NFSD: Remove NFSD_TCP kernel build option
[safe/jmp/linux-2.6] / fs / proc / base.c
index 1bd646d..c5e412a 100644 (file)
@@ -121,6 +121,10 @@ struct pid_entry {
        NOD(NAME, (S_IFREG|(MODE)),                     \
                NULL, &proc_info_file_operations,       \
                { .proc_read = &proc_##OTYPE } )
+#define ONE(NAME, MODE, OTYPE)                         \
+       NOD(NAME, (S_IFREG|(MODE)),                     \
+               NULL, &proc_single_file_operations,     \
+               { .proc_show = &proc_##OTYPE } )
 
 int maps_protect;
 EXPORT_SYMBOL(maps_protect);
@@ -149,7 +153,7 @@ static int get_nr_threads(struct task_struct *tsk)
        return count;
 }
 
-static int proc_cwd_link(struct inode *inode, struct dentry **dentry, struct vfsmount **mnt)
+static int proc_cwd_link(struct inode *inode, struct path *path)
 {
        struct task_struct *task = get_proc_task(inode);
        struct fs_struct *fs = NULL;
@@ -161,8 +165,8 @@ static int proc_cwd_link(struct inode *inode, struct dentry **dentry, struct vfs
        }
        if (fs) {
                read_lock(&fs->lock);
-               *mnt = mntget(fs->pwdmnt);
-               *dentry = dget(fs->pwd);
+               *path = fs->pwd;
+               path_get(&fs->pwd);
                read_unlock(&fs->lock);
                result = 0;
                put_fs_struct(fs);
@@ -170,7 +174,7 @@ static int proc_cwd_link(struct inode *inode, struct dentry **dentry, struct vfs
        return result;
 }
 
-static int proc_root_link(struct inode *inode, struct dentry **dentry, struct vfsmount **mnt)
+static int proc_root_link(struct inode *inode, struct path *path)
 {
        struct task_struct *task = get_proc_task(inode);
        struct fs_struct *fs = NULL;
@@ -182,8 +186,8 @@ static int proc_root_link(struct inode *inode, struct dentry **dentry, struct vf
        }
        if (fs) {
                read_lock(&fs->lock);
-               *mnt = mntget(fs->rootmnt);
-               *dentry = dget(fs->root);
+               *path = fs->root;
+               path_get(&fs->root);
                read_unlock(&fs->lock);
                result = 0;
                put_fs_struct(fs);
@@ -310,9 +314,12 @@ static int proc_pid_schedstat(struct task_struct *task, char *buffer)
 static int lstats_show_proc(struct seq_file *m, void *v)
 {
        int i;
-       struct task_struct *task = m->private;
-       seq_puts(m, "Latency Top version : v0.1\n");
+       struct inode *inode = m->private;
+       struct task_struct *task = get_proc_task(inode);
 
+       if (!task)
+               return -ESRCH;
+       seq_puts(m, "Latency Top version : v0.1\n");
        for (i = 0; i < 32; i++) {
                if (task->latency_record[i].backtrace[0]) {
                        int q;
@@ -337,32 +344,24 @@ static int lstats_show_proc(struct seq_file *m, void *v)
                }
 
        }
+       put_task_struct(task);
        return 0;
 }
 
 static int lstats_open(struct inode *inode, struct file *file)
 {
-       int ret;
-       struct seq_file *m;
-       struct task_struct *task = get_proc_task(inode);
-
-       ret = single_open(file, lstats_show_proc, NULL);
-       if (!ret) {
-               m = file->private_data;
-               m->private = task;
-       }
-       return ret;
+       return single_open(file, lstats_show_proc, inode);
 }
 
 static ssize_t lstats_write(struct file *file, const char __user *buf,
                            size_t count, loff_t *offs)
 {
-       struct seq_file *m;
-       struct task_struct *task;
+       struct task_struct *task = get_proc_task(file->f_dentry->d_inode);
 
