Re: install/58932: NetBSD-10.99.12-i386-install.img ACPI problems Compal DL-75 laptop

["Taylor R Campbell via gnats" <gnats-admin%NetBSD.org@localhost>]

The following reply was made to PR install/58932; it has been noted by GNATS.

From: Taylor R Campbell <riastradh%NetBSD.org@localhost>
To: Ramiro Aceves <ea1abz%gmail.com@localhost>
Cc: gnats-bugs%netbsd.org@localhost, install-manager%netbsd.org@localhost,
	gnats-admin%netbsd.org@localhost, netbsd-bugs%netbsd.org@localhost
Subject: Re: install/58932: NetBSD-10.99.12-i386-install.img ACPI problems
	Compal DL-75 laptop
Date: Thu, 26 Dec 2024 14:33:20 +0000

 This is a multi-part message in MIME format.
 --=_uYve6LJUJbusqt1t17g4r8OQ3Se97Nkl
 Content-Transfer-Encoding: quoted-printable
 
 > Date: Thu, 26 Dec 2024 08:25:02 +0100
 > From: Ramiro Aceves <ea1abz%gmail.com@localhost>
 >=20
 > I compiled the kernel with the patched file using ./build.sh command=20
 > like you explained (changing to i386). This way compilation succeded=20
 > fine, it build toolchain first after a long time in this weird, slow but=
 =20
 > beloved i386 machine.
 
 FYI, the command I gave you runs a cross-build, so you can do it on
 whatever is your most-capable machine -- no need to do it on a small,
 elderly i386 system!
 
 Also, once you've done `build.sh ... tools' once, no need to do it
 again; you can just do the `build.sh ... kernel=3DGENERIC' stage to
 build a new kernel.
 
 >                       It seems that the "manual way" method of building=20
 > the kernel as The Guide describe does not work in this case. I required=20
 > toolchain to be built even when there is no change in architecture.
 
 If you're building from (say) 10.0 to 10.99.x, you may still have to
 build a new toolchian -- e.g., right now, 10.99.x is on gcc-12, while
 10.0 is on gcc-10, and there are often also relevant changes in
 /usr/share/mk to handle changes in compiler warnings like the
 -Werror=3Daddress-of-packed-member failure you saw.
 
 So the instructions at
 <https://www.netbsd.org/docs/guide/en/chap-kernel.html#chap-kernel-building=
 -manually>
 (34.4 Building the kernel manually) are really only for building small
 updates to a kernel of the same major version (10.99.x should be
 thought of as 11.0 - epsilon rather than as 10.something), unless
 you're doing a cross-build.
 
 > I booted with the new kernel and it enters again in ddb, rebooted from=20
 > that and dmesg was saved. I attach dmesg below:
 
 Thanks, looks like the ichlpcib(4) timecounter might be busted in the
 same way -- and I suspect it's the same hardware underneath anyway
 (not really sure why we have two access paths to it).
 
 Can you try the attached patch on top of the last one I sent?
 
 --=_uYve6LJUJbusqt1t17g4r8OQ3Se97Nkl
 Content-Type: text/plain; charset="ISO-8859-1"; name="pr58932-testnoacpipmtimer"
 Content-Transfer-Encoding: quoted-printable
 Content-Disposition: attachment; filename="pr58932-testnoacpipmtimer.patch"
 
 diff -r 7c40a48ab841 sys/dev/ic/acpipmtimer.c
 --- a/sys/dev/ic/acpipmtimer.c	Tue Dec 24 12:13:05 2024 +0000
 +++ b/sys/dev/ic/acpipmtimer.c	Thu Dec 26 14:29:42 2024 +0000
 @@ -31,6 +31,8 @@ acpipmtimer_attach(device_t dev,
  		   bus_space_tag_t t, bus_space_handle_t h, bus_size_t off,
  		   int flags)
  {
 +	return NULL;
 +
  	struct hwtc *tc;
 =20
  	tc =3D malloc(sizeof(struct hwtc), M_DEVBUF, M_WAITOK|M_ZERO);
 
 --=_uYve6LJUJbusqt1t17g4r8OQ3Se97Nkl--