Stop shipping ARM32 builds for releases? #32375

issue laanwj opened this issue on April 29, 2025
  1. laanwj commented at 11:43 AM on April 29, 2025: member

    Clearly there is no hurry here. This is more of a brainstorming issue than a proposal.

    For:

    • 64-bit ARM is the standard nowadays for the type of computing hardware our users tend to have (modern Raspberry Pi's and ARM servers). While 32-bit ARM is only still used in low-power embedded systems, an unlikely combination with a bitcoin node.

    • 32-bit ARM is the last 32-bit architecture that we ship releases for. After removing this, we could fully focus on optimizing for large virtual address spaces. This means nearly-unrestricted mmap, and no need to worry about the virtual memory overhead of more thread stacks.

    • Is it still being properly tested at all? If no one really cares, things could break without us noticing.

    Against:

    • There are likely still users that use 2010-era ARM boxes for their nodes, though the number is likely to be dwindling (mine certainly died by now).

    • Requiring 64-bit even for self-built binaries is a step further, it might be wise to keep the 32-bit door open, at least i'm always hesitant with anything that limits potential hardware that can be used for bitcoin nodes.

      • OTOH we'd be in good company. Monero never supported 32-bit platforms for their node (because of using lmdb).

    See also bitcoin-core/bitcoincore.org#1126, where the default was recently changed to 64-bit.

  2. laanwj added the label Brainstorming on Apr 29, 2025
  3. laanwj added the label Linux/Unix on Apr 29, 2025
  4. achow101 commented at 5:11 PM on April 29, 2025: member

    IIRC, Raspberry Pi OS only started shipping 64-bit OSes a couple of years ago. For a while, you could buy a Pi 3 or 4 which had 64-bit hardware but the available Raspberry Pi OS image was only 32-bit. I would imagine that several of these machines are probably still in use. It also seems like the 32-bit images are still prominently shown on their website so I think it may be a bit premature to stop shipping 32-bit?

  5. willcl-ark commented at 9:56 AM on April 30, 2025: member

    I think I heard that that the Pi 5 ARM chips are unable to boot a 32bit kernel (although you can still hack about in a 32 bit userspace if you really want), so it seems that the day is definitely getting nearer.

    If I am correct above the question then perhaps becomes "how long do we want to ship release builds for Pi 3's and 4's".

  6. maflcko commented at 11:01 AM on April 30, 2025: member

    If this is dropped in the future, I guess we'll wholesale drop support for 32-bit architectures? If there is no release 32-bit build that we feel like is worth keeping, then I wonder if it is worth it to check and support it in CI.

  7. l0rinc commented at 11:49 AM on April 30, 2025: contributor

    As mentioned before, 32 bit support isn't tested as thoroughly as 64 (e.g. https://github.com/bitcoin/bitcoin/pull/28531/files#r2036954366), it would simplify a few scenarios if we didn't have walk on eggshells when doing storage size calculations (e.g. https://github.com/bitcoin/bitcoin/blob/master/src/util/byte_units.h#L15)

  8. laanwj commented at 12:43 PM on April 30, 2025: member

    IIRC, Raspberry Pi OS only started shipping 64-bit OSes a couple of years ago. For a while, you could buy a Pi 3 or 4 which had 64-bit hardware but the available Raspberry Pi OS image was only 32-bit. I would imagine that several of these machines are probably still in use. It also seems like the 32-bit images are still prominently shown on their website so I think it may be a bit premature to stop shipping 32-bit?

    Agree, there's no hurry.

    I think I heard that that the Pi 5 ARM chips are unable to boot a 32bit kernel (although you can still hack about in a 32 bit userspace if you really want), so it seems that the day is definitely getting nearer.

    Correct, it's more complicated for ARM than for x86*, but broadly modern 64-bit SoCs don't support 32-bit kernels. Most (but not all) do support 32-bit user-space, but there's less compatibility legacy to warrant keeping even that circuitry.

    Performance-wise, using 32-bit applications on hardware that supports 64-bit ARM is a really bad idea, and should be discouraged.

    * See "AArch32" column here https://gpages.juszkiewicz.com.pl/arm-socs-table/arm-socs.html

  9. sedited referenced this in commit 3f9c55426a on May 7, 2026
  10. maflcko commented at 8:00 AM on June 28, 2026: member

    I think the main issue is that there is no supported way to run the integer sanitizer on armhf. Running the clang integer sanitizer is essential, because on 32-bit architectures, size_t is only u32. A harmless-looking calculation like size_t{450_MiB}*10/100 will give a different result on different architectures.

    See the related commits: like 3789215f73466606eb111714f596a2a5e9bb1933, ac76d94117be70d2dcc23ba34b120b44aeb3b0c1, or 28a523fb94d333fd8a28ca101cce746157a90fb6 ...

    In theory one could use the i386 integer sanitizer as a proxy, but this is tedious, because one has to work around a bunch of irrelevant warnings about -D_TIME_BITS=64 missing or irrelevant warnings about not using fseeko/ftello.

    I am not saying that removal should be rushed, but I think we should acknowledge that a lot less quality assurance is happening on armhf.

  11. Kino1994 referenced this in commit f032783908 on Jun 28, 2026
  12. willcl-ark commented at 12:02 PM on June 29, 2026: member

    Another (small) note; 32 bit arm builds also don't have static PIE support on our current glibc version (although it should be in 2.44 according to this comment). Having to support old platforms like arm32 (and ppc64) will naturally hamper our ability to make beneficial changes (like -static-pie) occasionally, for arguably little benefit, as their support gets poorer and poorer.

