diff options
| author | Viresh Kumar <viresh.kumar@linaro.org> | 2022-01-06 13:00:43 +0530 |
|---|---|---|
| committer | Viresh Kumar <viresh.kumar@linaro.org> | 2022-01-06 13:00:43 +0530 |
| commit | 47c1c3285dbe93513a8fdece18d9cb634ce60423 (patch) | |
| tree | d2870c5953933b98b39a2855d12f30d8cb5067e2 | |
| parent | 051fc0d5e4bafd578f5adf88e2db284e78b6b870 (diff) | |
updates
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
| -rw-r--r-- | rust/.gpio.txt.swp | bin | 36864 -> 0 bytes | |||
| -rw-r--r-- | rust/gpio.html | 229 | ||||
| -rw-r--r-- | rust/gpio.txt | 231 | ||||
| -rw-r--r-- | rust/i2c.html | 993 | ||||
| -rw-r--r-- | rust/i2c.txt | 255 |
5 files changed, 1248 insertions, 460 deletions
diff --git a/rust/.gpio.txt.swp b/rust/.gpio.txt.swp Binary files differdeleted file mode 100644 index ec6da3d..0000000 --- a/rust/.gpio.txt.swp +++ /dev/null diff --git a/rust/gpio.html b/rust/gpio.html deleted file mode 100644 index 9012aa4..0000000 --- a/rust/gpio.html +++ /dev/null @@ -1,229 +0,0 @@ -<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
-<html>
-<head>
-<meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
-<meta name="generator" content="AsciiDoc 9.0.0rc1">
-<title>Rust based vhost-user I2C backend</title>
-</head>
-<body>
-<h1>Rust based vhost-user I2C backend</h1>
-<p>
-</p>
-<a name="preamble"></a>
-<p>There is a growing trend towards virtualization in areas other than the
-traditional server environment. The server environment is uniform in nature, but
-as we move towards a richer ecosystem in automotive, medical, general mobile and
-the IoT spaces, more device abstractions, and way richer organizations are
-needed. <a href="https://www.linaro.org/projects/#automotive_STR">Linaro’s Project
-Stratos</a> is working towards developing hypervisor agnostic abstract devices
-leveraging virtio and extending hypervisor interfaces and standards to allow all
-architectures.</p>
-<p>The Virtual Input/Output device (Virtio) standard provides an open interface for
-guest virtual machines (VMs) to access simplified "virtual" devices, such as
-network adapters and block devices, in a paravirtualized environment. Virtio
-provides a straightforward, efficient, standard and extensible mechanism for
-virtual devices, rather than a per-environment or per-OS mechanism.</p>
-<p>The backend (BE) virtio driver, implemented in the hypervisor running on the host,
-exposes the virtio device to the guest OS through a transport method, like PCI
-or MMIO. The virtio device, by design, looks like a physical device to the guest
-OS, which implements a frontend (FE) virtio driver compatible with the virtio
-device exposed by the hypervisor. The virtio device and driver communicate based
-on a set of predefined protocols as defined by the
-<a href="https://github.com/oasis-tcs/virtio-spec">virtio specification</a>, which is
-maintained by <a href="https://www.oasis-open.org/org/">OASIS</a>. The FE driver can
-implement zero or more Virtual queues (virtqueues), as defined by the virtio
-specification. The virtqueues are the mechanism of bulk data transport between
-FE (guest) and BE (host) drivers. These are normally implemented as standard
-ring buffers in the guest physical memory by the FE drivers. The BE drivers
-parse the virtqueues to obtain the request descriptors, process them and queue
-the response descriptors back to the virtqueue.</p>
-<p>The FE virtio driver at the guest and the virtio specification are normally
-independent of where the virtqueue processing happens at the host, in-kernel or
-userspace. The vhost protocol allows the virtio virtqueue processing at the
-host to be offloaded to another element, a user process or a kernel module. The
-vhost protocol when implemented in userspace is called as "vhost-user". Since
-Linaro’s Project Stratos is targeting hypervisor agnostic BE drivers, engineers
-at Linaro decided to work over the existing vhost-user protocol. This article
-focuses on the Rust based vhost-user implementation for I2C devices.</p>
-<hr>
-<h2><a name="_virtio_i2c_specification"></a>Virtio I2C Specification</h2>
-<p>The Virtio
-<a href="https://github.com/oasis-tcs/virtio-spec/blob/master/virtio-i2c.tex">specification</a>
-for I2C and the Linux
-<a href="https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/i2c/busses/i2c-virtio.c">i2c-virtio</a>
-driver are upstreamed by Jie Deng (Intel), who tested his work with the
-<a href="https://projectacrn.org">ACRN</a> hypervisor for IoT development. Both
-specification and driver received updates later on by Viresh Kumar (Linaro), to
-improve buffer management and allow zero-length transactions. Lets go through
-the I2C virtio specification briefly.</p>
-<p>virtio-i2c is a virtual I2C adapter device, which provides a way to flexibly
-organize and use the host I2C controlled devices from the guest. All
-communication between the FE and BE drivers happens over the "requestq"
-virtqueue. It is also mandatory for both the sides to implement the
-<code>VIRTIO_I2C_F_ZERO_LENGTH_REQUEST</code> feature, which allows zero-length transfers
-(like SMBus Quick) to take place. The I2C requests always originate at the guest
-FE driver, where the FE driver puts one or more I2C requests, represented by the
-<code>struct virtio_i2c_req</code>, on the requestq virtqueue. The I2C requests may or may
-not be be interdependent. If multiple requests are received together, then the
-host BE driver must process the requests in the order they are received on the
-virtqueue.</p>
-<table border="0" bgcolor="#e8e8e8" width="100%" cellpadding="4"><tr><td>
-<pre><code>struct virtio_i2c_req {
- struct virtio_i2c_out_hdr out_hdr;
- u8 buf[];
- struct virtio_i2c_in_hdr in_hdr;
-};</code></pre>
-</td></tr></table>
-<p>Each I2C virtio request consists of an <code>out_hdr</code> (set by the FE driver), followed by
-an optional buffer of some length (set by FE or BE driver based on if the
-transaction is write or read), followed by an <code>in_hdr</code> (set by the BE driver). The
-buffer is not sent for zero-length requests, like for the SMBus Quick command
-where no data is required to be sent or received.</p>
-<table border="0" bgcolor="#e8e8e8" width="100%" cellpadding="4"><tr><td>
-<pre><code>struct virtio_i2c_out_hdr {
- le16 addr;
- le16 padding;
- le32 flags;
-};</code></pre>
-</td></tr></table>
-<p>The <code>out_hdr</code> is represented by the <code>struct virtio_i2c_out_hdr</code>. The <code>addr</code>
-field of the header is the address of the I2C controlled device. Both 7-bit and
-10-bit address modes are supported by the specification (though only 7-bit mode
-is supported by the current implementation of the Linux FE driver). The <code>flags</code>
-field is used to mark a request "Read or write" (<code>VIRTIO_I2C_FLAGS_M_RD</code> (bit
-1)) or to show dependency between multiple requests
-(<code>VIRTIO_I2C_FLAGS_FAIL_NEXT</code> (bit 0)).</p>
-<p>As described earlier, the <code>buf</code> is optional. For "write" transactions, it is
-pre-filled by the FE driver and read by the BE driver. For "read" transactions,
-it is filled by the BE driver and read by the FE driver after the response is
-received.</p>
-<table border="0" bgcolor="#e8e8e8" width="100%" cellpadding="4"><tr><td>
-<pre><code>struct virtio_i2c_in_hdr {
- u8 status;
-};</code></pre>
-</td></tr></table>
-<p>The <code>in_hdr</code> is represented by the <code>struct virtio_i2c_in_hdr</code> and is used by the
-host BE driver to notify the guest with the status of the transfer with
-<code>VIRTIO_I2C_MSG_OK</code> or <code>VIRTIO_I2C_MSG_ERR</code>.</p>
-<p>Please refer the Virtio I2C
-<a href="https://github.com/oasis-tcs/virtio-spec/blob/master/virtio-i2c.tex">specification</a>
-of more details.</p>
-<hr>
-<h2><a name="_rust_based_i2c_backend"></a>Rust based I2C backend</h2>
-<p>Rust is the next big thing disrupting the Linux world and most of us are already
-aware of the <a href="https://github.com/Rust-for-Linux">Rust for Linux</a> project
-slowly making its way into the Linux kernel. Rust is a multi-paradigm,
-general-purpose programming language designed for performance and safety. It
-brings a lot of benefits to the table, especially
-<a href="https://en.wikipedia.org/wiki/Memory_safety">memory-safety</a> and safe
-<a href="https://en.wikipedia.org/wiki/Concurrency_(computer_science)">concurrency</a>.
-It was an easy choice to pick for developing hypervisor agnostic I2C BE driver.</p>
-<p>The <a href="https://github.com/rust-vmm">rust-vmm</a> project, an open-source
-initiative, was started back in late 2018, with the aim to share virtualization
-packages. The rust-vmm project lets one build custom
-<a href="https://en.wikipedia.org/wiki/Hypervisor">Virtual Machine Monitors (VMMs)
-and hypervisors</a>. This empowers other projects to quickly develop virtualization
-solutions, by reusing the components provided by rust-vmm, and better focus on
-key differentiators of their products. The rust-vmm project is organized as a
-shared ownership project that so far includes contributions from Alibaba, AWS,
-Cloud Base, Google, Intel, Linaro, Red Hat and other individual contributors.
