From mboxrd@z Thu Jan  1 00:00:00 1970
From: Gleb Natapov <gleb@redhat.com>
Subject: Re: [PATCH] kvm-unit-tests : The first version of VMX nested test
 case
Date: Tue, 16 Jul 2013 19:45:20 +0300
Message-ID: <20130716164519.GE8981@redhat.com>
References: <1373966841-22206-1-git-send-email-yzt356@gmail.com>
 <51E52035.1050401@redhat.com>
 <20130716152035.GB8981@redhat.com>
 <CABpY8M+k0KzN4EHZFuVz6WrGrzoEYSbw=Skin1sWPKKEV_nw2A@mail.gmail.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Cc: Paolo Bonzini <pbonzini@redhat.com>, kvm <kvm@vger.kernel.org>,
	Jan Kiszka <jan.kiszka@web.de>
To: Arthur Chunqi Li <yzt356@gmail.com>
Return-path: <kvm-owner@vger.kernel.org>
Received: from mx1.redhat.com ([209.132.183.28]:33164 "EHLO mx1.redhat.com"
	rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP
	id S933133Ab3GPQpX (ORCPT <rfc822;kvm@vger.kernel.org>);
	Tue, 16 Jul 2013 12:45:23 -0400
Content-Disposition: inline
In-Reply-To: <CABpY8M+k0KzN4EHZFuVz6WrGrzoEYSbw=Skin1sWPKKEV_nw2A@mail.gmail.com>
Sender: kvm-owner@vger.kernel.org
List-ID: <kvm.vger.kernel.org>

On Tue, Jul 16, 2013 at 11:29:20PM +0800, Arthur Chunqi Li wrote:
> On Tue, Jul 16, 2013 at 11:20 PM, Gleb Natapov <gleb@redhat.com> wrote:
> > On Tue, Jul 16, 2013 at 12:28:05PM +0200, Paolo Bonzini wrote:
> >> > +void vmx_exit(void)
> >> > +{
> >> > +   test_vmxoff();
> >> > +   printf("\nSUMMARY: %d tests, %d failures\n", tests, fails);
> >> > +   exit(fails ? -1 : 0);
> >> > +}
> >>
> >> Can you try to jump back to main, and do test_vmxoff there?  This will
> >> avoid having to write our tests in callback style, which is a pain.
> >> Basically something similar to setjmp/longjmp.  In main:
> >>
> >>       if (setjmp(jmpbuf) == 0) {
> >>               vmx_run();
> >>               /* Should not reach here */
> >>               report("test vmlaunch", 0);
> >>       }
> >>       test_vmxoff();
> >>
> >> exit:
> >>       printf("\nSUMMARY: %d tests, %d failures\n", tests, fails);
> >>       return fails ? 1 : 0;
> >>
> >> In vmx_handler:
> >>
> >>       case VMX_HLT:
> >>               printf("\nVM exit.\n");
> >>               longjmp(jmpbuf, 1);
> >>
> > Why not just make vmexit occur after vmlaunch/vmresume like KVM does. It
> > will make code much more straightforward and easer to follow.
> The concept "easier to follow" may have different meanings in
> different view. This achievement puts all the test cases in main
> function instead of scattering everywhere, which is another view to
> "easy to follow". As this is just a test case, I prefer this one.
> 
I do not see why what I propose will prevent you to put all tests into main.

vmx_run() will looks like that:

   vmlaunch
   while(1) {
       vmresume
         <---- vmexit jumps here
       switch(exit reason) {
          case reason1:
          break;
          case reason2:
          break;
          case HLT
          return;
       }
   }          
          
> Besides, this way we can start another VM following the previous one
> simply in main function. This is flexible if we want to test re-enter
> to VMX mode or so.
> 
That's what I am missing. How do one writes more tests now?

I was thinking about interface like that:

guest_func_test1()
{
}

tes1t_exit_handlers[] = {test1_handle_hlt, test1_handle_exception, ....}

main()
{

   init_vmcs(); /* generic stuff */
   init_vmcs_test1(); /* test1 related stuff */
   r = run_in_guest(guest_func_test1, test1_exit_handlers);
   report("test1", r);
}

--
			Gleb.