- Set the shader flag `DisableOptimizations` based on `optnone`
attribute of shader entry functions.
- Add DXIL Metadata Analysis pass as pre-requisite for Shader Flags pass
to obtain entry function information collected therein.
- Named module metadata `dx.disable_optimizations` is intended to
indicate disabling optimizations (`-O0`) via commandline flag. However,
its intent is fulfilled by `optnone` attribute of shader entry functions as
implemented in a recent change, and thus not needed. Delete
generation of named metadata and corresponding test file
`disable_opt.ll`.
- Add tests to verify correctness of setting shader flag.
Closes#112263
We should be able to use `spirv64` as a device variant match and it
should be considered a GPU.
Also add the triple to an RTTI check.
Signed-off-by: Sarnie, Nick <nick.sarnie@intel.com>
Replace some more nvvm.annotations with function attributes,
auto-upgrading the annotations as needed. These new attributes will be
more idiomatic and compile-time efficient than the annotations.
- !"maxclusterrank" / !"cluster_max_blocks" -> "nvvm.maxclusterrank"
- !"minctasm" -> "nvvm.minctasm"
- !"maxnreg" -> "nvvm.maxnreg"
This commit heavily refactors `out-of-bounds-constraint-check.c`:
1. The complex combinations of several `clang_analyzer_eval` calls were
replaced by `clang_analyzer_value`, which can directly query the range
of a symbol.
2. Testcases were renamed to a (hopefully) more consistent scheme.
3. The use of `size_t` as an argument type was replaced by `unsigned
long long`, which is usually a no-op, but seems to be a better choice if
I look for `64u` in the output of `clang_analyzer_value`.
4. The single "dynamic extent" case was generalized into a full set of
tests that use `malloc`.
5. Half of the testcases (the ones that don't use `malloc`) were changed
to use an `int[5]` array instead of a string literal. After this change
the tests in this file cover every functionality that was tested by the
testcases `test_assume_after_access{,2}` in the file `out-of-bounds.c`
so I was able to delete those two testcases (and therefore consolidate
the validation of these constraints within a single test file).
This is the second commit in a series that reorganizes the tests of
`security.ArrayBound` to system that's easier to understand and
maintain. (Note that this file wasn't significantly modified by the
recent commit 6e17ed9b04e5523cc910bf171c3122dcc64b86db which renamed
`alpha.security.ArrayBoundV2` to `security.ArrayBound`; but I still felt
that this cleanup may be useful.)
Multiple people reported flaky bot failures tied to
`clang/test/Analysis/live-stmts.cpp` I tried reproducing the flaky
behavior on my Linux x86_64 system, but the tests appears to be stable
in my context.
Only by looking at the failures reported, I could formulate a potential
diagnosis.
The output always looked almost the same, except that the Exprs dumped
per Basic block were shuffled compared to my expectation. This suggests
to me some ordering issue.
If you look at the backing storage of
`blocksEndToLiveness[B].liveExprs`,
it uses `llvm::ImmutableSet<const Expr *>`.
That container likely uses the pointer values as keys, thus the runtime
values of the addresses influence the iteration order.
To fix this, before dumping, I sort the expressions by their
"beginLocs". It should be efficient enough for a debug checker, where
there is no performance constraint.
This should hopefully fix the flaky behavior on systems where ASLR works
differently than (my) Linux system.
Hopefully fixes#126619
Hopefully fixes#126804
A proposed fix for the issue #95611, [OpenMP][SIMD] ordered has no
effect in a loop SIMD region as of LLVM 18.1.0
Changes:
- Implement new lowering behavior: Conservatively serialize "omp simd"
loops that have `omp simd ordered` directive to prevent incorrect
vectorization (which results in incorrect execution behavior of the
miscompiled program).
Implementation outline:
- We start with the optimistic default initial value of
`LoopStack.setParallel(/Enable=/true);` in
`CodeGenFunction::EmitOMPSimdInit(const OMPLoopDirective &D)`.
