# pfiles.stp
# Copyright (C) 2007-2010, 2012, 2015-2019 Red Hat, Inc., Eugene Teo
# <eteo@redhat.com>
#
# 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.

# Example:
# $ stap -g pfiles.stp `pgrep udevd`
# OR
# $ stap -g pfiles.stp -x `pgrep udevd`

#
# pfiles.stp report information for all open files by the process id.
#
# pfiles.stp is not a port of Solaris' pfiles tool. It was written based
# on the example outputs in:
# - https://bugzilla.redhat.com/show_bug.cgi?id=223489
# - http://blogs.sun.com/peteh/entry/pfiles_1_locked_files_and
# - http://lists.samba.org/archive/samba-technical/2001-May/014293.html
# - http://mail.gnome.org/archives/gconf-list/2004-December/msg00005.html
#
# pfiles.stp is able to:
# - report locked open files
# - report pathname information
# - report socket information (thanks Luis Henriques)
# - report sk_userlocks (thanks Arnaldo Carvalho de Melo)
# - report socket options
#
# Example:
# $ stap -g pfiles.stp `pgrep udevd`
#    787: udevd
#   Current rlimit: 32 file descriptors
#    0: S_IFCHR mode:0666 dev:0,15 ino:396 uid:0 gid:0 rdev:1,3
#       O_RDWR|O_LARGEFILE 
#       /dev/null
#    1: S_IFCHR mode:0666 dev:0,15 ino:396 uid:0 gid:0 rdev:1,3
#       O_RDWR|O_LARGEFILE 
#       /dev/null
#    2: S_IFCHR mode:0666 dev:0,15 ino:396 uid:0 gid:0 rdev:1,3
#       O_RDWR|O_LARGEFILE 
#       /dev/null
#    3: S_IFDIR mode:0600 dev:0,9 ino:1 uid:0 gid:0 rdev:0,0
#       O_RDONLY 
#       inotify
#    4: S_IFSOCK mode:0777 dev:0,4 ino:2353 uid:0 gid:0 rdev:0,0
#       O_RDWR 
#       socket:[2353]
#       SO_PASSCRED,SO_TYPE(2),SO_SNDBUF(111616),SO_RCVBUF(111616)
#         sockname: AF_UNIX
#    5: S_IFSOCK mode:0777 dev:0,4 ino:2354 uid:0 gid:0 rdev:0,0
#       O_RDWR 
#       socket:[2354]
#       SO_TYPE(2),SO_SNDBUF(111616),SO_RCVBUF(33554432)
#         ulocks: rcv
#    6: S_IFIFO mode:0600 dev:0,6 ino:2355 uid:0 gid:0 rdev:0,0
#       O_RDONLY|O_NONBLOCK 
#       pipe:[2355]
#    7: S_IFIFO mode:0600 dev:0,6 ino:2355 uid:0 gid:0 rdev:0,0
#       O_WRONLY|O_NONBLOCK 
#       pipe:[2355]
#
# I used Andrew Tridgell's locktst.c to test the advisory lock code.
# You can download it at http://samba.org/ftp/unpacked/junkcode/locktst.c
#
# $ ./locktst file &
# fcntl_lock 3 6 1 1 1
# $ pfiles 10126 | grep advisory -A1 -B2
#   3: S_IFREG mode:0644 dev:253,0 ino:15237159 uid:500 gid:500 rdev:0,0
#      O_RDWR 
#      advisory write lock set by process 10126
#      /home/eteo/pfiles/file
#

%{
#include <linux/file.h>
#include <net/sock.h>
#include <linux/un.h>
#include <linux/tcp.h>
#ifdef CONFIG_USER_NS
#include <linux/user_namespace.h>
#endif
#ifdef STAPCONF_LINUX_UIDGID_H
#include <linux/uidgid.h>
#endif
%}

# XXX: For the time being, mark this example as not working on PREEMPT_RT kernels:
%{
#if defined(CONFIG_PREEMPT_RT_FULL) || defined(CONFIG_PREEMPT_RT)
#error "pfiles.stp not supported on PREEMPT_RT kernels"
#endif
%}

function task_valid_file_handle:long (task:long, fd:long) %{ /* pure */
	struct task_struct *p = (struct task_struct *)((long)STAP_ARG_task);
	struct files_struct *files;
	struct file *filp;

	rcu_read_lock();
	if ((files = kread(&p->files))) {
		filp = fcheck_files(files, STAP_ARG_fd);
		STAP_RETVALUE = !!filp;
	}

