[linux-yocto] [PATCH 1/2] netatop: Initial integration

Bruce Ashfield bruce.ashfield at windriver.com
Mon Apr 9 12:19:52 PDT 2018


On 2018-04-05 3:08 PM, Bruce Ashfield wrote:
> On 2018-03-28 11:25 PM, He Zhe wrote:
>> This is a net statistics collector for atop from:
>> http://atoptool.nl/netatop.php
>> https://atoptool.nl/download/netatop-2.0.tar.gz
>>
>> Two changes:
>> Adjust pathes for netatop.h and netatopversion.h
>> Add module_init and module_exit
> 
> Which kernel versions have you tested this on ? Is there a way we could get
> a README that describes how to build and test it ?

Also, is there a recipe that builds this that I can see/test ?

Bruce

> 
> Bruce
> 
>>
>> Signed-off-by: He Zhe <zhe.he at windriver.com>
>> ---
>>   drivers/staging/netatop/netatop.c        | 1843 
>> ++++++++++++++++++++++++++++++
>>   drivers/staging/netatop/netatop.h        |   47 +
>>   drivers/staging/netatop/netatopversion.h |    2 +
>>   3 files changed, 1892 insertions(+)
>>   create mode 100644 drivers/staging/netatop/netatop.c
>>   create mode 100644 drivers/staging/netatop/netatop.h
>>   create mode 100644 drivers/staging/netatop/netatopversion.h
>>
>> diff --git a/drivers/staging/netatop/netatop.c 
>> b/drivers/staging/netatop/netatop.c
>> new file mode 100644
>> index 000000000000..3baa6520416a
>> --- /dev/null
>> +++ b/drivers/staging/netatop/netatop.c
>> @@ -0,0 +1,1843 @@
>> +/*
>> +** This module uses the netfilter interface to maintain statistics
>> +** about the network traffic per task, on level of thread group
>> +** and individual thread.
>> +**
>> +** General setup
>> +** -------------
>> +** Once the module is active, it is called for every packet that is
>> +** transmitted by a local process and every packet that is received
>> +** from an interface. Not only the packets that contain the user data
>> +** are passed but also the TCP related protocol packets (SYN, ACK, ...).
>> +**
>> +** When the module discovers a packet for a connection (TCP) or local
>> +** port (UDP) that is new, it creates a sockinfo structure. As soon as
>> +** possible the sockinfo struct will be connected to a taskinfo struct
>> +** that represents the proces or thread that is related to the socket.
>> +** However, the task can only be determined when a packet is 
>> transmitted,
>> +** i.e. the module is called during system call handling in the context
>> +** of the transmitting process. At that moment the tgid (process) and
>> +** pid (thread) can be obtained from the process administration to
>> +** be stored in the module's own taskinfo structs (one for the process,
>> +** one for the thread).
>> +** For the time that the sockinfo struct can not be related to a 
>> taskinfo
>> +** struct (e.g. when only packets are received), counters are maintained
>> +** temporarily in the sockinfo struct. As soon as a related taskinfo 
>> struct
>> +** is discovered when the task transmits, counters will be maintained in
>> +** the taskinfo struct itself.
>> +** When packets are only received for a socket (e.g. another machine is
>> +** sending UDP packets to the local machine) while the local task
>> +** never responds, no match to a process can be made and the packets
>> +** remain unidentified by the netatop module. At least one packet should
>> +** have been sent by a local process to be able to match packets for 
>> such
>> +** socket.
>> +** In the file /proc/netatop counters can be found that show the total
>> +** number of packets sent/received and how many of these packets were
>> +** unidentified (i.e. not accounted to a process/thread).
>> +**
>> +** Garbage collection
>> +** ------------------
>> +** The module uses a garbage collector to cleanup the unused sockinfo
>> +** structs if connections do not exist any more (TCP) or have not been
>> +** used for some time (TCP/UDP).
>> +** Furthermore, the garbage collector checks if the taskinfo structs
>> +** still represent existing processes or threads. If not, the 
>> taskinfo struct
>> +** is destroyed (in case of a thread) or it is moved to a separate 
>> list of
>> +** finished processes (in case of a process). Analysis programs can read
>> +** the taskinfo of such finished process. When the taskinfo of a 
>> finished
>> +** process is not read within 15 seconds, the taskinfo will be 
>> destroyed.
>> +**
>> +** A garbage collector cycle can be triggered by issueing a getsockopt
>> +** call from an analysis program (e.g. atop). Apart from that, a 
>> time-based
>> +** garbage collector cycle is issued anyhow every 15 seconds by the
>> +** knetatop kernel thread.
>> +**
>> +** Interface with user mode
>> +** ------------------------
>> +** Programs can open an IP socket and use the getsockopt() system call
>> +** to issue commands to this module. With the command ATOP_GETCNT_TGID
>> +** the current counters can be obtained on process level (thread group)
>> +** and with the command ATOP_GETCNT_PID the counters on thread level.
>> +** For both commands, the tgid/pid has to be passed of the required 
>> thread
>> +** (group). When the required thread (group) does not exist, an errno 
>> ESRCH
>> +** is given.
>> +**
>> +** The command ATOP_GETCNT_EXIT can be issued to obtain the counters of
>> +** an exited process. As stated above, such command has to be issued
>> +** within 15 seconds after a process has been declared 'finished' by
>> +** the garbage collector. Whenever this command is issued and no exited
>> +** process is in the exitlist, the requesting process is blocked until
>> +** an exited process is available.
>> +**
>> +** The command NETATOP_FORCE_GC activates the garbage collector of the
>> +** netatop module to  determine if sockinfo's of old connections/ports
>> +** can be destroyed and if taskinfo's of exited processes can be
>> +** The command NETATOP_EMPTY_EXIT can be issued to wait until the 
>> exitlist
>> +** with the taskinfo's of exited processes is empty.
>> +** 
>> ----------------------------------------------------------------------
>> +** Copyright (C) 2012    Gerlof Langeveld (gerlof.langeveld at atoptool.nl)
>> +**
>> +** This program is free software; you can redistribute it and/or modify
>> +** it under the terms of the GNU General Public License version 2 as
>> +** published by the Free Software Foundation.
>> +*/
>> +#include <linux/init.h>
>> +#include <linux/kernel.h>
>> +#include <linux/module.h>
>> +#include <linux/netfilter.h>
>> +#include <linux/netfilter_ipv4.h>
>> +#include <linux/sched.h>
>> +#include <linux/skbuff.h>
>> +#include <linux/types.h>
>> +#include <linux/file.h>
>> +#include <linux/kthread.h>
>> +#include <linux/seq_file.h>
>> +#include <linux/version.h>
>> +#include <net/sock.h>
>> +#include <linux/ip.h>
>> +#include <linux/tcp.h>
>> +#include <linux/udp.h>
>> +#include <linux/proc_fs.h>
>> +#include <net/ip.h>
>> +#include <net/tcp.h>
>> +#include <net/udp.h>
>> +
>> +#include "netatop.h"
>> +#include "netatopversion.h"
>> +
>> +MODULE_LICENSE("GPL");
>> +MODULE_AUTHOR("Gerlof Langeveld <gerlof.langeveld at atoptool.nl>");
>> +MODULE_DESCRIPTION("Per-task network statistics");
>> +MODULE_VERSION(NETATOPVERSION);
>> +
>> +#define    GCINTERVAL    (HZ*15)        // interval garbage collector 
>> (jiffies)
>> +#define    GCMAXUDP    (HZ*16)        // max inactivity for UDP     
>> (jiffies)
>> +#define    GCMAXTCP    (HZ*1800)    // max inactivity for TCP     
>> (jiffies)
>> +#define    GCMAXUNREF    (HZ*60)        // max time without taskref   
>> (jiffies)
>> +
>> +#define    SILIMIT        (4096*1024)    // maximum memory for 
>> sockinfo structs
>> +#define    TILIMIT        (2048*1024)    // maximum memory for 
>> taskinfo structs
>> +
>> +#define NF_IP_PRE_ROUTING       0
>> +#define NF_IP_LOCAL_IN          1
>> +#define NF_IP_FORWARD           2
>> +#define NF_IP_LOCAL_OUT         3
>> +#define NF_IP_POST_ROUTING      4
>> +
>> +/*
>> +** struct that maintains statistics about the network
>> +** traffic caused per thread or thread group
>> +*/
>> +struct chainer {
>> +    void            *next;
>> +    void             *prev;
>> +};
>> +
>> +struct taskinfobucket;
>> +
>> +struct taskinfo {
>> +    struct chainer        ch;
>> +
>> +    pid_t            id;        // tgid or pid
>> +    char            type;        // 'g' (thread group) or
>> +                        // 't' (thread)
>> +    unsigned char        state;        // see below
>> +    char            command[COMLEN];
>> +    unsigned long        btime;        // start time of process
>> +    unsigned long long    exittime;    // time inserted in exitlist
>> +
>> +    struct taskcount    tc;
>> +};
>> +
>> +static struct kmem_cache    *ticache;    // taskinfo cache
>> +
>> +// state values above
>> +#define    CHECKED        1    // verified that task still exists
>> +#define    INDELETE    2    // task exited but still in hash list
>> +#define    FINISHED    3    // task on exit list
>> +
>> +/*
>> +** hash tables to find a particular thread group or thread
>> +*/
>> +#define    TBUCKS        1024    // must be multiple of 2!
