Re: v2.26.0 fails to start with return code -1073741569 (0xC00000FD)

Jonas Hahnfeld via bug-lilypond <[email protected]> Sat, 25 Apr 2026 10:38:03 +0200
Newsgroups gmane.comp.gnu.lilypond.bugs
Message-ID <[email protected]>
--=-6OesZ7fcyQrzyE8+eebS
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Fri, 2026-04-24 at 14:11 -0500, [email protected] wrote:
> Context:
>=20
> Windows 11 25H2
>=20
> ARM processor
>=20
> Frescobaldi v4.0.5
>=20
> LilyPond v2.26.0

The ARM aspect may indeed be very relevant (thanks Karlin for pointing
out this aspect explicitly!): We only offer x86_64 binaries, so Windows
has to "emulate" / "translate" the machine instructions. Given this
worked so far and no problems were reported, I didn't think too much
about this.
However, in LilyPond 2.26.0 we switched to Guile 3.0. In principle,
this should be transparent to users but it now performs just-in-time
compilation (JIT). As a guess, this could cause problems when running
x86 binaries on ARM. As a test, can you set the environment variable
GUILE_JIT_THRESHOLD=3D-1 to disable the JIT, and verify with a second
variable GUILE_JIT_LOG=3D1 that indeed it's not running anymore. Maybe
this helps in the short term, but we might need to look into native
binaries for Windows + ARM...

Jonas

--=-6OesZ7fcyQrzyE8+eebS
Content-Type: application/pgp-signature; name="signature.asc"
Content-Description: This is a digitally signed message part

-----BEGIN PGP SIGNATURE-----

iHUEABYKAB0WIQRiF4DLRvon1vw5biCcVOmY/6OEDwUCaex9awAKCRCcVOmY/6OE
DzPZAQDWvYCnEKjfWq5jN4tzXHb1jas+6tAf2dORRaD5D9LSRwEAt7vf1pV2PSD0
KbudWSAWQQUfpumdclvxsxb853ZOyQU=
=2UcT
-----END PGP SIGNATURE-----

--=-6OesZ7fcyQrzyE8+eebS--