[PATCH bpf-next v4 0/3] bpf: misc performance improvements for cgroup hooks

[PATCH bpf-next v4 0/3] bpf: misc performance improvements for cgroup hooks

[Stanislav Fomichev]

First patch tries to remove kzalloc/kfree from getsockopt for the
common cases.

Second patch switches cgroup_bpf_enabled to be per-attach to
to add only overhead for the cgroup attach types used on the system.

No visible user-side changes.

v4:
- update performance numbers
- bypass_bpf_getsockopt (Martin KaFai Lau)

v3:
- remove extra newline, add comment about sizeof tcp_zerocopy_receive
  (Martin KaFai Lau)
- add another patch to remove lock_sock overhead from
  TCP_ZEROCOPY_RECEIVE; technically, this makes patch #1 obsolete,
  but I'd still prefer to keep it to help with other socket
  options

v2:
- perf numbers for getsockopt kmalloc reduction (Song Liu)
- (sk) in BPF_CGROUP_PRE_CONNECT_ENABLED (Song Liu)
- 128 -> 64 buffer size, BUILD_BUG_ON (Martin KaFai Lau)

Stanislav Fomichev (3):
  bpf: try to avoid kzalloc in cgroup/{s,g}etsockopt
  bpf: split cgroup_bpf_enabled per attach type
  bpf: remove extra lock_sock for TCP_ZEROCOPY_RECEIVE

 include/linux/bpf-cgroup.h                    |  61 ++++++---
 include/linux/filter.h                        |   5 +
 include/net/sock.h                            |   2 +
 include/net/tcp.h                             |   1 +
 kernel/bpf/cgroup.c                           | 122 ++++++++++++++----
 net/ipv4/af_inet.c                            |   9 +-
 net/ipv4/tcp.c                                |  14 ++
 net/ipv4/tcp_ipv4.c                           |   1 +
 net/ipv4/udp.c                                |   7 +-
 net/ipv6/af_inet6.c                           |   9 +-
 net/ipv6/tcp_ipv6.c                           |   1 +
 net/ipv6/udp.c                                |   7 +-
 .../selftests/bpf/prog_tests/sockopt_sk.c     |  22 ++++
 .../testing/selftests/bpf/progs/sockopt_sk.c  |  15 +++
 14 files changed, 217 insertions(+), 59 deletions(-)

-- 
2.29.2.729.g45daf8777d-goog

[PATCH bpf-next v4 1/3] bpf: try to avoid kzalloc in cgroup/{s,g}etsockopt

[Stanislav Fomichev]

When we attach a bpf program to cgroup/getsockopt any other getsockopt()
syscall starts incurring kzalloc/kfree cost. While, in general, it's
not an issue, sometimes it is, like in the case of TCP_ZEROCOPY_RECEIVE.
TCP_ZEROCOPY_RECEIVE (ab)uses getsockopt system call to implement
fastpath for incoming TCP, we don't want to have extra allocations in
there.

Let add a small buffer on the stack and use it for small (majority)
{s,g}etsockopt values. I've started with 128 bytes to cover
the options we care about (TCP_ZEROCOPY_RECEIVE which is 32 bytes
currently, with some planned extension to 64).

It seems natural to do the same for setsockopt, but it's a bit more
involved when the BPF program modifies the data (where we have to
kmalloc). The assumption is that for the majority of setsockopt
calls (which are doing pure BPF options or apply policy) this
will bring some benefit as well.

Collected some performance numbers using (on a 65k MTU localhost in a VM):
$ perf record -g -- ./tcp_mmap -s -z
$ ./tcp_mmap -H ::1 -z
$ ...
$ perf report --symbol-filter=__cgroup_bpf_run_filter_getsockopt

Without this patch:
     3.38%     0.07%  tcp_mmap  [kernel.kallsyms]  [k] __cgroup_bpf_run_filter_getsockopt
            |
             --3.30%--__cgroup_bpf_run_filter_getsockopt
                       |
                        --0.81%--__kmalloc

With the patch applied:
     1.87%     0.06%  tcp_mmap  [kernel.kallsyms]  [k] __cgroup_bpf_run_filter_getsockopt

So it saves about 1% of the system call. Unfortunately, we still get
2% of overhead due to another socket lock/unlock :-(

Signed-off-by: Stanislav Fomichev <sdf@google.com>
Cc: Martin KaFai Lau <kafai@fb.com>
Cc: Song Liu <songliubraving@fb.com>
---
 include/linux/filter.h |  3 +++
 kernel/bpf/cgroup.c    | 43 +++++++++++++++++++++++++++++++++++++++++-
 2 files changed, 45 insertions(+), 1 deletion(-)

diff --git a/include/linux/filter.h b/include/linux/filter.h
index 29c27656165b..54a4225f36d8 100644
--- a/include/linux/filter.h
+++ b/include/linux/filter.h
@@ -1281,6 +1281,8 @@ struct bpf_sysctl_kern {
 	u64 tmp_reg;
 };
 
+#define BPF_SOCKOPT_KERN_BUF_SIZE	64
+
 struct bpf_sockopt_kern {
 	struct sock	*sk;
 	u8		*optval;
@@ -1289,6 +1291,7 @@ struct bpf_sockopt_kern {
 	s32		optname;
 	s32		optlen;
 	s32		retval;
+	u8		buf[BPF_SOCKOPT_KERN_BUF_SIZE];
 };
 
 int copy_bpf_fprog_from_user(struct sock_fprog *dst, sockptr_t src, int len);
diff --git a/kernel/bpf/cgroup.c b/kernel/bpf/cgroup.c
index 6ec088a96302..ca6fa599a25d 100644
--- a/kernel/bpf/cgroup.c
+++ b/kernel/bpf/cgroup.c
@@ -16,6 +16,7 @@
 #include <linux/bpf-cgroup.h>
 #include <net/sock.h>
 #include <net/bpf_sk_storage.h>
+#include <uapi/linux/tcp.h> /* sizeof(struct tcp_zerocopy_receive) */
 
 #include "../cgroup/cgroup-internal.h"
 
@@ -1310,6 +1311,22 @@ static int sockopt_alloc_buf(struct bpf_sockopt_kern *ctx, int max_optlen)
 		max_optlen = PAGE_SIZE;
 	}
 
+	if (max_optlen <= sizeof(ctx->buf)) {
+		/* When the optval fits into BPF_SOCKOPT_KERN_BUF_SIZE
+		 * bytes avoid the cost of kzalloc.
+		 *
+		 * In order to remove extra allocations from the TCP
+		 * fast zero-copy path ensure that buffer covers
+		 * the size of struct tcp_zerocopy_receive.
+		 */
+		BUILD_BUG_ON(sizeof(struct tcp_zerocopy_receive) >
+			     BPF_SOCKOPT_KERN_BUF_SIZE);
+
+		ctx->optval = ctx->buf;
+		ctx->optval_end = ctx->optval + max_optlen;
+		return max_optlen;
+	}
+
 	ctx->optval = kzalloc(max_optlen, GFP_USER);
 	if (!ctx->optval)
 		return -ENOMEM;
@@ -1321,9 +1338,16 @@ static int sockopt_alloc_buf(struct bpf_sockopt_kern *ctx, int max_optlen)
 
 static void sockopt_free_buf(struct bpf_sockopt_kern *ctx)
 {
+	if (ctx->optval == ctx->buf)
+		return;
 	kfree(ctx->optval);
 }
 
+static bool sockopt_buf_allocated(struct bpf_sockopt_kern *ctx)
+{
+	return ctx->optval != ctx->buf;
+}
+
 int __cgroup_bpf_run_filter_setsockopt(struct sock *sk, int *level,
 				       int *optname, char __user *optval,
 				       int *optlen, char **kernel_optval)
@@ -1390,7 +1414,24 @@ int __cgroup_bpf_run_filter_setsockopt(struct sock *sk, int *level,
 		 */
 		if (ctx.optlen != 0) {
 			*optlen = ctx.optlen;
-			*kernel_optval = ctx.optval;
+			/* We've used bpf_sockopt_kern->buf as an intermediary
+			 * storage, but the BPF program indicates that we need
+			 * to pass this data to the kernel setsockopt handler.
+			 * No way to export on-stack buf, have to allocate a
+			 * new buffer.
+			 */
+			if (!sockopt_buf_allocated(&ctx)) {
+				void *p = kzalloc(ctx.optlen, GFP_USER);
+
+				if (!p) {
+					ret = -ENOMEM;
+					goto out;
+				}
+				memcpy(p, ctx.optval, ctx.optlen);
+				*kernel_optval = p;
+			} else {
+				*kernel_optval = ctx.optval;
+			}
 		}
 	}
 
-- 
2.29.2.729.g45daf8777d-goog

[PATCH bpf-next v4 2/3] bpf: split cgroup_bpf_enabled per attach type

[Stanislav Fomichev]

When we attach any cgroup hook, the rest (even if unused/unattached) start
to contribute small overhead. In particular, the one we want to avoid is
__cgroup_bpf_run_filter_skb which does two redirections to get to
the cgroup and pushes/pulls skb.

Let's split cgroup_bpf_enabled to be per-attach to make sure
only used attach types trigger.

I've dropped some existing high-level cgroup_bpf_enabled in some
places because BPF_PROG_CGROUP_XXX_RUN macros usually have another
cgroup_bpf_enabled check.

