feat: implement RFC 3553 to add SBOM support #13709

justahero · 2024-04-05T11:23:51Z

What does this PR try to resolve?

This PR is an implementation of RFC 3553 to add support to generate pre-cursor SBOM files for compiled artifacts in Cargo.

How should we test and review this PR?

The RFC 3553 adds a new option to Cargo to emit SBOM pre-cursor files. A project can be configured either by the new Cargo config field sbom.

# .cargo/config.toml
[build]
sbom = true

or using the environment variable CARGO_BUILD_SBOM=true. The sbom option is an unstable feature and requires the -Zsbom flag to enable it.

Check out this branch & compile Cargo. Pick a Cargo project to test it on, then run:

CARGO_BUILD_SBOM=true <path/to/compiled/cargo>/target/debug/cargo build -Zsbom

All generated *.cargo-sbom.json files are located in the target folder alongside their artifacts. To list all generated files use:

find ./target -name "*.cargo-sbom.json"

then check their content. To see the current output format, see these examples.

What does the PR not solve?

The PR leaves a task(s) open that are either out of scope or should be done in a follow-up PRs.

does not address Duplicate artifact tracking issue. #6313 (see comment)

Additional information

There are a few things that I would like to get feedback on, in particular the generated JSON format is not final. Currently it holds the information listed in the RFC 3553, but it could be further enriched with information only available during builds.

During the implementation a number of questions arose.

Is using the UnitGraph the right structure to determine all dependencies?
Is the location in the compile method to generate the SBOM files appropriate?
Is there important information missing from the generated output, some fields that should be omitted?
How best to check JSON output in the related tests in testsuite, are useful tests missing?
Are custom build script dependencies correctly resolved?

Thanks @arlosi, @RobJellinghaus and @lfrancke for initial guidance & feedback.

RFC: #RFC: cargo-sbom rfcs#3553
SBOM output examples (serde_json & axum-core crates)

rustbot · 2024-04-05T11:23:56Z

Thanks for the pull request, and welcome! The Rust team is excited to review your changes, and you should hear from @ehuss (or someone else) some time within the next two weeks.

Please see the contribution instructions for more information. Namely, in order to ensure the minimum review times lag, PR authors and assigned reviewers should ensure that the review label (S-waiting-on-review and S-waiting-on-author) stays updated, invoking these commands when appropriate:

@rustbot author: the review is finished, PR author should check the comments and take action accordingly
@rustbot review: the author is ready for a review, this PR will be queued again in the reviewer's queue

src/cargo/core/compiler/build_runner/mod.rs

heisen-li · 2024-04-07T12:12:11Z

Much respect for your contribution.

From my kind reminders, it seems appropriate to modify the documentation of the corresponding sections, e.g. Configuration, Environment Variables.

weihanglo · 2024-04-07T14:35:13Z

Thanks for the reminder, @heisen-li. Would love to see a doc update, though we should probably focus on the design discussion first, as the location of the configuration is not yet decided. (See rust-lang/rfcs#3553 (comment)).

epage · 2024-04-09T16:02:21Z

One approach for the docs (if this is looking to be merged) is to put the env and config documentation fragments in the Unstable docs.

src/cargo/core/compiler/mod.rs

src/cargo/core/compiler/build_runner/mod.rs

src/cargo/core/compiler/output_sbom.rs

weihanglo · 2024-05-02T20:35:25Z

tests/testsuite/sbom.rs

+    assert!(p.bin("foo").with_extension("cargo-sbom.json").is_file());
+    assert_eq!(
+        1,
+        p.glob(p.target_debug_dir().join("libfoo.cargo-sbom.json"))


I guess we might need to deal with different naming convention on different platform. (Windows specifically?)

Maybe the glob call can be simplified in a way that it works for all platforms.

tests/testsuite/sbom.rs

weihanglo

Just note that I reviewed this as-is, didn't really think too much for the design itself. Thank you for working on this!

src/cargo/core/compiler/build_config.rs

src/cargo/core/compiler/output_sbom.rs

bors · 2024-05-03T16:12:02Z

☔ The latest upstream changes (presumably #13571) made this pull request unmergeable. Please resolve the merge conflicts.

weihanglo

Now I like the idea of having this PR to explore SBOM format. I'll post back issues we've found so far to the RFC. Thank you :)

src/cargo/core/compiler/build_runner/mod.rs

src/doc/src/reference/unstable.md

tests/testsuite/sbom.rs

src/cargo/core/compiler/output_sbom.rs

weihanglo · 2024-05-08T01:22:32Z

src/cargo/core/compiler/output_sbom.rs

+}
+
+#[derive(Serialize)]
+struct Sbom {


Do we need a dependencies field for this top-level Sbom?

