* [PATCH v2 2/7] xen/arm: make pci_host_common_probe return the bridge
2025-03-11 10:24 [PATCH v2 0/7] Add support for R-Car Gen4 PCI host controller Mykyta Poturai
@ 2025-03-11 10:24 ` Mykyta Poturai
2025-03-11 10:24 ` [PATCH v2 1/7] xen/arm: allow PCI host bridge to have private data Mykyta Poturai
` (5 subsequent siblings)
6 siblings, 0 replies; 13+ messages in thread
From: Mykyta Poturai @ 2025-03-11 10:24 UTC (permalink / raw)
To: xen-devel@lists.xenproject.org
Cc: Oleksandr Andrushchenko, Stefano Stabellini, Julien Grall,
Bertrand Marquis, Michal Orzel, Volodymyr Babchuk, Mykyta Poturai
From: Oleksandr Andrushchenko <oleksandr_andrushchenko@epam.com>
Some of the PCI host bridges require additional processing during the
probe phase. For that they need to access struct bridge of the probed
host, so return pointer to the new bridge from pci_host_common_probe.
Signed-off-by: Oleksandr Andrushchenko <oleksandr_andrushchenko@epam.com>
Signed-off-by: Mykyta Poturai <mykyta_poturai@epam.com>
---
v1->v2:
* no change
---
xen/arch/arm/include/asm/pci.h | 8 +++++---
xen/arch/arm/pci/pci-host-common.c | 12 ++++++------
xen/arch/arm/pci/pci-host-generic.c | 2 +-
xen/arch/arm/pci/pci-host-zynqmp.c | 2 +-
4 files changed, 13 insertions(+), 11 deletions(-)
diff --git a/xen/arch/arm/include/asm/pci.h b/xen/arch/arm/include/asm/pci.h
index 44344ac8c1..e1f63d75e3 100644
--- a/xen/arch/arm/include/asm/pci.h
+++ b/xen/arch/arm/include/asm/pci.h
@@ -17,6 +17,8 @@
#ifdef CONFIG_HAS_PCI
+#include <xen/err.h>
+
#include <asm/p2m.h>
#define pci_to_dev(pcidev) (&(pcidev)->arch.dev)
@@ -95,9 +97,9 @@ struct pci_ecam_ops {
/* Default ECAM ops */
extern const struct pci_ecam_ops pci_generic_ecam_ops;
-int pci_host_common_probe(struct dt_device_node *dev,
- const struct pci_ecam_ops *ops,
- size_t priv_sz);
+struct pci_host_bridge *pci_host_common_probe(struct dt_device_node *dev,
+ const struct pci_ecam_ops *ops,
+ size_t priv_sz);
int pci_generic_config_read(struct pci_host_bridge *bridge, pci_sbdf_t sbdf,
uint32_t reg, uint32_t len, uint32_t *value);
int pci_generic_config_write(struct pci_host_bridge *bridge, pci_sbdf_t sbdf,
diff --git a/xen/arch/arm/pci/pci-host-common.c b/xen/arch/arm/pci/pci-host-common.c
index be7e6c3510..e4e05d1176 100644
--- a/xen/arch/arm/pci/pci-host-common.c
+++ b/xen/arch/arm/pci/pci-host-common.c
@@ -208,9 +208,9 @@ static int pci_bus_find_domain_nr(struct dt_device_node *dev)
return domain;
}
-int pci_host_common_probe(struct dt_device_node *dev,
- const struct pci_ecam_ops *ops,
- size_t priv_sz)
+struct pci_host_bridge *pci_host_common_probe(struct dt_device_node *dev,
+ const struct pci_ecam_ops *ops,
+ size_t priv_sz)
{
struct pci_host_bridge *bridge;
struct pci_config_window *cfg;
@@ -222,7 +222,7 @@ int pci_host_common_probe(struct dt_device_node *dev,
bridge = pci_alloc_host_bridge();
if ( !bridge )
- return -ENOMEM;
+ return ERR_PTR(-ENOMEM);
/* Parse and map our Configuration Space windows */
cfg = gen_pci_init(dev, ops);
@@ -257,7 +257,7 @@ int pci_host_common_probe(struct dt_device_node *dev,
pci_add_host_bridge(bridge);
- return 0;
+ return bridge;
err_priv:
xfree(bridge->priv);
@@ -265,7 +265,7 @@ err_priv:
err_exit:
xfree(bridge);
- return err;
+ return ERR_PTR(err);
}
/*
diff --git a/xen/arch/arm/pci/pci-host-generic.c b/xen/arch/arm/pci/pci-host-generic.c
index cc4bc70684..dde6a79a8e 100644
--- a/xen/arch/arm/pci/pci-host-generic.c
+++ b/xen/arch/arm/pci/pci-host-generic.c
@@ -29,7 +29,7 @@ static const struct dt_device_match __initconstrel gen_pci_dt_match[] =
static int __init pci_host_generic_probe(struct dt_device_node *dev,
const void *data)
{
- return pci_host_common_probe(dev, &pci_generic_ecam_ops, 0);
+ return PTR_RET(pci_host_common_probe(dev, &pci_generic_ecam_ops, 0));
}
DT_DEVICE_START(pci_gen, "PCI HOST GENERIC", DEVICE_PCI_HOSTBRIDGE)
diff --git a/xen/arch/arm/pci/pci-host-zynqmp.c b/xen/arch/arm/pci/pci-host-zynqmp.c
index 985a43c516..c796447ac4 100644
--- a/xen/arch/arm/pci/pci-host-zynqmp.c
+++ b/xen/arch/arm/pci/pci-host-zynqmp.c
@@ -47,7 +47,7 @@ static const struct dt_device_match __initconstrel nwl_pcie_dt_match[] =
static int __init pci_host_generic_probe(struct dt_device_node *dev,
const void *data)
{
- return pci_host_common_probe(dev, &nwl_pcie_ops, 0);
+ return PTR_RET(pci_host_common_probe(dev, &nwl_pcie_ops, 0));
}
DT_DEVICE_START(pci_gen, "PCI HOST ZYNQMP", DEVICE_PCI_HOSTBRIDGE)
--
2.34.1
^ permalink raw reply related [flat|nested] 13+ messages in thread* [PATCH v2 1/7] xen/arm: allow PCI host bridge to have private data
2025-03-11 10:24 [PATCH v2 0/7] Add support for R-Car Gen4 PCI host controller Mykyta Poturai
2025-03-11 10:24 ` [PATCH v2 2/7] xen/arm: make pci_host_common_probe return the bridge Mykyta Poturai
@ 2025-03-11 10:24 ` Mykyta Poturai
2025-03-11 15:05 ` Grygorii Strashko
2025-03-11 10:24 ` [PATCH v2 3/7] xen/arm: add support for PCI child bus Mykyta Poturai
` (4 subsequent siblings)
6 siblings, 1 reply; 13+ messages in thread
From: Mykyta Poturai @ 2025-03-11 10:24 UTC (permalink / raw)
To: xen-devel@lists.xenproject.org
Cc: Oleksandr Andrushchenko, Stefano Stabellini, Julien Grall,
Bertrand Marquis, Michal Orzel, Volodymyr Babchuk, Mykyta Poturai
From: Oleksandr Andrushchenko <oleksandr_andrushchenko@epam.com>
Some of the PCI host bridges require private data. Create a generic
approach for that, so such bridges may request the private data to
be allocated during initialization.
Signed-off-by: Oleksandr Andrushchenko <oleksandr_andrushchenko@epam.com>
Signed-off-by: Mykyta Poturai <mykyta_poturai@epam.com>
---
v1->v2:
* no change
---
xen/arch/arm/include/asm/pci.h | 4 +++-
xen/arch/arm/pci/pci-host-common.c | 18 +++++++++++++++++-
xen/arch/arm/pci/pci-host-generic.c | 2 +-
xen/arch/arm/pci/pci-host-zynqmp.c | 2 +-
4 files changed, 22 insertions(+), 4 deletions(-)
diff --git a/xen/arch/arm/include/asm/pci.h b/xen/arch/arm/include/asm/pci.h
index 7f77226c9b..44344ac8c1 100644
--- a/xen/arch/arm/include/asm/pci.h
+++ b/xen/arch/arm/include/asm/pci.h
@@ -66,6 +66,7 @@ struct pci_host_bridge {
uint16_t segment; /* Segment number */
struct pci_config_window* cfg; /* Pointer to the bridge config window */
const struct pci_ops *ops;
+ void *priv; /* Private data of the bridge. */
};
struct pci_ops {
@@ -95,7 +96,8 @@ struct pci_ecam_ops {
extern const struct pci_ecam_ops pci_generic_ecam_ops;
int pci_host_common_probe(struct dt_device_node *dev,
- const struct pci_ecam_ops *ops);
+ const struct pci_ecam_ops *ops,
+ size_t priv_sz);
int pci_generic_config_read(struct pci_host_bridge *bridge, pci_sbdf_t sbdf,
uint32_t reg, uint32_t len, uint32_t *value);
int pci_generic_config_write(struct pci_host_bridge *bridge, pci_sbdf_t sbdf,
diff --git a/xen/arch/arm/pci/pci-host-common.c b/xen/arch/arm/pci/pci-host-common.c
index c0faf0f436..be7e6c3510 100644
--- a/xen/arch/arm/pci/pci-host-common.c
+++ b/xen/arch/arm/pci/pci-host-common.c
@@ -209,7 +209,8 @@ static int pci_bus_find_domain_nr(struct dt_device_node *dev)
}
int pci_host_common_probe(struct dt_device_node *dev,
- const struct pci_ecam_ops *ops)
+ const struct pci_ecam_ops *ops,
+ size_t priv_sz)
{
struct pci_host_bridge *bridge;
struct pci_config_window *cfg;
@@ -241,11 +242,26 @@ int pci_host_common_probe(struct dt_device_node *dev,
printk(XENLOG_ERR "Inconsistent \"linux,pci-domain\" property in DT\n");
BUG();
}
+
bridge->segment = domain;
+
+ if ( priv_sz )
+ {
+ bridge->priv = xzalloc_bytes(priv_sz);
+ if ( !bridge->priv )
+ {
+ err = -ENOMEM;
+ goto err_priv;
+ }
+ }
+
pci_add_host_bridge(bridge);
return 0;
+err_priv:
+ xfree(bridge->priv);
+
err_exit:
xfree(bridge);
diff --git a/xen/arch/arm/pci/pci-host-generic.c b/xen/arch/arm/pci/pci-host-generic.c
index 46de6e43cc..cc4bc70684 100644
--- a/xen/arch/arm/pci/pci-host-generic.c
+++ b/xen/arch/arm/pci/pci-host-generic.c
@@ -29,7 +29,7 @@ static const struct dt_device_match __initconstrel gen_pci_dt_match[] =
static int __init pci_host_generic_probe(struct dt_device_node *dev,
const void *data)
{
- return pci_host_common_probe(dev, &pci_generic_ecam_ops);
+ return pci_host_common_probe(dev, &pci_generic_ecam_ops, 0);
}
DT_DEVICE_START(pci_gen, "PCI HOST GENERIC", DEVICE_PCI_HOSTBRIDGE)
diff --git a/xen/arch/arm/pci/pci-host-zynqmp.c b/xen/arch/arm/pci/pci-host-zynqmp.c
index 101edb8593..985a43c516 100644
--- a/xen/arch/arm/pci/pci-host-zynqmp.c
+++ b/xen/arch/arm/pci/pci-host-zynqmp.c
@@ -47,7 +47,7 @@ static const struct dt_device_match __initconstrel nwl_pcie_dt_match[] =
static int __init pci_host_generic_probe(struct dt_device_node *dev,
const void *data)
{
- return pci_host_common_probe(dev, &nwl_pcie_ops);
+ return pci_host_common_probe(dev, &nwl_pcie_ops, 0);
}
DT_DEVICE_START(pci_gen, "PCI HOST ZYNQMP", DEVICE_PCI_HOSTBRIDGE)
--
2.34.1
^ permalink raw reply related [flat|nested] 13+ messages in thread* Re: [PATCH v2 1/7] xen/arm: allow PCI host bridge to have private data
2025-03-11 10:24 ` [PATCH v2 1/7] xen/arm: allow PCI host bridge to have private data Mykyta Poturai
@ 2025-03-11 15:05 ` Grygorii Strashko
2025-03-11 19:24 ` Julien Grall
0 siblings, 1 reply; 13+ messages in thread
From: Grygorii Strashko @ 2025-03-11 15:05 UTC (permalink / raw)
To: Mykyta Poturai, xen-devel@lists.xenproject.org
Cc: Oleksandr Andrushchenko, Stefano Stabellini, Julien Grall,
Bertrand Marquis, Michal Orzel, Volodymyr Babchuk
On 11.03.25 12:24, Mykyta Poturai wrote:
> From: Oleksandr Andrushchenko <oleksandr_andrushchenko@epam.com>
>
> Some of the PCI host bridges require private data. Create a generic
> approach for that, so such bridges may request the private data to
> be allocated during initialization.
