[Webkit-unassigned] [Bug 128115] [Win] LLINT is not working.

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=128115





--- Comment #20 from Mark Lam <mark.lam at apple.com>  2014-02-05 15:58:01 PST ---
(From update of attachment 223229)
I can't run the benchmarks (to confirm that there is no perf regression) because with the patch, we're crashing or getting assertion failures.

The ToT baseline I'm using is r163476.  With this patch (your 5th one @ https://bugs.webkit.org/attachment.cgi?id=223229) without any changes, the jsc tests fails (crashes) in testapi.  With my suggested fix for _llint_function_for_call_prologue and _llint_function_for_construct_prologue, it gets past testapi but fails in the jsc-stress tests (I'm running on a Mac, not Windows).

Running with just the LLINT, I get assertion failures / crashes:
$ JSC_useJIT=false ./Tools/Scripts/run-javascriptcore-test --debug

Running with just the JIT, I get assertion failures / crashes as well:
$ JSC_useLLInt=false ./Tools/Scripts/run-javascriptcore-test --debug

Running the baseline (without your patch), I see only 4 timeouts.

Can you please do some debugging, or do another round of code self-review to see what you've changed that is causing these crashes / assertion failures?

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.