Fix a race with free'ing vmspaces at process exit when vmspaces are

shared.

Also introduce vm_endcopy instead of using pointer tricks when
initializing new vmspaces.

The race occured because of how the reference was utilized:
  test vmspace reference,
  possibly block,
  decrement reference

When sharing a vmspace between multiple processes it was possible
for two processes exiting at the same time to test the reference
count, possibly block and neither one free because they wouldn't
see the other's update.

Submitted by: green

1 files changed, 4 insertions, 3 deletions

diff --git a/sys/kern/kern_exit.c b/sys/kern/kern_exit.c
index 51b77c4..e8c5558 100644
--- a/sys/kern/kern_exit.c
+++ b/sys/kern/kern_exit.c
@@ -217,13 +217,14 @@ exit1(td, rv)
 	 * Can't free the entire vmspace as the kernel stack
 	 * may be mapped within that space also.
 	 */
-	if (vm->vm_refcnt == 1) {
+	if (--vm->vm_refcnt == 0) {
 		if (vm->vm_shm)
 			shmexit(p);
 		pmap_remove_pages(vmspace_pmap(vm), VM_MIN_ADDRESS,
 		    VM_MAXUSER_ADDRESS);
 		(void) vm_map_remove(&vm->vm_map, VM_MIN_ADDRESS,
 		    VM_MAXUSER_ADDRESS);
+		vm->vm_freer = p;
 	}
 
 	PROC_LOCK(p);
@@ -400,8 +401,8 @@ exit1(td, rv)
 	/*
 	 * Finally, call machine-dependent code to release the remaining
 	 * resources including address space, the kernel stack and pcb.
-	 * The address space is released by "vmspace_free(p->p_vmspace)"
-	 * in vm_waitproc();
+	 * The address space is released by "vmspace_exitfree(p)" in
+	 * vm_waitproc().
 	 */
 	cpu_exit(td);