I also had to copy-paste BPF_CGROUP_RUN_SA_PROG_LOCK for
GETPEERNAME/GETSOCKNAME because type for cgroup_bpf_enabled[type]
has to be constant and known at compile time.

Signed-off-by: Stanislav Fomichev <sdf@google.com>
Acked-by: Song Liu <songliubraving@fb.com>
---
 include/linux/bpf-cgroup.h | 36 +++++++++++++++++++-----------------
 kernel/bpf/cgroup.c        | 14 ++++++--------
 net/ipv4/af_inet.c         |  9 +++++----
 net/ipv4/udp.c             |  7 +++----
 net/ipv6/af_inet6.c        |  9 +++++----
 net/ipv6/udp.c             |  7 +++----
 6 files changed, 41 insertions(+), 41 deletions(-)

diff --git a/include/linux/bpf-cgroup.h b/include/linux/bpf-cgroup.h
index 72e69a0e1e8c..dd4b8e300746 100644
--- a/include/linux/bpf-cgroup.h
+++ b/include/linux/bpf-cgroup.h
@@ -23,8 +23,8 @@ struct ctl_table_header;
 
 #ifdef CONFIG_CGROUP_BPF
 
-extern struct static_key_false cgroup_bpf_enabled_key;
-#define cgroup_bpf_enabled static_branch_unlikely(&cgroup_bpf_enabled_key)
+extern struct static_key_false cgroup_bpf_enabled_key[MAX_BPF_ATTACH_TYPE];
+#define cgroup_bpf_enabled(type) static_branch_unlikely(&cgroup_bpf_enabled_key[type])
 
 DECLARE_PER_CPU(struct bpf_cgroup_storage*,
 		bpf_cgroup_storage[MAX_BPF_CGROUP_STORAGE_TYPE]);
@@ -185,7 +185,7 @@ int bpf_percpu_cgroup_storage_update(struct bpf_map *map, void *key,
 #define BPF_CGROUP_RUN_PROG_INET_INGRESS(sk, skb)			      \
 ({									      \
 	int __ret = 0;							      \
-	if (cgroup_bpf_enabled)						      \
+	if (cgroup_bpf_enabled(BPF_CGROUP_INET_INGRESS))		      \
 		__ret = __cgroup_bpf_run_filter_skb(sk, skb,		      \
 						    BPF_CGROUP_INET_INGRESS); \
 									      \
@@ -195,7 +195,7 @@ int bpf_percpu_cgroup_storage_update(struct bpf_map *map, void *key,
 #define BPF_CGROUP_RUN_PROG_INET_EGRESS(sk, skb)			       \
 ({									       \
 	int __ret = 0;							       \
-	if (cgroup_bpf_enabled && sk && sk == skb->sk) {		       \
+	if (cgroup_bpf_enabled(BPF_CGROUP_INET_EGRESS) && sk && sk == skb->sk) { \
 		typeof(sk) __sk = sk_to_full_sk(sk);			       \
 		if (sk_fullsock(__sk))					       \
 			__ret = __cgroup_bpf_run_filter_skb(__sk, skb,	       \
@@ -207,7 +207,7 @@ int bpf_percpu_cgroup_storage_update(struct bpf_map *map, void *key,
 #define BPF_CGROUP_RUN_SK_PROG(sk, type)				       \
 ({									       \
 	int __ret = 0;							       \
-	if (cgroup_bpf_enabled) {					       \
+	if (cgroup_bpf_enabled(type)) {					       \
 		__ret = __cgroup_bpf_run_filter_sk(sk, type);		       \
 	}								       \
 	__ret;								       \
@@ -228,7 +228,7 @@ int bpf_percpu_cgroup_storage_update(struct bpf_map *map, void *key,
 #define BPF_CGROUP_RUN_SA_PROG(sk, uaddr, type)				       \
 ({									       \
 	int __ret = 0;							       \
-	if (cgroup_bpf_enabled)						       \
+	if (cgroup_bpf_enabled(type))					       \
 		__ret = __cgroup_bpf_run_filter_sock_addr(sk, uaddr, type,     \
 							  NULL);	       \
 	__ret;								       \
@@ -237,7 +237,7 @@ int bpf_percpu_cgroup_storage_update(struct bpf_map *map, void *key,
 #define BPF_CGROUP_RUN_SA_PROG_LOCK(sk, uaddr, type, t_ctx)		       \
 ({									       \
 	int __ret = 0;							       \
-	if (cgroup_bpf_enabled)	{					       \
+	if (cgroup_bpf_enabled(type))	{				       \
 		lock_sock(sk);						       \
 		__ret = __cgroup_bpf_run_filter_sock_addr(sk, uaddr, type,     \
 							  t_ctx);	       \
@@ -252,8 +252,10 @@ int bpf_percpu_cgroup_storage_update(struct bpf_map *map, void *key,
 #define BPF_CGROUP_RUN_PROG_INET6_BIND_LOCK(sk, uaddr)			       \
 	BPF_CGROUP_RUN_SA_PROG_LOCK(sk, uaddr, BPF_CGROUP_INET6_BIND, NULL)
 
-#define BPF_CGROUP_PRE_CONNECT_ENABLED(sk) (cgroup_bpf_enabled && \
-					    sk->sk_prot->pre_connect)
+#define BPF_CGROUP_PRE_CONNECT_ENABLED(sk)				       \
+	((cgroup_bpf_enabled(BPF_CGROUP_INET4_CONNECT) ||		       \
+	  cgroup_bpf_enabled(BPF_CGROUP_INET6_CONNECT)) &&		       \
+	 (sk)->sk_prot->pre_connect)
 
 #define BPF_CGROUP_RUN_PROG_INET4_CONNECT(sk, uaddr)			       \
 	BPF_CGROUP_RUN_SA_PROG(sk, uaddr, BPF_CGROUP_INET4_CONNECT)
@@ -297,7 +299,7 @@ int bpf_percpu_cgroup_storage_update(struct bpf_map *map, void *key,
 #define BPF_CGROUP_RUN_PROG_SOCK_OPS_SK(sock_ops, sk)			\
 ({									\
 	int __ret = 0;							\
-	if (cgroup_bpf_enabled)						\
+	if (cgroup_bpf_enabled(BPF_CGROUP_SOCK_OPS))			\
 		__ret = __cgroup_bpf_run_filter_sock_ops(sk,		\
 							 sock_ops,	\
 							 BPF_CGROUP_SOCK_OPS); \
@@ -307,7 +309,7 @@ int bpf_percpu_cgroup_storage_update(struct bpf_map *map, void *key,
 #define BPF_CGROUP_RUN_PROG_SOCK_OPS(sock_ops)				       \
 ({									       \
 	int __ret = 0;							       \
-	if (cgroup_bpf_enabled && (sock_ops)->sk) {	       \
+	if (cgroup_bpf_enabled(BPF_CGROUP_SOCK_OPS) && (sock_ops)->sk) {       \
 		typeof(sk) __sk = sk_to_full_sk((sock_ops)->sk);	       \
 		if (__sk && sk_fullsock(__sk))				       \
 			__ret = __cgroup_bpf_run_filter_sock_ops(__sk,	       \
@@ -320,7 +322,7 @@ int bpf_percpu_cgroup_storage_update(struct bpf_map *map, void *key,
 #define BPF_CGROUP_RUN_PROG_DEVICE_CGROUP(type, major, minor, access)	      \
 ({									      \
 	int __ret = 0;							      \
-	if (cgroup_bpf_enabled)						      \
+	if (cgroup_bpf_enabled(BPF_CGROUP_DEVICE))			      \
 		__ret = __cgroup_bpf_check_dev_permission(type, major, minor, \
 							  access,	      \
 							  BPF_CGROUP_DEVICE); \
@@ -332,7 +334,7 @@ int bpf_percpu_cgroup_storage_update(struct bpf_map *map, void *key,
 #define BPF_CGROUP_RUN_PROG_SYSCTL(head, table, write, buf, count, pos)  \
 ({									       \
 	int __ret = 0;							       \
-	if (cgroup_bpf_enabled)						       \
+	if (cgroup_bpf_enabled(BPF_CGROUP_SYSCTL))			       \
 		__ret = __cgroup_bpf_run_filter_sysctl(head, table, write,     \
 						       buf, count, pos,        \
 						       BPF_CGROUP_SYSCTL);     \
@@ -343,7 +345,7 @@ int bpf_percpu_cgroup_storage_update(struct bpf_map *map, void *key,
 				       kernel_optval)			       \
 ({									       \
 	int __ret = 0;							       \
-	if (cgroup_bpf_enabled)						       \
+	if (cgroup_bpf_enabled(BPF_CGROUP_SETSOCKOPT))			       \
 		__ret = __cgroup_bpf_run_filter_setsockopt(sock, level,	       \
 							   optname, optval,    \
 							   optlen,	       \
@@ -354,7 +356,7 @@ int bpf_percpu_cgroup_storage_update(struct bpf_map *map, void *key,
 #define BPF_CGROUP_GETSOCKOPT_MAX_OPTLEN(optlen)			       \
 ({									       \
 	int __ret = 0;							       \
-	if (cgroup_bpf_enabled)						       \
+	if (cgroup_bpf_enabled(BPF_CGROUP_GETSOCKOPT))			       \
 		get_user(__ret, optlen);				       \
 	__ret;								       \
 })
@@ -363,7 +365,7 @@ int bpf_percpu_cgroup_storage_update(struct bpf_map *map, void *key,
 				       max_optlen, retval)		       \
 ({									       \
 	int __ret = retval;						       \
-	if (cgroup_bpf_enabled)						       \
+	if (cgroup_bpf_enabled(BPF_CGROUP_GETSOCKOPT))			       \
 		__ret = __cgroup_bpf_run_filter_getsockopt(sock, level,	       \
 							   optname, optval,    \
 							   optlen, max_optlen, \
@@ -427,7 +429,7 @@ static inline int bpf_percpu_cgroup_storage_update(struct bpf_map *map,
 	return 0;
 }
 
