[Buildroot] [PATCH 1/2] support/testing: update the ARM toolchain used for the systemd init tests

[Buildroot] [PATCH 1/2] support/testing: update the ARM toolchain used for the systemd init tests

[Romain Naour]

As reported on the mailing list, there is a build issue with systemd 245
when using gcc < 5.0:

http://lists.busybox.net/pipermail/buildroot/2020-April/278931.html

Update to the Bootling toolchain 2017.05 that meet the minimum requered
version of toolchain componant:
GCC           5.4.0
linux-headers 3.10.105

See:
https://toolchains.bootlin.com/downloads/releases/toolchains/armv5-eabi/readmes/armv5-eabi--glibc--stable-2017.05-toolchains-1-1.txt

Tested:
https://gitlab.com/kubu93/buildroot/pipelines/132890602

Fixes:
https://gitlab.com/buildroot.org/buildroot/-/jobs/498144403
https://gitlab.com/buildroot.org/buildroot/-/jobs/498144405
https://gitlab.com/buildroot.org/buildroot/-/jobs/498144406
https://gitlab.com/buildroot.org/buildroot/-/jobs/498144408
https://gitlab.com/buildroot.org/buildroot/-/jobs/498144410
https://gitlab.com/buildroot.org/buildroot/-/jobs/498144412

Signed-off-by: Romain Naour <romain.naour@gmail.com>
---
 support/testing/tests/init/test_systemd.py | 13 +++++++++++++
 1 file changed, 13 insertions(+)

diff --git a/support/testing/tests/init/test_systemd.py b/support/testing/tests/init/test_systemd.py
index 0ac18cb623..10634cbb1f 100644
--- a/support/testing/tests/init/test_systemd.py
+++ b/support/testing/tests/init/test_systemd.py
@@ -2,11 +2,24 @@ import infra.basetest
 from tests.init.base import InitSystemBase as InitSystemBase
 
 
+# External Toolchain fragment imported from:
+# https://toolchains.bootlin.com/downloads/releases/toolchains/armv5-eabi/fragments/armv5-eabi--glibc--stable-2017.05-toolchains-1-1.frag
 class InitSystemSystemdBase(InitSystemBase):
     config = \
         """
         BR2_arm=y
+        BR2_arm926t=y
+        BR2_ARM_CPU_ARMV5=y
+        BR2_ARM_EABI=y
+        BR2_ARM_SOFT_FLOAT=y
         BR2_TOOLCHAIN_EXTERNAL=y
+        BR2_TOOLCHAIN_EXTERNAL_CUSTOM=y
+        BR2_TOOLCHAIN_EXTERNAL_DOWNLOAD=y
+        BR2_TOOLCHAIN_EXTERNAL_URL="http://toolchains.bootlin.com/downloads/releases/toolchains/armv5-eabi--glibc--stable-2017.05-toolchains-1-1.tar.bz2"
+        BR2_TOOLCHAIN_EXTERNAL_GCC_5=y
+        BR2_TOOLCHAIN_EXTERNAL_HEADERS_3_10=y
+        BR2_TOOLCHAIN_EXTERNAL_HAS_SSP=y
+        BR2_TOOLCHAIN_EXTERNAL_CUSTOM_GLIBC=y
         BR2_INIT_SYSTEMD=y
         BR2_TARGET_GENERIC_GETTY_PORT="ttyAMA0"
         BR2_LINUX_KERNEL=y
-- 
2.25.1

[Buildroot] [PATCH 2/2] package/systemd: add gcc >= 5.x dependency

[Romain Naour]

As reported on the mailing list, there is a build issue with systemd 245
when using gcc < 5.0:

http://lists.busybox.net/pipermail/buildroot/2020-April/278931.html

Build issue:
../src/shared/gpt.c:7:9: error: initializer element is not constant
         { GPT_ROOT_X86,              "root-x86"              },

When testing with a toolchain using gcc 5.4.0 and the build is ok.
http://toolchains.bootlin.com/downloads/releases/toolchains/armv5-eabi--glibc--stable-2017.05-toolchains-1-1.tar.bz2

While searching for "error: initializer element is not constant" message, we
can notice a note about gcc 5 change about "Initializing statics with compound
literals":

https://gcc.gnu.org/gcc-5/porting_to.html

Add a dependency on gcc 5 to avoid using to old compiler.

