Re: non-automated test failure report! :)

[christos%astron.com@localhost (Christos Zoulas)]

In article <20162.41313.434442.283665%guava.gson.org@localhost>,
Andreas Gustafsson  <gson%gson.org@localhost> wrote:
>Jukka Ruohonen wrote:
>> On Tue, Nov 15, 2011 at 09:07:45AM -0800, Paul Goyette wrote:
>> > Original report indicated failure on i386:
>> > 
>> > lib/libtre/t_exhaust (388/485): 1 test cases
>> >     regcomp_too_big: UVM: pid 20107 (t_exhaust), uid 0 killed: out of swap
>> > Failed: Test program received signal 9
>> > 
>> > Looks like a test-bed failure?
>> 
>> Can you take a look, Andreas?
>
>It seems pretty clear to me what is happening: the test program is
>using a lot of memory and is being killed because the machine is
>running out of swap.
>
>What's not clear to me is why it's failing in this particular way
>instead of e.g. malloc() returning NULL, and what to do about it.
>
>I'm reluctant to work around this issue by increasing the memory size
>of the test VM from the current 32 MB, because having a limited amount
>of memory has helped find real regressions such as a huge memory leak
>in the /bin/sh yacc parser a while back.  If the test really needs
>that much memory, maybe it should be skipped on low-memory machines.

Should we add something like:

        atf_tc_set_md_var(tc, "required-memory", "256M");

and then have the driver not run the test if sysctl("hw.usermem64") < 256M
Seems useful to me instead of putting the check in the test?

christos