Johannes Doerfert [Mon, 11 May 2020 20:22:14 +0000 (15:22 -0500)]
[Attributor][FIX] Disallow function signature rewrite for casted calls
We will now ensure ensure the return type of called function is the type
of all call sites we are going to rewrite. This avoids a problem
partially fixed by D79680. The part that was not covered is a use of
this "weird" casted call site (see `@func3` in `misc_crash.ll`).
misc_crash.ll checks are auto-generated now.
Johannes Doerfert [Mon, 11 May 2020 17:22:41 +0000 (12:22 -0500)]
[Attributor] Make AAIsDead dependences optional to prevent top state
We should never give up on AAIsDead as it guards other AAs from
unreachable code (in which SSA properties are meaningless). We did
however use required dependences on some queries in AAIsDead which
caused us to invalidate AAIsDead if the queried AA got invalidated.
We now use optional dependences instead. The bug that exposed this is
added to the liveness.ll test and other test changes show the impact.
Bug report by @sdmitriev.
Johannes Doerfert [Mon, 11 May 2020 17:20:27 +0000 (12:20 -0500)]
[Attributor] Force update of "newly live" abstract attributes
During an update of AAIsDead, new instructions become live. If we query
information from them, the result is often just the initial state, e.g.,
for call site `noreturn` and `nounwind`. We will now trigger an update
for cached attributes during the AAIsDead update, though other AAs might
later use the same API.
Tim Keith [Mon, 11 May 2020 20:28:05 +0000 (13:28 -0700)]
[flang] Fix bug with IMPORT of USE of USE
When a module contained an import of a use-association of a
use-association, we weren't recognizing that the symbols was needed.
The fix is the follow all of the use-associations using `GetUltimate()`.
Differential Revision: https://reviews.llvm.org/D79737
LLVM GN Syncbot [Mon, 11 May 2020 20:09:12 +0000 (20:09 +0000)]
[gn build] Port
e97a3e5d9d4
Scott Constable [Mon, 11 May 2020 18:30:55 +0000 (11:30 -0700)]
[X86] Add Support for Load Hardening to Mitigate Load Value Injection (LVI)
After finding all such gadgets in a given function, the pass minimally inserts
LFENCE instructions in such a manner that the following property is satisfied:
for all SOURCE+SINK pairs, all paths in the CFG from SOURCE to SINK contain at
least one LFENCE instruction. The algorithm that implements this minimal
insertion is influenced by an academic paper that minimally inserts memory
fences for high-performance concurrent programs:
http://www.cs.ucr.edu/~lesani/companion/oopsla15/OOPSLA15.pdf
The algorithm implemented in this pass is as follows:
1. Build a condensed CFG (i.e., a GadgetGraph) consisting only of the following components:
-SOURCE instructions (also includes function arguments)
-SINK instructions
-Basic block entry points
-Basic block terminators
-LFENCE instructions
2. Analyze the GadgetGraph to determine which SOURCE+SINK pairs (i.e., gadgets) are already mitigated by existing LFENCEs. If all gadgets have been mitigated, go to step 6.
3. Use a heuristic or plugin to approximate minimal LFENCE insertion.
4. Insert one LFENCE along each CFG edge that was cut in step 3.
5. Go to step 2.
6. If any LFENCEs were inserted, return true from runOnFunction() to tell LLVM that the function was modified.
By default, the heuristic used in Step 3 is a greedy heuristic that avoids
inserting LFENCEs into loops unless absolutely necessary. There is also a
CLI option to load a plugin that can provide even better optimization,
inserting fewer fences, while still mitigating all of the LVI gadgets.
The plugin can be found here: https://github.com/intel/lvi-llvm-optimization-plugin,
and a description of the pass's behavior with the plugin can be found here:
https://software.intel.com/security-software-guidance/insights/optimized-mitigation-approach-load-value-injection.
Differential Revision: https://reviews.llvm.org/D75937
Scott Constable [Mon, 11 May 2020 17:25:35 +0000 (10:25 -0700)]
[X86] Add a Pass that builds a Condensed CFG for Load Value Injection (LVI) Gadgets
Adds a new data structure, ImmutableGraph, and uses RDF to find LVI gadgets and add them to a MachineGadgetGraph.
More specifically, a new X86 machine pass finds Load Value Injection (LVI) gadgets consisting of a load from memory (i.e., SOURCE), and any operation that may transmit the value loaded from memory over a covert channel, or use the value loaded from memory to determine a branch/call target (i.e., SINK).
Also adds a new target feature to X86: +lvi-load-hardening
The feature can be added via the clang CLI using -mlvi-hardening.
Differential Revision: https://reviews.llvm.org/D75936
jasonliu [Mon, 11 May 2020 20:03:37 +0000 (20:03 +0000)]
Move PowerPC specific test under PowerPC directive to fix build break
Fix build break in x86 platform which introduced by
https://reviews.llvm.org/D79127
Arthur Eubanks [Mon, 11 May 2020 01:06:55 +0000 (18:06 -0700)]
Run Coverage pass before other *San passes under new pass manager
Summary:
This fixes compiler-rt/test/msan/coverage-levels.cpp under the new pass manager (final check-msan test!).
Under the old pass manager, the coverage pass would run before the MSan pass. The opposite happened under the new pass manager. The MSan pass adds extra basic blocks, changing the number of coverage callbacks.
Reviewers: vitalybuka, leonardchan
Subscribers: cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D79698
Sean Silva [Sun, 10 May 2020 02:17:41 +0000 (19:17 -0700)]
[mlir][shape] Tidy up shape.shape_of
Summary:
- Mark it NoSideEffect
- Add custom parser/printer
Differential Revision: https://reviews.llvm.org/D79684
Sean Silva [Sun, 10 May 2020 00:52:35 +0000 (17:52 -0700)]
[mlir][Value] Add v.getDefiningOp<OpTy>()
Summary:
This makes a common pattern of
`dyn_cast_or_null<OpTy>(v.getDefiningOp())` more concise.
Differential Revision: https://reviews.llvm.org/D79681
jasonliu [Mon, 11 May 2020 18:43:07 +0000 (18:43 +0000)]
[XCOFF][AIX] Emit correct alignment for csect
Summary:
This patch tries to emit the correct alignment result for both
object file generation path and assembly path.
Reviewed by: hubert.reinterpretcast, DiggerLin, daltenty
Differential Revision: https://reviews.llvm.org/D79127
Jez Ng [Sun, 10 May 2020 21:15:42 +0000 (14:15 -0700)]
[lld-macho] Re-add dylink-lazy test
The initial attempt didn't work on Windows; apparently Powershell has a
different syntax for running commands sequentially and concatenating
their outputs. So I've created two temporary files instead.