	CATCH_DEREF_FAULT();
	rcu_read_unlock();
%}

function i_mode2str:string (i_mode:long) {
	if (%{ S_ISLNK(STAP_ARG_i_mode) %})
		return "S_IFLNK"
	if (%{ S_ISREG(STAP_ARG_i_mode) %})
		return "S_IFREG"
	if (%{ S_ISDIR(STAP_ARG_i_mode) %})
		return "S_IFDIR"
	if (%{ S_ISCHR(STAP_ARG_i_mode) %})
		return "S_IFCHR"
	if (%{ S_ISBLK(STAP_ARG_i_mode) %})
		return "S_IFBLK"
	if (%{ S_ISFIFO(STAP_ARG_i_mode) %})
		return "S_IFIFO"
	if (%{ S_ISSOCK(STAP_ARG_i_mode) %})
		return "S_IFSOCK"
}

function task_file_handle_i_mode:long (task:long, fd:long) %{ /* pure */
	struct task_struct *p = (struct task_struct *)((long)STAP_ARG_task);
	struct files_struct *files;
	struct file *filp;
	struct dentry *dentry;
	struct inode *inode;

	rcu_read_lock();
	if ((files = kread(&p->files)) &&
	    (filp = fcheck_files(files, STAP_ARG_fd)) &&
#ifdef STAPCONF_DPATH_PATH
	    (dentry = kread(&filp->f_path.dentry)) &&
#else
	    (dentry = kread(&filp->f_dentry)) &&
#endif
	    (inode = kread(&dentry->d_inode))) {
		STAP_RETVALUE = kread(&inode->i_mode);
	}

	CATCH_DEREF_FAULT();
	rcu_read_unlock();
%}

function task_file_handle_majmin_dev:string (task:long, fd:long) %{ /* pure */
	struct task_struct *p = (struct task_struct *)((long)STAP_ARG_task);
	struct files_struct *files;
	struct file *filp;
	struct dentry *dentry;
	struct inode *inode;
	struct super_block *sb;
	int dev_nr;

	rcu_read_lock();
	if ((files = kread(&p->files)) &&
	    (filp = fcheck_files(files, STAP_ARG_fd)) &&
#ifdef STAPCONF_DPATH_PATH
	    (dentry = kread(&filp->f_path.dentry)) &&
#else
	    (dentry = kread(&filp->f_dentry)) &&
#endif
	    (inode = kread(&dentry->d_inode)) &&
	    (sb = kread(&inode->i_sb))) {
		dev_nr = kread(&sb->s_dev);
		snprintf(STAP_RETVALUE, MAXSTRINGLEN,
			"%d,%d", MAJOR(dev_nr), MINOR(dev_nr));
	}

	CATCH_DEREF_FAULT();
	rcu_read_unlock();
%}

function task_file_handle_majmin_rdev:string (task:long, fd:long) %{ /* pure */
	struct task_struct *p = (struct task_struct *)((long)STAP_ARG_task);
	struct files_struct *files;
	struct file *filp;
	struct dentry *dentry;
	struct inode *inode;
	int rdev_nr;

	rcu_read_lock();
	if ((files = kread(&p->files)) &&
	    (filp = fcheck_files(files, STAP_ARG_fd)) &&
#ifdef STAPCONF_DPATH_PATH
	    (dentry = kread(&filp->f_path.dentry)) &&
#else
	    (dentry = kread(&filp->f_dentry)) &&
#endif
	    (inode = kread(&dentry->d_inode))) {
		rdev_nr = kread(&inode->i_rdev);
		snprintf(STAP_RETVALUE, MAXSTRINGLEN,
			"%d,%d", MAJOR(rdev_nr), MINOR(rdev_nr));
	}