-#define cgroup_bpf_enabled (0)
+#define cgroup_bpf_enabled(type) (0)
 #define BPF_CGROUP_RUN_SA_PROG_LOCK(sk, uaddr, type, t_ctx) ({ 0; })
 #define BPF_CGROUP_PRE_CONNECT_ENABLED(sk) (0)
 #define BPF_CGROUP_RUN_PROG_INET_INGRESS(sk,skb) ({ 0; })
diff --git a/kernel/bpf/cgroup.c b/kernel/bpf/cgroup.c
index ca6fa599a25d..adbecdcaa370 100644
--- a/kernel/bpf/cgroup.c
+++ b/kernel/bpf/cgroup.c
@@ -20,7 +20,7 @@
 
 #include "../cgroup/cgroup-internal.h"
 
-DEFINE_STATIC_KEY_FALSE(cgroup_bpf_enabled_key);
+DEFINE_STATIC_KEY_ARRAY_FALSE(cgroup_bpf_enabled_key, MAX_BPF_ATTACH_TYPE);
 EXPORT_SYMBOL(cgroup_bpf_enabled_key);
 
 void cgroup_bpf_offline(struct cgroup *cgrp)
@@ -129,7 +129,7 @@ static void cgroup_bpf_release(struct work_struct *work)
 			if (pl->link)
 				bpf_cgroup_link_auto_detach(pl->link);
 			kfree(pl);
-			static_branch_dec(&cgroup_bpf_enabled_key);
+			static_branch_dec(&cgroup_bpf_enabled_key[type]);
 		}
 		old_array = rcu_dereference_protected(
 				cgrp->bpf.effective[type],
@@ -500,7 +500,7 @@ int __cgroup_bpf_attach(struct cgroup *cgrp,
 	if (old_prog)
 		bpf_prog_put(old_prog);
 	else
-		static_branch_inc(&cgroup_bpf_enabled_key);
+		static_branch_inc(&cgroup_bpf_enabled_key[type]);
 	bpf_cgroup_storages_link(new_storage, cgrp, type);
 	return 0;
 
@@ -699,7 +699,7 @@ int __cgroup_bpf_detach(struct cgroup *cgrp, struct bpf_prog *prog,
 		cgrp->bpf.flags[type] = 0;
 	if (old_prog)
 		bpf_prog_put(old_prog);
-	static_branch_dec(&cgroup_bpf_enabled_key);
+	static_branch_dec(&cgroup_bpf_enabled_key[type]);
 	return 0;
 
 cleanup:
@@ -1364,8 +1364,7 @@ int __cgroup_bpf_run_filter_setsockopt(struct sock *sk, int *level,
 	 * attached to the hook so we don't waste time allocating
 	 * memory and locking the socket.
 	 */
-	if (!cgroup_bpf_enabled ||
-	    __cgroup_bpf_prog_array_is_empty(cgrp, BPF_CGROUP_SETSOCKOPT))
+	if (__cgroup_bpf_prog_array_is_empty(cgrp, BPF_CGROUP_SETSOCKOPT))
 		return 0;
 
 	/* Allocate a bit more than the initial user buffer for
@@ -1459,8 +1458,7 @@ int __cgroup_bpf_run_filter_getsockopt(struct sock *sk, int level,
 	 * attached to the hook so we don't waste time allocating
 	 * memory and locking the socket.
 	 */
-	if (!cgroup_bpf_enabled ||
-	    __cgroup_bpf_prog_array_is_empty(cgrp, BPF_CGROUP_GETSOCKOPT))
+	if (__cgroup_bpf_prog_array_is_empty(cgrp, BPF_CGROUP_GETSOCKOPT))
 		return retval;
 
 	ctx.optlen = max_optlen;
diff --git a/net/ipv4/af_inet.c b/net/ipv4/af_inet.c
index b94fa8eb831b..6ba2930ff49b 100644
--- a/net/ipv4/af_inet.c
+++ b/net/ipv4/af_inet.c
@@ -777,18 +777,19 @@ int inet_getname(struct socket *sock, struct sockaddr *uaddr,
 			return -ENOTCONN;
 		sin->sin_port = inet->inet_dport;
 		sin->sin_addr.s_addr = inet->inet_daddr;
+		BPF_CGROUP_RUN_SA_PROG_LOCK(sk, (struct sockaddr *)sin,
+					    BPF_CGROUP_INET4_GETPEERNAME,
+					    NULL);
 	} else {
 		__be32 addr = inet->inet_rcv_saddr;
 		if (!addr)
 			addr = inet->inet_saddr;
 		sin->sin_port = inet->inet_sport;
 		sin->sin_addr.s_addr = addr;
-	}
-	if (cgroup_bpf_enabled)
 		BPF_CGROUP_RUN_SA_PROG_LOCK(sk, (struct sockaddr *)sin,
-					    peer ? BPF_CGROUP_INET4_GETPEERNAME :
-						   BPF_CGROUP_INET4_GETSOCKNAME,
+					    BPF_CGROUP_INET4_GETSOCKNAME,
 					    NULL);
+	}
 	memset(sin->sin_zero, 0, sizeof(sin->sin_zero));
 	return sizeof(*sin);
 }
diff --git a/net/ipv4/udp.c b/net/ipv4/udp.c
index dece195f212c..fc3c2e75e400 100644
--- a/net/ipv4/udp.c
+++ b/net/ipv4/udp.c
@@ -1124,7 +1124,7 @@ int udp_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
 		rcu_read_unlock();
 	}
 
-	if (cgroup_bpf_enabled && !connected) {
+	if (cgroup_bpf_enabled(BPF_CGROUP_UDP4_SENDMSG) && !connected) {
 		err = BPF_CGROUP_RUN_PROG_UDP4_SENDMSG_LOCK(sk,
 					    (struct sockaddr *)usin, &ipc.addr);
 		if (err)
@@ -1858,9 +1858,8 @@ int udp_recvmsg(struct sock *sk, struct msghdr *msg, size_t len, int noblock,
 		memset(sin->sin_zero, 0, sizeof(sin->sin_zero));
 		*addr_len = sizeof(*sin);
 
-		if (cgroup_bpf_enabled)
-			BPF_CGROUP_RUN_PROG_UDP4_RECVMSG_LOCK(sk,
-							(struct sockaddr *)sin);
+		BPF_CGROUP_RUN_PROG_UDP4_RECVMSG_LOCK(sk,
+						      (struct sockaddr *)sin);
 	}
 
 	if (udp_sk(sk)->gro_enabled)
diff --git a/net/ipv6/af_inet6.c b/net/ipv6/af_inet6.c
index a7e3d170af51..fc985658dc91 100644
--- a/net/ipv6/af_inet6.c
+++ b/net/ipv6/af_inet6.c
@@ -527,18 +527,19 @@ int inet6_getname(struct socket *sock, struct sockaddr *uaddr,
 		sin->sin6_addr = sk->sk_v6_daddr;
 		if (np->sndflow)
 			sin->sin6_flowinfo = np->flow_label;
+		BPF_CGROUP_RUN_SA_PROG_LOCK(sk, (struct sockaddr *)sin,
+					    BPF_CGROUP_INET6_GETPEERNAME,
+					    NULL);
 	} else {
 		if (ipv6_addr_any(&sk->sk_v6_rcv_saddr))
 			sin->sin6_addr = np->saddr;
 		else
 			sin->sin6_addr = sk->sk_v6_rcv_saddr;
 		sin->sin6_port = inet->inet_sport;
-	}
-	if (cgroup_bpf_enabled)
 		BPF_CGROUP_RUN_SA_PROG_LOCK(sk, (struct sockaddr *)sin,
-					    peer ? BPF_CGROUP_INET6_GETPEERNAME :
-						   BPF_CGROUP_INET6_GETSOCKNAME,
+					    BPF_CGROUP_INET6_GETSOCKNAME,
 					    NULL);
+	}
 	sin->sin6_scope_id = ipv6_iface_scope_id(&sin->sin6_addr,
 						 sk->sk_bound_dev_if);
 	return sizeof(*sin);
diff --git a/net/ipv6/udp.c b/net/ipv6/udp.c
index 9008f5796ad4..50611bd63647 100644
--- a/net/ipv6/udp.c
+++ b/net/ipv6/udp.c
@@ -409,9 +409,8 @@ int udpv6_recvmsg(struct sock *sk, struct msghdr *msg, size_t len,
 		}
 		*addr_len = sizeof(*sin6);
 
