Merge pull request #2010 from DARMA-tasking/2009-release-1-2-1

2009: Create release 1.2.1
DARMA-tasking · Nov 9, 2022 · 7c16301 · 7c16301
2 parents 3ff11e0 + 672ebc1
commit 7c16301
Show file tree

Hide file tree

Showing 100 changed files with 4,073 additions and 3,438 deletions.
diff --git a/VERSION b/VERSION
@@ -1 +1 @@
-1.1.0
+1.2.1
diff --git a/ci/ctest_job_script.cmake b/ci/ctest_job_script.cmake
@@ -23,6 +23,8 @@ set(CTEST_SOURCE_DIRECTORY
 set(CTEST_BINARY_DIRECTORY
     "$ENV{VT_BUILD}"
 )
+set(CTEST_GIT_COMMAND "git")
+set(CTEST_UPDATE_VERSION_ONLY 1)
 
 if ( NOT DEFINED ENV{VT_TRACE_RUNTIME_ENABLED} )
   set(ENV{VT_TRACE_RUNTIME_ENABLED} "0")
@@ -169,8 +171,10 @@ set(configureOpts
 )
 #ctest_empty_binary_directory(${CTEST_BINARY_DIRECTORY})
 ctest_start(Continuous)
+ctest_update()
+ctest_submit(PARTS Start Update RETRY_COUNT 6 RETRY_DELAY 10)
 ctest_configure(OPTIONS "${configureOpts}" RETURN_VALUE ret_conf)
-ctest_submit(PARTS Start Configure RETRY_COUNT 6 RETRY_DELAY 10)
+ctest_submit(PARTS Configure RETRY_COUNT 6 RETRY_DELAY 10)
 if (NOT ret_conf)
   ctest_build(RETURN_VALUE ret_build)
   ctest_submit(PARTS Build RETRY_COUNT 6 RETRY_DELAY 10)

diff --git a/docs/md/lb-manager.md b/docs/md/lb-manager.md
@@ -15,19 +15,19 @@ instrumentation of work and communication performed by collection elements.
 To run a load balancer at runtime:
 
 - Pass `--vt_lb --vt_lb_name=<LB>` as a command line argument
-- Write a LB specification file `--vt_lb --vt_lb_file_name=<FILE>`
+- Write a LB config file `--vt_lb --vt_lb_file_name=<FILE>`
 - One can also pass `--vt_lb_self_migration` as a command line argument to allow load balancer to migrate objects to the same node
 
 Note that one should use either `--vt_lb_name` or `--vt_lb_file_name` option, not both.
 
-\section lb-specification-file LB Specification File
+\section lb-config-file LB Config File
 
-The LB specification file allows users to specify which load balancer along with
+The LB config file allows users to specify which load balancer along with
 which LB-specific configuration parameters are passed to the load balancer
 instance for any given phase. The order of the LB phase specification lines in
 the file disambiguates lines---higher precedence for earlier lines.
 
-The format of the LB specification file is:
+The format of the LB config file is:
 
 \code
 [%] <$phase> <$lbname> [$LB-specific-arg-1] ... [$LB-specific-arg-N]
@@ -42,7 +42,7 @@ balancer are allowed to customize how the load balancer is run with the format
 of `key=value`. These arguments are the equivalent of passing
 `--vt_lb_args="A=test B=test2"` on the command line.
 
-The following is an example LB specification:
+The following is an example LB config:
 
 \code
 %10 TemperedLB c=1 k=5 f=2 i=10
@@ -51,7 +51,7 @@ The following is an example LB specification:
 120 GreedyLB c=0 k=2 f=3 i=3
 \endcode
 
-To print LB specification during startup, use `--vt_lb_show_spec` command line flag.
+To print LB config during startup, use `--vt_lb_show_config` command line flag.
 
