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 2017-04-30

00:11:22 Quit lorenzo92 (Ping timeout: 260 seconds)
00:12:32lorenzo92_1saratoga: the language looks good ;-)
00:13:31lorenzo92_1lebellium: do you have a "list" of key mappings to be fixed? perhaps we can also track them in the bug tracker
00:32:54 Quit alucryd (Remote host closed the connection)
00:33:40__builtinprof_wolfff: after reading the logs, since there seem to be some issues with rbutil on mac, how about we just relegate it to unstable for 3.14?
00:34:38 Join alucryd [0] (~quassel@archlinux/developer/alucryd)
00:37:44 Join lorenzo92 [0] (~AndChat52@
00:40:18 Quit lorenzo92_1 (Ping timeout: 260 seconds)
00:57:25 Quit zoktar (Ping timeout: 240 seconds)
00:58:25 Quit olspookishmagus (Ping timeout: 240 seconds)
01:00:11 Join olspookishmagus [0] (
01:00:16 Join zoktar [0] (
01:00:16 Quit zoktar (Changing host)
01:00:16 Join zoktar [0] (~zoktar@unaffiliated/zoktar)
01:00:33 Nick olspookishmagus is now known as Guest80758 (
01:01:24 Quit JanC (Remote host closed the connection)
01:02:44 Quit PimpiN8 (Quit: My MacBook has gone to sleep. ZZZzzz…)
01:03:37 Join JanC [0] (~janc@lugwv/member/JanC)
01:04:20__builtinthere seems to be an incorrect mutex_lock() call somewhere on the Zen:,51757.msg239602/topicseen.html#msg239602
01:18:29 Quit lorenzo92 (Quit: Bye)
01:19:34***Saving seen data "./dancer.seen"
01:25:34 Quit cc___ (Ping timeout: 258 seconds)
01:26:20prof_wolfff__builtin: i agree, it will let bluebrother and myself more time to solve and testing the Mac and cryptopp issues, ATM i am a bit busy but anyways hope to do that soon as it is my priority task
01:28:21__builtingreat, I'll edit the release notes
01:30:43prof_wolfffanyway, actually users can use the mks5lboot tool to install/uninstall the bootloader on all platforms (for Mac it needs intel OS X 10.5+ or 10.6+, not sure)
01:55:16 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
02:18:10__builtinpamaury: someone found an incorrect cpu mode panic on the zen
02:19:59pamaury__builtin: hi, ah damn, I didn't bring my xfi2 with me and I'm away until the 20th of May. I'll try to have a look at what code could be responsible
02:21:52__builtindoes the build server have .map files available?
02:22:55pamauryI don't think so which is annoying :( Otherwise I can just send the guy a custom build of which I have the file.
02:25:40__builtinI'm going to assume that he's running the last available dev build
02:26:19pamaury__builtin: I think I know what the problem is
02:27:15pamaurybut I am not 100% sure I can fix the problem without testing on target. I'll have a try at it by sending a custom build to the guy on the forum
02:27:36__builtinok, great
02:27:38 Quit cephalopoid (Ping timeout: 260 seconds)
02:29:04__builtinseems to be in imx233_lradc_acquire_channel from some detective work
02:30:04__builtinwhich calls arbiter_acquire() which calls mutex_lock()
02:31:12 Quit JanC (Remote host closed the connection)
02:32:36 Join JanC [0] (~janc@lugwv/member/JanC)
02:34:48pamaurythe thing is that it calls arbiter_acquire with TIMEOUT_NOBLOCK, under the assumption that it would not mutex_lock, which is not the case
02:35:18pamauryI'll rewrite the code different and acquire a channel once at init
02:47:49 Join PurlingNayuki [0] (~Thunderbi@
02:52:12 Quit PurlingNayuki (Ping timeout: 258 seconds)
02:52:50 Quit paulk-collins (Ping timeout: 260 seconds)
02:55:06 Quit MrZeus (Ping timeout: 240 seconds)
03:01:36 Quit ZincAlloy (Quit: Leaving.)
