JIT: Update type when return temp is freshly created #111948

hez2010 · 2025-01-29T08:13:08Z

Together with #110827 and things we have done for #108913 so far, this unblocks us to fully de-abstract array enumerators even without PGO.

Contributes to #108913.

class ArrayTest
{
    static readonly int[] s_ro_array;
    static int[] newArr() => new int[512];
    static ArrayTest()
    {
        s_ro_array = newArr();
    }
    static IEnumerable<int> get_static_readonly_array() => s_ro_array;
    public int foreach_static_readonly_array_via_interface_property()
    {
        var e = get_static_readonly_array();
        int sum = 0;
        foreach (int i in e) sum += i;
        return sum;
    }
}

Before:

; Assembly listing for method ArrayTest:foreach_static_readonly_array_via_interface_property():int:this (FullOpts)

G_M22676_IG01:  ;; offset=0x0000
       push     rbp
       push     rdi
       push     rsi
       push     rbx
       sub      rsp, 56
       lea      rbp, [rsp+0x50]
       mov      qword ptr [rbp-0x30], rsp
                                                ;; size=17 bbWeight=1 PerfScore 5.75
G_M22676_IG02:  ;; offset=0x0011
       xor      ebx, ebx
       mov      rcx, 0x1A291800DF0      ; const ptr
       mov      rsi, gword ptr [rcx]
       mov      rcx, 0x7FFD07CD9AF8      ; System.SZGenericArrayEnumerator`1[int]
       call     CORINFO_HELP_NEWSFAST
       mov      rdi, rax
       mov      rcx, 0x200FFFFFFFF
       mov      qword ptr [rdi+0x08], rcx
       lea      rcx, bword ptr [rdi+0x10]
       mov      rdx, rsi
       call     CORINFO_HELP_ASSIGN_REF
       mov      gword ptr [rbp-0x20], rdi
                                                ;; size=63 bbWeight=1 PerfScore 8.00
G_M22676_IG03:  ;; offset=0x0050
       mov      rcx, rdi
       mov      r11, 0x7FFD06C80370      ; code for System.Collections.IEnumerator:MoveNext():ubyte:this
       call     [r11]System.Collections.IEnumerator:MoveNext():ubyte:this
       test     eax, eax
       je       SHORT G_M22676_IG05
                                                ;; size=20 bbWeight=8 PerfScore 38.00
G_M22676_IG04:  ;; offset=0x0064
       mov      rcx, rdi
       mov      r11, 0x7FFD06C80378      ; code for System.Collections.Generic.IEnumerator`1[int]:get_Current():int:this
       call     [r11]System.Collections.Generic.IEnumerator`1[int]:get_Current():int:this
       add      ebx, eax
       jmp      SHORT G_M22676_IG03
                                                ;; size=20 bbWeight=4 PerfScore 23.00
G_M22676_IG05:  ;; offset=0x0078
       mov      rcx, rdi
       mov      r11, 0x7FFD06C80380      ; code for System.IDisposable:Dispose():this
       call     [r11]System.IDisposable:Dispose():this
       mov      eax, ebx
                                                ;; size=18 bbWeight=1 PerfScore 3.75
G_M22676_IG06:  ;; offset=0x008A
       add      rsp, 56
       pop      rbx
       pop      rsi
       pop      rdi
       pop      rbp
       ret
                                                ;; size=9 bbWeight=1 PerfScore 3.25
G_M22676_IG07:  ;; offset=0x0093
       push     rbp
       push     rdi
       push     rsi
       push     rbx
       sub      rsp, 40
       mov      rbp, qword ptr [rcx+0x20]
       mov      qword ptr [rsp+0x20], rbp
       lea      rbp, [rbp+0x50]
                                                ;; size=21 bbWeight=0 PerfScore 0.00
G_M22676_IG08:  ;; offset=0x00A8
       mov      rcx, gword ptr [rbp-0x20]
       mov      r11, 0x7FFD06C80380      ; code for System.IDisposable:Dispose():this
       call     [r11]System.IDisposable:Dispose():this
       nop
                                                ;; size=18 bbWeight=0 PerfScore 0.00
G_M22676_IG09:  ;; offset=0x00BA
       add      rsp, 40
       pop      rbx
       pop      rsi
       pop      rdi
       pop      rbp
       ret
                                                ;; size=9 bbWeight=0 PerfScore 0.00