-The components provided by rust-vmm are already used by several projects, like
-Amazon’s <a href="https://github.com/firecracker-microvm/firecracker">Firecracker</a>
-and Intel’s <a href="https://github.com/cloud-hypervisor/cloud-hypervisor">Cloud
-Hypervisor</a>. The rust-vmm project currently roughly 30 repositories (or Rust
-crates, equivalent of a C library), where each crate plays a special role in the
-development of a fully functioning VMM.</p>
-<p>One such component provided by the rust-vmm project is the
-<a href="https://crates.io/crates/vhost-user-backend">vhost-user-backend</a> crate,
-which has recently made its way to <a href="https://crates.io/">crates.io</a>, the Rust
-community’s crate registry. The vhost-user-backend crate provides a framework to
-implement the vhost-user backend services. It provides necessary public APIs to
-support vhost-user backends, like a daemon control object (<code>VhostUserDaemon</code>) to
-start and stop the service daemon, a vhost-user backend trait
-(<code>VhostUserBackendMut</code>) to handle vhost-user control messages and virtio
-messages, and a vring access trait (<code>VringT</code>) to access virtio queues.</p>
-<p>A separate Rust workspace,
-<a href="https://github.com/rust-vmm/vhost-device">vhost-device</a>, is recently created
-in the rust-vmm project, to host per-device vhost-user backend crates. The only
-crate merged there as for now is for the I2C device, while there are others
-getting developed and reviewed as we speak, like GPIO, RNG, VSOCK, SCSI, and
-<a href="https://en.wikipedia.org/wiki/Replay_Protected_Memory_Block">RPMB</a>.</p>
-<p>The I2C vhost-device binary-crate (generates an executable upon build),
-developed by Viresh Kumar (Linaro), supports sharing host I2C busses (Adaptors)
-and client devices with multiple guest VMs at the same time with a single
-instance of an always running backend daemon. Once the vhost-device crate is
-compiled with <code>cargo build --release</code>, it generates the
-<code>target/release/vhost-device-i2c</code> executable. The <code>vhost-device-i2c</code> daemon
-communicates with guest VMs over Unix domain sockets, a unique socket for each
-VM.</p>
-<p>The daemon accepts three arguments:</p>
-<ul>
-<li>
-<p>
--s, --socket-path: Path of the vhost-user Unix domain sockets. This is
- suffixed with 0,1,2..socket_count-1 by the daemon to obtain actual socket
- paths.
-</p>
-</li>
-<li>
-<p>
--c, --socket-count: Number of sockets (guests) to connect to. This parameter
- is optional and defaults to 1.
-</p>
-</li>
-<li>
-<p>
--l, --device-list: List of I2C bus and clients in the format
- <bus>:<client_addr>[:<client_addr>][,<bus>:<client_addr>[:<client_addr>]]
-</p>
-</li>
-</ul>
-<p>As an example, consider the following command:</p>
-<table border="0" bgcolor="#e8e8e8" width="100%" cellpadding="4"><tr><td>
-<pre><code>./vhost-device-i2c -s ~/i2c.sock -c 6 -l 6:32:41,9:37:6</code></pre>
-</td></tr></table>
-<p>This will start the I2C backend daemon, which will create 6 Unix domain sockets
-(<sub>/i2c.sock0, .., </sub>/i2c.sock5), in order to communicate with 6 guest VMs, where
-communication with each VM happens in parallel with the help of a separate
-native OS thread. Once the threads are created by the daemon, the threads wait
-for a VM to start communicating on the thread’s designated socket. Later, when a
-VM shuts down, the respective thread starts waiting for a new VM to communicate
-on the same socket path. The daemon is also passed a list of host I2C busses and
-client devices, which are shared by all the VMs. The daemon can be modified
-later on, if required, to allow specific devices to be accessed only by a
-particular VM, this feature isn’t added in the current version of the daemon. In
-the above example, the devices shared by the host with the daemon are: devices
-with address 32 and 41 attached to I2C bus 6, 37 and 6 attached to I2C bus 9.
-The daemon extensively validates the device-list at initialization to avoid any
-failures later.</p>
-<p>The <code>vhost-user-i2c</code> daemon supports both I2C and SMBus protocols, only basic
-SMBus commands up to word-transfer. The backend provides the <code>pub trait
-I2cDevice</code>, a public Rust trait, which can be implemented for different host
-environments to provide access to the underlying I2C busses and devices. This is
-currently implemented only for the Linux userspace, where the I2C busses and
-devices are accessed via the <code>/dev/i2c-X</code> I2C device files. For the above
-example, the backend daemon will look for <code>/dev/i2c-6</code> and <code>/dev/i2c-9</code> device
-files. The users may need to load the <code>i2c-dev</code> kernel module, if not loaded
-already, for these device files to be available under <code>/dev/</code>. For a different
-host environment, like a bare-metal type 1 hypervisor, we need to add another
-implementation of the trait depending on how the I2C busses and devices are
-accessed.</p>
-<p>The <code>vhost-user-i2c</code> backend is truly a hypervisor agnostic solution that works
-with any hypervisor which understands the vhost-user protocol. It has been
-extensively tested with QEMU for example, with Linux userspace environment. Work
-is in progress to make Xen hypervisor vhost-user protocol compatible. Once that
-is done, we will be able to use the same <code>vhost-user-i2c</code> executable with both
-QEMU and Xen, for example, under the same host environments.</p>
-<p>Support for virtio-i2c is already merged in QEMU source, boilerplate stuff to
-create the virtio-i2c device in the guest kernel, and the virtio-i2c device can
-be created in the guest kernel by adding following command line arguments to
-your QEMU command:</p>
-<p><code>-chardev socket,path=~/i2c.sock0,id=vi2c -device vhost-user-i2c-device,chardev=vi2c,id=i2c</code></p>
-<p></p>
-<p></p>
-<hr><p><small>
-Last updated
- 2022-01-05 12:58:45 IST
-</small></p>
-</body>
-</html>
diff --git a/rust/gpio.txt b/rust/gpio.txt deleted file mode 100644 index 6148e6b..0000000 --- a/rust/gpio.txt +++ /dev/null @@ -1,231 +0,0 @@ -Rust based vhost-user I2C backend -================================= - -There is a growing trend towards virtualization in areas other than the -traditional server environment. The server environment is uniform in nature, but -as we move towards a richer ecosystem in automotive, medical, general mobile and -the IoT spaces, more device abstractions, and way richer organizations are -needed. link:https://www.linaro.org/projects/#automotive_STR[Linaro's Project -Stratos] is working towards developing hypervisor agnostic abstract devices -leveraging virtio and extending hypervisor interfaces and standards to allow all -architectures. - -The Virtual Input/Output device (Virtio) standard provides an open interface for -guest virtual machines (VMs) to access simplified "virtual" devices, such as -network adapters and block devices, in a paravirtualized environment. Virtio -provides a straightforward, efficient, standard and extensible mechanism for -virtual devices, rather than a per-environment or per-OS mechanism. - -The backend (BE) virtio driver, implemented in the hypervisor running on the host, -exposes the virtio device to the guest OS through a transport method, like PCI -or MMIO. The virtio device, by design, looks like a physical device to the guest -OS, which implements a frontend (FE) virtio driver compatible with the virtio -device exposed by the hypervisor. The virtio device and driver communicate based -on a set of predefined protocols as defined by the -link:https://github.com/oasis-tcs/virtio-spec[virtio specification], which is -maintained by link:https://www.oasis-open.org/org/[OASIS]. The FE driver can -implement zero or more Virtual queues (virtqueues), as defined by the virtio -specification. The virtqueues are the mechanism of bulk data transport between -FE (guest) and BE (host) drivers. These are normally implemented as standard -ring buffers in the guest physical memory by the FE drivers. The BE drivers -parse the virtqueues to obtain the request descriptors, process them and queue -the response descriptors back to the virtqueue. - -The FE virtio driver at the guest and the virtio specification are normally -independent of where the virtqueue processing happens at the host, in-kernel or -userspace. The vhost protocol allows the virtio virtqueue processing at the -host to be offloaded to another element, a user process or a kernel module. The -vhost protocol when implemented in userspace is called as "vhost-user". Since -Linaro's Project Stratos is targeting hypervisor agnostic BE drivers, engineers -at Linaro decided to work over the existing vhost-user protocol. This article -focuses on the Rust based vhost-user implementation for I2C devices. - -Virtio I2C Specification ------------------------- - -The Virtio -link:https://github.com/oasis-tcs/virtio-spec/blob/master/virtio-i2c.tex[specification] -for I2C and the Linux -link:https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/i2c/busses/i2c-virtio.c[i2c-virtio] -driver are upstreamed by Jie Deng (Intel), who tested his work with the -link:https://projectacrn.org[ACRN] hypervisor for IoT development. Both -specification and driver received updates later on by Viresh Kumar (Linaro), to -improve buffer management and allow zero-length transactions. Lets go through -the I2C virtio specification briefly. - -virtio-i2c is a virtual I2C adapter device, which provides a way to flexibly -organize and use the host I2C controlled devices from the guest. All -communication between the FE and