[PATCH] PCI: Add pci_assign_resource_fixed -- allow fixed address assignments

[PATCH] PCI: Add pci_assign_resource_fixed -- allow fixed address assignments

[Kumar Gala]

On some embedded systems the PCI address for hotplug devices are not only
known a priori but are required to be at a given PCI address for other
master in the system to be able to access.

An example of such a system would be an FPGA which is setup from user space
after the system has booted.  The FPGA may be access by DSPs in the system
and those DSPs expect the FPGA at a fixed PCI address.

Added pci_assign_resource_fixed() as a way to allow assignment of the PCI
devices's BARs at fixed PCI addresses.

Signed-off-by: Kumar Gala <galak@kernel.crashing.org>

---
commit cb0b39d6c2f3986dd86976062b0f899fa35358ae
tree 4b0af5b9d630113fd4fdcfb35b61313f2b3efb41
parent 2be4d50295e2b6f62c07b614e1b103e280dddb84
author Kumar Gala <galak@kernel.crashing.org> Thu, 27 Apr 2006 12:45:02 -0500
committer Kumar Gala <galak@kernel.crashing.org> Thu, 27 Apr 2006 12:45:02 -0500

 drivers/pci/pci.c       |    1 +
 drivers/pci/setup-res.c |   35 +++++++++++++++++++++++++++++++++++
 include/linux/pci.h     |    1 +
 3 files changed, 37 insertions(+), 0 deletions(-)

diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
index 2329f94..5dc7c14 100644
--- a/drivers/pci/pci.c
+++ b/drivers/pci/pci.c
@@ -947,6 +947,7 @@ EXPORT_SYMBOL_GPL(pci_intx);
 EXPORT_SYMBOL(pci_set_dma_mask);
 EXPORT_SYMBOL(pci_set_consistent_dma_mask);
 EXPORT_SYMBOL(pci_assign_resource);
+EXPORT_SYMBOL(pci_assign_resource_fixed);
 EXPORT_SYMBOL(pci_find_parent_resource);
 
 EXPORT_SYMBOL(pci_set_power_state);
diff --git a/drivers/pci/setup-res.c b/drivers/pci/setup-res.c
index ea9277b..f485958 100644
--- a/drivers/pci/setup-res.c
+++ b/drivers/pci/setup-res.c
@@ -155,6 +155,41 @@ int pci_assign_resource(struct pci_dev *
 	return ret;
 }
 
+int pci_assign_resource_fixed(struct pci_dev *dev, int resno)
+{
+	struct pci_bus *bus = dev->bus;
+	struct resource *res = dev->resource + resno;
+	unsigned int type_mask;
+	int i, ret = -EBUSY;
+
+	type_mask = IORESOURCE_IO | IORESOURCE_MEM | IORESOURCE_PREFETCH;
+
+	for (i = 0; i < PCI_BUS_NUM_RESOURCES; i++) {
+		struct resource *r = bus->resource[i];
+		if (!r)
+			continue;
+
+		/* type_mask must match */
+		if ((res->flags ^ r->flags) & type_mask)
+			continue;
+
+		ret = request_resource(r, res);
+
+		if (ret == 0)
+			break;
+	}
+
+	if (ret) {
+		printk(KERN_ERR "PCI: Failed to allocate %s resource #%d:%lx@%lx for %s\n",
+		       res->flags & IORESOURCE_IO ? "I/O" : "mem",
+		       resno, res->end - res->start + 1, res->start, pci_name(dev));
+	} else if (resno < PCI_BRIDGE_RESOURCES) {
+		pci_update_resource(dev, res, resno);
+	}
+
+	return ret;
+}
+
 /* Sort resources by alignment */
 void __devinit
 pdev_sort_resources(struct pci_dev *dev, struct resource_list *head)
diff --git a/include/linux/pci.h b/include/linux/pci.h
index 3a6a4e3..704dae3 100644
--- a/include/linux/pci.h
+++ b/include/linux/pci.h
@@ -496,6 +496,7 @@ int pci_set_dma_mask(struct pci_dev *dev
 int pci_set_consistent_dma_mask(struct pci_dev *dev, u64 mask);
 void pci_update_resource(struct pci_dev *dev, struct resource *res, int resno);
 int pci_assign_resource(struct pci_dev *dev, int i);
+int pci_assign_resource_fixed(struct pci_dev *dev, int i);
 void pci_restore_bars(struct pci_dev *dev);
 
 /* ROM control related routines */

Re: [PATCH] PCI: Add pci_assign_resource_fixed -- allow fixed address assignments

[Andrew Morton]

Kumar Gala <galak@kernel.crashing.org> wrote:
>
> On some embedded systems the PCI address for hotplug devices are not only
> known a priori but are required to be at a given PCI address for other
> master in the system to be able to access.
> 
> An example of such a system would be an FPGA which is setup from user space
> after the system has booted.  The FPGA may be access by DSPs in the system
> and those DSPs expect the FPGA at a fixed PCI address.
> 
> Added pci_assign_resource_fixed() as a way to allow assignment of the PCI
> devices's BARs at fixed PCI addresses.

Is there any sane way in which we can arrange for this function to not be
present in vmlinux's which don't need it?

Options would be

a) Put it in a .a file.

   - messy from a source perspective

   - doesn't work if the only reference is from a module

   - small gains anyway.

b) Use CONFIG_EMBEDDED.

?

