optimizer: Julia-level escape analysis

This commit ports [EscapeAnalysis.jl](https://github.com/aviatesk/EscapeAnalysis.jl) into Julia base.
You can find the documentation of this escape analysis at [this GitHub page](https://aviatesk.github.io/EscapeAnalysis.jl/dev/)[^1].

[^1]: The same documentation will be included into Julia's developer
      documentation by this commit.

This escape analysis will hopefully be an enabling technology for various
memory-related optimizations at Julia's high level compilation pipeline.
Possible target optimization includes mutable ϕ-node aware SROA,
`mutating_arrayfreeze` optimization (#42465), stack allocation of mutables,
dead array elimination, finalizer elision and so on[^2].

[^2]: It would be also interesting if LLVM-level optimizations can consume
      some IPO information derived by this escape analysis to broaden
      optimization possibilities.

The primary motivation for porting EA in this PR is to check its impact
on latency as well as to get feedbacks from a broader range of developers.
If there are no serious problems, EA will be used by #42465 for `mutating_arrayfreeze`
optimization for `ImmutableArray` construction.

This commit simply defines EA inside Julia base compiler and enables the
existing test suite with it. The plan is to validate EA's accuracy and
correctness and to check the latency impact by running EA before the
existing SROA pass and checking if it can detect all objects eliminated
by the SROA pass as "SROA-eliminatable".

base/Base.jl

@@ -397,6 +397,9 @@ include("util.jl")
 
 include("asyncmap.jl")
 
+include("compiler/ssair/EscapeAnalysis/EAUtils.jl")
+using .EAUtils: code_escapes
+
 # deprecated functions
 include("deprecated.jl")
 

base/compiler/bootstrap.jl

@@ -14,7 +14,7 @@ let
     fs = Any[
         # we first create caches for the optimizer, because they contain many loop constructions
         # and they're better to not run in interpreter even during bootstrapping
-        run_passes,
+        analyze_escapes, run_passes,
         # then we create caches for inference entries
         typeinf_ext, typeinf, typeinf_edge,
     ]

base/compiler/compiler.jl

@@ -95,6 +95,8 @@ ntuple(f, n) = (Any[f(i) for i = 1:n]...,)
 
 # core docsystem
 include("docs/core.jl")
+import Core.Compiler.CoreDocs
+Core.atdoc!(CoreDocs.docm)
 
 # sorting
 function sort end

base/compiler/optimize.jl

@@ -1,5 +1,45 @@
 # This file is a part of Julia. License is MIT: https://julialang.org/license
 
+#############
+# constants #
+#############
+
+# The slot has uses that are not statically dominated by any assignment
+# This is implied by `SLOT_USEDUNDEF`.
+# If this is not set, all the uses are (statically) dominated by the defs.
+# In particular, if a slot has `AssignedOnce && !StaticUndef`, it is an SSA.
+const SLOT_STATICUNDEF  = 1 # slot might be used before it is defined (structurally)
+const SLOT_ASSIGNEDONCE = 16 # slot is assigned to only once
+const SLOT_USEDUNDEF    = 32 # slot has uses that might raise UndefVarError
+# const SLOT_CALLED      = 64
+
+# NOTE make sure to sync the flag definitions below with julia.h and `jl_code_info_set_ir` in method.c
+
+const IR_FLAG_NULL        = 0x00
+# This statement is marked as @inbounds by user.
+# Ff replaced by inlining, any contained boundschecks may be removed.
+const IR_FLAG_INBOUNDS    = 0x01 << 0
+# This statement is marked as @inline by user
+const IR_FLAG_INLINE      = 0x01 << 1
+# This statement is marked as @noinline by user
+const IR_FLAG_NOINLINE    = 0x01 << 2
+const IR_FLAG_THROW_BLOCK = 0x01 << 3
+# This statement may be removed if its result is unused. In particular it must
+# thus be both pure and effect free.
+const IR_FLAG_EFFECT_FREE = 0x01 << 4
+
+# known to be always effect-free (in particular nothrow)
+const _PURE_BUILTINS = Any[tuple, svec, ===, typeof, nfields]
+
+# known to be effect-free if the are nothrow
+const _PURE_OR_ERROR_BUILTINS = [
+    fieldtype, apply_type, isa, UnionAll,
+    getfield, arrayref, const_arrayref, arraysize, isdefined, Core.sizeof,
+    Core.kwfunc, Core.ifelse, Core._typevar, (<:)
+]
+
+const TOP_TUPLE = GlobalRef(Core, :tuple)
+
 #####################
 # OptimizationState #
 #####################
@@ -52,7 +92,10 @@ function inlining_policy(interp::AbstractInterpreter, @nospecialize(src), stmt_f
     return nothing
 end
 
