**** BEGIN LOGGING AT Wed Jul 04 02:59:58 2012
Jul 04 07:53:25 <CIA-69> freesmartphone.org: 03morphis 07utilities * r5f668c67b034 10/android/image-utils/configure.ac: android: image-utils: rename to android-image-utils
Jul 04 07:53:25 <CIA-69> freesmartphone.org: 03morphis 07utilities * r6ddca47ec749 10/android/image-utils/ (5 files in 4 dirs): android: image-utils: get packaging bits right so we're ready for distribution
Jul 04 08:04:06 <CIA-69> freesmartphone.org: 03morphis 07utilities * r422062e9ff61 10/android/image-utils/ (ChangeLog NEWS README): android: image-utils: update documentation bits
Jul 04 09:53:19 <CIA-69> SHR: 03shr-devel 07buildhistory * ref7ba47904e2 10/packages/i686-nativesdk-oesdk-linux/ (352 files in 352 dirs): packages: Build 201207040848 of shr 20120704 for machine om-gta02 on opmbuild
Jul 04 12:17:19 <Alex[sp3dev]> paulk-desktop: hi
Jul 04 12:17:28 <paulk-desktop> hi Alex[sp3dev]!
Jul 04 12:18:30 <Alex[sp3dev]> looks like I've trashed my nvdata and have no backup ^_^. Not that I need it (and I can get it from ril log dumps), just unsure that I can test nvdata loading 
Jul 04 12:18:55 <paulk-desktop> ok
Jul 04 12:19:05 <Alex[sp3dev]> paulk-desktop: so, about ril socket
Jul 04 12:19:05 <paulk-desktop> I'll try to solve that odd issue I had yesterday
Jul 04 12:19:54 <Alex[sp3dev]> paulk-desktop: now I get permission denied. When I patched rild to run as root, it created server socket. I did a chmod 777 on /dev/socket and mediaserver connected. But I want to know of a proper way
Jul 04 12:20:42 <paulk-desktop> Alex[sp3dev], the way samsung always used is to run rild as root
Jul 04 12:21:01 <paulk-desktop> because for other phones and GB galaxys2, transport is over a net iface
Jul 04 12:21:11 <paulk-desktop> and it didn't work as well with radio user
Jul 04 12:21:27 <paulk-desktop> that's why CyanogenMod uses a "non-free" rild lib for these phones
Jul 04 12:21:34 <paulk-desktop> rild bin*
Jul 04 12:21:39 <Alex[sp3dev]> paulk-desktop: nope, they are not
Jul 04 12:22:03 <Alex[sp3dev]> maybe I need to strace the proprietary blob again?
Jul 04 12:22:14 <paulk-desktop> they are not what?
Jul 04 12:22:15 <Alex[sp3dev]> maybe some capability bits need to be cleared?
Jul 04 12:22:23 <Alex[sp3dev]> paulk-desktop: they are not using the closed rild
Jul 04 12:22:28 <paulk-desktop> not on ICS
Jul 04 12:22:31 <paulk-desktop> but on GB they sure do
Jul 04 12:22:59 <Alex[sp3dev]> yes, but. I have ICS. I replace the closed ril with ours. Socket fails. That pretty much shows we're doing something wrong
Jul 04 12:23:46 <paulk-desktop> or maybe they changed something in ICS? in GB it was possible to create a socket being non-root
Jul 04 12:23:59 <paulk-desktop> maybe they made it more restrictive
Jul 04 12:24:18 <paulk-desktop> I'll check /dev/socket group on GB
Jul 04 12:25:05 <Alex[sp3dev]> paulk-desktop: Also, I've pushed the hacked ril client branch to my fork on gitorious. On nexus, instead of HRilClient, a pointer to the socket fd is passed as the first argument
Jul 04 12:25:30 <Alex[sp3dev]> it more or less 'works' (except for permissions), but still no sound. gotta make more ril traces
Jul 04 12:26:20 <paulk-desktop> ok, good
Jul 04 12:26:35 <paulk-desktop> GB: drwxrwxr-x    2 root     net_raw        260 Jul  3 11:39 socket
Jul 04 12:26:44 <paulk-desktop> ICS: drwxr-xr-x root     root              2012-07-04 14:20 socket
Jul 04 12:26:56 <paulk-desktop> root:net_raw in GB, root:root in ICS
Jul 04 12:27:01 <Alex[sp3dev]> yep
Jul 04 12:28:12 <paulk-desktop> IMO the clean way is to make it back to net_raw in ueventd.rc
Jul 04 12:28:17 <Alex[sp3dev]> maybe we should implement ramdump support to be able to get the log from modem ram?
Jul 04 12:28:43 <paulk-desktop> I never tried it…
Jul 04 12:28:53 <paulk-desktop> what will it show?
Jul 04 12:29:02 <paulk-desktop> modem-side logs?
