diff mbox series

[v2,1/3] PCI / ACPI: Use cached ACPI device state to get PCI device power state

Message ID 20190618161858.77834-2-mika.westerberg@linux.intel.com (mailing list archive)
State Superseded, archived
Headers show
Series PCI / ACPI: Handle sibling devices sharing power resources | expand

Commit Message

Mika Westerberg June 18, 2019, 4:18 p.m. UTC
Intel Ice Lake has an integrated Thunderbolt controller which means that
the PCIe topology is extended directly from the two root ports (RP0 and
RP1). Power management is handled by ACPI power resources that are
shared between the root ports, Thunderbolt controller (NHI) and xHCI
controller.

The topology with the power resources (marked with []) looks like:

  Host bridge
    |
    +- RP0 ---\
    +- RP1 ---|--+--> [TBT]
    +- NHI --/   |
    |            |
    |            v
    +- xHCI --> [D3C]

Here TBT and D3C are the shared ACPI power resources. ACPI _PR3() method
returns either TBT or D3C or both.

Say we runtime suspend first the root ports RP0 and RP1, then NHI. Now
since the TBT power resource is still on when the root ports are runtime
suspended their dev->current_state is set to D3hot. When NHI is runtime
suspended TBT is finally turned off but state of the root ports remain
to be D3hot.

If the user now runs lspci for instance, the result is all 1's like in
the below output (07.0 is the first root port, RP0):

00:07.0 PCI bridge: Intel Corporation Device 8a1d (rev ff) (prog-if ff)
    !!! Unknown header type 7f
    Kernel driver in use: pcieport

I short the hardware state is not in sync with the software state
anymore. The exact same thing happens with the PME polling thread which
ends up bringing the root ports back into D0 after they are runtime
suspended.

ACPI core already sets the device state to be D3cold when it drops its
references to the power resources returned by _PR3 even if these power
resources are still physically on (other devices still reference them).
However, in PCI core we call acpi_device_get_power() to figure out the
power state and that returns the "real" power state based on the state
of its power resources.

To make it work with the shared power resources modify
acpi_pci_get_power_state() so that it reads the ACPI device power state
that was cached by the ACPI core. This makes the PCI device power state
match the ACPI device power state regardless of state of the shared
power resources that may still be on at this point.

Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
---
 drivers/pci/pci-acpi.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

Comments

Bjorn Helgaas June 19, 2019, 9:28 p.m. UTC | #1
On Tue, Jun 18, 2019 at 07:18:56PM +0300, Mika Westerberg wrote:
> Intel Ice Lake has an integrated Thunderbolt controller which means that
> the PCIe topology is extended directly from the two root ports (RP0 and
> RP1).

A PCIe topology is always extended directly from root ports,
regardless of whether a Thunderbolt controller is integrated, so I
guess I'm missing the point you're making.  It doesn't sound like this
is anything specific to Thunderbolt?

> Power management is handled by ACPI power resources that are
> shared between the root ports, Thunderbolt controller (NHI) and xHCI
> controller.
> 
> The topology with the power resources (marked with []) looks like:
> 
>   Host bridge
>     |
>     +- RP0 ---\
>     +- RP1 ---|--+--> [TBT]
>     +- NHI --/   |
>     |            |
>     |            v
>     +- xHCI --> [D3C]
> 
> Here TBT and D3C are the shared ACPI power resources. ACPI _PR3() method
> returns either TBT or D3C or both.
> 
> Say we runtime suspend first the root ports RP0 and RP1, then NHI. Now
> since the TBT power resource is still on when the root ports are runtime
> suspended their dev->current_state is set to D3hot. When NHI is runtime
> suspended TBT is finally turned off but state of the root ports remain
> to be D3hot.
> 
> If the user now runs lspci for instance, the result is all 1's like in
> the below output (07.0 is the first root port, RP0):
> 
> 00:07.0 PCI bridge: Intel Corporation Device 8a1d (rev ff) (prog-if ff)
>     !!! Unknown header type 7f
>     Kernel driver in use: pcieport
> 
> I short the hardware state is not in sync with the software state
> anymore. The exact same thing happens with the PME polling thread which
> ends up bringing the root ports back into D0 after they are runtime
> suspended.

s/I /In /
Mika Westerberg June 20, 2019, 8:27 a.m. UTC | #2
On Wed, Jun 19, 2019 at 04:28:01PM -0500, Bjorn Helgaas wrote:
> On Tue, Jun 18, 2019 at 07:18:56PM +0300, Mika Westerberg wrote:
> > Intel Ice Lake has an integrated Thunderbolt controller which means that
> > the PCIe topology is extended directly from the two root ports (RP0 and
> > RP1).
> 
> A PCIe topology is always extended directly from root ports,
> regardless of whether a Thunderbolt controller is integrated, so I
> guess I'm missing the point you're making.  It doesn't sound like this
> is anything specific to Thunderbolt?

The point I'm trying to make here is to explain why this is problem now
and not with the previous discrete controllers. With the previous there
was only a single ACPI power resource for the root port and the
Thunderbolt host router was connected to that root port. PCIe hierarchy
was extended through downstream ports (not root ports) of that
controller (which includes PCIe switch).

Now the thing is part of the SoC so power management is different and
causes problems in Linux.

> > Power management is handled by ACPI power resources that are
> > shared between the root ports, Thunderbolt controller (NHI) and xHCI
> > controller.
> > 
> > The topology with the power resources (marked with []) looks like:
> > 
> >   Host bridge
> >     |
> >     +- RP0 ---\
> >     +- RP1 ---|--+--> [TBT]
> >     +- NHI --/   |
> >     |            |
> >     |            v
> >     +- xHCI --> [D3C]
> > 
> > Here TBT and D3C are the shared ACPI power resources. ACPI _PR3() method
> > returns either TBT or D3C or both.
> > 
> > Say we runtime suspend first the root ports RP0 and RP1, then NHI. Now
> > since the TBT power resource is still on when the root ports are runtime
> > suspended their dev->current_state is set to D3hot. When NHI is runtime
> > suspended TBT is finally turned off but state of the root ports remain
> > to be D3hot.
> > 
> > If the user now runs lspci for instance, the result is all 1's like in
> > the below output (07.0 is the first root port, RP0):
> > 
> > 00:07.0 PCI bridge: Intel Corporation Device 8a1d (rev ff) (prog-if ff)
> >     !!! Unknown header type 7f
> >     Kernel driver in use: pcieport
> > 
> > I short the hardware state is not in sync with the software state
> > anymore. The exact same thing happens with the PME polling thread which
> > ends up bringing the root ports back into D0 after they are runtime
> > suspended.
> 
> s/I /In /

Thanks, I'll fix it.
Bjorn Helgaas June 20, 2019, 1:16 p.m. UTC | #3
On Thu, Jun 20, 2019 at 11:27:30AM +0300, Mika Westerberg wrote:
> On Wed, Jun 19, 2019 at 04:28:01PM -0500, Bjorn Helgaas wrote:
> > On Tue, Jun 18, 2019 at 07:18:56PM +0300, Mika Westerberg wrote:
> > > Intel Ice Lake has an integrated Thunderbolt controller which
> > > means that the PCIe topology is extended directly from the two
> > > root ports (RP0 and RP1).
> > 
> > A PCIe topology is always extended directly from root ports,
> > regardless of whether a Thunderbolt controller is integrated, so I
> > guess I'm missing the point you're making.  It doesn't sound like
> > this is anything specific to Thunderbolt?
>
> The point I'm trying to make here is to explain why this is problem
> now and not with the previous discrete controllers. With the
> previous there was only a single ACPI power resource for the root
> port and the Thunderbolt host router was connected to that root
> port. PCIe hierarchy was extended through downstream ports (not root
> ports) of that controller (which includes PCIe switch).

Sounds like you're using "PCIe topology extension" to mean
specifically something below a Thunderbolt controller, excluding a
subtree below a root port.  I don't think the PCI core is aware of
that distinction.

> Now the thing is part of the SoC so power management is different
> and causes problems in Linux.

The SoC is a physical packaging issue that really doesn't enter into
the specs directly.  I'm trying to get at the logical topology
questions in terms of the PCIe and ACPI specs.

I assume we could dream up a non-Thunderbolt topology that would show
the same problem?

