- When RS256, RS384, ES384, ES512 was specified as the JWT signing
algorithm they would generate RS512 and ES256 respectively.
- Added unit test.
(cherry picked from commit 7d59060dc6)
- This unifies the security behavior of enrolling security keys with
enrolling TOTP as a 2FA method. When TOTP is enrolled, you cannot use
basic authorization (user:password) to make API request on behalf of the
user, this is now also the case when you enroll security keys.
- The usage of access tokens are the only method to make API requests on
behalf of the user when a 2FA method is enrolled for the user.
- Integration test added.
(cherry picked from commit e6bbecb02d)
- Add a `purpose` column, this allows the `forgejo_auth_token` table to
be used by other parts of Forgejo, while still enjoying the
no-compromise architecture.
- Remove the 'roll your own crypto' time limited code functions and
migrate them to the `forgejo_auth_token` table. This migration ensures
generated codes can only be used for their purpose and ensure they are
invalidated after their usage by deleting it from the database, this
also should help making auditing of the security code easier, as we're
no longer trying to stuff a lot of data into a HMAC construction.
-Helper functions are rewritten to ensure a safe-by-design approach to
these tokens.
- Add the `forgejo_auth_token` to dbconsistency doctor and add it to the
`deleteUser` function.
- TODO: Add cron job to delete expired authorization tokens.
- Unit and integration tests added.
(cherry picked from commit 1ce33aa38d)
v9: Removed migration - XORM can handle this case automatically without
migration. Add `DEFAULT 'long_term_authorization'`.
- If the incoming mail feature is enabled, tokens are being sent with
outgoing mails. These tokens contains information about what type of
action is allow with such token (such as replying to a certain issue
ID), to verify these tokens the code uses the HMAC-SHA256 construction.
- The output of the HMAC is truncated to 80 bits, because this is
recommended by RFC2104, but RFC2104 actually doesn't recommend this. It
recommends, if truncation should need to take place, it should use
max(80, hash_len/2) of the leftmost bits. For HMAC-SHA256 this works out
to 128 bits instead of the currently used 80 bits.
- Update to token version 2 and disallow any usage of token version 1,
token version 2 are generated with 128 bits of HMAC output.
- Add test to verify the deprecation of token version 1 and a general
MAC check test.
(cherry picked from commit 9508aa7713)
- _Simply_ add `^$` to regexp that didn't had it yet, this avoids any
content being allowed that simply had the allowed content as a
substring.
- Fix file-preview regex to have `$` instead of `*`.
(cherry picked from commit 7067cc7da4)
v9: added fix for ref-issue, this is already fixed in forgejo branch but
not backported as it was part of a feature.
- Consider private/limited users in the `AccessibleRepositoryCondition`
query, previously this only considered private/limited organization.
This limits the ability for anomynous users to do code search on
private/limited user's repository
- Unit test added.
(cherry picked from commit b70196653f)
- The RSS and atom feed for branches exposes details about the code, it
therefore should be guarded by the requirement that the doer has access
to the code of that repository.
- Added integration testing.
(cherry picked from commit 3e3ef76808)
- If a repository is forked to a private or limited user/organization,
the fork should not be visible in the list of forks depending on the
doer requesting the list of forks.
- Added integration testing for web and API route.
(cherry picked from commit 061abe6004)
- Ensure that the specified push mirror ID belongs to the requested
repository, otherwise it is possible to modify the intervals of the push
mirrors that do not belong to the requested repository.
- Integration test added.
(cherry picked from commit 786dfc7fb8)
When the CI vars.ROLE is forgejo-coding, it is assumed to be the
repository where collaborative coding happens,
i.e. https://codeberg.org/forgejo/forgejo
When the CI vars.ROLE is forgejo-testing, it is assumed that only codebase
testing is to be run and no other tests such as release build
integration, label constraints, backporting etc.
(cherry picked from commit 068558accd)
Conflicts:
.forgejo/workflows/testing.yml
was in .forgejo/workflows/e2e.yml
When the CI vars.ROLE is forgejo-coding, it is assumed to be the
repository where collaborative coding happens,
i.e. https://codeberg.org/forgejo/forgejo
When the CI vars.ROLE is forgejo-testing, it is assumed that only codebase
testing is to be run and no other tests such as release build
integration, label constraints, backporting etc.
(cherry picked from commit f82840f1ea)
Conflicts:
.forgejo/workflows/merge-requirements.yml
- When a dependency is renamed, specified via `package="actual-name"` in
Cargo.toml, this should become the name of the depedency when the
package is retrieved from the registery by cargo and the old name should
be available in the `package` field.
- The reference implementation also does this: 490e66a9d6/src/controllers/krate/publish.rs (L702-L705)
- Resolves #5936
- Unit test added.
(cherry picked from commit bb93d3e6c8)
Notify https://code.forgejo.org/forgejo/forgejo that a new release was
published by setting the trigger label to
https://code.forgejo.org/forgejo/forgejo/issues/5.
It is only ever useful when a stable release is published, the
experimental releases are not mirrored. But it is triggered in all
cases. This will waste a few mirror check daily, when experimental
releases are built. This is an improvement compared to the current
situation where mirrors are checked hourly:
* Instead of being checked 24 times per day it will be down to less
than 5
* The mirror happens immediately after the release is published
instead of waiting for the next run of the cron job.
If a mirror operation is in progress, as evidenced by the presence of
the trigger label on the issure, it means two releases are being
published. Wait up to 1h for the mirror to complete and remove the
trigger label.
(cherry picked from commit 7492330721)
When a new commit is pushed to an existing pull request, the update of
the commit status will happen asynchronously, via the git hook.
--- FAIL: TestPullRequestCommitStatus/synchronize (2.14s)
actions_trigger_test.go:331:
Error Trace: /workspace/forgejo/forgejo/tests/integration/actions_trigger_test.go:331
Error: Should be true
Test: TestPullRequestCommitStatus/synchronize
(cherry picked from commit 983aed4268)
This css class was used to display the "forgot password"-link right and above the password field.
cd75519a0b moves this link, so this class is now unused
Previously hitting tab in the username field set the focus to the "forgot password" link. Only on the next hit the password field was selected.
This is an issue for some password managers (keepassdx android keyboard) and not as nice for accessibility.
Now the forgot link is below the sign up link at the bottom of the page.
Using "tabindex" didn't work properly with the templating engine because many elements get assigned a tabindex of "0" by default disrupting the tab selection sequence.
Functions modifying the labels in the database (DeleteIssueLabel,
NewIssueLabels, NewIssueLabel, ReplaceIssueLabels) need to force
reload them. Instead of:
issue.isLabelsLoaded = false
issue.Labels = nil
if err = issue.LoadLabels(ctx); err != nil {
return err
}
They can now use:
if err = issue.ReloadLabels(ctx); err != nil {
return err
}
(cherry picked from commit f06bdb0552)
When ReplaceIssueLabels calls issue.LoadLabels it was a noop because
issue.isLabelsLoaded is still set to true because of the call to
issue.LoadLabels that was done at the beginning of the function.
(cherry picked from commit c801838690)
When a workflow has
on:
pull_request:
types:
- labeled
- unlabeled
The outcome of the workflow (success or failure) must be associated
with the head sha commit status. Otherwise it cannot be used as a
requirement for merging the pull request (branch protections).
(cherry picked from commit 66c85b7d8b)
