Commit Graph

167548 Commits

Author SHA1 Message Date
Matt Arsenault ce6d61fba8 AMDGPU: Conversions always produce canonical results
Not sure why this was checking for denormals for f16.
My interpretation of the IEEE standard is conversions
should produce a canonical result, and the ISA manual
says denormals are created when appropriate.

llvm-svn: 339064
2018-08-06 21:51:52 +00:00
Philip Reames 94b29601ef [LICM] Further strengthen tests for hoisting guards and invariant.starts [NFC]
llvm-svn: 339062
2018-08-06 21:39:43 +00:00
Matt Arsenault f8768bfc84 AMDGPU: Fix implementation of isCanonicalized
If denormals are enabled, denormals are canonical.
Also fix a few other issues. minnum/maxnum are supposed
to canonicalize. Temporarily improve workaround for the
instruction behavior change in gfx9.

Handle selects and fcopysign.

The tests were also largely broken, since they were
checking for a flush used on some targets after the
store of the result.

llvm-svn: 339061
2018-08-06 21:38:27 +00:00
Reid Kleckner 5327805d7c Fix a -Wsign-compare
llvm-svn: 339059
2018-08-06 21:26:47 +00:00
Philip Reames 9d7bb2f700 [LICM] Strengthen invariant.start hoisting tests [NFC]
llvm-svn: 339057
2018-08-06 21:18:34 +00:00
Reid Kleckner 15e91c3235 [X86] Fix assertion in subreg extraction
This assert fires when attempting to extract a subregister from the
global PIC base register. This virtual register SD node is not in the
VRBaseMap, so we shouldn't call getVR to look it up there. If this is a
RegisterSDNode, we should be able to use the virtual register directly.

Fixes PR38385

llvm-svn: 339056
2018-08-06 21:16:16 +00:00
Philip Reames 81c7dc93d2 [LICM] Add tests highlighting missing hoists for intrinsics [NFC]
llvm-svn: 339054
2018-08-06 21:06:15 +00:00
Evandro Menezes 6e137cb9f0 [SLC] Fix shrinking of pow()
Properly shrink `pow()` to `powf()` as a binary function and, when no other
simplification applies, do not discard it.

Differential revision: https://reviews.llvm.org/D50113

llvm-svn: 339046
2018-08-06 19:40:17 +00:00
Alexandre Ganea 741cc3531a [llvm-pdbutil] Support PDBs without a DBI stream
Differential Revision: https://reviews.llvm.org/D50258

llvm-svn: 339045
2018-08-06 19:35:00 +00:00
Easwaran Raman 10fd92dd94 [X86] Recognize a splat of negate in isFNEG
Summary:
Expand isFNEG so that we generate the appropriate F(N)M(ADD|SUB)
instructions in more cases. For example, the following sequence
a = _mm256_broadcast_ss(f)
d = _mm256_fnmadd_ps(a, b, c)

generates an fsub and fma without this patch and an fnma with this
change.

Reviewers: craig.topper

Subscribers: llvm-commits, davidxl, wmi

Differential Revision: https://reviews.llvm.org/D48467

llvm-svn: 339043
2018-08-06 19:23:38 +00:00
Craig Topper 0076477a4c [X86] When using "and $0" and "orl $-1" to store 0 and -1 for minsize, make sure the store isn't volatile
If the store is volatile this might be a memory mapped IO access. In that case we shouldn't generate a load that didn't exist in the source

Differential Revision: https://reviews.llvm.org/D50270

llvm-svn: 339041
2018-08-06 18:44:26 +00:00
Craig Topper f8a8c746e3 [X86] Add test cases to show bad use of "and $0" and "orl $-1" for minsize when the store is volatile
If the store is volatile we shouldn't be adding a little that didn't exist in the source.

llvm-svn: 339040
2018-08-06 18:44:21 +00:00
Wei Mi 3c1c088500 [RegisterCoalescer] Delay live interval update work until the rematerialization
for all the uses from the same def is done.

We run into a compile time problem with flex generated code combined with
`-fno-jump-tables`. The cause is that machineLICM hoists a lot of invariants
outside of a big loop, and drastically increases the compile time in global
register splitting and copy coalescing.  https://reviews.llvm.org/D49353
relieves the problem in global splitting. This patch is to handle the problem
in copy coalescing.

About the situation where the problem in copy coalescing happens. After
machineLICM, we have several defs outside of a big loop with hundreds or
thousands of uses inside the loop. Rematerialization in copy coalescing
happens for each use and everytime rematerialization is done, shrinkToUses
will be called to update the huge live interval. Because we have 'n' uses
for a def, and each live interval update will have at least 'n' complexity,
the total update work is n^2.

