* [Xen-devel] a few xen swiotlb cleanups
@ 2019-04-11 7:19 Christoph Hellwig
2019-04-11 7:19 ` [PATCH 1/4] swiotlb-xen: make instances match their method names Christoph Hellwig
` (9 more replies)
0 siblings, 10 replies; 19+ messages in thread
From: Christoph Hellwig @ 2019-04-11 7:19 UTC (permalink / raw)
To: Konrad Rzeszutek Wilk, Boris Ostrovsky, Juergen Gross,
Stefano Stabellini
Cc: xen-devel, iommu
Hi all,
below are a couple of cleanups for swiotlb-xen.c. They were done in
preparation of eventually using the dma-noncoherent.h cache flushing
hooks, but that final goal will need some major work to the arm32 code
first. Until then I think these patches might be better in mainline
than in my local tree so that they don't bitrot.
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel
^ permalink raw reply [flat|nested] 19+ messages in thread
* [PATCH 1/4] swiotlb-xen: make instances match their method names
2019-04-11 7:19 [Xen-devel] a few xen swiotlb cleanups Christoph Hellwig
@ 2019-04-11 7:19 ` Christoph Hellwig
2019-04-11 7:19 ` [Xen-devel] " Christoph Hellwig
` (8 subsequent siblings)
9 siblings, 0 replies; 19+ messages in thread
From: Christoph Hellwig @ 2019-04-11 7:19 UTC (permalink / raw)
To: Konrad Rzeszutek Wilk, Boris Ostrovsky, Juergen Gross,
Stefano Stabellini
Cc: xen-devel, iommu
Just drop two pointless _attrs prefixes to make the code a little
more grep-able.
Signed-off-by: Christoph Hellwig <hch@lst.de>
---
drivers/xen/swiotlb-xen.c | 17 +++++++----------
1 file changed, 7 insertions(+), 10 deletions(-)
diff --git a/drivers/xen/swiotlb-xen.c b/drivers/xen/swiotlb-xen.c
index 877baf2a94f4..d4bc3aabd44d 100644
--- a/drivers/xen/swiotlb-xen.c
+++ b/drivers/xen/swiotlb-xen.c
@@ -504,9 +504,8 @@ xen_swiotlb_sync_single_for_device(struct device *hwdev, dma_addr_t dev_addr,
* concerning calls here are the same as for swiotlb_unmap_page() above.
*/
static void
-xen_swiotlb_unmap_sg_attrs(struct device *hwdev, struct scatterlist *sgl,
- int nelems, enum dma_data_direction dir,
- unsigned long attrs)
+xen_swiotlb_unmap_sg(struct device *hwdev, struct scatterlist *sgl, int nelems,
+ enum dma_data_direction dir, unsigned long attrs)
{
struct scatterlist *sg;
int i;
@@ -535,9 +534,8 @@ xen_swiotlb_unmap_sg_attrs(struct device *hwdev, struct scatterlist *sgl,
* same here.
*/
static int
-xen_swiotlb_map_sg_attrs(struct device *hwdev, struct scatterlist *sgl,
- int nelems, enum dma_data_direction dir,
- unsigned long attrs)
+xen_swiotlb_map_sg(struct device *hwdev, struct scatterlist *sgl, int nelems,
+ enum dma_data_direction dir, unsigned long attrs)
{
struct scatterlist *sg;
int i;
@@ -562,8 +560,7 @@ xen_swiotlb_map_sg_attrs(struct device *hwdev, struct scatterlist *sgl,
/* Don't panic here, we expect map_sg users
to do proper error handling. */
attrs |= DMA_ATTR_SKIP_CPU_SYNC;
- xen_swiotlb_unmap_sg_attrs(hwdev, sgl, i, dir,
- attrs);
+ xen_swiotlb_unmap_sg(hwdev, sgl, i, dir, attrs);
sg_dma_len(sgl) = 0;
return 0;
}
@@ -690,8 +687,8 @@ const struct dma_map_ops xen_swiotlb_dma_ops = {
.sync_single_for_device = xen_swiotlb_sync_single_for_device,
.sync_sg_for_cpu = xen_swiotlb_sync_sg_for_cpu,
.sync_sg_for_device = xen_swiotlb_sync_sg_for_device,
- .map_sg = xen_swiotlb_map_sg_attrs,
- .unmap_sg = xen_swiotlb_unmap_sg_attrs,
+ .map_sg = xen_swiotlb_map_sg,
+ .unmap_sg = xen_swiotlb_unmap_sg,
.map_page = xen_swiotlb_map_page,
.unmap_page = xen_swiotlb_unmap_page,
.dma_supported = xen_swiotlb_dma_supported,
--
2.20.1
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel
^ permalink raw reply related [flat|nested] 19+ messages in thread
* [Xen-devel] [PATCH 1/4] swiotlb-xen: make instances match their method names
2019-04-11 7:19 [Xen-devel] a few xen swiotlb cleanups Christoph Hellwig
2019-04-11 7:19 ` [PATCH 1/4] swiotlb-xen: make instances match their method names Christoph Hellwig
@ 2019-04-11 7:19 ` Christoph Hellwig
2019-04-15 22:55 ` Stefano Stabellini
2019-04-15 22:55 ` Stefano Stabellini
2019-04-11 7:19 ` [Xen-devel] [PATCH 2/4] swiotlb-xen: use ->map_page to implement ->map_sg Christoph Hellwig
` (7 subsequent siblings)
9 siblings, 2 replies; 19+ messages in thread
From: Christoph Hellwig @ 2019-04-11 7:19 UTC (permalink / raw)
To: Konrad Rzeszutek Wilk, Boris Ostrovsky, Juergen Gross,
Stefano Stabellini
Cc: xen-devel, iommu
Just drop two pointless _attrs prefixes to make the code a little
more grep-able.
Signed-off-by: Christoph Hellwig <hch@lst.de>
---
drivers/xen/swiotlb-xen.c | 17 +++++++----------
1 file changed, 7 insertions(+), 10 deletions(-)
diff --git a/drivers/xen/swiotlb-xen.c b/drivers/xen/swiotlb-xen.c
index 877baf2a94f4..d4bc3aabd44d 100644
--- a/drivers/xen/swiotlb-xen.c
+++ b/drivers/xen/swiotlb-xen.c
@@ -504,9 +504,8 @@ xen_swiotlb_sync_single_for_device(struct device *hwdev, dma_addr_t dev_addr,
* concerning calls here are the same as for swiotlb_unmap_page() above.
*/
static void
-xen_swiotlb_unmap_sg_attrs(struct device *hwdev, struct scatterlist *sgl,
- int nelems, enum dma_data_direction dir,
- unsigned long attrs)
+xen_swiotlb_unmap_sg(struct device *hwdev, struct scatterlist *sgl, int nelems,
+ enum dma_data_direction dir, unsigned long attrs)
{
struct scatterlist *sg;
int i;
@@ -535,9 +534,8 @@ xen_swiotlb_unmap_sg_attrs(struct device *hwdev, struct scatterlist *sgl,
* same here.
*/
static int
-xen_swiotlb_map_sg_attrs(struct device *hwdev, struct scatterlist *sgl,
- int nelems, enum dma_data_direction dir,
- unsigned long attrs)
+xen_swiotlb_map_sg(struct device *hwdev, struct scatterlist *sgl, int nelems,
+ enum dma_data_direction dir, unsigned long attrs)
{
struct scatterlist *sg;
int i;
@@ -562,8 +560,7 @@ xen_swiotlb_map_sg_attrs(struct device *hwdev, struct scatterlist *sgl,
/* Don't panic here, we expect map_sg users
to do proper error handling. */
attrs |= DMA_ATTR_SKIP_CPU_SYNC;
- xen_swiotlb_unmap_sg_attrs(hwdev, sgl, i, dir,
- attrs);
+ xen_swiotlb_unmap_sg(hwdev, sgl, i, dir, attrs);
sg_dma_len(sgl) = 0;
return 0;
}
@@ -690,8 +687,8 @@ const struct dma_map_ops xen_swiotlb_dma_ops = {
.sync_single_for_device = xen_swiotlb_sync_single_for_device,
.sync_sg_for_cpu = xen_swiotlb_sync_sg_for_cpu,
.sync_sg_for_device = xen_swiotlb_sync_sg_for_device,
- .map_sg = xen_swiotlb_map_sg_attrs,
- .unmap_sg = xen_swiotlb_unmap_sg_attrs,
+ .map_sg = xen_swiotlb_map_sg,
+ .unmap_sg = xen_swiotlb_unmap_sg,
.map_page = xen_swiotlb_map_page,
.unmap_page = xen_swiotlb_unmap_page,
.dma_supported = xen_swiotlb_dma_supported,
--
2.20.1
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel
^ permalink raw reply related [flat|nested] 19+ messages in thread
* [PATCH 2/4] swiotlb-xen: use ->map_page to implement ->map_sg
2019-04-11 7:19 [Xen-devel] a few xen swiotlb cleanups Christoph Hellwig
` (2 preceding siblings ...)
2019-04-11 7:19 ` [Xen-devel] [PATCH 2/4] swiotlb-xen: use ->map_page to implement ->map_sg Christoph Hellwig
@ 2019-04-11 7:19 ` Christoph Hellwig
2019-04-11 7:19 ` [PATCH 3/4] swiotlb-xen: simplify the DMA sync method implementations Christoph Hellwig
` (5 subsequent siblings)
9 siblings, 0 replies; 19+ messages in thread
From: Christoph Hellwig @ 2019-04-11 7:19 UTC (permalink / raw)
To: Konrad Rzeszutek Wilk, Boris Ostrovsky, Juergen Gross,
Stefano Stabellini
Cc: xen-devel, iommu
We can simply loop over the segments and map them, removing lots of
duplicate code.
Signed-off-by: Christoph Hellwig <hch@lst.de>
---
drivers/xen/swiotlb-xen.c | 68 ++++++---------------------------------
1 file changed, 10 insertions(+), 58 deletions(-)
diff --git a/drivers/xen/swiotlb-xen.c b/drivers/xen/swiotlb-xen.c
index d4bc3aabd44d..97a55c225593 100644
--- a/drivers/xen/swiotlb-xen.c
+++ b/drivers/xen/swiotlb-xen.c
@@ -517,24 +517,8 @@ xen_swiotlb_unmap_sg(struct device *hwdev, struct scatterlist *sgl, int nelems,
}
-/*
- * Map a set of buffers described by scatterlist in streaming mode for DMA.
