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

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

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

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

#rockbox log for 2017-01-14

00:00:14pamaurychrisjj: you'll be allowed to complain the day you implement a speaker setting in a plugin. Until then I don't see the point of arguing
00:02:36[Saint]nothing better to do, presumably.
00:02:51chrisjjI wasn't complaining. But thanks - I'll take that raincheck :-)
00:03:01chrisjjReturning to the ZEN BSoPO mystery, are you aware the ugly white full-screen flash on backlight wake is remains present on lcd_fix?
00:06:13chrisjjOK. The UWFSF on backlight /wake/ was cured, but interestingly by the bootloader update from V2Beta, not by the .rockbox update.
00:06:37pamauryUWFSF ?
00:07:01chrisjj ugly white full-screen flash :-)
00:09:01chrisjjI mention this for in case state persisting from the bootloader is not supposed to affect LCD performance under .rockbox.
00:09:45chrisjjI'd have thought it was not. I.e. I'd have thought .rockbox would completely setup the LCD state itself.
00:10:18pamauryit does
00:10:19chrisjjSorry. s/The UWFSF on backlight /wake/ was cured/The UWFSF on backlight /sleep/ was cured/
00:11:09chrisjjWell, if it does, I wonder how come I see different performance of LCD in .rockbox, after different bootloaders.
00:11:52chrisjjThis is on two different devices so in theory this could be the rumoured different LCDs... but I thought I'd ask before reflashing.
00:12:58pamauryThe old bootloader left the LCD in a weird state when giving the hand to .rockbox. It is possible that the code in .rockbox did not manage to recover from this. I don't know, I don't really care to be honest as look as the lcd works
00:16:33chrisjjThanks. I too wouldn't care, if the port ran reliably.
00:18:41chrisjjOn the UWFSF, is this something on which you'd appreciate further info?
00:19:21pamauryI probably know how to fix it
00:19:30 Quit TheLemonMan (Quit: "It's now safe to turn off your computer.")
00:21:15lebelliumgevaerts: where should I look for the shortname checking with WPS usage?
00:21:30lebelliummatching checkwps usage*
00:22:33chrisjjBTW, I just reflashed Unit G bootloader from V2Beta to lcd_fix... and the sleep UWFSF remains!
00:22:35gevaertslebellium: I'm pretty sure it's the name you can give to configure
00:22:40gevaertsNot sure if that helps :)
00:23:07lebelliumIs it necessarily the same name as the config file name. For example creativezen.h ?
00:23:46gevaertsArguably it *should* be, but those can be different
00:24:37gevaertsI suspect it's also the name used in tools/
00:25:28 Join alexweissman [0] (
00:26:56 Quit petur (Quit: Leaving)
00:28:57CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
00:28:57*pamaury thinks there is a bug in the speaker code for imx233. Enable > Disable > Enable leaves the speaker disabled
00:29:13chrisjjpamaury, i.e. On two units having the same bootloader and .rockbox, I see markedly different LCD performance.
00:30:16 Quit ender` (Quit: Trying to establish voice contact ... please yell into keyboard.)
00:30:33lebelliumgevaerts: here you go
00:31:12pamauryah great, freescale but a speaker powerdown bit but it cannot be cleared once it is set, seriously?
00:34:44chrisjj'markedly different LCD performance' Both with config.cfg deleted.
00:36:16gevaertslebellium: I've added them. Now running checkwps on all themes, I hope that's enough to actually make them get themes
00:38:17pamaurydamn, I wonder if the speaker also needs some magic like the headphones to start
00:39:41chrisjjFreescale: 'It is reset by a power-on reset only'
00:40:29***Saving seen data "./dancer.seen"
00:40:34chrisjjWhich does somewhat take the shine off 'Due to the differential output, the speaker can be powered up or down nearly instantaneously without any pop problems.' :-)
00:41:13gevaertslebellium: I imagine some of those have new resolutions
00:41:17lebelliumgevaerts: I guess we only have a problem with the 128x160 targets (NWZ-E370 and Zen Mozaic). The theme for Philips SA9200 should display for them, I guess
00:42:04*chrisjj assumes pamaury's talking about HW_AUDIOOUT_PWRDN bit 24.
00:42:24pamaurychrisjj: I assume the MUTE bit of the speaker control powers up/down
00:42:51lebelliumgevaerts: Samsung YH-820 is new resolution. It's the only one which should have 0 theme
00:43:52chrisjjpamaury: I failed to parse that.
00:44:37*pamaury is speaking about MUTE in HW_AUDIOOUT_SPEAKERCTRL
00:46:12__builtinoh hey gevaerts, I was wondering something
00:46:22gevaertsscorche: apparently the theme server doesn't have the necessary stuff installed to build checkwps these days
00:46:58__builtinwhat's the best way I should handle making and pushing multiple changes in git? I've got a bunch of changes on a branch and they're split into a couple commits
00:47:00scorchegevaerts: yeah - i saw those errors recently
00:47:12chrisjj' The speaker antipop startup/shutdown sequence should be followed before toggling this bit.' ...
00:47:28__builtinif I push them normally, the build server will take forever to finish, no?
00:47:58pamaurychrisjj: quoting the manual is not helping...
00:48:04gevaertsIt definitely won't build all revisions
00:48:09scorchegevaerts: you know of a quick fix or should it just wait until i completely re-do that server?
00:48:28chrisjjAre you following the speaker antipop startup/shutdown sequence?
00:48:39pamauryit is described nowhere so no
00:49:18chrisjjAh. Meets the definition of magic, so perhaps yes you do need to apply some :(
00:49:40gevaertsscorche: the first issue is it missing sdl-config
00:50:14gevaertsAnd then for some targets the android sdk apparently...
00:50:19gevaerts(and ndk)
00:50:30gevaertsNot sure why that's needed for checkwps, really
00:52:51*gevaerts suspects that's probably people being overly cautious/lazy when adding checks to configure
00:53:01chrisjjpamaury: Did you see the little antipop sequence description on 29-26? I promise not to quote it :-)
00:53:28*pamaury spots the obvious typo in the code that explains everything
00:54:39pamaurychrisjj: yes, that's for headphone only and it doesn't work
00:54:46pamauryI implement my own version of antipop
00:55:40chrisjjFor headphone only but can you translate it to use the same control bits for speaker?
00:56:35*chrisjj wonders if pamaury was joking about the typo.
00:56:57*__builtin doubts it
00:57:53*chrisjj wonders if this typoed code is Freescale's
00:58:28pamauryactually funny thing, the power down bit for speaker can be cleared and set at will, despite what the manual says
00:58:33pamauryand no typo was not a joke
00:59:05chrisjjSo please do tell us the explanation for everything!
00:59:10[Saint]heh - freescale not meeting parity with their documentation
00:59:17[Saint]I am both shocked and amazed
00:59:18gevaertsscorche: I suspect it's just the sdl dev stuff that's missing. The android stuff is a red herring, checkwps doesn't build for the android-based things anyway apparently
01:00:18fs-bluebotBuild Server message: 3New build round started. Revision 637c741, 255 builds, 15 clients.
01:01:34*chrisjj assumes Freescale's non-parity is for parity with Rockbox's non-parity.
01:02:42scorchegevaerts: libsdl1.2-dev?
01:02:53gevaertsYes, that's the one I expect
01:04:19scorchek - done
01:05:28gevaertsSeems to work
01:06:01gevaertsWell, the ones I expect to work. You didn't magically fix unmaintained broken half finished ports :)
01:11:29fs-bluebotBuild Server message: 3Build round completed after 670 seconds.
01:11:30fs-bluebotBuild Server message: 3Revision 637c741 result: 6 errors 1 warnings
01:11:53lebelliumgevaerts: should I see a difference now?
01:13:48gevaertslebellium: not yet :)
01:15:18chrisjjpamaury, re the mystery Unit N, I reinstalled bootloader and .rockbox, and still get LCD dots then data abort.
01:15:59chrisjjpamaury, I now have three units (G, N and Q) showing markedly different LCD performance from the same versions of bootloader (45697a0bf-161212) and .rockbox (again 45697a0bf-161212). Two have shown multiple BSoPOs though I've not yet reproduced the conditions for that. Let me know if you want any detail on this.
01:17:34gevaertsscorche: did I just kill apache?
01:17:44fs-bluebotBuild Server message: 3New build round started. Revision 79e8cd4, 255 builds, 15 clients.
01:18:08gevaertsOr just the entire server?
01:18:34__builtinwell, forums aren't working now :P
01:19:27scorchegevaerts: the latter - give it 3 minutes to reboot
01:22:16chrisjjCorrection: L, N and Q. Unit G shows the same as Q (Ugly White Full Screen Flash on backlight sleep.)