-       m = file->private_data;
-       task = m->private;
+       if (!task)
+               return -ESRCH;
        clear_all_latency_tracing(task);
+       put_task_struct(task);
 
        return count;
 }
@@ -412,6 +411,7 @@ static const struct limit_names lnames[RLIM_NLIMITS] = {
        [RLIMIT_MSGQUEUE] = {"Max msgqueue size", "bytes"},
        [RLIMIT_NICE] = {"Max nice priority", NULL},
        [RLIMIT_RTPRIO] = {"Max realtime priority", NULL},
+       [RLIMIT_RTTIME] = {"Max realtime timeout", "us"},
 };
 
 /* Display limits for a process */
@@ -502,17 +502,14 @@ static const struct inode_operations proc_def_inode_operations = {
        .setattr        = proc_setattr,
 };
 
-extern struct seq_operations mounts_op;
-struct proc_mounts {
-       struct seq_file m;
-       int event;
-};
-
-static int mounts_open(struct inode *inode, struct file *file)
+static int mounts_open_common(struct inode *inode, struct file *file,
+                             const struct seq_operations *op)
 {
        struct task_struct *task = get_proc_task(inode);
        struct nsproxy *nsp;
        struct mnt_namespace *ns = NULL;
+       struct fs_struct *fs = NULL;
+       struct path root;
        struct proc_mounts *p;
        int ret = -EINVAL;
 
@@ -525,40 +522,61 @@ static int mounts_open(struct inode *inode, struct file *file)
                                get_mnt_ns(ns);
                }
                rcu_read_unlock();
-
+               if (ns)
+                       fs = get_fs_struct(task);
                put_task_struct(task);
        }
 
-       if (ns) {
-               ret = -ENOMEM;
-               p = kmalloc(sizeof(struct proc_mounts), GFP_KERNEL);
-               if (p) {
-                       file->private_data = &p->m;
-                       ret = seq_open(file, &mounts_op);
-                       if (!ret) {
-                               p->m.private = ns;
-                               p->event = ns->event;
-                               return 0;
-                       }
-                       kfree(p);
-               }
-               put_mnt_ns(ns);
-       }
+       if (!ns)
+               goto err;
+       if (!fs)
+               goto err_put_ns;
+
+       read_lock(&fs->lock);
+       root = fs->root;
+       path_get(&root);
+       read_unlock(&fs->lock);
+       put_fs_struct(fs);
+
+       ret = -ENOMEM;
+       p = kmalloc(sizeof(struct proc_mounts), GFP_KERNEL);
+       if (!p)
+               goto err_put_path;
+
+       file->private_data = &p->m;
+       ret = seq_open(file, op);
+       if (ret)
+               goto err_free;
+
+       p->m.private = p;
+       p->ns = ns;
+       p->root = root;
+       p->event = ns->event;
+
+       return 0;
+
+ err_free:
+       kfree(p);
+ err_put_path:
+       path_put(&root);
+ err_put_ns:
+       put_mnt_ns(ns);
+ err:
        return ret;
 }
 
 static int mounts_release(struct inode *inode, struct file *file)
 {
-       struct seq_file *m = file->private_data;
-       struct mnt_namespace *ns = m->private;
-       put_mnt_ns(ns);
+       struct proc_mounts *p = file->private_data;
+       path_put(&p->root);
+       put_mnt_ns(p->ns);
        return seq_release(inode, file);
 }
 
 static unsigned mounts_poll(struct file *file, poll_table *wait)
 {
        struct proc_mounts *p = file->private_data;
-       struct mnt_namespace *ns = p->m.private;
+       struct mnt_namespace *ns = p->ns;
        unsigned res = 0;
 
        poll_wait(file, &ns->poll, wait);
@@ -573,6 +591,11 @@ static unsigned mounts_poll(struct file *file, poll_table *wait)
        return res;
 }
 
+static int mounts_open(struct inode *inode, struct file *file)
+{
+       return mounts_open_common(inode, file, &mounts_op);
+}
+
 static const struct file_operations proc_mounts_operations = {
        .open           = mounts_open,
        .read           = seq_read,
@@ -581,38 +604,22 @@ static const struct file_operations proc_mounts_operations = {
        .poll           = mounts_poll,
 };
 
