Skip to contentSkip to conversation

packet: Sock-diag extension for packet sockets (beginning)

PATCH In review Pavel Emelyanov 2 participants 6 messages Aug 13, 2012 – Aug 14, 2012

Subsystems (3)
Networking Network drivers Packet sockets
lore
Share & tools
lore

Pavel Emelyanov

author
#0 lore raw

cover packet: Sock-diag extension for packet sockets (beginning)

There's a set of stuff residing on a AF_PACKET socket which is currently write-only -- rings, copy_thresh and mclist. But we need to know these while doing the checkpoint-restore.

As a solution for the similar problem with unix sockets, the sock-diag engine was developed, so here's the basic implementation of the packet sockets extension. It doesn't report rings, fanout and stats, but it can be easily added later.

I'd like to mention, that it's not a strict requirement, that the diag is used for getting info about sockets, it would be perfectly fine just to fix the packet getsockopt callback to return the desired info. So, if the diag for packet sockets is for any reason unacceptable, just let me know.

Thanks,
Pavel

author
#1 lore raw

PATCH 1/4 packet: Introduce net/packet/internal.h header

The diag module will need to access some private packet_sock data, so move it to a header in advance. This file will be shared between the af_packet.c and the diag.c

Signed-off-by Pavel Emelyanov <[email protected]>
---
 net/packet/af_packet.c |   99 +---------------------------------------------
 net/packet/internal.h  |  103 ++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 105 insertions(+), 97 deletions(-)
 create mode 100644 net/packet/internal.h
net/packet/af_packet.c +2 -97
index ceaca7c..8a1605a 100644
@@ -93,6 +93,8 @@
 #include <net/inet_common.h>
 #endif
 
+#include "internal.h"
+
 /*
    Assumptions:
    - if device has no dev->hard_header routine, it adds and removes ll header

@@ -146,14 +148,6 @@ dev->hard_header == NULL (ll header is added by device, we cannot control it)
 
 /* Private packet socket structures. */
 
-struct packet_mclist {
-	struct packet_mclist	*next;
-	int			ifindex;
-	int			count;
-	unsigned short		type;
-	unsigned short		alen;
-	unsigned char		addr[MAX_ADDR_LEN];
-};
 /* identical to struct packet_mreq except it has
  * a longer address field.
  */

@@ -175,63 +169,7 @@ static int packet_set_ring(struct sock *sk, union tpacket_req_u *req_u,
 #define BLK_PLUS_PRIV(sz_of_priv) \
 	(BLK_HDR_LEN + ALIGN((sz_of_priv), V3_ALIGNMENT))
 
-/* kbdq - kernel block descriptor queue */
-struct tpacket_kbdq_core {
-	struct pgv	*pkbdq;
-	unsigned int	feature_req_word;
-	unsigned int	hdrlen;
-	unsigned char	reset_pending_on_curr_blk;
-	unsigned char   delete_blk_timer;
-	unsigned short	kactive_blk_num;
-	unsigned short	blk_sizeof_priv;
-
-	/* last_kactive_blk_num:
-	 * trick to see if user-space has caught up
-	 * in order to avoid refreshing timer when every single pkt arrives.
-	 */
-	unsigned short	last_kactive_blk_num;
-
-	char		*pkblk_start;
-	char		*pkblk_end;
-	int		kblk_size;
-	unsigned int	knum_blocks;
-	uint64_t	knxt_seq_num;
-	char		*prev;
-	char		*nxt_offset;
-	struct sk_buff	*skb;
-
-	atomic_t	blk_fill_in_prog;
-
-	/* Default is set to 8ms */
-#define DEFAULT_PRB_RETIRE_TOV	(8)
-
-	unsigned short  retire_blk_tov;
-	unsigned short  version;
-	unsigned long	tov_in_jiffies;
-
-	/* timer to retire an outstanding block */
-	struct timer_list retire_blk_timer;
-};
-
 #define PGV_FROM_VMALLOC 1
-struct pgv {
-	char *buffer;
-};
-
-struct packet_ring_buffer {
-	struct pgv		*pg_vec;
-	unsigned int		head;
-	unsigned int		frames_per_block;
-	unsigned int		frame_size;
-	unsigned int		frame_max;
-
-	unsigned int		pg_vec_order;
-	unsigned int		pg_vec_pages;
-	unsigned int		pg_vec_len;
-
-	struct tpacket_kbdq_core	prb_bdqc;
-	atomic_t		pending;
-};
 
 #define BLOCK_STATUS(x)	((x)->hdr.bh1.block_status)
 #define BLOCK_NUM_PKTS(x)	((x)->hdr.bh1.num_pkts)

