Simplify ResolveNTFunctionPtr (temporary)...

Simplify ResolveNTFunctionPtr (temporary)

I want to test the theory that the issues we are observing here
are actually a race condition.
- The race condtion would be related with 2 operations that
are not thread safe:
1- check/creation of the map
2- search/insert on the map

I would like to air this CL on dev channel for a week and
observe the crash rate.


BUG=11789
Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=25741

[cpu_(ooo_6.6-7.5), 2009-09-09 00:22:02]

Lots of crashes right now.

[rvargas (doing something else), 2009-09-09 00:29:51]

OK

[Nicolas Sylvain, 2009-09-09 03:30:57]

Don't think this is going to change anything, see comment inline. The problem is
not with the map, because it's never set when the crash happens. (i.e. it's the
first call for the given function).

http://codereview.chromium.org/199052/diff/4001/5001
File sandbox/src/win_utils.cc (right):

http://codereview.chromium.org/199052/diff/4001/5001#newcode203
Line 203: static HMODULE ntdll = ::GetModuleHandle(sandbox::kNtdllName);
This is going to fail the same way it was failing before. If you want to make it
work, you need to remove the "static" here.

[Nicolas Sylvain, 2009-09-09 03:31:14]

but i dont object to trying it out.

lgtm

[cpu_(ooo_6.6-7.5), 2009-09-09 17:02:55]

I am ok to have it failing the same way as before. I am going to try it as is
but I understand Nicolas reservations.


On 2009/09/09 03:31:14, Nicolas Sylvain wrote:
> but i dont object to trying it out.
> 
> lgtm