> What is your exact use case? Normally in android you should not need to > deploy applications by other (complicated from user point of view) means. Well, as I said before, I started compiling the source code given here: http://svn.pjsip.org/repos/pjproject/branches/projects/android/ I must say that this branch doesn't provide any kind of .apk, so I thought that I could directly execute the application /pjsip-apps/src/pjsua. The make-file will output the binary (not the .apk, though) in /pjsip-apps/bin. I must say that I've tried that application on an Android Emulator, on an Android Rooted Emulator and on a real device that hasn't been rooted yet (ie. cannot perform strace etc.)... Maybe i'll try to root it. Now, I've taken a look at android-ndk-r8b samples, and I've seen that some "native-samples", such as native-activity, doesn't have to carry any kind of .java code, even if they still have AndroidManifest.xml, Android.mk, Applications.mk, which are missing on the http://svn.pjsip.org/repos/pjproject/branches/projects/android/ original source code... Maybe I should recreate the same "file architecture" on the official version? > BTW, this said, the problem is maybe somewhere else. I sincerely hope so... M.P. On Wed, Jul 25, 2012 at 6:02 PM, R?gis Montoya <r3gis.3r at gmail.com> wrote: > > Hi, > > CSipSimple supports two backends for audio. One is the OpenSLES one (this > one has been integrated and probably improved in pjsip android branch) and > the other is the Java-JNI implementation. > OpenSLES is only valid from 2.3 and is my default setting from android 3.0 > (because some manufacturers did crappy things). The jni implementation is > used from android 1.6. > > If you dive in Android source code you'll see that both OpenSL-ES and Java > api relies on the same thing so in terms of performance what you win is the > garbage collection cycles and the fact there is no conversion to java arrays > (which doesn't have such a big impact). Other benefit of OpenSLES is that it > doesn't needs to keep a reference on the dalvik jvm and can be managed > directly in C independently of context. > If you want to try both options in CSipSimple, you can check the difference > by changing a simple option. > As OpenSLES (for now) rely on android media server, you are still > constrained by same things than a regular app. > Google engineers announced that they will rework libwilhelm (the one that > implements opensl-es api) to use lower api libraries (such as > libstagefright), but that's not yet the case, even in source code of 4.1... > So we are waiting for real time audio... > > In terms of permission you are constrained to the same things than someone > using the android audio server from a java app. Besides, normally even at > lower layer they will take care to not open a door to apps. It's probably > very easy to do by allowing rights to open audio device only to members of > one group. > > Normally strictly talking about unix permission it should not be a problem > since you are root when running your binary. However... I'm not fully sure > that the android media server doesn't takes an exclusive lock on > libstagefright. In which case your app running as binary would not be > allowed to do anything. That's just an idea, as I said in my first mail I > never tried that because it doesn't make many sense to