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 2017-01-15

00:03:17pamaurymassa: you can include it in your patch for cross compile
00:03:54 Quit johnb3 (Ping timeout: 240 seconds)
00:04:32massayes I'll do - after that I see some more warnings; e.g. that UNALIGNED is already defined - is using UNALIGNED as name for a macro a good idea?
00:09:11pamauryit's not a standard macro name, I don't see why it should be predefined somewhere
00:09:28pamaurywhere is it already defined ?
00:11:30massait's in /usr/i686-w64-mingw32/sys-root/mingw/include/_mingw.h
00:12:13massait's defined either as __unaligned or as "nothing"
00:12:40massa(the first in win64 environments, the latter in win32)...
00:16:01 Quit robertd1 (Ping timeout: 245 seconds)
00:16:06pamauryah, that's very unfortunate
00:19:25 Quit lebellium (Quit: ChatZilla 0.9.93 [Firefox 50.1.0/20161208153507])
00:21:46 Join robertd1 [0] (
00:24:59pamaurymassa: the way I see it, you could add code to #undef UNALIGNED if it is defined in the offending files
00:25:05pamaurywith a comment saying why
00:25:27 Quit n3m9 (Read error: Connection reset by peer)
00:25:37massait seems it's only one place affected - in firmware/asm/memset16.c
00:25:54massaand there I just rename UNALIGNED in RB_UNALIGNED ;-)
00:30:07massa(at least for my target device - when I search for UNALIGNED, there are a lot of places which define it...)
00:30:49 Quit cc___ (Ping timeout: 240 seconds)
00:30:56[Saint]grepping for it across the entire tree shows it is used rather extensively.
00:31:38massayeah, that's what I wanted to say with "a lot of places" :-)
00:32:47[Saint]You're doing the work that no one really bothered with at the moment.
00:32:47[Saint]Hard to argue that it isn't important, but it evidently wasn't anyone's pressing priority.
00:33:14[Saint]I never pay any attention to defined but unused or already defined warnings.
00:33:33[Saint]I usually go out of my way to get the compiler to just shut up about it.
00:33:33*pamaury thinks that defining UNALIGNED in mingw.h is big mistake
00:34:06[Saint]How so?
00:34:08pamaurybut that might be a Windows thing because windows.h defining a awful lot of things
00:34:42pamaurybecause it's a name you migth be tempted to use, that has no obvious meaning in C nor any standard construct
00:34:59[Saint]Oh. Right.
00:35:02 Quit robertd1 (Ping timeout: 260 seconds)
00:35:29pamaurythere is another one like it in POSIX that I run into from time to time, can't remember it
00:35:37 Join robertd1 [0] (
00:36:01massa[Saint]: you're right - actually it's totally unimportant to know if a define is already defined and gets redefined
00:36:10pamauryfor example, say you create an enum for errors:
00:36:16pamauryenum err_t { ..., UNALIGNED, ... };
00:36:19pamauryboom, compiled error
00:36:41pamaurythat's the level of badness I envision
00:36:52massapamaury: good point
00:37:53[Saint]In other projects I'll generally do what massa is doing now and use a project-specific prefix.
00:38:01[Saint]like PROJECT_UNALIGNED
00:38:03pamauryah yeah, the POSIX one is OVERFLOW
00:38:15pamaurywhich is #define OVERFLOW 1 or something like it
00:38:22[Saint]pamaury: ah, right - yeah I've bounced off that one too I believe.
00:38:48massa[Saint]: that's what I do, too - or if in C++, use namespaces :-)
00:39:24[Saint]I like mine to be verbose bordering on lunacy, on the other end of the scale some people get annoyed with that.
00:40:15[Saint]as an example instead of RB_UNALIGNED I'd likely do ROCKBOX_UNALIGNED to remove any doubt that RB wasn't come acrnym that refers to another esoteric function elsewhere.
00:40:31[Saint]bah - typing
00:40:36[Saint]*some acronym
00:41:00***Saving seen data "./dancer.seen"
00:41:40[Saint]Though in this context I sincerely doubt anyone would find RB to be an ambiguous prefix.
00:43:15[Saint]massa: I wouldn't say it's /totally/ unimportant, for what it's worth.
00:43:19massawhat if I add "-UUNALIGNED" to the GCCOPTS - would that help?
00:43:29[Saint]99.99% of the time it is absolutely intentional.
00:43:35[Saint]but that .01%...
00:44:10pamaurymassa: I would go for rename to RB_UNALIGNED
00:44:49[Saint]for GCCOPTS my usual method is "throw shit at it from stackoverflow top comments until it compiles".
00:44:55[Saint]It's a very scientific method.
00:46:06massadon't bother - that -UUNALIGNED would not help as it's defined in an include...
00:46:21*chrisjj finds that inserting the ZEN blanking plate in the SD slot causes RB to *PANIC*.
00:46:35pamaurychrisjj: that's not new
00:46:41*pamaury leaves
00:46:57massapamaury: you mean - search and replace every occurance of UNALIGNED to RB_UNALIGNED (or ROCKBOX_UNALIGNED)?
00:47:09 Quit robertd1 (Ping timeout: 240 seconds)
00:47:43chrisjjNew to me and RB Flyspray, I think.
00:48:43 Join robertd1 [0] (
00:51:04 Quit pamaury (Ping timeout: 240 seconds)
00:53:43*massa leaves too
00:54:04*massa says: good night - for those where it's night :-)
00:54:14 Quit massa ()
00:56:26[Saint]soooooo many 'bugs' on flyspray that are PEBCAK, honest mistake, or hardware error.
00:56:32[Saint]God I regret looking...
00:58:26 Part robertd1
00:59:41[Saint]need to go on a shitkicking spree and close out a bunch and ask a bunch of users if they can still reproduce.
01:16:53 Join jhMikeS [0] (
01:18:33 Quit girafe (Read error: Connection reset by peer)
01:25:49 Quit shmibs (Quit: leaving =o)
01:26:25 Join shmibs [0] (
01:38:01 Join CrashBash-Kun [0] (~CrashBash@unaffiliated/crashbash-kun)
02:08:43 Quit ender` (Quit: Hardware, n The part of the computer that you can kick.)
02:31:10 Quit ZincAlloy (Quit: Leaving.)
02:41:03***Saving seen data "./dancer.seen"
03:38:33 Join Senji_ [0] (~Senji@
03:40:59 Quit Senji (Ping timeout: 240 seconds)
04:14:36 Join Senji [0] (~Senji@
04:17:05 Quit Senji_ (Ping timeout: 240 seconds)
04:38:18 Quit Senji (Ping timeout: 260 seconds)
04:41:06***Saving seen data "./dancer.seen"
05:47:08__builtinwow, people still use flyspray?
