A DBG_VALUE between a two-address instruction and a following COPY would prevent rescheduleMIBelowKill optimization inside TwoAddressInstructionPass.
Details
- Reviewers
MatzeB aprantl - Group Reviewers
debug-info - Commits
- rG72b9deb21f12: [CodeGen] Skip over dbg-instr in twoaddr pass
rL350289: [CodeGen] Skip over dbg-instr in twoaddr pass
Diff Detail
Event Timeline
lib/CodeGen/TwoAddressInstructionPass.cpp | ||
---|---|---|
932 | For IR we have an iterator that skips over debug intrinsics automatically. Does the same thing exist for MIR, too? |
Thanks for the patch!
lib/CodeGen/TwoAddressInstructionPass.cpp | ||
---|---|---|
932 | Yes, I believe we have skipDebugInstructionsForward / skipDebugInstructionsBackward. It might be neat to use that here. | |
test/CodeGen/X86/twoaddr-dbg-value.mir | ||
12 | This is mostly an academic question, but: is this a valid DBG_VALUE? I'm a little surprised this parses (I'd expect the instruction to have more operands), but I like the simplicity of it. |
lib/CodeGen/TwoAddressInstructionPass.cpp | ||
---|---|---|
932 | If we were to use skipDebugInstructionsForward then it seems that the loop would have to be written as follows to preserve the semantics (and handle multiple COPYs with DBG_VALUEs interspersed in various ways). while (End != MBB->end()) { End = skipDebugInstructionsForward(End, MBB->end()) if (End->isCopy() && regOverlapsSet(Defs, End->getOperand(1).getReg(), TRI))) Defs.push_back(End->getOperand(0).getReg()); else break; ++End; } Would that be better? | |
test/CodeGen/X86/twoaddr-dbg-value.mir | ||
12 | It is probably not valid but does seem to be sufficient for the purpose of this test (where the presence of a DBG_VALUE is the only relevant thing and not the contents of it). I would be happy to make it more valid though if that is just a matter of adding the additional operands. |
lib/CodeGen/TwoAddressInstructionPass.cpp | ||
---|---|---|
932 | I personally find this easier to follow. Imho it's nice that the isDebugInstr check doesn't need to be duplicated. | |
test/CodeGen/X86/twoaddr-dbg-value.mir | ||
12 | Ah, I see. I think this is fine the way it is. It'd be a bit messy to add the debug-location metadata, and the benefit of having a fully-formed dbg_value here seems fairly marginal. |
lib/CodeGen/TwoAddressInstructionPass.cpp | ||
---|---|---|
933 | Of course if 'End == MBB->end()' after this point bad things would happen, but on the other hand that would also have been problematic in the original loop (and apparently it was not a problem in practice). Perhaps simply adding an assertion to indicate that this has been considered would be sufficient? |
lib/CodeGen/TwoAddressInstructionPass.cpp | ||
---|---|---|
933 | The last instruction pretty much has to be a terminator instruction and can't really be a debug intrinsic. |
For IR we have an iterator that skips over debug intrinsics automatically. Does the same thing exist for MIR, too?