OSDN Git Service
Chandler Carruth [Mon, 6 Aug 2018 09:46:59 +0000 (09:46 +0000)]
[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.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338997
91177308-0d34-0410-b5e6-
96231b3b80d8
Max Kazantsev [Mon, 6 Aug 2018 08:10:28 +0000 (08:10 +0000)]
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
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338994
91177308-0d34-0410-b5e6-
96231b3b80d8
Max Kazantsev [Mon, 6 Aug 2018 06:35:21 +0000 (06:35 +0000)]
Try to fix buildbot
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338991
91177308-0d34-0410-b5e6-
96231b3b80d8
Max Kazantsev [Mon, 6 Aug 2018 06:11:36 +0000 (06:11 +0000)]
[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
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338990
91177308-0d34-0410-b5e6-
96231b3b80d8
Max Kazantsev [Mon, 6 Aug 2018 05:49:19 +0000 (05:49 +0000)]
[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
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338988
91177308-0d34-0410-b5e6-
96231b3b80d8
Xin Tong [Mon, 6 Aug 2018 05:03:21 +0000 (05:03 +0000)]
[NFC] Fix typo
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338987
91177308-0d34-0410-b5e6-
96231b3b80d8
David Bolvansky [Mon, 6 Aug 2018 04:45:46 +0000 (04:45 +0000)]
[NFC] Fixed unused function warning
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338986
91177308-0d34-0410-b5e6-
96231b3b80d8
Hsiangkai Wang [Mon, 6 Aug 2018 03:59:47 +0000 (03:59 +0000)]
[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
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338984
91177308-0d34-0410-b5e6-
96231b3b80d8
Chandler Carruth [Mon, 6 Aug 2018 03:35:36 +0000 (03:35 +0000)]
[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.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338983
91177308-0d34-0410-b5e6-
96231b3b80d8
Chandler Carruth [Mon, 6 Aug 2018 02:30:01 +0000 (02:30 +0000)]
[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.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338982
91177308-0d34-0410-b5e6-
96231b3b80d8
Chandler Carruth [Mon, 6 Aug 2018 02:02:09 +0000 (02:02 +0000)]
[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.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338981
91177308-0d34-0410-b5e6-
96231b3b80d8
Chandler Carruth [Mon, 6 Aug 2018 01:41:25 +0000 (01:41 +0000)]
[docs] Fix an LLVM-syntax code block to actually be valid LLVM synatx.
Hopefully fixes an issue on the docs build bot.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338980
91177308-0d34-0410-b5e6-
96231b3b80d8
Chandler Carruth [Mon, 6 Aug 2018 01:19:43 +0000 (01:19 +0000)]
[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.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338978
91177308-0d34-0410-b5e6-
96231b3b80d8
Chandler Carruth [Mon, 6 Aug 2018 00:38:31 +0000 (00:38 +0000)]
[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
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338977
91177308-0d34-0410-b5e6-
96231b3b80d8
Lang Hames [Sun, 5 Aug 2018 23:55:35 +0000 (23:55 +0000)]
[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.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338975
91177308-0d34-0410-b5e6-
96231b3b80d8
Lang Hames [Sun, 5 Aug 2018 22:35:37 +0000 (22:35 +0000)]
[ORC] Change JITSymbolFlags debug output, add a function for getting a symbol
flags map from a buffer representing an object file.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338974
91177308-0d34-0410-b5e6-
96231b3b80d8
David Bolvansky [Sun, 5 Aug 2018 16:53:36 +0000 (16:53 +0000)]
[NFC] Fixed inliner tests - 2
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338973
91177308-0d34-0410-b5e6-
96231b3b80d8
David Bolvansky [Sun, 5 Aug 2018 16:30:46 +0000 (16:30 +0000)]
[NFC] Fixed inliner tests
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338972
91177308-0d34-0410-b5e6-
96231b3b80d8
David Bolvansky [Sun, 5 Aug 2018 14:53:08 +0000 (14:53 +0000)]
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
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338969
91177308-0d34-0410-b5e6-
96231b3b80d8
Eric Christopher [Sun, 5 Aug 2018 14:23:37 +0000 (14:23 +0000)]
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.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338968
91177308-0d34-0410-b5e6-
96231b3b80d8
Roman Lebedev [Sun, 5 Aug 2018 10:15:04 +0000 (10:15 +0000)]
[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.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338967
91177308-0d34-0410-b5e6-
96231b3b80d8
Roman Lebedev [Sun, 5 Aug 2018 08:53:40 +0000 (08:53 +0000)]
[NFC][InstCombine] Regenerate set.ll test
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338965
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Sun, 5 Aug 2018 06:25:01 +0000 (06:25 +0000)]
[X86] Remove stale comments from a test. NFC
The 16-bit case was recently fixed so this comment no longer applies.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338964
91177308-0d34-0410-b5e6-
96231b3b80d8
David Bolvansky [Sun, 5 Aug 2018 05:46:56 +0000 (05:46 +0000)]
[InstCombine] [NFC] Tests for strcmp to memcmp transformation
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338963
91177308-0d34-0410-b5e6-
96231b3b80d8
Chandler Carruth [Sat, 4 Aug 2018 08:17:26 +0000 (08:17 +0000)]
[ADT] Add an early-increment iterator-like type and range adaptor.
