Summary:
This is a temporary hack until we get around to remapping the vreg
numbers to local numbers. Dead vregs cause bad numbering and make
consumers sad.
We could also just look at debug info an use named locals instead, but
vregs have to work properly anyways so there!
Reviewers: binji, sunfish
Subscribers: jfb, llvm-commits, dschuff
Differential Revision: http://reviews.llvm.org/D13839
llvm-svn: 250594
Summary:
Some shared code for handling eh.exceptionpointer and eh.exceptioncode
needs to not share the part that truncates to 32 bits, which is intended
just for exception codes.
Reviewers: rnk
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D13747
llvm-svn: 250588
Our previous value of "16 + 8 + MaxCallFrameSize" for ParentFrameOffset
is incorrect when CSRs are involved. We were supposed to have a test
case to catch this, but it wasn't very rigorous.
The main effect here is that calling _CxxThrowException inside a
catchpad doesn't immediately crash on MOVAPS when you have an odd number
of CSRs.
llvm-svn: 250583
The motivation for this patch starts with PR20134:
https://llvm.org/bugs/show_bug.cgi?id=20134
void foo(int *a, int i) {
a[i] = a[i+1] + a[i+2];
}
It seems better to produce this (14 bytes):
movslq %esi, %rsi
movl 0x4(%rdi,%rsi,4), %eax
addl 0x8(%rdi,%rsi,4), %eax
movl %eax, (%rdi,%rsi,4)
Rather than this (22 bytes):
leal 0x1(%rsi), %eax
cltq
leal 0x2(%rsi), %ecx
movslq %ecx, %rcx
movl (%rdi,%rcx,4), %ecx
addl (%rdi,%rax,4), %ecx
movslq %esi, %rax
movl %ecx, (%rdi,%rax,4)
The most basic problem (the first test case in the patch combines constants) should also be fixed in InstCombine,
but it gets more complicated after that because we need to consider architecture and micro-architecture. For
example, AArch64 may not see any benefit from the more general transform because the ISA solves the sexting in
hardware. Some x86 chips may not want to replace 2 ADD insts with 1 LEA, and there's an attribute for that:
FeatureSlowLEA. But I suspect that doesn't go far enough or maybe it's not getting used when it should; I'm
also not sure if FeatureSlowLEA should also mean "slow complex addressing mode".
I see no perf differences on test-suite with this change running on AMD Jaguar, but I see small code size
improvements when building clang and the LLVM tools with the patched compiler.
A more general solution to the sext(add nsw(x, C)) problem that works for multiple targets is available
in CodeGenPrepare, but it may take quite a bit more work to get that to fire on all of the test cases that
this patch takes care of.
Differential Revision: http://reviews.llvm.org/D13757
llvm-svn: 250560
Summary:
We now use the block for the catchpad itself, rather than its normal
successor, as the funclet entry.
Putting the normal successor in the map leads downstream funclet
membership computations to erroneous results.
Reviewers: majnemer, rnk
Subscribers: rnk, llvm-commits
Differential Revision: http://reviews.llvm.org/D13798
llvm-svn: 250552
Summary: The syntax has changed a bit recently.
Reviewers: binji
Subscribers: llvm-commits, jfb, sunfish, dschuff
Differential Revision: http://reviews.llvm.org/D13821
llvm-svn: 250535
Summary:
When a cleanup's cleanupendpad or cleanupret targets a catchendpad, stop
trying to propagate the cleanup's parent's color to the catchendpad, since
what's needed is the cleanup's grandparent's color and the catchendpad
will get that color from the catchpad linkage already. We already had
this exclusion for invokes, but were missing it for
cleanupendpad/cleanupret.
Also add a missing line that tags cleanupendpads' states in the
EHPadStateMap, without with lowering invokes that target cleanupendpads
which unwind to other handlers (and so don't have the -1 state) will fail.
This fixes the reduced IR repro in PR25163.
Reviewers: majnemer, andrew.w.kaylor, rnk
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D13797
llvm-svn: 250534
Summary:
Follow the same syntax as for the spec repo. Both have evolved slightly
independently and need to converge again.
This, along with wasmate changes, allows me to do the following:
echo "int add(int a, int b) { return a + b; }" > add.c
./out/bin/clang -O2 -S --target=wasm32-unknown-unknown add.c -o add.wack
./experimental/prototype-wasmate/wasmate.py add.wack > add.wast
./sexpr-wasm-prototype/out/sexpr-wasm add.wast -o add.wasm
./sexpr-wasm-prototype/third_party/v8-native-prototype/v8/v8/out/Release/d8 -e "print(WASM.instantiateModule(readbuffer('add.wasm'), {print:print}).add(42, 1337));"
As you'd expect, the d8 shell prints out the right value.
Reviewers: sunfish
Subscribers: jfb, llvm-commits, dschuff
Differential Revision: http://reviews.llvm.org/D13712
llvm-svn: 250480
D4796 taught LLVM to fold some atomic integer operations into a single
instruction. The pattern was unaware that the instructions clobbered
flags. I fixed some of this issue in D13680 but had missed INC/DEC.
This patch adds the missing EFLAGS definition.
llvm-svn: 250438
Summary:
x86 codegen is clever about generating good code for relaxed
floating-point operations, but it was being silly when globals and
immediates were involved, forgetting where the global was and
loading/storing from/to the wrong place. The same applied to hard-coded
address immediates.
Don't let it forget about the displacement.
This fixes https://llvm.org/bugs/show_bug.cgi?id=25171
A very similar bug when doing floating-points atomics to the stack is
also fixed by this patch.
This fixes https://llvm.org/bugs/show_bug.cgi?id=25144
Reviewers: pete
Subscribers: llvm-commits, majnemer, rsmith
Differential Revision: http://reviews.llvm.org/D13749
llvm-svn: 250429
Summary:
The -mcpu=mips16 option caused the Integrated Assembler to crash because
it couldn't figure out the architecture revision number to write to the
.MIPS.abiflags section. This CPU definition has been removed because, like
microMIPS, MIPS16 is an ASE to a base architecture.
Reviewers: vkalintiris
Subscribers: rkotler, llvm-commits, dsanders
Differential Revision: http://reviews.llvm.org/D13656
llvm-svn: 250407
AVX-512 bit shuffle fails on 32 bit since we create a vector of 64-bit constants.
I split 8x64-bit const vector to 16x32 on 32-bit mode.
Differential Revision: http://reviews.llvm.org/D13644
llvm-svn: 250390
Recommit r250342: move coal-sections-powerpc.s to subdirectory for powerpc.
Some background on why we don't have to use *coal* sections anymore:
Long ago when C++ was new and "weak" had not been standardized, an attempt was
made in cctools to support C++ inlines that can be coalesced by putting them
into their own section (TEXT/textcoal_nt instead of TEXT/text).
The current macho linker supports the weak-def bit on any symbol to allow it to
be coalesced, but the compiler still puts weak-def functions/data into alternate
section names, which the linker must map back to the base section name.
This patch makes changes that are necessary to prevent the compiler from using
the "coal" sections and have it use the non-coal sections instead when the
target architecture is not powerpc:
TEXT/textcoal_nt instead use TEXT/text
TEXT/const_coal instead use TEXT/const
DATA/datacoal_nt instead use DATA/data
If the target is powerpc, we continue to use the *coal* sections since anyone
targeting powerpc is probably using an old linker that doesn't have support for
the weak-def bits.
Also, have the assembler issue a warning if it encounters a *coal* section in
the assembly file and inform the users to use the non-coal sections instead.
rdar://problem/14265330
Differential Revision: http://reviews.llvm.org/D13188
llvm-svn: 250370
Recommit r250342: add -arch=ppc32 to the RUN lines of powerpc tests.
Some background on why we don't have to use *coal* sections anymore:
Long ago when C++ was new and "weak" had not been standardized, an attempt was
made in cctools to support C++ inlines that can be coalesced by putting them
into their own section (TEXT/textcoal_nt instead of TEXT/text).
The current macho linker supports the weak-def bit on any symbol to allow it to
be coalesced, but the compiler still puts weak-def functions/data into alternate
section names, which the linker must map back to the base section name.
This patch makes changes that are necessary to prevent the compiler from using
the "coal" sections and have it use the non-coal sections instead when the
target architecture is not powerpc:
TEXT/textcoal_nt instead use TEXT/text
TEXT/const_coal instead use TEXT/const
DATA/datacoal_nt instead use DATA/data
If the target is powerpc, we continue to use the *coal* sections since anyone
targeting powerpc is probably using an old linker that doesn't have support for
the weak-def bits.
Also, have the assembler issue a warning if it encounters a *coal* section in
the assembly file and inform the users to use the non-coal sections instead.
rdar://problem/14265330
Differential Revision: http://reviews.llvm.org/D13188
llvm-svn: 250349
Some background on why we don't have to use *coal* sections anymore:
Long ago when C++ was new and "weak" had not been standardized, an attempt was
made in cctools to support C++ inlines that can be coalesced by putting them
into their own section (TEXT/textcoal_nt instead of TEXT/text).
The current macho linker supports the weak-def bit on any symbol to allow it to
be coalesced, but the compiler still puts weak-def functions/data into alternate
section names, which the linker must map back to the base section name.
