Re: [PATCH] PCI/ASPM: fix reverse ASPM L0s assignment of upstream and downstream
From: Bjorn Helgaas
Date: Wed May 25 2016 - 13:50:49 EST
On Wed, May 25, 2016 at 01:21:12PM -0400, Sinan Kaya wrote:
> Hi Bjorn,
> OK. I see that we are dealing with two different questions.
> > I thought you were talking about booting with
> > "pcie_aspm.policy=powersave", where pcie_aspm_set_policy() sets
> > aspm_policy = POLICY_POWERSAVE, then configures each link with
> > ASPM_STATE_ALL. But pcie_config_aspm_link() does AND the desired
> > state (ASPM_STATE_ALL) with link->aspm_capable, which only has
> > ASPM_STATE_L0S set if both the upstream and downstream components
> > advertised PCIE_LINK_STATE_L0S.
> > This path is very complicated, but I don't *think* it will enable L0s
> > if the other end of the link doesn't support it.
> Thanks for clarifying this. You are saying that if one side of the
> link doesn't support L0s, Linux doesn't enable L0s on the other side.
> This makes sense.
> I think there is a confusion between what supported means vs. when
> L0s can be enabled on which side of the link.
> You clarified the supported case above that code will not enable
> L0s if the other side doesn't support L0s.
> > Now we enable the downstream component's transmitter to enter L0s.
> > Per PCIe spec r3.0, sec 7.8.7, the receiver, i.e., the upstream
> > component, must be capable of entering L0s even when its transmitter
> > is disabled from entering L0s.
> Let's assume that both sides actually support L0s but the link parameters
> on one side is not compatible.
> You are saying that it is OK to enable L0s on just one side of the
> link as long as both sides support L0s.
I'm not sure what you mean by the link parameters not being
compatible, but I think it is legal to enable L0s on only one
> This part is a little bit misleading. I had HW people telling me
> that both sides need to enable L0s at about the same time.
I don't remember seeing anything like that in the spec. Do they have
a pointer? "At about the same time" is too hand-wavey to be useful to
> I'm actually seeing Linux enabling L0s on one side only as you
> described and both supports L0s.