Andreas Kohn ([email protected]) wrote: > On Mon, 2010-04-12 at 09:58 -0700, Kelly O'Hair wrote: > > You need some of the Sun Studio compiler patches made to 12u1 to build > > hotspot. > > Ah, ok. I'll stick to SS12 then.
I'd recommend that for now. There are some mods to hotspot required to compile with ss12u1+patches, including a work around for the assertion failure (the hotspot code was questionable anyway). I don't know if the fix for the assertion is available as a patch to the compiler backend; it may be. > > I'm not exactly sure what that list is yet. I'll CC John, he may know > > that list. Let me know and I can dig up the list. -John > > On Apr 11, 2010, at 10:48 AM, Andreas Kohn wrote: > > > > > Hi, > > > > > > I tried building the latest openjdk forest on opensolaris with the > > > SunStudio 12u1 (from packages), and it failed for both i586 and amd64 > > > builds with an assertion error. > > > Switching to an installation of SunStudio 12 worked without problems. > > > > > > [...] > > > CC -DSOLARIS -DSPARC_WORKS -DAMD64 -DPRODUCT -xF -I../generated/ > > > adfiles > > > -I../generated/jvmtifiles > > > -I/export/home/andreas/jdk7/jdk7/hotspot/src/share/vm/asm > > > -I/export/home/andreas/jdk7/jdk7/hotspot/src/share/vm/ci > > > -I/export/home/andreas/jdk7/jdk7/hotspot/src/share/vm/classfile > > > -I/export/home/andreas/jdk7/jdk7/hotspot/src/share/vm/code > > > -I/export/home/andreas/jdk7/jdk7/hotspot/src/share/vm/compiler > > > -I/export/home/andreas/jdk7/jdk7/hotspot/src/share/vm/ > > > gc_implementation > > > -I/export/home/andreas/jdk7/jdk7/hotspot/src/share/vm/ > > > gc_implementation/parNew -I/export/home/andreas/jdk7/jdk7/hotspot/ > > > src/share/vm/gc_implementation/concurrentMarkSweep -I/export/home/ > > > andreas/jdk7/jdk7/hotspot/src/share/vm/gc_implementation/ > > > parallelScavenge -I/export/home/andreas/jdk7/jdk7/hotspot/src/share/ > > > vm/gc_implementation/shared -I/export/home/andreas/jdk7/jdk7/hotspot/ > > > src/share/vm/gc_implementation/g1 -I/export/home/andreas/jdk7/jdk7/ > > > hotspot/src/share/vm/gc_interface -I/export/home/andreas/jdk7/jdk7/ > > > hotspot/src/share/vm/interpreter -I/export/home/andreas/jdk7/jdk7/ > > > hotspot/src/share/vm/libadt -I/export/home/andreas/jdk7/jdk7/hotspot/ > > > src/share/vm/memory -I/export/home/andreas/jdk7/jdk7/hotspot/src/ > > > share/vm/oops -I/export/home/andreas/jdk7/jdk7/hotspot/src/share/vm/ > > > opto -I/export/home/andreas/jdk7/jdk7/hotspot/src/share/vm/prims -I/ > > > export/home/andreas/jdk7/jdk7/hotspot/src/share/vm/runtime -I/export/ > > > home/andreas/jdk7/jdk7/hotspot/src/share/vm/services -I/export/home/ > > > andreas/jdk7/jdk7/hotspot/src/share/vm/utilities -I/export/home/ > > > andreas/jdk7/jdk7/hotspot/src/cpu/x86/vm -I/export/home/andreas/jdk7/ > > > jdk7/hotspot/src/os/solaris/vm -I/export/home/andreas/jdk7/jdk7/ > > > hotspot/src/os_cpu/solaris_x86/vm -I../generated - > > > DHOTSPOT_RELEASE_VERSION="\"18.0-b02\"" - > > > DHOTSPOT_BUILD_TARGET="\"product\"" -DHOTSPOT_BUILD_USER="\"andreas > > > \"" -DHOTSPOT_LIB_ARCH=\"amd64\" -DJRE_RELEASE_VERSION="\"1.7.0- > > > internal-andreas_2010_04_11_18_54-b00\"" - > > > DHOTSPOT_VM_DISTRO="\"OpenJDK\"" -DCOMPILER2 -DSOLARIS_7_OR_LATER - > > > m64 -m64 /export/home/andreas/jdk7/jdk7/hotspot/src/os_cpu/ > > > solaris_x86/vm/solaris_x86_64.il -template=no%extdef -features=no > > > %split_init -D_Crun_inline_placement -library=%none -KPIC -mt -xO4 > > > -DVM_LITTLE_ENDIAN -features=no%except -DHAVE_DTRACE_H - > > > DDTRACE_ENABLED -c -o accessFlags.o /export/home/andreas/jdk7/jdk7/ > > > hotspot/src/share/vm/utilities/accessFlags.cpp > > > > > > assertion failed in function cfg_coalesce_lf_blocks() @ cfg.c:4520 > > > assert(nd_equal_(lab, lf_opnd1_(x_last_lf))) > > > > > > > > > > > > I attached the full build log and sanity messages, here are the > > > versions > > > used: > > > > > > SunOS xentros 5.11 snv_134 i86pc i386 i86pc Solaris > > > cc: Sun C 5.10 SunOS_i386 2009/06/03 > > > CC: Sun C++ 5.10 SunOS_i386 128229-02 2009/09/21 > > > > > > Is this failure known? The error is reproducibly happening, always for > > > accessFlags.cpp. > > > > > > > > > Regards, > > > -- > > > Andreas > > > <typescript.gz><sanityCheckMessages.txt> > > > > -- > Never attribute to malice that which can be adequately explained by > stupidity. -- Hanlon's Razor
