use the generic dma-noncoherent code for microblaze

use the generic dma-noncoherent code for microblaze

[Christoph Hellwig]

Hi Michal,

can you review these patches to switch microblaze to use the generic
dma-noncoherent code?  All the requirements are in mainline already
and we've switched various architectures over to it already.

[PATCH 1/2] microblaze: use generic dma_noncoherent_ops

[Christoph Hellwig]

Switch to the generic noncoherent direct mapping implementation.

This removes the direction-based optimizations in
sync_{single,sg}_for_{cpu,device} which were marked untestested and
do not match the usually very well tested {un,}map_{single,sg}
implementations.

Signed-off-by: Christoph Hellwig <hch@lst.de>
---
 arch/microblaze/Kconfig                   |   4 +
 arch/microblaze/include/asm/Kbuild        |   1 +
 arch/microblaze/include/asm/dma-mapping.h |  28 -----
 arch/microblaze/include/asm/pgtable.h     |   2 -
 arch/microblaze/kernel/dma.c              | 144 ++--------------------
 arch/microblaze/mm/consistent.c           |   9 +-
 6 files changed, 22 insertions(+), 166 deletions(-)
 delete mode 100644 arch/microblaze/include/asm/dma-mapping.h

diff --git a/arch/microblaze/Kconfig b/arch/microblaze/Kconfig
index d14782100088..848e31a86ba5 100644
--- a/arch/microblaze/Kconfig
+++ b/arch/microblaze/Kconfig
@@ -1,6 +1,8 @@
 config MICROBLAZE
 	def_bool y
 	select ARCH_HAS_GCOV_PROFILE_ALL
+	select ARCH_HAS_SYNC_DMA_FOR_CPU
+	select ARCH_HAS_SYNC_DMA_FOR_DEVICE
 	select ARCH_MIGHT_HAVE_PC_PARPORT
 	select ARCH_NO_COHERENT_DMA_MMAP if !MMU
 	select ARCH_WANT_IPC_PARSE_VERSION
@@ -8,6 +10,8 @@ config MICROBLAZE
 	select TIMER_OF
 	select CLONE_BACKWARDS3
 	select COMMON_CLK
+	select DMA_NONCOHERENT_OPS
+	select DMA_NONCOHERENT_MMAP
 	select GENERIC_ATOMIC64
 	select GENERIC_CLOCKEVENTS
 	select GENERIC_CPU_DEVICES
diff --git a/arch/microblaze/include/asm/Kbuild b/arch/microblaze/include/asm/Kbuild
index fe6a6c6e5003..569ba9e670c1 100644
--- a/arch/microblaze/include/asm/Kbuild
+++ b/arch/microblaze/include/asm/Kbuild
@@ -5,6 +5,7 @@ generic-y += bugs.h
 generic-y += compat.h
 generic-y += device.h
 generic-y += div64.h
+generic-y += dma-mapping.h
 generic-y += emergency-restart.h
 generic-y += exec.h
 generic-y += extable.h
diff --git a/arch/microblaze/include/asm/dma-mapping.h b/arch/microblaze/include/asm/dma-mapping.h
deleted file mode 100644
index add50c1373bf..000000000000
--- a/arch/microblaze/include/asm/dma-mapping.h
+++ /dev/null
@@ -1,28 +0,0 @@
-/*
- * Implements the generic device dma API for microblaze and the pci
- *
- * Copyright (C) 2009-2010 Michal Simek <monstr@monstr.eu>
- * Copyright (C) 2009-2010 PetaLogix
- *
- * This file is subject to the terms and conditions of the GNU General
- * Public License. See the file COPYING in the main directory of this
- * archive for more details.
- *
- * This file is base on powerpc and x86 dma-mapping.h versions
- * Copyright (C) 2004 IBM
- */
-
-#ifndef _ASM_MICROBLAZE_DMA_MAPPING_H
-#define _ASM_MICROBLAZE_DMA_MAPPING_H
-
-/*
- * Available generic sets of operations
- */
-extern const struct dma_map_ops dma_nommu_ops;
-
-static inline const struct dma_map_ops *get_arch_dma_ops(struct bus_type *bus)
-{
-	return &dma_nommu_ops;
-}
-
-#endif	/* _ASM_MICROBLAZE_DMA_MAPPING_H */
diff --git a/arch/microblaze/include/asm/pgtable.h b/arch/microblaze/include/asm/pgtable.h
index db8b1fa83452..8a2e654b709f 100644
--- a/arch/microblaze/include/asm/pgtable.h
+++ b/arch/microblaze/include/asm/pgtable.h
@@ -553,8 +553,6 @@ void __init *early_get_page(void);
 
 extern unsigned long ioremap_bot, ioremap_base;
 