01:22:54pamaurychrisjj: see commits for typo
01:25:29chrisjjNot 'fix typo (nwz-zx100 -> nw-zx100)', I guess :-)
01:27:24chrisjjI think that rates as a 'codo' :-)
01:29:05chrisjjCongrats on the catch.
01:29:36fs-bluebotBuild Server message: 3Build round completed after 712 seconds.
01:29:37fs-bluebotBuild Server message: 3Revision 79e8cd4 result: 0 errors 1 warnings
01:32:10chrisjjpamaury: Does .rockbox on ZEN (plain speakerless ZEN model) ever read the audio jack status?
01:33:07*__builtin probably spends as much time wrangling with git as he does writing code... :(
01:35:26*chrisjj hopes __builtin finds comfort in the fact he's not alone.
01:35:59pamaurychrisjj: no, as far as I know, there is no way to detect jack on the ZEN
01:37:05chrisjjOK, thanks. I am sure I saw audio jack insert cause backlight wake (yes really) but it was at 4am so I guess I was asleep and having a nightmare.
01:37:49chrisjjA nightmare because audio jack is on my list of BSoPO factor candidates. Or was. Now removed.
01:38:06pamaurychrisjj: regarding the white flash on the ZEN, are you sure it happens on sleep ? It should happen on wake
01:38:28*chrisjj checks he's awake
01:39:43chrisjjYes I'm sure. I've seen it 30 times today.
01:40:35chrisjjUWFSF happens on wake too, but the two are different. Wake is clean-cut and about 100ms.
01:41:25chrisjjSleep's is sheared and about 40ms.
01:41:42 Quit lebellium (Quit: ChatZilla 0.9.93 [Firefox 50.1.0/20161208153507])
01:42:51pamauryI wonder if we are talking about the same thing. On the ZEN X-Fi, when backlight goes off (what I call sleep), there is no white screen, it fades slowly and then becomes dark. But when backlight goes on (what I call wake), you can briefly see a completely white screen
01:43:44chrisjjSounds like exactly the same thing I see on ZEN Unit L.
01:44:22chrisjjBut on e.g. Unit G I see a flash on sleep too.
01:45:03*pamaury thinks he may have an explanation but needs to check something
01:53:33 Quit ZincAlloy (Quit: Leaving.)
02:00:24*chrisjj goes to bed hoping not to have nightmares about audio jack power draw subtly disturbing cached memory DMA.
02:03:21 Join jhMikeS [0] (
02:21:51 Quit pamaury (Ping timeout: 255 seconds)
02:30:32 Quit Strife89 (Ping timeout: 240 seconds)
02:31:54 Join Strife89 [0] (
02:40:33***Saving seen data "./dancer.seen"
02:42:45 Quit Senji (Ping timeout: 252 seconds)
03:17:08 Join Senji [0] (~Senji@
03:22:32 Quit Senji (Ping timeout: 258 seconds)
03:56:37[Saint]...what the hell?
03:56:46[Saint]That boy ain't right.
04:12:29jhMikeShe's really KenM irl
04:35:15 Quit [Saint] (Remote host closed the connection)
04:35:50 Join [Saint] [0] (~sinner@rockbox/staff/saint)
04:40:36***Saving seen data "./dancer.seen"
05:23:00fs-bluebotBuild Server message: 3New build round started. Revision 823f726, 255 builds, 15 clients.
05:23:07__builtinfingers crossed here...
05:36:22fs-bluebotBuild Server message: 3Build round completed after 802 seconds.
05:36:23fs-bluebotBuild Server message: 3Revision 823f726 result: 51 errors 0 warnings
05:37:36__builtinmmh, it's the optimization flags
05:52:16fs-bluebotBuild Server message: 3New build round started. Revision c1b913b, 255 builds, 15 clients.
06:03:07fs-bluebotBuild Server message: 3Build round completed after 651 seconds.
06:03:08fs-bluebotBuild Server message: 3Revision c1b913b result: All green
06:25:52fs-bluebotBuild Server message: 3New build round started. Revision 0a5b0dd, 255 builds, 15 clients.
06:26:03 Quit TheSeven (Disconnected by services)
06:26:15 Join [7] [0] (~quassel@rockbox/developer/TheSeven)
06:37:30fs-bluebotBuild Server message: 3Build round completed after 698 seconds.
06:37:31fs-bluebotBuild Server message: 3Revision 0a5b0dd result: All green
06:40:37***Saving seen data "./dancer.seen"
07:36:18 Quit alexweissman (Remote host closed the connection)
07:38:52 Join alexweissman [0] (
07:43:13 Join [Sinner] [0] (~sinner@rockbox/staff/saint)
07:43:45 Quit [Saint] (Read error: Connection reset by peer)
07:45:36 Quit pixelma (Quit: .)
07:45:36 Quit amiconn (Quit: - Chat comfortably. Anywhere.)
07:45:50 Quit [Sinner] (Read error: Connection reset by peer)
07:47:08 Nick Bray90820_ is now known as bray90820 (
07:48:59 Join pixelma [0] (~pixelma@rockbox/staff/pixelma)
07:48:59 Join amiconn [0] (~amiconn@rockbox/developer/amiconn)
07:49:19 Join [Sinner] [0] (~sinner@rockbox/staff/saint)
07:51:42 Quit [Sinner] (Read error: Connection reset by peer)
07:52:12 Join [Saint] [0] (~sinner@rockbox/staff/saint)
08:40:39***Saving seen data "./dancer.seen"
10:14:41 Join cc___ [0] (~ac@2001:910:113f:1:6a05:caff:fe1c:1627)
10:19:45 Quit krnlyng (Ping timeout: 256 seconds)
10:23:05 Join Massa [0] (4fe8c676@gateway/web/freenode/ip.
10:23:25 Part robertd1
10:23:30 Join robertd1 [0] (
10:23:33MassaHello everybody!
10:28:07 Join smoke_fumus [0] (
10:31:30 Join krnlyng [0] (
10:40:40***Saving seen data "./dancer.seen"
10:43:47[Saint]Hiiiiiiiii D̶o̶c̶t̶o̶r̶ ̶N̶i̶c̶k̶, Massa.
10:48:48MassaYears ago I actively developed for rockbox - but I had a break for several years; but now I try get familiar again ;-)
10:49:21MassaI've a few question regarding login and wiki etc.
10:50:14MassaIt seems I never registered at the wiki pages - can't remember why; so know I did!
10:51:10MassaCan someone please add me to the WikiUsersGroup? My wiki name is MatthiasM
11:07:22 Quit smoke_fumus (Quit: KVIrc 4.2.0 Equilibrium
11:21:59cc___Hi Massa ! what were you working on ?
11:25:45MassaParts of the picture resizing and integrating of the pictures / albumart in wps
11:26:53Massaand I was involved in the wps redesign at that time - which leads to the viewport design
11:27:40MassaAnd some patches for games and other stuff to make it better work at the iRiver H340 devices...
11:28:31MassaCurrently I'm interested in iPod Video / Classic and SDXC / mSATA support
11:29:58MassaAnd I try to fix the windows simulator build so that it works with (my) current cygwin installation again...
11:30:35 Join ender` [0] (
11:34:45 Join lebellium [0] (
11:35:04 Join paulk-collins [0] (
11:40:05 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
11:42:42lebelliumpamaury: do you want real KAS for NW-ZX100? Something else too?
11:46:46pamaurylebellium: yes
11:47:00dongswhat's a KAS
11:47:02pamaurybasically I'd interested in everything not in;a=blob;f=utils/nwztools/upgtools/upg.c or in the not confirmed list
11:47:24pamauryKey And Signature
11:55:29lebelliumModel: NW-ZX100
11:55:30lebelliumSeries: NW-ZX100 Series
11:55:32lebelliumkas (node 11,key and signature):
11:55:33lebellium63 64 64 61 38 64 35 65 35 33 36 30 66 64 34 33 cdda8d5e5360fd43
11:55:35lebellium37 33 31 35 34 33 38 38 37 34 33 66 38 34 64 32 73154388743f84d2
11:55:36lebellium00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
11:55:38lebellium00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
11:55:46lebelliumdon't forget to fix the typo (NWZ/NW) when you pudate upg.c
11:57:14 Quit pamaury (Ping timeout: 240 seconds)
11:57:19 Join pamaury_ [0] (~pamaury@rockbox/developer/pamaury)
11:58:51pamaury_lebellium: thanks
11:58:54pamaury_which typo ?
11:59:00 Join johnb3 [0] (
11:59:06pamaury_ah in the upg structure rigt
11:59:08lebellium"nwz-zx100", false, "2c0bf029804f73e073154388743f84d2" },
11:59:25pamaury_Massa: I added you to the wiki (I think)
11:59:59johnb3Mihail: see my forum post. Which build should I try next with my clip+ var0: -7, -8 or something else?
