On Thu, 2011-02-10 at 16:57 -0800, Arun Sharma wrote:

On Thu, Feb 10, 2011 at 4:42 PM, Arnaldo Carvalho de Melo
[off-list ref] wrote:

quoted

Is it possible for you to test this with 2.6.38-rc4? At least the user
level tools, just do:

I tried the 2.6.38rc3 perf binary on two machines:

kernel = 2.6.37: The machine hung solid on the first try.
kernel = 2.6.38rc3: I could run it 10 times in a loop and no bad runs.

Looks like this is already fixed in newer kernels.

I couldn't actually reproduce on .37 with the given details, so while
its already fixed it would be nice to figure out what patch fixed it so
we can make sure it ends up in .37-stable.

Could you provide exact reproduction details or maybe bisect the thing?