| Title | awluthe test failure on OS X |
| Status | suspended |
| Priority | nice |
| Assigned user | Gareth Rees |
| Organization | Ravenbrook |
| Description | As of change 181788, this test case failure is reliably reproducible with the given seed: $ code/xc/Release/awluthe 946413326 code/xc/Release/awluthe: randomize(): resetting initial state (v3) to: 946413326. Strongly unreachable weak table entry found, slot 48. |
| Analysis | The test case was reliably reproducible yesterday (2013-05-15 -- and in fact it was failing not just at the seed given above, but at perhaps a third of seeds tested). However, today (2013-05-16) it's not failing at all under that seed or many others. Perhaps address space layout randomization is at fault? The awluthe test hashes on the addresses of objects so its behaviour will vary as the address space layout varies. But on investigation it seems that the MPS (with a virtual memory arena) defeats ASLR: heap addresses come out the same every time: see job003484. |
| How found | automated_test |
| Evidence | None |
| Observed in | 1.111.0 |
| Created by | Gareth Rees |
| Created on | 2013-05-15 16:38:48 |
| Last modified by | Gareth Rees |
| Last modified on | 2014-03-27 21:01:31 |
| History | 2013-05-15 GDR Created. 2014-03-25 GDR Suspended as it cannot be reproduced. |