Commit Graph

9118 Commits

Author SHA1 Message Date
nate-chandler
8c63134570 Merge pull request #78682 from nate-chandler/rdar142636711_2
[TempRValueOpt] Invalidate insts when completing.
2025-01-27 12:53:57 -08:00
eeckstein
95f1dfbf07 Merge pull request #78885 from eeckstein/fix-without-actually-escaping
Fix two problems with `withoutActuallyEscaping`
2025-01-27 07:59:02 +01:00
eeckstein
6b58f8f701 Merge pull request #78846 from eeckstein/fix-is-lexical
SIL: look through `end_init_let_ref` instructions when checking for `isLexical`
2025-01-27 07:58:27 +01:00
Andrew Trick
f0144ab7b1 Merge pull request #78830 from atrick/fix-lifedep-trivial-deadend
SILGenCleanup: extend to handle trivial local var scopes
2025-01-25 15:34:35 -08:00
Meghana Gupta
d2936a15bc Merge pull request #78824 from meg-gupta/fixmdi
Fix operand ownership of mark_dependence [nonescaping] of address values
2025-01-24 14:59:36 -08:00
Erik Eckstein
000eaedd03 SIL: look through end_init_let_ref instructions when checking for isLexical
`end_init_let_ref` instructions are inserted in class initializers after all fields are initialized.
It's important to look through such instructions when checking the `isLexical` property of a value.

Fixes a mis-compile due to shrinking a lexical liferange of a class.
Part of rdar://140229560
2025-01-24 20:10:05 +01:00
Erik Eckstein
3ec5d7de24 SIL: replace the is_escaping_closure instruction with destroy_not_escaped_closure
The problem with `is_escaping_closure` was that it didn't consume its operand and therefore reference count checks were unreliable.
For example, copy-propagation could break it.
As this instruction was always used together with an immediately following `destroy_value` of the closure, it makes sense to combine both into a `destroy_not_escaped_closure`.
It
1. checks the reference count and returns true if it is 1
2. consumes and destroys the operand
2025-01-24 19:23:27 +01:00
Meghana Gupta
79b649854b Fix operand ownership of mark_dependence [nonescaping] of address values 2025-01-24 00:49:32 -08:00
Andrew Trick
1b72c7bbf6 SILGenCleanup: extend to handle trivial local var scopes
Improves OSSALifetimeCompletion to handle trivial variables when running from
SILGenCleanup. This only affects lifetime dependence diagnostics.

For the purpose of lifetime diagnostics, trivial local variables are only valid
within their lexical scope. Sadly, SILGen only know how to insert cleanup code
on normal function exits. SILGenCleanup relies on lifetime completion to fix
lifetimes on dead end paths.

  %var = move_value [var_decl]
  try_apply %f() : $..., normal bb1, error error
error:
  extend_lifetime %var <=== insert this
  unreachable

This allows Span to depend on local unsafe pointers AND be used within
throwing closures:

    _ = a.withUnsafeBufferPointer {
      let buffer = $0
      let view = Span(_unsafeElements: buffer)
      return view.withUnsafeBufferPointer(\.count)
    }
2025-01-23 12:42:43 -08:00
Michael Gottesman
082b824a8e [rbi] Change Region Based Isolation for closures to not use the AST and instead just use SIL.
The reason why I am doing this is that in certain cases the AST captures indices
will never actually line up with partial apply capture indices since we seem to
"smush" together closures and locally defined functions.

NOTE: The reason for the really small amount of test changes is that this change
does not change the actual output by design. The only cases I had to change were
a case where we began to emit a better diagnostic and also where I added code
coverage around _ and let _ since those require ignored_use to be implemented so
that they would be diagnosed (previously we just did not emit anything so we
couldn't emit the diagnostic at the SIL level).

rdar://142661388
2025-01-22 21:12:36 -08:00
Michael Gottesman
6058b1d9bd [silgen] Change SILGen to emit ignored_user for emitIgnoredExpr and black hole initialization. 2025-01-22 21:12:36 -08:00
Michael Gottesman
7ae56aab2e [sil] Add a new instruction ignored_use.
This is used for synthetic uses like _ = x that do not act as a true use but
instead only suppress unused variable warnings. This patch just adds the
instruction.