> > > Power management is handled by ACPI power resources that are
> > > shared between the root ports, Thunderbolt controller (NHI) and xHCI
> > > controller.
> > > 
> > > The topology with the power resources (marked with []) looks like:
> > > 
> > >   Host bridge
> > >     |
> > >     +- RP0 ---\
> > >     +- RP1 ---|--+--> [TBT]
> > >     +- NHI --/   |
> > >     |            |
> > >     |            v
> > >     +- xHCI --> [D3C]
> > > 
> > > Here TBT and D3C are the shared ACPI power resources. ACPI
> > > _PR3() method returns either TBT or D3C or both.

I'm not very familiar with _PR3.  I guess this is under an ACPI object
representing a PCI device, e.g., \_SB.PCI0.RP0._PR3?

> > > Say we runtime suspend first the root ports RP0 and RP1, then
> > > NHI. Now since the TBT power resource is still on when the root
> > > ports are runtime suspended their dev->current_state is set to
> > > D3hot. When NHI is runtime suspended TBT is finally turned off
> > > but state of the root ports remain to be D3hot.

So in this example we might have:

  _SB.PCI0.RP0._PR3: TBT
  _SB.PCI0.RP1._PR3: TBT
  _SB.PCI0.NHI._PR3: TBT

And when Linux figures out that everything depending on TBT is in
D3hot, it evaluates TBT._OFF, which puts them all in D3cold?  And part
of the problem is that they're now in D3cold (where config access
doesn't work) but Linux still thinks they're in D3hot (where config
access would work)?

I feel like I'm missing something because I don't know how D3C is
involved, since you didn't mention suspending xHCI.

And I can't mentally match up the patch with the D3hot/D3cold state
change (if indeed that's the problem).  If we were updating the path
that evaluates _OFF so it changed the power state of all dependent
devices, *that* would make a lot of sense to me because it sounds like
that's where the physical change happens that makes things out of
sync.

> > > If the user now runs lspci for instance, the result is all 1's like in
> > > the below output (07.0 is the first root port, RP0):
> > > 
> > > 00:07.0 PCI bridge: Intel Corporation Device 8a1d (rev ff) (prog-if ff)
> > >     !!! Unknown header type 7f
> > >     Kernel driver in use: pcieport
Mika Westerberg June 20, 2019, 1:37 p.m. UTC | #4
On Thu, Jun 20, 2019 at 08:16:49AM -0500, Bjorn Helgaas wrote:
> On Thu, Jun 20, 2019 at 11:27:30AM +0300, Mika Westerberg wrote:
> > On Wed, Jun 19, 2019 at 04:28:01PM -0500, Bjorn Helgaas wrote:
> > > On Tue, Jun 18, 2019 at 07:18:56PM +0300, Mika Westerberg wrote:
> > > > Intel Ice Lake has an integrated Thunderbolt controller which
> > > > means that the PCIe topology is extended directly from the two
> > > > root ports (RP0 and RP1).
> > > 
> > > A PCIe topology is always extended directly from root ports,
> > > regardless of whether a Thunderbolt controller is integrated, so I
> > > guess I'm missing the point you're making.  It doesn't sound like
> > > this is anything specific to Thunderbolt?
> >
> > The point I'm trying to make here is to explain why this is problem
> > now and not with the previous discrete controllers. With the
> > previous there was only a single ACPI power resource for the root
> > port and the Thunderbolt host router was connected to that root
> > port. PCIe hierarchy was extended through downstream ports (not root
> > ports) of that controller (which includes PCIe switch).
> 
> Sounds like you're using "PCIe topology extension" to mean
> specifically something below a Thunderbolt controller, excluding a
> subtree below a root port.  I don't think the PCI core is aware of
> that distinction.

Right it is not.

> > Now the thing is part of the SoC so power management is different
> > and causes problems in Linux.
> 
> The SoC is a physical packaging issue that really doesn't enter into
> the specs directly.  I'm trying to get at the logical topology
> questions in terms of the PCIe and ACPI specs.
> 
> I assume we could dream up a non-Thunderbolt topology that would show
> the same problem?

Yes.

> > > > Power management is handled by ACPI power resources that are
> > > > shared between the root ports, Thunderbolt controller (NHI) and xHCI
> > > > controller.
> > > > 
> > > > The topology with the power resources (marked with []) looks like:
> > > > 
> > > >   Host bridge
> > > >     |
> > > >     +- RP0 ---\
> > > >     +- RP1 ---|--+--> [TBT]
> > > >     +- NHI --/   |
> > > >     |            |
> > > >     |            v
> > > >     +- xHCI --> [D3C]
> > > > 
> > > > Here TBT and D3C are the shared ACPI power resources. ACPI
> > > > _PR3() method returns either TBT or D3C or both.
> 
> I'm not very familiar with _PR3.  I guess this is under an ACPI object
> representing a PCI device, e.g., \_SB.PCI0.RP0._PR3?

Correct.

> > > > Say we runtime suspend first the root ports RP0 and RP1, then
> > > > NHI. Now since the TBT power resource is still on when the root
> > > > ports are runtime suspended their dev->current_state is set to
> > > > D3hot. When NHI is runtime suspended TBT is finally turned off
> > > > but state of the root ports remain to be D3hot.
> 
> So in this example we might have:
> 
>   _SB.PCI0.RP0._PR3: TBT
>   _SB.PCI0.RP1._PR3: TBT
>   _SB.PCI0.NHI._PR3: TBT

and also D3C.

> And when Linux figures out that everything depending on TBT is in
> D3hot, it evaluates TBT._OFF, which puts them all in D3cold?  And part
> of the problem is that they're now in D3cold (where config access
> doesn't work) but Linux still thinks they're in D3hot (where config
> access would work)?

Exactly.

> I feel like I'm missing something because I don't know how D3C is
> involved, since you didn't mention suspending xHCI.

That's another power resource so we will also have D3C turned off when
xHCI gets suspended but I did not want to complicate things too much in
the changelog.

> And I can't mentally match up the patch with the D3hot/D3cold state
> change (if indeed that's the problem).  If we were updating the path
> that evaluates _OFF so it changed the power state of all dependent
> devices, *that* would make a lot of sense to me because it sounds like
> that's where the physical change happens that makes things out of
> sync.

I did that in the first version [1] but Rafael pointed out that it is
racy one way or another [2].

[1] https://www.spinics.net/lists/linux-pci/msg83583.html
[2] https://www.spinics.net/lists/linux-pci/msg83600.html
Bjorn Helgaas June 20, 2019, 2:15 p.m. UTC | #5
On Thu, Jun 20, 2019 at 04:37:10PM +0300, Mika Westerberg wrote:
> On Thu, Jun 20, 2019 at 08:16:49AM -0500, Bjorn Helgaas wrote:
> > On Thu, Jun 20, 2019 at 11:27:30AM +0300, Mika Westerberg wrote:
> > > On Wed, Jun 19, 2019 at 04:28:01PM -0500, Bjorn Helgaas wrote:
> > > > On Tue, Jun 18, 2019 at 07:18:56PM +0300, Mika Westerberg wrote:
> > > > > Intel Ice Lake has an integrated Thunderbolt controller which
> > > > > means that the PCIe topology is extended directly from the two
> > > > > root ports (RP0 and RP1).
> > > > 
> > > > A PCIe topology is always extended directly from root ports,
> > > > regardless of whether a Thunderbolt controller is integrated, so I
> > > > guess I'm missing the point you're making.  It doesn't sound like
> > > > this is anything specific to Thunderbolt?
> > >
> > > The point I'm trying to make here is to explain why this is problem
> > > now and not with the previous discrete controllers. With the
> > > previous there was only a single ACPI power resource for the root
> > > port and the Thunderbolt host router was connected to that root
> > > port. PCIe hierarchy was extended through downstream ports (not root
> > > ports) of that controller (which includes PCIe switch).
> > 
> > Sounds like you're using "PCIe topology extension" to mean
> > specifically something below a Thunderbolt controller, excluding a
> > subtree below a root port.  I don't think the PCI core is aware of
> > that distinction.
> 
> Right it is not.
> 
> > > Now the thing is part of the SoC so power management is different
> > > and causes problems in Linux.
> > 
> > The SoC is a physical packaging issue that really doesn't enter into
> > the specs directly.  I'm trying to get at the logical topology
> > questions in terms of the PCIe and ACPI specs.
> > 
> > I assume we could dream up a non-Thunderbolt topology that would show
> > the same problem?
> 
> Yes.
> 
> > > > > Power management is handled by ACPI power resources that are
> > > > > shared between the root ports, Thunderbolt controller (NHI) and xHCI
> > > > > controller.
> > > > > 
> > > > > The topology with the power resources (marked with []) looks like:
> > > > > 
> > > > >   Host bridge
> > > > >     |
> > > > >     +- RP0 ---\
> > > > >     +- RP1 ---|--+--> [TBT]
> > > > >     +- NHI --/   |
> > > > >     |            |
> > > > >     |            v
> > > > >     +- xHCI --> [D3C]
> > > > > 
> > > > > Here TBT and D3C are the shared ACPI power resources. ACPI
> > > > > _PR3() method returns either TBT or D3C or both.
> > 
> > I'm not very familiar with _PR3.  I guess this is under an ACPI object
> > representing a PCI device, e.g., \_SB.PCI0.RP0._PR3?
> 
> Correct.
> 
> > > > > Say we runtime suspend first the root ports RP0 and RP1, then
> > > > > NHI. Now since the TBT power resource is still on when the root
> > > > > ports are runtime suspended their dev->current_state is set to
> > > > > D3hot. When NHI is runtime suspended TBT is finally turned off
> > > > > but state of the root ports remain to be D3hot.
> > 
> > So in this example we might have:
> > 
> >   _SB.PCI0.RP0._PR3: TBT
> >   _SB.PCI0.RP1._PR3: TBT
> >   _SB.PCI0.NHI._PR3: TBT
> 
> and also D3C.
> 
> > And when Linux figures out that everything depending on TBT is in
> > D3hot, it evaluates TBT._OFF, which puts them all in D3cold?  And part
> > of the problem is that they're now in D3cold (where config access
> > doesn't work) but Linux still thinks they're in D3hot (where config
> > access would work)?
> 
> Exactly.
> 
> > I feel like I'm missing something because I don't know how D3C is
> > involved, since you didn't mention suspending xHCI.
> 
> That's another power resource so we will also have D3C turned off when
> xHCI gets suspended but I did not want to complicate things too much in
> the changelog.

