In a subsequent commit, this is going to let me begin handling parameters with
actor regions in a nice way (and standardize all of the errors).
This is meant to be a refactoring commit that uses the current tests in tree to
make sure I did it correctly, so no tests need to be updated.
To keep this as an NFC commit, I only modeled initially actor isolated using
this. I am going to make it so that we properly treat global actor isolated
values as actor isolated/etc in a subsequent commit.
Previously, the lexical attribute on allock_stack instructions was used.
This doesn't work for values without lexical lifetimes which are
consumed, e.g. stdlib CoW types. Here, the new var_decl attribute on
alloc_stack is keyed off of instead. This flag encodes exactly that a
value corresponds to a source-level VarDecl, which is the condition
under which checking needs to run.
This makes it easier to debug/maintain/understand the code since the place that
emits the diagnostics is no longer split from the place that decides if a
diagnostic should be emitted. It also lets me eliminate the utility data
structure used to transfer data over.
Specifically, we previously would generate a list of errors to emit and then
have one centralized piece of code that went through the list and emitted them.
Instead, we now have a low level emitter struct that is just used to emit the
diagnostic at the immediate point where we would create previously an error
record. This places the diagnostic emission next to the piece of code that
triggered it being emitted making it easier to debug why an error was emitted
using -Xllvm -swift-diagnostics-assert-on-*.
I am going to do the same for transfer non transferrable in a little bit once
the async let patch has landed. It has some transfer non transferable error
additions that I want to get in so I am going to wait on that.
NFCI.
In embedded swift all de-serialized get public linkage because all the code is generated in the top-level module.
This change moves the point where we make de-serialized functions public to the end of the pipeline.
This allows dead function elimination to remove unused de-serialized functions.
For some stdlib functions (actually one: the Double initializer for a builtin integer) is essential, because codegen for embedded produces an unresolved symbol.
rdar://123772098
When we run RegionAnalysis, since it uses RPO order, we do not visit dead
blocks. This can create a problem when we emit diagnostics since we may merge in
a value into the region that was never actually defined. In this patch, if we
actually visit the block while performing dataflow, I mark a bit in its state
saying that it was live. Then when we emit diagnostics, I do not visit blocks
that were not marked live.
rdar://124042351
[region-isolation] Refactor out the stubify dead function if no longer used functionality from move only checker into its own pass and put it before region based isolation.
I am doing this since region based isolation hit the same issue that the move
checker did. So it makes sense to refactor the functionality into its own pass
and move it into a helper pass that runs before both.
It is very conservative and only stubifies functions that the specialization
passes explicitly mark as this being ok to be done to.
The comment stated that this was used to turn switches into consumes (which
is no longer relevant), but it looks like it is also necessary for the
move-only address checker to be able to properly understand partial consumes
of fields in some code patterns. Now that we surround borrowing switches
in plenty of opaque accesses to protect them from being pried apart by the
move checker, it should be safe to reenable this.
For years, optimizer engineers have been hitting a common bug caused by passes
assuming all SILValues have a parent function only to be surprised by SILUndef.
Generally we see SILUndef not that often so we see this come up later in
testing. This patch eliminates that problem by making SILUndef uniqued at the
function level instead of the module level. This ensures that it makes sense for
SILUndef to have a parent function, eliminating this possibility since we can
define an API to get its parent function.
rdar://123484595
With #70242 mark_dependence was added while emitting uninitialized array allocation
between the unsafe pointer and the array value returned.
Without it, we can end up with use after free of the unsafe pointer if the array's lifetime
is shortened. Along with this change, pattern matching of all array optimizations was updated to include
mark_dependence.
This change adds mark_dependence for uninitialized array allocation in OSLogOptimization which was left out previously.
With this, potential use-after-free of the unsafe pointer is prevented and all updated array optimizations apply
to the array created by OSLogOptimization.
Fixes rdar://122922902
This just cleans up the code by not exposing the details from the
VariableNameInferrer outside of the utilities. It also makes it easier to write
conditional code that uses these helpers by returning an optional, so instead of
having a long setup + an if statement, we just have an if statement optional
check.
I just did a full pass through. There were some cases around nonisolated
closures defined in methods and global actor isolated things where we are now
emitting the wrong message. I am going to fix that in subsequent commits.
NFC.
Before this commit, the diagnostic emission was split into utilities that were
used by one large emitter class. This could get confusing since there was some
duplication of names in between the utilities. Now instead, each part of the
main diagnostic emitter is in the MARK: section of the utilities that it uses.
Just making the pass easier for me to reason about.
We want to preserve the borrow scope during switch dispatch so that move-only
checking doesn't try to analyze destructures or consumes out of it. SILGen
should mark anywhere that's a potential possibility with its own marker so that
it gets borrow checked independently.