- * This is the scatter-gather version of the above xen_swiotlb_map_page
- * interface. Here the scatter gather list elements are each tagged with the
- * appropriate dma address and length. They are obtained via
- * sg_dma_{address,length}(SG).
- *
- * NOTE: An implementation may be able to use a smaller number of
- * DMA address/length pairs than there are SG table elements.
- * (for example via virtual mapping capabilities)
- * The routine returns the number of addr/length pairs actually
- * used, at most nents.
- *
- * Device ownership issues as mentioned above for xen_swiotlb_map_page are the
- * same here.
- */
static int
-xen_swiotlb_map_sg(struct device *hwdev, struct scatterlist *sgl, int nelems,
+xen_swiotlb_map_sg(struct device *dev, struct scatterlist *sgl, int nelems,
enum dma_data_direction dir, unsigned long attrs)
{
struct scatterlist *sg;
@@ -543,50 +527,18 @@ xen_swiotlb_map_sg(struct device *hwdev, struct scatterlist *sgl, int nelems,
BUG_ON(dir == DMA_NONE);
for_each_sg(sgl, sg, nelems, i) {
- phys_addr_t paddr = sg_phys(sg);
- dma_addr_t dev_addr = xen_phys_to_bus(paddr);
-
- if (swiotlb_force == SWIOTLB_FORCE ||
- xen_arch_need_swiotlb(hwdev, paddr, dev_addr) ||
- !dma_capable(hwdev, dev_addr, sg->length) ||
- range_straddles_page_boundary(paddr, sg->length)) {
- phys_addr_t map = swiotlb_tbl_map_single(hwdev,
- start_dma_addr,
- sg_phys(sg),
- sg->length,
- dir, attrs);
- if (map == DMA_MAPPING_ERROR) {
- dev_warn(hwdev, "swiotlb buffer is full\n");
- /* Don't panic here, we expect map_sg users
- to do proper error handling. */
- attrs |= DMA_ATTR_SKIP_CPU_SYNC;
- xen_swiotlb_unmap_sg(hwdev, sgl, i, dir, attrs);
- sg_dma_len(sgl) = 0;
- return 0;
- }
- dev_addr = xen_phys_to_bus(map);
- xen_dma_map_page(hwdev, pfn_to_page(map >> PAGE_SHIFT),
- dev_addr,
- map & ~PAGE_MASK,
- sg->length,
- dir,
- attrs);
- sg->dma_address = dev_addr;
- } else {
- /* we are not interested in the dma_addr returned by
- * xen_dma_map_page, only in the potential cache flushes executed
- * by the function. */
- xen_dma_map_page(hwdev, pfn_to_page(paddr >> PAGE_SHIFT),
- dev_addr,
- paddr & ~PAGE_MASK,
- sg->length,
- dir,
- attrs);
- sg->dma_address = dev_addr;
- }
+ sg->dma_address = xen_swiotlb_map_page(dev, sg_page(sg),
+ sg->offset, sg->length, dir, attrs);
+ if (sg->dma_address == DMA_MAPPING_ERROR)
+ goto out_unmap;
sg_dma_len(sg) = sg->length;
}
+
return nelems;
+out_unmap:
+ xen_swiotlb_unmap_sg(dev, sgl, i, dir, attrs | DMA_ATTR_SKIP_CPU_SYNC);
+ sg_dma_len(sgl) = 0;
+ return 0;
}
/*
--
2.20.1
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel
^ permalink raw reply related [flat|nested] 19+ messages in thread
* [Xen-devel] [PATCH 2/4] swiotlb-xen: use ->map_page to implement ->map_sg
2019-04-11 7:19 [Xen-devel] a few xen swiotlb cleanups Christoph Hellwig
2019-04-11 7:19 ` [PATCH 1/4] swiotlb-xen: make instances match their method names Christoph Hellwig
2019-04-11 7:19 ` [Xen-devel] " Christoph Hellwig
@ 2019-04-11 7:19 ` Christoph Hellwig
2019-04-15 22:55 ` Stefano Stabellini
2019-04-15 22:55 ` [Xen-devel] " Stefano Stabellini
2019-04-11 7:19 ` Christoph Hellwig
` (6 subsequent siblings)
9 siblings, 2 replies; 19+ messages in thread
From: Christoph Hellwig @ 2019-04-11 7:19 UTC (permalink / raw)
To: Konrad Rzeszutek Wilk, Boris Ostrovsky, Juergen Gross,
Stefano Stabellini
Cc: xen-devel, iommu
We can simply loop over the segments and map them, removing lots of
duplicate code.
Signed-off-by: Christoph Hellwig <hch@lst.de>
---
drivers/xen/swiotlb-xen.c | 68 ++++++---------------------------------
1 file changed, 10 insertions(+), 58 deletions(-)
diff --git a/drivers/xen/swiotlb-xen.c b/drivers/xen/swiotlb-xen.c
index d4bc3aabd44d..97a55c225593 100644
--- a/drivers/xen/swiotlb-xen.c
+++ b/drivers/xen/swiotlb-xen.c
@@ -517,24 +517,8 @@ xen_swiotlb_unmap_sg(struct device *hwdev, struct scatterlist *sgl, int nelems,
}
-/*
- * Map a set of buffers described by scatterlist in streaming mode for DMA.
- * This is the scatter-gather version of the above xen_swiotlb_map_page
- * interface. Here the scatter gather list elements are each tagged with the
- * appropriate dma address and length. They are obtained via
- * sg_dma_{address,length}(SG).
- *
- * NOTE: An implementation may be able to use a smaller number of
- * DMA address/length pairs than there are SG table elements.
- * (for example via virtual mapping capabilities)
- * The routine returns the number of addr/length pairs actually
- * used, at most nents.
- *
- * Device ownership issues as mentioned above for xen_swiotlb_map_page are the
- * same here.
- */
static int
-xen_swiotlb_map_sg(struct device *hwdev, struct scatterlist *sgl, int nelems,
+xen_swiotlb_map_sg(struct device *dev, struct scatterlist *sgl, int nelems,
enum dma_data_direction dir, unsigned long attrs)
{
struct scatterlist *sg;
@@ -543,50 +527,18 @@ xen_swiotlb_map_sg(struct device *hwdev, struct scatterlist *sgl, int nelems,
BUG_ON(dir == DMA_NONE);
for_each_sg(sgl, sg, nelems, i) {
- phys_addr_t paddr = sg_phys(sg);
- dma_addr_t dev_addr = xen_phys_to_bus(paddr);
-
- if (swiotlb_force == SWIOTLB_FORCE ||
- xen_arch_need_swiotlb(hwdev, paddr, dev_addr) ||
- !dma_capable(hwdev, dev_addr, sg->length) ||
- range_straddles_page_boundary(paddr, sg->length)) {
- phys_addr_t map = swiotlb_tbl_map_single(hwdev,
- start_dma_addr,
- sg_phys(sg),
- sg->length,
- dir, attrs);
- if (map == DMA_MAPPING_ERROR) {
- dev_warn(hwdev, "swiotlb buffer is full\n");
- /* Don't panic here, we expect map_sg users
- to do proper error handling. */
- attrs |= DMA_ATTR_SKIP_CPU_SYNC;
- xen_swiotlb_unmap_sg(hwdev, sgl, i, dir, attrs);
- sg_dma_len(sgl) = 0;
- return 0;
- }
- dev_addr = xen_phys_to_bus(map);
- xen_dma_map_page(hwdev, pfn_to_page(map >> PAGE_SHIFT),
- dev_addr,
- map & ~PAGE_MASK,
- sg->length,
- dir,
- attrs);
- sg->dma_address = dev_addr;
- } else {
- /* we are not interested in the dma_addr returned by
- * xen_dma_map_page, only in the potential cache flushes executed
- * by the function. */
- xen_dma_map_page(hwdev, pfn_to_page(paddr >> PAGE_SHIFT),
- dev_addr,
- paddr & ~PAGE_MASK,
- sg->length,
- dir,
- attrs);
- sg->dma_address = dev_addr;
- }
+ sg->dma_address = xen_swiotlb_map_page(dev, sg_page(sg),
+ sg->offset, sg->length, dir, attrs);
+ if (sg->dma_address == DMA_MAPPING_ERROR)
+ goto out_unmap;
sg_dma_len(sg) = sg->length;
}
+
return nelems;
+out_unmap:
+ xen_swiotlb_unmap_sg(dev, sgl, i, dir, attrs | DMA_ATTR_SKIP_CPU_SYNC);
+ sg_dma_len(sgl) = 0;
+ return 0;
}
/*
--
2.20.1
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel
^ permalink raw reply related [flat|nested] 19+ messages in thread
* [PATCH 3/4] swiotlb-xen: simplify the DMA sync method implementations
2019-04-11 7:19 [Xen-devel] a few xen swiotlb cleanups Christoph Hellwig
` (3 preceding siblings ...)
2019-04-11 7:19 ` Christoph Hellwig
@ 2019-04-11 7:19 ` Christoph Hellwig
2019-04-11 7:19 ` [Xen-devel] " Christoph Hellwig
` (4 subsequent siblings)
9 siblings, 0 replies; 19+ messages in thread
From: Christoph Hellwig @ 2019-04-11 7:19 UTC (permalink / raw)
To: Konrad Rzeszutek Wilk, Boris Ostrovsky, Juergen Gross,
Stefano Stabellini
Cc: xen-devel, iommu
Get rid of the grand multiplexer and implement the sync_single_for_cpu
and sync_single_for_device methods directly, and then loop over them
for the scatterlist based variants.
Note that this also loses a few comments related to highlevel DMA API
concepts, which have nothing to do with the swiotlb-xen implementation
details.
Signed-off-by: Christoph Hellwig <hch@lst.de>
---
drivers/xen/swiotlb-xen.c | 84 +++++++++++++--------------------------
1 file changed, 28 insertions(+), 56 deletions(-)
diff --git a/drivers/xen/swiotlb-xen.c b/drivers/xen/swiotlb-xen.c
index 97a55c225593..9a951504dc12 100644
--- a/drivers/xen/swiotlb-xen.c
+++ b/drivers/xen/swiotlb-xen.c
@@ -455,48 +455,28 @@ static void xen_swiotlb_unmap_page(struct device *hwdev, dma_addr_t dev_addr,
xen_unmap_single(hwdev, dev_addr, size, dir, attrs);
}
-/*
- * Make physical memory consistent for a single streaming mode DMA translation
- * after a transfer.
