From: Stewart Hildebrand <stewart.hildebrand@amd.com>
Enable the use of IOMMU + PCI in dom0 without having to specify
"pci-passthrough=yes". We rely on dom0 to initialize the PCI controller
and perform a PHYSDEVOP_pci_device_add call to add each device to SMMU.
Enable pci_init() for initializing Xen's internal PCI subsystem, and
allow PHYSDEVOP_pci_device_add when pci-passthrough is disabled.
Signed-off-by: Stewart Hildebrand <stewart.hildebrand@amd.com>
Signed-off-by: Mykyta Poturai <mykyta_poturai@epam.com>
---
hmm. Since
dec9e02f3190 ("xen: avoid generation of stub <asm/pci.h> header")
Should we also move is_pci_passthrough_enabled() back to xen/arch/arm/include/asm/pci.h ?
Not sure if PPC/RISC-V will plan on using this check.
v7->v8:
* bring back x86 definition of is_pci_passthrough_enabled()
v6->v7:
* remove x86 definition of is_pci_passthrough_enabled()
* update comments
* make pci_physdev_op checks stricter
v5->v6:
* new patch - this effectively replaces
("Revert "xen/arm: Add cmdline boot option "pci-passthrough = <boolean>""")
---
xen/arch/arm/pci/pci.c | 5 +++--
xen/drivers/pci/physdev.c | 4 ++--
2 files changed, 5 insertions(+), 4 deletions(-)
diff --git a/xen/arch/arm/pci/pci.c b/xen/arch/arm/pci/pci.c
index 78b97beaef..f2281e81aa 100644
--- a/xen/arch/arm/pci/pci.c
+++ b/xen/arch/arm/pci/pci.c
@@ -16,6 +16,7 @@
#include <xen/device_tree.h>
#include <xen/errno.h>
#include <xen/init.h>
+#include <xen/iommu.h>
#include <xen/param.h>
#include <xen/pci.h>
@@ -83,9 +84,9 @@ static int __init pci_init(void)
{
/*
* Enable PCI passthrough when has been enabled explicitly
- * (pci-passthrough=on).
+ * (pci-passthrough=on) or IOMMU is present and enabled.
*/
- if ( !pci_passthrough_enabled )
+ if ( !is_pci_passthrough_enabled() && !iommu_enabled )
return 0;
pci_segments_init();
diff --git a/xen/drivers/pci/physdev.c b/xen/drivers/pci/physdev.c
index 0161a85e1e..d8a49cadf3 100644
--- a/xen/drivers/pci/physdev.c
+++ b/xen/drivers/pci/physdev.c
@@ -19,7 +19,7 @@ ret_t pci_physdev_op(int cmd, XEN_GUEST_HANDLE_PARAM(void) arg)
struct pci_dev_info pdev_info;
nodeid_t node = NUMA_NO_NODE;
- if ( !is_pci_passthrough_enabled() )
+ if ( !is_pci_passthrough_enabled() && !iommu_enabled )
return -EOPNOTSUPP;
ret = -EFAULT;
@@ -57,7 +57,7 @@ ret_t pci_physdev_op(int cmd, XEN_GUEST_HANDLE_PARAM(void) arg)
case PHYSDEVOP_pci_device_remove: {
struct physdev_pci_device dev;
- if ( !is_pci_passthrough_enabled() )
+ if ( !is_pci_passthrough_enabled() && !iommu_enabled )
return -EOPNOTSUPP;
ret = -EFAULT;
--
2.34.1On 10.02.2025 11:30, Mykyta Poturai wrote:
> From: Stewart Hildebrand <stewart.hildebrand@amd.com>
>
> Enable the use of IOMMU + PCI in dom0 without having to specify
> "pci-passthrough=yes".
Why? It _is_ passing through, even if Dom0 is special.
> @@ -83,9 +84,9 @@ static int __init pci_init(void)
> {
> /*
> * Enable PCI passthrough when has been enabled explicitly
> - * (pci-passthrough=on).
> + * (pci-passthrough=on) or IOMMU is present and enabled.
> */
> - if ( !pci_passthrough_enabled )
> + if ( !is_pci_passthrough_enabled() && !iommu_enabled )
> return 0;
I can't reasonably judge on this adjustment, but ...
> --- a/xen/drivers/pci/physdev.c
> +++ b/xen/drivers/pci/physdev.c
> @@ -19,7 +19,7 @@ ret_t pci_physdev_op(int cmd, XEN_GUEST_HANDLE_PARAM(void) arg)
> struct pci_dev_info pdev_info;
> nodeid_t node = NUMA_NO_NODE;
>
> - if ( !is_pci_passthrough_enabled() )
> + if ( !is_pci_passthrough_enabled() && !iommu_enabled )
> return -EOPNOTSUPP;
>
> ret = -EFAULT;
> @@ -57,7 +57,7 @@ ret_t pci_physdev_op(int cmd, XEN_GUEST_HANDLE_PARAM(void) arg)
> case PHYSDEVOP_pci_device_remove: {
> struct physdev_pci_device dev;
>
> - if ( !is_pci_passthrough_enabled() )
> + if ( !is_pci_passthrough_enabled() && !iommu_enabled )
> return -EOPNOTSUPP;
>
> ret = -EFAULT;
... these two certainly look wrong to be made. If an Arm-specific adjustment
is needed (and can be justified), a per-arch hook may need introducing.
