Summary:
This patch introduces two things for offloading:
1. Asynchronous data transferring: those functions are suffix with `_async`. They have one more argument compared with their synchronous counterparts: `__tgt_async_info*`, which is a new struct that only has one field, `void *Identifier`. This struct is for information exchange between different asynchronous operations. It can be used for stream selection, like in this case, or operation synchronization, which is also used. We may expect more usages in the future.
2. Optimization of stream selection for data mapping. Previous implementation was using asynchronous device memory transfer but synchronizing after each memory transfer. Actually, if we say kernel A needs four memory copy to device and two memory copy back to host, then we can schedule these seven operations (four H2D, two D2H, and one kernel launch) into a same stream and just need synchronization after memory copy from device to host. In this way, we can save a huge overhead compared with synchronization after each operation.
Reviewers: jdoerfert, ye-luo
Reviewed By: jdoerfert
Subscribers: yaxunl, lildmh, guansong, openmp-commits
Tags: #openmp
Differential Revision: https://reviews.llvm.org/D77005
Summary:
Adds support for `ClangTidyCheck::OptionsView` to deteremine:
- If an option is found in the configuration.
- If an integer option read from configuration is parsable to an integer.
- Parse and Serialize enum configuration options directly using a mapping from `llvm::StringRef` to `EnumType`.
- If an integer or enum option isn't parseable but there is a default value it will issue a warning to stderr that the config value hasn't been used.
- If an enum option isn't parsable it can provide a hint if the value was a typo.
Reviewers: aaron.ballman, alexfh, gribozavr2
Reviewed By: aaron.ballman
Subscribers: xazax.hun, cfe-commits
Tags: #clang, #clang-tools-extra
Differential Revision: https://reviews.llvm.org/D77085
Update the sysroot expectation to match other targets and breakout
linux/musl toolchain tests into a new file.
Differential Revision: https://reviews.llvm.org/D77440
WG14 has adopted N2480 (http://www.open-std.org/jtc1/sc22/wg14/www/docs/n2480.pdf)
into C2x at the meetings last week, allowing parameter names of a function
definition to be elided. This patch relaxes the error so that C++ and C2x do not
diagnose this situation, and modes before C2x will allow it as an extension.
This also adds the same feature to ObjC blocks under the assumption that ObjC
wishes to follow the C standard in this regard.
This allows both the old and the new testing formats to handle these
tests with modules enabled.
We also include the modules flags in the %{flags} substitution, which
means that .sh.cpp tests in the old format and all tests in the new
format will use modules flags when enabled.
This patch extends existing constant folding in logical operations to
handle S_XNOR, S_NAND, S_NOR, S_ANDN2, S_ORN2, V_LSHL_ADD_U32 and
V_AND_OR_B32. Also added a couple of tests for existing folds.
This removes a call to getScalarType from a bunch of call sites.
It also makes the behavior consistent with SIGN_EXTEND_INREG.
Differential Revision: https://reviews.llvm.org/D77631
For implementing "remove obsolete debug info in lld", it is neccesary
to have DWARF generation code implementation. dsymutil uses DwarfStreamer
for that purpose. DwarfStreamer uses AsmPrinter. It is considered OK
to use AsmPrinter based code in lld(D74169). This patch moves
DwarfStreamer implementation into DWARFLinker, so that it could be reused
from lld.
Generally, a better place for such a common DWARF generation code would be
not DWARFLinker but an additional separate library. Such a library could
contain a single version of DWARF generation routines and could also
be independent of AsmPrinter. At the current moment, DwarfStreamer
does not pretend to be such a general implementation of DWARF generation.
So I decided to put it into DWARFLinker since it is the only user
of DwarfStreamer.
Testing: it passes "check-all" lit testing. MD5 checksum for clang .dSYM
bundle matches for the dsymutil with/without that patch.
Reviewed By: JDevlieghere
Differential revision: https://reviews.llvm.org/D77169
This reverts commit 21efb06f0a.
