Backport #40376 by @bircni
Plain repository creation marked a new repository as ready before its
git data, hooks and initial commit existed, and auto-init cloned the new
repository to build its initial commit on top of whatever it already
contained.
- Auto-init builds the initial commit in a freshly initialized temp
repository instead of a clone
- New repositories are inserted as being created and marked ready once
initialized, as fork, template generation, adoption and migrations
already do
---------
Co-authored-by: bircni <bircni@icloud.com>
Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
Co-authored-by: silverwind <me@silverwind.io>
Backport https://github.com/go-gitea/gitea/pull/39993
Gitea accepted unlimited wrong TOTP passcodes, so anyone holding an
account's password could enumerate the 6-digit code in minutes. This
allows 5 failed attempts per account, then refuses TOTP until 15 minutes
after the last one.
- The count lives in the cache, and the global lock spans the check,
validation and counting, so new sessions and concurrent requests cannot
widen it. With Redis, both are shared across instances
- Applies to every TOTP check: the web 2FA page, account recovery and
Basic auth with `X-Gitea-OTP` (API, git, packages)
- Only wrong or replayed passcodes count, server errors do not
- Scratch codes and WebAuthn stay usable as the recovery path while
locked
- Failures log `Failed authentication attempt for <user> from <ip>`,
which the fail2ban filter already matches
Co-authored-by: Sandiyo Christan <55909152+sandiyochristan@users.noreply.github.com>
Co-authored-by: silverwind <me@silverwind.io>
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
Backport https://github.com/go-gitea/gitea/pull/40322
Several API and background paths change repositories without writing the
audit event the web UI writes for the same change. This records the
existing actions on those paths, and webhook, secret and OAuth2
application events now also keep the impersonator.
Conflicts resolved for release/v28:
- The web project close/reopen handler uses
`NotFoundOrServerError`/`ServerError`, as `ServerErrorAuto` does not
exist on this branch.
- The repository webhook delete API keeps its documented 404 response.
- The notifier test keeps the release branch layout with the new cases
appended.
Signed-off-by: bircni <bircni@icloud.com>
Co-authored-by: silverwind <me@silverwind.io>
Co-authored-by: delvh <dev.lh@web.de>
Backport https://github.com/go-gitea/gitea/pull/40377
The web repo search response embedded the whole commit status model
although the dashboard only uses its state and target URL. It now
returns just those two fields.
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
Backport https://github.com/go-gitea/gitea/pull/40381
The API accepted a new name for an organization's Owners team. The owner
team is looked up by its name, so the organization had no owner team
afterwards and its owners lost access to it.
The API now rejects the rename, as the web UI already does.
Fixes#40380
Backport https://github.com/go-gitea/gitea/pull/40192
The repository issue search JSON route sits behind the issues reader
check only, and the milestone page is open to anyone who can read either
issues or pull requests, since the issue routes have to tolerate pull
request permissions. Neither handler narrows the issue type to what the
viewer may read: the search route takes `type=pulls` straight from the
query or lists both types when it is absent, and the milestone page
always lists both. A team member granted only the Issues unit on a
private repository can therefore fetch every pull request's title and
body as JSON and see pull requests on milestone pages, while a
pull-requests-only reader sees issues there. I noticed it while
comparing these routes with the API `ListIssues` handler and the web
issue suggestions endpoint, which both already narrow the type. The
narrowing now lives in one small helper used by all three handlers, with
the search route answering 404 for a requested type the viewer cannot
read, like the API does. A unit test covers both handlers with an
issues-only permission.
Co-authored-by: Basavaraj S m <basavaraj@digiscrypt.com>
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
Backport https://github.com/go-gitea/gitea/pull/39991
Changing a pull mirror address with an empty password field carried the
stored credentials over to the new address. The password is write-only,
so a repo admin who did not set up the mirror could send the owner's
token to a host they control, or pull another repository on the same
host that the token can read.
