Some CPUs are already overriding these sign extension instructions but we should be able to use the WriteALU schedule class by default.
Differential Revision: https://reviews.llvm.org/D39899
llvm-svn: 318308
Summary:
Added more remarks to SLP pass, in particular "missed" optimization remarks.
Also proposed several tests for new functionality.
Patch by Vladimir Miloserdov!
For reference you may look at: https://reviews.llvm.org/rL302811
Reviewers: anemet, fhahn
Reviewed By: anemet
Subscribers: javed.absar, lattner, petecoup, yakush, llvm-commits
Differential Revision: https://reviews.llvm.org/D38367
llvm-svn: 318307
Move the calling convention checks for tail-call eligibility for the 64-bit
SysV ABI into a separate function. This is so that it can be shared with
'mayBeEmittedAsTailCall' in a subsequent change.
llvm-svn: 318305
This is a recommit of r316869 which was speculatively reverted with r317444 and
subsequently shown to not be the cause of PR35210. That crash should be fixed
after r318237.
Original commit message:
The old PM sets the options of what used to be known as "latesimplifycfg" on the
instantiation after the vectorizers have run, so that's what we'redoing here.
FWIW, there's a later SimplifyCFGPass instantiation in both PMs where we do not
set the "late" options. I'm not sure if that's intentional or not.
Differential Revision: https://reviews.llvm.org/D39407
llvm-svn: 318299
Summary:
Prevent an issue where a diagnostic is reported multiple times by bailing out with a ParseFail if an invalid SVE register element qualifier/suffix is specified, for example:
<stdin>:10:18: error: invalid sve vector kind qualifier
add z20.h, z2.h, z31.x
^
<stdin>:10:18: error: invalid sve vector kind qualifier
add z20.h, z2.h, z31.x
...
<stdin>:10:18: error: invalid sve vector kind qualifier
add z20.h, z2.h, z31.x
^
Reviewers: fhahn, rengolin
Reviewed By: rengolin
Subscribers: aemerson, javed.absar, tschuett, llvm-commits, kristof.beyls
Differential Revision: https://reviews.llvm.org/D39894
llvm-svn: 318297
APInt is now used instead of uint64_t in function genConstMult() allowing
multiplication optimizations with constants of arbitrary length.
Patch by Milos Stojanovic.
Differential Revision: https://reviews.llvm.org/D38130
llvm-svn: 318296
This is a refactoring/cleanup of Arm `addrmode2` operand class. The patch
removes it completely.
Differential Revision: https://reviews.llvm.org/D39832
llvm-svn: 318291
In constructAbstractSubprogramScopeDIE there can be a potential mismatch
between `this` and the CU of ContextDIE when a scope is shared between
two DISubprograms belonging to a different CU. In that case, `this` is
the CU that was specified in the IR, but the CU of ContextDIE is that of
the first subprogram that was emitted. This patch fixes the mismatch by
looking up the CU of ContextDIE, and switching to use that.
This fixes PR35212 (https://bugs.llvm.org/show_bug.cgi?id=35212)
Patch by Philip Craig!
Differential revision: https://reviews.llvm.org/D39981
llvm-svn: 318289
Summary:
This fixes PR35241.
When using byval, the data is effectively copied as part of the call
anyway, so the pointer returned by the alloca will not be leaked to the
callee and thus there is no reason to issue a warning.
Reviewers: rnk
Reviewed By: rnk
Subscribers: Ka-Ka, llvm-commits
Differential Revision: https://reviews.llvm.org/D40009
llvm-svn: 318279
This allows us to remove extra extend creation during lowering and more accurately reflects the semantics of the instruction.
While there add an extra output VT to X86 masked gather node to better match the isel pattern predicate. Currently we're exploiting the fact that the isel table doesn't count how many output results a node actually has if the result type of any can be inferred from the first result and the type constraints defined in tablegen. I think we might ultimately want to lower all MGATHER/MSCATTER to an X86ISD node with the extra mask result and stop relying on this hole in the isel checking.
llvm-svn: 318278
Summary:
This patch optimizes a binop sandwiched between 2 selects with the same condition. Since we know its only used by the select we can propagate the appropriate input value from the earlier select.
As I'm writing this I realize I may need to avoid doing this for division in case the select was protecting a divide by zero?
Reviewers: spatel, majnemer
Reviewed By: majnemer
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D39999
llvm-svn: 318267
It crashes building sqlite; see reply on the llvm-commits thread.
> [SLPVectorizer] Failure to beneficially vectorize 'copyable' elements in integer binary ops.
>
> Patch tries to improve vectorization of the following code:
>
> void add1(int * __restrict dst, const int * __restrict src) {
> *dst++ = *src++;
> *dst++ = *src++ + 1;
> *dst++ = *src++ + 2;
> *dst++ = *src++ + 3;
> }
> Allows to vectorize even if the very first operation is not a binary add, but just a load.
>
> Fixed issues related to previous commit.
>
> Reviewers: spatel, mzolotukhin, mkuper, hfinkel, RKSimon, filcab, ABataev
>
> Reviewed By: ABataev, RKSimon
>
> Subscribers: llvm-commits, RKSimon
>
> Differential Revision: https://reviews.llvm.org/D28907
llvm-svn: 318239
Summary:
This patch adds another failure mode for `validateCFIProtection(..)`, wherein any register that affects the indirect control flow instruction is clobbered to between the CFI-check and the instruction's execution.
Also includes a modification to make MCInstrDesc::hasDefOfPhysReg public.
Reviewers: vlad.tsyrklevich
Reviewed By: vlad.tsyrklevich
Subscribers: llvm-commits, pcc, kcc
Differential Revision: https://reviews.llvm.org/D39820
llvm-svn: 318238
Simplifying a loop latch changes the IR and we need to make sure the pass manager knows to invalidate analysis passes if that happened.