-		if (cgroup_bpf_enabled)
-			BPF_CGROUP_RUN_PROG_UDP6_RECVMSG_LOCK(sk,
-						(struct sockaddr *)sin6);
+		BPF_CGROUP_RUN_PROG_UDP6_RECVMSG_LOCK(sk,
+						      (struct sockaddr *)sin6);
 	}
 
 	if (udp_sk(sk)->gro_enabled)
@@ -1462,7 +1461,7 @@ int udpv6_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
 		fl6.saddr = np->saddr;
 	fl6.fl6_sport = inet->inet_sport;
 
-	if (cgroup_bpf_enabled && !connected) {
+	if (cgroup_bpf_enabled(BPF_CGROUP_UDP6_SENDMSG) && !connected) {
 		err = BPF_CGROUP_RUN_PROG_UDP6_SENDMSG_LOCK(sk,
 					   (struct sockaddr *)sin6, &fl6.saddr);
 		if (err)
-- 
2.29.2.729.g45daf8777d-goog

[PATCH bpf-next v4 3/3] bpf: remove extra lock_sock for TCP_ZEROCOPY_RECEIVE

[Stanislav Fomichev]

Add custom implementation of getsockopt hook for TCP_ZEROCOPY_RECEIVE.
We skip generic hooks for TCP_ZEROCOPY_RECEIVE and have a custom
call in do_tcp_getsockopt using the on-stack data. This removes
2% overhead for locking/unlocking the socket.

Also:
- Removed BUILD_BUG_ON (zerocopy doesn't depend on the buf size anymore)
- Separated on-stack buffer into bpf_sockopt_buf and downsized to 32 bytes
  (let's keep it to help with the other options)

(I can probably split this patch into two: add new features and rework
 bpf_sockopt_buf; can follow up if the approach in general sounds
 good).

Without this patch:
     1.87%     0.06%  tcp_mmap  [kernel.kallsyms]  [k] __cgroup_bpf_run_filter_getsockopt

With the patch applied:
     0.52%     0.12%  tcp_mmap  [kernel.kallsyms]  [k] __cgroup_bpf_run_filter_getsockopt_kern

Signed-off-by: Stanislav Fomichev <sdf@google.com>
Cc: Martin KaFai Lau <kafai@fb.com>
Cc: Song Liu <songliubraving@fb.com>
Cc: Eric Dumazet <edumazet@google.com>
---
 include/linux/bpf-cgroup.h                    | 25 ++++-
 include/linux/filter.h                        |  6 +-
 include/net/sock.h                            |  2 +
 include/net/tcp.h                             |  1 +
 kernel/bpf/cgroup.c                           | 93 +++++++++++++------
 net/ipv4/tcp.c                                | 14 +++
 net/ipv4/tcp_ipv4.c                           |  1 +
 net/ipv6/tcp_ipv6.c                           |  1 +
 .../selftests/bpf/prog_tests/sockopt_sk.c     | 22 +++++
 .../testing/selftests/bpf/progs/sockopt_sk.c  | 15 +++
 10 files changed, 147 insertions(+), 33 deletions(-)

diff --git a/include/linux/bpf-cgroup.h b/include/linux/bpf-cgroup.h
index dd4b8e300746..cbba9c9ab073 100644
--- a/include/linux/bpf-cgroup.h
+++ b/include/linux/bpf-cgroup.h
@@ -147,6 +147,10 @@ int __cgroup_bpf_run_filter_getsockopt(struct sock *sk, int level,
 				       int __user *optlen, int max_optlen,
 				       int retval);
 
+int __cgroup_bpf_run_filter_getsockopt_kern(struct sock *sk, int level,
+					    int optname, void *optval,
+					    int *optlen, int retval);
+
 static inline enum bpf_cgroup_storage_type cgroup_storage_type(
 	struct bpf_map *map)
 {
@@ -366,10 +370,21 @@ int bpf_percpu_cgroup_storage_update(struct bpf_map *map, void *key,
 ({									       \
 	int __ret = retval;						       \
 	if (cgroup_bpf_enabled(BPF_CGROUP_GETSOCKOPT))			       \
-		__ret = __cgroup_bpf_run_filter_getsockopt(sock, level,	       \
-							   optname, optval,    \
-							   optlen, max_optlen, \
-							   retval);	       \
+		if (!(sock)->sk_prot->bpf_bypass_getsockopt ||		       \
+		    !(sock)->sk_prot->bpf_bypass_getsockopt(level, optname))   \
+			__ret = __cgroup_bpf_run_filter_getsockopt(	       \
+				sock, level, optname, optval, optlen,	       \
+				max_optlen, retval);			       \
+	__ret;								       \
+})
+
+#define BPF_CGROUP_RUN_PROG_GETSOCKOPT_KERN(sock, level, optname, optval,      \
+					    optlen, retval)		       \
+({									       \
+	int __ret = retval;						       \
+	if (cgroup_bpf_enabled(BPF_CGROUP_GETSOCKOPT))			       \
+		__ret = __cgroup_bpf_run_filter_getsockopt_kern(	       \
+			sock, level, optname, optval, optlen, retval);	       \
 	__ret;								       \
 })
 
@@ -454,6 +469,8 @@ static inline int bpf_percpu_cgroup_storage_update(struct bpf_map *map,
 #define BPF_CGROUP_GETSOCKOPT_MAX_OPTLEN(optlen) ({ 0; })
 #define BPF_CGROUP_RUN_PROG_GETSOCKOPT(sock, level, optname, optval, \
 				       optlen, max_optlen, retval) ({ retval; })
+#define BPF_CGROUP_RUN_PROG_GETSOCKOPT_KERN(sock, level, optname, optval, \
+					    optlen, retval) ({ retval; })
 #define BPF_CGROUP_RUN_PROG_SETSOCKOPT(sock, level, optname, optval, optlen, \
 				       kernel_optval) ({ 0; })
 
diff --git a/include/linux/filter.h b/include/linux/filter.h
index 54a4225f36d8..8739f1d4cac4 100644
--- a/include/linux/filter.h
+++ b/include/linux/filter.h
@@ -1281,7 +1281,10 @@ struct bpf_sysctl_kern {
 	u64 tmp_reg;
 };
 
-#define BPF_SOCKOPT_KERN_BUF_SIZE	64
+#define BPF_SOCKOPT_KERN_BUF_SIZE	32
+struct bpf_sockopt_buf {
+	u8		data[BPF_SOCKOPT_KERN_BUF_SIZE];
+};
 
 struct bpf_sockopt_kern {
 	struct sock	*sk;
@@ -1291,7 +1294,6 @@ struct bpf_sockopt_kern {
 	s32		optname;
 	s32		optlen;
 	s32		retval;
-	u8		buf[BPF_SOCKOPT_KERN_BUF_SIZE];
 };
 
 int copy_bpf_fprog_from_user(struct sock_fprog *dst, sockptr_t src, int len);
diff --git a/include/net/sock.h b/include/net/sock.h
index bdc4323ce53c..ebf44d724845 100644
--- a/include/net/sock.h
+++ b/include/net/sock.h
@@ -1174,6 +1174,8 @@ struct proto {
 
 	int			(*backlog_rcv) (struct sock *sk,
 						struct sk_buff *skb);
+	bool			(*bpf_bypass_getsockopt)(int level,
+							 int optname);
 
 	void		(*release_cb)(struct sock *sk);
 
diff --git a/include/net/tcp.h b/include/net/tcp.h
index 78d13c88720f..4bb42fb19711 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -403,6 +403,7 @@ __poll_t tcp_poll(struct file *file, struct socket *sock,
 		      struct poll_table_struct *wait);
 int tcp_getsockopt(struct sock *sk, int level, int optname,
 		   char __user *optval, int __user *optlen);
+bool tcp_bpf_bypass_getsockopt(int level, int optname);
 int tcp_setsockopt(struct sock *sk, int level, int optname, sockptr_t optval,
 		   unsigned int optlen);
 void tcp_set_keepalive(struct sock *sk, int val);
diff --git a/kernel/bpf/cgroup.c b/kernel/bpf/cgroup.c
index adbecdcaa370..e82df63aedc7 100644
--- a/kernel/bpf/cgroup.c
+++ b/kernel/bpf/cgroup.c
@@ -16,7 +16,6 @@
 #include <linux/bpf-cgroup.h>
 #include <net/sock.h>
 #include <net/bpf_sk_storage.h>
-#include <uapi/linux/tcp.h> /* sizeof(struct tcp_zerocopy_receive) */
 
 #include "../cgroup/cgroup-internal.h"
 
@@ -1299,7 +1298,8 @@ static bool __cgroup_bpf_prog_array_is_empty(struct cgroup *cgrp,
 	return empty;
 }
 
-static int sockopt_alloc_buf(struct bpf_sockopt_kern *ctx, int max_optlen)
+static int sockopt_alloc_buf(struct bpf_sockopt_kern *ctx, int max_optlen,
+			     struct bpf_sockopt_buf *buf)
 {
 	if (unlikely(max_optlen < 0))
 		return -EINVAL;
@@ -1311,18 +1311,11 @@ static int sockopt_alloc_buf(struct bpf_sockopt_kern *ctx, int max_optlen)
 		max_optlen = PAGE_SIZE;
 	}
 
