1============ 2Introduction 3============ 4 5The RapidIO standard is a packet-based fabric interconnect standard designed for 6use in embedded systems. Development of the RapidIO standard is directed by the 7RapidIO Trade Association (RTA). The current version of the RapidIO specification 8is publicly available for download from the RTA web-site [1]. 9 10This document describes the basics of the Linux RapidIO subsystem and provides 11information on its major components. 12 131 Overview 14========== 15 16Because the RapidIO subsystem follows the Linux device model it is integrated 17into the kernel similarly to other buses by defining RapidIO-specific device and 18bus types and registering them within the device model. 19 20The Linux RapidIO subsystem is architecture independent and therefore defines 21architecture-specific interfaces that provide support for common RapidIO 22subsystem operations. 23 242. Core Components 25================== 26 27A typical RapidIO network is a combination of endpoints and switches. 28Each of these components is represented in the subsystem by an associated data 29structure. The core logical components of the RapidIO subsystem are defined 30in include/linux/rio.h file. 31 322.1 Master Port 33--------------- 34 35A master port (or mport) is a RapidIO interface controller that is local to the 36processor executing the Linux code. A master port generates and receives RapidIO 37packets (transactions). In the RapidIO subsystem each master port is represented 38by a rio_mport data structure. This structure contains master port specific 39resources such as mailboxes and doorbells. The rio_mport also includes a unique 40host device ID that is valid when a master port is configured as an enumerating 41host. 42 43RapidIO master ports are serviced by subsystem specific mport device drivers 44that provide functionality defined for this subsystem. To provide a hardware 45independent interface for RapidIO subsystem operations, rio_mport structure 46includes rio_ops data structure which contains pointers to hardware specific 47implementations of RapidIO functions. 48 492.2 Device 50---------- 51 52A RapidIO device is any endpoint (other than mport) or switch in the network. 53All devices are presented in the RapidIO subsystem by corresponding rio_dev data 54structure. Devices form one global device list and per-network device lists 55(depending on number of available mports and networks). 56 572.3 Switch 58---------- 59 60A RapidIO switch is a special class of device that routes packets between its 61ports towards their final destination. The packet destination port within a 62switch is defined by an internal routing table. A switch is presented in the 63RapidIO subsystem by rio_dev data structure expanded by additional rio_switch 64data structure, which contains switch specific information such as copy of the 65routing table and pointers to switch specific functions. 66 67The RapidIO subsystem defines the format and initialization method for subsystem 68specific switch drivers that are designed to provide hardware-specific 69implementation of common switch management routines. 70 712.4 Network 72----------- 73 74A RapidIO network is a combination of interconnected endpoint and switch devices. 75Each RapidIO network known to the system is represented by corresponding rio_net 76data structure. This structure includes lists of all devices and local master 77ports that form the same network. It also contains a pointer to the default 78master port that is used to communicate with devices within the network. 79 802.5 Device Drivers 81------------------ 82 83RapidIO device-specific drivers follow Linux Kernel Driver Model and are 84intended to support specific RapidIO devices attached to the RapidIO network. 85 862.6 Subsystem Interfaces 87------------------------ 88 89RapidIO interconnect specification defines features that may be used to provide 90one or more common service layers for all participating RapidIO devices. These 91common services may act separately from device-specific drivers or be used by 92device-specific drivers. Example of such service provider is the RIONET driver 93which implements Ethernet-over-RapidIO interface. Because only one driver can be 94registered for a device, all common RapidIO services have to be registered as 95subsystem interfaces. This allows to have multiple common services attached to 96the same device without blocking attachment of a device-specific driver. 97 983. Subsystem Initialization 99=========================== 100 101In order to initialize the RapidIO subsystem, a platform must initialize and 102register at least one master port within the RapidIO network. To register mport 103within the subsystem controller driver's initialization code calls function 104rio_register_mport() for each available master port. 105 106After all active master ports are registered with a RapidIO subsystem, 107an enumeration and/or discovery routine may be called automatically or 108by user-space command. 109 110RapidIO subsystem can be configured to be built as a statically linked or 111modular component of the kernel (see details below). 112 1134. Enumeration and Discovery 114============================ 115 1164.1 Overview 117------------ 118 119RapidIO subsystem configuration options allow users to build enumeration and 120discovery methods as statically linked components or loadable modules. 121An enumeration/discovery method implementation and available input parameters 122define how any given method can be attached to available RapidIO mports: 123simply to all available mports OR individually to the specified mport device. 124 125Depending on selected enumeration/discovery build configuration, there are 126several methods to initiate an enumeration and/or discovery process: 127 128 (a) Statically linked enumeration and discovery process can be started 129 automatically during kernel initialization time using corresponding module 130 parameters. This was the original method used since introduction of RapidIO 131 subsystem. Now this method relies on enumerator module parameter which is 132 'rio-scan.scan' for existing basic enumeration/discovery method. 