fix: prevent RUSTFLAGS from leaking into downstream cargo-make rust-script builds by wmmc88 · Pull Request #629 · microsoft/windows-drivers-rs

wmmc88 · 2026-03-07T22:53:40Z

Problem

When downstream consumers of wdk-build (e.g. microsoft/Windows-rust-driver-samples) run cargo make, RUSTFLAGS=-D warnings leaks into rust-script compilation of the inline build scripts. This causes warnings in the published wdk-build crate to become hard errors, breaking downstream CI.

Example failure (from Windows-rust-driver-samples Build CI):

error: the feature `assert_matches` has been stable since 1.95.0-nightly
  --> wdk-build-0.5.1/src/lib.rs:13:40
   = note: `-D stable-features` implied by `-D warnings`

Root Cause

The rust-script tasks in rust-driver-makefile.toml specify wdk-build = { path = ".", version = "0.5.1" }. Cargo always uses path when both are specified, ignoring version. When load_rust_driver_makefile() symlinks the makefile from the registry cache, path = "." resolves to the registry source via --base-path, making wdk-build a path dependency. Cargo does not apply --cap-lints allow to path dependencies, so RUSTFLAGS=-D warnings applies directly to the published crate.

Normal cargo build is unaffected because wdk-build is resolved as a registry dependency there.

Fix

load_wdk_build_makefile() now checks package.source from cargo_metadata:
- Registry dep (source.is_some()): Copies the makefile and rewrites wdk-build = { path = "." } to wdk-build = "X.Y.Z" (version from metadata). This ensures cargo treats it as a registry dep and applies --cap-lints allow.
- Path/workspace dep (source.is_none()): Preserves existing symlink behavior so local development works correctly.
Removed unused version field from all 14 rust-script dep specs in rust-driver-makefile.toml and rust-driver-sample-makefile.toml, since path always takes precedence. Added a comment explaining the automatic rewrite by load_rust_driver_makefile().

Scenarios

Scenario	`package.source`	Behavior
Upstream WDR dev	N/A (uses `extend`, not `load_*`)	Unchanged
Downstream registry dep	`Some(registry+...)`	Copy + rewrite -> registry dep -> `--cap-lints`
Downstream path dep	`None`	Symlink preserved -> path dep (intentional)

Copilot

Pull request overview

This PR prevents downstream cargo make consumers of wdk-build from inheriting caller RUSTFLAGS (e.g. -D warnings) into rust-script compilation of embedded scripts by ensuring wdk-build is treated as a non-path dependency when loaded from the registry.

Changes:

Update load_wdk_build_makefile() to copy+rewrite embedded rust-script manifests from wdk-build = { path = "." } to wdk-build = "X.Y.Z" when wdk-build is resolved from a non-path source.
Preserve existing symlink behavior when wdk-build is a workspace/path dependency to keep local dev flows intact.
Remove redundant version = "...“ fields from rust-script embedded dependency specs and add an explanatory note in the main makefile.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 3 comments.

File	Description
crates/wdk-build/src/cargo_make.rs	Implements registry-vs-path behavior (copy+rewrite vs symlink) when materializing makefiles into `target/`.
crates/wdk-build/rust-driver-makefile.toml	Removes redundant embedded dependency versions and documents the downstream rewrite behavior.
crates/wdk-build/rust-driver-sample-makefile.toml	Removes redundant embedded dependency versions from sample tasks.

Comments suppressed due to low confidence (2)

crates/wdk-build/src/cargo_make.rs:934

The rewrite uses a simple replace without validating that any occurrences were actually patched. If the makefile format changes (spacing/comments/features), this will silently leave the path = "." dependency intact and the original lint-leak issue will persist. Consider checking whether patched_content differs from makefile_content and returning a ConfigError (or at least logging) when no replacement occurred.

        let version = &wdk_build_package.version;
        let patched_content = makefile_content.replace(
            r#"wdk-build = { path = "." }"#,
            &format!("wdk-build = \"{version}\""),
        );

crates/wdk-build/src/cargo_make.rs:927

load_wdk_build_makefile now sometimes writes a patched copy into target/ instead of always creating/updating a symlink. The rustdoc/comments above this function (and the public load_*_makefile wrappers) still describe the behavior as always symlinking; they should be updated to reflect the new copy+rewrite path for registry dependencies.

