**** BEGIN LOGGING AT Sun Aug 26 03:00:00 2012 Aug 26 08:03:45 SHR: 03shr-devel 07buildhistory * rb8d2a1e8c3ae 10/ (11 files in 11 dirs): packages: Build 201208260945 of shr 20120826 for machine om-gta02 on opmbuild Aug 26 08:23:02 SHR: 03shr-devel 07buildhistory * r4445d3c9af2b 10/packages/ (10 files in 10 dirs): packages: Build 201208261012 of shr 20120826 for machine nokia900 on opmbuild Aug 26 08:41:01 SHR: 03shr-devel 07buildhistory * r5aaf3dd4025f 10/packages/om_gta04-oe-linux-gnueabi/shr-version/ (latest shr-version/latest): packages: Build 201208261031 of shr 20120826 for machine om-gta04 on opmbuild Aug 26 09:00:27 SHR: 03shr-devel 07buildhistory * r0f9b0e7cfa93 10/packages/crespo-oe-linux-gnueabi/shr-version/ (latest shr-version/latest): packages: Build 201208261049 of shr 20120826 for machine crespo on opmbuild Aug 26 09:19:14 SHR: 03shr-devel 07buildhistory * r80966e40de56 10/packages/tuna-oe-linux-gnueabi/shr-version/ (latest shr-version/latest): packages: Build 201208261109 of shr 20120826 for machine tuna on opmbuild Aug 26 09:34:23 SHR: 03Martin.Jansa 07shr-chroot * r98ba3b9ad623 10/ (149 files in 20 dirs): system upgrade Aug 26 11:48:56 morphis, ping Aug 26 11:49:03 GNUtoo: pong Aug 26 11:49:24 If I remove framebuffer console from crespo it boots until Xorg Aug 26 11:50:02 and then? Aug 26 11:50:15 then it works Aug 26 11:50:22 but framebuffer console was a nice thing Aug 26 11:50:32 altough dropping the device is worse Aug 26 11:50:52 why it should be dropped from SHR? Aug 26 11:51:11 shr-image didn't boot Aug 26 11:51:17 it rebooted when Xorg started Aug 26 11:51:23 ah ok Aug 26 11:51:24 which caused filesystem corruption Aug 26 11:51:33 and at next boot xorg didn't start Aug 26 11:51:36 didn't tried shr-image on crespo for a long time Aug 26 11:51:46 which make it appear like a filesystem corruption problem Aug 26 11:51:47 and you know why it's diding? Aug 26 11:51:53 no Aug 26 11:51:55 s/diding/dieing/ Aug 26 11:51:56 no serial console... Aug 26 11:51:57 morphis meant: and you know why it's dieing? Aug 26 11:52:08 hm Aug 26 11:52:15 and nothing in cat /proc/last_kmsg Aug 26 11:52:22 I mean nothing interesting Aug 26 11:52:57 for the chvt I had an userspace workarround Aug 26 11:53:02 I did a samsung-fb-console recipe in meta-samsung for the tuna device which can be pulled into development images Aug 26 11:53:20 maybe we should do something similar for the crespo and just pull it into a shr-dev-image Aug 26 11:53:59 ah? Aug 26 11:54:01 what's that Aug 26 11:54:08 basically I went until that point Aug 26 11:54:14 I modified a systemd config file Aug 26 11:54:21 which permitted to start Xorg manually Aug 26 11:54:36 but after rebooting when xserver-nodm-init-systemd is installed Aug 26 11:54:45 it reboots again when Xorg start Aug 26 11:55:38 http://www.shr-project.org/trac/ticket/2054 warning the bug is long Aug 26 11:59:32 ah but that won't work Aug 26 11:59:39 your samsung script.... Aug 26 11:59:45 it's for the DSS subsystem Aug 26 11:59:50 for a different problem Aug 26 12:00:08 yes, we have to create a similar script which does it for the crespo Aug 26 12:00:34 samsung-fb-console is just a wrapper for a script to enable