[Git][ghc/ghc][wip/marge_bot_batch_merge_job] 5 commits: Inline join points for rhs without free vars
Marge Bot (@marge-bot)
gitlab at gitlab.haskell.org
Mon Feb 17 13:36:34 UTC 2025
Marge Bot pushed to branch wip/marge_bot_batch_merge_job at Glasgow Haskell Compiler / GHC
Commits:
6c05428b by Rodrigo Mesquita at 2025-02-17T08:36:13-05:00
Inline join points for rhs without free vars
While investigating #25170, we ran into a program (T16473) that allocated 67%
more because of a join point that failed to inline.
Note [Duplicating join points] explains why we want to be conservative
when inlining join points, using as an example a join point that
captures a free variable `f` that becomes available in the continuation
`blah` for further optimisations, as opposed to being lambda-abstracted.
However, when the RHS of the join point has no free variables and is
trivial, the same argument does not apply, and there's nothing to gain
from preserving it.
On the contrary, not inlining these trivial join points such as
$j f x = K f x |> co
can be actively harmful as they prevent useful optimisations from firing
on the known constructor application. #25723 is such an example.
Therefore, we've extended `uncondInlineJoin` to allow duplicating such closed
trivial join points. See the updated Note [Duplicating join points] for
further details.
Additionally, merge the guards in uncondInlineJoin for point DJ3(b) anad DJ3(c) of
Note [Duplicating join points] to avoid an unnecessary traversal in the
call to `collectArgs`; it's also more uniform.
Co-authored-by: Simon Peyton Jones <simon.peytonjones at gmail.com>
Fixes #25723
- - - - -
5731f6ff by M Farkas-Dyck at 2025-02-17T08:36:18-05:00
Scrub some partiality in `GHC.Tc.Gen.Match`.
In particular, we construct a list of the same length as another list, then loop over both and panic if their lengths are unequal. We can avoid this.
- - - - -
3aba62d6 by M Farkas-Dyck at 2025-02-17T08:36:18-05:00
Make list of `ParStmtBlock` in `ParStmt` `NonEmpty`.
In the ParStmt constructor Language.Haskell.Syntax.Expr.StmtLR, the 2nd argument, the list of ParStmtBlocks, must be NonEmpty; make it so.
- - - - -
86346338 by M Farkas-Dyck at 2025-02-17T08:36:18-05:00
GHC.Tc.Gen.Match: Added type signatures for `loop` functions.
- - - - -
d028b8de by Sylvain Henry at 2025-02-17T08:36:21-05:00
Remove a bunch of Makefiles from old build system
- - - - -
32 changed files:
- compiler/GHC/Core/Opt/Simplify/Iteration.hs
- compiler/GHC/Core/Unfold.hs
- compiler/GHC/Hs/Expr.hs
- compiler/GHC/Hs/Utils.hs
- compiler/GHC/HsToCore/ListComp.hs
- compiler/GHC/HsToCore/Quote.hs
- compiler/GHC/Parser.y
- compiler/GHC/Rename/Expr.hs
- compiler/GHC/Tc/Errors/Ppr.hs
- compiler/GHC/Tc/Errors/Types.hs
- compiler/GHC/Tc/Gen/Match.hs
- compiler/GHC/Tc/Zonk/Type.hs
- compiler/GHC/ThToHs.hs
- compiler/GHC/Types/Error/Codes.hs
- compiler/Language/Haskell/Syntax/Expr.hs
- docs/users_guide/9.14.1-notes.rst
- + testsuite/tests/perf/compiler/T25723.hs
- + testsuite/tests/perf/compiler/T25723.stdout
- testsuite/tests/perf/compiler/all.T
- testsuite/tests/simplCore/should_compile/T24229a.stderr
- testsuite/tests/simplCore/should_compile/T24229b.stderr
- + testsuite/tests/th/EmptyParStmt.hs
- + testsuite/tests/th/EmptyParStmt.stderr
- testsuite/tests/th/all.T
- − utils/deriveConstants/Makefile
- − utils/genprimopcode/Makefile
- − utils/ghc-pkg/Makefile
- − utils/hp2ps/Makefile
- − utils/iserv/Makefile
- − utils/remote-iserv/Makefile
- − utils/runghc/Makefile
- − utils/unlit/Makefile
The diff was not included because it is too large.
View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/161cd0543eac6297bae389aa3ab927cda81f0540...d028b8de5372468366c3f3a952d0c62d60652c60
--
View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/161cd0543eac6297bae389aa3ab927cda81f0540...d028b8de5372468366c3f3a952d0c62d60652c60
You're receiving this email because of your account on gitlab.haskell.org.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.haskell.org/pipermail/ghc-commits/attachments/20250217/9b10c499/attachment.html>
More information about the ghc-commits
mailing list