Re: [PATCH] PCI: Add pci_assign_resource_fixed -- allow fixed address assignments

[Kumar Gala]

On Apr 27, 2006, at 5:34 PM, Andrew Morton wrote:

> Kumar Gala <galak@kernel.crashing.org> wrote:
>>
>> On some embedded systems the PCI address for hotplug devices are  
>> not only
>> known a priori but are required to be at a given PCI address for  
>> other
>> master in the system to be able to access.
>>
>> An example of such a system would be an FPGA which is setup from  
>> user space
>> after the system has booted.  The FPGA may be access by DSPs in  
>> the system
>> and those DSPs expect the FPGA at a fixed PCI address.
>>
>> Added pci_assign_resource_fixed() as a way to allow assignment of  
>> the PCI
>> devices's BARs at fixed PCI addresses.
>
> Is there any sane way in which we can arrange for this function to  
> not be
> present in vmlinux's which don't need it?
>
> Options would be
>
> a) Put it in a .a file.
>
>    - messy from a source perspective
>
>    - doesn't work if the only reference is from a module
>
>    - small gains anyway.
>
> b) Use CONFIG_EMBEDDED.

I'm fine with wrapping it in a CONFIG_EMBEDDED, Greg?

- k

Re: [PATCH] PCI: Add pci_assign_resource_fixed -- allow fixed address assignments

[Greg KH]

On Thu, Apr 27, 2006 at 07:12:14PM -0500, Kumar Gala wrote:
> 
> On Apr 27, 2006, at 5:34 PM, Andrew Morton wrote:
> 
> >Kumar Gala <galak@kernel.crashing.org> wrote:
> >>
> >>On some embedded systems the PCI address for hotplug devices are  
> >>not only
> >>known a priori but are required to be at a given PCI address for  
> >>other
> >>master in the system to be able to access.
> >>
> >>An example of such a system would be an FPGA which is setup from  
> >>user space
> >>after the system has booted.  The FPGA may be access by DSPs in  
> >>the system
> >>and those DSPs expect the FPGA at a fixed PCI address.
> >>
> >>Added pci_assign_resource_fixed() as a way to allow assignment of  
> >>the PCI
> >>devices's BARs at fixed PCI addresses.
> >
> >Is there any sane way in which we can arrange for this function to  
> >not be
> >present in vmlinux's which don't need it?
> >
> >Options would be
> >
> >a) Put it in a .a file.
> >
> >   - messy from a source perspective
> >
> >   - doesn't work if the only reference is from a module
> >
> >   - small gains anyway.
> >
> >b) Use CONFIG_EMBEDDED.
> 
> I'm fine with wrapping it in a CONFIG_EMBEDDED, Greg?

That's fine with me.  Care to send me an updated patch?  I'll drop the
other one then.

thanks,

greg k-h

[PATCH][UPDATE] PCI: Add pci_assign_resource_fixed -- allow fixed address assignments

[Kumar Gala]

On some embedded systems the PCI address for hotplug devices are not only
known a priori but are required to be at a given PCI address for other
master in the system to be able to access.

An example of such a system would be an FPGA which is setup from user space
after the system has booted.  The FPGA may be access by DSPs in the system
and those DSPs expect the FPGA at a fixed PCI address.

Added pci_assign_resource_fixed() as a way to allow assignment of the PCI
devices's BARs at fixed PCI addresses.

Signed-off-by: Kumar Gala <galak@kernel.crashing.org>

---
commit 094733d858fc30c7a5412485b22f94d47b3efd86
tree 8d3b461566134baf72e085ee85d06bcc1fbd800f
parent 2be4d50295e2b6f62c07b614e1b103e280dddb84
author Kumar Gala <galak@kernel.crashing.org> Thu, 27 Apr 2006 23:30:43 -0500
committer Kumar Gala <galak@kernel.crashing.org> Thu, 27 Apr 2006 23:30:43 -0500

 drivers/pci/pci.c       |    3 +++
 drivers/pci/setup-res.c |   37 +++++++++++++++++++++++++++++++++++++
 include/linux/pci.h     |    3 +++
 3 files changed, 43 insertions(+), 0 deletions(-)

diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
index 2329f94..955a96e 100644
--- a/drivers/pci/pci.c
+++ b/drivers/pci/pci.c
@@ -947,6 +947,9 @@ EXPORT_SYMBOL_GPL(pci_intx);
 EXPORT_SYMBOL(pci_set_dma_mask);
 EXPORT_SYMBOL(pci_set_consistent_dma_mask);
 EXPORT_SYMBOL(pci_assign_resource);
+#ifdef CONFIG_EMBEDDED
+EXPORT_SYMBOL(pci_assign_resource_fixed);
+#endif
 EXPORT_SYMBOL(pci_find_parent_resource);
 
 EXPORT_SYMBOL(pci_set_power_state);
diff --git a/drivers/pci/setup-res.c b/drivers/pci/setup-res.c
index ea9277b..fbfb461 100644
--- a/drivers/pci/setup-res.c
+++ b/drivers/pci/setup-res.c
@@ -155,6 +155,43 @@ int pci_assign_resource(struct pci_dev *
 	return ret;
 }
 