Differential Revision: https://reviews.llvm.org/D79697
Melanie Blower [Mon, 11 May 2020 19:29:52 +0000 (12:29 -0700)]
FP LangOpts should not be dependent on CGOpt
This bug was observed by Apple since their compiler processes LangOpts and CGOpts in a different order.
Reviewed By: rjmccall
Differential Revision: https://reviews.llvm.org/D79735
Sanjay Patel [Mon, 11 May 2020 19:20:57 +0000 (15:20 -0400)]
[VectorCombine] account for extra uses in scalarization cost
Follow-up to D79452.
Mimics the extra use cost formula for the inverse transform with extracts.
Sanjay Patel [Mon, 11 May 2020 18:55:32 +0000 (14:55 -0400)]
[VectorCombine] add tests for possible scalarization with extra uses; NFC
Joel E. Denny [Mon, 4 May 2020 22:05:55 +0000 (18:05 -0400)]
[FileCheck] Support comment directives
Sometimes you want to disable a FileCheck directive without removing
it entirely, or you want to write comments that mention a directive by
name. The `COM:` directive makes it easy to do this. For example,
you might have:
```
; X32: pinsrd_1:
; X32: pinsrd $1, 4(%esp), %xmm0
; COM: FIXME: X64 isn't working correctly yet for this part of codegen, but
; COM: X64 will have something similar to X32:
; COM:
; COM: X64: pinsrd_1:
; COM: X64: pinsrd $1, %edi, %xmm0
```
Without this patch, you need to use some combination of rewording and
directive syntax mangling to prevent FileCheck from recognizing the
commented occurrences of `X32:` and `X64:` above as directives.
Moreover, FileCheck diagnostics have been proposed that might complain
about the occurrences of `X64` that don't have the trailing `:`
because they look like directive typos:
<http://lists.llvm.org/pipermail/llvm-dev/2020-April/140610.html>
I think dodging all these problems can prove tedious for test authors,
and directive syntax mangling already makes the purpose of existing
test code unclear. `COM:` can avoid all these problems.
This patch also updates the small set of existing tests that define
`COM` as a check prefix:
- clang/test/CodeGen/default-address-space.c
- clang/test/CodeGenOpenCL/addr-space-struct-arg.cl
- clang/test/Driver/hip-device-libs.hip
- llvm/test/Assembler/drop-debug-info-nonzero-alloca.ll
I think lit should support `COM:` as well. Perhaps `clang -verify`
should too.
Reviewed By: jhenderson, thopre
Differential Revision: https://reviews.llvm.org/D79276
Joel E. Denny [Mon, 11 May 2020 13:57:37 +0000 (09:57 -0400)]
[FileCheck] Make invalid prefix diagnostics more precise
This will prove especially helpful after D79276, which introduces
comment prefixes. Specifically, identifying whether there's a
uniqueness violation will be helpful as prefixes will be required to
be unique across both check prefixes and comment prefixes.
Also, remove a related comment about `cl::list` that no longer seems
relevant now that FileCheck is also a library.
Reviewed By: jhenderson, thopre
Differential Revision: https://reviews.llvm.org/D79375
Joel E. Denny [Mon, 11 May 2020 13:51:32 +0000 (09:51 -0400)]
[OpenMP][NFC] Fix `not` sustitution in tests
D78566 introduced a `\bnot\b` lit substitution in OpenMP test suites.
However, that would corrupt a command like
`FileCheck -implicit-check-not` or any file name like `%t.not`. We
could use lookbehind/lookahead assertions to avoid such cases, but
this patch switches to `%not` (suggested during the D78566 review) as
a safer option.
Reviewed By: jdoerfert
Differential Revision: https://reviews.llvm.org/D79529
Tim Keith [Mon, 11 May 2020 18:38:53 +0000 (11:38 -0700)]
[flang][NFC] Simplify semantics test scripts
There were several different ways of handling the option to f18 to
find predefined modules:
- test_errors.sh was created by cmake substituting
FLANG_INTRINSIC_MODULES_DIR into test_errors.sh.in
- some tests used the flang script which has the option built it
- some tests used %f18_with_includes which was replaced by the path
to f18 plus the -I option
- some included -I../../include/flang in their run command
To make this more consistent, change %f18 to include the
-intrinsic-module-directory option and use it everywhere, including
to replace %flang and %f18_with_includes. This requires changing all
of the invocations of the test scripts to put %f18 at the end so that
it can expand to more than one argument.
This eliminates the need to generate test_errors.sh which means we
don't need flang/test/Semantics/CMakeLists.txt or the %B substitution.
That makes the test_errors.sh command like the others, replacing
%B/test/Semantics/test_errors.sh with %S/test_errors.sh.
Also remove the OPTIONS: functionality as custom options can be included
in the RUN: command. And remove -I/../../include/flang as that is now
always included.
Differential Revision: https://reviews.llvm.org/D79634
Jessica Paquette [Sat, 9 May 2020 01:10:46 +0000 (18:10 -0700)]
[AArch64][GlobalISel] Make LR livein to entry in llvm.returnaddress selection
This fixes a couple verifier failures on this bot:
http://green.lab.llvm.org/green/job/test-suite-verify-machineinstrs-aarch64-globalisel-O0-g/
The failures show up in eeprof-1.c and pr17377.c in the GCC C Torture Suite.
Specifically:
*** Bad machine code: MBB has allocatable live-in, but isn't entry or landing-pad. ***
- function: foo
- basic block: %bb.3 if.end (0x7fac7106dfc8)
- p. register: $lr
and
*** Bad machine code: Using an undefined physical register ***
- function: f
- basic block: %bb.1 entry (0x7f8941092588)
- instruction: %18:gpr64 = COPY $lr
- operand 1: $lr
Unlike SDAG, we were setting LR as a live in to the block containing the
returnaddress.
Also, this ensures that we don't add LR as a livein to the entry block twice.
In MachineBasicBlock.h there's a comment saying
"Note that it is an error to add the same register to the same set more than
once unless the intention is to call sortUniqueLiveIns after all registers are
added."
so it's probably good to avoid adding LR twice.
Surprisingly the verifier doesn't complain about that. Maybe it should.
Differential Revision: https://reviews.llvm.org/D79657
Sam Elliott [Mon, 11 May 2020 17:52:42 +0000 (18:52 +0100)]
[RISCV] Support Constant Pools in Load/Store Peephole
Summary:
RISC-V uses a post-select peephole pass to optimise
`(load/store (ADDI $reg, %lo(addr)), 0)` into `(load/store $reg, %lo(addr))`.
