diff mbox series

PCI: Add missing link delays required by the PCIe spec

Message ID 20190821124519.71594-1-mika.westerberg@linux.intel.com (mailing list archive)
State New, archived
Delegated to: Bjorn Helgaas
Headers show
Series PCI: Add missing link delays required by the PCIe spec | expand

Commit Message

Mika Westerberg Aug. 21, 2019, 12:45 p.m. UTC
Currently Linux does not follow PCIe spec regarding the required delays
after reset. A concrete example is a Thunderbolt add-in-card that
consists of a PCIe switch and two PCIe endpoints:

  +-1b.0-[01-6b]----00.0-[02-6b]--+-00.0-[03]----00.0 TBT controller
                                  +-01.0-[04-36]-- DS hotplug port
                                  +-02.0-[37]----00.0 xHCI controller
                                  \-04.0-[38-6b]-- DS hotplug port

The root port (1b.0) and the PCIe switch downstream ports are all PCIe
gen3 so they support 8GT/s link speeds.

We wait for the PCIe hierarchy to enter D3cold (runtime):

  pcieport 0000:00:1b.0: power state changed by ACPI to D3cold

When it wakes up from D3cold, according to the PCIe 4.0 section 5.8 the
PCIe switch is put to reset and its power is re-applied. This means that
we must follow the rules in PCIe 4.0 section 6.6.1.

For the PCIe gen3 ports we are dealing with here, the following applies:

  With a Downstream Port that supports Link speeds greater than 5.0
  GT/s, software must wait a minimum of 100 ms after Link training
  completes before sending a Configuration Request to the device
  immediately below that Port. Software can determine when Link training
  completes by polling the Data Link Layer Link Active bit or by setting
  up an associated interrupt (see Section 6.7.3.3).

Translating this into the above topology we would need to do this (DLLLA
stands for Data Link Layer Link Active):

  pcieport 0000:00:1b.0: wait for 100ms after DLLLA is set before access to 0000:01:00.0
  pcieport 0000:02:00.0: wait for 100ms after DLLLA is set before access to 0000:03:00.0
  pcieport 0000:02:02.0: wait for 100ms after DLLLA is set before access to 0000:37:00.0

I've instrumented the kernel with additional logging so we can see the
actual delays the kernel performs:

  pcieport 0000:00:1b.0: power state changed by ACPI to D0
  pcieport 0000:00:1b.0: waiting for D3cold delay of 100 ms
  pcieport 0000:00:1b.0: waking up bus
  pcieport 0000:00:1b.0: waiting for D3hot delay of 10 ms
  pcieport 0000:00:1b.0: restoring config space at offset 0x2c (was 0x60, writing 0x60)
  ...
  pcieport 0000:00:1b.0: PME# disabled
  pcieport 0000:01:00.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
  ...
  pcieport 0000:01:00.0: PME# disabled
  pcieport 0000:02:00.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
  ...
  pcieport 0000:02:00.0: PME# disabled
  pcieport 0000:02:01.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
  ...
  pcieport 0000:02:01.0: restoring config space at offset 0x4 (was 0x100000, writing 0x100407)
  pcieport 0000:02:01.0: PME# disabled
  pcieport 0000:02:02.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
  ...
  pcieport 0000:02:02.0: PME# disabled
  pcieport 0000:02:04.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
  ...
  pcieport 0000:02:04.0: PME# disabled
  pcieport 0000:02:01.0: PME# enabled
  pcieport 0000:02:01.0: waiting for D3hot delay of 10 ms
  pcieport 0000:02:04.0: PME# enabled
  pcieport 0000:02:04.0: waiting for D3hot delay of 10 ms
  thunderbolt 0000:03:00.0: restoring config space at offset 0x14 (was 0x0, writing 0x8a040000)
  ...
  thunderbolt 0000:03:00.0: PME# disabled
  xhci_hcd 0000:37:00.0: restoring config space at offset 0x10 (was 0x0, writing 0x73f00000)
  ...
  xhci_hcd 0000:37:00.0: PME# disabled

For the switch upstream port (01:00.0) we wait for 100ms but not taking
into account the DLLLA requirement. We then wait 10ms for D3hot -> D0
transition of the root port and the two downstream hotplug ports. This
means that we deviate from what the spec requires.

Performing the same check for system sleep (s2idle) transitions we can
see following when resuming from s2idle:

  pcieport 0000:00:1b.0: power state changed by ACPI to D0
  pcieport 0000:00:1b.0: restoring config space at offset 0x2c (was 0x60, writing 0x60)
  ...
  pcieport 0000:01:00.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
  ...
  pcieport 0000:02:02.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
  pcieport 0000:02:02.0: restoring config space at offset 0x2c (was 0x0, writing 0x0)
  pcieport 0000:02:01.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
  pcieport 0000:02:04.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
  pcieport 0000:02:02.0: restoring config space at offset 0x28 (was 0x0, writing 0x0)
  pcieport 0000:02:00.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
  pcieport 0000:02:02.0: restoring config space at offset 0x24 (was 0x10001, writing 0x1fff1)
  pcieport 0000:02:01.0: restoring config space at offset 0x2c (was 0x0, writing 0x60)
  pcieport 0000:02:02.0: restoring config space at offset 0x20 (was 0x0, writing 0x73f073f0)
  pcieport 0000:02:04.0: restoring config space at offset 0x2c (was 0x0, writing 0x60)
  pcieport 0000:02:01.0: restoring config space at offset 0x28 (was 0x0, writing 0x60)
  pcieport 0000:02:00.0: restoring config space at offset 0x2c (was 0x0, writing 0x0)
  pcieport 0000:02:02.0: restoring config space at offset 0x1c (was 0x101, writing 0x1f1)
  pcieport 0000:02:04.0: restoring config space at offset 0x28 (was 0x0, writing 0x60)
  pcieport 0000:02:01.0: restoring config space at offset 0x24 (was 0x10001, writing 0x1ff10001)
  pcieport 0000:02:00.0: restoring config space at offset 0x28 (was 0x0, writing 0x0)
  pcieport 0000:02:02.0: restoring config space at offset 0x18 (was 0x0, writing 0x373702)
  pcieport 0000:02:04.0: restoring config space at offset 0x24 (was 0x10001, writing 0x49f12001)
  pcieport 0000:02:01.0: restoring config space at offset 0x20 (was 0x0, writing 0x73e05c00)
  pcieport 0000:02:00.0: restoring config space at offset 0x24 (was 0x10001, writing 0x1fff1)
  pcieport 0000:02:04.0: restoring config space at offset 0x20 (was 0x0, writing 0x89f07400)
  pcieport 0000:02:01.0: restoring config space at offset 0x1c (was 0x101, writing 0x5151)
  pcieport 0000:02:00.0: restoring config space at offset 0x20 (was 0x0, writing 0x8a008a00)
  pcieport 0000:02:02.0: restoring config space at offset 0xc (was 0x10000, writing 0x10020)
  pcieport 0000:02:04.0: restoring config space at offset 0x1c (was 0x101, writing 0x6161)
  pcieport 0000:02:01.0: restoring config space at offset 0x18 (was 0x0, writing 0x360402)
  pcieport 0000:02:00.0: restoring config space at offset 0x1c (was 0x101, writing 0x1f1)
  pcieport 0000:02:04.0: restoring config space at offset 0x18 (was 0x0, writing 0x6b3802)
  pcieport 0000:02:02.0: restoring config space at offset 0x4 (was 0x100000, writing 0x100407)
  pcieport 0000:02:00.0: restoring config space at offset 0x18 (was 0x0, writing 0x30302)
  pcieport 0000:02:01.0: restoring config space at offset 0xc (was 0x10000, writing 0x10020)
  pcieport 0000:02:04.0: restoring config space at offset 0xc (was 0x10000, writing 0x10020)
  pcieport 0000:02:00.0: restoring config space at offset 0xc (was 0x10000, writing 0x10020)
  pcieport 0000:02:01.0: restoring config space at offset 0x4 (was 0x100000, writing 0x100407)
  pcieport 0000:02:04.0: restoring config space at offset 0x4 (was 0x100000, writing 0x100407)
  pcieport 0000:02:00.0: restoring config space at offset 0x4 (was 0x100000, writing 0x100407)
  xhci_hcd 0000:37:00.0: restoring config space at offset 0x10 (was 0x0, writing 0x73f00000)
  ...
  thunderbolt 0000:03:00.0: restoring config space at offset 0x14 (was 0x0, writing 0x8a040000)

This is even worse. None of the mandatory delays are performed. If this
would be S3 instead of s2idle then according to PCI FW spec 3.2 section
4.6.8.  there is a specific _DSM that allows the OS to skip the delays
but this platform does not provide the _DSM and does not go to S3 anyway
so no firmware is involved that could already handle these delays.

In this particular Intel Coffee Lake platform these delays are not
actually needed because there is an additional delay as part of the ACPI
power resource that is used to turn on power to the hierarchy but since
that additional delay is not required by any of standards (PCIe, ACPI)
it is not present in the Intel Ice Lake, for example where missing the
mandatory delays causes pciehp to start tearing down the stack too early
(links are not yet trained).

There is also one reported case (see the bugzilla link below) where the
missing delay causes xHCI on a Titan Ridge controller fail to runtime
resume when USB-C dock is plugged.

For this reason, introduce a new function pcie_wait_downstream_accessible()
that is called on PCI core resume and runtime resume paths accordingly
if downstream/root port with device below entered D3cold.

This is second attempt to add the missing delays. The previous solution
in commit c2bf1fc212f7 ("PCI: Add missing link delays required by the
PCIe spec") was reverted because of two issues it caused:

  1. One system become unresponsive after S3 resume due to PME service
     spinning in pcie_pme_work_fn(). The root port in question reports
     that the xHCI sent PME but the xHCI device itself does not have PME
     status set. The PME status bit is never cleared in the root port
     resulting the indefinite loop in pcie_pme_work_fn().

  2. Slows down resume if the root/downstream port does not support
     Data Link Layer Active Reporting because pcie_wait_for_link_delay()
     waits 1100ms in that case.

This version should avoid the above issues because we restrict the delay
to happen only if the port went into D3cold (so it goes through reset)
and only when there is no firmware involved on resume path (so the
kernel is responsible for all the delays).

Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=203885
Reported-by: Kai-Heng Feng <kai.heng.feng@canonical.com>
Tested-by: Kai-Heng Feng <kai.heng.feng@canonical.com>
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
---
Hi all,

As the changelog says this is reworked version that tries to avoid reported
issues while at the same time fix the missing delays so we can get ICL
systems and at least the one system with Titan Ridge controller working
properly.

@Matthias, @Paul and @Nicholas: it would be great if you could try the
patch on top of v5.4-rc5+ and verify that it does not cause any issues on
your systems.

 drivers/pci/pci-driver.c |  19 ++++++
 drivers/pci/pci.c        | 127 ++++++++++++++++++++++++++++++++++++---
 drivers/pci/pci.h        |   1 +
 3 files changed, 137 insertions(+), 10 deletions(-)

Comments

Matthias Andree Aug. 22, 2019, 6:29 p.m. UTC | #1
Am 21.08.19 um 14:45 schrieb Mika Westerberg:
> Hi all,
>
> As the changelog says this is reworked version that tries to avoid reported
> issues while at the same time fix the missing delays so we can get ICL
> systems and at least the one system with Titan Ridge controller working
> properly.
>
> @Matthias, @Paul and @Nicholas: it would be great if you could try the
> patch on top of v5.4-rc5+ and verify that it does not cause any issues on
> your systems.
>
>  drivers/pci/pci-driver.c |  19 ++++++
>  drivers/pci/pci.c        | 127 ++++++++++++++++++++++++++++++++++++---
>  drivers/pci/pci.h        |   1 +
>  3 files changed, 137 insertions(+), 10 deletions(-)
>
> diff --git a/drivers/pci/pci-driver.c b/drivers/pci/pci-driver.c
> index a8124e47bf6e..9aec78ed8907 100644

...

Mika, Bjorn, Rafael,

quick smoke test, this test applied with git-am on top
v5.3-rc5-149-gbb7ba8069de9, reboot, suspend, wake, suspend, wake,
hibernate, wake, looks good to me. The top of git log --pretty-short is
shown below.

Couldn't test on v5.4-rc5 though as I've handed off my time machine for
repairs,
I seem to recall they said something about the flux compensator.

Thanks.

Regards,
Matthias

> commit 0fa4a33a010bdd24edb50b9179109c6639d5f6b5 (HEAD -> master)
> Author: Mika Westerberg <mika.westerberg@linux.intel.com>
>
>     PCI: Add missing link delays required by the PCIe spec
>
> commit bb7ba8069de933d69cb45dd0a5806b61033796a3 (origin/master,
> origin/HEAD)
> Merge: 2babd34df229 e4427372398c
> Author: Linus Torvalds <torvalds@linux-foundation.org>
>
>     Merge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm
>
Rafael J. Wysocki Aug. 22, 2019, 10:23 p.m. UTC | #2
On Wednesday, August 21, 2019 2:45:19 PM CEST Mika Westerberg wrote:
> Currently Linux does not follow PCIe spec regarding the required delays
> after reset. A concrete example is a Thunderbolt add-in-card that
> consists of a PCIe switch and two PCIe endpoints:
> 
>   +-1b.0-[01-6b]----00.0-[02-6b]--+-00.0-[03]----00.0 TBT controller
>                                   +-01.0-[04-36]-- DS hotplug port
>                                   +-02.0-[37]----00.0 xHCI controller
>                                   \-04.0-[38-6b]-- DS hotplug port
> 
> The root port (1b.0) and the PCIe switch downstream ports are all PCIe
> gen3 so they support 8GT/s link speeds.
> 
> We wait for the PCIe hierarchy to enter D3cold (runtime):
> 
>   pcieport 0000:00:1b.0: power state changed by ACPI to D3cold
> 
> When it wakes up from D3cold, according to the PCIe 4.0 section 5.8 the
> PCIe switch is put to reset and its power is re-applied. This means that
> we must follow the rules in PCIe 4.0 section 6.6.1.
> 
> For the PCIe gen3 ports we are dealing with here, the following applies:
> 
>   With a Downstream Port that supports Link speeds greater than 5.0
>   GT/s, software must wait a minimum of 100 ms after Link training
>   completes before sending a Configuration Request to the device
>   immediately below that Port. Software can determine when Link training
>   completes by polling the Data Link Layer Link Active bit or by setting
>   up an associated interrupt (see Section 6.7.3.3).
> 
> Translating this into the above topology we would need to do this (DLLLA
> stands for Data Link Layer Link Active):
> 
>   pcieport 0000:00:1b.0: wait for 100ms after DLLLA is set before access to 0000:01:00.0
>   pcieport 0000:02:00.0: wait for 100ms after DLLLA is set before access to 0000:03:00.0
>   pcieport 0000:02:02.0: wait for 100ms after DLLLA is set before access to 0000:37:00.0
> 
> I've instrumented the kernel with additional logging so we can see the
> actual delays the kernel performs:
> 
>   pcieport 0000:00:1b.0: power state changed by ACPI to D0
>   pcieport 0000:00:1b.0: waiting for D3cold delay of 100 ms
>   pcieport 0000:00:1b.0: waking up bus
>   pcieport 0000:00:1b.0: waiting for D3hot delay of 10 ms
>   pcieport 0000:00:1b.0: restoring config space at offset 0x2c (was 0x60, writing 0x60)
>   ...
>   pcieport 0000:00:1b.0: PME# disabled
>   pcieport 0000:01:00.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
>   ...
>   pcieport 0000:01:00.0: PME# disabled
>   pcieport 0000:02:00.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
>   ...
>   pcieport 0000:02:00.0: PME# disabled
>   pcieport 0000:02:01.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
>   ...
>   pcieport 0000:02:01.0: restoring config space at offset 0x4 (was 0x100000, writing 0x100407)
>   pcieport 0000:02:01.0: PME# disabled
>   pcieport 0000:02:02.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
>   ...
>   pcieport 0000:02:02.0: PME# disabled
>   pcieport 0000:02:04.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
>   ...
>   pcieport 0000:02:04.0: PME# disabled
>   pcieport 0000:02:01.0: PME# enabled
>   pcieport 0000:02:01.0: waiting for D3hot delay of 10 ms
>   pcieport 0000:02:04.0: PME# enabled
>   pcieport 0000:02:04.0: waiting for D3hot delay of 10 ms
>   thunderbolt 0000:03:00.0: restoring config space at offset 0x14 (was 0x0, writing 0x8a040000)
>   ...
>   thunderbolt 0000:03:00.0: PME# disabled
>   xhci_hcd 0000:37:00.0: restoring config space at offset 0x10 (was 0x0, writing 0x73f00000)
>   ...
>   xhci_hcd 0000:37:00.0: PME# disabled
> 
> For the switch upstream port (01:00.0) we wait for 100ms but not taking
> into account the DLLLA requirement. We then wait 10ms for D3hot -> D0
> transition of the root port and the two downstream hotplug ports. This
> means that we deviate from what the spec requires.
> 
> Performing the same check for system sleep (s2idle) transitions we can
> see following when resuming from s2idle:
> 
>   pcieport 0000:00:1b.0: power state changed by ACPI to D0
>   pcieport 0000:00:1b.0: restoring config space at offset 0x2c (was 0x60, writing 0x60)
>   ...
>   pcieport 0000:01:00.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
>   ...
>   pcieport 0000:02:02.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
>   pcieport 0000:02:02.0: restoring config space at offset 0x2c (was 0x0, writing 0x0)
>   pcieport 0000:02:01.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
>   pcieport 0000:02:04.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
>   pcieport 0000:02:02.0: restoring config space at offset 0x28 (was 0x0, writing 0x0)
>   pcieport 0000:02:00.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
>   pcieport 0000:02:02.0: restoring config space at offset 0x24 (was 0x10001, writing 0x1fff1)
>   pcieport 0000:02:01.0: restoring config space at offset 0x2c (was 0x0, writing 0x60)
>   pcieport 0000:02:02.0: restoring config space at offset 0x20 (was 0x0, writing 0x73f073f0)
>   pcieport 0000:02:04.0: restoring config space at offset 0x2c (was 0x0, writing 0x60)
>   pcieport 0000:02:01.0: restoring config space at offset 0x28 (was 0x0, writing 0x60)
>   pcieport 0000:02:00.0: restoring config space at offset 0x2c (was 0x0, writing 0x0)
>   pcieport 0000:02:02.0: restoring config space at offset 0x1c (was 0x101, writing 0x1f1)
>   pcieport 0000:02:04.0: restoring config space at offset 0x28 (was 0x0, writing 0x60)
>   pcieport 0000:02:01.0: restoring config space at offset 0x24 (was 0x10001, writing 0x1ff10001)
>   pcieport 0000:02:00.0: restoring config space at offset 0x28 (was 0x0, writing 0x0)
>   pcieport 0000:02:02.0: restoring config space at offset 0x18 (was 0x0, writing 0x373702)
>   pcieport 0000:02:04.0: restoring config space at offset 0x24 (was 0x10001, writing 0x49f12001)
>   pcieport 0000:02:01.0: restoring config space at offset 0x20 (was 0x0, writing 0x73e05c00)
>   pcieport 0000:02:00.0: restoring config space at offset 0x24 (was 0x10001, writing 0x1fff1)
>   pcieport 0000:02:04.0: restoring config space at offset 0x20 (was 0x0, writing 0x89f07400)
>   pcieport 0000:02:01.0: restoring config space at offset 0x1c (was 0x101, writing 0x5151)
>   pcieport 0000:02:00.0: restoring config space at offset 0x20 (was 0x0, writing 0x8a008a00)
>   pcieport 0000:02:02.0: restoring config space at offset 0xc (was 0x10000, writing 0x10020)
>   pcieport 0000:02:04.0: restoring config space at offset 0x1c (was 0x101, writing 0x6161)
>   pcieport 0000:02:01.0: restoring config space at offset 0x18 (was 0x0, writing 0x360402)
>   pcieport 0000:02:00.0: restoring config space at offset 0x1c (was 0x101, writing 0x1f1)
>   pcieport 0000:02:04.0: restoring config space at offset 0x18 (was 0x0, writing 0x6b3802)
>   pcieport 0000:02:02.0: restoring config space at offset 0x4 (was 0x100000, writing 0x100407)
>   pcieport 0000:02:00.0: restoring config space at offset 0x18 (was 0x0, writing 0x30302)
>   pcieport 0000:02:01.0: restoring config space at offset 0xc (was 0x10000, writing 0x10020)
>   pcieport 0000:02:04.0: restoring config space at offset 0xc (was 0x10000, writing 0x10020)
>   pcieport 0000:02:00.0: restoring config space at offset 0xc (was 0x10000, writing 0x10020)
>   pcieport 0000:02:01.0: restoring config space at offset 0x4 (was 0x100000, writing 0x100407)
>   pcieport 0000:02:04.0: restoring config space at offset 0x4 (was 0x100000, writing 0x100407)
>   pcieport 0000:02:00.0: restoring config space at offset 0x4 (was 0x100000, writing 0x100407)
>   xhci_hcd 0000:37:00.0: restoring config space at offset 0x10 (was 0x0, writing 0x73f00000)
>   ...
>   thunderbolt 0000:03:00.0: restoring config space at offset 0x14 (was 0x0, writing 0x8a040000)
> 
> This is even worse. None of the mandatory delays are performed. If this
> would be S3 instead of s2idle then according to PCI FW spec 3.2 section
> 4.6.8.  there is a specific _DSM that allows the OS to skip the delays
> but this platform does not provide the _DSM and does not go to S3 anyway
> so no firmware is involved that could already handle these delays.
> 
> In this particular Intel Coffee Lake platform these delays are not
> actually needed because there is an additional delay as part of the ACPI
> power resource that is used to turn on power to the hierarchy but since
> that additional delay is not required by any of standards (PCIe, ACPI)
> it is not present in the Intel Ice Lake, for example where missing the
> mandatory delays causes pciehp to start tearing down the stack too early
> (links are not yet trained).
> 
> There is also one reported case (see the bugzilla link below) where the
> missing delay causes xHCI on a Titan Ridge controller fail to runtime
> resume when USB-C dock is plugged.
> 
> For this reason, introduce a new function pcie_wait_downstream_accessible()
> that is called on PCI core resume and runtime resume paths accordingly
> if downstream/root port with device below entered D3cold.
> 
> This is second attempt to add the missing delays. The previous solution
> in commit c2bf1fc212f7 ("PCI: Add missing link delays required by the
> PCIe spec") was reverted because of two issues it caused:
> 
>   1. One system become unresponsive after S3 resume due to PME service
>      spinning in pcie_pme_work_fn(). The root port in question reports
>      that the xHCI sent PME but the xHCI device itself does not have PME
>      status set. The PME status bit is never cleared in the root port
>      resulting the indefinite loop in pcie_pme_work_fn().
> 
>   2. Slows down resume if the root/downstream port does not support
>      Data Link Layer Active Reporting because pcie_wait_for_link_delay()
>      waits 1100ms in that case.
> 
> This version should avoid the above issues because we restrict the delay
> to happen only if the port went into D3cold (so it goes through reset)
> and only when there is no firmware involved on resume path (so the
> kernel is responsible for all the delays).
> 
> Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=203885
> Reported-by: Kai-Heng Feng <kai.heng.feng@canonical.com>
> Tested-by: Kai-Heng Feng <kai.heng.feng@canonical.com>
> Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>

