try to fix PPC64 Qemu tests

diff --git a/.travis.yml b/.travis.yml
index ce69723d3187449a00cdcd0a703955dad3bbe757..ceb0f5a8e038b31b6716b82c53160d661b90b84e 100644 (file)
--- a/.travis.yml
+++ b/.travis.yml
@@ -203,12 +203,6 @@ matrix:
         - CC=clang-3.8 make tsan-test-zstream
         - CC=clang-3.8 make tsan-fuzztest
 
-    - name: Qemu PPC64 + Fuzz test  # ~13mn
-      if: branch = master
-      script:
-        - make ppcinstall
-        - make ppc64fuzz
-
     - name: PPC64LE + Fuzz test  # ~13mn
       if: branch = master
       arch: ppc64le
@@ -216,6 +210,13 @@ matrix:
         - cat /proc/cpuinfo
         - make test
 
+    - name: Qemu PPC64 + Fuzz test  # ~13mn, presumed Big-Endian (?)
+      dist: trusty  # note : PPC64 cross-compilation for Qemu tests seems broken on Xenial
+      if: branch = master
+      script:
+        - make ppcinstall
+        - make ppc64fuzz
+
     # note : we already have aarch64 tests on hardware
     - name: Qemu aarch64 + Fuzz Test (on Xenial)    # ~14mn
       if: branch = master