[FS] Make fstat work on file descriptors with no name in memfs

[hoodmane]

This makes fstat work on anonymous memfs file descriptors. It is split off from #23058.

[hoodmane]

So code size went down by a little. But now ./tools/maint/rebaseline_tests.py doesn't update anything when I run it locally but check-expectations is mad. hmm...

[hoodmane]

Well it's always possible to directly copy the ci output. Maybe check_expectations could print something like the following when it fails so I could paste it into the terminal and apply the updates that CI wanted:

echo 129211 > test/other/codesize/test_codesize_cxx_ctors1.size
echo 128623 > test/other/codesize/test_codesize_cxx_ctors2.size
echo 170879 > test/other/codesize/test_codesize_cxx_except.size
echo 144708 > test/other/codesize/test_codesize_cxx_except_wasm.size
echo 142162 > test/other/codesize/test_codesize_cxx_except_wasm_legacy.size
echo 232724 > test/other/codesize/test_codesize_cxx_mangle.size
echo 131774 > test/other/codesize/test_codesize_cxx_noexcept.size
echo 169181 > test/other/codesize/test_codesize_cxx_wasmfs.size
echo 15101 > test/other/codesize/test_codesize_hello_O0.size
echo 19394 > test/other/codesize/test_codesize_minimal_pthreads.size
echo 15101 > test/other/test_unoptimized_code_size.wasm.size
echo 12182 > test/other/test_unoptimized_code_size_no_asserts.wasm.size
echo 15101 > test/other/test_unoptimized_code_size_strict.wasm.size

[sbc100]

Well it's always possible to directly copy the ci output. Maybe check_expectations could print something like the following when it fails so I could paste it into the terminal and apply the updates that CI wanted:

echo 129211 > test/other/codesize/test_codesize_cxx_ctors1.size
echo 128623 > test/other/codesize/test_codesize_cxx_ctors2.size
echo 170879 > test/other/codesize/test_codesize_cxx_except.size
echo 144708 > test/other/codesize/test_codesize_cxx_except_wasm.size
echo 142162 > test/other/codesize/test_codesize_cxx_except_wasm_legacy.size
echo 232724 > test/other/codesize/test_codesize_cxx_mangle.size
echo 131774 > test/other/codesize/test_codesize_cxx_noexcept.size
echo 169181 > test/other/codesize/test_codesize_cxx_wasmfs.size
echo 15101 > test/other/codesize/test_codesize_hello_O0.size
echo 19394 > test/other/codesize/test_codesize_minimal_pthreads.size
echo 15101 > test/other/test_unoptimized_code_size.wasm.size
echo 12182 > test/other/test_unoptimized_code_size_no_asserts.wasm.size
echo 15101 > test/other/test_unoptimized_code_size_strict.wasm.size

I'm hoping to have the CI automatically add make those change perhaps?

In order to get the same results as the CI you need to make sure you do emsdk install tot and then emcc --clear-cache before rebaselineing.

[hoodmane]

In order to get the same results as the CI you need to make sure you do emsdk install tot and then emcc --clear-cache before rebaselineing.

I think I have been doing that, but it seems to work only some of the time...

[sbc100]

In order to get the same results as the CI you need to make sure you do emsdk install tot and then emcc --clear-cache before rebaselineing.

I think I have been doing that, but it seems to work only some of the time...

Oh, there is also the case when main can actually be out of date and needs updating.

This happens when llvm of binaryen changes effect code sizes. In that case I normally push a specific update commit, which I can look into doing now.

[sbc100]

Indeed if you looks at the failing CI test its called Check test expectation on target branch .. which means that expectations on main are failing, unrelated to your change.

I updated main in 12ca58d

[hoodmane]

Okay I rearranged the code in attempt to address your comments. Does it look better?

[sbc100]

Is this a separate fix?

[hoodmane]

Yeah