emacs: pin clock_gettime in fixrand shim for reproducible .pdmp by bryan-minimal · Pull Request #273 · gominimal/pkgs

bryan-minimal · 2026-06-18T18:04:23Z

What

Makes the emacs package reproducible by extending the existing fixrand.so LD_PRELOAD shim to also pin clock_gettime(CLOCK_REALTIME).

Why

After the current shim (which makes getrandom()//dev/urandom deterministic for hash-table seeding), the dumped emacs-*.pdmp still differed across builds — in exactly 11 bytes, in two clusters, both decoding to wall-clock timestamps captured at dump time via clock_gettime(CLOCK_REALTIME) (which the shim didn't intercept):

*scratch* buffer's buffer-display-time — an Emacs (TICKS . HZ) timestamp cons set by Fcurrent_time() during make_initial_frame (src/window.c / src/emacs.c). The two builds decoded to 2026-06-16 04:01:42 vs 04:02:43 — exactly the ~61s gap between them.
Vgc_elapsed — accumulated GC time (a float) in src/alloc.c, measured with current_timespec().

SOURCE_DATE_EPOCH doesn't help here because Emacs reads the live clock directly.

Fix

Add a clock_gettime interceptor to fixrand.c that returns SOURCE_DATE_EPOCH for CLOCK_REALTIME (and passes all other clocks, e.g. CLOCK_MONOTONIC, through untouched so timeouts/progress loops are unaffected). The build.sh compile line is unchanged — it already builds fixrand.c with -ldl.

This pins the display-time and makes GC start == end (elapsed 0.0), both deterministic. No literal SOURCE_DATE_EPOCH=0 (uses getenv + fallback), so it passes minimal-check.

Verification

Build-twice-and-diff on aarch64: two from-scratch builds are now byte-for-byte identical (repro-check diff, 4282/4282 files). Before, the .pdmp differed in 11 timestamp bytes.

🤖 Generated with Claude Code

Summary by CodeRabbit

Bug Fixes
- Improved reproducibility of Emacs builds by freezing system time readings during snapshot generation, ensuring consistent and deterministic output across different build environments and execution times.

After the existing fixrand.so shim (getrandom/urandom), the dumped .pdmp still differed in 11 bytes — two wall-clock timestamps read via clock_gettime(CLOCK_REALTIME), which the shim didn't cover: the *scratch* buffer's buffer-display-time (Fcurrent_time in make_initial_frame) and Vgc_elapsed GC timing (src/alloc.c). Extend the shim to return SOURCE_DATE_EPOCH for CLOCK_REALTIME (other clocks pass through). Verified on aarch64: two from-scratch builds are byte-identical (repro-check diff, 4282/4282 files). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

coderabbitai · 2026-06-18T18:04:38Z

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: b80a25de-cd70-4399-ab7b-5133dc793bf6

📥 Commits

Reviewing files that changed from the base of the PR and between fe301db and e4ccc06.

📒 Files selected for processing (1)

packages/emacs/fixrand.c

🚧 Files skipped from review as they are similar to previous changes (1)

packages/emacs/fixrand.c

📝 Walkthrough

Walkthrough

fixrand.c gains a clock_gettime LD_PRELOAD interposer that intercepts CLOCK_REALTIME calls and returns a fixed timestamp sourced from SOURCE_DATE_EPOCH (defaulting to 0), making wall-clock reads during Emacs .pdmp generation reproducible. All other clock IDs delegate to the real function.