05:47:23__builtinI don't even remember how to get there
05:54:22 Join idonob_ [0] (
06:25:05 Quit [7] (Ping timeout: 256 seconds)
06:25:22 Join TheSeven [0] (~quassel@rockbox/developer/TheSeven)
06:38:44 Quit CrashBash-Kun (Read error: Connection reset by peer)
06:40:51 Join JanC_ [0] (~janc@lugwv/member/JanC)
06:40:51 Nick JanC is now known as Guest24786 (~janc@lugwv/member/JanC)
06:40:51 Quit Guest24786 (Killed ( (Nickname regained by services)))
06:40:51 Nick JanC_ is now known as JanC (~janc@lugwv/member/JanC)
06:41:10***Saving seen data "./dancer.seen"
07:35:44 Join CrashBash-Kun [0] (~CrashBash@unaffiliated/crashbash-kun)
07:36:08 Quit CrashBash-Kun (Client Quit)
08:41:12***Saving seen data "./dancer.seen"
08:49:25 Join CrashBash-Kun [0] (~CrashBash@unaffiliated/crashbash-kun)
08:49:49 Quit CrashBash-Kun (Client Quit)
08:54:28 Join johnb3 [0] (
08:59:13 Quit johnb3 (Ping timeout: 256 seconds)
09:33:05 Join paulk-collins [0] (
09:36:04 Join ender` [0] (
09:43:33 Join johnb3 [0] (
10:09:48 Quit johnb3 (Quit: Nettalk6 -
10:19:44 Quit krnlyng (Ping timeout: 252 seconds)
10:20:30 Join xorly [0] (
10:33:02 Join krnlyng [0] (
10:33:06 Join ZincAlloy [0] (~Adium@2a02:8108:8b80:1700:19e:a62c:a52f:7163)
10:38:32 Join Senji [0] (~Senji@
10:41:13***Saving seen data "./dancer.seen"
10:45:37 Quit xorly (Ping timeout: 245 seconds)
11:03:16 Join cc___ [0] (~ac@2001:910:113f:1:6a05:caff:fe1c:1627)
11:05:36 Join johnb3 [0] (
11:07:08johnb3Mihail: can you please provide the source / patch for (the sd_only part), so I can do this on my own in the future?
11:08:15johnb3With this build however I have zero reception on FM! That's a variant 1.
11:10:17 Join lebellium [0] (
11:15:25lebellium__builtin: people just click on the Development>bugs from the home page.
11:18:10 Join Mihail [0] (25d48052@gateway/web/cgi-irc/
11:18:55johnb3No FM reception also with scaling-9, so probably with all of them. It does work well with the current dev builds.
11:19:52johnb3With rockbox-clip+-cvdd2_scaling-9 on my clip+ variant 0, FM works as normal!
11:20:55Mihailwe try fix fm later
11:20:58 Join smoke_fumus [0] (
11:21:46johnb3thanks a lot!
11:24:42johnb3How trivial magic can be ;-)
11:25:27Mihailjust hack :)
11:28:42 Quit cc___ (Ping timeout: 258 seconds)
11:29:26 Quit Senji (Quit: KVIrc 4.9.2 Aria
11:29:50 Join Senji [0] (~Senji@
11:43:05 Join pamaury [0] (~pamaury@rockbox/developer/pamaury)
11:45:04 Quit johnb3 (Ping timeout: 255 seconds)
11:47:07lebelliumpamaury: when "do_fw_upgrade", there is an empty "Result:" line. Is it supposed to display something when upgrade completed or is it a generic line for every scsitool command?
11:55:08pamauryit's normal iirc
11:58:27pamauryBut I can look/fix it if it's confusing
11:59:16lebelliumthat's a minor thing
11:59:40lebelliumBetter look on the sound issues :P
12:03:45pamauryI think it's because the old code used to display the error code unconditionally (OK being success) but the new one only does on error
12:13:54 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
12:23:54 Quit bertrik (Quit: Lost terminal)
12:26:24 Join bertrik [0] (
12:26:24 Quit bertrik (Changing host)
12:26:24 Join bertrik [0] (~bertrik@rockbox/developer/bertrik)
12:41:17***Saving seen data "./dancer.seen"
12:53:10 Join MrZeus_ [0] (~MrZeus@
12:54:57 Quit MrZeus (Ping timeout: 256 seconds)
13:01:19 Quit Mihail (Quit: - A hand crafted IRC client)
13:03:34 Quit idonob (Ping timeout: 240 seconds)
13:07:53 Join skapazzo [0] (~skapazzo@
13:11:04 Join idonob [0] (
13:14:44chrisjjpamaury, any particular reason why your battery benchmark procedure charging specified the power sources as PC rather than charger?
13:22:33chrisjjThanks. Any particular reason why it specifies charging from PC is to be by 'holding a key (so that usb mode is not triggered)'?
13:23:10 Quit idonob (Ping timeout: 240 seconds)
13:23:45pamauryyes, you don't want usb mode because otherwise you can't enable battery_bench
13:24:14pamaury(since supposedly, when you finish the discharge part, you're device cannot be powered from battery)
13:24:40pamauryalso battery_bench requires disk access
13:24:48pamaurythus it won't work with usb mode
13:24:58pamaury(I think)
13:25:25 Join idonob [0] (
13:28:49 Join johnb3 [0] (
13:32:35pamaurys/usb mode/usb disk mode/
13:33:24chrisjjFigures. Perhaps (despite that I am indeed avoiding USB mode) 'requires disc access' is the clue to why I'm not getting more than one battery-bench data point from this. Here's the end of file: . It could be me bungling it somehow, or again some unknown required variable.
13:33:26 Join JanC_ [0] (~janc@lugwv/member/JanC)
13:34:11chrisjjFor this memDFS benchmarking I've skipped this charge phase, and will investigate another time.
13:34:37 Quit JanC (Killed ( (Nickname regained by services)))
13:34:37 Nick JanC_ is now known as JanC (~janc@lugwv/member/JanC)
13:36:35chrisjj'usb disc mode'? Is this to imply that this mode does something else with USB data? I'd hoped it was doing nothing with USB data i.e. was using the USB for only power.
13:37:30 Quit johnb3 (Ping timeout: 240 seconds)
13:41:41 Quit smoke_fumus (Quit: KVIrc 4.2.0 Equilibrium
13:43:42 Join massa [0] (4fe8d32c@gateway/web/freenode/ip.
13:44:15massaHi everybody!
13:46:21massapamaury: I finished the configure changes and also managed to upload the changes for review to Gerrit; as I'm new to this - how do I trigger a review now? Which URL shall I present for reviews?
13:48:17pamaurymassa: for reviews I think you add people to the list of reviewers
13:48:25pamaurybut it's usually better to just come on IRC and ask people
13:48:37pamauryfor the link, you can write g#n where n the number of the task
13:48:49pamaury g#1497
13:48:51fs-bluebotGerrit review #1497 at : Renamed defines UNALIGNED to ROCKBOX_UNALIGNED - UNALIGNED is already by Matthias Mohr
13:48:58pamaury g#1496
13:49:00fs-bluebotGerrit review #1496 at : Fix configure for windows simulator builds in cygwin / mingw cross-compiling environm... by Matthias Mohr
13:50:10pamaurychrisjj: usb disk mode as in you get a disk over usb and thus the player cannot be operated. This is contrary to when you hold a key on usb plug and you don'get usb disk mode
13:52:06massapamaury: you mean the change numbers? it's 1496 and 1497 - reachable as
13:53:10chrisjjOK, got it.
13:53:28pamaurymassa: I'll try to review your patches today
13:54:34massapamaury: thanks - the patch for the puzzle function conflict (vsscanf) which is needed for compilation is not included - I thought it's __builtin's part ;-)
13:56:39chrisjjMy ZEN doing the memDFS benchmark has just done a Black Screen of Power Off, as before. (<< True not false alarm.) LED (that was on) is off; audio (that was sounding) is silent. Any other observations you'd like me to do?
14:04:47chrisjj(This a clean boot of a clean install to which has been added only audio files + battery_bench.txt. Unit Q.)
14:05:06massapamaury: This brings me to problem with the compiled 64-bit simulator. In the first place it works - but if you try e.g. to start playing some music it does not load the codec files - a SDL error appears, saying that the DLL initialising method failed...
14:05:41massaSo, how does the codecs get loaded in simulator environment?
14:06:12pamauryon linux it is compiled to .so (dynamic library), on windows it should be a dll
14:07:49 Quit paulk-collins (Remote host closed the connection)
14:10:36massapamaury: the file format must be O.K. - otherwise it would say "wrong file format" or something similary; it seems it tries to load it but the initialization fails...
14:10:38 Join n3m9 [0] (
14:11:36massaI enabled DEBUG and LOGF in my build - but I don't see a lot messages. Doesn't this work in simulator buidls?
