Le jeudi 12 janvier 2012 à 15:30 -0800, [email protected] a écrit : > The patch below does not apply to the 3.0-stable tree. > If someone wants it applied there, or to any other stable or longterm > tree, then please email the backport, including the original git commit > id to <[email protected]>. > > thanks, > > greg k-h > > ------------------ original commit in Linus's tree ------------------ > > From 73736e0387ba0e6d2b703407b4d26168d31516a7 Mon Sep 17 00:00:00 2001 > From: Eric Dumazet <[email protected]> > Date: Tue, 13 Dec 2011 04:57:06 +0100 > Subject: [PATCH] slub: fix a possible memleak in __slab_alloc()
Hi Greg, here is the backported patch for 3.0, sorry for the delay. I can exhaust memory (~10 pages/second) using "hackbench 10 thread 4000000" on my server, so this leak is real... Thanks ! [PATCH] slub: fix a possible memleak in __slab_alloc() commit 73736e0387ba0e6d2b703407b4d26168d31516a7 upstream. Zhihua Che reported a possible memleak in slub allocator on CONFIG_PREEMPT=y builds. It is possible current thread migrates right before disabling irqs in __slab_alloc(). We must check again c->freelist, and perform a normal allocation instead of scratching c->freelist. Many thanks to Zhihua Che for spotting this bug, introduced in 2.6.39 Cc: <[email protected]> [2.6.39+] Reported-by: Zhihua Che <[email protected]> Signed-off-by: Eric Dumazet <[email protected]> Acked-by: Christoph Lameter <[email protected]> Signed-off-by: Pekka Enberg <[email protected]> --- mm/slub.c | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/mm/slub.c b/mm/slub.c index 35f351f..0d0901e 100644 --- a/mm/slub.c +++ b/mm/slub.c @@ -1818,6 +1818,11 @@ static void *__slab_alloc(struct kmem_cache *s, gfp_t gfpflags, int node, if (unlikely(!node_match(c, node))) goto another_slab; + /* must check again c->freelist in case of cpu migration or IRQ */ + object = c->freelist; + if (object) + goto update_freelist; + stat(s, ALLOC_REFILL); load_freelist: @@ -1827,6 +1832,7 @@ load_freelist: if (kmem_cache_debug(s)) goto debug; +update_freelist: c->freelist = get_freepointer(s, object); page->inuse = page->objects; page->freelist = NULL; -- To unsubscribe from this list: send the line "unsubscribe stable" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html
