[PATCH 0/6] PCI: fix pci_remap_iospace() failure paths

[PATCH 0/6] PCI: fix pci_remap_iospace() failure paths

[Lorenzo Pieralisi]

On ARM/ARM64 architectures, PCI IO ports are emulated through memory
mapped IO, by reserving a chunk of virtual address space starting at
PCI_IOBASE and by mapping the PCI host bridges memory address space
driving PCI IO cycles to it.

PCI host bridge drivers that enable downstream PCI IO cycles map the
host bridge memory address responding to PCI IO cycles to the fixed
virtual address space through the pci_remap_iospace() API.

This means that if the pci_remap_iospace() function fails, the
corresponding host bridge PCI IO resource must be considered invalid, in
that there is no way for the kernel to actually drive PCI IO transactions
if the memory addresses responding to PCI IO cycles cannot be mapped into
the CPU virtual address space.

As a follow-up to the linux-pci mailing list thread ([0]), this series
fixes the pci_remap_iospace() failure paths by adding code that handles
the failures and takes the required actions.

Compile tested on all affected PCI host controllers, testing and
reviewing much appreciated.

[0] http://marc.info/?l=linux-pci&m=145816042427417&w=2

Lorenzo Pieralisi (6):
  drivers: pci: host: aardvark: fix pci_remap_iospace() failure path
  drivers: pci: host: designware: fix pci_remap_iospace() failure path
  drivers: pci: host: versatile: fix pci_remap_iospace() failure path
  drivers: pci: host: rcar: fix pci_remap_iospace() failure path
  drivers: pci: host: common: fix pci_remap_iospace() failure path
  drivers: pci: host: tegra: fix pci_remap_iospace() failure path

 drivers/pci/host/pci-aardvark.c    |  8 +++++---
 drivers/pci/host/pci-host-common.c |  8 +++++---
 drivers/pci/host/pci-tegra.c       |  7 +++++--
 drivers/pci/host/pci-versatile.c   |  8 +++++---
 drivers/pci/host/pcie-designware.c | 21 ++++++++++++---------
 drivers/pci/host/pcie-rcar.c       |  9 ++++++---
 6 files changed, 38 insertions(+), 23 deletions(-)

-- 
2.6.4

[PATCH 1/6] drivers: pci: host: aardvark: fix pci_remap_iospace() failure path

[Lorenzo Pieralisi]

On ARM/ARM64 architectures, PCI IO ports are emulated through memory
mapped IO, by reserving a chunk of virtual address space starting at
PCI_IOBASE and by mapping the PCI host bridges memory address space
driving PCI IO cycles to it.

PCI host bridge drivers that enable downstream PCI IO cycles map the
host bridge memory address responding to PCI IO cycles to the fixed
virtual address space through the pci_remap_iospace() API.

This means that if the pci_remap_iospace() function fails, the
corresponding host bridge PCI IO resource must be considered invalid, in
that there is no way for the kernel to actually drive PCI IO
transactions if the memory addresses responding to PCI
IO cycles cannot be mapped into the CPU virtual address space.

The PCI aardvark host bridge driver does not remove the PCI IO
resource from the host bridge resource windows if the
pci_remap_iospace() call fails; this is an actual bug in that the
PCI host bridge would consider the PCI IO resource valid (and possibly
assign it to downstream devices) even if the kernel was not able
to map the PCI host bridge memory address driving IO cycle to the
CPU virtual address space (ie pci_remap_iospace() failures).

Fix the PCI host bridge driver pci_remap_iospace() failure path, by
destroying the PCI host bridge PCI IO resources retrieved through
firmware when the pci_remap_iospace() function call fails, therefore
preventing the kernel from adding the respective PCI IO resource to the
list of PCI host bridge valid resources, fixing the issue.

Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Fixes: 8c39d710363c ("PCI: aardvark: Add Aardvark PCI host controller driver")
Cc: Bjorn Helgaas <bhelgaas@google.com>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
---
 drivers/pci/host/pci-aardvark.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/drivers/pci/host/pci-aardvark.c b/drivers/pci/host/pci-aardvark.c
index ef9893f..4f5e567 100644
--- a/drivers/pci/host/pci-aardvark.c
+++ b/drivers/pci/host/pci-aardvark.c
@@ -848,7 +848,7 @@ static int advk_pcie_parse_request_of_pci_ranges(struct advk_pcie *pcie)
 	int err, res_valid = 0;
 	struct device *dev = &pcie->pdev->dev;
 	struct device_node *np = dev->of_node;
-	struct resource_entry *win;
+	struct resource_entry *win, *tmp;
 	resource_size_t iobase;
 
 	INIT_LIST_HEAD(&pcie->resources);
@@ -862,7 +862,7 @@ static int advk_pcie_parse_request_of_pci_ranges(struct advk_pcie *pcie)
 	if (err)
 		goto out_release_res;
 
-	resource_list_for_each_entry(win, &pcie->resources) {
+	resource_list_for_each_entry_safe(win, tmp, &pcie->resources) {
 		struct resource *res = win->res;
 
 		switch (resource_type(res)) {
@@ -874,9 +874,11 @@ static int advk_pcie_parse_request_of_pci_ranges(struct advk_pcie *pcie)
 					     lower_32_bits(res->start),
 					     OB_PCIE_IO);
 			err = pci_remap_iospace(res, iobase);
-			if (err)
+			if (err) {
 				dev_warn(dev, "error %d: failed to map resource %pR\n",
 					 err, res);
+				resource_list_destroy_entry(win);
+			}
 			break;
 		case IORESOURCE_MEM:
 			advk_pcie_set_ob_win(pcie, 0,
-- 
2.6.4

[PATCH 2/6] drivers: pci: host: designware: fix pci_remap_iospace() failure path

[Lorenzo Pieralisi]

On ARM/ARM64 architectures, PCI IO ports are emulated through memory
mapped IO, by reserving a chunk of virtual address space starting at
PCI_IOBASE and by mapping the PCI host bridges memory address space
driving PCI IO cycles to it.

PCI host bridge drivers that enable downstream PCI IO cycles map the
host bridge memory address responding to PCI IO cycles to the fixed
virtual address space through the pci_remap_iospace() API.

This means that if the pci_remap_iospace() function fails, the
corresponding host bridge PCI IO resource must be considered invalid, in
that there is no way for the kernel to actually drive PCI IO
transactions if the memory addresses responding to PCI
IO cycles cannot be mapped into the CPU virtual address space.

The PCI designware host bridge driver does not remove the PCI IO
resource from the host bridge resource windows if the
pci_remap_iospace() call fails; this is an actual bug in that the
PCI host bridge would consider the PCI IO resource valid (and possibly
assign it to downstream devices) even if the kernel was not able
to map the PCI host bridge memory address driving IO cycle to the
CPU virtual address space (ie pci_remap_iospace() failures).

Fix the PCI host bridge driver pci_remap_iospace() failure path, by
destroying the PCI host bridge PCI IO resources retrieved through
firmware when the pci_remap_iospace() function call fails, therefore
preventing the kernel from adding the respective PCI IO resource to the
list of PCI host bridge valid resources, fixing the issue.

Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Fixes: cbce7900598c ("PCI: designware: Make driver arch-agnostic")
Cc: Bjorn Helgaas <bhelgaas@google.com>
Cc: Jingoo Han <jingoohan1@gmail.com>
Cc: Pratyush Anand <pratyush.anand@gmail.com>
---
 drivers/pci/host/pcie-designware.c | 21 ++++++++++++---------
 1 file changed, 12 insertions(+), 9 deletions(-)

diff --git a/drivers/pci/host/pcie-designware.c b/drivers/pci/host/pcie-designware.c
index 12afce1..2a500f2 100644
--- a/drivers/pci/host/pcie-designware.c
+++ b/drivers/pci/host/pcie-designware.c
@@ -436,7 +436,7 @@ int dw_pcie_host_init(struct pcie_port *pp)
 	struct resource *cfg_res;
 	int i, ret;
 	LIST_HEAD(res);
-	struct resource_entry *win;
+	struct resource_entry *win, *tmp;
 
 	cfg_res = platform_get_resource_byname(pdev, IORESOURCE_MEM, "config");
 	if (cfg_res) {
@@ -457,17 +457,20 @@ int dw_pcie_host_init(struct pcie_port *pp)
 		goto error;
 
 	/* Get the I/O and memory ranges from DT */
-	resource_list_for_each_entry(win, &res) {
+	resource_list_for_each_entry_safe(win, tmp, &res) {
 		switch (resource_type(win->res)) {
 		case IORESOURCE_IO:
-			pp->io = win->res;
-			pp->io->name = "I/O";
-			pp->io_size = resource_size(pp->io);
-			pp->io_bus_addr = pp->io->start - win->offset;
-			ret = pci_remap_iospace(pp->io, pp->io_base);
-			if (ret)
+			ret = pci_remap_iospace(win->res, pp->io_base);
+			if (ret) {
 				dev_warn(pp->dev, "error %d: failed to map resource %pR\n",
-					 ret, pp->io);
+					 ret, win->res);
+				resource_list_destroy_entry(win);
+			} else {
+				pp->io = win->res;
+				pp->io->name = "I/O";
+				pp->io_size = resource_size(pp->io);
+				pp->io_bus_addr = pp->io->start - win->offset;
+			}
 			break;
 		case IORESOURCE_MEM:
 			pp->mem = win->res;
-- 
2.6.4

[PATCH 3/6] drivers: pci: host: versatile: fix pci_remap_iospace() failure path

[Lorenzo Pieralisi]

On ARM/ARM64 architectures, PCI IO ports are emulated through memory
mapped IO, by reserving a chunk of virtual address space starting at
PCI_IOBASE and by mapping the PCI host bridges memory address space
driving PCI IO cycles to it.

PCI host bridge drivers that enable downstream PCI IO cycles map the
host bridge memory address responding to PCI IO cycles to the fixed
virtual address space through the pci_remap_iospace() API.

This means that if the pci_remap_iospace() function fails, the
corresponding host bridge PCI IO resource must be considered invalid, in
that there is no way for the kernel to actually drive PCI IO
transactions if the memory addresses responding to PCI
IO cycles cannot be mapped into the CPU virtual address space.

The PCI versatile host bridge driver does not remove the PCI IO
resource from the host bridge resource windows if the
pci_remap_iospace() call fails; this is an actual bug in that the
PCI host bridge would consider the PCI IO resource valid (and possibly
assign it to downstream devices) even if the kernel was not able
to map the PCI host bridge memory address driving IO cycle to the
CPU virtual address space (ie pci_remap_iospace() failures).

Fix the PCI host bridge driver pci_remap_iospace() failure path, by
destroying the PCI host bridge PCI IO resources retrieved through
firmware when the pci_remap_iospace() function call fails, therefore
preventing the kernel from adding the respective PCI IO resource to the
list of PCI host bridge valid resources, fixing the issue.

Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Fixes: b7e78170efd4 ("PCI: versatile: Add DT-based ARM Versatile PB PCIe
host driver")
Cc: Bjorn Helgaas <bhelgaas@google.com>
Cc: Rob Herring <robh@kernel.org>
---
 drivers/pci/host/pci-versatile.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/drivers/pci/host/pci-versatile.c b/drivers/pci/host/pci-versatile.c
index f2344057..b7dc070 100644
--- a/drivers/pci/host/pci-versatile.c
+++ b/drivers/pci/host/pci-versatile.c
@@ -74,7 +74,7 @@ static int versatile_pci_parse_request_of_pci_ranges(struct device *dev,
 	int err, mem = 1, res_valid = 0;
 	struct device_node *np = dev->of_node;
 	resource_size_t iobase;
-	struct resource_entry *win;
+	struct resource_entry *win, *tmp;
 
 	err = of_pci_get_host_bridge_resources(np, 0, 0xff, res, &iobase);
 	if (err)
@@ -84,15 +84,17 @@ static int versatile_pci_parse_request_of_pci_ranges(struct device *dev,
 	if (err)
 		goto out_release_res;
 
-	resource_list_for_each_entry(win, res) {
+	resource_list_for_each_entry_safe(win, tmp, res) {
 		struct resource *res = win->res;
 
 		switch (resource_type(res)) {
 		case IORESOURCE_IO:
 			err = pci_remap_iospace(res, iobase);
-			if (err)
+			if (err) {
 				dev_warn(dev, "error %d: failed to map resource %pR\n",
 					 err, res);
+				resource_list_destroy_entry(win);
+			}
 			break;
 		case IORESOURCE_MEM:
 			res_valid |= !(res->flags & IORESOURCE_PREFETCH);
-- 
2.6.4

[PATCH 4/6] drivers: pci: host: rcar: fix pci_remap_iospace() failure path

[Lorenzo Pieralisi]

On ARM/ARM64 architectures, PCI IO ports are emulated through memory
mapped IO, by reserving a chunk of virtual address space starting at
PCI_IOBASE and by mapping the PCI host bridges memory address space
driving PCI IO cycles to it.

PCI host bridge drivers that enable downstream PCI IO cycles map the
host bridge memory address responding to PCI IO cycles to the fixed
virtual address space through the pci_remap_iospace() API.

This means that if the pci_remap_iospace() function fails, the
corresponding host bridge PCI IO resource must be considered invalid, in
that there is no way for the kernel to actually drive PCI IO
transactions if the memory addresses responding to PCI
IO cycles cannot be mapped into the CPU virtual address space.

The PCI rcar host bridge driver does not remove the PCI IO
resource from the host bridge resource windows if the
pci_remap_iospace() call fails; this is an actual bug in that the
PCI host bridge would consider the PCI IO resource valid (and possibly
assign it to downstream devices) even if the kernel was not able
to map the PCI host bridge memory address driving IO cycle to the
CPU virtual address space (ie pci_remap_iospace() failures).

Fix the PCI host bridge driver pci_remap_iospace() failure path, by
destroying the PCI host bridge PCI IO resources retrieved through
firmware when the pci_remap_iospace() function call fails, therefore
preventing the kernel from adding the respective PCI IO resource to the
list of PCI host bridge valid resources, fixing the issue.

Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Fixes: 5d2917d469fa ("PCI: rcar: Convert to DT resource parsing API")
Cc: Phil Edworthy <phil.edworthy@renesas.com>
Cc: Bjorn Helgaas <bhelgaas@google.com>
Cc: Simon Horman <horms+renesas@verge.net.au>
---
 drivers/pci/host/pcie-rcar.c | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/drivers/pci/host/pcie-rcar.c b/drivers/pci/host/pcie-rcar.c
index 65db7a2..5f7fcc9 100644
--- a/drivers/pci/host/pcie-rcar.c
+++ b/drivers/pci/host/pcie-rcar.c
@@ -945,7 +945,7 @@ static int rcar_pcie_parse_request_of_pci_ranges(struct rcar_pcie *pci)
 	struct device *dev = pci->dev;
 	struct device_node *np = dev->of_node;
 	resource_size_t iobase;
-	struct resource_entry *win;
+	struct resource_entry *win, *tmp;
 
 	err = of_pci_get_host_bridge_resources(np, 0, 0xff, &pci->resources, &iobase);
 	if (err)
@@ -955,14 +955,17 @@ static int rcar_pcie_parse_request_of_pci_ranges(struct rcar_pcie *pci)
 	if (err)
 		goto out_release_res;
 
