Avoid adding data (vs compile_data) to Rustc actions (#3916)

relates to #3915
closes #3609

Author: UebelAndre

cargo/tests/cargo_build_script/transitive_data/BUILD.bazel

@@ -0,0 +1,5 @@
+load(":transitive_data_test.bzl", "transitive_cbs_data_test_suite")
+
+transitive_cbs_data_test_suite(
+    name = "transitive_cbs_data_test_suite",
+)

cargo/tests/cargo_build_script/transitive_data/transitive_data_test.bzl

@@ -0,0 +1,115 @@
+"""Tests documenting that cargo_build_script `data` files currently act as both
+compile_data and runtime data -- they appear in Rustc compile action inputs for
+both the direct library and transitive dependents.
+
+See https://github.com/bazelbuild/rules_rust/issues/3609 for context."""
+
+load("@bazel_skylib//lib:unittest.bzl", "analysistest", "asserts")
+load("@bazel_skylib//rules:write_file.bzl", "write_file")
+load("//cargo:defs.bzl", "cargo_build_script")
+load("//rust:defs.bzl", "rust_library")
+
+def _cbs_data_in_rustc_inputs_impl(ctx):
+    env = analysistest.begin(ctx)
+    target = analysistest.target_under_test(env)
+
+    rustc_action = None
+    for action in target.actions:
+        if action.mnemonic == "Rustc":
+            rustc_action = action
+            break
+
+    asserts.false(env, rustc_action == None, "Expected a Rustc action")
+
+    # cargo_build_script `data` currently flows into BuildInfo.compile_data
+    # via script_data, so it appears in Rustc action inputs for both the
+    # direct dependent and transitive dependents. This documents that
+    # CBS `data` effectively acts as both compile_data and data today.
+    data_inputs = [i for i in rustc_action.inputs.to_list() if "cbs_data_dep.txt" in i.path]
+    asserts.true(
+        env,
+        len(data_inputs) > 0,
+        "Expected CBS data file to appear in Rustc action inputs (CBS data currently acts as compile_data)",
+    )
+
+    return analysistest.end(env)
+
+cbs_data_in_rustc_inputs_test = analysistest.make(
+    _cbs_data_in_rustc_inputs_impl,
+)
+
+def _define_test_targets():
+    write_file(
+        name = "cbs_data_dep_file",
+        out = "cbs_data_dep.txt",
+        content = ["data for build script", ""],
+        newline = "unix",
+    )
+
+    write_file(
+        name = "build_rs_src",
+        out = "build.rs",
+        content = ["fn main() {}", ""],
+        newline = "unix",
+    )
+
+    cargo_build_script(
+        name = "build_script",
+        srcs = [":build.rs"],
+        data = [":cbs_data_dep.txt"],
+        edition = "2021",
+    )
+
+    write_file(
+        name = "lib_src",
+        out = "lib.rs",
+        content = ["pub fn hello() {}", ""],
+        newline = "unix",
+    )
+
+    rust_library(
+        name = "lib",
+        srcs = [":lib.rs"],
+        deps = [":build_script"],
+        edition = "2021",
+    )
+
+    write_file(
+        name = "bin_src",
+        out = "bin.rs",
+        content = ["extern crate lib;", ""],
+        newline = "unix",
+    )
+
+    rust_library(
+        name = "bin",
+        srcs = [":bin.rs"],
+        deps = [":lib"],
+        edition = "2021",
+    )
+
+def transitive_cbs_data_test_suite(name):
+    """Entry-point macro called from the BUILD file.
+
+    Args:
+        name (str): Name of the macro.
+    """
+    _define_test_targets()
+
+    cbs_data_in_rustc_inputs_test(
+        name = "cbs_data_in_lib_compile_inputs_test",
+        target_under_test = ":lib",
+    )
+
+    cbs_data_in_rustc_inputs_test(
+        name = "cbs_data_in_bin_compile_inputs_test",
+        target_under_test = ":bin",
+    )
+
+    native.test_suite(
+        name = name,
+        tests = [
+            ":cbs_data_in_lib_compile_inputs_test",
+            ":cbs_data_in_bin_compile_inputs_test",
+        ],
+    )

rust/private/rustc.bzl

@@ -858,6 +858,7 @@ def collect_inputs(
         build_info = build_info,
         dep_info = dep_info,
         include_link_flags = include_link_flags,
+        include_transitive_data = not toolchain._incompatible_do_not_include_transitive_data_in_compile_inputs,
     )
 
     # TODO(parkmycar): Cleanup the handling of lint_files here.
@@ -1994,13 +1995,15 @@ def add_edition_flags(args, crate):
 def _process_build_scripts(
         build_info,
         dep_info,
-        include_link_flags = True):
+        include_link_flags = True,
+        include_transitive_data = False):
     """Gathers the outputs from a target's `cargo_build_script` action.
 