This patch makes changes that are necessary to prevent the compiler from using
the "coal" sections and have it use the non-coal sections instead when the
target architecture is not powerpc:
TEXT/textcoal_nt instead use TEXT/text
TEXT/const_coal instead use TEXT/const
DATA/datacoal_nt instead use DATA/data
If the target is powerpc, we continue to use the *coal* sections since anyone
targeting powerpc is probably using an old linker that doesn't have support for
the weak-def bits.
Also, have the assembler issue a warning if it encounters a *coal* section in
the assembly file and inform the users to use the non-coal sections instead.
rdar://problem/14265330
Differential Revision: http://reviews.llvm.org/D13188
llvm-svn: 250342
PR25157 identifies a bug where a load plus a vector shuffle is
incorrectly converted into an LXVDSX instruction. That optimization
is only valid if the load is of a doubleword, and in the noted case,
it was not. This corrects that problem.
Joint patch with Eric Schweitz, who provided the bugpoint-reduced test
case.
llvm-svn: 250324
This patch teaches x86 fast-isel how to select nontemporal stores.
On x86, we can use MOVNTI for nontemporal stores of doublewords/quadwords.
Instructions (V)MOVNTPS/PD/DQ can be used for SSE2/AVX aligned nontemporal
vector stores.
Before this patch, fast-isel always selected 'movd/movq' instead of 'movnti'
for doubleword/quadword nontemporal stores. In the case of nontemporal stores
of aligned vectors, fast-isel always selected movaps/movapd/movdqa instead of
movntps/movntpd/movntdq.
With this patch, if we use SSE2/AVX intrinsics for nontemporal stores we now
always get the expected (V)MOVNT instructions.
The lack of fast-isel support for nontemporal stores was spotted when analyzing
the -O0 codegen for nontemporal stores.
Differential Revision: http://reviews.llvm.org/D13698
llvm-svn: 250285
Summary:
Emit the handler and clause locations immediately after the standard
xdata.
Clauses are emitted in the same order and format used to communiate them
to the CLR Execution Engine.
Add a lit test to verify correct table generation on a small but
interesting example function.
Reviewers: majnemer, andrew.w.kaylor, rnk
Subscribers: pgavlin, AndyAyers, llvm-commits
Differential Revision: http://reviews.llvm.org/D13451
llvm-svn: 250219
The comment says this was stopped because it was unlikely to be
profitable. This is not true if you want to combine vector loads
with multiple components.
For a simple case that looks like
t0 = load t0 ...
t1 = load t0 ...
t2 = load t0 ...
t3 = load t0 ...
t4 = store t0:1, t0:1
t5 = store t4, t1:0
t6 = store t5, t2:0
t7 = store t6, t3:0
We want to get all of these stores onto a chain
that is a TokenFactor of these N loads. This mostly
solves the AMDGPU merge-stores.ll regressions
with -combiner-alias-analysis for merging vector
stores of vector loads.
llvm-svn: 250138
Summary:
D4796 taught LLVM to fold some atomic integer operations into a single
instruction. The pattern was unaware that the instructions clobbered
flags.
This patch adds the missing EFLAGS definition.
Floating point operations don't set flags, the subsequent fadd
optimization is therefore correct. The same applies for surrounding
load/store optimizations.
Reviewers: rsmith, rtrieu
Subscribers: llvm-commits, reames, morisset
Differential Revision: http://reviews.llvm.org/D13680
llvm-svn: 250135
This basic combine was surprisingly missing.
AMDGPU legalizes many operations in terms of 32-bit vector components,
so not doing this results in many extra copies and subregister extracts
that need to be cleaned up later.
InstCombine already does this for the hasOneUse case. The target hook
is to fix a handful of tests which break (e.g. ARM/vmov.ll) which turn
from a vector materialize repeated immediate instruction to a constant
vector load with more scalar copies from it.
llvm-svn: 250129
When lowering invoke statement, all unwind destinations are directly added as successors of call site block, and the weight of those new edges are not assigned properly. Actually, default weight 16 are used for those edges. This patch calculates the proper edge weights for those edges when collecting all unwind destinations.
Differential revision: http://reviews.llvm.org/D13354
llvm-svn: 250119
We made them SP relative back in March (r233137) because that's the
value the runtime passes to EH functions. With the new cleanuppad IR,
funclets adjust their frame argument from SP to FP, so our offsets
should now be FP-relative.
llvm-svn: 250088
Function LowerVSETCC (in X86ISelLowering.cpp) worked under the wrong
assumption that for non-AVX512 targets, the source type and destination type
of a type-legalized setcc node were always the same type.
This assumption was unfortunately incorrect; the type legalizer is not always
able to promote the return type of a setcc to the same type as the first
operand of a setcc.
In the case of a vsetcc node, the legalizer firstly checks if the first input
operand has a legal type. If so, then it promotes the return type of the vsetcc
to that same type. Otherwise, the return type is promoted to the 'next legal
type', which, for vectors of MVT::i1 is always a 128-bit integer vector type.
Example (-mattr=+avx):
%0 = trunc <8 x i32> %a to <8 x i23>
%1 = icmp eq <8 x i23> %0, zeroinitializer
The initial selection dag for the code above is:
v8i1 = setcc t5, t7, seteq:ch
t5: v8i23 = truncate t2
t2: v8i32,ch = CopyFromReg t0, Register:v8i32 %vreg1
t7: v8i32 = build_vector of all zeroes.
The type legalizer would firstly check if 't5' has a legal type. If so, then it
would reuse that same type to promote the return type of the setcc node.
Unfortunately 't5' is of illegal type v8i23, and therefore it cannot be used to
promote the return type of the setcc node. Consequently, the setcc return type
is promoted to v8i16. Later on, 't5' is promoted to v8i32 thus leading to the
following dag node:
v8i16 = setcc t32, t25, seteq:ch
where t32 and t25 are now values of type v8i32.
Before this patch, function LowerVSETCC would have wrongly expanded the setcc
to a single X86ISD::PCMPEQ. Surprisingly, ISel was still able to match an
instruction. In our case, ISel would have matched a VPCMPEQWrr:
t37: v8i16 = X86ISD::VPCMPEQWrr t36, t25
However, t36 and t25 are both VR256, while the result type is instead of class
VR128. This inconsistency ended up causing the insertion of COPY instructions
like this:
%vreg7<def> = COPY %vreg3; VR128:%vreg7 VR256:%vreg3
Which is an invalid full copy (not a sub register copy).
Eventually, the backend would have hit an UNREACHABLE "Cannot emit physreg copy
instruction" in the attempt to expand the malformed pseudo COPY instructions.
This patch fixes the problem adding the missing logic in LowerVSETCC to handle
the corner case of a setcc with 128-bit return type and 256-bit operand type.
This problem was originally reported by Dimitry as PR25080. It has been latent
for a very long time. I have added the minimal reproducible from that bugzilla
as test setcc-lowering.ll.
Differential Revision: http://reviews.llvm.org/D13660
llvm-svn: 250085
Summary:
This removes unnecessary instructions when extracting from an undefined register
and also fixes a crash for O32 when passing undef to a double argument in
held in integer registers.
Reviewers: vkalintiris
Subscribers: llvm-commits, zoran.jovanovic, petarj
Differential Revision: http://reviews.llvm.org/D13467
llvm-svn: 250039
This patch fixes a problem in function 'combineX86ShuffleChain' that causes a
chain of shuffles to be wrongly folded away when the combined shuffle mask has
only one element.
We may end up with a combined shuffle mask of one element as a result of
multiple calls to function 'canWidenShuffleElements()'.
Function canWidenShuffleElements attempts to simplify a shuffle mask by widening
the size of the elements being shuffled.
For every pair of shuffle indices, function canWidenShuffleElements checks if
indices refer to adjacent elements. If all pairs refer to "adjacent" elements
then the shuffle mask is safely widened. As a consequence of widening, we end up
with a new shuffle mask which is half the size of the original shuffle mask.
The byte shuffle (pshufb) from test pr24562.ll has a mask of all SM_SentinelZero
indices. Function canWidenShuffleElements would combine each pair of
SM_SentinelZero indices into a single SM_SentinelZero index. So, in a
logarithmic number of steps (4 in this case), the pshufb mask is simplified to
a mask with only one index which is equal to SM_SentinelZero.
Before this patch, function combineX86ShuffleChain wrongly assumed that a mask
of size one is always equivalent to an identity mask. So, the entire shuffle
chain was just folded away as the combined shuffle mask was treated as a no-op
mask.
With this patch we know check if the only element of a combined shuffle mask is
SM_SentinelZero. In case, we propagate a zero vector.
Differential Revision: http://reviews.llvm.org/D13364
llvm-svn: 250027
The XOP vector integer comparisons can deal with all signed/unsigned comparison cases directly and can be easily commuted as well (D7646).
llvm-svn: 249976
expandPostRAPseudo():
STX -> 2 * STD: The first STD should not have the kill flag set for the address.
SystemZElimCompare:
BRC -> BRCT conversion: Don't forget to remove the CC<use,kill> operand.
Needed to make SystemZ/asm-17.ll pass with -verify-machineinstrs, which
now runs with this flag.
