Warnings in arch/arm64/mm/dma-mapping.c

Warnings in arch/arm64/mm/dma-mapping.c

[Christoph Hellwig]

On Mon, Oct 29, 2018 at 04:51:47PM -0700, Florian Fainelli wrote:
> Hi,
> 
> I have not a chance to run a bisection yet, but I suspect this is coming
> from Christoph's latest batch of DMA-API changes in ARM64 under the
> merge commit befa93633193e5327e4045d1e5fa29114580fa5d ("Merge tag
> 'dma-mapping-4.20-1' of git://git.infradead.org/users/hch/dma-mapping"),
> here are the warnings:

This should fix it:

---
>From 773c5e097c5cf3868443ffb79ae64d79f70e3ccf Mon Sep 17 00:00:00 2001
From: Christoph Hellwig <hch@lst.de>
Date: Tue, 30 Oct 2018 09:41:29 +0200
Subject: arm64: fix warnings without CONFIG_IOMMU_DMA

__swiotlb_get_sgtable_page and __swiotlb_mmap_pfn are not only misnamed
but also only used if CONFIG_IOMMU_DMA is set.  Just add a simple ifdef
for now, given that we plan to remove them entirely for the next merge
window.

Reported-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
---
 arch/arm64/mm/dma-mapping.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/arch/arm64/mm/dma-mapping.c b/arch/arm64/mm/dma-mapping.c
index d190612b8f33..560756a7290e 100644
--- a/arch/arm64/mm/dma-mapping.c
+++ b/arch/arm64/mm/dma-mapping.c
@@ -160,6 +160,7 @@ void arch_sync_dma_for_cpu(struct device *dev, phys_addr_t paddr,
 	__dma_unmap_area(phys_to_virt(paddr), size, dir);
 }
 
+#ifdef CONFIG_IOMMU_DMA
 static int __swiotlb_get_sgtable_page(struct sg_table *sgt,
 				      struct page *page, size_t size)
 {
@@ -188,6 +189,7 @@ static int __swiotlb_mmap_pfn(struct vm_area_struct *vma,
 
 	return ret;
 }
+#endif /* CONFIG_IOMMU_DMA */
 
 static int __init atomic_pool_init(void)
 {
-- 
2.19.1

Warnings in arch/arm64/mm/dma-mapping.c

[Will Deacon]

On Tue, Oct 30, 2018 at 08:43:44AM +0100, Christoph Hellwig wrote:
> On Mon, Oct 29, 2018 at 04:51:47PM -0700, Florian Fainelli wrote:
> > I have not a chance to run a bisection yet, but I suspect this is coming
> > from Christoph's latest batch of DMA-API changes in ARM64 under the
> > merge commit befa93633193e5327e4045d1e5fa29114580fa5d ("Merge tag
> > 'dma-mapping-4.20-1' of git://git.infradead.org/users/hch/dma-mapping"),
> > here are the warnings:
> 
> This should fix it:

Tested-by: Will Deacon <will.deacon@arm.com>

I guess Catalin can pick this up for -rc1, since there's also the kprobes
RO fix kicking around for arm64?

Will

> ---
> From 773c5e097c5cf3868443ffb79ae64d79f70e3ccf Mon Sep 17 00:00:00 2001
> From: Christoph Hellwig <hch@lst.de>
> Date: Tue, 30 Oct 2018 09:41:29 +0200
> Subject: arm64: fix warnings without CONFIG_IOMMU_DMA
> 
> __swiotlb_get_sgtable_page and __swiotlb_mmap_pfn are not only misnamed
> but also only used if CONFIG_IOMMU_DMA is set.  Just add a simple ifdef
> for now, given that we plan to remove them entirely for the next merge
> window.
> 
> Reported-by: Florian Fainelli <f.fainelli@gmail.com>
> Signed-off-by: Christoph Hellwig <hch@lst.de>
> ---
>  arch/arm64/mm/dma-mapping.c | 2 ++
>  1 file changed, 2 insertions(+)
> 
> diff --git a/arch/arm64/mm/dma-mapping.c b/arch/arm64/mm/dma-mapping.c
> index d190612b8f33..560756a7290e 100644
> --- a/arch/arm64/mm/dma-mapping.c
> +++ b/arch/arm64/mm/dma-mapping.c
> @@ -160,6 +160,7 @@ void arch_sync_dma_for_cpu(struct device *dev, phys_addr_t paddr,
>  	__dma_unmap_area(phys_to_virt(paddr), size, dir);
>  }
>  
> +#ifdef CONFIG_IOMMU_DMA
>  static int __swiotlb_get_sgtable_page(struct sg_table *sgt,
>  				      struct page *page, size_t size)
>  {
> @@ -188,6 +189,7 @@ static int __swiotlb_mmap_pfn(struct vm_area_struct *vma,
>  
>  	return ret;
>  }
> +#endif /* CONFIG_IOMMU_DMA */
>  
>  static int __init atomic_pool_init(void)
>  {
> -- 
> 2.19.1
> 

Warnings in arch/arm64/mm/dma-mapping.c

[Florian Fainelli]

On 10/30/18 3:13 AM, Will Deacon wrote:
> On Tue, Oct 30, 2018 at 08:43:44AM +0100, Christoph Hellwig wrote:
>> On Mon, Oct 29, 2018 at 04:51:47PM -0700, Florian Fainelli wrote:
>>> I have not a chance to run a bisection yet, but I suspect this is coming
>>> from Christoph's latest batch of DMA-API changes in ARM64 under the
>>> merge commit befa93633193e5327e4045d1e5fa29114580fa5d ("Merge tag
>>> 'dma-mapping-4.20-1' of git://git.infradead.org/users/hch/dma-mapping"),
>>> here are the warnings:
>>
>> This should fix it:
> 
> Tested-by: Will Deacon <will.deacon@arm.com>

Tested-by: Florian Fainelli <f.fainelli@gmail.com>

Thanks for the quick fix!

