bpo-21131: Fix faulthandler.register(chain=True) stack (GH-15276)

faulthandler now allocates a dedicated stack of SIGSTKSZ*2 bytes,
instead of just SIGSTKSZ bytes. Calling the previous signal handler
in faulthandler signal handler uses more than SIGSTKSZ bytes of stack
memory on some platforms.
(cherry picked from commit ac827ed)

Co-authored-by: Victor Stinner <[email protected]>

Author: miss-islington

Misc/NEWS.d/next/Library/2019-08-14-15-34-23.bpo-21131.0MMQRi.rst

@@ -0,0 +1,4 @@
+Fix ``faulthandler.register(chain=True)`` stack. faulthandler now allocates a
+dedicated stack of ``SIGSTKSZ*2`` bytes, instead of just ``SIGSTKSZ`` bytes.
+Calling the previous signal handler in faulthandler signal handler uses more
+than ``SIGSTKSZ`` bytes of stack memory on some platforms.

Modules/faulthandler.c

@@ -1322,7 +1322,11 @@ _PyFaulthandler_Init(int enable)
      * be able to allocate memory on the stack, even on a stack overflow. If it
      * fails, ignore the error. */
     stack.ss_flags = 0;
-    stack.ss_size = SIGSTKSZ;
+    /* bpo-21131: allocate dedicated stack of SIGSTKSZ*2 bytes, instead of just
+       SIGSTKSZ bytes. Calling the previous signal handler in faulthandler
+       signal handler uses more than SIGSTKSZ bytes of stack memory on some
+       platforms. */
+    stack.ss_size = SIGSTKSZ * 2;
     stack.ss_sp = PyMem_Malloc(stack.ss_size);
     if (stack.ss_sp != NULL) {
         err = sigaltstack(&stack, &old_stack);