03:01:46 Join ZincAlloy [0] (~Adium@2a02:8108:8b80:1700:25d4:2f0c:9afb:8c92)
03:04:30 Join jhMikeS [0] (
03:06:18 Quit ZincAlloy (Ping timeout: 252 seconds)
03:08:34 Quit xorly (Ping timeout: 245 seconds)
03:19:37***Saving seen data "./dancer.seen"
03:22:11 Quit snw (Remote host closed the connection)
03:28:11 Quit prg318 (Read error: Connection reset by peer)
03:30:39 Join prg318 [0] (~prg318@deadcodersociety/prg318)
03:48:46 Join paulk-collins [0] (~paulk@
04:21:36 Quit jhMikeS (Ping timeout: 260 seconds)
04:24:34 Quit mc2739 (Ping timeout: 240 seconds)
04:51:57 Join cephalopoid [0] (
04:52:47 Quit cephalopoid (Remote host closed the connection)
04:55:04 Quit JdGordon (Ping timeout: 240 seconds)
04:55:21 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
05:19:40***Saving seen data "./dancer.seen"
05:45:20 Join jhMikeS [0] (
06:08:06 Quit alexbobp (Quit: Lost terminal)
07:02:27 Quit TheSeven (Disconnected by services)
07:02:36 Join [7] [0] (~quassel@rockbox/developer/TheSeven)
07:19:41***Saving seen data "./dancer.seen"
08:24:11 Quit michaelni (Ping timeout: 260 seconds)
08:36:31 Join michaelni [0] (
08:44:25 Join alexbobp [0] (
08:44:47 Nick alexbobp is now known as Guest53237 (
09:03:06monotuxprof_wolfff: yes, I have dual SD cards instead. I'm not sure if I'm using the latest bootloader.
09:09:45 Quit jhMikeS (Ping timeout: 260 seconds)
09:19:44***Saving seen data "./dancer.seen"
09:41:45 Join jhMikeS [0] (
09:59:04 Quit Horrorcat (Quit: Gateway shutdown)
10:08:59 Join Horrorcat [0] (129994c4c6@2a01:4f8:171:1d56:1010:1010:3432:26)
10:08:59 Quit Horrorcat (Changing host)
10:08:59 Join Horrorcat [0] (129994c4c6@unaffiliated/horrorcat)
10:09:01 Part Horrorcat
10:11:28 Quit JdGordon (Ping timeout: 240 seconds)
10:50:41prof_wolfffmonotux: what is exactly your problem?, the power consumption peak occurs when HDD spin-up but should not happen on SD cards, does it work if you boot into Apple original firmware? (to enter original firmware press and hold MENU when the ipod is powered-off)
10:52:25monotuxI thought 3653 mW was pretty much juice left to run it with hence the question. haven't tried with original firmware.
10:55:25prof_wolfffif the battery voltage is too low then the bootloader waits until there is sufficient charge to boot, then it should boot into rockbox, what do you see exactly on the ipod screen?, maybe you USB cable is bad and the iPod is not charging, or the battery is broken...
10:57:39monotuxah, sorry if I'm not very clear. :) the bootloader says it wont start as the voltage is low (and gives a number of the current charge, for example 3653 mW) but I thought that the numbers given was more than enough to boot it. since the bootloader is unsupported I thought it was some kind of leftover from development etc.
10:58:10monotuxthe ipod charges OK (new battery) and everything
10:58:40monotuxso maybe this is simply caused by my lack of electricity knowledge and is indeed a feature. :)
10:59:44prof_wolfffif the ipod charges ok, then the 3653 mV should raise slowly until it reaches an appropiate charge level (cant recall exactly the level), and then it should boot into Rockbox, if the level does not raises then the battery is not being charged
11:01:26prof_wolfffi suppose you iPod is connected to your computer using the USB cable (or use a external charger)
11:02:34 Join lorenzo92_1 [0] (
11:03:51 Join johnb3 [0] (
11:06:14monotuxyes, it all works ok but the issue was that I thought that 3653 mW (or whatever value it gives when not starting due to a too low voltage) was more than necessary :)
11:08:07duo8would a stock ipod boot with that?
