Skip to contentSkip to conversation

: dma, x86, xen: reduce SWIOTLB usage in Xen guests

PATCH Superseded David Vrabel 2 participants 11 messages Nov 20, 2014 – Nov 21, 2014

Subsystems (6)

v4 of 5 Superseded Latest is v5 →

lore
Share & tools
lore
#0 lore raw

v4 cover : dma, x86, xen: reduce SWIOTLB usage in Xen guests

On systems where DMA addresses and physical addresses are not 1:1 (such as Xen PV guests), the generic dma_get_required_mask() will not return the correct mask (since it uses max_pfn).

Some device drivers (such as mptsas, mpt2sas) use dma_get_required_mask() to set the device's DMA mask to allow them to use only 32-bit DMA addresses in hardware structures. This results in unnecessary use of the SWIOTLB if DMA addresses are more than 32-bits, impacting performance significantly.

This series allows Xen PV guests to override the default dma_get_required_mask() with one that calculates the DMA mask from the maximum MFN (and not the PFN).

Changes in v4:
- Assume 64-bit mask is required.

Changes in v3:
- fix off-by-one in xen_dma_get_required_mask()
- split ia64 changes into separate patch.

Changes in v2:
- split x86 and xen changes into separate patches

David

#1 lore raw

PATCH v4 1/4 dma: add dma_get_required_mask_from_max_pfn()

A generic dma_get_required_mask() is useful even for architectures (such as ia64) that define ARCH_HAS_GET_REQUIRED_MASK.

---
 drivers/base/platform.c     |   10 ++++++++--
 include/linux/dma-mapping.h |    1 +
 2 files changed, 9 insertions(+), 2 deletions(-)
drivers/base/platform.c +8 -2
index b2afc29..f9f3930 100644
@@ -1009,8 +1009,7 @@ int __init platform_bus_init(void)
 	return error;
 }
 
-#ifndef ARCH_HAS_DMA_GET_REQUIRED_MASK
-u64 dma_get_required_mask(struct device *dev)
+u64 dma_get_required_mask_from_max_pfn(struct device *dev)
 {
 	u32 low_totalram = ((max_pfn - 1) << PAGE_SHIFT);
 	u32 high_totalram = ((max_pfn - 1) >> (32 - PAGE_SHIFT));
@@ -1028,6 +1027,13 @@ u64 dma_get_required_mask(struct device *dev)
 	}
 	return mask;
 }
+EXPORT_SYMBOL_GPL(dma_get_required_mask_from_max_pfn);
+
+#ifndef ARCH_HAS_DMA_GET_REQUIRED_MASK
+u64 dma_get_required_mask(struct device *dev)
+{
+	return dma_get_required_mask_from_max_pfn(dev);
+}
 EXPORT_SYMBOL_GPL(dma_get_required_mask);
 #endif
 
include/linux/dma-mapping.h +1 -0
index d5d3881..6e2fdfc 100644
@@ -127,6 +127,7 @@ static inline int dma_coerce_mask_and_coherent(struct device *dev, u64 mask)
 	return dma_set_mask_and_coherent(dev, mask);
 }
 
+extern u64 dma_get_required_mask_from_max_pfn(struct device *dev);
 extern u64 dma_get_required_mask(struct device *dev);
 
 #ifndef set_arch_dma_coherent_ops
Signature · 1 line
1.7.10.4
#2 lore raw

PATCH v4 1/4 dma: add dma_get_required_mask_from_max_pfn()

In reply to David Vrabel

A generic dma_get_required_mask() is useful even for architectures (such as ia64) that define ARCH_HAS_GET_REQUIRED_MASK.

