dotnet clr: I did some optimizations on my reflection emit, and now I get the following
message from PEVerify:

Stack height at all points must be determinable in a single forward scan of
IL

I'm not doing anything especially fancy like folding common code sections
together and branching from different locations in my method body. The
stack height is fixed for every instruction regardless of execution path.

The method initializes a BitArray field from an array of indexes to be set
(MyClass/Bits is an enum)
Here's the MSIL viewed with reflector (annotated with the stack height
before/after the instruction):
..method public specialname rtspecialname instance void .ctor([in]
MyClass/Bits[] bits) cil managed
{
.param [1]
.custom instance void [mscorlib]System.ParamArrayAttribute::.ctor()
.maxstack 6
.locals init (
[0] [mscorlib]System.Collections.BitArray array1,
[1] int32 num1)
0/1 L_0000: ldarg.0
1/2 L_0001: ldc.i4 16
2/2 L_0006: newobj instance void
[mscorlib]System.Collections.BitArray::.ctor(int32)
2/3 L_000b: ldarg.1
3/2 L_000c: brfalse L_002f
2/1 L_0011: stloc.0
1/2 L_0012: ldarg.1
2/2 L_0013: ldlen
2/1 L_0014: stloc.1
1/1 L_0015: br L_0028
# warning occurs here
1/2 L_001a: ldloc.1
2/3 L_001b: ldc.i4.1
3/2 L_001c: sub
2/1 L_001d: stloc.1
1/2 L_001e: ldloc.0
2/3 L_001f: ldarg.1
3/4 L_0020: ldloc.1
4/3 L_0021: ldelem.i4
3/4 L_0022: ldc.i4.1
4/1 L_0023: callvirt instance void
[mscorlib]System.Collections.BitArray::Set(int32, bool)
1/2 L_0028: ldloc.1
2/1 L_0029: brtrue L_001a
1/2 L_002e: ldloc.0
2/0 L_002f: stfld [mscorlib]System.Collections.BitArray
MyClass::internalValue
0/* L_0034: ret
}

What's more, even if PEVerify makes a linear pass, instead of following
unconditional branches, the constraint still holds (although I think I can
knock one instruction off if I relax that.

Are unconditional forward branches forbidden?

[quoted text, click to view]

Found it in the CIL Spec Part 3:

1.7.5 Backward Branch Constraints
It must be possible, with a single forward-pass through the CIL instruction
stream for any method, to infer the exact state of the evaluation stack at
every instruction (where by "state" we mean the number and type of each item
on the evaluation stack).
In particular, if that single-pass analysis arrives at an instruction, call
it location X, that immediately follows an unconditional branch, and where X
is not the target of an earlier branch instruction, then the state of the
evaluation stack at X, clearly, cannot be derived from existing information.
In this case, the CLI demands that the evaluation stack at X be empty.
Following on from this rule, it would clearly be invalid CIL if a later
branch instruction to X were to have a non-empty evaluation stack

The spec uses the word invalid. So, is my code not only unverifiable but
also illegal? However, I think I can deal with this with an unreachable
instruction setting the correct stack depth.

[quoted text, click to view]

Hi Ben,

Thanks for your post!

Yes, this "Backward Branch Constraints" is the root cause for this error
message. This is considered as an illegal. PEVerify will produce this error
message if you fail to obey the rule. Please refer to the link below:
"PEVerify does not respect Backward Branch Constraints. "
http://connect.microsoft.com/VisualStudio/feedback/ViewFeedback.aspx?Feedbac
kID=96968

Hope this helps!

Best regards,
Jeffrey Tan
Microsoft Online Community Support
==================================================
When responding to posts, please "Reply to Group" via your newsreader so
that others may learn and benefit from your issue.
==================================================
This posting is provided "AS IS" with no warranties, and confers no rights.