[Devel] [PATCH RH9 03/28] net/skbuff: Don't waste memory reserves

[Andrey Zhadchenko]

From: Andrey Ryabinin <aryabinin at virtuozzo.com>

We were observing network performance issues due to packets
being dropped by sk_filter_trim_cap() since the 'skb' was allocated
from pfmemalloc reserves:

    /*
     * If the skb was allocated from pfmemalloc reserves, only
     * allow SOCK_MEMALLOC sockets to use it as this socket is
     * helping free memory
     */
    if (skb_pfmemalloc(skb) && !sock_flag(sk, SOCK_MEMALLOC))
        return -ENOMEM;

Memalloc sockets are used for stuff like swap over NBD or NFS
and only memalloc sockets can process memalloc skbs. Since we
don't have any memalloc sockets in our setups we shouldn't have
memalloc skbs either. It simply doesn't make any sense to waste
memory reserves on skb only to drop packets later.

It appears that __dev_alloc_pages() unconditionally uses
__GFP_MEMALLOC, so unless caller added __GFP_NOMEMALLOC, the
__dev_alloc_pages() may dive into memory reserves.
Later build_skb() or __skb_fill_page_desc() sets skb->pfmemalloc = 1
so this skb always dropped by sk_filter_trim_cap().

Instead of wasting memory reserves we should simply avoid using
them in case of absence memalloc sockets in the system. Do this
by adding __GFP_MEMALLOC only when such socket is present in the
system.

https://pmc.acronis.com/browse/VSTOR-21390
Signed-off-by: Andrey Ryabinin <aryabinin at virtuozzo.com>

Cherry-picked from vz7 commit a6b8d2a78d83 ("net/skbuff: Don't waste memory
reserves")

Ported to VZ8 in scope of https://jira.sw.ru/browse/PSBM-127844.
Signed-off-by: Evgenii Shatokhin <eshatokhin at virtuozzo.com>

(cherry picked from vz8 commit 4253f1e95a73d4fcb10d56e199a0ee27b129d910)
Signed-off-by: Andrey Zhadchenko <andrey.zhadchenko at virtuozzo.com>
---
 include/linux/skbuff.h | 19 ++++++++++++++++++-
 include/net/sock.h     | 12 ------------
 2 files changed, 18 insertions(+), 13 deletions(-)

diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
index b2db9cd..03da2d2 100644
--- a/include/linux/skbuff.h
+++ b/include/linux/skbuff.h
@@ -2933,6 +2933,21 @@ static inline struct sk_buff *napi_alloc_skb(struct napi_struct *napi,
 void napi_skb_free_stolen_head(struct sk_buff *skb);
 void __kfree_skb_defer(struct sk_buff *skb);
 
+#ifdef CONFIG_NET
+DECLARE_STATIC_KEY_FALSE(memalloc_socks_key);
+static inline int sk_memalloc_socks(void)
+{
+	return static_branch_unlikely(&memalloc_socks_key);
+}
+#else
+
+static inline int sk_memalloc_socks(void)
+{
+	return 0;
+}
+
+#endif
+
 /**
  * __dev_alloc_pages - allocate page for network Rx
  * @gfp_mask: allocation priority. Set __GFP_NOMEMALLOC if not for network Rx
@@ -2953,7 +2968,9 @@ static inline struct page *__dev_alloc_pages(gfp_t gfp_mask,
 	 * 4.  __GFP_MEMALLOC is ignored if __GFP_NOMEMALLOC is set due to
 	 *     code in gfp_to_alloc_flags that should be enforcing this.
 	 */
-	gfp_mask |= __GFP_COMP | __GFP_MEMALLOC;
+	gfp_mask |= __GFP_COMP;
+	if (sk_memalloc_socks())
+		gfp_mask |= __GFP_MEMALLOC;
 
 	return alloc_pages_node(NUMA_NO_NODE, gfp_mask, order);
 }
diff --git a/include/net/sock.h b/include/net/sock.h
index 814e7d7..3a3ba11 100644
--- a/include/net/sock.h
+++ b/include/net/sock.h
@@ -905,20 +905,8 @@ static inline bool sock_flag(const struct sock *sk, enum sock_flags flag)
 }
 
 #ifdef CONFIG_NET
-DECLARE_STATIC_KEY_FALSE(memalloc_socks_key);
-static inline int sk_memalloc_socks(void)
-{
-	return static_branch_unlikely(&memalloc_socks_key);
-}
-
 void __receive_sock(struct file *file);
 #else
-
-static inline int sk_memalloc_socks(void)
-{
-	return 0;
-}
-
 static inline void __receive_sock(struct file *file)
 { }
 #endif
-- 
1.8.3.1