DMA, CMA: support alignment constraint on CMA region
authorJoonsoo Kim <iamjoonsoo.kim@lge.com>
Wed, 6 Aug 2014 23:05:21 +0000 (16:05 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Thu, 7 Aug 2014 01:01:16 +0000 (18:01 -0700)
PPC KVM's CMA area management needs alignment constraint on CMA region.
So support it to prepare generalization of CMA area management
functionality.

Additionally, add some comments which tell us why alignment constraint
is needed on CMA region.

Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Acked-by: Michal Nazarewicz <mina86@mina86.com>
Reviewed-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Cc: Alexander Graf <agraf@suse.de>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Cc: Gleb Natapov <gleb@kernel.org>
Acked-by: Marek Szyprowski <m.szyprowski@samsung.com>
Tested-by: Marek Szyprowski <m.szyprowski@samsung.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
drivers/base/dma-contiguous.c

index 9021762227a7a5db6ed992ee8dcdf5585f19f0f8..5f62c284072c8dac10d6b76ef290d468cf444866 100644 (file)
@@ -32,6 +32,7 @@
 #include <linux/swap.h>
 #include <linux/mm_types.h>
 #include <linux/dma-contiguous.h>
+#include <linux/log2.h>
 
 struct cma {
        unsigned long   base_pfn;
@@ -215,17 +216,16 @@ core_initcall(cma_init_reserved_areas);
 
 static int __init __dma_contiguous_reserve_area(phys_addr_t size,
                                phys_addr_t base, phys_addr_t limit,
+                               phys_addr_t alignment,
                                struct cma **res_cma, bool fixed)
 {
        struct cma *cma = &cma_areas[cma_area_count];
-       phys_addr_t alignment;
        int ret = 0;
 
-       pr_debug("%s(size %lx, base %08lx, limit %08lx)\n", __func__,
-                (unsigned long)size, (unsigned long)base,
-                (unsigned long)limit);
+       pr_debug("%s(size %lx, base %08lx, limit %08lx alignment %08lx)\n",
+               __func__, (unsigned long)size, (unsigned long)base,
+               (unsigned long)limit, (unsigned long)alignment);
 
-       /* Sanity checks */
        if (cma_area_count == ARRAY_SIZE(cma_areas)) {
                pr_err("Not enough slots for CMA reserved regions!\n");
                return -ENOSPC;
@@ -234,8 +234,17 @@ static int __init __dma_contiguous_reserve_area(phys_addr_t size,
        if (!size)
                return -EINVAL;
 
-       /* Sanitise input arguments */
-       alignment = PAGE_SIZE << max(MAX_ORDER - 1, pageblock_order);
+       if (alignment && !is_power_of_2(alignment))
+               return -EINVAL;
+
+       /*
+        * Sanitise input arguments.
+        * Pages both ends in CMA area could be merged into adjacent unmovable
+        * migratetype page by page allocator's buddy algorithm. In the case,
+        * you couldn't get a contiguous memory, which is not what we want.
+        */
+       alignment = max(alignment,
+               (phys_addr_t)PAGE_SIZE << max(MAX_ORDER - 1, pageblock_order));
        base = ALIGN(base, alignment);
        size = ALIGN(size, alignment);
        limit &= ~(alignment - 1);
@@ -299,7 +308,8 @@ int __init dma_contiguous_reserve_area(phys_addr_t size, phys_addr_t base,
 {
        int ret;
 
-       ret = __dma_contiguous_reserve_area(size, base, limit, res_cma, fixed);
+       ret = __dma_contiguous_reserve_area(size, base, limit, 0,
+                                               res_cma, fixed);
        if (ret)
                return ret;