net: filter: Just In Time compiler for x86-64

In order to speedup packet filtering, here is an implementation of a
JIT compiler for x86_64

It is disabled by default, and must be enabled by the admin.

echo 1 >/proc/sys/net/core/bpf_jit_enable

It uses module_alloc() and module_free() to get memory in the 2GB text
kernel range since we call helpers functions from the generated code.

EAX : BPF A accumulator
EBX : BPF X accumulator
RDI : pointer to skb   (first argument given to JIT function)
RBP : frame pointer (even if CONFIG_FRAME_POINTER=n)
r9d : skb->len - skb->data_len (headlen)
r8  : skb->data

To get a trace of generated code, use :

echo 2 >/proc/sys/net/core/bpf_jit_enable

Example of generated code :

# tcpdump -p -n -s 0 -i eth1 host 192.168.20.0/24

flen=18 proglen=147 pass=3 image=ffffffffa00b5000
JIT code: ffffffffa00b5000: 55 48 89 e5 48 83 ec 60 48 89 5d f8 44 8b 4f 60
JIT code: ffffffffa00b5010: 44 2b 4f 64 4c 8b 87 b8 00 00 00 be 0c 00 00 00
JIT code: ffffffffa00b5020: e8 24 7b f7 e0 3d 00 08 00 00 75 28 be 1a 00 00
JIT code: ffffffffa00b5030: 00 e8 fe 7a f7 e0 24 00 3d 00 14 a8 c0 74 49 be
JIT code: ffffffffa00b5040: 1e 00 00 00 e8 eb 7a f7 e0 24 00 3d 00 14 a8 c0
JIT code: ffffffffa00b5050: 74 36 eb 3b 3d 06 08 00 00 74 07 3d 35 80 00 00
JIT code: ffffffffa00b5060: 75 2d be 1c 00 00 00 e8 c8 7a f7 e0 24 00 3d 00
JIT code: ffffffffa00b5070: 14 a8 c0 74 13 be 26 00 00 00 e8 b5 7a f7 e0 24
JIT code: ffffffffa00b5080: 00 3d 00 14 a8 c0 75 07 b8 ff ff 00 00 eb 02 31
JIT code: ffffffffa00b5090: c0 c9 c3

BPF program is 144 bytes long, so native program is almost same size ;)

(000) ldh      [12]
(001) jeq      #0x800           jt 2    jf 8
(002) ld       [26]
(003) and      #0xffffff00
(004) jeq      #0xc0a81400      jt 16   jf 5
(005) ld       [30]
(006) and      #0xffffff00
(007) jeq      #0xc0a81400      jt 16   jf 17
(008) jeq      #0x806           jt 10   jf 9
(009) jeq      #0x8035          jt 10   jf 17
(010) ld       [28]
(011) and      #0xffffff00
(012) jeq      #0xc0a81400      jt 16   jf 13
(013) ld       [38]
(014) and      #0xffffff00
(015) jeq      #0xc0a81400      jt 16   jf 17
(016) ret      #65535
(017) ret      #0
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Arnaldo Carvalho de Melo <acme@infradead.org>
Cc: Ben Hutchings <bhutchings@solarflare.com>
Cc: Hagen Paul Pfeifer <hagen@jauu.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

Documentation/sysctl/net.txt

@@ -32,6 +32,17 @@ Table : Subdirectories in /proc/sys/net
 1. /proc/sys/net/core - Network core options
 -------------------------------------------------------
 
+bpf_jit_enable
+--------------
+
+This enables Berkeley Packet Filter Just in Time compiler.
+Currently supported on x86_64 architecture, bpf_jit provides a framework
+to speed packet filtering, the one used by tcpdump/libpcap for example.
+Values :
+	0 - disable the JIT (default value)
+	1 - enable the JIT
+	2 - enable the JIT and ask the compiler to emit traces on kernel log.
+
 rmem_default
 ------------
 

MAINTAINERS

@@ -4372,6 +4372,7 @@ S:	Maintained
 F:	net/ipv4/
 F:	net/ipv6/
 F:	include/net/ip*
+F:	arch/x86/net/*
 
