public inbox for systemtap@sourceware.org
 help / color / mirror / Atom feed
* stap make all check test failures
@ 2006-07-20 15:06 Kevin Feliksa
  2006-07-20 15:34 ` Frank Ch. Eigler
  0 siblings, 1 reply; 2+ messages in thread
From: Kevin Feliksa @ 2006-07-20 15:06 UTC (permalink / raw)
  To: systemtap

I found 4 of 211 tests fail during the "make all check"  testing phase of 
a SystemTap build from source.   Failing tests are:

.... /buildok/process_test.stp
.... /buildok/sched_test.stp
.... /buildok/scsi.stp
.... /buildok/seven.stp

Note that these tests are executed consecutively.  More information is 
provided below. 

Any suggestions?  I've seen these tests also fail during earlier source 
module builds on FC-5. 
Thanks.


Here's some info: 

System: 
IBM T30 laptop
Fedora core 5 ISO release (updated)
kernel-2.6.17-1.2157_FC5.i686
elfutils-0.122
kernel-devel-2.6.17-1.2157_FC5
kernel-debuginfo-2.6.17-1.2157_FC5
glib2-devel-2.10.3-1
gcc-4.1.1-1.fc5
gcc-c++-4.1.1-1.fc5
systemtap-20060715


From: 

.../testresults/buildok-process_test.stp.err
-------------------------------------------------------------
semantic error: no match for probe point
while: resolving probe point process.signal_handle
Pass 2: analysis failed.  Try again with more '-v' (verbose) options.


.../testresults/buildok-sched_test.stp.err
-------------------------------------------------------------
semantic error: no match for probe point
while: resolving probe point kernel.inline("pull_task")
semantic error: no match for probe point
while: resolving probe point scheduler.migrate
semantic error: no match for probe point
while: resolving probe point kernel.inline("idle_balance")
semantic error: no match for probe point
while: resolving probe point scheduler.balance
Pass 2: analysis failed.  Try again with more '-v' (verbose) options.


.../testresults/buildok-scsi.stp.err
-------------------------------------------------------------
semantic error: no match for probe point
while: resolving probe point 
module("*").function("scsi_prep_fn@drivers/scsi/scsi_lib.c")
semantic error: no match for probe point
while: resolving probe point scsi.ioentry
semantic error: no match for probe point
while: resolving probe point 
module("*").function("scsi_dispatch_cmd@drivers/scsi/scsi.c")
semantic error: no match for probe point
while: resolving probe point scsi.iodispatching
semantic error: no match for probe point
while: resolving probe point 
module("*").function("scsi_done@drivers/scsi/scsi.c")
semantic error: no match for probe point
while: resolving probe point scsi.iodone
semantic error: no match for probe point
while: resolving probe point 
module("*").function("scsi_io_completion@drivers/scsi/scsi_lib.c")
semantic error: no match for probe point
while: resolving probe point scsi.iocompleted
Pass 2: analysis failed.  Try again with more '-v' (verbose) options.


.../testresults/buildok-seven.stp.err
-------------------------------------------------------------
semantic error: unresolved target-symbol expression: identifier '$type' at 
././testsuite/buildok/seven.stp:11:30
semantic error: unresolved target-symbol expression: identifier '$type' at 
././testsuite/buildok/seven.stp:11:30
Pass 2: analysis failed.  Try again with more '-v' (verbose) options.


^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: stap make all check test failures
  2006-07-20 15:06 stap make all check test failures Kevin Feliksa
@ 2006-07-20 15:34 ` Frank Ch. Eigler
  0 siblings, 0 replies; 2+ messages in thread
From: Frank Ch. Eigler @ 2006-07-20 15:34 UTC (permalink / raw)
  To: Kevin Feliksa; +Cc: systemtap

Kevin Feliksa <feliksa@us.ibm.com> writes:

> I found 4 of 211 tests fail during the "make all check"  testing phase of 
> a SystemTap build from source.   Failing tests are:
> .... /buildok/process_test.stp
> .... /buildok/sched_test.stp
> .... /buildok/scsi.stp
> .... /buildok/seven.stp
> [...]

The tapset-related tests should be made more conditionalized so that
they succeed on machines e.g. without scsi.

In general, we need to improve the quality of the test suite.  We put
up with too many known failures, too much rigidity.  The tests should
pass on any supported platform, except where an actual bug is to be
inferred.

Perhaps it would help if the "make check" suite was converted to
dejagnu.  (Separely, I plan to pull over the "pass-5" test suite from
/tests/testsuite over into /src/testsuite also, though of course not
activated automatically with a "make check".)

- FChE

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2006-07-20 15:34 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2006-07-20 15:06 stap make all check test failures Kevin Feliksa
2006-07-20 15:34 ` Frank Ch. Eigler

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).