If D3C isn't essential to seeing this problem, you could just omit it
altogether.  I think stripping out anything that's not essential will
make it easier to think about the underlying issues.

> > And I can't mentally match up the patch with the D3hot/D3cold state
> > change (if indeed that's the problem).  If we were updating the path
> > that evaluates _OFF so it changed the power state of all dependent
> > devices, *that* would make a lot of sense to me because it sounds like
> > that's where the physical change happens that makes things out of
> > sync.
> 
> I did that in the first version [1] but Rafael pointed out that it is
> racy one way or another [2].
> 
> [1] https://www.spinics.net/lists/linux-pci/msg83583.html
> [2] https://www.spinics.net/lists/linux-pci/msg83600.html

Yeah, interesting.  It was definitely a much larger patch.  I don't
know enough to comment on the races.  I would wonder whether there's a
way to get rid of the caches that become stale, but that's just an
idle thought, not a suggestion.

Bjorn
Rafael J. Wysocki June 21, 2019, 10:32 a.m. UTC | #6
On Thu, Jun 20, 2019 at 4:15 PM Bjorn Helgaas <helgaas@kernel.org> wrote:
>
> On Thu, Jun 20, 2019 at 04:37:10PM +0300, Mika Westerberg wrote:
> > On Thu, Jun 20, 2019 at 08:16:49AM -0500, Bjorn Helgaas wrote:
> > > On Thu, Jun 20, 2019 at 11:27:30AM +0300, Mika Westerberg wrote:
> > > > On Wed, Jun 19, 2019 at 04:28:01PM -0500, Bjorn Helgaas wrote:
> > > > > On Tue, Jun 18, 2019 at 07:18:56PM +0300, Mika Westerberg wrote:
> > > > > > Intel Ice Lake has an integrated Thunderbolt controller which
> > > > > > means that the PCIe topology is extended directly from the two
> > > > > > root ports (RP0 and RP1).
> > > > >
> > > > > A PCIe topology is always extended directly from root ports,
> > > > > regardless of whether a Thunderbolt controller is integrated, so I
> > > > > guess I'm missing the point you're making.  It doesn't sound like
> > > > > this is anything specific to Thunderbolt?
> > > >
> > > > The point I'm trying to make here is to explain why this is problem
> > > > now and not with the previous discrete controllers. With the
> > > > previous there was only a single ACPI power resource for the root
> > > > port and the Thunderbolt host router was connected to that root
> > > > port. PCIe hierarchy was extended through downstream ports (not root
> > > > ports) of that controller (which includes PCIe switch).
> > >
> > > Sounds like you're using "PCIe topology extension" to mean
> > > specifically something below a Thunderbolt controller, excluding a
> > > subtree below a root port.  I don't think the PCI core is aware of
> > > that distinction.
> >
> > Right it is not.
> >
> > > > Now the thing is part of the SoC so power management is different
> > > > and causes problems in Linux.
> > >
> > > The SoC is a physical packaging issue that really doesn't enter into
> > > the specs directly.  I'm trying to get at the logical topology
> > > questions in terms of the PCIe and ACPI specs.
> > >
> > > I assume we could dream up a non-Thunderbolt topology that would show
> > > the same problem?
> >
> > Yes.
> >
> > > > > > Power management is handled by ACPI power resources that are
> > > > > > shared between the root ports, Thunderbolt controller (NHI) and xHCI
> > > > > > controller.
> > > > > >
> > > > > > The topology with the power resources (marked with []) looks like:
> > > > > >
> > > > > >   Host bridge
> > > > > >     |
> > > > > >     +- RP0 ---\
> > > > > >     +- RP1 ---|--+--> [TBT]
> > > > > >     +- NHI --/   |
> > > > > >     |            |
> > > > > >     |            v
> > > > > >     +- xHCI --> [D3C]
> > > > > >
> > > > > > Here TBT and D3C are the shared ACPI power resources. ACPI
> > > > > > _PR3() method returns either TBT or D3C or both.
> > >
> > > I'm not very familiar with _PR3.  I guess this is under an ACPI object
> > > representing a PCI device, e.g., \_SB.PCI0.RP0._PR3?
> >
> > Correct.
> >
> > > > > > Say we runtime suspend first the root ports RP0 and RP1, then
> > > > > > NHI. Now since the TBT power resource is still on when the root
> > > > > > ports are runtime suspended their dev->current_state is set to
> > > > > > D3hot. When NHI is runtime suspended TBT is finally turned off
> > > > > > but state of the root ports remain to be D3hot.
> > >
> > > So in this example we might have:
> > >
> > >   _SB.PCI0.RP0._PR3: TBT
> > >   _SB.PCI0.RP1._PR3: TBT
> > >   _SB.PCI0.NHI._PR3: TBT
> >
> > and also D3C.
> >
> > > And when Linux figures out that everything depending on TBT is in
> > > D3hot, it evaluates TBT._OFF, which puts them all in D3cold?  And part
> > > of the problem is that they're now in D3cold (where config access
> > > doesn't work) but Linux still thinks they're in D3hot (where config
> > > access would work)?
> >
> > Exactly.
> >
> > > I feel like I'm missing something because I don't know how D3C is
> > > involved, since you didn't mention suspending xHCI.
> >
> > That's another power resource so we will also have D3C turned off when
> > xHCI gets suspended but I did not want to complicate things too much in
> > the changelog.
>
> If D3C isn't essential to seeing this problem, you could just omit it
> altogether.  I think stripping out anything that's not essential will
> make it easier to think about the underlying issues.
>
> > > And I can't mentally match up the patch with the D3hot/D3cold state
> > > change (if indeed that's the problem).  If we were updating the path
> > > that evaluates _OFF so it changed the power state of all dependent
> > > devices, *that* would make a lot of sense to me because it sounds like
> > > that's where the physical change happens that makes things out of
> > > sync.
> >
> > I did that in the first version [1] but Rafael pointed out that it is
> > racy one way or another [2].
> >
> > [1] https://www.spinics.net/lists/linux-pci/msg83583.html
> > [2] https://www.spinics.net/lists/linux-pci/msg83600.html
>
> Yeah, interesting.  It was definitely a much larger patch.  I don't
> know enough to comment on the races.

Say two power resources are listed by _PR3 for one device (because why
not?) and you want to change the device's state to D3cold only if the
two power resources are both "off".  Then, you need some locking (or
equivalent) to synchronize two power resources with each other, so
that you can change the devices state when the last of them goes _OFF.
Currently, there is no such synchronization between power resources
other then the "system_level" value which may not be reliable enough
for this type of use.

