Re: Addressing SECURITY DEFINER Function Vulnerabilities in PostgreSQL Extensions

On Thu, 6 Jun 2024 at 20:10, Isaac Morland <isaac.morland@gmail.com> wrote:
>
> On Thu, 6 Jun 2024 at 12:53, Jeff Davis <pgsql@j-davis.com> wrote:
>
>>
>> > I didn't get you completely here. w.r.t extensions how will this have
>> > an impact if we set the search_path for definer functions.
>>
>> If we only set the search path for SECURITY DEFINER functions, I don't
>> think that solves the whole problem.
>
>
> Indeed. While the ability for a caller to set the search_path for a security definer functions introduces security problems that are different than for security invoker functions, it's still weird for the behaviour of a function to depend on the caller's search_path. It’s even weirder for the default search path behaviour to be different depending on whether or not the function is security definer.

+1

And +1 to the general idea and direction this thread is going in. I
definitely think we should be making extensions more secure by
default, and this is an important piece of it.

Even by default making the search_path "pg_catalog, pg_temp" for
functions created by extensions would be very useful.