This peephole wasn't firing for accesses to constant pools, which is how we
materialise most floating point constants.
This adds support for the constantpool case, which improves code generation for
lots of small FP loading examples. I have not added any tests because this
structure is well-covered by the `fp-imm.ll` testcases, as well as almost
all other uses of floating point constants in the RISC-V backend tests.
Reviewed By: luismarques, asb
Differential Revision: https://reviews.llvm.org/D79523
Florian Hahn [Mon, 11 May 2020 16:45:51 +0000 (17:45 +0100)]
[Matrix] Add matrix type to Clang.
This patch adds a matrix type to Clang as described in the draft
specification in clang/docs/MatrixSupport.rst. It introduces a new option
-fenable-matrix, which can be used to enable the matrix support.
The patch adds new MatrixType and DependentSizedMatrixType types along
with the plumbing required. Loads of and stores to pointers to matrix
values are lowered to memory operations on 1-D IR arrays. After loading,
the loaded values are cast to a vector. This ensures matrix values use
the alignment of the element type, instead of LLVM's large vector
alignment.
The operators and builtins described in the draft spec will will be added in
follow-up patches.
Reviewers: martong, rsmith, Bigcheese, anemet, dexonsmith, rjmccall, aaron.ballman
Reviewed By: rjmccall
Differential Revision: https://reviews.llvm.org/D72281
Nico Weber [Mon, 11 May 2020 17:46:21 +0000 (13:46 -0400)]
[gn build] (manually) port
bba38de50c9
Sanjay Patel [Mon, 11 May 2020 14:48:47 +0000 (10:48 -0400)]
[CGP][x86] add test for funnel-shift with cross-block splat shift-amount; NFC
Valentin Clement [Mon, 11 May 2020 17:13:58 +0000 (13:13 -0400)]
[flang] Fix compilation after rename of Loop dialect to scf
Summary: Flang was not compiling correctly after the renaming of Loop dialect to SCF. This patch fixes the problem.
Reviewers: ftynse, DavidTruby
Reviewed By: ftynse
Subscribers: aartbik, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D79723
Sam Elliott [Mon, 11 May 2020 17:13:41 +0000 (18:13 +0100)]
Revert "[RISCV] Support Constant Pools in Load/Store Peephole"
This reverts commit
fe69dfebcfa007e23bf528fff8b96ac63e593e34, due to
a slight change in the API.
Sean Fertile [Mon, 11 May 2020 17:03:45 +0000 (13:03 -0400)]
[PowerPC][NFC] Convert an if/else to a conditional.
Change an if else to use a conditional which is shorter. Also name the
conditonal value to make the code clearer.
Thomas Lively [Mon, 11 May 2020 16:55:43 +0000 (09:55 -0700)]
[WebAssembly] Add wasm-specific vector shuffle builtin and intrinsic
Summary:
Although using `__builtin_shufflevector` and the `shufflevector`
instruction works fine, they are not opaque to the optimizer. As a
result, DAGCombine can potentially reduce the number of shuffles and
change the shuffle masks. This is unexpected behavior for users of the
WebAssembly SIMD intrinsics who have crafted their shuffles to
optimize the code generated by engines. This patch solves the problem
by adding a new shuffle intrinsic that is opaque to the optimizers in
line with the decision of the WebAssembly SIMD contributors at
https://github.com/WebAssembly/simd/issues/196#issuecomment-
622494748. In
the future we may implement custom DAG combines to properly optimize
shuffles and replace this solution.
Reviewers: aheejin, dschuff
Subscribers: sbc100, jgravelle-google, hiraditya, sunfish, cfe-commits, llvm-commits
Tags: #clang, #llvm
Differential Revision: https://reviews.llvm.org/D66983
Sam Elliott [Mon, 11 May 2020 16:58:33 +0000 (17:58 +0100)]
[RISCV] Support Constant Pools in Load/Store Peephole
Summary:
RISC-V uses a post-select peephole pass to optimise
`(load/store (ADDI $reg, %lo(addr)), 0)` into `(load/store $reg, %lo(addr))`.
This peephole wasn't firing for accesses to constant pools, which is how we
materialise most floating point constants.
This adds support for the constantpool case, which improves code generation for
lots of small FP loading examples. I have not added any tests because this
structure is well-covered by the `fp-imm.ll` testcases, as well as almost
all other uses of floating point constants in the RISC-V backend tests.
Reviewed By: luismarques, asb
Differential Revision: https://reviews.llvm.org/D79523
Hongtao Yu [Mon, 11 May 2020 16:38:29 +0000 (09:38 -0700)]
Properly add out-of-module functions to the import list
This patch addresses two issues related to adding inline functions to the import list while recursively going through the profiling data.
1. For callsite samples, only add an inlined function to the import list if it's from outside of the module (i.e. only has a declaration inside the module).
2. For body samples, add each target function to the import list if it's from outside of the module (i.e. only has a declaration inside the module). Previously we were using getSubProgram() to check whether it has dbg info, which is inaccurate. This fix properly add imports and could improve the quality of the pass.
Added a few changes to the test to catch these cases.
Differential Revision: https://reviews.llvm.org/D79379
Vedant Kumar [Wed, 6 May 2020 23:08:19 +0000 (16:08 -0700)]
[lldb/test] Fix for flakiness in TestNSDictionarySynthetic
Summary:
TestNSDictionarySynthetic sets up an NSURL which does not initialize its
_baseURL member. When the test runs and we print out the NSURL, we print
out some garbage memory pointed-to by the _baseURL member, like:
```
_baseURL = 0x0800010020004029 @"d��qX"
```
and this can cause a python unicode decoding error like:
```
UnicodeDecodeError: 'utf8' codec can't decode byte 0xa0 in position
10309: invalid start byte
```
There's a discrepancy here because lldb's StringPrinter facility tries
to only print out "printable" sequences (see: isprint32()), whereas python
rejects the StringPrinter output as invalid utf8. For the specific error
seen above, lldb's `isprint32(0xa0) = true`, even though 0xa0 is not
really "printable" in the usual sense.
The problem is that lldb and python disagree on what exactly is
"printable". Both have dismayingly hand-rolled utf8 validation code
(c.f. _Py_DecodeUTF8Ex), and I can't really tell which one is more
correct.
I tried replacing lldb's isprint32() with a call to libc's iswprint():
this satisfied python, but broke emoji printing :|.
