This can occur when one of the inputs to the multiply is loop invariant. Though my test cases just use two basic blocks with an unconditional jump which we won't merge until after isel in the codegen pipeline.
For scalars, I believe SelectionDAGBuilder can add an AssertZExt to pass knowledge across basic blocks but its explicitly disabled for vectors.
llvm-svn: 347266
This works if DAG combiner is enabled, but without combining
we cannot select scalar_to_vector of <2 x half> and <2 x i16>.
Differential Revision: https://reviews.llvm.org/D54718
llvm-svn: 347259
Add methods to BasicBlock which make it easier to efficiently check
whether a block has N (or more) predecessors.
This can be more efficient than using pred_size(), which is a linear
time operation.
We might consider adding similar methods for successors. I haven't done
so in this patch because succ_size() is already O(1).
With this patch applied, I measured a 0.065% compile-time reduction in
user time for running `opt -O3` on the sqlite3 amalgamation (30 trials).
The change in mergeStoreIntoSuccessor alone saves 45 million linked list
iterations in a stage2 Release build of llc.
See llvm.org/PR39702 for a harder but more general way of achieving
similar results.
Differential Revision: https://reviews.llvm.org/D54686
llvm-svn: 347256
Consistently use (!LegalOperations || isOperationLegalOrCustom) for all node pairs.
Differential Revision: https://reviews.llvm.org/D53478
llvm-svn: 347255
Otherwise, the clang analyzer tests fail on Windows when attempting to
unpickle AnalyzerTest objects in the worker processes. The pattern of,
add to path, import, remove from path, serialize, deserialize, doesn't
work. Once something gets added to the path, if we want to move it
across the wire for multiprocessing, we need to keep the module on
sys.path.
llvm-svn: 347254
As discussed on D53794, for float types with ranges smaller than the destination integer type, then we should be able to just use a regular FP_TO_SINT opcode.
I thought we'd need to provide MSA test cases for very small integer types as well (fp16 -> i8 etc.), but it turns out that promotion will kick in so they're unnecessary.
Differential Revision: https://reviews.llvm.org/D54703
llvm-svn: 347251
We're seeing some issues internally where we sent some intrinsics into the cost model that the getTypeLegalizationCost call fails on, but X86 specific tables don't care about. Our base class implementation takes care of them. We'd just like X86 backend to ignore them.
This patch makes sure the switch returned something X86 cares about and skips the table lookups and type legalization call if not. Probably more efficient too since we don't go scanning the tables for every intrinsic we could possibly see.
Differential Revision: https://reviews.llvm.org/D54711
llvm-svn: 347248
SSE PSHUFB vector ctlz lowering works at the i4 nibble level. As detailed in PR39703, we were masking the lower nibble off but we only actually use it in the case where the upper nibble is known to be zero, making it safe to remove the mask and save an instruction.
Differential Revision: https://reviews.llvm.org/D54707
llvm-svn: 347242
Previously we split the vectors in half to allow the two halves to be any extended then concatenated the results back together.
This patch instead instead extends the v16i8 sse algorithm to extend half of each 128-bit lane using punpcklbw/punpckhbw. Multiplies all the low half lanes and high half lanes together in separate operations. Then merges the half lane results back together using packuswb.
Unfortunately, some of the cases in vector-reduce-mul.ll regress because we aren't narrowing the vector width of the multiplies as we reduce. The splitting was somewhat making up for that before by causing halves to be discarded after the split.
Differential Revision: https://reviews.llvm.org/D54668
llvm-svn: 347240
This will hold flags specific to subprograms. In the future
we could potentially free up scarce bits in DIFlags by moving
subprogram-specific flags from there to the new flags word.
This patch does not change IR/bitcode formats, that will be
done in a follow-up.
Differential Revision: https://reviews.llvm.org/D54597
llvm-svn: 347239
This allows to avoid scratch use or indirect VGPR addressing for
small vectors.
Differential Revision: https://reviews.llvm.org/D54606
llvm-svn: 347231
MachOObjectFile::getHostArch() returns a temporary, and getArchName
returns a StringRef pointing to a temporary std::string.
No tests since it doesn't trigger any errors except with the sanitizers.
llvm-svn: 347230
Summary:
This makes it easier/cleaner to generate a single signature from
this directive. Also:
- Adds the symbol name, such that we don't depend on the location
of this directive anymore.
- Actually constructs the signature in the assembler, and make the
assembler own it.
- Refactor the use of MVT vs ValType in the streamer and assembler
to require less conversions overall.
- Changed 700 or so tests to use it.
Reviewers: sbc100, dschuff
Subscribers: jgravelle-google, eraman, aheejin, sunfish, jfb, llvm-commits
Differential Revision: https://reviews.llvm.org/D54652
llvm-svn: 347228
Summary:
AMDGPUAsmPrinter has a getSTI function that derives a GCNSubtarget from the
TM. However, this means that overridden target features are not detected and can
result in incorrect behaviour.
Switch to using STM which is a GCNSubtarget derived from the MF (used elsewhere
in the same function).
Change-Id: Ib6328ad667b7fcdc87e9c06344e59859207db9b0
Subscribers: arsenm, kzhuravl, jvesely, wdng, nhaehnle, yaxunl, tpr, t-tye, llvm-commits
Differential Revision: https://reviews.llvm.org/D54301
llvm-svn: 347221
Summary:
Currently, when vectorizing stores to uniform addresses, the only
instance we prevent vectorization is if there are multiple stores to the
same uniform address causing an unsafe dependency.