Jul 04 12:29:22 <Alex[sp3dev]> paulk-desktop: no idea. looks like grindars has it implemented in his c++ fw loader for i9100 so it should be easy to replicate
Jul 04 12:29:28 <paulk-desktop> ok
Jul 04 12:29:57 <Alex[sp3dev]> what happens when you download NVDATA? the modem reboots?
Jul 04 12:30:16 <paulk-desktop> the whole phone reboots
Jul 04 12:30:30 <paulk-desktop> but it's not all the time
Jul 04 12:30:35 <paulk-desktop> sometimes it goes find
Jul 04 12:30:37 <paulk-desktop> fine*
Jul 04 12:30:39 <Alex[sp3dev]> that is very strange. probably a bug in kernel modemctl
Jul 04 12:30:50 <Alex[sp3dev]> it's not even a bug
Jul 04 12:31:04 <paulk-desktop> but it only happens when the nv_data.bin checksum is found to be wrong
Jul 04 12:31:19 <Alex[sp3dev]> iirc on s2, the driver calls BUG_ON on receiving some modem messages. That's quite moronic
Jul 04 12:31:30 <paulk-desktop> :)
Jul 04 12:34:07 <paulk-desktop> ok basically it happens when I do that: (non-free RIL running) -> setprop rild.libpath /system/lib/libsamsung-ril.so -> killall rild -> (free RIL works fine) -> killall rild -> reports bad md5 and crashes the phone
Jul 04 12:37:09 <paulk-desktop> now if I disable RFS messages, it works
Jul 04 12:44:21 <Alex[sp3dev]> huh. looks like I've found an nvdata backup
Jul 04 12:44:54 <paulk-desktop> good
Jul 04 12:46:49 <Alex[sp3dev]> paulk-desktop: http://pastebin.com/zqvE7PPH
Jul 04 12:49:05 <paulk-desktop> why would it read 0 ?
Jul 04 12:49:17 <Alex[sp3dev]> hmm. permissions?)
Jul 04 12:51:07 <paulk-desktop> that's odd, /dev/umts* should always be readable by radio…
Jul 04 12:54:03 <Alex[sp3dev]> [MODEM_IF] umts_ipc0 state changed: CRASH_EXIT
Jul 04 12:54:08 <Alex[sp3dev]> that's in dmesgf
Jul 04 12:54:15 <paulk-desktop> ah
Jul 04 12:54:19 <paulk-desktop> why that?
Jul 04 12:54:23 <Alex[sp3dev]> dunno
Jul 04 12:54:33 <Alex[sp3dev]> need to consult disassembly to figure out what's wrong
Jul 04 12:55:03 <paulk-desktop> ok
Jul 04 12:58:44 <Alex[sp3dev]> paulk-desktop: maybe we should try ignoring rfs errors and returning 0 as read data count to prevent ril from crashing and see if it works?
Jul 04 12:58:54 <paulk-desktop> I think I have a clue
Jul 04 12:58:58 <Alex[sp3dev]> on the other hand, looks like modem crashes
Jul 04 13:01:32 <paulk-desktop> yeah I think I spotted it
Jul 04 13:01:37 <paulk-desktop> it'a bout chunk size
Jul 04 13:01:42 <paulk-desktop> I used NV_DATA_SIZE / 10
Jul 04 13:02:05 <paulk-desktop> but it make sa decimal number
Jul 04 13:02:13 <paulk-desktop> so I wouldn't read it all when doing the md5
Jul 04 13:02:34 <paulk-desktop> while at boot, it checks with a non-decimal chunk size
Jul 04 13:03:45 <Alex[sp3dev]> paulk-desktop: I tried loading nvdata in fwloader without checking md5 at all and it failed. So you think it fails in rfs.c on rfs callback?
Jul 04 13:04:36 <paulk-desktop> well I don't know why the modem fails yet, but at least I see why there is a md5 mismatch
Jul 04 13:04:48 <paulk-desktop> I'll investigate more though
Jul 04 13:15:27 <paulk-desktop> fixed!
Jul 04 13:15:40 <paulk-desktop> that was definitely the buggy chunk size
Jul 04 13:16:18 <Alex[sp3dev]> yes
Jul 04 13:16:29 <Alex[sp3dev]> I changed the chunk size and it seems to work
Jul 04 13:16:52 <Alex[sp3dev]> why read by chunks at all? do you think we may run OOM if we request 2mb?