This allows us to model the common LLVM idiom of incrementing
immediately after dereferencing so that we can remove or update the
entity w/o losing our ability to reach the "next".
However, these are not real or proper iterators. They are just enough to
allow range based for loops and very simple range algorithms to work,
but should not be considered full general.
Differential Revision: https://reviews.llvm.org/D49956
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338955
91177308-0d34-0410-b5e6-
96231b3b80d8
Chijun Sima [Sat, 4 Aug 2018 08:13:47 +0000 (08:13 +0000)]
[TailCallElim] Preserve DT and PDT
Summary:
Previously, in the NewPM pipeline, TailCallElim recalculates the DomTree when it modifies any instruction in the Function.
For example,
```
CallInst *CI = dyn_cast<CallInst>(&I);
...
CI->setTailCall();
Modified = true;
...
if (!Modified || ...)
return PreservedAnalyses::all();
```
After applying this patch, the DomTree only recalculates if needed (plus an extra insertEdge() + an extra deleteEdge() call).
When optimizing SQLite with `-passes="default<O3>"` pipeline of the newPM, the number of DomTree recalculation decreases by 6.2%, the number of nodes visited by DFS decreases by 2.9%. The time used by DomTree will decrease approximately 1%~2.5% after applying the patch.
Statistics:
```
Before the patch:
23010 dom-tree-stats - Number of DomTree recalculations
489264 dom-tree-stats - Number of nodes visited by DFS -- DomTree
After the patch:
21581 dom-tree-stats - Number of DomTree recalculations
475088 dom-tree-stats - Number of nodes visited by DFS -- DomTree
```
Reviewers: kuhar, dmgreen, brzycki, grosser, davide
Reviewed By: kuhar, brzycki
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D49982
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338954
91177308-0d34-0410-b5e6-
96231b3b80d8
Fangrui Song [Sat, 4 Aug 2018 05:19:00 +0000 (05:19 +0000)]
[llvm-objdump] Remove continue after report_error which is unreachable
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338951
91177308-0d34-0410-b5e6-
96231b3b80d8
Chijun Sima [Sat, 4 Aug 2018 02:50:12 +0000 (02:50 +0000)]
[ADCE] Remove the need of DomTree
Summary: ADCE doesn't need to query domtree.
Reviewers: kuhar, brzycki, dmgreen, davide, grosser
Reviewed By: kuhar
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D49988
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338950
91177308-0d34-0410-b5e6-
96231b3b80d8
Galina Kistanova [Sat, 4 Aug 2018 01:59:12 +0000 (01:59 +0000)]
Reverted r338825 and all the following tries to fix issues introduced by that commit (r338826, r338827, r338829, r338880).
This commit has broken build bots and has been left unattended for too long.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338948
91177308-0d34-0410-b5e6-
96231b3b80d8
Aditya Nandakumar [Sat, 4 Aug 2018 01:22:12 +0000 (01:22 +0000)]
[GISel]: Add Opcodes for CTLZ/CTTZ/CTPOP
https://reviews.llvm.org/D48600
Added IRTranslator support to translate these known intrinsics into GISel opcodes.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338944
91177308-0d34-0410-b5e6-
96231b3b80d8
Rui Ueyama [Sat, 4 Aug 2018 00:23:37 +0000 (00:23 +0000)]
Fix buildbot breakage.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338940
91177308-0d34-0410-b5e6-
96231b3b80d8
Rui Ueyama [Sat, 4 Aug 2018 00:13:13 +0000 (00:13 +0000)]
Use the same constants as zlib to represent compression level.
This change allows users pass compression level that was not listed
in the enum. Also, I think using different values than zlib's
compression levels was just confusing.
Differential Revision: https://reviews.llvm.org/D50196
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338939
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Fri, 3 Aug 2018 22:08:30 +0000 (22:08 +0000)]
[X86] Add isel patterns for atomic_load+sub+atomic_sub.
Despite the comment removed in this patch, this is beneficial when the RHS of the sub is a register.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338930
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Fri, 3 Aug 2018 22:08:28 +0000 (22:08 +0000)]
[X86] Add test cases to show missed opportunity to use RMW for atomic_load+sub+atomic_store.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338929
91177308-0d34-0410-b5e6-
96231b3b80d8
Reid Kleckner [Fri, 3 Aug 2018 21:58:25 +0000 (21:58 +0000)]
[X86] Re-generate abi-isel.ll checks with update_llc_test_checks.py
These tests were clearly auto-generated when they were converted to
FileCheck back in r80019 (2009), but we didn't have a fancy script to
keep them up to date then. I've reviewed the diff, and we should be
generating the exact same code sequences we used to.
After this, I plan to commit a change that changes our output slightly,
but in a way that is still correct. It will generate a large diff, and I
want it to be clearly correct, so I am regenerating these checks in
preparation for that.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338928
91177308-0d34-0410-b5e6-
96231b3b80d8
Reid Kleckner [Fri, 3 Aug 2018 21:58:12 +0000 (21:58 +0000)]
[X86] Make abi-isel.ll like update_llc_test_checks.py output
- Remove -asm-verbose=0 from every llc command. The tests still pass.
- Reorder the RUN lines to match CHECKs.
