00:01:06 | pamaury | you can commit stuff to bootloaders since we do not distribute nightly builds for those |
00:01:34 | pamaury | so it's "okay" for them to be broken/untested (although you should avoid it as much as possible of course) |
00:02:14 | TheSeven | so... what do we do about android nightlies? |
00:02:49 | prof_wolfff | pamaury: i think the bootloader is tested, there are some patches for 'preparation' for bootloader but not bootloader itself, i was thinking on commiting these patches before tagging and then tag the bootloader itself, right? |
00:03:41 | pamaury | yes that's fine |
00:03:47 | prof_wolfff | nice! |
00:03:47 | TheSeven | [Saint]: around? |
00:04:01 | pamaury | as long as you only tag a well-known working and tested bootloader |
00:06:04 | prof_wolfff | well.. i think the bootloader is tested as much of possible at this point, if there are no complains i am going to commit it... |
00:08:06 | prof_wolfff | maybe we need some info on web, there are README files on bootloader, i will appreciate if someone could post this info on web, my english is a bit limited... |
00:08:33 | pamaury | what do you mean ? |
00:09:09 | pamaury | about README files on bootloader ? I don't know what you are referring to |
00:09:28 | prof_wolfff | that maybe we need to publish some info on how to install the bootloader on Rockbox web, as other target are doing... |
00:10:55 | pamaury | ah, well, we usually advise to use RBUtil |
00:11:11 | pamaury | but some targest (many targets) also document the manuel procedure on the wiki |
00:11:59 | pamaury | it's very device specific so if you know how it works, you are probably the best one to do it. If you want you can write down something on the wiki and I will cleanup the english after you ;) |
00:12:11 | prof_wolfff | there are patches that add install for Classic on RBUTIL, but i think that bluebrother should revise them |
00:12:47 | pamaury | ok, so how do you install it currently then ? |
00:14:41 | prof_wolfff | pamaury: i write instructions on mk6gboot/README, to use mk6gboot.exe utility, if you need more info please ask me |
00:15:04 | pamaury | ah, this is a bit unusual |
00:15:23 | prof_wolfff | sorry, please tell me another procedure |
00:15:50 | pamaury | we usually write it on the wiki as I said |
00:16:46 | pamaury | for example: http://www.rockbox.org/wiki/SansaFuzePlusPort in the Building the bootloader section, |
00:17:34 | prof_wolfff | well... i think that the info in README is the info needed on the wiki, but i would appreciate if some one could publish it and tell me if more info is needed |
00:17:37 | *** | Saving seen data "./dancer.seen" |
00:18:03 | prof_wolfff | i will send all the needed info |
00:18:15 | pamaury | what we tend to do is to have for each target (say it's called MyTarget) two wiki pages: MyTarget that documents the device, PCB, etc and MyTargetPort that documents port status, any useful dev information, install instruction for bootloader, troubleshooting, etc |
00:18:30 | pamaury | ok, where is the README file ? |
00:19:23 | pamaury | is it the one in g#1266 |
00:19:24 | pamaury | ? |
00:19:24 | fs-bluebot_ | Gerrit review #1266 at http://gerrit.rockbox.org/r/1266 : iPod Classic: dualboot-installer for bootloader (mk6gboot) by Cástor Muñoz |
00:20:19 | prof_wolfff | http://gerrit.rockbox.org/r/#/c/1266/4/rbutil/mk6gboot/README |
00:23:11 | pamaury | I see, you can keep the README file, that's a good idea. I can/copy/reformat part of this README and put it on the http://www.rockbox.org/wiki/IpodClassicPort page if you want, I'll add something that says the code is commited yet |
00:23:51 | prof_wolfff | nice, actually mk6gboot runs on Windows and Linux, i will try to borrow a Mac to implement it on Apple |
00:26:49 | | Quit girafe (Read error: Connection reset by peer) |
00:29:32 | | Join MasterplasterPL [0] (~WIN@188.146.80.167.nat.umts.dynamic.t-mobile.pl) |
00:30:14 | prof_wolfff | the idea on near future is to start a new target Nano3G that should be very similar to Classic, AFAIK the 'only' needed things are the PMU driver and the FLT, the later might be a bit complicated (7 knowns more about this), but once it is done, other 3G HW is like Classic |
00:34:22 | | Quit MasterplasterPL (Client Quit) |
00:41:37 | TheSeven | IIUC the PMU and codec chips are different, and probably also the LCD |
00:41:52 | TheSeven | (although the command set will likely be similar, the init sequence likely won't) |
00:42:11 | TheSeven | also the SDRAM config |
00:42:26 | TheSeven | PMU and flash stuff are the biggest problems I'd say |