 \section load-balancers Load balancers
 

diff --git a/docs/md/node-lb-data.md b/docs/md/node-lb-data.md
@@ -197,3 +197,10 @@ The type of communication lines up with the enum
 
 For all the broadcast-like edges, the communication logging will occur on the
 receive of the broadcast side (one entry per broadcast recipient).
+
+\section lb-spec-file LB Specification File
+In order to customize when LB output is enabled and disabled, a LB
+specification file can be passed to \vt via a command-line flag:
+`--vt_lb_spec --vt_lb_spec_file=filename.spec`.
+
+For details about vt's Specification File see \ref spec-file
diff --git a/docs/md/spec-file.md b/docs/md/spec-file.md
@@ -0,0 +1,48 @@
+\page spec-file Spec File
+\brief Specification File for vt components
+
+VT is using Specification Files for various components (trace and LB). Such files can be used to control when (on which Phases) these components should be enabled/disabled.
+
+The parser will read the following format:
+
+\code
+[%] <phase> <range negative> <range positive>
+\endcode
+
+The following is an example of a specification file:
+
+\code
+0 0 10
+%100 -3 3
+200 -5 5
+\endcode
+
+This specifies that component will be enabled on the following phases:
+
+\code
+{
+  [0,10], # phase 0 with offsets 0,+10 (subsumes [0,3] from %100 -3 3)
+  [97,103] # any phase % 100 with offset -3,+3
+  [195,205] # phase 200 with offsets -5,+5 (subsumes [197,203] from %100 -3 3)
+  [297,303] # any phase % 100 with offset -3,+3
+  [n%100-3,n%100+3] ... # any phase % 100 with offset -3,+3
+}
+\endcode
+
+The sets of mod-phase and phase-specific entries must be unique. There may be
+overlap across the two sets, but not within them. Having two entries that
+start with `%100` or two entries that start with `100` would be invalid and
+trigger a parsing error. But having a `%100` and `100` entry is valid.
+Whether component is enabled is calculated as an OR across all specification
+entries. Thus, if a given phase is contained in any spec line, it is
+enabled. Note that `0 % 100 = 0`. Therefore, if the above example did not
+contain the first line, component would be enabled as:
+
+\code
+{
+  [0,3], # any phase mod 100 from -3,+3
+  [97,103],
+  [195,205],
+  [297,303], ...
+}
+\endcode
diff --git a/docs/md/trace.md b/docs/md/trace.md
@@ -19,46 +19,4 @@ In order to customize when tracing is enabled and disabled, a trace
 specification file can be passed to \vt via a command-line flag:
 `--vt_trace_spec --vt_trace_spec_file=filename.spec`.
 
-The parser will read the  following format:
-
-\code
-[%] <phase> <range negative> <range positive>
-\endcode
-
-The following is an example of a trace specification:
-
-\code
-0 0 10
-%100 -3 3
-200 -5 5
-\endcode
-
-This specifies that tracing will be enabled on the following phases:
-
-\code
-{
-  [0,10], # phase 0 with offsets 0,+10 (subsumes [0,3] from %100 -3 3)
-  [97,103] # any phase % 100 with offset -3,+3
-  [195,205] # phase 200 with offsets -5,+5 (subsumes [197,203] from %100 -3 3)
-  [297,303] # any phase % 100 with offset -3,+3
-  [n%100-3,n%100+3] ... # any phase % 100 with offset -3,+3
-}
-\endcode
-
-The sets of mod-phase and phase-specific entries must be unique. There may be
-overlap across the two sets, but not within them. Having two entries that
-start with `%100` or two entries that start with `100` would be invalid and
-trigger a parsing error. But having a `%100` and `100` entry is valid.
-Whether tracing is enabled is calculated as an OR across all specification
-entries. Thus, if a given phase is contained in any spec line, it is
-enabled. Note that `0 % 100 = 0`. Therefore, if the above example did not
-contain the first line, tracing would be enabled as:
-
-\code
-{
-  [0,3], # any phase mod 100 from -3,+3
-  [97,103],
-  [195,205],
-  [297,303], ...
-}
-\endcode
+For details about vt's Specification File see \ref spec-file
diff --git a/scripts/report_logs_in_comment.sh b/scripts/report_logs_in_comment.sh
@@ -103,8 +103,9 @@ fi
 
 # Build comment
 commit_sha="$(git log --skip=1 -1  --pretty=format:%H)"
+commit_date="$(TZ=UTC0 git show -s --format=%cd --date=format-local:'%Y-%m-%d %H:%M:%S' "$commit_sha")"
 build_link='[Build log](https://dev.azure.com/DARMA-tasking/DARMA/_build/results?buildId='"$build_id"'&view=logs&j='"$job_id"'&t='"$task_id)"
-comment_body="Build for $commit_sha\n\n"'```'"\n$val\n"'```'"\n\n$build_link"
+comment_body="Build for $commit_sha ($commit_date UTC)\n\n"'```'"\n$val\n"'```'"\n\n$build_link"
 
