Summary:
According to
https://docs.nvidia.com/cuda/archive/10.0/ptx-writers-guide-to-interoperability/index.html#cuda-specific-dwarf,
the compiler should emit the DW_AT_address_class attribute for all
variable and parameter. It means, that DW_AT_address_class attribute
should be used in the non-standard way to support compatibility with the
cuda-gdb debugger.
Clang is able to generate the information about the variable address
class. This information is emitted as the expression sequence
`DW_OP_constu <DWARF Address Space> DW_OP_swap DW_OP_xderef`. The patch
tries to find all such expressions and transform them into
`DW_AT_address_class <DWARF Address Space>` if target is NVPTX and the debugger is gdb.
If the expression is not found, then default values are used. For the
local variables <DWARF Address Space> is set to ADDR_local_space(6), for
the globals <DWARF Address Space> is set to ADDR_global_space(5). The
values are taken from the table in the same section 5.2. CUDA-Specific
DWARF Definitions.
Reviewers: echristo, probinson
Subscribers: jholewinski, aprantl, llvm-commits
Differential Revision: https://reviews.llvm.org/D57157
llvm-svn: 353203
The v2i64 argument is lowered to a bitcast of v4i32 build_vector.
This would then attempt to use the i32-element as the source of the
vector truncate. This really would need to collect 2 elements from the
build_vector to produce the intended truncate.
llvm-svn: 353202
This patch removes some vendor-specific availability XFAILs from the
test suite. In the future, when a new feature is introduced in the
dylib, an availability macro should be created and a matching lit
feature should be created. That way, the test suite can XFAIL whenever
the implementation lacks the necessary feature instead of being
cluttered by vendor-specific annotations.
Right now, those vendor-specific annotations are still somewhat cluttering
the test suite by being in `config.py`, but at least they are localized.
In the future, we could design a way to define those less intrusively or
even automatically based on the availability macros that already exist
in <__config>.
llvm-svn: 353201
rL352997 enabled ZERO_EXTEND from non-shuffle-able value types. I've disabled it for now to fix a regression identified by @asbirlea until I can fix this properly.
llvm-svn: 353198
Change the format type of Value to PRIu64 since it is a uint64_t.
The problem was detected on mips boards building 32-bit binaries,
where it was printing junk values and causing test failure.
Patch by Milos Stojanovic.
Differential Revision: https://reviews.llvm.org/D57583
llvm-svn: 353194
Summary:
Adds the standard gauntlet of accessors for global indirect functions and updates the echo test.
Now it would be nice to have a target abstraction so one could know if they have access to a suitable ELF linker and runtime.
Reviewers: whitequark, deadalnix
Reviewed By: whitequark
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D56177
llvm-svn: 353193
Type punning through a union -> no good.
double to uint64 to double again -> no good either.
The nice side effect, other than silencing the sanitizer bot
is that it fixes the formatting of some dates, e.g. Jan 1st 1970.
<rdar://problem/47617983>
llvm-svn: 353191
We can't outline BTI instructions, because they need to be the very first
instruction executed after an indirect call or branch. If we outline them, then
an indirect call might go to the branch to the outlined function, which will
fault.
Differential revision: https://reviews.llvm.org/D57753
llvm-svn: 353190
The fix is to issue error messages if there are more than one
teams construct inside a target constructs.
#pragma omp target
{
#pragma omp teams
{ ... }
#pragma omp teams
{ ... }
}
llvm-svn: 353186
Emit{Nounwind,}RuntimeCall{,OrInvoke} have been modified to take a
FunctionCallee as an argument, and CreateRuntimeFunction has been
modified to return a FunctionCallee. All callers have been updated.
Additionally, CreateBuiltinFunction is removed, as it was redundant
with CreateRuntimeFunction after some previous changes.
Differential Revision: https://reviews.llvm.org/D57668
llvm-svn: 353184
edge cases.
Currently, EmitCall emits a call instruction with a function type
derived from the pointee-type of the callee. This *should* be the same
as the type created from the CallInfo parameter, but in some cases an
incorrect CallInfo was being passed.
All of these fixes were discovered by the addition of the assert in
EmitCall which verifies that the passed-in CallInfo matches the
Callee's function type.
As far as I know, these issues caused no bugs at the moment, as the
correct types were ultimately being emitted. But, some would become
problematic when pointee types are removed.
List of fixes:
* arrangeCXXConstructorCall was passing an incorrect value for the
number of Required args, when calling an inheriting constructor
where the inherited constructor is variadic. (The inheriting
constructor doesn't actually get passed any of the user's args, but
the code was calculating it as if it did).
* arrangeFreeFunctionLikeCall was not including the count of the
pass_object_size arguments in the count of required args.
* OpenCL uses other address spaces for the "this" pointer. However,
commonEmitCXXMemberOrOperatorCall was not annotating the address
space on the "this" argument of the call.
* Destructor calls were being created with EmitCXXMemberOrOperatorCall
instead of EmitCXXDestructorCall in a few places. This was a problem
because the calling convention sometimes has destructors returning
"this" rather than void, and the latter function knows about that,
and sets up the types properly (through calling
arrangeCXXStructorDeclaration), while the former does not.
* generateObjCGetterBody: the 'objc_getProperty' function returns type
'id', but was being called as if it returned the particular
property's type. (That is of course the *dynamic* return type, and
there's a downcast immediately after.)
* OpenMP user-defined reduction functions (#pragma omp declare
reduction) can be called with a subclass of the declared type. In
such case, the call was being setup as if the function had been
actually declared to take the subtype, rather than the base type.
Differential Revision: https://reviews.llvm.org/D57664
llvm-svn: 353181
Summary:
While the backend code of FileCheck relies on definition of variable
from the command-line to have an equal sign '=' and a variable name
before that, the frontend does not actually enforce it. This leads to
FileCheck crashing when invoked with invalid syntax for the -D option.
This patch adds the missing validation in the frontend. It also makes
the -D option an AlwaysPrefix option to be able to detect -D=FOO as
being a define without variable and -D as missing its value.
Copyright:
- Linaro (changes in version 2 of revision D55940)
- GraphCore (changes in later versions)
Reviewers: jdenny
Subscribers: JonChesterfield, hiraditya, kristina, probinson,
llvm-commits
Differential Revision: https://reviews.llvm.org/D55940
llvm-svn: 353173
Summary:
Add support for options that always prefix their value, giving an error
if the value is in the next argument or if the option is given a value
assignment (ie. opt=val). This is the desired behavior for the -D option
of FileCheck for instance.
Copyright:
- Linaro (changes in version 2 of revision D55940)
- GraphCore (changes in later versions and introduced when creating
D56549)
Reviewers: jdenny
Subscribers: llvm-commits, probinson, kristina, hiraditya,
JonChesterfield
Differential Revision: https://reviews.llvm.org/D56549
llvm-svn: 353172
DispatchStage should always delegate to an object of class RegisterFile the task
of updating data dependencies. ReadState and WriteState objects should not be
modified directly by DispatchStage.
This patch also renames stage IS_AVAILABLE to IS_DISPATCHED.
llvm-svn: 353170
Trivial fix: decode was not called for all subprocess.check_output calls.
Commited on behalf of Andrew Boyarshin
Differential Revision: https://reviews.llvm.org/D57505
llvm-svn: 353168
When Clang/LLVM is built with the CLANG_DEFAULT_STD_CXX CMake macro that sets
the default standard to something other than C++14, there are a number of lit
tests that fail as they rely on the C++14 default.
This patch just adds the language standard option explicitly to such test cases.
Differential revision: https://reviews.llvm.org/D57581
llvm-svn: 353163
In some cases, it is faster to just grow the set of 'Users' rather than
performing a llvm::find_if every time a new user is added to
the set. No functional change intended.
llvm-svn: 353162
When we attempt to add an addr space qual to a type already
qualified by an addr space ICE is triggered. Before creating
a type with new address space, remove the old addr space.
Fixing PR38614!
Differential Revision: https://reviews.llvm.org/D57524
llvm-svn: 353160
Some use cases are appearing where salvaging is needed that does not
correspond to an instruction being deleted -- for example an instruction
being sunk, or a Value not being available in a block being isel'd.
Enable more fine grained control over how salavging occurs by splitting
the logic into helper functions, separating things that are specific to
working on DbgVariableIntrinsics from those specific to interpreting IR
and building DIExpressions.
Differential Revision: https://reviews.llvm.org/D57696
llvm-svn: 353156
The test started failing for me recently. I don't see any changes around
this code, so maybe it's my local go version that changed or something.
The error seems real to me: we're trying to print an Attribute with %d.
