linux-kernel - Re: [RFC PATCH 1/3] kvm: x86: Rename KVM_DEBUGREG_RELOAD to KVM_DEBUGREG_NEED

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Mon, 27 Apr 2020 10:37:32 -0400
From:   Peter Xu <peterx@...hat.com>
To:     Paolo Bonzini <pbonzini@...hat.com>
Cc:     Sean Christopherson <sean.j.christopherson@...el.com>,
        Xiaoyao Li <xiaoyao.li@...el.com>, kvm@...r.kernel.org,
        Nadav Amit <namit@...technion.ac.il>,
        Vitaly Kuznetsov <vkuznets@...hat.com>,
        linux-kernel@...r.kernel.org
Subject: Re: [RFC PATCH 1/3] kvm: x86: Rename KVM_DEBUGREG_RELOAD to
 KVM_DEBUGREG_NEED_RELOAD

On Sat, Apr 25, 2020 at 09:48:17AM +0200, Paolo Bonzini wrote:
> On 24/04/20 22:21, Peter Xu wrote:
> > But then shouldn't DIRTY be set as long as KVM_DEBUGREG_BP_ENABLED is set every
> > time before vmenter?  Then it'll somehow go back to switch_db_regs, iiuc...
> > 
> > IIUC RELOAD actually wants to say "reload only for this iteration", that's why
> > it's cleared after each reload.  So maybe...  RELOAD_ONCE?
> > 
> > (Btw, do we have debug regs tests somewhere no matter inside guest or with
> >  KVM_SET_GUEST_DEBUG?)
> 
> What about KVM_DEBUGREG_EFF_DB_DIRTY?

The problem is iiuc we always reload eff_db[] no matter which bit in
switch_db_regs is set, so this may still not clearly identify this bit from the
rest of the two bits...

Actually I think eff_db[] is a bit confusing here in that it can be either the
host specified dbreg values or the guest specified depends on the dynamic value
of KVM_GUESTDBG_USE_HW_BP.

I am thinking maybe it's clearer to have host_db[] and guest_db[], then only
until vmenter do we load either of them by:

  if (KVM_GUESTDBG_USE_HW_BP)
    load(host_db[]);
  else
    load(gueet_db[]);

Then each db[] will be very clear on what's the data is about.  And we don't
need to check KVM_GUESTDBG_USE_HW_BP every time when accessing eff_db[].

> 
> We have them in kvm-unit-tests for debug regs inside the guest, but no
> selftests covering KVM_SET_GUEST_DEBUG.

I see!  Thanks,

-- 
Peter Xu