Change `-fprof-late` to insert cost centres after unfolding creation.

The former behaviour of adding cost centres after optimization but
before unfoldings are created is not available via the flag
`prof-late-inline` instead.

I also reduced the overhead of -fprof-late* by pushing the cost centres
into lambdas. This means the cost centres will only account for
execution of functions and not their partial application.

Further I made LATE_CC cost centres it's own CC flavour so they now
won't clash with user defined ones if a user uses the same string for
a custom scc.

LateCC: Don't put cost centres inside constructor workers.

With -fprof-late they are rarely useful as the worker is usually
inlined. Even if the worker is not inlined or we use -fprof-late-linline
they are generally not helpful but bloat compile and run time
significantly. So we just don't add sccs inside constructor workers.

-------------------------
Metric Decrease:
    T13701
-------------------------

5 files changed, 83 insertions, 0 deletions

diff --git a/testsuite/tests/profiling/should_compile/all.T b/testsuite/tests/profiling/should_compile/all.T
index 9e460e153b..1f4ae1430d 100644
--- a/testsuite/tests/profiling/should_compile/all.T
+++ b/testsuite/tests/profiling/should_compile/all.T
@@ -4,6 +4,8 @@ test('prof001', [only_ways(['normal']), req_profiling], compile_and_run, ['-prof
 test('prof002', [only_ways(['normal']), req_profiling], compile_and_run, ['-prof -fprof-cafs'])
 test('prof-late-cc', [only_ways(['normal']), req_profiling], compile, ['-prof -fprof-late'])
 test('prof-late-cc2', [only_ways(['normal']), req_profiling], compile_and_run, ['-prof -fprof-late'])
+test('prof-late-cc3', [only_ways(['normal']), req_profiling, grep_errmsg('scc')], compile, ['-prof -fprof-late -ddump-prep -ddump-simpl -dno-typeable-binds'])
+test('prof-late-cc4', [only_ways(['normal']), req_profiling, grep_errmsg('scc')], compile, ['-prof -fprof-late-inline -ddump-simpl -dno-typeable-binds -O'])
 
 test('T2410', [only_ways(['normal']), req_profiling], compile, ['-O2 -prof -fprof-cafs'])
 test('T5889', [only_ways(['normal']), req_profiling, extra_files(['T5889/A.hs', 'T5889/B.hs'])], multimod_compile, ['A B', '-O -prof -fno-prof-count-entries -v0'])
diff --git a/testsuite/tests/profiling/should_compile/prof-late-cc3.hs b/testsuite/tests/profiling/should_compile/prof-late-cc3.hs
new file mode 100644
index 0000000000..b7badbc216
--- /dev/null
+++ b/testsuite/tests/profiling/should_compile/prof-late-cc3.hs
@@ -0,0 +1,8 @@
+module M where
+
+-- There should be a cost center in core prep output but not in
+-- -ddump-simpl output with -fprof-late
+{-# INLINE doStuff #-}
+doStuff x = do
+    print x
+    return x
diff --git a/testsuite/tests/profiling/should_compile/prof-late-cc3.stderr b/testsuite/tests/profiling/should_compile/prof-late-cc3.stderr
new file mode 100644
index 0000000000..92d9b1cd8f
--- /dev/null
+++ b/testsuite/tests/profiling/should_compile/prof-late-cc3.stderr
@@ -0,0 +1,50 @@
+
+==================== Tidy Core ====================
+Result size of Tidy Core
+  = {terms: 12, types: 16, coercions: 0, joins: 0/0}
+
+-- RHS size: {terms: 11, types: 10, coercions: 0, joins: 0/0}
+doStuff [InlPrag=INLINE (sat-args=1)]
+  :: forall {b}. Show b => b -> IO b
+[GblId, Arity=2, Unf=OtherCon []]
+doStuff
+  = \ (@b_axP) ($dShow_aET :: Show b_axP) (eta_B0 :: b_axP) ->
+      >>
+        @IO
+        GHC.Base.$fMonadIO
+        @()
+        @b_axP
+        (print @b_axP $dShow_aET eta_B0)
+        (return @IO GHC.Base.$fMonadIO @b_axP eta_B0)
+
+
+
+
+==================== CorePrep ====================
+Result size of CorePrep
+  = {terms: 16, types: 20, coercions: 0, joins: 0/2}
+
+-- RHS size: {terms: 15, types: 14, coercions: 0, joins: 0/2}
+M.doStuff [InlPrag=INLINE (sat-args=1)]
+  :: forall {b}. GHC.Show.Show b => b -> GHC.Types.IO b
+[GblId, Arity=2, Unf=OtherCon []]
+M.doStuff
+  = \ (@b_axP)
+      ($dShow_sJs [Occ=OnceL1] :: GHC.Show.Show b_axP)
+      (eta_sJt :: b_axP) ->
+      scctick<doStuff>
+      let {
+        sat_sJv [Occ=Once1] :: GHC.Types.IO b_axP
+        [LclId]
+        sat_sJv
+          = GHC.Base.return
+              @GHC.Types.IO GHC.Base.$fMonadIO @b_axP eta_sJt } in
+      let {
+        sat_sJu [Occ=Once1] :: GHC.Types.IO ()
+        [LclId]
+        sat_sJu = System.IO.print @b_axP $dShow_sJs eta_sJt } in
+      GHC.Base.>>
+        @GHC.Types.IO GHC.Base.$fMonadIO @() @b_axP sat_sJu sat_sJv
+
+
+
diff --git a/testsuite/tests/profiling/should_compile/prof-late-cc4.hs b/testsuite/tests/profiling/should_compile/prof-late-cc4.hs
new file mode 100644
index 0000000000..45d628cd0f
--- /dev/null
+++ b/testsuite/tests/profiling/should_compile/prof-late-cc4.hs
@@ -0,0 +1,5 @@
+module M where
+
+-- There should be a cost center in the -ddump-simpl output
+addStuff x y = do
+    x + y :: Int
diff --git a/testsuite/tests/profiling/should_compile/prof-late-cc4.stderr b/testsuite/tests/profiling/should_compile/prof-late-cc4.stderr
new file mode 100644
index 0000000000..2de7dad615
--- /dev/null
+++ b/testsuite/tests/profiling/should_compile/prof-late-cc4.stderr
@@ -0,0 +1,18 @@
+
+==================== Tidy Core ====================
+Result size of Tidy Core
+  = {terms: 2, types: 3, coercions: 0, joins: 0/0}
+
+-- RHS size: {terms: 1, types: 0, coercions: 0, joins: 0/0}
+addStuff [InlPrag=INLINE (sat-args=2)] :: Int -> Int -> Int
+[GblId,
+ Str=<1!P(L)><1!P(L)>,
+ Cpr=1,
+ Unf=Unf{Src=InlineStable, TopLvl=True, Value=True, ConLike=True,
+         WorkFree=True, Expandable=True,
+         Guidance=ALWAYS_IF(arity=2,unsat_ok=False,boring_ok=True)
+         Tmpl= GHC.Num.$fNumInt_$c+}]
+addStuff = scctick<addStuff> GHC.Num.$fNumInt_$c+
+
+
+