474769 - TM: wrong number with nested type-unstable loops

Status: VERIFIED FIXED

(Core :: JavaScript Engine, defect)

Description

[Jesse Ruderman]

for each (b in [1, 1, 1, 1.5, 1, 1]) {
    (function() { for each (let h in [0, 0, 1.4, ""]) {} })()
}
print(b);

Result: -6.2774359784998874e+66

Expected: 1

Regression from bug 469044, I think.

Comment 1

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

Looks like two incompatible peers are being linked somewhere, it's trying to unbox 0x00000001 0xCDCDCDCD as a double whereas it's 0x1 as an integer.

Comment 2

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

Attachment: proposed fix

missing break tried to patch the same branch twice.  also, fixed a small memory leak.

Comment 3

[Andreas Gal :gal]

Comment on attachment 358320 [details] [diff] [review]
proposed fix

Nice. I think someone actually reported this leak as "leak in chrome".

Comment 4

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

Pushed fix as changeset: http://hg.mozilla.org/tracemonkey/rev/349fab40252a

Comment 5

[Bob Clary [:bc] (inactive)]

http://hg.mozilla.org/mozilla-central/rev/9a4f1433c04b
/cvsroot/mozilla/js/tests/js1_8/regress/regress-474769.js,v  <--  regress-474769.js
initial revision: 1.1

Comment 6

[Robert Sayre]

http://hg.mozilla.org/mozilla-central/rev/349fab40252a

Comment 7

[Robert Sayre]

http://hg.mozilla.org/releases/mozilla-1.9.1/rev/e7b59363129b

Comment 8

[Bob Clary [:bc] (inactive)]

http://hg.mozilla.org/mozilla-central/rev/e9226dd6c073

Comment 9

[Bob Clary [:bc] (inactive)]

v 1.9.1, 1.9.2