Or you can say that the device is in D3cold if at least one of the
power resources is _OFF, but IMO that may not really be consistent
with the view that the "logical" power state of the device should
reflect the physical reality accurately.

> I would wonder whether there's a way to get rid of the caches that become stale,

I guess what you mean is that the "cached" (or rather "logical" or
"expected") power state value may become different from what is
returned by acpi_device_get_power() for the device.

The problem here is that acpi_device_get_power() really only should be
used for two purposes: (1) To initialize adev->power.state, or to
update it via acpi_device_update_power(), and (2) by the
"real_power_state" sysfs attribute (of ACPI device objects).  The
adev->power.state value should be used anywhere else, in principle, so
the Mika's patch is correct.

[Note that adev->power.state cannot be updated after calling
acpi_device_get_power() to the value returned by it without updating
the reference counters of the power resources that are "on" *exactly*
because of the problem at hand here.]

> but that's just an idle thought, not a suggestion.

After the initialization of the ACPI subsystem, the authoritative
source of the ACPI device power state information is
adev->power.state.  The ACPI subsystem is expected to update this
value as needed going forward (including system-wide transitions like
resume from S3).
Rafael J. Wysocki June 21, 2019, 11:56 a.m. UTC | #7
On Tue, Jun 18, 2019 at 6:19 PM Mika Westerberg
<mika.westerberg@linux.intel.com> wrote:
>

Actually, to start with, you can say that the ACPI power state
returned by acpi_device_get_power() may depend on the configuration of
ACPI power resources in the system which may change at any time after
acpi_device_get_power() has returned, unless the reference counters of
the ACPI power resources in question are set to prevent that from
happening.  Thus it is invalid to use acpi_device_get_power() in
acpi_pci_get_power_state() the way it is done now and the value of the
power.state field in the corresponding struct acpi_device object
(which reflects the ACPI power resources reference counting, among
other things) should be used instead.

Then you can describe the particular issue below as an example.

IMO that would explain the rationale better here.

> Intel Ice Lake has an integrated Thunderbolt controller which means that
> the PCIe topology is extended directly from the two root ports (RP0 and
> RP1). Power management is handled by ACPI power resources that are
> shared between the root ports, Thunderbolt controller (NHI) and xHCI
> controller.
>
> The topology with the power resources (marked with []) looks like:
>
>   Host bridge
>     |
>     +- RP0 ---\
>     +- RP1 ---|--+--> [TBT]
>     +- NHI --/   |
>     |            |
>     |            v
>     +- xHCI --> [D3C]
>
> Here TBT and D3C are the shared ACPI power resources. ACPI _PR3() method
> returns either TBT or D3C or both.
>
> Say we runtime suspend first the root ports RP0 and RP1, then NHI. Now
> since the TBT power resource is still on when the root ports are runtime
> suspended their dev->current_state is set to D3hot. When NHI is runtime
> suspended TBT is finally turned off but state of the root ports remain
> to be D3hot.
>
> If the user now runs lspci for instance, the result is all 1's like in
> the below output (07.0 is the first root port, RP0):
>
> 00:07.0 PCI bridge: Intel Corporation Device 8a1d (rev ff) (prog-if ff)
>     !!! Unknown header type 7f
>     Kernel driver in use: pcieport
>
> I short the hardware state is not in sync with the software state
> anymore. The exact same thing happens with the PME polling thread which
> ends up bringing the root ports back into D0 after they are runtime
> suspended.
>
> ACPI core already sets the device state to be D3cold when it drops its
> references to the power resources returned by _PR3 even if these power
> resources are still physically on (other devices still reference them).
> However, in PCI core we call acpi_device_get_power() to figure out the
> power state and that returns the "real" power state based on the state
> of its power resources.
>
> To make it work with the shared power resources modify
> acpi_pci_get_power_state() so that it reads the ACPI device power state
> that was cached by the ACPI core. This makes the PCI device power state
> match the ACPI device power state regardless of state of the shared
> power resources that may still be on at this point.
>
> Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
> ---
>  drivers/pci/pci-acpi.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/pci/pci-acpi.c b/drivers/pci/pci-acpi.c
> index 1897847ceb0c..b782acac26c5 100644
> --- a/drivers/pci/pci-acpi.c
> +++ b/drivers/pci/pci-acpi.c
> @@ -685,7 +685,8 @@ static pci_power_t acpi_pci_get_power_state(struct pci_dev *dev)
>         if (!adev || !acpi_device_power_manageable(adev))
>                 return PCI_UNKNOWN;
>
> -       if (acpi_device_get_power(adev, &state) || state == ACPI_STATE_UNKNOWN)
> +       state = adev->power.state;
> +       if (state == ACPI_STATE_UNKNOWN)
>                 return PCI_UNKNOWN;
>
>         return state_conv[state];
> --
> 2.20.1
>
Bjorn Helgaas June 21, 2019, 1:09 p.m. UTC | #8
On Fri, Jun 21, 2019 at 12:32:22PM +0200, Rafael J. Wysocki wrote:
> On Thu, Jun 20, 2019 at 4:15 PM Bjorn Helgaas <helgaas@kernel.org> wrote:
> > On Thu, Jun 20, 2019 at 04:37:10PM +0300, Mika Westerberg wrote:
> > > On Thu, Jun 20, 2019 at 08:16:49AM -0500, Bjorn Helgaas wrote:
> > > > On Thu, Jun 20, 2019 at 11:27:30AM +0300, Mika Westerberg wrote:
> > > > > On Wed, Jun 19, 2019 at 04:28:01PM -0500, Bjorn Helgaas wrote:
> > > > > > On Tue, Jun 18, 2019 at 07:18:56PM +0300, Mika Westerberg wrote:
> > > > > > > Intel Ice Lake has an integrated Thunderbolt controller which
> > > > > > > means that the PCIe topology is extended directly from the two
> > > > > > > root ports (RP0 and RP1).
> > > > > >
> > > > > > A PCIe topology is always extended directly from root ports,
> > > > > > regardless of whether a Thunderbolt controller is integrated, so I
> > > > > > guess I'm missing the point you're making.  It doesn't sound like
> > > > > > this is anything specific to Thunderbolt?
> > > > >
> > > > > The point I'm trying to make here is to explain why this is problem
> > > > > now and not with the previous discrete controllers. With the
> > > > > previous there was only a single ACPI power resource for the root
> > > > > port and the Thunderbolt host router was connected to that root
> > > > > port. PCIe hierarchy was extended through downstream ports (not root
> > > > > ports) of that controller (which includes PCIe switch).
> > > >
> > > > Sounds like you're using "PCIe topology extension" to mean
> > > > specifically something below a Thunderbolt controller, excluding a
> > > > subtree below a root port.  I don't think the PCI core is aware of
> > > > that distinction.
> > >
> > > Right it is not.
> > >
> > > > > Now the thing is part of the SoC so power management is different
> > > > > and causes problems in Linux.
> > > >
> > > > The SoC is a physical packaging issue that really doesn't enter into
> > > > the specs directly.  I'm trying to get at the logical topology
> > > > questions in terms of the PCIe and ACPI specs.
> > > >
> > > > I assume we could dream up a non-Thunderbolt topology that would show
> > > > the same problem?
> > >
> > > Yes.
> > >
> > > > > > > Power management is handled by ACPI power resources that are
> > > > > > > shared between the root ports, Thunderbolt controller (NHI) and xHCI
> > > > > > > controller.
> > > > > > >
> > > > > > > The topology with the power resources (marked with []) looks like:
> > > > > > >
> > > > > > >   Host bridge
> > > > > > >     |
> > > > > > >     +- RP0 ---\
> > > > > > >     +- RP1 ---|--+--> [TBT]
> > > > > > >     +- NHI --/   |
> > > > > > >     |            |
> > > > > > >     |            v
> > > > > > >     +- xHCI --> [D3C]
> > > > > > >
> > > > > > > Here TBT and D3C are the shared ACPI power resources. ACPI
> > > > > > > _PR3() method returns either TBT or D3C or both.
> > > >
> > > > I'm not very familiar with _PR3.  I guess this is under an ACPI object
> > > > representing a PCI device, e.g., \_SB.PCI0.RP0._PR3?
> > >
> > > Correct.
> > >
> > > > > > > Say we runtime suspend first the root ports RP0 and RP1, then
> > > > > > > NHI. Now since the TBT power resource is still on when the root
> > > > > > > ports are runtime suspended their dev->current_state is set to
> > > > > > > D3hot. When NHI is runtime suspended TBT is finally turned off
> > > > > > > but state of the root ports remain to be D3hot.
> > > >
> > > > So in this example we might have:
> > > >
> > > >   _SB.PCI0.RP0._PR3: TBT
> > > >   _SB.PCI0.RP1._PR3: TBT
> > > >   _SB.PCI0.NHI._PR3: TBT
> > >
> > > and also D3C.
> > >
> > > > And when Linux figures out that everything depending on TBT is in
> > > > D3hot, it evaluates TBT._OFF, which puts them all in D3cold?  And part
> > > > of the problem is that they're now in D3cold (where config access
> > > > doesn't work) but Linux still thinks they're in D3hot (where config
> > > > access would work)?
> > >
> > > Exactly.
> > >
> > > > I feel like I'm missing something because I don't know how D3C is
> > > > involved, since you didn't mention suspending xHCI.
> > >
> > > That's another power resource so we will also have D3C turned off when
> > > xHCI gets suspended but I did not want to complicate things too much in
> > > the changelog.
> >
> > If D3C isn't essential to seeing this problem, you could just omit it
> > altogether.  I think stripping out anything that's not essential will
> > make it easier to think about the underlying issues.
> >
> > > > And I can't mentally match up the patch with the D3hot/D3cold state
> > > > change (if indeed that's the problem).  If we were updating the path
> > > > that evaluates _OFF so it changed the power state of all dependent
> > > > devices, *that* would make a lot of sense to me because it sounds like
> > > > that's where the physical change happens that makes things out of
> > > > sync.
> > >
> > > I did that in the first version [1] but Rafael pointed out that it is
> > > racy one way or another [2].
> > >
> > > [1] https://www.spinics.net/lists/linux-pci/msg83583.html
> > > [2] https://www.spinics.net/lists/linux-pci/msg83600.html
> >
> > Yeah, interesting.  It was definitely a much larger patch.  I don't
> > know enough to comment on the races.
> 
> Say two power resources are listed by _PR3 for one device (because why
> not?) and you want to change the device's state to D3cold only if the
> two power resources are both "off".  Then, you need some locking (or
> equivalent) to synchronize two power resources with each other, so
> that you can change the devices state when the last of them goes _OFF.
> Currently, there is no such synchronization between power resources
> other then the "system_level" value which may not be reliable enough
> for this type of use.
> 
> Or you can say that the device is in D3cold if at least one of the
> power resources is _OFF, but IMO that may not really be consistent
> with the view that the "logical" power state of the device should
> reflect the physical reality accurately.
> 
> > I would wonder whether there's a way to get rid of the caches that become stale,
> 
> I guess what you mean is that the "cached" (or rather "logical" or
> "expected") power state value may become different from what is
> returned by acpi_device_get_power() for the device.
> 
> The problem here is that acpi_device_get_power() really only should be
> used for two purposes: (1) To initialize adev->power.state, or to
> update it via acpi_device_update_power(), and (2) by the
> "real_power_state" sysfs attribute (of ACPI device objects).  The
> adev->power.state value should be used anywhere else, in principle, so
> the Mika's patch is correct.
> 
> [Note that adev->power.state cannot be updated after calling
> acpi_device_get_power() to the value returned by it without updating
> the reference counters of the power resources that are "on" *exactly*
> because of the problem at hand here.]
> 
> > but that's just an idle thought, not a suggestion.
> 
> After the initialization of the ACPI subsystem, the authoritative
> source of the ACPI device power state information is
> adev->power.state.  The ACPI subsystem is expected to update this
> value as needed going forward (including system-wide transitions like
> resume from S3).

