Re: [ros-svn] [arty] 13483: Patch to fix NtCreateSempahore, in the case where the initial lookup

[Thomas Weidenmueller]

[e-mail] wrote:
> Patch to fix NtCreateSempahore, in the case where the initial lookup
> succeeds.  We previously left the function without initializing 
> hSemaphore.  Patch suggested by me and executed by hpoussin.

To me it looks totally wrong. First of all, the handle in my tests is 
never written if the function fails (just create a testcase where your 
application tries to create thousands of semaphore objects to reproduce 
it). Second of all you initialize the semaphore object only if creating 
it failed, as in you initialize a semaphore object on a NULL pointer. 
Third of all you insert a handle to the object no matter if you 
previously could create it, so you would end up creating a handle to a 
null-pointer.

The previous version I wrote just had one small bug, it checked if 
ObCreateObject failed to continue initialization and handle creation 
instead of checking if it succeeded. I'm going to revert this version of 
NtCreateSemaphore as it introduces even more bugs and adds behavior 
neither documented nor reproducable. I guess in case an application 
failed due to this function it was because of the fact that the function 
succeeded and didn't return a handle to a semaphore object even though 
one was created.

Best Regards
Thomas

[art yerkes]

On Thu, 10 Feb 2005 10:18:05 +0100
Thomas Weidenmueller [e-mail] wrote:
> The previous version I wrote just had one small bug, it checked if 
> ObCreateObject failed to continue initialization and handle creation 
> instead of checking if it succeeded. I'm going to revert this version of 
> NtCreateSemaphore as it introduces even more bugs and adds behavior 
> neither documented nor reproducable. I guess in case an application 
> failed due to this function it was because of the fact that the function 
> succeeded and didn't return a handle to a semaphore object even though 
> one was created.

Well it is a bit unkind to make a new semaphore and not return the handle
to it.  Thanks for fixing it the right way.
-- 
Here's a simple experiment. Stand on a train track between two locomotives
which are pushing on you with equal force in opposite directions. You will
exhibit no net motion. None the less, you may soon begin to notice that
something important is happening.
-- Robert Stirniman