2
0
mirror of https://github.com/checkpoint-restore/criu synced 2025-08-22 18:07:57 +00:00
criu/sockets.c

1189 lines
23 KiB
C
Raw Normal View History

Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
#include <sys/socket.h>
#include <linux/netlink.h>
#include <linux/types.h>
#include <linux/net.h>
#include <linux/un.h>
#include <sys/types.h>
#include <sys/vfs.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <netinet/tcp.h>
#include <errno.h>
#include <unistd.h>
#include <limits.h>
#include <arpa/inet.h>
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
#include "types.h"
#include "libnetlink.h"
#include "sockets.h"
#include "unix_diag.h"
#include "image.h"
#include "crtools.h"
#include "util.h"
#include "inet_diag.h"
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
static char buf[4096];
#ifndef NETLINK_SOCK_DIAG
#define NETLINK_SOCK_DIAG NETLINK_INET_DIAG
#endif
#ifndef SOCK_DIAG_BY_FAMILY
#define SOCK_DIAG_BY_FAMILY 20
#endif
#ifndef SOCKFS_MAGIC
#define SOCKFS_MAGIC 0x534F434B
#endif
struct socket_desc {
unsigned int family;
unsigned int ino;
struct socket_desc *next;
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
};
struct unix_sk_desc {
struct socket_desc sd;
unsigned int type;
unsigned int state;
unsigned int peer_ino;
unsigned int rqlen;
unsigned int wqlen;
unsigned int namelen;
char *name;
unsigned int nr_icons;
unsigned int *icons;
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
};
struct unix_sk_listen_icon {
unsigned int peer_ino;
struct unix_sk_desc *sk_desc;
struct unix_sk_listen_icon *next;
};
struct unix_sk_listen {
unsigned int ino;
struct sockaddr_un addr;
unsigned int addrlen;
int type;
struct unix_sk_listen *next;
};
#define INET_ADDR_LEN 40
struct inet_sk_desc {
struct socket_desc sd;
unsigned int type;
unsigned int proto;
unsigned int src_port;
unsigned int state;
unsigned int rqlen;
unsigned int wqlen;
unsigned int src_addr[4];
};
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
#define SK_HASH_SIZE 32
#define SK_HASH_LINK(head, key, elem) \
do { \
(elem)->next = (head)[(key) % SK_HASH_SIZE]; \
(head)[(key) % SK_HASH_SIZE] = (elem); \
} while (0)
#define __gen_static_lookup_func(ret, name, head, _member, _type, _name)\
static ret *name(_type _name) { \
ret *d; \
for (d = head[_name % SK_HASH_SIZE]; d; d = d->next) { \
if (d->_member == _name) \
break; \
} \
return d; \
}
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
static struct socket_desc *sockets[SK_HASH_SIZE];
__gen_static_lookup_func(struct socket_desc, lookup_socket, sockets,
ino, int, ino);
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
static struct unix_sk_listen_icon *unix_listen_icons[SK_HASH_SIZE];
__gen_static_lookup_func(struct unix_sk_listen_icon,
lookup_unix_listen_icons,
unix_listen_icons,
peer_ino, unsigned int, ino);
static struct unix_sk_listen *unix_listen[SK_HASH_SIZE];
static struct unix_sk_listen *lookup_unix_listen(unsigned int ino, int type)
{
struct unix_sk_listen *l;
for (l = unix_listen[ino % SK_HASH_SIZE]; l != NULL; l = l->next)
if ((l->ino == ino) && (l->type == type))
return l;
return NULL;
}
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
static int sk_collect_one(int ino, int family, struct socket_desc *d)
{
d->ino = ino;
d->family = family;
SK_HASH_LINK(sockets, ino, d);
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
return 0;
}
static void show_one_inet(char *act, struct inet_sk_desc *sk)
{
char src_addr[INET_ADDR_LEN] = "<unknown>";
if (inet_ntop(AF_INET, (void *)sk->src_addr, src_addr,
INET_ADDR_LEN) == NULL) {
pr_err("Failed to translate address: %d\n", errno);
}
dprintk("\t%s: ino %d family %d type %d port %d "
"state %d src_addr %s\n",
act, sk->sd.ino, sk->sd.family, sk->type, sk->src_port,