>
> Signed-off-by: Oleksandr Andrushchenko <oleksandr_andrushchenko@epam.com>
> Signed-off-by: Mykyta Poturai <mykyta_poturai@epam.com>
> ---
> v1->v2:
> * no change
> ---
> xen/arch/arm/include/asm/pci.h | 4 +++-
> xen/arch/arm/pci/pci-host-common.c | 18 +++++++++++++++++-
> xen/arch/arm/pci/pci-host-generic.c | 2 +-
> xen/arch/arm/pci/pci-host-zynqmp.c | 2 +-
> 4 files changed, 22 insertions(+), 4 deletions(-)
>
> diff --git a/xen/arch/arm/include/asm/pci.h b/xen/arch/arm/include/asm/pci.h
> index 7f77226c9b..44344ac8c1 100644
> --- a/xen/arch/arm/include/asm/pci.h
> +++ b/xen/arch/arm/include/asm/pci.h
> @@ -66,6 +66,7 @@ struct pci_host_bridge {
> uint16_t segment; /* Segment number */
> struct pci_config_window* cfg; /* Pointer to the bridge config window */
> const struct pci_ops *ops;
> + void *priv; /* Private data of the bridge. */
> };
>
> struct pci_ops {
> @@ -95,7 +96,8 @@ struct pci_ecam_ops {
> extern const struct pci_ecam_ops pci_generic_ecam_ops;
>
> int pci_host_common_probe(struct dt_device_node *dev,
> - const struct pci_ecam_ops *ops);
> + const struct pci_ecam_ops *ops,
> + size_t priv_sz);
> int pci_generic_config_read(struct pci_host_bridge *bridge, pci_sbdf_t sbdf,
> uint32_t reg, uint32_t len, uint32_t *value);
> int pci_generic_config_write(struct pci_host_bridge *bridge, pci_sbdf_t sbdf,
> diff --git a/xen/arch/arm/pci/pci-host-common.c b/xen/arch/arm/pci/pci-host-common.c
> index c0faf0f436..be7e6c3510 100644
> --- a/xen/arch/arm/pci/pci-host-common.c
> +++ b/xen/arch/arm/pci/pci-host-common.c
> @@ -209,7 +209,8 @@ static int pci_bus_find_domain_nr(struct dt_device_node *dev)
> }
>
> int pci_host_common_probe(struct dt_device_node *dev,
> - const struct pci_ecam_ops *ops)
> + const struct pci_ecam_ops *ops,
> + size_t priv_sz)
> {
> struct pci_host_bridge *bridge;
> struct pci_config_window *cfg;
> @@ -241,11 +242,26 @@ int pci_host_common_probe(struct dt_device_node *dev,
> printk(XENLOG_ERR "Inconsistent \"linux,pci-domain\" property in DT\n");
> BUG();
> }
> +
> bridge->segment = domain;
> +
> + if ( priv_sz )
> + {
> + bridge->priv = xzalloc_bytes(priv_sz);
> + if ( !bridge->priv )
> + {
> + err = -ENOMEM;
> + goto err_priv;
> + }
> + }
> +
I'd like to propose to move allocation into pci_alloc_host_bridge()
to keep mallocs in one place and do it earlier, before proceeding
with other initialization steps.
Also the pci_alloc_host_bridge() could be made static, seems.
> pci_add_host_bridge(bridge);
>
> return 0;
>
> +err_priv:
> + xfree(bridge->priv);
> +
> err_exit:
> xfree(bridge);
>
[...]
-grygorii
^ permalink raw reply [flat|nested] 13+ messages in thread* Re: [PATCH v2 1/7] xen/arm: allow PCI host bridge to have private data
2025-03-11 15:05 ` Grygorii Strashko
@ 2025-03-11 19:24 ` Julien Grall
0 siblings, 0 replies; 13+ messages in thread
From: Julien Grall @ 2025-03-11 19:24 UTC (permalink / raw)
To: Grygorii Strashko, Mykyta Poturai, xen-devel@lists.xenproject.org
Cc: Oleksandr Andrushchenko, Stefano Stabellini, Bertrand Marquis,
Michal Orzel, Volodymyr Babchuk
Hi,
On 11/03/2025 15:05, Grygorii Strashko wrote:
>
>
> On 11.03.25 12:24, Mykyta Poturai wrote:
>> From: Oleksandr Andrushchenko <oleksandr_andrushchenko@epam.com>
>>
>> Some of the PCI host bridges require private data. Create a generic
>> approach for that, so such bridges may request the private data to
>> be allocated during initialization.
>>
>> Signed-off-by: Oleksandr Andrushchenko <oleksandr_andrushchenko@epam.com>
>> Signed-off-by: Mykyta Poturai <mykyta_poturai@epam.com>
>> ---
>> v1->v2:
>> * no change
>> ---
>> xen/arch/arm/include/asm/pci.h | 4 +++-
>> xen/arch/arm/pci/pci-host-common.c | 18 +++++++++++++++++-
>> xen/arch/arm/pci/pci-host-generic.c | 2 +-
>> xen/arch/arm/pci/pci-host-zynqmp.c | 2 +-
>> 4 files changed, 22 insertions(+), 4 deletions(-)
>>
>> diff --git a/xen/arch/arm/include/asm/pci.h b/xen/arch/arm/include/
>> asm/pci.h
>> index 7f77226c9b..44344ac8c1 100644
>> --- a/xen/arch/arm/include/asm/pci.h
>> +++ b/xen/arch/arm/include/asm/pci.h
>> @@ -66,6 +66,7 @@ struct pci_host_bridge {
>> uint16_t segment; /* Segment number */
>> struct pci_config_window* cfg; /* Pointer to the bridge config
>> window */
>> const struct pci_ops *ops;
>> + void *priv; /* Private data of the bridge. */
>> };
>> struct pci_ops {
>> @@ -95,7 +96,8 @@ struct pci_ecam_ops {
>> extern const struct pci_ecam_ops pci_generic_ecam_ops;
>> int pci_host_common_probe(struct dt_device_node *dev,
>> - const struct pci_ecam_ops *ops);
>> + const struct pci_ecam_ops *ops,
>> + size_t priv_sz);
>> int pci_generic_config_read(struct pci_host_bridge *bridge,
>> pci_sbdf_t sbdf,
>> uint32_t reg, uint32_t len, uint32_t
>> *value);
>> int pci_generic_config_write(struct pci_host_bridge *bridge,
>> pci_sbdf_t sbdf,
>> diff --git a/xen/arch/arm/pci/pci-host-common.c b/xen/arch/arm/pci/
>> pci-host-common.c
>> index c0faf0f436..be7e6c3510 100644
>> --- a/xen/arch/arm/pci/pci-host-common.c
>> +++ b/xen/arch/arm/pci/pci-host-common.c
>> @@ -209,7 +209,8 @@ static int pci_bus_find_domain_nr(struct
>> dt_device_node *dev)
>> }
>> int pci_host_common_probe(struct dt_device_node *dev,
>> - const struct pci_ecam_ops *ops)
>> + const struct pci_ecam_ops *ops,
>> + size_t priv_sz)
>> {
>> struct pci_host_bridge *bridge;
>> struct pci_config_window *cfg;
>> @@ -241,11 +242,26 @@ int pci_host_common_probe(struct dt_device_node
>> *dev,
>> printk(XENLOG_ERR "Inconsistent \"linux,pci-domain\"
>> property in DT\n");
>> BUG();
>> }
>> +
>> bridge->segment = domain;
>> +
>> + if ( priv_sz )
>> + {
>> + bridge->priv = xzalloc_bytes(priv_sz);
>> + if ( !bridge->priv )
>> + {
>> + err = -ENOMEM;
>> + goto err_priv;
>> + }
>> + }
>> +
>
> I'd like to propose to move allocation into pci_alloc_host_bridge()
> to keep mallocs in one place and do it earlier, before proceeding
> with other initialization steps.
In both cases, we need to pass the size of the structure in parameter
and the structure is not initialized until later. I would rather prefer
if the allocation is done by each PCI controller that needs it close to
the initialization of "priv".
Cheers,
--
Julien Grall
^ permalink raw reply [flat|nested] 13+ messages in thread
* [PATCH v2 3/7] xen/arm: add support for PCI child bus
2025-03-11 10:24 [PATCH v2 0/7] Add support for R-Car Gen4 PCI host controller Mykyta Poturai
2025-03-11 10:24 ` [PATCH v2 2/7] xen/arm: make pci_host_common_probe return the bridge Mykyta Poturai
2025-03-11 10:24 ` [PATCH v2 1/7] xen/arm: allow PCI host bridge to have private data Mykyta Poturai
@ 2025-03-11 10:24 ` Mykyta Poturai
2025-03-11 10:24 ` [PATCH v2 6/7] xen/arm: rcar4: add simple optimization to avoid ATU reprogramming Mykyta Poturai
` (3 subsequent siblings)
6 siblings, 0 replies; 13+ messages in thread
From: Mykyta Poturai @ 2025-03-11 10:24 UTC (permalink / raw)
To: xen-devel@lists.xenproject.org
Cc: Oleksandr Andrushchenko, Stefano Stabellini, Julien Grall,
Bertrand Marquis, Michal Orzel, Volodymyr Babchuk, Mykyta Poturai
From: Oleksandr Andrushchenko <oleksandr_andrushchenko@epam.com>
PCI host bridges often have different ways to access the root and child
bus configuration spaces. One of the examples is Designware's host bridge
and its multiple clones [1].
