LKML Archive on lore.kernel.org
help / color / mirror / Atom feed
* [PATCH 0/3 v5] dma: add dma_*map*_attrs() interfaces
@ 2008-03-20  1:07 akepner
  2008-03-28  3:18 ` Andrew Morton
  0 siblings, 1 reply; 3+ messages in thread
From: akepner @ 2008-03-20  1:07 UTC (permalink / raw)
  To: Tony Luck, Jesse Barnes, Jes Sorensen, Randy Dunlap,
	Roland Dreier, James Bottomley, David Miller,
	Benjamin Herrenschmidt, Grant Grundler, Michael Ellerman,
	Andrew Morton
  Cc: linux-kernel


v4-v5 changes:

- select HAVE_DMA_ATTRS via Kconfig
- changed the name DMA_ATTR_BARRIER to DMA_ATTR_WRITE_BARRIER
- changed struct dma_attrs to be a wrapper around a bitmap
- kernel-doc-ified some comments
- recombined the lib/swiotlb.c patch into the ia64-arch patch 
  (they need to go in together, or other forward decls are needed)

--- 

Introduce new interfaces, dma_*map*_attrs(), for passing 
architecture-specific attributes when memory is mapped and 
unmapped for DMA. Give the interfaces default implementations 
which ignore attributes. Also introduce the dma_{set|get}_attr() 
interfaces for setting and retrieving individual attributes. Define 
one attribute, DMA_ATTR_WRITE_BARRIER, in anticipation of its use 
by ia64/sn. Select whether architectures implement arch-specific 
versions of the dma_*map*_attrs() interfaces via HAVE_DMA_ATTRS in 
Kconfig.

Signed-off-by: Arthur Kepner <akepner@sgi.com>

---

 arch/Kconfig                |    3 +
 arch/ia64/Kconfig           |    1 
 include/linux/dma-attrs.h   |   74 ++++++++++++++++++++++++++++++++++++++++++++
 include/linux/dma-mapping.h |   35 ++++++++++++++++++++
 4 files changed, 113 insertions(+)

diff --git a/arch/Kconfig b/arch/Kconfig
index 694c9af..3ea332b 100644
--- a/arch/Kconfig
+++ b/arch/Kconfig
@@ -36,3 +36,6 @@ config HAVE_KPROBES
 
