Re: [PATCH] improve -Wrestrict for struct members (PR 85753)

[Martin Sebor <msebor at gmail dot com>]

On 05/14/2018 03:21 PM, Jeff Law wrote:
> On 05/11/2018 05:09 PM, Martin Sebor wrote:
> > The attached patch extends -Wrestrict to constrain valid offset
> > ranges into objects of struct types to the bounds of the object
> > type, the same way the warning already handles arrays.  This
> > makes it possible to detect overlapping accesses in cases like
> > the second call to memcpy below:
> >
> >   char a[16];
> >
> >   struct { char a[16]; } x;
> >
> >   void f (int i, int j)
> >   {
> >     memcpy (&a[i], &a[j], 9);   // -Wrestrict (good)
> >
> >     memcpy (&x.a[i], &x.a[j], 9);   // missing -Wrestrict
> >   }
> >
> > These is no way to copy 9 bytes within a 16 byte array without
> > either overlap or without accessing memory outside the bounaries
> > of the object.
> >
> > This is for GCC 9.
> >
> > Thanks
> > Martin
> >
> > gcc-85753.diff
> >
> >
> > PR tree-optimization/85753 - missing -Wrestrict on memcpy into a member array
> >
> > gcc/ChangeLog:
> >
> > 	PR tree-optimization/85753
> > 	* gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
> > 	RECORD_TYPE in addition to ARRAY_TYPE.
> >
> > gcc/testsuite/ChangeLog:
> >
> > 	PR tree-optimization/85753
> > 	* gcc.dg/Wrestrict-10.c: Adjust.
> > 	* gcc.dg/Wrestrict-16.c: New test.
> OK.
> jeff

It occurred to me that it doesn't matter if the type is an array,
struct, or a basic type like int.  They all should be treated
the same so after retesting I ended up committing a slightly
simplified version of the original: r260280.

Martin