Commit Graph

32 Commits

Author SHA1 Message Date
Slava Pestov
9ec80df97e SIL: Remove curried SILDeclRefs 2020-03-19 02:20:21 -04:00
Slava Pestov
53bfc767a3 SIL: Track target formal type for casts
SIL type lowering erases DynamicSelfType, so we generate
incorrect code when casting to DynamicSelfType. Fixing this
requires a fair amount of plumbing, but most of the
changes are mechanical.

Note that the textual SIL syntax for casts has changed
slightly; the target type is now a formal type without a '$',
not a SIL type.

Also, the unconditional_checked_cast_value and
checked_cast_value_br instructions now take the _source_
formal type as well, just like the *_addr forms they are
intended to replace.
2019-11-20 21:30:28 -05:00
Joe Groff
6588ee952e Dynamic cast optimizer: Consider CF conversions via superclasses.
A conversion from, e.g., CFMutableDictionary to NSDictionary is still a CF bridging conversion,
even though it doesn't go directly through CFMutableDictionary's own bridged class type.
Fixes rdar://problem/51078136.
2019-07-12 16:22:01 -07:00
Michael Gottesman
08c25da322 [cast-opt] Fix another code path in the cast optimizer for ownership and add an additional run of cast_folding_objc.swift with stripping ownership after serialization. 2019-05-20 00:51:26 -07:00
Michael Gottesman
de021aa32b [cast-opt] Eliminate a leak along the cast failure path when bridging from a swift to ns object derived class.
I added an interpreter test that exercises this code path using LifetimeTracked
to make sure we do not regress again.

rdar://49882071
2019-04-13 20:10:24 -07:00
Erik Eckstein
39bb14b094 change mangling prefix from $S to $s
This is the final ABI mangling prefix

rdar://problem/38471478
2018-09-19 13:55:11 -07:00
Erik Eckstein
8bc7fb860d Some improvements and simplifications regarding pass name printing in the pass manager.
* rename "Name" to "Description" in the pass definition, because it's not really the pass name, but the description of a pass
* remove the getName() from Transforms (which actually returned the description of a pass)
* in debug printing, print the pass ID and not the pass description. It makes it easier to correlate the debug output to the actual pass implementation.
* remove the iteration numbering in the pass manager, because we only run a single iteration anyway.
2018-01-09 15:35:26 -08:00
Erik Eckstein
cd3d50a5d9 ABI: Change the mangling prefix from _T0 to $S 2018-01-06 13:55:59 -08:00
Pavel Yaskevich
f90d943a29 [Mangling/ABI] NFC: Fix SILOptimizer tests to reflect label mangling changes 2017-12-18 15:45:50 -08:00
Roman Levenstein
a0891d64a3 [cast-optimizer] Fix casting of P.Protocol to P.Type
If the source and target are the same existential type, but the source is P.Protocol and the dest is P.Type, then we need to consider whether the protocol is self-conforming.
The only cases where a protocol self-conforms are objc protocols, but we're going to expect P.Type to hold a class object. And this case doesn't matter since for a self-conforming protocol type there can't be any type-level methods.

Thus we consider this kind of cast to always fail. The only exception from this rule is when the target is Any.Type, because *.Protocol can always be casted to Any.Type.

Fixes rdar://32682967
2017-06-15 19:00:25 -07:00
Roman Levenstein
4ef29e57af [cast-optimizer] Properly handle conditional bridged casts in optimizeBridgedSwiftToObjCCast
If there is a conditional bridged cast from a swift type to an objc type and this cast happens in two stages, where:
-  in the firs stage the Swift type is casted to its bridged ObjC class (e.g. String to NSString) and
-  in the second stage there is a downcast to a subclass of a bridged ObjC class (e.g. NSString to MyString)

then the second stage should use a conditional cast.