After:

; Assembly listing for method ArrayTest:foreach_static_readonly_array_via_interface_property():int:this (FullOpts)

G_M22676_IG01:  ;; offset=0x0000
       push     rbp
       mov      rbp, rsp
                                                ;; size=4 bbWeight=1 PerfScore 1.25
G_M22676_IG02:  ;; offset=0x0004
       xor      eax, eax
       mov      rcx, 0x205D7000DF0      ; const ptr
       mov      rcx, gword ptr [rcx]
       mov      edx, 0xFFFFFFFF
       jmp      SHORT G_M22676_IG04
       align    [6 bytes for IG03]
                                                ;; size=28 bbWeight=1 PerfScore 4.75
G_M22676_IG03:  ;; offset=0x0020
       mov      edx, edx
       add      eax, dword ptr [rcx+4*rdx+0x10]
                                                ;; size=6 bbWeight=16 PerfScore 52.00
G_M22676_IG04:  ;; offset=0x0026
       inc      edx
       cmp      edx, 512
       jb       SHORT G_M22676_IG03
                                                ;; size=10 bbWeight=8 PerfScore 12.00
G_M22676_IG05:  ;; offset=0x0030
       pop      rbp
       ret
                                                ;; size=2 bbWeight=1 PerfScore 1.50

dotnet-policy-service · 2025-01-29T08:13:52Z

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

hez2010 · 2025-01-29T08:43:53Z

Benchmark result:

Method	Job	Mean	Ratio	Allocated
foreach_static_readonly_array	Main	104.15 ns	1.00	-
foreach_static_readonly_array_in_loop	Main	103.18 ns	0.99	-
foreach_static_readonly_array_via_local	Main	102.56 ns	0.98	-
foreach_static_readonly_array_via_interface	Main	223.04 ns	2.14	-
foreach_static_readonly_array_via_interface_property	Main	1,593.66 ns	15.30	32 B
foreach_static_readonly_array_via_interface_property_in_loop	Main	1,693.10 ns	16.26	32 B
foreach_static_array	Main	104.16 ns	1.00	-
foreach_static_array_via_local	Main	104.03 ns	1.00	-
foreach_static_array_via_interface	Main	222.26 ns	2.13	-
foreach_static_array_via_interface_property	Main	1,695.46 ns	16.28	32 B
foreach_static_array_via_interface_property_in_loop	Main	1,683.28 ns	16.16	32 B
foreach_member_array	Main	105.42 ns	1.01	-
foreach_member_array_via_local	Main	103.96 ns	1.00	-
foreach_member_array_via_interface	Main	221.90 ns	2.13	-
foreach_member_array_via_interface_property	Main	1,669.04 ns	16.03	32 B
foreach_member_array_via_interface_property_in_loop	Main	1,674.84 ns	16.08	32 B
foreach_opaque_array_via_interface	Main	1,673.81 ns	16.07	32 B
foreach_opaque_array_via_interface_in_loop	Main	1,675.47 ns	16.09	32 B
sum_static_array_via_local	Main	29.66 ns	0.28	-

foreach_static_readonly_array	PR	102.92 ns	1.00	-
foreach_static_readonly_array_in_loop	PR	102.83 ns	1.00	-
foreach_static_readonly_array_via_local	PR	102.52 ns	1.00	-
foreach_static_readonly_array_via_interface	PR	221.10 ns	2.15	-
foreach_static_readonly_array_via_interface_property	PR	221.28 ns	2.15	-
foreach_static_readonly_array_via_interface_property_in_loop	PR	219.25 ns	2.13	32 B
foreach_static_array	PR	103.74 ns	1.01	-
foreach_static_array_via_local	PR	104.26 ns	1.01	-
foreach_static_array_via_interface	PR	222.43 ns	2.16	-
foreach_static_array_via_interface_property	PR	222.20 ns	2.16	-
foreach_static_array_via_interface_property_in_loop	PR	222.47 ns	2.16	32 B
foreach_member_array	PR	104.37 ns	1.01	-
foreach_member_array_via_local	PR	104.69 ns	1.02	-
foreach_member_array_via_interface	PR	222.59 ns	2.16	-
foreach_member_array_via_interface_property	PR	222.05 ns	2.16	-
foreach_member_array_via_interface_property_in_loop	PR	221.64 ns	2.15	32 B
foreach_opaque_array_via_interface	PR	1,285.20 ns	12.49	32 B
foreach_opaque_array_via_interface_in_loop	PR	1,285.92 ns	12.49	32 B
sum_static_array_via_local	PR	31.11 ns	0.30	-

