[PATCH] dma/direct: turn ARCH_ZONE_DMA_BITS into a variable

[PATCH] dma/direct: turn ARCH_ZONE_DMA_BITS into a variable

[Nicolas Saenz Julienne]

Some architectures, notably ARM, are interested in tweaking this
depending on their runtime DMA addressing limitations.

Signed-off-by: Nicolas Saenz Julienne <nsaenzjulienne@suse.de>
---

Changes since RFC:
 - Rebased to v5.4-rc6, fixed arm64 code.

NOTE: This will only apply to linux-next, where 
 arch/arm64/mm/init.c            |  5 +++++
 arch/powerpc/include/asm/page.h |  9 ---------
 arch/powerpc/mm/mem.c           | 20 +++++++++++++++-----
 arch/s390/include/asm/page.h    |  2 --
 arch/s390/mm/init.c             |  1 +
 include/linux/dma-direct.h      |  2 ++
 kernel/dma/direct.c             | 13 ++++++-------
 7 files changed, 29 insertions(+), 23 deletions(-)

diff --git a/arch/arm64/mm/init.c b/arch/arm64/mm/init.c
index 45c00a54909c..f716ea634804 100644
--- a/arch/arm64/mm/init.c
+++ b/arch/arm64/mm/init.c
@@ -20,6 +20,7 @@
 #include <linux/sort.h>
 #include <linux/of.h>
 #include <linux/of_fdt.h>
+#include <linux/dma-direct.h>
 #include <linux/dma-mapping.h>
 #include <linux/dma-contiguous.h>
 #include <linux/efi.h>
@@ -41,6 +42,8 @@
 #include <asm/tlb.h>
 #include <asm/alternative.h>
 
+#define ARM64_ZONE_DMA_BITS	30
+
 /*
  * We need to be able to catch inadvertent references to memstart_addr
  * that occur (potentially in generic code) before arm64_memblock_init()
@@ -424,6 +427,8 @@ void __init arm64_memblock_init(void)
 	else
 		arm64_dma_phys_limit = PHYS_MASK + 1;
 
+	zone_dma_bits = ARM64_ZONE_DMA_BITS;
+
 	reserve_crashkernel();
 
 	reserve_elfcorehdr();
diff --git a/arch/powerpc/include/asm/page.h b/arch/powerpc/include/asm/page.h
index c8bb14ff4713..f6c562acc3f8 100644
--- a/arch/powerpc/include/asm/page.h
+++ b/arch/powerpc/include/asm/page.h
@@ -329,13 +329,4 @@ struct vm_area_struct;
 #endif /* __ASSEMBLY__ */
 #include <asm/slice.h>
 
-/*
- * Allow 30-bit DMA for very limited Broadcom wifi chips on many powerbooks.
- */
-#ifdef CONFIG_PPC32
-#define ARCH_ZONE_DMA_BITS 30
-#else
-#define ARCH_ZONE_DMA_BITS 31
-#endif
-
 #endif /* _ASM_POWERPC_PAGE_H */
diff --git a/arch/powerpc/mm/mem.c b/arch/powerpc/mm/mem.c
index be941d382c8d..c95b7fe9f298 100644
--- a/arch/powerpc/mm/mem.c
+++ b/arch/powerpc/mm/mem.c
@@ -31,6 +31,7 @@
 #include <linux/slab.h>
 #include <linux/vmalloc.h>
 #include <linux/memremap.h>
+#include <linux/dma-direct.h>
 
 #include <asm/pgalloc.h>
 #include <asm/prom.h>
@@ -201,10 +202,10 @@ static int __init mark_nonram_nosave(void)
  * everything else. GFP_DMA32 page allocations automatically fall back to
  * ZONE_DMA.
  *
- * By using 31-bit unconditionally, we can exploit ARCH_ZONE_DMA_BITS to
- * inform the generic DMA mapping code.  32-bit only devices (if not handled
- * by an IOMMU anyway) will take a first dip into ZONE_NORMAL and get
- * otherwise served by ZONE_DMA.
+ * By using 31-bit unconditionally, we can exploit zone_dma_bits to inform the
+ * generic DMA mapping code.  32-bit only devices (if not handled by an IOMMU
+ * anyway) will take a first dip into ZONE_NORMAL and get otherwise served by
+ * ZONE_DMA.
  */
 static unsigned long max_zone_pfns[MAX_NR_ZONES];
 