- Use -LABEL like update_llc_test_checks.py does.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338927
91177308-0d34-0410-b5e6-
96231b3b80d8
Reid Kleckner [Fri, 3 Aug 2018 21:57:59 +0000 (21:57 +0000)]
[X86] Layout tests exactly as update_llc_test_checks.py would
Put the LLVM IR at the bottom of the function instead of the top. In my
next patch, I will run update_llc_test_checks.py on this file, and I
want to only highlight the diffs in the CHECK lines. Hopefully by doing
this change first, the patch will be more understandable.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338926
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Fri, 3 Aug 2018 21:40:44 +0000 (21:40 +0000)]
[X86] Remove RELEASE_ and ACQUIRE_ pseudo instructions. Use isel patterns and the normal instructions instead
At one point in time acquire implied mayLoad and mayStore as did release. Thus we needed separate pseudos that also carried that property. This appears to no longer be the case. I believe it was changed in 2012 with a comment saying that atomic memory accesses are marked volatile which preserves the ordering.
So from what I can tell we shouldn't need additional pseudos since they aren't carry any flags that are different from the normal instructions. The only thing I can think of is that we may consider them for load folding candidates in the peephole pass now where we didn't before. If that's important hopefully there's something in the memory operand we can check to prevent the folding without relying on pseudo instructions.
Differential Revision: https://reviews.llvm.org/D50212
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338925
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Fri, 3 Aug 2018 20:58:14 +0000 (20:58 +0000)]
[X86] Autogenerate complete checks. NFC
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338921
91177308-0d34-0410-b5e6-
96231b3b80d8
Anastasis Grammenos [Fri, 3 Aug 2018 20:27:13 +0000 (20:27 +0000)]
[TRE][DebugInfo] Preserve Debug Location in new branch instruction
There are two branch instructions created
so the new test covers them both.
Differential Revision: https://reviews.llvm.org/D50263
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338917
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Fri, 3 Aug 2018 20:14:18 +0000 (20:14 +0000)]
[SelectionDAG] Teach LegalizeVectorTypes to widen the mask input to a masked store.
The mask operand is visited before the data operand so we need to be able to widen it.
Fixes PR38436.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338915
91177308-0d34-0410-b5e6-
96231b3b80d8
Fangrui Song [Fri, 3 Aug 2018 19:37:49 +0000 (19:37 +0000)]
[Support] Don't initialize compressed buffer allocated by zlib::compress
resize() (zeroing) makes every allocated page resident. The actual size of the compressed buffer is usually much
smaller. Making every page resident is wasteful.
When linking a test binary with ~1.9GiB uncompressed debug info with LLD, this optimization decreases max RSS by ~1.5GiB.
Differential Revision: https://reviews.llvm.org/50223
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338913
91177308-0d34-0410-b5e6-
96231b3b80d8
Matt Arsenault [Fri, 3 Aug 2018 18:27:52 +0000 (18:27 +0000)]
DAG: Enhance isKnownNeverNaN
Add a parameter for testing specifically for
sNaNs - at least one instruction pattern on AMDGPU
needs to check specifically for this.
Also handle more cases, and add a target hook
for custom nodes, similar to the hooks for known
bits.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338910
91177308-0d34-0410-b5e6-
96231b3b80d8
Artem Belevich [Fri, 3 Aug 2018 18:05:24 +0000 (18:05 +0000)]
[NVPTX] Handle __nvvm_reflect("__CUDA_ARCH").
Summary:
libdevice in recent CUDA versions relies on __nvvm_reflect() to select
GPU-specific bitcode. This patch addresses the requirement.
Reviewers: jlebar
Subscribers: jholewinski, sanjoy, hiraditya, bixia, llvm-commits
Differential Revision: https://reviews.llvm.org/D50207
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338908
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Fri, 3 Aug 2018 18:00:29 +0000 (18:00 +0000)]
[X86] Add a DAG combine for the __builtin_parity idiom used by clang to enable better codegen
Clang uses "ctpop & 1" to implement __builtin_parity. If the popcnt instruction isn't supported this generates a large amount of code to calculate the population count. Instead we can bisect the data down to a single byte using xor and then check the parity flag.
Even when popcnt is supported, its still a good idea to split 64-bit data on 32-bit targets using an xor in front of a single popcnt. Otherwise we get two popcnts and an add before the and.
I've specifically targeted this at the sizes supported by clang builtins, but we could generalize this if we think that's useful.
Differential Revision: https://reviews.llvm.org/D50165
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338907
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Fri, 3 Aug 2018 18:00:23 +0000 (18:00 +0000)]
[X86] Add test cases for the current codegen of __builtin_parity.
Will be improved in a follow commit
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338906
91177308-0d34-0410-b5e6-
96231b3b80d8
Evandro Menezes [Fri, 3 Aug 2018 17:50:16 +0000 (17:50 +0000)]
[SLC] Refactor shrinking of functions (NFC)
Merge the helper functions for shrinking unary and binary functions into a
single one, while keeping all their functionality. Otherwise, NFC.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338905
91177308-0d34-0410-b5e6-
96231b3b80d8
Joel Galenson [Fri, 3 Aug 2018 17:12:23 +0000 (17:12 +0000)]
Fix crash in bounds checking.
