Modify structure type in SPIR-V dialect to support:
1) Multiple decorations per structure member
2) Key-value based decorations (e.g., MatrixStride)
This commit kept the Offset decoration separate from members'
decorations container for easier implementation and logical clarity.
As such, all references to Structure layoutinfo are now offsetinfo,
and any member layout defining decoration (e.g., RowMajor for Matrix)
will be add to the members' decorations container along with its
value if any.
Differential Revision: https://reviews.llvm.org/D81426
Following the previous revision `D81100`, this commit implements a templated class
that would provide conversion patterns for “straightforward” SPIR-V ops into
LLVM dialect. Templating allows to abstract away from concrete implementation
for each specific op. Those are mainly binary operations. Currently supported
and tested ops are:
- Arithmetic ops: `IAdd`, `ISub`, `IMul`, `FAdd`, `FSub`, `FMul`, `FDiv`, `FNegate`,
`SDiv`, `SRem` and `UDiv`
- Bitwise ops: `BitwiseAnd`, `BitwiseOr`, `BitwiseXor`
The implementation relies on `SPIRVToLLVMConversion` class that makes use of
`OpConversionPattern`.
Differential Revision: https://reviews.llvm.org/D81305
This change introduces an LLD switch --thinlto-single-module to allow compiling only a part of the input modules. This is specifically enables:
1. Fast investigating/debugging modules of interest without spending time on compiling unrelated modules.
2. Compiler debug dump with -mllvm -debug-only= for specific modules.
It will be useful for large applications which has 1K+ input modules for thinLTO.
The switch can be combined with `--lto-obj-path=` or `--lto-emit-asm` to obtain intermediate object files or assembly files. So far the module name matching is implemented as a fuzzy name lookup where the modules with name containing the switch value are compiled.
E.g,
Command:
ld.lld main.o thin.a --thinlto-single-module=thin.a --lto-obj-path=single.o
log:
[ThinLTO] Selecting thin.a(thin1.o at 168) to compile
[ThinLTO] Selecting thin.a(thin2.o at 228) to compile
Command:
ld.lld main.o thin.a --thinlto-single-module=thin1.o --lto-obj-path=single.o
log:
[ThinLTO] Selecting thin.a(thin1.o at 168) to compile
Differential Revision: https://reviews.llvm.org/D80406
@nikic raised an issue on D75936 that the added complexity to the O0 pipeline was causing noticeable slowdowns for `-O0` builds. This patch addresses the issue by adding a pass with equal security properties, but without any optimizations (and more importantly, without the need for expensive analysis dependencies).
Reviewers: nikic, craig.topper, mattdr
Reviewed By: craig.topper, mattdr
Differential Revision: https://reviews.llvm.org/D80964
These tests are flaky on the reproducer bot. I suspect it has something
to do with the module cache. Skipping the whole category while I
investigate the issue.
This brings over functionality from the xcodeproject that went missing during the CMake transition.
rdar://problem/63840635
Differential Revision: https://reviews.llvm.org/D81300
This reverts commit 2e009dbcb3.
Reverting since there were some test failures on buildbots that used the
new pass manager. ASan and MSan are also finding some bugs in this that
I'll need to address.
This change got missed while upstreaming
https://reviews.llvm.org/D77772. This is the part of that change that
actually passes the correct arguments when opening a PCM.
The test didn't catch this because it starts at the
`MemoryBuffer::getOpenFile` level. It's not really possible to test
`ModuleManager::addModule` itself to verify how the file was opened.
This will allow simplifying executors by always just copying the whole
%T, and assuming that all file dependencies are contained in it.
Superseeds https://reviews.llvm.org/D78245, which tried to make %T unique
in Lit, but which encountered push back.
The tutorial refers to invoking toyc with '-mlir-print-debuginfo' but
it wasn't registered anymore.
Differential Revision: https://reviews.llvm.org/D81604
This patch contains all of the clang changes from D72959.
- Generalize the relative vtables ABI such that it can be used by other targets.
- Add an enum VTableComponentLayout which controls whether components in the
vtable should be pointers to other structs or relative offsets to those structs.
Other ABIs can change this enum to restructure how components in the vtable
are laid out/accessed.
- Add methods to ConstantInitBuilder for inserting relative offsets to a
specified position in the aggregate being constructed.
See D72959 for background info.
Differential Revision: https://reviews.llvm.org/D77592
This is a cheap instruction. It's better to repeat it than to do
two separate operations.
