shylie/llvm-project
1
0
Fork 0
Code Issues Pull Requests Actions 6 Packages Projects Releases Wiki Activity
553,076 Commits 1,713 Branches 298 Tags
Commit Graph

32 Commits

Author SHA1 Message Date
Timm Baeder
f5ffedf81a
[clang][bytecode] Pass SourceInfo objects by value (#159532) 
They are only pointer-sized and copying them is cheaper than taking the
const ref.
 2025-09-18 13:08:42 +02:00
Timm Baeder
6a16f1a6f4
[clang][ExprConst] Assert that EvaluateAsInitializer has non-null VD (#159274) 
Both the expression (the initializer) as well as the VarDecl can't be
null here. Assert that.
 2025-09-17 13:23:02 +02:00
Timm Baeder
03c356c554
[clang][bytecode] Pass initializer along in evaluateAsInitializer() (#158056) 
We just called `getInit()`, which isn't always correct and used the
wrong initializer in the module test case.
 2025-09-15 07:39:11 +02:00
Timm Baeder
94d5c54a4f
[clang][bytecode] Don't update temporary in InitGlobalTemp* (#158022) 
We can save ourselves the conversion to an APValue here since we will do
that later in updateGlobalTemporaries() anyway.
 2025-09-11 12:32:58 +02:00
Timm Baeder
8ba0d0f72f
[clang][bytcode][NFC] Use UnsignedOrNone for global ids (#157328) 2025-09-08 08:56:46 +02:00
Timm Baeder
83ca877159
[clang][bytecode] Reject final ltor casts on string literals (#156669) 
Similar to what the current interpreter does.
 2025-09-04 14:08:04 +02:00
Timm Baeder
02ade22aa8
[clang][bytecode] Avoid copying APValue into EvaluationResult (#155757) 
Move the `APValue` into `EvaluationResult` instead.

For a large primitive array (`#embed` of the sqlite3 amalgamation), this
improves compile times by around 25%.
 2025-08-28 08:46:06 +02:00
Timm Baeder
e0acf6592b
[clang][bytecode] Call CheckFinalLoad in all language modes (#154496) 
Fixes #153997
 2025-08-21 08:24:09 +02:00
Timm Baeder
870aa979c4
[clang][bytecode] Use visitExpr() in interpretCall (#152857) 
This is the correct function to use and it will create a variable scope.

Fixes #152822
 2025-08-09 18:30:26 +02:00
Timm Baeder
c869ef6ebc
[clang][bytecode] Refactor Check* functions (#152300) 
... so we don't have to create Pointer instances when we don't need
them.
 2025-08-07 11:32:39 +02:00
Timm Baeder
dc7c3c2b2b
[clang][bytecode] Disable location tracking for implicit field inits (#150190) 2025-08-05 13:14:01 +02:00
Timm Baeder
d02714c85a
[clang][bytecode] Return success for pointers to locals (#151980) 
They aren't valid, but we leave them successful here and
CheckLValueConstantExpression will diagnose them later.
 2025-08-05 05:18:47 +02:00
Timm Baeder
76e38ca981
[clang][bytecode] Try to load primitive values directly (#151833) 
Instead of doing a GetPtrLocal + Load or GetPtrGlobal + Load pair, try
to load the value directly.
 2025-08-04 12:28:33 +02:00
Timm Baeder
e39ee62c5b
[clang][bytecode] Use OptPrimType instead of std::optional<PrimType> (#149812) 
We use this construct a lot. Use something similar to clang's
UnsignedOrNone.

This results in some slighy compile time improvements:
https://llvm-compile-time-tracker.com/compare.php?from=17a4b0399d161a3b89d8f0ce82add1638f23f5d4&to=a251d81ecd0ed45dd190462663155fdb303ef04d&stat=instructions:u
 2025-07-21 17:16:13 +02:00
Timm Baeder
ef49ed4829
[clang][bytecode] Use bytecode interpreter in isPotentialConstantExprU… (#149462) 
…nevaluated

Fake a function call to the given function and evaluate the given
expression as if it was part of that function call.

Fixes #149383
 2025-07-19 07:18:51 +02:00
Timm Baeder
172c2817e5
[clang][bytecode] Use a SmallVector for EvalEmitter's locals (#140513) 
The offset we return for them are just indices, so we can use a vector
here.
 2025-05-19 11:19:35 +02:00
Timm Baeder
ecc73a6f1c
[clang][bytecode] Use bytecode interpreter in EvaluateCharRangeAsString (#138461) 
This was always using the ast walker.
 2025-05-05 09:06:34 +02:00
Timm Baeder
2d63faead4
[clang][bytecode][NFC] Remove PT_FnPtr (#135947) 
We don't need this anymore since we don't return it from classify()
anymore.
 2025-04-16 13:21:25 +02:00
Timm Baeder
02f923f8e4
[clang][bytecode] Classify function pointers as PT_Ptr (#135026) 
The Pointer class already has the capability to be a function pointer,
but we still classifed function pointers as PT_FnPtr/FunctionPointer.
This means when converting from a Pointer to a FunctionPointer, we lost
the information of what the original Pointer pointed to.
 2025-04-10 06:40:54 +02:00
Timm Baeder
d08cf7900d
[clang][bytecode] Implement __builtin_constant_p (#130143) 
Use the regular code paths for interpreting.

