Re: [PATCH]: Lower the amount of pages used for buffering on hibernation

[Rafael J. Wysocki]

Hi,

On Saturday, March 17, 2012, Bojan Smojver wrote:
> Hi Rafael,
> 
> The patch below reduces the amount of pages that can used by hibernation
> code for buffering. Previously, we would take half of what was available
> as a starting point, now only a quarter. And, we clamp the value to an
> absolute maximum, just to be sure.
> 
> Also, the calculation of the maximum number of pages used for buffering
> does not depend on the size of the sector type, which it inadvertently
> did before.
> 
> In my testing, this had negligible effect on performance, but may
> provide more pages to other kernel needs during hibernation.

Looks good, please prepare a signed-off version.

Thanks,
Rafael


> ---------------------------------------
>  kernel/power/swap.c |   15 +++++++++++----
>  1 files changed, 11 insertions(+), 4 deletions(-)
> 
> diff --git a/kernel/power/swap.c b/kernel/power/swap.c
> index 8742fd0..6212351 100644
> --- a/kernel/power/swap.c
> +++ b/kernel/power/swap.c
> @@ -51,6 +51,9 @@
>  
>  #define MAP_PAGE_ENTRIES	(PAGE_SIZE / sizeof(sector_t) - 1)
>  
> +/* Maximum number of pages for write buffering. */
> +#define BUF_WRITE_PAGES		(1 << 15)
> +
>  struct swap_map_page {
>  	sector_t entries[MAP_PAGE_ENTRIES];
>  	sector_t next_swap;
> @@ -316,7 +319,9 @@ static int get_swap_writer(struct swap_map_handle *handle)
>  		goto err_rel;
>  	}
>  	handle->k = 0;
> -	handle->nr_free_pages = nr_free_pages() >> 1;
> +	handle->nr_free_pages = nr_free_pages() >> 2;
> +	handle->nr_free_pages = clamp_val(handle->nr_free_pages,
> +	                                  1, BUF_WRITE_PAGES);
>  	handle->written = 0;
>  	handle->first_sector = handle->cur_swap;
>  	return 0;
> @@ -404,7 +409,7 @@ static int swap_writer_finish(struct swap_map_handle *handle,
>  #define LZO_THREADS	3
>  
>  /* Maximum number of pages for read buffering. */
> -#define LZO_READ_PAGES	(MAP_PAGE_ENTRIES * 8)
> +#define LZO_READ_PAGES	(1 << 12)
>  
>  
>  /**
> @@ -618,7 +623,9 @@ static int save_image_lzo(struct swap_map_handle *handle,
>  	 * Adjust number of free pages after all allocations have been done.
>  	 * We don't want to run out of pages when writing.
>  	 */
> -	handle->nr_free_pages = nr_free_pages() >> 1;
> +	handle->nr_free_pages = nr_free_pages() >> 2;
> +	handle->nr_free_pages = clamp_val(handle->nr_free_pages,
> +	                                  1, BUF_WRITE_PAGES);
>  
>  	/*
>  	 * Start the CRC32 thread.
> @@ -1130,7 +1137,7 @@ static int load_image_lzo(struct swap_map_handle *handle,
>  	/*
>  	 * Adjust number of pages for read buffering, in case we are short.
>  	 */
> -	read_pages = (nr_free_pages() - snapshot_get_image_size()) >> 1;
> +	read_pages = (nr_free_pages() - snapshot_get_image_size()) >> 2;
>  	read_pages = clamp_val(read_pages, LZO_CMP_PAGES, LZO_READ_PAGES);
>  
>  	for (i = 0; i < read_pages; i++) {
> ---------------------------------------
> 
> 

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/