Bug 160832

Summary: Make JSValue::strictEqual() handle failures to resolve JSRopeStrings.
Product: WebKit Reporter: Mark Lam <mark.lam>
Component: JavaScriptCoreAssignee: Mark Lam <mark.lam>
Severity: Normal CC: benjamin, fpizlo, ggaren, keith_miller, msaboff, sbarati, webkit-bug-importer
Priority: P2 Keywords: InRadar
Version: WebKit Local Build   
Hardware: Unspecified   
OS: Unspecified   
Description Flags
proposed patch.
mark.lam: review-
proposed patch. ggaren: review+

Description Mark Lam 2016-08-12 16:57:34 PDT
Currently, JSValue::strictEqualSlowCaseInline() (and peers) will blindly try to access the StringImpl of a JSRopeString that fails to resolve its rope.  As a result, we'll crash with null pointer dereferences.  We should fix this.
Comment 1 Mark Lam 2016-08-12 17:01:20 PDT
Comment 2 Mark Lam 2016-08-12 17:20:47 PDT
Created attachment 285984 [details]
proposed patch.

Let's get some EWS testing and feedback.

I don't have a test because the only test case I have so far relies on allocating just the right amount of memory to run out of memory right at the moment of resolving a rope for a strict equality check.  The test is brittle and flaky.  So far, it only manifests the issue on ARM64, but not on x86_64 yet.  So, I think its of questionable value and will leave it out for now.
Comment 3 Mark Lam 2016-08-12 19:31:32 PDT
Comment on attachment 285984 [details]
proposed patch.

Will fix the build failure.
Comment 4 Mark Lam 2016-08-15 10:33:36 PDT
Created attachment 286065 [details]
proposed patch.
Comment 5 Geoffrey Garen 2016-08-15 14:30:20 PDT
Comment on attachment 286065 [details]
proposed patch.

Let's call this "equal" since the WTF function is "equal".

Comment 6 Mark Lam 2016-08-15 14:55:01 PDT
Thanks for the review.  I've replaced "equals" with "equal" (and ditto for the matching slow case function).

Landed in r204485: <http://trac.webkit.org/r204485>.