deps: upgrade to libuv 1.44.2 #42340

node/test/parallel/test-os.js

Line 215 in d4d0b09

assert.strictEqual(pwd.uid, -1);

pwd.uid is 4294967295 instead of -1. The culprit here seems to be libuv/libuv@f3e0bff.
node/test/parallel/test-timers-immediate-unref-nested-once.js

Line 8 in d4d0b09

setImmediate(common.mustNotCall()).unref();

The callback is called when it should not. The offending commit is libuv/libuv@939a056.

cc: @nodejs/libuv @vtjnash

RaisinTen · 2022-03-16T04:32:36Z

Yes, I can confirm that libuv/libuv#3193 caused the EPERM and EACCES mismatches and yes, this change should be labelled as semver-major because other modules like https://github.com/isaacs/rimraf also depends on the error code being EPERM instead of EACCES.

cc @isaacs

richardlau · 2022-03-16T11:08:03Z

Yes, I can confirm that libuv/libuv#3193 caused the EPERM and EACCES mismatches and yes, this change should be labelled as semver-major because other modules like https://github.com/isaacs/rimraf also depends on the error code being EPERM instead of EACCES.

This means we will be unable to update libuv in LTS to this or subsequent versions.

lpinca · 2022-03-16T12:10:36Z

This patch

diff --git a/deps/uv/src/win/core.c b/deps/uv/src/win/core.c
index c08dedc47c..cfd52e5148 100644
--- a/deps/uv/src/win/core.c
+++ b/deps/uv/src/win/core.c
@@ -401,7 +401,6 @@ int uv_loop_fork(uv_loop_t* loop) {
 static int uv__loop_alive(const uv_loop_t* loop) {
   return uv__has_active_handles(loop) ||
          uv__has_active_reqs(loop) ||
-         loop->pending_reqs_tail != NULL ||
          loop->endgame_handles != NULL;
 }

fixes the third issue (point 2 in my previous comment) but I'm not sure if it hides another issue/bug.

RaisinTen · 2022-03-16T15:34:18Z

This means we will be unable to update libuv in LTS to this or subsequent versions.

@richardlau is that something we are okay with or should we revert the change in the libuv repo?

aduh95 · 2022-03-17T01:07:05Z

621ae7f could land on a separate PR, as it is not semver-major.

node/doc/contributing/collaborator-guide.md

Lines 513 to 514 in a199387

    
           All commits should be self-contained, meaning every commit should pass all 
        
           tests. This makes it much easier when bisecting to find a breaking change.

We would need to either squash the commits when merging, or add a first commit that disables the test that the update break, or make an exception (see #42356 (comment)).

richardlau · 2022-03-17T01:22:06Z

This means we will be unable to update libuv in LTS to this or subsequent versions.

@richardlau is that something we are okay with or should we revert the change in the libuv repo?

On the one hand if we say "we'll never update libuv again in any of the existing LTS releases" then that's less work for releasers but I don't feel comfortable in making that statement. We could float patches but that's not particularly friendly to others who build their own packages of Node.js (Linux distros, homebrew etc) who dynamically link against a shared libuv library.

My question is, given the change would cause ecosystem breakage, is it worth it?

lpinca · 2022-03-17T06:02:02Z

621ae7f could land on a separate PR, as it is not semver-major.

It could, but it does not make sense without the libuv update.

targos · 2022-03-17T06:03:25Z

My question is, given the change would cause ecosystem breakage, is it worth it?

If it breaks things as popular as rimraf, I'd say it's not.

Notable changes: - Build regression fixes for various platform updates (libuv/libuv#3428, libuv/libuv#3419, libuv/libuv#3423, libuv/libuv#3413, libuv/libuv#3431) - Support for GNU/Hurd (libuv/libuv#3450) - Release tool improvements (libuv/libuv-release-tool#13) - Better performing rw locks on Win32 (libuv/libuv#3383) - Support for posix_spawn API (libuv/libuv#3257) - Fix regression on OpenBSD (libuv/libuv#3506) - Add uv_available_parallelism() (libuv/libuv#3499) - Don't use thread-unsafe strtok() (libuv/libuv#3524) - Fix hang after NOTE_EXIT (libuv/libuv#3521) - Better align order-of-events behavior between platforms (libuv/libuv#3598) - Fix fs event not fired if the watched file is moved/removed/recreated (libuv/libuv#3540) - Fix pipe resource leak if closed during connect (and other bugs) (libuv/libuv#3611) - Don't error when killing a zombie process (libuv/libuv#3625) - Avoid posix_spawnp() cwd bug (libuv/libuv#3597) - Skip EVFILT_PROC events when invalidating events for an fd (libuv/libuv#3629) Fixes: nodejs#42290 PR-URL: nodejs#42340 Reviewed-By: Michaël Zasso <targos@protonmail.com> Reviewed-By: Juan José Arboleda <soyjuanarbol@gmail.com> Reviewed-By: Yagiz Nizipli <yagiz@nizipli.com> Reviewed-By: Darshan Sen <raisinten@gmail.com> Reviewed-By: Antoine du Hamel <duhamelantoine1995@gmail.com>