14:13:18pamaurylogf needs to be enabled per file (by defining #define LOGF_ENABLED before #include "logf.h"). I think DEBUG should print the console
14:13:27pamauryHowever I wouldn't compile with DEBUG if I were you
14:14:02pamaurybecause last week I noticed that codecs seem broken with DEBUG, I couldn't identity why (it was in a real target, not simulator)
14:14:21pamauryI'd say, either use a debugger or printf
14:14:33pamauryI'll try it in my VM
14:18:41chrisjjmassa, If you'd like a test of your build of the Windows sims for ZEN and iPod Video, do post them.
14:18:43massapamaury: did you install the 64-bit mingw compiler chain? Lately you only had the 32-bit versions
14:19:00 Join petur [0] (~petur@rockbox/developer/petur)
14:20:14pamaurymassa: I installed both, but right now I'm busy, I'll debug that later if I can
14:24:23pamauryI don't really know what is done on DLL init, I am not sure we register any action on it, so it's mostly libc job I think
14:25:11massachrisjj: you mean to upload the binary packages to some cloud?
14:28:02pamaurymassa: you want to load into firmware/target/hosted/sdl/load_code-sdl.c
14:28:20pamaurythat's the code used to load a codec/plugin/code
14:28:41pamaury(on SDL)
14:28:47massapamaury: I'm already there - but there's only the call to SDL_LoadObject and that gives back a NULL handle -...
14:28:59massahave to find out why...
14:29:11pamauryah, that's going to be tough to debug
14:29:17pamauryI think you need a debugger
14:29:40chrisjjpamaury, this BSoPO'd ZEN showed no response to any key except power-on. Power-on launched RB as normal. So I assume the cause was a reset by angry watchdog, since no other potential cause is known.
14:31:24chrisjjZEN tester's Pro Tip: Remember that BSoPO loses the session's settings changes e.g. Repeat All for battery benchmarking.
14:31:54pamaurymassa: it works on mine, but I did a 32-bit build, let me check with a 64-bit one
14:33:15massapamaury: that's what I mean the 32-bit version is working, the 32-bit one is _new_, so maybe there is some loading or generating code which does not work for 64-bit...
14:33:31massaI meant: the 64-bit one is _new_
14:35:25 Join paulk-collins [0] (
14:36:25pamaurycompiling a 64-bit build, this will take a while
14:39:12pamauryone possibility (I'm not sure), is the "long" confusion on linux/windows
14:39:44massapamaury: the 32-bit version also works with DEBUG enabled - so that cannot be the reason...
14:39:45pamaurycause on Windows, long is still 32-bit in 64-bit mode
14:40:21pamaurymassa: don't be so sure though, I've got target that work in DEBUG and some that don't
14:40:22massapamaury: well, there were some pointer cast warnings in 64-bit
14:40:39pamaurybetter not silence them
14:41:20***Saving seen data "./dancer.seen"
14:41:32 Quit tchan (Quit: WeeChat 1.6)
14:42:53pamaurythere might be some fixing needed
14:46:25pamauryhum there are some suspicious ones
14:48:25*pamaury looks into it
14:51:16 Join johnb3 [0] (
14:53:50 Join tchan [0] (
14:53:51 Quit tchan (Changing host)
14:53:51 Join tchan [0] (~tchan@lunar-linux/developer/tchan)
14:55:42pamauryarg, why people use cast to long from pointers???
14:55:45pamaurythere intptr_t for that
15:02:39*pamaury fixes things
15:05:18pamauryso far only one of them is really dangerous, the one in filesystem-win32.c
15:05:49chrisjjBSoPO again. And this time the first few restarted BSoPOed after the BL text and before the RB logo. I've seen this before. Goodness only knows where the causal state is residing.
15:08:07chrisjjPerhaps battery_bench is itself the cause.
15:08:17chrisjjPerhaps a battery_bench.txt that is failing to take further entries, as reported above for the charging phase. FWIW, I've deleted battery_bench.txt before starting the plugin this time.
15:08:55chrisjjIf that too BSoPOs, I'll retry with the plugin not running.
15:09:20chrisjjs/the first few restarted BSoPOed after the BL text/the first few restartes BSoPOed after the BL text/
15:10:11*pamaury touches the skin_parser codes and tries not to break it
15:19:04 Quit bertrik (Remote host closed the connection)
15:22:37pamaurymassa: any progress ? I am fixing a bunch of warnings and although some are dangerous, they would simply crash the simulator if they were triggered
15:29:23pamauryarg, seriously libtremor, casting a pointer to unsigned long, I expected better
15:34:14massapamaury: no, I did some real world (tm) activity :-)
15:34:47 Join MT_ [0] (~MT@2601:482:4402:7b60:a8e4:1c64:d2d1:4ba6)
15:35:09 Quit munch (Quit: Let's blast this shit and get naked)
15:35:56 Join munch [0] (pls@gateway/shell/elitebnc/x-hjjqdjfndwontqqf)
15:36:25 Nick MT_ is now known as _mt_ (~MT@2601:482:4402:7b60:a8e4:1c64:d2d1:4ba6)
15:37:17 Quit johnb3 (Ping timeout: 245 seconds)
15:38:18pamauryhaha ok :-p
15:38:26pamauryI'm almost done
15:38:49pamauryI've added a comment to your gerrit task
15:38:58massachrisjj: which ZEN simulator do you want? Just "Zen"?
15:40:34chrisjjYes please.
15:41:00 Join johnb3 [0] (
15:41:30massapamaury: how can I update an existing patch at gerrit? Somehow with "git −−ammend", but how do I tell it which change# to use?
15:41:55massaI meant "git commit −−ammend"
15:42:24*chrisjj could seriously believe −−ammend in git
15:43:47pamaurymassa: you have two options:
15:43:47pamaury- use git −−amend and don't touch the Change-Id: .... line at the bottom of the message, then upload again
15:43:47pamaury- create a completely different commit but copy-paste the old ChangeId: ... line
15:44:10pamaurythis line:
15:44:10pamauryChange-Id: Ib5dd288c3f9a11be9b66e53e3406eadfa65f7604
15:44:21pamaurythat's the only thing that gerrit uses to track tasks
15:46:09*pamaury spots unsigned int wasted = (-(long) buffer) & 3; −−-> VERY suspicious
15:46:16chrisjjZEN Unit Q memDFS: Again BSoPO. After 3min. This time with a short battery_bench.txt, exonerating long battery_bench.txt.
15:47:37 Quit johnb3 (Ping timeout: 255 seconds)
15:50:20massapamaury: and how do I tell −−amend which change-Id / earlier commit I mean? E.g. if I do different changes on the same file? Or add changes to another file and want it to get to the same change#?
15:50:20pamauryis it me or:
15:50:20pamaury#define PTRDIFF(pe, ps) ((int) (((long) (pe) - (long) (ps)) / sizeof(*pe)))
15:50:20pamauryis exactly equivalent to:
15:50:20DBUGEnqueued KICK pamaury
15:50:20pamaurype - ps
15:51:17pamaurymassa: if you want to change just the last commit, use git −−amend. If you want to change an ealier commit (like the one before the last), it's more complicated, you need interactive rebase
15:53:22 Join cc___ [0] (~ac@2001:910:113f:1:6a05:caff:fe1c:1627)
15:53:33pamauryI suggest you document yourself on the internet, but basically it goes like this:
15:53:34pamaurygit rebase -i <commit before the one you want to edit> # for example git rebase -i HEAD^^ (the ^ means "one before")
15:53:34pamaurythen you get a text editor: change "pick" to "edit" for the commit(s) you want to change
15:53:34pamaurynow git does stuff and you get into a state where the first commit you want to "edit" is HEAD: do you changes, git commit −−amend
15:53:34***Alert Mode level 1
15:53:34pamaurythen continue by doing git rebase −−continue
15:53:45pamauryit's very pwoerful but a bit intimidating the first times
15:58:56pamauryas expected things like z80 and doom are horrible offenders
16:03:35***Alert Mode OFF
16:04:21massapamaury: I read some rebase documentation - but I'm lost :-(
16:05:27massaI did an interactive rebase as suggested, marked the corresponding commit as edit, put my changes in, commited it again and then tried to "rebase −−continue"
16:06:05 Quit tchan (Quit: WeeChat 1.6)
16:06:05massabut then it complains about "You have uncommitted changes in your working tree. Please, commit them first and then run 'git rebase −−continue' again."
