]>
Commit | Line | Data |
---|---|---|
7c673cae FG |
1 | The collector supports both incremental collection and threads under |
2 | Solaris 2. The incremental collector normally retrieves page dirty information | |
3 | through the appropriate /proc calls. But it can also be configured | |
4 | (by defining MPROTECT_VDB instead of PROC_VDB in gcconfig.h) to use mprotect | |
5 | and signals. This may result in shorter pause times, but it is no longer | |
6 | safe to issue arbitrary system calls that write to the heap. | |
7 | ||
8 | Under other UNIX versions, | |
9 | the collector normally obtains memory through sbrk. There is some reason | |
10 | to expect that this is not safe if the client program also calls the system | |
11 | malloc, or especially realloc. The sbrk man page strongly suggests this is | |
12 | not safe: "Many library routines use malloc() internally, so use brk() | |
13 | and sbrk() only when you know that malloc() definitely will not be used by | |
14 | any library routine." This doesn't make a lot of sense to me, since there | |
15 | seems to be no documentation as to which routines can transitively call malloc. | |
16 | Nonetheless, under Solaris2, the collector now (since 4.12) allocates | |
17 | memory using mmap by default. (It defines USE_MMAP in gcconfig.h.) | |
18 | You may want to reverse this decisions if you use -DREDIRECT_MALLOC=... | |
19 | ||
20 | ||
21 | SOLARIS THREADS: | |
22 | ||
23 | The collector must be compiled with -DGC_SOLARIS_THREADS (thr_ functions) | |
24 | or -DGC_THREADS to be thread safe. This assumes use of the pthread_ | |
25 | interface. Old style Solaris threads are no longer supported. | |
26 | ||
27 | It is also essential that gc.h be included in files that call thr_create, | |
28 | thr_join, thr_suspend, thr_continue, or dlopen. Gc.h macro defines | |
29 | these to also do GC bookkeeping, etc. Gc.h must be included with | |
30 | one or both of these macros defined, otherwise | |
31 | these replacements are not visible. | |
32 | A collector built in this way way only be used by programs that are | |
33 | linked with the threads library. | |
34 | ||
35 | Since 5.0 alpha5, dlopen disables collection temporarily, | |
36 | unless USE_PROC_FOR_LIBRARIES is defined. In some unlikely cases, this | |
37 | can result in unpleasant heap growth. But it seems better than the | |
38 | race/deadlock issues we had before. | |
39 | ||
40 | If solaris_threads are used on an X86 processor with malloc redirected to | |
41 | GC_malloc, it is necessary to call GC_thr_init explicitly before forking the | |
42 | first thread. (This avoids a deadlock arising from calling GC_thr_init | |
43 | with the allocation lock held.) | |
44 | ||
45 | It appears that there is a problem in using gc_cpp.h in conjunction with | |
46 | Solaris threads and Sun's C++ runtime. Apparently the overloaded new operator | |
47 | is invoked by some iostream initialization code before threads are correctly | |
48 | initialized. As a result, call to thr_self() in garbage collector | |
49 | initialization segfaults. Currently the only known workaround is to not | |
50 | invoke the garbage collector from a user defined global operator new, or to | |
51 | have it invoke the garbage-collector's allocators only after main has started. | |
52 | (Note that the latter requires a moderately expensive test in operator | |
53 | delete.) | |
54 | ||
55 | I encountered "symbol <unknown>: offet .... is non-aligned" errors. These | |
56 | appear to be traceable to the use of the GNU assembler with the Sun linker. | |
57 | The former appears to generate a relocation not understood by the latter. | |
58 | The fix appears to be to use a consistent tool chain. (As a non-Solaris-expert | |
59 | my solution involved hacking the libtool script, but I'm sure you can | |
60 | do something less ugly.) | |
61 | ||
62 | Hans-J. Boehm | |
63 | (The above contains my personal opinions, which are probably not shared | |
64 | by anyone else.) |