(Just a question. I don't really know if other SBOM formats need it to recover the dependency graph)

Ideally, yes. Copying my comment from the RFC:

Note that there are two ways of looking at dependencies: what each package needs, and the final resolved graph.

For example, if one package depends on rand with features = ["std", "getrandom"], and another with features = ["std", "simd_support"], the final resolved features will be ["std", "getrandom", "simd_support"]. Depending on the use case you may need either or both representations (direct package dependencies and the resolved graph).

cargo metadata exposes both (under "packages" and "resolve" fields), but inaccurately:

cargo-metadata always resolves features at the workspace level #7754

Under resolver v2, it conflates the normal and build-dependency trees

I think it would be best for the SBOM to also expose both, accurately this time.

So what I would like to see is two resolved dependency trees: one for normal dependencies and one for build dependencies, matching the way feature resolver v2 works.

src/cargo/core/compiler/output_sbom.rs

tests/testsuite/sbom.rs

src/cargo/core/compiler/build_runner/mod.rs

src/cargo/core/compiler/build_config.rs

src/cargo/core/compiler/output_sbom.rs

bors · 2024-10-04T19:05:52Z

☔ The latest upstream changes (presumably #14576) made this pull request unmergeable. Please resolve the merge conflicts.

Similar to the generation of `depinfo` files, a function is called to generated SBOM precursor file named `output_sbom`. It takes the `BuildRunner` & the current `Unit`. The `sbom` flag can be specified as a cargo build option, but it's currently not configured fully. To test the generation the flag is set to `true`. * use SBOM types to serialize data Output source, profile & dependencies Trying to fetch all dependencies This ignores dependencies for custom build scripts. The output should be similar to what `cargo tree` reports. Output package dependencies This is similar to what the `cargo metadata` command outputs. Extract logic to fetch sbom output files This extracts the logic to get the list of SBOM output file paths into its own function in `BuildRunner` for a given Unit. Add test file to check sbom output * add test to check project with bin & lib * extract sbom config into helper function Add build type to dependency Add test to read JSON Still needs to check output. Guard sbom logic behind unstable feature Add test with custom build script Integrate review feedback * disable `sbom` config when `-Zsbom` is not passed as unstable option * refactor tests * add test Expand end-to-end tests This expands the tests to reflect end-to-end tests by comparing the generated JSON output files with expected strings. * add test helper to compare actual & expected JSON content * refactor setup of packages in test Add 'sbom' section to unstable features doc Append SBOM file suffix instead of replacing Instead of replacing the file extension, the `.cargo-sbom.json` suffix is appended to the output file. This is to keep existing file extensions in place. * refactor logic to set `sbom` property from build config * expand build script related test to check JSON output Integrate review feedback * use `PackageIdSpec` instead of only `PackageId` in SBOM output * change `version` of a dependency to `Option<Version>` * output `Vec<CrateType>` instead of only the first found crate type * output rustc workspace wrapper * update 'warning' string in test using `[WARNING]` * use `serde_json::to_writer` to serialize SBOM * set sbom suffix in tests explicitely, instead of using `with_extension` Output additional fields to JSON In case a unit's profile differs from the profile information on root level, it's added to the package information to the JSON output. The verbose output for `rustc -vV` is also written to the `rustc` field in the SBOM. * rename `fetch_packages` to `collect_packages` * update JSON in tests to include profile information Add test to check multiple crate types Add test to check artifact name conflict Use SbomProfile to wrap Profile type This adds the `SbomProfile` to convert the existing `Profile` into, to expose relevant fields. For now it removes the `strip` field, while serializing all other fields. It should keep the output consistent, even when fields in the `Profile` change, e.g. new field added. Document package profile * only export `profile` field in case it differs from root profile Add test to check different features The added test uses a crate with multiple features. The main crate uses the dependency in the normal build & the custom build script with different features. Refactor storing of package dependencies All dependencies for a package are indices into the `packages` list now. This sets the correct association between a dependency & its associated package. * remove `SbomDependency` struct Refactor tests to use snapbox

epage · 2024-12-06T21:07:52Z

src/cargo/core/compiler/output_sbom.rs

+#[derive(Serialize, Clone, Debug, Copy)]
+#[serde(rename_all = "kebab-case")]
+enum SbomBuildType {
+    /// A package dependency
+    Normal,
+    /// A build script dependency
+    Build,
+}


