[00:08:03] * user890104 updates the TNT toolchain on the server [00:10:29] *** Quits: fmibot (~ircbot@freemyipod.org) (Remote host closed the connection) [00:13:52] ah.... OOM killer [00:13:59] gcc: internal compiler error: Killed (program cc1) [00:14:21] which also killed the bot :) [00:23:08] *** Joins: fmibot (~ircbot@freemyipod.org) [00:23:08] *** ChanServ sets mode: +o fmibot [00:45:44] ...which means that the bot consumes significant amounts of memory [00:52:48] no, the kernel got crazy and killed most of the processes (including nginx, mysqld and the bot) [00:53:05] that's because cc1 consumed way too much memory [01:02:00] it should actually kill cc1 in that case [01:02:30] the OOM killer should typically pick the biggest process first [01:05:18] * TheSeven is puzzled [01:05:29] fixing up the fifo definitions solves one of two problems [01:05:56] now little transfers work, but large transfers which end with a short packet never actually send that packet... [01:19:04] user890104: can you try downloading 0xfff00 bytes from 0x08000000? [01:19:16] er, 0xfff00 bytes [01:19:48] I'm getting this: Bulk data could not be fully received (16776704 of 16776960)! [02:00:42] apparently the USB core inserts a zero byte packet before the last packet if the last packet isn't full [02:00:54] but only after some specific number of packets have been sent [02:15:39] minimum size required to trigger it seems to be 0x200201 bytes [02:15:53] I'm wondering if this is somehow related to my USB bus topology [02:18:48] seems so... doesn't happen on another host [02:28:22] on that other host the minimum size to trigger some other issue is 0xff7f29... what a weird size! [02:29:11] so I guess I'll just tell libemcore to split everything into 8MB chinks [02:29:14] chunks* [03:09:50] *** Quits: Kebianizao (~kvirc@99.1.219.87.dynamic.jazztel.es) (Quit: Estaba usando KVIrc KVIrc Equilibrium 4.2.0, revision: 420, sources date: 20120701, built on: 2013-08-29 10:52:41 UTC 420 http://www.kvirc.net/) [04:11:22] *** Quits: bcobco (~bcobco@77.225.204.119) () [06:15:47] *** Quits: TheSeven (~quassel@rockbox/developer/TheSeven) (Ping timeout: 250 seconds) [06:17:23] *** Joins: TheSeven (~quassel@rockbox/developer/TheSeven) [06:38:27] *** Joins: steffengy (~quassel@p5088F391.dip0.t-ipconnect.de) [06:41:21] *** Quits: steffengy1 (~quassel@p57B4879B.dip0.t-ipconnect.de) (Ping timeout: 250 seconds) [09:06:42] *** Joins: shdwprince (~sp@130.185.3.151) [09:06:58] Hi there again. [09:07:52] Oh, I forgot that I should go into -support channel. [09:37:21] *** Quits: shdwprince (~sp@130.185.3.151) (Quit: WeeChat 0.4.3) [10:40:40] *** Joins: shdwprince (~sp@130.185.3.151) [10:54:18] *** Quits: shdwprince (~sp@130.185.3.151) (Quit: WeeChat 0.4.3) [11:01:40] *** Joins: shdwprince (~sp@130.185.3.151) [11:10:02] Trying to apply patch for USB, but I get "No rule to make target: .../lcd-color-common.c". Fetched patch from gerrit by copypasting command line. Where I am failed/ [11:10:04] *? [11:45:04] *** Quits: shdwprince (~sp@130.185.3.151) (Ping timeout: 260 seconds) [11:46:48] New commit by theseven (r964): emCORE/UMSboot: Fix USB concurrent IN transfer problems on S5L8702 and S5L8720 http://is.gd/X3cBXb [11:51:47] TheSeven: do you remember where did you get the "two-side bitten apple" image from? (n2g OF in boot menu) [11:52:03] i guess we'll need a bigger resolution of it [11:52:22] if we have it, it's in the tree that I sent you [11:53:06] I think it's a photo of a real apple used as the shape, and then some gradient fill + glow [11:53:22] ah, i see [11:53:37] New commit by theseven (r965): emCORE: iPod Classic: Fix several ATA driver bugs, which were causing compatibility issues with HDD/Flash mods http://is.gd/JT0Go8 [11:54:37] does OF still crash at some point after your most recent changes? [11:54:39] /archive/iLoader/iLoader Themes/Original Hires/appleos.png" [11:54:47] not