+function argextype end # imported by EscapeAnalysis
 include("compiler/ssair/driver.jl")
+using .EscapeAnalysis
+import .EscapeAnalysis: EscapeState
 
 mutable struct OptimizationState
     linfo::MethodInstance
@@ -121,46 +164,6 @@ function ir_to_codeinf!(opt::OptimizationState)
     return src
 end
 
-#############
-# constants #
-#############
-
-# The slot has uses that are not statically dominated by any assignment
-# This is implied by `SLOT_USEDUNDEF`.
-# If this is not set, all the uses are (statically) dominated by the defs.
-# In particular, if a slot has `AssignedOnce && !StaticUndef`, it is an SSA.
-const SLOT_STATICUNDEF  = 1 # slot might be used before it is defined (structurally)
-const SLOT_ASSIGNEDONCE = 16 # slot is assigned to only once
-const SLOT_USEDUNDEF    = 32 # slot has uses that might raise UndefVarError
-# const SLOT_CALLED      = 64
-
-# NOTE make sure to sync the flag definitions below with julia.h and `jl_code_info_set_ir` in method.c
-
-const IR_FLAG_NULL        = 0x00
-# This statement is marked as @inbounds by user.
-# Ff replaced by inlining, any contained boundschecks may be removed.
-const IR_FLAG_INBOUNDS    = 0x01 << 0
-# This statement is marked as @inline by user
-const IR_FLAG_INLINE      = 0x01 << 1
-# This statement is marked as @noinline by user
-const IR_FLAG_NOINLINE    = 0x01 << 2
-const IR_FLAG_THROW_BLOCK = 0x01 << 3
-# This statement may be removed if its result is unused. In particular it must
-# thus be both pure and effect free.
-const IR_FLAG_EFFECT_FREE = 0x01 << 4
-
-# known to be always effect-free (in particular nothrow)
-const _PURE_BUILTINS = Any[tuple, svec, ===, typeof, nfields]
-
-# known to be effect-free if the are nothrow
-const _PURE_OR_ERROR_BUILTINS = [
-    fieldtype, apply_type, isa, UnionAll,
-    getfield, arrayref, const_arrayref, arraysize, isdefined, Core.sizeof,
-    Core.kwfunc, Core.ifelse, Core._typevar, (<:)
-]
-
-const TOP_TUPLE = GlobalRef(Core, :tuple)
-
 #########
 # logic #
 #########
@@ -514,7 +517,8 @@ function run_passes(ci::CodeInfo, sv::OptimizationState)
     @timeit "Inlining"  ir = ssa_inlining_pass!(ir, ir.linetable, sv.inlining, ci.propagate_inbounds)
     # @timeit "verify 2" verify_ir(ir)
     @timeit "compact 2" ir = compact!(ir)
-    @timeit "SROA"      ir = sroa_pass!(ir)
+    nargs = let def = sv.linfo.def; isa(def, Method) ? Int(def.nargs) : 0; end
+    @timeit "SROA"      ir = sroa_pass!(ir, nargs)
     @timeit "ADCE"      ir = adce_pass!(ir)
     @timeit "type lift" ir = type_lift_pass!(ir)
     @timeit "compact 3" ir = compact!(ir)