  13. BigcoinBGC referenced this in commit 7f28737408 on Jun 30, 2026
  14. sedited referenced this in commit 5223cf1795 on Jul 9, 2026
  15. Kino1994 referenced this in commit 9a0fbb1cc3 on Jul 10, 2026
  16. TheBlueMatt commented at 6:36 PM on July 20, 2026: contributor

    In theory one could use the i386 integer sanitizer as a proxy, but this is tedious, because one has to work around a bunch of irrelevant warnings about -D_TIME_BITS=64 missing or irrelevant warnings about not using fseeko/ftello.

    Have you looked into testing on the x32 ABI? Its probably mostly unmaintained but...

  17. Kino1994 referenced this in commit 82e03f3471 on Aug 19, 2026
  18. Kino1994 referenced this in commit 6f5304fa64 on Aug 19, 2026
  19. willcl-ark commented at 9:11 AM on November 4, 2026: member

    @l0rinc added some more datapoints to this discussion in #36659, copied below here:

    We have found few recent reports of 32-bit Bitcoin Core node use, yet we regularly run into problems with architecture-dependent types such as size_t.

    Previous discussions focused on shipping 32-bit x86 Linux binaries and ARM32 release binaries. Keeping source support retains the code and testing burden even after binary distribution ends.

    Given these recurring costs, I propose requiring a 64-bit target for Bitcoin Core node builds, including builds from source. That would let us retire the 32-bit node CI and release configurations and simplify node code that exists only to accommodate 32-bit targets.

    <details> <summary>Further examples of the work created by 32-bit support</summary>

    • #33709 and #33724 replaced architecture-dependent serialized-size arithmetic after the published crash mitigation. The review discussion explains the recurring need to reason about intermediate overflow and checked narrowing.
    • #35097 changed byte-unit literals because size_t prevented expressing larger storage sizes and limits on 32-bit targets.
    • #7848 fixed UTXO hash results that differed between 32-bit and 64-bit builds because a size_t counter wrapped.
    • #24214 made interpreter arithmetic independent of host width and reduced integer-sanitizer suppressions.
    • #34110 exposed a 32-bit Base58 overflow during wallet fuzzing and documents the additional effort needed to reproduce it.
    • #30307 needed different handling for file-position errors that occurred earlier on 32-bit systems.
    • #29415's review revisited architecture-dependent counter types during private-broadcast work. #32497's review compared arithmetic across compilers and 32-bit/64-bit targets for a small allocation optimization.
    • #34108 and #34136 required changing a functional test whose memory-accounting assumptions failed on 32-bit builds. #34132's review needed different storage-corruption test setup because the 32-bit database path behaved differently.
    • #28531's test discussion describes the difficulty of maintaining memory-usage expectations across both widths. #35773 needed a sanitizer suppression for intentional hash truncation on 32-bit targets.
    • #16096 raised declining developer testing and assumptions about size_t, time_t, and long. #35466 later repaired ARM32 IPC test coverage that had been skipped because a dependency was missing.
    • #27879, #31965, and the later hosting discussion show ARM32 execution restricting CI hardware and provider choices. #33302 needed an additional runner/cache exception.
    • #35546 revisited an i686 compiler choice originally made to avoid memory exhaustion. #31772 records another 32-bit compiler/build-mode combination causing an IPC test failure.
    • #3324 documents longstanding virtual-address-space exhaustion. #28906 and #28913 show ARM32 alignment assumptions causing a later memory regression.
    • #33333 needed arithmetic ordering chosen to avoid 32-bit overflow. #34692 retained a separate 32-bit cache default. The unmerged #35620 also illustrates separate database-cache behavior complicating proposed changes.
    • #2203, #2236, #438, and #439 needed large-file accommodations for 32-bit builds.
    • #29980 and #29985 required dependency changes during the time64 transition. #25573's ARM32 discussion describes extra toolchain and IPC work for static PIE.
    • #28096 found a nominally 32-bit build producing a 64-bit executable after losing target flags. #22889 and #22923 prompted additional 32-bit Clang coverage after compilation failures.
    • #18686, #19735, and #22464 are smaller examples of warning and narrowing fixes needed specifically for 32-bit targets.
    • #15939 and #22908 retired Win32 release and build configurations. #18104 stopped default i686 release builds while explicitly retaining source builds and CI, leaving later cleanup in #19745.
    • #23744 removed an i686 Android target that had been broken unnoticed for months. #24448, #24639, and #26075 show the compatibility work and eventual cleanup around discontinued i686 release builds.

      </details>

  20. pinheadmz commented at 4:19 PM on November 4, 2026: member

    I agree that it's too much work to maintain. Note that Raspberry Pi is still releasing 32-bit OS, and as of last year has no plan to stop

    I personally still run the 32-bit on my RPi! When they first released the 64 bit I tried it and had issues. I'm sure it's fine by now but I continue using the 32 bit option pretty much just for the sake of bitcoin core development, having that platform available to run bitcoind, lnd, and some other stuff all together. SO count me as one of those last remaining hold outs.

    But also put another way, I'm waiting for bitcoin core to drop 32 bit support so I can upgrade my RPis to 64 bit OS


github-metadata-mirror

This is a metadata mirror of the GitHub repository bitcoin/bitcoin. This site is not affiliated with GitHub. Content is generated from a GitHub metadata backup.
generated: 2026-11-20 14:52 UTC

This site is hosted by @0xB10C
More mirrored repositories can be found on mirror.b10c.me