-	resource_list_for_each_entry(win, &pci->resources) {
+	resource_list_for_each_entry_safe(win, tmp, &pci->resources) {
 		struct resource *res = win->res;
 
 		if (resource_type(res) == IORESOURCE_IO) {
 			err = pci_remap_iospace(res, iobase);
-			if (err)
+			if (err) {
 				dev_warn(dev, "error %d: failed to map resource %pR\n",
 					 err, res);
+
+				resource_list_destroy_entry(win);
+			}
 		}
 	}
 
-- 
2.6.4

[PATCH 5/6] drivers: pci: host: common: fix pci_remap_iospace() failure path

[Lorenzo Pieralisi]

On ARM/ARM64 architectures, PCI IO ports are emulated through memory
mapped IO, by reserving a chunk of virtual address space starting at
PCI_IOBASE and by mapping the PCI host bridges memory address space
driving PCI IO cycles to it.

PCI host bridge drivers that enable downstream PCI IO cycles map the
host bridge memory address responding to PCI IO cycles to the fixed
virtual address space through the pci_remap_iospace() API.

This means that if the pci_remap_iospace() function fails, the
corresponding host bridge PCI IO resource must be considered invalid, in
that there is no way for the kernel to actually drive PCI IO
transactions if the memory addresses responding to PCI
IO cycles cannot be mapped into the CPU virtual address space.

The PCI common host bridge driver does not remove the PCI IO
resource from the host bridge resource windows if the
pci_remap_iospace() call fails; this is an actual bug in that the
PCI host bridge would consider the PCI IO resource valid (and possibly
assign it to downstream devices) even if the kernel was not able
to map the PCI host bridge memory address driving IO cycle to the
CPU virtual address space (ie pci_remap_iospace() failures).

Fix the PCI host bridge driver pci_remap_iospace() failure path, by
destroying the PCI host bridge PCI IO resources retrieved through
firmware when the pci_remap_iospace() function call fails, therefore
preventing the kernel from adding the respective PCI IO resource to the
list of PCI host bridge valid resources, fixing the issue.

Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Fixes: 4e64dbe226e7 ("PCI: generic: Expose pci_host_common_probe() for
use by other drivers")
Cc: Bjorn Helgaas <bhelgaas@google.com>
Cc: Will Deacon <will.deacon@arm.com>
---
 drivers/pci/host/pci-host-common.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/drivers/pci/host/pci-host-common.c b/drivers/pci/host/pci-host-common.c
index 9d9d34e..61eb4d4 100644
--- a/drivers/pci/host/pci-host-common.c
+++ b/drivers/pci/host/pci-host-common.c
@@ -29,7 +29,7 @@ static int gen_pci_parse_request_of_pci_ranges(struct device *dev,
 	int err, res_valid = 0;
 	struct device_node *np = dev->of_node;
 	resource_size_t iobase;
-	struct resource_entry *win;
+	struct resource_entry *win, *tmp;
 
 	err = of_pci_get_host_bridge_resources(np, 0, 0xff, resources, &iobase);
 	if (err)
@@ -39,15 +39,17 @@ static int gen_pci_parse_request_of_pci_ranges(struct device *dev,
 	if (err)
 		return err;
 
-	resource_list_for_each_entry(win, resources) {
+	resource_list_for_each_entry_safe(win, tmp, resources) {
 		struct resource *res = win->res;
 
 		switch (resource_type(res)) {
 		case IORESOURCE_IO:
 			err = pci_remap_iospace(res, iobase);
-			if (err)
+			if (err) {
 				dev_warn(dev, "error %d: failed to map resource %pR\n",
 					 err, res);
+				resource_list_destroy_entry(win);
+			}
 			break;
 		case IORESOURCE_MEM:
 			res_valid |= !(res->flags & IORESOURCE_PREFETCH);
-- 
2.6.4