sk->state, src_addr);
}
static void show_one_inet_img(char *act, struct inet_sk_entry *e)
{
char src_addr[INET_ADDR_LEN] = "<unknown>";
if (inet_ntop(AF_INET, (void *)e->src_addr, src_addr,
INET_ADDR_LEN) == NULL) {
pr_err("Failed to translate address: %d\n", errno);
}
dprintk("\t%s: fd %d family %d type %d proto %d port %d "
"state %d src_addr %s\n",
act, e->fd, e->family, e->type, e->proto, e->src_port,
e->state, src_addr);
}
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
static void show_one_unix(char *act, struct unix_sk_desc *sk)
{
dprintk("\t%s: ino %d type %d state %d name %s\n",
act, sk->sd.ino, sk->type, sk->state, sk->name);
if (sk->nr_icons) {
int i;
for (i = 0; i < sk->nr_icons; i++)
dprintk("\t\ticon: %4d\n", sk->icons[i]);
}
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
}
static void show_one_unix_img(char *act, struct unix_sk_entry *e)
{
dprintk("\t%s: fd %d type %d state %d name %d bytes\n",
act, e->fd, e->type, e->state, e->namelen);
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
}
static int can_dump_inet_sk(struct inet_sk_desc *sk)
{
if (sk->sd.family != AF_INET) {
pr_err("Only IPv4 sockets for now\n");
return 0;
}
if (sk->type != SOCK_STREAM) {
pr_err("Only stream inet sockets for now\n");
return 0;
}
switch (sk->state) {
case TCP_LISTEN:
if (sk->rqlen != 0) {
/*
* Currently the ICONS nla reports the conn
* requests for listen sockets. Need to pick
* those up and fix the connect job respectively
*/
pr_err("In-flight connection (l)\n");
return 0;
}
break;
default:
pr_err("Unknown state %d\n", sk->state);
return 0;
}
return 1;
}
static int dump_one_inet(struct socket_desc *_sk, int fd,
struct cr_fdset *cr_fdset)
{
struct inet_sk_desc *sk = (struct inet_sk_desc *)_sk;
struct inet_sk_entry ie;
if (!can_dump_inet_sk(sk))
goto err;
memset(&ie, 0, sizeof(ie));
ie.fd = fd;
ie.id = sk->sd.ino;
ie.family = sk->sd.family;
ie.type = sk->type;
ie.proto = sk->proto;
ie.state = sk->state;
ie.src_port = sk->src_port;
ie.backlog = sk->wqlen;
memcpy(ie.src_addr, sk->src_addr, sizeof(u32) * 4);
if (write_img(cr_fdset->fds[CR_FD_INETSK], &ie))
goto err;
pr_info("Dumping inet socket at %d\n", fd);
show_one_inet("Dumping", sk);
show_one_inet_img("Dumped", &ie);
return 0;
err:
return -1;
}
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
static int can_dump_unix_sk(struct unix_sk_desc *sk)
{
if (sk->type != SOCK_STREAM &&
sk->type != SOCK_DGRAM) {
pr_err("Only stream/dgram sockets for now\n");
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
return 0;
}
switch (sk->state) {
case TCP_LISTEN:
break;
case TCP_ESTABLISHED:
if (sk->rqlen) {
/*
* The hard case :( Currentl there's no way to
* clone the sk queue. Even the MSG_PEEK doesn't
* help, since it picks up the head of the queue
* always. Some more patches should go
*/
pr_err("Non empty queue\n");
return 0;
}
break;
case TCP_CLOSE:
if (sk->type != SOCK_DGRAM)
return 0;
break;
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
default:
pr_err("Unknown state %d\n", sk->state);
return 0;
}
return 1;
}
static int dump_one_unix(struct socket_desc *_sk, int fd,
struct cr_fdset *cr_fdset)
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
{
struct unix_sk_desc *sk = (struct unix_sk_desc *)_sk;
struct unix_sk_entry ue;
if (!can_dump_unix_sk(sk))
goto err;
ue.fd = fd;
ue.id = sk->sd.ino;
ue.type = sk->type;
ue.state = sk->state;
ue.namelen = sk->namelen;
ue.backlog = sk->wqlen;
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
ue.flags = 0;
ue.peer = sk->peer_ino;
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
/*
* If this is in-flight connection we need to figure
* out where to connect it on restore. Thus, tune up peer
* id by searching an existing listening socket.