Linux kernel implements this by instantiating a child bus when device
drivers provide not only the usual pci_ops to access ECAM space (this is
the case for the generic host bridge), but also means to access the child
bus which has a dedicated configuration space and own implementation for
accessing the bus, e.g. child_ops.
For Xen it is not feasible to fully implement PCI bus infrastructure as
Linux kernel does, but still child bus can be supported.
Add support for the PCI child bus which includes the following changes:
- introduce bus mapping functions depending on SBDF
- assign bus start and end for the child bus and re-configure the same for
the parent (root) bus
- make pci_find_host_bridge be aware of multiple busses behind the same bridge
- update pci_host_bridge_mappings, so it also doesn't map to guest the memory
spaces belonging to the child bus
- make pci_host_common_probe accept one more pci_ops structure for the child bus
- install MMIO handlers for the child bus
- re-work vpci_mmio_{write|read} with parent and child approach in mind
[1] https://elixir.bootlin.com/linux/v5.15/source/drivers/pci/controller/dwc
Signed-off-by: Oleksandr Andrushchenko <oleksandr_andrushchenko@epam.com>
Signed-off-by: Mykyta Poturai <mykyta_poturai@epam.com>
---
v1->v2:
* fixed compilation issues
---
xen/arch/arm/include/asm/pci.h | 14 ++++-
xen/arch/arm/pci/ecam.c | 17 ++++--
xen/arch/arm/pci/pci-access.c | 37 +++++++++++--
xen/arch/arm/pci/pci-host-common.c | 86 +++++++++++++++++++++++------
xen/arch/arm/pci/pci-host-generic.c | 2 +-
xen/arch/arm/pci/pci-host-zynqmp.c | 2 +-
xen/arch/arm/vpci.c | 83 ++++++++++++++++++++++------
7 files changed, 192 insertions(+), 49 deletions(-)
diff --git a/xen/arch/arm/include/asm/pci.h b/xen/arch/arm/include/asm/pci.h
index e1f63d75e3..0012c2ae9e 100644
--- a/xen/arch/arm/include/asm/pci.h
+++ b/xen/arch/arm/include/asm/pci.h
@@ -69,6 +69,9 @@ struct pci_host_bridge {
struct pci_config_window* cfg; /* Pointer to the bridge config window */
const struct pci_ops *ops;
void *priv; /* Private data of the bridge. */
+ /* Child bus */
+ struct pci_config_window* child_cfg;
+ const struct pci_ops *child_ops;
};
struct pci_ops {
@@ -97,15 +100,20 @@ struct pci_ecam_ops {
/* Default ECAM ops */
extern const struct pci_ecam_ops pci_generic_ecam_ops;
-struct pci_host_bridge *pci_host_common_probe(struct dt_device_node *dev,
- const struct pci_ecam_ops *ops,
- size_t priv_sz);
+struct pci_host_bridge *
+pci_host_common_probe(struct dt_device_node *dev,
+ const struct pci_ecam_ops *ops,
+ const struct pci_ecam_ops *child_ops,
+ size_t priv_sz);
int pci_generic_config_read(struct pci_host_bridge *bridge, pci_sbdf_t sbdf,
uint32_t reg, uint32_t len, uint32_t *value);
int pci_generic_config_write(struct pci_host_bridge *bridge, pci_sbdf_t sbdf,
uint32_t reg, uint32_t len, uint32_t value);
void __iomem *pci_ecam_map_bus(struct pci_host_bridge *bridge,
pci_sbdf_t sbdf, uint32_t where);
+void __iomem *pci_ecam_map_bus_generic(const struct pci_config_window *cfg,
+ const struct pci_ecam_ops *ops,
+ pci_sbdf_t sbdf, uint32_t where);
bool pci_ecam_need_p2m_hwdom_mapping(struct domain *d,
struct pci_host_bridge *bridge,
uint64_t addr);
diff --git a/xen/arch/arm/pci/ecam.c b/xen/arch/arm/pci/ecam.c
index 3987f96b01..5486881c5c 100644
--- a/xen/arch/arm/pci/ecam.c
+++ b/xen/arch/arm/pci/ecam.c
@@ -20,12 +20,10 @@
/*
* Function to implement the pci_ops->map_bus method.
*/
-void __iomem *pci_ecam_map_bus(struct pci_host_bridge *bridge,
- pci_sbdf_t sbdf, uint32_t where)
+void __iomem *pci_ecam_map_bus_generic(const struct pci_config_window *cfg,
+ const struct pci_ecam_ops *ops,
+ pci_sbdf_t sbdf, uint32_t where)
{
- const struct pci_config_window *cfg = bridge->cfg;
- const struct pci_ecam_ops *ops =
- container_of(bridge->ops, const struct pci_ecam_ops, pci_ops);
unsigned int devfn_shift = ops->bus_shift - 8;
void __iomem *base;
unsigned int busn = sbdf.bus;
@@ -39,6 +37,15 @@ void __iomem *pci_ecam_map_bus(struct pci_host_bridge *bridge,
return base + (sbdf.devfn << devfn_shift) + where;
}
+void __iomem *pci_ecam_map_bus(struct pci_host_bridge *bridge,
+ pci_sbdf_t sbdf, uint32_t where)
+{
+ const struct pci_ecam_ops *ops =
+ container_of(bridge->ops, const struct pci_ecam_ops, pci_ops);
+
+ return pci_ecam_map_bus_generic(bridge->cfg, ops, sbdf, where);
+}
+
bool __init pci_ecam_need_p2m_hwdom_mapping(struct domain *d,
struct pci_host_bridge *bridge,
uint64_t addr)
diff --git a/xen/arch/arm/pci/pci-access.c b/xen/arch/arm/pci/pci-access.c
index 9f9aac43d7..4da607ac3f 100644
--- a/xen/arch/arm/pci/pci-access.c
+++ b/xen/arch/arm/pci/pci-access.c
@@ -18,10 +18,31 @@
#define INVALID_VALUE (~0U)
#define PCI_ERR_VALUE(len) GENMASK(0, len * 8)
+static const struct pci_ops *get_ops(struct pci_host_bridge *bridge,
+ pci_sbdf_t sbdf)
+{
+ if ( bridge->child_ops )
+ {
+ struct pci_config_window* cfg = bridge->child_cfg;
+
+ if ( (sbdf.bus >= cfg->busn_start) && (sbdf.bus <= cfg->busn_end) )
+ return bridge->child_ops;
+ }
+ return bridge->ops;
+}
+
+static void __iomem *map_bus(struct pci_host_bridge *bridge, pci_sbdf_t sbdf,
+ uint32_t reg)
+{
+ const struct pci_ops *ops = get_ops(bridge, sbdf);
+
+ return ops->map_bus(bridge, sbdf, reg);
+}
+
int pci_generic_config_read(struct pci_host_bridge *bridge, pci_sbdf_t sbdf,
uint32_t reg, uint32_t len, uint32_t *value)
{
- void __iomem *addr = bridge->ops->map_bus(bridge, sbdf, reg);
+ void __iomem *addr = map_bus(bridge, sbdf, reg);
if ( !addr )
{
@@ -50,7 +71,7 @@ int pci_generic_config_read(struct pci_host_bridge *bridge, pci_sbdf_t sbdf,
int pci_generic_config_write(struct pci_host_bridge *bridge, pci_sbdf_t sbdf,
uint32_t reg, uint32_t len, uint32_t value)
{
- void __iomem *addr = bridge->ops->map_bus(bridge, sbdf, reg);
+ void __iomem *addr = map_bus(bridge, sbdf, reg);
if ( !addr )
return -ENODEV;
@@ -78,14 +99,16 @@ static uint32_t pci_config_read(pci_sbdf_t sbdf, unsigned int reg,
{
uint32_t val = PCI_ERR_VALUE(len);
struct pci_host_bridge *bridge = pci_find_host_bridge(sbdf.seg, sbdf.bus);
+ const struct pci_ops *ops;
if ( unlikely(!bridge) )
return val;
- if ( unlikely(!bridge->ops->read) )
+ ops = get_ops(bridge, sbdf);
+ if ( unlikely(!ops->read) )
return val;
- bridge->ops->read(bridge, sbdf, reg, len, &val);
+ ops->read(bridge, sbdf, reg, len, &val);
return val;
}
@@ -94,14 +117,16 @@ static void pci_config_write(pci_sbdf_t sbdf, unsigned int reg,
unsigned int len, uint32_t val)
{
struct pci_host_bridge *bridge = pci_find_host_bridge(sbdf.seg, sbdf.bus);
+ const struct pci_ops *ops;
if ( unlikely(!bridge) )
return;
- if ( unlikely(!bridge->ops->write) )
+ ops = get_ops(bridge, sbdf);
+ if ( unlikely(!ops->write) )
return;
- bridge->ops->write(bridge, sbdf, reg, len, val);
+ ops->write(bridge, sbdf, reg, len, val);
}
/*
diff --git a/xen/arch/arm/pci/pci-host-common.c b/xen/arch/arm/pci/pci-host-common.c
index e4e05d1176..3824146561 100644
--- a/xen/arch/arm/pci/pci-host-common.c
+++ b/xen/arch/arm/pci/pci-host-common.c
@@ -57,17 +57,12 @@ static void pci_ecam_free(struct pci_config_window *cfg)
xfree(cfg);
}
-static struct pci_config_window * __init
-gen_pci_init(struct dt_device_node *dev, const struct pci_ecam_ops *ops)
+static void __init gen_pci_init_bus_range(struct dt_device_node *dev,
+ struct pci_host_bridge *bridge,
+ struct pci_config_window *cfg)
{
- int err, cfg_reg_idx;
u32 bus_range[2];
- paddr_t addr, size;
- struct pci_config_window *cfg;
-
- cfg = xzalloc(struct pci_config_window);
- if ( !cfg )
- return NULL;
+ int err;
err = dt_property_read_u32_array(dev, "bus-range", bus_range,
ARRAY_SIZE(bus_range));
@@ -82,6 +77,36 @@ gen_pci_init(struct dt_device_node *dev, const struct pci_ecam_ops *ops)
if ( cfg->busn_end > cfg->busn_start + 0xff )
cfg->busn_end = cfg->busn_start + 0xff;
}
+}
+
+static void __init gen_pci_init_bus_range_child(struct dt_device_node *dev,
+ struct pci_host_bridge *bridge,
+ struct pci_config_window *cfg)
+{
+ cfg->busn_start = bridge->cfg->busn_start + 1;
+ cfg->busn_end = bridge->cfg->busn_end;
+ bridge->cfg->busn_end = bridge->cfg->busn_start;
+
+ printk(XENLOG_INFO "Root bus end updated: [bus %x-%x]\n",
+ bridge->cfg->busn_start, bridge->cfg->busn_end);
+}
+
+static struct pci_config_window * __init
+gen_pci_init(struct dt_device_node *dev, struct pci_host_bridge *bridge,