16:07:03massasorry, forget the last message - here's the correct one: "tools/configure: needs update You must edit all merge conflicts and then mark them as resolved using git add"
16:07:11pamaurymassa: you cannot rebase −−continue if you have uncommitted chanes
16:07:22pamauryie if git diff shows something
16:07:50pamauryah you might have a conflict
16:07:56pamaurytype git status
16:08:03pamauryit tells you what is the problem
16:10:33pamauryif you have a conflict, solve it (ie look at the file, correctr what you want). Then explicitely git add <file> and then commit
16:11:29 Join tchan [0] (~tchan@lunar-linux/developer/tchan)
16:11:49 Quit Moarc (Quit: i znowu NADMUCHAƁ BALONA)
16:12:16massapamaury: but it did commit already...
16:13:11massaI now did a git add tools/configure - and then it wanted that I commit that again - but without a change id; is that correct?
16:14:27chrisjjAgain, ZEN memDFS BSoPO.
16:14:35pamauryif it's a conflict because of rebase −−continue, I think it's correct. In the worse case, we can fix it later
16:14:54pamaurychrisjj: I think we understood that there is a problem, no need to spam the channel
16:15:28chrisjjPerhaps someone can explain this 'good practice' of code that auto-resets a stuck device, preventing me seeing the info on the screen that might show what went wrong.
16:16:18pamaurychrisjj: for the FUCKING MILLIONTH time, there is NO INFO TO SEE, all right ?
16:16:44 Join Moarc [0] (
16:16:59chrisjjIf you can't see the info, how do you know there's none to see?
16:17:05pamauryyou'll get a frozen screen, great
16:17:18pamaurywhat kind of info do you get from that ?
16:17:23chrisjjA frozen screen will tell me what the device was doing at the freeze.
16:17:44chrisjjWhat's the advantage in hiding it.
16:17:45pamaurysure, as in "it was playing sound" ?
16:18:11massapamaury: Hmm, I did commit it and then the rebase −−continue works - but trying to push it to gerrit leads to a "! [remote rejected] HEAD -> refs/for/master (no changes made)"
16:18:18chrisjjI don't know it was playing sound. I think you don't know either.
16:18:41pamauryI told you the advantage already: if the reset button is hard to reach, you device auto resets so you don't end up with a brick until you can get access to something small enough to reset the device
16:19:08pamaurymassa: it means you did not change anything compared to previously
16:19:08chrisjjIf it was playing sound, the screen will show me which file, where in the file, and (apropos fails in my previous batch of tests) the state of the backlight.
16:19:19pamaurymassa: if you type "git diff HEAD", does it show anything ?
16:19:58chrisjjCan't see much the advantage of making it easy to run the device to freeze again.
16:20:11chrisjjAnyway, the reset button is easy to reach in debugging.
16:20:36chrisjjThe information hidden by the reset is currently impossible to reach in debugging.
16:20:41pamaurychrisjj: the problem is that you consider the situation only from your point of view. The watchdog is not a debug mecanism
16:20:49 Quit _mt_ (Read error: Connection reset by peer)
16:21:04pamauryif you want to disable it from debug, go get a VM and disable it
16:21:08chrisjjI don't consider the watchdog to be a debug mechanism.
16:21:20massapamaury: no it does not show anything - but the file content of tools/configure is the changed one (I removed the -Wno flags for win64)
16:21:20chrisjjIt is a production mechanism which is a PITA during debugging.
16:22:12chrisjjHow can I disable it with a VM? You've not released the sources that would allow me to build it.
16:22:33pamaurychrisjj: I'll release them if you do builds
16:22:47pamaurysince at the moment you don't do any build, I'm not wasting any time
16:23:00pamauryand by the way this is not always true, all your recent tests are on gerrit
16:23:38pamaurymassa: this is strange, you can use "git log" and "git log -p" to review the changes recorded in each commit
16:23:40chrisjjHow about releasing them so that anyone can do builds? As we'd expect in this collaborative project.
16:24:43pamaury g#1483, g#1485, g#1489, g#1490
16:24:49fs-bluebotGerrit review #1483 at : imx233: add small framework for LED by Amaury Pouly
16:24:49fs-bluebotGerrit review #1485 at : zen/zenxfi: avoid white flash on backlight on by Amaury Pouly
16:24:50fs-bluebotGerrit review #1489 at : zen/zenxfi: make backlight sequence closer to that of the OF by Amaury Pouly
16:24:51DBUGEnqueued KICK fs-bluebot
16:24:51fs-bluebotGerrit review #1490 at : zen/zenxfi: hack to make white screen go away by Amaury Pouly
16:24:51chrisjjWhere's lcd_fix?
16:25:04pamaury g#1428
16:25:06fs-bluebotGerrit review #1428 at : imx233: workaround hardware dma bug by Amaury Pouly
16:26:18 Join _mt_ [0] (~MT@2601:482:4402:7b60:a8e4:1c64:d2d1:4ba6)
16:26:19pamauryarg no, why does cygwin changes the permissions of all files :'(
16:26:22chrisjjThanks. So why are these not in git?
16:27:50pamaurybecause I don't if they work obviously
16:29:14chrisjjWho says g#1428 doesn't work? Not me for sure.
16:29:16fs-bluebotGerrit review #1428 at : imx233: workaround hardware dma bug by Amaury Pouly
16:32:26pamauryWe this attitude, I would commit a lot of things that just work for me and I wouldn't try to fix the ZEN. I haven't tested it on all imx233 targets yet
16:32:32massapamaury: actually cygwin does not change it - it does only map the windows permissions to some linux based - and that does often not match very good (especially the x flag)
16:33:37pamaurymassa: I have uploaded g#1499 that tries to fix all casts
16:33:39fs-bluebotGerrit review #1499 at : Fix dangerous casts by Amaury Pouly
16:33:59 Quit paulk-collins (Ping timeout: 240 seconds)
16:34:05pamaurydid you solve your git problem ?
16:38:08pamaurymassa: I also get the dll init problem
16:39:26chrisjjpamaury, you've mistaken my attitude. I'm as cautious about testing before release as anyone you'll meet. If you tell me RB uses Gerrit Review for QA system test results (despite that seems to indicate Review is for source inspection and unit test results only), then I'll submit my results.
16:39:43 Quit dfkt (Disconnected by services)
16:40:19massachrisjj: you can download my compiled simulators here: but don't expect anything special, there's no changed code - just compiled in cygwin environment...
16:40:21chrisjjHowever does not offer me a Review button, despite that it lists me as a Reviewer.
16:41:13pamauryi've never seen a reviw button, one uses Reply
16:41:22***Saving seen data "./dancer.seen"
16:41:38massapamaury: no, I didn't solve the git problem - with "git log" I can see the commit message, but not on top...
16:43:06pamaurymassa: can you pastebin the output of git log -p HEAD^^^..HEAD
16:44:31chrisjjAh, silly me for going by the manual in which one uses Review, and no reply Reply exists! :-)
16:46:22massathe shown top commit is the UNALIGNED patch - then comes the new one which I want to push to gerrit...