 NETWORKING [LABELED] (NetLabel, CIPSO, Labeled IPsec, SECMARK)
 M:	Paul Moore <paul.moore@hp.com>

arch/x86/Kbuild

@@ -15,3 +15,4 @@ obj-y += vdso/
 obj-$(CONFIG_IA32_EMULATION) += ia32/
 
 obj-y += platform/
+obj-y += net/

arch/x86/Kconfig

@@ -72,6 +72,7 @@ config X86
 	select IRQ_FORCED_THREADING
 	select USE_GENERIC_SMP_HELPERS if SMP
 	select ARCH_NO_SYSDEV_OPS
+	select HAVE_BPF_JIT if X86_64
 
 config INSTRUCTION_DECODER
 	def_bool (KPROBES || PERF_EVENTS)

arch/x86/net/Makefile

+#
+# Arch-specific network modules
+#
+obj-$(CONFIG_BPF_JIT) += bpf_jit.o bpf_jit_comp.o

arch/x86/net/bpf_jit.S

+/* bpf_jit.S : BPF JIT helper functions
+ *
+ * Copyright (C) 2011 Eric Dumazet (eric.dumazet@gmail.com)
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; version 2
+ * of the License.
+ */
+#include <linux/linkage.h>
+#include <asm/dwarf2.h>
+
+/*
+ * Calling convention :
+ * rdi : skb pointer
+ * esi : offset of byte(s) to fetch in skb (can be scratched)
+ * r8  : copy of skb->data
+ * r9d : hlen = skb->len - skb->data_len
+ */
+#define SKBDATA	%r8
+
+sk_load_word_ind:
+	.globl	sk_load_word_ind
+
+	add	%ebx,%esi	/* offset += X */
+#	test    %esi,%esi	/* if (offset < 0) goto bpf_error; */
+	js	bpf_error
+
+sk_load_word:
+	.globl	sk_load_word
+
+	mov	%r9d,%eax		# hlen
+	sub	%esi,%eax		# hlen - offset
+	cmp	$3,%eax
+	jle	bpf_slow_path_word
+	mov     (SKBDATA,%rsi),%eax
+	bswap   %eax  			/* ntohl() */
+	ret
+
+
+sk_load_half_ind:
+	.globl sk_load_half_ind
+
+	add	%ebx,%esi	/* offset += X */
+	js	bpf_error
+
+sk_load_half:
+	.globl	sk_load_half
+
+	mov	%r9d,%eax
+	sub	%esi,%eax		#	hlen - offset
+	cmp	$1,%eax
+	jle	bpf_slow_path_half
+	movzwl	(SKBDATA,%rsi),%eax
+	rol	$8,%ax			# ntohs()
+	ret
+
+sk_load_byte_ind:
+	.globl sk_load_byte_ind
+	add	%ebx,%esi	/* offset += X */
+	js	bpf_error
+
+sk_load_byte:
+	.globl	sk_load_byte
+
+	cmp	%esi,%r9d   /* if (offset >= hlen) goto bpf_slow_path_byte */
+	jle	bpf_slow_path_byte
+	movzbl	(SKBDATA,%rsi),%eax
+	ret
+
+/**
+ * sk_load_byte_msh - BPF_S_LDX_B_MSH helper
+ *
+ * Implements BPF_S_LDX_B_MSH : ldxb  4*([offset]&0xf)
+ * Must preserve A accumulator (%eax)
+ * Inputs : %esi is the offset value, already known positive
+ */
+ENTRY(sk_load_byte_msh)
+	CFI_STARTPROC
+	cmp	%esi,%r9d      /* if (offset >= hlen) goto bpf_slow_path_byte_msh */
+	jle	bpf_slow_path_byte_msh
+	movzbl	(SKBDATA,%rsi),%ebx
+	and	$15,%bl
+	shl	$2,%bl
+	ret
+	CFI_ENDPROC
+ENDPROC(sk_load_byte_msh)
+
+bpf_error:
+# force a return 0 from jit handler
+	xor		%eax,%eax
+	mov		-8(%rbp),%rbx
+	leaveq
+	ret
+
+/* rsi contains offset and can be scratched */
+#define bpf_slow_path_common(LEN)		\
+	push	%rdi;    /* save skb */		\
+	push	%r9;				\
+	push	SKBDATA;			\
+/* rsi already has offset */			\
+	mov	$LEN,%ecx;	/* len */	\
+	lea	-12(%rbp),%rdx;			\
+	call	skb_copy_bits;			\
+	test    %eax,%eax;			\
+	pop	SKBDATA;			\
+	pop	%r9;				\
+	pop	%rdi
+
+
+bpf_slow_path_word:
+	bpf_slow_path_common(4)
+	js	bpf_error
+	mov	-12(%rbp),%eax
+	bswap	%eax
+	ret
+
+bpf_slow_path_half:
+	bpf_slow_path_common(2)
+	js	bpf_error
+	mov	-12(%rbp),%ax
+	rol	$8,%ax
+	movzwl	%ax,%eax
+	ret
+
+bpf_slow_path_byte:
+	bpf_slow_path_common(1)
+	js	bpf_error
+	movzbl	-12(%rbp),%eax
+	ret
+
+bpf_slow_path_byte_msh:
+	xchg	%eax,%ebx /* dont lose A , X is about to be scratched */
+	bpf_slow_path_common(1)
+	js	bpf_error
+	movzbl	-12(%rbp),%eax
+	and	$15,%al
+	shl	$2,%al
+	xchg	%eax,%ebx
+	ret

