[Buildroot] [PATCH 0/4] Fix BR2_HOSTARCH_NEEDS_IA32_* autobuilder issues

[Buildroot] [PATCH 0/4] Fix BR2_HOSTARCH_NEEDS_IA32_* autobuilder issues

[Peter Korsgaard]

Hi,

The following patch series fixes various autobuilder issues from packages
selecting BR2_HOSTARCH_NEEDS_IA32_* without ensuring hostarch is x86(-64),
breaking the build on the gcc112 ppc64le autobuilder.

--
Bye, Peter Korsgaard

[Buildroot] [PATCH 1/4] kvm-unit-test: x86-64 variant only available on x86-64 hosts

[Peter Korsgaard]

We use the host compiler for x86-64 builds, so ensure it is only available
on x86-64 hosts.

Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
---
 package/kvm-unit-tests/Config.in | 1 +
 1 file changed, 1 insertion(+)

diff --git a/package/kvm-unit-tests/Config.in b/package/kvm-unit-tests/Config.in
index d0d9e2ef96..8b7908eee9 100644
--- a/package/kvm-unit-tests/Config.in
+++ b/package/kvm-unit-tests/Config.in
@@ -15,6 +15,7 @@ config BR2_PACKAGE_KVM_UNIT_TESTS
 	# x86-64 we use the host gcc (see .mk file for details)
 	depends on BR2_TOOLCHAIN_GCC_AT_LEAST_4_5 || !BR2_i386
 	depends on BR2_HOST_GCC_AT_LEAST_4_5 || !BR2_x86_64
+	depends on BR2_HOSTARCH = "x86_64" || !BR2_x86_64
 	select BR2_HOSTARCH_NEEDS_IA32_COMPILER if BR2_x86_64=y
 	help
 	  kvm-unit-tests is a project as old as KVM. As its name
-- 
2.11.0

[Buildroot] [PATCH 2/4] ti-gfx: only available on x86(-64) hosts

[Peter Korsgaard]

ti-gfx is provided as a x86 self extracting executable, so it is only
available on x86(-64) hosts.

Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
---
 package/ti-gfx/Config.in | 1 +
 1 file changed, 1 insertion(+)

diff --git a/package/ti-gfx/Config.in b/package/ti-gfx/Config.in
index 6ec2f23117..dc49d822a2 100644
--- a/package/ti-gfx/Config.in
+++ b/package/ti-gfx/Config.in
@@ -1,6 +1,7 @@
 config BR2_PACKAGE_TI_GFX
 	bool "ti-gfx"
 	depends on BR2_LINUX_KERNEL && BR2_TOOLCHAIN_USES_GLIBC && BR2_arm
+	depends on BR2_HOSTARCH = "x86_64" || BR2_HOSTARCH = "x86"
 	select BR2_HOSTARCH_NEEDS_IA32_LIBS
 	select BR2_PACKAGE_HAS_LIBEGL
 	select BR2_PACKAGE_HAS_LIBGLES
-- 
2.11.0

[Buildroot] [PATCH 3/4] luajit: only available on x86(-64) hosts

[Peter Korsgaard]

The -m32 compiler flag is used for 32bit builds and host-luajit has limited
architecture support, so limit to x86(-64) hosts.

Fixes:
http://autobuild.buildroot.net/results/5f5b5edb058efe976c003678e21bcc28a87cc828/

Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
---
 package/luajit/Config.in | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/package/luajit/Config.in b/package/luajit/Config.in
index 1a27c434e1..6b9c6e6ae8 100644
--- a/package/luajit/Config.in
+++ b/package/luajit/Config.in
@@ -1,5 +1,8 @@
 config BR2_PACKAGE_LUAJIT_ARCH_SUPPORTS
 	bool
+	# -m32 flag is used for 32bit builds and host-luajit has
+	# limited architecture support
+	depends on BR2_HOSTARCH = "x86_64" || BR2_HOSTARCH = "x86"
 	default y if BR2_i386 || \
 		(BR2_x86_64 && BR2_HOSTARCH='x86_64') || \
 		BR2_powerpc || BR2_arm || BR2_armeb || \
-- 
2.11.0

[Buildroot] [PATCH 4/4] qt5webengine: only available on x86(-64) hosts

[Peter Korsgaard]

The -m32 compiler flag is used for 32bit builds and the host tools have
limited architecture support, so limit to x86(-64) hosts.

Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
---
 package/qt5/qt5webengine/Config.in | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/package/qt5/qt5webengine/Config.in b/package/qt5/qt5webengine/Config.in
index 6b68d1b2a8..d40b58062e 100644
--- a/package/qt5/qt5webengine/Config.in
+++ b/package/qt5/qt5webengine/Config.in
@@ -3,6 +3,9 @@ config BR2_PACKAGE_QT5WEBENGINE_ARCH_SUPPORTS
 	default y if BR2_aarch64 || BR2_arm || \
 		BR2_i386 || BR2_x86_64 || \
 		BR2_mips || BR2_mips64
+	# -m32 flag is used for 32bit builds and host tools have
+	# limited architecture support
+	depends on BR2_HOSTARCH = "x86_64" || BR2_HOSTARCH = "x86"
 	depends on BR2_PACKAGE_QT5_JSCORE_AVAILABLE # qt5declarative
 	depends on !BR2_BINFMT_FLAT # qt5base-icu
 	depends on BR2_USE_MMU # libglib2, qt5base-dbus
-- 
2.11.0

[Buildroot] [PATCH 3/4] luajit: only available on x86(-64) hosts

[Thomas Petazzoni]

Hello,

On Thu,  2 Nov 2017 11:45:19 +0100, Peter Korsgaard wrote:
> The -m32 compiler flag is used for 32bit builds and host-luajit has limited
> architecture support, so limit to x86(-64) hosts.
> 
> Fixes:
> http://autobuild.buildroot.net/results/5f5b5edb058efe976c003678e21bcc28a87cc828/
> 
> Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
> ---
>  package/luajit/Config.in | 3 +++
>  1 file changed, 3 insertions(+)
> 
> diff --git a/package/luajit/Config.in b/package/luajit/Config.in
> index 1a27c434e1..6b9c6e6ae8 100644
> --- a/package/luajit/Config.in
> +++ b/package/luajit/Config.in
> @@ -1,5 +1,8 @@
>  config BR2_PACKAGE_LUAJIT_ARCH_SUPPORTS
>  	bool
> +	# -m32 flag is used for 32bit builds and host-luajit has
> +	# limited architecture support
> +	depends on BR2_HOSTARCH = "x86_64" || BR2_HOSTARCH = "x86"
>  	default y if BR2_i386 || \
>  		(BR2_x86_64 && BR2_HOSTARCH='x86_64') || \
>  		BR2_powerpc || BR2_arm || BR2_armeb || \

How is this going to prevent other packages from depending on
host-luajit. BR2_PACKAGE_LUAJIT_ARCH_SUPPORTS is all about the target
variant of luajit, not the host variant.

Do we need a BR2_PACKAGE_HOST_LUAJIT_ARCH_SUPPORTS or something like
that ?

For example, if you enable moarvm, host-luajit will be built, but not
luajit.

Don't we have a problem here if your solution is used ?

Thanks!

Thomas
-- 
Thomas Petazzoni, CTO, Free Electrons
Embedded Linux and Kernel engineering
http://free-electrons.com

[Buildroot] [PATCH 0/4] Fix BR2_HOSTARCH_NEEDS_IA32_* autobuilder issues

[Peter Korsgaard]

Hmm, this apparently never made it to the list.

On Thu, Nov 2, 2017 at 11:45 AM, Peter Korsgaard <peter@korsgaard.com> wrote:
> Hi,
>
> The following patch series fixes various autobuilder issues from packages
> selecting BR2_HOSTARCH_NEEDS_IA32_* without ensuring hostarch is x86(-64),
> breaking the build on the gcc112 ppc64le autobuilder.
>
> --
> Bye, Peter Korsgaard



-- 
Bye, Peter Korsgaard

[Buildroot] [PATCH 0/4] Fix BR2_HOSTARCH_NEEDS_IA32_* autobuilder issues

[Baruch Siach]

Hi Peter,

On Thu, Nov 02, 2017 at 03:30:23PM +0100, Peter Korsgaard wrote:
> Hmm, this apparently never made it to the list.