Should we be consistent with cargo metadata wrt th schema for this?

epage · 2024-12-06T21:10:03Z

src/cargo/core/compiler/output_sbom.rs

+#[derive(Serialize, Clone, Debug)]
+struct SbomProfile {


No case is put on this. Is snake_case intentional? Looks like thats what we use for cargo metadata

epage · 2024-12-06T21:11:21Z

src/cargo/core/compiler/output_sbom.rs

+impl From<&Profile> for SbomProfile {
+    fn from(profile: &Profile) -> Self {
+        let rustflags = profile
+            .rustflags
+            .iter()
+            .map(|x| x.to_string())
+            .collect_vec();
+
+        Self {
+            name: profile.name.to_string(),
+            opt_level: profile.opt_level.to_string(),
+            lto: profile.lto,
+            codegen_backend: profile.codegen_backend.map(|x| x.to_string()),
+            codegen_units: profile.codegen_units.clone(),
+            debuginfo: profile.debuginfo.clone(),
+            split_debuginfo: profile.split_debuginfo.map(|x| x.to_string()),
+            debug_assertions: profile.debug_assertions,
+            overflow_checks: profile.overflow_checks,
+            rpath: profile.rpath,
+            incremental: profile.incremental,
+            panic: profile.panic.to_string(),
+            rustflags,
+        }
+    }
+}


A part of me wants to have us profile.clone() and destructure Profile so we make sure we update this whenever a new Profile field is added (and to make it easier to review to make sure all fields are present)

epage · 2024-12-06T21:12:05Z

src/cargo/core/compiler/output_sbom.rs

+    version: Option<Version>,
+    features: Vec<String>,
+    build_type: SbomBuildType,
+    extern_crate_name: String,


Why do we need this?

epage · 2024-12-06T21:12:49Z

src/cargo/core/compiler/output_sbom.rs

+    package_id: PackageIdSpec,
+    package: String,


whats package?

I guess its the name, but why include it (and version) when its in package_id?

epage · 2024-12-06T21:14:46Z

src/cargo/core/compiler/output_sbom.rs

+        let build_type = if unit_dep.unit.mode.is_run_custom_build() {
+            SbomBuildType::Build
+        } else {
+            SbomBuildType::Normal
+        };


The documentation for is_run_custom_build says

/// Returns `true` if this is the *execution* of a `build.rs` script.

So this isn't saying whether this is a normal or build instance but whether this will be run as a build script.

as in build scripts show up twice, once for the Build mode and once for the Run mode

epage · 2024-12-06T21:15:14Z

src/cargo/core/compiler/output_sbom.rs

+
+/// Describes a package dependency
+#[derive(Serialize, Clone, Debug)]
+struct SbomPackage {


Did we decide whether the sbom will track packages instead of crates?

epage · 2024-12-06T21:17:51Z

src/cargo/core/compiler/output_sbom.rs

+
+#[derive(Serialize)]
+struct Sbom {
+    format_version: SbomFormatVersion<1>,


Not-blocking: eventually we should move this out into cargo-util-schemas and provide Deserialize impls so people can use that to read this file. SbomFormatVersion isn't compatible with that.

epage · 2024-12-06T21:19:35Z

src/cargo/core/compiler/output_sbom.rs

+    package_id: PackageIdSpec,
+    name: String,
+    version: String,
+    source: String,
+    target: SbomTarget,
+    profile: SbomProfile,
+    packages: Vec<SbomPackage>,
+    features: Vec<String>,


Isn't most of this redundant with packages, is there a reason we don't just have a root_package field?

epage · 2024-12-06T21:20:30Z

src/cargo/core/compiler/output_sbom.rs

+    build_type: SbomBuildType,
+    extern_crate_name: String,
+    /// Indices into the `packages` array
+    dependencies: Vec<usize>,


