Release DACL after the named pipe was created

Release DACL after the named pipe was created

[Oleksandr, 2014-07-01 16:52:35]

This isn't pretty, but it's highly specialized case, and I don't really think we
should come up with anything fancier (like dedicated class for our own
SECURITY_DESCRIPTOR)

[Felix Dahlke, 2014-07-01 20:38:58]

On 2014/07/01 16:52:35, Oleksandr wrote:
> This isn't pretty, but it's highly specialized case, and I don't really think
we
> should come up with anything fancier (like dedicated class for our own
> SECURITY_DESCRIPTOR)

Does this fix any particular issue?

[Oleksandr, 2014-07-01 22:06:23]

Yes, named pipe creation fails quite often without this. It would say that ACL
in incorrect. Which is understandable, since we were releasing it before
actually using it. The fact that that memory was not usually overwritten yet by
that time was purely luck.

 
On 2014/07/01 20:38:58, Felix H. Dahlke wrote:
> On 2014/07/01 16:52:35, Oleksandr wrote:
> > This isn't pretty, but it's highly specialized case, and I don't really
think
> we
> > should come up with anything fancier (like dedicated class for our own
> > SECURITY_DESCRIPTOR)
> 
> Does this fix any particular issue?

[Oleksandr, 2014-07-01 22:08:17]

As for the specific issue, I've caught this while investigating this
https://issues.adblockplus.org/ticket/579. It doesn't fix that issue though. 

On 2014/07/01 22:06:23, Oleksandr wrote:
> Yes, named pipe creation fails quite often without this. It would say that ACL
> in incorrect. Which is understandable, since we were releasing it before
> actually using it. The fact that that memory was not usually overwritten yet
by
> that time was purely luck.
> 
>  
> On 2014/07/01 20:38:58, Felix H. Dahlke wrote:
> > On 2014/07/01 16:52:35, Oleksandr wrote:
> > > This isn't pretty, but it's highly specialized case, and I don't really
> think
> > we
> > > should come up with anything fancier (like dedicated class for our own
> > > SECURITY_DESCRIPTOR)
> > 
> > Does this fix any particular issue?