[PATCH v5 0/2] nvme-pci: add AMD PCIe quirk for NVMe simple suspend/resume

[PATCH v5 0/2] nvme-pci: add AMD PCIe quirk for NVMe simple suspend/resume

[Prike Liang]

Those patch series can handle NVMe can't suspend to D3 during s2idle
entry on some AMD platform. In this case, can be settld by assigning and
passing a PCIe bus flag to the armed device which need NVMe shutdown opt
in s2idle suspend and then use PCIe power setting to put the NVMe device
to D3.

Prike Liang (2):
  PCI: add AMD PCIe quirk for nvme shutdown opt
  nvme-pci: add AMD PCIe quirk for simple suspend/resume

 drivers/nvme/host/pci.c | 2 ++
 drivers/pci/probe.c     | 5 ++++-
 drivers/pci/quirks.c    | 7 +++++++
 include/linux/pci.h     | 2 ++
 4 files changed, 15 insertions(+), 1 deletion(-)

-- 
2.7.4


_______________________________________________
Linux-nvme mailing list
Linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme

[PATCH v5 1/2] PCI: add AMD PCIe quirk for nvme shutdown opt

[Prike Liang]

In the NVMe controller default suspend-resume seems only save/restore the
NVMe link state by APST opt and the NVMe remains in D0 during this time.
Then the NVMe device will be shutdown by SMU firmware in the s2idle entry
and then will lost the NVMe power context during s2idle resume.Finally,
the NVMe command queue request will be processed abnormally and result
in access timeout.This issue can be settled by using PCIe power set with
simple suspend-resume process path instead of APST get/set opt.

In this patch prepare a PCIe RC bus flag to identify the platform whether
need the quirk.

Cc: <stable@vger.kernel.org> # 5.10+
Signed-off-by: Prike Liang <Prike.Liang@amd.com>
Signed-off-by: Shyam Sundar S K <Shyam-sundar.S-k@amd.com>
[ck: split patches for nvme and pcie]
Signed-off-by: Chaitanya Kulkarni <chaitanya.kulkarni@wdc.com>
Suggested-by: Keith Busch <kbusch@kernel.org>
Acked-by: Keith Busch <kbusch@kernel.org>
---
Changes in v2:
Fix the patch format and check chip root complex DID instead of PCIe RP
to avoid the storage device plugged in internal PCIe RP by USB adaptor.

Changes in v3:
According to Christoph Hellwig do NVME PCIe related identify opt better in
PCIe quirk driver rather than in NVME module.

Changes in v4:
Split the fix to PCIe and NVMe part and then call the pci_dev_put() put
the device reference count and finally refine the commit info.

Changes in v5:
According to Christoph Hellwig and Keith Busch better use a passthrough device(bus)
gloable flag to identify the NVMe shutdown opt rather than look up the device BDF.
---
 drivers/pci/probe.c  | 5 ++++-
 drivers/pci/quirks.c | 7 +++++++
 include/linux/pci.h  | 2 ++
 3 files changed, 13 insertions(+), 1 deletion(-)

diff --git a/drivers/pci/probe.c b/drivers/pci/probe.c
index 953f15a..34ba691e 100644
--- a/drivers/pci/probe.c
+++ b/drivers/pci/probe.c
@@ -558,10 +558,13 @@ static struct pci_bus *pci_alloc_bus(struct pci_bus *parent)
 	INIT_LIST_HEAD(&b->resources);
 	b->max_bus_speed = PCI_SPEED_UNKNOWN;
 	b->cur_bus_speed = PCI_SPEED_UNKNOWN;
+	if (parent) {
 #ifdef CONFIG_PCI_DOMAINS_GENERIC
-	if (parent)
 		b->domain_nr = parent->domain_nr;
 #endif
+		if (parent->bus_flags & PCI_BUS_FLAGS_DISABLE_ON_S2I)
+			b->bus_flags |= PCI_BUS_FLAGS_DISABLE_ON_S2I;
+	}
 	return b;
 }
 
diff --git a/drivers/pci/quirks.c b/drivers/pci/quirks.c
index 653660e3..7c4bb8e 100644
--- a/drivers/pci/quirks.c
+++ b/drivers/pci/quirks.c
@@ -312,6 +312,13 @@ static void quirk_nopciamd(struct pci_dev *dev)
 }
 DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_AMD,	PCI_DEVICE_ID_AMD_8151_0,	quirk_nopciamd);
 
+static void quirk_amd_s2i_fixup(struct pci_dev *dev)
+{
+	dev->bus->bus_flags |= PCI_BUS_FLAGS_DISABLE_ON_S2I;
+	pci_info(dev, "AMD simple suspend opt enabled\n");
+}
+DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_AMD, 0x1630, quirk_amd_s2i_fixup);
+
 /* Triton requires workarounds to be used by the drivers */
 static void quirk_triton(struct pci_dev *dev)
 {
diff --git a/include/linux/pci.h b/include/linux/pci.h
index 53f4904..dc65219 100644
--- a/include/linux/pci.h
+++ b/include/linux/pci.h
@@ -240,6 +240,8 @@ enum pci_bus_flags {
 	PCI_BUS_FLAGS_NO_MMRBC	= (__force pci_bus_flags_t) 2,
 	PCI_BUS_FLAGS_NO_AERSID	= (__force pci_bus_flags_t) 4,
 	PCI_BUS_FLAGS_NO_EXTCFG	= (__force pci_bus_flags_t) 8,
+	/* Driver must pci_disable_device() for suspend-to-idle */
+	PCI_BUS_FLAGS_DISABLE_ON_S2I	= (__force pci_bus_flags_t) 16,
 };
 
 /* Values from Link Status register, PCIe r3.1, sec 7.8.8 */
-- 
2.7.4


_______________________________________________
Linux-nvme mailing list
Linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme

Re: [PATCH v5 1/2] PCI: add AMD PCIe quirk for nvme shutdown opt

[Bjorn Helgaas]

[+cc Rafael (probably nothing of interest to you), linux-pm]

On Tue, May 18, 2021 at 10:24:34AM +0800, Prike Liang wrote:
> In the NVMe controller default suspend-resume seems only save/restore the
> NVMe link state by APST opt and the NVMe remains in D0 during this time.
> Then the NVMe device will be shutdown by SMU firmware in the s2idle entry
> and then will lost the NVMe power context during s2idle resume.Finally,
> the NVMe command queue request will be processed abnormally and result
> in access timeout.This issue can be settled by using PCIe power set with
> simple suspend-resume process path instead of APST get/set opt.

I can't parse the paragraph above, sorry.  I'm sure this means
something to NVMe developers, but since you're adding this to the PCI
core, not the NVMe core, it needs to be intelligible to ordinary PCI
folks.

For example, since you only use this flag in the NVMe driver, you
should explain why the PCI core needs to keep track of the flag for
you.  Normally I would assume the driver could figure this out in its
.probe() function.

Quirks are usually used to work around a defect in a device.  What's
the defect in this case?  Ideally we can point to a section of the
PCIe spec with a requirement that the device violates.

What does "opt" mean?

What is SMU firmware?  Why is it relevant?

Is this a problem only with s2idle?  Why or why not?

The quirk applies to [1022:1630].  An lspci I found on the web says
this is a "00:00.0 Host bridge: AMD Renoir Root Complex" device.  So
it looks like this will result in PCI_BUS_FLAGS_DISABLE_ON_S2I being
set for every PCI bus in the entire system.  But the description talks
about an issue specifically with NVMe.

Is there a defect in this AMD PCIe controller that affects all
devices?

> In this patch prepare a PCIe RC bus flag to identify the platform whether
> need the quirk.
> 
> Cc: <stable@vger.kernel.org> # 5.10+
> Signed-off-by: Prike Liang <Prike.Liang@amd.com>
> Signed-off-by: Shyam Sundar S K <Shyam-sundar.S-k@amd.com>
> [ck: split patches for nvme and pcie]
> Signed-off-by: Chaitanya Kulkarni <chaitanya.kulkarni@wdc.com>
> Suggested-by: Keith Busch <kbusch@kernel.org>
> Acked-by: Keith Busch <kbusch@kernel.org>
> ---
> Changes in v2:
> Fix the patch format and check chip root complex DID instead of PCIe RP
> to avoid the storage device plugged in internal PCIe RP by USB adaptor.
> 
> Changes in v3:
> According to Christoph Hellwig do NVME PCIe related identify opt better in
> PCIe quirk driver rather than in NVME module.
> 
> Changes in v4:
> Split the fix to PCIe and NVMe part and then call the pci_dev_put() put
> the device reference count and finally refine the commit info.
> 
> Changes in v5:
> According to Christoph Hellwig and Keith Busch better use a passthrough device(bus)
> gloable flag to identify the NVMe shutdown opt rather than look up the device BDF.
> ---
>  drivers/pci/probe.c  | 5 ++++-
>  drivers/pci/quirks.c | 7 +++++++
>  include/linux/pci.h  | 2 ++
>  3 files changed, 13 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/pci/probe.c b/drivers/pci/probe.c
> index 953f15a..34ba691e 100644
> --- a/drivers/pci/probe.c
> +++ b/drivers/pci/probe.c
> @@ -558,10 +558,13 @@ static struct pci_bus *pci_alloc_bus(struct pci_bus *parent)
>  	INIT_LIST_HEAD(&b->resources);
>  	b->max_bus_speed = PCI_SPEED_UNKNOWN;
>  	b->cur_bus_speed = PCI_SPEED_UNKNOWN;
> +	if (parent) {
>  #ifdef CONFIG_PCI_DOMAINS_GENERIC
> -	if (parent)
>  		b->domain_nr = parent->domain_nr;
>  #endif
> +		if (parent->bus_flags & PCI_BUS_FLAGS_DISABLE_ON_S2I)
> +			b->bus_flags |= PCI_BUS_FLAGS_DISABLE_ON_S2I;
> +	}
>  	return b;
>  }
>  
> diff --git a/drivers/pci/quirks.c b/drivers/pci/quirks.c
> index 653660e3..7c4bb8e 100644
> --- a/drivers/pci/quirks.c
> +++ b/drivers/pci/quirks.c
> @@ -312,6 +312,13 @@ static void quirk_nopciamd(struct pci_dev *dev)
>  }
>  DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_AMD,	PCI_DEVICE_ID_AMD_8151_0,	quirk_nopciamd);
>  
> +static void quirk_amd_s2i_fixup(struct pci_dev *dev)
> +{
> +	dev->bus->bus_flags |= PCI_BUS_FLAGS_DISABLE_ON_S2I;
> +	pci_info(dev, "AMD simple suspend opt enabled\n");
> +}
> +DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_AMD, 0x1630, quirk_amd_s2i_fixup);
> +
>  /* Triton requires workarounds to be used by the drivers */
>  static void quirk_triton(struct pci_dev *dev)
>  {
> diff --git a/include/linux/pci.h b/include/linux/pci.h
> index 53f4904..dc65219 100644
> --- a/include/linux/pci.h
> +++ b/include/linux/pci.h
> @@ -240,6 +240,8 @@ enum pci_bus_flags {
>  	PCI_BUS_FLAGS_NO_MMRBC	= (__force pci_bus_flags_t) 2,
>  	PCI_BUS_FLAGS_NO_AERSID	= (__force pci_bus_flags_t) 4,
>  	PCI_BUS_FLAGS_NO_EXTCFG	= (__force pci_bus_flags_t) 8,
> +	/* Driver must pci_disable_device() for suspend-to-idle */
> +	PCI_BUS_FLAGS_DISABLE_ON_S2I	= (__force pci_bus_flags_t) 16,
>  };
>  
>  /* Values from Link Status register, PCIe r3.1, sec 7.8.8 */
> -- 
> 2.7.4
> 

_______________________________________________
Linux-nvme mailing list
Linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme

RE: [PATCH v5 1/2] PCI: add AMD PCIe quirk for nvme shutdown opt

[Liang, Prike]

[Public]

> From: Bjorn Helgaas <helgaas@kernel.org>
> Sent: Thursday, May 20, 2021 5:34 AM
> To: Liang, Prike <Prike.Liang@amd.com>
> Cc: linux-pci@vger.kernel.org; kbusch@kernel.org; axboe@fb.com;
> hch@lst.de; sagi@grimberg.me; linux-nvme@lists.infradead.org; Deucher,
> Alexander <Alexander.Deucher@amd.com>; stable@vger.kernel.org; S-k,
> Shyam-sundar <Shyam-sundar.S-k@amd.com>; Chaitanya Kulkarni
> <chaitanya.kulkarni@wdc.com>; Rafael J. Wysocki <rjw@rjwysocki.net>;
> linux-pm@vger.kernel.org
> Subject: Re: [PATCH v5 1/2] PCI: add AMD PCIe quirk for nvme shutdown opt
>
> [+cc Rafael (probably nothing of interest to you), linux-pm]
>
> On Tue, May 18, 2021 at 10:24:34AM +0800, Prike Liang wrote:
> > In the NVMe controller default suspend-resume seems only save/restore
> > the NVMe link state by APST opt and the NVMe remains in D0 during this
> time.
> > Then the NVMe device will be shutdown by SMU firmware in the s2idle
> > entry and then will lost the NVMe power context during s2idle
> > resume.Finally, the NVMe command queue request will be processed
> > abnormally and result in access timeout.This issue can be settled by
> > using PCIe power set with simple suspend-resume process path instead of
> APST get/set opt.
>
> I can't parse the paragraph above, sorry.  I'm sure this means something to
> NVMe developers, but since you're adding this to the PCI core, not the NVMe
> core, it needs to be intelligible to ordinary PCI folks.
>
[Prike]  I'm sorry to make confusion here. Those patches addressed a s2idle resume broken problem
that the NVMe driver's default suspend-resume policy of using NVMe APST during suspend-to-idle
prevents the PCI root port from going to D3.