+#ifdef CONFIG_EMBEDDED
+int pci_assign_resource_fixed(struct pci_dev *dev, int resno)
+{
+	struct pci_bus *bus = dev->bus;
+	struct resource *res = dev->resource + resno;
+	unsigned int type_mask;
+	int i, ret = -EBUSY;
+
+	type_mask = IORESOURCE_IO | IORESOURCE_MEM | IORESOURCE_PREFETCH;
+
+	for (i = 0; i < PCI_BUS_NUM_RESOURCES; i++) {
+		struct resource *r = bus->resource[i];
+		if (!r)
+			continue;
+
+		/* type_mask must match */
+		if ((res->flags ^ r->flags) & type_mask)
+			continue;
+
+		ret = request_resource(r, res);
+
+		if (ret == 0)
+			break;
+	}
+
+	if (ret) {
+		printk(KERN_ERR "PCI: Failed to allocate %s resource #%d:%lx@%lx for %s\n",
+		       res->flags & IORESOURCE_IO ? "I/O" : "mem",
+		       resno, res->end - res->start + 1, res->start, pci_name(dev));
+	} else if (resno < PCI_BRIDGE_RESOURCES) {
+		pci_update_resource(dev, res, resno);
+	}
+
+	return ret;
+}
+#endif
+
 /* Sort resources by alignment */
 void __devinit
 pdev_sort_resources(struct pci_dev *dev, struct resource_list *head)
diff --git a/include/linux/pci.h b/include/linux/pci.h
index 3a6a4e3..c7c5399 100644
--- a/include/linux/pci.h
+++ b/include/linux/pci.h
@@ -496,6 +496,9 @@ int pci_set_dma_mask(struct pci_dev *dev
 int pci_set_consistent_dma_mask(struct pci_dev *dev, u64 mask);
 void pci_update_resource(struct pci_dev *dev, struct resource *res, int resno);
 int pci_assign_resource(struct pci_dev *dev, int i);
+#ifdef CONFIG_EMBEDDED
+int pci_assign_resource_fixed(struct pci_dev *dev, int i);
+#endif
 void pci_restore_bars(struct pci_dev *dev);
 
 /* ROM control related routines */

Re: [PATCH][UPDATE] PCI: Add pci_assign_resource_fixed -- allow fixed address assignments

[Andrew Morton]

Kumar Gala <galak@kernel.crashing.org> wrote:
>
> diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
> index 2329f94..955a96e 100644
> --- a/drivers/pci/pci.c
> +++ b/drivers/pci/pci.c
> @@ -947,6 +947,9 @@ EXPORT_SYMBOL_GPL(pci_intx);
>  EXPORT_SYMBOL(pci_set_dma_mask);
>  EXPORT_SYMBOL(pci_set_consistent_dma_mask);
>  EXPORT_SYMBOL(pci_assign_resource);
> +#ifdef CONFIG_EMBEDDED
> +EXPORT_SYMBOL(pci_assign_resource_fixed);
> +#endif

This is a good argument for putting the export at the definition site ;)

> --- a/include/linux/pci.h
> +++ b/include/linux/pci.h
> @@ -496,6 +496,9 @@ int pci_set_dma_mask(struct pci_dev *dev
>  int pci_set_consistent_dma_mask(struct pci_dev *dev, u64 mask);
>  void pci_update_resource(struct pci_dev *dev, struct resource *res, int resno);
>  int pci_assign_resource(struct pci_dev *dev, int i);
> +#ifdef CONFIG_EMBEDDED
> +int pci_assign_resource_fixed(struct pci_dev *dev, int i);
> +#endif

Debatable - if we omit the ifdefs, it fails at link time or depmod time. 
The ifdefs will make it warn (but not fail) at compile-time.  Given that
the warning is non-fatal, the ifdefs don't add much value and are best
omitted.

Re: [PATCH][UPDATE] PCI: Add pci_assign_resource_fixed -- allow fixed address assignments

[Kumar Gala]

On Apr 27, 2006, at 11:43 PM, Andrew Morton wrote:

> Kumar Gala <galak@kernel.crashing.org> wrote:
>>
>> diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
>> index 2329f94..955a96e 100644
>> --- a/drivers/pci/pci.c
>> +++ b/drivers/pci/pci.c
>> @@ -947,6 +947,9 @@ EXPORT_SYMBOL_GPL(pci_intx);
>>  EXPORT_SYMBOL(pci_set_dma_mask);
>>  EXPORT_SYMBOL(pci_set_consistent_dma_mask);
>>  EXPORT_SYMBOL(pci_assign_resource);
>> +#ifdef CONFIG_EMBEDDED
>> +EXPORT_SYMBOL(pci_assign_resource_fixed);
>> +#endif
>
> This is a good argument for putting the export at the definition  
> site ;)
>
>> --- a/include/linux/pci.h
>> +++ b/include/linux/pci.h
>> @@ -496,6 +496,9 @@ int pci_set_dma_mask(struct pci_dev *dev
>>  int pci_set_consistent_dma_mask(struct pci_dev *dev, u64 mask);
>>  void pci_update_resource(struct pci_dev *dev, struct resource  
>> *res, int resno);
>>  int pci_assign_resource(struct pci_dev *dev, int i);
>> +#ifdef CONFIG_EMBEDDED
>> +int pci_assign_resource_fixed(struct pci_dev *dev, int i);
>> +#endif
>
> Debatable - if we omit the ifdefs, it fails at link time or depmod  
> time.
> The ifdefs will make it warn (but not fail) at compile-time.  Given  
> that
> the warning is non-fatal, the ifdefs don't add much value and are best
> omitted.

I'll make these changes.  I didn't care for the ifdef in the header  
but was following the example already in it for isa_bridge.

- k

[PATCH][UPDATE] PCI: Add pci_assign_resource_fixed -- allow fixed address assignments

[Kumar Gala]

PCI: Add pci_assign_resource_fixed -- allow fixed address assignments

On some embedded systems the PCI address for hotplug devices are not only
known a priori but are required to be at a given PCI address for other
master in the system to be able to access.

An example of such a system would be an FPGA which is setup from user space
after the system has booted.  The FPGA may be access by DSPs in the system
and those DSPs expect the FPGA at a fixed PCI address.