>> +#define    THASH(x, t)    (((x)+t)&(TBUCKS-1))
>> +
>> +struct taskinfobucket {
>> +    struct chainer    ch;
>> +    spinlock_t    lock;
>> +}    thash[TBUCKS];
>> +
>> +static unsigned long    nrt;     // current number of taskinfo allocated
>> +static unsigned long    nrt_ovf; // no taskinfo allocated due to 
>> overflow
>> +static DEFINE_SPINLOCK(nrtlock);
>> +
>> +
>> +static struct taskinfo    *exithead;    // linked list of exited 
>> processes
>> +static struct taskinfo    *exittail;
>> +static DEFINE_SPINLOCK(exitlock);
>> +
>> +static DECLARE_WAIT_QUEUE_HEAD(exitlist_filled);
>> +static DECLARE_WAIT_QUEUE_HEAD(exitlist_empty);
>> +
>> +static unsigned long    nre;    // current number of taskinfo on 
>> exitlist
>> +
>> +/*
>> +** structs that uniquely identify a TCP connection (host endian format)
>> +*/
>> +struct tcpv4_ident {
>> +        uint32_t    laddr;  /* local  IP  address */
>> +        uint32_t    raddr;  /* remote IP  address */
>> +        uint16_t    lport;  /* local  port number */
>> +        uint16_t    rport;  /* remote port number */
>> +};
>> +
>> +struct tcpv6_ident {
>> +        struct in6_addr    laddr;  /* local  IP  address */
>> +        struct in6_addr    raddr;  /* remote IP  address */
>> +        uint16_t    lport;  /* local  port number */
>> +        uint16_t    rport;  /* remote port number */
>> +};
>> +
>> +/*
>> +** struct to maintain the reference from a socket
>> +** to a thread and thread-group
>> +*/
>> +struct sockinfo {
>> +    struct chainer        ch;
>> +
>> +    unsigned char        last_state;    // last known state of socket
>> +        uint8_t            proto;      // protocol
>> +
>> +    union keydef {
>> +        uint16_t        udp;    // UDP ident (only portnumber)
>> +        struct tcpv4_ident    tcp4;    // TCP connection ident IPv4
>> +        struct tcpv6_ident    tcp6;    // TCP connection ident IPv6
>> +    } key;
>> +
>> +    struct taskinfo        *tgp;        // ref to thread group
>> +    struct taskinfo        *thp;        // ref to thread (or NULL)
>> +
>> +    short            tgh;        // hash number of thread group
>> +    short            thh;        // hash number of thread
>> +
>> +        unsigned int          sndpacks;    // temporary counters in case
>> +        unsigned int          rcvpacks;     // known yet
>> +        unsigned long          sndbytes;    // no relation to process is
>> +        unsigned long          rcvbytes;
>> +
>> +    unsigned long long    lastact;    // last updated (jiffies)
>> +};
>> +
>> +static struct kmem_cache    *sicache;    // sockinfo cache
>> +
>> +/*
>> +** hash table to find a socket reference
>> +*/
>> +#define    SBUCKS        1024    // must be multiple of 2!
>> +#define    SHASHTCP4(x)    (((x).raddr+(x).lport+(x).rport)&(SBUCKS-1))
>> +#define    SHASHUDP(x)    ((x)&(SBUCKS-1))
>> +
>> +struct {
>> +    struct chainer    ch;
>> +    spinlock_t    lock;
>> +}    shash[SBUCKS];
>> +
>> +static unsigned long    nrs;     // current number sockinfo allocated
>> +static unsigned long    nrs_ovf; // no sockinfo allocated due to 
>> overflow
>> +static DEFINE_SPINLOCK(nrslock);
>> +
>> +/*
>> +** various static counters
>> +*/
>> +static unsigned long    icmpsndbytes;
>> +static unsigned long    icmpsndpacks;
>> +static unsigned long    icmprcvbytes;
>> +static unsigned long    icmprcvpacks;
>> +
>> +static unsigned long    tcpsndpacks;
>> +static unsigned long    tcprcvpacks;
>> +static unsigned long    udpsndpacks;
>> +static unsigned long    udprcvpacks;
>> +static unsigned long    unidentudpsndpacks;
>> +static unsigned long    unidentudprcvpacks;
>> +static unsigned long    unidenttcpsndpacks;
>> +static unsigned long    unidenttcprcvpacks;
>> +
>> +static unsigned long    unknownproto;
>> +
>> +static DEFINE_MUTEX(gclock);
>> +static unsigned long long    gclast;    // last garbage collection 
>> (jiffies)
>> +
>> +static struct task_struct    *knetatop_task;
>> +
>> +static struct timespec    boottime;
>> +
>> +/*
>> +** function prototypes
>> +*/
>> +static void         analyze_tcpv4_packet(struct sk_buff *,
>> +                const struct net_device *, int, char,
>> +                struct iphdr *, void *);
>> +
>> +static void         analyze_udp_packet(struct sk_buff *,
>> +                const struct net_device *, int, char,
>> +                struct iphdr *, void *);
>> +
>> +static int        sock2task(char, struct sockinfo *,
>> +                     struct taskinfo **, short *,
>> +                struct sk_buff *, const struct net_device *,
>> +                int, char);
>> +
>> +static void        update_taskcounters(struct sk_buff *,
>> +                const struct net_device *,
>> +                struct taskinfo *, char);
>> +
>> +static void        update_sockcounters(struct sk_buff *,
>> +                const struct net_device *,
>> +                struct sockinfo *, char);
>> +
>> +static void        sock2task_sync(struct sk_buff *,
>> +                struct sockinfo *, struct taskinfo *);
>> +
>> +static void        register_unident(struct sockinfo *);
>> +
>> +static int        calc_reallen(struct sk_buff *,
>> +                         const struct net_device *);
>> +
>> +static void        get_tcpv4_ident(struct iphdr *, void *,
>> +                char, union keydef *);
>> +
>> +static struct sockinfo    *find_sockinfo(int, union keydef *, int, int);
>> +static struct sockinfo    *make_sockinfo(int, union keydef *, int, int);
>> +
>> +static void        wipesockinfo(void);
>> +static void        wipetaskinfo(void);
>> +static void        wipetaskexit(void);
>> +
>> +static void        garbage_collector(void);
>> +static void        gctaskexit(void);
>> +static void        gcsockinfo(void);
>> +static void        gctaskinfo(void);
>> +
>> +static void        move_taskinfo(struct taskinfo *);
>> +static void        delete_taskinfo(struct taskinfo *);
>> +static void        delete_sockinfo(struct sockinfo *);
>> +
>> +static struct taskinfo    *get_taskinfo(pid_t, char);
>> +
>> +static int        getsockopt(struct sock *, int, void *, int *);
>> +
>> +static int        netatop_open(struct inode *inode, struct file *file);
>> +
>> +/*
>> +** hook definitions
>> +*/
>> +static struct nf_hook_ops hookin_ipv4;
>> +static struct nf_hook_ops hookout_ipv4;
>> +
>> +/*
>> +** getsockopt definitions for communication with user space
>> +*/
>> +static struct nf_sockopt_ops sockopts = {
>> +        .pf             = PF_INET,
>> +        .get_optmin     = NETATOP_BASE_CTL,
>> +        .get_optmax     = NETATOP_BASE_CTL+6,
>> +        .get            = getsockopt,
>> +        .owner          = THIS_MODULE,
>> +};
>> +
>> +static struct file_operations netatop_proc_fops = {
>> +    .open           = netatop_open,
>> +    .read            = seq_read,
>> +    .llseek            = seq_lseek,
>> +    .release    = single_release,
>> +    .owner          = THIS_MODULE,
>> +};
>> +
>> +
>> +/*
>> +** hook function to be called for every incoming local packet
>> +*/
>> +#if LINUX_VERSION_CODE >= KERNEL_VERSION(3, 10, 0)
>> +#  if LINUX_VERSION_CODE >= KERNEL_VERSION(4, 4, 0)
>> +#    define    HOOK_ARG_TYPE    void *priv
>> +#  else
>> +#    define    HOOK_ARG_TYPE    const struct nf_hook_ops *ops
>> +#  endif
>> +#else
>> +#  define    HOOK_ARG_TYPE    unsigned int hooknum
>> +#endif
>> +
>> +#if LINUX_VERSION_CODE >= KERNEL_VERSION(4, 1, 0)
>> +#define HOOK_STATE_ARGS const struct nf_hook_state *state
>> +
>> +#define DEV_IN        state->in
>> +#define DEV_OUT        state->out
>> +#else
>> +#  ifndef __GENKSYMS__
>> +#    define HOOK_STATE_ARGS const struct net_device *in, \
>> +                const struct net_device *out, \
>> +                            const struct nf_hook_state *state
>> +#  else
>> +#    define HOOK_STATE_ARGS const struct net_device *in, \
>> +                const struct net_device *out, \
>> +                int (*okfn)(struct sk_buff *)
>> +#  endif
>> +#define DEV_IN        in
>> +#define DEV_OUT        out
>> +#endif
>> +
>> +
>> +static unsigned int
>> +ipv4_hookin(HOOK_ARG_TYPE, struct sk_buff *skb, HOOK_STATE_ARGS)
>> +{
>> +    struct iphdr    *iph;
>> +    void        *trh;
>> +
>> +    if (skb == NULL)        // useless socket buffer?
