Inlining with different target features

Hi llvm-dev,

I recently updated the WebAssembly TargetTransformInfo to allow functions with different target feature sets to be inlined into each other, but I ran into an issue I want to get the community’s opinion on.

Since WebAssembly modules have to be validated before they are run, it only makes sense to talk about WebAssembly features at module granularity rather than function granularity. The WebAssembly backend even runs a pass that applies the union of all used features to each function. That means that ideally inlining for the WebAssembly target would be able to disregard features entirely, since they will all be the same in the end.

However, right now I have to be more conservative than that and only allow a callee to be inlined into a caller if the callee has a subset of the caller’s features. Otherwise, a target intrinsic might end up being used in a function that does not have the necessary target features enabled, which would cause a validation failure.

The best solution I can think of for this problem would be to allow targets to opt-in to having a caller’s feature set updated to include the callee’s feature set when the callee is inlined into the caller. This could be implemented via a new TTI hook, but a more general solution might be to change the return type of areInlineCompatible to allow targets to control this behavior on a case-by-case basis. Does this general direction sound ok, and if so, would it be better to add a new hook or add functionality to the existing one?

Thanks,

Thomas

+echristo for thoughts on subtarget features and function inlining.

Maybe target features aren’t the right tool to model the WebAssembly situation? Perhaps you could model those with mergeable module-level metadata instead? Then the module would always have all the features and that would match the “we’re allowed to union all features across all functions anyway” without it being a delayed pass that happens later.

Thanks for your reply! Using module metadata is an interesting idea, but it would require frontends to make wasm-specific changes to how they handle target features, which would be unfortunate. Working around that by extracting target features into metadata somewhere seems like it would be at least as intrusive as having the inliner update target features. We would also lose out on other valuable utilities like the ability to limit target intrinsics to certain features.

Thanks for your reply! Using module metadata is an interesting idea, but it would require frontends to make wasm-specific changes to how they handle target features, which would be unfortunate. Working around that by extracting target features into metadata somewhere seems like it would be at least as intrusive as having the inliner update target features. We would also lose out on other valuable utilities like the ability to limit target intrinsics to certain features.

Fair enough all round. Though how does wasm work with this “spread the features all around” - does the resulting binary end up only being compatible with a target that has those features? Essentially making features impossible to use conditionally? (I couldn’t write some code that did a runtime test for certain features, then used them only under that condition - because my whole program would end up with the features enabled anyway?)

Another direction would be to require the features to be specified consistently for all components of the build, I guess - if that’s the net effect anyway. Would make portable libraries difficult, though - because they’d be linked into different things with different feature sets and that would violate the invariant.

Hi Thomas,

I’d prefer not to change areInlineCompatible because I think it reads fairly closely what is expected here (also see x86 for how this is used for subset inlining calculations). I think if you plan on updating all of the features for the functions to match you might just want to do that initially rather than try to update them piecemeal after or during inlining.

Thoughts?

-eric

David,

That’s right, WebAssembly does not have a way to conditionally use a feature or even do runtime feature testing right now. It’s on our roadmap of things to design and standardize, but it is still a long way off.

Another direction would be to require the features to be specified consistently for all components of the build, I guess - if that’s the net effect anyway. Would make portable libraries difficult, though - because they’d be linked into different things with different feature sets and that would violate the invariant.

I agree this would be reasonable. We already require separate builds of a library for each feature set the library wants to support, so this wouldn’t make that story any worse. The only reason I wouldn’t want to enforce this is because that would be another way targeting Wasm would be different from targeting other platforms for frontends.

Eric,

Updating all the features up front would indeed make the most sense. I haven’t seen a way for backends to specify LLVM IR passes that should be run early, though. Is that possible, or would frontends have to add this extra pass when targeting Wasm?

David,

That’s right, WebAssembly does not have a way to conditionally use a feature or even do runtime feature testing right now. It’s on our roadmap of things to design and standardize, but it is still a long way off.

Another direction would be to require the features to be specified consistently for all components of the build, I guess - if that’s the net effect anyway. Would make portable libraries difficult, though - because they’d be linked into different things with different feature sets and that would violate the invariant.

I agree this would be reasonable. We already require separate builds of a library for each feature set the library wants to support, so this wouldn’t make that story any worse. The only reason I wouldn’t want to enforce this is because that would be another way targeting Wasm would be different from targeting other platforms for frontends.

Eric,

Updating all the features up front would indeed make the most sense. I haven’t seen a way for backends to specify LLVM IR passes that should be run early, though. Is that possible, or would frontends have to add this extra pass when targeting Wasm?

+Alina Sbirlea

There are different hook points in the pipelines - I don’t see a problem with adding another one if we need to. :slight_smile:

-eric

David,

That’s right, WebAssembly does not have a way to conditionally use a feature or even do runtime feature testing right now. It’s on our roadmap of things to design and standardize, but it is still a long way off.

Another direction would be to require the features to be specified consistently for all components of the build, I guess - if that’s the net effect anyway. Would make portable libraries difficult, though - because they’d be linked into different things with different feature sets and that would violate the invariant.

I agree this would be reasonable. We already require separate builds of a library for each feature set the library wants to support, so this wouldn’t make that story any worse.

So if that’s already the case, what would change between that state and what I’m suggesting?

The difference is illustrated in this bug report: https://github.com/rust-lang/rust/issues/74320. Essentially, frontends like Rust want to allow users to opt-in to a feature by annotating a function that will use it. Your suggestion to require that all functions have the same feature set would prevent frontends from doing this sort of thing when targeting WebAsssembly. Which on one hand would be entirely reasonable, since annotating a single function with a feature misrepresents how WebAssembly features work, but on the other hand would make WebAssembly different from other targets, which I’d like to avoid.

That specific example, https://github.com/rust-lang/rust/issues/74320 seems like it would be covered by the superset check right?

The difference is illustrated in this bug report: https://github.com/rust-lang/rust/issues/74320. Essentially, frontends like Rust want to allow users to opt-in to a feature by annotating a function that will use it. Your suggestion to require that all functions have the same feature set would prevent frontends from doing this sort of thing when targeting WebAsssembly. Which on one hand would be entirely reasonable, since annotating a single function with a feature misrepresents how WebAssembly features work, but on the other hand would make WebAssembly different from other targets, which I’d like to avoid.

Ah, then I’m not sure I understand “We already require separate builds of a library for each feature set the library wants to support, so this wouldn’t make that story any worse.” then. Is that statement (“already require separate builds for a library for each feature set the library wants to support”) any different from, say x86? I guess for something like x86 you could have one object built with one feature, another object built with another feature, a 3rd that does a runtime test before dispatching to the appropriate object - and all that in a single library.

So, yeah, that statement makes it a bit less versatile than the x86/other target model - but it doesn’t make it as bad as I’m suggesting. So what I’m suggesting (that everything’d have to be rebuilt for any final target combination of features) would make things worse, by the sounds of it?