Reviewed by Ulrich Weigand.
llvm-svn: 249945
The new implementation works at least as well as the old implementation
did.
Also delete the associated preparation tests. They don't exercise
interesting corner cases of the new implementation. All the codegen
tests of the EH tables have already been ported.
llvm-svn: 249918
Also Fix a buglet where SEH tables had ranges that spanned funclets.
The remaining tests using the old landingpad IR are preparation tests,
and will be deleted along with the old preparation.
llvm-svn: 249917
x64 catchpads use rax to inform the unwinder where control should go
next. However, we must initialize rax before the epilogue sequence so
as to not perturb the unwinder.
llvm-svn: 249910
This occurred due to introducing the invalid i64 type after type
legalization had already finished, in an attempt to workaround bitcast
f64 -> v2i32 not doing constant folding.
The *right* thing is to actually fix bitcast, but that has other
complications. So, for now, just get rid of the broken workaround, and
check in a test-case showing that it doesn't crash, with TODOs for
emitting proper code.
llvm-svn: 249908
This wasn't very observable in execution tests, because usually there is
an invoke in the catchpad that unwinds the the catchendpad but never
actually throws.
llvm-svn: 249898
When running combine on an extract_vector_elt, it wants to look through
a bitcast to check if the argument to the bitcast was itself an
extract_vector_elt with particular operands.
However, it called getOperand() on the argument to the bitcast *before*
checking that the opcode was EXTRACT_VECTOR_ELT, assert-failing if there
were zero operands for the actual opcode.
Fix, and add trivial test.
llvm-svn: 249891
This patch corresponds to review:
http://reviews.llvm.org/D12032
This patch builds onto the patch that provided scalar to vector conversions
without stack operations (D11471).
Included in this patch:
- Vector element extraction for all vector types with constant element number
- Vector element extraction for v16i8 and v8i16 with variable element number
- Removal of some unnecessary COPY_TO_REGCLASS operations that ended up
unnecessarily moving things around between registers
Not included in this patch (will be in upcoming patch):
- Vector element extraction for v4i32, v4f32, v2i64 and v2f64 with
variable element number
- Vector element insertion for variable/constant element number
Testing is provided for all extractions. The extractions that are not
implemented yet are just placeholders.
llvm-svn: 249822
Accept r11 when targeting Windows on ARM rather than just low registers.
Because we are in a thumb-2 only mode, this may be slightly more expensive in
code size, but results in better code for the environment since it spills the
frame register, which is generally desired for fast stack walking as per the
ABI.
llvm-svn: 249804
Summary:
- Recurse from cleanupendpads to their cleanuppads, to make sure the
cleanuppad is visited if it has a cleanupendpad but no cleanupret.
- Check for and avoid double-processing cleanuppads, to allow for them to
have multiple cleanuprets (plus cleanupendpads).
- Update Cxx state numbering to visit toplevel cleanupendpads and to
recurse from cleanupendpads to their preds, to ensure we number any
funclets in inlined cleanups. SEH state numbering already did this.
Reviewers: rnk
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D13374
llvm-svn: 249792
its own variable.
This is needed so that we can explicitly turn off MMX without turning
off SSE and also so that we can diagnose feature set incompatibilities
that involve MMX without SSE.
Rationale:
// sse3
__m128d test_mm_addsub_pd(__m128d A, __m128d B) {
return _mm_addsub_pd(A, B);
}
// mmx
void shift(__m64 a, __m64 b, int c) {
_mm_slli_pi16(a, c);
_mm_slli_pi32(a, c);
_mm_slli_si64(a, c);
_mm_srli_pi16(a, c);
_mm_srli_pi32(a, c);
_mm_srli_si64(a, c);
_mm_srai_pi16(a, c);
_mm_srai_pi32(a, c);
}
clang -msse3 -mno-mmx file.c -c
For this code we should be able to explicitly turn off MMX
without affecting the compilation of the SSE3 function and then
diagnose and error on compiling the MMX function.
This matches the existing gcc behavior and follows the spirit of
the SSE/MMX separation in llvm where we can (and do) turn off
MMX code generation except in the presence of intrinsics.
Updated a couple of tests, but primarily tested with a couple of tests
for turning on only mmx and only sse.
This is paired with a patch to clang to take advantage of this behavior.
llvm-svn: 249731
o Before this patch, BPF backend will expand UNDEF node
to i64 constant 0.
o For second pass of dag combiner, legalizer will run through
each to-be-processed dag node.
o If any new SDNode is generated and has an undef operand,
dag combiner will put undef node, newly-generated constant-0 node,
and any node which uses these nodes in the working list.
o During this process, it is possible undef operand is
generated again, and this will form an infinite loop
for dag combiner pass2.
o This patch allows UNDEF to be a legal type.
Signed-off-by: Yonghong Song <yhs@plumgrid.com>
Signed-off-by: Alexei Starovoitov <ast@plumgrid.com>
llvm-svn: 249718
This fixes yet another scenario where tryBuildVectorShuffle would
attempt to create a BUILD_VECTOR node with an invalid combination
of types. This can happen if the incoming BUILD_VECTOR has elements
of a type different from the vector element type, which is allowed
in certain cases as long as they are all the same type.
When one of these elements is used in the residual vector, and
UNDEF elements are added to fill up the residual vector, those
UNDEFs then have to use the type of the original element, not
the vector element type, or else the resulting BUILD_VECTOR
will have an invalid type combination.
llvm-svn: 249706
We emit 1 compact unwind encoding per function, and this can’t represent
the varying stack pointer that will be generated by X86CallFrameOptimization.
Disable the optimization on Darwin.
(It might be possible to split the function into multiple ranges
and emit 1 compact unwind info per range. The compact unwind emission
code isn’t ready for that and this kind of info certainly isn’t
tested/used anywhere. It might be worth exploring this path if we want
to get the space savings at some point though)
llvm-svn: 249694
This instructions doesn't have intrincis.
Added tests for lowering and encoding.
Differential Revision: http://reviews.llvm.org/D12317
llvm-svn: 249688
This fixes two separate bugs:
1) The mask for the high lane was not set correctly. That fixes PR24532.
2) The transformation should bail out if it believes it involves more than
2 lanes, as it does not currently do anything sensible in this case.
Differential Revision: http://reviews.llvm.org/D13505
llvm-svn: 249669
Compare elimination extended to recognize load-and-test instructions used
for comparison and eliminate them the same way as with compare instructions.
Test case fp-cmp-05.ll updated to expect optimized results now also for z13.
The order of instruction shortening and compare elimination passes have been
changed so that opcodes do not have to be handled in both passes.
Reviewed by Ulrich Weigand.
llvm-svn: 249666
Since the LTxBRCompare instructions can't be used with vector registers, a
normal load-and-test instruction (with a modelled def operand) is used instead.
Reviewed by Ulrich Weigand.
llvm-svn: 249664
In particular, passing non-trivially copyable objects by value on win32
uses a dynamic alloca (inalloca). We would clobber ESP in the epilogue
and end up returning to outer space.
llvm-svn: 249637
The relocation for the filter funclet will be against a symbol table
entry for a function instead of the section, making it easier to
understand what is going on.
llvm-svn: 249621
The __CxxFrameHandler3 tables for 32-bit are supposed to hold stack
offsets relative to EBP, not ESP. I blindly updated the win-catchpad.ll
test case, and immediately noticed that 32-bit catching stopped working.
While I'm at it, move the frame index to frame offset WinEH table logic
out of PEI. PEI shouldn't have to know about WinEHFuncInfo. I realized
we can calculate frame index offsets just fine from the table printer.
llvm-svn: 249618
We remove unreachable blocks because it is pointless to consider them
for coloring. However, we still had stale pointers to these blocks in
some data structures after we removed them from the function.
Instead, remove the unreachable blocks before attempting to do anything
with the function.
This fixes PR25099.
llvm-svn: 249617
Summary:
This is necessary to keep the cloner from making bogus copies of debug
metadata attached to the IR it is cloning.
Also, avoid running RemapInstruction over all instructions in the common
case that no cloning was performed.
Reviewers: rnk, andrew.w.kaylor, majnemer
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D13514
llvm-svn: 249591
There was an off-by-one bug in ip2state tables which manifested when one
call immediately preceded the try-range of the next. The return address
of the previous call would appear to be within the try range of the next
scope, resulting in extra destructors or catches running.
We also computed the wrong offset for catch parameter stack objects. The
offset should be from RSP, not from RBP.
llvm-svn: 249578
I'll be using the function in a similar combine for AArch64. The helper was
also improved to handle undef values.
Part of http://reviews.llvm.org/D13442
llvm-svn: 249572
The ARM RTABI defines the half- to single-precision float conversion functions
with an __aeabi prefix, but libgcc only has them with a __gnu prefix. Therefore
we need to emit the __aeabi version when compiling with an eabi or eabihf
triple, and the __gnu version with a gnueabi or gnueabihf triple.
llvm-svn: 249565
Without an additional check for NEON, the compiler crashes during
legalization of NEON ldN/stN.
Differential Revision: http://reviews.llvm.org/D13508
llvm-svn: 249550
When outgoing function arguments are passed using push instructions, and EH
is enabled, we may need to indicate to the stack unwinder that the stack
pointer was adjusted before the call.
