Skip to contentSkip to conversation

PCI: fix pci-mvebu after conversion to common bridge emul code

PATCH Applied Thomas Petazzoni 5 participants 12 messages Feb 20, 2019 – Feb 22, 2019

Subsystems (4)
lore
Share & tools
lore

Thomas Petazzoni

author maintainer
#0 lore raw

cover PCI: fix pci-mvebu after conversion to common bridge emul code

Hello,

This set of two patches aim at fixing some regression reported by two users (Luis and Leigh), that were introduced by the conversion of the pci-mvebu driver to the common PCI bridge emulation code shared between pci-aardvark and pci-mvebu.

Due to this conversion, some registers of the PCI configuration that used to be read-only are now read-write, making the Linux PCI core believe that some features are implemented by the PCI bridge. Namely in the case of pci-mvebu, the prefetchable memory base/limit registers were read-only, while they are now read-write. Due to this, the Linux PCI core now believes it can configure a prefetchable memory area, but this is not supported by pci-mvebu, which does not have the logic to create the appropriate MBUs windows.

This set of two commits allow PCI controller code to tell the PCI bridge emulation logic about their capabilities. Because we envision that other capabilities than prefetchable memory support might need to be communicated from the PCI controller code to the PCI bridge emulation code in the future, we introduce a "flags" argument, even if for now only one flag is supported.

Lorenzo, let me know what you think about this approach. I am open to suggestions if the proposed approach is not satisfying.

Both patches have been confirmed by Luis and Leigh to fix the regression, but I'll let them give a formal Tested-by.

Best regards,

Thomas Petazzoni

Thomas Petazzoni (2):
PCI: pci-bridge-emul: Create per-bridge copy of register behavior
PCI: pci-bridge-emul: Extend pci_bridge_emul_init() with flags

 drivers/pci/controller/pci-aardvark.c |  2 +-
 drivers/pci/controller/pci-mvebu.c    |  2 +-
 drivers/pci/pci-bridge-emul.c         | 86 ++++++++++++++++++---------
 drivers/pci/pci-bridge-emul.h         | 13 +++-
 4 files changed, 73 insertions(+), 30 deletions(-)
Signature · 7 lines
2.20.1


_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
author maintainer
#1 lore raw

PATCH 1/2 PCI: pci-bridge-emul: Create per-bridge copy of register behavior

The behavior of the different registers of the PCI-to-PCI bridge is currently encoded in two global arrays, shared by all instances of PCI-to-PCI bridge emulation.

However, we will need to tweak the behavior on a per-bridge basis, to accommodate for different capabilities of the platforms where this code is used. In preparation for this, this commit creates a per-bridge copy of the register behavior arrays, so that they can later be tweaked on a per-bridge basis.

Cc Luís Mendes <[email protected]>
Cc Leigh Brown <[email protected]>
Fixes 1f08673eef123 ("PCI: mvebu: Convert to PCI emulated bridge config space")
---
 drivers/pci/pci-bridge-emul.c | 80 +++++++++++++++++++++++------------
 drivers/pci/pci-bridge-emul.h |  8 +++-
 2 files changed, 60 insertions(+), 28 deletions(-)
drivers/pci/pci-bridge-emul.c +53 -27
index 129738362d90..dd8d8060317e 100644
@@ -24,29 +24,6 @@
 #define PCI_CAP_PCIE_START	PCI_BRIDGE_CONF_END
 #define PCI_CAP_PCIE_END	(PCI_CAP_PCIE_START + PCI_EXP_SLTSTA2 + 2)
 
