Experiment: Enable multi-value #1459

nomeata · 2020-05-02T17:53:42Z

drun now uses wasmtime-0.15, which supports multi-value.

We can test all that on branches. Likely, though, allowing multi-value needs to
be part of a Public Spec release, so this can’t be the default for a while.

but `drun` does not support it out of box.

happens to be the version integrated into `drun`

nomeata · 2020-05-02T18:18:07Z

Ugh, we’d have to switch back to a fork of the wasm ocaml library that is not yet on Opam to get multi-value support. Not sure if it’s worth it.

nomeata · 2020-05-02T18:21:12Z

Maybe there is a benefit in not trying to use any not-yet-fully-approved wasm featues. Because image we support more than one (say, multi-value and bulk memory). Then we’d have to maintain our own merge of the wasm library.

Or we stop using the ocaml wasm library, for these reasons? It’s a reference implementation after all, and not necessarily a production ready, convenient-to-use, new features avaiable behind feature flags library… But it’s certainly convenient sometimes to use the wasm library.

@rossberg @ggreif what do you think?

crusso · 2020-05-02T21:33:22Z

What's the status on tail calls? I think that's actually a more important feature for us because of our cps conversion....

nomeata · 2020-05-03T09:23:00Z

Also some relevant more general discussion:
https://dfinity.slack.com/archives/CH4CADCJX/p1588453621167800

rossberg · 2020-05-04T07:07:25Z

Multi-value is fully approved and merged into the official spec at this point. It's also in the wasm interpreter. I merely didn't think of creating a new opam package. Will do today.

nomeata · 2020-05-04T08:22:53Z

Thanks! I missed that it had been merged. Also, the README of https://github.com/WebAssembly/multi-value/ maybe ought to indicate that.

rossberg · 2020-05-04T09:14:36Z

Good point. Added a note to the README.

rossberg · 2020-05-04T09:15:55Z

@crusso, tail calls are still at stage 3, awaiting a second implementation besides V8.

will help with #1459

will help with #1459 Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>

ggreif · 2020-05-05T08:44:10Z

This needs a rebase now, to reduce the diffs. Also note #1467, which is in the works.

…ulti-value

nomeata · 2020-05-05T09:33:43Z

Updated (I tend to avoid rebase these days)

nomeata · 2020-05-05T15:49:57Z

Nice, tests.run passes.
And so do the ic-ref-run tests…

inspired by the work in #1459 I figured it would be nice if Hydra runs the tests in `tests/run-drun` separately with `drun` and `ic-ref-run`. This way * We can see from the status list already if only one of the two break (e.g. after bumping dependencies or changing Wasm featurs) * We do less rebuilds (e.g. after bumping dependencies) * We have higher nix-level parallelism (relevant as `tests.run-drun` is one of the longest running tests) Thanks to #1216 this only needs a small change to `default.nix`.

* nix, tests: Split run-drun inspired by the work in #1459 I figured it would be nice if Hydra runs the tests in `tests/run-drun` separately with `drun` and `ic-ref-run`. This way * We can see from the status list already if only one of the two break (e.g. after bumping dependencies or changing Wasm featurs) * We do less rebuilds (e.g. after bumping dependencies) * We have higher nix-level parallelism (relevant as `tests.run-drun` is one of the longest running tests) Thanks to #1216 this only needs a small change to `default.nix`. * nix: Simplify name handling for test derivations Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>

…-value

nomeata · 2021-08-26T09:04:53Z

I was curious if recent updates to drun (in particular newer wasmtime) fixed this, but we still get

ingress Err: IC0505: Wasm module of canister rwlgt-iiaaa-aaaaa-aaaaa-cai is not valid: Failed to deserialize wasm module with Return types length should be 0 or 1

osa1 · 2021-08-26T11:05:46Z

I suspect the problem is parity-wasm, which is used for instrumentation. See paritytech/parity-wasm#292

rossberg · 2021-08-26T13:55:16Z

Other than @osa1's contribution of trunc_sat instructions, the parity impl seems to have made practically no progress over the last two years. That's rather concerning for a dependency. :(

…-value

nomeata · 2021-10-22T14:26:27Z

The base tests fall over this code

let as_block_type : stack_type -> block_type = function
  | [] -> ValBlockType None
  | [t] -> ValBlockType (Some t)
  | _ -> raise (Invalid_argument "instrList block combinators do not support multi-value yet")

which I introduced in 2ad8eed, three days after I branched off this branch. This will require a bit of refactoring. Thinking out loud: With multi-value returns, we need to produce a VarBlockType here, and we need to register the type with the type tabe, so as_block_type becomes effectful with regard to the wasm module. Either we put the burden on compile.ml, passing a block_type to G.if_ etc, or we make the G.t abstraction keep track of which types need to be added to the environment.

incidentially, this raise is not triggered by tests in tests/run, only the base tests. That is not good.

nomeata · 2021-10-22T15:29:14Z

Fixed that raise and added a simple test that exercised it. Also in #2853.