Changes since last attempt to land this patch:
- Sort files before deduplicating. This hopefully avoids some buildbot failures.
- Fix use of uninitialized variable when running without --use-analyzer.
- Remove the "REQUIRES: windows" item.
7aecf232 fixed the bug where we would miscompile, but we still generate
a crazy amount of code. Turn off the expansion until someone implements
an appropriate heuristic.
Differential Revision: https://reviews.llvm.org/D77599
We can only collapse adjacent SI_END_CF if outer statement
belongs to a simple SI_IF, otherwise correct mask is not in the
register we expect, but is an argument of an S_XOR instruction.
Even if SI_IF is simple it might be lowered using S_XOR because
lowering is dependent on a basic block layout. It is not
considered simple if instruction consuming its output is
not an SI_END_CF. Since that SI_END_CF might have already been
lowered to an S_OR isSimpleIf() check may return false.
This situation is an opportunity for a further optimization
of SI_IF lowering, but that is a separate optimization. In the
meanwhile move SI_END_CF post the lowering when we already know
how the rest of the CFG was lowered since a non-simple SI_IF
case still needs to be handled.
Differential Revision: https://reviews.llvm.org/D77610
This is an alternative design to D77512.
D45195 added --warn-backrefs to detect
* A. certain input orders which GNU ld either errors ("undefined reference")
or has different resolution semantics
* B. (byproduct) some latent multiple definition problems (-ldef1 -lref -ldef2) which I
call "linking sandwich problems". def2 may or may not be the same as def1.
When an archive appears more than once (-ldef -lref -ldef), lld and GNU
ld may have the same resolution but --warn-backrefs may warn. This is
not uncommon. For example, currently lld itself has such a problem:
```
liblldCommon.a liblldCOFF.a ... liblldCommon.a
_ZN3lld10DWARFCache13getDILineInfoEmm in liblldCOFF.a refers to liblldCommon.a(DWARF.cpp.o)
libLLVMSupport.a also appears twice and has a similar warning
```
glibc has such problems. It is somewhat destined because of its separate
libc/libpthread/... and arbitrary grouping. The situation is getting
improved over time but I have seen:
```
-lc __isnanl references -lm
-lc _IO_funlockfile references -lpthread
```
There are also various issues in interaction with other runtime
libraries such as libgcc_eh and libunwind:
```
-lc __gcc_personality_v0 references -lgcc_eh
-lpthread __gcc_personality_v0 references -lgcc_eh
-lpthread _Unwind_GetCFA references -lunwind
```
These problems are actually benign. We want --warn-backrefs to focus on
its main task A and defer task B (which is also useful) to a more
specific future feature (see gold --detect-odr-violations and
https://bugs.llvm.org/show_bug.cgi?id=43110).
Instead of warning immediately, we store the message and only report it
if no subsequent lazy definition exists.
The use of the static variable `backrefDiags` is similar to `undefs` in
Relocations.cpp
Reviewed By: grimar
Differential Revision: https://reviews.llvm.org/D77522
SymbolAssignment::addr stores the location counter. The type should be
uint64_t instead of unsigned. The upper half of the address space is
commonly used by operating system kernels.
Similarly, SymbolAssignment::size should be an uint64_t. A kernel linker
script can move the location counter from 0 to the upper half of the
address space.
Reviewed By: grimar
Differential Revision: https://reviews.llvm.org/D77445
If we have two back-to-back loops with block arguments, the OpPhi
instructions generated for the second loop's block arguments should
have use the merge block of the first SPIR-V loop structure as
their incoming parent block.
Differential Revision: https://reviews.llvm.org/D77543
These should not be assuming address space 0. Calling getPointerTy is
generally the wrong thing to do, since you should already know the
type from the incoming IR.
Summary:
This change adds DIFlagNonTrivial to forward declarations of
DICompositeType. It adds the flag to nontrivial types and types with
unknown triviality.