Add new instructions: `StartSpeculation` will reset the diagnostics
pointers to `nullptr`, which will keep us from reporting any diagnostics
during speculation. `EndSpeculation` will undo this.

The rest depends on what `Emitter` we use.

For `EvalEmitter`, we have no bytecode, so we implement `speculate()` by
simply visiting the first argument of `__builtin_constant_p`. If the
evaluation fails, we push a `0` on the stack, otherwise a `1`.

For `ByteCodeEmitter`, add another instrucion called `BCP`, that
interprets all the instructions following it until the next
`EndSpeculation` instruction. If any of those instructions fails, we
jump to the `EndLabel`, which brings us right before the
`EndSpeculation`. We then push the result on the stack.
 2025-03-08 06:06:14 +01:00
Timm Baeder
06130ed341
Reapply "[clang][bytecode] Stack-allocate bottom function frame" (#12… (#125349) 
…5325)

Move the BottomFrame to InterpState instead.
 2025-02-01 18:12:35 +01:00
Jorge Gorbe Moya
386af4a5c6
Revert "[clang][bytecode] Stack-allocate bottom function frame" (#125325) 
Reverts llvm/llvm-project#125253

It introduced an msan failure. Caught by a buildbot here:
https://lab.llvm.org/buildbot/#/builders/164/builds/6922/steps/17/logs/stdio
 2025-01-31 20:09:14 -08:00
Timm Baeder
f3549814f8
[clang][bytecode] Stack-allocate bottom function frame (#125253) 
Instead of heap-allocating it. This is similar to what the current
interpeter does. In C, we have no function calls, so the extra heap
allocation never makes sense.
 2025-01-31 20:08:58 +01:00
Timm Baeder
6b8d076b1f
[clang][bytecode] Return failure for dead non-temporary pointers (#125247) 
... even when we're not performing a lvalue-to-rvalue conversion.
 2025-01-31 20:06:29 +01:00
Timm Bäder
a0bd40e5a3 Revert "[clang][bytecode] Mark global decls with diagnostics invalid (#122895)" 
This reverts commit c1de9b920935e7d2fcc8dd7b39c1b7285783e948.

It looks like this breaks the memory-sanitizer builder:
https://lab.llvm.org/buildbot/#/builders/94/builds/3745
 2025-01-14 18:37:36 +01:00
Timm Baeder
c1de9b9209
[clang][bytecode] Mark global decls with diagnostics invalid (#122895) 
Even if their evaluation succeeds, mark them as invalid. This fixes some
long standing differences to the ast walker interpeter.
 2025-01-14 15:10:48 +01:00
Kazu Hirata
dec6324cb0
[AST] Remove unused includes (NFC) (#116549) 
Identified with misc-include-cleaner.
 2024-11-17 09:36:48 -08:00
Timm Baeder
3eaf4a9d1a
[clang][bytecode] Check for memory leaks after destroying global scope (#112868) 
The global scope we create when evaluating expressions might free some
of the dynamic memory allocations, so we can't check for memory leaks
before destroying it.
 2024-10-18 13:03:13 +02:00
Timm Baeder
83fea8b809
[clang][bytecode] Allow continuing when discarded MemberExpr Base fails (#107231) 
We don't need the value in this case, since we're discarding it anyway.
Allow continuing the interpretation but note the side effect.
 2024-09-07 09:33:27 +02:00
yronglin
d9e7286019
[NFC][clang][bytecode] Rename clang::interp::State::getCtx to clang::interp::State::getASTContext (#106071) 
The new constant interpreter's `clang::interp::InterpState` contains
both `clang::interp::Context` and `clang::ASTContext`. So using `S.Ctx`
and `S.getCtx()` was a bit confusing. This PR rename `getCtx()` to
`getASTContext` to make things more clearer.

Signed-off-by: yronglin <yronglin777@gmail.com>
 2024-08-26 22:23:07 +08:00
Timm Baeder
a14c730990
[clang][bytecode] Fix diagnostic in final ltor cast (#105292) 
Don't diagnose volatile reads but diagnose a few other accesses earlier.
 2024-08-21 08:08:32 +02:00
Timm Baeder
a07aba5d44
[clang] Rename all AST/Interp stuff to AST/ByteCode (#104552) 
"Interp" clashes with the clang interpreter and people often confuse
this.
 2024-08-16 17:13:12 +02:00