Summary
An authenticated backend user who can access one job can request an attachment identifier containing ../ segments and make the job attachment download endpoint read a file from another job directory inside var/job-attachments.
The controller authorizes only the jobUuid route parameter. The later attachment lookup joins that authorized job UUID with the attacker-controlled identifier, then passes the combined path to the virtual filesystem. VirtualFilesystem::resolve() canonicalizes the whole path and only rejects paths that escape the filesystem mount, so authorized-job/../victim-job/debug_log.csv becomes victim-job/debug_log.csv.
This is a cross-job authorization bypass for known job attachment paths. It is not a practical brute-force against unknown jobs because job directories are UUID v4 values.
Root Cause
JobsController::downloadJobAttachment() checks access to the route jobUuid before loading the attachment:
$job = $this->jobs->getByUuid($jobUuid);
if (!$job || !$this->jobs->hasAccess($job)) {
throw $this->createNotFoundException();
}
$attachment = $this->jobs->getAttachment($jobUuid, $identifier);Jobs::getAttachment() then resolves a path built from the authorized job UUID and the attacker-controlled identifier:
$fileItem = $this->jobAttachmentsStorage->get($this->getAttachmentIdentifier($job, $identifier));
return $job->getUuid().'/'.$identifier;
VirtualFilesystem::resolve() canonicalizes the combined path. It rejects absolute paths and paths that start with .., but it does not preserve the authorized job directory as a boundary:
$path = Path::canonicalize($location);
if (str_starts_with($path, '..')) {
throw new \OutOfBoundsException(...);
}
return Path::join($this->prefix, $path);Therefore:
<authorized-job>/../<victim-job>/debug_log.csv
canonicalizes to:
<victim-job>/debug_log.csv
which remains inside the job-attachments filesystem mount and is accepted.
Recommended Fix
Treat the attachment identifier as a filename, not a path:
- Reject
/, \, NUL, and dot-segment components in identifier.
- Add a route requirement that prevents slashes in
{identifier} if nested attachment paths are not intended.
- After resolving, assert the canonical relative path starts with
<authorized-job-uuid>/ before returning a FilesystemItem.
- Apply the same identifier validation in
Jobs::addAttachment() so future producers/extensions cannot write outside the owning job directory.
Impact
A low-privileged backend user can read another job's attachment if they know or obtain the target job UUID and attachment filename. Built-in crawler jobs attach CSV logs such as debug_log.csv, broken-link-checker_log.csv, and search-index_log.csv, which can contain crawled URLs, referring URLs, tags, and error messages.
References
Summary
An authenticated backend user who can access one job can request an attachment identifier containing
../segments and make the job attachment download endpoint read a file from another job directory insidevar/job-attachments.The controller authorizes only the
jobUuidroute parameter. The later attachment lookup joins that authorized job UUID with the attacker-controlledidentifier, then passes the combined path to the virtual filesystem.VirtualFilesystem::resolve()canonicalizes the whole path and only rejects paths that escape the filesystem mount, soauthorized-job/../victim-job/debug_log.csvbecomesvictim-job/debug_log.csv.This is a cross-job authorization bypass for known job attachment paths. It is not a practical brute-force against unknown jobs because job directories are UUID v4 values.
Root Cause
JobsController::downloadJobAttachment()checks access to the routejobUuidbefore loading the attachment:Jobs::getAttachment()then resolves a path built from the authorized job UUID and the attacker-controlled identifier:VirtualFilesystem::resolve()canonicalizes the combined path. It rejects absolute paths and paths that start with.., but it does not preserve the authorized job directory as a boundary:Therefore:
canonicalizes to:
which remains inside the
job-attachmentsfilesystem mount and is accepted.Recommended Fix
Treat the attachment identifier as a filename, not a path:
/,\, NUL, and dot-segment components inidentifier.{identifier}if nested attachment paths are not intended.<authorized-job-uuid>/before returning aFilesystemItem.Jobs::addAttachment()so future producers/extensions cannot write outside the owning job directory.Impact
A low-privileged backend user can read another job's attachment if they know or obtain the target job UUID and attachment filename. Built-in crawler jobs attach CSV logs such as
debug_log.csv,broken-link-checker_log.csv, andsearch-index_log.csv, which can contain crawled URLs, referring URLs, tags, and error messages.References