Messages in this thread | | | Date | Thu, 15 May 2014 12:51:36 -0700 | From | "H. Peter Anvin" <> | Subject | Re: [Xen-devel] [PATCH 7/9] x86: skip check for spurious faults for non-present faults |
| |
On 05/15/2014 12:22 PM, Keir Fraser wrote: >> >> Are we chasing hardware errata here? Or did someone go off and *assume* >> that the x86 hardware architecture work a certain way? Or is there >> something way more subtle going on? > > See Intel Developer's Manual Vol 3 Section 4.10.4.3, 3rd bullet... This > is expected behaviour, probably to make copy-on-write faults faster. >
Hm, yes. My memory of this comes from before these formal rules were written down... I guess there is some wiggle room in there, presumably as you say, for performance reasons (or implementation leeway, which is another way to say performance.)
This does make a P bit switch architecturally different from W or NX, so I'm okay with that, but I would like the patch adjusted in the following ways:
1. Put in an explicit comment about the architectural difference between the P bit on one hand and an W and NX on the other; an SDM reference is good, and *why* this makes the specific filtering correct.
2. Please use the standard format for multiline comments;
/* * blah * blah */
With that this should be okay.
-hpa
| |