Jul 04 13:17:13 <Alex[sp3dev]> either way, simple modular arithmetic would fix it
Jul 04 13:17:32 <Alex[sp3dev]> like (SIZE + CHUNK_SIZE - 1) / (CHUNK_SIZE)
Jul 04 13:17:59 <paulk-desktop> true
Jul 04 13:18:08 <paulk-desktop> I think we could read it all at once
Jul 04 13:18:15 <paulk-desktop> anyway, it works now…
Jul 04 13:18:56 <paulk-desktop> I pushed there: http://git.paulk.fr/gitweb/?p=libsamsung-ipc.git;a=shortlog;h=refs/heads/xmm6260
Jul 04 13:19:03 <Alex[sp3dev]> ok
Jul 04 13:19:05 <paulk-desktop> so nvdata/rfs is now fully working
Jul 04 13:19:26 <paulk-desktop> I'll do data next
Jul 04 13:20:14 <Alex[sp3dev]> btw, I've rebuilt cm9 for galaxy nexus with the free ril.. took just some half an hour on an i7 O_o
Jul 04 13:20:48 <Alex[sp3dev]> yay
Jul 04 13:20:56 <Alex[sp3dev]> I got my IMEI back :)
Jul 04 13:21:13 <Alex[sp3dev]> let me try your fix and see if it differs from mine
Jul 04 13:21:45 <paulk-desktop> ok
Jul 04 13:21:53 <paulk-desktop> half an hour!
Jul 04 13:22:09 <Alex[sp3dev]> yeah, that machine is a beast
Jul 04 13:25:01 <Alex[sp3dev]> paulk-desktop: oops. looks like you based your branch on my rfs hack while I rebased and pushed to xmm6260 one. I'll have to cherry pick then
Jul 04 13:25:12 <paulk-desktop> yep
Jul 04 13:25:14 <paulk-desktop> I saw that too
Jul 04 13:25:31 <paulk-desktop> I'll stick to your main dev branch
Jul 04 13:25:47 <Alex[sp3dev]> I will test your patches, add the fallback to radio nvdata on fail and push. And you can make a new branch based on that
Jul 04 13:26:29 <Alex[sp3dev]> So we only need to clean up code style and ready to submit a request to morphis?
Jul 04 13:29:04 <paulk-desktop> ok
Jul 04 13:29:40 <Alex[sp3dev]> I find my logging macros useful but ugly :)
Jul 04 13:30:06 <Alex[sp3dev]> btw, I've changed them to use libsamsung-ipc debug macro so now you can see the firmware loader log in logcat
Jul 04 13:32:01 <Alex[sp3dev]> paulk-desktop: if md5 is corrupted, will the rfs recalculate it and crash?
Jul 04 13:32:26 <paulk-desktop> nice
Jul 04 13:32:35 <paulk-desktop> Alex[sp3dev], I think it tries to restore the backup
Jul 04 13:32:57 <paulk-desktop> and if the backup has a wrong md5, it takes it antwats
Jul 04 13:32:59 <paulk-desktop> anyways*
Jul 04 13:33:16 <Alex[sp3dev]> ok, forget it. we don't need no generic solutions :). We won't load anything except nvdata from efs so I won't be inventing my bikeshed
Jul 04 13:45:45 <Alex[sp3dev]> paulk-desktop: sorry, I'm a bit slow with patching some stuff here. I'll be ready soon
Jul 04 13:45:52 <paulk-desktop> ok
Jul 04 13:45:59 <paulk-desktop> I figured things out for data
Jul 04 13:46:39 <Alex[sp3dev]> works on s2?
Jul 04 13:47:05 <paulk-desktop> will work very soon
Jul 04 13:47:33 <paulk-desktop> finally rebuilding a kernel with -eng and ro.secure=0 made the RIL log a lot more verbose
Jul 04 13:47:38 <Alex[sp3dev]> you're amazing. Now I need to be amazing and hack gps support into ril
Jul 04 13:47:41 <paulk-desktop> so It's easy to figure out now
Jul 04 13:47:46 <paulk-desktop> yep :)
Jul 04 13:48:05 <paulk-desktop> how is audio going also?
Jul 04 13:48:11 <Alex[sp3dev]> not going yet
Jul 04 13:48:24 <paulk-desktop> ok
Jul 04 13:48:56 <Alex[sp3dev]> ok, I thought to add code to load default nvdata if efs reading fails but looks like ril crashes if ipc_file_read returns error. let me look at it....
Jul 04 13:54:21 <paulk-desktop> ah damn, I pushed my /efs backup, though it was simlocked back then
Jul 04 13:55:01 <Alex[sp3dev]> want mine? xD
Jul 04 13:55:29 <paulk-desktop> nah I know how to SIM unlock it from nv_data
Jul 04 13:56:23 <Alex[sp3dev]> ok, I've pushed your changes and my log patch. Let's get to fallback data loading later. It is probably a big fuckup anyway if efs is missing
Jul 04 13:56:34 <paulk-desktop> ok
Jul 04 13:57:40 <Alex[sp3dev]> hmm. is alsa library loaded by audioflinger?
Jul 04 13:57:47 <Alex[sp3dev]> or audiopolicy?