have a pure binary > app. > The correct way to have a non platform app on android is to run it from an > android application. And android application are apk. Apk are in charge to > declare many things including permissions. And these permissions will indeed > be applied to the sandboxed user. > > What you could do however, if you really want to go the way of the binary, > is to package your binary inside your apk and to call it from a very simple > java app with an execute. > > What is your exact use case? Normally in android you should not need to > deploy applications by other (complicated from user point of view) means. > > BTW, this said, the problem is maybe somewhere else. I'll test soon apjsua > so if I encounter same problem maybe I'll be able to provide more feedback > :). > > > On 25/07/2012 17:07, Monty Python wrote: >> >> Thanks a lot for your reply!! >> >> I tried to launch the same application on a not-rooted device, and it >> gave me, unfortunately, the same result. >> >> I have also a question about CSipSimple. I've seen that that fork >> required to use a java support, maybe in order to avoid the problems >> with Android permissions', manifest files and so on. Anyway I am still >> perplexed, because NDK has the audio support through OpenSLES, and it >> gives also the support to sockets: so I initially thought that, >> working on C language, I should operate on Kernel level bypassing all >> the "Android Security Level". >> >> Next, I've read this article on source.android: >> >> >> http://source.android.com/tech/security/#android-platform-security-architecture >> >> I'm not misunderstanding, all the applications are run inside a >> "Application Sandbox", that it works in Kernel Mode. So, any attempt >> to run a "particular application" with audio, sockets ecc. is vain. Is >> there any way to bypass this "application sandbox"? Must I really use >> .apks in order to use Manifest permissions? Thanks again. >> >> On Tue, Jul 24, 2012 at 8:53 PM, Monty Python >> <pythonexperiment at gmail.com> wrote: >>> >>> Hi, >>> >>> Just some ideas, (I never had this issue) but ... : >>> >>> It could be something with unix rights : on android, applications get >>> permissions from their manifest and when android os creates the user of >>> the >>> sandbox for the app it adds this users to relevant groups. >>> Because could also be something with the media server of android os that >>> doesn't allow to use audio for not real android apps. So rather than >>> command line launch you should probably try to use the sample application >>> apk and install it properly. With "adb install your_apk_name.apk" .+ >>> >>> Another possible root cause is that the emulator doesn't always play well >>> with the audio layer. First it's absolutely not performant, and second >>> sometimes it doesn't detects at all the host audio hardware. I would >>> advise >>> you to use a real device if you have one. It's way better to develop for >>> the call / audio purpose. (However emulator is a solution to develop for >>> the user interface part). >>> >>> Best regards. >>>> >>>> 2012/7/25 Monty Python <pythonexperiment at gmail.com> >>>> I've also obtained the strace of the pjsua process, hoping that it >>>> could expound my problem. Thanks again. >>>> stat64("/vendor/lib/libstagefright_amrnb_common.so", 0xbea8f138) = -1 >>>> ENOENT (No such file or directory) >>>> stat64("/system/lib/libstagefright_amrnb_common.so", >>>> {st_mode=S_IFREG|0644, st_size=50760, ...