133 When automatic start of enumeration/discovery is used a user has to ensure 134 that all discovering endpoints are started before the enumerating endpoint 135 and are waiting for enumeration to be completed. 136 Configuration option CONFIG_RAPIDIO_DISC_TIMEOUT defines time that discovering 137 endpoint waits for enumeration to be completed. If the specified timeout 138 expires the discovery process is terminated without obtaining RapidIO network 139 information. NOTE: a timed out discovery process may be restarted later using 140 a user-space command as it is described below (if the given endpoint was 141 enumerated successfully). 142 143 (b) Statically linked enumeration and discovery process can be started by 144 a command from user space. This initiation method provides more flexibility 145 for a system startup compared to the option (a) above. After all participating 146 endpoints have been successfully booted, an enumeration process shall be 147 started first by issuing a user-space command, after an enumeration is 148 completed a discovery process can be started on all remaining endpoints. 149 150 (c) Modular enumeration and discovery process can be started by a command from 151 user space. After an enumeration/discovery module is loaded, a network scan 152 process can be started by issuing a user-space command. 153 Similar to the option (b) above, an enumerator has to be started first. 154 155 (d) Modular enumeration and discovery process can be started by a module 156 initialization routine. In this case an enumerating module shall be loaded 157 first. 158 159When a network scan process is started it calls an enumeration or discovery 160routine depending on the configured role of a master port: host or agent. 161 162Enumeration is performed by a master port if it is configured as a host port by 163assigning a host destination ID greater than or equal to zero. The host 164destination ID can be assigned to a master port using various methods depending 165on RapidIO subsystem build configuration: 166 167 (a) For a statically linked RapidIO subsystem core use command line parameter 168 "rapidio.hdid=" with a list of destination ID assignments in order of mport 169 device registration. For example, in a system with two RapidIO controllers 170 the command line parameter "rapidio.hdid=-1,7" will result in assignment of 171 the host destination ID=7 to the second RapidIO controller, while the first 172 one will be assigned destination ID=-1. 173 174 (b) If the RapidIO subsystem core is built as a loadable module, in addition 175 to the method shown above, the host destination ID(s) can be specified using 176 traditional methods of passing module parameter "hdid=" during its loading: 177 178 - from command line: "modprobe rapidio hdid=-1,7", or 179 - from modprobe configuration file using configuration command "options", 180 like in this example: "options rapidio hdid=-1,7". An example of modprobe 181 configuration file is provided in the section below. 182 183NOTES: 184 (i) if "hdid=" parameter is omitted all available mport will be assigned 185 destination ID = -1; 186 187 (ii) the "hdid=" parameter in systems with multiple mports can have 188 destination ID assignments omitted from the end of list (default = -1). 189 190If the host device ID for a specific master port is set to -1, the discovery 191process will be performed for it. 192 193The enumeration and discovery routines use RapidIO maintenance transactions 194to access the configuration space of devices. 195 196NOTE: If RapidIO switch-specific device drivers are built as loadable modules 197they must be loaded before enumeration/discovery process starts. 198This requirement is cased by the fact that enumeration/discovery methods invoke 199vendor-specific callbacks on early stages. 200 2014.2 Automatic Start of Enumeration and Discovery 202------------------------------------------------ 203 204Automatic enumeration/discovery start method is applicable only to built-in 205enumeration/discovery RapidIO configuration selection. To enable automatic 206enumeration/discovery start by existing basic enumerator method set use boot 207command line parameter "rio-scan.scan=1". 208 209This configuration requires synchronized start of all RapidIO endpoints that 210form a network which will be enumerated/discovered. Discovering endpoints have 211to be started before an enumeration starts to ensure that all RapidIO 212controllers have been initialized and are ready to be discovered. Configuration 213parameter CONFIG_RAPIDIO_DISC_TIMEOUT defines time (in seconds) which 214a discovering endpoint will wait for enumeration to be completed. 215 216When automatic enumeration/discovery start is selected, basic method's 217initialization routine calls rio_init_mports() to perform enumeration or 218discovery for all known mport devices. 219 220Depending on RapidIO network size and configuration this automatic 221enumeration/discovery start method may be difficult to use due to the 222requirement for synchronized start of all endpoints. 223 2244.3 User-space Start of Enumeration and Discovery 225------------------------------------------------- 226 227User-space start of enumeration and discovery can be used with built-in and 228modular build configurations. For user-space controlled start RapidIO subsystem 229creates the sysfs write-only attribute file '/sys/bus/rapidio/scan'. To initiate 230an enumeration or discovery process on specific mport device, a user needs to 231write mport_ID (not RapidIO destination ID) into that file. The mport_ID is a 232sequential number (0 ... RIO_MAX_MPORTS) assigned during mport device 233registration. For example for machine with single RapidIO controller, mport_ID 234for that controller always will be 0. 235 236To initiate RapidIO enumeration/discovery on all available mports a user may 237write '-1' (or RIO_MPORT_ANY) into the scan attribute file. 238 2394.4 Basic Enumeration Method 240---------------------------- 241 242This is an original enumeration/discovery method which is available since 243first release of RapidIO subsystem code. The enumeration process is 244implemented according to the enumeration algorithm outlined in the RapidIO 245Interconnect Specification: Annex I [1]. 