11:09:01prof_wolfffmonotux: if the charge level is too low then the bootloader does not launch Rockbox until you charge it, it is necessary because if the charge is low then the HDD cannot be powered-on and we want to avoid that
11:12:08prof_wolfffthe CF will boot with less charge but not the stock HDD
11:13:37prof_wolfffanyway if you boot with let's say 3500 mV, the running time will be very small and you will need to charge it
11:16:13 Join ZincAlloy [0] (~Adium@2a02:8108:8b80:1700:2d4a:d60c:97fc:48c6)
11:16:22prof_wolfffactually the bootloader does a slow charge (100mA) but in future i hope to implement the fast charge at 500mA, so the charge time in bootloader will be smaller than now
11:17:20monotuxah, cool. :)
11:18:43prof_wolfffyou can try to boot into original firmware, it does a similar thing, i think it charges also at 100mA but i am not sure, so try it and see
11:19:46***Saving seen data "./dancer.seen"
11:25:30 Join cc___ [0] (
11:44:28 Join MrZeus [0] (~MrZeus@2a02:c7f:7066:fb00:e4b3:dad8:e900:8f8b)
11:47:58 Quit johnb3 (Ping timeout: 268 seconds)
11:54:55 Join TheLemonMan [0] (~root@irssi/staff/TheLemonMan)
12:13:14lorenzo92_1lebellium: g#634 and g#1598 ... the only one it is worth testing is the one on ypr0, so that you can even start thinking about a theme for the rec screen :-)
12:13:17fs-bluebotGerrit review #634 at : Samsung YP-R0: record FM radio by Lorenzo Miori
12:13:17fs-bluebotGerrit review #1598 at : Samsung YP-R1: record FM radio and Microphone by Lorenzo Miori
12:13:54lorenzo92_1lebellium: on ypr1, keymaps are required and not playback recovery is yet working (i.e. needs reboot before listening to recorded files)
12:15:15lorenzo92_1as a general note: pathces are based off g#633 which is a generic alsa improvement, still work in progress, but aims to improve all targets that are using alsa as audio backend
12:15:16fs-bluebotGerrit review #633 at : pcm-alsa: recording support and improvements by Lorenzo Miori
12:15:34lorenzo92_1*aims at improving
12:21:52 Join lebellium_z3c [0] (~AndChat73@
12:22:57lebellium_z3clorenzo92_1: OK thanks, I'll try in 2 weeks
12:23:35lebellium_z3cAs for R0 keymap, I suggest you compare hold / stop / poweroff in an old vs fresh build
12:24:39lebellium_z3cNow Lisbon is waiting for me, bye :)
12:24:49 Quit lebellium_z3c (Client Quit)
12:28:40 Quit lorenzo92_1 (Ping timeout: 268 seconds)
12:41:34prof_wolfffmonotux: you are somewhat right about the battery charge threshold is too high, it was calculated for the worst case plus a margin: it should have enough power to carry on the second stage of the database initialization for a full 160GB HDD, it lasts for a few minutes (depending on the number of song files) while the HDD does reads/writes draining ~200mA from battery and 100mA from USB, but i see that due to some recent commit thi
12:41:34prof_wolfffs second stage DB initialization is gone!, so this threshold could be decreased probably 100mV or more
12:46:03 Quit jhMikeS (Ping timeout: 240 seconds)
12:48:52 Join johnb3 [0] (
13:02:01 Join xorly [0] (
13:04:42 Join lorenzo92_1 [0] (
13:08:35 Join robertd1 [0] (~root@
13:12:39 Join mc2739 [0] (~mc2739@rockbox/developer/mc2739)
13:17:03 Join jhMikeS [0] (
13:19:47***Saving seen data "./dancer.seen"
13:49:09 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
13:54:06 Quit pamaury (Ping timeout: 240 seconds)
13:59:57*lorenzo92_1 silly samsung spaghetti kernel, is pulling the pin low, it is now clear why the playback was unable to resume ^^
14:00:08*lorenzo92_1 *mute pin
14:00:33*lorenzo92_1 when performing the snd_pcm_close call
14:11:15 Quit jhMikeS (Ping timeout: 246 seconds)
14:18:00 Join PimpiN8 [0] (~textual@2a02:a454:38ea:1:a91c:61bd:d652:dc0f)
14:31:11 Quit lorenzo92_1 (Ping timeout: 246 seconds)
14:58:29 Part robertd1
14:59:03bluebrotherhmm. I cannot cross-compile mks5lboot for Windows because of the use of nanosleep(). Why is nanosleep used at all?
14:59:20bluebrotherand macro clashes with Windows headers ... :(
14:59:31 Quit johnb3 (Ping timeout: 255 seconds)
15:13:44bluebrotherwtf. Used Ipod 1G on ebay for 890EUR :o
15:16:47prof_wolfffbluebrother: first i used usleep() and Sleep() but people reports problems with Sleep() and usleep() seems deprecated and also got problems, i found nanosleep() the more portable function, note that i used minGW, i suppose you are using MSVC
15:17:33bluebrotherprof_wolfff: no, I'm using mingw-w64, currently cross compiling with MXE
15:18:04bluebrotherchecking the docs shows that nanosleep() does not exist in mingw, since that only has the functions that are in msvcrt.dll
15:18:21bluebrotherbut obviously you were able to compile that`
15:19:18bluebrotherbut why are there problems with Sleep()? Should do exactly the same ...
15:19:52***Saving seen data "./dancer.seen"
15:20:55prof_wolfffi am also very confused with mingw windows compilation, probably it depends on how the development system is configured
15:21:13prof_wolfffthe problem with Sleep() is that sometime it wont sleep
15:21:33prof_wolfffbetter said, on some compilations it wont sleep
15:21:51bluebrotheroh? Never heard of that
15:22:41prof_wolfffit works well for me but some executables compiled by other people got 0 seconds sleep, nanosleep() seems to be posix
15:23:18bluebrotherstrange. Sounds a but in their toolchain then
15:23:48prof_wolfffyes, probably it was the toolchain or the mingw version, or...
15:24:02bluebrotherbut anyway, I'll get rid of the use of uname in the Makefile. That's pretty likely to break (MinGW doesn't come with it, you need to install msys for that), and libtools.make already has something to handle that
15:24:17bluebrotherby querying the compiler, which also works for cross-compiles :)
15:26:01prof_wolfffi think it was copied from other target, really i am not used to windows compilations and almost all these things come from other targets
15:28:21bluebrotherI'll make some adjustments for Windows. Then lets see if others run into issues with that. I don't have any target to test this.
15:29:25prof_wolfffnice, push it to gerrit or send it to me so i can test it on XP with mingw
15:29:38prof_wolfffwhat about nanosleep(), what do you suggest?
15:30:23bluebrotheryes, that's the standard function on Windows for that
15:34:08 Join robertd1 [0] (~root@
16:09:51 Join johnb2 [0] (
16:16:37 Quit bertrik (Quit: "This incident will be reported")
16:17:22 Quit TheLemonMan (Remote host closed the connection)
16:20:24 Quit WakiMiko (Remote host closed the connection)
16:22:02 Quit JanC (Ping timeout: 260 seconds)
16:24:22 Join WakiMiko [0] (~WakiMiko@unaffiliated/wakimiko)
16:25:32 Quit WakiMiko (Client Quit)
16:33:25 Join WakiMiko [0] (~WakiMiko@unaffiliated/wakimiko)
16:34:28 Quit johnb2 (Ping timeout: 255 seconds)
16:37:44 Join johnb3 [0] (
17:18:30 Join Saratoga_ [0] (47e99464@gateway/web/freenode/ip.
17:19:05Saratoga___builtin: I think you don't need to mention unstable in the 3.14 release notes
17:19:45Saratoga_Usually we just put stuff related to the released devices, so either the classic is getting a release or it's not and we shouldn't mention it
17:19:55***Saving seen data "./dancer.seen"
17:21:25 Quit PimpiN8 (Quit: My MacBook has gone to sleep. ZZZzzz…)
17:24:55 Quit dys (Read error: Connection reset by peer)
17:33:17johnb3Need help on git: I have a local commit (German translation). git log shows the commit id, git status mentions nothing. How can I upload that commit to gerrit? I tried 'git commit −−amend' but then it finds pamaury's commit for Sony support and errors out because email names don't match.
17:33:36johnb3I had done a rebase in between.
17:35:29 Join dys [0] (
17:35:48 Quit Saratoga_ (Ping timeout: 260 seconds)
17:49:31__builtinsaratoga: alright
17:49:55 Quit robertd1 (Ping timeout: 240 seconds)
17:50:48__builtinI'll go ahead and start working my way down ReleaseChecklist
17:54:27__builtingevaerts: you around?
17:58:07__builtinhmm, there might be a slight issue with pushing tags
17:58:48__builtinI'm not in the Release Committers group, and I can't add myself :(
18:01:51 Join PurlingNayuki [0] (~Thunderbi@
18:11:50 Join PurlingNayuki1 [0] (~Thunderbi@
18:12:05 Quit PurlingNayuki (Remote host closed the connection)
18:12:05 Nick PurlingNayuki1 is now known as PurlingNayuki (~Thunderbi@
18:16:21 Join johnb2 [0] (
18:25:28 Join PurlingNayuki1 [0] (~Thunderbi@
18:25:46 Quit PurlingNayuki (Read error: Connection reset by peer)
18:25:47 Nick PurlingNayuki1 is now known as PurlingNayuki (~Thunderbi@
18:27:37johnb3My git problem is resolved: I was on my own branch and now changed back to master.
18:34:54 Join TheLemonMan [0] (~root@irssi/staff/TheLemonMan)
18:36:59gevaerts__builtin: I am, mostly
18:38:30gevaertsAh, I am in the Release Committers group. That *should* mean I can do tags and branches remotely (although probably not from the web interface)
18:39:42gevaertsSo just ask me, bluebrother, prof_wolfff, wodz, or funman for what you need, and that shouldn't be a blocker
18:54:21 Join JanC [0] (~janc@lugwv/member/JanC)
19:02:09 Join lorenzo92 [0] (~AndChat52@
19:02:13 Join PurlingNayuki1 [0] (~Thunderbi@
19:02:16 Quit PurlingNayuki (Remote host closed the connection)
19:02:16 Nick PurlingNayuki1 is now known as PurlingNayuki (~Thunderbi@
19:02:26 Quit johnb2 (Ping timeout: 240 seconds)
19:06:08__builtincan you create a v3.14 branch?