framebuffer console on boot Aug 26 12:01:59 ok Aug 26 12:02:06 GNUtoo: thats the idea Aug 26 12:02:18 you mean a script that disables framebuffer console? Aug 26 12:02:21 so we might have a shr-image without fb-console Aug 26 12:02:29 like just before Xorg starts? Aug 26 12:02:32 and a shr-dev-image with framebuffer-console disabled Aug 26 12:02:44 it should be enabled by the script Aug 26 12:02:51 how to disable framebuffer console? Aug 26 12:03:09 isn't there a kernel option for this? Aug 26 12:03:28 we could try to do a systemd unit Aug 26 12:03:36 that disables it before Xorg starting Aug 26 12:03:39 it may not work tough Aug 26 12:03:47 because it dies on VT_ACTIVATE Aug 26 12:04:01 and the fact that it dies or not seem governed by that: Aug 26 12:04:10 NAutoVTs=0 Aug 26 12:04:19 /etc/systemd/logind.conf Aug 26 12:04:28 if you put to 0 chvt works Aug 26 12:04:30 else it fails Aug 26 12:04:35 and the phone reboots Aug 26 12:05:34 and logind.conf is a config file for that program: Aug 26 12:06:32 /lib/systemd/bin/sytemd-logind or something like that Aug 26 12:06:42 I must boot under SHR to verify the path Aug 26 12:07:15 that program is responsible for the logins Aug 26 12:07:30 NAutoVTs=0 does that: Aug 26 12:08:48 How many virtual terminals to allocate by default and when switched to autospawn Aug 26 12:09:23 http://pastie.org/private/5wgrc7zp4il1u9xrvtenhg Aug 26 12:09:54 so when you chvt I guess it spawn something automatically Aug 26 12:09:58 and that makes it reboot Aug 26 12:10:01 same for Xorg Aug 26 12:10:10 because Xorg chvts Aug 26 12:10:59 morphis, how to disable framebuffer console from userspace? Aug 26 12:11:34 I really wonder what caused all that mess? Aug 26 12:11:37 systemd upgrade? Aug 26 12:11:41 GNUtoo: don't really know Aug 26 12:11:44 ok Aug 26 12:12:43 I'll read the kernel sources then Aug 26 12:12:49 because there may be an option for that Aug 26 12:15:32 GNUtoo: ok Aug 26 12:16:42 echo 0 > sys/class/vtconsole/vtcon1/bind Aug 26 12:16:45 I'll try that Aug 26 12:17:04 from Documentation/fb/fbcon.txt Aug 26 12:18:59 ok Aug 26 12:22:44 GNUtoo: I will start tomorrow with working on telephony bits for the tuna Aug 26 12:23:03 ok Aug 26 12:23:20 did you find the time to merge alex[sp3dev] 's code Aug 26 12:23:27 I will merge his code in samsung-ril Aug 26 12:23:36 but I've no access to libsamsung-ipc Aug 26 12:23:49 and did you merge my patch too in libsamsung-ipc? Aug 26 12:28:02 GNUtoo: will do tomorrow Aug 26 12:28:11 ok Aug 26 12:28:19 there are also gta04 patches Aug 26 12:28:29 for hardware routing on A4 Aug 26 12:28:45 yes, you already told me Aug 26 12:28:49 will merge them too Aug 26 12:29:22 ok nice Aug 26 12:36:52 SHR: 03morphis 07meta-smartphone * r293cb96a8110 10/meta-samsung/recipes-graphics/libgles-omap4/libgles-omap4_1.7.10.0.1.21.bb: meta-samsung: libgles-omap4: add initial recipe for opengl libraries Aug 26 12:37:58 thanks for adbd Aug 26 12:38:10 GNUtoo: you mean the recipe? Aug 26 12:38:13 yes Aug 26 12:38:27 I thought it might be helpful to use it instead of rndis Aug 26 12:38:31 in some situations Aug 26 12:38:51 yes Aug 26 12:39:18 about the libgles-omap I guess it's not used by default Aug 26 12:39:33 yes Aug 26 12:39:39 it's currently not used anywhere Aug 26 12:39:45 ok Aug 26 12:39:59 I don't think SHR will use it Aug 26 12:40:17 ok Aug 26 12:41:50 GNUtoo: I will add rndis and adb as default configuration in our rndissetup script Aug 26 12:41:55 ok Aug 26 12:44:08 SHR: 03morphis 07meta-smartphone * r47b6278075b3 10/meta-samsung/recipes-core/initscripts/ (initscripts/tuna/rndissetup.sh initscripts_1.0.bbappend): meta-samsung: initscripts: add one second sleep to wait for interface to come available Aug 26 13:09:15 GNUtoo, morphis : hi Aug 26 13:10:04 Alex[sp3dev]: heyho Aug 26 13:10:22 Alex[sp3dev], I've got it working under crespo Aug 26 13:10:40 it works with sound(one small patch is needed for the srs client) Aug 26 13:10:47 morphis: would you merge the request @ github? I've added some stuff, including boot marker for your radio Aug 26 13:11:44 Alex[sp3dev]: yes I will do tomorrow Aug 26 13:11:50 today I am busy with some other stuff Aug 26 13:11:52 GNUtoo: Post the diff then. I've found why the first call hangs on tuna. Turns out SRS cannot handle multiple clients. Well, I've also added pthread mutexes to RIL, so now we don't have memory corruption in some places and RSSI is correct Aug 26 13:12:09 ok Aug 26 13:13:41 I've ahttp://pastie.org/private/stldb6qgpyutwjudtnhedg Aug 26 13:13:42 oops Aug 26 13:13:43 http://pastie.org/private/stldb6qgpyutwjudtnhedg Aug 26 13:13:55 I've also tested on tuna without luck for the audio volume Aug 26 13:14:25 GNUtoo: so what was the problem on crespo? Aug 26 13:14:49 tuna changes http://pastie.org/private/o54jzd0ju4ao721xkqtw Aug 26 13:14:55 GNUtoo: I have the patch for GB now. You should try tuna-tracking branch and merge it if it works Aug 26 13:14:55 on crespo the problem was that: Aug 26 13:15:26 http://pastie.org/private/k5xrz6oeyry0flogakeg Aug 26 13:15:36 I'll push right after pushing your changes Aug 26 13:15:41 (for the srs changes) Aug 26 13:15:57 not a big issue Aug 26 13:16:08 and USSD do work on crespo!!!! Aug 26 13:16:20 GNUtoo: so, patching the SRS client worked on GB? that's good because we don't have to make yet another ifdef Aug 26 13:16:30 yes indeed Aug 26 13:16:54 I really want to push your changes Aug 26 13:17:00 what should I do now? Aug 26 13:17:04 yeah, USSD are nice. For latin USSD, I simply added some magic bytes (according to proprietary IPC dumps). for UTF8 I copied the decoding code from xdandroid ril Aug 26 13:17:09 ok Aug 26 13:17:14 GNUtoo: have you tested gprs? Aug 26 13:17:17 no Aug 26 13:17:24 let me look Aug 26 13:17:34 I'll have to find out how much it costs to test it Aug 26 13:17:52 if gprs works, you can just merge my tuna-tracking branch Aug 26 13:18:06 ok Aug 26 13:25:57 gprs deregisters crespo Aug 26 13:27:01 logcat or didn't happen, you know the rules :) Aug 26 13:29:17 http://gnutoo.homelinux.org/downloads/people/Alex%5bsp3dev%5d/gprs-crespo.log Aug 26 13:29:29 one second.... Aug 26 13:29:34 because the log are pretty long Aug 26 13:29:37 impossible to pastebin Aug 26 13:30:22 and for main: Aug 26 13:30:23 http://pastie.org/private/gzqef3g69xbvkp6q9tyaxq Aug 26 13:30:53 ipc_gprs_call_status Aug 26 13:31:12 SEGV_MAPERR), fault addr 00000001 gotta