Skip to content
Projects
Groups
Snippets
Help
Loading...
Help
Support
Keyboard shortcuts
?
Submit feedback
Contribute to GitLab
Sign in / Register
Toggle navigation
L
linux
Project overview
Project overview
Details
Activity
Releases
Repository
Repository
Files
Commits
Branches
Tags
Contributors
Graph
Compare
Issues
0
Issues
0
List
Boards
Labels
Milestones
Merge Requests
0
Merge Requests
0
Analytics
Analytics
Repository
Value Stream
Wiki
Wiki
Snippets
Snippets
Members
Members
Collapse sidebar
Close sidebar
Activity
Graph
Create a new issue
Commits
Issue Boards
Open sidebar
nexedi
linux
Commits
35a9a271
Commit
35a9a271
authored
Aug 31, 2003
by
Ville Nuorvala
Committed by
David S. Miller
Aug 31, 2003
Browse files
Options
Browse Files
Download
Email Patches
Plain Diff
[IPV6]: Remove sockets from ip6_tunnel.c
parent
71b1bb03
Changes
3
Hide whitespace changes
Inline
Side-by-side
Showing
3 changed files
with
130 additions
and
229 deletions
+130
-229
include/net/ip6_tunnel.h
include/net/ip6_tunnel.h
+2
-0
net/ipv6/ip6_tunnel.c
net/ipv6/ip6_tunnel.c
+127
-229
net/ipv6/ipv6_syms.c
net/ipv6/ipv6_syms.c
+1
-0
No files found.
include/net/ip6_tunnel.h
View file @
35a9a271
...
@@ -25,6 +25,8 @@ struct ip6_tnl {
...
@@ -25,6 +25,8 @@ struct ip6_tnl {
int
recursion
;
/* depth of hard_start_xmit recursion */
int
recursion
;
/* depth of hard_start_xmit recursion */
struct
ip6_tnl_parm
parms
;
/* tunnel configuration paramters */
struct
ip6_tnl_parm
parms
;
/* tunnel configuration paramters */
struct
flowi
fl
;
/* flowi template for xmit */
struct
flowi
fl
;
/* flowi template for xmit */
struct
dst_entry
*
dst_cache
;
/* cached dst */
u32
dst_cookie
;
};
};
/* Tunnel encapsulation limit destination sub-option */
/* Tunnel encapsulation limit destination sub-option */
...
...
net/ipv6/ip6_tunnel.c
View file @
35a9a271
...
@@ -23,7 +23,6 @@
...
@@ -23,7 +23,6 @@
#include <linux/module.h>
#include <linux/module.h>
#include <linux/errno.h>
#include <linux/errno.h>
#include <linux/types.h>
#include <linux/types.h>
#include <linux/socket.h>
#include <linux/sockios.h>
#include <linux/sockios.h>
#include <linux/if.h>
#include <linux/if.h>
#include <linux/in.h>
#include <linux/in.h>
...
@@ -37,12 +36,12 @@
...
@@ -37,12 +36,12 @@
#include <linux/init.h>
#include <linux/init.h>
#include <linux/route.h>
#include <linux/route.h>
#include <linux/rtnetlink.h>
#include <linux/rtnetlink.h>
#include <linux/netfilter_ipv6.h>
#include <asm/uaccess.h>
#include <asm/uaccess.h>
#include <asm/atomic.h>
#include <asm/atomic.h>
#include <net/ip.h>
#include <net/ip.h>
#include <net/sock.h>
#include <net/ipv6.h>
#include <net/ipv6.h>
#include <net/protocol.h>
#include <net/protocol.h>
#include <net/ip6_route.h>
#include <net/ip6_route.h>
...
@@ -63,22 +62,6 @@ MODULE_LICENSE("GPL");
...
@@ -63,22 +62,6 @@ MODULE_LICENSE("GPL");
#define IPV6_TCLASS_MASK (IPV6_FLOWINFO_MASK & ~IPV6_FLOWLABEL_MASK)
#define IPV6_TCLASS_MASK (IPV6_FLOWINFO_MASK & ~IPV6_FLOWLABEL_MASK)
/* socket(s) used by ip6ip6_tnl_xmit() for resending packets */
static
struct
socket
*
__ip6_socket
[
NR_CPUS
];
#define ip6_socket __ip6_socket[smp_processor_id()]
static
void
ip6_xmit_lock
(
void
)
{
local_bh_disable
();
if
(
unlikely
(
!
spin_trylock
(
&
ip6_socket
->
sk
->
sk_lock
.
slock
)))
BUG
();
}
static
void
ip6_xmit_unlock
(
void
)
{
spin_unlock_bh
(
&
ip6_socket
->
sk
->
sk_lock
.
slock
);
}
#define HASH_SIZE 32
#define HASH_SIZE 32
#define HASH(addr) (((addr)->s6_addr32[0] ^ (addr)->s6_addr32[1] ^ \
#define HASH(addr) (((addr)->s6_addr32[0] ^ (addr)->s6_addr32[1] ^ \
...
@@ -101,6 +84,33 @@ static struct ip6_tnl **tnls[2] = { tnls_wc, tnls_r_l };
...
@@ -101,6 +84,33 @@ static struct ip6_tnl **tnls[2] = { tnls_wc, tnls_r_l };
/* lock for the tunnel lists */
/* lock for the tunnel lists */
static
rwlock_t
ip6ip6_lock
=
RW_LOCK_UNLOCKED
;
static
rwlock_t
ip6ip6_lock
=
RW_LOCK_UNLOCKED
;
static
inline
struct
dst_entry
*
ip6_tnl_dst_check
(
struct
ip6_tnl
*
t
)
{
struct
dst_entry
*
dst
=
t
->
dst_cache
;
if
(
dst
&&
dst
->
obsolete
&&
dst
->
ops
->
check
(
dst
,
t
->
dst_cookie
)
==
NULL
)
{
t
->
dst_cache
=
NULL
;
return
NULL
;
}
return
dst
;
}
static
inline
void
ip6_tnl_dst_reset
(
struct
ip6_tnl
*
t
)
{
dst_release
(
t
->
dst_cache
);
t
->
dst_cache
=
NULL
;
}
static
inline
void
ip6_tnl_dst_store
(
struct
ip6_tnl
*
t
,
struct
dst_entry
*
dst
)
{
struct
rt6_info
*
rt
=
(
struct
rt6_info
*
)
dst
;
t
->
dst_cookie
=
rt
->
rt6i_node
?
rt
->
rt6i_node
->
fn_sernum
:
0
;
dst_release
(
t
->
dst_cache
);
t
->
dst_cache
=
dst
;
}
/**
/**
* ip6ip6_tnl_lookup - fetch tunnel matching the end-point addresses
* ip6ip6_tnl_lookup - fetch tunnel matching the end-point addresses
* @remote: the address of the tunnel exit-point
* @remote: the address of the tunnel exit-point
...
@@ -294,13 +304,16 @@ ip6ip6_tnl_locate(struct ip6_tnl_parm *p, struct ip6_tnl **pt, int create)
...
@@ -294,13 +304,16 @@ ip6ip6_tnl_locate(struct ip6_tnl_parm *p, struct ip6_tnl **pt, int create)
static
void
static
void
ip6ip6_tnl_dev_uninit
(
struct
net_device
*
dev
)
ip6ip6_tnl_dev_uninit
(
struct
net_device
*
dev
)
{
{
struct
ip6_tnl
*
t
=
dev
->
priv
;
if
(
dev
==
ip6ip6_fb_tnl_dev
)
{
if
(
dev
==
ip6ip6_fb_tnl_dev
)
{
write_lock_bh
(
&
ip6ip6_lock
);
write_lock_bh
(
&
ip6ip6_lock
);
tnls_wc
[
0
]
=
NULL
;
tnls_wc
[
0
]
=
NULL
;
write_unlock_bh
(
&
ip6ip6_lock
);
write_unlock_bh
(
&
ip6ip6_lock
);
}
else
{
}
else
{
ip6ip6_tnl_unlink
(
(
struct
ip6_tnl
*
)
dev
->
priv
);
ip6ip6_tnl_unlink
(
t
);
}
}
ip6_tnl_dst_reset
(
t
);
dev_put
(
dev
);
dev_put
(
dev
);
}
}
...
@@ -522,112 +535,34 @@ int ip6ip6_rcv(struct sk_buff **pskb, unsigned int *nhoffp)
...
