Summary:
When converting splat constants for nested sequential LLVM IR types wrapped in
MLIR, the constant conversion was erroneously assuming it was always possible
to recursively construct a constant of a sequential type given only one value.
Instead, wait until all sequential types are unpacked recursively before
constructing a scalar constant and wrapping it into the surrounding sequential
type.

Subscribers: mehdi_amini, rriddle, jpienaar, burmako, shauheen, antiagainst, nicolasvasilache, arpith-jacob, mgester, lucyrfox, aartbik, liufengdb, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D72688

Summary:
Previously, since these aggregates are > 2*XLen, Clang would think they
were being returned indirectly and thus would decrease the number of
available GPRs available by 1. For long argument lists this could lead
to a struct argument incorrectly being passed indirectly.

Reviewers: asb, lenary

Reviewed By: asb, lenary

Subscribers: luismarques, rbar, johnrusso, simoncook, apazos, sabuasal, niosHD, kito-cheng, shiva0217, zzheng, edward-jones, rogfer01, MartinMosbeck, brucehoult, the_o, rkruppe, PkmX, jocewei, psnobl, benna, Jim, lenary, s.egerton, pzheng, sameer.abuasal, cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D69590

Summary:
This is based on the use of code constantly checking for an attribute on
a model and instead represents the distinct operaion with a different
op. Instead, this op can be used to provide better filtering.

Reviewers: herhut, mravishankar, antiagainst, rriddle

Reviewed By: herhut, antiagainst, rriddle

Subscribers: liufengdb, aartbik, jholewinski, mgorny, mehdi_amini, rriddle, jpienaar, burmako, shauheen, antiagainst, nicolasvasilache, csigg, arpith-jacob, mgester, lucyrfox, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D72336

Makes this function exit early instead of nesting if statements.

Also removed all the if (tag_type->getDecl()) checks. If we created
a TagType with a nullptr as a Decl then Clang would have already
deferenced that nullptr during TagType creation so there is no point
in gracefully handling a nullptr here.

Summary:
I forgot to ran git-clang-format before committing.

Summary:
These operations can be used to specify a loop nest with a body that can
contain reductions. The iteration space can be iterated in any order.

RFC: https://groups.google.com/a/tensorflow.org/d/topic/mlir/pwtSgiKFPis/discussion

Differential Revision: https://reviews.llvm.org/D72394

Enables the masked gather pass to create a masked
gather loading from a base and vector of offsets.
This also enables v8i16 and v16i8 gather loads.

Differential Revision: https://reviews.llvm.org/D72330

Summary:
This allows you to make some of the defs in a multiclass or `foreach`
conditional on an expression computed from the parameters or iteration
variables.

It was already possible to simulate an if statement using a `foreach`
with a dummy iteration variable and a list constructed using `!if` so
that it had length 0 or 1 depending on the condition, e.g.

  foreach unusedIterationVar = !if(condition, [1], []<int>) in { ... }

But this syntax is nicer to read, and also more convenient because it
allows an else clause.

To avoid upheaval in the implementation, I've implemented `if` as pure
syntactic sugar on the `foreach` implementation: internally, `ParseIf`
actually does construct exactly the kind of foreach shown above (and
another reversed one for the else clause if present).

Reviewers: nhaehnle, hfinkel

Reviewed By: hfinkel

Subscribers: hiraditya, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D71474

Summary:
This allows you to define a global or local variable to an arbitrary
value, and refer to it in subsequent definitions.

The main use I anticipate for this is if you have to compute some
difficult function of the parameters of a multiclass, and then use it
many times. For example:

  multiclass Foo<int i, string s> {
    defvar op = !cast<BaseClass>("whatnot_" # s # "_" # i);
    def myRecord {
      dag a = (op this, (op that, the other), (op x, y, z));
      int b = op.subfield;
    }
    def myOtherRecord<"template params including", op>;
  }

There are a couple of ways to do this already, but they're not really
satisfactory. You can replace `defvar x = y` with a loop over a
singleton list, `foreach x = [y] in { ... }` - but that's unintuitive
to someone who hasn't seen that workaround idiom before, and requires
an extra pair of braces that you often didn't really want. Or you can
define a nested pair of multiclasses, with the inner one taking `x` as
a template parameter, and the outer one instantiating it just once
with the desired value of `x` computed from its other parameters - but
that makes it awkward to sequentially compute each value based on the
previous ones. I think `defvar` makes things considerably easier.

You can also use `defvar` at the top level, where it inserts globals
into the same map used by `defset`. That allows you to define global
constants without having to make a dummy record for them to live in:

  defvar MAX_BUFSIZE = 512;

  // previously:
  // def Dummy { int MAX_BUFSIZE = 512; }
  // and then refer to Dummy.MAX_BUFSIZE everywhere

Reviewers: nhaehnle, hfinkel

Reviewed By: hfinkel

Subscribers: hiraditya, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D71407

Summary:
Initialising a pointer from nullptr involves an implicit cast.
Ignore it after getting initialiser from InitListExpr.

