[Test] JIT: Prototype a simple inter-procedural analysis

[hez2010]

Prototype a simple inter-procedural analysis relying on tiering (thus no NAOT support).

1. Do analysis for tier-0 methods and save the analysis result.
2. Use the analysis result while compiling for tier-1.

To simplify the prototype, this only records local num less than 128.

Opening a draft PR to check the cost and benefits.

Example:

static int Test()
{
    var x = new Foo();
    SetValue(x);
    return x.Value;
}

[MethodImpl(MethodImplOptions.NoInlining)]
static void SetValue(Foo obj)
{
    obj.Value = 42;
}

class Foo
{
    public int Value { get; set; }
}

Codegen:

; Assembly listing for method Program:Test():int (Tier0)
; Emitting BLENDED_CODE for X64 with AVX - Windows
; Tier0 code
; rbp based frame
; partially interruptible
; compiling with minopt
; Final local variable assignments
;
;  V00 OutArgs      [V00    ] (  1,  1   )  struct (32) [rsp+0x00]  do-not-enreg[XS] addr-exposed "OutgoingArgSpace"
;  V01 tmp1         [V01    ] (  1,  1   )     ref  ->  [rbp-0x08]  do-not-enreg[] must-init class-hnd exact "NewObj constructor temp" <Program+Foo>
;
; Lcl frame size = 48

G_M58954_IG01:  ;; offset=0x0000
       push     rbp
       sub      rsp, 48
       lea      rbp, [rsp+0x30]
       xor      eax, eax
       mov      qword ptr [rbp-0x08], rax
                                                ;; size=16 bbWeight=1 PerfScore 3.00
G_M58954_IG02:  ;; offset=0x0010
       mov      rcx, 0x7FFAA3B61938      ; Program+Foo
       call     CORINFO_HELP_NEWSFAST
       mov      gword ptr [rbp-0x08], rax
       mov      rcx, gword ptr [rbp-0x08]
       call     [Program+Foo:.ctor():this]
       mov      rcx, gword ptr [rbp-0x08]
       call     [Program:SetValue(Program+Foo)]
       mov      rcx, gword ptr [rbp-0x08]
       cmp      dword ptr [rcx], ecx
       call     [Program+Foo:get_Value():int:this]
       nop
                                                ;; size=52 bbWeight=1 PerfScore 17.50
G_M58954_IG03:  ;; offset=0x0044
       add      rsp, 48
       pop      rbp
       ret
                                                ;; size=6 bbWeight=1 PerfScore 1.75

; Total bytes of code 74, prolog size 16, PerfScore 22.25, instruction count 19, allocated bytes for code 74 (MethodHash=347c19b5) for method Program:Test():int (Tier0)
; ============================================================

; Assembly listing for method Program:Test():int (Instrumented Tier0)
; Emitting BLENDED_CODE for X64 with AVX - Windows
; Instrumented Tier0 code
; rbp based frame
; partially interruptible
; compiling with minopt
; Final local variable assignments
;
;  V00 OutArgs      [V00    ] (  1,  1   )  struct (32) [rsp+0x00]  do-not-enreg[XS] addr-exposed "OutgoingArgSpace"
;  V01 tmp1         [V01    ] (  1,  1   )     ref  ->  [rbp-0x08]  do-not-enreg[] must-init class-hnd exact "NewObj constructor temp" <Program+Foo>
;
; Lcl frame size = 48

G_M58954_IG01:  ;; offset=0x0000
       push     rbp
       sub      rsp, 48
       lea      rbp, [rsp+0x30]
       xor      eax, eax
       mov      qword ptr [rbp-0x08], rax
                                                ;; size=16 bbWeight=1 PerfScore 3.00
G_M58954_IG02:  ;; offset=0x0010
       mov      rcx, 0x7FFAA3B61938      ; Program+Foo
       call     CORINFO_HELP_NEWSFAST
       mov      gword ptr [rbp-0x08], rax
       mov      rcx, gword ptr [rbp-0x08]
       call     [Program+Foo:.ctor():this]
       mov      rcx, gword ptr [rbp-0x08]
       call     [Program:SetValue(Program+Foo)]
       mov      rcx, gword ptr [rbp-0x08]
       cmp      dword ptr [rcx], ecx
       call     [Program+Foo:get_Value():int:this]
       nop
                                                ;; size=52 bbWeight=1 PerfScore 17.50
G_M58954_IG03:  ;; offset=0x0044
       add      rsp, 48
       pop      rbp
       ret
                                                ;; size=6 bbWeight=1 PerfScore 1.75

; Total bytes of code 74, prolog size 16, PerfScore 22.25, instruction count 19, allocated bytes for code 74 (MethodHash=347c19b5) for method Program:Test():int (Instrumented Tier0)
; ============================================================

; Assembly listing for method Program:Test():int (Tier1)
; Emitting BLENDED_CODE for X64 with AVX - Windows
; Tier1 code
; optimized code
; optimized using Synthesized PGO
; rsp based frame
; partially interruptible
; with Synthesized PGO: fgCalledCount is 100
; No PGO data
; 0 inlinees with PGO data; 3 single block inlinees; 0 inlinees without PGO data
; Final local variable assignments
;
;  V00 OutArgs      [V00    ] (  1,  1   )  struct (32) [rsp+0x00]  do-not-enreg[XS] addr-exposed "OutgoingArgSpace"
;* V01 tmp1         [V01    ] (  0,  0   )    long  ->  zero-ref    class-hnd exact "NewObj constructor temp" <Program+Foo>
;  V02 tmp2         [V02    ] (  4,  4   )  struct (16) [rsp+0x28]  do-not-enreg[XSF] addr-exposed "stack allocated ref class temp" <Program+Foo>
;
; Lcl frame size = 56

G_M58954_IG01:  ;; offset=0x0000
       sub      rsp, 56
                                                ;; size=4 bbWeight=1 PerfScore 0.25
G_M58954_IG02:  ;; offset=0x0004
       xor      ecx, ecx
       mov      qword ptr [rsp+0x28], rcx
       mov      dword ptr [rsp+0x30], ecx
       mov      rcx, 0x7FFAA3B61938      ; Program+Foo
       mov      qword ptr [rsp+0x28], rcx
       lea      rcx, [rsp+0x28]
       call     [Program:SetValue(Program+Foo)]
       mov      eax, dword ptr [rsp+0x30]
                                                ;; size=41 bbWeight=1 PerfScore 8.00
G_M58954_IG03:  ;; offset=0x002D
       add      rsp, 56
       ret
                                                ;; size=5 bbWeight=1 PerfScore 1.25

; Total bytes of code 50, prolog size 4, PerfScore 9.50, instruction count 11, allocated bytes for code 50 (MethodHash=347c19b5) for method Program:Test():int (Tier1)
; ============================================================

[dotnet-policy-service]

Tagging subscribers to this area: @JulieLeeMSFT, @jakobbotsch
See info in area-owners.md if you want to be subscribed.