PR35210 discovered a case where we failed to invalidate the post dominator tree after this simplification because we no changes other than simplifying the loop latch.
Fixes PR35210.
Differential Revision: https://reviews.llvm.org/D40035
llvm-svn: 318237
Summary:
In the mode when ASan shadow base is computed as the address of an
external global (__asan_shadow, currently on android/arm32 only),
regalloc prefers to rematerialize this value to save register spills.
Even in -Os. On arm32 it is rather expensive (2 loads + 1 constant
pool entry).
This changes adds an inline asm in the function prologue to suppress
this behavior. It reduces AsanTest binary size by 7%.
Reviewers: pcc, vitalybuka
Subscribers: aemerson, kristof.beyls, hiraditya, llvm-commits
Differential Revision: https://reviews.llvm.org/D40048
llvm-svn: 318235
This is no longer needed for any of the runtimes build and it breaks
in case we don't have the working compiler yet, e.g. when building
a compiler that uses compiler-rt and libc++ as a default runtime,
because these common options check whether these are available.
Differential Revision: https://reviews.llvm.org/D39932
llvm-svn: 318227
Even when building builtins and runtimes for the default target
we shouldn't assume that the just built compiler is already useable.
When the compiler uses compiler-rt and libc++ as the default runtime
and C++ library, it won't be usable until we finish building runtimes.
Differential Revision: https://reviews.llvm.org/D39715
llvm-svn: 318224
The test was doing -stop-after=isel, but that pass is actually the
AMDGPUDAGToDAGISel pass, which might not be built when targeting x86_64.
This changes the test to -stop-after=expand-isel-pseudos instead.
Follow-up to r318202.
llvm-svn: 318220
Allows users to view GraphResult objects in a DOT directed-graph format. This feature can be turned on through the --print-graphs flag.
Also enabled pretty-printing of instructions in output. Together these features make analysis of unprotected CF instructions much easier by providing a visual control flow graph.
Reviewers: pcc
Subscribers: llvm-commits, kcc, vlad.tsyrklevich
Differential Revision: https://reviews.llvm.org/D39819
llvm-svn: 318211
artifacts along with DCE
Legalization Artifacts are all those insts that are there to make the
type system happy. Currently, the target needs to say all combinations
of extends and truncs are legal and there's no way of verifying that
post legalization, we only have *truly* legal instructions. This patch
changes roughly the legalization algorithm to process all illegal insts
at one go, and then process all truncs/extends that were added to
satisfy the type constraints separately trying to combine trivial cases
until they converge. This has the added benefit that, the target
legalizerinfo can only say which truncs and extends are okay and the
artifact combiner would combine away other exts and truncs.
Updated legalization algorithm to roughly the following pseudo code.
WorkList Insts, Artifacts;
collect_all_insts_and_artifacts(Insts, Artifacts);
do {
for (Inst in Insts)
legalizeInstrStep(Inst, Insts, Artifacts);
for (Artifact in Artifacts)
tryCombineArtifact(Artifact, Insts, Artifacts);
} while(!Insts.empty());
Also, wrote a simple wrapper equivalent to SetVector, except for
erasing, it avoids moving all elements over by one and instead just
nulls them out.
llvm-svn: 318210
In addition to the current ON and OFF options, this adds the FORCE_ON
option, which causes a configuration error if libxml2 cannot be used.
Differential revision: https://reviews.llvm.org/D40050
llvm-svn: 318209
This adjusts the tests to hopfully pacify the
llvm-clang-x86_64-expensive-checks-win buildbot.
Unlike many other instructions, these instructions have aliases which
take coprocessor registers, gpr register, accumulator (and dsp accumulator)
registers, floating point registers, floating point control registers and
coprocessor 2 data and control operands.
For the moment, these aliases are treated as pseudo instructions which are
expanded into the underlying instruction. As a result, disassembling these
instructions shows the underlying instruction and not the alias.
Reviewers: slthakur, atanasyan
Differential Revision: https://reviews.llvm.org/D35253
llvm-svn: 318207
Summary:
Instcombine (and probably other passes) sometimes want to change the
type of an alloca. To do this, they generally create a new alloca with
the desired type, create a bitcast to make the new pointer type match
the old pointer type, replace all uses with the cast, and then simplify
the casts. We already knew how to salvage dbg.value instructions when
removing casts, but we can extend it to cover dbg.addr and dbg.declare.
Fixes a debug info quality issue uncovered in Chromium in
http://crbug.com/784609
Reviewers: aprantl, vsk
Subscribers: hiraditya, llvm-commits
Differential Revision: https://reviews.llvm.org/D40042
llvm-svn: 318203
This patch peels off the top case in switch statement into a branch if the
probability exceeds a threshold. This will help the branch prediction and
avoids the extra compares when lowering into chain of branches.
Differential Revision: http://reviews.llvm.org/D39262
llvm-svn: 318202
Clang implements the -finstrument-functions flag inherited from GCC, which
inserts calls to __cyg_profile_func_{enter,exit} on function entry and exit.
This is useful for getting a trace of how the functions in a program are
executed. Normally, the calls remain even if a function is inlined into another
function, but it is useful to be able to turn this off for users who are
interested in a lower-level trace, i.e. one that reflects what functions are
called post-inlining. (We use this to generate link order files for Chromium.)
LLVM already has a pass for inserting similar instrumentation calls to
mcount(), which it does after inlining. This patch renames and extends that
pass to handle calls both to mcount and the cygprofile functions, before and/or
after inlining as controlled by function attributes.
