|
02:22
xtt2 joined
04:07
viki joined
04:08
viki joined
06:09
xtt2 joined
08:18
domidumont joined
08:25
domidumont joined
08:58
domidumont joined
09:50
zakharyas joined
11:41
domidumont joined
12:37
FROGGS joined
15:08
domidumont joined
15:51
xtt2 joined
16:38
domidumont joined
17:28
vendethiel joined
18:06
lizmat_ joined
18:25
vendethiel joined
|
|||
| dalek | arVM: bf777f6 | (Jan-Olof Hendig)++ | src/strings/nfg.c: The function MVMNFGTrieNode incorrectly tries to free a memory block of an MVMNFGTrieNode. This works on 64-bit systems since both structs happens to be of the same size, something which is not the case on 32 bit systems. This resolves RT #130191 and is spectest clean. cygx++ for helping track this down. |
19:56 | |
| synopsebot6 | Link: rt.perl.org/rt3//Public/Bug/Displa...?id=130191 | ||
| MoarVM: ee0b54e | lizmat++ | src/strings/nfg.c: | |||
| MoarVM: Merge pull request #443 from dogbert17/master | |||
| jnthn | Nice catch:) | ||
| nwc10 | I wonder | 19:57 | |
| a) if there are any other bugs like that | |||
| b) if there's a way to systematically add a build option to prevent any ever happening again | |||
| jnthn | Yes, there's a debug mode in fixedsizealloc.c | 19:58 | |
| Which validates the size released matches the size allocated | |||
| nwc10 | aha right. but one would have to build on the 32 bit system | ||
| jnthn | Indeed. | 19:59 | |
|
22:06
mj41 joined
|
|||
| timotimo | hmm. a function "sizeofbutpretend32bitsystem" :D | 22:50 | |
| and then have to pass both to alloc and free | 22:51 | ||
| so there's two pairs of comparison | |||