Stored credentials are now reused only when the address is unchanged,
like a changed username already drops the stored password. The API
cannot change the address and push mirrors never reuse credentials, so
both are unaffected.
Co-authored-by: Basavaraj S m <basavaraj@digiscrypt.com>
Co-authored-by: silverwind <me@silverwind.io>
Backport of the `GetAffectedFiles` fix from main (#40326) to v28.
`git diff --name-only` C-quotes paths containing characters such as `"`
or non-ASCII bytes, so they never matched protected file patterns. Use
`diff-tree -r -z` and split on NUL. Renames now report both old and new
paths.
Assisted-by: Codet:claude
Backport https://github.com/go-gitea/gitea/pull/40042
`GET /users/{username}/orgs/{org}/permissions` returned a user's team
roles to anyone who could see the org, revealing concealed members and
their roles. Now only the user, unrestricted org members and site admins
get the real answer, everyone else gets the non-member response:
- Outsiders get no roles, for public members too.
- Restricted members only get their own roles.
- Public-only tokens get none, and site admins need an `all`-scoped
token.
v28 lacks the doer token helpers from
https://github.com/go-gitea/gitea/pull/39690, so the check reads
`ctx.PublicOnly` and the request's token scope instead.
Assisted-by: Claude Code:claude-opus-5-5
Co-authored-by: Basavaraj S m <basavaraj@digiscrypt.com>
Backport #40363 by @lazerg
The container registry rejected image names with `__` or runs of `-`
(like `a__b` or `a--b`) with `NAME_INVALID`, because the name check
still used the distribution-spec v1.0 grammar. This updates the pattern
to the v1.1 grammar, which docker and podman already accept.
Fixes: https://github.com/go-gitea/gitea/issues/40360
Assisted-by: Claude Code:claude-opus-5-5
Signed-off-by: wxiaoguang <wxiaoguang@gmail.com>
Co-authored-by: Lazizbek Ergashev <lazerg2@gmail.com>
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
Backport #40349 by @silverwind
Dropdowns initialized without an options object shared the default
Fomantic templates, so the ARIA patch kept wrapping `templates.menu` on
the defaults and retained every dropdown ever created. This grows memory
on the Actions run list, which re-renders its dropdowns on each refresh.
Each dropdown now gets its own templates copy, which the ARIA patch
modifies in place.
Fixes: https://github.com/go-gitea/gitea/issues/40347
Assisted-by: Claude Code:claude-opus-5-5
Co-authored-by: silverwind <me@silverwind.io>
Backport #40334 by @SudhanshuMatrix
Markdown renders `[X]` as a checked task like `[x]`, but parts of the
frontend only recognize lowercase `x`.
- Clicking a `[X]` checkbox now unchecks it instead of failing.
- Pressing Enter on a `- [X]` line in the editor continues the list with
`- [ ] ` instead of dropping the checkbox.
Assisted-by: Claude Code:claude-opus-5-5
Co-authored-by: Shudhanshu Singh <sudhanshuwriterblc@gmail.com>
Co-authored-by: silverwind <me@silverwind.io>
Backport https://github.com/go-gitea/gitea/pull/40131
Several code paths skipped checks that their siblings already perform.
- Code comment replies are rejected when they target a review of another
pull request or another user's pending review
- OAuth2 refuses tokens for users who can't sign in, and introspection
reports their tokens and already used refresh tokens as inactive
- AGit proc-receive runs the same pull request permission and wiki
checks as pre-receive
- Repository transfer drops the repository's owner-level scoped workflow
source registrations, and registrations whose source the owner no longer
owns are ignored
- The web editor's base branch copy into a fork and the fork's upstream
divergence info require read access to the base repository
- LFS downloads require sign-in under strict `REQUIRE_SIGNIN_VIEW`, like
git smart HTTP
- Unknown AGit target branches are reported as an unexpected ref again
Assisted-by: Claude:claude-opus-5-5
Backport https://github.com/go-gitea/gitea/pull/40261
The issue reference popup showed its labels without any spacing, and the
other label lists each used their own gap, with wrapped rows spaced
differently from columns.
