Add support for WASI based on Wasmtime's implementation

[OLUWAMUYIWA]

Closes #358

[OLUWAMUYIWA]

Hi @Robbepop .This basically depends on wasmtime's implementation of the WASI, meaning I didn't have to do much. Much of what I did is to integrate it with wasmi's Linker, and your work has made that very easy. Please check and request changes. I'm happy to see it through.

[paritytech-cicd-pr]

BENCHMARKS

	NATIVE			WASMTIME
BENCHMARK	MASTER	PR	DIFF	MASTER	PR	DIFF	WASMTIME OVERHEAD
execute/ bare_call_0	1.01ms	1.01ms	🟢 -0.03%	910.91µs	881.10µs	🟢 -3.24%	🟢 -13%
execute/ bare_call_0/typed	526.31µs	514.38µs	🟢 -2.21%	409.66µs	398.48µs	🟢 -2.65%	🟢 -23%
execute/ bare_call_1	1.06ms	1.05ms	🟢 -0.89%	1.08ms	1.08ms	🔴 0.01%	🟢 3%
execute/ bare_call_16	2.25ms	2.16ms	🟢 -4.53%	4.75ms	4.76ms	🟢 0.22%	🔴 120%
execute/ bare_call_16/typed	1.48ms	1.48ms	⚪ -0.71%	2.11ms	2.02ms	🟢 -4.51%	🟢 36%
execute/ bare_call_1/typed	609.11µs	611.42µs	⚪ 0.24%	686.88µs	726.24µs	🔴 5.54%	🟢 19%
execute/ bare_call_4	1.18ms	1.19ms	⚪ 0.28%	1.71ms	1.74ms	⚪ 1.21%	🟢 46%
execute/ bare_call_4/typed	626.42µs	626.41µs	⚪ 0.06%	814.07µs	815.16µs	⚪ 0.20%	🟢 30%
execute/ br_table	627.89µs	635.60µs	⚪ 1.48%	883.09µs	892.11µs	⚪ 1.10%	🟢 40%
execute/ count_until	697.15µs	651.91µs	🟢 -6.50%	2.00ms	2.06ms	🔴 2.60%	🔴 216%
execute/ factorial_iterative	307.83µs	304.73µs	🟢 -1.19%	844.39µs	845.50µs	⚪ -0.24%	🔴 177%
execute/ factorial_recursive	616.12µs	608.20µs	🟢 -1.33%	1.26ms	1.26ms	⚪ 0.24%	🔴 107%
execute/ fib_iterative	1.45ms	1.43ms	🟢 -1.19%	4.49ms	4.46ms	⚪ -0.56%	🔴 212%
execute/ fib_recursive	5.74ms	5.92ms	🔴 3.26%	11.39ms	11.30ms	⚪ -0.75%	🟡 91%
execute/ global_bump	950.25µs	950.23µs	⚪ -0.20%	3.11ms	3.16ms	🔴 1.59%	🔴 232%
execute/ global_const	717.68µs	716.46µs	⚪ -0.04%	2.32ms	2.32ms	⚪ -0.02%	🔴 224%
execute/ host_calls	30.13µs	28.97µs	🟢 -3.80%	37.46µs	38.98µs	🔴 3.98%	🟢 35%
execute/ memory_fill	1.28ms	1.28ms	⚪ -0.07%	3.90ms	3.98ms	🔴 1.87%	🔴 210%
execute/ memory_sum	1.81ms	1.25ms	🟢 -22.99%	3.93ms	3.95ms	⚪ 0.57%	🔴 216%
execute/ memory_vec_add	2.60ms	2.59ms	⚪ -0.15%	8.30ms	8.21ms	⚪ -1.14%	🔴 217%
execute/ recursive_is_even	1.13ms	1.11ms	⚪ -1.05%	2.10ms	2.10ms	⚪ -0.21%	🟡 89%
execute/ recursive_ok	148.57µs	146.23µs	🟢 -1.50%	302.09µs	293.55µs	🟢 -2.84%	🔴 101%
execute/ recursive_scan	182.07µs	179.50µs	🟢 -1.45%	371.56µs	375.10µs	⚪ 0.81%	🔴 109%
execute/ recursive_trap	14.76µs	14.44µs	🟢 -2.34%	29.60µs	28.61µs	🟢 -3.03%	🟡 98%
execute/ regex_redux	545.58µs	545.64µs	⚪ 0.05%	1.40ms	1.45ms	🔴 3.66%	🔴 166%
execute/ rev_complement	506.71µs	505.62µs	⚪ -0.26%	1.39ms	1.44ms	🔴 3.25%	🔴 184%
execute/ tiny_keccak	374.59µs	372.65µs	⚪ -0.72%	1.16ms	1.23ms	🔴 6.54%	🔴 230%
execute/ trunc_f2i	911.67µs	912.72µs	⚪ 0.07%	2.38ms	2.43ms	🔴 1.98%	🔴 166%
instantiate/ wasm_kernel	58.88µs	60.19µs	🔴 4.33%	95.16µs	67.67µs	🟢 -28.62%	🟢 12%
translate/ erc1155	209.52µs	207.67µs	⚪ -0.54%	375.61µs	377.63µs	⚪ 0.74%	🟡 82%
translate/ erc20	101.83µs	101.37µs	⚪ -0.49%	182.57µs	185.97µs	🔴 1.80%	🟡 83%
translate/ erc721	146.60µs	146.84µs	⚪ 0.02%	266.37µs	268.89µs	⚪ 0.91%	🟡 83%
translate/ spidermonkey	0.00ns	0.00ns	⚪ 0.47%	0.00ns	0.00ns	⚪ 0.91%	🟢 0%
translate/ wasm_kernel	3.79ms	3.77ms	⚪ -0.51%	7.17ms	7.19ms	⚪ 0.32%	🟡 91%

