[PATCH] D109052: [AMDGPU][GlobalISel] Fix waterfall loops

[Sebastian Neubauer via Phabricator via llvm-commits]

sebastian-ne added inline comments.


================
Comment at: llvm/lib/CodeGen/GlobalISel/Localizer.cpp:172
 
-    MachineBasicBlock::iterator II(MI);
+    MachineBasicBlock::instr_iterator II(MI);
     ++II;
----------------
foad wrote:
> sebastian-ne wrote:
> > foad wrote:
> > > Why is this change needed? Normally a BUNDLE instruction has use operands representing all the uses of the instructions inside the bundle, so I think this code should just work without any changes? See comment in lowerCall...
> > As far as I understand, it only has that after register allocation (or somewhere in these passes, like phi elimination. The pass that adds the BUNDLE instruction also adds the uses and defs).
> AMDGPU creates some bundles at instruction selection time, and it does use finalizeBundle, but they do not have any defs, so maybe that does not violate SSA? Example:
> ```
>   BUNDLE implicit killed %6:vgpr_32, implicit $m0, implicit $exec {
>     DS_GWS_BARRIER killed %6:vgpr_32, 0, implicit $m0, implicit $exec :: (load (s32) from custom "GWSResource")
>     S_WAITCNT 0
>   }
> ```
It definitely complains if I try to finalize the bundle :)

```
*** Bad machine code: Multiple virtual register defs in SSA form ***
- function:    test_indirect_call_sgpr_ptr
- basic block: %bb.1  (0x6d87e8)
- instruction: BUNDLE implicit-def $scc, implicit-def %16:_(p4), implicit-def %17:_(p4), implicit-def %19:_(p4), implicit-def %20:_(s64), implicit-def %18:_(p4), implicit-def %21:_(s64), implicit-def %22:_(s32), implicit-def %23:_(s32), implicit-def %24:_(s32), implicit-def %25:_(s32), implicit-def %26:_(s32), implicit-def %27:_(s32), implicit-def %28:_(s32), implicit-def %29:_(s32), implicit-def %30:_(s32), implicit-def %31:_(s32), implicit-def %32:_(s32), implicit-def %33:_(s32), implicit-def %34:_(<4 x s32>), implicit-def $sgpr0_sgpr1_sgpr2_sgpr3, implicit-def $sgpr0, implicit-def $sgpr0_lo16, implicit-def $sgpr0_hi16, implicit-def $sgpr1, implicit-def $sgpr1_lo16, implicit-def $sgpr1_hi16, implicit-def $sgpr2, implicit-def $sgpr2_lo16, implicit-def $sgpr2_hi16, implicit-def $sgpr3, implicit-def $sgpr3_lo16, implicit-def $sgpr3_hi16, implicit-def $sgpr0_sgpr1, implicit-def $sgpr0_sgpr1_sgpr2, implicit-def $sgpr2_sgpr3, implicit-def $sgpr4_sgpr5, implicit-def $sgpr4, implicit-def $sgpr4_lo16, implicit-def $sgpr4_hi16, implicit-def $sgpr5, implicit-def $sgpr5_lo16, implicit-def $sgpr5_hi16, implicit-def $sgpr6_sgpr7, implicit-def $sgpr6, implicit-def $sgpr6_lo16, implicit-def $sgpr6_hi16, implicit-def $sgpr7, implicit-def $sgpr7_lo16, implicit-def $sgpr7_hi16, implicit-def $sgpr8_sgpr9, implicit-def $sgpr8, implicit-def $sgpr8_lo16, implicit-def $sgpr8_hi16, implicit-def $sgpr9, implicit-def $sgpr9_lo16, implicit-def $sgpr9_hi16, implicit-def $sgpr10_sgpr11, …
```


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D109052/new/

https://reviews.llvm.org/D109052