-extern struct seq_operations mountstats_op;
-static int mountstats_open(struct inode *inode, struct file *file)
+static int mountinfo_open(struct inode *inode, struct file *file)
 {
-       int ret = seq_open(file, &mountstats_op);
-
-       if (!ret) {
-               struct seq_file *m = file->private_data;
-               struct nsproxy *nsp;
-               struct mnt_namespace *mnt_ns = NULL;
-               struct task_struct *task = get_proc_task(inode);
-
-               if (task) {
-                       rcu_read_lock();
-                       nsp = task_nsproxy(task);
-                       if (nsp) {
-                               mnt_ns = nsp->mnt_ns;
-                               if (mnt_ns)
-                                       get_mnt_ns(mnt_ns);
-                       }
-                       rcu_read_unlock();
+       return mounts_open_common(inode, file, &mountinfo_op);
+}
 
-                       put_task_struct(task);
-               }
+static const struct file_operations proc_mountinfo_operations = {
+       .open           = mountinfo_open,
+       .read           = seq_read,
+       .llseek         = seq_lseek,
+       .release        = mounts_release,
+       .poll           = mounts_poll,
+};
 
-               if (mnt_ns)
-                       m->private = mnt_ns;
-               else {
-                       seq_release(inode, file);
-                       ret = -EINVAL;
-               }
-       }
-       return ret;
+static int mountstats_open(struct inode *inode, struct file *file)
+{
+       return mounts_open_common(inode, file, &mountstats_op);
 }
 
 static const struct file_operations proc_mountstats_operations = {
@@ -658,6 +665,45 @@ static const struct file_operations proc_info_file_operations = {
        .read           = proc_info_read,
 };
 
+static int proc_single_show(struct seq_file *m, void *v)
+{
+       struct inode *inode = m->private;
+       struct pid_namespace *ns;
+       struct pid *pid;
+       struct task_struct *task;
+       int ret;
+
+       ns = inode->i_sb->s_fs_info;
+       pid = proc_pid(inode);
+       task = get_pid_task(pid, PIDTYPE_PID);
+       if (!task)
+               return -ESRCH;
+
+       ret = PROC_I(inode)->op.proc_show(m, ns, pid, task);
+
+       put_task_struct(task);
+       return ret;
+}
+
+static int proc_single_open(struct inode *inode, struct file *filp)
+{
+       int ret;
+       ret = single_open(filp, proc_single_show, NULL);
+       if (!ret) {
+               struct seq_file *m = filp->private_data;
+
+               m->private = inode;
+       }
+       return ret;
+}
+
+static const struct file_operations proc_single_file_operations = {
+       .open           = proc_single_open,
+       .read           = seq_read,
+       .llseek         = seq_lseek,
+       .release        = single_release,
+};
+
 static int mem_open(struct inode* inode, struct file* file)
 {
        file->private_data = (void*)((long)current->self_exec_id);
@@ -783,7 +829,7 @@ out_no_task:
 }
 #endif
 
-static loff_t mem_lseek(struct file * file, loff_t offset, int orig)
+loff_t mem_lseek(struct file *file, loff_t offset, int orig)
 {
        switch (orig) {
        case 0:
@@ -997,6 +1043,26 @@ static const struct file_operations proc_loginuid_operations = {
        .read           = proc_loginuid_read,
        .write          = proc_loginuid_write,
 };
+
+static ssize_t proc_sessionid_read(struct file * file, char __user * buf,
+                                 size_t count, loff_t *ppos)
+{
+       struct inode * inode = file->f_path.dentry->d_inode;
+       struct task_struct *task = get_proc_task(inode);
+       ssize_t length;
+       char tmpbuf[TMPBUFLEN];
+
+       if (!task)
+               return -ESRCH;
+       length = scnprintf(tmpbuf, TMPBUFLEN, "%u",
+                               audit_get_sessionid(task));
+       put_task_struct(task);
+       return simple_read_from_buffer(buf, count, ppos, tmpbuf, length);
+}
+
+static const struct file_operations proc_sessionid_operations = {
+       .read           = proc_sessionid_read,
+};
 #endif
 
 #ifdef CONFIG_FAULT_INJECTION
@@ -1121,39 +1187,36 @@ static void *proc_pid_follow_link(struct dentry *dentry, struct nameidata *nd)
        int error = -EACCES;
 