Differential Revision: https://reviews.llvm.org/D39287
llvm-svn: 318195
Patch tries to improve vectorization of the following code:
void add1(int * __restrict dst, const int * __restrict src) {
*dst++ = *src++;
*dst++ = *src++ + 1;
*dst++ = *src++ + 2;
*dst++ = *src++ + 3;
}
Allows to vectorize even if the very first operation is not a binary add, but just a load.
Fixed issues related to previous commit.
Reviewers: spatel, mzolotukhin, mkuper, hfinkel, RKSimon, filcab, ABataev
Reviewed By: ABataev, RKSimon
Subscribers: llvm-commits, RKSimon
Differential Revision: https://reviews.llvm.org/D28907
llvm-svn: 318193
I was being inconsistent with the way I was capitalizing help messages
for command line options. Additionally --remove-section wasn't using
value_desc even though it benefited from it.
Differential Revision: https://reviews.llvm.org/D39978
llvm-svn: 318190
In rare cases, common code will attempt to select an OR of two
constants. This confuses the logic in splitLargeImmediate,
causing an internal error during isel. Fixed by simply leaving
this case to common code to handle.
This fixes PR34859.
llvm-svn: 318187
They don't actually change nay behaviour, as llvm-strings currently
checks the whole object without looking at individual sections anyway.
This allows using llvm-strings in a context that explicitly passes
the -a option.
Differential Revision: https://reviews.llvm.org/D40020
llvm-svn: 318185
Summary:
Bypass of slow divs based on operand values is currently disabled for
-Os. Do the same when profile summary is available and the working set
size of the application is huge. This is similar to how loop peeling is
guarded by hasHugeWorkingSetSize. In the div bypass case, the generated
extra code (and the extra branch) tendss to outweigh the benefits of the
bypass. This results in noticeable performance improvement on an
internal application.
Reviewers: davidxl
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D39992
llvm-svn: 318179
Before using the 32-bit RISBMux set of instructions we need to
verify that the input bits are actually within range of the 32-bit
instruction. This fixer PR35289.
llvm-svn: 318177
D37065 (committed as rL317674) explicitly set hasSideEffects for all
TargetOpcode::* instructions where it was inferred previously. This is a
follow-up to that patch, setting hasSideEffects=0 for CFI_INSTRUCTION,
EH_LABEL, GC_LABEL and ANNOTATION_LABEL. All LLVM tests pass after this
change.
This patch also modifies MachineInstr::isLabel returns true for a
TargetOpcode::ANNOTATION_LABEL, which ensures that an annotation label won't
be incorrectly considered safe to move.
Differential Revision: https://reviews.llvm.org/D39941
llvm-svn: 318174
This change adds a new flag not present in GNU objcopy that we call
--strip-non-alloc.
Differential Revision: https://reviews.llvm.org/D39926
llvm-svn: 318168
TargetLowering::LowerCallTo assumes that sret value type corresponds to a
pointer in default address space, which is incorrect, since sret value type
should correspond to a pointer in alloca address space, which may not
be the default address space. This causes assertion for amdgcn target
in amdgiz environment.
This patch fixes that.
Differential Revision: https://reviews.llvm.org/D39996
llvm-svn: 318167
We haven't been supporting anything but ELF64LE since the start. Luckily
this was always accounted for and the change is pretty trivial. B35281
requests this change for ELF32LE. This change adds support for ELF32LE,
ELF64BE, and ELF32BE with all supported features that already existed
for ELF64LE.
Differential Revision: https://reviews.llvm.org/D39977
llvm-svn: 318166
This is a tablegen backend to generate documentation for the opcodes that exist
for each target. For each opcode, it lists the assembly string, the names and
types of all operands, and the flags and predicates that apply to the opcode.
Differential revision: https://reviews.llvm.org/D31025
llvm-svn: 318155
This patch is part of D38676.
The patch introduces two new Recipes to handle instructions whose vectorization
involves masking. These Recipes take VPlan-level masks in D38676, but still rely
on ILV's existing createEdgeMask(), createBlockInMask() in this patch.
VPBlendRecipe handles intra-loop phi nodes, which are vectorized as a sequence
of SELECTs. Its execute() code is refactored out of ILV::widenPHIInstruction(),
which now handles only loop-header phi nodes.
VPWidenMemoryInstructionRecipe handles load/store which are to be widened
(but are not part of an Interleave Group). In this patch it simply calls
ILV::vectorizeMemoryInstruction on execute().
Differential Revision: https://reviews.llvm.org/D39068
llvm-svn: 318149
Because the block-splitting code is multi-purpose, we have to meddle with the
branches when using it to fixup a conditional branch destination. We got the
code right, but forgot to update the CFG so the verifier complained when
expensive checks were on.
Probably harmless since constant-islands comes so late, but best to fix it
anyway.
llvm-svn: 318148
Get rid of the handwritten instruction selector code for handling
G_CONSTANT. This code wasn't checking all the preconditions correctly
anyway, so it's better to leave it to TableGen, which can handle at
least some cases correctly (e.g. MOVi, MOVi16, folding into binary
operations). Also add tests to cover those cases.
llvm-svn: 318146
When we emit a tail call for Armv8-M, but then discover that the caller needs to
save/restore `LR`, we convert the tail call to an ordinary one, since restoring
`LR` takes extra instructions, which may negate the benefits of the tail
call. If the callee, however, takes stack arguments, this conversion is
incorrect, since nothing has been done to pass the stack arguments.
Thus the patch reverts https://reviews.llvm.org/rL294000
Also, we improve the instruction sequence for popping `LR` in the case when we
couldn't immediately find a scratch low register, but we can use as a temporary
one of the callee-saved low registers and restore `LR` before popping other
callee-saves.