Jul 04 13:58:27 <paulk-desktop> huh, I have no idea
Jul 04 13:58:35 <paulk-desktop> I'll look
Jul 04 13:59:31 <paulk-desktop> ./audioflinger/AudioFlinger.cpp:    rc = hw_get_module_by_class(AUDIO_HARDWARE_MODULE_ID, if_name, mod);
Jul 04 13:59:52 <paulk-desktop> Alex[sp3dev], here is your answer
Jul 04 14:01:57 <Alex[sp3dev]> paulk-desktop: thank you, but I've chmodd'ed the socket manually for now
Jul 04 14:02:03 <Alex[sp3dev]> http://pastebin.com/0UXjez4z 
Jul 04 14:02:14 <Alex[sp3dev]> looks like the client works, but still no sound
Jul 04 14:02:17 <paulk-desktop> ok
Jul 04 14:02:32 <paulk-desktop> the most important thing is when it sends clock_start
Jul 04 14:02:37 <paulk-desktop> or something like that
Jul 04 14:03:38 <Alex[sp3dev]> probably the easiest thing is to compare the ipc trace of a proprietary and our library
Jul 04 14:05:35 <paulk-desktop> indeed
Jul 04 14:05:39 <paulk-desktop> I'll do that
Jul 04 14:05:46 <paulk-desktop> it should be shown in the logs at least
Jul 04 14:07:12 <paulk-desktop> mhh mediaserver sigsegv here
Jul 04 14:08:42 <Alex[sp3dev]> paulk-desktop: you'll have to disassemble something. or at least add debug prints. I think on s2 ril client is different from both nexus s and galaxy nexus
Jul 04 14:12:03 <paulk-desktop> ok call worked
Jul 04 14:12:07 <paulk-desktop> let's read the logs
Jul 04 14:12:11 <Alex[sp3dev]> with no sound?
Jul 04 14:12:35 <Alex[sp3dev]> paulk-desktop: I'm going home. Will be back in some 2 hours or so
Jul 04 14:12:40 <paulk-desktop> ok
Jul 04 14:12:45 <paulk-desktop> sound, with everything non-free
Jul 04 16:15:03 <CIA-69> SHR: 03Martin.Jansa 07shr-chroot * r0deebc124669 10/ (17 files in 11 dirs): bitbake upgrade
Jul 04 16:15:13 <CIA-69> SHR: 03Martin.Jansa 07shr-chroot * r8d2567d56ae0 10/ (304 files in 46 dirs): system upgrade
Jul 04 16:33:36 <CIA-69> SHR: 03Martin.Jansa 07meta-smartphone * r339fe768d084 10/meta-openmoko/recipes-graphics/drm/ (libdrm_2.4.34.bbappend libdrm_2.4.35.bbappend): libdrm: rename bbappend to match oe-core version
Jul 04 16:45:00 <CIA-69> SHR: 03Martin.Jansa 07shr-chroot * r3e97a2384362 10/ (9 files in 5 dirs): bitbake upgrade
Jul 04 17:08:12 <paulk-desktop> Alex[sp3dev], ping
Jul 04 17:08:40 <Alex[sp3dev]> paulk-desktop: pong
Jul 04 17:09:23 <paulk-desktop> Alex[sp3dev], data working
Jul 04 17:09:43 <paulk-desktop> actually it would have worked with no change, it's just because of ICS ril header that changed return format
Jul 04 17:10:13 <Alex[sp3dev]> paulk-desktop: nice. could you make the patch for libsamsung-ipc so that we can refactor and merge it?
Jul 04 17:10:21 <paulk-desktop> sure
Jul 04 17:10:50 <paulk-desktop> most stuff is about samsung-ril though
Jul 04 17:12:01 <Alex[sp3dev]> paulk-desktop: I feel kind of stupid. Installed the proprietary ril and still no sound. Either I missed ril client or audio library is fubar ^_^
Jul 04 17:12:17 <paulk-desktop> huh ^^
Jul 04 17:12:46 <paulk-desktop> it did the same to me a few hours ago. I was sure I didn't modify libsecril-client.so, though re-pushing it made it work
Jul 04 17:13:27 <paulk-desktop> pushed: http://git.paulk.fr/gitweb/?p=samsung-ril.git;a=shortlog;h=refs/heads/xmm6260
Jul 04 17:13:59 <paulk-desktop> now libsamsun-ipc…
Jul 04 17:16:32 <Alex[sp3dev]> wtf. is my phone fubar or what? X_X
Jul 04 17:18:06 <paulk-desktop> last 3 commits: http://git.paulk.fr/gitweb/?p=libsamsung-ipc.git;a=shortlog;h=refs/heads/xmm6260
Jul 04 17:19:05 <Alex[sp3dev]> I've got no money on my beeline sim card. Wonder if MTS one has gprs enabled
Jul 04 17:19:30 <paulk-desktop> ok
Jul 04 17:19:44 <paulk-desktop> if you can't test, don't bother
Jul 04 17:20:26 <Alex[sp3dev]> I need to test either way. But I can do it later
Jul 04 17:20:35 <paulk-desktop> ok
Jul 04 17:20:38 <Alex[sp3dev]> ok, this is no fun
Jul 04 17:20:43 <paulk-desktop> :D
Jul 04 17:20:47 <Alex[sp3dev]> the proprietary libs give me no sound
