| Age | Commit message (Collapse) | Author |
|---|
|
While the package is well documented for potential users (the pkgdesc
field is clear enough), it has almost zero documentation for potential
contributors, and the fact that pkgdesc is generated doesn't help
making it use clear.
This adds some documentation to help maintenance, and also documents
some use cases, as a potential user can easily understand why to use
or not use such kernel, but this can be less clear to contributors
wanting to modify/remove/replace the package in ways that could break
these use cases.
Signed-off-by: Denis 'GNUtoo' Carikli <GNUtoo@cyberdimension.org>
|
|
Signed-off-by: Denis 'GNUtoo' Carikli <GNUtoo@cyberdimension.org>
|
|
The status quo is that any Parabola hacker is expected to (be able to)
modify any packages, and having a single maintainer of a package
discourages that practice as people would typically send a patch to
the maintainer instead of pushing it directly.
So for a start we can add common maintainership on package lacking any
"Maintainer: " header for packages in repositories that are supposed
to be maintained.
As for finding who worked on a given package (in case it could be
needed), the git log should have all the information.
Signed-off-by: Denis 'GNUtoo' Carikli <GNUtoo@cyberdimension.org>
|
|
Without that fix, on Parabola armv7h we have:
| Stripping vmlinux...
| strip:
/build/linux-libre-64/pkg/linux-libre-64-headers/usr/lib/modules/5.11.11-gnu-2-64-ARCH/build/vmlinux:
file format not recognized
| ==> ERROR: A failure occurred in package_linux-libre-64-headers().
| Aborting...
==> Copying log and package files out of the chroot...
And after entering the chroot again, running file on vmlinux in
usr/lib/modules/5.11.11-gnu-2-64-ARCH/build/ in
build/linux-libre-64/pkg/linux-libre-64-headers/ gives the following:
ELF 64-bit LSB shared object, ARM aarch64,
version 1 (SYSV), statically linked,
BuildID[sha1]=61b2acea5733ab812f33e872c462a54e7e71fbeb, not stripped
So the file is not recognized because we use the host strip instead of
the cross compiler's strip.
Signed-off-by: Denis 'GNUtoo' Carikli <GNUtoo@cyberdimension.org>
|
|
Without that fix, when building for Parabola armv7h we have:
| ==> Starting package_linux-libre-64-headers()...
| Installing build files...
| Installing headers...
| cp: cannot stat 'arch/arm64/mach-dove/include': No such file or directory
| ==> ERROR: A failure occurred in package_linux-libre-64-headers().
| Aborting...
However these headers do exist for armv7h. As in this case we are
building an aarch64 kernel, we don't need armv7h headers anyway.
Signed-off-by: Denis 'GNUtoo' Carikli <GNUtoo@cyberdimension.org>
|
|
Signed-off-by: Denis 'GNUtoo' Carikli <GNUtoo@cyberdimension.org>
|
|
|
|
Instead of fixing checksums each time the kernel
configuration is updated, we can simply skip them as they
come from the same git repository than the PKGBUILD itself,
so it should introduce any security issues.
Signed-off-by: Denis 'GNUtoo' Carikli <GNUtoo@cyberdimension.org>
|
|
This regression was introduced by the following commits:
63359d4f8 kernels: re-enable the b43 and b43-legacy drivers
2e94284f5 kernels: re-enable the tg3 driver
Signed-off-by: Denis 'GNUtoo' Carikli <GNUtoo@cyberdimension.org>
|
|
The b43 and b43-legacy drivers were removed because their
corresponding upstream drivers loads a nonfree firmware.
However linux-libre-firmware has a free firmware
(ucode5.fw) that is used by both drivers.
The b43 and b43-legacy driver are also typically enabled
by Trisquel and Jason Self's linux-libre repository.
Note that some kernel parameters like qos=0 might be
necessary to make it work.
This regression was introduced in linux-libre by the
following commit:
32075365d upgpkg: libre/linux-libre 4.16.9_gnu-1
The drivers were added back by importing affected parts
from the following upstream kernels and configurations:
- config in linux-armv7 from Arch Linux ARM
- config.i686 in linux from Arch Linux 32
- config from linux from Arch Linux
Signed-off-by: Denis 'GNUtoo' Carikli <GNUtoo@cyberdimension.org>
|
|
The tg3 driver was removed because its upstream driver
loads a nonfree firmware. However with linux-libre, the
driver works fine without any nonfree firmware, at least
on some computers like the HP NC6320.
That driver is also typically enabled by Trisquel and
Jason Self's linux-libre repository.
This regression was introduced in linux-libre by the
following commit:
32075365d upgpkg: libre/linux-libre 4.16.9_gnu-1
The following configuration:
CONFIG_TIGON3=m
CONFIG_TIGON3_HWMON=y
is used in the following upstream kernels and
configurations:
- config in linux-armv7 from Arch Linux ARM
- config.i686 in linux from Arch Linux 32
- config from linux from Arch Linux
Signed-off-by: Denis 'GNUtoo' Carikli <GNUtoo@cyberdimension.org>
|
|
Signed-off-by: Denis 'GNUtoo' Carikli <GNUtoo@cyberdimension.org>
|
|
Signed-off-by: Denis 'GNUtoo' Carikli <GNUtoo@cyberdimension.org>
|
|
Note that even if it fixes packaging, in some case the
compiler segfauls. In that case you need to restart the
build until it succeddeds, which is usually around 1 or 2
times when you're unlucky.
Signed-off-by: Denis 'GNUtoo' Carikli <GNUtoo@cyberdimension.org>
|
|
Without that fix, we have the following issue:
Removing documentation...
cp: cannot stat 'arch/arm64/mach-dove/include': No such file or directory
==> ERROR: A failure occurred in package_linux-libre-64-headers().
Aborting...
What is strange is that this error didn't appear when building
linux-libre-64-5.6.12-1.
Signed-off-by: Denis 'GNUtoo' Carikli <GNUtoo@cyberdimension.org>
|
|
Signed-off-by: Denis 'GNUtoo' Carikli <GNUtoo@cyberdimension.org>
|
|
Signed-off-by: Denis 'GNUtoo' Carikli <GNUtoo@cyberdimension.org>
|
|
Signed-off-by: Denis 'GNUtoo' Carikli <GNUtoo@cyberdimension.org>
|
