summaryrefslogtreecommitdiff
path: root/crates/tor-dirserver/src
Commit message (Collapse)AuthorAgeFilesLines
* tor-netdoc: authcert: use TimerangeBound for UnverifiedAuthCert::verify (fmt)Ian Jackson2026-06-101-8/+8
| | | | Precisely the result of rustfmt.
* tor-netdoc: authcert: use TimerangeBound for UnverifiedAuthCert::verifyIan Jackson2026-06-101-2/+5
| | | | | | | | | | | | | | TimerangeBound is reasonably nice and this will fit in better when we want to verify votes. Adjust the one non-test call site (in tor-dirserver) using .and_then. In the tests: * Where we expected success, call .check_valid_at and add another .unwrap(). * Where we expected signature verification failure, delete the time parameters. * Where we expected timeliness failure, call .check_valid_at and map the error. * With nontrivial tolerance, add calls to `extend_[pre_]tolerance`.
* everywhere: Add #[allow(clippy::string_slice)]Clara Engler2026-06-091-0/+2
| | | | | | | | This commit adds #[allow(clippy::string_slice)] to all functions in the code where string slices are used, alongside a TODO comment. We do this add the function header to have it consistent, as things like expression based allow's are still experimental.
* maint: Run maint/add_warning to deny string slicesClara Engler2026-06-095-0/+5
| | | | | | | | | | | | This commit executes maint/add_warning with the just added change to deny string slices except in tests. I recommend auditing this by checking out the previous commit followed by running the script yourself and then verifying that the diff is identical to this commit. This commit makes cargo clippy fail. We will add exceptions in the next commit.
* tor-netdoc: Rename NetdocUnverified trait to NetdocParseableUnverifiedIan Jackson2026-06-021-2/+2
| | | | | | | | | | | | The NetdocParseableUnverified derive macro implements this trait (amongst other things). Traits and derive macros should have aligned names. This is only used for parsing, so let's keep the "Parseable" part of the name. I don't think the effort of deprecated alias, for downstream compatibility, is worth it, our compatibility policy notwithstanding.
* Merge branch 'directory-signature-3' into 'main'Clara Engler2026-05-131-16/+2
|\ | | | | | | | | tor-netdoc: Sort out netstatus signatures See merge request tpo/core/arti!3937
| * tor-netdoc: Apply deferred rustfmt churnIan Jackson2026-04-291-3/+1
| |
| * tor-netdoc: Abolish poc's netstatus signature typeIan Jackson2026-04-291-14/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | | | Use prod's Signature instead. This gets rid of: * The old parsing code. We have a new approach based on ItemValueParseable, KeywordOrString and and DigestAlgoInSignature. * The duplicate DirectorySignaturesHashesAccu and its temporary conversions. poc's verify_timeless function needs a little adjustment for the new struct layout.
* | tor-dirserver: Fix for missing_extra_infos() for NULLClara Engler2026-04-281-0/+1
| | | | | | | | | | | | | | | | | | | | If extra-infos is set to NULL, which might be the case for micro descriptors or even router descriptors because the field is optional there, this SQL query fails because it cannot LEFT JOIN server.extra_unsigned_sha1 when this field is NULL. To fix this, we simply add an additional clause to the WHERE statement that filters such rows out.
* | tor-dirserver: Make fingerprint optional in router_descriptorClara Engler2026-04-281-2/+2
|/ | | | | | | | This commit changes the schema to make the fingerprint optional. Reason for this is, that microdescriptors are also stored in this table and microdescriptors only contain an OPTIONAL onion-key, meaning it may not always be possible to determine this value from a microdescriptor, thereby making it silly to require it here.
* tor-dirserver: Make use of only one descriptor hash in missing selectionClara Engler2026-04-271-0/+2
| | | | | | | | | | | | | | | This commit makes use of the presence of only one descriptor hash inside consensus_router_descriptor_member by adding a respective AND clause to the select statements of missing descriptors to only select the rows with a non-null SHA-1 (or SHA-2 for the sake of microdescs) in the consensus_router_descriptor_member table. This is defensive programming and should not be required in practice because the docid as well as the "XOR" CHECK should already ensure that this value is always non-NULL. However, if it still happens and slips through, the statement would return NULL rows, which is not what we want and was previously prevented by ensuring this field was never NULL in the first place.
* tor-dirserver: Store precisely one descriptor hash in schemaClara Engler2026-04-273-24/+16
| | | | | | | | | | | | | | | | | | This commit modifies the consensus_router_descriptor_member table in the database schema, removing the NOT NULL constraint on unsigned_sha1 and unsigned_sha2 by replacing it with a new CHECK constraint that checks that either one of them is set but not both. The reason for this is as follows: We are going to use this table to compute the queue of missing descriptors, which means that we can only populate this table with the data we know from the consensus. The consensus however tells us only one of those hashes, namely sha1 in the case of a consensus-ns and sha2 in the case of a consensus-md. In other words: This commit can also be seen as an effort to change the design of the operation in such a way that the queue is obtained directly from the database and not computed at the start during state transition.
