Ok. I've got a build up for the iphone using the ipodsound.c method
with the 1.0.3 codebase. I've added a short routine as follows....
PJ_DEF(void) interruptionListenerCallback( void * inClientData,
UInt32 inInterruptionState)
{
//Put something here later.
//Handle stuff like people plugging in headphones etc.
}
PJ_DEF(pj_status_t) ipodsound_device_init() {
OSStatus status;
status = AudioSessionInitialize (NULL, NULL,
interruptionListenerCallback, NULL );
PJ_LOG(1, (THIS_FILE, "AudioSessionInitialise status %d", status));
UInt32 category = kAudioSessionCategory_PlayAndRecord;
status = AudioSessionSetProperty(kAudioSessionProperty_AudioCategory,
sizeof(category), &category);
PJ_LOG(1, (THIS_FILE, "AudioSessionSetProperty status %d", status));
status = AudioSessionSetActive(true);
PJ_LOG(1, (THIS_FILE, "AudioSessionSetActive status %d", status));
return status;
}
that is called before anything, and that seems to solve a problem with
the device audio initialization.
My compile script is as follows, since nobody seems to be uploading
particularly useful ones.
#!/bin/sh
find ./ -name "*.depend" | xargs rm -Rf
find ./ -name "*.o" | xargs rm -Rf
find ./ -name "*.a" | xargs rm -Rf
export DEV=/Developer/Platforms/iPhoneOS.platform/Developer
export SDK=${DEV}/SDKs/iPhoneOS3.0.sdk
pushd ${DEV}/usr/bin
ln -s arm-apple-darwin9-gcc-4.2.1 arm-apple-darwin9-gcc
ln -s arm-apple-darwin9-g++-4.2.1 arm-apple-darwin9-g++
ln -s ranlib arm-apple-darwin9-ranlib
popd
#change the -gstabs+ -Oo to just -O2 (or whatever) to remove xcode
debug compatibility.
export CFLAGS="-gstabs+ -O0 -arch armv6 -isysroot ${SDK} \
-I${SDK}/SDKs/iPhoneOS3.0.sdk/usr/include/gcc/darwin/4.2.1 \
-I/Developer/Platforms/iPhoneOS.platform/Developer/SDKs/
iPhoneOS3.0.sdk/System/Library/Frameworks/ \
-F${SDK}/System/Library/Frameworks"
export LDFLAGS="-O0 -arch armv6 -isysroot${SDK}\
-F${SDK}/System/Library/Frameworks"
export CPP="${DEV}/usr/bin/cpp"
export PATH=${DEV}/usr/bin:${PATH}
unexpand build/rules.mak
make distclean
autoconf aconfigure.ac > aconfigure
./aconfigure --host=arm-apple-darwin9 --disable-speex-aec \
--disable-speex-codec --disable-l16-codec --disable-g722-codec \
--disable-ilbc-codec # 2>&1 | more
unexpand build/rules.mak
echo "Stage #1"
make dep #2>&1 | grep [Ee]rror #uncomment this to filter out cruft
for xcode
echo "Stage #2"
make #2>&1 | grep [Ee]rror #uncomment this to filter out cruft for
xcode.
The instructions are roughly as per the Siphon wiki, but some mangling
might be needed of those instructions for linking.
Note you should comment out the SETJMP line in os_darwinwhateveritis
to get it to not crap itself when linking into your xcode stuff.
There might be some toolchain issues requiring some ln tomfoolery, but
thats all documented on the net. Be aware you might need to copy some
headers around as the sdk is missing a few crucial ones that ARE in
the simulator.
the -gstabs+ stuff lets you get sane output from gdb inside xcode.
If you add the script to xcode as an external
ANYWAY. . . . .
When I make a call, I get voice transmitted to the far end, but not
recieving any audio. After about 3 seconds that voice gets jittery and
then the whole phone locks up hard.
