Today I noticed a random bit-flip for the third time in my life! A "5" was changed to a "7" in the bank account number on invoices I sent. That could have had pretty big consequences!! Luckily the number was on the invoice twice and the buyer noticed the two were different.
I found out about this latest one because it was in a file. That file was generated from a template file, which also contained the flipped bit. So, it was either flipped on-disk in the template file or in memory when I last edited it and then saved to disk.
-
-
From what I understand bit-flips in cheap non-ECC memory are not that uncommon. Especially if you "hammer" the memory :). I guess the high temperatures from fuzzing at 100% 24/7 also does not help. (The later does not apply to my desktop OC).
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Cool. Do we have any assessment/guess on how likely a bit flip is to occur under normal (non-RowHammer) circumstances in a given timeframe on a given media? Is "bit-flip fishing" even theoretically a viable attack, e.g., for IP addresses, domain names, or "is_admin" variables?
-
I think there are too many factors affecting this to give a useful number or even a range that can be applied to a single person or machine. But you could squad bit-flipped domains like "http://googlu.com " to gather a number that applies to machines in general I guess.
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.