hez2010 · 2025-01-29T09:28:47Z

@MihuBot

src/coreclr/jit/fgopt.cpp

src/coreclr/jit/fgbasic.cpp

src/coreclr/jit/fgopt.cpp

Co-authored-by: Jakob Botsch Nielsen <[email protected]>

hez2010 · 2025-01-29T09:57:30Z

@MihuBot

jakobbotsch · 2025-01-29T12:32:30Z

Together with #110827, this allows us to fully de-abstract array enumerators even without PGO.

Just noting that this PR is handling the case where the type is obtained via inlining; the actual deabstraction when the type is known was done as part of #108913. This unlocks that optimization for more cases.

hez2010 · 2025-01-29T14:30:39Z

The test seems only fail on Linux-musl-arm, and I'm not able to reproduce it in my local environment (x64). From the test log I'm not sure whether it's introduced by this PR because I don't get how could it be related, or it's an existing bug that gets exposed by this PR.
I will try updating the branch for a test rerun.
Does this ring any bell?

hez2010 · 2025-01-29T15:59:42Z

The test seems only fail on Linux-musl-arm, and I'm not able to reproduce it in my local environment (x64). From the test log I'm not sure whether it's introduced by this PR because I don't get how could it be related, or it's an existing bug that gets exposed by this PR. I will try updating the branch for a test rerun. Does this ring any bell?

After updating to the latest branch, the failed tests passed this time.
Seems that it's indeed unrelated to this change.

src/coreclr/jit/lclvars.cpp

src/coreclr/jit/fgopt.cpp

AndyAyersMS · 2025-01-29T16:06:10Z

Also we should run jitstress and maybe some pgo stress on this, so it gets exposed to different inlining patterns.

hez2010 · 2025-01-29T18:49:45Z

CI is passing. (The remaining osx-arm64 two are likely to timeout like all other recent PRs).

Not too much interesting results from Diffs due to a lot of missing contexts. TP impact is not observable.

Diffs from MihuBot: https://gist.github.com/MihuBot/1674fcd3253ccef9fa1b24e541a0ab1d

hez2010 · 2025-01-30T08:53:41Z

Seems that the issue only happening on dynamic for COM objects. Deployed a workaround for this. (Or maybe we can ignore this one because it's too niche?)
PTAL.

src/coreclr/jit/importer.cpp

This reverts commit e94839b.

hez2010 · 2025-01-30T15:37:16Z

Okay. After a few hours of investigation, I scoped the issue to DynamicMetaObject and got this result with some logging points inserted by me.
When we are trying to inline System.Dynamic.DynamicMetaObject:Create into System.Dynamic.DynamicMetaObjectBinder:Bind, we set the spilled return temp V67 to exact, the logs are:

Setting return spill temp V67 to exact, when inline System.Dynamic.DynamicMetaObject:Create into System.Dynamic.DynamicMetaObjectBinder:Bind
--------
> Caller's block containing inline root STMT00250:
STMT00249 ( INL45 @ 0x015[E-] ... ??? ) <- INLRT @ 0x0AA[E-]
               [000873] DA-XG------                         *  STORE_LCL_VAR ref    V66 tmp57
               [000870] n--XG------                         \--*  IND       ref
               [000869] ---XG------                            \--*  INDEX_ADDR byref ref[]
               [000867] -----------                               +--*  LCL_VAR   ref    V01 arg1
               [000868] -----------                               \--*  LCL_VAR   int    V65 tmp56

STMT00250 ( INL45 @ ??? ... ??? ) <- INLRT @ 0x0AA[E-]
               [000876] I-C-G------                         *  CALL      ref    System.Dynamic.DynamicMetaObject:Create(System.Object,System.Linq.Expressions.Expression):System.Dynamic.DynamicMetaObject (exactContextHandle=0x00007FFAC8283729)
               [000874] ----------- arg0                    +--*  LCL_VAR   ref    V66 tmp57
               [000889] --CXG------ arg1                    \--*  CALLV stub ref    System.Collections.Generic.IList`1[System.__Canon]:get_Item(int):System.__Canon:this
               [000888] n--XG------ this                       +--*  IND       ref
               [000887] ---X-------                            |  \--*  FIELD_ADDR byref  System.Collections.ObjectModel.ReadOnlyCollection`1[System.__Canon]:list
               [000051] -----------                            |     \--*  LCL_VAR   ref    V02 arg2
               [000871] ----------- arg1                       \--*  LCL_VAR   int    V65 tmp56

