Ben Gamari authored 7 years ago and Ben Gamari committed 7 years ago

As noted in #14226, the common block elimination pass currently
implements an extremely strict equivalence relation, demanding that two
blocks are equivalent including the names of their local registers. This
is quite restrictive and severely hampers the effectiveness of the pass.

Here we allow the CBE pass to collapse blocks which are equivalent up to
alpha renaming of locally-bound local registers. This is completely safe
and catches many more duplicate blocks.

Test Plan: Validate

Reviewers: austin, simonmar, michalt

Reviewed By: michalt

Subscribers: rwbarton, thomie

GHC Trac Issues: #14226

Differential Revision: https://phabricator.haskell.org/D3973

Arnaud Spiwack authored 7 years ago and Ben Gamari committed 7 years ago

`mkCoreApps` re-implemented `mkCoreApp` in a recursive function,
rather than using a simple `foldl'` in order to avoid repeatingly
computing the type of the function argument. I've factored the two
logic into a new (internal) function `mkCoreType` which assumes that
the type is known. `mkCoreApp` and `mkCoreApps` are thin wrappers
around it.

Differences
- The assertion failure message of `mkCoreApps` has more
  information in it.
- `mkCoreApps` now special-cases coercion argument like
  `mkCoreApp` (previously they were given to `mk_val_app` instead)

Reviewers: austin, bgamari

Reviewed By: bgamari

Subscribers: rwbarton, thomie

Differential Revision: https://phabricator.haskell.org/D3971

Ben Gamari authored 7 years ago and Ben Gamari committed 7 years ago

Some variant of this should also be added to the migration guide.

[skip ci]

Test Plan: Read it

Reviewers: goldfire, austin

Reviewed By: goldfire

Subscribers: rwbarton, thomie

GHC Trac Issues: #13391

Differential Revision: https://phabricator.haskell.org/D3966

Niklas Hambüchen authored 7 years ago and Ben Gamari committed 7 years ago

Reviewers: bgamari, austin, hvr

Reviewed By: bgamari

Subscribers: rwbarton, thomie

Differential Revision: https://phabricator.haskell.org/D3960

Niklas Hambüchen authored 7 years ago and Ben Gamari committed 7 years ago

See the added comment for details.

It's "less likely" because it can still fail if the socket happens to
have an FD larger than 1023, which can happen if many files are opened.

Until now, basic socket programs that use `hWaitForInput` were broken on
Windows.

That is because on Windows `FD_SETSIZE` defaults to 64, but pretty much
all GHC programs seem to have > 64 FDs open, so you can't actually
create a socket on which you can `select()`.

It errors with `fdReady: fd is too big` even with an example as simple
as the following (in this case, on my machine the `fd` is `284`):

  {-# LANGUAGE OverloadedStrings #-}

  import Control.Monad (forever)
  import Network.Socket
  import System.IO

  -- Simple echo server: Reads up to 10 chars from network, echoes them back.
  -- Uses the Handle API so that `hWaitForInput` can be used.
  main :: IO ()
  main = do
    sock <- socket AF_INET Stream 0
    setSocketOption sock ReuseAddr 1
    bind sock (SockAddrInet 1234 0x0100007f)
      -- 0x0100007f == 127.0.0.1 localhost
    listen sock 2
    forever $ do
      (connSock, _connAddr) <- accept sock
      putStrLn "Got connection"

      h <- socketToHandle connSock ReadWriteMode
      hSetBuffering h NoBuffering

      ready <- hWaitForInput h (5 * 1000) -- 5 seconds
      putStrLn $ "Ready: " ++ show ready

      line <- hGetLine h
      putStrLn "Got line"
      hPutStrLn h ("Got: " ++ line)
      hClose h

I'm not sure how this was not discovered earlier; for #13525 (where
`fdReady()` breaking completely was also discovered late) at least it
failed only when the timeout was non-zero, which is not used in ghc
beyond in `hWaitForInput`, but in this Windows socket case it breaks
even on the 0-timeout.

Maybe there is not actually anybody who uses sockets as handles on
Windows?

