On Wed, Sep 16, 2015 at 12:58 PM, Stefan Weil <s...@weilnetz.de> wrote: > Am 16.09.2015 um 21:14 schrieb Peter Crosthwaite: >> On Wed, Sep 16, 2015 at 10:50 AM, Stefan Weil <s...@weilnetz.de> wrote: >>> Am 16.09.2015 um 14:29 schrieb Paolo Bonzini: >>>> From: Peter Crosthwaite <crosthwaitepe...@gmail.com> >>>> >>>> This requires global visibility to common code. Move to tcg-common. >>>> >>>> Cc: Stefan Weil <s...@weilnetz.de> >>>> Signed-off-by: Peter Crosthwaite <crosthwaite.pe...@gmail.com> >>>> Message-Id: >>>> <cb0340eba225ab4945aa6cf7c9013f33aa05bcf8.1441614289.git.crosthwaite.pe...@gmail.com> >>>> Signed-off-by: Paolo Bonzini <pbonz...@redhat.com> >>>> --- >>>> tcg/tcg-common.c | 4 ++++ >>>> tci.c | 6 ------ >>>> 2 files changed, 4 insertions(+), 6 deletions(-) >>>> >>> >>> The original versions of patches 15 and 16 did not catch >>> my attention because To / Cc did not include my name. >>> >>> Maybe a problem with the git configuration on the sender side? >>> >> >> Sorry. I have fixed that issue (sneaky suppress-cc on my send-email script). >> >> Is the patch ok? >> >> Regards, >> Peter > > It looks fine. > > I'd keep the comment which you removed in tci.c and move it to > the code location where tci_tb_ptr is used. But that can be done > in a later patch, no need to disturb Paolo's pull request. > > With your QEMU multi-arch modification and some smaller additional > patches it might become possible to build a single QEMU executable > which supports both interpreted and normal TCG, so the TCG interpreter > could be available by default. Users would need a command line option > like --enable-tci to enable it. >
Yep :) That one is on the multi arch radar but for a later phase. More a case of multiple backends though rather than target. Can the same system for differentiating tcg to KVM (accel API) be used here? Regards, Peter > Regards > Stefan >