- *
- * If you perform a xen_swiotlb_map_page() but wish to interrogate the buffer
- * using the cpu, yet do not wish to teardown the dma mapping, you must
- * call this function before doing so. At the next point you give the dma
- * address back to the card, you must first perform a
- * xen_swiotlb_dma_sync_for_device, and then the device again owns the buffer
- */
static void
-xen_swiotlb_sync_single(struct device *hwdev, dma_addr_t dev_addr,
- size_t size, enum dma_data_direction dir,
- enum dma_sync_target target)
+xen_swiotlb_sync_single_for_cpu(struct device *dev, dma_addr_t dma_addr,
+ size_t size, enum dma_data_direction dir)
{
- phys_addr_t paddr = xen_bus_to_phys(dev_addr);
+ phys_addr_t paddr = xen_bus_to_phys(dma_addr);
- BUG_ON(dir == DMA_NONE);
-
- if (target == SYNC_FOR_CPU)
- xen_dma_sync_single_for_cpu(hwdev, dev_addr, size, dir);
+ xen_dma_sync_single_for_cpu(dev, dma_addr, size, dir);
- /* NOTE: We use dev_addr here, not paddr! */
- if (is_xen_swiotlb_buffer(dev_addr))
- swiotlb_tbl_sync_single(hwdev, paddr, size, dir, target);
-
- if (target == SYNC_FOR_DEVICE)
- xen_dma_sync_single_for_device(hwdev, dev_addr, size, dir);
+ if (is_xen_swiotlb_buffer(dma_addr))
+ swiotlb_tbl_sync_single(dev, paddr, size, dir, SYNC_FOR_CPU);
}
-void
-xen_swiotlb_sync_single_for_cpu(struct device *hwdev, dma_addr_t dev_addr,
- size_t size, enum dma_data_direction dir)
+static void
+xen_swiotlb_sync_single_for_device(struct device *dev, dma_addr_t dma_addr,
+ size_t size, enum dma_data_direction dir)
{
- xen_swiotlb_sync_single(hwdev, dev_addr, size, dir, SYNC_FOR_CPU);
-}
+ phys_addr_t paddr = xen_bus_to_phys(dma_addr);
-void
-xen_swiotlb_sync_single_for_device(struct device *hwdev, dma_addr_t dev_addr,
- size_t size, enum dma_data_direction dir)
-{
- xen_swiotlb_sync_single(hwdev, dev_addr, size, dir, SYNC_FOR_DEVICE);
+ if (is_xen_swiotlb_buffer(dma_addr))
+ swiotlb_tbl_sync_single(dev, paddr, size, dir, SYNC_FOR_DEVICE);
+
+ xen_dma_sync_single_for_device(dev, dma_addr, size, dir);
}
/*
@@ -541,38 +521,30 @@ xen_swiotlb_map_sg(struct device *dev, struct scatterlist *sgl, int nelems,
return 0;
}
-/*
- * Make physical memory consistent for a set of streaming mode DMA translations
- * after a transfer.
- *
- * The same as swiotlb_sync_single_* but for a scatter-gather list, same rules
- * and usage.
- */
static void
-xen_swiotlb_sync_sg(struct device *hwdev, struct scatterlist *sgl,
- int nelems, enum dma_data_direction dir,
- enum dma_sync_target target)
+xen_swiotlb_sync_sg_for_cpu(struct device *dev, struct scatterlist *sgl,
+ int nelems, enum dma_data_direction dir)
{
struct scatterlist *sg;
int i;
- for_each_sg(sgl, sg, nelems, i)
- xen_swiotlb_sync_single(hwdev, sg->dma_address,
- sg_dma_len(sg), dir, target);
-}
-
-static void
-xen_swiotlb_sync_sg_for_cpu(struct device *hwdev, struct scatterlist *sg,
- int nelems, enum dma_data_direction dir)
-{
- xen_swiotlb_sync_sg(hwdev, sg, nelems, dir, SYNC_FOR_CPU);
+ for_each_sg(sgl, sg, nelems, i) {
+ xen_swiotlb_sync_single_for_cpu(dev, sg->dma_address,
+ sg->length, dir);
+ }
}
static void
-xen_swiotlb_sync_sg_for_device(struct device *hwdev, struct scatterlist *sg,
+xen_swiotlb_sync_sg_for_device(struct device *dev, struct scatterlist *sgl,
int nelems, enum dma_data_direction dir)
{
- xen_swiotlb_sync_sg(hwdev, sg, nelems, dir, SYNC_FOR_DEVICE);
+ struct scatterlist *sg;
+ int i;
+
+ for_each_sg(sgl, sg, nelems, i) {
+ xen_swiotlb_sync_single_for_device(dev, sg->dma_address,
+ sg->length, dir);
+ }
}
/*
--
2.20.1
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel
^ permalink raw reply related [flat|nested] 19+ messages in thread
* [Xen-devel] [PATCH 3/4] swiotlb-xen: simplify the DMA sync method implementations
2019-04-11 7:19 [Xen-devel] a few xen swiotlb cleanups Christoph Hellwig
` (4 preceding siblings ...)
2019-04-11 7:19 ` [PATCH 3/4] swiotlb-xen: simplify the DMA sync method implementations Christoph Hellwig
@ 2019-04-11 7:19 ` Christoph Hellwig
2019-04-15 22:55 ` Stefano Stabellini
2019-04-15 22:55 ` [Xen-devel] " Stefano Stabellini
2019-04-11 7:20 ` [Xen-devel] [PATCH 4/4] swiotlb-xen: ensure we have a single callsite for xen_dma_map_page Christoph Hellwig
` (3 subsequent siblings)
9 siblings, 2 replies; 19+ messages in thread
From: Christoph Hellwig @ 2019-04-11 7:19 UTC (permalink / raw)
To: Konrad Rzeszutek Wilk, Boris Ostrovsky, Juergen Gross,
Stefano Stabellini
Cc: xen-devel, iommu
Get rid of the grand multiplexer and implement the sync_single_for_cpu
and sync_single_for_device methods directly, and then loop over them
for the scatterlist based variants.
Note that this also loses a few comments related to highlevel DMA API
concepts, which have nothing to do with the swiotlb-xen implementation
details.
Signed-off-by: Christoph Hellwig <hch@lst.de>
---
drivers/xen/swiotlb-xen.c | 84 +++++++++++++--------------------------
1 file changed, 28 insertions(+), 56 deletions(-)
diff --git a/drivers/xen/swiotlb-xen.c b/drivers/xen/swiotlb-xen.c
index 97a55c225593..9a951504dc12 100644
--- a/drivers/xen/swiotlb-xen.c
+++ b/drivers/xen/swiotlb-xen.c
@@ -455,48 +455,28 @@ static void xen_swiotlb_unmap_page(struct device *hwdev, dma_addr_t dev_addr,
xen_unmap_single(hwdev, dev_addr, size, dir, attrs);
}
-/*
- * Make physical memory consistent for a single streaming mode DMA translation
- * after a transfer.
- *
- * If you perform a xen_swiotlb_map_page() but wish to interrogate the buffer
- * using the cpu, yet do not wish to teardown the dma mapping, you must
- * call this function before doing so. At the next point you give the dma
- * address back to the card, you must first perform a
- * xen_swiotlb_dma_sync_for_device, and then the device again owns the buffer
- */
static void
-xen_swiotlb_sync_single(struct device *hwdev, dma_addr_t dev_addr,
- size_t size, enum dma_data_direction dir,
- enum dma_sync_target target)
+xen_swiotlb_sync_single_for_cpu(struct device *dev, dma_addr_t dma_addr,
+ size_t size, enum dma_data_direction dir)
{
- phys_addr_t paddr = xen_bus_to_phys(dev_addr);
+ phys_addr_t paddr = xen_bus_to_phys(dma_addr);
- BUG_ON(dir == DMA_NONE);
-
- if (target == SYNC_FOR_CPU)
- xen_dma_sync_single_for_cpu(hwdev, dev_addr, size, dir);
+ xen_dma_sync_single_for_cpu(dev, dma_addr, size, dir);
- /* NOTE: We use dev_addr here, not paddr! */
- if (is_xen_swiotlb_buffer(dev_addr))
- swiotlb_tbl_sync_single(hwdev, paddr, size, dir, target);
-
- if (target == SYNC_FOR_DEVICE)
- xen_dma_sync_single_for_device(hwdev, dev_addr, size, dir);
+ if (is_xen_swiotlb_buffer(dma_addr))
+ swiotlb_tbl_sync_single(dev, paddr, size, dir, SYNC_FOR_CPU);
}
-void
-xen_swiotlb_sync_single_for_cpu(struct device *hwdev, dma_addr_t dev_addr,
- size_t size, enum dma_data_direction dir)
+static void
+xen_swiotlb_sync_single_for_device(struct device *dev, dma_addr_t dma_addr,
+ size_t size, enum dma_data_direction dir)
{
- xen_swiotlb_sync_single(hwdev, dev_addr, size, dir, SYNC_FOR_CPU);
-}
+ phys_addr_t paddr = xen_bus_to_phys(dma_addr);
-void
-xen_swiotlb_sync_single_for_device(struct device *hwdev, dma_addr_t dev_addr,
- size_t size, enum dma_data_direction dir)
-{
- xen_swiotlb_sync_single(hwdev, dev_addr, size, dir, SYNC_FOR_DEVICE);
+ if (is_xen_swiotlb_buffer(dma_addr))
+ swiotlb_tbl_sync_single(dev, paddr, size, dir, SYNC_FOR_DEVICE);
+
+ xen_dma_sync_single_for_device(dev, dma_addr, size, dir);
}
/*
@@ -541,38 +521,30 @@ xen_swiotlb_map_sg(struct device *dev, struct scatterlist *sgl, int nelems,
return 0;
}
-/*
- * Make physical memory consistent for a set of streaming mode DMA translations
- * after a transfer.
- *
- * The same as swiotlb_sync_single_* but for a scatter-gather list, same rules
- * and usage.
