This is the mail archive of the gdb-patches@sourceware.org mailing list for the GDB project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [resubmit] gdb.base, r*.exp thru w*.exp


Could you identify the subset of changes that behave erratically for you?

Frederic Riss wrote:
Hi,

I get some strange issues with this series of commits. Some tests
start to randomly fail. It's quite hard to reproduce, I only have 1
out of ~10 boxes that exposes the issue in an random manner. To take a
real example, let's take this hunk:

On 20 May 2010 22:29, Michael Snyder <msnyder@vmware.com> wrote:
Index: setshow.exp
--- setshow.exp 5 May 2010 18:06:58 -0000       1.14
+++ setshow.exp 20 May 2010 20:24:11 -0000
@@ -54,27 +54,25 @@ if { ![runto_main] } {
 #test set annotate 2
-send_gdb "set annotate 2\n"
-gdb_expect {
-       -re ".*\032\032pre-prompt.*$gdb_prompt .*\032\032prompt.*$" \
-                               { pass "set annotate 2" }
-       timeout                 { fail "(timeout) set annotate 2" }
+
+gdb_test_multiple "set annotate 2" "set annotate 2" {
+    -re ".*\032\032pre-prompt.*$gdb_prompt .*\032\032prompt.*$" {
+    pass "set annotate 2"
    }
+}

Before that commit, the test seemed 100% reliable. After the commit, if I run it in a loop on the affected box, I start to get failures like that in my runs:

set annotate 2

^Z^Zpre-prompt
(gdb) FAIL: gdb.base/setshow.exp: set annotate 2

^Z^Zprompt

As you see the required things are in the output, but expect fails to
match it;. There must be some subtle difference between gdb_expect and
gdb_test_multiple when matching patterns containing prompts. Does that
make any sense to someone? Note that I took this precise example, but
I have a few places in the testsuite that show similar random
failures.

Thanks,
Fred


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]