**** BEGIN LOGGING AT Mon Jul 02 02:59:58 2012 Jul 02 04:49:03 freesmartphone.org: 03morphis 07cornucopia * rd38dd81ba51e 10/fsogsmd/src/lib/modem.vala: fsosmd: lib: get everything compiling again Jul 02 10:06:04 paulk-desktop: hi Jul 02 10:06:13 hi Alex[sp3dev]! Jul 02 10:06:59 I did want to talk to you recently: congrats about Galaxy S2 and Galaxy Nexus modem work! Jul 02 10:07:09 paulk-desktop: I have almost implemented firmware loader and libsamsung-ipc support for galaxy s2 and galaxy nexus (except NVDATA loading from efs and rfs ipc callbacks). I have voice calls and sms working Jul 02 10:07:20 very nice! Jul 02 10:07:31 what do you use on top of libsamsung-ipc? Jul 02 10:07:47 paulk-desktop: what's the current status of data (working or not, what is needed to be done), GPS (does it work with proprietary libs and free ril?) and NFS (same as GPS) Jul 02 10:07:53 paulk-desktop: samsung-ril Jul 02 10:07:56 ok Jul 02 10:08:06 I'm working on data right now Jul 02 10:08:09 for galaxy s Jul 02 10:08:16 It's not very good, but works Jul 02 10:08:29 I can explain you more in details later Jul 02 10:08:52 I failed to compile samsung-ril on ics, had to revert the gprs patch. It probably depends on some modified libsamsung-ipc? Jul 02 10:09:02 definitely Jul 02 10:09:06 it's the aries-data branch Jul 02 10:09:12 I haven't pushed to master yet Jul 02 10:09:20 but I will soon Jul 02 10:09:40 you will probably need what's on that branch to have data working on galaxy s2 Jul 02 10:09:48 though, I'm very available to work on that these days Jul 02 10:09:59 so once I have galaxy s data fixed, I'll go with galaxy s2 Jul 02 10:10:17 and if you want, I can totally do rfs and nv_data Jul 02 10:10:21 I see. Could you please explain what to look into for getting audio/gps whatever working? I am currently finishing my integration work, only rfs and nvdata are left Jul 02 10:10:26 paulk-desktop: do you have s2? Jul 02 10:10:31 yes, I do Jul 02 10:10:38 but I didn't work on it alot Jul 02 10:10:49 paulk-desktop: then I can push my changes and you can try building the library and ril yourself Jul 02 10:10:56 I'd love that! Jul 02 10:11:25 about gps: we must look the protocol of the chip Jul 02 10:11:45 on previous samsung devices (galaxy s, nexus s, galaxy tab), it was unknown and undocumented Jul 02 10:11:57 let me find a log of what it looks like Jul 02 10:12:08 paulk-desktop: the standalone fwloader is https://github.com/astarasikov/samsung-xmm6260-fw-loader . Here is the libsamsung-ipc with it integrated, but rfs is missing https://github.com/astarasikov/libsamsung-ipc Jul 02 10:12:19 ok, great Jul 02 10:12:29 can you push to morphis's libsamsung-ipc? Jul 02 10:12:44 or do you want me to do it? Jul 02 10:12:47 paulk-desktop: note that my main target is galaxy nexus now. In terms of ril it is absolutely identical to galaxy s2, but sound and gps may be different Jul 02 10:12:57 ok Jul 02 10:13:06 paulk-desktop: I'd love to first implement rfs and do some tests and then rebase and submit a pull request Jul 02 10:13:16 ok, sure Jul 02 10:13:25 I can help you with that today if you want Jul 02 10:14:17 nexus s, galaxy s, and galaxy tab GPS protocol: http://pastie.org/4186543 Jul 02 10:14:19 with logs side Jul 02 10:14:43 paulk-desktop: ok. I'd like that you try out the ril yourself (you may need to flash gingerbread to s2 if you're building against a gb tree). There's some minor issue: when there is no data on ipc socket, i return 0, and the ril library spams the logcat with unknown code 0x0000 Jul 02 10:14:43 the raw hex with 00000000, 00000010 is a direct dump of what the chip sends Jul 02 10:15:00 ok Jul 02 10:15:07 I already have gingerbread Jul 02 10:15:12 paulk-desktop: if you use the standalone loader, please edit the makefile/header to define DEBUG. you'll get a very verbose log than Jul 02 10:15:19 great Jul 02 10:15:35 so the gps protocol was called "MEIF" Jul 02 10:15:41 and there is nothing you can do with it Jul 02 10:15:57 so we should trace android's GPSD to see how it is Jul 02 10:16:07 I have a bunch of tools to help: Jul 02 10:16:14 strace is the only tool Jul 02 10:16:19 https://gitorious.org/replicant/crespo-gps-utils Jul 02 10:16:41 yes but I have made a standalone HAL that permits to strace libgps Jul 02 10:16:51 or gps.