that I'm aware of [11:55:05] <[Saint]> iLoader. [11:55:08] the crash that I was chasing for the last few weeks turned out to be simply a firmware version incompatible with my hardware [11:55:09] <[Saint]> Wow...memories. [11:56:00] <[Saint]> Also, regarding the "4G" Classic, yeah, that is known. [11:56:04] <[Saint]> Not well known, but known. [11:56:06] /archive/iLoader/iLoader/Assets/apfel.psd" [11:56:26] I'm wondering if there's any real difference between 3.1g and 4g [11:56:48] besides the version number, any maybe a new model number for legal reasons [11:56:49] does booting OF mean that it would recognise our hdd layout? [11:57:09] no, you need to restore the OF's layout [11:57:21] and then restore again to run rockbox? [11:57:54] I haven't tried rockbox, but at least emcore seemed to somehow manage to mount apple's layout [11:58:04] even though I still don't get how [11:59:25] so let me quickly summarize our TODOs: [11:59:26] if i restore the OF using ipodscsi, then install emcore, it doesn't repartition the hdd [11:59:45] it should [12:00:04] ipodscsi will definitely repartition it [12:00:10] or did you mean our installer doesn't mess with it? [12:00:18] yes, but emcore accepted the partitioning [12:00:27] the installer leaves it alone if it's doing an update and was able to mount whatever was there [12:00:37] it was able to mount it [12:00:59] and just added the theme [12:01:16] iPod_Control, Notes, Videos and the other apple-stuff were there [12:01:57] i thought that it's not a bug but a feature :) [12:04:25] TODOs: [12:04:25] 1. decide how rockbox bootloader installation should work [12:04:25] 2. implement/fix the clickwheel driver in the rockbox bootloader [12:04:25] 3. port the dualboot code to the rockbox bootloader [12:04:26] 4. fix the LCD issue in the rockbox bootloader [12:04:28] 5. map button combinations sanely in the rockbox bootloader [12:04:31] 6. implement the bootloader flasher [12:04:33] 7. implement the PC-side DFU launcher (rbutil integration?) [12:04:35] 8. fix emCORE USB console [12:04:37] 9. implement dualboot in emCORE boot menu / installer [12:04:39] 10. implement apple disk mode in emCORE boot menu / installer [12:04:42] 11. implement SMART / ATA passthrough in emCORE disk mode [12:04:44] 12. port that to rockbox ;) [12:04:46] 13. iron out the last few issues with the designware driver [12:04:48] 14. remove those bitfields from it and make it compliant with rockbox's style guidelines [12:04:50] 15. actually get that thing committed [12:04:53] ...did I forget anything? [12:04:57] 16. get the rockbox guys to accept my other ipod classic fixes ;) [12:06:10] well, you used to have commit access, right? [12:07:11] talking to the right guy should grant you access again, and make things happen faster :) [12:07:18] I did, but don't have it anymore. I'm also not going to have much to commit anyway, and most of that could be controversial, so it's better to let it go through the review process anyway [12:07:51] <[Saint]> Even if he *could* commit directly....things like this should *always* go via gerrit. [12:07:59] <[Saint]> Always. [12:08:13] [Saint]: well, maybe not the ones that I was referring to in 16. [12:08:22] like that charging state detection two-liner fix ;) [12:08:38] which is trivial, and where the old code was just outright wrong [12:08:42] <[Saint]> well, yeah. I was referring more to the entire list. [12:09:09] so, yes, it would mostly help with the rather unimportant little nitpicks [12:09:19] <[Saint]> But, in saying that, passing trivial changes through gerrit should still happen anyway. [12:09:39] <[Saint]> Even if they just sit there for 24~48h before being puched directly. [12:09:53] <[Saint]> *pushed [12:10:28] while we're at it, can you rewiew those patches? ;) [12:10:52] do you happen to have a firewire ipod cable somewhere? [12:11:08] I can't verify if