For constant non-uniform cases we'll never introduce more and/andn/or selects than already occur in generic pre-SSE41 ISD::SRL lowering.
llvm-svn: 342352
This is a follow-up suggested in D51630 and originally proposed as an IR transform in D49040.
Copying the motivational statement by @evandro from that patch:
"This transformation helps some benchmarks in SPEC CPU2000 and CPU2006, such as 188.ammp,
447.dealII, 453.povray, and especially 300.twolf, as well as some proprietary benchmarks.
Otherwise, no regressions on x86-64 or A64."
I'm proposing to add only the minimum support for a DAG node here. Since we don't have an
LLVM IR intrinsic for cbrt, and there are no other DAG ways to create a FCBRT node yet, I
don't think we need to worry about DAG builder, legalization, a strict variant, etc. We
should be able to expand as needed when adding more functionality/transforms. For reference,
these are transform suggestions currently listed in SimplifyLibCalls.cpp:
// * cbrt(expN(X)) -> expN(x/3)
// * cbrt(sqrt(x)) -> pow(x,1/6)
// * cbrt(cbrt(x)) -> pow(x,1/9)
Also, given that we bail out on long double for now, there should not be any logical
differences between platforms (unless there's some platform out there that has pow()
but not cbrt()).
Differential Revision: https://reviews.llvm.org/D51753
llvm-svn: 342348
https://bugs.llvm.org/show_bug.cgi?id=38949
It's not clear to me that we even need a one-use check in this fold.
Ie, 2 independent loads might be better than a load+dependent shuffle.
Note that the existing re-use tests are not affected. We actually do form a
broadcast node in those tests now because there's no extra use of the
insert_subvector node in those cases. But something later in isel pattern
matching decides that it is not worth using a broadcast for the full load in
those tests:
Legalized selection DAG: %bb.0 'test_broadcast_2f64_4f64_reuse:'
t7: v2f64,ch = load<(load 16 from %ir.p0)> t0, t2, undef:i64
t4: i64,ch = CopyFromReg t0, Register:i64 %1
t10: ch = store<(store 16 into %ir.p1)> t7:1, t7, t4, undef:i64
t18: v4f64 = insert_subvector undef:v4f64, t7, Constant:i64<0>
t20: v4f64 = insert_subvector t18, t7, Constant:i64<2>
Becomes:
t7: v2f64,ch = load<(load 16 from %ir.p0)> t0, t2, undef:i64
t4: i64,ch = CopyFromReg t0, Register:i64 %1
t10: ch = store<(store 16 into %ir.p1)> t7:1, t7, t4, undef:i64
t21: v4f64 = X86ISD::SUBV_BROADCAST t7
ISEL: Starting selection on root node: t21: v4f64 = X86ISD::SUBV_BROADCAST t7
...
Created node: t27: v4f64 = INSERT_SUBREG IMPLICIT_DEF:v4f64, t7, TargetConstant:i32<7>
Morphed node: t21: v4f64 = VINSERTF128rr t27, t7, TargetConstant:i8<1>
llvm-svn: 342347
For people who use llvm-readelf as a replacement of GNU readelf, they would like to see -d -r ... listed in llvm-readelf -help. It also helps understanding the confusing -s (which is unfortunately different in semantics).
Reviewers: phosek, ruiu, echristo
Reviewed By: ruiu, echristo
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D52129
llvm-svn: 342339
Summary:
If the sub doesn't overflow in the original type we can move it above the sext/zext.
This is similar to what we do for add. The overflow checking for sub is currently weaker than add, so the test cases are constructed for what is supported.
Reviewers: spatel
Reviewed By: spatel
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D52075
llvm-svn: 342335
Naively computing the hash after the PDB data has been generated is in practice
as fast as other approaches I tried. I also tried online-computing the hash as
parts of the PDB were written out (https://reviews.llvm.org/D51887; that's also
where all the measuring data is) and computing the hash in parallel
(https://reviews.llvm.org/D51957). This approach here is simplest, without
being slower.
Differential Revision: https://reviews.llvm.org/D51956
llvm-svn: 342333
* Use same method of initializing the output stream and its buffer
* Allow a nullptr Status pointer
* Don't print the mangled name on demangling error
* Write to N (if it is non-nullptr)
Differential Revision: https://reviews.llvm.org/D52104
llvm-svn: 342330
The existing test was called extract_lane_insertps_5123 but it was in fact doing a <6,1,2,3> shuffle. I've fixed the name and added the <5,1,2,3> test case as well.
llvm-svn: 342328
Summary: This unfortunately adds a move, but isn't that better than going to the int domain and back?
Reviewers: RKSimon
Reviewed By: RKSimon
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D52134
llvm-svn: 342327
Summary:
MOVMSK only care about the sign bit so we don't need the setcc to fill the whole element with 0s/1s. We can just shift the bit we're looking for into the sign bit. This saves a constant pool load.
Inspired by PR38840.
Reviewers: RKSimon, spatel
Reviewed By: RKSimon
Subscribers: lebedev.ri, llvm-commits
Differential Revision: https://reviews.llvm.org/D52121
llvm-svn: 342326
A few changes found necessary for upcoming PassInstrumentation patch:
- name() methods made const
- properly forward arguments in AnalysisPassModel::run
Separated out of the main D47858 patch.
llvm-svn: 342325
Missing optimizations with blendv are shown in:
https://bugs.llvm.org/show_bug.cgi?id=38814
If this works, it's an easier and more powerful solution than adding pattern matching
for a few special cases in the backend. The potential danger with this transform in IR
is that the condition value can get separated from the select, and the backend might
not be able to make a blendv out of it again. I don't think that's too likely, but
I've kept this patch minimal with a 'TODO', so we can test that theory in the wild
before expanding the transform.
Differential Revision: https://reviews.llvm.org/D52059
llvm-svn: 342324
CodeGenPrepare has a transform that sinks {lshr, trunc} pairs to make it
easier for the backend to emit fancy extract-bits instructions (e.g UBFX).
