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 2012-09-27

00:00:33kugelthe array is filled only using strcopy
00:00:33lorenzo92kugel: the patch so far:
00:01:30lorenzo92kugel: it is lightweight, add something like 1% of cpu usage when being used...
00:04:41lorenzo92kugel: perhaps the crash is due to the different font!
00:04:45lorenzo92lebellium: ^
00:04:50 Quit n1s (Quit: Ex-Chat)
00:04:53 Quit Prodicus (Ping timeout: 265 seconds)
00:05:32kugelit would be nice to get this fixed, perhaps even before the release of 3.12 although that seems unlikely
00:07:13 Quit Wardo (Read error: Connection reset by peer)
00:07:23 Quit ender` (Quit: C++ is a modern language where your parent can't touch your privates but your friends can!)
00:07:24lorenzo92kugel: yes. moreover the patch has to be tested, I will also try using sleep(1), seems more rds-data-error-free
00:09:31lebelliumkugel: lorenzo did a test for me. If "%s%ac%?tz<%tz|>" is removed from my theme, it doesn't crash anymore. So it's definitely related to this %tz tag?!
00:10:17kugellorenzo92: could you document somewhere how you managed to get shell access?
00:10:38lorenzo92yes, will do that, in the meantime checkout this rom:
00:12:05lorenzo92install, turn on with cable inserted, when a screen is shown press UP arrow
00:12:22lorenzo92now open minicom at /dev/ttyACM0 and play ;)
00:15:45lorenzo92this is the serial via usb, then there is the real UART port, I'm waiting for ttl<->usb converter ;)
00:15:47 Quit lorenzo92 (Quit: ChatZilla 0.9.89 [Firefox 15.0.1/20120907231657])
00:19:16 Quit mgottschlag (Ping timeout: 264 seconds)
00:19:30 Quit kevku (Ping timeout: 260 seconds)
00:30:19 Quit lebellium (Quit: ChatZilla 0.9.89 [Firefox 16.0/20120919065210])
00:45:46 Join bzed_ [0] (
00:47:32 Quit bzed (Ping timeout: 260 seconds)
00:47:37 Nick bzed_ is now known as bzed (
00:51:38 Quit eckoit (Quit: eckoit)
00:55:11 Quit bertrik (Ping timeout: 244 seconds)
01:01:53 Join eckoit [0] (~ryan@
01:40:09 Join Epicanis [0] (
01:47:49 Join Prodicus [0] (
01:48:59 Quit Buschel (Quit: ChatZilla [Firefox 15.0.1/20120905151427])
01:50:48***Saving seen data "./dancer.seen"
02:02:00 Quit ser (Remote host closed the connection)
02:06:36 Join ser [0] (~ser@2610:28:3090:3001:a:a:a:dead)
02:11:53[Saint]lebellium: "%?tz<%tz|>" is really ugly :)
02:12:15[Saint]you don't need a false case here.
02:16:23 Quit eckoit (Quit: eckoit)
02:17:52[Saint]I also wonder if scrolling is messing it up.
02:18:13[Saint]the way rds works, it wouldn't surprise me.