Fixes: PR44440

Reviewers: aaron.ballman, alexfh, JonasToth

Reviewed By: JonasToth

Subscribers: xazax.hun, cfe-commits

Tags: #clang

Differential Revision: https://reviews.llvm.org/D72630

This addresses post commit review comments for D71766.

This change allows to model the height of the instruction
within a bundle for latency adjustment purposes.

Differential Revision: https://reviews.llvm.org/D72669

A bug was filed that these warnings should not be emitted as DevCom-883961. ( https://developercommunity.visualstudio.com/content/problem/883961/c4389-signedunsigned-mismatch-should-not-be-emitte.html )

Summary:
This renames the test `rdar-12481949` to `get-value-32bit-int` as it just tests that we return the
correct result get calling GetValueAsSigned/GetValueAsUnsigned on 32-bit integers.

It also deletes all the strange things going on in this test including resetting the data formatters (which are to my
knowledge not used to calculate scalar values) and testing Python's long integers (let's just assume that our Python
distribution works correctly). Also modernises the setup code.

Reviewers: labath, aprantl

Reviewed By: aprantl

Subscribers: JDevlieghere, lldb-commits

Tags: #lldb

Differential Revision: https://reviews.llvm.org/D72593

We do not have InstrItinerary so generic getInstLatency() was always
defaulting to return 1 cycle. We need to use TargetSchedModel instead
to compute an instruction's latency.

Differential Revision: https://reviews.llvm.org/D72655

Summary:
Whenever we cast an LLVM instruction to one of its subclasses, we do a double check if the RTTI
enum value actually allows us to cast the class. I don't see a way this can ever happen as even when
LLVM's RTTI system has some corrupt internal state (which we probably should not test in the first
place) we just reuse LLVM RTTI to do the second check.

This also means that if we ever make an actual programming error in this function (e.g., have a enum
value and then cast it to a different subclass), we just silently fall back to the JIT in our tests.

We also can't test this code in any reasonable way.

This removes the checks and uses `llvm::cast` instead which will raise a fatal error when casting fails.

Reviewers: labath, mib

Reviewed By: labath

Subscribers: abidh, JDevlieghere, lldb-commits

Tags: #lldb

Differential Revision: https://reviews.llvm.org/D72596

[lldb] Fix that SBThread.GetStopDescription is returning strings with uninitialized memory at the end.

Summary:
`SBThread.GetStopDescription` is a curious API as it takes a buffer length as a parameter that specifies
how many bytes the buffer we pass has. Then we fill the buffer until the specified length (or the length
of the stop description string) and return the string length. If the buffer is a nullptr however, we instead
return how many bytes we would have written to the buffer so that the user can allocate a buffer with
the right size and pass that size to a subsequent `SBThread.GetStopDescription` call.

Funnily enough, it is not possible to pass a nullptr via the Python SWIG bindings, so that might be the
first API in LLDB that is not only hard to use correctly but impossible to use correctly. The only way to
call this function via Python is to throw in a large size limit that is hopefully large enough to contain the
stop description (otherwise we only get the truncated stop description).

Currently passing a size limit that is smaller than the returned stop description doesn't cause the
Python bindings to return the stop description but instead the truncated stop description + uninitialized characters
at the end of the string. The reason for this is that we return the result of `snprintf` from the method
which returns the amount of bytes that *would* have been written (which is larger than the buffer).
This causes our Python bindings to return a string that is as large as full stop description but the
buffer that has been filled is only as large as the passed in buffer size.

This patch fixes this issue by just recalculating the string length in our buffer instead of relying on the wrong
return value. We also have to do this in a new type map as the old type map is also used for all methods
with the given argument pair `char *dst, size_t dst_len` (e.g. SBProcess.GetSTDOUT`). These methods have
different semantics for these arguments and don't null-terminate the returned buffer (they instead return the
size in bytes) so we can't change the existing typemap without breaking them.

Reviewers: labath, jingham

Reviewed By: labath

Subscribers: clayborg, shafik, abidh, JDevlieghere, lldb-commits

Tags: #lldb

Differential Revision: https://reviews.llvm.org/D72086

Summary: Depends on D72611.