Teach it to preserve debug locations and salvage debug values.
llvm-svn: 342319
Summary:
Implement shifts of vectors by i32. Since LLVM defines shifts as
binary operations between two vectors, this involves pattern matching
on splatted shift operands. For v2i64 shifts any i32 shift operands
have to be zero extended in the input and any i64 shift operands have
to be wrapped in the output. Depends on D52007.
Reviewers: aheejin, dschuff
Subscribers: sbc100, jgravelle-google, sunfish, llvm-commits
Differential Revision: https://reviews.llvm.org/D51906
llvm-svn: 342302
Before this fix, multiple invocations of testRoundTrip will create multiple
writers which share the same file as output destination. That could introduce
filesystem race issue when multiple subtests are executed concurrently. This
patch assign writers with different files as their output destinations.
llvm-svn: 342301
Currently if we got something like `const Foo` we'd ignore it and
just rely on printing the unmodified `Foo` later on. However,
for testing the native reading code we really would like to be able
to see these so that we can verify that the native reader can
actually handle them. Instead of printing out the full type though,
just print out the header.
llvm-svn: 342295
Similar to rL342278:
The test diffs are all cosmetic due to the change in
value naming, but I'm including that to show that the
new code does perform these folds rather than something
else in instcombine.
D52075 should be able to use this code too rather than
duplicating all of the logic.
llvm-svn: 342292
Summary:
Integer types smaller than i32 must be extended to i32 by default.
The feature "crbits" introduced at r202451 handles i1 as a special case,
but it did not extend properly.
The caller was, therefore, passing i1 stack arguments by writing 0/1 to
the first byte of the 4-byte stack object and callee was
reading the first byte for the value.
"crbits" is enabled if the optimization level is greater than 1,
which is very common in "release builds".
Such discrepancies with ABI specification also introduces
potential incompatibility with programs or libraries
built with other compilers e.g. GCC.
Fixes PR38661
Reviewers: hfinkel, cuviper
Subscribers: sylvestre.ledru, glaubitz, nagisa, nemanjai, kbarton, llvm-commits
Differential Revision: https://reviews.llvm.org/D51108
llvm-svn: 342288
Summary:
This change makes the tests more focused and avoids problematic
interactions between the testing modes and instruction encoding. This
change also allows the other tests to use less verbose output and
stricter checks.
Reviewers: aheejin, dschuff, aardappel
Subscribers: sbc100, jgravelle-google, sunfish, llvm-commits
Differential Revision: https://reviews.llvm.org/D52007
llvm-svn: 342287
Eventually we need to be able to support nested types, which don't
have an associated CVType record. To handle this, remove the
CVType from all of the record classes, and instead store the
deserialized record. Then move the deserialization up to the thing
that creates the type. This actually makes error handling better
anyway as we can return an invalid symbol instead of asserting false.
llvm-svn: 342284
The patch saves a function offset table which maps function name index to the
offset of its function profile to the start of the binary profile. By using
the function offset table, for those function profiles which will not be used
when compiling a module, the profile reader does't have to read them. For
profile size around 10~20M, it saves ~10% compile time.
Differential Revision: https://reviews.llvm.org/D51863
llvm-svn: 342283
This test constructs a non-readable file of mode 0111, which lingers in the test output directory and will cause EACCES to various tools (rg, rsync, ...)
llvm-svn: 342279
The test diffs are all cosmetic due to the change in
value naming, but I'm including that to show that the
new code does perform these folds rather than something
else in instcombine.
llvm-svn: 342278
The test used to fail with an invalid phi node: the two predecessors were outlined
and the SSA representation was left invalid. The patch adds the exit block to the
cold region.
llvm-svn: 342277
remove duplicate entries from isSingleEntrySingleExit: the Entry block is
already added by the loop over the dominance frontier.
Remove the heuristic from isOutlineCandidate that a region is too small when it
only contains a basic block. With this change we now grow regions starting from
a block and we continue adding to the ValidColdRegion. Check the heuristic just
before code generation.
llvm-svn: 342276
Also fix a problem in forward propagation:
const TerminatorInst *TI = It->getTerminator();
was set outside the while loop that iterates over It.
llvm-svn: 342275
Summary:
The multiprocess module uses pickling to transfer
information between processes and does not know how to pickle
the class created in the lit.cfg file and thus the example
fails.
Implement ManyTests in a separate file and import for the
example test passes
Patch by Nathan Lanza <nathan@lanza.io>
Differential Revision: https://reviews.llvm.org/D51328
llvm-svn: 342269
This patch fixes the debug info handling for SelectionDAG legalization
of DAG nodes with two results. When an replaced SDNode has more than
one result, transferDbgValues was always copying the SDDbgValue from
the first result and attaching them to all members. In reality
SelectionDAG::ReplaceAllUsesWith() is given an array of SDNodes
(though the type signature doesn't make this obvious (cf. the call
site code in ReplaceNode()).
rdar://problem/44162227
Differential Revision: https://reviews.llvm.org/D52112
llvm-svn: 342264
Summary:
During threaded thinLTO, it is possible that the entry for current
module doesn't exist in StringMaps (like ExportLists, ResolvedODR,
etc.). Using operator[] might trigger a rehash for the StringMap, which
might happen on multiple threads at the same time.
rdar://problem/43846199
Reviewers: tejohnson, mehdi_amini, kromanova, pcc
Reviewed By: tejohnson
Subscribers: dang, inglorion, eraman, dexonsmith, llvm-commits
Differential Revision: https://reviews.llvm.org/D52049
llvm-svn: 342263
Attempt to lower a shuffle as an unpack of elements from two inputs followed by a single-input (wider) permutation.
As long as the permutation is wider this is a win - there may be some circumstances where same size permutations would also be useful but I've left that for future work.
