Don't save quitcounter across longjmp in exec_byte_code

* src/bytecode.c (exec_byte_code): Reset quitcounter to 1 after longjmp;
the exact value isn't important.  This may reduce register pressure a
tiny bit or at least remove a useless stack slot.

1 files changed, 1 insertions, 3 deletions

diff --git a/src/bytecode.c b/src/bytecode.c
index debb08c2347..318b96a12be 100644
--- a/src/bytecode.c
+++ b/src/bytecode.c
@@ -526,7 +526,6 @@ exec_byte_code (Lisp_Object fun, ptrdiff_t args_template,
     for (ptrdiff_t i = nargs - rest; i < nonrest; i++)
       PUSH (Qnil);
 
-  unsigned char volatile saved_quitcounter;
 #if GCC_LINT && __GNUC__ && !__clang__
   Lisp_Object *volatile saved_vectorp;
   unsigned char const *volatile saved_bytestr_data;
@@ -960,7 +959,7 @@ exec_byte_code (Lisp_Object fun, ptrdiff_t args_template,
 
             if (sys_setjmp (c->jmp))
               {
-                quitcounter = saved_quitcounter;
+                quitcounter = 1;   /* no need to restore old value */
                 struct handler *c = handlerlist;
                 handlerlist = c->next;
                 top = c->bytecode_top;
@@ -990,7 +989,6 @@ exec_byte_code (Lisp_Object fun, ptrdiff_t args_template,
                 goto op_branch;
               }
 
-            saved_quitcounter = quitcounter;
 #if GCC_LINT && __GNUC__ && !__clang__
             saved_vectorp = vectorp;
             saved_bytestr_data = bytestr_data;