[all-commits] [llvm/llvm-project] 604a6c: [BPI] Transfer value-handles when assign/move cons...

[Jeremy Morse via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 604a6c409e8473b212952b8633d92bbdb22a45c9
      https://github.com/llvm/llvm-project/commit/604a6c409e8473b212952b8633d92bbdb22a45c9
  Author: Jeremy Morse <jeremy.morse at sony.com>
  Date:   2024-01-24 (Wed, 24 Jan 2024)

  Changed paths:
    M llvm/include/llvm/Analysis/BranchProbabilityInfo.h

  Log Message:
  -----------
  [BPI] Transfer value-handles when assign/move constructing BPI (#77774)

Background: BPI stores a collection of edge branch-probabilities, and
also a set of Callback value-handles for the blocks in the
edge-collection. When a block is deleted, BPI's eraseBlock method is
called to clear the edge-collection of references to that block, to
avoid dangling pointers.

However, when move-constructing or assigning a BPI object, the
edge-collection gets moved, but the value-handles are discarded. This
can lead to to stale entries in the edge-collection when blocks are
deleted without the callback -- not normally a problem, but if a new
block is allocated with the same address as an old block, spurious
branch probabilities will be recorded about it. The fix is to transfer
the handles from the source BPI object.

This was exposed by an unrelated debug-info change, it probably just
shifted around allocation orders to expose this. Detected as
nondeterminism and reduced by Zequan Wu:


https://github.com/llvm/llvm-project/commit/f1b0a544514f3d343f32a41de9d6fb0b6cbb6021#commitcomment-136737090

(No test because IMHO testing for a behaviour that varies with memory
allocators is likely futile; I can add the reproducer with a CHECK for
the relevant branch weights if it's desired though)