It is possible for copies or spills to be inserted in the middle of indirect
addressing sequences which use VGPR indexing. Spills to accvgprs could be
effected by the indexing mode.
Add new pseudo instructions that are expanded after register allocation to avoid
the problematic spill or copy placement.
Differential Revision: https://reviews.llvm.org/D91048
We can use these instructions for single bit immediates that are too large for ANDI/ORI/CLRI.
The _10 test cases are to make sure that we still use ANDI/ORI/CLRI for small immediates.
Differential Revision: https://reviews.llvm.org/D92262
While casting an (integral) pointer to an integer is obvious - you just get
the integral value of the pointer, casting an integer to an (integral) pointer
is deceivingly different. While you will get a pointer with that integral value,
if you got that integral value via a pointer-to-integer cast originally,
the new pointer will lack the provenance information from the original pointer.
So while (integral) pointer to integer casts are effectively no-ops,
and are transparent to the optimizer, integer to (integral) pointer casts
are *NOT* transparent, and may conceal information from optimizer.
While that may be the intention, it is not always so. For example,
let's take a look at a routine to align the pointer up to the multiple of 16:
The obvious, naive implementation for that is:
```
char* src(char* maybe_underbiased_ptr) {
uintptr_t maybe_underbiased_intptr = (uintptr_t)maybe_underbiased_ptr;
uintptr_t aligned_biased_intptr = maybe_underbiased_intptr + 15;
uintptr_t aligned_intptr = aligned_biased_intptr & (~15);
return (char*)aligned_intptr; // warning: avoid integer to pointer casts [misc-no-inttoptr]
}
```
The check will rightfully diagnose that cast.
But when provenance concealment is not the goal of the code, but an accident,
this example can be rewritten as follows, without using integer to pointer cast:
```
char*
tgt(char* maybe_underbiased_ptr) {
uintptr_t maybe_underbiased_intptr = (uintptr_t)maybe_underbiased_ptr;
uintptr_t aligned_biased_intptr = maybe_underbiased_intptr + 15;
uintptr_t aligned_intptr = aligned_biased_intptr & (~15);
uintptr_t bias = aligned_intptr - maybe_underbiased_intptr;
return maybe_underbiased_ptr + bias;
}
```
See also:
* D71499
* [[ https://www.cs.utah.edu/~regehr/oopsla18.pdf | Juneyoung Lee, Chung-Kil Hur, Ralf Jung, Zhengyang Liu, John Regehr, and Nuno P. Lopes. 2018. Reconciling High-Level Optimizations and Low-Level Code in LLVM. Proc. ACM Program. Lang. 2, OOPSLA, Article 125 (November 2018), 28 pages. ]]
Reviewed By: aaron.ballman
Differential Revision: https://reviews.llvm.org/D91055
This patch makes DWARF writer emit DW_AT_string_length using
the stringLengthExp operand of DIStringType.
This is part of the effort to add debug info support for
Fortran deferred length strings.
Also updated the tests to exercise the change.
Differential Revision: https://reviews.llvm.org/D92412
The LLVM_ENABLE_MODULES builds currently randomly fail due depending on the
headers generated by the intrinsics_gen target, but the current dependency only model
the non-modules dependencies:
```
While building module 'LLVM_ExecutionEngine' imported from llvm-project/llvm/lib/ExecutionEngine/Orc/Shared/TargetProcessControlTypes.cpp:13:
While building module 'LLVM_intrinsic_gen' imported from llvm-project/llvm/include/llvm/ExecutionEngine/Orc/ThreadSafeModule.h:17:
In file included from <module-includes>:1:
In file included from llvm-project/llvm/include/llvm/IR/Argument.h:18:
llvm/include/llvm/IR/Attributes.h:75:14: fatal error: 'llvm/IR/Attributes.inc' file not found
#include "llvm/IR/Attributes.inc"
^~~~~~~~~~~~~~~~~~~~~~~~
```
Depending on whether intrinsics_gen runs before compiling Orc/Shared files we either fail or include an outdated Attributes.inc
in module builds. The Clang modules require these additional dependencies as including/importing one module requires all
includes headers by that module to be parsable.