Differential Revision: https://reviews.llvm.org/D52043
llvm-svn: 342257
The folds are not limited to zext, and the real goal is width
reduction of a math op. D52075 is proposing to extend this to
subtracts.
llvm-svn: 342254
There was a bug in a check line regex that could cause the test to fail
with a naming difference. The auto-gen script seems to work as expected now.
llvm-svn: 342249
Using llvm::getInputFileDirectory() in unit tests is discouraged, so require an explicit opt-in.
This way, cmake also writes ~60 fewer unused files to disk.
Differential Revision: https://reviews.llvm.org/D52095
llvm-svn: 342248
Also added a C-interface function for large values, and updated
llvm-lto's --thinlto-cache-max-size-bytes switch to take a type larger
than int.
The maximum cache size in terms of bytes is a 64-bit number. However,
the methods to set it only took unsigned previously, which meant that
the maximum cache size could not be specified above 4GB. That's quite
small compared to the output of some projects, so it makes sense to
provide the ability to set larger values in that field.
We also needed a C-interface function that provides a greater range
than the existing thinlto_codegen_set_cache_size_bytes, which also only
takes an unsigned, so this change also adds
hinlto_codegen_set_cache_size_megabytes.
Reviewed by: mehdi_amini, tejohnson, steven_wu
Differential Revision: https://reviews.llvm.org/D52023
llvm-svn: 342233
Summary:
GFX9 and above support sin/cos instructions with a greater range and thus don't
require a fract instruction prior to invocation.
Added a subtarget feature to reflect this and added code to take advantage of
expanded range on GFX9+
Also updated the tests to check correct behaviour
Subscribers: arsenm, kzhuravl, jvesely, wdng, nhaehnle, yaxunl, tpr, t-tye, llvm-commits
Differential Revision: https://reviews.llvm.org/D51933
Change-Id: I1c1f1d3726a5ae32116646ca5cfa1ab4ef69e5b0
llvm-svn: 342222
On failing to find sequences that can be converted into dual macs,
try to find sequential 16-bit loads that are used by muls which we
can then use smultb, smulbt, smultt with a wide load.
Differential Revision: https://reviews.llvm.org/D51983
llvm-svn: 342210
As preparation for LoopInterchange becoming a loop pass, it needs to
preserve ScalarEvolution. Even though interchanging should not change
the trip count of the loop, it modifies loop entry, latch and exit
blocks.
I added -verify-scev to some loop interchange tests, but the verification does
not catch problems caused by missing invalidation of SE in loop interchange, as
the trip counts themselves do not change. So there might be potential to
make the SE verification covering more stuff in the future.
Reviewers: mkazantsev, efriedma, karthikthecool
Reviewed By: efriedma
Differential Revision: https://reviews.llvm.org/D52026
llvm-svn: 342209
After recent improvements which makes better use of LOC instead of IPM, the
TTI cost functions also needs to be updated to reflect this.
This involves sext, zext and xor of i1.
The tests were updated so that for z13 the new costs are expected, while the
old costs are still checked for on zEC12.
Review: Ulrich Weigand
https://reviews.llvm.org/D51339
llvm-svn: 342207
When reading directives from a .drectve section, the directives are
tokenized as a normal windows command line. However in these cases,
link.exe allows the directives to be separated by null bytes, not only by
spaces.
A test case for this change will be added in the lld repo.
Differential Revision: https://reviews.llvm.org/D52014
llvm-svn: 342204
Summary:
[VPlan] Implement vector code generation support for simple outer loops.
Context: Patch Series #1 for outer loop vectorization support in LV using VPlan. (RFC: http://lists.llvm.org/pipermail/llvm-dev/2017-December/119523.html).
This patch introduces vector code generation support for simple outer loops that are currently supported in the VPlanNativePath. Changes here essentially do the following:
- force vector code generation using explicit vectorize_width
- add conservative early returns in cost model and other places for VPlanNativePath
- add code for setting up outer loop inductions
- support for widening non-induction PHIs that can result from inner loops and uniform conditional branches
- support for generating uniform inner branches
We plan to add a handful C outer loop executable tests once the initial code generation support is committed. This patch is expected to be NFC for the inner loop vectorizer path. Since we are moving in the direction of supporting outer loop vectorization in LV, it may also be time to rename classes such as InnerLoopVectorizer.
Reviewers: fhahn, rengolin, hsaito, dcaballe, mkuper, hfinkel, Ayal
Reviewed By: fhahn, hsaito
Subscribers: dmgreen, bollu, tschuett, rkruppe, rogfer01, llvm-commits
Differential Revision: https://reviews.llvm.org/D50820
llvm-svn: 342197
Summary:
I accidentally left this behind in D50306, and it causes a build warning
when I build with gcc7.
Subscribers: arsenm, kzhuravl, jvesely, wdng, nhaehnle, yaxunl, dstuttard, t-tye, llvm-commits
Differential Revision: https://reviews.llvm.org/D52022
Change-Id: I30f7a47047e9d9d841f652da66d2fea19e74842c
llvm-svn: 342189
Summary:
Place global arrays in comdat sections with their associated functions.
This makes sure they are stripped along with the functions they
reference, even on the BFD linker.
Reviewers: eugenis
Reviewed By: eugenis
Subscribers: eraman, hiraditya, llvm-commits
Differential Revision: https://reviews.llvm.org/D51902
llvm-svn: 342186
See rL342148
This probably only shows up in BUILD_SHARED_LIBS=ON builds
which might explain how it crept in.
Differential Revision: https://reviews.llvm.org/D52054
llvm-svn: 342180
These functions previously passed nil for the location, which always resulted in a crash.
This is a signature breaking change, but I cannot see how they could have been used before.
Patch by Ben Clayton!
Differential Revision: https://reviews.llvm.org/D51970
llvm-svn: 342179
INLINEASM instructions use extra operands to carry flags. If a register operand is removed without removing the flag operand, then the flags will no longer make sense.
This patch fixes this by preventing the removal when a flag operand is present.
The included test case was generated by MS inline assembly. Longer term maybe we should fix the inline assembly parsing to not generate redundant operands.
