Bugfix in the mechanism of handling failed MP workers in the Event Service.
This patch covers the case when a worker fails to transition from the event
processing state into the finalization state after receiving "No more events"
message from the range scatterer. The idea here is that instead of releasing
a fixed number of workers (nprocs), the scatterer will keep releasing the
workers until it gets a signal that all of them have finished.

Update AnaAlgorithm.h: remove obsolete comment

See merge request atlas/athena!39395

TrigAnalysisExamples: Fix gcc11 warnings.

See merge request atlas/athena!39388

TrkVertexFitters: Fix use of uninitialized variable.

See merge request atlas/athena!39387

Hepmc3 nightly fixes 22122020 part 1

See merge request atlas/athena!39377

CaloG4Sim: Enable thread-safety checking.

See merge request atlas/athena!39386

update skel.GENtoTXT.py to be in agreements with skeleton in rel. 21.6

See merge request atlas/athena!39361

Ewelina Maria Lobodzinska authored 4 years ago and Walter Lampl committed 4 years ago

update gen_tf.py and skel.GENtoEVGEN.py to include the most recent modifications introduced for rel. 21.6

H7 art tests

See merge request atlas/athena!39312

RootStorageSvc+RootAuxDynIO: Fix memory leaks.

See merge request atlas/athena!39384

Scott Snyder authored 4 years ago and Walter Lampl committed 4 years ago

The IRootAuxDynReader objects were never deleted.
Change to hold them via unique_ptr.

MuonEventAthenaPool: Fix memory leak.

See merge request atlas/athena!39382

Jet hypo - support multiple FastReducer instances.

See merge request atlas/athena!39379

Better memory handling in PseudoJetContainer

See merge request atlas/athena!39374

Ntuple for validation in the seed making stage

See merge request atlas/athena!37387

Jay Ajitbhai Sandesara authored 4 years ago and Walter Lampl committed 4 years ago

Moved from writing ntuple per event to per seed, for seeds seen by the SiSPSeededTrackFinder algorithm.

Some cleanup in muon packages

See merge request atlas/athena!39364

Sweep previously ignored patch to 21.6 into master

See merge request atlas/athena!39389

MuonTruthAlgs: Fix memory leaks.

See merge request atlas/athena!39385

Scott Snyder authored 4 years ago and Walter Lampl committed 4 years ago

In MuonTruthDecorationAlg, firstPos and secondPos were not deleted
on all possible code paths.

Change them to use automatic allocation instead.

MuonCombinedTrackFindingTools: Fix memory leak.

See merge request atlas/athena!39383

Patch to ESDtoDPD configuration for IDTIDE

See merge request atlas/athena!39378

In the EVNT -> EVNT workflows where the run number needs adjusting, the
MC channel number requires the same correction. This will later propogte
into EventInfo clients such as xAOD::EventInfo, EventStreamInfo, and
FileMetaData.

Avoid spurious copies in for loops.

Don't pass an uninitialized vector; pass a zeroed vector instead.

Enable thread-safety checking.
Clean up resulting warnings.

Object returned from TrackSelectionProcessorTool::process() was being leaked.

In MdtCsmCnv_p1::persToTrans, we initialize the transient object via
assigment.  But it is a DataVector, so the assignment will leave
it as a view container, meaning that its contents will be leaked.
Need to convert it back to an owning container before populating it.

StoreGateBindings: Protect against missing hasStore().

See merge request !39371

ActsGeometry: Fix clang warnings.

See merge request atlas/athena!39369

TrkPatternParameters: Protect against passing null surface pointers.

See merge request atlas/athena!39368

TrigHLTJetHypo: Fix debug logging.

See merge request atlas/athena!39367

MuonTrackSteeringTools: Another attempt to fix dangling Surface references.

See merge request !39366

Update offline pattern rec. tracking for IDTrig

See merge request atlas/athena!39357

SGTools: Fix dummy proxy leak.

See merge request atlas/athena!39348

Scott Snyder authored 4 years ago and Walter Lampl committed 4 years ago

When we remove a proxy, be sure to decrement the reference count even
if it's not in m_storeMap.  A dummy proxy won't have been entered
there, but it will still have gotten a reference count.