*.so Jul 02 10:17:11 anyways, I'll look into gps a bit later Jul 02 10:17:21 what's about sound? Jul 02 10:18:14 paulk-desktop: I think it doesn't work, though I tested only once. And I am unsure what needs to be implemented if anything Jul 02 10:18:38 but it's ALSA, isn't it? Jul 02 10:18:58 though there are no sources for the audio module or libaudio for gb Jul 02 10:19:59 paulk-desktop: I'm working on galaxy nexus at work, we need to eliminate closed blobs from critical parts (gps, ril, bootloader) for one project so that's what I'm doing. So we need all stuff (gprs, gps, nfc) working fine. I think we may reuse these code later for replicant or SHR, when I have time to integrate Jul 02 10:20:20 ok, graet Jul 02 10:20:38 I think GPS might be a bit better on Galaxy Nexus than on Galaxy S2 Jul 02 10:20:44 I read somewhere that it's a known protocol Jul 02 10:20:49 but that needs to be checked Jul 02 10:21:36 paulk-desktop: I have not tried tracing it yet. I suggest it's a sirf star 4 which uses MEIF by default but can be switched to nmea Jul 02 10:22:11 that's probably not the case on galaxy s,nexus s, etx Jul 02 10:22:12 etc* Jul 02 10:22:21 because there is a whole program to translate MEIF to NMEA Jul 02 10:22:28 and then NMEA is sent back to the framework Jul 02 10:22:46 so I guess if it was possible to directly get NMEA out ouf the chip, they wouldn't have bothered Jul 02 10:22:53 paulk-desktop: ok. I will still be doing work on s2 in my free time because I want to get it running a full-blown ubuntu + FSO Jul 02 10:23:01 :) Jul 02 10:23:02 ok Jul 02 10:25:27 ok, done with galaxy s data Jul 02 10:25:28 paulk-desktop: do you know if it is possible to build (mmm) an android module with dependencies without first building the whole tree? Jul 02 10:25:30 it's reliable Jul 02 10:25:58 Alex[sp3dev], you can "make [module]" , that will only build its deps Jul 02 10:26:07 and the module in the end Jul 02 10:26:12 but not anything more Jul 02 10:26:30 paulk-desktop: since data is done, could you clean it up and push the changes to morphis libsamsung-ipc and replicant samsung-ril? I hope I can somehow reuse it Jul 02 10:26:38 sure Jul 02 10:27:00 ok. so are we going to add galaxy s2 and nexus to replicant? Jul 02 10:29:07 paulk-desktop: one last thing. you'll need a 3.X ICS kernel on your galaxy s2. They've replaced the phonet usb driver with a generic modemctl stuff. Besides, modemctl in kernel now handles HDLC. so we only read/write raw ipc packets to umts_ipc0 Jul 02 10:29:40 mhh I see Jul 02 10:34:21 all pushed Jul 02 10:35:36 paulk-desktop: ok. I've also not implemented power control yet. but I think we shouldn't hurry and do everything step by step. So now I'm writing RFS callbacks Jul 02 10:35:45 ok Jul 02 10:36:22 actually, power control is there, but only inside the fwloader. we just need to export a poweroff call to libsamsung-ipc Jul 02 12:05:34 paulk-desktop: what's rfs used for? I have currently not implemented it yet, but most stuff works. is it for IMEI? Jul 02 12:05:58 nope Jul 02 12:06:16 it is used for the modem to modify it's nv_data Jul 02 12:06:26 as it's stored on NAND Jul 02 12:06:32 the modem can't modify the file on itself Jul 02 12:06:56 for instance, if you SIM unlock the phone, the