[PATCH] rmem: support for dma_addr different from phys_addr

[PATCH] rmem: support for dma_addr different from phys_addr

[Vladimir Kondratiev]

On some systems, dma address differs from physical one due to
various reasons.

dma_init_coherent_memory is aware of it, it takes both
phys_addr_t and dma_addt_t arguments for memory block.
However, rmem_dma_device_init passes physical address for DMA one.

Fix this, using phys_to_dma(). This assumes the whole buffer has
same offset between physical and dma addresses.

Signed-off-by: Vladimir Kondratiev <vladimir.kondratiev@intel.com>
---
 drivers/base/dma-coherent.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/base/dma-coherent.c b/drivers/base/dma-coherent.c
index 640a7e63c453..5371cb19c918 100644
--- a/drivers/base/dma-coherent.c
+++ b/drivers/base/dma-coherent.c
@@ -292,7 +292,8 @@ static int rmem_dma_device_init(struct reserved_mem *rmem, struct device *dev)
 	struct dma_coherent_mem *mem = rmem->priv;
 
 	if (!mem &&
-	    !dma_init_coherent_memory(rmem->base, rmem->base, rmem->size,
+	    !dma_init_coherent_memory(rmem->base, phys_to_dma(dev, rmem->base),
+				      rmem->size,
 				      DMA_MEMORY_MAP | DMA_MEMORY_EXCLUSIVE,
 				      &mem)) {
 		pr_err("Reserved memory: failed to init DMA memory pool at %pa, size %ld MiB\n",
-- 
2.11.0

Re: [PATCH] rmem: support for dma_addr different from phys_addr

[Christoph Hellwig]

We've already got a different fix for this in the dma-mapping tree
for this merge window:

http://git.infradead.org/users/hch/dma-mapping.git/commitdiff/c41f9ea998f3ba61f38fa350eef28ec6caf0a07d

Also I think the subject line is rather confusing, I only read this mail
by accident..

[PATCH] rmem: support for dma_addr different from phys_addr

[Vladimir Kondratiev]

On some systems, dma address differs from physical one due to
various reasons.

dma_init_coherent_memory is aware of it, it takes both
phys_addr_t and dma_addt_t arguments for memory block.
However, rmem_dma_device_init passes physical address for DMA one.

Fix this, using phys_to_dma(). This assumes the whole buffer has
same offset between physical and dma addresses.

Change-Id: Ic709312941ee4e1a37afb2ea25cbd759eabc009b
Signed-off-by: Vladimir Kondratiev <vladimir.kondratiev@intel.com>
---
 drivers/base/dma-coherent.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/base/dma-coherent.c b/drivers/base/dma-coherent.c
index 640a7e63c453..5371cb19c918 100644
--- a/drivers/base/dma-coherent.c
+++ b/drivers/base/dma-coherent.c
@@ -292,7 +292,8 @@ static int rmem_dma_device_init(struct reserved_mem *rmem, struct device *dev)
 	struct dma_coherent_mem *mem = rmem->priv;
 
 	if (!mem &&
-	    !dma_init_coherent_memory(rmem->base, rmem->base, rmem->size,
+	    !dma_init_coherent_memory(rmem->base, phys_to_dma(dev, rmem->base),
+				      rmem->size,
 				      DMA_MEMORY_MAP | DMA_MEMORY_EXCLUSIVE,
 				      &mem)) {
 		pr_err("Reserved memory: failed to init DMA memory pool at %pa, size %ld MiB\n",
-- 
2.11.0

---------------------------------------------------------------------
Intel Israel (74) Limited

This e-mail and any attachments may contain confidential material for
the sole use of the intended recipient(s). Any review or distribution
by others is strictly prohibited. If you are not the intended
recipient, please contact the sender and delete all copies.

Re: [PATCH] rmem: support for dma_addr different from phys_addr

[Greg Kroah-Hartman]

On Tue, Jul 04, 2017 at 01:34:38PM +0300, Vladimir Kondratiev wrote:
> On some systems, dma address differs from physical one due to
> various reasons.
> 
> dma_init_coherent_memory is aware of it, it takes both
> phys_addr_t and dma_addt_t arguments for memory block.
> However, rmem_dma_device_init passes physical address for DMA one.
> 
> Fix this, using phys_to_dma(). This assumes the whole buffer has
> same offset between physical and dma addresses.
> 
> Change-Id: Ic709312941ee4e1a37afb2ea25cbd759eabc009b

What is that line for?  You didn't use checkpatch.pl :(

> Intel Israel (74) Limited
> 
> This e-mail and any attachments may contain confidential material for
> the sole use of the intended recipient(s). Any review or distribution
> by others is strictly prohibited. If you are not the intended
> recipient, please contact the sender and delete all copies.

Ok, this is getting really annoying, according to my lawyers, I have to
never respond to emails with this in it, and should just delete it.  But
I'll be nice, and tell you all to go fix all of Intel's email system to
never send this stuff out again, it keeps happening over and over and
over and is preventing you from contributing to open source projects.

email is now deleted.

greg k-h