Re: [Qemu-devel] [PATCH v2 0/9] hw/block/pflash_cfi01: Add DeviceReset() handler

[Laszlo Ersek <lersek@redhat.com>]

On 07/02/19 15:41, Laszlo Ersek wrote:
> On 07/02/19 13:52, Laszlo Ersek wrote:
>> Hi Phil,
>>
>> On 07/02/19 02:12, Philippe Mathieu-Daudé wrote:
>>> The pflash device lacks a reset() function.
>>> When a machine is resetted, the flash might be in an
>>> inconsistent state, leading to unexpected behavior:
>>> https://bugzilla.redhat.com/show_bug.cgi?id=1678713
>>> Resolve this issue by adding a DeviceReset() handler.
>>>
>>> Fix also two minor issues, and clean a bit the codebase.
>>>
>>> Since v1: https://lists.gnu.org/archive/html/qemu-devel/2019-05/msg00962.html
>>> - addressed Laszlo review comments
>>>
>>> Maintainers spam list from:
>>> ./scripts/get_maintainer.pl -f $(git grep -El '(pflash_cfi01_register|TYPE_PFLASH_CFI01)')
>>>
>>> Regards,
>>>
>>> Phil.
>>>
>>> Philippe Mathieu-Daudé (9):
>>>   hw/block/pflash_cfi01: Removed an unused timer
>>>   hw/block/pflash_cfi01: Use the correct READ_ARRAY value
>>>   hw/block/pflash_cfi01: Extract pflash_mode_read_array()
>>>   hw/block/pflash_cfi01: Start state machine as READY to accept commands
>>>   hw/block/pflash_cfi01: Add the DeviceReset() handler
>>>   hw/block/pflash_cfi01: Simplify CFI_QUERY processing
>>>   hw/block/pflash_cfi01: Improve command comments
>>>   hw/block/pflash_cfi01: Replace DPRINTF by qemu_log_mask(GUEST_ERROR)
>>>   hw/block/pflash_cfi01: Hold the PRI table offset in a variable
>>>
>>>  hw/block/pflash_cfi01.c | 140 +++++++++++++++++++++-------------------
>>>  hw/block/trace-events   |   1 +
>>>  2 files changed, 74 insertions(+), 67 deletions(-)
>>>
>>
>> I'll do some regression-tests with this, using OVMF and ArmVirtQemu.
>>
>> I don't think I can usefully review the patches without getting lost in
>> the related spec(s), and I don't have capacity for that.
>>
>> Until I have regression test results, one question: are the changes to
>> the device model transparent with regard to migration? (You are not
>> introducing any compat properties.)
> 
> I didn't test migration.
> 
> With OVMF, I performed my usual Linux guest tests (partly described at
> <https://github.com/tianocore/tianocore.github.io/wiki/Testing-SMM-with-QEMU,-KVM-and-libvirt#tests-to-perform-in-the-installed-guest-fedora-26-guest>).
> I found no problems / discrepancies, in either guest behavior or
> firmware logs.
> 
> With ArmVirtQemu, I meant to test on KVM (pflash used to be really
> sensitive to KVM<->TCG differences), but my aarch64 hardware is
> apparently dying, and I wouldn't like to spend a day just to provision a
> remote aarch64 box. So, no test results on aarch64.

Managed to run a light regression-test on aarch64 KVM too, using the
ArmVirtQemu fw and an F28 guest.  (boot+reboot, without the patches and
with the patches; logs compared, behavior compared.) Everything seems fine.

Thanks
Laszlo

> With those caveats:
> 
> Regression-tested-by: Laszlo Ersek <lersek@redhat.com>
> 
> Thanks
> Laszlo
>