It fixes adding the "CxxReturnUdt" flag to functions inconsistently,
since it is added based on whether the return type is marked NonTrivial, and
that changes if the return type was a forward declaration.
continues the discussion at https://reviews.llvm.org/D75215
Bug: https://bugs.llvm.org/show_bug.cgi?id=44785
Reviewers: rnk, dblaikie, aprantl
Subscribers: cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D77436
RDA sometimes needs to visit blocks twice, to take into account
reaching defs coming in along loop back edges. Currently it handles
repeated visitation the same way as usual, which means that it will
scan through all instructions and their reg unit defs again. Not
only is this very inefficient, it also means that all reaching defs
in loops are going to be inserted twice.
We can do much better than this. The only thing we need to handle
is a new reaching def from a predecessor, which either needs to be
prepended to the reaching definitions (if there was no reaching def
from a predecessor), or needs to replace an existing predecessor
reaching def, if it is more recent. Since D77508 we only store the
most recent predecessor reaching def, so that's the only one that
may need updating.
This also has the nice side-effect that reaching definitions are
now automatically sorted and unique, so drop the llvm::sort() call
in favor of an assertion.
Differential Revision: https://reviews.llvm.org/D77511
An instruction may define the same reg unit multiple times,
avoid inserting the same reaching def multiple times in that case.
Also print the reg unit, rather than the super-register, in the
debug code.
Summary:
- Remove the no longer used Darwin CalleeSavedRegs
- Combine the SVR464 callee saved regs and AIX64 since the two are (and should be) identical into PPC64
- Update tests for 64-bit CSR change
Reviewers: sfertile, ZarkoCA, cebowleratibm, jasonliu, #powerpc
Reviewed By: sfertile
Subscribers: wuzish, nemanjai, hiraditya, kbarton, shchenz, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D77235
This is part of the Propeller framework to do post link code layout
optimizations. Please see the RFC here:
https://groups.google.com/forum/#!msg/llvm-dev/ef3mKzAdJ7U/1shV64BYBAAJ and the
detailed RFC doc here:
https://github.com/google/llvm-propeller/blob/plo-dev/Propeller_RFC.pdf
This patch adds lld support for basic block sections and performs relaxations
after the basic blocks have been reordered.
After the linker has reordered the basic block sections according to the
desired sequence, it runs a relaxation pass to optimize jump instructions.
Currently, the compiler emits the long form of all jump instructions. AMD64 ISA
supports variants of jump instructions with one byte offset or a four byte
offset. The compiler generates jump instructions with R_X86_64 32-bit PC
relative relocations. We would like to use a new relocation type for these jump
instructions as it makes it easy and accurate while relaxing these instructions.
The relaxation pass does two things:
First, it deletes all explicit fall-through direct jump instructions between
adjacent basic blocks. This is done by discarding the tail of the basic block
section.
Second, If there are consecutive jump instructions, it checks if the first
conditional jump can be inverted to convert the second into a fall through and
delete the second.
The jump instructions are relaxed by using jump instruction mods, something
like relocations. These are used to modify the opcode of the jump instruction.
Jump instruction mods contain three values, instruction offset, jump type and
size. While writing this jump instruction out to the final binary, the linker
uses the jump instruction mod to determine the opcode and the size of the
modified jump instruction. These mods are required because the input object
files are memory-mapped without write permissions and directly modifying the
object files requires copying these sections. Copying a large number of basic
block sections significantly bloats memory.
Differential Revision: https://reviews.llvm.org/D68065
Summary:
- Use `device_builtin_surface` and `device_builtin_texture` for
surface/texture reference support. So far, both the host and device
use the same reference type, which could be revised later when
interface/implementation is stablized.
Reviewers: yaxunl
Subscribers: cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D77583
Shuffle combining can insert zero byte sized elements into the shuffle mask, which combineX86ShufflesConstants will attempt to fold without taking into account whether the byte-sized type is legal (e.g. AVX512F only targets).
If we have a full-zeroable vector then we should just return a zero version of the root type, otherwise if the type isn't valid we should bail.