-/*
- * Initialize a pci_bridge_emul structure to represent a fake PCI
- * bridge configuration space. The caller needs to have initialized
- * the PCI configuration space with whatever values make sense
- * (typically at least vendor, device, revision), the ->ops pointer,
- * and optionally ->data and ->has_pcie.
- */
-void pci_bridge_emul_init(struct pci_bridge_emul *bridge)
-{
-	bridge->conf.class_revision |= PCI_CLASS_BRIDGE_PCI << 16;
-	bridge->conf.header_type = PCI_HEADER_TYPE_BRIDGE;
-	bridge->conf.cache_line_size = 0x10;
-	bridge->conf.status = PCI_STATUS_CAP_LIST;
-
-	if (bridge->has_pcie) {
-		bridge->conf.capabilities_pointer = PCI_CAP_PCIE_START;
-		bridge->pcie_conf.cap_id = PCI_CAP_ID_EXP;
-		/* Set PCIe v2, root port, slot support */
-		bridge->pcie_conf.cap = PCI_EXP_TYPE_ROOT_PORT << 4 | 2 |
-			PCI_EXP_FLAGS_SLOT;
-	}
-}
-
 struct pci_bridge_reg_behavior {
 	/* Read-only bits */
 	u32 ro;
@@ -283,6 +260,55 @@ const static struct pci_bridge_reg_behavior pcie_cap_regs_behavior[] = {
 	},
 };
 
+/*
+ * Initialize a pci_bridge_emul structure to represent a fake PCI
+ * bridge configuration space. The caller needs to have initialized
+ * the PCI configuration space with whatever values make sense
+ * (typically at least vendor, device, revision), the ->ops pointer,
+ * and optionally ->data and ->has_pcie.
+ */
+int pci_bridge_emul_init(struct pci_bridge_emul *bridge)
+{
+	bridge->conf.class_revision |= PCI_CLASS_BRIDGE_PCI << 16;
+	bridge->conf.header_type = PCI_HEADER_TYPE_BRIDGE;
+	bridge->conf.cache_line_size = 0x10;
+	bridge->conf.status = PCI_STATUS_CAP_LIST;
+	bridge->pci_regs_behavior = kmemdup(pci_regs_behavior,
+					    sizeof(pci_regs_behavior),
+					    GFP_KERNEL);
+	if (!bridge->pci_regs_behavior)
+		return -ENOMEM;
+
+	if (bridge->has_pcie) {
+		bridge->conf.capabilities_pointer = PCI_CAP_PCIE_START;
+		bridge->pcie_conf.cap_id = PCI_CAP_ID_EXP;
+		/* Set PCIe v2, root port, slot support */
+		bridge->pcie_conf.cap = PCI_EXP_TYPE_ROOT_PORT << 4 | 2 |
+			PCI_EXP_FLAGS_SLOT;
+		bridge->pcie_cap_regs_behavior =
+			kmemdup(pcie_cap_regs_behavior,
+				sizeof(pcie_cap_regs_behavior),
+				GFP_KERNEL);
+		if (!bridge->pcie_cap_regs_behavior) {
+			kfree(bridge->pci_regs_behavior);
+			return -ENOMEM;
+		}
+	}
+
+	return 0;
+}
+
+/*
+ * Cleanup a pci_bridge_emul structure that was previously initilized
+ * using pci_bridge_emul_init().
+ */
+void pci_bridge_emul_cleanup(struct pci_bridge_emul *bridge)
+{
+	if (bridge->has_pcie)
+		kfree(bridge->pcie_cap_regs_behavior);
+	kfree(bridge->pci_regs_behavior);
+}
+
 /*
  * Should be called by the PCI controller driver when reading the PCI
  * configuration space of the fake bridge. It will call back the
@@ -312,11 +338,11 @@ int pci_bridge_emul_conf_read(struct pci_bridge_emul *bridge, int where,
 		reg -= PCI_CAP_PCIE_START;
 		read_op = bridge->ops->read_pcie;
 		cfgspace = (u32 *) &bridge->pcie_conf;
-		behavior = pcie_cap_regs_behavior;
+		behavior = bridge->pcie_cap_regs_behavior;
 	} else {
 		read_op = bridge->ops->read_base;
 		cfgspace = (u32 *) &bridge->conf;
-		behavior = pci_regs_behavior;
+		behavior = bridge->pci_regs_behavior;
 	}
 