The workaround for now is to increase `FD_SETSIZE` on Windows;
increasing it is possible on Windows and BSD, see

https://stackoverflow.com/questions/7976388/increasing-limit-of-fd-setsi
ze-and-select

A real fix would be to move to IO Completion Ports on Windows, and thus
get rid of the last uses of `select()` (the other platforms already use
`poll()` but Windows doesn't have that).

Reviewers: bgamari, austin, hvr, erikd, simonmar

Reviewed By: bgamari

Subscribers: rwbarton, thomie

Differential Revision: https://phabricator.haskell.org/D3959

Niklas Hambüchen authored 7 years ago and Ben Gamari committed 7 years ago

Reviewers: bgamari, austin, erikd, simonmar

Reviewed By: bgamari

Subscribers: rwbarton, thomie

Differential Revision: https://phabricator.haskell.org/D3958

Niklas Hambüchen authored 7 years ago and Ben Gamari committed 7 years ago

Reviewers: bgamari, austin, erikd, simonmar

Reviewed By: bgamari

Subscribers: rwbarton, thomie

Differential Revision: https://phabricator.haskell.org/D3957

Niklas Hambüchen authored 7 years ago and Ben Gamari committed 7 years ago

See https://ghc.haskell.org/trac/ghc/ticket/13497#comment:17

Until now, the program

  import System.IO
  main = hWaitForInput stdin (5 * 1000)

didn't wait 5 seconds for input on Winodws, it terminated immediately.

This was because the `PeekNamedPipe()` function introduced in commit
94fee9e7 really only peeks, it doesn't block.  So if there's no data,
`fdReady(fd, msec)` would return immediately even when the given `msec`
timeout is not zero.

This commit fixes it by looping around `PeekNamedPipe()` with a `sleep(1
ms)`.

Apparently there's no better way to do this on Windows without switching
to IOCP.

In any case, this change should be strictly better than what was there
before.

Reviewers: bgamari, austin, hvr

Reviewed By: bgamari

Subscribers: Phyx, rwbarton, thomie

Differential Revision: https://phabricator.haskell.org/D3956

Niklas Hambüchen authored 7 years ago and Ben Gamari committed 7 years ago

Reviewers: bgamari, austin, hvr

Reviewed By: bgamari

Subscribers: rwbarton, thomie

Differential Revision: https://phabricator.haskell.org/D3955

Niklas Hambüchen authored 7 years ago and Ben Gamari committed 7 years ago

This fixes #13497 for Windows -- at least for the `if (isSock)` part; I
haven't investigated the case where it's not a socket yet.

Solved by copying the new current-time based waiting logic from the
non-Windows implementation above.

Reviewers: bgamari, austin, hvr

Reviewed By: bgamari

Subscribers: rwbarton, thomie

Differential Revision: https://phabricator.haskell.org/D3954

Niklas Hambüchen authored 7 years ago and Ben Gamari committed 7 years ago

This is done by reusing the existing cross-platform
`getProcessElapsedTime()` function, which already provides nanosecond
monotonic clocks, and fallback for platforms that don't have those.

To do this, `getProcessElapsedTime()` had to be moved from a private RTS
symbol into the public interface.

Accuracy is improved in 2 ways:

* Use of the monotonic clock where available
* Measuring the total time spent waiting instead of a sum
  of intervals (between which there are small gaps)

Reviewers: bgamari, austin, hvr, erikd, simonmar

Reviewed By: bgamari

Subscribers: rwbarton, thomie

Differential Revision: https://phabricator.haskell.org/D3953

Herbert Valerio Riedel authored 7 years ago and Ben Gamari committed 7 years ago

This switches the compiler/ component to get compiled with
-XNoImplicitPrelude and a `import GhcPrelude` is inserted in all
modules.

This is motivated by the upcoming "Prelude" re-export of
`Semigroup((<>))` which would cause lots of name clashes in every
modulewhich imports also `Outputable`

Reviewers: austin, goldfire, bgamari, alanz, simonmar

Reviewed By: bgamari

Subscribers: goldfire, rwbarton, thomie, mpickering, bgamari

Differential Revision: https://phabricator.haskell.org/D3989

Ben Gamari authored 7 years ago and Ben Gamari committed 7 years ago

In general ticks are problematic for join point analysis as described
in #14242.  However, source notes are intended to be a best-effort
annotation which shouldn't interfere with optimization. Special-case
these to ensure that tail-call information is still correct, even in the
presence of source note
ticks.

Test Plan: Validate

Reviewers: simonpj, austin

Reviewed By: simonpj

Subscribers: rwbarton, thomie

GHC Trac Issues: #14242

Differential Revision: https://phabricator.haskell.org/D3978

Ben Gamari authored 7 years ago and Ben Gamari committed 7 years ago

This fixes #14221, where the NCG and the DWARF code were apparently
giving two different names to the same block.

Test Plan: Validate with DWARF support enabled.

Reviewers: simonmar, austin

Subscribers: rwbarton, thomie

GHC Trac Issues: #14221

Differential Revision: https://phabricator.haskell.org/D3977

Ben Gamari authored 7 years ago and Ben Gamari committed 7 years ago

orzo in `#ghc` reported seeing a crash due to the retainer profiler encountering
a BLOCKING_QUEUE closure, which isRetainer didn't know about. I performed an
audit to make sure that all of the valid closure types were listed; they
weren't. This is my guess of how they should appear.

Test Plan: Validate

Reviewers: simonmar, austin, erikd

Reviewed By: simonmar

Subscribers: rwbarton, thomie

GHC Trac Issues: #14235

Differential Revision: https://phabricator.haskell.org/D3967

This bug was shown up by Trac #14237.  It turned out
to be an outright error in TcSimplify.neededEvVars,
easily fixed.

I improved the comments.

This `#if 0`/`#endif` block has been around for over 10 years and
it became truly redundant in 36104d7a

This now becomes possible due to the introduction of the
Semigroup=>Monoid superclass relation (see #14191).

Reviewers: ekmett, RyanGlScott, austin, bgamari

Reviewed By: ekmett, RyanGlScott, bgamari

Subscribers: rwbarton, thomie

Differential Revision: https://phabricator.haskell.org/D3972

Summary:
This bumps our LLVM version requirement to LLVM5,
which will likely be released before GHC 8.4.

Test Plan: validate

Reviewers: austin, hvr, bgamari

Reviewed By: bgamari

Subscribers: rwbarton, thomie, erikd

Differential Revision: https://phabricator.haskell.org/D3797

This adds a function to the RTS linker API which lets the
user check the status of dynamically linked objects.

It was initially proposed by @afarmer in D2068.
It's useful for testing the linker and also for detecting retention
problems in production.

It takes a path, because it's easier to use path as key instead of producing
some stable handle.

It returns an enum instead of bool, because I see no reason for destroying
information. All the complexity is already out in the open, so there's
nothing to save the users from.

Test Plan: ./validate

Reviewers: simonmar, Phyx, bgamari, austin, erikd

Reviewed By: Phyx, bgamari

Subscribers: rwbarton, afarmer, thomie

Differential Revision: https://phabricator.haskell.org/D3963

f9c6d53f led to #14036. The
problem turned out to be rather simple: the `obj` pointer was
being tagged using `obj + arity`. Because this is C, that's done
with *pointer arithmetic*, which is not at all what we want. Add
appropriate casts.

Reviewers: austin, bgamari, erikd, simonmar

Reviewed By: bgamari

Subscribers: rwbarton, thomie

GHC Trac Issues: #14036

Differential Revision: https://phabricator.haskell.org/D3983

Ben Gamari authored 7 years ago and Ben Gamari committed 7 years ago

Test Plan: Validate

Reviewers: hvr, RyanGlScott, austin

Reviewed By: RyanGlScott

Subscribers: rwbarton, thomie

Differential Revision: https://phabricator.haskell.org/D3982

Otherwise compilation fails with,

    libraries/base/Data/Type/Equality.hs:145:4: error:
        • Data constructor ‘HRefl’ constrains the choice of kind parameter:
            k2 ~ k2
          Use TypeInType to allow this
        • In the definition of data constructor ‘HRefl’
          In the data type declaration for ‘:~~:’
        |
    145 |    HRefl :: a :~~: a
        |    ^