*
* Note the socket name will be found at restore stage,
* not now, just to reduce size of dump files.
*/
if (!ue.peer && ue.state == TCP_ESTABLISHED) {
struct unix_sk_listen_icon *e;
e = lookup_unix_listen_icons(ue.id);
if (!e) {
pr_err("Dangling in-flight connection %d\n", ue.id);
goto err;
}
/* e->sk_desc is _never_ NULL */
if (e->sk_desc->state != TCP_LISTEN) {
pr_err("In-flight connection on "
"non-listening socket %d\n", ue.id);
goto err;
}
ue.flags |= USK_INFLIGHT;
ue.peer = e->sk_desc->sd.ino;
dprintk("\t\tFixed inflight socket %d peer %d)\n",
ue.id, ue.peer);
}
if (write_img(cr_fdset->fds[CR_FD_UNIXSK], &ue))
goto err;
if (write_img_buf(cr_fdset->fds[CR_FD_UNIXSK], sk->name, ue.namelen))
goto err;
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
pr_info("Dumping unix socket at %d\n", fd);
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
show_one_unix("Dumping", sk);
show_one_unix_img("Dumped", &ue);
return 0;
err:
return -1;
}
int try_dump_socket(pid_t pid, int fd, struct cr_fdset *cr_fdset)
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
{
struct socket_desc *sk;
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
struct statfs fst;
struct stat st;
char path[64];
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
/*
* Sockets are tricky, we can't open it but can
* do stats over and check for sokets magic.
*/
snprintf(buf, sizeof(buf), "/proc/%d/fd/%d", pid, fd);
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
if (statfs(buf, &fst)) {
pr_err("Can't statfs %s\n", buf);
return -1;
}
if (stat(buf, &st)) {
pr_err("Can't stat %s\n", buf);
return -1;
}
if (fst.f_type != SOCKFS_MAGIC)
return 1; /* not a socket, proceed with caller error */
sk = lookup_socket(st.st_ino);
if (!sk) {
pr_err("Uncollected socket %ld\n", st.st_ino);
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
return -1;
}
switch (sk->family) {
case AF_UNIX:
return dump_one_unix(sk, fd, cr_fdset);
case AF_INET:
return dump_one_inet(sk, fd, cr_fdset);
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
default:
pr_err("BUG! Unknown socket collected\n");
break;
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
}
return -1;
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
}
static int inet_tcp_collect_one(struct inet_diag_msg *m, struct rtattr **tb)
{
struct inet_sk_desc *d;
d = xzalloc(sizeof(*d));
if (!d)
return -1;
d->type = SOCK_STREAM;
d->proto = IPPROTO_TCP;
d->src_port = ntohs(m->id.idiag_sport);
d->state = m->idiag_state;
d->rqlen = m->idiag_rqueue;
d->wqlen = m->idiag_wqueue;
memcpy(d->src_addr, m->id.idiag_src, sizeof(u32) * 4);
return sk_collect_one(m->idiag_inode, AF_INET, &d->sd);
}
static int inet_tcp_receive_one(struct nlmsghdr *h)
{
struct inet_diag_msg *m = NLMSG_DATA(h);
struct rtattr *tb[INET_DIAG_MAX+1];
parse_rtattr(tb, INET_DIAG_MAX, (struct rtattr *)(m + 1),
h->nlmsg_len - NLMSG_LENGTH(sizeof(*m)));
return inet_tcp_collect_one(m, tb);
}
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
static int unix_collect_one(struct unix_diag_msg *m, struct rtattr **tb)
{
struct unix_sk_desc *d, **h;
d = xzalloc(sizeof(*d));
if (!d)
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
return -1;
d->type = m->udiag_type;
d->state= m->udiag_state;
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
if (tb[UNIX_DIAG_PEER])
d->peer_ino = *(int *)RTA_DATA(tb[UNIX_DIAG_PEER]);
if (tb[UNIX_DIAG_NAME]) {
int len = RTA_PAYLOAD(tb[UNIX_DIAG_NAME]);
char *name = xmalloc(len + 1);