modem will send an RFS request telling to write the bytes telling it's unlocked in nv_data Jul 02 12:07:09 because it can't do it on his own Jul 02 12:07:09 paulk-desktop: what about samsung-ril-client? is it binary compatible to the proprietary one? should sound work if i just copy it over the blob? Jul 02 12:07:19 I think so Jul 02 12:07:28 just rename it Jul 02 12:07:34 to libsecril-client.so Jul 02 12:07:37 and it should work, yes Jul 02 12:10:41 nope, sound doesn't work and ussd call doesn't work. do you want a radio logcat? Jul 02 12:10:44 and dmesg Jul 02 12:15:31 paulk-desktop: here is the logcat http://pastebin.com/5TtmNd2r and here is the dmesg (with ipc log) http://pastebin.com/iVQRvrua . just in case you are interested. but you have an S2 so you can check yourself Jul 02 12:16:01 they're long and you probably only need the end of the files Jul 02 12:16:39 thanks a lot Jul 02 12:16:43 I know about USSD Jul 02 12:32:04 paulk-desktop: sorry to disturb you again, but are rfs messages just regular ipc messages with group set to IPC_GROUP_RFS and aseq to 0? Jul 02 12:32:46 on the devices I know, it's not Jul 02 12:32:50 it has its own header type Jul 02 12:33:07 is it defined somewhere? or is it per-board specific? Jul 02 12:33:08 but we convert it to the fmt header with IPC_GROUP_RFS on libsamsung-ipc Jul 02 12:33:15 let me see Jul 02 12:34:08 oh wait, that's it: Jul 02 12:34:33 * all the messages have a header that is: size, id, cmd and then data Jul 02 12:34:44 * on fmt, data has a header too (with aseq, mseq, etc) Jul 02 12:34:53 * on rfs, it's the raw data Jul 02 12:35:20 * on fmt, id and cmd (from the general-purpose header) aren't filled but these are on rfs Jul 02 12:35:31 thanks Jul 02 12:35:44 so, I just copy-paste my ipc code with minor changes Jul 02 12:37:16 ok Jul 02 12:38:02 if you want more details about how we translate RFS infos to the FMT header, look at devices/crespo/crespo_ipc.c:450 Jul 02 12:38:46 paulk-desktop: on galaxy nexus, nvdata is in /factory, not in /efs. Can we somehow fix libsamsung-ipc or should I just edit init.rc to add a symlink? Jul 02 12:39:48 mhh interesting Jul 02 12:40:30 the nv_data functions in rfs.c have /efs/ hardcoded Jul 02 12:40:37 but they all have the ipc client Jul 02 12:40:53 so we can make it device-specific Jul 02 12:41:06 yes, but that's not a priority. will do it at home later Jul 02 12:41:10 ok Jul 02 12:41:34 I did that for the gprs data iface: it was hardcoded and I made it device-specific Jul 02 12:41:47 so we can do it as well with nv_data location Jul 02 12:42:16 ok. btw, according to logcat and ifconfig, rmnet0 gets address and dials the APN. so it will probably just work when I merge your changes, right? Jul 02 12:42:50 with the non-free ril you mean? Jul 02 12:43:19 paulk-desktop: no, with replicant ril. maybe something non-free kicks rmnet, but I can give you the log Jul 02 12:43:32 oh, that's interesting Jul 02 12:43:40 you will have to adapt a few things with my changes Jul 02 12:43:52 but it should be all like crespo Jul 02 12:44:13 that: https://github.com/morphis/libsamsung-ipc/commit/63a3609348e04a1899b69a4f95e626408497f230 Jul 02 12:44:51 but you should try to trace/get logs from the non-free RIL to look at how it enables data Jul 02 12:45:04 that's a bit different across devices Jul 02 12:45:18 and maybe if it sort of worked, stuff may be missing Jul 02 12:45:19 should be more or less same. but I'll look into it Jul 02 12:45:27 s/maybe/even/ Jul 02 12:45:27 paulk-desktop meant: and even if it sort of worked, stuff may be missing Jul 02 12:45:47 I'll look for sgs2 Jul 02 12:47:23 paulk-desktop: here is the logcat http://pastebin.com/KcpcPnvg Jul 02 12:48:19 ok, we'll need to sort it out Jul 02 12:48:31 btw your devices are GSM, not