Commit | Line | Data |
72aaf631 |
1 | Building |
2 | |
d81a1b93 |
3 | If you want to build with multi-threading support and you are |
69ce17de |
4 | running one of the following: |
5 | * Linux 2.x (with the LinuxThreads library installed: that's |
6 | the linuxthreads and linuxthreads-devel RPMs for RedHat) |
7 | |
8 | * Digital UNIX 4.x |
9 | |
10 | * Solaris 2.x for recentish x (2.5 is OK) |
11 | |
12 | * IRIX 6.2 or newer. 6.2 will require a few os patches. |
13 | IMPORTANT: Without patch 2401, a kernel bug in IRIX 6.2 will |
14 | cause your machine to panic and crash when running threaded perl. |
15 | IRIX 6.3 and up should be OK. See lower down for patch details. |
16 | |
d81a1b93 |
17 | then you should be able to use |
18 | ./Configure -Dusethreads -Doptimize=-g -ders |
19 | make |
20 | and ignore the rest of this "Building" section. If it doesn't |
21 | work or you are using another platform which you believe supports |
22 | POSIX.1c threads then read on. |
23 | |
72aaf631 |
24 | Omit the -e from your ./Configure arguments. For example, use |
25 | ./Configure -drs |
26 | When it offers to let you change config.sh, do so. If you already |
27 | have a config.sh then you can edit it and do |
28 | ./Configure -S |
29 | to propagate the required changes. |
30 | In ccflags, insert -DUSE_THREADS (and probably -DDEBUGGING since |
31 | that's what I've been building with). Also insert any other |
32 | arguments in there that your compiler needs to use POSIX threads. |
33 | Change optimize to -g to give you better debugging information. |
34 | Include any necessary explicit libraries in libs and change |
35 | ldflags if you need any linker flags instead or as well. |
36 | |
37 | More explicitly, for Linux (when using the standard kernel-threads |
38 | based LinuxThreads library): |
39 | Add -DUSE_THREADS -D_REENTRANT -DDEBUGGING to ccflags and cppflags |
40 | Add -lpthread to libs |
41 | Change optimize to -g |
42 | For Digital Unix 4.x: |
43 | Add -pthread -DUSE_THREADS -DDEBUGGING to ccflags |
44 | Add -DUSE_THREADS -DDEBUGGING to cppflags |
45 | Add -pthread to ldflags |
46 | Change optimize to -g |
d81a1b93 |
47 | Add -lpthread -lc_r to lddlflags |
72aaf631 |
48 | For some reason, the extra includes for pthreads make Digital UNIX |
49 | complain fatally about the sbrk() delcaration in perl's malloc.c |
50 | so use the native malloc as follows: |
51 | Change usemymalloc to n |
52 | Zap mallocobj and mallocsrc (foo='') |
53 | Change d_mymalloc to undef |
d81a1b93 |
54 | For Solaris, do the same as for Linux above. |
eb1cfdd6 |
55 | For IRIX: |
56 | Add -DUSE_THREADS -DDEBUGGING to ccflags |
57 | Add -DUSE_THREADS -DDEBUGGING to cppflags |
58 | Add -lpthread to libs |
59 | Change optimize to -g |
60 | For IRIX 6.2, you have to have the following patches installed: |
61 | 1404 Irix 6.2 Posix 1003.1b man pages |
62 | 1645 IRIX 6.2 & 6.3 POSIX header file updates |
63 | 2000 Irix 6.2 Posix 1003.1b support modules |
64 | 2254 Pthread library fixes |
69ce17de |
65 | 2401 6.2 all platform kernel rollup |
66 | IMPORTANT: Without patch 2401, a kernel bug in IRIX 6.2 will |
67 | cause your machine to panic and crash when running threaded perl. |
68 | IRIX 6.3 and up should be OK. |
69 | |
eb1cfdd6 |
70 | For IRIX 6.3 and 6.4 the pthreads should work out of the box. |
71 | Thanks to Hannu Napari <Hannu.Napari@hut.fi> for the IRIX |
72 | pthreads patches information. |
72aaf631 |
73 | |
74 | Now you can do a |
d81a1b93 |
75 | make |
76 | |
72aaf631 |
77 | |
5756a3ac |
78 | O/S specific bugs |
79 | |
80 | Solaris qsort uses a hidden mutex for synchronisation. If you die() |
81 | while doing a sort() then the resulting longjmp() leaves the mutex |
82 | locked so you get a deadlock the next time you try to sort(). |
83 | |
84 | LinuxThreads 0.5 has a bug which can cause file descriptor 0 to be |
69ce17de |
85 | closed after a fork() leading to many strange symptoms. Version 0.6 |
