public inbox for isar-users@googlegroups.com
 help / color / mirror / Atom feed
From: "'MOESSBAUER, Felix' via isar-users" <isar-users@googlegroups.com>
To: "Kowalsky, Clara" <clara.kowalsky@siemens.com>,
	"isar-users@googlegroups.com" <isar-users@googlegroups.com>
Cc: "Gylstorff, Quirin" <quirin.gylstorff@siemens.com>
Subject: Re: [PATCH 2/2] rootfs-add-files: use privileged execution helper for rootless builds
Date: Wed, 21 Oct 2026 11:54:04 +0000	[thread overview]
Message-ID: <ce290c1686077ded51943110890522ac13dfa9db.camel@siemens.com> (raw)
In-Reply-To: <20261021112532.3824701-2-clara.kowalsky@siemens.com>

On Wed, 2026-10-21 at 13:25 +0200, 'Clara Kowalsky' via isar-users
wrote:
> rootfs_add_files uses host sudo to create directories and files.
> In unshare mode, package installation runs as root in a user namespace,
> where host-root ownership may not map to namespace root. Dpkg can then
> fail to replace a rootfs file with:

... and in rootless mode we don't require to have sudo installed at
all.

> 
> Unable to create `<path>.dpkg-new`: Permission denied.
> 
> Use RUN_PRIVILEGED_CMD for rootfs operations and disable ownership
> preservation when copying files in unshare mode.

The question is how to control ownership at all for these files.
The whole rootfs_add_files logic seems to be broken.

I even vote for just removing it. There can't be many users, given that
nobody noticed that it breaks under rootless.

> 
> Fixes: 4fedb1ae6f90 ("add support for fully rootless builds")
> Signed-off-by: Clara Kowalsky <clara.kowalsky@siemens.com>
> ---
>  meta/classes-recipe/rootfs-add-files.bbclass | 15 +++++++++++----
>  1 file changed, 11 insertions(+), 4 deletions(-)
> 
> diff --git a/meta/classes-recipe/rootfs-add-files.bbclass b/meta/classes-recipe/rootfs-add-files.bbclass
> index d33752f3..b21e282b 100644
> --- a/meta/classes-recipe/rootfs-add-files.bbclass
> +++ b/meta/classes-recipe/rootfs-add-files.bbclass
> @@ -16,6 +16,13 @@ ROOTFS_ADDITIONAL_FILES ??= ""
>  
>  
>  python rootfs_add_files() {
> +    import shlex
> +
> +    cmd = shlex.split(d.getVar('RUN_PRIVILEGED_CMD'))

bb.process.run executes the command in a shell if not passed as list.
With that, we don't need to split it.

Felix

> +    copy_args = ['-a']
> +    if d.getVar('ISAR_CHROOT_MODE') == 'unshare':
> +        copy_args.append('--no-preserve=ownership')
> +
>      postprocess_additional_files = d.getVar('ROOTFS_ADDITIONAL_FILES').split()
>      rootfsdir = d.getVar("ROOTFSDIR")
>  
> @@ -24,18 +31,18 @@ python rootfs_add_files() {
>          destination = d.getVarFlag(additional_file_entry, "destination") or ""
>          source = d.getVarFlag(additional_file_entry, "source") or ""
>          if os.path.exists(f"{rootfsdir}/{destination}"):
> -            bb.process.run(["/usr/bin/rm", "-f", f"{rootfsdir}/{destination}"])
> +            bb.process.run(cmd + ["/usr/bin/rm", "-f", f"{rootfsdir}/{destination}"])
>  
>          dest_dir = os.path.dirname(destination)
> -        bb.process.run(["sudo", "-E", "/usr/bin/mkdir", "-p", f"{rootfsdir}/{dest_dir}" ])
> +        bb.process.run(cmd + ["/usr/bin/mkdir", "-p", f"{rootfsdir}/{dest_dir}"])
>          # empty source creates only an empty destination file
>          if not source:
> -            bb.process.run(["sudo", "-E", "/usr/bin/touch", f"{rootfsdir}/{destination}" ])
> +            bb.process.run(cmd + ["/usr/bin/touch", f"{rootfsdir}/{destination}"])
>              return
>  
>          if not os.path.exists(f"{source}"):
>              bb.error(f"{source} does not exists and cannot be copied to the rootfs!")
>          # no recursive copy only single files
> -        bb.process.run(["sudo", "-E", "/usr/bin/cp", "-a",  f"{source}", f"{rootfsdir}/{destination}" ])
> +        bb.process.run(cmd + ["/usr/bin/cp"] + copy_args + [source, f"{rootfsdir}/{destination}"])
>  }
>  ROOTFS_INSTALL_COMMAND += "rootfs_add_files"
> -- 
> 2.55.0
> 
> -- 
> You received this message because you are subscribed to the Google Groups "isar-users" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to isar-users+unsubscribe@googlegroups.com.
> To view this discussion visit https://groups.google.com/d/msgid/isar-users/20261021112532.3824701-2-clara.kowalsky%40siemens.com.

-- 
You received this message because you are subscribed to the Google Groups "isar-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to isar-users+unsubscribe@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/isar-users/ce290c1686077ded51943110890522ac13dfa9db.camel%40siemens.com.

  reply	other threads:[~2026-10-21 11:54 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-10-21 11:25 [PATCH 1/2] rootfs-add-files: correctly remove files in ROOTFSDIR 'Clara Kowalsky' via isar-users
2026-10-21 11:25 ` [PATCH 2/2] rootfs-add-files: use privileged execution helper for rootless builds 'Clara Kowalsky' via isar-users
2026-10-21 11:54   ` 'MOESSBAUER, Felix' via isar-users [this message]
2026-10-21 12:21     ` 'Clara Kowalsky' via isar-users
2026-10-21 13:16       ` 'MOESSBAUER, Felix' via isar-users
2026-10-21 11:48 ` [PATCH 1/2] rootfs-add-files: correctly remove files in ROOTFSDIR 'MOESSBAUER, Felix' via isar-users

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=ce290c1686077ded51943110890522ac13dfa9db.camel@siemens.com \
    --to=isar-users@googlegroups.com \
    --cc=clara.kowalsky@siemens.com \
    --cc=felix.moessbauer@siemens.com \
    --cc=quirin.gylstorff@siemens.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox