The bug reports that a vectorized loop is stepped through 4 times and each step through the loop seemed to show a different path. I found two problems here:
A) An incorrect line number on a preheader block (for.body.preheader) instruction causes a step into the loop before it begins.
B) Instructions in the middle block have different line numbers which give the impression of another iteration.
In this patch I give all of the middle block instructions the line number of the scalar loop latch terminator branch. This seems to provide the smoothest debugging experience because the vectorized loops will always end on this line before dropping into the scalar loop. To solve problem A I have altered llvm::SplitBlockPredecessors to accommodate loop header blocks.
I have set up a separate review D61933 for a fix which is required for this patch.
Jay Foad [Wed, 19 Jun 2019 10:28:48 +0000 (10:28 +0000)]
[ConstantFolding] Fix assertion failure on non-power-of-two vector load.
Summary:
The test case does an (out of bounds) load from a global constant with
type <3 x float>. InstSimplify tried to turn this into an integer load
of the whole alloc size of the vector, which is 128 bits due to
alignment padding, and then bitcast this to <3 x vector> which failed
an assertion due to the type size mismatch.
The fix is to do an integer load of the normal size of the vector, with
no alignment padding.
Lewis Revill [Wed, 19 Jun 2019 10:11:13 +0000 (10:11 +0000)]
[RISCV] Fix failure to parse parenthesized immediates
Since the parser attempts to parse an operand as a register with
parentheses before parsing it as an immediate, immediates in
parentheses should not be parsed by parseRegister. However in the case
where the immediate does not start with an identifier, the LParen is not
unlexed and so the RParen causes an unexpected token error.
This patch adds the missing UnLex, and modifies the existing UnLex to
not use a buffered token, as it should always be unlexing an LParen.
Lewis Revill [Wed, 19 Jun 2019 08:40:59 +0000 (08:40 +0000)]
[RISCV] Add lowering of global TLS addresses
This patch adds lowering for global TLS addresses for the TLS models of
InitialExec, GlobalDynamic, LocalExec and LocalDynamic.
LocalExec support required using a 4-operand add instruction, which uses
the fourth operand to express a relocation on the symbol. The necessary
fixup is emitted when the instruction is emitted.
This allows targets to make more decisions about reserved registers
after isel. For example, now it should be certain there are calls or
stack objects in the frame or not, which could have been introduced by
legalization.
Thomas Lively [Wed, 19 Jun 2019 00:02:13 +0000 (00:02 +0000)]
[WebAssembly] Optimize ISel for SIMD Boolean reductions
Summary:
Converting the result *.{all,any}_true to a bool at the source level
generates LLVM IR that compares the result to 0. This check is
redundant since these instructions already return either 0 or 1 and
therefore conform to the BooleanContents setting for WebAssembly. This
CL adds patterns to detect and remove such redundant operations on the
result of Boolean reductions.
Daniel Sanders [Tue, 18 Jun 2019 23:34:46 +0000 (23:34 +0000)]
Re-commit r363744: [tblgen][disasm] Allow multiple encodings to disassemble to the same instruction
It seems macOS lets you have ArrayRef<const X> even though this is apparently
forbidden by the language standard (Thanks MSVC++ for the clear error message).
Removed the problematic const's to fix this.
(It also seems I'm not receiving buildbot emails anymore and I'm trying to find
out why. In the mean time I'll be polling lab.llvm.org to hopefully see if/when
failures occur)
Summary:
Historically llvm-objdump prints the path to a dylib as well as the
dylib's compatibility version and current version number. This change
extends this information by adding the kind of dylib load: weak,
reexport, etc.
Daniel Sanders [Tue, 18 Jun 2019 21:56:04 +0000 (21:56 +0000)]
[tblgen][disasm] Allow multiple encodings to disassemble to the same instruction
Summary:
Add an AdditionalEncoding class which can be used to define additional encodings
for a given instruction. This causes the disassembler to add an additional
encoding to its matching tables that map to the specified instruction.
let Size = 4;
let Inst{0-7} = Reg;
let Inst{8-14} = 0;
let Inst{15} = 1; // Continuation bit
let Inst{16-31} = 0;
...
}
def : AdditionalEncoding<ADD1> {
bits<8> Reg;
bits<16> Inst; // You can also have bits<32> and it will still be a 16-bit encoding
let Size = 2;
let Inst{0-3} = 0;
let Inst{4-7} = Reg;
let Inst{8-15} = 0;
...
