I ran coverity on the jdk11 hotspot sources and want to propose the following fixes. I scanned the linux x86_64 build. Some issues are similar to previous parfait fixes (check for NULL, add guarantees etc.) I also identified some issues I consider real problems:

Real issues
-----------

x86.ad
  For Op_CMoveVF the code always returned 'false'.

compileLog.cpp
  read returns -1 on unsuccesful completion. Use a
  singned variable and check for error before
  subtracting -1 from to_read etc.

disassembler.cpp
  check for _library.

jvmciCompilerToVM.cpp, getResolvedJavaType
  base_object is checked for null above, so
  check before printing the name, too.

simpleThresholdPolicy.cpp
  If I understand correctly, if the compile queue is empty,
  no max_task will be found. This is quite unlikely, though.



Useful code improvements:
-------------------------

codeBuffer.cpp
  Above, next_cs is checked for NULL. I would assume this
  check is boguous, but checking below, too, is on the
  safe side.

  The codebuffer should be aligned properly, so the
  first section does not need alignment.
  Make this more clear in the code. Else it looks
  as if prev_dest_cs is dereferenced when NULL.

c1_IR.cpp
  It's not obvious that n->subst() != n if false
  on the first iteration.

c1_LinearScan.cpp
  spill_candidate is used as index to an array in
  _mapping_from.at() and thus should not be negative.

  prev_cmp is dereferenced further down. change to guarantee.

codeCache.cpp
  Use format specifyer to print strings.

methodLiveness.cpp
  Turn assert into guarantee().

oopMap.cpp
  Change loop so that last is always initilaized.
  last gets dereferenced below.


jvmciCodeInstaller.cpp
  DebugInformationRecorder::describe_scope might pass NULL.

arraycopy.cpp
  If c == NULL the condition of the next if will crash.
  Change to guarantee.

bytecodeInfo.cpp
  If callee_method ever had been NULL, we would have
  seen a crach or assertion in post_inlining_event.

callnode.cpp
  Would fail if IR is broken.

compile.cpp
  Don't access arrays at negative indexes.

gcm.cpp
  There must be an LCA, as there should be at least one
  ancestor.

ifnode.cpp
  It's unclear that a proj is found in if (vop == Op_Phi)

indexSet.cpp
  Wrong data type used.

lcm.cpp
  Turn assertion into guarantee, don't access worklist at -1.

loopPredicate.cpp
  limit is dereferenced by _igvn.type(limit)

loopTransform.cpp
  if limit_n == NULL, we will skip the following if and do _igvn.type(limit_n)
  either return false, a guarantee or assert would do, too, I think.

macro.cpp
  igvn.replace_node(fallthroughproj, ctrl) would fail if ctrl == NULL.

memnode.cpp
  igvn is checked for NULL above, too. Protect all uses similarly.

  In the default case load == NULL would be dereferenced below.

node.hpp
  arr is checked for NULL above. Maybe remove (arr == NULL) check above?

output.cpp
  turn assertion into guarantee

parse1.cpp
  turn assertion into guarantee

reg_split.cpp
  turn assertion into guarantee

runtime.cpp
  print "no method" if m == NULL (assigned NULL above).

type.cpp
  One of the if blocks should be reached in any case.
  Change else if into a guarantee.


It's obviously pointless to scan adlc, but fixing the
issues does not hurt either I think.

archDesc.cpp
  Above, node is checked for NULL.

arena.cpp
  memset writes bytes.

dfa.cpp, valid_loc():
  breaks are missing.
dfa.cpp, compute_external():
  Use strncat (should also use snprintf, but we still have
  VS 2013 which does not support that)

filebuff.cpp
  breaks are missing.

formssel.cpp
  name_left was just checked for NULL.
  form2 is checked for NULL above, so check needed here, too.

main.cpp
  Wrong delete operator used.

output_c.cpp
  We cannot continue after this error, as below
  inst_rep_var is dereferenced.
  opc is assigned NULL if local == NULL. (in both cases).