[PATCH] mm: fix 'size' alignment in devm_memremap_pages()

[PATCH] mm: fix 'size' alignment in devm_memremap_pages()

[Dan Williams]

We need to align the end address, not just the size.

Cc: Toshi Kani <toshi.kani@hpe.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
---
Hi Andrew, one more fixup to devm_memremap_pages().  I was discussing
patch "mm: fix mixed zone detection in devm_memremap_pages" with Toshi
and noticed that it was mishandling the end-of-range alignment.  Please
apply or fold this into the existing patch.

 kernel/memremap.c |   12 +++++++-----
 1 file changed, 7 insertions(+), 5 deletions(-)

diff --git a/kernel/memremap.c b/kernel/memremap.c
index c0f11a498a5a..60baf4d3401e 100644
--- a/kernel/memremap.c
+++ b/kernel/memremap.c
@@ -270,14 +270,16 @@ struct dev_pagemap *find_dev_pagemap(resource_size_t phys)
 void *devm_memremap_pages(struct device *dev, struct resource *res,
 		struct percpu_ref *ref, struct vmem_altmap *altmap)
 {
-	resource_size_t align_start = res->start & ~(SECTION_SIZE - 1);
-	resource_size_t align_size = ALIGN(resource_size(res), SECTION_SIZE);
-	int is_ram = region_intersects(align_start, align_size, "System RAM");
-	resource_size_t key, align_end;
+	resource_size_t key, align_start, align_size, align_end;
 	struct dev_pagemap *pgmap;
 	struct page_map *page_map;
+	int error, nid, is_ram;
 	unsigned long pfn;
-	int error, nid;
+
+	align_start = res->start & ~(SECTION_SIZE - 1);
+	align_size = ALIGN(res->start + resource_size(res), SECTION_SIZE)
+		- align_start;
+	is_ram = region_intersects(align_start, align_size, "System RAM");
 
 	if (is_ram == REGION_MIXED) {
 		WARN_ONCE(1, "%s attempted on mixed region %pr\n",

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: [PATCH] mm: fix 'size' alignment in devm_memremap_pages()

[Toshi Kani]

On Tue, 2016-03-08 at 14:32 -0800, Dan Williams wrote:
> We need to align the end address, not just the size.
> 
> Cc: Toshi Kani <toshi.kani@hpe.com>
> Signed-off-by: Dan Williams <dan.j.williams@intel.com>

The change looks good.

Reviewed-by: Toshi Kani <toshi.kani@hpe.com>

Thanks,
-Toshi

> ---
> Hi Andrew, one more fixup to devm_memremap_pages().A A I was discussing
> patch "mm: fix mixed zone detection in devm_memremap_pages" with Toshi
> and noticed that it was mishandling the end-of-range alignment.A A Please
> apply or fold this into the existing patch.
> 
> A kernel/memremap.c |A A A 12 +++++++-----
> A 1 file changed, 7 insertions(+), 5 deletions(-)
> 
> diff --git a/kernel/memremap.c b/kernel/memremap.c
> index c0f11a498a5a..60baf4d3401e 100644
> --- a/kernel/memremap.c
> +++ b/kernel/memremap.c
> @@ -270,14 +270,16 @@ struct dev_pagemap
> *find_dev_pagemap(resource_size_t phys)
> A void *devm_memremap_pages(struct device *dev, struct resource *res,
> A 		struct percpu_ref *ref, struct vmem_altmap *altmap)
> A {
> -	resource_size_t align_start = res->start & ~(SECTION_SIZE - 1);
> -	resource_size_t align_size = ALIGN(resource_size(res),
> SECTION_SIZE);
> -	int is_ram = region_intersects(align_start, align_size, "System
> RAM");
> -	resource_size_t key, align_end;
> +	resource_size_t key, align_start, align_size, align_end;
> A 	struct dev_pagemap *pgmap;
> A 	struct page_map *page_map;
> +	int error, nid, is_ram;
> A 	unsigned long pfn;
> -	int error, nid;
> +
> +	align_start = res->start & ~(SECTION_SIZE - 1);
> +	align_size = ALIGN(res->start + resource_size(res),
> SECTION_SIZE)
> +		- align_start;
> +	is_ram = region_intersects(align_start, align_size, "System
> RAM");
> A 
> A 	if (is_ram == REGION_MIXED) {
> A 		WARN_ONCE(1, "%s attempted on mixed region %pr\n",
> 

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>