On Tue, 28 Feb 2012, Martin Jambor wrote:
> Hi,
>
> the second patch in the series handles MEM_REFs on LHS which are parts
> of a handled_component, usually a COMPONENT_REF. The failing testcase
> which requires it is not actually the one in the patch but it is
> gcc.c-torture/execute/mayalias-3.c which fails at -O1 without this
> change but with the third patch applied.
>
> The mechanism is the same as in the previous patch, it also
> piggy-backs a store_bit_field call onto generation of movmisalign
> operation, if that is not available but the access would be
> SLOW_UNALIGNED_ACCESS. I also needed to add
> !mem_ref_refers_to_non_mem_p condition for the same reasons.
> Additionally, complex numbers and accesses to their components are
> already handled by the existing code (the new testcase passes on all
> platforms I tried and so I added it to make sure it continues to) and
> actually messing with it further leads to failures (of
> g++.dg/torture/pr34099.C at -O1). Therefore I added a condition to
> punt on complex modes.
>
> Thanks,
>
> Martin
>
>
> 2012-02-28 Martin Jambor <mjam...@suse.cz>
>
> * expr.c (expand_assignment): Handle misaligned scalar writes to
> memory through MEM_REFs within handled_components by calling
> store_bit_field.
>
> * testsuite/gcc.dg/misaligned-expand-3.c: New test.
>
>
> Index: src/gcc/expr.c
> ===================================================================
> --- src.orig/gcc/expr.c
> +++ src/gcc/expr.c
> @@ -4575,7 +4575,7 @@ expand_assignment (tree to, tree from, b
> rtx result;
> enum machine_mode mode;
> unsigned int align;
> - enum insn_code icode;
> + enum insn_code icode = (enum insn_code) 0;
>
> /* Don't crash if the lhs of the assignment was erroneous. */
> if (TREE_CODE (to) == ERROR_MARK)
> @@ -4689,10 +4689,13 @@ expand_assignment (tree to, tree from, b
> mode = TYPE_MODE (TREE_TYPE (tem));
> if (TREE_CODE (tem) == MEM_REF
> && mode != BLKmode
> + && !COMPLEX_MODE_P (mode)
I think the COMPLEX_MODE_P check is not necessary, the cases we
explicitely handle would all fall under the mem_ref_refers_to_non_mem_p ()
case.
Ok for stage1 with that change if there are no further comments.
Thanks,
Richard.
> + && !mem_ref_refers_to_non_mem_p (tem)
> && ((align = get_object_or_type_alignment (tem))
> < GET_MODE_ALIGNMENT (mode))
> - && ((icode = optab_handler (movmisalign_optab, mode))
> - != CODE_FOR_nothing))
> + && (((icode = optab_handler (movmisalign_optab, mode))
> + != CODE_FOR_nothing)
> + || SLOW_UNALIGNED_ACCESS (mode, align)))
> {
> misalignp = true;
> to_rtx = gen_reg_rtx (mode);
> @@ -4871,13 +4874,18 @@ expand_assignment (tree to, tree from, b
> if (TREE_THIS_VOLATILE (tem))
> MEM_VOLATILE_P (mem) = 1;
>
> - create_fixed_operand (&ops[0], mem);
> - create_input_operand (&ops[1], to_rtx, mode);
> - /* The movmisalign<mode> pattern cannot fail, else the assignment
> - would silently be omitted. */
> - expand_insn (icode, 2, ops);
> + if (icode != CODE_FOR_nothing)
> + {
> + create_fixed_operand (&ops[0], mem);
> + create_input_operand (&ops[1], to_rtx, mode);
> + /* The movmisalign<mode> pattern cannot fail, else the assignment
> + would silently be omitted. */
> + expand_insn (icode, 2, ops);
> + }
> + else
> + store_bit_field (mem, GET_MODE_BITSIZE (mode),
> + 0, 0, 0, mode, to_rtx);
> }
> -
> if (result)
> preserve_temp_slots (result);
> free_temp_slots ();
> Index: src/gcc/testsuite/gcc.dg/misaligned-expand-3.c
> ===================================================================
> --- /dev/null
> +++ src/gcc/testsuite/gcc.dg/misaligned-expand-3.c
> @@ -0,0 +1,43 @@
> +/* Test that expand can generate correct stores to misaligned data of complex
> + type even on strict alignment platforms. */
> +
> +/* { dg-do run } */
> +/* { dg-options "-O0" } */
> +
> +extern void abort ();
> +
> +typedef _Complex float mycmplx __attribute__((aligned(1)));
> +
> +void
> +foo (mycmplx *p, float r, float i)
> +{
> + __real__ *p = r;
> + __imag__ *p = i;
> +}
> +
> +#define cvr 3.2f
> +#define cvi 2.5f
> +#define NUM 8
> +
> +struct blah
> +{
> + char c;
> + mycmplx x[NUM];
> +} __attribute__((packed));
> +
> +struct blah g;
> +
> +int
> +main (int argc, char **argv)
> +{
> + int k;
> +
> + for (k = 0; k < NUM; k++)
> + {
> + foo (&g.x[k], cvr, cvi);
> + if (__real__ g.x[k] != cvr
> + || __imag__ g.x[k] != cvi)
> + abort ();
> + }
> + return 0;
> +}
>
>
--
Richard Guenther <rguent...@suse.de>
SUSE / SUSE Labs
SUSE LINUX Products GmbH - Nuernberg - AG Nuernberg - HRB 16746
GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer
