After finding and removing the workaround I had still in place on lucid I was able to reproduce the bug with the current updated lucid kernel 2.6.32-25.
Testing under similar circumstances with the proposed 2.6.32-26 did not show any buggy behaviour on 7 suspend resume cycles.
On my hardware (same as in comment #3) this proposed kernel fixes the bug.
After finding and removing the workaround I had still in place on lucid I was able to reproduce the bug with the current updated lucid kernel 2.6.32-25.
Testing under similar circumstances with the proposed 2.6.32-26 did not show any buggy behaviour on 7 suspend resume cycles.
On my hardware (same as in comment #3) this proposed kernel fixes the bug.
Thanks to all.