Firefox 32-bit Linux Support to End in 2026

Original Article ↗ Hacker News Discussion ↗

Rationale and scale of impact

  • Telemetry suggests very few Firefox users are on 32‑bit at all; extrapolations put 32‑bit Linux x86 users at around 0.1% of Firefox users or “a few hundred to a few thousand.”
  • Many major distros and Chrome have already dropped full 32‑bit x86 support; by Mozilla’s cutoff date, most 32‑bit x86 distros will be in extended-support mode only.
  • Several commenters argue an open-source project still has to prioritize limited resources; others counter that open source typically supports any platform where volunteers keep it building.

Usability of 32‑bit-era hardware on the modern web

  • One camp: browsing on old 32‑bit CPUs is “miserable” because of RAM limits and slow JavaScript; examples include Gmail taking ~1 minute and ~500MB RAM on low-end hardware.
  • Another camp: with a lightweight Linux distro, an adblocker, and few tabs, even very old systems (Pentium, Atom netbooks, N270, etc.) can handle basic reading, email (non‑web), and niche protocols (Gopher, Gemini).
  • Several note that the browser itself is heavier now even on a blank page, and that “web browsing” is no longer a basic task because many sites are React SPAs overloaded with ads and video.

32‑bit vs 64‑bit: memory, stability, and user behavior

  • Some insist 32‑bit browsers genuinely use less RAM and that this matters on 2–4GB machines; this may explain the sizable share of 32‑bit Firefox on 64‑bit Windows.
  • Others report 32‑bit Firefox being unstable on 64‑bit systems with large, long‑lived profiles, likely due to 32‑bit address-space limits rather than missing libraries.
  • There’s concern that dropping 32‑bit will slowly break newer sites for these users, though polyfills defer that somewhat.

Dropping older x86‑64 CPUs and ISA baselines

  • Several argue Mozilla could also raise the x86‑64 baseline (e.g., x86‑64‑v2) and use instructions like CMPXCHG16B unconditionally, matching what modern OSes already require.
  • Discussion covers techniques like:
    • Multiple function implementations selected at runtime (glibc-style),
    • Kernel-style instruction patching (ALTERNATIVE),
    • Tradeoffs between portability, binary size, and small (< single‑digit %) performance gains.
  • Consensus: aggressive multi‑ISA support across the whole browser is complex and not “free.”

Who’s left on 32‑bit and what next

  • Remaining users include embedded systems, kiosks, some Raspberry Pi setups (though this drop is x86‑only), and retro/low‑power enthusiasts.
  • Many believe such users are technically savvy enough to:
    • Stay on ESR for the final year of security updates,
    • Switch to forks like Waterfox or Pale Moon,
    • Or rely on distros/communities (Gentoo, Arch32, Devuan/Derivatives) that keep 32‑bit ecosystems alive.
  • Several see this as a reasonable line to draw; others worry about “how small is too small” before a user group is effectively ignored.