Using Chandler's words from r265331:
This commit was greatly exacerbating PR17409 and effectively regressed
build time for lot of (very large) code when compiled with ASan or MSan.
PR17409 is fixed by r269249, so this is fine to reapply r263460.
Original commit message:
The bad behavior happens when we have a function with a long linear
chain of basic blocks, and have a live range spanning most of this
chain, but with very few uses.
Let say we have only 2 uses.
The Hopfield network is only seeded with two active blocks where the
uses are, and each iteration of the outer loop in
`RAGreedy::growRegion()` only adds two new nodes to the network due to
the completely linear shape of the CFG. Meanwhile,
`SpillPlacer->iterate()` visits the whole set of discovered nodes, which
adds up to a quadratic algorithm.
This is an historical accident effect from r129188.
When the Hopfield network is expanding, most of the action is happening
on the frontier where new nodes are being added. The internal nodes in
the network are not likely to be flip-flopping much, or they will at
least settle down very quickly. This means that while
`SpillPlacer->iterate()` is recomputing all the nodes in the network, it
is probably only the two frontier nodes that are changing their output.
Instead of recomputing the whole network on each iteration, we can
maintain a SparseSet of nodes that need to be updated:
- `SpillPlacement::activate()` adds the node to the todo list.
- When a node changes value (i.e., `update()` returns true), its
neighbors are added to the todo list.
- `SpillPlacement::iterate()` only updates the nodes in the list.
The result of Hopfield iterations is not necessarily exact. It should
converge to a local minimum, but there is no guarantee that it will find
a global minimum. It is possible that updating nodes in a different
order will cause us to switch to a different local minimum. In other
words, this is not NFC, but although I saw a few runtime improvements
and regressions when I benchmarked this change, those were side effects
and actually the performance change is in the noise as expected.
Huge thanks to Jakob Stoklund Olesen <stoklund@2pi.dk> for his
feedbacks, guidance and time for the review.
llvm-svn: 270149
Work around crashes in ``__sanitizer_malloc_hook()`` under Mac OSX.
Under Mac OSX we intercept calls to malloc before thread local
storage is initialised leading to a crash when accessing
``AllocTracer``. To workaround this ``AllocTracer`` is only accessed
in the hook under Linux. For symmetry ``__sanitizer_free_hook()``
is also modified in the same way.
To support this change a set of new macros
LIBFUZZER_LINUX and LIBFUZZER_APPLE has been defined which can be
used to check the target being compiled for.
Differential Revision: http://reviews.llvm.org/D20402
llvm-svn: 270145
This removes the subclasses of ProfileSummary, moves the members of the derived classes to the base class.
Differential Revision: http://reviews.llvm.org/D20390
llvm-svn: 270143
When matching an interleaved load to an ldN pattern, the interleaved access
pass checks that all users of the load are shuffles. If the load is used by an
instruction other than a shuffle, the pass gives up and an ldN is not
generated. This patch considers users of the load that are extractelement
instructions. It attempts to modify the extracts to use one of the available
shuffles rather than the load. After the transformation, the load is only used
by shuffles and will then be matched with an ldN pattern.
Differential Revision: http://reviews.llvm.org/D20250
llvm-svn: 270142
This splits ProfileSummary into two classes: a ProfileSummary class that has methods to convert from/to metadata and a ProfileSummaryBuilder class that computes the profiles summary which is in ProfileData.
Differential Revision: http://reviews.llvm.org/D20314
llvm-svn: 270136
This patch fixes https://llvm.org/bugs/show_bug.cgi?id=27703.
If there is a sequence of one or more load instructions, each loaded value is used as address of later load instruction, bitcast is necessary to change the value type, don't optimize it.
llvm-svn: 270135
This re-applies r270115.
Many of the MachO load commands can have data appended after the command structure. This data is frequently strings, but can actually be anything. This patch adds support for three optional fields on load command yaml descriptions.
The new PayloadString YAML field is populated with the data after load commands known to have strings as extra data.
The new ZeroPadBytes YAML field is a count of zero'd bytes after the end of the load command structure before the next command. This can apply anywhere in the file. MachO2YAML verifies that bytes are zero before populating this field, and YAML2MachO will add zero'd bytes.
