[patchV2 0/2] shpchp: no claim on pcie port device

[patchV2 0/2] shpchp: no claim on pcie port device

[Pingfan Liu]

The Linux Device Driver Model allows a physical device to be handled
by only a single driver. But at present, both shpchp and portdrv_pci
claim PCI_CLASS_BRIDGE_PCI. This cause problems, such as the wrong
shutdown seq. This series keeps shpchp driver away from pcie port device.

V1 -> V2:
implement the checking in the process of matching, instead of probing,
which is too late.

Pingfan Liu (2):
  pci: introduce an extra method for matching in pci_driver
  pci/shpchp: no claim on pcie port device

 drivers/pci/hotplug/shpchp_core.c | 13 +++++++++++++
 drivers/pci/pci-driver.c          |  4 ++++
 include/linux/pci.h               |  2 ++
 3 files changed, 19 insertions(+)

-- 
2.7.4

[patchV2 1/2] pci: introduce an extra method for matching in pci_driver

[Pingfan Liu]

In __driver_attach(), if a driver matches a device, the device will
be appended to the tail of devices_kset, no matter what the probing
result of the device. Hence in order to prevent a driver to append
a probed device to devices_kset, it requires a correct matching.

As for pci, pci driver uses pci_device_id to match a device. But it may
be not enough, since there is need for extra info such as pcie, which
can not be provided in pci_device_id. Such info is driver specific, and
this patch introduces a new method "extra_match" in pci_driver.
This is used by next patch.

Signed-off-by: Pingfan Liu <kernelfans@gmail.com>
---
 drivers/pci/pci-driver.c | 4 ++++
 include/linux/pci.h      | 2 ++
 2 files changed, 6 insertions(+)

diff --git a/drivers/pci/pci-driver.c b/drivers/pci/pci-driver.c
index b9a1311..151865f 100644
--- a/drivers/pci/pci-driver.c
+++ b/drivers/pci/pci-driver.c
@@ -273,6 +273,10 @@ static const struct pci_device_id *pci_match_device(struct pci_driver *drv,
 	if (!found_id && dev->driver_override)
 		found_id = &pci_device_id_any;
 
+	/* more strictly matching besides id_table */
+	if (drv->extra_match && drv->extra_match(dev) < 0)
+		found_id = NULL;
+
 	return found_id;
 }
 
diff --git a/include/linux/pci.h b/include/linux/pci.h
index 73178a2..6ed960d 100644
--- a/include/linux/pci.h
+++ b/include/linux/pci.h
@@ -738,10 +738,12 @@ struct pci_error_handlers {
 
 
 struct module;
+typedef int (*ematch)(struct pci_dev *pdev);
 struct pci_driver {
 	struct list_head	node;
 	const char		*name;
 	const struct pci_device_id *id_table;	/* Must be non-NULL for probe to be called */
+	ematch extra_match;	/* more strictly matching besides id_table */
 	int  (*probe)(struct pci_dev *dev, const struct pci_device_id *id);	/* New device inserted */
 	void (*remove)(struct pci_dev *dev);	/* Device removed (NULL if not a hot-plug capable driver) */
 	int  (*suspend)(struct pci_dev *dev, pm_message_t state);	/* Device suspended */
-- 
2.7.4

[patchV2 2/2] pci/shpchp: no claim on pcie port device

[Pingfan Liu]

The Linux Device Driver Model allows a physical device to be handled
by only a single driver. But at present, both shpchp and portdrv_pci
claim PCI_CLASS_BRIDGE_PCI, and touch devices_kset when the drivers are
loaded. This causes a few problems, one is the wrong shutdown seq of
devices, owing to the broken devices_kset. This patch keeps shpchp away
from pcie port device, using the extra matching method.

