This changes the behavior of constructing MLIRContext to no longer load globally
registered dialects on construction. Instead Dialects are only loaded explicitly
on demand:
- the Parser is lazily loading Dialects in the context as it encounters them
during parsing. This is the only purpose for registering dialects and not load
them in the context.
- Passes are expected to declare the dialects they will create entity from
(Operations, Attributes, or Types), and the PassManager is loading Dialects into
the Context when starting a pipeline.
This changes simplifies the configuration of the registration: a compiler only
need to load the dialect for the IR it will emit, and the optimizer is
self-contained and load the required Dialects. For example in the Toy tutorial,
the compiler only needs to load the Toy dialect in the Context, all the others
(linalg, affine, std, LLVM, ...) are automatically loaded depending on the
optimization pipeline enabled.
To adjust to this change, stop using the existing dialect registration: the
global registry will be removed soon.
1) For passes, you need to override the method:
virtual void getDependentDialects(DialectRegistry ®istry) const {}
and registery on the provided registry any dialect that this pass can produce.
Passes defined in TableGen can provide this list in the dependentDialects list
field.
2) For dialects, on construction you can register dependent dialects using the
provided MLIRContext: `context.getOrLoadDialect<DialectName>()`
This is useful if a dialect may canonicalize or have interfaces involving
another dialect.
3) For loading IR, dialect that can be in the input file must be explicitly
registered with the context. `MlirOptMain()` is taking an explicit registry for
this purpose. See how the standalone-opt.cpp example is setup:
mlir::DialectRegistry registry;
registry.insert<mlir::standalone::StandaloneDialect>();
registry.insert<mlir::StandardOpsDialect>();
Only operations from these two dialects can be in the input file. To include all
of the dialects in MLIR Core, you can populate the registry this way:
mlir::registerAllDialects(registry);
4) For `mlir-translate` callback, as well as frontend, Dialects can be loaded in
the context before emitting the IR: context.getOrLoadDialect<ToyDialect>()
Differential Revision: https://reviews.llvm.org/D85622
StackLifetime class collects lifetime marker of an `alloca` by collect
the user of `BitCast` who is the user of the `alloca`. However, either
the `alloca` itself could be used with the lifetime marker or the `BitCast`
of the `alloca` could be transformed to other instructions. (e.g.,
it may be transformed to all zero reps in `InstCombine` pass).
This patch tries to fix this process in `collectMarkers` functions.
Reviewed By: vitalybuka
Differential Revision: https://reviews.llvm.org/D85399
This greatly simplifies a large portion of the underlying infrastructure, allows for lookups of singleton classes to be much more efficient and always thread-safe(no locking). As a result of this, the dialect symbol registry has been removed as it is no longer necessary.
For users broken by this change, an alert was sent out(https://llvm.discourse.group/t/removing-kinds-from-attributes-and-types) that helps prevent a majority of the breakage surface area. All that should be necessary, if the advice in that alert was followed, is removing the kind passed to the ::get methods.
Differential Revision: https://reviews.llvm.org/D86121
Summary:
Caught by HWASAN on arm64 Android (which uses ld128 for long double). This
was running the existing fuzzer.
The specific minimized fuzz input to reproduce this is:
__cxa_demangle("1\006ILeeeEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEE", 0, 0, 0);
Reviewers: eugenis, srhines, #libc_abi!
Subscribers: kristof.beyls, danielkiss, libcxx-commits
Tags: #libc_abi
Differential Revision: https://reviews.llvm.org/D77924
Building on the backend support from D85165. This parses the command line option in the driver, passes it on to CC1 and adds a function attribute.
-Still need to support tune on the target attribute.
-Need to use "generic" as the tuning by default. But need to change generic in the backend first.
-Need to set tune if march is specified and mtune isn't.
-May need to disable getHostCPUName's ability to guess CPU name from features when it doesn't have a family/model match for mtune=native. That's what gcc appears to do.
Differential Revision: https://reviews.llvm.org/D85384
Now that we no longer require for this map to have stable iteration order,
we no longer need to pay for keeping the iteration order stable,
so switch from `SmallMapVector` to `SmallDenseMap`.
Summary:
Make exactly single NodeBuilder exists at any given time
Reviewers: NoQ, Szelethus, vsavchenko, xazax.hun
Reviewed By: NoQ
Subscribers: martong, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D85796
While it may seem like we can just "deduplicate" the case where
some basic block happens to be a predecessor more than once,
which happens for e.g. switches, that is not correct thing to do.
We must actually add a PHI operand for each predecessor.
This was initially reported to me by David Major
as a clang crash during gecko build for android.
Although it works fine with glibc, as currently implemented the
frameheader cache is incompatible with certain platforms with
slightly different locking semantics inside dl_iterate_phdr.