BE drivers happens over the "requestq" -virtqueue. It is also mandatory for both the sides to implement the -`VIRTIO_I2C_F_ZERO_LENGTH_REQUEST` feature, which allows zero-length transfers -(like SMBus Quick) to take place. The I2C requests always originate at the guest -FE driver, where the FE driver puts one or more I2C requests, represented by the -`struct virtio_i2c_req`, on the requestq virtqueue. The I2C requests may or may -not be be interdependent. If multiple requests are received together, then the -host BE driver must process the requests in the order they are received on the -virtqueue. - ----- -struct virtio_i2c_req { - struct virtio_i2c_out_hdr out_hdr; - u8 buf[]; - struct virtio_i2c_in_hdr in_hdr; -}; ----- - -Each I2C virtio request consists of an `out_hdr` (set by the FE driver), followed by -an optional buffer of some length (set by FE or BE driver based on if the -transaction is write or read), followed by an `in_hdr` (set by the BE driver). The -buffer is not sent for zero-length requests, like for the SMBus Quick command -where no data is required to be sent or received. - ----- -struct virtio_i2c_out_hdr { - le16 addr; - le16 padding; - le32 flags; -}; ----- - -The `out_hdr` is represented by the `struct virtio_i2c_out_hdr`. The `addr` -field of the header is the address of the I2C controlled device. Both 7-bit and -10-bit address modes are supported by the specification (though only 7-bit mode -is supported by the current implementation of the Linux FE driver). The `flags` -field is used to mark a request "Read or write" (`VIRTIO_I2C_FLAGS_M_RD` (bit -1)) or to show dependency between multiple requests -(`VIRTIO_I2C_FLAGS_FAIL_NEXT` (bit 0)). - -As described earlier, the `buf` is optional. For "write" transactions, it is -pre-filled by the FE driver and read by the BE driver. For "read" transactions, -it is filled by the BE driver and read by the FE driver after the response is -received. - ----- -struct virtio_i2c_in_hdr { - u8 status; -}; ----- - -The `in_hdr` is represented by the `struct virtio_i2c_in_hdr` and is used by the -host BE driver to notify the guest with the status of the transfer with -`VIRTIO_I2C_MSG_OK` or `VIRTIO_I2C_MSG_ERR`. - -Please refer the Virtio I2C -link:https://github.com/oasis-tcs/virtio-spec/blob/master/virtio-i2c.tex[specification] -of more details. - -Rust based I2C backend ----------------------- - -Rust is the next big thing disrupting the Linux world and most of us are already -aware of the link:https://github.com/Rust-for-Linux[Rust for Linux] project -slowly making its way into the Linux kernel. Rust is a multi-paradigm, -general-purpose programming language designed for performance and safety. It -brings a lot of benefits to the table, especially -link:https://en.wikipedia.org/wiki/Memory_safety[memory-safety] and safe -link:https://en.wikipedia.org/wiki/Concurrency_(computer_science)[concurrency]. -It was an easy choice to pick for developing hypervisor agnostic I2C BE driver. - -The link:https://github.com/rust-vmm[rust-vmm] project, an open-source -initiative, was started back in late 2018, with the aim to share virtualization -packages. The rust-vmm project lets one build custom -link:https://en.wikipedia.org/wiki/Hypervisor[Virtual Machine Monitors (VMMs) -and hypervisors]. This empowers other projects to quickly develop virtualization -solutions, by reusing the components provided by rust-vmm, and better focus on -key differentiators of their products. The rust-vmm project is organized as a -shared ownership project that so far includes contributions from Alibaba, AWS, -Cloud Base, Google, Intel, Linaro, Red Hat and other individual contributors. -The components provided by rust-vmm are already used by several projects, like -Amazon's link:https://github.com/firecracker-microvm/firecracker[Firecracker] -and Intel's link:https://github.com/cloud-hypervisor/cloud-hypervisor[Cloud -Hypervisor]. The rust-vmm project currently roughly 30 repositories (or Rust -crates, equivalent of a C library), where each crate plays a special role in the -development of a fully functioning VMM. - -One such component provided by the rust-vmm project is the -link:https://crates.io/crates/vhost-user-backend[vhost-user-backend] crate, -which has recently made its way to link:https://crates.io/[crates.io], the Rust -community’s crate registry. The vhost-user-backend crate provides a framework to -implement the vhost-user backend services. It provides necessary public APIs to -support vhost-user backends, like a daemon control object (`VhostUserDaemon`) to -start and stop the service daemon, a vhost-user backend trait -(`VhostUserBackendMut`) to handle vhost-user control messages and virtio -messages, and a vring access trait (`VringT`) to access virtio queues. - -A separate Rust workspace, -link:https://github.com/rust-vmm/vhost-device[vhost-device], is recently created -in the rust-vmm project, to host per-device vhost-user backend crates. The only -crate merged there as for now is for the I2C device, while there are others -getting developed and reviewed as we speak, like GPIO, RNG, VSOCK, SCSI, and -link:https://en.wikipedia.org/wiki/Replay_Protected_Memory_Block[RPMB]. - -The I2C vhost-device binary-crate (generates an executable upon build), -developed by Viresh Kumar (Linaro), supports sharing host I2C busses (Adaptors) -and client devices with multiple guest VMs at the same time with a single -instance of an always running backend daemon. Once the vhost-device crate is -compiled with `cargo build --release`, it generates the -`target/release/vhost-device-i2c` executable. The `vhost-device-i2c` daemon -communicates with guest VMs over Unix domain sockets, a unique socket for each -VM. - -The daemon accepts three arguments: - -* -s, --socket-path: Path of the vhost-user Unix domain sockets. This is - suffixed with 0,1,2..socket_count-1 by the daemon to obtain actual socket - paths. - -* -c, --socket-count: Number of sockets (guests) to connect to. This parameter - is optional and defaults to 1. - -* -l, --device-list: List of I2C bus and clients in the format - <bus>:<client_addr>[:<client_addr>][,<bus>:<client_addr>[:<client_addr>]] - -As an example, consider the following command: - ----- -./vhost-device-i2c -s ~/i2c.sock -c 6 -l 6:32:41,9:37:6 ----- - -This will start the I2C backend daemon, which will create 6 Unix domain sockets -(~/i2c.sock0, .., ~/i2c.sock5), in order to communicate with 6 guest VMs, where -communication with each VM happens in parallel with the help of a separate -native OS thread. Once the threads are created by the daemon, the threads wait -for a VM to start communicating on the thread's designated socket. Later, when a -VM shuts down, the respective thread starts waiting for a new VM to communicate -on the same socket path. The daemon is also passed a list of host I2C busses and -client devices, which are shared by all the VMs. The daemon can be modified -later on, if required, to allow specific devices to be accessed only by a -particular VM, this feature isn't added in the current version of the daemon. In -the above example, the devices shared by the host with the daemon are: devices -with address 32 and 41 attached to I2C bus 6, 37 and 6 attached to I2C bus 9. -The daemon extensively validates the device-list at initialization to avoid any -failures later. - -The `vhost-user-i2c` daemon supports both I2C and SMBus protocols, only basic -SMBus commands up to word-transfer. The backend provides the `pub trait -I2cDevice`, a public Rust trait, which can be implemented for different host -environments to provide access to the underlying I2C busses and devices. This is -currently implemented only for the Linux userspace, where the I2C busses and -devices are accessed via the `/dev/i2c-X` I2C device files. For the above -example, the backend daemon will look for `/dev/i2c-6` and `/dev/i2c-9` device -files. The users may need to load the `i2c-dev` kernel module, if not loaded -already, for these device files to be available under `/dev/`. For a different -host environment, like a bare-metal type 1 hypervisor, we need to add another -implementation of the trait depending on how the I2C busses and devices are -accessed. - -The `vhost-user-i2c` backend is truly a hypervisor agnostic solution that works -with any hypervisor which understands the vhost-user protocol. It has been -extensively tested with QEMU for example, with Linux userspace environment. Work -is in progress to make Xen hypervisor vhost-user protocol compatible. Once that -is done, we will be able to use the same `vhost-user-i2c` executable with both -QEMU and Xen, for example, under the same host environments. - -Support for virtio-i2c is already merged in QEMU source, boilerplate stuff to -create the virtio-i2c device in the guest kernel, and the virtio-i2c device can -be created in the guest kernel by adding following command line arguments to -your QEMU command: - ----- -`-chardev socket,path=~/i2c.sock0,id=vi2c -device vhost-user-i2c-device,chardev=vi2c,id=i2c` ----- diff --git a/rust/i2c.html b/rust/i2c.html new file mode 100644 index 0000000..4a7a254 --- /dev/null +++ b/rust/i2c.html @@ -0,0 +1,993 @@ +<!DOCTYPE html>
+<html lang="en">
+<head>
+<meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
+<meta name="generator" content="AsciiDoc 9.0.0rc1">
+<title>Rust based vhost-user I2C backend</title>
+<style type="text/css">
+/* Shared CSS for AsciiDoc xhtml11 and html5 backends */
+
+/* Default font. */
+body {
+ font-family: Georgia,serif;
+}
+
+/* Title font. */
+h1, h2, h3, h4, h5, h6,
