This patch detects vector reductions before instruction selection. Vector
reductions are vectorized reduction operations, and for such operations we have
freedom to reorganize the elements of the result as long as the reduction of them
stay unchanged. This will enable some reduction pattern recognition during
instruction combine such as SAD/dot-product on X86. A flag is added to
SDNodeFlags to mark those vector reduction nodes to be checked during instruction
combine.
To detect those vector reductions, we search def-use chains starting from the
given instruction, and check if all uses fall into two categories:
1. Reduction with another vector.
2. Reduction on all elements.
in which 2 is detected by recognizing the pattern that the loop vectorizer
generates to reduce all elements in the vector outside of the loop, which
includes several ShuffleVector and one ExtractElement instructions.
Differential revision: http://reviews.llvm.org/D15250
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@261070 91177308-0d34-0410-b5e6-96231b3b80d8
This reverts commit r261030 and r261036.
(The revision was marked "approved" on phabricator, but some concerns
were raised on the mailing list. Thanks D. Blaikie for notifying me.)
From: Mehdi Amini <mehdi.amini@apple.com>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@261055 91177308-0d34-0410-b5e6-96231b3b80d8
reference-edge SCCs.
This essentially builds a more normal call graph as a subgraph of the
"reference graph" that was the old model. This allows both to exist and
the different use cases to use the aspect which addresses their needs.
Specifically, the pass manager and other *ordering* constrained logic
can use the reference graph to achieve conservative order of visit,
while analyses reasoning about attributes and other properties derived
from reachability can reason about the direct call graph.
Note that this isn't necessarily complete: it doesn't model edges to
declarations or indirect calls. Those can be found by scanning the
instructions of the function if desirable, and in fact every user
currently does this in order to handle things like calls to instrinsics.
If useful, we could consider caching this information in the call graph
to save the instruction scans, but currently that doesn't seem to be
important.
An important realization for why the representation chosen here works is
that the call graph is a formal subset of the reference graph and thus
both can live within the same data structure. All SCCs of the call graph
are necessarily contained within an SCC of the reference graph, etc.
The design is to build 'RefSCC's to model SCCs of the reference graph,
and then within them more literal SCCs for the call graph.
The formation of actual call edge SCCs is not done lazily, unlike
reference edge 'RefSCC's. Instead, once a reference SCC is formed, it
directly builds the call SCCs within it and stores them in a post-order
sequence. This is used to provide a consistent platform for mutation and
update of the graph. The post-order also allows for very efficient
updates in common cases by bounding the number of nodes (and thus edges)
considered.
There is considerable common code that I'm still looking for the best
way to factor out between the various DFS implementations here. So far,
my attempts have made the code harder to read and understand despite
reducing the duplication, which seems a poor tradeoff. I've not given up
on figuring out the right way to do this, but I wanted to wait until
I at least had the system working and tested to continue attempting to
factor it differently.
This also requires introducing several new algorithms in order to handle
all of the incremental update scenarios for the more complex structure
involving two edge colorings. I've tried to comment the algorithms
sufficiently to make it clear how this is expected to work, but they may
still need more extensive documentation.
I know that there are some changes which are not strictly necessarily
coupled here. The process of developing this started out with a very
focused set of changes for the new structure of the graph and
algorithms, but subsequent changes to bring the APIs and code into
consistent and understandable patterns also ended up touching on other
aspects. There was no good way to separate these out without causing
*massive* merge conflicts. Ultimately, to a large degree this is
a rewrite of most of the core algorithms in the LCG class and so I don't
think it really matters much.
Many thanks to the careful review by Sanjoy Das!
Differential Revision: http://reviews.llvm.org/D16802
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@261040 91177308-0d34-0410-b5e6-96231b3b80d8
Summary: Loading IR with debug info improves MDString::get() from 19ms to 10ms.
Reviewers: dexonsmith
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D16597
From: Mehdi Amini <mehdi.amini@apple.com>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@261030 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
On the contrary to Full LTO, ThinLTO can afford to shift compile time
from the frontend to the linker: both phases are parallel (even if
it is not totally "free": projects like clang are reusing product
from the "compile phase" for multiple link, think about
libLLVMSupport reused for opt, llc, etc.).
This pipeline is based on the proposal in D13443 for full LTO. We
didn't move forward on this proposal because the LTO link was far too
long after that. We believe that we can afford it with ThinLTO.
The ThinLTO pipeline integrates in the regular O2/O3 flow:
- The compile phase perform the inliner with a somehow lighter
function simplification. (TODO: tune the inliner thresholds here)
This is intendend to simplify the IR and get rid of obvious things
like linkonce_odr that will be inlined.
- The link phase will run the pipeline from the start, extended with
some specific passes that leverage the augmented knowledge we have
during LTO. Especially after the inliner is done, a sequence of
globalDCE/globalOpt is performed, followed by another run of the
"function simplification" passes. It is not clear if this part
of the pipeline will stay as is, as the split model of ThinLTO
does not allow the same benefit as FullLTO without added tricks.
The measurements on the public test suite as well as on our internal
suite show an overall net improvement. The binary size for the clang
executable is reduced by 5%. We're still tuning it with the bringup
of ThinLTO and it will evolve, but this should provide a good starting
point.
