Simplify PCIe native ownership detection logic
Subsystems (3)
v8 cover Simplify PCIe native ownership detection logic
From: Kuppuswamy Sathyanarayanan <[email protected]>
Currently, PCIe capabilities ownership status is detected by verifying the status of pcie_ports_native, pcie_ports_dpc_native and _OSC negotiated results (cached in struct pci_host_bridge ->native_* members). But this logic can be simplified, and we can use only struct pci_host_bridge ->native_* members to detect it.
This patchset removes the distributed checks for pcie_ports_native, pcie_ports_dpc_native parameters.
Changes since v7:
* Fixed "fix array_size.cocci warnings".
Changes since v6:
* Created new patch for CONFIG_PCIEPORTBUS check in pci_init_host_bridge().
* Added warning message for a case when pcie_ports_native overrides _OSC negotiation result.
Changes since v5:
* Rebased on top of v5.8-rc1
Changes since v4:
* Changed the patch set title (Original link: https://lkml.org/lkml/2020/5/26/1710)
* Added AER/DPC dependency logic cleanup fixes.
Kuppuswamy Sathyanarayanan (5):
PCI: Conditionally initialize host bridge native_* members ACPI/PCI: Ignore _OSC negotiation result if pcie_ports_native is set. ACPI/PCI: Ignore _OSC DPC negotiation result if pcie_ports_dpc_native
is set.
PCI/portdrv: Remove redundant pci_aer_available() check in DPC enable
logic
PCI/DPC: Move AER/DPC dependency checks out of DPC driver
drivers/acpi/pci_root.c | 61 ++++++++++++++++++++++++++----- drivers/pci/hotplug/pciehp_core.c | 2 +- drivers/pci/pci-acpi.c | 3 -- drivers/pci/pcie/aer.c | 2 +- drivers/pci/pcie/dpc.c | 4 +- drivers/pci/pcie/portdrv.h | 2 - drivers/pci/pcie/portdrv_core.c | 13 +++---- drivers/pci/probe.c | 4 +- include/linux/pci.h | 2 + 9 files changed, 64 insertions(+), 29 deletions(-)
Signature · 1 line
2.17.1
PATCH v8 1/5 PCI: Conditionally initialize host bridge native_* members
From: Kuppuswamy Sathyanarayanan <[email protected]>
If CONFIG_PCIEPORTBUS is not enabled in kernel then initialing struct pci_host_bridge PCIe specific native_* members to "1" is incorrect. So protect the PCIe specific member initialization with CONFIG_PCIEPORTBUS.
--- drivers/pci/probe.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)
drivers/pci/probe.c +3 -1
@@ -588,12 +588,14 @@ static void pci_init_host_bridge(struct pci_host_bridge *bridge)* may implement its own AER handling and use _OSC to prevent the * OS from interfering. */ +#ifdef CONFIG_PCIEPORTBUS bridge->native_aer =1 ; bridge->native_pcie_hotplug =1 ; - bridge->native_shpc_hotplug =1 ; bridge->native_pme =1 ; bridge->native_ltr =1 ; bridge->native_dpc =1 ; +#endif + bridge->native_shpc_hotplug =1 ;device_initialize (&bridge->dev); }
Signature · 1 line
2.17.1
Re: [PATCH v8 1/5] PCI: Conditionally initialize host bridge native_* members
Quoted 6 lines · On Fri, Jul 24, 2020 at 08:58:52PM -0700…
From: Kuppuswamy Sathyanarayanan <[email protected]>
If
CONFIG_PCIEPORTBUSis not enabled in kernel then initialing structpci_host_bridgePCIe specific native_* members to "1" is incorrect. So protect the PCIe specific member initialization withCONFIG_PCIEPORTBUS.
s/initialing/initializing/
Quoted 10 lines
Signed-off-by Kuppuswamy Sathyanarayanan <[email protected]>--- drivers/pci/probe.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)drivers/pci/probe.c +0
-0
drivers/pci/probe.c static void pci_init_host_bridge(struct pci_host_bridge *bridge) · line 595
Quoted 9 lines
drivers/pci/probe.c +1
-1@@ -588,12 +588,14 @@ static void pci_init_host_bridge(struct pci_host_bridge *bridge)* may implement its own AER handling and use _OSC to prevent the * OS from interfering. */ +#ifdef CONFIG_PCIEPORTBUS bridge->native_aer =1 ; bridge->native_pcie_hotplug =1 ; - bridge->native_shpc_hotplug =1 ; bridge->native_pme =1 ; bridge->native_ltr =1 ;
native_ltr isn't dependent on PCIEPORTBUS either, is it? It's only used for ASPM.
Quoted 9 lines
drivers/pci/probe.c +2
-0bridge->native_dpc =1 ; +#endif + bridge->native_shpc_hotplug =1 ;device_initialize (&bridge->dev); }Signature · 1 line
2.17.1
Re: [PATCH v8 1/5] PCI: Conditionally initialize host bridge native_* members
Quoted 9 lines · Bjorn Helgaas
Quoted 6 lines · On Fri, Jul 24, 2020 at 08:58:52PM -0700…
From: Kuppuswamy Sathyanarayanan <[email protected]>
If
CONFIG_PCIEPORTBUSis not enabled in kernel then initialing structpci_host_bridgePCIe specific native_* members to "1" is incorrect. So protect the PCIe specific member initialization withCONFIG_PCIEPORTBUS.s/initialing/initializing/
will fix it in next version.
Quoted 15 lines
Quoted 15 lines
Signed-off-by Kuppuswamy Sathyanarayanan <[email protected]>--- drivers/pci/probe.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)drivers/pci/probe.c +1
-0@@ -588,12 +588,14 @@ static void pci_init_host_bridge(struct pci_host_bridge *bridge)* may implement its own AER handling and use _OSC to prevent the * OS from interfering. */ +#ifdef CONFIG_PCIEPORTBUS
Quoted 8 lines
drivers/pci/probe.c static void pci_init_host_bridge(struct pci_host_bridge *bridge) · line 595
Quoted 5 lines
drivers/pci/probe.c +0
-1bridge->native_aer =1 ; bridge->native_pcie_hotplug =1 ; - bridge->native_shpc_hotplug =1 ; bridge->native_pme =1 ; bridge->native_ltr =1 ;
native_ltrisn't dependent on PCIEPORTBUS either, is it? It's only used for ASPM.