Reviewed-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

> ---
> Hi all,
> 
> As the changelog says this is reworked version that tries to avoid reported
> issues while at the same time fix the missing delays so we can get ICL
> systems and at least the one system with Titan Ridge controller working
> properly.
> 
> @Matthias, @Paul and @Nicholas: it would be great if you could try the
> patch on top of v5.4-rc5+ and verify that it does not cause any issues on
> your systems.
> 
>  drivers/pci/pci-driver.c |  19 ++++++
>  drivers/pci/pci.c        | 127 ++++++++++++++++++++++++++++++++++++---
>  drivers/pci/pci.h        |   1 +
>  3 files changed, 137 insertions(+), 10 deletions(-)
> 
> diff --git a/drivers/pci/pci-driver.c b/drivers/pci/pci-driver.c
> index a8124e47bf6e..9aec78ed8907 100644
> --- a/drivers/pci/pci-driver.c
> +++ b/drivers/pci/pci-driver.c
> @@ -918,6 +918,7 @@ static int pci_pm_resume_noirq(struct device *dev)
>  {
>  	struct pci_dev *pci_dev = to_pci_dev(dev);
>  	struct device_driver *drv = dev->driver;
> +	pci_power_t state = pci_dev->current_state;
>  	int error = 0;
>  
>  	if (dev_pm_may_skip_resume(dev))
> @@ -947,6 +948,15 @@ static int pci_pm_resume_noirq(struct device *dev)
>  
>  	pcie_pme_root_status_cleanup(pci_dev);
>  
> +	/*
> +	 * If resume involves firmware assume it takes care of any delays
> +	 * for now. For suspend-to-idle case we need to do that here before
> +	 * resuming PCIe port services to keep pciehp from tearing down the
> +	 * downstream devices too early.
> +	 */
> +	if (state == PCI_D3cold && pm_suspend_no_platform())
> +		pcie_wait_downstream_accessible(pci_dev);
> +
>  	if (drv && drv->pm && drv->pm->resume_noirq)
>  		error = drv->pm->resume_noirq(dev);
>  
> @@ -1329,6 +1339,7 @@ static int pci_pm_runtime_resume(struct device *dev)
>  	int rc = 0;
>  	struct pci_dev *pci_dev = to_pci_dev(dev);
>  	const struct dev_pm_ops *pm = dev->driver ? dev->driver->pm : NULL;
> +	pci_power_t state = pci_dev->current_state;
>  
>  	/*
>  	 * Restoring config space is necessary even if the device is not bound
> @@ -1344,6 +1355,14 @@ static int pci_pm_runtime_resume(struct device *dev)
>  	pci_enable_wake(pci_dev, PCI_D0, false);
>  	pci_fixup_device(pci_fixup_resume, pci_dev);
>  
> +	/*
> +	 * If the hierarcy went into D3cold wait for the link to be
> +	 * reactivated before resuming PCIe port services to keep pciehp
> +	 * from tearing down the downstream devices too early.
> +	 */
> +	if (state == PCI_D3cold)
> +		pcie_wait_downstream_accessible(pci_dev);
> +
>  	if (pm && pm->runtime_resume)
>  		rc = pm->runtime_resume(dev);
>  
> diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
> index 1b27b5af3d55..9ac50710f1d4 100644
> --- a/drivers/pci/pci.c
> +++ b/drivers/pci/pci.c
> @@ -1025,15 +1025,11 @@ static void __pci_start_power_transition(struct pci_dev *dev, pci_power_t state)
>  	if (state == PCI_D0) {
>  		pci_platform_power_transition(dev, PCI_D0);
>  		/*
> -		 * Mandatory power management transition delays, see
> -		 * PCI Express Base Specification Revision 2.0 Section
> -		 * 6.6.1: Conventional Reset.  Do not delay for
> -		 * devices powered on/off by corresponding bridge,
> -		 * because have already delayed for the bridge.
> +		 * Mandatory power management transition delays are handled
> +		 * in pci_pm_runtime_resume() of the corresponding
> +		 * downstream/root port.
>  		 */
>  		if (dev->runtime_d3cold) {
> -			if (dev->d3cold_delay && !dev->imm_ready)
> -				msleep(dev->d3cold_delay);
>  			/*
>  			 * When powering on a bridge from D3cold, the
>  			 * whole hierarchy may be powered on into
> @@ -4607,14 +4603,17 @@ static int pci_pm_reset(struct pci_dev *dev, int probe)
>  
>  	return pci_dev_wait(dev, "PM D3->D0", PCIE_RESET_READY_POLL_MS);
>  }
> +
>  /**
> - * pcie_wait_for_link - Wait until link is active or inactive
> + * pcie_wait_for_link_delay - Wait until link is active or inactive
>   * @pdev: Bridge device
>   * @active: waiting for active or inactive?
> + * @delay: Delay to wait after link has become active (in ms)
>   *
>   * Use this to wait till link becomes active or inactive.
>   */
> -bool pcie_wait_for_link(struct pci_dev *pdev, bool active)
> +static bool pcie_wait_for_link_delay(struct pci_dev *pdev, bool active,
> +				     int delay)
>  {
>  	int timeout = 1000;
>  	bool ret;
> @@ -4651,13 +4650,121 @@ bool pcie_wait_for_link(struct pci_dev *pdev, bool active)
>  		timeout -= 10;
>  	}
>  	if (active && ret)
> -		msleep(100);
> +		msleep(delay);
>  	else if (ret != active)
>  		pci_info(pdev, "Data Link Layer Link Active not %s in 1000 msec\n",
>  			active ? "set" : "cleared");
>  	return ret == active;
>  }
>  
> +/**
> + * pcie_wait_for_link - Wait until link is active or inactive
> + * @pdev: Bridge device
> + * @active: waiting for active or inactive?
> + *
> + * Use this to wait till link becomes active or inactive.
> + */
> +bool pcie_wait_for_link(struct pci_dev *pdev, bool active)
> +{
> +	return pcie_wait_for_link_delay(pdev, active, 100);
> +}
> +
> +static int pcie_get_downstream_delay(struct pci_bus *bus)
> +{
> +	struct pci_dev *pdev;
> +	int min_delay = 100;
> +	int max_delay = 0;
> +
> +	list_for_each_entry(pdev, &bus->devices, bus_list) {
> +		if (pdev->imm_ready)
> +			min_delay = 0;
> +		else if (pdev->d3cold_delay < min_delay)
> +			min_delay = pdev->d3cold_delay;
> +		if (pdev->d3cold_delay > max_delay)
> +			max_delay = pdev->d3cold_delay;
> +	}
> +
> +	return max(min_delay, max_delay);
> +}
> +
> +/**
> + * pcie_wait_downstream_accessible - Wait downstream device to be accessible
> + * @pdev: PCIe port whose downstream device is waited
> + *
> + * Handle delays according to PCIe 4.0 section 6.6.1 before configuration
> + * access to the downstream device is permitted. If the port does not have
> + * any devices connected, does nothing.
> + *
> + * This is needed if the hierarchy below @pdev went through reset (after
> + * exit from D3cold back to D0uninitialized).
> + */
> +void pcie_wait_downstream_accessible(struct pci_dev *pdev)
> +{
> +	struct pci_dev *child;
> +	struct pci_bus *bus;
> +	int delay;
> +
> +	if (pci_pcie_type(pdev) != PCI_EXP_TYPE_ROOT_PORT &&
> +	    pci_pcie_type(pdev) != PCI_EXP_TYPE_DOWNSTREAM)
> +		return;
> +
> +	if (pci_dev_is_disconnected(pdev))
> +		return;
> +
> +	if (!pdev->bridge_d3)
> +		return;
> +
> +	bus = pdev->subordinate;
> +	if (!bus)
> +		return;
> +
> +	child = list_first_entry_or_null(&bus->devices, struct pci_dev,
> +					 bus_list);
> +	if (!child)
> +		return;
> +
> +	delay = pcie_get_downstream_delay(bus);
> +	if (!delay)
> +		return;
> +
> +	/*
> +	 * If downstream port does not support speeds greater than 5 GT/s
> +	 * need to wait minimum 100ms. For higher speeds (gen3) we need to
> +	 * wait first for the data link layer to become active.
> +	 *
> +	 * However, 100ms is the minimum and the spec says that the
> +	 * software must allow at least 1s before it can determine that the
> +	 * device that did not respond is a broken device. Also there is
> +	 * evidence that the 100ms is not always enough so what we do here
> +	 * is that we wait for the minimum 100ms (possibly after link
> +	 * training completes) and then probe for the device presence once.
> +	 * If we still don't get response we wait for another 100ms just to
> +	 * give it some additional time to complete its initialization.
> +	 */
> +	if (pcie_get_speed_cap(pdev) <= PCIE_SPEED_5_0GT) {
> +		dev_dbg(&pdev->dev, "waiting downstream link for %d ms\n",
> +			delay);
> +		msleep(delay);
> +	} else {
> +		dev_dbg(&pdev->dev,
> +			"waiting downstream link for %d ms after activation\n",
> +			delay);
> +		if (!pcie_wait_for_link_delay(pdev, true, delay)) {
> +			/*
> +			 * If the link did not train, no need to wait
> +			 * further the device is probably gone.
> +			 */
> +			return;
> +		}
> +	}
> +
> +	if (!pci_device_is_present(child)) {
> +		dev_dbg(&child->dev,
> +			"waiting for additional 100 ms for the device to become accessible\n");
> +		msleep(100);
> +	}
> +}
> +
>  void pci_reset_secondary_bus(struct pci_dev *dev)
>  {
>  	u16 ctrl;
> diff --git a/drivers/pci/pci.h b/drivers/pci/pci.h
> index d22d1b807701..9a83fcf612ca 100644
> --- a/drivers/pci/pci.h
> +++ b/drivers/pci/pci.h
> @@ -498,6 +498,7 @@ void pcie_do_recovery(struct pci_dev *dev, enum pci_channel_state state,
>  		      u32 service);
>  
>  bool pcie_wait_for_link(struct pci_dev *pdev, bool active);
> +void pcie_wait_downstream_accessible(struct pci_dev *pdev);
>  #ifdef CONFIG_PCIEASPM
>  void pcie_aspm_init_link_state(struct pci_dev *pdev);
>  void pcie_aspm_exit_link_state(struct pci_dev *pdev);
>
Mika Westerberg Aug. 23, 2019, 7:10 a.m. UTC | #3
On Thu, Aug 22, 2019 at 08:29:09PM +0200, Matthias Andree wrote:
> Am 21.08.19 um 14:45 schrieb Mika Westerberg:
> > Hi all,
> >
> > As the changelog says this is reworked version that tries to avoid reported
> > issues while at the same time fix the missing delays so we can get ICL
> > systems and at least the one system with Titan Ridge controller working
> > properly.
> >
> > @Matthias, @Paul and @Nicholas: it would be great if you could try the
> > patch on top of v5.4-rc5+ and verify that it does not cause any issues on
> > your systems.
> >
> >  drivers/pci/pci-driver.c |  19 ++++++
> >  drivers/pci/pci.c        | 127 ++++++++++++++++++++++++++++++++++++---
> >  drivers/pci/pci.h        |   1 +
> >  3 files changed, 137 insertions(+), 10 deletions(-)
> >
> > diff --git a/drivers/pci/pci-driver.c b/drivers/pci/pci-driver.c
> > index a8124e47bf6e..9aec78ed8907 100644
> 
> ...
> 
> Mika, Bjorn, Rafael,
> 
> quick smoke test, this test applied with git-am on top
> v5.3-rc5-149-gbb7ba8069de9, reboot, suspend, wake, suspend, wake,
> hibernate, wake, looks good to me. The top of git log --pretty-short is
> shown below.

Thanks for testing!

> Couldn't test on v5.4-rc5 though as I've handed off my time machine for
> repairs,
> I seem to recall they said something about the flux compensator.

:)
Bjorn Helgaas Aug. 24, 2019, 2:12 a.m. UTC | #4
Hi Mika,

I'm trying to figure out specifically why we need this and where it
should go.  Questions below.

On Wed, Aug 21, 2019 at 03:45:19PM +0300, Mika Westerberg wrote:
> Currently Linux does not follow PCIe spec regarding the required delays
> after reset. A concrete example is a Thunderbolt add-in-card that
> consists of a PCIe switch and two PCIe endpoints:
> 
>   +-1b.0-[01-6b]----00.0-[02-6b]--+-00.0-[03]----00.0 TBT controller
>                                   +-01.0-[04-36]-- DS hotplug port
>                                   +-02.0-[37]----00.0 xHCI controller
>                                   \-04.0-[38-6b]-- DS hotplug port
> 
> The root port (1b.0) and the PCIe switch downstream ports are all PCIe
> gen3 so they support 8GT/s link speeds.
> 
> We wait for the PCIe hierarchy to enter D3cold (runtime):
> 
>   pcieport 0000:00:1b.0: power state changed by ACPI to D3cold
> 
> When it wakes up from D3cold, according to the PCIe 4.0 section 5.8 the
> PCIe switch is put to reset and its power is re-applied. This means that
> we must follow the rules in PCIe 4.0 section 6.6.1.
> 
> For the PCIe gen3 ports we are dealing with here, the following applies:
> 
>   With a Downstream Port that supports Link speeds greater than 5.0
>   GT/s, software must wait a minimum of 100 ms after Link training
>   completes before sending a Configuration Request to the device
>   immediately below that Port. Software can determine when Link training
>   completes by polling the Data Link Layer Link Active bit or by setting
>   up an associated interrupt (see Section 6.7.3.3).
> 
> Translating this into the above topology we would need to do this (DLLLA
> stands for Data Link Layer Link Active):
> 
>   pcieport 0000:00:1b.0: wait for 100ms after DLLLA is set before access to 0000:01:00.0
>   pcieport 0000:02:00.0: wait for 100ms after DLLLA is set before access to 0000:03:00.0
>   pcieport 0000:02:02.0: wait for 100ms after DLLLA is set before access to 0000:37:00.0
> 
> I've instrumented the kernel with additional logging so we can see the
> actual delays the kernel performs:
> 
>   pcieport 0000:00:1b.0: power state changed by ACPI to D0
>   pcieport 0000:00:1b.0: waiting for D3cold delay of 100 ms
>   pcieport 0000:00:1b.0: waking up bus
>   pcieport 0000:00:1b.0: waiting for D3hot delay of 10 ms
>   pcieport 0000:00:1b.0: restoring config space at offset 0x2c (was 0x60, writing 0x60)
>   ...
>   pcieport 0000:00:1b.0: PME# disabled
>   pcieport 0000:01:00.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
>   ...
>   pcieport 0000:01:00.0: PME# disabled
>   pcieport 0000:02:00.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
>   ...
>   pcieport 0000:02:00.0: PME# disabled
>   pcieport 0000:02:01.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
>   ...
>   pcieport 0000:02:01.0: restoring config space at offset 0x4 (was 0x100000, writing 0x100407)
>   pcieport 0000:02:01.0: PME# disabled
>   pcieport 0000:02:02.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
>   ...
>   pcieport 0000:02:02.0: PME# disabled
>   pcieport 0000:02:04.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
>   ...
>   pcieport 0000:02:04.0: PME# disabled
>   pcieport 0000:02:01.0: PME# enabled
>   pcieport 0000:02:01.0: waiting for D3hot delay of 10 ms
>   pcieport 0000:02:04.0: PME# enabled
>   pcieport 0000:02:04.0: waiting for D3hot delay of 10 ms
>   thunderbolt 0000:03:00.0: restoring config space at offset 0x14 (was 0x0, writing 0x8a040000)
>   ...
>   thunderbolt 0000:03:00.0: PME# disabled
>   xhci_hcd 0000:37:00.0: restoring config space at offset 0x10 (was 0x0, writing 0x73f00000)
>   ...
>   xhci_hcd 0000:37:00.0: PME# disabled
> 
> For the switch upstream port (01:00.0) we wait for 100ms but not taking
> into account the DLLLA requirement. We then wait 10ms for D3hot -> D0
> transition of the root port and the two downstream hotplug ports. This
> means that we deviate from what the spec requires.
> 
> Performing the same check for system sleep (s2idle) transitions we can
> see following when resuming from s2idle:
> 
>   pcieport 0000:00:1b.0: power state changed by ACPI to D0
>   pcieport 0000:00:1b.0: restoring config space at offset 0x2c (was 0x60, writing 0x60)
>   ...
>   pcieport 0000:01:00.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
>   ...

