|
00:25
vendethiel joined
00:50
cognominal joined
00:51
cognominal joined
01:20
colomon joined
01:30
colomon joined
01:48
ilbot3 joined
01:55
cognominal joined
02:10
vendethiel joined
02:22
colomon joined
03:51
vendethiel joined
04:53
vendethiel joined
06:03
vendethiel joined
06:28
vendethiel joined
07:05
FROGGS joined
07:22
vendethiel joined
07:56
vendethiel joined
08:38
vendethiel joined
09:20
vendethiel joined
09:24
TEttinger joined
10:59
vendethiel joined
11:43
rurban joined
11:47
vendethiel joined
14:49
colomon joined
15:10
danaj joined,
vendethiel joined
16:11
vendethiel joined
17:47
FROGGS joined
18:56
zakharyas joined
19:00
ggoebel joined
19:08
rurban joined
19:36
TEttinger joined
19:45
brrt joined
|
|||
| brrt | \o | 19:45 | |
| seems to me i might have a usuable register allocator now | 19:53 | ||
| and a memory corruption, too ^^ | 19:54 | ||
| FROGGS | ó.ò | 20:28 | |
| brrt | well | ||
| that's what ASAN is for | |||
| i suspect an error in dynamic array implementation | 20:29 | ||
| brrt managed to crash ASAN | 20:40 | ||
|
20:42
brrt left
|
|||
| dalek | arVM/even-moar-jit: 77d3ef0 | brrt++ | src/jit/compile.c: I think this is a valid register allocation implemnetation. Use a 'register ring' for ensuring we can always take a valid new register; for each node ensure again that all children exist in a register. (That is, don't depend on having them after tree traversal, as the computation of one may expel another). |
20:42 | |
|
20:59
colomon joined
21:08
lizmat joined
21:10
colomon joined
21:22
ggoebel2 joined
21:29
brrt joined
23:35
arnsholt joined,
camelia joined
|
|||