Agreed. I was confused due to a comment in include/linux/pci.h
unsigned int native_ltr:1; /* OS may use PCIe LTR */
Quoted 10 lines
Quoted 10 lines
drivers/pci/probe.c +2
-0bridge->native_dpc =1 ; +#endif + bridge->native_shpc_hotplug =1 ;device_initialize (&bridge->dev); }Signature · 1 line
2.17.1
Signature · 2 lines
Sathyanarayanan Kuppuswamy Linux Kernel Developer
Re: [PATCH v8 1/5] PCI: Conditionally initialize host bridge native_* members
Quoted 36 lines · Sathyanarayanan
Quoted 9 lines · Bjorn Helgaas
Quoted 6 lines · On Fri, Jul 24, 2020 at 08:58:52PM -0700…
From: Kuppuswamy Sathyanarayanan <[email protected]>
If
CONFIG_PCIEPORTBUSis not enabled in kernel then initialing structpci_host_bridgePCIe specific native_* members to "1" is incorrect. So protect the PCIe specific member initialization withCONFIG_PCIEPORTBUS.s/initialing/initializing/
will fix it in next version.
Quoted 15 lines
Quoted 15 lines
Signed-off-by Kuppuswamy Sathyanarayanan <[email protected]>---
drivers/pci/probe.c| 4 +++-1 file changed, 3 insertions(+), 1 deletion(-)drivers/pci/probe.c +1
-0@@ -588,12 +588,14 @@ static void pci_init_host_bridge(struct pci_host_bridge *bridge)* may implement its own AER handling and use _OSC to prevent the * OS from interfering. */ +#ifdef CONFIG_PCIEPORTBUS Quoted 8 lines
drivers/pci/probe.c static void pci_init_host_bridge(struct pci_host_bridge *bridge) · line 595
Quoted 5 lines
drivers/pci/probe.c +0
-1bridge->native_aer = 1; bridge->native_pcie_hotplug = 1; - bridge->native_shpc_hotplug = 1; bridge->native_pme = 1; bridge->native_ltr = 1;
native_ltrisn't dependent on PCIEPORTBUS either, is it? It's only used for ASPM.
Quoted 3 lines
Agreed. I was confused due to a comment in
include/linux/pci.hunsigned int native_ltr:1; /* OS may use PCIe LTR */
Currently there is no code dependency between LTR and CONFIG_PCIEPORTBUS.
But I am wondering whether its correct to move LTR code under CONFIG_PCIEPORTBUS?. As per PCIe spec v5.0 sec 7.8.2, LTR is a optional PCIe extended capability. So why is not moved under drivers/pci/pcie/*. What is the criteria for code to be placed under drivers/pci/pcie/*
Quoted 11 lines
Quoted 10 lines
Quoted 10 lines
drivers/pci/probe.c +2
-0bridge->native_dpc = 1; +#endif + bridge->native_shpc_hotplug = 1; device_initialize(&bridge->dev); }Signature · 1 line
2.17.1
Signature · 2 lines
Sathyanarayanan Kuppuswamy Linux Kernel Developer
Re: [PATCH v8 1/5] PCI: Conditionally initialize host bridge native_* members
Quoted 39 lines · Sathyanarayanan
Quoted 34 lines · Sathyanarayanan
Quoted 9 lines · Bjorn Helgaas
Quoted 6 lines · On Fri, Jul 24, 2020 at 08:58:52PM -0700…
From: Kuppuswamy Sathyanarayanan <[email protected]>
If
CONFIG_PCIEPORTBUSis not enabled in kernel then initialing structpci_host_bridgePCIe specific native_* members to "1" is incorrect. So protect the PCIe specific member initialization withCONFIG_PCIEPORTBUS.s/initialing/initializing/
will fix it in next version.
Quoted 15 lines
Quoted 15 lines
Signed-off-by Kuppuswamy Sathyanarayanan <[email protected]>---
drivers/pci/probe.c| 4 +++-1 file changed, 3 insertions(+), 1 deletion(-)drivers/pci/probe.c +1
-0@@ -588,12 +588,14 @@ static void pci_init_host_bridge(struct pci_host_bridge *bridge)* may implement its own AER handling and use _OSC to prevent the * OS from interfering. */ +#ifdef CONFIG_PCIEPORTBUS Quoted 8 lines
drivers/pci/probe.c static void pci_init_host_bridge(struct pci_host_bridge *bridge) · line 595
Quoted 5 lines
drivers/pci/probe.c +0
-1bridge->native_aer = 1; bridge->native_pcie_hotplug = 1; - bridge->native_shpc_hotplug = 1; bridge->native_pme = 1; bridge->native_ltr = 1;
native_ltrisn't dependent on PCIEPORTBUS either, is it? It's only used for ASPM.Quoted 3 lines
Agreed. I was confused due to a comment in
include/linux/pci.hunsigned int native_ltr:1; /* OS may use PCIe LTR */
Quoted 7 lines
Currently there is no code dependency between LTR and
CONFIG_PCIEPORTBUS.But I am wondering whether its correct to move LTR code under
CONFIG_PCIEPORTBUS?. As per PCIe spec v5.0 sec 7.8.2, LTR is a optional PCIe extended capability. So why is not moved underdrivers/pci/pcie/*. What is the criteria for code to be placed underdrivers/pci/pcie/*
Some folks think drivers/pci/pcie/ should not exist, and I tend to agree, but it's a fair bit of churn to remove it. We do have quite a bit of PCIe extended capability support in drivers/pci -- ats.c, iov.c, vc.c.
There's no need to move LTR under CONFIG_PCIEPORTBUS because CONFIG_PCIEPORTBUS enables portdrv, and AFAIK there's nothing LTR-related that relies on portdrv.
The stuff currently in drivers/pci/pcie is mostly just portdrv and services that depend on it. aspm.c and ptm.c are exceptions and really should be in drivers/pci.
