blob: 425044ebef303e65534c4252fac220858cdda3f7 [file] [log] [blame]
Jeff Thompsona28eed82013-08-22 16:21:10 -07001// (C) Copyright Dustin Spicuzza 2009.
2// Adapted to vxWorks 6.9 by Peter Brockamp 2012.
3// Use, modification and distribution are subject to the
4// Boost Software License, Version 1.0. (See accompanying file
Jeff Thompsonf7d49942013-08-01 16:47:40 -07005// LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
6
7// See http://www.boost.org for most recent version.
8
Jeff Thompsona28eed82013-08-22 16:21:10 -07009// Since WRS does not yet properly support boost under vxWorks
10// and this file was badly outdated, but I was keen on using it,
11// I patched boost myself to make things work. This has been tested
12// and adapted by me for vxWorks 6.9 *only*, as I'm lacking access
13// to earlier 6.X versions! The only thing I know for sure is that
14// very old versions of vxWorks (namely everything below 6.x) are
15// absolutely unable to use boost. This is mainly due to the completely
16// outdated libraries and ancient compiler (GCC 2.96 or worse). Do
17// not even think of getting this to work, a miserable failure will
18// be guaranteed!
19// Equally, this file has been tested for RTPs (Real Time Processes)
20// only, not for DKMs (Downloadable Kernel Modules). These two types
21// of executables differ largely in the available functionality of
22// the C-library, STL, and so on. A DKM uses a library similar to those
23// of vxWorks 5.X - with all its limitations and incompatibilities
24// with respect to ANSI C++ and STL. So probably there might be problems
25// with the usage of boost from DKMs. WRS or any voluteers are free to
26// prove the opposite!
Jeff Thompsonf7d49942013-08-01 16:47:40 -070027
Jeff Thompsona28eed82013-08-22 16:21:10 -070028// ====================================================================
29//
30// Some important information regarding the usage of POSIX semaphores:
31// -------------------------------------------------------------------
32//
33// VxWorks as a real time operating system handles threads somewhat
34// different from what "normal" OSes do, regarding their scheduling!
35// This could lead to a scenario called "priority inversion" when using
36// semaphores, see http://en.wikipedia.org/wiki/Priority_inversion.
37//
38// Now, VxWorks POSIX-semaphores for DKM's default to the usage of
39// priority inverting semaphores, which is fine. On the other hand,
40// for RTP's it defaults to using non priority inverting semaphores,
41// which could easily pose a serious problem for a real time process,
42// i.e. deadlocks! To overcome this two possibilities do exist:
43//
44// a) Patch every piece of boost that uses semaphores to instanciate
45// the proper type of semaphores. This is non-intrusive with respect
46// to the OS and could relatively easy been done by giving all
47// semaphores attributes deviating from the default (for in-depth
48// information see the POSIX functions pthread_mutexattr_init()
49// and pthread_mutexattr_setprotocol()). However this breaks all
50// too easily, as with every new version some boost library could
51// all in a sudden start using semaphores, resurrecting the very
52// same, hard to locate problem over and over again!
53//
54// b) We could change the default properties for POSIX-semaphores
55// that VxWorks uses for RTP's and this is being suggested here,
56// as it will more or less seamlessly integrate with boost. I got
57// the following information from WRS how to do this, compare
58// Wind River TSR# 1209768:
59//
60// Instructions for changing the default properties of POSIX-
61// semaphores for RTP's in VxWorks 6.9:
62// - Edit the file /vxworks-6.9/target/usr/src/posix/pthreadLib.c
63// in the root of your Workbench-installation.
64// - Around line 917 there should be the definition of the default
65// mutex attributes:
66//
67// LOCAL pthread_mutexattr_t defaultMutexAttr =
68// {
69// PTHREAD_INITIALIZED_OBJ, PTHREAD_PRIO_NONE, 0,
70// PTHREAD_MUTEX_DEFAULT
71// };
72//
73// Here, replace PTHREAD_PRIO_NONE by PTHREAD_PRIO_INHERIT.
