From: Stefano Rivera <stefanor@debian.org>
Date: Wed, 13 Feb 2019 15:34:05 -0800
Subject: xfail TestOwnLib.test_modify_struct_value on some platforms

It is known to fail, and isn't a regression, but rather a pre-existing
bug, discovered in a new test.

Bug-Upstream: https://bitbucket.org/cffi/cffi/issues/401/testownlibtest_modify_struct_value-fails
---
 testing/cffi0/test_ownlib.py | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/testing/cffi0/test_ownlib.py b/testing/cffi0/test_ownlib.py
index f4f9ee2..65a27b4 100644
--- a/testing/cffi0/test_ownlib.py
+++ b/testing/cffi0/test_ownlib.py
@@ -1,4 +1,5 @@
 import py, sys, os
+import platform
 import subprocess, weakref
 from cffi import FFI
 from cffi.backend_ctypes import CTypesBackend
@@ -351,6 +352,8 @@ class TestOwnLib(object):
     def test_modify_struct_value(self):
         if self.module is None:
             py.test.skip("fix the auto-generation of the tiny test lib")
+        if platform.machine() in ('aarch64', 'parisc64', 'riscv64', 'sparc64'):
+            py.test.xfail("Known to fail on arm64, hppa, riscv64, and sparc64")
         ffi = FFI(backend=self.Backend())
         ffi.cdef("""
             typedef struct {