STMT00251 ( INL45 @ ??? ... ??? ) <- INLRT @ 0x0AA[E-]
               [000879] --CXG------                         *  CALL help void   CORINFO_HELP_ARRADDR_ST
               [000863] ----------- arg0                    +--*  LCL_VAR   ref    V63 tmp54
               [000878] ----------- arg1                    +--*  CAST      long <- int
               [000866] -----------                         |  \--*  SUB       int
               [000864] -----------                         |     +--*  LCL_VAR   int    V65 tmp56
               [000865] -----------                         |     \--*  CNS_INT   int    1
               [000877] --C-------- arg2                    \--*  RET_EXPR  ref   (for [000876]) -> [000945]

STMT00252 ( INL45 @ 0x029[E-] ... ??? ) <- INLRT @ 0x0AA[E-]
               [000883] DA---------                         *  STORE_LCL_VAR int    V65 tmp56
               [000882] -----------                         \--*  ADD       int
               [000880] -----------                            +--*  LCL_VAR   int    V65 tmp56
               [000881] -----------                            \--*  CNS_INT   int    1
--------
> Inlinee blocks:
STMT00254 ( 0x000[E-] ... ??? ) <- INL45 @ ??? <- INLRT @ 0x0AA[E-]
               [000893] I-C-G------                         *  CALL      void   System.ArgumentNullException:ThrowIfNull(System.Object,System.String) (exactContextHandle=0x00007FFAC73E7249)
               [000891] ----------- arg0                    +--*  LCL_VAR   ref    V68 tmp59
               [000892] ----------- arg1                    \--*  CNS_STR   ref   <string constant>

STMT00255 ( 0x00B[E-] ... ??? ) <- INL45 @ ??? <- INLRT @ 0x0AA[E-]
               [000896] DAC-G------                         *  STORE_LCL_VAR ref    V69 tmp60
               [000895] --C-G------                         \--*  CALL help ref    CORINFO_HELP_ISINSTANCEOFINTERFACE
               [000894] H------N--- arg0                       +--*  CNS_INT(h) long   0x7ffac8287b50 class System.Dynamic.IDynamicMetaObjectProvider
               [000874] ----------- arg1                       \--*  LCL_VAR   ref    V66 tmp57

STMT00256 ( 0x012[E-] ... ??? ) <- INL45 @ ??? <- INLRT @ 0x0AA[E-]
               [000900] -----------                         *  JTRUE     void
               [000899] -----------                         \--*  EQ        int
               [000897] -----------                            +--*  LCL_VAR   ref    V69 tmp60
               [000898] -----------                            \--*  CNS_INT   ref    null

STMT00270 ( 0x015[E-] ... ??? ) <- INL45 @ ??? <- INLRT @ 0x0AA[E-]
               [000951] ---X-------                         *  JTRUE     void
               [000950] ---X-------                         \--*  NE        int
               [000949] H----------                            +--*  CNS_INT(h) long   0x7ffac8507b30 class Microsoft.CSharp.RuntimeBinder.ComInterop.IDispatchComObject
               [000948] #--X-------                            \--*  IND       long
               [000947] -----------                               \--*  LCL_VAR   ref    V69 tmp60

STMT00271 ( ??? ... ??? )
               [000953] DA---------                         *  STORE_LCL_VAR ref    V73 tmp64
               [000952] -----------                         \--*  LCL_VAR   ref    V69 tmp60

STMT00272 ( 0x015[E-] ... ??? ) <- INL45 @ ??? <- INLRT @ 0x0AA[E-]
               [000954] I-C-G------                         *  CALL nullcheck ref    Microsoft.CSharp.RuntimeBinder.ComInterop.IDispatchComObject:System.Dynamic.IDynamicMetaObjectProvider.GetMetaObject(System.Linq.Expressions.Expression):System.Dynamic.DynamicMetaObject:this (exactContextHandle=0x00007FFAC8507B31)
               [000957] ----------- this                    +--*  LCL_VAR   ref    V73 tmp64
               [000956] ----------- arg1                    \--*  LCL_VAR   ref    V68 tmp59