86 | has this fixed but the following patch can be applied to 0.5 for now: |
5756a3ac |
87 | |
88 | ----------------------------- cut here ----------------------------- |
89 | --- linuxthreads-0.5/pthread.c.ORI Mon Oct 6 13:55:50 1997 |
90 | +++ linuxthreads-0.5/pthread.c Mon Oct 6 13:57:24 1997 |
91 | @@ -312,8 +312,10 @@ |
92 | free(pthread_manager_thread_bos); |
93 | pthread_manager_thread_bos = pthread_manager_thread_tos = NULL; |
94 | /* Close the two ends of the pipe */ |
95 | - close(pthread_manager_request); |
96 | - close(pthread_manager_reader); |
97 | + if (pthread_manager_request >= 0) { |
98 | + close(pthread_manager_request); |
99 | + close(pthread_manager_reader); |
100 | + } |
101 | pthread_manager_request = pthread_manager_reader = -1; |
102 | /* Update the pid of the main thread */ |
103 | self->p_pid = getpid(); |
104 | ----------------------------- cut here ----------------------------- |
105 | |
106 | |
72aaf631 |
107 | Building the Thread extension |
108 | |
5756a3ac |
109 | The Thread extension is now part of the main perl distribution tree. |
110 | If you did Configure -Dusethreads then it will have been added to |
111 | the list of extensions automatically. |
72aaf631 |
112 | |
5756a3ac |
113 | You can try some of the tests with |
114 | cd ext/Thread |
69ce17de |
115 | perl create.t |
116 | perl join.t |
117 | perl lock.t |
118 | perl io.t |
119 | etc. |
72aaf631 |
120 | The io one leaves a thread reading from the keyboard on stdin so |
121 | as the ping messages appear you can type lines and see them echoed. |
122 | |
123 | Try running the main perl test suite too. There are known |
69ce17de |
124 | failures for some of the DBM/DB extensions (if their underlying |
125 | libraries were not compiled to be thread-aware). |
72aaf631 |
126 | |
127 | |
128 | Bugs |
129 | |
72aaf631 |
130 | * FAKE_THREADS should produce a working perl but the Thread |
131 | extension won't build with it yet. |
132 | |
133 | * There's a known memory leak (curstack isn't freed at the end |
134 | of each thread because it causes refcount problems that I |
135 | haven't tracked down yet) and there are very probably others too. |
136 | |
5756a3ac |
137 | * There may still be races where bugs show up under contention. |
72aaf631 |
138 | |
43fe56be |
139 | * Need to document "lock", Thread.pm, Queue.pm, ... |
140 | |
72aaf631 |
141 | |
1304aa9d |
142 | Debugging |
143 | |
144 | Use the -DL command-line option to turn on debugging of the |
145 | multi-threading code. Under Linux, that also turns on a quick |
146 | hack I did to grab a bit of extra information from segfaults. |
147 | If you have a fancier gdb/threads setup than I do then you'll |
148 | have to delete the lines in perl.c which say |
149 | #if defined(DEBUGGING) && defined(USE_THREADS) && defined(__linux__) |
150 | DEBUG_L(signal(SIGSEGV, (void(*)(int))catch_sigsegv);); |
151 | #endif |
152 | |
153 | |
43fe56be |
154 | Background |
155 | |
156 | Some old globals (e.g. stack_sp, op) and some old per-interpreter |
157 | variables (e.g. tmps_stack, cxstack) move into struct thread. |
5756a3ac |
158 | All fields of struct thread which derived from original perl |
159 | variables have names of the form Tfoo. For example, stack_sp becomes |
43fe56be |
160 | the field Tstack_sp of struct thread. For those fields which moved |
161 | from original perl, thread.h does |
162 | #define foo (thr->Tfoo) |
163 | This means that all functions in perl which need to use one of these |
164 | fields need an (automatic) variable thr which points at the current |
165 | thread's struct thread. For pp_foo functions, it is passed around as |
166 | an argument, for other functions they do |
167 | dTHR; |
168 | which declares and initialises thr from thread-specific data |
169 | via pthread_getspecific. If a function fails to compile with an |
170 | error about "no such variable thr", it probably just needs a dTHR |