love these. very useful address :) Aug 26 13:31:25 ok, I'll try to fix that Aug 26 13:32:09 ok Aug 26 13:32:12 thanks a lot Aug 26 13:34:37 GNUtoo: could you start commenting out the free() calls in cleanup_ril_data_call_response? I think it's free(response->type) that's crashing it Aug 26 13:35:12 ok Aug 26 13:36:58 my thoughts so far: 1. we need to teach paulk to synchronize parallel code 2. we need to always bzero() data structures and 3. we should switch to C++ and use smart pointers Aug 26 13:37:30 ok for 1 and 2 Aug 26 13:37:33 not sure about 3 Aug 26 13:38:01 any good reason *not* to use c++? Aug 26 13:38:30 you can always export C-compatible API from C++ code via extern "C" Aug 26 13:39:20 less people know C++ Aug 26 13:39:28 I guess paulk doesn't know C++ Aug 26 13:39:31 neither do I Aug 26 13:40:40 my hard disks are slow.... Aug 26 13:40:50 GNUtoo: I'm also not a C++ expert but ignorance is not something to be proud of. I'm just sick of manual malloc/free and of the "if rc < 0 goto fail" paradigm. I'll read more on C++ and will let you know what I think later Aug 26 13:41:06 ok Aug 26 13:41:24 I guess vala is out of the question? Aug 26 13:41:33 that would be too complicated to port to android? Aug 26 13:41:45 + it often has bugs Aug 26 13:43:23 basically it's doing rootfs for fso2-console-image + the hdds are slow => the lib will compile in a while.... Aug 26 13:43:29 *libsamsung-ril Aug 26 13:43:39 besides, one good reason to use C++ is STL. STL offers data structures: vector (dynamic array), priority queue etc. With these structures we could replace hardcoded message buffers and make proper queues for IPC notifications. Of course, it's easy to write a data structure in C, but it doesn't feel OK to use your own library everywhere Aug 26 13:43:52 GNUtoo: use "mmm hardware/ril/samsung-ril" Aug 26 13:43:59 I did: Aug 26 13:44:03 make -j9 libsamsung-ril Aug 26 13:44:09 no Aug 26 13:44:13 use mmm Aug 26 13:44:18 make will scan the whole source tree Aug 26 13:44:23 ok Aug 26 13:44:32 I know mm Aug 26 13:44:39 I'll try mmm Aug 26 13:44:52 there's a "make help" command which explains mm/mmm Aug 26 13:44:59 ah it started compiling already Aug 26 13:45:32 well, AOSP make help is not verbose but the mmm command is there. Aug 26 13:49:23 I'll comment more code because of http://pastie.org/private/ylbuoprz9mbkiwve3my0g Aug 26 13:54:34 http://pastie.org/private/beoz8qtbaec1gtbzxybdmw Aug 26 13:54:39 it continues to crash with: Aug 26 13:55:01 http://pastie.org/private/kmkfhlzkjnwoit63ln2og Aug 26 13:55:45 basically, we should allocate the setup_data_call_response from heap, not on stack. I will see what can be done Aug 26 13:55:58 ok Aug 26 13:56:47 I hope to fix it before the credit finishes Aug 26 13:57:04 I meant: Aug 26 13:57:17 I hope it will be fixed by you before my credit finishes in trials Aug 26 14:17:19 GNUtoo: I'm out of good ideas. change "address = ip" at line 553 to address = strdup(ip) and try doing a malloc and memcpy of setup_data_call_response in ipc_gprs_call_status Aug 26 15:25:51 GNUtoo: I'm back Aug 26 15:25:56 ok Aug 26 15:25:59 I'm with gdb Aug 26 15:26:02 GNUtoo: read the jabber Aug 26 15:26:20 ok Aug 26 15:28:08 