Note:
I hit this bug on a Power9 machine, when "kexec -e", and see a ata-disk
behind a bridge can not write back buffer in flight due to the former
shutdown of the bridge which clears the BusMaster bit.

the device involved:
0004:00:00.0 PCI bridge: IBM Device 04c1 (prog-if 00 [Normal decode])
        Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR+ FastB2B- DisINTx-
        Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort- <TAbort- <MAbort- >SERR- <PERR- INTx-
        Latency: 0
        NUMA node: 0
        Bus: primary=00, secondary=01, subordinate=12, sec-latency=0
        I/O behind bridge: 00000000-00000fff
        Memory behind bridge: 80000000-ffefffff
        Prefetchable memory behind bridge: 0006024000000000-0006027f7fffffff
        Secondary status: 66MHz- FastB2B- ParErr- DEVSEL=fast >TAbort- <TAbort- <MAbort- <SERR- <PERR-
        BridgeCtl: Parity- SERR+ NoISA- VGA- MAbort- >Reset- FastB2B-
                PriDiscTmr- SecDiscTmr- DiscTmrStat- DiscTmrSERREn-
        Capabilities: [40] Power Management version 3
                Flags: PMEClk- DSI- D1- D2- AuxCurrent=0mA PME(D0+,D1-,D2-,D3hot+,D3cold+)
                Status: D0 NoSoftRst- PME-Enable- DSel=0 DScale=0 PME-
        Capabilities: [48] Express (v2) Root Port (Slot-), MSI 00
                DevCap: MaxPayload 512 bytes, PhantFunc 0
                        ExtTag- RBE+
                DevCtl: Report errors: Correctable- Non-Fatal- Fatal- Unsupported-
                        RlxdOrd- ExtTag- PhantFunc- AuxPwr- NoSnoop-
                        MaxPayload 256 bytes, MaxReadReq 128 bytes
                DevSta: CorrErr- UncorrErr- FatalErr- UnsuppReq- AuxPwr- TransPend-
                LnkCap: Port #0, Speed 8GT/s, Width x8, ASPM not supported, Exit Latency L0s <64ns, L1 <1us
                        ClockPM- Surprise- LLActRep+ BwNot+ ASPMOptComp-
                LnkCtl: ASPM Disabled; RCB 128 bytes Disabled- CommClk-
                        ExtSynch- ClockPM- AutWidDis- BWInt- AutBWInt-
                LnkSta: Speed 8GT/s, Width x4, TrErr- Train- SlotClk- DLActive+ BWMgmt- ABWMgmt+
                RootCtl: ErrCorrectable- ErrNon-Fatal- ErrFatal- PMEIntEna- CRSVisible-
                RootCap: CRSVisible-
                RootSta: PME ReqID 0000, PMEStatus- PMEPending-
                DevCap2: Completion Timeout: Range ABCD, TimeoutDis+, LTR-, OBFF Not Supported ARIFwd+
                DevCtl2: Completion Timeout: 16ms to 55ms, TimeoutDis+, LTR-, OBFF Disabled ARIFwd+
                LnkCtl2: Target Link Speed: 8GT/s, EnterCompliance- SpeedDis-
                         Transmit Margin: Normal Operating Range, EnterModifiedCompliance- ComplianceSOS-
                         Compliance De-emphasis: -6dB
                LnkSta2: Current De-emphasis Level: -3.5dB, EqualizationComplete+, EqualizationPhase1+
                         EqualizationPhase2+, EqualizationPhase3+, LinkEqualizationRequest-
        Capabilities: [100 v1] Advanced Error Reporting
                UESta:  DLP- SDES- TLP- FCP- CmpltTO- CmpltAbrt- UnxCmplt- RxOF- MalfTLP- ECRC- UnsupReq- ACSViol-
                UEMsk:  DLP- SDES- TLP+ FCP- CmpltTO+ CmpltAbrt+ UnxCmplt- RxOF- MalfTLP- ECRC+ UnsupReq- ACSViol-
                UESvrt: DLP- SDES- TLP- FCP- CmpltTO- CmpltAbrt- UnxCmplt- RxOF- MalfTLP- ECRC- UnsupReq- ACSViol-
                CESta:  RxErr- BadTLP- BadDLLP- Rollover- Timeout- NonFatalErr-
                CEMsk:  RxErr- BadTLP- BadDLLP- Rollover- Timeout- NonFatalErr-
                AERCap: First Error Pointer: 00, GenCap+ CGenEn+ ChkCap+ ChkEn+
        Capabilities: [148 v1] #19