In r337830 I added SCEV checks to enable us to insert fewer bounds checks. Unfortunately, this sometimes crashes when multiple bounds checks are added due to SCEV caching issues. This patch splits the bounds checking pass into two phases, one that computes all the conditions (using SCEV checks) and the other that adds the new instructions.
Differential Revision: https://reviews.llvm.org/D49946
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338902
91177308-0d34-0410-b5e6-
96231b3b80d8
Matt Davis [Fri, 3 Aug 2018 15:56:07 +0000 (15:56 +0000)]
[llvm-mca][docs] Move the code marker text into its own subsection. NFC.
Also fixed a few undecorated 'llvm-mca' references to be highlighted
with the 'program' emphasis.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338900
91177308-0d34-0410-b5e6-
96231b3b80d8
Simon Pilgrim [Fri, 3 Aug 2018 15:09:56 +0000 (15:09 +0000)]
[X86] Fix line endings.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338897
91177308-0d34-0410-b5e6-
96231b3b80d8
Graham Yiu [Fri, 3 Aug 2018 14:42:53 +0000 (14:42 +0000)]
[Partial Inlining] Fix small bug in detecting if we did something
- It's possible for 'Changed' to return as false even if we did
partial inline something. Fixed to accumulate return values
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338896
91177308-0d34-0410-b5e6-
96231b3b80d8
Nicholas Wilson [Fri, 3 Aug 2018 14:33:37 +0000 (14:33 +0000)]
[WebAssembly] Cleanup of the way globals and global flags are handled
Differential Revision: https://reviews.llvm.org/D44030
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338894
91177308-0d34-0410-b5e6-
96231b3b80d8
Andrea Di Biagio [Fri, 3 Aug 2018 12:55:28 +0000 (12:55 +0000)]
[llvm-mca] Speed up the computation of the wait/ready/issued sets in the Scheduler.
This patch is a follow-up to r338702.
We don't need to use a map to model the wait/ready/issued sets. It is much more
efficient to use a vector instead.
This patch gives us an average 7.5% speedup (on top of the ~12% speedup obtained
after r338702).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338883
91177308-0d34-0410-b5e6-
96231b3b80d8
Chijun Sima [Fri, 3 Aug 2018 12:45:29 +0000 (12:45 +0000)]
[Dominators] Make RemoveUnreachableBlocks return false if the BasicBlock is already awaiting deletion
Summary:
Previously, `removeUnreachableBlocks` still returns true (which indicates the CFG is changed) even when all the unreachable blocks found is awaiting deletion in the DDT class.
This makes code pattern like
```
// Code modified from lib/Transforms/Scalar/SimplifyCFGPass.cpp
bool EverChanged = removeUnreachableBlocks(F, nullptr, DDT);
...
do {
EverChanged = someMightHappenModifications();
EverChanged |= removeUnreachableBlocks(F, nullptr, DDT);
} while (EverChanged);
```
become a dead loop.
Fix this by detecting whether a BasicBlock is already awaiting deletion.
Reviewers: kuhar, brzycki, dmgreen, grosser, davide
Reviewed By: kuhar, brzycki
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D49738
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338882
91177308-0d34-0410-b5e6-
96231b3b80d8
Andrea Di Biagio [Fri, 3 Aug 2018 12:44:56 +0000 (12:44 +0000)]
[llvm-mca][docs] Improve the CommandLine documentation.
This patch replaces all the remaining occurrences of string "MCA" with
":program:`llvm-mca`". Somehow I missed those strings when I committed r338394.
This patch also improves section "Instruction Dispatch".
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338881
91177308-0d34-0410-b5e6-
96231b3b80d8
Nico Weber [Fri, 3 Aug 2018 12:21:54 +0000 (12:21 +0000)]
convert some tabs to spaces
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338880
91177308-0d34-0410-b5e6-
96231b3b80d8
Jonas Devlieghere [Fri, 3 Aug 2018 12:01:43 +0000 (12:01 +0000)]
[DebugInfo/Verifier] Don't emit error for missing module in index
We don't expect module names to be present in the index. This patch adds
DW_TAG_module to the blacklist.
Differential revision: https://reviews.llvm.org/D50237
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338878
91177308-0d34-0410-b5e6-
96231b3b80d8
Jonas Paulsson [Fri, 3 Aug 2018 10:43:05 +0000 (10:43 +0000)]
[SystemZ] Improve handling of instructions which expand to several groups
Some instructions expand to more than one decoder group.
This has been hitherto ignored, but is handled with this patch.
Review: Ulrich Weigand
https://reviews.llvm.org/D50187
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338849
91177308-0d34-0410-b5e6-
96231b3b80d8
Max Kazantsev [Fri, 3 Aug 2018 10:41:51 +0000 (10:41 +0000)]
[NFC] Add missing comment
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338848
91177308-0d34-0410-b5e6-
96231b3b80d8
Max Kazantsev [Fri, 3 Aug 2018 10:16:40 +0000 (10:16 +0000)]
[NFC] Move some methods into static functions
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338843
91177308-0d34-0410-b5e6-
96231b3b80d8
Jeremy Morse [Fri, 3 Aug 2018 10:13:35 +0000 (10:13 +0000)]
[Windows FS] Allow moving files in TempFile::keep
In r338216 / D49860 TempFile::keep was extended to allow keeping across
filesystems. The aim on Windows was to have this happen in rename_internal
using the existing system API. However, to fix an issue and preserve the
idea of "renaming" not being a move, put Windows keep-across-filesystem in
TempFile::keep.
