- Outlaw duplicate input files, fix driver, fix tests, and add test.
- Reflect that no buffer is present without a (possibly pseudo) named file.
- Reflect fact that every input has a (possible pseudo) name.
- Break up CompilerInstance::setup.
Don't bail on dups.
Unify actionHasOutput w/ doesActionProduceOutput & correct both.
Do not set an output filename if there is no output.
Add test to ensure no output is created in immediate mode.
Restore NFC module name computation.
passes
The test case looks for the Objective C ARC contraction pass. It uses
-Xllvm -time-passes and then greps for the pass. It seems that
-time-passes is not always printing the pass. Use -Xllvm
-debug-pass=Structure instead that should print the added passes
reliably.
rdar://35796399
Otherwise, we leave optimization opportunities on the table, and in
some cases even fail to remove marker intrinsics inserted by earlier
optimization.
Background: under -embed-bitcode, compilation happens in two stages: a
"frontend" job that compiles Swift code, generates LLVM IR, optimizes
it, and then emits a .bc file; and a "backend" job that takes that
.bc, converts it to assembly, and emits an object file with the
original bitcode embedded. However, there are actually optimization
passes that run before and during that "convert to assembly" step that
were getting completely skipped.
rdar://problem/34864094
When merging partial SIL modules we only link in function bodies defined
in the current module, so we might encounter functions with shared
linkage and no body.
Since pulling in these functions from other modules is a waste of time,
run the verifier in "single function" mode in this case.
Note that when the module is ultimately used and one of these functions
is deserialized, we should link in all downstream functions with shared
linkage, and failure to do so will be caught by the SIL verifier then.
Fixes <rdar://problem/34469704>.
* Unify the capitalization across all user-visible error messages (fatal errors, assertion failures, precondition failures) produced by the runtime, standard library and the compiler.
* Update some more tests to the new expectations.
Avoids a bogus "argument unused" warning, since bitcode compilation
doesn't bother to set up much of a Clang AST context. (-embed-bitcode
compiles in two steps: source -> bitcode, then bitcode -> object
file. This is about the second step.)
Patch by Bob Wilson, who's currently on vacation but wanted this to
get in sooner rather than later.
rdar://problem/31372950
Previously we would drop all serialized SIL from partial swiftmodule
files generated while compiling source in non-WMO mode; all that was
missing was linking it in.
This adds a frontend flag, and a test; driver change is coming up
next.
Progress on <rdar://problem/18913977>.
Add an -enforce-exclusivity=... flag to control enforcement of the law of
exclusivity. The flag takes one of four options:
"checked": Perform both static (compile-time) and dynamic (run-time) checks.
"unchecked": Perform only static enforcement. This is analogous to -Ounchecked.
"dynamic-only": Perform only dynamic checks. This is for staging purposes.
"none": Perform no checks at all. This is also for staging purposes.
The default, for now, is "none".
The intent is that in the fullness of time, "checked" and "unchecked" will
be the only legal options with "checked" the default. That is, static
enforcement will always be enabled and dynamic enforcement will be enabled
by default.
Instead of appending a character for each substitution, we now prefix the substitution with the repeat count, e.g.
AbbbbB -> A5B
The same is done for known-type substitutions, e.g.
SiSiSi -> S3i
This significantly shrinks mangled names which contain large lists of the same type, like
func foo(_ x: (Int, Int, Int, Int, Int, Int, Int, Int, Int, Int, Int, Int))
rdar://problem/30707433
The Swift compiler uses files with an extension of ".swiftdeps" to store
information about cross-file dependencies. These files are read in at the
start of compilation to compute a dependency graph, and updated as compilation
proceeds. However, because these files are updated on every build, an
issue with dependency analysis is hard to reproduce—the inputs have been
lost.
Address this by renaming swiftdeps files that are about to be
overwritten, to '.swiftdeps~'. This preserves dependency information
from the most recent compilation (but no further back).
The original support for embedded bitcode used appending linkage for the
magic internal variables that hold the bitcode and command line options,
but that private linkage is a better fit. The only real reason for
appending linkage was to prevent those variables from being optimized away.
r269706 limits the use of appending linkage so that it cannot be used for
those variables, so this switches to use private linkage and keep the
variables alive with llvm.compiler.used. This is basically copied from
clang r269679. rdar://problem/28685198.
As of the swift-3.1-branch versions of Clang/LLVM, embedded bitcode is now
working well enough that the tests can be reenabled. rdar://problem/26247134
Autolinking was added to the frontend in 22912bc3b. It was disabled on
Linux in 198402dcf, and further constrained to be disabled on "linux-gnu"
in 83b4384fa. Since then, more flavors of Linux have become supported
by Swift: "linux-gnueabihf" in 4bf81e09d, and "freebsd" in f41b791d4.
Autolinking most likely does not work on any of these platforms, so
mark it as unsupported for now.
Other tests that only mark "linux-gnu" as unsupported do so for similar
reasons. Ensure unsupported tests for "linux-gnu" are also unsupported
on similar platforms.
Finishes up https://bugs.swift.org/browse/SR-280. At this point, nothing
should be passing O(N) arguments on the command line, where N is the
number of input source files, unless N is small.
(There are other inputs which are passed through to subtools, and these
are not put into filelists. That's fine.)