}
with those definitions, llvm-mc will successfully disassemble both of these:
0x01 0x00
0x10 0x80 0x00 0x00
to:
ADD1 r1
Michael Liao [Tue, 18 Jun 2019 21:41:13 +0000 (21:41 +0000)]
Recommit [SROA] Enhance SROA to handle `addrspacecast`ed allocas
[SROA] Enhance SROA to handle `addrspacecast`ed allocas
- Fix typo in original change
- Add additional handling to ensure all return pointers are properly
casted.
Summary:
- After `addrspacecast` is allowed to be eliminated in SROA, the
adjusting of storage pointer (from `alloca) needs to handle the
potential different address spaces between the storage pointer (from
alloca) and the pointer being used.
Huihui Zhang [Tue, 18 Jun 2019 20:55:09 +0000 (20:55 +0000)]
[ARM] Comply with rules on ARMv8-A thumb mode partial deprecation of IT.
Summary:
When identifing instructions that can be folded into a MOVCC instruction,
checking for a predicate operand is not enough, also need to check for
thumb2 function, with restrict-IT, is the machine instruction eligible for
ARMv8 IT or not.
Notes in ARMv8-A Architecture Reference Manual, section "Partial deprecation of IT"
https://usermanual.wiki/Pdf/ARM20Architecture20Reference20ManualARMv8.1667877052.pdf
"ARMv8-A deprecates some uses of the T32 IT instruction. All uses of IT that apply to
instructions other than a single subsequent 16-bit instruction from a restricted set
are deprecated, as are explicit references to the PC within that single 16-bit
instruction. This permits the non-deprecated forms of IT and subsequent instructions
to be treated as a single 32-bit conditional instruction."
Sam Elliott [Tue, 18 Jun 2019 20:38:08 +0000 (20:38 +0000)]
[RISCV] Prevent re-ordering some adds after shifts
Summary:
DAGCombine will normally turn a `(shl (add x, c1), c2)` into `(add (shl x, c2), c1 << c2)`, where `c1` and `c2` are constants. This can be prevented by a callback in TargetLowering.
On RISC-V, materialising the constant `c1 << c2` can be more expensive than materialising `c1`, because materialising the former may take more instructions, and may use a register, where materialising the latter would not.
This patch implements the hook in RISCVTargetLowering to prevent this transform, in the cases where:
- `c1` fits into the immediate field in an `addi` instruction.
- `c1` takes fewer instructions to materialise than `c1 << c2`.
In future, DAGCombine could do the check to see whether `c1` fits into an add immediate, which might simplify more targets hooks than just RISC-V.
Adrian Prantl [Tue, 18 Jun 2019 20:09:09 +0000 (20:09 +0000)]
Add debug location verification for !llvm.loop attachments.
This patch teaches the Verifier how to detect broken !llvm.loop
attachments as discussed in https://reviews.llvm.org/D60831. This
allows LLVM to warn and strip out the broken debug info before
attempting an LTO compilation with input generated by LLVM predating
https://reviews.llvm.org/rL361149.
Adrian Prantl [Tue, 18 Jun 2019 19:42:29 +0000 (19:42 +0000)]
Add debug location verification for !llvm.loop attachments.
This patch teaches the Verifier how to detect broken !llvm.loop
attachments as discussed in https://reviews.llvm.org/D60831. This
allows LLVM to warn and strip out the broken debug info before
attempting an LTO compilation with input generated by LLVM predating
https://reviews.llvm.org/rL361149.
Reid Kleckner [Tue, 18 Jun 2019 19:41:25 +0000 (19:41 +0000)]
[PDB] Ignore .debug$S subsections with high bit set
Some versions of the Visual C++ 2015 runtime have line tables with the
subsection kind of 0x800000F2. In cvinfo.h, 0x80000000 is documented to
be DEBUG_S_IGNORE. This appears to implement the intended behavior.
Craig Topper [Tue, 18 Jun 2019 19:04:03 +0000 (19:04 +0000)]
[X86] Remove unnecessary line that makes v4f32 FP_ROUND Legal. NFC
FP_ROUND defaults to Legal for all MVT types and nothing changes
the v4f32 entry way from this default. If we needed this line
we'd also need one for v8f32 with AVX512 which we don't have.