Thanks, this is all very helpful!  Do you by any chance add
lore.kernel.org links to commit logs when applying patches?  This is a
case where I think the discussion could be useful in the future.

Link: https://lore.kernel.org/r/20190618161858.77834-2-mika.westerberg@linux.intel.com
Rafael J. Wysocki June 22, 2019, 8:51 a.m. UTC | #9
On Friday, June 21, 2019 3:09:20 PM CEST Bjorn Helgaas wrote:
> On Fri, Jun 21, 2019 at 12:32:22PM +0200, Rafael J. Wysocki wrote:
> > On Thu, Jun 20, 2019 at 4:15 PM Bjorn Helgaas <helgaas@kernel.org> wrote:
> > > On Thu, Jun 20, 2019 at 04:37:10PM +0300, Mika Westerberg wrote:
> > > > On Thu, Jun 20, 2019 at 08:16:49AM -0500, Bjorn Helgaas wrote:
> > > > > On Thu, Jun 20, 2019 at 11:27:30AM +0300, Mika Westerberg wrote:
> > > > > > On Wed, Jun 19, 2019 at 04:28:01PM -0500, Bjorn Helgaas wrote:
> > > > > > > On Tue, Jun 18, 2019 at 07:18:56PM +0300, Mika Westerberg wrote:
> > > > > > > > Intel Ice Lake has an integrated Thunderbolt controller which
> > > > > > > > means that the PCIe topology is extended directly from the two
> > > > > > > > root ports (RP0 and RP1).
> > > > > > >
> > > > > > > A PCIe topology is always extended directly from root ports,
> > > > > > > regardless of whether a Thunderbolt controller is integrated, so I
> > > > > > > guess I'm missing the point you're making.  It doesn't sound like
> > > > > > > this is anything specific to Thunderbolt?
> > > > > >
> > > > > > The point I'm trying to make here is to explain why this is problem
> > > > > > now and not with the previous discrete controllers. With the
> > > > > > previous there was only a single ACPI power resource for the root
> > > > > > port and the Thunderbolt host router was connected to that root
> > > > > > port. PCIe hierarchy was extended through downstream ports (not root
> > > > > > ports) of that controller (which includes PCIe switch).
> > > > >
> > > > > Sounds like you're using "PCIe topology extension" to mean
> > > > > specifically something below a Thunderbolt controller, excluding a
> > > > > subtree below a root port.  I don't think the PCI core is aware of
> > > > > that distinction.
> > > >
> > > > Right it is not.
> > > >
> > > > > > Now the thing is part of the SoC so power management is different
> > > > > > and causes problems in Linux.
> > > > >
> > > > > The SoC is a physical packaging issue that really doesn't enter into
> > > > > the specs directly.  I'm trying to get at the logical topology
> > > > > questions in terms of the PCIe and ACPI specs.
> > > > >
> > > > > I assume we could dream up a non-Thunderbolt topology that would show
> > > > > the same problem?
> > > >
> > > > Yes.
> > > >
> > > > > > > > Power management is handled by ACPI power resources that are
> > > > > > > > shared between the root ports, Thunderbolt controller (NHI) and xHCI
> > > > > > > > controller.
> > > > > > > >
> > > > > > > > The topology with the power resources (marked with []) looks like:
> > > > > > > >
> > > > > > > >   Host bridge
> > > > > > > >     |
> > > > > > > >     +- RP0 ---\
> > > > > > > >     +- RP1 ---|--+--> [TBT]
> > > > > > > >     +- NHI --/   |
> > > > > > > >     |            |
> > > > > > > >     |            v
> > > > > > > >     +- xHCI --> [D3C]
> > > > > > > >
> > > > > > > > Here TBT and D3C are the shared ACPI power resources. ACPI
> > > > > > > > _PR3() method returns either TBT or D3C or both.
> > > > >
> > > > > I'm not very familiar with _PR3.  I guess this is under an ACPI object
> > > > > representing a PCI device, e.g., \_SB.PCI0.RP0._PR3?
> > > >
> > > > Correct.
> > > >
> > > > > > > > Say we runtime suspend first the root ports RP0 and RP1, then
> > > > > > > > NHI. Now since the TBT power resource is still on when the root
> > > > > > > > ports are runtime suspended their dev->current_state is set to
> > > > > > > > D3hot. When NHI is runtime suspended TBT is finally turned off
> > > > > > > > but state of the root ports remain to be D3hot.
> > > > >
> > > > > So in this example we might have:
> > > > >
> > > > >   _SB.PCI0.RP0._PR3: TBT
> > > > >   _SB.PCI0.RP1._PR3: TBT
> > > > >   _SB.PCI0.NHI._PR3: TBT
> > > >
> > > > and also D3C.
> > > >
> > > > > And when Linux figures out that everything depending on TBT is in
> > > > > D3hot, it evaluates TBT._OFF, which puts them all in D3cold?  And part
> > > > > of the problem is that they're now in D3cold (where config access
> > > > > doesn't work) but Linux still thinks they're in D3hot (where config
> > > > > access would work)?
> > > >
> > > > Exactly.
> > > >
> > > > > I feel like I'm missing something because I don't know how D3C is
> > > > > involved, since you didn't mention suspending xHCI.
> > > >
> > > > That's another power resource so we will also have D3C turned off when
> > > > xHCI gets suspended but I did not want to complicate things too much in
> > > > the changelog.
> > >
> > > If D3C isn't essential to seeing this problem, you could just omit it
> > > altogether.  I think stripping out anything that's not essential will
> > > make it easier to think about the underlying issues.
> > >
> > > > > And I can't mentally match up the patch with the D3hot/D3cold state
> > > > > change (if indeed that's the problem).  If we were updating the path
> > > > > that evaluates _OFF so it changed the power state of all dependent
> > > > > devices, *that* would make a lot of sense to me because it sounds like
> > > > > that's where the physical change happens that makes things out of
> > > > > sync.
> > > >
> > > > I did that in the first version [1] but Rafael pointed out that it is
> > > > racy one way or another [2].
> > > >
> > > > [1] https://www.spinics.net/lists/linux-pci/msg83583.html
> > > > [2] https://www.spinics.net/lists/linux-pci/msg83600.html
> > >
> > > Yeah, interesting.  It was definitely a much larger patch.  I don't
> > > know enough to comment on the races.
> > 
> > Say two power resources are listed by _PR3 for one device (because why
> > not?) and you want to change the device's state to D3cold only if the
> > two power resources are both "off".  Then, you need some locking (or
> > equivalent) to synchronize two power resources with each other, so
> > that you can change the devices state when the last of them goes _OFF.
> > Currently, there is no such synchronization between power resources
> > other then the "system_level" value which may not be reliable enough
> > for this type of use.
> > 
> > Or you can say that the device is in D3cold if at least one of the
> > power resources is _OFF, but IMO that may not really be consistent
> > with the view that the "logical" power state of the device should
> > reflect the physical reality accurately.
> > 
> > > I would wonder whether there's a way to get rid of the caches that become stale,
> > 
> > I guess what you mean is that the "cached" (or rather "logical" or
> > "expected") power state value may become different from what is
> > returned by acpi_device_get_power() for the device.
> > 
> > The problem here is that acpi_device_get_power() really only should be
> > used for two purposes: (1) To initialize adev->power.state, or to
> > update it via acpi_device_update_power(), and (2) by the
> > "real_power_state" sysfs attribute (of ACPI device objects).  The
> > adev->power.state value should be used anywhere else, in principle, so
> > the Mika's patch is correct.
> > 
> > [Note that adev->power.state cannot be updated after calling
> > acpi_device_get_power() to the value returned by it without updating
> > the reference counters of the power resources that are "on" *exactly*
> > because of the problem at hand here.]
> > 
> > > but that's just an idle thought, not a suggestion.
> > 
> > After the initialization of the ACPI subsystem, the authoritative
> > source of the ACPI device power state information is
> > adev->power.state.  The ACPI subsystem is expected to update this
> > value as needed going forward (including system-wide transitions like
> > resume from S3).
> 
> Thanks, this is all very helpful!  Do you by any chance add
> lore.kernel.org links to commit logs when applying patches?  This is a
> case where I think the discussion could be useful in the future.
> 
> Link: https://lore.kernel.org/r/20190618161858.77834-2-mika.westerberg@linux.intel.com