 config HAVE_KRETPROBES
 	def_bool n
+
+config HAVE_DMA_ATTRS
+	def_bool n
diff --git a/arch/ia64/Kconfig b/arch/ia64/Kconfig
index 8fa3faf..985b267 100644
--- a/arch/ia64/Kconfig
+++ b/arch/ia64/Kconfig
@@ -19,6 +19,7 @@ config IA64
 	select HAVE_OPROFILE
 	select HAVE_KPROBES
 	select HAVE_KRETPROBES
+	select HAVE_DMA_ATTRS
 	default y
 	help
 	  The Itanium Processor Family is Intel's 64-bit successor to
diff --git a/include/linux/dma-attrs.h b/include/linux/dma-attrs.h
index e69de29..5f151fb 100644
--- a/include/linux/dma-attrs.h
+++ b/include/linux/dma-attrs.h
@@ -0,0 +1,74 @@
+#ifndef _DMA_ATTR_H
+#define _DMA_ATTR_H
+
+#include <linux/bitmap.h>
+#include <linux/bitops.h>
+#include <linux/bug.h>
+
+/**
+ * an enum dma_attr represents an attribute associated with a DMA
+ * mapping. The semantics of each attribute should be defined in
+ * Documentation/DMA-attributes.txt.
+ */
+enum dma_attr {
+	DMA_ATTR_WRITE_BARRIER,
+	DMA_ATTR_MAX,
+};
+
+#define __DMA_ATTRS_LONGS BITS_TO_LONGS(DMA_ATTR_MAX)
+
+/**
+ * struct dma_attrs - an opaque container for DMA attributes
+ * @flags - bitmask representing a collection of enum dma_attr
+ */
+struct dma_attrs {
+	unsigned long flags[__DMA_ATTRS_LONGS];
+};
+
+#define DEFINE_DMA_ATTRS(x) 					\
+	struct dma_attrs x = {					\
+		.flags = { [0 ... __DMA_ATTRS_LONGS-1] = 0 },	\
+	}
+
+static inline void init_dma_attrs(struct dma_attrs *attrs)
+{
+	bitmap_zero(attrs->flags, __DMA_ATTRS_LONGS);
+}
+
+#ifdef CONFIG_HAVE_DMA_ATTRS
+/**
+ * dma_set_attr - set a specific attribute
+ * @attr: attribute to set
+ * @attrs: struct dma_attrs (may be NULL)
+ */
+void dma_set_attr(enum dma_attr attr, struct dma_attrs *attrs)
+{
+	if (attrs == NULL)
+		return;
+	BUG_ON(attr >= DMA_ATTR_MAX);
+	__set_bit(attr, attrs->flags);
+}
+
+/**
+ * dma_get_attr - check for a specific attribute
+ * @attr: attribute to set
+ * @attrs: struct dma_attrs (may be NULL)
+ */
+int dma_get_attr(enum dma_attr attr, struct dma_attrs *attrs)
+{
+	if (attrs == NULL)
+		return 0;
+	BUG_ON(attr >= DMA_ATTR_MAX);
+	return test_bit(attr, attrs->flags);
+}
+#else /* !CONFIG_HAVE_DMA_ATTRS */
+static inline void dma_set_attr(enum dma_attr attr, struct dma_attrs *attrs)
+{
+}
+
+static inline int dma_get_attr(enum dma_attr attr, struct dma_attrs *attrs)
+{
+	return 0;
+}
+#endif /* CONFIG_HAVE_DMA_ATTRS */
+#endif /* _DMA_ATTR_H */
diff --git a/include/linux/dma-mapping.h b/include/linux/dma-mapping.h
index 3320307..f757d63 100644
--- a/include/linux/dma-mapping.h
+++ b/include/linux/dma-mapping.h
@@ -146,4 +146,39 @@ static inline void dmam_release_declared_memory(struct device *dev)
 }
 #endif /* ARCH_HAS_DMA_DECLARE_COHERENT_MEMORY */
 
+#ifndef CONFIG_HAVE_DMA_ATTRS
+struct dma_attrs;
+
+static inline dma_addr_t dma_map_single_attrs(struct device *dev,
+					      void *cpu_addr, size_t size,
+					      enum dma_data_direction dir,
+					      struct dma_attrs *attrs)
+{
+	return dma_map_single(dev, cpu_addr, size, dir);
+}
+
+static inline void dma_unmap_single_attrs(struct device *dev,
+					  dma_addr_t dma_addr, size_t size,
+					  enum dma_data_direction dir,
+					  struct dma_attrs *attrs)
+{
+	return dma_unmap_single(dev, dma_addr, size, dir);
+}
+
+static inline int dma_map_sg_attrs(struct device *dev, struct scatterlist *sgl,
+				   int nents, enum dma_data_direction dir,
+				   struct dma_attrs *attrs)
+{
+	return dma_map_sg(dev, sgl, nents, dir);
+}
+
+static inline void dma_unmap_sg_attrs(struct device *dev,
+				      struct scatterlist *sgl, int nents,
+				      enum dma_data_direction dir,
+				      struct dma_attrs *attrs)
+{
+	return dma_unmap_sg(dev, sgl, nents, dir);
+}
+#endif /* CONFIG_HAVE_DMA_ATTRS */
+
 #endif

^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: [PATCH 0/3 v5] dma: add dma_*map*_attrs() interfaces
  2008-03-20  1:07 [PATCH 0/3 v5] dma: add dma_*map*_attrs() interfaces akepner
@ 2008-03-28  3:18 ` Andrew Morton
  2008-04-01  1:31   ` akepner
  0 siblings, 1 reply; 3+ messages in thread
From: Andrew Morton @ 2008-03-28  3:18 UTC (permalink / raw)
  To: akepner
  Cc: Tony Luck, Jesse Barnes, Jes Sorensen, Randy Dunlap,
	Roland Dreier, James Bottomley, David Miller,
	Benjamin Herrenschmidt, Grant Grundler, Michael Ellerman,
	linux-kernel

On Wed, 19 Mar 2008 18:07:09 -0700 akepner@sgi.com wrote:

