This is the mail archive of the
mailing list for the GCC project.
Re: ACATS c380004
- From: kenner at vlsi1 dot ultra dot nyu dot edu (Richard Kenner)
- To: duncan dot sands at math dot u-psud dot fr
- Cc: gcc at gcc dot gnu dot org
- Date: Fri, 7 Jan 05 10:28:36 EST
- Subject: Re: ACATS c380004
Well, the information from valgrind shows that bad code is being
produced (eg: silly arg (-2147483640) to malloc()).
Right. It's trying to allocate an amount of memory that overflows because
it thinks it needs to allocate an array of 2**32 entries. This can either
end up allocating zero bytes, a negative number, or a large positive
number depending on lot of random things.
By the way, gcc at the pre-ssa tag doesn't fail this test, and
valgrind doesn't show any problems for it.
Try it on a 64-bit machine.
As I said, this is a well-understood issue and has to do with an obscure
part of language semantics. (Sufficiently-obscure that I can't explain it.
Arno, can you try?)