[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [RFC PATCH v2 0/4] vfio/pci: Atomic Ops completer support
From: |
Philippe Mathieu-Daudé |
Subject: |
Re: [RFC PATCH v2 0/4] vfio/pci: Atomic Ops completer support |
Date: |
Fri, 2 Jun 2023 16:02:20 +0200 |
User-agent: |
Mozilla/5.0 (Macintosh; Intel Mac OS X 10.15; rv:102.0) Gecko/20100101 Thunderbird/102.11.0 |
Hi Markus, Marc-André,
On 1/6/23 00:24, Alex Williamson wrote:
Policy decisions like that are generally left to management tools, so
there would always be a means to enable or disable the feature. In
fact, that's specifically why I test that the Atomic Op completer bits
are unset in the root port before changing them so that this automatic
enablement could live alongside a command line option to statically
enable some bits.
That does however remind me that it is often good with these sorts of
"clever" automatic features to have an opt-out, so I'll likely add an
x-no-rp-atomics device option in the next version to provide that.
Still thinking about this series I remembered since commit a3c45b3e62
("qapi: New special feature flag "unstable"") the "x-" prefix is
obsolete (superseded) in QAPI generated code.
I wonder about device properties:
$ git grep -F '"x-' hw | wc -l
130
$ git grep -F '"x-' hw/vfio/pci.c
hw/vfio/pci.c:2987: error_setg(errp, "x-balloon-allowed only
potentially compatible "
hw/vfio/pci.c:3335:
DEFINE_PROP_ON_OFF_AUTO("x-pre-copy-dirty-page-tracking", VFIOPCIDevice,
hw/vfio/pci.c:3342: DEFINE_PROP_UINT32("x-intx-mmap-timeout-ms",
VFIOPCIDevice,
hw/vfio/pci.c:3344: DEFINE_PROP_BIT("x-vga", VFIOPCIDevice, features,
hw/vfio/pci.c:3346: DEFINE_PROP_BIT("x-req", VFIOPCIDevice, features,
hw/vfio/pci.c:3348: DEFINE_PROP_BIT("x-igd-opregion", VFIOPCIDevice,
features,
hw/vfio/pci.c:3350: DEFINE_PROP_BOOL("x-enable-migration", VFIOPCIDevice,
hw/vfio/pci.c:3352: DEFINE_PROP_BOOL("x-no-mmap", VFIOPCIDevice,
vbasedev.no_mmap, false),
hw/vfio/pci.c:3353: DEFINE_PROP_BOOL("x-balloon-allowed", VFIOPCIDevice,
hw/vfio/pci.c:3355: DEFINE_PROP_BOOL("x-no-kvm-intx", VFIOPCIDevice,
no_kvm_intx, false),
hw/vfio/pci.c:3356: DEFINE_PROP_BOOL("x-no-kvm-msi", VFIOPCIDevice,
no_kvm_msi, false),
hw/vfio/pci.c:3357: DEFINE_PROP_BOOL("x-no-kvm-msix", VFIOPCIDevice,
no_kvm_msix, false),
hw/vfio/pci.c:3358: DEFINE_PROP_BOOL("x-no-geforce-quirks",
VFIOPCIDevice,
hw/vfio/pci.c:3360: DEFINE_PROP_BOOL("x-no-kvm-ioeventfd",
VFIOPCIDevice, no_kvm_ioeventfd,
hw/vfio/pci.c:3362: DEFINE_PROP_BOOL("x-no-vfio-ioeventfd",
VFIOPCIDevice, no_vfio_ioeventfd,
hw/vfio/pci.c:3364: DEFINE_PROP_UINT32("x-pci-vendor-id",
VFIOPCIDevice, vendor_id, PCI_ANY_ID),
hw/vfio/pci.c:3365: DEFINE_PROP_UINT32("x-pci-device-id",
VFIOPCIDevice, device_id, PCI_ANY_ID),
hw/vfio/pci.c:3366: DEFINE_PROP_UINT32("x-pci-sub-vendor-id",
VFIOPCIDevice,
hw/vfio/pci.c:3368: DEFINE_PROP_UINT32("x-pci-sub-device-id",
VFIOPCIDevice,
hw/vfio/pci.c:3370: DEFINE_PROP_UINT32("x-igd-gms", VFIOPCIDevice,
igd_gms, 0),
hw/vfio/pci.c:3371:
DEFINE_PROP_UNSIGNED_NODEFAULT("x-nv-gpudirect-clique", VFIOPCIDevice,
hw/vfio/pci.c:3374: DEFINE_PROP_OFF_AUTO_PCIBAR("x-msix-relocation",
VFIOPCIDevice, msix_relo,
Is there a plan to use something similar for QOM properties?
Is it OK to keep using the "x-" prefix there?
Thanks,
Phil.