Differential Revision: https://reviews.llvm.org/D39599
llvm-svn: 318143
This test was originally added when an old bug was fixed that caused
broken iterator code to break basic block placement.
The issue has an extremely low chance of every being a problem again.
This specific test is very flaky and fails often due to upstream
changes.
I have removed this test because it negates more value than it returns.
llvm-svn: 318134
If the register from the copy from exec was spilled,
the copy before the spill was deleted leaving a spill
of undefined register verifier error and miscompiling.
Check for other use instructions of the copy register.
llvm-svn: 318132
Registers it and everything, updates all the references, etc.
Next patch will add support to Clang's `-fexperimental-new-pass-manager`
path to actually enable BoundsChecking correctly.
Differential Revision: https://reviews.llvm.org/D39084
llvm-svn: 318128
a legacy and new PM pass.
This essentially moves the class state to parameters and re-shuffles the
code to make that reasonable. It also does some minor cleanups along the
way and leaves some comments.
Differential Revision: https://reviews.llvm.org/D39081
llvm-svn: 318124
For now at least. We clearly need some kind of comdat or
linkonce_odr support for wasm but currently COMDAT is not
supported.
Disable COMDAT support in the same way we do the Mach-O. This
also causes clang not to generated COMDATs.
Differential Revision: https://reviews.llvm.org/D39873
llvm-svn: 318123
This reverts the aforementioned patch and 2 subsequent follow-ups,
as some buildbots are still failing 2 tests because of it.
Investigation is ongoing into the cause of the failures.
llvm-svn: 318112
std::error_code can represent success, so we don't need a
Optional<std::error_code>.
Rename the variable to avoid confusion with the type Error.
llvm-svn: 318111
Similar to r315841, GlobalISel and SelectionDAG require different code for the
common atomic predicates due to differences in the representation.
Even without that, differences in the IR (SDNode vs MachineInstr) require
differences in the C++ predicate.
This patch moves the implementation of the common atomic predicates related to
ordering into tablegen so that it can handle these differences.
It's NFC for SelectionDAG since it emits equivalent code and it's NFC for
GlobalISel since the rules involving the relevant predicates are still
rejected by the importer.
llvm-svn: 318102
Similar to r315841, GlobalISel and SelectionDAG require different code for the
common atomic predicates due to differences in the representation.
Even without that, differences in the IR (SDNode vs MachineInstr) require
differences in the C++ predicate.
This patch moves the implementation of the common atomic predicates related to
memory type into tablegen so that it can handle these differences.
It's NFC for SelectionDAG since it emits equivalent code and it's NFC for
GlobalISel since the rules involving the relevant predicates are still
rejected by the importer.
llvm-svn: 318095
Many projects use this option. There are two ways to use it. You can
either a) Just use --strip-debug and keep the old file with debug
content or b) you can use --strip-debug, --only-keep-debug, and
--add-gnu-debuglink all in conjunction to create two separate files, the
stripped file and the debug file. --only-keep-debug is more complicated
than --strip-debug because it keeps the section headers without keeping
section contents. That's not really supported by llvm-objcopy at the
moment but I plan on adding it. So this change just supports a) and
options to support b) will come soon.
Differential Revision: https://reviews.llvm.org/D39919
llvm-svn: 318094
This change adds a slightly less extreme form of stripping. It should
remove any section that starts with ".debug" and should remove any
symbol table or relocations. In general this strips out most of the
stuff you don't need to execute but leaves a number of things around.
This behavior has been designed to be compatible with GNU strip/objcopy
--strip-all so that anywhere you currently use --strip-all you should be
able to use llvm-objcopy as a drop in replacement.
Differential Revision: https://reviews.llvm.org/D39769
llvm-svn: 318092
If the first values in Value.def is the range of constant, then the code
generated by `isa<Constant>` is smaller by one operation (basically, an add is
removed). It turns out this small optimization reduces the size of the
statically linked clang binary by 400ko on my laptop. The theoritical
performance gain is non visible from my benchmarks, but the size dropdown is.
Differential Revision: https://reviews.llvm.org/D39373
llvm-svn: 318083
Summary:
This fixes PR35221.
Use pseudo-instructions to let MachineCSE hoist global address computation.
Subscribers: aemerson, javed.absar, kristof.beyls, llvm-commits, hiraditya
Differential Revision: https://reviews.llvm.org/D39871
llvm-svn: 318081
I don't believe this was a problem in practice, as it's likely that the
boolean wasn't checked unless the backend condition was non-null.
llvm-svn: 318073
This just adds a TempFile class and replaces the use in
FileOutputBuffer with it.
The only difference for now is better error handling. Followup work includes:
- Convert other user of temporary files to it.
- Add support for automatically deleting on windows.
- Add a createUnnamed method that returns a potentially unnamed
file. It would be actually unnamed on modern linux and have a
unknown name on windows.
llvm-svn: 318069
Some alias instructions are printed with an extra space after the tab
character. Fix this by skipping that space when the tab character is printed
so that the instructions are aligned with the rest of the code.
Patch by Milos Stojanovic.
Differential Revision: https://reviews.llvm.org/D35946
llvm-svn: 318059
If the base of our gather corresponds to something contained in X86ISD::Wrapper we should be able to fold it into the address.
This patch refactors some of the address matching to more fully use the X86ISelAddressMode struct and the getAddressOperands helper. A new helper function matchVectorAddress is added to call matchWrapper or fall back to matchAddressBase.
We should also be able to support constant offsets from a wrapper, but I'll look into that in a future patch. We may even be able to completely reuse matchAddress here, but I wanted to start simple and work up to it.