00:42:52 | TheSeven | I vaguely remember the nano3g PMU being more similar to nano4g than classic |
00:44:00 | | Nick suYin is now known as suYin`OFF (mysuyin@server2.shellfire.net) |
00:47:54 | TheSeven | [Saint], pamaury, gevaerts, whoever else might be concerned... any objections vs. switching to statically linked libsdl for crosscompiled windows sims? |
00:48:17 | TheSeven | http://pastie.org/10836352 |
00:48:22 | pamaury | no, tat sounds like a good idea |
00:48:39 | pamaury | I don't even know why we were no doing it before |
00:48:42 | pamaury | *not |
00:48:46 | prof_wolfff | TheSeven: probably the hard thing on Nano3G will be the NAND, but the other HW seems very similar to Classic, the Classicbootloader was written thinking on other s5l8702 targets, do you think this NAND could be similar to Nano2G? |
00:49:02 | TheSeven | pamaury: can you just commit that then? ;) |
00:49:14 | pamaury | TheSeven: sure |
00:50:30 | TheSeven | prof_wolfff: I failed to even find the relevant code in the (messy) OF disassembly |
00:51:09 | prof_wolfff | BTW, i am testing the Designware USB driver on Nano2G, i think that it need a couple of functions, but if the HW behaves as it should then it will work!!! |
00:52:25 | prof_wolfff | i also did a small probe of designware USB driver on Sansa and it also works |
00:53:46 | prof_wolfff | with a small probe i am referring to storage+usb_debug+usb_hid concurrently |
00:55:23 | | Quit lebellium (Quit: ChatZilla 0.9.92 [Firefox 46.0.1/20160502172042]) |
00:56:36 | pamaury | TheSeven: am I right that this patch is statically link everything ? not just sdl |
00:57:28 | pamaury | like libgcc |
00:57:56 | TheSeven | yes |
00:58:03 | TheSeven | libgcc would be statically linked anyway |
00:58:36 | TheSeven | but dynamic libsdl also linked dynamic libgcc, which was a problem because that was missing from the zip (and figuring out where to pull that from is non-trivial) |
00:59:58 | TheSeven | so just making one self-contained .exe file seems easiest and best for everything IMO |
01:00 |
01:02:51 | TheSeven | hm, where do I get rbclient.pl? |
01:03:34 | prof_wolfff | The Seven: about the BBT code on Classic HDD, really i am surprised about it, discovered it a bit time ago, it is necessary or can be wiped? |
01:03:37 | pamaury | TheSeven: g#1317 looks fine ? |
01:03:38 | fs-bluebot_ | Gerrit review #1317 at http://gerrit.rockbox.org/r/1317 : configure: statically link sdl on win32 cross compile by Amaury Pouly |
01:04:01 | pamaury | TheSeven: I think it's on another repository |
01:04:26 | pamaury | http://git.rockbox.org/?p=www.git;a=blob_plain;f=buildserver/rbclient.pl;hb=HEAD |
01:04:45 | pamaury | in www repo: http://git.rockbox.org/?p=www.git;a=summary |
01:07:26 | TheSeven | prof_wolfff: that can be wiped, it's something that we invented to work around some semi-broken disks |
01:07:35 | TheSeven | only present on a handful of ipods |
01:08:07 | prof_wolfff | yes, i saw it on emCore nano code, so supposed i was intended for NAND, righ? |
01:09:06 | TheSeven | no, that's ipod classic hdd stuff |
01:09:28 | TheSeven | nand BBT is handled deeply buried in the FTL, completely different technique |
01:09:43 | TheSeven | argh... |
01:09:44 | TheSeven | 2016-05-14 01:08:35 Error: You specified arch mipsel but the output of 'mipsel-elf-gcc −−version' did not include '4.1.2'. |
01:09:51 | prof_wolfff | oh!, but... it is not supposed that HDD firmware will deal with bad HDD sector??? |
01:10:05 | TheSeven | this is for disks where that ran out of spares |
01:10:11 | TheSeven | because it has just 4095 spares |
01:10:27 | TheSeven | and we were using some devices with ~300k bad sectors for development |
01:11:01 | prof_wolfff | ah!, really 4095 sector are not so much for devices that are used for running... |
01:11:25 | TheSeven | pamaury: looks fine |
01:11:38 | TheSeven | my build client should also be back up now, so we can see if it joins the round |
01:11:52 | TheSeven | (still lacking mipsel due to version check, and android) |
01:12:42 | | Quit smoke_fumus (Read error: Connection reset by peer) |
01:13:47 | | Quit ender` (Quit: I just found out what Alt+F4 does.) |
01:14:25 | prof_wolfff | do anyone have documentation for the Designware driver, i am using s3c6400 and other old docs found on the web, ATM i think i know all functionally, but always is good to have official docs |
01:15:13 | TheSeven | the stm32f2 reference manual has some reasoably useful docs |
01:15:28 | TheSeven | probably for a slightly newer core version |