- We only disable the loop parallel memory access assumption with `if
(HasOrderedDirective) LoopStack.setParallel(/Enable=/false);` using the
`HasOrderedDirective` (which tests for the presence of an
`OMPOrderedDirective`).
- This results in no longer incorrectly vectorizing the loop when the
`omp simd ordered` directive is present.
Motivation: We'd like to prevent incorrect vectorization of the loops
marked with the `#pragma omp ordered simd` directive which has
previously resulted in miscompiled code.
At the same time, we'd like the usage outside of the `#pragma omp
ordered simd` context to remain unaffected: Note that in the test
"clang/test/OpenMP/ordered_codegen.cpp" we only "lose" the
`!llvm.access.group` metadata in `foo_simd` alone.
This is conservative, in that it's possible some of the loops would be
possible to vectorize, but we prefer to avoid miscompilation of the
loops that are currently illegal to vectorize.
A concrete example follows:
```cpp
// "test.c"
#include <float.h>
#include <math.h>
#include <omp.h>
#include <stdio.h>
#include <stdlib.h>
#include <time.h>
int compare_float(float x1, float x2, float scalar) {
const float diff = fabsf(x1 - x2);
x1 = fabsf(x1);
x2 = fabsf(x2);
const float l = (x2 > x1) ? x2 : x1;
if (diff <= l * scalar * FLT_EPSILON)
return 1;
else
return 0;
}
#define ARRAY_SIZE 256
__attribute__((noinline)) void initialization_loop(
float X[ARRAY_SIZE][ARRAY_SIZE], float Y[ARRAY_SIZE][ARRAY_SIZE]) {
const float max = 1000.0;
srand(time(NULL));
for (int r = 0; r < ARRAY_SIZE; r++) {
for (int c = 0; c < ARRAY_SIZE; c++) {
X[r][c] = ((float)rand() / (float)(RAND_MAX)) * max;
Y[r][c] = X[r][c];
}
}
}
__attribute__((noinline)) void omp_simd_loop(float X[ARRAY_SIZE][ARRAY_SIZE]) {
for (int r = 1; r < ARRAY_SIZE; ++r) {
for (int c = 1; c < ARRAY_SIZE; ++c) {
#pragma omp simd
for (int k = 2; k < ARRAY_SIZE; ++k) {
#pragma omp ordered simd
X[r][k] = X[r][k - 2] + sinf((float)(r / c));
}
}
}
}
__attribute__((noinline)) int comparison_loop(float X[ARRAY_SIZE][ARRAY_SIZE],
float Y[ARRAY_SIZE][ARRAY_SIZE]) {
int totalErrors_simd = 0;
const float scalar = 1.0;
for (int r = 1; r < ARRAY_SIZE; ++r) {
for (int c = 1; c < ARRAY_SIZE; ++c) {
for (int k = 2; k < ARRAY_SIZE; ++k) {
Y[r][k] = Y[r][k - 2] + sinf((float)(r / c));
}
}
// check row for simd update
for (int k = 0; k < ARRAY_SIZE; ++k) {
if (!compare_float(X[r][k], Y[r][k], scalar)) {
++totalErrors_simd;
}
}
}
return totalErrors_simd;
}
int main(void) {
float X[ARRAY_SIZE][ARRAY_SIZE];
float Y[ARRAY_SIZE][ARRAY_SIZE];
initialization_loop(X, Y);
omp_simd_loop(X);
const int totalErrors_simd = comparison_loop(X, Y);
if (totalErrors_simd) {
fprintf(stdout, "totalErrors_simd: %d \n", totalErrors_simd);
fprintf(stdout, "%s : %d - FAIL: error in ordered simd computation.\n",
__FILE__, __LINE__);
} else {
fprintf(stdout, "Success!\n");
}
return totalErrors_simd;
}
```
Before:
```
$ clang -fopenmp-simd -O3 -ffast-math -lm test.c -o test && ./test
totalErrors_simd: 15408
test.c : 76 - FAIL: error in ordered simd computation.