Make `os.userInfo()` convert the `uid` and `gid` fields to 32-bit signed integers on Windows. PR-URL: #42340 Refs: libuv/libuv@f3e0bffcb14 Reviewed-By: Michaël Zasso <targos@protonmail.com> Reviewed-By: Juan José Arboleda <soyjuanarbol@gmail.com> Reviewed-By: Yagiz Nizipli <yagiz@nizipli.com> Reviewed-By: Darshan Sen <raisinten@gmail.com> Reviewed-By: Antoine du Hamel <duhamelantoine1995@gmail.com>

The `'error'` event is not emitted because the socket is already destroyed when `socket.write()` is called. Use the `socket.write()` callback instead. PR-URL: #42340 Refs: #42340 (comment) Reviewed-By: Michaël Zasso <targos@protonmail.com> Reviewed-By: Juan José Arboleda <soyjuanarbol@gmail.com> Reviewed-By: Yagiz Nizipli <yagiz@nizipli.com> Reviewed-By: Darshan Sen <raisinten@gmail.com> Reviewed-By: Antoine du Hamel <duhamelantoine1995@gmail.com>

Do not delay the call to `stream.end()` too much. PR-URL: #42340 Refs: #42340 (comment) Reviewed-By: Michaël Zasso <targos@protonmail.com> Reviewed-By: Juan José Arboleda <soyjuanarbol@gmail.com> Reviewed-By: Yagiz Nizipli <yagiz@nizipli.com> Reviewed-By: Darshan Sen <raisinten@gmail.com> Reviewed-By: Antoine du Hamel <duhamelantoine1995@gmail.com>

Do not immediately check the `tcpserver` hook invocations when it closes. Do it in the next iteration of the event loop. PR-URL: #42340 Refs: #42340 (comment) Reviewed-By: Michaël Zasso <targos@protonmail.com> Reviewed-By: Juan José Arboleda <soyjuanarbol@gmail.com> Reviewed-By: Yagiz Nizipli <yagiz@nizipli.com> Reviewed-By: Darshan Sen <raisinten@gmail.com> Reviewed-By: Antoine du Hamel <duhamelantoine1995@gmail.com>

Notable changes: - Build regression fixes for various platform updates (libuv/libuv#3428, libuv/libuv#3419, libuv/libuv#3423, libuv/libuv#3413, libuv/libuv#3431) - Support for GNU/Hurd (libuv/libuv#3450) - Release tool improvements (libuv/libuv-release-tool#13) - Better performing rw locks on Win32 (libuv/libuv#3383) - Support for posix_spawn API (libuv/libuv#3257) - Fix regression on OpenBSD (libuv/libuv#3506) - Add uv_available_parallelism() (libuv/libuv#3499) - Don't use thread-unsafe strtok() (libuv/libuv#3524) - Fix hang after NOTE_EXIT (libuv/libuv#3521) - Better align order-of-events behavior between platforms (libuv/libuv#3598) - Fix fs event not fired if the watched file is moved/removed/recreated (libuv/libuv#3540) - Fix pipe resource leak if closed during connect (and other bugs) (libuv/libuv#3611) - Don't error when killing a zombie process (libuv/libuv#3625) - Avoid posix_spawnp() cwd bug (libuv/libuv#3597) - Skip EVFILT_PROC events when invalidating events for an fd (libuv/libuv#3629) Fixes: #42290 PR-URL: #42340 Reviewed-By: Michaël Zasso <targos@protonmail.com> Reviewed-By: Juan José Arboleda <soyjuanarbol@gmail.com> Reviewed-By: Yagiz Nizipli <yagiz@nizipli.com> Reviewed-By: Darshan Sen <raisinten@gmail.com> Reviewed-By: Antoine du Hamel <duhamelantoine1995@gmail.com>