-	if (max_optlen <= sizeof(ctx->buf)) {
+	if (max_optlen <= sizeof(buf->data)) {
 		/* When the optval fits into BPF_SOCKOPT_KERN_BUF_SIZE
 		 * bytes avoid the cost of kzalloc.
-		 *
-		 * In order to remove extra allocations from the TCP
-		 * fast zero-copy path ensure that buffer covers
-		 * the size of struct tcp_zerocopy_receive.
 		 */
-		BUILD_BUG_ON(sizeof(struct tcp_zerocopy_receive) >
-			     BPF_SOCKOPT_KERN_BUF_SIZE);
-
-		ctx->optval = ctx->buf;
+		ctx->optval = buf->data;
 		ctx->optval_end = ctx->optval + max_optlen;
 		return max_optlen;
 	}
@@ -1336,16 +1329,18 @@ static int sockopt_alloc_buf(struct bpf_sockopt_kern *ctx, int max_optlen)
 	return max_optlen;
 }
 
-static void sockopt_free_buf(struct bpf_sockopt_kern *ctx)
+static void sockopt_free_buf(struct bpf_sockopt_kern *ctx,
+			     struct bpf_sockopt_buf *buf)
 {
-	if (ctx->optval == ctx->buf)
+	if (ctx->optval == buf->data)
 		return;
 	kfree(ctx->optval);
 }
 
-static bool sockopt_buf_allocated(struct bpf_sockopt_kern *ctx)
+static bool sockopt_buf_allocated(struct bpf_sockopt_kern *ctx,
+				  struct bpf_sockopt_buf *buf)
 {
-	return ctx->optval != ctx->buf;
+	return ctx->optval != buf->data;
 }
 
 int __cgroup_bpf_run_filter_setsockopt(struct sock *sk, int *level,
@@ -1353,6 +1348,7 @@ int __cgroup_bpf_run_filter_setsockopt(struct sock *sk, int *level,
 				       int *optlen, char **kernel_optval)
 {
 	struct cgroup *cgrp = sock_cgroup_ptr(&sk->sk_cgrp_data);
+	struct bpf_sockopt_buf buf = {};
 	struct bpf_sockopt_kern ctx = {
 		.sk = sk,
 		.level = *level,
@@ -1373,7 +1369,7 @@ int __cgroup_bpf_run_filter_setsockopt(struct sock *sk, int *level,
 	 */
 	max_optlen = max_t(int, 16, *optlen);
 
-	max_optlen = sockopt_alloc_buf(&ctx, max_optlen);
+	max_optlen = sockopt_alloc_buf(&ctx, max_optlen, &buf);
 	if (max_optlen < 0)
 		return max_optlen;
 
@@ -1419,7 +1415,7 @@ int __cgroup_bpf_run_filter_setsockopt(struct sock *sk, int *level,
 			 * No way to export on-stack buf, have to allocate a
 			 * new buffer.
 			 */
-			if (!sockopt_buf_allocated(&ctx)) {
+			if (!sockopt_buf_allocated(&ctx, &buf)) {
 				void *p = kzalloc(ctx.optlen, GFP_USER);
 
 				if (!p) {
@@ -1436,7 +1432,7 @@ int __cgroup_bpf_run_filter_setsockopt(struct sock *sk, int *level,
 
 out:
 	if (ret)
-		sockopt_free_buf(&ctx);
+		sockopt_free_buf(&ctx, &buf);
 	return ret;
 }
 
@@ -1445,15 +1441,20 @@ int __cgroup_bpf_run_filter_getsockopt(struct sock *sk, int level,
 				       int __user *optlen, int max_optlen,
 				       int retval)
 {
-	struct cgroup *cgrp = sock_cgroup_ptr(&sk->sk_cgrp_data);
-	struct bpf_sockopt_kern ctx = {
-		.sk = sk,
-		.level = level,
-		.optname = optname,
-		.retval = retval,
-	};
+	struct bpf_sockopt_kern ctx;
+	struct bpf_sockopt_buf buf;
+	struct cgroup *cgrp;
 	int ret;
 
+	memset(&buf, 0, sizeof(buf));
+	memset(&ctx, 0, sizeof(ctx));
+
+	cgrp = sock_cgroup_ptr(&sk->sk_cgrp_data);
+	ctx.sk = sk;
+	ctx.level = level;
+	ctx.optname = optname;
+	ctx.retval = retval;
+
 	/* Opportunistic check to see whether we have any BPF program
 	 * attached to the hook so we don't waste time allocating
 	 * memory and locking the socket.
@@ -1463,7 +1464,7 @@ int __cgroup_bpf_run_filter_getsockopt(struct sock *sk, int level,
 
 	ctx.optlen = max_optlen;
 
-	max_optlen = sockopt_alloc_buf(&ctx, max_optlen);
+	max_optlen = sockopt_alloc_buf(&ctx, max_optlen, &buf);
 	if (max_optlen < 0)
 		return max_optlen;
 
@@ -1521,9 +1522,47 @@ int __cgroup_bpf_run_filter_getsockopt(struct sock *sk, int level,
 	ret = ctx.retval;
 
 out:
-	sockopt_free_buf(&ctx);
+	sockopt_free_buf(&ctx, &buf);
 	return ret;
 }
+
+int __cgroup_bpf_run_filter_getsockopt_kern(struct sock *sk, int level,
+					    int optname, void *optval,
+					    int *optlen, int retval)
+{
+	struct cgroup *cgrp = sock_cgroup_ptr(&sk->sk_cgrp_data);
+	struct bpf_sockopt_kern ctx = {
+		.sk = sk,
+		.level = level,
+		.optname = optname,
+		.retval = retval,
+		.optlen = *optlen,
+		.optval = optval,
+		.optval_end = optval + *optlen,
+	};
+	int ret;
+
+	ret = BPF_PROG_RUN_ARRAY(cgrp->bpf.effective[BPF_CGROUP_GETSOCKOPT],
+				 &ctx, BPF_PROG_RUN);
+	if (!ret)
+		return -EPERM;
+
+	if (ctx.optlen > *optlen)
+		return -EFAULT;
+
+	/* BPF programs only allowed to set retval to 0, not some
+	 * arbitrary value.
+	 */
+	if (ctx.retval != 0 && ctx.retval != retval)
+		return -EFAULT;
+
+	/* BPF programs can shrink the buffer, export the modifications.
+	 */
+	if (ctx.optlen != 0)
+		*optlen = ctx.optlen;
+
+	return ctx.retval;
+}
 #endif
 
 static ssize_t sysctl_cpy_dir(const struct ctl_dir *dir, char **bufp,
diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
index ed42d2193c5c..ef3c895b66c1 100644
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c
@@ -4098,6 +4098,8 @@ static int do_tcp_getsockopt(struct sock *sk, int level,
 			return -EFAULT;
 		lock_sock(sk);
 		err = tcp_zerocopy_receive(sk, &zc);
+		err = BPF_CGROUP_RUN_PROG_GETSOCKOPT_KERN(sk, level, optname,
+							  &zc, &len, err);
 		release_sock(sk);
 		if (len >= offsetofend(struct tcp_zerocopy_receive, err))
 			goto zerocopy_rcv_sk_err;
@@ -4132,6 +4134,18 @@ static int do_tcp_getsockopt(struct sock *sk, int level,
 	return 0;
 }
 
+bool tcp_bpf_bypass_getsockopt(int level, int optname)
+{
+	/* TCP do_tcp_getsockopt has optimized getsockopt implementation
+	 * to avoid extra socket lock for TCP_ZEROCOPY_RECEIVE.
+	 */
+	if (level == SOL_TCP && optname == TCP_ZEROCOPY_RECEIVE)
+		return true;
+
+	return false;
+}
+EXPORT_SYMBOL(tcp_bpf_bypass_getsockopt);
+
 int tcp_getsockopt(struct sock *sk, int level, int optname, char __user *optval,
 		   int __user *optlen)
 {
diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
index 58207c7769d0..8b4906980fce 100644
--- a/net/ipv4/tcp_ipv4.c
+++ b/net/ipv4/tcp_ipv4.c
@@ -2792,6 +2792,7 @@ struct proto tcp_prot = {
 	.shutdown		= tcp_shutdown,
 	.setsockopt		= tcp_setsockopt,
 	.getsockopt		= tcp_getsockopt,
+	.bpf_bypass_getsockopt	= tcp_bpf_bypass_getsockopt,
 	.keepalive		= tcp_set_keepalive,
 	.recvmsg		= tcp_recvmsg,
 	.sendmsg		= tcp_sendmsg,
diff --git a/net/ipv6/tcp_ipv6.c b/net/ipv6/tcp_ipv6.c
index e254569a3005..6624eccff85b 100644
--- a/net/ipv6/tcp_ipv6.c
+++ b/net/ipv6/tcp_ipv6.c
@@ -2121,6 +2121,7 @@ struct proto tcpv6_prot = {
 	.shutdown		= tcp_shutdown,
 	.setsockopt		= tcp_setsockopt,
 	.getsockopt		= tcp_getsockopt,
+	.bpf_bypass_getsockopt	= tcp_bpf_bypass_getsockopt,
 	.keepalive		= tcp_set_keepalive,
 	.recvmsg		= tcp_recvmsg,
 	.sendmsg		= tcp_sendmsg,
diff --git a/tools/testing/selftests/bpf/prog_tests/sockopt_sk.c b/tools/testing/selftests/bpf/prog_tests/sockopt_sk.c
index b25c9c45c148..6bb18b1d8578 100644
--- a/tools/testing/selftests/bpf/prog_tests/sockopt_sk.c
+++ b/tools/testing/selftests/bpf/prog_tests/sockopt_sk.c
@@ -11,6 +11,7 @@ static int getsetsockopt(void)
 		char u8[4];
 		__u32 u32;
 		char cc[16]; /* TCP_CA_NAME_MAX */
+		struct tcp_zerocopy_receive zc;
 	} buf = {};
 	socklen_t optlen;
 	char *big_buf = NULL;
@@ -154,6 +155,27 @@ static int getsetsockopt(void)
 		goto err;
 	}
 
