On Wed, Nov 26, 2014 at 05:07:20PM +0100, Andrew Jones wrote:

On Tue, Nov 25, 2014 at 04:10:02PM +0000, Alex Benn?e wrote:

quoted

This adds support for SW breakpoints inserted by userspace.

First we need to trap all BKPT exceptions in the hypervisor (ELS). This
in controlled through the MDCR_EL2 register. I've added a new field to
the vcpu structure to hold this value. There should be scope to
rationlise this with the VCPU_DEBUG_FLAGS/KVM_ARM64_DEBUG_DIRTY_SHIFT
manipulation in later patches.

I think we should start using the new mdcr_el2 field everywhere we plan
to within the same series that it is introduced. Otherwise it's hard
to tell if we need an mdcr_el2 field, or if a more generic field would
suffice. We can always translate bits of a more generic field to
mdcr_el2 bits when necessary, but not the reverse.

Agreed, this is getting messy already with this patch.

-Christoffer