I think the important thing in all the above logging is that it
doesn't show any delay, right?  If that's the case, you can just say
that in one line; I trust you even without 40 lines of config space
restore debug output :)

>   xhci_hcd 0000:37:00.0: restoring config space at offset 0x10 (was 0x0, writing 0x73f00000)
>   ...
>   thunderbolt 0000:03:00.0: restoring config space at offset 0x14 (was 0x0, writing 0x8a040000)
> 
> This is even worse. None of the mandatory delays are performed. If this
> would be S3 instead of s2idle then according to PCI FW spec 3.2 section
> 4.6.8.  there is a specific _DSM that allows the OS to skip the delays
> but this platform does not provide the _DSM and does not go to S3 anyway
> so no firmware is involved that could already handle these delays.
> 
> In this particular Intel Coffee Lake platform these delays are not
> actually needed because there is an additional delay as part of the ACPI
> power resource that is used to turn on power to the hierarchy but since
> that additional delay is not required by any of standards (PCIe, ACPI)

So it sounds like this Coffee Lake accidentally works because of
unrelated firmware delay that's not actually required, or at least not
related to the delay required by PCIe?

I did notice that we don't implement all of _DSM function 9 and the
parts we're missing look like they could be relevant.

> it is not present in the Intel Ice Lake, for example where missing the
> mandatory delays causes pciehp to start tearing down the stack too early
> (links are not yet trained).

I'm guessing the Coffee Lake/Ice Lake distinction is not really
relevant and the important thing is that something about Ice Lake is
faster and reduces the accidental delay to the point that things stop
working.

So probably the same thing could happen on Coffee Lake or any other
system if it had different firmware.

It *would* be interesting to mention what the pciehp/link training
issue looks like to a user, just to help people who see it find this
fix.

> There is also one reported case (see the bugzilla link below) where the
> missing delay causes xHCI on a Titan Ridge controller fail to runtime
> resume when USB-C dock is plugged.
> 
> For this reason, introduce a new function pcie_wait_downstream_accessible()
> that is called on PCI core resume and runtime resume paths accordingly
> if downstream/root port with device below entered D3cold.
> 
> This is second attempt to add the missing delays. The previous solution
> in commit c2bf1fc212f7 ("PCI: Add missing link delays required by the
> PCIe spec") was reverted because of two issues it caused:

c2bf1fc212f7 was merged for v5.3 (it appeared in v5.3-rc1).  I *guess*
it addressed https://bugzilla.kernel.org/show_bug.cgi?id=203885, which
was reported on v5.2-rc4, though c2bf1fc212f7 doesn't actually mention
the bugzilla?

0617bdede511 ("Revert "PCI: Add missing link delays required by the
PCIe spec"") appeared in v5.3-rc4.  If c2bf1fc212f7 was supposed to
address BZ 203885, I'm confused about why you asked Kai-Heng to test
v5.3-rc4, where c2bf1fc212f7 had already been reverted.

Or maybe c2bf1fc212f7 wasn't connected with BZ 203885 in the first
place?

The net result is that I think v5.3-rc4 is equivalent to v5.2 with
respect to this issue.  It *is* a pretty serious usability issue, no
question, but it's not completely obvious that this fix needs to be in
v5.3 since it's not something we broke during the v5.3 merge window.
So if we *do* want it in v5.3, we need to think about how to justify
that, e.g., if this issue affects shipping systems that are likely to
run an upstream kernel, etc.

>   1. One system become unresponsive after S3 resume due to PME service
>      spinning in pcie_pme_work_fn(). The root port in question reports
>      that the xHCI sent PME but the xHCI device itself does not have PME
>      status set. The PME status bit is never cleared in the root port
>      resulting the indefinite loop in pcie_pme_work_fn().

I don't see the connection between PME and either c2bf1fc212f7 or this
patch.  Is there a fix for this pcie_pme_work_fn() infinite loop?

I do see that BZ 203885 mentions a flood of "PME: Spurious native
interrupt!" messages, but I don't see how we've fixed that.

>   2. Slows down resume if the root/downstream port does not support
>      Data Link Layer Active Reporting because pcie_wait_for_link_delay()
>      waits 1100ms in that case.

I don't see the slowdown mentioned in BZ 203885; is there another link
to these reports?

> This version should avoid the above issues because we restrict the delay
> to happen only if the port went into D3cold (so it goes through reset)
> and only when there is no firmware involved on resume path (so the
> kernel is responsible for all the delays).
> 
> Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=203885
> Reported-by: Kai-Heng Feng <kai.heng.feng@canonical.com>
> Tested-by: Kai-Heng Feng <kai.heng.feng@canonical.com>
> Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
> ---
> Hi all,
> 
> As the changelog says this is reworked version that tries to avoid reported
> issues while at the same time fix the missing delays so we can get ICL
> systems and at least the one system with Titan Ridge controller working
> properly.
> 
> @Matthias, @Paul and @Nicholas: it would be great if you could try the
> patch on top of v5.4-rc5+ and verify that it does not cause any issues on
> your systems.
> 
>  drivers/pci/pci-driver.c |  19 ++++++
>  drivers/pci/pci.c        | 127 ++++++++++++++++++++++++++++++++++++---
>  drivers/pci/pci.h        |   1 +
>  3 files changed, 137 insertions(+), 10 deletions(-)
> 
> diff --git a/drivers/pci/pci-driver.c b/drivers/pci/pci-driver.c
> index a8124e47bf6e..9aec78ed8907 100644
> --- a/drivers/pci/pci-driver.c
> +++ b/drivers/pci/pci-driver.c
> @@ -918,6 +918,7 @@ static int pci_pm_resume_noirq(struct device *dev)
>  {
>  	struct pci_dev *pci_dev = to_pci_dev(dev);
>  	struct device_driver *drv = dev->driver;
> +	pci_power_t state = pci_dev->current_state;
>  	int error = 0;
>  
>  	if (dev_pm_may_skip_resume(dev))
> @@ -947,6 +948,15 @@ static int pci_pm_resume_noirq(struct device *dev)
>  
>  	pcie_pme_root_status_cleanup(pci_dev);
>  
> +	/*
> +	 * If resume involves firmware assume it takes care of any delays
> +	 * for now. For suspend-to-idle case we need to do that here before
> +	 * resuming PCIe port services to keep pciehp from tearing down the
> +	 * downstream devices too early.
> +	 */
> +	if (state == PCI_D3cold && pm_suspend_no_platform())
> +		pcie_wait_downstream_accessible(pci_dev);

Aren't these paths used for Conventional PCI devices as well as PCIe
devices?

I think the D3cold and pm_suspend_no_platform() checks should move
inside the new interface, whatever it's called.  I'm not sure what
that means for the fact that you don't check pm_suspend_no_platform()
in the runtime-resume path; maybe it needs a flag or something.

But the "wait downstream" part seems a little too specific to be at
the .resume_noirq and .runtime_resume level.

Do we descend the hierarchy and call .resume_noirq and .runtime_resume
for the children of the bridge, too?

>  	if (drv && drv->pm && drv->pm->resume_noirq)
>  		error = drv->pm->resume_noirq(dev);
>  
> @@ -1329,6 +1339,7 @@ static int pci_pm_runtime_resume(struct device *dev)
>  	int rc = 0;
>  	struct pci_dev *pci_dev = to_pci_dev(dev);
>  	const struct dev_pm_ops *pm = dev->driver ? dev->driver->pm : NULL;
> +	pci_power_t state = pci_dev->current_state;
>  
>  	/*
>  	 * Restoring config space is necessary even if the device is not bound
> @@ -1344,6 +1355,14 @@ static int pci_pm_runtime_resume(struct device *dev)
>  	pci_enable_wake(pci_dev, PCI_D0, false);
>  	pci_fixup_device(pci_fixup_resume, pci_dev);
>  
> +	/*
> +	 * If the hierarcy went into D3cold wait for the link to be
> +	 * reactivated before resuming PCIe port services to keep pciehp
> +	 * from tearing down the downstream devices too early.

s/hierarcy/hierarchy/

> +	 */
> +	if (state == PCI_D3cold)
> +		pcie_wait_downstream_accessible(pci_dev);
> +
>  	if (pm && pm->runtime_resume)
>  		rc = pm->runtime_resume(dev);
>  
> diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
> index 1b27b5af3d55..9ac50710f1d4 100644
> --- a/drivers/pci/pci.c
> +++ b/drivers/pci/pci.c
> @@ -1025,15 +1025,11 @@ static void __pci_start_power_transition(struct pci_dev *dev, pci_power_t state)
>  	if (state == PCI_D0) {
>  		pci_platform_power_transition(dev, PCI_D0);
>  		/*
> -		 * Mandatory power management transition delays, see
> -		 * PCI Express Base Specification Revision 2.0 Section
> -		 * 6.6.1: Conventional Reset.  Do not delay for
> -		 * devices powered on/off by corresponding bridge,
> -		 * because have already delayed for the bridge.
> +		 * Mandatory power management transition delays are handled
> +		 * in pci_pm_runtime_resume() of the corresponding
> +		 * downstream/root port.
>  		 */
>  		if (dev->runtime_d3cold) {
> -			if (dev->d3cold_delay && !dev->imm_ready)
> -				msleep(dev->d3cold_delay);
>  			/*
>  			 * When powering on a bridge from D3cold, the
>  			 * whole hierarchy may be powered on into
> @@ -4607,14 +4603,17 @@ static int pci_pm_reset(struct pci_dev *dev, int probe)
>  
>  	return pci_dev_wait(dev, "PM D3->D0", PCIE_RESET_READY_POLL_MS);
>  }
> +
>  /**
> - * pcie_wait_for_link - Wait until link is active or inactive
> + * pcie_wait_for_link_delay - Wait until link is active or inactive
>   * @pdev: Bridge device
>   * @active: waiting for active or inactive?
> + * @delay: Delay to wait after link has become active (in ms)
>   *
>   * Use this to wait till link becomes active or inactive.
>   */
> -bool pcie_wait_for_link(struct pci_dev *pdev, bool active)
> +static bool pcie_wait_for_link_delay(struct pci_dev *pdev, bool active,
> +				     int delay)
>  {
>  	int timeout = 1000;
>  	bool ret;
> @@ -4651,13 +4650,121 @@ bool pcie_wait_for_link(struct pci_dev *pdev, bool active)
>  		timeout -= 10;
>  	}
>  	if (active && ret)
> -		msleep(100);
> +		msleep(delay);
>  	else if (ret != active)
>  		pci_info(pdev, "Data Link Layer Link Active not %s in 1000 msec\n",
>  			active ? "set" : "cleared");
>  	return ret == active;
>  }
>  
> +/**
> + * pcie_wait_for_link - Wait until link is active or inactive
> + * @pdev: Bridge device
> + * @active: waiting for active or inactive?
> + *
> + * Use this to wait till link becomes active or inactive.
> + */
> +bool pcie_wait_for_link(struct pci_dev *pdev, bool active)
> +{
> +	return pcie_wait_for_link_delay(pdev, active, 100);
> +}

This part (adding pcie_wait_for_link_delay() to make the delay time
configurable) could be a separate patch to reduce the size of this
patch and make the important pieces a little more obvious.

> +static int pcie_get_downstream_delay(struct pci_bus *bus)
> +{
> +	struct pci_dev *pdev;
> +	int min_delay = 100;
> +	int max_delay = 0;
> +
> +	list_for_each_entry(pdev, &bus->devices, bus_list) {
> +		if (pdev->imm_ready)
> +			min_delay = 0;
> +		else if (pdev->d3cold_delay < min_delay)
> +			min_delay = pdev->d3cold_delay;
> +		if (pdev->d3cold_delay > max_delay)
> +			max_delay = pdev->d3cold_delay;
> +	}
> +
> +	return max(min_delay, max_delay);
> +}
> +
> +/**
> + * pcie_wait_downstream_accessible - Wait downstream device to be accessible
> + * @pdev: PCIe port whose downstream device is waited
> + *
> + * Handle delays according to PCIe 4.0 section 6.6.1 before configuration
> + * access to the downstream device is permitted. If the port does not have
> + * any devices connected, does nothing.
> + *
> + * This is needed if the hierarchy below @pdev went through reset (after
> + * exit from D3cold back to D0uninitialized).

D3cold -> D0 is a cold reset by definition, isn't it?

> + */
> +void pcie_wait_downstream_accessible(struct pci_dev *pdev)
> +{
> +	struct pci_dev *child;
> +	struct pci_bus *bus;
> +	int delay;

IIUC pdev is not guaranteed to be a PCIe device here.

Do we need to observe the Trhfa requirements for Conventional PCI and
PCI-X devices here?  If we don't do it here, where is it observed?

> +	if (pci_pcie_type(pdev) != PCI_EXP_TYPE_ROOT_PORT &&
> +	    pci_pcie_type(pdev) != PCI_EXP_TYPE_DOWNSTREAM)
> +		return;
> +
> +	if (pci_dev_is_disconnected(pdev))
> +		return;
> +
> +	if (!pdev->bridge_d3)
> +		return;
> +
> +	bus = pdev->subordinate;
> +	if (!bus)
> +		return;
> +
> +	child = list_first_entry_or_null(&bus->devices, struct pci_dev,
> +					 bus_list);
> +	if (!child)
> +		return;

I'm not convinced yet about skipping this if there's no subordinate
bus or no children.  Don't we have to assume that the caller may
immediately *probe* for children as soon as we return?

> +	delay = pcie_get_downstream_delay(bus);
> +	if (!delay)
> +		return;

I'm not sold on the idea that this delay depends on what's *below* the
bridge.  We're using sec 6.6.1 to justify the delay, and that section
doesn't say anything about downstream devices.

If we call .resume_noirq/.runtime_resume for the downstream devices
themselves, could we use d3cold_delay *there*?

> +	/*
> +	 * If downstream port does not support speeds greater than 5 GT/s
> +	 * need to wait minimum 100ms. For higher speeds (gen3) we need to
> +	 * wait first for the data link layer to become active.
> +	 *
> +	 * However, 100ms is the minimum and the spec says that the
> +	 * software must allow at least 1s before it can determine that the
> +	 * device that did not respond is a broken device. Also there is
> +	 * evidence that the 100ms is not always enough so what we do here

This sort of "there is evidence" statement needs a specific reference.
Otherwise it ends up meaning "we don't know why it matters and we
can't test any changes in this area".

> +	 * is that we wait for the minimum 100ms (possibly after link
> +	 * training completes) and then probe for the device presence once.
> +	 * If we still don't get response we wait for another 100ms just to
> +	 * give it some additional time to complete its initialization.
> +	 */
> +	if (pcie_get_speed_cap(pdev) <= PCIE_SPEED_5_0GT) {
> +		dev_dbg(&pdev->dev, "waiting downstream link for %d ms\n",

pci_dbg()
s/waiting downstream link for %d ms/waiting %d ms for downstream link/

> +			delay);
> +		msleep(delay);
> +	} else {
> +		dev_dbg(&pdev->dev,
> +			"waiting downstream link for %d ms after activation\n",
> +			delay);
> +		if (!pcie_wait_for_link_delay(pdev, true, delay)) {
> +			/*
> +			 * If the link did not train, no need to wait
> +			 * further the device is probably gone.
> +			 */
> +			return;
> +		}
> +	}
> +
> +	if (!pci_device_is_present(child)) {
> +		dev_dbg(&child->dev,
> +			"waiting for additional 100 ms for the device to become accessible\n");
> +		msleep(100);

This seems like magic.  Can we relate this back to anything in the
spec?  If we need this, maybe we're seeing CRRS?  I suppose this is
related to the note above about evidence that 100ms may not be enough?

> +	}
> +}
> +
>  void pci_reset_secondary_bus(struct pci_dev *dev)
>  {
>  	u16 ctrl;
> diff --git a/drivers/pci/pci.h b/drivers/pci/pci.h
> index d22d1b807701..9a83fcf612ca 100644
> --- a/drivers/pci/pci.h
> +++ b/drivers/pci/pci.h
> @@ -498,6 +498,7 @@ void pcie_do_recovery(struct pci_dev *dev, enum pci_channel_state state,
>  		      u32 service);
>  
>  bool pcie_wait_for_link(struct pci_dev *pdev, bool active);
> +void pcie_wait_downstream_accessible(struct pci_dev *pdev);
>  #ifdef CONFIG_PCIEASPM
>  void pcie_aspm_init_link_state(struct pci_dev *pdev);
>  void pcie_aspm_exit_link_state(struct pci_dev *pdev);
> -- 
> 2.23.0.rc1
>
Mika Westerberg Aug. 26, 2019, 10:17 a.m. UTC | #5
On Fri, Aug 23, 2019 at 09:12:54PM -0500, Bjorn Helgaas wrote:
> Hi Mika,

Hi,

> I'm trying to figure out specifically why we need this and where it
> should go.  Questions below.

Thanks for looking at this.

