Diagnostic feedback when disassembly fails is needed

[kg]

After a lot of blindly stumbling around in the dark I got DisassemblyDiagnoser to work for a benchmark I'm iterating on, and then a random change to the code being measured made it suddenly stop disassembling all my benchmarks. The only information or diagnostic output I get is:

No benchmarks were disassembled

It would be awesome if there was some way to get diagnostic information here. Like did decoding the pdb fail? Did nothing get jitted during benchmark execution? etc.

[ThatBusFromSpeed]

I'm suffering from this same issue right now. I want to look at what some F# code I have is generating, but I've been slamming my head against the wall for about two weeks trying to find any clues as to why disassembly is failing.

[adamsitnik]

Providing such an information would be very tricky, as we (BDN) use a low level debugger API (ClrMd) to obtain the disassembly.

For example over here you can check what it takes to get in on Intel hardware: https://github.com/dotnet/BenchmarkDotNet/blob/master/src/BenchmarkDotNet/Disassemblers/IntelDisassembler.cs

And here is the arch-agnostic part:
https://github.com/dotnet/BenchmarkDotNet/blob/master/src/BenchmarkDotNet/Disassemblers/ClrMdV3Disassembler.cs

The problem is that even if we add an else block everywhere and print some info I am not sure if this would be actionable from the end user perspective. From my experience, when sth does not work it's usually a runtime or debugger bug...

Over here you can find some docs that explain how to debug it in our repo: https://benchmarkdotnet.org/articles/contributing/disassembler.html#how-to-debug-the-disassembler

[timcassell]

I don't know much about issues with disassembly, but it's possible whatever issues you had in v0.14.0 are fixed in v.0.14.1, since ClrMD was updated to a newer version. You could try with the latest nightly version.