Reviewers: austin, bgamari

Reviewed By: bgamari

Subscribers: rwbarton, thomie

Differential Revision: https://phabricator.haskell.org/D3970

Alan Zimmerman authored 7 years ago and Ben Gamari committed 7 years ago

Prior to this, in the RenamedSource for

    module Renaming.RenameInExportedType
      (
      MyType (NT)
      ) where

    data MyType = MT Int | NT

The (NT) was given the location of MyType earlier on the line in the
export list.

Also the location was discarded for any field labels, and replaced with
a `noLoc`.

Test Plan: ./validate

Reviewers: bgamari, austin

Reviewed By: bgamari

Subscribers: rwbarton, thomie

GHC Trac Issues: #14189

Differential Revision: https://phabricator.haskell.org/D3968

Niklas Hambüchen authored 7 years ago and Ben Gamari committed 7 years ago

Reviewers: bgamari, austin, hvr

Reviewed By: bgamari

Subscribers: rwbarton, thomie

Differential Revision: https://phabricator.haskell.org/D3952

`isIrrefutableHsPat` should always return `False` for unboxed sum
patterns (`SumPat`s), since they always have at least one other
corresponding pattern of the same arity (since the minimum arity for a
`SumPat` is 2). Failure to do so causes incorrect code to be generated
for pattern synonyms that use unboxed sums, as shown in #14228.

Test Plan: make test TEST=T14228

Reviewers: austin, bgamari, simonpj

Reviewed By: simonpj

Subscribers: simonpj, rwbarton, thomie

GHC Trac Issues: #14228

Differential Revision: https://phabricator.haskell.org/D3951

Test Plan: make test TEST=T13870

Reviewers: RyanGlScott, austin, bgamari, mpickering

Reviewed By: mpickering

Subscribers: mpickering, rwbarton, thomie, RyanGlScott

Tags: #ghc

GHC Trac Issues: #13870

Differential Revision: https://phabricator.haskell.org/D3940

Tamar Christina authored 7 years ago and Ben Gamari committed 7 years ago

ignore all untracked changes in all submodules from the top level.
This matches the older settings for the other submodules and stops
litering `git status` output with files created by ./boot and make.

Reviewers: austin, bgamari

Reviewed By: bgamari

Subscribers: hvr, rwbarton, thomie

Differential Revision: https://phabricator.haskell.org/D3899

[skip ci]

This will unblock perf.haskell.org, which got stuck after 8ae263ce.

Richard Eisenberg authored 7 years ago and Ben Gamari committed 7 years ago

This commit fixes several things:

1. RuntimeRep arg suppression was overeager for *visibly*-quantified
RuntimeReps, which should remain.

2. The choice of whether to used a Named TyConBinder or an anonymous
was sometimes wrong. Now, we do an extra little pass right before
constructing the tycon to fix these.

3. TyCons that normally cannot appear unsaturated can appear unsaturated
in :kind. But this fact was not propagated into the type checker.
It now is.

Richard Eisenberg authored 7 years ago and Ben Gamari committed 7 years ago

test case: dependent/should_compile/T13938

Richard Eisenberg authored 7 years ago and Ben Gamari committed 7 years ago

test case: typecheck/should_fail/T13929

Richard Eisenberg authored 7 years ago and Ben Gamari committed 7 years ago

This was a lurking bug discovered on the hunt for #13910, but
it doesn't fix that bug. The old version of rejigConRes was
just wrong, forgetting to propagate a kind-change.

Richard Eisenberg authored 7 years ago and Ben Gamari committed 7 years ago

GHC was complaining about numbers of arguments when the real
problem is impredicativity.

test case: typecheck/should_fail/T13909

grep was considering the output to be binary on my machine, resulting in a
spurious test failure.