| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Date: Sat, 25 Apr 2009 21:54:59 +0900
From: Daisuke Nishimura <d-nishimura@....biglobe.ne.jp>
To: KAMEZAWA Hiroyuki <kamezawa.hiroyu@...fujitsu.com>
Cc: "linux-mm@...ck.org" <linux-mm@...ck.org>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
"balbir@...ux.vnet.ibm.com" <balbir@...ux.vnet.ibm.com>,
"hugh@...itas.com" <hugh@...itas.com>,
d-nishimura@....biglobe.ne.jp,
Daisuke Nishimura <nishimura@....nes.nec.co.jp>
Subject: Re: [RFC][PATCH] fix swap entries is not reclaimed in proper way
for memg v3.
> +static void memcg_fixup_stale_swapcache(struct work_struct *work)
> +{
> + int pos = 0;
> + swp_entry_t entry;
> + struct page *page;
> + int forget, ret;
> +
> + while (ssc.num) {
> + spin_lock(&ssc.lock);
> + pos = find_next_bit(ssc.usemap, STALE_ENTS, pos);
> + spin_unlock(&ssc.lock);
> +
> + if (pos >= STALE_ENTS)
> + break;
> +
> + entry = ssc.ents[pos];
> +
> + forget = 1;
> + page = lookup_swap_cache(entry);
> + if (page) {
> + lock_page(page);
> + ret = try_to_free_swap(page);
> + /* If it's still under I/O, don't forget it */
> + if (!ret && PageWriteback(page))
> + forget = 0;
> + unlock_page(page);
I think we need page_cache_release().
lookup_swap_cache() gets the page.
> + }
> + if (forget) {
> + spin_lock(&ssc.lock);
> + clear_bit(pos, ssc.usemap);
> + ssc.num--;
> + if (ssc.num < STALE_ENTS/2)
> + ssc.congestion = 0;
> + spin_unlock(&ssc.lock);
> + }
> + pos++;
> + }
> + if (ssc.num) /* schedule me again */
> + schedule_delayed_work(&ssc.gc_work, HZ/10);
"if (ssc.congestion)" would be better ?
> + return;
> +}
> +
(snip)
> Index: mmotm-2.6.30-Apr21/mm/vmscan.c
> ===================================================================
> --- mmotm-2.6.30-Apr21.orig/mm/vmscan.c
> +++ mmotm-2.6.30-Apr21/mm/vmscan.c
> @@ -661,6 +661,9 @@ static unsigned long shrink_page_list(st
> if (PageAnon(page) && !PageSwapCache(page)) {
> if (!(sc->gfp_mask & __GFP_IO))
> goto keep_locked;
> + /* avoid making more stale swap caches */
> + if (memcg_stale_swap_congestion())
> + goto keep_locked;
> if (!add_to_swap(page))
> goto activate_locked;
> may_enter_fs = 1;
>
Hmm, I don't think this can avoid type-2 stale swap caches.
IIUC, this can only avoid add_to_swap() if the number of stale swap caches
exceeds some threshold, but type-2 swap caches(set !PageCgroupUsed by the
owner process via page_remove_rmap()->mem_cgroup_uncharge_page() before
beeing add to swap cache) is not handled as 'stale'.
In fact, I can see the usage of SwapCache increasing gradually.
Can you add a patch like bellow ?
Thanks,
Daisuke Nishimura.
===
From: Daisuke Nishimura <nishimura@....nes.nec.co.jp>
Instead of checking memcg_stale_swap_congestion() before add_to_swap(),
free "unused" swap cache after add_to_swap().
IMHO, it would be better to let shrink_page_list() free as much pages
as possible, so free type-2 stale swap caches directly, instead of
handling them in lazy manner.
shrink_page_list() calls try_to_free_swap() already in some paths.
(e.g. pageout()->swap_writepage()->try_to_free_swap())
Signed-off-by: Daisuke Nishimura <nishimura@....nes.nec.co.jp>
---
diff --git a/include/linux/swap.h b/include/linux/swap.h
index 1e6519c..51c6985 100644
--- a/include/linux/swap.h
+++ b/include/linux/swap.h
@@ -339,6 +339,7 @@ extern void mem_cgroup_uncharge_swapcache(struct page *page, swp_entry_t ent);
extern void memcg_mark_swapent_stale(swp_entry_t ent);
extern void memcg_sanity_check_swapin(struct page *page, swp_entry_t ent);
extern int memcg_stale_swap_congestion(void);
+extern int memcg_free_unused_swapcache(struct page *page);
#else
static inline void
mem_cgroup_uncharge_swapcache(struct page *page, swp_entry_t ent)
@@ -357,6 +358,11 @@ static inline int memcg_stale_swap_congestion(void)
{
return 0;
}
+
+static int memcg_free_unused_swapcache(struct page *page)
+{
+ return 0;
+}
#endif
#ifdef CONFIG_CGROUP_MEM_RES_CTLR_SWAP
extern void mem_cgroup_uncharge_swap(swp_entry_t ent);
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index ccc69b4..822a914 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -1754,6 +1754,17 @@ static void setup_stale_swapcache_control(void)
INIT_DELAYED_WORK(&ssc.gc_work, memcg_fixup_stale_swapcache);
}
+int memcg_free_unused_swapcache(struct page *page)
+{
+ VM_BUG_ON(!PageSwapCache(page));
+ VM_BUG_ON(!PageLocked(page));
+
+ if (mem_cgroup_disabled())
+ return 0;
+ if (!PageAnon(page) || page_mapped(page))
+ return 0;
+ return try_to_free_swap(page); /* checks page_swapcount */
+}
#else
int memcg_stale_swap_congestion(void)
@@ -1765,6 +1776,10 @@ static void setup_stale_swapcache_control(void)
{
}
+int memcg_free_unused_swapcache(struct page *page)
+{
+ return 0;
+}
#endif /* CONFIG_SWAP */
static DEFINE_MUTEX(set_limit_mutex);
diff --git a/mm/vmscan.c b/mm/vmscan.c
index 054ed38..5b9aa8e 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -654,11 +654,16 @@ static unsigned long shrink_page_list(struct list_head *page_list,
if (PageAnon(page) && !PageSwapCache(page)) {
if (!(sc->gfp_mask & __GFP_IO))
goto keep_locked;
- /* avoid making more stale swap caches */
- if (memcg_stale_swap_congestion())
- goto keep_locked;
if (!add_to_swap(page))
goto activate_locked;
+ /*
+ * The owner process might have uncharged the page
+ * (by page_remove_rmap()) before it has been added
+ * to swap cache.
+ * Check it here to avoid making it stale.
+ */
+ if (memcg_free_unused_swapcache(page))
+ goto keep_locked;
may_enter_fs = 1;
}
===
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists