From: Roland Dreier on
> +#ifdef CONFIG_PCI_IOV
> + if (dev->is_virtfn)
> + dev = dev->physfn;
> +#endif

Seems we would prefer to avoid this ifdef... if we had a function like

struct pci_dev *pci_physfn(struct pci_dev *virtfn)

as an inline in <linux/pci.h> that just returns NULL if PCI_IOV is not
enabled, then we could write this as

if (dev->is_virtfn)
dev = pci_physfn(dev)

without any #ifdef.
--
Roland Dreier <rolandd(a)cisco.com> || For corporate legal information go to:
http://www.cisco.com/web/about/doing_business/legal/cri/index.html
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
From: Roland Dreier on
Looks good to me.
--
Roland Dreier <rolandd(a)cisco.com> || For corporate legal information go to:
http://www.cisco.com/web/about/doing_business/legal/cri/index.html
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
From: Yinghai on
please check

Subject: [PATCH] pci/dmar/sriov: use physfn to search drhd for VF

When virtfn is used, we should use physfn to find correct drhd

-v2: add pci_physfn() Suggested by Roland Dreier <rdreier(a)cisco.com>
do can remove ifdef in dmar.c

Signed-off-by: Yinghai Lu <yinghai(a)kernel.org>

---
drivers/pci/dmar.c | 2 ++
include/linux/pci.h | 10 ++++++++++
2 files changed, 12 insertions(+)

Index: linux-2.6/drivers/pci/dmar.c
===================================================================
--- linux-2.6.orig/drivers/pci/dmar.c
+++ linux-2.6/drivers/pci/dmar.c
@@ -534,6 +534,8 @@ dmar_find_matched_drhd_unit(struct pci_d
struct dmar_drhd_unit *dmaru = NULL;
struct acpi_dmar_hardware_unit *drhd;

+ dev = pci_physfn(dev);
+
list_for_each_entry(dmaru, &dmar_drhd_units, list) {
drhd = container_of(dmaru->hdr,
struct acpi_dmar_hardware_unit,
Index: linux-2.6/include/linux/pci.h
===================================================================
--- linux-2.6.orig/include/linux/pci.h
+++ linux-2.6/include/linux/pci.h
@@ -334,6 +334,16 @@ struct pci_dev {
#endif
};

+static inline struct pci_dev *pci_physfn(struct pci_dev *dev)
+{
+#ifdef CONFIG_PCI_IOV
+ if (dev->is_virtfn)
+ dev = dev->physfn;
+#endif
+
+ return dev;
+}
+
extern struct pci_dev *alloc_pci_dev(void);

#define pci_dev_b(n) list_entry(n, struct pci_dev, bus_list)
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
From: Chris Wright on
* Yinghai (yinghai.lu(a)oracle.com) wrote:
> --- linux-2.6.orig/drivers/pci/dmar.c
> +++ linux-2.6/drivers/pci/dmar.c
> @@ -534,6 +534,8 @@ dmar_find_matched_drhd_unit(struct pci_d
> struct dmar_drhd_unit *dmaru = NULL;
> struct acpi_dmar_hardware_unit *drhd;
>
> + dev = pci_physfn(dev);
> +

Yeah, we typically don't have enough VF's to wrap bus numbers, or we're
under a catchall IOMMU. In the catchall case both vf->bus and vf->pf->bus
will have the same domain (segment) regardless of whether we have large
VF count or big offset/stride. So, I suppose this could be done inside
of dmar_pci_device_match().

Otherwise, I think you'd want to add the same thing to
dmar_find_matched_atsr_unit() since it's the same device scopes there.

thanks,
-chris
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
From: Yinghai on
On 04/08/2010 04:24 PM, Chris Wright wrote:
> * Yinghai (yinghai.lu(a)oracle.com) wrote:
>> --- linux-2.6.orig/drivers/pci/dmar.c
>> +++ linux-2.6/drivers/pci/dmar.c
>> @@ -534,6 +534,8 @@ dmar_find_matched_drhd_unit(struct pci_d
>> struct dmar_drhd_unit *dmaru = NULL;
>> struct acpi_dmar_hardware_unit *drhd;
>>
>> + dev = pci_physfn(dev);
>> +
>
> Yeah, we typically don't have enough VF's to wrap bus numbers, or we're
> under a catchall IOMMU. In the catchall case both vf->bus and vf->pf->bus
> will have the same domain (segment) regardless of whether we have large
> VF count or big offset/stride. So, I suppose this could be done inside
> of dmar_pci_device_match().
>
> Otherwise, I think you'd want to add the same thing to
> dmar_find_matched_atsr_unit() since it's the same device scopes there.

please check

Subject: [PATCH] pci/dmar/sriov: use physfn to search drhd for VF

When virtfn is used, we should use physfn to find correct drhd

-v2: add pci_physfn() Suggested by Roland Dreier <rdreier(a)cisco.com>
do can remove ifdef in dmar.c
-v3: Chris pointed out we need that for dma_find_matched_atsr_unit too
also change dmar_pci_device_match() static

Signed-off-by: Yinghai Lu <yinghai(a)kernel.org>
Acked-by: Roland Dreier <rdreier(a)cisco.com>

---
drivers/pci/dmar.c | 6 +++++-
include/linux/pci.h | 10 ++++++++++
2 files changed, 15 insertions(+), 1 deletion(-)

Index: linux-2.6/drivers/pci/dmar.c
===================================================================
--- linux-2.6.orig/drivers/pci/dmar.c
+++ linux-2.6/drivers/pci/dmar.c
@@ -313,6 +313,8 @@ int dmar_find_matched_atsr_unit(struct p
struct acpi_dmar_atsr *atsr;
struct dmar_atsr_unit *atsru;

+ dev = pci_physfn(dev);
+
list_for_each_entry(atsru, &dmar_atsr_units, list) {
atsr = container_of(atsru->hdr, struct acpi_dmar_atsr, header);
if (atsr->segment == pci_domain_nr(dev->bus))
@@ -511,7 +513,7 @@ parse_dmar_table(void)
return ret;
}

-int dmar_pci_device_match(struct pci_dev *devices[], int cnt,
+static int dmar_pci_device_match(struct pci_dev *devices[], int cnt,
struct pci_dev *dev)
{
int index;
@@ -534,6 +536,8 @@ dmar_find_matched_drhd_unit(struct pci_d
struct dmar_drhd_unit *dmaru = NULL;
struct acpi_dmar_hardware_unit *drhd;

+ dev = pci_physfn(dev);
+
list_for_each_entry(dmaru, &dmar_drhd_units, list) {
drhd = container_of(dmaru->hdr,
struct acpi_dmar_hardware_unit,
Index: linux-2.6/include/linux/pci.h
===================================================================
--- linux-2.6.orig/include/linux/pci.h
+++ linux-2.6/include/linux/pci.h
@@ -334,6 +334,16 @@ struct pci_dev {
#endif
};

+static inline struct pci_dev *pci_physfn(struct pci_dev *dev)
+{
+#ifdef CONFIG_PCI_IOV
+ if (dev->is_virtfn)
+ dev = dev->physfn;
+#endif
+
+ return dev;
+}
+
extern struct pci_dev *alloc_pci_dev(void);

#define pci_dev_b(n) list_entry(n, struct pci_dev, bus_list)
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/