 	if (read_op)
@@ -383,11 +409,11 @@ int pci_bridge_emul_conf_write(struct pci_bridge_emul *bridge, int where,
 		reg -= PCI_CAP_PCIE_START;
 		write_op = bridge->ops->write_pcie;
 		cfgspace = (u32 *) &bridge->pcie_conf;
-		behavior = pcie_cap_regs_behavior;
+		behavior = bridge->pcie_cap_regs_behavior;
 	} else {
 		write_op = bridge->ops->write_base;
 		cfgspace = (u32 *) &bridge->conf;
-		behavior = pci_regs_behavior;
+		behavior = bridge->pci_regs_behavior;
 	}
 
 	/* Keep all bits, except the RW bits */
drivers/pci/pci-bridge-emul.h +7 -1
index 9d510ccf738b..f04637bb3222 100644
@@ -107,15 +107,21 @@ struct pci_bridge_emul_ops {
 			   u32 old, u32 new, u32 mask);
 };
 
+struct pci_bridge_reg_behavior;
+
 struct pci_bridge_emul {
 	struct pci_bridge_emul_conf conf;
 	struct pci_bridge_emul_pcie_conf pcie_conf;
 	struct pci_bridge_emul_ops *ops;
+	struct pci_bridge_reg_behavior *pci_regs_behavior;
+	struct pci_bridge_reg_behavior *pcie_cap_regs_behavior;
 	void *data;
 	bool has_pcie;
 };
 
-void pci_bridge_emul_init(struct pci_bridge_emul *bridge);
+int pci_bridge_emul_init(struct pci_bridge_emul *bridge);
+void pci_bridge_emul_cleanup(struct pci_bridge_emul *bridge);
+
 int pci_bridge_emul_conf_read(struct pci_bridge_emul *bridge, int where,
 			      int size, u32 *value);
 int pci_bridge_emul_conf_write(struct pci_bridge_emul *bridge, int where,
Signature · 1 line
2.20.1
Tested-by on v1 1/2
#2 lore raw

Re: [PATCH 1/2] PCI: pci-bridge-emul: Create per-bridge copy of register behavior

Tested patch against Vanilla kernels 4.20.11 and 5.0-rc7. It is working properly on both kernels.

Quoted 17 lines · Thomas Petazzoni

The behavior of the different registers of the PCI-to-PCI bridge is currently encoded in two global arrays, shared by all instances of PCI-to-PCI bridge emulation.

However, we will need to tweak the behavior on a per-bridge basis, to accommodate for different capabilities of the platforms where this code is used. In preparation for this, this commit creates a per-bridge copy of the register behavior arrays, so that they can later be tweaked on a per-bridge basis.

Cc Luís Mendes <[email protected]>
Cc Leigh Brown <[email protected]>
Fixes 1f08673eef123 ("PCI: mvebu: Convert to PCI emulated bridge config space")
author maintainer
#3 lore raw

PATCH 2/2 PCI: pci-bridge-emul: Extend pci_bridge_emul_init() with flags

In reply to Thomas Petazzoni

Depending on the capabilities of the PCI controller/platform, the PCI-to-PCI bridge emulation behavior might need to be different. For example, on platforms that use the pci-mvebu code, we currently don't support prefetchable memory BARs, so the corresponding fields in the PCI-to-PCI bridge configuration space should be read-only.

To implement this, this commit extends pci_bridge_emul_init() to take a "flags" argument, with currently one flag supported:
PCI_BRIDGE_EMUL_NO_PREFETCHABLE_BAR, that will make the prefetchable memory base and limit registers read-only.

The pci-mvebu and pci-aardvark drivers are updated accordingly.

Cc Luís Mendes <[email protected]>
Cc Leigh Brown <[email protected]>
Fixes 1f08673eef123 ("PCI: mvebu: Convert to PCI emulated bridge config space")
---
 drivers/pci/controller/pci-aardvark.c | 2 +-
 drivers/pci/controller/pci-mvebu.c    | 2 +-
 drivers/pci/pci-bridge-emul.c         | 8 +++++++-
 drivers/pci/pci-bridge-emul.h         | 7 ++++++-
 4 files changed, 15 insertions(+), 4 deletions(-)
drivers/pci/controller/pci-aardvark.c +1 -1
index 750081c1cb48..6eecae447af3 100644
@@ -499,7 +499,7 @@ static void advk_sw_pci_bridge_init(struct advk_pcie *pcie)
 	bridge->data = pcie;
 	bridge->ops = &advk_pci_bridge_emul_ops;
 
-	pci_bridge_emul_init(bridge);
+	pci_bridge_emul_init(bridge, 0);
 
 }
 
drivers/pci/controller/pci-mvebu.c +1 -1
index fa0fc46edb0c..d3a0419e42f2 100644
@@ -583,7 +583,7 @@ static void mvebu_pci_bridge_emul_init(struct mvebu_pcie_port *port)
 	bridge->data = port;
 	bridge->ops = &mvebu_pci_bridge_emul_ops;
 
-	pci_bridge_emul_init(bridge);
+	pci_bridge_emul_init(bridge, PCI_BRIDGE_EMUL_NO_PREFETCHABLE_BAR);
 }
 
 static inline struct mvebu_pcie *sys_to_pcie(struct pci_sys_data *sys)
drivers/pci/pci-bridge-emul.c +7 -1
index dd8d8060317e..83fb077d0b41 100644
@@ -267,7 +267,8 @@ const static struct pci_bridge_reg_behavior pcie_cap_regs_behavior[] = {
  * (typically at least vendor, device, revision), the ->ops pointer,
  * and optionally ->data and ->has_pcie.
  */
-int pci_bridge_emul_init(struct pci_bridge_emul *bridge)
+int pci_bridge_emul_init(struct pci_bridge_emul *bridge,
+			 unsigned int flags)
 {
 	bridge->conf.class_revision |= PCI_CLASS_BRIDGE_PCI << 16;
 	bridge->conf.header_type = PCI_HEADER_TYPE_BRIDGE;
@@ -295,6 +296,11 @@ int pci_bridge_emul_init(struct pci_bridge_emul *bridge)
 		}
 	}
 
+	if (flags & PCI_BRIDGE_EMUL_NO_PREFETCHABLE_BAR) {
+		bridge->pci_regs_behavior[PCI_PREF_MEMORY_BASE / 4].ro = ~0;
+		bridge->pci_regs_behavior[PCI_PREF_MEMORY_BASE / 4].rw = 0;
+	}
+
 	return 0;
 }
 
drivers/pci/pci-bridge-emul.h +6 -1
index f04637bb3222..e65b1b79899d 100644
@@ -119,7 +119,12 @@ struct pci_bridge_emul {
 	bool has_pcie;
 };
 
-int pci_bridge_emul_init(struct pci_bridge_emul *bridge);
+enum {
+	PCI_BRIDGE_EMUL_NO_PREFETCHABLE_BAR = BIT(0),
+};
+
+int pci_bridge_emul_init(struct pci_bridge_emul *bridge,
+			 unsigned int flags);
 void pci_bridge_emul_cleanup(struct pci_bridge_emul *bridge);
 
 int pci_bridge_emul_conf_read(struct pci_bridge_emul *bridge, int where,
Signature · 7 lines
2.20.1


_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
Tested-by on v1 2/2
#4 lore raw

Re: [PATCH 2/2] PCI: pci-bridge-emul: Extend pci_bridge_emul_init() with flags

Tested patch against Vanilla kernels 4.20.11 and 5.0-rc7. It is working properly on both kernels.

