Re: [PATCH v2] rootfs class lock undefined in corner cases fix

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

On 04.01.23 20:46, roberto.foglietta@gmail.com wrote:
> From: "Roberto A. Foglietta" <roberto.foglietta@gmail.com>
> 
> rootfs class lock undefined in corner cases fix
> 
> Signed-off-by: Roberto A. Foglietta <roberto.foglietta@gmail.com>
> ---
>  meta/classes/rootfs.bbclass | 5 ++++-
>  1 file changed, 4 insertions(+), 1 deletion(-)
> 
> diff --git a/meta/classes/rootfs.bbclass b/meta/classes/rootfs.bbclass
> index b3faade..63735d4 100644
> --- a/meta/classes/rootfs.bbclass
> +++ b/meta/classes/rootfs.bbclass
> @@ -198,11 +198,14 @@ python do_rootfs_install() {
>          if (d.getVarFlag(cmd, 'isar-apt-lock') or "") == "acquire-before":
>              lock = bb.utils.lockfile(d.getVar("REPO_ISAR_DIR") + "/isar.lock",
>                                       shared=True)
> +        else:
> +            lock = ""
>  
>          bb.build.exec_func(cmd, d)
>  
>          if (d.getVarFlag(cmd, 'isar-apt-lock') or "") == "release-after":
> -            bb.utils.unlockfile(lock)
> +            if (lock != ""):
> +                bb.utils.unlockfile(lock)
>      progress_reporter.finish()
>  }
>  addtask rootfs_install before do_rootfs_postprocess after do_unpack

I thinks you didn't understand the logic yet when you are proposing such
a change and you are rather papering over a downstream issue: missing
"acquire-before" in a commend chain that later on has a "release-after".

We can improve detecting such bugs by asserting on lock being defined
when running into a "release-after".

Jan

-- 
Siemens AG, Technology
Competence Center Embedded Linux