16:47:48massapamaury: if you solved the pointer casts and still have the loading problem there must be another cause for this
16:47:54pamaurymassa: ok, I guess gerrit is confused and thinks there is no change. What you can do is merge the two commits that touch configure
16:49:17 Join paulk-collins [0] (
16:49:33pamauryassume your log is:
16:49:33pamauryconfigure patch #2
16:49:33pamauryconfigure patch #1
16:49:33pamaurythen you can do:
16:49:35pamaurygit rebase -i HEAD^^^^
16:49:37pamauryit will show a file like this:
16:49:39pamaurypick configure patch #1
16:49:41pamaurypick configure patch #2
16:49:43pamaurypick unaligned
16:49:59pamaurynow replace the second line by:
16:49:59pamaurysquash configure patch #2
16:50:59pamauryit will create a single commit for the two configure patches and let you edit the message. Make sure you keep only the original change id:
16:50:59pamauryChange-Id: Ib5dd288c3f9a11be9b66e53e3406eadfa65f7604
16:50:59pamauryand now you should be done
16:53:17chrisjjmassa, thanks for the sim ZIPs. Did you actually manage to run the content yourself?? ISTM the .exe is missing.
16:53:51chrisjjHere's a compare of the latest working ZEN sim here (from rasher) against your ZIP .
16:54:55pamaurymassa: I'm trying to use gdb to find the problem
16:55:33pamaurybut it's very weird
16:57:12massachrisjj: your're right - it seems "make fullzip" does not put the simulator in the zip... - I redo with "make zip" and reupload it...
16:59:02massapamaury: I'm neary sure it has something to do with the generation of the codecs - there must be some DLLMain or similar for the simulators - maybe that's missing for 64-bit builds
16:59:34pamaurywhat's really weird is this: this is the backtrace
17:00:36pamauryit looks like if _FindPESectionByName (a mingw thing) calls strlen() that calls rockbox strlen that calls ci->strlen()
17:00:42pamaurybut of course ci is not initialsied that early
17:00:47pamauryso might be a symbol problem
17:01:26CtcpIgnored 2 channel CTCP requests in 34 minutes and 31 seconds at the last flood
17:01:26*pamaury is installing tons of debug symbols
17:03:13 Quit b-t` (Read error: Connection reset by peer)
17:04:32massapamaury: there seems to be something wrong with the sections in the codec - I think there's still some alignment problem; when I solved my git problem I'll also have a look at it...
17:04:46pamauryand in codeclib.h there is this comment:
17:04:47pamaury * so confuses the cygwin linker, and maybe others. These functions need to
17:04:47pamaury * be implemented elsewhere */
17:05:16pamauryI'm going to rename more function like it's done for malloc
17:06:58chrisjjmassa, that'll be some strange meaning of the word 'full' with which we are not familiar... :-)
17:08:20massachrisjj: neither "make zip" nor "make fullzip" does put the simulator files into the zip - I'll have to either do it by hand or investigate in it - but later, I'm currently busy with other things...
17:09:08pamaurymassa: chrisjj: it makes sense, make zip takes .rockbox and zip it basically. And make fullzip does the same but adds font. Simulator executale is not in .rockbox/ and it's probably an oversight
17:10:21pamauryarg, developping on windows would be easier if windows update didn't take 100% of the CPU and disk
17:11:17chrisjjIf the automatic ZIPping doesn't use a custom script, perhaps there's a standard one you've missed. 'fullerzip' perhaps :-)
17:18:02massapamaury: now it uploads something to gerrit - but I can't see any difference? Maybe I should start allover and do a fresh clone?
17:21:17 Quit _mt_ (Ping timeout: 256 seconds)
17:22:04pamaurymassa: what si the state of configure in your editor ? Do you still have the warnings removed ?
17:22:15pamaury(and make sure to reload to file to check what is really on disk)
17:22:24massapamaury: no, they're gone...
17:23:10pamauryok, let's try again, slowly this time:
17:23:10pamaurygit rebase -i HEAD^^
17:23:10pamauryit should show two commits: unaligned and configure. Change the configure one to edit
17:23:43massapamaury: no, it now shows other commits (puzzle, dircache and others)...
17:24:12pamaurymassa: you mean the last two commits are not yours ?
17:25:27pamauryok, let's restart from a clean state (you'll loose any change, except what is saved on gerrit already):
17:25:28pamaurygit checkout master
17:25:28pamaurygit reset −−hard origin/HEAD
17:26:07pamaurynow go to g#1497: there is a "Download" meny at the top right, it gives you the command to get a copy of those changes. I suggest you use checkout
17:26:09fs-bluebotGerrit review #1497 at : Renamed defines UNALIGNED to ROCKBOX_UNALIGNED - UNALIGNED is already by Matthias Mohr
17:27:05pamaurynow if you type "git log", you commits should be the last two, right ?
17:27:57massapamaury: do I have to remove everything at first? Because when I do a "git checkout master" it says "Already on 'master' Your branch is ahead of 'origin/master' by 2 commits. (use "git push" to publish your local commits)"
17:28:48*pamaury is confused. If it prints that, then the last two commits shown in git log should be yours
17:29:14pamaurycan you pastebin the output of "git log" ?
17:29:25pamauryand "git status"
17:29:34massapamaury: in "git log" the two last are mine - but not in "git rebase HEAD^^^^"
17:29:54pamauryah that's normal: the number of "^" determines how far to go
17:30:00pamauryHEAD^^: go back two in times ;)
17:30:10pamaury(and it prints them backwards)
17:30:31pamauryso if you do git rebase -i HEAD^^ then it should print only two which are yours
17:32:11massapamaury: So I could now just edit "tools/configure" again and the "commit -a" and paste the old Change-Id to the commit message?
17:32:50pamaurydid you do the interactive rebase ?
17:33:14pamaurymassa: ^ if so, then edit configure, then:
17:33:14pamaurygit commit -a −−amend
17:33:40pamauryit will add your changes to the previous commit and normally you don't have to edit the message (you get the same message as the previous one).
17:33:49pamaurythen git rebase −−continue
17:34:18pamaury(NB: the -a means "add all changes" and −−amend means "do not create a new commit but add it to the previous changes")
17:38:39massapamaury: now it seems to work - including the push to gerrit :-)
17:38:52 Join johnb3 [0] (
17:39:44pamauryyes :)
17:40:07pamaurygit can be confusing, I suggest you read a bit online how it works, it's quite logical in the end, when you understand what's going on
17:41:54massayes, it seems I have to do this 8-)
17:42:12pamauryI'm still compiling the new build with my fix...
17:42:18pamauryhopefully it will fix the sim
17:45:39 Quit johnb3 (Ping timeout: 240 seconds)
17:45:45massapamaury: sorry for bugging you - here's my next question: I added myself as reviewer to your g#1499 - but how do I now officially "review"? And add review comments to some files?
17:45:47fs-bluebotGerrit review #1499 at : Fix dangerous casts by Amaury Pouly
17:46:34 Quit paulk-collins (Ping timeout: 240 seconds)
17:46:50massaand how do I add your change to my list of "watched changes"?
17:47:03pamaurymassa: Reply button
17:47:34pamauryI don't know about watched changes
17:50:37massapamaury: I saw your comment directly at the lines of the source - how did you do this?
17:51:03*massa is away for a while
17:51:21pamaurymassa: click on a line and press "c"
17:51:28pamaurymassa: the simulator is working now
17:51:32pamauryI will upload the fix
17:51:42 Nick massa is now known as massa_ (4fe8d32c@gateway/web/freenode/ip.
17:51:51massa_pamaury: great!