- */
static void
-xen_swiotlb_sync_sg(struct device *hwdev, struct scatterlist *sgl,
- int nelems, enum dma_data_direction dir,
- enum dma_sync_target target)
+xen_swiotlb_sync_sg_for_cpu(struct device *dev, struct scatterlist *sgl,
+ int nelems, enum dma_data_direction dir)
{
struct scatterlist *sg;
int i;
- for_each_sg(sgl, sg, nelems, i)
- xen_swiotlb_sync_single(hwdev, sg->dma_address,
- sg_dma_len(sg), dir, target);
-}
-
-static void
-xen_swiotlb_sync_sg_for_cpu(struct device *hwdev, struct scatterlist *sg,
- int nelems, enum dma_data_direction dir)
-{
- xen_swiotlb_sync_sg(hwdev, sg, nelems, dir, SYNC_FOR_CPU);
+ for_each_sg(sgl, sg, nelems, i) {
+ xen_swiotlb_sync_single_for_cpu(dev, sg->dma_address,
+ sg->length, dir);
+ }
}
static void
-xen_swiotlb_sync_sg_for_device(struct device *hwdev, struct scatterlist *sg,
+xen_swiotlb_sync_sg_for_device(struct device *dev, struct scatterlist *sgl,
int nelems, enum dma_data_direction dir)
{
- xen_swiotlb_sync_sg(hwdev, sg, nelems, dir, SYNC_FOR_DEVICE);
+ struct scatterlist *sg;
+ int i;
+
+ for_each_sg(sgl, sg, nelems, i) {
+ xen_swiotlb_sync_single_for_device(dev, sg->dma_address,
+ sg->length, dir);
+ }
}
/*
--
2.20.1
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel
^ permalink raw reply related [flat|nested] 19+ messages in thread
* [PATCH 4/4] swiotlb-xen: ensure we have a single callsite for xen_dma_map_page
2019-04-11 7:19 [Xen-devel] a few xen swiotlb cleanups Christoph Hellwig
` (6 preceding siblings ...)
2019-04-11 7:20 ` [Xen-devel] [PATCH 4/4] swiotlb-xen: ensure we have a single callsite for xen_dma_map_page Christoph Hellwig
@ 2019-04-11 7:20 ` Christoph Hellwig
2019-04-25 18:55 ` a few xen swiotlb cleanups Konrad Rzeszutek Wilk
2019-04-25 18:55 ` [Xen-devel] " Konrad Rzeszutek Wilk
9 siblings, 0 replies; 19+ messages in thread
From: Christoph Hellwig @ 2019-04-11 7:20 UTC (permalink / raw)
To: Konrad Rzeszutek Wilk, Boris Ostrovsky, Juergen Gross,
Stefano Stabellini
Cc: xen-devel, iommu
Refactor the code a bit to make further changes easier.
Signed-off-by: Christoph Hellwig <hch@lst.de>
---
drivers/xen/swiotlb-xen.c | 31 ++++++++++++++++---------------
1 file changed, 16 insertions(+), 15 deletions(-)
diff --git a/drivers/xen/swiotlb-xen.c b/drivers/xen/swiotlb-xen.c
index 9a951504dc12..5dcb06fe9667 100644
--- a/drivers/xen/swiotlb-xen.c
+++ b/drivers/xen/swiotlb-xen.c
@@ -391,13 +391,8 @@ static dma_addr_t xen_swiotlb_map_page(struct device *dev, struct page *page,
if (dma_capable(dev, dev_addr, size) &&
!range_straddles_page_boundary(phys, size) &&
!xen_arch_need_swiotlb(dev, phys, dev_addr) &&
- (swiotlb_force != SWIOTLB_FORCE)) {
- /* we are not interested in the dma_addr returned by
- * xen_dma_map_page, only in the potential cache flushes executed
- * by the function. */
- xen_dma_map_page(dev, page, dev_addr, offset, size, dir, attrs);
- return dev_addr;
- }
+ swiotlb_force != SWIOTLB_FORCE)
+ goto done;
/*
* Oh well, have to allocate and map a bounce buffer.
@@ -410,19 +405,25 @@ static dma_addr_t xen_swiotlb_map_page(struct device *dev, struct page *page,
return DMA_MAPPING_ERROR;
dev_addr = xen_phys_to_bus(map);
- xen_dma_map_page(dev, pfn_to_page(map >> PAGE_SHIFT),
- dev_addr, map & ~PAGE_MASK, size, dir, attrs);
/*
* Ensure that the address returned is DMA'ble
*/
- if (dma_capable(dev, dev_addr, size))
- return dev_addr;
-
- attrs |= DMA_ATTR_SKIP_CPU_SYNC;
- swiotlb_tbl_unmap_single(dev, map, size, dir, attrs);
+ if (unlikely(!dma_capable(dev, dev_addr, size))) {
+ swiotlb_tbl_unmap_single(dev, map, size, dir,
+ attrs | DMA_ATTR_SKIP_CPU_SYNC);
+ return DMA_MAPPING_ERROR;
+ }
- return DMA_MAPPING_ERROR;
+ page = pfn_to_page(map >> PAGE_SHIFT);
+ offset = map & ~PAGE_MASK;
+done:
+ /*
+ * we are not interested in the dma_addr returned by xen_dma_map_page,
+ * only in the potential cache flushes executed by the function.
+ */
+ xen_dma_map_page(dev, page, dev_addr, offset, size, dir, attrs);
+ return dev_addr;
}
/*
--
2.20.1
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel
^ permalink raw reply related [flat|nested] 19+ messages in thread
* [Xen-devel] [PATCH 4/4] swiotlb-xen: ensure we have a single callsite for xen_dma_map_page
2019-04-11 7:19 [Xen-devel] a few xen swiotlb cleanups Christoph Hellwig
` (5 preceding siblings ...)
2019-04-11 7:19 ` [Xen-devel] " Christoph Hellwig
@ 2019-04-11 7:20 ` Christoph Hellwig
2019-04-15 22:56 ` Stefano Stabellini
2019-04-15 22:56 ` Stefano Stabellini
2019-04-11 7:20 ` Christoph Hellwig
` (2 subsequent siblings)
9 siblings, 2 replies; 19+ messages in thread
From: Christoph Hellwig @ 2019-04-11 7:20 UTC (permalink / raw)
To: Konrad Rzeszutek Wilk, Boris Ostrovsky, Juergen Gross,
Stefano Stabellini
Cc: xen-devel, iommu
Refactor the code a bit to make further changes easier.
Signed-off-by: Christoph Hellwig <hch@lst.de>
---
drivers/xen/swiotlb-xen.c | 31 ++++++++++++++++---------------
1 file changed, 16 insertions(+), 15 deletions(-)
diff --git a/drivers/xen/swiotlb-xen.c b/drivers/xen/swiotlb-xen.c
index 9a951504dc12..5dcb06fe9667 100644
--- a/drivers/xen/swiotlb-xen.c
+++ b/drivers/xen/swiotlb-xen.c
@@ -391,13 +391,8 @@ static dma_addr_t xen_swiotlb_map_page(struct device *dev, struct page *page,
if (dma_capable(dev, dev_addr, size) &&
!range_straddles_page_boundary(phys, size) &&
!xen_arch_need_swiotlb(dev, phys, dev_addr) &&
- (swiotlb_force != SWIOTLB_FORCE)) {
- /* we are not interested in the dma_addr returned by
- * xen_dma_map_page, only in the potential cache flushes executed
- * by the function. */
- xen_dma_map_page(dev, page, dev_addr, offset, size, dir, attrs);
- return dev_addr;
- }
+ swiotlb_force != SWIOTLB_FORCE)
+ goto done;
/*
* Oh well, have to allocate and map a bounce buffer.
@@ -410,19 +405,25 @@ static dma_addr_t xen_swiotlb_map_page(struct device *dev, struct page *page,
return DMA_MAPPING_ERROR;
dev_addr = xen_phys_to_bus(map);
- xen_dma_map_page(dev, pfn_to_page(map >> PAGE_SHIFT),
- dev_addr, map & ~PAGE_MASK, size, dir, attrs);
/*
* Ensure that the address returned is DMA'ble
*/
- if (dma_capable(dev, dev_addr, size))
- return dev_addr;
-
- attrs |= DMA_ATTR_SKIP_CPU_SYNC;
- swiotlb_tbl_unmap_single(dev, map, size, dir, attrs);
+ if (unlikely(!dma_capable(dev, dev_addr, size))) {
+ swiotlb_tbl_unmap_single(dev, map, size, dir,
+ attrs | DMA_ATTR_SKIP_CPU_SYNC);
+ return DMA_MAPPING_ERROR;
+ }
- return DMA_MAPPING_ERROR;
+ page = pfn_to_page(map >> PAGE_SHIFT);
+ offset = map & ~PAGE_MASK;
+done:
+ /*
+ * we are not interested in the dma_addr returned by xen_dma_map_page,
+ * only in the potential cache flushes executed by the function.
+ */
+ xen_dma_map_page(dev, page, dev_addr, offset, size, dir, attrs);
+ return dev_addr;
}
/*
--
2.20.1
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel
^ permalink raw reply related [flat|nested] 19+ messages in thread
* Re: [PATCH 1/4] swiotlb-xen: make instances match their method names
2019-04-11 7:19 ` [Xen-devel] " Christoph Hellwig
2019-04-15 22:55 ` Stefano Stabellini
@ 2019-04-15 22:55 ` Stefano Stabellini
1 sibling, 0 replies; 19+ messages in thread
From: Stefano Stabellini @ 2019-04-15 22:55 UTC (permalink / raw)
To: Christoph Hellwig
Cc: Juergen Gross, Stefano Stabellini, Konrad Rzeszutek Wilk, iommu,
xen-devel, Boris Ostrovsky
On Thu, 11 Apr 2019, Christoph Hellwig wrote:
> Just drop two pointless _attrs prefixes to make the code a little
> more grep-able.
>
> Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
> ---
> drivers/xen/swiotlb-xen.c | 17 +++++++----------
> 1 file changed, 7 insertions(+), 10 deletions(-)
>
> diff --git a/drivers/xen/swiotlb-xen.c b/drivers/xen/swiotlb-xen.c
> index 877baf2a94f4..d4bc3aabd44d 100644
> --- a/drivers/xen/swiotlb-xen.c
> +++ b/drivers/xen/swiotlb-xen.c
> @@ -504,9 +504,8 @@ xen_swiotlb_sync_single_for_device(struct device *hwdev, dma_addr_t dev_addr,
> * concerning calls here are the same as for swiotlb_unmap_page() above.