Quoted 27 lines · Thomas Petazzoni

Depending on the capabilities of the PCI controller/platform, the PCI-to-PCI bridge emulation behavior might need to be different. For example, on platforms that use the pci-mvebu code, we currently don't support prefetchable memory BARs, so the corresponding fields in the PCI-to-PCI bridge configuration space should be read-only.

To implement this, this commit extends pci_bridge_emul_init() to take a "flags" argument, with currently one flag supported:
PCI_BRIDGE_EMUL_NO_PREFETCHABLE_BAR, that will make the prefetchable memory base and limit registers read-only.

The pci-mvebu and pci-aardvark drivers are updated accordingly.

Cc Luís Mendes <[email protected]>
Cc Leigh Brown <[email protected]>
Fixes 1f08673eef123 ("PCI: mvebu: Convert to PCI emulated bridge config space")
---
 drivers/pci/controller/pci-aardvark.c | 2 +-
 drivers/pci/controller/pci-mvebu.c    | 2 +-
 drivers/pci/pci-bridge-emul.c         | 8 +++++++-
 drivers/pci/pci-bridge-emul.h         | 7 ++++++-
 4 files changed, 15 insertions(+), 4 deletions(-)
Acked-by on the v1 series
#5 lore raw
In reply to Thomas Petazzoni
Quoted 37 lines · Thomas Petazzoni

Hello,

This set of two patches aim at fixing some regression reported by two users (Luis and Leigh), that were introduced by the conversion of the pci-mvebu driver to the common PCI bridge emulation code shared between pci-aardvark and pci-mvebu.

Due to this conversion, some registers of the PCI configuration that used to be read-only are now read-write, making the Linux PCI core believe that some features are implemented by the PCI bridge. Namely in the case of pci-mvebu, the prefetchable memory base/limit registers were read-only, while they are now read-write. Due to this, the Linux PCI core now believes it can configure a prefetchable memory area, but this is not supported by pci-mvebu, which does not have the logic to create the appropriate MBUs windows.

This set of two commits allow PCI controller code to tell the PCI bridge emulation logic about their capabilities. Because we envision that other capabilities than prefetchable memory support might need to be communicated from the PCI controller code to the PCI bridge emulation code in the future, we introduce a "flags" argument, even if for now only one flag is supported.

Lorenzo, let me know what you think about this approach. I am open to suggestions if the proposed approach is not satisfying.

Both patches have been confirmed by Luis and Leigh to fix the regression, but I'll let them give a formal Tested-by.

Best regards,

Thomas Petazzoni

Thomas Petazzoni (2):
PCI: pci-bridge-emul: Create per-bridge copy of register behavior
PCI: pci-bridge-emul: Extend pci_bridge_emul_init() with flags

Quoted 5 lines
 drivers/pci/controller/pci-aardvark.c |  2 +-
 drivers/pci/controller/pci-mvebu.c    |  2 +-
 drivers/pci/pci-bridge-emul.c         | 86 ++++++++++++++++++---------
 drivers/pci/pci-bridge-emul.h         | 13 +++-
 4 files changed, 73 insertions(+), 30 deletions(-)

Lorenzo, I assume you'll take both of these, so for both patches,

Maybe we should add a stable tag (v4.20+)?

#6 lore raw
Quoted 45 lines · Bjorn Helgaas
Quoted 37 lines · Thomas Petazzoni

Hello,

This set of two patches aim at fixing some regression reported by two users (Luis and Leigh), that were introduced by the conversion of the pci-mvebu driver to the common PCI bridge emulation code shared between pci-aardvark and pci-mvebu.

Due to this conversion, some registers of the PCI configuration that used to be read-only are now read-write, making the Linux PCI core believe that some features are implemented by the PCI bridge. Namely in the case of pci-mvebu, the prefetchable memory base/limit registers were read-only, while they are now read-write. Due to this, the Linux PCI core now believes it can configure a prefetchable memory area, but this is not supported by pci-mvebu, which does not have the logic to create the appropriate MBUs windows.

