Re: [Xen-devel] [PATCH V4] x86/altp2m: Fix crash with INVALID_ALTP2M EPTP index

[Razvan Cojocaru <rcojocaru@xxxxxxxxxxxxxxx>]

On 06/27/2018 06:04 PM, Jan Beulich wrote:
>>>> On 27.06.18 at 15:12, <rcojocaru@xxxxxxxxxxxxxxx> wrote:
>> xc_altp2m_set_vcpu_enable_notify() ends up calling
>> altp2m_vcpu_update_vmfunc_ve(), which sets the
>> SECONDARY_EXEC_ENABLE_VIRT_EXCEPTIONS bit on
>> vmx_secondary_exec_control. A subsequent call to
>> xc_altp2m_set_domain_state(..., false) (i.e. disabling altp2m
>> for the domain) ends up calling altp2m_vcpu_destroy(), which
>> calls (in this order) altp2m_vcpu_reset() (which sets the
>> current EPTP index to INVALID_ALTP2M), altp2m_vcpu_update_p2m()
>> (which __vmwrite()s EPTP_INDEX as INVALID_ALTP2M if
>> SECONDARY_EXEC_ENABLE_VIRT_EXCEPTIONS is set), and
>> altp2m_vcpu_update_vmfunc_ve() (which finally clears
>> SECONDARY_EXEC_ENABLE_VIRT_EXCEPTIONS).
> 
> I continue to consider this paragraph unreadable, but perhaps it's
> just me. The rest of the description looks reasonable to me now.

If you (or anyone else) have something specific in mind I'd be happy to
change it to that, otherwise I can also try again (the only concern is
that I might unwantedly continue to be unable to guess correctly at the
desired balance between technical clarity (detail) and general readability).

Is "A VM entry handler executed immediately after enabling #VE might
find a stale __vmsave()d EPTP_INDEX, stored by calling
altp2m_vcpu_destroy() when SECONDARY_EXEC_ENABLE_VIRT_EXCEPTIONS had
been set by altp2m_vcpu_update_vmfunc_ve()." a better first paragraph
perhaps?

> Please allow some more time than a single day between versions
> though, so others also have a chance to respond.

Sorry about that, I misremembered that I had sent V3 yesterday.


Thanks,
Razvan

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/xen-devel