     Args:
         build_info (BuildInfo): The target Build's dependency info.
         dep_info (DepInfo): The Depinfo provider form the target Crate's set of inputs.
         include_link_flags (bool, optional): Whether to include flags like `-l` that instruct the linker to search for a library.
+        include_transitive_data (bool, optional): Whether to include transitive data dependencies in compile inputs.
 
     Returns:
         tuple: A tuple: A tuple of the following items:
@@ -2010,7 +2013,9 @@ def _process_build_scripts(
             - (depset[File]): All direct and transitive build flags from the current build info.
     """
     direct_inputs = []
-    transitive_inputs = [dep_info.link_search_path_files, dep_info.transitive_data]
+    transitive_inputs = [dep_info.link_search_path_files]
+    if include_transitive_data:
+        transitive_inputs.append(dep_info.transitive_data)
 
     # Arguments to the commandline line wrapper that are going to be used
     # to create the final command line
@@ -2037,6 +2042,7 @@ def _process_build_scripts(
     for dep_build_info in dep_info.transitive_build_infos.to_list():
         if dep_build_info.out_dir:
             direct_inputs.append(dep_build_info.out_dir)
+        transitive_inputs.append(dep_build_info.compile_data)
 
     out_dir_compile_inputs = depset(
         direct_inputs,

rust/settings/BUILD.bazel

@@ -27,6 +27,7 @@ load(
     "extra_rustc_flags",
     "incompatible_change_clippy_error_format",
     "incompatible_do_not_include_data_in_compile_data",
+    "incompatible_do_not_include_transitive_data_in_compile_inputs",
     "lto",
     "no_std",
     "pipelined_compilation",
@@ -111,6 +112,8 @@ incompatible_change_clippy_error_format()
 
 incompatible_do_not_include_data_in_compile_data()
 
+incompatible_do_not_include_transitive_data_in_compile_inputs()
+
 lto()
 
 no_std()

rust/settings/settings.bzl

@@ -535,6 +535,15 @@ def incompatible_do_not_include_data_in_compile_data():
         issue = "https://github.com/bazelbuild/rules_rust/issues/2977",
     )
 
+# buildifier: disable=unnamed-macro
+def incompatible_do_not_include_transitive_data_in_compile_inputs():
+    """A flag to control whether transitive data dependencies are included in compile inputs."""
+    incompatible_flag(
+        name = "incompatible_do_not_include_transitive_data_in_compile_inputs",
+        build_setting_default = True,
+        issue = "https://github.com/bazelbuild/rules_rust/issues/3915",
+    )
+
 def codegen_units():
     """The default value for `--codegen-units` which also affects resource allocation for rustc actions.
 

rust/toolchain.bzl

@@ -623,6 +623,7 @@ def _rust_toolchain_impl(ctx):
         _experimental_use_coverage_metadata_files = ctx.attr._experimental_use_coverage_metadata_files[BuildSettingInfo].value,
         _toolchain_generated_sysroot = ctx.attr._toolchain_generated_sysroot[BuildSettingInfo].value,
         _incompatible_do_not_include_data_in_compile_data = ctx.attr._incompatible_do_not_include_data_in_compile_data[IncompatibleFlagInfo].enabled,
+        _incompatible_do_not_include_transitive_data_in_compile_inputs = ctx.attr._incompatible_do_not_include_transitive_data_in_compile_inputs[IncompatibleFlagInfo].enabled,
         _no_std = no_std,
         _codegen_units = ctx.attr._codegen_units[BuildSettingInfo].value,
         _experimental_use_allocator_libraries_with_mangled_symbols = ctx.attr.experimental_use_allocator_libraries_with_mangled_symbols,
@@ -866,6 +867,10 @@ rust_toolchain = rule(
             default = Label("//rust/settings:incompatible_do_not_include_data_in_compile_data"),
             doc = "Label to a boolean build setting that controls whether to include data files in compile_data.",
         ),
+        "_incompatible_do_not_include_transitive_data_in_compile_inputs": attr.label(
+            default = Label("//rust/settings:incompatible_do_not_include_transitive_data_in_compile_inputs"),
+            doc = "Label to a boolean build setting that controls whether to include transitive data dependencies in compile inputs.",
+        ),
         "_linker_preference": attr.label(
             default = Label("//rust/settings:toolchain_linker_preference"),
         ),