[Git][ghc/ghc][wip/az/epa-remove-addepann] 11 commits: driver: fix runWorkerLimit on wasm

Alan Zimmerman (@alanz) gitlab at gitlab.haskell.org
Mon Oct 7 19:53:29 UTC 2024



Alan Zimmerman pushed to branch wip/az/epa-remove-addepann at Glasgow Haskell Compiler / GHC


Commits:
ceca9efb by Cheng Shao at 2024-10-06T02:18:31+00:00
driver: fix runWorkerLimit on wasm

This commit fixes link-time unresolved symbol errors for sem_open etc
on wasm, by making runWorkerLimit always behave single-threaded. This
avoids introducing the jobserver logic into the final wasm module and
thus avoids referencing the posix semaphore symbols.

- - - - -
135fd1ac by Torsten Schmits at 2024-10-06T02:18:31+00:00
Parallelize getRootSummary computations in dep analysis downsweep

This reuses the upsweep step's infrastructure to process batches of
modules in parallel.

I benchmarked this by running `ghc -M` on two sets of 10,000 modules;
one with a linear dependency chain and the other with a binary tree.
Comparing different values for the number of modules per thread
suggested an optimum at `length targets `div` (n_cap * 2)`, with results
similar to this one (6 cores, 12 threads):

```
Benchmark 1: linear 1 jobs
  Time (mean ± σ):      1.775 s ±  0.026 s    [User: 1.377 s, System: 0.399 s]
  Range (min … max):    1.757 s …  1.793 s    2 runs

Benchmark 2: linear 6 jobs
  Time (mean ± σ):     876.2 ms ±  20.9 ms    [User: 1833.2 ms, System: 518.6 ms]
  Range (min … max):   856.2 ms … 898.0 ms    3 runs

Benchmark 3: linear 12 jobs
  Time (mean ± σ):     793.5 ms ±  23.2 ms    [User: 2318.9 ms, System: 718.6 ms]
  Range (min … max):   771.9 ms … 818.0 ms    3 runs
```

Results don't differ much when the batch size is reduced to a quarter
of that, but there's significant thread scheduling overhead for a size
of 1:

```
Benchmark 1: linear 1 jobs
  Time (mean ± σ):      2.611 s ±  0.029 s    [User: 2.851 s, System: 0.783 s]
  Range (min … max):    2.591 s …  2.632 s    2 runs

Benchmark 2: linear 6 jobs
  Time (mean ± σ):      1.189 s ±  0.007 s    [User: 2.707 s, System: 1.103 s]
  Range (min … max):    1.184 s …  1.194 s    2 runs

Benchmark 3: linear 12 jobs
  Time (mean ± σ):      1.097 s ±  0.006 s    [User: 2.938 s, System: 1.300 s]
  Range (min … max):    1.093 s …  1.101 s    2 runs
```

Larger batches also slightly worsen performance.

- - - - -
535a2117 by Daniel Díaz at 2024-10-06T09:51:46-04:00
Clarify the meaning of "exactly once" in LinearTypes

Solves documentaion issue #25084.

- - - - -
92f8939a by Krzysztof Gogolewski at 2024-10-06T09:52:22-04:00
Only allow (a => b) :: Constraint rather than CONSTRAINT rep

Fixes #25243

- - - - -
4a2f0f13 by Alan Zimmerman at 2024-10-07T05:16:54-04:00
EPA: Remove unused hsCaseAnnsRest

We never populate it, so remove it.

- - - - -
5099057b by John Paul Adrian Glaubitz at 2024-10-07T05:17:40-04:00
rts: Fix invocation of __ieee_set_fp_control() on alpha-linux

Fixes the following error when building GHC on alpha-linux:

rts/posix/Signals.c: In function ‘initDefaultHandlers’:

rts/posix/Signals.c:709:5: error:
     error: implicit declaration of function ‘ieee_set_fp_control’ [-Wimplicit-function-declaration]
      709 |     ieee_set_fp_control(0);
          |     ^~~~~~~~~~~~~~~~~~~
    |
709 |     ieee_set_fp_control(0);
    |

- - - - -
c9590ba0 by Teo Camarasu at 2024-10-07T05:18:17-04:00
Add changelog entries for !12479

- - - - -
bf9c9566 by Matthew Pickering at 2024-10-07T13:19:30-04:00
javascript: Read fields of ObjectBlock lazily

When linking a module with a large dependency footprint too much of the
object files were forced during linking. This lead to a large amount of
memory taken up by thunks which would never be forced

On the PartialDownsweep test this halves the memory required (from 25G
to 13G).

