rend-spec: add a note about replay detection
1 unresolved thread
1 unresolved thread
Explain why we're doing it, and the extent to which we do (or do not) need to be perfect.
Closes #246 (closed).
Based on discussion at arti#725 (comment 2959820)
Merge request reports
Activity
assigned to @nickm
requested review from @ahf
473 so long as the opportunity for replay attacks remains limited. 474 For example, it is not necessary to fsync() data to disk 475 after each request. 476 477 > Rationale: 478 > 479 > The main reason we prevent INTRODUCE replays 480 > is to detect attempts by introduction points to mount replay attacks. 481 > Such attacks would cause the onion service to make a second circuit 482 > to the client's chosen rendezvous point. 483 > If the attacker controls both the introduction point 484 > and the rendezvous point, 485 > they can use this to learn which original user circuit 486 > corresponded to the replayed request. 487 > This likely helps with traffic analysis somewhat, 488 > but is unlikely to changed this line in version 2 of the diff
@dgoulet @Diziet would one of you like to review this, since we were the ones discussion on arti#725 (closed)?
- Resolved by David Goulet
added 1 commit
- 3d07bf66 - fixup! rend-spec: add a note about replay detection
marked this merge request as draft from nickm/torspec@3d07bf66
@dgoulet Any comments to Nick's changes? Otherwise, I think we should merge this.
added 35 commits
-
3d07bf66...33ef7615 - 34 commits from branch
tpo/core:main
- d1daab72 - rend-spec: add a note about replay detection
-
3d07bf66...33ef7615 - 34 commits from branch
enabled an automatic merge when all merge checks for d1daab72 pass
mentioned in commit 3d46c77e
Please register or sign in to reply