 # Fix new lines
 new_line="\n"

diff --git a/src/CMakeLists.txt b/src/CMakeLists.txt
@@ -97,7 +97,7 @@ set(
     serialization
       serialization/messaging serialization/traits serialization/auto_dispatch
       serialization/sizing
-    utils/demangle utils/container utils/bits utils/mutex
+    utils/demangle utils/container utils/bits utils/mutex utils/file_spec
     utils/hash utils/atomic utils/tls utils/static_checks utils/string
     utils/memory utils/mpi_limits utils/compress utils/json utils/strong
     registry/auto
@@ -107,7 +107,6 @@ set(
       objgroup/proxy objgroup/holder objgroup/active_func objgroup/dispatch
       objgroup/type_registry
     trace
-      trace/file_spec
 )
 list(APPEND PROJECT_SUBDIRS_LIST ${TOP_LEVEL_SUBDIRS})
 

diff --git a/src/vt/collective/collective_ops.cc b/src/vt/collective/collective_ops.cc
@@ -125,7 +125,7 @@ void printOverwrittens(
   printIfOverwritten(vt_trace_event_polling);
   printIfOverwritten(vt_trace_irecv_polling);
   printIfOverwritten(vt_lb);
-  printIfOverwritten(vt_lb_show_spec);
+  printIfOverwritten(vt_lb_show_config);
   printIfOverwritten(vt_lb_quiet);
   printIfOverwritten(vt_lb_file_name);
   printIfOverwritten(vt_lb_name);

diff --git a/src/vt/configs/arguments/app_config.h b/src/vt/configs/arguments/app_config.h
@@ -136,7 +136,7 @@ struct AppConfig {
   bool vt_trace_irecv_polling     = false;
 
   bool vt_lb                     = false;
-  bool vt_lb_show_spec           = false;
+  bool vt_lb_show_config           = false;
   bool vt_lb_quiet               = false;
   std::string vt_lb_file_name    = "";
   std::string vt_lb_name         = "NoLB";
@@ -155,6 +155,9 @@ struct AppConfig {
   std::string vt_lb_statistics_dir = "";
   bool vt_help_lb_args           = false;
   bool vt_lb_self_migration      = false;
+  bool vt_lb_spec                = false;
+  std::string vt_lb_spec_file    = "";
+
 
   bool vt_no_detect_hang       = false;
   bool vt_print_no_progress    = true;
@@ -306,7 +309,7 @@ struct AppConfig {
       | vt_trace_irecv_polling
 
       | vt_lb
-      | vt_lb_show_spec
+      | vt_lb_show_config
       | vt_lb_quiet
       | vt_lb_file_name
       | vt_lb_name