This should fix the exception handling issues in PR24792.
Differential Revision: http://reviews.llvm.org/D13132
llvm-svn: 249522
This stops using an unknown reg class operand.
Currently build_vector selection has a broken looking check
where it tries to use a VGPR reg class and an SGPR one if it
sees an SGPR use.
With the source operand has an explicit VGPR class,
illegal copies will be inserted that SIFixSGPRCopies will take care
of normally later, which will allow removing the weird check
of build_vector users. Without this, when removed v_movrels_b32 would
still be emitted even though all of the values were only stored in
SGPRs.
llvm-svn: 249494
Our current emission strategy is to emit the funclet prologue in the
CatchPad's normal destination. This is problematic because
intra-funclet control flow to the normal destination is not erroneous
and results in us reevaluating the prologue if said control flow is
taken.
Instead, use the CatchPad's location for the funclet prologue. This
correctly models our desire to have unwind edges evaluate the prologue
but edges to the normal destination result in typical control flow.
Differential Revision: http://reviews.llvm.org/D13424
llvm-svn: 249483
Summary:
We currently ignore the calling convention, so there is no real reason to
assert on the calling convention of functions.
Reviewers: arsenm
Subscribers: arsenm, llvm-commits
Differential Revision: http://reviews.llvm.org/D13367
llvm-svn: 249468
Summary:
This fixes 7 tests during fast LLVM test-suite run:
* MultiSource/Benchmarks/McCat/18-imp/imp
* MultiSource/Applications/oggenc/oggenc
* MultiSource/Benchmarks/MallocBench/gs/gs
* MultiSource/Benchmarks/MiBench/automotive-susan/automotive-susan
* MultiSource/Benchmarks/VersaBench/beamformer/beamformer
* MultiSource/Benchmarks/MiBench/consumer-lame/consumer-lame
* MultiSource/Benchmarks/Bullet/bullet
Error message was in the form of:
fatal error: error in backend: Cannot select: 0x95c3288: f32 = fsqrt 0x95c0190 [ORD=9] [ID=18]
0x95c0190: f32 = fadd 0x95bef30, 0x95c4d00 [ORD=8] [ID=17]
0x95bef30: f32 = fmul 0x95c4988, 0x95c4988 [ORD=5] [ID=16]
...
There was problem with selecting sqrt instruction in LLVM backend.
To fix the issue changes are made in TableGen definition for sqrt instruction in MipsInstrFPU.td and new test file sqrt.ll is added to LLVM regression tests.
Patch by Zlatko Buljan
Reviewers: zoran.jovanovic, hvarga, dsanders
Subscribers: llvm-commits, petarj
Differential Revision: http://reviews.llvm.org/D13235
llvm-svn: 249416
For the program like below
struct key_t {
int pid;
char name[16];
};
extern void test1(char *);
int test() {
struct key_t key = {};
test1(key.name);
return 0;
}
For key.name, the llc/bpf may generate the below code:
R1 = R10 // R10 is the frame pointer
R1 += -24 // framepointer adjustment
R1 |= 4 // R1 is then used as the first parameter of test1
OR operation is not recognized by in-kernel verifier.
This patch introduces an intermediate FI_ri instruction and
generates the following code that can be properly verified:
R1 = R10
R1 += -20
Patch by Yonghong Song <yhs@plumgrid.com>
llvm-svn: 249371
Most importantly, this keeps constant hoisting from preventing instruction selections ability to turn an AND with 0xffffffff into a move into a 32-bit subregister.
llvm-svn: 249370
This new syntax is built around putting each instruction on its own line
in a "mnemonic op, op, op" like syntax. It also uses conventional data
section directives like ".byte" and so on rather than requiring everything
to be in hierarchical S-expression format. This is a more natural syntax
for a ".s" file format from the perspective of LLVM MC and related tools,
while remaining easy to translate into other forms as needed.
llvm-svn: 249364
We were previously codegen'ing memcpy as regular load/store operations and
hoping that the register allocator would allocate registers in ascending order
so that we could apply an LDM/STM combine after register allocation. According
to the commit that first introduced this code (r37179), we planned to teach the
register allocator to allocate the registers in ascending order. This never got
implemented, and up to now we've been stuck with very poor codegen.
A much simpler approach for achieving better codegen is to create MEMCPY pseudo
instructions, attach scratch virtual registers to them and then, post register
allocation, expand the MEMCPYs into LDM/STM pairs using the scratch registers.
The register allocator will have picked arbitrary registers which we sort when
expanding the MEMCPY. This approach also avoids the need to repeatedly calculate
offsets which ultimately ought to be eliminated pre-RA in order to decrease
register pressure.
Fixes PR9199 and PR23768.
[This is based on Peter Collingbourne's r238473 which was reverted.]
Differential Revision: http://reviews.llvm.org/D13239
Change-Id: I727543c2e94136e0f80b8e22d5642d7b9ee5b458
Author: Peter Collingbourne <peter@pcc.me.uk>
llvm-svn: 249322
Track which basic blocks belong to which funclets. Permit branch
folding to fire but only if it can prove that doing so will not cause
code in one funclet to be reused in another.
llvm-svn: 249257
This patch teaches FastIsel the following two things:
1) On SSE2, no instructions are needed for bitcasts between 128-bit vector types;
2) On AVX, no instructions are needed for bitcasts between 256-bit vector types.
Example:
%1 = bitcast <4 x i31> %V to <2 x i64>
Before (-fast-isel -fast-isel-abort=1):
FastIsel miss: %1 = bitcast <4 x i31> %V to <2 x i64>
Now we don't fall back to SelectionDAG and we correctly fold that computation
propagating the register associated to %V.
Originally reviewed here: http://reviews.llvm.org/D13347
llvm-svn: 249147
This patch teaches FastIsel the following two things:
1) On SSE2, no instructions are needed for bitcasts between 128-bit vector types;
2) On AVX, no instructions are needed for bitcasts between 256-bit vector types.
Example:
%1 = bitcast <4 x i31> %V to <2 x i64>
Before (-fast-isel -fast-isel-abort=1):
FastIsel miss: %1 = bitcast <4 x i31> %V to <2 x i64>
Now we don't fall back to SelectionDAG and we correctly fold that computation
propagating the register associated to %V.
Differential Revision: http://reviews.llvm.org/D13347
llvm-svn: 249121
We emit denormalized tables, where every range of invokes in the same
state gets a complete list of EH action entries. This is significantly
simpler than trying to infer the correct nested scoping structure from
the MI. Fortunately, for SEH, the nesting structure is really just a
size optimization.
With this, some basic __try / __except examples work.
llvm-svn: 249078
Summary:
Instead of asserting when the kernel metadata is different than we expect,
we should just skip lowering that function. This fixes assertion
failures with OpenCL argument metadata from older LLVM releases.
Reviewers: arsenm
Subscribers: arsenm, llvm-commits
Differential Revision: http://reviews.llvm.org/D13356
llvm-svn: 249073
Catchret transfers control from a catch funclet to an earlier funclet.
However, it is not completely clear which funclet the catchret target is
part of. Make this clear by stapling the catchret target's funclet
membership onto the CATCHRET SDAG node.
llvm-svn: 249052
Add generic instructions for load complement, load negative and load positive
for fp32 and fp64, and let isel prefer them. They do not clobber CC, and so
give scheduler more freedom. SystemZElimCompare pass will convert them when it
can to the CC-setting variants.
Regression tests updated to expect the new opcodes in places where the old ones
where used. New test case SystemZ/fp-cmp-05.ll checks that
SystemZCompareElim.cpp can handle the new opcodes.
README.txt updated (bullet removed).
Note that fp128 is not yet handled, because it is relatively rare, and is a
bit trickier, because of the fact that l.dfr would operate on the sign bit of
one of the subregisters of a fp128, but we would not want to copy the other
sub-reg in case src and dst regs are not the same.
Reviewed by Ulrich Weigand.
llvm-svn: 249046
v2: Add test (Matt).
Fix capitalization of isEOP (Matt).
Move pattern to class parameter (Matt).
Make the instruction available to Cayman (Matt).
Change name from MEM_RAT WRITE_TYPED to MEM_RAT STORE_TYPED.
Patch by: Zoltan Gilian
llvm-svn: 249042
The custom code produces incorrect results if later reassociated.
Since r221657, on x86, vNi32 uitofp is lowered using an optimized
sequence:
movdqa LCPI0_0(%rip), %xmm1 ## xmm1 = [65535, ...]
pand %xmm0, %xmm1
por LCPI0_1(%rip), %xmm1 ## [0x4b000000, ...]
psrld $16, %xmm0
por LCPI0_2(%rip), %xmm0 ## [0x53000000, ...]
addps LCPI0_3(%rip), %xmm0 ## [float -5.497642e+11, ...]
addps %xmm1, %xmm0
Since r240361, the machine combiner opportunistically reassociates
2-instruction sequences (with -ffast-math). In the new code sequence,
the ADDPS' are eligible. In isolation, for simple examples (without
reassociable users), this makes no performance difference (the goal
being to enable reassociation of longer chains).
In the trivial example (just one uitofp), the reassociation doesn't
happen, because (I think) it would require the emission of a separate
movaps for a constantpool load (instead of folding it into addps).