+div.title, caption.title,
+thead, p.table.header,
+#toctitle,
+#author, #revnumber, #revdate, #revremark,
+#footer {
+ font-family: Arial,Helvetica,sans-serif;
+}
+
+body {
+ margin: 1em 5% 1em 5%;
+}
+
+a {
+ color: blue;
+ text-decoration: underline;
+}
+a:visited {
+ color: fuchsia;
+}
+
+em {
+ font-style: italic;
+ color: navy;
+}
+
+strong {
+ font-weight: bold;
+ color: #083194;
+}
+
+h1, h2, h3, h4, h5, h6 {
+ color: #527bbd;
+ margin-top: 1.2em;
+ margin-bottom: 0.5em;
+ line-height: 1.3;
+}
+
+h1, h2, h3 {
+ border-bottom: 2px solid silver;
+}
+h2 {
+ padding-top: 0.5em;
+}
+h3 {
+ float: left;
+}
+h3 + * {
+ clear: left;
+}
+h5 {
+ font-size: 1.0em;
+}
+
+div.sectionbody {
+ margin-left: 0;
+}
+
+hr {
+ border: 1px solid silver;
+}
+
+p {
+ margin-top: 0.5em;
+ margin-bottom: 0.5em;
+}
+
+ul, ol, li > p {
+ margin-top: 0;
+}
+ul > li { color: #aaa; }
+ul > li > * { color: black; }
+
+.monospaced, code, pre {
+ font-family: "Courier New", Courier, monospace;
+ font-size: inherit;
+ color: navy;
+ padding: 0;
+ margin: 0;
+}
+pre {
+ white-space: pre-wrap;
+}
+
+#author {
+ color: #527bbd;
+ font-weight: bold;
+ font-size: 1.1em;
+}
+#email {
+}
+#revnumber, #revdate, #revremark {
+}
+
+#footer {
+ font-size: small;
+ border-top: 2px solid silver;
+ padding-top: 0.5em;
+ margin-top: 4.0em;
+}
+#footer-text {
+ float: left;
+ padding-bottom: 0.5em;
+}
+#footer-badges {
+ float: right;
+ padding-bottom: 0.5em;
+}
+
+#preamble {
+ margin-top: 1.5em;
+ margin-bottom: 1.5em;
+}
+div.imageblock, div.exampleblock, div.verseblock,
+div.quoteblock, div.literalblock, div.listingblock, div.sidebarblock,
+div.admonitionblock {
+ margin-top: 1.0em;
+ margin-bottom: 1.5em;
+}
+div.admonitionblock {
+ margin-top: 2.0em;
+ margin-bottom: 2.0em;
+ margin-right: 10%;
+ color: #606060;
+}
+
+div.content { /* Block element content. */
+ padding: 0;
+}
+
+/* Block element titles. */
+div.title, caption.title {
+ color: #527bbd;
+ font-weight: bold;
+ text-align: left;
+ margin-top: 1.0em;
+ margin-bottom: 0.5em;
+}
+div.title + * {
+ margin-top: 0;
+}
+
+td div.title:first-child {
+ margin-top: 0.0em;
+}
+div.content div.title:first-child {
+ margin-top: 0.0em;
+}
+div.content + div.title {
+ margin-top: 0.0em;
+}
+
+div.sidebarblock > div.content {
+ background: #ffffee;
+ border: 1px solid #dddddd;
+ border-left: 4px solid #f0f0f0;
+ padding: 0.5em;
+}
+
+div.listingblock > div.content {
+ border: 1px solid #dddddd;
+ border-left: 5px solid #f0f0f0;
+ background: #f8f8f8;
+ padding: 0.5em;
+}
+
+div.quoteblock, div.verseblock {
+ padding-left: 1.0em;
+ margin-left: 1.0em;
+ margin-right: 10%;
+ border-left: 5px solid #f0f0f0;
+ color: #888;
+}
+
+div.quoteblock > div.attribution {
+ padding-top: 0.5em;
+ text-align: right;
+}
+
+div.verseblock > pre.content {
+ font-family: inherit;
+ font-size: inherit;
+}
+div.verseblock > div.attribution {
+ padding-top: 0.75em;
+ text-align: left;
+}
+/* DEPRECATED: Pre version 8.2.7 verse style literal block. */
+div.verseblock + div.attribution {
+ text-align: left;
+}
+
+div.admonitionblock .icon {
+ vertical-align: top;
+ font-size: 1.1em;
+ font-weight: bold;
+ text-decoration: underline;
+ color: #527bbd;
+ padding-right: 0.5em;
+}
+div.admonitionblock td.content {
+ padding-left: 0.5em;
+ border-left: 3px solid #dddddd;
+}
+
+div.exampleblock > div.content {
+ border-left: 3px solid #dddddd;
+ padding-left: 0.5em;
+}
+
+div.imageblock div.content { padding-left: 0; }
+span.image img { border-style: none; vertical-align: text-bottom; }
+a.image:visited { color: white; }
+
+dl {
+ margin-top: 0.8em;
+ margin-bottom: 0.8em;
+}
+dt {
+ margin-top: 0.5em;
+ margin-bottom: 0;
+ font-style: normal;
+ color: navy;
+}
+dd > *:first-child {
+ margin-top: 0.1em;
+}
+
+ul, ol {
+ list-style-position: outside;
+}
+ol.arabic {
+ list-style-type: decimal;
+}
+ol.loweralpha {
+ list-style-type: lower-alpha;
+}
+ol.upperalpha {
+ list-style-type: upper-alpha;
+}
+ol.lowerroman {
+ list-style-type: lower-roman;
+}
+ol.upperroman {
+ list-style-type: upper-roman;
+}
+
+div.compact ul, div.compact ol,
+div.compact p, div.compact p,
+div.compact div, div.compact div {
+ margin-top: 0.1em;
+ margin-bottom: 0.1em;
+}
+
+tfoot {
+ font-weight: bold;
+}
+td > div.verse {
+ white-space: pre;
+}
+
+div.hdlist {
+ margin-top: 0.8em;
+ margin-bottom: 0.8em;
+}
+div.hdlist tr {
+ padding-bottom: 15px;
+}
+dt.hdlist1.strong, td.hdlist1.strong {
+ font-weight: bold;
+}
+td.hdlist1 {
+ vertical-align: top;
+ font-style: normal;
+ padding-right: 0.8em;
+ color: navy;
+}
+td.hdlist2 {
+ vertical-align: top;
+}
+div.hdlist.compact tr {
+ margin: 0;
+ padding-bottom: 0;
+}
+
+.comment {
+ background: yellow;
+}
+
+.footnote, .footnoteref {
+ font-size: 0.8em;
+}
+
+span.footnote, span.footnoteref {
+ vertical-align: super;
+}
+
+#footnotes {
+ margin: 20px 0 20px 0;
+ padding: 7px 0 0 0;
+}
+
+#footnotes div.footnote {
+ margin: 0 0 5px 0;
+}
+
+#footnotes hr {
+ border: none;
+ border-top: 1px solid silver;
+ height: 1px;
+ text-align: left;
+ margin-left: 0;
+ width: 20%;
+ min-width: 100px;
+}
+
+div.colist td {
+ padding-right: 0.5em;
+ padding-bottom: 0.3em;
+ vertical-align: top;
+}
+div.colist td img {
+ margin-top: 0.3em;
+}
+
+@media print {
+ #footer-badges { display: none; }
+}
+
+#toc {
+ margin-bottom: 2.5em;
+}
+
+#toctitle {
+ color: #527bbd;
+ font-size: 1.1em;
+ font-weight: bold;
+ margin-top: 1.0em;
+ margin-bottom: 0.1em;
+}
+
+div.toclevel0, div.toclevel1, div.toclevel2, div.toclevel3, div.toclevel4 {
+ margin-top: 0;
+ margin-bottom: 0;
+}
+div.toclevel2 {
+ margin-left: 2em;
+ font-size: 0.9em;
+}
+div.toclevel3 {
+ margin-left: 4em;
+ font-size: 0.9em;
+}
+div.toclevel4 {
+ margin-left: 6em;
+ font-size: 0.9em;
+}
+
+span.aqua { color: aqua; }
+span.black { color: black; }
+span.blue { color: blue; }
+span.fuchsia { color: fuchsia; }
+span.gray { color: gray; }
+span.green { color: green; }
+span.lime { color: lime; }
+span.maroon { color: maroon; }
+span.navy { color: navy; }
+span.olive { color: olive; }
+span.purple { color: purple; }
+span.red { color: red; }
+span.silver { color: silver; }
+span.teal { color: teal; }
+span.white { color: white; }
+span.yellow { color: yellow; }
+
+span.aqua-background { background: aqua; }
+span.black-background { background: black; }
+span.blue-background { background: blue; }
+span.fuchsia-background { background: fuchsia; }
+span.gray-background { background: gray; }
+span.green-background { background: green; }
+span.lime-background { background: lime; }
+span.maroon-background { background: maroon; }
+span.navy-background { background: navy; }
+span.olive-background { background: olive; }
+span.purple-background { background: purple; }
+span.red-background { background: red; }
+span.silver-background { background: silver; }
+span.teal-background { background: teal; }
+span.white-background { background: white; }
+span.yellow-background { background: yellow; }
+
+span.big { font-size: 2em; }
+span.small { font-size: 0.6em; }
+
+span.underline { text-decoration: underline; }
+span.overline { text-decoration: overline; }
+span.line-through { text-decoration: line-through; }
+
+div.unbreakable { page-break-inside: avoid; }
+
+
+/*
+ * xhtml11 specific
+ *
+ * */
+
+div.tableblock {
+ margin-top: 1.0em;
+ margin-bottom: 1.5em;
+}
+div.tableblock > table {
+ border: 3px solid #527bbd;
+}
+thead, p.table.header {
+ font-weight: bold;
+ color: #527bbd;
+}
+p.table {
+ margin-top: 0;
+}
+/* Because the table frame attribute is overridden by CSS in most browsers. */
+div.tableblock > table[frame="void"] {
+ border-style: none;
+}
+div.tableblock > table[frame="hsides"] {
+ border-left-style: none;
+ border-right-style: none;
+}
+div.tableblock > table[frame="vsides"] {
+ border-top-style: none;
+ border-bottom-style: none;
+}
+
+
+/*
+ * html5 specific
+ *
+ * */
+
+table.tableblock {
+ margin-top: 1.0em;
+ margin-bottom: 1.5em;
+}
+thead, p.tableblock.header {
+ font-weight: bold;
+ color: #527bbd;
+}
+p.tableblock {
+ margin-top: 0;
+}
+table.tableblock {
+ border-width: 3px;
+ border-spacing: 0px;
+ border-style: solid;
+ border-color: #527bbd;
+ border-collapse: collapse;
+}
+th.tableblock, td.tableblock {
+ border-width: 1px;
+ padding: 4px;
+ border-style: solid;
+ border-color: #527bbd;
+}
+
+table.tableblock.frame-topbot {
+ border-left-style: hidden;
+ border-right-style: hidden;
+}
+table.tableblock.frame-sides {
+ border-top-style: hidden;
+ border-bottom-style: hidden;
+}
+table.tableblock.frame-none {
+ border-style: hidden;
+}
+
+th.tableblock.halign-left, td.tableblock.halign-left {
+ text-align: left;
+}
+th.tableblock.halign-center, td.tableblock.halign-center {
+ text-align: center;
+}
+th.tableblock.halign-right, td.tableblock.halign-right {
+ text-align: right;
+}
+
+th.tableblock.valign-top, td.tableblock.valign-top {
+ vertical-align: top;
+}
+th.tableblock.valign-middle, td.tableblock.valign-middle {
+ vertical-align: middle;
+}
+th.tableblock.valign-bottom, td.tableblock.valign-bottom {
+ vertical-align: bottom;
+}
+
+
+/*
+ * manpage specific
+ *
+ * */
+
+body.manpage h1 {
+ padding-top: 0.5em;
+ padding-bottom: 0.5em;
+ border-top: 2px solid silver;
+ border-bottom: 2px solid silver;
+}
+body.manpage h2 {
+ border-style: none;
+}
+body.manpage div.sectionbody {
+ margin-left: 3em;
+}
+
+@media print {
+ body.manpage div#toc { display: none; }
+}
+
+
+</style>
+<script type="text/javascript">
+/*<+'])');
+ // Function that scans the DOM tree for header elements (the DOM2
+ // nodeIterator API would be a better technique but not supported by all
+ // browsers).