-void *consistent_alloc(gfp_t gfp, size_t size, dma_addr_t *dma_handle);
-void consistent_free(size_t size, void *vaddr);
 void consistent_sync(void *vaddr, size_t size, int direction);
 void consistent_sync_page(struct page *page, unsigned long offset,
 	size_t size, int direction);
diff --git a/arch/microblaze/kernel/dma.c b/arch/microblaze/kernel/dma.c
index 3145e7dc8ab1..71032cf64669 100644
--- a/arch/microblaze/kernel/dma.c
+++ b/arch/microblaze/kernel/dma.c
@@ -8,29 +8,15 @@
  */
 
 #include <linux/device.h>
-#include <linux/dma-mapping.h>
+#include <linux/dma-noncoherent.h>
 #include <linux/gfp.h>
 #include <linux/dma-debug.h>
 #include <linux/export.h>
 #include <linux/bug.h>
 #include <asm/cacheflush.h>
 
-static void *dma_nommu_alloc_coherent(struct device *dev, size_t size,
-				       dma_addr_t *dma_handle, gfp_t flag,
-				       unsigned long attrs)
-{
-	return consistent_alloc(flag, size, dma_handle);
-}
-
-static void dma_nommu_free_coherent(struct device *dev, size_t size,
-				     void *vaddr, dma_addr_t dma_handle,
-				     unsigned long attrs)
-{
-	consistent_free(size, vaddr);
-}
-
-static inline void __dma_sync(unsigned long paddr,
-			      size_t size, enum dma_data_direction direction)
+static void __dma_sync(struct device *dev, phys_addr_t paddr, size_t size,
+		enum dma_data_direction direction)
 {
 	switch (direction) {
 	case DMA_TO_DEVICE:
@@ -45,113 +31,21 @@ static inline void __dma_sync(unsigned long paddr,
 	}
 }
 
-static int dma_nommu_map_sg(struct device *dev, struct scatterlist *sgl,
-			     int nents, enum dma_data_direction direction,
-			     unsigned long attrs)
+void arch_sync_dma_for_device(struct device *dev, phys_addr_t paddr,
+		size_t size, enum dma_data_direction dir)
 {
-	struct scatterlist *sg;
-	int i;
-
-	/* FIXME this part of code is untested */
-	for_each_sg(sgl, sg, nents, i) {
-		sg->dma_address = sg_phys(sg);
-
-		if (attrs & DMA_ATTR_SKIP_CPU_SYNC)
-			continue;
-
-		__dma_sync(sg_phys(sg), sg->length, direction);
-	}
-
-	return nents;
-}
-
-static inline dma_addr_t dma_nommu_map_page(struct device *dev,
-					     struct page *page,
-					     unsigned long offset,
-					     size_t size,
-					     enum dma_data_direction direction,
-					     unsigned long attrs)
-{
-	if (!(attrs & DMA_ATTR_SKIP_CPU_SYNC))
-		__dma_sync(page_to_phys(page) + offset, size, direction);
-	return page_to_phys(page) + offset;
+	__dma_sync(dev, paddr, size, dir);
 }
 
-static inline void dma_nommu_unmap_page(struct device *dev,
-					 dma_addr_t dma_address,
-					 size_t size,
-					 enum dma_data_direction direction,
-					 unsigned long attrs)
+void arch_sync_dma_for_cpu(struct device *dev, phys_addr_t paddr,
+		size_t size, enum dma_data_direction dir)
 {
-/* There is not necessary to do cache cleanup
- *
- * phys_to_virt is here because in __dma_sync_page is __virt_to_phys and
- * dma_address is physical address
- */
-	if (!(attrs & DMA_ATTR_SKIP_CPU_SYNC))
-		__dma_sync(dma_address, size, direction);
+	__dma_sync(dev, paddr, size, dir);
 }
 