    if wdk_build_package.source.is_some() {
        // wdk-build is a registry (or git) dependency. Rewrite path dependencies in
        // the makefile's rust-script embedded manifests to version-only dependencies so
        // that cargo treats wdk-build as a registry dep and applies --cap-lints. This
        // prevents the caller's RUSTFLAGS (e.g. -D warnings) from leaking into the
        // published wdk-build crate's compilation.
        let makefile_content = std::fs::read_to_string(&rust_driver_makefile_toml_path)

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-03-07T22:57:47Z

crates/wdk-build/src/cargo_make.rs

+    if wdk_build_package.source.is_some() {
+        // wdk-build is a registry (or git) dependency. Rewrite path dependencies in
+        // the makefile's rust-script embedded manifests to version-only dependencies so
+        // that cargo treats wdk-build as a registry dep and applies --cap-lints. This
+        // prevents the caller's RUSTFLAGS (e.g. -D warnings) from leaking into the
+        // published wdk-build crate's compilation.
+        let makefile_content = std::fs::read_to_string(&rust_driver_makefile_toml_path)
+            .map_err(|source| IoError::with_path(&rust_driver_makefile_toml_path, source))?;
+
+        let version = &wdk_build_package.version;
+        let patched_content = makefile_content.replace(
+            r#"wdk-build = { path = "." }"#,
+            &format!("wdk-build = \"{version}\""),
+        );
+
+        // Only write if content changed or file doesn't exist, to avoid unnecessary
+        // rebuilds from rust-script cache invalidation
+        let should_write = if destination_path.exists() {
+            let existing_content = std::fs::read_to_string(&destination_path)
+                .map_err(|source| IoError::with_path(&destination_path, source))?;
+            existing_content != patched_content
+        } else {
+            true
+        };
+
+        if should_write {
+            if destination_path.exists() {
+                std::fs::remove_file(&destination_path)
+                    .map_err(|source| IoError::with_path(&destination_path, source))?;
+            }
+            std::fs::write(&destination_path, patched_content)
+                .map_err(|source| IoError::with_path(&destination_path, source))?;
+        }
+    } else {


This change introduces new behavior (copying and patching the makefile based on cargo_metadata source) but there are no tests covering the rewrite decision or the resulting file content. Since this module already has unit tests, consider adding tests (e.g., using temp dirs + a small fixture makefile, and injecting/stubbing metadata) to lock in registry-vs-path behavior and ensure the wdk-build dep is rewritten as intended.

This issue also appears on line 930 of the same file.

Copilot · 2026-03-07T22:57:47Z

crates/wdk-build/src/cargo_make.rs

+    if wdk_build_package.source.is_some() {
+        // wdk-build is a registry (or git) dependency. Rewrite path dependencies in
+        // the makefile's rust-script embedded manifests to version-only dependencies so
+        // that cargo treats wdk-build as a registry dep and applies --cap-lints. This
+        // prevents the caller's RUSTFLAGS (e.g. -D warnings) from leaking into the
+        // published wdk-build crate's compilation.
+        let makefile_content = std::fs::read_to_string(&rust_driver_makefile_toml_path)
+            .map_err(|source| IoError::with_path(&rust_driver_makefile_toml_path, source))?;
+
+        let version = &wdk_build_package.version;
+        let patched_content = makefile_content.replace(
+            r#"wdk-build = { path = "." }"#,
+            &format!("wdk-build = \"{version}\""),
+        );


wdk_build_package.source.is_some() groups registry and git dependencies together, but rewriting to wdk-build = "{version}" will break consumers using a git dependency (Cargo will try to resolve that version from the registry). Consider detecting source kind (e.g., registry+... vs git+...) and only performing the version-only rewrite for registry sources; for git sources, either preserve the symlink behavior or rewrite to a git dependency in the embedded manifests.

This issue also appears on line 921 of the same file.

Copilot · 2026-03-07T22:57:47Z

crates/wdk-build/rust-driver-makefile.toml

+//! # `load_rust_driver_makefile()`, this path dependency is automatically
+//! # rewritten to a version-only registry dependency. This ensures cargo
+//! # applies `--cap-lints allow` to wdk-build, preventing the caller's
+//! # RUSTFLAGS from leaking into the published crate's compilation.


The NOTE says the path = "." dependency is rewritten when loaded by downstream consumers, but the implementation only rewrites when wdk-build is resolved from a non-path source (registry/git). Downstream users can also depend on wdk-build via a local path, in which case the rewrite intentionally does not happen. Consider clarifying the note to mention it only applies when wdk-build is a registry dependency.