Agreed, and thanks for the URL.

I guess Mika can add this tag to the patch changelog.
Mika Westerberg June 24, 2019, 10:57 a.m. UTC | #10
On Sat, Jun 22, 2019 at 10:51:28AM +0200, Rafael J. Wysocki wrote:
> > Thanks, this is all very helpful!  Do you by any chance add
> > lore.kernel.org links to commit logs when applying patches?  This is a
> > case where I think the discussion could be useful in the future.
> > 
> > Link: https://lore.kernel.org/r/20190618161858.77834-2-mika.westerberg@linux.intel.com
> 
> Agreed, and thanks for the URL.
> 
> I guess Mika can add this tag to the patch changelog.

Sure I'll add it.
Mika Westerberg June 24, 2019, 10:58 a.m. UTC | #11
On Fri, Jun 21, 2019 at 01:56:49PM +0200, Rafael J. Wysocki wrote:
> On Tue, Jun 18, 2019 at 6:19 PM Mika Westerberg
> <mika.westerberg@linux.intel.com> wrote:
> >
> 
> Actually, to start with, you can say that the ACPI power state
> returned by acpi_device_get_power() may depend on the configuration of
> ACPI power resources in the system which may change at any time after
> acpi_device_get_power() has returned, unless the reference counters of
> the ACPI power resources in question are set to prevent that from
> happening.  Thus it is invalid to use acpi_device_get_power() in
> acpi_pci_get_power_state() the way it is done now and the value of the
> power.state field in the corresponding struct acpi_device object
> (which reflects the ACPI power resources reference counting, among
> other things) should be used instead.
> 
> Then you can describe the particular issue below as an example.
> 
> IMO that would explain the rationale better here.

Thanks! I'll update the changelog accordingly.
Rafael J. Wysocki June 24, 2019, 11:14 a.m. UTC | #12
On Friday, June 21, 2019 12:32:22 PM CEST Rafael J. Wysocki wrote:
> On Thu, Jun 20, 2019 at 4:15 PM Bjorn Helgaas <helgaas@kernel.org> wrote:
> >
> > On Thu, Jun 20, 2019 at 04:37:10PM +0300, Mika Westerberg wrote:
> > > On Thu, Jun 20, 2019 at 08:16:49AM -0500, Bjorn Helgaas wrote:
> > > > On Thu, Jun 20, 2019 at 11:27:30AM +0300, Mika Westerberg wrote:
> > > > > On Wed, Jun 19, 2019 at 04:28:01PM -0500, Bjorn Helgaas wrote:
> > > > > > On Tue, Jun 18, 2019 at 07:18:56PM +0300, Mika Westerberg wrote:
> > > > > > > Intel Ice Lake has an integrated Thunderbolt controller which
> > > > > > > means that the PCIe topology is extended directly from the two
> > > > > > > root ports (RP0 and RP1).
> > > > > >
> > > > > > A PCIe topology is always extended directly from root ports,
> > > > > > regardless of whether a Thunderbolt controller is integrated, so I
> > > > > > guess I'm missing the point you're making.  It doesn't sound like
> > > > > > this is anything specific to Thunderbolt?
> > > > >
> > > > > The point I'm trying to make here is to explain why this is problem
> > > > > now and not with the previous discrete controllers. With the
> > > > > previous there was only a single ACPI power resource for the root
> > > > > port and the Thunderbolt host router was connected to that root
> > > > > port. PCIe hierarchy was extended through downstream ports (not root
> > > > > ports) of that controller (which includes PCIe switch).
> > > >
> > > > Sounds like you're using "PCIe topology extension" to mean
> > > > specifically something below a Thunderbolt controller, excluding a
> > > > subtree below a root port.  I don't think the PCI core is aware of
> > > > that distinction.
> > >
> > > Right it is not.
> > >
> > > > > Now the thing is part of the SoC so power management is different
> > > > > and causes problems in Linux.
> > > >
> > > > The SoC is a physical packaging issue that really doesn't enter into
> > > > the specs directly.  I'm trying to get at the logical topology
> > > > questions in terms of the PCIe and ACPI specs.
> > > >
> > > > I assume we could dream up a non-Thunderbolt topology that would show
> > > > the same problem?
> > >
> > > Yes.
> > >
> > > > > > > Power management is handled by ACPI power resources that are
> > > > > > > shared between the root ports, Thunderbolt controller (NHI) and xHCI
> > > > > > > controller.
> > > > > > >
> > > > > > > The topology with the power resources (marked with []) looks like:
> > > > > > >
> > > > > > >   Host bridge
> > > > > > >     |
> > > > > > >     +- RP0 ---\
> > > > > > >     +- RP1 ---|--+--> [TBT]
> > > > > > >     +- NHI --/   |
> > > > > > >     |            |
> > > > > > >     |            v
> > > > > > >     +- xHCI --> [D3C]
> > > > > > >
> > > > > > > Here TBT and D3C are the shared ACPI power resources. ACPI
> > > > > > > _PR3() method returns either TBT or D3C or both.
> > > >
> > > > I'm not very familiar with _PR3.  I guess this is under an ACPI object
> > > > representing a PCI device, e.g., \_SB.PCI0.RP0._PR3?
> > >
> > > Correct.
> > >
> > > > > > > Say we runtime suspend first the root ports RP0 and RP1, then
> > > > > > > NHI. Now since the TBT power resource is still on when the root
> > > > > > > ports are runtime suspended their dev->current_state is set to
> > > > > > > D3hot. When NHI is runtime suspended TBT is finally turned off
> > > > > > > but state of the root ports remain to be D3hot.
> > > >
> > > > So in this example we might have:
> > > >
> > > >   _SB.PCI0.RP0._PR3: TBT
> > > >   _SB.PCI0.RP1._PR3: TBT
> > > >   _SB.PCI0.NHI._PR3: TBT
> > >
> > > and also D3C.
> > >
> > > > And when Linux figures out that everything depending on TBT is in
> > > > D3hot, it evaluates TBT._OFF, which puts them all in D3cold?  And part
> > > > of the problem is that they're now in D3cold (where config access
> > > > doesn't work) but Linux still thinks they're in D3hot (where config
> > > > access would work)?
> > >
> > > Exactly.
> > >
> > > > I feel like I'm missing something because I don't know how D3C is
> > > > involved, since you didn't mention suspending xHCI.
> > >
> > > That's another power resource so we will also have D3C turned off when
> > > xHCI gets suspended but I did not want to complicate things too much in
> > > the changelog.
> >
> > If D3C isn't essential to seeing this problem, you could just omit it
> > altogether.  I think stripping out anything that's not essential will
> > make it easier to think about the underlying issues.
> >
> > > > And I can't mentally match up the patch with the D3hot/D3cold state
> > > > change (if indeed that's the problem).  If we were updating the path
> > > > that evaluates _OFF so it changed the power state of all dependent
> > > > devices, *that* would make a lot of sense to me because it sounds like
> > > > that's where the physical change happens that makes things out of
> > > > sync.
> > >
> > > I did that in the first version [1] but Rafael pointed out that it is
> > > racy one way or another [2].
> > >
> > > [1] https://www.spinics.net/lists/linux-pci/msg83583.html
> > > [2] https://www.spinics.net/lists/linux-pci/msg83600.html
> >
> > Yeah, interesting.  It was definitely a much larger patch.  I don't
> > know enough to comment on the races.
> 
> Say two power resources are listed by _PR3 for one device (because why
> not?) and you want to change the device's state to D3cold only if the
> two power resources are both "off".  Then, you need some locking (or
> equivalent) to synchronize two power resources with each other, so
> that you can change the devices state when the last of them goes _OFF.
> Currently, there is no such synchronization between power resources
> other then the "system_level" value which may not be reliable enough
> for this type of use.
> 
> Or you can say that the device is in D3cold if at least one of the
> power resources is _OFF, but IMO that may not really be consistent
> with the view that the "logical" power state of the device should
> reflect the physical reality accurately.
> 
> > I would wonder whether there's a way to get rid of the caches that become stale,
> 
> I guess what you mean is that the "cached" (or rather "logical" or
> "expected") power state value may become different from what is
> returned by acpi_device_get_power() for the device.
> 
> The problem here is that acpi_device_get_power() really only should be
> used for two purposes: (1) To initialize adev->power.state, or to
> update it via acpi_device_update_power(), and (2) by the
> "real_power_state" sysfs attribute (of ACPI device objects).  The
> adev->power.state value should be used anywhere else, in principle, so
> the Mika's patch is correct.