> On Wed, Aug 21, 2019 at 03:45:19PM +0300, Mika Westerberg wrote:
> > Currently Linux does not follow PCIe spec regarding the required delays
> > after reset. A concrete example is a Thunderbolt add-in-card that
> > consists of a PCIe switch and two PCIe endpoints:
> > 
> >   +-1b.0-[01-6b]----00.0-[02-6b]--+-00.0-[03]----00.0 TBT controller
> >                                   +-01.0-[04-36]-- DS hotplug port
> >                                   +-02.0-[37]----00.0 xHCI controller
> >                                   \-04.0-[38-6b]-- DS hotplug port
> > 
> > The root port (1b.0) and the PCIe switch downstream ports are all PCIe
> > gen3 so they support 8GT/s link speeds.
> > 
> > We wait for the PCIe hierarchy to enter D3cold (runtime):
> > 
> >   pcieport 0000:00:1b.0: power state changed by ACPI to D3cold
> > 
> > When it wakes up from D3cold, according to the PCIe 4.0 section 5.8 the
> > PCIe switch is put to reset and its power is re-applied. This means that
> > we must follow the rules in PCIe 4.0 section 6.6.1.
> > 
> > For the PCIe gen3 ports we are dealing with here, the following applies:
> > 
> >   With a Downstream Port that supports Link speeds greater than 5.0
> >   GT/s, software must wait a minimum of 100 ms after Link training
> >   completes before sending a Configuration Request to the device
> >   immediately below that Port. Software can determine when Link training
> >   completes by polling the Data Link Layer Link Active bit or by setting
> >   up an associated interrupt (see Section 6.7.3.3).
> > 
> > Translating this into the above topology we would need to do this (DLLLA
> > stands for Data Link Layer Link Active):
> > 
> >   pcieport 0000:00:1b.0: wait for 100ms after DLLLA is set before access to 0000:01:00.0
> >   pcieport 0000:02:00.0: wait for 100ms after DLLLA is set before access to 0000:03:00.0
> >   pcieport 0000:02:02.0: wait for 100ms after DLLLA is set before access to 0000:37:00.0
> > 
> > I've instrumented the kernel with additional logging so we can see the
> > actual delays the kernel performs:
> > 
> >   pcieport 0000:00:1b.0: power state changed by ACPI to D0
> >   pcieport 0000:00:1b.0: waiting for D3cold delay of 100 ms
> >   pcieport 0000:00:1b.0: waking up bus
> >   pcieport 0000:00:1b.0: waiting for D3hot delay of 10 ms
> >   pcieport 0000:00:1b.0: restoring config space at offset 0x2c (was 0x60, writing 0x60)
> >   ...
> >   pcieport 0000:00:1b.0: PME# disabled
> >   pcieport 0000:01:00.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
> >   ...
> >   pcieport 0000:01:00.0: PME# disabled
> >   pcieport 0000:02:00.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
> >   ...
> >   pcieport 0000:02:00.0: PME# disabled
> >   pcieport 0000:02:01.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
> >   ...
> >   pcieport 0000:02:01.0: restoring config space at offset 0x4 (was 0x100000, writing 0x100407)
> >   pcieport 0000:02:01.0: PME# disabled
> >   pcieport 0000:02:02.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
> >   ...
> >   pcieport 0000:02:02.0: PME# disabled
> >   pcieport 0000:02:04.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
> >   ...
> >   pcieport 0000:02:04.0: PME# disabled
> >   pcieport 0000:02:01.0: PME# enabled
> >   pcieport 0000:02:01.0: waiting for D3hot delay of 10 ms
> >   pcieport 0000:02:04.0: PME# enabled
> >   pcieport 0000:02:04.0: waiting for D3hot delay of 10 ms
> >   thunderbolt 0000:03:00.0: restoring config space at offset 0x14 (was 0x0, writing 0x8a040000)
> >   ...
> >   thunderbolt 0000:03:00.0: PME# disabled
> >   xhci_hcd 0000:37:00.0: restoring config space at offset 0x10 (was 0x0, writing 0x73f00000)
> >   ...
> >   xhci_hcd 0000:37:00.0: PME# disabled
> > 
> > For the switch upstream port (01:00.0) we wait for 100ms but not taking
> > into account the DLLLA requirement. We then wait 10ms for D3hot -> D0
> > transition of the root port and the two downstream hotplug ports. This
> > means that we deviate from what the spec requires.
> > 
> > Performing the same check for system sleep (s2idle) transitions we can
> > see following when resuming from s2idle:
> > 
> >   pcieport 0000:00:1b.0: power state changed by ACPI to D0
> >   pcieport 0000:00:1b.0: restoring config space at offset 0x2c (was 0x60, writing 0x60)
> >   ...
> >   pcieport 0000:01:00.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
> >   ...
> 
> I think the important thing in all the above logging is that it
> doesn't show any delay, right?

Right.

> If that's the case, you can just say
> that in one line; I trust you even without 40 lines of config space
> restore debug output :)

OK :)

> >   xhci_hcd 0000:37:00.0: restoring config space at offset 0x10 (was 0x0, writing 0x73f00000)
> >   ...
> >   thunderbolt 0000:03:00.0: restoring config space at offset 0x14 (was 0x0, writing 0x8a040000)
> > 
> > This is even worse. None of the mandatory delays are performed. If this
> > would be S3 instead of s2idle then according to PCI FW spec 3.2 section
> > 4.6.8.  there is a specific _DSM that allows the OS to skip the delays
> > but this platform does not provide the _DSM and does not go to S3 anyway
> > so no firmware is involved that could already handle these delays.
> > 
> > In this particular Intel Coffee Lake platform these delays are not
> > actually needed because there is an additional delay as part of the ACPI
> > power resource that is used to turn on power to the hierarchy but since
> > that additional delay is not required by any of standards (PCIe, ACPI)
> 
> So it sounds like this Coffee Lake accidentally works because of
> unrelated firmware delay that's not actually required, or at least not
> related to the delay required by PCIe?

Correct. The root port ACPI Power Resource includes quite long delay
which allows the links to be trained before the _ON method is finished.

> I did notice that we don't implement all of _DSM function 9 and the
> parts we're missing look like they could be relevant.

AFAICT that _DSM is only for lowering the delays, not increasing them.
If the platform does not provide it the OS must adhere to all the timing
requirements of the PCIe spec (PCI FW 3.2 section 4.6.9). Here we are
actually trying to do just that :)

> > it is not present in the Intel Ice Lake, for example where missing the
> > mandatory delays causes pciehp to start tearing down the stack too early
> > (links are not yet trained).
> 
> I'm guessing the Coffee Lake/Ice Lake distinction is not really
> relevant and the important thing is that something about Ice Lake is
> faster and reduces the accidental delay to the point that things stop
> working.
> 
> So probably the same thing could happen on Coffee Lake or any other
> system if it had different firmware.

Indeed.

> It *would* be interesting to mention what the pciehp/link training
> issue looks like to a user, just to help people who see it find this
> fix.

OK.

> > There is also one reported case (see the bugzilla link below) where the
> > missing delay causes xHCI on a Titan Ridge controller fail to runtime
> > resume when USB-C dock is plugged.
> > 
> > For this reason, introduce a new function pcie_wait_downstream_accessible()
> > that is called on PCI core resume and runtime resume paths accordingly
> > if downstream/root port with device below entered D3cold.
> > 
> > This is second attempt to add the missing delays. The previous solution
> > in commit c2bf1fc212f7 ("PCI: Add missing link delays required by the
> > PCIe spec") was reverted because of two issues it caused:
> 
> c2bf1fc212f7 was merged for v5.3 (it appeared in v5.3-rc1).  I *guess*
> it addressed https://bugzilla.kernel.org/show_bug.cgi?id=203885, which
> was reported on v5.2-rc4, though c2bf1fc212f7 doesn't actually mention
> the bugzilla?

I think c2bf1fc212f7 was merged before I was aware of that bug (or I
just missed it).

> 0617bdede511 ("Revert "PCI: Add missing link delays required by the
> PCIe spec"") appeared in v5.3-rc4.  If c2bf1fc212f7 was supposed to
> address BZ 203885, I'm confused about why you asked Kai-Heng to test
> v5.3-rc4, where c2bf1fc212f7 had already been reverted.

I was hoping that the revert in v5.3-rc4 would help here as well but
after futher investigation it turned to be the exact opposite.

> Or maybe c2bf1fc212f7 wasn't connected with BZ 203885 in the first
> place?
> 
> The net result is that I think v5.3-rc4 is equivalent to v5.2 with
> respect to this issue.  It *is* a pretty serious usability issue, no
> question, but it's not completely obvious that this fix needs to be in
> v5.3 since it's not something we broke during the v5.3 merge window.
> So if we *do* want it in v5.3, we need to think about how to justify
> that, e.g., if this issue affects shipping systems that are likely to
> run an upstream kernel, etc.

I don't think it needs to be in v5.3. I was thinking more like v5.4 so
we get some more testing coverage.

> >   1. One system become unresponsive after S3 resume due to PME service
> >      spinning in pcie_pme_work_fn(). The root port in question reports
> >      that the xHCI sent PME but the xHCI device itself does not have PME
> >      status set. The PME status bit is never cleared in the root port
> >      resulting the indefinite loop in pcie_pme_work_fn().
> 
> I don't see the connection between PME and either c2bf1fc212f7 or this
> patch.  Is there a fix for this pcie_pme_work_fn() infinite loop?

No.

> I do see that BZ 203885 mentions a flood of "PME: Spurious native
> interrupt!" messages, but I don't see how we've fixed that.

It was not completely root caused yet. To me it looks like the root port
keeps the PME status and pending bits set and that causes
pcie_pme_work_fn() spin forever. From the debugging output logs I got
from Matthias, it is the xHCI that seems to send the PME but it does not
have PME status set for some reason.

> >   2. Slows down resume if the root/downstream port does not support
> >      Data Link Layer Active Reporting because pcie_wait_for_link_delay()
> >      waits 1100ms in that case.
> 
> I don't see the slowdown mentioned in BZ 203885; is there another link
> to these reports?

Nicholas for example reported it here:

  https://lore.kernel.org/linux-pci/SL2P216MB01878BBCD75F21D882AEEA2880C60@SL2P216MB0187.KORP216.PROD.OUTLOOK.COM/

I should have added that link to the changelog.

> > This version should avoid the above issues because we restrict the delay
> > to happen only if the port went into D3cold (so it goes through reset)
> > and only when there is no firmware involved on resume path (so the
> > kernel is responsible for all the delays).
> > 
> > Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=203885
> > Reported-by: Kai-Heng Feng <kai.heng.feng@canonical.com>
> > Tested-by: Kai-Heng Feng <kai.heng.feng@canonical.com>
> > Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
> > ---
> > Hi all,
> > 
> > As the changelog says this is reworked version that tries to avoid reported
> > issues while at the same time fix the missing delays so we can get ICL
> > systems and at least the one system with Titan Ridge controller working
> > properly.
> > 
> > @Matthias, @Paul and @Nicholas: it would be great if you could try the
> > patch on top of v5.4-rc5+ and verify that it does not cause any issues on
> > your systems.
> > 
> >  drivers/pci/pci-driver.c |  19 ++++++
> >  drivers/pci/pci.c        | 127 ++++++++++++++++++++++++++++++++++++---
> >  drivers/pci/pci.h        |   1 +
> >  3 files changed, 137 insertions(+), 10 deletions(-)
> > 
> > diff --git a/drivers/pci/pci-driver.c b/drivers/pci/pci-driver.c
> > index a8124e47bf6e..9aec78ed8907 100644
> > --- a/drivers/pci/pci-driver.c
> > +++ b/drivers/pci/pci-driver.c
> > @@ -918,6 +918,7 @@ static int pci_pm_resume_noirq(struct device *dev)
> >  {
> >  	struct pci_dev *pci_dev = to_pci_dev(dev);
> >  	struct device_driver *drv = dev->driver;
> > +	pci_power_t state = pci_dev->current_state;
> >  	int error = 0;
> >  
> >  	if (dev_pm_may_skip_resume(dev))
> > @@ -947,6 +948,15 @@ static int pci_pm_resume_noirq(struct device *dev)
> >  
> >  	pcie_pme_root_status_cleanup(pci_dev);
> >  
> > +	/*
> > +	 * If resume involves firmware assume it takes care of any delays
> > +	 * for now. For suspend-to-idle case we need to do that here before
> > +	 * resuming PCIe port services to keep pciehp from tearing down the
> > +	 * downstream devices too early.
> > +	 */
> > +	if (state == PCI_D3cold && pm_suspend_no_platform())
> > +		pcie_wait_downstream_accessible(pci_dev);
> 
> Aren't these paths used for Conventional PCI devices as well as PCIe
> devices?

Indeed they are.

> I think the D3cold and pm_suspend_no_platform() checks should move
> inside the new interface, whatever it's called.  I'm not sure what
> that means for the fact that you don't check pm_suspend_no_platform()
> in the runtime-resume path; maybe it needs a flag or something.

It is not needed for runtime-resume path since we are not entering
system suspend. For runtime-resume path we need to perform the delays
always.

Here we just assume the firmware actually handles the delays which is
not entirely accurate. According to PCI FW 3.2 section 4.6.8 there is a
_DSM that the firmware can use to indicate to the OS that it can skip
the reset delays. So I think more correct would be to check for that in
resume path. However, since it looks like that _DSM is not implemented
in too many systems I've seen, it may slow down resume times
unnecessarily if the firmware actually handles the delays.

> But the "wait downstream" part seems a little too specific to be at
> the .resume_noirq and .runtime_resume level.
> 
> Do we descend the hierarchy and call .resume_noirq and .runtime_resume
> for the children of the bridge, too?

We do but at that time it is too late as we have already resumed pciehp
of the parent downstream port and it may have already started tearing
down the device stack below.

I'm open to any better ideas where this delay could be added, though :)

> 
> >  	if (drv && drv->pm && drv->pm->resume_noirq)
> >  		error = drv->pm->resume_noirq(dev);
> >  
> > @@ -1329,6 +1339,7 @@ static int pci_pm_runtime_resume(struct device *dev)
> >  	int rc = 0;
> >  	struct pci_dev *pci_dev = to_pci_dev(dev);
> >  	const struct dev_pm_ops *pm = dev->driver ? dev->driver->pm : NULL;
> > +	pci_power_t state = pci_dev->current_state;
> >  
> >  	/*
> >  	 * Restoring config space is necessary even if the device is not bound
> > @@ -1344,6 +1355,14 @@ static int pci_pm_runtime_resume(struct device *dev)
> >  	pci_enable_wake(pci_dev, PCI_D0, false);
> >  	pci_fixup_device(pci_fixup_resume, pci_dev);
> >  
> > +	/*
> > +	 * If the hierarcy went into D3cold wait for the link to be
> > +	 * reactivated before resuming PCIe port services to keep pciehp
> > +	 * from tearing down the downstream devices too early.
> 
> s/hierarcy/hierarchy/

OK.

> > +	 */
> > +	if (state == PCI_D3cold)
> > +		pcie_wait_downstream_accessible(pci_dev);
> > +
> >  	if (pm && pm->runtime_resume)
> >  		rc = pm->runtime_resume(dev);
> >  
> > diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
> > index 1b27b5af3d55..9ac50710f1d4 100644
> > --- a/drivers/pci/pci.c
> > +++ b/drivers/pci/pci.c
> > @@ -1025,15 +1025,11 @@ static void __pci_start_power_transition(struct pci_dev *dev, pci_power_t state)
> >  	if (state == PCI_D0) {
> >  		pci_platform_power_transition(dev, PCI_D0);
> >  		/*
> > -		 * Mandatory power management transition delays, see
> > -		 * PCI Express Base Specification Revision 2.0 Section
> > -		 * 6.6.1: Conventional Reset.  Do not delay for
> > -		 * devices powered on/off by corresponding bridge,
> > -		 * because have already delayed for the bridge.
> > +		 * Mandatory power management transition delays are handled
> > +		 * in pci_pm_runtime_resume() of the corresponding
> > +		 * downstream/root port.
> >  		 */
> >  		if (dev->runtime_d3cold) {
> > -			if (dev->d3cold_delay && !dev->imm_ready)
> > -				msleep(dev->d3cold_delay);
> >  			/*
> >  			 * When powering on a bridge from D3cold, the
> >  			 * whole hierarchy may be powered on into
> > @@ -4607,14 +4603,17 @@ static int pci_pm_reset(struct pci_dev *dev, int probe)
> >  
> >  	return pci_dev_wait(dev, "PM D3->D0", PCIE_RESET_READY_POLL_MS);
> >  }
> > +
> >  /**
> > - * pcie_wait_for_link - Wait until link is active or inactive
> > + * pcie_wait_for_link_delay - Wait until link is active or inactive
> >   * @pdev: Bridge device
> >   * @active: waiting for active or inactive?
> > + * @delay: Delay to wait after link has become active (in ms)
> >   *
> >   * Use this to wait till link becomes active or inactive.
> >   */
> > -bool pcie_wait_for_link(struct pci_dev *pdev, bool active)
> > +static bool pcie_wait_for_link_delay(struct pci_dev *pdev, bool active,
> > +				     int delay)
> >  {
> >  	int timeout = 1000;
> >  	bool ret;
> > @@ -4651,13 +4650,121 @@ bool pcie_wait_for_link(struct pci_dev *pdev, bool active)
> >  		timeout -= 10;
> >  	}
> >  	if (active && ret)
> > -		msleep(100);
> > +		msleep(delay);
> >  	else if (ret != active)
> >  		pci_info(pdev, "Data Link Layer Link Active not %s in 1000 msec\n",
> >  			active ? "set" : "cleared");
> >  	return ret == active;
> >  }
> >  
> > +/**
> > + * pcie_wait_for_link - Wait until link is active or inactive
> > + * @pdev: Bridge device
> > + * @active: waiting for active or inactive?
> > + *
> > + * Use this to wait till link becomes active or inactive.
> > + */
> > +bool pcie_wait_for_link(struct pci_dev *pdev, bool active)
> > +{
> > +	return pcie_wait_for_link_delay(pdev, active, 100);
> > +}
> 
> This part (adding pcie_wait_for_link_delay() to make the delay time
> configurable) could be a separate patch to reduce the size of this
> patch and make the important pieces a little more obvious.

Indeed, will split it into a separate patch.

> 
> > +static int pcie_get_downstream_delay(struct pci_bus *bus)
> > +{
> > +	struct pci_dev *pdev;
> > +	int min_delay = 100;
> > +	int max_delay = 0;
> > +
> > +	list_for_each_entry(pdev, &bus->devices, bus_list) {
> > +		if (pdev->imm_ready)
> > +			min_delay = 0;
> > +		else if (pdev->d3cold_delay < min_delay)
> > +			min_delay = pdev->d3cold_delay;
> > +		if (pdev->d3cold_delay > max_delay)
> > +			max_delay = pdev->d3cold_delay;
> > +	}
> > +
> > +	return max(min_delay, max_delay);
> > +}
> > +
> > +/**
> > + * pcie_wait_downstream_accessible - Wait downstream device to be accessible
> > + * @pdev: PCIe port whose downstream device is waited
> > + *
> > + * Handle delays according to PCIe 4.0 section 6.6.1 before configuration
> > + * access to the downstream device is permitted. If the port does not have
> > + * any devices connected, does nothing.
> > + *
> > + * This is needed if the hierarchy below @pdev went through reset (after
> > + * exit from D3cold back to D0uninitialized).
> 
> D3cold -> D0 is a cold reset by definition, isn't it?

Yes.

> > + */
> > +void pcie_wait_downstream_accessible(struct pci_dev *pdev)
> > +{
> > +	struct pci_dev *child;
> > +	struct pci_bus *bus;
> > +	int delay;
> 
> IIUC pdev is not guaranteed to be a PCIe device here.

Good point, here we should check for pci_is_pcie() first.

> Do we need to observe the Trhfa requirements for Conventional PCI and
> PCI-X devices here?  If we don't do it here, where is it observed?