This patch teaches LAA to avoid vectorizing loops that have an unsafe
cross-iteration dependency between a load and a store to the same uniform address.
Fixes PR39653.
Reviewers: Ayal, efriedma
Subscribers: rkruppe, llvm-commits
Differential Revision: https://reviews.llvm.org/D54538
llvm-svn: 347220
Recently I tried to port LLDB's lit configuration files over to use a
on the surface, but broke some cases that weren't broken before and also
exposed some additional problems with the old approach that we were just
getting lucky with.
When we set up a lit environment, the goal is to make it as hermetic as
possible. We should not be relying on PATH and enabling the use of
arbitrary shell commands. Instead, only whitelisted commands should be
allowed. These are, generally speaking, the lit builtins such as echo,
cd, etc, as well as anything for which substitutions have been
explicitly set up for. These substitutions should map to the build
output directory, but in some cases it's useful to be able to override
this (for example to point to an installed tools directory).
This is, of course, how it's supposed to work. What was actually
happening is that we were bringing in PATH and LD_LIBRARY_PATH and then
just running the given run line as a shell command. This led to problems
such as finding the wrong version of clang-cl on PATH since it wasn't
even a substitution, and flakiness / non-determinism since the
environment the tests were running in would change per-machine. On the
other hand, it also made other things possible. For example, we had some
tests that were explicitly running cl.exe and link.exe instead of
clang-cl and lld-link and the only reason it worked at all is because it
was finding them on PATH. Unfortunately we can't entirely get rid of
these tests, because they support a few things in debug info that
clang-cl and lld-link don't (notably, the LF_UDT_MOD_SRC_LINE record
which makes some of the tests fail.
The high level changes introduced in this patch are:
1. Removal of functionality - The lit test suite no longer respects
LLDB_TEST_C_COMPILER and LLDB_TEST_CXX_COMPILER. This means there is no
more support for gcc, but nobody was using this anyway (note: The
functionality is still there for the dotest suite, just not the lit test
suite). There is no longer a single substitution %cxx and %cc which maps
to <arbitrary-compiler>, you now explicitly specify the compiler with a
substitution like %clang or %clangxx or %clang_cl. We can revisit this
in the future when someone needs gcc.
2. Introduction of the LLDB_LIT_TOOLS_DIR directory. This does in spirit
what LLDB_TEST_C_COMPILER and LLDB_TEST_CXX_COMPILER used to do, but now
more friendly. If this is not specified, all tools are expected to be
the just-built tools. If it is specified, the tools which are not
themselves being tested but are being used to construct and run checks
(e.g. clang, FileCheck, llvm-mc, etc) will be searched for in this
directory first, then the build output directory.
3. Changes to core llvm lit files. The use_lld() and use_clang()
functions were introduced long ago in anticipation of using them in
lldb, but since they were never actually used anywhere but their
respective problems, there were some issues to be resolved regarding
generality and ability to use them outside their project.
4. Changes to .test files - These are all just replacing things like
clang-cl with %clang_cl and %cxx with %clangxx, etc.
5. Changes to lit.cfg.py - Previously we would load up some system
environment variables and then add some new things to them. Then do a
bunch of work building out our own substitutions. First, we delete the
system environment variable code, making the environment hermetic. Then,
we refactor the substitution logic into two separate helper functions,
one which sets up substitutions for the tools we want to test (which
must come from the build output directory), and another which sets up
substitutions for support tools (like compilers, etc).
6. New substitutions for MSVC -- Previously we relied on location of
MSVC by bringing in the entire parent's PATH and letting
subprocess.Popen just run the command line. Now we set up real
substitutions that should have the same effect. We use PATH to find
them, and then look for INCLUDE and LIB to construct a substitution
command line with appropriate /I and /LIBPATH: arguments. The nice thing
about this is that it opens the door to having separate %msvc-cl32 and
%msvc-cl64 substitutions, rather than only requiring the user to run
vcvars first. Because we can deduce the path to 32-bit libraries from
64-bit library directories, and vice versa. Without these substitutions
this would have been impossible.
Differential Revision: https://reviews.llvm.org/D54567
llvm-svn: 347216
Legacy loop pass manager is issuing "Made Modification" message after each Loop Pass
run, however condition for issuing it is accumulated among all the runs.
That leads to confusing 'modification' messages as soon as the first modification is done.
Changing condition to be "current pass made modifications", similar to how
it is being done in all other pass managers.
llvm-svn: 347215
This patch defines an interleaved-load-combine pass. The pass searches
for ShuffleVector instructions that represent interleaved loads. Matches are
converted such that they will be captured by the InterleavedAccessPass.
The pass extends LLVMs capabilities to use target specific instruction
selection of interleaved load patterns (e.g.: ld4 on Aarch64
architectures).
Differential Revision: https://reviews.llvm.org/D52653
llvm-svn: 347208
Summary:
As it was pointed out in D54388+D54390, the maximal size of `Neighbors` is known,
it will contain at most Points_.size() minus one (the center of the cluster)
While that is the upper bound, meaning in the most cases, the actual count
will be much smaller, since D54390 made the allocation persistent,
we no longer have to worry about overly-optimistically `reserve()`ing.