I got the entire set.

It's archived at 
http://lists.busybox.net/pipermail/buildroot/2017-November/205961.html.

> On Thu, Nov 2, 2017 at 11:45 AM, Peter Korsgaard <peter@korsgaard.com> wrote:
> > Hi,
> >
> > The following patch series fixes various autobuilder issues from packages
> > selecting BR2_HOSTARCH_NEEDS_IA32_* without ensuring hostarch is x86(-64),
> > breaking the build on the gcc112 ppc64le autobuilder.
> >
> > --
> > Bye, Peter Korsgaard

-- 
     http://baruch.siach.name/blog/                  ~. .~   Tk Open Systems
=}------------------------------------------------ooO--U--Ooo------------{=
   - baruch at tkos.co.il - tel: +972.52.368.4656, http://www.tkos.co.il -

[Buildroot] [PATCH 0/4] Fix BR2_HOSTARCH_NEEDS_IA32_* autobuilder issues

[Peter Korsgaard]

>>>>> "Baruch" == Baruch Siach <baruch@tkos.co.il> writes:

 > Hi Peter,
 > On Thu, Nov 02, 2017 at 03:30:23PM +0100, Peter Korsgaard wrote:
 >> Hmm, this apparently never made it to the list.

 > I got the entire set.

 > It's archived at 
 > http://lists.busybox.net/pipermail/buildroot/2017-November/205961.html.

Indeed it is, somehow I didn't see it.

-- 
Bye, Peter Korsgaard

[Buildroot] [PATCH 3/4] luajit: only available on x86(-64) hosts

[Peter Korsgaard]

>>>>> "Thomas" == Thomas Petazzoni <thomas.petazzoni@free-electrons.com> writes:

Hi,

 >> +	# -m32 flag is used for 32bit builds and host-luajit has
 >> +	# limited architecture support
 >> +	depends on BR2_HOSTARCH = "x86_64" || BR2_HOSTARCH = "x86"
 >> default y if BR2_i386 || \
 >> (BR2_x86_64 && BR2_HOSTARCH='x86_64') || \
 >> BR2_powerpc || BR2_arm || BR2_armeb || \

 > How is this going to prevent other packages from depending on
 > host-luajit. BR2_PACKAGE_LUAJIT_ARCH_SUPPORTS is all about the target
 > variant of luajit, not the host variant.

Correct.

 > Do we need a BR2_PACKAGE_HOST_LUAJIT_ARCH_SUPPORTS or something like
 > that ?

 > For example, if you enable moarvm, host-luajit will be built, but not
 > luajit.

Ahh, I had missed moarvm. Moarvm seems quite special as no other
packages need host-luajit but no luajit on the target.

Looking at the upstream project, it doesn't seem to explicitly need
host-luajit, just host-lua would do. From Configure.pl:

=item --lua=path/to/lua/executable
Path to a lua executable. (Used during the build when JIT is enabled).

This configure option is even dropped in 2017.10 with the following
information in the changelog:

+ Removal of Configure.pl --lua and --has-dynasm options, compiling with these
included 3rdparty libs is now required

Where the embedded lua copy is a tweaked lua-5.1.5:

https://github.com/MoarVM/dynasm/blob/d84e11714c9460bbb032de6cb1e06c3cb1c31f07/minilua.c

And indeed, moarvm still builds if moarvm.mk is changed to pull in
host-lua and the --lua path is adjusted. Francois, is there a specific
reason you went with host-luajit over host-lua?

So I would prefer to do that change instead of adding
BR2_PACKAGE_HOST_LUAJIT_ARCH_SUPPORTS.

I'll send a patch for moarvm.