We probably should but I intented this to be PCIe specific since there
we have issues. For conventional PCI/PCI-X things "seem" to work and we
don't power manage those bridges anyway.

I'm not aware if Trhfa is handled in anywhere in the PCI stack
currently.

> > +	if (pci_pcie_type(pdev) != PCI_EXP_TYPE_ROOT_PORT &&
> > +	    pci_pcie_type(pdev) != PCI_EXP_TYPE_DOWNSTREAM)
> > +		return;
> > +
> > +	if (pci_dev_is_disconnected(pdev))
> > +		return;
> > +
> > +	if (!pdev->bridge_d3)
> > +		return;
> > +
> > +	bus = pdev->subordinate;
> > +	if (!bus)
> > +		return;
> > +
> > +	child = list_first_entry_or_null(&bus->devices, struct pci_dev,
> > +					 bus_list);
> > +	if (!child)
> > +		return;
> 
> I'm not convinced yet about skipping this if there's no subordinate
> bus or no children.  Don't we have to assume that the caller may
> immediately *probe* for children as soon as we return?

Yes, you are right. I was trying to avoid any unnecessary delays but
looks like we can't avoid them here.

> > +	delay = pcie_get_downstream_delay(bus);
> > +	if (!delay)
> > +		return;
> 
> I'm not sold on the idea that this delay depends on what's *below* the
> bridge.  We're using sec 6.6.1 to justify the delay, and that section
> doesn't say anything about downstream devices.

6.6.1 indeed talks about Downstream Ports and devices immediately below
them.

> If we call .resume_noirq/.runtime_resume for the downstream devices
> themselves, could we use d3cold_delay *there*?

Otherwise probably but since we already have resumed the downstream port
itself including pciehp it may notice that the link is not up anymore
and remove the device below.

> > +	/*
> > +	 * If downstream port does not support speeds greater than 5 GT/s
> > +	 * need to wait minimum 100ms. For higher speeds (gen3) we need to
> > +	 * wait first for the data link layer to become active.
> > +	 *
> > +	 * However, 100ms is the minimum and the spec says that the
> > +	 * software must allow at least 1s before it can determine that the
> > +	 * device that did not respond is a broken device. Also there is
> > +	 * evidence that the 100ms is not always enough so what we do here
> 
> This sort of "there is evidence" statement needs a specific reference.
> Otherwise it ends up meaning "we don't know why it matters and we
> can't test any changes in this area".

Sure I can refer to that bugzilla entry where the xHCI seems to require
more than 100ms.

> > +	 * is that we wait for the minimum 100ms (possibly after link
> > +	 * training completes) and then probe for the device presence once.
> > +	 * If we still don't get response we wait for another 100ms just to
> > +	 * give it some additional time to complete its initialization.
> > +	 */
> > +	if (pcie_get_speed_cap(pdev) <= PCIE_SPEED_5_0GT) {
> > +		dev_dbg(&pdev->dev, "waiting downstream link for %d ms\n",
> 
> pci_dbg()
> s/waiting downstream link for %d ms/waiting %d ms for downstream link/

OK.

> > +			delay);
> > +		msleep(delay);
> > +	} else {
> > +		dev_dbg(&pdev->dev,
> > +			"waiting downstream link for %d ms after activation\n",
> > +			delay);
> > +		if (!pcie_wait_for_link_delay(pdev, true, delay)) {
> > +			/*
> > +			 * If the link did not train, no need to wait
> > +			 * further the device is probably gone.
> > +			 */
> > +			return;
> > +		}
> > +	}
> > +
> > +	if (!pci_device_is_present(child)) {
> > +		dev_dbg(&child->dev,
> > +			"waiting for additional 100 ms for the device to become accessible\n");
> > +		msleep(100);
> 
> This seems like magic.  Can we relate this back to anything in the
> spec?  If we need this, maybe we're seeing CRRS? 

I assume CRRS == Configuration Request Retry Status? Unfortunately
that does not seem to be the case here because the root ports in
question on that bugzilla entry do not support CRS software visibility.

Only thing I can relate to the spec regarding this (again in 6.6.1) is
that it says "Software must wait minimum of 100ms". The only other limit
it gives is 1.0s that the software must allow before it can determine
the device is broken (well or missing, I suppose). So it pretty much
gives us a range of 100ms - 1000ms.

> I suppose this is related to the note above about evidence that 100ms
> may not be enough?

Yes indeed it is. I figured that we only need this for certain devices
so no need to do it for all.

We could also assume something like doubling the 100ms delay to be on
the safe side or so but that slows down resume paths even more.

> > +	}
> > +}
> > +
> >  void pci_reset_secondary_bus(struct pci_dev *dev)
> >  {
> >  	u16 ctrl;
> > diff --git a/drivers/pci/pci.h b/drivers/pci/pci.h
> > index d22d1b807701..9a83fcf612ca 100644
> > --- a/drivers/pci/pci.h
> > +++ b/drivers/pci/pci.h
> > @@ -498,6 +498,7 @@ void pcie_do_recovery(struct pci_dev *dev, enum pci_channel_state state,
> >  		      u32 service);
> >  
> >  bool pcie_wait_for_link(struct pci_dev *pdev, bool active);
> > +void pcie_wait_downstream_accessible(struct pci_dev *pdev);
> >  #ifdef CONFIG_PCIEASPM
> >  void pcie_aspm_init_link_state(struct pci_dev *pdev);
> >  void pcie_aspm_exit_link_state(struct pci_dev *pdev);
> > -- 
> > 2.23.0.rc1
> >
Bjorn Helgaas Aug. 26, 2019, 2:07 p.m. UTC | #6
On Mon, Aug 26, 2019 at 01:17:26PM +0300, Mika Westerberg wrote:
> On Fri, Aug 23, 2019 at 09:12:54PM -0500, Bjorn Helgaas wrote:
> > Hi Mika,
> 
> Hi,
> 
> > I'm trying to figure out specifically why we need this and where it
> > should go.  Questions below.
> 
> Thanks for looking at this.
> 
> > On Wed, Aug 21, 2019 at 03:45:19PM +0300, Mika Westerberg wrote:
> > > Currently Linux does not follow PCIe spec regarding the required delays
> > > after reset. A concrete example is a Thunderbolt add-in-card that
> > > consists of a PCIe switch and two PCIe endpoints:
> > > 
> > >   +-1b.0-[01-6b]----00.0-[02-6b]--+-00.0-[03]----00.0 TBT controller
> > >                                   +-01.0-[04-36]-- DS hotplug port
> > >                                   +-02.0-[37]----00.0 xHCI controller
> > >                                   \-04.0-[38-6b]-- DS hotplug port
> > > 
> > > The root port (1b.0) and the PCIe switch downstream ports are all PCIe
> > > gen3 so they support 8GT/s link speeds.
> > > 
> > > We wait for the PCIe hierarchy to enter D3cold (runtime):
> > > 
> > >   pcieport 0000:00:1b.0: power state changed by ACPI to D3cold
> > > 
> > > When it wakes up from D3cold, according to the PCIe 4.0 section 5.8 the
> > > PCIe switch is put to reset and its power is re-applied. This means that
> > > we must follow the rules in PCIe 4.0 section 6.6.1.
> > > 
> > > For the PCIe gen3 ports we are dealing with here, the following applies:
> > > 
> > >   With a Downstream Port that supports Link speeds greater than 5.0
> > >   GT/s, software must wait a minimum of 100 ms after Link training
> > >   completes before sending a Configuration Request to the device
> > >   immediately below that Port. Software can determine when Link training
> > >   completes by polling the Data Link Layer Link Active bit or by setting
> > >   up an associated interrupt (see Section 6.7.3.3).
> > > 
> > > Translating this into the above topology we would need to do this (DLLLA
> > > stands for Data Link Layer Link Active):
> > > 
> > >   pcieport 0000:00:1b.0: wait for 100ms after DLLLA is set before access to 0000:01:00.0
> > >   pcieport 0000:02:00.0: wait for 100ms after DLLLA is set before access to 0000:03:00.0
> > >   pcieport 0000:02:02.0: wait for 100ms after DLLLA is set before access to 0000:37:00.0
> > > 
> > > I've instrumented the kernel with additional logging so we can see the
> > > actual delays the kernel performs:
> > > ...

> > >   xhci_hcd 0000:37:00.0: restoring config space at offset 0x10 (was 0x0, writing 0x73f00000)
> > >   ...
> > >   thunderbolt 0000:03:00.0: restoring config space at offset 0x14 (was 0x0, writing 0x8a040000)
> > > 
> > > This is even worse. None of the mandatory delays are performed. If this
> > > would be S3 instead of s2idle then according to PCI FW spec 3.2 section
> > > 4.6.8.  there is a specific _DSM that allows the OS to skip the delays
> > > but this platform does not provide the _DSM and does not go to S3 anyway
> > > so no firmware is involved that could already handle these delays.
> > > 
> > > In this particular Intel Coffee Lake platform these delays are not
> > > actually needed because there is an additional delay as part of the ACPI
> > > power resource that is used to turn on power to the hierarchy but since
> > > that additional delay is not required by any of standards (PCIe, ACPI)
> > 
> > So it sounds like this Coffee Lake accidentally works because of
> > unrelated firmware delay that's not actually required, or at least not
> > related to the delay required by PCIe?
> 
> Correct. The root port ACPI Power Resource includes quite long delay
> which allows the links to be trained before the _ON method is finished.
> 
> > I did notice that we don't implement all of _DSM function 9 and the
> > parts we're missing look like they could be relevant.
> 
> AFAICT that _DSM is only for lowering the delays, not increasing them.
> If the platform does not provide it the OS must adhere to all the timing
> requirements of the PCIe spec (PCI FW 3.2 section 4.6.9). Here we are
> actually trying to do just that :)

Yep, true.

> > > it is not present in the Intel Ice Lake, for example where missing the
> > > mandatory delays causes pciehp to start tearing down the stack too early
> > > (links are not yet trained).
> > 
> > I'm guessing the Coffee Lake/Ice Lake distinction is not really
> > relevant and the important thing is that something about Ice Lake is
> > faster and reduces the accidental delay to the point that things stop
> > working.
> > 
> > So probably the same thing could happen on Coffee Lake or any other
> > system if it had different firmware.
> 
> Indeed.

Thanks for confirming that.  I would probably abstract this somehow in
the commit log so people don't have the impression that we're fixing
something specific to Ice Lake.

> > > There is also one reported case (see the bugzilla link below) where the
> > > missing delay causes xHCI on a Titan Ridge controller fail to runtime
> > > resume when USB-C dock is plugged.
> > > 
> > > For this reason, introduce a new function pcie_wait_downstream_accessible()
> > > that is called on PCI core resume and runtime resume paths accordingly
> > > if downstream/root port with device below entered D3cold.
> > > 
> > > This is second attempt to add the missing delays. The previous solution
> > > in commit c2bf1fc212f7 ("PCI: Add missing link delays required by the
> > > PCIe spec") was reverted because of two issues it caused:
> > 
> > c2bf1fc212f7 was merged for v5.3 (it appeared in v5.3-rc1).  I *guess*
> > it addressed https://bugzilla.kernel.org/show_bug.cgi?id=203885, which
> > was reported on v5.2-rc4, though c2bf1fc212f7 doesn't actually mention
> > the bugzilla?
> 
> I think c2bf1fc212f7 was merged before I was aware of that bug (or I
> just missed it).

Ah, so there's some other issue that prompted this patch in the first
place.

> > 0617bdede511 ("Revert "PCI: Add missing link delays required by the
> > PCIe spec"") appeared in v5.3-rc4.  If c2bf1fc212f7 was supposed to
> > address BZ 203885, I'm confused about why you asked Kai-Heng to test
> > v5.3-rc4, where c2bf1fc212f7 had already been reverted.
> 
> I was hoping that the revert in v5.3-rc4 would help here as well but
> after futher investigation it turned to be the exact opposite.
> 
> > Or maybe c2bf1fc212f7 wasn't connected with BZ 203885 in the first
> > place?
> > 
> > The net result is that I think v5.3-rc4 is equivalent to v5.2 with
> > respect to this issue.  It *is* a pretty serious usability issue, no
> > question, but it's not completely obvious that this fix needs to be in
> > v5.3 since it's not something we broke during the v5.3 merge window.
> > So if we *do* want it in v5.3, we need to think about how to justify
> > that, e.g., if this issue affects shipping systems that are likely to
> > run an upstream kernel, etc.
> 
> I don't think it needs to be in v5.3. I was thinking more like v5.4 so
> we get some more testing coverage.

Sounds good.

> > >   1. One system become unresponsive after S3 resume due to PME service
> > >      spinning in pcie_pme_work_fn(). The root port in question reports
> > >      that the xHCI sent PME but the xHCI device itself does not have PME
> > >      status set. The PME status bit is never cleared in the root port
> > >      resulting the indefinite loop in pcie_pme_work_fn().
> > 
> > I don't see the connection between PME and either c2bf1fc212f7 or this
> > patch.  Is there a fix for this pcie_pme_work_fn() infinite loop?
> 
> No.
> 
> > I do see that BZ 203885 mentions a flood of "PME: Spurious native
> > interrupt!" messages, but I don't see how we've fixed that.
> 
> It was not completely root caused yet. To me it looks like the root port
> keeps the PME status and pending bits set and that causes
> pcie_pme_work_fn() spin forever. From the debugging output logs I got
> from Matthias, it is the xHCI that seems to send the PME but it does not
> have PME status set for some reason.

I'm not sure what the value of mentioning this PME issue is.  If this
patch happens to mask the PME issue, maybe it's worth a brief mention
just to say "this patch masks but does not fix this problem".
Otherwise it's a distraction.

> > >   2. Slows down resume if the root/downstream port does not support
> > >      Data Link Layer Active Reporting because pcie_wait_for_link_delay()
> > >      waits 1100ms in that case.
> > 
> > I don't see the slowdown mentioned in BZ 203885; is there another link
> > to these reports?
> 
> Nicholas for example reported it here:
> 
>   https://lore.kernel.org/linux-pci/SL2P216MB01878BBCD75F21D882AEEA2880C60@SL2P216MB0187.KORP216.PROD.OUTLOOK.COM/
> 
> I should have added that link to the changelog.
> 
> > > This version should avoid the above issues because we restrict the delay
> > > to happen only if the port went into D3cold (so it goes through reset)
> > > and only when there is no firmware involved on resume path (so the
> > > kernel is responsible for all the delays).
> > > 
> > > Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=203885
> > > Reported-by: Kai-Heng Feng <kai.heng.feng@canonical.com>
> > > Tested-by: Kai-Heng Feng <kai.heng.feng@canonical.com>
> > > Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
> > > ---
> > > Hi all,
> > > 
> > > As the changelog says this is reworked version that tries to avoid reported
> > > issues while at the same time fix the missing delays so we can get ICL
> > > systems and at least the one system with Titan Ridge controller working
> > > properly.
> > > 
> > > @Matthias, @Paul and @Nicholas: it would be great if you could try the
> > > patch on top of v5.4-rc5+ and verify that it does not cause any issues on
> > > your systems.
> > > 
> > >  drivers/pci/pci-driver.c |  19 ++++++
> > >  drivers/pci/pci.c        | 127 ++++++++++++++++++++++++++++++++++++---
> > >  drivers/pci/pci.h        |   1 +
> > >  3 files changed, 137 insertions(+), 10 deletions(-)
> > > 
> > > diff --git a/drivers/pci/pci-driver.c b/drivers/pci/pci-driver.c
> > > index a8124e47bf6e..9aec78ed8907 100644
> > > --- a/drivers/pci/pci-driver.c
> > > +++ b/drivers/pci/pci-driver.c
> > > @@ -918,6 +918,7 @@ static int pci_pm_resume_noirq(struct device *dev)
> > >  {
> > >  	struct pci_dev *pci_dev = to_pci_dev(dev);
> > >  	struct device_driver *drv = dev->driver;
> > > +	pci_power_t state = pci_dev->current_state;
> > >  	int error = 0;
> > >  
> > >  	if (dev_pm_may_skip_resume(dev))
> > > @@ -947,6 +948,15 @@ static int pci_pm_resume_noirq(struct device *dev)
> > >  
> > >  	pcie_pme_root_status_cleanup(pci_dev);
> > >  
> > > +	/*
> > > +	 * If resume involves firmware assume it takes care of any delays
> > > +	 * for now. For suspend-to-idle case we need to do that here before
> > > +	 * resuming PCIe port services to keep pciehp from tearing down the
> > > +	 * downstream devices too early.
> > > +	 */
> > > +	if (state == PCI_D3cold && pm_suspend_no_platform())
> > > +		pcie_wait_downstream_accessible(pci_dev);
> > 
> > Aren't these paths used for Conventional PCI devices as well as PCIe
> > devices?
> 
> Indeed they are.
> 
> > I think the D3cold and pm_suspend_no_platform() checks should move
> > inside the new interface, whatever it's called.  I'm not sure what
> > that means for the fact that you don't check pm_suspend_no_platform()
> > in the runtime-resume path; maybe it needs a flag or something.
> 
> It is not needed for runtime-resume path since we are not entering
> system suspend. For runtime-resume path we need to perform the delays
> always.

From the hardware device point of view, I don't think the
system/runtime resume distinction is important.  The only thing the
devices know is that they're changing from D3->D0, and it doesn't
matter whether it's because of a runtime resume or an entire system
resume.

> Here we just assume the firmware actually handles the delays which is
> not entirely accurate. According to PCI FW 3.2 section 4.6.8 there is a
> _DSM that the firmware can use to indicate to the OS that it can skip
> the reset delays. So I think more correct would be to check for that in
> resume path. However, since it looks like that _DSM is not implemented
> in too many systems I've seen, it may slow down resume times
> unnecessarily if the firmware actually handles the delays.
> 
> > But the "wait downstream" part seems a little too specific to be at
> > the .resume_noirq and .runtime_resume level.
> > 
> > Do we descend the hierarchy and call .resume_noirq and .runtime_resume
> > for the children of the bridge, too?
> 
> We do but at that time it is too late as we have already resumed pciehp
> of the parent downstream port and it may have already started tearing
> down the device stack below.
> 
> I'm open to any better ideas where this delay could be added, though :)

So we resume pciehp *before* we're finished resuming the Downstream
Port?  That sounds wrong.

> > > +static int pcie_get_downstream_delay(struct pci_bus *bus)
> > > +{
> > > +	struct pci_dev *pdev;
> > > +	int min_delay = 100;
> > > +	int max_delay = 0;
> > > +
> > > +	list_for_each_entry(pdev, &bus->devices, bus_list) {
> > > +		if (pdev->imm_ready)
> > > +			min_delay = 0;
> > > +		else if (pdev->d3cold_delay < min_delay)
> > > +			min_delay = pdev->d3cold_delay;
> > > +		if (pdev->d3cold_delay > max_delay)
> > > +			max_delay = pdev->d3cold_delay;
> > > +	}
> > > +
> > > +	return max(min_delay, max_delay);
> > > +}

> > > + */
> > > +void pcie_wait_downstream_accessible(struct pci_dev *pdev)

> > Do we need to observe the Trhfa requirements for Conventional PCI and
> > PCI-X devices here?  If we don't do it here, where is it observed?
> 
> We probably should but I intented this to be PCIe specific since there
> we have issues. For conventional PCI/PCI-X things "seem" to work and we
> don't power manage those bridges anyway.
> 
> I'm not aware if Trhfa is handled in anywhere in the PCI stack
> currently.