diff --git a/src/vt/configs/arguments/args.cc b/src/vt/configs/arguments/args.cc
@@ -313,7 +313,7 @@ void addTraceArgs(CLI::App& app, AppConfig& appConfig) {
   auto qy = app.add_flag("--vt_trace_sys_collection",    appConfig.vt_trace_sys_collection,    tsyscoll);
   auto qz = app.add_flag("--vt_trace_sys_serial_msg",    appConfig.vt_trace_sys_serial_msg,    tsyssmsg);
   auto qza = app.add_flag("--vt_trace_spec",             appConfig.vt_trace_spec,              tspec);
-  auto qzb = app.add_option("--vt_trace_spec_file",      appConfig.vt_trace_spec_file,         tspecfile)->capture_default_str();
+  auto qzb = app.add_option("--vt_trace_spec_file",      appConfig.vt_trace_spec_file,         tspecfile)->capture_default_str()->check(CLI::ExistingFile);
   auto qzc = app.add_flag("--vt_trace_memory_usage",     appConfig.vt_trace_memory_usage,      tmemusage);
   auto qzd = app.add_flag("--vt_trace_event_polling",    appConfig.vt_trace_event_polling,     tpolled);
   auto qze = app.add_flag("--vt_trace_irecv_polling",    appConfig.vt_trace_irecv_polling,     tirecv);
@@ -459,11 +459,11 @@ void addLbArgs(CLI::App& app, AppConfig& appConfig) {
   /*
    * Flags for enabling load balancing and configuring it
    */
-  auto lb            = "Enable load balancing";
-  auto lb_args       = "Arguments pass to LB: \"x=0 y=1\"; try --vt_help_lb_args";
-  auto lb_quiet      = "Silence load balancing output";
-  auto lb_file_name  = "LB specification file to read";
-  auto lb_show_spec  = "Show LB specification during startup";
+  auto lb             = "Enable load balancing";
+  auto lb_args        = "Arguments pass to LB: \"x=0 y=1\"; try --vt_help_lb_args";
+  auto lb_quiet       = "Silence load balancing output";
+  auto lb_file_name   = "LB config file to read";
+  auto lb_show_config = "Show LB config during startup";
   auto lb_name       = "Name of the load balancer to use";
   auto lb_interval   = "Load balancing interval";
   auto lb_keep_last_elm = "Do not migrate last element in collection";
@@ -478,10 +478,12 @@ void addLbArgs(CLI::App& app, AppConfig& appConfig) {
   auto lb_statistics_file = "Load balancing statistics output file name";
   auto lb_statistics_dir  = "Load balancing statistics output directory name";
   auto lb_self_migration = "Allow load balancer to migrate objects to the same node";
+  auto lb_spec      = "Enable LB spec file (defines which phases output LB data)";
+  auto lb_spec_file = "File containing LB spec; --vt_lb_spec to enable";
   auto s  = app.add_flag("--vt_lb", appConfig.vt_lb, lb);
   auto t1 = app.add_flag("--vt_lb_quiet", appConfig.vt_lb_quiet, lb_quiet);
   auto u  = app.add_option("--vt_lb_file_name", appConfig.vt_lb_file_name, lb_file_name)->capture_default_str()->check(CLI::ExistingFile);
-  auto u1 = app.add_flag("--vt_lb_show_spec", appConfig.vt_lb_show_spec, lb_show_spec);
+  auto u1 = app.add_flag("--vt_lb_show_config", appConfig.vt_lb_show_config, lb_show_config);
   auto v  = app.add_option("--vt_lb_name", appConfig.vt_lb_name, lb_name)->capture_default_str();
   auto v1 = app.add_option("--vt_lb_args", appConfig.vt_lb_args, lb_args)->capture_default_str();
   auto w  = app.add_option("--vt_lb_interval", appConfig.vt_lb_interval, lb_interval)->capture_default_str();
@@ -496,7 +498,9 @@ void addLbArgs(CLI::App& app, AppConfig& appConfig) {
   auto yy = app.add_flag("--vt_lb_statistics_compress", appConfig.vt_lb_statistics_compress, lb_statistics_comp);
   auto yz = app.add_option("--vt_lb_statistics_file",   appConfig.vt_lb_statistics_file,     lb_statistics_file)->capture_default_str();
   auto zz = app.add_option("--vt_lb_statistics_dir",    appConfig.vt_lb_statistics_dir,      lb_statistics_dir)->capture_default_str();
-  auto lbasm = app.add_flag("--vt_lb_self_migration", appConfig.vt_lb_self_migration, lb_self_migration);
+  auto lbasm = app.add_flag("--vt_lb_self_migration",   appConfig.vt_lb_self_migration,      lb_self_migration);
+  auto lbspec = app.add_flag("--vt_lb_spec",            appConfig.vt_lb_spec,                lb_spec);
+  auto lbspecfile = app.add_option("--vt_lb_spec_file", appConfig.vt_lb_spec_file,           lb_spec_file)->capture_default_str()->check(CLI::ExistingFile);
 
   auto debugLB = "Load Balancing";
   s->group(debugLB);
@@ -518,6 +522,8 @@ void addLbArgs(CLI::App& app, AppConfig& appConfig) {
   yz->group(debugLB);
   zz->group(debugLB);
   lbasm->group(debugLB);
+  lbspec->group(debugLB);
+  lbspecfile->group(debugLB);
 
