On 01/06/17 13:11, Jan Beulich wrote:
> Commit aac1df3d03 ("x86/HVM: introduce hvm_get_cpl() and respective
> hook") went too far in one aspect: When emulating a task switch we
> really shouldn't be looking at what hvm_get_cpl() returns, as we're
> switching all segment registers.
>
> However, instead of reverting the relevant parts of that commit, have
> the caller tell the segment loading function what the new CPL is. This
> at once fixes ES being loaded before CS so far having had its checks
> done against the old CPL.
>
> Reported-by: Andrew Cooper <andrew.coop...@citrix.com>
> Signed-off-by: Jan Beulich <jbeul...@suse.com>On further consideration, wouldn't it be better to audit all segment registers, before updating any of them in the vmcs/vmcb? This would leave us with a far lower chance of other vmentry failures. Loading the segment registers is beyond the commit point of a task switch, and the manual says that the processor will try to skip further segmentation checks in an attempt to deliver a fault in the new context. ~Andrew _______________________________________________ Xen-devel mailing list Xen-devel@lists.xen.org https://lists.xen.org/xen-devel