>> +        return NF_ACCEPT;
>> +
>> +    /*
>> +    ** get pointer to IP header and transport header
>> +    */
>> +    iph = (struct iphdr *)skb_network_header(skb);
>> +    trh = ((char *)iph + (iph->ihl * 4));
>> +
>> +    /*
>> +    ** react on protocol number
>> +    */
>> +    switch (iph->protocol) {
>> +       case IPPROTO_TCP:
>> +        tcprcvpacks++;
>> +        analyze_tcpv4_packet(skb, DEV_IN, 0, 'i', iph, trh);
>> +        break;
>> +
>> +       case IPPROTO_UDP:
>> +        udprcvpacks++;
>> +        analyze_udp_packet(skb, DEV_IN, 0, 'i', iph, trh);
>> +        break;
>> +
>> +       case IPPROTO_ICMP:
>> +        icmprcvpacks++;
>> +        icmprcvbytes += skb->len + DEV_IN->hard_header_len + 4;
>> +        break;
>> +
>> +       default:
>> +        unknownproto++;
>> +    }
>> +
>> +    // accept every packet after stats gathering
>> +    return NF_ACCEPT;
>> +}
>> +
>> +/*
>> +** hook function to be called for every outgoing local packet
>> +*/
>> +static unsigned int
>> +ipv4_hookout(HOOK_ARG_TYPE, struct sk_buff *skb, HOOK_STATE_ARGS)
>> +{
>> +    int        in_syscall = !in_interrupt();
>> +    struct iphdr    *iph;
>> +    void        *trh;
>> +
>> +    if (skb == NULL)        // useless socket buffer?
>> +        return NF_ACCEPT;
>> +
>> +    /*
>> +    ** get pointer to IP header and transport header
>> +    */
>> +    iph = (struct iphdr *)skb_network_header(skb);
>> +    trh = skb_transport_header(skb);
>> +
>> +    /*
>> +    ** react on protocol number
>> +    */
>> +    switch (iph->protocol) {
>> +       case IPPROTO_TCP:
>> +        tcpsndpacks++;
>> +        analyze_tcpv4_packet(skb, DEV_OUT, in_syscall, 'o', iph, trh);
>> +        break;
>> +
>> +       case IPPROTO_UDP:
>> +        udpsndpacks++;
>> +        analyze_udp_packet(skb, DEV_OUT, in_syscall, 'o', iph, trh);
>> +        break;
>> +
>> +       case IPPROTO_ICMP:
>> +        icmpsndpacks++;
>> +        icmpsndbytes += skb->len + DEV_OUT->hard_header_len + 4;
>> +        break;
>> +
>> +       default:
>> +        unknownproto++;
>> +    }
>> +
>> +    // accept every packet after stats gathering
>> +    return NF_ACCEPT;
>> +}
>> +
>> +/*
>> +** generic function (for input and output) to analyze the current packet
>> +*/
>> +static void
>> +analyze_tcpv4_packet(struct sk_buff *skb,
>> +    const struct net_device *ndev,    // interface description
>> +    int in_syscall,            // called during system call?
>> +    char direction,            // incoming ('i') or outgoing ('o')
>> +    struct iphdr *iph, void *trh)
>> +{
>> +    union keydef        key;
>> +    struct sockinfo        *sip;
>> +    int            bs;    // hash bucket for sockinfo
>> +    unsigned long        sflags;
>> +
>> +    /*
>> +    ** determine tcpv4_ident that identifies this TCP packet
>> +    ** and calculate hash bucket in sockinfo hash
>> +    */
>> +    get_tcpv4_ident(iph, trh, direction, &key);
>> +
>> +    /*
>> +    ** check if we have seen this tcpv4_ident before with a
>> +    ** corresponding thread and thread group
>> +    */
>> +    bs = SHASHTCP4(key.tcp4);
>> +
>> +    spin_lock_irqsave(&shash[bs].lock, sflags);
>> +
>> +    if ( (sip = find_sockinfo(IPPROTO_TCP, &key, sizeof key.tcp4, bs))
>> +                                == NULL) {
>> +        // no sockinfo yet: create one
>> +        if ( (sip = make_sockinfo(IPPROTO_TCP, &key,
>> +                    sizeof key.tcp4, bs)) == NULL) {
>> +            if (direction == 'i')
>> +                unidenttcprcvpacks++;
>> +            else
>> +                unidenttcpsndpacks++;
>> +            goto unlocks;
>> +        }
>> +    }
>> +
>> +    if (skb->sk)
>> +        sip->last_state = skb->sk->sk_state;
>> +
>> +    /*
>> +    ** if needed (re)connect the sockinfo to a taskinfo and update
>> +    ** the counters
>> +    */
>> +
>> +    // connect to thread group and update
>> +    if (sock2task('g', sip, &sip->tgp, &sip->tgh,
>> +                skb, ndev, in_syscall, direction)) {
>> +        // connect to thread and update
>> +        (void) sock2task('t', sip, &sip->thp, &sip->thh,
>> +                skb, ndev, in_syscall, direction);
>> +    }
>> +
>> +unlocks:
>> +    spin_unlock_irqrestore(&shash[bs].lock, sflags);
>> +}
>> +
>> +
>> +/*
>> +** generic function (for input and output) to analyze the current packet
>> +*/
>> +static void
>> +analyze_udp_packet(struct sk_buff *skb,
>> +    const struct net_device *ndev,    // interface description
>> +    int in_syscall,            // called during system call?
>> +    char direction,            // incoming ('i') or outgoing ('o')
>> +    struct iphdr *iph, void *trh)
>> +{
>> +    struct udphdr    *udph = (struct udphdr *)trh;
>> +    uint16_t    udplocal = (direction == 'i' ?