Old: (D54393)
```
Performance counter stats for './bin/llvm-exegesis -mode=analysis -analysis-epsilon=100000 -benchmarks-file=/tmp/benchmarks.yaml -analysis-inconsistencies-output-file=/tmp/clusters.html' (16 runs):
6553.167456 task-clock (msec) # 1.000 CPUs utilized ( +- 0.21% )
...
6.5547 +- 0.0134 seconds time elapsed ( +- 0.20% )
```
New:
```
Performance counter stats for './bin/llvm-exegesis -mode=analysis -analysis-epsilon=100000 -benchmarks-file=/tmp/benchmarks.yaml -analysis-inconsistencies-output-file=/tmp/clusters.html' (16 runs):
6315.057872 task-clock (msec) # 0.999 CPUs utilized ( +- 0.24% )
...
6.3187 +- 0.0160 seconds time elapsed ( +- 0.25% )
```
And that is another -~4%.
Since this is the last (as of this moment) patch in this patch series,
it is a good time to summarize:
Old: (svn trunk, as stated in D54381)
```
$ time ./bin/llvm-exegesis -mode=analysis -analysis-epsilon=100000 -benchmarks-file=/tmp/benchmarks.yaml -analysis-inconsistencies-output-file=/tmp/clusters.html &> /dev/null
real 0m24.884s
user 0m24.099s
sys 0m0.785s
```
So these patches, on a given benchmark,
has decreased llvm-exegesis analysis time by 74.62%.
There surely is more room for further improvements.
D54514 may improve thins by -11.5% more (relative to this patch).
Parallelization may improve things further significantly, too.
Reviewers: courbet, MaskRay, RKSimon, gchatelet, john.brawn
Reviewed By: courbet, MaskRay
Subscribers: tschuett, llvm-commits
Differential Revision: https://reviews.llvm.org/D54415
llvm-svn: 347204
Summary:
Test data: 500kLOC of benchmark.yaml, 23Mb. (that is a subset of the actual uops benchmark i was trying to analyze!)
Old time: (D54381)
```
$ time ./bin/llvm-exegesis -mode=analysis -analysis-epsilon=100000 -benchmarks-file=/tmp/benchmarks.yaml -analysis-inconsistencies-output-file=/tmp/clusters.html &> /dev/null
real 0m10.487s
user 0m9.745s
sys 0m0.740s
```
New time:
```
$ time ./bin/llvm-exegesis -mode=analysis -analysis-epsilon=100000 -benchmarks-file=/tmp/benchmarks.yaml -analysis-inconsistencies-output-file=/tmp/clusters.html &> /dev/null
real 0m9.599s
user 0m8.824s
sys 0m0.772s
```
Not that much, around -9%. But that is not the good part yet, again.
Old:
* calls to allocation functions: 3347676
* temporary allocations: 277818
* bytes allocated in total (ignoring deallocations): 10.52 GB
New:
* calls to allocation functions: 2109712 (-36%)
* temporary allocations: 33112 (-88%)
* bytes allocated in total (ignoring deallocations): 4.43 GB (-58% *sic*)
Reviewers: courbet, MaskRay, RKSimon, gchatelet, john.brawn
Reviewed By: courbet, MaskRay
Subscribers: tschuett, llvm-commits
Differential Revision: https://reviews.llvm.org/D54382
llvm-svn: 347198
Summary:
Test data: 500kLOC of benchmark.yaml, 23Mb. (that is a subset of the actual uops benchmark i was trying to analyze!)
Old time:
```
$ time ./bin/llvm-exegesis -mode=analysis -analysis-epsilon=100000 -benchmarks-file=/tmp/benchmarks.yaml -analysis-inconsistencies-output-file=/tmp/clusters.html &> /dev/null
real 0m24.884s
user 0m24.099s
sys 0m0.785s
```
New time:
```
$ time ./bin/llvm-exegesis -mode=analysis -analysis-epsilon=100000 -benchmarks-file=/tmp/benchmarks.yaml -analysis-inconsistencies-output-file=/tmp/clusters.html &> /dev/null
real 0m10.469s
user 0m9.797s
sys 0m0.672s
```
So -60%. And that isn't the good bit yet.
Old:
* calls to allocation functions: 106560180 (yes, 107 *million* allocations.)
* bytes allocated in total (ignoring deallocations): 12.17 GB
New:
* calls to allocation functions: 3347676 (-96.86%) (just 3 mil)
* bytes allocated in total (ignoring deallocations): 10.52 GB (~2GB less)
---
Two points i want to raise:
* `std::unordered_set<>` should not have been used there in the first place.
It is banned by the https://llvm.org/docs/ProgrammersManual.html#other-set-like-container-options
* There is no tests, so i'm not fully sure this is correct.
Since it was unordered set, i guess there are zero restrictions on the order, and anything will be ok?
* I tried other containers suggested in https://llvm.org/docs/ProgrammersManual.html#set-like-containers-std-set-smallset-setvector-etc,
this `llvm::SetVector<>` seems to be best here.
Reviewers: courbet, MaskRay, RKSimon, gchatelet, john.brawn
Reviewed By: courbet
Subscribers: kristina, bobsayshilol, tschuett, llvm-commits
Differential Revision: https://reviews.llvm.org/D54381
llvm-svn: 347197
Truncs are treated as sources if their produce a value of the same
type as the one we currently trying to promote. Truncs used to be
considered as a sink if their operand was the same value type.
We now allow smaller types in the search, so we should search through
truncs that produce a smaller value. These truncs can then be
converted to an AND mask.
This leaves sinks as being:
- points where the value in the register is being observed, such as
an icmp, switch or store.
