This is the mail archive of the systemtap@sourceware.org mailing list for the systemtap 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: S390x Systemtap testing on s390x


David Smith wrote:

Thanks for the comments David.

David Wilder wrote:

This is a re-send, some people could not read my original message, my mailer messed up..
Here is the results from todays testing on S390 with RHEL5 RC. This test is based on the systemtap included in the RC.


I took a look at these results, comments below. Could you also re-run the tests with the latest version of systemtap for comparison purposes?

Yea, I do that regularly the results are much better. I will post the results soon. Many of the failures I understand and have fixes for. Not all have been checked in however.



=== systemtap Summary ===

# of unexpected failures 34


Wow, 34 unexpected failures. I've never seen such a high number (without something like debuginfo not being installed).

I am down to about 17 failures on later builds.

Running /usr/src/redhat/SOURCES/systemtap-0.5.12/testsuite/systemtap.maps/pmap_agg_overflow.exp ...
ERROR: tcl error sourcing /usr/src/redhat/SOURCES/systemtap-0.5.12/testsuite/systemtap.maps/pmap_agg_overflow.exp.


ERROR: spawn_id: spawn id exp8 not open
while executing
"close"
(file "/usr/src/redhat/SOURCES/systemtap-0.5.12/testsuite/systemtap.maps/pmap_agg_overflow.exp" line 26)
invoked from within
"source /usr/src/redhat/SOURCES/systemtap-0.5.12/testsuite/systemtap.maps/pmap_agg_overflow.exp"


("uplevel" body line 1)
invoked from within
"uplevel #0 source /usr/src/redhat/SOURCES/systemtap-0.5.12/testsuite/systemtap.maps/pmap_agg_overflow.exp"


    invoked from within
"catch "uplevel #0 source $test_file_name""


I'm unsure as to what the above is - test suite oddity or real problem.

FAIL: buildok/scsi.stp


Here's some of the output from systemtap.log for buildok/scsi.stp:

=============
Running /usr/src/redhat/SOURCES/systemtap-0.5.12/testsuite/buildok/scsi.stp
starting /usr/src/redhat/SOURCES/systemtap-0.5.12/testsuite/buildok/scsi.stp
/usr/src/redhat/SOURCES/systemtap-0.5.12/testsuite/buildok/scsi.stp: line 43: 11026 Segmentation fault ./stap -p4 - <<EOF
=============


I'm not sure if I've ever seen stap with a segmentation fault (that wasn't my fault as I was developing). Can you give us a backtrace here? A script shouldn't be able to crash stap.

This is not failing on later builds.

FAIL: buildok/seventeen.stp


Here's some of the output from systemtap.log for buildok/seventeen.stp:
=============
Running /usr/src/redhat/SOURCES/systemtap-0.5.12/testsuite/buildok/seventeen.stp
starting /usr/src/redhat/SOURCES/systemtap-0.5.12/testsuite/buildok/seventeen.stp
semantic error: libdwfl failure (dwfl_addrmodule): Operation not permitted: identifier '$write_fifo_fops' at /usr/src/redhat/SOURCES/systemtap-0.5.12/testsuite/buildok/seventeen.stp:8:19


Pass 2: analysis failed.  Try again with more '-v' (verbose) options.
=============

This one might be a S390x elfutils/gcc problem.

FAIL: buildok/twenty.stp


All we get from systemtap.log for buildok/twenty.stp is:

=============
Running /usr/src/redhat/SOURCES/systemtap-0.5.12/testsuite/buildok/twenty.stp
starting /usr/src/redhat/SOURCES/systemtap-0.5.12/testsuite/buildok/twenty.stp
Pass 4: compilation failed. Try again with more '-v' (verbose) options.
=============


You'll have to run that one by hand and up the number of '-v' options so we can see what is going on.

FAIL: buildok/twentythree.stp


The buildok/twentythree.stp output looks exactly like buildok/twenty.stp. You'll have to run it by hand and up the number of '-v' options.

Running /usr/src/redhat/SOURCES/systemtap-0.5.12/testsuite/systemtap.stress/current.exp ...
FAIL: /usr/src/redhat/SOURCES/systemtap-0.5.12/testsuite/systemtap.stress/current.stp compilation


My guess on this one is that you need to add some S390x specific code to the test.

Running /usr/src/redhat/SOURCES/systemtap-0.5.12/testsuite/systemtap.syscall/test.exp ...
FAIL: access
FAIL: acct
FAIL: alarm
FAIL: chmod
FAIL: clock
FAIL: dir
FAIL: forkwait
FAIL: itimer
FAIL: link
FAIL: mmap
FAIL: mount
FAIL: net1
FAIL: openclose
FAIL: readwrite
FAIL: rt_signal
FAIL: sendfile
FAIL: signal
FAIL: stat
FAIL: statfs
FAIL: swap
FAIL: sync
FAIL: timer
FAIL: trunc
FAIL: uid
UNSUPPORTED: uid16 not supported on this arch
FAIL: umask
FAIL: unlink


I'd bet most of these failures are tapset problems (that have subsequently been fixed). There is at least one compile error in the above tests (clock) that you might want to look into.

I hope this helps.


All the last set of failures are because sys.stp will not build because it cant find the 5th argument to probed functions. I think I have that one handled.


--
David Wilder
IBM Linux Technology Center
Beaverton, Oregon, USA dwilder@us.ibm.com
(503)578-3789



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