Added pci_assign_resource_fixed() as a way to allow assignment of the PCI
devices's BARs at fixed PCI addresses.

Signed-off-by: Kumar Gala <galak@kernel.crashing.org>

---
commit a6dbd2dcbccd760291bb098d5ee4a72cd4ccd45b
tree 7b10a03750e293901a2ce192d7d8055381810e2d
parent 2be4d50295e2b6f62c07b614e1b103e280dddb84
author Kumar Gala <galak@kernel.crashing.org> Mon, 01 May 2006 10:46:00 -0500
committer Kumar Gala <galak@kernel.crashing.org> Mon, 01 May 2006 10:46:00 -0500

 drivers/pci/setup-res.c |   40 ++++++++++++++++++++++++++++++++++++++++
 include/linux/pci.h     |    1 +
 2 files changed, 41 insertions(+), 0 deletions(-)

diff --git a/drivers/pci/setup-res.c b/drivers/pci/setup-res.c
index ea9277b..5ae09d2 100644
--- a/drivers/pci/setup-res.c
+++ b/drivers/pci/setup-res.c
@@ -155,6 +155,46 @@ int pci_assign_resource(struct pci_dev *
 	return ret;
 }
 
+#ifdef CONFIG_EMBEDDED
+int pci_assign_resource_fixed(struct pci_dev *dev, int resno)
+{
+	struct pci_bus *bus = dev->bus;
+	struct resource *res = dev->resource + resno;
+	unsigned int type_mask;
+	int i, ret = -EBUSY;
+
+	type_mask = IORESOURCE_IO | IORESOURCE_MEM | IORESOURCE_PREFETCH;
+
+	for (i = 0; i < PCI_BUS_NUM_RESOURCES; i++) {
+		struct resource *r = bus->resource[i];
+		if (!r)
+			continue;
+
+		/* type_mask must match */
+		if ((res->flags ^ r->flags) & type_mask)
+			continue;
+
+		ret = request_resource(r, res);
+
+		if (ret == 0)
+			break;
+	}
+
+	if (ret) {
+		printk(KERN_ERR "PCI: Failed to allocate %s resource "
+				"#%d:%llx@%llx for %s\n",
+			res->flags & IORESOURCE_IO ? "I/O" : "mem",
+			resno, (unsigned long long)(res->end - res->start + 1),
+			(unsigned long long)res->start, pci_name(dev));
+	} else if (resno < PCI_BRIDGE_RESOURCES) {
+		pci_update_resource(dev, res, resno);
+	}
+
+	return ret;
+}
+EXPORT_SYMBOL(pci_assign_resource_fixed);
+#endif
+
 /* Sort resources by alignment */
 void __devinit
 pdev_sort_resources(struct pci_dev *dev, struct resource_list *head)
diff --git a/include/linux/pci.h b/include/linux/pci.h
index 3a6a4e3..704dae3 100644
--- a/include/linux/pci.h
+++ b/include/linux/pci.h
@@ -496,6 +496,7 @@ int pci_set_dma_mask(struct pci_dev *dev
 int pci_set_consistent_dma_mask(struct pci_dev *dev, u64 mask);
 void pci_update_resource(struct pci_dev *dev, struct resource *res, int resno);
 int pci_assign_resource(struct pci_dev *dev, int i);
+int pci_assign_resource_fixed(struct pci_dev *dev, int i);
 void pci_restore_bars(struct pci_dev *dev);
 
 /* ROM control related routines */

Re: [PATCH][UPDATE] PCI: Add pci_assign_resource_fixed -- allow fixed address assignments

[Pavel Machek]

Hi!

> On some embedded systems the PCI address for hotplug devices are not only
> known a priori but are required to be at a given PCI address for other
> master in the system to be able to access.

I'm afraid that CONFIG_EMBEDDED is wrong option for this, I'm
afraid. It was supposed to mean "enable memory-saving options".

-- 
Thanks for all the (sleeping) penguins.

[PATCH] PCI: Add pci_assign_resource_fixed -- allow fixed address assignments

[Kumar Gala]

On some embedded systems the PCI address for hotplug devices are not only
known a priori but are required to be at a given PCI address for other
master in the system to be able to access.

An example of such a system would be an FPGA which is setup from user space
after the system has booted.  The FPGA may be access by DSPs in the system
and those DSPs expect the FPGA at a fixed PCI address.

Added pci_assign_resource_fixed() as a way to allow assignment of the PCI
devices's BARs at fixed PCI addresses.

Signed-off-by: Kumar Gala <galak@kernel.crashing.org>

---
commit 45d4a23317c459865ec740c80b6e2a2ad9f53fd3
tree 432b5e41ef5f231dd57eb1a98f103239c62d63a0
parent 8176dee014ec6ad1039b8c0075c9c1d02147c2c8
author Kumar Gala <galak@kernel.crashing.org> Thu, 09 Mar 2006 12:34:25 -0600
committer Kumar Gala <galak@kernel.crashing.org> Thu, 09 Mar 2006 12:34:25 -0600

 drivers/pci/pci.c       |    1 +
 drivers/pci/setup-res.c |   35 +++++++++++++++++++++++++++++++++++
 include/linux/pci.h     |    1 +
 3 files changed, 37 insertions(+), 0 deletions(-)

diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
index d2d1879..2557e86 100644
--- a/drivers/pci/pci.c
+++ b/drivers/pci/pci.c
@@ -935,6 +935,7 @@ EXPORT_SYMBOL_GPL(pci_intx);
 EXPORT_SYMBOL(pci_set_dma_mask);
 EXPORT_SYMBOL(pci_set_consistent_dma_mask);
 EXPORT_SYMBOL(pci_assign_resource);
+EXPORT_SYMBOL(pci_assign_resource_fixed);
 EXPORT_SYMBOL(pci_find_parent_resource);
 