Differential Revision: https://reviews.llvm.org/D50048
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338841
91177308-0d34-0410-b5e6-
96231b3b80d8
Simon Pilgrim [Fri, 3 Aug 2018 10:00:54 +0000 (10:00 +0000)]
[TargetLowering] Generalise BuildSDIV function
First step towards a BuildSDIV equivalent to D49248 for non-uniform vector support - this just pushes the splat detection down into TargetLowering::BuildSDIV where its still used.
Differential Revision: https://reviews.llvm.org/D50185
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338838
91177308-0d34-0410-b5e6-
96231b3b80d8
Guillaume Chatelet [Fri, 3 Aug 2018 09:29:38 +0000 (09:29 +0000)]
[llvm-exegesis] Renaming classes and functions.
Summary: Functional No Op.
Reviewers: gchatelet
Subscribers: tschuett, courbet, llvm-commits
Differential Revision: https://reviews.llvm.org/D50231
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338836
91177308-0d34-0410-b5e6-
96231b3b80d8
Sjoerd Meijer [Fri, 3 Aug 2018 09:24:29 +0000 (09:24 +0000)]
[ARM] FP16: support vector zip and unzip
This is addressing PR38404.
Differential Revision: https://reviews.llvm.org/D50186
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338835
91177308-0d34-0410-b5e6-
96231b3b80d8
Dean Michael Berris [Fri, 3 Aug 2018 09:21:31 +0000 (09:21 +0000)]
[XRay][tools] Use Support/JSON.h in llvm-xray convert
Summary:
This change removes the ad-hoc implementation used by llvm-xray's
`convert` subcommand to generate JSON encoded catapult (AKA Chrome
Trace Viewer) trace output, to instead use the JSON encoder now in the
Support library.
Reviewers: kpw, zturner, eizan
Reviewed By: kpw
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D50129
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338834
91177308-0d34-0410-b5e6-
96231b3b80d8
Simon Pilgrim [Fri, 3 Aug 2018 09:20:02 +0000 (09:20 +0000)]
[X86] Add example of 'zero shift' guards on rotation patterns (PR34924)
Basic pattern that leaves an unnecessary select on a rotation by zero result. This variant is trivial - the more general case with a compare+branch to prevent execution of undefined shifts is more tricky.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338833
91177308-0d34-0410-b5e6-
96231b3b80d8
Sjoerd Meijer [Fri, 3 Aug 2018 09:12:56 +0000 (09:12 +0000)]
[ARM] FP16: support VFMA
This is addressing PR38404.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338830
91177308-0d34-0410-b5e6-
96231b3b80d8
Dean Michael Berris [Fri, 3 Aug 2018 09:06:11 +0000 (09:06 +0000)]
[XRay] fixup: add one more missing std::move(...)
Follow up to D48370.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338829
91177308-0d34-0410-b5e6-
96231b3b80d8
Dean Michael Berris [Fri, 3 Aug 2018 07:54:37 +0000 (07:54 +0000)]
[XRay] fixup: Add missing std::move(...)
Follow up to D48370.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338827
91177308-0d34-0410-b5e6-
96231b3b80d8
Dean Michael Berris [Fri, 3 Aug 2018 07:41:34 +0000 (07:41 +0000)]
[XRay] Fixup: remove 'noexcept' in defaulted move members
This is to appease stage1 builds using gcc.
Follow-up to D48370.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338826
91177308-0d34-0410-b5e6-
96231b3b80d8
Dean Michael Berris [Fri, 3 Aug 2018 07:18:39 +0000 (07:18 +0000)]
[XRay][llvm] Load XRay Profiles
Summary:
This change implements the profile loading functionality in LLVM to
support XRay's profiling mode in compiler-rt.
We introduce a type named `llvm::xray::Profile` which allows building a
profile representation. We can load an XRay profile from a file to build
Profile instances, or do it manually through the Profile type's API.
The intent is to get the `llvm-xray` tool to generate `Profile`
instances and use that as the common abstraction through which all
conversion and analysis can be done. In the future we can generate
`Profile` instances from `Trace` instances as well, through conversion
functions.
Some of the key operations supported by the `Profile` API are:
- Path interning (`Profile::internPath(...)`) which returns a unique path
identifier.
- Block appending (`Profile::addBlock(...)`) to add thread-associated
profile information.
- Path ID to Path lookup (`Profile::expandPath(...)`) to look up a
PathID and return the original interned path.
- Block iteration.
A 'Path' in this context represents the function call stack in
leaf-to-root order. This is represented as a path in an internally
managed prefix tree in the `Profile` instance. Having a handle (PathID)
to identify the unique Paths we encounter for a particular Profile
allows us to reduce the amount of memory required to associate profile
data to a particular Path.
This is the first of a series of patches to migrate the `llvm-stacks`
tool towards using a single profile representation.
Depends on D48653.
