We can use *Set::insert_range to collapse:
for (auto Elem : Range)
Set.insert(E);
down to:
Set.insert_range(Range);
In some cases, we can further fold that into the set declaration.
Option becomes: -instruction-tables=`<level>`
The choice of `<level>` controls number of printed information.
`<level>` may be `none` (default), `normal`, `full`.
Note: If the option is used without `<label>`, default is `normal`
(legacy).
When `<level>` is `full`, additional information are:
- `<Bypass Latency>`: Latency when a bypass is implemented between
operands
in pipelines (see SchedReadAdvance).
- `<LLVM Opcode Name>`: mnemonic plus operands identifier.
- `<Resources units>`: Used resources associated with LLVM Opcode.
- `<instruction comment>`: reports comment if any from source assembly.
Level `full` can be used to better check scheduling info when TableGen
is modified.
LLVM Opcode name help to find right instruction regexp to fix TableGen
Scheduling Info.
-instruction-tables=full option is validated on
AArch64/Neoverse/V1-sve-instructions.s
Follow up of MR #126703
---------
Co-authored-by: Julien Villette <julien.villette@sipearl.com>
So far I've been unsuccessful in finding an example where the used constant
value is directly observed in the output. This avoids an assert in an intermediate
step of value replacement.
Summary:
There were a few issues with the first one, leading to some errors and
warnings. Most importantly, this was building on MSVC which isn't
supported.
DenseSet, SmallPtrSet, SmallSet, SetVector, and StringSet recently
gained C++23-style insert_range. This patch replaces:
Dest.insert(Src.begin(), Src.end());
with:
Dest.insert_range(Src);
This patch does not touch custom begin like succ_begin for now.
Summary:
These tools `amdhsa-loader` and `nvptx-loader` are used to execute unit
tests directly on the GPU. We use this for `libc` and `libcxx` unit
tests as well as general GPU experimentation. It looks like this.
```console
> clang++ main.cpp --target=amdgcn-amd-amdhsa -mcpu=native -flto -lc ./lib/amdgcn-amd-amdhsa/crt1.o
> llvm-gpu-loader a.out
Hello World!
```
Currently these are a part of the `libc` project, but this creates
issues as `libc` itself depends on them to run tests. Right now we get
around this by force-including the `libc` project prior to running the
runtimes build so that this dependency can be built first. We should
instead just make this a simple LLVM tool so it's always available.
This has the effect of installing these by default now instead of just
when `libc` was enabled, but they should be relatively small. Right now
this only supports a 'static' configuration. That is, we locate the CUDA
and HSA dependencies at LLVM compile time. In the future we should be
able to provide this by default using `dlopen` and some API.
I don't know if it's required to reformat all of these names since they
used the `libc` naming convention so I just left it for now.
Before this patch, whole program devirtualization is suppressed on a
class if any superclass is visible to regular object files, by recording
the class GUID in `VisibleToRegularObjSymbols`.
This patch suppresses whole program devirtualization on a class if the
LTO unit doesn't have the prevailing definition of this class (e.g., the
prevailing definition is in a shared library)
Implementation summaries:
1. In llvm/lib/LTO/LTO.cpp, `IsVisibleToRegularObj` is updated to look
at the global resolution's `IsPrevailing` bit for ThinLTO and
regularLTO.
2. In llvm/tools/llvm-lto2/llvm-lto2.cpp,
- three command line options are added so `llvm-lto2` can override
`Conf.HasWholeProgramVisibility`, `Conf.ValidateAllVtablesHaveTypeInfos`
and `Conf.AllVtablesHaveTypeInfos`.
The test case is reduced from a small C++ program (main.cc, lib.cc/h
pasted below in [1]). To reproduce the program failure without this
patch, compile lib.cc into a shared library, and provide it to a ThinLTO
build of main.cc (commands are pasted in [2]).
[1]
* lib.h
```
#include <cstdio>
class Derived {
public:
void dispatch();
virtual void print();
virtual void sum();
};
void Derived::dispatch() {
static_cast<Derived*>(this)->print();
static_cast<Derived*>(this)->sum();
}
void Derived::sum() {
printf("Derived::sum\n");
}
__attribute__((noinline)) void* create(int i);
__attribute__((noinline)) void* getPtr(int i);
```
* lib.cc
```
#include "lib.h"
#include <cstdio>
#include <iostream>
class Derived2 : public Derived {
public:
void print() override {
printf("DerivedSharedLib\n");
}
void sum() override {
printf("DerivedSharedLib::sum\n");
}
};
void Derived::print() {
printf("Derived\n");
}
__attribute__((noinline)) void* create(int i) {
if (i & 1)
return new Derived2();
return new Derived();
}
```
* main.cc
```
cat main.cc
#include "lib.h"
class DerivedN : public Derived {
public:
};
__attribute__((noinline)) void* getPtr(int x) {
return new DerivedN();
}
int main() {
Derived*b = static_cast<Derived*>(create(201));
b->dispatch();
delete b;
Derived* a = static_cast<Derived*>(getPtr(202));
a->dispatch();
delete a;
return 0;
}
```
[2]
