This is to simplify an upcoming change where we distinguish between
flavors of libc++ by adding an apple-libc++ Lit feature.
Differential Revision: https://reviews.llvm.org/D110870
This patch makes instruction-referencing accepts an additional scenario
where values can be read from physical registers at the start of blocks. As
far as I was aware, this only happened:
* With arguments in the entry block,
* With constant physical registers,
To which this patch adds a third case:
* With exception-handling landing-pad blocks
In the attached test: the operand of the dbg.value traces back to the
"landingpad" instruction, which becomes some copies from physregs. Right
now, that's deemed unacceptable, and the assertion fires. The fix is to
just accept this scenario; this is a case where the value in question is
defined by a register and a position, not by an instruction that defines
it. Reading it with a DBG_PHI is the correct behaviour, there isn't a
non-copy instruction that we can refer to.
Differential Revision: https://reviews.llvm.org/D109005
We keep a map from function name to source location so we don't have to
do it via looking up a source location from the AST. However, since
function names can be long, we actually use a hash of the function name
as the key.
Additionally, we can't rely on Clang's printing of function names via
the AST, so we just demangle the name instead.
This is necessary to implement
https://lists.llvm.org/pipermail/cfe-dev/2021-September/068930.html.
Reviewed By: dblaikie
Differential Revision: https://reviews.llvm.org/D110665
When we check if a load is loop invariant by finding a dominating
invariant.start call, we strip bitcasts until we get to an i8* Value,
and look for an invariant.start use of the i8* Value.
We may accidentally end up at an i8 global and look at a global's uses,
which we shouldn't do in a loop pass. Although we could make this
logic work with globals, that's not currently intended.
Reviewed By: nikic
Differential Revision: https://reviews.llvm.org/D111098
Per the GCC info page:
If the function is declared 'extern', then this definition of the
function is used only for inlining. In no case is the function
compiled as a standalone function, not even if you take its address
explicitly. Such an address becomes an external reference, as if
you had only declared the function, and had not defined it.
Respect that behavior for inline builtins: keep the original definition, and
generate a copy of the declaration suffixed by '.inline' that's only referenced
in direct call.
This fixes holes in c3717b6858.
Differential Revision: https://reviews.llvm.org/D111009
This fixes some typos in OpDefinitions.md and DeclarativeRewrites.md,
and wrap function/class names in backticks.
Reviewed By: mehdi_amini
Differential Revision: https://reviews.llvm.org/D110582
A bunch of MTE tests like ./ScudoUnitTest-aarch64-Test/MemtagTest.StoreTags
can fail on aarch64-linux if the kernel doesn't support the tagged address ABI. It looks like
the call to prctl(PR_GET_TAGGED_ADDR_CTRL, 0, 0, 0, 0) can return -1, which
casted to an unsigned int and masked will return a value not equal to
PR_MTE_TCF_NONE, meaning systemDetectsMemoryTagFaultsTestOnly can return an incorrect value.
This updates the check to account for a failing prctl call.
Differential Revision: https://reviews.llvm.org/D110888
When determining NoAlias based on object size and dereferenceability
information, we can ignore frees for the same reason we can ignore
possible null pointers (if null is not a valid pointer): Actually
accessing the null pointer / freed pointer would be immediate UB,
and AA results are only valid under the assumption of an access.
This addresses a minor regression from D110745.
Differential Revision: https://reviews.llvm.org/D111028
The builtins: `__compare_and_swaplp`, `__fetch_and_addlp`,
` __fetch_and_andlp`, `__fetch_and_orlp`, `__fetch_and_swaplp` are
64 bit only. This patch ensures the compiler produces an error in 32 bit mode.
Reviewed By: #powerpc, nemanjai
Differential Revision: https://reviews.llvm.org/D110824
The delayed stack protector feature which is currently used for SDAG (and thus
allows for more commonly generating tail calls) depends on being able to extract
the tail call into a separate return block. To do this it also has to extract
the vreg->physreg copies that set up the call's arguments, since if it doesn't
then the call inst ends up using undefined physregs in it's new spliced block.
SelectionDAG implementations can do this because they delay emitting register
copies until *after* the stack arguments are set up. GISel however just
processes and emits the arguments in IR order, so stack arguments always end up
last, and thus this breaks the code that looks for any register arg copies that
precede the call instruction.
This patch adds a thunk argument to the assignValueToReg() and custom assignment
hooks. For outgoing arguments, register assignments use this return param to
return a thunk that does the actual generating of the copies. We collect these
until all the outgoing stack assignments have been done and then execute them,
so that the copies (and perhaps some artifacts like G_SEXTs) are placed after
any stores.
Differential Revision: https://reviews.llvm.org/D110610
The discussion in https://reviews.llvm.org/D110425 demonstrated that "packing"
may be a confusing term to define the behavior of this op in presence of the
attribute. Instead, indicate the intended effect of preventing the folder from
being applied.
Reviewed By: nicolasvasilache, silvas
Differential Revision: https://reviews.llvm.org/D111046
This provides better support for `TypeLoc`s to allow `TypeLoc`-related
matchers to feature stricter typing and to avoid relying on the dynamic
casting of `TypeLoc`s in matchers.