+ const struct pci_ecam_ops *ops,
+ void (*init_bus_range)(struct dt_device_node *dev,
+ struct pci_host_bridge *bridge,
+ struct pci_config_window *cfg))
+{
+ int err, cfg_reg_idx;
+ paddr_t addr, size;
+ struct pci_config_window *cfg;
+
+ cfg = xzalloc(struct pci_config_window);
+ if ( !cfg )
+ return NULL;
+
+ init_bus_range(dev, bridge, cfg);
if ( ops->cfg_reg_index )
{
@@ -208,9 +233,11 @@ static int pci_bus_find_domain_nr(struct dt_device_node *dev)
return domain;
}
-struct pci_host_bridge *pci_host_common_probe(struct dt_device_node *dev,
- const struct pci_ecam_ops *ops,
- size_t priv_sz)
+struct pci_host_bridge *
+pci_host_common_probe(struct dt_device_node *dev,
+ const struct pci_ecam_ops *ops,
+ const struct pci_ecam_ops *child_ops,
+ size_t priv_sz)
{
struct pci_host_bridge *bridge;
struct pci_config_window *cfg;
@@ -225,7 +252,7 @@ struct pci_host_bridge *pci_host_common_probe(struct dt_device_node *dev,
return ERR_PTR(-ENOMEM);
/* Parse and map our Configuration Space windows */
- cfg = gen_pci_init(dev, ops);
+ cfg = gen_pci_init(dev, bridge, ops, gen_pci_init_bus_range);
if ( !cfg )
{
err = -ENOMEM;
@@ -245,6 +272,21 @@ struct pci_host_bridge *pci_host_common_probe(struct dt_device_node *dev,
bridge->segment = domain;
+ if ( child_ops )
+ {
+ /* Parse and map child's Configuration Space windows */
+ cfg = gen_pci_init(dev, bridge, child_ops,
+ gen_pci_init_bus_range_child);
+ if ( !cfg )
+ {
+ err = -ENOMEM;
+ goto err_child;
+ }
+
+ bridge->child_cfg = cfg;
+ bridge->child_ops = &child_ops->pci_ops;
+ }
+
if ( priv_sz )
{
bridge->priv = xzalloc_bytes(priv_sz);
@@ -262,6 +304,9 @@ struct pci_host_bridge *pci_host_common_probe(struct dt_device_node *dev,
err_priv:
xfree(bridge->priv);
+err_child:
+ xfree(bridge->cfg);
+
err_exit:
xfree(bridge);
@@ -296,9 +341,11 @@ struct pci_host_bridge *pci_find_host_bridge(uint16_t segment, uint8_t bus)
{
if ( bridge->segment != segment )
continue;
- if ( (bus < bridge->cfg->busn_start) || (bus > bridge->cfg->busn_end) )
- continue;
- return bridge;
+ if ( bridge->child_cfg && (bus >= bridge->child_cfg->busn_start) &&
+ (bus <= bridge->child_cfg->busn_end) )
+ return bridge;
+ if ( (bus >= bridge->cfg->busn_start) && (bus <= bridge->cfg->busn_end) )
+ return bridge;
}
return NULL;
@@ -364,6 +411,7 @@ int __init pci_host_bridge_mappings(struct domain *d)
{
const struct dt_device_node *dev = bridge->dt_node;
unsigned int i;
+ bool need_mapping;
for ( i = 0; i < dt_number_of_address(dev); i++ )
{
@@ -379,7 +427,11 @@ int __init pci_host_bridge_mappings(struct domain *d)
return err;
}
- if ( bridge->ops->need_p2m_hwdom_mapping(d, bridge, addr) )
+ need_mapping = bridge->ops->need_p2m_hwdom_mapping(d, bridge, addr);
+ if ( need_mapping && bridge->child_ops )
+ need_mapping = bridge->child_ops->
+ need_p2m_hwdom_mapping(d, bridge, addr);
+ if ( need_mapping )
{
err = map_range_to_domain(dev, addr, size, &mr_data);
if ( err )
diff --git a/xen/arch/arm/pci/pci-host-generic.c b/xen/arch/arm/pci/pci-host-generic.c
index dde6a79a8e..08d94879b7 100644
--- a/xen/arch/arm/pci/pci-host-generic.c
+++ b/xen/arch/arm/pci/pci-host-generic.c
@@ -29,7 +29,7 @@ static const struct dt_device_match __initconstrel gen_pci_dt_match[] =
static int __init pci_host_generic_probe(struct dt_device_node *dev,
const void *data)
{
- return PTR_RET(pci_host_common_probe(dev, &pci_generic_ecam_ops, 0));
+ return PTR_RET(pci_host_common_probe(dev, &pci_generic_ecam_ops, NULL, 0));
}
DT_DEVICE_START(pci_gen, "PCI HOST GENERIC", DEVICE_PCI_HOSTBRIDGE)
diff --git a/xen/arch/arm/pci/pci-host-zynqmp.c b/xen/arch/arm/pci/pci-host-zynqmp.c
index c796447ac4..0b51ff6bd9 100644
--- a/xen/arch/arm/pci/pci-host-zynqmp.c
+++ b/xen/arch/arm/pci/pci-host-zynqmp.c
@@ -47,7 +47,7 @@ static const struct dt_device_match __initconstrel nwl_pcie_dt_match[] =
static int __init pci_host_generic_probe(struct dt_device_node *dev,
const void *data)
{
- return PTR_RET(pci_host_common_probe(dev, &nwl_pcie_ops, 0));
+ return PTR_RET(pci_host_common_probe(dev, &nwl_pcie_ops, NULL, 0));
}
DT_DEVICE_START(pci_gen, "PCI HOST ZYNQMP", DEVICE_PCI_HOSTBRIDGE)
diff --git a/xen/arch/arm/vpci.c b/xen/arch/arm/vpci.c
index b63a356bb4..ed665e9f26 100644
--- a/xen/arch/arm/vpci.c
+++ b/xen/arch/arm/vpci.c
@@ -8,15 +8,17 @@
#include <asm/mmio.h>
static pci_sbdf_t vpci_sbdf_from_gpa(const struct pci_host_bridge *bridge,
- paddr_t gpa)
+ paddr_t gpa, bool use_root)
{
pci_sbdf_t sbdf;
if ( bridge )
{
- sbdf.sbdf = VPCI_ECAM_BDF(gpa - bridge->cfg->phys_addr);
+ const struct pci_config_window *cfg = use_root ? bridge->cfg :
+ bridge->child_cfg;
+ sbdf.sbdf = VPCI_ECAM_BDF(gpa - cfg->phys_addr);
sbdf.seg = bridge->segment;
- sbdf.bus += bridge->cfg->busn_start;
+ sbdf.bus += cfg->busn_start;
}
else
sbdf.sbdf = VPCI_ECAM_BDF(gpa - GUEST_VPCI_ECAM_BASE);
@@ -24,11 +26,9 @@ static pci_sbdf_t vpci_sbdf_from_gpa(const struct pci_host_bridge *bridge,
return sbdf;
}
-static int vpci_mmio_read(struct vcpu *v, mmio_info_t *info,
- register_t *r, void *p)
+static int vpci_mmio_read(struct vcpu *v, mmio_info_t *info, register_t *r,
+ pci_sbdf_t sbdf)
{
- struct pci_host_bridge *bridge = p;
- pci_sbdf_t sbdf = vpci_sbdf_from_gpa(bridge, info->gpa);
const unsigned int access_size = (1U << info->dabt.size) * 8;
const register_t invalid = GENMASK_ULL(access_size - 1, 0);
/* data is needed to prevent a pointer cast on 32bit */
@@ -46,31 +46,78 @@ static int vpci_mmio_read(struct vcpu *v, mmio_info_t *info,
return 0;
}
-static int vpci_mmio_write(struct vcpu *v, mmio_info_t *info,
- register_t r, void *p)
+static int vpci_mmio_read_root(struct vcpu *v, mmio_info_t *info,
+ register_t *r, void *p)
{
struct pci_host_bridge *bridge = p;
- pci_sbdf_t sbdf = vpci_sbdf_from_gpa(bridge, info->gpa);
+ pci_sbdf_t sbdf = vpci_sbdf_from_gpa(bridge, info->gpa, true);
+ return vpci_mmio_read(v, info, r, sbdf);
+}
+
+static int vpci_mmio_read_child(struct vcpu *v, mmio_info_t *info,
+ register_t *r, void *p)
+{
+ struct pci_host_bridge *bridge = p;
+ pci_sbdf_t sbdf = vpci_sbdf_from_gpa(bridge, info->gpa, false);
+
+ return vpci_mmio_read(v, info, r, sbdf);
+}
+
+static int vpci_mmio_write(struct vcpu *v, mmio_info_t *info,
+ register_t r, pci_sbdf_t sbdf)
+{
return vpci_ecam_write(sbdf, ECAM_REG_OFFSET(info->gpa),
1U << info->dabt.size, r);
}
+static int vpci_mmio_write_root(struct vcpu *v, mmio_info_t *info,
+ register_t r, void *p)
+{
+ struct pci_host_bridge *bridge = p;
+ pci_sbdf_t sbdf = vpci_sbdf_from_gpa(bridge, info->gpa, true);
+
+ return vpci_mmio_write(v, info, r, sbdf);
+}
+
+static int vpci_mmio_write_child(struct vcpu *v, mmio_info_t *info,
+ register_t r, void *p)
+{
+ struct pci_host_bridge *bridge = p;
+ pci_sbdf_t sbdf = vpci_sbdf_from_gpa(bridge, info->gpa, false);
+
+ return vpci_mmio_write(v, info, r, sbdf);
+}
+
static const struct mmio_handler_ops vpci_mmio_handler = {
- .read = vpci_mmio_read,
- .write = vpci_mmio_write,
+ .read = vpci_mmio_read_root,
+ .write = vpci_mmio_write_root,
+};
+
+static const struct mmio_handler_ops vpci_mmio_handler_child = {
+ .read = vpci_mmio_read_child,
+ .write = vpci_mmio_write_child,
};
static int vpci_setup_mmio_handler_cb(struct domain *d,
struct pci_host_bridge *bridge)
{
struct pci_config_window *cfg = bridge->cfg;
+ int count = 1;
register_mmio_handler(d, &vpci_mmio_handler,
cfg->phys_addr, cfg->size, bridge);
- /* We have registered a single MMIO handler. */
- return 1;
+ if ( bridge->child_ops )
+ {
+ struct pci_config_window *cfg = bridge->child_cfg;
+
+ register_mmio_handler(d, &vpci_mmio_handler_child,
+ cfg->phys_addr, cfg->size, bridge);
+ count++;
+ }
+
+ return count;
}
int domain_vpci_init(struct domain *d)
@@ -101,8 +148,12 @@ int domain_vpci_init(struct domain *d)
static int vpci_get_num_handlers_cb(struct domain *d,
struct pci_host_bridge *bridge)
{
- /* Each bridge has a single MMIO handler for the configuration space. */
- return 1;
+ int count = 1;
+
+ if ( bridge->child_cfg )
+ count++;
+
+ return count;
}
unsigned int domain_vpci_get_num_mmio_handlers(struct domain *d)
--
2.34.1
^ permalink raw reply related [flat|nested] 13+ messages in thread* [PATCH v2 6/7] xen/arm: rcar4: add simple optimization to avoid ATU reprogramming
2025-03-11 10:24 [PATCH v2 0/7] Add support for R-Car Gen4 PCI host controller Mykyta Poturai
` (2 preceding siblings ...)