 EXPORT_SYMBOL(pci_set_power_state);
diff --git a/drivers/pci/setup-res.c b/drivers/pci/setup-res.c
index ea9277b..f485958 100644
--- a/drivers/pci/setup-res.c
+++ b/drivers/pci/setup-res.c
@@ -155,6 +155,41 @@ int pci_assign_resource(struct pci_dev *
 	return ret;
 }
 
+int pci_assign_resource_fixed(struct pci_dev *dev, int resno)
+{
+	struct pci_bus *bus = dev->bus;
+	struct resource *res = dev->resource + resno;
+	unsigned int type_mask;
+	int i, ret = -EBUSY;
+
+	type_mask = IORESOURCE_IO | IORESOURCE_MEM | IORESOURCE_PREFETCH;
+
+	for (i = 0; i < PCI_BUS_NUM_RESOURCES; i++) {
+		struct resource *r = bus->resource[i];
+		if (!r)
+			continue;
+
+		/* type_mask must match */
+		if ((res->flags ^ r->flags) & type_mask)
+			continue;
+
+		ret = request_resource(r, res);
+
+		if (ret == 0)
+			break;
+	}
+
+	if (ret) {
+		printk(KERN_ERR "PCI: Failed to allocate %s resource #%d:%lx@%lx for %s\n",
+		       res->flags & IORESOURCE_IO ? "I/O" : "mem",
+		       resno, res->end - res->start + 1, res->start, pci_name(dev));
+	} else if (resno < PCI_BRIDGE_RESOURCES) {
+		pci_update_resource(dev, res, resno);
+	}
+
+	return ret;
+}
+
 /* Sort resources by alignment */
 void __devinit
 pdev_sort_resources(struct pci_dev *dev, struct resource_list *head)
diff --git a/include/linux/pci.h b/include/linux/pci.h
index fe1a2b0..0db1e2d 100644
--- a/include/linux/pci.h
+++ b/include/linux/pci.h
@@ -460,6 +460,7 @@ int pci_set_dma_mask(struct pci_dev *dev
 int pci_set_consistent_dma_mask(struct pci_dev *dev, u64 mask);
 void pci_update_resource(struct pci_dev *dev, struct resource *res, int resno);
 int pci_assign_resource(struct pci_dev *dev, int i);
+int pci_assign_resource_fixed(struct pci_dev *dev, int i);
 void pci_restore_bars(struct pci_dev *dev);
 
 /* ROM control related routines */

Re: [PATCH] PCI: Add pci_assign_resource_fixed -- allow fixed address assignments

[Randy.Dunlap]

On Fri, 10 Mar 2006 09:06:32 -0600 (CST) Kumar Gala wrote:

> On some embedded systems the PCI address for hotplug devices are not only
> known a priori but are required to be at a given PCI address for other
> master in the system to be able to access.
> 
> An example of such a system would be an FPGA which is setup from user space
> after the system has booted.  The FPGA may be access by DSPs in the system
> and those DSPs expect the FPGA at a fixed PCI address.
> 
> Added pci_assign_resource_fixed() as a way to allow assignment of the PCI
> devices's BARs at fixed PCI addresses.
> 
> Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
> 
> ---
> commit 45d4a23317c459865ec740c80b6e2a2ad9f53fd3
> tree 432b5e41ef5f231dd57eb1a98f103239c62d63a0
> parent 8176dee014ec6ad1039b8c0075c9c1d02147c2c8
> author Kumar Gala <galak@kernel.crashing.org> Thu, 09 Mar 2006 12:34:25 -0600
> committer Kumar Gala <galak@kernel.crashing.org> Thu, 09 Mar 2006 12:34:25 -0600
> 
>  drivers/pci/pci.c       |    1 +
>  drivers/pci/setup-res.c |   35 +++++++++++++++++++++++++++++++++++
>  include/linux/pci.h     |    1 +
>  3 files changed, 37 insertions(+), 0 deletions(-)
> 
> diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
> index d2d1879..2557e86 100644
> --- a/drivers/pci/pci.c
> +++ b/drivers/pci/pci.c
> @@ -935,6 +935,7 @@ EXPORT_SYMBOL_GPL(pci_intx);
>  EXPORT_SYMBOL(pci_set_dma_mask);
>  EXPORT_SYMBOL(pci_set_consistent_dma_mask);
>  EXPORT_SYMBOL(pci_assign_resource);
> +EXPORT_SYMBOL(pci_assign_resource_fixed);
>  EXPORT_SYMBOL(pci_find_parent_resource);
>  
>  EXPORT_SYMBOL(pci_set_power_state);
> diff --git a/drivers/pci/setup-res.c b/drivers/pci/setup-res.c
> index ea9277b..f485958 100644
> --- a/drivers/pci/setup-res.c
> +++ b/drivers/pci/setup-res.c
> @@ -155,6 +155,41 @@ int pci_assign_resource(struct pci_dev *
>  	return ret;
>  }
>  
> +int pci_assign_resource_fixed(struct pci_dev *dev, int resno)
> +{
> +	struct pci_bus *bus = dev->bus;
> +	struct resource *res = dev->resource + resno;
> +	unsigned int type_mask;
> +	int i, ret = -EBUSY;
> +
> +	type_mask = IORESOURCE_IO | IORESOURCE_MEM | IORESOURCE_PREFETCH;

If type_mask must match (as in comment below), should it be a
parameter instead of hard-coded here?  Does this match your FPGA
resource?  It may not match my <hypothetical> resource.

