From: Hideo AOKI <haoki@redhat.com>
To: Herbert Xu <herbert@gondor.apana.org.au>,
netdev <netdev@vger.kernel.org>
Cc: David Miller <davem@davemloft.net>,
Satoshi Oshima <satoshi.oshima.fk@hitachi.com>,
Bill Fink <billfink@mindspring.com>,
Andi Kleen <andi@firstfloor.org>,
Evgeniy Polyakov <johnpol@2ka.mipt.ru>,
Stephen Hemminger <shemminger@linux-foundation.org>,
yoshfuji@linux-ipv6.org,
Yumiko Sugita <yumiko.sugita.yf@hitachi.com>,
haoki@redhat.com
Subject: [PATCH 2/4] datagram: mem_scheudle functions
Date: Wed, 28 Nov 2007 13:52:59 -0500 [thread overview]
Message-ID: <474DB90B.9090107@redhat.com> (raw)
In-Reply-To: <474DB80E.5070403@redhat.com>
This patch introduces datagram memory accounting functions. Owing to
call memory schedule functions from IP layer, sk_wmem_schedule() is
also added.
Cc: Satoshi Oshima <satoshi.oshima.fk@hitachi.com>
signed-off-by: Hideo Aoki <haoki@redhat.com>
---
include/net/sock.h | 31 +++++++++++++++++++++++++++++++
net/core/datagram.c | 34 ++++++++++++++++++++++++++++++++++
2 files changed, 65 insertions(+)
diff -pruN net-2.6-udp-take9a2-p1/include/net/sock.h net-2.6-udp-take9a2-p2/include/net/sock.h
--- net-2.6-udp-take9a2-p1/include/net/sock.h 2007-11-20 10:29:40.000000000 -0500
+++ net-2.6-udp-take9a2-p2/include/net/sock.h 2007-11-27 11:11:38.000000000 -0500
@@ -778,6 +778,37 @@ static inline int sk_stream_wmem_schedul
sk_stream_mem_schedule(sk, size, 0);
}
+extern int sk_datagram_mem_schedule(struct sock *sk, int size, int kind);
+
+#define SK_DATAGRAM_MEM_QUANTUM ((unsigned int)PAGE_SIZE)
+
+static inline int sk_datagram_pages(int amt)
+{
+ /* Cast to unsigned as an optimization, since amt is always positive. */
+ return DIV_ROUND_UP((unsigned int)amt, SK_DATAGRAM_MEM_QUANTUM);
+}
+
+static inline int sk_datagram_rmem_schedule(struct sock *sk,
+ struct sk_buff *skb)
+{
+ return sk_datagram_mem_schedule(sk, skb->truesize, 1);
+}
+
+static inline int sk_datagram_wmem_schedule(struct sock *sk, int size)
+{
+ return sk_datagram_mem_schedule(sk, size, 0);
+}
+
+static inline int sk_wmem_schedule(struct sock *sk, int size)
+{
+ if (sk->sk_type == SOCK_STREAM)
+ return sk_stream_wmem_schedule(sk, size);
+ else if (sk->sk_type == SOCK_DGRAM)
+ return sk_datagram_wmem_schedule(sk, size);
+ else
+ return 1;
+}
+
/* Used by processes to "lock" a socket state, so that
* interrupts and bottom half handlers won't change it
* from under us. It essentially blocks any incoming
diff -pruN net-2.6-udp-take9a2-p1/net/core/datagram.c net-2.6-udp-take9a2-p2/net/core/datagram.c
--- net-2.6-udp-take9a2-p1/net/core/datagram.c 2007-11-14 10:49:06.000000000 -0500
+++ net-2.6-udp-take9a2-p2/net/core/datagram.c 2007-11-28 12:10:05.000000000 -0500
@@ -484,6 +484,39 @@ fault:
}
/**
+ * sk_datagram_mem_schedule - memory accounting for datagram protocls
+ * @sk: socket
+ * @size: memory size to allocate
+ * @kind: allocation type
+ *
+ * If kind is 0, it means wmem allocation. Otherwise it means rmem
+ * allocation.
+ */
+int sk_datagram_mem_schedule(struct sock *sk, int size, int kind)
+{
+ int amt;
+ struct proto *prot = sk->sk_prot;
+
+ /* Don't account and limit memory if protocol doesn't support. */
+ if (prot->memory_allocated == NULL)
+ return 1;
+
+ amt = sk_datagram_pages(size);
+ atomic_add(amt, prot->memory_allocated);
+ if (kind &&
+ (atomic_read(prot->memory_allocated) < prot->sysctl_mem[0] ||
+ atomic_read(&sk->sk_rmem_alloc) + size < prot->sysctl_rmem[0]))
+ return 1;
+ else if (atomic_read(prot->memory_allocated) < prot->sysctl_mem[0] ||
+ atomic_read(&sk->sk_wmem_alloc) + size < prot->sysctl_wmem[0])
+ return 1;
+
+ /* Undo changes. */
+ atomic_sub(amt, prot->memory_allocated);
+ return 0;
+}
+
+/**
* datagram_poll - generic datagram poll
* @file: file struct
* @sock: socket
@@ -542,3 +575,4 @@ EXPORT_SYMBOL(skb_copy_and_csum_datagram
EXPORT_SYMBOL(skb_copy_datagram_iovec);
EXPORT_SYMBOL(skb_free_datagram);
EXPORT_SYMBOL(skb_recv_datagram);
+EXPORT_SYMBOL(sk_datagram_mem_schedule);
--
Hitachi Computer Products (America) Inc.
next prev parent reply other threads:[~2007-11-28 18:54 UTC|newest]
Thread overview: 17+ messages / expand[flat|nested] mbox.gz Atom feed top
2007-11-28 18:48 [PATCH 0/4] UDP memory accounting and limitation (take 9) Hideo AOKI
2007-11-28 18:52 ` [PATCH 1/4] udp: fix send buffer check Hideo AOKI
2007-11-28 18:52 ` Hideo AOKI [this message]
2007-12-01 12:09 ` [PATCH 2/4] datagram: mem_scheudle functions Herbert Xu
2007-12-04 0:10 ` Hideo AOKI
2007-12-15 14:45 ` Herbert Xu
2007-12-18 17:02 ` Hideo AOKI
2007-11-28 18:53 ` [PATCH 3/4] udp: add udp_mem, udp_rmem_min and udp_wmem_min Hideo AOKI
2007-11-28 18:53 ` [PATCH 4/4] udp: memory accounting in IPv4 Hideo AOKI
2007-12-01 12:21 ` Herbert Xu
2007-12-01 13:08 ` Eric Dumazet
2007-12-01 13:16 ` Herbert Xu
2007-12-04 0:14 ` Hideo AOKI
2007-12-04 0:26 ` Herbert Xu
2007-12-06 4:28 ` Hideo AOKI
2007-12-10 9:22 ` Herbert Xu
2007-12-11 1:28 ` Hideo AOKI
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=474DB90B.9090107@redhat.com \
--to=haoki@redhat.com \
--cc=andi@firstfloor.org \
--cc=billfink@mindspring.com \
--cc=davem@davemloft.net \
--cc=herbert@gondor.apana.org.au \
--cc=johnpol@2ka.mipt.ru \
--cc=netdev@vger.kernel.org \
--cc=satoshi.oshima.fk@hitachi.com \
--cc=shemminger@linux-foundation.org \
--cc=yoshfuji@linux-ipv6.org \
--cc=yumiko.sugita.yf@hitachi.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).