<div dir="ltr">Hi Sanne,<div>For the record I am able to recreate this issue if I create a custom variant and omit jvmti feature using the following configure command:</div><div><br></div><div><font face="monospace">bash ./configure --with-conf-name=no-jvmti-release --with-jvm-features=cds,compiler1,compiler2,g1gc,serialgc,jfr,jni-check,jvmci,management,services --with-jvm-variants=custom<br></font></div><div><br><div>I tried to fix it by enclosing the code in SCCache.cpp with INCLUDE_JVMTI but the build failed again in premain specific code with this error:</div><div><br></div><div><font face="monospace">/home/asmehra/data/ashu-mehra/leyden/src/hotspot/share/interpreter/interpreterRuntime.cpp:1804:23: error: ‘_perf_InterpreterRuntime_member_name_arg_or_null_timer’ was not declared in this scope; did you mean ‘_perf_InterpreterRuntime_prepare_native_call_timer’?<br> 1804 |   NEWPERFTICKCOUNTERS(_perf_##sub##_##name##_timer, SUN_CI, #sub "::" #name); \<br>      |                       ^~~~~~<br>/home/asmehra/data/ashu-mehra/leyden/src/hotspot/share/runtime/perfData.hpp:854:4: note: in definition of macro ‘NEWPERFTICKCOUNTERS’<br>  854 |   {counter = PerfDataManager::create_tick_counters(counter_ns, counter_name, counter_name "_elapsed_time", \<br>      |    ^~~~~~~<br>/home/asmehra/data/ashu-mehra/leyden/src/hotspot/share/interpreter/interpreterRuntime.cpp:1798:3: note: in expansion of macro ‘INIT_COUNTER’<br> 1798 |   macro(InterpreterRuntime, member_name_arg_or_null)<br>      |   ^~~~~<br>/home/asmehra/data/ashu-mehra/leyden/src/hotspot/share/interpreter/interpreterRuntime.cpp:1812:5: note: in expansion of macro ‘DO_JVMTI_COUNTERS’<br> 1812 |     DO_JVMTI_COUNTERS(INIT_COUNTER)<br>      |     ^~~~~~~~~~~~~~~~~<br></font><br><div><div dir="ltr" class="gmail_signature" data-smartmail="gmail_signature"><div>I worked around it and it next failed with following errors:</div><div><br></div><div><font face="monospace">/usr/bin/ld: /home/asmehra/data/ashu-mehra/leyden/build/premain-release/hotspot/variant-custom/libjvm/objs/jfrJvmtiAgent.o: in function `JvmtiUtil::error_name(int)':<br>make/hotspot/src/hotspot/share/prims/jvmtiUtil.hpp:51: undefined reference to `JvmtiUtil::_error_names'<br>/usr/bin/ld: /home/asmehra/data/ashu-mehra/leyden/build/premain-release/hotspot/variant-custom/libjvm/objs/jfrJvmtiAgent.o: in function `JvmtiEnvBase::get_phase()':<br>make/hotspot/src/hotspot/share/prims/jvmtiEnvBase.hpp:77: undefined reference to `JvmtiEnvBase::_phase'<br>/usr/bin/ld: /home/asmehra/data/ashu-mehra/leyden/build/premain-release/hotspot/variant-custom/libjvm/objs/jfrPeriodic.o: in function `JfrPeriodicEventSet::requestJavaAgent()':<br>make/hotspot/src/hotspot/share/jfr/periodic/jfrPeriodic.cpp:294: undefined reference to `JvmtiAgentList::java_agents()'<br>/usr/bin/ld: /home/asmehra/data/ashu-mehra/leyden/build/premain-release/hotspot/variant-custom/libjvm/objs/jfrPeriodic.o: in function `JfrPeriodicEventSet::requestNativeAgent()':<br>make/hotspot/src/hotspot/share/jfr/periodic/jfrPeriodic.cpp:314: undefined reference to `JvmtiAgentList::native_agents()'<br>/usr/bin/ld: make/hotspot/src/hotspot/share/jfr/periodic/jfrPeriodic.cpp:316: undefined reference to `JvmtiAgentList::xrun_agents()'<br>/usr/bin/ld: /home/asmehra/data/ashu-mehra/leyden/build/premain-release/hotspot/variant-custom/libjvm/objs/jvmciCompilerToVMInit.o: in function `CompilerToVM::Data::initialize(JVMCIEnv*)':<br>make/hotspot/src/hotspot/share/jvmci/jvmciCompilerToVMInit.cpp:212: undefined reference to `JvmtiExport::_should_notify_object_alloc'<br>collect2: error: ld returned 1 exit status</font><br></div><div><font face="monospace"><br></font></div><div><font face="arial, sans-serif">This is not specific to premain, and I can reproduce it with mainline as well. It doesn't look like it is straightforward to remove the jvmti feature. Seems like there may be other components (like jfr) missing INCLUDE_JVMTI macro.</font></div><div><font face="arial, sans-serif">My suggestion, if you are interested in pursuing it further, would be to try and wrinkle out the problems with the mainline first.</font></div><div><font face="arial, sans-serif"><br></font></div><div><font face="arial, sans-serif">Meanwhile I will push the changes to use </font><span style="font-family:arial,sans-serif">INCLUDE_JVMTI in the premain specific code.</span></div><div><font face="arial, sans-serif"><br></font></div><div dir="ltr">- Ashutosh Mehra</div></div></div><br></div></div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Mon, Jul 8, 2024 at 11:17 AM Ashutosh Mehra <<a href="mailto:asmehra@redhat.com">asmehra@redhat.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="ltr">Hi Sanne, thanks for reporting this. It looks like a bug. It can be fixed by enclosing the relevant code under the "INCLUDE_JVMTI" macro.<div><br></div><div><div><div dir="ltr" class="gmail_signature"><div dir="ltr">- Ashutosh Mehra</div></div></div><br></div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Mon, Jul 8, 2024 at 9:58 AM Sanne Grinovero <<a href="mailto:sanne@redhat.com" target="_blank">sanne@redhat.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="ltr">Hello all,<div><br></div><div>I'm experimenting building Leyden locally with non-default configurations, with the intent of comparing its potential as an alternative to GraalVM native-images, and one of my goals would be to produce a minimal JVM build which exclusively includes the components that our applications are going to need at runtime.</div><div><br></div><div>This implies being very selective with the JVM features I choose at configuration time; I've already opened some minor issues identified during such experiments.</div><div><br></div><div>Today I noticed today that if I exclude the "jvmti" feature, the premain branch of Leyden doesn't compile:</div><div><br></div><div><br></div><div><div>=== Output from failing command(s) repeated here ===<br>* For target hotspot_variant-custom_libjvm_objs_SCCache.o:<br>/home/sanne/sources/leyden/src/hotspot/share/code/SCCache.cpp: In member function ‘void SCAddressTable::init_opto()’:<br>/home/sanne/sources/leyden/src/hotspot/share/code/SCCache.cpp:3815:39: error: ‘notify_jvmti_vthread_start’ is not a member of ‘OptoRuntime’<br> 3815 |   SET_ADDRESS(_C2_blobs, OptoRuntime::notify_jvmti_vthread_start());<br>      |                                       ^~~~~~~~~~~~~~~~~~~~~~~~~~<br>/home/sanne/sources/leyden/src/hotspot/share/code/SCCache.cpp:3511:47: note: in definition of macro ‘SET_ADDRESS’<br> 3511 |     type##_addr[type##_length++] = (address) (addr);      \<br>      |                                               ^~~~<br>/home/sanne/sources/leyden/src/hotspot/share/code/SCCache.cpp:3816:39: error: ‘notify_jvmti_vthread_end’ is not a member of ‘OptoRuntime’<br> 3816 |   SET_ADDRESS(_C2_blobs, OptoRuntime::notify_jvmti_vthread_end());<br>      |                                       ^~~~~~~~~~~~~~~~~~~~~~~~<br>/home/sanne/sources/leyden/src/hotspot/share/code/SCCache.cpp:3511:47: note: in definition of macro ‘SET_ADDRESS’<br> 3511 |     type##_addr[type##_length++] = (address) (addr);      \<br>      |                                               ^~~~<br>/home/sanne/sources/leyden/src/hotspot/share/code/SCCache.cpp:3817:39: error: ‘notify_jvmti_vthread_mount’ is not a member of ‘OptoRuntime’<br> 3817 |   SET_ADDRESS(_C2_blobs, OptoRuntime::notify_jvmti_vthread_mount());<br>      |                                       ^~~~~~~~~~~~~~~~~~~~~~~~~~<br>/home/sanne/sources/leyden/src/hotspot/share/code/SCCache.cpp:3511:47: note: in definition of macro ‘SET_ADDRESS’<br> 3511 |     type##_addr[type##_length++] = (address) (addr);      \<br>      |                                               ^~~~<br>/home/sanne/sources/leyden/src/hotspot/share/code/SCCache.cpp:3818:39: error: ‘notify_jvmti_vthread_unmount’ is not a member of ‘OptoRuntime’<br> 3818 |   SET_ADDRESS(_C2_blobs, OptoRuntime::notify_jvmti_vthread_unmount());<br>      |                                       ^~~~~~~~~~~~~~~~~~~~~~~~~~~~<br>/home/sanne/sources/leyden/src/hotspot/share/code/SCCache.cpp:3511:47: note: in definition of macro ‘SET_ADDRESS’<br> 3511 |     type##_addr[type##_length++] = (address) (addr);      \<br>      |                                               ^~~~<br>   ... (rest of output omitted)</div><div><br></div><div>I'm wondering if I should open an issue for this case? I appreciate it's early days and that such aspects might not be a priority currently!</div><div><br></div><div>Thanks,</div><div>Sanne<br><br></div></div></div>
</blockquote></div>
</blockquote></div>