This is the mail archive of the
mailing list for the GCC project.
Re: [PATCH] Fix digest_init vector handling (PRs middle-end/19164,c/17297)
- From: Mark Mitchell <mark at codesourcery dot com>
- To: Jakub Jelinek <jakub at redhat dot com>
- Cc: Richard Henderson <rth at redhat dot com>, gcc-patches at gcc dot gnu dot org
- Date: Tue, 18 Jan 2005 22:00:43 -0800
- Subject: Re: [PATCH] Fix digest_init vector handling (PRs middle-end/19164,c/17297)
- Organization: CodeSourcery, LLC
- References: <20050113181430.GA10340@devserv.devel.redhat.com>
Jakub Jelinek wrote:
PR middle-end/19164 ICEs because digest_init handles assumes there is a
CONSTRUCTOR while there is actually a NOP_EXPR <INTEGER_CST> (or there
can be VECTOR_CST etc.).
PR c/17297 fails because although the constructor is TREE_CONSTANT,
it contains TREE_CONSTANT RDIV_EXPRs (division by 0), but everything
in the compiler expects VECTOR_CST to only contain *_CSTs.
Ok to commit if testing succeeds, both to HEAD and 3.4?
Yes, this OK.
I've run into this mess before in other contexts. I can't remember why
we allow division by zero in the input program; perhaps the standards
say that this is undefined behavior only if the program actually reaches