Deterministic SIGSEGV crash-loop in extension-process at relative offset 0x2ad266 — survives full profile wipe + reinstall
-
davilarjbr last edited by
Describe the bug
Opera crashes in rapid bursts (dozens of renderer crashes within ~10 seconds, then self-stabilizes without a full restart) with SIGSEGV at the exact same relative/PIE code offset0x2ad266inside the mainoperabinary, always in a renderer process launched with--extension-process.To Reproduce
Crash occurs spontaneously during normal browsing; I have not yet isolated it to one specific user action. Confirmed via local minidump analysis (Pythonminidumplibrary) across dozens of.dmpfiles from two separate crash episodes — the large majority match the identical exception address.Expected behavior
Browser should not crash-loop; a single renderer crash should not cascade into dozens of crashes in a few seconds.Environment
- OS: Debian GNU/Linux 13 (trixie), amd64
- Opera: reproduced identically on both 132.0.5905.37 and 140.0.6074.18
- GPU: AMD Carrizo R7 integrated, ozone-platform=x11
- Extensions: only Opera's own built-in components present (Rich Hints Agent, Continue On Support, Browser Operator, Opera AI, Opera Shopping) — no third-party extensions installed
Key detail ruling out local/profile causes
This exact crash signature (same code offset, same process type) was already present before I did a fullapt purgeof opera-stable plus a wipe of~/.config/operaand~/.cache/opera, followed by a clean reinstall on 2026-11-13. The crash-loop recurred identically on the fresh profile a few days later, on the new version. This rules out profile corruption, stale experiment/flag state, and any user-installed extension as the cause — it looks like a genuine bug in the browser binary itself, possibly related to this GPU/driver combination.Happy to upload raw .dmp files or run further diagnostics if that helps triage.