rdar://32319580
2017-05-22 13:51:01 -07:00
Roman Levenstein
62e1944d7b [cast-optimizer] Bail out on a conditional downcast to a subclass of a bridged ObjC class
This is very conservative fix for rdar://problem/32316838.
A more aggressive fix would introduce a conditional cast.
2017-05-21 22:51:28 -07:00
Slava Pestov
b5721e8d8e AST: Remove AnyObject protocol 2017-05-02 19:45:00 -07:00
Erik Eckstein
c4a11f4c92 tests: remove the now unused option -new-mangling-for-tests 2017-03-22 11:28:43 -07:00
Erik Eckstein
1d3724666f tests: convert about 400 tests to the new mangling by using the -new-mangling-for-tests option
When the new mangling is enabled permanently, the option can be removed from the RUN command lines again.
2017-01-24 15:27:45 -08:00
practicalswift
fa7fbdb8b0 [gardening] Remove redundant nil-initialization of optional variable
From the Swift documentation:

"If you define an optional variable without providing a default value,
 the variable is automatically set to nil for you."
2016-09-18 07:40:07 +02:00
Dmitri Gribenko
d175b3b66d Migrate FileCheck to %FileCheck in tests 2016-08-10 23:52:02 -07:00
Erik Eckstein
84565932c6 SILOptimizer: fixed bugs which cause memory leaks.
*) cast optimizer: when a bridging cast is replaced with a function call and the owning convention of the instruction and the call parameter do not match, compensating retain/release instructions must be inserted.
*) cast optimizer: when a consuming dead cast instruction is removed a compensating release instruction must be inserted
*) mem2reg: An alloc_stack location which contains a destroy_addr must not be considered as a write-only location. The destroy_addr must be preserved.

rdar://problem/27601057
2016-08-04 15:39:14 -07:00
Joe Groff
2a5b9d6559 Remove now-dead _isBridgedToObjectiveC() declarations in _ObjectiveCBridgeable conformers. 2016-07-28 12:29:47 -07:00
Arnold Schwaighofer
f6e3f67293 Fix ASTContext::getBridgedToObjC to not return None under id-as-any
This would lead us to conclude that a check-cast-branch fails when it does not.

rdar://27536049
2016-07-26 14:19:03 -07:00
Doug Gregor
852686b225 Manually add _unconditionallyBridgeFromObjectiveC throughout test cases. NFC 2016-05-09 00:28:56 -07:00
Xin Tong
1a4f567685 More conservative about when we can move a release across an instruction
We now consider effect of deinit in addition to the released value.

rdar://25362826

This is the only 10%+ regression i measured on my machine. no performance improvement.

Sim2DArray                                              | 326     | 366     | +12.3%    | **0.89x**
2016-04-12 20:39:30 -07:00
Chris Willmore
73d5077bae Add API notes for Foundation following internal Swift 3 naming audit. 2016-04-07 20:35:18 -07:00
Manav Gabhawala
7928140f79 [SE-0046] Implements consistent function parameter labels by discarding extraneous parameter names and adding _ where necessary 2016-04-06 20:21:58 -04:00
Xin Tong
f95d9b3c92 Change FSO heuristic.
FSO functions that have high potential but does not have caller inside
current module.

The thunk can then be inlined into the module calling the function and
the function would get the benefit of FSO.

The heuristic for selecting such function is
1. Have no indirect caller. This would introduce a thunk.
2. Have potential to give better performance. i.e. function argument can
be O2G.

