In ExprEngine::bindReturnValue() we cast an SVal to DefinedOrUnknownSVal,
however this SVal can also be Undefined, which leads to an assertion failure.
Fixes: #56873
Differential Revision: https://reviews.llvm.org/D130974
As progress towards having FileManager::getFileRef() return the path
as-requested by default, return a FileEntryRef that can use
getNameAsRequested() to retrieve this path, with the ultimate goal that
this should be the behaviour of getName() and clients should explicitly
request the "external" name if they need to (see comment in
FileManager::getFileRef). For now, getName() continues to return the
external path by looking through the redirects.
For now, the new function is only used in unit tests.
Differential Revision: https://reviews.llvm.org/D131004
Refactor the mechanism for sending interrupt packets to send them
from async thread (i.e. the same thread that sends the continue packet
preceding them and that waits for the response), rather than from
the thread requesting the interrupt. This is going to become especially
important when using the vCtrlC packet as part of the non-stop protocol,
as -- unlike the simple ^c sent in the all-stop mode -- this packet
involves an explicit reply.
Suggested by Pavel Labath in D126614.
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.llvm.org/D131075
https://github.com/llvm/llvm-project/issues/56884
The root problem is in isOpenMPRebuildMemberExpr, it is only need to rebuild
for field expression. No need for member function call.
The fix is to check field for member expression and skip rebuild for member
function call.
Differential Revision: https://reviews.llvm.org/D131024
This patch modifies context-sensitive analysis of functions to use a cache,
rather than recreate the `ControlFlowContext` from a function decl on each
encounter. However, this is just step 1 (of N) in adding support for a
configurable map of "modeled" function decls (see issue #56879). The map will go
from the actual function decl to the `ControlFlowContext` used to model it. Only
functions pre-configured in the map will be modeled in a context-sensitive way.
We start with a cache because it introduces the desired map, while retaining the
current behavior. Here, functions are mapped to their actual implementations
(when available).
Differential Revision: https://reviews.llvm.org/D131039
This is mostly a stylistic change to make the uniform memop widening cost
code fit more naturally with the sourounding code. Its not strictly
speaking NFC as I added in the store with invariant value case, and we
could in theory have a target where a gather/scatter is cheaper than a
single load/store... but it's probably NFC in practice. Note that the
scatter/gather result can still be overriden later if the result is
uniform-by-parts.
When doing that NFC refactor, I'd messed up how char8_t was reported,
which resulted in it being considered a 'future' keyword, without the
corresponding diagnostic, which lead to an assert. This patch corrects
the char8_t to ONLY be future in C++ mode.
Currently, lldb-server was opening the executable file to determine the
process architecture (to differentiate between 32 and 64 bit
architecture flavours). This isn't a particularly trustworthy source of
information (the file could have been changed since the process was
started) and it is not always available (file could be deleted or
otherwise inaccessible).
Unfortunately, ptrace does not give us a direct API to access the
process architecture, but we can still infer it via some of its
responses -- given that the general purpose register set of 64-bit
applications is larger [citation needed] than the GPR set of 32-bit
ones, we can just ask for the application GPR set and check its size.
This is what this patch does.
Differential Revision: https://reviews.llvm.org/D130985
The getKeywordStatus function is a horrible mess of inter-dependent 'if'
statements that depend significantly on the ORDER of the checks. This
patch removes the dependency on order by checking each set-flag only
once.
It does this by looping through each of the set bits, and checks each
individual flag for its effect, then combines them at the end.
This might slow down startup performance slightly, as there are only a
few hundred keywords, and a vast majority will only get checked 1x
still.
This patch ALSO removes the KEYWORD_CONCEPTS flag, because it has since
become synonymous with C++20.
Differential Revision: https://reviews.llvm.org/D131007
The SystemZ ABI says that 128 bit integers should be aligned to only 8 bytes.
Reviewed By: Ulrich Weigand, Nikita Popov
Differential Revision: https://reviews.llvm.org/D130900
Trying to be generic didn't work properly because we had to special-case
some interface libraries that we didn't want in the linker script. Instead,
only look at the ABI and the unwinding libraries explicitly.
This should solve the issue reported by @dim in [1].
[1]: https://discourse.llvm.org/t/15-0-0-rc1-has-been-tagged/64174/22
Differential Revision: https://reviews.llvm.org/D131037
It must have been a copy-paste error, since cxx-include is never defined
by the libunwind config.
Differential Revision: https://reviews.llvm.org/D131030
This patch ensures consistency in the construction of FP_ROUND nodes
such that they always use ISD::TargetConstant instead of ISD::Constant.
This additionally fixes a bug in the AArch64 SVE backend where patterns
were matching against TargetConstant nodes and sometimes failing when
passed a Constant node.
Reviewed By: paulwalker-arm
Differential Revision: https://reviews.llvm.org/D130370
An unnecessary sext.w is generated when masking the result of the
riscv_masked_cmpxchg_i64 intrinsic. Implementing handling of the
intrinsic in ComputeNumSignBitsForTargetNode allows it to be removed.
Although this isn't a particularly important optimisation, removing the
sext.w simplifies implementation of an additional cmpxchg-related
optimisation in D130192.
Although I can't produce a test with different codegen for the other
atomics intrinsics, these are added as well for completeness.