2025-03-11 10:24 ` [PATCH v2 3/7] xen/arm: add support for PCI child bus Mykyta Poturai
@ 2025-03-11 10:24 ` Mykyta Poturai
2025-03-12 10:04 ` Grygorii Strashko
2025-03-11 10:24 ` [PATCH v2 4/7] xen/arm: add support for R-Car Gen4 PCI host controller Mykyta Poturai
` (2 subsequent siblings)
6 siblings, 1 reply; 13+ messages in thread
From: Mykyta Poturai @ 2025-03-11 10:24 UTC (permalink / raw)
To: xen-devel@lists.xenproject.org
Cc: Volodymyr Babchuk, Stefano Stabellini, Julien Grall,
Bertrand Marquis, Michal Orzel, Volodymyr Babchuk, Mykyta Poturai
From: Volodymyr Babchuk <volodymyr_babchuk@epam.com>
There are high chances that there will be a number of a consecutive
accesses to configuration space of one device. To speed things up,
we can program ATU only during first access.
This is mostly beneficial taking into account the previous patch that
adds 1ms delay after ATU reprogramming.
Signed-off-by: Volodymyr Babchuk <volodymyr_babchuk@epam.com>
Signed-off-by: Mykyta Poturai <mykyta_poturai@epam.com>
---
v1->v2:
* rebased
---
xen/arch/arm/pci/pci-designware.c | 8 ++++++++
1 file changed, 8 insertions(+)
diff --git a/xen/arch/arm/pci/pci-designware.c b/xen/arch/arm/pci/pci-designware.c
index def2c12d63..cec52cf81a 100644
--- a/xen/arch/arm/pci/pci-designware.c
+++ b/xen/arch/arm/pci/pci-designware.c
@@ -272,6 +272,14 @@ static void dw_pcie_prog_outbound_atu(struct pci_host_bridge *pci, int index,
int type, uint64_t cpu_addr,
uint64_t pci_addr, uint64_t size)
{
+ static uint64_t prev_addr = ~0;
+
+ /* Simple optimization to not-program ATU for every transaction */
+ if (prev_addr == pci_addr)
+ return;
+
+ prev_addr = pci_addr;
+
__dw_pcie_prog_outbound_atu(pci, 0, index, type,
cpu_addr, pci_addr, size);
}
--
2.34.1
^ permalink raw reply related [flat|nested] 13+ messages in thread* Re: [PATCH v2 6/7] xen/arm: rcar4: add simple optimization to avoid ATU reprogramming
2025-03-11 10:24 ` [PATCH v2 6/7] xen/arm: rcar4: add simple optimization to avoid ATU reprogramming Mykyta Poturai
@ 2025-03-12 10:04 ` Grygorii Strashko
0 siblings, 0 replies; 13+ messages in thread
From: Grygorii Strashko @ 2025-03-12 10:04 UTC (permalink / raw)
To: Mykyta Poturai, xen-devel@lists.xenproject.org
Cc: Volodymyr Babchuk, Stefano Stabellini, Julien Grall,
Bertrand Marquis, Michal Orzel
On 11.03.25 12:24, Mykyta Poturai wrote:
> From: Volodymyr Babchuk <volodymyr_babchuk@epam.com>
>
> There are high chances that there will be a number of a consecutive
> accesses to configuration space of one device. To speed things up,
> we can program ATU only during first access.
>
> This is mostly beneficial taking into account the previous patch that
> adds 1ms delay after ATU reprogramming.
>
> Signed-off-by: Volodymyr Babchuk <volodymyr_babchuk@epam.com>
> Signed-off-by: Mykyta Poturai <mykyta_poturai@epam.com>
> ---
> v1->v2:
> * rebased
> ---
> xen/arch/arm/pci/pci-designware.c | 8 ++++++++
> 1 file changed, 8 insertions(+)
>
> diff --git a/xen/arch/arm/pci/pci-designware.c b/xen/arch/arm/pci/pci-designware.c
> index def2c12d63..cec52cf81a 100644
> --- a/xen/arch/arm/pci/pci-designware.c
> +++ b/xen/arch/arm/pci/pci-designware.c
> @@ -272,6 +272,14 @@ static void dw_pcie_prog_outbound_atu(struct pci_host_bridge *pci, int index,
> int type, uint64_t cpu_addr,
> uint64_t pci_addr, uint64_t size)
> {
First, using (hiding) static var inside the function deserve big, fat comment on why and what.
> + static uint64_t prev_addr = ~0;
Second, from an experience, static vars in functions is source of potential problems,
it's kinda land mines in code. For example, lets assume there are two pci_host_bridge
instances in Xen and they are accessed concurrently.
Best to get rid of static var here - may be move in pci_host_bridge, like cached_pci_addr? not sure
> +
> + /* Simple optimization to not-program ATU for every transaction */
> + if (prev_addr == pci_addr)
> + return;
> +
> + prev_addr = pci_addr;
> +
> __dw_pcie_prog_outbound_atu(pci, 0, index, type,
> cpu_addr, pci_addr, size);
> }
BR,
-grygorii
^ permalink raw reply [flat|nested] 13+ messages in thread
* [PATCH v2 4/7] xen/arm: add support for R-Car Gen4 PCI host controller
2025-03-11 10:24 [PATCH v2 0/7] Add support for R-Car Gen4 PCI host controller Mykyta Poturai
` (3 preceding siblings ...)
2025-03-11 10:24 ` [PATCH v2 6/7] xen/arm: rcar4: add simple optimization to avoid ATU reprogramming Mykyta Poturai
@ 2025-03-11 10:24 ` Mykyta Poturai
2025-03-11 10:24 ` [PATCH v2 5/7] xen/arm: rcar4: add delay after programming ATU Mykyta Poturai
2025-03-11 10:24 ` [PATCH v2 7/7] xen/arm: rcar4: program ATU to accesses to all functions Mykyta Poturai
6 siblings, 0 replies; 13+ messages in thread
From: Mykyta Poturai @ 2025-03-11 10:24 UTC (permalink / raw)
To: xen-devel@lists.xenproject.org
Cc: Oleksandr Andrushchenko, Stefano Stabellini, Julien Grall,
Bertrand Marquis, Michal Orzel, Volodymyr Babchuk, Mykyta Poturai
From: Oleksandr Andrushchenko <oleksandr_andrushchenko@epam.com>
Add support for Renesas R-Car Gen4 PCI host controller, specifically
targeting the S4 and V4H SoCs. The implementation includes configuration
read/write operations for both root and child buses. For accessing the
child bus, iATU is used for address translation.
Code common to all DesignWare PCI host controllers is located in a
separate file to allow for easy reuse in other DesignWare-based PCI
host controllers.
Signed-off-by: Oleksandr Andrushchenko <oleksandr_andrushchenko@epam.com>
Signed-off-by: Mykyta Poturai <mykyta_poturai@epam.com>
---
v1->v2:
* move designware code in a separate file
---
xen/arch/arm/pci/Makefile | 2 +
xen/arch/arm/pci/pci-designware.c | 409 ++++++++++++++++++++++++++++++
xen/arch/arm/pci/pci-designware.h | 105 ++++++++
xen/arch/arm/pci/pci-host-rcar4.c | 104 ++++++++
4 files changed, 620 insertions(+)
create mode 100644 xen/arch/arm/pci/pci-designware.c
create mode 100644 xen/arch/arm/pci/pci-designware.h
create mode 100644 xen/arch/arm/pci/pci-host-rcar4.c
diff --git a/xen/arch/arm/pci/Makefile b/xen/arch/arm/pci/Makefile
index 1d045ade01..ca6135e282 100644
--- a/xen/arch/arm/pci/Makefile
+++ b/xen/arch/arm/pci/Makefile
@@ -4,3 +4,5 @@ obj-y += pci-host-generic.o
obj-y += pci-host-common.o
obj-y += ecam.o
obj-y += pci-host-zynqmp.o
+obj-y += pci-designware.o
+obj-y += pci-host-rcar4.o
diff --git a/xen/arch/arm/pci/pci-designware.c b/xen/arch/arm/pci/pci-designware.c
new file mode 100644
index 0000000000..6ab03cf9b0
--- /dev/null
+++ b/xen/arch/arm/pci/pci-designware.c
@@ -0,0 +1,409 @@
+/*
+ * Based on Linux drivers/pci/controller/pci-host-common.c
+ * Based on Linux drivers/pci/controller/pci-host-generic.c
+ * Based on xen/arch/arm/pci/pci-host-generic.c
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <xen/delay.h>
+#include <asm/io.h>
+
+#include "pci-designware.h"
+/**
+ * upper_32_bits - return bits 32-63 of a number
+ * @n: the number we're accessing
+ *
+ * A basic shift-right of a 64- or 32-bit quantity. Use this to suppress
+ * the "right shift count >= width of type" warning when that quantity is
+ * 32-bits.
