|
how is the thread suspended?
Is it through an external called of Suspend() method or by waiting for a handle through WaitForSingleObject, WaitForMultipleObjects,MsgWaitForMultipleObjects functions?
|
|
|
|
|
Extermal Suspend()
|
|
|
|
|
Hi all,
I have a windowless application that has a simple main() function with well-defined starting and ending points.
One of my customers claims that when this application is running within the SYSTEM account, he can see it "hang" in memory when execution should actually be complete and should terminate. The same program however, when running within a "regular" user account, runs fine and exits as expected. I don't doubt his claim, but I haven't been able to reproduce this particular problem.
To put it mildly, I'm stumped. I'm at a loss for things to look for or even specific questions to ask the customer.
I'm willing to entertain any sort of speculation.
|
|
|
|
|
How is he running it as SYSTEM? Is this a service?
|
|
|
|
|
> How is he running it as SYSTEM?
"SYSTEM", as in, when you look at the process in Task Manager, the User Name shows SYSTEM instead of any "regular" user profile.
The customer runs my program through SMS, which, apparently, runs within the SYSTEM context (I'm not all that familiar with it). In any case, he says he can reproduce the setup like this (so SMS shouldn't really be a factor):
- From the command line, run "at 00:00 /INTERACTIVE cmd" (without the quotes)
- This sets up a new scheduled task at midnight, which will simply launch CMD
- Navigate to the Scheduled Tasks folder in Control Panel. Right-click on the new task (at1), select Run
- This creates an instance of CMD running within the SYSTEM account, since that's how Task Scheduler runs everything (you can confirm this with Task Manager)
- He then runs my app from within this console window. It's also running within the SYSTEM context (I can confirm this behavior so far).
At this point, he says my program never shuts down. It does, however, on all the machines I tried this on.
I'm still trying to get some basic information out of the customer, but in the meantime, I'd like to hear some feedback so I can further experiment on my own...
|
|
|
|
|
First guess is that it's popping up a message box, and waiting for you to click it, except the message box is on a "hidden" desktop, so it'll wait forever.
If he allows the Scheduler service to interact with the desktop, and runs it via the scheduler (ie using the at command as you describe), you might see the messagebox (if there is one)
|
|
|
|
|
> First guess is that it's popping up a message box, and waiting for you to click it,
I thought about that as well, only if this was the case, he'd be seeing that message when NOT running as SYSTEM...there's nothing in the program that would try to display a message box when running as SYSTEM but not as a regular user.
Besides, with the /INTERACTIVE switch specified for the scheduled job, it *should* already be able to display a message box...
|
|
|
|
|
it will let me do debug (F5) and create report and every will be fine, but if I click "execute program (Ctrl +F5)"...It will have "application error, the instruction at "0v00490c3c " referenced memory at "0x000000cc". The memory could not be "read", click ok to terminate the program or cancle to debug"
but in release version, everything works, F5 or Ctrl + F5, could anyone tell me what to look at this error?
Thank you!
|
|
|
|
|
I would just fix the bug.
8bc7c0ec02c0e404c0cc0680f7018827ebee
|
|
|
|
|
Thanks for your reply, I am trying to fix the bug, i think my question is what's the difference between execute and go?
thank you
|
|
|
|
|
Do you have any program arguments in the debug settings? That could make a difference.
In any case, you should be able to break into the debugger after the exception and take a look at the call stack, variable values, etc., so as to determine where and why the program is producing an access violation...
--
jlr
http://jlamas.blogspot.com/[^]
|
|
|
|
|
Hi,
I'm not sure but I think the main difference is that when pressing F5 the application will be loaded into the debugger. Which on his turn allocates memory for the programma and fills it up (the not initialize) with 0xCC. When pressing CTRL + F5 the application is started like a normal program without the extra byte filling. This can lead to different behaviuor when there are unintialized variables that are use directly.
Hope that this clarifies it a bit
Best regards,
Kurt Pattyn
codito ergo sum
|
|
|
|
|
When I lauch my C++ program, my CPU is only used at 50%. While watching my CU history, I can see that there is 2 graphics. All the other system that I used to work with had only one graphic showing the CPU. I don't remember having 2 CPU in my system...
My program (DLL) lauches a simulation program and is the main process on my computer. In other words, I'm using my computer only for that task, start DLL, launch a simulation program and calculate...I want full CPU usage for that task.
How can I use 100% of my CPU while lauching ma C++ program? I presume that my program would go faster!
My system is a Pentium(R) 4 CPU 3.2 GHz
Thanks
|
|
|
|
|
If you're checking the CPU Usage in the task manager and it's showing 50%, you're using way too much of your computer, in all likelihood.. you should try putting a few discreet sleep calls into your code to allow some process switching, it sounds like your program is hogging a lot of your computer resources. Unless you're writing a game or something that you intend to be used without any other programs running, 50% is a huge amount of processor power for one program to use.
|
|
|
|
|
The program calls a simulation program. It's the simulator that takes 50% of the CPU, I want 100% to accelerate the process. Every process uses 50% max of the CPU. If a start my program twice, the simulator is open twice and only then may I be able to use 100% of my CPU. My systems is working like if I had two CPU...but it's not the case.
|
|
|
|
|
Hi,
The computer you are working with is probably working with a Hyperthreading enabled processer. This type of processors has 2 sets of pips, 2 sets of registers. This reflects in a behaviour that with the taskmanager you will see, like there are two processors.
Because of this it is inpossible to make a 'single threaded program' use 100% of CPU. There will always be 1 set free for the other programs. This results in a more stable OS. Even if 1 program hangs and uses 50% (in old terms 100% CPU) the rest of the OS is still active.
Normaly you could check if the processer is Hyperthreaded. You could check the Intel Logo on your computer it should mention HT in it.
Hope that I was clear enough
Regards
Kurt Pattyn
codito ergo sum
|
|
|
|
|
I have HT on the logo. thanks for that clarification. I will look on the thread command to use 100% of my system.
If I'm usign 50%, do I calculate at 50%, twice the time that if I used 100%??
Thanks
|
|
|
|
|
It depends on your code and your processor. A hyperthreaded processor duplicates some of the processor internals, but not all of them so it doesn't provide the performance boost of dual processors or even dual core processors.
When you look at using multiple threads, the biggest consideration is how you share data between the threads and usually requires a change in algorithmic approach. Performance is usually best when you can break your calculation into completely separate operations of signifigant size. If you can't do that, then a single threaded approach might even be faster in most cases.
|
|
|
|
|
hi, when i run on on of my application in visual studio, it takes up to 100% of the usage sometime, does it mean the exe might just crashed?
|
|
|
|
|
No, that's normal. (Anonymous is, erm, wrong)
Please also see my other reply
Pandoras Gift #44: Hope. The one that keeps you on suffering.
aber.. "Wie gesagt, der Scheiss is' Therapie"
boost your code || Fold With Us! || sighist | doxygen
|
|
|
|
|
That is not right.
Even a low priority program will get 100% of CPU power when no other program is running.
Reasons why a program does not use 100% of CPU:
- it is waiting for user input
- it is waiting for external hardware (e.g. hard drive)
- It "gives up" CPU cycles (Sleep(), WaitFor* etc.) for one or another reason
The first two are the most common why on a modern system you rarely reach 100% CPU power e.g. with office.
Pandoras Gift #44: Hope. The one that keeps you on suffering.
aber.. "Wie gesagt, der Scheiss is' Therapie"
boost your code || Fold With Us! || sighist | doxygen
|
|
|
|
|
You are on a dual processor machine. The only way to use 100% would to have two threads RUNNING INDEPENDENTLY.
Tim Smith
I'm going to patent thought. I have yet to see any prior art.
|
|
|
|
|
Your CPU has hyper-threading which makes it show up as a dual processor system. In order to take full advantage of the CPU, you need to use mutiple threads in your app, which is usually much more difficult, unless approach is naturally parallelizable.
|
|
|
|
|
j_s_lacroix wrote:
I can see that there is 2 graphics
Pentium 4 processors have Hyperthreading (try google). The OS sees that as two CPUs.
Do you have 100% on one, 50% on one, or 50% on both?
100% on one:
The program is single threaded, and uses one CPU to the maximum. Only a multithreaded program can emplo more than one CPU.
50% one one (or both):
Two possibilities: (A) the task is regulary giving up CPU power (e.g. by Sleep(), or WaitFor*() for thread synchronization). (B) some other component such as File I/O is slowing you down.
if the simulation heavily works with external files, or does graphics, it is probably (B)
Pandoras Gift #44: Hope. The one that keeps you on suffering.
aber.. "Wie gesagt, der Scheiss is' Therapie"
boost your code || Fold With Us! || sighist | doxygen
|
|
|
|
|
the program will works fine on small set of data just fine......but in large data, the dialog window will come up and with just 0s then started processed after 5 minutes.
this only happened on release version, debug version works fine no matter what size of data.
could anyone tell me what to look? I have generated the debug info on release version, but it's really hard to tell where went wrong, because it didn't really crash, it's just slow....
if I turn off the optimazation, it will be even slower -- take 10 minutes to start!
any info will be greatly appreciated!
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