this is actually fixed now [12:13:00] TheSeven: the "iLoader apple" is not the one i'm looking for. it's this one: http://svn.freemyipod.org/apps/bootmenu-ipodnano2g/images/crapple.png [12:14:50] TheSeven: what about adding diagmode to the boot menu? [12:14:52] oh right, we were using a colorful one [12:15:14] but shouldn't this one have sufficient resolution? [12:15:44] i'll try to use it anyway (so starting work on 9. and 10.) [12:15:45] hm, can we squeeze all of that in? emcore, bootmenu with all those additional graphics, RB fallback image, disk mode, diag mode? [12:16:01] do we need rockbox fallback if we have OF disk mode? [12:16:27] hm, I consider it more important than the (mostly useless) diagmode [12:16:44] especially once we have proper SMART I don't see a single sane use case for diagmode anymore [12:17:30] on the nano it contained some useful stuff, but the classic diagmore is horrible. and if a developer needs it, he can just runfirmware it or push it into umsboot [12:17:48] also watch out that different ipod generations have differently sized disk/diagmodes [12:18:29] can we use a better compression then? [12:18:52] I don't think you can get much better than UPX [12:19:05] maybe LZMA, but that's MUCH slower and uses awful lots of RAM [12:19:06] unless we're completely out of space, i'd vote for keeping these additional firmware images [12:19:12] <[Saint]> Firewire? [12:19:18] <[Saint]> The fuck? [12:19:33] <[Saint]> I thought that died out with the Color/Photo/4G [12:19:53] <[Saint]> (and, yeah, I do have a firewire cable around...somewhere) [12:19:58] the classic can still charge via firewire, and someone complained that this didn't work recently [12:20:12] <[Saint]> I'd say fuck 'em. [12:20:16] and it should be easy to fix [12:20:18] <[Saint]> It still can't sync. [12:20:25] <[Saint]> So...what's the point? [12:20:38] it was mostly "eh, I dunno what this is, so I'm not allowing more than 100mA" [12:20:43] <[Saint]> Firewire is dead. Let it die a noble death. :) [12:21:11] <[Saint]> It might also cause some confusion. [12:21:20] <[Saint]> As no other iPod targets support it. [12:21:31] really? [12:21:36] <[Saint]> Really really. [12:21:42] how do you know? [12:21:52] <[Saint]> I mean in context of Rockbox. [12:22:09] on the classic it would "magically work" if rockbox wasn't trying to be clever about USB host detection [12:22:23] * user890104 wonders where he can get a nice shiny decrypted working 2G-classic OF image [12:22:45] grab the IPSW, extract it [12:22:53] <[Saint]> ^ [12:23:05] then locate the image from that index table [12:23:23] check the payload size in the 8702 header [12:25:07] then grab that many bytes from the beginning of the encrypted payload [12:25:16] upload that to the ipod and aesdecrypt it with key 1 [12:25:42] i've used extract2g and have a file that begins with 87021.0 [12:26:13] then a mostly-NULL 2k area [12:26:18] in that case check the payload size (at 0xc IIRC), and grab that many bytes from offset 0x800 [12:26:24] * [Saint] had a differently worded variant of that saved squirreled away but TheSeven made a damn fine job out of explaining it and enabling me to not look for it [12:28:40] 0xA06634 = 10511924 [12:29:39] seems reasonable [12:29:57] you'll have to round that up to the next 16 byte boundary for decryption [12:30:21] (but can throw away that padding afterwards) [12:30:46] will ipodcrypt do this for me? [12:30:55] possibly [12:30:59] I did it manually yesterday [12:31:54] i guess s5l8702-decryptnor is what i need [12:32:40] ah, it even skips that 2k header [12:33:02] it doesn't know about the signature at the end though [12:33:09] so you'll at least get some trailing garbage [12:33:47] i hope that apple's code won't leak the program counter past the end of their code [12:38:27] FS mount error: 0x0800000A [12:47:11] user890104: generated how? [12:47:23] ran the decrypted