Assertion failed '!"Missing flags on tree"'

[EgorBo]

[MethodImpl(MethodImplOptions.NoInlining)]
static unsafe void PrintFPtr()
{
    delegate*<void> ptr = &Foo;
    Console.WriteLine(*(long*)ptr);
}

[MethodImpl(MethodImplOptions.NoInlining)]
static void Foo()
{
}

Missing flags on tree [000001]: #------------
               [000001] n---G+-----                         *  IND       long  
               [000000] H----+-----                         \--*  CNS_INT(h) long   0x7ffb48ff03c0 ftn

Assert failure(PID 2888 [0x00000b48], Thread: 50468 [0xc524]): Assertion failed '!"Missing flags on tree"' in 'Tests:PrintFPtr()' during 'Morph - Global' (IL size 13; hash 0x403dcc3a; Tier0)

    File: C:\prj\runtime-base\src\coreclr\jit\fgdiagnostic.cpp Line: 3319
    Image: C:\prj\runtime-base\artifacts\bin\coreclr\windows.x64.Checked\CoreRun.exe

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

Issue Details

---

[MethodImpl(MethodImplOptions.NoInlining)]
static unsafe void PrintFPtr()
{
    delegate*<void> ptr = &Foo;
    Console.WriteLine(*(long*)ptr);
}

[MethodImpl(MethodImplOptions.NoInlining)]
static void Foo()
{
}

Missing flags on tree [000001]: #------------
               [000001] n---G+-----                         *  IND       long  
               [000000] H----+-----                         \--*  CNS_INT(h) long   0x7ffb48ff03c0 ftn

Assert failure(PID 2888 [0x00000b48], Thread: 50468 [0xc524]): Assertion failed '!"Missing flags on tree"' in 'Tests:PrintFPtr()' during 'Morph - Global' (IL size 13; hash 0x403dcc3a; Tier0)

    File: C:\prj\runtime-base\src\coreclr\jit\fgdiagnostic.cpp Line: 3319
    Image: C:\prj\runtime-base\artifacts\bin\coreclr\windows.x64.Checked\CoreRun.exe

Author:	EgorBo
Assignees:	-
Labels:	area-CodeGen-coreclr
Milestone:	-

[JulieLeeMSFT]

@AndyAyersMS, PTAL.

[AndyAyersMS]

Sounds like @MichalPetryka knows what this is... assigning to @EgorBo for now.

[MichalPetryka]

Sounds like @MichalPetryka knows what this is...

The issue basically is that the JIT expects the places that can create indirs over handles to be known, which is problematic in cases like this where the code can get both a handle and a normal pointer or when optimizations later fold non handle indirs into handle ones like in #88860. The assert here isn't a real correctness issue, it's just a check against missed places that can create handle indirs.