- points where value types have to match, such as calls and returns.
- zext are included to ease the transformation and are generally
removed later on.
During this change, it also became apart from truncating sinks was
broken: if a sink used a source, its type information had already
been lost by the time the truncation happens. So I've changed the
method of caching the type information.
Differential Revision: https://reviews.llvm.org/D54515
llvm-svn: 347191
The general approach taken is to make note of loop invariant branches, then when
we see something conditional on that branch, such as a phi, we create a copy of
the branch and (empty versions of) its successors and hoist using that.
This has no impact by itself that I've been able to see, as LICM typically
doesn't see such phis as they will have been converted into selects by the time
LICM is run, but once we start doing phi-to-select conversion later it will be
important.
Differential Revision: https://reviews.llvm.org/D52827
llvm-svn: 347190
There is no variable-length shifts on MSP430. Therefore
"eat" 8 bits of shift via bswap & ext.
Path by Kristina Bessonova!
Differential Revision: https://reviews.llvm.org/D54623
llvm-svn: 347187
The shift requires a copy to avoid clobbering a register. Comparing with 0 uses an xor to produce 0 that will be overwritten with the compare results. So still requires 2 instructions, but should be one byte shorter since it doesn't need to encode an immediate.
llvm-svn: 347185
This patch introduces infrastructure and the simplest case for constant-folding
of branch and switch instructions within loop into unconditional branches.
It is useful as a cleanup for such passes as loop unswitching that sometimes
produce such branches.
Only the simplest case supported in this patch: after the folding, no block
should become dead or stop being part of the loop. Support for more
sophisticated cases will go separately in follow-up patches.
Differential Revision: https://reviews.llvm.org/D54021
Reviewed By: anna
llvm-svn: 347183
Every Analysis pass has a get method that returns a reference of the Result of
the Analysis, for example, BlockFrequencyInfo
&BlockFrequencyInfoWrapperPass::getBFI(). I believe that
ProfileSummaryInfo::getPSI() is the only exception to that, as it was returning
a pointer.
Another change is renaming isHotBB and isColdBB to isHotBlock and isColdBlock,
respectively. Most methods use BB as the argument of variable names while
methods usually refer to Basic Blocks as Blocks, instead of BB. For example,
Function::getEntryBlock, Loop:getExitBlock, etc.
I also fixed one of the comments.
Patch by Rodrigo Caetano Rocha!
Differential Revision: https://reviews.llvm.org/D54669
llvm-svn: 347182
Previously we used an arithmetic shift right by 31, but that requires a copy to preserve the input. So we might as well materialize a zero and compare to it since the comparison will overwrite the register that contains the zeros. This should be one byte shorter.
llvm-svn: 347181
Leave just the v4i8->v4i64 and v8i8->v8i64, but only enable them on pre-sse4.1 targets when 64-bit mode is enabled. In those cases we end up creating sext loads that get scalarized to code that looks better than what we get from loading into a vector register and doing a multiple step sign extend using unpacks and shifts.
llvm-svn: 347180
Pre-SSE4.1 sext_invec for v2i64 is complicated because we don't have a v2i64 sra instruction. So instead we sign extend to i32 using unpack and sra, then copy the elements and do a v4i32 sra to fill with sign bits, then interleave the i32 sign extend and the sign bits. So really we're doing to two sign extends but only using half of the v4i32 intermediate result.
When the result is more than 128 bits, default type legalization would prefer to split the destination type all the way down to v2i64 with shuffles followed by v16i8/v8i16->v2i64 sext_inreg operations. This results in more instructions than necessary because we are only utilizing the lower 2 elements of the v4i32 intermediate result. Instead we can custom split a v4i8/v4i16->v4i64 sign_extend. Then we can sign extend v4i8/v4i16->v4i32 invec producing a full v4i32 result. Create the sign bit vector as a v4i32 then split and interleave with the sign bits using an punpackldq and punpackhdq.
llvm-svn: 347176
Some of these sequeces look pretty bad since we have to copy the sign bit from a 32 bit register to a 64 bit register to finish a sign extend.
llvm-svn: 347175
If we widen illegal types instead of promoting, we should be able to rely on the type legalizer to create the vector_inreg operations for us with some caveats.
This patch disables combineToExtendVectorInReg when we are using widening.
I've enabled custom legalization for v8i8->v8i64 extends under avx512f since the type legalizer would want to create a vector_inreg with a v64i8 input type which isn't legal without avx512bw. So we go to v16i8 with custom code using the relaxation of rules we get from D54346.
I've also enable custom legalization of v8i64 and v16i32 operations with with AVX. When the input type is 128 bits, the default splitting legalization would extend first 128->256, then do the a split to two 128 pieces. Extend each half to 256 and then concat the result. The custom legalization I've added instead uses a 128->256 bit vector_inreg extend that only reads the lower 64-bits for the low half of the split. Then shuffles the high 64-bits to the low 64-bits and does another vector_inreg extend.
llvm-svn: 347172
Summary: This is an improvement over the two pshufbs and punpcklqdq we'd get otherwise.
Reviewers: RKSimon, spatel
Reviewed By: RKSimon
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D54671
llvm-svn: 347171
Sadly, this duplicates (twice) the logic from InstSimplify. There
might be some way to at least share the DAG versions of the code,
but copying the folds seems to be the standard method to ensure
that we don't miss these folds.