>> +                    ntohs(udph->dest) : ntohs(udph->source));
>> +    int        bs;    // hash bucket for sockinfo
>> +
>> +    union keydef     key;
>> +    struct sockinfo    *sip;
>> +    unsigned long    sflags;
>> +
>> +    /*
>> +    ** check if we have seen this local UDP port before with a
>> +    ** corresponding thread and thread group
>> +    */
>> +    key.udp    = udplocal;
>> +    bs      = SHASHUDP(udplocal);
>> +
>> +    spin_lock_irqsave(&shash[bs].lock, sflags);
>> +
>> +    if ( (sip = find_sockinfo(IPPROTO_UDP, &key, sizeof key.udp, bs))
>> +                                == NULL) {
>> +        // no sockinfo yet: create one
>> +        if ( (sip = make_sockinfo(IPPROTO_UDP, &key,
>> +                sizeof key.udp, bs)) == NULL) {
>> +            if (direction == 'i')
>> +                unidentudprcvpacks++;
>> +            else
>> +                unidentudpsndpacks++;
>> +            goto unlocks;
>> +        }
>> +    }
>> +
>> +    /*
>> +    ** if needed (re)connect the sockinfo to a taskinfo and update
>> +    ** the counters
>> +    */
>> +
>> +    // connect to thread group and update
>> +    if (sock2task('g', sip, &sip->tgp, &sip->tgh,
>> +                skb, ndev, in_syscall, direction)) {
>> +        // connect to thread and update
>> +        (void) sock2task('t', sip, &sip->thp, &sip->thh,
>> +                skb, ndev, in_syscall, direction);
>> +    }
>> +
>> +unlocks:
>> +    spin_unlock_irqrestore(&shash[bs].lock, sflags);
>> +}
>> +
>> +/*
>> +** connect the sockinfo to the correct taskinfo and update the counters
>> +*/
>> +static int
>> +sock2task(char idtype, struct sockinfo *sip, struct taskinfo **tipp,
>> +    short *hash, struct sk_buff *skb, const struct net_device *ndev,
>> +    int in_syscall, char direction)
>> +{
>> +    pid_t        curid;
>> +    unsigned long    tflags;
>> +
>> +    if (*tipp == NULL) {
>> +        /*
>> +        ** no taskinfo connected yet for this reference from
>> +        ** sockinfo; to connect to a taskinfo, we must
>> +        ** be in system call handling now --> verify
>> +        */
>> +        if (!in_syscall) {
>> +            if (idtype == 'g')
>> +                update_sockcounters(skb, ndev, sip, direction);
>> +
>> +            return 0;    // failed
>> +        }
>> +
>> +        /*
>> +        ** try to find existing taskinfo or create new taskinfo
>> +        */
>> +        curid = (idtype == 'g' ? current->tgid : current->pid);
>> +
>> +        *hash = THASH(curid, idtype);        // calc hashQ
>> +
>> +        spin_lock_irqsave(&thash[*hash].lock, tflags);
>> +
>> +        if ( (*tipp = get_taskinfo(curid, idtype)) == NULL) {
>> +            /*
>> +            ** not possible to connect
>> +            */
>> +            spin_unlock_irqrestore(&thash[*hash].lock, tflags);
>> +
>> +            if (idtype == 'g')
>> +                update_sockcounters(skb, ndev, sip, direction);
>> +
>> +            return 0;            // failed
>> +        }
>> +
>> +        /*
>> +        ** new connection made:
>> +        ** update task counters with sock counters
>> +        */
>> +        sock2task_sync(skb, sip, *tipp);
>> +    } else {
>> +        /*
>> +        ** already related to thread group or thread
>> +        ** lock existing task
>> +        */
>> +        spin_lock_irqsave(&thash[*hash].lock, tflags);
>> +
>> +        /*
>> +        ** check if socket has been passed to another process in the
>> +        ** meantime, like programs as xinetd use to do
>> +        ** if so, connect sockinfo to the new task
>> +        */
>> +        if (in_syscall) {
>> +            curid = (idtype == 'g' ? current->tgid : current->pid);
>> +
>> +            if ((*tipp)->id != curid) {
>> +                spin_unlock_irqrestore(&thash[*hash].lock,
>> +                                    tflags);
>> +                *hash = THASH(curid, idtype);
>> +
>> +                spin_lock_irqsave(&thash[*hash].lock, tflags);
>> +
>> +                if ( (*tipp = get_taskinfo(curid, idtype))
>> +                                == NULL) {
>> +                    spin_unlock_irqrestore(
>> +                        &thash[*hash].lock, tflags);
>> +                    return 0;
>> +                }
>> +            }
>> +        }
>> +    }
>> +
>> +    update_taskcounters(skb, ndev, *tipp, direction);
>> +
>> +    spin_unlock_irqrestore(&thash[*hash].lock, tflags);
>> +
>> +    return 1;
>> +}
>> +
>> +/*
>> +** update the statistics of a particular thread group or thread
>> +*/
>> +static void
>> +update_taskcounters(struct sk_buff *skb, const struct net_device *ndev,
>> +        struct taskinfo *tip, char direction)
>> +{
>> +    struct iphdr    *iph = (struct iphdr *)skb_network_header(skb);
>> +    int        reallen = calc_reallen(skb, ndev);
>> +
>> +    switch (iph->protocol) {
>> +       case IPPROTO_TCP:
>> +        if (direction == 'i') {
>> +            tip->tc.tcprcvpacks++;
>> +            tip->tc.tcprcvbytes += reallen;
>> +        } else {
>> +            tip->tc.tcpsndpacks++;
>> +            tip->tc.tcpsndbytes += reallen;
>> +        }
>> +        break;
>> +
>> +       case IPPROTO_UDP:
>> +        if (direction == 'i') {
>> +            tip->tc.udprcvpacks++;
>> +            tip->tc.udprcvbytes += reallen;
>> +        } else {
>> +            tip->tc.udpsndpacks++;
>> +            tip->tc.udpsndbytes += reallen;
>> +        }
>> +     }
>> +}
>> +
>> +/*
>> +** update the statistics of a sockinfo without a connected task
>> +*/
>> +static void
>> +update_sockcounters(struct sk_buff *skb, const struct net_device *ndev,
>> +        struct sockinfo *sip, char direction)
>> +{
>> +    int    reallen = calc_reallen(skb, ndev);
>> +
>> +    if (direction == 'i') {
>> +        sip->rcvpacks++;
>> +        sip->rcvbytes += reallen;
>> +    } else {
>> +        sip->sndpacks++;
>> +        sip->sndbytes += reallen;
>> +    }
>> +}
>> +
>> +/*
>> +** add the temporary counters in the sockinfo to the new connected task
>> +*/
>> +static void
>> +sock2task_sync(struct sk_buff *skb, struct sockinfo *sip, struct 
>> taskinfo *tip)
>> +{
>> +    struct iphdr    *iph = (struct iphdr *)skb_network_header(skb);
>> +
>> +    switch (iph->protocol) {
>> +       case IPPROTO_TCP:
>> +        tip->tc.tcprcvpacks    += sip->rcvpacks;
>> +        tip->tc.tcprcvbytes     += sip->rcvbytes;
>> +        tip->tc.tcpsndpacks    += sip->sndpacks;
>> +        tip->tc.tcpsndbytes    += sip->sndbytes;
>> +        break;
>> +
>> +       case IPPROTO_UDP:
>> +        tip->tc.udprcvpacks    += sip->rcvpacks;
>> +        tip->tc.udprcvbytes     += sip->rcvbytes;
>> +        tip->tc.udpsndpacks    += sip->sndpacks;
>> +        tip->tc.udpsndbytes    += sip->sndbytes;
>> +    }
>> +}
>> +
>> +static void
>> +register_unident(struct sockinfo *sip)
>> +{
>> +    switch (sip->proto) {
>> +       case IPPROTO_TCP:
>> +        unidenttcprcvpacks += sip->rcvpacks;
>> +        unidenttcpsndpacks += sip->sndpacks;
>> +        break;
>> +
>> +       case IPPROTO_UDP:
>> +        unidentudprcvpacks += sip->rcvpacks;
>> +        unidentudpsndpacks += sip->sndpacks;
>> +    }
>> +}
>> +
>> +/*
>> +** calculate the number of bytes that are really sent or received
>> +*/
>> +static int
>> +calc_reallen(struct sk_buff *skb, const struct net_device *ndev)
>> +{
>> +    /*
>> +    ** calculate the real load of this packet on the network:
>> +    **
>> +    **  - length of IP header, TCP/UDP header and data (skb->len)
>> +    **
>> +    **    since packet assembly/disassembly is done by the IP layer
>> +    **    (we get an input packet that has been assembled already and
>> +    **    an output packet that still has to be disassembled), 
>> additional
>> +    **    IP headers/interface headers and interface headers have
>> +    **    to be calculated for packets that are larger than the mtu
>> +    **
>> +    **  - interface header length + 4 bytes crc
>> +    */
>> +    int reallen = skb->len;
>> +
>> +    if (reallen > ndev->mtu)
>> +        reallen += (reallen / ndev->mtu) *
>> +               (sizeof(struct iphdr) + ndev->hard_header_len + 4);
>> +
>> +    reallen += ndev->hard_header_len + 4;
>> +
>> +    return reallen;
>> +}
>> +
>> +/*
>> +** find the tcpv4_ident for the current packet, represented by
>> +** the skb_buff
>> +*/
>> +static void
>> +get_tcpv4_ident(struct iphdr *iph, void *trh, char direction, union 
>> keydef *key)
>> +{
>> +    struct tcphdr    *tcph = (struct tcphdr *)trh;
>> +
>> +    memset(key, 0, sizeof *key);     // important for memcmp later on
>> +
>> +    /*
>> +    ** determine local/remote IP address and
>> +    ** determine local/remote port number
>> +    */
>> +    switch (direction) {
>> +       case 'i':    // incoming packet
>> +        key->tcp4.laddr = ntohl(iph->daddr);
>> +        key->tcp4.raddr = ntohl(iph->saddr);
>> +        key->tcp4.lport = ntohs(tcph->dest);
>> +        key->tcp4.rport = ntohs(tcph->source);
>> +        break;
>> +
>> +       case 'o':    // outgoing packet
>> +        key->tcp4.laddr = ntohl(iph->saddr);
>> +        key->tcp4.raddr = ntohl(iph->daddr);
>> +        key->tcp4.lport = ntohs(tcph->source);
>> +        key->tcp4.rport = ntohs(tcph->dest);
>> +    }
>> +}
>> +
>> +/*
>> +** search for the sockinfo holding the given address info