Regression
TEST                                                    | OLD_MIN | NEW_MIN | DELTA (%) | SPEEDUP
---                                                     | ---     | ---     | ---       | ---
BenchLangCallingCFunction                               | 184     | 211     | +14.7%    | **0.87x**
Calculator                                              | 55      | 59      | +7.3%     | **0.93x**
DeadArray                                               | 687     | 741     | +7.9%     | **0.93x**
MonteCarloPi                                            | 39275   | 41669   | +6.1%     | **0.94x**

Improvement
TEST                                                    | OLD_MIN | NEW_MIN | DELTA (%) | SPEEDUP
---                                                     | ---     | ---     | ---       | ---
LuhnAlgoLazy                                            | 2478    | 2327    | -6.1%     | **1.06x**
OpenClose                                               | 54      | 51      | -5.6%     | **1.06x**
SortLettersInPlace                                      | 1016    | 946     | -6.9%     | **1.07x**
ObjectiveCBridgeFromNSDictionaryAnyObjectToStringForced | 149993  | 139755  | -6.8%     | **1.07x**
Phonebook                                               | 9666    | 8992    | -7.0%     | **1.07x**
ObjectiveCBridgeFromNSDictionaryAnyObjectToString       | 222713  | 206538  | -7.3%     | **1.08x**
LuhnAlgoEager                                           | 2393    | 2226    | -7.0%     | **1.08x**
Dictionary                                              | 1307    | 1196    | -8.5%     | **1.09x**
JSONHelperDeserialize                                   | 3808    | 3492    | -8.3%     | **1.09x**
StdlibSort                                              | 7310    | 4084    | -44.1%    | **1.79x**

I see 0.15% increase in code size for Benchmark_O.

Thanks @gottesmm for suggesting this opportunity.

rdar://25345056
2016-03-29 23:04:36 -07:00
Slava Pestov
fded0d5bc2 SIL: Mark externally-visible witness tables for fixed-layout types as [fragile]
Also, mark witness thunks for [fragile] witnesses as [fragile].

This allows us to serialize the witness table, as well as any thunks
for witnesses declared @_transparent and @inline(__always).
2016-03-28 14:14:49 -07:00
Doug Gregor
1889ea2657 Remove the default implementation of _ObjectiveCBridgeable._unconditionallyBridgeFromObjectiveC.
SILGen can't handle it yet, so pull it out until SILGen is fixed.
2016-03-24 17:10:26 -07:00
Doug Gregor
67dad6041c [Runtime] Use the _ObjectiveCBridgeable._ObjectiveCType type witness directly.
This lets us eliminate the _getObjectiveCType() value witness, which
was working around the lack of proper type witness metadata in witness
tables. Boilerplate -= 1.
2016-03-18 16:11:38 -07:00
Daniel Duan
2bc78b8c09 [stdlib] update for 'inout' adjustment (SE-0031) 2016-02-26 12:02:29 -08:00
Adrian Prantl
c9927f66f0 Serialize debug scope and location info in the SIL assembler language.
At the moment it is only possible to test the effects that SIL
optimization passes have on debug information by observing the
effects of a full .swift -> LLVM IR compilation. This change enable us
to write targeted testcases for single SIL optimization passes.

The new syntax is as follows:

 sil-scope-ref ::= 'scope' [0-9]+
 sil-scope ::= 'sil_scope' [0-9]+ '{'
                 sil-loc
                 'parent' scope-parent
                 ('inlined_at' sil-scope-ref )?
               '}'
 scope-parent ::= sil-function-name ':' sil-type
 scope-parent ::= sil-scope-ref
 sil-loc ::= 'loc' string-literal ':' [0-9]+ ':' [0-9]+

Each instruction may have a debug location and a SIL scope reference
at the end.  Debug locations consist of a filename, a line number, and
a column number.  If the debug location is omitted, it defaults to the
location in the SIL source file.  SIL scopes describe the position
inside the lexical scope structure that the Swift expression a SIL
instruction was generated from had originally. SIL scopes also hold
inlining information.

<rdar://problem/22706994>
2016-02-26 10:46:29 -08:00
practicalswift
36d7072013 Remove immediately adjacent repeated words ("the the", "for for", "an an", etc.). 2015-12-21 22:16:04 +01:00
Andrew Trick
bd35b4789c Move test/SILOptimizer files to reflect lib/SILOptimizer. 2015-12-11 15:53:22 -08:00