Therefore only enable it when it is turned on explicitly with
a configure-time option.
Differential Revision: https://reviews.llvm.org/D86163
These instructions weren't in the initial version of MMX, but
were added when SSE1 was introduced. We already have the intrinsic
named correctly to include sse and the frontened header enforces
sse. We have one place in the backend where we DAG combine to
this intrinsic, but that's also qualified. So don't know of anything
currently broken unless someone writes their own IR and doesn't
set the sse feature.
This changes the behavior of constructing MLIRContext to no longer load globally
registered dialects on construction. Instead Dialects are only loaded explicitly
on demand:
- the Parser is lazily loading Dialects in the context as it encounters them
during parsing. This is the only purpose for registering dialects and not load
them in the context.
- Passes are expected to declare the dialects they will create entity from
(Operations, Attributes, or Types), and the PassManager is loading Dialects into
the Context when starting a pipeline.
This changes simplifies the configuration of the registration: a compiler only
need to load the dialect for the IR it will emit, and the optimizer is
self-contained and load the required Dialects. For example in the Toy tutorial,
the compiler only needs to load the Toy dialect in the Context, all the others
(linalg, affine, std, LLVM, ...) are automatically loaded depending on the
optimization pipeline enabled.
To adjust to this change, stop using the existing dialect registration: the
global registry will be removed soon.
1) For passes, you need to override the method:
virtual void getDependentDialects(DialectRegistry ®istry) const {}
and registery on the provided registry any dialect that this pass can produce.
Passes defined in TableGen can provide this list in the dependentDialects list
field.
2) For dialects, on construction you can register dependent dialects using the
provided MLIRContext: `context.getOrLoadDialect<DialectName>()`
This is useful if a dialect may canonicalize or have interfaces involving
another dialect.
3) For loading IR, dialect that can be in the input file must be explicitly
registered with the context. `MlirOptMain()` is taking an explicit registry for
this purpose. See how the standalone-opt.cpp example is setup:
mlir::DialectRegistry registry;
mlir::registerDialect<mlir::standalone::StandaloneDialect>();
mlir::registerDialect<mlir::StandardOpsDialect>();
Only operations from these two dialects can be in the input file. To include all
of the dialects in MLIR Core, you can populate the registry this way:
mlir::registerAllDialects(registry);
4) For `mlir-translate` callback, as well as frontend, Dialects can be loaded in
the context before emitting the IR: context.getOrLoadDialect<ToyDialect>()
LinalgDistribution options to allow more general distributions.
Changing the signature of the callback to send in the ranges for all
the parallel loops and expect a vector with the Value to use for the
processor-id and number-of-processors for each of the parallel loops.
Differential Revision: https://reviews.llvm.org/D86095
Checking if an object file is in memory should use the ObjectFile::IsInMemory(), not test ObjectFile::BaseAddress(). ObjectFile::BaseAddress() is designed to be overridden by all classes and is for mach-o, ELF and COFF plug-ins. They find the header base adddress and return that as a section offset address. The default implementation of ObjectFile::BaseAddress() does try and make an Address() from the ObjectFile::m_memory_addr, but I switched it to a correct function call.
Differential Revision: https://reviews.llvm.org/D86122
We probably want to introduce pseudo-instructions at some point, like
we have for binary operations, but this seems okay for now.
One thing I'm not sure about is whether we should be doing this as a
DAGCombine instead of directly pattern-matching it. I don't see any big
downside to doing it this way, though.
Differential Revision: https://reviews.llvm.org/D85681
This isn't necessaary for ACLE, but could be useful in other situations.
And the change is simple.
Differential Revision: https://reviews.llvm.org/D85251
SemaBuiltinConstantArg has an early exit for that case that doesn't
produce an error and doesn't update the APInt. We need to detect that
case and not use the APInt value.
While there delete the signature of CheckX86BuiltinTileArgumentsRange
that takes a single Argument index to check. There's another version
that takes an ArrayRef and single value is convertible to an ArrayRef.
It's annoying to have to maintain multiple, nearly identical chains of if
statements which all set the same attributes.
Add a helper function, `addFlagsUsingAttrFn` which performs the attribute
setting.
Then, use wrappers for that function in `lowerCall` and `setArgFlags`.
(Note that the flag-setting code in `setArgFlags` was missing the returned
attribute. There's no selection for this yet, so no test. It's an example of
the kind of thing this lets us avoid, though.)
Differential Revision: https://reviews.llvm.org/D86159
Similar to this commit:
faf8065a99
Testcase is pretty much the same as
test/CodeGen/AArch64/tailcall-explicit-sret.ll
Except it uses i64 (since we don't handle the i1024 return values yet), and
doesn't have indirect tail call testcases (because we can't translate those
yet).
