linux/Documentation/remoteproc.txt
<<
>>
Prefs
   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
  54  struct rproc *rproc_get_by_phandle(phandle phandle)
  55    - Find an rproc handle using a device tree phandle. Returns the rproc
  56      handle on success, and NULL on failure. This function increments
  57      the remote processor's refcount, so always use rproc_put() to
  58      decrement it back once rproc isn't needed anymore.
  59
  603. Typical usage
  61
  62#include <linux/remoteproc.h>
  63
  64/* in case we were given a valid 'rproc' handle */
  65int dummy_rproc_example(struct rproc *my_rproc)
  66{
  67        int ret;
  68
  69        /* let's power on and boot our remote processor */
  70        ret = rproc_boot(my_rproc);
  71        if (ret) {
  72                /*
  73                 * something went wrong. handle it and leave.
  74                 */
  75        }
  76
  77        /*
  78         * our remote processor is now powered on... give it some work
  79         */
  80
  81        /* let's shut it down now */
  82        rproc_shutdown(my_rproc);
  83}
  84
  854. API for implementors
  86
  87  struct rproc *rproc_alloc(struct device *dev, const char *name,
  88                                const struct rproc_ops *ops,
  89                                const char *firmware, int len)
  90    - Allocate a new remote processor handle, but don't register
  91      it yet. Required parameters are the underlying device, the
  92      name of this remote processor, platform-specific ops handlers,
  93      the name of the firmware to boot this rproc with, and the
  94      length of private data needed by the allocating rproc driver (in bytes).
  95
  96      This function should be used by rproc implementations during
  97      initialization of the remote processor.
  98      After creating an rproc handle using this function, and when ready,
  99      implementations should then call rproc_add() to complete
 100      the registration of the remote processor.
 101      On success, the new rproc is returned, and on failure, NULL.
 102
 103      Note: _never_ directly deallocate @rproc, even if it was not registered
 104      yet. Instead, when you need to unroll rproc_alloc(), use rproc_free().
 105
 106  void rproc_free(struct rproc *rproc)
 107    - Free an rproc handle that was allocated by rproc_alloc.
 108      This function essentially unrolls rproc_alloc(), by decrementing the
 109      rproc's refcount. It doesn't directly free rproc; that would happen
 110      only if there are no other references to rproc and its refcount now
 111      dropped to zero.
 112
 113  int rproc_add(struct rproc *rproc)
 114    - Register @rproc with the remoteproc framework, after it has been
 115      allocated with rproc_alloc().
 116      This is called by the platform-specific rproc implementation, whenever
 117      a new remote processor device is probed.
 118      Returns 0 on success and an appropriate error code otherwise.
 119      Note: this function initiates an asynchronous firmware loading
 120      context, which will look for virtio devices supported by the rproc's
 121      firmware.
 122      If found, those virtio devices will be created and added, so as a result
 123      of registering this remote processor, additional virtio drivers might get
 124      probed.
 125
 126  int rproc_del(struct rproc *rproc)
 127    - Unroll rproc_add().
 128      This function should be called when the platform specific rproc
 129      implementation decides to remove the rproc device. it should
 130      _only_ be called if a previous invocation of rproc_add()
 131      has completed successfully.
 132
 133      After rproc_del() returns, @rproc is still valid, and its
 134      last refcount should be decremented by calling rproc_free().
 135
 136      Returns 0 on success and -EINVAL if @rproc isn't valid.
 137
 138  void rproc_report_crash(struct rproc *rproc, enum rproc_crash_type type)
 139    - Report a crash in a remoteproc
 140      This function must be called every time a crash is detected by the
 141      platform specific rproc implementation. This should not be called from a
 142      non-remoteproc driver. This function can be called from atomic/interrupt
 143      context.
 144
 1455. Implementation callbacks
 146
 147These callbacks should be provided by platform-specific remoteproc
 148drivers:
 149
 150/**
 151 * struct rproc_ops - platform-specific device handlers
 152 * @start:      power on the device and boot it
 153 * @stop:       power off the device
 154 * @kick:       kick a virtqueue (virtqueue id given as a parameter)
 155 */
 156struct rproc_ops {
 157        int (*start)(struct rproc *rproc);
 158        int (*stop)(struct rproc *rproc);
 159        void (*kick)(struct rproc *rproc, int vqid);
 160};
 161
 162Every remoteproc implementation should at least provide the ->start and ->stop
 163handlers. If rpmsg/virtio functionality is also desired, then the ->kick handler
 164should be provided as well.
 165
 166The ->start() handler takes an rproc handle and should then power on the
 167device and boot it (use rproc->priv to access platform-specific private data).
 168The boot address, in case needed, can be found in rproc->bootaddr (remoteproc
 169core puts there the ELF entry point).
 170On success, 0 should be returned, and on failure, an appropriate error code.
 171
 172The ->stop() handler takes an rproc handle and powers the device down.
 173On success, 0 is returned, and on failure, an appropriate error code.