+	/* TCP_ZEROCOPY_RECEIVE triggers */
+	memset(&buf, 0, sizeof(buf));
+	optlen = sizeof(buf.zc);
+	err = getsockopt(fd, SOL_TCP, TCP_ZEROCOPY_RECEIVE, &buf, &optlen);
+	if (err) {
+		log_err("Unexpected getsockopt(TCP_ZEROCOPY_RECEIVE) err=%d errno=%d",
+			err, errno);
+		goto err;
+	}
+
+	memset(&buf, 0, sizeof(buf));
+	buf.zc.address = 12345; /* rejected by BPF */
+	optlen = sizeof(buf.zc);
+	errno = 0;
+	err = getsockopt(fd, SOL_TCP, TCP_ZEROCOPY_RECEIVE, &buf, &optlen);
+	if (errno != EPERM) {
+		log_err("Unexpected getsockopt(TCP_ZEROCOPY_RECEIVE) err=%d errno=%d",
+			err, errno);
+		goto err;
+	}
+
 	free(big_buf);
 	close(fd);
 	return 0;
diff --git a/tools/testing/selftests/bpf/progs/sockopt_sk.c b/tools/testing/selftests/bpf/progs/sockopt_sk.c
index 712df7b49cb1..c726f0763a13 100644
--- a/tools/testing/selftests/bpf/progs/sockopt_sk.c
+++ b/tools/testing/selftests/bpf/progs/sockopt_sk.c
@@ -57,6 +57,21 @@ int _getsockopt(struct bpf_sockopt *ctx)
 		return 1;
 	}
 
+	if (ctx->level == SOL_TCP && ctx->optname == TCP_ZEROCOPY_RECEIVE) {
+		/* Verify that TCP_ZEROCOPY_RECEIVE triggers.
+		 * It has a custom implementation for performance
+		 * reasons.
+		 */
+
+		if (optval + sizeof(struct tcp_zerocopy_receive) > optval_end)
+			return 0; /* EPERM, bounds check */
+
+		if (((struct tcp_zerocopy_receive *)optval)->address != 0)
+			return 0; /* EPERM, unexpected data */
+
+		return 1;
+	}
+
 	if (ctx->level == SOL_IP && ctx->optname == IP_FREEBIND) {
 		if (optval + 1 > optval_end)
 			return 0; /* EPERM, bounds check */
-- 
2.29.2.729.g45daf8777d-goog

Re: [PATCH bpf-next v4 3/3] bpf: remove extra lock_sock for TCP_ZEROCOPY_RECEIVE

[Martin KaFai Lau]

On Thu, Jan 07, 2021 at 10:43:05AM -0800, Stanislav Fomichev wrote:
> Add custom implementation of getsockopt hook for TCP_ZEROCOPY_RECEIVE.
> We skip generic hooks for TCP_ZEROCOPY_RECEIVE and have a custom
> call in do_tcp_getsockopt using the on-stack data. This removes
> 2% overhead for locking/unlocking the socket.
> 
> Also:
> - Removed BUILD_BUG_ON (zerocopy doesn't depend on the buf size anymore)
> - Separated on-stack buffer into bpf_sockopt_buf and downsized to 32 bytes
>   (let's keep it to help with the other options)
> 
> (I can probably split this patch into two: add new features and rework
>  bpf_sockopt_buf; can follow up if the approach in general sounds
>  good).
> 
> Without this patch:
>      1.87%     0.06%  tcp_mmap  [kernel.kallsyms]  [k] __cgroup_bpf_run_filter_getsockopt
> 
> With the patch applied:
>      0.52%     0.12%  tcp_mmap  [kernel.kallsyms]  [k] __cgroup_bpf_run_filter_getsockopt_kern
> 
> Signed-off-by: Stanislav Fomichev <sdf@google.com>
> Cc: Martin KaFai Lau <kafai@fb.com>
> Cc: Song Liu <songliubraving@fb.com>
> Cc: Eric Dumazet <edumazet@google.com>
> ---
>  include/linux/bpf-cgroup.h                    | 25 ++++-
>  include/linux/filter.h                        |  6 +-
>  include/net/sock.h                            |  2 +
>  include/net/tcp.h                             |  1 +
>  kernel/bpf/cgroup.c                           | 93 +++++++++++++------
>  net/ipv4/tcp.c                                | 14 +++
>  net/ipv4/tcp_ipv4.c                           |  1 +
>  net/ipv6/tcp_ipv6.c                           |  1 +
>  .../selftests/bpf/prog_tests/sockopt_sk.c     | 22 +++++
>  .../testing/selftests/bpf/progs/sockopt_sk.c  | 15 +++
>  10 files changed, 147 insertions(+), 33 deletions(-)
>

[ ... ]

> @@ -454,6 +469,8 @@ static inline int bpf_percpu_cgroup_storage_update(struct bpf_map *map,
>  #define BPF_CGROUP_GETSOCKOPT_MAX_OPTLEN(optlen) ({ 0; })
>  #define BPF_CGROUP_RUN_PROG_GETSOCKOPT(sock, level, optname, optval, \
>  				       optlen, max_optlen, retval) ({ retval; })
> +#define BPF_CGROUP_RUN_PROG_GETSOCKOPT_KERN(sock, level, optname, optval, \
> +					    optlen, retval) ({ retval; })
>  #define BPF_CGROUP_RUN_PROG_SETSOCKOPT(sock, level, optname, optval, optlen, \
>  				       kernel_optval) ({ 0; })
>  
> diff --git a/include/linux/filter.h b/include/linux/filter.h
> index 54a4225f36d8..8739f1d4cac4 100644
> --- a/include/linux/filter.h
> +++ b/include/linux/filter.h
> @@ -1281,7 +1281,10 @@ struct bpf_sysctl_kern {
>  	u64 tmp_reg;
>  };
>  
> -#define BPF_SOCKOPT_KERN_BUF_SIZE	64
> +#define BPF_SOCKOPT_KERN_BUF_SIZE	32
It is reduced from patch 1 because there is no
need to use the buf (and copy from/to buf) in TCP_ZEROCOPY_RECEIVE?

Patch 1 is still desired (and kept in this set) because it may still
benefit other optname?

> +struct bpf_sockopt_buf {
> +	u8		data[BPF_SOCKOPT_KERN_BUF_SIZE];
> +};
>  
>  struct bpf_sockopt_kern {
>  	struct sock	*sk;
> @@ -1291,7 +1294,6 @@ struct bpf_sockopt_kern {
>  	s32		optname;
>  	s32		optlen;
>  	s32		retval;
> -	u8		buf[BPF_SOCKOPT_KERN_BUF_SIZE];
It is better to pick one way to do thing to avoid code
churn like this within the same series.

>  };
>  
>  int copy_bpf_fprog_from_user(struct sock_fprog *dst, sockptr_t src, int len);
> diff --git a/include/net/sock.h b/include/net/sock.h
> index bdc4323ce53c..ebf44d724845 100644
> --- a/include/net/sock.h
> +++ b/include/net/sock.h
> @@ -1174,6 +1174,8 @@ struct proto {
>  
>  	int			(*backlog_rcv) (struct sock *sk,
>  						struct sk_buff *skb);
> +	bool			(*bpf_bypass_getsockopt)(int level,
> +							 int optname);
>  
>  	void		(*release_cb)(struct sock *sk);
>  
> diff --git a/include/net/tcp.h b/include/net/tcp.h
> index 78d13c88720f..4bb42fb19711 100644
> --- a/include/net/tcp.h
> +++ b/include/net/tcp.h
> @@ -403,6 +403,7 @@ __poll_t tcp_poll(struct file *file, struct socket *sock,
>  		      struct poll_table_struct *wait);
>  int tcp_getsockopt(struct sock *sk, int level, int optname,
>  		   char __user *optval, int __user *optlen);
> +bool tcp_bpf_bypass_getsockopt(int level, int optname);
>  int tcp_setsockopt(struct sock *sk, int level, int optname, sockptr_t optval,
>  		   unsigned int optlen);
>  void tcp_set_keepalive(struct sock *sk, int val);
> diff --git a/kernel/bpf/cgroup.c b/kernel/bpf/cgroup.c
> index adbecdcaa370..e82df63aedc7 100644
> --- a/kernel/bpf/cgroup.c
> +++ b/kernel/bpf/cgroup.c
> @@ -16,7 +16,6 @@
>  #include <linux/bpf-cgroup.h>
>  #include <net/sock.h>
>  #include <net/bpf_sk_storage.h>
> -#include <uapi/linux/tcp.h> /* sizeof(struct tcp_zerocopy_receive) */
Can the patches be re-ordered a little to avoid code churn like this
in the same series?