74// - Around line 1236 there should be a definition for the function
75// pthread_mutexattr_init(). A couple of lines below you should
76// find a block of code like this:
77//
78// pAttr->mutexAttrStatus = PTHREAD_INITIALIZED_OBJ;
79// pAttr->mutexAttrProtocol = PTHREAD_PRIO_NONE;
80// pAttr->mutexAttrPrioceiling = 0;
81// pAttr->mutexAttrType = PTHREAD_MUTEX_DEFAULT;
82//
83// Here again, replace PTHREAD_PRIO_NONE by PTHREAD_PRIO_INHERIT.
84// - Finally, rebuild your VSB. This will create a new VxWorks kernel
85// with the changed properties. That's it! Now, using boost should
86// no longer cause any problems with task deadlocks!
87//
88// And here's another useful piece of information concerning VxWorks'
89// POSIX-functionality in general:
90// VxWorks is not a genuine POSIX-OS in itself, rather it is using a
91// kind of compatibility layer (sort of a wrapper) to emulate the
92// POSIX-functionality by using its own resources and functions.
93// At the time a task (thread) calls it's first POSIX-function during
94// runtime it is being transformed by the OS into a POSIX-thread.
95// This transformation does include a call to malloc() to allocate the
96// memory required for the housekeeping of POSIX-threads. In a high
97// priority RTP this malloc() call may be highly undesirable, as its
98// timing is more or less unpredictable (depending on what your actual
99// heap looks like). You can circumvent this problem by calling the
100// function thread_self() at a well defined point in the code of the
101// task, e.g. shortly after the task spawns up. Thereby you are able
102// to define the time when the task-transformation will take place and
103// you could shift it to an uncritical point where a malloc() call is
104// tolerable. So, if this could pose a problem for your code, remember
105// to call thread_self() from the affected task at an early stage.
106//
107// ====================================================================
Jeff Thompsonf7d49942013-08-01 16:47:40 -0700108
Jeff Thompsona28eed82013-08-22 16:21:10 -0700109// Block out all versions before vxWorks 6.x, as these don't work:
110// Include header with the vxWorks version information and query them
111#include <version.h>
112#if !defined(_WRS_VXWORKS_MAJOR) || (_WRS_VXWORKS_MAJOR < 6)
113# error "The vxWorks version you're using is so badly outdated,\
114 it doesn't work at all with boost, sorry, no chance!"
Jeff Thompsonf7d49942013-08-01 16:47:40 -0700115#endif
116
Jeff Thompsona28eed82013-08-22 16:21:10 -0700117// Handle versions above 5.X but below 6.9
118#if (_WRS_VXWORKS_MAJOR == 6) && (_WRS_VXWORKS_MINOR < 9)
119// TODO: Starting from what version does vxWorks work with boost?
120// We can't reasonably insert a #warning "" as a user hint here,
121// as this will show up with every file including some boost header,
122// badly bugging the user... So for the time being we just leave it.
123#endif
124
125// vxWorks specific config options:
126// --------------------------------
Jeff Thompson3d613fd2013-10-15 15:39:04 -0700127#define NDNBOOST_PLATFORM "vxWorks"
Jeff Thompsona28eed82013-08-22 16:21:10 -0700128
129// Special behaviour for DKMs:
130#ifdef _WRS_KERNEL
131 // DKMs do not have the <cwchar>-header,
132 // but apparently they do have an intrinsic wchar_t meanwhile!