Unlike in IR, we don't run DAGCombiner to fixpoint, so there's no
way to ensure that we do these kinds of simplifications unless the
code is repeated at node creation time and during combines.
There were other tests that would become worthless with this
improvement that I changed as pre-commits:
rL347161
rL347164
rL347165
rL347166
rL347167
I'm not sure how to salvage the remaining tests (diffs in this patch).
So the x86 tests verify that the new code is working as intended.
The AMDGPU test is actually similar to my motivating case: we have
some undef value that has survived to machine IR in an x86 test, and
then it gets folded in some weird way, or we crash if we don't transfer
the undef flag. But we would have been better off never getting to that
point by doing these simplifications.
This will lead back to PR32023 someday...
https://bugs.llvm.org/show_bug.cgi?id=32023
llvm-svn: 347170
Refactor towards making this recursive (necessary for PR38243 rotation splat detection).
IsSplatVector returns the original vector source of the splat and the splat index.
GetSplatValue returns the scalar splatted value as an extraction from IsSplatVector.
llvm-svn: 347168
We were using the 'normalized' shuffle mask from resolveTargetShuffleInputs, which replaces zero/undef inputs with sentinel values. For SimplifyDemandedVectorElts we need the raw mask so we can correctly demand those 'zero' inputs that got normalized away, this requires an extra bit of logic to locally normalize undef inputs.
llvm-svn: 347158
Summary:
NetBSD ships with native curses(3) and -ltinfo is a part of ncurses.
Set -lterminfo before -ltinfo, as it allows to prioritize native curses
libraries. Mixing curses and ncurses does not work well, especially
in software built on top of llvm.
Original patch by Ryo Onodera (NetBSD) in pkgsrc.
Reviewers: labath, dim, mgorny
Reviewed By: dim, mgorny
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D54650
llvm-svn: 347156
Summary:
This was missing in D54096. Independent tests for this is not available
here, because these are used in lld.
Reviewers: sbc100
Subscribers: dschuff, jgravelle-google, sunfish, llvm-commits
Differential Revision: https://reviews.llvm.org/D54662
llvm-svn: 347154
The zero extend will require two stages of unpacks to implement. So its better to shrink the multiply using pmullw and then extend that result back to v4i32 using a single unpack.
llvm-svn: 347149
Fix all of the missing debug location errors in CVP found by debugify.
This includes the missing-location-after-udiv-truncation case described
in llvm.org/PR38178.
llvm-svn: 347147
The #if check around the statistics computation gave an error about
the statistic being an unused variable. Instead, guard with
AreStatisticsEnabled().
llvm-svn: 347146
Summary:
Follow up to D49362 ([ThinLTO] Internalize read only globals). Add a
statistic on the number of read only variables (only counting live
variables since dead variables will be dropped anyway).
Reviewers: evgeny777
Subscribers: mehdi_amini, inglorion, eraman, steven_wu, dexonsmith, arphaman, llvm-commits
Differential Revision: https://reviews.llvm.org/D54642
llvm-svn: 347145
Especially for symbolizer it can be efficient to have to search through
the entire index when it isn't needed - llvm-symbolizer looks up only a
few CUs & already has an index available in getUnitForEntry, once it's
passed down to DWARFUnitHeader::extract then there's no need for it to
call getFromOffset.
llvm-svn: 347134
See "GN build roundtable summary; adding GN build files to the repo" on
llvm-dev and cfe-dev for discussion.
In particular, this build is completely unsupported. People adding new files to
LLVM are not expected to update the GN build files, and reviewers are not
supposed to request the gn build files to be updated.
This adds just enough to be able to build llvm/lib/Demangle. It requires using
a monorepo.
This adds a few build config options you can set in args.gn
(`gn args out/foo --list` for all):
- is_debug = true to enable debug builds (defaults to release)
- llvm_enable_assertions to toggle assertions (defaults to true)
- clang_base_path, if set an absolute path to a locally-built clang to be used
as host compiler
Differential Revision: https://reviews.llvm.org/D54345
llvm-svn: 347128
Comments in llc outputs are printed differently on different
platforms, some with '#', some with '##'. Removed non-essential
part of the checks.
llvm-svn: 347112
This tries to force the result type to vXi32 followed by a truncate. This can help avoid scalarization that would otherwise occur.
There's some annoying examples of an avx512 truncate instruction followed by a packus where we should really be able to just use one truncate. But overall this is still a net improvement.
llvm-svn: 347105
Also, support modifications to toplevel files in git (which need to be
committed to "monorepo-root" in svn).
Differential Revision: https://reviews.llvm.org/D54341
llvm-svn: 347103
In one case probably you have be using it, in the other it
looks like it was redundant.
Differential Revision: https://reviews.llvm.org/D54644
llvm-svn: 347098
We need to control exponential behavior of loop-unswitch so we do not get
run-away compilation.
Suggested solution is to introduce a multiplier for an unswitch cost that
makes cost prohibitive as soon as there are too many candidates and too
many sibling loops (meaning we have already started duplicating loops
by unswitching).
It does solve the currently known problem with compile-time degradation
(PR 39544).
Tests are built on top of a recently implemented CHECK-COUNT-<num>
FileCheck directives.
Reviewed By: chandlerc, mkazantsev
Differential Revision: https://reviews.llvm.org/D54223
llvm-svn: 347097
For example, on X86 we emit a sign_extend_vector_inreg from LowerLoad and without sse4.1 this node will need further legalization. Previously this sign_extend_vector_inreg was being custom lowered during DAG legalization instead of vector op legalization.