Reviewed By: ymandel, tdl-g, sbenza
Differential Revision: https://reviews.llvm.org/D110586
fir.array_update is only handling intrinsic assignments.
They are two big differences with user defined assignments:
1. The LHS and RHS types may not match, this does not play well
with fir.array_update that relies on both the merge and the
updated element to have the same type.
2. user defined assignment has a call semantics, with potential
side effects. So if a fir.array_update can hide a call, it traits
would need to be updated.
Instead of hiding more semantic in the fir.array_update, introduce
a new fir.array_modify op that allows de-correlating indicating that
an array value element is modified, and how it is modified.
This allow the ArrayValueCopy pass to still perform copy elision
while not having to implement the call itself, and could in general
be used for all kind of assignments (e.g. character assignment).
Update the alias analysis to not rely on the merge arguments (since
fir.array_modify has none).
Instead, analyze what is done with the element address.
This implies adding the ability to follow the users of fir.array_modify,
as well as being able to go through fir.store that may be generated to
store the RHS value in order to pass it to a user define routine.
This is done by adding a ReachCollector class to gather all array
accesses.
This patch is part of the upstreaming effort from fir-dev branch.
Reviewed By: schweitz
Differential Revision: https://reviews.llvm.org/D110928
Co-authored-by: Valentin Clement <clementval@gmail.com>
Fix some clang-tidy wrning in flang/Optimizer/Support and
remove explicit number of inlined elements for SmallVector. This
is mostly to sync with the changes from fir-dev.
This patch is part of the upstreaming effort from fir-dev branch.
Reviewed By: schweitz
Differential Revision: https://reviews.llvm.org/D111044
Both ports are required for BitScan ops. Update the port usage and distributed throughput based off the most recent llvm-exegesis captures (PR36895) and what Intel AoM / Agner reports as well.
Also remove some redundancy because the source and result
types of any multiply are always the same.
Differential Revision: https://reviews.llvm.org/D110926
This splits out the logic from shouldChangeType() that
currently allows 8/16/32-bit transforms even if those
types are not listed as legal in the data layout.
This could be useful as a predicate for vector
insert/extract transforms.
Note that this leaves the subsequent checks in
shouldChangeType() unchanged. We may want to merge
the checks for i1 and/or "ToLegal" into "isDesirable",
but that may alter existing transforms.
We were previously just ignoring unreachable, but targets like Darwin want to
keep unreachable instructions as traps.
Differential Revision: https://reviews.llvm.org/D110603
The LE Power sanitizer bot fails when testing standalone compiler-rt due to
an MSAN test warning introduced by -Wbitwise-instead-of-logical. As this option
along with -Werror is enabled on the bot, the test failure occurs.
This patch updates msan_test.cpp to fix the warning introduced by the
-Wbitwise-instead-of-logical.
In working on D106362 I found that a few more tests were needed. I've
been asked to pre-push the tests for that ticket. This should complete
the tests needed for now.
AArch64StorePairSuppress will prevent the creation of LDP's based on
scheduling info. This shouldn't apply when optimizing for size though,
where the size decrease should be considered more important.
Differential Revision: https://reviews.llvm.org/D110809
The `fallback` setting for import-std-module is supposed to allow running
expression that require an imported C++ module without causing any regressions
for users (neither in terms of functionality nor performance). This is done by
first trying to normally parse/evaluate an expression and when an error occurred
during this first attempt, we retry with the loaded 'std' module.
When we run into a system with a 'std' module that for some reason doesn't build
or otherwise causes parse errors, then this currently means that the second
parse attempt will overwrite the error diagnostics of the first parse attempt.
Given that the module build errors are outside of the scope of what the user can
influence, it makes more sense to show the errors from the first parse attempt
that are only concerned with the actual user input.
Reviewed By: aprantl
Differential Revision: https://reviews.llvm.org/D110696
The THEN keyword in the "ELSE IF (test) THEN" statement is useless
syntactically, and to omit it is a common error (at least for me!)
that has poor error recovery. This patch changes the parser to
cough up a simple "expected 'THEN'" and still recognize the rest of
the IF construct.
Differential Revision: https://reviews.llvm.org/D110952
The first line of flang/include/flang/Runtime/magic-numbers.h
got split into two somehow; join it back up.
Differential Revision: https://reviews.llvm.org/D110965
In some build environments, the C++ compiler is unable to infer the
correct type for the DenseMap::insert in isErrorBlock. Typing out
std::make_pair helps.
This patch is mainly to propogate location attribute from spv.GlobalVariable to llvm.mlir.global.
It also contains three small changes.
1. Remove the restriction on UniformConstant In SPIRVToLLVM.cpp;
2. Remove the errorCheck on relaxedPrecision when deserializering SPIR-V in Deserializer.cpp
3. In SPIRVOps.cpp, let ConstantOp take signedInteger too.
Co-authered: Alan Liu <alanliu.yf@gmail.com> and Xinyi Liu <xyliuhelen@gmail.com>
Reviewed by:antiagainst
Differential revision: https://reviews.llvm.org/D110207
The objc_clsopt_v16_t struct does not match up with the macOS/iOS15
dyld_shared_cache ObjC runtime structures. A struct field was seemingly
omitted.
Differential revision: https://reviews.llvm.org/D110477
