msvs/ninja win: Fix support for ImageHasSafeExceptionHandlers

msvs/ninja win: Fix support for ImageHasSafeExceptionHandlers

For ninja, we now emulate msvs by passing /SAFESEH:NO
when ImageHasSafeExceptionHandlers is false, and passing
/SAFESEH otherwise.

For msvs, we now translate "MSBuildOnly" properties from
msvs_settings "as-is".  Also, the msvs generator has been
updated so support the MASM tool, so that the
"gyptest-link-safeseh.py" test can be enabled.

BUG=
R=scottmg@chromium.org

Committed: https://code.google.com/p/gyp/source/detail?r=2026

[Shezan Baig (Bloomberg), 2015-01-16 23:31:26]

sbaig1@bloomberg.net changed reviewers:
+ scottmg@chromium.org

[Shezan Baig (Bloomberg), 2015-01-16 23:31:27]

Hmm, there seems to be something wrong with the try servers.

[scottmg, 2015-01-19 20:17:14]

https://codereview.chromium.org/790213007/diff/1/pylib/gyp/msvs_emulation.py
File pylib/gyp/msvs_emulation.py (right):

https://codereview.chromium.org/790213007/diff/1/pylib/gyp/msvs_emulation.py#...
pylib/gyp/msvs_emulation.py:610: map={'false': ':NO', 'true': ''},
prefix='/SAFESEH')
Does test\win\gyptest-link-safeseh.py work on msvs now, and could it be enabled?

[Shezan Baig (Bloomberg), 2015-01-20 16:05:25]

On 2015/01/19 20:17:14, scottmg wrote:
> https://codereview.chromium.org/790213007/diff/1/pylib/gyp/msvs_emulation.py
> File pylib/gyp/msvs_emulation.py (right):
> 
>
https://codereview.chromium.org/790213007/diff/1/pylib/gyp/msvs_emulation.py#...
> pylib/gyp/msvs_emulation.py:610: map={'false': ':NO', 'true': ''},
> prefix='/SAFESEH')
> Does test\win\gyptest-link-safeseh.py work on msvs now, and could it be
enabled?

Almost :)

I had to add support for 'MASM' in the msvs generator.  Now, the
'test_safeseh_yes' and 'test_safeseh_no' targets succeed.  However, the
'test_safeseh_default' is failing with:

    error LNK2026: module unsafe for SAFESEH image.
    fatal error LNK1281: Unable to generate SAFESEH image.

It looks like the *default* value for Link.ImageHasSafeExceptionHandlers in msvs
is true, and there doesn't seem to be a way to make it use neither /SAFESEH nor
/SAFESEH:NO -- it is always one or the other.

I think I will need to change msvs_emulation.py to always use /SAFESEH, unless
ImageHasSafeExceptionHandlers is false, in which case it will use /SAFESEH:NO. 
This will make it behave more like msvs.  In this case, I think
test_safeseh_default should be removed/adjusted.  WDYT?

[scottmg, 2015-01-20 17:00:24]

On 2015/01/20 16:05:25, Shezan Baig (Bloomberg) wrote:
> On 2015/01/19 20:17:14, scottmg wrote:
> > https://codereview.chromium.org/790213007/diff/1/pylib/gyp/msvs_emulation.py
> > File pylib/gyp/msvs_emulation.py (right):
> > 
> >
>
https://codereview.chromium.org/790213007/diff/1/pylib/gyp/msvs_emulation.py#...
> > pylib/gyp/msvs_emulation.py:610: map={'false': ':NO', 'true': ''},
> > prefix='/SAFESEH')
> > Does test\win\gyptest-link-safeseh.py work on msvs now, and could it be
> enabled?
> 
> Almost :)
> 
> I had to add support for 'MASM' in the msvs generator.  Now, the
> 'test_safeseh_yes' and 'test_safeseh_no' targets succeed.  However, the
> 'test_safeseh_default' is failing with:
> 
>     error LNK2026: module unsafe for SAFESEH image.
>     fatal error LNK1281: Unable to generate SAFESEH image.
> 
> It looks like the *default* value for Link.ImageHasSafeExceptionHandlers in
msvs
> is true, and there doesn't seem to be a way to make it use neither /SAFESEH
nor
> /SAFESEH:NO -- it is always one or the other.
> 
> I think I will need to change msvs_emulation.py to always use /SAFESEH, unless
> ImageHasSafeExceptionHandlers is false, in which case it will use /SAFESEH:NO.

> This will make it behave more like msvs.  In this case, I think
> test_safeseh_default should be removed/adjusted.  WDYT?

Strange, the docs claim it defaults to enabling it if it's possible. But I guess
that's link's default vs. the IDE's default.

Sounds good to me to adjust the _default test, probably better to keep it to
document this behaviour, but if you can't get it working reasonably, removing is
fine too.

[Shezan Baig (Bloomberg), 2015-01-20 17:46:28]

PTAL -- There seems to be something wrong with the trybot servers.  But I ran
the test locally and it works.

[scottmg, 2015-01-20 18:05:32]

lgtm

[Shezan Baig (Bloomberg), 2015-01-20 21:03:53]

Committed patchset #2 (id:20001) manually as 2026 (presubmit successful).