19:07:55gevaertsBased on master as it is now?
19:09:21__builtinnot yet, actually
19:09:50__builtinwell, it works
19:10:04*gevaerts just did it :)
19:10:39__builtinI just bumped all the versions in the scripts
19:12:30gevaertsAs long as things don't diverge, git merge trivially handles that sort of thing
19:17:03__builtinalright, now what?
19:19:56***Saving seen data "./dancer.seen"
19:20:39 Quit johnb3 (Quit: Nettalk6 -
19:22:09 Join PurlingNayuki1 [0] (~Thunderbi@
19:22:09 Quit PurlingNayuki (Remote host closed the connection)
19:22:10 Nick PurlingNayuki1 is now known as PurlingNayuki (~Thunderbi@
19:22:15gevaertsIs tools/ up to date?
19:23:00 Quit cc___ (Quit: WeeChat 1.6)
19:23:29__builtinhmm, crap
19:23:42__builtinpushing that seems to have automatically updated the website too
19:23:47__builtinbut the links are all broken
19:24:27__builtinI'll have to revert the last commit on master
19:24:32gevaertsThat looks at master, so you should be able to revert it there
19:26:21__builtinI guess all that's left is to build the bins
19:26:33__builtin(and hope they actually compile)
19:27:29gevaertsYes, I'm trying to run tools/release/ here
19:27:43gevaertsWe'll see what happens :)
19:36:14gevaertsSo I guess the plan is to release 3.14, hope nothing annoying is broken, and if anything is (and someone fixes it!) release 3.14.1?
19:37:28 Join PurlingNayuki1 [0] (~Thunderbi@
19:37:28__builtinyep! :D
19:37:35 Quit PurlingNayuki (Remote host closed the connection)
19:37:35 Nick PurlingNayuki1 is now known as PurlingNayuki (~Thunderbi@
19:38:28gevaertsIt's all a question of names, really. Do you call it .0 or release candidate? :)
19:39:18*__builtin is confused
19:39:26__builtinit's 3.14, nothing else
19:39:51gevaertsWell yes
19:40:06gevaertsWe don't do .0 on intial releases, because we're optimistic
19:40:21*gevaerts grumbles. Version number are not decimal!
19:40:27__builtinwe could always go the way of TeX...
19:40:57gevaertsWe could, but that's been done, really
19:41:15gevaertsAnd I'm fairly sure it confuses more people than it amuses
19:41:43__builtinwho has access to the server?
19:42:06__builtinjust the swedes?
19:46:08gevaerts__builtin: things don't build due to something in a plugin called "puzzles". Do you know who might know about that one?
19:46:55__builtinwhat's the error, and what target?
19:48:13gevaertsIt doesn't find keymaps.h, on (for now, alphabetical order I believe) creativezenmozaic, creativezenxfi, creativezenxfistyle, creativezenxfi3
19:49:18*__builtin remembers fixing that!
19:50:39__builtinone sec
19:56:59__builtinpushed a fix
19:58:17gevaertsOK, starting a new build run
20:02:08user890104does anyone know why i get "error - Unsupported compression 335675392" when trying to run bmp2rb?
20:02:34user890104the compression type is 0x00000000 according to my hex editor
20:02:36__builtinis the bmp compressed?
20:02:53user890104it's tools/iaudio_bl_flash.bmp
20:02:55user890104in git
20:03:26__builtinworks for me
20:03:30user890104trying to build in mingw64
20:03:34user890104on windows 7 x64
20:03:53user890104gcc 6.3.0
20:05:00__builtinah, I think I know what it is
20:05:13__builtinint readint(void* value) subscripts a void* pointer, which is undefined
20:05:26__builtintry making it a char* pointer
20:05:41gevaertsBah, random errors due to makefile race conditions :(
20:05:48*gevaerts will retry the failed builds later
20:05:53user890104uhm... what's wrong with void* ?
20:06:49*__builtin is stupid
20:06:52__builtinnever mind
20:08:15user890104in hex it's 0x14020000
20:08:38user890104open() is called with O_BINARY, so shouldn't be a cr/lf problem
20:09:15__builtinit shouldn't be an endianness issue
20:10:18__builtincould it be structure padding?
20:10:56 Join johnb2 [0] (
20:11:13gevaertsSounds like it
20:11:18user890104looks like it reads at an offset +2
20:11:37user89010400 00 02 14 is what i get
20:11:53user890104the struct has STRUCT_PACKED
20:12:34user890104it's defined to the gcc attribute only if __GNUC__ is defined
20:12:47__builtinis it?