http://pastie.org/private/jjfo4vkda2kbp8dlgghxa Aug 26 15:28:21 Alex[sp3dev], let's continue on IRC Aug 26 15:28:28 GNUtoo: ol Aug 26 15:28:29 ok Aug 26 15:30:06 do you want me to run some gdb stuff? Aug 26 15:30:41 well, no, I have some, ugh, idea Aug 26 15:31:39 ah? Aug 26 15:32:04 GNUtoo: please edit ril/libril/ril.cpp and in the responseDataCallList function print the pointers and strings inside the loop (before writeStringToParcel). I think it fails for some reason there Aug 26 15:32:47 ok Aug 26 15:32:55 http://pastie.org/private/ef3jsou03wqoazdrgnitpq Aug 26 15:32:58 I'll do that Aug 26 15:35:23 I print that: Aug 26 15:35:25 p.writeInt32(p_cur[i].cid); Aug 26 15:35:26 p.writeInt32(p_cur[i].active); Aug 26 15:35:26 ? Aug 26 15:35:53 GNUtoo: print first the pointer (%p) and then the string (%s) Aug 26 15:36:02 ok Aug 26 15:42:43 http://pastie.org/private/uugcq6fsoluowibwc3f8iq has: Aug 26 15:42:47 E/RILC ( 725): [DEBUGME pointer1] 0x1 Aug 26 15:42:51 here you go Aug 26 15:42:55 GNUtoo: you have to print .type and .address, args to writeStringToParcel Aug 26 15:42:58 ok Aug 26 15:43:19 isn't 0x1 the address we got before Aug 26 15:43:47 + LOGE("[DEBUGME pointer1] %p",p_cur[i]); Aug 26 15:43:47 + LOGE("[DEBUGME pointer1] %s",p_cur[i].cid); Aug 26 15:43:47 + LOGE("[DEBUGME pointer1] %s",p_cur[i].active); Aug 26 15:43:50 here's the diff Aug 26 15:43:50 GNUtoo: it fails inside writeStringToParcel (actually, strdup8to16) Aug 26 15:43:54 ok Aug 26 15:44:03 so we need the type and address strings Aug 26 15:44:15 I suspect it's a bug in GB which we haven't hit before Aug 26 15:44:22 ok Aug 26 15:44:29 but....0x1 isn't a valid address Aug 26 15:44:45 and that's what we hit before Aug 26 15:44:46 right? Aug 26 15:44:52 GNUtoo: cid is not a pointer, it's an integer Aug 26 15:45:18 ah ok sorry Aug 26 15:45:31 but p[cur[i]] ? Aug 26 15:45:34 oops Aug 26 15:45:39 p_cur[i] Aug 26 15:45:45 I'll improve the debugging Aug 26 15:45:53 I didn't slept well at all yesterday sorry Aug 26 15:46:00 why? Aug 26 15:46:37 LOGE("%s: %p", __func__, p_cur[i].type);LOGE("%s: %p", __func__, p_cur[i].address);LOGE("%s: %s", __func__, p_cur[i].type);LOGE("%s: %s", __func__, p_cur[i].address); Aug 26 15:46:49 ok Aug 26 15:47:56 ah you want the content.... Aug 26 15:48:15 but could p_cur[i] be wrong? Aug 26 15:48:34 don't think so Aug 26 15:49:26 ok Aug 26 15:50:50 main: http://pastie.org/private/fisi4ni7uhzb7kjbsof2kq Aug 26 15:51:21 radio: http://pastie.org/private/jzlgq5afag5wvalvgga Aug 26 15:52:08 hmmm Aug 26 15:52:16 0x0 (null) etc... Aug 26 15:52:34 GNUtoo: let me guess. you've printed the .apn field? Aug 26 15:52:59 GNUtoo: I have opened GB source and turns out .apn field is used but it is unused in ICS. maybe we need to fill it Aug 26 15:53:05 ok Aug 26 16:00:16 what should I do exactly with reguard to APN? Aug 26 16:00:20 GNUtoo: try adding to the ipc_gprs_connection_enable around line 553 (after setup_data_call_response->address = ip) Aug 26 16:00:28 asprintf(&setup_data_call_response->apn, "%s", gprs_connection->define_context.apn); Aug 26 16:00:56 ok same than with version 6 then Aug 26 16:01:08 no, there's no apn for version 6 Aug 26 16:01:16 ah sorry Aug 26 16:01:19 