[all-commits] [llvm/llvm-project] 3bf6ec: [flang][PFT-to-MLIR] Wrap unstructured Fortran con...

Thu Jul 16 07:03:01 PDT 2026

  Branch: refs/heads/users/ergawy/enclode_in_scf.execute_region
  Home:   https://github.com/llvm/llvm-project
  Commit: 3bf6ec5d581affe3b25a7069b57b95774749e4e4
      https://github.com/llvm/llvm-project/commit/3bf6ec5d581affe3b25a7069b57b95774749e4e4
  Author: ergawy <kareem.ergawy at gmail.com>
  Date:   2026-07-16 (Thu, 16 Jul 2026)

  Changed paths:
    M flang/include/flang/Lower/PFTBuilder.h
    M flang/lib/Lower/Bridge.cpp
    M flang/lib/Lower/PFTBuilder.cpp
    M flang/lib/Lower/Runtime.cpp
    M flang/test/HLFIR/assumed_shape_with_value_keyword.f90
    M flang/test/HLFIR/optional_dummy.f90
    M flang/test/Integration/OpenMP/parallel-private-reduction-worstcase.f90
    M flang/test/Lower/HLFIR/intrinsic-subroutines.f90
    M flang/test/Lower/MIF/change_team2.f90
    M flang/test/Lower/OpenACC/Todo/acc-unstructured-combined-construct.f90
    M flang/test/Lower/OpenACC/Todo/do-loops-to-acc-loops-todo.f90
    M flang/test/Lower/OpenACC/acc-terminator.f90
    M flang/test/Lower/OpenACC/acc-unstructured.f90
    M flang/test/Lower/OpenMP/loop-compound.f90
    M flang/test/Lower/OpenMP/parallel-reduction3.f90
    M flang/test/Lower/OpenMP/stop-stmt-in-region.f90
    M flang/test/Lower/OpenMP/unstructured.f90
    M flang/test/Lower/OpenMP/wsloop-unstructured.f90
    M flang/test/Lower/branching-directive.f90
    A flang/test/Lower/do_loop_execute_region_wrap.f90
    M flang/test/Lower/do_loop_unstructured.f90
    M flang/test/Lower/fail_image.f90
    M flang/test/Lower/ifconvert.f90
    M flang/test/Lower/mixed_loops.f90
    M flang/test/Lower/nsw.f90
    M flang/test/Lower/pre-fir-tree02.f90
    M flang/test/Lower/while_loop.f90

  Log Message:
  -----------
  [flang][PFT-to-MLIR] Wrap unstructured Fortran constructs in scf.execute_region

Extend the PFT-to-MLIR (HLFIR/FIR) lowering so unstructured DO and IF
constructs are emitted inside scf.execute_region, hiding their multi-block
CFG behind a single op. OpenACC and OpenMP lowerings that reject
multi-block content (e.g. the "unstructured do loop in combined acc
construct" TODO in OpenACC.cpp) now see a structured op instead.

Flag: -mmlir --wrap-unstructured-constructs-in-execute-region (default on).

An evaluation is wrappable iff all of the following hold:

  * wrap flag on
  * eval is parser::DoConstruct or parser::IfConstruct
  * eval.isUnstructured
  * branchesAreInternal(eval) -- every controlSuccessor in the subtree
    targets a nested eval or the constructExit
  * !hasIncomingBranch(eval) -- no outside eval branches into the body
    (PFT's synthetic IfConstruct around `if(c) goto X` absorbs label
    targets between the IF and X; the incoming-branch check excludes
    such wrappers when an outer GOTO names one of those labels)
  * does not contain a ReturnStmt -- its lowering creates the function's
    final block in the current region, which would mis-parent func.return
  * not an infinite DO and does not contain one (no LoopControl in any
    nested DO): the wrap's yield is unreachable and the body has no
    write-shaped side effects, so RegionDCE treats the wrap as trivially
    dead and drops it. Excluding the whole enclosing construct keeps
    such infinite loops visible in the parent CFG.
  * not the body DO of an enclosing OpenACCLoopConstruct or
    OpenACCCombinedConstruct -- nor one of the N collapsed iterator DOs
    reached by walking down through `collapse(N)`. Such DOs are driven
    directly into acc.loop by the OpenACC lowering, so wrapping them
    would hide the iteration from the acc.loop op.

Diagnostics
-----------
Emit per-wrap and per-function diagnostics on stderr so the wrapping is
observable from a single compile invocation:

  [wrap-unstructured] wrapped DO at <loc>
  [wrap-unstructured] wrapped IF at <loc>
  [wrap-unstructured] summary: N execute_region(s) wrapping unstructured
                      constructs at <loc>

The counter is reset per function and the summary is suppressed when no
wraps fire.

Co-authored-by: Claude Opus 4.7 <noreply at anthropic.com>

  Commit: 734638f0c7222917809905c4d6212e0836390cd7
      https://github.com/llvm/llvm-project/commit/734638f0c7222917809905c4d6212e0836390cd7
  Author: ergawy <kareem.ergawy at gmail.com>
  Date:   2026-07-16 (Thu, 16 Jul 2026)

  Changed paths:
    M flang/lib/Lower/PFTBuilder.cpp
    M flang/test/Lower/do_loop_execute_region_wrap.f90

  Log Message:
  -----------
  [flang][PFT] check every branch target in wrappability analyses

Co-Authored-By: Claude Sonnet 4.6 <noreply at anthropic.com>

  Commit: 1d99ead9459fe79e0f14240ab2837981d2efd459
      https://github.com/llvm/llvm-project/commit/1d99ead9459fe79e0f14240ab2837981d2efd459
  Author: ergawy <kareem.ergawy at gmail.com>
  Date:   2026-07-16 (Thu, 16 Jul 2026)

  Changed paths:
    M flang/lib/Lower/Bridge.cpp

  Log Message:
  -----------
  [flang][PFT-to-MLIR] pull the scf.execute_region wrap inside the DO lowering

wrapUnstructuredConstruct was previously invoked at the top of
genFIR(DoConstruct), before any of the per-branch preheader/header/body
blocks had been settled. That layout worked but was fragile: the wrap
happened before createEmptyBlocks (called inside
wrapUnstructuredConstruct) had a chance to reallocate doStmtEval.block
into the scf.execute_region entry block, and the follow-on setup
recomputed beginBlock from a stale value.

Introduce a maybeWrapAndRecalc lambda that is called in each of the DO
variants (infinite / while / increment / concurrent) after
maybeStartBlock(preheaderBlock). It:

  - conditionally creates the scf.execute_region wrap,
  - notices whether createEmptyBlocks allocated a fresh preheader
    inside the new region, and, if so, branches the scf entry to it
    and uses it as beginBlock; otherwise it falls back to the scf
    entry block itself,
  - refreshes bodyBlock/exitBlock so the rest of the DO lowering
    lands inside the wrap.

headerBlock is now computed per-branch after the wrap so that the
"step beginBlock through preheader/header/mask" walk starts from the
correct scf-internal block. No functional change when
--wrap-unstructured-constructs-in-execute-region is off (unstructured
DOs still lower to flat CFG).

Co-Authored-By: Claude Sonnet 4.6 <noreply at anthropic.com>

Compare: https://github.com/llvm/llvm-project/compare/119fbaeffa97...1d99ead9459f

To unsubscribe from these emails, change your notification settings at https://github.com/llvm/llvm-project/settings/notifications