Jeff Thompson3d613fd2013-10-15 15:39:04 -0700133# define NDNBOOST_NO_CWCHAR
Jeff Thompsona28eed82013-08-22 16:21:10 -0700134
135 // Lots of wide-functions and -headers are unavailable for DKMs as well:
Jeff Thompson3d613fd2013-10-15 15:39:04 -0700136# define NDNBOOST_NO_CWCTYPE
137# define NDNBOOST_NO_SWPRINTF
138# define NDNBOOST_NO_STD_WSTRING
139# define NDNBOOST_NO_STD_WSTREAMBUF
Jeff Thompsona28eed82013-08-22 16:21:10 -0700140#endif
141
142// Generally available headers:
Jeff Thompson3d613fd2013-10-15 15:39:04 -0700143#define NDNBOOST_HAS_UNISTD_H
144#define NDNBOOST_HAS_STDINT_H
145#define NDNBOOST_HAS_DIRENT_H
146#define NDNBOOST_HAS_SLIST
Jeff Thompsonf7d49942013-08-01 16:47:40 -0700147
Jeff Thompsona28eed82013-08-22 16:21:10 -0700148// vxWorks does not have installed an iconv-library by default,
149// so unfortunately no Unicode support from scratch is available!
150// Thus, instead it is suggested to switch to ICU, as this seems
151// to be the most complete and portable option...
Jeff Thompson3d613fd2013-10-15 15:39:04 -0700152#define NDNBOOST_LOCALE_WITH_ICU
Jeff Thompsonf7d49942013-08-01 16:47:40 -0700153
Jeff Thompsona28eed82013-08-22 16:21:10 -0700154// Generally available functionality:
Jeff Thompson3d613fd2013-10-15 15:39:04 -0700155#define NDNBOOST_HAS_THREADS
156#define NDNBOOST_HAS_NANOSLEEP
157#define NDNBOOST_HAS_GETTIMEOFDAY
158#define NDNBOOST_HAS_CLOCK_GETTIME
159#define NDNBOOST_HAS_MACRO_USE_FACET
Jeff Thompsona28eed82013-08-22 16:21:10 -0700160
161// Generally unavailable functionality, delivered by boost's test function:
Jeff Thompson3d613fd2013-10-15 15:39:04 -0700162//#define NDNBOOST_NO_DEDUCED_TYPENAME // Commented this out, boost's test gives an errorneous result!
163#define NDNBOOST_NO_CXX11_EXTERN_TEMPLATE
164#define NDNBOOST_NO_CXX11_VARIADIC_MACROS
Jeff Thompsona28eed82013-08-22 16:21:10 -0700165
166// Generally available threading API's:
Jeff Thompson3d613fd2013-10-15 15:39:04 -0700167#define NDNBOOST_HAS_PTHREADS
168#define NDNBOOST_HAS_SCHED_YIELD
169#define NDNBOOST_HAS_SIGACTION
Jeff Thompsona28eed82013-08-22 16:21:10 -0700170
171// Functionality available for RTPs only:
172#ifdef __RTP__
Jeff Thompson3d613fd2013-10-15 15:39:04 -0700173# define NDNBOOST_HAS_PTHREAD_MUTEXATTR_SETTYPE
174# define NDNBOOST_HAS_LOG1P
175# define NDNBOOST_HAS_EXPM1
Jeff Thompsona28eed82013-08-22 16:21:10 -0700176#endif
177
178// Functionality available for DKMs only:
179#ifdef _WRS_KERNEL
180 // Luckily, at the moment there seems to be none!
181#endif
182
183// These #defines allow posix_features to work, since vxWorks doesn't
184// #define them itself for DKMs (for RTPs on the contrary it does):
185#ifdef _WRS_KERNEL
186# ifndef _POSIX_TIMERS
187# define _POSIX_TIMERS 1
188# endif
189# ifndef _POSIX_THREADS
190# define _POSIX_THREADS 1
191# endif
192#endif
193
194// vxWorks doesn't work with asio serial ports:
Jeff Thompson3d613fd2013-10-15 15:39:04 -0700195#define NDNBOOST_ASIO_DISABLE_SERIAL_PORT
Jeff Thompsona28eed82013-08-22 16:21:10 -0700196// TODO: The problem here seems to bee that vxWorks uses its own, very specific
197// ways to handle serial ports, incompatible with POSIX or anything...