Well, it is an improvement, but it is not sufficient.

> [Note that adev->power.state cannot be updated after calling
> acpi_device_get_power() to the value returned by it without updating
> the reference counters of the power resources that are "on" *exactly*
> because of the problem at hand here.]

That is obviously correct, but ->

> > but that's just an idle thought, not a suggestion.
> 
> After the initialization of the ACPI subsystem, the authoritative
> source of the ACPI device power state information is
> adev->power.state.  The ACPI subsystem is expected to update this
> value as needed going forward (including system-wide transitions like
> resume from S3).

-> the "resume from S3 or hibernation" case needs special handling, because
in that case the device power state need not reflect the information the ACPI
subsystem has.  That only matters if adev->power.state is ACPI_STATE_D0 and
the device is actually *not* in D0, because in that case acpi_device_set_power()
will not work.  So that case is not covered currently (it should be rare in practice,
though, if it happens at all), so something like the patch below (untested) may
be needed in addition to the Mika's patch.

Still, there is also the "power state not matching" case in pci_pm_complete() that's
need to be covered and the non-PCI ACPI PM has a similar issue in theory, so I
need to think about this a bit more.

---
 drivers/pci/pci-acpi.c |   14 +++++++++++++-
 drivers/pci/pci-mid.c  |    3 ++-
 drivers/pci/pci.c      |    9 +++++----
 drivers/pci/pci.h      |    2 +-
 4 files changed, 21 insertions(+), 7 deletions(-)

Index: linux-pm/drivers/pci/pci-acpi.c
===================================================================
--- linux-pm.orig/drivers/pci/pci-acpi.c
+++ linux-pm/drivers/pci/pci-acpi.c
@@ -632,7 +632,8 @@ static bool acpi_pci_power_manageable(st
 	return adev ? acpi_device_power_manageable(adev) : false;
 }
 
-static int acpi_pci_set_power_state(struct pci_dev *dev, pci_power_t state)
+static int acpi_pci_set_power_state(struct pci_dev *dev, pci_power_t state,
+				    bool force)
 {
 	struct acpi_device *adev = ACPI_COMPANION(&dev->dev);
 	static const u8 state_conv[] = {
@@ -657,6 +658,17 @@ static int acpi_pci_set_power_state(stru
 		}
 		/* Fall through */
 	case PCI_D0:
+		if (force) {
+			int acpi_state;
+
+			error = acpi_device_update_power(adev, &acpi_state);
+			if (error)
+				return error;
+
+			if (acpi_state == ACPI_STATE_D0)
+				return 0;
+		}
+		/* fall through */
 	case PCI_D1:
 	case PCI_D2:
 	case PCI_D3hot:
Index: linux-pm/drivers/pci/pci-mid.c
===================================================================
--- linux-pm.orig/drivers/pci/pci-mid.c
+++ linux-pm/drivers/pci/pci-mid.c
@@ -21,7 +21,8 @@ static bool mid_pci_power_manageable(str
 	return true;
 }
 
-static int mid_pci_set_power_state(struct pci_dev *pdev, pci_power_t state)
+static int mid_pci_set_power_state(struct pci_dev *pdev, pci_power_t state,
+				   bool not_used)
 {
 	return intel_mid_pci_set_power_state(pdev, state);
 }
Index: linux-pm/drivers/pci/pci.c
===================================================================
--- linux-pm.orig/drivers/pci/pci.c
+++ linux-pm/drivers/pci/pci.c
@@ -767,9 +767,10 @@ static inline bool platform_pci_power_ma
 }
 
 static inline int platform_pci_set_power_state(struct pci_dev *dev,
-					       pci_power_t t)
+					       pci_power_t state, bool force)
 {
-	return pci_platform_pm ? pci_platform_pm->set_state(dev, t) : -ENOSYS;
+	return pci_platform_pm ?
+		pci_platform_pm->set_state(dev, state, force) : -ENOSYS;
 }
 
 static inline pci_power_t platform_pci_get_power_state(struct pci_dev *dev)
@@ -944,7 +945,7 @@ void pci_update_current_state(struct pci
 void pci_power_up(struct pci_dev *dev)
 {
 	if (platform_pci_power_manageable(dev))
-		platform_pci_set_power_state(dev, PCI_D0);
+		platform_pci_set_power_state(dev, PCI_D0, true);
 
 	pci_raw_set_power_state(dev, PCI_D0);
 	pci_update_current_state(dev, PCI_D0);
@@ -960,7 +961,7 @@ static int pci_platform_power_transition
 	int error;
 