The new PayloadBytes YAML field stores all bytes after the end of the load command structure before the next command if they are non-zero. This is a catch all for all unhandled bytes. If MachO2Yaml populates PayloadBytes it will not populate ZeroPadBytes, instead zero'd bytes will be in the PayloadBytes structure.
llvm-svn: 270124
Many of the MachO load commands can have data appended after the command structure. This data is frequently strings, but can actually be anything. This patch adds support for three optional fields on load command yaml descriptions.
The new PayloadString YAML field is populated with the data after load commands known to have strings as extra data.
The new ZeroPadBytes YAML field is a count of zero'd bytes after the end of the load command structure before the next command. This can apply anywhere in the file. MachO2YAML verifies that bytes are zero before populating this field, and YAML2MachO will add zero'd bytes.
The new PayloadBytes YAML field stores all bytes after the end of the load command structure before the next command if they are non-zero. This is a catch all for all unhandled bytes. If MachO2Yaml populates PayloadBytes it will not populate ZeroPadBytes, instead zero'd bytes will be in the PayloadBytes structure.
llvm-svn: 270115
Since the calls don't return, the instruction afterwards will never run,
and is just taking up unnecessary space in the binary.
Differential Revision: http://reviews.llvm.org/D20406
llvm-svn: 270109
According to Cuda Programming guide (v7.5, E2.3.1):
> __device__, __constant__ and __shared__ variables defined in namespace
> scope, that are of class type, cannot have a non-empty constructor or a
> non-empty destructor.
Clang already deals with device-side constructors (see D15305).
This patch enforces similar rules for destructors.
Differential Revision: http://reviews.llvm.org/D20140
llvm-svn: 270108
Codegen tests for device-side variable initialization are subset of test
cases used to verify Sema's part of the job.
Including CodeGenCUDA/device-var-init.cu from SemaCUDA makes it easier to
keep both sides in sync.
Differential Revision: http://reviews.llvm.org/D20139
llvm-svn: 270107
A baby step toward translating DIType records to CodeView.
This does not (yet) combine the record length with the record data. I'm going back and forth trying to determine if that's a good idea.
llvm-svn: 270106
getRegAsmName ends up making a copy of the register's name in order to
make a lower-case version of it. This is bad because
getRegForInlineAsmConstraint, it's sole caller, does a lowercase
comparison anyway.
This resulted in a significant regression in compile time for the Linux
kernel because getRegAsmName is called in a loop by
getRegForInlineAsmConstraint.
Instead, forgo the call to lower in getRegAsmName and have it return a
StringRef.
No functionality change is intended.
llvm-svn: 270099
This matches default nvcc behavior and gives substantial
performance boost on GPU where fmad is much cheaper compared to add+mul.
Differential Revision: http://reviews.llvm.org/D20341
llvm-svn: 270094
the linker script. The cycle in the ELF/LinkerScript.cpp:assignAddresses()
routine will be used to go through all the sections and set all the
addresses correctly.
Add new test to check this case.
llvm-svn: 270090
Previously, specifying -post-RA-scheduler=true had the side effect of
disabling the antidependency breaker, yielding different behavior than
if the post-RA-scheduler was enabled via the scheduling model.
Differential Revision: http://reviews.llvm.org/D20186
llvm-svn: 270077
It broke buildbot:
http://lab.llvm.org:8011/builders/clang-s390x-linux/builds/4817/steps/ninja%20check%201/logs/stdio
Actually it is just because D20273 not yet commited, but these 2 were crossing with each other,
and I`ll better find the way to land them separatelly soon.
Initial commit message:
[llvm-mc] - Teach llvm-mc to generate compressed debug sections in zlib style.
Before this patch llvm-mc generated zlib-gnu styled sections.
That means no SHF_COMPRESSED flag was set, magic 'zlib' signature
was used in combination with full size field. Sections were renamed to "*.z*".
This patch reimplements the compression style to zlib one as zlib-gnu looks
to be depricated everywhere.
Differential revision: http://reviews.llvm.org/D20331
llvm-svn: 270075
There are at least 2 places (DAGCombiner, X86ISelLowering) where this could be used instead
of ad-hoc and watered down code that is trying to match a power-of-2 pattern.
Differential Revision: http://reviews.llvm.org/D20439
llvm-svn: 270073
Quentin Colombet