02:19:53 Quit Thra11_ (Quit: kthxbai)
02:26:47 Join t0rc [0] (~t0rc@unaffiliated/t0rc/x-5233201)
02:36:39 Quit pystar89 (Ping timeout: 245 seconds)
02:39:49 Quit Xerion (Quit: )
02:40:32 Join eckoit [0] (
02:41:37 Join Xerion [0] (
02:56:56 Join pedro_angelo [0] (
02:58:12 Quit ender| (Read error: Operation timed out)
03:12:50 Join ender| [0] (
03:14:28 Quit eckoit (Ping timeout: 264 seconds)
03:14:30 Join Gallomimia [0] (
03:14:30 Quit Gallomimia (Excess Flood)
03:16:53 Quit thegeek (Read error: Connection reset by peer)
03:19:04 Join Gallomimia [0] (
03:38:19 Quit guymann (Ping timeout: 245 seconds)
03:40:27 Join guymann [0] (~c@unaffiliated/guymann)
03:43:27 Join eckoit [0] (~ryan@
03:46:12 Quit guymann (Ping timeout: 244 seconds)
03:48:27 Join guymann [0] (~c@unaffiliated/guymann)
03:50:31 Quit n17ikh (Ping timeout: 246 seconds)
03:50:53***Saving seen data "./dancer.seen"
03:51:27 Join n17ikh [0] (~n17ikh@
03:51:28 Quit n17ikh (Changing host)
03:51:28 Join n17ikh [0] (~n17ikh@unaffiliated/n17ikh)
03:57:32 Quit t0rc (Quit: WeeChat 0.3.8)
04:04:27 Join amayer [0] (
04:12:58 Quit ukleinek (Ping timeout: 255 seconds)
04:19:01 Join amiconn_ [0] (quassel@rockbox/developer/amiconn)
04:19:01 Quit amiconn (Disconnected by services)
04:19:01 Quit pixelma (Disconnected by services)
04:19:02 Join pixelma_ [0] (pixelma@rockbox/staff/pixelma)
04:19:04 Nick pixelma_ is now known as pixelma (pixelma@rockbox/staff/pixelma)
04:19:06 Nick amiconn_ is now known as amiconn (quassel@rockbox/developer/amiconn)
04:19:57 Join ukleinek [0] (
04:31:56 Part amayer
04:38:10 Join TheSphinX_ [0] (
04:40:59 Quit Gallomimia (Ping timeout: 244 seconds)
04:41:42 Quit TheSphinX^ (Ping timeout: 252 seconds)
04:48:56 Quit Epicanis (Quit: sleepytime)
04:49:36 Quit ukleinek (Ping timeout: 245 seconds)
04:56:53 Join ukleinek [0] (
05:05:00 Quit eckoit (Quit: eckoit)
05:15:22 Join Guinness` [0] (
05:18:20 Quit Guinness (Ping timeout: 256 seconds)
05:25:36 Quit [7] (Disconnected by services)
05:25:44 Join TheSeven [0] (~quassel@rockbox/developer/TheSeven)
05:28:52 Quit Horscht (Ping timeout: 248 seconds)
05:37:30 Join shai_ [0] (
05:40:17 Quit shai (Ping timeout: 264 seconds)
05:50:55***Saving seen data "./dancer.seen"
06:19:45 Join eckoit [0] (~ryan@
06:26:13 Quit [Saint] (Read error: Connection reset by peer)
06:26:18 Join [Saint_] [0] (~saint@rockbox/user/saint)
06:55:05 Join davo [0] (
07:15:46 Join mortalis [0] (
07:18:35 Quit Rower85 (Read error: Connection reset by peer)
07:20:01 Join shai [0] (
07:20:44 Quit shai_ (Ping timeout: 256 seconds)
07:28:42 Join kevku [0] (
07:50:57***Saving seen data "./dancer.seen"
07:57:16 Nick [Saint_] is now known as [Saint] (~saint@rockbox/user/saint)
08:04:23 Nick crwll is now known as crwl (
08:13:31 Join Buschel [0] (
08:14:05[Saint]JdGordon: is it just me...or, are RTL fonts just broken?
08:14:57[Saint]wrt: display (some words looks really weird and broken, even without being able to read the language and alignment is fudged)
08:18:45[Saint]why did I say fonts? whoops, I meant "RTL langs", general.
08:19:11[Saint]every one I try seems to have it's own weird issue.
08:20:29[Saint]some words are screwy and jumbled and I've got a feeling the direction specific alignment tags aren't helping things.
08:20:57[Saint]dunno, need to test more myself, it just doesn't look right.
08:21:46 Quit XavierGr (Ping timeout: 240 seconds)
08:24:19 Join wodz [0] (
08:29:48 Join n1s [0] (
08:29:49 Quit n1s (Changing host)
08:29:49 Join n1s [0] (~n1s@rockbox/developer/n1s)
08:37:20 Join bertrik [0] (~quassel@rockbox/developer/bertrik)
08:45:13 Quit [Saint] (Remote host closed the connection)
08:46:57 Join [Saint] [0] (~saint@rockbox/user/saint)
08:47:04davoI've damaged my sansa fuze first gen and I need a replacement. any suggestions on a comparable model?
