Re: [PATCH 3/4] powerpc/powernv: Enable TM without suspend if possible

[Michael Neuling <mikey@neuling.org>]

On Thu, 2017-10-19 at 11:34 -0200, Tulio Magno Quites Machado Filho wrote:
> Forwarding some comments from Adhemerval sent to libc-alpha [1]...
>=20
> Adhemerval Zanella <adhemerval.zanella@linaro.org> writes:
> > Florian Weimer <fweimer@redhat.com> writes:
> >=20
> > > On 10/12/2017 12:17 PM, Michael Ellerman wrote:
> > > > +	pr_info("Enabling TM (Transactional Memory) with Suspend
> > > > Disabled\n");
> > > > +	cur_cpu_spec->cpu_features |=3D CPU_FTR_TM;
> > > > +	cur_cpu_spec->cpu_user_features2 |=3D
> > > > PPC_FEATURE2_HTM_NO_SUSPEND;
> > > > +	tm_suspend_disabled =3D true;
> > >=20
> > > This doesn't look right because if suspend is not available, you need=
 to=C2=A0
> > > clear the original PPC_FEATURE2_HTM flag because the semantics are no=
t=C2=A0
> > > right, so that applications can use fallback code.=C2=A0=C2=A0Otherwi=
se,=C2=A0
> > > applications may incorrectly select the HTM code and break if running=
 on=C2=A0
> > > a system which supports HTM, but without the suspend state.
> > >=20
> > > The new flag should say that HTM is supported, but without the suspen=
d=C2=A0
> > > state, and it should be always set if PPC_FEATURE2_HTM is set.
> >=20
> > Will it also change TEXARS with the abort information?
>=20
> It should, with a permanent error cause so that old applications entering
> suspended state can adopt another technique.
> Michael, could you clarify if this is indeed happening, please?

It will.

If we hit a tsuspend and we are in this mode, the hardware will abort us an=
d set
the Failure Persistent bit (unless there is some other exception pending at=
 the
same time which causes the abort).

Mikey