Now, I believe that lldb (and python too) ought to just call into some
battle-tested utf library, and that we shouldn't aim for compatibility
with python's strict unicode decoding mode until then.
FWIW I ran this test under an ASanified lldb hundreds of times but
didn't turn up any other issues.
rdar://
62941711
Reviewers: JDevlieghere, jingham, shafik
Subscribers: lldb-commits
Tags: #lldb
Differential Revision: https://reviews.llvm.org/D79645
Julian Lettner [Thu, 7 May 2020 02:05:31 +0000 (19:05 -0700)]
[compile-rt] Reduce #ifdef noise for ptrauth
Create a sanitizer_ptrauth.h header that #includes <ptrauth> when
available and defines just the required macros as "no ops" otherwise.
This should avoid the need for excessive #ifdef'ing.
Follow-up to and discussed in: https://reviews.llvm.org/D79132
Reviewed By: delcypher
Differential Revision: https://reviews.llvm.org/D79540
LLVM GN Syncbot [Mon, 11 May 2020 16:35:33 +0000 (16:35 +0000)]
[gn build] Port
bf95cf4a681
Zola Bridges [Mon, 11 May 2020 16:33:55 +0000 (09:33 -0700)]
[x86][seses] Introduce SESES pass for LVI
This is an implementation of Speculative Execution Side Effect
Suppression which is intended as a last resort mitigation against Load
Value Injection, LVI, a newly disclosed speculative execution side
channel vulnerability.
One pager:
https://software.intel.com/security-software-guidance/software-guidance/load-value-injection
Deep dive:
https://software.intel.com/security-software-guidance/insights/deep-dive-load-value-injection
The mitigation consists of a compiler pass that inserts an LFENCE before
each memory read instruction, memory write instruction, and the first
branch instruction in a group of terminators at the end of a basic
block. The goal is to prevent speculative execution, potentially based
on misspeculated conditions and/or containing secret data, from leaking
that data via side channels embedded in such instructions.
This is something of a last-resort mitigation: it is expected to have
extreme performance implications and it may not be a complete mitigation
due to trying to enumerate side channels.
In addition to the full version of the mitigation, this patch
implements three flags to turn off part of the mitigation. These flags
are disabled by default. The flags are not intended to result in a
secure variant of the mitigation. The flags are intended to be used by
users who would like to experiment with improving the performance of
the mitigation. I ran benchmarks with each of these flags enabled in
order to find if there was any room for further optimization of LFENCE
placement with respect to LVI.
Performance Testing Results
When applying this mitigation to BoringSSL, we see the following
results. These are a summary/aggregation of the performance changes when
this mitigation is applied versus when no mitigation is applied.
Fully Mitigated vs Baseline
Geometric mean
0.071 (Note: This can be read as the ops/s of the mitigated
program was 7.1% of the ops/s of the unmitigated program.)
Minimum
0.041
Quartile 1
0.060
Median
0.063
Quartile 3
0.077
Maximum
0.230
Reviewed By: george.burgess.iv
Differential Revision: https://reviews.llvm.org/D75939
Nicolas Vasilache [Mon, 11 May 2020 16:09:18 +0000 (12:09 -0400)]
[mlir] Simplify and better document std.view semantics
This [discussion](https://llvm.discourse.group/t/viewop-isnt-expressive-enough/991/2) raised some concerns with ViewOp.
In particular, the handling of offsets is incorrect and does not match the op description.
Note that with an elemental type change, offsets cannot be part of the type in general because sizeof(srcType) != sizeof(dstType).
Howerver, offset is a poorly chosen term for this purpose and is renamed to byte_shift.
Additionally, for all intended purposes, trying to support non-identity layouts for this op does not bring expressive power but rather increases code complexity.
This revision simplifies the existing semantics and implementation.
This simplification effort is voluntarily restrictive and acts as a stepping stone towards supporting richer semantics: treat the non-common cases as YAGNI for now and reevaluate based on concrete use cases once a round of simplification occurred.
Differential revision: https://reviews.llvm.org/D79541
Zola Bridges [Mon, 4 May 2020 18:51:16 +0000 (11:51 -0700)]
[llvm][utils] Remove git-svn folder + scripts
Summary:
These tools are no longer useful since we've migrated off of SVN, so
this patch deletes them.
Link: http://lists.llvm.org/pipermail/llvm-dev/2020-May/141386.html
Unless there is opposition in the RFC thread, I'll submit the patch on
May 10, 2020.
I searched through the repo to confirm there were no mentions of the scripts
in other scripts or documentation.
Reviewed By: echristo, tstellar, MaskRay
Differential Revision: https://reviews.llvm.org/D79348
Kostya Kortchinsky [Mon, 11 May 2020 16:16:54 +0000 (09:16 -0700)]
Add vendor identity check for Hygon Dhyana processor in Scudo
Summary:
The Hygon Dhyana processor supports hardware CRC32.
Related link:
https://reviews.llvm.org/D78874
Result of "make check":
Testing Time: 1364.04s
Unsupported Tests: 317
Expected Passes : 36802
Expected Failures: 161
[100%] Built target check-llvm
[100%] Built target check
Reviewers: cryptoad
Reviewed By: cryptoad
Subscribers: craig.topper, cryptoad, cfe-commits, #sanitizers, llvm-commits
Tags: #clang, #sanitizers, #llvm
Differential Revision: https://reviews.llvm.org/D62368
LLVM GN Syncbot [Mon, 11 May 2020 16:11:59 +0000 (16:11 +0000)]
[gn build] Port
48fa355ed4b
Mircea Trofin [Fri, 8 May 2020 03:35:08 +0000 (20:35 -0700)]
[llvm][NFC] Move inlining decision-related APIs in InliningAdvisor.
Summary: Factoring out in preparation to https://reviews.llvm.org/D79042
Reviewers: dblaikie, davidxl
Subscribers: mgorny, eraman, hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D79613
Pavel Labath [Mon, 11 May 2020 15:51:46 +0000 (17:51 +0200)]
[lldb] Speculative fix for the entry values test on arm64
The bot fails with a message which seems to indicate a problem in option
parsing. Avoid grouping the options to see if that helps.
Sergey Dmitriev [Mon, 11 May 2020 15:05:51 +0000 (08:05 -0700)]
[Attributor] Fix for a crash on RAUW when rewriting function signature
Reviewers: jdoerfert, sstefan1, uenoku
Reviewed By: uenoku
Subscribers: hiraditya, uenoku, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D79680
Kang Zhang [Mon, 11 May 2020 15:02:09 +0000 (15:02 +0000)]
[PowerPC] Use PredictableSelectIsExpensive to enable select to branch in CGP
Summary:
This patch will set the variable PredictableSelectIsExpensive to do the
select to if based on BranchProbability in CodeGenPrepare.
