On Wed, Jun 25, 2014 at 12:48:02AM +0200, Michal Nazarewicz wrote:
> With [f495d26: “generalize CMA reserved area management
> functionality”] patch CMA has its place under mm directory now so
> there is no need to shoehorn a highly CMA specific functions inside of
> page_alloc.c.
>
> As such move init_cma_reserved_pageblock from mm/page_alloc.c to
> mm/cma.c, rename it to cma_init_reserved_pageblock and refactor
> a little.
>
> Most importantly, if a !pfn_valid(pfn) is encountered, just
> return -EINVAL instead of warning and trying to continue the
> initialisation of the area. It's not clear, to me at least, what good
> is continuing the work on a PFN that is known to be invalid.
>
> Signed-off-by: Michal Nazarewicz <min...@mina86.com>
Acked-by: Joonsoo Kim <iamjoonsoo....@lge.com>
One question below.
> ---
> include/linux/gfp.h | 3 --
> mm/cma.c | 85
> +++++++++++++++++++++++++++++++++++++++++------------
> mm/page_alloc.c | 31 -------------------
> 3 files changed, 66 insertions(+), 53 deletions(-)
>
> diff --git a/include/linux/gfp.h b/include/linux/gfp.h
> index 5e7219d..107793e9 100644
> --- a/include/linux/gfp.h
> +++ b/include/linux/gfp.h
> @@ -415,9 +415,6 @@ extern int alloc_contig_range(unsigned long start,
> unsigned long end,
> unsigned migratetype);
> extern void free_contig_range(unsigned long pfn, unsigned nr_pages);
>
> -/* CMA stuff */
> -extern void init_cma_reserved_pageblock(struct page *page);
> -
> #endif
>
> #endif /* __LINUX_GFP_H */
> diff --git a/mm/cma.c b/mm/cma.c
> index c17751c..843b2b6 100644
> --- a/mm/cma.c
> +++ b/mm/cma.c
> @@ -28,11 +28,14 @@
> #include <linux/err.h>
> #include <linux/mm.h>
> #include <linux/mutex.h>
> +#include <linux/page-isolation.h>
> #include <linux/sizes.h>
> #include <linux/slab.h>
> #include <linux/log2.h>
> #include <linux/cma.h>
>
> +#include "internal.h"
> +
> struct cma {
> unsigned long base_pfn;
> unsigned long count;
> @@ -83,37 +86,81 @@ static void cma_clear_bitmap(struct cma *cma, unsigned
> long pfn, int count)
> mutex_unlock(&cma->lock);
> }
>
> +/* Free whole pageblock and set its migration type to MIGRATE_CMA. */
> +static int __init cma_init_reserved_pageblock(struct zone *zone,
> + unsigned long pageblock_pfn)
> +{
> + unsigned long pfn, nr_pages, i;
> + struct page *page, *p;
> + unsigned order;
> +
> + pfn = pageblock_pfn;
> + if (!pfn_valid(pfn))
> + goto invalid_pfn;
> + page = pfn_to_page(pfn);
> +
> + p = page;
> + i = pageblock_nr_pages;
> + do {
> + if (!pfn_valid(pfn))
> + goto invalid_pfn;
> +
> + /*
> + * alloc_contig_range requires the pfn range specified to be
> + * in the same zone. Make this simple by forcing the entire
> + * CMA resv range to be in the same zone.
> + */
> + if (page_zone(p) != zone) {
> + pr_err("pfn %lu belongs to %s, expecting %s\n",
> + pfn, page_zone(p)->name, zone->name);
> + return -EINVAL;
> + }
> +
> + __ClearPageReserved(p);
> + set_page_count(p, 0);
> + } while (++p, ++pfn, --i);
So, when we meet fail condition, __ClearPageReserved, set_page_count()
are already executed for some pages. Is that no problem?
Thanks.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
