Re: Libm core dump with SIGILL

[Paul Koning <paulkoning%comcast.net@localhost>]

> On Aug 21, 2015, at 11:10 AM, Greg Stark <stark%mit.edu@localhost> wrote:
> 
> On Fri, Aug 21, 2015 at 3:53 PM, Anders Magnusson <ragge%ludd.ltu.se@localhost> wrote:
>> Well, it is telling the truth.
>> But more interesting is the manpage for infnan that mentions the ability to
>> replace infnan() by something else, depending on the needs in the program.
>> This may be usable for Greg's tests I assume?
> 
> Not really. I suppose I could but it doesn't get called for normal
> arithmetic overflows so I would still need an FPE handler anyways.
> 
> I'm trying to figure out how this API is intended to be used. It
> sounds like I could maybe just set SIGILL to be ignored and check the
> errno after any math call. Or I could add the FPE handler to SIGILL as
> well and treat them as synonyms.

You should just be able to supply your own function infnan(arg), which inspects arg and decides what to do.

	paul