It feels like this patch 3 should be the first patch instead.
The current patch 1 should be the second patch
but it can still use the tcp_mmap to show potential
benefit for other optnames.

>  
>  #include "../cgroup/cgroup-internal.h"
>  
> @@ -1299,7 +1298,8 @@ static bool __cgroup_bpf_prog_array_is_empty(struct cgroup *cgrp,
>  	return empty;
>  }
>  
> -static int sockopt_alloc_buf(struct bpf_sockopt_kern *ctx, int max_optlen)
> +static int sockopt_alloc_buf(struct bpf_sockopt_kern *ctx, int max_optlen,
> +			     struct bpf_sockopt_buf *buf)
>  {
>  	if (unlikely(max_optlen < 0))
>  		return -EINVAL;
> @@ -1311,18 +1311,11 @@ static int sockopt_alloc_buf(struct bpf_sockopt_kern *ctx, int max_optlen)
>  		max_optlen = PAGE_SIZE;
>  	}
>  
> -	if (max_optlen <= sizeof(ctx->buf)) {
> +	if (max_optlen <= sizeof(buf->data)) {
>  		/* When the optval fits into BPF_SOCKOPT_KERN_BUF_SIZE
>  		 * bytes avoid the cost of kzalloc.
> -		 *
> -		 * In order to remove extra allocations from the TCP
> -		 * fast zero-copy path ensure that buffer covers
> -		 * the size of struct tcp_zerocopy_receive.
>  		 */
> -		BUILD_BUG_ON(sizeof(struct tcp_zerocopy_receive) >
> -			     BPF_SOCKOPT_KERN_BUF_SIZE);
> -
> -		ctx->optval = ctx->buf;
> +		ctx->optval = buf->data;
>  		ctx->optval_end = ctx->optval + max_optlen;
>  		return max_optlen;
>  	}
> @@ -1336,16 +1329,18 @@ static int sockopt_alloc_buf(struct bpf_sockopt_kern *ctx, int max_optlen)
>  	return max_optlen;
>  }
>  
> -static void sockopt_free_buf(struct bpf_sockopt_kern *ctx)
> +static void sockopt_free_buf(struct bpf_sockopt_kern *ctx,
> +			     struct bpf_sockopt_buf *buf)
>  {
> -	if (ctx->optval == ctx->buf)
> +	if (ctx->optval == buf->data)
>  		return;
>  	kfree(ctx->optval);
>  }
>  
> -static bool sockopt_buf_allocated(struct bpf_sockopt_kern *ctx)
> +static bool sockopt_buf_allocated(struct bpf_sockopt_kern *ctx,
> +				  struct bpf_sockopt_buf *buf)
>  {
> -	return ctx->optval != ctx->buf;
> +	return ctx->optval != buf->data;
>  }
>  
>  int __cgroup_bpf_run_filter_setsockopt(struct sock *sk, int *level,
> @@ -1353,6 +1348,7 @@ int __cgroup_bpf_run_filter_setsockopt(struct sock *sk, int *level,
>  				       int *optlen, char **kernel_optval)
>  {
>  	struct cgroup *cgrp = sock_cgroup_ptr(&sk->sk_cgrp_data);
> +	struct bpf_sockopt_buf buf = {};
>  	struct bpf_sockopt_kern ctx = {
>  		.sk = sk,
>  		.level = *level,
> @@ -1373,7 +1369,7 @@ int __cgroup_bpf_run_filter_setsockopt(struct sock *sk, int *level,
>  	 */
>  	max_optlen = max_t(int, 16, *optlen);
>  
> -	max_optlen = sockopt_alloc_buf(&ctx, max_optlen);
> +	max_optlen = sockopt_alloc_buf(&ctx, max_optlen, &buf);
>  	if (max_optlen < 0)
>  		return max_optlen;
>  
> @@ -1419,7 +1415,7 @@ int __cgroup_bpf_run_filter_setsockopt(struct sock *sk, int *level,
>  			 * No way to export on-stack buf, have to allocate a
>  			 * new buffer.
>  			 */
> -			if (!sockopt_buf_allocated(&ctx)) {
> +			if (!sockopt_buf_allocated(&ctx, &buf)) {
>  				void *p = kzalloc(ctx.optlen, GFP_USER);
>  
>  				if (!p) {
> @@ -1436,7 +1432,7 @@ int __cgroup_bpf_run_filter_setsockopt(struct sock *sk, int *level,
>  
>  out:
>  	if (ret)
> -		sockopt_free_buf(&ctx);
> +		sockopt_free_buf(&ctx, &buf);
>  	return ret;
>  }
>  
> @@ -1445,15 +1441,20 @@ int __cgroup_bpf_run_filter_getsockopt(struct sock *sk, int level,
>  				       int __user *optlen, int max_optlen,
>  				       int retval)
>  {
> -	struct cgroup *cgrp = sock_cgroup_ptr(&sk->sk_cgrp_data);
> -	struct bpf_sockopt_kern ctx = {
> -		.sk = sk,
> -		.level = level,
> -		.optname = optname,
> -		.retval = retval,
> -	};
This change looks unnecessary?

> +	struct bpf_sockopt_kern ctx;
> +	struct bpf_sockopt_buf buf;
> +	struct cgroup *cgrp;
>  	int ret;
>  
> +	memset(&buf, 0, sizeof(buf));
> +	memset(&ctx, 0, sizeof(ctx));
> +
> +	cgrp = sock_cgroup_ptr(&sk->sk_cgrp_data);
> +	ctx.sk = sk;
> +	ctx.level = level;
> +	ctx.optname = optname;
> +	ctx.retval = retval;
> +

Re: [PATCH bpf-next v4 3/3] bpf: remove extra lock_sock for TCP_ZEROCOPY_RECEIVE

[Stanislav Fomichev]

On Thu, Jan 7, 2021 at 5:09 PM Martin KaFai Lau <kafai@fb.com> wrote:
>
> On Thu, Jan 07, 2021 at 10:43:05AM -0800, Stanislav Fomichev wrote:
> > Add custom implementation of getsockopt hook for TCP_ZEROCOPY_RECEIVE.
> > We skip generic hooks for TCP_ZEROCOPY_RECEIVE and have a custom
> > call in do_tcp_getsockopt using the on-stack data. This removes
> > 2% overhead for locking/unlocking the socket.
> >
> > Also:
> > - Removed BUILD_BUG_ON (zerocopy doesn't depend on the buf size anymore)
> > - Separated on-stack buffer into bpf_sockopt_buf and downsized to 32 bytes
> >   (let's keep it to help with the other options)
> >
> > (I can probably split this patch into two: add new features and rework
> >  bpf_sockopt_buf; can follow up if the approach in general sounds
> >  good).
> >
> > Without this patch:
> >      1.87%     0.06%  tcp_mmap  [kernel.kallsyms]  [k] __cgroup_bpf_run_filter_getsockopt
> >
> > With the patch applied:
> >      0.52%     0.12%  tcp_mmap  [kernel.kallsyms]  [k] __cgroup_bpf_run_filter_getsockopt_kern
> >
> > Signed-off-by: Stanislav Fomichev <sdf@google.com>
> > Cc: Martin KaFai Lau <kafai@fb.com>
> > Cc: Song Liu <songliubraving@fb.com>
> > Cc: Eric Dumazet <edumazet@google.com>
> > ---
> >  include/linux/bpf-cgroup.h                    | 25 ++++-
> >  include/linux/filter.h                        |  6 +-
> >  include/net/sock.h                            |  2 +
> >  include/net/tcp.h                             |  1 +
> >  kernel/bpf/cgroup.c                           | 93 +++++++++++++------
> >  net/ipv4/tcp.c                                | 14 +++
> >  net/ipv4/tcp_ipv4.c                           |  1 +
> >  net/ipv6/tcp_ipv6.c                           |  1 +
> >  .../selftests/bpf/prog_tests/sockopt_sk.c     | 22 +++++
> >  .../testing/selftests/bpf/progs/sockopt_sk.c  | 15 +++
> >  10 files changed, 147 insertions(+), 33 deletions(-)
> >
>
> [ ... ]
>
> > @@ -454,6 +469,8 @@ static inline int bpf_percpu_cgroup_storage_update(struct bpf_map *map,
> >  #define BPF_CGROUP_GETSOCKOPT_MAX_OPTLEN(optlen) ({ 0; })
> >  #define BPF_CGROUP_RUN_PROG_GETSOCKOPT(sock, level, optname, optval, \
> >                                      optlen, max_optlen, retval) ({ retval; })
> > +#define BPF_CGROUP_RUN_PROG_GETSOCKOPT_KERN(sock, level, optname, optval, \
> > +                                         optlen, retval) ({ retval; })
> >  #define BPF_CGROUP_RUN_PROG_SETSOCKOPT(sock, level, optname, optval, optlen, \
> >                                      kernel_optval) ({ 0; })
> >
> > diff --git a/include/linux/filter.h b/include/linux/filter.h
> > index 54a4225f36d8..8739f1d4cac4 100644
> > --- a/include/linux/filter.h
> > +++ b/include/linux/filter.h
> > @@ -1281,7 +1281,10 @@ struct bpf_sysctl_kern {
> >       u64 tmp_reg;
> >  };
> >
> > -#define BPF_SOCKOPT_KERN_BUF_SIZE    64
> > +#define BPF_SOCKOPT_KERN_BUF_SIZE    32
> It is reduced from patch 1 because there is no
> need to use the buf (and copy from/to buf) in TCP_ZEROCOPY_RECEIVE?
>
> Patch 1 is still desired (and kept in this set) because it may still
> benefit other optname?
Right, it seems like a good idea to keep it to help with the (majority?)
of small socket options.

