[PATCH] kernel: set COMPATIBLE_HOST to *-linux

[PATCH] kernel: set COMPATIBLE_HOST to *-linux

[Ross Burton]

The target system triple contains whether the target is Linux or not,
so use it to avoid situations where you can attempt to build a kernel
for systems which don't support Linux.

Signed-off-by: Ross Burton <ross.burton@arm.com>
---
 meta/classes/kernel.bbclass              | 2 ++
 meta/recipes-kernel/linux/linux-dummy.bb | 2 +-
 2 files changed, 3 insertions(+), 1 deletion(-)

diff --git a/meta/classes/kernel.bbclass b/meta/classes/kernel.bbclass
index 1a444efabf..2f1006ddf0 100644
--- a/meta/classes/kernel.bbclass
+++ b/meta/classes/kernel.bbclass
@@ -1,5 +1,7 @@
 inherit linux-kernel-base kernel-module-split
 
+COMPATIBLE_HOST = ".*-linux"
+
 KERNEL_PACKAGE_NAME ??= "kernel"
 KERNEL_DEPLOYSUBDIR ??= "${@ "" if (d.getVar("KERNEL_PACKAGE_NAME") == "kernel") else d.getVar("KERNEL_PACKAGE_NAME") }"
 
diff --git a/meta/recipes-kernel/linux/linux-dummy.bb b/meta/recipes-kernel/linux/linux-dummy.bb
index 649fc04dd1..95dc85ff2f 100644
--- a/meta/recipes-kernel/linux/linux-dummy.bb
+++ b/meta/recipes-kernel/linux/linux-dummy.bb
@@ -26,7 +26,7 @@ DESCRIPTION_kernel-vmlinux = "Kernel vmlinux meta package"
 
 INHIBIT_DEFAULT_DEPS = "1"
 
-#COMPATIBLE_MACHINE = "your_machine"
+COMPATIBLE_HOST = ".*-linux"
 
 PR = "r1"
 
-- 
2.25.1

Re: [OE-core] [PATCH] kernel: set COMPATIBLE_HOST to *-linux

[Khem Raj]

On 12/11/20 8:09 AM, Ross Burton wrote:
> The target system triple contains whether the target is Linux or not,
> so use it to avoid situations where you can attempt to build a kernel
> for systems which don't support Linux.
> 
> Signed-off-by: Ross Burton <ross.burton@arm.com>
> ---
>   meta/classes/kernel.bbclass              | 2 ++
>   meta/recipes-kernel/linux/linux-dummy.bb | 2 +-
>   2 files changed, 3 insertions(+), 1 deletion(-)
> 
> diff --git a/meta/classes/kernel.bbclass b/meta/classes/kernel.bbclass
> index 1a444efabf..2f1006ddf0 100644
> --- a/meta/classes/kernel.bbclass
> +++ b/meta/classes/kernel.bbclass
> @@ -1,5 +1,7 @@
>   inherit linux-kernel-base kernel-module-split
>   
> +COMPATIBLE_HOST = ".*-linux"

does this cover linux-gnueabi and such tuples too ?

> +
>   KERNEL_PACKAGE_NAME ??= "kernel"
>   KERNEL_DEPLOYSUBDIR ??= "${@ "" if (d.getVar("KERNEL_PACKAGE_NAME") == "kernel") else d.getVar("KERNEL_PACKAGE_NAME") }"
>   
> diff --git a/meta/recipes-kernel/linux/linux-dummy.bb b/meta/recipes-kernel/linux/linux-dummy.bb
> index 649fc04dd1..95dc85ff2f 100644
> --- a/meta/recipes-kernel/linux/linux-dummy.bb
> +++ b/meta/recipes-kernel/linux/linux-dummy.bb
> @@ -26,7 +26,7 @@ DESCRIPTION_kernel-vmlinux = "Kernel vmlinux meta package"
>   
>   INHIBIT_DEFAULT_DEPS = "1"
>   
> -#COMPATIBLE_MACHINE = "your_machine"
> +COMPATIBLE_HOST = ".*-linux"
>   
>   PR = "r1"
>   
> 
> 
> 
> 
> 

Re: [OE-core] [PATCH] kernel: set COMPATIBLE_HOST to *-linux

[Ross Burton]

On Fri, 11 Dec 2020 at 16:53, Khem Raj <raj.khem@gmail.com> wrote:

> > +COMPATIBLE_HOST = ".*-linux"
>
> does this cover linux-gnueabi and such tuples too ?

Maybe?  Depends on the matching in anchored.  If it does by accident
then explicitly matching would be better.  What machine has a triple
like that for testing?

Ross

Re: [OE-core] [PATCH] kernel: set COMPATIBLE_HOST to *-linux

[Richard Purdie]

On Sun, 2020-12-13 at 22:20 +0000, Ross Burton wrote:
> On Fri, 11 Dec 2020 at 16:53, Khem Raj <raj.khem@gmail.com> wrote:
> 
> > > +COMPATIBLE_HOST = ".*-linux"
> > 
> > does this cover linux-gnueabi and such tuples too ?
> 
> Maybe?  Depends on the matching in anchored.  If it does by accident
> then explicitly matching would be better.  What machine has a triple
> like that for testing?

Most arm machines I think? :)

It is unanchored, something which I have to wonder about. I also don't
like the overhead adding COMPATIBLE_* adds but I guess we just have to
live with that :(.

Cheers,

Richard

Re: [OE-core] [PATCH] kernel: set COMPATIBLE_HOST to *-linux

[Ross Burton]

On Sun, 13 Dec 2020 at 23:09, Richard Purdie
<richard.purdie@linuxfoundation.org> wrote:
> > Maybe?  Depends on the matching in anchored.  If it does by accident
> > then explicitly matching would be better.  What machine has a triple
> > like that for testing?
>
> Most arm machines I think? :)

Only old ones, which are dead to me. ;)

> It is unanchored, something which I have to wonder about. I also don't
> like the overhead adding COMPATIBLE_* adds but I guess we just have to
> live with that :(.

So prior art in meta:

meta/recipes-core/systemd/systemd-boot_246.6.bb:COMPATIBLE_HOST =
"(x86_64.*|i.86.*)-linux"
meta/recipes-devtools/dmidecode/dmidecode_3.3.bb:COMPATIBLE_HOST =
"(i.86|x86_64|aarch64|arm|powerpc|powerpc64).*-linux"
meta/recipes-devtools/gcc/gcc-sanitizers.inc:COMPATIBLE_HOST =
'(x86_64|i.86|powerpc|sparc|s390|arm|aarch64).*-linux'
meta/recipes-devtools/syslinux/syslinux_6.04-pre2.bb:COMPATIBLE_HOST =
'(x86_64|i.86).*-(linux|freebsd.*)'
meta/recipes-devtools/valgrind/valgrind_3.16.1.bb:COMPATIBLE_HOST =
'(i.86|x86_64|arm|aarch64|mips|powerpc|powerpc64).*-linux'

35 users in total and none are anchored, they all implicitly let the
expression match -linux-gnueabi.

Ross