Jul 04 17:21:16 <paulk-desktop> I'll take a break before refactoring the code
Jul 04 17:21:42 <Alex[sp3dev]> paulk-desktop: I'd prefer to take a break till friday and work on my other projects
Jul 04 17:22:36 <paulk-desktop> oh sure, go ahead
Jul 04 17:22:42 <paulk-desktop> we're done anyways
Jul 04 17:23:27 <Alex[sp3dev]> yes. half a year of moaning and dreaming and 2 weeks of coding. the same old story
Jul 04 17:34:55 <CIA-69> SHR: 03Martin.Jansa 07meta-smartphone * r929cb6c1708b 10/meta-openmoko/recipes-graphics/drm/ (libdrm/glamo.patch libdrm_2.4.35.bbappend): meta-openmoko: libdrm: update glamo.patch to apply on 2.4.35
Jul 04 17:55:32 <GNUtoo-hplaptop> mrmoku, ping
Jul 04 17:57:31 <Alex[sp3dev]> now I have one of the most stupid theories to try
Jul 04 17:57:47 <Alex[sp3dev]> maybe uboot sets up pinmux wrong and speaker gpio is not set ^^
Jul 04 17:58:36 <GNUtoo-hplaptop> maybe that's the patch? wm8753: allow setting DAI mode even while pcm is active
Jul 04 17:59:20 <Alex[sp3dev]> this is official. I am ultra stupid today
Jul 04 18:00:08 <Alex[sp3dev]> it was uboot indeed. maybe I should really make my lazy ass rewrite the pinmux stuff
Jul 04 18:06:19 <Alex[sp3dev]> ok, so sound is working now. only data needs testing and socket permissions need review
Jul 04 18:23:30 <CIA-69> SHR: 03shr-devel 07buildhistory * rdc943e1ee0d6 10/packages/ (833 files in 833 dirs): packages: Build 201207041902 of shr 20120704 for machine om-gta02 on opmbuild
Jul 04 18:30:53 <GNUtoo-hplaptop> hi paulk-desktop 
Jul 04 18:31:28 <GNUtoo-hplaptop> I'll retry but I think libsamsung-ipc fails to compile in SHR
Jul 04 18:31:37 <paulk-desktop> autorev?
Jul 04 18:31:48 <GNUtoo-hplaptop> yes
Jul 04 18:32:22 <GNUtoo-hplaptop> | ipc_util.c:30:25: fatal error: mtd/mtd-abi.h: No such file or directoryutil.c:30:25: fatal error: mtd/mtd-abi.h: No such file or directory
Jul 04 18:32:30 <paulk-desktop> ah
Jul 04 18:32:34 <paulk-desktop> that is not related to a recent change
Jul 04 18:32:41 <GNUtoo-hplaptop> ok
Jul 04 18:32:44 <GNUtoo-hplaptop> let me retry
Jul 04 18:34:18 <GNUtoo-hplaptop> http://pastie.org/private/imkpc7u9b9ytxbqufxiala
Jul 04 18:34:40 <paulk-desktop> ok
Jul 04 18:34:49 <paulk-desktop> I don't know where mtd/mtd-abi.h should be 
Jul 04 18:34:53 <GNUtoo-hplaptop> I guess it's a missing dep
Jul 04 18:35:17 <paulk-desktop> probably, yes
Jul 04 18:35:23 <paulk-desktop> ask morphis, I think he knows best
Jul 04 18:37:55 <GNUtoo-hplaptop> ok
Jul 04 18:38:31 <GNUtoo-hplaptop> maybe you know the aptitude command that find to which package a file belongs to?
Jul 04 18:38:44 <GNUtoo-hplaptop> because I have theses files on trisquel
Jul 04 18:39:31 <paulk-desktop> I'll search the command
Jul 04 18:39:50 <paulk-desktop> apt-file
Jul 04 18:47:35 <GNUtoo-hplaptop> doesn't that command require a database?
Jul 04 18:49:41 <Alex[sp3dev]> paulk-desktop: so sound is an uboot issue. Works fine now with my patched ril client. So, I have only three tasks for friday: refactor libsamsung-ipc, patch uboot and fix socket permissions
Jul 04 18:50:59 <Alex[sp3dev]> btw, I've pushed the gprs patches applied on top of my tree so that we can submit a pull request to morphis when ready
Jul 04 19:02:08 <GNUtoo-hplaptop> ah yes it's the thing that requires a database to be created
Jul 04 19:25:45 <GNUtoo-hplaptop> linux-libc-dev
Jul 04 19:25:47 <GNUtoo-hplaptop> hmmm
Jul 04 19:25:55 <GNUtoo-hplaptop> I'll clean the toolchain
Jul 04 19:25:57 <GNUtoo-hplaptop> and rebuild ti
Jul 04 19:56:52 <nschle85> GNUtoo-hplaptop: hello dennis do you have some seconds for me ? i have a  small C problem
Jul 04 19:57:11 <GNUtoo-hplaptop> nschle85, hi
Jul 04 20:00:57 <nschle85> GNUtoo-hplaptop: http://pastebin.com/HShB0xSi i get a segmantation fault in the last line (sscanf)
Jul 04 20:01:48 <nschle85> GNUtoo-hplaptop: but  what can cause that problem ?