However, when we have multiple uitofp sequences, and the constantpool
loads are CSE'd earlier, the machine combiner can do the reassociation.
When the ADDPS' are reassociated, the resulting sequence isn't correct
anymore, as we'd be adding large (2**39) constants with comparatively
smaller values (~2**23). Given that two of the three inputs are powers
of 2 larger than 2**16, and that ulp(2**39) == 2**(39-24) == 2**15,
the reassociated chain will produce 0 for any input in [0, 2**14[.
In my testing, it also produces wrong results for 99.5% of [0, 2**32[.
Avoid this by disabling the new lowering when -ffast-math. It does
mean that we'll get slower code than without it, but at least we
won't get egregiously incorrect code.
One might argue that, considering -ffast-math is all but meaningless,
uitofp producing wrong results isn't a compiler bug. But it really is.
Fixes PR24512.
...though this is really more of a workaround.
Ideally, we'd have some sort of Machine FMF, but that's a problem
that's not worth tackling until we do more with machine IR.
llvm-svn: 248965
The Win64 unwinder disassembles forwards from each PC to try to
determine if this PC is in an epilogue. If so, it skips calling the EH
personality function for that frame. Typically, this means you cannot
catch an exception in the same frame that you threw it, because 'throw'
calls a noreturn runtime function.
Previously we avoided this problem with the TrapUnreachable
TargetOption, but that's a much bigger hammer than we need. All we need
is a 1 byte non-epilogue instruction right after the call. Instead,
what we got was an unconditional branch to a shared block containing the
ud2, potentially 7 bytes instead of 1. So, this reverts r206684, which
added TrapUnreachable, and replaces it with something better.
The new code pattern matches for invoke/call followed by unreachable and
inserts an int3 into the DAG. To be 100% watertight, we would need to
insert SEH_Epilogue instructions into all basic blocks ending in a call
with no terminators or successors, but in practice this is unlikely to
come up.
llvm-svn: 248959
Unscaled load/store combining has been enabled since the initial ARM64 port. No
need for a redundance run. Also, add CHECK-LABEL directives.
llvm-svn: 248945
Previously, the index was constrained to the size of the memory operation for
no apparent reason. This change removes that constraint so that we can form
pre-index instructions with any valid offset.
llvm-svn: 248931
This commit changes the interface of the vld[1234], vld[234]lane, and vst[1234],
vst[234]lane ARM neon intrinsics and associates an address space with the
pointer that these intrinsics take. This changes, e.g.,
<2 x i32> @llvm.arm.neon.vld1.v2i32(i8*, i32)
to
<2 x i32> @llvm.arm.neon.vld1.v2i32.p0i8(i8*, i32)
This change ensures that address spaces are fully taken into account in the ARM
target during lowering of interleaved loads and stores.
Differential Revision: http://reviews.llvm.org/D12985
llvm-svn: 248887
The XOP shifts just have logical/arithmetic versions and the left/right shifts are controlled by whether the value is positive/negative. Because of this I've added new X86ISD nodes instead of trying to force them to use the existing shift nodes.
Additionally Excavator cores (bdver4) support XOP and AVX2 - meaning that it should use the AVX2 shifts when it can and fall back to XOP in other cases.
Differential Revision: http://reviews.llvm.org/D8690
llvm-svn: 248878
The x64 ABI requires that epilogues do not contain code other than stack
adjustments and some limited control flow. However, we'd insert code to
initialize the return address after stack adjustments. Instead, insert
EAX/RAX with the current value before we create the stack adjustments in
the epilogue.
llvm-svn: 248839
HHVM calling convention, hhvmcc, is used by HHVM JIT for
functions in translated cache. We currently support LLVM back end to
generate code for X86-64 and may support other architectures in the
future.
In HHVM calling convention any GP register could be used to pass and
return values, with the exception of R12 which is reserved for
thread-local area and is callee-saved. Other than R12, we always
pass RBX and RBP as args, which are our virtual machine's stack pointer
and frame pointer respectively.
When we enter translation cache via hhvmcc function, we expect
the stack to be aligned at 16 bytes, i.e. skewed by 8 bytes as opposed
to standard ABI alignment. This affects stack object alignment and stack
adjustments for function calls.
One extra calling convention, hhvm_ccc, is used to call C++ helpers from
HHVM's translation cache. It is almost identical to standard C calling
convention with an exception of first argument which is passed in RBP
(before we use RDI, RSI, etc.)
Differential Revision: http://reviews.llvm.org/D12681
llvm-svn: 248832
Summary:
Funclets have been turned into functions by the time they hit the object
file. Make sure that they have decent names for the symbol table and
CFI directives explaining how to reason about their prologues.
Differential Revision: http://reviews.llvm.org/D13261
llvm-svn: 248824
alignment requirements, for example in the case of vectors.
These requirements are exploited by the code generator by using
move instructions that have similar alignment requirements, e.g.,
movaps on x86.
Although the code generator properly aligns the arguments with
respect to the displacement of the stack pointer it computes,
the displacement itself may cause misalignment. For example if
we have
%3 = load <16 x float>, <16 x float>* %1, align 64
call void @bar(<16 x float> %3, i32 0)
the x86 back-end emits:
movaps 32(%ecx), %xmm2
movaps (%ecx), %xmm0
movaps 16(%ecx), %xmm1
movaps 48(%ecx), %xmm3
subl $20, %esp <-- if %esp was 16-byte aligned before this instruction, it no longer will be afterwards
movaps %xmm3, (%esp) <-- movaps requires 16-byte alignment, while %esp is not aligned as such.
movl $0, 16(%esp)
calll __bar
To solve this, we need to make sure that the computed value with which
the stack pointer is changed is a multiple af the maximal alignment seen
during its computation. With this change we get proper alignment:
subl $32, %esp
movaps %xmm3, (%esp)
Differential Revision: http://reviews.llvm.org/D12337
llvm-svn: 248786
When AA is being used, non-aliasing stores are canonicalized to use the same
chain, and DAGCombiner::getStoreMergeAndAliasCandidates can take advantage of
this by looking only as users of a store's chain operand. However, user
iteration is not result-number specific, we need to check that the use is as a
chain operand, and not via some other operand. It is certainly possible to have
another potentially-aliasing store, which shares the first's base pointer, and
uses the first's chain's node via some other operand.
Failure to catch this situation caused, at least in the included test case, an
assert later because the relative sequence-number ordering caused later
replacement to create a cycle in the DAG.
llvm-svn: 248698
Summary:
Factor the code that rewrites invokes to calls and rewrites WinEH
terminators to their "unwind to caller" equivalents into a helper in
Utils/Local, and use it in the three places I'm aware of that need to do
this.
Reviewers: andrew.w.kaylor, majnemer, rnk
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D13152
llvm-svn: 248677
Trying to use the version with the explicit output operand
would complain because of the missing WriteSALU. I'm not sure
why it doesn't complain about this with the implicit VCC def.
llvm-svn: 248646
BranchProbability now is represented by its numerator and denominator in uint32_t type. This patch changes this representation into a fixed point that is represented by the numerator in uint32_t type and a constant denominator 1<<31. This is quite similar to the representation of BlockMass in BlockFrequencyInfoImpl.h. There are several pros and cons of this change:
Pros:
1. It uses only a half space of the current one.
2. Some operations are much faster like plus, subtraction, comparison, and scaling by an integer.
Cons:
1. Constructing a probability using arbitrary numerator and denominator needs additional calculations.
2. It is a little less precise than before as we use a fixed denominator. For example, 1 - 1/3 may not be exactly identical to 1 / 3 (this will lead to many BranchProbability unit test failures). This should not matter when we only use it for branch probability. If we use it like a rational value for some precise calculations we may need another construct like ValueRatio.
One important reason for this change is that we propose to store branch probabilities instead of edge weights in MachineBasicBlock. We also want clients to use probability instead of weight when adding successors to a MBB. The current BranchProbability has more space which may be a concern.
Differential revision: http://reviews.llvm.org/D12603
llvm-svn: 248633
This is a redo of D7208 ( r227242 - http://llvm.org/viewvc/llvm-project?view=revision&revision=227242 ).
The patch was reverted because an AArch64 target could infinite loop after the change in DAGCombiner
to merge vector stores. That happened because AArch64's allowsMisalignedMemoryAccesses() wasn't telling
the truth. It reported all unaligned memory accesses as fast, but then split some 128-bit unaligned
accesses up in performSTORECombine() because they are slow.
This patch attempts to fix the problem in AArch's allowsMisalignedMemoryAccesses() while preserving
existing (perhaps questionable) lowering behavior.
The x86 test shows that store merging is working as intended for a target with fast 32-byte unaligned
stores.
Differential Revision: http://reviews.llvm.org/D12635
llvm-svn: 248622
The algorithm would not modify the live-in list of blocks below the save
block point which is correct unless it happens to be a restore point at
the same time.
Also fixes the benign issue of live-in registers being added twice in
some cases.
The testcase is based on a test submitted by Kit Barton.
Differential Revision: http://reviews.llvm.org/D13176
llvm-svn: 248620
If a virtual register is copied and another copy was already
seen, replace with the previous copy. This only handles the
simplest cases for now.
