This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH] Fix PR81782
- From: Richard Biener <rguenther at suse dot de>
- To: Jakub Jelinek <jakub at redhat dot com>
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Fri, 08 Dec 2017 16:51:06 +0100
- Subject: Re: [PATCH] Fix PR81782
- Authentication-results: sourceware.org; auth=none
- References: <alpine.LSU.2.20.1712081543220.12252@zhemvz.fhfr.qr> <20171208145811.GX2353@tucnak>
On December 8, 2017 3:58:11 PM GMT+01:00, Jakub Jelinek <jakub@redhat.com> wrote:
>On Fri, Dec 08, 2017 at 03:44:03PM +0100, Richard Biener wrote:
>>
>> The following fixes spurious uninit warnings for zero-sized arrays.
>>
>> Bootstrapped and tested on x86_64-unknown-linux-gnu, applied.
>>
>> Richard.
>>
>> 2017-12-08 Richard Biener <rguenther@suse.de>
>>
>> PR middle-end/81782
>> * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
>> handle accesses outside of zero-sized vars.
>
>Is ref.max_size == -1 always guaranteeing no access before offset, only
>after it?
Yes.
>What I fear is e.g. ARRAY_REF with non-constant index with base of
>MEM_REF with & of a middle of array or something similar?
>struct S { int x[64]; } foo;
>MEM_REF[&foo, 40][i] with negative i.
The array type constraints the access. Invalid accesses invoke undefined behavior.
>Anyway, in this case it is about not printing a warning, so we can give
>up
>even if it is theoretically possible.
>
>> --- gcc/tree-ssa-uninit.c (revision 255499)
>> +++ gcc/tree-ssa-uninit.c (working copy)
>> @@ -296,8 +296,8 @@ warn_uninitialized_vars (bool warn_possi
>> variable. */
>> if (DECL_P (base)
>> && ref.size != -1
>> - && ref.max_size == ref.size
>> - && (ref.offset + ref.size <= 0
>> + && ((ref.max_size == ref.size
>> + && ref.offset + ref.size <= 0)
>> || (ref.offset >= 0
>> && DECL_SIZE (base)
>> && TREE_CODE (DECL_SIZE (base)) == INTEGER_CST
>
> Jakub