The following (trivial) patch against 2.4.0 (but should apply cleanly
against 241p1) makes do_munmap only null our precious cache if the
pointer contained therein is invalidated by the unmapping. I doubt it
will make any measurable difference but it seems straightforward
enough to be done anyway.
Running kernel compiles and starting X/gnome with this patch raises my
cache hit rate a couple of percentage points, and I have a feeling that
there might be work loads out there that hit this kernel path more often
that these.
Comments?
diff -aur linux-2.4.0-clean/mm/mmap.c linux/mm/mmap.c
--- linux-2.4.0-clean/mm/mmap.c Sat Dec 30 18:35:19 2000
+++ linux/mm/mmap.c Tue Jan 9 23:30:03 2001
@@ -712,7 +712,10 @@
if (mm->mmap_avl)
avl_remove(mpnt, &mm->mmap_avl);
}
- mm->mmap_cache = NULL; /* Kill the cache. */
+ if (mm->mmap_cache && mm->mmap_cache->vm_start < addr+len
+ && mm->mmap_cache->vm_end > addr)
+ mm->mmap_cache = NULL; /* Kill the cache. */
+
spin_unlock(&mm->page_table_lock);
/* Ok - we have the memory areas we should free on the 'free' list,
-- Regards, Rasmus(rasmus@jaquet.dk)"It's like an Alcatraz around my neck." -Boston mayor Menino on the shortage of city parking spaces - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org Please read the FAQ at http://www.tux.org/lkml/