>> +** the appropriate hash bucket must have been locked before calling
>> +*/
>> +static struct sockinfo *
>> +find_sockinfo(int proto, union keydef *identp, int identsz, int hash)
>> +{
>> +    struct sockinfo    *sip = shash[hash].ch.next;
>> +
>> +    /*
>> +    ** search for appropriate struct
>> +    */
>> +    while (sip != (void *)&shash[hash].ch) {
>> +        if ( memcmp(&sip->key, identp, identsz) == 0 &&
>> +                        sip->proto == proto) {
>> +            sip->lastact = jiffies_64;
>> +            return sip;
>> +        }
>> +
>> +        sip = sip->ch.next;
>> +    }
>> +
>> +    return NULL;    // not existing
>> +}
>> +
>> +/*
>> +** create a new sockinfo and fill
>> +** the appropriate hash bucket must have been locked before calling
>> +*/
>> +static struct sockinfo *
>> +make_sockinfo(int proto, union keydef *identp, int identsz, int hash)
>> +{
>> +    struct sockinfo    *sip;
>> +    unsigned long    flags;
>> +
>> +    /*
>> +    ** check if the threshold of memory used for sockinfo structs
>> +    ** is reached to avoid that a fork bomb of processes opening
>> +    ** a socket leads to memory overload
>> +    */
>> +    if ( (nrs+1) * sizeof(struct sockinfo) > SILIMIT) {
>> +        spin_lock_irqsave(&nrslock, flags);
>> +        nrs_ovf++;
>> +        spin_unlock_irqrestore(&nrslock, flags);
>> +        return NULL;
>> +    }
>> +
>> +    if ( (sip = kmem_cache_alloc(sicache, GFP_ATOMIC)) == NULL)
>> +        return NULL;
>> +
>> +    spin_lock_irqsave(&nrslock, flags);
>> +    nrs++;
>> +    spin_unlock_irqrestore(&nrslock, flags);
>> +
>> +    /*
>> +    ** insert new struct in doubly linked list
>> +    */
>> +    memset(sip, '\0', sizeof *sip);
>> +
>> +    sip->ch.next         = &shash[hash].ch;
>> +    sip->ch.prev         =  shash[hash].ch.prev;
>> +    ((struct sockinfo *)shash[hash].ch.prev)->ch.next = sip;
>> +    shash[hash].ch.prev     = sip;
>> +
>> +    sip->proto         = proto;
>> +    sip->lastact         = jiffies_64;
>> +    sip->key         = *identp;
>> +
>> +    return sip;
>> +}
>> +
>> +/*
>> +** search the taskinfo structure holding the info about the given 
>> id/type
>> +** if such taskinfo is not yet present, create a new one
>> +*/
>> +static struct taskinfo *
>> +get_taskinfo(pid_t id, char type)
>> +{
>> +    int        bt = THASH(id, type);
>> +    struct taskinfo    *tip = thash[bt].ch.next;
>> +    unsigned long    tflags;
>> +
>> +    /*
>> +    ** search if id exists already
>> +    */
>> +    while (tip != (void *)&thash[bt].ch) {
>> +        if (tip->id == id && tip->type == type)
>> +            return tip;
>> +
>> +        tip = tip->ch.next;
>> +    }
>> +
>> +    /*
>> +    ** check if the threshold of memory used for taskinfo structs
>> +    ** is reached to avoid that a fork bomb of processes opening
>> +    ** a socket lead to memory overload
>> +    */
>> +    if ( (nre+nrt+1) * sizeof(struct taskinfo) > TILIMIT) {
>> +        spin_lock_irqsave(&nrtlock, tflags);
>> +        nrt_ovf++;
>> +        spin_unlock_irqrestore(&nrtlock, tflags);
>> +        return NULL;
>> +    }
>> +
>> +    /*
>> +    ** id not known yet
>> +    ** add new entry to hash list
>> +    */
>> +    if ( (tip = kmem_cache_alloc(ticache, GFP_ATOMIC)) == NULL)
>> +        return NULL;
>> +
>> +    spin_lock_irqsave(&nrtlock, tflags);
>> +    nrt++;
>> +    spin_unlock_irqrestore(&nrtlock, tflags);
>> +
>> +    /*
>> +    ** insert new struct in doubly linked list
>> +    ** and fill values
>> +    */
>> +    memset(tip, '\0', sizeof *tip);
>> +
>> +    tip->ch.next         = &thash[bt].ch;
>> +    tip->ch.prev         =  thash[bt].ch.prev;
>> +    ((struct taskinfo *)thash[bt].ch.prev)->ch.next = tip;
>> +    thash[bt].ch.prev      = tip;
>> +
>> +    tip->id        = id;
>> +    tip->type        = type;
>> +
>> +#if LINUX_VERSION_CODE >= KERNEL_VERSION(3, 17, 0)
>> +    tip->btime    = div_u64((current->real_start_time +
>> +                (boottime.tv_sec * NSEC_PER_SEC +
>> +                    boottime.tv_sec)), NSEC_PER_SEC);
>> +#else
>> +    // current->real_start_time is type u64
>> +    tip->btime     = current->real_start_time.tv_sec + boottime.tv_sec;
>> +
>> +    if (current->real_start_time.tv_nsec + boottime.tv_nsec > 
>> NSEC_PER_SEC)
>> +        tip->btime++;
>> +#endif
>> +
>> +    strncpy(tip->command, current->comm, COMLEN);
>> +
>> +    return tip;
>> +}
>> +
>> +/*
>> +** garbage collector that removes:
>> +** - exited tasks that are no longer used by user mode programs
>> +** - sockinfo's that are not used any more
>> +** - taskinfo's that do not exist any more
>> +**
>> +** a mutex avoids that the garbage collector runs several times in 
>> parallel
>> +**
>> +** this function may only be called in process context!
>> +*/
>> +static void
>> +garbage_collector(void)
>> +{
>> +    mutex_lock(&gclock);
>> +
>> +    if (jiffies_64 < gclast + (HZ/2)) { // maximum 2 GC cycles per 
>> second
>> +        mutex_unlock(&gclock);
>> +        return;
>> +    }
>> +
>> +    gctaskexit();    // remove remaining taskinfo structs from exit list
>> +
>> +    gcsockinfo();    // clean up sockinfo structs in shash list
>> +
>> +    gctaskinfo();    // clean up taskinfo structs in thash list
>> +
>> +    gclast = jiffies_64;
>> +
>> +    mutex_unlock(&gclock);
>> +}
>> +
>> +/*
>> +** tasks in the exitlist can be read by a user mode process for a 
>> limited
>> +** amount of time; this function removes all taskinfo structures that 
>> have
>> +** not been read within that period of time
>> +** notice that exited processes are chained to the tail, so the oldest
>> +** can be found at the head
>> +*/
>> +static void
>> +gctaskexit()
>> +{
>> +    unsigned long    flags;
>> +    struct taskinfo    *tip;
>> +
>> +    spin_lock_irqsave(&exitlock, flags);
>> +
>> +    for (tip=exithead; tip;) {
>> +        if (jiffies_64 < tip->exittime + GCINTERVAL)
>> +            break;
>> +
>> +        // remove taskinfo from exitlist
>> +        exithead = tip->ch.next;
>> +        kmem_cache_free(ticache, tip);
>> +        nre--;
>> +        tip = exithead;
>> +    }
>> +
>> +    /*
>> +    ** if list empty now, then exithead and exittail both NULL
>> +    ** wakeup waiters for emptylist
>> +    */
>> +    if (nre == 0) {
>> +        exittail = NULL;
>> +        wake_up_interruptible(&exitlist_empty);
>> +    }
>> +
>> +    spin_unlock_irqrestore(&exitlock, flags);
>> +}
>> +
>> +/*
>> +** cleanup sockinfo structures that are connected to finished processes
>> +*/
>> +static void
>> +gcsockinfo()
>> +{
>> +    int        i;
>> +    struct sockinfo    *sip, *sipsave;
>> +    unsigned long    sflags, tflags;
>> +    struct pid    *pid;
>> +
>> +    /*
>> +    ** go through all sockinfo hash buckets
>> +    */
>> +    for (i=0; i < SBUCKS; i++) {
>> +        if (shash[i].ch.next == (void *)&shash[i].ch)
>> +            continue;    // quick return without lock
>> +
>> +        spin_lock_irqsave(&shash[i].lock, sflags);
>> +
>> +        sip  = shash[i].ch.next;
>> +
>> +        /*
>> +        ** search all sockinfo structs chained in one bucket
>> +        */
>> +        while (sip != (void *)&shash[i].ch) {
>> +            /*
>> +             ** TCP connections that were not in
>> +            ** state ESTABLISHED or LISTEN can be
>> +            ** eliminated
>> +            */
>> +            if (sip->proto == IPPROTO_TCP) {
>> +                switch (sip->last_state) {
>> +                   case TCP_ESTABLISHED:
>> +                   case TCP_LISTEN:
>> +                    break;
>> +
>> +                   default:
>> +                    sipsave = sip->ch.next;
>> +                    delete_sockinfo(sip);
>> +                    sip = sipsave;
>> +                    continue;
>> +                }
>> +            }
>> +
>> +            /*
>> +            ** check if this sockinfo has no relation
>> +            ** for a while with a thread group
>> +            ** if so, delete the sockinfo
>> +            */
>> +            if (sip->tgp == NULL) {
>> +                if (sip->lastact + GCMAXUNREF < jiffies_64) {
>> +                    register_unident(sip);
>> +                    sipsave = sip->ch.next;
>> +                    delete_sockinfo(sip);
>> +                    sip = sipsave;
>> +                } else {
>> +                    sip = sip->ch.next;
>> +                }
>> +                continue;
>> +            }
>> +
>> +            /*
>> +            ** check if referred thread group is
>> +            ** already marked as 'indelete' during this
>> +            ** sockinfo search
>> +            ** if so, delete this sockinfo
>> +            */
>> +            spin_lock_irqsave(&thash[sip->tgh].lock, tflags);
>> +
>> +            if (sip->tgp->state == INDELETE) {
>> +                spin_unlock_irqrestore(&thash[sip->tgh].lock,
>> +                                    tflags);
>> +                sipsave = sip->ch.next;
>> +                delete_sockinfo(sip);
>> +                sip = sipsave;
>> +                continue;
>> +            }
>> +
>> +            /*
>> +            ** check if referred thread group still exists;
>> +            ** this step will be skipped if we already verified
