Re: [Qemu-devel] [RFC][PATCH v2 2/4] tcg: add extended MMU helpers to softmmu targets

[Yeongkyoon Lee]

On 2012년 07월 06일 03:49, Blue Swirl wrote:
> On Thu, Jul 5, 2012 at 1:43 PM, Peter Maydell <address@hidden> wrote:
> > On 5 July 2012 14:23, Yeongkyoon Lee <address@hidden> wrote:
> > > Add extended MMU helpers to softmmu targets, where the targets are alpha, arm, 
> > > cris, i386, lm32, m68k, microblaze, mips, ppc, s390x, sh4, sparc and xtensa.
> > >
> > > Signed-off-by: Yeongkyoon Lee <address@hidden>
> > > ---
> > >   target-alpha/mem_helper.c     |   22 ++++++++++++++++++++++
> > >   target-arm/op_helper.c        |   23 +++++++++++++++++++++++
> > >   target-cris/op_helper.c       |   22 ++++++++++++++++++++++
> > >   target-i386/mem_helper.c      |   22 ++++++++++++++++++++++
> > >   target-lm32/op_helper.c       |   23 ++++++++++++++++++++++-
> > >   target-m68k/op_helper.c       |   22 ++++++++++++++++++++++
> > >   target-microblaze/op_helper.c |   22 ++++++++++++++++++++++
> > >   target-mips/op_helper.c       |   22 ++++++++++++++++++++++
> > >   target-ppc/mem_helper.c       |   22 ++++++++++++++++++++++
> > >   target-s390x/op_helper.c      |   22 ++++++++++++++++++++++
> > >   target-sh4/op_helper.c        |   22 ++++++++++++++++++++++
> > >   target-sparc/ldst_helper.c    |   23 +++++++++++++++++++++++
> > >   target-xtensa/op_helper.c     |   22 ++++++++++++++++++++++
> > >   13 files changed, 288 insertions(+), 1 deletions(-)
> > This makes the already slightly repetitive inclusion of the
> > softmmu_templates even more repetitive. Perhaps we could abstract
> > it all out into a single header which the targets can include?
> I'd just replace standard versions with extended versions
> unconditionally, no CONFIG_*. Both AREG0 and !AREG0 cases must be
> handled.

I've only modified the code related to the MMU call from generated code 
because this patch is focused optimize that kind of runtime generated 
code. As I mentioned another thread, basically I agree to apply AREG0 
case if ldst optimization is accepted as default (no macros).
But when replacing standard versions with extended versions, there is an 
issue of non-x86 (and x64) hosts which has no ldst optimization impl. It 
needs to conserve the usage of conditional macro for standard versions, 
however, it looks better have a type of version per a host, which is 
different from current my patch.
How do you think about it?