This allows us to use PSHUFB for v8i16/v4i32 and VPERMD/PERMPS for v4i64/v4f64 variable shuffles.
Differential Revision: https://reviews.llvm.org/D42487
llvm-svn: 323987
Summary: Now that v2i1/v4i1 are legal without VLX. And v32i1 is legalized by splitting rather than widening. And isVectorLoadExtDesirable returns false for vXi1. It appears this handling is dead because the operations simply don't exist.
Reviewers: RKSimon, zvi, guyblank, delena, spatel
Reviewed By: delena
Subscribers: llvm-commits, rengolin
Differential Revision: https://reviews.llvm.org/D42781
llvm-svn: 323983
Summary:
EmitTest sometimes creates X86ISD::AND specifically to hide the AND from DAG combine. But this prevents isel patterns that look for (cmp (and X, Y), 0) from being able to see it. So we end up with an AND and a TEST. The TEST gets removed by compare instruction optimization during the peephole pass.
This patch attempts to fix this by converting X86ISD::AND with no flag users back into ISD::AND during the DAG preprocessing just before isel.
In order to do this correctly I had to make the X86ISD::AND node created by EmitTest in this case really have a flag output. Which arguably it should have had anyway so that the number of operands would be consistent for the opcode in all cases. Then I had to modify the ReplaceAllUsesWith to understand that we might be looking at an instruction with 2 outputs. Though in this case there are no uses to replace since we just created the node, but that's what the code did before so I just made it keep working.
Reviewers: spatel, RKSimon, niravd, deadalnix
Reviewed By: RKSimon
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D42764
llvm-svn: 323982
As shown in the example in PR34994:
https://bugs.llvm.org/show_bug.cgi?id=34994
...we can return a very wrong answer (inf instead of 0.0) for square root when
using a reciprocal square root estimate instruction.
Here, I've conditionalized the filtering out of denorms based on the function
having "denormal-fp-math"="ieee" in its attributes. The other options for this
attribute are 'preserve-sign' and 'positive-zero'.
So we don't generate this extra code by default with just '-ffast-math' (because
then there's no denormal attribute string at all), but it works if you specify
'-ffast-math -fdenormal-fp-math=ieee' from clang.
As noted in the review, there may be other problems in clang that affect the
results depending on platform (Linux x86 at least), but this should allow
creating the desired codegen.
Differential Revision: https://reviews.llvm.org/D42323
llvm-svn: 323981
https://reviews.llvm.org/D42757
The method ThreadsInTeam is used to determine the number of threads to be used in a parallel region under SPMD mode (see line 127 of supporti.h in libomptarget/deviceRTLs/nvptx/src/). This patch fixes the corresponding debug print upon initialization of the kernel in SPMD mode.
llvm-svn: 323978
Summary:
In Instruction Selection UpdateChains replaces all matched Nodes'
chain references including interior token factors and deletes them.
This may allow nodes which depend on these interior nodes but are not
part of the set of matched nodes to be left with a dangling dependence.
Avoid this by doing the replacement for matched non-TokenFactor nodes.
Fixes PR36164.
Reviewers: jonpa, RKSimon, bogner
Subscribers: llvm-commits, hiraditya
Differential Revision: https://reviews.llvm.org/D42754
llvm-svn: 323977
Currently ICF information is output through stderr if the "--verbose"
flag is used. This differs to Gold for example, which uses an explicit
flag to output this to stdout. This commit adds the
"--print-icf-sections" and "--no-print-icf-sections" flags and changes
the output message format for clarity and consistency with
"--print-gc-sections". These messages are still output to stderr if
using the verbose flag. However to avoid intermingled message output to
console, this will not occur when the "--print-icf-sections" flag is
used.
Existing tests have been modified to expect the new message format from
stderr.
Patch by Owen Reynolds.
Differential Revision: https://reviews.llvm.org/D42375
Reviewers: ruiu, rafael
Reviewed by:
llvm-svn: 323976
This avoids the warnings when building with LLVM_ENABLE_LIBCXX
which automatically adds -stdlib=libc++ to CMAKE_CXX_FLAGS.