@@ -522,112 +535,34 @@ int ip6ip6_rcv(struct sk_buff **pskb, unsigned int *nhoffp)
return
0
;
return
0
;
}
}
/**
static
inline
struct
ipv6_txoptions
*
create_tel
(
__u8
encap_limit
)
* txopt_len - get necessary size for new &struct ipv6_txoptions
* @orig_opt: old options
*
* Return:
* Size of old one plus size of tunnel encapsulation limit option
**/
static
inline
int
txopt_len
(
struct
ipv6_txoptions
*
orig_opt
)
{
int
len
=
sizeof
(
*
orig_opt
)
+
8
;
if
(
orig_opt
&&
orig_opt
->
dst0opt
)
len
+=
ipv6_optlen
(
orig_opt
->
dst0opt
);
return
len
;
}
/**
* merge_options - add encapsulation limit to original options
* @encap_limit: number of allowed encapsulation limits
* @orig_opt: original options
*
* Return:
* Pointer to new &struct ipv6_txoptions containing the tunnel
* encapsulation limit
**/
static
struct
ipv6_txoptions
*
merge_options
(
struct
sock
*
sk
,
__u8
encap_limit
,
struct
ipv6_txoptions
*
orig_opt
)
{
{
struct
ipv6_tlv_tnl_enc_lim
*
tel
;
struct
ipv6_tlv_tnl_enc_lim
*
tel
;
struct
ipv6_txoptions
*
opt
;
struct
ipv6_txoptions
*
opt
;
__u8
*
raw
;
__u8
*
raw
;
__u8
pad_to
=
8
;
int
opt_len
=
txopt_len
(
orig_opt
);
if
(
!
(
opt
=
sock_kmalloc
(
sk
,
opt_len
,
GFP_ATOMIC
)))
{
int
opt_len
=
sizeof
(
*
opt
)
+
8
;
if
(
!
(
opt
=
kmalloc
(
opt_len
,
GFP_ATOMIC
)))
{
return
NULL
;
return
NULL
;
}
}
memset
(
opt
,
0
,
opt_len
);
memset
(
opt
,
0
,
opt_len
);
opt
->
tot_len
=
opt_len
;
opt
->
tot_len
=
opt_len
;
opt
->
dst0opt
=
(
struct
ipv6_opt_hdr
*
)
(
opt
+
1
);
opt
->
dst0opt
=
(
struct
ipv6_opt_hdr
*
)
(
opt
+
1
);
opt
->
opt_nflen
=
8
;
opt
->
opt_nflen
=
8
;
raw
=
(
__u8
*
)
opt
->
dst0opt
;
tel
=
(
struct
ipv6_tlv_tnl_enc_lim
*
)
(
opt
->
dst0opt
+
1
);
tel
=
(
struct
ipv6_tlv_tnl_enc_lim
*
)
(
opt
->
dst0opt
+
1
);
tel
->
type
=
IPV6_TLV_TNL_ENCAP_LIMIT
;
tel
->
type
=
IPV6_TLV_TNL_ENCAP_LIMIT
;
tel
->
length
=
1
;
tel
->
length
=
1
;
tel
->
encap_limit
=
encap_limit
;
tel
->
encap_limit
=
encap_limit
;
if
(
orig_opt
)
{
raw
=
(
__u8
*
)
opt
->
dst0opt
;
__u8
*
orig_raw
;
opt
->
hopopt
=
orig_opt
->
hopopt
;
/* Keep the original destination options properly
aligned and merge possible old paddings to the
new padding option */
if
((
orig_raw
=
(
__u8
*
)
orig_opt
->
dst0opt
)
!=
NULL
)
{
__u8
type
;
int
i
=
sizeof
(
struct
ipv6_opt_hdr
);
pad_to
+=
sizeof
(
struct
ipv6_opt_hdr
);
while
(
i
<
ipv6_optlen
(
orig_opt
->
dst0opt
))
{
type
=
orig_raw
[
i
++
];
if
(
type
==
IPV6_TLV_PAD0
)
pad_to
++
;
else
if
(
type
==
IPV6_TLV_PADN
)
{
int
len
=
orig_raw
[
i
++
];
i
+=
len
;
pad_to
+=
len
+
2
;
}
else
{
break
;
}
}
opt
->
dst0opt
->
hdrlen
=
orig_opt
->
dst0opt
->
hdrlen
+
1
;
memcpy
(
raw
+
pad_to
,
orig_raw
+
pad_to
-
8
,
opt_len
-
sizeof
(
*
opt
)
-
pad_to
);
}
opt
->
srcrt
=
orig_opt
->
srcrt
;
opt
->
opt_nflen
+=
orig_opt
->
opt_nflen
;
opt
->
dst1opt
=
orig_opt
->
dst1opt
;
opt
->
auth
=
orig_opt
->
auth
;
opt
->
opt_flen
=
orig_opt
->
opt_flen
;
}
raw
[
5
]
=
IPV6_TLV_PADN
;
raw
[
5
]
=
IPV6_TLV_PADN
;
raw
[
6
]
=
1
;
/* subtract lengths of destination suboption header,
tunnel encapsulation limit and pad N header */
raw
[
6
]
=
pad_to
-
7
;
return
opt
;
return
opt
;
}
}
static
int
ip6ip6_getfrag
(
void
*
from
,
char
*
to
,
int
offset
,
int
len
,
int
odd
,
struct
sk_buff
*
skb
)
{
memcpy
(
to
,
(
char
*
)
from
+
offset
,
len
);
return
0
;
}
/**
/**
* ip6ip6_tnl_addr_conflict - compare packet addresses to tunnel's own
* ip6ip6_tnl_addr_conflict - compare packet addresses to tunnel's own
* @t: the outgoing tunnel device
* @t: the outgoing tunnel device
...
