Forums before death by AOL, social media and spammers... "We can't have nice things"
|    comp.lang.asm.x86    |    Ahh, the lost art of x86 assembly    |    4,675 messages    |
[   << oldest   |   < older   |   list   |   newer >   |   newest >>   ]
|    Message 3,837 of 4,675    |
|    Terje Mathisen to All    |
|    Re: x86 memory testing - dealing with ca    |
|    05 Apr 19 21:33:13    |
      From: terje.mathisen@nospicedham.tmsw.no              src153 wrote:       > On Fri, 5 Apr 2019 09:04:35 +0200, Terje Mathisen wrote:       >       >> You have of course considered Rowhammer and similar attacks?       >       > They say Rowhammer can defeat ECC too, with a 3-bit flip, because most       > ECC can only detect 1 or 2 bit errors.       >       >       >> shown to cause errors in otherwise perfectly OK memory chips       >       > It's like airplanes. Just when they tell you it's safe to fly, whoops       > there goes another nosedive.       >       >       Ouch.              The preliminary report from the ET flight (the second crash) indicates       that the pilots knew about the emergency procedures to be used in case       of MCAS trouble, applied them and still could not avoid a nosedive.              As programmers we really need to consider the possible error scenarios       our programs could experience, and what the consequences could be.              Just once in my life have I been called in to trouble-shoot a sw error       which was a real life-or-death problem:              A new hospital in Oslo had experienced a total failure of the emergency       notification system which should gather the crash surgical team, in this       case an immediate C-section was delayed because the surgeon could not be       contacted.              Thankfully I did figure that one out within a few days, used that       knowledge to get a workaround in place and flew from Norway to the US to       spend a week in the lab of one of the involved vendors.              We developed, tested out and verified a solid sw fix for a problem which       was really caused by another, larger (Canadian) vendor and got that       installed on all the 1200-1500 personal devices used by the medical staff.              Terje              --       - |
[   << oldest   |   < older   |   list   |   newer >   |   newest >>   ]
(c) 1994, bbs@darkrealms.ca