>> +            ** the existance of the thread group earlier during
>> +            ** this garbage collection cycle
>> +            */
>> +            if (sip->tgp->state != CHECKED) {
>> +                /*
>> +                ** connected thread group not yet verified
>> +                ** during this cycle, so check if it still
>> +                ** exists
>> +                ** if not, mark the thread group as 'indelete'
>> +                ** (it can not be deleted right now because
>> +                ** we might find other sockinfo's referring
>> +                ** to this thread group during the current
>> +                ** cycle) and delete this sockinfo
>> +                ** if the thread group exists, just mark
>> +                ** it  as 'checked' for this cycle
>> +                */
>> +                rcu_read_lock();
>> +                pid = find_vpid(sip->tgp->id);
>> +                rcu_read_unlock();
>> +
>> +                if (pid == NULL) {
>> +                    sip->tgp->state = INDELETE;
>> +                    spin_unlock_irqrestore(
>> +                        &thash[sip->tgh].lock, tflags);
>> +
>> +                    sipsave = sip->ch.next;
>> +                    delete_sockinfo(sip);
>> +                    sip = sipsave;
>> +                    continue;
>> +                } else {
>> +                    sip->tgp->state = CHECKED;
>> +                }
>> +            }
>> +
>> +            spin_unlock_irqrestore(&thash[sip->tgh].lock, tflags);
>> +
>> +            /*
>> +            ** check if this sockinfo has a relation with a thread
>> +            ** if not, skip further handling of this sockinfo
>> +            */
>> +            if (sip->thp == NULL) {
>> +                sip = sip->ch.next;
>> +                continue;
>> +            }
>> +
>> +            /*
>> +            ** check if referred thread is already marked
>> +            ** as 'indelete' during this sockinfo search
>> +            ** if so, break connection
>> +            */
>> +            spin_lock_irqsave(&thash[sip->thh].lock, tflags);
>> +
>> +            if (sip->thp->state == INDELETE) {
>> +                spin_unlock_irqrestore(&thash[sip->thh].lock,
>> +                                    tflags);
>> +                sip->thp = NULL;
>> +                sip = sip->ch.next;
>> +                continue;
>> +            }
>> +
>> +            /*
>> +            ** check if referred thread is already checked
>> +            ** during this sockinfo search
>> +            */
>> +            if (sip->thp->state == CHECKED) {
>> +                spin_unlock_irqrestore(&thash[sip->thh].lock,
>> +                                tflags);
>> +                sip = sip->ch.next;
>> +                continue;
>> +            }
>> +
>> +            /*
>> +            ** connected thread not yet verified
>> +            ** check if it still exists
>> +            ** if not, mark it as 'indelete' and break connection
>> +            ** if thread exists, mark it 'checked'
>> +            */
>> +            rcu_read_lock();
>> +            pid = find_vpid(sip->thp->id);
>> +            rcu_read_unlock();
>> +
>> +            if (pid == NULL) {
>> +                sip->thp->state = INDELETE;
>> +                sip->thp = NULL;
>> +            } else {
>> +                sip->thp->state = CHECKED;
>> +            }
>> +
>> +            spin_unlock_irqrestore(&thash[sip->thh].lock, tflags);
>> +
>> +            /*
>> +            ** check if a TCP port has not been used
>> +            ** for some time --> destroy even if the thread
>> +            ** (group) is still there
>> +            */
>> +            if (sip->proto == IPPROTO_TCP &&
>> +                sip->lastact + GCMAXTCP < jiffies_64) {
>> +                sipsave = sip->ch.next;
>> +                delete_sockinfo(sip);
>> +                sip = sipsave;
>> +                continue;
>> +            }
>> +
>> +            /*
>> +            ** check if a UDP port has not been used
>> +            ** for some time --> destroy even if the thread
>> +            ** (group) is still there
>> +            ** e.g. outgoing DNS requests (to remote port 53) are
>> +            ** issued every time with another source port being
>> +            ** a new object that should not be kept too long;
>> +            ** local well-known ports are useful to keep
>> +            */
>> +            if (sip->proto == IPPROTO_UDP &&
>> +                sip->lastact + GCMAXUDP < jiffies_64 &&
>> +                sip->key.udp > 1024)                {
>> +                sipsave = sip->ch.next;
>> +                delete_sockinfo(sip);
>> +                sip = sipsave;
>> +                continue;
>> +            }
>> +
>> +            sip = sip->ch.next;
>> +        }
>> +
>> +        spin_unlock_irqrestore(&shash[i].lock, sflags);
>> +    }
>> +}
>> +
>> +/*
>> +** remove taskinfo structures of finished tasks from hash list
>> +*/
>> +static void
>> +gctaskinfo()
>> +{
>> +    int        i;
>> +    struct taskinfo    *tip, *tipsave;
>> +    unsigned long    tflags;
>> +    struct pid    *pid;
>> +
>> +    /*
>> +    ** go through all taskinfo hash buckets
>> +    */
>> +    for (i=0; i < TBUCKS; i++) {
>> +        if (thash[i].ch.next == (void *)&thash[i].ch)
>> +            continue;    // quick return without lock
>> +
>> +        spin_lock_irqsave(&thash[i].lock, tflags);
>> +
>> +        tip = thash[i].ch.next;
>> +
>> +        /*
>> +        ** check all taskinfo structs chained to this bucket
>> +        */
>> +        while (tip != (void *)&thash[i].ch) {
>> +            switch (tip->state) {
>> +                /*
>> +                ** remove INDELETE tasks from the hash buckets
>> +                ** -- move thread group to exitlist
>> +                ** -- destroy thread right away
>> +                */
>> +               case INDELETE:
>> +                tipsave = tip->ch.next;
>> +
>> +                if (tip->type == 'g')
>> +                    move_taskinfo(tip);    // thread group
>> +                else
>> +                    delete_taskinfo(tip);    // thread
>> +
>> +                tip = tipsave;
>> +                break;
>> +
>> +               case CHECKED:
>> +                tip->state = 0;
>> +                tip = tip->ch.next;
>> +                break;
>> +
>> +               default:    // not checked yet
>> +                rcu_read_lock();
>> +                pid = find_vpid(tip->id);
>> +                rcu_read_unlock();
>> +
>> +                if (pid == NULL) {
>> +                    tipsave = tip->ch.next;
>> +
>> +                    if (tip->type == 'g')
>> +                        move_taskinfo(tip);
>> +                    else
>> +                        delete_taskinfo(tip);
>> +
>> +                    tip = tipsave;
>> +                } else {
>> +                    tip = tip->ch.next;
>> +                }
>> +            }
>> +        }
>> +
>> +        spin_unlock_irqrestore(&thash[i].lock, tflags);
>> +    }
>> +}
>> +
>> +
>> +/*
>> +** remove all sockinfo structs
>> +*/
>> +static void
>> +wipesockinfo()
>> +{
>> +    struct sockinfo    *sip, *sipsave;
>> +    int         i;
>> +    unsigned long    sflags;
>> +
>> +    for (i=0; i < SBUCKS; i++) {
>> +        spin_lock_irqsave(&shash[i].lock, sflags);
>> +
>> +        sip = shash[i].ch.next;
>> +
>> +        /*
>> +        ** free all structs chained in one bucket
>> +        */
>> +        while (sip != (void *)&shash[i].ch) {
>> +            sipsave = sip->ch.next;
>> +            delete_sockinfo(sip);
>> +            sip = sipsave;
>> +        }
>> +
>> +        spin_unlock_irqrestore(&shash[i].lock, sflags);
>> +    }
>> +}
>> +
>> +/*
>> +** remove all taskinfo structs from hash list
>> +*/
>> +static void
>> +wipetaskinfo()
>> +{
>> +    struct taskinfo    *tip, *tipsave;
>> +    int         i;
>> +    unsigned long    tflags;
>> +
>> +    for (i=0; i < TBUCKS; i++) {
>> +        spin_lock_irqsave(&thash[i].lock, tflags);
>> +
>> +        tip = thash[i].ch.next;
>> +
>> +        /*
>> +        ** free all structs chained in one bucket
>> +        */
>> +        while (tip != (void *)&thash[i].ch) {
>> +            tipsave = tip->ch.next;
>> +            delete_taskinfo(tip);
>> +            tip = tipsave;
>> +        }
>> +
>> +        spin_unlock_irqrestore(&thash[i].lock, tflags);
>> +    }
>> +}
>> +
>> +/*
>> +** remove all taskinfo structs from exit list
>> +*/
>> +static void
>> +wipetaskexit()
>> +{
>> +    gctaskexit();
>> +}
>> +
>> +/*
>> +** move one taskinfo struct from hash bucket to exitlist
>> +*/
>> +static void
>> +move_taskinfo(struct taskinfo *tip)
>> +{
>> +    unsigned long     flags;
>> +
>> +    /*
>> +    ** remove from hash list
>> +    */
>> +    ((struct taskinfo *)tip->ch.next)->ch.prev = tip->ch.prev;
>> +    ((struct taskinfo *)tip->ch.prev)->ch.next = tip->ch.next;
>> +
>> +    spin_lock_irqsave(&nrtlock, flags);
>> +    nrt--;
>> +    spin_unlock_irqrestore(&nrtlock, flags);
>> +
>> +    /*
>> +    ** add to exitlist
>> +    */
>> +    tip->ch.next     = NULL;
>> +    tip->state       = FINISHED;
>> +    tip->exittime    = jiffies_64;
>> +
>> +    spin_lock_irqsave(&exitlock, flags);
>> +
>> +    if (exittail) {            // list filled?
