Change macro definitions for android/arm and add srtp tests into gyp

third_party/libsrtp/libsrtp.gyp

Index: third_party/libsrtp/libsrtp.gyp
===================================================================
--- third_party/libsrtp/libsrtp.gyp	(revision 130067)
+++ third_party/libsrtp/libsrtp.gyp	(working copy)
@@ -29,10 +29,51 @@
         ],
       }],
       ['target_arch=="arm"', {
-        'defines': [
-          'CPU_RISC',
-          'SIZEOF_UNSIGNED_LONG=4',
-          'SIZEOF_UNSIGNED_LONG_LONG=8',
+        'conditions': [
+          ['OS!="android"', {
+            'defines': [
+              'CPU_RISC',
+              'SIZEOF_UNSIGNED_LONG=4',
+              'SIZEOF_UNSIGNED_LONG_LONG=8',
+            ],
+          }],
+          ['OS=="android"', {
+            'defines': [
+              'CPU_CISC',

[Sergey Ulanov, 2012/04/02 18:44:36]

This seems wrong to me. ARM is always RISC architecture, even when you run
android on it :-)

[leozwang, 2012/04/02 21:14:57]

It seemed not a right flag to me at the beginning. srtp uses this flag to define
different data on different platforms, particularly in aes.c. I'm not familiar
with how srtp works internally, but this macro set seems work in my test, AOSP
uses same macro defines and it works fine. I can dig into it if needed.

[Sergey Ulanov, 2012/04/02 23:36:23]

As far as I understand libsrtp uses that to different AES implementations
depending on CPU architecture - one algorithm works better RISC, and the other
one - on CISC. So even if you use CPU_CISC on ARM I would expect it would still
compile, but would be slower. 
It's also possible that the RISC implementation is broken because not many
people use it, and that's the reason it didn't work for you. It shouldn't be
hard to verify. If that's the problem then we will need to track down the issue
and fix it in libsrtp, but it's OK to keep CPU_CISC here.

+              'DEV_URANDOM="/dev/urandom"',
+              'HAVE_ARPA_INET_H',

[Sergey Ulanov, 2012/04/02 18:44:36]

Do we need to have all these defines? Many of them seem to be common with other
platforms so it would be wrong to redefine them specifically for android.

[leozwang, 2012/04/02 21:14:57]

I was also debating about if I should keep this long list. Because I don't have
clear picture of what other platforms will be and what will be supported on
other arm processor based platforms, I decided to keep this long list for
android. If a defines is clearly redefined, I will remove from the list.
Also, I decided to keep this long list because seems these defines are generated
by configure script, I'm not sure if removing any of them could cause any
potential problem. If there is a test or test set that can verify these flags, I
can do more test to make this list shorter.

[Sergey Ulanov, 2012/04/02 23:36:23]

I don't think any of these can cause any problems as long as it compiles. Please
try to find out of CPU_CISC/CPU_RISC is what causing the problem.

+              'HAVE_BYTESWAP_H',
+              'HAVE_INET_ATON',
+              'HAVE_INT16_T',
+              'HAVE_INT32_T',
+              'HAVE_INT8_T',
+              'HAVE_INTTYPES_H',
+              'HAVE_MEMORY_H',
+              'HAVE_NETINET_IN_H',
+              'HAVE_SOCKET',
+              'HAVE_STDINT_H',
+              'HAVE_STDLIB_H',
+              'HAVE_STRINGS_H',
+              'HAVE_STRING_H',
+              'HAVE_SYSLOG_H',
+              'HAVE_SYS_SOCKET_H',
+              'HAVE_SYS_STAT_H',
+              'HAVE_SYS_TYPES_H',
+              'HAVE_SYS_UIO_H',
+              'HAVE_UINT16_T',
+              'HAVE_UINT32_T',
+              'HAVE_UINT64_T',
+              'HAVE_UINT8_T',
+              'HAVE_UNISTD_H',
+              'HAVE_USLEEP',
+              'PACKAGE_BUGREPORT=""',
+              'PACKAGE_NAME=""',
+              'PACKAGE_STRING=""',
+              'PACKAGE_TARNAME=""',
+              'PACKAGE_VERSION=""',
+              'STDC_HEADERS=1',
+            ],
+          }],
         ],
       }],
       ['OS!="win"', {
@@ -71,10 +112,51 @@
           ],
         }],
         ['target_arch=="arm"', {
-          'defines': [
-            'CPU_RISC',
-            'SIZEOF_UNSIGNED_LONG=4',
-            'SIZEOF_UNSIGNED_LONG_LONG=8',
+          'conditions': [
+            ['OS!="android"', {
+              'defines': [
+                'CPU_RISC',
+                'SIZEOF_UNSIGNED_LONG=4',
+                'SIZEOF_UNSIGNED_LONG_LONG=8',
+              ],
+            }],
+            ['OS=="android"', {
+              'defines': [
+                'CPU_CISC',
+                'DEV_URANDOM="/dev/urandom"',
+                'HAVE_ARPA_INET_H',
+                'HAVE_BYTESWAP_H',
+                'HAVE_INET_ATON',
+                'HAVE_INT16_T',
+                'HAVE_INT32_T',
+                'HAVE_INT8_T',
+                'HAVE_INTTYPES_H',
+                'HAVE_MEMORY_H',
+                'HAVE_NETINET_IN_H',
+                'HAVE_SOCKET',
+                'HAVE_STDINT_H',
+                'HAVE_STDLIB_H',
+                'HAVE_STRINGS_H',
+                'HAVE_STRING_H',
+                'HAVE_SYSLOG_H',
+                'HAVE_SYS_SOCKET_H',
+                'HAVE_SYS_STAT_H',
+                'HAVE_SYS_TYPES_H',
+                'HAVE_SYS_UIO_H',
+                'HAVE_UINT16_T',
+                'HAVE_UINT32_T',
+                'HAVE_UINT64_T',
+                'HAVE_UINT8_T',
+                'HAVE_UNISTD_H',
+                'HAVE_USLEEP',
+                'PACKAGE_BUGREPORT=""',
+                'PACKAGE_NAME=""',
+                'PACKAGE_STRING=""',
+                'PACKAGE_TARNAME=""',
+                'PACKAGE_VERSION=""',
+                'STDC_HEADERS=1',
+              ],
+            }],
           ],
         }],
         ['OS!="win"', {