When the BranchProbability more than MinPercentageForPredictableBranch,
PPC will convert SELECT to branch.
Reviewed By: nemanjai
Differential Revision: https://reviews.llvm.org/D71883
Kamlesh Kumar [Mon, 11 May 2020 14:47:42 +0000 (10:47 -0400)]
[Clang] Fix the incorrect return type of atomic_is_lock_free
Fixing the return type of atomic_is_lock_free as per
https://en.cppreference.com/w/c/atomic/atomic_is_lock_free
Differential Revision: https://reviews.llvm.org/D79504
Nicolas Vasilache [Mon, 11 May 2020 14:39:16 +0000 (10:39 -0400)]
[mlir][Linalg] NFC - Refactor and simplify Promotion
Summary: This revision introduces LinalgPromotionOptions to more easily control the application of promotion patterns. It also simplifies the different entry points into Promotion in preparation for some behavior change in subsequent revisions.
Differential Revision: https://reviews.llvm.org/D79489
Sam McCall [Mon, 11 May 2020 14:44:01 +0000 (16:44 +0200)]
Revert "[DAGCombine] Remove the getNegatibleCost to avoid the out of sync with getNegatedExpression"
This reverts commit
3c44c441db0f8d7e210806b5b221cd9ed66f2d7b.
Causes infloops on some inputs, see https://reviews.llvm.org/D77319 for repro
Sanjay Patel [Mon, 11 May 2020 14:39:24 +0000 (10:39 -0400)]
[x86] add test for funnel shift in loop with cross-block splat variable; NFC
Pavel Labath [Thu, 7 May 2020 11:06:02 +0000 (13:06 +0200)]
[lldb/test] Make "inline" tests handle multiple statements at the same location
Summary:
The test machinery translates each continuous block of "//%" comments
into a single breakpoint. If there's no code between the blocks the
breakpoints will end up at the same location in the program. When the
process stops at a breakpoint lldb correctly reports all breakpoint IDs,
but the test machinery only looks at the first one. This results in a
very dangerous situation as it means some checks can be silently
stopped.
This patch fixes that by making the test machinery iterate through all
breakpoints at a given location and execute all commands.
Reviewers: vsk, JDevlieghere
Subscribers: lldb-commits
Tags: #lldb
Differential Revision: https://reviews.llvm.org/D79563
Pavel Labath [Wed, 6 May 2020 14:06:42 +0000 (16:06 +0200)]
[lldb] Revive TestBasicEntryValuesX86_64
Summary:
This function rewrites the test to be (hopefully) less susceptible to
codegen changes and re-enables it.
The most interesting changes are:
- use an __attribute__((optnone)) function instead of a volatile asm to
"use" a value. This isn't strictly necessary, but it makes the
function simpler while achieving the same effect.
- use a call to a function with the exact same signature instead of a
volatile asm to "destroy" arguments. This makes the independent of the
ABI, and (together with avoiding the usage of the arguments after the
call) ensures that the compiler has no reason to move the argument
from its initial register (previously we needed to guess where will
the compiler store the arguments).
Reviewers: vsk, djtodoro, dblaikie
Subscribers: lldb-commits
Tags: #lldb
Differential Revision: https://reviews.llvm.org/D79491
Artem Dergachev [Mon, 11 May 2020 14:24:09 +0000 (17:24 +0300)]
[analyzer] SATestBuild.py: Allow comments in run_static_analyzer.cmd.
Because those can get really weird sometimes.
rtayl [Fri, 8 May 2020 18:20:47 +0000 (14:20 -0400)]
[mlir][rocdl] Add xdlops intrinsics to rocdl dialect
Summary: This adds xdlops (mfma) to the rocdl dialect and also tests the translation to llvm ir.
Reviewers: ftynse
Subscribers: mehdi_amini, rriddle, jpienaar, shauheen, antiagainst, nicolasvasilache, arpith-jacob, mgester, lucyrfox, liufengdb, stephenneuendorffer, Joonsoo, grosul1, frgossen, Kayjukh, llvm-commits
Tags: #llvm #mlir
Differential Revision: https://reviews.llvm.org/D79642
Raphael Isemann [Mon, 11 May 2020 14:02:17 +0000 (16:02 +0200)]
[lldb] Add missing nullptr checks to AppleObjCRuntime::GetBacktraceThreadFromException
Summary:
We got a few crash reports where LLDB crashes while derefencing the `frames_value` shared_ptr in the AppleObjCRuntime::GetBacktraceThreadFromException. `GetChildMemberWithName` returns a nullptr when an error occurs, so this seems to be just a missing nullptr check.
This patch adds that nullptr check and the other ones in the similar code directly below.
Fixes rdar://
62174039
Reviewers: jingham, kubamracek
Reviewed By: jingham
Subscribers: abidh, JDevlieghere
Differential Revision: https://reviews.llvm.org/D78798
Gongyu Deng [Mon, 11 May 2020 13:54:13 +0000 (15:54 +0200)]
Tab completion for breakpoint write and breakpoint name add/delete
Summary: Apply the common completion created in [[ https://reviews.llvm.org/D75418 | Revision D75418 ]] to the commands `breakpoint write` and `breakpoint name add/delete`.
Reviewers: teemperor, JDevlieghere
Reviewed By: teemperor
Tags: #lldb
Differential Revision: https://reviews.llvm.org/D79686
Alex Zinenko [Mon, 11 May 2020 13:26:11 +0000 (15:26 +0200)]
[mlir] More support for detached regions in affine symbol checkers
Add documentation to `isToLevelValue` explaining its behavior for
detached regions, and fix the overloaded version that accepts `Region`.
Peter Smith [Fri, 8 May 2020 12:19:12 +0000 (13:19 +0100)]
[ELF][ARM] Support /DISCARD/ of subset of .ARM.exidx sections
Both the .ARM.exidx and .eh_frame sections have a custom SyntheticSection
that acts as a container for the InputSections. The InputSections are added
to the SyntheticSection prior to /DISCARD/ which limits the affect a
/DISCARD/ can have to the whole SyntheticSection. In the majority of cases
this is sufficient as it is not common to discard subsets of the
InputSections. The Linux kernel has one of these scripts which has something
like:
/DISCARD/ : { *(.ARM.exidx.exit.text) *(.ARM.extab.exit.text) ... }
The .ARM.exidx.exit.text are not discarded because the InputSection has been
transferred to the Synthetic Section. The *(.ARM.extab.exit.text) sections
have not so they are discarded. When we come to write out the .ARM.exidx
sections the dangling references from .ARM.exidx.exit.text to
.ARM.extab.exit.text currently cause relocation out of range errors, but
could as easily cause a fatal error message if we check for dangling
references at relocation time.