	CATCH_DEREF_FAULT();
	rcu_read_unlock();
%}

function task_file_handle_i_node:long (task:long, fd:long) %{ /* pure */
	struct task_struct *p = (struct task_struct *)((long)STAP_ARG_task);
	struct files_struct *files;
	struct file *filp;
	struct dentry *dentry;
	struct inode *inode;

	rcu_read_lock();
	if ((files = kread(&p->files)) &&
	    (filp = fcheck_files(files, STAP_ARG_fd)) &&
#ifdef STAPCONF_DPATH_PATH
	    (dentry = kread(&filp->f_path.dentry)) &&
#else
	    (dentry = kread(&filp->f_dentry)) &&
#endif
	    (inode = kread(&dentry->d_inode))) {
		STAP_RETVALUE = kread(&inode->i_ino);
	}

	CATCH_DEREF_FAULT();
	rcu_read_unlock();
%}

function task_file_handle_uid:long (task:long, fd:long) %{ /* pure */
	struct task_struct *p = (struct task_struct *)((long)STAP_ARG_task);
	struct files_struct *files;
	struct file *filp;
#ifdef CONFIG_USER_NS
	struct user_namespace *ns = NULL;
#endif

	rcu_read_lock();
	if ((files = kread(&p->files)) &&
	    (filp = fcheck_files(files, STAP_ARG_fd))) {
#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,29)
		/* git commit d76b0d9b */
		const struct cred *cred;
		if ((cred = kread(&filp->f_cred))) {
#ifdef CONFIG_USER_NS
#if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 29)
			ns = p->nsproxy->user_ns;
#else
#if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 39)
			ns = (task_cred_xxx(p, user))->user_ns;
#else /* LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 39) */
			ns = task_cred_xxx(p, user_ns);
#endif
#endif
			if (ns) {
				get_user_ns(ns);
				// We call kderef_buffer() here to
				// ensure the memory at the kuid_t
				// location is valid to read. We can't
				// use kderef()/kread(), since they
				// only handle data with a size of 1,
				// 2, 4, or 8 bytes.
				kderef_buffer(NULL,  &cred->fsuid,
					      sizeof(cred->fsuid));
				STAP_RETVALUE = from_kuid_munged(ns,
								 cred->fsuid);
			}
			else
				STAP_RETVALUE = -1;

#else	/* ! CONFIG_USER_NS */
			STAP_RETVALUE = kread(&cred->fsuid);
#endif	/* ! CONFIG_USER_NS */
		}
#else
		STAP_RETVALUE = kread(&filp->f_uid);
#endif
	}

	CATCH_DEREF_FAULT();
#ifdef CONFIG_USER_NS
	if (ns)
		put_user_ns(ns);
#endif
	rcu_read_unlock();
%}

function task_file_handle_gid:long (task:long, fd:long) %{ /* pure */
	struct task_struct *p = (struct task_struct *)((long)STAP_ARG_task);
	struct files_struct *files;
	struct file *filp;
#ifdef CONFIG_USER_NS
	struct user_namespace *ns = NULL;
#endif

	rcu_read_lock();
	if ((files = kread(&p->files)) &&
	    (filp = fcheck_files(files, STAP_ARG_fd))) {
#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,29)
		/* git commit d76b0d9b */
		const struct cred *cred;
		if ((cred = kread(&filp->f_cred))) {
#ifdef CONFIG_USER_NS
#if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 29)
			ns = p->nsproxy->user_ns;
#else
#if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 39)
			ns = (task_cred_xxx(p, user))->user_ns;
#else /* LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 39) */
			ns = task_cred_xxx(p, user_ns);
#endif
#endif
			if (ns) {
				get_user_ns(ns);
				// We call kderef_buffer() here to
				// ensure the memory at the kgid_t
				// location is valid to read. We can't
				// use kderef()/kread(), since they
				// only handle data with a size of 1,
				// 2, 4, or 8 bytes.
				kderef_buffer(NULL,  &cred->fsgid,
					      sizeof(cred->fsgid));
				STAP_RETVALUE = from_kgid_munged(ns,
								 cred->fsgid);
			}
			else
				STAP_RETVALUE = -1;

#else	/* ! CONFIG_USER_NS */
			STAP_RETVALUE = kread(&cred->fsgid);
#endif	/* ! CONFIG_USER_NS */
		}
#else
		STAP_RETVALUE = kread(&filp->f_gid);
#endif
	}