There is the same issue with host-systemd with host gcc 4.9
(tested with Debian Jessie). So, add a dependency on host gcc >= 5.x.

Fixes:
http://autobuild.buildroot.org/results/520/520dab2253f4cbe408a8177a6587dcb38c6ba215
http://autobuild.buildroot.org/results/e0e/e0e0512de822864d670b5d176798a24ab09eed2d
http://autobuild.buildroot.org/results/f56/f5660b2711627fcee4086e096e4ec4d9ba190ab6

Signed-off-by: Romain Naour <romain.naour@gmail.com>
Cc: Adam Duskett <aduskett@gmail.com>
Cc: Maxime Hadjinlian <maxime.hadjinlian@gmail.com>
Cc: Yann E. MORIN <yann.morin.1998@free.fr>
---
 package/systemd/Config.in      | 2 ++
 package/systemd/Config.in.host | 1 +
 system/Config.in               | 6 ++++--
 3 files changed, 7 insertions(+), 2 deletions(-)

diff --git a/package/systemd/Config.in b/package/systemd/Config.in
index bd638477a0..4afe694413 100644
--- a/package/systemd/Config.in
+++ b/package/systemd/Config.in
@@ -19,6 +19,8 @@ menuconfig BR2_PACKAGE_SYSTEMD
 	depends on !BR2_STATIC_LIBS # kmod
 	depends on BR2_TOOLCHAIN_HAS_THREADS # dbus
 	depends on BR2_USE_MMU # dbus
+	depends on BR2_TOOLCHAIN_GCC_AT_LEAST_5
+	depends on BR2_HOST_GCC_AT_LEAST_5 # host-systemd
 	select BR2_PACKAGE_HAS_UDEV
 	select BR2_PACKAGE_DBUS # runtime dependency only
 	select BR2_PACKAGE_LIBCAP
diff --git a/package/systemd/Config.in.host b/package/systemd/Config.in.host
index 219f24239e..8e00bcf5c1 100644
--- a/package/systemd/Config.in.host
+++ b/package/systemd/Config.in.host
@@ -1,3 +1,4 @@
 # Select this if you need host systemd tools (e.g. systemctl)
 config BR2_PACKAGE_HOST_SYSTEMD
 	bool
+	depends on BR2_HOST_GCC_AT_LEAST_5
diff --git a/system/Config.in b/system/Config.in
index c8c5be40e0..25ead1489e 100644
--- a/system/Config.in
+++ b/system/Config.in
@@ -121,15 +121,17 @@ config BR2_INIT_SYSTEMD
 	depends on BR2_USE_MMU
 	depends on !BR2_STATIC_LIBS
 	depends on BR2_TOOLCHAIN_HEADERS_AT_LEAST_3_10
+	depends on BR2_TOOLCHAIN_GCC_AT_LEAST_5
 	select BR2_ROOTFS_MERGED_USR
 	select BR2_PACKAGE_SYSTEMD
 
-comment "systemd needs a glibc toolchain w/ SSP, headers >= 3.10"
+comment "systemd needs a glibc toolchain w/ SSP, headers >= 3.10, gcc >= 5"
 	depends on BR2_PACKAGE_SYSTEMD_ARCH_SUPPORTS
 	depends on BR2_USE_MMU
 	depends on !BR2_TOOLCHAIN_USES_GLIBC || \
 		!BR2_TOOLCHAIN_HAS_SSP || \
-		!BR2_TOOLCHAIN_HEADERS_AT_LEAST_3_10
+		!BR2_TOOLCHAIN_HEADERS_AT_LEAST_3_10 || \
+		!BR2_TOOLCHAIN_GCC_AT_LEAST_5
 
 config BR2_INIT_NONE
 	bool "None"
-- 
2.25.1

[Buildroot] [PATCH 1/2] support/testing: update the ARM toolchain used for the systemd init tests

[Yann E. MORIN]

Romain, All,

On 2020-04-05 00:02 +0200, Romain Naour spake thusly:
> As reported on the mailing list, there is a build issue with systemd 245
> when using gcc < 5.0:
> 
> http://lists.busybox.net/pipermail/buildroot/2020-April/278931.html
> 
> Update to the Bootling toolchain 2017.05 that meet the minimum requered
> version of toolchain componant:
> GCC           5.4.0
> linux-headers 3.10.105
> 
> See:
> https://toolchains.bootlin.com/downloads/releases/toolchains/armv5-eabi/readmes/armv5-eabi--glibc--stable-2017.05-toolchains-1-1.txt
> 
> Tested:
> https://gitlab.com/kubu93/buildroot/pipelines/132890602
> 
> Fixes:
> https://gitlab.com/buildroot.org/buildroot/-/jobs/498144403
> https://gitlab.com/buildroot.org/buildroot/-/jobs/498144405
> https://gitlab.com/buildroot.org/buildroot/-/jobs/498144406
> https://gitlab.com/buildroot.org/buildroot/-/jobs/498144408
> https://gitlab.com/buildroot.org/buildroot/-/jobs/498144410
> https://gitlab.com/buildroot.org/buildroot/-/jobs/498144412
> 
> Signed-off-by: Romain Naour <romain.naour@gmail.com>
> ---
>  support/testing/tests/init/test_systemd.py | 13 +++++++++++++
>  1 file changed, 13 insertions(+)
> 
> diff --git a/support/testing/tests/init/test_systemd.py b/support/testing/tests/init/test_systemd.py
> index 0ac18cb623..10634cbb1f 100644
> --- a/support/testing/tests/init/test_systemd.py
> +++ b/support/testing/tests/init/test_systemd.py
> @@ -2,11 +2,24 @@ import infra.basetest
>  from tests.init.base import InitSystemBase as InitSystemBase
>  
>  
> +# External Toolchain fragment imported from:
> +# https://toolchains.bootlin.com/downloads/releases/toolchains/armv5-eabi/fragments/armv5-eabi--glibc--stable-2017.05-toolchains-1-1.frag
>  class InitSystemSystemdBase(InitSystemBase):
>      config = \
>          """
>          BR2_arm=y
> +        BR2_arm926t=y
> +        BR2_ARM_CPU_ARMV5=y
> +        BR2_ARM_EABI=y
> +        BR2_ARM_SOFT_FLOAT=y

Those last four options are useless, as they are the default for
BR2_arm.

>          BR2_TOOLCHAIN_EXTERNAL=y
> +        BR2_TOOLCHAIN_EXTERNAL_CUSTOM=y
> +        BR2_TOOLCHAIN_EXTERNAL_DOWNLOAD=y
> +        BR2_TOOLCHAIN_EXTERNAL_URL="http://toolchains.bootlin.com/downloads/releases/toolchains/armv5-eabi--glibc--stable-2017.05-toolchains-1-1.tar.bz2"

I'd rather we use one of the known pre-built toolchains that is
available from the menuconfig. Unfortiunately, for armv5, we have no
better than the codesourcery one. But for armv7, we have a darn-recent
toolchain from ARM.

So I think we just need to add BR2_cortex_a7 to this defconfig:

    diff --git a/support/testing/tests/init/test_systemd.py b/support/testing/tests/init/test_systemd.py
    index 0ac18cb623..d679081768 100644
    --- a/support/testing/tests/init/test_systemd.py
    +++ b/support/testing/tests/init/test_systemd.py
    @@ -6,6 +6,7 @@ class InitSystemSystemdBase(InitSystemBase):
         config = \
             """
             BR2_arm=y
    +        BR2_cortex_a7=y
             BR2_TOOLCHAIN_EXTERNAL=y
             BR2_INIT_SYSTEMD=y
             BR2_TARGET_GENERIC_GETTY_PORT="ttyAMA0"

Note that I am not saying the bootlin toolchains are bad. I am instead
saying that we should test with what is readily available in menuconfig,
so that we know users can have a working starting point.

Runtime tests are not (IMHO) about testing corner cases, but to ensure
that the packages do build and run as expected in standard conditions.

Bootlin toolchains are not in the menuconfig, while the ARM toolchain
is, and that is what users will use 99.99% of the time. Hence we should
test with that.

Regards,
Yann E. MORIN.

> +        BR2_TOOLCHAIN_EXTERNAL_GCC_5=y
> +        BR2_TOOLCHAIN_EXTERNAL_HEADERS_3_10=y
> +        BR2_TOOLCHAIN_EXTERNAL_HAS_SSP=y
> +        BR2_TOOLCHAIN_EXTERNAL_CUSTOM_GLIBC=y
>          BR2_INIT_SYSTEMD=y
>          BR2_TARGET_GENERIC_GETTY_PORT="ttyAMA0"
>          BR2_LINUX_KERNEL=y
> -- 
> 2.25.1
> 
> _______________________________________________
> buildroot mailing list
> buildroot at busybox.net
> http://lists.busybox.net/mailman/listinfo/buildroot

-- 
.-----------------.--------------------.------------------.--------------------.
|  Yann E. MORIN  | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: |
| +33 662 376 056 | Software  Designer | \ / CAMPAIGN     |  ___               |
| +33 561 099 427 `------------.-------:  X  AGAINST      |  \e/  There is no  |
| http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL    |   v   conspiracy.  |
'------------------------------^-------^------------------^--------------------'

[Buildroot] [PATCH 1/2] support/testing: update the ARM toolchain used for the systemd init tests

[Yann E. MORIN]

Romain, All,

On 2020-04-05 10:44 +0200, Yann E. MORIN spake thusly:
> On 2020-04-05 00:02 +0200, Romain Naour spake thusly:
> > As reported on the mailing list, there is a build issue with systemd 245
> > when using gcc < 5.0:
> > http://lists.busybox.net/pipermail/buildroot/2020-April/278931.html
[--SNIP--]
> >          BR2_TOOLCHAIN_EXTERNAL=y
> > +        BR2_TOOLCHAIN_EXTERNAL_CUSTOM=y
> > +        BR2_TOOLCHAIN_EXTERNAL_DOWNLOAD=y
> > +        BR2_TOOLCHAIN_EXTERNAL_URL="http://toolchains.bootlin.com/downloads/releases/toolchains/armv5-eabi--glibc--stable-2017.05-toolchains-1-1.tar.bz2"
> I'd rather we use one of the known pre-built toolchains that is
> available from the menuconfig. Unfortiunately, for armv5, we have no
> better than the codesourcery one. But for armv7, we have a darn-recent
> toolchain from ARM.

Of course, I mistakenly replied to your v1 of your patchset, but this is
also valid for your v2...

Regards,
Yann E. MORIN.