Suggested change

//! # `load_rust_driver_makefile()`, this path dependency is automatically

//! # rewritten to a version-only registry dependency. This ensures cargo

//! # applies `--cap-lints allow` to wdk-build, preventing the caller's

//! # RUSTFLAGS from leaking into the published crate's compilation.

//! # `load_rust_driver_makefile()`, and `wdk-build` is resolved from a

//! # non-path source (e.g., registry or git), this path dependency is

//! # automatically rewritten to a version-only registry dependency. This

//! # ensures cargo applies `--cap-lints allow` to wdk-build, preventing the

//! # caller's RUSTFLAGS from leaking into the published crate's compilation.

codecov-commenter · 2026-03-07T23:19:52Z

Codecov Report

❌ Patch coverage is 93.21429% with 19 lines in your changes missing coverage. Please review.
✅ Project coverage is 78.63%. Comparing base (666875a) to head (bcdf058).

Files with missing lines	Patch %	Lines
crates/wdk-build/src/cargo_make.rs	93.21%	19 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #629      +/-   ##
==========================================
+ Coverage   77.42%   78.63%   +1.21%     
==========================================
  Files          24       24              
  Lines        4851     5117     +266     
  Branches     4851     5117     +266     
==========================================
+ Hits         3756     4024     +268     
+ Misses        978      976       -2     
  Partials      117      117

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

…t builds Unset RUSTFLAGS for cargo-make and cargo-wdk steps in build.yaml to prevent -D warnings from leaking into rust-script compilation of wdk-build. This is a temporary workaround. The proper fix is in microsoft/windows-drivers-rs#629 which rewrites path deps to version-only deps in load_wdk_build_makefile() for registry consumers. Root cause: wdk-build = { path = '.' } in rust-driver-makefile.toml makes wdk-build a path dependency even when loaded from the registry cache via symlink. Cargo does not apply --cap-lints allow to path deps, so RUSTFLAGS leaks through. Tracked in ADO Bug #5396440.

Copilot

Pull request overview

Copilot reviewed 3 out of 3 changed files in this pull request and generated 2 comments.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-03-08T01:00:26Z

crates/wdk-build/src/cargo_make.rs

+        let version = &wdk_build_package.version;
+        let patched_content = makefile_content.replace(
+            r#"wdk-build = { path = "." }"#,
+            &format!("wdk-build = \"{version}\""),


The rewritten dependency uses wdk-build = "{version}", which is a caret requirement and may allow rust-script to resolve a newer compatible 0.5.x than the wdk-build version that provided this makefile. To keep the embedded scripts and the loaded makefile in lockstep (and make downstream behavior deterministic/offline-friendly), consider rewriting to an exact requirement like wdk-build = "={version}" instead.

Suggested change

&format!("wdk-build = \"{version}\""),

&format!("wdk-build = \"={version}\""),

Copilot · 2026-03-08T01:00:26Z

crates/wdk-build/src/cargo_make.rs

+    let is_registry_source = wdk_build_package
+        .source
+        .as_ref()
+        .is_some_and(|s| s.repr.starts_with("registry+") || s.repr.starts_with("sparse+"));
+
+    if is_registry_source {
+        // wdk-build is a registry dependency. Rewrite path dependencies in the
+        // makefile's rust-script embedded manifests to version-only dependencies
+        // so that cargo treats wdk-build as a registry dep and applies
+        // --cap-lints. This prevents the caller's RUSTFLAGS (e.g. -D warnings)
+        // from leaking into the published wdk-build crate's compilation.
+        let makefile_content = std::fs::read_to_string(&rust_driver_makefile_toml_path)
+            .map_err(|source| IoError::with_path(&rust_driver_makefile_toml_path, source))?;
+
+        let version = &wdk_build_package.version;
+        let patched_content = makefile_content.replace(
+            r#"wdk-build = { path = "." }"#,
+            &format!("wdk-build = \"{version}\""),
+        );
+
+        if patched_content == makefile_content {
+            warn!(
+                "No wdk-build path dependency found to rewrite in {makefile_name:?}. The makefile \
+                 format may have changed."
+            );
+        }
+
+        // Only write if content changed or file doesn't exist, to avoid
+        // unnecessary rebuilds from rust-script cache invalidation
+        let should_write = if destination_path.exists() {
+            let existing_content = std::fs::read_to_string(&destination_path)
+                .map_err(|source| IoError::with_path(&destination_path, source))?;
+            existing_content != patched_content
+        } else {
+            true
+        };
+
+        if should_write {
+            if destination_path.exists() {
+                std::fs::remove_file(&destination_path)
+                    .map_err(|source| IoError::with_path(&destination_path, source))?;
+            }
+            std::fs::write(&destination_path, patched_content)
+                .map_err(|source| IoError::with_path(&destination_path, source))?;
+        }
+    } else {