```
clang 19.1.0: https://godbolt.org/z/6EvhxqEhe
After:
```
$ clang -fopenmp-simd -O3 -ffast-math test.c -o test && ./test
Success!
```
Co-authored-by: Matt P. Dziubinski <matt-p.dziubinski@hpe.com>
This commit adds the new analyzer option
`assume-at-least-one-iteration`, which is `false` by default, but can be
set to `true` to ensure that the analyzer always assumes at least one
iteration in loops.
In some situations this "loop is skipped" execution path is an important
corner case that may evade the notice of the developer and hide
significant bugs -- however, there are also many situations where it's
guaranteed that at least one iteration will happen (e.g. some data
structure is always nonempty), but the analyzer cannot realize this and
will produce false positives when it assumes that the loop is skipped.
This commit refactors some logic around the implementation of the new
feature, but the only functional change is introducing the new analyzer
option. If the new option is left in its default state (false), then the
analysis is functionally equivalent to an analysis done with a version
before this commit.
- This reverts commit
0c6c4a9993.
- Add '-mcode-object-version=5' as to explicitly use code object
version 5 to match with 'FAIL' diagnostic.
- Add Requires directive to support lit test run on platforms
registered with x86_64 and amdgpu.
My last change made the test not run when the `spirv-tools` feature is
not available, which is always the case in CI for clang tests, but it
fails if `spirv-tools` is available for the following reasons:
1) We didn't build `spirv-link` as part of the internal `SPIRV-Tools`
build, which is required by the `clang` call in `clang-linker-wrapper`,
I already fixed that
[here](https://github.com/llvm/llvm-project/pull/126319).
2) We didn't depend on the `SPIRV-Tools` CMake targets in clang tests,
so depending on what CMake targets were built before running
`check-clang`, `SPIRV-Tools` might not have been built.
3) We didn't check for `llvm-spirv` being available, which is not part
of `SPIRV-Tools` but is currently required for SPIR-V compilation.
Manually confirmed this works. This test is the bane of my existence.
---------
Signed-off-by: Sarnie, Nick <nick.sarnie@intel.com>
Allow operator T&() in a member function which returns a const member
variable.
In particular, this will allow UniqueRef::operator T&() and
Ref::operator T&() to be treated as a safe pointer origin when they're
called on a const member.
This PR #125646 added this test and it was failing in Android's compiler
and on my machine locally. I removed the "Build config" check and it
passes now.
Don't return true here in InvalidNewDeleteExpr just because we are in
C++26 mode. This invalid there as well.
Testcase reduced from
libcxx/test/std/utilities/smartptr/unique.ptr/unique.ptr.create/make_unique_for_overwrite.pass.cpp
Summary:
The `clang-offload-packager` records the architecture of the job.
Currently there are cases where this will be empty. SYCL, CPU, and when
the user manually overrides it to be empty. In these cases we should
alwas consider it 'generic'. Adding this string both makes it clear how
it behaves and triggers the special handling for this architecture,
allowing it to bind to different architectures.
Example from the issue:
```c++
void Func(int x) {
switch (x) {
[[likely]] case 0:
case 1:
int i = 3;
case 2:
break;
}
}
```
Clang checks if ``case 2`` can be reachable by checking its scope. The
variable declaration should create a scope containing ``case 2``, but
due to the structure of the AST, ``case 2`` gets the scope of the
``likely`` statement, but not ``int i = 3;``. Therefore, I changed this
code to force attribute statement not to be scope parents.
Fixes#84072
Before commit 6e17ed9b04e5523cc910bf171c3122dcc64b86db the test file
`no-outofbounds.c` tested the behavior of the old alpha checker
`alpha.security.ArrayBound` (V1); then that commit converted it into a
test for the checker `security.ArrayBound` which was previously called
`alpha.security.ArrayBoundV2`.
This commit removes this small separate test file and adds some tests to
the "native" test files of `security.ArrayBound` to ensure that the same
functionality is still tested.
This is intended as the first commit in a series that reorganizes the
tests of `security.ArrayBound` to system that's easier to understand and
maintain.
