Fix spurious EINTR while exec() within fakeroot

The scenario is:

- process running inside fakeroot exec()s
- fakeroot forwards the exec call
- exec starts thrashing the previous process, and notably close its
  ports
- the send port that pointed to fakeroot is thus closed
- fakeroot thus receives a no-senders notification
- its libports thus interrupts the current RPC on it, i.e. the exec(),
  because it thinks that the caller is no more, but that was exactly the
  point of the exec()...
- exec tries to interrupt its own RPCs, at best the execution will fail,
  at worse the process is already thrashed and there's nothing better
  than burying it for good.

The symptom is processes running inside fakeroot seeming randomly
crashing with SIGKILL.

Making fakeroot keep a send right avoids the no-senders notification,
thus the whole interruption (which was meaningless).

1 files changed, 8 insertions, 0 deletions

diff --git a/trans/fakeroot.c b/trans/fakeroot.c
index 944fafdb..e70cb280 100644
--- a/trans/fakeroot.c
+++ b/trans/fakeroot.c
@@ -939,6 +939,11 @@ netfs_S_file_exec_paths (struct protid *user,
 			      file, MACH_PORT_RIGHT_SEND, 1);
   pthread_mutex_unlock (&user->po->np->lock);
 
+  /* Add a gratuitous send right on the protid to avoid a no-sender, and thus
+     interrupt the exec, just because we are precisely replacing the calling
+     process!  */
+  mach_port_t gratuitous = ports_get_send_right (user);
+
   if (!err)
     {
 #ifdef HAVE_FILE_EXEC_PATHS
@@ -977,6 +982,9 @@ netfs_S_file_exec_paths (struct protid *user,
       for (i = 0; i < portarraylen; ++i)
 	mach_port_deallocate (mach_task_self (), portarray[i]);
     }
+
+  mach_port_deallocate (mach_task_self (), gratuitous);
+
   return err;
 }