This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [PATCH v2] Improve analysis of racy testcases
- From: Pedro Alves <palves at redhat dot com>
- To: Sergio Durigan Junior <sergiodj at redhat dot com>, GDB Patches <gdb-patches at sourceware dot org>
- Date: Tue, 1 Mar 2016 12:41:36 +0000
- Subject: Re: [PATCH v2] Improve analysis of racy testcases
- Authentication-results: sourceware.org; auth=none
- References: <87r3gcgm91 dot fsf at redhat dot com> <87d1rewxst dot fsf at redhat dot com>
On 03/01/2016 05:51 AM, Sergio Durigan Junior wrote:
# Note that we must resort to a recursive make invocation here,
@@ -190,6 +197,26 @@ DO_RUNTEST = \
check-single:
$(DO_RUNTEST) $(RUNTESTFLAGS) $(expanded_tests_or_none)
+check-single-racy:
(...)
+ $(srcdir)/analyze-racy-logs.py \
+ `ls racy_outputs/gdb*.sum` > racy.sum; \
This ls invocation needs updating.
+ sed -n '/=== gdb Summary ===/,$$ p' racy.sum
+
diff --git a/gdb/testsuite/README b/gdb/testsuite/README
index 6b59027..043a8bd 100644
--- a/gdb/testsuite/README
+++ b/gdb/testsuite/README
@@ -50,6 +50,51 @@ to any non-empty value:
If you want to use runtest directly instead of using the Makefile, see
the description of GDB_PARALLEL below.
+Racy testcases
+**************
+
+Sometimes, new testcases are added to the testsuite that are not
+entirely deterministic, and can randomly pass or fail. We call them
+"racy testcases", and they can be bothersome when one is comparing
+different testsuite runs. In order to help identifying them, it is
+possible to run the tests several times in a row and ask the testsuite
+machinery to analyze the results. To do that, you need to specify the
+RACY_ITER environment variable to make:
Nicer, thanks. :-)
+
+ # Print the header.
+ print "\t\t=== gdb tests ===\n"
"gdb racy tests" ?
+
+ # Print each test.
+ for line in sorted (racy_tests):
+ print line
+
+ # Print the summary.
+ print "\n"
+ print "\t\t=== gdb Summary ===\n"
+ print "# of racy testcases:\t\t%d" % len (racy_tests)
AFAICS, it's # of racy tests, not # of racy testcases, right?
Feel free to push with the above addressed.
Thanks,
Pedro Alves