[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [PATCH] acpi: Add Windows ACPI Emulated Device Table (WAET)

From: Liran Alon
Subject: Re: [PATCH] acpi: Add Windows ACPI Emulated Device Table (WAET)
Date: Thu, 12 Mar 2020 20:48:48 +0200
User-agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.13; rv:68.0) Gecko/20100101 Thunderbird/68.5.0

On 12/03/2020 18:35, Igor Mammedov wrote:
On Thu, 12 Mar 2020 14:55:50 +0200
Liran Alon <address@hidden> wrote:

On 12/03/2020 14:19, Michael S. Tsirkin wrote:
On Thu, Mar 12, 2020 at 01:30:01PM +0200, Liran Alon wrote:
On 12/03/2020 8:12, Michael S. Tsirkin wrote:
On Thu, Mar 12, 2020 at 01:20:02AM +0200, Liran Alon wrote:
But this is just a good practice in general and in the past it was said by
maintainers that this is one of the main reasons that ACPI and SMBIOS
generation have moved from SeaBIOS to QEMU.
I think a flag to disable this might make sense though. For example,
some guests might behave differently and get broken.
Right. That's why I think it's a good practice to have this flag and tie it
to machine-type.
Tying things to the machine type is not what I had in mind.
A separate flag would also be helpful so users can tweak this
for new machine types, too.
I think it's unnecessary, given how common WAET ACPI table is exposed by
default by other hypervisors.

But if you insist, I can add such flag on a separate commit in v2...
Where do you want to have such flag? It cannot be a property of some
qdev object.
So you want to add a new QEMU_OPTION_no_weat in vl.c?
If it doesn't break any windows guests we probably don't need an option.
Can you test if old guests are booting fine with new table, to confirm
that it's fine? (starting with XPsp3)

Old guests boot fine with the new WAET table.
We are running with this table in production for many years with many Windows XP guests (and much more esoteric guests)

Just to verify, I've just now run it with a WinXP SP3 VM and it works just fine. So should I remove the flag completely or remain with the current functionality I have that makes sure WAET is only exposed on new machine-types?


Guest-visible changes shouldn't be exposed to old machine-types.
Well almost any change in qemu is guest visible to some level.
Even optimizations are guest visible.
We made changes in ACPI without versioning in the past but I'm not
opposed to versioning here. However in that case pls do add a bit
of documentation about why this is done here.
I remember that maintainers have explicitly specified that ACPI/SMBIOS
should not be changed between machine-types.
This have been one of the reasons to move ACPI/SMBIOS generation from
SeaBIOS to QEMU control.

What can of documentation you want me to add and where?
The only thing I can say is that I tie it to machine-type because I do
not think a given machine-type should suddenly change BIOS exposed info
to guest.
But that's kinda generic. I haven't found similar documentation in other
ACPI-disable flags to copy from (E.g. do_not_add_smb_acpi).

What I am asking about is whether we need a flag to disable
this as part of the stable interface.
I personally think not. But if you think otherwise, can you provide
guidance of where you suggest to add this flag?
As the only place I see fit is adding a new QEMU_OPTION_no_weat.

reply via email to

[Prev in Thread] Current Thread [Next in Thread]