I'm not familiar with the behaviour of cmake exports so I don't feel able to review this but FWIW, since it goes on to check libraries against LLVM_AVAILABLE_LIBS line 289 it doesn't sound wrong to check against that a little earlier.

LGTM, Thanks!

LGTM (we should wait for Matt to confirm he's ok with dealing with the high-level issue later though)

In D103352#2791465, @bogner wrote:

How much work is left to just remove this completely?

LGTM too. Thanks!

LGTM

LGTM. Sorry for the long wait

LGTM with the check in ProcessBlockPRE

The change LGTM but could you add a test case? There probably aren't many convergent instructions upstream but it should be possible to make a test case using AMDGPU instructions or via G_INTRINSIC

I'm not aware of anything still using it either. LGTM

It turns out someone already added C++11 default initializers but they're not used in this case. MachineModuleInfo could use some better commenting about its lifetime so I'll add that in a follow-up at the same time as renaming it

Patch by Roman Tereshin (it seems arcanist loses that info for the review)

I just pushed my bug fix. Let's wait a day until I know it doesn't bust the build, then push this one.

Sounds good to me, thanks. I've also unreverted the original downstream and applied your fix there too.

Yep, ./bin/llvm-lit -svv ../llvm-project/llvm/test/TableGen and ninja check passes

LGTM

The patches I mentioned are in D100253

I'll post a patch for the Rec10.Zero fix in a moment

Ah, okay, that sounds reasonable. Now what is the circumstance under which ? is coerced to 0?

In rG14580ce2fdd1898d130b20d9eb21bc4281868e7c#988453, @Paul-C-Anagnostopoulos wrote:

@dsanders

Okay, there is a lot going on here.

If you look at the two lines I added to Record.cpp, you'll see that I did an awesome job of it. It always fetches element 0 of the list, regardless of the actual subscript. So the last two fields in Z7LstBits will have element 0, the ?.

You're right the bit slices haven't changed behaviour, the one that has is list slices. Sorry about that, the same test has bit slices nearby and it's trying to test bit/list slices are consistent w.r.t our !iscomplete() predicate.

Hi Paul,

LGTM

LGTM

LGTM

Looks like phabricator noticed the revert but not the recommit. Here's the recommit:

bc21c7baaf00 [globalisel][unittests] Make it a compiler error to not call setUp() to avoid future rotten green tests

I forgot to update my local commit message to the phabricator one before committing the first time around

LGTM

In D99692#2662702, @bogner wrote:

The title/summary seems a bit confusing. The interesting part of this change is having the set up function return a target machine so it's impossible to miss - IMO renaming it is incidental. This seems reasonable to do, but I think it'd be better titled something along the lines of "Return a target machine from the test setup so that it's structurally required"

Thanks for fixing this, I just ran into the same problem after basing a new test off of one of these on an older branch and wondering why it passed before I'd written the result string. I did a bit of digging before finding this commit and I think part of the problem here is that gtest will call the virtual function SetUp() automatically but not setUp(). I think we should also rename setUp() to something that doesn't clash with gtest (createTargetAndModule, initLLVM, etc.)

In D74588#2660144, @aaronpuchert wrote:

In D74588#2659652, @dsanders wrote:

All I wanted was a subdirectory to organize the sources specific to the GISel tablegen passes :-).

Not only that, you also want to use them in TableGenTests, and this I think is why can't just have them in llvm-tblgen.

In D74588#2659067, @aaronpuchert wrote:

In D74588#2658871, @beanz wrote:

Yep, and that's because of a different bug in how LLVMTableGenGlobalISel was setup.

This patch would address the issues: [...]

Ok, that allows me to remove the DISABLE_LLVM_LINK_LLVM_DYLIB, but

• maybe it would be confusing to have an LLVM component in llvm/utils instead of llvm/lib, and
• maybe this shouldn't be part of LLVM because it's only needed for llvm-tblgen? In other words maybe this isn't actually a component of LLVM proper but just a tool for building LLVM. If we make it a component library it will be part of libLLVM.so.

In D74588#2657461, @beanz wrote:

In D74588#2654289, @aaronpuchert wrote:

In D74588#2654185, @beanz wrote:

I think this bug is the other way around. LLVMTableGenGlobalISel should not be using DISABLE_LLVM_LINK_LLVM_DYLIB.

The commit message of rL373651 says

Fixed the -DLLVM_LINK_LLVM_DYLIB=ON using DISABLE_LLVM_LINK_LLVM_DYLIB when creating the library. Apparently it automatically links to libLLVM.dylib and we don't want that from tablegen.