        /* We don't need a base pointer in the /proc filesystem */
-       path_release(nd);
+       path_put(&nd->path);
 
        /* Are we allowed to snoop on the tasks file descriptors? */
        if (!proc_fd_access_allowed(inode))
                goto out;
 
-       error = PROC_I(inode)->op.proc_get_link(inode, &nd->dentry, &nd->mnt);
+       error = PROC_I(inode)->op.proc_get_link(inode, &nd->path);
        nd->last_type = LAST_BIND;
 out:
        return ERR_PTR(error);
 }
 
-static int do_proc_readlink(struct dentry *dentry, struct vfsmount *mnt,
-                           char __user *buffer, int buflen)
+static int do_proc_readlink(struct path *path, char __user *buffer, int buflen)
 {
-       struct inode * inode;
        char *tmp = (char*)__get_free_page(GFP_TEMPORARY);
-       char *path;
+       char *pathname;
        int len;
 
        if (!tmp)
                return -ENOMEM;
 
-       inode = dentry->d_inode;
-       path = d_path(dentry, mnt, tmp, PAGE_SIZE);
-       len = PTR_ERR(path);
-       if (IS_ERR(path))
+       pathname = d_path(path, tmp, PAGE_SIZE);
+       len = PTR_ERR(pathname);
+       if (IS_ERR(pathname))
                goto out;
-       len = tmp + PAGE_SIZE - 1 - path;
+       len = tmp + PAGE_SIZE - 1 - pathname;
 
        if (len > buflen)
                len = buflen;
-       if (copy_to_user(buffer, path, len))
+       if (copy_to_user(buffer, pathname, len))
                len = -EFAULT;
  out:
        free_page((unsigned long)tmp);
@@ -1164,20 +1227,18 @@ static int proc_pid_readlink(struct dentry * dentry, char __user * buffer, int b
 {
        int error = -EACCES;
        struct inode *inode = dentry->d_inode;
-       struct dentry *de;
-       struct vfsmount *mnt = NULL;
+       struct path path;
 
        /* Are we allowed to snoop on the tasks file descriptors? */
        if (!proc_fd_access_allowed(inode))
                goto out;
 
-       error = PROC_I(inode)->op.proc_get_link(inode, &de, &mnt);
+       error = PROC_I(inode)->op.proc_get_link(inode, &path);
        if (error)
                goto out;
 
-       error = do_proc_readlink(de, mnt, buffer, buflen);
-       dput(de);
-       mntput(mnt);
+       error = do_proc_readlink(&path, buffer, buflen);
+       path_put(&path);
 out:
        return error;
 }
@@ -1404,8 +1465,7 @@ out:
 
 #define PROC_FDINFO_MAX 64
 
-static int proc_fd_info(struct inode *inode, struct dentry **dentry,
-                       struct vfsmount **mnt, char *info)
+static int proc_fd_info(struct inode *inode, struct path *path, char *info)
 {
        struct task_struct *task = get_proc_task(inode);
        struct files_struct *files = NULL;
@@ -1424,10 +1484,10 @@ static int proc_fd_info(struct inode *inode, struct dentry **dentry,
                spin_lock(&files->file_lock);
                file = fcheck_files(files, fd);
                if (file) {
-                       if (mnt)
-                               *mnt = mntget(file->f_path.mnt);
-                       if (dentry)
-                               *dentry = dget(file->f_path.dentry);
+                       if (path) {
+                               *path = file->f_path;
+                               path_get(&file->f_path);
+                       }
                        if (info)
                                snprintf(info, PROC_FDINFO_MAX,
                                         "pos:\t%lli\n"
@@ -1444,10 +1504,9 @@ static int proc_fd_info(struct inode *inode, struct dentry **dentry,
        return -ENOENT;
 }
 
