qemu-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Qemu-devel] [PATCH 1/7] target-tricore: Add FPU infrastructure


From: Richard Henderson
Subject: Re: [Qemu-devel] [PATCH 1/7] target-tricore: Add FPU infrastructure
Date: Tue, 1 Mar 2016 09:46:07 -0800
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:38.0) Gecko/20100101 Thunderbird/38.6.0

On 03/01/2016 08:24 AM, Bastian Koppelmann wrote:
> +static inline void f_update_psw_flags(CPUTriCoreState *env, bool calc_z)

You probably need it for compiling this intermediate patch, but probably drop
inline and let the compiler choose.  This is quite a bit of code after all...

> +{
> +    int8_t flags = env->fp_status.float_exception_flags;
> +    int32_t some_excp = 0;

You need to set float_exception_flags to zero after reading, so that you don't
re-copy the flags during the next fp insn.

> +#define FPU_FS PSW_USB_C
> +#define FPU_FI PSW_USB_V
> +#define FPU_FV PSW_USB_SV
> +#define FPU_FZ PSW_USB_AV
> +#define FPU_FU PSW_USB_SAV

What an unfortunate spec.  This is an incredibly broken way to implement fp
exception flags.

Exception flags are the sort of thing that's supposed to be collected across a
whole subroutine, without having to worry about the fp exception flags being
clobbered by the surrounding pointer arithmetic.

In order to implement proper IEEE support with this chip, one would have to
implement the fp exception word in software, clearing the PSW bits before each
group of fp instructions and storing the PSW bits after each such group.

Oh well.  You're doing what the manual says...


r~



reply via email to

[Prev in Thread] Current Thread [Next in Thread]