17:55:09 Join johnb3 [0] (
17:58:46pamaurymassa_: I've split the rbwrapper change and put it into another commit
18:12:03 Join paulk-collins [0] (
18:12:14 Join MrZeus [0] (~MrZeus@
18:13:11 Quit MrZeus_ (Ping timeout: 260 seconds)
18:22:30 Quit johnb3 (Ping timeout: 240 seconds)
18:25:46 Join Bilgus [0] (
18:27:11__builtinpamaury: can I push g#1502?
18:27:13fs-bluebotGerrit review #1502 at : Fix compilation of puzzles on Windows by Amaury Pouly
18:29:00 Nick massa_ is now known as massa (4fe8d32c@gateway/web/freenode/ip.
18:29:05*massa is here again
18:29:42pamaury__builtin: yes, but make sure you don't push the dependencies with it
18:29:53 Join TheEaterOfSouls [0] (~souls@unaffiliated/theeaterofsouls)
18:30:54fs-bluebotBuild Server message: New build round started. Revision 793c041, 255 builds, 16 clients.
18:39:09__builtinmassa: I'll push the change to rbendian.h that you suggested, too
18:41:25***Saving seen data "./dancer.seen"
18:42:45fs-bluebotBuild Server message: Build round completed after 712 seconds.
18:42:46fs-bluebotBuild Server message: Revision 793c041 result: All green
18:42:47fs-bluebotBuild Server message: New build round started. Revision 42e54c5, 255 builds, 14 clients.
18:43:52 Join webguest073 [0] (
18:45:37massapamaury: your patch g#1499 looks good now! How did you find out the problems in g#1502?
18:45:40fs-bluebotGerrit review #1499 at : Fix dangerous casts by Amaury Pouly
18:45:40fs-bluebotGerrit review #1502 at : Fix compilation of puzzles on Windows by Amaury Pouly
18:46:31massa__builtin: thumbs up! Forgot about this ;-)
18:48:48 Quit webguest073 (Quit: CGI:IRC (Ping timeout))
18:49:06pamaurymassa: you mean g#1501 ?
18:49:08fs-bluebotGerrit review #1501 at : Fix codecs in simulator builds on Windows by Amaury Pouly
18:51:34massapamaury: yes, of course - I'm a bit confused :-)
18:52:03pamauryI used gdb, I posted the offending backtrace
18:53:18massaaah yes, it says something about strlen - I didn't believe it had something to do with this - I would've searched on other places...
18:58:28fs-bluebotBuild Server message: Build round completed after 943 seconds.
18:58:29fs-bluebotBuild Server message: Revision 42e54c5 result: All green
19:04:24massapamaury: how can I integrate your changes to my local git repo? Is cherry picking at gerrit the correct way?
19:04:45pamaurymassa: yes cherry pick
19:08:35massapamaury: "git cherry-pick" or use the button "cherry-pick" at your change in gerrit? If the latter, what do I enter as "Cherry Pick to Branch"?
19:09:28pamaurymassa: re your comment: the #undef is necessary
19:09:44pamauryas for the other methods, yeah we could rename all of them I guess is currently unreachable for me - down?
19:14:04__builtinmassa: yeah, it's a known issue
19:17:39 Join Ishka [0] (
19:17:51pamaurymassa: the downside of renaming everything is that it increases the risk of clashes like the one with strcmp
19:20:01massapamaury: so you suggest to keep it as is and only rename those methods if it's necessary?
19:20:43chrisjjpamaury, does the 'Code-Review +2 Amaury Pouly' that just appeared on g#1428 mean that this change will now go to the master?
19:20:44fs-bluebotGerrit review #1428 at : imx233: workaround hardware dma bug by Amaury Pouly
19:21:57massaI also don't like the #defines there - it's a shame we cannot use namespaces - would be perfect for that purpose :-)
19:22:29massapamaury: did you read my second question about the cherry-picking?
19:23:00IshkaIs there a problem with the 'Themes' site? just times out.
19:23:45massaIshka: forums are also unreachable atm...
19:23:51__builtinthey're on the same server
19:23:59pamaurymassa: which one ?
19:24:03__builtinchrisjj: not necessarily
19:24:35massapamaury: "git cherry-pick" or use the button "cherry-pick" at your change in gerrit? If the latter, what do I enter as "Cherry Pick to Branch"?
19:24:59pamaurymassa: those are two different things
19:25:14pamaurymassa: don't use the cerry pick button unless you understand it ;)
19:25:28pamaurymassa: copy the text from the Download > Cherry-pick menu
19:26:30__builtinchrisjj: it means that a dev has reviewed it, but that doesn't automatically submit it to master
19:28:29Ishkapamaury: I'm looking for a highly visible theme that I can read without glasses.
19:29:50Ishkamassa: I didn't understand your message. "Ishka: forums are also unreachable atm..."
19:31:13__builtinIshka: there's an issue with the server that hosts both the forums and the theme site
19:31:14massaIshka: I meant the site are also unreachable at the moment...
19:32:30Ishkamassa Thank you very much, I'll try later.
19:32:33 Join johnb3 [0] (
19:33:21pamaurymassa: I suggest we push g#1501 as-is
19:33:23fs-bluebotGerrit review #1501 at : Fix codecs in simulator builds on Windows by Amaury Pouly
19:33:43pamauryafter all, it wasn't that hard to find the problem, should that happen in the future
19:33:52pamauryfor other functions
19:34:58massapamaury: it's O.K. - I'm currently in the progress to build it at my local machine
19:36:33massaif that finished and works I'll give you my +1 score (that's all I have) :-)
19:36:34 Quit johnb3 (Ping timeout: 240 seconds)
19:38:16massapamaury: btw, I cannot find any make target to put the simulator binaries into the zip - I assume rasher does use his own script for that...
19:39:23 Join mark___ [0] (6d9f5f8c@gateway/web/freenode/ip.
19:39:31pamaurymassa: apparently make zip doesn't put it in the zip, I'll look into it
19:39:45massapamaury: I still have cast warnings - e.g. in simple_viewer.c
19:40:10pamauryah, maybe I missed some of them, can you pastebin all of them ?
19:41:26pamauryI will double check this is not breaking linux or normal builds
19:42:48mark___Can anybody tell me why Themes and the Forums seem to have disapeard from
19:43:13gevaertsBecause that server is down
19:43:53mark___for maintainance?
19:44:11__builtinit's a bug IIUC
19:46:18massapamaury: I'll pastebin it - when it finished the build...
19:46:44mark___I don't know what ITUC means. Is this somthing that will be sorted in the next few days?
19:47:12__builtin= "if I understand correctly"
19:47:24__builtinand no, probably not
19:50:33pamauryhum apparently g#1501 is breaking normal builds
19:50:35fs-bluebotGerrit review #1501 at : Fix codecs in simulator builds on Windows by Amaury Pouly
19:52:06 Quit Rondom (Remote host closed the connection)
19:54:38 Join Rondom [0] (
19:57:14massapamaury: your're right - my normal ipodvideo build does not work anymore...
19:57:36pamauryit should be simple enough to fix
19:59:10massapamaury: but the 64-bit simulator works now :-)
20:00:49gevaertsbah, sorry
20:01:01gevaertsshell tab-completion doesn't work here...
20:09:08*__builtin really needs to refine his git workflow
20:16:33pamaurymassa: I updadted g#1501
20:16:35fs-bluebotGerrit review #1501 at : Fix codecs in simulator builds on Windows by Amaury Pouly
20:16:42pamaurycan you redo a windows build to check it still works ?
20:16:54 Quit mark___ (Ping timeout: 260 seconds)
20:18:16massapamaury: o.k. - but it still compiles the old 32-bit version...