198// Maybe a specific implementation would be possible, but until the
199// straight need arises... This implementation would presumably consist
200// of some vxWorks specific ioctl-calls, etc. Any voluteers?
Jeff Thompsonf7d49942013-08-01 16:47:40 -0700201
Jeff Thompsona28eed82013-08-22 16:21:10 -0700202// vxWorks-around: <time.h> #defines CLOCKS_PER_SEC as sysClkRateGet() but
203// miserably fails to #include the required <sysLib.h> to make
204// sysClkRateGet() available! So we manually include it here.
205#ifdef __RTP__
206# include <time.h>
207# include <sysLib.h>
208#endif
209
210// vxWorks-around: In <stdint.h> the macros INT32_C(), UINT32_C(), INT64_C() and
211// UINT64_C() are defined errorneously, yielding not a signed/
212// unsigned long/long long type, but a signed/unsigned int/long
213// type. Eventually this leads to compile errors in ratio_fwd.hpp,
214// when trying to define several constants which do not fit into a
215// long type! We correct them here by redefining.
216#include <cstdint>
217
218// Some macro-magic to do the job
219#define VX_JOIN(X, Y) VX_DO_JOIN(X, Y)
220#define VX_DO_JOIN(X, Y) VX_DO_JOIN2(X, Y)
221#define VX_DO_JOIN2(X, Y) X##Y
222
223// Correctly setup the macros
224#undef INT32_C
225#undef UINT32_C
226#undef INT64_C
227#undef UINT64_C
228#define INT32_C(x) VX_JOIN(x, L)
229#define UINT32_C(x) VX_JOIN(x, UL)
230#define INT64_C(x) VX_JOIN(x, LL)
231#define UINT64_C(x) VX_JOIN(x, ULL)
232
233// #include Libraries required for the following function adaption
234#include <ioLib.h>
235#include <tickLib.h>
236#include <sys/time.h>
237
238// Use C-linkage for the following helper functions
239extern "C" {
240
241// vxWorks-around: The required functions getrlimit() and getrlimit() are missing.
242// But we have the similar functions getprlimit() and setprlimit(),
243// which may serve the purpose.
244// Problem: The vxWorks-documentation regarding these functions
245// doesn't deserve its name! It isn't documented what the first two
246// parameters idtype and id mean, so we must fall back to an educated
247// guess - null, argh... :-/
248
249// TODO: getprlimit() and setprlimit() do exist for RTPs only, for whatever reason.
250// Thus for DKMs there would have to be another implementation.
251#ifdef __RTP__
252 inline int getrlimit(int resource, struct rlimit *rlp){
253 return getprlimit(0, 0, resource, rlp);
254 }
255
256 inline int setrlimit(int resource, const struct rlimit *rlp){
257 return setprlimit(0, 0, resource, const_cast<struct rlimit*>(rlp));
258 }
259#endif
260
261// vxWorks has ftruncate() only, so we do simulate truncate():
262inline int truncate(const char *p, off_t l){
263 int fd = open(p, O_WRONLY);
264 if (fd == -1){
265 errno = EACCES;
266 return -1;
267 }
268 if (ftruncate(fd, l) == -1){
269 close(fd);
270 errno = EACCES;
271 return -1;
272 }
273 return close(fd);
274}
275
276// Fake symlink handling by dummy functions:
277inline int symlink(const char*, const char*){
278 // vxWorks has no symlinks -> always return an error!
279 errno = EACCES;
280 return -1;
281}
282
283inline ssize_t readlink(const char*, char*, size_t){
284 // vxWorks has no symlinks -> always return an error!