binary [12:48:24] it was unhappy then [12:48:39] apple typically corrupts the FS and reboots if it failed to mount it somehow [12:49:02] lol [12:49:11] nice error handling, apple! [12:50:59] "we found an unknown filesystem. let's corrupt it!" [12:55:05] yes, to make windows/itunes format it when it reconnects in disk mode ;) [12:55:37] just like the nano2g keeps wiping its FTL for most kinds of corruption [12:55:55] however I've seen nano2gs softbrick themselves by failing to do so :P [12:57:14] <[Saint]> Once upon a time: [12:57:26] <[Saint]> hard reset Rockbox? [12:57:30] <[Saint]> Bam. Trashed FTL. [12:57:51] <[Saint]> Every. Fucking. Time. [12:58:01] I guess that still happens occasionally if you don't have emcore [12:58:18] apple's code is just unable to fix up some possible situations after unclean shutdowns [13:03:03] how bad idea would be to runfirmware aupd? [13:03:31] will probably crash somehow [13:05:45] at least it shows an apple logo [13:06:22] I think I remember that on nano2g we couldn't just run AUPD to restore, and instead had to parse the payloads out of it and do the restore ourselves [13:07:15] it didn't touch emcore at all [13:07:32] did it show a status bar? [13:07:38] no [13:07:48] ok so it didn't even try to flash something [13:08:05] that's what it looks like [13:08:31] ok, back to osos. i'll repartition using ipodscsi first this time [13:08:36] what are you trying to do with aupd? [13:08:51] you can typically just use the recovery dfu image instead of it [13:08:52] restore norboot [13:09:08] just trying to see if it works at all [13:09:31] uhm, how do i extract disk mode? [13:09:47] from where? [13:09:54] from OSOS i guess [13:09:59] nah [13:10:19] it's in the bootflash, so it's in aupd and in the recovery dfu image [13:10:21] i can't find it in the firmware's directory table [13:10:33] it's easiest to grab it from a factory bootflash image [13:10:51] it's just aupd, osos, rsrc and hash (which is just 0xFFs) [13:10:56] yes [13:11:11] disk mode isn't on the disk ;) [13:11:31] i guess diskmode from different firmware version would also work? [13:11:46] it should, as long as it's the same hardware generation [13:12:44] ok then, what should i look for? [13:13:11] what are you trying to do? [13:13:26] boot disk mode, then repartition with ipodscsi [13:13:36] final goal: tethered boot of osos [13:13:36] which HW generation? [13:13:42] 2G [13:14:05] i found only 1g and 3g in your archive [13:14:43] hslfksid [13:14:47] ah, here it is [13:15:04] what's the format? 8 bytes label, then? [13:15:56] 000FFE00 68 73 6C 66 6B 73 69 64 00 00 00 00 30 B3 0C 00 hslfksid 0³ [13:15:56] 000FFE10 C4 48 03 00 00 00 00 08 00 00 00 00 00 00 00 00 ÄH [13:15:56] 000FFE20 00 50 01 00 24 BA 11 08 P $º [13:17:27] 0xCB330 looks like an offset, and 0x348C4 looks like a size [13:18:22] you could try running an 1g disk mode on it [13:18:35] ok [13:18:52] it might have LCD problems, but should work USB-wise [13:19:05] it might even fully work depending on your LCD model [13:19:24] and yes, those numbers seem correct [13:19:39] no lcd issues so far [13:20:33] on my 3g it comes up with a white screen [13:20:52] (it just has no driver for that one) [13:21:19] and, unlike the OF, handles that properly instead of crashing after calling a three times dereferenced null pointer ;) [13:21:42] (3.0g OSOS on 3.1g hardware) [13:22:44] can you integrate the linux version of ipodscsi into ours? [13:23:12] does it make sense? [13:23:19] how different are they? [13:23:40] i.e. is the ifdef hell worth it, or should we better keep it separate? [13:24:00] it uses ioctl instead of the windows api [13:24:40] we can make two implementations of the device access code, and only compile the one we need for that platform [13:24:53] while still keeping the main logic apart [13:27:49] yay! apple firmware :) [13:29:05] seems to crash on usb unplug [13:29:10] or even sooner [13:29:33] uh? [13:30:20] it crashed if booted with usb plugged in [13:30:34] hm, weird [13:31:15] never seen it crash after starting to draw things on the LCD [13:31:25] maybe just the UI thread, because usb replugs work [13:31:35] but the Connected icon doesn't go away [13:32:22] er... [13:32:29] it's capable of sending its usb descriptors to the host, but windows cannot find any volumes [13:32:33] I guess you didn't properly restore the database? [13:32:41] it doesn't show up in disk manager at all [13:32:58] i just repartitioned, but didn't format the partition [13:33:16] hm, not sure what it would do in that situation [13:33:31] actually I'm puzzled that it doesn't reboot into disk mode [13:33:44] uhm, there's no disk mode to boot into :P [13:33:53] well it would end up in emcore [13:34:14] even the lock switch doesn't work in this condition [13:34:25] but i can hear clicks when i press a button [13:34:33] I guess something hangs while trying to mount the FS and read the database [13:36:04] what fat unit size should i choose? [13:37:19] all valid ones should work [13:37:35] for music files going for the maximum is probably most sane [13:37:42] so probably 32K [13:39:24] how do i initialize the database sans-itunes? [13:40:31] you can't [13:40:51] maybe with some 3rd party ipod manager tools, but most of those can't build a new database, just edit an existing one [13:41:13] and apple signs the database with a device specific key to avoid copying it to another ipod [13:41:26] i can cheat and restore the files from my oldest backup (before any emcore-related experiments) [13:41:41] yes, that's what I did for my tests [13:42:05] actually I just restored the factory HDD image (which I captured before even booting that ipod for the first time) [13:42:48] is disk mode expected to reboot after unplugging like on n2g, or it needs some help (n3g/n4g)? [13:43:56] 335 MB used / 111 GB free [13:43:59] that's better [13:44:45] ... unless you connect an USB cable [13:45:13] then it has the same behaviour that i explained earlier [13:47:58] looks like i'll need to do a DFU restore [14:02:41] TheSeven: i got my ipod's hdd to a state which both emcore disk mode recognise [14:03:11] you need to repartition and reinstal using ipodscsi (not itunes), then format using guiformat.exe [14:03:25] can you try to reproduce it on your device? [14:04:17] then you can compare the partition/fs layout, so we can make a OF-compatible repartitioning/reformatting [14:07:19] I don't see how ipodscsi vs. itunes would make a difference there [14:07:36] and I'll try to get away without any repartitioning at all [14:08:57] it's probably formatting the partition in some way we don't like [14:09:44] and yes, installing rockbox without formatting the hdd would be the best option [14:10:04] where are we going to store OF? as a file on the main partition? [14:11:40] <[Saint]> That's how the other iPods do it, no? [14:12:07] <[Saint]> Oh, my mistake, only with the fancy ipodpatcher install methods. [14:12:35] <[Saint]> seems we'll look for apple_os.bin [14:13:05] or leave the firmware partition, extract the image from it then jump inside [14:13:34] i prefer this way, because the user cannot delete the original firmware by reformatting the ipod [14:13:39] RB bootloader does the latter [14:13:49] emcore does the former [14:14:21] on the classic we will probably have to keep the firmware partition intact anyway [14:14:37] can we also do this on n2g? [14:14:45] rockbox does it there already [14:15:00] and I'm not going to rework emcore on the 2g [14:15:39] <[Saint]> For the one guy that uses it. [14:15:49] <[Saint]> (probably user890104 ;)) [14:15:51] that thing has next to zero user base, and that little user base does typically care about flash space [14:15:51] well, i can do it if you don't mind [14:16:21] so wasting 100MB for no reason doesn't seem to make much sense [14:16:24] <[Saint]> Yeah, on the n2g, its best to not shit things into the already limited storage if possible. [14:16:31] on the classic space is less of an issue [14:16:39] <[Saint]> Much so. [14:16:43] well, apple does exactly that [14:16:44] although I don't get why apple has to waste 300MB there [14:16:49] for a 10MB firmware image [14:16:59] and maybe 50MB of other crap [14:17:08] <[Saint]> see also: Samsungs multi-GB /cache partitons [14:17:34] <[Saint]> 4GB for a /cache partition on your 32GB device? Sure...why the fuck not. [14:17:53] <[Saint]> We'll...maybe....use it. [14:18:16] on android the cache partition should usually be roughly the same size as the system partition [14:18:37] <[Saint]> I _think_ it does this so that, potentially, a full system image can be dumped and extracted there as an OTA. [14:18:40] otherwise you might get into trouble with major release OTA updates [14:18:58] yes, OTA updates are downloaded to that partition [14:19:33] (I see no good reason why this couldn't be on /data on modern devices though) [14:20:12] <[Saint]> On modern modern devices, the partition sizes are fairly arbitrary and can be resized dynamically by my understanding. [14:20:51] not for typical end users [14:21:05] <[Saint]> Right, indeed not. [14:21:07] you usually can't resize them without wiping all user data [14:21:11] so not appropriate for OTA [14:21:21] on my SGN3 (32GB) the cache partition is actually suspiciously small [14:21:42] system is 2.2GB (2.0GB used), cache is 300MB [14:22:09] TheSeven: can you also add a check for menu+select held down just before powering on the HDD, that goes into an inifnite loop until they are released? [14:22:10] <[Saint]> On my Galaxy Ace 3, its something like 128MB [14:22:17] leaving a total of 26.4GB for user data [14:23:08] another item for the todo list... [14:30:34] http://www.freemyipod.org/wiki/Todo_list#1G.2F2G.2F3G_Classic [14:31:09] <[Saint]> Love the last 2 for the wiki [14:33:21] Freemyipod is a project for reverse-engineering iPods with clickwheel (no iOS devices) [14:33:23] well ... [14:33:43] this is half-wrong :) [14:34:11] well I think there's exactly one person that cares about the ipt2g :P [14:34:22] [Saint]: did you manage boot OF on any of your classics, from within emcore? [14:34:35] <[Saint]> Freemyipod is a project for reverse-engineering iPods with clickwheel and some iOS devices, but, don;t get too excited end users...you'll be doing absolutely nothing useful with it. [14:35:22] (until i make my HIDAccelGamepad app) [14:35:55] <[Saint]> Yeah, uses'll be all up in that shizzle, yo. ;) [14:36:03] <[Saint]> *users [14:36:35] well, it would be slightly more useful than it is at the moment [14:37:14] so "absolutely nothing useful" won't be true anymore [14:37:18] <[Saint]> I would guess that Android dominates the (small) market of "turning things that aren't controllers into controllers" [14:40:38] well, after creating my first HID host and device software, i liked that idea [14:41:04] (it's an usb thermometer btw) [14:42:13] <[Saint]> Entirely unrelated... [14:42:27] <[Saint]> Why the flying fuck does notify-send not work as I expected with ssh? [14:43:01] * [Saint] wanted to log in and craft amusing fake system errors with notify-send...but *noooooooo*, apparently, that's not a thing. [14:50:42] *** Quits: [Saint] (~saint@rockbox/staff/saint) (Quit: going down for system upgrade) [14:55:49] *** Joins: [Saint] (~saint@rockbox/staff/saint) [22:53:07] grr, seems like UMSboot is broken [22:53:30] now the question is whether one of yesterday's commits is the culprit, or if it was broken even before that [23:07:58] are you talking about the case where it locks up right after being launched? [23:29:47] not quite sure what it's doing, but it's showing its splash screen and not mounting [23:29:51] didn't look into the details yet