12:01:06pamaury_lebellium: you made the typo too :-o
12:01:28pamaury_oh no you just copied my wrong line
12:02:01lebelliumI can't claim I never make the typo NWZ vs NW but I try not to!
12:02:18pamaury_beware, quoting leads to the chrisjj side of #rockbox
12:03:16fs-bluebotBuild Server message: 3New build round started. Revision 0cabc1f, 255 builds, 16 clients.
12:05:34johnb3pamaury_: This week I received my E585. As I had never seen it in real life before, I am astonished how tiny it is. I can do testing in the future if you want me to. My presence on IRC is limited however ...
12:06:20pamaury_johnb3: I think at the moment we are good, me and lebellium have a E580 to test on, but when audio is finally working you can test it :)
12:08:04Massapamaury_: thanks - yes you did :-)
12:11:44pixelmapamaury_: you added "MatthiasM" as is? I haven't looked at it but that wiki name doesn't comply with the rules...)
12:12:24pamaury_a that's right, the wiki name rule, I always forget about that
12:12:31fs-bluebotBuild Server message: 3Build round completed after 555 seconds.
12:12:32fs-bluebotBuild Server message: 3Revision 0cabc1f result: 0 errors 1 warnings
12:12:35*lebellium doesn't want to debate about this stupid real name policy again
12:12:59pamaury_Massa: we have a real name policy, so you must put your real name on the wiki
12:13:48pixelmalebellium: calling it stupid gives incentive to start a debate...
12:13:57lebelliumthat's true
12:14:02 Nick pamaury_ is now known as pamaury (~pamaury@rockbox/developer/pamaury)
12:14:16Massapamaury_: I added my real name to the irc nicks list:
12:14:20*chrisjj wishes the E585 screen wasn't so tiny.
12:14:37pamaurypixelma: is that enough ?
12:15:10pamauryor must the name itself be the full name ?
12:15:26*pamaury knows some people are very picky about this and prefers to ask
12:15:53johnb3chrisjj: for me the screen is no problem but it doesn't really fill the palm of my hand.
12:16:23pamaurychrisjj: can you test a build for me on the ZEN ?
12:16:46johnb3So i feel I don't have a secure grip ;-) Maybe I will learn ...
12:17:26Massapamaury: Can I change the WikiName?
12:17:27chrisjjjohnb3: You could glue it to the back of the slightly larger ZEN :-)
12:17:47pamauryMassa: I don't know, I think not, the simplest way is to recreate an account :-/
12:17:58pamaurygevaerts: can one change a wiki name ?
12:18:22johnb3the only zen I had was a Creative NOMAD Jukebox Zen Xtra and didn't like it too much. sold.
12:18:22chrisjjWhat astonished me about the E585 was the battery runtime.
12:19:15pamauryit is an attempt at fixing the white flash on backlight
12:20:12Massapamaury: O.K. I added another WikiUser "MatthiasMohr" - could you please add me to the group (and remove MatthiasM)?
12:22:03MassaIs it possible to sign in to gerrit with a GitHub account?
12:22:07Massapamaury: thanks!
12:23:05pamauryMassa: in theory there a github oauth integration
12:23:33pamaurybut I think last time I tried it ended up in a 404 on github :-/
12:23:53Massayes - but it does not work; when I click at the link and log in to my GitHub account, I only get a 404 on github :-(
12:24:11Massayou were faster than me ;-)
12:24:23pamauryoauth is really annoying, it breaks all the time
12:26:24pamauryI will see with bjorn if we can fix gerrit github
12:26:35pamaurybut he is usually not very responsive
12:27:37MassaWhere can I easily get an OpenID just for that? I don't want to combine my google or wordpress accounts with that and also don't want to sign up e.g. to myspace just for that...
12:28:13chrisjjpamaury: No change. I.e. Clean boot of that build Version 3ad3145b4-170114 on a ZEN that previously showed backlight sleep and wake white flash (Unit G) still shows that.
12:28:24MassaI personally think that combining several accounts is a bad idea (privacy related)
12:30:32pamauryMassa: iirc Launchpad
12:32:53chrisjjZEN testers' Pro Tip: To revert Settings changes made since last power-on, press reset.
12:32:56Massapamaury: it's not in the list at OpenId (or I'm unable to see it)!? Do you have a link?
12:33:59lebelliumchrisjj: aren't you the only one?
12:35:22MassaSomeone here with knowledge about the configure script? Especially with building the simulator?
12:35:31chrisjj... or plug USB to PC.
12:36:40pamauryMassa: it's on the Sign-In page for gerrit (for me)
12:37:04pamauryMassa: with configure script yes, simulator less
12:37:43MassaI think there are some script bugs related to it, especially with detecting the sdl-config
12:38:13pamaurychrisjj: what about this
12:38:49MassaI tried to make it more reliable and fixed it already - but now I have some strange error when it tries to identify the architecture
12:39:31pamaurycan you pastebin ? It's a bit vague :)
12:39:32MassaIt's strange, because I can enter the exact commands with their arguments at my command line by hand and they work - but not inside the script!
12:39:37chrisjjlebellium: We live in hope :-)
12:39:53pamauryMassa: beware that configure is sh based
12:40:07pamaury(for reasons I don't understand, we might as well move it to bash)
12:40:43pamaurythere are subtle but important difference between bash and sh. What is the command that fails ?
12:40:44***Saving seen data "./dancer.seen"
12:44:22Massafirst of all, you see line 3 there ( line 348 in script)?
12:45:00Massait directly calls "sdl-config" and not the one which get's searched and found earlier (in variable $sdl)
12:46:05Massathat's the first bug - it should be changed to "$sdl"
12:47:28Massathe second bug (at least for cross-compiling environments) is, that it searches for sdl-config in the PATH - and not in the predefined cross compiling environment
12:47:55pamaurywell it does that:
12:47:56pamauryno ?
12:48:11pamaurywith comment # sdl-config might (not) be prefixed for cross compiles so try both.
12:48:32chrisjjpamaury: On that rockbox_zen_blfix_ugly 03895dd28M-170114, the backlight flashes remain.
12:49:38*pamaury cannot believe it and gives on flash for now
12:50:01pamauryMassa: did you manage to setup a gerrit account or should I push a first fix ?
12:51:39chrisjjI can send video proof if required :-)
12:52:11chrisjjBut I'll be posting the devices to you hopefully Monday.
12:52:49pamauryI need to see it myself, I don't understand the physics of this flash, the display cannot flash when backlight is off
12:53:53chrisjjI think the backlight is not off. I think the white flash is the backlight On and the pixels White.
12:54:44chrisjjAnd surely you did see it yourself on lcd_fix when you have a real device. Albeit a month back.
12:55:17pamauryThis builds turn off backlight and wait for an extra 100ms before turning off the LCD. If backlights takes more than 100ms to turn off that's bad
12:55:35pamauryI don't have this flash on sleep.
12:56:00pamauryI only have a flash on wake and this solves it for me
12:56:12chrisjjI thought you don't have the flash because you don't have the device :-)
12:56:19pamauryon the X-Fi
12:56:37pamauryIt's the same LCD as far as I know
12:57:33chrisjjAh. X-Fi. Well perhaps that where the rumoured different LCD is. Which is not to say it is likely that the LCD here takes >100ms for backlight to go off.
12:59:11chrisjjDid you look for flash on ZEN classic when you had one?
13:01:33Massapamaury: now I logged in successfully to gerrit with an launchpad Id - but it's never wrong to fix the GitHub OAuth login ;-)
13:01:55pamauryMassa: I sent an email to Bjorn to see if he can look into this.
13:02:05pamauryDon't forgot to read UsingGit on the wiki
13:02:21pamauryyou need a small hook in your git to generate Change-Id on commit
13:02:30 Join ZincAlloy [0] (~Adium@2a02:8108:8b80:1700:71cb:3547:d2da:f5e6)
13:02:49MassaO.K., now back to configure...
13:04:46MassaIn my cygwin64 / mingw environment the correct sdl-config can be found as "sdl-config" inside the cross environment; e.g. in /i686-w64-mingw32/sys-root/mingw/bin
13:05:41Massaso with the current search method it searches for i686-w64-mingw32-sdl-config through the PATH and didn't find it, but it'll find a (native) sdl-config
13:05:56Massawhich is the wrong one
13:06:40pamauryhum, that's tricky
13:07:19Massaso it's fine to first search for $(CROSS_COMPILE)sdl-config - but it should search inside the cross environment first and then in the PATH
13:07:40pamauryyeah but how does it know the cross environment location ?