   // help options deliberately omitted from the debugLB group above so that
   // they appear grouped with --vt_help when --vt_help is used

diff --git a/src/vt/messaging/active.cc b/src/vt/messaging/active.cc
@@ -151,9 +151,8 @@ void ActiveMessenger::initialize() {
 }
 
 void ActiveMessenger::startup() {
-  auto const this_node = theContext()->getNode();
   bare_handler_dummy_elm_id_for_lb_data_ =
-    elm::ElmIDBits::createBareHandler(this_node);
+    elm::ElmIDBits::createBareHandler(this_node_);
 
 #if vt_check_enabled(lblite)
   // Hook to collect LB data about objgroups
@@ -225,7 +224,7 @@ EventType ActiveMessenger::sendMsgBytesWithPut(
   }
 
   vtWarnIf(
-    dest == theContext()->getNode() &&
+    dest == this_node_ &&
     not is_bcast &&
     not theConfig()->vt_lb_self_migration,
     fmt::format("Destination {} should != this node", dest)
@@ -375,7 +374,6 @@ EventType ActiveMessenger::sendMsgMPI(
       "sendMsgMPI: (multi): size={}\n", msg_size
     );
     auto tag = allocateNewTag();
-    auto this_node = theContext()->getNode();
 
     // Send the actual data in multiple chunks
     PtrLenPairType tup = std::make_tuple(untyped_msg, msg_size);
@@ -387,7 +385,7 @@ EventType ActiveMessenger::sendMsgMPI(
     mpi_event->setManagedMessage(base.to<ShortMessage>());
 
     // Send the control message to receive the multiple chunks of data
-    auto m = makeMessage<MultiMsg>(info, this_node, msg_size);
+    auto m = makeMessage<MultiMsg>(info, this_node_, msg_size);
     sendMsg<MultiMsg, chunkedMultiMsg>(dest, m);
 
     return event_id;
@@ -491,9 +489,8 @@ EventType ActiveMessenger::doMessageSend(
   );
 
   // Don't go through MPI with self-send, schedule the message locally instead
-  auto const this_node = theContext()->getNode();
   if (deliver) {
-    if (dest != this_node) {
+    if (dest != this_node_) {
       sendMsgBytesWithPut(dest, base, send_tag);
     } else {
       recordLBDataCommForSend(dest, base, base.size());
@@ -611,7 +608,7 @@ std::tuple<EventType, int> ActiveMessenger::sendDataMPI(
   }
 
   if (events.size() > 1) {
-    ret_event = theEvent()->createParentEvent(theContext()->getNode());
+    ret_event = theEvent()->createParentEvent(this_node_);
     auto& holder = theEvent()->getEventHolder(ret_event);
     for (auto&& child_event : events) {
       holder.get_event()->addEventToList(child_event);
@@ -864,8 +861,9 @@ void ActiveMessenger::recordLBDataCommForSend(
   NodeType const dest, MsgSharedPtr<BaseMsgType> const& base,
   MsgSizeType const msg_size
 ) {
-  if (theContext()->getTask() != nullptr) {
-    auto lb = theContext()->getTask()->get<ctx::LBData>();
+  auto the_task = theContext()->getTask();
+  if (the_task != nullptr) {
+    auto lb = the_task->get<ctx::LBData>();
 
     if (lb) {
       auto const& msg = base.get();
@@ -874,7 +872,7 @@ void ActiveMessenger::recordLBDataCommForSend(
 
       if (not already_recorded) {
         auto dest_elm_id = elm::ElmIDBits::createBareHandler(dest);
-        theContext()->getTask()->send(dest_elm_id, msg_size);
+        the_task->send(dest_elm_id, msg_size);
       }
     }
   }

diff --git a/src/vt/messaging/active.impl.h b/src/vt/messaging/active.impl.h
@@ -169,7 +169,7 @@ ActiveMessenger::PendingSendType ActiveMessenger::sendMsgCopyableImpl(
   }
 
   if (is_bcast) {
-    dest = theContext()->getNode();
+    dest = this_node_;
   }
   if (tag != no_tag) {
     envelopeSetTag(rawMsg->env, tag);

diff --git a/src/vt/runtime/runtime.cc b/src/vt/runtime/runtime.cc
@@ -642,6 +642,7 @@ void Runtime::setup() {
 # if vt_check_enabled(trace_enabled)
   theTrace->loadAndBroadcastSpec();
 # endif
+  theNodeLBData->loadAndBroadcastSpec();
 
   if (theConfig()->vt_pause) {
     pauseForDebugger();