>> +        exittail->ch.next = tip;
>> +        exittail          = tip;
>> +    } else {            // list empty
>> +        exithead = exittail = tip;
>> +    }
>> +
>> +    nre++;
>> +
>> +    wake_up_interruptible(&exitlist_filled);
>> +
>> +    spin_unlock_irqrestore(&exitlock, flags);
>> +}
>> +
>> +/*
>> +** remove one taskinfo struct for the hash bucket chain
>> +*/
>> +static void
>> +delete_taskinfo(struct taskinfo *tip)
>> +{
>> +    unsigned long    flags;
>> +
>> +    ((struct taskinfo *)tip->ch.next)->ch.prev = tip->ch.prev;
>> +    ((struct taskinfo *)tip->ch.prev)->ch.next = tip->ch.next;
>> +
>> +    kmem_cache_free(ticache, tip);
>> +
>> +    spin_lock_irqsave(&nrtlock, flags);
>> +    nrt--;
>> +    spin_unlock_irqrestore(&nrtlock, flags);
>> +}
>> +
>> +/*
>> +** remove one sockinfo struct for the hash bucket chain
>> +*/
>> +static void
>> +delete_sockinfo(struct sockinfo *sip)
>> +{
>> +    unsigned long    flags;
>> +
>> +    ((struct sockinfo *)sip->ch.next)->ch.prev = sip->ch.prev;
>> +    ((struct sockinfo *)sip->ch.prev)->ch.next = sip->ch.next;
>> +
>> +    kmem_cache_free(sicache, sip);
>> +
>> +    spin_lock_irqsave(&nrslock, flags);
>> +    nrs--;
>> +    spin_unlock_irqrestore(&nrslock, flags);
>> +}
>> +
>> +/*
>> +** read function for /proc/netatop
>> +*/
>> +static int
>> +netatop_show(struct seq_file *m, void *v)
>> +{
>> +    seq_printf(m, "tcpsndpacks:  %12lu (unident: %9lu)\n"
>> +                        "tcprcvpacks:  %12lu (unident: %9lu)\n"
>> +                        "udpsndpacks:  %12lu (unident: %9lu)\n"
>> +                        "udprcvpacks:  %12lu (unident: %9lu)\n\n"
>> +                        "icmpsndpacks: %12lu\n"
>> +                        "icmprcvpacks: %12lu\n\n"
>> +                        "#sockinfo:    %12lu (overflow: %8lu)\n"
>> +                        "#taskinfo:    %12lu (overflow: %8lu)\n"
>> +                        "#taskexit:    %12lu\n\n"
>> +                        "modversion: %14s\n",
>> +                tcpsndpacks,  unidenttcpsndpacks,
>> +                tcprcvpacks,  unidenttcprcvpacks,
>> +                udpsndpacks,  unidentudpsndpacks,
>> +                udprcvpacks,  unidentudprcvpacks,
>> +                icmpsndpacks, icmprcvpacks,
>> +                 nrs,          nrs_ovf,
>> +                nrt,          nrt_ovf,
>> +                nre,          NETATOPVERSION);
>> +    return 0;
>> +}
>> +
>> +static int
>> +netatop_open(struct inode *inode, struct file *file)
>> +{
>> +    return single_open(file, netatop_show, NULL);
>> +}
>> +
>> +/*
>> +** called when user spce issues system call getsockopt()
>> +*/
>> +static int
>> +getsockopt(struct sock *sk, int cmd, void __user *user, int *len)
>> +{
>> +    int            bt;
>> +    struct taskinfo        *tip;
>> +    char            tasktype = 't';
>> +    struct netpertask    npt;
>> +    unsigned long        tflags;
>> +
>> +    /*
>> +    ** verify the proper privileges
>> +    */
>> +    if (!capable(CAP_NET_ADMIN))
>> +        return -EPERM;
>> +
>> +    /*
>> +    ** react on command
>> +    */
>> +    switch (cmd) {
>> +        case NETATOP_PROBE:
>> +        break;
>> +
>> +        case NETATOP_FORCE_GC:
>> +        garbage_collector();
>> +        break;
>> +
>> +        case NETATOP_EMPTY_EXIT:
>> +        while (nre > 0) {
>> +            if (wait_event_interruptible(exitlist_empty, nre == 0))
>> +                return -ERESTARTSYS;
>> +        }
>> +        break;
>> +
>> +        case NETATOP_GETCNT_EXIT:
>> +        if (nre == 0)
>> +            wake_up_interruptible(&exitlist_empty);
>> +
>> +        if (*len < sizeof(pid_t))
>> +            return -EINVAL;
>> +
>> +        if (*len > sizeof npt)
>> +            *len = sizeof npt;
>> +
>> +        spin_lock_irqsave(&exitlock, tflags);
>> +
>> +        /*
>> +        ** check if an exited process is present
>> +        ** if not, wait for it...
>> +        */
>> +        while (nre == 0) {
>> +            spin_unlock_irqrestore(&exitlock, tflags);
>> +
>> +            if ( wait_event_interruptible(exitlist_filled, nre > 0))
>> +                return -ERESTARTSYS;
>> +
>> +            spin_lock_irqsave(&exitlock, tflags);
>> +        }
>> +
>> +        /*
>> +        ** get first eprocess from exitlist and remove it from there
>> +        */
>> +        tip = exithead;
>> +
>> +        if ( (exithead = tip->ch.next) == NULL)
>> +            exittail = NULL;
>> +
>> +        nre--;
>> +
>> +        spin_unlock_irqrestore(&exitlock, tflags);
>> +
>> +        /*
>> +        ** pass relevant info to user mode
>> +        ** and free taskinfo struct
>> +        */
>> +        npt.id        = tip->id;
>> +        npt.tc        = tip->tc;
>> +        npt.btime    = tip->btime;
>> +        memcpy(npt.command, tip->command, COMLEN);
>> +
>> +        if (copy_to_user(user, &npt, *len) != 0)
>> +            return -EFAULT;
>> +
>> +        kmem_cache_free(ticache, tip);
>> +
>> +        return 0;
>> +
>> +        case NETATOP_GETCNT_TGID:
>> +        tasktype = 'g';
>> +
>> +        case NETATOP_GETCNT_PID:
>> +        if (*len < sizeof(pid_t))
>> +            return -EINVAL;
>> +
>> +        if (*len > sizeof npt)
>> +            *len = sizeof npt;
>> +
>> +        if (copy_from_user(&npt, user, *len) != 0)
>> +            return -EFAULT;
>> +
>> +        /*
>> +        ** search requested id in taskinfo hash
>> +        */
>> +        bt = THASH(npt.id, tasktype);    // calculate hash
>> +
>> +        if (thash[bt].ch.next == (void *)&thash[bt].ch)
>> +            return -ESRCH;        // quick return without lock
>> +
>> +        spin_lock_irqsave(&thash[bt].lock, tflags);
>> +
>> +        tip = thash[bt].ch.next;
>> +
>> +        while (tip != (void *)&thash[bt].ch) {
>> +            // is this the one?