@@ -655,7 +590,7 @@ ip6ip6_tnl_addr_conflict(struct ip6_tnl *t, struct ipv6hdr *hdr)
...
@@ -655,7 +590,7 @@ ip6ip6_tnl_addr_conflict(struct ip6_tnl *t, struct ipv6hdr *hdr)
*
*
* Description:
* Description:
* Build new header and do some sanity checks on the packet before sending
* Build new header and do some sanity checks on the packet before sending
* it
to ip6_build_xmit()
.
* it.
*
*
* Return:
* Return:
* 0
* 0
...
@@ -666,18 +601,17 @@ int ip6ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev)
...
@@ -666,18 +601,17 @@ int ip6ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev)
struct
ip6_tnl
*
t
=
(
struct
ip6_tnl
*
)
dev
->
priv
;
struct
ip6_tnl
*
t
=
(
struct
ip6_tnl
*
)
dev
->
priv
;
struct
net_device_stats
*
stats
=
&
t
->
stat
;
struct
net_device_stats
*
stats
=
&
t
->
stat
;
struct
ipv6hdr
*
ipv6h
=
skb
->
nh
.
ipv6h
;
struct
ipv6hdr
*
ipv6h
=
skb
->
nh
.
ipv6h
;
struct
ipv6_txoptions
*
orig_opt
=
NULL
;
struct
ipv6_txoptions
*
opt
=
NULL
;
struct
ipv6_txoptions
*
opt
=
NULL
;
int
encap_limit
=
-
1
;
int
encap_limit
=
-
1
;
__u16
offset
;
__u16
offset
;
struct
flowi
fl
;
struct
flowi
fl
;
struct
ip6_flowlabel
*
fl_lbl
=
NULL
;
int
err
=
0
;
struct
dst_entry
*
dst
;
struct
dst_entry
*
dst
;
int
link_failure
=
0
;
struct
net_device
*
tdev
;
struct
sock
*
sk
=
ip6_socket
->
sk
;
struct
ipv6_pinfo
*
np
=
inet6_sk
(
sk
);
int
mtu
;
int
mtu
;
int
max_headroom
=
sizeof
(
struct
ipv6hdr
);
u8
proto
;
int
err
;
int
pkt_len
;
if
(
t
->
recursion
++
)
{
if
(
t
->
recursion
++
)
{
stats
->
collisions
++
;
stats
->
collisions
++
;
...
@@ -700,58 +634,39 @@ int ip6ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev)
...
@@ -700,58 +634,39 @@ int ip6ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev)
}
else
if
(
!
(
t
->
parms
.
flags
&
IP6_TNL_F_IGN_ENCAP_LIMIT
))
{
}
else
if
(
!
(
t
->
parms
.
flags
&
IP6_TNL_F_IGN_ENCAP_LIMIT
))
{
encap_limit
=
t
->
parms
.
encap_limit
;
encap_limit
=
t
->
parms
.
encap_limit
;
}
}
ip6_xmit_lock
();
memcpy
(
&
fl
,
&
t
->
fl
,
sizeof
(
fl
));
memcpy
(
&
fl
,
&
t
->
fl
,
sizeof
(
fl
));
proto
=
fl
.
proto
;
if
((
t
->
parms
.
flags
&
IP6_TNL_F_USE_ORIG_TCLASS
))
if
((
t
->
parms
.
flags
&
IP6_TNL_F_USE_ORIG_TCLASS
))
fl
.
fl6_flowlabel
|=
(
*
(
__u32
*
)
ipv6h
&
IPV6_TCLASS_MASK
);
fl
.
fl6_flowlabel
|=
(
*
(
__u32
*
)
ipv6h
&
IPV6_TCLASS_MASK
);
if
((
t
->
parms
.