13:07:54pamauryI can only think of $(CROSS_PREFIX)gcc −−print-sysroot
13:08:20pamauryor you need to tell the script
13:08:50pamauryme thinks
13:08:50pamaury`$(CROSS_PREFIX)gcc −−print-sysroot`/bin
13:08:50pamauryis a valid search location
13:09:40Massathat's a nice idea - didn't think of that :-o
13:09:44pamaurybut it might not work, mingw tends to do weird things, like putting prefixed tools in /bin and not prefixed tools in /mingw/bin
13:10:02pamaurymaybe adding
13:10:02pamaury`$(CROSS_PREFIX)gcc −−print-sysroot`/mingw/bin
13:10:02pamaurydoesn't hurt
13:10:46MassaI wrote a function which searches in /usr, /usr/local for the cross_prefix directory - with cross gcc it would be easier; but only if it's in PATH...
13:11:09Massaonly the mingw subdirectory is somehow "incorrect"
13:12:30pamauryMassa: currently how to specify your cross compiler ?
13:13:13pamauryyou pass CROSS_COMPILE=blabla ./configure blabla ?
13:14:03Massathis is my currently used complete configre script:
13:15:04pamaurya diff would be helpful ;)
13:15:51pamauryor upload to gerrit for review
13:20:58Massathis is the diff:
13:21:57MassaI currently don't want it to upload to gerrit, because it's not finished and has problems which I don't know what's going on...
13:23:51Massa - it also contains a lot "echo" - to find the reason for my problems ;-)
13:24:39pamaurythat looks complicated, why do you need all those include and libs dir ? The cross compiler is supposed to know that already
13:25:43pamauryah ok some are commented
13:25:47pamaurythere is a lot commented code
13:28:08Massathere is a lot "test code" ;-)
13:28:55Massathe problem is, that the generated GCCOPTS make the test compile to detect the architecture fails with "unknown command line option"
13:29:40Massaand they don't seem to be wrong - when I enter the CC command by hand, with that exact parameters it works!
13:30:19Massahere's the output of the 64-bit mingw compiler, called by the script:
13:30:23Massax86_64-w64-mingw32-cpp: error: unrecognized command line option ‘-W -Wall -Wundef -O -nostdlib -ffreestanding -Wstrict-prototypes -pipe -std=gnu99’
13:30:55pamaurycan you remind where the test is done in configure ?
13:32:08Massain line 375 in original configure, line 487 in mine
13:33:10Massathat was bullshit - have to search it..
13:34:06Massaline 4277 in original script, 4398 in mine
13:34:54 Quit johnb3 (Ping timeout: 240 seconds)
13:35:06pamauryso you are saying that
13:35:06pamaurycpp_defines=$(echo "" | $CPP $GCCOPTS -dD)
13:35:06pamauryfails ?
13:37:51pamauryweird, on my (linux) machine, this works:
13:37:51pamauryCROSS_COMPILE=i686-w64-mingw32- ../tools/configure
13:37:51pamaury(I haven't tried to actually compiled the sim)
13:40:15pamaurywhat is your environment ? linux ? cygwin ?
13:47:46Massathe really strange thing is: when I add a line with GCCOPTS= directly above the $CPP command in the original script with the determined parameters of my script it does work!
13:49:26pamaurycan you replace /bin/sh by /bin§bash no line 1 ?
13:49:33Massayes, it seems to work - but it does find the wrong sdl-config and therefore it would use the wrong library and include directories...
13:50:18pamauryyes I understand there is a problem with sdl but are you saying that sh vs bash makes a difference at this line ?
13:51:31Massano, I don't think it's a bash vs. sh problem; as said - when I paste the parameter line in the original script it does work...
13:51:37 Join Senji [0] (~Senji@
13:52:07pamauryand what is the content of GCCOPTS normally ?
13:52:17MassaI thought it could be some kind of invisible special characters which gets removed when pasting...
13:52:44pamauryI suspect it's more of an expansion problem
13:52:47Massawhat do you mean by "normally"?
13:53:07pamauryif you don't change it and let the script run without modification
13:53:19pamauryI suspect it treats $CC $GCCOPTS
13:53:25pamauryas having one argument only
13:53:58 Quit TorC (Read error: Connection reset by peer)
13:54:59MassaGCCOPTS=-W -Wall -O -Wstrict-prototypes -pipe -std=gnu99 -fno-builtin -g -Wno-unused-result -mmmx -mno-ms-bitfields -I/usr/x86_64-w64-mingw32/sys-root/mingw/include/SDL -D_GNU_SOURCE=1 -Dmain=SDL_main -I$(SIMDIR) -Wno-pointer-sign -Wno-override-init
13:55:44 Join TorC [0] (
13:55:45 Quit TorC (Changing host)
13:55:45 Join TorC [0] (~TorC@fsf/member/TorC)
13:57:49Massaon the first sight it's the same - on the second sight: it does onky have one space between "-Wall" and "-O" and between "-O" and "-Wstrict-prototypes" whereas the other version has two...
13:58:12pamauryspaces you be irrelevant
13:58:18pamauryand what is the error message again ?
13:58:24Massathat's the reason why I think there could be some strange special characters which are not visible...
13:58:50Massai686-w64-mingw32-cpp: error: unrecognized command line option ‘-W -Wall -O -Wstrict-prototypes -pipe -std=gnu99 -fno-builtin -g -Wno-unused-result -mmmx -mno-ms-bitfields -I/usr/i686-w64-mingw32/sys-root/mingw/include/SDL -D_GNU_SOURCE=1 -Dmain=SDL_main -I$(SIMDIR) -Wno-pointer-sign -Wno-override-init’
13:59:00pamaurywhat's weird in this message
13:59:12pamauryis that it *seems* it is trated the entire line as one argument
13:59:50Massayes, you're right - how could that be?
13:59:51pamaurybecause cpp print an error *per switch* when it does recognize it, not the entire line
14:00:29pamaurythis is going to sound crazy but
14:00:38pamaurywhat happens if you replace the line by
14:00:38pamaurycpp_defines=`echo "" | $CPP $GCCOPTS -dD`
14:02:46Massayou mean backticks instead of "$()"? No change!
14:04:56MassaBTW, when I hardcoded paste the GCCOPTS line in my version of the script the error remains :-(
14:06:05pamauryhum, ok give me an hour, I'll install cygwin in a VM
14:06:28Massabtw, how do you setup a cross compiling environment in you linux? I could not manage to compile the SDL library...
14:06:40MassaI use cygwin64 if that matters...
14:07:14Massa(with both: mingw64 and mingw32 - so I could in principal compile it for both "worlds")
14:09:25pamauryI have never compiled sdl for cross compile but I don't think it's a problem, you first install mingw on linux and then compile sdl using it. There is a project called "mxe" that contains scripts to help that I think
14:14:12pamauryMassa: downloading, it may take a small while, I'll ping you when I have a working setup
14:15:00MassaIn the meantime I try to remove all what is currently unneeded in my script version and I try to integrate your idea by getting the correct directory with the cross gcc...
14:18:11 Join johnb3 [0] (
14:26:06chrisjjpamaury: Re ZEN, do you recall seeing anything supporting the notion here that there's at least two distinct LCDs used?
14:29:57pamauryI maintain there is no evidence for that though
14:30:06pamauryCreative has code for one LCD only
14:33:18 Join Senji_ [0] (~Senji@
14:37:11 Quit Senji (Ping timeout: 258 seconds)
14:38:55 Join Senji [0] (~Senji@
14:39:54 Quit johnb3 (Quit: Nettalk6 -
14:40:48***Saving seen data "./dancer.seen"
14:42:06 Quit Senji_ (Ping timeout: 255 seconds)
14:42:31 Quit StaticAmbience (Remote host closed the connection)
14:46:12 Join StaticAmbience [0] (
14:47:02 Quit shmibs (Quit: leaving =o)
14:50:55chrisjjOK, then I'll scratch that as a candidate cause for the LCD differences across units.
14:52:26pamauryI mean as usual take it with a grain of salt, reverse engineering is an art, the potential for mistakes is high
14:53:44pamauryI have found a difference between ZEN and ZEN X-Fi code, it's bit puzzling
14:54:23chrisjjDo we have any better candidate explanation for the three different LCD performances I see?
14:56:15chrisjjI mean: better than different LCDs.
14:57:44pamauryI don't have an explanation yet, except that the code must be doing something wrong and leads to unpredictable behavior
15:00:23 Join shmibs [0] (
15:00:56 Quit shmibs (Client Quit)
15:01:41 Join StaticAmbience_ [0] (
15:02:22 Join shmibs [0] (
15:02:31chrisjjWrong code can given different results only if the hardware is different.
15:02:35chrisjjSo it sounds to me like different LCDs is the best explanation we have.
15:03:44chrisjjYou suggest Creative has code for one LCD only, but we don't know actually know that.
15:04:04 Quit StaticAmbience (Ping timeout: 240 seconds)
15:04:25chrisjjWe know only that the Creative code disassembly shows no detection of different LCDs.
15:04:58chrisjjIt may be that the same Creative code drives different LCDs fine.
15:06:05chrisjjWithout detecting the LCD type.