@@ -237,9 +238,18 @@ void __init paging_init(void)
 	printk(KERN_DEBUG "Memory hole size: %ldMB\n",
 	       (long int)((top_of_ram - total_ram) >> 20));
 
+	/*
+	 * Allow 30-bit DMA for very limited Broadcom wifi chips on many
+	 * powerbooks.
+	 */
+	if (IS_ENABLED(CONFIG_PPC32))
+		zone_dma_bits = 30;
+	else
+		zone_dma_bits = 31;
+
 #ifdef CONFIG_ZONE_DMA
 	max_zone_pfns[ZONE_DMA]	= min(max_low_pfn,
-				      1UL << (ARCH_ZONE_DMA_BITS - PAGE_SHIFT));
+				      1UL << (zone_dma_bits - PAGE_SHIFT));
 #endif
 	max_zone_pfns[ZONE_NORMAL] = max_low_pfn;
 #ifdef CONFIG_HIGHMEM
diff --git a/arch/s390/include/asm/page.h b/arch/s390/include/asm/page.h
index 823578c6b9e2..a4d38092530a 100644
--- a/arch/s390/include/asm/page.h
+++ b/arch/s390/include/asm/page.h
@@ -177,8 +177,6 @@ static inline int devmem_is_allowed(unsigned long pfn)
 #define VM_DATA_DEFAULT_FLAGS	(VM_READ | VM_WRITE | \
 				 VM_MAYREAD | VM_MAYWRITE | VM_MAYEXEC)
 
-#define ARCH_ZONE_DMA_BITS	31
-
 #include <asm-generic/memory_model.h>
 #include <asm-generic/getorder.h>
 
diff --git a/arch/s390/mm/init.c b/arch/s390/mm/init.c
index a124f19f7b3c..f0ce22220565 100644
--- a/arch/s390/mm/init.c
+++ b/arch/s390/mm/init.c
@@ -118,6 +118,7 @@ void __init paging_init(void)
 
 	sparse_memory_present_with_active_regions(MAX_NUMNODES);
 	sparse_init();
+	zone_dma_bits = 31;
 	memset(max_zone_pfns, 0, sizeof(max_zone_pfns));
 	max_zone_pfns[ZONE_DMA] = PFN_DOWN(MAX_DMA_ADDRESS);
 	max_zone_pfns[ZONE_NORMAL] = max_low_pfn;
diff --git a/include/linux/dma-direct.h b/include/linux/dma-direct.h
index adf993a3bd58..d03af3605460 100644
--- a/include/linux/dma-direct.h
+++ b/include/linux/dma-direct.h
@@ -5,6 +5,8 @@
 #include <linux/dma-mapping.h>
 #include <linux/mem_encrypt.h>
 
+extern unsigned int zone_dma_bits;
+
 #ifdef CONFIG_ARCH_HAS_PHYS_TO_DMA
 #include <asm/dma-direct.h>
 #else
diff --git a/kernel/dma/direct.c b/kernel/dma/direct.c
index 8402b29c280f..0b67c04e531b 100644
--- a/kernel/dma/direct.c
+++ b/kernel/dma/direct.c
@@ -16,12 +16,11 @@
 #include <linux/swiotlb.h>
 