08:47:51 Join thegeek [0] (
08:50:54 Join mgottschlag [0] (~quassel@reactos/tester/phoenix64)
08:53:26 Quit n1s (Read error: Connection timed out)
08:54:23 Join ender` [0] (
08:54:40 Quit Prodicus (Ping timeout: 244 seconds)
08:56:53 Join Zagor [0] (
08:56:53 Quit Zagor (Changing host)
08:56:53 Join Zagor [242] (~bjst@rockbox/developer/Zagor)
09:00:10 Join petur [0] (~petur@rockbox/developer/petur)
09:02:44 Quit nosa-j (Ping timeout: 248 seconds)
09:04:47 Quit bertrik (Ping timeout: 265 seconds)
09:11:15 Quit factor (Ping timeout: 248 seconds)
09:17:18 Quit scorche (Disconnected by services)
09:17:22 Join scorche` [0] (~scorche@rockbox/administrator/scorche)
09:18:02 Quit davo (Ping timeout: 256 seconds)
09:22:25 Quit pedro_angelo (Remote host closed the connection)
09:25:14 Join LinusN [0] (
09:26:56 Part LinusN
09:37:15 Quit bzed (Remote host closed the connection)
09:37:22 Join bzed [0] (
09:43:53wodzwhere is CIA?
09:50:59***Saving seen data "./dancer.seen"
09:55:25wodzanyway lcd rework for rk27xx is big step forward :-)
09:55:49wodzn1s: (logs) your last commit causes warnings on MIPS
10:16:55 Quit wodz (Quit: Leaving)
10:23:30 Join Thra11 [0] (
10:50:25funmanwodz: cia is dead
10:51:10 Join wodz [0] (
10:52:02Zagorfunman: :-(
10:52:40Zagorok, time to make logbot take up this job instead
10:54:31gevaertsZagor: what's the status of commit emails these days?
10:54:36 Join n1s [0] (
10:54:36 Quit n1s (Changing host)
10:54:36 Join n1s [0] (~n1s@rockbox/developer/n1s)
10:55:37Zagorgevaerts: I thought they worked?
10:55:56gevaertsMaybe there's something wrong with my subscription then
10:55:58ZagorI'm not subscribed myself though
10:56:05ZagorI have to verify
10:56:57wodzI don't get commit mails also
10:57:09gevaertsThe web thing says I'm subscribed to rockbox-cvs
10:57:25Zagorweird. I'll look at it.
10:57:40wodzmortalis: Thank you for cooperation. lcd dma stuff is big step forward
10:58:33 Join gxk [0] (
10:59:31 Quit funman (Ping timeout: 248 seconds)
11:00:00 Join funman [0] (~fun@rockbox/developer/funman)
11:01:52mortalisanytime :)
11:02:04 Join LinusN [0] (
11:02:09wodzmortalis: Do you plan to work on something particular now?
11:03:28mortalisi added rbutil support, but it not tested yet. After that i want to try with ftl
11:04:13 Join lorenzo92 [0] (
11:05:05mortalisbtw, typecast of LCD_BUFF still incorrect.
11:05:20wodzmortalis: I can send you hacked skyeye simulator. I used it to trace execution path of bootloader's FTL
11:05:39wodzmortalis: you mean lack of volatile? I realized this after I pushed the button
11:06:24mortaliswodz: yes,
11:06:52wodzmortalis: unfortunately with disk crash I lost most of my notes from FTL reverse engineer sessions.
11:08:00wodzmortalis: go ahead and commit this. I don't have git/gerrit stuff setup on this machine.
11:10:00mortalisgtg, i'll push it later
11:13:13[Saint]gevaerts: you're right, rockbox-cvs is a wasteland
11:13:18[Saint]it hasn't worked for ages.
11:14:51n1swodz: ah, missed the warning since i didn't expect it to differ between archs, i suppose it's the ancient gcc used for mips
11:17:49Buscheln1s: with algorithmic changes to combfilter and lots of hacky IRAM stuff I reach 81 MHz on PP now for the 64k opus file
11:19:07n1sBuschel: cool, i made a somewhat hacky patch that puts the whole decoder state in iram, that together with putting the hottest functions in iram and asm for the deemphasis loop makes the 64kbps file 102% or so realtime
11:20:03n1si took the iram alloc stuff from tremor but it's fairly pointless when just allocating a single buffer
11:20:31BuschelI chose an even hackier solution :)
11:21:24n1sthe variable size is a bit annoying but all the test files alloc the same amount so just allocating a static buffer and using it as long as it's big enough ought to work just as well
11:22:24Buschelcould you just exchange your comb_filter() with the following implementation and measure on cf? ->
11:26:41Buscheldid you also move the variables freq and X (from celt_decode_with_ec()) to IRAM? this resulted in a reasonable speedup. same for f and f2 from clt_mdct_backward()
11:30:25n1sonly the f in clt_mdct_backward() (i pushed that change yesterday :)) didn't think the rest looked frequently used enough
11:31:27lorenzo92possible that radio chip uses lot of energy? battery benchmark lasted 10 hrs versus 24 hrs playback!