> +	for (i = 0; i < PCI_BUS_NUM_RESOURCES; i++) {
> +		struct resource *r = bus->resource[i];
> +		if (!r)
> +			continue;
> +
> +		/* type_mask must match */
> +		if ((res->flags ^ r->flags) & type_mask)
> +			continue;
> +
> +		ret = request_resource(r, res);
> +
> +		if (ret == 0)
> +			break;
> +	}
> +
> +	if (ret) {
> +		printk(KERN_ERR "PCI: Failed to allocate %s resource #%d:%lx@%lx for %s\n",
> +		       res->flags & IORESOURCE_IO ? "I/O" : "mem",
> +		       resno, res->end - res->start + 1, res->start, pci_name(dev));
> +	} else if (resno < PCI_BRIDGE_RESOURCES) {
> +		pci_update_resource(dev, res, resno);
> +	}

braces not needed.

> +
> +	return ret;
> +}
> +


---
~Randy
Please use an email client that implements proper (compliant) threading.
(You know who you are.)

Re: [PATCH] PCI: Add pci_assign_resource_fixed -- allow fixed address assignments

[Kumar Gala]

On Mar 10, 2006, at 3:04 PM, Randy.Dunlap wrote:

> On Fri, 10 Mar 2006 09:06:32 -0600 (CST) Kumar Gala wrote:
>
>> On some embedded systems the PCI address for hotplug devices are  
>> not only
>> known a priori but are required to be at a given PCI address for  
>> other
>> master in the system to be able to access.
>>
>> An example of such a system would be an FPGA which is setup from  
>> user space
>> after the system has booted.  The FPGA may be access by DSPs in  
>> the system
>> and those DSPs expect the FPGA at a fixed PCI address.
>>
>> Added pci_assign_resource_fixed() as a way to allow assignment of  
>> the PCI
>> devices's BARs at fixed PCI addresses.
>>
>> Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
>>
>> ---
>> commit 45d4a23317c459865ec740c80b6e2a2ad9f53fd3
>> tree 432b5e41ef5f231dd57eb1a98f103239c62d63a0
>> parent 8176dee014ec6ad1039b8c0075c9c1d02147c2c8
>> author Kumar Gala <galak@kernel.crashing.org> Thu, 09 Mar 2006  
>> 12:34:25 -0600
>> committer Kumar Gala <galak@kernel.crashing.org> Thu, 09 Mar 2006  
>> 12:34:25 -0600
>>
>>  drivers/pci/pci.c       |    1 +
>>  drivers/pci/setup-res.c |   35 +++++++++++++++++++++++++++++++++++
>>  include/linux/pci.h     |    1 +
>>  3 files changed, 37 insertions(+), 0 deletions(-)
>>
>> diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
>> index d2d1879..2557e86 100644
>> --- a/drivers/pci/pci.c
>> +++ b/drivers/pci/pci.c
>> @@ -935,6 +935,7 @@ EXPORT_SYMBOL_GPL(pci_intx);
>>  EXPORT_SYMBOL(pci_set_dma_mask);
>>  EXPORT_SYMBOL(pci_set_consistent_dma_mask);
>>  EXPORT_SYMBOL(pci_assign_resource);
>> +EXPORT_SYMBOL(pci_assign_resource_fixed);
>>  EXPORT_SYMBOL(pci_find_parent_resource);
>>
>>  EXPORT_SYMBOL(pci_set_power_state);
>> diff --git a/drivers/pci/setup-res.c b/drivers/pci/setup-res.c
>> index ea9277b..f485958 100644
>> --- a/drivers/pci/setup-res.c
>> +++ b/drivers/pci/setup-res.c
>> @@ -155,6 +155,41 @@ int pci_assign_resource(struct pci_dev *
>>  	return ret;
>>  }
>>
>> +int pci_assign_resource_fixed(struct pci_dev *dev, int resno)
>> +{
>> +	struct pci_bus *bus = dev->bus;
>> +	struct resource *res = dev->resource + resno;
>> +	unsigned int type_mask;
>> +	int i, ret = -EBUSY;
>> +
>> +	type_mask = IORESOURCE_IO | IORESOURCE_MEM | IORESOURCE_PREFETCH;
>
> If type_mask must match (as in comment below), should it be a
> parameter instead of hard-coded here?  Does this match your FPGA
> resource?  It may not match my <hypothetical> resource.

I was trying to ensure an exact match between the bus and device  
resource types.  I figured if you were calling this API you should be  
able to ensure that the resource types match exactly.

>> +	for (i = 0; i < PCI_BUS_NUM_RESOURCES; i++) {
>> +		struct resource *r = bus->resource[i];
>> +		if (!r)
>> +			continue;
>> +
>> +		/* type_mask must match */
>> +		if ((res->flags ^ r->flags) & type_mask)
>> +			continue;
>> +
>> +		ret = request_resource(r, res);
>> +
>> +		if (ret == 0)
>> +			break;
>> +	}
>> +
>> +	if (ret) {
>> +		printk(KERN_ERR "PCI: Failed to allocate %s resource #%d:%lx@% 
>> lx for %s\n",
>> +		       res->flags & IORESOURCE_IO ? "I/O" : "mem",
>> +		       resno, res->end - res->start + 1, res->start, pci_name 
>> (dev));
>> +	} else if (resno < PCI_BRIDGE_RESOURCES) {
>> +		pci_update_resource(dev, res, resno);
>> +	}
>
> braces not needed.

Fair, need to go find where I stole this from and possibly fix extra  
braces there as well.