-static int proc_fd_link(struct inode *inode, struct dentry **dentry,
-                       struct vfsmount **mnt)
+static int proc_fd_link(struct inode *inode, struct path *path)
 {
-       return proc_fd_info(inode, dentry, mnt, NULL);
+       return proc_fd_info(inode, path, NULL);
 }
 
 static int tid_fd_revalidate(struct dentry *dentry, struct nameidata *nd)
@@ -1574,7 +1633,6 @@ static int proc_readfd_common(struct file * filp, void * dirent,
        unsigned int fd, ino;
        int retval;
        struct files_struct * files;
-       struct fdtable *fdt;
 
        retval = -ENOENT;
        if (!p)
@@ -1597,9 +1655,8 @@ static int proc_readfd_common(struct file * filp, void * dirent,
                        if (!files)
                                goto out;
                        rcu_read_lock();
-                       fdt = files_fdtable(files);
                        for (fd = filp->f_pos-2;
-                            fd < fdt->max_fds;
+                            fd < files_fdtable(files)->max_fds;
                             fd++, filp->f_pos++) {
                                char name[PROC_NUMBUF];
                                int len;
@@ -1641,7 +1698,7 @@ static ssize_t proc_fdinfo_read(struct file *file, char __user *buf,
                                      size_t len, loff_t *ppos)
 {
        char tmp[PROC_FDINFO_MAX];
-       int err = proc_fd_info(file->f_path.dentry->d_inode, NULL, NULL, tmp);
+       int err = proc_fd_info(file->f_path.dentry->d_inode, NULL, tmp);
        if (!err)
                err = simple_read_from_buffer(buf, len, ppos, tmp, strlen(tmp));
        return err;
@@ -2058,15 +2115,23 @@ static const struct file_operations proc_coredump_filter_operations = {
 static int proc_self_readlink(struct dentry *dentry, char __user *buffer,
                              int buflen)
 {
+       struct pid_namespace *ns = dentry->d_sb->s_fs_info;
+       pid_t tgid = task_tgid_nr_ns(current, ns);
        char tmp[PROC_NUMBUF];
-       sprintf(tmp, "%d", task_tgid_vnr(current));
+       if (!tgid)
+               return -ENOENT;
+       sprintf(tmp, "%d", tgid);
        return vfs_readlink(dentry,buffer,buflen,tmp);
 }
 
 static void *proc_self_follow_link(struct dentry *dentry, struct nameidata *nd)
 {
+       struct pid_namespace *ns = dentry->d_sb->s_fs_info;
+       pid_t tgid = task_tgid_nr_ns(current, ns);
        char tmp[PROC_NUMBUF];
-       sprintf(tmp, "%d", task_tgid_vnr(current));
+       if (!tgid)
+               return ERR_PTR(-ENOENT);
+       sprintf(tmp, "%d", task_tgid_nr_ns(current, ns));
        return ERR_PTR(vfs_follow_link(nd,tmp));
 }
 