15:06:25pamaurythat is incredibly unlikely though, and even if was true, I don't see the point because we couldn't possibly know that thus there is no difference one lcd or several lcd.
15:06:26pamauryAlso if you do something unpredictable, the result can be different on each device, but no two LCD are exactly identical anyway
15:07:18 Quit shmibs (Quit: leaving =o)
15:08:42 Join shmibs [0] (
15:10:16pamauryso the real only difference I can see betwen ZEN and ZEN X-Fi code is that the X-Fi resets the LCD but the ZEN apparetly does not
15:12:00 Join skapazzo [0] (~skapazzo@
15:12:04 Quit StaticAmbience_ (Remote host closed the connection)
15:14:36 Join StaticAmbience [0] (
15:15:05Massa__builtin: are you here?
15:15:12 Quit dfkt (Ping timeout: 245 seconds)
15:17:02pamauryMassa: cygwin installed, I need to get rockbox and see if I miss libs
15:23:01pamaurychrisjj: one thing we can try to match closer to the OF is to avoid powering the LCD on/off but rather putting it in standby mode
15:23:23pamauryI'll send you a build to try
15:24:23pamaury(I think OF uses standby)
15:24:50pamaury(or might not power down at all)
15:28:25gevaertspamaury: I have no idea!
15:28:39pamaurygevaerts: about what ?
15:28:42*pamaury forgot
15:28:49gevaertsThe wiki
15:29:03pamauryok, I thought we ran into this issue previously
15:29:08gevaertsWell, maybe
15:29:23gevaertsBut I don't know anything more about the wiki than how to add people
15:30:49lebelliumgevaerts: compared to yesterday we lost themes for Zen X-Fi and Zen X-Fi Style on the theme website
15:32:48 Join Senji_ [0] (~Senji@
15:33:23gevaertslebellium: the server went away every time I tried to re-run checkwps, so I went to sleep
15:34:24gevaertsAlso, technically yesterday the Zen X-Fi and Zen X-Fi Style weren't on the theme site at all!
15:34:46lebelliumI mean once they were
15:35:01gevaertsYes, but that was after midnight :)
15:35:07 Join n3m9 [0] (
15:35:10lebelliumok you win
15:35:24 Quit Senji (Ping timeout: 256 seconds)
15:37:12*gevaerts has another go
15:41:42 Join dfkt [0] (~dfkt@unaffiliated/dfkt)
15:42:21 Quit prg318 (Ping timeout: 258 seconds)
15:43:47chrisjjpamaury: I don't see why the same Creative code driving different LCDs is incredibly unlikely. The difference could be immaterial to the particular way the Creative codes drives, especially if that coding was deliberate to avoid differences.
15:46:06chrisjjThe point of considering this is that it recommends RB drive very closely follows the OF drive. Which is not to say your RB drive doesn't already, but perhaps closer is needed.
15:46:49Massapamaury: I've a new version of the script - with that (much easier) I was able to compile a running x86 simulator - x64 version is in progress :-)
15:46:50pamauryI don't think you know how LCD works. There are litteraly no two different LCD that have the same init sequence
15:47:21pamauryMassa: I finally rockbox and cygwin so let me try a sim build
15:47:28chrisjjAnd of course your suggestion of standing-by the LCD fits with that.
15:48:09chrisjj... provided the difference we see is after the current power down.
15:48:40chrisjjpamaury: I accept there are literally no two different LCD that have the same init sequence known to you.
15:48:41pamauryMassa: I'm not familar with cygwin. What is the different between i686-w64-mingw32 and i686-pc-mingw32 ?
15:48:53Massa(I got a lot of warnings - mostly format warnings and an error in puzzles - which I fixed by a small edit in rbwrappers)
15:49:05chrisjjBut while there's no better explanation known to you, I think this has to stand as the best explanation.
15:49:46pamauryMassa: ok so if I run:
15:49:46pamauryCROSS_COMPILE=i686-w64-mingw32- ../tools/configure
15:49:46pamauryI get
15:49:46DBUGEnqueued KICK pamaury
15:49:46pamauryCygwin host detected
15:49:47***Alert Mode level 1
15:49:47pamauryconfigure didn't find sdl-config, which indicates that you
15:49:51Massapamaury: Huh? I've i686-w64-mingw32 and x86_64-w64-mingw32
15:50:12Massathe latter is the 64-bit version...
15:50:16 Quit StaticAmbience (Remote host closed the connection)
15:50:17pamauryI didn't install the 64-bit compiler, but I have an extra "pc" one
15:51:13chrisjjBTW, one of the display differences could be due to non-LCD hardware differences. This is the dots.
15:51:13pamaury i686-pc-cygwin-gcc
15:51:14pamauryI guess it targets the cygwin environment
15:52:53 Join StaticAmbience [0] (
15:53:17pamauryMassa: can you pastebin your diff to fix sdl-config ?
15:53:50 Quit robertd1 (Ping timeout: 240 seconds)
15:57:45Massapamaury: did you install the 32- or the 64- version of cygwin?
15:58:50pamauryI installed cygwin64 but I must have chosen the 32-bit compiler in the (very long) list
15:58:55pamauryI don't think it matters anyway
15:59:48***Alert Mode OFF
16:00:13pamauryMassa: how do I tell cygwin to install more package ? Do I have to re-run the installer ?
16:00:45MassaYes, re-run the installer anytime you want to install more packages or to update existing ones...
16:01:14Massathe 64-bit compilation stops now with an error:
16:01:19Massamake: *** [/cygdrive/d/Prog/Projekt/RockBox/rockbox.git/lib/rbcodec/codecs/codecs.make:189: /cygdrive/d/Prog/Projekt/RockBox/rockbox.git/build.ipodvideo.sim/lib/rbcodec/codecs/libopus/silk/resampler_private_AR2.o] Error 127
16:01:37 Join robertd1 [0] (
16:01:54 Part robertd1
16:01:59 Join robertd1 [0] (
16:03:34MassaSo do you think i686-pc-mingw32 is the 32-bit version for 32-bit compiler? And i686-w64-mingw32 is the one for a 64-bit compiler?
16:04:53MassaIf yes, how can we distinguish / detect which one is present? And what about the old i586-mingw32msvc which is currently present in the script?
16:06:30pamauryno, I think i686-w64-mingw32 is the windows toolchain that targets windows (without dependency on cygwin) whereas i686-pc-mingw32 targets cygwin
16:06:58 Quit StaticAmbience (Ping timeout: 255 seconds)
16:07:14pamaury i586-mingw32msvc probably doesn't exists anymore, now afaik, there are only two toolchains: i686-w64-mingw32 and x86_64-w64-mingw32
16:08:03Massaat my installation x86_64-pc-cygwin seems to target cygwin64, so I assumed that i686-pc-cygwin would target 32-bit cygwin?
16:09:01pamauryyeah, but better use the w64-mingw32 I think
16:09:04pamauryanyway, in your diff
16:09:15pamauryI don't like the 64-bit option
16:09:20MassaBTW, I already tried to compile a "native" cygwin simulator - that shows a lot compile problems because of duplicated headers in rockbox' "firmware/libc" and /usr/include...
16:09:30pamaurythe script should detect 64-bit and use override it if necessary
16:10:27Massano, I don't think so - in prinicpal I'm able to choose between 64-bit and 32-bit in my environment (and the 64-bit compile does not work...)
16:10:45pamauryarg, /me hates windows
16:10:46 Join StaticAmbience [0] (
16:11:55Massayou can do the same in 64-bit Linux!
16:12:20MassaHuh? What does this mean:
16:12:30Massa 0 [main] make 4384 fork: child -1 - forked process 1008 died unexpectedly, retry 0, exit code 0xC0000142, errno 11 make: /cygdrive/d/Prog/Projekt/RockBox/rockbox.git/apps/plugins/rockboy/rockboy.make:18: fork: Resource temporarily unavailable 1
16:12:56Massa1091979 [main] make 4384 fork: child -1 - forked process 5144 died unexpectedly, retry 0, exit code 0xC0000142, errno 11
16:13:01pamauryyes but that's my point: if you don't say anything, it uses gcc (whatever default it uses). If you want to target *something else*, you have to do something (change compiler)
16:13:24pamauryadmitedly cygwin doesn't make it easy because there is no easily detectable compiler
16:14:15Massayes, you're right - I also wondered why they didn't integrate all the different versions in _one_ gcc (with target options)...
16:14:33MassaMaybe that's possible - but we don't know it ;-)
16:15:35pamauryyou can build a multilib gcc yes
16:15:37pamaurylike on linux
16:17:17Massado you know what all those "forked process died unexpectedly" mean which I get, when I try to build the 64-bit simulator version?
