Messages in this thread | | | Subject | Re: [GIT PULL] cputime patch for 2.6.30-rc6 | From | Peter Zijlstra <> | Date | Tue, 19 May 2009 11:00:35 +0200 |
| |
On Tue, 2009-05-19 at 10:49 +0200, Martin Schwidefsky wrote:
> Well, don't do stupid things like that. That falls into the same > category as programs calling gettimeofday all the time.
Is gtod a global DoS? The worry I have about these cputime for_each_cpu() loops in the proc code is that any odd unpriv user can take the whole machine down (provided the machine is large enough).
Granted, this would only be a real problem for SGI and their insane boxes atm, but it might very well be a problem for us in the near future, given the rate Intel seems to increase cores.
So, I'm really not objecting too much to the patch at hand, but I'd love to find a solution to this problem.
My personal favourite is to lower the resolution/accuracy of the answer the bigger the box gets.
| |