Fix the binary layout of ProcessThreadImpl.

We apparently hit an obscure problem on mac where seemingly an unaligned mutex causes memory corruption.
The effect was that the |modules_| list became corrupt and we crashed.  At this point I'm not exactly
sure what the alignment requirements are but for now, I've fixed up the layout in a way that doesn't cause these same issues.

I'm also changing auto->proper type at the request of drive by reviewers from my previous cl in the same file.

TBR=pbos@webrtc.org

Review URL: https://webrtc-codereview.appspot.com/38989004

Cr-Commit-Position: refs/heads/master@{#8286}
git-svn-id: http://webrtc.googlecode.com/svn/trunk@8286 4adac7df-926f-26a2-2b94-8c16560cd09d
This commit is contained in:
tommi@webrtc.org
2015-02-08 00:48:10 +00:00
parent ec499beaf5
commit 103f3289b5
2 changed files with 37 additions and 19 deletions

View File

@ -32,29 +32,44 @@ class ProcessThreadImpl : public ProcessThread {
void WakeUp(Module* module) override;
int32_t RegisterModule(Module* module);
int32_t DeRegisterModule(const Module* module);
int32_t RegisterModule(Module* module) override;
int32_t DeRegisterModule(const Module* module) override;
protected:
static bool Run(void* obj);
bool Process();
private:
rtc::ThreadChecker thread_checker_;
const rtc::scoped_ptr<EventWrapper> wake_up_;
rtc::scoped_ptr<ThreadWrapper> thread_;
struct ModuleCallback {
ModuleCallback() : module(nullptr), next_callback(0) {}
ModuleCallback(const ModuleCallback& cb)
: module(cb.module), next_callback(cb.next_callback) {}
ModuleCallback(Module* module) : module(module), next_callback(0) {}
bool operator==(const ModuleCallback& cb) const {
return cb.module == module;
}
Module* const module;
int64_t next_callback; // Absolute timestamp.
private:
ModuleCallback& operator=(ModuleCallback&);
};
rtc::CriticalSection lock_; // Used to guard modules_ and stop_.
typedef std::list<ModuleCallback> ModuleList;
// Warning: For some reason, if |lock_| comes immediately before |modules_|
// with the current class layout, we will start to have mysterious crashes
// on Mac 10.9 debug. I (Tommi) suspect we're hitting some obscure alignemnt
// issues, but I haven't figured out what they are, if there are alignment
// requirements for mutexes on Mac or if there's something else to it.
// So be careful with changing the layout.
rtc::CriticalSection lock_; // Used to guard modules_ and stop_.
rtc::ThreadChecker thread_checker_;
const rtc::scoped_ptr<EventWrapper> wake_up_;
rtc::scoped_ptr<ThreadWrapper> thread_;
ModuleList modules_;
bool stop_;
};