CDMA, aren't they? Jul 02 12:48:35 paulk-desktop: yes Jul 02 12:48:52 (cause there is zero support for CDMA in samsung-ril and libsamsung-ipc) Jul 02 12:49:38 paulk-desktop: I'd recommend that you update S2 to ICS because of the new kernel. And because we'll need to update some defines in ril for ics (some structures were renamed) Jul 02 12:50:17 okay, but replicant will probably be gingerbread for galaxy s2 Jul 02 12:50:39 I'll find a way to have both supported Jul 02 12:50:41 paulk-desktop: then we'll have to make sure gingerbread works with ics kernel 3.X Jul 02 12:51:18 or I can write other code, based on yours, to work with gingerbread kernel transport interfaces Jul 02 12:51:28 and we integrate both Jul 02 12:51:34 that shouldn't be so hard Jul 02 12:51:53 paulk-desktop: that's not a good idea. why duplicate code. I think the best solution (in case 3.X kernel doesn't work) is to backport modemctl Jul 02 12:52:10 that's doable too Jul 02 12:52:35 paulk-desktop: does ICS work without opengl es blob? with software only? Jul 02 12:52:49 I only tried on nexus s Jul 02 12:52:54 and it was worse than gingerbread Jul 02 12:53:14 but maybe with some work to have adapted gralloc and hwcomposer modules, it could work as well Jul 02 12:54:49 I have to leave for a couple of hours Jul 02 12:54:50 well, we (i mean, here, at our work project) can accept the proprietary blobs except the bootloader and radio-related things. we can probably add some hacks to the kernel MMU to restrict its access Jul 02 12:54:59 paulk-desktop: ok, have fun. see you later then Jul 02 12:55:11 ok Jul 02 14:10:38 a colleague just tells me that german providers now check *each* IMEI for valid certified device class on book-in Jul 02 14:11:22 I wonder "who" the FR IMEIs are pretending to be Jul 02 14:11:53 book-in? Jul 02 14:12:36 the suggestion of my colleague been all IMEI worldwide are registered by ICAN or whatever. Anybody got a pointer? Jul 02 14:12:50 Apsu: sign-in to net Jul 02 14:13:35 we finally might *need* a IMEI changer for GTA02 ;-P Jul 02 14:14:10 any reports of FR not booking in anymore? Jul 02 14:14:50 when you mean check, you mean check and refuse, or check and signal to someone to look at it ? Jul 02 14:14:59 wpwrak: any historical facts on that topic from your 'archive'? Jul 02 14:15:16 refuse Jul 02 14:15:54 I recall the first ffew 100 GTAxx had IMEI 0000000000 XP Jul 02 14:16:19 or some few at least Jul 02 14:30:28 wpwrak: ping Jul 02 14:33:17 On my Freerunner the IMEI Manufacturer Code is 01, i.e. AEG I think. Jul 02 14:35:31 could you give me whole imei please? Jul 02 14:35:53 *06# ;-) Jul 02 14:41:19 DocScrutinizer: i know nothing of how IMEIs are assigned, sorry Jul 02 14:41:58 do you recall which one range OM used for gta02? Jul 02 14:44:55 https://terminal-profile.osmocom.org/ Jul 02 14:45:47 Openmoko GTA02 354651-01-xxxxxx-x Jul 02 14:46:58 My Freerunner has an IMEI beginning like that printed under the battery. Jul 02 14:48:40 15 numbers long, so that is some old ime assigned prior 2004. Jul 02 14:48:53 s/ime/imei/ Jul 02 14:48:54 Apsu meant: 15 numbers long, so that is some old imei assigned prior 2004. Jul 02 14:50:07 DocScrutinizer: neither that. i don't remember hearing the purchase/registration of an IMEI code point mentioned. if OM didn't have its own codepoint (and if IMEIs are actually organized that way), then it may have inherited some from FIC. but as is said, i don't know much about the IMEIs. Jul 02 14:50:26 why? wiki says new past2004 are also 15 digits Jul 02 14:51:08 wpwrak: anyway thanks Jul 02 14:52:32 but yeah, the format suggests 6digit TAC Jul 02 14:52:42 so pre-2002 Jul 02 14:56:24 Hmm... You are right: both old and new IMEIs are 15 digits long. Jul 02 15:01:27 http://www.mulliner.org/tacdb/feed/babt35_clean.csv