16:17:33pamaurynope :(
16:18:07pamauryMassa: reading your sysroot handling, I suspect it's going to fail on linux
16:18:25pamauryor other targets
16:18:40pamaurybecause I am not 100% that a sysroot is mandatory
16:19:04pamauryhum scratch that, it just prints a warning
16:20:24MassaNo, you're right - I should surround it by a if [ "$winbuild" = "yes" ]...
16:21:58MassaOr maybe it's not bad to keep it as is - and to only surround the warn messages with that "if"...
16:22:42pamaurydamn, cygwin is slooooooooow
16:22:54pamauryand doing things, I don't know what
16:23:14Massawhat do you mean?
16:24:39pamauryI don't know, the whole is just super slow
16:25:35MassaIt's slow - but it's also slow to compile in a VM...
16:26:26 Quit n3m9 (Read error: Connection reset by peer)
16:30:05 Quit pamaury (Ping timeout: 255 seconds)
16:30:15 Join pamaury_ [0] (~pamaury@rockbox/developer/pamaury)
16:31:54pamaury_Massa: your patch works, there are two things I don't like: 1) it touches unrelated lines (probably spaces) 2) in findsdl, I'm not a big fan of setting IFS=":" once, I find it clearer to set IFS before every loop
16:32:15pamaury_I'm trying to compile now (32-bit)
16:33:59 Join Senji [0] (~Senji@
16:34:22 Quit Petri152 (Ping timeout: 245 seconds)
16:35:13pamaury_Massa: did you compile with make -j by any chance ?
16:36:42 Join StaticAmbience_ [0] (
16:36:45 Quit Senji_ (Ping timeout: 248 seconds)
16:39:24pamaury_ah damn, make fails because it tries to call gcc
16:40:32chrisjjpamaury: The ZEN Unit N colours dots could be due to differences in SoC, not LCD. The fact it shows in the RB bootloader indicates it is not provoked by your LCD power-down.
16:40:43 Quit StaticAmbience (Ping timeout: 255 seconds)
16:40:52***Saving seen data "./dancer.seen"
16:41:19pamaury_Massa: did you run this issue as well ?
16:41:57 Join Petri152 [0] (
16:43:17MassaNo, I didn't have this issue
16:43:54chrisjjFor your suggested build to try, I vote for backlight sleep ding nothing else to the LCD i.e. no standby or powerdown. If that cures the flash (and I bet it will) then we can benchmark the battery runtime to see if there'd be any gain from a debugged LCD powerdown/standby.
16:44:19Massaabout touching the unrelated lines - that's because I configured the editor to remove trailing spaces when saving...
16:45:39Massaabout IFS - it never gets set back - do you think the correct way is to set it before the loop, reset it after it and again for the next loop?
16:46:36MassaI now try to compile the 64-bit version with "make -j 1" - and no, I didn't gave it a "-j" option before...
16:46:57pamaury_Massa: no, the correct way is for each loop to set it to what it wants. Because if you set it for several loops and of one them changes, you siltently break the other, not good
16:47:20pamaury_Massa: ok, I was thinking the -j could make cygwin run out of resource for processes but apparently that's not it
16:49:27Massais there any make configuration or environment variable which could change the default behaviour of "-j 1"?
16:49:32 Join guest6538 [0] (~user@
16:54:32 Quit guest6538 (Read error: Connection reset by peer)
16:54:32 Quit TorC (Read error: Connection reset by peer)
16:55:49Massapamaury_: with "make -j 1" my 64-bit compile also works and produces a running executable :-)
16:56:00 Quit StaticAmbience_ (Ping timeout: 240 seconds)
16:56:18 Join StaticAmbience [0] (
16:56:38 Join TorC [0] (~TorC@fsf/member/TorC)
17:08:10 Quit Strife89 (Ping timeout: 240 seconds)
17:08:13 Quit Petri152 (Ping timeout: 248 seconds)
17:08:56 Join Strife89 [0] (
17:34:03 Nick Massa is now known as massa_ (4fe8c676@gateway/web/freenode/ip.
17:47:09 Join Petri152 [0] (
17:47:29pamaury_massa_: so what was the problem ?
17:48:24pamaury_massa_: in your cygwin env, do you have a default compiler, ie if you type "gcc -v", does it print something ?
17:48:57massa_pamaury_: it seems my make does get called (somehow) with -j n (n > 1); the call with "make -j 1" fixes this...
17:49:35massa_pamaury_: yes, e.g. it prints "Target: x86_64-pc-cygwin"
17:49:39pamaury_chrisjj: I will send you a build with no lcd sleep. It will solve the white flash but not the lcd dot problem. And I'm not really sure about battery life
17:49:55pamaury_massa_: ah, I don't have such a compiler, that's why it fails for me
17:50:40massa_pamaury_: oh, do you think it should be possible to build without a "native" gcc?
17:51:23pamaury_no clearly it's not, it's just that in this case, we are not really "cross" compiling, we could use the same compiler for host and target (ie mingw)
17:53:16massa_I have to do some cleanups for the script changes and then I can put it to gerrit (if I manage how to do this) for review.
17:53:39massa_But not now - I'm away, at least for a few hours...
17:53:57pamaury_did you make changes to other files ?
17:54:41massa_BTW, I think the correct way in findsdl is to wrap the sysroot things with if [ -n "$CROSS_COMPILE" ]
17:54:51 Quit ZincAlloy (Quit: Leaving.)
17:56:10massa_pamaury_: there are a lot of warnings (for whom I want to have a look later) - but only one error in "apps/plugins/puzzles/rbwrappers.c"
17:56:37pamaury_yes sysroot should be conditional to cross compile
17:56:47 Quit Bilgus (Quit: Leaving)
17:57:32massa_pamaury_: that's why I wanted to talk to __builtin, there is a method named "vsscanf" in it - which conflicts with a system one...
17:58:35massa_pamaury_: it's only called once - a few lines above, so I just changed the name of that function from "vsscanf" to "sscanf_vsscanf" - that was it...
18:00:13massa_pamaury_: thanks for your help! When I'm back and cleaned up the changes at the scrip I'll ping you again - maybe today, maybe tomorrow :-)
18:00:44 Quit massa_ ()
18:05:03__builtindang it
18:05:05__builtintoo late
18:06:42 Join prg318 [0] (~prg318@deadcodersociety/prg318)
18:37:02pamaury_chrisjj: there ?
18:37:04 Nick pamaury_ is now known as pamaury (~pamaury@rockbox/developer/pamaury)
18:40:01 Join johnb2 [0] (
18:40:54***Saving seen data "./dancer.seen"
18:44:09 Quit johnb2 (Ping timeout: 240 seconds)
18:45:45pamaurychrisjj: could you check if fixed the lcd with coloured dots ?
18:58:16 Quit jhMikeS (Ping timeout: 256 seconds)
19:24:03chrisjjme returns.
19:24:38chrisjjpamaury: shall I test both, or one?
19:25:02pamaurychrisjj: both please
19:25:36pamaurynolcdsleep is the one where I disabled lcd sleep on backlight, so I expected to white flash but worse battery life and still coloured lcd dots
19:25:50pamaurydotclkpolfix is the one that tries to fix the coloured lcd dots
19:28:19pamauryI found a "bug"/mismatch in the LCD configuration and soc configuration that could result in unpredictable behavior and potentialy corrupted lcd data
19:36:54chrisjjOn Unit G (on which lcd_fix shows flash on backlight wake and sleep), rockbox_zen_dotclkpolfix 93c2fbb0d-170114 again shows flash on both.
19:37:23 Join johnb2 [0] (
19:37:33pamaurychrisjj: dotclkpolfix doesn't fix flash
19:38:05pamauryit's only supposed to fix the unit where you get funny dots (could you upload a picture of that btw ?)
19:41:45chrisjjOn Unit G, rockbox_zen_nolcdsleep 0cabc1fc5M-170114 shows no flash on sleep and a shorter and dimmer flash on wake. On sleep, the backlight fade doesn't reach zero.
19:42:17chrisjjNow I'll test on for coloured dots, requiring Unit N.
19:44:30 Join n3m9 [0] (
19:47:23chrisjjOn Unit N (on which the lcd_fix bootloader and .rockbox show coloured dots), rockbox_zen_dotclkpolfix 93c2fbb0d-170114 shows no change. And like other recent builds, it shows a data abort after the RB logo.
19:50:57chrisjjOn Unit N, rockbox_zen_nolcdsleep 0cabc1fc5M-170114 shows no change.
19:51:50chrisjjAnd like other recent builds, it shows a data abort after the RB logo, though two out of five runs showed a RB logo with keys non-responsive.
19:53:30 Quit johnb2 (Ping timeout: 240 seconds)
19:53:52 Quit __builtin (Ping timeout: 260 seconds)
19:54:21 Join __builtin [0] (
19:54:44 Nick __builtin is now known as Guest43679 (
19:55:57 Quit Guest43679 (Changing host)
19:55:57 Join Guest43679 [0] (~xray@rockbox/developer/builtin)
19:56:27pamaurychrisjj: can you take a picture of the data abort with dotclkpolfix ?
