inline, optimize, and call stack visibility?

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

inline, optimize, and call stack visibility?

Subject: inline, optimize, and call stack visibility?
From: "David M. Cotter" <email@hidden>
Date: Tue, 25 Jan 2011 15:01:01 -0800

In my app, if it crashes on a user, it generates a crash report which gets sent to me.

i have a class that's created on different threads

i want to differentiate between each thread, so i can READ which thread it is by looking at the call stack

in my debug version, the thread stacks look like this:

thread 3

3 com.kjams.kjams 0x001095d4 CSemaphore::wait(long) + 1288

4 com.kjams.kjams 0x00287835 CMutex_ThreadQue::CB_ActualRunQue() + 471

5 com.kjams.kjams 0x00237d71 CMutex_ThreadQue_PushMeta::CB_RunQue() + 17

6 com.kjams.kjams 0x0028547d CMutex_ThreadQue::CB_S_RunQue(void*, long) + 17

7 com.kjams.kjams 0x00286617 CThreads::CB_Fork(APP_ForkInfo*) + 703

8 com.kjams.kjams 0x00286a71 CB_S_Fork_Preemptive(void*) + 25

9 com.kjams.kjams 0x00289ccd boost::detail::thread_data<CFork_Preemptive>::run() + 23

10 libboost_thread.dylib 0x007bca3a thread_proxy + 138

11 libSystem.B.dylib 0x95a6b85d _pthread_start + 345

12 libSystem.B.dylib 0x95a6b6e2 thread_start + 34

thread 5

3 com.kjams.kjams 0x001095d4 CSemaphore::wait(long) + 1288

4 com.kjams.kjams 0x00287835 CMutex_ThreadQue::CB_ActualRunQue() + 471

5 com.kjams.kjams 0x00237d85 CMutex_ThreadQue_DiscEvent::CB_RunQue() + 17

6 com.kjams.kjams 0x0028547d CMutex_ThreadQue::CB_S_RunQue(void*, long) + 17

7 com.kjams.kjams 0x00286617 CThreads::CB_Fork(APP_ForkInfo*) + 703

8 com.kjams.kjams 0x00286a71 CB_S_Fork_Preemptive(void*) + 25

9 com.kjams.kjams 0x00289ccd boost::detail::thread_data<CFork_Preemptive>::run() + 23

10 libboost_thread.dylib 0x007bca3a thread_proxy + 138

11 libSystem.B.dylib 0x95a6b85d _pthread_start + 345

12 libSystem.B.dylib 0x95a6b6e2 thread_start + 34

see how you can see the difference?

but in the release version, that is optimized out! looks like this:

thread 3

4 com.kjams.kjams 0x001095d4 CSemaphore::wait(long) + 1288

5 com.kjams.kjams 0x00287835 CMutex_ThreadQue::CB_ActualRunQue() + 471

6 com.kjams.kjams 0x0028547d CMutex_ThreadQue::CB_S_RunQue(void*, long) + 17

7 com.kjams.kjams 0x00286617 CThreads::CB_Fork(APP_ForkInfo*) + 703

8 com.kjams.kjams 0x00286a71 CB_S_Fork_Preemptive(void*) + 25

9 com.kjams.kjams 0x00289ccd boost::detail::thread_data<CFork_Preemptive>::run() + 23

10 libboost_thread.dylib 0x007bca3a thread_proxy + 138

11 libSystem.B.dylib 0x95a6b85d _pthread_start + 345

12 libSystem.B.dylib 0x95a6b6e2 thread_start + 34

thread 5

4 com.kjams.kjams 0x001095d4 CSemaphore::wait(long) + 1288

5 com.kjams.kjams 0x00287835 CMutex_ThreadQue::CB_ActualRunQue() + 471

6 com.kjams.kjams 0x0028547d CMutex_ThreadQue::CB_S_RunQue(void*, long) + 17

7 com.kjams.kjams 0x00286617 CThreads::CB_Fork(APP_ForkInfo*) + 703

8 com.kjams.kjams 0x00286a71 CB_S_Fork_Preemptive(void*) + 25

9 com.kjams.kjams 0x00289ccd boost::detail::thread_data<CFork_Preemptive>::run() + 23

10 libboost_thread.dylib 0x007bca3a thread_proxy + 138

11 libSystem.B.dylib 0x95a6b85d _pthread_start + 345

12 libSystem.B.dylib 0x95a6b6e2 thread_start + 34

looks the same. :( so i tried really hard to force it to NOT inline:

class CMutex_ThreadQue_PushMeta : public CMutex_ThreadQue {

#pragma no_inline

void CB_RunQue() __attribute__ ((noinline)) { CB_ActualRunQue(); }

#pragma inline

public:

CMutex_ThreadQue_PushMeta() : CMutex_ThreadQue("Push Meta Que", false, true) {}

};

i also tried turning off "inline methods hidden"

all TO NO AVAIL! so maybe this isn't an inlining issue?

so then i tried turning down the optimizations:

started with:

-Os - fail

-O3 - fail

-O2 - fail

-O1 - WIN!!

so, i have to dial down all the way to -O1?

this is for my release app, so i'd definitely rather not do that.

is there any way i can keep "-Os" and still see that function name in the call stack of each thread?

 _______________________________________________
Do not post admin requests to the list. They will be ignored.
Xcode-users mailing list      (email@hidden)
Help/Unsubscribe/Update your Subscription:

This email sent to email@hidden

Follow-Ups:
- Re: inline, optimize, and call stack visibility?
  - From: "Sean McBride" <email@hidden>
- Re: inline, optimize, and call stack visibility?
  - From: Wim Lewis <email@hidden>

Prev by Date: Re: no previous prototype for function that is defined in header?
Next by Date: Re: Why doesn't NSMakePoint work from XCode/gdb?
Previous by thread: Re: Why doesn't NSMakePoint work from XCode/gdb?
Next by thread: Re: inline, optimize, and call stack visibility?
Index(es):
- Date
- Thread