Cygwin / Linux core dumps.

[John E]

After a long period of inactivity, I need to revisit a program that I once compiled under Cygwin/gcc. In theory it should still be the same as I left it but after rebuilding it this morning, I got a dreaded segfault at run time (before the program had even become visible). The only help I got was this core dump:-

Exception: STATUS_ACCESS_VIOLATION at eip=106DC252
eax=00000000 ebx=00000000 ecx=11866B30 edx=0022B5D0 esi=0022B620 edi=00820F50
ebp=0022B488 esp=0022B488 program=C:\cygwin\path\to\my\program.exe, pid 3676, thread main
cs=001B ds=0023 es=0023 fs=003B gs=0000 ss=0023
Stack trace:
Frame Function Args
0022B488 106DC252 (00000000, 1185CC60, 0022B4B8, 6AFD16D9)
0022B528 1067A40E (0022B620, 119266C0, 0022B5D0, 0022B610)
0022B688 110F567E (117D7170, 11171430, 00000009, 0022B700)
0022B748 00431DA3 (117D7170, 11171430, 00000009, 0022B7E0)
0022B858 110F5238 (117D7170, 00000001, 11926674, FFFFFFFF)
0022B878 110F4E5C (117D7170, 00000001, 11926674, 0081DC78)
0022B898 0A262611 (11935774, 0A2A9114, 0022BA2C, 0A293116)
0022B8B8 0A262412 (11935770, 0A2A9114, 0022BA2C, 0A25D5C3)
0022B8D8 0A25CE60 (11935758, 0A2A9114, 0022BA2C, 0A2565AB)
0022B988 0A25D4B2 (117D6E88, 0A2A9114, 0022BA2C, 0022BA40)
0022B9A8 0A26269E (0A2A912C, 0A2A9114, 0022BA2C, 00000000)
0022B9C8 0A2626BF (0A2A912C, 0A2A9114, 0022BA2C, 0A248A97)
0022BA68 0A23FB30 (0A2A90E0, 00A6EB8C, 00A6DA18, 00000008)
0022BA98 00838B38 (0A2A90E8, 00838A94, 0022BB60, 00895D32)
0022BB88 00628736 (119495AC, 0022C080, 0022C0C0, 00000001)
0022C108 0061C46E (119495AC, 00A72C24, 0022C9A0, 00000000)
End of stack trace (more stack frames may be present)

Does this tell me anything useful? FWIW the original executable (which I backed up before rebuilding it today) still runs fine.

[John E]

I managed to find the problem. Although the source itself was still intact, a couple of the compiler settings had changed and that's what was causing the difference in compilation.

Having said that, I'm still curious to know why Linux and Cygwin always seem to produce this report whenever a program crashes. AFAICT it isn't even remotely helpful in tracking down the problem.