-static inline void
-dma_nommu_sync_single_for_cpu(struct device *dev,
-			       dma_addr_t dma_handle, size_t size,
-			       enum dma_data_direction direction)
-{
-	/*
-	 * It's pointless to flush the cache as the memory segment
-	 * is given to the CPU
-	 */
-
-	if (direction == DMA_FROM_DEVICE)
-		__dma_sync(dma_handle, size, direction);
-}
-
-static inline void
-dma_nommu_sync_single_for_device(struct device *dev,
-				  dma_addr_t dma_handle, size_t size,
-				  enum dma_data_direction direction)
-{
-	/*
-	 * It's pointless to invalidate the cache if the device isn't
-	 * supposed to write to the relevant region
-	 */
-
-	if (direction == DMA_TO_DEVICE)
-		__dma_sync(dma_handle, size, direction);
-}
-
-static inline void
-dma_nommu_sync_sg_for_cpu(struct device *dev,
-			   struct scatterlist *sgl, int nents,
-			   enum dma_data_direction direction)
-{
-	struct scatterlist *sg;
-	int i;
-
-	/* FIXME this part of code is untested */
-	if (direction == DMA_FROM_DEVICE)
-		for_each_sg(sgl, sg, nents, i)
-			__dma_sync(sg->dma_address, sg->length, direction);
-}
-
-static inline void
-dma_nommu_sync_sg_for_device(struct device *dev,
-			      struct scatterlist *sgl, int nents,
-			      enum dma_data_direction direction)
-{
-	struct scatterlist *sg;
-	int i;
-
-	/* FIXME this part of code is untested */
-	if (direction == DMA_TO_DEVICE)
-		for_each_sg(sgl, sg, nents, i)
-			__dma_sync(sg->dma_address, sg->length, direction);
-}
-
-static
-int dma_nommu_mmap_coherent(struct device *dev, struct vm_area_struct *vma,
-			     void *cpu_addr, dma_addr_t handle, size_t size,
-			     unsigned long attrs)
+int arch_dma_mmap(struct device *dev, struct vm_area_struct *vma,
+		void *cpu_addr, dma_addr_t handle, size_t size,
+		unsigned long attrs)
 {
 #ifdef CONFIG_MMU
 	unsigned long user_count = vma_pages(vma);
@@ -170,17 +64,3 @@ int dma_nommu_mmap_coherent(struct device *dev, struct vm_area_struct *vma,
 	return -ENXIO;
 #endif
 }
-
-const struct dma_map_ops dma_nommu_ops = {
-	.alloc			= dma_nommu_alloc_coherent,
-	.free			= dma_nommu_free_coherent,
-	.mmap			= dma_nommu_mmap_coherent,
-	.map_sg			= dma_nommu_map_sg,
-	.map_page		= dma_nommu_map_page,
-	.unmap_page		= dma_nommu_unmap_page,
-	.sync_single_for_cpu	= dma_nommu_sync_single_for_cpu,
-	.sync_single_for_device	= dma_nommu_sync_single_for_device,
-	.sync_sg_for_cpu	= dma_nommu_sync_sg_for_cpu,
-	.sync_sg_for_device	= dma_nommu_sync_sg_for_device,
-};
-EXPORT_SYMBOL(dma_nommu_ops);
diff --git a/arch/microblaze/mm/consistent.c b/arch/microblaze/mm/consistent.c
index b06c3a7faf20..b9a9c8c3397b 100644
--- a/arch/microblaze/mm/consistent.c
+++ b/arch/microblaze/mm/consistent.c
@@ -33,6 +33,7 @@
 #include <linux/pci.h>
 #include <linux/interrupt.h>
 #include <linux/gfp.h>
+#include <linux/dma-noncoherent.h>
 
 #include <asm/pgalloc.h>
 #include <linux/io.h>
@@ -59,7 +60,8 @@
  * uncached region.  This will no doubt cause big problems if memory allocated
  * here is not also freed properly. -- JW
  */
-void *consistent_alloc(gfp_t gfp, size_t size, dma_addr_t *dma_handle)
+void *arch_dma_alloc(struct device *dev, size_t size, dma_addr_t *dma_handle,
+		gfp_t gfp, unsigned long attrs)
 {
 	unsigned long order, vaddr;
 	void *ret;
@@ -154,7 +156,6 @@ void *consistent_alloc(gfp_t gfp, size_t size, dma_addr_t *dma_handle)
 
 	return ret;
 }
-EXPORT_SYMBOL(consistent_alloc);
 
 #ifdef CONFIG_MMU
 static pte_t *consistent_virt_to_pte(void *vaddr)
@@ -178,7 +179,8 @@ unsigned long consistent_virt_to_pfn(void *vaddr)
 /*
  * free page(s) as defined by the above mapping.
  */
-void consistent_free(size_t size, void *vaddr)
+void arch_dma_free(struct device *dev, size_t size, void *vaddr,
+		dma_addr_t dma_addr, unsigned long attrs)
 {
 	struct page *page;
 
@@ -218,7 +220,6 @@ void consistent_free(size_t size, void *vaddr)
 	flush_tlb_all();
 #endif
 }
-EXPORT_SYMBOL(consistent_free);
 
 /*
  * make an area consistent.
-- 
2.18.0

[PATCH 2/2] microblaze: remove consistent_sync and consistent_sync_page

[Christoph Hellwig]

Both unused.