Jan
On 10.02.25 12:56, Jan Beulich wrote:
> On 10.02.2025 11:30, Mykyta Poturai wrote:
>> From: Stewart Hildebrand <stewart.hildebrand@amd.com>
>>
>> Enable the use of IOMMU + PCI in dom0 without having to specify
>> "pci-passthrough=yes".
>
> Why? It _is_ passing through, even if Dom0 is special.
Do you think it would be better to drop this completely and require
pci-passthrough=yes for PCI to work in Dom0?
>
>> @@ -83,9 +84,9 @@ static int __init pci_init(void)
>> {
>> /*
>> * Enable PCI passthrough when has been enabled explicitly
>> - * (pci-passthrough=on).
>> + * (pci-passthrough=on) or IOMMU is present and enabled.
>> */
>> - if ( !pci_passthrough_enabled )
>> + if ( !is_pci_passthrough_enabled() && !iommu_enabled )
>> return 0;
>
> I can't reasonably judge on this adjustment, but ...
>
>> --- a/xen/drivers/pci/physdev.c
>> +++ b/xen/drivers/pci/physdev.c
>> @@ -19,7 +19,7 @@ ret_t pci_physdev_op(int cmd, XEN_GUEST_HANDLE_PARAM(void) arg)
>> struct pci_dev_info pdev_info;
>> nodeid_t node = NUMA_NO_NODE;
>>
>> - if ( !is_pci_passthrough_enabled() )
>> + if ( !is_pci_passthrough_enabled() && !iommu_enabled )
>> return -EOPNOTSUPP;
>>
>> ret = -EFAULT;
>> @@ -57,7 +57,7 @@ ret_t pci_physdev_op(int cmd, XEN_GUEST_HANDLE_PARAM(void) arg)
>> case PHYSDEVOP_pci_device_remove: {
>> struct physdev_pci_device dev;
>>
>> - if ( !is_pci_passthrough_enabled() )
>> + if ( !is_pci_passthrough_enabled() && !iommu_enabled )
>> return -EOPNOTSUPP;
>>
>> ret = -EFAULT;
>
> ... these two certainly look wrong to be made. If an Arm-specific adjustment
> is needed (and can be justified), a per-arch hook may need introducing.
This should not affect x86 in any way if I'm not missing something, as
!is_pci_passthrough_enabled() will always be false. Or are you concerned
about something else?
>
> Jan
--
Mykyta
On 10.02.2025 12:28, Mykyta Poturai wrote:
> On 10.02.25 12:56, Jan Beulich wrote:
>> On 10.02.2025 11:30, Mykyta Poturai wrote:
>>> From: Stewart Hildebrand <stewart.hildebrand@amd.com>
>>>
>>> Enable the use of IOMMU + PCI in dom0 without having to specify
>>> "pci-passthrough=yes".
>>
>> Why? It _is_ passing through, even if Dom0 is special.
>
> Do you think it would be better to drop this completely and require
> pci-passthrough=yes for PCI to work in Dom0?
From an abstract perspective: Yes. I don't know any of the Arm background,
though.
>>> @@ -83,9 +84,9 @@ static int __init pci_init(void)
>>> {
>>> /*
>>> * Enable PCI passthrough when has been enabled explicitly
>>> - * (pci-passthrough=on).
>>> + * (pci-passthrough=on) or IOMMU is present and enabled.
>>> */
>>> - if ( !pci_passthrough_enabled )
>>> + if ( !is_pci_passthrough_enabled() && !iommu_enabled )
>>> return 0;
>>
>> I can't reasonably judge on this adjustment, but ...
>>
>>> --- a/xen/drivers/pci/physdev.c
>>> +++ b/xen/drivers/pci/physdev.c
>>> @@ -19,7 +19,7 @@ ret_t pci_physdev_op(int cmd, XEN_GUEST_HANDLE_PARAM(void) arg)
>>> struct pci_dev_info pdev_info;
>>> nodeid_t node = NUMA_NO_NODE;
>>>
>>> - if ( !is_pci_passthrough_enabled() )
>>> + if ( !is_pci_passthrough_enabled() && !iommu_enabled )
>>> return -EOPNOTSUPP;
>>>
>>> ret = -EFAULT;
>>> @@ -57,7 +57,7 @@ ret_t pci_physdev_op(int cmd, XEN_GUEST_HANDLE_PARAM(void) arg)
>>> case PHYSDEVOP_pci_device_remove: {
>>> struct physdev_pci_device dev;
>>>
>>> - if ( !is_pci_passthrough_enabled() )
>>> + if ( !is_pci_passthrough_enabled() && !iommu_enabled )
>>> return -EOPNOTSUPP;
>>>
>>> ret = -EFAULT;
>>
>> ... these two certainly look wrong to be made. If an Arm-specific adjustment
>> is needed (and can be justified), a per-arch hook may need introducing.
>
> This should not affect x86 in any way if I'm not missing something, as
> !is_pci_passthrough_enabled() will always be false. Or are you concerned
> about something else?
Indeed I am - the wrong look of it. Readers like me will have the immediate
impression of there being something fishy here.
Jan
© 2016 - 2025 Red Hat, Inc.