`.labels-list` is now a wrapping flex list with `--gap-inline` by
default, giving the popup, issue sidebar and project cards equal gaps in
both directions. Lists inline with text, in issue list titles and
timeline events, stay inline so their labels still wrap one by one.
<img width="600" alt="Issue reference popup, issue sidebar and issue
list labels, before and after"
src="https://github.com/user-attachments/assets/fc4f0b9e-2b9d-4060-a2fb-c1bd3b7391db"
/>
Assisted-by: Claude Code:claude-opus-5-5
Backport #40315 by @GiteaBot
This PR contains the following updates:
| Package | Type | Update | Change |
|---|---|---|---|
| [go](https://go.dev/)
([source](https://redirect.github.com/golang/go)) | toolchain | patch |
`1.27.3` → `1.27.4` |
---
### Configuration
📅 **Schedule**: (UTC)
- Branch creation
- At any time (no schedule defined)
- Automerge
- At any time (no schedule defined)
🚦 **Automerge**: Disabled by config. Please merge this manually once you
are satisfied.
♻ **Rebasing**: Whenever PR becomes conflicted, or you tick the
rebase/retry checkbox.
🔕 **Ignore**: Close this PR and you won't be reminded about this update
again.
---
- [ ] <!-- rebase-check -->If you want to rebase/retry this PR, check
this box
---
This PR has been generated by [Mend Renovate
CLI](https://redirect.github.com/renovatebot/renovate).
<!--renovate-debug:eyJjcmVhdGVkSW5WZXIiOiI0NC4yMTAuMCIsInVwZGF0ZWRJblZlciI6IjQ0LjIxMC4wIiwidGFyZ2V0QnJhbmNoIjoibWFpbiIsImxhYmVscyI6WyJkZXBlbmRlbmNpZXMiXX0=-->
Backport #40291 by @lunny
`PUT/DELETE /repos/{owner}/{repo}/issues/{index}/lock` passed
`ctx.ContextUser` as the doer of `LockIssue` / `UnlockIssue`. In the
repo routes that is the owner of the repository, so on a repository of
an organization the lock and unlock timeline comments were created by
the organization instead of the user who did it. The web handler already
uses the signed-in user; this uses `ctx.Doer` as well.
Adds a subtest `DoerOnOrganizationRepo` to the existing
`TestAPILockIssue`: a member of an organization locks and unlocks an
issue of an organization repository, and the comments must be posted by
that member.
Found while testing audit logging of the issue lock events downstream.
_Authored by Codet (Claude)._
Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
When a file, README or wiki page is rendered as markup, the HTML is read
through a 32 KiB buffer to escape invisible and ambiguous characters. A
multi-byte character cut by that buffer right after a tag was shown as
broken characters, e.g. `<p>보기</p>` became `<p>���기</p>`. Such a part is
now kept for the next read. Two related cases are fixed as well:
- Invalid bytes right before a tag are marked as broken instead of
passing through raw as part of the tag.
- A BOM split across reads is kept instead of being escaped as
invisible.
To reproduce, view a Markdown file with 32734 `A` characters, a blank
line and `보기`, which puts `보` across the first 32 KiB of the rendered
HTML:
[split-rune-after-tag.md](https://github.com/user-attachments/files/34361104/split-rune-after-tag.md)
https://github.com/go-gitea/gitea/pull/39937 replaces this reader with a
single in-memory pass. This small fix can be backported to the release
branches.
Assisted-by: Claude Code:claude-opus-5-5
----
| Before (2ed6a6f61b) | After
(2a1d09755f) |
| --- | --- |
|

|

|
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
Update clippie to 4.3.1, which fixes copy buttons in browsers that have
`navigator.clipboard.write` but no `ClipboardItem`, such as Pale Moon.
Text is now copied through `writeText`, with a copy event as fallback.
The same update is part of https://github.com/go-gitea/gitea/pull/40292
on main.
Fixes: https://github.com/go-gitea/gitea/issues/40287
Assisted-by: Claude Code:claude-opus-5-5
Backport #40053 by @silverwind
Copying lines from a diff, file view, blame, code search or code preview
and pasting them into a comment on Gitea or GitHub produced a markdown
table with escaped pipes and stripped indentation. Both use
`@github/paste-markdown`, which converts copied HTML tables unless they
carry `data-paste-markdown-skip`, as GitHub's own code tables do.
Gitea's code tables now set it too.
The test is a bit more complicated because of a upstream bug which
causes `paste-markdown` to insert pasted content twice, once
https://github.com/github/paste-markdown/pull/128 is merged and
released, the negative assertion could be removed.
Signed-off-by: silverwind <me@silverwind.io>
Co-authored-by: silverwind <me@silverwind.io>
Backport #40260 by @lazerg
Pending reviews created through the API went through `SubmitReview`,
which added a timeline entry and sent notifications before the review
was submitted, so submitting it later showed the review twice.
- Pending reviews skip the timeline entry and notifications until
submitted, like on GitHub
- `CreatePullReview` rejects comments without a `path` or a positive
line with 422, like GitHub
- Already duplicated reviews stay as they are
Fixes https://github.com/go-gitea/gitea/issues/40259
Assisted-by: Claude Code:claude-opus-5-5
Co-authored-by: Lazizbek Ergashev <lazerg2@gmail.com>
Co-authored-by: silverwind <me@silverwind.io>
Backport #40120 by @silverwind
Several code paths skipped checks their siblings perform, or cleaned up
more than they validated.
- Official team review requests now count, and removing one no longer
marks an unrelated review official
- Release edits only delete files of validated attachments
- Rejected container uploads no longer delete an existing blob
- Attachment removal only removes the uploader's own pending uploads
- OpenID sign-in checks the allow and block lists against the normalized
identifier
- HTTP git access is rejected while a repository is being created, like
SSH
Signed-off-by: silverwind <me@silverwind.io>
Co-authored-by: silverwind <me@silverwind.io>
Co-authored-by: delvh <dev.lh@web.de>
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
Backport #40102 by @Zettat123
The post-receive hook syncs pushed branches to the database, but
`is_empty` and the default branch of an empty repository were only
updated later by the `push_update` queue. Until then the API treated the
repository as empty (branch list `null`, `GET /branches/{branch}` 500),
and the queue overwrote the default branch chosen by the hook or set via
the API.
`SyncBranchesToDB` now initializes an empty repository and repairs a
default branch missing in git in its locked transaction, so the state is
final when `git push` returns. The push queue and web file edits retry
it while the repository is still empty.
- The default branch is the configured one if it exists, else `main`,
else `master`, else the first branch by name, like GitLab. HEAD follows
it.
- A first push of several branches no longer depends on push order. The
push that initializes a repository still prints no pull request hints,
like GitHub.
---------
Co-authored-by: Zettat123 <zettat123@gmail.com>
Co-authored-by: silverwind <me@silverwind.io>
Backport https://github.com/go-gitea/gitea/pull/40110
The post-receive hook wrote each push's new commit into the branch
table. Each push runs its own hook after the ref is updated, so with
pushes in close succession, like CI pushing back or a quick force-push,
a slower first hook can finish last. The table then keeps the older
commit, or marks a re-created branch as deleted.
- The sync reads each pushed branch's state from git under a repo-scoped
lock
- Deletions go through the same sync, so a re-created branch stays
active
- A branch that fails to sync is logged and skipped instead of aborting
the batch
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
Co-authored-by: Giteabot <teabot@gitea.io>
Backport #40252 by @silverwind
The `alpine:3.24` base image lags behind fixes published to the Alpine
repository.
Both images now run `apk upgrade` before `apk add`. Packages from `apk
add` are already current, so this only affects the preinstalled base
packages.
_Written by Claude (Opus 5.5)._
Co-authored-by: silverwind <me@silverwind.io>
Backport #40237 by @Shadow0Bit
Fixes#40230
Cookbooks without dependencies are served with `"dependencies": null`
from the universe and version metadata endpoints. This happens because
`omitempty` drops the empty map when the metadata is stored. Chef
clients iterate over this value and Policyfile resolution crashes.
Supermarket returns `{}` for these cookbooks, Gitea now does too. The
fix checks whether dependencies are `nil` while building response,
returns an empty map if true. This way cookbooks that are already stored
are covered without a re-upload.
Tested with the Chef integration test and `chef-cli install`.
Assisted by Claude Code (claude-opus-5-5).
Co-authored-by: rafał przypek <58443512+Shadow0Bit@users.noreply.github.com>
Backport #40207 by @silverwind
The issue reference popup built its `/info` request from the raw link
path, so a link ending in `/owner/repo/issues/N` queried an unrelated
location.
It now builds the request from the owner, repo, type and index parsed
from the link, with `parseIssueHref` anchored to the pathname under
`appSubUrl`.
Co-authored-by: silverwind <me@silverwind.io>
Backport #40126 by @silverwind
The job emitter resolves a dependency chain in several rounds in memory,
but a job's `if:` reads its needs' results from the database, where a
need skipped or failed earlier in the same pass was still blocked. A job
with `if: always() && needs.build.result == 'skipped'` was therefore
skipped instead of started.
Finished statuses are now persisted at the end of each round, which also
replaces the separate handling of skipped and failed reusable workflow
callers.
Co-authored-by: silverwind <me@silverwind.io>
On release/v28, `PATCH /repos/{owner}/{repo}/milestones/{id}` with
`"due_on": null`, the documented way to remove a deadline, returns HTTP
500. The backport https://github.com/go-gitea/gitea/pull/39921 changed
`EditMilestoneOption.Deadline` to `optional.Option[*string]`, which
decodes `null` as a slice holding a nil pointer, and the go-chi/binding
validator panics on that element (`reflect: call of
reflect.Value.Interface on zero Value`).
The field has no binding rules, so tag it `form:"-"` to skip validation.
main is unaffected since https://github.com/go-gitea/gitea/pull/39528
moved the binding into Gitea.
Backport #40166 by @perfectra1n
Concurrent `notification-service` workers can deadlock in the
transaction for-loop.
The transaction is useless, so it is removed to avoid deadlock.
Co-authored-by: Jon Fuller <jonfuller2012@gmail.com>
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
Backport #40167 by @perfectra1n
The `repo_stats_update` queue only deduplicates waiting items, so
several workers can index the same repository at once, then
`UpdateLanguageStats` can lead to transaction deadlock.
Use a global lock to avoid concurrent updates, and also helps
https://github.com/go-gitea/gitea/pull/39342 to avoid duplicate updates.
Co-authored-by: Jon Fuller <jonfuller2012@gmail.com>
Co-authored-by: silverwind <me@silverwind.io>
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
Backport https://github.com/go-gitea/gitea/pull/40127
The starred and watched repository API lists converted repositories
without loading their attributes, so `language` was always empty. They
are now loaded in one batch, owners included.
Three of these handlers also continued after an internal error and wrote
a second response body, they now return.
Backport #40130 by @bircni
Requiring branches to be up to date reruns the full CI matrix on every
waiting PR after each merge, which saturates our 20 concurrent jobs. A
merge queue still tests against the latest `main`, but once per batch.
This runs the PR workflows on `merge_group` and reports `giteabot/lgtm`
on queue commits, since GiteaBot only sets it on PR heads. Mergers queue
PRs with "Merge when ready" instead of `reviewed/wait-merge`, so
giteabot no longer updates branches. Nothing changes until the
`merge-queue` ruleset is enabled and "Require branches to be up to date"
is turned off. Backport to `release/v28` before the ruleset covers it.
Co-authored-by: bircni <bircni@icloud.com>
Co-authored-by: silverwind <me@silverwind.io>
Backport #40125 by @silverwind
A failed contributor stats generation never released its lock, so after
the cached error expired, the contributors and code frequency pages kept
saying stats are being generated until a restart. A generation
outlasting the 5 second wait also leaked its goroutine on an unread
channel.
The lock is now claimed atomically and released together with the
completion signal when generation ends, and a cache miss after the wait
counts as still generating.
Co-authored-by: silverwind <me@silverwind.io>
Backport #40060 by @soujasK
Fixes https://github.com/go-gitea/gitea/issues/40056
The admin web UI lists every account type except organizations, but
`gitea admin user list` and `GET /api/v1/admin/users` only returned
individual users. That hid bots created for CI, even though the CLI and
the admin API can both create and manage them.
- The admin web UI, the admin API and the CLI share one type list:
individual, reserved, bot and remote accounts.
- `gitea admin user list` (and `--admin`) gains a `Type` column showing
`User`, `Bot`, `Reserved` or `Remote`.
Signed-off-by: Soujas K <kudchadkarsoujas@gmail.com>
Co-authored-by: soujasK <121026960+soujasK@users.noreply.github.com>
Co-authored-by: silverwind <me@silverwind.io>
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
Backport https://github.com/go-gitea/gitea/pull/40121
With `[oauth2_client].ACCOUNT_LINKING = auto`, the link account sign-up
form linked the external identity to an existing account matching the
submitted user name or email and signed in as it. Automatic linking now
only happens when the existing user matches the external identity,
otherwise the linking login page asks for the account's password.
`release/v28` lacks `TestOAuth2AvatarFromPicture`, so only the new test
is added.
Co-authored-by: bircni <bircni@icloud.com>
Co-authored-by: Giteabot <teabot@gitea.io>
Backport #39930 by @lunny
Pages showing commit statuses loaded the repository of every status
separately, so a PR with ~75 statuses ran ~75 identical repository
queries per merge box render (about 3.4s in a production trace).
`statusRepoCanReadActions` only loaded the repository inside the
permission cache loader, which never runs once `RepoAssignment` has
published the permission.
- `CommitStatus.loadRepository` resolves through the request context
cache
- `statusRepoCanReadActions` loads the repository before the permission
lookup
- `RepoAssignment` publishes its repository to the context cache next to
the permission
- `GetCommitActionsStatusMap` drops its own repository map
Queries per request with 50 statuses (45 Actions, 5 external) on the
head commit, SQLite:
| Page | Repository queries | Total queries |
|---|---|---|
| PR view | 61 → 11 | 157 → 107 |
| Commit | 6 → 5 | 24 → 23 |
| Branches | 6 → 5 | 33 → 32 |
| Commits | 6 → 5 | 23 → 22 |
| Repo home | 8 → 7 | 35 → 34 |
| `/pulls` | 15 → 15 | 37 → 37 |
Co-authored-by: Lunny Xiao <xiaolunwen@gmail.com>
Co-authored-by: silverwind <me@silverwind.io>
Backport #40114 by @silverwind
The repository issue and pull request lists filter `access` by
`repo_id`, which no index covers, so every load scans the whole table.
Issue counts on larger repositories merge three single-column indices,
and the API pull request list loads the latest reviews with `id IN
(SELECT max(id) ... GROUP BY ...)`, which MySQL runs as a scan of every
review, about 1.1s on
https://gitea.com/api/v1/repos/gitea/gitea-mcp/pulls?limit=1.
- Index `access.repo_id`
- Index `team_repo.repo_id`
- Replace the single `repo_id`, `is_closed` and `is_pull` indices with
one `issue` index on `(repo_id, is_closed, is_pull)`
- Join the latest review ids instead of `id IN (subquery)`
Median times in ms on MySQL 8.4 with tables sized like gitea.com (1M
access rows, 500k issues, 200k reviews), for a copy of
https://gitea.com/gitea/gitea-mcp and a repository with 9.7k issues:
| Page | Before | After |
|---|---|---|
| `/{owner}/{repo}/issues` | 107 | 40 |
| `/{owner}/{repo}/pulls` | 106 | 32 |
| `/{owner}/{repo}/pulls?state=closed` | 153 | 35 |
| `/{owner}/{repo}/issues/{index}` | 141 | 60 |
| `/{owner}/{repo}/issues`, 9.7k issues | 303 | 53 |
| `/api/v1/repos/{owner}/{repo}/pulls?limit=50` | 180 | 46 |
| `/api/v1/repos/{owner}/{repo}/pulls?limit=1` | 137 | 16 |
| `/api/v1/repos/{owner}/{repo}/issues`, 9.7k issues | 248 | 43 |
Co-authored-by: silverwind <me@silverwind.io>
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
Backport #40058 by @silverwind
A review's text is stored on both the review and its timeline comment,
but editing it in the web UI or through the issue comment API only
updated the comment. The reviews API kept returning the old body
afterwards.
Extracted out of https://github.com/go-gitea/gitea/pull/39865.
*Written by Claude.*
Co-authored-by: silverwind <me@silverwind.io>
Backport #40016 by @basavaraj-sm05
Project board cards list the pull requests that reference their issue,
but loaded them without checking the viewer's access. A private pull
request mentioning a public issue showed its title, number and state on
that issue's board to anyone who can view the board, anonymous visitors
included.
Repository and organization boards now skip linked pull requests in
repositories where the viewer cannot read pull requests, matching how
the issue page already hides such references.
Co-authored-by: Basavaraj S m <basavaraj@digiscrypt.com>
Co-authored-by: silverwind <me@silverwind.io>
Co-authored-by: bircni <bircni@icloud.com>
Backport https://github.com/go-gitea/gitea/pull/39945
API tokens and Actions task tokens could see data beyond their scope
through user-scoped API routes and web profile pages.
- Actions task tokens authenticate as a shared system user with a
negative ID, which org, project and SSH/GPG key filters treated as "no
owner filter". These filters now check `!= 0`, so system users own
nothing. The `/api/v1/user` routes except `GET /api/v1/user` only accept
individual and bot accounts, and repo search no longer skips access
checks for the owner.
- `/{username}` only accepts tokens for RSS/Atom feeds. HTML profile and
org pages render as anonymous for token requests.
- User and org feeds require `read:user` or `read:organization` from
tokens, the same way repo feeds require `read:repository`.
`release/v28` lacks https://github.com/go-gitea/gitea/pull/39967, so
`/{username}` enables basic auth only for feed requests through a
middleware. The profile page token tests from
https://github.com/go-gitea/gitea/pull/39853 are folded into
`TestFeedUser`, because profile pages no longer accept tokens.
Co-authored-by: bircni <bircni@icloud.com>
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
Co-authored-by: Giteabot <teabot@gitea.io>
Backport https://github.com/go-gitea/gitea/pull/40106
Repository-level and instance-level runners still picked up waiting jobs
of repositories with Actions disabled.
https://github.com/go-gitea/gitea/pull/30331 only covered owner-level
runners, by filtering them out in SQL.
Pickup now checks the Actions unit of each job it is about to claim, for
every runner scope, and cancels the job instead when Actions is
disabled, also honoring `DISABLED_REPO_UNITS`. Disabling Actions
therefore cancels its queued jobs once a matching runner polls, instead
of leaving them waiting until `ABANDONED_JOB_TIMEOUT`.
Co-authored-by: silverwind <me@silverwind.io>