773588 - IonMonkey: Assertion failure: [barrier verifier] Unmarked edge: <unknown>, at jsgc.cpp:4711

Status: VERIFIED FIXED

(Core :: JavaScript Engine, defect)

Description

[Christian Holler (:decoder)]

The following testcase asserts on ionmonkey revision a29f6c635516 (run with --ion -n -m --ion-eager):


gcPreserveCode()
gczeal(4);
function f() {
    var a = [], i, N = 10;
    for (i = 0; i < N; i++)
        a[i] = {
		m: function() { return 0; }, 
		m: function() { return (false  ); }
		};
        schedulegc(10);
        assertEq(f(), 1);
}
f();

Comment 1

[David Anderson [:dvander] - inactive, e-mail if emergency]

Attachment: fix

INITPROP can initialize the same element twice, meaning we may need a write barrier.

Comment 2

[David Anderson [:dvander] - inactive, e-mail if emergency]

CC'ing Bill and Brian since I think JM+TI might have the same bug, but I can't reproduce it with this test case.

Comment 3

[Bill McCloskey [inactive unless it's an emergency] (:billm)]

I tried to write a JM test case for this, but for some reason gczeal(4) seems to be totally disabling the methodjit. That didn't used to happen. I'll look into it tomorrow.

Comment 4

[Jan de Mooij [:jandem]]

Comment on attachment 642811 [details] [diff] [review]
fix

Review of attachment 642811 [details] [diff] [review]:
-----------------------------------------------------------------

Please add the testcase too, or do you want to do that after JM is fixed?

Comment 5

[David Anderson [:dvander] - inactive, e-mail if emergency]

http://hg.mozilla.org/projects/ionmonkey/rev/699ab277c0b8 (test case unfortunately iloops)

Bill, should I keep this open or file a separate bug for JM?

Comment 6

[Bill McCloskey [inactive unless it's an emergency] (:billm)]

I cloned this into bug 774859.

Comment 7

[David Anderson [:dvander] - inactive, e-mail if emergency]

Thanks!

Comment 8

[Christian Holler (:decoder)]

JSBugMon: This bug has been automatically verified fixed.