}) = 0 >>>> open("/system/lib/libstagefright_amrnb_common.so", O_RDONLY|O_LARGEFILE) >>>> = >>>> 3 >>>> lseek(3, 0, SEEK_SET) = 0 >>>> read(3, "\177ELF\1\1\1\0\0\0\0\0\0\0\0\0\3\0(\0\1\0\0\0\0\0\0\000"..., >>>> 4096) = 4096 >>>> lseek(3, -8, SEEK_END) = 50752 >>>> read(3, "\1\0\0\0\0\0\0\0", 8) = 8 >>>> mmap2(NULL, 53248, PROT_READ|PROT_EXEC, MAP_PRIVATE|MAP_ANONYMOUS, -1, >>>> 0) = 0x40b0e000 >>>> mmap2(0x40b0e000, 46908, PROT_READ|PROT_EXEC, MAP_PRIVATE|MAP_FIXED, >>>> 3, 0) = 0x40b0e000 >>>> mprotect(0x40b0e000, 49152, PROT_READ|PROT_WRITE|PROT_EXEC) = 0 >>>> mmap2(0x40b1a000, 444, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_FIXED, 3, >>>> 0xc) = 0x40b1a000 >>>> close(3) = 0 >>>> mprotect(0x40b0e000, 49152, PROT_READ|PROT_EXEC) = 0 >>>> stat64("/vendor/lib/libstagefright_enc_common.so", 0xbea8f138) = -1 >>>> ENOENT (No such file or directory) >>>> stat64("/system/lib/libstagefright_enc_common.so", >>>> {st_mode=S_IFREG|0644, st_size=5420, ...}) = 0 >>>> open("/system/lib/libstagefright_enc_common.so", O_RDONLY|O_LARGEFILE) = >>>> 3 >>>> lseek(3, 0, SEEK_SET) = 0 >>>> read(3, "\177ELF\1\1\1\0\0\0\0\0\0\0\0\0\3\0(\0\1\0\0\0\0\0\0\000"..., >>>> 4096) = 4096 >>>> lseek(3, -8, SEEK_END) = 5412 >>>> read(3, "\1\0\0\0\0\0\0\0", 8) = 8 >>>> mmap2(NULL, 8192, PROT_READ|PROT_EXEC, MAP_PRIVATE|MAP_ANONYMOUS, -1, >>>> 0) = 0x40b1b000 >>>> mmap2(0x40b1b000, 1700, PROT_READ|PROT_EXEC, MAP_PRIVATE|MAP_FIXED, 3, >>>> 0) = 0x40b1b000 >>>> mprotect(0x40b1b000, 4096, PROT_READ|PROT_WRITE|PROT_EXEC) = 0 >>>> mmap2(0x40b1c000, 260, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_FIXED, 3, >>>> 0x1) = 0x40b1c000 >>>> close(3) = 0 >>>> mprotect(0x40b1b000, 4096, PROT_READ|PROT_EXEC) = 0 >>>> stat64("/vendor/lib/libstagefright_avc_common.so", 0xbea8f138) = -1 >>>> ENOENT (No such file or directory) >>>> stat64("/system/lib/libstagefright_avc_common.so", >>>> {st_mode=S_IFREG|0644, st_size=21848, ...}) = 0 >>>> open("/system/lib/libstagefright_avc_common.so", O_RDONLY|O_LARGEFILE) = >>>> 3 >>>> lseek(3, 0, SEEK_SET) = 0 >>>> read(3, "\177ELF\1\1\1\0\0\0\0\0\0\0\0\0\3\0(\0\1\0\0\0\0\0\0\000"..., >>>> 4096) = 4096 >>>> lseek(3, -8, SEEK_END) = 21840 >>>> read(3, "\1\0\0\0\0\0\0\0", 8) = 8 >>>> mmap2(NULL, 24576, PROT_READ|PROT_EXEC, MAP_PRIVATE|MAP_ANONYMOUS, -1, >>>> 0) = 0x40b1d000 >>>> mmap2(0x40b1d000, 19024, PROT_READ|PROT_EXEC, MAP_PRIVATE|MAP_FIXED, >>>> 3, 0) = 0x40b1d000 >>>> mprotect(0x40b1d000, 20480, PROT_READ|PROT_WRITE|PROT_EXEC) = 0 >>>> mmap2(0x40b22000, 256, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_FIXED, 3, >>>> 0x5) = 0x40b22000 >>>> close(3) = 0 >>>> mprotect(0x40b1d000, 20480, PROT_READ|PROT_EXEC) = 0 >>>> mprotect(0x40574000, 1159168, PROT_READ|PROT_EXEC) = 0 >>>> mprotect(0x40106000, 4096, PROT_READ|PROT_WRITE) = 0 >>>> mprotect(0x40106000, 4096, PROT_READ) = 0 >>>> mprotect(0x40106000, 4096, PROT_READ|PROT_WRITE) = 0 >>>> mprotect(0x40106000, 4096, PROT_READ) = 0 >>>> mprotect(0x40106000, 4096, PROT_READ|PROT_WRITE) = 0 >>>> mprotect(0x40106000, 4096, PROT_READ) = 0 >>>> mprotect(0x40106000, 4096, PROT_READ|PROT_WRITE) = 0 >>>> mprotect(0x40106000, 4096, PROT_READ) = 0 >>>> mprotect(0x40106000, 4096, PROT_READ|PROT_WRITE) = 0 >>>> mprotect(0x40106000, 