diff options
author | Anders Roxell <anders.roxell@linaro.org> | 2014-07-25 09:46:56 +0200 |
---|---|---|
committer | Maxim Uvarov <maxim.uvarov@linaro.org> | 2014-07-29 22:58:06 +0400 |
commit | b5278f60aaade8dd60bce612be278c9e0876b948 (patch) | |
tree | a1be7a0a91aab540815526e67697ccaefb506801 /doc | |
parent | 8f4f080625e2db810da3f13770ba418c72b08ede (diff) |
doxygen reorganisation
1. To make the structure better
2. To have a Users Guide
3. Remove out of date content
Signed-off-by: Anders Roxell <anders.roxell@linaro.org>
Tested-by: Mike Holmes <mike.holmes@linaro.org>
Diffstat (limited to 'doc')
-rw-r--r-- | doc/api/api.dox | 11 | ||||
-rw-r--r-- | doc/doxygen.cfg | 32 | ||||
-rw-r--r-- | doc/odp.dox | 184 | ||||
-rw-r--r-- | doc/users-guide/guide.dox | 18 |
4 files changed, 61 insertions, 184 deletions
diff --git a/doc/api/api.dox b/doc/api/api.dox new file mode 100644 index 000000000..752227057 --- /dev/null +++ b/doc/api/api.dox @@ -0,0 +1,11 @@ +/* Copyright (c) 2014, Linaro Limited + * All rights reserved + * + * SPDX-License-Identifier: BSD-3-Clause + */ + +/** + * + * @page api Application Programming Interface (API) Guide + * + */ diff --git a/doc/doxygen.cfg b/doc/doxygen.cfg new file mode 100644 index 000000000..b3a033d10 --- /dev/null +++ b/doc/doxygen.cfg @@ -0,0 +1,32 @@ +PROJECT_NAME = "API Reference Manual" +PROJECT_LOGO = $(SRCDIR)/doc/images/ODP-Logo-HQ.png +QUIET = YES +OUTPUT_DIRECTORY = $(DOCDIR) +FULL_PATH_NAMES = NO +JAVADOC_AUTOBRIEF = YES +OPTIMIZE_OUTPUT_FOR_C = YES +TYPEDEF_HIDES_STRUCT = YES +EXTRACT_STATIC = YES +SORT_MEMBER_DOCS = NO +WARN_NO_PARAMDOC = YES +INPUT = $(SRCDIR)/doc $(SRCDIR)/doc/users-guide $(SRCDIR)/doc/api $(SRCDIR)/include $(SRCDIR)/example +FILE_PATTERNS = odp*.h odp*.c *.dox +RECURSIVE = YES +SOURCE_BROWSER = YES +REFERENCED_BY_RELATION = YES +REFERENCES_RELATION = YES +ALPHABETICAL_INDEX = NO +QHP_NAMESPACE = +GENERATE_TREEVIEW = YES +PAPER_TYPE = a4wide +CLASS_DIAGRAMS = NO +HAVE_DOT = YES +CALL_GRAPH = YES +DOT_MULTI_TARGETS = NO +EXAMPLE_PATH = $(SRCDIR)/example $(SRCDIR)/platform +EXAMPLE_PATTERNS = *.c +EXAMPLE_RECURSIVE = YES +IMAGE_PATH = $(SRCDIR)/doc/images +HTML_EXTRA_STYLESHEET = $(SRCDIR)/doc/odpdoxygen.css +PREDEFINED = __GNUC__ +INTERNAL_DOCS = YES diff --git a/doc/odp.dox b/doc/odp.dox index 328c7bb71..73d1f371b 100644 --- a/doc/odp.dox +++ b/doc/odp.dox @@ -68,188 +68,4 @@ * - The git repo is https://git.linaro.org/lng/odp.git * - Bug tracking https://launchpad.net/linaro-odp * - * - * @section sec_2 User guide - * - * @subsection sub2_1 The ODP API - * - * This file (odp.h) is the main ODP API file. User should include only this - * file to keep portability since structure and naming of sub header files - * may be change between implementations. - * - * @subsection sub2_2 Threading - * - * ODP does not specify a threading model. Applications can use - * processes or pthreads, or Roll-Your-Own (RYO) threading/fibre - * mechanisms for multi-threading as needed. Creation and control of - * threads is the responsibility of the ODP application. For optimal - * performance on many-core SoCs, it is recommended that threads be - * run on dedicated cores. ODP provides high-level APIs for core - * enumeration and assignment while the corresponding ODP - * implementation layer provides the appropriate mechanisms to realize - * these functions. - * - * Threads used for ODP processing should be pinned into separate cores. - * Commonly these threads process packets in a run-to-completion loop. - * Application should avoid blocking threads used for ODP processing, - * since it may cause blocking on other threads/cores. - * - * @subsection sub2_3 ODP initialisation - * - * Before calling any other ODP API functions, ODP library must be - * initialised by calling odp_init_global() once and odp_init_local() - * on each of the cores sharing the same ODP environment (instance). - * - * @subsection sub2_4 API Categories - * - * APIs provided by ODP cover the following areas: - * - * @subsubsection memory_management Memory Management - * - * This includes macros and other APIs to control memory alignments - * of data structures as well as allocation/deallocation services - * for ODP-managed objects. Note that ODP does not wrapper malloc() - * or similar platform specific APIs for the sake of wrappering. - * - * @subsubsection buffer_management Buffer Management - * - * This includes APIs for defining and managing buffer pools used - * for packets and other bulk purposes. Note that the allocation - * and release of buffers from buffer pools is not something done - * explicitly by ODP applications, but rather by APIs that use these - * buffers. This is because in most SoCs, actual buffer allocation - * and release is accelerated and performed by hardware. Software's - * role in buffer management is normally reserved to allocating - * large chunks of memory which are then given to hardware for - * automatic management as pools of buffers. In this way the ODP - * application operates independent of how buffers are managed by - * the underlying ODP implementation. - * - * @subsubsection packet_management Packet Management - * - * This includes APIs and accessor functions for packet descriptors - * as well as packet receipt and transmission. - * - * @subsubsection syncronisation Synchronization - * - * This includes APIs and related functions for synchronization - * involving other ODP APIs, such as barriers and related atomics. - * Again, as ODP does not specify a threading model applications - * make use whatever synchronization primitives are native to the - * model