11:31:53[Saint]some, yes. that much? Hmmm...methinks something's wrong.
11:32:10[Saint]Fm enabled should halve the battery life.
11:32:16Buscheln1s: f2 saved another 0.8 MHz here, freq/X about 1.5 MHz (on PP)
11:32:46n1sBuschel: ah, then they should definitely be interesting on cf
11:33:02Buschelyep, this should scale even better on cf
11:33:44Buscheln1s: did I get you right that you made a proposal for an optimization of deemphasis()?
11:34:14lorenzo92lebellium: can you please do a battery benchmark while using radio on R0 WITHOUT rds...and then we'll see one with radio rds
11:36:09n1si just did what jmspeex suggested (removing the downsampling stuff since we don't use it and drop a multiplication by 0) and used a better 16 bit saturation, saves about 1.3MHz on cf but i also wrote a version in inline asm that's faster but i'm not quite happy with it yet
11:36:31n1sso no big algorithmic change
11:38:42n1syour comb_filter is about 1.5MHz faster on cf, but it changes the output, is that expected?
11:39:21Buschelgood result. yes, the output change is expected. this is caused by addition of two operands before the multiplication
11:40:09Buschellike changing (a*x)>>15 + (a*y)>>15 => (a*(x+y)=>>15
11:41:16n1sas long as it's not overflowing
11:44:52n1si've gtg, bbl (i'll look at that mips warning then)
11:45:03Buschelsee you later
11:46:15 Quit fs-bluebot (Ping timeout: 246 seconds)
11:46:15 Quit bluebrother (Ping timeout: 246 seconds)
11:47:30 Join fs-bluebot [0] (
11:48:17 Join bluebrother [0] (~dom@rockbox/developer/bluebrother)
11:51:00***Saving seen data "./dancer.seen"
12:05:38 Quit kevku (Quit: KVIrc 4.2.0 Equilibrium
12:15:12 Join kevku [0] (
12:17:01kugellorenzo92: sleep(1) seems low. did you try higher values?
12:18:13kugellorenzo92: btw you can add me as a reviewer on gerrit then I would have bene notified by mail
12:22:04lorenzo92kugel: okay, for the sleep(1), well sleep(5) seems too high (incomplete rds data). 2 was fine too, but since cpu usage doesn't raise I've put 1.
12:23:49lorenzo92kugel: btw did you try serial console? :)
12:26:08kugelnot yet
12:26:31lorenzo92kugel: okay, so what do you think about the patch?
12:34:34 Join JdGord [0] (~AndChat80@
12:36:52kugellorenzo92: how does work? it puts the thread to sleep forever if radio/rds isnt enabled
12:37:24 Quit JdGord (Read error: Connection reset by peer)
12:38:13 Quit mgottschlag (Ping timeout: 240 seconds)
12:38:20 Quit crwl (Quit: muutto)
12:38:24lorenzo92kugel: nice! I had a similar idea, but we need to test out if it really works
12:38:49lorenzo92how many seconds are TIMEOUT_BLOCK?
12:39:18kugelforever :)
12:39:29lorenzo92ah yeah sorry :p
12:40:01lorenzo92kugel: indeed the best approach unlike sleeping for a second
12:40:27lorenzo92kugel: queues aren't a problem in RaaA?
12:41:11lorenzo92good then, can I test it?
12:41:49kugelyes please
12:42:02kugelapply it ontop of your gerrit patch
12:45:30kugel(btw I didn't test it)
12:45:38lorenzo92kugel: :( crash on start
12:45:54kugelhow does it crash?
12:46:23lorenzo92kugel: nothing on the screen after logo, now investigating with serial
12:47:10lorenzo92uhm no it doesn't crash
12:47:20lorenzo92but doesn't start either!
12:47:55kugelfreeze or something?
12:48:32lorenzo92well I don't know really, because the serial console is still working when I place rockbox in background
12:49:31kugeltry gdb
12:49:36lorenzo92kugel: 915 root RW 40292 8904 99.6 64.4 rockbox
12:50:24lorenzo92kugel: doesn't tell anything, just running
12:50:42kugelhit ctrl+c to see where it's stuck
12:50:51kugel(+ bt afterwards)
12:51:02lorenzo92if I hit CTRL+Z this is ouputted:
12:51:04lorenzo92[New LWP 12619]
12:51:06lorenzo92[LWP 12619 exited]
12:51:23lorenzo92always with different codes
12:51:57kugelctrl+c. not +z
12:52:23lorenzo92kugel: ctrl+c does the same!
12:52:49mortaliswodz: Do i recall correctly that rk28 nand driver is the same as in rk27? what about ftl?
12:53:02wodzmortalis: It is similar not the same
12:53:37wodzThey shuffled structs, some are merged. Identification number differ etc.
12:53:45mortalisis ftl similar too?
12:54:57wodzJudging from the code it should be similar. I saw many similar constructs. But as I don't know exactly how rk27xx ftl works (even initialization) I can't tell for sure
12:54:59lorenzo92kugel: hope no bugs with queues since they're used also in usb context (I will understand then why my usb code crashes)
12:56:16kugelqueues work fine
12:56:58wodzelf2flt leaks mem like a hell. It is by no means the most horrible piece of code I ever saw. No question why it is so fragile.
12:57:03kugelbut I can't see the problem
12:58:29kugellorenzo92: what if you enter bt?
12:58:51lorenzo92kugel: oh! now's happening something odd. it continues printing out LWP #### exited]7
12:58:58lorenzo92kugel: no way to insert bt
13:00:06 Quit petur (Quit: *plop*)
13:01:30kugelafter ctrl+c, you should get to the gdb prompt
13:01:46lorenzo92kugel: doesn't happen
13:02:10lorenzo92kugel: indeed in various occasions ctrl+c doesn't work
13:02:17lorenzo92perhaps it's a setting of the serial console
13:04:56 Join dfkt [0] (dfkt@unaffiliated/dfkt)
13:09:55kugellorenzo92: so, rockbox as a whole doesnt start, not only radio?
13:10:29wodzis there standard way to tell if machine is big endian or little endian I mean some standard function or something?
13:10:44lorenzo92kugel: as whole
13:11:39kugelwodz: BYTE_ORDER macro
13:12:31kugelor __BYTE_ORDER
13:27:46 Quit lorenzo92 (Ping timeout: 260 seconds)
13:32:30kugelmortalis: can't you just do memset?
13:33:41wodzkugel: memseting volatile * is undefined behavior
13:33:49mortaliskugel: memset needs non volatile pointer
13:35:46kugelare you sure?
13:36:16kugelcould gcc optimize the call away?
13:37:08wodzmortalis: If you are bored you can play with putting cpu in sleep mode (writing SCU_CPUPD = 0xdeedbabe). It works with caches turned off. With caches on it hangs or crashes
13:42:48mortaliskugel: In our case I think nothing bad will happen if refer to LCD_BUFF through non volatile pointer, but anyway it's undefined behavior, so it's not very good practice
13:42:56mortalis"kugel: "If an attempt is made to refer to an object defined with a volatile-qualified type through use of an lvalue with non-volatile-qualified type, the behavior is undefined.""
13:43:54mortaliswodz: not today
13:51:02***Saving seen data "./dancer.seen"
13:56:49 Quit hype (Read error: Connection reset by peer)
14:08:55kugelwodz: doesnt arm have a standard method to sleep?
14:09:38Torneit has a semi-standard WFI call
14:09:39kugel0xdeedbabe looks like a type!
14:09:49Tornebut that doesn't normally affect the rest of the SoC
14:09:58Torneputting more of the device into sleep is implementation-defined
14:10:01TorneWFI is only for the actual core
14:10:19n1snow why would casting a pointer to an array be strict align ok but not casting the array directly?
14:11:35 Join Prodicus [0] (
14:11:39n1sdoes anyone use mips or have a toolchain set up?