flags
&
IP6_TNL_F_USE_ORIG_FLOWLABEL
))
if
((
t
->
parms
.
flags
&
IP6_TNL_F_USE_ORIG_FLOWLABEL
))
fl
.
fl6_flowlabel
|=
(
*
(
__u32
*
)
ipv6h
&
IPV6_FLOWLABEL_MASK
);
fl
.
fl6_flowlabel
|=
(
*
(
__u32
*
)
ipv6h
&
IPV6_FLOWLABEL_MASK
);
if
(
fl
.
fl6_flowlabel
)
{
if
(
encap_limit
>=
0
&&
(
opt
=
create_tel
(
encap_limit
))
==
NULL
)
fl_lbl
=
fl6_sock_lookup
(
sk
,
fl
.
fl6_flowlabel
);
goto
tx_err
;
if
(
fl_lbl
)
orig_opt
=
fl_lbl
->
opt
;
}
if
(
encap_limit
>=
0
)
{
if
(
!
(
opt
=
merge_options
(
sk
,
encap_limit
,
orig_opt
)))
{
goto
tx_err_free_fl_lbl
;
}
}
else
{
opt
=
orig_opt
;
}
dst
=
__sk_dst_check
(
sk
,
np
->
dst_cookie
);
if
(
dst
)
{
if
((
dst
=
ip6_tnl_dst_check
(
t
))
!=
NULL
)
if
(
np
->
daddr_cache
==
NULL
||
dst_hold
(
dst
);
ipv6_addr_cmp
(
&
fl
.
fl6_dst
,
np
->
daddr_cache
)
||
else
(
fl
.
oif
&&
fl
.
oif
!=
dst
->
dev
->
ifindex
))
{
dst
=
ip6_route_output
(
NULL
,
&
fl
);
dst
=
NULL
;
}
if
(
dst
->
error
||
xfrm_lookup
(
&
dst
,
&
fl
,
NULL
,
0
)
<
0
)
}
goto
tx_err_link_failure
;
if
(
dst
==
NULL
)
{
dst
=
ip6_route_output
(
sk
,
&
fl
);
tdev
=
dst
->
dev
;
if
(
dst
->
error
)
{
stats
->
tx_carrier_errors
++
;
if
(
tdev
==
dev
)
{
link_failure
=
1
;
stats
->
collisions
++
;
goto
tx_err_dst_release
;
if
(
net_ratelimit
())
}
printk
(
KERN_WARNING
/* local routing loop */
"%s: Local routing loop detected!
\n
"
,
if
(
dst
->
dev
==
dev
)
{
t
->
parms
.
name
);
stats
->
collisions
++
;
goto
tx_err_dst_release
;
if
(
net_ratelimit
())
printk
(
KERN_WARNING
"%s: Local routing loop detected!
\n
"
,
t
->
parms
.
name
);
goto
tx_err_dst_release
;
}
ipv6_addr_copy
(
&
np
->
daddr
,
&
fl
.
fl6_dst
);
ipv6_addr_copy
(
&
np
->
saddr
,
&
fl
.
fl6_src
);
}
}
mtu
=
dst_pmtu
(
dst
)
-
sizeof
(
*
ipv6h
);
mtu
=
dst_pmtu
(
dst
)
-
sizeof
(
*
ipv6h
);
if
(
opt
)
{
if
(
opt
)
{
mtu
-=
(
opt
->
opt_nflen
+
opt
->
opt_flen
);
max_headroom
+=
8
;
mtu
-=
8
;
}
}
if
(
mtu
<
IPV6_MIN_MTU
)
if
(
mtu
<
IPV6_MIN_MTU
)
mtu
=
IPV6_MIN_MTU
;
mtu
=
IPV6_MIN_MTU
;
...
@@ -764,41 +679,71 @@ int ip6ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev)
...
@@ -764,41 +679,71 @@ int ip6ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev)
icmpv6_send
(
skb
,
ICMPV6_PKT_TOOBIG
,
0
,
mtu
,
dev
);
icmpv6_send
(
skb
,
ICMPV6_PKT_TOOBIG
,
0
,
mtu
,
dev
);
goto
tx_err_dst_release
;
goto
tx_err_dst_release
;
}
}
err
=
ip6_append_data
(
sk
,
ip6ip6_getfrag
,
skb
->
nh
.
raw
,
skb
->
len
,
0
,
skb
->
h
.
raw
=
skb
->
nh
.
raw
;
t
->
parms
.
hop_limit
,
opt
,
&
fl
,
(
struct
rt6_info
*
)
dst
,
MSG_DONTWAIT
);
if
(
err
)
{
/*
ip6_flush_pending_frames
(
sk
);
* Okay, now see if we can stuff it in the buffer as-is.