I think we should make this agnostic of the Conventional/PCIe
difference if possible.  I assume we can tell if we're dealing with a
D3->D0 transition and we only add delays in that case.  If we don't
power manage Conventional PCI devices, I assume we won't see D3->D0
transitions for runtime resume so there won't be any harm.

Making it PCIe-specific seems like it adds extra code ("dev-is-PCIe
checks") with no obvious reason for existence and an implicit
dependency on the fact that we only power manage PCIe devices.  If we
ever *did* support runtime power-management for Conventional PCI, we'd
trip over that implicit dependency and probably debug this issue
again.

But I guess it might slow down system resume for Conventional PCI
devices.  If we rely on delays in firmware, I wonder if there's
any point during resume where we could grab an early timestamp, then
take another timestamp here and deduce that we've already delayed the
difference?

> > > +	if (pci_pcie_type(pdev) != PCI_EXP_TYPE_ROOT_PORT &&
> > > +	    pci_pcie_type(pdev) != PCI_EXP_TYPE_DOWNSTREAM)
> > > +		return;
> > > +
> > > +	if (pci_dev_is_disconnected(pdev))
> > > +		return;
> > > +
> > > +	if (!pdev->bridge_d3)
> > > +		return;
> > > +
> > > +	bus = pdev->subordinate;
> > > +	if (!bus)
> > > +		return;
> > > +
> > > +	child = list_first_entry_or_null(&bus->devices, struct pci_dev,
> > > +					 bus_list);
> > > +	if (!child)
> > > +		return;
> > 
> > I'm not convinced yet about skipping this if there's no subordinate
> > bus or no children.  Don't we have to assume that the caller may
> > immediately *probe* for children as soon as we return?
> 
> Yes, you are right. I was trying to avoid any unnecessary delays but
> looks like we can't avoid them here.
> 
> > > +	delay = pcie_get_downstream_delay(bus);
> > > +	if (!delay)
> > > +		return;
> > 
> > I'm not sold on the idea that this delay depends on what's *below* the
> > bridge.  We're using sec 6.6.1 to justify the delay, and that section
> > doesn't say anything about downstream devices.
> 
> 6.6.1 indeed talks about Downstream Ports and devices immediately below
> them.

Wait, I don't think we're reading this the same way.  6.6.1 mentions
Downstream Ports: "With a Downstream Port that does not support Link
speeds greater than 5.0 GT/s, software must wait a minimum of 100 ms
before sending a Configuration Request to the device immediately below
that Port."

This says we have to delay before sending a config request to a device
below a Downstream Port, but it doesn't say anything about the
characteristics of that device.  In particular, I don't think it says
the delay can be shortened if that device supports Immediate Readiness
or implements a readiness _DSM.

I don't think this delay has anything to do with devices downstream
from the Port.  I think this is about giving the Downstream Port time
to bring up the link.  That way config requests may fail because
there's no device below the Port or it's not ready yet, but they
shouldn't fail simply because the link isn't up yet.

> > If we call .resume_noirq/.runtime_resume for the downstream devices
> > themselves, could we use d3cold_delay *there*?
> 
> Otherwise probably but since we already have resumed the downstream port
> itself including pciehp it may notice that the link is not up anymore
> and remove the device below.

I think this sounds like a problem with pciehp or the order in which
we resume devices and portdrv services.

Bjorn
Mika Westerberg Aug. 26, 2019, 2:42 p.m. UTC | #7
On Mon, Aug 26, 2019 at 09:07:12AM -0500, Bjorn Helgaas wrote:
> On Mon, Aug 26, 2019 at 01:17:26PM +0300, Mika Westerberg wrote:
> > On Fri, Aug 23, 2019 at 09:12:54PM -0500, Bjorn Helgaas wrote:
> > > Hi Mika,
> > 
> > Hi,
> > 
> > > I'm trying to figure out specifically why we need this and where it
> > > should go.  Questions below.
> > 
> > Thanks for looking at this.
> > 
> > > On Wed, Aug 21, 2019 at 03:45:19PM +0300, Mika Westerberg wrote:
> > > > Currently Linux does not follow PCIe spec regarding the required delays
> > > > after reset. A concrete example is a Thunderbolt add-in-card that
> > > > consists of a PCIe switch and two PCIe endpoints:
> > > > 
> > > >   +-1b.0-[01-6b]----00.0-[02-6b]--+-00.0-[03]----00.0 TBT controller
> > > >                                   +-01.0-[04-36]-- DS hotplug port
> > > >                                   +-02.0-[37]----00.0 xHCI controller
> > > >                                   \-04.0-[38-6b]-- DS hotplug port
> > > > 
> > > > The root port (1b.0) and the PCIe switch downstream ports are all PCIe
> > > > gen3 so they support 8GT/s link speeds.
> > > > 
> > > > We wait for the PCIe hierarchy to enter D3cold (runtime):
> > > > 
> > > >   pcieport 0000:00:1b.0: power state changed by ACPI to D3cold
> > > > 
> > > > When it wakes up from D3cold, according to the PCIe 4.0 section 5.8 the
> > > > PCIe switch is put to reset and its power is re-applied. This means that
> > > > we must follow the rules in PCIe 4.0 section 6.6.1.
> > > > 
> > > > For the PCIe gen3 ports we are dealing with here, the following applies:
> > > > 
> > > >   With a Downstream Port that supports Link speeds greater than 5.0
> > > >   GT/s, software must wait a minimum of 100 ms after Link training
> > > >   completes before sending a Configuration Request to the device
> > > >   immediately below that Port. Software can determine when Link training
> > > >   completes by polling the Data Link Layer Link Active bit or by setting
> > > >   up an associated interrupt (see Section 6.7.3.3).
> > > > 
> > > > Translating this into the above topology we would need to do this (DLLLA
> > > > stands for Data Link Layer Link Active):
> > > > 
> > > >   pcieport 0000:00:1b.0: wait for 100ms after DLLLA is set before access to 0000:01:00.0
> > > >   pcieport 0000:02:00.0: wait for 100ms after DLLLA is set before access to 0000:03:00.0
> > > >   pcieport 0000:02:02.0: wait for 100ms after DLLLA is set before access to 0000:37:00.0
> > > > 
> > > > I've instrumented the kernel with additional logging so we can see the
> > > > actual delays the kernel performs:
> > > > ...
> 
> > > >   xhci_hcd 0000:37:00.0: restoring config space at offset 0x10 (was 0x0, writing 0x73f00000)
> > > >   ...
> > > >   thunderbolt 0000:03:00.0: restoring config space at offset 0x14 (was 0x0, writing 0x8a040000)
> > > > 
> > > > This is even worse. None of the mandatory delays are performed. If this
> > > > would be S3 instead of s2idle then according to PCI FW spec 3.2 section
> > > > 4.6.8.  there is a specific _DSM that allows the OS to skip the delays
> > > > but this platform does not provide the _DSM and does not go to S3 anyway
> > > > so no firmware is involved that could already handle these delays.
> > > > 
> > > > In this particular Intel Coffee Lake platform these delays are not
> > > > actually needed because there is an additional delay as part of the ACPI
> > > > power resource that is used to turn on power to the hierarchy but since
> > > > that additional delay is not required by any of standards (PCIe, ACPI)
> > > 
> > > So it sounds like this Coffee Lake accidentally works because of
> > > unrelated firmware delay that's not actually required, or at least not
> > > related to the delay required by PCIe?
> > 
> > Correct. The root port ACPI Power Resource includes quite long delay
> > which allows the links to be trained before the _ON method is finished.
> > 
> > > I did notice that we don't implement all of _DSM function 9 and the
> > > parts we're missing look like they could be relevant.
> > 
> > AFAICT that _DSM is only for lowering the delays, not increasing them.
> > If the platform does not provide it the OS must adhere to all the timing
> > requirements of the PCIe spec (PCI FW 3.2 section 4.6.9). Here we are
> > actually trying to do just that :)
> 
> Yep, true.
> 
> > > > it is not present in the Intel Ice Lake, for example where missing the
> > > > mandatory delays causes pciehp to start tearing down the stack too early
> > > > (links are not yet trained).
> > > 
> > > I'm guessing the Coffee Lake/Ice Lake distinction is not really
> > > relevant and the important thing is that something about Ice Lake is
> > > faster and reduces the accidental delay to the point that things stop
> > > working.
> > > 
> > > So probably the same thing could happen on Coffee Lake or any other
> > > system if it had different firmware.
> > 
> > Indeed.
> 
> Thanks for confirming that.  I would probably abstract this somehow in
> the commit log so people don't have the impression that we're fixing
> something specific to Ice Lake.

OK.

> > > > There is also one reported case (see the bugzilla link below) where the
> > > > missing delay causes xHCI on a Titan Ridge controller fail to runtime
> > > > resume when USB-C dock is plugged.
> > > > 
> > > > For this reason, introduce a new function pcie_wait_downstream_accessible()
> > > > that is called on PCI core resume and runtime resume paths accordingly
> > > > if downstream/root port with device below entered D3cold.
> > > > 
> > > > This is second attempt to add the missing delays. The previous solution
> > > > in commit c2bf1fc212f7 ("PCI: Add missing link delays required by the
> > > > PCIe spec") was reverted because of two issues it caused:
> > > 
> > > c2bf1fc212f7 was merged for v5.3 (it appeared in v5.3-rc1).  I *guess*
> > > it addressed https://bugzilla.kernel.org/show_bug.cgi?id=203885, which
> > > was reported on v5.2-rc4, though c2bf1fc212f7 doesn't actually mention
> > > the bugzilla?
> > 
> > I think c2bf1fc212f7 was merged before I was aware of that bug (or I
> > just missed it).
> 
> Ah, so there's some other issue that prompted this patch in the first
> place.

Yes, the Ice Lake issue :)

> > > 0617bdede511 ("Revert "PCI: Add missing link delays required by the
> > > PCIe spec"") appeared in v5.3-rc4.  If c2bf1fc212f7 was supposed to
> > > address BZ 203885, I'm confused about why you asked Kai-Heng to test
> > > v5.3-rc4, where c2bf1fc212f7 had already been reverted.
> > 
> > I was hoping that the revert in v5.3-rc4 would help here as well but
> > after futher investigation it turned to be the exact opposite.
> > 
> > > Or maybe c2bf1fc212f7 wasn't connected with BZ 203885 in the first
> > > place?
> > > 
> > > The net result is that I think v5.3-rc4 is equivalent to v5.2 with
> > > respect to this issue.  It *is* a pretty serious usability issue, no
> > > question, but it's not completely obvious that this fix needs to be in
> > > v5.3 since it's not something we broke during the v5.3 merge window.
> > > So if we *do* want it in v5.3, we need to think about how to justify
> > > that, e.g., if this issue affects shipping systems that are likely to
> > > run an upstream kernel, etc.
> > 
> > I don't think it needs to be in v5.3. I was thinking more like v5.4 so
> > we get some more testing coverage.
> 
> Sounds good.
> 
> > > >   1. One system become unresponsive after S3 resume due to PME service
> > > >      spinning in pcie_pme_work_fn(). The root port in question reports
> > > >      that the xHCI sent PME but the xHCI device itself does not have PME
> > > >      status set. The PME status bit is never cleared in the root port
> > > >      resulting the indefinite loop in pcie_pme_work_fn().
> > > 
> > > I don't see the connection between PME and either c2bf1fc212f7 or this
> > > patch.  Is there a fix for this pcie_pme_work_fn() infinite loop?
> > 
> > No.
> > 
> > > I do see that BZ 203885 mentions a flood of "PME: Spurious native
> > > interrupt!" messages, but I don't see how we've fixed that.
> > 
> > It was not completely root caused yet. To me it looks like the root port
> > keeps the PME status and pending bits set and that causes
> > pcie_pme_work_fn() spin forever. From the debugging output logs I got
> > from Matthias, it is the xHCI that seems to send the PME but it does not
> > have PME status set for some reason.
> 
> I'm not sure what the value of mentioning this PME issue is.  If this
> patch happens to mask the PME issue, maybe it's worth a brief mention
> just to say "this patch masks but does not fix this problem".
> Otherwise it's a distraction.

OK.

> > > >   2. Slows down resume if the root/downstream port does not support
> > > >      Data Link Layer Active Reporting because pcie_wait_for_link_delay()
> > > >      waits 1100ms in that case.
> > > 
> > > I don't see the slowdown mentioned in BZ 203885; is there another link
> > > to these reports?
> > 
> > Nicholas for example reported it here:
> > 
> >   https://lore.kernel.org/linux-pci/SL2P216MB01878BBCD75F21D882AEEA2880C60@SL2P216MB0187.KORP216.PROD.OUTLOOK.COM/
> > 
> > I should have added that link to the changelog.
> > 
> > > > This version should avoid the above issues because we restrict the delay
> > > > to happen only if the port went into D3cold (so it goes through reset)
> > > > and only when there is no firmware involved on resume path (so the
> > > > kernel is responsible for all the delays).
> > > > 
> > > > Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=203885
> > > > Reported-by: Kai-Heng Feng <kai.heng.feng@canonical.com>
> > > > Tested-by: Kai-Heng Feng <kai.heng.feng@canonical.com>
> > > > Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
> > > > ---
> > > > Hi all,
> > > > 
> > > > As the changelog says this is reworked version that tries to avoid reported
> > > > issues while at the same time fix the missing delays so we can get ICL
> > > > systems and at least the one system with Titan Ridge controller working
> > > > properly.
> > > > 
> > > > @Matthias, @Paul and @Nicholas: it would be great if you could try the
> > > > patch on top of v5.4-rc5+ and verify that it does not cause any issues on
> > > > your systems.
> > > > 
> > > >  drivers/pci/pci-driver.c |  19 ++++++
> > > >  drivers/pci/pci.c        | 127 ++++++++++++++++++++++++++++++++++++---
> > > >  drivers/pci/pci.h        |   1 +
> > > >  3 files changed, 137 insertions(+), 10 deletions(-)
> > > > 
> > > > diff --git a/drivers/pci/pci-driver.c b/drivers/pci/pci-driver.c
> > > > index a8124e47bf6e..9aec78ed8907 100644
> > > > --- a/drivers/pci/pci-driver.c
> > > > +++ b/drivers/pci/pci-driver.c
> > > > @@ -918,6 +918,7 @@ static int pci_pm_resume_noirq(struct device *dev)
> > > >  {
> > > >  	struct pci_dev *pci_dev = to_pci_dev(dev);
> > > >  	struct device_driver *drv = dev->driver;
> > > > +	pci_power_t state = pci_dev->current_state;
> > > >  	int error = 0;
> > > >  
> > > >  	if (dev_pm_may_skip_resume(dev))
> > > > @@ -947,6 +948,15 @@ static int pci_pm_resume_noirq(struct device *dev)
> > > >  
> > > >  	pcie_pme_root_status_cleanup(pci_dev);
> > > >  
> > > > +	/*
> > > > +	 * If resume involves firmware assume it takes care of any delays
> > > > +	 * for now. For suspend-to-idle case we need to do that here before
> > > > +	 * resuming PCIe port services to keep pciehp from tearing down the
> > > > +	 * downstream devices too early.
> > > > +	 */
> > > > +	if (state == PCI_D3cold && pm_suspend_no_platform())
> > > > +		pcie_wait_downstream_accessible(pci_dev);
> > > 
> > > Aren't these paths used for Conventional PCI devices as well as PCIe
> > > devices?
> > 
> > Indeed they are.
> > 
> > > I think the D3cold and pm_suspend_no_platform() checks should move
> > > inside the new interface, whatever it's called.  I'm not sure what
> > > that means for the fact that you don't check pm_suspend_no_platform()
> > > in the runtime-resume path; maybe it needs a flag or something.
> > 
> > It is not needed for runtime-resume path since we are not entering
> > system suspend. For runtime-resume path we need to perform the delays
> > always.
> 
> >From the hardware device point of view, I don't think the
> system/runtime resume distinction is important.  The only thing the
> devices know is that they're changing from D3->D0, and it doesn't
> matter whether it's because of a runtime resume or an entire system
> resume.
> 
> > Here we just assume the firmware actually handles the delays which is
> > not entirely accurate. According to PCI FW 3.2 section 4.6.8 there is a
> > _DSM that the firmware can use to indicate to the OS that it can skip
> > the reset delays. So I think more correct would be to check for that in
> > resume path. However, since it looks like that _DSM is not implemented
> > in too many systems I've seen, it may slow down resume times
> > unnecessarily if the firmware actually handles the delays.
> > 
> > > But the "wait downstream" part seems a little too specific to be at
> > > the .resume_noirq and .runtime_resume level.
> > > 
> > > Do we descend the hierarchy and call .resume_noirq and .runtime_resume
> > > for the children of the bridge, too?
> > 
> > We do but at that time it is too late as we have already resumed pciehp
> > of the parent downstream port and it may have already started tearing
> > down the device stack below.
> > 
> > I'm open to any better ideas where this delay could be added, though :)
> 
> So we resume pciehp *before* we're finished resuming the Downstream
> Port?  That sounds wrong.

I mean once we resume the downstream port (the bridge) we also resume
"PCIe port services" including pciehp and only then descend to whatever
device is physically connected to that port.

> > > > +static int pcie_get_downstream_delay(struct pci_bus *bus)
> > > > +{
> > > > +	struct pci_dev *pdev;
> > > > +	int min_delay = 100;
> > > > +	int max_delay = 0;
> > > > +
> > > > +	list_for_each_entry(pdev, &bus->devices, bus_list) {
> > > > +		if (pdev->imm_ready)
> > > > +			min_delay = 0;
> > > > +		else if (pdev->d3cold_delay < min_delay)
> > > > +			min_delay = pdev->d3cold_delay;
> > > > +		if (pdev->d3cold_delay > max_delay)
> > > > +			max_delay = pdev->d3cold_delay;
> > > > +	}
> > > > +
> > > > +	return max(min_delay, max_delay);
> > > > +}
> 
> > > > + */
> > > > +void pcie_wait_downstream_accessible(struct pci_dev *pdev)
> 
> > > Do we need to observe the Trhfa requirements for Conventional PCI and
> > > PCI-X devices here?  If we don't do it here, where is it observed?
> > 
> > We probably should but I intented this to be PCIe specific since there
> > we have issues. For conventional PCI/PCI-X things "seem" to work and we
> > don't power manage those bridges anyway.
> > 
> > I'm not aware if Trhfa is handled in anywhere in the PCI stack
> > currently.
> 
> I think we should make this agnostic of the Conventional/PCIe
> difference if possible.  I assume we can tell if we're dealing with a
> D3->D0 transition and we only add delays in that case.  If we don't
> power manage Conventional PCI devices, I assume we won't see D3->D0
> transitions for runtime resume so there won't be any harm.
>
> Making it PCIe-specific seems like it adds extra code ("dev-is-PCIe
> checks") with no obvious reason for existence and an implicit
> dependency on the fact that we only power manage PCIe devices.  If we
> ever *did* support runtime power-management for Conventional PCI, we'd
> trip over that implicit dependency and probably debug this issue
> again.
> 
> But I guess it might slow down system resume for Conventional PCI
> devices.  If we rely on delays in firmware, I wonder if there's
> any point during resume where we could grab an early timestamp, then
> take another timestamp here and deduce that we've already delayed the
> difference?