14:14:13wodzn1s: I seem to have toolchain installed
14:14:53 Join mgottschlag [0] (~quassel@reactos/tester/phoenix64)
14:17:29n1swodz: could you test build a warning fix if i come up with something simple that's better than just disabling strict aliasing for mips
14:18:35wodzsure. I can build withing 15 minutes or so and then I will be offline for ~1h
14:19:05 Join einhirn [0] (
14:19:39wodz"ARMv5 (for example the ARM926EJ-S) does not include the WFI instruction." So this will not work on rk27xx
14:19:44 Join amayer_ [0] (
14:22:31kugelI think we use it on these targets
14:22:45kugelperhaps it's optional on these?
14:22:46Tornewodz: not the WFI instruction
14:22:59Tornethere's a cp15 write that is WFI on older ARMs
14:23:04Torneat least, on those with cp15
14:23:15Tornethe WFI instruction is a later addition to make that less cryptic
14:24:11wodzrk27xx doesn't have cp15
14:24:19n1swodz: could you test build this
14:24:37Tornewell, then it doesn't have the standard in-core mechanism
14:24:47Tornedevices without cp15 lack all that stuff
14:24:54Torneso that's why it's a peripheral register :)
14:32:22 Join XavierGr [0] (XavierGr@rockbox/staff/XavierGr)
14:33:18Buscheln1s: -> 80.2 MHz on PP now. Would be interesting to have a quick test on cf as well. this also contains all dirty hacks for IRAM usage of variables.
14:35:34n1sBuschel: did you test the various iram things individually?
14:37:12Buscheln1s: yesterday night. I would like to retest them though because I've changed other stuff in the meantime
14:37:21 Quit kevku (Ping timeout: 272 seconds)
14:38:44n1soverflows plugin iram by 6k
14:39:48wodzn1s: patch does not apply
14:40:43Buscheln1s: argh... just remembering that PP has quite large IRAM... :/
14:41:04 Quit wodz (Quit: Leaving)
14:42:25Buschelanyway, gtg now. see you later
14:42:27n1sBuschel: yeah it can be tight on 5249 (only 48k available), putting some code in iram also gave me a speedup of ~10MHz on cf
14:42:28 Quit Buschel (Quit: ChatZilla [Firefox 15.0.1/20120905151427])
14:47:59 Join t0rc [0] (~t0rc@unaffiliated/t0rc/x-5233201)
14:55:17 Join kevku [0] (
15:08:58 Quit the-kyle (Read error: Connection reset by peer)
15:11:14 Join factor [0] (
15:21:09amayer_is there a way to see if a specific target was effected by updates?
15:24:27n1samayer_: look at the changes to the code
15:27:43amayer_i have an RSS feed of all the changes. but i am not familiar with the hardware in my iPod Classic.
15:27:45amayer_I think its arm based but im not sure.
15:27:47amayer_I was just wondering if there was a way to know if the updates effected me or not.
15:33:36Torneif you wait a while, then almost certainly it will affect you :)
15:33:52Torneas a first approximation, anything not specifically identified as being a different target probably affects you
15:34:25Torneso if there's more than a few commits that are new then probably at least *one* of them impacts your build.
15:34:27gevaertsUnless it's about a feature you don't use
15:34:52TorneWell, it still affects your build :)
15:35:56amayer_Torne: thank you.
15:35:58amayer_I understand that anything general affects me. I was just wondering when a commit is made that talks about processor drivers and LCD updates if they effect me.
15:36:20gevaertsYou could compare the resulting binaries :)
15:38:34amayer_I have to do all my updates manually. Im not compliaing i love rockbox and am willing to go the extra mile to keep it up to date.
15:38:36amayer_i was just wondering if there was a way to tell.
15:38:38amayer_taking the time to download a new commit and manually installing the update seems like a lot of work if the update doesnt effect me.
15:39:30gevaertsUpdating after every commit is madness anyway
15:39:39amayer_thats what im saying
15:39:40Torneright, but the point is that probably 50%+ of commits affect you
15:39:41gevaertsI'd recommend just updating weekly or something like that
15:40:29Tornemaybe that's not actually true, hm.
15:40:38Tornea significant enough proportion that it's not worth really caring, though
15:40:38amayer_gevaerts: i usually do that.