	CATCH_DEREF_FAULT();
#ifdef CONFIG_USER_NS
	if (ns)
		put_user_ns(ns);
#endif
	rcu_read_unlock();
%}

function task_file_handle_f_flags:long (task:long, fd:long) %{ /* pure */
	struct task_struct *p = (struct task_struct *)((long)STAP_ARG_task);
	struct files_struct *files;
	struct file *filp;

	rcu_read_lock();
	if ((files = kread(&p->files)) &&
	    (filp = fcheck_files(files, STAP_ARG_fd))) {
		STAP_RETVALUE = kread(&filp->f_flags);
	}

	CATCH_DEREF_FAULT();
	rcu_read_unlock();
%}

function task_file_handle_fd_flags:string (task:long, fd:long) %{ /* pure */
	struct task_struct *p = (struct task_struct *)((long)STAP_ARG_task);
	struct files_struct *files;
	struct fdtable *fdt;
	int gcoe;

	rcu_read_lock();
	if ((files = kread(&p->files)) &&
	    (fdt = files_fdtable(files))) {
		/* Use test_bit directly instead of FD_ISSET which might
		   not be defined on some kernels.  */
		unsigned long *bs = (unsigned long *)kread(&fdt->close_on_exec);
		gcoe = test_bit(STAP_ARG_fd, bs);
		snprintf(STAP_RETVALUE, MAXSTRINGLEN,
			"%s", gcoe ? "FD_CLOEXEC" : "");
	}

	CATCH_DEREF_FAULT();
	rcu_read_unlock();
%}

function task_file_handle_flock:string (task:long, fd:long) %{ /* pure */
	struct task_struct *p = (struct task_struct *)((long)STAP_ARG_task);
	struct files_struct *files;
	struct file_lock *flock;
	int fl_type = -1;
	int fl_pid = -1;
	struct file *filp;
	struct dentry *dentry;
	struct inode *inode;
#if LINUX_VERSION_CODE >= KERNEL_VERSION(4,0,0)
	struct file_lock_context *ctx;
	int flc_lock_set = 0;
#endif

	rcu_read_lock();
	if ((files = kread(&p->files)) &&
	    (filp = fcheck_files(files, STAP_ARG_fd)) &&
#ifdef STAPCONF_DPATH_PATH
	    (dentry = kread(&filp->f_path.dentry)) &&
#else
	    (dentry = kread(&filp->f_dentry)) &&
#endif
	    (inode = kread(&dentry->d_inode))) {
// In kernels >= 4.0, we've now got a list of file_lock structures.
#if LINUX_VERSION_CODE >= KERNEL_VERSION(4,0,0)
		if ((ctx = kread(&inode->i_flctx))) {
			struct file_lock *lock;

			// Make sure the file_lock_context struct is valid.
			kderef_buffer(NULL, ctx,
				      sizeof(struct file_lock_context));

			// We'll just look at the first lock.
			spin_lock(&ctx->flc_lock);
			flc_lock_set = 1;
			list_for_each_entry(lock, &ctx->flc_flock, fl_list) {
				fl_type = kread(&lock->fl_type);
				fl_pid = kread(&lock->fl_pid);
				break;
			}
			spin_unlock(&ctx->flc_lock);
			flc_lock_set = 0;
		}
#else
		if ((flock = kread(&inode->i_flock))) {
			fl_type = kread(&flock->fl_type);
			fl_pid = kread(&flock->fl_pid);
		}
#endif

		if (fl_type != -1) { /* !flock */
			snprintf(STAP_RETVALUE, MAXSTRINGLEN,
				"      advisory %s lock set by process %d",
				fl_type ? "write" : "read", fl_pid);
		} else {
			snprintf(STAP_RETVALUE, MAXSTRINGLEN, "NULL");
		}
	}

