summaryrefslogtreecommitdiff
path: root/doc/dev/notes
Commit message (Collapse)AuthorAgeFilesLines
...
* tor-hsservice: Remove impl Clone for PowManager.Wesley Aptekar-Cassels2025-01-161-2/+0
| | | | Now that it's not in a Arc internally it shouldn't have this.
* tor-hsservice: Update PowManager persistence plan.Wesley Aptekar-Cassels2025-01-161-7/+19
|
* tor-hsservice: Make PowManager not use internal Arc.Wesley Aptekar-Cassels2025-01-161-4/+4
| | | | | | | | | | Per review feedback: > We have discovered that as a general rule of thumb, it is a good idea > to expose the `Arc`. This has a number of advantages; for example, if a > caller wants a `Weak` for any reason, that's right there. The typical > pattern is for the constructor to return `Arc<Self>` and methods to take > `self: &Arc<Self>`.
* tor-hsservice: Make PoW seed expiration be per-TP.Wesley Aptekar-Cassels2025-01-161-7/+6
| | | | | | This is something we wanted to do in general for security reasons, and ad I was implementing I discovered that it actually makes the code simpler, rather than more complicated.
* tor-hsservice: Update notes about PowManager persistence.Wesley Aptekar-Cassels2025-01-161-1/+1
|
* tor-hsservice: Add KeyMgr and HsNickname to PowManager.Wesley Aptekar-Cassels2025-01-161-0/+6
| | | | | These are needed to allow the PowManager to get the blinded ID needed to construct a PoW verifier.
* tor-hsservice: Give PowManager its own update loop.Wesley Aptekar-Cassels2025-01-161-14/+19
|
* tor-hsservice: Make PowManager use Arc internally.Wesley Aptekar-Cassels2025-01-161-1/+3
| | | | It is cleaner to not force callers to wrap it.
* fix: fix typosDimitris Apostolou2025-01-062-2/+2
|
* Merge branch 'dirauth-0' into 'main'Ian Jackson2024-12-181-0/+242
|\ | | | | | | | | Initial sketch of dirauth structure See merge request tpo/core/arti!2635
| * dirauth sketch: discuss offline KS_dirauth_id_*Ian Jackson2024-12-171-0/+3
| |
| * dirauth sketch: mention shared random protocolIan Jackson2024-12-171-0/+1
| |
| * dirauth sketch: dircache, expand a bitIan Jackson2024-12-171-0/+5
| | | | | | | | | | From https://gitlab.torproject.org/tpo/core/arti/-/merge_requests/2635#note_3131826
| * dirauth sketch: dircache: mention connection to relaysIan Jackson2024-12-171-1/+2
| |
| * dirauth deployment plans: Updates apropos commentsIan Jackson2024-12-171-5/+14
| |
| * dirauth-sketch: Fix typoIan Jackson2024-12-171-1/+1
| |
| * dirauth deployment plans: formattingIan Jackson2024-12-171-1/+4
| |
| * dirauth deployment plans: typosIan Jackson2024-12-171-2/+2
| |
| * dirauth deployment plans: fix formattingIan Jackson2024-12-171-3/+3
| |
| * dirauth deployment plans: detailed transition planIan Jackson2024-12-171-0/+129
| |
| * dirauth deployment plans: add rationale for pluginIan Jackson2024-12-171-0/+20
| |
| * dirauth deployment plans: fix some hard tabsIan Jackson2024-12-171-2/+2
| |
| * Initial sketch of dirauth structureIan Jackson2024-12-171-0/+71
| |
* | docs: Update PoW proposal with lock hierarchy.Wesley Aptekar-Cassels2024-12-171-0/+4
| |
* | docs: Fix proposed PoW struct visibility.Wesley Aptekar-Cassels2024-12-171-1/+1
| |
* | docs: Update proposed PoW locking and add explanation.Wesley Aptekar-Cassels2024-12-171-10/+56
| |
* | docs: Move PoW next_expiration_time into PowVerifiers.Wesley Aptekar-Cassels2024-12-171-1/+1
| |
* | docs: Add note about ReplayLog cleanup.Wesley Aptekar-Cassels2024-12-171-1/+10
| |
* | docs: Update PoW replay log type.Wesley Aptekar-Cassels2024-12-171-1/+1
| |
* | docs: Put all datastructures for PoW seeds in the same struct.Wesley Aptekar-Cassels2024-12-171-5/+7
| | | | | | | | | | This will prevent problems when trying to acquire multiple locks to update this data in a consistent way.