> For example, since you only use this flag in the NVMe driver, you should
> explain why the PCI core needs to keep track of the flag for you.  Normally I
> would assume the driver could figure this out in its
> .probe() function.
>
[Prike] Yeah, we can assign the quirk flag in the .probe function or add it in nvme_id_table and this also
the primary solution we tried out. However, that seems not possible to enumerate every uncertain NVMe device then assign quirk flag to them. In this case, in order to handle various NVMe device we can use the root complex device ID to identify the question platform.

> Quirks are usually used to work around a defect in a device.  What's the
> defect in this case?  Ideally we can point to a section of the PCIe spec with a
> requirement that the device violates.
>
[Prike] In this case the quirk is only used to identify the question platform which requires the NVMe
device go to D3 in the s2idle suspend.
> What does "opt" mean?
>
[Prike] I'm also not dedicate working on the NVMe driver, but from the software perspective the APST
opt is used for handling the power state S&R without PCI interfering during s2idle legacy suspend-resume.

> What is SMU firmware?  Why is it relevant?
>
[Prike] SMU firmware is a proprietary micro component which responsible for device power management. Without the quirk flag, NVMe device will not enter D3 during s2idle suspend then SMU firmware will shut down the NVMe device, unfortunately since NVMe is a third-party device the SMU firmware only restore NVMe root port power state during s2ilde wake up process. Eventually, the NVMe device power state will be lost when back to OS s2idle resume  and then result in NVMe command request failed.

> Is this a problem only with s2idle?  Why or why not?
>
[Prike] Yeah, this issue is only found in the s2idle scenario, and that's because s2idle will check whether
each device will enter its own minimum power level defined in the LPI constrains table.

> The quirk applies to [1022:1630].  An lspci I found on the web says this is a
> "00:00.0 Host bridge: AMD Renoir Root Complex" device.  So it looks like this
> will result in PCI_BUS_FLAGS_DISABLE_ON_S2I being set for every PCI bus in
> the entire system.  But the description talks about an issue specifically with
> NVMe.
>
> Is there a defect in this AMD PCIe controller that affects all devices?
>
[Prike] In this solution by checking root complex DID to identify the question platform which need
the quirk flag. So far, only NVMe device need check this flag for special processing of NVMe
s2idle suspend.

> > In this patch prepare a PCIe RC bus flag to identify the platform
> > whether need the quirk.
> >
> > Cc: <stable@vger.kernel.org> # 5.10+
> > Signed-off-by: Prike Liang <Prike.Liang@amd.com>
> > Signed-off-by: Shyam Sundar S K <Shyam-sundar.S-k@amd.com>
> > [ck: split patches for nvme and pcie]
> > Signed-off-by: Chaitanya Kulkarni <chaitanya.kulkarni@wdc.com>
> > Suggested-by: Keith Busch <kbusch@kernel.org>
> > Acked-by: Keith Busch <kbusch@kernel.org>
> > ---
> > Changes in v2:
> > Fix the patch format and check chip root complex DID instead of PCIe
> > RP to avoid the storage device plugged in internal PCIe RP by USB adaptor.
> >
> > Changes in v3:
> > According to Christoph Hellwig do NVME PCIe related identify opt
> > better in PCIe quirk driver rather than in NVME module.
> >
> > Changes in v4:
> > Split the fix to PCIe and NVMe part and then call the pci_dev_put()
> > put the device reference count and finally refine the commit info.
> >
> > Changes in v5:
> > According to Christoph Hellwig and Keith Busch better use a
> > passthrough device(bus) gloable flag to identify the NVMe shutdown opt
> rather than look up the device BDF.
> > ---
> >  drivers/pci/probe.c  | 5 ++++-
> >  drivers/pci/quirks.c | 7 +++++++
> >  include/linux/pci.h  | 2 ++
> >  3 files changed, 13 insertions(+), 1 deletion(-)
> >
> > diff --git a/drivers/pci/probe.c b/drivers/pci/probe.c index
> > 953f15a..34ba691e 100644
> > --- a/drivers/pci/probe.c
> > +++ b/drivers/pci/probe.c
> > @@ -558,10 +558,13 @@ static struct pci_bus *pci_alloc_bus(struct
> pci_bus *parent)
> >     INIT_LIST_HEAD(&b->resources);
> >     b->max_bus_speed = PCI_SPEED_UNKNOWN;
> >     b->cur_bus_speed = PCI_SPEED_UNKNOWN;
> > +   if (parent) {
> >  #ifdef CONFIG_PCI_DOMAINS_GENERIC
> > -   if (parent)
> >             b->domain_nr = parent->domain_nr;
> >  #endif
> > +           if (parent->bus_flags & PCI_BUS_FLAGS_DISABLE_ON_S2I)
> > +                   b->bus_flags |= PCI_BUS_FLAGS_DISABLE_ON_S2I;
> > +   }
> >     return b;
> >  }
> >
> > diff --git a/drivers/pci/quirks.c b/drivers/pci/quirks.c index
> > 653660e3..7c4bb8e 100644
> > --- a/drivers/pci/quirks.c
> > +++ b/drivers/pci/quirks.c
> > @@ -312,6 +312,13 @@ static void quirk_nopciamd(struct pci_dev *dev)
> > }
> >  DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_AMD,
>       PCI_DEVICE_ID_AMD_8151_0,       quirk_nopciamd);
> >
> > +static void quirk_amd_s2i_fixup(struct pci_dev *dev) {
> > +   dev->bus->bus_flags |= PCI_BUS_FLAGS_DISABLE_ON_S2I;
> > +   pci_info(dev, "AMD simple suspend opt enabled\n"); }
> > +DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_AMD, 0x1630,
> > +quirk_amd_s2i_fixup);
> > +
> >  /* Triton requires workarounds to be used by the drivers */  static
> > void quirk_triton(struct pci_dev *dev)  { diff --git
> > a/include/linux/pci.h b/include/linux/pci.h index 53f4904..dc65219
> > 100644
> > --- a/include/linux/pci.h
> > +++ b/include/linux/pci.h
> > @@ -240,6 +240,8 @@ enum pci_bus_flags {
> >     PCI_BUS_FLAGS_NO_MMRBC  = (__force pci_bus_flags_t) 2,
> >     PCI_BUS_FLAGS_NO_AERSID = (__force pci_bus_flags_t) 4,
> >     PCI_BUS_FLAGS_NO_EXTCFG = (__force pci_bus_flags_t) 8,
> > +   /* Driver must pci_disable_device() for suspend-to-idle */
> > +   PCI_BUS_FLAGS_DISABLE_ON_S2I    = (__force pci_bus_flags_t) 16,
> >  };
> >
> >  /* Values from Link Status register, PCIe r3.1, sec 7.8.8 */
> > --
> > 2.7.4
> >

_______________________________________________
Linux-nvme mailing list
Linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme

Re: [PATCH v5 1/2] PCI: add AMD PCIe quirk for nvme shutdown opt

[Bjorn Helgaas]

On Thu, May 20, 2021 at 06:57:41AM +0000, Liang, Prike wrote:
> > From: Bjorn Helgaas <helgaas@kernel.org>
> > Sent: Thursday, May 20, 2021 5:34 AM
> > To: Liang, Prike <Prike.Liang@amd.com>
> > Cc: linux-pci@vger.kernel.org; kbusch@kernel.org; axboe@fb.com;
> > hch@lst.de; sagi@grimberg.me; linux-nvme@lists.infradead.org; Deucher,
> > Alexander <Alexander.Deucher@amd.com>; stable@vger.kernel.org; S-k,
> > Shyam-sundar <Shyam-sundar.S-k@amd.com>; Chaitanya Kulkarni
> > <chaitanya.kulkarni@wdc.com>; Rafael J. Wysocki <rjw@rjwysocki.net>;
> > linux-pm@vger.kernel.org
> > Subject: Re: [PATCH v5 1/2] PCI: add AMD PCIe quirk for nvme shutdown opt
> >
> > [+cc Rafael (probably nothing of interest to you), linux-pm]
> >
> > On Tue, May 18, 2021 at 10:24:34AM +0800, Prike Liang wrote:
> > > In the NVMe controller default suspend-resume seems only
> > > save/restore the NVMe link state by APST opt and the NVMe
> > > remains in D0 during this time.  Then the NVMe device will be
> > > shutdown by SMU firmware in the s2idle entry and then will lost
> > > the NVMe power context during s2idle resume.Finally, the NVMe
> > > command queue request will be processed abnormally and result in
> > > access timeout.This issue can be settled by using PCIe power set
> > > with simple suspend-resume process path instead of
> > > APST get/set opt.
> >
> > I can't parse the paragraph above, sorry.  I'm sure this means
> > something to NVMe developers, but since you're adding this to the
> > PCI core, not the NVMe core, it needs to be intelligible to
> > ordinary PCI folks.
> >
> [Prike]  I'm sorry to make confusion here. Those patches addressed a
> s2idle resume broken problem that the NVMe driver's default
> suspend-resume policy of using NVMe APST during suspend-to-idle
> prevents the PCI root port from going to D3.
>
> > For example, since you only use this flag in the NVMe driver, you
> > should explain why the PCI core needs to keep track of the flag
> > for you.  Normally I would assume the driver could figure this out
> > in its .probe() function.
> >
> [Prike] Yeah, we can assign the quirk flag in the .probe function or
> add it in nvme_id_table and this also the primary solution we tried
> out. However, that seems not possible to enumerate every uncertain
> NVMe device then assign quirk flag to them. In this case, in order
> to handle various NVMe device we can use the root complex device ID
> to identify the question platform.
> 
> > Quirks are usually used to work around a defect in a device.
> > What's the defect in this case?  Ideally we can point to a section
> > of the PCIe spec with a requirement that the device violates.
>
> [Prike] In this case the quirk is only used to identify the question
> platform which requires the NVMe device go to D3 in the s2idle
> suspend.
>
> > What does "opt" mean?
> >
> [Prike] I'm also not dedicate working on the NVMe driver, but from
> the software perspective the APST opt is used for handling the power
> state S&R without PCI interfering during s2idle legacy
> suspend-resume.
>
> > What is SMU firmware?  Why is it relevant?
> >
> [Prike] SMU firmware is a proprietary micro component which
> responsible for device power management. Without the quirk flag,
> NVMe device will not enter D3 during s2idle suspend then SMU
> firmware will shut down the NVMe device, unfortunately since NVMe is
> a third-party device the SMU firmware only restore NVMe root port
> power state during s2ilde wake up process. Eventually, the NVMe
> device power state will be lost when back to OS s2idle resume  and
> then result in NVMe command request failed.
> 
> > Is this a problem only with s2idle?  Why or why not?
> >
> [Prike] Yeah, this issue is only found in the s2idle scenario, and
> that's because s2idle will check whether each device will enter its
> own minimum power level defined in the LPI constrains table.
>
> > The quirk applies to [1022:1630].  An lspci I found on the web
> > says this is a "00:00.0 Host bridge: AMD Renoir Root Complex"
> > device.  So it looks like this will result in
> > PCI_BUS_FLAGS_DISABLE_ON_S2I being set for every PCI bus in the
> > entire system.  But the description talks about an issue
> > specifically with NVMe.
> >
> > Is there a defect in this AMD PCIe controller that affects all
> > devices?
> >
> [Prike] In this solution by checking root complex DID to identify
> the question platform which need the quirk flag. So far, only NVMe
> device need check this flag for special processing of NVMe s2idle
> suspend.

We're really not getting anywhere.  The commit log needs to explain
how this is related to PCI.  Apparently the issue is related to NVMe
APST and NVMe device state being lost.  AFAICT, APST has to do with
power states of the NVMe controller itself.  Those states are internal
to NVMe and are not directly visible to the Root Port.

Maybe there's a connection with ASPM or the Link state, or putting the
device in D3cold, or ...?  

Ideally it would describe something about this AMD PCIe controller
that doesn't work according to spec.

It should say something about why this flag needs to apply to *all*
devices below this controller, even though we currently only use it
for NVMe.