 /*
- * Most architectures use ZONE_DMA for the first 16 Megabytes, but
- * some use it for entirely different regions:
+ * Most architectures use ZONE_DMA for the first 16 Megabytes, but some use it
+ * it for entirely different regions. In that case the arch code needs to
+ * override the variable below for dma-direct to work properly.
  */
-#ifndef ARCH_ZONE_DMA_BITS
-#define ARCH_ZONE_DMA_BITS 24
-#endif
+unsigned int zone_dma_bits __ro_after_init = 24;
 
 static void report_addr(struct device *dev, dma_addr_t dma_addr, size_t size)
 {
@@ -69,7 +68,7 @@ static gfp_t __dma_direct_optimal_gfp_mask(struct device *dev, u64 dma_mask,
 	 * Note that GFP_DMA32 and GFP_DMA are no ops without the corresponding
 	 * zones.
 	 */
-	if (*phys_mask <= DMA_BIT_MASK(ARCH_ZONE_DMA_BITS))
+	if (*phys_mask <= DMA_BIT_MASK(zone_dma_bits))
 		return GFP_DMA;
 	if (*phys_mask <= DMA_BIT_MASK(32))
 		return GFP_DMA32;
@@ -395,7 +394,7 @@ int dma_direct_supported(struct device *dev, u64 mask)
 	u64 min_mask;
 
 	if (IS_ENABLED(CONFIG_ZONE_DMA))
-		min_mask = DMA_BIT_MASK(ARCH_ZONE_DMA_BITS);
+		min_mask = DMA_BIT_MASK(zone_dma_bits);
 	else
 		min_mask = DMA_BIT_MASK(32);
 
-- 
2.23.0

Re: [PATCH] dma/direct: turn ARCH_ZONE_DMA_BITS into a variable

[Christoph Hellwig]

On Thu, Oct 31, 2019 at 04:28:37PM +0100, Nicolas Saenz Julienne wrote:
> Some architectures, notably ARM, are interested in tweaking this
> depending on their runtime DMA addressing limitations.
> 
> Signed-off-by: Nicolas Saenz Julienne <nsaenzjulienne@suse.de>
> ---
> 
> Changes since RFC:
>  - Rebased to v5.4-rc6, fixed arm64 code.
> 
> NOTE: This will only apply to linux-next, where 

missing end of the sentence.  But only applying to linux-next isn't
going to help anyone..

> diff --git a/arch/arm64/mm/init.c b/arch/arm64/mm/init.c
> index 45c00a54909c..f716ea634804 100644
> --- a/arch/arm64/mm/init.c
> +++ b/arch/arm64/mm/init.c
> @@ -20,6 +20,7 @@
>  #include <linux/sort.h>
>  #include <linux/of.h>
>  #include <linux/of_fdt.h>
> +#include <linux/dma-direct.h>
>  #include <linux/dma-mapping.h>
>  #include <linux/dma-contiguous.h>
>  #include <linux/efi.h>
> @@ -41,6 +42,8 @@
>  #include <asm/tlb.h>
>  #include <asm/alternative.h>
>  
> +#define ARM64_ZONE_DMA_BITS	30
> +
>  /*
>   * We need to be able to catch inadvertent references to memstart_addr
>   * that occur (potentially in generic code) before arm64_memblock_init()
> @@ -424,6 +427,8 @@ void __init arm64_memblock_init(void)
>  	else
>  		arm64_dma_phys_limit = PHYS_MASK + 1;
>  
> +	zone_dma_bits = ARM64_ZONE_DMA_BITS;
> +
>  	reserve_crashkernel();

This actually adds a new limit, as there wasn't one before for arm64.

Re: [PATCH] dma/direct: turn ARCH_ZONE_DMA_BITS into a variable

[Nicolas Saenz Julienne]

[-- Attachment #1: Type: text/plain, Size: 1823 bytes --]

On Thu, 2019-10-31 at 16:47 +0100, Christoph Hellwig wrote:
> On Thu, Oct 31, 2019 at 04:28:37PM +0100, Nicolas Saenz Julienne wrote:
> > Some architectures, notably ARM, are interested in tweaking this
> > depending on their runtime DMA addressing limitations.
> > 
> > Signed-off-by: Nicolas Saenz Julienne <nsaenzjulienne@suse.de>
> > ---
> > 
> > Changes since RFC:
> >  - Rebased to v5.4-rc6, fixed arm64 code.
> > 
> > NOTE: This will only apply to linux-next, where 
> 
> missing end of the sentence.  But only applying to linux-next isn't
> going to help anyone..

