[PATCH] D68092: [AMDGPU] Invert the handling of skip insertion.
Matt Arsenault via Phabricator via llvm-commits
llvm-commits at lists.llvm.org
Wed Nov 13 02:33:27 PST 2019
arsenm added inline comments.
================
Comment at: lib/Target/AMDGPU/SIRemoveShortExecBranches.cpp:63-75
+static bool isForwardBranch(MachineBasicBlock *StartMBB,
+ MachineBasicBlock *EndMBB) {
+ MachineFunction::const_iterator MBBI = StartMBB->getIterator();
+ MachineFunction::const_iterator MBBE = StartMBB->getParent()->end();
+ MachineFunction::const_iterator MBBToI = EndMBB->getIterator();
+
+ while (++MBBI != MBBE) {
----------------
You're already doing a forward scan through the blocks in mustRetainExeczBranch, so it seems like this should just be checked there
================
Comment at: lib/Target/AMDGPU/SIRemoveShortExecBranches.cpp:82-84
+ // analyzeBranch doesn't update the FalseMBB when it is the fallthrough block.
+ if (!From)
+ From = SrcMBB.getNextNode();
----------------
I was thinking this would go in a separate wrapper function along with the analyzeBranch call. The connection to analyzeBranch isn't obvious at this point
Repository:
rL LLVM
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D68092/new/
https://reviews.llvm.org/D68092
More information about the llvm-commits
mailing list