New locking/refcounting for fs_struct

* all changes of current->fs are done under task_lock and write_lock of
  old fs->lock
* refcount is not atomic anymore (same protection)
* its decrements are done when removing reference from current; at the
  same time we decide whether to free it.
* put_fs_struct() is gone
* new field - ->in_exec.  Set by check_unsafe_exec() if we are trying to do
  execve() and only subthreads share fs_struct.  Cleared when finishing exec
  (success and failure alike).  Makes CLONE_FS fail with -EAGAIN if set.
* check_unsafe_exec() may fail with -EAGAIN if another execve() from subthread
  is in progress.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

1 files changed, 26 insertions, 5 deletions

diff --git a/fs/exec.c b/fs/exec.c
index c5128fbc9165..07a059664b73 100644
--- a/fs/exec.c
+++ b/fs/exec.c
@@ -1056,16 +1056,18 @@ EXPORT_SYMBOL(install_exec_creds);
  * - the caller must hold current->cred_exec_mutex to protect against
  *   PTRACE_ATTACH
  */
-void check_unsafe_exec(struct linux_binprm *bprm)
+int check_unsafe_exec(struct linux_binprm *bprm)
 {
         struct task_struct *p = current, *t;
         unsigned long flags;
         unsigned n_fs, n_sighand;
+        int res = 0;
 
         bprm->unsafe = tracehook_unsafe_exec(p);
 
         n_fs = 1;
         n_sighand = 1;
+        write_lock(&p->fs->lock);
         lock_task_sighand(p, &flags);
         for (t = next_thread(p); t != p; t = next_thread(t)) {
                 if (t->fs == p->fs)
@@ -1073,11 +1075,19 @@ void check_unsafe_exec(struct linux_binprm *bprm)
                 n_sighand++;
         }
 
-        if (atomic_read(&p->fs->count) > n_fs ||
-            atomic_read(&p->sighand->count) > n_sighand)
+        if (p->fs->users > n_fs ||
+            atomic_read(&p->sighand->count) > n_sighand) {
                 bprm->unsafe |= LSM_UNSAFE_SHARE;
+        } else {
+                if (p->fs->in_exec)
+                        res = -EAGAIN;
+                p->fs->in_exec = 1;
+        }
 
         unlock_task_sighand(p, &flags);
+        write_unlock(&p->fs->lock);
+
+        return res;
 }
 
 /* 
@@ -1296,12 +1306,15 @@ int do_execve(char * filename,
         bprm->cred = prepare_exec_creds();
         if (!bprm->cred)
                 goto out_unlock;
-        check_unsafe_exec(bprm);
+
+        retval = check_unsafe_exec(bprm);
+        if (retval)
+                goto out_unlock;
 
         file = open_exec(filename);
         retval = PTR_ERR(file);
         if (IS_ERR(file))
-                goto out_unlock;
+                goto out_unmark;
 
         sched_exec();
 
@@ -1344,6 +1357,9 @@ int do_execve(char * filename,
                 goto out;
 
         /* execve succeeded */
+        write_lock(&current->fs->lock);
+        current->fs->in_exec = 0;
+        write_unlock(&current->fs->lock);
         current->in_execve = 0;
         mutex_unlock(&current->cred_exec_mutex);
         acct_update_integrals(current);
@@ -1362,6 +1378,11 @@ out_file:
                 fput(bprm->file);
         }
 
+out_unmark:
+        write_lock(&current->fs->lock);
+        current->fs->in_exec = 0;
+        write_unlock(&current->fs->lock);
+
 out_unlock:
         current->in_execve = 0;
         mutex_unlock(&current->cred_exec_mutex);