Rebase and address lints

Ping

Replicate the leader table insertion ordering of the old linked list.

In D125205#3508443, @nikic wrote:

It looks like this change ends up being a slightly negative in terms of instruction count: http://llvm-compile-time-tracker.com/compare.php?from=bf1b81d076f89bd56e86189b013f27dcf4d73ae8&to=efa882a79c027d27f6deff14e75bd9f558dd95d0&stat=instructions

Format

Fix accidentally walking off of the end of the BB vector.

Fix condition.wq

Address review feedback.

Ping

In D120112#3334823, @dmgreen wrote:

You have the operands in the wrong order,

Fix operand order.

Ping

Rebase

Ping

@ldionne Ping

In D116338#3236613, @majnemer wrote:

Do we also need to test for __aarch64__? https://developer.arm.com/documentation/dui0774/g/chr1383660321827 says that __arm__ is only defined for 32-bit targets.

Add arch64 to detect AArch64 properly.

Which ones were not working? It might be reasonable to XFAIL that part of the test for non mainstream compilers? Technically we only support recent Clangs and GCCs, which I would assume should handle this properly.

Remove the platform verification test. Unfortunately, getting this to work consistently
across different compilers isn't really tractable given the state of strict floating point
support in various compilers/versions.

Fix oversight where std::nextafter was always specialized on double rather than the type under test.

Update tests to account for the change in behavior, and add a
specific test that checks the value of tinyness_before against
the actual platform behavior.

Squash

• Teach InstCombine to remove the realloc from free(realloc(...)).

Add negative test for ptrtoint.

Add a negative test for non-inbounds GEPs.

Update for comments.

Update for comments.

Update for comments.

In D109565#2995449, @lebedev.ri wrote:

Won't this also skip non-zero-offset GEP's?

Update based on comments to preserve original behavior of shouldBuildLookupTablesForConstant().

Update tests for comments.

Update for comments.

Use CHECK-LABEL

Do we need to handle the same pattern for non IN_REG variations?

I'll be the dissenting voice: I don't think this is the right approach. "It's what gas does" is not a sufficient reason to break user code. I think a warning and *possibly* removing support some time well in the future is the right solution here.

Did the concept of using the static DAG not work?

Nice catch, LGTM.

LGTM

If the maps are typically small, would SmallDenseMap make sense?