version < 6 Aug 26 16:01:25 I've that: Aug 26 16:01:33 #if RIL_VERSION < 6 Aug 26 16:01:39 resp->address = addr; Aug 26 16:01:49 asprintf(&(resp->apn), "%s",gprs_connection->define_context.apn); Aug 26 16:01:58 hmmm Aug 26 16:02:03 GNUtoo: hey. I told you to add it to ipc_gprs_connection_enable and you're looking at ipc_gprs_pdp_context Aug 26 16:02:28 ok sorry Aug 26 16:02:38 I just grepped for that Aug 26 16:02:41 with // Aug 26 16:03:09 I'll add what you tell me.... Aug 26 16:03:34 In the meanwhile I still need to fix SRS mess Aug 26 16:04:11 ok Aug 26 16:05:13 should I remove the free(gateway) ? Aug 26 16:05:26 I'm telling only stupid things today it seems tough Aug 26 16:05:56 GNUtoo: no, leave it there. should not matter Aug 26 16:06:02 ok Aug 26 16:08:54 main: http://pastie.org/private/esilplwfjoo0kbtoxxnpa Aug 26 16:09:29 radio: http://pastie.org/private/pwg4nkat8zlt4jhihpsg Aug 26 16:10:46 Alex[sp3dev], is there a way to get the line? Aug 26 16:10:56 GNUtoo: __LINE__ Aug 26 16:10:57 like if I push a non-stripped version of samsung-ril? Aug 26 16:11:03 I know that Aug 26 16:11:08 addr2line Aug 26 16:11:10 but I mean in the debug Aug 26 16:11:18 yes but do they give the hex? Aug 26 16:11:20 hm.. "where" in gdb? Aug 26 16:11:27 ah yes Aug 26 16:11:28 http://pastie.org/private/pwg4nkat8zlt4jhihpsg Aug 26 16:11:29 oops Aug 26 16:11:29 GNUtoo: yes, the hex is in the logcat Aug 26 16:11:34 00003b1c Aug 26 16:11:44 I'll addr2line the samsung-ril then Aug 26 16:11:57 addr2line -f -e binary 00003b1c Aug 26 16:12:11 that's what I do with the kernel Aug 26 16:12:17 I'll try with libsamsung-ril.so Aug 26 16:12:19 $((0x3b1c)) I think Aug 26 16:14:02 /home/gnutoo/embedded/android/replicant-2.3/external/libsamsung-ipc/samsung-ipc/device/aries/aries_ipc.c:100 Aug 26 16:14:05 strange.... Aug 26 16:16:13 because it's libril.so and not libsamsung-ril.so Aug 26 16:16:25 you should really take a rest :) Aug 26 16:17:02 I have no idea why all the pointers in the ril are null. according to your log Aug 26 16:17:31 ok Aug 26 16:17:36 I've taken a 5 min rest Aug 26 16:19:24 what if I do exactly what you tell me and try to stop thinking or acting on my own for now? Aug 26 16:20:09 what comand should I type to get libril addr2line? Aug 26 16:21:15 (arm-eabi-addr2line -f -e out/target/product/crespo/obj/SHARED_LIBRARIES/libril_intermediates/LINKED/libril.so $((0x3b1c)) fails....) Aug 26 16:22:06 GNUtoo: maybe, remove the -f argument Aug 26 16:22:25 ok Aug 26 16:22:40 same Aug 26 16:22:46 ??:0 Aug 26 16:22:51 instead of : Aug 26 16:22:58 ?? Aug 26 16:22:58 ??:0 Aug 26 16:23:23 maybe I use the wrong dir Aug 26 16:23:39 I need unstripped but complete lib Aug 26 16:25:06 GNUtoo: I have no good ideas, maybe I've misused the asprintf. could you try one last hack for today> Aug 26 16:25:07 ? Aug 26 16:25:17 yes Aug 26 16:25:31 I can try all the stuff you want Aug 26 16:25:40 just you have to guide more than ususal..... Aug 26 16:25:54 GNUtoo: in the ipc_gprs_connection_enable, use strdup when assigning setup_data_call_response->{address, apn} Aug 26 16:26:05 ok Aug 26 16:26:11 well, that's probably not it Aug 26 16:26:19 see, we strdup the type ("IP") Aug 26 16:26:28 