[syscall] Accept O_NONBLOCK flag in pipe2

[thiblahute]

pipe2 only accepted O_CLOEXEC, rejecting O_NONBLOCK with ENOTSUP. Accept O_NONBLOCK and set it on the created pipe streams, it is a no-op in practice, but there is no reason to not allow it.

[sbc100]

Can you update the test for pipe2 to include this flag?

[sbc100]

lgtm otherwise! Thanks for working on this

[thiblahute]

Added a test

[sbc100]

For the codesize test updates you will want to do emsdk install tot then ./tools/maint/rebaseline_tests.py

[thiblahute]

From https://github.com/emscripten-core/emscripten/actions/runs/23765785637/job/69244759709?pr=26501:

The following (30) test expectation files were updated by
running the tests with `--rebaseline`:


codesize/test_codesize_cxx_ctors1.json: 151832 => 151832 [+0 bytes / +0.00%]
codesize/test_codesize_cxx_ctors2.json: 151235 => 151235 [+0 bytes / +0.00%]
codesize/test_codesize_cxx_except.json: 195690 => 195690 [+0 bytes / +0.00%]
codesize/test_codesize_cxx_except_wasm.json: 166948 => 166948 [+0 bytes / +0.00%]
codesize/test_codesize_cxx_except_wasm_legacy.json: 164829 => 164829 [+0 bytes / +0.00%]
codesize/test_codesize_cxx_lto.json: 120519 => 120519 [+0 bytes / +0.00%]
codesize/test_codesize_cxx_mangle.json: 262181 => 262181 [+0 bytes / +0.00%]
codesize/test_codesize_cxx_noexcept.json: 153855 => 153855 [+0 bytes / +0.00%]
codesize/test_codesize_cxx_wasmfs.json: 179737 => 179737 [+0 bytes / +0.00%]
codesize/test_codesize_file_preload.json: 23789 => 23789 [+0 bytes / +0.00%]
codesize/test_codesize_files_js_fs.json: 18215 => 18215 [+0 bytes / +0.00%]
codesize/test_codesize_files_wasmfs.json: 63883 => 63883 [+0 bytes / +0.00%]
codesize/test_codesize_hello_O0.json: 39111 => 39111 [+0 bytes / +0.00%]
codesize/test_codesize_hello_O1.json: 8875 => 8875 [+0 bytes / +0.00%]
codesize/test_codesize_hello_dylink.json: 43853 => 43853 [+0 bytes / +0.00%]
codesize/test_codesize_mem_O3.json: 9614 => 9614 [+0 bytes / +0.00%]
codesize/test_codesize_mem_O3_grow.json: 9900 => 9900 [+0 bytes / +0.00%]
codesize/test_codesize_minimal_O0.json: 20467 => 20467 [+0 bytes / +0.00%]
codesize/test_codesize_minimal_pthreads.json: 26409 => 26409 [+0 bytes / +0.00%]
codesize/test_codesize_minimal_pthreads_memgrowth.json: 26812 => 26812 [+0 bytes / +0.00%]
codesize/test_minimal_runtime_code_size_hello_embind.json: 14909 => 14909 [+0 bytes / +0.00%]
codesize/test_minimal_runtime_code_size_hello_embind_val.json: 11642 => 11642 [+0 bytes / +0.00%]
codesize/test_minimal_runtime_code_size_hello_webgl2_wasm.json: 13200 => 13200 [+0 bytes / +0.00%]
codesize/test_minimal_runtime_code_size_hello_webgl2_wasm2js.json: 18534 => 18534 [+0 bytes / +0.00%]
codesize/test_minimal_runtime_code_size_hello_webgl2_wasm_singlefile.json: 15046 => 15046 [+0 bytes / +0.00%]
codesize/test_minimal_runtime_code_size_hello_webgl_wasm.json: 12738 => 12738 [+0 bytes / +0.00%]
codesize/test_minimal_runtime_code_size_hello_webgl_wasm2js.json: 18060 => 18060 [+0 bytes / +0.00%]
codesize/test_minimal_runtime_code_size_random_printf_wasm.json: 10815 => 10815 [+0 bytes / +0.00%]
codesize/test_minimal_runtime_code_size_random_printf_wasm2js.json: 17114 => 17114 [+0 bytes / +0.00%]
codesize/test_unoptimized_code_size.json: 180230 => 180230 [+0 bytes / +0.00%]

Average change: +0.00% (+0.00% - +0.00%)

This looks weird to me

[sbc100]

This looks weird to me

Hmm.. yes that does looks strange. That can happen sometimes when the script doesn't know how to parse the json file in question correctly.

Are you able to run ./test/runner codesize --rebase on the main branch without generating any diffs? Unless you can do that there is no hope of generated the correct results on your PR branch.

If you are not able to do that, and you have tried ./emsdk install tot again, and it still doesn't work I can also upload the new expections to this PR if you like.

[kleisauke]

Regarding the codesize expectations, you may also find this Dockerfile relevant:
#23990 (comment)

The issue is that gzip-compressed data may not be bit-identical across different zlib versions (it looks like you're using zlib-ng as well).

[thiblahute]

Regarding the codesize expectations, you may also find this Dockerfile relevant: #23990 (comment)

The issue is that gzip-compressed data may not be bit-identical across different zlib versions (it looks like you're using zlib-ng as well).

I used the docker file to regenerate with

 podman run --rm -v $(pwd):/src ./test/runner emscripten-codesize codesize --rebase

and it is still failing

[sbc100]

I doubt there are docker images available for the tot build which is what you need to be rebasing against.

[thiblahute]

I doubt there are docker images available for the tot build which is what you need to be rebasing against.

I built it myself based on this dockerfile

Let me rebuild in case it was bad timing 🤷

[sbc100]

I pushed a change to this branch with the updated codesizes.