Jul 04 20:02:08 <GNUtoo-hplaptop> let me look
Jul 04 20:02:10 <lindi-> sscanf sounds scary
Jul 04 20:02:34 <GNUtoo-hplaptop> indeed
Jul 04 20:03:34 <GNUtoo-hplaptop> nschle85, what is that code part of?
Jul 04 20:03:42 <GNUtoo-hplaptop> because it's not standalone....
Jul 04 20:03:53 <lindi-> nschle85: that strncpy should really use "50" as the maximum length
Jul 04 20:04:01 <lindi-> nschle85: or you risk corrupting the stack
Jul 04 20:05:23 <nschle85> GNUtoo-hplaptop:  its from here:  http://caruizdiaz.com/?p=294
Jul 04 20:05:40 <GNUtoo-hplaptop> indeed but even with  strncpy he should check the output
Jul 04 20:05:50 <GNUtoo-hplaptop> because if I remember well when it's over the len
Jul 04 20:05:55 <GNUtoo-hplaptop> the \0 doesn't get written
Jul 04 20:06:05 <GNUtoo-hplaptop> and that can causes issues if the string is reused
Jul 04 20:06:22 <GNUtoo-hplaptop> ok let me look
Jul 04 20:06:46 <nschle85> GNUtoo-hplaptop: i made a small makefile
Jul 04 20:08:08 <nschle85> GNUtoo-hplaptop: http://pastebin.com/XYiZs4pJ
Jul 04 20:13:23 <nschle85> GNUtoo-hplaptop: lindi-: here i added a console output of result http://pastebin.com/kJ5Bvknb
Jul 04 20:24:25 <CIA-69> SHR: 03shr-devel 07buildhistory * reee63bb840f1 10/packages/ (843 files in 843 dirs): packages: Build 201207042104 of shr 20120704 for machine nokia900 on opmbuild
Jul 04 20:26:40 <GNUtoo-hplaptop> nschle85, did you replace with strncpy?
Jul 04 20:27:00 <GNUtoo-hplaptop> I should reread the coed maybe
Jul 04 20:27:07 <GNUtoo-hplaptop> I was talking trough irc
Jul 04 20:37:38 <nschle85> JaMa: hello martin,  i have a small question  belonging to  hwclock.sh  and systemd .   i created a patch http://pastebin.com/Zngpu4XG but i have some questions:
Jul 04 20:38:40 <JaMa> yes I remember seeing this patch before
Jul 04 20:39:17 <nschle85> JaMa: 1. should i reuse  hwclock.sh  and depend on it ?  (but its only provided by busybox)    ?
Jul 04 20:40:39 <nschle85> 2. or should the service file contain  all the necessary information,  so it can use any hwclock command provider ? 
Jul 04 20:41:46 <JaMa> no strong opinion about 1) probably wont harm as it's busybox-hwclock-systemd anyway
Jul 04 20:43:45 <JaMa> and maybe you should take this http://patchwork.openembedded.org/patch/29609/ into account
Jul 04 20:43:53 <GNUtoo-hplaptop> hmmm now fsogsmd fails
Jul 04 20:43:54 <nschle85> JaMa: thats the next problem, i would like to add rdepends  to busybox-hwclock-systemd .... for shr systemd based images but this means always install busybox.
Jul 04 20:43:57 <JaMa> as that patchset should land in meta-oe reasonably soon
Jul 04 20:44:34 <JaMa> that's the case already.. so no big deal
Jul 04 20:45:59 <nschle85> JaMa: so this i did not understand in your mail: should i create a new recipe for hwclock-sytemd ?  which is independent from busybox ?
Jul 04 20:47:36 <nschle85> JaMa: or what is the strategy ? reuse hwclock.sh or not ?
Jul 04 20:48:09 <JaMa> there is no strategy and I dont mind either way
Jul 04 20:49:41 <JaMa> hwclock is provided by util-linux or busybox, so one of those has to be in image anyway and hwclock.sh doesn't make much difference
Jul 04 20:50:55 <nschle85> JaMa: hwclock.sh contails some logic about timezones and addinitional configuration...
Jul 04 20:51:56 <nschle85> JaMa: ok, ill make the things depend on busybox and working 
Jul 04 20:52:47 <JaMa> and prepare it as .bbappend which can be easily merged with http://patchwork.openembedded.org/patch/29609/
Jul 04 20:52:48 <GNUtoo-hplaptop> paulk-desktop, do you know if that's caused by a libsamsung-ipc update: http://pastie.org/private/wzcch6moegf7dn581jr7cq ?