> > > In this patch prepare a PCIe RC bus flag to identify the
> > > platform whether need the quirk.
> > >
> > > Cc: <stable@vger.kernel.org> # 5.10+
> > > Signed-off-by: Prike Liang <Prike.Liang@amd.com>
> > > Signed-off-by: Shyam Sundar S K <Shyam-sundar.S-k@amd.com>
> > > [ck: split patches for nvme and pcie]
> > > Signed-off-by: Chaitanya Kulkarni <chaitanya.kulkarni@wdc.com>
> > > Suggested-by: Keith Busch <kbusch@kernel.org>
> > > Acked-by: Keith Busch <kbusch@kernel.org>
> > > ---
> > > Changes in v2:
> > > Fix the patch format and check chip root complex DID instead of PCIe
> > > RP to avoid the storage device plugged in internal PCIe RP by USB adaptor.
> > >
> > > Changes in v3:
> > > According to Christoph Hellwig do NVME PCIe related identify opt
> > > better in PCIe quirk driver rather than in NVME module.
> > >
> > > Changes in v4:
> > > Split the fix to PCIe and NVMe part and then call the pci_dev_put()
> > > put the device reference count and finally refine the commit info.
> > >
> > > Changes in v5:
> > > According to Christoph Hellwig and Keith Busch better use a
> > > passthrough device(bus) gloable flag to identify the NVMe shutdown opt
> > > rather than look up the device BDF.

The changelog above bears little resemblance to reality.  I dug up all
the previous postings, hoping there was a hint about why this is
relevant to PCI, but I didn't find anything.  For the archives, here
are the versions I found and notes about what really changed:

  v1 2021-04-14  2:18 https://lore.kernel.org/r/1618366694-14092-1-git-send-email-Prike.Liang@amd.com/

  v2 2021-04-14  6:19 https://lore.kernel.org/r/1618381200-14856-1-git-send-email-Prike.Liang@amd.com/
    (Not tagged as v2 in the posting.)
    Check result of pci_get_domain_bus_and_slot() for NULL.

  v3 2021-04-14  8:18 https://lore.kernel.org/r/1618388281-15629-1-git-send-email-Prike.Liang@amd.com/
    (Not tagged as v3 in the posting.)
    Drop reference acquired by pci_get_domain_bus_and_slot().
    Return "true" (not NVME_QUIRK_SIMPLE_SUSPEND) from bool
    nvme_acpi_storage_d3().

  v4 2021-04-15  3:52 https://lore.kernel.org/r/1618458725-17164-1-git-send-email-Prike.Liang@amd.com/
    Reorder Signed-off-by tags.
    No code change at all.

  v5 2021-04-16  6:54 https://lore.kernel.org/r/1618556075-24589-1-git-send-email-Prike.Liang@amd.com/
    Move flag from pci_dev_flags to pci_bus_flags.
    Rename flag to PCI_BUS_FLAGS_DISABLE_ON_S2I.
    Inherit PCI_BUS_FLAGS_DISABLE_ON_S2I in all child buses of AMD
    0x1630.
    Check dev->bus->bus_flags instead of using
    pci_get_domain_bus_and_slot().

> > > ---
> > >  drivers/pci/probe.c  | 5 ++++-
> > >  drivers/pci/quirks.c | 7 +++++++
> > >  include/linux/pci.h  | 2 ++
> > >  3 files changed, 13 insertions(+), 1 deletion(-)
> > >
> > > diff --git a/drivers/pci/probe.c b/drivers/pci/probe.c index
> > > 953f15a..34ba691e 100644
> > > --- a/drivers/pci/probe.c
> > > +++ b/drivers/pci/probe.c
> > > @@ -558,10 +558,13 @@ static struct pci_bus *pci_alloc_bus(struct
> > pci_bus *parent)
> > >     INIT_LIST_HEAD(&b->resources);
> > >     b->max_bus_speed = PCI_SPEED_UNKNOWN;
> > >     b->cur_bus_speed = PCI_SPEED_UNKNOWN;
> > > +   if (parent) {
> > >  #ifdef CONFIG_PCI_DOMAINS_GENERIC
> > > -   if (parent)
> > >             b->domain_nr = parent->domain_nr;
> > >  #endif
> > > +           if (parent->bus_flags & PCI_BUS_FLAGS_DISABLE_ON_S2I)
> > > +                   b->bus_flags |= PCI_BUS_FLAGS_DISABLE_ON_S2I;
> > > +   }
> > >     return b;
> > >  }
> > >
> > > diff --git a/drivers/pci/quirks.c b/drivers/pci/quirks.c index
> > > 653660e3..7c4bb8e 100644
> > > --- a/drivers/pci/quirks.c
> > > +++ b/drivers/pci/quirks.c
> > > @@ -312,6 +312,13 @@ static void quirk_nopciamd(struct pci_dev *dev)
> > > }
> > >  DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_AMD,
> >       PCI_DEVICE_ID_AMD_8151_0,       quirk_nopciamd);
> > >
> > > +static void quirk_amd_s2i_fixup(struct pci_dev *dev) {
> > > +   dev->bus->bus_flags |= PCI_BUS_FLAGS_DISABLE_ON_S2I;
> > > +   pci_info(dev, "AMD simple suspend opt enabled\n"); }
> > > +DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_AMD, 0x1630,
> > > +quirk_amd_s2i_fixup);
> > > +
> > >  /* Triton requires workarounds to be used by the drivers */  static
> > > void quirk_triton(struct pci_dev *dev)  { diff --git
> > > a/include/linux/pci.h b/include/linux/pci.h index 53f4904..dc65219
> > > 100644
> > > --- a/include/linux/pci.h
> > > +++ b/include/linux/pci.h
> > > @@ -240,6 +240,8 @@ enum pci_bus_flags {
> > >     PCI_BUS_FLAGS_NO_MMRBC  = (__force pci_bus_flags_t) 2,
> > >     PCI_BUS_FLAGS_NO_AERSID = (__force pci_bus_flags_t) 4,
> > >     PCI_BUS_FLAGS_NO_EXTCFG = (__force pci_bus_flags_t) 8,
> > > +   /* Driver must pci_disable_device() for suspend-to-idle */
> > > +   PCI_BUS_FLAGS_DISABLE_ON_S2I    = (__force pci_bus_flags_t) 16,
> > >  };
> > >
> > >  /* Values from Link Status register, PCIe r3.1, sec 7.8.8 */
> > > --
> > > 2.7.4
> > >

_______________________________________________
Linux-nvme mailing list
Linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme

RE: [PATCH v5 1/2] PCI: add AMD PCIe quirk for nvme shutdown opt

[Deucher, Alexander]

[AMD Public Use]

> -----Original Message-----
> From: Bjorn Helgaas <helgaas@kernel.org>
> Sent: Thursday, May 20, 2021 12:59 PM
> To: Liang, Prike <Prike.Liang@amd.com>
> Cc: linux-pci@vger.kernel.org; kbusch@kernel.org; axboe@fb.com;
> hch@lst.de; sagi@grimberg.me; linux-nvme@lists.infradead.org; Deucher,
> Alexander <Alexander.Deucher@amd.com>; stable@vger.kernel.org; S-k,
> Shyam-sundar <Shyam-sundar.S-k@amd.com>; Chaitanya Kulkarni
> <chaitanya.kulkarni@wdc.com>; Rafael J. Wysocki <rjw@rjwysocki.net>;
> linux-pm@vger.kernel.org
> Subject: Re: [PATCH v5 1/2] PCI: add AMD PCIe quirk for nvme shutdown opt
>
> On Thu, May 20, 2021 at 06:57:41AM +0000, Liang, Prike wrote:
> > > From: Bjorn Helgaas <helgaas@kernel.org>
> > > Sent: Thursday, May 20, 2021 5:34 AM
> > > To: Liang, Prike <Prike.Liang@amd.com>
> > > Cc: linux-pci@vger.kernel.org; kbusch@kernel.org; axboe@fb.com;
> > > hch@lst.de; sagi@grimberg.me; linux-nvme@lists.infradead.org;
> > > Deucher, Alexander <Alexander.Deucher@amd.com>;
> > > stable@vger.kernel.org; S-k, Shyam-sundar
> > > <Shyam-sundar.S-k@amd.com>; Chaitanya Kulkarni
> > > <chaitanya.kulkarni@wdc.com>; Rafael J. Wysocki <rjw@rjwysocki.net>;
> > > linux-pm@vger.kernel.org
> > > Subject: Re: [PATCH v5 1/2] PCI: add AMD PCIe quirk for nvme
> > > shutdown opt
> > >
> > > [+cc Rafael (probably nothing of interest to you), linux-pm]
> > >
> > > On Tue, May 18, 2021 at 10:24:34AM +0800, Prike Liang wrote:
> > > > In the NVMe controller default suspend-resume seems only
> > > > save/restore the NVMe link state by APST opt and the NVMe remains
> > > > in D0 during this time.  Then the NVMe device will be shutdown by
> > > > SMU firmware in the s2idle entry and then will lost the NVMe power
> > > > context during s2idle resume.Finally, the NVMe command queue
> > > > request will be processed abnormally and result in access
> > > > timeout.This issue can be settled by using PCIe power set with
> > > > simple suspend-resume process path instead of APST get/set opt.
> > >
> > > I can't parse the paragraph above, sorry.  I'm sure this means
> > > something to NVMe developers, but since you're adding this to the
> > > PCI core, not the NVMe core, it needs to be intelligible to ordinary
> > > PCI folks.
> > >
> > [Prike]  I'm sorry to make confusion here. Those patches addressed a
> > s2idle resume broken problem that the NVMe driver's default
> > suspend-resume policy of using NVMe APST during suspend-to-idle
> > prevents the PCI root port from going to D3.
> >
> > > For example, since you only use this flag in the NVMe driver, you
> > > should explain why the PCI core needs to keep track of the flag for
> > > you.  Normally I would assume the driver could figure this out in
> > > its .probe() function.
> > >
> > [Prike] Yeah, we can assign the quirk flag in the .probe function or
> > add it in nvme_id_table and this also the primary solution we tried
> > out. However, that seems not possible to enumerate every uncertain
> > NVMe device then assign quirk flag to them. In this case, in order to
> > handle various NVMe device we can use the root complex device ID to
> > identify the question platform.
> >
> > > Quirks are usually used to work around a defect in a device.
> > > What's the defect in this case?  Ideally we can point to a section
> > > of the PCIe spec with a requirement that the device violates.
> >
> > [Prike] In this case the quirk is only used to identify the question
> > platform which requires the NVMe device go to D3 in the s2idle
> > suspend.
> >
> > > What does "opt" mean?
> > >
> > [Prike] I'm also not dedicate working on the NVMe driver, but from the
> > software perspective the APST opt is used for handling the power state
> > S&R without PCI interfering during s2idle legacy suspend-resume.
> >
> > > What is SMU firmware?  Why is it relevant?
> > >
> > [Prike] SMU firmware is a proprietary micro component which
> > responsible for device power management. Without the quirk flag, NVMe
> > device will not enter D3 during s2idle suspend then SMU firmware will
> > shut down the NVMe device, unfortunately since NVMe is a third-party
> > device the SMU firmware only restore NVMe root port power state during
> > s2ilde wake up process. Eventually, the NVMe device power state will
> > be lost when back to OS s2idle resume  and then result in NVMe command
> > request failed.
> >
> > > Is this a problem only with s2idle?  Why or why not?
> > >
> > [Prike] Yeah, this issue is only found in the s2idle scenario, and
> > that's because s2idle will check whether each device will enter its
> > own minimum power level defined in the LPI constrains table.
> >
> > > The quirk applies to [1022:1630].  An lspci I found on the web says
> > > this is a "00:00.0 Host bridge: AMD Renoir Root Complex"
> > > device.  So it looks like this will result in
> > > PCI_BUS_FLAGS_DISABLE_ON_S2I being set for every PCI bus in the
> > > entire system.  But the description talks about an issue
> > > specifically with NVMe.
> > >
> > > Is there a defect in this AMD PCIe controller that affects all
> > > devices?
> > >
> > [Prike] In this solution by checking root complex DID to identify the
> > question platform which need the quirk flag. So far, only NVMe device
> > need check this flag for special processing of NVMe s2idle suspend.
>
> We're really not getting anywhere.  The commit log needs to explain how this
> is related to PCI.  Apparently the issue is related to NVMe APST and NVMe
> device state being lost.  AFAICT, APST has to do with power states of the
> NVMe controller itself.  Those states are internal to NVMe and are not
> directly visible to the Root Port.
>
> Maybe there's a connection with ASPM or the Link state, or putting the
> device in D3cold, or ...?
>
> Ideally it would describe something about this AMD PCIe controller that
> doesn't work according to spec.
>
> It should say something about why this flag needs to apply to *all* devices
> below this controller, even though we currently only use it for NVMe.
>

It doesn't really have anything to do with PCI.  The PCI link is just a proxy for specific AMD platforms.  It's platform firmware behavior we are catering to.  This was originally posted as an nvme quirk, but during the review it was recommended to move the quirk into PCI because the quirk is not specific a particular NVMe device, but rather a set of AMD platforms.  Lots of other platforms seems to do similar things in the nvme driver based on ACPI or DMI flags, etc.  On our hardware this nvme flag is required for all cezanne and renoir platforms.

