marcello kernel &ppc64

[Troy Benjegerdes]

On Fri, Nov 14, 2003 at 03:07:27PM -0600, Stephanie Glass wrote:
>
> Tony wrote:
> >
> >> Which takes me to another, ahem, politically touchy point, I suspect.
> >> There's a test lab here that pounds the crap out of SuSE kernels.
> >> Speaking from personal experience, 9 out of 10 or 19 out of 20 of the
> >> kernel crashes & hangs that they find are *not* in the ppc64 code,
> >> but are in the generic linux kernel code. These are races of various
> >> sorts, missing locks, data corruptions, you name it, I've seen it.
> >> These get fixed in the SuSE kernel.
> >
> >What we need is this testing lab also pounding the crap out of marcelo
> >kernels..
> >
> >I was hoping the Linux Test project and the stuff OSDL is doing would
> >wind up with a 'test suite' that mere mortals with real work to do
> >could easily run.
> >
> >Personally, I'd love to be able to build every revision that gets
> >pushed into the ameslab tree, and then run regression tests on it. And
> >then do the same thing for the mainline kernels.
> >
> >It would be a lot easier to defend patches when you can show test logs
> >of kernel.org kernels failing. OSDL and other internal testing labs
> >are nice and all, but if we want the mainline kernel fixed, we have to
> >either test mainline kernels, or make it easy for LKML people to get
> >access to test results.
>
> I am not sure what you want Tony. Right now, the LTP is run on the x-86
> mainline kernel every night. We think that the LTP is something that can
> be easily run by mere mortals. If you have any specific problem, please
> let us know and we will work to correct. We do not run nightly on ppc64
> due to a lack of hardware not desire.

Well, honestly, I haven't tried it. But what I'd really like to be able
to do is:

Download 'linux-test-thingy.tar.gz' (or get via cvs/bk/whatever)

extract it somewhere, export via nfs/whatever, then set up DHCP and tftp
to to support netboot of some target machines, and maybe a short shell
script to be able to power-cycle the target machine, as well as
recording serial console output.

Then, point it to a kernel tree to start running tests.

Provided I already have DHCP, tftp, console, and power cycling set up,
I'd like this to take less than an hour or so to get running.

Otherwise, I'd be happy with a RedHat rpm or debian package I can
install on one of my cluster nodes that installs all the tests and some
reporting tools or somesuch.

My impression of where things are now is it would take me at least a
couple of days to get something usable up and running.

> The LTC Test group right now picks up the Rochester p kernel builds (2.6
> based) as they come out and runs many different tests on them. Is that
> the tree you want runs done on or is there another place to pick up the
> kernels you are talking about. If you contact me directly and give me
> the details we will try and do everything we can to help.


** Sent via the linuxppc64-dev mail list. See http://lists.linuxppc.org/