This splits collecting information from actually performing the transformation, so that we can add a heuristic in between the two.
NFC.
Differential Revision: http://reviews.llvm.org/D7497
llvm-svn: 228817
The NodeMetadata are maintained in an incremental way. When an edge between
2 nodes has its cost updated, in the course of graph reduction for example,
the NodeMetadata need first to have the old edge cost removed, then the new
edge cost added. Only once the NodeMetadata have been fully updated, it
becomes safe to consider promoting the nodes to the
ConservativelyAllocatable or OptimallyReducible sets. Previously, this
promotion was occuring right after the removing the old cost, and this was
breaking the assumption that a ConservativelyAllocatable should not be
spilled.
This patch also adds asserts to:
- enforces the invariant that a node's reduction can not be downgraded,
- only not provably allocatable or optimally reducible nodes can be spilled.
llvm-svn: 228816
This is a slightly convoluted workaround. GCC does not support the
__has_feature extension of clang, and this results in some issues with
static_asserts. config.h defines static_assert as a macro with a C-specific
trickery. This then propagates into the C++ headers included after config.h,
which are used with C++11 mode, enabling constexpr constructors. The macro'ed
static_assert does not get treated as the static_assert builtin, and will cause
an error due to a non-empty constexpr constructor. Tweaking the include order
permits the use of libc++ headers to build libunwind with GCC on Linux.
llvm-svn: 228809
Mark that the functions always return or abort if the register class is
unhandled. Avoid placing the abort in the switch to permit -Wswitch-cover to
catch missing values.
llvm-svn: 228808
This allows IDEs to recognize the entire set of header files for
each of the core LLVM projects.
Differential Revision: http://reviews.llvm.org/D7526
Reviewed By: Chris Bieneman
llvm-svn: 228798
We're still using global constructors when not on linux, so this was
causing warnings on Darwin. Reverting for now.
This reverts r228384.
llvm-svn: 228795
This is a test for the llvm change in r228793. We need to make sure
that names referred to by coverage end up in the right section, or the
coverage tools won't work.
llvm-svn: 228794
Add handling for __llvm_coverage_mapping to the InstrProfiling
pass. We need to make sure the constant and any profile names it
refers to are in the correct sections, which is easier and cleaner to
do here where we have to know about profiling sections anyway.
This is really tricky to test without a frontend, so I'm committing
the test for the fix in clang. If anyone knows a good way to test this
within LLVM, please let me know.
Fixes PR22531.
llvm-svn: 228793
(or of a lambda init-capture, which is sort-of such a variable). The semantics
of such constructs will change when we implement N3922, so we intend to warn on
this in Clang 3.6 then change the semantics in Clang 3.7.
llvm-svn: 228792
A runtime support value is a ValueObject whose only purpose is to support some language runtime's operation, but it does not directly provide any user-visible benefit
As such, unless the user is working on the runtime support, it is mostly safe for them not to see such a value when debugging
It is a language runtime's job to check whether a ValueObject is a support value, and that - in conjunction with a target setting - is used by frame variable and target variable
SBFrame::GetVariables gets a new overload with yet another flag to dictate whether to return those support values to the caller - that which defaults to the setting's value
rdar://problem/15539930
llvm-svn: 228791
Summary:
Hello Howard,
While running the libc++ tests on our ARM boards, we encounter sporadic failures of the two tests:
test/std/thread/futures/futures.shared_future/wait_until.pass.cpp
test/std/thread/futures/futures.unique_future/wait_until.pass.cpp
The worker thread might not finish yet when the main thread checks its result.
I filed the bug 21998 for this case: http://llvm.org/bugs/show_bug.cgi?id=21998
Would you be able to review this please?
Thank you.
Oleg
Reviewers: howard.hinnant, mclow.lists, danalbert, jroelofs, EricWF
Reviewed By: jroelofs, EricWF
Subscribers: EricWF, mclow.lists, aemerson, llvm-commits
Differential Revision: http://reviews.llvm.org/D6750
llvm-svn: 228783
If the landingpad of the invoke is using a personality function that
catches asynch exceptions, then it can catch a trap.
Also add some landingpads to invalid LLVM IR test cases that lack them.
Over-the-shoulder reviewed by David Majnemer.
llvm-svn: 228782
Because types are not reliably protected against the death of their owners, having ValueObjects lurking around like that past the useful lifetime of their owner processes is a potential source of crashes
That is - in itself - worth fixing at some point, but for this case, watchpoints holding on to old values don't offer enough value to make the larger fix worth
Fixes rdar://19788756
llvm-svn: 228777
Summary:
Make sure we don't print the error report from -fsanitize=function
twice for the same source location, as we do in another UBSan handlers.
Test Plan: check-ubsan test suite
Reviewers: rsmith, pcc
Reviewed By: pcc
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D7524
llvm-svn: 228772
- you have a type that contains a typedef to a VectorType or an ExtVectorType
- that type is returned from an ARM function that LLDB steps over so we try to figure out the return type
- we try to determine if the type is a homogeneous aggregate type and we crash
We get not using getAs() when we should have been and using llvm::cast caused an assertion crash when the typedef type didn't return a valid VectorType or ExtVectorType.
<rdar://problem/19646550>
llvm-svn: 228771
Make sure clear_cache() builtin has an appropriate definition for Linux.
Call syscall(NR_cacheflush, ...).
Differential Revision: http://reviews.llvm.org/D7205
llvm-svn: 228767
The isSigned argument of makeLibCall function was hard-coded to false
(unsigned). This caused zero extension on MIPS64 soft float.
As the result SingleSource/Benchmarks/Stanford/FloatMM test and
SingleSource/UnitTests/2005-07-17-INT-To-FP test failed.
The solution was to use the proper argument.
Patch by Strahinja Petrovic.
Differential Revision: http://reviews.llvm.org/D7292
llvm-svn: 228765
table entry. This happens when SROA splits up an alloca and the resulting
allocas cannot be lowered to SSA values because their address is passed
to a function.
Fixes PR22502.
llvm-svn: 228764
David Majnemer