01:16:44 | pamaury | the stm32 is the best one, but double check/triple the config registers, there are many variations in the cores |
01:17:18 | pamaury | it's also not very clear to me what are the differences between the versions |
01:17:49 | prof_wolfff | nice!, i was using the Linux driver, RT3050 and xmd4500, realized that Nano2G uses NO shared fifos, but i think in a couple of days it should be solved |
01:18:09 | TheSeven | mostly different TX fifo handling, slight variations in the transfer queueing |
01:18:32 | prof_wolfff | sorry, Nano2G uses shared Tx fifos, it is causing headache for me |
01:19:04 | TheSeven | yup, and IIUC that has some nasty hardware race conditions |
01:20:07 | prof_wolfff | yes, and the periodic queue is also broken, i need to reset it everytime a packed is sent for it to work correctly, ATM i need a couple of functions to deal with flush |
01:20:56 | prof_wolfff | but i think it will work on RB |
01:21:57 | fs-bluebot_ | Build Server message: New build round started. Revision 60fb707, 255 builds, 12 clients. |
01:37:27 | TheSeven | what's going on with the build? |
01:37:51 | TheSeven | my client isn't doing anything anymore since 2 minutes, yet the round isn't done |
01:40:03 | TheSeven | hmmmm... have we broken something? |
01:45:56 | pamaury | I don't know, I know sometimes the build server gets stuck but I don't manage it so I don't really know how it works |
01:47:07 | TheSeven | I'm the one to blame :P |
01:47:13 | TheSeven | I was the only remaining build client that could do android |
01:47:24 | TheSeven | and I haven't set that up yet on the new system |
01:49:04 | pamaury | ah, good point |
01:49:19 | pamaury | that ought to be fixed |
01:49:52 | * | TheSeven waits for instructions from [Saint] |
01:56:16 | TheSeven | pamaury: what should we do about the mips version thing? |
01:56:43 | pamaury | fix configure I guess, why ? |
01:57:06 | TheSeven | the build client complains that it's a wrong version |
01:57:18 | TheSeven | so I guess we need to patch that as well |
01:57:42 | pamaury | hum, I didn't know the build clients was making checks like this |
01:57:43 | TheSeven | which raises the question whether we want to force build clients to use the new one or use a mix |
01:58:47 | pamaury | yeah, that's a good question, I think the idea is that we have a somewhat reproducible setup by using the same version everywhere |
01:59:12 | pamaury | but you could just change the version in rbclient.pl |
01:59:21 | pamaury | (the build server won't know :-p) |
01:59:23 | TheSeven | that script self-updates |
01:59:37 | pamaury | ah... |
01:59:55 | pamaury | well, then I think we should commit a change to configure and rbclient |
01:59:59 | TheSeven | so we might do something like with m86k-gcc452 |
02:00 |
02:00:12 | TheSeven | or just tolerate both versions |
02:00:38 | pamaury | yes that's a good idea, who has a mips toolchain on the build bots these days ? is anyone still using 4.1.2 |
02:00:46 | TheSeven | changing just the version check without changing the compiler name will kill all clients that have the old one, and prevent them from building any other targets as well |
02:01:22 | pamaury | yes, that's why I was wondering who still has the old one |
02:01:33 | TheSeven | which targets does that build? |
02:02:28 | pamaury | some ondas |
02:02:36 | pamaury | wodz uses it for atj |
02:02:45 | pamaury | and I will use it for fiio x1 and xduoo x3 |
02:03:03 | pamaury | though at the moment only ondas are built |
02:03:13 | pamaury | and probably no one has them anymore |
02:03:15 | TheSeven | ok, I'll flick through some logs, which clients appear there |
02:03:27 | TheSeven | someone must still have them besides me |
02:03:30 | TheSeven | these didn't fail |
02:05:19 | pamaury | then we should commit a new line to rblient to support two versions |
02:06:25 | | Quit prof_wolfff (Read error: Connection reset by peer) |
02:06:54 | | Join prof_wolfff [0] (~prof_wolf@82.159.0.123.dyn.user.ono.com) |
02:09:33 | | Quit TheSeven (Disconnected by services) |
02:09:42 | | Join [7] [0] (~quassel@rockbox/developer/TheSeven) |
02:13:57 | [7] | lillebror1-zagor n07-roolku b0hoon-vm1-b0hoon homepc-petur Type-R-ej0rge n16-roolku n17-roolku just to name a few |
02:14:04 | [7] | seems to be part of the "standard setup" |
02:17:38 | *** | Saving seen data "./dancer.seen" |
02:17:46 | | Join JdGordon_ [0] (~jonno@rockbox/developer/JdGordon) |
02:20:07 | | Quit JdGordon (Ping timeout: 246 seconds) |
02:27:35 | | Join bluebrother [0] (~dom@rockbox/developer/bluebrother) |
02:28:29 | | Join fs-bluebot [0] (~fs-bluebo@xd9baf450.dyn.telefonica.de) |