20:19:44 Join johnb3 [0] (
20:21:53pamaurymassa: I don't understand what you mean
20:22:29 Quit johnb3 (Client Quit)
20:22:37massapamaury: I will start a new compile when the current one has been finished (with the old version of the patch)
20:23:30pamauryI will long into the other warnings
20:27:37massapamaury: the 32-bit build (with old patch version) has finished - it produced one (new?) warning:
20:29:53fs-bluebotBuild Server message: New build round started. Revision 0bdb255, 255 builds, 16 clients.
20:34:23 Quit JdGordon (Ping timeout: 255 seconds)
20:34:39 Join JdGordon [0] (~jonno@rockbox/developer/JdGordon)
20:34:41massapamaury: even the new version does not build the ipodvide:
20:39:04 Join xorly [0] (
20:40:44pamaury g#1501 doesn't touch tlsf.c, it must come from the dangerous cast fix
20:40:46fs-bluebotGerrit review #1501 at : Fix codecs in simulator builds on Windows by Amaury Pouly
20:40:51pamauryis it a native build or sim ?
20:41:29***Saving seen data "./dancer.seen"
20:43:24fs-bluebotBuild Server message: Build round completed after 811 seconds.
20:43:25fs-bluebotBuild Server message: Revision 0bdb255 result: All green
20:43:26fs-bluebotBuild Server message: New build round started. Revision a724dbb, 255 builds, 16 clients.
20:51:18massapamaury: native build
20:51:43massapamaury: the simulator build 32- and 64-bits both compile and work
20:53:54chrisjj__builtin, Does a +2 require that also verification has been made? Or has the RB project removed the verification requirement stated in the docs? > I
20:54:22pamaurymassa: I have updated g#1499 to fix the compile issues
20:54:24fs-bluebotGerrit review #1499 at : Fix dangerous casts by Amaury Pouly
20:54:33pamauryit works on native and linux sim here
20:54:39fs-bluebotBuild Server message: Build round completed after 675 seconds.
20:54:40fs-bluebotBuild Server message: Revision a724dbb result: 0 errors 1 warnings
20:56:16pamaurythe lua warnings won't go away, I don't plan to fix them
20:56:33massapamaury: what does the "Cannot Merge" sign mean at g#1499
20:56:35fs-bluebotGerrit review #1499 at : Fix dangerous casts by Amaury Pouly
20:58:27pamaurymassa: it just means that gerrit has detected a potential conflict if this fix is merged but don't worry about it
20:58:27__builtinchrisjj: there's no requirement for anything
20:58:35pamauryit would be a long to explain the details ;)
21:02:13pamaurybasically rockbox decided to have a "no merge" policy, ie no merge commits. Gerrit enforces that but still let you push commits that are not rebased on HEAD but only if it's obvious that there no conflicts (basically it's going to rebase them and then push). But if it thinks the rebase may fail, it show "Cannot merge" and thus you have to explicitely rebase to convince gerrit that it cleany applies
21:07:31 Join wodz [0] (
21:10:25massapamaury: and gerrit's right - I'm unable to apply the latest version of g#1499: " * branch refs/changes/99/1499/4 -> FETCH_HEAD error: could not apply de9a212... Fix dangerous casts"
21:10:26fs-bluebotGerrit review #1499 at : Fix dangerous casts by Amaury Pouly
21:10:32wodzpamaury: As you have good karma with gdb today, maybe you could help me with g#1389. It crashes because pointers passed to dsp_process become NULL in the beginning of processing. I tried to set watchpoints to find offender but I get "Watchpoint 2 deleted because the program has left the block in which its expression is valid." and then crash again.
21:10:33fs-bluebotGerrit review #1389 at : Port of picoTTS by Marcin Bukat
21:10:54pamaurymassa: you should do a checkout for g#1499
21:10:56fs-bluebotGerrit review #1499 at : Fix dangerous casts by Amaury Pouly
21:13:22pamaurywodz: I'll see what I can do. I'd like to cleanup a bit the windows simulator situation first
21:13:33wodzpamaury: sure
21:13:50pamauryalso it didn't involved the dsp stuff ;)
21:14:03wodzbam, valgrind doesn't like rockbox simulator :/
21:14:31wodzpamaury: I have a feeling the crash is related to the threading but can't find why.
21:17:09pamaurymassa: could you cleanup g#1463 ? I'd like to push it but there are some ugly comments
21:17:11fs-bluebotGerrit review #1463 at : nwztools/scsitool: add option to force/specify series by Amaury Pouly
21:17:16pamaurylike #echo sdl-ccopts="${sdlccopts}"
21:17:20wodzhum, it behaves differently with sdl threading model
21:17:40__builtinwodz: I tried to get the sim to work with valgrind a while back
21:17:56__builtin... to no avail
21:18:08wodz__builtin: It kinda works - you need to pass configure −−sdl-threads
21:18:19__builtinooh, really?
21:19:21__builtinnice... :)
21:19:42massapamaury: I thought I could keep them - to reenable them when needed; but you're right I should clean up it...
21:20:17wodz__builtin: this threading mode makes debugging with gdb easier as well as you don't need to pass handle SIGUSR1 nostop to gdb as well :-)
21:21:00pamaurywodz: I take it that picotts crashes in sim ?
21:21:23wodzpamaury: yes
21:22:55wodzpamaury: but not always. Sometimes it simply stops outputing samples
21:23:52pamauryhow do I test it ?
21:23:58*pamaury doesn't promise anything
21:25:48massapamaury: are those few #echo the only lines you want to cleanup?
21:25:57pamaurymassa: yes I think
21:26:12 Join _mt_ [0] (~MT@2601:482:4402:7b60:a8e4:1c64:d2d1:4ba6)
21:27:15wodzpamaury: get the g#1389, build sim, copy lang_pico folder with voice definitions to simdisk/.rockbox. Copy some text file for tts (first chapter of alice in wonderland is the standard test :-)). Then fire up simulator, in browser long select on txt file (or whatever key is mapped to open context menu) and select open with -> picotts
21:27:17massapamaury: finished
21:27:17fs-bluebotGerrit review #1389 at : Port of picoTTS by Marcin Bukat
21:28:46pamaurywodz: I don't promise I'll look into it tonight but probably more tomorrow
21:28:52pamauryI've spent the day debugging stuff ;)
21:29:00wodzpamaury: thanks anyway :-)
21:29:41pamaurymassa: should I push g#1496 then ?
21:29:43fs-bluebotGerrit review #1496 at : Fix configure for windows simulator builds in cygwin / mingw cross-compiling environm... by Matthias Mohr
21:31:03pamauryjust noticed that if I use rockbox simulator in a windows VM, the sound is crap, as if it was constantly underflowing, maybe a problem with sdl audio
21:31:50massapamaury: yes, and g#1497 also
21:31:52fs-bluebotGerrit review #1497 at : Renamed defines UNALIGNED to ROCKBOX_UNALIGNED - UNALIGNED is already by Matthias Mohr
21:32:26fs-bluebotBuild Server message: New build round started. Revision 955be5b, 255 builds, 16 clients.
21:33:16pamaurycool thanks :)
21:33:31pamauryand I'll push g#1501 unless you object ?
21:33:33fs-bluebotGerrit review #1501 at : Fix codecs in simulator builds on Windows by Amaury Pouly
21:38:47pamauryrasher: ping
21:40:08 Join girafe [0] (
21:41:35 Nick __builtin is now known as _ (~xray@rockbox/developer/builtin)
21:42:04 Nick _ is now known as Guest93277 (~xray@rockbox/developer/builtin)
21:42:28 Quit _mt_ (Read error: Connection reset by peer)
21:43:37 Part Guest93277 (" - Chat comfortably. Anywhere.")
21:45:06fs-bluebotBuild Server message: Build round completed after 760 seconds.