Differential Revision: https://reviews.llvm.org/D86148
Parsing DWARFv5 debug_loclist offsets when a CU is parsed is weighing
down memory usage of symbolizers that don't need to parse this data at
all. There's not much benefit to caching these anyway - since they are
O(1) lookup and reading once you know where the offset list starts (and
can do bounds checking with the offset list size too).
In general, I think it might be time to start paying down some of the
technical debt of loc/loclist/range/rnglist parsing to try to unify it a
bit more.
eg:
* Currently DWARFUnit has: RangeSection, RangeSectionBase, LocSection,
LocSectionBase, LocTable, RngListTable, LoclistTableHeader (be nice if
these were all wrapped up in two variables - one for loclists, one for
rnglists)
* rnglists and loclists are handled differently (see:
LoclistTableHeader, but no RnglistTableHeader)
* maybe all these types could be less stateful - lazily parse what they
need to, even reparsing rather than caching because it doesn't seem
too expensive, for instance. (though admittedly so long as it's
constantcost/overead per compilatiton that's probably adequate)
* Maybe implementing and using a DWARFDataExtractor that can be
sub-ranged (so we could slice it up to just the single contribution) -
though maybe that's not so useful because loc/ranges need to refer to
it by absolute, not contribution-relative mechanisms
Differential Revision: https://reviews.llvm.org/D86110
When a procedure name was used on the RHS of an assignment we were not
reporting the error. When one was used in an expression the error
message wasn't very good (e.g. "Operands of + must be numeric; have
INTEGER(4) and untyped").
Detect these cases in ArgumentAnalyzer and emit better messages,
depending on whether the named procedure is a function or subroutine.
Procedure names may appear as actual arguments to function and
subroutine calls so don't report errors in those cases. That is the same
case where assumed type arguments are allowed, so rename `isAssumedType_`
to `isProcedureCall_` and use that to decide if it is an error.
Differential Revision: https://reviews.llvm.org/D86107
This is restricted to single use loads, which if we fold to sextloads we can
find more optimal addressing modes on AArch64.
This also fixes an overload the MachineFunction::getMachineMemOperand() method
which was incorrectly using the MF alignment instead of the MMO alignment.
Differential Revision: https://reviews.llvm.org/D85966
By detecting this sign extend pattern early, we can uncover opportunities for
more optimizations.
Differential Revision: https://reviews.llvm.org/D85965
There should be an equivalent std.floor op to std.ceil. This includes
matching lowerings for SPIRV, NVVM, ROCDL, and LLVM.
Reviewed By: ftynse
Differential Revision: https://reviews.llvm.org/D85940
Create a reduction pass that accepts an optimization pass as argument
and only replaces the golden module in the pipeline if the output of the
optimization pass is smaller than the input and still exhibits the
interesting behavior.
Add a -test-pass option to test individual passes in the MLIR Reduce
tool.
Reviewed By: jpienaar
Differential Revision: https://reviews.llvm.org/D84783
We're (temporarily) disabling ExtInt for the '__atomic' builtins so we can better design their behavior later. The idea is until we do an audit/design for the way atomic builtins are supposed to work with _ExtInt, we should leave them restricted so they don't limit our future options, such as by binding us to a sub-optimal implementation via ABI.
Example after this change:
$ cat test.c
void f(_ExtInt(64) *ptr) {
__atomic_fetch_add(ptr, 1, 0);
}
$ clang -c test.c
test.c:2:22: error: argument to atomic builtin of type '_ExtInt' is not supported
__atomic_fetch_add(ptr, 1, 0);
^
1 error generated.
Differential Revision: https://reviews.llvm.org/D84049
VLD2/4 instructions cannot be predicated, so we cannot tail predicate
them from autovec. From intrinsics though, they should be valid as they
will just end up loading extra values into off vector lanes, not
effecting the on lanes. The same is true for loads in general where so
long as we are not using the other vector lanes, an unpredicated load
can be converted to a predicated one.
This marks VLD2 and VLD4 instructions as validForTailPredication and
allows any unpredicated load in tail predication loop, which seems to be
valid given the other checks we have.
Differential Revision: https://reviews.llvm.org/D86022
D85968 started to use `split-file` and while buildbots run fine while
doing `make check-lldb` by hand I get:
.../llvm-monorepo-clangassert/tools/lldb/test/SymbolFile/DWARF/Output/DW_AT_declaration-with-children.s.script: line 2: split-file: command not found
failed:
lldb-shell :: SymbolFile/DWARF/DW_AT_declaration-with-children.s
Differential Revision: https://reviews.llvm.org/D86144
There are some cases where the instruction that sets up the iteration
count for a tail predicated loop cannot be moved before the dlstp,
stopping tail predication entirely. This patch checks if the mov operand
can be used and if so, uses that instead.