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
if (!name)
goto err;
memcpy(name, RTA_DATA(tb[UNIX_DIAG_NAME]), len);
name[len] = '\0';
if (name[0] != '\0' && d->state == TCP_LISTEN) {
struct unix_diag_vfs *uv;
struct stat st;
if (name[0] != '/') {
pr_warning("Relative bind path '%s' "
"unsupported\n", name);
xfree(name);
xfree(d);
return 0;
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
}
if (!tb[UNIX_DIAG_VFS]) {
pr_err("Bound socket w/o inode %d\n",
m->udiag_ino);
goto err;
}
uv = RTA_DATA(tb[UNIX_DIAG_VFS]);
if (stat(name, &st)) {
pr_perror("Can't stat socket %d(%s)",
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
m->udiag_ino, name);
goto err;
}
if ((st.st_ino != uv->udiag_vfs_ino) ||
(st.st_dev != kdev_to_odev(uv->udiag_vfs_dev))) {
pr_info("unix: Dropping path for "
"unlinked bound "
"sk %x.%x real %x.%x\n",
(int)st.st_dev,
(int)st.st_ino,
(int)uv->udiag_vfs_dev,
(int)uv->udiag_vfs_ino);
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
/*
* When a listen socket is bound to
* unlinked file, we just drop his name,
* since noone will access it via one.
*/
xfree(name);
len = 0;
name = NULL;
}
}
d->namelen = len;
d->name = name;
}
if (tb[UNIX_DIAG_ICONS]) {
int len = RTA_PAYLOAD(tb[UNIX_DIAG_ICONS]);
int i;
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
d->icons = xmalloc(len);
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
if (!d->icons)
goto err;
memcpy(d->icons, RTA_DATA(tb[UNIX_DIAG_ICONS]), len);
d->nr_icons = len / sizeof(u32);
/*
* Remember these sockets, we will need them
* to fix up in-flight sockets peers.
*/
for (i = 0; i < d->nr_icons; i++) {
struct unix_sk_listen_icon *e;
int n;
e = xzalloc(sizeof(*e));
if (!e)
goto err;
SK_HASH_LINK(unix_listen_icons, d->icons[i], e);
dprintk("\t\tCollected icon %d\n", d->icons[i]);
e->peer_ino = d->icons[i];
e->sk_desc = d;
}
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
}
if (tb[UNIX_DIAG_RQLEN]) {
struct unix_diag_rqlen *rq;
rq = (struct unix_diag_rqlen *)RTA_DATA(tb[UNIX_DIAG_RQLEN]);
d->rqlen = rq->udiag_rqueue;
d->wqlen = rq->udiag_wqueue;
}
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
sk_collect_one(m->udiag_ino, AF_UNIX, &d->sd);
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
show_one_unix("Collected", d);
return 0;
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
err:
xfree(d->icons);
xfree(d->name);
xfree(d);
return -1;
}
static int unix_receive_one(struct nlmsghdr *h)
{
struct unix_diag_msg *m = NLMSG_DATA(h);
struct rtattr *tb[UNIX_DIAG_MAX+1];
parse_rtattr(tb, UNIX_DIAG_MAX, (struct rtattr *)(m + 1),
h->nlmsg_len - NLMSG_LENGTH(sizeof(*m)));
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
return unix_collect_one(m, tb);
}
static int collect_sockets_nl(int nl, void *req, int size,
int (*receive_callback)(struct nlmsghdr *h))
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
{
struct msghdr msg;
struct sockaddr_nl nladdr;
struct iovec iov;
memset(&msg, 0, sizeof(msg));
msg.msg_name = &nladdr;
msg.msg_namelen = sizeof(nladdr);
msg.msg_iov = &iov;
msg.msg_iovlen = 1;
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
memset(&nladdr, 0, sizeof(nladdr));
nladdr.nl_family= AF_NETLINK;
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
iov.iov_base = req;
iov.iov_len = size;
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
if (sendmsg(nl, &msg, 0) < 0) {
pr_perror("Can't send request message");
goto err;
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
}
iov.iov_base = buf;
iov.iov_len = sizeof(buf);
