The GAIT and all of its entries must be represented by physical
addresses as this structure is shared with underlying firmware.
We can keep a virtual address of the GAIT origin in order to
handle processing in the kernel, but when traversing the entries
we must again convert the physical AISB stored in that GAIT entry
into a virtual address in order to process it.
Note: this currently doesn't fix a real bug, since virtual addresses
are indentical to physical ones.
Reviewed-by: Pierre Morel <[email protected]>
Acked-by: Nico Boehr <[email protected]>
Signed-off-by: Matthew Rosato <[email protected]>
---
arch/s390/kvm/interrupt.c | 2 +-
arch/s390/kvm/pci.c | 2 +-
2 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/arch/s390/kvm/interrupt.c b/arch/s390/kvm/interrupt.c
index b9c944b262c7..ab569faf0df2 100644
--- a/arch/s390/kvm/interrupt.c
+++ b/arch/s390/kvm/interrupt.c
@@ -3324,7 +3324,7 @@ static void aen_host_forward(unsigned long si)
if (gaite->count == 0)
return;
if (gaite->aisb != 0)
- set_bit_inv(gaite->aisbo, (unsigned long *)gaite->aisb);
+ set_bit_inv(gaite->aisbo, phys_to_virt(gaite->aisb));
kvm = kvm_s390_pci_si_to_kvm(aift, si);
if (!kvm)
diff --git a/arch/s390/kvm/pci.c b/arch/s390/kvm/pci.c
index bb8c335d17b9..8cfa0b03ebbb 100644
--- a/arch/s390/kvm/pci.c
+++ b/arch/s390/kvm/pci.c
@@ -71,7 +71,7 @@ static int zpci_setup_aipb(u8 nisc)
rc = -ENOMEM;
goto free_sbv;
}
- aift->gait = (struct zpci_gaite *)page_to_phys(page);
+ aift->gait = (struct zpci_gaite *)page_to_virt(page);
zpci_aipb->aipb.faisb = virt_to_phys(aift->sbv->vector);
zpci_aipb->aipb.gait = virt_to_phys(aift->gait);
--
2.37.3
On Wed, 7 Sep 2022 11:59:52 -0400
Matthew Rosato <[email protected]> wrote:
> The GAIT and all of its entries must be represented by physical
> addresses as this structure is shared with underlying firmware.
> We can keep a virtual address of the GAIT origin in order to
> handle processing in the kernel, but when traversing the entries
> we must again convert the physical AISB stored in that GAIT entry
> into a virtual address in order to process it.
>
> Note: this currently doesn't fix a real bug, since virtual addresses
> are indentical to physical ones.
>
> Reviewed-by: Pierre Morel <[email protected]>
> Acked-by: Nico Boehr <[email protected]>
> Signed-off-by: Matthew Rosato <[email protected]>
Reviewed-by: Claudio Imbrenda <[email protected]>
> ---
> arch/s390/kvm/interrupt.c | 2 +-
> arch/s390/kvm/pci.c | 2 +-
> 2 files changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/arch/s390/kvm/interrupt.c b/arch/s390/kvm/interrupt.c
> index b9c944b262c7..ab569faf0df2 100644
> --- a/arch/s390/kvm/interrupt.c
> +++ b/arch/s390/kvm/interrupt.c
> @@ -3324,7 +3324,7 @@ static void aen_host_forward(unsigned long si)
> if (gaite->count == 0)
> return;
> if (gaite->aisb != 0)
> - set_bit_inv(gaite->aisbo, (unsigned long *)gaite->aisb);
> + set_bit_inv(gaite->aisbo, phys_to_virt(gaite->aisb));
>
> kvm = kvm_s390_pci_si_to_kvm(aift, si);
> if (!kvm)
> diff --git a/arch/s390/kvm/pci.c b/arch/s390/kvm/pci.c
> index bb8c335d17b9..8cfa0b03ebbb 100644
> --- a/arch/s390/kvm/pci.c
> +++ b/arch/s390/kvm/pci.c
> @@ -71,7 +71,7 @@ static int zpci_setup_aipb(u8 nisc)
> rc = -ENOMEM;
> goto free_sbv;
> }
> - aift->gait = (struct zpci_gaite *)page_to_phys(page);
> + aift->gait = (struct zpci_gaite *)page_to_virt(page);
>
> zpci_aipb->aipb.faisb = virt_to_phys(aift->sbv->vector);
> zpci_aipb->aipb.gait = virt_to_phys(aift->gait);
On 9/7/22 17:59, Matthew Rosato wrote:
> The GAIT and all of its entries must be represented by physical
> addresses as this structure is shared with underlying firmware.
> We can keep a virtual address of the GAIT origin in order to
> handle processing in the kernel, but when traversing the entries
> we must again convert the physical AISB stored in that GAIT entry
> into a virtual address in order to process it.
>
> Note: this currently doesn't fix a real bug, since virtual addresses
> are indentical to physical ones.
>
> Reviewed-by: Pierre Morel <[email protected]>
> Acked-by: Nico Boehr <[email protected]>
> Signed-off-by: Matthew Rosato <[email protected]>
Thanks, applied