> > +struct bpf_sockopt_buf {
> > +     u8              data[BPF_SOCKOPT_KERN_BUF_SIZE];
> > +};
> >
> >  struct bpf_sockopt_kern {
> >       struct sock     *sk;
> > @@ -1291,7 +1294,6 @@ struct bpf_sockopt_kern {
> >       s32             optname;
> >       s32             optlen;
> >       s32             retval;
> > -     u8              buf[BPF_SOCKOPT_KERN_BUF_SIZE];
> It is better to pick one way to do thing to avoid code
> churn like this within the same series.
Agreed. I pointed it out in the commit description that it might be a
good idea to separate those changes.
I wasn't sure about the fate of this patch when I first sent it out
and didn't spend too much time on this sort of stuff.
Let me simplify/reorder as you suggested below and resend.

> >  };
> >
> >  int copy_bpf_fprog_from_user(struct sock_fprog *dst, sockptr_t src, int len);
> > diff --git a/include/net/sock.h b/include/net/sock.h
> > index bdc4323ce53c..ebf44d724845 100644
> > --- a/include/net/sock.h
> > +++ b/include/net/sock.h
> > @@ -1174,6 +1174,8 @@ struct proto {
> >
> >       int                     (*backlog_rcv) (struct sock *sk,
> >                                               struct sk_buff *skb);
> > +     bool                    (*bpf_bypass_getsockopt)(int level,
> > +                                                      int optname);
> >
> >       void            (*release_cb)(struct sock *sk);
> >
> > diff --git a/include/net/tcp.h b/include/net/tcp.h
> > index 78d13c88720f..4bb42fb19711 100644
> > --- a/include/net/tcp.h
> > +++ b/include/net/tcp.h
> > @@ -403,6 +403,7 @@ __poll_t tcp_poll(struct file *file, struct socket *sock,
> >                     struct poll_table_struct *wait);
> >  int tcp_getsockopt(struct sock *sk, int level, int optname,
> >                  char __user *optval, int __user *optlen);
> > +bool tcp_bpf_bypass_getsockopt(int level, int optname);
> >  int tcp_setsockopt(struct sock *sk, int level, int optname, sockptr_t optval,
> >                  unsigned int optlen);
> >  void tcp_set_keepalive(struct sock *sk, int val);
> > diff --git a/kernel/bpf/cgroup.c b/kernel/bpf/cgroup.c
> > index adbecdcaa370..e82df63aedc7 100644
> > --- a/kernel/bpf/cgroup.c
> > +++ b/kernel/bpf/cgroup.c
> > @@ -16,7 +16,6 @@
> >  #include <linux/bpf-cgroup.h>
> >  #include <net/sock.h>
> >  #include <net/bpf_sk_storage.h>
> > -#include <uapi/linux/tcp.h> /* sizeof(struct tcp_zerocopy_receive) */
> Can the patches be re-ordered a little to avoid code churn like this
> in the same series?
>
> It feels like this patch 3 should be the first patch instead.
> The current patch 1 should be the second patch
> but it can still use the tcp_mmap to show potential
> benefit for other optnames.
>
> >
> >  #include "../cgroup/cgroup-internal.h"
> >
> > @@ -1299,7 +1298,8 @@ static bool __cgroup_bpf_prog_array_is_empty(struct cgroup *cgrp,
> >       return empty;
> >  }
> >
> > -static int sockopt_alloc_buf(struct bpf_sockopt_kern *ctx, int max_optlen)
> > +static int sockopt_alloc_buf(struct bpf_sockopt_kern *ctx, int max_optlen,
> > +                          struct bpf_sockopt_buf *buf)
> >  {
> >       if (unlikely(max_optlen < 0))
> >               return -EINVAL;
> > @@ -1311,18 +1311,11 @@ static int sockopt_alloc_buf(struct bpf_sockopt_kern *ctx, int max_optlen)
> >               max_optlen = PAGE_SIZE;
> >       }
> >
> > -     if (max_optlen <= sizeof(ctx->buf)) {
> > +     if (max_optlen <= sizeof(buf->data)) {
> >               /* When the optval fits into BPF_SOCKOPT_KERN_BUF_SIZE
> >                * bytes avoid the cost of kzalloc.
> > -              *
> > -              * In order to remove extra allocations from the TCP
> > -              * fast zero-copy path ensure that buffer covers
> > -              * the size of struct tcp_zerocopy_receive.
> >                */
> > -             BUILD_BUG_ON(sizeof(struct tcp_zerocopy_receive) >
> > -                          BPF_SOCKOPT_KERN_BUF_SIZE);
> > -
> > -             ctx->optval = ctx->buf;
> > +             ctx->optval = buf->data;
> >               ctx->optval_end = ctx->optval + max_optlen;
> >               return max_optlen;
> >       }
> > @@ -1336,16 +1329,18 @@ static int sockopt_alloc_buf(struct bpf_sockopt_kern *ctx, int max_optlen)
> >       return max_optlen;
> >  }
> >
> > -static void sockopt_free_buf(struct bpf_sockopt_kern *ctx)
> > +static void sockopt_free_buf(struct bpf_sockopt_kern *ctx,
> > +                          struct bpf_sockopt_buf *buf)
> >  {
> > -     if (ctx->optval == ctx->buf)
> > +     if (ctx->optval == buf->data)
> >               return;
> >       kfree(ctx->optval);
> >  }
> >
> > -static bool sockopt_buf_allocated(struct bpf_sockopt_kern *ctx)
> > +static bool sockopt_buf_allocated(struct bpf_sockopt_kern *ctx,
> > +                               struct bpf_sockopt_buf *buf)
> >  {
> > -     return ctx->optval != ctx->buf;
> > +     return ctx->optval != buf->data;
> >  }
> >
> >  int __cgroup_bpf_run_filter_setsockopt(struct sock *sk, int *level,
> > @@ -1353,6 +1348,7 @@ int __cgroup_bpf_run_filter_setsockopt(struct sock *sk, int *level,
> >                                      int *optlen, char **kernel_optval)
> >  {
> >       struct cgroup *cgrp = sock_cgroup_ptr(&sk->sk_cgrp_data);
> > +     struct bpf_sockopt_buf buf = {};
> >       struct bpf_sockopt_kern ctx = {
> >               .sk = sk,
> >               .level = *level,
> > @@ -1373,7 +1369,7 @@ int __cgroup_bpf_run_filter_setsockopt(struct sock *sk, int *level,
> >        */
> >       max_optlen = max_t(int, 16, *optlen);
> >
> > -     max_optlen = sockopt_alloc_buf(&ctx, max_optlen);
> > +     max_optlen = sockopt_alloc_buf(&ctx, max_optlen, &buf);
> >       if (max_optlen < 0)
> >               return max_optlen;
> >
> > @@ -1419,7 +1415,7 @@ int __cgroup_bpf_run_filter_setsockopt(struct sock *sk, int *level,
> >                        * No way to export on-stack buf, have to allocate a
> >                        * new buffer.
> >                        */
> > -                     if (!sockopt_buf_allocated(&ctx)) {
> > +                     if (!sockopt_buf_allocated(&ctx, &buf)) {
> >                               void *p = kzalloc(ctx.optlen, GFP_USER);
> >
> >                               if (!p) {
> > @@ -1436,7 +1432,7 @@ int __cgroup_bpf_run_filter_setsockopt(struct sock *sk, int *level,
> >
> >  out:
> >       if (ret)
> > -             sockopt_free_buf(&ctx);
> > +             sockopt_free_buf(&ctx, &buf);
> >       return ret;
> >  }
> >
> > @@ -1445,15 +1441,20 @@ int __cgroup_bpf_run_filter_getsockopt(struct sock *sk, int level,
> >                                      int __user *optlen, int max_optlen,
> >                                      int retval)
> >  {
> > -     struct cgroup *cgrp = sock_cgroup_ptr(&sk->sk_cgrp_data);
> > -     struct bpf_sockopt_kern ctx = {
> > -             .sk = sk,
> > -             .level = level,
> > -             .optname = optname,
> > -             .retval = retval,
> > -     };
> This change looks unnecessary?
>
> > +     struct bpf_sockopt_kern ctx;
> > +     struct bpf_sockopt_buf buf;
> > +     struct cgroup *cgrp;
> >       int ret;
> >
> > +     memset(&buf, 0, sizeof(buf));
> > +     memset(&ctx, 0, sizeof(ctx));
> > +
> > +     cgrp = sock_cgroup_ptr(&sk->sk_cgrp_data);
> > +     ctx.sk = sk;
> > +     ctx.level = level;
> > +     ctx.optname = optname;
> > +     ctx.retval = retval;
> > +