19:57:25 Nick Guest43679 is now known as __builtin (~xray@rockbox/developer/builtin)
19:57:32chrisjjBy Screendump, unlikely, since this function broke a while back.
19:58:12pamauryyou can't a screendump of a data abort, I mean picture, litteraly
20:02:24 Join ZincAlloy [0] (~Adium@2a02:8108:8b80:1700:8c6b:961d:b210:157a)
20:04:51chrisjjOK, fetching camera...
20:06:28pamauryI still don't understand the data abort, and why specifically on this device, why is it special ?
20:06:53chrisjjWebcam blinded
20:07:25pamaurycan you pastebin the whole text ?
20:07:38pamauryor at least all the pc address and stack trace if any ?
20:10:59chrisjjPhone blinded too.
20:10:59chrisjjconfig.cgf brightness is ignored by the Data Abort screen
20:11:16chrisjjTranscribing to pastebin...
20:11:34pamaurydata abort cannot access config.cfg...
20:30:23 Join johnb2 [0] (
20:31:08chrisjjI guessed data abort does not access config.cfg, but I'd hoped it would retain the brightness previously set by .rockbox from config.cfg.
20:31:46pamaurychrisjj: no because if a data abort happens when lcd is off there is no way to know the value of the setting, thus setting backlight to 100% is the only safe option
20:31:56pamaurychrisjj: does the data abort happen in the bootloader or in .rockbox ?
20:32:30chrisjjYup, makes sense. I was just hoping otherwise. Though I'd say 30% would do fine too :-)
20:32:55chrisjjRe the Data Abrt, as teh pastebin says, it follows the RB logo.
20:33:21pamauryah yeah the BL doesn't display a logo
20:33:51pamauryare you using a custom theme ?
20:34:30 Quit johnb2 (Ping timeout: 240 seconds)
20:35:13 Quit paulk-collins (Quit: Leaving)
20:39:19chrisjjTo the pastebin I added a description of all screen updates prior to the Abort.
20:39:33chrisjjNo, I am using factory defaults.
20:40:03chrisjjI.e. I am using your .ZIPs .rockbox with no added config.cfg.
20:40:57chrisjjThe coloured dots show in the BL and .rockbox, but I have seen the DA only in .rockbox.
20:40:58***Saving seen data "./dancer.seen"
20:42:26pamaurythe data aborts occurs in the skin engine
20:42:37pamaurydo you remove .rockbox entirely between tests ?
20:42:58chrisjjI rename it.
20:43:25chrisjjSkin engine, eh? As with previous builds, the DA varies (e.g. can be 'at 6004C7F8'). And occasionally is absent, whereupon I see the RB logo but keys are unresponsive.
20:44:04chrisjjThere's no BSoPO as one would expect from a freeze and angry watchdog.
20:45:22chrisjjAnd the coloured dots are flashing, as you'd expect if RB had got stuck in a loop that pacified the watchdog AND the dots are bytes from working memory getting fed to LCD by wayward DMA controller.
20:49:36chrisjjFWIW, deleting all cabbieV2 makes no difference.
20:50:58 Join Strife1989 [0] (~Strife89@
20:51:17 Nick Strife1989 is now known as Strife89|laptop (~Strife89@
20:51:21CtcpIgnored 1 channel CTCP requests in 0 seconds at the last flood
20:51:21*pamaury wonders what makes this particular unit special
20:52:16pamauryyou can try to benchmark the nolcdsleep if you want
20:52:22pamaurysee if that makes a big difference battery wise
20:56:08chrisjjpamaury, no thing I know makes this unit special except that it is the only one that shows the coloured dots and data abort on RB. On OF, it operates exactly as does other units, inc. other units that don't show dots and DA on RB.
20:58:07chrisjjAn interesting thing. If I start the unit from reset by connecting it to PC and then disconnecting, sometimes I get no DA and .rockbox runs.
20:59:06chrisjjDots are present. Menus are operable. WPS causes DA.
20:59:12chrisjjI have it running now. Any diagnostics I should extract?
21:01:08pamauryI can't think of anything
21:01:17pamaurythere must be something wrong but I have no idea what
21:01:38chrisjjOops, entering Text Editor gave DA. Undefined instruction at 61EC4500.
21:04:52chrisjjI reverted to a build in which Screendump is less broken, and with dots on screen got this: , .
21:06:16chrisjjThat clearly indicates the dots aren't in the the main memory frame buffer and hence are being created by e.g. the DMA, no?
21:10:26pamauryno I think are created by the LCD, or more specifically by a subtle mismtach in the timing with the lcd, but that's just a theory
21:10:49chrisjjSounds probable to me.
21:11:47chrisjjShame I can't show you an image. I think you'd find the dot pattern very illuminating.
21:12:50chrisjjZEN tester's Pro Tip: Though Screendump broken on recent builds and appears broken on Build 3f55f01-131201, that specific build does save the .bmp you want despite then saving a stream of repeats under successive filenames until it gives a Data Abort.
21:13:59chrisjjBTW, what means the "Scanning disc" popup that appears before the crash on some builds?
21:14:10 Join jhMikeS [0] (
21:14:57__builtinit's the database, I thinkj
21:18:30chrisjjWhat could trigger database activity on an install with no config.cfg? Some other user file??
21:21:36 Quit rudi_s (Quit: leaving)
21:21:41chrisjjpamaury, the dot pattern is invariant in layout, differing (and continuously) changing in colouration. There are typically 381 dots, arranged in 16 columns evenly spaced at 20 pixels, in approx. 24 rows.
21:23:42chrisjjEach dot is one non-black pixel, or two horizontally separated by about three pixels.
21:25:06 Join rudi_s [0] (
21:25:07chrisjjTwo pixels in the same dot are always different colour. Vertically adjacent dots are mostly the same colour. Horizontally adjacent dots always differ in colour.
21:27:31chrisjjThe first dot is on pixel row 0 or 1 (counting from top). The last dot is on approx pixel row 192.
21:30:18chrisjjVertically adjacent dots are eight pixels apart. Each dot is on the pixel row /below/ the dot immediately to its left.
21:32:51chrisjjThis means that the dot positions are as if something stepped through a 320-pixel-wide row-by-row frame buffer in memory order, each step being 336 pixels, and each footfall leaving a coloured dot.
21:32:54fs-bluebot3Gerrit review #336 at : 3SH gcc 4.6.3 with link-time optimization, for Archos targets by Boris Gjenero
21:33:30__builting 123
21:33:32fs-bluebot3Gerrit review #123 at : 3Add an alternative analogic touchpad sensitivity setting by Jean-Louis Biasini
21:33:38__builting 123
21:33:40fs-bluebot3Gerrit review #123 at : 3Add an alternative analogic touchpad sensitivity setting by Jean-Louis Biasini
21:34:03chrisjjPerhaps some of those numbers ring bells from the DMA process.
21:42:08pamaurydma doesn't do anything but pushes the pixel to the lcdif. It's the lcdif that controls how the data is sent. I am not entirely sure I visualize the thing correctly.
21:43:56 Join johnb2 [0] (
22:09:22chrisjjPushing pixels to the LCDIF is the only operation that can cause such a messup, surely?
22:10:10pamaurychrisjj: not so sure, if I had to choose, I would blame the lcdif rather than the DMA but who knows
22:10:10chrisjjDoes one frame's push occur in a single DMA operation, do you know? Or in smaller blocks.
22:10:20pamaurychrisjj: are you planning to ship me this unit ?
22:11:07chrisjjI was planning to ship you a unit that works best, but perhaps I should put this one that works worst in too.
22:11:22 Join massa [0] (4fe8c676@gateway/web/freenode/ip.
22:11:42pamauryno each frame uses, iirc, 4 linked DMA transfers
22:11:55chrisjj4? Why 4?
22:12:14massapamaury: I'm back again :-)
22:12:44chrisjjTo fit inside a 64K byte limit, perhaps?
22:13:51__builtinhello massa
22:13:53pamauryeach transfer can do 64K and the frame is 320 x 230 x 3 bytes
22:13:59__builtinsorry I wasn't around earlier
22:14:12massapamaury: I did some more changes in my script - and suddenly the error occured again...
22:14:47pamaurymassa: which changes ? I managed to build the simulator fine, after the vscanf problem
22:15:11massa__builtin: Hi, did you read the content of the chat of this day?
22:16:36 Quit jhMikeS (Ping timeout: 256 seconds)
22:16:40massapamaury: as I said - some cleanup and also some checks for the different possible CROSS_COMPILE variants for x86
22:16:56__builtinmassa: parts of it ;)
22:17:40__builtinI see that there's a naming conflict in sgt-puzzles, I can fix that
22:17:40massapamaury: and I found out, what the problem is - you were right, the CC call treats the GCCOPTS as _one_ parameter!