4096, PROT_READ) = 0 >>>> mprotect(0x40106000, 4096, PROT_READ|PROT_WRITE) = 0 >>>> mprotect(0x40106000, 4096, PROT_READ) = 0 >>>> mprotect(0x40071000, 155648, PROT_READ|PROT_EXEC) = 0 >>>> mprotect(0x40106000, 4096, PROT_READ|PROT_WRITE) = 0 >>>> mprotect(0x40106000, 4096, PROT_READ) = 0 >>>> mprotect(0x40106000, 4096, PROT_READ|PROT_WRITE) = 0 >>>> mprotect(0x40106000, 4096, PROT_READ) = 0 >>>> mprotect(0x4006e000, 8192, PROT_READ|PROT_EXEC) = 0 >>>> mprotect(0x8000, 1933312, PROT_READ|PROT_EXEC) = 0 >>>> sigaction(49264, {0xb00144c4, [], SA_RESTART}, {0xb00144c4, [], >>>> SA_RESTART}, 0) = -1 EINVAL (Invalid argument) >>>> --- SIGSEGV (Segmentation fault) @ 0 (f0) --- >>>> open("/dev/log/main", O_WRONLY|O_LARGEFILE) = 3 >>>> writev(3, [{"\7", 1}, {"libc\0", 5}, {"Fatal signal 11 (SIGSEGV) at >>>> 0x0"..., 49}], 3) = 55 >>>> SYS_224(0x37, 0xbea8f850, 0x3, 0x38) = 465 >>>> socket(PF_UNIX, SOCK_STREAM, 0) = 4 >>>> connect(4, {sa_family=AF_UNIX, path=@android:debuggerd}, 20) = 0 >>>> write(4, "?\1\0\0", 4) = 4 >>>> read(4, 0xbea8f888, 1) = ? ERESTARTSYS (To be >>>> restarted) >>>> --- SIGCONT (Continue) @ 0 (0) --- >>>> read(4, "?", 1) = 1 >>>> close(4) = 0 >>>> sigaction(SIGSEGV, {SIG_DFL}, {0xb0005b29, [], SA_RESTART|SA_SIGINFO}, >>>> 0) >>>> = 0 >>>> rt_sigreturn(0) = -1 EPERM (Operation not >>>> permitted) >>>> --- SIGSEGV (Segmentation fault) @ 0 (f0) --- >>>> +++ killed by SIGSEGV +++ >>>> >>>> On Tue, Jul 24, 2012 at 8:53 PM, Monty Python >>>> <pythonexperiment at gmail.com> wrote: >>>>> >>>>> Greetings to everyone! >>>>> >>>>> I'm trying to compile pjsua using the following branch: >>>>> http://svn.pjsip.org/repos/pjproject/branches/projects/android/. I've >>>>> tried to do a push (adb push pjsua /data/local/) to my android-sdk >>>>> emulator but, when I've tried to execute it via adb shell, the android >>>>> logcat gave me the following SIGFAULT error: where am I wrong? Thanks >>>>> in advance. >>>>> >>>>> F/libc ( 464): Fatal signal 11 (SIGSEGV) at 0x000000f0 (code=1) >>>>> I/DEBUG ( 33): *** *** *** *** *** *** *** *** *** *** *** *** *** >>>>> *** *** *** >>>>> I/DEBUG ( 33): Build fingerprint: >>>>> 'generic/sdk/generic:4.0.3/MR1/237985:eng/test-keys' >>>>> I/DEBUG ( 33): pid: 464, tid: 464 >>> >>>>> ./pjsua-arm-unknown-linux-androideabi <<< >>>>> I/DEBUG ( 33): signal 11 (SIGSEGV), code 1 (SEGV_MAPERR), fault >>>>> addr 000000f0 >>>>> I/DEBUG ( 33): r0 ffffffff r1 beef7c30 r2 beef7c30 r3 ffffffff >>>>> I/DEBUG ( 33): r4 00000000 r5 00000000 r6 00000000 r7 00000000 >>>>> I/DEBUG ( 33): r8 00000000 r9 00000000 10 00000000 fp 00000000 >>>>> I/DEBUG ( 33): ip 4003c4b9 sp beef7c60 lr 4003c4b1 pc b000469a >>>>> cpsr 00000030 >>>>> I/DEBUG ( 33): d0 00000000bd6bc8e3 d1 0000000000000000 >>>>> I/DEBUG ( 33): d2 0000000000000000 d3 0000000000000000 >>>>> I/DEBUG ( 33): d4 0000000000000000 d5 41c8f0a46e800000 >>>>> I/DEBUG ( 33): d6 3f50624dd2f1a9fc d7 c18af9670cce266f >>>>> I/DEBUG ( 33): d8 0000000000000000 d9 0000000000000000 >>>>> I/DEBUG ( 33): d10 0000000000000000 d11 0000000000000000 >>>>> I/DEBUG ( 33): d12 0000000000000000 d13 0000000000000000 >>>>> I/DEBUG ( 33): d14 0000000000000000 d15 0000000000000000 >>>>> I/DEBUG ( 33): scr 00000010 >>>>> I/DEBUG ( 33): >>>>> I/DEBUG ( 33): #00 pc b000469a /system/bin/linker >>>>> I/DEBUG ( 33): #01 pc 000264ac /system/lib/libc.so >>>> >>>> (__set_errno) >>>>> >>>>> I/DEBUG ( 33): >>>>> I/DEBUG ( 33): code around pc: >>>>> I/DEBUG ( 33): b0004678 95004840 44784a40 4d414b40 447b447a >>>>> @H.. at JxD at KAMzD{D >>>>> I/DEBUG ( 33): b0004688 682d447d f44f9103 95017140 f0009402 >>>>> }D-h..O. at q...... >>>>> I/DEBUG ( 33): b0004698 f8d4ff67 b10330f0 f8d44798 b17000e0 >>>>> g....0...G....p. >>>>> I/DEBUG ( 33): b00046a8 10e4f8d4 f7ff2200 2000f9b5 f8d4e007 >>>>> ....."