Messages in this thread Patch in this message | | | Date | Wed, 10 Jan 2001 21:57:01 +0100 | From | Rasmus Andersen <> | Subject | [uPATCH] Nulling mmap_cache iff needed (2.4.0) |
| |
Hi.
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/
| |