Arrgh, excuse me, I meant to delete that line.

> > diff --git a/arch/arm64/mm/init.c b/arch/arm64/mm/init.c
> > index 45c00a54909c..f716ea634804 100644
> > --- a/arch/arm64/mm/init.c
> > +++ b/arch/arm64/mm/init.c
> > @@ -20,6 +20,7 @@
> >  #include <linux/sort.h>
> >  #include <linux/of.h>
> >  #include <linux/of_fdt.h>
> > +#include <linux/dma-direct.h>
> >  #include <linux/dma-mapping.h>
> >  #include <linux/dma-contiguous.h>
> >  #include <linux/efi.h>
> > @@ -41,6 +42,8 @@
> >  #include <asm/tlb.h>
> >  #include <asm/alternative.h>
> >  
> > +#define ARM64_ZONE_DMA_BITS	30
> > +
> >  /*
> >   * We need to be able to catch inadvertent references to memstart_addr
> >   * that occur (potentially in generic code) before arm64_memblock_init()
> > @@ -424,6 +427,8 @@ void __init arm64_memblock_init(void)
> >  	else
> >  		arm64_dma_phys_limit = PHYS_MASK + 1;
> >  
> > +	zone_dma_bits = ARM64_ZONE_DMA_BITS;
> > +
> >  	reserve_crashkernel();
> 
> This actually adds a new limit, as there wasn't one before for arm64.

Well, as zone_dma_bits is only relevant in dma/direct when ZONE_DMA is defined
I figured it doesn't matter if the variable is set conditionally to ZONE_DMA or
not.


[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: [PATCH] dma/direct: turn ARCH_ZONE_DMA_BITS into a variable

[Christoph Hellwig]

On Thu, Oct 31, 2019 at 04:53:13PM +0100, Nicolas Saenz Julienne wrote:
> > > +#define ARM64_ZONE_DMA_BITS	30
> > > +
> > >  /*
> > >   * We need to be able to catch inadvertent references to memstart_addr
> > >   * that occur (potentially in generic code) before arm64_memblock_init()
> > > @@ -424,6 +427,8 @@ void __init arm64_memblock_init(void)
> > >  	else
> > >  		arm64_dma_phys_limit = PHYS_MASK + 1;
> > >  
> > > +	zone_dma_bits = ARM64_ZONE_DMA_BITS;
> > > +
> > >  	reserve_crashkernel();
> > 
> > This actually adds a new limit, as there wasn't one before for arm64.
> 
> Well, as zone_dma_bits is only relevant in dma/direct when ZONE_DMA is defined
> I figured it doesn't matter if the variable is set conditionally to ZONE_DMA or
> not.

I'd much prefer that to do separately.

Re: [PATCH] dma/direct: turn ARCH_ZONE_DMA_BITS into a variable

[Nicolas Saenz Julienne]

[-- Attachment #1: Type: text/plain, Size: 1356 bytes --]

On Thu, 2019-10-31 at 16:57 +0100, Christoph Hellwig wrote:
> On Thu, Oct 31, 2019 at 04:53:13PM +0100, Nicolas Saenz Julienne wrote:
> > > > +#define ARM64_ZONE_DMA_BITS	30
> > > > +
> > > >  /*
> > > >   * We need to be able to catch inadvertent references to memstart_addr
> > > >   * that occur (potentially in generic code) before
> > > > arm64_memblock_init()
> > > > @@ -424,6 +427,8 @@ void __init arm64_memblock_init(void)
> > > >  	else
> > > >  		arm64_dma_phys_limit = PHYS_MASK + 1;
> > > >  
> > > > +	zone_dma_bits = ARM64_ZONE_DMA_BITS;
> > > > +
> > > >  	reserve_crashkernel();
> > > 
> > > This actually adds a new limit, as there wasn't one before for arm64.
> > 
> > Well, as zone_dma_bits is only relevant in dma/direct when ZONE_DMA is
> > defined
> > I figured it doesn't matter if the variable is set conditionally to ZONE_DMA
> > or
> > not.
> 
> I'd much prefer that to do separately.