Eventually, we can use it to move the unused variable warning from Sema to SIL
slimmming the type checker down a little bit... but for now I am using it so
that other diagnostic passes can have a SIL instruction (with SIL location) so
that we can emit diagnostics on code like _ = x. Today we just do not emit
anything at all for that case so a diagnostic SIL pass would not see any
instruction that it could emit a diagnostic upon. In the next patch of this
series, I am going to add SILGen support to do that.
2025-01-22 21:12:36 -08:00
Becca Royal-Gordon
0ce403d680 [NFC] Add some pretty stack traces
These would have helped us to debug rdar://142693093 more quickly.
2025-01-21 17:33:09 -08:00
Nate Chandler
2eb3a86928 [NFC] SIL: Replaced a recursion with a worklist. 2025-01-16 08:18:30 -08:00
Nate Chandler
8c7f1484bb [Gardening] SIL: Move comment.
Move it to the lambda which it talks about.  The old location is in a
function which now enjoys multiple callers, and the comment applies to
only one.
2025-01-16 08:18:29 -08:00
nate-chandler
b59280c0f0 Merge pull request #78624 from nate-chandler/rdar142520491_2
[OSSACanonicalizeOwned] Fix liveness passed to completion.
2025-01-15 19:38:21 -08:00
Gábor Horváth
4ae5685536 Merge pull request #78544 from swiftlang/gaborh/calling-conv
[cxx-interop] Fix wrong calling convention for some C++ types
2025-01-14 10:12:29 +00:00
Nate Chandler
4135540f62 [NFC] PrunedLiveness: Permit postDomBlocks dupes.
According to the documentation,

```
It's ok if postDomBlocks has duplicates or extraneous blocks, as long
as they jointly post-dominate all live blocks that aren't on dead-end
paths.
```

Previously, though, duplicates resulted in trapping: each element of
`postDomBlocks` is pushed into a `BasicBlockWorklist`; when the second
occurrence of an element in `postDomBlocks` was encountered,
`BasicBlockWorklist::push` would trigger an assertion failure.

Here, `pushIfNotVisited` is called instead.
2025-01-13 17:00:30 -08:00
Erik Eckstein
cc308c97cf embedded: make sure to generate witness tables which are imported from other modules
In embedded swift all the code is generated in the top-level module.
De-serialized witness tables for class existentials must be code-gen'd and therefore made non-external.

Fixes an unresolved symbol linker error.
rdar://142561676
2025-01-13 14:08:35 +01:00
Nate Chandler
9f87e0a8c6 [SIL] Print AST types with print-module-on-error. 2025-01-12 15:32:54 -08:00
Ellie Shin
727fb8c32d Merge pull request #78258 from swiftlang/elsh/disallow-bypass-deser-check
Package CMO: add deserialization checks to ensure correct memory layout
2025-01-11 05:40:49 -08:00
Meghana Gupta
00b9dd58c5 Merge pull request #78557 from meg-gupta/fixcse
Look through forwarding instructions while querying ownership root
2025-01-10 19:45:45 -08:00
Meghana Gupta
6f375aa6b6 Look through forwarding instructions while querying ownership root
Fixes rdar://142507746
2025-01-10 11:52:06 -08:00
Gabor Horvath
8cde635433 [cxx-interop] Fix wrong calling convention for some C++ types
We erroneously used @autoreleased convention for C++ types in some
scenarios.
2025-01-10 12:42:47 +00:00
eeckstein
6b8c341b64 Merge pull request #78464 from eeckstein/sil-linker-error
SILLinker: convert an assert to a compiler error message
2025-01-09 07:44:47 +01:00
Joe Groff
5ccb26b16c Merge pull request #78493 from jckarter/addressable-for-dependencies-3
SIL: Consider Builtin.FixedArray and @_rawLayout types to be structurally addressable-for-dependencies.
2025-01-08 20:11:00 -08:00
Joe Groff
b291cf42a1 SIL: Consider Builtin.FixedArray and @_rawLayout types to be structurally addressable-for-dependencies.
A primary purpose of these features is to enable persistent references into their storage, so it's
reasonable to assume this property for these types and types containing them.
2025-01-08 10:17:14 -08:00
Joe Groff
bc405c5f0a Merge pull request #78457 from jckarter/addressable-for-depedencies-2
`@_addressableForDependencies` fixes
2025-01-08 07:49:56 -08:00
Erik Eckstein
67b8c08ef1 SILLinker: convert an assert to a compiler error message
Having a wrong linkage can happen if the user "re-defines" an existing function with a wrong linkage, e.g. using `@_cdecl`.
2025-01-08 11:00:27 +01:00
elsh
c03abed00d Package optimization allows bypassing resilience, but that assumes the memory layout of the
decl being accessed is correct. When this assumption fails due to a deserialization error
of its members, the use site accesses the layout with a wrong field offset, resulting in
UB or a crash. The deserialization error is currently not caught at compile time due to
LangOpts.EnableDeserializationRecovery being enabled by default to allow for recovery of some
of the deserialization errors at a later time. In case of member deserialization, however,
it's not necessarily recovered later on.

This PR tracks whether member deserialization had an error by recursively loading members and
checking for deserialization error, and fails and emits a diagnostic. It provides a way to
prevent resilience bypassing when the deserialized decl's layout is incorrect.