+ var iterate = function (el) {
+ for (var i = el.firstChild; i != null; i = i.nextSibling) {
+ if (i.nodeType == 1 /* Node.ELEMENT_NODE */) {
+ var mo = re.exec(i.tagName);
+ if (mo && (i.getAttribute("class") || i.getAttribute("className")) != "float") {
+ result[result.length] = new TocEntry(i, getText(i), mo[1]-1);
+ }
+ iterate(i);
+ }
+ }
+ }
+ iterate(el);
+ return result;
+ }
+
+ var toc = document.getElementById("toc");
+ if (!toc) {
+ return;
+ }
+
+ // Delete existing TOC entries in case we're reloading the TOC.
+ var tocEntriesToRemove = [];
+ var i;
+ for (i = 0; i < toc.childNodes.length; i++) {
+ var entry = toc.childNodes[i];
+ if (entry.nodeName.toLowerCase() == 'div'
+ && entry.getAttribute("class")
+ && entry.getAttribute("class").match(/^toclevel/))
+ tocEntriesToRemove.push(entry);
+ }
+ for (i = 0; i < tocEntriesToRemove.length; i++) {
+ toc.removeChild(tocEntriesToRemove[i]);
+ }
+
+ // Rebuild TOC entries.
+ var entries = tocEntries(document.getElementById("content"), toclevels);
+ for (var i = 0; i < entries.length; ++i) {
+ var entry = entries[i];
+ if (entry.element.id == "")
+ entry.element.id = "_toc_" + i;
+ var a = document.createElement("a");
+ a.href = "#" + entry.element.id;
+ a.appendChild(document.createTextNode(entry.text));
+ var div = document.createElement("div");
+ div.appendChild(a);
+ div.className = "toclevel" + entry.toclevel;
+ toc.appendChild(div);
+ }
+ if (entries.length == 0)
+ toc.parentNode.removeChild(toc);
+},
+
+
+/////////////////////////////////////////////////////////////////////
+// Footnotes generator
+/////////////////////////////////////////////////////////////////////
+
+/* Based on footnote generation code from:
+ * http://www.brandspankingnew.net/archive/2005/07/format_footnote.html
+ */
+
+footnotes: function () {
+ // Delete existing footnote entries in case we're reloading the footnodes.
+ var i;
+ var noteholder = document.getElementById("footnotes");
+ if (!noteholder) {
+ return;
+ }
+ var entriesToRemove = [];
+ for (i = 0; i < noteholder.childNodes.length; i++) {
+ var entry = noteholder.childNodes[i];
+ if (entry.nodeName.toLowerCase() == 'div' && entry.getAttribute("class") == "footnote")
+ entriesToRemove.push(entry);
+ }
+ for (i = 0; i < entriesToRemove.length; i++) {
+ noteholder.removeChild(entriesToRemove[i]);
+ }
+
+ // Rebuild footnote entries.
+ var cont = document.getElementById("content");
+ var spans = cont.getElementsByTagName("span");
+ var refs = {};
+ var n = 0;
+ for (i=0; i<spans.length; i++) {
+ if (spans[i].className == "footnote") {
+ n++;
+ var note = spans[i].getAttribute("data-note");
+ if (!note) {
+ // Use [\s\S] in place of . so multi-line matches work.
+ // Because JavaScript has no s (dotall) regex flag.
+ note = spans[i].innerHTML.match(/\s*\[([\s\S]*)]\s*/)[1];
+ spans[i].innerHTML =
+ "[<a id='_footnoteref_" + n + "' href='#_footnote_" + n +
+ "' title='View footnote' class='footnote'>" + n + "</a>]";
+ spans[i].setAttribute("data-note", note);
+ }
+ noteholder.innerHTML +=
+ "<div class='footnote' id='_footnote_" + n + "'>" +
+ "<a href='#_footnoteref_" + n + "' title='Return to text'>" +
+ n + "</a>. " + note + "</div>";
+ var id =spans[i].getAttribute("id");
+ if (id != null) refs["#"+id] = n;
+ }
+ }
+ if (n == 0)
+ noteholder.parentNode.removeChild(noteholder);
+ else {
+ // Process footnoterefs.
+ for (i=0; i<spans.length; i++) {
+ if (spans[i].className == "footnoteref") {
+ var href = spans[i].getElementsByTagName("a")[0].getAttribute("href");
+ href = href.match(/#.*/)[0]; // Because IE return full URL.
+ n = refs[href];
+ spans[i].innerHTML =
+ "[<a href='#_footnote_" + n +
+ "' title='View footnote' class='footnote'>" + n + "</a>]";
+ }
+ }
+ }
+},
+
+install: function(toclevels) {
+ var timerId;
+
+ function reinstall() {
+ asciidoc.footnotes();
+ if (toclevels) {
+ asciidoc.toc(toclevels);
+ }
+ }
+
+ function reinstallAndRemoveTimer() {
+ clearInterval(timerId);
+ reinstall();
+ }
+
+ timerId = setInterval(reinstall, 500);
+ if (document.addEventListener)
+ document.addEventListener("DOMContentLoaded", reinstallAndRemoveTimer, false);
+ else
+ window.onload = reinstallAndRemoveTimer;
+}
+
+}
+asciidoc.install();
+/*]]>*/
+</script>
+</head>
+<body class="article" style="max-width:60em">
+<div id="header">
+<h1>Rust based vhost-user I2C backend</h1>
+</div>
+<div id="content">
+<div id="preamble">
+<div class="sectionbody">
+<div class="paragraph"><p>There is a growing trend towards virtualization in areas other than the
+traditional server environment. The server environment is uniform in nature, but
+as we move towards a richer ecosystem in automotive, medical, general mobile,
+and the IoT spaces, more device abstractions, and way richer organizations are
+needed. Linaro’s <a href="https://www.linaro.org/projects/#automotive_STR">Project
+Stratos</a> is working towards developing hypervisor agnostic abstract devices
+leveraging virtio and extending hypervisor interfaces and standards to allow all
+architectures.</p></div>
+<div class="paragraph"><p>The Virtual Input/Output device (Virtio) standard provides an open interface for
+guest <a href="https://en.wikipedia.org/wiki/Virtual_machine">virtual machines</a> (VMs)
+to access simplified "virtual" devices, such as network adapters and block
+devices, in a paravirtualized environment. Virtio provides a straightforward,
+efficient, standard, and extensible mechanism for virtual devices, rather than a
+per-environment or per-OS mechanism.</p></div>
+<div class="paragraph"><p>Virtio adopts a frontend-backend architecture that enables a simple but flexible
+framework. The backend (BE) virtio driver, implemented by the hypervisor running
+on the host, exposes the virtio device to the guest OS through a standard
+transport method, like
+<a href="https://en.wikipedia.org/wiki/Peripheral_Component_Interconnect">PCI</a> or
+<a href="https://en.wikipedia.org/wiki/Memory-mapped_I/O">MMIO</a>. This virtio device,
+by design, looks like a physical device to the guest OS, which implements a
+frontend (FE) virtio driver compatible with the virtio device exposed by the
+hypervisor. The virtio device and driver communicate based on a set of
+predefined protocols as defined by the
+<a href="https://github.com/oasis-tcs/virtio-spec">virtio specification</a>, which is
+maintained by <a href="https://www.oasis-open.org/org/">OASIS</a>. The FE driver may
+implement zero or more Virtual queues (virtqueues), as defined by the virtio
+specification. The virtqueues are the mechanism of bulk data transport between
+FE (guest) and BE (host) drivers. These are normally implemented as standard
+ring buffers in the guest physical memory space. The BE drivers parse the
+virtqueues to obtain the request descriptors, process them and queue the
+response descriptors back to the virtqueue.</p></div>
+<div class="paragraph"><p>The FE virtio driver, at the guest, and the virtio specification are normally
+independent of where the virtqueue processing happens at the host, in-kernel or
+userspace. The virtio vhost protocol allows the virtio virtqueue processing at
+the host to be offloaded to another element, a user process or a kernel module.