> */
> static void
> -xen_swiotlb_unmap_sg_attrs(struct device *hwdev, struct scatterlist *sgl,
> - int nelems, enum dma_data_direction dir,
> - unsigned long attrs)
> +xen_swiotlb_unmap_sg(struct device *hwdev, struct scatterlist *sgl, int nelems,
> + enum dma_data_direction dir, unsigned long attrs)
> {
> struct scatterlist *sg;
> int i;
> @@ -535,9 +534,8 @@ xen_swiotlb_unmap_sg_attrs(struct device *hwdev, struct scatterlist *sgl,
> * same here.
> */
> static int
> -xen_swiotlb_map_sg_attrs(struct device *hwdev, struct scatterlist *sgl,
> - int nelems, enum dma_data_direction dir,
> - unsigned long attrs)
> +xen_swiotlb_map_sg(struct device *hwdev, struct scatterlist *sgl, int nelems,
> + enum dma_data_direction dir, unsigned long attrs)
> {
> struct scatterlist *sg;
> int i;
> @@ -562,8 +560,7 @@ xen_swiotlb_map_sg_attrs(struct device *hwdev, struct scatterlist *sgl,
> /* Don't panic here, we expect map_sg users
> to do proper error handling. */
> attrs |= DMA_ATTR_SKIP_CPU_SYNC;
> - xen_swiotlb_unmap_sg_attrs(hwdev, sgl, i, dir,
> - attrs);
> + xen_swiotlb_unmap_sg(hwdev, sgl, i, dir, attrs);
> sg_dma_len(sgl) = 0;
> return 0;
> }
> @@ -690,8 +687,8 @@ const struct dma_map_ops xen_swiotlb_dma_ops = {
> .sync_single_for_device = xen_swiotlb_sync_single_for_device,
> .sync_sg_for_cpu = xen_swiotlb_sync_sg_for_cpu,
> .sync_sg_for_device = xen_swiotlb_sync_sg_for_device,
> - .map_sg = xen_swiotlb_map_sg_attrs,
> - .unmap_sg = xen_swiotlb_unmap_sg_attrs,
> + .map_sg = xen_swiotlb_map_sg,
> + .unmap_sg = xen_swiotlb_unmap_sg,
> .map_page = xen_swiotlb_map_page,
> .unmap_page = xen_swiotlb_unmap_page,
> .dma_supported = xen_swiotlb_dma_supported,
> --
> 2.20.1
>
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [Xen-devel] [PATCH 1/4] swiotlb-xen: make instances match their method names
2019-04-11 7:19 ` [Xen-devel] " Christoph Hellwig
@ 2019-04-15 22:55 ` Stefano Stabellini
2019-04-15 22:55 ` Stefano Stabellini
1 sibling, 0 replies; 19+ messages in thread
From: Stefano Stabellini @ 2019-04-15 22:55 UTC (permalink / raw)
To: Christoph Hellwig
Cc: Juergen Gross, Stefano Stabellini, Konrad Rzeszutek Wilk, iommu,
xen-devel, Boris Ostrovsky
On Thu, 11 Apr 2019, Christoph Hellwig wrote:
> Just drop two pointless _attrs prefixes to make the code a little
> more grep-able.
>
> Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
> ---
> drivers/xen/swiotlb-xen.c | 17 +++++++----------
> 1 file changed, 7 insertions(+), 10 deletions(-)
>
> diff --git a/drivers/xen/swiotlb-xen.c b/drivers/xen/swiotlb-xen.c
> index 877baf2a94f4..d4bc3aabd44d 100644
> --- a/drivers/xen/swiotlb-xen.c
> +++ b/drivers/xen/swiotlb-xen.c
> @@ -504,9 +504,8 @@ xen_swiotlb_sync_single_for_device(struct device *hwdev, dma_addr_t dev_addr,
> * concerning calls here are the same as for swiotlb_unmap_page() above.
> */
> static void
> -xen_swiotlb_unmap_sg_attrs(struct device *hwdev, struct scatterlist *sgl,
> - int nelems, enum dma_data_direction dir,
> - unsigned long attrs)
> +xen_swiotlb_unmap_sg(struct device *hwdev, struct scatterlist *sgl, int nelems,
> + enum dma_data_direction dir, unsigned long attrs)
> {
> struct scatterlist *sg;
> int i;
> @@ -535,9 +534,8 @@ xen_swiotlb_unmap_sg_attrs(struct device *hwdev, struct scatterlist *sgl,
> * same here.
> */
> static int
> -xen_swiotlb_map_sg_attrs(struct device *hwdev, struct scatterlist *sgl,
> - int nelems, enum dma_data_direction dir,
> - unsigned long attrs)
> +xen_swiotlb_map_sg(struct device *hwdev, struct scatterlist *sgl, int nelems,
> + enum dma_data_direction dir, unsigned long attrs)
> {
> struct scatterlist *sg;
> int i;
> @@ -562,8 +560,7 @@ xen_swiotlb_map_sg_attrs(struct device *hwdev, struct scatterlist *sgl,
> /* Don't panic here, we expect map_sg users
> to do proper error handling. */
> attrs |= DMA_ATTR_SKIP_CPU_SYNC;
> - xen_swiotlb_unmap_sg_attrs(hwdev, sgl, i, dir,
> - attrs);
> + xen_swiotlb_unmap_sg(hwdev, sgl, i, dir, attrs);
> sg_dma_len(sgl) = 0;
> return 0;
> }
> @@ -690,8 +687,8 @@ const struct dma_map_ops xen_swiotlb_dma_ops = {
> .sync_single_for_device = xen_swiotlb_sync_single_for_device,
> .sync_sg_for_cpu = xen_swiotlb_sync_sg_for_cpu,
> .sync_sg_for_device = xen_swiotlb_sync_sg_for_device,
> - .map_sg = xen_swiotlb_map_sg_attrs,
> - .unmap_sg = xen_swiotlb_unmap_sg_attrs,
> + .map_sg = xen_swiotlb_map_sg,
> + .unmap_sg = xen_swiotlb_unmap_sg,
> .map_page = xen_swiotlb_map_page,
> .unmap_page = xen_swiotlb_unmap_page,
> .dma_supported = xen_swiotlb_dma_supported,
> --
> 2.20.1
>
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [PATCH 2/4] swiotlb-xen: use ->map_page to implement ->map_sg
2019-04-11 7:19 ` [Xen-devel] [PATCH 2/4] swiotlb-xen: use ->map_page to implement ->map_sg Christoph Hellwig
@ 2019-04-15 22:55 ` Stefano Stabellini
2019-04-15 22:55 ` [Xen-devel] " Stefano Stabellini
1 sibling, 0 replies; 19+ messages in thread
From: Stefano Stabellini @ 2019-04-15 22:55 UTC (permalink / raw)
To: Christoph Hellwig
Cc: Juergen Gross, Stefano Stabellini, Konrad Rzeszutek Wilk, iommu,
xen-devel, Boris Ostrovsky
On Thu, 11 Apr 2019, Christoph Hellwig wrote:
> We can simply loop over the segments and map them, removing lots of
> duplicate code.
Right, the only difference is the additional dma_capable check which is
good to have.
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
> Signed-off-by: Christoph Hellwig <hch@lst.de>
> ---
> drivers/xen/swiotlb-xen.c | 68 ++++++---------------------------------
> 1 file changed, 10 insertions(+), 58 deletions(-)
>
> diff --git a/drivers/xen/swiotlb-xen.c b/drivers/xen/swiotlb-xen.c
> index d4bc3aabd44d..97a55c225593 100644
> --- a/drivers/xen/swiotlb-xen.c
> +++ b/drivers/xen/swiotlb-xen.c
> @@ -517,24 +517,8 @@ xen_swiotlb_unmap_sg(struct device *hwdev, struct scatterlist *sgl, int nelems,
>
> }
>
> -/*
> - * Map a set of buffers described by scatterlist in streaming mode for DMA.
> - * This is the scatter-gather version of the above xen_swiotlb_map_page
> - * interface. Here the scatter gather list elements are each tagged with the
> - * appropriate dma address and length. They are obtained via
> - * sg_dma_{address,length}(SG).
> - *
> - * NOTE: An implementation may be able to use a smaller number of
> - * DMA address/length pairs than there are SG table elements.
> - * (for example via virtual mapping capabilities)
> - * The routine returns the number of addr/length pairs actually
> - * used, at most nents.
> - *
> - * Device ownership issues as mentioned above for xen_swiotlb_map_page are the
> - * same here.