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
while (1) {
int err;
memset(&msg, 0, sizeof(msg));
msg.msg_name = &nladdr;
msg.msg_namelen = sizeof(nladdr);
msg.msg_iov = &iov;
msg.msg_iovlen = 1;
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
err = recvmsg(nl, &msg, 0);
if (err < 0) {
if (errno == EINTR)
continue;
else {
pr_perror("Error receiving nl report");
goto err;
}
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
}
if (err == 0)
break;
err = nlmsg_receive(buf, err, receive_callback);
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
if (err < 0)
goto err;
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
if (err == 0)
break;
}
return 0;
err:
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
return -1;
}
int collect_sockets(void)
{
int err = 0, tmp;
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
int nl;
int supp_type = 0;
struct {
struct nlmsghdr hdr;
union {
struct unix_diag_req u;
struct inet_diag_req_v2 i;
} r;
} req;
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
nl = socket(PF_NETLINK, SOCK_RAW, NETLINK_SOCK_DIAG);
if (nl < 0) {
pr_err("Can't create sock diag socket\n");
return -1;
}
memset(&req, 0, sizeof(req));
req.hdr.nlmsg_len = sizeof(req);
req.hdr.nlmsg_type = SOCK_DIAG_BY_FAMILY;
req.hdr.nlmsg_flags = NLM_F_DUMP | NLM_F_REQUEST;
req.hdr.nlmsg_seq = CR_NLMSG_SEQ;
/* Collect UNIX sockets */
req.r.u.sdiag_family = AF_UNIX;
req.r.u.udiag_states = -1; /* All */
req.r.u.udiag_show = UDIAG_SHOW_NAME | UDIAG_SHOW_VFS |
UDIAG_SHOW_PEER | UDIAG_SHOW_ICONS |
UDIAG_SHOW_RQLEN;
tmp = collect_sockets_nl(nl, &req, sizeof(req), unix_receive_one);
if (tmp)
err = tmp;
/* Collect IPv4 TCP sockets */
req.r.i.sdiag_family = AF_INET;
req.r.i.sdiag_protocol = IPPROTO_TCP;
req.r.i.idiag_ext = 0;
/* Only listening sockets supported yet */
req.r.i.idiag_states = 1 << TCP_LISTEN;
tmp = collect_sockets_nl(nl, &req, sizeof(req), inet_tcp_receive_one);
if (tmp)
err = tmp;
out:
close(nl);
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
return err;
}
static void prep_conn_addr(int id, struct sockaddr_un *addr, int *addrlen)
{
addr->sun_family = AF_UNIX;
addr->sun_path[0] = '\0';
snprintf(addr->sun_path + 1, UNIX_PATH_MAX - 1, "crtools-sk-%10d", id);
*addrlen = sizeof(addr->sun_family) + sizeof("crtools-sk-") + 10;
}
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
struct unix_conn_job {
int fd;
unsigned int peer;
int type;
struct unix_conn_job *next;
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
};
enum {
CJ_DGRAM,
CJ_STREAM,
CJ_STREAM_INFLIGHT,
};
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
static void unix_show_job(char *type, int fd, int id)
{
dprintk("%s job fd %d id %d\n", type, fd, id);
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
}
static struct unix_conn_job *conn_jobs;
static int schedule_conn_job(int type, struct unix_sk_entry *ue)
{
struct unix_conn_job *cj;
cj = xmalloc(sizeof(*cj));
if (!cj)
return -1;
cj->type = type;
cj->peer = ue->peer;
cj->fd = ue->fd;
cj->next = conn_jobs;
conn_jobs = cj;
unix_show_job("Sched conn", ue->fd, ue->peer);
return 0;
}
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
static int run_connect_jobs(void)
{
struct unix_conn_job *cj, *next;
int i;
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
cj = conn_jobs;
while (cj) {
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
int attempts = 8;
struct sockaddr_un addr;
int addrlen;
/*
* Might need to resolve in-flight connection name.
