[PATCH] ACPI/IORT: fix build regression without IOMMU

[PATCH] ACPI/IORT: fix build regression without IOMMU

[Arnd Bergmann]

A recent change reintroduced a bug that had previously been
fixed by commit d49f2dedf33b ("ACPI/IORT: Fix CONFIG_IOMMU_API
dependency"):

drivers/acpi/arm64/iort.c: In function 'iort_iommu_configure':
drivers/acpi/arm64/iort.c:829:26: error: 'struct iommu_fwspec' has no member named 'ops'

This does the same change again, replacing another direct
reference to iommu_fwspec->ops with a helper function call.

Fixes: bc8648d49a95 ("ACPI/IORT: Handle PCI aliases properly for IOMMUs")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
---
 drivers/acpi/arm64/iort.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/acpi/arm64/iort.c b/drivers/acpi/arm64/iort.c
index 9cdd49289786..9565d572f8dd 100644
--- a/drivers/acpi/arm64/iort.c
+++ b/drivers/acpi/arm64/iort.c
@@ -826,7 +826,7 @@ const struct iommu_ops *iort_iommu_configure(struct device *dev)
 	 * add_device callback for dev, replay it to get things in order.
 	 */
 	if (!err) {
-		ops = dev->iommu_fwspec->ops;
+		ops = iort_fwspec_iommu_ops(dev->iommu_fwspec);
 		err = iort_add_device_replay(ops, dev);
 	}
 
-- 
2.9.0

Re: [PATCH] ACPI/IORT: fix build regression without IOMMU

[Lorenzo Pieralisi]

On Thu, Aug 10, 2017 at 02:11:00PM +0200, Arnd Bergmann wrote:
> A recent change reintroduced a bug that had previously been
> fixed by commit d49f2dedf33b ("ACPI/IORT: Fix CONFIG_IOMMU_API
> dependency"):
> 
> drivers/acpi/arm64/iort.c: In function 'iort_iommu_configure':
> drivers/acpi/arm64/iort.c:829:26: error: 'struct iommu_fwspec' has no member named 'ops'
> 
> This does the same change again, replacing another direct
> reference to iommu_fwspec->ops with a helper function call.
> 
> Fixes: bc8648d49a95 ("ACPI/IORT: Handle PCI aliases properly for IOMMUs")
> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
> ---
>  drivers/acpi/arm64/iort.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)

Sorry Arnd, thanks, I will send it to Catalin straight away.

Lorenzo

> diff --git a/drivers/acpi/arm64/iort.c b/drivers/acpi/arm64/iort.c
> index 9cdd49289786..9565d572f8dd 100644
> --- a/drivers/acpi/arm64/iort.c
> +++ b/drivers/acpi/arm64/iort.c
> @@ -826,7 +826,7 @@ const struct iommu_ops *iort_iommu_configure(struct device *dev)
>  	 * add_device callback for dev, replay it to get things in order.
>  	 */
>  	if (!err) {
> -		ops = dev->iommu_fwspec->ops;
> +		ops = iort_fwspec_iommu_ops(dev->iommu_fwspec);
>  		err = iort_add_device_replay(ops, dev);
>  	}
>  
> -- 
> 2.9.0
> 

Re: [PATCH] ACPI/IORT: fix build regression without IOMMU

[Thorsten Leemhuis]

On 10.08.2017 16:24, Lorenzo Pieralisi wrote:
> On Thu, Aug 10, 2017 at 02:11:00PM +0200, Arnd Bergmann wrote:
>> A recent change reintroduced a bug that had previously been
>> fixed by commit d49f2dedf33b ("ACPI/IORT: Fix CONFIG_IOMMU_API
>> dependency"):
>>
>> drivers/acpi/arm64/iort.c: In function 'iort_iommu_configure':
>> drivers/acpi/arm64/iort.c:829:26: error: 'struct iommu_fwspec' has no member named 'ops'
>>
>> This does the same change again, replacing another direct
>> reference to iommu_fwspec->ops with a helper function call.
>>
>> Fixes: bc8648d49a95 ("ACPI/IORT: Handle PCI aliases properly for IOMMUs")
>> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
>> ---
>>  drivers/acpi/arm64/iort.c | 2 +-
>>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> Sorry Arnd, thanks, I will send it to Catalin straight away.

Was there any progress? Afaics that patch hasn't hit mainline yet (or
was it fixed by a different patch?). Asking because the issue is
mentioned in the regression report for 4.13.

Ciao, Thorsten

Re: [PATCH] ACPI/IORT: fix build regression without IOMMU

[Lorenzo Pieralisi]

On Mon, Aug 28, 2017 at 11:27:20AM +0200, Thorsten Leemhuis wrote:
> On 10.08.2017 16:24, Lorenzo Pieralisi wrote:
> > On Thu, Aug 10, 2017 at 02:11:00PM +0200, Arnd Bergmann wrote:
> >> A recent change reintroduced a bug that had previously been
> >> fixed by commit d49f2dedf33b ("ACPI/IORT: Fix CONFIG_IOMMU_API
> >> dependency"):
> >>
> >> drivers/acpi/arm64/iort.c: In function 'iort_iommu_configure':
> >> drivers/acpi/arm64/iort.c:829:26: error: 'struct iommu_fwspec' has no member named 'ops'
> >>
> >> This does the same change again, replacing another direct
> >> reference to iommu_fwspec->ops with a helper function call.
> >>
> >> Fixes: bc8648d49a95 ("ACPI/IORT: Handle PCI aliases properly for IOMMUs")
> >> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
> >> ---
> >>  drivers/acpi/arm64/iort.c | 2 +-
> >>  1 file changed, 1 insertion(+), 1 deletion(-)
> > 
> > Sorry Arnd, thanks, I will send it to Catalin straight away.
> 
> Was there any progress? Afaics that patch hasn't hit mainline yet (or
> was it fixed by a different patch?). Asking because the issue is
> mentioned in the regression report for 4.13.

I do not understand how it can be mentioned in the 4.13 regression
report given that it is code queued for 4.14 and it is not in the
mainline that triggered the regression. Anyway, fix is queued along
with other patches and it is in -next (coming from arm64 tree):

commit: 4d36037a9a07

Does this answer your question ?

Lorenzo

Re: [PATCH] ACPI/IORT: fix build regression without IOMMU

[Thorsten Leemhuis]

Lo! On 28.08.2017 11:57, Lorenzo Pieralisiwrote:
> On Mon, Aug 28, 2017 at 11:27:20AM +0200, Thorsten Leemhuis wrote:
>
>> Was there any progress? Afaics that patch hasn't hit mainline yet (or
>> was it fixed by a different patch?). Asking because the issue is
>> mentioned in the regression report for 4.13.
> I do not understand how it can be mentioned in the 4.13 regression
> report given that it is code queued for 4.14 […]

Ohh, there is a simple explanation: Because I'm stupid and missed that
the commit Arnd mentioned was not in mainline :-/ I could have sworn I
checked that, but seems I did not.

Sorry for the noise & many thx for the answer!

Ciao, Thorsten