-- 
Bye, Peter Korsgaard

[Buildroot] [PATCH 1/4] kvm-unit-test: x86-64 variant only available on x86-64 hosts

[Arnout Vandecappelle]

On 02-11-17 11:45, Peter Korsgaard wrote:
> We use the host compiler for x86-64 builds, so ensure it is only available
> on x86-64 hosts.
> 
> Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
> ---
>  package/kvm-unit-tests/Config.in | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/package/kvm-unit-tests/Config.in b/package/kvm-unit-tests/Config.in
> index d0d9e2ef96..8b7908eee9 100644
> --- a/package/kvm-unit-tests/Config.in
> +++ b/package/kvm-unit-tests/Config.in
> @@ -15,6 +15,7 @@ config BR2_PACKAGE_KVM_UNIT_TESTS
>  	# x86-64 we use the host gcc (see .mk file for details)
>  	depends on BR2_TOOLCHAIN_GCC_AT_LEAST_4_5 || !BR2_i386
>  	depends on BR2_HOST_GCC_AT_LEAST_4_5 || !BR2_x86_64
> +	depends on BR2_HOSTARCH = "x86_64" || !BR2_x86_64

 I don't think this is correct: it will also build correctly for x86_64 if the
host is i386, because it uses -m32 (which is the reason to use the host compiler
instead of the target compiler).

 Regards,
 Arnout

>  	select BR2_HOSTARCH_NEEDS_IA32_COMPILER if BR2_x86_64=y
>  	help
>  	  kvm-unit-tests is a project as old as KVM. As its name
> 

-- 
Arnout Vandecappelle                          arnout at mind be
Senior Embedded Software Architect            +32-16-286500
Essensium/Mind                                http://www.mind.be
G.Geenslaan 9, 3001 Leuven, Belgium           BE 872 984 063 RPR Leuven
LinkedIn profile: http://www.linkedin.com/in/arnoutvandecappelle
GPG fingerprint:  7493 020B C7E3 8618 8DEC 222C 82EB F404 F9AC 0DDF

[Buildroot] [PATCH 3/4] luajit: only available on x86(-64) hosts

[Arnout Vandecappelle]

On 02-11-17 11:45, Peter Korsgaard wrote:
> The -m32 compiler flag is used for 32bit builds and host-luajit has limited
> architecture support, so limit to x86(-64) hosts.

 I really don't see why this is needed. What we need is that the host and target
bitness is the same. On x86_64, we can always do this by compiling with -m32
(which is what we do in the .mk file). But I see no reason at all why building
for a 32-bit target would not be possible on ARM, MIPS or PowerPC (32-bit).

 That said, we can't easily test that and nobody cares anyway, so I've committed
with an updated commit message.

 Regards,
 Arnout

> 
> Fixes:
> http://autobuild.buildroot.net/results/5f5b5edb058efe976c003678e21bcc28a87cc828/
> 
> Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
> ---
>  package/luajit/Config.in | 3 +++
>  1 file changed, 3 insertions(+)
> 
> diff --git a/package/luajit/Config.in b/package/luajit/Config.in
> index 1a27c434e1..6b9c6e6ae8 100644
> --- a/package/luajit/Config.in
> +++ b/package/luajit/Config.in
> @@ -1,5 +1,8 @@
>  config BR2_PACKAGE_LUAJIT_ARCH_SUPPORTS
>  	bool
> +	# -m32 flag is used for 32bit builds and host-luajit has
> +	# limited architecture support
> +	depends on BR2_HOSTARCH = "x86_64" || BR2_HOSTARCH = "x86"
>  	default y if BR2_i386 || \
>  		(BR2_x86_64 && BR2_HOSTARCH='x86_64') || \
>  		BR2_powerpc || BR2_arm || BR2_armeb || \
> 

-- 
Arnout Vandecappelle                          arnout at mind be
Senior Embedded Software Architect            +32-16-286500
Essensium/Mind                                http://www.mind.be
G.Geenslaan 9, 3001 Leuven, Belgium           BE 872 984 063 RPR Leuven
LinkedIn profile: http://www.linkedin.com/in/arnoutvandecappelle
GPG fingerprint:  7493 020B C7E3 8618 8DEC 222C 82EB F404 F9AC 0DDF

[Buildroot] [PATCH 2/4] ti-gfx: only available on x86(-64) hosts

[Arnout Vandecappelle]

On 02-11-17 11:45, Peter Korsgaard wrote:
> ti-gfx is provided as a x86 self extracting executable, so it is only
> available on x86(-64) hosts.
> 
> Signed-off-by: Peter Korsgaard <peter@korsgaard.com>


 Applied to master, thanks.