Differential Revision: https://reviews.llvm.org/D86087
This patch adds more op/type conversion support
necessary for `spirv-runner`:
- EntryPoint/ExecutionMode: currently removed since we assume
having only one kernel function in the kernel module.
- StorageBuffer storage class is now supported. We are not
concerned with multithreading so this is fine for now.
- Type conversion enhanced, now regular offsets and strides
for structs and arrays are supported (based on
`VulkanLayoutUtils`).
- Support of `spc.AccessChain` that is modelled with GEP op
in LLVM dialect.
Reviewed By: mravishankar
Differential Revision: https://reviews.llvm.org/D86109
The CrossOver mutator is meant to cross over two given buffers (referred to as
the first/second buffer henceforth). Previously InsertPartOf/CopyPartOf calls
used in the CrossOver mutator incorrectly inserted/copied part of the second
buffer into a "scratch buffer" (MutateInPlaceHere of the size
CurrentMaxMutationLen), rather than the first buffer. This is not intended
behavior, because the scratch buffer does not always (i) contain the content of
the first buffer, and (ii) have the same size as the first buffer;
CurrentMaxMutationLen is typically a lot larger than the size of the first
buffer. This patch fixes the issue by using the first buffer instead of the
scratch buffer in InsertPartOf/CopyPartOf calls.
A FuzzBench experiment was run to make sure that this change does not
inadvertently degrade the performance. The performance is largely the same; more
details can be found at:
https://storage.googleapis.com/fuzzer-test-suite-public/fixcrossover-report/index.html
This patch also adds two new tests, namely "cross_over_insert" and
"cross_over_copy", which specifically target InsertPartOf and CopyPartOf,
respectively.
- cross_over_insert.test checks if the fuzzer can use InsertPartOf to trigger
the crash.
- cross_over_copy.test checks if the fuzzer can use CopyPartOf to trigger the
crash.
These newly added tests were designed to pass with the current patch, but not
without the it (with 790878f291 these tests do not
pass). To achieve this, -max_len was intentionally given a high value. Without
this patch, InsertPartOf/CopyPartOf will generate larger inputs, possibly with
unpredictable data in it, thereby failing to trigger the crash.
The test pass condition for these new tests is narrowed down by (i) limiting
mutation depth to 1 (i.e., a single CrossOver mutation should be able to trigger
the crash) and (ii) checking whether the mutation sequence of "CrossOver-" leads
to the crash.
Also note that these newly added tests and an existing test (cross_over.test)
all use "-reduce_inputs=0" flags to prevent reducing inputs; it's easier to
force the fuzzer to keep original input string this way than tweaking
cov-instrumented basic blocks in the source code of the fuzzer executable.
Differential Revision: https://reviews.llvm.org/D85554
There is an untested but useful case: `this` (even if not written) is counted as a
source variable.
Reviewed By: dblaikie
Differential Revision: https://reviews.llvm.org/D86044
This is a non-functional-change to generalize the printIR routines so that
the output can be saved and manipulated rather than being directly output
to dbgs(). This is a prerequisite change for many upcoming changes that
allow new ways of examining changes made to the IR in the new pass manager.
Reviewed By: aeubanks (Arthur Eubanks)
Differential Revision: https://reviews.llvm.org/D85999
* GNU ld places non-SHF_ALLOC sections after SHF_ALLOC sections. This has the
advantage that the file offsets of a non-SHF_ALLOC cannot be contained in
a PT_LOAD. This patch matches the behavior.
* For non-SHF_ALLOC non-orphan sections, GNU ld may assign non-zero sh_addr and
treat them similar to SHT_NOBITS (not advance location counter). This
is an alternative approach to what we have done in D85100.
By placing non-SHF_ALLOC sections at the end, we can drop special
cases in createSection and findOrphanPos added by D85100.
Different from GNU ld, we set sh_addr to 0 for non-SHF_ALLOC sections. 0
arguably is better because non-SHF_ALLOC sections don't appear in the memory
image.
ELF spec says:
> sh_addr - If the section will appear in the memory image of a process, this
> member gives the address at which the section's first byte should
> reside. Otherwise, the member contains 0.
D85100 appeared to take a detour. If we take a combined view on D85100 and this
patch, the overall complexity slightly increases (one more 3-line loop) and
compatibility with GNU ld improves.
The behavior we don't want to match is the special treatment of .symtab
.shstrtab .strtab: they can be matched in LLD but not in GNU ld.
Reviewed By: jhenderson, psmith
Differential Revision: https://reviews.llvm.org/D85867
Mehdi Amini