[Lldb-commits] [PATCH] D112691: Include target settings in "statistics dump" output.

Thu Oct 28 15:07:56 PDT 2021

clayborg added a comment.

In D112691#3093889 <https://reviews.llvm.org/D112691#3093889>, @JDevlieghere wrote:

> I understand the need for something like this to make some of the statistics more meaningful, but this is stretching the notion of statistics. Conceptually, this is approaching something like a dump of the debugger for issue/performance analysis. I think that idea is really exciting, and from that perspective there's a lot more useful information we could add to it. Long term, I can see this output be something that we ask users to include in every bug report.

That is kind of what I am going for with this command. One main issue is when breakpoints fail to resolve we often don't know why, and knowing if we a) have debug info enabled and b) have the right settings to remap things really will help. I was thinking the same thing though: this info is not really statistics.

Would adding a "--target-state" option that would only include target stuff if requested make more sense? See my comment below as well

> TL;DR: Can we make this more generic and keep the separation between statistics/metrics and a more general concept that includes these target settings?

I am open to ideas on how to do this kind of thing. We can always gather this information on our own or with a separate command.

We could do a "target read" and "target write" just like we do for breakpoints? The idea would be to serialize all settings needed for a target into JSON and be able to load a target from JSON. This could be really handy for bug reports. If we have a command that does this, we could easily add an option to the "statistics dump" command to include this state with --target-state so we can still get it all in one place, but the user must ask for it? The description for --target-state could be something like:

  -t (--target-state)
      Include the serialization of the target in this information to help with being able to reproduce the conditions that led to the statistics

Lemme know what you think.

Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D112691/new/

https://reviews.llvm.org/D112691