Other than adding consistent demanded elts handling which was a trivial addition, the other differences in functionality will be added in later patches.
Michael Liao [Tue, 18 Jun 2019 17:58:49 +0000 (17:58 +0000)]
[SROA] Enhance SROA to handle `addrspacecast`ed allocas
Summary:
- After `addrspacecast` is allowed to be eliminated in SROA, the
adjusting of storage pointer (from `alloca) needs to handle the
potential different address spaces between the storage pointer (from
alloca) and the pointer being used.
Other than adding consistent demanded elts handling which was a trivial addition, the other differences in functionality will be added in later patches.
Simon Atanasyan [Tue, 18 Jun 2019 16:59:57 +0000 (16:59 +0000)]
[mips] Add PTR_64 and GPR_64 predicates to some MIPS 64-bit instructions
Add `IsGP64bit` and `IsPTR64bit` to the list of `UnsupportedFeatures`
of the P5600 scheduling definitions. Also mark some MIPS 64-bit
instructions by PTR_64 and GPR_64 predicates. This reduces number
of "No schedule information for" and "lacks information for" errors
in case of marking this scheduler model as complete.
This patch is one of a series of patches. The goal is to make P5600
scheduler model complete and turn on the `CompleteModel` flag.
Simon Atanasyan [Tue, 18 Jun 2019 16:59:47 +0000 (16:59 +0000)]
[mips] Set the hasNoSchedulingInfo flag for the `MipsAsmPseudoInst`
Set the hasNoSchedulingInfo flag for the`MipsAsmPseudoInst`. These
pseudo-instructions are never used by codegen. This flag allows to
reduce number of "No schedule information for" and "lacks information
for" errors in case of marking a scheduler model as complete.
This patch is one of a series of patches. The goal is to make P5600
scheduler model complete and turn on the `CompleteModel` flag.
Adrian McCarthy [Tue, 18 Jun 2019 16:36:57 +0000 (16:36 +0000)]
Fix some lit test ResourceWarnings on Windows
When running LLDB lit tests on Windows, the system selects a debug version
of Python, which was issuing lots of ResourceWarnings about files that
weren't closed. There are two kinds of them, and each test triggered one
of each.
This patch fixes one kind by ensuring TestRunner explicitly close the
temporary files created for routing stderr. This is important on Windows
but has no net effect on Posix systems.
The remaining ResourceWarnings are more elusive; the bug may lie in
the Python library subprocess.py, and it may be Windows-specific.
Simon Tatham [Tue, 18 Jun 2019 16:19:59 +0000 (16:19 +0000)]
[ARM] Add MVE vector shift instructions.
This includes saturating and non-saturating shifts, both with
immediate shift count and with the shift counts given by another
vector register; VSHLC (in which the bits shifted out of each active
vector lane are shifted in to the next active lane); and also VMOVL,
which is enough like an immediate shift that it didn't fit too badly
in this category.
Summary:
These form a small family of their own, to go with the floating-point
VMINNM/VMAXNM instructions added in a previous commit.
They introduce the first of many special cases in the mnemonic
recognition code, because VMIN with the E suffix used by the VPT
predication system needs to avoid being interpreted as the nonexistent
instruction 'VMI' with an ordinary 'NE' condition suffix.
Part of fixing the X86 regression noted in D63281 - I've split this into X86 and generic parts - the generic commit will be coming shortly and will fix the vector-reduce-mul-widen.ll regression introduced here.
Simon Tatham [Tue, 18 Jun 2019 15:05:42 +0000 (15:05 +0000)]
[ARM] Rename MVE instructions in Tablegen for consistency.
Summary:
Their names began with a mishmash of `MVE_`, `t2` and no prefix at
all. Now they all start with `MVE_`, which seems like a reasonable
choice on the grounds that (a) NEON is the thing they're most at risk
of being confused with, and (b) MVE implies Thumb-2, so a prefix
indicating MVE is strictly more specific than one indicating Thumb-2.
Fangrui Song [Tue, 18 Jun 2019 14:01:03 +0000 (14:01 +0000)]
[llvm-readobj] Allow --hex-dump/--string-dump to dump multiple sections
1) `-x foo` currently dumps one `foo`. This change makes it dump all `foo`.
2) `-x foo -x foo` currently dumps `foo` twice. This change makes it dump `foo` once.