This makes no sense. The tablegen tool build disables linking the dylib, which should be sufficient. We should never need to force it on a static archive. I’m 99% sure, that rL373651 is just wrong here, and that the solution to all of this is to just remove DISABLE_LLVM_LINK_LLVM_DYLIB from the library.

In D99033#2645272, @aemerson wrote:

Relaxing G_ADD semantics seems fine to me. Maybe we should still canonicalize on G_PTR_ADD during translation, and allow targets to legalize them into plain G_ADDs.

LGTM

In D98046#2629692, @dsanders wrote:

I didn't try to implement the fix because I don't have a target that needs it so I wouldn't be able to test it. @dsanders do you?

Sorry, I've gone to try and confirm this a couple times in the last week and things kept coming up. I'll take another look a moment

I didn't try to implement the fix because I don't have a target that needs it so I wouldn't be able to test it. @dsanders do you?

I was able to switch our downstream tools to use the MCContext diagnostic handler. For one case I added an empty SourceMgr when the MCContext was created, if you don't do that then SrcMgr can be nullptr even though it's a reference. For the other, I had a little trouble with MCContext's handler being able to store some opaque data and give it back when it calls the handler but I was able to resolve this by wrapping my handler in a capturing lambda and capturing the variable I needed.

LGTM

It looks like I have a second problem hiding behind the first which I'm not sure how to solve yet. It only appears in our downstream tests because the test has a warning it's not supposed to have but has been hidden by llvm-lit -s. That warning is currently fatal (it calls llvm_unreachable) because MCContext::diagnose() expects SrcMgr or InlineSrcMgr to be non-null but we have neither because our MCContext was set up for disassembling an executable so we didn't have a SourceMgr. I can work around it by removing the llvm_unreachable() in MCContext::diagnose() but that results in passing a nullptr being passed to a const SourceMgr &. I think it might need to be made a const SourceMgr * to support this case

In rG955365524aeea122243cb8ef201cd2687e7694b4#980419, @ychen wrote:

Sorry for the confusion. 5de2d189e6ad was supposed to reduce but not cause more confusion.

Thanks for this workaround. The change also caused one of our downstream unittests to fail because errors stopped being delivered to our ASSERT_EQ code. I haven't quite got to the bottom of where our DiagHandler gets set up yet but cherry-picking this caused our unittest to pass again and now I know what I'm looking for

Note: I went with undef in this patch because the keyword already exists but it occurred to me while posting that that could also be confusing. The location is defined, we just don't know what it is. We might want to consider something more direct like 'unknown'. Any thoughts on the spelling to use?

LGTM with a couple nitpicks

LGTM

FWIW, I still feel the encoder state shouldn't live inside the iterator but I am a bit happier that the iterator doesn't own the non-trivial encoder logic anymore. It's more consistent with pointing into a container rather than being the container, even if the pointer is unusually descriptive and the container is really more of a description of all the possible containers disambiguated by references to a specific element. I do wonder why do the logic and not the storage though. All that said, pragmatically I don't really want to insist on that as it's been in review for a very long time now so I think the middle ground is, LGTM as it currently is and if someone needs to use it in a generator-like style at some point they'll be the ones to put detail::encodeLEB128Byte() and the state in an object and have the iterators use that. Similarly, if we grow more places where iterators don't make sense they can either use the existing functions or do the same transformation just described.

I think lifting the if (DstTy.isVector()) without testing all the cases it affects is a bit risky. LGTM if that's confirmed safe or if the check is sunk to the cases you didn't mean to affect

LGTM with the changingInstr() to match the changedInstr() on the setReg() path.

LGTM

A sys::SmartMutex equivalent of this would LGTM. Let me know if there's a reason we should stick to std::mutex

Mostly LGTM but I'd rather move the vector_extract back to SelectionDAGCompat.td as having it there doesn't break anything.

LGTM, Thanks

Thanks!

In D93208#2456177, @Paul-C-Anagnostopoulos wrote:

@dsanders Which target is this?

In D93208#2455809, @Paul-C-Anagnostopoulos wrote:

Here are the Phabricator discussions. Daniel Sanders says that he knows of out-of-tree users of these data structures.

https://reviews.llvm.org/D63320

https://reviews.llvm.org/rL366278

But since everyone else have had to do this for other targets, I suppose I will be forced to do the same.

LGTM

LGTM with @gjain's suggestions. I think SIRegisterInfo.cpp could also propagate MCRegister a bit further if you wanted but I didn't check all the users of SubReg

Peng asked me to commit this off-list