Am Montag, den 10.01.2011, 18:38 +0100 schrieb Vincent Rivière <vincent.riviere@freesbee.fr>:
It is very easy to investigate that point. Put a printf() at the very beginning of main(). If you see your trace, it is the proof that it does not crash before main(). Then add other traces in strategical places to find where it actually crashes. If you don't see your trace, there is a very big problem with the compiler, linker or startup code.
Well (or not),I have to admit that I found out yesterday, that it is not crashing before main. I tried to run NetSurf on plain Falcon with plain TOS - which resultet in 4 Bombs, or 2, I don't remember now... Then I did a next try..., this time with ADEBUG... and then I found out that it did not find the resource file :/ This time it was exiting with an error message, like it should... I don't know how it happened that NS was showing bombs, but I know that the version I tried first maybe had -02 code within..., and now I'm just forgetting about that and try to resolve the current issue (line f emulation stuff) - which happens when I open the Hotlist window ( no network code involved ...).
One thing that I find exciting: the linef triggering instruction is located at an address that is not "disassembled as start of instruction" - so you can't see that address within the ASM Listing. Let's say the bad-boy address ist at A... but the disassembler just shows instructions at Addresses 8 and C. so when PC hit's the badboy instruction, it means that there was an jump to an location which isn't aligned as the debugger expected, right? (Of course the debugger now disassembles the instruction from A..., but before the PC pointed to it, it was part of another instruction, Do you know what I mean?) Maybe it is just an disassembling problem... but maybe this is another indicator for an Overflow...