---
 drivers/base/platform.c     |   10 ++++++++--
 include/linux/dma-mapping.h |    1 +
 2 files changed, 9 insertions(+), 2 deletions(-)
drivers/base/platform.c +8 -2
index b2afc29..f9f3930 100644
@@ -1009,8 +1009,7 @@ int __init platform_bus_init(void)
 	return error;
 }
 
-#ifndef ARCH_HAS_DMA_GET_REQUIRED_MASK
-u64 dma_get_required_mask(struct device *dev)
+u64 dma_get_required_mask_from_max_pfn(struct device *dev)
 {
 	u32 low_totalram = ((max_pfn - 1) << PAGE_SHIFT);
 	u32 high_totalram = ((max_pfn - 1) >> (32 - PAGE_SHIFT));
@@ -1028,6 +1027,13 @@ u64 dma_get_required_mask(struct device *dev)
 	}
 	return mask;
 }
+EXPORT_SYMBOL_GPL(dma_get_required_mask_from_max_pfn);
+
+#ifndef ARCH_HAS_DMA_GET_REQUIRED_MASK
+u64 dma_get_required_mask(struct device *dev)
+{
+	return dma_get_required_mask_from_max_pfn(dev);
+}
 EXPORT_SYMBOL_GPL(dma_get_required_mask);
 #endif
 
include/linux/dma-mapping.h +1 -0
index d5d3881..6e2fdfc 100644
@@ -127,6 +127,7 @@ static inline int dma_coerce_mask_and_coherent(struct device *dev, u64 mask)
 	return dma_set_mask_and_coherent(dev, mask);
 }
 
+extern u64 dma_get_required_mask_from_max_pfn(struct device *dev);
 extern u64 dma_get_required_mask(struct device *dev);
 
 #ifndef set_arch_dma_coherent_ops
Signature · 1 line
1.7.10.4
#3 lore raw

PATCH v4 2/4 ia64: use common dma_get_required_mask_from_pfn()

In reply to David Vrabel
---
 arch/ia64/include/asm/machvec.h      |    2 +-
 arch/ia64/include/asm/machvec_init.h |    1 -
 arch/ia64/pci/pci.c                  |   20 --------------------
 3 files changed, 1 insertion(+), 22 deletions(-)
arch/ia64/include/asm/machvec.h +1 -1
index 9c39bdf..beaa47d 100644
@@ -287,7 +287,7 @@ extern struct dma_map_ops *dma_get_ops(struct device *);
 # define platform_dma_get_ops		dma_get_ops
 #endif
 #ifndef platform_dma_get_required_mask
-# define  platform_dma_get_required_mask	ia64_dma_get_required_mask
+# define  platform_dma_get_required_mask	dma_get_required_mask_from_max_pfn
 #endif
 #ifndef platform_irq_to_vector
 # define platform_irq_to_vector		__ia64_irq_to_vector
arch/ia64/include/asm/machvec_init.h +0 -1
index 37a4698..ef964b2 100644
@@ -3,7 +3,6 @@
 
 extern ia64_mv_send_ipi_t ia64_send_ipi;
 extern ia64_mv_global_tlb_purge_t ia64_global_tlb_purge;
-extern ia64_mv_dma_get_required_mask ia64_dma_get_required_mask;
 extern ia64_mv_irq_to_vector __ia64_irq_to_vector;
 extern ia64_mv_local_vector_to_irq __ia64_local_vector_to_irq;
 extern ia64_mv_pci_get_legacy_mem_t ia64_pci_get_legacy_mem;
arch/ia64/pci/pci.c +0 -20
index 291a582..79da21b 100644
@@ -791,26 +791,6 @@ static void __init set_pci_dfl_cacheline_size(void)
 	pci_dfl_cache_line_size = (1 << cci.pcci_line_size) / 4;
 }
 
-u64 ia64_dma_get_required_mask(struct device *dev)
-{
-	u32 low_totalram = ((max_pfn - 1) << PAGE_SHIFT);
-	u32 high_totalram = ((max_pfn - 1) >> (32 - PAGE_SHIFT));
-	u64 mask;
-
-	if (!high_totalram) {
-		/* convert to mask just covering totalram */
-		low_totalram = (1 << (fls(low_totalram) - 1));
-		low_totalram += low_totalram - 1;
-		mask = low_totalram;
-	} else {
-		high_totalram = (1 << (fls(high_totalram) - 1));
-		high_totalram += high_totalram - 1;
-		mask = (((u64)high_totalram) << 32) + 0xffffffff;
-	}
-	return mask;
-}
-EXPORT_SYMBOL_GPL(ia64_dma_get_required_mask);
-
 u64 dma_get_required_mask(struct device *dev)
 {
 	return platform_dma_get_required_mask(dev);
Signature · 1 line
1.7.10.4
#4 lore raw

