[openib-general] Re: Re: Userspace testing results (manykernels, many svn trees)

[Nishanth Aravamudan]

On 24.01.2006 [11:24:02 -0800], Roland Dreier wrote:
>     Michael> Thats what trunk head does, but Nishanth Aravamudan here
>     Michael> sees
> 
>     Michael> 1 sec = 5.37731e+14 usec
> 
>     Michael> which seems to indicate something's still wrong.
> 
> Nish, did everything get rebuilt?  Since the makefile doesn't have
> dependencies, I can do things like:
> 
>     $ make clock_test
>     cc -Wall -g -D_GNU_SOURCE -O2    clock_test.c get_clock.c   -o clock_test
>     $ touch get_clock.h
>     $ make clock_test
>     make: `clock_test' is up to date.
> 
> so a simple "svn up" followed by "make" make not be enough to pick up
> all changes.

I'm *almost* certain it got rebuilt, as this is the current flow:

I submit a job to the testing grid:

	The grid sends the job to the two machines:

		Each machine does:

			A fresh kernel build, and reboots into it, then

			Grabs a tarball of the svn tree and untars, then
			runs make inside, first rm -rf the old tree.

So, in essence, it starts from scratch *every* time. I'll run the test
again and see what happens.

Thanks,
Nish