Here's the original nvme patch discussions:
http://lists.infradead.org/pipermail/linux-nvme/2021-March/023973.html
http://lists.infradead.org/pipermail/linux-nvme/2021-March/024017.html
http://lists.infradead.org/pipermail/linux-nvme/2021-April/024291.html

Alex

> > > > In this patch prepare a PCIe RC bus flag to identify the platform
> > > > whether need the quirk.
> > > >
> > > > Cc: <stable@vger.kernel.org> # 5.10+
> > > > Signed-off-by: Prike Liang <Prike.Liang@amd.com>
> > > > Signed-off-by: Shyam Sundar S K <Shyam-sundar.S-k@amd.com>
> > > > [ck: split patches for nvme and pcie]
> > > > Signed-off-by: Chaitanya Kulkarni <chaitanya.kulkarni@wdc.com>
> > > > Suggested-by: Keith Busch <kbusch@kernel.org>
> > > > Acked-by: Keith Busch <kbusch@kernel.org>
> > > > ---
> > > > Changes in v2:
> > > > Fix the patch format and check chip root complex DID instead of
> > > > PCIe RP to avoid the storage device plugged in internal PCIe RP by USB
> adaptor.
> > > >
> > > > Changes in v3:
> > > > According to Christoph Hellwig do NVME PCIe related identify opt
> > > > better in PCIe quirk driver rather than in NVME module.
> > > >
> > > > Changes in v4:
> > > > Split the fix to PCIe and NVMe part and then call the
> > > > pci_dev_put() put the device reference count and finally refine the
> commit info.
> > > >
> > > > Changes in v5:
> > > > According to Christoph Hellwig and Keith Busch better use a
> > > > passthrough device(bus) gloable flag to identify the NVMe shutdown
> > > > opt rather than look up the device BDF.
>
> The changelog above bears little resemblance to reality.  I dug up all the
> previous postings, hoping there was a hint about why this is relevant to PCI,
> but I didn't find anything.  For the archives, here are the versions I found and
> notes about what really changed:
>
>   v1 2021-04-14  2:18
> https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Flore.
> kernel.org%2Fr%2F1618366694-14092-1-git-send-email-
> Prike.Liang%40amd.com%2F&amp;data=04%7C01%7CAlexander.Deucher%4
> 0amd.com%7C7773f0f2af054b0ab45408d91bb08ad8%7C3dd8961fe4884e608e
> 11a82d994e183d%7C0%7C0%7C637571267348438733%7CUnknown%7CTWFp
> bGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVC
> I6Mn0%3D%7C1000&amp;sdata=KFiTsmHEIbv9Qe7UI4yjKjXr1mWSusA2LYV
> M07GUGAY%3D&amp;reserved=0
>
>   v2 2021-04-14  6:19
> https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Flore.
> kernel.org%2Fr%2F1618381200-14856-1-git-send-email-
> Prike.Liang%40amd.com%2F&amp;data=04%7C01%7CAlexander.Deucher%4
> 0amd.com%7C7773f0f2af054b0ab45408d91bb08ad8%7C3dd8961fe4884e608e
> 11a82d994e183d%7C0%7C0%7C637571267348438733%7CUnknown%7CTWFp
> bGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVC
> I6Mn0%3D%7C1000&amp;sdata=I2rKN%2Bixmn2NcXcZ20seu9LQC%2BekwH
> azH9oqzym7eeE%3D&amp;reserved=0
>     (Not tagged as v2 in the posting.)
>     Check result of pci_get_domain_bus_and_slot() for NULL.
>
>   v3 2021-04-14  8:18
> https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Flore.
> kernel.org%2Fr%2F1618388281-15629-1-git-send-email-
> Prike.Liang%40amd.com%2F&amp;data=04%7C01%7CAlexander.Deucher%4
> 0amd.com%7C7773f0f2af054b0ab45408d91bb08ad8%7C3dd8961fe4884e608e
> 11a82d994e183d%7C0%7C0%7C637571267348438733%7CUnknown%7CTWFp
> bGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVC
> I6Mn0%3D%7C1000&amp;sdata=7T9VcWinB8XQTR4tVKMZ4WkCI5oGKBBuH
> n33wxIZzbI%3D&amp;reserved=0
>     (Not tagged as v3 in the posting.)
>     Drop reference acquired by pci_get_domain_bus_and_slot().
>     Return "true" (not NVME_QUIRK_SIMPLE_SUSPEND) from bool
>     nvme_acpi_storage_d3().
>
>   v4 2021-04-15  3:52
> https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Flore.
> kernel.org%2Fr%2F1618458725-17164-1-git-send-email-
> Prike.Liang%40amd.com%2F&amp;data=04%7C01%7CAlexander.Deucher%4
> 0amd.com%7C7773f0f2af054b0ab45408d91bb08ad8%7C3dd8961fe4884e608e
> 11a82d994e183d%7C0%7C0%7C637571267348438733%7CUnknown%7CTWFp
> bGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVC
> I6Mn0%3D%7C1000&amp;sdata=aTgZtIfWMeoWZ3HoHcjclrJPW8qknIsuoVaz
> hZnoDHo%3D&amp;reserved=0
>     Reorder Signed-off-by tags.
>     No code change at all.
>
>   v5 2021-04-16  6:54
> https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Flore.
> kernel.org%2Fr%2F1618556075-24589-1-git-send-email-
> Prike.Liang%40amd.com%2F&amp;data=04%7C01%7CAlexander.Deucher%4
> 0amd.com%7C7773f0f2af054b0ab45408d91bb08ad8%7C3dd8961fe4884e608e
> 11a82d994e183d%7C0%7C0%7C637571267348438733%7CUnknown%7CTWFp
> bGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVC
> I6Mn0%3D%7C1000&amp;sdata=5yzUyXfIDKxTE%2B%2FJy5fgOS9yCsU10uw
> n4CjGd2mLLc0%3D&amp;reserved=0
>     Move flag from pci_dev_flags to pci_bus_flags.
>     Rename flag to PCI_BUS_FLAGS_DISABLE_ON_S2I.
>     Inherit PCI_BUS_FLAGS_DISABLE_ON_S2I in all child buses of AMD
>     0x1630.
>     Check dev->bus->bus_flags instead of using
>     pci_get_domain_bus_and_slot().
>
> > > > ---
> > > >  drivers/pci/probe.c  | 5 ++++-
> > > >  drivers/pci/quirks.c | 7 +++++++
> > > >  include/linux/pci.h  | 2 ++
> > > >  3 files changed, 13 insertions(+), 1 deletion(-)
> > > >
> > > > diff --git a/drivers/pci/probe.c b/drivers/pci/probe.c index
> > > > 953f15a..34ba691e 100644
> > > > --- a/drivers/pci/probe.c
> > > > +++ b/drivers/pci/probe.c
> > > > @@ -558,10 +558,13 @@ static struct pci_bus *pci_alloc_bus(struct
> > > pci_bus *parent)
> > > >     INIT_LIST_HEAD(&b->resources);
> > > >     b->max_bus_speed = PCI_SPEED_UNKNOWN;
> > > >     b->cur_bus_speed = PCI_SPEED_UNKNOWN;
> > > > +   if (parent) {
> > > >  #ifdef CONFIG_PCI_DOMAINS_GENERIC
> > > > -   if (parent)
> > > >             b->domain_nr = parent->domain_nr;  #endif
> > > > +           if (parent->bus_flags & PCI_BUS_FLAGS_DISABLE_ON_S2I)
> > > > +                   b->bus_flags |= PCI_BUS_FLAGS_DISABLE_ON_S2I;
> > > > +   }
> > > >     return b;
> > > >  }
> > > >
> > > > diff --git a/drivers/pci/quirks.c b/drivers/pci/quirks.c index
> > > > 653660e3..7c4bb8e 100644
> > > > --- a/drivers/pci/quirks.c
> > > > +++ b/drivers/pci/quirks.c
> > > > @@ -312,6 +312,13 @@ static void quirk_nopciamd(struct pci_dev
> > > > *dev) }  DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_AMD,
> > >       PCI_DEVICE_ID_AMD_8151_0,       quirk_nopciamd);
> > > >
> > > > +static void quirk_amd_s2i_fixup(struct pci_dev *dev) {
> > > > +   dev->bus->bus_flags |= PCI_BUS_FLAGS_DISABLE_ON_S2I;
> > > > +   pci_info(dev, "AMD simple suspend opt enabled\n"); }
> > > > +DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_AMD, 0x1630,
> > > > +quirk_amd_s2i_fixup);
> > > > +
> > > >  /* Triton requires workarounds to be used by the drivers */
> > > > static void quirk_triton(struct pci_dev *dev)  { diff --git
> > > > a/include/linux/pci.h b/include/linux/pci.h index 53f4904..dc65219
> > > > 100644
> > > > --- a/include/linux/pci.h
> > > > +++ b/include/linux/pci.h
> > > > @@ -240,6 +240,8 @@ enum pci_bus_flags {
> > > >     PCI_BUS_FLAGS_NO_MMRBC  = (__force pci_bus_flags_t) 2,
> > > >     PCI_BUS_FLAGS_NO_AERSID = (__force pci_bus_flags_t) 4,
> > > >     PCI_BUS_FLAGS_NO_EXTCFG = (__force pci_bus_flags_t) 8,
> > > > +   /* Driver must pci_disable_device() for suspend-to-idle */
> > > > +   PCI_BUS_FLAGS_DISABLE_ON_S2I    = (__force pci_bus_flags_t) 16,
> > > >  };
> > > >
> > > >  /* Values from Link Status register, PCIe r3.1, sec 7.8.8 */
> > > > --
> > > > 2.7.4
> > > >

_______________________________________________
Linux-nvme mailing list
Linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme

Re: [PATCH v5 1/2] PCI: add AMD PCIe quirk for nvme shutdown opt

[Keith Busch]

On Thu, May 20, 2021 at 05:40:54PM +0000, Deucher, Alexander wrote:
> It doesn't really have anything to do with PCI.  The PCI link is just a proxy
> for specific AMD platforms.  It's platform firmware behavior we are catering
> to.  This was originally posted as an nvme quirk, but during the review it
> was recommended to move the quirk into PCI because the quirk is not specific
> a particular NVMe device, but rather a set of AMD platforms.  Lots of other
> platforms seems to do similar things in the nvme driver based on ACPI or DMI
> flags, etc.  On our hardware this nvme flag is required for all cezanne and
> renoir platforms.

The quirk was initially presented as specific to the pci root. Does it make
more sense for nvme to recognize the limitation from querying a different
platform component instead of the pci bus?

_______________________________________________
Linux-nvme mailing list
Linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme

RE: [PATCH v5 1/2] PCI: add AMD PCIe quirk for nvme shutdown opt

[Deucher, Alexander]

[AMD Public Use]

> -----Original Message-----
> From: Keith Busch <kbusch@kernel.org>
> Sent: Thursday, May 20, 2021 2:04 PM
> To: Deucher, Alexander <Alexander.Deucher@amd.com>
> Cc: Bjorn Helgaas <helgaas@kernel.org>; Liang, Prike
> <Prike.Liang@amd.com>; linux-pci@vger.kernel.org; axboe@fb.com;
> hch@lst.de; sagi@grimberg.me; linux-nvme@lists.infradead.org;
> stable@vger.kernel.org; S-k, Shyam-sundar <Shyam-sundar.S-k@amd.com>;
> Chaitanya Kulkarni <chaitanya.kulkarni@wdc.com>; Rafael J. Wysocki
> <rjw@rjwysocki.net>; linux-pm@vger.kernel.org
> Subject: Re: [PATCH v5 1/2] PCI: add AMD PCIe quirk for nvme shutdown opt
> 
> On Thu, May 20, 2021 at 05:40:54PM +0000, Deucher, Alexander wrote:
> > It doesn't really have anything to do with PCI.  The PCI link is just
> > a proxy for specific AMD platforms.  It's platform firmware behavior
> > we are catering to.  This was originally posted as an nvme quirk, but
> > during the review it was recommended to move the quirk into PCI
> > because the quirk is not specific a particular NVMe device, but rather
> > a set of AMD platforms.  Lots of other platforms seems to do similar
> > things in the nvme driver based on ACPI or DMI flags, etc.  On our
> > hardware this nvme flag is required for all cezanne and renoir platforms.
> 
> The quirk was initially presented as specific to the pci root. Does it make
> more sense for nvme to recognize the limitation from querying a different
> platform component instead of the pci bus?

Maybe.  I'm not sure what the best way to tie this to a specific platform is.  @Limonciello, Mario?

