On Tue, 2014-01-28 at 07:49 -0500, Brian Davis wrote: > Tristan wrote: > To clarify: > > The original issue[1] that I found with regards to bug16782 was a > missing bound check message, and abnormal program exit, when Windows > ghdl-mcode encounters an **index** out of range. > > Brian Drummond then started patching the overflow exception path in an > attempt to address this problem, and asked me to test his patch. > ( Which I thought sounded reasonable, as IIRC I have seen bounds test > implemented with trapping math against the bounds. ) > > My recent test of **integer** out of range (rather than index out of > range), which works as expected, seems to show the generated bounds > check code does not lead though the particular overflow handling code > that he was patching.
Let me understand : that patch actually fixed *one* of the two reported testcases? If so, that wasn't clear from your earlier report. What is it that "works as expected" in the sentence above? Integer overflow but not index overflow? And does that (i.e. integer overflow) get through the SEH with an additional printf message at its start? If *integer* overflow produces the expected message via SEH but index overflow doesn't, I'll take another look and see if I can figure out why. I was under the impression that the patch fixed nothing because neither testcase got to the handler. - Brian _______________________________________________ Ghdl-discuss mailing list [email protected] https://mail.gna.org/listinfo/ghdl-discuss