15:40:52Torneupdate whenever you get around to it
15:40:57Tornethere is little reason to try and be on the latest commit all the time
15:41:05Torneunless there is a fix for a problem you are actually having
15:41:18Tornei update my devices maybe every couple of months :)
15:41:29[Saint]basically, if you're happy with it, and no new features are added (or nothing you're interested in), updating at all between releases is needless.
15:41:50Tornenot even on purpose, half the time, but just because i accidentally blew something away I didn't mean to and the easiest way to get it working again is to install the current build :p
15:44:23[Saint]I have some players that still have a svn-revisioned build on :)
15:46:05n1si usually just update when i'm testing code i'm working on, which had the annoying tendency of leaving me with a broken vorbis codec, which i actually use
15:46:51n1sit's nice to work on codecs you don't use
15:48:56[Saint]I'm mostly lucky insofar as everything I work on wrt rockbox I can test in the sdl app
15:51:06***Saving seen data "./dancer.seen"
15:52:20amayer_is there a script that runs on the build server that could output a checksum of the binary after it is built?
15:52:22amayer_then it would be easy to write a perl script to match the checksum against your current binary to see if the updates effect you
15:53:32amayer_i guess i could just write a perl script to download the update, unzip it, then checksum both binarys.
15:54:27n1samayer_: the git rev sha is built into the binary so the checksums will always differ
15:55:13amayer_n1s: you just killed my dreams. jk. thank you very much for the info
15:56:20mortalisamayer_: you could look at file size deltas of the binaries
15:57:19mortalisif it changed then probably build affects you
15:57:35 Join linuxstb [0] (
15:57:40amayer_what do you mean file size _deltas_?
15:58:11amayer_does that just mean difference from the last build?
15:58:39mortalissecond table shows you how size of the binary was changed
15:59:56amayer_mortalis: thank you very much. i think that is exactly what i was looking for
16:01:30 Quit mortalis (Quit: Leaving)
16:02:07 Join the-kyle [0] (
16:37:20 Join Rower85 [0] (
16:41:52 Quit mgottschlag (Ping timeout: 255 seconds)
16:49:07 Quit factor (Read error: Connection reset by peer)
16:50:07 Join factor [0] (
16:51:08 Quit eckoit (Quit: eckoit)
17:07:47 Join pamaury [0] (~quassel@
17:07:47 Quit pamaury (Changing host)
17:07:47 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
17:08:10 Quit t0rc (Quit: WeeChat 0.3.8)
17:17:01 Quit pamaury (Ping timeout: 252 seconds)
17:36:19 Join pedro_angelo [0] (
17:36:39 Join y4n [0] (~y4n@unaffiliated/y4ndexx)
17:37:16 Join mgottschlag [0] (~quassel@reactos/tester/phoenix64)
17:49:26 Quit Zagor (Quit: Clint excited)
17:51:07***Saving seen data "./dancer.seen"
18:11:06 Join mortalis [0] (4d6c62b1@gateway/web/freenode/ip.
18:21:49 Join TheLemonMan [0] (~LemonBoy@unaffiliated/thelemonman)
18:33:39 Quit n17ikh (Ping timeout: 265 seconds)
18:33:58 Join n17ikh [0] (~n17ikh@
18:33:58 Quit n17ikh (Changing host)
18:33:58 Join n17ikh [0] (~n17ikh@unaffiliated/n17ikh)
18:37:23 Part LinusN
18:49:18 Join pretty_function [0] (~sigBART@
18:49:22 Quit the-kyle (Quit: Leaving.)
18:53:02 Quit mgottschlag (Ping timeout: 245 seconds)
18:56:51 Quit Rower85 (Quit: Hmmm...)
19:05:20 Join eckoit [0] (~ryan@
19:13:44 Quit einhirn (Quit: Miranda IM! Smaller, Faster, Easier.
19:16:10 Join the-kyle [0] (
19:23:44 Join Buschel [0] (
19:34:48 Join Rower85 [0] (
19:35:12 Quit dfkt (Quit: -= SysReset 2.55=- Sic gorgiamus allos subjectatos nunc.)
19:51:09***Saving seen data "./dancer.seen"
19:59:39 Join lorenzo92 [0] (
20:00:19 Quit TheLemonMan (Ping timeout: 255 seconds)
20:02:21lorenzo92kugel: back to your patch...did you find something interesting for the issue?