Alex

_______________________________________________
Linux-nvme mailing list
Linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme

RE: [PATCH v5 1/2] PCI: add AMD PCIe quirk for nvme shutdown opt

[Limonciello, Mario]

[AMD Public Use]

> > -----Original Message-----
> > From: Keith Busch <kbusch@kernel.org>
> > Sent: Thursday, May 20, 2021 2:04 PM
> > To: Deucher, Alexander <Alexander.Deucher@amd.com>
> > Cc: Bjorn Helgaas <helgaas@kernel.org>; Liang, Prike
> > <Prike.Liang@amd.com>; linux-pci@vger.kernel.org; axboe@fb.com;
> > hch@lst.de; sagi@grimberg.me; linux-nvme@lists.infradead.org;
> > stable@vger.kernel.org; S-k, Shyam-sundar <Shyam-sundar.S-
> k@amd.com>;
> > Chaitanya Kulkarni <chaitanya.kulkarni@wdc.com>; Rafael J. Wysocki
> > <rjw@rjwysocki.net>; linux-pm@vger.kernel.org
> > Subject: Re: [PATCH v5 1/2] PCI: add AMD PCIe quirk for nvme shutdown
> opt
> >
> > On Thu, May 20, 2021 at 05:40:54PM +0000, Deucher, Alexander wrote:
> > > It doesn't really have anything to do with PCI.  The PCI link is just
> > > a proxy for specific AMD platforms.  It's platform firmware behavior
> > > we are catering to.  This was originally posted as an nvme quirk, but
> > > during the review it was recommended to move the quirk into PCI
> > > because the quirk is not specific a particular NVMe device, but rather
> > > a set of AMD platforms.  Lots of other platforms seems to do similar
> > > things in the nvme driver based on ACPI or DMI flags, etc.  On our
> > > hardware this nvme flag is required for all cezanne and renoir platforms.
> >
> > The quirk was initially presented as specific to the pci root. Does it make
> > more sense for nvme to recognize the limitation from querying a different
> > platform component instead of the pci bus?
> 
> Maybe.  I'm not sure what the best way to tie this to a specific platform is.
> @Limonciello, Mario?
> 

I'll just remind folks that Prike mentioned this is a problem specific to the
Renoir and Cezanne ASICs.  These were the first ones that S2idle was used.
"Future" designs the problems that cause the need for this change should be fixed.

With that in mind, I can see the argument from Bjorn to not over-engineer it and
claim a PCI quirk that applies to all the downstream PCIe devices when this is just
needed for NVME devices.  The PCI device id selected (0x1630) is the root complex associated
specifically to these ASICs.  

Since these are mobile platforms that don't contain any way to connect other external
PCIe devices I think another way to safely do it could be an if #ifdef CONFIG_X86 and then
check if set for doing s2i and if so do a x86_cpu_match() to the model and families
matching the CPUs.

To me this seems like a fine compromise given there is a precedent for dmi_match on
OEM platforms and enumerating "all" of the OEM platforms that contain CZN/RN and enable
S2I may be an exercise in futility.

_______________________________________________
Linux-nvme mailing list
Linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme

RE: [PATCH v5 1/2] PCI: add AMD PCIe quirk for nvme shutdown opt

[Liang, Prike]

[Public]

> -----Original Message-----
> From: Limonciello, Mario <Mario.Limonciello@amd.com>
> Sent: Friday, May 21, 2021 4:34 AM
> To: Deucher, Alexander <Alexander.Deucher@amd.com>; Keith Busch
> <kbusch@kernel.org>
> Cc: Bjorn Helgaas <helgaas@kernel.org>; Liang, Prike
> <Prike.Liang@amd.com>; linux-pci@vger.kernel.org; axboe@fb.com;
> hch@lst.de; sagi@grimberg.me; linux-nvme@lists.infradead.org;
> stable@vger.kernel.org; S-k, Shyam-sundar <Shyam-sundar.S-k@amd.com>;
> Chaitanya Kulkarni <chaitanya.kulkarni@wdc.com>; Rafael J. Wysocki
> <rjw@rjwysocki.net>; linux-pm@vger.kernel.org
> Subject: RE: [PATCH v5 1/2] PCI: add AMD PCIe quirk for nvme shutdown opt
>
> [AMD Public Use]
>
> > > -----Original Message-----
> > > From: Keith Busch <kbusch@kernel.org>
> > > Sent: Thursday, May 20, 2021 2:04 PM
> > > To: Deucher, Alexander <Alexander.Deucher@amd.com>
> > > Cc: Bjorn Helgaas <helgaas@kernel.org>; Liang, Prike
> > > <Prike.Liang@amd.com>; linux-pci@vger.kernel.org; axboe@fb.com;
> > > hch@lst.de; sagi@grimberg.me; linux-nvme@lists.infradead.org;
> > > stable@vger.kernel.org; S-k, Shyam-sundar <Shyam-sundar.S-
> > k@amd.com>;
> > > Chaitanya Kulkarni <chaitanya.kulkarni@wdc.com>; Rafael J. Wysocki
> > > <rjw@rjwysocki.net>; linux-pm@vger.kernel.org
> > > Subject: Re: [PATCH v5 1/2] PCI: add AMD PCIe quirk for nvme
> > > shutdown
> > opt
> > >
> > > On Thu, May 20, 2021 at 05:40:54PM +0000, Deucher, Alexander wrote:
> > > > It doesn't really have anything to do with PCI.  The PCI link is
> > > > just a proxy for specific AMD platforms.  It's platform firmware
> > > > behavior we are catering to.  This was originally posted as an
> > > > nvme quirk, but during the review it was recommended to move the
> > > > quirk into PCI because the quirk is not specific a particular NVMe
> > > > device, but rather a set of AMD platforms.  Lots of other
> > > > platforms seems to do similar things in the nvme driver based on
> > > > ACPI or DMI flags, etc.  On our hardware this nvme flag is required for
> all cezanne and renoir platforms.
> > >
> > > The quirk was initially presented as specific to the pci root. Does
> > > it make more sense for nvme to recognize the limitation from
> > > querying a different platform component instead of the pci bus?
> >
> > Maybe.  I'm not sure what the best way to tie this to a specific platform is.
> > @Limonciello, Mario?
> >
>
> I'll just remind folks that Prike mentioned this is a problem specific to the
> Renoir and Cezanne ASICs.  These were the first ones that S2idle was used.
> "Future" designs the problems that cause the need for this change should be
> fixed.
>
> With that in mind, I can see the argument from Bjorn to not over-engineer it
> and claim a PCI quirk that applies to all the downstream PCIe devices when
> this is just needed for NVME devices.  The PCI device id selected (0x1630) is
> the root complex associated specifically to these ASICs.
>
> Since these are mobile platforms that don't contain any way to connect other
> external PCIe devices I think another way to safely do it could be an if #ifdef
> CONFIG_X86 and then check if set for doing s2i and if so do a
> x86_cpu_match() to the model and families matching the CPUs.
>
> To me this seems like a fine compromise given there is a precedent for
> dmi_match on OEM platforms and enumerating "all" of the OEM platforms
> that contain CZN/RN and enable S2I may be an exercise in futility.

Thanks for all proposal and will draft a new patch for this.



_______________________________________________
Linux-nvme mailing list
Linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme

Re: [PATCH v5 1/2] PCI: add AMD PCIe quirk for nvme shutdown opt

[Bjorn Helgaas]

On Thu, May 20, 2021 at 05:40:54PM +0000, Deucher, Alexander wrote:
> [AMD Public Use]
> 
> > -----Original Message-----
> > From: Bjorn Helgaas <helgaas@kernel.org>
> > Sent: Thursday, May 20, 2021 12:59 PM
> > To: Liang, Prike <Prike.Liang@amd.com>
> > Cc: linux-pci@vger.kernel.org; kbusch@kernel.org; axboe@fb.com;
> > hch@lst.de; sagi@grimberg.me; linux-nvme@lists.infradead.org; Deucher,
> > Alexander <Alexander.Deucher@amd.com>; stable@vger.kernel.org; S-k,
> > Shyam-sundar <Shyam-sundar.S-k@amd.com>; Chaitanya Kulkarni
> > <chaitanya.kulkarni@wdc.com>; Rafael J. Wysocki <rjw@rjwysocki.net>;
> > linux-pm@vger.kernel.org
> > Subject: Re: [PATCH v5 1/2] PCI: add AMD PCIe quirk for nvme shutdown opt
> >
> > On Thu, May 20, 2021 at 06:57:41AM +0000, Liang, Prike wrote:
> > > > From: Bjorn Helgaas <helgaas@kernel.org>
> > > > Sent: Thursday, May 20, 2021 5:34 AM
> > > > To: Liang, Prike <Prike.Liang@amd.com>
> > > > Cc: linux-pci@vger.kernel.org; kbusch@kernel.org; axboe@fb.com;
> > > > hch@lst.de; sagi@grimberg.me; linux-nvme@lists.infradead.org;
> > > > Deucher, Alexander <Alexander.Deucher@amd.com>;
> > > > stable@vger.kernel.org; S-k, Shyam-sundar
> > > > <Shyam-sundar.S-k@amd.com>; Chaitanya Kulkarni
> > > > <chaitanya.kulkarni@wdc.com>; Rafael J. Wysocki <rjw@rjwysocki.net>;
> > > > linux-pm@vger.kernel.org
> > > > Subject: Re: [PATCH v5 1/2] PCI: add AMD PCIe quirk for nvme
> > > > shutdown opt
> > > >
> > > > [+cc Rafael (probably nothing of interest to you), linux-pm]
> > > >
> > > > On Tue, May 18, 2021 at 10:24:34AM +0800, Prike Liang wrote:
> > > > > In the NVMe controller default suspend-resume seems only
> > > > > save/restore the NVMe link state by APST opt and the NVMe remains
> > > > > in D0 during this time.  Then the NVMe device will be shutdown by
> > > > > SMU firmware in the s2idle entry and then will lost the NVMe power
> > > > > context during s2idle resume.Finally, the NVMe command queue
> > > > > request will be processed abnormally and result in access
> > > > > timeout.This issue can be settled by using PCIe power set with
> > > > > simple suspend-resume process path instead of APST get/set opt.
> > > >
> > > > I can't parse the paragraph above, sorry.  I'm sure this means
> > > > something to NVMe developers, but since you're adding this to the
> > > > PCI core, not the NVMe core, it needs to be intelligible to ordinary
> > > > PCI folks.
> > > >
> > > [Prike]  I'm sorry to make confusion here. Those patches addressed a
> > > s2idle resume broken problem that the NVMe driver's default
> > > suspend-resume policy of using NVMe APST during suspend-to-idle
> > > prevents the PCI root port from going to D3.
> > >
> > > > For example, since you only use this flag in the NVMe driver, you
> > > > should explain why the PCI core needs to keep track of the flag for
> > > > you.  Normally I would assume the driver could figure this out in
> > > > its .probe() function.
> > > >
> > > [Prike] Yeah, we can assign the quirk flag in the .probe function or
> > > add it in nvme_id_table and this also the primary solution we tried
> > > out. However, that seems not possible to enumerate every uncertain
> > > NVMe device then assign quirk flag to them. In this case, in order to
> > > handle various NVMe device we can use the root complex device ID to
> > > identify the question platform.
> > >
> > > > Quirks are usually used to work around a defect in a device.
> > > > What's the defect in this case?  Ideally we can point to a section
> > > > of the PCIe spec with a requirement that the device violates.
> > >
> > > [Prike] In this case the quirk is only used to identify the question
> > > platform which requires the NVMe device go to D3 in the s2idle
> > > suspend.
> > >
> > > > What does "opt" mean?
> > > >
> > > [Prike] I'm also not dedicate working on the NVMe driver, but from the
> > > software perspective the APST opt is used for handling the power state
> > > S&R without PCI interfering during s2idle legacy suspend-resume.
> > >
> > > > What is SMU firmware?  Why is it relevant?
> > > >
> > > [Prike] SMU firmware is a proprietary micro component which
> > > responsible for device power management. Without the quirk flag, NVMe
> > > device will not enter D3 during s2idle suspend then SMU firmware will
> > > shut down the NVMe device, unfortunately since NVMe is a third-party
> > > device the SMU firmware only restore NVMe root port power state during
> > > s2ilde wake up process. Eventually, the NVMe device power state will
> > > be lost when back to OS s2idle resume  and then result in NVMe command
> > > request failed.
> > >
> > > > Is this a problem only with s2idle?  Why or why not?
> > > >
> > > [Prike] Yeah, this issue is only found in the s2idle scenario, and
> > > that's because s2idle will check whether each device will enter its
> > > own minimum power level defined in the LPI constrains table.
> > >
> > > > The quirk applies to [1022:1630].  An lspci I found on the web says
> > > > this is a "00:00.0 Host bridge: AMD Renoir Root Complex"
> > > > device.  So it looks like this will result in
> > > > PCI_BUS_FLAGS_DISABLE_ON_S2I being set for every PCI bus in the
> > > > entire system.  But the description talks about an issue
> > > > specifically with NVMe.
> > > >
> > > > Is there a defect in this AMD PCIe controller that affects all
> > > > devices?
> > > >
> > > [Prike] In this solution by checking root complex DID to identify the
> > > question platform which need the quirk flag. So far, only NVMe device
> > > need check this flag for special processing of NVMe s2idle suspend.
> >
> > We're really not getting anywhere.  The commit log needs to explain how this
> > is related to PCI.  Apparently the issue is related to NVMe APST and NVMe
> > device state being lost.  AFAICT, APST has to do with power states of the
> > NVMe controller itself.  Those states are internal to NVMe and are not
> > directly visible to the Root Port.
> >
> > Maybe there's a connection with ASPM or the Link state, or putting the
> > device in D3cold, or ...?
> >
> > Ideally it would describe something about this AMD PCIe controller that
> > doesn't work according to spec.
> >
> > It should say something about why this flag needs to apply to *all* devices
> > below this controller, even though we currently only use it for NVMe.
> 
> It doesn't really have anything to do with PCI.  The PCI link is
> just a proxy for specific AMD platforms.  It's platform firmware
> behavior we are catering to.  This was originally posted as an nvme
> quirk, but during the review it was recommended to move the quirk
> into PCI because the quirk is not specific a particular NVMe device,
> but rather a set of AMD platforms.  Lots of other platforms seems to
> do similar things in the nvme driver based on ACPI or DMI flags,
> etc.  On our hardware this nvme flag is required for all cezanne and
> renoir platforms.