	CATCH_DEREF_FAULT();
#if LINUX_VERSION_CODE >= KERNEL_VERSION(4,0,0)
	if (flc_lock_set)
		spin_unlock(&ctx->flc_lock);
#endif
	rcu_read_unlock();
%}

function task_file_handle_d_path:string (task:long, fd:long) %{ /* pure */
	struct task_struct *p = (struct task_struct *)((long)STAP_ARG_task);
	struct files_struct *files;
	char *page = NULL;
	struct file *filp;
	struct dentry *dentry;
	struct vfsmount *vfsmnt;
	char *path = NULL;

	rcu_read_lock();
	if ((files = kread(&p->files)) &&
	    // We need GFP_ATOMIC since we're inside a lock so we
	    // can't sleep.
	    (page = (char *)__get_free_page(GFP_ATOMIC)) &&
	    (filp = fcheck_files(files, STAP_ARG_fd))) {

#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,26)
		/* git commit 9d1bc601 */
		path = d_path(&filp->f_path, page, PAGE_SIZE);
#else
		dentry = kread(&filp->f_dentry);
		vfsmnt = kread(&filp->f_vfsmnt);

		if (dentry && vfsmnt) {
			path = d_path(dentry, vfsmnt, page, PAGE_SIZE);
		}
#endif
		if (path && !IS_ERR(path)) {
			snprintf(STAP_RETVALUE, MAXSTRINGLEN, "%s", path);
		}
	}
	CATCH_DEREF_FAULT();

	if (page) free_page((unsigned long)page);

	rcu_read_unlock();
%}

function task_file_handle_socket:long (task:long, fd:long) %{ /* pure */
	struct task_struct *p = (struct task_struct *)((long)STAP_ARG_task);
	struct files_struct *files;
	struct file *filp;
	struct dentry *dentry;
	struct inode *inode;

	rcu_read_lock();
	if ((files = kread(&p->files)) &&
	    (filp = fcheck_files(files, STAP_ARG_fd)) &&
#ifdef STAPCONF_DPATH_PATH
	    (dentry = kread(&filp->f_path.dentry)) &&
#else
	    (dentry = kread(&filp->f_dentry)) &&
#endif
	    (inode = kread(&dentry->d_inode))) {
		if (S_ISSOCK(kread(&inode->i_mode)))
			STAP_RETVALUE = (long)SOCKET_I(inode);
	}

	CATCH_DEREF_FAULT();
	rcu_read_unlock();
%}

function socket_userlocks:string (sock:long) %{ /* pure */
	struct socket *sock = (struct socket *)((long)STAP_ARG_sock);
	struct sock *sk = (struct sock *)kread(&sock->sk);
	unsigned char locks = sk->sk_userlocks;
	int printed = 0;

	STAP_RETVALUE[0] = '\0';

	if (locks & SOCK_RCVBUF_LOCK)
		printed = snprintf(STAP_RETVALUE, MAXSTRINGLEN,
			"ulocks: rcv");

	if (locks & SOCK_SNDBUF_LOCK)
		snprintf(STAP_RETVALUE, MAXSTRINGLEN,
			"ulocks: %ssnd",
			printed ? "rcv, " : "");

	CATCH_DEREF_FAULT();
%}

function socket_optname:string (sock:long) %{ /* pure */
    struct socket *sock = (struct socket *)((long)STAP_ARG_sock);
    int optval;
    int optlen;
    int ret, a;
    char str[30];
    struct linger ling;
    struct optname_item {
        int optfam;
        int optnum;
        const char *optname;
        int show_val;
    } const *p;
    static const struct optname_item optname_entries[] = {
        { SOL_SOCKET, SO_DEBUG,     "SO_DEBUG",     0 },
        { SOL_SOCKET, SO_REUSEADDR, "SO_REUSEADDR", 0 },
        { SOL_SOCKET, SO_DONTROUTE, "SO_DONTROUTE", 0 },
        { SOL_SOCKET, SO_BROADCAST, "SO_BROADCAST", 0 },
        { SOL_SOCKET, SO_KEEPALIVE, "SO_KEEPALIVE", 0 },
        { SOL_SOCKET, SO_OOBINLINE, "SO_OOBINLINE", 0 },
        { SOL_SOCKET, SO_PASSCRED,  "SO_PASSCRED",  0 },
/*      { SOL_SOCKET, SO_SNDLOWAT,  "SO_SNDLOWAT",  0 },*/
        { SOL_SOCKET, SO_TYPE,      "SO_TYPE",      1 },
        { SOL_SOCKET, SO_ERROR,     "SO_ERROR",     1 },
        { SOL_SOCKET, SO_SNDBUF,    "SO_SNDBUF",    1 },
        { SOL_SOCKET, SO_RCVBUF,    "SO_RCVBUF",    1 },
        { SOL_SOCKET, SO_NO_CHECK,  "SO_NO_CHECK",  1 },
        { SOL_TCP,    TCP_NODELAY,  "TCP_NODELAY",  0 },
        { 0, 0, NULL, 0 },
    };