+ */
+#define upper_32_bits(n) ((uint32_t)(((n) >> 16) >> 16))
+
+/**
+ * lower_32_bits - return bits 0-31 of a number
+ * @n: the number we're accessing
+ */
+#define lower_32_bits(n) ((uint32_t)((n) & 0xffffffff))
+
+static int dw_pcie_read(void __iomem *addr, int size, uint32_t *val)
+{
+ if ( !IS_ALIGNED((uintptr_t)addr, size) )
+ {
+ *val = 0;
+ return PCIBIOS_BAD_REGISTER_NUMBER;
+ }
+
+ if (size == 4)
+ *val = readl(addr);
+ else if (size == 2)
+ *val = readw(addr);
+ else if (size == 1)
+ *val = readb(addr);
+ else
+ {
+ *val = 0;
+ return PCIBIOS_BAD_REGISTER_NUMBER;
+ }
+
+ return PCIBIOS_SUCCESSFUL;
+}
+
+static int dw_pcie_write(void __iomem *addr, int size, uint32_t val)
+{
+ if ( !IS_ALIGNED((uintptr_t)addr, size) )
+ return PCIBIOS_BAD_REGISTER_NUMBER;
+
+ if (size == 4)
+ writel(val, addr);
+ else if (size == 2)
+ writew(val, addr);
+ else if (size == 1)
+ writeb(val, addr);
+ else
+ return PCIBIOS_BAD_REGISTER_NUMBER;
+
+ return PCIBIOS_SUCCESSFUL;
+}
+
+static uint32_t dw_pcie_read_dbi(struct pci_host_bridge *bridge,
+ uint32_t reg, size_t size)
+{
+ void __iomem *addr = bridge->cfg->win + reg;
+ uint32_t val;
+
+ dw_pcie_read(addr, size, &val);
+ return val;
+}
+
+static void dw_pcie_write_dbi(struct pci_host_bridge *bridge,
+ uint32_t reg, size_t size, uint32_t val)
+{
+ void __iomem *addr = bridge->cfg->win + reg;
+
+ dw_pcie_write(addr, size, val);
+}
+
+static uint32_t dw_pcie_readl_dbi(struct pci_host_bridge *bridge, uint32_t reg)
+{
+ return dw_pcie_read_dbi(bridge, reg, sizeof(uint32_t));
+}
+
+static void dw_pcie_writel_dbi(struct pci_host_bridge *pci, uint32_t reg,
+ uint32_t val)
+{
+ dw_pcie_write_dbi(pci, reg, sizeof(uint32_t), val);
+}
+
+static void dw_pcie_read_iatu_unroll_enabled(struct pci_host_bridge *bridge)
+{
+ struct dw_pcie_priv *priv = bridge->priv;
+ uint32_t val;
+
+ val = dw_pcie_readl_dbi(bridge, PCIE_ATU_VIEWPORT);
+ if (val == 0xffffffff)
+ priv->iatu_unroll_enabled = true;
+
+ printk(XENLOG_DEBUG "%s iATU unroll: %sabled\n",
+ dt_node_full_name(bridge->dt_node),
+ priv->iatu_unroll_enabled ? "en" : "dis");
+}
+
+static uint32_t dw_pcie_readl_atu(struct pci_host_bridge *pci, uint32_t reg)
+{
+ struct dw_pcie_priv *priv = pci->priv;
+ int ret;
+ uint32_t val;
+
+ ret = dw_pcie_read(priv->atu_base + reg, 4, &val);
+ if ( ret )
+ printk(XENLOG_ERR "Read ATU address failed\n");
+
+ return val;
+}
+
+static void dw_pcie_writel_atu(struct pci_host_bridge *pci, uint32_t reg,
+ uint32_t val)
+{
+ struct dw_pcie_priv *priv = pci->priv;
+ int ret;
+
+ ret = dw_pcie_write(priv->atu_base + reg, 4, val);
+ if (ret)
+ printk(XENLOG_ERR "Write ATU address failed\n");
+}
+
+static uint32_t dw_pcie_readl_ob_unroll(struct pci_host_bridge *pci,
+ uint32_t index, uint32_t reg)
+{
+ uint32_t offset = PCIE_GET_ATU_OUTB_UNR_REG_OFFSET(index);
+
+ return dw_pcie_readl_atu(pci, offset + reg);
+}
+
+static void dw_pcie_writel_ob_unroll(struct pci_host_bridge *pci,
+ uint32_t index, uint32_t reg, uint32_t val)
+{
+ uint32_t offset = PCIE_GET_ATU_OUTB_UNR_REG_OFFSET(index);
+
+ dw_pcie_writel_atu(pci, offset + reg, val);
+}
+
+static uint32_t dw_pcie_enable_ecrc(uint32_t val)
+{
+ ASSERT_UNREACHABLE();
+ return 0;
+}
+
+static void dw_pcie_prog_outbound_atu_unroll(struct pci_host_bridge *pci,
+ uint8_t func_no, int index,
+ int type, uint64_t cpu_addr,
+ uint64_t pci_addr, uint64_t size)
+{
+ struct dw_pcie_priv *priv = pci->priv;
+ uint32_t retries, val;
+ uint64_t limit_addr = cpu_addr + size - 1;
+
+ dw_pcie_writel_ob_unroll(pci, index, PCIE_ATU_UNR_LOWER_BASE,
+ lower_32_bits(cpu_addr));
+ dw_pcie_writel_ob_unroll(pci, index, PCIE_ATU_UNR_UPPER_BASE,
+ upper_32_bits(cpu_addr));
+ dw_pcie_writel_ob_unroll(pci, index, PCIE_ATU_UNR_LOWER_LIMIT,
+ lower_32_bits(limit_addr));
+ dw_pcie_writel_ob_unroll(pci, index, PCIE_ATU_UNR_UPPER_LIMIT,
+ upper_32_bits(limit_addr));
+ dw_pcie_writel_ob_unroll(pci, index, PCIE_ATU_UNR_LOWER_TARGET,
+ lower_32_bits(pci_addr));
+ dw_pcie_writel_ob_unroll(pci, index, PCIE_ATU_UNR_UPPER_TARGET,
+ upper_32_bits(pci_addr));
+ val = type | PCIE_ATU_FUNC_NUM(func_no);
+ val = upper_32_bits(size - 1) ? val | PCIE_ATU_INCREASE_REGION_SIZE : val;
+ if (priv->version == 0x490A)
+ val = dw_pcie_enable_ecrc(val);
+ dw_pcie_writel_ob_unroll(pci, index, PCIE_ATU_UNR_REGION_CTRL1, val);
+ dw_pcie_writel_ob_unroll(pci, index, PCIE_ATU_UNR_REGION_CTRL2,
+ PCIE_ATU_ENABLE);
+
+ /*
+ * Make sure ATU enable takes effect before any subsequent config
+ * and I/O accesses.
+ */
+ for (retries = 0; retries < LINK_WAIT_MAX_IATU_RETRIES; retries++)
+ {
+ val = dw_pcie_readl_ob_unroll(pci, index,
+ PCIE_ATU_UNR_REGION_CTRL2);
+ if (val & PCIE_ATU_ENABLE)
+ return;
+
+ mdelay(LINK_WAIT_IATU);
+ }
+ printk(XENLOG_ERR "Outbound iATU is not being enabled\n");
+}
+
+static void __dw_pcie_prog_outbound_atu(struct pci_host_bridge *pci,
+ uint8_t func_no, int index, int type,
+ uint64_t cpu_addr, uint64_t pci_addr,
+ uint64_t size)
+{
+ struct dw_pcie_priv *priv = pci->priv;
+ uint32_t retries, val;
+
+ if (priv->iatu_unroll_enabled)
+ {
+ dw_pcie_prog_outbound_atu_unroll(pci, func_no, index, type,
+ cpu_addr, pci_addr, size);
+ return;
+ }
+
+ dw_pcie_writel_dbi(pci, PCIE_ATU_VIEWPORT,
+ PCIE_ATU_REGION_OUTBOUND | index);
+ dw_pcie_writel_dbi(pci, PCIE_ATU_LOWER_BASE,
+ lower_32_bits(cpu_addr));
+ dw_pcie_writel_dbi(pci, PCIE_ATU_UPPER_BASE,
+ upper_32_bits(cpu_addr));
+ dw_pcie_writel_dbi(pci, PCIE_ATU_LIMIT,
+ lower_32_bits(cpu_addr + size - 1));
+ if (priv->version >= 0x460A)
+ dw_pcie_writel_dbi(pci, PCIE_ATU_UPPER_LIMIT,
+ upper_32_bits(cpu_addr + size - 1));
+ dw_pcie_writel_dbi(pci, PCIE_ATU_LOWER_TARGET,
+ lower_32_bits(pci_addr));
+ dw_pcie_writel_dbi(pci, PCIE_ATU_UPPER_TARGET,
+ upper_32_bits(pci_addr));
+ val = type | PCIE_ATU_FUNC_NUM(func_no);
+ val = ((upper_32_bits(size - 1)) && (priv->version >= 0x460A)) ?
+ val | PCIE_ATU_INCREASE_REGION_SIZE : val;
+ if (priv->version == 0x490A)
+ val = dw_pcie_enable_ecrc(val);
+ dw_pcie_writel_dbi(pci, PCIE_ATU_CR1, val);
+ dw_pcie_writel_dbi(pci, PCIE_ATU_CR2, PCIE_ATU_ENABLE);
+
+ /*
+ * Make sure ATU enable takes effect before any subsequent config
+ * and I/O accesses.
+ */
+ for (retries = 0; retries < LINK_WAIT_MAX_IATU_RETRIES; retries++)
+ {
+ val = dw_pcie_readl_dbi(pci, PCIE_ATU_CR2);
+ if (val & PCIE_ATU_ENABLE)
+ return;
+
+ mdelay(LINK_WAIT_IATU);
+ }
+ printk(XENLOG_ERR "Outbound iATU is not being enabled\n");
+}
+
+static void dw_pcie_prog_outbound_atu(struct pci_host_bridge *pci, int index,
+ int type, uint64_t cpu_addr,
+ uint64_t pci_addr, uint64_t size)
+{
+ __dw_pcie_prog_outbound_atu(pci, 0, index, type,
+ cpu_addr, pci_addr, size);
+}
+
+void dw_pcie_set_version(struct pci_host_bridge *bridge, unsigned int version)
+{
+ struct dw_pcie_priv *priv = bridge->priv;
+
+ priv->version = version;
+}
+
+void __iomem *dw_pcie_child_map_bus(struct pci_host_bridge *bridge,
+ pci_sbdf_t sbdf, uint32_t where)
+{
+ uint32_t busdev;
+
+ busdev = PCIE_ATU_BUS(sbdf.bus) | PCIE_ATU_DEV(PCI_SLOT(sbdf.devfn)) |
+ PCIE_ATU_FUNC(PCI_FUNC(sbdf.devfn));
+
+ /* FIXME: Parent is the root bus, so use PCIE_ATU_TYPE_CFG0. */
+ dw_pcie_prog_outbound_atu(bridge, PCIE_ATU_REGION_INDEX1,
+ PCIE_ATU_TYPE_CFG0,
+ bridge->child_cfg->phys_addr,
+ busdev, bridge->child_cfg->size);
+
+ return bridge->child_cfg->win + where;
+}
+
+int dw_pcie_child_config_read(struct pci_host_bridge *bridge,
+ pci_sbdf_t sbdf, uint32_t reg,
+ uint32_t len, uint32_t *value)
+{
+ struct dw_pcie_priv *priv = bridge->priv;
+ int ret;
+
+ /*
+ * FIXME: we cannot read iATU settings at the early initialization
+ * (probe) as the host's HW is not yet initialized at that phase.
+ * This read operation is the very first thing Domain-0 will do
+ * during its initialization, so take this opportunity and read
+ * iATU setting now.
+ */
+ if ( unlikely(!priv->iatu_unroll_initilized) )
+ {
+ dw_pcie_read_iatu_unroll_enabled(bridge);
+ priv->iatu_unroll_initilized = true;
+ }
+
+ ret = pci_generic_config_read(bridge, sbdf, reg, len, value);
+ if ( !ret && (priv->num_viewport <= 2) )
+ dw_pcie_prog_outbound_atu(bridge, PCIE_ATU_REGION_INDEX1,
+ PCIE_ATU_TYPE_IO,
+ bridge->child_cfg->phys_addr,
+ 0, bridge->child_cfg->size);
+
+ return ret;
+}
+
+int dw_pcie_child_config_write(struct pci_host_bridge *bridge,
+ pci_sbdf_t sbdf, uint32_t reg,
+ uint32_t len, uint32_t value)
+{
+ struct dw_pcie_priv *priv = bridge->priv;
+ int ret;
+
+ ret = pci_generic_config_write(bridge, sbdf, reg, len, value);
+ if ( !ret && (priv->num_viewport <= 2) )
+ dw_pcie_prog_outbound_atu(bridge, PCIE_ATU_REGION_INDEX1,
+ PCIE_ATU_TYPE_IO,
+ bridge->child_cfg->phys_addr,
+ 0, bridge->child_cfg->size);
+ return ret;
+}
+
+bool __init dw_pcie_child_need_p2m_hwdom_mapping(struct domain *d,
+ struct pci_host_bridge *bridge,
+ uint64_t addr)
+{
+ struct pci_config_window *cfg = bridge->child_cfg;
+
+ /*
+ * We do not want ECAM address space to be mapped in Domain-0's p2m,
+ * so we can trap access to it.