Differential Revision: https://reviews.llvm.org/D92873
Update all reference from the specification to the new OpenACC 3.1
document.
Reviewed By: SouraVX
Differential Revision: https://reviews.llvm.org/D92120
-Reject an "mf1" lmul
-Make sure tail agnostic is exactly "tu" or "ta" not just that it starts with "tu" or "ta"
-Make sure mask agnostic is exactly "mu" or "ma" not just that it starts with "mu" or "ma"
Differential Revision: https://reviews.llvm.org/D92805
APInt's string constructor asserts on error. Since this is the parser and we don't yet know if the string is a valid integer we shouldn't use that.
Instead use StringRef::getAsInteger which returns a bool to indicate success or failure.
Since we no longer need APInt, use 'unsigned' instead.
Differential Revision: https://reviews.llvm.org/D92801
Add restriction on loop construct associated with DO CONCURRENT. Add couple of tests to ensure
clause validity checks.
Reviewed By: sameeranjoshi
Differential Revision: https://reviews.llvm.org/D92533
We can't expand lambda types anyway. Now we simply not offer the code
action instead of showing it and then returning an error in apply().
Differential Revision: https://reviews.llvm.org/D92847
This implements the following folds:
```
G_SELECT cc, (G_SUB 0, %x), %false -> CSNEG %x, %false, inv_cc
G_SELECT cc, (G_XOR x, -1), %false -> CSINV %x, %false, inv_cc
```
This is similar to the folds introduced in
5bc0bd05e6.
In 5bc0bd05e6 I mentioned that we may prefer to do
this in AArch64PostLegalizerLowering.
I think that it's probably better to do this in the selector. The way we select
G_SELECT depends on what register banks end up being assigned to it. If we did
this in AArch64PostLegalizerLowering, then we'd end up checking *every* G_SELECT
to see if it's worth swapping operands. Doing it in the selector allows us to
restrict the optimization to only relevant G_SELECTs.
Also fix up some comments in `TryFoldBinOpIntoSelect` which are kind of
confusing IMO.
Example IR: https://godbolt.org/z/3qPGca
Differential Revision: https://reviews.llvm.org/D92860
clang was complaing about this code:
llvm/include/llvm/IR/PassManager.h:715:17: error: ISO C++20 considers use of overloaded operator '!=' to be ambiguous despite there being a unique best viable function with non-reversed arguments [-Werror,-Wambiguous-reversed-operator]
if (IMapI != IsResultInvalidated.end())
~~~~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~
llvm/include/llvm/ADT/DenseMap.h:1253:8: note: candidate function with non-reversed arguments
bool operator!=(const ConstIterator &RHS) const {
^
llvm/include/llvm/ADT/DenseMap.h:1246:8: note: ambiguous candidate function with reversed arguments
bool operator==(const ConstIterator &RHS) const {
^
The warning is triggered when the DenseMapIterator (lhs) is not const and so
the == operator is applied to different types on lhs/rhs.
Using a template allows the function to be available for both const/non-const
iterator types and gets rid of the warning
Also error out if we find anything other than an object or bitcode file
in the archive.
Note that we were previously inserting the symbols and sections of the
unpacked ObjFile into the containing ArchiveFile. This was actually
unnecessary -- we can just insert the ObjectFile (or BitcodeFile) into
the `inputFiles` vector. This is the approach taken by LLD-ELF.
Reviewed By: thakis
Differential Revision: https://reviews.llvm.org/D92539
Additionally:
1. Move the helper functions in InputSection.h below the definition of
`InputSection`, so the important stuff is on top
2. Remove unnecessary `explicit`
Reviewed By: #lld-macho, compnerd
Differential Revision: https://reviews.llvm.org/D92453
Add an option to pass the number of worker threads to select the number of async regions for parallel for transformation.
