Page MenuHomePhabricator

Filter all input to the usable subset of UTF-8
Closed, WontfixPublic

Description

After D8316, we decline to construct queries including UTF-8 outside the basic multilingual plane. It would be slightly better to reject this data sooner, in request handling: if you submit a GET/POST with HTTP data that isn't in the BMP, we can kick you out before doing any sort of processing.

This is mostly nice-to-have and may break a few things (browsers with weird encoding rules?) but is something I plan to pursue after the query BMP stuff settles down.

Event Timeline

epriestley raised the priority of this task from to Low.
epriestley updated the task description. (Show Details)
epriestley added a project: Security.
epriestley added a subscriber: epriestley.
epriestley claimed this task.
  • T1191 is a better approach to this ("make it work").
  • D8316 seems to have stuck and be working well.

We also aren't seeing issues with D8316 interacting badly with regular users (e.g., browsers doing bad stuff), except in expected places (e.g., non-utf8 repositories).