Jul 04 20:54:08 <paulk-desktop> error: FsoGsm.IsiCallHandler.conference: no suitable method found to override
Jul 04 20:54:20 <paulk-desktop> I think it's unrelated
Jul 04 20:54:27 <paulk-desktop> but I don't know the code, so… it's just a guess
Jul 04 20:54:36 <GNUtoo-hplaptop> oops
Jul 04 20:54:37 <GNUtoo-hplaptop> sorry
Jul 04 20:54:43 <GNUtoo-hplaptop> it's for n900
Jul 04 20:54:44 <GNUtoo-hplaptop> indeed
Jul 04 20:54:49 <GNUtoo-hplaptop> I looked too fast
Jul 04 20:54:59 <GNUtoo-hplaptop> I'm a bit tired, still ill with cold 
Jul 04 20:56:10 <nschle85> GNUtoo-hplaptop: ??? WHAT ARE YOU TALKING ABOUT ?
Jul 04 20:56:26 <nschle85> GNUtoo-hplaptop: ups i hit caps lock
Jul 04 20:56:37 <GNUtoo-hplaptop> It's summer and I managed to take cold
Jul 04 20:57:38 <nschle85> no idea about segfault ?
Jul 04 20:59:17 <CIA-69> SHR: 03shr-devel 07buildhistory * raefde30577aa 10/packages/om_gta04-oe-linux-gnueabi/shr-version/ (latest shr-version/latest): packages: Build 201207042239 of shr 20120704 for machine om-gta04 on opmbuild
Jul 04 21:08:02 <nschle85> GNUtoo-hplaptop: I know you are ill and its late. Can you provide me an irc channel  where people can help me ? (segfault problem)
Jul 04 21:08:11 <GNUtoo-hplaptop> #C ?
Jul 04 21:08:45 <PaulFertser> nschle85: just get a coredump, do "backtrace full" in gdb, and you'll see exactly where it faults and why.
Jul 04 21:12:08 <nschle85> PaulFertser: this i already tried , but i cannot find any core dump file
Jul 04 21:12:23 <PaulFertser> nschle85: if you say it segfault on sscanf, i'd bet it's because cvoice is NULL (or invalid).
Jul 04 21:13:13 <PaulFertser> nschle85: ``ulimit -c unlimited'' before starting the app. Or just "gdb /path/to/your/app" and no need for a coredump.
Jul 04 21:14:55 <nschle85> PaulFertser: http://pastebin.com/kJ5Bvknb
Jul 04 21:15:52 <PaulFertser> nschle85: now type "bt"
Jul 04 21:16:12 <PaulFertser> then "frame <number>" where number is the one that has that at_cvoice_parse_response()
Jul 04 21:16:17 <PaulFertser> nschle85: then "p cvoice"
Jul 04 21:17:17 <nschle85> PaulFertser: http://pastebin.com/XBhQ30hV
Jul 04 21:17:36 <PaulFertser> nschle85: "frame 3" before "p"
Jul 04 21:17:48 <PaulFertser> Ah, no need, i see cvoice without it.
Jul 04 21:18:20 <PaulFertser> But i wonder if "p *cvoice" would give any error
Jul 04 21:18:58 <nschle85> PaulFertser: i did not compile all parts with debug symbols
Jul 04 21:19:17 <nschle85> PaulFertser: may be  that the problem ?
Jul 04 21:19:31 <PaulFertser> nschle85: no
Jul 04 21:19:48 <PaulFertser> nschle85: have you tried "frame 3" and then "p *cvoice"?
Jul 04 21:19:51 <nschle85> ill show you where cvoice is initialized
Jul 04 21:20:05 <nschle85> no i did not ry
Jul 04 21:20:43 <PaulFertser> nschle85: i'd guess cvoice pointer is invalid, e.g. already freed.
Jul 04 21:21:23 <PaulFertser> nschle85: either that, or result is not null-terminated.
Jul 04 21:21:46 <nschle85> http://pastebin.com/9uRazkUb
Jul 04 21:22:16 <PaulFertser> nschle85: first: you should be passing actual buffer size to strncpy, i.e. sizeof(result).
Jul 04 21:22:34 <PaulFertser> nschle85: second: if using sscanf, you must ensure result is null-terminated.
Jul 04 21:22:59 <PaulFertser> nschle85: ah, invalid pointer, my first guess is correct.
Jul 04 21:23:09 <PaulFertser> nschle85: most probably use-after-free.
Jul 04 21:23:24 <nschle85> PaulFertser: http://pastebin.com/GhtfC3Ka
Jul 04 21:24:19 <nschle85> PaulFertser: i dont think so
Jul 04 21:24:50 <PaulFertser> nschle85: yes, in this case it looks more like a stack corruption.
Jul 04 21:25:22 <PaulFertser> nschle85: do "b at_cvoice_parse_response" and then "r" (to restart the program)
Jul 04 21:25:59 <PaulFertser> nschle85: then check when you expect the segfault to happen, step-by-step that function and notice if "cvoice" in the end is the same as was passed.
