Doing so could cause a bug where the linker tries to remap a function
"reimported" from the current module when materializing it, causing a
lookup assert in the type mappings.
Such threads can cause false leak reports,
but often it's hard to diagnose the reason of
failed PTRACE_ATTACH. Maybe we can find
a clue from `/proc/*/task/*/status`
Temporarily Revert until Chelsea can look at this. With a clean build,
SBLanguages.h won't be generated in the build directory at the point
when it is included by lldb-enumerations when compiling e.g.
Broadcaster.cpp. On a clean build (no pre-existing build directory),
the dependency ordering is not explicitly stated so the build will fail.
An incremental build will succeed.
This reverts commit b3554265f2.
Need to track changes with the repeated reduced value, since it might be
vectorized in the next attempt for reduction vectorization, to correctly
generate the code and avoid compiler crash.
Fixes#111887
This fixes the following assertion: "Cannot create Expected<T> from
Error success value." The problem was that GetFrameBaseValue return
false without updating the Status argument. This patch eliminates the
opportunity for mistakes by returning an llvm:Error.
This commit only adds support for the
`SBProcess::ReverseContinue()` API. A user-accessible command for this
will follow in a later commit.
This feature depends on a gdbserver implementation (e.g. `rr`) providing
support for the `bc` and `bs` packets. `lldb-server` does not support
those packets, and there is no plan to change that. So, for testing
purposes, `lldbreverse.py` wraps `lldb-server` with a Python
implementation of *very limited* record-and-replay functionality for use
by *tests only*.
The majority of this PR is test infrastructure (about 700 of the 950
lines added).
Update the llvm/docs/Coroutines.rst docs to include a full description
of Custom ABI objects. This documentation describes the how ABI objects
allow users (plugin libraries) to create custom ABI objects for their
needs.
Before CUDA 12.3 `ptxas` did not recognize that the trap instruction
terminates a basic block. Instead, it would assume that control flow
continued to the next instruction. The next instruction could be in the
block that's lexically below it. This would lead to phantom CFG edges
being created within ptxas.
[NVPTX: Lower unreachable to exit to allow ptxas to accurately
reconstruct the
CFG.](1ee4d880e8)
added the LowerUnreachable pass to NVPTX to work around this. Several
other WAR patches followed.
This bug in `ptxas` was fixed in CUDA 12.3 and is thus impossible to
encounter when targeting PTX ISA v8.3+
This commit reverts the WARs for the `ptxas` bug when targeting PTX ISA
v8.3+
CC @maleadt
First, ReadlanePieces should be in the scope of each MachineOperand. It
is not correct if we declare in a outer scope without clearing after the
use for a MachineOperand.
Additionally, we do not need the OrigBB argyment for
emitLoadScalarOpsFromVGPRLoop, since MachineFunction (the only use) can
be obtained from LoopBB (or BodyBB).
Update fixupIVUsers to compute the value for escaped inductions using
the already computed end value of the induction (EndValue), but
subtracting the step.
This results in slightly simpler codegen, as we avoid computing the full
transformed index at VectorTripCount - 1.
PR: https://github.com/llvm/llvm-project/pull/110576
To consolidate behavior of function mangling and limit the number of
places that ABI changes will need to be made, this switches the DirectX
target used for HLSL to use the Itanium ABI from the Microsoft ABI. The
Itanium ABI has greater flexibility in decisions regarding mangling of
new types of which we have more than a few yet to add.
One effect of this will be that linking library shaders compiled with
DXC will not be possible with shaders compiled with clang. That isn't
considered a terribly interesting use case and one that would likely
have been onerous to maintain anyway.
This involved adding a function to call all global destructors as the
Microsoft ABI had done.
This requires a few changes to tests. Most notably the mangling style
has changed which accounts for most of the changes. In making those
changes, I took the opportunity to harmonize some very similar tests for
greater consistency. I also shaved off some unneeded run flags that had
probably been copied over from one test to another.
Other changes effected by using the new ABI include using different
types when manipulating smaller bitfields, eliminating an unnecessary
alloca in one instance in this-assignment.hlsl, changing the way static
local initialization is guarded, and changing the order of inout
parameters getting copied in and out. That last is a subtle change in
functionality, but one where there was sufficient inconsistency in the
past that standardizing is important, but the particular direction of
the standardization is less important for the sake of existing shaders.
fixes#110736
- add additional lowering for directx backend in CGBuiltin.cpp
- add directx intrinsic to IntrinsicsDirectX.td
- add semantic check of arguments in SemaHLSL.cpp
- add mapping to DXIL op in DXIL.td
- add testing of semantics in WaveGetLaneIndex-errors.hlsl
- add testing of dxil lowering in WaveGetLaneIndex.ll
Resolves#70105
This commit changes the libc++ frame recognizer to hide implementation
details of libc++ more aggressively. The applied heuristic is rather
straightforward: We consider every function name starting with `__` as
an implementation detail.