To fix the problem, we try to do the live interval update work in a collective
way. If a def has many copylike uses larger than a threshold, each time
rematerialization is done for one of those uses, we won't do the live interval
update in time but delay that work until rematerialization for all those uses
are completed, so we only have to do the live interval update work once.

Delaying the live interval update could potentially change the copy coalescing
result, so we hope to limit that change to those defs with many
(like above a hundred) copylike uses, and the cutoff can be adjusted by the
option -mllvm -late-remat-update-threshold=xxx.

Differential Revision: https://reviews.llvm.org/D49519

llvm-svn: 339035
2018-08-06 17:30:45 +00:00
Daniil Fukalov 0d3b0f4f72 Fix typo in the MSVC Visualizer for SmallVector class
llvm-svn: 339029
2018-08-06 16:47:24 +00:00
Owen Reynolds a489d11158 Fix raw_fd_ostream::write_impl hang due to an infinite loop with large output
On windows when raw_fd_ostream::write_impl calls write, a 32 bit input is required for character count. As a variable with size_t is used for this argument, on x64 integral demotion occurs. In the case of large files an infinite loop follows. 
See: https://bugs.llvm.org/show_bug.cgi?id=37926
This fix allows the output of files larger than the previous int32 limit.

Differential Revision: https://reviews.llvm.org/D48948

llvm-svn: 339027
2018-08-06 16:21:41 +00:00
Matt Arsenault 0d1b3934e2 AMDGPU: Fold v_lshl_or_b32 with 0 src0
Appears from expansion of some packed cases.

llvm-svn: 339025
2018-08-06 15:40:20 +00:00
Matt Arsenault 56b31d8d75 ValueTracking: Handle canonicalize in CannotBeNegativeZero
Also fix apparently missing test coverage for any of the
handling here.

llvm-svn: 339023
2018-08-06 15:16:26 +00:00
Matt Arsenault dbf77c5b41 AMDGPU: Rename check prefixes in test
Will avoid noisy diff in future change.

llvm-svn: 339022
2018-08-06 15:16:12 +00:00
David Bolvansky 1e51e6896f [NFC] Fixed unused function warnings
llvm-svn: 339021
2018-08-06 15:09:15 +00:00
David Bolvansky 3d2653bd39 Revert unused function fix
llvm-svn: 339020
2018-08-06 15:05:51 +00:00
David Bolvansky 6bca938bf0 [NFC] Fixed unused function warning
llvm-svn: 339019
2018-08-06 14:42:07 +00:00
Bryan Chan e023706471 [AArch64] Fix assertion failure on widened f16 BUILD_VECTOR
Summary:
Ensure that NormalizedBuildVector returns a BUILD_VECTOR with operands of the
same type. This fixes an assertion failure in VerifySDNode.

Reviewers: SjoerdMeijer, t.p.northover, javed.absar

Reviewed By: SjoerdMeijer

Subscribers: kristof.beyls, llvm-commits

Differential Revision: https://reviews.llvm.org/D50202

llvm-svn: 339013
2018-08-06 14:14:41 +00:00
Tim Northover b376645b83 Fix modules build with different technique to suppress Knuth debugging
Currently we use #pragma push_macro(LLVM_DEBUG) to fiddle with the LLVM_DEBUG
macro so that we can silence debugging the Knuth division algorithm unless it's
actually desired. Unfortunately this is incompatible with enabling modules
while building LLVM (via LLVM_ENABLE_MODULES=ON), probably due to a bug being
fixed by D33004.

llvm-svn: 339009
2018-08-06 11:43:11 +00:00
Tim Northover 9956e4a24b ARM-MachO: don't add Thumb bit for addend to non-external relocation.
ld64 supplies its own Thumb bit for Thumb functions, and intentionally zeroes
out that part of any addend in an object file. But it only does that for
symbols marked N_EXT -- i.e. external symbols. So LLVM should avoid setting
that extra bit in other cases.

llvm-svn: 339007
2018-08-06 11:32:44 +00:00
Max Kazantsev 2dbbd64cb7 Re-enable "[ValueTracking] Teach isKnownNonNullFromDominatingCondition about AND"
The patch was reverted because of bug detected by sanitizer. The bug is fixed,
respective tests added.