Jul 04 21:26:07 <PaulFertser> nschle85: but first thing: fix strncpy!
Jul 04 21:27:02 <PaulFertser> That's the most likely candidate for the breakage.
Jul 04 21:27:18 <Alex[sp3dev]> PaulFertser: what's up
Jul 04 21:27:23 <PaulFertser> Alex[sp3dev]: yo :)
Jul 04 21:27:30 <PaulFertser> Going to sleep
Jul 04 21:27:38 <Alex[sp3dev]> ah, me too
Jul 04 21:27:40 <PaulFertser> But wanted to try to help nschle85 first :)
Jul 04 21:27:46 <PaulFertser> I hope i already did though :)
Jul 04 21:28:21 <Alex[sp3dev]> spent today setting up git server and then fixing ril. turned out the problem with sound was wrong pinmux in uboot. huh.
Jul 04 21:28:44 <Alex[sp3dev]> guess I should better spend my time learning gdb and jtag before someone steals my jlink and panda
Jul 04 21:28:47 <nschle85> PaulFertser: cvoice is different
Jul 04 21:29:00 <PaulFertser> Alex[sp3dev]: gdb + jtag is awesome!
Jul 04 21:29:21 <Alex[sp3dev]> PaulFertser: my friend told me step-by-step debugging kernel is amazing
Jul 04 21:29:26 <PaulFertser> Alex[sp3dev]: but you can get jtag for cheap with Olimex ARM OCD, or build it yourself with ft232r or the like.
Jul 04 21:30:01 <Alex[sp3dev]> PaulFertser: I have jlink at work. and jlink can handle 1.2-5V while olimex is 3.3V. Panda is 1.8V
Jul 04 21:30:01 <nschle85> Alex[sp3dev]: i have a panda... did not fulfill my expectations
Jul 04 21:30:02 <PaulFertser> nschle85: fix strncpy, and add result[sizeof(result)-1] = '\0';
Jul 04 21:30:31 <PaulFertser> Alex[sp3dev]: ughugh
Jul 04 21:30:32 <PaulFertser> Well
Jul 04 21:30:36 <PaulFertser> :)
Jul 04 21:31:30 <PaulFertser> Olimex is 2.0-5V, probably will work with an 1.8 target too.
Jul 04 21:31:35 <nschle85> PaulFertser: before you go to bed, what do you mean with fix strncopy ? limit to 50 ?
Jul 04 21:31:45 <PaulFertser> nschle85: of course
Jul 04 21:31:54 <PaulFertser> nschle85: limit to sizeof(result)
Jul 04 21:32:31 <nschle85> PaulFertser: sorry for my stupid questions but in java i have not such meta problems :-)
Jul 04 21:32:33 <Alex[sp3dev]> PaulFertser: btw, me and paulk have finally finished adding galaxy s2 and galaxy nexus support to the free ril and libsamsung-ipc
Jul 04 21:32:36 <lindi-> PaulFertser: it would be more awesome if gdb actually understood linux threads
Jul 04 21:33:18 <Alex[sp3dev]> either way, I'm heading to bed now. Good night everyone. Though I'll probably monitor the channel for another 5 minutes or so
Jul 04 21:33:36 <PaulFertser> Alex[sp3dev]: i've seen your chat here, congrats!
Jul 04 21:33:43 <PaulFertser> Alex[sp3dev]: and good night!
Jul 04 21:33:57 <PaulFertser> lindi-: but it does with OpenOCD 5.0 and later :)
Jul 04 21:34:16 <lindi-> PaulFertser: and openocd seemed really buggy for me :/
Jul 04 21:34:46 <PaulFertser> lindi-: i use it extensively at work and though sometimes i face minor issues, i'm very grateful for its existance.
Jul 04 21:34:56 <lindi-> PaulFertser: well it's the only thing we have
Jul 04 21:35:11 <CIA-69> SHR: 03shr-devel 07buildhistory * r61d8b5f99eaa 10/packages/crespo-oe-linux-gnueabi/shr-version/ (latest shr-version/latest): packages: Build 201207042315 of shr 20120704 for machine crespo on opmbuild
Jul 04 22:09:22 <CIA-69> SHR: 03shr-devel 07buildhistory * r6f0c3fb25288 10/packages/palmpre-oe-linux-gnueabi/shr-version/ (latest shr-version/latest): packages: Build 201207042350 of shr 20120704 for machine palmpre on opmbuild
Jul 04 22:50:19 <CIA-69> SHR: 03shr-devel 07buildhistory * r72a832814ea3 10/packages/palmpre2-oe-linux-gnueabi/shr-version/ (latest shr-version/latest): packages: Build 201207050025 of shr 20120704 for machine palmpre2 on opmbuild
**** ENDING LOGGING AT Thu Jul 05 02:59:58 2012