On Wed, Dec 04, 2024 at 02:01:02PM -0500, Stefan Berger wrote:
>
>
> On 12/3/24 6:34 PM, Mimi Zohar wrote:
> > Like direct file execution (e.g. ./script.sh), indirect file exection
>
> typo: execution
>
> > (e.g. sh script.sh) needs to be measured and appraised. Instantiate
>
> If I understand the underlying patches correctly then 'sh script.sh' would
> be evaluated with execveat(AT_CHECK) but this requires the execute flag to
> be set. To maintain backwards compatibility sh cannot assume that script.sh
> has the execute flag set since it doesn't need today:
>
> $ echo 'echo hi' > foo.sh
> $ sh foo.sh
> hi
>
> the same is true for python:
>
> $ echo 'print("hi")' > foo.py
> $ python foo.py
> hi
>
> I am not sure which interpreters are going to be able to take advantage of
> this or whether they will behave differently if the x bit is set versus when
> it is not set...?
This is a valid concern handled with two new securebits. See the
related patch series and documentation:
https://lore.kernel.org/all/20241112191858.162021-3-mic@xxxxxxxxxxx/