Quoted 13 lines
Quoted 9 lines
Quoted 8 lines
Quoted 8 lines
drivers/pci/probe.c +2
-0bridge->native_dpc = 1; +#endif + bridge->native_shpc_hotplug = 1; device_initialize(&bridge->dev); }Signature · 1 line
2.17.1Signature · 2 lines
Sathyanarayanan Kuppuswamy Linux Kernel Developer
Re: [PATCH v8 1/5] PCI: Conditionally initialize host bridge native_* members
Quoted 19 lines · Bjorn Helgaas
Quoted 7 lines · Sathyanarayanan
Quoted 4 lines · Sathyanarayanan
Quoted 3 lines · Bjorn Helgaas
Quoted 2 lines · On Fri, Jul 24, 2020 at 08:58:52PM -0700…
From: Kuppuswamy Sathyanarayanan <[email protected]>
Quoted 5 lines
But I am wondering whether its correct to move LTR code under
CONFIG_PCIEPORTBUS?. As per PCIe spec v5.0 sec 7.8.2, LTR is a optional PCIe extended capability. So why is not moved underdrivers/pci/pcie/*. What is the criteria for code to be placed underdrivers/pci/pcie/*Some folks think
drivers/pci/pcie/should not exist, and I tend to agree, but it's a fair bit of churn to remove it. We do have quite a bit of PCIe extended capability support indrivers/pci--ats.c,iov.c,vc.c.
Quoted 7 lines
There's no need to move LTR under
CONFIG_PCIEPORTBUSbecauseCONFIG_PCIEPORTBUSenables portdrv, and AFAIK there's nothing LTR-related that relies on portdrv.The stuff currently in
drivers/pci/pcieis mostly just portdrv and services that depend on it.aspm.candptm.care exceptions and really should be indrivers/pci.
Thanks for the clarification. I will remove the CONFIG_PCIEPORTBUS dependency.
Quoted 10 lines
Quoted 10 lines
Quoted 6 lines
Quoted 5 lines
Quoted 5 lines
Signature · 2 lines
-- 2.17.1Signature · 2 lines
Sathyanarayanan Kuppuswamy Linux Kernel Developer
Signature · 2 lines
Sathyanarayanan Kuppuswamy Linux Kernel Developer
PATCH v8 2/5 ACPI/PCI: Ignore _OSC negotiation result if pcie_ports_native is set.
In reply to [email protected]From: Kuppuswamy Sathyanarayanan <[email protected]>
pcie_ports_native is set only if user requests native handling of PCIe capabilities via pcie_port_setup command line option. User input takes precedence over _OSC based control negotiation result. So consider the _OSC negotiated result only if pcie_ports_native is unset.
Also, since struct pci_host_bridge ->native_* members caches the ownership status of various PCIe capabilities, use them instead of distributed checks for pcie_ports_native.
--- drivers/acpi/pci_root.c | 61 ++++++++++++++++++++++++++----- drivers/pci/hotplug/pciehp_core.c | 2 +- drivers/pci/pci-acpi.c | 3 -- drivers/pci/pcie/aer.c | 2 +- drivers/pci/pcie/portdrv_core.c | 9 ++--- 5 files changed, 56 insertions(+), 21 deletions(-)
drivers/acpi/pci_root.c +51 -10
@@ -145,6 +145,17 @@ static struct pci_osc_bit_struct pci_osc_control_bit[] = { { OSC_PCI_EXPRESS_DPC_CONTROL, "DPC" }, }; +static char *get_osc_desc (u32 bit) +{ +int i =0 ; + +for (i =0 ; i <ARRAY_SIZE (pci_osc_control_bit); i++) +if (bit == pci_osc_control_bit[i].bit) +return pci_osc_control_bit[i].desc; + +return NULL ; +} +static void decode_osc_bits (struct acpi_pci_root *root,char *msg,u32 word,struct pci_osc_bit_struct *table,int size) {
@@ -914,18 +925,48 @@ struct pci_bus *acpi_pci_root_create(struct acpi_pci_root *root, goto out_release_info; host_bridge =to_pci_host_bridge (bus->bridge); -if (!(root->osc_control_set &OSC_PCI_EXPRESS_NATIVE_HP_CONTROL )) - host_bridge->native_pcie_hotplug =0 ; +if (!(root->osc_control_set &OSC_PCI_EXPRESS_NATIVE_HP_CONTROL )) { +if (!pcie_ports_native) + host_bridge->native_pcie_hotplug =0 ; +else +dev_warn (&bus->dev,"OS overrides %s firmware control" , +get_osc_desc (OSC_PCI_EXPRESS_NATIVE_HP_CONTROL )); + } +if (!(root->osc_control_set &OSC_PCI_SHPC_NATIVE_HP_CONTROL )) host_bridge->native_shpc_hotplug =0 ; -if (!(root->osc_control_set &OSC_PCI_EXPRESS_AER_CONTROL )) - host_bridge->native_aer =0 ; -if (!(root->osc_control_set &OSC_PCI_EXPRESS_PME_CONTROL )) - host_bridge->native_pme =0 ; -if (!(root->osc_control_set &OSC_PCI_EXPRESS_LTR_CONTROL )) - host_bridge->native_ltr =0 ; -if (!(root->osc_control_set &OSC_PCI_EXPRESS_DPC_CONTROL )) - host_bridge->native_dpc =0 ; + +if (!(root->osc_control_set &OSC_PCI_EXPRESS_AER_CONTROL )) { +if (!pcie_ports_native) + host_bridge->native_aer =0 ; +else +dev_warn (&bus->dev,"OS overrides %s firmware control" , +get_osc_desc (OSC_PCI_EXPRESS_AER_CONTROL )); + } + +if (!(root->osc_control_set &OSC_PCI_EXPRESS_PME_CONTROL )) { +if (!pcie_ports_native) + host_bridge->native_pme =0 ; +else +dev_warn (&bus->dev,"OS overrides %s firmware control" , +get_osc_desc (OSC_PCI_EXPRESS_PME_CONTROL )); + } + +if (!(root->osc_control_set &OSC_PCI_EXPRESS_LTR_CONTROL )) { +if (!pcie_ports_native) + host_bridge->native_ltr =0 ; +else +dev_warn (&bus->dev,"OS overrides %s firmware control" , +get_osc_desc (OSC_PCI_EXPRESS_LTR_CONTROL )); + } + +if (!(root->osc_control_set &OSC_PCI_EXPRESS_DPC_CONTROL )) { +if (!pcie_ports_native) + host_bridge->native_dpc =0 ; +else +dev_warn (&bus->dev,"OS overrides %s firmware control" , +get_osc_desc (OSC_PCI_EXPRESS_DPC_CONTROL )); + }/* * Evaluate the "PCI Boot Configuration" _DSM Function. If it
