UtchybannTheSeven: this one was 2048 :)
23:56:27 Join midgey [0] (
23:57:16TheSevengevaerts: you just applied FS #10656?
23:57:26Utchybanngevaerts: you just close FS #10656...
23:57:30*Utchybann is too slow
23:57:39gevaertsTheSeven: yes
23:57:51 Join shaggy-h [0] (
23:58:48 Join sansaAMS [0] (
23:58:50kugelpixelma: would you be able to insert some panicf's to find out where it crashes?

Previous day | Next day