STMT00273 ( ??? ... ??? )
               [000959] DAC--------                         *  STORE_LCL_VAR ref    V72 tmp63
               [000958] --C--------                         \--*  RET_EXPR  ref   (for [000954])

STMT00274 ( 0x015[E-] ... ??? ) <- INL45 @ ??? <- INLRT @ 0x0AA[E-]
               [000960] DAC-G------                         *  STORE_LCL_VAR ref    V72 tmp63
               [000915] --C-G------                         \--*  CALLV stub ref    System.Dynamic.IDynamicMetaObjectProvider:GetMetaObject(System.Linq.Expressions.Expression):System.Dynamic.DynamicMetaObject:this
               [000913] ----------- this                       +--*  LCL_VAR   ref    V69 tmp60
               [000914] ----------- arg1                       \--*  LCL_VAR   ref    V68 tmp59

STMT00261 ( 0x015[E-] ... ??? ) <- INL45 @ ??? <- INLRT @ 0x0AA[E-]
               [000917] DAC--------                         *  STORE_LCL_VAR ref    V71 tmp62
               [000946] -----------                         \--*  LCL_VAR   ref    V72 tmp63

STMT00262 ( 0x01D[E-] ... ??? ) <- INL45 @ ??? <- INLRT @ 0x0AA[E-]
               [000921] -----------                         *  JTRUE     void
               [000920] -----------                         \--*  EQ        int
               [000918] -----------                            +--*  LCL_VAR   ref    V71 tmp62
               [000919] -----------                            \--*  CNS_INT   ref    null

STMT00264 ( 0x020[E-] ... ??? ) <- INL45 @ ??? <- INLRT @ 0x0AA[E-]
               [000927] I-C-G------                         *  CALL nullcheck int    System.Dynamic.DynamicMetaObject:get_HasValue():ubyte:this (exactContextHandle=0x00007FFAC8283729)
               [000926] ----------- this                    \--*  LCL_VAR   ref    V71 tmp62

STMT00265 ( 0x020[E-] ... ??? ) <- INL45 @ ??? <- INLRT @ 0x0AA[E-]
               [000931] --C--------                         *  JTRUE     void
               [000930] --C--------                         \--*  EQ        int
               [000928] --C--------                            +--*  RET_EXPR  int   (for [000927])
               [000929] -----------                            \--*  CNS_INT   int    0

STMT00266 ( 0x028[E-] ... ??? ) <- INL45 @ ??? <- INLRT @ 0x0AA[E-]
               [000936] --C-G------                         *  JTRUE     void
               [000935] --C-G------                         \--*  EQ        int
               [000933] --C-G------                            +--*  CALL nullcheck ref    System.Dynamic.DynamicMetaObject:get_Value():System.Object:this
               [000932] ----------- this                       |  \--*  LCL_VAR   ref    V71 tmp62
               [000934] -----------                            \--*  CNS_INT   ref    null

STMT00267 ( 0x030[E-] ... ??? ) <- INL45 @ ??? <- INLRT @ 0x0AA[E-]
               [000938] I-C-G------                         *  CALL nullcheck ref    System.Dynamic.DynamicMetaObject:get_Expression():System.Linq.Expressions.Expression:this (exactContextHandle=0x00007FFAC8283729)
               [000937] ----------- this                    \--*  LCL_VAR   ref    V71 tmp62

STMT00268 ( 0x030[E-] ... ??? ) <- INL45 @ ??? <- INLRT @ 0x0AA[E-]
               [000942] --C--------                         *  JTRUE     void
               [000941] --C--------                         \--*  EQ        int
               [000939] --C--------                            +--*  RET_EXPR  ref   (for [000938])
               [000940] -----------                            \--*  LCL_VAR   ref    V68 tmp59

STMT00263 ( 0x039[E-] ... ??? ) <- INL45 @ ??? <- INLRT @ 0x0AA[E-]
               [000925] --CXG------                         *  CALL help void   CORINFO_HELP_THROW
               [000924] --CXG------ arg0                    \--*  CALL      ref    System.Linq.Expressions.Error:InvalidMetaObjectCreated(System.Object):System.InvalidOperationException
               [000923] --CX------- arg0                       \--*  INTRINSIC ref    objGetType
               [000922] -----------                               \--*  LCL_VAR   ref    V69 tmp60