Differential Revision: https://reviews.llvm.org/D51829
llvm-svn: 342176
When replacing a named register input to the appropriately sized
sub/super-register. In the case of a 64-bit value being assigned to a
register in 32-bit mode, match GCC's assignment.
Reviewers: eli.friedman, craig.topper
Subscribers: nickdesaulniers, llvm-commits, hiraditya
Differential Revision: https://reviews.llvm.org/D51502
llvm-svn: 342175
Summary:
It is sometimes important to check that some newly-computed value
is non-negative and only n bits wide (where n is a variable.)
There are many ways to check that:
https://godbolt.org/z/o4RB8D
The last variant seems best?
(I'm sure there are some other variations i haven't thought of..)
More complicated, canonical pattern:
https://rise4fun.com/Alive/uhA
We do need to have two `switch()`'es like this,
to not mismatch the swappable predicates.
https://bugs.llvm.org/show_bug.cgi?id=38708
Reviewers: spatel, craig.topper, RKSimon
Reviewed By: spatel
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D52001
llvm-svn: 342173
This adds DebugCounter support to the PartiallyInlineLibCalls pass,
which should make debugging/automated bisection easier in the future.
Patch by Zhizhou Yang!
Differential Revision: https://reviews.llvm.org/D50093
llvm-svn: 342172
Came up in https://reviews.llvm.org/D52001#1233827
While we don't do a good job here, we at least want to make
sure that we don't have any inf-loops.
llvm-svn: 342171
add a tool to generate symbol remapping files.
Summary:
The new tool llvm-cxxmap builds a symbol mapping table from a file containing
a description of partial equivalences to apply to mangled names and files
containing old and new symbol tables.
Reviewers: davidxl
Subscribers: mgorny, llvm-commits
Differential Revision: https://reviews.llvm.org/D51470
llvm-svn: 342168
Summary:
The hash computed for an ArrayType was different when first constructed
versus when later profiled due to the constructor default argument, and
we were not tracking constructor / destructor variant as part of the
mangled name AST, leading to incorrect equivalences.
Reviewers: erik.pilkington
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D51463
llvm-svn: 342166
This allows the xor to be removed completely.
This might help with recomitting r341674, but seems good regardless.
Coincidentally fixes PR38915.
Differential Revision: https://reviews.llvm.org/D51964
llvm-svn: 342163
Summary:
Fixed assertions due to invalid fixup when encoding compressed instructions
(c.addi, c.addiw, c.li, c.andi) with bare symbols with/without modifiers.
This matches GAS behavior as well.
This bug was uncovered by a LLVM MC Disassembler Protocol Buffer Fuzzer
for the RISC-V assembly language.
Reviewers: asb
Reviewed By: asb
Subscribers: rbar, johnrusso, simoncook, sabuasal, niosHD, kito-cheng, shiva0217, zzheng, edward-jones, mgrang, rogfer01, MartinMosbeck, brucehoult, the_o, rkruppe, PkmX, jocewei, asb
Differential Revision: https://reviews.llvm.org/D52005
llvm-svn: 342160
Summary:
The illegal instruction 0x00 0x00 is being wrongly decoded as
c.addi4spn with 0 immediate.
The invalid instruction 0x01 0x61 is being wrongly decoded as
c.addi16sp with 0 immediate.
This bug was uncovered by a LLVM MC Disassembler Protocol Buffer Fuzzer
for the RISC-V assembly language.
Reviewers: asb
Reviewed By: asb
Subscribers: rbar, johnrusso, simoncook, sabuasal, niosHD, kito-cheng, shiva0217, zzheng, edward-jones, mgrang, rogfer01, MartinMosbeck, brucehoult, the_o, rkruppe, PkmX, jocewei, asb
Differential Revision: https://reviews.llvm.org/D51815
llvm-svn: 342159
Also, add a check to ensure that when main has the expected signature
we do not create a wrapper.
Differential Revision: https://reviews.llvm.org/D51562
llvm-svn: 342157
I accidentally committed this diff with rL342147 because
I had applied D51964. We probably do need those checks,
but D51964 has tests and more discussion/motivation,
so they should be re-added with that patch.
llvm-svn: 342149
I don't have a test case for this, but it's motivated by
the discussion in D51964, and I've added TODO comments for
the better fix - move simplifications into instsimplify
because that's more efficient and reduces risk of infinite
loops in instcombine caused by transforms trying to do the
opposite folds.
In this case, we know that the transform that tries to move
'not' through min/max can be fooled by the multiple uses
of a value in another min/max, so try to squash the
foldSPFofSPF() patterns first.
llvm-svn: 342147
We previously only allowed truncs as sinks, but now allow them as
sources too. We do this by checking that the result type is the
narrow type that we're trying to optimise for.
Differential Revision: https://reviews.llvm.org/D51978
llvm-svn: 342141
Part of FixConsts wrongly assumes either a 8- or 16-bit constant
which can result in the wrong constants being generated during
promotion.
Differential Revision: https://reviews.llvm.org/D52032
llvm-svn: 342140
In r319995, we fixed the line table format to version 2 on Darwin
because dsymutil didn't yet understand the new format which caused test
failures for the LLDB bots. This has been resolved in the meantime so
there's no reason to keep this limitation.
rdar://problem/35968332
llvm-svn: 342136
If an argument was passed on the stack, this
was using the default alignment.
I'm not sure there's an observable change from this. This
was observable due to bugs in expansion of unaligned
loads and stores, but since that is fixed I don't think
this matters much.
llvm-svn: 342133
The Technical Reference Manuals for these two CPUs state that branching
to an unaligned 32-bit instruction incurs an extra pipeline reload
penalty. That's bad.
This also enables the optimization at -Os since it costs on average one
byte per loop in return for 1 cycle per iteration, which is pretty good
going.
llvm-svn: 342127
Summary:
This change has a number of fixes for FDR mode in compiler-rt along with
changes to the tooling handling the traces in llvm.
