Fix wireguard-go-rs build for different Android targets

[MarkusPettersson98]

This PR addresses several issues with building libwg.so for the supported Android targets. It removes wireguard-go-rs/libwg/build-android.sh in favor of wireguard-go-rs/build.rs.

The biggest change is that only the actual target is built, and not all 4 Android targets which was the case for build-android.sh.

---

This change is 

[Pururun]

Reviewed 3 of 3 files at r1, all commit messages.
Reviewable status: complete! all files reviewed, all discussions resolved

[Rawa]

Reviewed all commit messages.
Reviewable status: all files reviewed, 2 unresolved discussions (waiting on @MarkusPettersson98)

---

wireguard-go-rs/build.rs line 166 at r1 (raw file):

/// Compile libwg as a dynamic library for android and place it in `OUT_DIR`.
// NOTE: We use dynamic linking as Go cannot produce static binaries specifically for Android.
fn build_android_dynamic_lib(daita: bool) -> anyhow::Result<()> {

Does it make sense to toggle daita? I don't think we adapt the UI to this feature flag, so it might be problems if we generate a binary without the daita support while UI thinks we do have it?

---

wireguard-go-rs/build.rs line 226 at r1 (raw file):

    };

    // Create the directory with RWX permissions for all users so that the build server can clean

Do we need RWX, would RW- be enough?

[MarkusPettersson98]

Reviewable status: all files reviewed, 2 unresolved discussions (waiting on @Rawa)

---

wireguard-go-rs/build.rs line 166 at r1 (raw file):

Previously, Rawa (David Göransson) wrote…

Does it make sense to toggle daita? I don't think we adapt the UI to this feature flag, so it might be problems if we generate a binary without the daita support while UI thinks we do have it?

I mostly did it like this to be in line with how libwg is built for linux and macOS (see build_static_lib). Since adding support for DAITA is not part of this PR, the daita: bool argument is just preparatory work for #6684. Either the argument is simply set to true in that PR, or we remove this argument and make the subsequent changes #6684 such that we always assume that libwg should be compiled with DAITA 😊

---

wireguard-go-rs/build.rs line 226 at r1 (raw file):

Previously, Rawa (David Göransson) wrote…

Do we need RWX, would RW- be enough?

RW- ought to be enough, but I would prefer if we didn't need to muck with permissions at all 🤔

[Rawa]

Reviewable status: 2 of 3 files reviewed, all discussions resolved (waiting on @Pururun)

[albin-mullvad]

Reviewed 2 of 3 files at r1, 1 of 1 files at r2, all commit messages.
Reviewable status: all files reviewed, 1 unresolved discussion (waiting on @MarkusPettersson98)

---

wireguard-go-rs/build.rs line 59 at r2 (raw file):

            "armv7-linux-androideabi" => Ok(Armv7),
            "i686-linux-android" => Ok(I686),
            _ => bail!("{input} is not a support android target!"),

typo: -> supported

Code quote:

support

[MarkusPettersson98]

Reviewable status: 2 of 3 files reviewed, all discussions resolved (waiting on @albin-mullvad)

---

wireguard-go-rs/build.rs line 59 at r2 (raw file):

Previously, albin-mullvad wrote…

typo: -> supported

Done

[Rawa]

Reviewed 1 of 1 files at r3, all commit messages.
Reviewable status: all files reviewed, 1 unresolved discussion (waiting on @MarkusPettersson98)

---

wireguard-go-rs/build.rs line 221 at r3 (raw file):

/// Note: This function will create the parent directory/directories to `output` if necessary.
fn android_move_binary(binary: &Path, output: &Path) -> anyhow::Result<()> {
    // Create the directory with RWX permissions for all users so that the build server can clean

Should we remove/update the comment?

[MarkusPettersson98]

Reviewable status: complete! all files reviewed, all discussions resolved

---

wireguard-go-rs/build.rs line 221 at r3 (raw file):

Previously, Rawa (David Göransson) wrote…

Should we remove/update the comment?

🤦
Fixed!

[hulthe]

Reviewable status: 2 of 3 files reviewed, 2 unresolved discussions (waiting on @MarkusPettersson98 and @Rawa)

---

wireguard-go-rs/build.rs line 210 at r1 (raw file):

    // If daita is enabled, also enable the corresponding rust feature flag
    if daita {

since daita isn't implemented yet I think we should bail! here just to make that clear

---

wireguard-go-rs/libwg/Android.mk line 7 at r3 (raw file):

DESTDIR ?= $(OUT_DIR)
CARGO_TARGET_DIR ?=
TARGET ?=

What is TARGET and CARGO_TARGET_DIR used for?

[MarkusPettersson98]

Reviewable status: 2 of 3 files reviewed, 2 unresolved discussions (waiting on @hulthe and @Rawa)

---

wireguard-go-rs/build.rs line 210 at r1 (raw file):

Previously, hulthe (Joakim Hulthe) wrote…

since daita isn't implemented yet I think we should bail! here just to make that clear

While you are right, it will be enabled in #6684 so the window of potential confusion is rather small 😊

Do you think I should remove the bool?

---

wireguard-go-rs/libwg/Android.mk line 7 at r3 (raw file):

Previously, hulthe (Joakim Hulthe) wrote…

What is TARGET and CARGO_TARGET_DIR used for?

Preparatory work, could be pushed to #6684 I suppose 🤷

[hulthe]

Reviewable status: 2 of 3 files reviewed, 2 unresolved discussions (waiting on @MarkusPettersson98 and @Rawa)

---

wireguard-go-rs/build.rs line 210 at r1 (raw file):

Previously, MarkusPettersson98 (Markus Pettersson) wrote…

While you are right, it will be enabled in #6684 so the window of potential confusion is rather small 😊

Do you think I should remove the bool?

Ah ok, If it's going to be fixed soon then I don't mind

---

wireguard-go-rs/libwg/Android.mk line 7 at r3 (raw file):

Previously, MarkusPettersson98 (Markus Pettersson) wrote…

Preparatory work, could be pushed to #6684 I suppose 🤷

Ah ok, If it's going to be fixed soon then I don't mind

[hulthe]

Reviewable status: 2 of 3 files reviewed, all discussions resolved (waiting on @Rawa)