Re: [PATCH v2] base: Fix HOST_ARCH for native builds

[Jan Kiszka <jan.kiszka@siemens.com>]

On 29.09.23 19:33, Uladzimir Bely wrote:
> On Fri, 2023-09-29 at 20:25 +0300, Uladzimir Bely wrote:
>> On Fri, 2023-09-29 at 19:36 +0300, Uladzimir Bely wrote:
>>> On Mon, 2023-09-25 at 13:41 +0200, Jan Kiszka wrote:
>>>> From: Jan Kiszka <jan.kiszka@siemens.com>
>>>>
>>>> HOST_ARCH must be DISTRO_ARCH when we are not cross-building.
>>>> Otherwise,
>>>> recipes that set PACKAGE_ARCH to it will fail in native builds.
>>>>
>>>> Use late expansion for ISAR_CROSS_COMPILE to avoid recursion
>>>> issues.
>>>>
>>>> Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
>>>> ---
>>>>
>>>> That "expand=False" seems to work fine, so let's move forward
>>>> with
>>>> it.
>>>>
>>>>  meta/classes/base.bbclass | 6 ++++--
>>>>  1 file changed, 4 insertions(+), 2 deletions(-)
>>>>
>>>> diff --git a/meta/classes/base.bbclass
>>>> b/meta/classes/base.bbclass
>>>> index 88004120..8ca089a2 100644
>>>> --- a/meta/classes/base.bbclass
>>>> +++ b/meta/classes/base.bbclass
>>>> @@ -49,13 +49,15 @@ def oe_import(d):
>>>>  # We need the oe module name space early (before INHERITs get
>>>> added)
>>>>  OE_IMPORTED := "${@oe_import(d)}"
>>>>  
>>>> -def get_deb_host_arch():
>>>> +def get_deb_host_arch(d):
>>>>      import subprocess
>>>> +    if d.getVar("ISAR_CROSS_COMPILE", expand=False) != "1":
>>>> +        return d.getVar("DISTRO_ARCH")
>>>>      host_arch = subprocess.check_output(
>>>>          ["dpkg", "--print-architecture"]
>>>>      ).decode('utf-8').strip()
>>>>      return host_arch
>>>> -HOST_ARCH ??= "${@get_deb_host_arch()}"
>>>> +HOST_ARCH ??= "${@get_deb_host_arch(d)}"
>>>>  HOST_DISTRO ??= "${DISTRO}"
>>>>  
>>>>  die() {
>>>
>>> For some reason this doesn't properly work for some targets in case
>>> of
>>> cross-compile mode.
>>>
>>> Without patch (default local.conf with ISAR_CROSS_COMPILE set):
>>> ```
>>> bitbake mc:qemuarm64-focal:isar-image-base -e | grep "^HOST_ARCH="
>>> HOST_ARCH="amd64"
>>> bitbake mc:rpi-arm64-v8-bullseye:isar-image-base -e | grep
>>> "^HOST_ARCH="
>>> HOST_ARCH="amd64"
>>> bitbake mc:rpi-arm-v7-bullseye:isar-image-base -e | grep
>>> "^HOST_ARCH="
>>> HOST_ARCH="amd64"
>>> bitbake mc:stm32mp15x-bullseye:isar-image-base -e | grep
>>> "^HOST_ARCH="
>>> HOST_ARCH="amd64"
>>> ```
>>>
>>> With patch:
>>> ```
>>> bitbake mc:qemuarm64-focal:isar-image-base -e | grep "^HOST_ARCH="
>>> HOST_ARCH="amd64"
>>> bitbake mc:rpi-arm64-v8-bullseye:isar-image-base -e | grep
>>> "^HOST_ARCH="
>>> HOST_ARCH="amd64"
>>> bitbake mc:rpi-arm-v7-bullseye:isar-image-base -e | grep
>>> "^HOST_ARCH="
>>> HOST_ARCH="armhf"
>>> bitbake mc:stm32mp15x-bullseye:isar-image-base -e | grep
>>> "^HOST_ARCH="
>>> HOST_ARCH="armhf"
>>> ```
>>>
>>> For some reasons this affects only "armhf" targets...
>>>
>>
>> ... And it looks to be caused by line 69 in "imagetypes.bbclass":
>> ISAR_CROSS_COMPILE:armhf = "${@bb.utils.contains('IMAGE_BASETYPES',
>> 'ubifs', '1', '${THIS_ISAR_CROSS_COMPILE}', d)}"
>>
> ... And it seems extactly this line caused the issue with parsing
> "armhf" multiconfigs in patch v1.
> 

OK, good to know this. I have no idea yet how to fix all this.

I just tried to avoid the issue by switching the affected recipes to
-native. Would be cleaner by now anyway. However, that mechanism also
uses the broken HOST_ARCH.

crossvars has BUILD_HOST_ARCH which actually seems to have the correct
values. But why do we have to have two of such vars?

Jan

-- 
Siemens AG, Technology
Linux Expert Center