There are probably more cases like this, but this one was reported
as a regression in our internal benchmarking.
We relied on the fact that the iterators walks through the elements of a
DenseSet in a deterministic order (which is not true). This caused
ThinLTO cache misses. This patch addresses this problem.
See PR 45819 for additional information
https://bugs.llvm.org/show_bug.cgi?id=45819
Differential Revision: https://reviews.llvm.org/D79772
If we probe *after* each static stack allocation, we need to probe *before* each
dynamic stack allocation. Provide a scheme to describe the possible scenario.
Thanks a lot to @jonpa for motivating this fix.
Differential Revision: https://reviews.llvm.org/D81067
`noderef` was failing to trigger warnings in some cases related to c++ style
casting. This patch addresses them.
Differential Revision: https://reviews.llvm.org/D77836
The in, out, and sbi/cbi family of instructions seem to require a custom
decoder. I'm not exactly sure why and would prefer to convince TableGen
to provide the correct decoders for these, but I can't seem to convince
it to do so. They simply disassemble without any operands.
Differential Revision: https://reviews.llvm.org/D74049
There is an invalid cast produced when a pointee is a pointer
and the alloca type is cast to a pointer to int.
Differential Revision: https://reviews.llvm.org/D81606
This patch allows for usage of the @PLT modifier in AArch64 assembly which
lowers to an R_AARCH64_PLT32 relocation. See D81184 for handling this
relocation in lld.
Differential Revision: https://reviews.llvm.org/D81446
The expected strings would previously not catch bugs when redzones were
added when they were not actually expected. Fix by adding "global "
before the type.
Summary:
With this change, "ninja check-libc" on linux/aarch64 succeeds.
However, all entrypoints with machine dependent implementations
have been skipped. A good number of these skipped entrypoints can
be enabled once we have aarch64 syscall support available.
Reviewers: abrachet, asteinhauser
Differential Revision: https://reviews.llvm.org/D81533
The feature check should probably be enhanced for non-x86 architectures,
but this change shields them from x86 specific pieces until then.
This patch has been split out from https://reviews.llvm.org/D81533.
Summary:
- In HIP, just as the regular device-only compilation, the device-only
relocatable code compilation should not involve offload bundle.
- In addition, that device-only relocatable code compilation should have
the similar 3 steps, namely preprocessor, compile, and backend, to the
regular code generation with `-emit-llvm`.
Reviewers: yaxunl, tra
Subscribers: cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D81427
Summary:
New file include to support platform dependent grid constants. It will be
used by clang, libomptarget plugins, and deviceRTLs to access constant
values consistently and with fast access in the deviceRTLs.
Originally authored by Greg Rodgers (@gregrodgers).
Reviewers: arsenm, sameerds, jdoerfert, yaxunl, b-sumner, scchan, JonChesterfield
Reviewed By: arsenm
Subscribers: llvm-commits, pdhaliwal, jholewinski, jvesely, wdng, nhaehnle, guansong, kerbowa, sstefan1, cfe-commits, ronlieb, gregrodgers
Tags: #clang, #llvm
Differential Revision: https://reviews.llvm.org/D80917
Some platforms do not support aliases. Split the test, and pass explicit
triple to avoid test failure.
Reviewed By: thakis
Tags: #clang
Differential Revision: https://reviews.llvm.org/D81591
This was promoting booleans to i32 to perform a comparison against
them to feed to a select condition. Just use the booleans
directly. This produces the same final code, since the combiner is
unable to undo the mess this creates. I untangled this logic when I
ported this code to GlobalISel, so port the cleanups back.
MIRBuilder was in the middle of of a bunch of methods and not group
with the other member variables, which made it harder to see what
state this carries around. Move these to the top as is the usual
convention.
The complex expression helper class can be used to generate small,
short-lived instances of a "helper" that can be used to aid the
construction of complex expressions in FIR. The helper class bundles
together these functionally related operations.
Included in this diff is the header for the FIR builder. The
implementation has other dependences and will follow.
Differential Revision: https://reviews.llvm.org/D81287
Before 539b47c9 this test was not actually using the debug_names section
because the -gdwarf added by Makefile.rules on windows overrode the
-gdwarf-5 flag from CFLAGS_EXTRAS. Now that -gdwarf-5 is respected, the
test is failing.
If the target explicitly requested custom legalization, it should be
required to implement this. Also move default legalizeIntrinsic
implementation into the header so it's next to the related
legalizeCustom.
Craig Topper