02:30:52 | | Quit fs-bluebot_ (Ping timeout: 260 seconds) |
02:31:05 | | Quit bluebrother^ (Ping timeout: 260 seconds) |
02:41:24 | [Saint] | [7]: sorry - shits been pretty hectic here, ummm... |
02:41:24 | [Saint] | http://pastebin.com/Bj8TRB5z |
02:42:23 | [Saint] | that's just a big inelegant lump of all things Android. |
02:44:09 | [7] | what kind of stuff do I need to set up for that to work? |
02:44:24 | [7] | I assume that I'll at least need some NDK? |
02:45:07 | [Saint] | Just the NDK and the SDK as they come. |
02:45:12 | pamaury | [7]: you'll need sdk and ndk |
02:45:25 | [Saint] | Then you just need to grab SDK platform 19. |
02:45:27 | pamaury | debian has the sdk packaged, maybe also ubuntu |
02:45:34 | [Saint] | and build-tools 19.1 |
02:45:49 | pamaury | search for android-tools and android-sdk in aptitude |
02:46:13 | pamaury | you'll need to download the ndk though |
02:46:33 | [Saint] | this creates a chicken/egg problem with the build system. |
02:48:08 | [Saint] | you also need openjdk 7 or 8. |
02:48:34 | [Saint] | (honestly I'm surprised we can build under openjdk8 without it bitching) |
02:49:18 | [Saint] | then just make sure that the ANDROID_SDK_PATH and ANDROID_NDK_PATH env variables are set. |
02:52:51 | [Saint] | can anyone think of a clean way to do "always use the most modern android build tools present"? |
02:53:47 | [Saint] | there's no real reason why we prefer to use $(BUILD_TOOLS_VERSION)/* |
02:54:25 | [Saint] | It really doesn't need to be tied to the platform target. |
02:54:59 | [Saint] | it can't be lesser than the platform target, but backwards compat ensures it can be greater than. |
02:59:40 | | Quit pamaury (Ping timeout: 260 seconds) |
03:00 |
03:07:46 | | Quit [7] (Ping timeout: 276 seconds) |
03:16:10 | | Quit ZincAlloy (Quit: Leaving.) |
03:25:15 | | Join TheSeven [0] (~quassel@rockbox/developer/TheSeven) |
03:28:16 | | Quit krabador (Remote host closed the connection) |
03:44:59 | TheSeven | [Saint]: looks like the ubuntu android-sdk package doesn't contain an sdk manager, it just pulls in build-tools etc. |
03:45:09 | TheSeven | so how do I get the SDK platform installed then? |
03:46:03 | [Saint] | https://dl.google.com/android/android-sdk_r24.4.1-linux.tgz |
03:46:27 | [Saint] | then run 'android' from the tools directory. |
03:46:45 | TheSeven | ok so what does the android-sdk ubuntu package even do then? |
03:46:56 | [Saint] | exist pointlessly. |
03:48:12 | [Saint] | I can't really think of any situation where you'd want to just blindly pull in all the platforms and build tools. |
03:56:15 | | Join JdGordon [0] (~jonno@rockbox/developer/JdGordon) |
03:59:19 | | Quit JdGordon_ (Ping timeout: 244 seconds) |
04:00 |
04:08:30 | | Quit prof_wolfff (Ping timeout: 265 seconds) |
04:17:41 | *** | Saving seen data "./dancer.seen" |
04:21:37 | | Join prof_wolfff [0] (~prof_wolf@82.159.0.123.dyn.user.ono.com) |
04:28:42 | | Join JanC_ [0] (~janc@lugwv/member/JanC) |
04:29:43 | | Nick JanC is now known as Guest71745 (~janc@lugwv/member/JanC) |
04:29:43 | | Nick JanC_ is now known as JanC (~janc@lugwv/member/JanC) |
04:30:47 | | Quit Guest71745 (Ping timeout: 276 seconds) |
05:00 |
05:22:16 | | Quit nlogex (Quit: WeeChat 1.5) |
05:51:11 | | Quit michaelni (Ping timeout: 246 seconds) |
06:00 |
06:03:49 | | Join michaelni [0] (~michael@chello213047041020.graz.surfer.at) |
06:17:45 | *** | Saving seen data "./dancer.seen" |
06:35:21 | duo8 | is it absolutely not possible to do anything on atj players? |
06:39:05 | [Saint] | Not as far as what you know Rockbox today to be is concerned, not really, no. |
06:39:54 | duo8 | so rockbox would be able to do as much as on the old archos player/recorder? |
06:40:27 | [Saint] | I am shortening the scope of the conversation as I assume you're talking about this in relation to the Clip Jam/Sport. |
06:41:16 | [Saint] | and, less, believe it or not. |
06:41:58 | duo8 | wow |
06:42:23 | duo8 | new mips atj, doesn't seem too different from 2006 z80 atj |
06:43:19 | [Saint] | from what I understand of the Sport/Jam, the theme engine alone, without any other core code, would overflow the available resources. |
06:43:53 | duo8 | how'd sandisk get the fw on there? doesn't look like standard atj provided stuff |
06:44:56 | [Saint] | it looks like someone chewed up a generic s1mp3 firmware and threw it up. |
06:47:08 | [Saint] | Basically, if Rockbox ever were to happen for those devices, it would be completely and totally unrecognizable as the Rockbox we know today. I would think. |
