Re: [Qemu-devel] [PATCH] openrisc: terminate qemu process upon receiving a halt signal.

[Richard Henderson <rth@twiddle.net>]

On 04/02/2015 09:50 AM, Hesham ALMatary wrote:
> or1ksim simulator currently handles "l.nop 0xC" instruction as
> a halt signal. Do the same for QEMU.
> 
> Signed-off-by: Hesham ALMatary  <heshamelmatary@gmail.com>
> ---
>  target-openrisc/translate.c | 5 +++++
>  1 file changed, 5 insertions(+)
> 
> diff --git a/target-openrisc/translate.c b/target-openrisc/translate.c
> index dc76789..5fa8ede 100644
> --- a/target-openrisc/translate.c
> +++ b/target-openrisc/translate.c
> @@ -750,6 +750,11 @@ static void dec_misc(DisasContext *dc, uint32_t insn)
>          switch (op1) {
>          case 0x01:    /* l.nop */
>              LOG_DIS("l.nop %d\n", I16);
> +
> +            if(I16 == 0xC) {
> +                exit(0);
> +            }
> +

Ug.  IMO l.nop was a poor choice; l.trap would have been better.
But whatever...

Further:

 (1) This is translation time, not execution time.  You can't do
     the exit here, only when the instruction is executed.

 (2) This, and the other 10 things hidden under l_nop are
     similar to the "-semihosting" switch available for arm,
     lm32, m68k, and xtensa.

     Although semihosting is typically even more complex,
     generally implementing the exit/read/write/open/close/lseek/stat
     families of syscalls.


r~