Messages in this thread | | | Date | Thu, 7 Apr 2011 18:42:45 +0200 | From | Andi Kleen <> | Subject | Re: [RFT/PATCH v2 2/6] x86-64: Optimize vread_tsc's barriers |
| |
> But before that, I would check that the second lfence is needed AT > ALL. I don't know whether we ever tried the "only barrier before" > version, because quite frankly, if you call this thing in a loop, then
> a single barrier will still mean that there are barriers in between > iterations. So...
The testers will call it in a loop, but real users (e.g. using it for a logging timestamp) may not.
I'm sure a single barrier would have fixed the testers, as you point out, but the goal wasn't to only fix testers.
-Andi -- ak@linux.intel.com -- Speaking for myself only.
| |