Differential Revision: https://reviews.llvm.org/D50172

llvm-svn: 339005
2018-08-06 11:14:18 +00:00
Chandler Carruth c3cc1ec6b1 [docs] Continue working around broken Sphinx parsing of LLVM code blocks
by switching to `text` in another place.

llvm-svn: 339001
2018-08-06 10:20:05 +00:00
Chandler Carruth f98035ce65 [docs] Remove the `dso_local` tag from these functions.
The sphinx build bot is erroring on these examples for some unknown
reason, and really the dso_local doesn't seem to be relevant to the
example in any way so its cleaner to omit it. And now they will look
a bit more like other (successful) IR examples.

llvm-svn: 338998
2018-08-06 10:03:25 +00:00
Chandler Carruth 3a56e3f028 [docs] Stop trying to parse the ThinLTO summary IR fragments with the
`llvm` syntax in Sphinx. This appears to just fail and create errors on
the docs buildbot.

llvm-svn: 338997
2018-08-06 09:46:59 +00:00
Max Kazantsev 3271f379a9 Revert rL338990 to see if it causes sanitizer failures
Multiple failues reported by sanitizer-x86_64-linux, seem to be caused by this
patch. Reverting to see if they sustain without it.

Differential Revision: https://reviews.llvm.org/D50172

llvm-svn: 338994
2018-08-06 08:10:28 +00:00
Max Kazantsev 778f62bb46 Try to fix buildbot
llvm-svn: 338991
2018-08-06 06:35:21 +00:00
Max Kazantsev 34b0666be9 [ValueTracking] Teach isKnownNonNullFromDominatingCondition about AND
`isKnownNonNullFromDominatingCondition` is able to prove non-null basing on `br` or `guard`
by `%p != null` condition, but is unable to do so basing on `(%p != null) && %other_cond`.
This patch allows it to do so.

Differential Revision: https://reviews.llvm.org/D50172
Reviewed By: reames

llvm-svn: 338990
2018-08-06 06:11:36 +00:00
Max Kazantsev eded4abef8 [GuardWidening] Widen guards with conditions of frequently taken dominated branches
If there is a frequently taken branch dominated by a guard, and its condition is available
at the point of the guard, we can widen guard with condition of this branch and convert
the branch into unconditional:

  guard(cond1)
  if (cond2) {
    // taken in 99.9% cases
    // do something
  } else {
    // do something else    
  }

Converts to

  guard(cond1 && cond2)
  // do something

Differential Revision: https://reviews.llvm.org/D49974
Reviewed By: reames

llvm-svn: 338988
2018-08-06 05:49:19 +00:00
Xin Tong 77931caec5 [NFC] Fix typo
llvm-svn: 338987
2018-08-06 05:03:21 +00:00
David Bolvansky 1a56ac790a [NFC] Fixed unused function warning
llvm-svn: 338986
2018-08-06 04:45:46 +00:00
Hsiangkai Wang ef72e481ea [DebugInfo] Refactor DbgInfoIntrinsic class hierarchy.
In the past, DbgInfoIntrinsic has a strong assumption that these
intrinsics all have variables and expressions attached to them.
However, it is too strong to derive the class for other debug entities.
Now, it has problems for debug labels.

In order to make DbgInfoIntrinsic as a base class for 'debug info', I
create a class for 'variable debug info', DbgVariableIntrinsic.

DbgDeclareInst, DbgAddrIntrinsic, and DbgValueInst will be derived from it.

Differential Revision: https://reviews.llvm.org/D50220

llvm-svn: 338984
2018-08-06 03:59:47 +00:00
Chandler Carruth 4a73aa112b [docs] Switch debug info metadata blocks to use `text` instead of `llvm`
highlighting syntax.

Most of them already were like this, and the Sphinx runs on the docs
build bot seems to be substantially more picky and/or not have support
for a bunch of the syntax here. Hopefully this will let it progress past
this.

My previous attempt to fix the syntax made the `opt` tool happy, but no
idea what the Sphinx stuff is really looking for, and the fact that
other blocks already just use `text` led me to this solution.

llvm-svn: 338983
2018-08-06 03:35:36 +00:00
Chandler Carruth 24dd211f05 [docs] Correct the basic syntax structure of the DISubrange example.
Notably, just close two of the debug info metadata nodes early rather
than leaving them open with `...` which won't ever lex correctly. And
add the missing `:` on the count labels.

Slowly progressing through all of the warnings on the documentation
build bot. Sorry to do this one commit at a time, but despite my best
efforts I can't trigger these errors locally.

llvm-svn: 338982
2018-08-06 02:30:01 +00:00
Chandler Carruth 297620d337 [docs] Remove an example that isn't well formed LLVM IR and trips up the
Sphinx syntax highlighter.

