[RFC PATCH v1 1/1] mm: tuning hardcoded reserved memory

From: Andrew Shewmaker
Date: Mon Feb 25 2013 - 18:23:34 EST

__vm_enough_memory reserves a hardcoded 3% of free memory for other processes
when overcommit is disabled. However, 3% is becoming excessive as memory sizes
increase and memory cgroups provide a more flexible way to manage the issue anyway.
This patch against 3.8 removes the "other" reserve.

I've found that reducing this reserve is beneficial in the case where a system
with overcommit disabled has one primary user that wants to allocate as much
memory as possible with a just few processes.

An additional hardcoded 3% is reserved for root, both when overcommit is enabled
and when it is disabled. I've made it tunable in private patches, and I plan on
submitting some version of them, but I can't decide whether a ratio or a byte
count would be more acceptable. What would people prefer see?

Signed-off-by: Andrew Shewmaker <agshew@xxxxxxxxx>

diff --git a/mm/mmap.c b/mm/mmap.c
index 09da0b2..eef9505 100644
--- a/mm/mmap.c
+++ b/mm/mmap.c
@@ -183,11 +183,6 @@ int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin)
allowed -= allowed / 32;
allowed += total_swap_pages;

- /* Don't let a single process grow too big:
- leave 3% of the size of this process for other processes */
- if (mm)
- allowed -= mm->total_vm / 32;
if (percpu_counter_read_positive(&vm_committed_as) < allowed)
return 0;
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/