 Regards,
 Arnout

> ---
>  package/ti-gfx/Config.in | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/package/ti-gfx/Config.in b/package/ti-gfx/Config.in
> index 6ec2f23117..dc49d822a2 100644
> --- a/package/ti-gfx/Config.in
> +++ b/package/ti-gfx/Config.in
> @@ -1,6 +1,7 @@
>  config BR2_PACKAGE_TI_GFX
>  	bool "ti-gfx"
>  	depends on BR2_LINUX_KERNEL && BR2_TOOLCHAIN_USES_GLIBC && BR2_arm
> +	depends on BR2_HOSTARCH = "x86_64" || BR2_HOSTARCH = "x86"
>  	select BR2_HOSTARCH_NEEDS_IA32_LIBS
>  	select BR2_PACKAGE_HAS_LIBEGL
>  	select BR2_PACKAGE_HAS_LIBGLES
> 

-- 
Arnout Vandecappelle                          arnout at mind be
Senior Embedded Software Architect            +32-16-286500
Essensium/Mind                                http://www.mind.be
G.Geenslaan 9, 3001 Leuven, Belgium           BE 872 984 063 RPR Leuven
LinkedIn profile: http://www.linkedin.com/in/arnoutvandecappelle
GPG fingerprint:  7493 020B C7E3 8618 8DEC 222C 82EB F404 F9AC 0DDF

[Buildroot] [PATCH 4/4] qt5webengine: only available on x86(-64) hosts

[Arnout Vandecappelle]

On 02-11-17 11:45, Peter Korsgaard wrote:
> The -m32 compiler flag is used for 32bit builds and the host tools have
> limited architecture support, so limit to x86(-64) hosts.

 Same story here: there are combinations that should work, e.g. building for
x86_64 target on an aarch64 host. However, it's equally likely to break, and the
Config.in expressions would be quite complicated to allow it. So I've applied
after extending the commit message a little.

 Regards,
 Arnout

> 
> Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
> ---
>  package/qt5/qt5webengine/Config.in | 3 +++
>  1 file changed, 3 insertions(+)
> 
> diff --git a/package/qt5/qt5webengine/Config.in b/package/qt5/qt5webengine/Config.in
> index 6b68d1b2a8..d40b58062e 100644
> --- a/package/qt5/qt5webengine/Config.in
> +++ b/package/qt5/qt5webengine/Config.in
> @@ -3,6 +3,9 @@ config BR2_PACKAGE_QT5WEBENGINE_ARCH_SUPPORTS
>  	default y if BR2_aarch64 || BR2_arm || \
>  		BR2_i386 || BR2_x86_64 || \
>  		BR2_mips || BR2_mips64
> +	# -m32 flag is used for 32bit builds and host tools have
> +	# limited architecture support
> +	depends on BR2_HOSTARCH = "x86_64" || BR2_HOSTARCH = "x86"
>  	depends on BR2_PACKAGE_QT5_JSCORE_AVAILABLE # qt5declarative
>  	depends on !BR2_BINFMT_FLAT # qt5base-icu
>  	depends on BR2_USE_MMU # libglib2, qt5base-dbus
> 

-- 
Arnout Vandecappelle                          arnout at mind be
Senior Embedded Software Architect            +32-16-286500
Essensium/Mind                                http://www.mind.be
G.Geenslaan 9, 3001 Leuven, Belgium           BE 872 984 063 RPR Leuven
LinkedIn profile: http://www.linkedin.com/in/arnoutvandecappelle
GPG fingerprint:  7493 020B C7E3 8618 8DEC 222C 82EB F404 F9AC 0DDF

[Buildroot] [PATCH 3/4] luajit: only available on x86(-64) hosts

[Peter Korsgaard]

>>>>> "Arnout" == Arnout Vandecappelle <arnout@mind.be> writes:

 > On 02-11-17 11:45, Peter Korsgaard wrote:
 >> The -m32 compiler flag is used for 32bit builds and host-luajit has limited
 >> architecture support, so limit to x86(-64) hosts.

