Support for showing memory usage of 64-bit IE in a 32-bit Chromium process....

Support for showing memory usage of 64-bit IE in a 32-bit Chromium process.
Note that the product version won't be shown and the virtual memory columns will be "N/A".

* memory_details.{cc,h}
  Switch to CreateToolhelp32Snapshot()/Process32First()/Process32Next() from EnumProcess() because EnumProcess() in a 32-bit process doesn't enumerate 64-bit processes.
  Note: GetModuleFileNameEx() for a 64-bit process fails.

* GetCommittedKBytes() in process_util_win.cc
  Don't hung up for a 64-bit process.
  It sets zeros as results for a 64-bit process.

* about_memory.html
  If a value is zero, shows "N/A".

BUG=191
TEST=manual check for the following cases:
  about:memory had no behavior change in 32-bit Windows.
  about:memory had no behavior change in 64-bit Windows if 64-bit IE was not running.
  about:memory showed "IE (64bit)" row in 64-bit Windows if 64-bit IE was running.
  about:memory showed both of "IE" and "IE (64bit)" rows in 64-bit Windows if 64-bit IE and 32-bit IE were running.

[tkent_google.com, 2009-04-16 07:15:53]

Would you take a look at this please?

[Mike Belshe, 2009-04-21 18:37:42]

Thanks for fixing this!

Mostly looks good - just a couple of questions about the 64bit stuff.

http://codereview.chromium.org/75031/diff/1008/1009
File base/process_util_win.cc (right):

http://codereview.chromium.org/75031/diff/1008/1009#newcode439
Line 439: MEMORY_BASIC_INFORMATION mbi = {0};
Should you be using MEMORY_BASIC_INFORMATION64 here?

This might avoid the hack problem you have below?

http://codereview.chromium.org/75031/diff/1008/1012
File chrome/browser/memory_details.cc (right):

http://codereview.chromium.org/75031/diff/1008/1012#newcode129
Line 129: process_entry.szExeFile) != 0)
According to http://msdn.microsoft.com/en-us/library/ms684839(VS.85).aspx

you can't use szExeFile to look at the process name of a 64bit process and you
must use QueryFullProcessImageName?

[tkent_google.com, 2009-04-22 17:27:24]

Thank you for reviewing.

http://codereview.chromium.org/75031/diff/1008/1009
File base/process_util_win.cc (right):

http://codereview.chromium.org/75031/diff/1008/1009#newcode439
Line 439: MEMORY_BASIC_INFORMATION mbi = {0};
On 2009/04/21 18:37:42, Mike Belshe wrote:
> Should you be using MEMORY_BASIC_INFORMATION64 here?
> 
> This might avoid the hack problem you have below?

We can't use it.  If we get 64-bit values by M_B_I64, we can not continute
iterations because the 2nd parameter of VirtualQueryEx() is always 32-bit in a
32-bit process.

http://codereview.chromium.org/75031/diff/1008/1012
File chrome/browser/memory_details.cc (right):

http://codereview.chromium.org/75031/diff/1008/1012#newcode129
Line 129: process_entry.szExeFile) != 0)
On 2009/04/21 18:37:42, Mike Belshe wrote:
> According to http://msdn.microsoft.com/en-us/library/ms684839(VS.85).aspx
> 
> you can't use szExeFile to look at the process name of a 64bit process and you
> must use QueryFullProcessImageName?

szExeFile is ok here because we compare it with basenames of known browsers such
as "iexplorer.exe".

[Mike Belshe, 2009-05-22 20:12:14]

sorry for leaving this hanging.

LGTM