lkml.org 
[lkml]   [2018]   [Oct]   [19]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
From
Subject[RFC PATCH 2/2] mm: add priority threshold to __purge_vmap_area_lazy()
Date
commit 763b218ddfaf ("mm: add preempt points into
__purge_vmap_area_lazy()")

introduced some preempt points, one of those is making
an allocation more prioritized.

Prioritizing an allocation over freeing does not work
well all the time, i.e. it should be rather a compromise.

1) Number of lazy pages directly influence on busy list
length thus on operations like: allocation, lookup, unmap,
remove, etc.

2) Under heavy simultaneous allocations/releases there may
be a situation when memory usage grows too fast hitting
out_of_memory -> panic.

Establish a threshold passing which the freeing path is
prioritized over allocation creating a balance between both.

Signed-off-by: Uladzislau Rezki (Sony) <urezki@gmail.com>
---
mm/vmalloc.c | 14 ++++++++------
1 file changed, 8 insertions(+), 6 deletions(-)

diff --git a/mm/vmalloc.c b/mm/vmalloc.c
index a7f257540a05..bbafcff6632b 100644
--- a/mm/vmalloc.c
+++ b/mm/vmalloc.c
@@ -1124,23 +1124,23 @@ static bool __purge_vmap_area_lazy(unsigned long start, unsigned long end)
struct llist_node *valist;
struct vmap_area *va;
struct vmap_area *n_va;
- bool do_free = false;
+ int resched_threshold;

lockdep_assert_held(&vmap_purge_lock);

valist = llist_del_all(&vmap_purge_list);
+ if (unlikely(valist == NULL))
+ return false;
+
llist_for_each_entry(va, valist, purge_list) {
if (va->va_start < start)
start = va->va_start;
if (va->va_end > end)
end = va->va_end;
- do_free = true;
}

- if (!do_free)
- return false;
-
flush_tlb_kernel_range(start, end);
+ resched_threshold = (int) lazy_max_pages() << 1;

spin_lock(&vmap_area_lock);
llist_for_each_entry_safe(va, n_va, valist, purge_list) {
@@ -1148,7 +1148,9 @@ static bool __purge_vmap_area_lazy(unsigned long start, unsigned long end)

__free_vmap_area(va);
atomic_sub(nr, &vmap_lazy_nr);
- cond_resched_lock(&vmap_area_lock);
+
+ if (atomic_read(&vmap_lazy_nr) < resched_threshold)
+ cond_resched_lock(&vmap_area_lock);
}
spin_unlock(&vmap_area_lock);
return true;
--
2.11.0
\
 
 \ /
  Last update: 2018-10-19 19:37    [W:0.152 / U:2.676 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site