 >  I really don't see why this is needed. What we need is that the host and target
 > bitness is the same. On x86_64, we can always do this by compiling with -m32
 > (which is what we do in the .mk file). But I see no reason at all why building
 > for a 32-bit target would not be possible on ARM, MIPS or PowerPC (32-bit).

Correct, I opted for the simplest possible fix as building on !x86
really isn't a very realistic use case.

 >  That said, we can't easily test that and nobody cares anyway, so I've committed
 > with an updated commit message.

Great, thanks!

-- 
Bye, Peter Korsgaard

[Buildroot] [PATCH 1/4] kvm-unit-test: x86-64 variant only available on x86-64 hosts

[Peter Korsgaard]

>>>>> "Arnout" == Arnout Vandecappelle <arnout@mind.be> writes:

 > On 02-11-17 11:45, Peter Korsgaard wrote:
 >> We use the host compiler for x86-64 builds, so ensure it is only available
 >> on x86-64 hosts.
 >> 
 >> Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
 >> ---
 >> package/kvm-unit-tests/Config.in | 1 +
 >> 1 file changed, 1 insertion(+)
 >> 
 >> diff --git a/package/kvm-unit-tests/Config.in b/package/kvm-unit-tests/Config.in
 >> index d0d9e2ef96..8b7908eee9 100644
 >> --- a/package/kvm-unit-tests/Config.in
 >> +++ b/package/kvm-unit-tests/Config.in
 >> @@ -15,6 +15,7 @@ config BR2_PACKAGE_KVM_UNIT_TESTS
 >> # x86-64 we use the host gcc (see .mk file for details)
 >> depends on BR2_TOOLCHAIN_GCC_AT_LEAST_4_5 || !BR2_i386
 >> depends on BR2_HOST_GCC_AT_LEAST_4_5 || !BR2_x86_64
 >> +	depends on BR2_HOSTARCH = "x86_64" || !BR2_x86_64

 >  I don't think this is correct: it will also build correctly for x86_64 if the
 > host is i386, because it uses -m32 (which is the reason to use the host compiler
 > instead of the target compiler).

But some parts of kvm-unit-tests presumably are 64bit when building for
x86-64, which wouldn't work when using the host compiler on a x86
system, right? I don't think x86 system compilers are 32/64bit multilib.

But I haven't actually used a x86 distribution in the last 10 years or
so, so perhaps I'm missing something?

-- 
Bye, Peter Korsgaard

[Buildroot] [PATCH 1/4] kvm-unit-test: x86-64 variant only available on x86-64 hosts

[Arnout Vandecappelle]

On 05-11-17 20:27, Peter Korsgaard wrote:
>>>>>> "Arnout" == Arnout Vandecappelle <arnout@mind.be> writes:
> 
>  > On 02-11-17 11:45, Peter Korsgaard wrote:
>  >> We use the host compiler for x86-64 builds, so ensure it is only available
>  >> on x86-64 hosts.
>  >> 
>  >> Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
>  >> ---
>  >> package/kvm-unit-tests/Config.in | 1 +
>  >> 1 file changed, 1 insertion(+)
>  >> 
>  >> diff --git a/package/kvm-unit-tests/Config.in b/package/kvm-unit-tests/Config.in
>  >> index d0d9e2ef96..8b7908eee9 100644
>  >> --- a/package/kvm-unit-tests/Config.in
>  >> +++ b/package/kvm-unit-tests/Config.in
>  >> @@ -15,6 +15,7 @@ config BR2_PACKAGE_KVM_UNIT_TESTS
>  >> # x86-64 we use the host gcc (see .mk file for details)
>  >> depends on BR2_TOOLCHAIN_GCC_AT_LEAST_4_5 || !BR2_i386
>  >> depends on BR2_HOST_GCC_AT_LEAST_4_5 || !BR2_x86_64
>  >> +	depends on BR2_HOSTARCH = "x86_64" || !BR2_x86_64
> 
>  >  I don't think this is correct: it will also build correctly for x86_64 if the
>  > host is i386, because it uses -m32 (which is the reason to use the host compiler
>  > instead of the target compiler).
> 
> But some parts of kvm-unit-tests presumably are 64bit when building for
> x86-64, 

 OK, I assumed everything was built with -m32. If some parts are indeed still
64-bit then it won't work in i386 host.

