Commit Graph

58686 Commits

Author SHA1 Message Date
Simon Pilgrim 44feb4a87b [CostModel][X86] Add XOP icmp cost tests (PR40376)
llvm-svn: 351741
2019-01-21 11:33:52 +00:00
Dmitry Venikov 119cf66fa5 [llvm-symbolizer] Add -no-demangle as alias for -demangle=false
Summary: Provides -no-demangle as alias for -demangle=false. Motivation: https://bugs.llvm.org/show_bug.cgi?id=40075

Reviewers: jhenderson, ruiu

Reviewed By: jhenderson

Subscribers: erik.pilkington, rupprecht, llvm-commits

Differential Revision: https://reviews.llvm.org/D56773

llvm-svn: 351735
2019-01-21 10:00:57 +00:00
Craig Topper f608dc1f57 [X86] Remove and autoupgrade vpmovqd/vpmovwb intrinsics using trunc+select.
llvm-svn: 351729
2019-01-21 08:16:59 +00:00
Kito Cheng 5e8798f987 [RISCV] Add R_RISCV_RELAX relocation to all possible relax candidates.
Summary:
Add R_RISCV_RELAX relocation to all possible relax candidates and
update corresponding testcase.

Reviewers: asb, apazos

Differential Revision: https://reviews.llvm.org/D46677

llvm-svn: 351723
2019-01-21 05:27:09 +00:00
Dylan McKay 5c23410fdf [AVR] Insert unconditional branch when inserting MBBs between blocks with fallthrough
This updates the AVR Select8/Select16 expansion code so that, when
inserting the two basic blocks for true and false conditions, any
existing fallthrough on the previous block is preserved.

Prior to this patch, if the block before the Select pseudo fell through
to the subsequent block, two new basic blocks would be inserted at the
prior fallthrough point, changing the fallthrough destination.

The predecessor or successor lists were not updated, causing the
BranchFolding pass at -O1 and above the rearrange basic blocks, causing
an infinite loop. Not to mention the unconditional fallthrough to the
true block is incorrect in of itself.

This patch modifies the Select8/16 expansion so that, if inserting true
and false basic blocks at a fallthrough point, the implicit branch is
preserved by means of an explicit, unconditional branch to the previous
fallthrough destination.

Thanks to Carl Peto for reporting this bug.

This fixes avr-rust bug https://github.com/avr-rust/rust/issues/123.

llvm-svn: 351721
2019-01-21 04:32:02 +00:00
Dylan McKay ce0ab06353 Revert "[AVR] Insert unconditional branch when inserting MBBs between blocks with fallthrough"
This reverts commit r351718.

Carl pointed out that the unit test could be improved.

This patch will be recommitted once the test is made more resilient.

llvm-svn: 351719
2019-01-21 02:46:13 +00:00
Dylan McKay 33acba43f0 [AVR] Insert unconditional branch when inserting MBBs between blocks with fallthrough
This updates the AVR Select8/Select16 expansion code so that, when
inserting the two basic blocks for true and false conditions, any
existing fallthrough on the previous block is preserved.

Prior to this patch, if the block before the Select pseudo fell through
to the subsequent block, two new basic blocks would be inserted at the
prior fallthrough point, changing the fallthrough destination.

The predecessor or successor lists were not updated, causing the
BranchFolding pass at -O1 and above the rearrange basic blocks, causing
an infinite loop. Not to mention the unconditional fallthrough to the
true block is incorrect in of itself.

This patch modifies the Select8/16 expansion so that, if inserting true
and false basic blocks at a fallthrough point, the implicit branch is
preserved by means of an explicit, unconditional branch to the previous
fallthrough destination.

Thanks to Carl Peto for reporting this bug.

This fixes avr-rust bug https://github.com/avr-rust/rust/issues/123.

llvm-svn: 351718
2019-01-21 02:44:09 +00:00
Matt Arsenault 7ac79ed8f0 AMDGPU: Legalize more bitcasts
llvm-svn: 351700
2019-01-20 19:45:18 +00:00
Matt Arsenault 46ffe68d77 AMDGPU/GlobalISel: Really legalize exts from i1
There is a combine that was hiding these tests
not actually testing what they should be, although
they were producing the expected end result.

llvm-svn: 351698
2019-01-20 19:28:20 +00:00
Simon Pilgrim e1143c1322 [X86] Auto upgrade VPCOM/VPCOMU intrinsics to generic integer comparisons
This causes a couple of changes in the upgrade tests as signed/unsigned eq/ne are equivalent and we constant fold true/false codes, these changes are the same as what we already do for avx512 cmp/ucmp.