Towards #25324

-------------------------
Metric Increase:
    size_hello_obj
-------------------------

- - - - -
571329df by Matthew Pickering at 2024-10-07T13:20:06-04:00
ci: Run the i386 validation job when i386 label is set

This is helpful when making changes to base and must update the
javascript and i386 base exports files.

- - - - -
e68f9aaf by Matthew Pickering at 2024-10-07T13:20:42-04:00
Rewrite partitionByWorkerSize to avoid pattern match checker bug

With `-g3` the pattern match checker would warn about these incomplete
patterns. This affects the debug_info builds on CI.

```
    Pattern match(es) are non-exhaustive
    In an equation for ‘go’:
        Patterns of type ‘[a]’, ‘[a]’, ‘[SpecFailWarning]’ not matched:
            (_:_) _ _
     |
2514 |         go [] small warnings = (small, warnings)
     |         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^...
```

Workaround for #25338

- - - - -
7602d6ed by Alan Zimmerman at 2024-10-07T20:52:52+01:00
EPA: Remove [AddEpAnn] from (most of) HsExpr

EPA: introduce EpAnnLam for lambda annotationsi, and remove `glAA`
from `Parser.y`, it is the same as `glR`

EPA: Remove unused annotation from XOpApp

EPA: Use EpToken for XNPat and XNegApp

EPA: specific anns for XExplicitTuple / XTuplePat / sumPatParens.

EPA: Use specific annotation for MultiIf

EPA: Move annotations into FunRhs

EPA: Remove [AddEpAnn] from SigPat and ExprWithTySig

EPA: Remove [AddEpAnn] from ArithSeq

EPA: Remove [AddEpAnn] from HsProc

EPA: Remove [AddEpAnn] from HsStatic

EPA: Remove [AddEpAnn] from BindStmt

EPA: Remove [AddEpAnn] from TransStmt

EPA: Remove [AddEpAnn] from HsTypedSplice

EPA: Remove [AddEpAnn] from HsUntypedSpliceExpr

- - - - -


30 changed files:

- .gitlab/generate-ci/gen_ci.hs
- .gitlab/jobs.yaml
- compiler/GHC/Core/Opt/SpecConstr.hs
- compiler/GHC/Driver/Make.hs
- compiler/GHC/Hs/Dump.hs
- compiler/GHC/Hs/Expr.hs
- compiler/GHC/Hs/Pat.hs
- compiler/GHC/Hs/Utils.hs
- compiler/GHC/HsToCore/Arrows.hs
- compiler/GHC/HsToCore/Binds.hs
- compiler/GHC/HsToCore/Expr.hs
- compiler/GHC/HsToCore/Match.hs
- compiler/GHC/Parser.y
- compiler/GHC/Parser/Annotation.hs
- compiler/GHC/Parser/PostProcess.hs
- compiler/GHC/Rename/Bind.hs
- compiler/GHC/Rename/Utils.hs
- compiler/GHC/StgToJS/Object.hs
- compiler/GHC/StgToJS/Types.hs
- compiler/GHC/Tc/Deriv/Functor.hs
- compiler/GHC/Tc/Deriv/Generate.hs
- compiler/GHC/Tc/Gen/Arrow.hs
- compiler/GHC/Tc/Gen/HsType.hs
- compiler/GHC/Tc/Gen/Match.hs
- compiler/GHC/Tc/Module.hs
- compiler/GHC/Tc/TyCl/Instance.hs
- compiler/GHC/Tc/TyCl/PatSyn.hs
- compiler/GHC/Tc/TyCl/Utils.hs
- compiler/GHC/ThToHs.hs
- compiler/GHC/Unit/Finder.hs


The diff was not included because it is too large.


View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/1ecfc64d67184d31c332076022cc8650eba588a6...7602d6ed59fe5dd28046182837d82bcb07476b42

-- 
View it on GitLab: https://gitlab.haskell.org/ghc/ghc/-/compare/1ecfc64d67184d31c332076022cc8650eba588a6...7602d6ed59fe5dd28046182837d82bcb07476b42
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/20241007/ea38ec6b/attachment-0001.html>


More information about the ghc-commits mailing list