171 | at the top. |
172 | |
173 | |
174 | Fake threads |
175 | |
176 | For FAKE_THREADS, thr is a global variable and perl schedules threads |
177 | by altering thr in between appropriate ops. The next and prev fields |
178 | of struct thread keep all fake threads on a doubly linked list and |
179 | the next_run and prev_run fields keep all runnable threads on a |
180 | doubly linked list. Mutexes are stubs for FAKE_THREADS. Condition |
181 | variables are implemented as a list of waiting threads. |
182 | |
183 | |
184 | Mutexes and condition variables |
185 | |
186 | The API is via macros MUTEX_{INIT,LOCK,UNLOCK,DESTROY} and |
5756a3ac |
187 | COND_{INIT,WAIT,SIGNAL,BROADCAST,DESTROY}. |
188 | |
189 | A mutex is only required to be a simple, fast mutex (e.g. it does not |
190 | have to be recursive). It is only ever held across very short pieces |
191 | of code. Condition variables are only ever signalled/broadcast while |
192 | their associated mutex is held. (This constraint simplifies the |
193 | implementation of condition variables in certain porting situations.) |
194 | For POSIX threads, perl mutexes and condition variables correspond to |
195 | POSIX ones. For FAKE_THREADS, mutexes are stubs and condition variables |
196 | are implmented as lists of waiting threads. For FAKE_THREADS, a thread |
43fe56be |
197 | waits on a condition variable by removing itself from the runnable |
198 | list, calling SCHEDULE to change thr to the next appropriate |
199 | runnable thread and returning op (i.e. the new threads next op). |
200 | This means that fake threads can only block while in PP code. |
201 | A PP function which contains a COND_WAIT must be prepared to |
202 | handle such restarts and can use the field "private" of struct |
203 | thread to record its state. For fake threads, COND_SIGNAL and |
204 | COND_BROADCAST work by putting back all the threads on the |
205 | condition variables list into the run queue. Note that a mutex |
206 | must *not* be held while returning from a PP function. |
207 | |
c7848ba1 |
208 | Perl locks and condition variables are both implemented as a |
209 | condpair_t structure, containing a mutex, an "owner" condition |
210 | variable, an owner thread field and another condition variable). |
211 | The structure is attached by 'm' magic to any SV. pp_lock locks |
212 | such an object by waiting on the ownercond condition variable until |
213 | the owner field is zero and then setting the owner field to its own |
214 | thread pointer. The lock is semantically recursive so if the owner |
215 | field already matches the current thread then pp_lock returns |
216 | straight away. If the owner field has to be filled in then |
217 | unlock_condpair is queued as an end-of-block destructor and |
218 | that function zeroes out the owner field and signals the ownercond |
219 | condition variable, thus waking up any other thread that wants to |
220 | lock it. When used as a condition variable, the condpair is locked |
221 | (involving the above wait-for-ownership and setting the owner field) |
222 | and the spare condition variable field is used for waiting on. |
223 | |
224 | |
225 | Thread states |
226 | |
227 | |
228 | $t->join |
229 | R_JOINABLE ---------------------> R_JOINED >----\ |
230 | | \ pthread_join(t) | ^ | |
231 | | \ | | join | pthread_join |
232 | | \ | | | |
233 | | \ | \------/ |
234 | | \ | |
235 | | \ | |
236 | | $t->detach\ pthread_detach | |
237 | | _\| | |
238 | ends| R_DETACHED ends | unlink |
239 | | \ | |
240 | | ends \ unlink | |
241 | | \ | |
242 | | \ | |
243 | | \ | |
244 | | \ | |
245 | | \ | |
246 | V join detach _\| V |
247 | ZOMBIE ----------------------------> DEAD |
248 | pthread_join pthread_detach |
249 | and unlink and unlink |
250 | |
43fe56be |
251 | |
252 | |
72aaf631 |
253 | Malcolm Beattie |
254 | mbeattie@sable.ox.ac.uk |
69ce17de |
255 | Last updated: 27 November 1997 |