> 
> Introduce new interfaces, dma_*map*_attrs(), for passing 
> architecture-specific attributes when memory is mapped and 
> unmapped for DMA. Give the interfaces default implementations 
> which ignore attributes. Also introduce the dma_{set|get}_attr() 
> interfaces for setting and retrieving individual attributes. Define 
> one attribute, DMA_ATTR_WRITE_BARRIER, in anticipation of its use 
> by ia64/sn. Select whether architectures implement arch-specific 
> versions of the dma_*map*_attrs() interfaces via HAVE_DMA_ATTRS in 
> Kconfig.

alpha allmodconfig:


In file included from include/asm/pci.h:7,
                 from include/linux/pci.h:953,
                 from include/asm/core_mcpcia.h:9,
                 from include/asm/io.h:226,
                 from arch/alpha/kernel/asm-offsets.c:11:
include/linux/dma-mapping.h: In function 'dma_map_single_attrs':
include/linux/dma-mapping.h:157: error: implicit declaration of function 'pci_map_single'
include/linux/dma-mapping.h:157: error: implicit declaration of function 'alpha_gendev_to_pci'
include/linux/dma-mapping.h: In function 'dma_unmap_single_attrs':
include/linux/dma-mapping.h:165: error: implicit declaration of function 'pci_unmap_single'
include/linux/dma-mapping.h:165: warning: 'return' with a value, in function returning void
include/linux/dma-mapping.h: In function 'dma_map_sg_attrs':
include/linux/dma-mapping.h:172: error: implicit declaration of function 'pci_map_sg'
include/linux/dma-mapping.h: In function 'dma_unmap_sg_attrs':
include/linux/dma-mapping.h:180: error: implicit declaration of function 'pci_unmap_sg'
include/linux/dma-mapping.h:180: warning: 'return' with a value, in function returning void
In file included from include/linux/pci.h:953,
                 from include/asm/core_mcpcia.h:9,
                 from include/asm/io.h:226,
                 from arch/alpha/kernel/asm-offsets.c:11:


We're in include hell, I think.  alpha defined CONFIG_HAS_DMA, so it has
already included asm/dma-mapping.h, which defines pci_map_single().  But I
suspect that the compiler skipped that inclusion because it was recursive. 
Or something.

I will point you at http://userweb.kernel.org/~akpm/cross-compilers/ and
run away, sorry.


^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: [PATCH 0/3 v5] dma: add dma_*map*_attrs() interfaces
  2008-03-28  3:18 ` Andrew Morton
@ 2008-04-01  1:31   ` akepner
  0 siblings, 0 replies; 3+ messages in thread
From: akepner @ 2008-04-01  1:31 UTC (permalink / raw)
  To: Andrew Morton
  Cc: Tony Luck, Jesse Barnes, Jes Sorensen, Randy Dunlap,
	Roland Dreier, James Bottomley, David Miller,
	Benjamin Herrenschmidt, Grant Grundler, Michael Ellerman,
	linux-kernel

On Thu, Mar 27, 2008 at 08:18:43PM -0700, Andrew Morton wrote:
> On Wed, 19 Mar 2008 18:07:09 -0700 akepner@sgi.com wrote:
> 
> alpha allmodconfig:
> 
> 
> In file included from include/asm/pci.h:7,
>                  from include/linux/pci.h:953,
>                  from include/asm/core_mcpcia.h:9,
>                  from include/asm/io.h:226,
>                  from arch/alpha/kernel/asm-offsets.c:11:
> include/linux/dma-mapping.h: In function 'dma_map_single_attrs':
> include/linux/dma-mapping.h:157: error: implicit declaration of function 'pci_map_single'
> ....
> 
> We're in include hell, I think....

Yeah.

Alpha did some macro substitutions differently than on the 
other platforms I had built and tested with.

The simple fix is to convert the new _attrs()* functions 
in include/linux/dma-mapping.h into macros. 

An updated patch follows. Built for alpha, ia64, x86_64. 
Tested on ia64 and x86_64.

-- 
Arthur


^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2008-04-01  1:33 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2008-03-20  1:07 [PATCH 0/3 v5] dma: add dma_*map*_attrs() interfaces akepner
2008-03-28  3:18 ` Andrew Morton
2008-04-01  1:31   ` akepner

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).