Re: Systemtap snap:b6371390 test on kernel 2.6.30-rc3-git2

[Josh Stone <jistone@redhat.com>]

Mark Wielaard wrote:
> On Tue, 2009-04-28 at 11:30 +0530, Pavan Naregundi wrote:
>> Running /home/pavan/systemtap/src/testsuite/systemtap.base/cast.exp ...
>> executing: stap /home/pavan/systemtap/src/testsuite/systemtap.base/cast.stp
>> -g
>> FAIL: systemtap.base/cast.stp
>> line 4: expected "tv_sec OK"
>> Got "tv_sec 42 != 0"
>> testcase /home/pavan/systemtap/src/testsuite/systemtap.base/cast.exp
>> completed in 8 seconds
> 
> That is strange, maybe struct timeval is different on ppc64 from what we
> expect in the @cast or get_timeval() in the script. Maybe Josh has a
> hunch what is going on (he wrote the test originally).

The scenario where this would fail is a big-endian machine where the
size of "long tv_sec" is different between kernel and user mode.  AFAIK,
ppc64 is one such machine.

I would say that the @cast is working fine here -- it's just a bad
assumption on my part in writing the test.  I'll see if there's another
struct type I can use that will be more robust...

Josh