This pattern shows up from various operand restrictions
AMDGPU has which require inserting copies depending
on the register class of the operands.
llvm-svn: 248611
This fixes a select error when the i64 source was also
bitcasted to v2i32 in the original source.
Instead of awkwardly trying to select the modified source value and
the store, replace before isel begins.
Uses a worklist to avoid possible problems from mutating the DAG,
although it seems to work OK without it.
llvm-svn: 248589
These were all using the default 32-bit VALU write class,
but the i64/f64 compares are half rate.
I'm not sure this is really correct, because they are still using
the write to VALU write class, even though they really write
to the SALU.
llvm-svn: 248582
We now emit the compiler generated divide by zero check that was needed for the
MSVC routines. We construct a psuedo-instruction for the DBZ check as the
operation requires splitting up the BB. For the 64-bit operations, we need to
custom expand the node as we need to insert the DBZ check and then emit the
libcall to the appropriate name. Because this is target specific, it seemed
better to reproduce the expansion operation from the target-agnostic type
legalization rather than sink this there to avoid the duplication. The division
library calls now match MSVC semantically.
llvm-svn: 248561
Fix for D12561 - we weren't correctly ensuring that the base element for extension was moved to start on a boundary suitable for UNPCKL/H
llvm-svn: 248536
These are necessary for implementing mem_fence for
OpenCL 2.0.
The VI assembler tests are disabled since it seems to be
using the wrong encoding or opcode.
llvm-svn: 248532
Fixes the overflow case of llvm.*absdiff intrinsic also updats the tests and LangRef.rst accordingly.
Differential Revision: http://reviews.llvm.org/D11678
llvm-svn: 248483
Allow a target to do something other than search for copies
that will avoid cross register bank copies.
Implement for SI by only rewriting the most basic copies,
so it should look through anything like a subregister extract.
I'm not entirely satisified with this because it seems like
eliminating a reg_sequence that isn't fully used should work
generically for all targets without them having to override
something. However, it seems to be tricky to have a simple
implementation of this without rewriting to invalid kinds
of subregister copies on some targets.
I'm not sure if there is currently a generic way to easily check
if a subregister index would be valid for the current use.
The current set of TargetRegisterInfo::get*Class functions don't
quite behave like I would expect (e.g. getSubClassWithSubReg
returns the maximal register class rather than the minimal), so
I'm not sure how to make the generic test keep searching if
SrcRC:SrcSubReg is a valid replacement for DefRC:DefSubReg. Making
the default implementation to check for simple copies breaks
a variety of ARM and x86 tests by producing illegal subregister uses.
The ARM tests are not actually changed since it should still be using
the same sharesSameRegisterFile implementation, this just relaxes
them to not check for specific registers.
llvm-svn: 248478
If the stores are storing values from loads which partially
alias the stores, we could end up placing the merged loads
and stores on the same chain which has the potential to break.
Each store may have a different chain dependency on only some
of the original loads. Create a new TokenFactor to capture all
of the required dependencies of the stores rather than assuming
all stores can use the same chain.
The testcase is a situation where this happens, although
it does not have an observable change from this. The DAG nodes
just happened to not be reordered before despite this missing
chain dependency.
This is based on an off-list report for an out of tree target
which regressed due to r246307 and I haven't managed to find a case
where the nodes do end up reordered with an in tree target.
llvm-svn: 248468
Instead of always inserting a copy in case
the super register is itself a subregister,
only extract to the super reg class if this is
actually the case.
This shouldn't really change codegen, but
makes looking at the output of SIFixSGPRCopies
easier to read.
llvm-svn: 248467
This time, the issue is that we weren't accounting for the possibility that
aligned DPRs could have been stored after the final "push" in a prologue. When
that happened we effectively moved a "sub sp, #N" from below the aligned stores
to above them, and everything went to pot.
To make it worse, I'd actually committed something testing that we produced
wrong code, so the test update is tiny.
llvm-svn: 248437
This patch changes the order of GEPs generated by Splitting GEPs
pass, specially when one of the GEPs has constant and the base is
loop invariant, then we will generate the GEP with constant first
when beneficial, to expose more cases for LICM.
If originally Splitting GEP generate the following:
do.body.i:
%idxprom.i = sext i32 %shr.i to i64
%2 = bitcast %typeD* %s to i8*
%3 = shl i64 %idxprom.i, 2
%uglygep = getelementptr i8, i8* %2, i64 %3
%uglygep7 = getelementptr i8, i8* %uglygep, i64 1032
...
Now it genereates:
do.body.i:
%idxprom.i = sext i32 %shr.i to i64
%2 = bitcast %typeD* %s to i8*
%3 = shl i64 %idxprom.i, 2
%uglygep = getelementptr i8, i8* %2, i64 1032
%uglygep7 = getelementptr i8, i8* %uglygep, i64 %3
...
For no-loop cases, the original way of generating GEPs seems to
expose more CSE cases, so we don't change the logic for no-loop
cases, and only limit our change to the specific case we are
interested in.
llvm-svn: 248420
This patches removes the x86.sse41.pmovsx* intrinsics, provides a suitable upgrade path and updates relevant tests to sign extend a subvector instead.
LLVM counterpart to D12835
Differential Revision: http://reviews.llvm.org/D13002
llvm-svn: 248368
We may have subregister defs which are unused but not discovered and
cleaned up prior to liveness analysis. This creates multiple connected
components in the resulting live range which are forbidden in the
MachineVerifier because they would unnecesarily constrain the register
allocator. Rewrite those dead definitions to define a newly created
virtual register.
Differential Revision: http://reviews.llvm.org/D13035
llvm-svn: 248335
ARM counterpart to r248291:
In the comparison failure block of a cmpxchg expansion, the initial
ldrex/ldxr will not be followed by a matching strex/stxr.
On ARM/AArch64, this unnecessarily ties up the execution monitor,
which might have a negative performance impact on some uarchs.
Instead, release the monitor in the failure block.
The clrex instruction was designed for this: use it.
Also see ARMARM v8-A B2.10.2:
"Exclusive access instructions and Shareable memory locations".
Differential Revision: http://reviews.llvm.org/D13033
llvm-svn: 248294
In the comparison failure block of a cmpxchg expansion, the initial
ldrex/ldxr will not be followed by a matching strex/stxr.
On ARM/AArch64, this unnecessarily ties up the execution monitor,
which might have a negative performance impact on some uarchs.
Instead, release the monitor in the failure block.
The clrex instruction was designed for this: use it.
Also see ARMARM v8-A B2.10.2:
"Exclusive access instructions and Shareable memory locations".
Differential Revision: http://reviews.llvm.org/D13033
llvm-svn: 248291
The C standard has historically not specified whether or not these functions should raise the inexact flag. Traditionally on Darwin, these functions *did* raise inexact, and the llvm lowerings followed that conventions. n1778 (C bindings for IEEE-754 (2008)) clarifies that these functions should not set inexact. This patch brings the lowerings for arm64 and x86 in line with the newly specified behavior. This also lets us fold some logic into TD patterns, which is nice.
Differential Revision: http://reviews.llvm.org/D12969
llvm-svn: 248266
This patch generalizes the lowering of shuffles as zero extensions to allow extensions that don't start from the first element. It now recognises extensions starting anywhere in the lower 128-bits or at the start of any higher 128-bit lane.
The motivation was to reduce the number of high cost pshufb calls, but it also improves the SSE2 case as well.
Differential Revision: http://reviews.llvm.org/D12561
llvm-svn: 248250
This patch adds support for combining patterns such as (FMUL(FADD(1.0, x), y)) and (FMUL(FSUB(x, 1.0), y)) to their FMA equivalents.
This is useful in particular for linear interpolation cases such as (FADD(FMUL(x, t), FMUL(y, FSUB(1.0, t))))
Differential Revision: http://reviews.llvm.org/D13003
llvm-svn: 248210
The vext pseudo-instruction takes the number of elements that need to be
extracted, not the number of bytes. Hence, use the number of elements
directly instead of scaling them with a factor.
Reviewers: Silviu Baranga, James Molloy
(not reflected in the differential revision)
Differential Revision: http://reviews.llvm.org/D12974
llvm-svn: 248208
The ISD::FPOW and ISD::FSINCOS opcodes default to Legal, but there
is no legal instruction for those on SystemZ. This could cause
LLVM internal errors. Fixed by setting the operation action to
Expand for those opcodes.
Also added test cases for all other LLVM IR intrinsics that should
generate a library call. (Those already work correctly since the
default operation action is fine.)
llvm-svn: 248180
If storing multiple FP constants, some subset of the stores
would be replaced with integers due to visit order, so
MergeConsecutiveStores would only partially merge
these.
llvm-svn: 248169
Now that we have fast vector CTPOP implementations we can use this to speed up vector CTTZ using the pattern (cttz(x) = ctpop((x & -x) - 1))
Additionally, for AVX512CD that provides lzcnt instructions we can use the pattern (cttz_undef(x) = (width - 1) - ctlz(x & -x))
Differential Revision: http://reviews.llvm.org/D12663
llvm-svn: 248091
In if-conversion, there is a utility function MergeBlocks() that is used to merge blocks. However, when new edges are built in this function the edge weight is either not provided or not updated properly, leading to a modified CFG with incorrect edge weights. This patch corrects this issue.
