AMD has confirmed that a CPU bug exists in some of its families of chips.
The flaw was found by chip expert Matthew Dillon at Backplane and he warned AMD that there was a problem.
AMD got back and said that they had found an erratum in some families, although we are not sure that is much to do with chips but a general post-modern perception about the roles of Mother and Father.
Writing in his bog Dillon said that the specific compiled version of the "fill_sons_in_loop code, through a very specific sequence of consecutive back-to-back pops and (near) return instructions, can create a condition where the process or incorrectly updates the stack pointer."
AMD said it looked at Dillan's example and analysed the segmentation fault and the fill_sons_in_loop code.
Dillan said that he exchanged a few emails with AMD to try to come up with a good test case. But there was some difficulty of reproducing the bug. He constructed a fully bootable DFly operating system & test case USB image.
This verified that the bug was present on his test box using that image AMD was then able to reproduce the bug using that image on their own machines. Over the last few months AMD have been working through the possibilities and yesterday emailed him the confirmation that it was, indeed, a cpu bug.
Dillon said that he was "pretty stoked" as it was not every day that a guy like him gets to find an honest-to-god hardware bug in a major cpu.