```
std::unique_ptr<OperationPass<FuncOp>> createAsyncParallelForPass(int numWorkerThreads);
```
Reviewed By: mehdi_amini
Differential Revision: https://reviews.llvm.org/D92835
Define Fortran derived types that describe the characteristics
of derived types, and instantiations of parameterized derived
types, that are of relevance to the runtime language support
library. Define a suite of corresponding C++ structure types
for the runtime library to use to interpret instances of the
descriptions.
Create instances of these description types in Semantics as
static initializers for compiler-created objects in the scopes
that define or instantiate user derived types.
Delete obsolete code from earlier attempts to package runtime
type information.
Differential Revision: https://reviews.llvm.org/D92802
This node returns 2 results and uses a chain. As long as we use a DAG as part of the pseudo instruction definition where we can use the "set" operator, it looks like tablegen can handle use a pattern for this without a problem. I believe the original implementation was copied from PowerPC.
This also fixes the pseudo instruction so that it is marked as having side effects to match the definition of CSRRS and the RV64 instruction. And we don't need to explicitly clear mayLoad/mayStore since those can be inferred now.
Differential Revision: https://reviews.llvm.org/D92786
Currently, -ftime-report + new pass manager emits one line of report for each
pass run. This potentially causes huge output text especially with regular LTO
or large single file (Obeserved in private tests and was reported in D51276).
The behaviour of -ftime-report + legacy pass manager is
emitting one line of report for each pass object which has relatively reasonable
text output size. This patch adds a flag `-ftime-report=` to control time report
aggregation for new pass manager.
The flag is for new pass manager only. Using it with legacy pass manager gives
an error. It is a driver and cc1 flag. `per-pass` is the new default so
`-ftime-report` is aliased to `-ftime-report=per-pass`. Before this patch,
functionality-wise `-ftime-report` is aliased to `-ftime-report=per-pass-run`.
* Adds an boolean variable TimePassesHandler::PerRun to control per-pass vs per-pass-run.
* Adds a new clang CodeGen flag CodeGenOptions::TimePassesPerRun to work with the existing CodeGenOptions::TimePasses.
* Remove FrontendOptions::ShowTimers, its uses are replaced by the existing CodeGenOptions::TimePasses.
* Remove FrontendTimesIsEnabled (It was introduced in D45619 which was largely reverted.)
Differential Revision: https://reviews.llvm.org/D92436
The check 'cppcoreguidelines-narrowing-conversions' does not detect conversions
involving typedef. This notably includes the standard fixed-width integer types
like int32_t, uint64_t, etc. Now look through the typedefs at the desugared type.
On RH66, timespec_get is not available. Use clock_gettime instead.
This problem was introduced with D87120
Reviewed By: tejohnson
Differential Revision: https://reviews.llvm.org/D91687
phi-empty.ll does not pass under the new PM because the NPM runs
-loop-simplify. Running -loop-simplify ends up not reproing
https://llvm.org/PR48296.
Verified that this test fails when 9eb2c011 is reverted.
Reviewed By: spatel
Differential Revision: https://reviews.llvm.org/D92807
LLVM intrinsic llvm.maxnum|minnum is overloaded intrinsic, can be used on any
floating-point or vector of floating-point type.
This patch extends current infrastructure to support scalable vector type.
This patch also fix a warning message of incorrect use of EVT::getVectorNumElements()
for scalable type, when DAGCombiner trying to split scalable vector.
Reviewed By: sdesmalen
Differential Revision: https://reviews.llvm.org/D92607
Function Parser::ParseAvailabilityAttribute checks that the message string of
an availability attribute is not a wide string literal. Test case
clang/test/Parser/attr-availability.c specifies that a string literal is
expected.
