On Tue, 17 Jul 2007, Andrew Morton wrote:
> On Tue, 17 Jul 2007 18:26:14 +0100 (BST) Hugh Dickins <[EMAIL PROTECTED]>
> wrote:
> > > I think I like the lock ;)
> >
> > I hate to waste your time, but I'm still puzzled. Wasn't the race fixed
> > by your changeover from use of "static int nid" throughout, to setting
> > local "int nid" from "static int prev_nid", working with nid, then
> > setting prev_nid from nid at the end? What does the lock add to that?
>
> There are still minor races without the lock - two CPUs will allocate
> from the first node, and prev_nid can occasionally go backwards.
>
> I agree that they are sufficiently minor that we could remove the lock.
Phew, yes, that's what I meant - thanks.
[PATCH] Remove nid_lock from alloc_fresh_huge_page
The fix to the race in alloc_fresh_huge_page() which could give an illegal
node ID did not need nid_lock at all: the fix was to replace static int nid
by static int prev_nid and do the work on local int nid. nid_lock did make
sure that racers strictly roundrobin the nodes, but that's not something we
actually need to enforce strictly. Kill that nid_lock.
Signed-off-by: Hugh Dickins <[EMAIL PROTECTED]>
---
mm/hugetlb.c | 3 ---
1 file changed, 3 deletions(-)
--- 2.6.22-git9/mm/hugetlb.c 2007-07-17 20:29:33.000000000 +0100
+++ linux/mm/hugetlb.c 2007-07-17 20:32:51.000000000 +0100
@@ -107,15 +107,12 @@ static int alloc_fresh_huge_page(void)
{
static int prev_nid;
struct page *page;
- static DEFINE_SPINLOCK(nid_lock);
int nid;
- spin_lock(&nid_lock);
nid = next_node(prev_nid, node_online_map);
if (nid == MAX_NUMNODES)
nid = first_node(node_online_map);
prev_nid = nid;
- spin_unlock(&nid_lock);
page = alloc_pages_node(nid, htlb_alloc_mask|__GFP_COMP|__GFP_NOWARN,
HUGETLB_PAGE_ORDER);
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