> So I think we just need to add BR2_cortex_a7 to this defconfig:
> 
>     diff --git a/support/testing/tests/init/test_systemd.py b/support/testing/tests/init/test_systemd.py
>     index 0ac18cb623..d679081768 100644
>     --- a/support/testing/tests/init/test_systemd.py
>     +++ b/support/testing/tests/init/test_systemd.py
>     @@ -6,6 +6,7 @@ class InitSystemSystemdBase(InitSystemBase):
>          config = \
>              """
>              BR2_arm=y
>     +        BR2_cortex_a7=y
>              BR2_TOOLCHAIN_EXTERNAL=y
>              BR2_INIT_SYSTEMD=y
>              BR2_TARGET_GENERIC_GETTY_PORT="ttyAMA0"
> 
> Note that I am not saying the bootlin toolchains are bad. I am instead
> saying that we should test with what is readily available in menuconfig,
> so that we know users can have a working starting point.
> 
> Runtime tests are not (IMHO) about testing corner cases, but to ensure
> that the packages do build and run as expected in standard conditions.
> 
> Bootlin toolchains are not in the menuconfig, while the ARM toolchain
> is, and that is what users will use 99.99% of the time. Hence we should
> test with that.
> 
> Regards,
> Yann E. MORIN.
> 
> > +        BR2_TOOLCHAIN_EXTERNAL_GCC_5=y
> > +        BR2_TOOLCHAIN_EXTERNAL_HEADERS_3_10=y
> > +        BR2_TOOLCHAIN_EXTERNAL_HAS_SSP=y
> > +        BR2_TOOLCHAIN_EXTERNAL_CUSTOM_GLIBC=y
> >          BR2_INIT_SYSTEMD=y
> >          BR2_TARGET_GENERIC_GETTY_PORT="ttyAMA0"
> >          BR2_LINUX_KERNEL=y
> > -- 
> > 2.25.1
> > 
> > _______________________________________________
> > buildroot mailing list
> > buildroot at busybox.net
> > http://lists.busybox.net/mailman/listinfo/buildroot
> 
> -- 
> .-----------------.--------------------.------------------.--------------------.
> |  Yann E. MORIN  | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: |
> | +33 662 376 056 | Software  Designer | \ / CAMPAIGN     |  ___               |
> | +33 561 099 427 `------------.-------:  X  AGAINST      |  \e/  There is no  |
> | http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL    |   v   conspiracy.  |
> '------------------------------^-------^------------------^--------------------'
> _______________________________________________
> buildroot mailing list
> buildroot at busybox.net
> http://lists.busybox.net/mailman/listinfo/buildroot

-- 
.-----------------.--------------------.------------------.--------------------.
|  Yann E. MORIN  | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: |
| +33 662 376 056 | Software  Designer | \ / CAMPAIGN     |  ___               |
| +33 561 099 427 `------------.-------:  X  AGAINST      |  \e/  There is no  |
| http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL    |   v   conspiracy.  |
'------------------------------^-------^------------------^--------------------'

[Buildroot] [PATCH 1/2] support/testing: update the ARM toolchain used for the systemd init tests

[Romain Naour]

Yann, All,

Le 05/04/2020 ? 11:09, Yann E. MORIN a ?crit?:
> Romain, All,
> 
> On 2020-04-05 10:44 +0200, Yann E. MORIN spake thusly:
>> On 2020-04-05 00:02 +0200, Romain Naour spake thusly:
>>> As reported on the mailing list, there is a build issue with systemd 245
>>> when using gcc < 5.0:
>>> http://lists.busybox.net/pipermail/buildroot/2020-April/278931.html
> [--SNIP--]
>>>          BR2_TOOLCHAIN_EXTERNAL=y
>>> +        BR2_TOOLCHAIN_EXTERNAL_CUSTOM=y
>>> +        BR2_TOOLCHAIN_EXTERNAL_DOWNLOAD=y
>>> +        BR2_TOOLCHAIN_EXTERNAL_URL="http://toolchains.bootlin.com/downloads/releases/toolchains/armv5-eabi--glibc--stable-2017.05-toolchains-1-1.tar.bz2"
>> I'd rather we use one of the known pre-built toolchains that is
>> available from the menuconfig. Unfortiunately, for armv5, we have no
>> better than the codesourcery one. But for armv7, we have a darn-recent
>> toolchain from ARM.

Indeed, If we are agree to use armv7 instead of armv5 for the testsuite we
should use the ARM arm toolchain.

> 
> Of course, I mistakenly replied to your v1 of your patchset, but this is
> also valid for your v2...
> 
> Regards,
> Yann E. MORIN.
> 
>> So I think we just need to add BR2_cortex_a7 to this defconfig:
>>
>>     diff --git a/support/testing/tests/init/test_systemd.py b/support/testing/tests/init/test_systemd.py
>>     index 0ac18cb623..d679081768 100644
>>     --- a/support/testing/tests/init/test_systemd.py
>>     +++ b/support/testing/tests/init/test_systemd.py
>>     @@ -6,6 +6,7 @@ class InitSystemSystemdBase(InitSystemBase):
>>          config = \
>>              """
>>              BR2_arm=y
>>     +        BR2_cortex_a7=y
>>              BR2_TOOLCHAIN_EXTERNAL=y
>>              BR2_INIT_SYSTEMD=y
>>              BR2_TARGET_GENERIC_GETTY_PORT="ttyAMA0"
>>
>> Note that I am not saying the bootlin toolchains are bad. I am instead
>> saying that we should test with what is readily available in menuconfig,
>> so that we know users can have a working starting point.