Noticed while cleaning up vector integer comparison costs for PR40376.

llvm-svn: 351697
2019-01-20 19:27:40 +00:00
Matt Arsenault 745fd9f547 GlobalISel: Implement widenScalar for basic FP ops
llvm-svn: 351696
2019-01-20 19:10:31 +00:00
Matt Arsenault cfd9e7f594 AMDGPU/GlobalISel: Legalize f32->f16 fptrunc
llvm-svn: 351695
2019-01-20 19:10:26 +00:00
Matt Arsenault ff6a9a275b AMDGPU/GlobalISel: Fix some crashs in g_unmerge_values/g_merge_values
This was crashing in the predicate function assuming the value
is a vector.

Copy more of what AArch64 uses. This probably needs more refinement
later, but I don't exactly understand what it means in some cases,
particularly since any legalization for these seems to be missing.

llvm-svn: 351693
2019-01-20 18:40:36 +00:00
Matt Arsenault 2a2086b830 AMDGPU/GlobalISel: Regbank select for fpext
llvm-svn: 351692
2019-01-20 18:35:41 +00:00
Matt Arsenault 24563ef628 AMDGPU/GlobalISel: Cleanup legality for extensions
llvm-svn: 351691
2019-01-20 18:34:24 +00:00
Simon Pilgrim b590e4f7e5 [X86] Auto upgrade old style VPCOM/VPCOMU intrinsics to generic integer comparisons
We were upgrading these to the new style VPCOM/VPCOMU intrinsics (which includes the condition code immediate), but we'll be getting rid of those shortly, so convert these to generics first.

This causes a couple of changes in the upgrade tests as signed/unsigned eq/ne are equivalent and we constant fold true/false codes, these changes are the same as what we already do for avx512 cmp/ucmp.

Noticed while cleaning up vector integer comparison costs for PR40376.

llvm-svn: 351690
2019-01-20 17:36:22 +00:00
Simon Pilgrim 4fd2459c4d [X86] Replace VPCOM/VPCOMU with generic integer comparisons (llvm)
These intrinsics can always be replaced with generic integer comparisons without any regression in codegen, even for -O0/-fast-isel cases.

Noticed while cleaning up vector integer comparison costs for PR40376.

A future commit will remove/autoupgrade the existing VPCOM/VPCOMU llvm intrinsics.

llvm-svn: 351688
2019-01-20 16:40:44 +00:00
Simon Pilgrim c934d3a01b [CostModel][X86] Add explicit vector select costs
Prior to SSE41 (and sometimes on AVX1), vector select has to be performed as a ((X & C)|(Y & ~C)) bit select.

Exposes a couple of issues with the min/max reduction costs (which only go down to SSE42 for some reason).

The increase pre-SSE41 selection costs also prevent a couple of tests from firing any longer, so I've either tweaked the target or added AVX tests as well to the existing SSE2 tests.

llvm-svn: 351685
2019-01-20 13:55:01 +00:00
Simon Pilgrim 1231904c48 [CostModel][X86] Add explicit fcmp costs for pre-SSE42 targets
Typical throughputs: cmpss/cmpps = 1cy and cmpsd/cmppd = 2cy before the Core2 era

llvm-svn: 351684
2019-01-20 13:21:43 +00:00
Simon Pilgrim 60e5a3accb [CostModel][X86] Split icmp/fcmp costs tests and test all comparison codes
llvm-svn: 351682
2019-01-20 12:10:42 +00:00
Simon Pilgrim 5d7182ecb6 [CostModel][X86] Add masked load/store/gather/scatter tests for SSE2/SSE42/AVX1 targets
llvm-svn: 351681
2019-01-20 11:23:01 +00:00
Simon Pilgrim a8b009fd14 [CostModel][X86] Add non-constant vselect cost tests
Also add AVX512 costs at the same time

llvm-svn: 351680
2019-01-20 11:19:35 +00:00
Dylan McKay a6241a5dc0 [AVR] Remove unneeded XFAILs from the Generic CodeGen tests
These have been in place for quite a while now.

Several bugs have since been fixed, and these tests now pass.

llvm-svn: 351679
2019-01-20 11:16:58 +00:00
Dylan McKay 6afef286d9 [AVR] Fix codegen bug in 16-bit loads
Prior to this patch, the AVR::LDWRdPtr instruction was always lowered to
instructions of this pattern:

    ld  $GPR8, [PTR:XYZ]+
    ld  $GPR8, [PTR]+1

This has a problem; the [PTR] is incremented in-place once, but never
decremented.

Future uses of the same pointer will use the now clobbered value,
leading to the pointer being incorrect by an offset of one.

This patch modifies the expansion code of the LDWRdPtr pseudo
instruction so that the pointer variable is not silently clobbered in
future uses in the same live range.

Bug first reported by Keshav Kini.

Patch by Kaushik Phatak.

llvm-svn: 351673
2019-01-20 03:41:08 +00:00
Dylan McKay 52846ab09a Revert "[AVR] Fix codegen bug in 16-bit loads"
This reverts commit r351544.