This patch attempts to respect the /DISCARD/ command by running it on the
.ARM.exidx InputSections stored in the SyntheticSection.
The .eh_frame is in theory affected by this problem, but I don't think that
there is a dangling reference problem that can happen with these sections.
Fixes remaining part of pr44824
Differential Revision: https://reviews.llvm.org/D79687
Gongyu Deng [Mon, 11 May 2020 13:21:37 +0000 (15:21 +0200)]
Complete breakpoint enable/disable/delete/modify with a list of breakpoint IDs
Summary:
1. A new common completion `CommandCompletions::Breakpoints` to provide a list of the breakpoints of the current context;
2. Apply the completion above to the commands breakpoint enable/disable/delete/modify;
3. Unit test.
Reviewers: teemperor, JDevlieghere
Reviewed By: teemperor
Tags: #lldb
Differential Revision: https://reviews.llvm.org/D79666
Alex Zinenko [Mon, 11 May 2020 13:00:48 +0000 (15:00 +0200)]
[mlir] NFC: Rename LoopOps dialect to SCF (Structured Control Flow)
This dialect contains various structured control flow operaitons, not only
loops, reflect this in the name. Drop the Ops suffix for consistency with other
dialects.
Note that this only moves the files and changes the C++ namespace from 'loop'
to 'scf'. The visible IR prefix remains the same and will be updated
separately. The conversions will also be updated separately.
Differential Revision: https://reviews.llvm.org/D79578
Kirill Bobyrev [Mon, 11 May 2020 10:27:49 +0000 (12:27 +0200)]
[clangd] Fix remote index build for macOS with Homebrew-installed gRPC and Protobuf
Reviewers: sammccall
Reviewed By: sammccall
Subscribers: mgorny, ilya-biryukov, MaskRay, jkorous, arphaman, kadircet, usaxena95, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D79387
Tyker [Mon, 11 May 2020 09:50:35 +0000 (11:50 +0200)]
[AssumeBundles] fix crashes
Summary:
this patch fixe crash/asserts found in the test-suite.
the AssumeptionCache cannot be assumed to have all assumes contrary to what i tought.
prevent generation of information for terminators, because this can create broken IR in transfromation where we insert the new terminator before removing the old one.
Reviewers: jdoerfert
Reviewed By: jdoerfert
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D79458
OCHyams [Thu, 7 May 2020 11:41:20 +0000 (12:41 +0100)]
[NFC][DwarfDebug] Add test for variables with a single location which
don't span their entire scope.
The previous commit (
6d1c40c171e) is an older version of the test.
Reviewed By: aprantl, vsk
Differential Revision: https://reviews.llvm.org/D79573
OCHyams [Thu, 7 May 2020 11:41:20 +0000 (12:41 +0100)]
[NFC][DwarfDebug] Add test for variables with a single location which
don't span their entire scope.
The previous commit (
6d1c40c171e) is an older version of the test.
Reviewed By: aprantl, vsk
Differential Revision: https://reviews.llvm.org/D79573
OCHyams [Mon, 4 May 2020 12:28:07 +0000 (13:28 +0100)]
[NFC][DwarfDebug] Add test for variables with a single location which
don't span their entire scope.
Reviewed By: aprantl, vsk
Differential Revision: https://reviews.llvm.org/D79573
Sander de Smalen [Mon, 11 May 2020 07:52:02 +0000 (08:52 +0100)]
[SveEmitter] Add builtins for svmovlb and svmovlt
These builtins are expanded in CGBuiltin to use intrinsics
for (signed/unsigned) shift left long top/bottom.
Reviewers: efriedma, SjoerdMeijer
Reviewed By: efriedma
Differential Revision: https://reviews.llvm.org/D79579
Kadir Cetinkaya [Sun, 10 May 2020 21:29:27 +0000 (23:29 +0200)]
[clangd] Make version in PublishDiagnosticsParams optional
Summary: We were serializing it no matter what, which was against the spec
Reviewers: sammccall
Subscribers: ilya-biryukov, MaskRay, jkorous, arphaman, usaxena95, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D79692
Hanhan Wang [Mon, 11 May 2020 08:29:08 +0000 (01:29 -0700)]
[mlir][StandardToLLVM] Add support for lowering FPToSIOp to LLVM.
Summary: Depends On D79374
Differential Revision: https://reviews.llvm.org/D79455
Hanhan Wang [Mon, 11 May 2020 08:26:57 +0000 (01:26 -0700)]
[mlir][StandardToSPIRV] Add support for lowering FPToSIOp to SPIR-V.
Summary: Depends On D79373
Differential Revision: https://reviews.llvm.org/D79374
Hanhan Wang [Mon, 11 May 2020 08:25:45 +0000 (01:25 -0700)]
[mlir] Add FPToSIOp to Standard dialect.
Summary:
Cast from a value interpreted as floating-point to the corresponding signed
integer value. Similar to an element-wise `static_cast` in C++, performs an
element-wise conversion operation.
Differential Revision: https://reviews.llvm.org/D79373
Djordje Todorovic [Mon, 11 May 2020 07:30:36 +0000 (09:30 +0200)]
[NFC][DwarfDebug] Avoid default capturing when using lambdas
It is bad practice to capture by default (via [&] in this case) when
using lambdas, so we should avoid that as much as possible.
This patch fixes that in the getForwardingRegsDefinedByMI
from DwarfDebug module.
Differential Revision: https://reviews.llvm.org/D79616
Haojian Wu [Mon, 11 May 2020 07:23:32 +0000 (09:23 +0200)]
Add a missing test file for recovery expressions.
The test was missed in
8222107aa9249aada81334c922a2d284042242.
Haojian Wu [Mon, 11 May 2020 07:19:42 +0000 (09:19 +0200)]
Fix -Wunused compiler warning.
Djordje Todorovic [Fri, 8 May 2020 10:04:45 +0000 (12:04 +0200)]
[NFC][DwarfDebug] Prefer explicit to auto type deduction
We should use explicit type instead of auto type deduction when
the type is so obvious. In addition, we remove ambiguity, since auto
type deduction sometimes is not that intuitive, so that could lead
us to some unwanted behavior.
This patch fixes that in the collectCallSiteParameters() from
DwarfDebug module.