20:13:46gevaerts maybe?
20:14:30user890104looks like it is
20:14:53igorsk_add explicit padding?
20:14:53gevaertsTry adding -mno-ms-bitfields somewhere
20:15:22gevaertsigorsk_: the idea here is that there should be *no* padding
20:15:31 Join MrZeus1 [0] (~MrZeus@2a02:c7f:7066:fb00:e4b3:dad8:e900:8f8b)
20:15:56__builtingevaerts: what were the race conditions, exactly? something with sysfont.h?
20:16:10gevaertsOh, look at , kugel saw it too :)
20:16:13gevaerts__builtin: yes
20:16:41gevaertsAlways something in puzzles as well, so I suspect the makefile bit in there has something wrong in it
20:17:04user890104gevaerts: -mno-ms-bitfields works fine :)
20:17:58 Quit MrZeus (Remote host closed the connection)
20:19:39user890104it might be a good idea to add it when doing a mingw build
20:19:46gevaertsYes, probably
20:21:06user890104any idea what generates apps/lang.h?
20:21:38user890104compiling fails at apps/action.c
20:23:36gevaertstools/genlang I think
20:26:30__builtinthe puzzles makefile is a monstrosity even compared with doom
20:26:33 Quit PurlingNayuki (Remote host closed the connection)
20:27:06__builtinI really need to simplify it at some point
20:27:47*gevaerts declares it a 4.0 release goal
20:31:20 Join lorenzo92_1 [0] (~lorenzo@
20:38:06 Quit MrZeus1 (Ping timeout: 240 seconds)
20:38:52user890104 - apps/lang.h is not generated for some reason
20:47:26 Join krabador [0] (~krabador@unaffiliated/krabador)
20:52:02 Join webguest449 [0] ( for ARM toolchain also fails on mingw - binutils-2.20.1/binutils/rename.c fails to build
20:56:16 Join webguest42 [0] (
20:56:46lorenzo92_1I am thinking about enabling rbutils to patch samsung ypr0 and ypr1 ROM files but I am (and always been) very puzzled on the cramfs patching. There is no real clean way to extract - patch - compress on Windows. Therefore, I had an idea: providing a binary patch
20:56:58 Quit webguest449 (Quit: CGI:IRC (Ping timeout))
20:57:15__builtindoes anyone ever use the cgi irc service anymore?
20:57:54lorenzo92_1do you think guys that this is feasible (legal issues, etc). Basically, rbtuils would 1) download the original ROM from samsung servers locally 2) use the tools I have developed in plain C to decrypt ROM 3) apply binary patch 4) encrypt ROM
20:58:27 Join webguest756 [0] (
21:01:14 Quit webguest42 (Quit: CGI:IRC (Ping timeout))
21:02:52 Quit yosafbridge (Quit: Leaving)
21:03:22 Quit webguest756 (Quit: CGI:IRC (Ping timeout))
21:05:25 Join ZincAlloy1 [0] (
21:05:33gevaertslorenzo92_1: historically we've always had the user download
21:05:33 Quit ZincAlloy (Read error: Connection reset by peer)
21:06:06lorenzo92_1gevaerts: fine, never used :-) what about binary patching?
21:07:30*lorenzo92_1 I need to actually check what a binary patch could look like...
21:07:38*gevaerts doesn't know the details
21:08:00gevaertsHave a look at H1x0, H3x0, and all sansas
21:08:35lorenzo92_1gevaerts: thanks, will do
21:12:20 Join yosafbridge [0] (
21:16:30 Join Boltermor [0] (
21:20:00***Saving seen data "./dancer.seen"
21:21:07saratogaAMS devices have the user download a firmware, hand it to rbutil, and then it patches and copies it over
21:23:08 Quit lorenzo92 (Ping timeout: 246 seconds)
21:31:27 Quit johnb2 (Quit: Nettalk6 -
21:31:35lorenzo92_1saratoga: i see, so as long as I provide a way for one to recreate the binary patch, providing such a binary diff shouldn't be a problem. I'll now perform some tests on the subject
21:32:28saratogayeah sure, there are no legal problems with a user patching something themselves
21:32:43saratogait'd get weird if we were hosting patched binaries of someone else's software though
21:39:44 Nick Guest53237 is now known as alexbobp (
21:50:34*user890104 is going to run a rockbox build client on the freemyipod server
21:51:07*__builtin doesn't think that'll be of much use at the moment
21:57:27user890104well i was going to run it at some point, then forgot
21:59:28__builtingevaerts: how are the builds going?