has a raw: "35465100,HTC,ST20F, T-Mobile - SDA Music,,," Jul 02 15:01:53 lol Jul 02 15:02:39 freesmartphone.org: 03morphis 07cornucopia * r564e3406f177 10/fsogsmd/src/ (lib/mediator.vala plugins/dbus_service/gsm_call_service.vala): fsogsmd: lib: add mediators for call conference and joining calls Jul 02 15:02:41 freesmartphone.org: 03morphis 07cornucopia * re03cb07d04c5 10/fsogsmd/src/ (4 files in 4 dirs): fsogsmd: lib: extend our call manager for conference/join support Jul 02 15:02:41 freesmartphone.org: 03morphis 07cornucopia * r5cd36d1aed71 10/fsogsmd/src/lib/at/ (atcallmediators.vala atmediatorhelpers.vala): fsogsmd: lib: at: implement call mediators for conference/join Jul 02 15:02:41 freesmartphone.org: 03morphis 07cornucopia * r168f9dbc4daf 10/fsogsmd/src/ (5 files in 4 dirs): fsogsmd: lib: activating a conference call needs the id of a call Jul 02 15:02:42 freesmartphone.org: 03morphis 07cornucopia * r953129ab9883 10/fsogsmd/src/lib/ (at/atcall.vala call.vala): fsogsmd: lib: at: move call id validation to our base class Jul 02 15:02:43 freesmartphone.org: 03morphis 07cornucopia * r34dfbe398e1d 10/fsogsmd/src/lib/at/atcall.vala: fsogsmd: lib: at: implement activating conference calls Jul 02 16:33:00 paulk-desktop: ping Jul 02 16:33:38 Alex[sp3dev], pong Jul 02 16:33:45 paulk-desktop: still busy? Jul 02 16:34:16 nope, I'm good to test on galaxy s2 Jul 02 16:34:46 so I'll flash CM9 Jul 02 16:34:51 fine. can you build against ics or do you want me to build the ril? Jul 02 16:35:15 mhh, I don't really have an ICS tree Jul 02 16:35:21 btw. morphis's ril should support runtime board detection if I edit the makefile, right? Jul 02 16:35:25 but maybe GB binaries will work? Jul 02 16:35:43 Alex[sp3dev], it's done at libsamsung-ipc level Jul 02 16:35:51 paulk-desktop: they should. but you'll still have to flash ics and kernel 3.x Jul 02 16:35:59 good Jul 02 16:36:11 basically, we designed things to be device-specific at libsamsung-ipc and to be generic on upper layer (samsung-ril or fsogsmd) Jul 02 16:36:41 and runtime detection is made looking at /proc/cpuinfo Jul 02 16:36:53 yes, saw that. but it is ifdeffed, though Jul 02 16:37:16 right, you can either have it set at build or at runtime Jul 02 16:37:28 (it being the device to use) Jul 02 16:37:53 so on Android we make it at build time but when you go with SHR, you'll have to make it work at run time Jul 02 16:44:11 nearly done downloading, I'll build the latest heimdall… Jul 02 16:47:20 i've recompiled libsamsung-ipc with runtime detection and am trying to get it running on sgs2 Jul 02 16:50:14 mhh now I can't find a working recovery kernel Jul 02 16:51:42 working as in what Jul 02 16:51:48 as in "boots" Jul 02 16:51:56 Alex[sp3dev]: I never wrote a ril, just libsamsung-ipc Jul 02 16:52:21 morphis: huh? did I say you did? Jul 02 16:52:32 " btw. morphis's ril should support runtime board detection if I edit the makefile, right?" Jul 02 16:52:42 ah, ok. i meant libsamsung-ipc Jul 02 16:53:26 ok :) Jul 02 16:54:00 let me take a look Jul 02 16:54:20 I didn't touched libsamsung-ipc for some time Jul 02 16:54:53 paulk-desktop: btw. do you mind if I am moving libsamsung-ipc from github to git.freesmartphone.org? Jul 02 16:55:11 morphis, go ahead, as long as I can still push to it :) Jul 02 16:56:01 Alex[sp3dev], where is your libsamsung-ipc code already? Jul 02 16:56:16 paulk-desktop: @my github Jul 02 16:56:22 https://github.com/astarasikov/libsamsung-ipc Jul 02 16:56:31 thanks Jul 02 16:56:43 Alex[sp3dev]: tell if I should merge your changes in Jul 02 16:56:49 paulk-desktop: there's another project at my github, the standalone loader Jul 02 16:57:00 yes, I have seen it already Jul 02 16:57:13 morphis: not yet. we need to implement rfs callbacks