Unfortunately, this doesn't seem to matter for the output of any existing lit tests.
llvm-svn: 347094
This NFC patch just adds test cases for conversions that currently
require scalarization of vectors. An updcoming patch will change
the legalization for these and it is more suitable on the review
to show the diferences in code gen rather than just the new code gen.
llvm-svn: 347090
When unwinding past a function that uses shadow call stack, we must
subtract 8 from the value of the x18 register. This patch causes us
to emit a call frame instruction that causes that to happen.
Differential Revision: https://reviews.llvm.org/D54609
llvm-svn: 347089
Summary:
Experience has shown that the functionality is useful. It makes linking
optimized clang with debug info for me a lot faster, 20s to 13s. The
type merging phase of PDB writing goes from 10s to 3s.
This removes the LLVM cl::opt and replaces it with a metadata flag.
After this change, users can do the following to use ghash:
- add -gcodeview-ghash to compiler flags
- replace /DEBUG with /DEBUG:GHASH in linker flags
Reviewers: zturner, hans, thakis, takuto.ikuta
Subscribers: aprantl, hiraditya, JDevlieghere, llvm-commits
Differential Revision: https://reviews.llvm.org/D54370
llvm-svn: 347072
Summary: Allow code-signing with entitlements. FORCE may be used to avoid an error when replacing existing signatures.
Reviewers: beanz, bogner
Reviewed By: beanz
Subscribers: mgorny, llvm-commits, lldb-commits
Differential Revision: https://reviews.llvm.org/D54443
llvm-svn: 347068
Summary:
This is for supporting custom stack map formats, where the
custom printer can access the stack map data.
Patch by Cherry Zhang <cherryyz@google.com>.
Related: https://reviews.llvm.org/D53892
Reviewers: thanm, apilipenko
Reviewed By: apilipenko
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D54224
llvm-svn: 347061
Summary:
StructRet attribute is not allowed in vararg calls. The statepoint
intrinsic is vararg, but the wrapped function may be not. Allow
calls of statepoint with StructRet arg, as long as the wrapped
function is not vararg.
Reviewers: thanm, anna
Reviewed By: anna
Subscribers: anna, llvm-commits
Differential Revision: https://reviews.llvm.org/D53602
llvm-svn: 347050
This is a follow-up to r346715. Use PRIx64 to formatted print of 64-bit
value in the `DWARFDebugLoclists::LocationList::dump` to escape problem
on big-endian hosts.
llvm-svn: 347049
Summary:
As discussed in previous review, and noted in the FIXME, if `X` is actually an `lshr Y, Z` (logical!),
we can fold the `Z` into 'control`, and let the `BEXTR` do this too.
We could just insert those 8 bits of shift amount into control,
but it is better to instead zero-extend them, and 'or' them in place.
We can only do this for `lshr`, not `ashr`, because we do not know that the mask cover only the bits of `Y`,
and not any of the sign-extended bits.
The obvious question is, is this actually legal to do?
I believe it is. Relevant quotes, from `Intel® 64 and IA-32 Architectures Software Developer’s Manual`, `BEXTR — Bit Field Extract`:
* `Bit 7:0 of the second source operand specifies the starting bit position of bit extraction.`
* `A START value exceeding the operand size will not extract any bits from the second source operand.`
* `Only bit positions up to (OperandSize -1) of the first source operand are extracted.`
* `All higher order bits in the destination operand (starting at bit position LENGTH) are zeroed.`
* `The destination register is cleared if no bits are extracted.`
FIXME: if we can do this, i wonder if we should prefer `BEXTR` over `BZHI` in such cases.
Reviewers: RKSimon, craig.topper, spatel, andreadb
Reviewed By: RKSimon, craig.topper, andreadb
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D54095
llvm-svn: 347048
The RISC-V ISA manual was updated on 2018-11-07 (commit 00557c3) to define a
new compressed instruction format, RVC format CA (no actual instruction
encodings were changed). This patch updates the RISC-V backend to define the
new format, and to use it in the relevant instructions.
Differential Revision: https://reviews.llvm.org/D54302
Patch by Luís Marques.
llvm-svn: 347043
This commit introduces support for materialising 64-bit constants for RV64I,
making use of the RISCVMatInt::generateInstSeq helper in order to share logic
for immediate materialisation with the MC layer (where it's used for the li
pseudoinstruction).
test/CodeGen/RISCV/imm.ll is updated to test RV64, and gains new 64-bit
constant tests. It would be preferable if anyext constant returns were sign
rather than zero extended (see PR39092). This patch simply adds an explicit
signext to the returns in imm.ll.
Further optimisations for constant materialisation are possible, most notably
for mask-like values which can be generated my loading -1 and shifting right.
A future patch will standardise on the C++ codepath for immediate selection on
RV32 as well as RV64, and then add further such optimisations to
RISCVMatInt::generateInstSeq in order to benefit both RV32 and RV64 for
codegen and li expansion.
Differential Revision: https://reviews.llvm.org/D52962
llvm-svn: 347042
Introduces support for '.refsym' assembler directive.
From GCC docs (for MSP430):
'.refsym' - This directive instructs assembler to add an undefined reference
to the symbol following the directive. No relocation is created for this symbol;
it will exist purely for pulling in object files from archives.
Patch by Kristina Bessonova!
Differential Revision: https://reviews.llvm.org/D54618
llvm-svn: 347041
An attempt to recommit r346584 after failure on OSX build bot.