*/
if (cj->type == CJ_STREAM)
prep_conn_addr(cj->peer, &addr, &addrlen);
else {
struct unix_sk_listen *e;
if (cj->type == CJ_STREAM_INFLIGHT)
e = lookup_unix_listen(cj->peer, SOCK_STREAM);
else /* if (cj->type == CJ_DGRAM) */
e = lookup_unix_listen(cj->peer, SOCK_DGRAM);
if (!e) {
pr_err("Bad in-flight socket peer %d\n",
cj->peer);
return -1;
}
memcpy(&addr, &e->addr, sizeof(addr));
addrlen = e->addrlen;
}
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
unix_show_job("Run conn", cj->fd, -1);
try_again:
if (connect(cj->fd, (struct sockaddr *)&addr, addrlen) < 0) {
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
if (attempts) {
usleep(1000);
attempts--;
goto try_again; /* FIXME use avagin@ waiters */
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
}
pr_perror("Can't restore connection (c)");
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
return -1;
}
unix_show_job("Fin conn", cj->fd, -1);
next = cj->next;
xfree(cj);
cj = next;
}
/*
* Free collected listening sockets,
* we don't need them anymore.
*/
for (i = 0; i < SK_HASH_SIZE; i++) {
struct unix_sk_listen *h = unix_listen[i];
struct unix_sk_listen *e;
while (h) {
e = h->next;
xfree(h);
h = e;
}
}
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
return 0;
}
struct unix_accept_job {
int fd;
struct unix_accept_job *next;
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
};
static struct unix_accept_job *accept_jobs;
static int schedule_acc_job(int sk, struct unix_sk_entry *ue)
{
struct sockaddr_un addr;
int len;
struct unix_accept_job *aj;
prep_conn_addr(ue->id, &addr, &len);
if (bind(sk, (struct sockaddr *)&addr, len) < 0) {
pr_perror("Can't bind socket");
goto err;
}
if (listen(sk, 1) < 0) {
pr_perror("Can't listen socket");
goto err;
}
aj = xmalloc(sizeof(*aj));
if (aj == NULL)
goto err;
aj->fd = ue->fd;
aj->next = accept_jobs;
accept_jobs = aj;
unix_show_job("Sched acc", ue->fd, ue->id);
return 0;
err:
return -1;
}
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
static int run_accept_jobs(void)
{
struct unix_accept_job *aj, *next;
aj = accept_jobs;
while (aj) {
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
int fd;
unix_show_job("Run acc", aj->fd, -1);
fd = accept(aj->fd, NULL, NULL);
if (fd < 0) {
pr_perror("Can't restore connection (s)");
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
return -1;
}
if (reopen_fd_as_nocheck(aj->fd, fd))
return -1;
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
unix_show_job("Fin acc", aj->fd, -1);
next = aj->next;
xfree(aj);
aj = next;
}
return 0;
}
static int bind_unix_sk_to_addr(int sk, struct sockaddr_un *addr, int addrlen,
int id, int type)
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
{
struct unix_sk_listen *e;
if (bind(sk, (struct sockaddr *)addr, addrlen) < 0) {
pr_perror("Can't bind socket");
goto err;
}
/*
* Just remember it and connect() if needed.
*/
e = xmalloc(sizeof(*e));
if (!e)
goto err;
memcpy(&e->addr, addr, sizeof(e->addr));
e->addrlen = addrlen;
e->ino = id;
e->type = type;
SK_HASH_LINK(unix_listen, e->ino, e);
return 0;
err:
return -1;
}
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
static int bind_unix_sk(int sk, struct unix_sk_entry *ue, int img_fd)
{
struct sockaddr_un addr;
if (!ue->namelen || ue->namelen >= UNIX_PATH_MAX) {
pr_err("Bad unix name len %d\n", ue->namelen);
return -1;
}
memset(&addr, 0, sizeof(addr));
addr.sun_family = AF_UNIX;
if (read_img_buf(img_fd, &addr.sun_path, ue->namelen) < 0)
return -1;
if (addr.sun_path[0] != '\0')
unlink(addr.sun_path);
return bind_unix_sk_to_addr(sk, &addr,
sizeof(addr.sun_family) + ue->namelen,
ue->id, ue->type);
}
static int open_unix_sk_dgram(int sk, struct unix_sk_entry *ue, int img_fd)
{
int ret = 0;
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
if (ue->namelen)
ret = bind_unix_sk(sk, ue, img_fd);
else if (ue->peer) {
struct sockaddr_un addr;
int addrlen;
/*
* dgram socket without name, but with peer
* this is only possible for those created
* by socketpair call
*/
prep_conn_addr(ue->id, &addr, &addrlen);
ret = bind_unix_sk_to_addr(sk, &addr, addrlen, ue->id,
SOCK_DGRAM);
}
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
if (!ret && ue->peer)
ret = schedule_conn_job(CJ_DGRAM, ue);
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
return ret;
}
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
static int open_unix_sk_stream(int sk, struct unix_sk_entry *ue, int img_fd)
{
int ret;
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
if (ue->state == TCP_LISTEN) {
ret = bind_unix_sk(sk, ue, img_fd);
if (ret < 0)
goto out;
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
ret = listen(sk, ue->backlog);
if (ret < 0) {
pr_perror("Can't listen socket");
goto out;
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
}
} else if (ue->state == TCP_ESTABLISHED) {
/*
* If a connection is established we need
* two separate steps -- one peer become
* a server and do bind()/listen(), then
* it deferred to accept() later, while
* another peer become a client and
* deferred to connect() later.
