Bug 188811 – [MSVC] Stop disabling /O2 features.

• Component: New Bugs
• Status: RESOLVED FIXED

[Ross Kirsling]

[MSVC] Stop disabling /O2 features.

[Ross Kirsling]

Created attachment 347686 [details]
Patch

[Ross Kirsling]

/O2 should equate to all of these:
/Og /Oi /Ot /Oy /Ob2 /Gs /GF /Gy

But we have a few disabled:
/Oy- /GF- /Gy-

These appear to have been necessary for AppleWin back at the time of https://trac.webkit.org/changeset/190361.

But WinCairo doesn't seem to need to disable these optimizations, and I'm hoping the same is true for AppleWin.

This patch fixes bug 187737 as a side effect.

[Ross Kirsling]

Given that the AppleWin build passed and this doesn't seem to be making the existing test failures any worse, I'd say we're safe as hoped! :D

[EWS Watchlist]

Comment on attachment 347686 [details]
Patch

Attachment 347686 [details] did not pass win-ews (win):
Output: https://webkit-queues.webkit.org/results/8943713

New failing tests:
http/tests/security/local-video-source-from-remote.html

[EWS Watchlist]

Created attachment 347793 [details]
Archive of layout-test-results from ews202 for win-future

The attached test failures were seen while running run-webkit-tests on the win-ews.
Bot: ews202  Port: win-future  Platform: CYGWIN_NT-6.1-2.10.0-0.325-5-3-x86_64-64bit

[Per Arne Vollan]

I seem to remember that JSC needs the frame pointer. Do you see any JSC test regressions when applying this patch?

[Ross Kirsling]

(In reply to Per Arne Vollan from comment #6)
> I seem to remember that JSC needs the frame pointer. Do you see any JSC test
> regressions when applying this patch?

Ahh -- I was hoping any issues would be made evident by Layout/API/Bindings/Built-ins tests, as we've yet to fully stabilize our JSC testing for WinCairo. (Actually in the past, JIT issues have often been reproable just by running MiniBrowser on a couple of pages.) If there are certain edge cases that require run-javascriptcore-tests to reliably encounter, would it be possible to have somebody test this with AppleWin?

Otherwise I suppose we could try it, keep a close eye on the AppleWin Debug Tests bot, and revert that part if things go unexpectedly south?

[Ross Kirsling]

Created attachment 347838 [details]
Patch

[Per Arne Vollan]

Comment on attachment 347838 [details]
Patch

R=me. Before landing, please check that changing these settings does not make the WebKit build significantly slower (both debug and release).

[WebKit Commit Bot]

Comment on attachment 347838 [details]
Patch

Clearing flags on attachment: 347838

Committed r235203: <https://trac.webkit.org/changeset/235203>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.

[Radar WebKit Bug Importer]

<rdar://problem/43623013>