These Wasm nodes need to be annotated with their return type, and if the return type is more than nullary or unary, then this type needs to be defined in the Wasm module type table. This refactors the story around `if_` etc. to allow that: `E.if_ env tys thn else` will register the types as needed But in many places we know statically that there is only zero or one return type. For them, `G.if0` and `G.if1` can be used for cleaner code and to emphasize their “purity”. Extracted from #1459, but I’d like to have it on master to prevent merge conflicts in #1459 (which is a long-running branch since May 2020…)

nomeata · 2021-10-22T15:45:20Z

BTW, Claudio, do you (internally) already have canisters that use this? Becaues “multi-value compilation” is still an open issue in the Rust repo: rust-lang/rust#73755. It seems dangerous to turn on this feature if it isn’t fully supported by upstream rust already

These Wasm nodes need to be annotated with their return type, and if the return type is more than nullary or unary, then this type needs to be defined in the Wasm module type table. This refactors the story around `if_` etc. to allow that: `E.if_ env tys thn else` will register the types as needed But in many places we know statically that there is only zero or one return type. For them, `G.if0` and `G.if1` can be used for cleaner code and to emphasize their “purity”. In a similar vein, `G.loop0` is introduced; since our loops never produce values on the stack, there is no `loop_` with multi-value result. Extracted from #1459, but I’d like to have it on master to prevent merge conflicts in #1459 (which is a long-running branch since May 2020…)

crusso · 2021-10-24T20:51:37Z

BTW, Claudio, do you (internally) already have canisters that use this? Becaues “multi-value compilation” is still an open issue in the Rust repo: rust-lang/rust#73755. It seems dangerous to turn on this feature if it isn’t fully supported by upstream rust already

No, the only reason I was asking about multi-value is that I've been asked to support the 128 bit cycles API which appears to use multi-value returns for System functions. I assume the replica has some tests, but perhaps those were hand-crafted wasm, not rust compiled.

…-value

nomeata · 2021-12-12T14:44:32Z

Gave it another shot. Fun fact: tests.drun all succeed, but tests.perf fails in qr.mo with

ingress Err: IC0505: Wasm module of canister rwlgt-iiaaa-aaaaa-aaaaa-cai is not valid: Failed to deserialize wasm module with Invalid or unknown value type 34

The fact that this only shows up in perf means our test suite is incomplete. I found a simple reproducer, and added it to this PR. It seems that the replica does not yet accept multi-value files after all.

extracted from #1459

So far, the IC does not support wasm using multi-value returns from functions and blocks (see #1459). The backend was originally written with that feature in mind, and later we added a work-around (module `FakeMultiVal`). For functions we work around it by storing multiple values in fixed globals, and reading them after the call. No allocations needed. For blocks with multiple values, however, we allocate a tuple (array) on the heap, write the values there, and read them later. This adds unnecessary heap churn and fills up memory faster. So this PR uses the mechanism from the functions also for blocks. Goes well with #3556

@1

So far, the IC does not support wasm using multi-value returns from functions and blocks (see #1459). The backend was originally written with that feature in mind, and later we added a work-around (module `FakeMultiVal`). For functions we work around it by storing multiple values in fixed globals, and reading them after the call. No allocations needed. For blocks with multiple values, however, we allocate a tuple (array) on the heap, write the values there, and read them later. This adds unnecessary heap churn and fills up memory faster. So this PR uses the mechanism from the functions also for blocks. Goes well with #3556 Here is a (particularly bad) example: In ```motoko func multi_value_blocks(n : Nat) : (Nat, Nat) { return ( if (n == 0) { returns_tuple() } else { if (n == 1) { returns_tuple() } else { returns_tuple() } } ); }; ``` Now no allocation happens (and all the reads and writes to the global variables are eliminated due to #3556): ```diff (func $multi_value_blocks (type 1) (param $clos i32) (param $n i32) local.get $n i32.const 0 call $B_eq - if (result i32) ;; label = @1 + if ;; label = @1 i32.const 0 call $returns_tuple - global.get 4 - global.get 5 - call $to_2_tuple else local.get $n i32.const 2 call $B_eq - if (result i32) ;; label = @2 + if ;; label = @2 i32.const 0 call $returns_tuple - global.get 4 - global.get 5 - call $to_2_tuple else i32.const 0 call $returns_tuple - global.get 4 - global.get 5 - call $to_2_tuple end end - call $from_2_tuple return) ```

nomeata · 2023-03-06T15:44:20Z

The Haskell to WebAssembly story hasn't enabled multi-value yet due to llvm/llvm-project#59095. Worth keeping an eye on.

nomeata added 6 commits May 2, 2020 19:43

Experiment: Try enabling multi-value again

d718193

but `drun` does not support it out of box.

