This is the mail archive of the java-patches@gcc.gnu.org mailing list for the Java project.
| Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
|---|---|---|
| Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |
| Other format: | [Raw text] | |
Andreas Tobler writes:
I tried this patch on solaris8 sparc and I saw an improvement in the libgcj test (32 and 64 bit). But I hang in the gctest. I might debug next week.
After manually working around the hang with the alternate thread library in /usr/lib/lwp, I've now a clean patch which uses it. One can easily use gcc -print-multi-os-directory to detect the /usr/lib/lwp subdir corresponding to the current multilib.
Here's the revised patch, tested with
make configure-target-boehm-gc make check
on Solaris 8 for both multilibs. I have to explicitly link with -lthread, otherwise the default version in /usr/lib is used: libpthread is a filter library, and its dependencies are only searched in LD_LIBRARY_PATH and the RPATH/RUNPATH encoded in the affected library (libpthread, i.e. none). This is confirmed on threads(3THR), which specifies
cc -mt ... -lpthread ... -R/usr/lib/lwp
where -mt expands to -lthread.
The other change is to use THREADLIBS instead of THREADDLLIBS in gctest_LDADD so the proper flags are passed to the gctest link. This is probably a merge error from upstream, where the variable is called THREADDLLIBS in gc6.8.
I suppose this modified patch is still ok for 4.1/4.2/4.3?
| Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
|---|---|---|
| Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |