Comment 11 for bug 855199

For reference's sake, one patch that does not seem to be the culprit:

From the list of 2.6.39 changes at kernelnewbies.org, the one IRQ-related change I could pick out was this:

  genirq: Provide forced interrupt threading

  http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=8d32a307e4faa8b123dc8a9cd56d1a7525f69ad3

I built a 2.6.39 with that patch reverted. It made no difference; the regression was present.