STMT00269 ( 0x045[E-] ... ??? ) <- INL45 @ ??? <- INLRT @ 0x0AA[E-]
               [000944] DA---------                         *  STORE_LCL_VAR ref    V67 tmp58
               [000943] -----------                         \--*  LCL_VAR   ref    V71 tmp62

STMT00257 ( 0x047[E-] ... ??? ) <- INL45 @ ??? <- INLRT @ 0x0AA[E-]
               [000907] DA---------                         *  STORE_LCL_VAR ref    V70 tmp61
               [000906] -----------                         \--*  ALLOCOBJ  ref
               [000905] H----------                            \--*  CNS_INT(h) long   0x7ffac8283728 class System.Dynamic.DynamicMetaObject

STMT00258 ( ??? ... ??? ) <- INL45 @ ??? <- INLRT @ 0x0AA[E-]
               [000909] I-C-G------                         *  CALL      void   System.Dynamic.DynamicMetaObject:.ctor(System.Linq.Expressions.Expression,System.Dynamic.BindingRestrictions,System.Object):this (exactContextHandle=0x00007FFAC8283729)
               [000908] ----------- this                    +--*  LCL_VAR   ref    V70 tmp61
               [000901] ----------- arg1                    +--*  LCL_VAR   ref    V68 tmp59
               [000903] #---------- arg2                    +--*  IND       ref
               [000902] H----------                         |  \--*  CNS_INT(h) long   0x18aa4400f18 const ptr Fseq[Empty]
               [000904] ----------- arg3                    \--*  LCL_VAR   ref    V66 tmp57

STMT00259 ( 0x053[--] ... ??? ) <- INL45 @ ??? <- INLRT @ 0x0AA[E-]
               [000911] DA---------                         *  STORE_LCL_VAR ref    V67 tmp58
               [000910] -----------                         \--*  LCL_VAR   ref    V70 tmp61

So here V67 can either be DynamicMetaObject or an object obtained from System.Dynamic.IDynamicMetaObjectProvider:GetMetaObject. They have the same type but the former is exact, while the later is from a virtual call so it's not exact.
I think this is why we are getting the test failure, and we don't get test failures without GDV because we don't even try inlining System.Dynamic.DynamicMetaObject:Create into System.Dynamic.DynamicMetaObjectBinder:Bind.

cc: @jakobbotsch

hez2010 · 2025-01-30T15:46:19Z

So I think the issue is in the importer:

runtime/src/coreclr/jit/importer.cpp

Lines 11066 to 11087 in 7afd807

    
           if (info.compRetType == TYP_REF) 
        
           { 
        
               bool                 isExact      = false; 
        
               bool                 isNonNull    = false; 
        
               CORINFO_CLASS_HANDLE returnClsHnd = gtGetClassHandle(op2, &isExact, &isNonNull); 
        
               if (inlRetExpr->gtSubstExpr == nullptr) 
        
               { 
        
                   // This is the first return, so best known type is the type 
        
                   // of this return value. 
        
                   impInlineInfo->retExprClassHnd        = returnClsHnd; 
        
                   impInlineInfo->retExprClassHndIsExact = isExact; 
        
               } 
        
               else if (impInlineInfo->retExprClassHnd != returnClsHnd) 
        
               { 
        
                   // This return site type differs from earlier seen sites, 
        
                   // so reset the info and we'll fall back to using the method's 
        
                   // declared return type for the return spill temp. 
        
                   impInlineInfo->retExprClassHnd        = nullptr; 
        
                   impInlineInfo->retExprClassHndIsExact = false; 
        
               } 
        
           }

We didn't check the difference of the type exactness here. If we see multiple sites have different exactness, we should reset it.
What do you think?

jakobbotsch · 2025-01-30T15:47:04Z

Yes, that looks like the actual problem to me as well.

hez2010 · 2025-01-30T15:48:28Z

Tests should pass this time. PTAL.

jakobbotsch · 2025-01-30T15:48:52Z

/azp run runtime-coreclr jitstress, runtime-coreclr libraries-jitstress, runtime-coreclr pgostress

azure-pipelines · 2025-01-30T15:49:09Z

Azure Pipelines successfully started running 3 pipeline(s).

hez2010 · 2025-01-30T19:27:58Z

Test failure seems unrelated.

src/coreclr/jit/importer.cpp

hez2010 · 2025-01-31T13:51:58Z

Failures due to #112021.