    for (p = optname_entries; p->optname; ++p) {
       optlen = sizeof(optval);
       ret = kernel_getsockopt(sock, p->optfam, p->optnum, (char *)&optval, &optlen);
       if (ret == 0 && optval != 0) {
           if (!p->show_val)
               snprintf(str, MAXSTRINGLEN, "%s,", p->optname);
           else
               snprintf(str, MAXSTRINGLEN, "%s(%d),", p->optname, optval);
           strcat(STAP_RETVALUE, str);
       }
    }

    optlen = sizeof(ling);
    ret = kernel_getsockopt(sock, SOL_SOCKET, SO_LINGER, (void *)&ling, &optlen);
    if (ret == 0 && ling.l_onoff != 0) {
        snprintf(str, MAXSTRINGLEN, "SO_LINGER(%d),", ling.l_linger);
        strcat(STAP_RETVALUE, str);
    }

    STAP_RETVALUE[strlen(STAP_RETVALUE) - 1] = '\0';
%}

function socket_family:long (sock:long) %{ /* pure */
	struct socket *sock = (struct socket *)((long)STAP_ARG_sock);
	const struct proto_ops *ops = kread(&sock->ops);
	STAP_RETVALUE = (long)kread(&ops->family);
	CATCH_DEREF_FAULT();
%}

function socket_unix_sockname:string (sock:long) %{ /* pure */
	struct socket *sock = (struct socket *)((long)STAP_ARG_sock);
	const struct proto_ops *ops = kread(&sock->ops);
	struct sockaddr_un un_addr;
	int err, len;

#if LINUX_VERSION_CODE >= KERNEL_VERSION(4,17,0)
	err = ops->getname(sock, (struct sockaddr *)(&un_addr), 0);
#else
	err = ops->getname(sock, (struct sockaddr *)(&un_addr), &len, 0);
#endif
//	err = kernel_getsockname(kread(&sock), (struct sockaddr *)(&un_addr), &len);
	if (!(err < 0)) {
		if (kread(&un_addr.sun_path[0]) != 0)
			snprintf(STAP_RETVALUE, MAXSTRINGLEN,
				"        sockname: AF_UNIX %s", un_addr.sun_path);
		else
			snprintf(STAP_RETVALUE, MAXSTRINGLEN, "        sockname: AF_UNIX");
	}
	CATCH_DEREF_FAULT();
%}

function socket_unix_peername:string (sock:long) %{ /* pure */
	struct socket *sock = (struct socket *)((long)STAP_ARG_sock);
	const struct proto_ops *ops = kread(&sock->ops);
	struct sockaddr_un un_addr;
	int err, len;

#if LINUX_VERSION_CODE >= KERNEL_VERSION(4,17,0)
	err = ops->getname(sock, (struct sockaddr *)(&un_addr), 1);
#else
	err = ops->getname(sock, (struct sockaddr *)(&un_addr), &len, 1);
#endif
//	err = kernel_getpeername(kread(&sock), (struct sockaddr *)(&un_addr), &len);
	if (!(err < 0)) {
		if (un_addr.sun_path[0] != 0)
			snprintf(STAP_RETVALUE, MAXSTRINGLEN,
				"        peername: AF_UNIX %s", un_addr.sun_path);
		else
			snprintf(STAP_RETVALUE, MAXSTRINGLEN, "        peername: AF_UNIX");
	}
	CATCH_DEREF_FAULT();
%}