That sounds rather complex, to be honest ;-)

According to PCI FW 3.2 the OS is responsible for those delays if the
platform firmware does not provide that _DSM. So simpler way would be
always do the delays if the _DSM is not there.

> > > > +	if (pci_pcie_type(pdev) != PCI_EXP_TYPE_ROOT_PORT &&
> > > > +	    pci_pcie_type(pdev) != PCI_EXP_TYPE_DOWNSTREAM)
> > > > +		return;
> > > > +
> > > > +	if (pci_dev_is_disconnected(pdev))
> > > > +		return;
> > > > +
> > > > +	if (!pdev->bridge_d3)
> > > > +		return;
> > > > +
> > > > +	bus = pdev->subordinate;
> > > > +	if (!bus)
> > > > +		return;
> > > > +
> > > > +	child = list_first_entry_or_null(&bus->devices, struct pci_dev,
> > > > +					 bus_list);
> > > > +	if (!child)
> > > > +		return;
> > > 
> > > I'm not convinced yet about skipping this if there's no subordinate
> > > bus or no children.  Don't we have to assume that the caller may
> > > immediately *probe* for children as soon as we return?
> > 
> > Yes, you are right. I was trying to avoid any unnecessary delays but
> > looks like we can't avoid them here.
> > 
> > > > +	delay = pcie_get_downstream_delay(bus);
> > > > +	if (!delay)
> > > > +		return;
> > > 
> > > I'm not sold on the idea that this delay depends on what's *below* the
> > > bridge.  We're using sec 6.6.1 to justify the delay, and that section
> > > doesn't say anything about downstream devices.
> > 
> > 6.6.1 indeed talks about Downstream Ports and devices immediately below
> > them.
> 
> Wait, I don't think we're reading this the same way.  6.6.1 mentions
> Downstream Ports: "With a Downstream Port that does not support Link
> speeds greater than 5.0 GT/s, software must wait a minimum of 100 ms
> before sending a Configuration Request to the device immediately below
> that Port."
> 
> This says we have to delay before sending a config request to a device
> below a Downstream Port, but it doesn't say anything about the
> characteristics of that device.  In particular, I don't think it says
> the delay can be shortened if that device supports Immediate Readiness
> or implements a readiness _DSM.

Well it says this:

  To allow components to perform internal initialization, system software
  must wait a specified minimum period following the end of a Conventional
  Reset of one or more devices before it is permitted to issue
  Configuration Requests to those devices, unless Readiness Notifications
  mechanisms are used

My understanding of the above (might be wrong) is that Readiness
Notification can shorten the delay.

> I don't think this delay has anything to do with devices downstream
> from the Port.  I think this is about giving the Downstream Port time
> to bring up the link.  That way config requests may fail because
> there's no device below the Port or it's not ready yet, but they
> shouldn't fail simply because the link isn't up yet.

My understanding (again might be wrong) is that there are two factors
here. One is to get the link trained and the other is to give the
downstream device some time to perform its internal initialization so
that it is ready to answer config requests.

> > > If we call .resume_noirq/.runtime_resume for the downstream devices
> > > themselves, could we use d3cold_delay *there*?
> > 
> > Otherwise probably but since we already have resumed the downstream port
> > itself including pciehp it may notice that the link is not up anymore
> > and remove the device below.
> 
> I think this sounds like a problem with pciehp or the order in which
> we resume devices and portdrv services.

Well the order is from parent to children and in this case parent is the
PCIe downstream port including its port services. I'm not entirely sure
which order they get resumed but I think they are still resumed before
any of the children (probably depends on which order port services get
registered vs. the children gets enumerated).
Matthias Andree Aug. 26, 2019, 4:16 p.m. UTC | #8
Am 26.08.19 um 16:42 schrieb Mika Westerberg:
> According to PCI FW 3.2 the OS is responsible for those delays if the
> platform firmware does not provide that _DSM. So simpler way would be
> always do the delays if the _DSM is not there.

Well, that is

* unless we trip over delays that make other parts of the kernel spin
crazy in unterminated loops because there's this mismatch between the
PME and the readyness world or whatever you analyzed my system PME
interrupt storms for.

Note we normally resume devices asynchronously these days unless someone
zeroes out /sys/power/pm_async.

* assuming we're not seeing complaints from I think it was Nicholas's
system with wakeup/resume deadlines.
Bjorn Helgaas Aug. 26, 2019, 10:05 p.m. UTC | #9
On Mon, Aug 26, 2019 at 05:42:42PM +0300, Mika Westerberg wrote:
> On Mon, Aug 26, 2019 at 09:07:12AM -0500, Bjorn Helgaas wrote:
> > On Mon, Aug 26, 2019 at 01:17:26PM +0300, Mika Westerberg wrote:
> > > On Fri, Aug 23, 2019 at 09:12:54PM -0500, Bjorn Helgaas wrote:

> > > > But the "wait downstream" part seems a little too specific to be at
> > > > the .resume_noirq and .runtime_resume level.
> > > > 
> > > > Do we descend the hierarchy and call .resume_noirq and .runtime_resume
> > > > for the children of the bridge, too?
> > > 
> > > We do but at that time it is too late as we have already resumed pciehp
> > > of the parent downstream port and it may have already started tearing
> > > down the device stack below.
> > > 
> > > I'm open to any better ideas where this delay could be added, though :)
> > 
> > So we resume pciehp *before* we're finished resuming the Downstream
> > Port?  That sounds wrong.
> 
> I mean once we resume the downstream port (the bridge) we also resume
> "PCIe port services" including pciehp and only then descend to whatever
> device is physically connected to that port.

That order sounds right.  I guess I'd have to see more details about
what's happening with pciehp to understand this.  Do you happen to
have a trace (dmesg, backtrace, etc) of pciehp tearing things down?

> > > > > +static int pcie_get_downstream_delay(struct pci_bus *bus)
> > > > > +{
> > > > > +	struct pci_dev *pdev;
> > > > > +	int min_delay = 100;
> > > > > +	int max_delay = 0;
> > > > > +
> > > > > +	list_for_each_entry(pdev, &bus->devices, bus_list) {
> > > > > +		if (pdev->imm_ready)
> > > > > +			min_delay = 0;
> > > > > +		else if (pdev->d3cold_delay < min_delay)
> > > > > +			min_delay = pdev->d3cold_delay;
> > > > > +		if (pdev->d3cold_delay > max_delay)
> > > > > +			max_delay = pdev->d3cold_delay;
> > > > > +	}
> > > > > +
> > > > > +	return max(min_delay, max_delay);
> > > > > +}
> > 
> > > > > + */
> > > > > +void pcie_wait_downstream_accessible(struct pci_dev *pdev)
> > 
> > > > Do we need to observe the Trhfa requirements for Conventional PCI and
> > > > PCI-X devices here?  If we don't do it here, where is it observed?
> > > 
> > > We probably should but I intended this to be PCIe specific since there
> > > we have issues. For conventional PCI/PCI-X things "seem" to work and we
> > > don't power manage those bridges anyway.
> > > 
> > > I'm not aware if Trhfa is handled in anywhere in the PCI stack
> > > currently.
> > 
> > I think we should make this agnostic of the Conventional/PCIe
> > difference if possible.  I assume we can tell if we're dealing with a
> > D3->D0 transition and we only add delays in that case.  If we don't
> > power manage Conventional PCI devices, I assume we won't see D3->D0
> > transitions for runtime resume so there won't be any harm.
> >
> > Making it PCIe-specific seems like it adds extra code ("dev-is-PCIe
> > checks") with no obvious reason for existence and an implicit
> > dependency on the fact that we only power manage PCIe devices.  If we
> > ever *did* support runtime power-management for Conventional PCI, we'd
> > trip over that implicit dependency and probably debug this issue
> > again.
> > 
> > But I guess it might slow down system resume for Conventional PCI
> > devices.  If we rely on delays in firmware, I wonder if there's
> > any point during resume where we could grab an early timestamp, then
> > take another timestamp here and deduce that we've already delayed the
> > difference?
> 
> That sounds rather complex, to be honest ;-)

Maybe so, I was just trying to brainstorm possibilities for making
sure we observe the delay requirements without slowing down resume.

For example, if we have several devices on the same bus, we shouldn't
have to do the delays serially; we should be able to take advantage of
the fact that the Trhfa period starts at the same time for all of
them.

> > > > > +	delay = pcie_get_downstream_delay(bus);
> > > > > +	if (!delay)
> > > > > +		return;
> > > > 
> > > > I'm not sold on the idea that this delay depends on what's *below* the
> > > > bridge.  We're using sec 6.6.1 to justify the delay, and that section
> > > > doesn't say anything about downstream devices.
> > > 
> > > 6.6.1 indeed talks about Downstream Ports and devices immediately below
> > > them.
> > 
> > Wait, I don't think we're reading this the same way.  6.6.1 mentions
> > Downstream Ports: "With a Downstream Port that does not support Link
> > speeds greater than 5.0 GT/s, software must wait a minimum of 100 ms
> > before sending a Configuration Request to the device immediately below
> > that Port."
> > 
> > This says we have to delay before sending a config request to a device
> > below a Downstream Port, but it doesn't say anything about the
> > characteristics of that device.  In particular, I don't think it says
> > the delay can be shortened if that device supports Immediate Readiness
> > or implements a readiness _DSM.
> 
> Well it says this:
> 
>   To allow components to perform internal initialization, system software
>   must wait a specified minimum period following the end of a Conventional
>   Reset of one or more devices before it is permitted to issue
>   Configuration Requests to those devices, unless Readiness Notifications
>   mechanisms are used
> 
> My understanding of the above (might be wrong) is that Readiness
> Notification can shorten the delay.

Yeeesss, but if we're talking about transitioning device X from
D3->D0, I think this is talking about config requests to device X,
not to something downstream from X.

And we have no support for Readiness Notifications, although maybe the
_DSM stuff qualifies as "a mechanism that supersedes FRS and/or DRS"
(as mentioned in 6.23).

If device X was in D3cold, don't we have to assume that devices
downstream from X may have been added/removed while X was in D3cold?

> > I don't think this delay has anything to do with devices downstream
> > from the Port.  I think this is about giving the Downstream Port time
> > to bring up the link.  That way config requests may fail because
> > there's no device below the Port or it's not ready yet, but they
> > shouldn't fail simply because the link isn't up yet.
> 
> My understanding (again might be wrong) is that there are two factors
> here. One is to get the link trained and the other is to give the
> downstream device some time to perform its internal initialization so
> that it is ready to answer config requests.

Yeah, maybe you're right.  The second item about waiting 100ms after
Link training completes (for Ports that support speeds greater than
5.0 GT/s) does suggest that the delay is related to the downstream
device, since the Link is already up before the 100ms delay.

Bjorn
Mika Westerberg Aug. 27, 2019, 9:25 a.m. UTC | #10
On Mon, Aug 26, 2019 at 05:05:02PM -0500, Bjorn Helgaas wrote:
> On Mon, Aug 26, 2019 at 05:42:42PM +0300, Mika Westerberg wrote:
> > On Mon, Aug 26, 2019 at 09:07:12AM -0500, Bjorn Helgaas wrote:
> > > On Mon, Aug 26, 2019 at 01:17:26PM +0300, Mika Westerberg wrote:
> > > > On Fri, Aug 23, 2019 at 09:12:54PM -0500, Bjorn Helgaas wrote:
> 
> > > > > But the "wait downstream" part seems a little too specific to be at
> > > > > the .resume_noirq and .runtime_resume level.
> > > > > 
> > > > > Do we descend the hierarchy and call .resume_noirq and .runtime_resume
> > > > > for the children of the bridge, too?
> > > > 
> > > > We do but at that time it is too late as we have already resumed pciehp
> > > > of the parent downstream port and it may have already started tearing
> > > > down the device stack below.
> > > > 
> > > > I'm open to any better ideas where this delay could be added, though :)
> > > 
> > > So we resume pciehp *before* we're finished resuming the Downstream
> > > Port?  That sounds wrong.
> > 
> > I mean once we resume the downstream port (the bridge) we also resume
> > "PCIe port services" including pciehp and only then descend to whatever
> > device is physically connected to that port.
> 
> That order sounds right.  I guess I'd have to see more details about
> what's happening with pciehp to understand this.  Do you happen to
> have a trace (dmesg, backtrace, etc) of pciehp tearing things down?

Here are the relevant parts from ICL. I'll send you the full dmesg as
a separate email. The topology is such that I have 2 Titan Ridge devices
connected in chain (each include PCIe switch + xHCI). I wait for the
whole hierarchy to enter D3cold:

[   50.485411] thunderbolt 0000:00:0d.3: power state changed by ACPI to D3cold

Here I plug in normal type-C memory stick to last Titan Ridge port which
wakes up the hierarchy:

[   63.404860] thunderbolt 0000:00:0d.3: power state changed by ACPI to D0
[   63.408558] thunderbolt 0000:00:0d.2: power state changed by ACPI to D0
[   63.512696] pcieport 0000:00:07.3: power state changed by ACPI to D0
[   63.512921] pcieport 0000:00:07.0: power state changed by ACPI to D0
[   63.512960] pcieport 0000:00:07.2: power state changed by ACPI to D0
[   63.519567] pcieport 0000:00:07.1: power state changed by ACPI to D0
[   63.524365] thunderbolt 0000:00:0d.3: PME# disabled
[   63.524379] thunderbolt 0000:00:0d.3: control channel starting...
[   63.524382] thunderbolt 0000:00:0d.3: starting TX ring 0
[   63.524388] thunderbolt 0000:00:0d.3: enabling interrupt at register 0x38200 bit 0 (0x0 -> 0x1)
[   63.524390] thunderbolt 0000:00:0d.3: starting RX ring 0
[   63.524396] thunderbolt 0000:00:0d.3: enabling interrupt at register 0x38200 bit 12 (0x1 -> 0x1001)
[   63.525777] thunderbolt 0000:00:0d.2: PME# disabled
[   63.592667] thunderbolt 0000:00:0d.3: ICM rtd3 veto=0x00000001
[   63.627014] pcieport 0000:00:07.3: restoring config space at offset 0x2c (was 0x60, writing 0x60)
[   63.627018] pcieport 0000:00:07.0: restoring config space at offset 0x2c (was 0x60, writing 0x60)
[   63.627026] pcieport 0000:00:07.0: restoring config space at offset 0x28 (was 0x60, writing 0x60)
[   63.627028] pcieport 0000:00:07.3: restoring config space at offset 0x28 (was 0x60, writing 0x60)
[   63.627034] pcieport 0000:00:07.3: restoring config space at offset 0x24 (was 0x7bf16001, writing 0x7bf16001)
[   63.627036] pcieport 0000:00:07.0: restoring config space at offset 0x24 (was 0x1bf10001, writing 0x1bf10001)
[   63.632439] pcieport 0000:00:07.0: PME# disabled
[   63.632465] pcieport 0000:00:07.2: restoring config space at offset 0x2c (was 0x60, writing 0x60)
[   63.632467] pcieport 0000:00:07.2: restoring config space at offset 0x28 (was 0x60, writing 0x60)
[   63.632468] pcieport 0000:00:07.2: restoring config space at offset 0x24 (was 0x5bf14001, writing 0x5bf14001)
[   63.634219] pcieport 0000:00:07.3: PME# disabled
[   63.634324] pcieport 0000:82:00.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
[   63.634333] pcieport 0000:82:00.0: restoring config space at offset 0x2c (was 0x0, writing 0x60)
[   63.634337] pcieport 0000:82:00.0: restoring config space at offset 0x28 (was 0x0, writing 0x60)
[   63.634341] pcieport 0000:82:00.0: restoring config space at offset 0x24 (was 0x10001, writing 0x7bf16001)
[   63.634344] pcieport 0000:82:00.0: restoring config space at offset 0x20 (was 0x0, writing 0x5c105000)
[   63.634348] pcieport 0000:82:00.0: restoring config space at offset 0x1c (was 0x101, writing 0xb171)
[   63.634352] pcieport 0000:82:00.0: restoring config space at offset 0x18 (was 0x0, writing 0xac8382)
[   63.634363] pcieport 0000:82:00.0: restoring config space at offset 0x4 (was 0x100000, writing 0x100007)
[   63.634852] pcieport 0000:00:07.1: restoring config space at offset 0x2c (was 0x60, writing 0x60)
[   63.634854] pcieport 0000:00:07.1: restoring config space at offset 0x28 (was 0x60, writing 0x60)
[   63.634856] pcieport 0000:00:07.1: restoring config space at offset 0x24 (was 0x3bf12001, writing 0x3bf12001)
[   63.636163] pcieport 0000:00:07.2: PME# disabled
[   63.636320] pcieport 0000:82:00.0: PME# disabled
[   63.636451] pcieport 0000:83:02.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
[   63.636473] pcieport 0000:83:02.0: restoring config space at offset 0x2c (was 0x0, writing 0x60)
[   63.636480] pcieport 0000:83:02.0: restoring config space at offset 0x28 (was 0x0, writing 0x60)
[   63.636488] pcieport 0000:83:02.0: restoring config space at offset 0x24 (was 0x10001, writing 0x60116011)
[   63.636495] pcieport 0000:83:02.0: restoring config space at offset 0x20 (was 0x0, writing 0x50105010)
[   63.636502] pcieport 0000:83:02.0: restoring config space at offset 0x1c (was 0x101, writing 0x8181)
[   63.636510] pcieport 0000:83:02.0: restoring config space at offset 0x18 (was 0x0, writing 0x858583)
[   63.636529] pcieport 0000:83:02.0: restoring config space at offset 0x4 (was 0x100000, writing 0x100407)
[   63.636575] pcieport 0000:83:04.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
[   63.636597] pcieport 0000:83:04.0: restoring config space at offset 0x2c (was 0x0, writing 0x60)
[   63.636603] pcieport 0000:83:04.0: restoring config space at offset 0x28 (was 0x0, writing 0x60)
[   63.636609] pcieport 0000:83:04.0: restoring config space at offset 0x24 (was 0x10001, writing 0x7bf16021)
[   63.636615] pcieport 0000:83:04.0: restoring config space at offset 0x20 (was 0x0, writing 0x5c105020)
[   63.636621] pcieport 0000:83:04.0: restoring config space at offset 0x1c (was 0x101, writing 0xb191)
[   63.636627] pcieport 0000:83:04.0: restoring config space at offset 0x18 (was 0x0, writing 0xac8683)
[   63.636641] pcieport 0000:83:04.0: restoring config space at offset 0x4 (was 0x100000, writing 0x100407)
[   63.639915] pcieport 0000:83:01.0: restoring config space at offset 0x3c (was 0x1ff, writing 0x201ff)
[   63.639926] pcieport 0000:83:01.0: restoring config space at offset 0x2c (was 0x0, writing 0x60)
[   63.639931] pcieport 0000:83:01.0: restoring config space at offset 0x28 (was 0x0, writing 0x60)
[   63.639936] pcieport 0000:83:01.0: restoring config space at offset 0x24 (was 0x10001, writing 0x60016001)
[   63.639941] pcieport 0000:83:01.0: restoring config space at offset 0x20 (was 0x0, writing 0x50005000)
[   63.639946] pcieport 0000:83:01.0: restoring config space at offset 0x1c (was 0x101, writing 0x7171)
[   63.639951] pcieport 0000:83:01.0: restoring config space at offset 0x18 (was 0x0, writing 0x848483)
[   63.639963] pcieport 0000:83:01.0: restoring config space at offset 0x4 (was 0x100000, writing 0x100407)
[   63.640473] pcieport 0000:00:07.1: PME# disabled
[   63.640515] pcieport 0000:83:04.0: PME# disabled
[   63.640537] pcieport 0000:83:04.0: pciehp: Slot(4): Card not present

Here pciehp notices the card is not present and starts tearing down the
devices below.