Signed-off-by: Christoph Hellwig <hch@lst.de>
---
 arch/microblaze/include/asm/pgtable.h |  3 --
 arch/microblaze/mm/consistent.c       | 45 ---------------------------
 2 files changed, 48 deletions(-)

diff --git a/arch/microblaze/include/asm/pgtable.h b/arch/microblaze/include/asm/pgtable.h
index 8a2e654b709f..7b650ab14fa0 100644
--- a/arch/microblaze/include/asm/pgtable.h
+++ b/arch/microblaze/include/asm/pgtable.h
@@ -553,9 +553,6 @@ void __init *early_get_page(void);
 
 extern unsigned long ioremap_bot, ioremap_base;
 
-void consistent_sync(void *vaddr, size_t size, int direction);
-void consistent_sync_page(struct page *page, unsigned long offset,
-	size_t size, int direction);
 unsigned long consistent_virt_to_pfn(void *vaddr);
 
 void setup_memory(void);
diff --git a/arch/microblaze/mm/consistent.c b/arch/microblaze/mm/consistent.c
index b9a9c8c3397b..c9a278ac795a 100644
--- a/arch/microblaze/mm/consistent.c
+++ b/arch/microblaze/mm/consistent.c
@@ -220,48 +220,3 @@ void arch_dma_free(struct device *dev, size_t size, void *vaddr,
 	flush_tlb_all();
 #endif
 }
-
-/*
- * make an area consistent.
- */
-void consistent_sync(void *vaddr, size_t size, int direction)
-{
-	unsigned long start;
-	unsigned long end;
-
-	start = (unsigned long)vaddr;
-
-	/* Convert start address back down to unshadowed memory region */
-#ifdef CONFIG_XILINX_UNCACHED_SHADOW
-	start &= ~UNCACHED_SHADOW_MASK;
-#endif
-	end = start + size;
-
-	switch (direction) {
-	case PCI_DMA_NONE:
-		BUG();
-	case PCI_DMA_FROMDEVICE:	/* invalidate only */
-		invalidate_dcache_range(start, end);
-		break;
-	case PCI_DMA_TODEVICE:		/* writeback only */
-		flush_dcache_range(start, end);
-		break;
-	case PCI_DMA_BIDIRECTIONAL:	/* writeback and invalidate */
-		flush_dcache_range(start, end);
-		break;
-	}
-}
-EXPORT_SYMBOL(consistent_sync);
-
-/*
- * consistent_sync_page makes memory consistent. identical
- * to consistent_sync, but takes a struct page instead of a
- * virtual address
- */
-void consistent_sync_page(struct page *page, unsigned long offset,
-	size_t size, int direction)
-{
-	unsigned long start = (unsigned long)page_address(page) + offset;
-	consistent_sync((void *)start, size, direction);
-}
-EXPORT_SYMBOL(consistent_sync_page);
-- 
2.18.0

Re: use the generic dma-noncoherent code for microblaze

[Michal Simek]

Hi,

On 19.7.2018 14:54, Christoph Hellwig wrote:
> Hi Michal,
> 
> can you review these patches to switch microblaze to use the generic
> dma-noncoherent code?  All the requirements are in mainline already
> and we've switched various architectures over to it already.
> 

I can't see any issue with them.
Both applied to next.

Thanks,
Michal

Re: use the generic dma-noncoherent code for microblaze

[Michal Simek]

Hi,

On 19.7.2018 14:54, Christoph Hellwig wrote:
> Hi Michal,
> 
> can you review these patches to switch microblaze to use the generic
> dma-noncoherent code?  All the requirements are in mainline already
> and we've switched various architectures over to it already.
> 

I can't see any issue with them.
Both applied to next.

Thanks,
Michal

Re: use the generic dma-noncoherent code for microblaze

[Christoph Hellwig]

On Fri, Jul 20, 2018 at 02:33:44PM +0200, Michal Simek wrote:
> Hi,
> 
> On 19.7.2018 14:54, Christoph Hellwig wrote:
> > Hi Michal,
> > 
> > can you review these patches to switch microblaze to use the generic
> > dma-noncoherent code?  All the requirements are in mainline already
> > and we've switched various architectures over to it already.
> > 
> 
> I can't see any issue with them.
> Both applied to next.

!