06:47:29 | [Saint] | You would need to make so many sacrifices you wouldn't even recognize it. |
06:48:23 | duo8 | wasn't there debate on how they got album art working? |
06:50:05 | [Saint] | Yeah. From what I know of the available resources it doesn't make a lot of sense at all. |
06:50:23 | [Saint] | I would guess the LCD has its own cache. |
08:00 |
08:17:48 | *** | No seen item changed, no save performed. |
09:00 |
09:46:53 | | Join paulk-collins [0] (~paulk@gagarine.paulk.fr) |
09:58:38 | | Join xorly [0] (~xorly@ip-89-176-117-132.net.upcbroadband.cz) |
10:00 |
10:17:50 | *** | No seen item changed, no save performed. |
10:27:19 | | Join rela [0] (~x@pdpc/supporter/active/rela) |
10:44:27 | TheSeven | I can't really believe that this thing has only 96k of RAM |
10:44:56 | TheSeven | that just seems prohibitive if that needs to contain both code and data of an FTL, audio decoder and LCD UI |
10:48:39 | | Join paulk-collins_ [0] (~paulk@gagarine.paulk.fr) |
10:49:02 | | Quit paulk-collins (Ping timeout: 246 seconds) |
11:00 |
11:02:33 | duo8 | audio is probably h/w? |
11:15:36 | TheSeven | not according to the datasheet that I found |
11:19:28 | | Join freenoodle [0] (~jr@p57A08D67.dip0.t-ipconnect.de) |
11:37:54 | | Join ender` [0] (krneki@foo.eternallybored.org) |
11:55:54 | TheSeven | /home/rockbox/dailybuilds/rockbox/apps/action.c:26:18: fatal error: lang.h: No such file or directory |
11:56:02 | TheSeven | what's the matter with that? |
12:00 |
12:04:40 | | Join smoke_fumus [0] (~smoke_fum@188.35.176.90) |
12:04:48 | TheSeven | *drumroll* |
12:06:56 | fs-bluebot | Build Server message: Build round completed after 38700 seconds. |
12:06:58 | fs-bluebot | Build Server message: Revision 60fb707 result: 170 errors 1 warnings |
12:09:21 | | Join freenoodle1 [0] (~jr@p57A08E98.dip0.t-ipconnect.de) |
12:11:21 | | Quit freenoodle (Ping timeout: 265 seconds) |
12:15:02 | TheSeven | should I make an image of that box available? |
12:17:53 | *** | Saving seen data "./dancer.seen" |
12:34:19 | | Join lebellium [0] (~chatzilla@89-93-179-187.hfc.dyn.abo.bbox.fr) |
12:48:20 | | Nick suYin`OFF is now known as suYin (mysuyin@server2.shellfire.net) |
12:50:39 | | Quit PurlingNayuki (Remote host closed the connection) |
12:51:13 | | Join PurlingNayuki [0] (~Thunderbi@v163-44-154-238.a00f.g.sin1.static.cnode.io) |
12:55:12 | | Join Rower [0] (husvagn@d83-183-134-99.cust.tele2.se) |
13:00 |
13:03:20 | | Join petur [0] (~petur@rockbox/developer/petur) |
13:04:50 | | Join pamaury [0] (~pamaury@rockbox/developer/pamaury) |
13:09:41 | | Quit freenoodle1 (Ping timeout: 276 seconds) |
13:27:16 | pamaury | duo8: TheSeven: re atj: first the atj seems to have a huge ROM with many functions like all the libc, then my guess is that they do like on s1mp3: split the binary in many smalls sections and load them when needed. It's just super super slow |
13:27:59 | pamaury | as for the lcd and album art, you just need to store it in the lcd controller, you don't even need to put it in RAM |
13:32:02 | pamaury | for example if you look at https://github.com/Suber/PD196_ATJ2127 which is the best leaked source of information about ATJ2127, you can clearly see that the drivers are manually "banked" |
14:00 |
14:01:02 | | Quit pamaury (Ping timeout: 276 seconds) |
14:17:54 | *** | Saving seen data "./dancer.seen" |
14:29:57 | | Join pamaury [0] (~pamaury@rockbox/developer/pamaury) |
14:39:37 | TheSeven | ouch |
14:39:57 | TheSeven | I still think that something like an FTL would need tens of KB of RAM structures though |
14:53:41 | | Join krabador [0] (~krabador@host108-63-dynamic.44-79-r.retail.telecomitalia.it) |
14:53:56 | | Quit krabador (Changing host) |
14:53:56 | | Join krabador [0] (~krabador@unaffiliated/krabador) |
14:58:27 | | Quit Rower (Ping timeout: 252 seconds) |
15:00 |
15:03:33 | | Quit jtdesigns01 (Remote host closed the connection) |
15:17:02 | | Join jtdesigns01 [0] (~quassel@2601:400:8000:34f5:230:bdff:fe71:cebd) |
15:38:48 | TheSeven | so could anyone make use of a prebuilt build bot VM image? |
15:38:53 | TheSeven | (i.e. rootfs tarball) |
15:39:58 | | Quit xorly (Read error: Connection reset by peer) |
15:40:26 | TheSeven | or could any of the other build client owners please add the android19 toolchain? |
15:40:26 | | Join xorly [0] (~xorly@ip-89-176-117-132.net.upcbroadband.cz) |
15:40:37 | TheSeven | that would reduce the reliance of the build system on my server quite a bit |
15:55:22 | TheSeven | I'm having some issues with the android builds |
15:55:39 | TheSeven | looks like they're failing intermittently with lang.h missing |