In that commit, I had mistakenly misattributed the issue submitter as
the patch author, Kaushik Phatak.

The patch will be recommitted immediately with the correct attribution.

llvm-svn: 351672
2019-01-20 03:41:00 +00:00
Martin Storsjo e8305175b0 [llvm-objcopy] [COFF] Implement --only-section
Differential Revision: https://reviews.llvm.org/D56873

llvm-svn: 351663
2019-01-19 19:42:54 +00:00
Martin Storsjo 1868d88b2e [llvm-objcopy] [COFF] Implement --only-keep-debug
Differential Revision: https://reviews.llvm.org/D56840

llvm-svn: 351662
2019-01-19 19:42:48 +00:00
Martin Storsjo 78a0b418b4 [llvm-objcopy] [COFF] Implement --strip-debug
Also remove sections similarly for --strip-all, --discard-all,
--strip-unneeded.

Differential Revision: https://reviews.llvm.org/D56839

llvm-svn: 351661
2019-01-19 19:42:41 +00:00
Martin Storsjo f9e1434ef4 [llvm-objcopy] [COFF] Add support for removing sections
Differential Revision: https://reviews.llvm.org/D56683

llvm-svn: 351660
2019-01-19 19:42:35 +00:00
Martin Storsjo e9f62f62ce [llvm-objcopy] [COFF] Add a testcase for patching the debug directory. NFC.
The debug directory contains the rwa file address of itself,
which is updated on write. Add a testcase for this existing
functionality.

Differential Revision: https://reviews.llvm.org/D56876

llvm-svn: 351659
2019-01-19 19:42:27 +00:00
Martin Storsjo f11509ab11 [llvm-objcopy] [COFF] Rename a test from .yaml to .test. NFC.
Tests named .yaml aren't executed by default in this directory
(while they are within e.g. LLD).

llvm-svn: 351657
2019-01-19 19:42:19 +00:00
Nikita Popov 6515db205a [InstCombine] Simplify cttz/ctlz + icmp ugt/ult
Followup to D55745, this time handling comparisons with ugt and ult
predicates (which are the canonical forms for non-equality predicates).

For ctlz we can convert into a simple icmp, for cttz we can convert
into a mask check.

Differential Revision: https://reviews.llvm.org/D56355

llvm-svn: 351645
2019-01-19 09:56:01 +00:00
Johannes Doerfert 36872b5db9 Enable IPConstantPropagation to work with abstract call sites
This modification of the currently unused inter-procedural constant
propagation pass (IPConstantPropagation) shows how abstract call sites
enable optimization of callback calls alongside direct and indirect
calls. Through minimal changes, mostly dealing with the partial mapping
of callbacks, inter-procedural constant propagation was enabled for
callbacks, e.g., OpenMP runtime calls or pthreads_create.

Differential Revision: https://reviews.llvm.org/D56447

llvm-svn: 351628
2019-01-19 05:19:12 +00:00
Johannes Doerfert 18251842c6 AbstractCallSite -- A unified interface for (in)direct and callback calls
An abstract call site is a wrapper that allows to treat direct,
  indirect, and callback calls the same. If an abstract call site
  represents a direct or indirect call site it behaves like a stripped
  down version of a normal call site object. The abstract call site can
  also represent a callback call, thus the fact that the initially
  called function (=broker) may invoke a third one (=callback callee).
  In this case, the abstract call side hides the middle man, hence the
  broker function. The result is a representation of the callback call,
  inside the broker, but in the context of the original instruction that
  invoked the broker.

  Again, there are up to three functions involved when we talk about
  callback call sites. The caller (1), which invokes the broker
  function. The broker function (2), that may or may not invoke the
  callback callee. And finally the callback callee (3), which is the
  target of the callback call.

  The abstract call site will handle the mapping from parameters to
  arguments depending on the semantic of the broker function. However,
  it is important to note that the mapping is often partial. Thus, some
  arguments of the call/invoke instruction are mapped to parameters of
  the callee while others are not. At the same time, arguments of the
  callback callee might be unknown, thus "null" if queried.

  This patch introduces also !callback metadata which describe how a
  callback broker maps from parameters to arguments. This metadata is
  directly created by clang for known broker functions, provided through
  source code attributes by the user, or later deduced by analyses.

For motivation and additional information please see the corresponding
talk (slides/video)
  https://llvm.org/devmtg/2018-10/talk-abstracts.html#talk20
as well as the LCPC paper
  http://compilers.cs.uni-saarland.de/people/doerfert/par_opt_lcpc18.pdf

Differential Revision: https://reviews.llvm.org/D54498

llvm-svn: 351627
2019-01-19 05:19:06 +00:00
Roman Tereshin a0383d6c1f Reapply "[CGP] Check for existing inttotpr before creating new one"
Original commit: r351582

llvm-svn: 351626
2019-01-19 03:37:25 +00:00
Vedant Kumar b537b946b8 [MergeFunc] Allow merging identical vararg functions using aliases
Thanks to Nikita Popov for pointing out this missed case.