If it has nothing to do with PCI, and you need to detect the platform,
why not a DMI or similar quirk?

> Here's the original nvme patch discussions:
> http://lists.infradead.org/pipermail/linux-nvme/2021-March/023973.html
> http://lists.infradead.org/pipermail/linux-nvme/2021-March/024017.html
> http://lists.infradead.org/pipermail/linux-nvme/2021-April/024291.html

These contain no additional information.  Certainly nothing that would
explain why this should be a PCI quirk.

> > > > > In this patch prepare a PCIe RC bus flag to identify the platform
> > > > > whether need the quirk.
> > > > >
> > > > > Cc: <stable@vger.kernel.org> # 5.10+
> > > > > Signed-off-by: Prike Liang <Prike.Liang@amd.com>
> > > > > Signed-off-by: Shyam Sundar S K <Shyam-sundar.S-k@amd.com>
> > > > > [ck: split patches for nvme and pcie]
> > > > > Signed-off-by: Chaitanya Kulkarni <chaitanya.kulkarni@wdc.com>
> > > > > Suggested-by: Keith Busch <kbusch@kernel.org>
> > > > > Acked-by: Keith Busch <kbusch@kernel.org>
> > > > > ---
> > > > > Changes in v2:
> > > > > Fix the patch format and check chip root complex DID instead of
> > > > > PCIe RP to avoid the storage device plugged in internal PCIe RP by USB
> > adaptor.
> > > > >
> > > > > Changes in v3:
> > > > > According to Christoph Hellwig do NVME PCIe related identify opt
> > > > > better in PCIe quirk driver rather than in NVME module.
> > > > >
> > > > > Changes in v4:
> > > > > Split the fix to PCIe and NVMe part and then call the
> > > > > pci_dev_put() put the device reference count and finally refine the
> > commit info.
> > > > >
> > > > > Changes in v5:
> > > > > According to Christoph Hellwig and Keith Busch better use a
> > > > > passthrough device(bus) gloable flag to identify the NVMe shutdown
> > > > > opt rather than look up the device BDF.
> >
> > The changelog above bears little resemblance to reality.  I dug up all the
> > previous postings, hoping there was a hint about why this is relevant to PCI,
> > but I didn't find anything.  For the archives, here are the versions I found and
> > notes about what really changed:
> >
> >   v1 2021-04-14  2:18
> > https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Flore.
> > kernel.org%2Fr%2F1618366694-14092-1-git-send-email-
> > Prike.Liang%40amd.com%2F&amp;data=04%7C01%7CAlexander.Deucher%4
> > 0amd.com%7C7773f0f2af054b0ab45408d91bb08ad8%7C3dd8961fe4884e608e
> > 11a82d994e183d%7C0%7C0%7C637571267348438733%7CUnknown%7CTWFp
> > bGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVC
> > I6Mn0%3D%7C1000&amp;sdata=KFiTsmHEIbv9Qe7UI4yjKjXr1mWSusA2LYV
> > M07GUGAY%3D&amp;reserved=0
> >
> >   v2 2021-04-14  6:19
> > https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Flore.
> > kernel.org%2Fr%2F1618381200-14856-1-git-send-email-
> > Prike.Liang%40amd.com%2F&amp;data=04%7C01%7CAlexander.Deucher%4
> > 0amd.com%7C7773f0f2af054b0ab45408d91bb08ad8%7C3dd8961fe4884e608e
> > 11a82d994e183d%7C0%7C0%7C637571267348438733%7CUnknown%7CTWFp
> > bGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVC
> > I6Mn0%3D%7C1000&amp;sdata=I2rKN%2Bixmn2NcXcZ20seu9LQC%2BekwH
> > azH9oqzym7eeE%3D&amp;reserved=0
> >     (Not tagged as v2 in the posting.)
> >     Check result of pci_get_domain_bus_and_slot() for NULL.
> >
> >   v3 2021-04-14  8:18
> > https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Flore.
> > kernel.org%2Fr%2F1618388281-15629-1-git-send-email-
> > Prike.Liang%40amd.com%2F&amp;data=04%7C01%7CAlexander.Deucher%4
> > 0amd.com%7C7773f0f2af054b0ab45408d91bb08ad8%7C3dd8961fe4884e608e
> > 11a82d994e183d%7C0%7C0%7C637571267348438733%7CUnknown%7CTWFp
> > bGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVC
> > I6Mn0%3D%7C1000&amp;sdata=7T9VcWinB8XQTR4tVKMZ4WkCI5oGKBBuH
> > n33wxIZzbI%3D&amp;reserved=0
> >     (Not tagged as v3 in the posting.)
> >     Drop reference acquired by pci_get_domain_bus_and_slot().
> >     Return "true" (not NVME_QUIRK_SIMPLE_SUSPEND) from bool
> >     nvme_acpi_storage_d3().
> >
> >   v4 2021-04-15  3:52
> > https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Flore.
> > kernel.org%2Fr%2F1618458725-17164-1-git-send-email-
> > Prike.Liang%40amd.com%2F&amp;data=04%7C01%7CAlexander.Deucher%4
> > 0amd.com%7C7773f0f2af054b0ab45408d91bb08ad8%7C3dd8961fe4884e608e
> > 11a82d994e183d%7C0%7C0%7C637571267348438733%7CUnknown%7CTWFp
> > bGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVC
> > I6Mn0%3D%7C1000&amp;sdata=aTgZtIfWMeoWZ3HoHcjclrJPW8qknIsuoVaz
> > hZnoDHo%3D&amp;reserved=0
> >     Reorder Signed-off-by tags.
> >     No code change at all.
> >
> >   v5 2021-04-16  6:54
> > https://nam11.safelinks.protection.outlook.com/?url=https%3A%2F%2Flore.
> > kernel.org%2Fr%2F1618556075-24589-1-git-send-email-
> > Prike.Liang%40amd.com%2F&amp;data=04%7C01%7CAlexander.Deucher%4
> > 0amd.com%7C7773f0f2af054b0ab45408d91bb08ad8%7C3dd8961fe4884e608e
> > 11a82d994e183d%7C0%7C0%7C637571267348438733%7CUnknown%7CTWFp
> > bGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVC
> > I6Mn0%3D%7C1000&amp;sdata=5yzUyXfIDKxTE%2B%2FJy5fgOS9yCsU10uw
> > n4CjGd2mLLc0%3D&amp;reserved=0
> >     Move flag from pci_dev_flags to pci_bus_flags.
> >     Rename flag to PCI_BUS_FLAGS_DISABLE_ON_S2I.
> >     Inherit PCI_BUS_FLAGS_DISABLE_ON_S2I in all child buses of AMD
> >     0x1630.
> >     Check dev->bus->bus_flags instead of using
> >     pci_get_domain_bus_and_slot().
> >
> > > > > ---
> > > > >  drivers/pci/probe.c  | 5 ++++-
> > > > >  drivers/pci/quirks.c | 7 +++++++
> > > > >  include/linux/pci.h  | 2 ++
> > > > >  3 files changed, 13 insertions(+), 1 deletion(-)
> > > > >
> > > > > diff --git a/drivers/pci/probe.c b/drivers/pci/probe.c index
> > > > > 953f15a..34ba691e 100644
> > > > > --- a/drivers/pci/probe.c
> > > > > +++ b/drivers/pci/probe.c
> > > > > @@ -558,10 +558,13 @@ static struct pci_bus *pci_alloc_bus(struct
> > > > pci_bus *parent)
> > > > >     INIT_LIST_HEAD(&b->resources);
> > > > >     b->max_bus_speed = PCI_SPEED_UNKNOWN;
> > > > >     b->cur_bus_speed = PCI_SPEED_UNKNOWN;
> > > > > +   if (parent) {
> > > > >  #ifdef CONFIG_PCI_DOMAINS_GENERIC
> > > > > -   if (parent)
> > > > >             b->domain_nr = parent->domain_nr;  #endif
> > > > > +           if (parent->bus_flags & PCI_BUS_FLAGS_DISABLE_ON_S2I)
> > > > > +                   b->bus_flags |= PCI_BUS_FLAGS_DISABLE_ON_S2I;
> > > > > +   }
> > > > >     return b;
> > > > >  }
> > > > >
> > > > > diff --git a/drivers/pci/quirks.c b/drivers/pci/quirks.c index
> > > > > 653660e3..7c4bb8e 100644
> > > > > --- a/drivers/pci/quirks.c
> > > > > +++ b/drivers/pci/quirks.c
> > > > > @@ -312,6 +312,13 @@ static void quirk_nopciamd(struct pci_dev
> > > > > *dev) }  DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_AMD,
> > > >       PCI_DEVICE_ID_AMD_8151_0,       quirk_nopciamd);
> > > > >
> > > > > +static void quirk_amd_s2i_fixup(struct pci_dev *dev) {
> > > > > +   dev->bus->bus_flags |= PCI_BUS_FLAGS_DISABLE_ON_S2I;
> > > > > +   pci_info(dev, "AMD simple suspend opt enabled\n"); }
> > > > > +DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_AMD, 0x1630,
> > > > > +quirk_amd_s2i_fixup);
> > > > > +
> > > > >  /* Triton requires workarounds to be used by the drivers */
> > > > > static void quirk_triton(struct pci_dev *dev)  { diff --git
> > > > > a/include/linux/pci.h b/include/linux/pci.h index 53f4904..dc65219
> > > > > 100644
> > > > > --- a/include/linux/pci.h
> > > > > +++ b/include/linux/pci.h
> > > > > @@ -240,6 +240,8 @@ enum pci_bus_flags {
> > > > >     PCI_BUS_FLAGS_NO_MMRBC  = (__force pci_bus_flags_t) 2,
> > > > >     PCI_BUS_FLAGS_NO_AERSID = (__force pci_bus_flags_t) 4,
> > > > >     PCI_BUS_FLAGS_NO_EXTCFG = (__force pci_bus_flags_t) 8,
> > > > > +   /* Driver must pci_disable_device() for suspend-to-idle */
> > > > > +   PCI_BUS_FLAGS_DISABLE_ON_S2I    = (__force pci_bus_flags_t) 16,
> > > > >  };
> > > > >
> > > > >  /* Values from Link Status register, PCIe r3.1, sec 7.8.8 */
> > > > > --
> > > > > 2.7.4
> > > > >

_______________________________________________
Linux-nvme mailing list
Linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme

[PATCH v5 2/2] nvme-pci: add AMD PCIe quirk for simple suspend/resume

[Prike Liang]

In the NVMe controller default suspend-resume seems only save/restore the
NVMe link state by APST opt and the NVMe remains in D0 during this time.
Then the NVMe device will be shutdown by SMU firmware in the s2idle entry
and then will lost the NVMe power context during s2idle resume.Finally,
the NVMe command queue request will be processed abnormally and result
in access timeout.This issue can be settled by using PCIe power set with
simple suspend-resume process path instead of APST get/set opt.

This patch is updating the nvme_acpi_storage_d3() with previously
added quirk.

Cc: <stable@vger.kernel.org> # 5.10+
Signed-off-by: Prike Liang <Prike.Liang@amd.com>
Signed-off-by: Shyam Sundar S K <Shyam-sundar.S-k@amd.com>
[ck: split patches for nvme and pcie]
Signed-off-by: Chaitanya Kulkarni <chaitanya.kulkarni@wdc.com>
Suggested-by: Keith Busch <kbusch@kernel.org>
Acked-by: Keith Busch <kbusch@kernel.org>
---
Changes in v2:
Fix the patch format and check chip root complex DID instead of PCIe RP
to avoid the storage device plugged in internal PCIe RP by USB adaptor.

Changes in v3:
According to Christoph Hellwig do NVME PCIe related identify opt better in
PCIe quirk driver rather than in NVME module.

Changes in v4:
Split the fix to PCIe and NVMe part and then call the pci_dev_put() put
the device reference count and finally refine the commit info.

Changes in v5:
According to Christoph Hellwig and Keith Busch better use a passthrough device(bus)
gloable flag to identify the NVMe shutdown opt rather than look up the device BDF.
---
 drivers/nvme/host/pci.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/nvme/host/pci.c b/drivers/nvme/host/pci.c
index 6bad4d4..617256e 100644
--- a/drivers/nvme/host/pci.c
+++ b/drivers/nvme/host/pci.c
@@ -2836,6 +2836,8 @@ static bool nvme_acpi_storage_d3(struct pci_dev *dev)
 	acpi_status status;
 	u8 val;
 
+	if (dev->bus->bus_flags & PCI_BUS_FLAGS_DISABLE_ON_S2I)
+		return true;
 	/*
 	 * Look for _DSD property specifying that the storage device on the port
 	 * must use D3 to support deep platform power savings during
-- 
2.7.4


_______________________________________________
Linux-nvme mailing list
Linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme

Re: [PATCH v5 0/2] nvme-pci: add AMD PCIe quirk for NVMe simple suspend/resume

[Christoph Hellwig]

This looks ok to me now.