Reviewers: kpw, eizan
Reviewed By: kpw
Subscribers: mgorny, llvm-commits, hiraditya
Differential Revision: https://reviews.llvm.org/D48370
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338825
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Fri, 3 Aug 2018 07:01:10 +0000 (07:01 +0000)]
[X86] Remove all the vector NOP bitcast patterns. Use a few lines of code in the Select method in X86ISelDAGToDAG.cpp instead.
There are a lot of permutations of types here generating a lot of patterns in the isel table. It's more efficient to just ReplaceUses and RemoveDeadNode from the Select function.
The test changes are because we have a some shuffle patterns that have a bitcast as their root node. But the behavior is identical to another instruction whose pattern doesn't start with a bitcast. So this isn't a functional change.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338824
91177308-0d34-0410-b5e6-
96231b3b80d8
Hans Wennborg [Fri, 3 Aug 2018 07:00:08 +0000 (07:00 +0000)]
build_llvm_package.bat: Add OpenMP back
After r338721, it builds again.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338823
91177308-0d34-0410-b5e6-
96231b3b80d8
Chijun Sima [Fri, 3 Aug 2018 06:51:35 +0000 (06:51 +0000)]
[Dominators] Refine the logic of recalculate() in the DomTreeUpdater
Summary:
This patch refines the logic of `recalculate()` in the `DomTreeUpdater` in the following two aspects:
1. Previously, `recalculate()` tests whether there are pending updates/BBs awaiting deletion and then do recalculation under Lazy UpdateStrategy; and do recalculation immediately under Eager UpdateStrategy. (The former behavior is inherited from the `DeferredDominance` class). This is an inconsistency between two strategies and there is no obvious reason to do this. So the behavior is changed to always recalculate available trees when calling `recalculate()`.
2. Fix the issue of when DTU under Lazy UpdateStrategy holds nothing but with BBs awaiting deletion, after calling `recalculate()`, BBs awaiting deletion aren't flushed. An additional unittest is added to cover this case.
Reviewers: kuhar, dmgreen, brzycki, grosser, davide
Reviewed By: kuhar
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D50173
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338822
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Fri, 3 Aug 2018 06:12:56 +0000 (06:12 +0000)]
[X86] Support fp128 and/or/xor/load/store with VEX and EVEX encoded instructions.
Move all the patterns to X86InstrVecCompiler.td so we can keep SSE/AVX/AVX512 all in one place.
To save some patterns we'll use an existing DAG combine to convert f128 fand/for/fxor to integer when sse2 is enabled. This allows use to reuse all the existing patterns for v2i64.
I believe this now makes SHA instructions the only case where VEX/EVEX and legacy encoded instructions could be generated simultaneously.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338821
91177308-0d34-0410-b5e6-
96231b3b80d8
Hiroshi Inoue [Fri, 3 Aug 2018 05:39:48 +0000 (05:39 +0000)]
[InstSimplify] fold extracting from std::pair (2/2)
This is the second patch of the series which intends to enable jump threading for an inlined method whose return type is std::pair<int, bool> or std::pair<bool, int>.
The first patch is https://reviews.llvm.org/rL338485.
This patch handles code sequences that merges two values using `shl` and `or`, then extracts one value using `and`.
Differential Revision: https://reviews.llvm.org/D49981
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338817
91177308-0d34-0410-b5e6-
96231b3b80d8
Chijun Sima [Fri, 3 Aug 2018 05:08:17 +0000 (05:08 +0000)]
[Dominators] Convert existing passes and utils to use the DomTreeUpdater class
Summary:
This patch is the second in a series of patches related to the [[ http://lists.llvm.org/pipermail/llvm-dev/2018-June/123883.html | RFC - A new dominator tree updater for LLVM ]].
It converts passes (e.g. adce/jump-threading) and various functions which currently accept DDT in local.cpp and BasicBlockUtils.cpp to use the new DomTreeUpdater class.
These converted functions in utils can accept DomTreeUpdater with either UpdateStrategy and can deal with both DT and PDT held by the DomTreeUpdater.
Reviewers: brzycki, kuhar, dmgreen, grosser, davide
Reviewed By: brzycki
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D48967
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338814
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Fri, 3 Aug 2018 04:49:42 +0000 (04:49 +0000)]
[X86] When post-processing the DAG to remove zero extending moves for YMM/ZMM, make sure the producing instruction is VEX/XOP/EVEX encoded.
If the producing instruction is legacy encoded it doesn't implicitly zero the upper bits. This is important for the SHA instructions which don't have a VEX encoded version. We might also be able to hit this with the incomplete f128 support that hasn't been ported to VEX.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338812
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Fri, 3 Aug 2018 04:49:41 +0000 (04:49 +0000)]
[X86] Autogenerate complete checks. NFC
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338811
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Fri, 3 Aug 2018 03:45:19 +0000 (03:45 +0000)]
[X86] Add R13D to the isInefficientLEAReg in FixupLEAs.
I'm assuming the R13 restriction extends to R13D. Guessing this restriction is related to the funny encoding of this register as base always requiring a displacement to be encoded.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338806
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Fri, 3 Aug 2018 01:28:12 +0000 (01:28 +0000)]
[X86] Autogenerate complete checks. NFC
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338802
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Fri, 3 Aug 2018 01:20:32 +0000 (01:20 +0000)]
[X86] Autogenerate complete checks. NFC
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338799
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Fri, 3 Aug 2018 00:37:34 +0000 (00:37 +0000)]
[X86] Prevent promotion of i16 add/sub/and/or/xor to i32 if we can fold an atomic load and atomic store.
