Messages in this thread |  | | | Date | Fri, 07 Mar 1997 12:42:55 -0500 | | From | Joerg Pommnitz <> | | Subject | Re: x86 MSRs - example program |
| |
Matthias Urlichs wrote: > > In linux.dev.kernel, article <m0w1e9H-000AA6C@cerebro>, > mlehmann@hildesheim.sgh-net.de (Marc Lehmann) writes: > > >> ON the other hand... security hole????? how???????? > > >> > > >Cryptography. If you know exactly how long a given algorithm takes for > > >different inputs, you can sometimes infer clues about the secret key. > > > > gettimeofyday is always available, and in that scenario is just > > as good as rdtsc. > > > gettimeofday isn't nearly as accurate. While that's not an absolute > barrier, it raises the number of iterations required to get accurate data.
A few random length busy loops are enough to protect against a Kocher attack. This is certainly not a reason to prevent ordinary users from access to high resolution timers.
Best regards Joerg -- Joerg Pommnitz |Joerg@Raleigh.ibm.com| Never attribute to malloc Mobile/Wireless |Dept UMRA | that which can be adequately T/L (919)254-6397|Office B502/E117 | explained by stupidity.
|  |