On 25/07/18 21:38, Luc Van Oostenryck wrote:
> A zero-extension has no effect on the result of a comparison with 0 or 1.
>
> Optimize away the extension.
>
> Note: this simplification was already done but only when the
> original size was 1 but it can be done for all sizes.
>
> Signed-off-by: Luc Van Oostenryck <luc.vanoostenryck@xxxxxxxxx>
> ---
> simplify.c | 17 +++++++----------
> validation/optim/bool-zext-test.c | 1 -
> 2 files changed, 7 insertions(+), 11 deletions(-)
>
> diff --git a/simplify.c b/simplify.c
> index b6c07ad9f..fb4bb3429 100644
> --- a/simplify.c
> +++ b/simplify.c
> @@ -608,16 +608,13 @@ static int simplify_seteq_setne(struct instruction *insn, long long value)
> return REPEAT_CSE;
>
> case OP_ZEXT:
> - if (def->orig_type->bit_size == 1) {
> - // Convert:
> - // zext.m %s <- (1) %a
> - // setne.1 %r <- %s, $0
> - // into:
> - // setne.1 %s <- %a, $0
> - // and same for setne/eq ... 0/1
> - return replace_pseudo(insn, &insn->src1, def->src1);
> - }
> - break;
> + // Convert:
> + // *ext.m %s <- (1) %a
why *ext.m?, this is only zext - Ah, next patch you add sext! ;-)
Hmm, maybe keep it as zext until next patch?
> + // setne.1 %r <- %s, $0
> + // into:
> + // setne.1 %s <- %a, $0
> + // and same for setne/eq ... 0/1
> + return replace_pseudo(insn, &insn->src1, def->src);
Hmm, so this isn't just removing the conditional. In the context
I can see that the last argument to replace_pseudo was def->src1
rather than def->src used here. I haven't given it much thought,
just reading the patch text, but ...
ATB,
Ramsay Jones
> }
> return 0;
> }
> diff --git a/validation/optim/bool-zext-test.c b/validation/optim/bool-zext-test.c
> index f837ace20..138938b0a 100644
> --- a/validation/optim/bool-zext-test.c
> +++ b/validation/optim/bool-zext-test.c
> @@ -6,7 +6,6 @@ _Bool neu1(unsigned char a) { return a != 1; }
> /*
> * check-name: bool-zext-test
> * check-command: test-linearize -Wno-decl $file
> - * check-known-to-fail
> *
> * check-output-ignore
> * check-output-excludes: zext\\.
>
--
To unsubscribe from this list: send the line "unsubscribe linux-sparse" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