Differential Revision: https://reviews.llvm.org/D130191
Enable SGPRs for the following operands of these opcodes:
- src operands of VOP3 variant.
- src2 operand of DPP variants.
Differential Revision: https://reviews.llvm.org/D130989
The reconciliation pass has been improved to introduce the support for chains of casts, thus not limiting anymore the reconciliation to just consider pairs of unrealized casts.
Reviewed By: ftynse
Differential Revision: https://reviews.llvm.org/D130711
As reported in Issue #56850, mlir/lib/Analysis/Presburger/Utils.cpp doesn't
compile on Solaris 11.4/SPARC with the bundled GCC 11, as seen when testing
LLVM 15.0.0 rc1:
/var/llvm/reltest/llvm-15.0.0-rc1/rc1/llvm-project/mlir/include/mlir/Analysis/Presburger/MPInt.h:260:47:
error: inlining failed in call to ‘always_inline’ ‘int64_t
mlir::presburger::int64FromMPInt(const mlir::presburger::MPInt&)’: indirect
function call with a yet undetermined callee
This patch hacks around this and allowed the build to finish.
Tested on `sparcv9-sun-solaris2.11`.
Differential Revision: https://reviews.llvm.org/D131060
`BMI` new instruction `tzcnt` has better performance than `bsf` on new
processors. Its encoding has a mandatory prefix '0xf3' compared to
`bsf`. If we force emit `rep` prefix for `bsf`, we will gain better
performance when the same code run on new processors.
GCC has already done this way: https://c.godbolt.org/z/6xere6fs1Fixes#34191
Reviewed By: skan
Differential Revision: https://reviews.llvm.org/D130956
Unfortunately, this overflow is extremely hard to reproduce reliably (in fact, I was unable to do so). The issue is that:
- getOperandsToCreate sometimes skips creating an SCEV for the LHS
- then, createSCEV is called for the BinaryOp
- ... which calls getNoWrapFlagsFromUB
- ... which under certain circumstances calls isSCEVExprNeverPoison
- ... which under certain circumstances requires the SCEVs of all operands
For certain deep dependency trees, this causes a stack overflow.
Reviewed By: bkramer, fhahn
Differential Revision: https://reviews.llvm.org/D129745
`split-file` is essentially a development tool used for writing tests. Other related tooling for this purpose such as `FileCheck`, `count`, `not` and more are already created via `add_llvm_utility` instead of `add_llvm_tool`. The later is more meant as LLVM tools used as part of a toolchain (eg. `llvm-ar`), not for tools for the development of LLVM itself.
The main semantic difference this makes is that `split-file` is now built and installed via the `LLVM_INSTALL_UTILS` and `LLVM_BUILD_UTILS` instead of `LLVM_BUILD_TOOLS` and `LLVM_INSTALL_TOOLS` cmake flags. That way one can use `LLVM_BUILD_TOOLS=OFF LLVM_INSTALL_UTILS=ON` and have `split-file` installed along side the other testing tools.
Differential Revision: https://reviews.llvm.org/D130977
VisitCXXConstructExpr() will later do something similar, but for large
arrays, we should try to do it once an not for every element.
Fixes#56774
Differential Revision: https://reviews.llvm.org/D130791
In SmallVector unit test, referring test suite members every time using 'this'
(due to how non-dependent name lookups works) is quite cumbersome and can be improved with aliases.
Some test cases already use this approach, in this change we expand that approach to whole test file.
Differential Revision: https://reviews.llvm.org/D129990
Improve the cycle definition, by avoiding usage of not yet defined
or only vaguely defined terminology inside definitions.
More precisely, the existing definition defined "outermost cycles",
and then proceeded to use the term "cycles" for further definitions,
which in turn were used to actually define "cycles".
Now, instead only define "cycles". This does not change the meaning
of a cycle, which depends on the chosen surrounding (subgraph) of a CFG.
Also mention the function CFG in the first definition, because later
later definitions require it anyways.
Also slightly improve the definition of a closed path, by explicitly
requiring the inner nodes to be distinct.
Differential Revision: https://reviews.llvm.org/D130891
Mark ModRefInfo as a bitmask enum, which allows using normal
& and | operators on it. This supersedes various functions like
unionModRef() and intersectModRef(). I think this makes the code
cleaner than going through helper functions...
Differential Revision: https://reviews.llvm.org/D130870
Sometimes SCEV cannot infer nuw/nsw from something as simple as
```
len in [0, MAX_INT]
...
iv = phi(0, iv.next)
guard(iv <s len)
guard(iv <u len)
iv.next = iv + 1
```
just because flag strenthening only relies on definition and does not use local facts.
This patch adds support for the simplest case: inference of flags of `add(x, constant)`
if we can contextually prove that `x <= max_int - constant`.
In case if it has negative CT impact, we can add an option to switch it off. I woudln't
expect that though.
Differential Revision: https://reviews.llvm.org/D129643
Reviewed By: apilipenko
Rename `DataflowAnalysisContext::getStableStorageLocation(QualType)`
to `createStorageLocation`, to make it clear that it doesn't return
a stable storage location.
Differential Revision: https://reviews.llvm.org/D131021
Reviewed-by: ymandel, xazax.hun, gribozavr2
This patch adds constant folder for AtanOp which only supports single and double precision floating-point.
Differential Revision: https://reviews.llvm.org/D130983
Jonas Devlieghere