drivers/pci/hotplug/pciehp_core.c +1 -1
@@ -255,7 +255,7 @@ static bool pme_is_native(struct pcie_device *dev)const struct pci_host_bridge *host; host =pci_find_host_bridge (dev->port->bus); -return pcie_ports_native || host->native_pme; +return host->native_pme; }static void pciehp_disable_interrupt (struct pcie_device *dev)
drivers/pci/pci-acpi.c +0 -3
@@ -800,9 +800,6 @@ bool pciehp_is_native(struct pci_dev *bridge)if (!(slot_cap &PCI_EXP_SLTCAP_HPC ))return false; -if (pcie_ports_native) -return true; - host =pci_find_host_bridge (bridge->bus);return host->native_pcie_hotplug; }
drivers/pci/pcie/aer.c +1 -1
@@ -219,7 +219,7 @@ int pcie_aer_is_native(struct pci_dev *dev)if (!dev->aer_cap)return 0 ; -return pcie_ports_native || host->native_aer; +return host->native_aer; }int pci_enable_pcie_error_reporting (struct pci_dev *dev)
drivers/pci/pcie/portdrv_core.c +3 -6
@@ -208,8 +208,7 @@ static int get_port_device_capability(struct pci_dev *dev)struct pci_host_bridge *host =pci_find_host_bridge (dev->bus);int services =0 ; -if (dev->is_hotplug_bridge && - (pcie_ports_native || host->native_pcie_hotplug)) { +if (dev->is_hotplug_bridge && host->native_pcie_hotplug) { services |=PCIE_PORT_SERVICE_HP ;/*
@@ -221,8 +220,7 @@ static int get_port_device_capability(struct pci_dev *dev) }#ifdef CONFIG_PCIEAER -if (dev->aer_cap &&pci_aer_available () && - (pcie_ports_native || host->native_aer)) { +if (dev->aer_cap &&pci_aer_available () && host->native_aer) { services |=PCIE_PORT_SERVICE_AER ;/*
@@ -238,8 +236,7 @@ static int get_port_device_capability(struct pci_dev *dev)* Event Collectors can also generate PMEs, but we don't handle * those yet. */ -if (pci_pcie_type (dev) ==PCI_EXP_TYPE_ROOT_PORT && - (pcie_ports_native || host->native_pme)) { +if (pci_pcie_type (dev) ==PCI_EXP_TYPE_ROOT_PORT && host->native_pme) { services |=PCIE_PORT_SERVICE_PME ;/*
Signature · 1 line
2.17.1
Re: [PATCH v8 2/5] ACPI/PCI: Ignore _OSC negotiation result if pcie_ports_native is set.
Quoted 4 lines · On Sat, Jul 25, 2020 at 7:01 AM <[email protected]>…
From: Kuppuswamy Sathyanarayanan <[email protected]>
pcie_ports_nativeis set only if user requests native handling
the user
Quoted 8 lines
of PCIe capabilities via
pcie_port_setupcommand line option. User input takes precedence over_OSCbased control negotiation result. So consider the_OSCnegotiated result only ifpcie_ports_nativeis unset.Also, since struct
pci_host_bridge->native_* members caches the ownership status of various PCIe capabilities, use them instead of distributed checks forpcie_ports_native.
...
drivers/acpi/pci_root.c static struct pci_osc_bit_struct pci_osc_control_bit[] = { · line 150
Quoted 4 lines
drivers/acpi/pci_root.c +3
-0+static char *get_osc_desc (u32 bit) +{ +int i =0 ;
Unneeded assignment.
Quoted 6 lines
drivers/acpi/pci_root.c +6
-0+for (i =0 ; i <ARRAY_SIZE (pci_osc_control_bit); i++) +if (bit == pci_osc_control_bit[i].bit) +return pci_osc_control_bit[i].desc; + +return NULL ; +}
...
drivers/acpi/pci_root.c struct pci_bus *acpi_pci_root_create(struct acpi_pci_root *root,
Quoted 9 lines
host_bridge = to_pci_host_bridge(bus->bridge);drivers/acpi/pci_root.c +5
-2-if (!(root->osc_control_set &OSC_PCI_EXPRESS_NATIVE_HP_CONTROL )) - host_bridge->native_pcie_hotplug =0 ; +if (!(root->osc_control_set &OSC_PCI_EXPRESS_NATIVE_HP_CONTROL )) { +if (!pcie_ports_native) + host_bridge->native_pcie_hotplug =0 ; +else + dev_warn(&bus->dev, "OS overrides %s firmware control",
pci_warn() ?
drivers/acpi/pci_root.c struct pci_bus *acpi_pci_root_create(struct acpi_pci_root *root,
Quoted 6 lines
drivers/acpi/pci_root.c +3
-0+ get_osc_desc(OSC_PCI_EXPRESS_NATIVE_HP_CONTROL)); + } +if (!(root->osc_control_set &OSC_PCI_SHPC_NATIVE_HP_CONTROL )) host_bridge->native_shpc_hotplug =0 ;
Group them in the same way you did in the previous patch.
Quoted 6 lines
drivers/acpi/pci_root.c +0
-6-if (!(root->osc_control_set &OSC_PCI_EXPRESS_AER_CONTROL )) - host_bridge->native_aer =0 ; -if (!(root->osc_control_set &OSC_PCI_EXPRESS_PME_CONTROL )) - host_bridge->native_pme =0 ; -if (!(root->osc_control_set &OSC_PCI_EXPRESS_LTR_CONTROL )) - host_bridge->native_ltr =0 ;
drivers/acpi/pci_root.c struct pci_bus *acpi_pci_root_create(struct acpi_pci_root *root,
Quoted 10 lines
drivers/acpi/pci_root.c +7
-2-if (!(root->osc_control_set &OSC_PCI_EXPRESS_DPC_CONTROL )) - host_bridge->native_dpc =0 ; + +if (!(root->osc_control_set &OSC_PCI_EXPRESS_AER_CONTROL )) { +if (!pcie_ports_native) + host_bridge->native_aer =0 ; +else +dev_warn (&bus->dev,"OS overrides %s firmware control" , +get_osc_desc (OSC_PCI_EXPRESS_AER_CONTROL ));
Looks like a lot of duplication here. Perhaps split out a helper ?