include/linux/filter.h

@@ -135,6 +135,8 @@ struct sk_filter
 {
 	atomic_t		refcnt;
 	unsigned int         	len;	/* Number of filter blocks */
+	unsigned int		(*bpf_func)(const struct sk_buff *skb,
+					    const struct sock_filter *filter);
 	struct rcu_head		rcu;
 	struct sock_filter     	insns[0];
 };
@@ -153,6 +155,80 @@ extern unsigned int sk_run_filter(const struct sk_buff *skb,
 extern int sk_attach_filter(struct sock_fprog *fprog, struct sock *sk);
 extern int sk_detach_filter(struct sock *sk);
 extern int sk_chk_filter(struct sock_filter *filter, int flen);
+
+#ifdef CONFIG_BPF_JIT
+extern void bpf_jit_compile(struct sk_filter *fp);
+extern void bpf_jit_free(struct sk_filter *fp);
+#define SK_RUN_FILTER(FILTER, SKB) (*FILTER->bpf_func)(SKB, FILTER->insns)
+#else
+static inline void bpf_jit_compile(struct sk_filter *fp)
+{
+}
+static inline void bpf_jit_free(struct sk_filter *fp)
+{
+}
+#define SK_RUN_FILTER(FILTER, SKB) sk_run_filter(SKB, FILTER->insns)
+#endif
+
+enum {
+	BPF_S_RET_K = 1,
+	BPF_S_RET_A,
+	BPF_S_ALU_ADD_K,
+	BPF_S_ALU_ADD_X,
+	BPF_S_ALU_SUB_K,
+	BPF_S_ALU_SUB_X,
+	BPF_S_ALU_MUL_K,
+	BPF_S_ALU_MUL_X,
+	BPF_S_ALU_DIV_X,
+	BPF_S_ALU_AND_K,
+	BPF_S_ALU_AND_X,
+	BPF_S_ALU_OR_K,
+	BPF_S_ALU_OR_X,
+	BPF_S_ALU_LSH_K,
+	BPF_S_ALU_LSH_X,
+	BPF_S_ALU_RSH_K,
+	BPF_S_ALU_RSH_X,
+	BPF_S_ALU_NEG,
+	BPF_S_LD_W_ABS,
+	BPF_S_LD_H_ABS,
+	BPF_S_LD_B_ABS,
+	BPF_S_LD_W_LEN,
+	BPF_S_LD_W_IND,
+	BPF_S_LD_H_IND,
+	BPF_S_LD_B_IND,
+	BPF_S_LD_IMM,
+	BPF_S_LDX_W_LEN,
+	BPF_S_LDX_B_MSH,
+	BPF_S_LDX_IMM,
+	BPF_S_MISC_TAX,
+	BPF_S_MISC_TXA,
+	BPF_S_ALU_DIV_K,
+	BPF_S_LD_MEM,
+	BPF_S_LDX_MEM,
+	BPF_S_ST,
+	BPF_S_STX,
+	BPF_S_JMP_JA,
+	BPF_S_JMP_JEQ_K,
+	BPF_S_JMP_JEQ_X,
+	BPF_S_JMP_JGE_K,
+	BPF_S_JMP_JGE_X,
+	BPF_S_JMP_JGT_K,
+	BPF_S_JMP_JGT_X,
+	BPF_S_JMP_JSET_K,
+	BPF_S_JMP_JSET_X,
+	/* Ancillary data */
+	BPF_S_ANC_PROTOCOL,
+	BPF_S_ANC_PKTTYPE,
+	BPF_S_ANC_IFINDEX,
+	BPF_S_ANC_NLATTR,
+	BPF_S_ANC_NLATTR_NEST,
+	BPF_S_ANC_MARK,
+	BPF_S_ANC_QUEUE,
+	BPF_S_ANC_HATYPE,
+	BPF_S_ANC_RXHASH,
+	BPF_S_ANC_CPU,
+};
+
 #endif /* __KERNEL__ */
 