In the runtime, we do the following:
- Advance the "last record" pointer appropriately when writing the
custom event data in the log.
- Add XRAY_NEVER_INSTRUMENT in the rewinding routine.
- When collecting the argument of functions appropriately marked, we
should not attempt to rewind them (and reset the counts of functions
that can be re-wound).
In the tooling, we do the following:
- Remove the state logic in BlockIndexer and instead rely on the
presence/absence of records to indicate blocks.
- Move the verifier into a loop associated with each block.
Reviewers: mboerger, eizan
Subscribers: llvm-commits, hiraditya
Differential Revision: https://reviews.llvm.org/D51965
llvm-svn: 342122
This implements suggesting alternative mnemonics when an invalid one is
specified. For example `addru $9, $6, 17767` leads to the following
error message:
error: unknown instruction, did you mean: add, addiu, addu, maddu?
Differential revision: https://reviews.llvm.org/D40646
llvm-svn: 342119
Summary:
The snippet-generation part goes to the SnippetGenerator class.
This will allow benchmarking arbitrary code (see PR38437).
Reviewers: gchatelet
Subscribers: mgorny, tschuett, llvm-commits
Differential Revision: https://reviews.llvm.org/D51979
llvm-svn: 342117
Summary:
Previously we type legalized v2i32 div/rem by promoting to v2i64. But we don't support div/rem of vectors so op legalization would then scalarize it using i64 scalar ops since it doesn't know about the original promotion. 64-bit scalar divides on Intel hardware are known to be slow and in 32-bit mode they require a libcall.
This patch switches type legalization to do the scalarizing itself using i32.
It looks like the division by power of 2 optimization is still kicking in and leaving the code as a vector. The division by other constant optimization doesn't kick in pre type legalization since it ignores illegal types. And previously, after type legalization we scalarized the v2i64 since we don't have v2i64 MULHS/MULHU support.
Another option might be to widen v2i32 to v4i32 so we could do division by constant optimizations, but we'd have to be careful to only do that for constant divisors or we risk scalaring to 4 scalar divides.
Reviewers: RKSimon, spatel
Reviewed By: spatel
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D51325
llvm-svn: 342114
The `IMAGE_REL_ARM_BRANCH20T` applies only to a `b.w` instruction. A
thumb-2 `bl` should be relocated using a `IMAGE_REL_ARM_BRANCH24T`.
Correct the relocation that we emit in such a case.
Resolves PR38620! Based on the patch by Jordan Rhee!
llvm-svn: 342109
Shufflevector instructions in LLVM IR that extract a subset of elements
of a longer input into a shorter vector can be done using VECTOR_SHUFFLEs.
This will avoid expanding them into constly extracts and inserts.
llvm-svn: 342091
This is available on most platforms (Linux/Mac/Win/BSD) with no extra syscalls.
On other platforms (e.g. Solaris) we stat() if this information is requested.
This will allow switching clang's VFS to efficiently expose (path, type) when
traversing a directory. Currently it exposes an entire Status, but does so by
calling fs::status() on all platforms.
Almost all callers only need the path, and all callers only need (path, type).
Patch by sammccall (Sam McCall)
Differential Revision: https://reviews.llvm.org/D51918
llvm-svn: 342089
template methods in JITDylib out-of-line.
This also splits JITDylib::define into a pair of template methods, one taking an
lvalue reference and the other an rvalue reference. This simplifies the
templates at the cost of a small amount of code duplication.
llvm-svn: 342087
construction, a new convenience lookup method, and add-to layer methods.
ExecutionSession now creates a special 'main' JITDylib upon construction. All
subsequently created JITDylibs are added to the main JITDylib's search order by
default (controlled by the AddToMainDylibSearchOrder parameter to
ExecutionSession::createDylib). The main JITDylib's search order will be used in
the future to properly handle cross-JITDylib weak symbols, with the first
definition in this search order selected.
This commit also adds a new ExecutionSession::lookup convenience method that
performs a blocking lookup using the main JITDylib's search order, as this will
be a very common operation for clients.
Finally, new convenience overloads of IRLayer and ObjectLayer's add methods are
introduced that add the given program representations to the main dylib, which
is likely to be the common case.
llvm-svn: 342086
Summary:
rL341389 broke code with tied register operands in inline assembly. For
example, `asm("" : "=r"(var) : "0"(var));`
The code above specifies the input operand to be in the same register
with the output operand, tying the two register. This patch makes this
kind of code work again.
Reviewers: dschuff
Subscribers: sbc100, jgravelle-google, eraman, sunfish, llvm-commits
Differential Revision: https://reviews.llvm.org/D51991
llvm-svn: 342084
Summary:
Some FPMathOperators succeed and the retrieve FMF context when they never have it, we should omit these cases to keep from removing FMF context.
For instance when we visit some FPMathOperator mapped Instructions which never have FMF flags and a Node was associated which does have FMF flags, that Node today will have all its flags cleared via the intersect operation. With this change, we exclude associating Nodes that never have FPMathOperator status under FMF.
Reviewers: spatel, wristow, arsenm, hfinkel, aemerson
Reviewed By: spatel
Subscribers: llvm-commits, wdng
Differential Revision: https://reviews.llvm.org/D51145
llvm-svn: 342081
r342003 added support for emitting FPO data from the
DEBUG_S_FRAMEDATA subsection of the .debug$S section to the PDB
file. However, that is not the end of the story. FPO can end
up in two different destinations in a PDB, each corresponding to
a different FPO data source.
The case handled by r342003 involves copying data from the
DEBUG_S_FRAMEDATA subsection of the .debug$S section to the
"New FPO" stream in the PDB, which is then referred to by the
DBI stream. The case handled by this patch involves copying
records from the .debug$F section of an object file to the "FPO"
stream (or perhaps more aptly, the "Old FPO" stream) in the PDB
file, which is also referred to by the DBI stream.