This works pretty neatly for `std::invoke`, `std::function`,
`std::sort`, `std::map::emplace` and many others. Also, this should
align quite nicely with libc++'s general coding convention of using the
`__` for their implementation details, thereby keeping the future
maintenance effort low.
However, this heuristic by itself does not work in 100% of the cases:
E.g., `std::ranges::sort` is not a function, but an object with an
overloaded `operator()`, which means that there is no actual call
`std::ranges::sort` in the call stack. Instead, there is a
`std::ranges::__sort::operator()` call. To make sure that we don't hide
this stack frame, we never hide the frame which represents the entry
point from user code into libc++ code
The semantics utility GetAllNames has declarations in two header files
and a definition that really should be in the common utilities source
file. Remove the redudant declaration from resolve-names-utils.h and
move code from resolve-names-utils.cpp into Semantics/tools.cpp.
ProgramTree instances are created as the value of a local variable in
the Pre(const parser::ProgramUnit &) member function in name resolution.
But references to these ProgramTree instances can persist in
SubprogramNameDetails symbol table entries that might survive that
function call's lifetime, and lead to trouble later when (e.g.)
expression semantics needs to deal with a possible forward reference in
a function reference in an expression being processed later in
expression checking.
So put those ProgramTree instances into a longer-lived linked list
within the SemanticsContext.
Might fix some weird crashes reported on big-endian targets (AIX &
Solaris).
When an OPEN statement with a unit number fails in a recoverable manner,
the runtime needs to delete the ExternalFileUnit instance that was
created in the unit map. And we do this too soon -- that instance still
holds some of the I/O statement state that will be used by a later call
into the runtime for EndIoStatement.
Move the code that deletes the unit after a failed but recoverable OPEN
into ExternalIoStatementBase::EndIoStatement, and don't do things
afterwards that would need the I/O statement state that has been
destroyed.
Fixes https://github.com/llvm/llvm-project/issues/111404.
In [[NVPTX] Improve lowering of
v4i8](cbafb6f2f5)
@Artem-B add the ability to lower ISD::BUILD_VECTOR with bfi PTX
instructions. @Artem-B did this because:
([source](https://github.com/llvm/llvm-project/pull/67866#discussion_r1343066911))
> Under the hood byte extraction/insertion ends up as BFI/BFE
instructions, so we may as well do that in PTX, too.
https://godbolt.org/z/Tb3zWbj9b
However, the example that @Artem-B linked was targeting sm_52. On modern
architectures, ptxas uses prmt.b32.
[Example](https://godbolt.org/z/Ye4W1n84o).
Thus, remove uses of NVPTXISD::BFI in favor of NVPTXISD::PRMT.
This is a purely mechanical commit for fixing the indentation of the
runtimes' CMakeLists files after #80007. That PR didn't update the
indentation in order to make the diff easier to review and for merge
conflicts to be easier to resolve (for downstream changes).
This doesn't change any code, it only reindents it.
A COMMON block is a named area of memory that holds a collection of
variables. Fortran subprograms may map the COMMON block memory area to a
list of variables. A common block is represented in LLVM debug by
DICommonBlock.
This PR adds support for this in MLIR. The changes are mostly mechanical
apart from small change to access the DICompileUnit when the scope of
the variable is DICommonBlock.
---------
Co-authored-by: Tobias Gysi <tobias.gysi@nextsilicon.com>
LLVM now triggers an assertion when the format string and arguments
don't match. Fix a variety of incorrect format strings I discovered when
enabling logging with a debug build.
In tosa valiation pass, change the type of profile option to ListOption.
Now TOSA profiles is turned from hierarchical to composable. Each
profile is an independent set, i.e. an target can implement multiple
profiles.
Set the profile option to none by default, and limit to profiles if
requested.
The profiles can be specified via command line, e.g.
$ mlir-opt ... --tosa-validate="profile=bi,mi" which tells the valiation
pass that BI and MI are enabled.
Change-Id: I1fb8d0c1b27eccd768349b6eb4234093313efb57
By allowing AnnotateAttr to be applied to statements, users can place arbitrary information in the AST for later use.
For example, this can be used for HW-targeted language extensions that involve specialized loop annotations.