15:55:49 | TheSeven | wondering if that's related to MAKEFLAGS=-j4 |
15:56:28 | TheSeven | rasher.dk/rockbox/android/">http://rasher.dk/rockbox/android/ shows that only roughly half of them built |
16:00 |
16:09:02 | | Quit petur (Quit: Leaving) |
16:17:55 | *** | Saving seen data "./dancer.seen" |
16:22:34 | | Join andiator [0] (5492fc99@gateway/web/freenode/ip.84.146.252.153) |
16:24:51 | andiator | Hi, how do I enable the debug log file on the RB? I managed to get the logf on the screen, but I can only see few short lines. |
16:25:20 | andiator | I'm trying to figure out, why the Cowon D2 does crash |
16:30:51 | | Join kuba-orlik_ [0] (~quassel@188.146.141.41.nat.umts.dynamic.t-mobile.pl) |
16:36:42 | pamaury | andiator: first enable logf when you configure the build |
16:37:09 | pamaury | then in a given file, to enable logf, first #define LOGF_ENABLE before #include "logf.h" and then use logf() normally |
16:37:19 | pamaury | you can see the log file in System > Debug > Show Log file |
16:38:42 | andiator | pamaury: thanks, do you know where this file is created and how is it named? |
16:38:53 | pamaury | it is not stored in a file |
16:38:57 | pamaury | unless you use logfdisk |
16:39:05 | andiator | My problem is, the D2 crashes on writing empty files |
16:39:10 | pamaury | but you can dump if to a file using Debug > Debug > Dump log file |
16:39:25 | pamaury | crash as in panic ? |
16:39:33 | andiator | yes, panic |
16:39:48 | pamaury | panic screen displays the last lines of logf |
16:40:31 | pamaury | there is no solution to this problem, if the system crash you cannot dump the log |
16:40:48 | pamaury | the only way I see would be to make the log buffer scrollable in the panic screen |
16:41:33 | TheSeven | what's the panic message |
16:42:23 | andiator | well with logf on the screen I know that it panics trying to write the cash to NAND, which on D2 is RO. |
16:42:57 | andiator | Strange thing is, it works if the the file is not empty. RB on D2 is running on the SD card |
16:43:34 | andiator | cash is cache |
16:43:55 | TheSeven | what kind of cache? |
16:44:25 | andiator | disk cache I suppose? |
16:44:36 | TheSeven | so some kind of cache flush call? |
16:44:43 | TheSeven | that should be a no-op on readonly nand... |
16:45:15 | | Quit PurlingNayuki (Remote host closed the connection) |
16:45:38 | andiator | In the storage.c in the function storage_write_sectors() this piece of code is called: |
16:45:39 | andiator | #if (CONFIG_STORAGE & STORAGE_NAND) case STORAGE_NAND: return nand_write_sectors(IF_MD(ldrive,)start,count,buf); #endif |
16:45:59 | andiator | and this one returns -1, which RB doesn't like and panic |
16:46:11 | TheSeven | hm, do you know who's calling that? |
16:46:19 | pamaury | which function calls it ? the panic screen shows a backtrace |
16:47:31 | andiator | On the very first start, without an existing file: |
16:47:32 | andiator | fat_create_file("nvram.bin",200F01E4,200F0234) cache_commit() dc_commit_all() dc_write_callback() −−-> *PANIC* |
16:48:47 | andiator | on the next start the file does exists, but is empty |
16:48:51 | andiator | dc_cach_probe() dc_write_callback() −−-> *PANIC* |
16:49:21 | andiator | I don't see more calls, because the screen is too small |
16:50:34 | TheSeven | hm, do files without a disk number refer to nand or sd on cowond2? |
16:51:04 | pamaury | they refer to the main storage, I think on cowond2 it's supposed to be SD... |
16:51:31 | andiator | IRC the SD card is the main storage |
16:52:03 | andiator | If I copy this file(s) from the old version, there is no panic even if they are written, so settings are working, for example |
16:52:13 | pamaury | but probably the problem is the cache flush, not the file itself ? |
16:52:38 | TheSeven | I can't seem to see any check for readonly storage when opening a file... that might cause the error to be reported rather late (and hard to track down) |
16:53:13 | andiator | strange thing it happens only on empty files (size = 0) |
16:53:37 | TheSeven | you mean if nvram.bin is empty, or are you testing with some other particular file? |
16:54:06 | andiator | This files are even created on SD, but the next write crashes |
16:54:13 | andiator | not only nvram.bin |
16:54:41 | andiator | but settings.cfg too and now playing position or so |
16:55:03 | | Join T-rack [0] (4d638c8d@gateway/web/freenode/ip.77.99.140.141) |
16:55:36 | T-rack | Hi |
16:55:39 | andiator | If I copy this files, RB is running for some minutes, I can change the settings etc, but it crashes later |