Should we newtype the usize? On one hand, it will make it easy to find all all usizes that fill those role. On the other hand, it will make indexing more annoying.

epage · 2024-12-06T21:24:27Z

src/cargo/core/compiler/output_sbom.rs

+    for (index, package) in visited_dependencies.iter().enumerate() {
+        let dependencies: BTreeSet<&UnitDep> = unit_graph[&package.unit].iter().collect();
+
+        let mut indices = dependencies
+            .iter()
+            .filter_map(|dep| {
+                visited_dependencies
+                    .iter()
+                    .position(|unit_dep| dep == unit_dep)
+            })
+            .collect::<Vec<_>>();
+
+        if let Some(package) = packages.get_mut(index) {
+            package.dependencies.append(&mut indices);


This assumes that the order we iterate through visitied_dependencies (BTreeSet) is the same order things appear in packages (Vec`).

Should visited_dependencies be an IndexSet?

epage · 2024-12-06T21:25:54Z

src/cargo/core/compiler/output_sbom.rs

+    for (index, package) in visited_dependencies.iter().enumerate() {
+        let dependencies: BTreeSet<&UnitDep> = unit_graph[&package.unit].iter().collect();
+
+        let mut indices = dependencies
+            .iter()
+            .filter_map(|dep| {
+                visited_dependencies
+                    .iter()
+                    .position(|unit_dep| dep == unit_dep)


Is this O(n^3)? Is the equality check doing a straight pointer comparison or a deep check?

epage · 2024-12-06T21:26:41Z

src/doc/src/reference/unstable.md

@@ -396,6 +397,32 @@ It's possible to update `my-dependency` to a pre-release with `update -Zunstable
 This is because `0.1.2-pre.0` is considered compatible with `0.1.1`.
 It would not be possible to upgrade to `0.2.0-pre.0` from `0.1.1` in the same way.

+## sbom


Should we document the schema in here? yes, the RFC will have it but that will then be a snapshot in time and eventually we'll want to have it documented.

epage · 2024-12-06T21:28:14Z

tests/testsuite/sbom/mod.rs

+
+    let file = with_sbom_suffix(&p.bin("foo"));
+    let output = std::fs::read_to_string(file).unwrap();
+    assert_e2e().eq(output, snapbox::file!["build_sbom_using_cargo_config.json"]);


Hmm, having external snapshots will make it a lot harder to review these tests because I have to find the right test and right file and jump back and forth between them

epage · 2024-12-06T21:29:45Z

tests/testsuite/sbom/mod.rs

+    assert_eq!(
+        2,
+        p.glob(p.target_debug_dir().join("*.cargo-sbom.json"))
+            .count()
+    );


Under what conditions is an rlib sbom created?

epage · 2024-12-06T21:30:40Z

tests/testsuite/sbom/mod.rs

+fn configured_project() -> ProjectBuilder {
+    project().file(
+        ".cargo/config.toml",
+        r#"
+            [build]
+            sbom = true
+        "#,
+    )
+}


Personally, I find helpers like this not too helpful

They obfuscate what the test is doing (it should at least say what the config is)

They don't scale if you then want to do something else with the config

epage · 2024-12-06T21:31:20Z

tests/testsuite/sbom/mod.rs

+#[cargo_test]
+fn build_sbom_with_artifact_name_conflict() {
+    Package::new("deps", "0.1.0")
+        .file(
+            "Cargo.toml",
+            r#"
+                [package]
+                name = "deps" # name conflict
+                version = "0.1.0"


What is this test showing?

epage · 2024-12-06T21:35:29Z

src/cargo/core/compiler/output_sbom.rs

+        let profile = if &unit_dep.unit.profile != root_profile {
+            Some((&unit_dep.unit.profile).into())
+        } else {
+            None
+        };


Not seeing a test that demonstrates this conditional profile

rustbot assigned ehuss Apr 5, 2024

rustbot added A-build-execution Area: anything dealing with executing the compiler A-configuration Area: cargo config files and env vars A-unstable Area: nightly unstable support S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Apr 5, 2024

justahero commented Apr 5, 2024

View reviewed changes

src/cargo/core/compiler/build_runner/mod.rs Show resolved Hide resolved

justahero force-pushed the rfc3553/cargo-sbom-support branch from 74dafa0 to 190682e Compare April 6, 2024 14:31

justahero force-pushed the rfc3553/cargo-sbom-support branch from 190682e to ae0881c Compare May 2, 2024 19:54

weihanglo reviewed May 2, 2024

View reviewed changes

tests/testsuite/sbom.rs Outdated Show resolved Hide resolved

weihanglo reviewed May 2, 2024

View reviewed changes

src/cargo/core/compiler/build_config.rs Outdated Show resolved Hide resolved

src/cargo/core/compiler/build_config.rs Outdated Show resolved Hide resolved

src/cargo/core/compiler/output_sbom.rs Outdated Show resolved Hide resolved

justahero force-pushed the rfc3553/cargo-sbom-support branch 4 times, most recently from 1cfd71a to 376fe1e Compare May 6, 2024 13:42

rustbot added the A-documenting-cargo-itself Area: Cargo's documentation label May 6, 2024

justahero force-pushed the rfc3553/cargo-sbom-support branch 4 times, most recently from 67332d6 to 0aa10e9 Compare May 7, 2024 11:13

justahero marked this pull request as ready for review May 7, 2024 11:53

weihanglo reviewed May 8, 2024

View reviewed changes

weihanglo mentioned this pull request May 8, 2024

RFC: cargo-sbom rust-lang/rfcs#3553

Open

arlosi reviewed May 8, 2024

View reviewed changes

justahero force-pushed the rfc3553/cargo-sbom-support branch from c8e1bc8 to 8d5fa4d Compare May 13, 2024 12:33

justahero force-pushed the rfc3553/cargo-sbom-support branch 6 times, most recently from adf4f19 to 0c7f60f Compare August 5, 2024 08:17

justahero requested a review from arlosi August 5, 2024 13:56

justahero force-pushed the rfc3553/cargo-sbom-support branch from 496d62e to bc95299 Compare August 6, 2024 12:45

Shnatsel mentioned this pull request Nov 8, 2024

cargo metadata: provide an option not to collect data about dev-dependencies #14794

Open

arlosi force-pushed the rfc3553/cargo-sbom-support branch from bc95299 to 848cc32 Compare November 26, 2024 15:56

epage reviewed Dec 6, 2024

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat: implement RFC 3553 to add SBOM support #13709

feat: implement RFC 3553 to add SBOM support #13709

justahero commented Apr 5, 2024 •

edited

Loading

rustbot commented Apr 5, 2024

heisen-li commented Apr 7, 2024

weihanglo commented Apr 7, 2024

epage commented Apr 9, 2024

weihanglo May 2, 2024

justahero May 6, 2024

weihanglo left a comment

bors commented May 3, 2024

weihanglo left a comment

weihanglo May 8, 2024

Shnatsel May 8, 2024

bors commented Oct 4, 2024

epage Dec 6, 2024

epage Dec 6, 2024

epage Dec 6, 2024

epage Dec 6, 2024

epage Dec 6, 2024

epage Dec 6, 2024

epage Dec 6, 2024

epage Dec 6, 2024

epage Dec 6, 2024

epage Dec 6, 2024

epage Dec 6, 2024

epage Dec 6, 2024

epage Dec 6, 2024

epage Dec 6, 2024

epage Dec 6, 2024

epage Dec 6, 2024

epage Dec 6, 2024

epage Dec 6, 2024

epage Dec 6, 2024

epage Dec 6, 2024

feat: implement RFC 3553 to add SBOM support #13709

Are you sure you want to change the base?

feat: implement RFC 3553 to add SBOM support #13709

Conversation

justahero commented Apr 5, 2024 • edited Loading

What does this PR try to resolve?

How should we test and review this PR?

What does the PR not solve?

Additional information

rustbot commented Apr 5, 2024

heisen-li commented Apr 7, 2024

weihanglo commented Apr 7, 2024

epage commented Apr 9, 2024

Choose a reason for hiding this comment

Choose a reason for hiding this comment

weihanglo left a comment

Choose a reason for hiding this comment

bors commented May 3, 2024

weihanglo left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

bors commented Oct 4, 2024

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

justahero commented Apr 5, 2024 •

edited

Loading