Root/Documentation/remoteproc.txt

1Remote Processor Framework
2
31. Introduction
4
5Modern SoCs typically have heterogeneous remote processor devices in asymmetric
6multiprocessing (AMP) configurations, which may be running different instances
7of operating system, whether it's Linux or any other flavor of real-time OS.
8
9OMAP4, for example, has dual Cortex-A9, dual Cortex-M3 and a C64x+ DSP.
10In a typical configuration, the dual cortex-A9 is running Linux in a SMP
11configuration, and each of the other three cores (two M3 cores and a DSP)
12is running its own instance of RTOS in an AMP configuration.
13
14The remoteproc framework allows different platforms/architectures to
15control (power on, load firmware, power off) those remote processors while
16abstracting the hardware differences, so the entire driver doesn't need to be
17duplicated. In addition, this framework also adds rpmsg virtio devices
18for remote processors that supports this kind of communication. This way,
19platform-specific remoteproc drivers only need to provide a few low-level
20handlers, and then all rpmsg drivers will then just work
21(for more information about the virtio-based rpmsg bus and its drivers,
22please read Documentation/rpmsg.txt).
23Registration of other types of virtio devices is now also possible. Firmwares
24just need to publish what kind of virtio devices do they support, and then
25remoteproc will add those devices. This makes it possible to reuse the
26existing virtio drivers with remote processor backends at a minimal development
27cost.
28
292. User API
30
31  int rproc_boot(struct rproc *rproc)
32    - Boot a remote processor (i.e. load its firmware, power it on, ...).
33      If the remote processor is already powered on, this function immediately
34      returns (successfully).
35      Returns 0 on success, and an appropriate error value otherwise.
36      Note: to use this function you should already have a valid rproc
37      handle. There are several ways to achieve that cleanly (devres, pdata,
38      the way remoteproc_rpmsg.c does this, or, if this becomes prevalent, we
39      might also consider using dev_archdata for this).
40
41  void rproc_shutdown(struct rproc *rproc)
42    - Power off a remote processor (previously booted with rproc_boot()).
43      In case @rproc is still being used by an additional user(s), then
44      this function will just decrement the power refcount and exit,
45      without really powering off the device.
46      Every call to rproc_boot() must (eventually) be accompanied by a call
47      to rproc_shutdown(). Calling rproc_shutdown() redundantly is a bug.
48      Notes:
49      - we're not decrementing the rproc's refcount, only the power refcount.
50        which means that the @rproc handle stays valid even after
51        rproc_shutdown() returns, and users can still use it with a subsequent
52        rproc_boot(), if needed.
53
543. Typical usage
55
56#include <linux/remoteproc.h>
57
58/* in case we were given a valid 'rproc' handle */
59int dummy_rproc_example(struct rproc *my_rproc)
60{
61    int ret;
62
63    /* let's power on and boot our remote processor */
64    ret = rproc_boot(my_rproc);
65    if (ret) {
66        /*
67         * something went wrong. handle it and leave.
68         */
69    }
70
71    /*
72     * our remote processor is now powered on... give it some work
73     */
74
75    /* let's shut it down now */
76    rproc_shutdown(my_rproc);
77}
78
794. API for implementors
80
81  struct rproc *rproc_alloc(struct device *dev, const char *name,
82                const struct rproc_ops *ops,
83                const char *firmware, int len)
84    - Allocate a new remote processor handle, but don't register
85      it yet. Required parameters are the underlying device, the
86      name of this remote processor, platform-specific ops handlers,
87      the name of the firmware to boot this rproc with, and the
88      length of private data needed by the allocating rproc driver (in bytes).
89
90      This function should be used by rproc implementations during
91      initialization of the remote processor.
92      After creating an rproc handle using this function, and when ready,
93      implementations should then call rproc_add() to complete
94      the registration of the remote processor.
95      On success, the new rproc is returned, and on failure, NULL.
96
97      Note: _never_ directly deallocate @rproc, even if it was not registered
98      yet. Instead, when you need to unroll rproc_alloc(), use rproc_put().
99
100  void rproc_put(struct rproc *rproc)
101    - Free an rproc handle that was allocated by rproc_alloc.
102      This function essentially unrolls rproc_alloc(), by decrementing the
103      rproc's refcount. It doesn't directly free rproc; that would happen
104      only if there are no other references to rproc and its refcount now
105      dropped to zero.
106
107  int rproc_add(struct rproc *rproc)
108    - Register @rproc with the remoteproc framework, after it has been
109      allocated with rproc_alloc().
110      This is called by the platform-specific rproc implementation, whenever
111      a new remote processor device is probed.
112      Returns 0 on success and an appropriate error code otherwise.
113      Note: this function initiates an asynchronous firmware loading
114      context, which will look for virtio devices supported by the rproc's
115      firmware.
116      If found, those virtio devices will be created and added, so as a result
117      of registering this remote processor, additional virtio drivers might get
118      probed.
119
120  int rproc_del(struct rproc *rproc)
121    - Unroll rproc_add().
122      This function should be called when the platform specific rproc
123      implementation decides to remove the rproc device. it should
124      _only_ be called if a previous invocation of rproc_add()
125      has completed successfully.
126
127      After rproc_del() returns, @rproc is still valid, and its
128      last refcount should be decremented by calling rproc_put().
129
130      Returns 0 on success and -EINVAL if @rproc isn't valid.
131
1325. Implementation callbacks
133
134These callbacks should be provided by platform-specific remoteproc
135drivers:
136
137/**
138 * struct rproc_ops - platform-specific device handlers
139 * @start: power on the device and boot it
140 * @stop: power off the device
141 * @kick: kick a virtqueue (virtqueue id given as a parameter)
142 */
143struct rproc_ops {
144    int (*start)(struct rproc *rproc);
145    int (*stop)(struct rproc *rproc);
