From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: From: tstark@linux.microsoft.com Subject: [PATCH v2 1/1] virtio-pmem: Support describing pmem as shared memory region Date: Mon, 26 Jul 2021 21:29:07 -0700 Message-Id: <20210727042907.24315-2-tstark@linux.microsoft.com> In-Reply-To: <20210727042907.24315-1-tstark@linux.microsoft.com> References: <20210727042907.24315-1-tstark@linux.microsoft.com> MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="US-ASCII" To: virtio-comment@lists.oasis-open.org Cc: grahamwo@microsoft.com, benhill@microsoft.com, mst@redhat.com, pankaj.gupta@ionos.com, tstark@microsoft.com, david@redhat.com, cohuck@redhat.com List-ID: From: Taylor Stark Update the virtio-pmem RFC spec to add support for describing the pmem regi= on as a shared memory window. This is required to support virtio-pmem in Hyper= -V, since Hyper-V only allows PCI devices to operate on memory ranges defined v= ia BARs. When using the virtio PCI transport, shared memory regions are descri= bed via PCI BARs. Signed-off-by: Taylor Stark --- conformance.tex | 1 + virtio-pmem.tex | 30 +++++++++++++++++++++++++----- 2 files changed, 26 insertions(+), 5 deletions(-) diff --git a/conformance.tex b/conformance.tex index 33fdac9..8e414eb 100644 --- a/conformance.tex +++ b/conformance.tex @@ -201,6 +201,7 @@ \section{Conformance Targets}\label{sec:Conformance / C= onformance Targets} =20 \begin{itemize} \item \ref{devicenormative:Device Types / PMEM Device / Device Initializat= ion} +\item \ref{drivernormative:Device Types / PMEM Driver / Driver Initializat= ion} \item \ref{drivernormative:Device Types / PMEM Driver / Driver Initializat= ion / Direct access} \item \ref{drivernormative:Device Types / PMEM Driver / Driver Initializat= ion / Virtio flush} \item \ref{drivernormative:Device Types / PMEM Driver / Driver Operation /= Virtqueue command} diff --git a/virtio-pmem.tex b/virtio-pmem.tex index 04e07bb..9f25664 100644 --- a/virtio-pmem.tex +++ b/virtio-pmem.tex @@ -19,7 +19,10 @@ \subsection{Virtqueues}\label{sec:Device Types / PMEM De= vice / Virtqueues} =20 \subsection{Feature bits}\label{sec:Device Types / PMEM Device / Feature b= its} =20 -There are currently no feature bits defined for this device. +\begin{description} +\item[VIRTIO_PMEM_F_SHMEM_REGION (0)] The guest physical address range wil= l be +indicated as a shared memory region. +\end{description} =20 \subsection{Device configuration layout}\label{sec:Device Types / PMEM Dev= ice / Device configuration layout} =20 @@ -38,11 +41,11 @@ \subsection{Device configuration layout}\label{sec:Devi= ce Types / PMEM Device / \subsection{Device Initialization}\label{sec:Device Types / PMEM Device / = Device Initialization} =20 Device hotplugs physical memory to guest address space. Persistent memory = device -is emulated with file backed memory at host side. - +is emulated with file backed memory at host side. The device indicates the= guest +physical address to the driver in one of two ways: \begin{enumerate} -\item Guest vpmem start is read from \field{start}. -\item Guest vpmem end is read from \field{size}. +\item As a guest absolute address. +\item As a shared memory region. \end{enumerate} =20 \devicenormative{\subsubsection}{Device Initialization}{Device Types / PME= M Device / Device Initialization} @@ -50,12 +53,29 @@ \subsection{Device Initialization}\label{sec:Device Typ= es / PMEM Device / Device File backed memory SHOULD be memory mapped to guest address space with SHA= RED memory mapping. =20 +If VIRTIO_PMEM_F_SHMEM_REGION has been negotiated, the device MUST indicat= e the +guest physical address as a shared memory region. The device MUST use shar= ed +memory region ID 0. The device SHOULD set \field{start} and \field{size} t= o zero. + +If VIRTIO_PMEM_F_SHMEM_REGION has not been negotiated, the device MUST ind= icate +the guest physical address as a guest absolute address. The device MUST se= t +\field{start} to the absolute address and \field{size} to the size of the +address range, in bytes. + \subsection{Driver Initialization}\label{sec:Device Types / PMEM Driver / = Driver Initialization} =20 Driver hotplugs the physical memory and registers associated region with the pmem API. Also, configures a flush callback function with the corresponding region. =20 +\drivernormative{\subsubsection}{Driver Initialization}{Device Types / PME= M Driver / Driver Initialization} + +The driver SHOULD query the physical address ranges where the pmem was map= ped. +If VIRTIO_PMEM_F_SHMEM_REGION has been negotiated, the driver SHOULD query +shared memory ID 0 for the physical address ranges and MUST NOT use \field= {start} +or \field{size}. Else, the driver SHOULD read the physical address ranges = from +\field{start} and \field{size}. + \drivernormative{\subsubsection}{Driver Initialization: Filesystem direct = access}{Device Types / PMEM Driver / Driver Initialization / Direct access} =20 Driver SHOULD enable filesystem direct access operations for --=20 2.32.0