On Fri, Feb 25, 2011 at 01:36:00PM -0800, Greg Thelen wrote:
> If the current process is in a non-root memcg, then
> balance_dirty_pages() will consider the memcg dirty limits as well as
> the system-wide limits. This allows different cgroups to have distinct
> dirty limits which trigger direct and background writeback at different
> levels.
>
> If called with a mem_cgroup, then throttle_vm_writeout() should query
> the given cgroup for its dirty memory usage limits.
>
> Signed-off-by: Andrea Righi <arighi@xxxxxxxxxxx>
> Signed-off-by: Greg Thelen <gthelen@xxxxxxxxxx>
> Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@xxxxxxxxxxxxxx>
> Acked-by: Wu Fengguang <fengguang.wu@xxxxxxxxx>
> ---
<snip>
> /*
> @@ -477,12 +502,14 @@ unsigned long bdi_dirty_limit(struct backing_dev_info *bdi, unsigned long dirty)
> * data. It looks at the number of dirty pages in the machine and will force
> * the caller to perform writeback if the system is over `vm_dirty_ratio'.
> * If we're over `background_thresh' then the writeback threads are woken to
> - * perform some writeout.
> + * perform some writeout. The current task may have per-memcg dirty
> + * limits, which are also checked.
> */
> static void balance_dirty_pages(struct address_space *mapping,
> unsigned long write_chunk)
> {
> struct dirty_info sys_info;
> + struct dirty_info memcg_info;
> long bdi_nr_reclaimable;
> long bdi_nr_writeback;
> unsigned long bdi_thresh;
> @@ -500,18 +527,27 @@ static void balance_dirty_pages(struct address_space *mapping,
> };
>
> global_dirty_info(&sys_info);
> + if (!memcg_dirty_info(NULL, &memcg_info))
> + memcg_info = sys_info;
Sigh.
I don't like dobule check in case of no-memcg configuration or root memcgroup.
"Dobule check" means 1) sys_info check and 2) memcg_info check but it's same so
second check is redundant.
In addition, we always need same logic between global and memcg.
It adds binary bloating.
>
> /*
> * Throttle it only when the background writeback cannot
> * catch-up. This avoids (excessively) small writeouts
> * when the bdi limits are ramping up.
> */
> - if (dirty_info_reclaimable(&sys_info) + sys_info.nr_writeback <=
> + if ((dirty_info_reclaimable(&sys_info) +
> + sys_info.nr_writeback <=
> (sys_info.background_thresh +
> - sys_info.dirty_thresh) / 2)
> + sys_info.dirty_thresh) / 2) &&
> + (dirty_info_reclaimable(&memcg_info) +
> + memcg_info.nr_writeback <=
> + (memcg_info.background_thresh +
> + memcg_info.dirty_thresh) / 2))
> break;
>
> - bdi_thresh = bdi_dirty_limit(bdi, sys_info.dirty_thresh);
> + bdi_thresh = bdi_dirty_limit(bdi,
> + min(sys_info.dirty_thresh,
> + memcg_info.dirty_thresh));
> bdi_thresh = task_dirty_limit(current, bdi_thresh);
>
> /*
> @@ -541,7 +577,9 @@ static void balance_dirty_pages(struct address_space *mapping,
> dirty_exceeded =
> (bdi_nr_reclaimable + bdi_nr_writeback > bdi_thresh)
> || (dirty_info_reclaimable(&sys_info) +
> - sys_info.nr_writeback > sys_info.dirty_thresh);
> + sys_info.nr_writeback > sys_info.dirty_thresh)
> + || (dirty_info_reclaimable(&memcg_info) +
> + memcg_info.nr_writeback > memcg_info.dirty_thresh);
>
> if (!dirty_exceeded)
> break;
> @@ -559,7 +597,9 @@ static void balance_dirty_pages(struct address_space *mapping,
> * up.