AndyAyersMS

Thanks, this looks good.

AndyAyersMS · 2025-01-31T16:19:36Z

Not clear why build analysis is red here. I think this change is good. but let's wait for the current CI problem to get resolved and merge up and run the tests once more.

hez2010 · 2025-01-31T19:55:34Z

It seems that the infra issue is causing every PR to fail the CI. Looking at the logs it seems that the tests are passing, but the test environment is lacking some python dependencies so that the result was failed to upload correctly.

AndyAyersMS · 2025-01-31T20:30:23Z

Yep. We are expecting the CI issues to be fixed fairly soon, and I will trigger new tests when that happens.

AndyAyersMS · 2025-01-31T21:13:29Z

Should be fixed...

AndyAyersMS · 2025-02-01T02:02:19Z

/azp run runtime-coreclr pgostress

azure-pipelines · 2025-02-01T02:02:35Z

Azure Pipelines successfully started running 1 pipeline(s).

hez2010 · 2025-02-01T08:48:51Z

Test failures are unrelated.

AndyAyersMS · 2025-02-01T16:12:04Z

Thanks again @hez2010 !

Teach the JIT about System.SZArrayHelper.GetEnumerator

39a4204

dotnet-issue-labeler bot added the area-CodeGen-coreclr CLR JIT compiler in src/coreclr/src/jit and related components such as SuperPMI label Jan 29, 2025

dotnet-policy-service bot added the community-contribution Indicates that the PR has been added by a community member label Jan 29, 2025

hez2010 added 2 commits January 29, 2025 18:23

A better approach

6ec367f

Format JIT

126a042

MihuBot mentioned this pull request Jan 29, 2025

[JitDiff X64] [hez2010] JIT: Teach the inliner about NI_System_SZArrayHelper_GetEnumerator MihuBot/runtime-utils#961

Open

jakobbotsch reviewed Jan 29, 2025

View reviewed changes

src/coreclr/jit/fgopt.cpp Outdated Show resolved Hide resolved

jakobbotsch reviewed Jan 29, 2025

View reviewed changes

src/coreclr/jit/fgbasic.cpp Outdated Show resolved Hide resolved

Use lvaInlineeReturnSpillTempFreshlyCreated

0e91fd0

hez2010 changed the title ~~JIT: Teach the inliner about NI_System_SZArrayHelper_GetEnumerator~~ JIT: Update type when return temp is freshly created Jan 29, 2025

jakobbotsch reviewed Jan 29, 2025

View reviewed changes

src/coreclr/jit/fgopt.cpp Outdated Show resolved Hide resolved

hez2010 and others added 3 commits January 29, 2025 18:53

Update src/coreclr/jit/fgopt.cpp

c43cf22

Co-authored-by: Jakob Botsch Nielsen <[email protected]>

Update assertion

1c45b3e

Nit

02c808b

MihuBot mentioned this pull request Jan 29, 2025

[JitDiff X64] [hez2010] JIT: Update type when return temp is freshly created MihuBot/runtime-utils#962

Open

Nit

3ae7287

Merge branch 'main' into deabstract-enumerator

193451d

AndyAyersMS reviewed Jan 29, 2025

View reviewed changes

src/coreclr/jit/lclvars.cpp Outdated Show resolved Hide resolved

src/coreclr/jit/fgopt.cpp Show resolved Hide resolved

hez2010 added 2 commits January 30, 2025 01:18

Address review feedbacks

f5fd84e

Make the check always available

22bd9cf

hez2010 mentioned this pull request Jan 29, 2025

JIT: De-abstraction in .NET 10 #108913

Open

Deploy a fix for COM objects

e94839b

Merge branch 'main' into deabstract-enumerator

c4351c2

jakobbotsch reviewed Jan 30, 2025

View reviewed changes

src/coreclr/jit/importer.cpp Outdated Show resolved Hide resolved

Revert "Deploy a fix for COM objects"

7844b8f

This reverts commit e94839b.

hez2010 force-pushed the deabstract-enumerator branch from 90c769a to 7844b8f Compare January 30, 2025 13:46