> - */
> static int
> -xen_swiotlb_map_sg(struct device *hwdev, struct scatterlist *sgl, int nelems,
> +xen_swiotlb_map_sg(struct device *dev, struct scatterlist *sgl, int nelems,
> enum dma_data_direction dir, unsigned long attrs)
> {
> struct scatterlist *sg;
> @@ -543,50 +527,18 @@ xen_swiotlb_map_sg(struct device *hwdev, struct scatterlist *sgl, int nelems,
> BUG_ON(dir == DMA_NONE);
>
> for_each_sg(sgl, sg, nelems, i) {
> - phys_addr_t paddr = sg_phys(sg);
> - dma_addr_t dev_addr = xen_phys_to_bus(paddr);
> -
> - if (swiotlb_force == SWIOTLB_FORCE ||
> - xen_arch_need_swiotlb(hwdev, paddr, dev_addr) ||
> - !dma_capable(hwdev, dev_addr, sg->length) ||
> - range_straddles_page_boundary(paddr, sg->length)) {
> - phys_addr_t map = swiotlb_tbl_map_single(hwdev,
> - start_dma_addr,
> - sg_phys(sg),
> - sg->length,
> - dir, attrs);
> - if (map == DMA_MAPPING_ERROR) {
> - dev_warn(hwdev, "swiotlb buffer is full\n");
> - /* Don't panic here, we expect map_sg users
> - to do proper error handling. */
> - attrs |= DMA_ATTR_SKIP_CPU_SYNC;
> - xen_swiotlb_unmap_sg(hwdev, sgl, i, dir, attrs);
> - sg_dma_len(sgl) = 0;
> - return 0;
> - }
> - dev_addr = xen_phys_to_bus(map);
> - xen_dma_map_page(hwdev, pfn_to_page(map >> PAGE_SHIFT),
> - dev_addr,
> - map & ~PAGE_MASK,
> - sg->length,
> - dir,
> - attrs);
> - sg->dma_address = dev_addr;
> - } else {
> - /* we are not interested in the dma_addr returned by
> - * xen_dma_map_page, only in the potential cache flushes executed
> - * by the function. */
> - xen_dma_map_page(hwdev, pfn_to_page(paddr >> PAGE_SHIFT),
> - dev_addr,
> - paddr & ~PAGE_MASK,
> - sg->length,
> - dir,
> - attrs);
> - sg->dma_address = dev_addr;
> - }
> + sg->dma_address = xen_swiotlb_map_page(dev, sg_page(sg),
> + sg->offset, sg->length, dir, attrs);
> + if (sg->dma_address == DMA_MAPPING_ERROR)
> + goto out_unmap;
> sg_dma_len(sg) = sg->length;
> }
> +
> return nelems;
> +out_unmap:
> + xen_swiotlb_unmap_sg(dev, sgl, i, dir, attrs | DMA_ATTR_SKIP_CPU_SYNC);
> + sg_dma_len(sgl) = 0;
> + return 0;
> }
>
> /*
> --
> 2.20.1
>
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [Xen-devel] [PATCH 2/4] swiotlb-xen: use ->map_page to implement ->map_sg
2019-04-11 7:19 ` [Xen-devel] [PATCH 2/4] swiotlb-xen: use ->map_page to implement ->map_sg Christoph Hellwig
2019-04-15 22:55 ` Stefano Stabellini
@ 2019-04-15 22:55 ` Stefano Stabellini
1 sibling, 0 replies; 19+ messages in thread
From: Stefano Stabellini @ 2019-04-15 22:55 UTC (permalink / raw)
To: Christoph Hellwig
Cc: Juergen Gross, Stefano Stabellini, Konrad Rzeszutek Wilk, iommu,
xen-devel, Boris Ostrovsky
On Thu, 11 Apr 2019, Christoph Hellwig wrote:
> We can simply loop over the segments and map them, removing lots of
> duplicate code.
Right, the only difference is the additional dma_capable check which is
good to have.
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
> Signed-off-by: Christoph Hellwig <hch@lst.de>
> ---
> drivers/xen/swiotlb-xen.c | 68 ++++++---------------------------------
> 1 file changed, 10 insertions(+), 58 deletions(-)
>
> diff --git a/drivers/xen/swiotlb-xen.c b/drivers/xen/swiotlb-xen.c
> index d4bc3aabd44d..97a55c225593 100644
> --- a/drivers/xen/swiotlb-xen.c
> +++ b/drivers/xen/swiotlb-xen.c
> @@ -517,24 +517,8 @@ xen_swiotlb_unmap_sg(struct device *hwdev, struct scatterlist *sgl, int nelems,
>
> }
>
> -/*
> - * Map a set of buffers described by scatterlist in streaming mode for DMA.
> - * This is the scatter-gather version of the above xen_swiotlb_map_page
> - * interface. Here the scatter gather list elements are each tagged with the
> - * appropriate dma address and length. They are obtained via
> - * sg_dma_{address,length}(SG).
> - *
> - * NOTE: An implementation may be able to use a smaller number of
> - * DMA address/length pairs than there are SG table elements.
> - * (for example via virtual mapping capabilities)
> - * The routine returns the number of addr/length pairs actually
> - * used, at most nents.
> - *
> - * Device ownership issues as mentioned above for xen_swiotlb_map_page are the
> - * same here.
> - */
> static int
> -xen_swiotlb_map_sg(struct device *hwdev, struct scatterlist *sgl, int nelems,
> +xen_swiotlb_map_sg(struct device *dev, struct scatterlist *sgl, int nelems,
> enum dma_data_direction dir, unsigned long attrs)
> {
> struct scatterlist *sg;
> @@ -543,50 +527,18 @@ xen_swiotlb_map_sg(struct device *hwdev, struct scatterlist *sgl, int nelems,
> BUG_ON(dir == DMA_NONE);
>
> for_each_sg(sgl, sg, nelems, i) {
> - phys_addr_t paddr = sg_phys(sg);
> - dma_addr_t dev_addr = xen_phys_to_bus(paddr);
> -
> - if (swiotlb_force == SWIOTLB_FORCE ||
> - xen_arch_need_swiotlb(hwdev, paddr, dev_addr) ||
> - !dma_capable(hwdev, dev_addr, sg->length) ||
> - range_straddles_page_boundary(paddr, sg->length)) {
> - phys_addr_t map = swiotlb_tbl_map_single(hwdev,
> - start_dma_addr,
> - sg_phys(sg),
> - sg->length,
> - dir, attrs);
> - if (map == DMA_MAPPING_ERROR) {
> - dev_warn(hwdev, "swiotlb buffer is full\n");
> - /* Don't panic here, we expect map_sg users
> - to do proper error handling. */
> - attrs |= DMA_ATTR_SKIP_CPU_SYNC;
> - xen_swiotlb_unmap_sg(hwdev, sgl, i, dir, attrs);
> - sg_dma_len(sgl) = 0;
> - return 0;
> - }
> - dev_addr = xen_phys_to_bus(map);
> - xen_dma_map_page(hwdev, pfn_to_page(map >> PAGE_SHIFT),
> - dev_addr,
> - map & ~PAGE_MASK,
> - sg->length,
> - dir,
> - attrs);
> - sg->dma_address = dev_addr;
> - } else {
> - /* we are not interested in the dma_addr returned by
> - * xen_dma_map_page, only in the potential cache flushes executed
> - * by the function. */
> - xen_dma_map_page(hwdev, pfn_to_page(paddr >> PAGE_SHIFT),
> - dev_addr,
> - paddr & ~PAGE_MASK,
> - sg->length,
> - dir,
> - attrs);
> - sg->dma_address = dev_addr;
> - }
> + sg->dma_address = xen_swiotlb_map_page(dev, sg_page(sg),
> + sg->offset, sg->length, dir, attrs);
> + if (sg->dma_address == DMA_MAPPING_ERROR)
> + goto out_unmap;
> sg_dma_len(sg) = sg->length;
> }
> +
> return nelems;
> +out_unmap:
> + xen_swiotlb_unmap_sg(dev, sgl, i, dir, attrs | DMA_ATTR_SKIP_CPU_SYNC);
> + sg_dma_len(sgl) = 0;
> + return 0;
> }
>
> /*
> --
> 2.20.1
>
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [PATCH 3/4] swiotlb-xen: simplify the DMA sync method implementations
2019-04-11 7:19 ` [Xen-devel] " Christoph Hellwig
@ 2019-04-15 22:55 ` Stefano Stabellini
2019-04-15 22:55 ` [Xen-devel] " Stefano Stabellini
1 sibling, 0 replies; 19+ messages in thread
From: Stefano Stabellini @ 2019-04-15 22:55 UTC (permalink / raw)
To: Christoph Hellwig
Cc: Juergen Gross, Stefano Stabellini, Konrad Rzeszutek Wilk, iommu,
xen-devel, Boris Ostrovsky
On Thu, 11 Apr 2019, Christoph Hellwig wrote:
> Get rid of the grand multiplexer and implement the sync_single_for_cpu
> and sync_single_for_device methods directly, and then loop over them
> for the scatterlist based variants.
>
> Note that this also loses a few comments related to highlevel DMA API
> concepts, which have nothing to do with the swiotlb-xen implementation
> details.
>
> Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
> ---
> drivers/xen/swiotlb-xen.c | 84 +++++++++++++--------------------------
> 1 file changed, 28 insertions(+), 56 deletions(-)
>
> diff --git a/drivers/xen/swiotlb-xen.c b/drivers/xen/swiotlb-xen.c
> index 97a55c225593..9a951504dc12 100644
> --- a/drivers/xen/swiotlb-xen.c
> +++ b/drivers/xen/swiotlb-xen.c
> @@ -455,48 +455,28 @@ static void xen_swiotlb_unmap_page(struct device *hwdev, dma_addr_t dev_addr,
> xen_unmap_single(hwdev, dev_addr, size, dir, attrs);
> }
>
> -/*
> - * Make physical memory consistent for a single streaming mode DMA translation
> - * after a transfer.
> - *
> - * If you perform a xen_swiotlb_map_page() but wish to interrogate the buffer
> - * using the cpu, yet do not wish to teardown the dma mapping, you must
> - * call this function before doing so. At the next point you give the dma
> - * address back to the card, you must first perform a
> - * xen_swiotlb_dma_sync_for_device, and then the device again owns the buffer
> - */
> static void
> -xen_swiotlb_sync_single(struct device *hwdev, dma_addr_t dev_addr,
> - size_t size, enum dma_data_direction dir,
> - enum dma_sync_target target)
> +xen_swiotlb_sync_single_for_cpu(struct device *dev, dma_addr_t dma_addr,
> + size_t size, enum dma_data_direction dir)
> {
> - phys_addr_t paddr = xen_bus_to_phys(dev_addr);
> + phys_addr_t paddr = xen_bus_to_phys(dma_addr);
>
> - BUG_ON(dir == DMA_NONE);
> -
> - if (target == SYNC_FOR_CPU)
> - xen_dma_sync_single_for_cpu(hwdev, dev_addr, size, dir);
> + xen_dma_sync_single_for_cpu(dev, dma_addr, size, dir);
>
> - /* NOTE: We use dev_addr here, not paddr! */
> - if (is_xen_swiotlb_buffer(dev_addr))
> - swiotlb_tbl_sync_single(hwdev, paddr, size, dir, target);
> -
> - if (target == SYNC_FOR_DEVICE)
> - xen_dma_sync_single_for_device(hwdev, dev_addr, size, dir);
> + if (is_xen_swiotlb_buffer(dma_addr))
> + swiotlb_tbl_sync_single(dev, paddr, size, dir, SYNC_FOR_CPU);
> }
>
> -void
> -xen_swiotlb_sync_single_for_cpu(struct device *hwdev, dma_addr_t dev_addr,
> - size_t size, enum dma_data_direction dir)
> +static void
> +xen_swiotlb_sync_single_for_device(struct device *dev, dma_addr_t dma_addr,
> + size_t size, enum dma_data_direction dir)
> {
> - xen_swiotlb_sync_single(hwdev, dev_addr, size, dir, SYNC_FOR_CPU);
> -}
> + phys_addr_t paddr = xen_bus_to_phys(dma_addr);
>
> -void
> -xen_swiotlb_sync_single_for_device(struct device *hwdev, dma_addr_t dev_addr,
> - size_t size, enum dma_data_direction dir)
> -{
> - xen_swiotlb_sync_single(hwdev, dev_addr, size, dir, SYNC_FOR_DEVICE);
> + if (is_xen_swiotlb_buffer(dma_addr))
> + swiotlb_tbl_sync_single(dev, paddr, size, dir, SYNC_FOR_DEVICE);
> +
> + xen_dma_sync_single_for_device(dev, dma_addr, size, dir);
> }
>
> /*
> @@ -541,38 +521,30 @@ xen_swiotlb_map_sg(struct device *dev, struct scatterlist *sgl, int nelems,
> return 0;
> }
>
> -/*
> - * Make physical memory consistent for a set of streaming mode DMA translations
> - * after a transfer.
