feat: thin lto for non consuming artifacts

[Aaalibaba42]

Fat LTO + codegen-units=1 makes the Generate profiling FFI step slow for no reason, this doesn't ship an artifact anyone consumes. Override to thin LTO / codegen-units=16 via CARGO_PROFILE_RELEASE_* env vars for this build only

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: ac022a4654

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

[chatgpt-codex-connector]

Include the profile overrides in the Rust cache key

When the ffi job restores an existing main-branch cache, these variables are visible only to the later Generate step, so the preceding Swatinem/rust-cache action computes the same key as for the fat-LTO configuration. Cargo then invalidates and rebuilds the restored release artifacts, while the immutable exact-key cache prevents the newly built thin-LTO artifacts from replacing them; consequently subsequent runs can repeatedly rebuild the dependency graph and lose the intended speedup. Put these overrides at job scope or otherwise incorporate them into the cache key.

Useful? React with 👍 / 👎.

[datadog-datadog-prod-us1-2]

🎉 All green!

🧪 All tests passed
❄️ No new flaky tests detected

🎯 Code Coverage (details)
• Patch Coverage: 100.00%
• Overall Coverage: 76.63% (-0.02%)

_{This comment will be updated automatically if new data arrives.
🔗 Commit SHA: ac022a4 | Docs | View more details | Give us feedback!}

[pr-commenter]

Benchmarks

Comparison

Benchmark execution time: 2026-08-24 13:21:12

Comparing candidate commit ac022a4 in PR branch jwiriath/APMSP-3950-thin-lto with baseline commit 9d6b2ff in branch main.

Found 1 performance improvements and 5 performance regressions! Performance is the same for 146 metrics, 0 unstable metrics.

Explanation

This is an A/B test comparing a candidate commit's performance against that of a baseline commit. Performance changes are noted in the tables below as:

• 🟩 = significantly better candidate vs. baseline
• 🟥 = significantly worse candidate vs. baseline

We compute a confidence interval (CI) over the relative difference of means between metrics from the candidate and baseline commits, considering the baseline as the reference.

If the CI is entirely outside the configured SIGNIFICANT_IMPACT_THRESHOLD (or the deprecated UNCONFIDENCE_THRESHOLD), the change is considered significant.

Feel free to reach out to #apm-benchmarking-platform on Slack if you have any questions.

More details about the CI and significant changes

You can imagine this CI as a range of values that is likely to contain the true difference of means between the candidate and baseline commits.

CIs of the difference of means are often centered around 0%, because often changes are not that big:

---------------------------------(------|---^--------)-------------------------------->
                              -0.6%    0%  0.3%     +1.2%
                                 |          |        |
         lower bound of the CI --'          |        |
sample mean (center of the CI) -------------'        |
         upper bound of the CI ----------------------'

As described above, a change is considered significant if the CI is entirely outside the configured SIGNIFICANT_IMPACT_THRESHOLD (or the deprecated UNCONFIDENCE_THRESHOLD).

For instance, for an execution time metric, this confidence interval indicates a significantly worse performance:

----------------------------------------|---------|---(---------^---------)---------->
                                       0%        1%  1.3%      2.2%      3.1%
                                                  |   |         |         |
       significant impact threshold --------------'   |         |         |
                      lower bound of CI --------------'         |         |
       sample mean (center of the CI) --------------------------'         |
                      upper bound of CI ----------------------------------'

scenario:flagevaluation_evp/payloads/stress/10flags_1000users_250fields

• 🟥 execution_time [+4.363ms; +4.599ms] or [+5.144%; +5.423%]
• 🟥 throughput [-608.276op/s; -576.570op/s] or [-5.158%; -4.890%]

scenario:msgpack_decoder::v05/low_sharing/10000

• 🟥 execution_time [+1.095ms; +1.102ms] or [+6.301%; +6.338%]
• 🟥 throughput [-34295.502op/s; -34089.816op/s] or [-5.962%; -5.926%]

scenario:normalization/normalize_trace/test_trace

• 🟩 execution_time [-56.562ns; -51.553ns] or [-18.583%; -16.937%]

scenario:vec_map/as_deduped_map/needs_dedup_1_in_4/8

• 🟥 execution_time [+19.825ns; +20.022ns] or [+5.050%; +5.100%]

Benchmark execution time: 2026-08-24 13:28:06

Comparing candidate commit ac022a4 in PR branch jwiriath/APMSP-3950-thin-lto with baseline commit 9d6b2ff in branch main.

Found 2 performance improvements and 2 performance regressions! Performance is the same for 164 metrics, 10 unstable metrics.

Explanation

This is an A/B test comparing a candidate commit's performance against that of a baseline commit. Performance changes are noted in the tables below as:

• 🟩 = significantly better candidate vs. baseline
• 🟥 = significantly worse candidate vs. baseline

We compute a confidence interval (CI) over the relative difference of means between metrics from the candidate and baseline commits, considering the baseline as the reference.

If the CI is entirely outside the configured SIGNIFICANT_IMPACT_THRESHOLD (or the deprecated UNCONFIDENCE_THRESHOLD), the change is considered significant.

Feel free to reach out to #apm-benchmarking-platform on Slack if you have any questions.

More details about the CI and significant changes

You can imagine this CI as a range of values that is likely to contain the true difference of means between the candidate and baseline commits.

CIs of the difference of means are often centered around 0%, because often changes are not that big:

---------------------------------(------|---^--------)-------------------------------->
                              -0.6%    0%  0.3%     +1.2%
                                 |          |        |
         lower bound of the CI --'          |        |
sample mean (center of the CI) -------------'        |
         upper bound of the CI ----------------------'

As described above, a change is considered significant if the CI is entirely outside the configured SIGNIFICANT_IMPACT_THRESHOLD (or the deprecated UNCONFIDENCE_THRESHOLD).

