Fixes #8456. Previous version of control flow optimisations
did not update the list of block predecessors, leading to
unnecessary duplication of blocks in some cases. See Trac
and comments in the code for more details.

The only substantive change here is to change "==" into ">=" in
the Note [Always false stack check] code.  This is semantically
correct, but won't have any practical impact.

This patches fixes two separate instances of the bug,

 * one in tc_ax_branches (Trac #8449)

 * one in type/kind applications in IfaceExpr
   (hence the new tcIfaceApps)

The latter was reported by Iavor, no ticket

Fix a bug introduced in 94125c97.
See Note [Always false stack check]

This fixes #8452.

Compiled HscMain.o is now smaller.

I am removing old loopification code that has been commented out
for long long time. We now have loopification implemented in
the code generator (see Note [Self-recursive tail calls]) so we
won't need to resurect this old code.

When compiling a function we can determine how much stack space it will
use. We therefore need to perform only a single stack check at the beginning
of a function to see if we have enough stack space. Instead of referring
directly to Sp - as we used to do in the past - the code generator uses
(old + 0) in the stack check. Stack layout phase turns (old + 0) into Sp.

The idea here is that, while we need to perform only one stack check for
each function, we could in theory place more stack checks later in the
function. They would be redundant, but not incorrect (in a sense that they
should not change program behaviour). We need to make sure however that a
stack check inserted after incrementing the stack pointer checks for a
respectively smaller stack space. This would not be the case if the code
generator produced direct references to Sp. By referencing (old + 0) we make
sure that we always check for a correct amount of stack: when converting
(old + 0) to Sp the stack layout phase takes into account changes already
made to stack pointer. The idea for this change came from observations made
while debugging #8275.

krakrjak authored 11 years ago and Krzysztof Gogolewski committed 11 years ago

Version of Happy required is now 1.19.

Richard Eisenberg authored 11 years ago and Krzysztof Gogolewski committed 11 years ago

We no longer check capitalization (or colons) in names that come
from TH, according to the commentary in #7667.

Remove remains of old role syntax

For details see the comment on `interactTopSub`.

This is actually a no-op wrt the source-code and is only performed to show
that upstream is identical to our lagged repository.

Signed-off-by: Herbert Valerio Riedel <hvr@gnu.org>

This commit moves the check from parser to renamer.

Spotted by Bertram Felgenhauer.

Signed-off-by: Austin Seipp <austin@well-typed.com>

Signed-off-by: Austin Seipp <austin@well-typed.com>

Mac OS X 10.9 mkdir is apparently stricter than the Mac OS X 10.8
mkdir about which paths are considered valid arguments. For example,
in a typical build on Mac OS X 10.9, the first of the following
invocations of mkdirhier.sh will succeed but the second will fail:

"inplace/bin/mkdirhier"   utils/ghc-cabal/dist/build/tmp//.  # WORKS
"inplace/bin/mkdirhier"   bootstrapping/.                    # FAILS

Simply prefixing the path arguments with "./" causes both to succeed:

"inplace/bin/mkdirhier" ./utils/ghc-cabal/dist/build/tmp//.  # WORKS
"inplace/bin/mkdirhier" ./bootstrapping/.                    # WORKS

Testing indicates failure on paths satisfying all of these criteria:

- path is suffixed with "/."
- path is only 1 level deep (e.g., "foo/."; _not_ "foo/bar/.")
- path is _not_ prefixed with "./"

This workaround prefixes "./" to the path argument passed to mkdir.

Signed-off-by: Austin Seipp <austin@well-typed.com>

Authored-by: Gergely Risko <gergely@risko.hu>
Signed-off-by: Austin Seipp <austin@well-typed.com>

Authored-by: Gergely Risko <gergely@risko.hu>
Signed-off-by: Austin Seipp <austin@well-typed.com>

Signed-off-by: Edward Z. Yang <ezyang@mit.edu>

Patch by errge

rwbarton authored 11 years ago and Erik de Castro Lopo committed 11 years ago

Signed-off-by: Erik de Castro Lopo <erikd@mega-nerd.com>

Signed-off-by: Edward Z. Yang <ezyang@mit.edu>