476691 - Testcase ecma3/Type/e8_5 has incorrect expected values and tamarin is producing them

Status: VERIFIED FIXED

(Tamarin Graveyard :: Virtual Machine, defect)

Description

[Brent Baker]

The following 2 tests in the ecma2/Type/e8_5.as testcase have an expected value that is incorrect, and tamarin is producing the incorrect value:

Positive finite Non zero values where e is 971
Negative finite Non zero values where e is 971

Finite nonzero values  that are Normalised having the form s*m*2**e where s is +1 or -1, m is a positive integer less than 2**53 but not less than s**52 and e is an integer ranging from -1074 to 971

Postive:
1*(Math.pow(2,52))*(Math.pow(2,971))
spidermonkey: 8.98846567431158e+307
tamarin: 1e+308

Negative:
-1*(Math.pow(2,52))*(Math.pow(2,971))
spidermonkey: -8.98846567431158e+307
tamarin: -e+308

Comment 1

[Brent Baker]

This testcase is unstable on x64, passing and failing intermittently, skipping the test until this is resolved to remove unwanted noise in the build system.

Comment 2

[Lars T Hansen]

The tests are unstable because the code works on uninitialized memory in the case of negative values.

Comment 3

[Lars T Hansen]

Attachment: If working on negative values, skip the uninitialized space set aside for the minus sign when scanning leading zeroes

Comment 4

[Steven Johnson]

Comment on attachment 360713 [details] [diff] [review]
If working on negative values, skip the uninitialized space set aside for the minus sign when scanning leading zeroes

Yow! Nice find.

Comment 5

[Lars T Hansen]

redux changeset:   1406:6790439dd3cd

Comment 6

[Brent Baker]

removed the skip from the testconfig (1407:0c0f3cf40ad0)