}
else
{
*/
err
=
ip6_push_pending_frames
(
sk
);
max_headroom
+=
LL_RESERVED_SPACE
(
tdev
);
err
=
(
err
<
0
?
err
:
0
);
if
(
skb_headroom
(
skb
)
<
max_headroom
||
skb_cloned
(
skb
)
||
skb_shared
(
skb
))
{
struct
sk_buff
*
new_skb
;
if
(
!
(
new_skb
=
skb_realloc_headroom
(
skb
,
max_headroom
)))
goto
tx_err_dst_release
;
if
(
skb
->
sk
)
skb_set_owner_w
(
new_skb
,
skb
->
sk
);
kfree_skb
(
skb
);
skb
=
new_skb
;
}
}
if
(
!
err
)
{
dst_release
(
skb
->
dst
);
stats
->
tx_bytes
+=
skb
->
len
;
skb
->
dst
=
dst_clone
(
dst
);
if
(
opt
)
ipv6_push_nfrag_opts
(
skb
,
opt
,
&
proto
,
NULL
);
skb
->
nh
.
raw
=
skb_push
(
skb
,
sizeof
(
struct
ipv6hdr
));
ipv6h
=
skb
->
nh
.
ipv6h
;
*
(
u32
*
)
ipv6h
=
fl
.
fl6_flowlabel
|
htonl
(
0x60000000
);
ipv6h
->
payload_len
=
htons
(
skb
->
len
-
sizeof
(
struct
ipv6hdr
));
ipv6h
->
hop_limit
=
t
->
parms
.
hop_limit
;
ipv6h
->
nexthdr
=
proto
;
ipv6_addr_copy
(
&
ipv6h
->
saddr
,
&
fl
.
fl6_src
);
ipv6_addr_copy
(
&
ipv6h
->
daddr
,
&
fl
.
fl6_dst
);
#ifdef CONFIG_NETFILTER
nf_conntrack_put
(
skb
->
nfct
);
skb
->
nfct
=
NULL
;
#ifdef CONFIG_NETFILTER_DEBUG
skb
->
nf_debug
=
0
;
#endif
#endif
pkt_len
=
skb
->
len
;
err
=
NF_HOOK
(
PF_INET6
,
NF_IP6_LOCAL_OUT
,
skb
,
NULL
,
skb
->
dst
->
dev
,
dst_output
);
if
(
err
==
NET_XMIT_SUCCESS
||
err
==
NET_XMIT_CN
)
{
stats
->
tx_bytes
+=
pkt_len
;
stats
->
tx_packets
++
;
stats
->
tx_packets
++
;
}
else
{
}
else
{
stats
->
tx_errors
++
;
stats
->
tx_errors
++
;
stats
->
tx_aborted_errors
++
;
stats
->
tx_aborted_errors
++
;
}
}
if
(
opt
&&
opt
!=
orig_opt
)
ip6_tnl_dst_store
(
t
,
dst
);
sock_kfree_s
(
sk
,
opt
,
opt
->
tot_len
);
if
(
opt
)
kfree
(
opt
);
fl6_sock_release
(
fl_lbl
);
ip6_dst_store
(
sk
,
dst
,
&
np
->
daddr
);
ip6_xmit_unlock
();
kfree_skb
(
skb
);
t
->
recursion
--
;
t
->
recursion
--
;
return
0
;
return
0
;
tx_err_link_failure:
stats
->
tx_carrier_errors
++
;
dst_link_failure
(
skb
);
tx_err_dst_release:
tx_err_dst_release:
dst_release
(
dst
);
dst_release
(
dst
);
if
(
opt
&&
opt
!=
orig_opt
)
if
(
opt
)
sock_kfree_s
(
sk
,
opt
,
opt
->
tot_len
);
kfree
(
opt
);
tx_err_free_fl_lbl:
fl6_sock_release
(
fl_lbl
);
ip6_xmit_unlock
();
if
(
link_failure
)
dst_link_failure
(
skb
);
tx_err:
tx_err:
stats
->
tx_errors
++
;
stats
->
tx_errors
++
;
stats
->
tx_dropped
++
;
stats
->
tx_dropped
++
;
...
@@ -850,13 +795,12 @@ static void ip6ip6_tnl_link_config(struct ip6_tnl *t)
...