%{
#ifndef NIPQUAD
#define NIPQUAD(addr) \
	((unsigned char *)&addr)[0], \
	((unsigned char *)&addr)[1], \
	((unsigned char *)&addr)[2], \
	((unsigned char *)&addr)[3]
#endif
#ifndef NIPQUAD_FMT
#define NIPQUAD_FMT "%u.%u.%u.%u"
#endif
%}

function socket_ipv4_sockname:string (sock:long) %{ /* pure */
	struct socket *sock = (struct socket *)((long)STAP_ARG_sock);
	const struct proto_ops *ops = kread(&sock->ops);
	struct sockaddr_in in_addr;
	__be32 addr, port;
	int err, len;

#if LINUX_VERSION_CODE >= KERNEL_VERSION(4,17,0)
	err = ops->getname (sock, (struct sockaddr*)(&in_addr), 0);
#else
	err = ops->getname (sock, (struct sockaddr*)(&in_addr), &len, 0);
#endif
	if (!(err < 0)) {
		addr = in_addr.sin_addr.s_addr;
		port = htons(in_addr.sin_port);
		snprintf(STAP_RETVALUE, MAXSTRINGLEN,
			"        sockname: AF_INET " NIPQUAD_FMT "  port: %d",
			NIPQUAD(addr), port);
	}
	CATCH_DEREF_FAULT();
%}

function socket_ipv4_peername:string (sock:long) %{ /* pure */
	struct socket *sock = (struct socket *)((long)STAP_ARG_sock);
	const struct proto_ops *ops = kread(&sock->ops);
	struct sockaddr_in in_addr;
	__be32 addr, port;
	int err, len;

#if LINUX_VERSION_CODE >= KERNEL_VERSION(4,17,0)
	err = ops->getname (sock, (struct sockaddr*)(&in_addr), 1);
#else
	err = ops->getname (sock, (struct sockaddr*)(&in_addr), &len, 1);
#endif
	if (!(err < 0)) {
		addr = in_addr.sin_addr.s_addr;
		port = htons(in_addr.sin_port);
		snprintf(STAP_RETVALUE, MAXSTRINGLEN,
			"        peername: AF_INET " NIPQUAD_FMT "  port: %d",
			NIPQUAD(addr), port);
	}
	CATCH_DEREF_FAULT();
%}

%{
#ifndef NIP6_FMT
#define NIP6_FMT "%04x:%04x:%04x:%04x:%04x:%04x:%04x:%04x"
#endif
#ifndef NIP6
#define NIP6(addr) \
    ntohs((addr).s6_addr16[0]), \
    ntohs((addr).s6_addr16[1]), \
    ntohs((addr).s6_addr16[2]), \
    ntohs((addr).s6_addr16[3]), \
    ntohs((addr).s6_addr16[4]), \
    ntohs((addr).s6_addr16[5]), \
    ntohs((addr).s6_addr16[6]), \
    ntohs((addr).s6_addr16[7])
#endif

%}

function socket_ipv6_sockname:string (sock:long) %{ /* pure */
	struct socket *sock = (struct socket *)((long)STAP_ARG_sock);
	const struct proto_ops *ops = kread(&sock->ops);
	struct sockaddr_in6 in_addr;
	__be32 port;
	int err, len;

#if LINUX_VERSION_CODE >= KERNEL_VERSION(4,17,0)
	err = ops->getname (sock, (struct sockaddr*)(&in_addr), 0);
#else
	err = ops->getname (sock, (struct sockaddr*)(&in_addr), &len, 0);
#endif
	if (!(err < 0)) {
		port = htons(in_addr.sin6_port);
		snprintf(STAP_RETVALUE, MAXSTRINGLEN,
			"        sockname: AF_INET6 " NIP6_FMT "  port: %d",
			NIP6(in_addr.sin6_addr), port);
	}
	CATCH_DEREF_FAULT();
%}

function socket_ipv6_peername:string (sock:long) %{ /* pure */
	struct socket *sock = (struct socket *)((long)STAP_ARG_sock);
	const struct proto_ops *ops = kread(&sock->ops);
	struct sockaddr_in6 in_addr;
	__be32 port;
	int err, len;