Fixes PR45443
The new format should be equivalent to the old format, and it is now the
default format when running the libc++ tests. This commit changes the
libc++abi tests to use the new format by default too. If unexpected failures
are discovered, it should be fine to revert this commit until they are
addressed.
Also note that it is still possible to use the old format by passing
`--param=use_old_format=True` when running Lit for the time being.
Summary:
Same restrictions apply as in the other direction: macro arguments are
not supported yet, only full macro expansions can be mapped.
Taking over from https://reviews.llvm.org/D72581.
Reviewers: gribozavr2, sammccall
Reviewed By: gribozavr2
Subscribers: cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D77209
Summary:
Our previous definition of "top-level" was too informal, and didn't
allow for overlapping macros that each directly produce expanded tokens.
See D77507 for previous discussion.
Fixes http://bugs.llvm.org/show_bug.cgi?id=45428
Reviewers: kadircet, vabridgers
Subscribers: cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D77615
Currently Clang does not respect -fno-unroll-loops during LTO. During
D76916 it was suggested to respect -fno-unroll-loops on a TU basis.
This patch uses the existing llvm.loop.unroll.disable metadata to
disable loop unrolling explicitly for each loop in the TU if
unrolling is disabled. This should ensure that loops from TUs compiled
with -fno-unroll-loops are skipped by the unroller during LTO.
This also means that if a loop from a TU with -fno-unroll-loops
gets inlined into a TU without this option, the loop won't be
unrolled.
Due to the fact that some transforms might drop loop metadata, there
potentially are cases in which we still unroll loops from TUs with
-fno-unroll-loops. I think we should fix those issues rather than
introducing a function attribute to disable loop unrolling during LTO.
Improving the metadata handling will benefit other use cases, like
various loop pragmas, too. And it is an improvement to clang completely
ignoring -fno-unroll-loops during LTO.
If that direction looks good, we can use a similar approach to also
respect -fno-vectorize during LTO, at least for LoopVectorize.
In the future, this might also allow us to remove the UnrollLoops option
LLVM's PassManagerBuilder.
Reviewers: Meinersbur, hfinkel, dexonsmith, tejohnson
Reviewed By: Meinersbur, tejohnson
Differential Revision: https://reviews.llvm.org/D77058
Summary:
The motivation here is fixing https://bugs.llvm.org/show_bug.cgi?id=45428, see
D77507. The fundamental problem is that a "top-level" expansion wasn't precisely
defined. Repairing this concept means that TokenBuffer's "top-level expansion"
may not correspond to a single macro expansion. Example:
```
M(2); // expands to 1+2
```
The expansions overlap, but neither expansion alone yields all the tokens.
We need a TokenBuffer::Mapping that corresponds to their union.
This is fairly easy to fix in CollectPPExpansions, but the current design of
TokenCollector::Builder needs a fix too as it relies on the macro's expansion
range rather than the captured expansion bounds. This fix is hard to make due
to the way code is reused within Builder. And honestly, I found that code pretty
hard to reason about too.
The new approach doesn't use the expansion range, but only the expansion
location: it assumes an expansion is the contiguous set of expanded tokens with
the same expansion location, which seems like a reasonable formalization of
the "top-level" notion.
And hopefully the control flow is easier to follow too, it's considerably
shorter even with more documentation.
Reviewers: kadircet
Subscribers: cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D77614
Otherwise, files don't link when using a GNU linker, which is more
sensitive on the order of the source file relative to the various
linked libraries. See http://c-faq.com/lib/libsearch.html for an
explanation of the problem.
Summary:
FileInputs are only written by ASTWorker thread, therefore it is safe
to read them without the lock inside that thread. It can still be read by other
threads through ASTWorker::getCurrentCompileCommand though.
This patch also gets rid of the smart pointer wrapping FileInputs as there is
never mutliple owners.
Reviewers: sammccall
Subscribers: ilya-biryukov, javed.absar, MaskRay, jkorous, arphaman, usaxena95, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D77309
Shilei Tian