This is the mail archive of the ecos-discuss@sources.redhat.com mailing list for the eCos 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: problem with multi_lo_select.c


sandeep <shimple0@yahoo.com> writes:

> following problem is even possible with non-smp configuration (patches
> for it have been sent two times in past years in the bundle of smp
> patches).
> 
> in dummy thread, s_s1 is static variable and access to it is
> unprotected. this can lead to more than 1 dummy threads enter "if
> (s_s1 < 0) { ...." and some part of test not behaving as expected.
> 
> in the wake of recent weeks bugs with ecos/ecos-tests, it would be
> fruitful to also examine cases for unprotected static/global variable
> accesses of similar nature in both, from no-smp and smp point of views.

I'll agree that that program is not particularly good in either SMP or
non-SMP systems.

Some of the tests will never work properly in an SMP environment,
since they test the behaviour of eCos by testing that certain threads
execute in an expected order, which may not be true on SMP.

I'm aware that you have submitted some patches which have not yet been
committed. I think that the best way forward now is to wait until you
are happy with your current tests and then for you to submit a single
patch against the current CVS. Depending on the size and scope of the
changes we may need an assignment from you to commit it. However, I am
keen to get the results of your work into the source base.

-- 
Nick Garnett                    eCos Kernel Architect
http://www.ecoscentric.com/     The eCos and RedBoot experts


-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss


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