[PATCH, for-4.19] dma-mapping: add the missing ARCH_HAS_SYNC_DMA_FOR_CPU_ALL declaration

[PATCH, for-4.19] dma-mapping: add the missing ARCH_HAS_SYNC_DMA_FOR_CPU_ALL declaration

[Christoph Hellwig]

The patch adding the infrastructure failed to actually add the symbol
declaration, oops..

Fixes: faef87723a ("dma-noncoherent: add a arch_sync_dma_for_cpu_all hook")
Signed-off-by: Christoph Hellwig <hch-jcswGhMUV9g@public.gmane.org>
---
 kernel/dma/Kconfig | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/kernel/dma/Kconfig b/kernel/dma/Kconfig
index 9bd54304446f..1b1d63b3634b 100644
--- a/kernel/dma/Kconfig
+++ b/kernel/dma/Kconfig
@@ -23,6 +23,9 @@ config ARCH_HAS_SYNC_DMA_FOR_CPU
 	bool
 	select NEED_DMA_MAP_STATE
 
+config ARCH_HAS_SYNC_DMA_FOR_CPU_ALL
+	bool
+
 config DMA_DIRECT_OPS
 	bool
 	depends on HAS_DMA
-- 
2.18.0

Re: [PATCH, for-4.19] dma-mapping: add the missing ARCH_HAS_SYNC_DMA_FOR_CPU_ALL declaration

[Christoph Hellwig]

Aby chance to get a review for this?

On Tue, Sep 11, 2018 at 11:00:49AM +0200, Christoph Hellwig wrote:
> The patch adding the infrastructure failed to actually add the symbol
> declaration, oops..
> 
> Fixes: faef87723a ("dma-noncoherent: add a arch_sync_dma_for_cpu_all hook")
> Signed-off-by: Christoph Hellwig <hch-jcswGhMUV9g@public.gmane.org>
> ---
>  kernel/dma/Kconfig | 3 +++
>  1 file changed, 3 insertions(+)
> 
> diff --git a/kernel/dma/Kconfig b/kernel/dma/Kconfig
> index 9bd54304446f..1b1d63b3634b 100644
> --- a/kernel/dma/Kconfig
> +++ b/kernel/dma/Kconfig
> @@ -23,6 +23,9 @@ config ARCH_HAS_SYNC_DMA_FOR_CPU
>  	bool
>  	select NEED_DMA_MAP_STATE
>  
> +config ARCH_HAS_SYNC_DMA_FOR_CPU_ALL
> +	bool
> +
>  config DMA_DIRECT_OPS
>  	bool
>  	depends on HAS_DMA
> -- 
> 2.18.0
> 
> _______________________________________________
> iommu mailing list
> iommu-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA@public.gmane.org
> https://lists.linuxfoundation.org/mailman/listinfo/iommu
---end quoted text---

Re: [PATCH, for-4.19] dma-mapping: add the missing ARCH_HAS_SYNC_DMA_FOR_CPU_ALL declaration

[Robin Murphy]

On 14/09/18 11:08, Christoph Hellwig wrote:
> Aby chance to get a review for this?

So without this, the select does nothing, 
CONFIG_ARCH_HAS_SYNC_DMA_FOR_CPU_ALL is never defined, and BMIPS gets 
the static inline stub and never flushes the RAC when it should? I don't 
know enough MIPS to consider even compile-testing to check it, but that 
sounds like a legitimate fix to me.

Robin.

> On Tue, Sep 11, 2018 at 11:00:49AM +0200, Christoph Hellwig wrote:
>> The patch adding the infrastructure failed to actually add the symbol
>> declaration, oops..
>>
>> Fixes: faef87723a ("dma-noncoherent: add a arch_sync_dma_for_cpu_all hook")
>> Signed-off-by: Christoph Hellwig <hch-jcswGhMUV9g@public.gmane.org>
>> ---
>>   kernel/dma/Kconfig | 3 +++
>>   1 file changed, 3 insertions(+)
>>
>> diff --git a/kernel/dma/Kconfig b/kernel/dma/Kconfig
>> index 9bd54304446f..1b1d63b3634b 100644
>> --- a/kernel/dma/Kconfig
>> +++ b/kernel/dma/Kconfig
>> @@ -23,6 +23,9 @@ config ARCH_HAS_SYNC_DMA_FOR_CPU
>>   	bool
>>   	select NEED_DMA_MAP_STATE
>>   
>> +config ARCH_HAS_SYNC_DMA_FOR_CPU_ALL
>> +	bool
>> +
>>   config DMA_DIRECT_OPS
>>   	bool
>>   	depends on HAS_DMA
>> -- 
>> 2.18.0
>>
>> _______________________________________________
>> iommu mailing list
>> iommu-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA@public.gmane.org
>> https://lists.linuxfoundation.org/mailman/listinfo/iommu
> ---end quoted text---
> 

Re: [PATCH, for-4.19] dma-mapping: add the missing ARCH_HAS_SYNC_DMA_FOR_CPU_ALL declaration

[Christoph Hellwig]

On Fri, Sep 14, 2018 at 04:44:15PM +0100, Robin Murphy wrote:
> On 14/09/18 11:08, Christoph Hellwig wrote:
>> Aby chance to get a review for this?
>
> So without this, the select does nothing, 
> CONFIG_ARCH_HAS_SYNC_DMA_FOR_CPU_ALL is never defined, and BMIPS gets the 
> static inline stub and never flushes the RAC when it should?

Exactly!