This example also doesn't really make sense. There is no control flow or
clarification of what the `Safe:` block exists to do... If we want
examples here, we should make them much more clear in addition to making
them well formed IR sequences.

llvm-svn: 338981
2018-08-06 02:02:09 +00:00
Chandler Carruth ef705b7de1 [docs] Fix an LLVM-syntax code block to actually be valid LLVM synatx.
Hopefully fixes an issue on the docs build bot.

llvm-svn: 338980
2018-08-06 01:41:25 +00:00
Chandler Carruth 343a87ac8d [docs] Turn of `nasm` highlighting for a code block.
This appears to produce a warning on the docs build bot. It doesn't
reproduce for me, likely because I have a newer (or more full featured)
pygments install.

llvm-svn: 338978
2018-08-06 01:19:43 +00:00
Chandler Carruth 1230d22c59 [docs] Reinstate r337730 - Add support for Markdown documentation in
Sphinx.

We think the bot is updated now, so trying this again. I'm landing it
(with permission) as Michael is at a con at the moment.

Actual patch largely by Michael Spencer.

Differential Revision: https://reviews.llvm.org/D44910

llvm-svn: 338977
2018-08-06 00:38:31 +00:00
Lang Hames efd72d6a85 [ORC] Remove an incorrect use of 'cantFail'.
This code was moved out from BasicObjectLayerMaterializationUnit, which required
the supplied object to be well formed. The getObjectSymbolFlags function does
not require a well-formed object, so we have to propagate the error here.

llvm-svn: 338975
2018-08-05 23:55:35 +00:00
Lang Hames c0514f00bf [ORC] Change JITSymbolFlags debug output, add a function for getting a symbol
flags map from a buffer representing an object file.

llvm-svn: 338974
2018-08-05 22:35:37 +00:00
David Bolvansky b7fcd10700 [NFC] Fixed inliner tests - 2
llvm-svn: 338973
2018-08-05 16:53:36 +00:00
David Bolvansky 2f1f3b10ad [NFC] Fixed inliner tests
llvm-svn: 338972
2018-08-05 16:30:46 +00:00
David Bolvansky c0aa4b75a4 Enrich inline messages
Summary:
This patch improves Inliner to provide causes/reasons for negative inline decisions.
1. It adds one new message field to InlineCost to report causes for Always and Never instances. All Never and Always instantiations must provide a simple message.
2. Several functions that used to return the inlining results as boolean are changed to return InlineResult which carries the cause for negative decision.
3. Changed remark priniting and debug output messages to provide the additional messages and related inline cost.
4. Adjusted tests for changed printing.

Patch by: yrouban (Yevgeny Rouban)


Reviewers: craig.topper, sammccall, sgraenitz, NutshellySima, shchenz, chandlerc, apilipenko, javed.absar, tejohnson, dblaikie, sanjoy, eraman, xbolva00

Reviewed By: tejohnson, xbolva00

Subscribers: xbolva00, llvm-commits, arsenm, mehdi_amini, eraman, haicheng, steven_wu, dexonsmith

Differential Revision: https://reviews.llvm.org/D49412

llvm-svn: 338969
2018-08-05 14:53:08 +00:00
Eric Christopher 9855a5a0a1 Revert "Add a warning if someone attempts to add extra section flags to sections"
There are a bunch of edge cases and inconsistencies in how we're emitting sections
cause this warning to fire and it needs more work.

This reverts commit r335558.

llvm-svn: 338968
2018-08-05 14:23:37 +00:00
Roman Lebedev 365fa96055 [NFC][InstCombine] Add tests for sinking 'not' into 'xor' (PR38446)
https://rise4fun.com/Alive/IT3

Comes up in the [most ugliest]  signed int -> signed char  case of
-fsanitize=implicit-conversion (https://reviews.llvm.org/D50250)

Not sure if we want to do it always, or only when it is free to invert.

llvm-svn: 338967
2018-08-05 10:15:04 +00:00
Roman Lebedev 656a478e98 [NFC][InstCombine] Regenerate set.ll test
llvm-svn: 338965
2018-08-05 08:53:40 +00:00
Craig Topper fb33181038 [X86] Remove stale comments from a test. NFC
The 16-bit case was recently fixed so this comment no longer applies.

llvm-svn: 338964
2018-08-05 06:25:01 +00:00