Re: [PATCH 5/6] drivers: pci: host: common: fix pci_remap_iospace() failure path

[Will Deacon]

On Mon, Aug 15, 2016 at 05:50:45PM +0100, Lorenzo Pieralisi wrote:
> On ARM/ARM64 architectures, PCI IO ports are emulated through memory
> mapped IO, by reserving a chunk of virtual address space starting at
> PCI_IOBASE and by mapping the PCI host bridges memory address space
> driving PCI IO cycles to it.
> 
> PCI host bridge drivers that enable downstream PCI IO cycles map the
> host bridge memory address responding to PCI IO cycles to the fixed
> virtual address space through the pci_remap_iospace() API.
> 
> This means that if the pci_remap_iospace() function fails, the
> corresponding host bridge PCI IO resource must be considered invalid, in
> that there is no way for the kernel to actually drive PCI IO
> transactions if the memory addresses responding to PCI
> IO cycles cannot be mapped into the CPU virtual address space.
> 
> The PCI common host bridge driver does not remove the PCI IO
> resource from the host bridge resource windows if the
> pci_remap_iospace() call fails; this is an actual bug in that the
> PCI host bridge would consider the PCI IO resource valid (and possibly
> assign it to downstream devices) even if the kernel was not able
> to map the PCI host bridge memory address driving IO cycle to the
> CPU virtual address space (ie pci_remap_iospace() failures).
> 
> Fix the PCI host bridge driver pci_remap_iospace() failure path, by
> destroying the PCI host bridge PCI IO resources retrieved through
> firmware when the pci_remap_iospace() function call fails, therefore
> preventing the kernel from adding the respective PCI IO resource to the
> list of PCI host bridge valid resources, fixing the issue.
> 
> Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
> Fixes: 4e64dbe226e7 ("PCI: generic: Expose pci_host_common_probe() for
> use by other drivers")
> Cc: Bjorn Helgaas <bhelgaas@google.com>
> Cc: Will Deacon <will.deacon@arm.com>
> ---
>  drivers/pci/host/pci-host-common.c | 8 +++++---
>  1 file changed, 5 insertions(+), 3 deletions(-)

Looks good to me:

Acked-by: Will Deacon <will.deacon@arm.com>

Will

[PATCH 6/6] drivers: pci: host: tegra: fix pci_remap_iospace() failure path

[Lorenzo Pieralisi]

On ARM/ARM64 architectures, PCI IO ports are emulated through memory
mapped IO, by reserving a chunk of virtual address space starting at
PCI_IOBASE and by mapping the PCI host bridges memory address space
driving PCI IO cycles to it.

PCI host bridge drivers that enable downstream PCI IO cycles map the
host bridge memory address responding to PCI IO cycles to the fixed
virtual address space through the pci_remap_iospace() API.

This means that if the pci_remap_iospace() function fails, the
corresponding host bridge PCI IO resource must be considered invalid, in
that there is no way for the kernel to actually drive PCI IO
transactions if the memory addresses responding to PCI
IO cycles cannot be mapped into the CPU virtual address space.

The PCI tegra host bridge driver adds the PCI IO resource retrieved
from firmware to the host bridge resource windows even if the
pci_remap_iospace() call fails; this is an actual bug in that the PCI
host bridge would consider the PCI IO resource valid (and possibly
assign it to downstream devices) even if the kernel was not able to map
the PCI host bridge memory address driving IO cycle to the CPU virtual
address space (ie pci_remap_iospace() failures).

Add the PCI host bridge driver pci_remap_iospace() failure path and
do not add the corresponding PCI host bridge PCI IO resources
retrieved through firmware when the pci_remap_iospace() function
call fails, fixing the issue.

Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
Fixes: e6e9f471f5fe ("PCI: tegra: Use generic pci_remap_iospace() rather than ARM32-specific one")
Cc: Bjorn Helgaas <bhelgaas@google.com>
Cc: Thierry Reding <treding@nvidia.com>
---
 drivers/pci/host/pci-tegra.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/drivers/pci/host/pci-tegra.c b/drivers/pci/host/pci-tegra.c
index 6de0757..8c2590d 100644
--- a/drivers/pci/host/pci-tegra.c
+++ b/drivers/pci/host/pci-tegra.c
@@ -621,7 +621,11 @@ static int tegra_pcie_setup(int nr, struct pci_sys_data *sys)
 	if (err < 0)
 		return err;
 
-	pci_add_resource_offset(&sys->resources, &pcie->pio, sys->io_offset);
+	err = pci_remap_iospace(&pcie->pio, pcie->io.start);
+	if (!err)
+		pci_add_resource_offset(&sys->resources, &pcie->pio,
+					sys->io_offset);
+
 	pci_add_resource_offset(&sys->resources, &pcie->mem, sys->mem_offset);
 	pci_add_resource_offset(&sys->resources, &pcie->prefetch,
 				sys->mem_offset);
@@ -631,7 +635,6 @@ static int tegra_pcie_setup(int nr, struct pci_sys_data *sys)
 	if (err < 0)
 		return err;
 
-	pci_remap_iospace(&pcie->pio, pcie->io.start);
 	return 1;
 }
 
-- 
2.6.4

Re: [PATCH 0/6] PCI: fix pci_remap_iospace() failure paths

[Bjorn Helgaas]

On Mon, Aug 15, 2016 at 05:50:40PM +0100, Lorenzo Pieralisi wrote:
> On ARM/ARM64 architectures, PCI IO ports are emulated through memory
> mapped IO, by reserving a chunk of virtual address space starting at
> PCI_IOBASE and by mapping the PCI host bridges memory address space
> driving PCI IO cycles to it.
> 
> PCI host bridge drivers that enable downstream PCI IO cycles map the
> host bridge memory address responding to PCI IO cycles to the fixed
> virtual address space through the pci_remap_iospace() API.
> 
> This means that if the pci_remap_iospace() function fails, the
> corresponding host bridge PCI IO resource must be considered invalid, in
> that there is no way for the kernel to actually drive PCI IO transactions
> if the memory addresses responding to PCI IO cycles cannot be mapped into
> the CPU virtual address space.
> 
> As a follow-up to the linux-pci mailing list thread ([0]), this series
> fixes the pci_remap_iospace() failure paths by adding code that handles
> the failures and takes the required actions.
> 
> Compile tested on all affected PCI host controllers, testing and
> reviewing much appreciated.
> 
> [0] http://marc.info/?l=linux-pci&m=145816042427417&w=2
> 
> Lorenzo Pieralisi (6):
>   drivers: pci: host: aardvark: fix pci_remap_iospace() failure path
>   drivers: pci: host: designware: fix pci_remap_iospace() failure path
>   drivers: pci: host: versatile: fix pci_remap_iospace() failure path
>   drivers: pci: host: rcar: fix pci_remap_iospace() failure path
>   drivers: pci: host: common: fix pci_remap_iospace() failure path
>   drivers: pci: host: tegra: fix pci_remap_iospace() failure path
> 
>  drivers/pci/host/pci-aardvark.c    |  8 +++++---
>  drivers/pci/host/pci-host-common.c |  8 +++++---
>  drivers/pci/host/pci-tegra.c       |  7 +++++--
>  drivers/pci/host/pci-versatile.c   |  8 +++++---
>  drivers/pci/host/pcie-designware.c | 21 ++++++++++++---------
>  drivers/pci/host/pcie-rcar.c       |  9 ++++++---
>  6 files changed, 38 insertions(+), 23 deletions(-)

Applied to pci/enumeration for v4.9, thanks!