I'm having issues trying to debug this, as I really have no idea how
to get the phone to eject a core dump, the logs really are not useful,
the thing just refuses to cooperate with instruments or shark, and
thus I really don't know how to find out whats going haywire.
I'll note two other things. SipPhone works beautifully. Siphon seems
to have no sound. Perhaps the Siphon guys should look into that code I
posted up the top. btw It'd be cool if you guys could post your pjsip
directory with your modifications and build script. That would be
*very* instructive.
With the script above, the phone DOES build in such a way that GDB can
interrogate the library on the iphone, but I'm still trying to divine
a strategy to get the phone to chill out when it gets wonky and let me
try and work out whats going wrong.
I note that the sinewave test indicates audio out DOES work, it
suggests thus that with that ammendment I posted the ipodsound driver
does work, so I *think* the ipodsound driver works.
I'm getting this when I run the pjsua code (The Its a trap! is a
deactivated breakpoint I've added with a log message to indicate where
it would normally fire)
22:12:48.715 ipodsound.c AudioSessionInitialise status 0
22:12:48.737 ipodsound.c AudioSessionSetProperty status 0
22:12:48.848 ipodsound.c AudioSessionSetActive status 0
22:12:48.934 os_core_unix.c pjlib 1.0.3 for POSIX initialized
22:12:49.036 sip_endpoint.c Creating endpoint instance...
22:12:49.047 pjlib select() I/O Queue created (0x831614)
22:12:49.047 sip_endpoint.c Module "mod-msg-print" registered
22:12:49.048 sip_transport. Transport manager created.
22:12:49.051 sip_endpoint.c Module "mod-tsx-layer" registered
22:12:49.051 sip_endpoint.c Module "mod-stateful-util" registered
22:12:49.052 sip_endpoint.c Module "mod-ua" registered
22:12:49.054 sip_endpoint.c Module "mod-100rel" registered
22:12:49.055 sip_endpoint.c Module "mod-pjsua" registered
22:12:49.056 sip_endpoint.c Module "mod-invite" registered
22:12:49.061 pjlib select() I/O Queue created (0x830214)
22:12:49.067 sip_endpoint.c Module "mod-evsub" registered
22:12:49.069 sip_endpoint.c Module "mod-presence" registered
22:12:49.070 sip_endpoint.c Module "mod-refer" registered
22:12:49.070 sip_endpoint.c Module "mod-pjsua-pres" registered
22:12:49.071 sip_endpoint.c Module "mod-pjsua-im" registered
22:12:49.072 sip_endpoint.c Module "mod-pjsua-options" registered
22:12:49.073 pjsua_core.c 1 SIP worker threads created
22:12:49.073 pjsua_core.c pjsua version 1.0.3 for arm-apple-
darwin9 initialized
22:12:49.094 pjsua_core.c SIP UDP socket reachable at 10.1.1.4:5060
22:12:49.096 udp0x855a00 SIP UDP transport started, published
address is 10.1.1.4:5060
22:12:49.109 pjsua_media.c RTP socket reachable at 10.1.1.4:4000
22:12:49.113 pjsua_media.c RTCP socket reachable at 10.1.1.4:4001
22:12:49.123 pjsua_media.c RTP socket reachable at 10.1.1.4:4002
22:12:49.126 pjsua_media.c RTCP socket reachable at 10.1.1.4:4003
22:12:49.134 pjsua_media.c RTP socket reachable at 10.1.1.4:4004
22:12:49.137 pjsua_media.c RTCP socket reachable at 10.1.1.4:4005
22:12:49.146 pjsua_media.c RTP socket reachable at 10.1.1.4:4006
22:12:49.149 pjsua_media.c RTCP socket reachable at 10.1.1.4:4007
2009-07-26 22:12:49.152 sipContract[157:207] Status = 0
2009-07-26 22:12:49.181 sipContract[157:207] Here we go now!