This is a follow-up to r351411, which disabled function merging for
vararg functions outright due to a miscompile (see llvm.org/PR40345).

Differential Revision: https://reviews.llvm.org/D56865

llvm-svn: 351624
2019-01-19 02:46:22 +00:00
Vedant Kumar b755a2df51 [HotColdSplit] Mark inherently cold functions as such
If an inherently cold function is found, mark it as cold. For now this
means applying the `cold` and `minsize` attributes.

As a drive-by, revisit and clean up the criteria for considering a
function for splitting. Add tests.

llvm-svn: 351623
2019-01-19 02:38:47 +00:00
Vedant Kumar 17d9f14bff [CodeExtractor] Emit lifetime markers around reloads of outputs
CodeExtractor permits extracting a region of blocks from a function even
when values defined within the region are used outside of it.

This is typically done by creating an alloca in the original function
and reloading the alloca after a call to the extracted function.

Wrap the reload in lifetime start/end markers to promote stack coloring.

Suggested by Sergei Kachkov!

Differential Revision: https://reviews.llvm.org/D56045

llvm-svn: 351621
2019-01-19 02:37:59 +00:00
Roman Tereshin 022bf3e8e7 Revert "Reapply "[CGP] Check for existing inttotpr before creating new one""
This reverts commit r351618.

Compiler RT + ASAN tests are failing for PowerPC. Not sure
how would I reproduce these on macOS, so reverting (again)
until I do.

llvm-svn: 351619
2019-01-19 01:53:26 +00:00
Roman Tereshin dd6f9f68bb Reapply "[CGP] Check for existing inttotpr before creating new one"
Original commit: r351582

llvm-svn: 351618
2019-01-19 01:41:03 +00:00
Amara Emerson d5015edb37 Revert r351584: "GlobalISel: Verify g_zextload and g_sextload"
This new assertion triggered on the AArch64 GlobalISel bots. Reverting while it's being investigated.

llvm-svn: 351617
2019-01-19 00:36:11 +00:00
Nico Weber 63fd07ce07 Use llvm_canonicalize_cmake_booleans for LLVM_LIBXML2_ENABLED [llvm]
r291284 added a nice mechanism to consistently pass CMake on/off toggles to
lit. This change uses it for LLVM_LIBXML2_ENABLED too (which was added around
the same time and doesn't use the new system yet).

Also alphabetically sort the list passed to llvm_canonicalize_cmake_booleans()
in llvm/test/CMakeLists.txt.

No intended behavior change.

Differential Revision: https://reviews.llvm.org/D56912

llvm-svn: 351615
2019-01-19 00:10:54 +00:00
Matt Arsenault 96e4701401 AMDGPU/GlobalISel: Legalize more types for select
llvm-svn: 351599
2019-01-18 21:42:55 +00:00
Roman Tereshin 86ac532687 Revert "[CGP] Check for existing inttotpr before creating new one"
This reverts commit r351582.

Bots are failing. Reverting this to fix and re-commit later.

llvm-svn: 351598
2019-01-18 21:38:44 +00:00
Matt Arsenault 4599159ac3 AMDGPU/GlobalISel: Legalize illegal g_constant
llvm-svn: 351596
2019-01-18 21:33:50 +00:00
Matt Arsenault bd3a5b29cb GlobalISel: Verify G_BITCAST
llvm-svn: 351594
2019-01-18 21:04:59 +00:00
Armando Montanez 56d18121e2 [elfabi] Add support for reading DT_NEEDED from binaries
This patch gives elfabi the ability to read DT_NEEDED entries from ELF binaries
to populate NeededLibs in TextAPI's ELFStub.

Differential Revision: https://reviews.llvm.org/D55852

llvm-svn: 351592
2019-01-18 20:56:03 +00:00
Matt Arsenault 215c4f68f6 GlobalISel: Verify G_ICMP/G_FCMP vector types
llvm-svn: 351591
2019-01-18 20:49:17 +00:00
Sanjay Patel 4453e4292d [x86] add more movmsk tests; NFC
The existing tests already show a sub-optimal transform,
but this should make it clear that we can't just match
an 'and' op when creating movmsk instructions.

llvm-svn: 351590
2019-01-18 20:42:12 +00:00
Teresa Johnson 723636ee8c Make ThinLTO test run single threaded to try to avoid flakiness
To see if this helps flaky bot failures in PR40351.

llvm-svn: 351589
2019-01-18 20:41:49 +00:00