Changes

clock_gettime determinism shim

Layer / File(s)	Summary
clock_gettime interposition: headers, docs, and implementation `packages/emacs/fixrand.c`	File comment extended to document `CLOCK_REALTIME` freezing via `SOURCE_DATE_EPOCH`; `<stdlib.h>` and `<time.h>` headers added; new `clock_gettime` interposer lazily resolves the real function via `dlsym`, overrides `CLOCK_REALTIME` with `SOURCE_DATE_EPOCH` parsed as seconds (`tv_nsec=0`), and passes all other clock IDs through unchanged.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~8 minutes

Suggested reviewers

twitchyliquid64

Poem

🐇 Tick-tock, the clock stands still,
SOURCE_DATE_EPOCH bends time to my will.
No random wall-clock sneaking in,
The .pdmp dump can now begin!
Reproducible builds — a rabbit's thrill. 🕰️

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 50.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately describes the main change: extending the fixrand shim to intercept clock_gettime for reproducible .pdmp files in Emacs builds.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

📝 Generate docstrings

Create stacked PR
Commit on current branch

🧪 Generate unit tests (beta)

Create PR with unit tests
Commit unit tests in branch bryan/emacs-reproducible

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

coderabbitai

Actionable comments posted: 1

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@packages/emacs/fixrand.c`:
- Around line 75-85: The real_clock_gettime function pointer can remain NULL if
dlsym fails to load the symbol, which will cause a NULL pointer dereference when
called for non-CLOCK_REALTIME clock IDs. Add a NULL check for real_clock_gettime
before calling it on the final return statement, and either log an error or
return an appropriate error code (such as -1 with errno set to ENOSYS) if the
symbol lookup failed.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

Push a commit to this branch (recommended)
Create a new PR with the fixes

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: 9567d1f9-d904-47db-ac7c-949595cb483a

📥 Commits

Reviewing files that changed from the base of the PR and between b265510 and fe301db.

📒 Files selected for processing (1)

packages/emacs/fixrand.c

Defensive NULL check before deref of real_clock_gettime: if dlsym(RTLD_NEXT, "clock_gettime") ever returns NULL, fail with ENOSYS instead of segfaulting. The CLOCK_REALTIME path already returns before this point, so this only affects pass-through clocks (CLOCK_MONOTONIC, etc.) — but it makes the shim robust regardless. Folds in the one valid point from the Edge Delta bot's #278. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

coderabbitai Bot reviewed Jun 18, 2026

View reviewed changes

Comment thread packages/emacs/fixrand.c

This was referenced Jun 19, 2026

build hardening: filter ldd copies, guard sed patches, validate strtoll #274

Closed

emacs/fixrand.c: guard against NULL real_clock_gettime from dlsym failure #278

Closed

edge-delta Bot mentioned this pull request Jun 20, 2026

emacs: resolve dlsym symbols in constructor to prevent clock_gettime deadlock #281

Closed

6 tasks

mitodrummer approved these changes Jun 22, 2026

View reviewed changes

bryan-minimal added this pull request to the merge queue Jun 22, 2026

Merged via the queue into main with commit 838b255 Jun 22, 2026
4 checks passed

bryan-minimal deleted the bryan/emacs-reproducible branch June 22, 2026 17:29

Navigation Menu

Uh oh!

emacs: pin clock_gettime in fixrand shim for reproducible .pdmp - #273

emacs: pin clock_gettime in fixrand shim for reproducible .pdmp#273
bryan-minimal merged 2 commits into
mainfrom
bryan/emacs-reproducible

bryan-minimal commented Jun 18, 2026 •

edited by coderabbitai Bot

Loading

Uh oh!

coderabbitai Bot commented Jun 18, 2026 •

edited

Loading

Walkthrough

Changes

Estimated code review effort

Suggested reviewers

Poem

❌ Failed checks (1 warning)

Uh oh!

coderabbitai Bot left a comment

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Conversation

bryan-minimal commented Jun 18, 2026 • edited by coderabbitai Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

What

Why

Fix

Verification

Summary by CodeRabbit

Uh oh!

coderabbitai Bot commented Jun 18, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Walkthrough

Changes

Estimated code review effort

Suggested reviewers

Poem

❌ Failed checks (1 warning)

Uh oh!

coderabbitai Bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

bryan-minimal commented Jun 18, 2026 •

edited by coderabbitai Bot

Loading

coderabbitai Bot commented Jun 18, 2026 •

edited

Loading