Reviewers: rampitec, arsenm

Subscribers: kzhuravl, jvesely, wdng, nhaehnle, yaxunl, dstuttard, tpr, t-tye, jfb, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D72616

Summary:
This adds checks that the expected error was actually reported against
the correct instruction, and fixes a couple of problems that that showed
up: one incorrect W32-ERR:

 v_cmp_class_f16_sdwa vcc, v1, v2 src0_sel:DWORD src1_sel:DWORD
 // W64: encoding: [0xf9,0x04,0x1e,0x7d,0x01,0x00,0x06,0x06]
-// W32-ERR: error: invalid operand for instruction
+// W32-ERR: error: {{instruction not supported on this GPU|invalid operand for instruction}}

and one missing W32-ERR:

 v_cmp_class_f16_sdwa s[6:7], v1, v2 src0_sel:DWORD src1_sel:DWORD
 // W64: encoding: [0xf9,0x04,0x1e,0x7d,0x01,0x86,0x06,0x06]
+// W32-ERR: error: invalid operand for instruction

Reviewers: rampitec, arsenm

Subscribers: kzhuravl, jvesely, wdng, nhaehnle, yaxunl, dstuttard, tpr, t-tye, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D72611

Summary:
This adds assembler tests for cases that were previously only in the
disassembler tests, and vice versa.

Reviewers: rampitec, arsenm, nhaehnle

Subscribers: kzhuravl, jvesely, wdng, yaxunl, dstuttard, tpr, t-tye, jfb, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D72592

The OpenMP runtime is not instrumented, so entering the runtime leaves no hint
on the source line of the pragma on ThreadSanitizer's function stack.

This patch adds function entry/exit annotations for OpenMP parallel regions,
and synchronization regions (barrier, taskwait, taskgroup).

Reviewed By: jdoerfert

Differential Revision: https://reviews.llvm.org/D70408

If the openmp project is built standalone, the test compiler is feature tested for an available -fsanitize=thread flag.
If the openmp project is built as part of llvm, the target tsan is needed to test archer.

An additional line (requires tsan) was introduced to the tests, this patch updates the line numbers for the race.

Follow-up for 77ad98c8

Reviewed By: jdoerfert

Differential Revision: https://reviews.llvm.org/D71914

This patch is a joint work by Rui Ueyama and me based on D58102 by Xiang Zhang.

It adds Intel CET (Control-flow Enforcement Technology) support to lld.
The implementation follows the draft version of psABI which you can
download from https://github.com/hjl-tools/x86-psABI/wiki/X86-psABI.

CET introduces a new restriction on indirect jump instructions so that
you can limit the places to which you can jump to using indirect jumps.

In order to use the feature, you need to compile source files with
-fcf-protection=full.

* IBT is enabled if all input files are compiled with the flag. To force enabling ibt, pass -z force-ibt.
* SHSTK is enabled if all input files are compiled with the flag, or if -z shstk is specified.

IBT-enabled executables/shared objects have two PLT sections, ".plt" and
".plt.sec".  For the details as to why we have two sections, please read
the comments.

Reviewed By: xiangzhangllvm

Differential Revision: https://reviews.llvm.org/D59780

I used the codemod python tool to do this with the following commands:

codemod 'tensorflow/mlir/blob/master/include' 'llvm/llvm-project/blob/master/mlir/include'
codemod 'tensorflow/mlir/blob/master' 'llvm/llvm-project/blob/master/mlir'
codemod 'tensorflow/mlir' 'llvm-project/llvm'

Differential Revision: https://reviews.llvm.org/D72244

[X86] Copy the nofpexcept flag when folding a load into an instruction using the load folding tables./

Fix riscv-toolchain-extra tests to pass when CLANG_RESOURCE_DIR is set
to another value than the default.

Differential Revision: https://reviews.llvm.org/D72591

Reviewers: nicolasvasilache, andydavis1, rriddle

Reviewed By: nicolasvasilache, rriddle

Subscribers: merge_guards_bot, mehdi_amini, rriddle, jpienaar, burmako, shauheen, antiagainst, nicolasvasilache, arpith-jacob, mgester, lucyrfox, liufengdb, llvm-commits

Tags: #llvm

Differential Revision: https://reviews.llvm.org/D72664

This fixes an error thrown by clang 3.8 that no viable conversion from
returned value to the function return type.

This causes an error with older versions of clang: constructor for
'llvm::exegesis::InstructionsCache' must explicitly initialize the const
member 'BVC'

We're planning to remove the shufflemask operand from ShuffleVectorInst
(D72467); fix GlobalISel so it doesn't depend on that Constant.

The change to prelegalizercombiner-shuffle-vector.mir happens because
the input contains a literal "-1" in the mask (so the parser/verifier
weren't really handling it properly). We now treat it as equivalent to
"undef" in all contexts.

Differential Revision: https://reviews.llvm.org/D72663

These were temporarily disabled in 2013 and we apparently forgot to
ever turn them back on again.

Fix spelling of flag to llvm-mc in recently-added test that wasn't
actually being run due to this.