>
>> +
>> +	return ret;
>> +}
>> +
>
>
> ---
> ~Randy
> Please use an email client that implements proper (compliant)  
> threading.
> (You know who you are.)

Re: [PATCH] PCI: Add pci_assign_resource_fixed -- allow fixed address assignments

[Kumar Gala]

Greg,

Any comments on this?

- kumar

On Mar 10, 2006, at 3:30 PM, Kumar Gala wrote:

>
> On Mar 10, 2006, at 3:04 PM, Randy.Dunlap wrote:
>
>> On Fri, 10 Mar 2006 09:06:32 -0600 (CST) Kumar Gala wrote:
>>
>>> On some embedded systems the PCI address for hotplug devices are  
>>> not only
>>> known a priori but are required to be at a given PCI address for  
>>> other
>>> master in the system to be able to access.
>>>
>>> An example of such a system would be an FPGA which is setup from  
>>> user space
>>> after the system has booted.  The FPGA may be access by DSPs in  
>>> the system
>>> and those DSPs expect the FPGA at a fixed PCI address.
>>>
>>> Added pci_assign_resource_fixed() as a way to allow assignment of  
>>> the PCI
>>> devices's BARs at fixed PCI addresses.
>>>
>>> Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
>>>
>>> ---
>>> commit 45d4a23317c459865ec740c80b6e2a2ad9f53fd3
>>> tree 432b5e41ef5f231dd57eb1a98f103239c62d63a0
>>> parent 8176dee014ec6ad1039b8c0075c9c1d02147c2c8
>>> author Kumar Gala <galak@kernel.crashing.org> Thu, 09 Mar 2006  
>>> 12:34:25 -0600
>>> committer Kumar Gala <galak@kernel.crashing.org> Thu, 09 Mar 2006  
>>> 12:34:25 -0600
>>>
>>>  drivers/pci/pci.c       |    1 +
>>>  drivers/pci/setup-res.c |   35 +++++++++++++++++++++++++++++++++++
>>>  include/linux/pci.h     |    1 +
>>>  3 files changed, 37 insertions(+), 0 deletions(-)
>>>
>>> diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
>>> index d2d1879..2557e86 100644
>>> --- a/drivers/pci/pci.c
>>> +++ b/drivers/pci/pci.c
>>> @@ -935,6 +935,7 @@ EXPORT_SYMBOL_GPL(pci_intx);
>>>  EXPORT_SYMBOL(pci_set_dma_mask);
>>>  EXPORT_SYMBOL(pci_set_consistent_dma_mask);
>>>  EXPORT_SYMBOL(pci_assign_resource);
>>> +EXPORT_SYMBOL(pci_assign_resource_fixed);
>>>  EXPORT_SYMBOL(pci_find_parent_resource);
>>>
>>>  EXPORT_SYMBOL(pci_set_power_state);
>>> diff --git a/drivers/pci/setup-res.c b/drivers/pci/setup-res.c
>>> index ea9277b..f485958 100644
>>> --- a/drivers/pci/setup-res.c
>>> +++ b/drivers/pci/setup-res.c
>>> @@ -155,6 +155,41 @@ int pci_assign_resource(struct pci_dev *
>>>  	return ret;
>>>  }
>>>
>>> +int pci_assign_resource_fixed(struct pci_dev *dev, int resno)
>>> +{
>>> +	struct pci_bus *bus = dev->bus;
>>> +	struct resource *res = dev->resource + resno;
>>> +	unsigned int type_mask;
>>> +	int i, ret = -EBUSY;
>>> +
>>> +	type_mask = IORESOURCE_IO | IORESOURCE_MEM | IORESOURCE_PREFETCH;
>>
>> If type_mask must match (as in comment below), should it be a
>> parameter instead of hard-coded here?  Does this match your FPGA
>> resource?  It may not match my <hypothetical> resource.
>
> I was trying to ensure an exact match between the bus and device  
> resource types.  I figured if you were calling this API you should  
> be able to ensure that the resource types match exactly.
>
>>> +	for (i = 0; i < PCI_BUS_NUM_RESOURCES; i++) {
>>> +		struct resource *r = bus->resource[i];
>>> +		if (!r)
>>> +			continue;
>>> +
>>> +		/* type_mask must match */
>>> +		if ((res->flags ^ r->flags) & type_mask)
>>> +			continue;
>>> +
>>> +		ret = request_resource(r, res);
>>> +
>>> +		if (ret == 0)
>>> +			break;
>>> +	}
>>> +
>>> +	if (ret) {
>>> +		printk(KERN_ERR "PCI: Failed to allocate %s resource #%d:%lx@% 
>>> lx for %s\n",
>>> +		       res->flags & IORESOURCE_IO ? "I/O" : "mem",
>>> +		       resno, res->end - res->start + 1, res->start, pci_name 
>>> (dev));
>>> +	} else if (resno < PCI_BRIDGE_RESOURCES) {
>>> +		pci_update_resource(dev, res, resno);
>>> +	}
>>
>> braces not needed.
>
> Fair, need to go find where I stole this from and possibly fix  
> extra braces there as well.
>
>>
>>> +
>>> +	return ret;
>>> +}
>>> +
>>
>>
>> ---
>> ~Randy
>> Please use an email client that implements proper (compliant)  
>> threading.
>> (You know who you are.)
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux- 
> kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/

Re: [PATCH] PCI: Add pci_assign_resource_fixed -- allow fixed address assignments

[Kumar Gala]

Greg,

My weekly maintainer ping :)

- k

On Mar 20, 2006, at 10:09 AM, Kumar Gala wrote:

