Skip to contentSkip to conversation

make vm_enough_memory more efficient

PATCH In review Martin J. Bligh 2 participants 3 messages Jan 16, 2003

Subsystems (2)
Memory management Memory mapping
lore
Share & tools
lore

Martin J. Bligh

author
#0 lore raw

PATCH 1/1 make vm_enough_memory more efficient

vm_enough_memory seems to call si_meminfo just to get the total RAM, which seems far too expensive. This replaces the comment saying "this is crap" with some code that's less crap.

Not heavily tested (compiles and boots), but seems pretty obvious.

M.

mm/mmap.c +1 -7
diff -urpN -X /home/fletch/.diff.exclude virgin/mm/mmap.c vm_enough_memory/mm/mmap.c
@@ -72,7 +72,6 @@ inline void vm_unacct_memory(long pages)
 int vm_enough_memory(long pages)
 {
 	unsigned long free, allowed;
-	struct sysinfo i;
 
 	atomic_add(pages, &vm_committed_space);
 

@@ -113,12 +112,7 @@ int vm_enough_memory(long pages)
 		return 0;
 	}
 
-	/*
-	 * FIXME: need to add arch hooks to get the bits we need
-	 * without this higher overhead crap
-	 */
-	si_meminfo(&i);
-	allowed = i.totalram * sysctl_overcommit_ratio / 100;
+	allowed = totalram_pages * sysctl_overcommit_ratio / 100;
 	allowed += total_swap_pages;
 
 	if (atomic_read(&vm_committed_space) < allowed)
#1 lore raw
Quoted 6 lines · Martin J. Bligh

vm_enough_memory seems to call si_meminfo just to get the total RAM, which seems far too expensive. This replaces the comment saying "this is crap" with some code that's less crap.

Not heavily tested (compiles and boots), but seems pretty obvious.

Yup, obviously correct.

The really hurtful part of vm_enough_memory() is the call to get_page_cache_size(), which has to go over every CPU's local VM statistics in get_page_state().

But I guess you're running with sysctl_overcommit_memory != 0.

author
#2 lore raw
Quoted 7 lines
Quoted 5 lines

vm_enough_memory seems to call si_meminfo just to get the total RAM, which seems far too expensive. This replaces the comment saying "this is crap" with some code that's less crap.

Not heavily tested (compiles and boots), but seems pretty obvious.

Yup, obviously correct.

Cool.

Quoted 5 lines

The really hurtful part of vm_enough_memory() is the call to get_page_cache_size(), which has to go over every CPU's local VM statistics in get_page_state().

But I guess you're running with sysctl_overcommit_memory != 0.

Yup, I manually disable that because it's so expensive. I'll see if I can make the default case cheaper as well.

M.