Re: [PATCH 2/2] arm/virt/acpi: remove _ADR from devices identified by _HID

From: Guoheyi <guoheyi@huawei.com>
To: Peter Maydell <peter.maydell@linaro.org>,
	"Michael S. Tsirkin" <mst@redhat.com>
Cc: Corey Minyard <cminyard@mvista.com>,
	QEMU Developers <qemu-devel@nongnu.org>,
	Shannon Zhao <shannon.zhaosl@gmail.com>,
	qemu-arm <qemu-arm@nongnu.org>,
	Igor Mammedov <imammedo@redhat.com>,
	wanghaibin.wang@huawei.com
Subject: Re: [PATCH 2/2] arm/virt/acpi: remove _ADR from devices identified by _HID
Date: Wed, 15 Jan 2020 10:03:13 +0800	[thread overview]
Message-ID: <bbf347fe-c4a4-c5bb-19c2-3a18571b658f@huawei.com> (raw)
In-Reply-To: <CAFEAcA-sduqVO3rrG2V1VsysE2chgd0SnSySvEXFfue-aZN8dg@mail.gmail.com>

Hi Peter, Michael,

Have we come to conclusion on how to submit patches for ARM ACPI tables?

Some rough thoughts: is it possible to use the disassembled ASL file as 
the 'golden master' data? One problem I can imagine is that this may 
introduce dependency on the version of iASL tool. If so, how about 
adding acpica source code as a submodule, just like what we did for the 
device tree compile "dtc".

There may be much more which I missed; looking forward to your comments.

Thanks,

Heyi

在 2020/1/6 23:51, Peter Maydell 写道:
> On Sun, 5 Jan 2020 at 12:34, Michael S. Tsirkin <mst@redhat.com> wrote:
>> On Thu, Dec 19, 2019 at 02:47:59PM +0800, Heyi Guo wrote:
>>>   hw/arm/virt-acpi-build.c          |   8 --------
>>>   tests/data/acpi/virt/DSDT         | Bin 18449 -> 18426 bytes
>>>   tests/data/acpi/virt/DSDT.memhp   | Bin 19786 -> 19763 bytes
>>>   tests/data/acpi/virt/DSDT.numamem | Bin 18449 -> 18426 bytes
>>>   4 files changed, 8 deletions(-)
>> Please do not include binary changes in acpi patches.
>>
>> See comment at the top of tests/bios-tables-test.c for documentation
>> on how to update these.
> If you want the patches not to include binary changes then
> you will need to take these yourself through your own tree.
> As the Arm subtree maintainer I am not going to follow a
> specific process for acpi related patches that requires me
>   to do anything other than "apply patches from email, test
> them, send pull request". I also have no way to
> identify whether any differences that I might see if I
> disassembled the ACPI tables make sense, as that comment
> suggests I should be doing. The differences in the tables need
> to be checked by the people reviewing the patches, which will
> not be me for anything ACPI related -- I just don't know
> enough about the ACPI specs.
>
> Patches should be self contained, including updating test
> cases as required. The underlying problem here is that
> the 'golden master' data for the acpi tests is a pile
> of binary blobs rather than something that's human
> readable and reviewable as part of a patch.
>
> thanks
> -- PMM
>
> .