From: gneuner2@nospicedham.comcast.net   
      
   On Sat, 13 Jan 2018 22:59:50 +0100, "Walter H."   
    wrote:   
      
   >On 03.01.2018 19:38, Rod Pemberton wrote:   
   >>   
   >> Apparently, Intel processor's for over the past decade are allowing   
   >> speculative execution of code without any privilege checks.  The exact   
   >> specifics of the flaw are apparently still secret.  The software   
   >> patches to fix these issues for Windows, Linux, and Mac's are believed   
   >> to kill processor performance by up to 30%.   
   >   
   >you're talking about meltdown and/or spectre;   
   >   
   >just a primitive question:   
   >   
   >as you just mentioned that there will be patches for Windows, Linux, ...   
   >so it is assumed the problem doesn't exist on patched systems;   
   >   
   >but: isn't this a problem of these operating systems itself?   
      
   No, it's a problem of information leakage by speculative excution   
   through covert side channels in the hardware.   
      
   It is true that the operating systems did nothing to mitigate the   
   problem(s), but then the OS developers didn't know about them.   
      
      
   Meltdown - which is a [more or less] Intel specific OoO speculative   
   execution issue - can be addressed in software by a combination of   
   using separate kernel and user space page tables, and changing the   
   layout of the kernel entry trampoline functions.   
   [i.e. there will be 2 sets of page tables per process.]   
      
      
   Spectre is much more general problem that requires hardware fixes (or   
   redesigns) to prevent information leakage from speculation via the   
   BTB, the TLB, the caches, etc. - essentially any system resource that   
   is shared by all processes.   
      
   Intel's current solution is to flush the BTB at every context switch.   
   But that drastically impacts branch prediction and more importantly it   
   only mitigates the *known* attacks.  There may be other exploitable   
   attack vectors that it does not fix.   
      
   There's very little that software developers can do about Spectre on   
   their own.  Some compiler changes have been suggested that can make   
   the *known* attack vectors harder to exploit from compiled code ...   
   but even given new compilers and all software having been recompiled,   
   that STILL would not address hand written ASM or any as yet *unknown*   
   attacks based on Spectre.   
      
      
   >or in other words, I found a programme   
   >https://github.com/ionescu007/SpecuCheck   
   >which gives informations if the problem exists or not;   
   >   
   >but what does it say, when it has requirements, the OS hasn't - e.g. on   
   >old OS like Windows XP?   
   >   
   >I can't compile it for WinXP ...   
      
   Haven't looked at it.   
      
   >Thanks,   
   >Walter   
      
   George   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)