Added -debug-assert-immediately and -debug-crash-immediately, which cause an
llvm_unreachable or LLVM_BUILTIN_TRAP to execute during argument parsing.
Added -debug-assert-after-parse and -debug-crash-after-parse, which cause an
llvm_unreachable or LLVM_BUILTIN_TRAP to execute after calling
CompilerInstance::performParse().
This fixes <rdar://problem/16013025>.
Swift SVN r13653
specialize on polymorphic arguments.
This can be enabled with: -sil-devirt-threshold 500.
It currently improves RC4 (when enabled) by 20%, but will be much more
important after Michael's load elimination with alias analysis lands.
This implementation is suitable for experimentation. Superficial code
reviews are also welcome. Although be warned that the design is overly
complex and I plan to rewrite it. I initially abandoned the idea of
incrementally specializing one function at a time, thinking that we
need to analyze full chains. However, I since realized after talking
to Nadav that the incremental approach can be made to work. A lot of
book-keeping will go away with that change.
TODO:
- Resolve protocol argument types. Currently we assume they can be
reinitialized at applies, but I don't think they can unless they are
@inouts. This is an issue with the existing local devirtualizer
that prevents it working across calls.
- Properly mangle the specialized methods. Find existing
specializations by demangling rather than maintaining a map.
- Rewrite the logic for specializing chains for simplicity.
- Enable by default.
Swift SVN r13642
Because this is useful in testing, I've left in a frontend option
-enable-source-import for both swift and swift-ide-test that sidesteps the
module restriction. Right now, though, this is the right thing to avoid
users running into strange issues when they import another file within
their module and Swift treats it as a separate module.
<rdar://problem/15937521>
Swift SVN r13248
Part of the migration to the new driver. With this commit, the only
failures in the test suite using the new frontend are features we don't
intend to port over. Hooray!
Swift SVN r13198
SILSerializeAll and EmitVerboseSIL are /not/ being moved because they are
options controlling the output, not about SILGen and SIL passes.
No functionality change.
Swift SVN r13197
Plumbing this through to the inliner necessitated the creation of a
SILOptions class (like FrontendOptions and IRGenOptions). I'll move
more things into this soon.
One change: for compatibility with the new driver, the option must be
specified as "-sil-inline-threshold 50" instead of "-sil-inline-threshold=50".
(We're really trying to be consistent about joined-equals vs. separate
in the new frontend.)
Swift SVN r13193
...and probably should not be emitted before optimization passes.
The new frontend got this wrong, and the old one pretended to allow it but
then silently didn't write a module at all.
Swift SVN r13189
Also, restructure so that the option isn't declared in a random library file.
(And do the same with "-sil-link-all".)
Part of the migration to the new driver.
Swift SVN r13184
Added DiagnosticsDriver.def and DiagnosticsDriver.h for driver-only diagnostics.
(Diagnostics which are shared with the frontend remain in
DiagnosticsFrontend.{def,h}.)
Added a DiagnosticEngine& to Compilation, so that it can emit diagnostics for
events which occur while performing Jobs.
Replaced all of the locations where we were manually printing error messages to
emitting real diagnostics, adding diagnostics if necessary.
Updated Driver::buildCompilation() so that it fails early if any errors were
encountered.
Updated test/Driver/actions.swift to pass a -module-name for multi-input tests.
Swift SVN r13175
This keeps us from having to deal with fat swiftmodules for now.
In the long run we're hoping to solve this problem with build configurations,
so that a single module file can support multiple architectures.
(See <rdar://problem/15056323>)
<rdar://problem/15204953>
Swift SVN r13135
This adds some ugliness in the current swift binary because we weren't
bothering to set a requested action before, but it keeps things simpler
elsewhere. (Thanks, Connor.)
Swift SVN r13067
This activates a special mode where code is also generated for input files.
We don't really care about this for -i mode, but it makes -repl mode a bit
nicer.
Swift SVN r13044
Adjusted how this option is handled: it is now set in IRGenOptions. If set,
this prevents the relevant passes from being added at all, instead of making
them no-ops.
Swift SVN r13005
If the OutputFilename points to a directory, fall back to the first
InputFilename.
This matches the behavior of the swift executable, and matches what would be
chosen if we could resolve the correct OutputFilename without knowing the module
name.
Swift SVN r12935
If the user did not specify an OutputFilename, or if the user specified a
directory with -o, calculate a default OutputFilename based on the inputs,
module name, and requested action.
This matches the behavior of the swift executable.
Swift SVN r12880
TargetOptions only contained the target triple, which was duplicated in
IRGenOptions. These could get out-of-sync, which would cause issues during
IRGen. Since nothing was using TargetOptions other than CompilerInvocation,
removed TargetOptions in favor of making IRGenOptions the canonical home of the
target triple.
Swift SVN r12869
This option was never honored, and the main functionality that this option
would have implemented has been subsumed by -output-file-map.
Swift SVN r12800
The new format is "-<option>" and "-<option>-path". As a result:
- Renamed -serialized-diagnostics-path to -serialize-diagnostics-path.
- Renamed -module-output-path to -emit-module-path.
Swift SVN r12614
Thanks to the way we've set up our diagnostics engine, there's not actually
a reason for /everything/ to get rebuilt when /one/ diagnostic changes.
I've split them up into five categories for now: Parse, Sema, SIL, IRGen,
and Frontend, plus a set of "Common" diagnostics that are used in multiple
areas of the compiler. We can massage this later.
No functionality change, but should speed up compile times!
Swift SVN r12438
Added support for the -emit-module and -module-output-path options. -emit-module
and -module-output-path function similarly to -serialize-diagnostics and
-serialized-diagnostics-path: notably, -module-output-path implies -emit-module,
but -emit-module can be passed without -module-output-path and a default path
will be determined based on other inputs.
Added support for the -module-link-name option, which specifies the name of the
library to link against when importing the generated module.
Note that support for these options is only implemented in the integrated
frontend; driver-level support for these options is forthcoming.
Swift SVN r12301
Reworked the -serialize-diagnostics option, so that it is now just a flag
indicating that the frontend should generate serialized diagnostics. The path at
which the diagnostics will be serialized is now specified by the
-serialized-diagnostics-path option, which is a frontend-only option. (The
frontend treats -serialized-diagnostics-path as implying -serialize-diagnostics.)
If -serialize-diagnostics is passed but -serialized-diagnostics-path is not
passed, the frontend will choose an output path from a few default values:
- If the frontend has a non-stdout output path, replace that path's extension
with .dia.
- If there is a primary input filename, use that input's base name as the base
name for the serialized diagnostics file.
- Otherwise, use the module name as the base name for the serialized diagnostics
file.
Added support for setting up a serialized diagnostics DiagnosticConsumer in
frontend_main() if FrontendOptions::SerializedDiagnosticsPath is non-empty.
Swift SVN r12251
Added checks to determine if the input should be treated as SIL, Library, REPL, or Main, based on the command line arguments and inputs.
Swift SVN r12176
We should be able to cut out another layer of IRGen grime now.
This does XFAIL one test, test/Prototypes/TextFormatting.swift, which fails because of a weird archetype ordering in a nested substitution list. This should get sorted out by switching to interface types, so I'm going to let it go until then.
Swift SVN r11618
MainInputFilename and OutputFilename are pulled from the FrontendOptions, while Triple is parsed directly from the ArgList.
Eventually, MainInputFilename and OutputFilename should likely be removed from IRGenOptions, in favor of the values in FrontendOptions, while Triple will be in a separate TargetOptions class.
Swift SVN r11455
