Re: [Qemu-devel] Sparc32: Mask writes to the WIM register

[Blue Swirl]

On 8/19/08, Luis Pureza <address@hidden> wrote:
> On Tue, Aug 19, 2008 at 5:48 PM, Blue Swirl <address@hidden> wrote:
>  > On 8/19/08, Luis Pureza <address@hidden> wrote:
>  >> The following patch, for Sparc32, correctly masks the WIM register, so
>  >>  that bits relative to non-existent windows remain cleared.
>  >>  This fixes a problem I was having with RTEMS.
>  >
>  > Thanks. The wim mask can be calculated at translation time, so perhaps
>  > you could propose a new patch where only an "andi_i32" is added?
>
>
> That was my first try, but then I realized I can't access
>  env->nwindows from within disas_sparc_insn(). So I'm not really sure
>  what's the best approach to somehow store the result of (1 <<
>  env->windows) - 1 in a way that's accessible inside
>  disas_sparc_insn(). What do you think?

At gen_intermediate_code_internal(), env is still accessible, so
nwindows could be copied to a new field of dc there.

>  > Could you give an URL to an .iso image or something so I could add
>  > RTEMS to my set of tests?
>
>
> I'm afraid it's not that simple, as the RTEMS tasks I'm running depend
>  on a lot of stuff that's not yet into QEMU trunk (namely support for
>  the LEON2 and ERC32 processors). But you might be able to compile
>  something for the sun4 architecture and execute it from there.
>
>  By the way, it is my wish to submit this work to QEMU. However, I
>  faced some "integration issues" and I had to modify some TARGET_SPARC
>  bits that were too coupled with the sun4 architecture. And I wanted to
>  discuss with you how to deal with these differences. Do you mind if I
>  send you a private email one of these days detailing the issues?

No problem.