Bob Wilson [Tue, 17 Jan 2017 19:18:57 +0000 (19:18 +0000)]
Revert r291640 change to fold X86 comparison with atomic_load_add.
Even with the fix from r291630, this still causes problems. I get
widespread assertion failures in the Swift runtime's WeakRefCount::increment()
function. I sent a reduced testcase in reply to the commit.
Chandler Carruth [Tue, 17 Jan 2017 19:18:12 +0000 (19:18 +0000)]
[PM] Teach the LoopPassManager to automatically canonicalize loops by
runnig LCSSA over them prior to running the loop pipeline.
This also teaches the loop PM to verify that LCSSA form is preserved
throughout the pipeline's run across the loop nest.
Most of the test updates just leverage this new functionality. One has to be
relaxed with the new PM as IVUsers is less powerful when it sees LCSSA input.
Kuba Mracek [Tue, 17 Jan 2017 17:15:02 +0000 (17:15 +0000)]
[lit] Limit parallelism of sanitizer tests on Darwin [llvm part]
Running lit tests and unit tests of ASan and TSan on macOS has very bad performance when running with a high number of threads. This is caused by xnu (the macOS kernel), which currently doesn't handle mapping and unmapping of sanitizer shadow regions (reserved VM which are several terabytes large) very well. The situation is so bad that increasing the number of threads actually makes the total testing time larger. The macOS buildbots are affected by this. Note that we can't easily limit the number of sanitizer testing threads without affecting the rest of the tests.
This patch adds a special "group" into lit, and limits the number of concurrently running tests in this group. This helps solve the contention problem, while still allowing other tests to run in full, that means running lit with -j8 will still with 8 threads, and parallelism is only limited in sanitizer tests.
George Rimar [Tue, 17 Jan 2017 15:45:07 +0000 (15:45 +0000)]
Recommit r292214 "[Support/Compression] - Change zlib API to return Error instead of custom status"
No any changes, will follow up with D28807 commit containing APLi change for clang
to fix build issues happened.
Original commit message:
[Support/Compression] - Change zlib API to return Error instead of custom status.
Previously API returned custom enum values.
Patch changes it to return Error with string description.
That should help users to report errors in universal way.
[llvm-objdump tests] Copy the inputs of tests closer to tests.
Summary:
Tests under tools/llvm-objdump should not use inputs from Object. Copied the
required inputs and aligned the new tests to be more consistent with the existing
tests in this respect.
George Rimar [Tue, 17 Jan 2017 13:27:58 +0000 (13:27 +0000)]
Revert r292214 "[Support/Compression] - Change zlib API to return Error instead of custom status."
It broked clang:
http://lab.llvm.org:8080/green//job/clang-stage1-cmake-RA-incremental_build/34218/consoleFull#46141505449ba4694-19c4-4d7e-bec5-911270d8a58c
George Rimar [Tue, 17 Jan 2017 13:20:17 +0000 (13:20 +0000)]
[Support/Compression] - Change zlib API to return Error instead of custom status.
Previously API returned custom enum values.
Patch changes it to return Error with string description.
That should help users to report errors in universal way.
Serge Rogatch [Tue, 17 Jan 2017 11:52:10 +0000 (11:52 +0000)]
[XRay][Arm] Repair XRay table emission on Arm32 and add tests to identify such problem earlier
Summary:
Emission of XRay table was occasionally disabled for Arm32, but this bug was not then detected because earlier (also by mistake) testing of XRay was occasionally disabled on 32-bit Arm targets. This patch should fix that problem and detect such problems in the future.
This patch is one of a series, see also
- https://reviews.llvm.org/D28623
Emit error when BPF backend sees a call to a global function or to an external symbol.
The kernel verifier only allows calls to predefined helpers from bpf.h
which are defined in 'enum bpf_func_id'. Such calls in assembler must
look like 'call [1-9]+' where number matches bpf_func_id.
Shoaib Meenai [Tue, 17 Jan 2017 07:10:55 +0000 (07:10 +0000)]
[utils] Add libc++ and libc++abi config to llvm-lit
This allows us to use bin/llvm-lit to run individual libc++ and
libc++abi tests without having to explicitly specify the site config
paths, similar to other projects.
Craig Topper [Tue, 17 Jan 2017 06:49:59 +0000 (06:49 +0000)]
[AVX-512] Add support for taking a bitcast between a SUBV_BROADCAST and VSELECT and moving it to the input of the SUBV_BROADCAST if it will help with using a masked operation.