Resolves rdar://132411524
2025-01-07 21:51:49 -08:00
Joe Groff
d9ee3f4de8 Make behavior changes for addressable types unconditional.
With the other fixes, it is now possible to build the stdlib without conditionalizing these
behaviors. This will allow libraries to adopt addressability as an experimental feature
without breaking ABI when interacting with other code.
2025-01-07 14:59:46 -08:00
Joe Groff
31ec17556c SIL: Only lower nominal types marked with @_addressableForDependencies as such.
I accidentally typoed the RecursiveProperties constructor so that every type
was getting treated as addressable-for-dependencies. Oops.
2025-01-06 18:16:28 -08:00
Allan Shortlidge
d0f63a0753 AST: Split Availability.h into multiple headers.
Put AvailabilityRange into its own header with very few dependencies so that it
can be included freely in other headers that need to use it as a complete type.

NFC.
2025-01-03 18:36:04 -08:00
Joe Groff
5c4406b5e8 Add an @_addressableForDependencies type attribute.
This attribute makes it so that a parameter of the annotated type, as well as
any type structurally containing that type as a field, becomes passed as
if `@_addressable` if the return value of the function has a dependency on
the parameter. This allows nonescapable values to take interior pointers into
such types.
2025-01-02 21:33:51 -08:00
Michael Gottesman
56f38c4172 [concurrency] Add support in SILGen/SIL for emitting and calling implicit leading parameters for CallerInheritingIsolation isolation.
This does not change region isolation yet to recognize these as effectively
nonisolated.
2025-01-02 13:18:54 -08:00
Michael Gottesman
b2a4f7e3cc [sil] Add to SIL and SILGen the ability to emit/represent implicit parameters.
I need this today to add the implicit isolated parameter... but I can imagine us
adding more implicit parameters in the future, so it makes sense to formalize it
so it is easier to do in the future.
2025-01-02 13:18:11 -08:00
Erik Eckstein
41dd3dc453 DeadObjectElimination: handle OSSA instructions when analyzing class destructors
Handle typical patterns which are generated by SILGen.
2024-12-24 12:00:22 +01:00
eeckstein
e6b9cb986f Merge pull request #78290 from eeckstein/sil-infrastructure
Some AST and SIL infrastructure additions and improvements
2024-12-20 07:18:32 +01:00
Erik Eckstein
4fdf16de1b SIL: add CanonicalType.canBeClass
And move the implementation of `SIL.Type.canBeClass` to the AST Type. The SIL Type just calls the AST Type implementation.
Also rename `SIL.Type.canonicalASTType` -> `SIL.Type.astType`.
2024-12-19 20:53:46 +01:00
Allan Shortlidge
17dfbf5053 AST: Adopt SemanticAvailableAttr for SpecializeAttr. 2024-12-19 08:40:00 -08:00
fahadnayyar
0f2637c32e Merge pull request #78230 from fahadnayyar/returns-retained-objc-support
[cxx-interop] Add support for SWIFT_RETURNS_(UN)RETAINED for ObjC APIs returning C++ FRT
2024-12-18 07:27:41 -08:00
Fahad Nayyar
426d471d1e [cxx-interop] Add support for SWIFT_RETURNS_(UN)RETAINED for ObjC APIs returning C++ FRT
rdar://135360972
2024-12-17 17:31:47 -08:00
nate-chandler
c6e8590837 Merge pull request #78229 from omochi/fix-variadic-optional-crash
Fix a crash in type lowering verification when using variadic optional tuples
2024-12-17 16:33:27 -08:00
Allan Shortlidge
95cc7319a1 SIL: Revert public linkage for @_silgen_name forward declarations.
It turns out that the stdlib build depends on `internal` functions with
`@_silgen_name` getting hidden linkage in some configurations. Instead of
messing with the linkage computation, just fix the `stdlib/Error.swift` test by
making `setWillThrowHandler` `public` to give it the right linkage.

Resolves rdar://141590619.
2024-12-17 08:35:01 -08:00
eeckstein
22b31dcaa8 Merge pull request #78221 from eeckstein/pointers_in_c_unions
TypeLowering: assume that C unions can contain a pointer
2024-12-17 06:59:24 +01:00
omochimetaru
37c1775b78 vanish single tuple on lowering verification 2024-12-17 10:43:21 +09:00
Meghana Gupta
349c3f19a6 Merge pull request #78217 from meg-gupta/fixreborrowverify
Fix reborrow verifier to look through borrowed from
2024-12-16 15:58:53 -08:00
Erik Eckstein
0441359b37 TypeLowering: assume that C unions can contain a pointer
C unions are imported as opaque types. Therefore we have to assume that a union contains a pointer.
This is important for alias analysis to catch escaping pointers via C unions.

Fixes a miscompile.
rdar://141555290
2024-12-16 19:37:54 +01:00
Meghana Gupta
4cab04e190 Fix reborrow verifier to look through borrowed from 2024-12-16 10:23:18 -08:00
Allan Shortlidge
8f5e37c57c Merge pull request #78183 from tshortli/silgen-name-linkage
SIL: Always give `@_silgen_name` forward declarations public linkage
2024-12-14 10:12:03 -08:00