and check runtime detection. I'll submit a pull request this week Jul 02 16:57:22 Alex[sp3dev]: ok Jul 02 16:57:39 Alex[sp3dev]: but please send it to morphis at gravedo.de Jul 02 16:58:06 or better to smartphones-userland@linuxtogo.org Jul 02 16:58:28 paulk-desktop: ok, I will move it to git.freesmartphone.org tomorrow Jul 02 16:58:33 ok Jul 02 16:58:44 paulk-desktop: can you send me your ssh pub key again so I can include it there too? Jul 02 16:59:21 http://download.paulk.fr/fso-shr.pub Jul 02 17:00:18 tell me when you're good so that I can remove it from there Jul 02 17:02:47 paulk-desktop: done Jul 02 17:02:59 thanks Jul 02 17:04:16 Alex[sp3dev], CM9 installing Jul 02 17:04:38 paulk-desktop: nice. libsamsung-ipc fails at runtime detection via cpuinfo, I'm debugging Jul 02 17:05:07 booting, fine :) Jul 02 17:05:27 it's the table in ipc_devices.c Jul 02 17:05:54 and the 2nd string is the one that should be in /proc/cpuinfo Jul 02 17:05:55 paulk-desktop: yes, I see that. just give me a couple minutes Jul 02 17:06:12 ok Jul 02 17:06:13 here it's SMDK4210 Jul 02 17:06:16 yes Jul 02 17:07:57 oh i see my fuckup Jul 02 17:08:09 the lookup function expects board_name to be lowercase Jul 02 17:10:42 we should use strcasestr Jul 02 17:10:55 instead of strstr to compare the strings Jul 02 17:11:22 can you make that change? Jul 02 17:11:44 paulk-desktop: I've already lowercased the board names and pushed it Jul 02 17:11:53 ok, nevermind then Jul 02 17:14:05 have you pushed all your work? Jul 02 17:14:07 I'll build now Jul 02 17:14:30 paulk-desktop: I've built it, rebooting. will upload the binaries Jul 02 17:14:47 it would be nice if it works on both galaxies with runtime detection Jul 02 17:15:13 yeah, just add another line with the other board name Jul 02 17:15:57 somehow it doesn't work. either ipc just fails on s2 or something else. will try the same binary on nexus Jul 02 17:16:14 now that firmware loader is done it should be easy Jul 02 17:17:28 ok, it works on nexus. need to strace then Jul 02 17:21:58 paulk-desktop: so, here is what is left. [fix i9100, implement RFS callbacks, implement nvdata checking, implement nvdata loading from emmc (just a hook to the fwloader)]. I tried loading nvdata from emmc instead of radio image and it failed on nexus so I gave up for today Jul 02 17:23:08 ok Jul 02 17:23:21 I'll have it running and then do RFS and nv_data Jul 02 17:23:37 http://www.mediafire.com/?n8cvoak6adyyyzc Jul 02 17:23:41 paulk-desktop: here are the binaries Jul 02 17:23:43 then I'll probably look at data Jul 02 17:23:44 thanks Jul 02 17:23:45 works on i9100 Jul 02 17:25:08 morphis, Alex[sp3dev], btw we have some retry-loop-on-failure code for some devices (aries and crespo) but I wonder if it's really a good idea to have them at libsamsung-ipc level Jul 02 17:25:48 paulk-desktop: yes, it is a good idea. but you should better remove the retry from crespo and aries and move it to the ipc code which calls bootstrap Jul 02 17:25:55 samsung-ril will already try like 10 times to boot when it fails, so if libsamsung-ipc tries 10 times before returning, we'll end up trying 100 times… Jul 02 17:26:09 yes, that was my thought Jul 02 17:26:17 keeping it all in samsung-ril and fsogsmd Jul 02 17:26:18 paulk-desktop: I would suggest that bootstrap should both reset the device,enable power and load firmware and the power off callback should disable power. poweron callback will be a NOP Jul 02 17:27:55 actually on nexus s, the power on function is actually "modem start" Jul 02 17:28:03 it's not really to power it on Jul 02 17:28:59 paulk-desktop: so. I've accepted the voice call on i9100 http://rghost.ru/private/38993954/39247223973cffe87b6290ac7affe9b0.view Jul 02 17:29:17 nice :) Jul 02 17:29:33 first call ever with