I was suggesting to use the upcoming Bootlin ARM toolchain 2020.02-2, but I'm
fine with ARM arm toolchain.

>>
>> Runtime tests are not (IMHO) about testing corner cases, but to ensure
>> that the packages do build and run as expected in standard conditions.

That's why I recently updated the basic toolchain used by the testsuite :)

>>
>> Bootlin toolchains are not in the menuconfig, while the ARM toolchain
>> is, and that is what users will use 99.99% of the time. Hence we should
>> test with that.

So it also means that 99,99% of users using the ARM toolchain will use a glibc
based toolchain. But we are testing with an uClibc-ng toolchain most of the
time. Maybe we should do more tests with glibc and musl toolchain also with 2
gcc version (the default version and the latest one) ?

See the discussion:
http://lists.busybox.net/pipermail/buildroot/2020-April/278786.html

Best regards,
Romain

>>
>> Regards,
>> Yann E. MORIN.
>>
>>> +        BR2_TOOLCHAIN_EXTERNAL_GCC_5=y
>>> +        BR2_TOOLCHAIN_EXTERNAL_HEADERS_3_10=y
>>> +        BR2_TOOLCHAIN_EXTERNAL_HAS_SSP=y
>>> +        BR2_TOOLCHAIN_EXTERNAL_CUSTOM_GLIBC=y
>>>          BR2_INIT_SYSTEMD=y
>>>          BR2_TARGET_GENERIC_GETTY_PORT="ttyAMA0"
>>>          BR2_LINUX_KERNEL=y
>>> -- 
>>> 2.25.1
>>>
>>> _______________________________________________
>>> buildroot mailing list
>>> buildroot at busybox.net
>>> http://lists.busybox.net/mailman/listinfo/buildroot
>>
>> -- 
>> .-----------------.--------------------.------------------.--------------------.
>> |  Yann E. MORIN  | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: |
>> | +33 662 376 056 | Software  Designer | \ / CAMPAIGN     |  ___               |
>> | +33 561 099 427 `------------.-------:  X  AGAINST      |  \e/  There is no  |
>> | http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL    |   v   conspiracy.  |
>> '------------------------------^-------^------------------^--------------------'
>> _______________________________________________
>> buildroot mailing list
>> buildroot at busybox.net
>> http://lists.busybox.net/mailman/listinfo/buildroot
> 

[Buildroot] [PATCH 1/2] support/testing: update the ARM toolchain used for the systemd init tests

[Yann E. MORIN]

romain, All,

On 2020-04-05 11:35 +0200, Romain Naour spake thusly:
> Le 05/04/2020 ? 11:09, Yann E. MORIN a ?crit?:
> > Romain, All,
> > 
> > On 2020-04-05 10:44 +0200, Yann E. MORIN spake thusly:
> >> On 2020-04-05 00:02 +0200, Romain Naour spake thusly:
> >>> As reported on the mailing list, there is a build issue with systemd 245
> >>> when using gcc < 5.0:
> >>> http://lists.busybox.net/pipermail/buildroot/2020-April/278931.html
> > [--SNIP--]
> >>>          BR2_TOOLCHAIN_EXTERNAL=y
> >>> +        BR2_TOOLCHAIN_EXTERNAL_CUSTOM=y
> >>> +        BR2_TOOLCHAIN_EXTERNAL_DOWNLOAD=y
> >>> +        BR2_TOOLCHAIN_EXTERNAL_URL="http://toolchains.bootlin.com/downloads/releases/toolchains/armv5-eabi--glibc--stable-2017.05-toolchains-1-1.tar.bz2"
> >> I'd rather we use one of the known pre-built toolchains that is
> >> available from the menuconfig. Unfortiunately, for armv5, we have no
> >> better than the codesourcery one. But for armv7, we have a darn-recent
> >> toolchain from ARM.
> Indeed, If we are agree to use armv7 instead of armv5 for the testsuite we
> should use the ARM arm toolchain.

