[PATCH 02/10] exec: Fix creating executable stacks

[Sergey Bugaev]

The previous logic had two independent issues:

* We need to make the stack executable if either the program or its ELF
  interpreter requires executable stack.  In practice, it's common for
  the program itself to not require executable stack, but ld.so (glibc)
  needs it.

* mach_setup_thread () allocates stacks with a simple vm_allocate (),
  which creates non-executable memory.  So if an executable stack is
  required, the stack has to be vm_protect'ed to enable execution, not
  the other way around.  As the comment suggest, it would've been better
  to use vm_map () with the desired protection directly.
---
 exec/exec.c | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/exec/exec.c b/exec/exec.c
index 639564cb..f6788520 100644
--- a/exec/exec.c
+++ b/exec/exec.c
@@ -1335,11 +1335,12 @@ do_exec (file_t file,
   if (e.error)
     goto out;
 
-  /* It would probably be better to change mach_setup_thread so
-     it does a vm_map with the right permissions to start with.  */
-  if (!e.info.elf.execstack)
+  /* mach_setup_thread () creates non-executable stacks (with vm_allocate ()).
+     It would probably be better to change mach_setup_thread () so it does
+     a vm_map () with the right permissions to start with.  */
+  if (e.info.elf.execstack || (e.interp.section && interp.info.elf.execstack))
     e.error = vm_protect (newtask, boot->stack_base, boot->stack_size,
-                         0, VM_PROT_READ | VM_PROT_WRITE);
+                         0, VM_PROT_READ | VM_PROT_WRITE | VM_PROT_EXECUTE);
 
   if (oldtask != newtask && oldtask != MACH_PORT_NULL)
     {
-- 
2.44.0