22:12:49.209 pjsua_acc.c Account sip:iphone at shayneoneilldevelopment.com
added with id 0
22:12:49.232 pjsua_acc.c Registration sent
22:12:50.349 pjsua_acc.c sip:iphone at shayneoneilldevelopment.com:
registration success, status=200 (OK), will re-register in 300 seconds
22:12:50.349 pjsua_acc.c Keep-alive timer started for acc 0,
destination:174.143.243.107:5060, interval:15s
2009-07-26 22:12:59.656 sipContract[157:207] RINGING sip:2001 at shayneoneilldevelopment.com
22:12:59.663 pjsua_media.c pjsua_set_snd_dev(): attempting to open
devices @16000 Hz
22:12:59.674 ipodsound.c clock rate 16000
22:12:59.675 ipodsound.c pjmedia_snd_stream_start : play back
starting...
22:13:00.240 ipodsound.c Allocated 0 buffer , status code 0
22:13:00.243 ipodsound.c Enqueued with result 0
22:13:00.244 ipodsound.c Allocated 1 buffer , status code 0
22:13:00.245 ipodsound.c Enqueued with result 0
22:13:00.247 ipodsound.c Allocated 2 buffer , status code 0
22:13:00.247 ipodsound.c Enqueued with result 0
22:13:00.249 ipodsound.c Allocated 3 buffer , status code 0
22:13:00.250 ipodsound.c Enqueued with result 0
22:13:00.251 ipodsound.c Allocated 4 buffer , status code 0
22:13:00.252 ipodsound.c Enqueued with result 0
22:13:00.253 ipodsound.c Allocated 5 buffer , status code 0
22:13:00.254 ipodsound.c Enqueued with result 0
22:13:00.261 ipodsound.c New thread!
22:13:00.265 ec0x44aef0 Echo suppressor created,
clock_rate=16000, channel=1, samples per frame=320, tail length=200
ms, latency=20 ms
22:13:00.266 pjsua_call.c Making call with acc #0 to sip:2001 at shayneoneilldevelopment.com
22:13:00.362 ec0x44aef0 Underflow, buf_cnt=0, will generate 1
frame
22:13:00.552 pjsua_media.c Media index 0 selected for call 0
22:13:00.573 APP Call 0 state=CALLING
22:13:01.101 APP Call 0 state=CALLING
22:13:02.121 ipodsound.c Its a trap!
22:13:03.111 APP Call 0 state=EARLY
22:13:04.141 ipodsound.c Its a trap!
22:13:04.569 ec0x44aef0 Underflow, buf_cnt=0, will generate 1
frame
22:13:05.306 sound_port.c EC suspended because of inactivity
22:13:06.141 ipodsound.c Its a trap!
22:13:08.136 ipodsound.c Its a trap!
22:13:10.137 ipodsound.c Its a trap!
22:13:10.997 APP Call 0 state=CONNECTING
22:13:11.002 strm0x87fd74 VAD temporarily disabled
22:13:11.005 strm0x87fd74 Encoder stream started
22:13:11.005 strm0x87fd74 Decoder stream started
22:13:11.006 pjsua_media.c Media updates, stream #0: GSM (sendrecv)
22:13:11.007 conference.c Port 1 (sip:2001@**account details
deleted**) transmitting to port 0 (iPod Sound Device)
22:13:11.007 conference.c Port 0 (iPod Sound Device) transmitting
to port 1 (sip:2001@**account details deleted**)
22:13:11.009 APP Call 0 state=CONFIRMED
(Phone goes haywire at about this point)
Can anyone suggest a strategy to try and track down this bug? I really
just don't know where to start looking.
Thanks, and I hope these instructions I've posted help people navigate
the confusing lack of documentation on the net. Be aware they should
be read in conjunction with the siphon pj build documentation :)
Shayne.
===================================
Shayne O'Neill Development
Mobile, Web and Business process integration.
shayne.oneill at gmail.com 0400247091
Ask me about how Alfresco can help your business grow.
