[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] Fwd: [PATCH v5] xen/arm : emulation of arm's PSCI v0.2 standard

On Fri, 2014-07-25 at 09:54 +0100, Ian Campbell wrote:
> > > The upper bits of power_state also describe an affinity level to put to
> > > sleep, which you also don't handle AFAICT. Perhaps since we don't
> > > actually have affinities > 0 what you've done is OK, what do you think?
> > I think it is better to put a comment explicitly stating that we have
> > ignored the value because affinities > 0 are not valid for xen or we
> > can check for affinity and return if it is greater than zero.
> > I think i'll go with check, will take care in next patchset
> I'm not 100% what the spec allows/requires us to do here. I've asked
> ARM. I suspect shutting down only the local core would be permissible.

I've asked ARM and this is fine. If the caller asks us to power down the
"cluster" we don't have to actually go that deep and since we are
virtualised it doesn't really matter.

BTW, WRT to stand by vs power down (bit 16). For standby you always
return to the next instruction after the SMC where as for power down you
can return either to the next instruction or the entry point address.
Since we don't actually implement hardware sleep states which don't
preserve the full CPU state we may as well take the easy option of
always returning to the next instruction.


Xen-devel mailing list



Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.