Will you resend an updated patch series, then? Or do you want me to just
fix it in place while I'm applying your patch?

Note that I am currently runing the test with just this BR2_cortex_a7=y
change, so I can fix it locally if you're okay.

> So it also means that 99,99% of users using the ARM toolchain will use a glibc
> based toolchain. But we are testing with an uClibc-ng toolchain most of the
> time. Maybe we should do more tests with glibc and musl toolchain also with 2
> gcc version (the default version and the latest one) ?

Ececpt in this case, systemd is a glibc-only package to befing with, so
we can't test uClibc or musl anyway.

Regards,
Yann E. MORIN.

-- 
.-----------------.--------------------.------------------.--------------------.
|  Yann E. MORIN  | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: |
| +33 662 376 056 | Software  Designer | \ / CAMPAIGN     |  ___               |
| +33 561 099 427 `------------.-------:  X  AGAINST      |  \e/  There is no  |
| http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL    |   v   conspiracy.  |
'------------------------------^-------^------------------^--------------------'

[Buildroot] [PATCH 1/2] support/testing: update the ARM toolchain used for the systemd init tests

[Romain Naour]

Yann, All,

Le 05/04/2020 ? 11:54, Yann E. MORIN a ?crit?:
> romain, All,
> 
> On 2020-04-05 11:35 +0200, Romain Naour spake thusly:
>> Le 05/04/2020 ? 11:09, Yann E. MORIN a ?crit?:
>>> Romain, All,
>>>
>>> On 2020-04-05 10:44 +0200, Yann E. MORIN spake thusly:
>>>> On 2020-04-05 00:02 +0200, Romain Naour spake thusly:
>>>>> As reported on the mailing list, there is a build issue with systemd 245
>>>>> when using gcc < 5.0:
>>>>> http://lists.busybox.net/pipermail/buildroot/2020-April/278931.html
>>> [--SNIP--]
>>>>>          BR2_TOOLCHAIN_EXTERNAL=y
>>>>> +        BR2_TOOLCHAIN_EXTERNAL_CUSTOM=y
>>>>> +        BR2_TOOLCHAIN_EXTERNAL_DOWNLOAD=y
>>>>> +        BR2_TOOLCHAIN_EXTERNAL_URL="http://toolchains.bootlin.com/downloads/releases/toolchains/armv5-eabi--glibc--stable-2017.05-toolchains-1-1.tar.bz2"
>>>> I'd rather we use one of the known pre-built toolchains that is
>>>> available from the menuconfig. Unfortiunately, for armv5, we have no
>>>> better than the codesourcery one. But for armv7, we have a darn-recent
>>>> toolchain from ARM.
>> Indeed, If we are agree to use armv7 instead of armv5 for the testsuite we
>> should use the ARM arm toolchain.
> 
> Will you resend an updated patch series, then? Or do you want me to just
> fix it in place while I'm applying your patch?
> 
> Note that I am currently runing the test with just this BR2_cortex_a7=y
> change, so I can fix it locally if you're okay.

I'm fine if you fix it locally but you have to remove the link to my gitlab test
since it is based on a armv5 toolchain:

https://gitlab.com/kubu93/buildroot/pipelines/132890602

> 
>> So it also means that 99,99% of users using the ARM toolchain will use a glibc
>> based toolchain. But we are testing with an uClibc-ng toolchain most of the
>> time. Maybe we should do more tests with glibc and musl toolchain also with 2
>> gcc version (the default version and the latest one) ?
> 
> Ececpt in this case, systemd is a glibc-only package to befing with, so
> we can't test uClibc or musl anyway.

Indeed for systemd, but my comment is for all the testsuite :)
We have the basic arm toolchain using uClibc-ng to run most of the tests present
in the testsuite.

Best regards,
Romain

> 
> Regards,
> Yann E. MORIN.
>