From Newsgroup: muc.lists.freebsd.ports


--3yTvd2OI+oJYJ/n3
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

It's possible that I'm doing something stupid (or just "wrong")....

On my main laptop, I track stable/15 & head (different slices -- I
started doing this over 20 years ago & update to a new laptop by copying
the disk drive -- the laptop is only about 10 years old).

In /etc/src.conf, I have:

g1-120(15.1-P)[2] grep '^PORT' /etc/src.conf PORTS_MODULES+=3Dgraphics/drm-66-kmod PORTS_MODULES+=3Dgraphics/gpu-firmware-amd-kmod PORTS_MODULES+=3Dgraphics/gpu-firmware-intel-kmod PORTS_MODULES+=3Dx11/nvidia-kmod-390

so the various kmod ports get rebuilt when I rebuild the kernel.

Today, I noticed the update from drm-66-kmod-6.6.25.1500508_8 to drm-66-kmod-6.6.25.1500508_9; no problem (so I thought).

I had built and run

FreeBSD g1-120.catwhisker.org 15.1-PRERELEASE FreeBSD 15.1-PRERELEASE #798 = stable/15-n283363-8d584f064885: Tue Apr 28 10:39:26 UTC 2026 root@g1-12= 0.catwhisker.org:/common/S1/obj/usr/src/amd64.amd64/sys/CANARY amd64 150050=
8 1500508

yesterday (without issue), but after updating sources to stable/15-n283375-bf1529d83918, then firing up the usual "buildworld"
& friends, the kernel build failed while attempting to build
drm-66-kmod, with whines about "taking address of packed member..."
of a class or structure:

--- intel_guc_ct.o --- /common/S1/obj/usr/src/amd64.amd64/sys/CANARY/common/ports/graphics/drm-66-= kmod/work/drm-kmod-drm_v6.6.25_10/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c= :606:19: error: taking address of packed member 'head' of class or structur=
e 'guc_ct_buffer_desc' may result in an unaligned pointer value [-Werror,-W= address-of-packed-member]
606 | head =3D READ_ONCE(desc->head);
| ^~~~~~~~~~ /usr/src/sys/compat/linuxkpi/common/include/linux/compiler.h:81:37: note: e= xpanded from macro 'READ_ONCE'
81 | (*(const volatile __typeof(x) *)&(x)); \
| ^


Any suggestions for suitable evasive action?

Thanks!

Peace,
david
--=20
David H. Wolfskill david@catwhisker.org

See https://www.catwhisker.org/~david/publickey.gpg for my public key.

--3yTvd2OI+oJYJ/n3
Content-Type: application/pgp-signature; name=signature.asc

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

iNUEARYKAH0WIQRCec5RsK7Enudh3yGB9MJ9AwUELQUCafHnf18UgAAAAAAuAChp c3N1ZXItZnByQG5vdGF0aW9ucy5vcGVucGdwLmZpZnRoaG9yc2VtYW4ubmV0NDI3 OUNFNTFCMEFFQzQ5RUU3NjFERjIxODFGNEMyN0QwMzA1MDQyRAAKCRCB9MJ9AwUE Lbi3AQD5kIrhrxtcWEyAeJtuY11s7aq+Cs14TI/cqLQxRXpxBwEAszEggcslSbXY yC6MOgsLD+4PVZ9zdLrRD2iy5IetAQk=
=jQ8G
-----END PGP SIGNATURE-----

--3yTvd2OI+oJYJ/n3--


--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-admin@muc.de
--- Synchronet 3.21f-Linux NewsLink 1.2

From Newsgroup: muc.lists.freebsd.ports

On Wed, 29 Apr 2026 04:11:59 -0700
David Wolfskill <david@catwhisker.org> wrote:

It's possible that I'm doing something stupid (or just "wrong")....

On my main laptop, I track stable/15 & head (different slices -- I
started doing this over 20 years ago & update to a new laptop by copying
the disk drive -- the laptop is only about 10 years old).

In /etc/src.conf, I have:

g1-120(15.1-P)[2] grep '^PORT' /etc/src.conf PORTS_MODULES+=graphics/drm-66-kmod PORTS_MODULES+=graphics/gpu-firmware-amd-kmod PORTS_MODULES+=graphics/gpu-firmware-intel-kmod PORTS_MODULES+=x11/nvidia-kmod-390

so the various kmod ports get rebuilt when I rebuild the kernel.

Today, I noticed the update from drm-66-kmod-6.6.25.1500508_8 to drm-66-kmod-6.6.25.1500508_9; no problem (so I thought).

I had built and run

FreeBSD g1-120.catwhisker.org 15.1-PRERELEASE FreeBSD 15.1-PRERELEASE #798 stable/15-n283363-8d584f064885: Tue Apr 28 10:39:26 UTC 2026 root@g1-120.catwhisker.org:/common/S1/obj/usr/src/amd64.amd64/sys/CANARY amd64 1500508 1500508

yesterday (without issue), but after updating sources to stable/15-n283375-bf1529d83918, then firing up the usual "buildworld"
& friends, the kernel build failed while attempting to build
drm-66-kmod, with whines about "taking address of packed member..."
of a class or structure:

--- intel_guc_ct.o --- /common/S1/obj/usr/src/amd64.amd64/sys/CANARY/common/ports/graphics/drm-66-kmod/work/drm-kmod-drm_v6.6.25_10/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c:606:19: error: taking address of packed member 'head' of class or structure 'guc_ct_buffer_desc' may result in an unaligned pointer value [-Werror,-Waddress-of-packed-member]
606 | head = READ_ONCE(desc->head);
| ^~~~~~~~~~ /usr/src/sys/compat/linuxkpi/common/include/linux/compiler.h:81:37: note: expanded from macro 'READ_ONCE'
81 | (*(const volatile __typeof(x) *)&(x)); \
| ^

Any suggestions for suitable evasive action?

Thanks!

Peace,
david
--
David H. Wolfskill david@catwhisker.org

See https://www.catwhisker.org/~david/publickey.gpg for my public key.

For anyone not yet noticed it (David and manu@ is already there):
I've filed Bug 294870 for this.

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=294870

graphics/drm-61-kmod has the same issue, too.

Noticed just before submitting updates for NVIDIA driver sets
595.71.05, pending it for now (need regenerating patch anyway).

Regards.
--
Tomoaki AOKI <junchoon@dec.sakura.ne.jp>


--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-admin@muc.de
--- Synchronet 3.21f-Linux NewsLink 1.2