Re: [PATCH 03/14] KVM: x86 emulator: move memop, memopp into emulation context

[Avi Kivity <avi@redhat.com>]

On 09/14/2011 01:41 PM, Takuya Yoshikawa wrote:
> On Tue, 13 Sep 2011 10:45:40 +0300
> Avi Kivity<avi@redhat.com>  wrote:
>
> >  Simplifies further generalization of decode.
> >
> >  Signed-off-by: Avi Kivity<avi@redhat.com>
> >  ---
> >   arch/x86/include/asm/kvm_emulate.h |    2 ++
> >   arch/x86/kvm/emulate.c             |   34 +++++++++++++++++-----------------
> >   2 files changed, 19 insertions(+), 17 deletions(-)
> >
> >  diff --git a/arch/x86/include/asm/kvm_emulate.h b/arch/x86/include/asm/kvm_emulate.h
> >  index 6040d11..56bac3e 100644
> >  --- a/arch/x86/include/asm/kvm_emulate.h
> >  +++ b/arch/x86/include/asm/kvm_emulate.h
> >  @@ -275,6 +275,8 @@ struct x86_emulate_ctxt {
> >   	unsigned long _eip;
> >   	/* Fields above regs are cleared together. */
> >   	unsigned long regs[NR_VCPU_REGS];
> >  +	struct operand memop;
> >  +	struct operand *memopp;
> >   	struct fetch_cache fetch;
> >   	struct read_cache io_read;
> >   	struct read_cache mem_read;
>
> Once the emulator context gets stablized, some comments will be nice
> to know which ones are supposed to be accessed from outside of the
> emulator, and which ones are only for the emulator internal usage.
>
> Practically, knowing each member's lifetime, decode stage only or
> emulation stage only or throughout the emulation, will make it easy
> to avoid extra ctxt/regs initialization and ... maybe more.
>

Nothing should be accessed from outside the emulator, except via 
accessors.  We should move initialization to the emulator as well (or 
just initialize from x86_decode_insn() - any reason not to?)

-- 
error compiling committee.c: too many arguments to function