 #endif /* __LINUX_FILTER_H__ */

include/linux/netdevice.h

@@ -2514,6 +2514,7 @@ extern struct rtnl_link_stats64 *dev_get_stats(struct net_device *dev,
 extern int		netdev_max_backlog;
 extern int		netdev_tstamp_prequeue;
 extern int		weight_p;
+extern int		bpf_jit_enable;
 extern int		netdev_set_master(struct net_device *dev, struct net_device *master);
 extern int netdev_set_bond_master(struct net_device *dev,
 				  struct net_device *master);

include/linux/skbuff.h

@@ -391,8 +391,8 @@ struct sk_buff {
 
 	__u32			rxhash;
 
+	__u16			queue_mapping;
 	kmemcheck_bitfield_begin(flags2);
-	__u16			queue_mapping:16;
 #ifdef CONFIG_IPV6_NDISC_NODETYPE
 	__u8			ndisc_nodetype:2;
 #endif

net/Kconfig

@@ -232,6 +232,19 @@ config XPS
 	depends on SMP && SYSFS && USE_GENERIC_SMP_HELPERS
 	default y
 
+config HAVE_BPF_JIT
+	bool
+
+config BPF_JIT
+	bool "enable BPF Just In Time compiler"
+	depends on HAVE_BPF_JIT
+	---help---
+	  Berkeley Packet Filter filtering capabilities are normally handled
+	  by an interpreter. This option allows kernel to generate a native
+	  code when filter is loaded in memory. This should speedup
+	  packet sniffing (libpcap/tcpdump). Note : Admin should enable
+	  this feature changing /proc/sys/net/core/bpf_jit_enable
+
 menu "Network testing"
 