Differential Revision: https://reviews.llvm.org/D39927
llvm-svn: 318057
Summary:
If a compare instruction is same or inverse of the compare in the
branch of the loop latch, then return a constant evolution node.
This shall facilitate computations of loop exit counts in cases
where compare appears in the evolution chain of induction variables.
Will fix PR 34538
Reviewers: sanjoy, hfinkel, junryoungju
Reviewed By: sanjoy, junryoungju
Subscribers: javed.absar, llvm-commits
Differential Revision: https://reviews.llvm.org/D38494
llvm-svn: 318050
Make one of the legalizer tests a bit more robust by making sure all
values we're interested in are used (either in a store or a return) and
by using loads instead of constants for obtaining values on fewer than
32 bits. This should make the test less fragile to changes in the
legalize combiner, since those loads are legal (as opposed to the
constants, which were being widened and thus produced opportunities for
the legalize combiner).
llvm-svn: 318047
In more recent Linux kernels (including those with 47 bit VMAs) the layout of
virtual memory for powerpc64 changed causing the memory sanitizer to not
work properly. This patch adjusts a bit mask in the memory sanitizer to work
on the newer kernels while continuing to work on the older ones as well.
This is the non-runtime part of the patch and finishes it. ref: r317802
Tested on several 4.x and 3.x kernel releases.
llvm-svn: 318045
Remove builtins from llvm and add AutoUpgrade support.
Also add fast-isel tests for the TEST and TESTN instructions.
Differential Revision: https://reviews.llvm.org/D38736
llvm-svn: 318036
When generating table jump code for switch statements, place the jump
table label as the first operand in the various addition instructions
in order to enable addressing mode selectors to better match index
computation and possibly fold them into the addressing mode of the
table entry load instruction.
Differential revision: https://reviews.llvm.org/D39752
llvm-svn: 318033
CodeGenPrepare sinks address computations from one basic block to another
and attempts to reuse address computations that have already been sunk. If
the same address computation appears twice with the first instance as an
operand of a load whose result is an operand to a simplifable select,
CodeGenPrepare simplifies the select and recursively erases the now dead
instructions. CodeGenPrepare then attempts to use the erased address
computation for the second load.
Fix this by erasing the cached address value if it has zero uses before
looking for the address value in the sunken address map.
This partially resolves PR35209.
Thanks to Alexander Richardson for reporting the issue!
Reviewers: john.brawn
Differential Revision: https://reviews.llvm.org/D39841
llvm-svn: 318032
Summary:
This patch extends the partial inliner to support inlining parts of
vararg functions, if the vararg handling is done in the outlined part.
It adds a `ForwardVarArgsTo` argument to InlineFunction. If it is
non-null, all varargs passed to the inlined function will be added to
all calls to `ForwardVarArgsTo`.
The partial inliner takes care to only pass `ForwardVarArgsTo` if the
varargs handing is done in the outlined function. It checks that vastart
is not part of the function to be inlined.
`test/Transforms/CodeExtractor/PartialInlineNoInline.ll` (already part
of the repo) checks we do not do partial inlining if vastart is used in
a basic block that will be inlined.
Reviewers: davide, davidxl, grosser
Reviewed By: davide, davidxl, grosser
Subscribers: gyiu, grosser, eraman, llvm-commits
Differential Revision: https://reviews.llvm.org/D39607
llvm-svn: 318028
This patch, together with a matching clang patch (https://reviews.llvm.org/D38672), implements the lowering of X86 shuffle i/f intrinsics to IR.
Differential Revision: https://reviews.llvm.org/D38671
Change-Id: I1e7d359a74743e995ec356237a85214ce55d3661
llvm-svn: 318026
Updated the scheduling information of the SKX subtarget in the file X86SchedSkylakeServer.td under lib/Target/X86 to:
1. add regular opcodes in addition to the suffixed "_Int" opcodes
2. add the (V)MAXCPD/MAXCPS/MAXCSD/MAXCSS/MINCPD/MINCPS/MINCSD/MINCSS
instructions that are equivalent to their counterparts without the 'C' as they are part of a hack to
make floating point min/max commutable under fast math.
Reviewers: zvi, RKSimon, craig.topper
Differential Revision: https://reviews.llvm.org/D39833
Change-Id: Ie13702a5ce1b1a08af91ca637a52b6962881e7d6
llvm-svn: 318024
This was using a custom function that didn't handle the
addressing modes properly for private. Use
isLegalAddressingMode to avoid duplicating this.
Additionally, skip the combine if there is only one use
since the standard combine will handle it.
llvm-svn: 318013
The VRNDSCALE instructions implement a superset of the (V)ROUND instructions. They are equivalent if the upper 4-bits of the immediate are 0.
This patch lowers the legacy intrinsics to the VRNDSCALE ISD node and masks the upper bits of the immediate to 0. This allows us to take advantage of the larger register encoding space.
We should maybe consider converting VRNDSCALE back to VROUND in the EVEX to VEX pass if the extended registers are not being used.
I notice some load folding opportunities being missed for the VRNDSCALESS/SD instructions that I'll try to fix in future patches.
llvm-svn: 318008
I want to reuse the VRNDSCALE node for the legacy SSE rounding intrinsics so that those intrinsics can use EVEX instructions. All of these nodes share tablegen multiclasses so I split them all so that they all remain similar in their implementations.
llvm-svn: 318007
Summary:
This patch adds an early out to visitICmpInst if we are looking at a compare as part of an integer absolute value idiom. Similar is already done for min/max.
In the particular case I observed in a benchmark we had an absolute value of a load from an indexed global. We simplified the compare using foldCmpLoadFromIndexedGlobal into a magic bit vector, a shift, and an and. But the load result was still used for the select and the negate part of the absolute valute idiom. So we overcomplicated the code and lost the ability to recognize it as an absolute value.
I've chosen a simpler case for the test here.
Reviewers: spatel, davide, majnemer
Reviewed By: spatel
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D39766
llvm-svn: 317994
matchBinOpReduction currently matches against a single opcode, but we already have a case where we repeat calls to try to match against AND/OR and I'll be shortly adding another case for SMAX/SMIN/UMAX/UMIN (D39729).
This NFCI patch alters matchBinOpReduction to try and pattern match against any of the provided list of candidate bin ops at once to save time.
Differential Revision: https://reviews.llvm.org/D39726
llvm-svn: 317985
This is consistent with out normal implementation of scalar instructions.
While there disable load folding for the patterns with IMPLICIT_DEF unless optimizing for size which is also our standard practice.
llvm-svn: 317977
Allow a pattern rewriter to be installed in CodeGenDAGPatterns and use it to
correct situations where SelectionDAG and GlobalISel disagree on
representation. For example, it would rewrite:
(sextload:i32 $ptr)<<unindexedload>><<sextload>><<sextloadi16>
to:
(sext:i32 (load:i16 $ptr)<<unindexedload>>)
I'd have preferred to replace the fragments and have the expansion happen
naturally as part of PatFrag expansion but the type inferencing system can't
cope with loads of types narrower than those mentioned in register classes.
This is because the SDTCisInt's on the sext constrain both the result and
operand to the 'legal' integer types (where legal is defined as 'a register
class can contain the type') which immediately rules the narrower types out.
Several targets (those with only one legal integer type) would then go on to
crash on the SDTCisOpSmallerThanOp<> when it removes all the possible types
for the result of the extend.
Also, improve isObviouslySafeToFold() slightly to automatically return true for
neighbouring instructions. There can't be any re-ordering problems if
re-ordering isn't happenning. We'll need to improve it further to handle
sign/zero-extending loads when the extend and load aren't immediate neighbours
though.
llvm-svn: 317971
This is a fix for a bug in r317947. We were supposed to check that all the indices are are constant 0, but instead we're only make sure that indices that are constant are 0. Non-constant indices are being ignored.
llvm-svn: 317950
Currently we can only get a uniform base from a simple GEP with 2 operands. This causes us to miss address folding opportunities for simple global array accesses as the test case shows.
This patch adds support for larger GEPs if the other indices are 0 since those don't require any additional computations to be inserted.
We may also want to handle constant splats of zero here, but I'm leaving that for future work when I have a real world example.
Differential Revision: https://reviews.llvm.org/D39911
llvm-svn: 317947
Summary:
The following kernel change has moved ET_DYN base to 0x4000000 on arm32:
https://marc.info/?l=linux-kernel&m=149825162606848&w=2
Switch to dynamic shadow base to avoid such conflicts in the future.
Reserve shadow memory in an ifunc resolver, but don't use it in the instrumentation
until PR35221 is fixed. This will eventually let use save one load per function.
Reviewers: kcc
Subscribers: aemerson, srhines, kubamracek, kristof.beyls, hiraditya, llvm-commits
Differential Revision: https://reviews.llvm.org/D39393
llvm-svn: 317943
Also change some default cases into llvm_unreachable in
WindowsResourceCOFFWriter, to make it easier to find if they
are triggerd from within e.g. lld, which supported ARM64 earlier
than llvm-cvtres did.
Differential Revision: https://reviews.llvm.org/D39892
llvm-svn: 317942
Refactors the behaviour of building graphs out of FileAnalysis, allowing for analysis of the GraphResult by the callee without having to rebuild the graph. Means when we want to analyse the constructed graph (planned for later revisions), we don't do repeated work.
Also makes CFI verification in FileAnalysis now return an enum that allows us to differentiate why something failed, not just that it did/didn't fail.
Reviewers: vlad.tsyrklevich
Subscribers: kcc, pcc, llvm-commits
Differential Revision: https://reviews.llvm.org/D39764
llvm-svn: 317927
Previously, debuginfo-tests was expected to be checked out into
clang/test and then the tests would automatically run as part of
check-clang. This is not a standard workflow for handling
external projects, and it brings with it some serious drawbacks
such as the inability to depend on things other than clang, which
we will need going forward.
The goal of this patch is to migrate towards a more standard
workflow. To ease the transition for build bot maintainers,
this patch tries not to break the existing workflow, but instead
simply deprecate it to give maintainers a chance to update
the build infrastructure.
Differential Revision: https://reviews.llvm.org/D39605
llvm-svn: 317925
Summary:
The specification of the @llvm.memcpy.element.unordered.atomic intrinsic requires
that the pointer arguments have alignments of at least the element size. The existing
IRBuilder interface to create a call to this intrinsic does not allow for providing
the alignment of these pointer args. Having an interface that makes it easy to
construct invalid intrinsic calls doesn't seem sensible, so this patch simply
adds the requirement that one provide the argument alignments when using IRBuilder
to create atomic memcpy calls.
llvm-svn: 317918
Summary:
This adds logic to CVP to remove some overflow checks. It uses LVI to remove
operations with at least one constant. Specifically, this can remove many
overflow intrinsics immediately following an overflow check in the source code,
such as:
if (x < INT_MAX)
... x + 1 ...
Patch by Joel Galenson!
Reviewers: sanjoy, regehr
Reviewed By: sanjoy
Subscribers: fhahn, pirama, srhines, llvm-commits
Differential Revision: https://reviews.llvm.org/D39483
llvm-svn: 317911
This patch adds the ability to include the member function declarations
in the instruction selector class separately from the member bodies.