22:17:53massapamaury: and I know why this happens :-)
22:18:18pamaurymassa: haha! I suspected that but I don't know why ?
22:18:26pamaurybecause it works here
22:18:44massa__builtin: I found a problem in apps/plugins/puzzles/rbwrappers.c
22:18:53__builtinanything else?
22:19:56massa__builtin: no, I did currently not see your post :-)
22:20:27massapamaury: the reason is somehow setting IFS=":"
22:22:26massapamaury: I added another for loop with checks and a leading IFS=":" - and an echo of GCCOPTS before and after - before it's good and after it's not good...
22:23:00pamauryhum, interesting
22:23:12pamauryI wonder if IFS influences argument splitting in commands
22:23:43massaseems so - when I set back IFS to the initial value, everything is fine!
22:24:26pamauryit definitely influences echo
22:24:45pamaurymassa: I have a theory, switch to bash instead of sh
22:26:48massadoes not change a thing
22:27:17massamaybe IFS gets automatically restored after leaving a function?
22:27:20pamaurysomehow I would expect the IFS change to be local, but that may depend on how you write it
22:27:39massa(and my newer changes are not inside a function)
22:29:30pamaurymassa: then the simple solution is to unset IFS
22:29:38pamauryunset IFS
22:30:04massathe simplest solution in my case is not to set IFS and use space separated name lists :-)
22:32:17pamaurythat will paths with spaces though (which could break for a whole lot of other reasons I admit) but yeah you can do that
22:34:57massano, no - the loops over the PATH are all inside functions - I mean my new one with a list of potential CROSS_COMPILE names
22:35:35lebelliumgevaerts: I assume we have to give up :)
22:37:43massapamaury: I just checked it, after leaving a function the IFS is unset again...
22:40:59***Saving seen data "./dancer.seen"
22:41:33*[Saint] thinks if you're using paths with spaces in you deserve to get your shit broken
22:41:57[Saint](also - yes, field separators create some odd magic at times)
22:45:39 Quit skapazzo (Quit: Lost terminal)
22:48:21massaSaint: I don't think any cross compiler environment name (e.g. i686-w64-mingw32) will contain a space - so keeping them in a space separated list should imho O.K.
22:50:39[Saint]AFAIK, you're right, but you can't account for end users doing "weird shit" (TM).
22:50:44massapamaury: do you still have your VM environment with no native gcc?
22:50:56[Saint]But IMO spaces in paths is a cardinal sin.
22:51:33massapamaury: here's the configure diff of my latest version:
22:51:56massapamaury: this should also work in pure cross hosting environments...
22:53:20massa[Saint]: well, I assume possible spaces and other strange characters are the reason why the PATH separator is a ":" :-)
22:54:33massaSo you always have to expect the worst when handling with filenames and directory names...
22:56:27massa__builtin: BTW, in my self compiled (64-bit) simulator the keypresses works :-)
22:56:54__builtinmassa: sorry, in what context, again?
22:59:49massa__builtin: that was the reason why I started the whole thing - I had problems with the simulators compiled by rasher and then tried to compile them myself...
22:59:59__builtinohh, yeah
23:00:12__builtinI forgot for a second
23:00:14__builtinnice job!
23:01:23 Join johnb3 [0] (
23:01:39pamaurymassa: no I installed the pc-cygwin-blabla and I can just gcc in cygwin now
23:02:00 Quit johnb2 (Ping timeout: 258 seconds)
23:02:17pamaurymassa: thanks, I'll have a look at the diff later
23:02:30[Saint]we found the one guy who uses cygwin folks.
23:02:30[Saint]the searchis over.
23:02:42massapamaury: shit, then my two-liner to fix it for those environments cannot be tested :-)
23:07:42massa[Saint]: Well, I first tried to use a recent version of Ubuntu - but this doesn't worked either, I couldn't build the simulators...
23:08:19[Saint]wow - two people using cygwin in one day.
23:08:26[Saint]we are truly blessed by a miracle.
23:08:42 Quit Strife89|laptop (Quit: Leaving)
23:09:03massa[Saint]: So I gave cygwin a try again - nowadays I most of the time use windows at my laptop (had been different in the past)...
23:09:39massa[Saint]: oh, you didn't talk to me?
23:09:54[Saint]But...yeah, compiling for Windows is a royal pain in the tits and the environment can break if you look at it sideways.
23:10:08[Saint]massa: no, it was originally directed at pamaury
23:10:39massa[Saint]: pamaury only installed cygwin in a VM to help me find and fix the cross compiling issues :-)
23:10:57[Saint]aha, right.
23:11:11pamauryI usually don't compile on windows, I cross compile from linux but hey
23:12:48massapamaury: and you also compile the simulators for windows in your linux cross environment?
23:13:14pamauryI don't compile simulators for windows but I don't see why it can't be done
23:13:32pamauryyou need to cross compile sdl and then cross compile the sim
23:13:51pamaurythat doesn't sound too horrible, I mean compared to RBUtil for example
23:15:37[Saint]Oh, absolutely. It's just annoying, not particularly difficult.
23:16:04[Saint]Not having the statically linked SDL is what I think gets most people.
23:16:27[Saint]and perhaps the absolute myriad of params that govern it.
23:17:26massaHmm, I couldn't manage it in the first place and then searched in the Internet for somebody who managed it - and it seems to be difficult 8-)
23:18:15massaI wonder why there aren't precompiled packages for that - you can get packages for a lot of cross-compiled libraries, but not for SDL...
23:18:38massaBTW, which of the SDL libraries does rockbox use?
23:19:14[Saint]I legitimately have no idea and I made absolutely zero effort to find out.
23:19:27[Saint]I just compiled _ALL THE THINGS_ in my static binary.
23:19:51[Saint]bluebrother would be able to tell you.
23:20:25[Saint]oh, actually, maybe not.
23:20:40[Saint]I imagine rbutil and the sim have radically different requirements.
23:20:41massaAnd what's difficult with rbutil?
23:21:07pamaurymassa: I think we just use the basic SDL library
23:21:21pamaurymassa: for rbutil you need all of Qt
23:22:15massaThat's easy - I often compiled Qt in a lot of environments :-)
23:24:04pamauryWell apparently bluebrother runs into problem with accessibility stuff
23:24:17massaBut I never tried to compile rbutil - never saw a reason to do that...
23:25:58massaWhat about all those warnings, especially in DEBUGF formats, when I compile it? Don't they come in linux?
23:26:09[Saint]the current reason is if you have an iPod 6G and want to test the bootloader installation automation.
23:26:43[Saint]They do. But warnings are warnings. If it was a problem they'd be errors.
23:27:02[Saint]I generally just tell it to shut up about warns.
23:27:27massa[Saint]: yes I have one and tried to put SDXC cards in it - I gave up (nothing to do with rockbox)
23:28:27[Saint]Lots of people seem to have a wide range of issues with SD or CF based solid state conversion in the ipods.
23:28:34[Saint]I guess I've been relatively lucky.
23:28:57massa[Saint]: about warnings: I have another opinion about warnings - they have a reason and point to potential problems; if they can be fixed they should
23:29:24[Saint]That reminds me that I still want to try an ipod6g build with multivolume in a large format disk and see if the OF is fine running off the first partition.
23:30:54massa[Saint]: I tried that - and the OF still corrupted my disk; everything is fine if you use cards below the magical 128G limit - but you'll have trouble if you use bigger ones.
23:31:29pamaurymassa: warning of formats are really annoying
23:32:07massa[Saint]: All that was the reason why I bought an old iPod Video 5.5 - with that everything works (but with really really slow transfer rates from PC to Pod)
23:32:11[Saint]massa: yeah, bummer, I kinda had a feeling it would do that...well, you saved me from wasting that time I guess.
23:32:16pamaurybecause windows and linux don't handle the same format, and rockbox itself probably doesn't have a very standard implementation anyway
23:32:57 Quit user890104 (Quit: .)
23:33:18 Join user890104 [0] (Venci@unaffiliated/user890104)
23:33:34massapamaury: what is that z prefix in the format (e.g. %zu)?
23:35:57massaaah, I see it's actually a C99 addition for size_t
23:38:23massaso why are the global CCOPTS (which contain -std=gnu99) not used in my case?
23:39:30massaI mean it's used, but does not show a effect?
23:41:29pamaurymassa: iirc, windows's printf doesn't know about %zu
23:42:08pamauryI'm not 100% sure
23:42:32pamauryhonestly that's really really low priority
23:48:46massapamaury: normally it should link against a mingw gcc libc version which should support it or not?
23:52:35massamaybe we should just add "-Wno-format" to get rid of those warnings?
23:56:59massaat least in the windows cross compiles...
23:57:15massa(and that we see other, more important warnings...)

Previous day | Next day