The cttz/ctlz intrinsics have a parameter specifying whether the
result is undefined for zero. cttz(x, false) can be relaxed to
cttz(x, true) if x is known non-zero, and in fact such an optimization
is already performed. However, this currently doesn't work if x is
non-zero as a result of a select rather than an explicit branch.
This patch adds handling for this case, thus allowing
x != 0 ? cttz(x, false) : y to simplify to x != 0 ? cttz(x, true) : y.
Differential Revision: https://reviews.llvm.org/D55786
llvm-svn: 350463
This enables x86 relaxation by default. This depends on a linker new
enough to support the new reloc types but since we default to lld we
don't worry about host system linkers that might be too old to support
the new reloc types.
Differential Revision: https://reviews.llvm.org/D56349
llvm-svn: 350460
This has some minor optimizations to shouldBeDeferred. This is not
strictly NFC because the early exit inside the loop assumes
TotalSecondaryCost is monotonically non-decreasing, which is not true if
the threshold used by CostAnalyzer is negative. AFAICT the thresholds do
not go below 0 for the default values of the various options we use.
llvm-svn: 350456
We don't need to require the first operand to be an integer because we already said it was the same type as the result which we also constrained to an integer.
llvm-svn: 350455
This is slightly ugly for three reasons:
- The shlib needs to go next to the binary to be found on all platforms, so the
build files refer to target_out_dir
- The explicit -fPIC flag needed on the shared lib side, and the -rdynamic flag
needed on the host side, on Linux
- Plugins that refer to LLVM code and assume that the host will resolve them
don't work on Windows -- PluginsTests won't test anything on Windows (but
DynamicLibraryTests will, since the dll here doesn't call LLVM code)
If we get lots more of these plugin / plugin host targets it might make sense
to add a template for them. But for now, these are the last ones we need.
(We're at 6 plugin hosts, 2 of them tests, and at 6 shared libraries, 2 of them
tests as well. clang is a plugin host by default in the CMake build but not
(yet?) in the GN build.)
Differential Revision: https://reviews.llvm.org/D56330
llvm-svn: 350454
Saves up to 1.3 sec on large PDBs.
Figures below are for the "Globals Stream Layout" pass:
Before This patch
Large EXE (PDB is ~2 GB) 3330 ms 2022 ms
Large EXE (PDB is ~2 GB) 2680 ms 1608 ms
Large DLL (PDB is ~1 GB) 1455 ms 938 ms
Large DLL (PDB is ~800 MB) 1215 ms 800 ms
Small DLL (PDB is ~200 MB) 224 ms 146 ms
Differential Revision: https://reviews.llvm.org/D56334
llvm-svn: 350452
Moves the code added in r350340 around a bit, to hopefully make the existing
plugin tests pass when clang is built with examples enabled.
llvm-svn: 350451
I'm pretty unhappy this patch: DebugInfoPDBTests uses an API that requires some
magic txt file to be next to the unit test executable that stores the absolute
path to the LLVM source root.
The choices here are:
1. Don't use the unittest() template for DebugInfoPDBTests and set output_dir
for unit tests in two places (the gni file for every test but this one, and the
BUILD.gn file for this specific test).
2. Add another unittest_foo() template variation for this one test.
I went with the former, and added a comment to the template to look out for
this.
(The CMake build has the same issue.)
Differential Revision: https://reviews.llvm.org/D56324
llvm-svn: 350447
We still need the interceptor on non-aarch64 to untag the pthread_t
and pthread_attr_t pointers and disable tagging on allocations done
internally by glibc.
llvm-svn: 350445
Summary:
The default setting kTabSizeLog=20 results in an 8Mb global hash table,
almost all of it in private pages. That is not a sane setting in a
mobile, system-wide use case: with ~150 concurrent processes stack
depot will account for more than 1Gb of RAM.
Reviewers: kcc, pcc
Subscribers: srhines, kubamracek, llvm-commits
Differential Revision: https://reviews.llvm.org/D56333
llvm-svn: 350443
In LTO or Thin-lto mode (though linker plugin), the module
names are of temp file names which are different for
different compilations. Using SourceFileName avoids the issue.
This should not change any functionality for current PGO as
all the current callers of getPGOFuncName() is before LTO.
llvm-svn: 350442
As discussed in D56113, this patch refactors the implementation of the
const restriction for linear to reuse a function introduced by D56113.
A side effect is that, if a variable has mutable members, this
diagnostic is now skipped, and the diagnostic for the variable not
being an integer or pointer is reported instead.
Reviewed By: ABataev
Differential Revision: https://reviews.llvm.org/D56299
llvm-svn: 350441
As discussed in D56113, this patch refactors the implementation of the
const restriction for reductions to reuse a function introduced by
D56113. A side effect is that diagnostics sometimes now say
"variable" instead of "list item" when a list item is a variable.