OK, I see what you mean now. It's wrong indeed.

The trouble is the ZONE_DMA series[1] in arm64, also due for v5.5, will be
affected by this patch. I don't know the right way to approach this problem
since depending on the merge order, this patch should be updated or the arm64
ZONE_DMA series fixed.

Maybe it's easier to just wait for v5.6.

Regards,
Nicolas

[1] https://lkml.org/lkml/2019/9/11/734


[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: [PATCH] dma/direct: turn ARCH_ZONE_DMA_BITS into a variable

[Christoph Hellwig]

On Thu, Oct 31, 2019 at 05:22:59PM +0100, Nicolas Saenz Julienne wrote:
> OK, I see what you mean now. It's wrong indeed.
> 
> The trouble is the ZONE_DMA series[1] in arm64, also due for v5.5, will be
> affected by this patch. I don't know the right way to approach this problem
> since depending on the merge order, this patch should be updated or the arm64
> ZONE_DMA series fixed.
> 
> Maybe it's easier to just wait for v5.6.

Ok, I can wait.  Or the arm64 maintainers can pick up this patch if
you want to add it to that series.

Re: [PATCH] dma/direct: turn ARCH_ZONE_DMA_BITS into a variable

[Catalin Marinas]

On Thu, Oct 31, 2019 at 05:58:53PM +0100, Christoph Hellwig wrote:
> On Thu, Oct 31, 2019 at 05:22:59PM +0100, Nicolas Saenz Julienne wrote:
> > OK, I see what you mean now. It's wrong indeed.
> > 
> > The trouble is the ZONE_DMA series[1] in arm64, also due for v5.5, will be
> > affected by this patch. I don't know the right way to approach this problem
> > since depending on the merge order, this patch should be updated or the arm64
> > ZONE_DMA series fixed.
> > 
> > Maybe it's easier to just wait for v5.6.
> 
> Ok, I can wait.  Or the arm64 maintainers can pick up this patch if
> you want to add it to that series.

This branch is stable (may add a fix but not I'm not rebasing it) if you
want to base this patch on top:

https://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux.git/log/?h=for-next/zone-dma

Otherwise, with your ack, I can add it on top of the above branch (aimed
at 5.5).

-- 
Catalin

Re: [PATCH] dma/direct: turn ARCH_ZONE_DMA_BITS into a variable

[Christoph Hellwig]

On Thu, Oct 31, 2019 at 06:06:19PM +0000, Catalin Marinas wrote:
> On Thu, Oct 31, 2019 at 05:58:53PM +0100, Christoph Hellwig wrote:
> > On Thu, Oct 31, 2019 at 05:22:59PM +0100, Nicolas Saenz Julienne wrote:
> > > OK, I see what you mean now. It's wrong indeed.
> > > 
> > > The trouble is the ZONE_DMA series[1] in arm64, also due for v5.5, will be
> > > affected by this patch. I don't know the right way to approach this problem
> > > since depending on the merge order, this patch should be updated or the arm64
> > > ZONE_DMA series fixed.
> > > 
> > > Maybe it's easier to just wait for v5.6.
> > 
> > Ok, I can wait.  Or the arm64 maintainers can pick up this patch if
> > you want to add it to that series.
> 
> This branch is stable (may add a fix but not I'm not rebasing it) if you
> want to base this patch on top:
> 
> https://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux.git/log/?h=for-next/zone-dma
> 
> Otherwise, with your ack, I can add it on top of the above branch (aimed
> at 5.5).

Please go ahead:

Acked-by: Christoph Hellwig <hch@lst.de>