Review comments addressed.
Bug fix for not propagating fast math flags and attributes.

Review comments addressed.
Corrected a bug where arguments to fused intrinsics were unnecessarily swapped.

Removed IR combines as they will go to another patch.
Addressed review comments.

Implemented IR combines to convert _m intrinsics that take an all active predicate to their equivalent _u intrinsic.

Some refactoring. More tests added.

Corrections to the very first patch.
Added tests as well.

Corrected the commit message.
Added tests are now in alphabetical order.

Minor reordering after review comment.

Test adjustment for Windows. Disabling read permissions using chmod does not seem to work on Windows.

Adjusting the test for Windows. Setting the read bits only for the config file as Windows does not have full support for chmod.

In D144179#4146599, @MaskRay wrote:

This looks like introducing a footgun (when something behaves differently from an upstream Clang, it would be difficult for toolchain maintainers to know why).
Why can't your user specify CLANG_CONFIG_FILE_SYSTEM_DIR?

Adjusting the test for Windows OS again.

Adjusted the test for Windows OS. Once again.

In D144179#4132597, @MaskRay wrote:

Behaviors due to a new environment variable should be very careful. How is this useful? If you want this, you can add a wrapper around clang to specify --config= by yourself.

Think about this: if the user sets CONFIG_FILE= to something, it will be difficult for toolchain maintainers to triage a bug.

Extended the test to cover Windows OS.

Addressing one more review comment.

Yet another clean up after review comments.

Code clean up after review comments.

Simplified the code in accordance with review comments.

Allowing integer-sized pointers to be forwarded.

Added opaque pointer versions of previously added tests.

In D132239#3735560, @mgabka wrote:

Could you change the commit message? or at least add [LoopLoadElim] ? it does not make clear that these are LoopLoadElimination tests.

Added tests for int-pointer mismatch and int-vector mismatch.

1. Correcting errand assumption that the size of a pointer is 0.
2. Removing vector bitcasting tests as they were adding more confusion than value.
3. Regenerating CHECK lines with update_test_checks.py script.

In D130233#3667850, @nikic wrote:

Not all types are bitcastable. Please add tests where the load/store are pointer and integer (in which case you need inttoptr or ptrtoint -- and watch out for non-integral address spaces) and one where an aggregate is involved (cannot be bitcast at all).

1. Guarding against non-valid bitcasts.
2. Load-store forwarding works with vector types as well.
3. Added a test for load-store forwarding with vector types.
4. Added tests confirming bitcasting vectors won't corrupt the values stored.

@tahonermann, I addressed your comments in https://reviews.llvm.org/D129373. I just need to remove a tab that crept in and it seems to be more difficult than I thought as my code does not seem to have any tab and reformatting the code does not change the file.

Removing a tab.

Correcting computation of RealTypeUsesObjCFPRetMask.

Removed an unnecessary local variable.

Removing an unnecessary test change from aarch64-target-features.c
as init-aarch64.c provides enough coverage.

Updated Release Notes.
Renamed RealTypeUsesObjCFPRet to RealTypeUsesObjCFPRetMask.

Correcting my buggy computation of RealTypeUsesObjCFPRet bit-field.

Removing the implicit dependency on the FloatModeKind enumerator values.

Addressing review comments:

1. Added the test for mode(HF)
2. Moved Half before Float in FloatModeKind enum class to preserve precision ordering

Addressing review comments.

Removed legacy invocation and require<tbaa>.
Removed -store-to-load-forwarding-conflict-detection flag
as it is set by default and simplified check labels.
Added test where TypeSize != StoreSize != AllocaSize.

Adjusted indentation and changed CHECK to CHECK-NEXT in depend_diff_types.ll