Hi Bastian,
allright thank you. A fix is already on the way.
Best regards,
Andreas
> -----Ursprüngliche Nachricht-----
> Von: Bastian Koppelmann <kbast...@mail.uni-paderborn.de>
> Gesendet: Mittwoch, 19. Juni 2019 14:25
> An: David Brenken <david.bren...@efs-auto.org>; qemu-devel@nongnu.org
> Cc: Biermanski, Lars (EFS-GH3) <lars.bierman...@efs-auto.de>; Hofstetter,
> Georg (EFS-GH2) <georg.hofstet...@efs-auto.de>; Brenken, David (EFS-GH2)
> <david.bren...@efs-auto.de>; Rasche, Robert (EFS-GH2) <robert.rasche@efs-
> auto.de>; Konopik, Andreas (EFS-GH2) <andreas.kono...@efs-auto.de>
> Betreff: Re: [Qemu-devel] [PATCH v2 4/5] tricore: add QSEED instruction
>
>
> On 6/19/19 9:56 AM, David Brenken wrote:
> > +
> > + result = 0;
> > + result = deposit32(result, 31, 1, new_S);
> > + result = deposit32(result, 23, 8, new_E);
> > + result = deposit32(result, 15, 8, new_M);
> > + }
> > +
> > + if (float32_is_any_nan(arg1) || result == float32_sqrt_nan) {
>
>
> You need float32_is_signaling_nan, since only signaling nan raises the invalid
> flag.
>
>
> > + env->FPU_FI = 1;
>
> env->FPU_FI = 1 << 31;
>
> See f_update_psw_flags(). FPU_FI and PSW_V are the same and we defined
> bit 31 as the V bit for optimization purposes.
>
> Cheers,
>
> Bastian