but in the ril callback, it is NULL (where you have printed it) Aug 26 16:26:30 ok Aug 26 16:26:37 GNUtoo: which branch are you using? Aug 26 16:26:42 let me look Aug 26 16:26:47 GNUtoo: can you try my latest tuna-tracking? Aug 26 16:26:59 Fix GPRS on ICS/JB Aug 26 16:27:01 ok Aug 26 16:27:13 but I should revert all my changes or something like that? Aug 26 16:27:18 if it works, kill paulk. if it doesn't, kill me :) Aug 26 16:27:39 ok Aug 26 16:27:41 lol Aug 26 16:27:53 GNUtoo: revert samsung-ril changes but leave the debugging in libril Aug 26 16:27:56 ok Aug 26 16:28:23 where's your branch? Aug 26 16:28:29 gitorious Aug 26 16:28:31 ok Aug 26 16:28:42 git://gitorious.org/~ast/replicant/asts-hardware_ril_samsung-ril.git Aug 26 16:28:49 git://gitorious.org/~ast/replicant/asts-hardware_ril_samsung-ril.git ? Aug 26 16:28:51 ok Aug 26 16:31:06 still crashes Aug 26 16:31:28 http://pastie.org/private/wajbrzikgmhazvj6kit2eg Aug 26 16:31:47 http://pastie.org/private/purvvne5fdezcx7hegdmqq Aug 26 16:33:25 hmmm Aug 26 16:33:34 arm-eabi-addr2line -e out/target/product/crespo/symbols/system/lib/libril.so 0x3b1c gives: Aug 26 16:33:38 /home/gnutoo/embedded/android/replicant-2.3/hardware/ril/libril/ril.cpp:2724 Aug 26 16:33:55 ret = pRI->pCI->responseFunction(p, response, responselen); Aug 26 16:34:12 inside RIL_onRequestComplete Aug 26 16:37:39 Alex[sp3dev], ^^^ Aug 26 16:37:46 do you have a new idea? Aug 26 16:37:55 GNUtoo: I'm tired of it for today. Let's wait for paulk and let him debug. I'm not sure it is good to merge broken code Aug 26 16:38:08 GNUtoo: Idea: switch crespo to ICS ^=^ Aug 26 16:38:19 it has USSD and breaks GPRS..... Aug 26 16:38:30 so what's better? USSD or GPRS? Aug 26 16:38:32 the rest works Aug 26 16:38:50 I guess most people wants GPRS Aug 26 16:39:46 well, let's just postpone the merge. I'll try to look at it and fix SRS. let's get back to it on wednesday or when paulk is back Aug 26 16:39:55 ok Aug 26 16:40:06 he should be back very soon Aug 26 16:40:33 GNUtoo: see you. I'll go to shower and be away from the internets. Maybe I'll try building GB libril and ril on ICS Aug 26 16:41:28 wifi cafee is closing anyway Aug 26 16:41:30 bye Aug 26 21:08:50 paulk-desktop: hey Aug 26 21:30:33 Alex[sp3dev], hi! Aug 26 21:30:59 paulk-desktop: I rate your POSIX skills 3/10 :) Aug 26 21:31:27 lol Aug 26 21:31:38 paulk-desktop: please, be careful and always do bzero/memset, check null pointers and use pthread mutexes when using threads :) Aug 26 21:31:49 paulk-desktop: and yes, 0 is a valid file descriptor :) Aug 26 21:31:57 I know Aug 26 21:32:03 but it's usually stdin / stdout Aug 26 21:32:38 though I know I'm not an highly skilled programmer Aug 26 21:32:47 I didn't learn that at school nor in books Aug 26 21:32:58 I'm 17 yo for the record Aug 26 21:33:23 these are my excuses Aug 26 21:33:27 paulk-desktop: I know that story. I also didn't know C before I was 16 and started hacking linux for the first time :) Aug 26 21:33:29 but you're right, I should be more careful Aug 26 21:34:16 paulk-desktop: still I recommend reading about threading and POSIX. I recommend "advanced programming in the unix environment" and tannenbaum's book about Operating