@@ -850,13 +795,12 @@ static void ip6ip6_tnl_link_config(struct ip6_tnl *t)
{
{
struct
net_device
*
dev
=
t
->
dev
;
struct
net_device
*
dev
=
t
->
dev
;
struct
ip6_tnl_parm
*
p
=
&
t
->
parms
;
struct
ip6_tnl_parm
*
p
=
&
t
->
parms
;
struct
flowi
*
fl
;
struct
flowi
*
fl
=
&
t
->
fl
;
memcpy
(
&
dev
->
dev_addr
,
&
p
->
laddr
,
sizeof
(
struct
in6_addr
));
memcpy
(
&
dev
->
dev_addr
,
&
p
->
laddr
,
sizeof
(
struct
in6_addr
));
memcpy
(
&
dev
->
broadcast
,
&
p
->
raddr
,
sizeof
(
struct
in6_addr
));
memcpy
(
&
dev
->
broadcast
,
&
p
->
raddr
,
sizeof
(
struct
in6_addr
));
/* Set up flowi template */
/* Set up flowi template */
fl
=
&
t
->
fl
;
ipv6_addr_copy
(
&
fl
->
fl6_src
,
&
p
->
laddr
);
ipv6_addr_copy
(
&
fl
->
fl6_src
,
&
p
->
laddr
);
ipv6_addr_copy
(
&
fl
->
fl6_dst
,
&
p
->
raddr
);
ipv6_addr_copy
(
&
fl
->
fl6_dst
,
&
p
->
raddr
);
fl
->
oif
=
p
->
link
;
fl
->
oif
=
p
->
link
;
...
@@ -881,10 +825,7 @@ static void ip6ip6_tnl_link_config(struct ip6_tnl *t)
...
@@ -881,10 +825,7 @@ static void ip6ip6_tnl_link_config(struct ip6_tnl *t)
if
(
rt
==
NULL
)
if
(
rt
==
NULL
)
return
;
return
;
/* as long as tunnels use the same socket for transmission,
if
(
rt
->
rt6i_dev
)
{
locally nested tunnels won't work */
if
(
rt
->
rt6i_dev
&&
rt
->
rt6i_dev
->
type
!=
ARPHRD_TUNNEL6
)
{
dev
->
iflink
=
rt
->
rt6i_dev
->
ifindex
;
dev
->
iflink
=
rt
->
rt6i_dev
->
ifindex
;
dev
->
hard_header_len
=
rt
->
rt6i_dev
->
hard_header_len
+
dev
->
hard_header_len
=
rt
->
rt6i_dev
->
hard_header_len
+
...
@@ -1139,7 +1080,7 @@ ip6ip6_tnl_dev_init(struct net_device *dev)
...
@@ -1139,7 +1080,7 @@ ip6ip6_tnl_dev_init(struct net_device *dev)
int
ip6ip6_fb_tnl_dev_init
(
struct
net_device
*
dev
)
int
ip6ip6_fb_tnl_dev_init
(
struct
net_device
*
dev
)
{
{
struct
ip6_tnl
*
t
=
dev
->
priv
;
struct
ip6_tnl
*
t
=
dev
->
priv
;
ip6ip6_tnl_dev_init_gen
(
dev
);
ip6ip6_tnl_dev_init_gen
(
dev
);
dev_hold
(
dev
);
dev_hold
(
dev
);
tnls_wc
[
0
]
=
t
;
tnls_wc
[
0
]
=
t
;
return
0
;
return
0
;
...
@@ -1159,61 +1100,28 @@ static struct inet6_protocol ip6ip6_protocol = {
...
@@ -1159,61 +1100,28 @@ static struct inet6_protocol ip6ip6_protocol = {
int
__init
ip6_tunnel_init
(
void
)
int
__init
ip6_tunnel_init
(
void
)
{
{
int
i
,
j
,
err
;
int
err
;
struct
sock
*
sk
;
struct
ipv6_pinfo
*
np
;
for
(
i
=
0
;
i
<
NR_CPUS
;
i
++
)
{
if
(
!
cpu_possible
(
i
))
continue
;
err
=
sock_create
(
PF_INET6
,
SOCK_RAW
,
IPPROTO_IPV6
,
&
__ip6_socket
[
i
]);
if
(
err
<
0
)
{
printk
(
KERN_ERR
"Failed to create the IPv6 tunnel socket "
"(err %d).