+The vhost protocol, when implemented in userspace is called as "vhost-user".
+Since Linaro’s Project Stratos is targeting hypervisor agnostic BE solutions,
+engineers at Linaro decided to work over the existing vhost-user protocol. This
+article focuses on the Rust based vhost-user implementation for the
+<a href="https://en.wikipedia.org/wiki/I%C2%B2C">I2C</a> (or Inter-Integrated Circuit)
+devices.</p></div>
+</div>
+</div>
+<div class="sect1">
+<h2 id="_virtio_i2c_specification">Virtio I2C Specification</h2>
+<div class="sectionbody">
+<div class="paragraph"><p>The Virtio
+<a href="https://github.com/oasis-tcs/virtio-spec/blob/master/virtio-i2c.tex">specification</a>
+for I2C and the Linux
+<a href="https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/i2c/busses/i2c-virtio.c">i2c-virtio</a>
+driver are upstreamed by Jie Deng (Intel), who tested his work with the
+<a href="https://projectacrn.org">ACRN</a> hypervisor designed for IoT development. Both
+specification and driver received updates later on by Viresh Kumar (Linaro), to
+improve buffer management and allow zero-length transactions. Lets briefly go
+through the virtio I2C specification.</p></div>
+<div class="paragraph"><p><span class="monospaced">virtio-i2c</span> is a virtual I2C adapter device, which provides a way to flexibly
+organize and use the host I2C controlled devices from the guest. All
+communication between the FE and BE drivers happens over the <span class="monospaced">requestq</span>
+virtqueue. The I2C requests always originate at the guest FE driver, where the
+FE driver puts one or more I2C requests, represented by the <span class="monospaced">struct
+virtio_i2c_req</span>, on the <span class="monospaced">requestq</span> virtqueue. The I2C requests may or may not be
+be interdependent. If multiple requests are received together, then the host BE
+driver must process the requests in the order they are received on the
+virtqueue.</p></div>
+<div class="listingblock">
+<div class="content monospaced">
+<pre>struct virtio_i2c_req {
+ struct virtio_i2c_out_hdr out_hdr;
+ u8 buf[];
+ struct virtio_i2c_in_hdr in_hdr;
+};</pre>
+</div></div>
+<div class="paragraph"><p>Each I2C virtio request consists of an <span class="monospaced">out_hdr</span>, followed by an optional data
+buffer of some length, followed by an <span class="monospaced">in_hdr</span>. The buffer is not sent for the
+zero-length requests, like for the SMBus <span class="monospaced">QUICK</span> command where no data is
+required to be sent or received.</p></div>
+<div class="listingblock">
+<div class="content monospaced">
+<pre>struct virtio_i2c_out_hdr {
+ le16 addr;
+ le16 padding;
+ le32 flags;
+};</pre>
+</div></div>
+<div class="paragraph"><p>The <span class="monospaced">out_hdr</span> is represented by the <span class="monospaced">struct virtio_i2c_out_hdr</span> and is always
+set by the FE driver. The <span class="monospaced">addr</span> field of the header is set with the address of
+the I2C controlled device. Both 7-bit and 10-bit address modes are supported by
+the specification, though only 7-bit mode is supported by the current
+implementation of the Linux FE driver. The <span class="monospaced">flags</span> field is used to show
+dependency between multiple requests, by setting <span class="monospaced">VIRTIO_I2C_FLAGS_FAIL_NEXT</span>
+(0b01), or to mark a request <span class="monospaced">READ</span> or <span class="monospaced">WRITE</span>, by setting
+<span class="monospaced">VIRTIO_I2C_FLAGS_M_RD</span> (0b10) for <span class="monospaced">READ</span> operation.</p></div>
+<div class="paragraph"><p>As described earlier, <span class="monospaced">buf</span> is optional. The virtio specification for I2C
+defines a feature for zero-length transfers, <span class="monospaced">VIRTIO_I2C_F_ZERO_LENGTH_REQUEST</span>
+(0b01). It is mandatory for both FE and BE drivers to implement this feature,
+which allows zero-length transfers (like SMBus <span class="monospaced">QUICK</span> command) to take place.</p></div>
+<div class="paragraph"><p>For <span class="monospaced">WRITE</span> transactions, the buffer is set by the FE driver and read by the BE
+driver. For <span class="monospaced">READ</span> transactions, it is set by the BE driver and read by the FE
+driver after the response is received. The amount of the data to transfer is
+inferred by the size of the buffer descriptor.</p></div>
+<div class="listingblock">
+<div class="content monospaced">
+<pre>struct virtio_i2c_in_hdr {
+ u8 status;
+};</pre>
+</div></div>
+<div class="paragraph"><p>The <span class="monospaced">in_hdr</span> is represented by the <span class="monospaced">struct virtio_i2c_in_hdr</span> and is used by the
+host BE driver to notify the guest with the status of the transfer with
+<span class="monospaced">VIRTIO_I2C_MSG_OK</span> (0) or <span class="monospaced">VIRTIO_I2C_MSG_ERR</span> (1).</p></div>
+<div class="paragraph"><p>Please refer the Virtio I2C
+<a href="https://github.com/oasis-tcs/virtio-spec/blob/master/virtio-i2c.tex">specification</a>
+of more details.</p></div>
+</div>
+</div>
+<div class="sect1">
+<h2 id="_rust_based_i2c_backend">Rust based I2C backend</h2>
+<div class="sectionbody">
+<div class="paragraph"><p>Rust is the next big thing disrupting the Linux world. Most of us are already
+aware of the <a href="https://github.com/Rust-for-Linux">Rust for Linux</a> project
+slowly making its way into the Linux kernel. Rust is a multi-paradigm,
+general-purpose programming language designed for performance and safety. It
+brings a lot of benefits to the table, especially
+<a href="https://en.wikipedia.org/wiki/Memory_safety">memory-safety</a> and safe
+<a href="https://en.wikipedia.org/wiki/Concurrency_(computer_science)">concurrency</a>.</p></div>
+<div class="paragraph"><p>The <a href="https://github.com/rust-vmm">rust-vmm</a> project, an open-source
+initiative, was started back in late 2018, with the aim to share virtualization
+packages. The rust-vmm project lets one build custom
+<a href="https://en.wikipedia.org/wiki/Hypervisor">Virtual Machine Monitors (VMMs)
+and hypervisors</a>. This empowers other projects to quickly develop virtualization
+solutions, by reusing the components provided by rust-vmm, and better focus on
+key differentiators of their products. The rust-vmm project is organized as a
+shared ownership project, that so far includes contributions from Alibaba, AWS,
+Cloud Base, Google, Intel, Linaro, Red Hat and other individual contributors.
+The components provided by rust-vmm are already used by several projects, like
+Amazon’s <a href="https://github.com/firecracker-microvm/firecracker">Firecracker</a>
+and Intel’s <a href="https://github.com/cloud-hypervisor/cloud-hypervisor">Cloud
+Hypervisor</a>. The rust-vmm project currently hosts ~30 repositories (or Rust
+crates, equivalent of a C library), where each crate plays a specialized role in
+the development of a fully functioning VMM.</p></div>
+<div class="paragraph"><p>One such component provided by the rust-vmm project is the
+<a href="https://crates.io/crates/vhost-user-backend">vhost-user-backend</a> crate,
+which has recently made its way to <a href="https://crates.io/">crates.io</a>, the Rust
+community’s crate registry. The vhost-user-backend crate provides a framework to
+implement the vhost-user backend services. It provides necessary public APIs to
+support vhost-user backends, like a daemon control object (<span class="monospaced">VhostUserDaemon</span>) to
+start and stop the service daemon, a vhost-user backend trait
+(<span class="monospaced">VhostUserBackendMut</span>) to handle vhost-user control messages and virtio
+messages, and a vring access trait (<span class="monospaced">VringT</span>) to access virtio queues. A Rust
+trait tells the Rust compiler about functionality a particular type has and can
+share with other types.</p></div>
+<div class="paragraph"><p>A separate Rust workspace,
+<a href="https://github.com/rust-vmm/vhost-device">vhost-device</a>, is recently created
+in the rust-vmm project, to host per-device vhost-user backend crates. The only
+crate merged there as for now is the I2C device crate, while there are others
+getting developed and reviewed as we speak, like GPIO, RNG, VSOCK, SCSI, and
+<a href="https://en.wikipedia.org/wiki/Replay_Protected_Memory_Block">RPMB</a>.</p></div>
+<div class="paragraph"><p>The I2C vhost-device binary-crate (generates an executable upon build),
+developed by Viresh Kumar (Linaro), supports sharing host I2C busses (Adaptors)
+and client devices with multiple guest VMs at the same time with a single
+instance of the backend daemon. Once the vhost-device crate is compiled with
+<span class="monospaced">cargo build --release</span> command, it generates the
+<span class="monospaced">target/release/vhost-device-i2c</span> executable. The <span class="monospaced">vhost-device-i2c</span> daemon
+communicates with guest VMs over Unix domain sockets, a unique socket for each
+VM.</p></div>
+<div class="paragraph"><p>The daemon accepts these arguments:</p></div>
+<div class="ulist"><ul>
+<li>
+<p>
+-s, --socket-path: Path of the vhost-user Unix domain sockets. This is
+ suffixed with 0,1,2..socket_count-1 by the daemon to obtain actual socket
+ paths.