Reapplying changes from https://github.com/llvm/llvm-project/pull/125638
after buildbot failures.
Buildbot failures fixed in 029e7e98dc9956086adc6c1dfb0c655a273fbee6,
latest commit on this PR. It was a problem with a declared class member
with same name as its type. Sorry!
Close https://github.com/llvm/llvm-project/issues/126373
Although the root problems should be we shouldn't place the friend
declaration to the incorrect module, let's avoid bleeding the edge by
stoping diagnosing declarations not in file scope.
Updates layout struct to be `struct` with public fields instead of
`class` with private fields and chang the name prefix to `__cblayout_`.
Also adds Packed attribute to prevent struct padding and filters
additional types from the layout that were missed previously (arrays of
resources and groupshared declarations).
Fixes https://github.com/llvm/llvm-project/issues/123459.
Previously, when the StackAddrEscapeChecker checked return values, it
did not scan into the structure of the return SVal. Now it does, and we
can catch some more false negatives that were already mocked out in the
tests in addition to those mentioned in
https://github.com/llvm/llvm-project/issues/123459.
The warning message at the moment for these newly caught leaks is not
great. I think they would be better if they had a better trace of why
and how the region leaks. If y'all are happy with these changes, I would
try to improve these warnings and work on normalizing this SVal checking
on the `checkEndFunction` side of the checker also.
Two of the stack address leak test cases now have two warnings, one
warning from return expression checking and another from`
checkEndFunction` `iterBindings` checking. For these two cases, I prefer
the warnings from the return expression checking, but I couldn't figure
out a way to drop the `checkEndFunction` without breaking other
`checkEndFunction` warnings that we do want. Thoughts here?
Summary:
Currently we conditionally enable NVPTX lowering depending on the
language (C/C++/OpenMP). Unfortunately this causes problems because this
option is only present if the backend was enabled, which causes this to
error if you try to make LLVM-IR.
This patch instead makes it the only accepted lowering. The reason we
had it as opt-in before is because it is not handled by CUDA. So, this
pach also introduces diagnostics to prevent *all* creation of
device-side global constructors and destructors. We already did this for
variables, now we do it for attributes as well.
This inverts the responsibility of blocking this from the backend to the
langauage like it should be given that support for this is language
dependent.
As Intel is working to add support for SPIR-V OpenMP device offloading
in upstream clang/liboffload, we need to modify the OpenMP frontend to
allow SPIR-V as well as generate valid IR for SPIR-V. For example, we
need the frontend to generate code to define and interact with device
globals used in the DeviceRTL.
This is the beginning of what I expect will be (many) other changes, but
let's get started with something simple.
---------
Signed-off-by: Sarnie, Nick <nick.sarnie@intel.com>
Fixes#103016
This is the last missing piece for the C++20 CTAD alias feature. No
release note being added in this PR yet, I will send out a follow-up
patch to mark this feature done.
(Since the release 20 branch is cut, I think we should target on
clang21).
Also move the -fno-strict-overflow option definition next to the
-fstrict-overflow one while here.
Also add test coverage for f(no-)wrapv-pointer being a clang-cl option.
For `new A[N]{1,2,3}`, we need to allocate N elements of type A, and
initialize the first three with the given InitListExpr elements.
However, if N is larger than 3, we need to initialize the remaining
elements with the InitListExpr array filler.
Similarly, for `new A[N];`, we need to initilize all fields with the
constructor of A. The initializer type is a CXXConstructExpr of
IncompleteArrayType in this case, which we can't generally handle.
When we launch hipcc with multiple offload architectures along with -MF
dep_file flag, the clang compilation invocations for host and device
offloads write to the same dep_file, and can lead to collision during
file IO operations. This can typically happen during large workloads.
This commit provides a fix to generate dep_file only in host
compilation.
---------
Co-authored-by: anikelal <anikelal@amd.com>
In my previous PR (#123656) to update the names of AVX10.2 intrinsics
and mnemonics, I have erroneously deleted `_ph` from few intrinsics.
This PR corrects this.