* | docs: s/replay_log_dir/pow_replay_log_dir/ for clarity.Wesley Aptekar-Cassels2024-12-171-1/+1
| |
* | docs: Add note about ReplayLog rotation.Wesley Aptekar-Cassels2024-12-171-0/+1
| |
* | docs: Add note about PowManagerRecord and StateRecord.Wesley Aptekar-Cassels2024-12-171-0/+1
| |
* | docs: Adjust check_solve visibility.Wesley Aptekar-Cassels2024-12-171-1/+1
| |
* | docs: We want nonces to be unique, not solution bytes.Wesley Aptekar-Cassels2024-12-171-1/+1
| |
* | docs: Add details of ReplayLog to service side PoW notes.Wesley Aptekar-Cassels2024-12-171-3/+29
| |
* | docs: Update notes on service-side PoWWesley Aptekar-Cassels2024-12-171-22/+24
| | | | | | | | | | | | | | | | | | | | | | This makes a few major changes: * Removes the IptPowInstance type, instead moving that functionality to PowManager, using a RwLock. * Adds detail about the priority queue types / APIs * Makes all members of PowManager wrapped by a Mutex/RwLock, so that the caller doesn't need to worry about that. The caller will still likely need to put it in a Arc, though.
* | docs: Rework service side PoW draft.Wesley Aptekar-Cassels2024-12-171-74/+93
| |
* | docs: Add note on plan for service side PoW.Wesley Aptekar-Cassels2024-12-171-0/+107
|/
* doc/dev/notes: cell-lifecycle: Say that stream management details are elided.Gabriela Moldovan2024-11-211-0/+4
|
* doc/dev/notes: cell-lifecycle: We only describe the unbounded buffers.Gabriela Moldovan2024-11-211-3/+3
|
* doc/dev/notes: cell-lifecycle: Say that the reactor awaits rather than blocks.Gabriela Moldovan2024-11-211-1/+1
|
* doc/dev/notes: cell-lifecycle: Describe the current state of things.Gabriela Moldovan2024-11-201-0/+319
| | | | | | | | | | | | | | | | | These are my notes and observations about the circuit and channel reactors. In hindsight, this whole document is just stating the "obvious"[^1], so I'm not sure if it's going to be useful for anyone else. Also, since this isn't part of the code docs or tor-proto crate-level docs, it's very likely to become outdated in the near future. At the end of the document, I also wrote a very rough KIST implementation plan based on the discussion from #1712. Closes #1747 [^1]: scare quotes because the "obvious" is only obvious after you've stared at the circ/chan reactors for a while
* doc/dev/notes: keymgr-certificates: Clarify how keymgr might generate the cert.Gabriela Moldovan2024-11-141-2/+3
| | | | | This is not necessarily set in stone, but it seems like a plausible way of going about this.
* doc/dev/notes: keymgr-certificates: Emphasize that we won't manage arbitrary ↵Gabriela Moldovan2024-11-141-3/+8
| | | | certs.
* doc/dev/notes: keymgr-certificates: Abolish get_cert_entry.Gabriela Moldovan2024-11-141-16/+0
| | | | This also removes the reference to the `get_entry()` function.
* doc/dev/notes: keymgr-certificates: Allow the cert denotators to be absent.Gabriela Moldovan2024-11-141-11/+9
|
* doc/dev/notes: keymgr-certificates: Update proposed renaming.Gabriela Moldovan2024-11-141-28/+19
| | | | | Based on our discussion in https://gitlab.torproject.org/tpo/core/arti/-/merge_requests/2565#note_3129065
* doc/dev/notes: keymgr-certificates: Apply key format clarifications from Diziet.Gabriela Moldovan2024-11-141-0/+7
|
* doc/dev/notes: keymgr-certificates: Apply suggestions from Diziet.gabi-2502024-11-141-2/+2
|