@@ -2229,16 +2294,19 @@ static const struct pid_entry tgid_base_stuff[] = {
        DIR("task",       S_IRUGO|S_IXUGO, task),
        DIR("fd",         S_IRUSR|S_IXUSR, fd),
        DIR("fdinfo",     S_IRUSR|S_IXUSR, fdinfo),
+#ifdef CONFIG_NET
+       DIR("net",        S_IRUGO|S_IXUGO, net),
+#endif
        REG("environ",    S_IRUSR, environ),
        INF("auxv",       S_IRUSR, pid_auxv),
-       INF("status",     S_IRUGO, pid_status),
+       ONE("status",     S_IRUGO, pid_status),
        INF("limits",     S_IRUSR, pid_limits),
 #ifdef CONFIG_SCHED_DEBUG
        REG("sched",      S_IRUGO|S_IWUSR, pid_sched),
 #endif
        INF("cmdline",    S_IRUGO, pid_cmdline),
-       INF("stat",       S_IRUGO, tgid_stat),
-       INF("statm",      S_IRUGO, pid_statm),
+       ONE("stat",       S_IRUGO, tgid_stat),
+       ONE("statm",      S_IRUGO, pid_statm),
        REG("maps",       S_IRUGO, maps),
 #ifdef CONFIG_NUMA
        REG("numa_maps",  S_IRUGO, numa_maps),
@@ -2248,10 +2316,12 @@ static const struct pid_entry tgid_base_stuff[] = {
        LNK("root",       root),
        LNK("exe",        exe),
        REG("mounts",     S_IRUGO, mounts),
+       REG("mountinfo",  S_IRUGO, mountinfo),
        REG("mountstats", S_IRUSR, mountstats),
-#ifdef CONFIG_MMU
+#ifdef CONFIG_PROC_PAGE_MONITOR
        REG("clear_refs", S_IWUSR, clear_refs),
        REG("smaps",      S_IRUGO, smaps),
+       REG("pagemap",    S_IRUSR, pagemap),
 #endif
 #ifdef CONFIG_SECURITY
        DIR("attr",       S_IRUGO|S_IXUGO, attr_dir),
@@ -2275,6 +2345,7 @@ static const struct pid_entry tgid_base_stuff[] = {
        REG("oom_adj",    S_IRUGO|S_IWUSR, oom_adjust),
 #ifdef CONFIG_AUDITSYSCALL
        REG("loginuid",   S_IWUSR|S_IRUGO, loginuid),
+       REG("sessionid",  S_IRUSR, sessionid),
 #endif
 #ifdef CONFIG_FAULT_INJECTION
        REG("make-it-fail", S_IRUGO|S_IWUSR, fault_inject),
@@ -2320,7 +2391,8 @@ static void proc_flush_task_mnt(struct vfsmount *mnt, pid_t pid, pid_t tgid)
        name.len = snprintf(buf, sizeof(buf), "%d", pid);
        dentry = d_hash_and_lookup(mnt->mnt_root, &name);
        if (dentry) {
-               shrink_dcache_parent(dentry);
+               if (!(current->flags & PF_EXITING))
+                       shrink_dcache_parent(dentry);
                d_drop(dentry);
                dput(dentry);
        }
@@ -2560,14 +2632,14 @@ static const struct pid_entry tid_base_stuff[] = {
        DIR("fdinfo",    S_IRUSR|S_IXUSR, fdinfo),
        REG("environ",   S_IRUSR, environ),
        INF("auxv",      S_IRUSR, pid_auxv),
-       INF("status",    S_IRUGO, pid_status),
+       ONE("status",    S_IRUGO, pid_status),
        INF("limits",    S_IRUSR, pid_limits),
 #ifdef CONFIG_SCHED_DEBUG
        REG("sched",     S_IRUGO|S_IWUSR, pid_sched),
 #endif
        INF("cmdline",   S_IRUGO, pid_cmdline),
-       INF("stat",      S_IRUGO, tid_stat),
-       INF("statm",     S_IRUGO, pid_statm),
+       ONE("stat",      S_IRUGO, tid_stat),
+       ONE("statm",     S_IRUGO, pid_statm),
        REG("maps",      S_IRUGO, maps),
 #ifdef CONFIG_NUMA
        REG("numa_maps", S_IRUGO, numa_maps),
@@ -2577,9 +2649,11 @@ static const struct pid_entry tid_base_stuff[] = {
        LNK("root",      root),
        LNK("exe",       exe),
        REG("mounts",    S_IRUGO, mounts),
-#ifdef CONFIG_MMU
+       REG("mountinfo",  S_IRUGO, mountinfo),
+#ifdef CONFIG_PROC_PAGE_MONITOR
        REG("clear_refs", S_IWUSR, clear_refs),
        REG("smaps",     S_IRUGO, smaps),
+       REG("pagemap",    S_IRUSR, pagemap),
 #endif
 #ifdef CONFIG_SECURITY
        DIR("attr",      S_IRUGO|S_IXUGO, attr_dir),
@@ -2603,6 +2677,7 @@ static const struct pid_entry tid_base_stuff[] = {
        REG("oom_adj",   S_IRUGO|S_IWUSR, oom_adjust),
 #ifdef CONFIG_AUDITSYSCALL
        REG("loginuid",  S_IWUSR|S_IRUGO, loginuid),
+       REG("sessionid",  S_IRUSR, sessionid),
 #endif
 #ifdef CONFIG_FAULT_INJECTION
        REG("make-it-fail", S_IRUGO|S_IWUSR, fault_inject),