Fixed cache key computation in ThinLTOCodeGenerator and added
test case
llvm-svn: 347033
By early promoting the multiply to use an i16 element type we can avoid op legalization emit a second multiply for the 8 upper elements of the v16i8 type we would otherwise get.
llvm-svn: 347032
If a block had one of the _term instructions used for gluing
exec modifying instructions to the end of the block,
analyzeBranch would fail, preventing the verifier from catching
a broken successor list.
llvm-svn: 347027
When using multi-stage builds, we would like support cross-compilation.
Example is 2-stage build when the first stage is compiled for host while
the second stage is compiled for the target.
Normally, the second stage would be also used for compiling runtimes,
but that's not possible when cross-compiling, so we use the first stage
compiler instead. However, we still want to use the second stage paths.
To do so, we set the -resource-dir of the first stage compiler to point
to the resource directory of the second stage.
We also need compiler tools that support the target architecture. These
tools are not guaranteed to be present on the host, but in case of
multi-stage build, we can build these tools in the first stage.
Differential Revision: https://reviews.llvm.org/D54461
llvm-svn: 347025
Originally we created our 64-bit UID scheme by using the first byte as
sort of a "tag" to represent what kind of symbol this was, and we
re-used the PDB_SymType enumeration for this. For native pdb support,
this is not really the right abstraction layer, because what we really
want is something that tells us *how* to find the symbol. This means,
specifically, is in the globals stream / public stream / module stream /
TPI stream / etc, and for whichever one it is in, where is it within
that stream?
A good example of why the old namespacing scheme was insufficient is
that it is more or less impossible to create a uid for a field list
member of a class/struction/union/enum that tells you how to locate
the original record.
With this new scheme, the first byte is no longer a PDB_SymType enum
but a new enum created specifically to identify where in the PDB
this record lives. This gives us much better flexibility in
what kinds of symbols the uids can identify.
llvm-svn: 347018
We aren't going to use the upper bits of the multiply result that the extend would effect. So we don't need a specific type of extend.
This makes some reduction test cases shorter because we were previously trying to sign_extend a truncate which we can't eliminate.
llvm-svn: 347011
It fixes the case when Objective-C framework is added as a subframework
through a symlink. When parent framework infers a module map and fails
to detect a symlink, it would add a subframework as a submodule. And
when we parse module map for the subframework, we would encounter an
error like
> error: umbrella for module 'WithSubframework.Foo' already covers this directory
By implementing `getRealPath` "an egregious but useful hack" in
`ModuleMap::inferFrameworkModule` works as expected.
rdar://problem/45821279
Reviewers: bruno, benlangmuir, erik.pilkington
Reviewed By: bruno
Subscribers: hiraditya, dexonsmith, JDevlieghere, cfe-commits, llvm-commits
Differential Revision: https://reviews.llvm.org/D54245
llvm-svn: 347009
Add a pass to fixup various vector ISel issues.
Currently we handle converting GLOBAL_{LOAD|STORE}_*
and GLOBAL_Atomic_* instructions into their _SADDR variants.
This involves feeding the sreg into the saddr field of the new instruction.
llvm-svn: 347008
Summary:
`throw` instruction is a terminator in wasm, but BBs were not splitted
after `throw` instructions, causing machine instruction verifier to
fail.
This patch
- Splits BBs after `throw` instructions in WasmEHPrepare and adding an
unreachable instruction after `throw`, which will be deleted in
LateEHPrepare pass
- Refactors WasmEHPrepare into two member functions
- Changes the semantics of `eraseBBsAndChildren` in LateEHPrepare pass
to match that of WasmEHPrepare pass, which is newly added. Now
`eraseBBsAndChildren` does not delete BBs with remaining predecessors.
- Fixes style nits, making static function names conform to clang-tidy
- Re-enables the test temporarily disabled by rL346840 && rL346845
Reviewers: dschuff
Subscribers: sbc100, jgravelle-google, sunfish, llvm-commits
Differential Revision: https://reviews.llvm.org/D54571
llvm-svn: 347003
Removing this code doesn't affect any lit tests so it doesn't appear to be tested anymore. I assume it was when it was added, but I guess something else changed? Code coverage report also says its unused.
I mostly didn't like that it seemed to count the sign bits as if it was a sign_extend, but then set isPositive as if it was a zero_extend. It feels like we should have picked one interpretation?
Differential Revision: https://reviews.llvm.org/D54596
llvm-svn: 346995
* Add amdhsa prefix to names to allow other tools to use the metadata
without collision.
* Make names consistent.
* Simplify structure.
* Change note record ID.
* Switch from YAML to MsgPack format.
* Document metadata assembler directive.
Patch By: t-tye (Tony Tye)
Differential Revision: https://reviews.llvm.org/D53445
llvm-svn: 346992
Summary:
This fixes libLLVM.so ABI mismatches between llvm compiled with clang
and llvm compiled with gcc (PR39427).
Reviewers: bkramer, sylvestre.ledru, mgorny, hans
Reviewed By: bkramer, hans
Subscribers: dexonsmith, kristina, llvm-commits
Differential Revision: https://reviews.llvm.org/D54540
llvm-svn: 346985
Use unsigned to calculate the subvector index to avoid a cast.
Remove an unnecessary condition and replace it with a stronger assert.