Defining GET_DAGISEL_DECL macro to any value will only include the member
declarations. To include bodies, define GET_DAGISEL_BODY macro to be the
selector class name. Example:
class FooDAGToDAGISel : public SelectionDAGISel {
// Pull in declarations only.
#define GET_DAGISEL_DECL
#include "FooISelDAGToDAG.inc"
};
// Include the function bodies (with names qualified with the provided
// class name).
#define GET_DAGISEL_BODY FooDAGToDAGISel
#include "FooISelDAGToDAG.inc"
When neither of the two macros are defined, the function bodies are emitted
inline (in the same way as before this patch).
Differential Revision: https://reviews.llvm.org/D39596
llvm-svn: 317903
Summary:
This eliminates the boilerplate implementation of the iterator interface in
mapped_iterator.
This patch also adds unit tests that verify that the mapped function is applied
by operator* and operator->, and that references returned by the map function
are returned via operator*.
Reviewers: dblaikie, chandlerc
Subscribers: llvm-commits, mgorny
Differential Revision: https://reviews.llvm.org/D39855
llvm-svn: 317902
Summary:
I want to leverage this to clean up some of the code in clang. This will allow us to simplify D39521 which was trying to do some of the same.
If we accurately keep the code in Host.cpp synced with new CPUs added to compile-rt/libgcc we should be able to use this file as a proxy for what's implemented in the libraries.
The entries for the CPUs recognized by the libraries use separate macros that define additional parameters like the name for __builtin_cpu_is and an alias string for the couple cases where __builtin_cpu_is accepts two different names.
All of the macros contain an ARCHNAME that is usually the same as the __builtin_cpu_is string, but sometimes isn't. This represents the name recognized by X86.td and -march.
I'm following the precedent set by ARM and AArch64 and adding this information to lib/Support/TargetParser.cpp
Reviewers: erichkeane, echristo, asbirlea
Reviewed By: echristo
Subscribers: llvm-commits, aemerson, kristof.beyls
Differential Revision: https://reviews.llvm.org/D39782
llvm-svn: 317900
Summary:
LTO/Caching.cpp uses file rename to atomically set the value for a
cache key. On Windows, this fails when the destination file already
exists. Previously, LLVM would report_fatal_error in such
cases. However, because the old and the new value for the cache key
are supposed to be equivalent, it actually doesn't matter which one we
keep. This change makes it so that failing the rename when an openable
file with the desired name already exists causes us to report success
instead of fataling.
Reviewers: pcc, hans
Subscribers: mehdi_amini, llvm-commits, hiraditya
Differential Revision: https://reviews.llvm.org/D39874
llvm-svn: 317899
Change the test format for SVE assembler/disassembler tests to be less verbose and have both tests in the same file.
The tests check the following:
* All instructions are assembled correctly into the right encoding.
* All instructions are disassembled correctly (into the preferred assembly format)
* Without -mattr=+sve the instructions are not assembled.
* Without -mattr=+sve the instructions are not disassembled.
This patch also adds several negative tests for SVE add/sub.
Patch by Sander De Smalen.
Reviewed by: rengolin, fhahn
Differential Revision: https://reviews.llvm.org/D39792
llvm-svn: 317894
This change adds generic fuzzing tools capable of running libFuzzer tests on
any optimization pass or combination of them.
Differential Revision: https://reviews.llvm.org/D39555
llvm-svn: 317883
Summary:
The associated debug value is updated when the virtual source register
of a copy is completely eliminated and replaced with a rematerialize
value in the defed register of the copy. As the debug value now is
associated with another register it also need to be moved, otherwise
the debug value isn't valid.
Reviewers: aprantl
Reviewed By: aprantl
Subscribers: MatzeB, llvm-commits, qcolombet
Differential Revision: https://reviews.llvm.org/D38024
llvm-svn: 317880
* The method getRegAllocationHints() is now of bool type instead of void. If
true is returned, regalloc (AllocationOrder) will *only* try to allocate the
hints, as opposed to merely trying them before non-hinted registers.
* TargetRegisterInfo::getRegAllocationHints() is implemented for SystemZ with
an increase in number of LOCRs.
In this case, it is desired to force the hints even though there is a slight
increase in spilling, because if a non-hinted register would be allocated,
the LOCRMux pseudo would have to be expanded with a jump sequence. The LOCR
(Load On Condition) SystemZ instruction must have both operands in either the
low or high part of the 64 bit register.
Reviewers: Quentin Colombet and Ulrich Weigand
https://reviews.llvm.org/D36795
llvm-svn: 317879
r600 uses dummy pointer info for lowering load/store. Since dummy pointer info
assumes address space 0, this causes isel failure when temporary load/store SDNodes
are generated for amdgiz environment.
Since the offest is not constant, FixedStack pseudo source value cannot be used
to create the pointer info. This patch creates pointer info using llvm undef value.
At least this provides correct address space so that isel can be done correctly.
Differential Revision: https://reviews.llvm.org/D39698
llvm-svn: 317862
The pointer info for pseudo source for r600 is not correct when
alloca addr space is not 0, which causes invalid SDNode for r600---amdgiz.
This patch fixes that.
Differential Revision: https://reviews.llvm.org/D39670
llvm-svn: 317861
- Use ELF header flags to identify processor.
- Remove isa note record.
- Add target feature section.
- Make metadata for NumVGPRs, NumSGPRs and MaxFlatWorkGroupSize required.
- Add FixedWorkGroupSize to CodeProps metadata.
- Add ReqdWorkGroupSize* to kernel descriptor and move MaxFlatWorkGroupSize to be adjacent.
- Move IsXNACKEnabled in the kernel descriptor to be at the end of the unused flags.
- Remove IsDynamicCallStack from the metadata and kernel descriptor.