The code checked that the first token in a string concatenation is a string
literal, and then that the concatenated string consists of 1-byte characters.
On a target where wide character is 1 byte, a string concatenation "a" L"b"
passes both those checks, but L"b" alone is rejected. More generally, "a" u8"b"
passes the checks, but u8"b" alone is rejected.
So check isAscii() instead of character size.
ScalarizeMaskedMemIntrinsic is currently a codeGen level pass. The pass
is actually operating on IR level and does not use any code gen specific
passes. It is useful to move it into transforms directory so that it
can be more widely used as a mid-level transform as well (apart from
usage in codegen pipeline).
In particular, we have a usecase downstream where we would like to use
this pass in our mid-level pipeline which operates on IR level.
The next change will be to add support for new PM.
Reviewers: craig.topper, apilipenko, skatkov
Reviewed-By: skatkov
Differential Revision: https://reviews.llvm.org/D92407
We didn't have selector support for these.
Selection code is similar to `getAArch64XALUOOp` in AArch64ISelLowering. Similar
to that code, this returns the AArch64CC and the instruction produced. In SDAG,
this is used to optimize select + overflow and condition branch + overflow
pairs. (See `AArch64TargetLowering::LowerBR_CC` and
`AArch64TargetLowering::LowerSelect`)
(G_USUBO should be easy to add here, but it isn't legalized right now.)
This also factors out the existing G_UADDO selection code, and removes an
unnecessary check for s32/s64. AFAIK, we shouldn't ever get anything other than
s32/s64. It makes more sense for this to be handled by the type assertion in
`emitAddSub`.
Differential Revision: https://reviews.llvm.org/D92610
This should address the FIXME about clang3.9 dervied to base unique_ptr constructor not working.
Reviewed By: sammccall
Differential Revision: https://reviews.llvm.org/D91925
This is a rework of D85812, which didn't land.
When callee coroutine function is inlined into caller coroutine function before coro-split pass, llvm will emits "coroutine should have exactly one defining @llvm.coro.begin". It seems that coro-early pass can not handle this quiet well.
So we believe that unsplited coroutine function should not be inlined.
This patch fix such issue by not inlining function if it has attribute "coroutine.presplit" (it means the function has not been splited) to fix this issue
test plan: check-llvm, check-clang
In D85812, there was suggestions on moving the macros to Attributes.td to avoid circular header dependency issue.
I believe it's not worth doing just to be able to use one constant string in one place.
Today, there are already 3 possible attribute values for "coroutine.presplit": c6543cc6b8/llvm/lib/Transforms/Coroutines/CoroInternal.h (L40-L42)
If we move them into Attributes.td, we would be adding 3 new attributes to EnumAttr, just to support this, which I think is an overkill.
Instead, I think the best way to do this is to add an API in Function class that checks whether this function is a coroutine, by checking the attribute by name directly.
Differential Revision: https://reviews.llvm.org/D92706
This reverts commit 02c9050155.
Painted myself into a corner with -Wvirtual_overload, private access, and final.
Differential Revision: https://reviews.llvm.org/D92855
Fix static analyzer warnings - castAs<> will assert the type is correct, but getAs<> just returns null, which would just result in a dereferenced null pointer.
This is needed so a listener hears all changes during the dialect
conversion to allow correct rollbacks upon failure.
Differential Revision: https://reviews.llvm.org/D92685
The interesting change here is that we no longer consider `__convert_to_integral`
an ADL customization point for the user's types. I think the new behavior
is defensible. The old behavior had come from D7449, where Marshall explicitly
said "people can't define their own [`__convert_to_integral` overloads]."
Differential Revision: https://reviews.llvm.org/D92814
It adds coverage for back-deploying to a system that contains the
filesystem library, which 10.9 (currently our only back-deployment
target in the CI) does not have.
Differential Revision: https://reviews.llvm.org/D92794
Austin Kerbow