This new registry-vs-non-registry behavior (copy+rewrite vs symlink) is not covered by tests, but this module already has unit tests for other helpers. Consider extracting the rewrite decision/content patching into a small pure helper and adding tests that (a) registry sources produce rewritten content and (b) non-registry sources preserve the symlink/update behavior.

…t builds Unset RUSTFLAGS for cargo-make and cargo-wdk steps in build.yaml to prevent -D warnings from leaking into rust-script compilation of wdk-build. Temporary workaround until microsoft/windows-drivers-rs#629 is merged and a new wdk-build is published. Tracked in ADO Bug #5396440.

Unset RUSTFLAGS for the cargo-make step in build.yaml to prevent -D warnings from leaking into rust-script compilation of wdk-build. Only cargo-make is affected since it uses rust-script to compile inline build scripts where wdk-build is a path dependency (via symlink), bypassing cargo's --cap-lints. cargo-wdk and cargo install are normal cargo operations with proper --cap-lints handling and do not need this workaround. Temporary workaround until microsoft/windows-drivers-rs#629 is merged and a new wdk-build is published. Tracked in ADO Bug #5396440.

## Changes ### 1. Fix nightly clippy + rustfmt failures Edits to pass new nightly `clippy::unnecessary_trailing_comma` [lint](https://rust-lang.github.io/rust-clippy/master/index.html?search=clippy%3A%3Aunnecessary_trailing_comma#unnecessary_trailing_comma) and nightly rustfmt import reordering. - Remove 4 unnecessary trailing commas in `format!`/`println!` macros (`general/echo/kmdf/exe/src/main.rs`) - Reorder underscore-prefixed imports to match nightly rustfmt sorting (`device.rs`, `queue.rs`, `driver.rs`) **Failed CI run:** https://github.com/microsoft/Windows-rust-driver-samples/actions/runs/22519532366 ### 2. Workaround RUSTFLAGS leaking into cargo-make/cargo-wdk builds Unset `RUSTFLAGS` for `cargo make`, `cargo wdk install`, and `cargo wdk build` steps in `build.yaml`. This prevents `-D warnings` from leaking into `rust-script` compilation of `wdk-build` (which is treated as a path dependency via symlink, bypassing cargo's `--cap-lints allow`). Temporary workaround until microsoft/windows-drivers-rs#629 is merged and a new `wdk-build` is published.

…cript builds When downstream consumers (e.g. Windows-rust-driver-samples) run cargo make, RUSTFLAGS=-D warnings leaks into rust-script compilation of wdk-build because the makefile dep specs use path = '.' which makes wdk-build a path dependency. Cargo does not apply --cap-lints to path dependencies, so -D warnings applies directly to the published crate, causing upstream lint issues to break downstream builds. Fix: load_wdk_build_makefile() now detects whether wdk-build is a registry dependency (via package.source from cargo_metadata). When it is, the makefile is copied with path deps rewritten to version-only deps, ensuring cargo applies --cap-lints allow. When wdk-build is a local path dep, the existing symlink behavior is preserved. Also removes the unused version field from all 14 rust-script dep specs in the makefiles, since path always takes precedence when both are specified.

… makefiles

Copilot

Pull request overview

Copilot reviewed 3 out of 3 changed files in this pull request and generated 2 comments.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-03-27T23:53:32Z

crates/wdk-build/src/cargo_make.rs

+/// The version of `wdk-build` from which the file comes is determined by the
+/// working directory of the process that invokes this function. For example,
+/// if this function is ultimately executing in a `cargo_make` `load_script`,
+/// the files will come from the `wdk-build` version that is in the
+/// `Cargo.lock` file, and not the `wdk-build` version specified in the
+/// `load_script`.


The updated doc comment explains the new copy+rewrite behavior, but the # Errors list for this function (below) still describes only symlink-related IO errors and doesn’t mention that NoWdkBuildCrateDetected can be returned. Please align the error docs with the current behavior (read/write/symlink + missing wdk-build).