21:45:07fs-bluebotBuild Server message: Revision 955be5b result: All green
21:45:08fs-bluebotBuild Server message: New build round started. Revision d984725, 255 builds, 16 clients.
21:45:13 Join _mt_ [0] (~MT@2601:482:4402:7b60:a8e4:1c64:d2d1:4ba6)
21:50:55 Join __builtin [0] (~xray@rockbox/developer/builtin)
21:56:59fs-bluebotBuild Server message: Build round completed after 712 seconds.
21:57:00fs-bluebotBuild Server message: Revision d984725 result: All green
21:58:10massapamaury: of course you can push it...
22:00:01*pamaury thinks we need to add a win32 and win64 builds to the build server, at least one of each
22:00:33massapamaury: I currently fight again with (or better against?) git... :-(
22:00:46*massa thinks the same as pamaury...
22:01:47pamaurymassa: what is your git problem ?
22:02:43massachrisjj: I updated the simulator build binaries for ipod video and zen in win32 and win64 version at
22:03:08massapamaury: well, it's a bit difficult to explain - but I try :-)
22:04:08 Quit wodz (Quit: Leaving)
22:04:47 Quit michaelni (Ping timeout: 245 seconds)
22:04:52fs-bluebotBuild Server message: New build round started. Revision 16d1788, 255 builds, 16 clients.
22:05:02massaI initially created my local rockbox repository in a directory rockbox.gitorig - from that I also did the gerrit pushes
22:06:08massathen I created another directory rockbox.gitdev and made a clone of the rockbox.gitorig (a local clone)
22:07:33massain that directory I cherry-picked all the changes of your gerrit # - and I also used that directory to build (in subdirectories)
22:08:34massanow I'm unable to get them both to the same (last) rockbox change.
22:09:07massarockbox.gitorig seems to be fine - at least I can see all commits with "git log"
22:10:31pamaurymassa: to get back normal, I suggest: (warning: you'll loose any change):
22:10:31pamaurygit reset −−hard HEAD # delete any changes compared to current HEAD
22:10:31pamaurygit checkout master # make sure you are on master branch
22:10:31pamaurygit reset −−hard origin/HEAD # make sure you are at the same commit as remote
22:10:34pamaury# and then possibly update:
22:10:37pamaurygit pull −−rebase
22:14:27massaI already did "git reset −−hard" but without additional parameter
22:14:42massaand why do I need several reset commands?
22:15:47pamauryif you reset −−hard, you get back to HEAD but 1) HEAD might be different from remote HEAD (for example if you cherry-pick commits). So in general you want git reset −−hard origin/HEAD
22:16:30pamauryBUT, that only works if you have to local changes, thus you need the first reset −−hard. Also if you did a checkout previously, you're not on master anymore, thus the checkout to master to make sure you're back on the right branch
22:16:47massaCould you recommend any good documentation about that git basics? Not the basic basics - I know what revision control etc. is - I use it every day, with svn and hg but I'm really confused with that git handling...
22:17:12 Join michaelni [0] (
22:18:36fs-bluebotBuild Server message: Build round completed after 824 seconds.
22:18:37fs-bluebotBuild Server message: Revision 16d1788 result: All green
22:19:26pamaurymassa: as I recall: is quite nice with picture
22:19:53pamaury(there is a menu on the left with different topics)
22:24:52massapamaury: thanks - will have a deeper look at them - could it be that the windows TortoiseGit is not the best client to use? I'm very used to the Tortoise-Clients (long ago for CVS, now SVN and HG) - but the GIT version seems to be way behind the other...
22:26:14massaNormally I do the daily work with a GUI and only special things with command line - but here it seems that it's better to do everything with command line
22:26:16pamaurymassa: I don't know, I've only ever used the command line git
22:26:28pamaurygit is not very UI friendly imo
22:27:43*__builtin sometimes uses `git gui'
22:27:45__builtinand `gitk'
22:28:39 Quit shmibs (Quit: leaving =o)
22:29:39 Quit TheEaterOfSouls (Quit: ChatZilla 0.9.93 [Firefox 50.1.0/20161208153507])
22:29:44massapamaury: currently only g#1499 is need as cherry pick now?
22:29:46fs-bluebotGerrit review #1499 at : Fix dangerous casts by Amaury Pouly
22:30:51pamaurymassa: yes
22:33:35 Join shmibs [0] (
22:40:30 Quit shmibs (Quit: leaving =o)
22:41:30***Saving seen data "./dancer.seen"
22:51:12massapamaury: win64 built has finished - no more warnings and it still works :-)
22:53:33pamaurygood :)
23:00:23 Quit petur (Quit: Leaving)
23:00:46massapamaury: win32 built has also finished - no warnings and works!
23:06:42pamaurymassa: cool thanks
23:07:12pamauryI'll still wait a bit more, I sent an email on the ML about it, to make sure other have a chance at reviewing it since it touches many things
23:07:59 Quit lebellium (Quit: ChatZilla 0.9.93 [Firefox 50.1.0/20161208153507])
23:17:48massapamaury: native ipodvideo build has also finished without warnings - I'm now tranferring it to my device
23:24:55massapamaury: the native build does _NOT_ work - it crashes directly after the start :-(
23:27:37massapamaury: message is: data abort at 0008BF90 (0)
23:27:54massapc: 0008BF90 sp:4060AE58
23:28:00massaA: 0008D21C
23:28:08massaA: 000347C4
23:28:20massaA: 00033344
23:29:09massabt end
23:31:10pamaurymassa: without symols the backtrace is useless though. Did you make sure that HEAD works (ie without g#1499) ?
23:31:12fs-bluebotGerrit review #1499 at : Fix dangerous casts by Amaury Pouly
23:32:37 Quit skapazzo (Quit: leaving)
23:32:48 Quit JanC (Remote host closed the connection)
23:34:07pamaurymassa: I'll try it on my fuze+ to see if I can find the problem
23:35:19massapamaury: that's what I do now - compile a HEAD ipodvideo build - we'll see (but that's the last action for today - then I'll leave and unfortunately I don
23:35:34massaI don't think I'll find much time next week...
23:35:52 Quit cc___ (Ping timeout: 256 seconds)
23:36:04massa(including next weekend - I'm skiing ;-)
23:37:02massapamaury: can I force a backtrace / core or similar at the device?
23:37:28pamaurymassa: not sure what you mean
23:38:04massaI mean can rockbox produce a crashlog or a core file or something similar?
23:38:17pamauryno, just a backtrace
23:38:42pamaurywhich is only useful if you have the file, which you did but I stupidly ask you to try another build ;)
23:38:57pamaurylet me see if I can reproduce it on my fuze+, that would be the simplest way to fix it
23:40:46pamauryit works on the fuze+ :-/
23:41:48pamaurymassa: what could help is if you redo the fault build, post the backtrace and upload somewhere
23:43:43massapamaury: the HEAD build works :-)
23:43:57massapamaury: O.K. I'll redo the build...
23:44:12pamauryok good, at least we know g#1499 is faulty
23:44:13fs-bluebotGerrit review #1499 at : Fix dangerous casts by Amaury Pouly
23:44:33massaor my build was faulty...
23:44:45pamauryin doubt, make clean
23:45:10 Join JanC [0] (~janc@lugwv/member/JanC)
23:48:32massapamaury: I always "make clean" to make sure everything got build ;-)
23:48:53pamaurydid you use Rolo when you put the new firmware ?
23:52:37 Quit n3m9 (Read error: Connection reset by peer)
23:55:40 Quit paulk-collins (Quit: Leaving)
23:57:49massapamaury: Do I use what? :-) I create a Zip and unzip it at the device which is connect as mass storage device...
23:58:42pamaurymassa: rolo is when you see the message "the firmware has been changed, do you want to Reload" (ROLO=reload)

Previous day | Next day