Differential Revision: https://reviews.llvm.org/D42238
llvm-svn: 323969
Commit r323512 introduced an optimisation in LowerReturn for half-precision
return values. A missing check caused a crash when the return value is "undef"
(i.e. a node that has no operands).
Differential Revision: https://reviews.llvm.org/D42743
llvm-svn: 323968
This patch includes EVA instructions in the Std2MicroMips mapping
tables, which is required for direct object emission.
Differential Revision: https://reviews.llvm.org/D41771
llvm-svn: 323958
This fixes bugzilla 33011
https://bugs.llvm.org/show_bug.cgi?id=33011
Defines bits {19-16} as zero or unpredictable as specified by the ARM ARM in
sections A8.8.116 and A8.8.117.
It fixes also the usage of PC register as destination register for MVN
register-shifted register version as specified in A8.8.117.
Differential Revision: https://reviews.llvm.org/D41905
llvm-svn: 323954
The mock server was sending acks back in response to spurious acks from
the client, but the client was not prepared to handle these. Most of the
time this would work because the only time the client was sending
unsolicited acks is after the initial connection, and there reply-ack
would get ignored in the "flush all packets from the server" loop which
came after the ack. However, this loop had only a 10ms delay, and
sometimes this was not enough to catch the reply (which meant the
connection got out of sync, and test failed).
Since this behavior not consistent with how lldb-server handles this
situation (it just ignores the ack), I fix the mock server to do the
same.
llvm-svn: 323953
Summary:
This patch enables debugging of C99 VLA types by generating more precise
LLVM Debug metadata, using the extended DISubrange 'count' field that
takes a DIVariable.
This should implement:
Bug 30553: Debug info generated for arrays is not what GDB expects (not as good as GCC's)
https://bugs.llvm.org/show_bug.cgi?id=30553
Reviewers: echristo, aprantl, dexonsmith, clayborg, pcc, kristof.beyls, dblaikie
Reviewed By: aprantl
Subscribers: jholewinski, schweitz, davide, fhahn, JDevlieghere, cfe-commits
Differential Revision: https://reviews.llvm.org/D41698
llvm-svn: 323952
This, in instcombine, allows conversions to i8/i16/i32 (very
common cases) even if the resulting type is not legal according
to the data layout. This can often open up extra combine
opportunities.
Differential Revision: https://reviews.llvm.org/D42424
llvm-svn: 323951
Summary:
thread_local has nice syntax and semantics, but requires __cxa_thread_atexit,
and some not-ancient runtime libraries don't provide it.
The clang-x86_64-linux-selfhost-modules buildbot is one example :-)
It works on windows, and the other platforms clang-tools-extra supports should
all have the relevant pthread API. So we just use that if it's available,
falling back to thread_local (so if a platform has neither, we'll fail to link).
The fallback should really be the other way, that would require cmake changes.
Reviewers: ilya-biryukov, bkramer
Subscribers: klimek, jkorous-apple, ioeric, cfe-commits
Differential Revision: https://reviews.llvm.org/D42742
llvm-svn: 323949
Summary:
Before emitting code for scaled registers, we prevent
SCEVExpander from hoisting any scaled addressing mode
by emitting all the bases first. However, these bases
are being forced to the final type, resulting in some
odd code.
For example, if the type of the base is an integer and
the final type is a pointer, we will emit an inttoptr
for the base, a ptrtoint for the scale, and then a
'reverse' GEP where the GEP pointer is actually the base
integer and the index is the pointer. It's more intuitive
to use the pointer as a pointer and the integer as index.
Patch by: Bevin Hansson
Reviewers: atrick, qcolombet, sanjoy
Reviewed By: qcolombet
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D42103
llvm-svn: 323946
This patch fixes a bug in CGRecordLowering::accumulateBitFields where it
unconditionally starts a new run and emits a storage field when it sees
a zero-sized bitfield, which causes an assertion in insertPadding to
fail when -fno-bitfield-type-align is used.
