Messages in this thread | | | Date | Thu, 14 Jun 2012 14:13:55 -0700 | From | "H. Peter Anvin" <> | Subject | Re: bugs in page colouring code |
| |
On 06/14/2012 02:03 PM, Rik van Riel wrote: > On 06/14/2012 04:58 PM, H. Peter Anvin wrote: >> On 06/14/2012 05:57 AM, Rik van Riel wrote: >>> >>> However, I expect that on x86 many applications expect >>> MAP_FIXED to just work, and enforcing that would be >>> more trouble than it's worth. >>> >> >> MAP_FIXED, is well, fixed. It means that performance be screwed, if we >> can fulfill the request we MUST do so. > > My codebase now has a separate arch_align_addr function > for x86, which is surprisingly similar to the generic > one :)
I am much more skeptical to disabling page coloring in the !PF_RANDOMIZE case when no address hint is proposed. I would like to at least try running without it, perhaps with a chicken bit in a sysctl.
-hpa
| |