This set of two commits allow PCI controller code to tell the PCI bridge emulation logic about their capabilities. Because we envision that other capabilities than prefetchable memory support might need to be communicated from the PCI controller code to the PCI bridge emulation code in the future, we introduce a "flags" argument, even if for now only one flag is supported.

Lorenzo, let me know what you think about this approach. I am open to suggestions if the proposed approach is not satisfying.

Both patches have been confirmed by Luis and Leigh to fix the regression, but I'll let them give a formal Tested-by.

Best regards,

Thomas Petazzoni

Thomas Petazzoni (2):
PCI: pci-bridge-emul: Create per-bridge copy of register behavior
PCI: pci-bridge-emul: Extend pci_bridge_emul_init() with flags

Quoted 6 lines
 drivers/pci/controller/pci-aardvark.c |  2 +-
 drivers/pci/controller/pci-mvebu.c    |  2 +-
 drivers/pci/pci-bridge-emul.c         | 86 
++++++++++++++++++---------
 drivers/pci/pci-bridge-emul.h         | 13 +++-
 4 files changed, 73 insertions(+), 30 deletions(-)
Quoted 5 lines

Lorenzo, I assume you'll take both of these, so for both patches,

Maybe we should add a stable tag (v4.20+)?

That would be great, as it would get that kernel version working again.

Regards,

Leigh.

author maintainer
#7 lore raw
In reply to Bjorn Helgaas

Hello Bjorn,

Quoted 3 lines · On Thu, 21 Feb 2019 17:23:17 -0600 Bjorn Helgaas <[email protected]> wrote:

Lorenzo, I assume you'll take both of these, so for both patches,

Thanks!

Quoted 1 line

Maybe we should add a stable tag (v4.20+)?

Both patches have a Fixes: tag, so they should be automatically be backported to the applicable stable kernels. From Documentation/process/submitting-patches.rst.

A Fixes: tag indicates that the patch fixes an issue in a previous commit. It is used to make it easy to determine where a bug originated, which can help review a bug fix. This tag also assists the stable kernel team in determining which stable kernel versions should receive your fix. This is the preferred method for indicating a bug fixed by the patch. See :ref:describe_changes for more details.

Isn't that sufficient ?

Thanks,

Thomas

Signature · 3 lines
Thomas Petazzoni, CTO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com
#8 lore raw
Quoted 25 lines · Thomas Petazzoni

Hello Bjorn,

Quoted 3 lines · On Thu, 21 Feb 2019 17:23:17 -0600 Bjorn Helgaas <[email protected]> wrote:

Lorenzo, I assume you'll take both of these, so for both patches,

Thanks!

Quoted 1 line

Maybe we should add a stable tag (v4.20+)?

Both patches have a Fixes: tag, so they should be automatically be backported to the applicable stable kernels. From Documentation/process/submitting-patches.rst.

A Fixes: tag indicates that the patch fixes an issue in a previous commit. It is used to make it easy to determine where a bug originated, which can help review a bug fix. This tag also assists the stable kernel team in determining which stable kernel versions should receive your fix. This is the preferred method for indicating a bug fixed by the patch. See :ref:describe_changes for more details.

Quoted 1 line

Isn't that sufficient ?

Adding a CC: stable won't hurt so that it is certain it will get stable kernels attention, that's what I will do.

Thanks,
Lorenzo

author maintainer
#9 lore raw

Hello Lorenzo,

Quoted 4 lines · On Fri, 22 Feb 2019 11:05:12 +0000 Lorenzo Pieralisi…
Quoted 1 line

Isn't that sufficient ?

Adding a CC: stable won't hurt so that it is certain it will get stable kernels attention, that's what I will do.

Ah, yes, indeed. Sorry for forgetting about these, and thanks for applying with those Cc: stable added!