21:59:46 Quit krabador (Remote host closed the connection)
22:00:00gevaerts45 done. I'm not sure how much there are in total
22:00:21gevaertsAnd then I have to re-do the failed ones
22:01:00gevaertsAnd then the manuals and source tarball
22:05:43 Quit Boltermor (Quit: Leaving)
22:06:33__builtin50 total, it seems
22:11:20gevaerts48 "done", 23 of those failed
22:12:37*__builtin looks into the sysfont.h issue
22:12:45*gevaerts should not have set -j
22:12:56gevaertsI thought the number of failures would be a lot lower
22:13:13 Join jhMikeS [0] (
22:13:26__builtinit works if you do a single-threaded build, usually
22:13:36gevaertsYes, exactly
22:14:19__builtindo you have the exact error message?
22:20:02__builtinI think I know what's causing it
22:20:44__builtinI had to utilize some crazy hacks to get puzzles to build without touching the upstream sources at all, one of those is not including any rockbox headers in the imported sources
22:21:00__builtinso instead I include a "rbcompat.h" on the command line
22:21:10__builtinand that's not detected by `make dep'
22:24:05__builtinif I add apps/plugin.h as a dependency, it ought to work, no?
22:24:52*gevaerts is looking
22:28:08gevaertsI wonder what happens if you just add rbcompat to the explicit dependencies in the three pattern rules in puzzles.make
22:31:39__builtinwell, it builds
22:32:06gevaertsWhich one?
22:32:37__builtinadding rbcompat.h as a dependency
22:32:46gevaertsI just started a -j3 series with
22:32:48gevaertsWe'll see
22:33:15gevaerts(while at the same time running no-j builds for the ones that failed in the previous run)
22:33:30__builtinthat should hopefully work
22:33:35gevaertsPoor laptop fan!
22:33:48__builtinyou're building on a laptop?
22:34:15*gevaerts nods
22:37:15__builtinpresuming that the build server was working, is there any reason why we couldn't use that to do the builds?
22:37:32gevaertsStill fails
22:38:06prof_wolfffbluebrother: after googling it seems that Sleep() is not the proper function to use when you want 'precise' sleep for a few milliseconds, you should use nanosleep() if available or a custom implementation when it is not available, or some kind of udelay() if you don't mind busy wait
22:38:23prof_wolffffrom MSDN: If dwMilliseconds is less than the resolution of the system clock, the thread may sleep for less than the specified length of time. If dwMilliseconds is greater than one tick but less than two, the wait can be anywhere between one and two ticks, and so on.
22:38:55gevaertsYes, actually. While the build server can build arbitrary revisions (although I don't remember *how*), it doesn't do the VERSION= thing, and it doesn't keep elf and map files
22:39:40prof_wolfffobviously we are not expecting always precise sleep on a non real-time OS, but i was expecting to sleep at least the supplied time, not a resulting 0 ms sleep when 1 ms is required, windows world is fascinating :)
22:40:19gevaertsAlso, building the entire set doesn't take *that* long. My CPU was also doing other things
22:40:28prof_wolfffwhy Sleep() was working for me and not for other people? ATM this is a mystery for me, let's see if your build works
22:41:37gevaertsccache *really* helps for this second run to catch up. A minute per build now
22:42:34__builtinah, nice
22:46:40gevaertsOK, 49 builds done (archos recorder fails of course)
22:46:46gevaertsI'm now looking at the manual
22:47:42__builtinhow are you going to transfer them to the download server?
22:48:10gevaertsUpload them to my webspace, tell you where that is, and let you ask zagor :)
22:48:19gevaertsYou're the release manager after all!
22:48:55*__builtin puts on his official hat
22:49:33gevaertsOK, so the machine running my build client manages to build the manual
22:54:37gevaertsOh, right, voices too
22:59:21gevaerts__builtin: manuals are also around a target per minute. Voices I don't know yet (I think the thing is intelligent and uses a cache so it only has to do each string once), but I guess I'll have all files in half an hour to an hour
23:01:19*gevaerts thinks he'll upload a tar file with everything in it so zagor can unpack things easily
23:02:22*__builtin tallies up the places to announce it
23:02:57__builtinthe front page, forums, mailing list... what else?
23:05:11gevaertsI think that's it
23:05:53*__builtin drafts a forum announcement
23:06:35gevaertsOK, so the tarball (well, 7z) script wants there to be a -final tag. Are you ok with me making and pushing that, meaning that *any* error after now means 3.14.1?
