[PATCH] D55281: debuginfo: Use symbol difference for CU length to simplify assembly reading/editing

[Paul Robinson via Phabricator via llvm-commits]

probinson added a comment.

In D55281#1333552 <https://reviews.llvm.org/D55281#1333552>, @dblaikie wrote:

> @aprantl @probinson - you folks haev any feelings about this? Looks like these flags are essentially unused (except for testing, which could equally be accomplished by tests that specify an NVPTX triple, I think (except on builds that aren't building an NVPTX backend)) & are a collection of features that are all only needed on NVPTX. I'd lean towards removing them all & gating this on NVPTX. Or on one flag rather than several fine-grained ones, if/when someone starts playing with a ptx assembler that can cope with these features.
>
> But yeah, no big deal I guess - leads to a bit of confusion about who these flags are meant for, under what conditions some of these features might be enabled but not others, etc.


In a way this reminds me of the discussions around how to do "debugger tuning" and there were some very strong voices in favor of unpacking the tuning setting into separate flags.  I believe that the DwarfDebug ctor is the only place that directly checks tuning, or that was the goal on the LLVM side anyhow.

I guess there are already feature flags for some of these ptxas quirks; the regular assembler doesn't care which way the flag is set, and ptxas depends on it being a certain way.  I think as long as the regular assembler is agnostic about the setting, then having a feature flag instead of a triple check is the pattern we have been following.  And that's the case here, right? Turning a hard-coded constant into an expression that the assembler will compute.
So I would lean in favor of a feature flag, defaulted based on the triple.


Repository:
  rL LLVM

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D55281/new/

https://reviews.llvm.org/D55281