> Greg,
>
> Any comments on this?
>
> - kumar
>
> On Mar 10, 2006, at 3:30 PM, Kumar Gala wrote:
>
>>
>> On Mar 10, 2006, at 3:04 PM, Randy.Dunlap wrote:
>>
>>> On Fri, 10 Mar 2006 09:06:32 -0600 (CST) Kumar Gala wrote:
>>>
>>>> On some embedded systems the PCI address for hotplug devices are  
>>>> not only
>>>> known a priori but are required to be at a given PCI address for  
>>>> other
>>>> master in the system to be able to access.
>>>>
>>>> An example of such a system would be an FPGA which is setup from  
>>>> user space
>>>> after the system has booted.  The FPGA may be access by DSPs in  
>>>> the system
>>>> and those DSPs expect the FPGA at a fixed PCI address.
>>>>
>>>> Added pci_assign_resource_fixed() as a way to allow assignment  
>>>> of the PCI
>>>> devices's BARs at fixed PCI addresses.
>>>>
>>>> Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
>>>>
>>>> ---
>>>> commit 45d4a23317c459865ec740c80b6e2a2ad9f53fd3
>>>> tree 432b5e41ef5f231dd57eb1a98f103239c62d63a0
>>>> parent 8176dee014ec6ad1039b8c0075c9c1d02147c2c8
>>>> author Kumar Gala <galak@kernel.crashing.org> Thu, 09 Mar 2006  
>>>> 12:34:25 -0600
>>>> committer Kumar Gala <galak@kernel.crashing.org> Thu, 09 Mar  
>>>> 2006 12:34:25 -0600
>>>>
>>>>  drivers/pci/pci.c       |    1 +
>>>>  drivers/pci/setup-res.c |   35 +++++++++++++++++++++++++++++++++++
>>>>  include/linux/pci.h     |    1 +
>>>>  3 files changed, 37 insertions(+), 0 deletions(-)
>>>>
>>>> diff --git a/drivers/pci/pci.c b/drivers/pci/pci.c
>>>> index d2d1879..2557e86 100644
>>>> --- a/drivers/pci/pci.c
>>>> +++ b/drivers/pci/pci.c
>>>> @@ -935,6 +935,7 @@ EXPORT_SYMBOL_GPL(pci_intx);
>>>>  EXPORT_SYMBOL(pci_set_dma_mask);
>>>>  EXPORT_SYMBOL(pci_set_consistent_dma_mask);
>>>>  EXPORT_SYMBOL(pci_assign_resource);
>>>> +EXPORT_SYMBOL(pci_assign_resource_fixed);
>>>>  EXPORT_SYMBOL(pci_find_parent_resource);
>>>>
>>>>  EXPORT_SYMBOL(pci_set_power_state);
>>>> diff --git a/drivers/pci/setup-res.c b/drivers/pci/setup-res.c
>>>> index ea9277b..f485958 100644
>>>> --- a/drivers/pci/setup-res.c
>>>> +++ b/drivers/pci/setup-res.c
>>>> @@ -155,6 +155,41 @@ int pci_assign_resource(struct pci_dev *
>>>>  	return ret;
>>>>  }
>>>>
>>>> +int pci_assign_resource_fixed(struct pci_dev *dev, int resno)
>>>> +{
>>>> +	struct pci_bus *bus = dev->bus;
>>>> +	struct resource *res = dev->resource + resno;
>>>> +	unsigned int type_mask;
>>>> +	int i, ret = -EBUSY;
>>>> +
>>>> +	type_mask = IORESOURCE_IO | IORESOURCE_MEM | IORESOURCE_PREFETCH;
>>>
>>> If type_mask must match (as in comment below), should it be a
>>> parameter instead of hard-coded here?  Does this match your FPGA
>>> resource?  It may not match my <hypothetical> resource.
>>
>> I was trying to ensure an exact match between the bus and device  
>> resource types.  I figured if you were calling this API you should  
>> be able to ensure that the resource types match exactly.
>>
>>>> +	for (i = 0; i < PCI_BUS_NUM_RESOURCES; i++) {
>>>> +		struct resource *r = bus->resource[i];
>>>> +		if (!r)
>>>> +			continue;
>>>> +
>>>> +		/* type_mask must match */
>>>> +		if ((res->flags ^ r->flags) & type_mask)
>>>> +			continue;
>>>> +
>>>> +		ret = request_resource(r, res);
>>>> +
>>>> +		if (ret == 0)
>>>> +			break;
>>>> +	}
>>>> +
>>>> +	if (ret) {
>>>> +		printk(KERN_ERR "PCI: Failed to allocate %s resource #%d:%lx@% 
>>>> lx for %s\n",
>>>> +		       res->flags & IORESOURCE_IO ? "I/O" : "mem",
>>>> +		       resno, res->end - res->start + 1, res->start, pci_name 
>>>> (dev));
>>>> +	} else if (resno < PCI_BRIDGE_RESOURCES) {
>>>> +		pci_update_resource(dev, res, resno);
>>>> +	}
>>>
>>> braces not needed.
>>
>> Fair, need to go find where I stole this from and possibly fix  
>> extra braces there as well.
>>
>>>
>>>> +
>>>> +	return ret;
>>>> +}
>>>> +
>>>
>>>
>>> ---
>>> ~Randy
>>> Please use an email client that implements proper (compliant)  
>>> threading.
>>> (You know who you are.)
>>
>> -
>> To unsubscribe from this list: send the line "unsubscribe linux- 
>> kernel" in
>> the body of a message to majordomo@vger.kernel.org
>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>> Please read the FAQ at  http://www.tux.org/lkml/
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux- 
> kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/