Prevent possible search_path attacks#918
Merged
knizhnik merged 1 commit intoREL_15_STABLE_neonfrom Mar 17, 2026
Merged
Conversation
mmeent-databricks
approved these changes
Mar 16, 2026
mmeent-databricks
left a comment
mmeent-databricks
left a comment
There was a problem hiding this comment.
One nit, apart from that LGTM
knizhnik
force-pushed
the
search_path_attack_v15
branch
2 times, most recently
from
ed95d65 to
08ce83f
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Problem
See https://databricks.slack.com/archives/C092W8NBXC0/p1771421660357509?thread_ts=1771421653.582579&cid=C092W8NBXC0
We have discussed it many times:
https://databricks.slack.com/archives/C092W8NBXC0/p1769006207119519
last time at this Monday compute sync.
Still not quite clear to me if it is possible or not to prevent all such SECs but just by providing safe search_path.
Let me share my current vision of the problem.
Extensions quite often work with schema public and neon users are allows to create object in schema public . So user can always place some object in schema public which overrides (hides) standard function.
Since PostgreSQL 9.0+, functions are collected from all schemas in the path before resolution. This means adding a schema to search_path can introduce unexpected overload candidates that change which function gets called.
Solution
While selecting function candidates always prefer ones belonging to superuser
Always prepend
pg_catalogtosearch_path(Postgres is doing it only if it is not explicitly include in search path)