Standard Solaris 10 X86 machine setup procedure with the JPRT systems is to run ( cd /usr/lib/dtrace && gzip mpi.d ) as root user.
I gave up a long time ago ever expecting a fix for this. -kto On Jul 31, 2013, at 5:52 PM, David Holmes wrote: > Hi Ioi, > > This is an old known issue. Simplest fix if you have permission is to rename > /usr/lib/dtrace/mpi.d to not be a .d file. > > David > > On 1/08/2013 7:52 AM, Ioi Lam wrote: >> |Hi, when I try to build for solaris/x86/32bit/client, I get this error >> (on a 64-bit host):|| >> || >> However, I can successfully build the 64-bit VM. >> >> --- make log --- >> ||Making libjvm_db.so|| >> ||cc -m32 -DCOMPILER1 -I. -I../generated \|| >> || -G -M >> /net/sc11136754/scratch/iklam/jdk/vtab/make/solaris/makefiles/mapfile-vers-jvm_db >> -mt -KPIC -xnolib -o libjvm_db.so >> /net/sc11136754/scratch/iklam/jdk/vtab/src/os/solaris/dtrace/libjvm_db.c >> -lc|| >> ||dtrace: failed to compile script dtrace.d: "/usr/lib/dtrace/mpi.d", >> line 68: syntax error near "umpiconninfo_t"|| >> ||--- >> >> The failing command is this >> >> ------------ >> $ (cd >> /net/sc11136754/scratch/iklam/jdk/vtab/build/solaris/solaris_i486_compiler1/product; >> /usr/sbin/dtrace -C -I. -G -xlazyload -o dtrace.o -s dtrace.d ciEnv.o >> classLoadingService.o compileBroker.o hashtable.o instanceKlass.o java.o >> jni.o jvm.o memoryManager.o nmethod.o objectMonitor.o runtimeService.o >> sharedRuntime.o synchronizer.o thread.o unsafe.o vmThread.o >> vmCMSOperations.o vmPSOperations.o vmGCOperations.o ) >> dtrace: failed to compile script dtrace.d: "/usr/lib/dtrace/mpi.d", line >> 68: syntax error near "umpiconninfo_t" >> ----------- >> >> ||Any hints? >> >> Thanks >> - Ioi >> |
