1. Patchew
  2. Xen
  3. View series

[PATCH v1] iommu/ipmmu-vmsa: Fix build with HAS_PCI=n

Mykyta Poturai posted 1 patch 2 months, 1 week ago
Patches applied successfully (tree, apply log)
git fetch https://gitlab.com/xen-project/patchew/xen tags/patchew/f6a8095483ba54e1eabeca5e4b08138312fa822b.1755773176.git.mykyta._5Fpoturai@epam.com
There is a newer version of this series
xen/drivers/passthrough/arm/ipmmu-vmsa.c | 2 ++
1 file changed, 2 insertions(+)
[PATCH v1] iommu/ipmmu-vmsa: Fix build with HAS_PCI=n
Posted by Mykyta Poturai 2 months, 1 week ago 
With PCI disabled the build fails due to undefined struct
pci_host_bridge.

Add ifdef guard around pci-host-rcar4.h to not include it when PCI
support is disabled.

Signed-off-by: Mykyta Poturai <mykyta_poturai@epam.com>
---
This patch can be squashed with iommu/ipmmu-vmsa: Implement basic PCIE-IPMMU OSID support
---
 xen/drivers/passthrough/arm/ipmmu-vmsa.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/xen/drivers/passthrough/arm/ipmmu-vmsa.c b/xen/drivers/passthrough/arm/ipmmu-vmsa.c
index ea9fa9ddf3..49f149e222 100644
--- a/xen/drivers/passthrough/arm/ipmmu-vmsa.c
+++ b/xen/drivers/passthrough/arm/ipmmu-vmsa.c
@@ -51,7 +51,9 @@
 #include <asm/device.h>
 #include <asm/io.h>
 #include <asm/iommu_fwspec.h>
+#ifdef CONFIG_HAS_PCI
 #include "../arch/arm/pci/pci-host-rcar4.h"
+#endif
 
 #define dev_name(dev) dt_node_full_name(dev_to_dt(dev))
 
-- 
2.34.1
Re: [PATCH v1] iommu/ipmmu-vmsa: Fix build with HAS_PCI=n
Posted by Mykola Kvach 2 months, 1 week ago 
Hi Mykyta,

On Thu, Aug 21, 2025 at 1:51 PM Mykyta Poturai <Mykyta_Poturai@epam.com> wrote:
>
> With PCI disabled the build fails due to undefined struct
> pci_host_bridge.
>
> Add ifdef guard around pci-host-rcar4.h to not include it when PCI
> support is disabled.
>
> Signed-off-by: Mykyta Poturai <mykyta_poturai@epam.com>
> ---
> This patch can be squashed with iommu/ipmmu-vmsa: Implement basic PCIE-IPMMU OSID support
> ---
>  xen/drivers/passthrough/arm/ipmmu-vmsa.c | 2 ++
>  1 file changed, 2 insertions(+)
>
> diff --git a/xen/drivers/passthrough/arm/ipmmu-vmsa.c b/xen/drivers/passthrough/arm/ipmmu-vmsa.c
> index ea9fa9ddf3..49f149e222 100644
> --- a/xen/drivers/passthrough/arm/ipmmu-vmsa.c
> +++ b/xen/drivers/passthrough/arm/ipmmu-vmsa.c
> @@ -51,7 +51,9 @@
>  #include <asm/device.h>
>  #include <asm/io.h>
>  #include <asm/iommu_fwspec.h>
> +#ifdef CONFIG_HAS_PCI
>  #include "../arch/arm/pci/pci-host-rcar4.h"
> +#endif

As far as I know, the IPMMU VMSA driver can also be used on other
R-Car platforms. Is there a chance we could have HAS_PCI enabled
together with IPMMU VMSA, but not be on an R-Car Gen4 platform,
for example, on R-Car3 or possibly some future R-Car variants?

Wouldn't it be better to use:
#ifdef CONFIG_RCAR4
?

>
>  #define dev_name(dev) dt_node_full_name(dev_to_dt(dev))
>
> --
> 2.34.1
>

Best regards,
Mykola
Re: [PATCH v1] iommu/ipmmu-vmsa: Fix build with HAS_PCI=n
Posted by Mykyta Poturai 2 months, 1 week ago 
On 21.08.25 15:03, Mykola Kvach wrote:
> Hi Mykyta,
> 
> On Thu, Aug 21, 2025 at 1:51 PM Mykyta Poturai <Mykyta_Poturai@epam.com> wrote:
>>
>> With PCI disabled the build fails due to undefined struct
>> pci_host_bridge.
>>
>> Add ifdef guard around pci-host-rcar4.h to not include it when PCI
>> support is disabled.
>>
>> Signed-off-by: Mykyta Poturai <mykyta_poturai@epam.com>
>> ---
>> This patch can be squashed with iommu/ipmmu-vmsa: Implement basic PCIE-IPMMU OSID support
>> ---
>>   xen/drivers/passthrough/arm/ipmmu-vmsa.c | 2 ++
>>   1 file changed, 2 insertions(+)
>>
>> diff --git a/xen/drivers/passthrough/arm/ipmmu-vmsa.c b/xen/drivers/passthrough/arm/ipmmu-vmsa.c
>> index ea9fa9ddf3..49f149e222 100644
>> --- a/xen/drivers/passthrough/arm/ipmmu-vmsa.c
>> +++ b/xen/drivers/passthrough/arm/ipmmu-vmsa.c
>> @@ -51,7 +51,9 @@
>>   #include <asm/device.h>
>>   #include <asm/io.h>
>>   #include <asm/iommu_fwspec.h>
>> +#ifdef CONFIG_HAS_PCI
>>   #include "../arch/arm/pci/pci-host-rcar4.h"
>> +#endif
> 
> As far as I know, the IPMMU VMSA driver can also be used on other
> R-Car platforms. Is there a chance we could have HAS_PCI enabled
> together with IPMMU VMSA, but not be on an R-Car Gen4 platform,
> for example, on R-Car3 or possibly some future R-Car variants?
> 
> Wouldn't it be better to use:
> #ifdef CONFIG_RCAR4
> ?

Gen5 boards will use SMMU, so this will not be applicable to them. In 
any case, checking for RCAR4 will not help here, as PCI can be disabled 
regardless of the platform revision.

> 
> Best regards,
> Mykola

-- 
Mykyta
Re: [PATCH v1] iommu/ipmmu-vmsa: Fix build with HAS_PCI=n
Posted by Mykola Kvach 2 months, 1 week ago 
On Thu, Aug 21, 2025 at 4:09 PM Mykyta Poturai <Mykyta_Poturai@epam.com> wrote:
>
> On 21.08.25 15:03, Mykola Kvach wrote:
> > Hi Mykyta,
> >
> > On Thu, Aug 21, 2025 at 1:51 PM Mykyta Poturai <Mykyta_Poturai@epam.com> wrote:
> >>
> >> With PCI disabled the build fails due to undefined struct
> >> pci_host_bridge.
> >>
> >> Add ifdef guard around pci-host-rcar4.h to not include it when PCI
> >> support is disabled.
> >>
> >> Signed-off-by: Mykyta Poturai <mykyta_poturai@epam.com>
> >> ---
> >> This patch can be squashed with iommu/ipmmu-vmsa: Implement basic PCIE-IPMMU OSID support
> >> ---
> >>   xen/drivers/passthrough/arm/ipmmu-vmsa.c | 2 ++
> >>   1 file changed, 2 insertions(+)
> >>
> >> diff --git a/xen/drivers/passthrough/arm/ipmmu-vmsa.c b/xen/drivers/passthrough/arm/ipmmu-vmsa.c
> >> index ea9fa9ddf3..49f149e222 100644
> >> --- a/xen/drivers/passthrough/arm/ipmmu-vmsa.c
> >> +++ b/xen/drivers/passthrough/arm/ipmmu-vmsa.c
> >> @@ -51,7 +51,9 @@
> >>   #include <asm/device.h>
> >>   #include <asm/io.h>
> >>   #include <asm/iommu_fwspec.h>
> >> +#ifdef CONFIG_HAS_PCI
> >>   #include "../arch/arm/pci/pci-host-rcar4.h"
> >> +#endif
> >
> > As far as I know, the IPMMU VMSA driver can also be used on other
> > R-Car platforms. Is there a chance we could have HAS_PCI enabled
> > together with IPMMU VMSA, but not be on an R-Car Gen4 platform,
> > for example, on R-Car3 or possibly some future R-Car variants?
> >
> > Wouldn't it be better to use:
> > #ifdef CONFIG_RCAR4
> > ?
>
> Gen5 boards will use SMMU, so this will not be applicable to them. In
> any case, checking for RCAR4 will not help here, as PCI can be disabled
> regardless of the platform revision.

I see your point that CONFIG_RCAR4 alone isn’t enough, since PCI can be disabled
even on Gen4. My concern is that we’re adding Gen4-specific code into a generic
IPMMU VMSA driver, which is used on multiple R-Car generations.

>
> >
> > Best regards,
> > Mykola
>
> --
> Mykyta
Re: [PATCH v1] iommu/ipmmu-vmsa: Fix build with HAS_PCI=n
Posted by Jan Beulich 2 months, 1 week ago 
On 21.08.2025 12:50, Mykyta Poturai wrote:
> With PCI disabled the build fails due to undefined struct
> pci_host_bridge.
> 
> Add ifdef guard around pci-host-rcar4.h to not include it when PCI
> support is disabled.
> 
> Signed-off-by: Mykyta Poturai <mykyta_poturai@epam.com>

Oh, and - Reported-by: ... please.

Jan
Re: [PATCH v1] iommu/ipmmu-vmsa: Fix build with HAS_PCI=n
Posted by Jan Beulich 2 months, 1 week ago 
On 21.08.2025 12:50, Mykyta Poturai wrote:
> --- a/xen/drivers/passthrough/arm/ipmmu-vmsa.c
> +++ b/xen/drivers/passthrough/arm/ipmmu-vmsa.c
> @@ -51,7 +51,9 @@
>  #include <asm/device.h>
>  #include <asm/io.h>
>  #include <asm/iommu_fwspec.h>
> +#ifdef CONFIG_HAS_PCI
>  #include "../arch/arm/pci/pci-host-rcar4.h"
> +#endif
>  
>  #define dev_name(dev) dt_node_full_name(dev_to_dt(dev))

Hmm, while this may take care of the immediate issue, I wonder if it wouldn't
better be the header itself which is modified. Then perhaps also taking care
of the other (Misra) issue there: The header guard check not being first in
the file, but sitting after an #include.

I further assume there is a good reason why a private header is included here
from some other subtree.

Jan

Patchew 2.1-devel-b67e4c2

© 2016 - 2025 Red Hat, Inc.