20:11:03Buscheln1s: just remeasured the IRAM changes. OpusDecoder (26468 byte, 2.75 MHz saving), f2 (3840 byte, 1.15 MHz saving), f (3840 byte, 0.39 MHz saving), x (3840 byte, 1.02 MHZ saving), freq (7680 byte, 1.24 MHz saving), s_x (4320 byte, 0.43 MHz saving). The variables names are taken from the patch I've posted before)))
20:13:29Buscheln1s: we should definately move OpusDecoder, freq, x and f2 to IRAM. The other changes could be dropped for targets with less IRAM.
20:18:27 Join Wardo [0] (
20:19:56 Join pamaury [0] (~quassel@
20:19:56 Quit pamaury (Changing host)
20:19:56 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
20:25:18 Quit guymann (Ping timeout: 265 seconds)
20:28:49 Join bertrik [0] (~quassel@rockbox/developer/bertrik)
20:46:47 Join guymann [0] (~c@unaffiliated/guymann)
20:53:57 Join Strife89 [0] (
20:56:38 Join Strife1989 [0] (~Strife89@
20:59:17 Quit Strife89 (Ping timeout: 245 seconds)
21:11:34 Quit kevku (Ping timeout: 260 seconds)
21:13:31 Quit mortalis (Quit: Page closed)
21:23:25 Join kevku [0] (
21:25:39 Quit pretty_function (Ping timeout: 246 seconds)
21:31:21 Quit lorenzo92 (Quit: ChatZilla 0.9.89 [Firefox 15.0.1/20120907231657])
21:35:57 Join wodz [0] (
21:39:56wodzbFLT on MIPS will be nightmare. There are lots of pairs R_MIPS_HI16 and R_MIPS_LO16 generated which cannot be represented as simple addend. R_MIPS_26 also. Relocation entry will need to keep its type at least.
21:44:39 Quit wodz (Quit: Leaving)
21:47:38 Nick Strife1989 is now known as Strife89 (~Strife89@
21:48:42 Quit y4n (Quit: Assumption is the mother of all fuckups)
21:49:18 Quit pamaury (Ping timeout: 260 seconds)
21:51:10***Saving seen data "./dancer.seen"
21:56:01 Join nosa-j [0] (~m00k@
22:02:57 Quit guymann (Ping timeout: 246 seconds)
22:03:36 Join pamaury [0] (~quassel@
22:03:36 Quit pamaury (Changing host)
22:03:36 Join pamaury [0] (~quassel@rockbox/developer/pamaury)
22:03:41 Quit Buschel (Quit: ChatZilla [Firefox 15.0.1/20120905151427])
22:04:56 Join guymann [0] (~c@unaffiliated/guymann)
22:10:56bluebrotherAlexP: what's the timeframe for 3.12?
22:11:41AlexPbluebrother: My plan such as it is is to reassess where we are this weekend, and hopefully get it out then
22:14:28 Quit efyx (Ping timeout: 264 seconds)
22:16:35bluebrotherhmm, cia not being around is quite odd
22:17:55 Quit pamaury (Ping timeout: 245 seconds)
22:18:19amiconnMissed some other i386 lib. Fixed now (re borked android builds on saturn)
22:18:49amiconnHmm, and did we loose some roolku clients? Builds are quite slow
22:23:35bluebrotherwould be nice if the build client would not only show the time needed after the build finished but also the result (i.e. number of errors / warnings)
22:27:16 Join efyx [0] (~efyx@
22:28:01 Quit Strife89 (Quit: Connection reset by deer.)
22:33:33 Quit eckoit (Quit: eckoit)
22:44:13 Join eckoit [0] (~ryan@
22:49:14n1sBuschel: nice, putting OpusDecoder in iram saved ~24MHz on cf and f saved ~32MHz, did you test against current git as of yesterday or from before i made f be on the stack, which is in iram?
23:03:31n1sanyway, i'll test it out tomorrow
23:10:44 Quit pedro_angelo (Remote host closed the connection)
23:10:58 Part amayer_
23:15:33 Join mgottschlag [0] (~quassel@reactos/tester/phoenix64)
23:18:40 Quit bertrik (Read error: Operation timed out)
23:24:27 Quit kevku (Quit: KVIrc 4.2.0 Equilibrium
23:42:16 Quit Viper^ ()
23:51:13***Saving seen data "./dancer.seen"
23:58:40 Join amayer [0] (

Previous day | Next day