Differential Revision: https://reviews.llvm.org/D79624
Haojian Wu [Fri, 8 May 2020 14:13:17 +0000 (16:13 +0200)]
[AST] Preserve the type in RecoveryExprs for broken function calls.
RecoveryExprs are modeled as dependent type to prevent bogus diagnostics
and crashes in clang.
This patch allows to preseve the type for broken calls when the
RecoveryEprs have a known type, e.g. a broken non-overloaded call, a
overloaded call when the all candidates have the same return type, so
that more features (code completion still work on "take2args(x).^") still
work.
However, adding the type is risky, which may result in more clang code being
affected leading to new crashes and hurt diagnostic, and it requires large
effort to minimize the affect (update all sites in clang to handle errorDepend
case), so we add a new flag (off by default) to allow us to develop/test
them incrementally.
This patch also has some trivial fixes to suppress diagnostics (to prevent regressions).
Tested:
all existing tests are passed (when both "-frecovery-ast", "-frecovery-ast-type" flags are flipped on);
Reviewed By: sammccall
Subscribers: rsmith, arphaman, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D79160
Muhammad Omair Javaid [Mon, 11 May 2020 05:18:50 +0000 (10:18 +0500)]
[LLDB] Fix broken testsuite due to Xfail decorators
Following test cases need minor adjustment in order to accomodate xfail
decorator:
lldb/test/Shell/SymbolFile/NativePDB/break-by-line.cpp
lldb/test/Shell/SymbolFile/NativePDB/source-list.cpp
Xun Li [Mon, 11 May 2020 04:50:32 +0000 (21:50 -0700)]
Remove an unused Module param
Summary:
In D65848 the function getFuncNameInModule was refactored to no longer use module.
This diff removes the parameter and rename the function name to avoid confusion.
Reviewers: wenlei, wmi, davidxl
Reviewed By: wenlei
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D79310
Xing GUO [Mon, 11 May 2020 05:02:39 +0000 (13:02 +0800)]
[Object] Remove unused variable after D79560. NFC.
Fangrui Song [Mon, 11 May 2020 00:16:01 +0000 (17:16 -0700)]
[gcov] Implement --stdout -t
gcov by default prints to a .gcov file. With --stdout, stdout is used.
Some summary information is omitted. There is no separator for multiple
source files.
Johannes Doerfert [Fri, 17 Apr 2020 01:32:46 +0000 (20:32 -0500)]
[Attributor] Merge the query set into AbstractAttribute
The old QuerriedAAs contained two vectors, one for required one for
optional dependences (=queries). We now use a single vector and encode
the kind directly in the pointer.
This reduces memory consumption and makes the connection between
abstract attributes and their dependences clearer.
No functional change is intended, changes in the test are due to
different order in the query map. Neither the order before nor now is in
any way special.
---
Single run of the Attributor module and then CGSCC pass (oldPM)
for SPASS/clause.c (~10k LLVM-IR loc):
Before:
```
calls to allocation functions: 543734 (329735/s)
temporary memory allocations: 105895 (64217/s)
peak heap memory consumption: 19.19MB
peak RSS (including heaptrack overhead): 102.26MB
total memory leaked: 269.10KB
```
After:
```
calls to allocation functions: 513292 (341511/s)
temporary memory allocations: 106028 (70544/s)
peak heap memory consumption: 13.35MB
peak RSS (including heaptrack overhead): 95.64MB
total memory leaked: 269.10KB
```
Difference:
```
calls to allocation functions: -30442 (208506/s)
temporary memory allocations: 133 (-910/s)
peak heap memory consumption: -5.84MB
peak RSS (including heaptrack overhead): 0B
total memory leaked: 0B
```
---
Reviewed By: uenoku
Differential Revision: https://reviews.llvm.org/D78729
QingShan Zhang [Mon, 11 May 2020 02:41:10 +0000 (02:41 +0000)]
[DAGCombine] Remove the getNegatibleCost to avoid the out of sync with getNegatedExpression
We have the getNegatibleCost/getNegatedExpression to evaluate the cost and negate the expression.
However, during negating the expression, the cost might change as we are changing the DAG,
and then, hit the assertion if we negated the wrong expression as the cost is not trustful anymore.
This patch is target to remove the getNegatibleCost to avoid the out of sync with getNegatedExpression,
and check the cost during negating the expression. It also reduce the duplicated code between
getNegatibleCost and getNegatedExpression. And fix the crash for the test in D76638
Reviewed By: RKSimon, spatel
Differential Revision: https://reviews.llvm.org/D77319
Muhammad Omair Javaid [Mon, 11 May 2020 02:01:39 +0000 (07:01 +0500)]
[LLDB] Decorate tests failing on arm-linux buildbot
Tests impacted by these decorators fail on arm-linux-gnueabihf.
These have been triaged and appropriate bugs have been filed.
Muhammad Omair Javaid [Sun, 10 May 2020 23:44:57 +0000 (04:44 +0500)]
Remove XFAIL arm-linux decorator from passing tests
Muhammad Omair Javaid [Sun, 10 May 2020 23:40:23 +0000 (04:40 +0500)]
XFAIL NativePDB tests on arm-linux
NativePDB tests fail on arm-linux buildbot. clang-cl driver crashes with
-m32 option. Bug files http://llvm.org/pr45867
Fangrui Song [Mon, 11 May 2020 01:21:18 +0000 (18:21 -0700)]
[X86] Fix combineVectorCompareAndMaskUnaryOp regression after
0e8e731449d8fbfa6ba4523de928b4812ef9f100
Lang Hames [Mon, 11 May 2020 00:33:48 +0000 (17:33 -0700)]
[examples] Fix the SpeculativeJIT and ThinLtoJIT examples for
41379f1ec46.
Nico Weber [Mon, 11 May 2020 00:15:29 +0000 (20:15 -0400)]
Remove a comment that is no longer true after
d03838343f2.
Nico Weber [Mon, 11 May 2020 00:13:05 +0000 (20:13 -0400)]
Remove a variable write that is not needed after
d03838343f2.
Johannes Doerfert [Sat, 9 May 2020 17:48:29 +0000 (12:48 -0500)]
[Attributor][FIX] Carefully handle/ignore/forget `argmemonly`
When we have an existing `argmemonly` or `inaccessiblememorargmemonly`
we used to "know" that information. However, interprocedural constant
propagation can invalidate these attributes. We now ignore and remove
these attributes for internal functions (which may be affected by IP
constant propagation), if we are deriving new attributes for the
function.
