fma4 instructions zero the upper bits of the xmm register. fma3 instructions leave the bits unmodified. This requires separate builtins for the different semantics.
While we're cleaning up the scalar builtins this also removes the fma3 fmsub/fnmadd/fnmsub builtins by using negates in the header file.
llvm-svn: 318985
Summary:
These instructions zero the non-scalar part of the lower 128-bits which makes them different than the FMA3 instructions which pass through the non-scalar part of the lower 128-bits.
I've only added fmadd because we should be able to derive all other variants using operand negation in the intrinsic header like we do for AVX512.
I think there are still some missed negate folding opportunities with the FMA4 instructions in light of this behavior difference that I hadn't noticed before.
I've split the tests so that we can use different intrinsics for scalar testing between the two. I just copied the tests split the RUN lines and changed out the scalar intrinsics.
fma4-fneg-combine.ll is a new test to make sure we negate the fma4 intrinsics correctly though there are a couple TODOs in it.
Reviewers: RKSimon, spatel
Reviewed By: RKSimon
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D39851
llvm-svn: 318984
Summary:
This adds a new fast gather feature bit to cover all CPUs that support fast gather that we can use independent of whether the AVX512 feature is enabled. I'm only using this new bit to qualify AVX2 codegen. AVX512 is still implicitly assuming fast gather to keep tests working and to match the scatter behavior.
Test command lines have been added for these two cases.
Reviewers: magabari, delena, RKSimon, zvi
Reviewed By: RKSimon
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D40282
llvm-svn: 318983
Summary:
Currently ScalarizeVecRes_SETCC checks for the result type being a vector and jumps to ScalarizeVecRes_VSETCC. But if we're scalarizing a vector result, aren't we guaranteed to be looking at a vector type?
This patch deletes the current ScalarizeVecRes_SETCC and renames ScalarizeVecRes_VSETCC to ScalarizeVecRes_SETCC.
Reviewers: RKSimon, arsenm, eladcohen, zvi
Reviewed By: RKSimon
Subscribers: wdng, llvm-commits
Differential Revision: https://reviews.llvm.org/D40452
llvm-svn: 318982
Summary:
InitializeSwiftDemangler() attempts to resolve the
swift_demangle symbol. If this is not available, we
observe dlerror message leak.
Caught on NetBSD/amd64 in TSan.
Sponsored by <The NetBSD Foundation>
Reviewers: joerg, kubamracek, vitalybuka, dvyukov, eugenis
Reviewed By: dvyukov
Subscribers: llvm-commits, #sanitizers
Tags: #sanitizers
Differential Revision: https://reviews.llvm.org/D40382
llvm-svn: 318980
Teach the retain-count checker that CoreMedia reference types use
CoreFoundation-style reference counting. This enables the checker
to catch leaks and over releases of those types.
rdar://problem/33599757
llvm-svn: 318979
Summary:
clang-format does not collapse short records, interfaces, unions, etc.,
but fails to do so if the record is preceded by certain modifiers
(export, default, abstract, declare). This change skips over all
modifiers, and thus handles all record definitions uniformly.
Before:
export class Foo { bar: string; }
class Baz {
bam: string;
}
After:
export class Foo {
bar: string;
}
class Baz {
bam: string;
}
Reviewers: djasper
Subscribers: klimek
Differential Revision: https://reviews.llvm.org/D40430
llvm-svn: 318976
Summary:
TypeScript generic type arguments can contain object (literal) types,
which in turn can contain semicolons:
const x: Array<{a: number; b: string;} = [];
Previously, clang-format would incorrectly categorize the braced list as
a block and terminate the line at the openening `{`, and then format the
entire expression badly.
With this change, clang-format recognizes `<` preceding a `{` as
introducing a type expression. In JS, `<` comparison with an object
literal can never be true, so the chance of introducing false positives
here is very low.
Reviewers: djasper
Subscribers: klimek, cfe-commits
Differential Revision: https://reviews.llvm.org/D40424
llvm-svn: 318975
Summary:
Automatic Semicolon Insertion in clang-format tries to guess if a line
wrap should insert an implicit semicolong. The previous heuristic would
not trigger ASI if a token was immediately preceded by an `@` sign:
function foo(@Bar // <-- does not trigger due to preceding @
baz) {}
However decorators can have arbitrary parameters:
function foo(@Bar(param, param, param) // <-- precending @ missed
baz) {}
While it would be possible to precisely find the matching `@`, just
conversatively disabling ASI for the entire line is simpler, while also
not regressing ASI substatially.
