Messages in this thread |  | | Subject | Re: [RFC][PATCH] slub: Check for page NULL before doing the node_match check | From | Steven Rostedt <> | Date | Thu, 17 Jan 2013 18:10:45 -0500 |
| |
On Thu, 2013-01-17 at 17:46 -0500, Steven Rostedt wrote: > On Thu, 2013-01-17 at 21:51 +0000, Christoph Lameter wrote: > > > This is dealing with the same cpu being interrupted. Some of these > > segments are in interrupt disable sections so they are not affected. > > Except that we are not always on the same CPU. Now I'm looking at > mainline (non modified by -rt):
Because there's also nothing to keep page related to object either, we may just need to do:
> > From slab_alloc_node(): > > /* > * Must read kmem_cache cpu data via this cpu ptr. Preemption is > * enabled. We may switch back and forth between cpus while > * reading from one cpu area. That does not matter as long > * as we end up on the original cpu again when doing the cmpxchg. > */
local_irq_save(flags);
> c = __this_cpu_ptr(s->cpu_slab); > > /* > * The transaction ids are globally unique per cpu and per operation on > * a per cpu queue. Thus they can be guarantee that the cmpxchg_double > * occurs on the right processor and that there was no operation on the > * linked list in between. > */ > tid = c->tid; > barrier(); > > object = c->freelist; > page = c->page;
r = !object || !node_match(page, node); local_irq_restore(flags);
if (unlikely(r)) {
> object = __slab_alloc(s, gfpflags, node, addr, c); >
I was thinking at first we could use preempt_disable(), but if an interrupt comes in after we set object = c->freelist, and changes c->page, then we disassociate the freelist and page again.
-- Steve
|  |