Copilot · 2026-03-27T23:53:33Z

crates/wdk-build/src/cargo_make.rs

+    let wdk_build_package = &wdk_build_package_matches[0];
+
+    let rust_driver_makefile_toml_path = wdk_build_package
        .manifest_path
        .parent()


rust_driver_makefile_toml_path is used for both rust-driver and rust-driver-sample makefiles (it’s derived from the generic makefile_name). Renaming it to something neutral like makefile_path/source_makefile_path would avoid confusion as this function grows.

Copilot AI review requested due to automatic review settings March 7, 2026 22:53

Copilot started reviewing on behalf of wmmc88 March 7, 2026 22:54 View session

Copilot AI reviewed Mar 7, 2026

View reviewed changes

wmmc88 marked this pull request as draft March 8, 2026 00:26

wmmc88 force-pushed the fix/cargo-make-rustflags-leak branch from 4df0dc5 to 9a89265 Compare March 8, 2026 00:33

wmmc88 mentioned this pull request Mar 8, 2026

ci: workaround RUSTFLAGS leaking into cargo-make/cargo-wdk builds microsoft/Windows-rust-driver-samples#51

Closed

wmmc88 force-pushed the fix/cargo-make-rustflags-leak branch from 9a89265 to 3522897 Compare March 8, 2026 00:56

Copilot AI review requested due to automatic review settings March 8, 2026 00:56

Copilot started reviewing on behalf of wmmc88 March 8, 2026 00:56 View session

Copilot AI reviewed Mar 8, 2026

View reviewed changes

wmmc88 force-pushed the fix/cargo-make-rustflags-leak branch from 3522897 to a28a18b Compare March 8, 2026 01:18

wmmc88 mentioned this pull request Mar 8, 2026

fix: nightly CI failures + workaround RUSTFLAGS cargo-make leak microsoft/Windows-rust-driver-samples#49

Merged

wmmc88 added 2 commits March 27, 2026 16:45

fix: update documentation to clarify wdk-build dependency behavior in…

bcdf058

… makefiles

Copilot AI review requested due to automatic review settings March 27, 2026 23:45

wmmc88 force-pushed the fix/cargo-make-rustflags-leak branch from a28a18b to bcdf058 Compare March 27, 2026 23:45

Copilot started reviewing on behalf of wmmc88 March 27, 2026 23:47 View session

Copilot AI reviewed Mar 27, 2026

View reviewed changes

Copilot AI mentioned this pull request Mar 28, 2026

Revert RUSTFLAGS env var leak workaround in build.yaml microsoft/Windows-rust-driver-samples#60

Draft

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

fix: prevent RUSTFLAGS from leaking into downstream cargo-make rust-script builds#629

fix: prevent RUSTFLAGS from leaking into downstream cargo-make rust-script builds#629
wmmc88 wants to merge 2 commits intomicrosoft:mainfrom
wmmc88:fix/cargo-make-rustflags-leak

wmmc88 commented Mar 7, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Mar 7, 2026

Uh oh!

Copilot AI Mar 7, 2026

Uh oh!

Copilot AI Mar 7, 2026

Uh oh!

codecov-commenter commented Mar 7, 2026 •

edited

Loading

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Mar 8, 2026

Uh oh!

Copilot AI Mar 8, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Mar 27, 2026

Uh oh!

Copilot AI Mar 27, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

-//! # `load_rust_driver_makefile()`, this path dependency is automatically
-//! # rewritten to a version-only registry dependency. This ensures cargo
-//! # applies `--cap-lints allow` to wdk-build, preventing the caller's
-//! # RUSTFLAGS from leaking into the published crate's compilation.
+//! # `load_rust_driver_makefile()`, and `wdk-build` is resolved from a
+//! # non-path source (e.g., registry or git), this path dependency is
+//! # automatically rewritten to a version-only registry dependency. This
+//! # ensures cargo applies `--cap-lints allow` to wdk-build, preventing the
+//! # caller's RUSTFLAGS from leaking into the published crate's compilation.

	&format!("wdk-build = \"{version}\""),
	&format!("wdk-build = \"={version}\""),

Conversation

wmmc88 commented Mar 7, 2026

Problem

Root Cause

Fix

Scenarios

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI Mar 7, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 7, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 7, 2026

Choose a reason for hiding this comment

Uh oh!

codecov-commenter commented Mar 7, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Copilot AI Mar 8, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 8, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Copilot AI Mar 27, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 27, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

codecov-commenter commented Mar 7, 2026 •

edited

Loading