285 errno = EACCES;
286 return -1;
287}
288
289// vxWorks claims to implement gettimeofday in sys/time.h
290// but nevertheless does not provide it! See
291// https://support.windriver.com/olsPortal/faces/maintenance/techtipDetail_noHeader.jspx?docId=16442&contentId=WR_TECHTIP_006256
292// We implement a surrogate version here via clock_gettime:
293inline int gettimeofday(struct timeval *tv, void * /*tzv*/) {
294 struct timespec ts;
295 clock_gettime(CLOCK_MONOTONIC, &ts);
296 tv->tv_sec = ts.tv_sec;
297 tv->tv_usec = ts.tv_nsec / 1000;
298 return 0;
299}
300
301// vxWorks does provide neither struct tms nor function times()!
302// We implement an empty dummy-function, simply setting the user
303// and system time to the half of thew actual system ticks-value
304// and the child user and system time to 0.
305// Rather ugly but at least it suppresses compiler errors...
306// Unfortunately, this of course *does* have an severe impact on
307// dependant libraries, actually this is chrono only! Here it will
308// not be possible to correctly use user and system times! But
309// as vxWorks is lacking the ability to calculate user and system
310// process times there seems to be no other possible solution.
311struct tms{
312 clock_t tms_utime; // User CPU time
313 clock_t tms_stime; // System CPU time
314 clock_t tms_cutime; // User CPU time of terminated child processes
315 clock_t tms_cstime; // System CPU time of terminated child processes
316};
317
318inline clock_t times(struct tms *t){
319 struct timespec ts;
320 clock_gettime(CLOCK_THREAD_CPUTIME_ID, &ts);
321 clock_t ticks(static_cast<clock_t>(static_cast<double>(ts.tv_sec) * CLOCKS_PER_SEC +
322 static_cast<double>(ts.tv_nsec) * CLOCKS_PER_SEC / 1000000.0));
323 t->tms_utime = ticks/2U;
324 t->tms_stime = ticks/2U;
325 t->tms_cutime = 0; // vxWorks is lacking the concept of a child process!
326 t->tms_cstime = 0; // -> Set the wait times for childs to 0
327 return ticks;
328}
329
330} // extern "C"
331
332// Put the selfmade functions into the std-namespace, just in case
333namespace std {
334# ifdef __RTP__
335 using ::getrlimit;
336 using ::setrlimit;
337# endif
338 using ::truncate;
339 using ::symlink;
340 using ::readlink;
341 using ::times;
342 using ::gettimeofday;
343}
344
345// Some more macro-magic:
346// vxWorks-around: Some functions are not present or broken in vxWorks
347// but may be patched to life via helper macros...
348
349// Include signal.h which might contain a typo to be corrected here
350#include <signal.h>
351
352#define getpagesize() sysconf(_SC_PAGESIZE) // getpagesize is deprecated anyway!
353#ifndef S_ISSOCK
354# define S_ISSOCK(mode) ((mode & S_IFMT) == S_IFSOCK) // Is file a socket?
355#endif
356#define lstat(p, b) stat(p, b) // lstat() == stat(), as vxWorks has no symlinks!
357#ifndef FPE_FLTINV
358# define FPE_FLTINV (FPE_FLTSUB+1) // vxWorks has no FPE_FLTINV, so define one as a dummy
359#endif
360#if !defined(BUS_ADRALN) && defined(BUS_ADRALNR)
361# define BUS_ADRALN BUS_ADRALNR // Correct a supposed typo in vxWorks' <signal.h>
362#endif
363//typedef int locale_t; // locale_t is a POSIX-extension, currently unpresent in vxWorks!
364
365// #include boilerplate code:
Jeff Thompson2277ce52013-08-01 17:34:11 -0700366#include <ndnboost/config/posix_features.hpp>
Jeff Thompsona28eed82013-08-22 16:21:10 -0700367
368// vxWorks lies about XSI conformance, there is no nl_types.h:
Jeff Thompson3d613fd2013-10-15 15:39:04 -0700369#undef NDNBOOST_HAS_NL_TYPES_H