dma-mapping: fix ia64 and m32r build error and warnings

From: Niklas Söderlund <[email protected]>

Hi Vindo,

This small series fixes the build error and warnings for ia64 and m32r which kbuild test robot found and where introduced in the series '[PATCHv9 0/6] dmaengine: rcar-dmac: add iommu support for slave transfers'.

It is based ontop of:

git://git.kernel.org/pub/scm/linux/kernel/git/vkoul/slave-dma.git next

The fixes are only compile tested for the ia64 and m32r architecture to show that the error and warnings are fixed. It is tested on ARM to prove that the intended functionality still works.

Kbuild test robot found one more class of warnings for the patch series mentioned above for the balackfin arch.

include/linux/dma-mapping.h: In function 'dma_map_resource':

    #define pfn_valid(pfn)  ((pfn) >= ARCH_PFN_OFFSET && ((pfn) - ARCH_PFN_OFFSET) < max_mapnr)
                                   ^

These warnings are not addressed in this series since they are not new or unique to original patch. They come from how pfn_valid() is implemented in include/asm-generic/page.h, if ARCH_PFN_OFFSET is 0 the warning is triggered. On the kbuild blackfin build it is defined as '(0x0 >> 12)' so any invocation of pfn_valid() will trigger it.

I tried to fix it but was unable to find a good solution. There is a MIPS specific fix for this in commit 95486e4 (MIPS: Fix flood of warnings about comparsion being always true.), but a similar fix for asm-generic seems to tricky for me to attempt since the macro needs to be turned into a static inline function and that might be harmful. I also toyed with the solution of '#if ARCH_PFN_OFFSET == 0' to provided an alternative pfn_valid() macro, but since ARCH_PFN_OFFSET is defined as an expression and not a constant that wont work. If you can think of another way I be happy to try that out and see if I can make it work.

Niklas Söderlund (2):
dma-mapping: fix ia64 build, use PHYS_PFN
dma-mapping: fix m32r build warning

 include/linux/dma-mapping.h | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

2.9.3

From: Niklas Söderlund <[email protected]>

kbuild test robot reports:

In file included from include/linux/skbuff.h:34:0,
from include/linux/tcp.h:21,
from drivers/net/ethernet/amd/xgbe/xgbe-drv.c:119:
include/linux/dma-mapping.h: In function 'dma_map_resource':

     unsigned long pfn = __phys_to_pfn(phys_addr);
                         ^~~~~~~~~~~~~

ia64 does not provide __phys_to_pfn(), use the PHYS_PFN() alias.

---
 include/linux/dma-mapping.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

index 6e00c7f..ff7c87f 100644
@@ -271,7 +271,7 @@ static inline dma_addr_t dma_map_resource(struct device *dev,
 					  unsigned long attrs)
 {
 	struct dma_map_ops *ops = get_dma_ops(dev);
-	unsigned long pfn = __phys_to_pfn(phys_addr);
+	unsigned long pfn = PHYS_PFN(phys_addr);
 	dma_addr_t addr;
 
 	BUG_ON(!valid_dma_direction(dir));

2.9.3

_______________________________________________
iommu mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/iommu

From: Niklas Söderlund <[email protected]>

kbuild test robot reports:

In file included from include/linux/skbuff.h:34:0,

                    from include/linux/icmpv6.h:4,
                    from include/linux/ipv6.h:75,
                    from include/net/ipv6.h:16,
                    from include/linux/sunrpc/clnt.h:27,
                    from include/linux/nfs_fs.h:30,
                    from fs/lockd/clntlock.c:13:

include/linux/dma-mapping.h: In function 'dma_map_resource':

     unsigned long pfn = __phys_to_pfn(phys_addr);
                   ^~~

The pfn value is only used once in the call to pfn_valid(), remove the variable and calculate the pfn when it's needed. Note that the kbuild report is old and PHYS_PFN() is now used instead of __phys_to_pfn() to calculate the pfn.

---
 include/linux/dma-mapping.h | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

index ff7c87f..642cb4c 100644
@@ -271,13 +271,12 @@ static inline dma_addr_t dma_map_resource(struct device *dev,
 					  unsigned long attrs)
 {
 	struct dma_map_ops *ops = get_dma_ops(dev);
-	unsigned long pfn = PHYS_PFN(phys_addr);
 	dma_addr_t addr;
 
 	BUG_ON(!valid_dma_direction(dir));
 
 	/* Don't allow RAM to be mapped */
-	BUG_ON(pfn_valid(pfn));
+	BUG_ON(pfn_valid(PHYS_PFN(phys_addr)));
 
 	addr = phys_addr;
 	if (ops->map_resource)

2.9.3

:(

Applied both

Thanks

~Vinod

I'm sorry, my hands sometimes moves faster then my head.

Thanks.

Regards,
Niklas Söderlund