Differential Revision: http://reviews.llvm.org/D12513
llvm-svn: 248030
In ARMBaseInstrInfo::isProfitableToIfCvt(), there is a simple cost model in which the number of cycles is scaled by a probability to estimate the cost. However, when the number of cycles is small (which is usually the case), there is a precision issue after the computation. To avoid this issue, this patch scales those cycles by 1024 (chosen to make the multiplication a litter faster) before they are scaled by the probability. Other variables are also scaled up for the final comparison.
Differential Revision: http://reviews.llvm.org/D12742
llvm-svn: 248018
Summary:
For bitfield insert OR matching, check both operands for larger pattern
first before checking for smaller pattern.
Add pattern for unsigned bitfield insert-in-zero done with SHL+AND.
Resolves PR21631.
Reviewers: jmolloy, t.p.northover
Subscribers: aemerson, rengolin, llvm-commits, mcrosier
Differential Revision: http://reviews.llvm.org/D12908
llvm-svn: 248006
- Strenghten the logic to be sure we hoist the restore point out of the current
loop. (The fixes a bug with infinite loop, added as part of the patch.)
- Walk over the exit blocks of the current loop to conver to the desired restore
point in one iteration of the update loop.
llvm-svn: 247958
Windows EH funclets need to be contiguous. The FuncletLayout pass will
ensure that the funclets are together and begin with a funclet entry MBB.
Differential Revision: http://reviews.llvm.org/D12943
llvm-svn: 247937
This makes catchret look more like a branch, and less like a weird use
of BlockAddress. It also lets us get away from
llvm.x86.seh.restoreframe, which relies on the old parentfpoffset label
arithmetic.
llvm-svn: 247936
getLandingPadSuccessor assumes that each invoke can have at most one EH
pad successor, but WinEH invokes can have more than one. Two out of
three callers of getLandingPadSuccessor don't use the returned
landingpad, so we can make them use this simple predicate instead.
Eventually we'll have to circle back and fix SplitKit.cpp so that
register allocation works. Baby steps.
llvm-svn: 247904
AVX-512 does not provide an instruction that shuffles mask register. So I do the following way:
mask-2-simd , shuffle simd , simd-2-mask
Differential Revision: http://reviews.llvm.org/D12727
llvm-svn: 247876
The MSVC doesn't really support exception specifications so let's just
turn these into cleanuppads. Later, we might use terminatepad to more
efficiently encode the "noexcept"-ness of a function body.
llvm-svn: 247848
Clang now passes the adjectives as an argument to catchpad.
Getting the CatchObj working is simply a matter of threading another
static alloca through codegen, first as an alloca, then as a frame
index, and finally as a frame offset.
llvm-svn: 247844
We are experimenting with a new approach to saving and restoring SSA
values used across funclets: let the register allocator do the dirty
work for us.
However, this means that we need to be able to clone commoned blocks
without relying on demotion.
llvm-svn: 247835
Otherwise we'd try to emit the thunk that passes the LSDA to
__CxxFrameHandler3. We don't emit the LSDA if there were no landingpads,
so we'd end up with an assembler error when trying to write the COFF
object.
llvm-svn: 247820
This pass implements a simple algorithm for conversion from CFG to
wasm's structured control flow. It doesn't yet handle multiple-entry
loops; that will be added in a future patch.
It also adds initial support for switch statements.
Differential Revision: http://reviews.llvm.org/D12735
llvm-svn: 247818
After D10403, we had FMF in the DAG but disabled by default. Nick reported no crashing errors after some stress testing,
so I enabled them at r243687. However, Escha soon notified us of a bug not covered by any in-tree regression tests:
if we don't propagate the flags, we may fail to CSE DAG nodes because differing FMF causes them to not match. There is
one test case in this patch to prove that point.
This patch hopes to fix or leave a 'TODO' for all of the in-tree places where we create nodes that are FMF-capable. I
did this by putting an assert in SelectionDAG.getNode() to find any FMF-capable node that was being created without FMF
( D11807 ). I then ran all regression tests and test-suite and confirmed that everything passes.
This patch exposes remaining work to get DAG FMF to be fully functional: (1) add the flags to non-binary nodes such as
FCMP, FMA and FNEG; (2) add the flags to intrinsics; (3) use the flags as conditions for transforms rather than the
current global settings.
Differential Revision: http://reviews.llvm.org/D12095
llvm-svn: 247815
When trying emit a stack adjustments using pops, frame lowering selects an
arbitrary free GPR. It should always select one from an appropriate class...
This fixes PR24649.
Patch by: amjad.aboud@intel.com
Differential Revision: http://reviews.llvm.org/D12609
llvm-svn: 247785
When building LLVM as a (potentially dynamic) library that can be linked against
by multiple compilers, the default triple is not really meaningful.
We allow to explicitely set it to an empty string when configuring LLVM.
In this case, said "target independent" tests in the test suite that are using
the default triple are disabled by matching the newly available feature
"default_triple".
Reviewers: probinson, echristo
Differential Revision: http://reviews.llvm.org/D12660
From: Mehdi Amini <mehdi.amini@apple.com>
llvm-svn: 247775
Under certain circumstances, tryBuildVectorShuffle would attempt to
create a BUILD_VECTOR node with an invalid combination of types.
This happened when one of the components of the original BUILD_VECTOR
was itself a TRUNCATE node. That TRUNCATE was stripped off during
intermediate processing to simplify code, but when adding the node
back to the result vector, we still need it to get the type right.
llvm-svn: 247694
Turning (op x (mul y k)) into (op x (lsl (mul y k>>n) n)) is beneficial when
we can do the lsl as a shifted operand and the resulting multiply constant is
simpler to generate.
Do this by doing the transformation when trying to select a shifted operand,
as that ensures that it actually turns out better (the alternative would be to
do it in PreprocessISelDAG, but we don't know for sure there if extracting the
shift would allow a shifted operand to be used).
Differential Revision: http://reviews.llvm.org/D12196
llvm-svn: 247569
KNL does not have VXORPS, VORPS for 512-bit values.
I use integer VPXOR, VPOR that actually do the same.
X86ISD::FXOR/FOR are generated as a result of FSUB combining.
Differential Revision: http://reviews.llvm.org/D12753
llvm-svn: 247523
The changes in:
test/CodeGen/X86/machine-cp.ll
are just due to scheduling differences after some logic instructions were reassociated.
llvm-svn: 247516
realignment should be forced.
With this commit, we can now force stack realignment when doing LTO and
do so on a per-function basis. Also, add a new cl::opt option
"stackrealign" to CommandFlags.h which is used to force stack
realignment via llc's command line.
Out-of-tree projects currently using -force-align-stack to force stack
realignment should make changes to attach the attribute to the functions
in the IR.
Differential Revision: http://reviews.llvm.org/D11814
llvm-svn: 247450
We used different conditions to determine if we should emit startproc vs
endproc. Use the same condition to ensure that they will always be
paired.
This fixes PR24374.
llvm-svn: 247435
The rest of the EH pads are fine, since they have at most one label and
take fewer operands for the personality.
Old catchpad vs. new:
%5 = catchpad [i8* bitcast (i32 ()* @"\01?filt$0@0@main@@" to i8*)] to label %__except.ret.10 unwind label %catchendblock.9
-----
%5 = catchpad [i8* bitcast (i32 ()* @"\01?filt$0@0@main@@" to i8*)]
to label %__except.ret.10 unwind label %catchendblock.9
llvm-svn: 247433
The Win32 EH runtime caller does not preserve EBP, even though it does
preserve the CSRs (EBX, ESI, EDI) for us. The result was that each
finally funclet call would leave the frame pointer off by 12 bytes.
llvm-svn: 247348
The (mostly-deprecated) SelectionDAG-based ILPListDAGScheduler scheduler
was making poor scheduling decisions, causing high register pressure and
extraneous register spills.
Switching to the newer machine scheduler generates better code -- even
without there being a machine model defined for SPARC yet.
(Actually committing the test changes too, this time, unlike r247315)
llvm-svn: 247343
Summary:
The coloring code in WinEHPrepare queues cleanuprets' successors with the
correct color (the parent one) when it sees their cleanuppad, and so later
when iterating successors knows to skip processing cleanuprets since
they've already been queued. This latter check was incorrectly under an
'else' condition and so inadvertently was not kicking in for single-block
cleanups. This change sinks the check out of the 'else' to fix the bug.
Reviewers: majnemer, andrew.w.kaylor, rnk
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D12751
llvm-svn: 247299
order.
The implicit register verifier in the MIR parser should only check if the
instruction's default implicit operands are present in the instruction. It
should not check the order in which they occur.
llvm-svn: 247283
Summary:
The BUILD_VECTOR node will truncate its operators to match the
type. We need to take this into account when constant folding -
we need to perform a truncation before constant folding the elements.
This is because the upper bits can change the result, depending on
the operation type (for example this is the case for min/max).
This change also adds a regression test.