16:56:06 | | Quit krabador (Remote host closed the connection) |
16:57:13 | TheSeven | I'm a bit puzzled that the various calls in your backtrace don't mention a volume ID, which suggests that this is not a multivolume build |
16:57:35 | pamaury | actually in config/cowond2.h I can't fine any mention of sd being the main storage, or maybe I am missing something |
16:57:35 | TheSeven | in which case I don't get why it would *ever* try to write to nand |
16:57:59 | pamaury | T-rack: ask your question if you have one |
16:58:58 | TheSeven | then again the dc_writeback_callback will always have args |
16:59:02 | TheSeven | so maybe that's just a backtrace thing |
16:59:04 | pamaury | TheSeven: cowond2 has HAVE_MULTIDRIVE which implies multi volume |
16:59:10 | andiator | TheSeven: I think it was 0, but my problem is I can only see a few lines on the screen |
16:59:25 | andiator | and the NAND is readable |
17:00 |
17:00:04 | pamaury | andiator: maybe you could try, as an ugly hack, to make nand_write_sectors print a warning to logf but return 0, see if that helps debugging |
17:00:19 | TheSeven | I'd probably try to add another panic here for debugging: http://git.rockbox.org/?p=rockbox.git;a=blob;f=firmware/common/disk_cache.c;h=0e842e77963bc5895732427684759e9d4fa6e678;hb=HEAD#l235 |
17:00:31 | * | pamaury does not remember if he still has the D2 |
17:01:09 | TheSeven | in an if (cache_entry[index].volume == whatever) of course |
17:01:20 | TheSeven | to catch the source of any attempt to dirty nand cache |
17:01:37 | andiator | OK, I will try this |
17:02:04 | andiator | my idea was to do alot of logf to a file, to see which path is taken |
17:02:10 | TheSeven | not sure which volume ID nand is |
17:02:29 | TheSeven | it will 99% certainly go through there, and point you to the culprit in the backtrace |
17:02:52 | TheSeven | writing to a file might not be the best tool if you're debugging storage-related crashes |
17:11:26 | | Join freenoodle [0] (~jr@p5B058186.dip0.t-ipconnect.de) |
17:12:05 | | Quit duo8 (Quit: ZNC 1.6.3 - http://znc.in) |
17:14:27 | | Quit T-rack (Quit: Page closed) |
17:17:46 | | Quit freenoodle (Ping timeout: 260 seconds) |
17:23:50 | | Quit kuba-orlik_ (Remote host closed the connection) |
17:26:32 | | Quit ender` (Ping timeout: 240 seconds) |
17:32:38 | | Join JdGordon_ [0] (~jonno@rockbox/developer/JdGordon) |
17:32:41 | andiator | pamaury: with returning a 0 there is no panic |
17:33:07 | andiator | and the RB seems to working normally |
17:33:24 | pamaury | andiator: yeah but the point is to find why rockbox tries to write to read-only storage |
17:33:31 | pamaury | I suggest you try what TheSeven suggested |
17:35:23 | | Quit JdGordon (Ping timeout: 246 seconds) |
17:35:51 | | Quit krnlyng (Ping timeout: 260 seconds) |
17:40:00 | | Join dfkt [0] (~dfkt@unaffiliated/dfkt) |
17:48:45 | | Join krnlyng [0] (~liar@77.116.78.87.wireless.dyn.drei.com) |
17:54:43 | | Join freenoodle [0] (~jr@p5B058186.dip0.t-ipconnect.de) |
17:57:02 | | Quit freenoodle (Client Quit) |
17:57:05 | | Join freenoodle1 [0] (~jr@p5B058186.dip0.t-ipconnect.de) |
17:58:21 | andiator | Trying to create a DB, these logf are made by me, so the line numbers are a little bit off and I had to type them anyway, so here it goes.. |
17:58:29 | andiator | *PANIC* write to nand: drive = 0 DC:2006832C sp:200D1D88 A: 2005C068 A: 2005CE0C bt end |
17:58:36 | andiator | start of logf data |
17:58:42 | andiator | ata-nand-telechips.c:865 nand_write_sectors(): drive = 0 |
17:58:51 | andiator | storage.c:109 storage_write_sectors(): write to NAND |
17:59:00 | andiator | disk_cache.c:209 dc_cache_probe() |
17:59:07 | andiator | storage.c:103 storage_write_sectors() write to SD start = 42742... copies = 1 |
17:59:08 | | Quit dan- (Read error: Connection reset by peer) |
17:59:16 | andiator | disk_cache.c:271 dc_commit_all(): volume = 0 |
17:59:23 | andiator | storage.c:103 storage_write_sectors() write to SD start = 8193, copies =1 |
17:59:31 | andiator | disk_cache.c:271 dc_commit_all(): volume = 0 |
17:59:38 | andiator | storage.c:103 storage_write_sectors() write to SD start = 42742... copies = 1 |
17:59:44 | andiator | disk_cache.c:271 dc_commit_all(): volume = 0 |
17:59:51 | andiator | fat.c:305 cache_commit() |
17:59:58 | andiator | fat.c:2037 fat_create_file() |
18:00 |
18:00:05 | andiator | far_create_file("database_tmp.tcd", 200D285C, 200D28AC... |
18:00:15 | andiator | Codec: entering run state |
18:00:21 | andiator | Codec: calling entrypoint |
18:00:28 | andiator | storage.c:103 storage_write_sectors() write to SD start = 8193, copies =1 |