Link to pipeline

[Robbepop]

@OLUWAMUYIWA from a design perspective this PR looks as I expected it on a surface level which is good.
I have not taken a deep look but the general structure looks decent. I am a bit worried about those extreme regressions since I could not find a single piece of diff in this PR that would justify those regressions.

Could you please update to the most recent master branch before I do the proper review?

[OLUWAMUYIWA]

Thanks @Robbepop and @athei .
Much has been discussed by @Robbepop and I in chats. And I think i have a fairly good understanding of what he expects.

But first, I suppose I need to create another PR addressing #517 . Namely, Traps.

[codecov-commenter]

Codecov Report

Merging #557 (6b999ac) into master (53eac5c) will increase coverage by 0.34%.
The diff coverage is 88.46%.

@@            Coverage Diff             @@
##           master     #557      +/-   ##
==========================================
+ Coverage   79.73%   80.07%   +0.34%     
==========================================
  Files          75       77       +2     
  Lines        6251     6304      +53     
==========================================
+ Hits         4984     5048      +64     
+ Misses       1267     1256      -11     

Impacted Files	Coverage Δ
crates/wasi/src/snapshots/preview_1.rs	81.25% <81.25%> (ø)
crates/wasi/tests/wasi_wat.rs	100.00% <100.00%> (ø)
crates/core/src/trap.rs	41.50% <0.00%> (-0.80%)	⬇️
crates/wasmi/src/memory/mod.rs	62.69% <0.00%> (+2.38%)	⬆️
crates/wasmi/src/func/into_func.rs	93.18% <0.00%> (+4.54%)	⬆️
crates/wasmi/src/store.rs	74.04% <0.00%> (+4.58%)	⬆️
crates/wasmi/src/func/caller.rs	52.94% <0.00%> (+41.17%)	⬆️

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[Robbepop]

To fix the no_std CI failures we also need to add --exclude wasmi_wasi to GitHub Action's rust.yml:

      - name: Build wasmi itself as no_std
        uses: actions-rs/cargo@v1
        with:
          command: build
          args: --workspace --lib --no-default-features --target thumbv7em-none-eabi --exclude wasmi_cli

[Robbepop]

LGTM

Thanks a ton for all the hard work to get partial WASI support for wasmi @OLUWAMUYIWA !

Let us merge this. 🥳