>> +            if (tip->id == npt.id && tip->type == tasktype) {
>> +                /*
>> +                ** found: copy results to user space
>> +                */
>> +                memcpy(npt.command, tip->command, COMLEN);
>> +                npt.tc    = tip->tc;
>> +                npt.btime = tip->btime;
>> +
>> +                spin_unlock_irqrestore(&thash[bt].lock, tflags);
>> +
>> +                if (copy_to_user(user, &npt, *len) != 0)
>> +                    return -EFAULT;
>> +                else
>> +                    return 0;
>> +            }
>> +
>> +            tip = tip->ch.next;
>> +        }
>> +
>> +        spin_unlock_irqrestore(&thash[bt].lock, tflags);
>> +        return -ESRCH;
>> +
>> +       default:
>> +        printk(KERN_INFO "unknown getsockopt command %d\n", cmd);
>> +        return -EINVAL;
>> +    }
>> +
>> +    return 0;
>> +}
>> +
>> +/*
>> +** kernel mode thread: initiate garbage collection every N seconds
>> +*/
>> +static int netatop_thread(void *dummy)
>> +{
>> +    while (!kthread_should_stop()) {
>> +        /*
>> +         ** do garbage collection
>> +        */
>> +        garbage_collector();
>> +
>> +        /*
>> +        ** wait a while
>> +        */
>> +        (void) schedule_timeout_interruptible(GCINTERVAL);
>> +    }
>> +
>> +    return 0;
>> +}
>> +
>> +#if LINUX_VERSION_CODE >= KERNEL_VERSION(4, 13, 0)
>> +static int __net_init netatop_nf_register(struct net *net)
>> +{
>> +    int err;
>> +
>> +    err = nf_register_net_hooks(net, &hookin_ipv4,  1);
>> +    if (err)
>> +        return err;
>> +
>> +    err = nf_register_net_hooks(net, &hookout_ipv4, 1);
>> +        if (err)
>> +        nf_unregister_net_hooks(net, &hookin_ipv4,  1);
>> +
>> +    return err;
>> +}
>> +
>> +static void __net_exit netatop_nf_unregister(struct net *net)
>> +{
>> +        nf_unregister_net_hooks(net, &hookin_ipv4,  1);
>> +        nf_unregister_net_hooks(net, &hookout_ipv4, 1);
>> +}
>> +
>> +static struct pernet_operations netatop_net_ops = {
>> +        .init = netatop_nf_register,
>> +        .exit = netatop_nf_unregister,
>> +};
>> +#endif
>> +
>> +/*
>> +** called when module loaded
>> +*/
>> +int
>> +init_module()
>> +{
>> +    int i;
>> +
>> +    /*
>> +    ** initialize caches for taskinfo and sockinfo
>> +    */
>> +    ticache = kmem_cache_create("Netatop_taskinfo",
>> +                    sizeof (struct taskinfo), 0, 0, NULL);
>> +    if (!ticache)
>> +        return -EFAULT;
>> +
>> +    sicache = kmem_cache_create("Netatop_sockinfo",
>> +                    sizeof (struct sockinfo), 0, 0, NULL);
>> +    if (!sicache) {
>> +        kmem_cache_destroy(ticache);
>> +        return -EFAULT;
>> +    }
>> +
>> +    /*
>> +    ** initialize hash table for taskinfo and sockinfo
>> +    */
>> +    for (i=0; i < TBUCKS; i++) {
>> +        thash[i].ch.next = &thash[i].ch;
>> +        thash[i].ch.prev = &thash[i].ch;
>> +        spin_lock_init(&thash[i].lock);
>> +    }
>> +
>> +    for (i=0; i < SBUCKS; i++) {
>> +        shash[i].ch.next = &shash[i].ch;
>> +        shash[i].ch.prev = &shash[i].ch;
>> +        spin_lock_init(&shash[i].lock);
>> +    }
>> +
>> +    getboottime(&boottime);
>> +
>> +    /*
>> +    ** register getsockopt for user space communication
>> +    */
>> +    if (nf_register_sockopt(&sockopts) < 0) {
>> +        kmem_cache_destroy(ticache);
>> +        kmem_cache_destroy(sicache);
>> +        return -1;
>> +    }
>> +
>> +    /*
>> +     ** create a new kernel mode thread for time-driven garbage 
>> collection
>> +    ** after creation, the thread waits until it is woken up
>> +    */
>> +    knetatop_task = kthread_create(netatop_thread, NULL, "knetatop");
>> +
>> +    if (IS_ERR(knetatop_task)) {
>> +        nf_unregister_sockopt(&sockopts);
>> +        kmem_cache_destroy(ticache);
>> +        kmem_cache_destroy(sicache);
>> +        return -1;
>> +    }
>> +
>> +    /*
>> +    ** prepare hooks for input and output packets
>> +    */
>> +    hookin_ipv4.hooknum    = NF_IP_LOCAL_IN;    // input packs
>> +    hookin_ipv4.hook    = ipv4_hookin;        // func to call
>> +    hookin_ipv4.pf        = PF_INET;        // IPV4 packets
>> +    hookin_ipv4.priority    = NF_IP_PRI_FIRST;    // highest prio
>> +
>> +    hookout_ipv4.hooknum    = NF_IP_LOCAL_OUT;    // output packs
>> +    hookout_ipv4.hook    = ipv4_hookout;        // func to call
>> +    hookout_ipv4.pf        = PF_INET;        // IPV4 packets
>> +    hookout_ipv4.priority    = NF_IP_PRI_FIRST;    // highest prio
>> +
>> +#if LINUX_VERSION_CODE >= KERNEL_VERSION(4, 13, 0)
>> +    i = register_pernet_subsys(&netatop_net_ops);
>> +    if (i)
>> +        return i;
>> +#else
>> +    nf_register_hook(&hookin_ipv4);            // register hook
>> +    nf_register_hook(&hookout_ipv4);        // register hook
>> +#endif
>> +
>> +    /*
>> +    ** create a /proc-entry to produce status-info on request
>> +    */
>> +    proc_create("netatop", 0444, NULL, &netatop_proc_fops);
>> +
>> +    /*
>> +     ** all admi prepared; kick off kernel mode thread
>> +    */
>> +    wake_up_process(knetatop_task);
>> +
>> +    return 0;        // return success
>> +}
>> +
>> +/*
>> +** called when module unloaded
>> +*/
>> +void
>> +cleanup_module()
>> +{
>> +    /*
>> +     ** tell kernel daemon to stop
>> +    */
>> +    kthread_stop(knetatop_task);
>> +
>> +    /*
>> +     ** unregister netfilter hooks and other miscellaneous stuff
>> +    */
>> +#if LINUX_VERSION_CODE >= KERNEL_VERSION(4, 13, 0)
>> +    unregister_pernet_subsys(&netatop_net_ops);
>> +#else
>> +    nf_unregister_hook(&hookin_ipv4);
>> +    nf_unregister_hook(&hookout_ipv4);
>> +#endif
>> +
>> +    remove_proc_entry("netatop", NULL);
>> +
>> +    nf_unregister_sockopt(&sockopts);
>> +
>> +    /*
>> +    ** destroy allocated stats
>> +    */
>> +    wipesockinfo();
>> +    wipetaskinfo();
>> +    wipetaskexit();
>> +
>> +    /*
>> +     ** destroy caches
>> +    */
>> +    kmem_cache_destroy(ticache);
>> +    kmem_cache_destroy(sicache);
>> +}
>> +
>> +module_init(init_module);
>> +module_exit(cleanup_module);
>> diff --git a/drivers/staging/netatop/netatop.h 
>> b/drivers/staging/netatop/netatop.h
>> new file mode 100644
>> index 000000000000..a14e6fa5db52
>> --- /dev/null
>> +++ b/drivers/staging/netatop/netatop.h
>> @@ -0,0 +1,47 @@
>> +#define    COMLEN    16
>> +
>> +struct taskcount {
>> +    unsigned long long    tcpsndpacks;
>> +    unsigned long long    tcpsndbytes;
>> +    unsigned long long    tcprcvpacks;
>> +    unsigned long long    tcprcvbytes;
>> +
>> +    unsigned long long    udpsndpacks;
>> +    unsigned long long    udpsndbytes;
>> +    unsigned long long    udprcvpacks;
>> +    unsigned long long    udprcvbytes;
>> +
>> +    /* space for future extensions */
>> +};
>> +
>> +struct netpertask {
>> +    pid_t            id;    // tgid or tid (depending on command)
>> +    unsigned long        btime;
>> +    char            command[COMLEN];
>> +
>> +    struct taskcount    tc;
>> +};
>> +
>> +
>> +/*
>> +** getsocktop commands
>> +*/
>> +#define NETATOP_BASE_CTL       15661
>> +
>> +// just probe if the netatop module is active
>> +#define NETATOP_PROBE        (NETATOP_BASE_CTL)
>> +
>> +// force garbage collection to make finished processes available
>> +#define NETATOP_FORCE_GC    (NETATOP_BASE_CTL+1)
>> +
>> +// wait until all finished processes are read (blocks until done)
>> +#define NETATOP_EMPTY_EXIT    (NETATOP_BASE_CTL+2)
>> +
>> +// get info for finished process (blocks until available)
>> +#define NETATOP_GETCNT_EXIT    (NETATOP_BASE_CTL+3)
>> +
>> +// get counters for thread group (i.e. process):  input is 'id' (pid)
>> +#define NETATOP_GETCNT_TGID    (NETATOP_BASE_CTL+4)
>> +
>> +// get counters for thread:  input is 'id' (tid)
>> +#define NETATOP_GETCNT_PID     (NETATOP_BASE_CTL+5)
>> diff --git a/drivers/staging/netatop/netatopversion.h 
>> b/drivers/staging/netatop/netatopversion.h
>> new file mode 100644
>> index 000000000000..eb011f5a13c9
>> --- /dev/null
>> +++ b/drivers/staging/netatop/netatopversion.h
>> @@ -0,0 +1,2 @@
>> +#define    NETATOPVERSION    "2.0"
>> +#define    NETATOPDATE    "2017/09/30 12:06:19"
>>
> 



More information about the linux-yocto mailing list