1. Use shuffle to insert element i1 into vector. The previous implementation was incorrect ( dest_bit OR src_bit , it doesn't clear the bit if src_bit=0 )
2. Improve shuffle i1 vector, use CVT2MASK if supported instead TRUNCATE.
Differential Revision: http://reviews.llvm.org/D23347
llvm-svn: 278623
LowerTargetConstantPool is not properly setting the TargetFlag to indicate
desired relocation. Coding error, the offset parameter was omitted, so the
TargetFlag was used as the offset, and the TargetFlag defaulted to zero.
This only affects -fpic compilation, and only those items created in a
Constant Pool, for example a vector of constants. Halide ran into this issue.
llvm-svn: 278614
Summary:
This test was resulting in asan/valgrind failures due to undefined
DWARF register mappings for WebAssembly, and was disabled in r278495.
These have been resolved.
Reviewers: sunfish, dschuff
Subscribers: bkramer, llvm-commits, jfb
Differential Revision: https://reviews.llvm.org/D23459
llvm-svn: 278576
This bring LLVM-generated PTX closer to what nvcc generates and avoids
triggering issues in ptxas.
For instance, ptxas does not accept .s16 (or .u16) registers as operands
for .fp16 instructions.
Differential Revision: https://reviews.llvm.org/D23460
llvm-svn: 278568
Trunk would try to create something like "stp x9, x8, [x0], #512", which isn't actually a valid instruction.
Differential revision: https://reviews.llvm.org/D23368
llvm-svn: 278559
Currently X86ISelLowering has a similar transformation for sexts:
sext(add_nsw(x, C)) --> add(sext(x), C_sext)
In this change I extend this code to handle zexts as well.
Reviewed By: spatel
Differential Revision: https://reviews.llvm.org/D23359
llvm-svn: 278520
...and the two followup commits:
Revert "[Sparc][Leon] Missed resetting option flags from check-in 278489."
Revert "[Sparc][Leon] Errata fixes for various errata in different
versions of the Leon variants of the Sparc 32 bit processor."
This reverts commit r274856, r278489, and r278492.
llvm-svn: 278511
The PALIGNR target shuffle decode was not taking into account that DecodePALIGNRMask (rather oddly) expects the operands to be in reverse order, nor was it detecting unary patterns, causing combines to combine with the incorrect input.
The cgbuiltin, auto upgrade and instruction comments code correctly swap the operands so are not affected.
llvm-svn: 278494
The nature of the errata are listed in the comments preceding the errata fix passes. Relevant unit tests are implemented for each of these.
These changes update older versions of these errata fixes with improvements to code and unit tests.
Differential Revision: https://reviews.llvm.org/D21960
llvm-svn: 278489
"insert_subreg, subreg_to_reg, and reg_sequence" instructions' after
adjusting some unittest checks.
This is to solve PR28852. The restriction was added at 2010 to make better register
coalescing. We assumed that it was not necessary any more. Testing results on x86
supported the assumption.
We will look closely to any performance impact it will bring and will be prepared
to help analyzing performance problem found on other architectures.
Differential Revision: https://reviews.llvm.org/D23210
llvm-svn: 278466
To fix PR28014, this patch restricts tail merging to blocks that belong to the
same loop after MBP.
Differential Revision: https://reviews.llvm.org/D23191
llvm-svn: 278463
This helped to improved memory-folding and register coalescing optimizations.
Also, this patch fixed the tracker #17229.
Reviewer: Craig Topper.
Differential Revision: https://reviews.llvm.org/D23108
llvm-svn: 278431
It's sharing the integer G_CONSTANT for now since I don't *think* it creates
any ambiguity (even on weird archs). If that turns out wrong we can create a
G_PTRCONSTANT or something.
llvm-svn: 278423
subreg_to_reg, and reg_sequence" instructions.
This is to solve PR28852. The restriction was added at 2010 to make better register
coalescing. We assumed that it was not necessary any more. Testing results on x86
supported the assumption.
We will look closely to any performance impact it will bring and will be prepared
to help analyzing performance problem found on other architectures.
Differential Revision: https://reviews.llvm.org/D23210
llvm-svn: 278384
From the point of view of register assignment, byval parameters are
ignored: a byval parameter is not going to be assigned to a register,
and it will not affect the assignments of subsequent parameters.
When matching registers with parameters in the bit tracker, make sure
to skip byval parameters before advancing the registers.
llvm-svn: 278375
Summary: Some backends, like WebAssembly, use virtual registers instead of physical registers. This crashes the DbgValueHistoryCalculator pass, which assumes that all registers are physical. Instead, skip virtual registers when iterating aliases, and assume that they are clobbered.
Reviewers: dexonsmith, dschuff, aprantl
Subscribers: yurydelendik, llvm-commits, jfb, sunfish
Differential Revision: https://reviews.llvm.org/D22590
llvm-svn: 278371
The previous implementation (not custom) doesn't enforce zeroing off upper bits. The assumption is that i1 PRODUCER (truncate and extractelement) must zero all upper bits, so i1 CONSUMER instructions ( test, zext, save, etc) can be done without additional zeroing.
Make extractelement i1 lowering custom for all vector i1.
Differential Revision: http://reviews.llvm.org/D23246
llvm-svn: 278328
This patch helps avoid false dependencies on undef registers by updating the machine instructions' undef operand to use a register that the instruction is truly dependent on, or use a register with clearance higher than Pref.