The formats are largely similar, and the difference is mostly
only visible in masm generated object files, such as some of the
low-level CRT object files like memcpy. MASM doesn't appear to
support writing the DEBUG_S_FRAMEDATA subsection, and instead
just writes these records to the .debug$F section.
Although clang-cl does not emit a .debug$F section ever, lld still
needs to support it so we have good debugging for CRT functions.
Differential Revision: https://reviews.llvm.org/D51958
llvm-svn: 342080
Scalarization of a shuffle will break up the source vectors into individual
elements, and use them to assemble the resulting vector. An element type of
a legal vector type may not necessarily be a legal scalar type, so make
sure that the extracted values are extended to a legal scalar type.
llvm-svn: 342079
Move isa version determination into TargetParser.
Also switch away from target features to CPU string when
determining isa version. This fixes an issue when we
output wrong isa version in the object code when features
of a particular CPU are altered (i.e. gfx902 w/o xnack
used to result in gfx900).
llvm-svn: 342069
Summary:
It is sometimes important to check that some newly-computed value
is non-negative and only `n` bits wide (where `n` is a variable.)
There are **many** ways to check that:
https://godbolt.org/z/o4RB8D
The last variant seems best?
(I'm sure there are some other variations i haven't thought of..)
Let's handle the second variant first, since it is much simpler.
https://rise4fun.com/Alive/LYjYhttps://bugs.llvm.org/show_bug.cgi?id=38708
Reviewers: spatel, craig.topper, RKSimon
Reviewed By: spatel
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D51985
llvm-svn: 342067
Submitted on behalf of Armando Montanez (amontanez@google.com).
Objects with unused program headers copied by objcopy would always have
nonzero values for program header offset and program header entry size.
While technically valid, this atypical behavior triggers warnings in some
tools. This change sets the two fields to zero when the program header is
unused, better fitting the general expectations for unused program header
data.
Section headers behaved somewhat similarly (though only with the entry size),
and are fixed in this revision as well.
Differential Revision: https://reviews.llvm.org/D51961
llvm-svn: 342065
Summary:
Match the ordering semantics of non-vector comparisons. For
floating point comparisons that do not correspond to instructions, the
tests check that some vector comparison instruction was emitted but do
not care about the full implementation.
Reviewers: aheejin, dschuff
Subscribers: sbc100, jgravelle-google, sunfish, llvm-commits
Differential Revision: https://reviews.llvm.org/D51765
llvm-svn: 342064
There's no advantage to this instruction unless you need to avoid touching other flag bits. It's encoding is longer, it can't fold an immediate, it doesn't write all the flags.
I don't think gcc will generate this instruction either.
Fixes PR38852.
Differential Revision: https://reviews.llvm.org/D51754
llvm-svn: 342059
Currently we have a few duplicated matcher classes, which all do pretty
much the same thing. This patch introduces generic
One,Tow,ThreeOps_match classes which take the opcode the match as
template argument.
Reviewers: SjoerdMeijer, dneilson, spatel, arsenm
Reviewed By: SjoerdMeijer
Differential Revision: https://reviews.llvm.org/D51044
llvm-svn: 342058
Fix for https://bugs.llvm.org/show_bug.cgi?id=38912.
In GVNHoist::computeInsertionPoints() we iterate over the Value
Numbers and calculate the Iterated Dominance Frontiers without
clearing the IDFBlocks vector first. IDFBlocks ends up accumulating
an insane number of basic blocks, which bloats the compilation time
of SemaChecking.cpp with ubsan enabled.
Differential Revision: https://reviews.llvm.org/D51980
llvm-svn: 342055
This patch adds codegen support for the saving/restoring
V8-V23 for functions specified with the aarch64_vector_pcs
calling convention attribute, as added in patch D51477.
Reviewers: t.p.northover, gberry, thegameg, rengolin, javed.absar, MatzeB
Reviewed By: thegameg
Differential Revision: https://reviews.llvm.org/D51479
llvm-svn: 342049
Eliminating some duplication of rangelist dumping code at the expense of
some version-dependent code in dump and extract routines.
Reviewer: dblaikie, JDevlieghere, vleschuk
Differential revision: https://reviews.llvm.org/D51081
llvm-svn: 342048
The output of splitLargeGEPOffsets does not appear to be deterministic because
of the way that we iterate over a DenseMap. I've changed it to a MapVector for
consistent output.
The test here isn't particularly great, only showing a consmetic difference in
output. The original reproducer is much larger but show a diffierence in
instruction ordering, leading to different codegen.
Differential Revision: https://reviews.llvm.org/D51851
llvm-svn: 342043
Previously the alignment on the newly created switch table data was not set,
meaning that DataLayout::getPreferredAlignment was free to overalign it to 16
bytes. This causes unnecessary code bloat.
Differential Revision: https://reviews.llvm.org/D51800
llvm-svn: 342039
This patch refactors several parts of AArch64FrameLowering
so that it can be easily extended to support saving/restoring
of FPR128 (Q) registers.
Reviewers: t.p.northover, gberry, thegameg, rengolin, javed.absar
Reviewed By: thegameg
Differential Revision: https://reviews.llvm.org/D51478
llvm-svn: 342038
SMLAD and SMLALD instructions also come in the form of SMLADX and
SMLALDX which perform an exchange on their second operand. To support
this, more of the loads in the MAC candidates are compared for
sequential access and a boolean value has been added to BinOpChain.
AddMACCandiate has been refactored into a small pattern matching
state machine to reduce the amount of duplicated code, but also to
enable the matching to be more flexible. CreateParallelMACPairs now
iterates through all the candidates to find parallel ones.
Differential Revision: https://reviews.llvm.org/D51424
llvm-svn: 342033
This patch adds parsing support for the 'aarch64_vector_pcs'
calling convention attribute to calls and function declarations.