> */
> trace_wbc_balance_dirty_start(&wbc, bdi);
> - if (bdi_nr_reclaimable > bdi_thresh) {
> + if ((bdi_nr_reclaimable > bdi_thresh) ||
> + (dirty_info_reclaimable(&memcg_info) >
> + memcg_info.dirty_thresh)) {
Why does memcg need this check?
I guess bdi_thresh is the f(x), x = dirty_thresh and dirty_thresh is already memcg's one.
So isn't it enough?
I don't know the logic well but as I look at the comment, at least you change the behavior.
Please write down why you need it and if you need it, please change comment, too.
"
* Only move pages to writeback if this bdi is over its
* threshold otherwise wait until the disk writes catch
* up.
"
> writeback_inodes_wb(&bdi->wb, &wbc);
> pages_written += write_chunk - wbc.nr_to_write;
> trace_wbc_balance_dirty_written(&wbc, bdi);
> @@ -594,8 +634,10 @@ static void balance_dirty_pages(struct address_space *mapping,
> * background_thresh, to keep the amount of dirty memory low.
> */
> if ((laptop_mode && pages_written) ||
> - (!laptop_mode && (dirty_info_reclaimable(&sys_info) >
> - sys_info.background_thresh)))
> + (!laptop_mode && ((dirty_info_reclaimable(&sys_info) >
> + sys_info.background_thresh) ||
> + (dirty_info_reclaimable(&memcg_info) >
> + memcg_info.background_thresh))))
> bdi_start_background_writeback(bdi);
> }
>
> @@ -653,12 +695,20 @@ void balance_dirty_pages_ratelimited_nr(struct address_space *mapping,
> }
> EXPORT_SYMBOL(balance_dirty_pages_ratelimited_nr);
>
> -void throttle_vm_writeout(gfp_t gfp_mask)
> +/*
> + * Throttle the current task if it is near dirty memory usage limits.
> + * If @mem_cgroup is NULL or the root_cgroup, then use global dirty memory
> + * information; otherwise use the per-memcg dirty limits.
> + */
> +void throttle_vm_writeout(gfp_t gfp_mask, struct mem_cgroup *mem_cgroup)
> {
> struct dirty_info sys_info;
> + struct dirty_info memcg_info;
>
> for ( ; ; ) {
> global_dirty_info(&sys_info);
> + if (!memcg_dirty_info(mem_cgroup, &memcg_info))
> + memcg_info = sys_info;
>
> /*
> * Boost the allowable dirty threshold a bit for page
> @@ -666,9 +716,13 @@ void throttle_vm_writeout(gfp_t gfp_mask)
> */
> sys_info.dirty_thresh +=
> sys_info.dirty_thresh / 10; /* wheeee... */
> + memcg_info.dirty_thresh +=
> + memcg_info.dirty_thresh / 10; /* wheeee... */
>
> - if (sys_info.nr_unstable_nfs +
> - sys_info.nr_writeback <= sys_info.dirty_thresh)
> + if ((sys_info.nr_unstable_nfs +
> + sys_info.nr_writeback <= sys_info.dirty_thresh) &&
> + (memcg_info.nr_unstable_nfs +
> + memcg_info.nr_writeback <= memcg_info.dirty_thresh))
> break;
> congestion_wait(BLK_RW_ASYNC, HZ/10);
>
> diff --git a/mm/vmscan.c b/mm/vmscan.c
> index ba11e28..f723242 100644
> --- a/mm/vmscan.c
> +++ b/mm/vmscan.c
> @@ -1927,7 +1927,7 @@ restart:
> sc->nr_scanned - nr_scanned, sc))
> goto restart;
>
> - throttle_vm_writeout(sc->gfp_mask);
> + throttle_vm_writeout(sc->gfp_mask, sc->mem_cgroup);
> }
>
> /*
> --
> 1.7.3.1
>
--
Kind regards,
Minchan Kim
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@xxxxxxxxxx For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>