In addition, if foo has section index 9, `-x foo -x 9` dumps `foo` once.
3) Give a warning instead of an error if `foo` does not exist.
The new behaviors match GNU readelf.
Also, print a new line as a separator between two section dumps.
GNU readelf uses two lines, but one seems good enough.
Andrea Di Biagio [Tue, 18 Jun 2019 12:59:46 +0000 (12:59 +0000)]
[MCA] Slightly refactor the bottleneck analysis view. NFCI
This patch slightly refactors data structures internally used by the bottleneck
analysis to track data and resource dependencies.
This patch also updates methods used to print out information about dependency
edges when in debug mode.
This is the last of a sequence of commits done in preparation for an upcoming
patch that fixes PR37494. No functional change intended.
Matt Arsenault [Tue, 18 Jun 2019 12:48:36 +0000 (12:48 +0000)]
AMDGPU: Change API for checking for exec modification
Invert the name and return value to better reflect the imprecise
nature.
Force passing in the DefMI, since it's known in the 2 users and could
possibly fail for an arbitrary vreg.
Allow specifying a specific user instruction. Scan through use
instructions, instead of use operands. Add scan thresholds instead of
searching infinitely.
Stop using a set to track seen uses. I didn't understand this usage,
or why it would not check the last use. I don't think the use list has
any particular order.
Graham Hunter [Tue, 18 Jun 2019 10:11:56 +0000 (10:11 +0000)]
[SVE][IR] Scalable Vector IR Type with pr42210 fix
Recommit of D32530 with a few small changes:
- Stopped recursively walking through aggregates in
the verifier, so that we don't impose too much
overhead on large modules under LTO (see PR42210).
- Changed tests to match; the errors are slightly
different since they only report the array or
struct that actually contains a scalable vector,
rather than all aggregates which contain one in
a nested member.
- Corrected an older comment
Diogo N. Sampaio [Tue, 18 Jun 2019 10:04:36 +0000 (10:04 +0000)]
[NFC] Improve triple match of scripts that update tests
Summary:
The prior behavior of the triple matcher would stop
in the first matched triple. It was not possible to
create specific matches for sub-sets of a triple
(e.g aarch64-apple-darwin would never be used after
aarch64 was matched).
This patch:
1) Allows that specialized triples take priority,
considering that the string lenght of the triple
indentifies how specialized a triple is. If two
triples of same lenght match, the one matched first
prevails, preserving the old behavior.
2) Remove 20 duplicated triples of arm, thumb,
aarch64 options with same arguments, matching
the common prefix (aarch64, arm, thumb) of them.
3) Creates three new function matching regexes and
five triple options for arm64-apple-ios,
(arm|thumb)-apple-ios and thumb(v5)?-macho
Simon Pilgrim [Tue, 18 Jun 2019 09:50:13 +0000 (09:50 +0000)]
[X86] Replace any_extend* vector extensions with zero_extend* equivalents
First step toward addressing the vector-reduce-mul-widen.ll regression in D63281 - we should replace ANY_EXTEND/ANY_EXTEND_VECTOR_INREG in X86ISelDAGToDAG to avoid having to add duplicate patterns when treating any extensions as legal.
In future patches this will also allow us to keep any extension nodes around a lot longer in the DAG, which should mean that we can keep better track of undef elements that otherwise become zeros that we think we have to keep......
Jeremy Morse [Tue, 18 Jun 2019 08:52:38 +0000 (08:52 +0000)]
[DebugInfo][Docs] Document that prologue/epilogue variable location changes are ignored
This patch documents that LLVM does not describe all changes in variable
locations during the prologue and the epilogue. The debugger doesn't /
shouldn't step through that portion of the function anyway, and describing
every location through such stages would bloat location lists.
Perform some minor cleanup at the same time,
* Fix an enumerated list
* Document that dbg.declare intrinsics have their variable location recorded
in a MachineFunction table, not with DBG_VALUE meta-insts
* Adds frame-indexes to the list of things that can be operands to
DBG_VALUEs.
Craig Topper [Tue, 18 Jun 2019 03:23:11 +0000 (03:23 +0000)]
[X86] Introduce new MOVSSrm/MOVSDrm opcodes that use VR128 register class.
Rename the old versions that use FR32/FR64 to MOVSSrm_alt/MOVSDrm_alt.
