On 12/13/21 4:13 AM, matheus.fe...@eldorado.org.br wrote:
From: Matheus Ferst <matheus.fe...@eldorado.org.br>
The non-signalling versions of VSX scalar convert to shorter/longer
precision insns doesn't silence SNaNs in the hardware. We are currently
honoring this behavior in xscvdpspn, since helper_xscvdpspn handles the
conversion with extracts/deposits/etc. OTOH, xscvspdpn uses
float32_to_float64 that calls parts_float_to_float, which uses
parts_return_nan that finally calls parts_silence_nan if the input is an
SNaN.
To address this problem, this patch adds a new float_status flag
(return_snan) to avoid the call to parts_silence_nan in the
float_class_snan case of parts_return_nan.
Signed-off-by: Matheus Ferst <matheus.fe...@eldorado.org.br>
I think you shouldn't be using softfloat at all for this, but use the existing
helper_todouble function.
r~