+ */
+ return cfg->phys_addr != addr;
+}
+
+struct pci_host_bridge * __init
+dw_pcie_host_probe(struct dt_device_node *dev, const void *data,
+ const struct pci_ecam_ops *ops,
+ const struct pci_ecam_ops *child_ops)
+{
+
+ struct pci_host_bridge *bridge;
+ struct dw_pcie_priv *priv;
+
+ paddr_t atu_phys_addr;
+ paddr_t atu_size;
+ int atu_idx, ret;
+
+ bridge = pci_host_common_probe(dev, ops, child_ops, sizeof(*priv));
+ if ( IS_ERR(bridge) )
+ return bridge;
+
+ priv = bridge->priv;
+
+ atu_idx = dt_property_match_string(dev, "reg-names", "atu");
+ if ( atu_idx < 0 )
+ {
+ printk(XENLOG_ERR "Cannot find \"atu\" range index in device tree\n");
+ return ERR_PTR(atu_idx);
+ }
+ ret = dt_device_get_address(dev, atu_idx, &atu_phys_addr, &atu_size);
+ if ( ret )
+ {
+ printk(XENLOG_ERR "Cannot find \"atu\" range in device tree\n");
+ return ERR_PTR(ret);
+ }
+ printk("iATU at [mem 0x%" PRIpaddr "-0x%" PRIpaddr "]\n",
+ atu_phys_addr, atu_phys_addr + atu_size - 1);
+ priv->atu_base = ioremap_nocache(atu_phys_addr, atu_size);
+ if ( !priv->atu_base )
+ {
+ printk(XENLOG_ERR "iATU ioremap failed\n");
+ return ERR_PTR(ENXIO);
+ }
+
+ if ( !dt_property_read_u32(dev, "num-viewport", &priv->num_viewport) )
+ priv->num_viewport = 2;
+
+ /*
+ * FIXME: we cannot read iATU unroll enable now as the host bridge's
+ * HW is not yet initialized by Domain-0: leave it for later.
+ */
+
+ printk(XENLOG_INFO "%s number of view ports: %d\n", dt_node_full_name(dev),
+ priv->num_viewport);
+
+ return bridge;
+}
diff --git a/xen/arch/arm/pci/pci-designware.h b/xen/arch/arm/pci/pci-designware.h
new file mode 100644
index 0000000000..f00ebc6be4
--- /dev/null
+++ b/xen/arch/arm/pci/pci-designware.h
@@ -0,0 +1,105 @@
+/*
+ * Based on Linux drivers/pci/controller/pci-host-common.c
+ * Based on Linux drivers/pci/controller/pci-host-generic.c
+ * Based on xen/arch/arm/pci/pci-host-generic.c
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <xen/pci.h>
+#include <xen/init.h>
+
+#ifndef __PCI_DESIGNWARE_H__
+#define __PCI_DESIGNWARE_H__
+
+#define PCIBIOS_SUCCESSFUL 0x00
+#define PCIBIOS_BAD_REGISTER_NUMBER 0x87
+
+#define PCIE_ATU_VIEWPORT 0x900
+#define PCIE_ATU_REGION_OUTBOUND 0
+#define PCIE_ATU_CR1 0x904
+#define PCIE_ATU_INCREASE_REGION_SIZE BIT(13, UL)
+#define PCIE_ATU_CR2 0x908
+#define PCIE_ATU_ENABLE BIT(31, UL)
+#define PCIE_ATU_LOWER_BASE 0x90C
+#define PCIE_ATU_UPPER_BASE 0x910
+#define PCIE_ATU_LIMIT 0x914
+#define PCIE_ATU_LOWER_TARGET 0x918
+#define PCIE_ATU_UPPER_TARGET 0x91C
+#define PCIE_ATU_UPPER_LIMIT 0x924
+
+#define PCIE_ATU_REGION_INDEX1 0x1
+#define PCIE_ATU_TYPE_IO 0x2
+#define PCIE_ATU_TYPE_CFG0 0x4
+
+#define FIELD_PREP(_mask, _val) \
+ (((typeof(_mask))(_val) << (ffs64(_mask) - 1)) & (_mask))
+
+#define PCIE_ATU_BUS(x) FIELD_PREP(GENMASK(31, 24), x)
+#define PCIE_ATU_DEV(x) FIELD_PREP(GENMASK(23, 19), x)
+#define PCIE_ATU_FUNC(x) FIELD_PREP(GENMASK(18, 16), x)
+
+/* Register address builder */
+#define PCIE_GET_ATU_OUTB_UNR_REG_OFFSET(region) \
+ ((region) << 9)
+
+/*
+ * iATU Unroll-specific register definitions
+ * From 4.80 core version the address translation will be made by unroll
+ */
+#define PCIE_ATU_UNR_REGION_CTRL1 0x00
+#define PCIE_ATU_UNR_REGION_CTRL2 0x04
+#define PCIE_ATU_UNR_LOWER_BASE 0x08
+#define PCIE_ATU_UNR_UPPER_BASE 0x0C
+#define PCIE_ATU_UNR_LOWER_LIMIT 0x10
+#define PCIE_ATU_UNR_LOWER_TARGET 0x14
+#define PCIE_ATU_UNR_UPPER_TARGET 0x18
+#define PCIE_ATU_UNR_UPPER_LIMIT 0x20
+
+#define PCIE_ATU_FUNC_NUM(pf) ((pf) << 20)
+
+/* Parameters for the waiting for iATU enabled routine */
+#define LINK_WAIT_MAX_IATU_RETRIES 5
+#define LINK_WAIT_IATU 9
+
+struct dw_pcie_priv
+{
+ uint32_t num_viewport;
+ bool iatu_unroll_initilized;
+ bool iatu_unroll_enabled;
+ void __iomem *atu_base;
+ unsigned int version;
+};
+
+void dw_pcie_set_version(struct pci_host_bridge *bridge, unsigned int version);
+
+void __iomem *dw_pcie_child_map_bus(struct pci_host_bridge *bridge,
+ pci_sbdf_t sbdf, uint32_t where);
+
+int dw_pcie_child_config_read(struct pci_host_bridge *bridge,
+ pci_sbdf_t sbdf, uint32_t reg,
+ uint32_t len, uint32_t *value);
+
+int dw_pcie_child_config_write(struct pci_host_bridge *bridge,
+ pci_sbdf_t sbdf, uint32_t reg,
+ uint32_t len, uint32_t value);
+
+bool __init dw_pcie_child_need_p2m_hwdom_mapping(struct domain *d,
+ struct pci_host_bridge *bridge,
+ uint64_t addr);
+
+struct pci_host_bridge * __init
+dw_pcie_host_probe(struct dt_device_node *dev, const void *data,
+ const struct pci_ecam_ops *ops,
+ const struct pci_ecam_ops *child_ops);
+#endif /* __PCI_DESIGNWARE_H__ */
diff --git a/xen/arch/arm/pci/pci-host-rcar4.c b/xen/arch/arm/pci/pci-host-rcar4.c
new file mode 100644
index 0000000000..f6ae3c5685
--- /dev/null
+++ b/xen/arch/arm/pci/pci-host-rcar4.c
@@ -0,0 +1,104 @@
+/*
+ * Based on Linux drivers/pci/controller/pci-host-common.c
+ * Based on Linux drivers/pci/controller/pci-host-generic.c
+ * Based on xen/arch/arm/pci/pci-host-generic.c
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <xen/init.h>
+#include <xen/pci.h>
+
+#include <asm/device.h>
+#include <asm/io.h>
+#include <asm/pci.h>
+
+#include "pci-designware.h"
+
+#define RCAR4_DWC_VERSION 0x520A
+
+/*
+ * PCI host bridges often have different ways to access the root and child
+ * bus config spaces:
+ * "dbi" : the aperture where root port's own configuration registers
+ * are available.
+ * "config": child's configuration space
+ * "atu" : iATU registers for DWC version 4.80 or later
+ */
+static int __init rcar4_cfg_reg_index(struct dt_device_node *np)
+{
+ return dt_property_match_string(np, "reg-names", "dbi");
+}
+
+static int __init rcar4_child_cfg_reg_index(struct dt_device_node *np)
+{
+ return dt_property_match_string(np, "reg-names", "config");
+}
+
+/* ECAM ops */
+const struct pci_ecam_ops rcar4_pcie_ops = {
+ .bus_shift = 20,
+ .cfg_reg_index = rcar4_cfg_reg_index,
+ .pci_ops = {
+ .map_bus = pci_ecam_map_bus,
+ .read = pci_generic_config_read,
+ .write = pci_generic_config_write,
+ .need_p2m_hwdom_mapping = pci_ecam_need_p2m_hwdom_mapping,
+ }
+};
+
+const struct pci_ecam_ops rcar4_pcie_child_ops = {
+ .bus_shift = 20,
+ .cfg_reg_index = rcar4_child_cfg_reg_index,
+ .pci_ops = {
+ .map_bus = dw_pcie_child_map_bus,
+ .read = dw_pcie_child_config_read,
+ .write = dw_pcie_child_config_write,
+ .need_p2m_hwdom_mapping = dw_pcie_child_need_p2m_hwdom_mapping,
+ }
+};
+
+static const struct dt_device_match __initconstrel rcar4_pcie_dt_match[] =
+{
+ { .compatible = "renesas,r8a779f0-pcie" },
+ { .compatible = "renesas,r8a779g0-pcie" },
+ { },
+};
+
+static int __init pci_host_generic_probe(struct dt_device_node *dev,
+ const void *data)
+{
+ struct pci_host_bridge *bridge;
+
+ bridge = dw_pcie_host_probe(dev, data, &rcar4_pcie_ops,
+ &rcar4_pcie_child_ops);
+
+ dw_pcie_set_version(bridge, RCAR4_DWC_VERSION);
+
+ return 0;
+}
+
+DT_DEVICE_START(pci_gen, "PCI HOST R-CAR GEN4", DEVICE_PCI_HOSTBRIDGE)
+.dt_match = rcar4_pcie_dt_match,
+.init = pci_host_generic_probe,
+DT_DEVICE_END
+
+/*
+ * Local variables:
+ * mode: C
+ * c-file-style: "BSD"
+ * c-basic-offset: 4
+ * tab-width: 4
+ * indent-tabs-mode: nil
+ * End:
+ */
--
2.34.1
^ permalink raw reply related [flat|nested] 13+ messages in thread* [PATCH v2 5/7] xen/arm: rcar4: add delay after programming ATU
2025-03-11 10:24 [PATCH v2 0/7] Add support for R-Car Gen4 PCI host controller Mykyta Poturai
` (4 preceding siblings ...)