* dirserver: Skip warnings re SaturatingTime::saturating_duration_sinceNick Mathewson2026-04-211-0/+2
| | | | | Rust 1.95 warns us that this is an experimental API that could be stabilized in the future.
* tor-netdoc: Rename `AuthCertUnverified::verify_self_signed`Ian Jackson2026-03-311-1/+1
| | | | | | | | | | | This method verifies all the signatures, and checks that the signing authority is in the provided list. Anyway, authcerts aren't really self-signed: they're a signature by KS_auth_id_rsa on KP_auth_sign_rsa. Note that there is also a `verify_selfcert` method which does only some of the checks, and has some code duplication. That will be cleaned up later.
* Merge branch 'weak_table_explicit_hasher' into 'main'David Goulet2026-03-241-0/+5
|\ | | | | | | | | | | | | Force use of standard hasher with weak_tables. Closes #2418 See merge request tpo/core/arti!3801
| * Force use of standard hasher with weak_tables.Nick Mathewson2026-03-241-0/+5
| | | | | | | | | | | | | | | | | | | | | | | | | | Closes #2418. Fixes TROVE-2026-005, where we would use a less cryptographically secure (and probably less DoS resistant) hash function for these tables if: - We are built alongside another crate that uses `weak-table` - That crate enables the `weak-table/ahash` feature. - We are running on a system without hardware AES. Severity: Low
* | Fix typosTobias Stoeckmann2026-03-247-23/+23
|/ | | | Typos found with codespell
* tor-dirserver: Avoid using NetdocParseable for consensusesIan Jackson2026-03-191-4/+8
| | | | | | | | | | | | We want to stop deriving NetdocParseable directly for body structs. Doing so reveals a call site here in tor-dirmirror where a consensus is parsed and the body data used, but without verifying the signatures. Do this explicitly with the hoop-jumping which is going to become deliberately unavoidable. Add a TODO comment because I'm not sure we have decided explicitloy that this is OK.
* tor-dirserver: Avoid using NetdocParseable for consensuses (prep)Ian Jackson2026-03-191-4/+10
| | | | Formatting changes which make the next commit more readable.
* tor-netdoc: parse2: Introduce SignatureData structIan Jackson2026-03-191-2/+2
| | | | This is going to contain body information, and the hashes, too.
* tor-dirserver: Require AuthCerts to make progressClara Engler2026-03-091-37/+54
| | | | | | This commit changes the functionality of the AuthCerts state to only report a success when at least a single certificate was included in the response.
* tor-dirserver: Add retry logic POC TODOClara Engler2026-03-091-0/+4
| | | | | Adds a small TODO with regard to a potentially broken retry logic in the proof-of-concept.
* tor-dirserver: SQL comment for unsigned hashesClara Engler2026-03-091-0/+11
| | | | | This commit documents why and how we use the `unsigned_` fields in the `consensus_router_descriptor_member` table alongside SQL limitations.
* tor-dirserver: Move POC to own moduleClara Engler2026-03-092-73/+91
| | | | | This commit moves dirserver POC code to an own module to semantically indicate it is not production ready.
* tor-dirserver: Document stream dropClara Engler2026-03-091-1/+2
| | | | | | This commit documents why we drop the HTTP TCP stream and why this is fine, namely because this is compliant HTTP/1.0 behavior where there is no connection reuse.
* tor-dirserver: Link to discussion regarding TODOClara Engler2026-03-091-0/+3
| | | | | This commit links the discussion for the TODO for the dirmirror's handling of forward compatibility with netdocs.
* tor-dirserver: Use collect instead of mutable forClara Engler2026-03-091-18/+9
|
* tor-dirserver: Remove fiddly SQL CHECKClara Engler2026-03-091-5/+1
|
* tor-dirserver: Use tor-llcrypto for digestsClara Engler2026-03-091-4/+4
| | | | | | | | This commit replaces the uses of sha1, sha2, and sha3 with their respective pedants from tor-llcrypto for better consistency. Internally, they still use the same logic and underlying crates but let's use this encapsulation nonetheless.
* tor-dirserver: Improve hash wrapper macro docsClara Engler2026-03-091-0/+19
|
* tor-dirserver: Move schema to own fileClara Engler2026-03-092-170/+167
| | | | | | | | This commit moves the database schema into schema_v1.sql and uses include_str to include it. For now, the schema lives in the `src/` directory. If this becomes a problem because we get more schemas and schema upgrades, we can move it into another sub directory, but let's not overengineer the hierarchy there.
* tor-dirserver: PoC for FSM main loopClara Engler2026-03-091-2/+73
| | | | | This commit implements a PoC serving as the main loop for the FSM, demonstrating how invocation and error handling works.
* tor-dirserver: TODO a torspec DoS issueClara Engler2026-03-091-0/+9
| | | | Discussed with nickm on IRC, there will be a torspec issue soon.
* tor-dirserver: Implement `State::AuthCerts`Clara Engler2026-03-092-7/+229
| | | | | | | | | | | | | | This commit implements the logic required for retrieving, validating, and storing authority certificates. The implementation determines the missing certificates by looking at the signatories of the unvalidated consensus and checking them in the db. Afterwards, they will be queried and individually filtered and verified before being inserted into the database. A return of this implementation notably DOES NOT imply all missing certificates have been downloaded. This was chosen for a simplified retry logic.