Reviewed By: ABataev
Differential Revision: https://reviews.llvm.org/D56298
llvm-svn: 350440
The following appears in OpenMP 3.1 sec. 2.9.1.1 as a predetermined
data-sharing attribute:
> Variables with const-qualified type having no mutable member are
> shared.
It does not appear in OpenmP 4.0, 4.5, or 5.0. This patch removes the
implementation of that attribute when the requested OpenMP version is
greater than 3.1.
One effect of that removal is that `default(none)` affects const
variables without mutable members.
Also, without this patch, if a const variable without mutable members
was explicitly lastprivate or private, it was an error because it was
predetermined shared. Now, clang instead complains that it's const
without mutable fields, which is a more intelligible diagnostic. That
should be fine for all of the above versions because they all have
something like the following, which is quoted from OpenMP 5.0
sec. 2.19.3:
> A variable that is privatized must not have a const-qualified type
> unless it is of class type with a mutable member. This restriction does
> not apply to the firstprivate clause.
reduction and linear clauses already have separate checks for const
variables. Future patches will merge the implementations.
Reviewed By: ABataev
Differential Revision: https://reviews.llvm.org/D56113
llvm-svn: 350439
In addition to finding dead uses of instructions, also find dead uses
of function arguments, and replace them with zero as well.
I'm changing the way the known bits are computed here to remove the
coupling between the transfer function and the algorithm. It previously
relied on the first op being visited first and computing known bits --
unless the first op is not an instruction, in which case they're computed
on the second op. I could have adjusted this to check for "instruction
or argument", but I think it's better to avoid the repeated calculation
with an explicit flag.
Differential Revision: https://reviews.llvm.org/D56247
llvm-svn: 350435
This adds support for calculating sign bits of insert_subvector. I based it on the computeKnownBits.
My motivating case is propagating sign bits information across basic blocks on AVX targets where concatenating using insert_subvector is common.
Differential Revision: https://reviews.llvm.org/D56283
llvm-svn: 350432
The problem is similar to D55986 but for threads: a process with the
interceptor hwasan library loaded might have some threads started by
instrumented libraries and some by uninstrumented libraries, and we
need to be able to run instrumented code on the latter.
The solution is to perform per-thread initialization lazily. If a
function needs to access shadow memory or add itself to the per-thread
ring buffer its prologue checks to see whether the value in the
sanitizer TLS slot is null, and if so it calls __hwasan_thread_enter
and reloads from the TLS slot. The runtime does the same thing if it
needs to access this data structure.
This change means that the code generator needs to know whether we
are targeting the interceptor runtime, since we don't want to pay
the cost of lazy initialization when targeting a platform with native
hwasan support. A flag -fsanitize-hwaddress-abi={interceptor,platform}
has been introduced for selecting the runtime ABI to target. The
default ABI is set to interceptor since it's assumed that it will
be more common that users will be compiling application code than
platform code.
Because we can no longer assume that the TLS slot is initialized,
the pthread_create interceptor is no longer necessary, so it has
been removed.
Ideally, lazy initialization should only cost one instruction in the
hot path, but at present the call may cause us to spill arguments
to the stack, which means more instructions in the hot path (or
theoretically in the cold path if the spills are moved with shrink
wrapping). With an appropriately chosen calling convention for
the per-thread initialization function (TODO) the hot path should
always need just one instruction and the cold path should need two
instructions with no spilling required.
Differential Revision: https://reviews.llvm.org/D56038
llvm-svn: 350429
The Android dynamic loader has a non-standard feature that allows
libraries such as the hwasan runtime to interpose symbols even after
the symbol already has a value. The new value of the symbol is used to
relocate libraries loaded after the interposing library, but existing
libraries keep the old value. This behaviour is activated by the
DF_1_GLOBAL flag in DT_FLAGS_1, which is set by passing -z global to
the linker, which is what we already do to link the hwasan runtime.
What this means in practice is that if we have .so files that depend
on interceptor-mode hwasan without the main executable depending on
it, some of the libraries in the process will be using the hwasan
allocator and some will be using the system allocator, and these
allocators need to interact somehow. For example, if an instrumented
library calls a function such as strdup that allocates memory on
behalf of the caller, the instrumented library can reasonably expect
to be able to call free to deallocate the memory.
We can handle that relatively easily with hwasan by using tag 0 to
represent allocations from the system allocator. If hwasan's realloc
or free functions are passed a pointer with tag 0, the system allocator
is called.