> which wouldn't work when using the host compiler on a x86
> system, right? I don't think x86 system compilers are 32/64bit multilib

 Actually, on Debian i386 it is possible to install 64-bit multilib I think. But
obviously we can't count on that.

 BTW, what happens on x32, does anybody know that?

> But I haven't actually used a x86 distribution in the last 10 years or
> so, so perhaps I'm missing something?

 Yeah, it's pretty unlikely as well I guess.

 I'll apply this patch as well then.

 Regards,
 Arnout

-- 
Arnout Vandecappelle                          arnout at mind be
Senior Embedded Software Architect            +32-16-286500
Essensium/Mind                                http://www.mind.be
G.Geenslaan 9, 3001 Leuven, Belgium           BE 872 984 063 RPR Leuven
LinkedIn profile: http://www.linkedin.com/in/arnoutvandecappelle
GPG fingerprint:  7493 020B C7E3 8618 8DEC 222C 82EB F404 F9AC 0DDF

[Buildroot] [PATCH 1/4] kvm-unit-test: x86-64 variant only available on x86-64 hosts

[Peter Korsgaard]

>>>>> "Arnout" == Arnout Vandecappelle <arnout@mind.be> writes:

Hi,

 >> But some parts of kvm-unit-tests presumably are 64bit when building for
 >> x86-64, 

 >  OK, I assumed everything was built with -m32. If some parts are indeed still
 > 64-bit then it won't work in i386 host.

Exactly.

 >> which wouldn't work when using the host compiler on a x86
 >> system, right? I don't think x86 system compilers are 32/64bit multilib

 >  Actually, on Debian i386 it is possible to install 64-bit multilib I think. But
 > obviously we can't count on that.

 >  BTW, what happens on x32, does anybody know that?

Is x32 really used by any mainstream distro? I would guess it is as rare
as building on a !x86 machine.


 >> But I haven't actually used a x86 distribution in the last 10 years or
 >> so, so perhaps I'm missing something?

 >  Yeah, it's pretty unlikely as well I guess.

 >  I'll apply this patch as well then.

Great, thanks!

-- 
Bye, Peter Korsgaard

[Buildroot] [PATCH 2/4] ti-gfx: only available on x86(-64) hosts

[Peter Korsgaard]

>>>>> "Peter" == Peter Korsgaard <peter@korsgaard.com> writes:

 > ti-gfx is provided as a x86 self extracting executable, so it is only
 > available on x86(-64) hosts.

 > Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

Committed to 2017.02.x and 2017.08.x, thanks.

-- 
Bye, Peter Korsgaard

[Buildroot] [PATCH 3/4] luajit: only available on x86(-64) hosts

[Peter Korsgaard]

>>>>> "Peter" == Peter Korsgaard <peter@korsgaard.com> writes:

 > The -m32 compiler flag is used for 32bit builds and host-luajit has limited
 > architecture support, so limit to x86(-64) hosts.

 > Fixes:
 > http://autobuild.buildroot.net/results/5f5b5edb058efe976c003678e21bcc28a87cc828/

 > Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

Committed to 2017.02.x and 2017.08.x, thanks.

-- 
Bye, Peter Korsgaard

[Buildroot] [PATCH 1/4] kvm-unit-test: x86-64 variant only available on x86-64 hosts

[Peter Korsgaard]

>>>>> "Peter" == Peter Korsgaard <peter@korsgaard.com> writes:

 > We use the host compiler for x86-64 builds, so ensure it is only available
 > on x86-64 hosts.

 > Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

Committed to 2017.08.x, thanks.

-- 
Bye, Peter Korsgaard