Quoted 1 line
+ }
Signature · 2 lines
With Best Regards, Andy Shevchenko
Re: [PATCH v8 2/5] ACPI/PCI: Ignore _OSC negotiation result if pcie_ports_native is set.
In reply to [email protected]Quoted 45 lines · On Fri, Jul 24, 2020 at 08:58:53PM -0700…
From: Kuppuswamy Sathyanarayanan <[email protected]>
pcie_ports_nativeis set only if user requests native handling of PCIe capabilities viapcie_port_setupcommand line option. User input takes precedence over_OSCbased control negotiation result. So consider the_OSCnegotiated result only ifpcie_ports_nativeis unset.Also, since struct
pci_host_bridge->native_* members caches the ownership status of various PCIe capabilities, use them instead of distributed checks forpcie_ports_native.Signed-off-by Kuppuswamy Sathyanarayanan <[email protected]>--- drivers/acpi/pci_root.c | 61 ++++++++++++++++++++++++++----- drivers/pci/hotplug/pciehp_core.c | 2 +- drivers/pci/pci-acpi.c | 3 -- drivers/pci/pcie/aer.c | 2 +- drivers/pci/pcie/portdrv_core.c | 9 ++--- 5 files changed, 56 insertions(+), 21 deletions(-)drivers/acpi/pci_root.c +11
-0@@ -145,6 +145,17 @@ static struct pci_osc_bit_struct pci_osc_control_bit[] = { { OSC_PCI_EXPRESS_DPC_CONTROL, "DPC" }, }; +static char *get_osc_desc (u32 bit) +{ +int i =0 ; + +for (i =0 ; i <ARRAY_SIZE (pci_osc_control_bit); i++) +if (bit == pci_osc_control_bit[i].bit) +return pci_osc_control_bit[i].desc; + +return NULL ; +} +static void decode_osc_bits (struct acpi_pci_root *root,char *msg,u32 word,struct pci_osc_bit_struct *table,int size) {@@ -914,18 +925,48 @@ struct pci_bus *acpi_pci_root_create(struct acpi_pci_root *root, goto out_release_info;
drivers/acpi/pci_root.c struct pci_bus *acpi_pci_root_create(struct acpi_pci_root *root, · line 933
Quoted 10 lines
drivers/acpi/pci_root.c +6
-2host_bridge =to_pci_host_bridge (bus->bridge); -if (!(root->osc_control_set &OSC_PCI_EXPRESS_NATIVE_HP_CONTROL )) - host_bridge->native_pcie_hotplug =0 ; +if (!(root->osc_control_set &OSC_PCI_EXPRESS_NATIVE_HP_CONTROL )) { +if (!pcie_ports_native) + host_bridge->native_pcie_hotplug =0 ; +else +dev_warn (&bus->dev,"OS overrides %s firmware control" , +get_osc_desc (OSC_PCI_EXPRESS_NATIVE_HP_CONTROL ));
There's got to be a way to write this more concisely. Maybe something like this?
#define OSC_OWNER(ctrl, bit, flag) \
if (!(ctrl & bit)) \
flag = 0;
if (pcie_ports_native)
decode_osc_control(root, "OS forcibly taking over", ~0);
else {
ctrl = root->osc_control_set;
OSC_OWNER(ctrl, OSC_PCI_EXPRESS_AER_CONTROL, host_bridge->native_aer);
OSC_OWNER(ctrl, OSC_PCI_EXPRESS_PME_CONTROL, host_bridge->native_pme);
...
}Quoted 52 lines
drivers/acpi/pci_root.c +34
-8+ } +if (!(root->osc_control_set &OSC_PCI_SHPC_NATIVE_HP_CONTROL )) host_bridge->native_shpc_hotplug =0 ; -if (!(root->osc_control_set &OSC_PCI_EXPRESS_AER_CONTROL )) - host_bridge->native_aer =0 ; -if (!(root->osc_control_set &OSC_PCI_EXPRESS_PME_CONTROL )) - host_bridge->native_pme =0 ; -if (!(root->osc_control_set &OSC_PCI_EXPRESS_LTR_CONTROL )) - host_bridge->native_ltr =0 ; -if (!(root->osc_control_set &OSC_PCI_EXPRESS_DPC_CONTROL )) - host_bridge->native_dpc =0 ; + +if (!(root->osc_control_set &OSC_PCI_EXPRESS_AER_CONTROL )) { +if (!pcie_ports_native) + host_bridge->native_aer =0 ; +else +dev_warn (&bus->dev,"OS overrides %s firmware control" , +get_osc_desc (OSC_PCI_EXPRESS_AER_CONTROL )); + } + +if (!(root->osc_control_set &OSC_PCI_EXPRESS_PME_CONTROL )) { +if (!pcie_ports_native) + host_bridge->native_pme =0 ; +else +dev_warn (&bus->dev,"OS overrides %s firmware control" , +get_osc_desc (OSC_PCI_EXPRESS_PME_CONTROL )); + } + +if (!(root->osc_control_set &OSC_PCI_EXPRESS_LTR_CONTROL )) { +if (!pcie_ports_native) + host_bridge->native_ltr =0 ; +else +dev_warn (&bus->dev,"OS overrides %s firmware control" , +get_osc_desc (OSC_PCI_EXPRESS_LTR_CONTROL )); + } + +if (!(root->osc_control_set &OSC_PCI_EXPRESS_DPC_CONTROL )) { +if (!pcie_ports_native) + host_bridge->native_dpc =0 ; +else +dev_warn (&bus->dev,"OS overrides %s firmware control" , +get_osc_desc (OSC_PCI_EXPRESS_DPC_CONTROL )); + }/* * Evaluate the "PCI Boot Configuration" _DSM Function. If it drivers/pci/hotplug/pciehp_core.c +0
-0
drivers/pci/hotplug/pciehp_core.c static bool pme_is_native(struct pcie_device *dev) · line 258
Quoted 5 lines
drivers/pci/hotplug/pciehp_core.c +1
-1@@ -255,7 +255,7 @@ static bool pme_is_native(struct pcie_device *dev)const struct pci_host_bridge *host; host =pci_find_host_bridge (dev->port->bus); -return pcie_ports_native || host->native_pme; +return host->native_pme;
I love this part!
