Proposed fix to protect the kernel from Spectre (& Meltdown).pic.twitter.com/qQZr4Iw2yw
You can add location information to your Tweets, such as your city or precise location, from the web and via third-party applications. You always have the option to delete your Tweet location history. Learn more
I actually once helped a friend in India with an old PC that was "hanging" starting Windows. Slowness of the initial logo fade-in or whatever gave me a clue...
Somehow the BIOS setup had an option to disable cache and it had gotten disabled. I think it eventually booted after 6 hours or so.
no, you use MTRRs to restrict which mappings are cacheable depending on user or kernel context, then do a CLFLUSH when entering/exiting kernel
MTRRs are for physical ranges, not virtual, no? But yes just flushing cache on kernel enter/return would work if you disable HT.
yes, physical, but the memory manager can handle the details. using PATs would be even more flexible
It really can't. That completely breaks logical layering and makes it hopelessly complex and therefore bug-ridden. And it probably exceeds limit on # of MTRRs.
honestly CLFLUSH would probably be "good enough" if we could CLFLUSH the sibling threads (on the same physical core) too. i wonder if there is a way to synchronize sysenter/sysexit in this way using the topology data we already have
As long as HT is there, attacker will have opportunities to observe cache effects. Exploit runs on both threads, synchronizing via shared mem spinlocks, executes syscall on one while probing cache from other.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.