More information describing the vector ABI and procedure call standard
can be found here:
https://developer.arm.com/products/software-development-tools/\
hpc/arm-compiler-for-hpc/vector-function-abi
Reviewers: t.p.northover, rnk, rengolin, javed.absar, thegameg, SjoerdMeijer
Reviewed By: SjoerdMeijer
Differential Revision: https://reviews.llvm.org/D51477
llvm-svn: 342030
Move the 2 classes out of LoopVectorize.cpp to make it easier to re-use
them for VPlan outside LoopVectorize.cpp
Reviewers: Ayal, mssimpso, rengolin, dcaballe, mkuper, hsaito, hfinkel, xbolva00
Reviewed By: rengolin, xbolva00
Differential Revision: https://reviews.llvm.org/D49488
llvm-svn: 342027
This fixes a layering violation:
Analysis/IVDescrtors.cpp can't include Transforms/Utils/BasicBlockUtils.h,
since TransformUtils depends on Analysis.
llvm-svn: 342024
I suspect this became unecessary when the CSE of mgather was fixed in r338080. It may still be possible to hit this if we widen the element type of a gather outside of type legalization and the promote the mask of a separate gather node so they become the same. But that seems pretty unlikely.
llvm-svn: 342022
Summary:
The InductionDescriptor and RecurrenceDescriptor classes basically analyze the IR to identify the respective IVs. So, it is better to have them in the "Analysis" directory instead of the "Transforms" directory.
The rationale for this is to make the Induction and Recurrence descriptor classes available for analysis passes. Currently including them in an analysis pass produces link error (http://lists.llvm.org/pipermail/llvm-dev/2018-July/124456.html).
Induction and Recurrence descriptors are moved from Transforms/Utils/LoopUtils.h|cpp to Analysis/IVDescriptors.h|cpp.
Reviewers: dmgreen, llvm-commits, hfinkel
Reviewed By: dmgreen
Subscribers: mgorny
Differential Revision: https://reviews.llvm.org/D51153
llvm-svn: 342016
Summary:
In GNUX23, is64BitMode returns true, but pointers are 32-bits. So we shouldn't copy pointer values into RSI/RDI since the widths don't match.
Fixes PR38865 despite what the title says. I think the llvm_unreachable in the copyPhysReg code tricked the optimizer and made the fatal error trigger.
Reviewers: rnk, efriedma, MatzeB, echristo
Reviewed By: efriedma
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D51893
llvm-svn: 342015
Since the outliner is a module pass, it doesn't get codegen size remarks like
the other codegen passes do. This adds size remarks *to* the outliner.
This is kind of a workaround, so it's peppered with FIXMEs; size remarks
really ought to not ever be handled by the pass itself. However, since the
outliner is the only "MachineModulePass", this works for now. Since the
entire purpose of the MachineOutliner is to produce code size savings, it
really ought to be included in codgen size remarks.
If we ever go ahead and make a MachineModulePass (say, something similar to
MachineFunctionPass), then all of this ought to be moved there.
llvm-svn: 342009
Name: op_ugt_sum
%a = add i8 %x, %y
%r = icmp ugt i8 %x, %a
=>
%notx = xor i8 %x, -1
%r = icmp ugt i8 %y, %notx
Name: sum_ult_op
%a = add i8 %x, %y
%r = icmp ult i8 %a, %x
=>
%notx = xor i8 %x, -1
%r = icmp ugt i8 %y, %notx
https://rise4fun.com/Alive/ZRxI
AFAICT, this doesn't interfere with any add-saturation patterns
because those have >1 use for the 'add'. But this should be
better for IR analysis and codegen in the basic cases.
This is another fold inspired by PR14613:
https://bugs.llvm.org/show_bug.cgi?id=14613
llvm-svn: 342004
Summary:
Previously, check-llvm on my Windows 10 workstation took about 300s to
run, and it would lock up my mouse. Now, it takes just under 60s.
Previously running the tests only used about 15% of the available CPU
time, now it uses up to 60%.
Shell32.dll and ole32.dll have direct dependencies on user32.dll and
gdi32.dll. These DLLs are mostly used to for Windows GUI functionality,
which most LLVM tools don't need. It seems that these DLLs acquire and
release some system resources on startup and exit, and that requires
acquiring the same highly contended lock discussed in this post:
https://randomascii.wordpress.com/2017/07/09/24-core-cpu-and-i-cant-move-my-mouse/
Most LLVM tools still have a transitive dependency on
SHGetKnownFolderPathW, which is used to look up the user home directory
and local AppData directory. We also use SHFileOperationW to recursively
delete a directory, but only LLDB and clang-doc use this today. At some
point, we might consider removing these last shell32.dll deps, but for
now, I would like to take this free performance.
Many thanks to Bruce Dawson for suggesting this fix. He looked at an ETW
trace of an LLVM test run, noticed these DLLs, and suggested avoiding
them.
Reviewers: zturner, pcc, thakis
Subscribers: mgorny, llvm-commits, hiraditya
Differential Revision: https://reviews.llvm.org/D51952
llvm-svn: 342002
This reverts rL341954.
The builder `sanitizer-x86_64-linux-bootstrap-ubsan` has been
failing with timeouts at stage2 clang/ubsan:
[3065/3073] Linking CXX executable bin/lld
command timed out: 1200 seconds without output running python
../sanitizer_buildbot/sanitizers/buildbot_selector.py,
attempting to kill
llvm-svn: 342001
Summary:
There are two registers encoded in the S_FRAMEPROC flags: one for locals
and one for parameters. The encoding is described by the
ExpandEncodedBasePointerReg function in cvinfo.h. Two bits are used to
indicate one of four possible values:
0: no register - Used when there are no variables.
1: SP / standard - Variables are stored relative to the standard SP
for the ISA.
2: FP - Variables are addressed relative to the ISA frame
pointer, i.e. EBP on x86. If realignment is required, parameters
use this. If a dynamic alloca is used, locals will be EBP relative.
3: Alternative - Variables are stored relative to some alternative
third callee-saved register. This is required to address highly
aligned locals when there are dynamic stack adjustments. In this
case, both the incoming SP saved in the standard FP and the current
SP are at some dynamic offset from the locals. LLVM uses ESI in
this case, MSVC uses EBX.
Most of the changes in this patch are to pass around the CPU so that we
can decode these into real, named architectural registers.
Subscribers: hiraditya
Differential Revision: https://reviews.llvm.org/D51894
llvm-svn: 341999
In this diff we adjust the code of getSymbols to avoid creating LLVMContext when it's not necessary.
Without this patch when the function getSymbols was called on a MachO object with a __bitcode section
it was parsing the embedded bitcode and then ignoring the result.
Test plan: make check-all
Differential revision: https://reviews.llvm.org/D51759
llvm-svn: 341998
These are the folds in Alive;
Name: xor_ult
Pre: isPowerOf2(-C1)
%xor = xor i8 %x, C1
%r = icmp ult i8 %xor, C1
=>
%r = icmp ugt i8 %x, ~C1
Name: xor_ugt
Pre: isPowerOf2(C1+1)
%xor = xor i8 %x, C1
%r = icmp ugt i8 %xor, C1
=>
%r = icmp ugt i8 %x, C1
https://rise4fun.com/Alive/Vty
The ugt case in its simplest form was already handled by DemandedBits,
but that's not ideal as shown in the multi-use test.
I'm not sure if these are all of the symmetrical folds, but I adjusted
the existing code for one of the folds to try to show the similarities.
There's no obvious connection, but this is another preliminary step
for PR14613...
https://bugs.llvm.org/show_bug.cgi?id=14613
llvm-svn: 341997
Summary: Initial support for nsw, nuw and exact flags in MI
Reviewers: spatel, hfinkel, wristow
Reviewed By: spatel
Subscribers: nlopes
Differential Revision: https://reviews.llvm.org/D51738
llvm-svn: 341996
Fixes at_file.c test failure caused by r341988. We may want to change
how we treat \n in our tokenizer, but this is probably a good fix
regardless, since we can invoke all kinds of programs with different
interpretations of the command line quoting rules.
llvm-svn: 341992
Summary:
Shell32.dll depends on gdi32.dll and user32.dll, which are mostly DLLs
for Windows GUI functionality. LLVM's utilities don't typically need GUI
functionality, and loading these DLLs seems to be slowing down startup.
Also, we already have an implementation of Windows command line
tokenization in cl::TokenizeWindowsCommandLine, so we can just use it.
The goal is to get the original argv in UTF-8, so that it can pass
through most LLVM string APIs. A Windows process starts life with a
UTF-16 string for its command line, and it can be retreived with
GetCommandLineW from kernel32.dll.
Previously, we would:
1. Get the wide command line
2. Call CommandLineToArgvW to handle quoting rules and separate it into
arguments.
3. For each wide argument, expand wildcards (* and ?) using
FindFirstFileW.
4. Convert each argument to UTF-8
Now we:
1. Get the wide command line, convert the whole thing to UTF-8
2. Tokenize the UTF-8 command line with cl::TokenizeWindowsCommandLine
3. For each argument, expand wildcards if present
- This requires converting back to UTF-16 to call FindFirstFileW
- Results of FindFirstFileW must be converted back to UTF-8
Reviewers: zturner
Subscribers: hiraditya, llvm-commits
Differential Revision: https://reviews.llvm.org/D51941
llvm-svn: 341988
Summary:
Place global arrays in comdat sections with their associated functions.
This makes sure they are stripped along with the functions they
reference, even on the BFD linker.
Reviewers: eugenis
Reviewed By: eugenis
Subscribers: eraman, hiraditya, llvm-commits
Differential Revision: https://reviews.llvm.org/D51902
llvm-svn: 341987
Summary:
Update MemorySSA in old LoopUnswitch pass.
Actual dependency and update is disabled by default.
Subscribers: sanjoy, jlebar, Prazek, llvm-commits
Differential Revision: https://reviews.llvm.org/D45301
llvm-svn: 341984
into TargetParser.
Also switch away from target features to CPU string when
determining isa version. This fixes an issue when we
output wrong isa version in the object code when features
of a particular CPU are altered (i.e. gfx902 w/o xnack
used to result in gfx900).
Differential Revision: https://reviews.llvm.org/D51890
llvm-svn: 341982
I noticed that we were not back-propagating undef lanes to shuffle masks when we have a
shuffle that reduces the vector width. This is part of investigating/solving PR38691:
https://bugs.llvm.org/show_bug.cgi?id=38691
The DAG equivalent was proposed with:
D51696
Differential Revision: https://reviews.llvm.org/D51433
llvm-svn: 341981
Summary:
This patch removes the storing of accumulated floating point data
within the llvm-mca library.
This patch splits-up the two quantities: cycles and number of resource units.
By splitting-up these two quantities, we delay the calculation of "cycles per resource unit"
until that value is read, reducing the chance of accumulating floating point error.
I considered using the APFloat, but after measuring performance, for a large (many iteration)
sample, I decided to go with this faster solution.
Reviewers: andreadb, courbet, RKSimon
Reviewed By: andreadb
Subscribers: llvm-commits, javed.absar, tschuett, gbedwell
Differential Revision: https://reviews.llvm.org/D51903
llvm-svn: 341980
Right now, the counters are added in regards of the number of successors
for a given BasicBlock: it's good when we've only 1 or 2 successors (at
least with BranchInstr). But in the case of a switch statement, the
BasicBlock after switch has several predecessors and we need know from
which BB we're coming from.
So the idea is to revert what we're doing: add a PHINode in each block
which will select the counter according to the incoming BB. They're
several pros for doing that:
- we fix the "switch" bug
- we remove the function call to "__llvm_gcov_indirect_counter_increment"
and the lookup table stuff
- we replace by PHINodes, so the optimizer will probably makes a better
job.
Patch by calixte!
Differential Revision: https://reviews.llvm.org/D51619
llvm-svn: 341977
Simon Pilgrim