Johannes Doerfert [Sun, 10 May 2020 04:29:43 +0000 (23:29 -0500)]
[Attributor] Use "simplify to constant" in genericValueTraversal
As we replace values with constants interprocedurally, we also need to
do this "look-through" step during the generic value traversal or we
would derive properties from replaced values. While this is often not
problematic, it is when we use the "kind" of a value for reasoning,
e.g., accesses to arguments allow `argmemonly`.
Johannes Doerfert [Sun, 10 May 2020 21:53:07 +0000 (16:53 -0500)]
[Attributor] Ignore illegal accesses to `null`
When we categorize a pointer value we bailed at `null` before. If we
know `null` is not a valid memory location we can ignore it as there
won't be an access at all.
Johannes Doerfert [Sun, 10 May 2020 18:16:49 +0000 (13:16 -0500)]
[Attributor] Use existing helpers to determine IR facts
We now use getPointerDereferenceableBytes to determine `nonnull` and
`dereferenceable` facts from the IR. We also use getPointerAlignment in
AAAlign for the same reason. The latter can interfere with callbacks so
we do restrict it to non-function-pointers for now.
Johannes Doerfert [Sat, 9 May 2020 17:49:39 +0000 (12:49 -0500)]
[Attributor][NFC] Clang format Attributor*.cpp
Lang Hames [Sun, 10 May 2020 18:34:04 +0000 (11:34 -0700)]
[ORC] Share ownership of JITDylibs between ExecutionSession and
MaterializationResponsibility.
MaterializationResponsibility objects provide a connection between a
materialization process (compiler, jit linker, etc.) and the JIT state held in
the ExecutionSession and JITDylib objects. Switching to shared ownership
extends the lifetime of JITDylibs to ensure they remain accessible until all
materializers targeting them have completed. This will allow (in a follow-up
patch) JITDylibs to be removed from the ExecutionSession and placed in a
pending-destruction state while they are kept alive to communicate errors
to/from any still-runnning materialization processes. The intent is to enable
JITDylibs to be safely removed even if they have running compiles targeting
them.
Fangrui Song [Sun, 10 May 2020 19:47:45 +0000 (12:47 -0700)]
[gcov] Default coverage version to '407*' and delete CC1 option -coverage-cfg-checksum
Defaulting to -Xclang -coverage-version='407*' makes .gcno/.gcda
compatible with gcov [4.7,8)
In addition, delete clang::CodeGenOptionsBase::CoverageExtraChecksum and GCOVOptions::UseCfgChecksum.
We can infer the information from the version.
With this change, .gcda files produced by `clang --coverage a.o` linked executable can be read by gcov 4.7~7.
We don't need other -Xclang -coverage* options.
There may be a mismatching version warning, though.
(Note, GCC r173147 "split checksum into cfg checksum and line checksum"
made gcov 4.7 incompatible with previous versions.)
Ricky Zhou [Sun, 10 May 2020 20:51:01 +0000 (16:51 -0400)]
[examples] Fix llvm.memset prototype in BrainF example.
Commit
1e68724d24ba38de7c7cdb2e1939d78c8b37cc0d removed the alignment
argument from the memset intrinsic. Update the BrainF example to match.
Reviewed By: jyknight
Differential Revision: https://reviews.llvm.org/D79601
Matt Arsenault [Sun, 10 May 2020 19:27:40 +0000 (15:27 -0400)]
AMDGPU/GlobalISel: Remove -global-isel-abort=0 from tests
Richard Smith [Sun, 10 May 2020 21:13:25 +0000 (14:13 -0700)]
Fix clangd test that depended on a diagnostic removed in Clang.
Richard Smith [Sun, 10 May 2020 20:58:42 +0000 (13:58 -0700)]
Enforce the C++11 anonymous enum bitfields check even for
Objective-C++11 and under MS extensions.
This matches the MSVC behavior, and means that Objective-C behaves as a
set of extensions to the base language, rather than replacing the base
language rule with a different one.
Richard Smith [Sun, 10 May 2020 20:39:23 +0000 (13:39 -0700)]
Fix typo in enum-base disambiguation.
Martin Storsjö [Fri, 8 May 2020 08:01:09 +0000 (11:01 +0300)]
[LLD] [COFF] Fix def file exporting of symbols containing periods
This fixes an accidental breakage of exporting symbols using def
files, when the symbol name contains a period, since commit
0ca06f7950e5, mixing up a symbol name containing a period with
the case of exporting a symbol as a forward to another dll.
Differential Revision: https://reviews.llvm.org/D79619
Richard Smith [Sun, 10 May 2020 20:14:51 +0000 (13:14 -0700)]
Properly implement 'enum class' parsing.
The 'class' or 'struct' keyword is only permitted as part of either an
enum definition or a standalone opaque-enum-declaration, not as part of
an elaborated type specifier. We previously failed to diagnose this, and
generally didn't properly implement the restrictions on elaborated type
specifiers for enumeration types.
In passing, also fixed incorrect parsing for enum-bases, which we
previously parsed as a type-name, but are actually a type-specifier-seq.
This matters for cases like 'enum E : int *p;', which is valid as a
Microsoft extension.
Plus some minor parse diagnostic improvements.
Bumped the recently-added ExtWarn for 'enum E : int x;' to be
DefaultError; this is not an intentional extension, so producing an
error by default seems appropriate, but the warning flag to disable it
may still be useful for code written against old Clang. The same
treatment is given here to the diagnostic for 'enum class E x;', which
we similarly have incorrectly accepted for many years. These diagnostics
continue to be suppressed under -fms-extensions and when compiling
Objective-C code. We will need to decide separately whether Objective-C
should follow the C++ rules or the (older) MSVC rules.
Fangrui Song [Sun, 10 May 2020 17:50:20 +0000 (10:50 -0700)]
[gcov] Delete CC1 option -coverage-no-function-names-in-data
rL144865 incorrectly wrote function names for GCOV_TAG_FUNCTION
(this might be part of the reasons the header says
"We emit files in a corrupt version of GCOV's "gcda" file format").
rL176173 and rL177475 realized the problem and introduced -coverage-no-function-names-in-data
to work around the issue. (However, the description is wrong.
libgcov never writes function names, even before GCC 4.2).
In reality, the linker command line has to look like:
clang --coverage -Xclang -coverage-version='407*' -Xclang -coverage-cfg-checksum -Xclang -coverage-no-function-names-in-data
Failing to pass -coverage-no-function-names-in-data can make gcov 4.7~7
either produce wrong results (for one gcov-4.9 program, I see "No executable lines")
or segfault (gcov-7).
(gcov-8 uses an incompatible format.)
This patch deletes -coverage-no-function-names-in-data and the related
function names support from libclang_rt.profile