+</p>
+</li>
+<li>
+<p>
+-c, --socket-count: Number of sockets (guests) to connect to. This parameter
+ is optional and defaults to 1.
+</p>
+</li>
+<li>
+<p>
+-l, --device-list: List of I2C busses and clients in the format
+ <bus>:<client_addr>[:<client_addr>][,<bus>:<client_addr>[:<client_addr>]]
+</p>
+</li>
+</ul></div>
+<div class="paragraph"><p>As an example, consider the following command:</p></div>
+<div class="listingblock">
+<div class="content monospaced">
+<pre>./vhost-device-i2c -s ~/i2c.sock -c 6 -l 6:32:41,9:37:6</pre>
+</div></div>
+<div class="paragraph"><p>This will start the I2C backend daemon, which will create 6 Unix domain sockets
+(<span class="monospaced">~/i2c.sock0</span>, .. <span class="monospaced">~/i2c.sock5</span>), in order to communicate with 6 guest VMs,
+where communication with each VM happens in parallel with the help of a separate
+native OS thread. Each thread, once created by the daemon, will wait for a VM to
+start communicating over the thread’s designated socket. Once a VM is found for
+the thread, the thread registers a <span class="monospaced">vhost-user-backend</span> instance and starts
+processing the requests on the <span class="monospaced">requestq</span> virtqueue. At a later point of time,
+once the VM shuts down, the respective thread starts waiting for a new VM to
+communicate on the same socket path. In the above example, the daemon is also
+passed a list of host I2C busses and client devices, which are shared among the
+VMs. This is how sharing is defined in the daemon’s implementation for now,
+though it can be modified later on, if required, to allow specific devices
+to be accessed only by a particular VM. In the above example, the devices
+provided by the host to the daemon are: devices with address 32 and 41 attached
+to I2C bus 6, and 37 and 6 attached to I2C bus 9. The daemon extensively
+validates the device-list at initialization to avoid any failures later,
+especially for duplicate entries.</p></div>
+<div class="paragraph"><p>The <span class="monospaced">vhost-user-i2c</span> daemon supports both I2C and SMBus protocols, only basic
+SMBus commands up to word-transfer though. The backend provides the <span class="monospaced">pub trait
+I2cDevice</span>, a public Rust trait, which can be implemented for different host
+environments to provide access to the underlying I2C busses and devices. This is
+currently implemented only for the Linux userspace, where the I2C busses and
+devices are accessed via the <span class="monospaced">/dev/i2c-X</span> device files. For the above example,
+the backend daemon will look for <span class="monospaced">/dev/i2c-6</span> and <span class="monospaced">/dev/i2c-9</span> device files. The
+users may need to load the standard <span class="monospaced">i2c-dev</span> kernel module on the host machine,
+if not loaded already, for these device files to be available under <span class="monospaced">/dev/</span>. For
+a different host environment, like with a bare-metal type 1 hypervisor, we need
+to add another implementation of the trait depending on how the I2C busses and
+devices are accessed.</p></div>
+<div class="paragraph"><p>The <span class="monospaced">vhost-user-i2c</span> backend is truly a hypervisor agnostic solution that works
+with any hypervisor which understands the vhost-user protocol. It has been
+extensively tested with QEMU for example, with Linux userspace environment. Work
+is in progress to make Xen hypervisor vhost-user protocol compatible. Once that
+is done, we will be able to use the same <span class="monospaced">vhost-user-i2c</span> executable with both
+QEMU and Xen, for example, under the same host environment.</p></div>
+<div class="paragraph"><p>Support for i2c-virtio is already merged in QEMU source, boilerplate stuff to
+create the i2c-virtio device in the guest kernel, and the i2c-virtio device can
+be created in the guest kernel by adding following command line arguments to
+your QEMU command:</p></div>
+<div class="listingblock">
+<div class="content monospaced">
+<pre>-chardev socket,path=~/i2c.sock0,id=vi2c -device vhost-user-i2c-device,chardev=vi2c,id=i2c</pre>
+</div></div>
+<div class="paragraph"><p>We have come a long way forward with the I2C vhost-user device implementation in
+the <a href="https://github.com/rust-vmm/vhost-device">vhost-device</a> workspace. But
+there is still a lot to do, specially testing the same vhost-user backend
+executables with multiple hypervisors, which makes this a truly hypervisor
+agnostic solution. As said earlier, there is work going on currently in that
+area. Moreover, this workspace will receive more device specific crates in
+future.</p></div>
+</div>
+</div>
+</div>
+<div id="footnotes"><hr></div>
+<div id="footer">
+<div id="footer-text">
+Last updated
+ 2022-01-05 16:22:33 IST
+</div>
+</div>
+</body>
+</html>
diff --git a/rust/i2c.txt b/rust/i2c.txt new file mode 100644 index 0000000..2741af7 --- /dev/null +++ b/rust/i2c.txt @@ -0,0 +1,255 @@ +Rust based vhost-user I2C backend +================================= + +There is a growing trend towards virtualization in areas other than the +traditional server environment. The server environment is uniform in nature, but +as we move towards a richer ecosystem in automotive, medical, general mobile, +and the IoT spaces, more device abstractions, and way richer organizations are +needed. Linaro's link:https://www.linaro.org/projects/#automotive_STR[Project +Stratos] is working towards developing hypervisor agnostic abstract devices +leveraging virtio and extending hypervisor interfaces and standards to allow all +architectures. + +The Virtual Input/Output device (Virtio) standard provides an open interface for +guest link:https://en.wikipedia.org/wiki/Virtual_machine[virtual machines] (VMs) +to access simplified "virtual" devices, such as network adapters and block +devices, in a paravirtualized environment. Virtio provides a straightforward, +efficient, standard, and extensible mechanism for virtual devices, rather than a +per-environment or per-OS mechanism. + +Virtio adopts a frontend-backend architecture that enables a simple but flexible +framework. The backend (BE) virtio driver, implemented by the hypervisor running +on the host, exposes the virtio device to the guest OS through a standard +transport method, like +link:https://en.wikipedia.org/wiki/Peripheral_Component_Interconnect[PCI] or +link:https://en.wikipedia.org/wiki/Memory-mapped_I/O[MMIO]. This virtio device, +by design, looks like a physical device to the guest OS, which implements a +frontend (FE) virtio driver compatible with the virtio device exposed by the +hypervisor. The virtio device and driver communicate based on a set of +predefined protocols as defined by the +link:https://github.com/oasis-tcs/virtio-spec[virtio specification], which is +maintained by link:https://www.oasis-open.org/org/[OASIS]. The FE driver may +implement zero or more Virtual queues (virtqueues), as defined by the virtio +specification. The virtqueues are the mechanism of bulk data transport between +FE (guest) and BE (host) drivers. These are normally implemented as standard +ring buffers in the guest physical memory space. The BE drivers parse the +virtqueues to obtain the request descriptors, process them and queue the +response descriptors back to the virtqueue. + +The FE virtio driver, at the guest, and the virtio specification are normally +independent of where the virtqueue processing happens at the host, in-kernel or +userspace. The virtio vhost protocol allows the virtio virtqueue processing at +the host to be offloaded to another element, a user process or a kernel module. +The vhost protocol, when implemented in userspace is called as "vhost-user". +Since Linaro's Project Stratos is targeting hypervisor agnostic BE solutions, +engineers at Linaro decided to work over the existing vhost-user protocol. This +article focuses on the Rust based vhost-user implementation for the +link:https://en.wikipedia.org/wiki/I%C2%B2C[I2C] (or Inter-Integrated Circuit) +devices. + +Virtio I2C Specification +------------------------ + +The Virtio +link:https://github.com/oasis-tcs/virtio-spec/blob/master/virtio-i2c.tex[specification] +for I2C and the Linux +link:https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/i2c/busses/i2c-virtio.c[i2c-virtio] +driver are upstreamed by Jie Deng (Intel), who tested his work with the +link:https://projectacrn.org[ACRN] hypervisor designed for IoT development. Both +specification and driver received updates later on by Viresh Kumar (Linaro), to +improve buffer management and allow zero-length transactions. Lets briefly go +through the virtio I2C specification. + +`virtio-i2c` is a virtual I2C adapter device, which provides a way to flexibly +organize and use the host I2C controlled devices from the guest. All +communication between the FE and BE drivers happens over the `requestq` +virtqueue. The I2C requests always originate at the guest FE driver, where the +FE driver puts one or more I2C requests, represented by the `struct +virtio_i2c_req`, on the `requestq` virtqueue. The I2C requests may or may not be +be interdependent. If multiple requests are received together, then the host BE +driver must process the requests in the order they are received on the +virtqueue. + +---- +struct virtio_i2c_req { + struct virtio_i2c_out_hdr out_hdr; + u8 buf[]; + struct virtio_i2c_in_hdr in_hdr; +}; +---- + +Each I2C virtio request