Quoted 67 lines
drivers/pci/hotplug/pciehp_core.c +0
-0}static void pciehp_disable_interrupt (struct pcie_device *dev)drivers/pci/pci-acpi.c +0
-3@@ -800,9 +800,6 @@ bool pciehp_is_native(struct pci_dev *bridge)if (!(slot_cap &PCI_EXP_SLTCAP_HPC ))return false; -if (pcie_ports_native) -return true; - host =pci_find_host_bridge (bridge->bus);return host->native_pcie_hotplug; }drivers/pci/pcie/aer.c +1
-1@@ -219,7 +219,7 @@ int pcie_aer_is_native(struct pci_dev *dev)if (!dev->aer_cap)return 0 ; -return pcie_ports_native || host->native_aer; +return host->native_aer; }int pci_enable_pcie_error_reporting (struct pci_dev *dev)drivers/pci/pcie/portdrv_core.c +3
-6@@ -208,8 +208,7 @@ static int get_port_device_capability(struct pci_dev *dev)struct pci_host_bridge *host =pci_find_host_bridge (dev->bus);int services =0 ; -if (dev->is_hotplug_bridge && - (pcie_ports_native || host->native_pcie_hotplug)) { +if (dev->is_hotplug_bridge && host->native_pcie_hotplug) { services |=PCIE_PORT_SERVICE_HP ;/* @@ -221,8 +220,7 @@ static int get_port_device_capability(struct pci_dev *dev) }#ifdef CONFIG_PCIEAER -if (dev->aer_cap &&pci_aer_available () && - (pcie_ports_native || host->native_aer)) { +if (dev->aer_cap &&pci_aer_available () && host->native_aer) { services |=PCIE_PORT_SERVICE_AER ;/* @@ -238,8 +236,7 @@ static int get_port_device_capability(struct pci_dev *dev)* Event Collectors can also generate PMEs, but we don't handle * those yet. */ -if (pci_pcie_type (dev) ==PCI_EXP_TYPE_ROOT_PORT && - (pcie_ports_native || host->native_pme)) { +if (pci_pcie_type (dev) ==PCI_EXP_TYPE_ROOT_PORT && host->native_pme) { services |=PCIE_PORT_SERVICE_PME ;/* Signature · 1 line
2.17.1
Re: [PATCH v8 2/5] ACPI/PCI: Ignore _OSC negotiation result if pcie_ports_native is set.
Quoted 38 lines · Bjorn Helgaas
Quoted 26 lines · On Fri, Jul 24, 2020 at 08:58:53PM -0700…
From: Kuppuswamy Sathyanarayanan <[email protected]>
pcie_ports_nativeis set only if user requests native handling of PCIe capabilities viapcie_port_setupcommand line option. User input takes precedence over_OSCbased control negotiation result. So consider the_OSCnegotiated result only ifpcie_ports_nativeis unset.Also, since struct
pci_host_bridge->native_* members caches the ownership status of various PCIe capabilities, use them instead of distributed checks forpcie_ports_native.Signed-off-by Kuppuswamy Sathyanarayanan <[email protected]>--- drivers/acpi/pci_root.c | 61 ++++++++++++++++++++++++++----- drivers/pci/hotplug/pciehp_core.c | 2 +- drivers/pci/pci-acpi.c | 3 -- drivers/pci/pcie/aer.c | 2 +- drivers/pci/pcie/portdrv_core.c | 9 ++--- 5 files changed, 56 insertions(+), 21 deletions(-)drivers/acpi/pci_root.c +0
-0drivers/acpi/pci_root.c static struct pci_osc_bit_struct pci_osc_control_bit[] = { · line 151
Quoted 7 lines
drivers/acpi/pci_root.c +3
-0@@ -145,6 +145,17 @@ static struct pci_osc_bit_struct pci_osc_control_bit[] = { { OSC_PCI_EXPRESS_DPC_CONTROL, "DPC" }, }; + else + dev_warn(&bus->dev, "OS overrides %s firmware control", + get_osc_desc(OSC_PCI_EXPRESS_NATIVE_HP_CONTROL));There's got to be a way to write this more concisely. Maybe something like this?
Quoted 6 lines
#define OSC_OWNER(ctrl, bit, flag) \
if (!(ctrl & bit)) \
flag = 0;
if (pcie_ports_native)
decode_osc_control(root, "OS forcibly taking over", ~0);BIT1 and BIT6 does not have PCIe dependency. And BIT7-31 are reserved. So we can't force all _OSC bits based on pcie_ports_native value. So, IM0, its better to handle PCIe features seperatly.
Quoted 6 lines
else {
ctrl = root->osc_control_set;
OSC_OWNER(ctrl, OSC_PCI_EXPRESS_AER_CONTROL, host_bridge->native_aer);
OSC_OWNER(ctrl, OSC_PCI_EXPRESS_PME_CONTROL, host_bridge->native_pme);
...
}Signature · 2 lines
Sathyanarayanan Kuppuswamy Linux Kernel Developer
Re: [PATCH v8 2/5] ACPI/PCI: Ignore _OSC negotiation result if pcie_ports_native is set.
Quoted 42 lines · Sathyanarayanan
Quoted 41 lines · Bjorn Helgaas
Quoted 26 lines · On Fri, Jul 24, 2020 at 08:58:53PM -0700…
From: Kuppuswamy Sathyanarayanan <[email protected]>
pcie_ports_nativeis set only if user requests native handling of PCIe capabilities viapcie_port_setupcommand line option. User input takes precedence over_OSCbased control negotiation result. So consider the_OSCnegotiated result only ifpcie_ports_nativeis unset.Also, since struct
pci_host_bridge->native_* members caches the ownership status of various PCIe capabilities, use them instead of distributed checks forpcie_ports_native.Signed-off-by Kuppuswamy Sathyanarayanan <[email protected]>--- drivers/acpi/pci_root.c | 61 ++++++++++++++++++++++++++----- drivers/pci/hotplug/pciehp_core.c | 2 +- drivers/pci/pci-acpi.c | 3 -- drivers/pci/pcie/aer.c | 2 +- drivers/pci/pcie/portdrv_core.c | 9 ++--- 5 files changed, 56 insertions(+), 21 deletions(-)drivers/acpi/pci_root.c +0
-0drivers/acpi/pci_root.c static struct pci_osc_bit_struct pci_osc_control_bit[] = { · line 150
Quoted 6 lines
drivers/acpi/pci_root.c +3
-0@@ -145,6 +145,17 @@ static struct pci_osc_bit_struct pci_osc_control_bit[] = { { OSC_PCI_EXPRESS_DPC_CONTROL, "DPC" }, }; + else + dev_warn(&bus->dev, "OS overrides %s firmware control", + get_osc_desc(OSC_PCI_EXPRESS_NATIVE_HP_CONTROL));There's got to be a way to write this more concisely. Maybe something like this?