@@ -269,34 +207,6 @@ static void prb_fill_vlan_info(struct tpacket_kbdq_core *,
 		struct tpacket3_hdr *);
 static void packet_flush_mclist(struct sock *sk);
 
-struct packet_fanout;
-struct packet_sock {
-	/* struct sock has to be the first member of packet_sock */
-	struct sock		sk;
-	struct packet_fanout	*fanout;
-	struct tpacket_stats	stats;
-	union  tpacket_stats_u	stats_u;
-	struct packet_ring_buffer	rx_ring;
-	struct packet_ring_buffer	tx_ring;
-	int			copy_thresh;
-	spinlock_t		bind_lock;
-	struct mutex		pg_vec_lock;
-	unsigned int		running:1,	/* prot_hook is attached*/
-				auxdata:1,
-				origdev:1,
-				has_vnet_hdr:1;
-	int			ifindex;	/* bound device		*/
-	__be16			num;
-	struct packet_mclist	*mclist;
-	atomic_t		mapped;
-	enum tpacket_versions	tp_version;
-	unsigned int		tp_hdrlen;
-	unsigned int		tp_reserve;
-	unsigned int		tp_loss:1;
-	unsigned int		tp_tstamp;
-	struct packet_type	prot_hook ____cacheline_aligned_in_smp;
-};
-
 #define PACKET_FANOUT_MAX	256
 
 struct packet_fanout {

@@ -334,11 +244,6 @@ struct packet_skb_cb {
 	(((x)->kactive_blk_num < ((x)->knum_blocks-1)) ? \
 	((x)->kactive_blk_num+1) : 0)
 
-static struct packet_sock *pkt_sk(struct sock *sk)
-{
-	return (struct packet_sock *)sk;
-}
-
 static void __fanout_unlink(struct sock *sk, struct packet_sock *po);
 static void __fanout_link(struct sock *sk, struct packet_sock *po);
net/packet/internal.h new file +103 -0

103 lines not on this page · raw message

Signature · 1 line
1.7.6.5
author
#2 lore raw

PATCH 2/4 packet: Diag core and basic socket info dumping

In reply to Pavel Emelyanov

The diag module can be built independently from the af_packet.ko one, just like it's done in unix sockets.

The core dumping message carries the info available at socket creation time, i.e. family, type and protocol (in the same byte order as shown in the proc file).

The socket inode number and cookie is reserved for future per-socket info retrieving. The per-protocol filtering is also reserved for future by requiring the sdiag_protocol to be zero.

Signed-off-by Pavel Emelyanov <[email protected]>
---
 include/linux/Kbuild        |    1 +
 include/linux/packet_diag.h |   24 ++++++++++
 net/packet/Kconfig          |    8 +++
 net/packet/Makefile         |    2 +
 net/packet/diag.c           |  104 +++++++++++++++++++++++++++++++++++++++++++
 5 files changed, 139 insertions(+), 0 deletions(-)
 create mode 100644 include/linux/packet_diag.h
 create mode 100644 net/packet/diag.c
include/linux/Kbuild +1 -0
index d9a7544..d823d60 100644
@@ -195,6 +195,7 @@ header-y += in_route.h
 header-y += sock_diag.h
 header-y += inet_diag.h
 header-y += unix_diag.h
+header-y += packet_diag.h
 header-y += inotify.h
 header-y += input.h
 header-y += ioctl.h
include/linux/packet_diag.h new file +24 -0
new file mode 100644
index 0000000..fa19300
@@ -0,0 +1,24 @@
+#ifndef __PACKET_DIAG_H__
+#define __PACKET_DIAG_H__
+
+#include <linux/types.h>
+
+struct packet_diag_req {
+	__u8	sdiag_family;
+	__u8	sdiag_protocol;
+	__u16	pad;
+	__u32	pdiag_ino;
+	__u32	pdiag_show;
+	__u32	pdiag_cookie[2];
+};
+
+struct packet_diag_msg {
+	__u8	pdiag_family;
+	__u8	pdiag_type;
+	__u16	pdiag_num;
+
+	__u32	pdiag_ino;
+	__u32	pdiag_cookie[2];
+};
+
+#endif
net/packet/Kconfig +8 -0
index 0060e3b..cc55b35 100644
@@ -14,3 +14,11 @@ config PACKET
 	  be called af_packet.
 