Bump wasmtime to 0.15

8127150

happens to be the version integrated into `drun`

Bump dfinity to enable multi-value

33246c9

-wasi-system-api: export _start

93374cb

Update test output (new wasmtime version)

89bd890

Stackrep stuff

001b54c

nomeata added a commit that referenced this pull request May 4, 2020

Bump ic-ref

933510d

will help with #1459

nomeata mentioned this pull request May 4, 2020

Bump ic-ref #1462

Merged

nomeata added a commit that referenced this pull request May 4, 2020

Bump ic-ref

45adb24

will help with #1459

mergify bot added a commit that referenced this pull request May 4, 2020

Bump ic-ref (#1462)

c6394a5

will help with #1459 Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>

Merge branch 'master' of github.com:dfinity-lab/motoko into joachim/m…

4b8f57e

…ulti-value

ghost assigned nomeata May 5, 2020

ghost added the P3 low priority, resolve when there is time label May 5, 2020

Merge branch 'master' into joachim/multi-value

fc0e3e6

nomeata mentioned this pull request May 6, 2020

nix, tests: Split run-drun #1479

Merged

Merge branch 'master' into joachim/multi-value

c48c46d

Merge branch 'master' into joachim/multi-value

fccfd76

nomeata added 2 commits August 26, 2021 11:00

Merge branch 'master' of github.com:dfinity/motoko into joachim/multi…

baf1250

…-value

Update command line flags

75f6d1a

nomeata mentioned this pull request Oct 21, 2021

Adopt 128 bit cycles API #2833

Closed

Merge branch 'master' of github.com:dfinity/motoko into joachim/multi…

3fdb00b

…-value

nomeata mentioned this pull request Oct 22, 2021

Backend: Prepare if_ and block_ for multi-value #2853

Merged

nomeata force-pushed the joachim/multi-value branch from 5ffe568 to 5171f7a Compare October 22, 2021 19:38

Merge branch 'master' of github.com:dfinity/motoko into joachim/multi…

1b4a4f7

…-value

nomeata force-pushed the joachim/multi-value branch from 5171f7a to 1b4a4f7 Compare December 12, 2021 12:36

Simple multi-value test case

ae36c8f

nomeata added a commit that referenced this pull request Dec 13, 2021

tests: Code that uses multi-value block results

2b27ae5

extracted from #1459

nomeata mentioned this pull request Dec 13, 2021

tests: Code that uses multi-value block results #2992

Merged

mergify bot pushed a commit that referenced this pull request Dec 13, 2021

tests: Code that uses multi-value block results (#2992)

7b9c8ba

extracted from #1459

nomeata mentioned this pull request Nov 10, 2022

Multi-value emulation: Use globals, not allocated tuples #3566

Merged

Merge remote-tracking branch 'origin/master' into joachim/multi-value

d4d854e

dfinity-droid-prod bot added the cla:agreed label Nov 11, 2022

update bench output

7c80e7f

nomeata mentioned this pull request Mar 8, 2023

AssocList.replace: Even less allocations dfinity/motoko-base#541

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Experiment: Enable multi-value #1459

Experiment: Enable multi-value #1459

nomeata commented May 2, 2020 •

edited

Loading

nomeata commented May 2, 2020

nomeata commented May 2, 2020

crusso commented May 2, 2020

nomeata commented May 3, 2020

rossberg commented May 4, 2020

nomeata commented May 4, 2020

rossberg commented May 4, 2020

rossberg commented May 4, 2020

ggreif commented May 5, 2020 •

edited

Loading

nomeata commented May 5, 2020

nomeata commented May 5, 2020 •

edited

Loading

nomeata commented Aug 26, 2021

osa1 commented Aug 26, 2021 •

edited

Loading

rossberg commented Aug 26, 2021

nomeata commented Oct 22, 2021 •

edited

Loading

nomeata commented Oct 22, 2021 •

edited

Loading

nomeata commented Oct 22, 2021 •

edited

Loading

crusso commented Oct 24, 2021

nomeata commented Dec 12, 2021

nomeata commented Mar 6, 2023

Experiment: Enable multi-value #1459

Are you sure you want to change the base?

Experiment: Enable multi-value #1459

Conversation

nomeata commented May 2, 2020 • edited Loading

nomeata commented May 2, 2020

nomeata commented May 2, 2020

crusso commented May 2, 2020

nomeata commented May 3, 2020

rossberg commented May 4, 2020

nomeata commented May 4, 2020

rossberg commented May 4, 2020

rossberg commented May 4, 2020

ggreif commented May 5, 2020 • edited Loading

nomeata commented May 5, 2020

nomeata commented May 5, 2020 • edited Loading

nomeata commented Aug 26, 2021

osa1 commented Aug 26, 2021 • edited Loading

rossberg commented Aug 26, 2021

nomeata commented Oct 22, 2021 • edited Loading

nomeata commented Oct 22, 2021 • edited Loading

nomeata commented Oct 22, 2021 • edited Loading

crusso commented Oct 24, 2021

nomeata commented Dec 12, 2021

nomeata commented Mar 6, 2023

nomeata commented May 2, 2020 •

edited

Loading

ggreif commented May 5, 2020 •

edited

Loading

nomeata commented May 5, 2020 •

edited

Loading

osa1 commented Aug 26, 2021 •

edited

Loading

nomeata commented Oct 22, 2021 •

edited

Loading

nomeata commented Oct 22, 2021 •

edited

Loading

nomeata commented Oct 22, 2021 •

edited

Loading