 config NET_PKTGEN

net/core/filter.c

@@ -39,65 +39,6 @@
 #include <linux/filter.h>
 #include <linux/reciprocal_div.h>
 
-enum {
-	BPF_S_RET_K = 1,
-	BPF_S_RET_A,
-	BPF_S_ALU_ADD_K,
-	BPF_S_ALU_ADD_X,
-	BPF_S_ALU_SUB_K,
-	BPF_S_ALU_SUB_X,
-	BPF_S_ALU_MUL_K,
-	BPF_S_ALU_MUL_X,
-	BPF_S_ALU_DIV_X,
-	BPF_S_ALU_AND_K,
-	BPF_S_ALU_AND_X,
-	BPF_S_ALU_OR_K,
-	BPF_S_ALU_OR_X,
-	BPF_S_ALU_LSH_K,
-	BPF_S_ALU_LSH_X,
-	BPF_S_ALU_RSH_K,
-	BPF_S_ALU_RSH_X,
-	BPF_S_ALU_NEG,
-	BPF_S_LD_W_ABS,
-	BPF_S_LD_H_ABS,
-	BPF_S_LD_B_ABS,
-	BPF_S_LD_W_LEN,
-	BPF_S_LD_W_IND,
-	BPF_S_LD_H_IND,
-	BPF_S_LD_B_IND,
-	BPF_S_LD_IMM,
-	BPF_S_LDX_W_LEN,
-	BPF_S_LDX_B_MSH,
-	BPF_S_LDX_IMM,
-	BPF_S_MISC_TAX,
-	BPF_S_MISC_TXA,
-	BPF_S_ALU_DIV_K,
-	BPF_S_LD_MEM,
-	BPF_S_LDX_MEM,
-	BPF_S_ST,
-	BPF_S_STX,
-	BPF_S_JMP_JA,
-	BPF_S_JMP_JEQ_K,
-	BPF_S_JMP_JEQ_X,
-	BPF_S_JMP_JGE_K,
-	BPF_S_JMP_JGE_X,
-	BPF_S_JMP_JGT_K,
-	BPF_S_JMP_JGT_X,
-	BPF_S_JMP_JSET_K,
-	BPF_S_JMP_JSET_X,
-	/* Ancillary data */
-	BPF_S_ANC_PROTOCOL,
-	BPF_S_ANC_PKTTYPE,
-	BPF_S_ANC_IFINDEX,
-	BPF_S_ANC_NLATTR,
-	BPF_S_ANC_NLATTR_NEST,
-	BPF_S_ANC_MARK,
-	BPF_S_ANC_QUEUE,
-	BPF_S_ANC_HATYPE,
-	BPF_S_ANC_RXHASH,
-	BPF_S_ANC_CPU,
-};
-
 /* No hurry in this branch */
 static void *__load_pointer(const struct sk_buff *skb, int k, unsigned int size)
 {
@@ -145,7 +86,7 @@ int sk_filter(struct sock *sk, struct sk_buff *skb)
 	rcu_read_lock();
 	filter = rcu_dereference(sk->sk_filter);
 	if (filter) {
-		unsigned int pkt_len = sk_run_filter(skb, filter->insns);
+		unsigned int pkt_len = SK_RUN_FILTER(filter, skb);
 
 		err = pkt_len ? pskb_trim(skb, pkt_len) : -EPERM;
 	}
@@ -638,6 +579,7 @@ void sk_filter_release_rcu(struct rcu_head *rcu)
 {
 	struct sk_filter *fp = container_of(rcu, struct sk_filter, rcu);
 
+	bpf_jit_free(fp);
 	kfree(fp);
 }
 EXPORT_SYMBOL(sk_filter_release_rcu);
@@ -672,6 +614,7 @@ int sk_attach_filter(struct sock_fprog *fprog, struct sock *sk)
 
 	atomic_set(&fp->refcnt, 1);
 	fp->len = fprog->len;
+	fp->bpf_func = sk_run_filter;
 
 	err = sk_chk_filter(fp->insns, fp->len);
 	if (err) {
@@ -679,6 +622,8 @@ int sk_attach_filter(struct sock_fprog *fprog, struct sock *sk)
 		return err;
 	}
 
+	bpf_jit_compile(fp);
+
 	old_fp = rcu_dereference_protected(sk->sk_filter,
 					   sock_owned_by_user(sk));
 	rcu_assign_pointer(sk->sk_filter, fp);

net/core/sysctl_net_core.c

@@ -122,6 +122,15 @@ static struct ctl_table net_core_table[] = {
 		.mode		= 0644,
 		.proc_handler	= proc_dointvec
 	},
+#ifdef CONFIG_BPF_JIT
+	{
+		.procname	= "bpf_jit_enable",
+		.data		= &bpf_jit_enable,
+		.maxlen		= sizeof(int),
+		.mode		= 0644,
+		.proc_handler	= proc_dointvec
+	},
+#endif
 	{
 		.procname	= "netdev_tstamp_prequeue",
 		.data		= &netdev_tstamp_prequeue,

net/packet/af_packet.c

@@ -538,7 +538,7 @@ static inline unsigned int run_filter(const struct sk_buff *skb,
 	rcu_read_lock();
 	filter = rcu_dereference(sk->sk_filter);
 	if (filter != NULL)
-		res = sk_run_filter(skb, filter->insns);
+		res = SK_RUN_FILTER(filter, skb);
 	rcu_read_unlock();
 
 	return res;