PATCH v4 2/4 ia64: use common dma_get_required_mask_from_pfn()

In reply to David Vrabel
---
 arch/ia64/include/asm/machvec.h      |    2 +-
 arch/ia64/include/asm/machvec_init.h |    1 -
 arch/ia64/pci/pci.c                  |   20 --------------------
 3 files changed, 1 insertion(+), 22 deletions(-)
arch/ia64/include/asm/machvec.h +1 -1
index 9c39bdf..beaa47d 100644
@@ -287,7 +287,7 @@ extern struct dma_map_ops *dma_get_ops(struct device *);
 # define platform_dma_get_ops		dma_get_ops
 #endif
 #ifndef platform_dma_get_required_mask
-# define  platform_dma_get_required_mask	ia64_dma_get_required_mask
+# define  platform_dma_get_required_mask	dma_get_required_mask_from_max_pfn
 #endif
 #ifndef platform_irq_to_vector
 # define platform_irq_to_vector		__ia64_irq_to_vector
arch/ia64/include/asm/machvec_init.h +0 -1
index 37a4698..ef964b2 100644
@@ -3,7 +3,6 @@
 
 extern ia64_mv_send_ipi_t ia64_send_ipi;
 extern ia64_mv_global_tlb_purge_t ia64_global_tlb_purge;
-extern ia64_mv_dma_get_required_mask ia64_dma_get_required_mask;
 extern ia64_mv_irq_to_vector __ia64_irq_to_vector;
 extern ia64_mv_local_vector_to_irq __ia64_local_vector_to_irq;
 extern ia64_mv_pci_get_legacy_mem_t ia64_pci_get_legacy_mem;
arch/ia64/pci/pci.c +0 -20
index 291a582..79da21b 100644
@@ -791,26 +791,6 @@ static void __init set_pci_dfl_cacheline_size(void)
 	pci_dfl_cache_line_size = (1 << cci.pcci_line_size) / 4;
 }
 
-u64 ia64_dma_get_required_mask(struct device *dev)
-{
-	u32 low_totalram = ((max_pfn - 1) << PAGE_SHIFT);
-	u32 high_totalram = ((max_pfn - 1) >> (32 - PAGE_SHIFT));
-	u64 mask;
-
-	if (!high_totalram) {
-		/* convert to mask just covering totalram */
-		low_totalram = (1 << (fls(low_totalram) - 1));
-		low_totalram += low_totalram - 1;
-		mask = low_totalram;
-	} else {
-		high_totalram = (1 << (fls(high_totalram) - 1));
-		high_totalram += high_totalram - 1;
-		mask = (((u64)high_totalram) << 32) + 0xffffffff;
-	}
-	return mask;
-}
-EXPORT_SYMBOL_GPL(ia64_dma_get_required_mask);
-
 u64 dma_get_required_mask(struct device *dev)
 {
 	return platform_dma_get_required_mask(dev);
Signature · 1 line
1.7.10.4
#5 lore raw

PATCH v4 3/4 x86: allow dma_get_required_mask() to be overridden

In reply to David Vrabel

Use dma_ops->get_required_mask() if provided, defaulting to dma_get_requried_mask_from_max_pfn().

This is needed on systems (such as Xen PV guests) where the DMA address and the physical address are not equal.

ARCH_HAS_DMA_GET_REQUIRED_MASK is defined in asm/device.h instead of asm/dma-mapping.h because linux/dma-mapping.h uses the define before including asm/dma-mapping.h

---
 arch/x86/include/asm/device.h |    2 ++
 arch/x86/kernel/pci-dma.c     |    8 ++++++++
 2 files changed, 10 insertions(+)
arch/x86/include/asm/device.h +2 -0
index 03dd729..10bc628 100644
@@ -13,4 +13,6 @@ struct dev_archdata {
 struct pdev_archdata {
 };
 
+#define ARCH_HAS_DMA_GET_REQUIRED_MASK
+
 #endif /* _ASM_X86_DEVICE_H */
arch/x86/kernel/pci-dma.c +8 -0
index a25e202..5154400 100644
@@ -140,6 +140,14 @@ void dma_generic_free_coherent(struct device *dev, size_t size, void *vaddr,
 		free_pages((unsigned long)vaddr, get_order(size));
 }
 