The test talks about "attribute masks" I'm not sure what that refers to,
but I suppose we could print the raw pointer value, since that's
what the test seems to be comparing.
Differential revision: https://reviews.llvm.org/D57672
llvm-svn: 353155
If we have broadcasts of different vector widths, keep the longest vector width and extract subvectors for the shorter vectors (which should be free).
Differential Revision: https://reviews.llvm.org/D57663
llvm-svn: 353154
This patch improves code generation for some AArch64 ACLE intrinsics. It adds
support to CGP to duplicate and sink operands to their user, if they can be
folded into a target instruction, like zexts and sub into usubl. It adds a
TargetLowering hook shouldSinkOperands, which looks at the operands of
instructions to see if sinking is profitable.
I decided to add a new target hook, as for the sinking to be profitable,
at least on AArch64, we have to look at multiple operands of an
instruction, instead of looking at the users of a zext for example.
The sinking is done in CGP, because it works around an instruction
selection limitation. If instruction selection is not limited to a
single basic block, this patch should not be needed any longer.
Alternatively this could be done in the LoopSink pass, which tries to
undo LICM for instructions in blocks that are not executed frequently.
Note that we do not force the operands to sink to have a single user,
because we duplicate them before sinking. Therefore this is only
desirable if they really can be done for free. Additionally we could
consider the impact on live ranges later on.
This should fix https://bugs.llvm.org/show_bug.cgi?id=40025.
As for performance, we have internal code that uses intrinsics and can
be speed up by 10% by this change.
Reviewers: SjoerdMeijer, t.p.northover, samparker, efriedma, RKSimon, spatel
Reviewed By: samparker
Differential Revision: https://reviews.llvm.org/D57377
llvm-svn: 353152
Summary:
This patch adds support of expression evaluation in a context of some object.
Consider the following example:
```
struct S {
int a = 11;
int b = 12;
};
int main() {
S s;
int a = 1;
int b = 2;
// We have stopped here
return 0;
}
```
This patch allows to do something like that:
```
lldb.frame.FindVariable("s").EvaluateExpression("a + b")
```
and the result will be `33` (not `3`) because fields `a` and `b` of `s` will be
used (not locals `a` and `b`).
This is achieved by replacing of `this` type and object for the expression. This
has some limitations: an expression can be evaluated only for values located in
the debuggee process memory (they must have an address of `eAddressTypeLoad`
type).
Reviewers: teemperor, clayborg, jingham, zturner, labath, davide, spyffe, serge-sans-paille
Reviewed By: jingham
Subscribers: abidh, lldb-commits, leonid.mashinskiy
Tags: #lldb
Differential Revision: https://reviews.llvm.org/D55318
llvm-svn: 353149
LLVM_ENABLED_PROJECT and reconfigured it had no effect on what
projects were actually built. This was very confusing behaviour. The
reason for this is that the value of the `LLVM_TOOL_<PROJECT>_BUILD`
variables are already set.
The problem here is that we have two sources of truth:
* The projects listed in LLVM_ENABLE_PROJECTS.
* The projects enabled/disabled with LLVM_TOOL_<PROJECT>_BUILD.
At configure time we have no real way of knowing which source of truth
the user wants so we apply the following heuristic:
If the user ever sets `LLVM_ENABLE_PROJECTS` in the CMakeCache then that
is used as the single source of truth and we force the
`LLVM_TOOL_<PROJECT>_BUILD` CMake cache variables to have the
appropriate values that match the contents of the
`LLVM_ENABLE_PROJECTS`. If the user never sets `LLVM_ENABLE_PROJECTS`
then they can continue to use and set the `LLVM_TOOL_<PROJECT>_BUILD`
variables as the "source of truth".
The problem with this approach is that if the user ever tries to use
both `LLVM_ENABLE_PROJECTS` and `LLVM_TOOL_<PROJECT>_BUILD` for the same
build directory then any user set value for `LLVM_TOOL_<PROJECT>_BUILD`
variables will get overwriten, likely without the user noticing.
Hopefully the above shouldn't matter in practice because the
LLVM_TOOL_<PROJECT>_BUILD variables are not documented, but
LLVM_ENABLE_PROJECTS is.
We should probably deprecate the `LLVM_TOOL_<PROJECT>_BUILD`
variables at some point by turning them into to regular CMake
variables that don't live in the CMake cache.
Differential Revision: https://reviews.llvm.org/D57535
llvm-svn: 353148
Alexey Bataev