Commit be46142
committed
AMDGPU: Handle V->A MFMA copy from case with immediate src2
Handle a special case for copies from AGPR VGPR on the MFMA inputs.
If the "input" is really a subregister def, we will not see the
usual copy to VGPR for src2, only the read of the subregister def.
Not sure if this pattern appears in practice.1 parent 37c8164 commit be46142
File tree
2 files changed
+6
-3
lines changed- llvm
- lib/Target/AMDGPU
- test/CodeGen/AMDGPU
2 files changed
+6
-3
lines changed| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
370 | 370 | | |
371 | 371 | | |
372 | 372 | | |
| 373 | + | |
| 374 | + | |
| 375 | + | |
373 | 376 | | |
374 | | - | |
| 377 | + | |
375 | 378 | | |
376 | 379 | | |
377 | 380 | | |
| |||
Lines changed: 2 additions & 2 deletions
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
187 | 187 | | |
188 | 188 | | |
189 | 189 | | |
190 | | - | |
191 | | - | |
| 190 | + | |
| 191 | + | |
192 | 192 | | |
193 | 193 | | |
194 | 194 | | |
| |||
0 commit comments