+u64 dma_get_required_mask(struct device *dev)
+{
+	if (dma_ops->get_required_mask)
+		return dma_ops->get_required_mask(dev);
+	return dma_get_required_mask_from_max_pfn(dev);
+}
+EXPORT_SYMBOL_GPL(dma_get_required_mask);
+
 /*
  * See <Documentation/x86/x86_64/boot-options.txt> for the iommu kernel
  * parameter documentation.
Signature · 1 line
1.7.10.4
#6 lore raw

PATCH v4 3/4 x86: allow dma_get_required_mask() to be overridden

In reply to David Vrabel

Use dma_ops->get_required_mask() if provided, defaulting to dma_get_requried_mask_from_max_pfn().

This is needed on systems (such as Xen PV guests) where the DMA address and the physical address are not equal.

ARCH_HAS_DMA_GET_REQUIRED_MASK is defined in asm/device.h instead of asm/dma-mapping.h because linux/dma-mapping.h uses the define before including asm/dma-mapping.h

---
 arch/x86/include/asm/device.h |    2 ++
 arch/x86/kernel/pci-dma.c     |    8 ++++++++
 2 files changed, 10 insertions(+)
arch/x86/include/asm/device.h +2 -0
index 03dd729..10bc628 100644
@@ -13,4 +13,6 @@ struct dev_archdata {
 struct pdev_archdata {
 };
 
+#define ARCH_HAS_DMA_GET_REQUIRED_MASK
+
 #endif /* _ASM_X86_DEVICE_H */
arch/x86/kernel/pci-dma.c +8 -0
index a25e202..5154400 100644
@@ -140,6 +140,14 @@ void dma_generic_free_coherent(struct device *dev, size_t size, void *vaddr,
 		free_pages((unsigned long)vaddr, get_order(size));
 }
 
+u64 dma_get_required_mask(struct device *dev)
+{
+	if (dma_ops->get_required_mask)
+		return dma_ops->get_required_mask(dev);
+	return dma_get_required_mask_from_max_pfn(dev);
+}
+EXPORT_SYMBOL_GPL(dma_get_required_mask);
+
 /*
  * See <Documentation/x86/x86_64/boot-options.txt> for the iommu kernel
  * parameter documentation.
Signature · 1 line
1.7.10.4
#7 lore raw

PATCH v4 4/4 x86/xen: assume a 64-bit DMA mask is required

In reply to David Vrabel

On a Xen PV guest the DMA addresses and physical addresses are not 1:1 (such as Xen PV guests) and the generic dma_get_required_mask() does not return the correct mask (since it uses max_pfn).

Some device drivers (such as mptsas, mpt2sas) use dma_get_required_mask() to set the device's DMA mask to allow them to use only 32-bit DMA addresses in hardware structures. This results in unnecessary use of the SWIOTLB if DMA addresses are more than 32-bits, impacting performance significantly.

We could base the DMA mask on the maximum MFN but:

a) The hypercall op to get the maximum MFN (XENMEM_maximum_ram_page) will truncate the result to an int in 32-bit guests.

b) Future uses of the IOMMU in Xen may map frames at bus addresses above the end of RAM.

So, just assume a 64-bit DMA mask is always required.

---
 arch/x86/xen/pci-swiotlb-xen.c |    1 +
 drivers/xen/swiotlb-xen.c      |    9 +++++++++
 include/xen/swiotlb-xen.h      |    4 ++++
 3 files changed, 14 insertions(+)
arch/x86/xen/pci-swiotlb-xen.c +1 -0
index 0e98e5d..a5d180a 100644
@@ -31,6 +31,7 @@ static struct dma_map_ops xen_swiotlb_dma_ops = {
 	.map_page = xen_swiotlb_map_page,
 	.unmap_page = xen_swiotlb_unmap_page,
 	.dma_supported = xen_swiotlb_dma_supported,
+	.get_required_mask = xen_swiotlb_get_required_mask,
 };
 