 	  If unsure, say Y.
+
+config PACKET_DIAG
+	tristate "Packet: sockets monitoring interface"
+	depends on PACKET
+	default n
+	---help---
+	  Support for PF_PACKET sockets monitoring interface used by the ss tool.
+	  If unsure, say Y.
net/packet/Makefile +2 -0
index 81183ea..9df6134 100644
@@ -3,3 +3,5 @@
 #
 
 obj-$(CONFIG_PACKET) += af_packet.o
+obj-$(CONFIG_PACKET_DIAG) += af_packet_diag.o
+af_packet_diag-y += diag.o
net/packet/diag.c new file +104 -0

104 lines not on this page · raw message

Signature · 1 line
1.7.6.5
author
#3 lore raw

PATCH 2/4 packet: Report more packet sk info via diag module

In reply to Pavel Emelyanov

This reports in one rtattr message all the other scalar values, that can be set on a packet socket with setsockopt.

Signed-off-by Pavel Emelyanov <[email protected]>
---
 include/linux/packet_diag.h |   23 +++++++++++++++++++++++
 net/packet/diag.c           |   33 +++++++++++++++++++++++++++++++++
 2 files changed, 56 insertions(+), 0 deletions(-)
include/linux/packet_diag.h +23 -0
index fa19300..3781ea4 100644
@@ -12,6 +12,8 @@ struct packet_diag_req {
 	__u32	pdiag_cookie[2];
 };
 
+#define PACKET_SHOW_INFO	0x00000001 /* Basic packet_sk information */
+
 struct packet_diag_msg {
 	__u8	pdiag_family;
 	__u8	pdiag_type;

@@ -21,4 +23,25 @@ struct packet_diag_msg {
 	__u32	pdiag_cookie[2];
 };
 
+enum {
+	PACKET_DIAG_INFO,
+
+	PACKET_DIAG_MAX,
+};
+
+struct packet_diag_info {
+	__u32	pdi_index;
+	__u32	pdi_version;
+	__u32	pdi_reserve;
+	__u32	pdi_copy_thresh;
+	__u32	pdi_tstamp;
+	__u32	pdi_flags;
+
+#define PDI_RUNNING	0x1
+#define PDI_AUXDATA	0x2
+#define PDI_ORIGDEV	0x4
+#define PDI_VNETHDR	0x8
+#define PDI_LOSS	0x10
+};
+
 #endif
net/packet/diag.c +33 -0
index ff2f7f5..d5bd037 100644
@@ -7,6 +7,31 @@
 
 #include "internal.h"
 
+static int pdiag_put_info(const struct packet_sock *po, struct sk_buff *nlskb)
+{
+	struct packet_diag_info pinfo;
+
+	pinfo.pdi_index = po->ifindex;
+	pinfo.pdi_version = po->tp_version;
+	pinfo.pdi_reserve = po->tp_reserve;
+	pinfo.pdi_copy_thresh = po->copy_thresh;
+	pinfo.pdi_tstamp = po->tp_tstamp;
+
+	pinfo.pdi_flags = 0;
+	if (po->running)
+		pinfo.pdi_flags |= PDI_RUNNING;
+	if (po->auxdata)
+		pinfo.pdi_flags |= PDI_AUXDATA;
+	if (po->origdev)
+		pinfo.pdi_flags |= PDI_ORIGDEV;
+	if (po->has_vnet_hdr)
+		pinfo.pdi_flags |= PDI_VNETHDR;
+	if (po->tp_loss)
+		pinfo.pdi_flags |= PDI_LOSS;
+
+	return nla_put(nlskb, PACKET_DIAG_INFO, sizeof(pinfo), &pinfo);
+}
+
 static int sk_diag_fill(struct sock *sk, struct sk_buff *skb, struct packet_diag_req *req,
 		u32 pid, u32 seq, u32 flags, int sk_ino)
 {

@@ -25,7 +50,15 @@ static int sk_diag_fill(struct sock *sk, struct sk_buff *skb, struct packet_diag
 	rp->pdiag_ino = sk_ino;
 	sock_diag_save_cookie(sk, rp->pdiag_cookie);
 
+	if ((req->pdiag_show & PACKET_SHOW_INFO) &&
+			pdiag_put_info(po, skb))
+		goto out_nlmsg_trim;
+
 	return nlmsg_end(skb, nlh);
+
+out_nlmsg_trim:
+	nlmsg_cancel(skb, nlh);
+	return -EMSGSIZE;
 }
 
 static int packet_diag_dump(struct sk_buff *skb, struct netlink_callback *cb)
Signature · 1 line
1.7.6.5
author
#4 lore raw

PATCH 4/4 packet: Report socket mclist info via diag module

In reply to Pavel Emelyanov

The info is reported as an array of packet_diag_mclist structures. Each includes not only the directly configured values (index, type, etc), but also the "count".