Thomas

Signature · 3 lines
Thomas Petazzoni, CTO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com
Tested-by on the v1 series
#10 lore raw
In reply to Thomas Petazzoni
Quoted 37 lines · Thomas Petazzoni

Hello,

This set of two patches aim at fixing some regression reported by two users (Luis and Leigh), that were introduced by the conversion of the pci-mvebu driver to the common PCI bridge emulation code shared between pci-aardvark and pci-mvebu.

Due to this conversion, some registers of the PCI configuration that used to be read-only are now read-write, making the Linux PCI core believe that some features are implemented by the PCI bridge. Namely in the case of pci-mvebu, the prefetchable memory base/limit registers were read-only, while they are now read-write. Due to this, the Linux PCI core now believes it can configure a prefetchable memory area, but this is not supported by pci-mvebu, which does not have the logic to create the appropriate MBUs windows.

This set of two commits allow PCI controller code to tell the PCI bridge emulation logic about their capabilities. Because we envision that other capabilities than prefetchable memory support might need to be communicated from the PCI controller code to the PCI bridge emulation code in the future, we introduce a "flags" argument, even if for now only one flag is supported.

Lorenzo, let me know what you think about this approach. I am open to suggestions if the proposed approach is not satisfying.

Both patches have been confirmed by Luis and Leigh to fix the regression, but I'll let them give a formal Tested-by.

Best regards,

Thomas Petazzoni

Thomas Petazzoni (2):
PCI: pci-bridge-emul: Create per-bridge copy of register behavior
PCI: pci-bridge-emul: Extend pci_bridge_emul_init() with flags

Quoted 5 lines
 drivers/pci/controller/pci-aardvark.c |  2 +-
 drivers/pci/controller/pci-mvebu.c    |  2 +-
 drivers/pci/pci-bridge-emul.c         | 86 ++++++++++++++++++---------
 drivers/pci/pci-bridge-emul.h         | 13 +++-
 4 files changed, 73 insertions(+), 30 deletions(-)

Tested on WRT1900ACS router against 4.20.7 and 5.0.0-rc.

Regards,

Leigh.

#11 lore raw
In reply to Thomas Petazzoni
Quoted 37 lines · Thomas Petazzoni

Hello,

This set of two patches aim at fixing some regression reported by two users (Luis and Leigh), that were introduced by the conversion of the pci-mvebu driver to the common PCI bridge emulation code shared between pci-aardvark and pci-mvebu.

Due to this conversion, some registers of the PCI configuration that used to be read-only are now read-write, making the Linux PCI core believe that some features are implemented by the PCI bridge. Namely in the case of pci-mvebu, the prefetchable memory base/limit registers were read-only, while they are now read-write. Due to this, the Linux PCI core now believes it can configure a prefetchable memory area, but this is not supported by pci-mvebu, which does not have the logic to create the appropriate MBUs windows.

This set of two commits allow PCI controller code to tell the PCI bridge emulation logic about their capabilities. Because we envision that other capabilities than prefetchable memory support might need to be communicated from the PCI controller code to the PCI bridge emulation code in the future, we introduce a "flags" argument, even if for now only one flag is supported.

Lorenzo, let me know what you think about this approach. I am open to suggestions if the proposed approach is not satisfying.

Both patches have been confirmed by Luis and Leigh to fix the regression, but I'll let them give a formal Tested-by.

Best regards,

Thomas Petazzoni

Thomas Petazzoni (2):
PCI: pci-bridge-emul: Create per-bridge copy of register behavior
PCI: pci-bridge-emul: Extend pci_bridge_emul_init() with flags

Quoted 5 lines
 drivers/pci/controller/pci-aardvark.c |  2 +-
 drivers/pci/controller/pci-mvebu.c    |  2 +-
 drivers/pci/pci-bridge-emul.c         | 86 ++++++++++++++++++---------
 drivers/pci/pci-bridge-emul.h         | 13 +++-
 4 files changed, 73 insertions(+), 30 deletions(-)

I have applied the series to pci/misc for v5.1, thanks.

Lorenzo