On 03.12.2021 12:48, Andrew Cooper wrote: > On 03/12/2021 11:21, Jan Beulich wrote: >> The SDM explicitly permits this, and since that's sensible behavior >> don't special case AMD (where the PM doesn't explicitly say so). > > APM explicitly says so too. > > "The CLFLUSH instruction executes at any privilege level. CLFLUSH > performs all the segmentation and paging checks that a 1-byte read would > perform, except that it also allows references to execute-only segments." > > and > > "The CLFLUSHOPT instruction executes at any privilege level. CLFLUSHOPT > performs all the segmentation and paging checks that a 1-byte read would > perform, except that it also allows references to execute-only segments."
Somehow I didn't read further after the page table related paragraph, perhaps on the assumption that like in the SDM it would be all in one paragraph. >> Fixes: 52dba7bd0b36 ("x86emul: generalize wbinvd() hook") >> Reported-by: Andrew Cooper <andrew.coop...@citrix.com> >> Signed-off-by: Jan Beulich <jbeul...@suse.com> > > With the commit message tweaked, Reviewed-by: Andrew Cooper > <andrew.coop...@citrix.com>. Far less invasive than I was fearing. Thanks. I've switched to simply saying "Both SDM and PM explicitly permit this." Jan