146    void (*kick)(struct rproc *rproc, int vqid);
147};
148
149Every remoteproc implementation should at least provide the ->start and ->stop
150handlers. If rpmsg/virtio functionality is also desired, then the ->kick handler
151should be provided as well.
152
153The ->start() handler takes an rproc handle and should then power on the
154device and boot it (use rproc->priv to access platform-specific private data).
155The boot address, in case needed, can be found in rproc->bootaddr (remoteproc
156core puts there the ELF entry point).
157On success, 0 should be returned, and on failure, an appropriate error code.
158
159The ->stop() handler takes an rproc handle and powers the device down.
160On success, 0 is returned, and on failure, an appropriate error code.
161
162The ->kick() handler takes an rproc handle, and an index of a virtqueue
163where new message was placed in. Implementations should interrupt the remote
164processor and let it know it has pending messages. Notifying remote processors
165the exact virtqueue index to look in is optional: it is easy (and not
166too expensive) to go through the existing virtqueues and look for new buffers
167in the used rings.
168
1696. Binary Firmware Structure
170
171At this point remoteproc only supports ELF32 firmware binaries. However,
172it is quite expected that other platforms/devices which we'd want to
173support with this framework will be based on different binary formats.
174
175When those use cases show up, we will have to decouple the binary format
176from the framework core, so we can support several binary formats without
177duplicating common code.
178
179When the firmware is parsed, its various segments are loaded to memory
180according to the specified device address (might be a physical address
181if the remote processor is accessing memory directly).
182
183In addition to the standard ELF segments, most remote processors would
184also include a special section which we call "the resource table".
185
186The resource table contains system resources that the remote processor
187requires before it should be powered on, such as allocation of physically
188contiguous memory, or iommu mapping of certain on-chip peripherals.
189Remotecore will only power up the device after all the resource table's
190requirement are met.
191
192In addition to system resources, the resource table may also contain
193resource entries that publish the existence of supported features
194or configurations by the remote processor, such as trace buffers and
195supported virtio devices (and their configurations).
196
197The resource table begins with this header:
198
199/**
200 * struct resource_table - firmware resource table header
201 * @ver: version number
202 * @num: number of resource entries
203 * @reserved: reserved (must be zero)
204 * @offset: array of offsets pointing at the various resource entries
205 *
206 * The header of the resource table, as expressed by this structure,
207 * contains a version number (should we need to change this format in the
208 * future), the number of available resource entries, and their offsets
209 * in the table.
210 */
211struct resource_table {
212    u32 ver;
213    u32 num;
214    u32 reserved[2];
215    u32 offset[0];
216} __packed;
217
218Immediately following this header are the resource entries themselves,
219each of which begins with the following resource entry header:
220
221/**
222 * struct fw_rsc_hdr - firmware resource entry header
223 * @type: resource type
224 * @data: resource data
225 *
226 * Every resource entry begins with a 'struct fw_rsc_hdr' header providing
227 * its @type. The content of the entry itself will immediately follow
228 * this header, and it should be parsed according to the resource type.
229 */
230struct fw_rsc_hdr {
231    u32 type;
232    u8 data[0];
233} __packed;
234
235Some resources entries are mere announcements, where the host is informed
236of specific remoteproc configuration. Other entries require the host to
237do something (e.g. allocate a system resource). Sometimes a negotiation
238is expected, where the firmware requests a resource, and once allocated,
239the host should provide back its details (e.g. address of an allocated
240memory region).
241
242Here are the various resource types that are currently supported:
243
244/**
245 * enum fw_resource_type - types of resource entries
246 *
247 * @RSC_CARVEOUT: request for allocation of a physically contiguous
248 * memory region.
249 * @RSC_DEVMEM: request to iommu_map a memory-based peripheral.
250 * @RSC_TRACE: announces the availability of a trace buffer into which
251 * the remote processor will be writing logs.
252 * @RSC_VDEV: declare support for a virtio device, and serve as its
253 * virtio header.
254 * @RSC_LAST: just keep this one at the end
255 *
256 * Please note that these values are used as indices to the rproc_handle_rsc
257 * lookup table, so please keep them sane. Moreover, @RSC_LAST is used to
258 * check the validity of an index before the lookup table is accessed, so
259 * please update it as needed.
260 */
261enum fw_resource_type {
262    RSC_CARVEOUT = 0,
263    RSC_DEVMEM = 1,
264    RSC_TRACE = 2,
265    RSC_VDEV = 3,
266    RSC_LAST = 4,
267};
268
269For more details regarding a specific resource type, please see its
270dedicated structure in include/linux/remoteproc.h.
271
272We also expect that platform-specific resource entries will show up
273at some point. When that happens, we could easily add a new RSC_PLATFORM
274type, and hand those resources to the platform-specific rproc driver to handle.
275
2767. Virtio and remoteproc
277
278The firmware should provide remoteproc information about virtio devices
279that it supports, and their configurations: a RSC_VDEV resource entry
280should specify the virtio device id (as in virtio_ids.h), virtio features,
281virtio config space, vrings information, etc.
282
283When a new remote processor is registered, the remoteproc framework
284will look for its resource table and will register the virtio devices
285it supports. A firmware may support any number of virtio devices, and
286of any type (a single remote processor can also easily support several
287rpmsg virtio devices this way, if desired).
288
289Of course, RSC_VDEV resource entries are only good enough for static
290allocation of virtio devices. Dynamic allocations will also be made possible
291using the rpmsg bus (similar to how we already do dynamic allocations of
292rpmsg channels; read more about it in rpmsg.txt).
293

Archive Download this file



interactive