Using gdb insted of idb

[alexey_medvedev]

Hello!

I have a problem with debugging my software package, compliled with icc 9.0 on ia64 (Version 9.0 Build 20051201). When all C++ source files are compiled with -g -mp and without any -Ox options and any other specific options, eidb says: "Warning: bad source correlation found. Further instances ignored.". When I start program, it doesn't seems to work at all, just says: "Thread received signal 32
stopped at [ __pthread_sigsuspend(...) 0x200000000005b8e1]".

As far as I can understand it is a known problem with eidb, so I try to guess how to find the workaround here. I tried to start the program under gdb (GNU gdb 6.3-debian), and it seems to work great, but I'm not sure if it is quite reliable and safe to do so.

Can I rely on the results of such a debugging?

[TimP]

Perhaps it's evident, but your suggestions all involve unsupported software combinations, so are undertaken at your own risk. I would not be surprised that you found gdb more satisfactory than idb on debian, and you have more potential resources to sort out problems with gdb. You might do well to try more up to date versions of icc and idb. many improvements and bug fixes have come in the last 19 months.

[Intel_C_Intel]

Intel compilers produce debug information compatible with gdb. If you find that gdb is providing satisfacory debugging for your application, then please do proceed in that way.

You may also want to try a later baselevel of IDB. You may do so by downloading the latest update of the 9.0 compiler (or an evaluation copy of the 9.1 compiler) and installing the IDB that comes with that package. You won't need a serial number for it (and it won't expire at the end of the evaluation period).

-- G