Reviewers: jmolloy
Subscribers: jmolloy, llvm-commits
Differential Revision: http://reviews.llvm.org/D12697
llvm-svn: 247265
The tests in isVTRNMask and isVTRN_v_undef_Mask should also check that the elements of the upper and lower half of the vectorshuffle occur in the correct order when both halves are used. Without this test the code assumes that it is correct to use vector transpose (vtrn) for the masks <1, 1, 0, 0> and <1, 3, 0, 2>, among others, but the transpose actually incorrectly generates shuffles for <0, 0, 1, 1> and <0, 2, 1, 3> in this case.
Patch by Jeroen Ketema!
llvm-svn: 247254
The changes in this patch are as follows:
1. Modify the emitPrologue and emitEpilogue methods to work properly when the prologue and epilogue blocks are not the first/last blocks in the function
2. Fix a bug in PPCEarlyReturn optimization caused by an empty entry block in the function
3. Override the runShrinkWrap PredicateFtor (defined in TargetMachine) to check whether shrink wrapping should run:
Shrink wrapping will run on PPC64 (Little Endian and Big Endian) unless -enable-shrink-wrap=false is specified on command line
A new test case, ppc-shrink-wrapping.ll was created based on the existing shrink wrapping tests for x86, arm, and arm64.
Phabricator review: http://reviews.llvm.org/D11817
llvm-svn: 247237
First, we need to teach isFrameOffsetLegal about STNP.
It already knew about the STP/LDP variants, but those were probably
never exercised, because it's only the load/store optimizer that
generates STP/LDP, and the only user of the method is frame lowering,
which runs earlier.
The STP/LDP cases were wrong: they didn't take into account the fact
that they return two results, not one, so the immediate offset will be
the 4th operand, not the 3rd.
Follow-up to r247234.
llvm-svn: 247236
We could go through the load/store optimizer and match STNP where
we would have matched a nontemporal-annotated STP, but that's not
reliable enough, as an opportunistic optimization.
Insetad, we can guarantee emitting STNP, by matching them at ISel.
Since there are no single-input nontemporal stores, we have to
resort to some high-bits-extracting trickery to generate an STNP
from a plain store.
Also, we need to support another, LDP/STP-specific addressing mode,
base + signed scaled 7-bit immediate offset.
For now, only match the base. Let's make it smart separately.
Part of PR24086.
llvm-svn: 247231
All of the complexity is in cleanupret, and it mostly follows the same
codepaths as catchret, except it doesn't take a return value in RAX.
This small example now compiles and executes successfully on win32:
extern "C" int printf(const char *, ...) noexcept;
struct Dtor {
~Dtor() { printf("~Dtor\n"); }
};
void has_cleanup() {
Dtor o;
throw 42;
}
int main() {
try {
has_cleanup();
} catch (int) {
printf("caught it\n");
}
}
Don't try to put the cleanup in the same function as the catch, or Bad
Things will happen.
llvm-svn: 247219
The 32-bit tables don't actually contain PC range data, so emitting them
is incredibly simple.
The 64-bit tables, on the other hand, use the same table for state
numbering as well as label ranges. This makes things more difficult, so
it will be implemented later.
llvm-svn: 247192
Summary:
This helps mostly when we use add instructions for address calculations
that contain immediates.
Reviewers: arsenm
Subscribers: arsenm, llvm-commits
Differential Revision: http://reviews.llvm.org/D12256
llvm-svn: 247157
Summary:
One of the vector splitting paths for extract_vector_elt tries to lower:
define i1 @via_stack_bug(i8 signext %idx) {
%1 = extractelement <2 x i1> <i1 false, i1 true>, i8 %idx
ret i1 %1
}
to:
define i1 @via_stack_bug(i8 signext %idx) {
%base = alloca <2 x i1>
store <2 x i1> <i1 false, i1 true>, <2 x i1>* %base
%2 = getelementptr <2 x i1>, <2 x i1>* %base, i32 %idx
%3 = load i1, i1* %2
ret i1 %3
}
However, the elements of <2 x i1> are not byte-addressible. The result of this
is that the getelementptr expands to '%base + %idx * (1 / 8)' which simplifies
to '%base + %idx * 0', and then simply '%base' causing all values of %idx to
extract element zero.
This commit fixes this by promoting the vector elements of <8-bits to i8 before
splitting the vector.
This fixes a number of test failures in pocl.
Reviewers: pekka.jaaskelainen
Subscribers: pekka.jaaskelainen, llvm-commits
Differential Revision: http://reviews.llvm.org/D12591
llvm-svn: 247128
Currently this hits an assert that extload should
always be supported, which assumes integer extloads.
This moves a hack out of SI's argument lowering and
is covered by existing tests.
llvm-svn: 247113
Typically these are catchpads, which hold data used to decide whether to
catch the exception or continue unwinding. We also shouldn't create MBBs
for catchendpads, cleanupendpads, or terminatepads, since no real code
can live in them.
This fixes a problem where MI passes (like the register allocator) would
try to put code into catchpad blocks, which are not executed by the
runtime. In the new world, blocks ending in invokes now have many
possible successors.
llvm-svn: 247102
Summary:
32-bit funclets have short prologues that allocate enough stack for the
largest call in the whole function. The runtime saves CSRs for the
funclet. It doesn't restore CSRs after we finally transfer control back
to the parent funciton via a CATCHRET, but that's a separate issue.
32-bit funclets also have to adjust the incoming EBP value, which is
what llvm.x86.seh.recoverframe does in the old model.
64-bit funclets need to spill CSRs as normal. For simplicity, this just
spills the same set of CSRs as the parent function, rather than trying
to compute different CSR sets for the parent function and each funclet.
64-bit funclets also allocate enough stack space for the largest
outgoing call frame, like 32-bit.
Reviewers: majnemer
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D12546
llvm-svn: 247092
Summary: This patch modifies X86TargetLowering::LowerVASTART so that
struct va_list is initialized with 32 bit pointers in x32. It also
includes tests that call @llvm.va_start() for x32.
Patch by João Porto
Subscribers: llvm-commits, hjl.tools
Differential Revision: http://reviews.llvm.org/D12346
llvm-svn: 247069
The old implementation assumed LP64 which is broken for x32. Specifically, the
MOVE8rm_NOREX and MOVE8mr_NOREX, when selected, would cause a 'Cannot emit
physreg copy instruction' error message to be reported.
This patch also enable the h-register*ll tests for x32.
Differential Revision: http://reviews.llvm.org/D12336
Patch by João Porto
llvm-svn: 247058
sub C, x - > add (sub 0, x), C for DS offsets.
This is mostly to fix regressions that show up when
SeparateConstOffsetFromGEP is enabled.
llvm-svn: 247054
Summary: And define them to have noop casts with address spaces 0-255.
Reviewers: pekka.jaaskelainen
Subscribers: pekka.jaaskelainen, llvm-commits
Differential Revision: http://reviews.llvm.org/D12678
llvm-svn: 246990
In searching for a fix for the underlying code-quality bug highlighted by
r246937 (that SDAG simplification can lead to us generating an ISD::OR node
with a constant zero LHS), I ran across this:
We generically canonicalize commutative binary-operation nodes in SDAG getNode
so that, if only one operand is a constant, it will be on the RHS. However, we
were doing this only after a bunch of constant-based simplification checks that
all assume this canonical form (that any constant will be on the RHS). Moving
the operand-swapping canonicalization prior to these checks seems like the
right thing to do (and, as it turns out, causes SDAG to completely fold away the
computation in test/CodeGen/ARM/2012-11-14-subs_carry.ll, just like InstCombine
would do).
llvm-svn: 246938
To commute a trivial rlwimi instructions (meaning one with a full mask and zero
shift), we'd need to ability to form an all-zero mask (instead of an all-one
mask) using rlwimi. We can't represent this, however, and we'll miscompile code
if we try.
The code quality problem that this highlights (that SDAG simplification can
lead to us generating an ISD::OR node with a constant zero LHS) will be fixed
as a follow-up.
Fixes PR24719.
llvm-svn: 246937
PPCISelDAGToDAG has a transformation that generates a rlwimi instruction from
an input pattern that looks like this:
and(or(x, c1), c2)
but the associated logic does not work if there are bits that are 1 in c1 but 0
in c2 (these are normally canonicalized away, but that can't happen if the 'or'
has other users. Make sure we abort the transformation if such bits are
discovered.
Fixes PR24704.
llvm-svn: 246900
When forming permutation-based unaligned vector loads, we need to know whether
it is valid to read ahead of the requested address by a full vector length.
Doing so is more efficient (and allows for more CSE with later loads), but
could trigger a page fault if invalid. To determine validity, we look for other
loads in the same block that access the relevant address range.
The relevant point here is that we need to do this as part of the process of
forming permutation-based vector loads, and this happens quite early in the
SDAG pipeline - specifically before many of the address calculations are fully
canonicalized. As a result, we need to try harder to recognize base+offset
address computations, because they still might appear as chain of adds
(base+offset+offset, for example). To account for this, we'll look through
chains of adds, accumulating the constant offsets.
llvm-svn: 246813
If you compute the MMO offset using unsigned arithmetic, you end up with a
large positive offset instead of a small negative one. In theory, this could
cause bad instruction-scheduling decisions later.
I noticed this by inspection from the debug output, and using that for the
regression test is the best I can do right now.
llvm-svn: 246805
JF Bastien