 /*
drivers/xen/swiotlb-xen.c +9 -0
index ebd8f21..767d048 100644
@@ -42,9 +42,11 @@
 #include <xen/page.h>
 #include <xen/xen-ops.h>
 #include <xen/hvc-console.h>
+#include <xen/interface/memory.h>
 
 #include <asm/dma-mapping.h>
 #include <asm/xen/page-coherent.h>
+#include <asm/xen/hypercall.h>
 
 #include <trace/events/swiotlb.h>
 /*
@@ -683,3 +685,10 @@ xen_swiotlb_set_dma_mask(struct device *dev, u64 dma_mask)
 	return 0;
 }
 EXPORT_SYMBOL_GPL(xen_swiotlb_set_dma_mask);
+
+u64
+xen_swiotlb_get_required_mask(struct device *dev)
+{
+	return DMA_BIT_MASK(64);
+}
+EXPORT_SYMBOL_GPL(xen_swiotlb_get_required_mask);
include/xen/swiotlb-xen.h +4 -0
index 8b2eb93..6408888 100644
@@ -58,4 +58,8 @@ xen_swiotlb_dma_supported(struct device *hwdev, u64 mask);
 
 extern int
 xen_swiotlb_set_dma_mask(struct device *dev, u64 dma_mask);
+
+extern u64
+xen_swiotlb_get_required_mask(struct device *dev);
+
 #endif /* __LINUX_SWIOTLB_XEN_H */
Signature · 1 line
1.7.10.4
#8 lore raw

PATCH v4 4/4 x86/xen: assume a 64-bit DMA mask is required

In reply to David Vrabel

On a Xen PV guest the DMA addresses and physical addresses are not 1:1 (such as Xen PV guests) and the generic dma_get_required_mask() does not return the correct mask (since it uses max_pfn).

Some device drivers (such as mptsas, mpt2sas) use dma_get_required_mask() to set the device's DMA mask to allow them to use only 32-bit DMA addresses in hardware structures. This results in unnecessary use of the SWIOTLB if DMA addresses are more than 32-bits, impacting performance significantly.

We could base the DMA mask on the maximum MFN but:

a) The hypercall op to get the maximum MFN (XENMEM_maximum_ram_page) will truncate the result to an int in 32-bit guests.

b) Future uses of the IOMMU in Xen may map frames at bus addresses above the end of RAM.

So, just assume a 64-bit DMA mask is always required.

---
 arch/x86/xen/pci-swiotlb-xen.c |    1 +
 drivers/xen/swiotlb-xen.c      |    9 +++++++++
 include/xen/swiotlb-xen.h      |    4 ++++
 3 files changed, 14 insertions(+)
arch/x86/xen/pci-swiotlb-xen.c +1 -0
index 0e98e5d..a5d180a 100644
@@ -31,6 +31,7 @@ static struct dma_map_ops xen_swiotlb_dma_ops = {
 	.map_page = xen_swiotlb_map_page,
 	.unmap_page = xen_swiotlb_unmap_page,
 	.dma_supported = xen_swiotlb_dma_supported,
+	.get_required_mask = xen_swiotlb_get_required_mask,
 };
 
 /*
drivers/xen/swiotlb-xen.c +9 -0
index ebd8f21..767d048 100644
@@ -42,9 +42,11 @@
 #include <xen/page.h>
 #include <xen/xen-ops.h>
 #include <xen/hvc-console.h>
+#include <xen/interface/memory.h>
 
 #include <asm/dma-mapping.h>
 #include <asm/xen/page-coherent.h>
+#include <asm/xen/hypercall.h>
 
 #include <trace/events/swiotlb.h>
 /*
@@ -683,3 +685,10 @@ xen_swiotlb_set_dma_mask(struct device *dev, u64 dma_mask)
 	return 0;
 }
 EXPORT_SYMBOL_GPL(xen_swiotlb_set_dma_mask);
+
+u64
+xen_swiotlb_get_required_mask(struct device *dev)
+{
+	return DMA_BIT_MASK(64);
+}
+EXPORT_SYMBOL_GPL(xen_swiotlb_get_required_mask);
include/xen/swiotlb-xen.h +4 -0
index 8b2eb93..6408888 100644
@@ -58,4 +58,8 @@ xen_swiotlb_dma_supported(struct device *hwdev, u64 mask);
 
 extern int
 xen_swiotlb_set_dma_mask(struct device *dev, u64 dma_mask);
+
+extern u64
+xen_swiotlb_get_required_mask(struct device *dev);
+
 #endif /* __LINUX_SWIOTLB_XEN_H */
Signature · 1 line
1.7.10.4
#9 lore raw