*/
if (ue->peer < ue->id && !(ue->flags & USK_INFLIGHT))
ret = schedule_acc_job(sk, ue);
else
ret = schedule_conn_job((ue->flags & USK_INFLIGHT) ?
CJ_STREAM_INFLIGHT : CJ_STREAM, ue);
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
} else {
pr_err("Unknown state %d\n", ue->state);
ret = -1;
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
}
out:
return ret;
}
static int open_unix_sk(struct unix_sk_entry *ue, int *img_fd)
{
int sk;
show_one_unix_img("Restore", ue);
sk = socket(PF_UNIX, ue->type, 0);
if (sk < 0) {
pr_perror("Can't create unix socket");
return -1;
}
switch (ue->type) {
case SOCK_STREAM:
if (open_unix_sk_stream(sk, ue, *img_fd))
goto err;
break;
case SOCK_DGRAM:
if (open_unix_sk_dgram(sk, ue, *img_fd))
goto err;
break;
default:
pr_err("Unsupported socket type: %d\n", ue->type);
goto err;
}
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
if (move_img_fd(img_fd, ue->fd))
return -1;
return reopen_fd_as(ue->fd, sk);
err:
close(sk);
return -1;
}
static int prepare_unix_sockets(int pid)
{
int usk_fd, ret = -1;
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
usk_fd = open_image_ro(CR_FD_UNIXSK, pid);
if (usk_fd < 0)
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
return -1;
while (1) {
struct unix_sk_entry ue;
ret = read_img_eof(usk_fd, &ue);
if (ret <= 0)
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
break;
ret = open_unix_sk(&ue, &usk_fd);
if (ret)
break;
}
err:
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
close(usk_fd);
if (!ret)
ret = run_connect_jobs();
if (!ret)
ret = run_accept_jobs();
return ret;
}
static int open_inet_sk(struct inet_sk_entry *ie, int *img_fd)
{
int sk;
struct sockaddr_in addr;
show_one_inet_img("Restore", ie);
if (ie->family != AF_INET) {
pr_err("Unsupported socket family: %d\n", ie->family);
return -1;
}
if (ie->type != SOCK_STREAM) {
pr_err("Unsupported socket type: %d\n", ie->type);
return -1;
}
sk = socket(ie->family, ie->type, ie->proto);
if (sk < 0) {
pr_perror("Can't create unix socket");
return -1;
}
/*
* Listen sockets are easiest ones -- simply
* bind() and listen(), and that's all.