246 247This method can be configured as statically linked or loadable module. 248The method's single parameter "scan" allows to trigger the enumeration/discovery 249process from module initialization routine. 250 251This enumeration/discovery method can be started only once and does not support 252unloading if it is built as a module. 253 254The enumeration process traverses the network using a recursive depth-first 255algorithm. When a new device is found, the enumerator takes ownership of that 256device by writing into the Host Device ID Lock CSR. It does this to ensure that 257the enumerator has exclusive right to enumerate the device. If device ownership 258is successfully acquired, the enumerator allocates a new rio_dev structure and 259initializes it according to device capabilities. 260 261If the device is an endpoint, a unique device ID is assigned to it and its value 262is written into the device's Base Device ID CSR. 263 264If the device is a switch, the enumerator allocates an additional rio_switch 265structure to store switch specific information. Then the switch's vendor ID and 266device ID are queried against a table of known RapidIO switches. Each switch 267table entry contains a pointer to a switch-specific initialization routine that 268initializes pointers to the rest of switch specific operations, and performs 269hardware initialization if necessary. A RapidIO switch does not have a unique 270device ID; it relies on hopcount and routing for device ID of an attached 271endpoint if access to its configuration registers is required. If a switch (or 272chain of switches) does not have any endpoint (except enumerator) attached to 273it, a fake device ID will be assigned to configure a route to that switch. 274In the case of a chain of switches without endpoint, one fake device ID is used 275to configure a route through the entire chain and switches are differentiated by 276their hopcount value. 277 278For both endpoints and switches the enumerator writes a unique component tag 279into device's Component Tag CSR. That unique value is used by the error 280management notification mechanism to identify a device that is reporting an 281error management event. 282 283Enumeration beyond a switch is completed by iterating over each active egress 284port of that switch. For each active link, a route to a default device ID 285(0xFF for 8-bit systems and 0xFFFF for 16-bit systems) is temporarily written 286into the routing table. The algorithm recurs by calling itself with hopcount + 1 287and the default device ID in order to access the device on the active port. 288 289After the host has completed enumeration of the entire network it releases 290devices by clearing device ID locks (calls rio_clear_locks()). For each endpoint 291in the system, it sets the Discovered bit in the Port General Control CSR 292to indicate that enumeration is completed and agents are allowed to execute 293passive discovery of the network. 294 295The discovery process is performed by agents and is similar to the enumeration 296process that is described above. However, the discovery process is performed 297without changes to the existing routing because agents only gather information 298about RapidIO network structure and are building an internal map of discovered 299devices. This way each Linux-based component of the RapidIO subsystem has 300a complete view of the network. The discovery process can be performed 301simultaneously by several agents. After initializing its RapidIO master port 302each agent waits for enumeration completion by the host for the configured wait 303time period. If this wait time period expires before enumeration is completed, 304an agent skips RapidIO discovery and continues with remaining kernel 305initialization. 306 3074.5 Adding New Enumeration/Discovery Method 308------------------------------------------- 309 310RapidIO subsystem code organization allows addition of new enumeration/discovery 311methods as new configuration options without significant impact to the core 312RapidIO code. 313 314A new enumeration/discovery method has to be attached to one or more mport 315devices before an enumeration/discovery process can be started. Normally, 316method's module initialization routine calls rio_register_scan() to attach 317an enumerator to a specified mport device (or devices). The basic enumerator 318implementation demonstrates this process. 319 3204.6 Using Loadable RapidIO Switch Drivers 321----------------------------------------- 322 323In the case when RapidIO switch drivers are built as loadable modules a user 324must ensure that they are loaded before the enumeration/discovery starts. 325This process can be automated by specifying pre- or post- dependencies in the 326RapidIO-specific modprobe configuration file as shown in the example below. 327 328File /etc/modprobe.d/rapidio.conf:: 329 330 # Configure RapidIO subsystem modules 331 332 # Set enumerator host destination ID (overrides kernel command line option) 333 options rapidio hdid=-1,2 334 335 # Load RapidIO switch drivers immediately after rapidio core module was loaded 336 softdep rapidio post: idt_gen2 idtcps tsi57x 337 338 # OR : 339 340 # Load RapidIO switch drivers just before rio-scan enumerator module is loaded 341 softdep rio-scan pre: idt_gen2 idtcps tsi57x 342 343 -------------------------- 344 345NOTE: 346 In the example above, one of "softdep" commands must be removed or 347 commented out to keep required module loading sequence. 348 3495. References 350============= 351 352[1] RapidIO Trade Association. RapidIO Interconnect Specifications. 353 http://www.rapidio.org. 354 355[2] Rapidio TA. Technology Comparisons. 356 http://www.rapidio.org/education/technology_comparisons/ 357 358[3] RapidIO support for Linux. 359 http://lwn.net/Articles/139118/ 360 361[4] Matt Porter. RapidIO for Linux. Ottawa Linux Symposium, 2005 362 http://www.kernel.org/doc/ols/2005/ols2005v2-pages-43-56.pdf 363