 174
 175The ->kick() handler takes an rproc handle, and an index of a virtqueue
 176where new message was placed in. Implementations should interrupt the remote
 177processor and let it know it has pending messages. Notifying remote processors
 178the exact virtqueue index to look in is optional: it is easy (and not
 179too expensive) to go through the existing virtqueues and look for new buffers
 180in the used rings.
 181
 1826. Binary Firmware Structure
 183
 184At this point remoteproc only supports ELF32 firmware binaries. However,
 185it is quite expected that other platforms/devices which we'd want to
 186support with this framework will be based on different binary formats.
 187
 188When those use cases show up, we will have to decouple the binary format
 189from the framework core, so we can support several binary formats without
 190duplicating common code.
 191
 192When the firmware is parsed, its various segments are loaded to memory
 193according to the specified device address (might be a physical address
 194if the remote processor is accessing memory directly).
 195
 196In addition to the standard ELF segments, most remote processors would
 197also include a special section which we call "the resource table".
 198
 199The resource table contains system resources that the remote processor
 200requires before it should be powered on, such as allocation of physically
 201contiguous memory, or iommu mapping of certain on-chip peripherals.
 202Remotecore will only power up the device after all the resource table's
 203requirement are met.
 204
 205In addition to system resources, the resource table may also contain
 206resource entries that publish the existence of supported features
 207or configurations by the remote processor, such as trace buffers and
 208supported virtio devices (and their configurations).
 209
 210The resource table begins with this header:
 211
 212/**
 213 * struct resource_table - firmware resource table header
 214 * @ver: version number
 215 * @num: number of resource entries
 216 * @reserved: reserved (must be zero)
 217 * @offset: array of offsets pointing at the various resource entries
 218 *
 219 * The header of the resource table, as expressed by this structure,
 220 * contains a version number (should we need to change this format in the
 221 * future), the number of available resource entries, and their offsets
 222 * in the table.
 223 */
 224struct resource_table {
 225        u32 ver;
 226        u32 num;
 227        u32 reserved[2];
 228        u32 offset[0];
 229} __packed;
 230
 231Immediately following this header are the resource entries themselves,
 232each of which begins with the following resource entry header:
 233
 234/**
 235 * struct fw_rsc_hdr - firmware resource entry header
 236 * @type: resource type
 237 * @data: resource data
 238 *
 239 * Every resource entry begins with a 'struct fw_rsc_hdr' header providing
 240 * its @type. The content of the entry itself will immediately follow
 241 * this header, and it should be parsed according to the resource type.
 242 */
 243struct fw_rsc_hdr {
 244        u32 type;
 245        u8 data[0];
 246} __packed;
 247
 248Some resources entries are mere announcements, where the host is informed
 249of specific remoteproc configuration. Other entries require the host to
 250do something (e.g. allocate a system resource). Sometimes a negotiation
 251is expected, where the firmware requests a resource, and once allocated,
 252the host should provide back its details (e.g. address of an allocated
 253memory region).
 254
 255Here are the various resource types that are currently supported:
 256
 257/**
 258 * enum fw_resource_type - types of resource entries
 259 *
 260 * @RSC_CARVEOUT:   request for allocation of a physically contiguous
 261 *                  memory region.
 262 * @RSC_DEVMEM:     request to iommu_map a memory-based peripheral.
 263 * @RSC_TRACE:      announces the availability of a trace buffer into which
 264 *                  the remote processor will be writing logs.
 265 * @RSC_VDEV:       declare support for a virtio device, and serve as its
 266 *                  virtio header.
 267 * @RSC_LAST:       just keep this one at the end
 268 *
 269 * Please note that these values are used as indices to the rproc_handle_rsc
 270 * lookup table, so please keep them sane. Moreover, @RSC_LAST is used to
 271 * check the validity of an index before the lookup table is accessed, so
 272 * please update it as needed.
 273 */
 274enum fw_resource_type {
 275        RSC_CARVEOUT    = 0,
 276        RSC_DEVMEM      = 1,
 277        RSC_TRACE       = 2,
 278        RSC_VDEV        = 3,
 279        RSC_LAST        = 4,
 280};
 281
 282For more details regarding a specific resource type, please see its
 283dedicated structure in include/linux/remoteproc.h.
 284
 285We also expect that platform-specific resource entries will show up
 286at some point. When that happens, we could easily add a new RSC_PLATFORM
 287type, and hand those resources to the platform-specific rproc driver to handle.
 288
 2897. Virtio and remoteproc
 290
 291The firmware should provide remoteproc information about virtio devices
 292that it supports, and their configurations: a RSC_VDEV resource entry
 293should specify the virtio device id (as in virtio_ids.h), virtio features,
 294virtio config space, vrings information, etc.
 295
 296When a new remote processor is registered, the remoteproc framework
 297will look for its resource table and will register the virtio devices
 298it supports. A firmware may support any number of virtio devices, and
 299of any type (a single remote processor can also easily support several
 300rpmsg virtio devices this way, if desired).
 301
 302Of course, RSC_VDEV resource entries are only good enough for static
 303allocation of virtio devices. Dynamic allocations will also be made possible
 304using the rpmsg bus (similar to how we already do dynamic allocations of
 305rpmsg channels; read more about it in rpmsg.txt).
 306