#define OSC_OWNER(ctrl, bit, flag) \ if (!(ctrl & bit)) \ flag = 0;
Quoted 6 lines
Quoted 2 lines
if (pcie_ports_native) decode_osc_control(root, "OS forcibly taking over", ~0);BIT1 and BIT6 does not have PCIe dependency. And BIT7-31 are reserved. So we can't force all
_OSCbits based onpcie_ports_nativevalue. So, IM0, its better to handle PCIe features seperatly.
Yes, we may need to handle a few bits specially. But we need to figure out a nicer-looking way of coding this. It's too cumbersome to check pcie_ports_native and log a message for every _OSC bit individually.
Quoted 11 lines
Quoted 6 lines
else { ctrl = root->osc_control_set; OSC_OWNER(ctrl, OSC_PCI_EXPRESS_AER_CONTROL, host_bridge->native_aer); OSC_OWNER(ctrl, OSC_PCI_EXPRESS_PME_CONTROL, host_bridge->native_pme); ... }Signature · 2 lines
Sathyanarayanan Kuppuswamy Linux Kernel Developer
Re: [PATCH v8 2/5] ACPI/PCI: Ignore _OSC negotiation result if pcie_ports_native is set.
Quoted 46 lines · Bjorn Helgaas
Quoted 45 lines · Sathyanarayanan
Quoted 43 lines · Bjorn Helgaas
Quoted 26 lines · On Fri, Jul 24, 2020 at 08:58:53PM -0700…
From: Kuppuswamy Sathyanarayanan <[email protected]>
pcie_ports_nativeis set only if user requests native handling of PCIe capabilities viapcie_port_setupcommand line option. User input takes precedence over_OSCbased control negotiation result. So consider the_OSCnegotiated result only ifpcie_ports_nativeis unset.Also, since struct
pci_host_bridge->native_* members caches the ownership status of various PCIe capabilities, use them instead of distributed checks forpcie_ports_native.Signed-off-by Kuppuswamy Sathyanarayanan <[email protected]>--- drivers/acpi/pci_root.c | 61 ++++++++++++++++++++++++++----- drivers/pci/hotplug/pciehp_core.c | 2 +- drivers/pci/pci-acpi.c | 3 -- drivers/pci/pcie/aer.c | 2 +- drivers/pci/pcie/portdrv_core.c | 9 ++--- 5 files changed, 56 insertions(+), 21 deletions(-)drivers/acpi/pci_root.c +0
-0drivers/acpi/pci_root.c static struct pci_osc_bit_struct pci_osc_control_bit[] = { · line 150
Quoted 6 lines
drivers/acpi/pci_root.c +3
-0@@ -145,6 +145,17 @@ static struct pci_osc_bit_struct pci_osc_control_bit[] = { { OSC_PCI_EXPRESS_DPC_CONTROL, "DPC" }, }; + else + dev_warn(&bus->dev, "OS overrides %s firmware control", + get_osc_desc(OSC_PCI_EXPRESS_NATIVE_HP_CONTROL));There's got to be a way to write this more concisely. Maybe something like this?
#define OSC_OWNER(ctrl, bit, flag) \ if (!(ctrl & bit)) \ flag = 0; if (pcie_ports_native) decode_osc_control(root, "OS forcibly taking over", ~0);
Quoted 8 lines
Quoted 3 lines
BIT1 and BIT6 does not have PCIe dependency. And BIT7-31 are reserved. So we can't force all
_OSCbits based onpcie_ports_nativevalue. So, IM0, its better to handle PCIe features seperatly.Yes, we may need to handle a few bits specially. But we need to figure out a nicer-looking way of coding this. It's too cumbersome to check
pcie_ports_nativeand log a message for every_OSCbit individually.
ok. Let me check how to simplify it.
Quoted 12 lines
Quoted 12 lines
Quoted 7 lines
else { ctrl = root->osc_control_set; OSC_OWNER(ctrl, OSC_PCI_EXPRESS_AER_CONTROL, host_bridge->native_aer); OSC_OWNER(ctrl, OSC_PCI_EXPRESS_PME_CONTROL, host_bridge->native_pme); ... }Signature · 2 lines
Sathyanarayanan Kuppuswamy Linux Kernel Developer
Signature · 2 lines
Sathyanarayanan Kuppuswamy Linux Kernel Developer
PATCH v8 3/5 ACPI/PCI: Ignore _OSC DPC negotiation result if pcie_ports_dpc_native is set.
In reply to [email protected]From: Kuppuswamy Sathyanarayanan <[email protected]>
pcie_ports_dpc_native is set only if user requests native handling of PCIe DPC capability via pcie_port_setup command line option. User input takes precedence over _OSC based control negotiation result. So consider the _OSC negotiated result for DPC ownership only if pcie_ports_dpc_native is unset.