The real fix

3c5a41b

AndyAyersMS reviewed Jan 30, 2025

View reviewed changes

src/coreclr/jit/importer.cpp Show resolved Hide resolved

hez2010 added 2 commits January 31, 2025 15:49

Address review feedback

b177f77

Fix typo

1ea346a

build-analysis bot mentioned this pull request Jan 31, 2025

[Linux-arm64] ModuleNotFoundError: No module named 'requests' #112021

Open

AndyAyersMS approved these changes Jan 31, 2025

View reviewed changes

Merge branch 'main' into deabstract-enumerator

eb9e993

AndyAyersMS merged commit 44167e0 into dotnet:main Feb 1, 2025
127 of 129 checks passed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

JIT: Update type when return temp is freshly created #111948

JIT: Update type when return temp is freshly created #111948

hez2010 commented Jan 29, 2025 •

edited

Loading

dotnet-policy-service bot commented Jan 29, 2025

hez2010 commented Jan 29, 2025 •

edited

Loading

hez2010 commented Jan 29, 2025

hez2010 commented Jan 29, 2025

jakobbotsch commented Jan 29, 2025 •

edited

Loading

hez2010 commented Jan 29, 2025 •

edited

Loading

hez2010 commented Jan 29, 2025 •

edited

Loading

AndyAyersMS commented Jan 29, 2025

hez2010 commented Jan 29, 2025 •

edited

Loading

hez2010 commented Jan 30, 2025 •

edited

Loading

hez2010 commented Jan 30, 2025 •

edited

Loading

hez2010 commented Jan 30, 2025

jakobbotsch commented Jan 30, 2025

hez2010 commented Jan 30, 2025

jakobbotsch commented Jan 30, 2025

azure-pipelines bot commented Jan 30, 2025

hez2010 commented Jan 30, 2025

hez2010 commented Jan 31, 2025

AndyAyersMS left a comment

AndyAyersMS commented Jan 31, 2025

hez2010 commented Jan 31, 2025

AndyAyersMS commented Jan 31, 2025

AndyAyersMS commented Jan 31, 2025

AndyAyersMS commented Feb 1, 2025

azure-pipelines bot commented Feb 1, 2025

hez2010 commented Feb 1, 2025

AndyAyersMS commented Feb 1, 2025

JIT: Update type when return temp is freshly created #111948

JIT: Update type when return temp is freshly created #111948

Conversation

hez2010 commented Jan 29, 2025 • edited Loading

dotnet-policy-service bot commented Jan 29, 2025

hez2010 commented Jan 29, 2025 • edited Loading

hez2010 commented Jan 29, 2025

hez2010 commented Jan 29, 2025

jakobbotsch commented Jan 29, 2025 • edited Loading

hez2010 commented Jan 29, 2025 • edited Loading

hez2010 commented Jan 29, 2025 • edited Loading

AndyAyersMS commented Jan 29, 2025

hez2010 commented Jan 29, 2025 • edited Loading

hez2010 commented Jan 30, 2025 • edited Loading

hez2010 commented Jan 30, 2025 • edited Loading

hez2010 commented Jan 30, 2025

jakobbotsch commented Jan 30, 2025

hez2010 commented Jan 30, 2025

jakobbotsch commented Jan 30, 2025

azure-pipelines bot commented Jan 30, 2025

hez2010 commented Jan 30, 2025

hez2010 commented Jan 31, 2025

AndyAyersMS left a comment

Choose a reason for hiding this comment

AndyAyersMS commented Jan 31, 2025

hez2010 commented Jan 31, 2025

AndyAyersMS commented Jan 31, 2025

AndyAyersMS commented Jan 31, 2025

AndyAyersMS commented Feb 1, 2025

azure-pipelines bot commented Feb 1, 2025

hez2010 commented Feb 1, 2025

AndyAyersMS commented Feb 1, 2025

hez2010 commented Jan 29, 2025 •

edited

Loading

hez2010 commented Jan 29, 2025 •

edited

Loading

jakobbotsch commented Jan 29, 2025 •

edited

Loading

hez2010 commented Jan 29, 2025 •

edited

Loading

hez2010 commented Jan 29, 2025 •

edited

Loading

hez2010 commented Jan 29, 2025 •

edited

Loading

hez2010 commented Jan 30, 2025 •

edited

Loading

hez2010 commented Jan 30, 2025 •

edited

Loading