Summary:
This drastically reduces lit test execution startup time on Windows. Our
previous strategy was to manually create one Process per job and manage
the worker pool ourselves. Instead, let's use the worker pool provided
by multiprocessing. multiprocessing.Pool(jobs) returns almost
immediately, and initializes the appropriate number of workers, so they
can all start executing tests immediately. This avoids the ramp-up
period that the old implementation suffers from. This appears to speed
up small test runs.
Here are some timings of the llvm-readobj tests on Windows using the
various execution strategies:
# multiprocessing.Pool:
$ for i in `seq 1 3`; do tim python ./bin/llvm-lit.py -sv ../llvm/test/tools/llvm-readobj/ --use-process-pool |& grep real: ; done
real: 0m1.156s
real: 0m1.078s
real: 0m1.094s
# multiprocessing.Process:
$ for i in `seq 1 3`; do tim python ./bin/llvm-lit.py -sv ../llvm/test/tools/llvm-readobj/ --use-processes |& grep real: ; done
real: 0m6.062s
real: 0m5.860s
real: 0m5.984s
# threading.Thread:
$ for i in `seq 1 3`; do tim python ./bin/llvm-lit.py -sv ../llvm/test/tools/llvm-readobj/ --use-threads |& grep real: ; done
real: 0m9.438s
real: 0m10.765s
real: 0m11.079s
I kept the old code to launch processes in case this change doesn't work
on all platforms that LLVM supports, but at some point I would like to
remove both the threading and old multiprocessing execution strategies.
Reviewers: modocache, rafael
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D31677
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@299560 91177308-0d34-0410-b5e6-96231b3b80d8
Commit r298799 changed code that made the XFAIL on MachineBranchProb.ll
irrelevant, but some configurations still failed. I can't reproduce it
locally, so I'm hoping that enabling this will tell me if some
configurations will really fail or if they were just too slow.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@299558 91177308-0d34-0410-b5e6-96231b3b80d8
This test case depends on the loop being vectorized without forcing the
vectorization factor. If the profitability ever changes in the future (due to
cost model improvements), the test may no longer work as intended. Instead of
checking the resulting IR, we should just check the instruction costs. The
costs will be computed regardless if vectorization is profitable.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@299545 91177308-0d34-0410-b5e6-96231b3b80d8
This is a generic combine enabled via target hook to reduce icmp logic as discussed in:
https://bugs.llvm.org/show_bug.cgi?id=32401
It's likely that other targets will want to enable this hook for scalar transforms,
and there are probably other patterns that can use bitwise logic to reduce comparisons.
Note that we are missing an IR canonicalization for these patterns, and we will probably
prefer the pair-of-compares form in IR (shorter, more likely to fold).
Differential Revision: https://reviews.llvm.org/D31483
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@299542 91177308-0d34-0410-b5e6-96231b3b80d8
When DAGCombiner visits a SIGN_EXTEND_INREG of a BUILD_VECTOR with
constant operands, a new BUILD_VECTOR node will be created transformed
constants.
Llvm-stress found a case where the new BUILD_VECTOR had constant operands
of an illegal type, because the (legal) element type is in fact not a legal
scalar type.
This patch changes this so that the new BUILD_VECTOR has the same operand
type as the old one.
Review: Eli Friedman, Nirav Dave
https://bugs.llvm.org//show_bug.cgi?id=32422
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@299540 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
Temporaries are now allocated to operands instead of predicates and this
allocation is used to correctly pair up the rendered operands with the
matched operands.
Previously, ComplexPatterns were allocated temporaries independently in the
Src Pattern and Dst Pattern, leading to mismatches. Additionally, the Dst
Pattern failed to account for the allocated index and therefore always used
temporary 0, 1, ... when it should have used base+0, base+1, ...
Thanks to Aditya Nandakumar for noticing the bug.
Depends on D30539
Reviewers: ab, t.p.northover, qcolombet, rovka, aditya_nandakumar
Reviewed By: rovka
Subscribers: igorb, dberris, kristof.beyls, llvm-commits
Differential Revision: https://reviews.llvm.org/D31054
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@299538 91177308-0d34-0410-b5e6-96231b3b80d8
A number of backends (AArch64, MIPS, ARM) have been using
MCContext::reportError to report issues such as out-of-range fixup values in
their TgtAsmBackend. This is great, but because MCContext couldn't easily be
threaded through to the adjustFixupValue helper function from its usual
callsite (applyFixup), these backends ended up adding an MCContext* argument
and adding another call to applyFixup to processFixupValue. Adding an
MCContext parameter to applyFixup makes this unnecessary, and even better -
applyFixup can take a reference to MCContext rather than a potentially null
pointer.
Differential Revision: https://reviews.llvm.org/D30264
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@299529 91177308-0d34-0410-b5e6-96231b3b80d8
This is a latent bug that's been hanging around for a while. For a loop-invariant
pointer, expandBounds would return the range {Ptr, Ptr}, but this was interpreted
as a half-open range, not a closed range. So we ended up planting incorrect
bounds checks. Even worse, they were tautological, so we ended up incorrectly
executing the optimized loop.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@299526 91177308-0d34-0410-b5e6-96231b3b80d8
Fix a bug in ARC contract pass where an iterator that pointed to a
deleted instruction was dereferenced.
