Gentle ping.

In D87045#2274009, @Whitney wrote:

In D87045#2265645, @TaWeiTu wrote:

In D87045#2264236, @ychen wrote:

• How about using PreservedAnalyses run(Loop &, LoopAnalysisManager &, LoopStandardAnalysisResults &, LPMUpdater &, LoopNest &); for the interface. I think it should work better because it is still a loop pass interface.

Sounds like a good idea to me. Let' see what @Whitney and @etiotto think about the new interface first, and I'll update the patch later if they agree on that. Thanks!

I am not strongly against it, however is not clear to me what the extra argument Loop & provide, as we can get the loop from LoopNest & by LN.getOutermostLoop().

Gentle ping to @Whitney and @etiotto on the opinion about the new interface for loop-nest pass suggested above. Thanks!

Oh, I forgot about it. --author="Ta-Wei Tu <tu.da.wei@gmail.com>", thanks!

Hi @arsenm, thank you for the review! Can you commit this for me? Because I don't have the commit access yet. Thanks!

In D87045#2264236, @ychen wrote:

• How about using PreservedAnalyses run(Loop &, LoopAnalysisManager &, LoopStandardAnalysisResults &, LPMUpdater &, LoopNest &); for the interface. I think it should work better because it is still a loop pass interface.

Add PassInstrumentation callback tests.

Replace SFINAE helper with llvm::is_detected.

Address (some) code review comments.

ping :)

Add unit test.

Hi, thanks for your comments and suggestions!

In D86133#2236462, @Whitney wrote:

In D86133#2236450, @TaWeiTu wrote:

Hi @etiotto, @Whitney thanks for your reviews!
Can you commit this for me? --author="Ta-Wei Tu <tu.da.wei@gmail.com>" as mentioned above looks good to me, thanks!

Sure. I am swamped with other work, finally have time to get to it today.

Hi @etiotto, @Whitney thanks for your reviews!
Can you commit this for me? --author="Ta-Wei Tu <tu.da.wei@gmail.com>" as mentioned above looks good to me, thanks!

Replace unnecessary [&] with [].

In D86133#2227414, @Whitney wrote:

--author="Ta-Wei Tu <tu.da.wei@gmail.com>" Good?

Source -> IncomingBlock

Hi @Whitney, @fhahn, thanks for your review and comment!
Please let me know if there's any other problem with the patch.
Also, if the patch is ready to land, is it possible for you to commit it instead (I believe I don't have the commit access)?

Rename variables. Use BasicBlock::phis(). Explain what "extra Phi block" means in the comment.

Check the existence of the extra block only when the inner loop is guarded and the inner loop exit contains LCSSA Phi instructions.

Hi @Whitney, @fhahn, @ychen, thank you all for your comments and suggestions! Sorry for the late reply.
I was trying to convert the LoopInterchange pass into a loop-nest pass. However, there seems to be no corresponding loop pass for the NPM.
Any particular reason for this?

Fix LCSSA block checking. Rename loops in the test file.

@ychen Again, thanks for your comment!

@ychen Thank you for your comment! I've added some unittests. Please let me know if there's anything unclear about the intended usage.

Test the functionality of addNewLoopNests().
Fix LoopNestAnalysis invalidation bugs.

Fix bug of holding dangling LoopNest references.

Add tests about invalidating the analysis results and clearing the result cache.

Add basic unit-tests to test and demonstrate the functionality of loop-nest-pass-related infrastructure (LoopNestPassManagerTest.cpp). Still working on more tests.

Fix LoopRotate/pr35210.ll (sync with commit b36c39260edcded47436f344e48f78cfbedac494).

Propagate the requirement of MemorySSA from loop passes to loop nest passes.

Fix loop deletion and rotation tests.

Fix LOOP_NEST_ANALYSIS macro, clang-tidy warnings. Rebase.

Fix several pre-merge checks.