they use. - * - * @subsubsection core_enumeration Core Enumeration and managment - * - * This includes APIs to allow applications to enumerate and - * reference cores and per-core data structures. - * - * @subsection sub2_5 Miscellaneous Facilities - * - * ODP includes miscellaneous facilities for compiler hints and - * optimizations common in GCC. [Not sure if we want to consider - * these an "API" per se]. - * - * @subsection sub2_6 Application Programming Model - * - * ODP supports applications that execute using a "run to completion" - * programming model. This means that once dispatched, application - * threads are not interrupted by the kernel or other scheduling - * entity. - * - * Application threads receive work requests as \a events that are - * delivered on application and/or implementation defined - * \a queues. ODP application code would thus normally be - * structured as follows: - * - * @code - * #include <odp.h> - * ...other needed #includes - * - * int main (int argc, char *argv[]) - * { - * ...application-specific initialization - * odp_init_global(); - * - * ...launch threads - * ...wait for threads to terminate - * } - * - * void worker_thread (parameters) - * { - * odp_init_local(); - * - * while (1) { - * do_work(get_work()); // Replace with ODP calls when defined - * } - * - * } - * @endcode - * - * Events are receved on input queues and are processed until they are - * placed on an output queue of some sort. The thread then gets the - * next event to be processed from an input queue and repeats the - * process. - * - * @subsection sub3_1 Asynchronous Operations - * - * Note that work to be performed by a thread may require access to an - * asynchronous function that takes a significant amount of time to - * complete. In such cases the event is forwarded to another worker - * thread or hardware accelerator, depending on the implementation, by - * placing it on anothert queue, which is an output queue of the - * thread making the request. This event in turn is received and - * processed by the thread/accelerator that handles it via its input - * queue. When this aysynchronous event is complete, the event is - * placed on the handler's output queue, which feeds back to the - * original requestor's input queue. When the requesting thread next - * receives this event it resumes processing of the event following - * the asynchronous event and works on it either until it is ready for - * final disposition, or until another asynchronous operation is - * required to process the event. - * - * @subsection sub3_2 Queue Linkages - * - * The mapping of input and output queues that connect worker threads - * to accelerators and related offload functions is a cooperation - * between the implementation and the ODP application. The - * implementation defines the service funtions that are available to - * worker threads (e.g., cypto offload services) and as part of that - * definition defines the queue structure that connects requests to - * those services as well as the outputs from those services that - * connect back to the requesting workers. The ODP application, in - * turn, defines the number of worker threads and how they cooperate - * among themselves. Note that the application may use ODP core - * enumeration APIs to decide how many such worker threads should be - * deployed. - * - * @subsection sub3_3 Packet I/O - * - * In ODP packet I/O is implicit by reading from and writing to queues - * associated with interfaces. An ODP application receives packets by - * dequeuing an event from an input queue associated with an I/O - * interface. This either triggers a packet read or (more likely) - * simply provides the next (queued) packet from the associated - * interface. The actual mechanism used to effect the receipt of the - * packet is left to the ODP implementation and may involve any - * combination of sofware and/or hardware operations. - * - * Similarly, packet transmission is performed by writing a packet to - * an output queue associated with an I/O interface. Again, this - * schedules the packet for output using some combination of software - * and/or hardware as determined by the implementation. ODP applications - * themselves, therefore, are freed from the details of how packet I/O - * is performed or buffered to minimize latencies. The latter is the - * concern of the ODP implementation to achieve optimal results for - * the platform supporting the implementation. - * - * @subsection How to Use this Reference - * - * This reference provides an overview of each data structure and API - * function, along with a graphical representation of the various - * structural dependencies among them. When using the HTML version of - * this reference, all links are dynamic and provide access to the - * underlying implementation source files as well, thus providing both - * a ready reference to API parameters and syntax, as well as - * convenient access to the actual implementation behind them to - * further programmer understandng. */ diff --git a/doc/users-guide/guide.dox b/doc/users-guide/guide.dox new file mode 100644 index 000000000..314d2952f --- /dev/null +++ b/doc/users-guide/guide.dox @@ -0,0 +1,18 @@ +/* Copyright (c) 2014, Linaro Limited + * All rights reserved + * + * SPDX-License-Identifier: BSD-3-Clause + */ + +/** + * + * @page users_guide Users Guide + * + * @section sec_gene Linux Generic + * @verbinclude linux-generic/README + * @section sec_dpdk Linux DPDK + * @verbinclude linux-dpdk/README + * @section sec_keys Linux Keystone2 + * @verbinclude linux-keystone2/README + * + */ |