It appears that tryToContractReleaseIntoStoreStrong was incorrectly
assuming that a call to objc_retain would not immediately follow a call
to objc_release.
rdar://problem/25276306
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@299507 91177308-0d34-0410-b5e6-96231b3b80d8
Before r294774, there was a problem when lowering broadcasts to use
128-bit subvectors.
When we looked through a bitcast to find the broadcast input, we'd keep
using the original type, so you'd end up with things like:
(v8f32 (broadcast
(v4f32 (extract_subvector
(v8i32 V),
...))
))
r294774 fixed it to always emit subvectors with the scalar type of the
original source.
It also introduced some asserts, to check that we use scalars with
the same size, and vectors with the same number of elements.
The scalar size equality is checked earlier when looking through bitcasts,
and is a useful assert.
However, the number of elements don't have to be identical: we're always
going to extract a 128-bit subvector, and we can have different size
inputs if we looked through a concat_vector to find a 256-bit source.
Relax the overzealous assert.
Replace it with a check of the original source vector being 256 or 512
bits. If it's 128 bits, we can't extract_subvector from it.
Fixes PR32371.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@299490 91177308-0d34-0410-b5e6-96231b3b80d8
Decouple this setting from EnableIRPA.
To support function calls on AMDGPU, it is necessary to
report the global register usage throughout the kernel's
call graph, so callees need to be handled first.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@299487 91177308-0d34-0410-b5e6-96231b3b80d8
stores with some fixes.
Summary:
This enables us to cache the clobbering access for stores, despite the
fact that we can't rewrite the use-def chains themselves.
Early testing shows that, after this change, for larger testcases, it
will be a significant net positive (memory and time) to remove the
walker caching.
Reviewers: george.burgess.iv, davide
Subscribers: Prazek, llvm-commits
Differential Revision: https://reviews.llvm.org/D31567
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@299486 91177308-0d34-0410-b5e6-96231b3b80d8
Set correct default flags and section type based on its name for .text,
.data, .bss, .init_array, .fini_array, .preinit_array, .tdata, and .tbss
and support section name suffixes for .data.*, .rodata.*, .text.*,
.bss.*, .tdata.* and .tbss.* which matches the behavior of GAS.
Fixes PR31888.
Differential Revision: https://reviews.llvm.org/D30229
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@299484 91177308-0d34-0410-b5e6-96231b3b80d8
This improves upon r246462: that prevented FMOVs from being emitted
for the cross-class INSERT_SUBREGs by disabling the formation of
INSERT_SUBREGs of LOAD. But the ld1.s that we started selecting
caused us to introduce partial dependencies on the vector register.
Avoid that by using SCALAR_TO_VECTOR: it's a first-class citizen that
is folded away by many patterns, including the scalar LDRS that we
want in this case.
Credit goes to Adam for finding the issue!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@299482 91177308-0d34-0410-b5e6-96231b3b80d8
GAS uses "m" as a compatibility alias for "M" (SHF_MERGE).
"o" is free, except on ia64, where it already means SHF_LINK_ORDER.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@299479 91177308-0d34-0410-b5e6-96231b3b80d8
If an instruction has a true dependency, it makes sense for to use that
register for any undef read operands in the same instruction (we'll have
to wait for that register to become available anyway). This logic
was already implemented. However, the code would then still try to
revisit that instruction and break the dependency (and always fail,
since by definition a true dependency has to be live before the
instruction). Avoid revisiting such instructions as a performance
optimization. No functional change.
Differential Revision: https://reviews.llvm.org/D30173
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@299467 91177308-0d34-0410-b5e6-96231b3b80d8
This shares detection logic with ARM(32), since AArch64 capable CPUs may
also run in 32-bit system mode.
We observe weird /proc/cpuinfo output for MSM8992 and MSM8994, where
they report all CPU cores as one single model, depending on which CPU
core the kernel is running on. As a workaround, we hardcode the known
CPU part name for these SoCs.
For big.LITTLE systems, this patch would only return the part name of
the first core (usually the little core). Proper support will be added
in a follow-up change.
Differential Revision: D31675
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@299458 91177308-0d34-0410-b5e6-96231b3b80d8
When the ProcessAllSections flag (introduced in r204398) is set RuntimeDyld is
supposed to make a call to the client's memory manager for every section in each
object that is loaded. Due to some missing checks, this was not happening in all
cases. This patch adds the missing cases, and fixes the Orc unit test that
verifies correct behavior for ProcessAllSections (The unit test had been
silently bailing out due to an ordering issue: a change in the test order meant
that this unit-test was running before the native target was registered. This
issue has also been fixed in this patch).
This fixes <rdar://problem/22789965>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@299449 91177308-0d34-0410-b5e6-96231b3b80d8
