MTTCG start-up tweaks

[Qemu-devel] [PATCH v1 0/3] MTTCG start-up tweaks

Alex Bennée posted 3 patches 7 years, 1 month ago

Download series mbox

Patches applied successfully (tree, apply log)
git fetch https://github.com/patchew-project/qemu tags/patchew/20170228150709.27453-1-alex.bennee@linaro.org

Test checkpatch passed

Test docker passed

Test s390x passed

cpus.c            | 11 +++++++----
target/i386/cpu.h |  3 +++
vl.c              |  7 ++-----
3 files changed, 12 insertions(+), 9 deletions(-)

Expand all Fold all

[Qemu-devel] [PATCH v1 0/3] MTTCG start-up tweaks

Posted by Alex Bennée 7 years, 1 month ago

Hi,

These are a couple of quick tweaks to command line handling for MTTCG.
The first patch moves the paring of --accel tcg to after icount and
simplifies the logic to disable mttcg when icount is enabled. There is
still a regression to icount to be fixed (caused by 8d04fb5).

Finally there is a patch to declare the TCG_GUEST_DEFAULT_MO for x86
guests to avoid the bogus warning about memory order when forcing
mttcg on. A new warning has been added specifically for guests that
haven't declared TARGET_SUPPORT_MTTCG which means the front-end has
had atomics/barrier/helpers converted for a MTTCG world.

Alex Bennée (3):
  vl/cpus: be smarter with icount and MTTCG
  target/i386/cpu.h: declare TCG_GUEST_DEFAULT_MO
  cpus.c: add additional error_report when !TARGET_SUPPORT_MTTCG

 cpus.c            | 11 +++++++----
 target/i386/cpu.h |  3 +++
 vl.c              |  7 ++-----
 3 files changed, 12 insertions(+), 9 deletions(-)

-- 
2.11.0

Re: [Qemu-devel] [PATCH v1 0/3] MTTCG start-up tweaks

Posted by Richard Henderson 7 years, 1 month ago

On 03/01/2017 02:07 AM, Alex Bennée wrote:
> Hi,
>
> These are a couple of quick tweaks to command line handling for MTTCG.
> The first patch moves the paring of --accel tcg to after icount and
> simplifies the logic to disable mttcg when icount is enabled. There is
> still a regression to icount to be fixed (caused by 8d04fb5).
>
> Finally there is a patch to declare the TCG_GUEST_DEFAULT_MO for x86
> guests to avoid the bogus warning about memory order when forcing
> mttcg on. A new warning has been added specifically for guests that
> haven't declared TARGET_SUPPORT_MTTCG which means the front-end has
> had atomics/barrier/helpers converted for a MTTCG world.
>
> Alex Bennée (3):
>   vl/cpus: be smarter with icount and MTTCG
>   target/i386/cpu.h: declare TCG_GUEST_DEFAULT_MO
>   cpus.c: add additional error_report when !TARGET_SUPPORT_MTTCG
>
>  cpus.c            | 11 +++++++----
>  target/i386/cpu.h |  3 +++
>  vl.c              |  7 ++-----
>  3 files changed, 12 insertions(+), 9 deletions(-)
>

Reviewed-by: Richard Henderson <rth@twiddle.net>


r~