For instance, for an execution time metric, this confidence interval indicates a significantly worse performance:

----------------------------------------|---------|---(---------^---------)---------->
                                       0%        1%  1.3%      2.2%      3.1%
                                                  |   |         |         |
       significant impact threshold --------------'   |         |         |
                      lower bound of CI --------------'         |         |
       sample mean (center of the CI) --------------------------'         |
                      upper bound of CI ----------------------------------'

scenario:datadog_sample_span/name_pattern_rule_not_matching/wall_time

• 🟥 execution_time [+23.945ns; +24.028ns] or [+14.897%; +14.949%]

scenario:no_profiler/short_circuit/4096

• 🟩 execution_time [-13.891ns; -13.761ns] or [-13.401%; -13.276%]

scenario:profiler_attached/fast_path_system/4096

• 🟥 execution_time [+6.147ns; +6.310ns] or [+6.229%; +6.395%]

scenario:profiler_attached/slow_path_system/4096

• 🟩 execution_time [-25.442ns; -25.350ns] or [-16.555%; -16.495%]

Candidate

Omitted due to size.

Baseline

Omitted due to size.

[dd-octo-sts]

Artifact Size Benchmark Report

aarch64-alpine-linux-musl

Artifact	Baseline	Commit	Change
/aarch64-alpine-linux-musl/lib/libdatadog_profiling.a	89.52 MB	89.52 MB	0% (0 B) 👌
/aarch64-alpine-linux-musl/lib/libdatadog_profiling.so	8.26 MB	8.26 MB	0% (0 B) 👌

aarch64-unknown-linux-gnu

Artifact	Baseline	Commit	Change
/aarch64-unknown-linux-gnu/lib/libdatadog_profiling.a	100.70 MB	100.70 MB	0% (0 B) 👌
/aarch64-unknown-linux-gnu/lib/libdatadog_profiling.so	11.06 MB	11.06 MB	0% (0 B) 👌

libdatadog-x64-windows

Artifact	Baseline	Commit	Change
/libdatadog-x64-windows/debug/dynamic/datadog_profiling_ffi.dll	26.64 MB	26.64 MB	0% (0 B) 👌
/libdatadog-x64-windows/debug/dynamic/datadog_profiling_ffi.lib	94.96 KB	94.96 KB	0% (0 B) 👌
/libdatadog-x64-windows/debug/dynamic/datadog_profiling_ffi.pdb	180.73 MB	180.73 MB	+0% (+8.00 KB) 👌
/libdatadog-x64-windows/debug/static/datadog_profiling_ffi.lib	772.85 MB	772.85 MB	0% (0 B) 👌
/libdatadog-x64-windows/release/dynamic/datadog_profiling_ffi.dll	8.73 MB	8.73 MB	0% (0 B) 👌
/libdatadog-x64-windows/release/dynamic/datadog_profiling_ffi.lib	94.96 KB	94.96 KB	0% (0 B) 👌
/libdatadog-x64-windows/release/dynamic/datadog_profiling_ffi.pdb	25.65 MB	25.65 MB	0% (0 B) 👌
/libdatadog-x64-windows/release/static/datadog_profiling_ffi.lib	51.09 MB	51.09 MB	0% (0 B) 👌

libdatadog-x86-windows

Artifact	Baseline	Commit	Change
/libdatadog-x86-windows/debug/dynamic/datadog_profiling_ffi.dll	23.22 MB	23.22 MB	0% (0 B) 👌
/libdatadog-x86-windows/debug/dynamic/datadog_profiling_ffi.lib	96.45 KB	96.45 KB	0% (0 B) 👌
/libdatadog-x86-windows/debug/dynamic/datadog_profiling_ffi.pdb	185.66 MB	185.66 MB	0% (0 B) 👌
/libdatadog-x86-windows/debug/static/datadog_profiling_ffi.lib	759.09 MB	759.09 MB	0% (0 B) 👌
/libdatadog-x86-windows/release/dynamic/datadog_profiling_ffi.dll	6.74 MB	6.74 MB	0% (0 B) 👌
/libdatadog-x86-windows/release/dynamic/datadog_profiling_ffi.lib	96.45 KB	96.45 KB	0% (0 B) 👌
/libdatadog-x86-windows/release/dynamic/datadog_profiling_ffi.pdb	27.59 MB	27.59 MB	0% (0 B) 👌
/libdatadog-x86-windows/release/static/datadog_profiling_ffi.lib	48.62 MB	48.62 MB	0% (0 B) 👌

x86_64-alpine-linux-musl

Artifact	Baseline	Commit	Change
/x86_64-alpine-linux-musl/lib/libdatadog_profiling.a	79.78 MB	79.78 MB	0% (0 B) 👌
/x86_64-alpine-linux-musl/lib/libdatadog_profiling.so	9.16 MB	9.16 MB	0% (0 B) 👌

x86_64-unknown-linux-gnu

Artifact	Baseline	Commit	Change
/x86_64-unknown-linux-gnu/lib/libdatadog_profiling.a	95.42 MB	95.42 MB	0% (0 B) 👌
/x86_64-unknown-linux-gnu/lib/libdatadog_profiling.so	11.17 MB	11.17 MB	0% (0 B) 👌

[hoolioh]

Instead of hardcoding these values in the workflow files we could setup a profile for the CI in the Cargo.toml the downside is that you would need to include some plumbering in the builder crate in order to pass the profile. That way we could reuse that profile in libdatadog, libddprof-build and also downstream projects that use the builder. Wdyt? It can be done in a subsequent PR if you want to test this approach first.
Also another advantage is that you can include the profile in the cache key rather than manually setting them depending on the job.