It shouldn't emit new storage if UseZeroLengthBitfieldAlignment and
UseBitFieldTypeAlignment are both false.
rdar://problem/36762205
llvm-svn: 323943
Summary:
This change expands the amount of registers stashed by the entry and
`__xray_CustomEvent` trampolines.
We've found that since the `__xray_CustomEvent` trampoline calls can show up in
situations where the scratch registers are being used, and since we don't
typically want to affect the code-gen around the disabled
`__xray_customevent(...)` intrinsic calls, that we need to save and restore the
state of even the scratch registers in the handling of these custom events.
Reviewers: pcc, pelikan, dblaikie, eizan, kpw, echristo, chandlerc
Reviewed By: echristo
Subscribers: chandlerc, echristo, hiraditya, davide, dblaikie, llvm-commits
Differential Revision: https://reviews.llvm.org/D40894
llvm-svn: 323940
Fix the infinite loop reported in PR35809. It can occur with GCC-style
EH table assembly, where the compiler relies on the assembler to
calculate the offsets in the EH table.
Also see https://sourceware.org/bugzilla/show_bug.cgi?id=4029 for the
equivalent issue in the GNU assembler.
Patch by Ryan Prichard!
llvm-svn: 323934
For very, very large global initializers which can be statically evaluated, the
code would create vectors of temporary Constants, modifying them in place,
before committing the resulting Constant aggregate to the global's initializer
value. This had effectively O(n^2) complexity in the size of the global
initializer and would cause memory and non-termination issues compiling some
workloads.
This change performs the static initializer evaluation and creation in batches,
once for each global in the evaluated IR memory. The existing code is maintained
as a last resort when the initializers are more complex than simple values in a
large aggregate. This should theoretically by NFC, no test as the example case
is massive. The existing test cases pass with this, as well as the llvm test
suite.
To give an example, consider the following C++ code adapted from the clang
regression tests:
struct S {
int n = 10;
int m = 2 * n;
S(int a) : n(a) {}
};
template<typename T>
struct U {
T *r = &q;
T q = 42;
U *p = this;
};
U<S> e;
The global static constructor for 'e' will need to initialize 'r' and 'p' of
the outer struct, while also initializing the inner 'q' structs 'n' and 'm'
members. This batch algorithm will simply use general CommitValueTo() method
to handle the complex nested S struct initialization of 'q', before
processing the outermost members in a single batch. Using CommitValueTo() to
handle member in the outer struct is inefficient when the struct/array is
very large as we end up creating and destroy constant arrays for each
initialization.
For the above case, we expect the following IR to be generated:
%struct.U = type { %struct.S*, %struct.S, %struct.U* }
%struct.S = type { i32, i32 }
@e = global %struct.U { %struct.S* gep inbounds (%struct.U, %struct.U* @e,
i64 0, i32 1),
%struct.S { i32 42, i32 84 }, %struct.U* @e }
The %struct.S { i32 42, i32 84 } inner initializer is treated as a complex
constant expression, while the other two elements of @e are "simple".
Differential Revision: https://reviews.llvm.org/D42612
llvm-svn: 323933
Don't include type signatures that are not referenced by
some relocation.
We don't include this in the -gc-sections settings since
we are always building the type section from scratch,
just like we do the table elements.
In the future we might want to unify the relocation
processing which is currently done once for gc-sections
and then again for building the sympathetic type and
table sections.
Differential Revision: https://reviews.llvm.org/D42747
llvm-svn: 323931
Summary:
r323164 made lld-link not overwrite import libraries when their
contents haven't changed. MSVC's link.exe does this only when
performing incremental linking. This change makes lld-link's import
library overwriting similarly dependent on whether or not incremental
linking is being performed. This is controlled by the /incremental or
/incremental:no options. In addition, /opt:icf, /opt:ref, and /order
turn off /incremental and issue a warning if /incremental was
specified on the command line.
Reviewers: rnk, ruiu, zturner
Reviewed By: ruiu
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D42716
llvm-svn: 323930
Changpeng Fang