Reviewers: djasper
Subscribers: klimek, cfe-commits
Differential Revision: https://reviews.llvm.org/D40410
llvm-svn: 318973
v4i32 isn't a legal type with sse1 only and would end up getting scalarized otherwise.
This isn't completely ideal as it doesn't handle cases like v8i32 that would get split to v4i32. But it at least helps with code written using the clang intrinsic header.
llvm-svn: 318967
The previous commit had the condition in the do/while backwards.
Debug builds currently print out low level details of the Knuth division algorithm when -debug is used. This information isn't useful in most cases and just adds noise to the log.
This adds a new preprocessor flag to enable the prints in the knuth division code in APInt.
Differential Revision: https://reviews.llvm.org/D40404
llvm-svn: 318966
This optimization can occur after type legalization and emit a vselect with v4i32 type. But that type is not legal with sse1. This ultimately gets scalarized by the second type legalization that runs after vector op legalization, but that's really intended to handle the scalar types that might be introduced by legalizing vector ops.
For now just stop this from happening by disabling the optimization with sse1.
llvm-svn: 318965
Debug builds currently print out low level details of the Knuth division algorithm when -debug is used. This information isn't useful in most cases and just adds noise to the log.
This adds a new preprocessor flag to enable the prints in the knuth division code in APInt.
Differential Revision: https://reviews.llvm.org/D40404
llvm-svn: 318963
Summary:
Wrapping between the type name and the array type indicator creates
invalid syntax in TypeScript.
Before:
const xIsALongIdent:
YJustBarelyFitsLinex
[]; // illegal syntax.
After:
const xIsALongIdent:
YJustBarelyFitsLinex[];
Reviewers: djasper
Subscribers: klimek
Differential Revision: https://reviews.llvm.org/D40436
llvm-svn: 318959
Summary: The same rules apply as for `return`.
Reviewers: djasper
Subscribers: klimek
Differential Revision: https://reviews.llvm.org/D40431
llvm-svn: 318958
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!
This fixed version relands r318032 which was reverted in r318049 due to
sanitizer buildbot failures.
Reviewers: john.brawn
Differential Revision: https://reviews.llvm.org/D39841
llvm-svn: 318956
This patch extends the recent work in optimizeMemoryInst to make it able to
combine more ExtAddrMode fields than just the BaseReg.
This fixes some benchmark regressions introduced by r309397, where GVN PRE is
hoisting a getelementptr such that it can no longer be combined into the
addressing mode of the load or store that uses it.
Differential Revision: https://reviews.llvm.org/D38133
llvm-svn: 318949
This patch fixes an issue where microMIPS ASE flag is not set
when a function has micromips attribute or when .set micromips
directive is used.
Differential Revision: https://reviews.llvm.org/D40316
llvm-svn: 318948
Summary: Preambles are large and we should avoid copying them.
Reviewers: bkramer, klimek
Reviewed By: bkramer
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D40302
llvm-svn: 318945
Summary:
In-memory preambles will not be copied anymore, so we need to make
sure they outlive the AST.
Reviewers: bkramer, sammccall, klimek
Reviewed By: sammccall
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D40301
llvm-svn: 318944
Summary: Provide default implementations so that only getCompileCommands() is mandatory.
Reviewers: ioeric
Subscribers: cfe-commits, bkramer, klimek
Differential Revision: https://reviews.llvm.org/D40409
llvm-svn: 318943
Summary:
Previously, clang-format would drop a space character between `of` and
then following (non-identifier) token if the preceding token was part of
a destructuring assignment (`}` or `]`).
Before:
for (const [a, b] of[]) {}
After:
for (const [a, b] of []) {}
Reviewers: djasper
Subscribers: klimek
Differential Revision: https://reviews.llvm.org/D40411
llvm-svn: 318942
LLD uses .bss.rel.ro for read-only copy relocations whereas the ld.bfd and
gold linkers use .data.rel.ro. In some linker scripts including ld.bfd's
internal linker script, the relro sections are placed sequentially assuming
.data.rel.ro is used. LLD's use of .bss.rel.ro means that the copy
relocations get matched into the .bss section causing the relro sections to
be non-contiguous.
This change checks for a .data.rel.ro OutputSection when a linker script
with the SECTIONS command is used. The section will match in the
.data.rel.ro output section and will maintain contiguous relro.
Differential Revision: https://reviews.llvm.org/D40365
Fixes PR35265
llvm-svn: 318940
It now has a DT_NEEDED that could be removed by --gc-sections and one
that cannot. Without this all tests would pass if --gc-sections just
removed all DT_NEEDED.
llvm-svn: 318937
Craig Topper