--- drivers/acpi/pci_root.c | 2 +- drivers/pci/pcie/dpc.c | 3 ++- drivers/pci/pcie/portdrv.h | 2 -- drivers/pci/pcie/portdrv_core.c | 2 +- include/linux/pci.h | 2 ++ 5 files changed, 6 insertions(+), 5 deletions(-)
drivers/acpi/pci_root.c +1 -1
@@ -961,7 +961,7 @@ struct pci_bus *acpi_pci_root_create(struct acpi_pci_root *root, }if (!(root->osc_control_set &OSC_PCI_EXPRESS_DPC_CONTROL )) { -if (!pcie_ports_native) +if (!pcie_ports_native && !pcie_ports_dpc_native) host_bridge->native_dpc =0 ;else dev_warn(&bus->dev, "OS overrides %s firmware control",
drivers/pci/pcie/dpc.c +2 -1
@@ -280,11 +280,12 @@ void pci_dpc_init(struct pci_dev *pdev)static int dpc_probe (struct pcie_device *dev) {struct pci_dev *pdev = dev->port; +struct pci_host_bridge *host =pci_find_host_bridge (pdev->bus);struct device *device = &dev->device;int status;u16 ctl, cap; -if (!pcie_aer_is_native (pdev) && !pcie_ports_dpc_native) +if (!pcie_aer_is_native (pdev) && !host->native_dpc)return -ENOTSUPP ; status = devm_request_threaded_irq(device, dev->irq, dpc_irq,
drivers/pci/pcie/portdrv.h +0 -2
@@ -25,8 +25,6 @@#define PCIE_PORT_DEVICE_MAXSERVICES 5 -extern bool pcie_ports_dpc_native; -#ifdef CONFIG_PCIEAER int pcie_aer_init (void );int pcie_aer_is_native (struct pci_dev *dev);
drivers/pci/pcie/portdrv_core.c +1 -1
@@ -253,7 +253,7 @@ static int get_port_device_capability(struct pci_dev *dev)*/ if (pci_find_ext_capability (dev,PCI_EXT_CAP_ID_DPC ) &&pci_aer_available () && - (pcie_ports_dpc_native || (services &PCIE_PORT_SERVICE_AER ))) + (host->native_dpc || (services &PCIE_PORT_SERVICE_AER ))) services |=PCIE_PORT_SERVICE_DPC ; if (pci_pcie_type(dev) == PCI_EXP_TYPE_DOWNSTREAM ||
include/linux/pci.h +2 -0
@@ -1556,9 +1556,11 @@ static inline int pci_irqd_intx_xlate(struct irq_domain *d,#ifdef CONFIG_PCIEPORTBUS extern bool pcie_ports_disabled;extern bool pcie_ports_native; +extern bool pcie_ports_dpc_native;#else #define pcie_ports_disabled true#define pcie_ports_native false +#define pcie_ports_dpc_native false#endif #define PCIE_LINK_STATE_L0S BIT(0)
Signature · 1 line
2.17.1
PATCH v8 4/5 PCI/portdrv: Remove redundant pci_aer_available() check in DPC enable logic
In reply to [email protected]From: Kuppuswamy Sathyanarayanan <[email protected]>
In DPC service enable logic, check for
services & PCIE_PORT_SERVICE_AER implies pci_aer_available() is true. So there is no need to explicitly check it again.
Also, passing pcie_ports=dpc-native in kernel command line implies DPC needs to be enabled in native mode irrespective of AER ownership status. So checking for pci_aer_available() without checking for pcie_ports status is incorrect.
--- drivers/pci/pcie/portdrv_core.c | 1 - 1 file changed, 1 deletion(-)
drivers/pci/pcie/portdrv_core.c +0 -1
@@ -252,7 +252,6 @@ static int get_port_device_capability(struct pci_dev *dev)* permission to use AER. */ if (pci_find_ext_capability (dev,PCI_EXT_CAP_ID_DPC ) && -pci_aer_available () && (host->native_dpc || (services &PCIE_PORT_SERVICE_AER ))) services |=PCIE_PORT_SERVICE_DPC ;
Signature · 1 line
2.17.1
PATCH v8 5/5 PCI/DPC: Move AER/DPC dependency checks out of DPC driver
In reply to [email protected]From: Kuppuswamy Sathyanarayanan <[email protected]>
Currently, AER and DPC Capabilities dependency checks is distributed between DPC and portdrv service drivers. So move them out of DPC driver.
Also, since services & PCIE_PORT_SERVICE_AER check already ensures AER native ownership, no need to add additional pcie_aer_is_native() check.
--- drivers/pci/pcie/dpc.c | 3 --- drivers/pci/pcie/portdrv_core.c | 1 + 2 files changed, 1 insertion(+), 3 deletions(-)
drivers/pci/pcie/dpc.c +0 -3
@@ -285,9 +285,6 @@ static int dpc_probe(struct pcie_device *dev)int status;u16 ctl, cap; -if (!pcie_aer_is_native (pdev) && !host->native_dpc) -return -ENOTSUPP ; - status =devm_request_threaded_irq (device, dev->irq, dpc_irq, dpc_handler,IRQF_SHARED ,"pcie-dpc" , pdev);
drivers/pci/pcie/portdrv_core.c +1 -0
@@ -252,6 +252,7 @@ static int get_port_device_capability(struct pci_dev *dev)* permission to use AER. */ if (pci_find_ext_capability (dev,PCI_EXT_CAP_ID_DPC ) && + host->native_dpc && (host->native_dpc || (services &PCIE_PORT_SERVICE_AER ))) services |=PCIE_PORT_SERVICE_DPC ;
Signature · 1 line
2.17.1
Re: [PATCH v8 5/5] PCI/DPC: Move AER/DPC dependency checks out of DPC driver
Quoted 4 lines · On Sat, Jul 25, 2020 at 7:01 AM <[email protected]>…
From: Kuppuswamy Sathyanarayanan <[email protected]>
Currently, AER and DPC Capabilities dependency checks is
are
Quoted 6 lines
distributed between DPC and portdrv service drivers. So move them out of DPC driver.
Also, since services &
PCIE_PORT_SERVICE_AERcheck already ensures AER native ownership, no need to add additionalpcie_aer_is_native()check.
...
Quoted 4 lines
if (pci_find_ext_capability(dev, PCI_EXT_CAP_ID_DPC) &&
+ host->native_dpc &&
(host->native_dpc || (services & PCIE_PORT_SERVICE_AER)))
services |= PCIE_PORT_SERVICE_DPC;Can you elaborate this change, please?
Signature · 2 lines
With Best Regards, Andy Shevchenko
Hi Bjorn,
Quoted 8 lines · On 7/24/20 8:58 PM, [email protected] wrote:
From: Kuppuswamy Sathyanarayanan <[email protected]>
Currently, PCIe capabilities ownership status is detected by verifying the status of
pcie_ports_native,pcie_ports_dpc_nativeand_OSCnegotiated results (cached in structpci_host_bridge->native_* members). But this logic can be simplified, and we can use only structpci_host_bridge->native_* members to detect it.
Any comments on this patch set?
Signature · 2 lines
Sathyanarayanan Kuppuswamy Linux Kernel Developer