Repository navigation
Commit 68ff5f5
committed
[aarch64] Fix vector FSUB subtracting the wrong registers
When intrinsics are not preferred, vector `FSUB` is lifted lane by lane.
That lifting read `Vd` and `Vn` instead of `Vn` and `Vm`, so
`fsub v0.4s, v1.4s, v2.4s` computed `v0 - v1`. It now computes `Vn - Vm`.1 parent bcfaac2 commit 68ff5f5
1 file changed
Lines changed: 5 additions & 4 deletions
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
1805 | 1805 | | |
1806 | 1806 | | |
1807 | 1807 | | |
1808 | | - | |
| 1808 | + | |
1809 | 1809 | | |
1810 | | - | |
1811 | | - | |
| 1810 | + | |
| 1811 | + | |
| 1812 | + | |
1812 | 1813 | | |
1813 | 1814 | | |
1814 | 1815 | | |
1815 | 1816 | | |
1816 | | - | |
| 1817 | + | |
1817 | 1818 | | |
1818 | 1819 | | |
1819 | 1820 | | |
| |||
0 commit comments