\n
"
,
err
);
goto
fail
;
}
sk
=
__ip6_socket
[
i
]
->
sk
;
sk
->
sk_allocation
=
GFP_ATOMIC
;
np
=
inet6_sk
(
sk
);
np
->
hop_limit
=
255
;
np
->
mc_loop
=
0
;
sk
->
sk_prot
->
unhash
(
sk
);
}
if
((
err
=
inet6_add_protocol
(
&
ip6ip6_protocol
,
IPPROTO_IPV6
))
<
0
)
{
if
((
err
=
inet6_add_protocol
(
&
ip6ip6_protocol
,
IPPROTO_IPV6
))
<
0
)
{
printk
(
KERN_ERR
"Failed to register IPv6 protocol
\n
"
);
printk
(
KERN_ERR
"Failed to register IPv6 protocol
\n
"
);
goto
fail
;
return
err
;
}
}
ip6ip6_fb_tnl_dev
=
alloc_netdev
(
sizeof
(
struct
ip6_tnl
),
"ip6tnl0"
,
ip6ip6_fb_tnl_dev
=
alloc_netdev
(
sizeof
(
struct
ip6_tnl
),
"ip6tnl0"
,
ip6ip6_tnl_dev_setup
);
ip6ip6_tnl_dev_setup
);
if
(
!
ip6ip6_fb_tnl_dev
)
{
if
(
!
ip6ip6_fb_tnl_dev
)
{
err
=
-
ENOMEM
;
err
=
-
ENOMEM
;
goto
tnl_
fail
;
goto
fail
;
}
}
ip6ip6_fb_tnl_dev
->
init
=
ip6ip6_fb_tnl_dev_init
;
ip6ip6_fb_tnl_dev
->
init
=
ip6ip6_fb_tnl_dev_init
;
if
((
err
=
register_netdev
(
ip6ip6_fb_tnl_dev
)))
{
if
((
err
=
register_netdev
(
ip6ip6_fb_tnl_dev
)))
{
kfree
(
ip6ip6_fb_tnl_dev
);
kfree
(
ip6ip6_fb_tnl_dev
);
goto
tnl_
fail
;
goto
fail
;
}
}
return
0
;
return
0
;
tnl_fail:
inet6_del_protocol
(
&
ip6ip6_protocol
,
IPPROTO_IPV6
);
fail:
fail:
for
(
j
=
0
;
j
<
i
;
j
++
)
{
inet6_del_protocol
(
&
ip6ip6_protocol
,
IPPROTO_IPV6
);
if
(
!
cpu_possible
(
j
))
continue
;
sock_release
(
__ip6_socket
[
j
]);
__ip6_socket
[
j
]
=
NULL
;
}
return
err
;
return
err
;
}
}
...
@@ -1223,18 +1131,8 @@ int __init ip6_tunnel_init(void)
...
@@ -1223,18 +1131,8 @@ int __init ip6_tunnel_init(void)
void
ip6_tunnel_cleanup
(
void
)
void
ip6_tunnel_cleanup
(
void
)
{
{
int
i
;
unregister_netdev
(
ip6ip6_fb_tnl_dev
);
unregister_netdev
(
ip6ip6_fb_tnl_dev
);
inet6_del_protocol
(
&
ip6ip6_protocol
,
IPPROTO_IPV6
);
inet6_del_protocol
(
&
ip6ip6_protocol
,
IPPROTO_IPV6
);
for
(
i
=
0
;
i
<
NR_CPUS
;
i
++
)
{
if
(
!
cpu_possible
(
i
))
continue
;
sock_release
(
__ip6_socket
[
i
]);
__ip6_socket
[
i
]
=
NULL
;
}
}
}
#ifdef MODULE
#ifdef MODULE
...
...
net/ipv6/ipv6_syms.c
View file @
35a9a271
...
@@ -45,3 +45,4 @@ EXPORT_SYMBOL(ipv6_ext_hdr);
...
@@ -45,3 +45,4 @@ EXPORT_SYMBOL(ipv6_ext_hdr);
EXPORT_SYMBOL
(
ip6_append_data
);
EXPORT_SYMBOL
(
ip6_append_data
);
EXPORT_SYMBOL
(
ip6_flush_pending_frames
);
EXPORT_SYMBOL
(
ip6_flush_pending_frames
);
EXPORT_SYMBOL
(
ip6_push_pending_frames
);
EXPORT_SYMBOL
(
ip6_push_pending_frames
);
EXPORT_SYMBOL
(
ipv6_push_nfrag_opts
);
Write
Preview
Markdown
is supported
0%
Try again
or
attach a new file
Attach a file
Cancel
You are about to add
0
people
to the discussion. Proceed with caution.
Finish editing this message first!
Cancel
Please
register
or
sign in
to comment