* tor-dirserver: Add `StaticEngine::send_request()`Clara Engler2026-03-092-47/+104
| | | | | | | | | | | | | This commit adds a new method to static engine that serves as a convenience wrapper around `tor_dirclient::send_request`. The reason for that is, that fetch_consensus is not the only method that requires performing download requests, so it makes sense to generalize it. Besides, it also adds support for parsing multiple netdocs alongside storing their raw variant, which is required for inserting them into the database at one point eventually.
* tor-dirserver: Add IsFatal traitClara Engler2026-03-092-19/+24
| | | | | This commit adds the IsFatal trait to the err module for having a generic signature for the fatality of certain error variants.
* tor-dirservert: Implement FetchConsensus stateClara Engler2026-03-092-2/+167
| | | | | | | | | | | | This commit implements the `FetchConsensus` state by adding a method to `StaticEngine` called `fetch_consensus`, which retrieves the consensus from an upstream directory authority. Likewise, it also implements a new error type called `AuthorityRequestError`. The retry logic is handled externally which will be done in later commits.
* tor-dirserver: Derive PartialEq and Eq for StateClara Engler2026-03-091-2/+2
| | | | Required to test state transitions properly.
* tor-dirserver: PreferredRuntime in StaticEngineClara Engler2026-03-091-2/+9
| | | | This is required for compatibility with other crates in arti.
* tor-dirserver: Remove unused error variantsClara Engler2026-03-091-62/+0
|
* tor-dirserver: Remove download moduleClara Engler2026-03-092-367/+0
| | | | | This commit removes the download manager module because it does not fit well into the mental model of our current finite state machine anymore.
* tor-dirserver: Remove `preferred` from dataClara Engler2026-03-091-9/+1
| | | | | | | | | | This commit removes the `preferred` member field from the `Unverified` and `Verified` variant in `ConsensusBoundData` while adding it as a parameter to `StaticEngine::execute`, with the idea being that the retry logic is handled by the caller anyways, involving the selection of authorities. Right now, I am still a bit unsure how this will play out.
* tor-dirserver: Fix rustdoc commentsClara Engler2026-03-092-5/+5
|
* tor-dirserver: Implement consensus loadingClara Engler2026-03-091-16/+205
| | | | | | This commit implements the consensus loading mechanism by glueing together the logic from the database module with regard to querying missing descriptors.
* tor-dirserver: Query missing descriptor logicClara Engler2026-03-091-1/+394
| | | | | | | | | | | | | | | | | | | This commit implements the logic for querying missing descriptors by adding three new methods to ConsensusMeta: * missing_servers * missing_extras * missing_micros All of them essentially work the same, namely by querying the consensus_router_descriptor_member database table alongside the docid of the current consensus, left joining the respective router_descriptor and returning the digests on all rows where the left join resulteed in a NULL. A small exception are extra-info descriptors. There, we can only return the ones where we have an accompanying server descriptor, hence why we perform an inner join with the server descriptors between the from and the left join.
* tor-dirserver: Rename `md` to `microdesc`Clara Engler2026-03-091-2/+2
| | | | | This is better because it is returned by `ConsensusFlavor::Microdesc::name()`.
* tor-dirserver: Fix extra info schemaClara Engler2026-03-091-7/+10
| | | | | | | | | | | | This commit fixes some parts in the extra-info related part of the database schema. Most notably, it changes the semantic to indicate that the sha1 represents the unsigned part of the extra-info document. Likewise, it also ensures that the reference to an extra-info document must not be null with a plain consensus but null with a microdescriptor consensus.
* tor-dirserver: Store unsigned SHA for descsClara Engler2026-03-091-12/+13
| | | | | | | | | | | This commit modifies the database schema to rename `sha1` and `sha2` to `unsigned_sha1` and `unsigned_sha2`. Network status consensuses refer to server descriptors by their unsigned sha1. Microdescriptor consensuses refer to microdescriptors by their sha2, which is always unsigned, so we adapt that name for consistency.
* tor-dirserver: Remove foreign key from descriptor memberClara Engler2026-03-091-4/+8
| | | | | | | | | | | This commit removes the foreign key constraints from the `consensus_router_descriptor_member` in order to allow the insertion of the router descriptors contained in a consensus before the respective descriptors were fetched. This also allows to directly compute the missing descriptors in SQL, using a left join on this table on `router_descriptor` and obtaining the entries that are NULL.
* tor-dirserver: Store sha1 for extra-infoClara Engler2026-03-091-2/+6
| | | | | | | | | | | This commit modifies the database schema by tracking the sha1 of the extra-info instead of the rowid. This makes generating queues and other things easier. It also removes the foreign key constraint in order to allow for an asynchronous retrieval and storage. Otherwise it would not be possible to store a router descriptor without having obtained the extra info first.
generated by cgit v1.2.3 (git 2.46.0) at 2026-08-26 22:25:59 +0000