On Tue, Apr 17, 2018 at 11:00:35AM +0200, Eric Botcazou wrote:
> Right, they were originally alike, but someone you know very well changed it:
>
> 2013-12-17 Jakub Jelinek
>
> * expr.c (convert_modes): For SUBREG_PROMOTED_VAR_P use SUBREG_REG (x)
> instead of x as last gen_lowpart
On Tue, Apr 17, 2018 at 09:32:31AM +0200, Eric Botcazou wrote:
> > The following testcase FAILs, because cse_local sees
> > (zero_extend:TI (subreg/s/v:DI (reg:TI ...) 0))
> > inside of REG_EQUAL note, and simplify-rtx.c attempts to optimize it.
> > case ZERO_EXTEND:
> > /* Check for a ze
> Yes, convert_modes does this, on the other side e.g. convert_move a few
> hundred lines above it doesn't:
> if (GET_CODE (from) == SUBREG
> && SUBREG_PROMOTED_VAR_P (from)
> && is_a (to_mode, &to_int_mode)
> && (GET_MODE_PRECISION (subreg_promoted_mode (from))
>
>
> The following testcase FAILs, because cse_local sees
> (zero_extend:TI (subreg/s/v:DI (reg:TI ...) 0))
> inside of REG_EQUAL note, and simplify-rtx.c attempts to optimize it.
> case ZERO_EXTEND:
> /* Check for a zero extension of a subreg of a promoted
> variable, where the pro
Hi!
The following testcase FAILs, because cse_local sees
(zero_extend:TI (subreg/s/v:DI (reg:TI ...) 0))
inside of REG_EQUAL note, and simplify-rtx.c attempts to optimize it.
case ZERO_EXTEND:
/* Check for a zero extension of a subreg of a promoted
variable, where the promotion