This makes them consistent with i8/i32/i64. Which still seems to be more aggressive on folding than icc, gcc, or MSVC.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338795
91177308-0d34-0410-b5e6-
96231b3b80d8
Philip Reames [Fri, 3 Aug 2018 00:21:56 +0000 (00:21 +0000)]
[LICM] Remove unneccessary safety check to increase sinking effectiveness
This one requires a bit of explaination. It's not every day you simply delete code to implement an optimization. :)
The transform in question is sinking an instruction from a loop to the uses in loop exiting blocks. We know (from LCSSA) that all of the uses outside the loop must be phi nodes, and after predecessor splitting, we know all phi users must have a single operand. Since the use must be strictly dominated by the def, we know from the definition of dominance/ssa that the exit block must execute along a (non-strict) subset of paths which reach the def. As a result, duplicating a potentially faulting instruction can not *introduce* a fault that didn't previously exist in the program.
The full story is that this patch builds on "rL338671: [LICM] Factor out fault legality from canHoistOrSinkInst [NFC]" which pulled this logic out of a common helper routine. As best I can tell, this check was originally added to the helper function for hoisting legality, later an incorrect fastpath for loads/calls was added, and then the bug was fixed by duplicating the fault safety check in the hoist path. This left the redundant check in the common code to pessimize sinking for no reason. I split it out in an NFC, and am not removing the unneccessary check. I wanted there to be something easy to revert in case I missed something.
Reviewed by: Anna Thomas (in person)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338794
91177308-0d34-0410-b5e6-
96231b3b80d8
Dave Lee [Fri, 3 Aug 2018 00:06:38 +0000 (00:06 +0000)]
objdump: Better handling of Mach-O universal binaries
Summary:
With Mach-O, there is a flag requirement discrepancy between working with
universal binaries and thin binaries. Many flags that don't require the `-macho`
flag (for example `-private-headers` and `-disassemble`) fail to work on
universal binaries unless `-macho` is given. When this happens, the error
message is unhelpful, stating:
The file was not recognized as a valid object file.
Which can lead to confusion.
This change allows generic flags to be used on universal binaries with and
without the `-macho` flag. This means flags that can be used for thin files can
be used consistently with fat files too.
To do this, the universal binary support within `ParseInputMachO()` is extracted
into a new function. This new function is called directly from `DumpInput()`
when the input binary is universal. Additionally the `-arch` flag validation in
`ParseInputMachO()` was extracted to be reused.
Reviewers: compnerd
Reviewed By: compnerd
Subscribers: keith, llvm-commits
Differential Revision: https://reviews.llvm.org/D48702
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338792
91177308-0d34-0410-b5e6-
96231b3b80d8
Eli Friedman [Thu, 2 Aug 2018 23:54:16 +0000 (23:54 +0000)]
[GlobalMerge] Allow merging globals with explicit section markings.
At least on ELF, it's impossible to tell from the object file whether
two globals with the same section marking were merged: the merged global
uses "private" linkage to hide its symbol, and the aliases look like
regular symbols. I can't think of any other reason to disallow it.
(Of course, we can only merge globals in the same section.)
The weird alignment handling matches AsmPrinter; our alignment handling
for global variables should probably be refactored.
Differential Revision: https://reviews.llvm.org/D49822
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338791
91177308-0d34-0410-b5e6-
96231b3b80d8
Tim Renouf [Thu, 2 Aug 2018 23:33:01 +0000 (23:33 +0000)]
[AMDGPU] Minor change to d16 buffer load implementation
Summary:
By not reconstructing the operand list of the SDNode, this change makes
it easier to add the forthcoming new tbuffer and buffer intrinsics.
Subscribers: arsenm, kzhuravl, wdng, nhaehnle, yaxunl, dstuttard, t-tye, llvm-commits
Differential Revision: https://reviews.llvm.org/D49995
Change-Id: I0cb79ef0801532645d7dd954a6d7355139db7b38
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338784
91177308-0d34-0410-b5e6-
96231b3b80d8
Tim Renouf [Thu, 2 Aug 2018 23:31:32 +0000 (23:31 +0000)]
[AMDGPU] Reworked SIFixWWMLiveness
Summary:
I encountered some problems with SIFixWWMLiveness when WWM is in a loop:
1. It sometimes gave invalid MIR where there is some control flow path
to the new implicit use of a register on EXIT_WWM that does not pass
through any def.
2. There were lots of false positives of registers that needed to have
an implicit use added to EXIT_WWM.
3. Adding an implicit use to EXIT_WWM (and adding an implicit def just
before the WWM code, which I tried in order to fix (1)) caused lots
of the values to be spilled and reloaded unnecessarily.
This commit is a rework of SIFixWWMLiveness, with the following changes:
1. Instead of considering any register with a def that can reach the WWM
code and a def that can be reached from the WWM code, it now
considers three specific cases that need to be handled.
2. A register that needs liveness over WWM to be synthesized now has it
done by adding itself as an implicit use to defs other than the
dominant one.
Also added the following fixmes:
FIXME: We should detect whether a register in one of the above
categories is already live at the WWM code before deciding to add the
implicit uses to synthesize its liveness.
