Re: Misplaced superuser check in pg_log_backend_memory_contexts()
Julien Rouhaud <rjuju123@gmail.com>
From: Julien Rouhaud <rjuju123@gmail.com>
To: Michael Paquier <michael@paquier.xyz>
Cc: Postgres hackers <pgsql-hackers@lists.postgresql.org>, Fujii Masao <masao.fujii@oss.nttdata.com>
Date: 2021-06-06T07:13:12Z
Lists: pgsql-hackers
On Sun, Jun 06, 2021 at 03:53:10PM +0900, Michael Paquier wrote: > > While reading the code of pg_log_backend_memory_contexts(), I have > been surprised to see that the code would attempt to look at a PROC > entry based on the given input PID *before* checking if the function > has been called by a superuser. This does not strike me as a good > idea as this allows any users to call this function and to take > ProcArrayLock in shared mode, freely. It doesn't seem like a huge problem as at least GetSnapshotData also acquires ProcArrayLock in shared mode. Knowing if a specific pid is a postgres backend or not isn't privileged information either, and anyone can check that using pg_stat_activity as an unprivileged user (which will also acquire ProcArrayLock in shared mode). > > It seems to me that we had better check for a superuser at the > beginning of the function, like in the attached. However +1 for the patch, as it seems more consistent to always get a permission failure if you're not a superuser.
Commits
-
Reorder superuser check in pg_log_backend_memory_contexts()
- 4e47b0283482 14.0 landed