- Remove legacy debugger metadata.
- Remove old XNACK enabled processor names.
Differential Revision: https://reviews.llvm.org/D39828
llvm-svn: 317855
This change doesn't fix the root cause of the miscompile PR34966 as the root
cause is in the linker ld64. This change makes call graph more complete
allowing to have better module imports/exports.
rdar://problem/35344706
Reviewers: tejohnson
Reviewed By: tejohnson
Subscribers: mehdi_amini, inglorion, eraman, llvm-commits, hiraditya
Differential Revision: https://reviews.llvm.org/D39356
llvm-svn: 317853
Summary:
zturner suggested that mapped_file_region::init() on Windows seems to
create mappings that are larger than they need to be: Offset+Size
instead of Size. Indeed, that appears to be the case. I confirmed that
tests pass with mappings of just Size bytes, and fail with Size-1
bytes, suggesting that Size is indeed the correct value.
Reviewers: amccarth, zturner
Reviewed By: zturner
Subscribers: hiraditya, llvm-commits
Differential Revision: https://reviews.llvm.org/D39876
llvm-svn: 317850
Summary:
This wrapper checks if there is at least one non-zero weight before
setting the metadata.
Reviewers: davidxl
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D39872
llvm-svn: 317845
With this patch, you can now cross-compile for Windows
on non-Windows hosts.
Differential Revision: https://reviews.llvm.org/D39814
This allows cross-compiling for windows on other platforms.
llvm-svn: 317830
When the Constant Hoisting pass moves expensive constants into a
common block, it would assign a debug location equal to the last use
of that constant. While this is certainly intuitive, it places the
constant in an out-of-order location, according to the debug location
information. This produces out-of-order stepping when debugging
programs affected by this pass.
This patch creates in-order stepping behavior by merging the debug
locations for hoisted constants, and the new insertion point.
Patch by Matthew Voss!
Differential Revision: https://reviews.llvm.org/D38088
llvm-svn: 317827
Whenever LLVM_ENABLE_ABI_BREAKING_CHECKS is enabled, which
is usually the case for example when asserts are enabled,
Error's destructor does some additional checking to make sure
that that it does not represent an error condition and that it
was checked.
However, this is -- by definition -- not the likely codepath.
Some profiling shows that at least with some compilers, simply
calling assertIsChecked -- in a release build with full
optimizations -- can account for up to 15% of the entire
runtime of the program, even though this function should almost
literally be a no-op.
The problem is that the assertIsChecked function can be considered
too big to inline depending on the compiler's inliner. Since it's
unlikely to ever need to failure path though, we can move it out
of line and force it to not be inlined, so that the fast path
can be inlined.
In my test (using lld to link clang with CMAKE_BUILD_TYPE=Release
and LLVM_ENABLE_ASSERTIONS=ON), this reduces link time from 27
seconds to 23.5 seconds, which is a solid 15% gain.
llvm-svn: 317824
Summary:
The analysis of the store sequence goes in straight order - from the
first store to the last. Bu the best opportunity for vectorization will
happen if we're going to use reverse order - from last store to the
first. It may be best because usually users have some initialization
part + further processing and this first initialization may confuse
SLP vectorizer.
Reviewers: RKSimon, hfinkel, mkuper, spatel
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D39606
llvm-svn: 317821
The toxic stew of created values named 'tmp' and tests that already have
values named 'tmp' and CHECK lines looking for values named 'tmp' causes
bad things to happen in our test line auto-generation scripts because it
wants to use 'TMP' as a prefix for unnamed values. Use less 'tmp' to
avoid that.
llvm-svn: 317818
Summary: This fixes failure in CodeGen/AArch64/global-merge-group-by-use.ll uncovered by D39245.
Reviewers: ab, asl
Reviewed By: ab
Subscribers: aemerson, llvm-commits, kristof.beyls
Differential Revision: https://reviews.llvm.org/D39635
llvm-svn: 317817
The reassociate pass generates named values such as "%tmp2" which trips up the script's regex's
because the script uses a 'TMP' prefix for unnamed values (%2).
llvm-svn: 317810
We don't really need any special handling of "offsettable"
memory addresses, but since some existing code uses inline
asm statements with the "o" constraint, add support for this
constraint for compatibility purposes.
llvm-svn: 317807
Correct the definition of 'j' as being unavailable for microMIPS32R6 and
provide the 'b' assembly idiom for codegen purposes for microMIPS32r3.
Provide the necessary 'br' pattern for microMIPS32R6 as it now longer
incorrectly uses the 'j' instruction.
Reviewers: atanasyan
Differential Revision: https://reviews.llvm.org/D39741
llvm-svn: 317801
Summary:
This change allows yaml input to control the order of implicitly added sections
(`.symtab`, `.strtab`, `.shstrtab`). The order is controlled by adding a
placeholder section of the given name to the Sections field.
This change is to support changes in D39582, where it is desirable to control
the location of the `.dynsym` section.
This reapplied version fixes:
1. use of a function call within an assert
2. failing lld test which has an unnamed section
3. incorrect section count when given an unnamed section
Additionally, one more test to cover the unnamed section failure.
Reviewers: compnerd, jakehehrlich
Reviewed By: jakehehrlich
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D39749
llvm-svn: 317789
No existing processor has both so it doesn't really matter what we do here. But we were previously just relying on pattern order which gave FMA4 priority.
llvm-svn: 317775
Summary:
This will let ORC JIT clients plug in custom logic for the mmap, munmap and
mprotect paths.
Reviewers: loladiro, dblaikie
Subscribers: mcrosier, llvm-commits
Differential Revision: https://reviews.llvm.org/D39300
llvm-svn: 317770
Vedant Kumar