Systems Aug 26 21:34:26 okay Aug 26 21:34:29 thanks Aug 26 21:34:53 I'm 20 btw :) Aug 26 21:35:09 paulk-desktop: me and GNUToo have been trying to merge my galaxy nexus changes to replicant upstream Aug 26 21:35:17 okay Aug 26 21:35:22 so there is a lot to process there Aug 26 21:35:50 about ICS inclusion in general, I decided to make use of the RIL_VERSION and ifdef all the code Aug 26 21:35:52 paulk-desktop: in short, I've added the compat.h header to ifdef stuff on ICS/GB releases. I have fixed GPRS on ICS. I want you to build and test my tuna-tracking branch on crespo GB because GPRS crashes there and GNUToo failed to debug it Aug 26 21:35:58 paulk-desktop: ^_^ Aug 26 21:36:02 paulk-desktop: that's what I've done Aug 26 21:36:06 ok Aug 26 21:36:13 I know Aug 26 21:36:19 I just, well, took the same decision Aug 26 21:36:33 paulk-desktop: I've also fixed RSSI (well, just added mutexes) and sound (reworked SRS to make it accept multiple clients) Aug 26 21:36:42 okay nice Aug 26 21:37:01 btw apparently sound on tuna doesn't require anything more than ALSA routing Aug 26 21:37:04 so if you debug the null pointers in GB GPRS, it is ready to merge Aug 26 21:37:15 ok I'll look at it all Aug 26 21:38:22 paulk-desktop: after we merge, I would like that we implemented flight mode. I mean, instead of going LPM, we should close IPC channels and turn off the modem and re-init everything on exit. I want it mainly for security reasons and that's going to be hard to implement Aug 26 21:39:08 and yes, GNUToo reports USSD working on crespo now with my 'magic' hack Aug 26 21:39:19 ah nice Aug 26 21:39:33 we should probably merge it to libsamsung-ipc Aug 26 21:39:42 that's where all the crappy stuff goes Aug 26 21:39:51 later maybe Aug 26 21:39:51 s/crappy/magic/ Aug 26 21:39:52 paulk-desktop meant: that's where all the magic stuff goes Aug 26 21:40:09 basically, all that we can't explain… Aug 26 21:40:13 well, in terms of crapiness samsung-ril is surely winning the libsamsung-ipc Aug 26 21:40:20 agreed Aug 26 21:40:30 though I don't get that much segfaults Aug 26 21:43:08 paulk-desktop: how much time are you going to have this week? are you ready to fix gprs and merge my code? Aug 26 21:43:34 probably yes, but there are other (non-samsung) devices I'd like to work on first Aug 26 21:43:52 GNUtoo gave me a GeeksPhone One Aug 26 21:44:16 so I'd like to play a bit with its modem Aug 26 21:44:16 paulk-desktop: well, gprs is probably easy, it's just a null pointer. I just want to merge it and forget about it because if someone commits in between, we'll have to rebase and test again Aug 26 21:44:27 ok Aug 26 21:44:36 I'll look tomorrow then Aug 26 21:44:38 GeeksPhone One is some msm7200 crap? Aug 26 21:44:47 indeed Aug 26 21:45:02 I have a lot of experience with msm7200A Aug 26 21:45:20 I'm not going to spend a lot of time on it though Aug 26 21:45:27 as I don't like MSM/QSD at all Aug 26 21:45:42 I have ported htc kovsky (windows mobile device) from 2.6.27 to 2.6.35 and fixed a lot of stuff (wl1251 driver). I have also ported the LK bootloader. That were fun times Aug 26 21:45:59 nice Aug 26 22:05:43 gotta go, bye **** ENDING LOGGING AT Mon Aug 27 02:59:57 2012