On Tue, Sep 11, 2012 at 5:34 PM, Aurelien Jarno wrote:
> On Sun, Sep 09, 2012 at 08:19:59PM -0400, crwu...@gmail.com wrote:
> > From: Chris Wulff
> >
> > Signed-off-by: Chris Wulff
>
> > +tcg_gen_movi_tl(dc->cpu_R[R_RA], dc->pc + 4);
> > +tcg_gen_movi_tl(dc->cpu_R[R_PC],
> > +
Am 10.09.2012 02:19, schrieb crwu...@gmail.com:
> From: Chris Wulff
>
> Signed-off-by: Chris Wulff
> ---
> target-nios2/Makefile.objs |5 +
> target-nios2/altera_iic.c | 100 +++
> target-nios2/cpu-qom.h | 69 +++
> target-nios2/cpu.c | 83 +++
> target-nios2/cpu.h
Thanks for the feedback. I'll see about getting things updated once I have
the chance. I wasn't too terribly concerned with performance as the target
processor being emulated is much slower than the host (about 100MHz). Even
the way things are now, Linux runs several times faster in emulation that
Thanks for the feedback. Nothing jumps out at me that I disagree with. I'll
get it updated and another patch out once I get the chance to incorporate
the info.
> +#include "hw/sysbus.h"
> +#include "dyngen-exec.h"
> +#include "cpu.h"
> +
> +struct altera_iic {
CamelCase
Yep. I missed that one
Somehow the original patch set never arrived here. Replying indirectly...
> On Sun, Sep 09, 2012 at 08:19:59PM -0400, crwu...@gmail.com wrote:
>> diff --git a/target-nios2/exec.h b/target-nios2/exec.h
...
>> +static inline int cpu_has_work(CPUState *env)
>> +{
>> +return env->interrupt_reques
On 09/11/2012 02:34 PM, Aurelien Jarno wrote:
> Minor nitpick: it's better to write:
>
>tcg_gen_add_tl(addr, addr, dc->cpu_R[instr->a]);
>
> as the code generator on non-RISC hosts usually generate a slightly tiny
> better code.
You know... we already swap operands for constants. It wou