Signed-off-by: Pingfan Liu <kernelfans@gmail.com>
---
 drivers/pci/hotplug/shpchp_core.c | 13 +++++++++++++
 1 file changed, 13 insertions(+)

diff --git a/drivers/pci/hotplug/shpchp_core.c b/drivers/pci/hotplug/shpchp_core.c
index 1f0f969..85b4665 100644
--- a/drivers/pci/hotplug/shpchp_core.c
+++ b/drivers/pci/hotplug/shpchp_core.c
@@ -336,6 +336,18 @@ static void shpc_remove(struct pci_dev *dev)
 	kfree(ctrl);
 }
 
+static int shpc_extra_match(struct pci_dev *pdev)
+{
+	/* do not claim pcie port device */
+	if (pci_is_pcie(pdev) &&
+	    (pci_pcie_type(pdev) == PCI_EXP_TYPE_ROOT_PORT ||
+	     pci_pcie_type(pdev) == PCI_EXP_TYPE_UPSTREAM ||
+	     pci_pcie_type(pdev) == PCI_EXP_TYPE_DOWNSTREAM))
+		return -ENODEV;
+
+	return 0;
+}
+
 static const struct pci_device_id shpcd_pci_tbl[] = {
 	{PCI_DEVICE_CLASS(((PCI_CLASS_BRIDGE_PCI << 8) | 0x00), ~0)},
 	{ /* end: all zeroes */ }
@@ -345,6 +357,7 @@ MODULE_DEVICE_TABLE(pci, shpcd_pci_tbl);
 static struct pci_driver shpc_driver = {
 	.name =		SHPC_MODULE_NAME,
 	.id_table =	shpcd_pci_tbl,
+	.extra_match =	shpc_extra_match,
 	.probe =	shpc_probe,
 	.remove =	shpc_remove,
 };
-- 
2.7.4

Re: [patchV2 1/2] pci: introduce an extra method for matching in pci_driver

[Christoph Hellwig]

If we successfull call ->probe twice on a device we have a much
deeper problem than we can fix in the PCI layer or shpchp.

Please explain the actual race or other condition that can lead to
this and report it to the driver core maintainer.

Re: [patchV2 1/2] pci: introduce an extra method for matching in pci_driver

[Pingfan Liu]

On Wed, Jun 13, 2018 at 2:37 PM Christoph Hellwig <hch@infradead.org> wrote:
>
> If we successfull call ->probe twice on a device we have a much
> deeper problem than we can fix in the PCI layer or shpchp.
>
OK, see your point.

> Please explain the actual race or other condition that can lead to
> this and report it to the driver core maintainer.
Oh, then it should be. I will add more debug info to narrow down the
problem, and report it to driver core maintainer.

Thanks for your kindly review.

Regards,
Pingfan

Re: [patchV2 2/2] pci/shpchp: no claim on pcie port device

[Bjorn Helgaas]

On Wed, Jun 13, 2018 at 02:29:57PM +0800, Pingfan Liu wrote:
> The Linux Device Driver Model allows a physical device to be handled
> by only a single driver. But at present, both shpchp and portdrv_pci
> claim PCI_CLASS_BRIDGE_PCI, and touch devices_kset when the drivers are
> loaded. This causes a few problems, one is the wrong shutdown seq of
> devices, owing to the broken devices_kset. This patch keeps shpchp away
> from pcie port device, using the extra matching method.

As Christoph pointed out, something seems wrong if we add to
devices_kset even when the probe fails.  We will call shpc_probe() for
the device below, but there's no SHPC capability, so the probe should
fail when shpc_init() fails.