..... .... >>>>> I/DEBUG ( 33): b00046b8 f04f20a8 f04230ff f8c40102 b00710a8 . >>>>> O..0B......... >>>>> I/DEBUG ( 33): >>>>> I/DEBUG ( 33): code around lr: >>>>> I/DEBUG ( 33): 4003c490 f240b507 9300736c 33fff04f 466b9301 >>>>> .. at .ls..O..3..kF >>>>> I/DEBUG ( 33): 4003c4a0 fd80f7ff bf00bd0e 4604b510 fe90f7ec >>>>> ...........F.... >>>>> I/DEBUG ( 33): 4003c4b0 f04f6004 bd1030ff 0ffff110 db02b510 >>>>> .`O..0.......... >>>>> I/DEBUG ( 33): 4003c4c0 f7ff4240 bd10fff1 48214603 4478b5f0 >>>>> @B.......F!H..xD >>>>> I/DEBUG ( 33): 4003c4d0 b0976800 68022150 4620ac01 92154e1d >>>>> .h..P!.h.. F.N.. >>>>> I/DEBUG ( 33): >>>>> I/DEBUG ( 33): stack: >>>>> I/DEBUG ( 33): beef7c20 00000000 >>>>> I/DEBUG ( 33): beef7c24 4003c4c7 /system/lib/libc.so >>>>> I/DEBUG ( 33): beef7c28 00000000 >>>>> I/DEBUG ( 33): beef7c2c 4002f477 /system/lib/libc.so >>>>> I/DEBUG ( 33): beef7c30 b00144c4 >>>>> I/DEBUG ( 33): beef7c34 00000000 >>>>> I/DEBUG ( 33): beef7c38 10000000 >>>>> I/DEBUG ( 33): beef7c3c 00000000 >>>>> I/DEBUG ( 33): beef7c40 00000000 >>>>> I/DEBUG ( 33): beef7c44 4002f49b /system/lib/libc.so >>>>> I/DEBUG ( 33): beef7c48 00000000 >>>>> I/DEBUG ( 33): beef7c4c 0000c090 >>>>> /data/local/pjsua-arm-unknown-linux-androideabi >>>>> I/DEBUG ( 33): beef7c50 b00144c4 >>>>> I/DEBUG ( 33): beef7c54 0000c070 >>>>> /data/local/pjsua-arm-unknown-linux-androideabi >>>>> I/DEBUG ( 33): beef7c58 df0027ad >>>>> I/DEBUG ( 33): beef7c5c 00000000 >>>>> I/DEBUG ( 33): #01 beef7c60 00000001 >>>>> I/DEBUG ( 33): beef7c64 beef7d47 [stack] >>>>> I/DEBUG ( 33): beef7c68 00000000 >>>>> I/DEBUG ( 33): beef7c6c beef7d6d [stack] >>>>> I/DEBUG ( 33): beef7c70 beef7d82 [stack] >>>>> I/DEBUG ( 33): beef7c74 beef7d92 [stack] >>>>> I/DEBUG ( 33): beef7c78 beef7dba [stack] >>>>> I/DEBUG ( 33): beef7c7c beef7df7 [stack] >>>>> I/DEBUG ( 33): beef7c80 beef7e10 [stack] >>>>> I/DEBUG ( 33): beef7c84 beef7e2a [stack] >>>>> I/DEBUG ( 33): beef7c88 beef7f55 [stack] >>>>> I/DEBUG ( 33): beef7c8c beef7f68 [stack] >>>>> I/DEBUG ( 33): beef7c90 beef7f83 [stack] >>>>> I/DEBUG ( 33): beef7c94 beef7fa0 [stack] >>>>> I/DEBUG ( 33): beef7c98 beef7fb3 [stack] >>>>> I/DEBUG ( 33): beef7c9c 00000000 >>>>> I/DEBUG ( 33): beef7ca0 00000010 >>>>> I/DEBUG ( 33): beef7ca4 000030d7 >>>>> >>>>> M.P. >>>> >>>> _______________________________________________ >>>> Visit our blog: http://blog.pjsip.org >>>> >>>> pjsip mailing list >>>> pjsip at lists.pjsip.org >>>> http://lists.pjsip.org/mailman/listinfo/pjsip_lists.pjsip.org >>>> >> _______________________________________________ >> Visit our blog: http://blog.pjsip.org >> >> pjsip mailing list >> pjsip at lists.pjsip.org >> http://lists.pjsip.org/mailman/listinfo/pjsip_lists.pjsip.org > > > > _______________________________________________ > Visit our blog: http://blog.pjsip.org > > pjsip mailing list > pjsip at lists.pjsip.org > http://lists.pjsip.org/mailman/listinfo/pjsip_lists.pjsip.org