As a note to the people in AMD who came up with this scheme:  having
to shut down the controller every time s2i is entered not only generally
uses more power than using power states, but also means every time
s2i is entered the SSD needs to write to the media, which causes massive
endurance shortfalls.

On Tue, May 18, 2021 at 10:24:33AM +0800, Prike Liang wrote:
> Those patch series can handle NVMe can't suspend to D3 during s2idle
> entry on some AMD platform. In this case, can be settld by assigning and
> passing a PCIe bus flag to the armed device which need NVMe shutdown opt
> in s2idle suspend and then use PCIe power setting to put the NVMe device
> to D3.
> 
> Prike Liang (2):
>   PCI: add AMD PCIe quirk for nvme shutdown opt
>   nvme-pci: add AMD PCIe quirk for simple suspend/resume
> 
>  drivers/nvme/host/pci.c | 2 ++
>  drivers/pci/probe.c     | 5 ++++-
>  drivers/pci/quirks.c    | 7 +++++++
>  include/linux/pci.h     | 2 ++
>  4 files changed, 15 insertions(+), 1 deletion(-)
> 
> -- 
> 2.7.4
---end quoted text---

_______________________________________________
Linux-nvme mailing list
Linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme

[PATCH v5 0/2] nvme-pci: add AMD PCIe quirk for NVMe simple suspend/resume

[Prike Liang]

Those patch series can handle NVMe can't suspend to D3 during s2idle
entry on some AMD platform. In this case, can be settld by assigning and
passing a PCIe bus flag to the armed device which need NVMe shutdown opt
in s2idle suspend and then use PCIe power setting to put the NVMe device
to D3.

Prike Liang (2):
  PCI: add AMD PCIe quirk for nvme shutdown opt
  nvme-pci: add AMD PCIe quirk for simple suspend/resume

 drivers/nvme/host/pci.c | 2 ++
 drivers/pci/probe.c     | 5 ++++-
 drivers/pci/quirks.c    | 7 +++++++
 include/linux/pci.h     | 2 ++
 4 files changed, 15 insertions(+), 1 deletion(-)

-- 
2.7.4


_______________________________________________
Linux-nvme mailing list
Linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme

[PATCH v5 1/2] PCI: add AMD PCIe quirk for nvme shutdown opt

[Prike Liang]

In the NVMe controller default suspend-resume seems only save/restore the
NVMe link state by APST opt and the NVMe remains in D0 during this time.
Then the NVMe device will be shutdown by SMU firmware in the s2idle entry
and then will lost the NVMe power context during s2idle resume.Finally,
the NVMe command queue request will be processed abnormally and result
in access timeout.This issue can be settled by using PCIe power set with
simple suspend-resume process path instead of APST get/set opt.

In this patch prepare a PCIe RC bus flag to identify the platform whether
need the quirk.

Cc: <stable@vger.kernel.org> # 5.10+
Signed-off-by: Prike Liang <Prike.Liang@amd.com>
Signed-off-by: Shyam Sundar S K <Shyam-sundar.S-k@amd.com>
[ck: split patches for nvme and pcie]
Signed-off-by: Chaitanya Kulkarni <chaitanya.kulkarni@wdc.com>
Suggested-by: Keith Busch <kbusch@kernel.org>
Acked-by: Keith Busch <kbusch@kernel.org>
---
Changes in v2:
Fix the patch format and check chip root complex DID instead of PCIe RP
to avoid the storage device plugged in internal PCIe RP by USB adaptor.

Changes in v3:
According to Christoph Hellwig do NVME PCIe related identify opt better in
PCIe quirk driver rather than in NVME module.

Changes in v4:
Split the fix to PCIe and NVMe part and then call the pci_dev_put() put
the device reference count and finally refine the commit info.

Changes in v5:
According to Christoph Hellwig and Keith Busch better use a passthrough device(bus)
gloable flag to identify the NVMe shutdown opt rather than look up the device BDF.
---
 drivers/pci/probe.c  | 5 ++++-
 drivers/pci/quirks.c | 7 +++++++
 include/linux/pci.h  | 2 ++
 3 files changed, 13 insertions(+), 1 deletion(-)

diff --git a/drivers/pci/probe.c b/drivers/pci/probe.c
index 953f15a..34ba691e 100644
--- a/drivers/pci/probe.c
+++ b/drivers/pci/probe.c
@@ -558,10 +558,13 @@ static struct pci_bus *pci_alloc_bus(struct pci_bus *parent)
 	INIT_LIST_HEAD(&b->resources);
 	b->max_bus_speed = PCI_SPEED_UNKNOWN;
 	b->cur_bus_speed = PCI_SPEED_UNKNOWN;
+	if (parent) {
 #ifdef CONFIG_PCI_DOMAINS_GENERIC
-	if (parent)
 		b->domain_nr = parent->domain_nr;
 #endif
+		if (parent->bus_flags & PCI_BUS_FLAGS_DISABLE_ON_S2I)
+			b->bus_flags |= PCI_BUS_FLAGS_DISABLE_ON_S2I;
+	}
 	return b;
 }
 
diff --git a/drivers/pci/quirks.c b/drivers/pci/quirks.c
index 653660e3..7c4bb8e 100644
--- a/drivers/pci/quirks.c
+++ b/drivers/pci/quirks.c
@@ -312,6 +312,13 @@ static void quirk_nopciamd(struct pci_dev *dev)
 }
 DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_AMD,	PCI_DEVICE_ID_AMD_8151_0,	quirk_nopciamd);
 
+static void quirk_amd_s2i_fixup(struct pci_dev *dev)
+{
+	dev->bus->bus_flags |= PCI_BUS_FLAGS_DISABLE_ON_S2I;
+	pci_info(dev, "AMD simple suspend opt enabled\n");
+}
+DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_AMD, 0x1630, quirk_amd_s2i_fixup);
+
 /* Triton requires workarounds to be used by the drivers */
 static void quirk_triton(struct pci_dev *dev)
 {
diff --git a/include/linux/pci.h b/include/linux/pci.h
index 53f4904..dc65219 100644
--- a/include/linux/pci.h
+++ b/include/linux/pci.h
@@ -240,6 +240,8 @@ enum pci_bus_flags {
 	PCI_BUS_FLAGS_NO_MMRBC	= (__force pci_bus_flags_t) 2,
 	PCI_BUS_FLAGS_NO_AERSID	= (__force pci_bus_flags_t) 4,
 	PCI_BUS_FLAGS_NO_EXTCFG	= (__force pci_bus_flags_t) 8,
+	/* Driver must pci_disable_device() for suspend-to-idle */
+	PCI_BUS_FLAGS_DISABLE_ON_S2I	= (__force pci_bus_flags_t) 16,
 };
 
 /* Values from Link Status register, PCIe r3.1, sec 7.8.8 */
-- 
2.7.4


_______________________________________________
Linux-nvme mailing list
Linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme

RE: [PATCH v5 1/2] PCI: add AMD PCIe quirk for nvme shutdown opt

[Liang, Prike]

[AMD Public Use]

+ linux-pci for further review.

Thanks,
> -----Original Message-----
> From: Liang, Prike <Prike.Liang@amd.com>
> Sent: Thursday, April 22, 2021 9:19 AM
> To: linux-nvme@lists.infradead.org; kbusch@kernel.org; hch@infradead.org;
> Chaitanya.Kulkarni@wdc.com; gregkh@linuxfoundation.org
> Cc: stable@vger.kernel.org; Deucher, Alexander
> <Alexander.Deucher@amd.com>; S-k, Shyam-sundar <Shyam-sundar.S-
> k@amd.com>; Liang, Prike <Prike.Liang@amd.com>; Chaitanya Kulkarni
> <chaitanya.kulkarni@wdc.com>
> Subject: [PATCH v5 1/2] PCI: add AMD PCIe quirk for nvme shutdown opt
>
> In the NVMe controller default suspend-resume seems only save/restore the
> NVMe link state by APST opt and the NVMe remains in D0 during this time.
> Then the NVMe device will be shutdown by SMU firmware in the s2idle entry
> and then will lost the NVMe power context during s2idle resume.Finally, the
> NVMe command queue request will be processed abnormally and result in
> access timeout.This issue can be settled by using PCIe power set with simple
> suspend-resume process path instead of APST get/set opt.
>
> In this patch prepare a PCIe RC bus flag to identify the platform whether
> need the quirk.
>
> Cc: <stable@vger.kernel.org> # 5.10+
> Signed-off-by: Prike Liang <Prike.Liang@amd.com>
> Signed-off-by: Shyam Sundar S K <Shyam-sundar.S-k@amd.com>
> [ck: split patches for nvme and pcie]
> Signed-off-by: Chaitanya Kulkarni <chaitanya.kulkarni@wdc.com>
> Suggested-by: Keith Busch <kbusch@kernel.org>
> Acked-by: Keith Busch <kbusch@kernel.org>
> ---
> Changes in v2:
> Fix the patch format and check chip root complex DID instead of PCIe RP to
> avoid the storage device plugged in internal PCIe RP by USB adaptor.
>
> Changes in v3:
> According to Christoph Hellwig do NVME PCIe related identify opt better in
> PCIe quirk driver rather than in NVME module.
>
> Changes in v4:
> Split the fix to PCIe and NVMe part and then call the pci_dev_put() put the
> device reference count and finally refine the commit info.
>
> Changes in v5:
> According to Christoph Hellwig and Keith Busch better use a passthrough
> device(bus) gloable flag to identify the NVMe shutdown opt rather than look
> up the device BDF.
> ---
>  drivers/pci/probe.c  | 5 ++++-
>  drivers/pci/quirks.c | 7 +++++++
>  include/linux/pci.h  | 2 ++
>  3 files changed, 13 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/pci/probe.c b/drivers/pci/probe.c index
> 953f15a..34ba691e 100644
> --- a/drivers/pci/probe.c
> +++ b/drivers/pci/probe.c
> @@ -558,10 +558,13 @@ static struct pci_bus *pci_alloc_bus(struct pci_bus
> *parent)
>  INIT_LIST_HEAD(&b->resources);
>  b->max_bus_speed = PCI_SPEED_UNKNOWN;
>  b->cur_bus_speed = PCI_SPEED_UNKNOWN;
> +if (parent) {
>  #ifdef CONFIG_PCI_DOMAINS_GENERIC
> -if (parent)
>  b->domain_nr = parent->domain_nr;
>  #endif
> +if (parent->bus_flags & PCI_BUS_FLAGS_DISABLE_ON_S2I)
> +b->bus_flags |= PCI_BUS_FLAGS_DISABLE_ON_S2I;
> +}
>  return b;
>  }
>
> diff --git a/drivers/pci/quirks.c b/drivers/pci/quirks.c index
> 653660e3..7c4bb8e 100644
> --- a/drivers/pci/quirks.c
> +++ b/drivers/pci/quirks.c
> @@ -312,6 +312,13 @@ static void quirk_nopciamd(struct pci_dev *dev)  }
>  DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_AMD,
> PCI_DEVICE_ID_AMD_8151_0,quirk_nopciamd);
>
> +static void quirk_amd_s2i_fixup(struct pci_dev *dev) {
> +dev->bus->bus_flags |= PCI_BUS_FLAGS_DISABLE_ON_S2I;
> +pci_info(dev, "AMD simple suspend opt enabled\n"); }
> +DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_AMD, 0x1630,
> +quirk_amd_s2i_fixup);
> +
>  /* Triton requires workarounds to be used by the drivers */  static void
> quirk_triton(struct pci_dev *dev)  { diff --git a/include/linux/pci.h
> b/include/linux/pci.h index 53f4904..dc65219 100644
> --- a/include/linux/pci.h
> +++ b/include/linux/pci.h
> @@ -240,6 +240,8 @@ enum pci_bus_flags {
>  PCI_BUS_FLAGS_NO_MMRBC= (__force pci_bus_flags_t) 2,
>  PCI_BUS_FLAGS_NO_AERSID= (__force pci_bus_flags_t) 4,
>  PCI_BUS_FLAGS_NO_EXTCFG= (__force pci_bus_flags_t) 8,
> +/* Driver must pci_disable_device() for suspend-to-idle */
> +PCI_BUS_FLAGS_DISABLE_ON_S2I= (__force pci_bus_flags_t) 16,
>  };
>
>  /* Values from Link Status register, PCIe r3.1, sec 7.8.8 */
> --
> 2.7.4


_______________________________________________
Linux-nvme mailing list
Linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme

RE: [PATCH v5 1/2] PCI: add AMD PCIe quirk for nvme shutdown opt

[Liang, Prike]

[AMD Public Use]

Ping... Could I get some comments or RB on this latest series.


