[Qemu-devel] [PATCH 1/4] linux-user: Disallow setting newsp for fork

[Richard Henderson]

Or really, just clone devolving into fork.  This should not ever happen
in practice.  We do want to reserve calling cpu_clone_regs for the case
in which we are actually performing a clone.

Signed-off-by: Richard Henderson <address@hidden>
---
 linux-user/syscall.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/linux-user/syscall.c b/linux-user/syscall.c
index dfc851cc35..5bf8d13de7 100644
--- a/linux-user/syscall.c
+++ b/linux-user/syscall.c
@@ -6502,10 +6502,14 @@ static int do_fork(CPUArchState *env, unsigned int 
flags, abi_ulong newsp,
         pthread_mutex_destroy(&info.mutex);
         pthread_mutex_unlock(&clone_lock);
     } else {
-        /* if no CLONE_VM, we consider it is a fork */
+        /* If no CLONE_VM, we consider it is a fork.  */
         if (flags & CLONE_INVALID_FORK_FLAGS) {
             return -TARGET_EINVAL;
         }
+        /* As a fork, setting a new sp does not make sense.  */
+        if (newsp) {
+            return -TARGET_EINVAL;
+        }
 
         /* We can't support custom termination signals */
         if ((flags & CSIGNAL) != TARGET_SIGCHLD) {
@@ -6520,7 +6524,6 @@ static int do_fork(CPUArchState *env, unsigned int flags, 
abi_ulong newsp,
         ret = fork();
         if (ret == 0) {
             /* Child Process.  */
-            cpu_clone_regs(env, newsp);
             fork_end(1);
             /* There is a race condition here.  The parent process could
                theoretically read the TID in the child process before the child
-- 
2.17.1