the free RIL? Jul 02 17:29:53 paulk-desktop: yes, on the i9100. on the nexus the first call was on friday night Jul 02 17:30:00 good :) Jul 02 17:30:04 still no sound though? Jul 02 17:30:11 need to check Jul 02 17:31:53 to show off I could make a call from nexus to s2 ;) Jul 02 17:34:01 paulk-desktop: no sound ;) Jul 02 17:36:07 ok, the RIL works here Jul 02 17:36:13 I'll try with a SIM to see how calls go Jul 02 17:36:42 D/gps_sirf( 3073): ### Samsung U1-ICS, CSR SiRF Software 2.0, B16 2012,01,17 ### . I see that in logcat on S2.. Which makes me think it ain't bcm47511 Jul 02 17:37:21 yay! SIRF is good Jul 02 17:37:51 paulk-desktop: so you just used my binaries? Jul 02 17:37:56 yep Jul 02 17:37:59 ok Jul 02 17:38:01 but I'll build now Jul 02 17:38:12 btw ICS has restricted adb privileges on CM9 Jul 02 17:38:18 that's quite a pain in the ass Jul 02 17:38:34 paulk-desktop: you can change it in settings Jul 02 17:38:53 developer options -> root access Jul 02 17:39:23 I set it to both apps and adb Jul 02 17:39:28 but still adb shell shows as user Jul 02 17:39:30 and I have to use su Jul 02 17:39:55 i see Jul 02 17:40:39 and so adb push to /system fails Jul 02 17:41:00 paulk-desktop: I'm done with ril for today and you can play with nvdata and rfs. ping me if you have news, I'll be reading some books and coding some dsp algos Jul 02 17:41:12 paulk-desktop: push to /data and do "su -c \ Jul 02 17:41:23 su -c \" cp /data/foo /system\" Jul 02 17:41:34 that's what I did Jul 02 17:41:48 Alex[sp3dev], ok, have fun! Jul 02 17:42:03 http://pastebin.com/2PjTnfDw I used this hack during testing fwloader ;) Jul 02 17:46:09 huh Jul 02 17:46:10 thanks Jul 02 17:46:40 apparently my libsamsung-ril-client isn't compatible with libsecril-client.so (the audio module fails) Jul 02 17:46:51 but as I plan to replace that audio module, it'll be fine Jul 02 17:47:27 paulk-desktop: I kindly let you fix that ;). If you implement RFS, either make a pull request to github or rebase it yourself into a single patch and push to morphis Jul 02 17:48:08 Alex[sp3dev], last thing, then I stop disturbing you: do you mind if I make the log output look like the output of other devices and also if I rename/reorder some functions to match our coding style? Jul 02 17:48:13 it would be nice to make it binary compatible to the proprietary one. even though audio is free and open on galaxy nexus, we could at least make it so that we don't have to patch sources Jul 02 17:48:25 indeed Jul 02 17:48:58 the only change we do is usually this: https://gitorious.org/replicant/device_samsung_crespo/commit/26ee25777e8263ca1129a189a5e1d2e3ff649dee Jul 02 17:49:06 paulk-desktop: I am not extremely fond of the idea but you may modify the functions in log.h. I just tried to move debugging macros to a separate file so that you can hook them to android log/stdout or whatever Jul 02 17:49:35 ok I'll look Jul 02 17:50:21 paulk-desktop: it might be easier to just fix the makefile to build it under the same name as the proprietary one. you see, it's not a problem to fix the other libraries, but the less changes we make, the easier it is to maintain code and reuse cm code Jul 02 19:42:18 paulk-desktop: ping Jul 02 20:24:37 Alex[sp3dev], pong Jul 02 20:25:03 I have read the code and I have many comments Jul 02 20:25:18 paulk-desktop: you are welcome Jul 02 20:25:28 for instance, I think we should make maguro and galaxys2 two separate devices Jul 02 20:25:57 as they don't share much (there are dedicated boot functions for each, etc) Jul 02 20:25:58 paulk-desktop: why? there are only minor differences, and the ipc protocol is the same Jul 02 20:26:29 I am ready to refactor the code.. but I'd prefer that we have gprs and sound working first Jul 02 20:26:46 ok, sure Jul 02 20:26:59 