Re: [PATCH 5/7] riscv: Use by-pieces to do overlapping accesses in block_move_straight

["Christoph Müllner" <christoph.muellner@vrull.eu>]

[-- Attachment #1: Type: text/plain, Size: 1799 bytes --]

On Mon, Nov 14, 2022 at 6:16 PM Jeff Law <jeffreyalaw@gmail.com> wrote:

>
> On 11/13/22 16:05, Christoph Muellner wrote:
> > From: Christoph Müllner <christoph.muellner@vrull.eu>
> >
> > The current implementation of riscv_block_move_straight() emits a couple
> > of load-store pairs with maximum width (e.g. 8-byte for RV64).
> > The remainder is handed over to move_by_pieces(), which emits code based
> > target settings like slow_unaligned_access and overlap_op_by_pieces.
> >
> > move_by_pieces() will emit overlapping memory accesses with maximum
> > width only if the given length exceeds the size of one access
> > (e.g. 15-bytes for 8-byte accesses).
> >
> > This patch changes the implementation of riscv_block_move_straight()
> > such, that it preserves a remainder within the interval
> > [delta..2*delta) instead of [0..delta), so that overlapping memory
> > access may be emitted (if the requirements for them are given).
> >
> > gcc/ChangeLog:
> >
> >       * config/riscv/riscv-string.c (riscv_block_move_straight):
> >         Adjust range for emitted load/store pairs.
>
> The change to riscv_expand_block_move isn't noted in the ChangeLog.  OK
> with that fixed (I'm assuming you want to attempt to use overlapping
> word ops for that case).
>

The change in riscv_expand_block_move is a code cleanup.
At the beginning of riscv_expand_block_move we do the following:
  unsigned HOST_WIDE_INT length = UINTVAL (length);
The signature of riscv_block_move_straight wants a "unsigned HOST_WIDE_INT
length".
So we can simply reuse length instead of doing "INTVAL (length)", which is
redundant
and uses the wrong signess (INTVAL vs UINTVAL).

Also, the ChangeLog entry for the test was missing.

Thanks,
Christoph


>
>
> jeff
>
>
>