Allow to drop existing fd cache of cgroup path
A process can give up the permission to set cgroup. If we still
keep the fd that was cached before losing the permission, when
the process sets scheduling group, it will write to the cached
fd without checking if is accessible and lead to sepolicy denied.
Bug: 123043091
Test: Build and boot.
Test: A new process from zygote set cgroup and drop fd cache, and
then specializes to app domain. There is no sepolicy denied
when the process creates new thread.
(android::thread_data_t::trampoline)
Change-Id: I285ee91424ea965ea9c670fc0f6662948e3e2ce5
diff --git a/libprocessgroup/task_profiles.cpp b/libprocessgroup/task_profiles.cpp
index edc316a..aee5f0c 100644
--- a/libprocessgroup/task_profiles.cpp
+++ b/libprocessgroup/task_profiles.cpp
@@ -173,6 +173,15 @@
fd_ = std::move(fd);
}
+void SetCgroupAction::DropResourceCaching() {
+ std::lock_guard<std::mutex> lock(fd_mutex_);
+ if (fd_ == FDS_NOT_CACHED) {
+ return;
+ }
+
+ fd_.reset(FDS_NOT_CACHED);
+}
+
bool SetCgroupAction::AddTidToCgroup(int tid, int fd) {
if (tid <= 0) {
return true;
@@ -292,6 +301,24 @@
res_cached_ = true;
}
+void TaskProfile::DropResourceCaching() {
+ if (!res_cached_) {
+ return;
+ }
+
+ for (auto& element : elements_) {
+ element->DropResourceCaching();
+ }
+
+ res_cached_ = false;
+}
+
+void TaskProfiles::DropResourceCaching() const {
+ for (auto& iter : profiles_) {
+ iter.second->DropResourceCaching();
+ }
+}
+
TaskProfiles& TaskProfiles::GetInstance() {
// Deliberately leak this object to avoid a race between destruction on
// process exit and concurrent access from another thread.