[Bug testsuite/5094] gtod.exp fails on ppc64

["wcohen at redhat dot com" <sourceware-bugzilla at sourceware dot org>]

------- Additional Comments From wcohen at redhat dot com  2008-01-15 20:05 -------
I noticed a similar failure on an i686 machine running a stock kernel,
2.6.18-53.1.4EL with the current stap_snap_testing (20080112).  The gtod.exp
test has the following matching:

	-re {^[0-9]+ \:([0-9]+) appl\r\n[0-9]+ \:\1 kern\r\n[0-9]+ \:\1 prog\r\n} {
incr ok; exp_continue }

the \1 before "kern" and "prog" expect need something like the following to match:

1200411986605858 :00 appl

1200411986605862 :00 kern

1200411986605873 :00 prog


However, there are places in the output where the number entries are not grouped
together (notice the 78 before the last 77 and the 79 between the 78's):

1200411986606250 :77 kern
1200411986606252 :77 appl
1200411986606255 :78 kern
1200411986606256 :77 prog
1200411986606257 :78 appl
1200411986606260 :79 kern
1200411986606261 :78 prog
1200411986606262 :79 appl
1200411986606265 :79 prog


It appears that the get_time_of_day is not being monotonic. The particular i686
machine was a laptop, which could have changes in the clock frequency. I wonder
if the hypervisor on the ppc64 could be skewing the results.

-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=5094

------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.