Ahmed Bougacha [Tue, 17 Jan 2017 03:10:06 +0000 (03:10 +0000)]
[TLI] Robustize SDAG proto checking by merging it into TLI.
SelectionDAGBuilder recognizes libfuncs using some homegrown
parameter type-checking.
Use TLI instead, removing another heap of redundant code.
This isn't strictly NFC, as the SDAG code was too lax.
Concretely, this means changes are required to two tests:
- calling a non-variadic function via a variadic prototype isn't OK;
it just happens to work on x86_64 (but not on, e.g., aarch64).
- mempcpy has a size_t parameter; the SDAG code accepts any integer
type, which meant using i32 on x86_64 worked.
I don't think it's worth supporting either of these (IMO) broken
testcases. Instead, fix them to be more correct.
Ahmed Bougacha [Tue, 17 Jan 2017 03:10:02 +0000 (03:10 +0000)]
[TLI] Add prototype checking for all remaining LibFuncs.
This is another step towards unifying all LibFunc prototype checks.
This work started in r267758 (D19469); add the remaining checks.
Also add a unittest that checks each libfunc declared with a known-valid
and known-invalid prototype. New libfuncs added in the future are
required to have prototype checking in place; the known-valid test will
fail otherwise.
David Blaikie [Mon, 16 Jan 2017 20:36:26 +0000 (20:36 +0000)]
[XRay] Implement the `llvm-xray graph` subcommand
Here we define the `graph` subcommand which generates a graph from the function
call information and uses it to present the call information graphically with
additional annotations.
Tony Jiang [Mon, 16 Jan 2017 20:12:26 +0000 (20:12 +0000)]
[PowerPC] Expand ISEL instruction into if-then-else sequence.
Generally, the ISEL is expanded into if-then-else sequence, in some
cases (like when the destination register is the same with the true
or false value register), it may just be expanded into just the if
or else sequence.
Sanjay Patel [Mon, 16 Jan 2017 20:05:26 +0000 (20:05 +0000)]
[InstCombine] fix names in canEvaluateShiftedShift(); NFC
It's not clear what 'First' and 'Second' mean, so use 'Inner' and 'Outer'
to match foldShiftedShift() and add comments with formulas, so it's easier
to see what's going on.
Falkor only partially implements the ARMv8.1a extensions, so this patch
refactors the support for the SQRDML[A|S]H instruction into a separate
feature.
Hal Finkel [Mon, 16 Jan 2017 15:22:01 +0000 (15:22 +0000)]
Fix use-after-free bug in AffectedValueCallbackVH::allUsesReplacedWith
When transferring affected values in the cache from an old value, identified by
the value of the current callback, to the specified new value we might need to
insert a new entry into the DenseMap which constitutes the cache. Doing so
might delete the current callback object. Move the copying logic into a new
function, a member of the assumption cache itself, so that we don't run into UB
should the callback handle itself be removed mid-copy.
Daniel Sanders [Mon, 16 Jan 2017 15:20:43 +0000 (15:20 +0000)]
[globalisel] Tablegen-erate current Register Bank Information
Summary:
Adds a RegisterBank tablegen class that can be used to declare the register
banks and an associated tablegen pass to generate the necessary code.
Tony Jiang [Mon, 16 Jan 2017 14:43:12 +0000 (14:43 +0000)]
[PowerPC] Expand ISEL instruction into if-then-else sequence.
Generally, the ISEL is expanded into if-then-else sequence, in some
cases (like when the destination register is the same with the true
or false value register), it may just be expanded into just the if
or else sequence.
Craig Topper [Mon, 16 Jan 2017 00:55:58 +0000 (00:55 +0000)]
[AVX-512] Correct memory operand size for VPGATHERQPS and VPGATHERQD
with ZMM index. Similar for SCATTER and the prefetch gather and scatter
instructions.
Justin Lebar [Sun, 15 Jan 2017 16:54:35 +0000 (16:54 +0000)]
[NVPTX] Let there be One True Way to set NVVMReflect params.
Summary:
Previously there were three ways to inform the NVVMReflect pass whether
you wanted to flush denormals to zero:
* An LLVM command-line option
* Parameters to the NVVMReflect constructor
* Metadata on the module itself.
This change removes the first two, leaving only the third.
The motivation for this change, aside from simplifying things, is that
we want LLVM to be aware of whether it's operating in FTZ mode, so other
passes can use this information. Ideally we'd have a target-generic
piece of metadata on the module. This change moves us in that
direction.
projects / llvm / log