18:00:35 | andiator | disk_cache.c:271 dc_commit_all(): volume = 0 |
18:00:42 | andiator | storage.c:103 storage_write_sectors() write to SD start = 42742 |
18:00:44 | | Quit freenoodle1 (Client Quit) |
18:00:45 | | Join freenoodle [0] (~jr@p5B058186.dip0.t-ipconnect.de) |
18:00:53 | | Join ZincAlloy [0] (~Adium@pD9FB78C2.dip0.t-ipconnect.de) |
18:01:06 | andiator | ... and there is no more data to see |
18:11:42 | pamaury | andiator: did you try TheSeven idea of putting panic in dc_dirty_buf ? |
18:12:29 | | Quit freenoodle (Ping timeout: 246 seconds) |
18:12:36 | pamaury | the backtrace of nand_write seems useless because the error is differed, some buffers is marked as dirty and later written to disk it appears |
18:12:43 | * | pamaury needs to go |
18:12:45 | | Join dan- [0] (~d@101.165.168.135) |
18:12:45 | | Quit dan- (Changing host) |
18:12:45 | | Join dan- [0] (~d@unaffiliated/danneh-/x-7505085) |
18:15:50 | | Join duo8 [0] (~ZNC-SRV-H@117.6.50.149) |
18:16:08 | | Join PurlingNayuki [0] (~Thunderbi@v163-44-154-238.a00f.g.sin1.static.cnode.io) |
18:17:02 | | Quit pamaury (Ping timeout: 240 seconds) |
18:17:22 | | Join nlogex [0] (~filip@dhcp-108-168-15-53.cable.user.start.ca) |
18:17:56 | *** | Saving seen data "./dancer.seen" |
18:19:47 | andiator | pamaury: I will try these too, but it seems it is always the same volume −−> 0 |
18:24:22 | | Join MasterplasterPL [0] (~WIN@188.146.80.167.nat.umts.dynamic.t-mobile.pl) |
18:47:40 | | Join freenoodle [0] (~jr@p5B058186.dip0.t-ipconnect.de) |
19:00 |
19:04:06 | | Quit freenoodle (Ping timeout: 265 seconds) |
19:27:58 | | Quit MasterplasterPL (Ping timeout: 276 seconds) |
19:46:09 | | Join ender` [0] (krneki@foo.eternallybored.org) |
19:46:41 | | Quit nlogex (Quit: WeeChat 1.5) |
19:48:37 | fs-bluebot | Build Server message: New build round started. Revision 9a4cd2e, 255 builds, 11 clients. |
19:51:27 | | Quit ZincAlloy (Quit: Leaving.) |
20:00 |
20:13:26 | | Part Neffi |
20:18:00 | *** | Saving seen data "./dancer.seen" |
20:53:14 | | Join Petri152 [0] (~Petri152@petritrebs.ca) |
21:00 |
21:04:46 | | Nick suYin is now known as suYin`OFF (mysuyin@server2.shellfire.net) |
21:16:20 | | Join naleo [0] (~naleo@unaffiliated/naleo) |
22:00 |
22:04:08 | | Join Grohl [0] (~textual@pool-108-52-16-36.phlapa.fios.verizon.net) |
22:04:48 | Grohl | Having some troubles and am trying to reformat my iPod Classic. For some reason, it hangs on the "reformatting data partition..." screen. |
22:07:24 | | Join pamaury [0] (~pamaury@rockbox/developer/pamaury) |
22:11:44 | | Join petur [0] (~petur@rockbox/developer/petur) |
22:18:03 | *** | No seen item changed, no save performed. |
22:39:18 | | Join nlogex [0] (~filip@dhcp-108-168-15-53.cable.user.start.ca) |
22:41:17 | | Nick suYin`OFF is now known as suYin (mysuyin@server2.shellfire.net) |
22:44:47 | | Quit nlogex (Quit: WeeChat 1.5) |
22:54:42 | | Quit paulk-collins_ (Quit: Leaving) |
22:58:43 | | Join ZincAlloy [0] (~Adium@pD9FB78C2.dip0.t-ipconnect.de) |
23:00 |
23:04:31 | | Nick suYin is now known as suYin`OFF (mysuyin@server2.shellfire.net) |
23:06:04 | | Join saratoga [0] (123e11e0@gateway/web/freenode/ip.18.62.17.224) |
23:06:44 | saratoga | i'm pretty sure GUI boost screws up test_codec since the system will be boosted if any buttons are pressed |
23:06:49 | saratoga | is there some way to force it to not boost? |
23:07:13 | | Quit petur (Quit: Leaving) |
23:07:15 | saratoga | i can do cpu_boost(false) at the start, but it'll still boost if a button is pressed |
23:10:39 | | Quit pamaury (Read error: Connection reset by peer) |
23:11:54 | | Join pamaury [0] (~pamaury@rockbox/developer/pamaury) |
23:12:42 | fs-bluebot | Build Server message: New build round started. Revision ee7dfb5, 255 builds, 11 clients. |
23:18:29 | saratoga | do we have code anywhere to scroll text up as new lines are written (like a command line)? |
23:25:01 | fs-bluebot | Build Server message: Build round completed after 738 seconds. |
23:25:05 | fs-bluebot | Build Server message: Revision ee7dfb5 result: 70 errors 0 warnings |
23:43:17 | | Quit pamaury (Read error: Connection reset by peer) |
23:44:32 | | Join pamaury [0] (~pamaury@rockbox/developer/pamaury) |
23:45:16 | | Quit pamaury (Read error: Connection reset by peer) |
23:46:31 | | Join pamaury [0] (~pamaury@rockbox/developer/pamaury) |
23:53:43 | | Quit Elfish_ (Ping timeout: 246 seconds) |
23:56:38 | | Join Elfish [0] (amba@2001:1608:12:1:13:3:3:7) |
23:57:44 | | Join JdGordon [0] (~jonno@rockbox/developer/JdGordon) |
23:58:46 | | Quit pamaury (Read error: Connection reset by peer) |