Use the InVT variable we updated when we extracted instead of grabbing it from the In SDValue.
llvm-svn: 346983
In reduceVMULWidth, we no longer need to worry about extending the vector to 128 bits first. Regular widening of extends, muls and shuffles will take care of that for us.
In combineMulToPMADDWD, we can handle v2i32 multiplies and allow the VPMADDWD to be widened to v4i32 during type legalization by adding custom widening like we do have for AVG/ADDUS/SUBUS. I had to modify that code a little to allow different and output VTs.
Differential Revision: https://reviews.llvm.org/D54512
llvm-svn: 346980
Summary:
The old return type did not allow for correct error reporting and was
causing a compiler warning.
Reviewers: aheejin
Subscribers: dschuff, sbc100, jgravelle-google, sunfish, llvm-commits
Differential Revision: https://reviews.llvm.org/D54586
llvm-svn: 346979
Add data structure to represent MessagePack "documents" and convert
to/from both MessagePack and YAML encodings.
Differential Revision: https://reviews.llvm.org/D48175
llvm-svn: 346978
We were adding the entire scalarization extraction cost for reductions, which returns the total cost of extracting every element of a vector type.
For reductions we don't need to do this - we just need to extract the 0'th element after the reduction pattern has completed.
Fixes PR37731
Differential Revision: https://reviews.llvm.org/D54585
llvm-svn: 346970
There's a potential miscompile here. It's unlikely in the real
world because this transform is guarded with shouldChangeType(),
but this test file doesn't include a standard data-layout for
some reason (despite including a custom 1), so we can see the bug.
llvm-svn: 346966
This fixes -filetype=null support when compiling for a Win32 target and the module has a CodeView flag.
The only places changed are the uses of getTargetStreamer function - this patch guards both of them with null checks.
Committed on behalf of @eush (Eugene Sharygin)
Differential Revision: https://reviews.llvm.org/D54008
llvm-svn: 346962
This is a problem seen in common rotate idioms as noted in:
https://bugs.llvm.org/show_bug.cgi?id=34924
Note that we are not canonicalizing standard IR (shifts and logic) to the intrinsics yet.
(Although I've written this before...) I think this is the last step before we enable
that transform. Ie, we could regress code by doing that transform without this
simplification in place.
In PR34924, I questioned whether this is a valid transform for target-independent IR,
but I convinced myself this is ok. If we're speculating a funnel shift by turning cmp+br
into select, then SimplifyCFG has already determined that the transform is justified.
It's possible that SimplifyCFG is not taking into account profile or other metadata,
but if that's true, then it's a bug independent of funnel shifts.
Also, we do have CGP code to restore a guard like this around an intrinsic if it can't
be lowered cheaply. But that isn't necessary for funnel shift because the default
expansion in SelectionDAGBuilder includes this same cmp+select.
Differential Revision: https://reviews.llvm.org/D54552
llvm-svn: 346960
C.EBREAK was defined with hasSideEffects = 0, which is incorrect and
inconsistent with the non-compressed instruction form. This patch corrects
this oversight.
This wouldn't cause codegen issues, as compressed instructions are only ever
generated by converting the non-compressed form as an MCInst. But having
correct flags is still worthwhile.
Differential Revision: https://reviews.llvm.org/D54256
Patch by Luís Marques.
llvm-svn: 346959
Mark the FREM SelectionDAG node as Expand, which is necessary in order to
support the frem IR instruction on RISC-V. This is expanded into a library
call. Adds the corresponding test. Previously, this would have triggered an
assertion at instruction selection time.
Differential Revision: https://reviews.llvm.org/D54159
Patch by Luís Marques.
llvm-svn: 346958
Reapply r346374 with the fixes for modules build.
Original summary:
This change implements assembler parser, code emitter, ELF object writer
and disassembler for the MSP430 ISA. Also, more instruction forms are added
to the target description.
Patch by Michael Skvortsov!
llvm-svn: 346948
Summary:
According to `MaskRay`, use `auto` for type inference, according to coding standards.
Delete some comments, because these comments can be easily inferred from codes.
Reviewers: jhenderson, MaskRay
Reviewed By: jhenderson
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D54573
llvm-svn: 346946
Logic to load 32-bit and 64-bit immediates is currently present in
RISCVAsmParser::emitLoadImm in order to support the li pseudoinstruction. With
the introduction of RV64 codegen, there is a greater benefit of sharing
immediate materialisation logic between the MC layer and codegen. The
generateInstSeq helper allows this by producing a vector of simple structs
representing the chosen instructions. This can then be consumed in the MC
layer to produce MCInsts or at instruction selection time to produce
appropriate SelectionDAG node. Sharing this logic means that both the li
pseudoinstruction and codegen can benefit from future optimisations, and
that this logic can be used for materialising constants during RV64 codegen.
This patch does contain a behaviour change: addi will now be produced on RV64
when no lui is necessary to materialise the constant. In that case addiw takes
x0 as the source register, so is semantically identical to addi.
Differential Revision: https://reviews.llvm.org/D52961
llvm-svn: 346937
This avoids some nasty shuffles when we have avx512. It will also prevent using zmm truncate instructions when a ymm instruction that zeroes part of an xmm register will do. Also avoid using avx512 truncate instructions when the input is 128 bits or less. These instructions are 2 uops on skx so we can probably find a better single uop shuffle like pshufb.
llvm-svn: 346936
The narrow types end up requesting widening, but generic legalization will end up scalaring and using a build_vector to do the widening.
llvm-svn: 346916
Craig Topper