[   63.640539] pcieport 0000:87:04.0: PME# disabled
[   63.648104] pcieport 0000:83:04.0: pciehp: pciehp_unconfigure_device: domain:bus:dev = 0000:86:00
[   63.648105] pcieport 0000:86:00.0: Refused to change power state, currently in D3
[   63.656858] pcieport 0000:86:00.0: restoring config space at offset 0x3c (was 0xffffffff, writing 0x201ff)
[   63.656858] pcieport 0000:86:00.0: restoring config space at offset 0x38 (was 0xffffffff, writing 0x0)
[   63.656859] pcieport 0000:86:00.0: restoring config space at offset 0x34 (was 0xffffffff, writing 0x80)
[   63.656859] pcieport 0000:86:00.0: restoring config space at offset 0x30 (was 0xffffffff, writing 0x0)
[   63.656860] pcieport 0000:86:00.0: restoring config space at offset 0x2c (was 0xffffffff, writing 0x60)
[   63.656860] pcieport 0000:86:00.0: restoring config space at offset 0x28 (was 0xffffffff, writing 0x60)
[   63.656861] pcieport 0000:86:00.0: restoring config space at offset 0x24 (was 0xffffffff, writing 0x7bf16021)
[   63.656862] pcieport 0000:86:00.0: restoring config space at offset 0x20 (was 0xffffffff, writing 0x5c105020)
[   63.656862] pcieport 0000:86:00.0: restoring config space at offset 0x1c (was 0xffffffff, writing 0xb191)
[   63.656863] pcieport 0000:86:00.0: restoring config space at offset 0x18 (was 0xffffffff, writing 0xac8786)
[   63.656863] pcieport 0000:86:00.0: restoring config space at offset 0x14 (was 0xffffffff, writing 0x0)
[   63.656864] pcieport 0000:86:00.0: restoring config space at offset 0x10 (was 0xffffffff, writing 0x0)
[   63.656864] pcieport 0000:86:00.0: restoring config space at offset 0xc (was 0xffffffff, writing 0x10000)
[   63.656865] pcieport 0000:86:00.0: restoring config space at offset 0x8 (was 0xffffffff, writing 0x6040006)
[   63.656865] pcieport 0000:86:00.0: restoring config space at offset 0x4 (was 0xffffffff, writing 0x100007)
[   63.656866] pcieport 0000:86:00.0: restoring config space at offset 0x0 (was 0xffffffff, writing 0x15ef8086)
[   63.656883] pcieport 0000:86:00.0: PME# disabled
[   63.656884] pcieport 0000:87:04.0: Refused to change power state, currently in D3

There is also another case which does not involve pciehp. The xHCI issue
Kai-Heng reported. In that case PCI core tries to access xHCI too soon
and fails to resume it:

[   74.100492] pcieport 0000:04:02.0: restoring config space at offset 0xc (was 0x10000, writing 0x10020)
[   74.100498] pcieport 0000:04:02.0: restoring config space at offset 0x4 (was 0x100000, writing 0x100406)
[   74.100545] pcieport 0000:04:02.0: PME# disabled
[   74.100554] xhci_hcd 0000:39:00.0: Refused to change power state, currently in D3
[   74.102769] xhci_hcd 0000:39:00.0: restoring config space at offset 0x3c (was 0xffffffff, writing 0x1ff)
[   74.102774] xhci_hcd 0000:39:00.0: restoring config space at offset 0x38 (was 0xffffffff, writing 0x0)
[   74.102778] xhci_hcd 0000:39:00.0: restoring config space at offset 0x34 (was 0xffffffff, writing 0x80)
[   74.102782] xhci_hcd 0000:39:00.0: restoring config space at offset 0x30 (was 0xffffffff, writing 0x0)
[   74.102787] xhci_hcd 0000:39:00.0: restoring config space at offset 0x2c (was 0xffffffff, writing 0x9261028)

the full dmesg is here:

  https://bugzilla.kernel.org/attachment.cgi?id=284427

> > > > > > +static int pcie_get_downstream_delay(struct pci_bus *bus)
> > > > > > +{
> > > > > > +	struct pci_dev *pdev;
> > > > > > +	int min_delay = 100;
> > > > > > +	int max_delay = 0;
> > > > > > +
> > > > > > +	list_for_each_entry(pdev, &bus->devices, bus_list) {
> > > > > > +		if (pdev->imm_ready)
> > > > > > +			min_delay = 0;
> > > > > > +		else if (pdev->d3cold_delay < min_delay)
> > > > > > +			min_delay = pdev->d3cold_delay;
> > > > > > +		if (pdev->d3cold_delay > max_delay)
> > > > > > +			max_delay = pdev->d3cold_delay;
> > > > > > +	}
> > > > > > +
> > > > > > +	return max(min_delay, max_delay);
> > > > > > +}
> > > 
> > > > > > + */
> > > > > > +void pcie_wait_downstream_accessible(struct pci_dev *pdev)
> > > 
> > > > > Do we need to observe the Trhfa requirements for Conventional PCI and
> > > > > PCI-X devices here?  If we don't do it here, where is it observed?
> > > > 
> > > > We probably should but I intended this to be PCIe specific since there
> > > > we have issues. For conventional PCI/PCI-X things "seem" to work and we
> > > > don't power manage those bridges anyway.
> > > > 
> > > > I'm not aware if Trhfa is handled in anywhere in the PCI stack
> > > > currently.
> > > 
> > > I think we should make this agnostic of the Conventional/PCIe
> > > difference if possible.  I assume we can tell if we're dealing with a
> > > D3->D0 transition and we only add delays in that case.  If we don't
> > > power manage Conventional PCI devices, I assume we won't see D3->D0
> > > transitions for runtime resume so there won't be any harm.
> > >
> > > Making it PCIe-specific seems like it adds extra code ("dev-is-PCIe
> > > checks") with no obvious reason for existence and an implicit
> > > dependency on the fact that we only power manage PCIe devices.  If we
> > > ever *did* support runtime power-management for Conventional PCI, we'd
> > > trip over that implicit dependency and probably debug this issue
> > > again.
> > > 
> > > But I guess it might slow down system resume for Conventional PCI
> > > devices.  If we rely on delays in firmware, I wonder if there's
> > > any point during resume where we could grab an early timestamp, then
> > > take another timestamp here and deduce that we've already delayed the
> > > difference?
> > 
> > That sounds rather complex, to be honest ;-)
> 
> Maybe so, I was just trying to brainstorm possibilities for making
> sure we observe the delay requirements without slowing down resume.
> 
> For example, if we have several devices on the same bus, we shouldn't
> have to do the delays serially; we should be able to take advantage of
> the fact that the Trhfa period starts at the same time for all of
> them.

Also we do async suspend these days for PCI devices so I think sibling
devices are already resumed concurrently.

> > > > > > +	delay = pcie_get_downstream_delay(bus);
> > > > > > +	if (!delay)
> > > > > > +		return;
> > > > > 
> > > > > I'm not sold on the idea that this delay depends on what's *below* the
> > > > > bridge.  We're using sec 6.6.1 to justify the delay, and that section
> > > > > doesn't say anything about downstream devices.
> > > > 
> > > > 6.6.1 indeed talks about Downstream Ports and devices immediately below
> > > > them.
> > > 
> > > Wait, I don't think we're reading this the same way.  6.6.1 mentions
> > > Downstream Ports: "With a Downstream Port that does not support Link
> > > speeds greater than 5.0 GT/s, software must wait a minimum of 100 ms
> > > before sending a Configuration Request to the device immediately below
> > > that Port."
> > > 
> > > This says we have to delay before sending a config request to a device
> > > below a Downstream Port, but it doesn't say anything about the
> > > characteristics of that device.  In particular, I don't think it says
> > > the delay can be shortened if that device supports Immediate Readiness
> > > or implements a readiness _DSM.
> > 
> > Well it says this:
> > 
> >   To allow components to perform internal initialization, system software
> >   must wait a specified minimum period following the end of a Conventional
> >   Reset of one or more devices before it is permitted to issue
> >   Configuration Requests to those devices, unless Readiness Notifications
> >   mechanisms are used
> > 
> > My understanding of the above (might be wrong) is that Readiness
> > Notification can shorten the delay.
> 
> Yeeesss, but if we're talking about transitioning device X from
> D3->D0, I think this is talking about config requests to device X,
> not to something downstream from X.
> 
> And we have no support for Readiness Notifications, although maybe the
> _DSM stuff qualifies as "a mechanism that supersedes FRS and/or DRS"
> (as mentioned in 6.23).
> 
> If device X was in D3cold, don't we have to assume that devices
> downstream from X may have been added/removed while X was in D3cold?

Yes, I think so.
Mika Westerberg Sept. 19, 2019, 6:57 a.m. UTC | #11
Hi Bjorn,

On Tue, Aug 27, 2019 at 12:25:47PM +0300, Mika Westerberg wrote:
> Here are the relevant parts from ICL. I'll send you the full dmesg as
> a separate email. The topology is such that I have 2 Titan Ridge devices
> connected in chain (each include PCIe switch + xHCI). I wait for the
> whole hierarchy to enter D3cold:

Did you had time to check the log I sent you?

Also how should we proceed with this patch? I can update it according to
comments and send a new version after v5.4-rc1 is released. Of course
unless you have better way we can fix this issue.

Thanks!
diff mbox series

Patch

diff --git a/drivers/pci/pci-driver.c b/drivers/pci/pci-driver.c
index a8124e47bf6e..9aec78ed8907 100644
--- a/drivers/pci/pci-driver.c
+++ b/drivers/pci/pci-driver.c
@@ -918,6 +918,7 @@  static int pci_pm_resume_noirq(struct device *dev)
 {
 	struct pci_dev *pci_dev = to_pci_dev(dev);
 	struct device_driver *drv = dev->driver;
+	pci_power_t state = pci_dev->current_state;
 	int error = 0;
 
 	if (dev_pm_may_skip_resume(dev))
@@ -947,6 +948,15 @@  static int pci_pm_resume_noirq(struct device *dev)
 
 	pcie_pme_root_status_cleanup(pci_dev);
 
+	/*
+	 * If resume involves firmware assume it takes care of any delays
+	 * for now. For suspend-to-idle case we need to do that here before
+	 * resuming PCIe port services to keep pciehp from tearing down the
+	 * downstream devices too early.
+	 */
+	if (state == PCI_D3cold && pm_suspend_no_platform())
+		pcie_wait_downstream_accessible(pci_dev);
+
 	if (drv && drv->pm && drv->pm->resume_noirq)
 		error = drv->pm->resume_noirq(dev);
 
@@ -1329,6 +1339,7 @@  static int pci_pm_runtime_resume(struct device *dev)
 	int rc = 0;
 	struct pci_dev *pci_dev = to_pci_dev(dev);
 	const struct dev_pm_ops *pm = dev->driver ? dev->driver->pm : NULL;
+	pci_power_t state = pci_dev->current_state;
 
 	/*
 	 * Restoring config space is necessary even if the device is not bound
@@ -1344,6 +1355,14 @@  static int pci_pm_runtime_resume(struct device *dev)
 	pci_enable_wake(pci_dev, PCI_D0, false);
 	pci_fixup_device(pci_fixup_resume, pci_dev);
 
+	/*
+	 * If the hierarcy went into D3cold wait for the link to be
+	 * reactivated before resuming PCIe port services to keep pciehp
+	 * from tearing down the downstream devices too early.
+	 */
+	if (state == PCI_D3cold)
+		pcie_wait_downstream_accessible(pci_dev);
+
 	if (pm && pm->runtime_resume)
 		rc = pm->runtime_resume(dev);
 
diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
index 1b27b5af3d55..9ac50710f1d4 100644
--- a/drivers/pci/pci.c
+++ b/drivers/pci/pci.c
@@ -1025,15 +1025,11 @@  static void __pci_start_power_transition(struct pci_dev *dev, pci_power_t state)
 	if (state == PCI_D0) {
 		pci_platform_power_transition(dev, PCI_D0);
 		/*
-		 * Mandatory power management transition delays, see
-		 * PCI Express Base Specification Revision 2.0 Section
-		 * 6.6.1: Conventional Reset.  Do not delay for
-		 * devices powered on/off by corresponding bridge,
-		 * because have already delayed for the bridge.
+		 * Mandatory power management transition delays are handled
+		 * in pci_pm_runtime_resume() of the corresponding
+		 * downstream/root port.
 		 */
 		if (dev->runtime_d3cold) {
-			if (dev->d3cold_delay && !dev->imm_ready)
-				msleep(dev->d3cold_delay);
 			/*
 			 * When powering on a bridge from D3cold, the
 			 * whole hierarchy may be powered on into
@@ -4607,14 +4603,17 @@  static int pci_pm_reset(struct pci_dev *dev, int probe)
 
 	return pci_dev_wait(dev, "PM D3->D0", PCIE_RESET_READY_POLL_MS);
 }
+
 /**
- * pcie_wait_for_link - Wait until link is active or inactive
+ * pcie_wait_for_link_delay - Wait until link is active or inactive
  * @pdev: Bridge device
  * @active: waiting for active or inactive?
+ * @delay: Delay to wait after link has become active (in ms)
  *
  * Use this to wait till link becomes active or inactive.
  */
-bool pcie_wait_for_link(struct pci_dev *pdev, bool active)
+static bool pcie_wait_for_link_delay(struct pci_dev *pdev, bool active,
+				     int delay)
 {
 	int timeout = 1000;
 	bool ret;
@@ -4651,13 +4650,121 @@  bool pcie_wait_for_link(struct pci_dev *pdev, bool active)
 		timeout -= 10;
 	}
 	if (active && ret)
-		msleep(100);
+		msleep(delay);
 	else if (ret != active)
 		pci_info(pdev, "Data Link Layer Link Active not %s in 1000 msec\n",
 			active ? "set" : "cleared");
 	return ret == active;
 }
 
+/**
+ * pcie_wait_for_link - Wait until link is active or inactive
+ * @pdev: Bridge device
+ * @active: waiting for active or inactive?
+ *
+ * Use this to wait till link becomes active or inactive.
+ */
+bool pcie_wait_for_link(struct pci_dev *pdev, bool active)
+{
+	return pcie_wait_for_link_delay(pdev, active, 100);
+}
+
+static int pcie_get_downstream_delay(struct pci_bus *bus)
+{
+	struct pci_dev *pdev;
+	int min_delay = 100;
+	int max_delay = 0;
+
+	list_for_each_entry(pdev, &bus->devices, bus_list) {
+		if (pdev->imm_ready)
+			min_delay = 0;
+		else if (pdev->d3cold_delay < min_delay)
+			min_delay = pdev->d3cold_delay;
+		if (pdev->d3cold_delay > max_delay)
+			max_delay = pdev->d3cold_delay;
+	}
+
+	return max(min_delay, max_delay);
+}
+
+/**
+ * pcie_wait_downstream_accessible - Wait downstream device to be accessible
+ * @pdev: PCIe port whose downstream device is waited
+ *
+ * Handle delays according to PCIe 4.0 section 6.6.1 before configuration
+ * access to the downstream device is permitted. If the port does not have
+ * any devices connected, does nothing.
+ *
+ * This is needed if the hierarchy below @pdev went through reset (after
+ * exit from D3cold back to D0uninitialized).
+ */
+void pcie_wait_downstream_accessible(struct pci_dev *pdev)
+{
+	struct pci_dev *child;
+	struct pci_bus *bus;
+	int delay;
+
+	if (pci_pcie_type(pdev) != PCI_EXP_TYPE_ROOT_PORT &&
+	    pci_pcie_type(pdev) != PCI_EXP_TYPE_DOWNSTREAM)
+		return;
+
+	if (pci_dev_is_disconnected(pdev))
+		return;
+
+	if (!pdev->bridge_d3)
+		return;
+
+	bus = pdev->subordinate;
+	if (!bus)
+		return;
+
+	child = list_first_entry_or_null(&bus->devices, struct pci_dev,
+					 bus_list);
+	if (!child)
+		return;
+
+	delay = pcie_get_downstream_delay(bus);
+	if (!delay)
+		return;
+
+	/*
+	 * If downstream port does not support speeds greater than 5 GT/s
+	 * need to wait minimum 100ms. For higher speeds (gen3) we need to
+	 * wait first for the data link layer to become active.
+	 *
+	 * However, 100ms is the minimum and the spec says that the
+	 * software must allow at least 1s before it can determine that the
+	 * device that did not respond is a broken device. Also there is
+	 * evidence that the 100ms is not always enough so what we do here
+	 * is that we wait for the minimum 100ms (possibly after link
+	 * training completes) and then probe for the device presence once.
+	 * If we still don't get response we wait for another 100ms just to
+	 * give it some additional time to complete its initialization.
+	 */
+	if (pcie_get_speed_cap(pdev) <= PCIE_SPEED_5_0GT) {
+		dev_dbg(&pdev->dev, "waiting downstream link for %d ms\n",
+			delay);
+		msleep(delay);
+	} else {
+		dev_dbg(&pdev->dev,
+			"waiting downstream link for %d ms after activation\n",
+			delay);
+		if (!pcie_wait_for_link_delay(pdev, true, delay)) {
+			/*
+			 * If the link did not train, no need to wait
+			 * further the device is probably gone.
+			 */
+			return;
+		}
+	}
+
+	if (!pci_device_is_present(child)) {
+		dev_dbg(&child->dev,
+			"waiting for additional 100 ms for the device to become accessible\n");
+		msleep(100);
+	}
+}
+
 void pci_reset_secondary_bus(struct pci_dev *dev)
 {
 	u16 ctrl;
diff --git a/drivers/pci/pci.h b/drivers/pci/pci.h
index d22d1b807701..9a83fcf612ca 100644
--- a/drivers/pci/pci.h
+++ b/drivers/pci/pci.h
@@ -498,6 +498,7 @@  void pcie_do_recovery(struct pci_dev *dev, enum pci_channel_state state,
 		      u32 service);
 
 bool pcie_wait_for_link(struct pci_dev *pdev, bool active);
+void pcie_wait_downstream_accessible(struct pci_dev *pdev);
 #ifdef CONFIG_PCIEASPM
 void pcie_aspm_init_link_state(struct pci_dev *pdev);
 void pcie_aspm_exit_link_state(struct pci_dev *pdev);