I do think the current structure where portdrv and shpchp are both
"drivers" that claim bridges is broken.  This has caused problems
before, e.g., some PCIe switches have performance counters, and a
driver for those switches ought to be able to claim the device and use
the counters, but portdrv is in the way.

I think it would be better if portdrv (and maybe shpchp; not sure
about that) were integrated into the PCIe core and did not register as
a driver.  But this is a big project, far beyond the scope of this
current issue.

> Note:
> I hit this bug on a Power9 machine, when "kexec -e", and see a ata-disk
> behind a bridge can not write back buffer in flight due to the former
> shutdown of the bridge which clears the BusMaster bit.
> 
> the device involved:
> 0004:00:00.0 PCI bridge: IBM Device 04c1 (prog-if 00 [Normal decode])
>         Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR+ FastB2B- DisINTx-
>         Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort- <TAbort- <MAbort- >SERR- <PERR- INTx-
>         Latency: 0
>         NUMA node: 0
>         Bus: primary=00, secondary=01, subordinate=12, sec-latency=0
>         I/O behind bridge: 00000000-00000fff
>         Memory behind bridge: 80000000-ffefffff
>         Prefetchable memory behind bridge: 0006024000000000-0006027f7fffffff
>         Secondary status: 66MHz- FastB2B- ParErr- DEVSEL=fast >TAbort- <TAbort- <MAbort- <SERR- <PERR-
>         BridgeCtl: Parity- SERR+ NoISA- VGA- MAbort- >Reset- FastB2B-
>                 PriDiscTmr- SecDiscTmr- DiscTmrStat- DiscTmrSERREn-
>         Capabilities: [40] Power Management version 3
>                 Flags: PMEClk- DSI- D1- D2- AuxCurrent=0mA PME(D0+,D1-,D2-,D3hot+,D3cold+)
>                 Status: D0 NoSoftRst- PME-Enable- DSel=0 DScale=0 PME-
>         Capabilities: [48] Express (v2) Root Port (Slot-), MSI 00
>                 DevCap: MaxPayload 512 bytes, PhantFunc 0
>                         ExtTag- RBE+
>                 DevCtl: Report errors: Correctable- Non-Fatal- Fatal- Unsupported-
>                         RlxdOrd- ExtTag- PhantFunc- AuxPwr- NoSnoop-
>                         MaxPayload 256 bytes, MaxReadReq 128 bytes
>                 DevSta: CorrErr- UncorrErr- FatalErr- UnsuppReq- AuxPwr- TransPend-
>                 LnkCap: Port #0, Speed 8GT/s, Width x8, ASPM not supported, Exit Latency L0s <64ns, L1 <1us
>                         ClockPM- Surprise- LLActRep+ BwNot+ ASPMOptComp-
>                 LnkCtl: ASPM Disabled; RCB 128 bytes Disabled- CommClk-
>                         ExtSynch- ClockPM- AutWidDis- BWInt- AutBWInt-
>                 LnkSta: Speed 8GT/s, Width x4, TrErr- Train- SlotClk- DLActive+ BWMgmt- ABWMgmt+