```
# compile lib.o in a shared library.
$ ./bin/clang++ -O2 -fPIC -c lib.cc -o lib.o
$ ./bin/clang++ -shared -o libdata.so lib.o
# Provide the shared library in `-ldata`
$ ./bin/clang++ -v -g -ldata --save-temps -fno-discard-value-names -Wl,-mllvm,-print-before=wholeprogramdevirt -Wl,-mllvm,-wholeprogramdevirt-check=trap -Rpass=wholeprogramdevirt -Wl,--lto-whole-program-visibility -Wl,--lto-validate-all-vtables-have-type-infos -mllvm -disable-icp=true -Wl,-mllvm,-disable-icp=false -flto=thin -fwhole-program-vtables -fno-split-lto-unit -fuse-ld=lld main.cc -L . -o main >/tmp/wholeprogramdevirt.ir 2>&1
# Run the program hits a segmentation fault with `-Wl,-mllvm,-wholeprogramdevirt-check=trap`
$ LD_LIBRARY_PATH=. ./main
DerivedSharedLib
Trace/breakpoint trap (core dumped)
```
Rather than requiring users to pass `-m64` to the `llvm-ml` driver to
get 64-bit behavior, we add the `llvm-ml64` alias, matching the behavior
of `ML.EXE` and `ML64.EXE`.
The original flavor/bitness flags still work, but the alias should make
some workflows easier.
NOTE: The logic for this already existed in the code; we're just finally
adding the build/install instructions to match.
RPaths are basically search paths for how to load dependent libraries.
The order they appear is the order the linker will search, we should
preserve that order in tbd files.
* Additionally add this level of detection to llvm-readtapi.
resolves: rdar://145603347
Currently, `DIContext::getLineInfoForAddress` and
`DIContext::getLineInfoForDataAddress` returns empty DILineInfo when the
debug info is missing for the given address. This is not differentiable
with the case when debug info is found for the given address but the
debug info is default value (filename:linenum is <invalid>:0).
This change wraps the return types of `DIContext::getLineInfoForAddress`
and `DIContext::getLineInfoForDataAddress` with `std::optional`.
Until recently checkFeature was quite slow. #130936
I was curious where we use checkFeature and noticed these. I thought we
could use hasFeature instead of going through strings.
These date back to when the non-intrinsic format of variable locations
was still being tested and was behind a compile-time flag, so not all
builds / bots would correctly run them. The solution at the time, to get
at least some test coverage, was to have tests opt-in to non-intrinsic
debug-info if it was built into LLVM.
Nowadays, non-intrinsic format is the default and has been on for more
than a year, there's no need for this flag to exist.
(I've downgraded the flag from "try" to explicitly requesting
non-intrinsic format in some places, so that we can deal with tests that
are explicitly about non-intrinsic format in their own commit).
This is meant as a preparation for PR #130988 "[AMDGPU] Implement IR
expansion for frem instruction" which implements the expansion of
another instruction in this pass. The more general name seems more
appropriate given this change and quite reasonable even without it.
Current implementation (for AArch64) only supports the GRP32, GPR64,
FPR64/128, PPR16 and ZPR128 register classes. This adds support for
the other floating point register classes to initialize registers and avoid
the "setReg is not implemented" warning for these cases.
This patch adds SYCL Module splitting - the necessary step in the SYCL
compilation pipeline. Only 2 splitting modes are being added in this
patch: by kernel and by source.
Refactor readobj to integrate AArch64 Build Attributes under
ELFAttributeParser. ELFAttributeParser now serves as a base class for:
- ELFCompactAttrParser, handling Arm-style attributes with a single
build attribute subsection.
- ELFExtendedAttrParser, handling AArch64-style attributes with multiple
build attribute subsections. This improves code organization and better
aligns with the attribute parsing model.
Add support for parsing AArch64 Build Attributes.
This reverts commit b0baa1d8bd68a2ce2f7c5f2b62333e410e9122a1.
Reverting follow-up commit to ce9e1d3c15ed6290f1cb07b482939976fa8115cd since the original commit test is flaky.
…Stream.
CachedFileStream has previously performed the commit step in its
destructor, but this means its only recourse for error handling is
report_fatal_error. Modify this to add an explicit commit() method, and
call this in the appropriate places with appropriate error handling for
the location.
Currently the destructor of CacheStream gives an assert failure in Debug
builds if commit() was not called. This will help track down any
remaining uses of the API that assume the old destructior behaviour. In
Release builds we fall back to the previous behaviour and call
report_fatal_error if the commit fails.
This change means that llvm-strip no longer exits immediately upon
encountering an error when modifying a file and will instead continue
modifying the other inputs. Fixes#129412
The module currently stores the target triple as a string. This means
that any code that wants to actually use the triple first has to
instantiate a Triple, which is somewhat expensive. The change in #121652
caused a moderate compile-time regression due to this. While it would be
easy enough to work around, I think that architecturally, it makes more
sense to store the parsed Triple in the module, so that it can always be
directly queried.
For this change, I've opted not to add any magic conversions between
std::string and Triple for backwards-compatibilty purses, and instead
write out needed Triple()s or str()s explicitly. This is because I think
a decent number of them should be changed to work on Triple as well, to
avoid unnecessary conversions back and forth.
The only interesting part in this patch is that the default triple is
Triple("") instead of Triple() to preserve existing behavior. The former
defaults to using the ELF object format instead of unknown object
format. We should fix that as well.
Current implementation (for Aarch64) in llvm-exegesis only supports
GRP32 and GPR64 bit register class, thus for opcodes variants which used
FPR64/128, PPR16 and ZPR128, llvm-exegesis throws warning "setReg is not
implemented". This code will handle the above register class and
initialize the registers using appropriate base instruction class.