Re: [PATCH, for-4.19] dma-mapping: add the missing ARCH_HAS_SYNC_DMA_FOR_CPU_ALL declaration

[Paul Burton]

Hi Christoph,

On Fri, Sep 21, 2018 at 08:10:52AM +0200, Christoph Hellwig wrote:
> Paul, other mips folks, any comments?
> 
> At this point I'm tempted to just move it to 4.20 and add a stable
> tag given that no one cared so far.

Copying Kevin & Florian as maintainers of the affected BMIPS systems in
case they have anything to add.

> On Tue, Sep 11, 2018 at 11:00:49AM +0200, Christoph Hellwig wrote:
> > The patch adding the infrastructure failed to actually add the symbol
> > declaration, oops..
> > 
> > Fixes: faef87723a ("dma-noncoherent: add a arch_sync_dma_for_cpu_all hook")
> > Signed-off-by: Christoph Hellwig <hch-jcswGhMUV9g@public.gmane.org>
> > ---
> >  kernel/dma/Kconfig | 3 +++
> >  1 file changed, 3 insertions(+)
> > 
> > diff --git a/kernel/dma/Kconfig b/kernel/dma/Kconfig
> > index 9bd54304446f..1b1d63b3634b 100644
> > --- a/kernel/dma/Kconfig
> > +++ b/kernel/dma/Kconfig
> > @@ -23,6 +23,9 @@ config ARCH_HAS_SYNC_DMA_FOR_CPU
> >  	bool
> >  	select NEED_DMA_MAP_STATE
> >  
> > +config ARCH_HAS_SYNC_DMA_FOR_CPU_ALL
> > +	bool
> > +
> >  config DMA_DIRECT_OPS
> >  	bool
> >  	depends on HAS_DMA

The change looks reasonable to me, so feel free to take your choice of:

    Reviewed-by: Paul Burton <paul.burton-8NJIiSa5LzA@public.gmane.org>
    Acked-by: Paul Burton <paul.burton-8NJIiSa5LzA@public.gmane.org>

My thought would be that it would be ideal to fix in 4.19 since that's
where the breakage is, but having said that I don't have much insight
into how bad the breakage is for the affected systems.

Thanks,
    Paul

Re: [PATCH, for-4.19] dma-mapping: add the missing ARCH_HAS_SYNC_DMA_FOR_CPU_ALL declaration

[Florian Fainelli]

On September 21, 2018 12:42:20 PM PDT, Paul Burton <paul.burton-8NJIiSa5LzA@public.gmane.org> wrote:
>Hi Christoph,
>
>On Fri, Sep 21, 2018 at 08:10:52AM +0200, Christoph Hellwig wrote:
>> Paul, other mips folks, any comments?
>> 
>> At this point I'm tempted to just move it to 4.20 and add a stable
>> tag given that no one cared so far.
>
>Copying Kevin & Florian as maintainers of the affected BMIPS systems in
>case they have anything to add.
>
>> On Tue, Sep 11, 2018 at 11:00:49AM +0200, Christoph Hellwig wrote:
>> > The patch adding the infrastructure failed to actually add the
>symbol
>> > declaration, oops..
>> > 
>> > Fixes: faef87723a ("dma-noncoherent: add a
>arch_sync_dma_for_cpu_all hook")
>> > Signed-off-by: Christoph Hellwig <hch-jcswGhMUV9g@public.gmane.org>
>> > ---
>> >  kernel/dma/Kconfig | 3 +++
>> >  1 file changed, 3 insertions(+)
>> > 
>> > diff --git a/kernel/dma/Kconfig b/kernel/dma/Kconfig
>> > index 9bd54304446f..1b1d63b3634b 100644
>> > --- a/kernel/dma/Kconfig
>> > +++ b/kernel/dma/Kconfig
>> > @@ -23,6 +23,9 @@ config ARCH_HAS_SYNC_DMA_FOR_CPU
>> >  	bool
>> >  	select NEED_DMA_MAP_STATE
>> >  
>> > +config ARCH_HAS_SYNC_DMA_FOR_CPU_ALL
>> > +	bool
>> > +
>> >  config DMA_DIRECT_OPS
>> >  	bool
>> >  	depends on HAS_DMA
>
>The change looks reasonable to me, so feel free to take your choice of:
>
>    Reviewed-by: Paul Burton <paul.burton-8NJIiSa5LzA@public.gmane.org>
>    Acked-by: Paul Burton <paul.burton-8NJIiSa5LzA@public.gmane.org>
>
>My thought would be that it would be ideal to fix in 4.19 since that's
>where the breakage is, but having said that I don't have much insight
>into how bad the breakage is for the affected systems.


AFAIR, DMA corruption is typically what you would observe, which could prove difficult if you are not exactly looking for it. In any case, I don't have a BMIPS system running 4.19 right now to exercise this on, but this looks correct:

Acked-by: Florian Fainelli <f.fainelli-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
-- 
Florian

Re: [PATCH, for-4.19] dma-mapping: add the missing ARCH_HAS_SYNC_DMA_FOR_CPU_ALL declaration

[Christoph Hellwig]

On Sat, Sep 22, 2018 at 12:24:33PM -0700, Florian Fainelli wrote:
> >My thought would be that it would be ideal to fix in 4.19 since that's
> >where the breakage is, but having said that I don't have much insight
> >into how bad the breakage is for the affected systems.
> 
> 
> AFAIR, DMA corruption is typically what you would observe, which could prove difficult if you are not exactly looking for it. In any case, I don't have a BMIPS system running 4.19 right now to exercise this on, but this looks correct:

Well, I originally intended it for 4.19, but without reviews from
the affected maintainers I'd fell bad about it.  Now that we've got
agreement I'll queue it for 4.19.