The collector supports both incremental collection and threads under Solaris 2. The incremental collector normally retrieves page dirty information through the appropriate /proc calls. But it can also be configured (by defining MPROTECT_VDB instead of PROC_VDB in gcconfig.h) to use mprotect and signals. This may result in shorter pause times, but it is no longer safe to issue arbitrary system calls that write to the heap. Under other UNIX versions, the collector normally obtains memory through sbrk. There is some reason to expect that this is not safe if the client program also calls the system malloc, or especially realloc. The sbrk man page strongly suggests this is not safe: "Many library routines use malloc() internally, so use brk() and sbrk() only when you know that malloc() definitely will not be used by any library routine." This doesn't make a lot of sense to me, since there seems to be no documentation as to which routines can transitively call malloc. Nonetheless, under Solaris2, the collector now allocates memory using mmap by default. (It defines USE_MMAP in gcconfig.h.) You may want to reverse this decisions if you use -DREDIRECT_MALLOC=... Note: Before you run "make check", you need to set your LD_LIBRARY_PATH correctly (e.g., to "/usr/local/lib") so that tests can find the shared library libgcc_s.so.1. Alternatively, you can configure with --disable-shared. SOLARIS THREADS: Unless --disable-threads option is given, threads support is on by default in configure. This causes the collector to be compiled with -D GC_THREADS ensuring thread safety. This assumes use of the pthread_ interface; old-style Solaris threads are no longer supported. Thread-local allocation is on by default. Parallel marking is on by default (it could be disabled manually by configure --disable-parallel-mark option). It is also essential that gc.h be included in files that call pthread_create, pthread_join, pthread_detach, or dlopen. gc.h macro defines these to also do GC bookkeeping, etc. gc.h must be included with GC_THREADS macro defined first, otherwise these replacements are not visible. A collector built in this way may only be used by programs that are linked with the threads library. Unless USE_PROC_FOR_LIBRARIES is defined, dlopen disables collection temporarily. In some unlikely cases, this can result in unpleasant heap growth. But it seems better than the race/deadlock issues we had before. If threads are used on an X86 processor with malloc redirected to GC_malloc, it is necessary to call GC_INIT explicitly before forking the first thread. (This avoids a deadlock arising from calling GC_thr_init with the allocation lock held.) There could be an issue when using gc_cpp.h in conjunction with Solaris threads and Sun's C++ runtime. Apparently the overloaded new operator may be invoked by some iostream initialization code before threads are correctly initialized. This may cause a SIGSEGV during initialization of the garbage collector. Currently the only known workaround is to not invoke the garbage collector from a user defined global operator new, or to have it invoke the garbage-collector's allocators only after main has started. (Note that the latter requires a moderately expensive test in operator delete.) I encountered "symbol <unknown>: offset .... is non-aligned" errors. These appear to be traceable to the use of the GNU assembler with the Sun linker. The former appears to generate a relocation not understood by the latter. The fix appears to be to use a consistent toolchain. (As a non-Solaris-expert my solution involved hacking the libtool script, but I'm sure you can do something less ugly.) Hans-J. Boehm (The above contains my personal opinions, which are probably not shared by anyone else.)
Generated by dwww version 1.15 on Mon Jun 17 16:33:11 CEST 2024.