"Robin Dapp" <rdapp....@gmail.com> writes:
> This patch amends the documentation for masked loads (maskload,
> vec_mask_load_lanes, and mask_gather_load as well as their len
> counterparts) with an else operand.
>
> gcc/ChangeLog:
>
> * doc/md.texi: Document masked load else operand.
> ---
> gcc/doc/md.texi | 60 +++++++++++++++++++++++++++++++------------------
> 1 file changed, 38 insertions(+), 22 deletions(-)
>
> diff --git a/gcc/doc/md.texi b/gcc/doc/md.texi
> index 5dc0d55edd6..4047d8f58fe 100644
> --- a/gcc/doc/md.texi
> +++ b/gcc/doc/md.texi
> @@ -5017,8 +5017,9 @@ This pattern is not allowed to @code{FAIL}.
> @item @samp{vec_mask_load_lanes@var{m}@var{n}}
> Like @samp{vec_load_lanes@var{m}@var{n}}, but takes an additional
> mask operand (operand 2) that specifies which elements of the destination
> -vectors should be loaded. Other elements of the destination
> -vectors are set to zero. The operation is equivalent to:
> +vectors should be loaded. Operand 3 is an else operand similar to the one
> +in @code{maskload}.
How about:
Other elements of the destination vectors are taken from operand 3,
which is an else operand similar to the one in @code{maskload}.
> +The operation is equivalent to:
>
> @smallexample
> int c = GET_MODE_SIZE (@var{m}) / GET_MODE_SIZE (@var{n});
> @@ -5028,7 +5029,7 @@ for (j = 0; j < GET_MODE_NUNITS (@var{n}); j++)
> operand0[i][j] = operand1[j * c + i];
> else
> for (i = 0; i < c; i++)
> - operand0[i][j] = 0;
> + operand0[i][j] = operand3;
It looks from the other patches like operand 3 is actually a vector
(which is good), so I suppose this should be operand3[j] instead.
Same for the others.
> @end smallexample
>
> This pattern is not allowed to @code{FAIL}.
> [...]
> @@ -5368,8 +5379,12 @@ Operands 4 and 5 have a target-dependent scalar
> integer mode.
> @cindex @code{maskload@var{m}@var{n}} instruction pattern
> @item @samp{maskload@var{m}@var{n}}
> Perform a masked load of vector from memory operand 1 of mode @var{m}
> -into register operand 0. Mask is provided in register operand 2 of
> -mode @var{n}.
> +into register operand 0. The mask is provided in register operand 2 of
> +mode @var{n}. Operand 3 (the "else value") specifies which value is loaded
> +when the mask is unset.
I think this should clarify that operand 3 has mode @var{m}.
> The predicate of operand 3 must only accept
> +the else values that the target actually supports. Currently two values
> +are attempted, zero and undefined. GCC handles an else value of zero more
> +efficiently than an undefined one.
It looks like the code has support for all-ones as well.
>
> This pattern is not allowed to @code{FAIL}.
>
> @@ -5435,15 +5450,16 @@ Operands 0 and 1 have mode @var{m}, which must be a
> vector mode. Operand 3
> has whichever integer mode the target prefers. A mask is specified in
> operand 2 which must be of type @var{n}. The mask has lower precedence than
> the length and is itself subject to length masking,
> -i.e. only mask indices < (operand 3 + operand 4) are used.
> +i.e. only mask indices < (operand 4 + operand 5) are used.
> +Operand 3 is an else operand similar to the one in @code{maskload}.
> Operand 4 conceptually has mode @code{QI}.
>
> -Operand 2 can be a variable or a constant amount. Operand 4 specifies a
> +Operand 2 can be a variable or a constant amount. Operand 5 specifies a
IIUC, the old text should have said "Operand 3 can be ..." and the new
text should say "Operand 4 can be ...". Operand 2 is the mask.
> constant bias: it is either a constant 0 or a constant -1. The predicate on
> -operand 4 must only accept the bias values that the target actually supports.
> +operand 5 must only accept the bias values that the target actually supports.
> GCC handles a bias of 0 more efficiently than a bias of -1.
>
> -If (operand 2 + operand 4) exceeds the number of elements in mode
> +If (operand 2 + operand 4 + operand 5) exceeds the number of elements in mode
Similarly, I think the old text should have said "operand 3 + operand 4"
and the new one should say "operand 4 + operand 5". There isn't a third
term to add.
Thanks,
Richard
> @var{m}, the behavior is undefined.
>
> If the target prefers the length to be measured in bytes