>                 RootCtl: ErrCorrectable- ErrNon-Fatal- ErrFatal- PMEIntEna- CRSVisible-
>                 RootCap: CRSVisible-
>                 RootSta: PME ReqID 0000, PMEStatus- PMEPending-
>                 DevCap2: Completion Timeout: Range ABCD, TimeoutDis+, LTR-, OBFF Not Supported ARIFwd+
>                 DevCtl2: Completion Timeout: 16ms to 55ms, TimeoutDis+, LTR-, OBFF Disabled ARIFwd+
>                 LnkCtl2: Target Link Speed: 8GT/s, EnterCompliance- SpeedDis-
>                          Transmit Margin: Normal Operating Range, EnterModifiedCompliance- ComplianceSOS-
>                          Compliance De-emphasis: -6dB
>                 LnkSta2: Current De-emphasis Level: -3.5dB, EqualizationComplete+, EqualizationPhase1+
>                          EqualizationPhase2+, EqualizationPhase3+, LinkEqualizationRequest-
>         Capabilities: [100 v1] Advanced Error Reporting
>                 UESta:  DLP- SDES- TLP- FCP- CmpltTO- CmpltAbrt- UnxCmplt- RxOF- MalfTLP- ECRC- UnsupReq- ACSViol-
>                 UEMsk:  DLP- SDES- TLP+ FCP- CmpltTO+ CmpltAbrt+ UnxCmplt- RxOF- MalfTLP- ECRC+ UnsupReq- ACSViol-
>                 UESvrt: DLP- SDES- TLP- FCP- CmpltTO- CmpltAbrt- UnxCmplt- RxOF- MalfTLP- ECRC- UnsupReq- ACSViol-
>                 CESta:  RxErr- BadTLP- BadDLLP- Rollover- Timeout- NonFatalErr-
>                 CEMsk:  RxErr- BadTLP- BadDLLP- Rollover- Timeout- NonFatalErr-
>                 AERCap: First Error Pointer: 00, GenCap+ CGenEn+ ChkCap+ ChkEn+
>         Capabilities: [148 v1] #19
> 
> Signed-off-by: Pingfan Liu <kernelfans@gmail.com>
> ---
>  drivers/pci/hotplug/shpchp_core.c | 13 +++++++++++++
>  1 file changed, 13 insertions(+)
> 
> diff --git a/drivers/pci/hotplug/shpchp_core.c b/drivers/pci/hotplug/shpchp_core.c
> index 1f0f969..85b4665 100644
> --- a/drivers/pci/hotplug/shpchp_core.c
> +++ b/drivers/pci/hotplug/shpchp_core.c
> @@ -336,6 +336,18 @@ static void shpc_remove(struct pci_dev *dev)
>  	kfree(ctrl);
>  }
>  
> +static int shpc_extra_match(struct pci_dev *pdev)
> +{
> +	/* do not claim pcie port device */
> +	if (pci_is_pcie(pdev) &&
> +	    (pci_pcie_type(pdev) == PCI_EXP_TYPE_ROOT_PORT ||
> +	     pci_pcie_type(pdev) == PCI_EXP_TYPE_UPSTREAM ||
> +	     pci_pcie_type(pdev) == PCI_EXP_TYPE_DOWNSTREAM))
> +		return -ENODEV;
> +
> +	return 0;
> +}
> +
>  static const struct pci_device_id shpcd_pci_tbl[] = {
>  	{PCI_DEVICE_CLASS(((PCI_CLASS_BRIDGE_PCI << 8) | 0x00), ~0)},
>  	{ /* end: all zeroes */ }
> @@ -345,6 +357,7 @@ MODULE_DEVICE_TABLE(pci, shpcd_pci_tbl);
>  static struct pci_driver shpc_driver = {
>  	.name =		SHPC_MODULE_NAME,
>  	.id_table =	shpcd_pci_tbl,
> +	.extra_match =	shpc_extra_match,
>  	.probe =	shpc_probe,
>  	.remove =	shpc_remove,
>  };
> -- 
> 2.7.4
> 

Re: [patchV2 2/2] pci/shpchp: no claim on pcie port device

[Pingfan Liu]

On Wed, Jun 13, 2018 at 9:13 PM Bjorn Helgaas <helgaas@kernel.org> wrote:
>
> On Wed, Jun 13, 2018 at 02:29:57PM +0800, Pingfan Liu wrote:
> > The Linux Device Driver Model allows a physical device to be handled
> > by only a single driver. But at present, both shpchp and portdrv_pci
> > claim PCI_CLASS_BRIDGE_PCI, and touch devices_kset when the drivers are
> > loaded. This causes a few problems, one is the wrong shutdown seq of
> > devices, owing to the broken devices_kset. This patch keeps shpchp away
> > from pcie port device, using the extra matching method.
>
> As Christoph pointed out, something seems wrong if we add to
> devices_kset even when the probe fails.  We will call shpc_probe() for
> the device below, but there's no SHPC capability, so the probe should
> fail when shpc_init() fails.
>
> I do think the current structure where portdrv and shpchp are both
> "drivers" that claim bridges is broken.  This has caused problems
> before, e.g., some PCIe switches have performance counters, and a
> driver for those switches ought to be able to claim the device and use
> the counters, but portdrv is in the way.
>
> I think it would be better if portdrv (and maybe shpchp; not sure
> about that) were integrated into the PCIe core and did not register as
> a driver.  But this is a big project, far beyond the scope of this
> current issue.
>
Yes, I agree. But just leaving this for later fix, and try to bring
out a fix in drivers/core firstly.

Regards,
Pingfan

> > Note:
> > I hit this bug on a Power9 machine, when "kexec -e", and see a ata-disk
> > behind a bridge can not write back buffer in flight due to the former
> > shutdown of the bridge which clears the BusMaster bit.
> >
> > the device involved:
> > 0004:00:00.0 PCI bridge: IBM Device 04c1 (prog-if 00 [Normal decode])
> >         Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR+ FastB2B- DisINTx-
> >         Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort- <TAbort- <MAbort- >SERR- <PERR- INTx-
> >         Latency: 0
> >         NUMA node: 0
> >         Bus: primary=00, secondary=01, subordinate=12, sec-latency=0
> >         I/O behind bridge: 00000000-00000fff
> >         Memory behind bridge: 80000000-ffefffff
> >         Prefetchable memory behind bridge: 0006024000000000-0006027f7fffffff
> >         Secondary status: 66MHz- FastB2B- ParErr- DEVSEL=fast >TAbort- <TAbort- <MAbort- <SERR- <PERR-
> >         BridgeCtl: Parity- SERR+ NoISA- VGA- MAbort- >Reset- FastB2B-
> >                 PriDiscTmr- SecDiscTmr- DiscTmrStat- DiscTmrSERREn-
> >         Capabilities: [40] Power Management version 3
> >                 Flags: PMEClk- DSI- D1- D2- AuxCurrent=0mA PME(D0+,D1-,D2-,D3hot+,D3cold+)
> >                 Status: D0 NoSoftRst- PME-Enable- DSel=0 DScale=0 PME-
> >         Capabilities: [48] Express (v2) Root Port (Slot-), MSI 00
> >                 DevCap: MaxPayload 512 bytes, PhantFunc 0
> >                         ExtTag- RBE+
> >                 DevCtl: Report errors: Correctable- Non-Fatal- Fatal- Unsupported-
> >                         RlxdOrd- ExtTag- PhantFunc- AuxPwr- NoSnoop-
> >                         MaxPayload 256 bytes, MaxReadReq 128 bytes
> >                 DevSta: CorrErr- UncorrErr- FatalErr- UnsuppReq- AuxPwr- TransPend-
> >                 LnkCap: Port #0, Speed 8GT/s, Width x8, ASPM not supported, Exit Latency L0s <64ns, L1 <1us
> >                         ClockPM- Surprise- LLActRep+ BwNot+ ASPMOptComp-
> >                 LnkCtl: ASPM Disabled; RCB 128 bytes Disabled- CommClk-
> >                         ExtSynch- ClockPM- AutWidDis- BWInt- AutBWInt-
> >                 LnkSta: Speed 8GT/s, Width x4, TrErr- Train- SlotClk- DLActive+ BWMgmt- ABWMgmt+
> >                 RootCtl: ErrCorrectable- ErrNon-Fatal- ErrFatal- PMEIntEna- CRSVisible-
> >                 RootCap: CRSVisible-
> >                 RootSta: PME ReqID 0000, PMEStatus- PMEPending-
> >                 DevCap2: Completion Timeout: Range ABCD, TimeoutDis+, LTR-, OBFF Not Supported ARIFwd+
> >                 DevCtl2: Completion Timeout: 16ms to 55ms, TimeoutDis+, LTR-, OBFF Disabled ARIFwd+
> >                 LnkCtl2: Target Link Speed: 8GT/s, EnterCompliance- SpeedDis-
> >                          Transmit Margin: Normal Operating Range, EnterModifiedCompliance- ComplianceSOS-
> >                          Compliance De-emphasis: -6dB
> >                 LnkSta2: Current De-emphasis Level: -3.5dB, EqualizationComplete+, EqualizationPhase1+
> >                          EqualizationPhase2+, EqualizationPhase3+, LinkEqualizationRequest-
> >         Capabilities: [100 v1] Advanced Error Reporting
> >                 UESta:  DLP- SDES- TLP- FCP- CmpltTO- CmpltAbrt- UnxCmplt- RxOF- MalfTLP- ECRC- UnsupReq- ACSViol-
> >                 UEMsk:  DLP- SDES- TLP+ FCP- CmpltTO+ CmpltAbrt+ UnxCmplt- RxOF- MalfTLP- ECRC+ UnsupReq- ACSViol-
> >                 UESvrt: DLP- SDES- TLP- FCP- CmpltTO- CmpltAbrt- UnxCmplt- RxOF- MalfTLP- ECRC- UnsupReq- ACSViol-
> >                 CESta:  RxErr- BadTLP- BadDLLP- Rollover- Timeout- NonFatalErr-
> >                 CEMsk:  RxErr- BadTLP- BadDLLP- Rollover- Timeout- NonFatalErr-
> >                 AERCap: First Error Pointer: 00, GenCap+ CGenEn+ ChkCap+ ChkEn+
> >         Capabilities: [148 v1] #19
> >
> > Signed-off-by: Pingfan Liu <kernelfans@gmail.com>
> > ---
> >  drivers/pci/hotplug/shpchp_core.c | 13 +++++++++++++
> >  1 file changed, 13 insertions(+)
> >
> > diff --git a/drivers/pci/hotplug/shpchp_core.c b/drivers/pci/hotplug/shpchp_core.c
> > index 1f0f969..85b4665 100644
> > --- a/drivers/pci/hotplug/shpchp_core.c
> > +++ b/drivers/pci/hotplug/shpchp_core.c
> > @@ -336,6 +336,18 @@ static void shpc_remove(struct pci_dev *dev)
> >       kfree(ctrl);
> >  }
> >
> > +static int shpc_extra_match(struct pci_dev *pdev)
> > +{
> > +     /* do not claim pcie port device */
> > +     if (pci_is_pcie(pdev) &&
> > +         (pci_pcie_type(pdev) == PCI_EXP_TYPE_ROOT_PORT ||
> > +          pci_pcie_type(pdev) == PCI_EXP_TYPE_UPSTREAM ||
> > +          pci_pcie_type(pdev) == PCI_EXP_TYPE_DOWNSTREAM))
> > +             return -ENODEV;
> > +
> > +     return 0;
> > +}
> > +
> >  static const struct pci_device_id shpcd_pci_tbl[] = {
> >       {PCI_DEVICE_CLASS(((PCI_CLASS_BRIDGE_PCI << 8) | 0x00), ~0)},
> >       { /* end: all zeroes */ }
> > @@ -345,6 +357,7 @@ MODULE_DEVICE_TABLE(pci, shpcd_pci_tbl);
> >  static struct pci_driver shpc_driver = {
> >       .name =         SHPC_MODULE_NAME,
> >       .id_table =     shpcd_pci_tbl,
> > +     .extra_match =  shpc_extra_match,
> >       .probe =        shpc_probe,
> >       .remove =       shpc_remove,
> >  };
> > --
> > 2.7.4
> >