FIXME: I believe this whole scheme may be flawed due to the possibility
of the register allocator doing live interval splitting.
Subscribers: arsenm, kzhuravl, wdng, nhaehnle, yaxunl, dstuttard, t-tye, llvm-commits
Differential Revision: https://reviews.llvm.org/D46756
Change-Id: Ie7fba0ede0378849181df3f1a9a7a39ed1a94a94
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338783
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Thu, 2 Aug 2018 23:30:38 +0000 (23:30 +0000)]
[X86] Allow 'atomic_store (neg/not atomic_load)' to isel to a RMW instruction.
There was a FIXMe in the td file about a type inference issue that was easy to fix.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338782
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Thu, 2 Aug 2018 23:30:31 +0000 (23:30 +0000)]
[X86] Add NEG and NOT test cases to atomic_mi.ll in preparation for fixing the FIXME in X86InstrCompiler.td to make these work for atomic load/store.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338781
91177308-0d34-0410-b5e6-
96231b3b80d8
Tim Renouf [Thu, 2 Aug 2018 22:53:57 +0000 (22:53 +0000)]
[AMDGPU] Avoid using divergent value in mubuf addr64 descriptor
Summary:
This fixes a problem where a load from global+idx generated incorrect
code on <=gfx7 when the index is divergent.
Subscribers: arsenm, kzhuravl, wdng, nhaehnle, yaxunl, dstuttard, t-tye, llvm-commits
Differential Revision: https://reviews.llvm.org/D47383
Change-Id: Ib4d177d6254b1dd3f8ec0203fdddec94bd8bc5ed
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338779
91177308-0d34-0410-b5e6-
96231b3b80d8
Zachary Turner [Thu, 2 Aug 2018 22:37:40 +0000 (22:37 +0000)]
[MS Demangler] Fix some tests that are no longer broken.
These were fixed with earlier patches, but had not yet been
re-enabled.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338778
91177308-0d34-0410-b5e6-
96231b3b80d8
Krzysztof Parzyszek [Thu, 2 Aug 2018 22:17:53 +0000 (22:17 +0000)]
[Hexagon] Simplify CFG after atomic expansion
This will remove suboptimal branching from the generated ll/sc loops.
The extra simplification pass affects a lot of testcases, which have
been modified to accommodate this change: either by modifying the
test to become immune to the CFG simplification, or (less preferablt)
by adding option -hexagon-initial-cfg-clenaup=0.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338774
91177308-0d34-0410-b5e6-
96231b3b80d8
Heejin Ahn [Thu, 2 Aug 2018 21:44:24 +0000 (21:44 +0000)]
[WebAssembly] Support for atomic.wait / atomic.wake instructions
Summary:
This adds support for atomic.wait / atomic.wake instructions in the wasm
thread proposal.
Reviewers: dschuff
Subscribers: sbc100, jgravelle-google, sunfish, llvm-commits
Differential Revision: https://reviews.llvm.org/D49395
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338770
91177308-0d34-0410-b5e6-
96231b3b80d8
Krzysztof Parzyszek [Thu, 2 Aug 2018 20:45:06 +0000 (20:45 +0000)]
[Unittests] Fix returning string in SolveQuadraticEquationWrap
Returning a Twine caused runtime failures. Convert it to std::string
before retuning.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338768
91177308-0d34-0410-b5e6-
96231b3b80d8
Craig Topper [Thu, 2 Aug 2018 20:28:45 +0000 (20:28 +0000)]
[X86] Autogenerate complete checks. NFC
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338765
91177308-0d34-0410-b5e6-
96231b3b80d8
Lang Hames [Thu, 2 Aug 2018 20:13:58 +0000 (20:13 +0000)]
[ORC] Add a re-exports fallback definition generator.
An instance of ReexportsFallbackDefinitionGenerator can be attached to a VSO
(via setFallbackDefinitionGenerator) to re-export symbols on demandy from a
backing VSO.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338764
91177308-0d34-0410-b5e6-
96231b3b80d8
George Burgess IV [Thu, 2 Aug 2018 19:50:27 +0000 (19:50 +0000)]
[Support] Add an enable bit to our DebugCounters
r337748 made us start incrementing DebugCounters all of the time. This
makes tsan unhappy in multithreaded environments.
Since it doesn't make much sense to use DebugCounters with multiple
threads, this patch makes us only count anything if the user passed a
-debug-counter option or if some other piece of code explicitly asks
for it (e.g. the pass in D50031).
The amount of global state here makes writing a unittest for this
behavior somewhat awkward. So, no test is provided.
Differential Revision: https://reviews.llvm.org/D50150
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338762
91177308-0d34-0410-b5e6-
96231b3b80d8
Krzysztof Parzyszek [Thu, 2 Aug 2018 19:38:18 +0000 (19:38 +0000)]
Unbreak build after r338758: specify lambda return type explicitly
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338760
91177308-0d34-0410-b5e6-
96231b3b80d8
Paul Robinson [Thu, 2 Aug 2018 19:29:38 +0000 (19:29 +0000)]
[DebugInfo/DWARF] Remove redundant iterator type. NFC
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@338759
91177308-0d34-0410-b5e6-
96231b3b80d8