have you looked into it yet? Jul 02 20:27:12 I did tests with rfs, not working yet Jul 02 20:27:25 for data, I think I know how to proceed Jul 02 20:27:44 and for sound, there are two solutions: Jul 02 20:27:46 have you figured out rfs format? look at the gingerbread s2 kernel or tizen Jul 02 20:28:01 I can't get any message from RFS ATM Jul 02 20:28:21 though I'll strace the non-free RIL to see how it's dealt with Jul 02 20:28:44 so about sound: we can either wait until we have a free module, then it'll be easy to have it working Jul 02 20:29:07 paulk-desktop: which buffer size are you using? It behaves strangely if you leave data in buffer, so you should try reading as much as possible (4096 for safety) and see how many it returns Jul 02 20:29:20 like for ipc_fmt Jul 02 20:29:24 sure, that's what we do for other devices Jul 02 20:29:53 paulk-desktop: you've probably seen that http://pastebin.com/8Ez2qsha . there's the rfs header structure at the end Jul 02 20:30:18 /* RFS IPC Frame */ Jul 02 20:30:21 that's the usual one Jul 02 20:30:27 len, cmd, id Jul 02 20:30:52 so it should just work Jul 02 20:31:44 btw. on galaxy nexus, audio is free and open. so I can make tests if you want to look into it. You can ping me during the day on mon/wed/fri or in the evening on other days (+4 GMT) Jul 02 20:31:58 ok, good Jul 02 20:33:17 any more complaints except splitting s2/nexus? Probably some compilation warnings need looking into, but that's not a priority. We can always clean the code when we have stuff working, but until that it's better to not touch it Jul 02 20:34:42 also, defines should be in spearate headers Jul 02 21:00:51 RFS rx working Jul 02 21:01:20 and TX? Jul 02 21:01:33 doing it now Jul 02 21:01:39 but should work as well Jul 02 21:03:08 could you show the code? I'm so tired today. it's been too hot and stuffy in Moscow. I think I'll go to bed soon Jul 02 21:03:45 do you mind if I paste? Jul 02 21:03:56 I'd like to commit both rx and tx at the same time Jul 02 21:04:04 yes, sure Jul 02 21:04:14 http://pastie.org/4189343 Jul 02 21:04:40 and you must set that function at xmm6260_rfs_ops Jul 02 21:05:49 paulk-desktop: why don't you check the return byte count? Jul 02 21:06:14 the one from client->handlers->read ? Jul 02 21:06:27 yes Jul 02 21:06:39 if (rc < 0) it returns Jul 02 21:06:43 I thought that would be enough Jul 02 21:06:53 no, check that it is at least sizeof(rfs_header) large Jul 02 21:07:08 i'm paranoid about such things Jul 02 21:07:13 ok, good idea Jul 02 21:07:22 I assume it's that large later in the code Jul 02 21:07:26 so you're right to check Jul 02 21:08:15 and also. don't use #ifdef DEBUG in future. it is a good idea to do like in linux - define a macro like pr_debug that will evaluate to "do {} while 0" if DEBUG is not defined. I just avoid ifdefs, because they tend to get stale and broken when no one compiles them ;) Jul 02 21:08:45 anyway, thank you for your help with galaxy s2 and gprs Jul 02 21:08:53 mhh, that's what we decided to use with morphis, but we can discuss that later Jul 02 21:09:04 I'll have gprs working this evening I think Jul 02 21:09:09 I'll let you know tomorrow Jul 02 21:09:25 I think I'll add the missing functions to the ril client so that it works on s2 and nexus. Probably just export some symbols Jul 02 21:09:54 ok Jul 02 21:10:31 Also from the brief glance it looks like proprietary gps and nfc libs want some functions from ril, so I'll probably implement them. Not that I want to have blobs, but having all hardware working is a must. Although non-free gps is very insecure Jul 02 21:11:55 about GPS, I think it's for AGPS Jul 02 21:11:58 to get cellid, etc Jul 02 21:12:19 I have seen gpsd asking the ril for these, yes **** ENDING LOGGING AT Tue Jul 03 02:59:57 2012