The pc-file output from the exp-bbv tool is mildly implementation dependent.

This causes the regression test to fail with the "million" benchmark
on a Fedora 9 system.  For now, remove the pc-file part of the test,
and later I'll come up with a better way of testing this feature.

git-svn-id: svn://svn.valgrind.org/valgrind/trunk@10571

diff --git a/exp-bbv/tests/amd64-linux/million.post.exp b/exp-bbv/tests/amd64-linux/million.post.exp
index 30bdd298c5f4cbdd8a29bbbf2775e40bc1046953..260eee6d8cc365e9158900f2c4e9e632b6462fc5 100644 (file)
--- a/exp-bbv/tests/amd64-linux/million.post.exp
+++ b/exp-bbv/tests/amd64-linux/million.post.exp
@@ -16,6 +16,3 @@ T:2:100000
 #   Unique reps: 0
 #   Total fldcw instructions: 0
 
-F:1:400078:
-F:2:400085:
-F:3:40008a:

diff --git a/exp-bbv/tests/amd64-linux/million.vgtest b/exp-bbv/tests/amd64-linux/million.vgtest
index 969a636813eb37ecdc0040e5c9af1498cbabadd9..c366a8bafd6ce60ba8c4f3e409572eaba359c633 100644 (file)
--- a/exp-bbv/tests/amd64-linux/million.vgtest
+++ b/exp-bbv/tests/amd64-linux/million.vgtest
@@ -1,5 +1,5 @@
 prog: million 
-vgopts: --interval-size=100000 --bb-out-file=million.out.bb --pc-out-file=million.out.pc
-post:  cat million.out.bb million.out.pc
-cleanup: rm million.out.bb million.out.pc
+vgopts: --interval-size=100000 --bb-out-file=million.out.bb
+post:  cat million.out.bb
+cleanup: rm million.out.bb