Call NtTerminateProcess directly in Process::Terminate

Call ntdll.dll!NtTerminateProcess directly, without going through the
import table for kernel31.dll!TerminateProcess to determine if TerminateProcess is being hooked.

This is an attempt to diagnose the bug referenced below.

BUG=81449
Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=96807

[apatrick_chromium, 2011-08-12 00:15:24]

Argh.

[darin (slow to review), 2011-08-12 00:19:59]

LGTM

On Thu, Aug 11, 2011 at 5:15 PM, <apatrick@chromium.org> wrote:

> Reviewers: darin,
>
> Message:
> Argh.
>
> Description:
> Call ntdll.dll!NtTerminateProcess directly, without going through the
> import table for kernel31.dll!TerminateProcess to determine if
> TerminateProcess
> is being hooked.
>
> This is an attempt to diagnose the bug referenced below.
>
> BUG=81449
>
> Please review this at
http://codereview.chromium.**org/7640008/<http://codereview.chromium.org/7640...
>
> SVN Base: svn://chrome-svn/chrome/trunk/**src/
>
> Affected files:
>  M     base/process_win.cc
>
>
> Index: base/process_win.cc
> ==============================**==============================**=======
> --- base/process_win.cc (revision 96401)
> +++ base/process_win.cc (working copy)
> @@ -22,7 +22,18 @@
>  void Process::Terminate(int result_code) {
>   if (!process_)
>     return;
> -  ::TerminateProcess(process_, result_code);
> +
> +  // TODO(apatrick): Call NtTerminateProcess directly, without going
> through the
> +  // import table to determine if TerminateProcess is being hooked.
> +  // http://crbug.com/81449.
> +  HMODULE module = GetModuleHandle(L"ntdll.dll");
> +  typedef UINT (WINAPI *TerminateStatusPtr)(HANDLE handle, UINT code);
> +  TerminateStatusPtr terminate_process = reinterpret_cast<**
> TerminateStatusPtr>(
> +      GetProcAddress(module, "NtTerminateProcess"));
> +  terminate_process(process_, result_code);
> +
> +  // It used to look like this:
> +  //::TerminateProcess(process_, result_code);
>  }
>
>  bool Process::**IsProcessBackgrounded() const {
>
>
>

[apatrick_chromium, 2011-08-12 00:25:16]

On 2011/08/12 00:19:59, darin wrote:
> LGTM
> 

rvargas suggested it might be Process::CloseHandle so I'll so that one as well.

[darin (slow to review), 2011-08-12 04:09:20]

OK... have at it.  Feel free to TBR this sort of change.

On Thu, Aug 11, 2011 at 5:25 PM, <apatrick@chromium.org> wrote:

> On 2011/08/12 00:19:59, darin wrote:
>
>> LGTM
>>
>
>
> rvargas suggested it might be Process::CloseHandle so I'll so that one as
> well.
>
>
>
http://codereview.chromium.**org/7640008/<http://codereview.chromium.org/7640...
>