From mboxrd@z Thu Jan  1 00:00:00 1970
From: Alexandru Elisei <alexandru.elisei@arm.com>
Subject: Re: [kvm-unit-tests PATCH 5/7] lib: arm: Fallback to
 psci_system_off() in exit()
Date: Fri, 25 Jan 2019 15:51:03 +0000
Message-ID: <15da35cb-c047-6e88-d0b7-5d2e68e34133@arm.com>
References: <20190124111634.4727-1-alexandru.elisei@arm.com>
 <20190124111634.4727-6-alexandru.elisei@arm.com>
 <20190124130020.5gjoblt5lqqujzqw@kamzik.brq.redhat.com>
 <20190124133511.3izvap4y2crqrbbp@kamzik.brq.redhat.com>
 <6cf2fe9d-9c35-ae42-6cb4-d34dfdc2df3b@arm.com>
 <20190125153137.dakwla2ozmd22vq2@kamzik.brq.redhat.com>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
Cc: andre.przywara@arm.com, kvmarm@lists.cs.columbia.edu, kvm@vger.kernel.org
To: Andrew Jones <drjones@redhat.com>
Return-path: <kvmarm-bounces@lists.cs.columbia.edu>
In-Reply-To: <20190125153137.dakwla2ozmd22vq2@kamzik.brq.redhat.com>
Content-Language: en-US
List-Unsubscribe: <https://lists.cs.columbia.edu/mailman/options/kvmarm>,
 <mailto:kvmarm-request@lists.cs.columbia.edu?subject=unsubscribe>
List-Archive: <https://lists.cs.columbia.edu/pipermail/kvmarm>
List-Post: <mailto:kvmarm@lists.cs.columbia.edu>
List-Help: <mailto:kvmarm-request@lists.cs.columbia.edu?subject=help>
List-Subscribe: <https://lists.cs.columbia.edu/mailman/listinfo/kvmarm>,
 <mailto:kvmarm-request@lists.cs.columbia.edu?subject=subscribe>
Errors-To: kvmarm-bounces@lists.cs.columbia.edu
Sender: kvmarm-bounces@lists.cs.columbia.edu
List-Id: kvm.vger.kernel.org


On 1/25/19 3:31 PM, Andrew Jones wrote:
> On Fri, Jan 25, 2019 at 02:56:30PM +0000, Alexandru Elisei wrote:
>> On 1/24/19 1:35 PM, Andrew Jones wrote:
>>> On Thu, Jan 24, 2019 at 02:00:20PM +0100, Andrew Jones wrote:
>>>> [..]
>>>> chr_testdev_init() ensures vcon is NULL if it fails to initialize.
>>>> chr_testdev_exit() immediately returns if vcon is NULL. This was
>>>> done by design to allow fallback exits to be placed below the
>>>> chr_testdev_exit call, e.g. halt().
>>>>
>>>> We should be able to drop patch 3/7 and change exit() to this
>>>>
>>>>   void exit(int code)
>>>>   {
>>>>       chr_testdev_exit(code);
>>>>       psci_system_off();
>>>>       halt(code);
>>>>       __builtin_unreachable();
>>>>   }
>>>>
>>> There's also a framework for exits that can't return status codes. powerpc
>>> uses it. Before exiting with psci_system_off we need to make this print
>>> statement
>>>
>>>  printf("\nEXIT: STATUS=%d\n", ((code) << 1) | 1); 
>>>
>>> And run_qemu in arm/run needs to be changed to run_qemu_status. It's
>>> hacky, but maybe we can live with it until kvmtool offers some sort of
>>> debug exit.
>>>
>>> Thanks,
>>> drew
>> I can make the change, but if I understand the scripts/arch-run.bash code
>> correctly, run_qemu() will check if QEMU was terminated because of a signal and
>> adjust the return code to take that into account. Using run_qemu_status() means
>> that check won't be made when the tests are run under QEMU, is that acceptable?
>>
> run_qemu_status() first calls run_qemu(). If the return value from
> run_qemu() is 1, then it'll change the value to whatever the EXIT:
> status line says it should be. If run_qemu() detected that a signal
> caused the exit, then the return value won't be 1. In that case
> run_qemu_status() will simply propagate the value to the caller.
>
> It might be worth doing a few tests to ensure that all works out as
> designed, but I'm pretty sure it should.
>
> Thanks,
> drew

Now it makes more sense, thank you. I'll make the change and run some QEMU and
kvmtool tests.