> - *
> - * The same as swiotlb_sync_single_* but for a scatter-gather list, same rules
> - * and usage.
> - */
> static void
> -xen_swiotlb_sync_sg(struct device *hwdev, struct scatterlist *sgl,
> - int nelems, enum dma_data_direction dir,
> - enum dma_sync_target target)
> +xen_swiotlb_sync_sg_for_cpu(struct device *dev, struct scatterlist *sgl,
> + int nelems, enum dma_data_direction dir)
> {
> struct scatterlist *sg;
> int i;
>
> - for_each_sg(sgl, sg, nelems, i)
> - xen_swiotlb_sync_single(hwdev, sg->dma_address,
> - sg_dma_len(sg), dir, target);
> -}
> -
> -static void
> -xen_swiotlb_sync_sg_for_cpu(struct device *hwdev, struct scatterlist *sg,
> - int nelems, enum dma_data_direction dir)
> -{
> - xen_swiotlb_sync_sg(hwdev, sg, nelems, dir, SYNC_FOR_CPU);
> + for_each_sg(sgl, sg, nelems, i) {
> + xen_swiotlb_sync_single_for_cpu(dev, sg->dma_address,
> + sg->length, dir);
> + }
> }
>
> static void
> -xen_swiotlb_sync_sg_for_device(struct device *hwdev, struct scatterlist *sg,
> +xen_swiotlb_sync_sg_for_device(struct device *dev, struct scatterlist *sgl,
> int nelems, enum dma_data_direction dir)
> {
> - xen_swiotlb_sync_sg(hwdev, sg, nelems, dir, SYNC_FOR_DEVICE);
> + struct scatterlist *sg;
> + int i;
> +
> + for_each_sg(sgl, sg, nelems, i) {
> + xen_swiotlb_sync_single_for_device(dev, sg->dma_address,
> + sg->length, dir);
> + }
> }
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [Xen-devel] [PATCH 3/4] swiotlb-xen: simplify the DMA sync method implementations
2019-04-11 7:19 ` [Xen-devel] " Christoph Hellwig
2019-04-15 22:55 ` Stefano Stabellini
@ 2019-04-15 22:55 ` Stefano Stabellini
1 sibling, 0 replies; 19+ messages in thread
From: Stefano Stabellini @ 2019-04-15 22:55 UTC (permalink / raw)
To: Christoph Hellwig
Cc: Juergen Gross, Stefano Stabellini, Konrad Rzeszutek Wilk, iommu,
xen-devel, Boris Ostrovsky
On Thu, 11 Apr 2019, Christoph Hellwig wrote:
> Get rid of the grand multiplexer and implement the sync_single_for_cpu
> and sync_single_for_device methods directly, and then loop over them
> for the scatterlist based variants.
>
> Note that this also loses a few comments related to highlevel DMA API
> concepts, which have nothing to do with the swiotlb-xen implementation
> details.
>
> Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
> ---
> drivers/xen/swiotlb-xen.c | 84 +++++++++++++--------------------------
> 1 file changed, 28 insertions(+), 56 deletions(-)
>
> diff --git a/drivers/xen/swiotlb-xen.c b/drivers/xen/swiotlb-xen.c
> index 97a55c225593..9a951504dc12 100644
> --- a/drivers/xen/swiotlb-xen.c
> +++ b/drivers/xen/swiotlb-xen.c
> @@ -455,48 +455,28 @@ static void xen_swiotlb_unmap_page(struct device *hwdev, dma_addr_t dev_addr,
> xen_unmap_single(hwdev, dev_addr, size, dir, attrs);
> }
>
> -/*
> - * Make physical memory consistent for a single streaming mode DMA translation
> - * after a transfer.
> - *
> - * If you perform a xen_swiotlb_map_page() but wish to interrogate the buffer
> - * using the cpu, yet do not wish to teardown the dma mapping, you must
> - * call this function before doing so. At the next point you give the dma
> - * address back to the card, you must first perform a
> - * xen_swiotlb_dma_sync_for_device, and then the device again owns the buffer
> - */
> static void
> -xen_swiotlb_sync_single(struct device *hwdev, dma_addr_t dev_addr,
> - size_t size, enum dma_data_direction dir,
> - enum dma_sync_target target)
> +xen_swiotlb_sync_single_for_cpu(struct device *dev, dma_addr_t dma_addr,
> + size_t size, enum dma_data_direction dir)
> {
> - phys_addr_t paddr = xen_bus_to_phys(dev_addr);
> + phys_addr_t paddr = xen_bus_to_phys(dma_addr);
>
> - BUG_ON(dir == DMA_NONE);
> -
> - if (target == SYNC_FOR_CPU)
> - xen_dma_sync_single_for_cpu(hwdev, dev_addr, size, dir);
> + xen_dma_sync_single_for_cpu(dev, dma_addr, size, dir);
>
> - /* NOTE: We use dev_addr here, not paddr! */
> - if (is_xen_swiotlb_buffer(dev_addr))
> - swiotlb_tbl_sync_single(hwdev, paddr, size, dir, target);
> -
> - if (target == SYNC_FOR_DEVICE)
> - xen_dma_sync_single_for_device(hwdev, dev_addr, size, dir);
> + if (is_xen_swiotlb_buffer(dma_addr))
> + swiotlb_tbl_sync_single(dev, paddr, size, dir, SYNC_FOR_CPU);
> }
>
> -void
> -xen_swiotlb_sync_single_for_cpu(struct device *hwdev, dma_addr_t dev_addr,
> - size_t size, enum dma_data_direction dir)
> +static void
> +xen_swiotlb_sync_single_for_device(struct device *dev, dma_addr_t dma_addr,
> + size_t size, enum dma_data_direction dir)
> {
> - xen_swiotlb_sync_single(hwdev, dev_addr, size, dir, SYNC_FOR_CPU);
> -}
> + phys_addr_t paddr = xen_bus_to_phys(dma_addr);
>
> -void
> -xen_swiotlb_sync_single_for_device(struct device *hwdev, dma_addr_t dev_addr,
> - size_t size, enum dma_data_direction dir)
> -{
> - xen_swiotlb_sync_single(hwdev, dev_addr, size, dir, SYNC_FOR_DEVICE);
> + if (is_xen_swiotlb_buffer(dma_addr))
> + swiotlb_tbl_sync_single(dev, paddr, size, dir, SYNC_FOR_DEVICE);
> +
> + xen_dma_sync_single_for_device(dev, dma_addr, size, dir);
> }
>
> /*
> @@ -541,38 +521,30 @@ xen_swiotlb_map_sg(struct device *dev, struct scatterlist *sgl, int nelems,
> return 0;
> }
>
> -/*
> - * Make physical memory consistent for a set of streaming mode DMA translations
> - * after a transfer.
> - *
> - * The same as swiotlb_sync_single_* but for a scatter-gather list, same rules
> - * and usage.
> - */
> static void
> -xen_swiotlb_sync_sg(struct device *hwdev, struct scatterlist *sgl,
> - int nelems, enum dma_data_direction dir,
> - enum dma_sync_target target)
> +xen_swiotlb_sync_sg_for_cpu(struct device *dev, struct scatterlist *sgl,
> + int nelems, enum dma_data_direction dir)
> {
> struct scatterlist *sg;
> int i;
>
> - for_each_sg(sgl, sg, nelems, i)
> - xen_swiotlb_sync_single(hwdev, sg->dma_address,
> - sg_dma_len(sg), dir, target);
> -}
> -
> -static void
> -xen_swiotlb_sync_sg_for_cpu(struct device *hwdev, struct scatterlist *sg,
> - int nelems, enum dma_data_direction dir)
> -{
> - xen_swiotlb_sync_sg(hwdev, sg, nelems, dir, SYNC_FOR_CPU);
> + for_each_sg(sgl, sg, nelems, i) {
> + xen_swiotlb_sync_single_for_cpu(dev, sg->dma_address,
> + sg->length, dir);
> + }
> }
>
> static void
> -xen_swiotlb_sync_sg_for_device(struct device *hwdev, struct scatterlist *sg,
> +xen_swiotlb_sync_sg_for_device(struct device *dev, struct scatterlist *sgl,
> int nelems, enum dma_data_direction dir)
> {
> - xen_swiotlb_sync_sg(hwdev, sg, nelems, dir, SYNC_FOR_DEVICE);
> + struct scatterlist *sg;
> + int i;
> +
> + for_each_sg(sgl, sg, nelems, i) {
> + xen_swiotlb_sync_single_for_device(dev, sg->dma_address,
> + sg->length, dir);
> + }
> }
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [PATCH 4/4] swiotlb-xen: ensure we have a single callsite for xen_dma_map_page
2019-04-11 7:20 ` [Xen-devel] [PATCH 4/4] swiotlb-xen: ensure we have a single callsite for xen_dma_map_page Christoph Hellwig
2019-04-15 22:56 ` Stefano Stabellini
@ 2019-04-15 22:56 ` Stefano Stabellini
1 sibling, 0 replies; 19+ messages in thread
From: Stefano Stabellini @ 2019-04-15 22:56 UTC (permalink / raw)
To: Christoph Hellwig
Cc: Juergen Gross, Stefano Stabellini, Konrad Rzeszutek Wilk, iommu,
xen-devel, Boris Ostrovsky
On Thu, 11 Apr 2019, Christoph Hellwig wrote:
> Refactor the code a bit to make further changes easier.
>
> Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
> ---
> drivers/xen/swiotlb-xen.c | 31 ++++++++++++++++---------------
> 1 file changed, 16 insertions(+), 15 deletions(-)
>
> diff --git a/drivers/xen/swiotlb-xen.c b/drivers/xen/swiotlb-xen.c
> index 9a951504dc12..5dcb06fe9667 100644
> --- a/drivers/xen/swiotlb-xen.c
> +++ b/drivers/xen/swiotlb-xen.c
> @@ -391,13 +391,8 @@ static dma_addr_t xen_swiotlb_map_page(struct device *dev, struct page *page,
> if (dma_capable(dev, dev_addr, size) &&
> !range_straddles_page_boundary(phys, size) &&
> !xen_arch_need_swiotlb(dev, phys, dev_addr) &&
> - (swiotlb_force != SWIOTLB_FORCE)) {
> - /* we are not interested in the dma_addr returned by
> - * xen_dma_map_page, only in the potential cache flushes executed
> - * by the function. */
> - xen_dma_map_page(dev, page, dev_addr, offset, size, dir, attrs);
> - return dev_addr;
> - }
> + swiotlb_force != SWIOTLB_FORCE)
> + goto done;
>
> /*
> * Oh well, have to allocate and map a bounce buffer.
> @@ -410,19 +405,25 @@ static dma_addr_t xen_swiotlb_map_page(struct device *dev, struct page *page,
> return DMA_MAPPING_ERROR;
>
> dev_addr = xen_phys_to_bus(map);
> - xen_dma_map_page(dev, pfn_to_page(map >> PAGE_SHIFT),
> - dev_addr, map & ~PAGE_MASK, size, dir, attrs);
>
> /*
> * Ensure that the address returned is DMA'ble
> */
> - if (dma_capable(dev, dev_addr, size))
> - return dev_addr;
> -
> - attrs |= DMA_ATTR_SKIP_CPU_SYNC;
> - swiotlb_tbl_unmap_single(dev, map, size, dir, attrs);
> + if (unlikely(!dma_capable(dev, dev_addr, size))) {
> + swiotlb_tbl_unmap_single(dev, map, size, dir,
> + attrs | DMA_ATTR_SKIP_CPU_SYNC);
> + return DMA_MAPPING_ERROR;
> + }
>
> - return DMA_MAPPING_ERROR;
> + page = pfn_to_page(map >> PAGE_SHIFT);
> + offset = map & ~PAGE_MASK;
> +done:
> + /*
> + * we are not interested in the dma_addr returned by xen_dma_map_page,
> + * only in the potential cache flushes executed by the function.
> + */
> + xen_dma_map_page(dev, page, dev_addr, offset, size, dir, attrs);
> + return dev_addr;
> }
>
> /*
> --
> 2.20.1
>
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [Xen-devel] [PATCH 4/4] swiotlb-xen: ensure we have a single callsite for xen_dma_map_page
2019-04-11 7:20 ` [Xen-devel] [PATCH 4/4] swiotlb-xen: ensure we have a single callsite for xen_dma_map_page Christoph Hellwig
@ 2019-04-15 22:56 ` Stefano Stabellini
2019-04-15 22:56 ` Stefano Stabellini
1 sibling, 0 replies; 19+ messages in thread
From: Stefano Stabellini @ 2019-04-15 22:56 UTC (permalink / raw)
To: Christoph Hellwig
Cc: Juergen Gross, Stefano Stabellini, Konrad Rzeszutek Wilk, iommu,
xen-devel, Boris Ostrovsky
On Thu, 11 Apr 2019, Christoph Hellwig wrote:
> Refactor the code a bit to make further changes easier.
>
> Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
> ---
> drivers/xen/swiotlb-xen.c | 31 ++++++++++++++++---------------
> 1 file changed, 16 insertions(+), 15 deletions(-)
>
> diff --git a/drivers/xen/swiotlb-xen.c b/drivers/xen/swiotlb-xen.c
> index 9a951504dc12..5dcb06fe9667 100644
> --- a/drivers/xen/swiotlb-xen.c
> +++ b/drivers/xen/swiotlb-xen.c
> @@ -391,13 +391,8 @@ static dma_addr_t xen_swiotlb_map_page(struct device *dev, struct page *page,
> if (dma_capable(dev, dev_addr, size) &&
> !range_straddles_page_boundary(phys, size) &&
> !xen_arch_need_swiotlb(dev, phys, dev_addr) &&
> - (swiotlb_force != SWIOTLB_FORCE)) {
> - /* we are not interested in the dma_addr returned by
> - * xen_dma_map_page, only in the potential cache flushes executed
> - * by the function. */
> - xen_dma_map_page(dev, page, dev_addr, offset, size, dir, attrs);
> - return dev_addr;
> - }
> + swiotlb_force != SWIOTLB_FORCE)
> + goto done;
>
> /*
> * Oh well, have to allocate and map a bounce buffer.
> @@ -410,19 +405,25 @@ static dma_addr_t xen_swiotlb_map_page(struct device *dev, struct page *page,
> return DMA_MAPPING_ERROR;
>
> dev_addr = xen_phys_to_bus(map);
> - xen_dma_map_page(dev, pfn_to_page(map >> PAGE_SHIFT),
> - dev_addr, map & ~PAGE_MASK, size, dir, attrs);
>
> /*
> * Ensure that the address returned is DMA'ble
> */
> - if (dma_capable(dev, dev_addr, size))
> - return dev_addr;
> -
> - attrs |= DMA_ATTR_SKIP_CPU_SYNC;
> - swiotlb_tbl_unmap_single(dev, map, size, dir, attrs);
> + if (unlikely(!dma_capable(dev, dev_addr, size))) {
> + swiotlb_tbl_unmap_single(dev, map, size, dir,
> + attrs | DMA_ATTR_SKIP_CPU_SYNC);
> + return DMA_MAPPING_ERROR;
> + }
>
> - return DMA_MAPPING_ERROR;
> + page = pfn_to_page(map >> PAGE_SHIFT);
> + offset = map & ~PAGE_MASK;
> +done:
> + /*
> + * we are not interested in the dma_addr returned by xen_dma_map_page,
> + * only in the potential cache flushes executed by the function.
> + */
> + xen_dma_map_page(dev, page, dev_addr, offset, size, dir, attrs);
> + return dev_addr;
> }
>
> /*
> --
> 2.20.1
>
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: a few xen swiotlb cleanups
2019-04-11 7:19 [Xen-devel] a few xen swiotlb cleanups Christoph Hellwig
` (7 preceding siblings ...)
2019-04-11 7:20 ` Christoph Hellwig
@ 2019-04-25 18:55 ` Konrad Rzeszutek Wilk
2019-04-25 18:55 ` [Xen-devel] " Konrad Rzeszutek Wilk
9 siblings, 0 replies; 19+ messages in thread
From: Konrad Rzeszutek Wilk @ 2019-04-25 18:55 UTC (permalink / raw)
To: Christoph Hellwig
Cc: Juergen Gross, xen-devel, Boris Ostrovsky, Stefano Stabellini, iommu
On Thu, Apr 11, 2019 at 09:19:56AM +0200, Christoph Hellwig wrote:
> Hi all,
I will slurp these up.. right after I test them for correctness.
>
> below are a couple of cleanups for swiotlb-xen.c. They were done in
> preparation of eventually using the dma-noncoherent.h cache flushing
> hooks, but that final goal will need some major work to the arm32 code
> first. Until then I think these patches might be better in mainline
> than in my local tree so that they don't bitrot.
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [Xen-devel] a few xen swiotlb cleanups
2019-04-11 7:19 [Xen-devel] a few xen swiotlb cleanups Christoph Hellwig
` (8 preceding siblings ...)
2019-04-25 18:55 ` a few xen swiotlb cleanups Konrad Rzeszutek Wilk
@ 2019-04-25 18:55 ` Konrad Rzeszutek Wilk
9 siblings, 0 replies; 19+ messages in thread
From: Konrad Rzeszutek Wilk @ 2019-04-25 18:55 UTC (permalink / raw)
To: Christoph Hellwig
Cc: Juergen Gross, xen-devel, Boris Ostrovsky, Stefano Stabellini, iommu
On Thu, Apr 11, 2019 at 09:19:56AM +0200, Christoph Hellwig wrote:
> Hi all,
I will slurp these up.. right after I test them for correctness.
>
> below are a couple of cleanups for swiotlb-xen.c. They were done in
> preparation of eventually using the dma-noncoherent.h cache flushing
> hooks, but that final goal will need some major work to the arm32 code
> first. Until then I think these patches might be better in mainline
> than in my local tree so that they don't bitrot.
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel
^ permalink raw reply [flat|nested] 19+ messages in thread
end of thread, other threads:[~2019-04-25 18:57 UTC | newest]
Thread overview: 19+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2019-04-11 7:19 [Xen-devel] a few xen swiotlb cleanups Christoph Hellwig
2019-04-11 7:19 ` [PATCH 1/4] swiotlb-xen: make instances match their method names Christoph Hellwig
2019-04-11 7:19 ` [Xen-devel] " Christoph Hellwig
2019-04-15 22:55 ` Stefano Stabellini
2019-04-15 22:55 ` Stefano Stabellini
2019-04-11 7:19 ` [Xen-devel] [PATCH 2/4] swiotlb-xen: use ->map_page to implement ->map_sg Christoph Hellwig
2019-04-15 22:55 ` Stefano Stabellini
2019-04-15 22:55 ` [Xen-devel] " Stefano Stabellini
2019-04-11 7:19 ` Christoph Hellwig
2019-04-11 7:19 ` [PATCH 3/4] swiotlb-xen: simplify the DMA sync method implementations Christoph Hellwig
2019-04-11 7:19 ` [Xen-devel] " Christoph Hellwig
2019-04-15 22:55 ` Stefano Stabellini
2019-04-15 22:55 ` [Xen-devel] " Stefano Stabellini
2019-04-11 7:20 ` [Xen-devel] [PATCH 4/4] swiotlb-xen: ensure we have a single callsite for xen_dma_map_page Christoph Hellwig
2019-04-15 22:56 ` Stefano Stabellini
2019-04-15 22:56 ` Stefano Stabellini
2019-04-11 7:20 ` Christoph Hellwig
2019-04-25 18:55 ` a few xen swiotlb cleanups Konrad Rzeszutek Wilk
2019-04-25 18:55 ` [Xen-devel] " Konrad Rzeszutek Wilk
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).