 	if (platform_pci_power_manageable(dev)) {
-		error = platform_pci_set_power_state(dev, state);
+		error = platform_pci_set_power_state(dev, state, false);
 		if (!error)
 			pci_update_current_state(dev, state);
 	} else
Index: linux-pm/drivers/pci/pci.h
===================================================================
--- linux-pm.orig/drivers/pci/pci.h
+++ linux-pm/drivers/pci/pci.h
@@ -67,7 +67,7 @@ int pci_bus_error_reset(struct pci_dev *
 struct pci_platform_pm_ops {
 	bool (*bridge_d3)(struct pci_dev *dev);
 	bool (*is_manageable)(struct pci_dev *dev);
-	int (*set_state)(struct pci_dev *dev, pci_power_t state);
+	int (*set_state)(struct pci_dev *dev, pci_power_t state, bool force);
 	pci_power_t (*get_state)(struct pci_dev *dev);
 	pci_power_t (*choose_state)(struct pci_dev *dev);
 	int (*set_wakeup)(struct pci_dev *dev, bool enable);
Mika Westerberg June 25, 2019, 9:45 a.m. UTC | #13
On Mon, Jun 24, 2019 at 01:14:47PM +0200, Rafael J. Wysocki wrote:
> > The problem here is that acpi_device_get_power() really only should be
> > used for two purposes: (1) To initialize adev->power.state, or to
> > update it via acpi_device_update_power(), and (2) by the
> > "real_power_state" sysfs attribute (of ACPI device objects).  The
> > adev->power.state value should be used anywhere else, in principle, so
> > the Mika's patch is correct.
> 
> Well, it is an improvement, but it is not sufficient.
> 
> > [Note that adev->power.state cannot be updated after calling
> > acpi_device_get_power() to the value returned by it without updating
> > the reference counters of the power resources that are "on" *exactly*
> > because of the problem at hand here.]
> 
> That is obviously correct, but ->
> 
> > > but that's just an idle thought, not a suggestion.
> > 
> > After the initialization of the ACPI subsystem, the authoritative
> > source of the ACPI device power state information is
> > adev->power.state.  The ACPI subsystem is expected to update this
> > value as needed going forward (including system-wide transitions like
> > resume from S3).
> 
> -> the "resume from S3 or hibernation" case needs special handling, because
> in that case the device power state need not reflect the information the ACPI
> subsystem has.  That only matters if adev->power.state is ACPI_STATE_D0 and
> the device is actually *not* in D0, because in that case acpi_device_set_power()
> will not work. 

I guess you are talking about the special-cased devices that we leave in
D0 when system suspend (via firmware) is entered?

> So that case is not covered currently (it should be rare in practice,
> though, if it happens at all), so something like the patch below (untested) may
> be needed in addition to the Mika's patch.

Looks good to me.

> Still, there is also the "power state not matching" case in pci_pm_complete() that's
> need to be covered and the non-PCI ACPI PM has a similar issue in theory, so I
> need to think about this a bit more.

Do you want me to hold off sending an updated version of the patch
series while we figure this one out or is it fine if I send it out now
and we can add further details on top?
Rafael J. Wysocki June 25, 2019, 10 a.m. UTC | #14
On Tue, Jun 25, 2019 at 11:46 AM Mika Westerberg
<mika.westerberg@linux.intel.com> wrote:
>
> On Mon, Jun 24, 2019 at 01:14:47PM +0200, Rafael J. Wysocki wrote:
> > > The problem here is that acpi_device_get_power() really only should be
> > > used for two purposes: (1) To initialize adev->power.state, or to
> > > update it via acpi_device_update_power(), and (2) by the
> > > "real_power_state" sysfs attribute (of ACPI device objects).  The
> > > adev->power.state value should be used anywhere else, in principle, so
> > > the Mika's patch is correct.
> >
> > Well, it is an improvement, but it is not sufficient.
> >
> > > [Note that adev->power.state cannot be updated after calling
> > > acpi_device_get_power() to the value returned by it without updating
> > > the reference counters of the power resources that are "on" *exactly*
> > > because of the problem at hand here.]
> >
> > That is obviously correct, but ->
> >
> > > > but that's just an idle thought, not a suggestion.
> > >
> > > After the initialization of the ACPI subsystem, the authoritative
> > > source of the ACPI device power state information is
> > > adev->power.state.  The ACPI subsystem is expected to update this
> > > value as needed going forward (including system-wide transitions like
> > > resume from S3).
> >
> > -> the "resume from S3 or hibernation" case needs special handling, because
> > in that case the device power state need not reflect the information the ACPI
> > subsystem has.  That only matters if adev->power.state is ACPI_STATE_D0 and
> > the device is actually *not* in D0, because in that case acpi_device_set_power()
> > will not work.
>
> I guess you are talking about the special-cased devices that we leave in
> D0 when system suspend (via firmware) is entered?
>
> > So that case is not covered currently (it should be rare in practice,
> > though, if it happens at all), so something like the patch below (untested) may
> > be needed in addition to the Mika's patch.
>
> Looks good to me.

I actually decided to address this issue in acpi_device_set_power() as
it may affect devices beyond PCI in principle.  I will send a patch
for that shortly.

> > Still, there is also the "power state not matching" case in pci_pm_complete() that's
> > need to be covered and the non-PCI ACPI PM has a similar issue in theory, so I
> > need to think about this a bit more.
>
> Do you want me to hold off sending an updated version of the patch
> series while we figure this one out or is it fine if I send it out now
> and we can add further details on top?

It is independent of the other fix, so it can be sent now just fine IMO.
Mika Westerberg June 25, 2019, 10:08 a.m. UTC | #15
On Tue, Jun 25, 2019 at 12:00:57PM +0200, Rafael J. Wysocki wrote:
> On Tue, Jun 25, 2019 at 11:46 AM Mika Westerberg
> <mika.westerberg@linux.intel.com> wrote:
> >
> > On Mon, Jun 24, 2019 at 01:14:47PM +0200, Rafael J. Wysocki wrote:
> > > > The problem here is that acpi_device_get_power() really only should be
> > > > used for two purposes: (1) To initialize adev->power.state, or to
> > > > update it via acpi_device_update_power(), and (2) by the
> > > > "real_power_state" sysfs attribute (of ACPI device objects).  The
> > > > adev->power.state value should be used anywhere else, in principle, so
> > > > the Mika's patch is correct.
> > >
> > > Well, it is an improvement, but it is not sufficient.
> > >
> > > > [Note that adev->power.state cannot be updated after calling
> > > > acpi_device_get_power() to the value returned by it without updating
> > > > the reference counters of the power resources that are "on" *exactly*
> > > > because of the problem at hand here.]
> > >
> > > That is obviously correct, but ->
> > >
> > > > > but that's just an idle thought, not a suggestion.
> > > >
> > > > After the initialization of the ACPI subsystem, the authoritative
> > > > source of the ACPI device power state information is
> > > > adev->power.state.  The ACPI subsystem is expected to update this
> > > > value as needed going forward (including system-wide transitions like
> > > > resume from S3).
> > >
> > > -> the "resume from S3 or hibernation" case needs special handling, because
> > > in that case the device power state need not reflect the information the ACPI
> > > subsystem has.  That only matters if adev->power.state is ACPI_STATE_D0 and
> > > the device is actually *not* in D0, because in that case acpi_device_set_power()
> > > will not work.
> >
> > I guess you are talking about the special-cased devices that we leave in
> > D0 when system suspend (via firmware) is entered?
> >
> > > So that case is not covered currently (it should be rare in practice,
> > > though, if it happens at all), so something like the patch below (untested) may
> > > be needed in addition to the Mika's patch.
> >
> > Looks good to me.
> 
> I actually decided to address this issue in acpi_device_set_power() as
> it may affect devices beyond PCI in principle.  I will send a patch
> for that shortly.

Thanks!

> > > Still, there is also the "power state not matching" case in pci_pm_complete() that's
> > > need to be covered and the non-PCI ACPI PM has a similar issue in theory, so I
> > > need to think about this a bit more.
> >
> > Do you want me to hold off sending an updated version of the patch
> > series while we figure this one out or is it fine if I send it out now
> > and we can add further details on top?
> 
> It is independent of the other fix, so it can be sent now just fine IMO.

OK, I'll send it out in a minute.
diff mbox series

Patch

diff --git a/drivers/pci/pci-acpi.c b/drivers/pci/pci-acpi.c
index 1897847ceb0c..b782acac26c5 100644
--- a/drivers/pci/pci-acpi.c
+++ b/drivers/pci/pci-acpi.c
@@ -685,7 +685,8 @@  static pci_power_t acpi_pci_get_power_state(struct pci_dev *dev)
 	if (!adev || !acpi_device_power_manageable(adev))
 		return PCI_UNKNOWN;
 
-	if (acpi_device_get_power(adev, &state) || state == ACPI_STATE_UNKNOWN)
+	state = adev->power.state;
+	if (state == ACPI_STATE_UNKNOWN)
 		return PCI_UNKNOWN;
 
 	return state_conv[state];