[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: exception crash
From: |
Guy Harrison |
Subject: |
Re: exception crash |
Date: |
Tue, 18 Jan 2005 17:00:09 GMT |
User-agent: |
KNode/0.8.1 |
John Max Skaller wrote:
> On Fri, 07 Jan 2005 23:00:10 +0000, Guy Harrison wrote:
>>
>> Don't quite follow you there. If what follows doesn't help please
>> elaborate with code.
>>
>>> BTW: the error is known and expected, it should
>>> print a diagnostic and terminate gracefully, but
>>> aborts instead. The same behaviour occurs for all
>>> errors.
>>
>> int
>> main(/*...*/)
>> {
>> try {
>> return Main();
>> }
>> //other catches
>> catch (...) {
>> std::cerr << "*** UNHANDLED EXCEPTION ***\n";
>> return EXIT_FAILURE;
>> }
>> }
>
> My mainline does the equivalent of the above.
> Doesn't help. I'm getting an unhandled exception,
> which calls terminate, so the catch(...) isn't relevant.
Hmm, the only other "obvious" situation where that can occur is a
double-exception. I only mention that in case...
> The error only occurs with -O3 -fomit-frame-pointer.
> So I think this is either this is either a bug in g++ or
> in the documentation for failing to specify the above
> combination doesn't work.
...the optimisation above is causing code generation to create that
situation. Nevertheless, it's apparent you have a bug with even "throw 42"
failing (you may even be able to follow the asm for that by hand). Perhaps
RedHat already know of it, or else you should submit a report.