Use the new versions in patterns that previously used a COPY_TO_REGCLASS
to VR128. These patterns expect the upper bits to be zero. The
current set up appears to work, but I'm not sure we should be
enforcing upper bits being zero through a COPY_TO_REGCLASS.
I wanted to flip the arrangement and use a COPY_TO_REGCLASS to
FR32/FR64 for the patterns that need an f32/f64 result, but that
complicated fastisel and globalisel.
I've been doing some experiments with reducing some isel patterns
and ended up in a situation where I had a
(SUBREG_TO_REG (COPY_TO_RECLASS (VMOVSSrm), VR128)) and our
post-isel peephole was unable to avoid using an instruction for
the SUBREG_TO_REG due to the COPY_TO_REGCLASS. Having a VR128
instruction removes the COPY_TO_REGCLASS that was breaking this.
Tom Stellard [Tue, 18 Jun 2019 02:05:06 +0000 (02:05 +0000)]
GlobalISel: Remove redundant pass initialization
Summary:
All the GlobalISel passes are initialized when the target calls
initializeGlobalISel(), so we don't need to call the initializers
from the pass constructors.
Alex Brachet [Tue, 18 Jun 2019 00:39:10 +0000 (00:39 +0000)]
[llvm-strip] Error when using stdin twice
Summary: Implements bug [[ https://bugs.llvm.org/show_bug.cgi?id=42204 | 42204 ]]. llvm-strip now warns when the same input file is used more than once, and errors when stdin is used more than once.
hwasan: Use bits [3..11) of the ring buffer entry address as the base stack tag.
This saves roughly 32 bytes of instructions per function with stack objects
and causes us to preserve enough information that we can recover the original
tags of all stack variables.
Now that stack tags are deterministic, we no longer need to pass
-hwasan-generate-tags-with-calls during check-hwasan. This also means that
the new stack tag generation mechanism is exercised by check-hwasan.
hwasan: Add a tag_offset DWARF attribute to instrumented stack variables.
The goal is to improve hwasan's error reporting for stack use-after-return by
recording enough information to allow the specific variable that was accessed
to be identified based on the pointer's tag. Currently we record the PC and
lower bits of SP for each stack frame we create (which will eventually be
enough to derive the base tag used by the stack frame) but that's not enough
to determine the specific tag for each variable, which is the stack frame's
base tag XOR a value (the "tag offset") that is unique for each variable in
a function.
In IR, the tag offset is most naturally represented as part of a location
expression on the llvm.dbg.declare instruction. However, the presence of the
tag offset in the variable's actual location expression is likely to confuse
debuggers which won't know about tag offsets, and moreover the tag offset
is not required for a debugger to determine the location of the variable on
the stack, so at the DWARF level it is represented as an attribute so that
it will be ignored by debuggers that don't know about it.
Amara Emerson [Mon, 17 Jun 2019 23:20:29 +0000 (23:20 +0000)]
[GlobalISel][Localizer] Rewrite localizer to run in 2 phases, inter & intra block.
Inter-block localization is the same as what currently happens, except now it
only runs on the entry block because that's where the problematic constants with
long live ranges come from.
The second phase is a new intra-block localization phase which attempts to
re-sink the already localized instructions further right before one of the
multiple uses.
One additional change is to also localize G_GLOBAL_VALUE as they're constants
too. However, on some targets like arm64 it takes multiple instructions to
materialize the value, so some additional heuristics with a TTI hook have been
introduced attempt to prevent code size regressions when localizing these.
Overall, these changes improve CTMark code size on arm64 by 1.2%.
This is part of the approved D63204 pending parent revision.
This small change is in fact a part of the VOP2b legalization which
does not technically belong to wave32 support, so extracted
separately.
Philip Reames [Mon, 17 Jun 2019 21:06:17 +0000 (21:06 +0000)]
Teach getSCEVAtScope how to handle loop phis w/invariant operands in loops w/taken backedges
This patch really contains two pieces:
Teach SCEV how to fold a phi in the header of a loop to the value on the backedge when a) the backedge is known to execute at least once, and b) the value is safe to use globally within the scope dominated by the original phi.
Teach IndVarSimplify's rewriteLoopExitValues to allow loop invariant expressions which already exist (and thus don't need new computation inserted) even in loops where we can't optimize away other uses.
projects / llvm / log