#if LINUX_VERSION_CODE >= KERNEL_VERSION(4,17,0)
	err = ops->getname (sock, (struct sockaddr*)(&in_addr), 1);
#else
	err = ops->getname (sock, (struct sockaddr*)(&in_addr), &len, 1);
#endif
	if (!(err < 0)) {
		port = htons(in_addr.sin6_port);
		snprintf(STAP_RETVALUE, MAXSTRINGLEN,
			"        peername: AF_INET6 " NIP6_FMT "  port: %d",
			NIP6(in_addr.sin6_addr), port);
	}
	CATCH_DEREF_FAULT();
%}

function print_file_handle_general(task, fd) {
	printf("%4d: %s mode:%04o dev:%s ino:%d uid:%d gid:%d rdev:%s\n",
		fd, i_mode2str(task_file_handle_i_mode(task, fd)),
		task_file_handle_i_mode(task, fd) & 0777,
		task_file_handle_majmin_dev(task, fd),
		task_file_handle_i_node(task, fd),
		task_file_handle_uid(task, fd),
		task_file_handle_gid(task, fd),
		task_file_handle_majmin_rdev(task, fd));
}

function print_file_handle_flags(task, fd) {
	printf("      %s %s\n",
		_sys_open_flag_str(task_file_handle_f_flags(task, fd)),
		task_file_handle_fd_flags(task, fd));
}

function print_file_handle_flock(task, fd) {
	flock = task_file_handle_flock(task, fd)
	if (!isinstr(flock, "NULL"))
		printf("%s\n", flock)
}

function print_file_handle_d_path(task, fd) {
	printf("      %s\n", task_file_handle_d_path(task, fd))
}

function print_socket_userlocks(sock) {
	slocks = socket_userlocks(sock)
	if (strlen(slocks) > 0)
		printf("        %s\n", slocks)
}

function print_unix_socket(sock) {
	sockname = socket_unix_sockname(sock)
	peername = socket_unix_peername(sock)
	printf("%s%s", strlen(sockname) > 0 ? sockname . "\n" : "",
		strlen(peername) > 0 ? peername . "\n" : "")
        try { # skip line in case of null pointers
	printf("        peercred pid: %d\n",
            @defined(@cast(sock, "socket", "kernel")->sk->sk_peer_pid) ?
	    @cast(sock, "socket", "kernel")->sk->sk_peer_pid->numbers[0]->nr :
	    @cast(sock, "socket", "kernel")->sk->sk_peercred->pid ); } catch { }
}

function print_ipv4_socket(sock) {
	sockname = socket_ipv4_sockname(sock)
	peername = socket_ipv4_peername(sock)
	printf("%s%s", strlen(sockname) > 0 ? sockname . "\n" : "",
		strlen(peername) > 0 ? peername . "\n" : "")
}

function print_ipv6_socket(sock) {
	sockname = socket_ipv6_sockname(sock)
	peername = socket_ipv6_peername(sock)
	printf("%s%s", strlen(sockname) > 0 ? sockname . "\n" : "",
		strlen(peername) > 0 ? peername . "\n" : "")
}

function print_socket_optname(sock) {
    str = socket_optname(sock)
    printf("      %s", strlen(str) > 0 ? str . "\n" : "")
}

probe begin {
%( $# < 1
        %? pid = target()
        %: pid = $1
%)
   task = pid2task(pid)
   if (task == 0) error (sprintf("Process-id %d is invalid, please provide valid PID as $1 or -x PID", pid))
   
	max_fds = task_max_file_handles(task)

	printf("%6d: %s\n", pid, pid2execname(pid))
	printf("  Current rlimit: %d file descriptors\n", max_fds)

	for (fd = 0; fd < max_fds; fd++) {
		if (task_valid_file_handle(task, fd)) {
			print_file_handle_general(task, fd)
			print_file_handle_flags(task, fd)
			print_file_handle_flock(task, fd)
			print_file_handle_d_path(task, fd)
			sock = task_file_handle_socket(task, fd)
			if (sock) {
				print_socket_optname(sock)
				fam = socket_family(sock)
				if (fam == %{ AF_UNIX %}) {
					print_unix_socket(sock)
				} else if (fam == %{ AF_INET %}) {
					print_ipv4_socket(sock)
				} else if (fam == %{ AF_INET6 %}) {
					print_ipv6_socket(sock)
				}
				print_socket_userlocks(sock)
			}
		}
	}

	exit()
}