[Bug tree-optimization/94675] [9/10 regression] -Warray-bounds false positive with -O2 since r9-1948

["msebor at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94675

--- Comment #14 from Martin Sebor <msebor at gcc dot gnu.org> ---
I can think of only one way the warning code could avoid triggering here: by
assuming that the difference between two pointers into the same array is less
than or equal the size of the array (with non-array objects being viewed as
arrays of a single element).  That's necessarily true in any valid C/C++
program so it would be a safe assumption to make here as well.

Since there is no array subscripting involved here, issuing -Warray-bounds
might also be considered inappropriate, and the warning could be made not to
trigger.  But that would just mask the problem which would come back if
__ps_skip attempted to access ps->s[len].  It would also come back if/when GCC
started to warn more diligently for forming out-of-bounds pointers (I already
submitted one patch to do that and the work I'm doing in the path isolation
pass is an opportunity to revisit the feature).

So we're back to deriving assumptions about the results of pointer arithmetic
based the sizes of pointed-to objects.  The warning would need to work quite
hard to figure this out in general, so hard that I don't think it would be
worth the effort unless it benefited code generation as well, or at least all
other warnings like it (-Warray-bounds isn't the only one that can be triggered
-- a number of others could be made to).  Which was also the point of my
comment #1 (and related to Richard's observation in comment #4 about an missed
optimization opportunity).

That said and codegen improvements aside, I think the submitted test case is
sufficiently tricky that I don't see issuing a warning for it as a problem. 
All flow-based warnings have a non-zero rate of false positives (as do many
front-end warnings) and there are mechanisms to avoid them.  Compared to some
of the other false positives we have, avoiding this one seems like a low
priority to me.