2025-03-11 10:24 ` [PATCH v2 4/7] xen/arm: add support for R-Car Gen4 PCI host controller Mykyta Poturai
@ 2025-03-11 10:24 ` Mykyta Poturai
2025-03-12 10:12 ` Grygorii Strashko
2025-03-11 10:24 ` [PATCH v2 7/7] xen/arm: rcar4: program ATU to accesses to all functions Mykyta Poturai
6 siblings, 1 reply; 13+ messages in thread
From: Mykyta Poturai @ 2025-03-11 10:24 UTC (permalink / raw)
To: xen-devel@lists.xenproject.org
Cc: Volodymyr Babchuk, Stefano Stabellini, Julien Grall,
Bertrand Marquis, Michal Orzel, Volodymyr Babchuk, Mykyta Poturai
From: Volodymyr Babchuk <volodymyr_babchuk@epam.com>
For some reason, we need a delay before accessing ATU region after
we programmed it. Otherwise, we'll get erroneous TLP.
There is a code below, which should do this in proper way, by polling
CTRL2 register, but according to documentation, hardware does not
change this ATU_ENABLE bit at all.
Signed-off-by: Volodymyr Babchuk <volodymyr_babchuk@epam.com>
Signed-off-by: Mykyta Poturai <mykyta_poturai@epam.com>
---
v1->v2:
* rebased
---
xen/arch/arm/pci/pci-designware.c | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/xen/arch/arm/pci/pci-designware.c b/xen/arch/arm/pci/pci-designware.c
index 6ab03cf9b0..def2c12d63 100644
--- a/xen/arch/arm/pci/pci-designware.c
+++ b/xen/arch/arm/pci/pci-designware.c
@@ -194,6 +194,11 @@ static void dw_pcie_prog_outbound_atu_unroll(struct pci_host_bridge *pci,
dw_pcie_writel_ob_unroll(pci, index, PCIE_ATU_UNR_REGION_CTRL2,
PCIE_ATU_ENABLE);
+ /*
+ * HACK: We need to delay there, because the next code does not
+ * work as expected on S4
+ */
+ mdelay(1);
/*
* Make sure ATU enable takes effect before any subsequent config
* and I/O accesses.
--
2.34.1
^ permalink raw reply related [flat|nested] 13+ messages in thread* Re: [PATCH v2 5/7] xen/arm: rcar4: add delay after programming ATU
2025-03-11 10:24 ` [PATCH v2 5/7] xen/arm: rcar4: add delay after programming ATU Mykyta Poturai
@ 2025-03-12 10:12 ` Grygorii Strashko
2025-03-12 11:29 ` Mykyta Poturai
0 siblings, 1 reply; 13+ messages in thread
From: Grygorii Strashko @ 2025-03-12 10:12 UTC (permalink / raw)
To: Mykyta Poturai, xen-devel@lists.xenproject.org
Cc: Volodymyr Babchuk, Stefano Stabellini, Julien Grall,
Bertrand Marquis, Michal Orzel
On 11.03.25 12:24, Mykyta Poturai wrote:
> From: Volodymyr Babchuk <volodymyr_babchuk@epam.com>
>
> For some reason, we need a delay before accessing ATU region after
> we programmed it. Otherwise, we'll get erroneous TLP.
>
> There is a code below, which should do this in proper way, by polling
> CTRL2 register, but according to documentation, hardware does not
> change this ATU_ENABLE bit at all.
>
> Signed-off-by: Volodymyr Babchuk <volodymyr_babchuk@epam.com>
> Signed-off-by: Mykyta Poturai <mykyta_poturai@epam.com>
> ---
> v1->v2:
> * rebased
> ---
> xen/arch/arm/pci/pci-designware.c | 5 +++++
> 1 file changed, 5 insertions(+)
>
> diff --git a/xen/arch/arm/pci/pci-designware.c b/xen/arch/arm/pci/pci-designware.c
> index 6ab03cf9b0..def2c12d63 100644
> --- a/xen/arch/arm/pci/pci-designware.c
> +++ b/xen/arch/arm/pci/pci-designware.c
> @@ -194,6 +194,11 @@ static void dw_pcie_prog_outbound_atu_unroll(struct pci_host_bridge *pci,
> dw_pcie_writel_ob_unroll(pci, index, PCIE_ATU_UNR_REGION_CTRL2,
> PCIE_ATU_ENABLE);
>
> + /*
> + * HACK: We need to delay there, because the next code does not
> + * work as expected on S4
> + */
> + mdelay(1);
It seems like a HACK to WA some platform issue, but in its current form it
will affect all DW PCI compatible platforms.
So, it is required a proper solution for the affected platform to be found, or
some sort of DW PCI "quirk"s processing code be introduced.
I'd recommend to drop this patch for now from this series.
> /*
> * Make sure ATU enable takes effect before any subsequent config
> * and I/O accesses.
BR,
-grygorii
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: [PATCH v2 5/7] xen/arm: rcar4: add delay after programming ATU
2025-03-12 10:12 ` Grygorii Strashko
@ 2025-03-12 11:29 ` Mykyta Poturai
0 siblings, 0 replies; 13+ messages in thread
From: Mykyta Poturai @ 2025-03-12 11:29 UTC (permalink / raw)
To: Grygorii Strashko, xen-devel@lists.xenproject.org
Cc: Volodymyr Babchuk, Stefano Stabellini, Julien Grall,
Bertrand Marquis, Michal Orzel
On 12.03.25 12:12, Grygorii Strashko wrote:
>
>
> On 11.03.25 12:24, Mykyta Poturai wrote:
>> From: Volodymyr Babchuk <volodymyr_babchuk@epam.com>
>>
>> For some reason, we need a delay before accessing ATU region after
>> we programmed it. Otherwise, we'll get erroneous TLP.
>>
>> There is a code below, which should do this in proper way, by polling
>> CTRL2 register, but according to documentation, hardware does not
>> change this ATU_ENABLE bit at all.
>>
>> Signed-off-by: Volodymyr Babchuk <volodymyr_babchuk@epam.com>
>> Signed-off-by: Mykyta Poturai <mykyta_poturai@epam.com>
>> ---
>> v1->v2:
>> * rebased
>> ---
>> xen/arch/arm/pci/pci-designware.c | 5 +++++
>> 1 file changed, 5 insertions(+)
>>
>> diff --git a/xen/arch/arm/pci/pci-designware.c
>> b/xen/arch/arm/pci/pci-designware.c
>> index 6ab03cf9b0..def2c12d63 100644
>> --- a/xen/arch/arm/pci/pci-designware.c
>> +++ b/xen/arch/arm/pci/pci-designware.c
>> @@ -194,6 +194,11 @@ static void
>> dw_pcie_prog_outbound_atu_unroll(struct pci_host_bridge *pci,
>> dw_pcie_writel_ob_unroll(pci, index, PCIE_ATU_UNR_REGION_CTRL2,
>> PCIE_ATU_ENABLE);
>> + /*
>> + * HACK: We need to delay there, because the next code does not
>> + * work as expected on S4
>> + */
>> + mdelay(1);
>
> It seems like a HACK to WA some platform issue, but in its current form it
> will affect all DW PCI compatible platforms.
>
> So, it is required a proper solution for the affected platform to be
> found, or
> some sort of DW PCI "quirk"s processing code be introduced.
>
> I'd recommend to drop this patch for now from this series.
>
>> /*
>> * Make sure ATU enable takes effect before any subsequent config
>> * and I/O accesses.
>
> BR,
> -grygorii
Hi Grygorii
After some further investigations I have retested this on V4H, and it
seems to work fine without delays. Considering this and the issue with
static variables and multiple PCI hosts I think I will drop ATU related
workarounds until some proper quirks handling system is established.
--
Mykyta
^ permalink raw reply [flat|nested] 13+ messages in thread
* [PATCH v2 7/7] xen/arm: rcar4: program ATU to accesses to all functions
2025-03-11 10:24 [PATCH v2 0/7] Add support for R-Car Gen4 PCI host controller Mykyta Poturai
` (5 preceding siblings ...)
2025-03-11 10:24 ` [PATCH v2 5/7] xen/arm: rcar4: add delay after programming ATU Mykyta Poturai
@ 2025-03-11 10:24 ` Mykyta Poturai
6 siblings, 0 replies; 13+ messages in thread
From: Mykyta Poturai @ 2025-03-11 10:24 UTC (permalink / raw)
To: xen-devel@lists.xenproject.org
Cc: Volodymyr Babchuk, Stefano Stabellini, Julien Grall,
Bertrand Marquis, Michal Orzel, Volodymyr Babchuk, Mykyta Poturai
From: Volodymyr Babchuk <volodymyr_babchuk@epam.com>
According to ATU documentation, bits [18:16] of accessed memory
address correspond to a function number. This is somewhat similar to
ECAM, but with huge holes between regions.
We can use this to minimize number of ATU re-programmings: configure
ATU to access BDF with F=0 and adjust memory address with function
number.
Taking into account the previous patch, that optimizes ATU
reprogramming by skipping call to __dw_pcie_prog_outbound_atu() if we
already configured pci_address, we can be sure that accesses to all
functions of one device will not trigger ATU reprogramming at all.
Signed-off-by: Volodymyr Babchuk <volodymyr_babchuk@epam.com>
Signed-off-by: Mykyta Poturai <mykyta_poturai@epam.com>
---
v1->v2:
* rebased
---
xen/arch/arm/pci/pci-designware.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/xen/arch/arm/pci/pci-designware.c b/xen/arch/arm/pci/pci-designware.c
index cec52cf81a..fec8c7296d 100644
--- a/xen/arch/arm/pci/pci-designware.c
+++ b/xen/arch/arm/pci/pci-designware.c
@@ -297,7 +297,7 @@ void __iomem *dw_pcie_child_map_bus(struct pci_host_bridge *bridge,
uint32_t busdev;
busdev = PCIE_ATU_BUS(sbdf.bus) | PCIE_ATU_DEV(PCI_SLOT(sbdf.devfn)) |
- PCIE_ATU_FUNC(PCI_FUNC(sbdf.devfn));
+ PCIE_ATU_FUNC(0);
/* FIXME: Parent is the root bus, so use PCIE_ATU_TYPE_CFG0. */
dw_pcie_prog_outbound_atu(bridge, PCIE_ATU_REGION_INDEX1,
@@ -305,7 +305,7 @@ void __iomem *dw_pcie_child_map_bus(struct pci_host_bridge *bridge,
bridge->child_cfg->phys_addr,
busdev, bridge->child_cfg->size);
- return bridge->child_cfg->win + where;
+ return bridge->child_cfg->win + ((uint32_t)sbdf.fn << 16) + where;
}
int dw_pcie_child_config_read(struct pci_host_bridge *bridge,
--
2.34.1
^ permalink raw reply related [flat|nested] 13+ messages in thread