23:06:51gevaertsI think I'll wait for the manuals before I do that
23:07:52__builtinwell, we can always delete the tag ;)
23:08:06__builtinand it doesn't have to be pushed to the remote repo
23:08:46gevaertsVoices done, manuals still pending
23:09:28 Join smoke_fumus [0] (~smoke_fum@
23:12:03*__builtin remembers the missing Sony and Creative manuals
23:12:49gevaertsWe can of course always add those later
23:13:37__builtinwasn't the plan to just change the "stable" criteria temporarily?
23:14:02gevaertsYes, but even then I
23:14:13gevaerts'd still consider adding those later
23:20:02***Saving seen data "./dancer.seen"
23:21:49__builtingevaerts: regarding tagging, I pushed a slightly different commit than yours to fix puzzles
23:22:08gevaertsIs it just a build fix?
23:22:53__builtinso when you're tagging, just pull and tag my commit instead of yours
23:23:08gevaertsMy tree is without any makefile fix right now, and that's what I built with, so I'd prefer to tag that
23:23:31gevaertsIt probably doesn't matter much
23:23:36__builtinyeah, you're right
23:24:40gevaertsWe'll have the fix in master and for possible 3.14.1, which is fine really
23:24:54gevaertsAnd the workaround is easy, "try again or remove -j" :)
23:25:09gevaertsSo I basically just missed one place to add the dependency?
23:25:28__builtinI fixed a typo along with the dependency fix ;)
23:25:42__builtinpuzzles/unfinshed should be puzzles/src/unfinished
23:25:58*lorenzo92_1 bsdiff looks promising, I have also found a nice library version. Raw cramfs file is about 14mb, raw patch files is also about 14 mb but mostly zeros. Therefore, I plan to gzip the data as well (qt should have the requisite classes)
23:26:42gevaertsHmmm, no
23:27:00gevaerts__builtin: I don't see any real difference compared to what I did, and that one still failed?
23:27:06gevaertsI'll try your fix too
23:27:37__builtinwhat failed?
23:28:00*lorenzo92_1 gzipped patch is ~100KiB, which is ideal to generate a "nice" .c patch file for every target that needs this kind of operation (at least, ypr0 and ypr1)
23:28:33gevaertsSame thing. Let's see
23:32:02__builtinhmm, I know
23:32:37__builtinit doesn't generate dependencies for rbcompat.h
23:33:36__builtinI'll just try making it depend on $(BUILDDIR)/sysfont.h
23:34:01 Quit TheLemonMan (Quit: "It's now safe to turn off your computer.")
23:34:24gevaertsMaybe just add a dependency line with rbcompat.h's includes?
23:34:45 Quit jhMikeS (Ping timeout: 255 seconds)
23:34:49__builtinno, I don't think it autogenerates dependencies for .h files at all
23:35:28gevaertsSo add them explicitly!
23:35:42 Join jhMikeS [0] (
23:36:36__builtinI think just sysfont.h is enough
23:37:08gevaerts"I think just * is enough" is what caused all previous makefile race conditions :)
23:39:00__builtinadding a $(PUZZLES_SRCDIR)/rbcompat.h: $(APPSDIR)/plugin.h $(BUILDDIR)/sysfont.h $(PUZZLES_SRCDIR)/rbassert.h ought to fix everything ;)
23:40:04gevaertsWhy not actally the *direct* includes?
23:40:49__builtinwhat do you mean?
23:41:18gevaertsplugin.h, rbassert.h, lib/pluginlib_exit.h and tlsf.h
23:41:42gevaertsOf course, for unexplained reasons, your current code in git has now built 27 targets without issue
23:41:48gevaertsSo we might be fine already
23:42:11*gevaerts doesn't get how that differs from what he tried, but he might be too sleepy
23:42:27__builtinpure luck?
23:42:46gevaertsI had around 50% failure before
23:43:10gevaertsI was doing other things as well on the CPU, and things are faster with ccache, but still
23:43:27gevaertsWith my fix, conditions were the same as now and it failed within five builds
23:43:45gevaertsSo my assumption is that I did something wrong
23:44:05__builtinwell, I don't see anything blatantly wrong with your patch
23:44:48gevaertsNo, but there might have been something wrong with how I tested it
23:46:35gevaertsAh, "good" news :)
23:47:03gevaertsIt just failed again with your in-git fix :)
23:47:29__builtinpure luck it is
23:51:09__builtinI just managed to revert bumping the version, lol
23:57:39gevaerts__builtin: it's probably a good idea to do a few sanity checks (like running it on at least one player) before officially releasing. I didn't do *any* of that :)
23:58:17gevaertsAnyway, tarball with all the builds, voices, manuals, and source 7z uploaded, and tag pushed, so with any luck I'm done

Previous day | Next day