Re: [PATCH v2] arm64/mm: Correct the cache line size warning with non coherent device

[Catalin Marinas]

On Sat, Jun 15, 2019 at 10:44:33AM +0800, Zhangshaokun wrote:
> On 2019/6/14 21:11, Masayoshi Mizuma wrote:
> > diff --git a/arch/arm64/kernel/cacheinfo.c b/arch/arm64/kernel/cacheinfo.c
> > index 6eaf1c07aa4e..7fa6828bb488 100644
> > --- a/arch/arm64/kernel/cacheinfo.c
> > +++ b/arch/arm64/kernel/cacheinfo.c
> > @@ -19,12 +19,10 @@
> >  
> >  int cache_line_size(void)
> >  {
> > -	u32 cwg = cache_type_cwg();
> > -
> >  	if (coherency_max_size != 0)
> >  		return coherency_max_size;
> >  
> > -	return cwg ? 4 << cwg : ARCH_DMA_MINALIGN;
> > +	return cache_line_size_of_cpu();
> >  }
> 
> How about simplify it as this?
> 
> int cache_line_size(void)
> {
>         return coherency_max_size ? coherency_max_size :
>                 cache_line_size_of_cpu();
> }

I don't see this as a simplification, easier to read with explicit 'if'.

> >  EXPORT_SYMBOL_GPL(cache_line_size);
> >  
> > diff --git a/arch/arm64/mm/dma-mapping.c b/arch/arm64/mm/dma-mapping.c
> > index 1669618db08a..379589dc7113 100644
> > --- a/arch/arm64/mm/dma-mapping.c
> > +++ b/arch/arm64/mm/dma-mapping.c
> > @@ -38,10 +38,6 @@ void arch_dma_prep_coherent(struct page *page, size_t size)
> >  
> >  static int __init arm64_dma_init(void)
> >  {
> > -	WARN_TAINT(ARCH_DMA_MINALIGN < cache_line_size(),
> > -		   TAINT_CPU_OUT_OF_SPEC,
> > -		   "ARCH_DMA_MINALIGN smaller than CTR_EL0.CWG (%d < %d)",
> > -		   ARCH_DMA_MINALIGN, cache_line_size());
> >  	return dma_atomic_pool_init(GFP_DMA32, __pgprot(PROT_NORMAL_NC));
> >  }
> >  arch_initcall(arm64_dma_init);
> > @@ -56,7 +52,17 @@ void arch_teardown_dma_ops(struct device *dev)
> >  void arch_setup_dma_ops(struct device *dev, u64 dma_base, u64 size,
> >  			const struct iommu_ops *iommu, bool coherent)
> >  {
> > +	int cls = cache_line_size_of_cpu();
> 
> whether we need this local variable, how about use cache_line_size_of_cpu
> directly in WARN_TAINT just like before.

The reason being?

Anyway, I'll queue v2 of this patch as is for 5.3. Thanks.

-- 
Catalin