Re: [PATCH 4/4] x86/xen: assume a 64-bit DMA mask is required

Quoted 12 lines · David Vrabel

On a Xen PV guest the DMA addresses and physical addresses are not 1:1 (such as Xen PV guests) and the generic dma_get_required_mask() does not return the correct mask (since it uses max_pfn).

Some device drivers (such as mptsas, mpt2sas) use dma_get_required_mask() to set the device's DMA mask to allow them to use only 32-bit DMA addresses in hardware structures. This results in unnecessary use of the SWIOTLB if DMA addresses are more than 32-bits, impacting performance significantly.

We could base the DMA mask on the maximum MFN but:

Quoted 7 lines

a) The hypercall op to get the maximum MFN (XENMEM_maximum_ram_page) will truncate the result to an int in 32-bit guests.

b) Future uses of the IOMMU in Xen may map frames at bus addresses above the end of RAM.

So, just assume a 64-bit DMA mask is always required.

That's OK for me but at this point I am unsure whether swiotlb-xen is the right place to put this function. It feels more like something that should be introduced in pci-swiotlb-xen.c or arch/x86/xen/mmu.c?

If I were to introduce it on ARM, I would add it to arch/arm/xen/mm.c.

Quoted 68 lines
 arch/x86/xen/pci-swiotlb-xen.c |    1 +
 drivers/xen/swiotlb-xen.c      |    9 +++++++++
 include/xen/swiotlb-xen.h      |    4 ++++
 3 files changed, 14 insertions(+)
arch/x86/xen/pci-swiotlb-xen.c +1 -0
index 0e98e5d..a5d180a 100644
@@ -31,6 +31,7 @@ static struct dma_map_ops xen_swiotlb_dma_ops = {
 	.map_page = xen_swiotlb_map_page,
 	.unmap_page = xen_swiotlb_unmap_page,
 	.dma_supported = xen_swiotlb_dma_supported,
+	.get_required_mask = xen_swiotlb_get_required_mask,
 };
 
 /*
drivers/xen/swiotlb-xen.c +9 -0
index ebd8f21..767d048 100644
@@ -42,9 +42,11 @@
 #include <xen/page.h>
 #include <xen/xen-ops.h>
 #include <xen/hvc-console.h>
+#include <xen/interface/memory.h>
 
 #include <asm/dma-mapping.h>
 #include <asm/xen/page-coherent.h>
+#include <asm/xen/hypercall.h>
 
 #include <trace/events/swiotlb.h>
 /*
@@ -683,3 +685,10 @@ xen_swiotlb_set_dma_mask(struct device *dev, u64 dma_mask)
 	return 0;
 }
 EXPORT_SYMBOL_GPL(xen_swiotlb_set_dma_mask);
+
+u64
+xen_swiotlb_get_required_mask(struct device *dev)
+{
+	return DMA_BIT_MASK(64);
+}
+EXPORT_SYMBOL_GPL(xen_swiotlb_get_required_mask);
include/xen/swiotlb-xen.h +4 -0
index 8b2eb93..6408888 100644
@@ -58,4 +58,8 @@ xen_swiotlb_dma_supported(struct device *hwdev, u64 mask);
 
 extern int
 xen_swiotlb_set_dma_mask(struct device *dev, u64 dma_mask);
+
+extern u64
+xen_swiotlb_get_required_mask(struct device *dev);
+
 #endif /* __LINUX_SWIOTLB_XEN_H */
Signature · 7 lines
1.7.10.4

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://secure-web.cisco.com/1uIZMIeWCQQAJEnGQc6XX-CEjdCVm87GER9t4JEG7wumHKaVAK-6pwSilKvT3RCicfLK8CEakvw_GaO39BTHmmtTqFXSad9_K_6hMXHETF0twK9R1Pdq_kGKFl6aw081mC3L6qVNyOQdb-Hyqn38-_ESRf6rGMDflx4RA3XBEjGA/http%3A%2F%2Fwww.tux.org%2Flkml%2F
#10 lore raw

Re: [PATCH 4/4] x86/xen: assume a 64-bit DMA mask is required

In reply to David Vrabel
Quoted 12 lines · David Vrabel

On a Xen PV guest the DMA addresses and physical addresses are not 1:1 (such as Xen PV guests) and the generic dma_get_required_mask() does not return the correct mask (since it uses max_pfn).

Some device drivers (such as mptsas, mpt2sas) use dma_get_required_mask() to set the device's DMA mask to allow them to use only 32-bit DMA addresses in hardware structures. This results in unnecessary use of the SWIOTLB if DMA addresses are more than 32-bits, impacting performance significantly.

We could base the DMA mask on the maximum MFN but:

Quoted 7 lines

a) The hypercall op to get the maximum MFN (XENMEM_maximum_ram_page) will truncate the result to an int in 32-bit guests.

b) Future uses of the IOMMU in Xen may map frames at bus addresses above the end of RAM.

So, just assume a 64-bit DMA mask is always required.

That's OK for me but at this point I am unsure whether swiotlb-xen is the right place to put this function. It feels more like something that should be introduced in pci-swiotlb-xen.c or arch/x86/xen/mmu.c?

If I were to introduce it on ARM, I would add it to arch/arm/xen/mm.c.

Quoted 68 lines
 arch/x86/xen/pci-swiotlb-xen.c |    1 +
 drivers/xen/swiotlb-xen.c      |    9 +++++++++
 include/xen/swiotlb-xen.h      |    4 ++++
 3 files changed, 14 insertions(+)
arch/x86/xen/pci-swiotlb-xen.c +1 -0
index 0e98e5d..a5d180a 100644
@@ -31,6 +31,7 @@ static struct dma_map_ops xen_swiotlb_dma_ops = {
 	.map_page = xen_swiotlb_map_page,
 	.unmap_page = xen_swiotlb_unmap_page,
 	.dma_supported = xen_swiotlb_dma_supported,
+	.get_required_mask = xen_swiotlb_get_required_mask,
 };
 
 /*
drivers/xen/swiotlb-xen.c +9 -0
index ebd8f21..767d048 100644
@@ -42,9 +42,11 @@
 #include <xen/page.h>
 #include <xen/xen-ops.h>
 #include <xen/hvc-console.h>
+#include <xen/interface/memory.h>
 
 #include <asm/dma-mapping.h>
 #include <asm/xen/page-coherent.h>
+#include <asm/xen/hypercall.h>
 
 #include <trace/events/swiotlb.h>
 /*
@@ -683,3 +685,10 @@ xen_swiotlb_set_dma_mask(struct device *dev, u64 dma_mask)
 	return 0;
 }
 EXPORT_SYMBOL_GPL(xen_swiotlb_set_dma_mask);
+
+u64
+xen_swiotlb_get_required_mask(struct device *dev)
+{
+	return DMA_BIT_MASK(64);
+}
+EXPORT_SYMBOL_GPL(xen_swiotlb_get_required_mask);
include/xen/swiotlb-xen.h +4 -0
index 8b2eb93..6408888 100644
@@ -58,4 +58,8 @@ xen_swiotlb_dma_supported(struct device *hwdev, u64 mask);
 
 extern int
 xen_swiotlb_set_dma_mask(struct device *dev, u64 dma_mask);
+
+extern u64
+xen_swiotlb_get_required_mask(struct device *dev);
+
 #endif /* __LINUX_SWIOTLB_XEN_H */
Signature · 7 lines
1.7.10.4

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://secure-web.cisco.com/1uIZMIeWCQQAJEnGQc6XX-CEjdCVm87GER9t4JEG7wumHKaVAK-6pwSilKvT3RCicfLK8CEakvw_GaO39BTHmmtTqFXSad9_K_6hMXHETF0twK9R1Pdq_kGKFl6aw081mC3L6qVNyOQdb-Hyqn38-_ESRf6rGMDflx4RA3XBEjGA/http%3A%2F%2Fwww.tux.org%2Flkml%2F