Thanks,
Prike
> -----Original Message-----
> From: Liang, Prike
> Sent: Thursday, May 6, 2021 11:08 AM
> To: linux-nvme@lists.infradead.org; kbusch@kernel.org; hch@infradead.org;
> Chaitanya.Kulkarni@wdc.com; gregkh@linuxfoundation.org; linux-
> pci@vger.kernel.org; Bjorn Helgaas <helgaas@kernel.org>
> Cc: stable@vger.kernel.org; Deucher, Alexander
> <Alexander.Deucher@amd.com>; S-k, Shyam-sundar <Shyam-sundar.S-
> k@amd.com>; Chaitanya Kulkarni <chaitanya.kulkarni@wdc.com>;
> rjw@rjwysocki.net
> Subject: RE: [PATCH v5 1/2] PCI: add AMD PCIe quirk for nvme shutdown opt
>
> + linux-pci for further review.
>
> Thanks,
> > -----Original Message-----
> > From: Liang, Prike <Prike.Liang@amd.com>
> > Sent: Thursday, April 22, 2021 9:19 AM
> > To: linux-nvme@lists.infradead.org; kbusch@kernel.org;
> > hch@infradead.org; Chaitanya.Kulkarni@wdc.com;
> > gregkh@linuxfoundation.org
> > Cc: stable@vger.kernel.org; Deucher, Alexander
> > <Alexander.Deucher@amd.com>; S-k, Shyam-sundar <Shyam-sundar.S-
> > k@amd.com>; Liang, Prike <Prike.Liang@amd.com>; Chaitanya Kulkarni
> > <chaitanya.kulkarni@wdc.com>
> > Subject: [PATCH v5 1/2] PCI: add AMD PCIe quirk for nvme shutdown opt
> >
> > In the NVMe controller default suspend-resume seems only save/restore
> > the NVMe link state by APST opt and the NVMe remains in D0 during this
> time.
> > Then the NVMe device will be shutdown by SMU firmware in the s2idle
> > entry and then will lost the NVMe power context during s2idle
> > resume.Finally, the NVMe command queue request will be processed
> > abnormally and result in access timeout.This issue can be settled by
> > using PCIe power set with simple suspend-resume process path instead of
> APST get/set opt.
> >
> > In this patch prepare a PCIe RC bus flag to identify the platform
> > whether need the quirk.
> >
> > Cc: <stable@vger.kernel.org> # 5.10+
> > Signed-off-by: Prike Liang <Prike.Liang@amd.com>
> > Signed-off-by: Shyam Sundar S K <Shyam-sundar.S-k@amd.com>
> > [ck: split patches for nvme and pcie]
> > Signed-off-by: Chaitanya Kulkarni <chaitanya.kulkarni@wdc.com>
> > Suggested-by: Keith Busch <kbusch@kernel.org>
> > Acked-by: Keith Busch <kbusch@kernel.org>
> > ---
> > Changes in v2:
> > Fix the patch format and check chip root complex DID instead of PCIe
> > RP to avoid the storage device plugged in internal PCIe RP by USB adaptor.
> >
> > Changes in v3:
> > According to Christoph Hellwig do NVME PCIe related identify opt
> > better in PCIe quirk driver rather than in NVME module.
> >
> > Changes in v4:
> > Split the fix to PCIe and NVMe part and then call the pci_dev_put()
> > put the device reference count and finally refine the commit info.
> >
> > Changes in v5:
> > According to Christoph Hellwig and Keith Busch better use a
> > passthrough
> > device(bus) gloable flag to identify the NVMe shutdown opt rather than
> > look up the device BDF.
> > ---
> >  drivers/pci/probe.c  | 5 ++++-
> >  drivers/pci/quirks.c | 7 +++++++
> >  include/linux/pci.h  | 2 ++
> >  3 files changed, 13 insertions(+), 1 deletion(-)
> >
> > diff --git a/drivers/pci/probe.c b/drivers/pci/probe.c index
> > 953f15a..34ba691e 100644
> > --- a/drivers/pci/probe.c
> > +++ b/drivers/pci/probe.c
> > @@ -558,10 +558,13 @@ static struct pci_bus *pci_alloc_bus(struct
> > pci_bus
> > *parent)
> >     INIT_LIST_HEAD(&b->resources);
> >     b->max_bus_speed = PCI_SPEED_UNKNOWN;
> >     b->cur_bus_speed = PCI_SPEED_UNKNOWN;
> > +   if (parent) {
> >  #ifdef CONFIG_PCI_DOMAINS_GENERIC
> > -   if (parent)
> >             b->domain_nr = parent->domain_nr;
> >  #endif
> > +           if (parent->bus_flags & PCI_BUS_FLAGS_DISABLE_ON_S2I)
> > +                   b->bus_flags |= PCI_BUS_FLAGS_DISABLE_ON_S2I;
> > +   }
> >     return b;
> >  }
> >
> > diff --git a/drivers/pci/quirks.c b/drivers/pci/quirks.c index
> > 653660e3..7c4bb8e 100644
> > --- a/drivers/pci/quirks.c
> > +++ b/drivers/pci/quirks.c
> > @@ -312,6 +312,13 @@ static void quirk_nopciamd(struct pci_dev *dev)
> > }  DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_AMD,
> >     PCI_DEVICE_ID_AMD_8151_0,       quirk_nopciamd);
> >
> > +static void quirk_amd_s2i_fixup(struct pci_dev *dev) {
> > +   dev->bus->bus_flags |= PCI_BUS_FLAGS_DISABLE_ON_S2I;
> > +   pci_info(dev, "AMD simple suspend opt enabled\n"); }
> > +DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_AMD, 0x1630,
> > +quirk_amd_s2i_fixup);
> > +
> >  /* Triton requires workarounds to be used by the drivers */  static
> > void quirk_triton(struct pci_dev *dev)  { diff --git
> > a/include/linux/pci.h b/include/linux/pci.h index 53f4904..dc65219
> > 100644
> > --- a/include/linux/pci.h
> > +++ b/include/linux/pci.h
> > @@ -240,6 +240,8 @@ enum pci_bus_flags {
> >     PCI_BUS_FLAGS_NO_MMRBC  = (__force pci_bus_flags_t) 2,
> >     PCI_BUS_FLAGS_NO_AERSID = (__force pci_bus_flags_t) 4,
> >     PCI_BUS_FLAGS_NO_EXTCFG = (__force pci_bus_flags_t) 8,
> > +   /* Driver must pci_disable_device() for suspend-to-idle */
> > +   PCI_BUS_FLAGS_DISABLE_ON_S2I    = (__force pci_bus_flags_t) 16,
> >  };
> >
> >  /* Values from Link Status register, PCIe r3.1, sec 7.8.8 */
> > --
> > 2.7.4


_______________________________________________
Linux-nvme mailing list
Linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme

[PATCH v5 0/2] nvme-pci: add AMD PCIe quirk for NVMe simple suspend/resume

[Prike Liang]

Those patch series can handle NVMe can't suspend to D3 during s2idle
entry on some AMD platform. In this case, can be settld by assigning and
passing a PCIe bus flag to the armed device which need NVMe shutdown opt
in s2idle suspend and then use PCIe power setting to put the NVMe device
to D3.

Prike Liang (2):
  PCI: add AMD PCIe quirk for nvme shutdown opt
  nvme-pci: add AMD PCIe quirk for simple suspend/resume

 drivers/nvme/host/pci.c | 2 ++
 drivers/pci/probe.c     | 5 ++++-
 drivers/pci/quirks.c    | 7 +++++++
 include/linux/pci.h     | 2 ++
 4 files changed, 15 insertions(+), 1 deletion(-)

-- 
2.7.4


_______________________________________________
Linux-nvme mailing list
Linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme

[PATCH v5 1/2] PCI: add AMD PCIe quirk for nvme shutdown opt

[Prike Liang]

In the NVMe controller default suspend-resume seems only save/restore the
NVMe link state by APST opt and the NVMe remains in D0 during this time.
Then the NVMe device will be shutdown by SMU firmware in the s2idle entry
and then will lost the NVMe power context during s2idle resume.Finally,
the NVMe command queue request will be processed abnormally and result
in access timeout.This issue can be settled by using PCIe power set with
simple suspend-resume process path instead of APST get/set opt.

In this patch prepare a PCIe RC bus flag to identify the platform whether
need the quirk.

Cc: <stable@vger.kernel.org> # 5.11+
Signed-off-by: Prike Liang <Prike.Liang@amd.com>
Signed-off-by: Shyam Sundar S K <Shyam-sundar.S-k@amd.com>
[ck: split patches for nvme and pcie]
Signed-off-by: Chaitanya Kulkarni <chaitanya.kulkarni@wdc.com>
Signed-off-by: Keith Busch <kbusch@kernel.org>
---
Changes in v2:
Fix the patch format and check chip root complex DID instead of PCIe RP
to avoid the storage device plugged in internal PCIe RP by USB adaptor.

Changes in v3:
According to Christoph Hellwig do NVME PCIe related identify opt better in
PCIe quirk driver rather than in NVME module.

Changes in v4:
Split the fix to PCIe and NVMe part and then call the pci_dev_put() put
the device reference count and finally refine the commit info.

Changes in v5:
According to Christoph Hellwig and Keith Busch better use a passthrough device(bus)
gloable flag to identify the NVMe shutdown opt rather than look up the device BDF.
---
 drivers/pci/probe.c  | 5 ++++-
 drivers/pci/quirks.c | 7 +++++++
 include/linux/pci.h  | 2 ++
 3 files changed, 13 insertions(+), 1 deletion(-)

diff --git a/drivers/pci/probe.c b/drivers/pci/probe.c
index 953f15a..34ba691e 100644
--- a/drivers/pci/probe.c
+++ b/drivers/pci/probe.c
@@ -558,10 +558,13 @@ static struct pci_bus *pci_alloc_bus(struct pci_bus *parent)
 	INIT_LIST_HEAD(&b->resources);
 	b->max_bus_speed = PCI_SPEED_UNKNOWN;
 	b->cur_bus_speed = PCI_SPEED_UNKNOWN;
+	if (parent) {
 #ifdef CONFIG_PCI_DOMAINS_GENERIC
-	if (parent)
 		b->domain_nr = parent->domain_nr;
 #endif
+		if (parent->bus_flags & PCI_BUS_FLAGS_DISABLE_ON_S2I)
+			b->bus_flags |= PCI_BUS_FLAGS_DISABLE_ON_S2I;
+	}
 	return b;
 }
 
diff --git a/drivers/pci/quirks.c b/drivers/pci/quirks.c
index 653660e3..7c4bb8e 100644
--- a/drivers/pci/quirks.c
+++ b/drivers/pci/quirks.c
@@ -312,6 +312,13 @@ static void quirk_nopciamd(struct pci_dev *dev)
 }
 DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_AMD,	PCI_DEVICE_ID_AMD_8151_0,	quirk_nopciamd);
 
+static void quirk_amd_s2i_fixup(struct pci_dev *dev)
+{
+	dev->bus->bus_flags |= PCI_BUS_FLAGS_DISABLE_ON_S2I;
+	pci_info(dev, "AMD simple suspend opt enabled\n");
+}
+DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_AMD, 0x1630, quirk_amd_s2i_fixup);
+
 /* Triton requires workarounds to be used by the drivers */
 static void quirk_triton(struct pci_dev *dev)
 {
diff --git a/include/linux/pci.h b/include/linux/pci.h
index 53f4904..dc65219 100644
--- a/include/linux/pci.h
+++ b/include/linux/pci.h
@@ -240,6 +240,8 @@ enum pci_bus_flags {
 	PCI_BUS_FLAGS_NO_MMRBC	= (__force pci_bus_flags_t) 2,
 	PCI_BUS_FLAGS_NO_AERSID	= (__force pci_bus_flags_t) 4,
 	PCI_BUS_FLAGS_NO_EXTCFG	= (__force pci_bus_flags_t) 8,
+	/* Driver must pci_disable_device() for suspend-to-idle */
+	PCI_BUS_FLAGS_DISABLE_ON_S2I	= (__force pci_bus_flags_t) 16,
 };
 
 /* Values from Link Status register, PCIe r3.1, sec 7.8.8 */
-- 
2.7.4


_______________________________________________
Linux-nvme mailing list
Linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme

Re: [PATCH v5 1/2] PCI: add AMD PCIe quirk for nvme shutdown opt

[Keith Busch]

On Fri, Apr 16, 2021 at 02:54:34PM +0800, Prike Liang wrote:
> In the NVMe controller default suspend-resume seems only save/restore the
> NVMe link state by APST opt and the NVMe remains in D0 during this time.
> Then the NVMe device will be shutdown by SMU firmware in the s2idle entry
> and then will lost the NVMe power context during s2idle resume.Finally,
> the NVMe command queue request will be processed abnormally and result
> in access timeout.This issue can be settled by using PCIe power set with
> simple suspend-resume process path instead of APST get/set opt.
> 
> In this patch prepare a PCIe RC bus flag to identify the platform whether
> need the quirk.
> 
> Cc: <stable@vger.kernel.org> # 5.11+
> Signed-off-by: Prike Liang <Prike.Liang@amd.com>
> Signed-off-by: Shyam Sundar S K <Shyam-sundar.S-k@amd.com>
> [ck: split patches for nvme and pcie]
> Signed-off-by: Chaitanya Kulkarni <chaitanya.kulkarni@wdc.com>
> Signed-off-by: Keith Busch <kbusch@kernel.org>

Just a "Suggested-by:" from me is fine. I'm glad you were able to
confirm this is successful, so I can add my Ack as well

Acked-by: Keith Busch <kbusch@kernel.org>

_______________________________________________
Linux-nvme mailing list
Linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme