Re: [PATCH 8/8] vm: Add an tuning knob for vm.max_writeback_mb

[Richard Kennedy]

On 04/09/09 08:46, Jens Axboe wrote:
> From: Theodore Ts'o<tytso@xxxxxxx>
>
> Originally, MAX_WRITEBACK_PAGES was hard-coded to 1024 because of a
> concern of not holding I_SYNC for too long.  (At least, that was the
> comment previously.)  This doesn't make sense now because the only
> time we wait for I_SYNC is if we are calling sync or fsync, and in
> that case we need to write out all of the data anyway.  Previously
> there may have been other code paths that waited on I_SYNC, but not
> any more.
>
> According to Christoph, the current writeback size is way too small,
> and XFS had a hack that bumped out nr_to_write to four times the value
> sent by the VM to be able to saturate medium-sized RAID arrays.  This
> value was also problematic for ext4 as well, as it caused large files
> to be come interleaved on disk by in 8 megabyte chunks (we bumped up
> the nr_to_write by a factor of two).
>
> So, in this patch, we make the MAX_WRITEBACK_PAGES a tunable,
> max_writeback_mb, and set it to a default value of 128 megabytes.
>
> http://bugzilla.kernel.org/show_bug.cgi?id=13930
>
> Signed-off-by: "Theodore Ts'o"<tytso@xxxxxxx>
> Signed-off-by: Jens Axboe<jens.axboe@xxxxxxxxxx>
> ---
>   fs/fs-writeback.c         |    9 +--------
>   include/linux/writeback.h |    1 +
>   kernel/sysctl.c           |    8 ++++++++
>   mm/page-writeback.c       |    6 ++++++
>   4 files changed, 16 insertions(+), 8 deletions(-)
>
> diff --git a/fs/fs-writeback.c b/fs/fs-writeback.c
> index ce68f60..790d379 100644
> --- a/fs/fs-writeback.c
> +++ b/fs/fs-writeback.c
> @@ -641,14 +641,7 @@ void writeback_inodes_wbc(struct writeback_control *wbc)
>   	writeback_inodes_wb(&bdi->wb, wbc);
>   }
>
> -/*
> - * The maximum number of pages to writeout in a single bdi flush/kupdate
> - * operation.  We do this so we don't hold I_SYNC against an inode for
> - * enormous amounts of time, which would block a userspace task which has
> - * been forced to throttle against that inode.  Also, the code reevaluates
> - * the dirty each time it has written this many pages.
> - */
> -#define MAX_WRITEBACK_PAGES     1024
> +#define MAX_WRITEBACK_PAGES	(max_writeback_mb<<  (20 - PAGE_SHIFT))
>
>   static inline bool over_bground_thresh(void)
>   {
> diff --git a/include/linux/writeback.h b/include/linux/writeback.h
> index 78b1e46..fbed759 100644
> --- a/include/linux/writeback.h
> +++ b/include/linux/writeback.h
> @@ -104,6 +104,7 @@ extern int vm_dirty_ratio;
>   extern unsigned long vm_dirty_bytes;
>   extern unsigned int dirty_writeback_interval;
>   extern unsigned int dirty_expire_interval;
> +extern unsigned int max_writeback_mb;
>   extern int vm_highmem_is_dirtyable;
>   extern int block_dump;
>   extern int laptop_mode;
> diff --git a/kernel/sysctl.c b/kernel/sysctl.c
> index 58be760..315fc30 100644
> --- a/kernel/sysctl.c
> +++ b/kernel/sysctl.c
> @@ -1104,6 +1104,14 @@ static struct ctl_table vm_table[] = {
>   		.proc_handler	=&proc_dointvec,
>   	},
>   	{
> +		.ctl_name	= CTL_UNNUMBERED,
> +		.procname	= "max_writeback_mb",
> +		.data		=&max_writeback_mb,
> +		.maxlen		= sizeof(max_writeback_mb),
> +		.mode		= 0644,
> +		.proc_handler	=&proc_dointvec,
> +	},
> +	{
>   		.ctl_name	= VM_NR_PDFLUSH_THREADS,
>   		.procname	= "nr_pdflush_threads",
>   		.data		=&nr_pdflush_threads,
> diff --git a/mm/page-writeback.c b/mm/page-writeback.c
> index 2c287d9..38fe4e8 100644
> --- a/mm/page-writeback.c
> +++ b/mm/page-writeback.c
> @@ -55,6 +55,12 @@ static inline long sync_writeback_pages(void)
>   /* The following parameters are exported via /proc/sys/vm */
>
>   /*
> + * The maximum amount of memory (in megabytes) to write out in a
> + * single bdflush/kupdate operation.
> + */
> +unsigned int max_writeback_mb = 128;
> +
> +/*
>    * Start background writeback (via pdflush) at this percentage
>    */
>   int dirty_background_ratio = 10;

Hi Jens,

I've been testing this & it works pretty well here, but setting 
max_writeback_mb to 128 seems much too large for normal desktop machines.

Because it is so large the background writes don't stop when they get 
down to the background threshold, but just keep on writing. 
background_threshold on my machine is only about 300Mb so it can 
undershoot by quite a bit. This could impact random write workloads 
significantly.

Would making the tunable a percentage of dirty_threshold be better for 
most people? At least it would scale with the size of the system memory. 
I'm guessing that machines with RAID arrays also have large memories.

Or can the check for the background threshold be pushed further down 
into writeback_inodes_wb and just check it every N pages? I think this 
would do a better job but make the code even more complex.

regards
Richard
--
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/