*/
memset(&addr, 0, sizeof(addr));
addr.sin_family = ie->family;
addr.sin_port = htons(ie->src_port);
memcpy(&addr.sin_addr.s_addr, ie->src_addr, sizeof(unsigned int) * 4);
if (bind(sk, (struct sockaddr *) &addr, sizeof(addr)) == -1) {
pr_perror("Inet socket bind failed");
goto err;
}
if (listen(sk, ie->backlog) == -1) {
pr_perror("listen() failed on %d", sk);
goto err;
}
if (move_img_fd(img_fd, ie->fd))
return -1;
return reopen_fd_as(ie->fd, sk);
err:
close(sk);
return -1;
}
static int prepare_inet_sockets(int pid)
{
int isk_fd, ret = -1;
isk_fd = open_image_ro(CR_FD_INETSK, pid);
if (isk_fd < 0)
return -1;
while (1) {
struct inet_sk_entry ie;
ret = read_img_eof(isk_fd, &ie);
if (ret <= 0)
break;
ret = open_inet_sk(&ie, &isk_fd);
if (ret)
break;
}
err:
close(isk_fd);
return ret;
}
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
int prepare_sockets(int pid)
{
int err;
pr_info("%d: Opening sockets\n", pid);
err = prepare_unix_sockets(pid);
if (err)
return err;
return prepare_inet_sockets(pid);
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
}
void show_inetsk(int fd)
{
struct inet_sk_entry ie;
int ret = 0;
pr_img_head(CR_FD_INETSK);
while (1) {
char src_addr[INET_ADDR_LEN] = "<unknown>";
ret = read_img_eof(fd, &ie);
if (ret <= 0)
goto out;
if (inet_ntop(AF_INET, (void *)ie.src_addr, src_addr,
INET_ADDR_LEN) == NULL) {
pr_err("Failed to translate address: %d\n", errno);
}
pr_info("fd %d family %d type %d proto %d port %d state %d "
"--> %s\n", ie.fd, ie.family, ie.type, ie.proto,
ie.src_port, ie.state, src_addr);
}
out:
if (ret)
pr_info("\n");
pr_img_tail(CR_FD_INETSK);
}
void show_unixsk(int fd)
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
{
struct unix_sk_entry ue;
int ret = 0;
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
pr_img_head(CR_FD_UNIXSK);
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
while (1) {
ret = read_img_eof(fd, &ue);
if (ret <= 0)
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
goto out;
pr_info("fd %4d type %2d state %2d namelen %4d backlog %4d "
"id %6d peer %6d",
ue.fd, ue.type, ue.state, ue.namelen, ue.namelen,
ue.id, ue.peer);
if (ue.namelen) {
ret = read_img_buf(fd, buf, ue.namelen);
if (ret < 0) {
pr_info("\n");
goto out;
}
if (!buf[0])
buf[0] = '@';
pr_info(" --> %s\n", buf);
} else
pr_info("\n");
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
}
out:
pr_img_tail(CR_FD_UNIXSK);
Unix sockets initial support Currently it can only work with stream sockets, which have no skbs in queues (listening or established -- both work OK). The cpt part uses the sock_diag engine that was merged to Dave recently to collect sockets. Then it dumps sockets by checking the filesystem ID of a failed-to-open through /proc/pid/fd descriptors (sockets do not allow for such tricks with opens through proc) against SOCKFS_TYPE. The rst part is more tricky. Listen sockets are just restored, this is simple. Connected sockets are restored like this: 1. One end establishes a listening anon socket at the desired descriptor; 2. The other end just creates a socket at the desired descriptor; 3. All sockets, that are to be connect()-ed call connect. Unix sockets do not block connect() till the accept() time and thus we continue with... 4. ... all listening sockets call accept() and ... dup2 the new fd into the accepting end. There's a problem with this approach -- socket names are not preserved, but looking into our OpenVZ implementation I think this is OK for existing apps. What should be done next is: 1. Need to merge the file IDs patches in our tree and make Andrey to support files sharing. This will solve the sk = socket(); fork(); case. Currently it simply doesn't work :( 2. Need to add support for DGRAM sockets -- I wrote comment how to do it in the can_dump_unix_sk() 3. Need to add support for in-flight connections 4. Implement support for UDP sockets (quite simple) 5. Implement support for listening TCP sockets (also not very complex) 6. Implement support for connected TCP scokets (hard one, Tejun's patches are not very good for this from my POV) Cyrill, plz, apply this patch and put the above descriptions onto wiki docs (do we have the plans page yet?). Andrey, plz, take care of unix sockets tests in zdtm. Most likely it won't work till you do the shared files support for sockets. Signed-off-by: Pavel Emelyanov <xemul@parallels.com> Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
2011-12-26 22:12:03 +04:00
}