> 
> I guess Catalin can pick this up for -rc1, since there's also the kprobes
> RO fix kicking around for arm64?
> 
> Will
> 
>> ---
>> From 773c5e097c5cf3868443ffb79ae64d79f70e3ccf Mon Sep 17 00:00:00 2001
>> From: Christoph Hellwig <hch@lst.de>
>> Date: Tue, 30 Oct 2018 09:41:29 +0200
>> Subject: arm64: fix warnings without CONFIG_IOMMU_DMA
>>
>> __swiotlb_get_sgtable_page and __swiotlb_mmap_pfn are not only misnamed
>> but also only used if CONFIG_IOMMU_DMA is set.  Just add a simple ifdef
>> for now, given that we plan to remove them entirely for the next merge
>> window.
>>
>> Reported-by: Florian Fainelli <f.fainelli@gmail.com>
>> Signed-off-by: Christoph Hellwig <hch@lst.de>
>> ---
>>  arch/arm64/mm/dma-mapping.c | 2 ++
>>  1 file changed, 2 insertions(+)
>>
>> diff --git a/arch/arm64/mm/dma-mapping.c b/arch/arm64/mm/dma-mapping.c
>> index d190612b8f33..560756a7290e 100644
>> --- a/arch/arm64/mm/dma-mapping.c
>> +++ b/arch/arm64/mm/dma-mapping.c
>> @@ -160,6 +160,7 @@ void arch_sync_dma_for_cpu(struct device *dev, phys_addr_t paddr,
>>  	__dma_unmap_area(phys_to_virt(paddr), size, dir);
>>  }
>>  
>> +#ifdef CONFIG_IOMMU_DMA
>>  static int __swiotlb_get_sgtable_page(struct sg_table *sgt,
>>  				      struct page *page, size_t size)
>>  {
>> @@ -188,6 +189,7 @@ static int __swiotlb_mmap_pfn(struct vm_area_struct *vma,
>>  
>>  	return ret;
>>  }
>> +#endif /* CONFIG_IOMMU_DMA */
>>  
>>  static int __init atomic_pool_init(void)
>>  {
>> -- 
>> 2.19.1
>>


-- 
Florian

Warnings in arch/arm64/mm/dma-mapping.c

[Catalin Marinas]

On Tue, Oct 30, 2018 at 10:13:42AM +0000, Will Deacon wrote:
> On Tue, Oct 30, 2018 at 08:43:44AM +0100, Christoph Hellwig wrote:
> > On Mon, Oct 29, 2018 at 04:51:47PM -0700, Florian Fainelli wrote:
> > > I have not a chance to run a bisection yet, but I suspect this is coming
> > > from Christoph's latest batch of DMA-API changes in ARM64 under the
> > > merge commit befa93633193e5327e4045d1e5fa29114580fa5d ("Merge tag
> > > 'dma-mapping-4.20-1' of git://git.infradead.org/users/hch/dma-mapping"),
> > > here are the warnings:
> > 
> > This should fix it:
> 
> Tested-by: Will Deacon <will.deacon@arm.com>
> 
> I guess Catalin can pick this up for -rc1, since there's also the kprobes
> RO fix kicking around for arm64?

I can pick it up at -rc1 as I don't have Christoph's patches in my tree
(and don't want to base a branch on some random point in the merging
window).

Alternatively, it could go via Christoph's tree for -rc1.

Acked-by: Catalin Marinas <catalin.marinas@arm.com>

Warnings in arch/arm64/mm/dma-mapping.c

[Christoph Hellwig]

On Fri, Nov 02, 2018 at 05:47:24PM +0000, Catalin Marinas wrote:
> On Tue, Oct 30, 2018 at 10:13:42AM +0000, Will Deacon wrote:
> > On Tue, Oct 30, 2018 at 08:43:44AM +0100, Christoph Hellwig wrote:
> > > On Mon, Oct 29, 2018 at 04:51:47PM -0700, Florian Fainelli wrote:
> > > > I have not a chance to run a bisection yet, but I suspect this is coming
> > > > from Christoph's latest batch of DMA-API changes in ARM64 under the
> > > > merge commit befa93633193e5327e4045d1e5fa29114580fa5d ("Merge tag
> > > > 'dma-mapping-4.20-1' of git://git.infradead.org/users/hch/dma-mapping"),
> > > > here are the warnings:
> > > 
> > > This should fix it:
> > 
> > Tested-by: Will Deacon <will.deacon@arm.com>
> > 
> > I guess Catalin can pick this up for -rc1, since there's also the kprobes
> > RO fix kicking around for arm64?
> 
> I can pick it up at -rc1 as I don't have Christoph's patches in my tree
> (and don't want to base a branch on some random point in the merging
> window).
> 
> Alternatively, it could go via Christoph's tree for -rc1.
> 
> Acked-by: Catalin Marinas <catalin.marinas@arm.com>

I'll send it on tomorrow.