Signed-off-by Pavel Emelyanov <[email protected]>
---
 include/linux/packet_diag.h |   10 ++++++++++
 net/packet/diag.c           |   39 +++++++++++++++++++++++++++++++++++++++
 2 files changed, 49 insertions(+), 0 deletions(-)
include/linux/packet_diag.h +10 -0
index 3781ea4..ea2e8923 100644
@@ -13,6 +13,7 @@ struct packet_diag_req {
 };
 
 #define PACKET_SHOW_INFO	0x00000001 /* Basic packet_sk information */
+#define PACKET_SHOW_MCLIST	0x00000002 /* A set of packet_diag_mclist-s */
 
 struct packet_diag_msg {
 	__u8	pdiag_family;

@@ -25,6 +26,7 @@ struct packet_diag_msg {
 
 enum {
 	PACKET_DIAG_INFO,
+	PACKET_DIAG_MCLIST,
 
 	PACKET_DIAG_MAX,
 };

@@ -44,4 +46,12 @@ struct packet_diag_info {
 #define PDI_LOSS	0x10
 };
 
+struct packet_diag_mclist {
+	__u32	pdmc_index;
+	__u32	pdmc_count;
+	__u16	pdmc_type;
+	__u16	pdmc_alen;
+	__u8	pdmc_addr[MAX_ADDR_LEN];
+};
+
 #endif
net/packet/diag.c +39 -0
index d5bd037..3dda4ec 100644
@@ -1,6 +1,7 @@
 #include <linux/module.h>
 #include <linux/sock_diag.h>
 #include <linux/net.h>
+#include <linux/netdevice.h>
 #include <linux/packet_diag.h>
 #include <net/net_namespace.h>
 #include <net/sock.h>

@@ -32,6 +33,40 @@ static int pdiag_put_info(const struct packet_sock *po, struct sk_buff *nlskb)
 	return nla_put(nlskb, PACKET_DIAG_INFO, sizeof(pinfo), &pinfo);
 }
 
+static int pdiag_put_mclist(const struct packet_sock *po, struct sk_buff *nlskb)
+{
+	struct nlattr *mca;
+	struct packet_mclist *ml;
+
+	mca = nla_nest_start(nlskb, PACKET_DIAG_MCLIST);
+	if (!mca)
+		return -EMSGSIZE;
+
+	rtnl_lock();
+	for (ml = po->mclist; ml; ml = ml->next) {
+		struct packet_diag_mclist *dml;
+
+		dml = nla_reserve_nohdr(nlskb, sizeof(*dml));
+		if (!dml) {
+			rtnl_unlock();
+			nla_nest_cancel(nlskb, mca);
+			return -EMSGSIZE;
+		}
+
+		dml->pdmc_index = ml->ifindex;
+		dml->pdmc_type = ml->type;
+		dml->pdmc_alen = ml->alen;
+		dml->pdmc_count = ml->count;
+		BUILD_BUG_ON(sizeof(dml->pdmc_addr) != sizeof(ml->addr));
+		memcpy(dml->pdmc_addr, ml->addr, sizeof(ml->addr));
+	}
+
+	rtnl_unlock();
+	nla_nest_end(nlskb, mca);
+
+	return 0;
+}
+
 static int sk_diag_fill(struct sock *sk, struct sk_buff *skb, struct packet_diag_req *req,
 		u32 pid, u32 seq, u32 flags, int sk_ino)
 {

@@ -54,6 +89,10 @@ static int sk_diag_fill(struct sock *sk, struct sk_buff *skb, struct packet_diag
 			pdiag_put_info(po, skb))
 		goto out_nlmsg_trim;
 
+	if ((req->pdiag_show & PACKET_SHOW_MCLIST) &&
+			pdiag_put_mclist(po, skb))
+		goto out_nlmsg_trim;
+
 	return nlmsg_end(skb, nlh);
 
 out_nlmsg_trim:
Signature · 1 line
1.7.6.5
maintainer
#5 lore raw
In reply to Pavel Emelyanov

From: Pavel Emelyanov <[email protected]>
Date: Mon, 13 Aug 2012 19:47:40 +0400

Quoted 9 lines

There's a set of stuff residing on a AF_PACKET socket which is currently write-only -- rings, copy_thresh and mclist. But we need to know these while doing the checkpoint-restore.

As a solution for the similar problem with unix sockets, the sock-diag engine was developed, so here's the basic implementation of the packet sockets extension. It doesn't report rings, fanout and stats, but it can be easily added later.

Quoted 4 lines

I'd like to mention, that it's not a strict requirement, that the diag is used for getting info about sockets, it would be perfectly fine just to fix the packet getsockopt callback to return the desired info. So, if the diag for packet sockets is for any reason unacceptable, just let me know.

All applied, looks fine.

Please number your patches properly next time, patch #3 was listed as 2/4