My Java app has a JNI layer for communicating with a native C++ library. When the app is idle for approximately three (3) minutes, the app crashes, with LogCat showing the following stack trace:
07-13 13:21:35.876: A/libc(9889): Fatal signal 11 (SIGSEGV) at 0x606b0ba4 (code=1), thread 9889 (nc.myapp)
07-13 13:21:35.980: I/DEBUG(4710): *** *** *** *** *** *** *** *** *** *** *** *** *** *** *** ***
07-13 13:21:35.980: I/DEBUG(4710): Build fingerprint: 'motorola/ghost_retail/ghost:4.4.3/KXA21.12-L1.21/23:user/release-keys'
07-13 13:21:35.981: I/DEBUG(4710): Revision: 'p300'
07-13 13:21:35.983: I/DEBUG(4710): pid: 9889, tid: 9889, name: nc.myapp >>> com.company.myapp <<<
07-13 13:21:35.984: I/DEBUG(4710): signal 11 (SIGSEGV), code 1 (SEGV_MAPERR), fault addr 606b0ba4
07-13 13:21:36.093: I/DEBUG(4710): r0 00000001 r1 0018c020 r2 60b8ec1c r3 00001204
07-13 13:21:36.093: I/DEBUG(4710): r4 00000001 r5 60485008 r6 60b70ce8 r7 00000000
07-13 13:21:36.093: I/DEBUG(4710): r8 606b0cd6 r9 00000001 sl 60485010 fp 00000000
07-13 13:21:36.094: I/DEBUG(4710): ip 00000000 sp 606b0b98 lr 6099ed68 pc 6099ed74 cpsr 200d0010
07-13 13:21:36.094: I/DEBUG(4710): d0 0000000000000001 d1 0000001100000012
07-13 13:21:36.094: I/DEBUG(4710): d2 0000000000000000 d3 000000040000005c
07-13 13:21:36.094: I/DEBUG(4710): d4 0041002e00770065 d5 0061005300730062
07-13 13:21:36.094: I/DEBUG(4710): d6 0053006400650076 d7 0065007400610074
07-13 13:21:36.094: I/DEBUG(4710): d8 440b00000000022c d9 0000000042fe0000
07-13 13:21:36.095: I/DEBUG(4710): d10 0000000000000000 d11 0000000000000000
07-13 13:21:36.095: I/DEBUG(4710): d12 0000000000000000 d13 0000000000000000
07-13 13:21:36.095: I/DEBUG(4710): d14 0000000000000000 d15 0000000000000000
07-13 13:21:36.095: I/DEBUG(4710): d16 0074006900760069 d17 006e0061004d0079
07-13 13:21:36.095: I/DEBUG(4710): d18 0061006c006f0072 d19 007800690070002e
07-13 13:21:36.096: I/DEBUG(4710): d20 00690070006c0065 d21 0049002e00650070
07-13 13:21:36.096: I/DEBUG(4710): d22 0065007800690050 d23 007000690050006c
07-13 13:21:36.096: I/DEBUG(4710): d24 000e000d000c000b d25 0010000f000d000e
07-13 13:21:36.096: I/DEBUG(4710): d26 0000000000000000 d27 0000000000000000
07-13 13:21:36.097: I/DEBUG(4710): d28 0148014701450146 d29 0149014a014a0149
07-13 13:21:36.097: I/DEBUG(4710): d30 0010001000100010 d31 0000000000000000
07-13 13:21:36.097: I/DEBUG(4710): scr 20000013
07-13 13:21:36.098: I/DEBUG(4710): backtrace:
07-13 13:21:36.098: I/DEBUG(4710): #00 pc 000b1d74 /data/app-lib/com.company.myapp-1/libwrappersjni.so
07-13 13:21:36.098: I/DEBUG(4710): #01 pc 000b1d64 /data/app-lib/com.v-1/libwrappersjni.so
The crash seems to indicate the crash is occurring in my shared library, libwrappersjni.so. I have a linker map file, creating through using the -XLinker and -Map gcc linker options. The map file lists symbols with a relative offset, for example:
.ARM.extab.text.Java_com_hyperlync_myapp_WrappersApi_logout
0x00224e1c 0xc ./obj/local/armeabi/objs/wrappersjni/jni_Wrappers.o
.ARM.extab.text.Java_com_hyperlync_myapp_WrappersApi_signout
0x00224e28 0xc ./obj/local/armeabi/objs/wrappersjni/jni_Wrappers.o
.ARM.extab.text.Java_com_hyperlync_myapp_WrappersApi_uploadFileInt
0x00224e34 0x20 ./obj/local/armeabi/objs/wrappersjni/jni_Wrappers.o
What I seem to be missing is the actual address my app was loaded into memory. if I knew that I would subtract the loading address from the pc (Program Counter) register to pinpoint the crash code location.
Is there an intuitive method for manually mapping the absolute crash address found in the pc register to an entry in the linker map file?