Reviewers: tejohnson
Differential Revision: http://reviews.llvm.org/D17115
From: Mehdi Amini <mehdi.amini@apple.com>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@261029 91177308-0d34-0410-b5e6-96231b3b80d8
It is intended to contains the passes run over a function after the
inliner is done with a function and before it moves to its callers.
From: Mehdi Amini <mehdi.amini@apple.com>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@261028 91177308-0d34-0410-b5e6-96231b3b80d8
The usual way to get a 32-bit relocation is to use a constant extender which doubles the size of the instruction, 4 bytes to 8 bytes.
Another way is to put a .word32 and mix code and data within a function. The disadvantage is it's not a valid instruction encoding and jumping over it causes prefetch stalls inside the hardware.
This relocation packs a 23-bit value in to an "r0 = add(rX, #a)" instruction by overwriting the source register bits. Since r0 is the return value register, if this instruction is placed after a function call which return void, r0 will be filled with an undefined value, the prefetch won't be confused, and the callee can access the constant value by way of the link register.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@261006 91177308-0d34-0410-b5e6-96231b3b80d8
Original message:
Get rid of the ifdefs in TargetLowering.
Introduce a new API used only by GlobalISel: CallLowering.
This API will contain target hooks dedicated to call lowering.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@260998 91177308-0d34-0410-b5e6-96231b3b80d8
Original messages:
Revert "[readobj] Handle ELF files with no section table or with no program headers."
Revert "[readobj] Dump DT_JMPREL relocations when outputting dynamic relocations."
r260489 depends on r260488 and among other issues r260488 deleted error
handling code.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@260962 91177308-0d34-0410-b5e6-96231b3b80d8
r260925 introduced a version of the *trim methods which is preferable
when trimming a single kind of character. Update all users in llvm.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@260926 91177308-0d34-0410-b5e6-96231b3b80d8
Add support for trimming a single kind of character from a StringRef.
This makes the common case of trimming null bytes much neater. It's also
probably a bit speedier too, since it avoids creating a std::bitset in
find_{first,last}_not_of.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@260925 91177308-0d34-0410-b5e6-96231b3b80d8
Introduce a new API used only by GlobalISel: CallLowering.
This API will contain target hooks dedicated to call lowering.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@260922 91177308-0d34-0410-b5e6-96231b3b80d8
This patch avoids the initial memset at the cost of making iterators
slightly more complex. This should be beneficial as most SmallPtrSets
hold no or only a few elements, while iterating over them is less
common.
Differential Revision: http://reviews.llvm.org/D16672
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@260913 91177308-0d34-0410-b5e6-96231b3b80d8
This ensures that all of the various pieces are working. The next patch
will wire up commandline-driven alias analysis chain building and allow
BasicAA to work with the AAManager.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@260838 91177308-0d34-0410-b5e6-96231b3b80d8
into the new pass manager and fix the latent bugs there.
This lets everything live together nicely, but it isn't really useful
yet. I never finished wiring the AA layer up for the new pass manager,
and so subsequent patches will change this to do that wiring and get AA
stuff more fully integrated into the new pass manager. Turns out this is
necessary even to get functionattrs ported over. =]
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@260836 91177308-0d34-0410-b5e6-96231b3b80d8
r180893 added an indirect include of llvm/Config/Targets.def to
llvm/Support/CodeGen.h, which in turn is included by things like
llvm/IR/Module.h. After a full build of LLVM and Clang, ninja had to
rebuild 1274 files after reconfiguring.
This commit strips CodeGen.h back down to just a pile of enums and moves
the expensive includes over to CodeGenCWrappers.h (which is only
included in two places). This gets ninja down to 88 files if you
reconfigure with, e.g., -DLLVM_TARGETS_TO_BUILD=X86.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@260835 91177308-0d34-0410-b5e6-96231b3b80d8
This intrinsic will be used to expose dpp functionality to higher-level
languages. It will map to the dpp version of v_mov_b32.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@260792 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
Export the CloneDebugInfoMetadata utility, which clones all debug info
associated with a function into the first module. Also use this function
in CloneModule on each function we clone (the CloneFunction entrypoint
already does this).
Without this, cloning a module will lead to DI quality regressions,
especially since r252219 reversed the Function <-> DISubprogram edge
(before we could get lucky and have this edge preserved if the
DISubprogram itself was, e.g. due to location metadata).
This was verified to fix missing debug information in julia and
a unittest to verify the new behavior is included.
Patch by Yichao Yu! Thanks!
Reviewers: loladiro, pcc
Differential Revision: http://reviews.llvm.org/D17165
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@260791 91177308-0d34-0410-b5e6-96231b3b80d8
As support expands to more runtimes, we'll need to
distinguish between more than just HSA and unknown.
This also lets us stop using unknown everywhere.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@260790 91177308-0d34-0410-b5e6-96231b3b80d8
These provide direct access to the hardware instruction without
the unit version required like llvm.sin/llvm.cos lowering requires.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@260782 91177308-0d34-0410-b5e6-96231b3b80d8
MSan adds a constructor to each translation unit that calls
__msan_init, and does nothing else. The idea is to run __msan_init
before any instrumented code. This results in multiple constructors
and multiple .init_array entries in the final binary, one per
translation unit. This is absolutely unnecessary; one would be
enough.
This change moves the constructors to a comdat group in order to drop
the extra ones.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@260632 91177308-0d34-0410-b5e6-96231b3b80d8