One limitation is that this scheme doesn't work in reverse: if an
instrumented library allocates memory, it must free the memory itself
and cannot pass ownership to a system library. In a future change,
we may want to expose an API for calling the system allocator so
that instrumented libraries can safely transfer ownership of memory
to system libraries.
Differential Revision: https://reviews.llvm.org/D55986
llvm-svn: 350427
The offload bundler action should not unbundle the input file types that does not match the action type. This fixes an issue where .so files are unbundled when the action type is object files.
Reviewers: yaxunl
Differential Revision: https://reviews.llvm.org/D56321
llvm-svn: 350426
The offload bundler action should not unbundle the input file types that does not match the action type. This fixes an issue where .so files are unbundled when the action type is object files.
llvm-svn: 350425
Summary:
As with NameAnonGlobals, invoke the new CanonicalizeAliases via clang
when using the new PM.
Depends on D54507.
Reviewers: pcc, davidxl
Subscribers: mehdi_amini, inglorion, steven_wu, dexonsmith, cfe-commits
Differential Revision: https://reviews.llvm.org/D55620
llvm-svn: 350424
At -O0, globalopt is not run during the compile step, and we can have a
chain of an alias having an immediate aliasee of another alias. The
summaries are constructed assuming aliases in a canonical form
(flattened chains), and as a result only the base object but no
intermediate aliases were preserved.
Fix by adding a pass that canonicalize aliases, which ensures each
alias is a direct alias of the base object.
Reviewers: pcc, davidxl
Subscribers: mehdi_amini, inglorion, eraman, steven_wu, dexonsmith, arphaman, llvm-commits
Differential Revision: https://reviews.llvm.org/D54507
llvm-svn: 350423
This attribute, called "objc_externally_retained", exposes clang's
notion of pseudo-__strong variables in ARC. Pseudo-strong variables
"borrow" their initializer, meaning that they don't retain/release
it, instead assuming that someone else is keeping their value alive.
If a function is annotated with this attribute, implicitly strong
parameters of that function aren't implicitly retained/released in
the function body, and are implicitly const. This is useful to expose
for performance reasons, most functions don't need the extra safety
of the retain/release, so programmers can opt out as needed.
This attribute can also apply to declarations of local variables,
with similar effect.
Differential revision: https://reviews.llvm.org/D55865
llvm-svn: 350422
The 1st try for this was at rL350369, but it caused IR-level diffs because
our cost models differentiate custom vs. legal/promote lowering. So that was
reverted at rL350373. The cost models were fixed independently at rL350403,
so this is effectively the same patch as last time.
Original commit message:
This would show up if we fix horizontal reductions to narrow as they go along,
but it's an improvement for size and/or Jaguar (fast-hops) independent of that.
We need to do this late to not interfere with other pattern matching of larger
horizontal sequences.
We can extend this to integer ops in a follow-up patch.
Differential Revision: https://reviews.llvm.org/D56011
llvm-svn: 350421
Lifetime markers which reference inputs to the extraction region are not
safe to extract. Example ('rhs' will be extracted):
```
entry:
+------------+
| x = alloca |
| y = alloca |
+------------+
/ \
lhs: rhs:
+-------------------+ +-------------------+
| lifetime_start(x) | | lifetime_start(x) |
| use(x) | | lifetime_start(y) |
| lifetime_end(x) | | use(x, y) |
| lifetime_start(y) | | lifetime_end(y) |
| use(y) | | lifetime_end(x) |
| lifetime_end(y) | +-------------------+
+-------------------+
```
Prior to extraction, the stack coloring pass sees that the slots for 'x'
and 'y' are in-use at the same time. After extraction, the coloring pass
infers that 'x' and 'y' are *not* in-use concurrently, because markers
from 'rhs' are no longer available to help decide otherwise.
This leads to a miscompile, because the stack slots actually are in-use
concurrently in the extracted function.
Fix this by moving lifetime start/end markers for memory regions defined
in the calling function around the call to the extracted function.
Fixes llvm.org/PR39671 (rdar://45939472).
Differential Revision: https://reviews.llvm.org/D55967
llvm-svn: 350420
Similar to rL350199 - there are no known analysis/codegen holes for
funnel shift intrinsics now, so we can canonicalize the 6+ regular
instructions to funnel shift to improve vectorization, inlining,
unrolling, etc.
llvm-svn: 350419
This patch adds #pragma clang loop pipeline and #pragma clang loop pipeline_initiation_interval for debugging or reducing compile time purposes. It is possible to disable SWP for concrete loops to save compilation time or to find bugs by not doing SWP to certain loops. It is possible to set value of initiation interval to concrete number to save compilation time by not doing extra pipeliner passes or to check created schedule for specific initiation interval.
Patch by Alexey Lapshin.
llvm-svn: 350414
Martin Storsjo