consists of an `out_hdr`, followed by an optional data +buffer of some length, followed by an `in_hdr`. The buffer is not sent for the +zero-length requests, like for the SMBus `QUICK` command where no data is +required to be sent or received. + +---- +struct virtio_i2c_out_hdr { + le16 addr; + le16 padding; + le32 flags; +}; +---- + +The `out_hdr` is represented by the `struct virtio_i2c_out_hdr` and is always +set by the FE driver. The `addr` field of the header is set with the address of +the I2C controlled device. Both 7-bit and 10-bit address modes are supported by +the specification, though only 7-bit mode is supported by the current +implementation of the Linux FE driver. The `flags` field is used to show +dependency between multiple requests, by setting `VIRTIO_I2C_FLAGS_FAIL_NEXT` +(0b01), or to mark a request `READ` or `WRITE`, by setting +`VIRTIO_I2C_FLAGS_M_RD` (0b10) for `READ` operation. + +As described earlier, `buf` is optional. The virtio specification for I2C +defines a feature for zero-length transfers, `VIRTIO_I2C_F_ZERO_LENGTH_REQUEST` +(0b01). It is mandatory for both FE and BE drivers to implement this feature, +which allows zero-length transfers (like SMBus `QUICK` command) to take place. + +For `WRITE` transactions, the buffer is set by the FE driver and read by the BE +driver. For `READ` transactions, it is set by the BE driver and read by the FE +driver after the response is received. The amount of the data to transfer is +inferred by the size of the buffer descriptor. + +---- +struct virtio_i2c_in_hdr { + u8 status; +}; +---- + +The `in_hdr` is represented by the `struct virtio_i2c_in_hdr` and is used by the +host BE driver to notify the guest with the status of the transfer with +`VIRTIO_I2C_MSG_OK` (0) or `VIRTIO_I2C_MSG_ERR` (1). + +Please refer the Virtio I2C +link:https://github.com/oasis-tcs/virtio-spec/blob/master/virtio-i2c.tex[specification] +of more details. + +Rust based I2C backend +---------------------- + +Rust is the next big thing disrupting the Linux world. Most of us are already +aware of the link:https://github.com/Rust-for-Linux[Rust for Linux] project +slowly making its way into the Linux kernel. Rust is a multi-paradigm, +general-purpose programming language designed for performance and safety. It +brings a lot of benefits to the table, especially +link:https://en.wikipedia.org/wiki/Memory_safety[memory-safety] and safe +link:https://en.wikipedia.org/wiki/Concurrency_(computer_science)[concurrency]. + +The link:https://github.com/rust-vmm[rust-vmm] project, an open-source +initiative, was started back in late 2018, with the aim to share virtualization +packages. The rust-vmm project lets one build custom +link:https://en.wikipedia.org/wiki/Hypervisor[Virtual Machine Monitors (VMMs) +and hypervisors]. This empowers other projects to quickly develop virtualization +solutions, by reusing the components provided by rust-vmm, and better focus on +key differentiators of their products. The rust-vmm project is organized as a +shared ownership project, that so far includes contributions from Alibaba, AWS, +Cloud Base, Google, Intel, Linaro, Red Hat and other individual contributors. +The components provided by rust-vmm are already used by several projects, like +Amazon's link:https://github.com/firecracker-microvm/firecracker[Firecracker] +and Intel's link:https://github.com/cloud-hypervisor/cloud-hypervisor[Cloud +Hypervisor]. The rust-vmm project currently hosts ~30 repositories (or Rust +crates, equivalent of a C library), where each crate plays a specialized role in +the development of a fully functioning VMM. + +One such component provided by the rust-vmm project is the +link:https://crates.io/crates/vhost-user-backend[vhost-user-backend] crate, +which has recently made its way to link:https://crates.io/[crates.io], the Rust +community’s crate registry. The vhost-user-backend crate provides a framework to +implement the vhost-user backend services. It provides necessary public APIs to +support vhost-user backends, like a daemon control object (`VhostUserDaemon`) to +start and stop the service daemon, a vhost-user backend trait +(`VhostUserBackendMut`) to handle vhost-user control messages and virtio +messages, and a vring access trait (`VringT`) to access virtio queues. A Rust +trait tells the Rust compiler about functionality a particular type has and can +share with other types. + +A separate Rust workspace, +link:https://github.com/rust-vmm/vhost-device[vhost-device], is recently created +in the rust-vmm project, to host per-device vhost-user backend crates. The only +crate merged there as for now is the +link:https://github.com/rust-vmm/vhost-device/tree/main/i2c[I2C] device crate, +while there are others getting developed and reviewed as we speak, like GPIO, +RNG, VSOCK, SCSI, and +link:https://en.wikipedia.org/wiki/Replay_Protected_Memory_Block[RPMB]. + +The I2C vhost-device binary-crate (generates an executable upon build), +developed by Viresh Kumar (Linaro), supports sharing host I2C busses (Adaptors) +and client devices with multiple guest VMs at the same time with a single +instance of the backend daemon. Once the vhost-device crate is compiled with +`cargo build --release` command, it generates the +`target/release/vhost-device-i2c` executable. The `vhost-device-i2c` daemon +communicates with guest VMs over Unix domain sockets, a unique socket for each +VM. + +The daemon accepts these arguments: + +* -s, --socket-path: Path of the vhost-user Unix domain sockets. This is + suffixed with 0,1,2..socket_count-1 by the daemon to obtain actual socket + paths. + +* -c, --socket-count: Number of sockets (guests) to connect to. This parameter + is optional and defaults to 1. + +* -l, --device-list: List of I2C busses and clients in the format + <bus>:<client_addr>[:<client_addr>][,<bus>:<client_addr>[:<client_addr>]] + +As an example, consider the following command: + +---- +./vhost-device-i2c -s ~/i2c.sock -c 6 -l 6:32:41,9:37:6 +---- + +This will start the I2C backend daemon, which will create 6 Unix domain sockets +(`~/i2c.sock0`, .. `~/i2c.sock5`), in order to communicate with 6 guest VMs, +where communication with each VM happens in parallel with the help of a separate +native OS thread. Each thread, once created by the daemon, will wait for a VM to +start communicating over the thread's designated socket. Once a VM is found for +the thread, the thread registers a `vhost-user-backend` instance and starts +processing the requests on the `requestq` virtqueue. At a later point of time, +once the VM shuts down, the respective thread starts waiting for a new VM to +communicate on the same socket path. In the above example, the daemon is also +passed a list of host I2C busses and client devices, which are shared among the +VMs. This is how sharing is defined in the daemon's implementation for now, +though it can be modified later on, if required, to allow specific devices +to be accessed only by a particular VM. In the above example, the devices +provided by the host to the daemon are: devices with address 32 and 41 attached +to I2C bus 6, and 37 and 6 attached to I2C bus 9. The daemon extensively +validates the device-list at initialization to avoid any failures later, +especially for duplicate entries. + +The `vhost-user-i2c` daemon supports both I2C and SMBus protocols, only basic +SMBus commands up to word-transfer though. The backend provides the `pub trait +I2cDevice`, a public Rust trait, which can be implemented for different host +environments to provide access to the underlying I2C busses and devices. This is +currently implemented only for the Linux userspace, where the I2C busses and +devices are accessed via the `/dev/i2c-X` device files. For the above example, +the backend daemon will look for `/dev/i2c-6` and `/dev/i2c-9` device files. The +users may need to load the standard `i2c-dev` kernel module on the host machine, +if not loaded already, for these device files to be available under `/dev/`. For +a different host environment, like with a bare-metal type 1 hypervisor, we need +to add another implementation of the trait depending on how the I2C busses and +devices are accessed. + +The `vhost-user-i2c` backend is truly a hypervisor agnostic solution that works +with any hypervisor which understands the vhost-user protocol. It has been +extensively tested with QEMU for example, with Linux userspace environment. Work +is in progress to make Xen hypervisor vhost-user protocol compatible. Once that +is done, we will be able to use the same `vhost-user-i2c` executable with both +QEMU and Xen, for example, under the same host environment. + +Support for i2c-virtio is already merged in QEMU source, boilerplate stuff to +create the i2c-virtio device in the guest kernel, and the i2c-virtio device can +be created in the guest kernel by adding following command line arguments to +your QEMU command: + +---- +-chardev socket,path=~/i2c.sock0,id=vi2c -device vhost-user-i2c-device,chardev=vi2c,id=i2c +---- + +We have come a long way forward with the I2C vhost-user device implementation in +the link:https://github.com/rust-vmm/vhost-device[vhost-device] workspace. But +there is still a lot to do, specially testing the same vhost-user backend +executables with multiple hypervisors, which makes this a truly hypervisor +agnostic solution. As said earlier, there is work going on currently in that +area. Moreover, this workspace will receive more device specific crates in +future. |