Pseudo example:
loop:
xmm0 = ...
xmm1 = vcvtsi2sdl eax, xmm0<undef>
... = inst xmm0
jmp loop
In this example, selecting xmm0 as the undef register creates false dependency between loop iterations.
This false dependency cannot be solved by inserting an xor before vcvtsi2sdl because xmm0 is alive at the point of the vcvtsi2sdl instruction.
Selecting a different register instead of xmm0, especially a register that is not used in the loop, will eliminate this problem.
Differential Revision: https://reviews.llvm.org/D22466
llvm-svn: 278321
It's more than just inttoptr, but the others can't be tested until we have
support for non-trivial constants (they currently get unavoidably folded to a
ConstantInt).
llvm-svn: 278303
Summary:
This patch define and implement amdgcn image intrinsics with sampler.
1. define vdata type to be llvm_anyfloat_ty, address type to be llvm_anyfloat_ty,
and rsrc type to be llvm_anyint_ty. As a result, we expect the intrinsics name
to have three suffixes to overload each of these three types;
2. D128 as well as two other flags are implied in the three types, for example,
if you use v8i32 as resource type, then r128 is 0!
3. don't expose TFE flag, and other flags are exposed in the instruction order:
unrm, glc, slc, lwe and da.
Differential Revision: http://reviews.llvm.org/D22838
Reviewed by:
arsenm and tstellarAMD
llvm-svn: 278291
If AnalyzeBranch can't analyze a block and it is possible to
fallthrough, then duplicating the block doesn't make sense, as only one
block can be the layout predecessor for the un-analyzable fallthrough.
Submitted wit a test case, but NOTE: the test case doesn't currently
fail. However, the test case fails with D20505 and would have saved me
some time debugging.
llvm-svn: 278288
Insert before the skip branch if one is created.
This is a somewhat more natural placement relative
to the skip branches, and makes it possible to implement
analyzeBranch for skip blocks.
The test changes are mostly due to a quirk where
the block label is not emitted if there is a terminator
that is not also a branch.
llvm-svn: 278273
Summary:
See the new test case for one that was (non-deterministically) crashing
on trunk and deterministically hit the assertion that I added in D23302.
Basically, the machine function contains a sequence
DS_WRITE_B32 %vreg4, %vreg14:sub0, ...
DS_WRITE_B32 %vreg4, %vreg14:sub0, ...
%vreg14:sub1<def> = COPY %vreg14:sub0
and SILoadStoreOptimizer::mergeWrite2Pair merges the two DS_WRITE_B32
instructions into one before calling repairIntervalsInRange.
Now repairIntervalsInRange wants to repair %vreg14, in particular, and
ends up trying to repair %vreg14:sub1 as well, but that only becomes
active _after_ the range that is to be repaired, hence the crash due
to LR.find(...) == LR.begin() at the start of repairOldRegInRange.
I believe that just skipping those subrange is fine, but again, not too
familiar with that code.
Reviewers: MatzeB, kparzysz, tstellarAMD
Subscribers: llvm-commits, MatzeB
Differential Revision: https://reviews.llvm.org/D23303
llvm-svn: 278268
This change makes it possible for tail-duplication and tail-merging to
be disjoint. By being less aggressive when merging during layout, there are no
overlapping cases between tail-duplication and tail-merging, provided the
thresholds are disjoint.
There is a remaining TODO to benchmark the succ_size() test for non-layout tail
merging.
llvm-svn: 278265
If the value produced by the bitcast hasn't been referenced yet, we can simply
reuse the input register avoiding an unnecessary COPY instruction.
llvm-svn: 278245
Floating point instructions use general purpose registers, so the few
instructions that can put floating point immediates into registers are,
in fact, integer instruction. Use them explicitly instead of having
pseudo-instructions specifically for dealing with floating point values.
Simplify the constant loading instructions (from sdata) to have only two:
one for 32-bit values and one for 64-bit values: CONST32 and CONST64.
llvm-svn: 278244
If the input vector to INSERT_SUBVECTOR is another INSERT_SUBVECTOR, and this inserted subvector replaces the last insertion, then insert into the common source vector.
i.e.
INSERT_SUBVECTOR( INSERT_SUBVECTOR( Vec, SubOld, Idx ), SubNew, Idx ) --> INSERT_SUBVECTOR( Vec, SubNew, Idx )
Differential Revision: https://reviews.llvm.org/D23330
llvm-svn: 278211
Created a Thumb2 predicated pattern matcher that uses Thumb2 and
HasT2ExtractPack and used it to redefine the patterns for sxta{b|h}
and uxta{b|h}. Also used the similar patterns to fill in isel pattern
gaps for the corresponding instructions in the ARM backend.
The patch is mainly changes to tests since most of this functionality
appears not to have been tested.
Differential Revision: https://reviews.llvm.org/D23273
llvm-svn: 278207
This patch adds -emscripten-cxx-exceptions-whitelist option to
WebAssemblyLowerEmscriptenExceptions pass. This options is the list of
function names in which Emscripten-style exception handling is enabled.
This is to support emscripten's EXCEPTION_CATCHING_WHITELIST which
exists because of the performance impact of emscripten's non-zero-cost
EH method.
Patch by Heejin Ahn
Differential Revision: https://reviews.llvm.org/D23292
llvm-svn: 278171
For now put them all in the entry block. This should be correct but may give
poor runtime performance. Hopefully MachineSinking combined with
isReMaterializable can solve those issues, but if not the interface is sound
enough to support alternatives.
llvm-svn: 278168
A UD2 might make its way into the program via a call to @llvm.trap.
Obviously, calls are not terminators. However, we modeled the X86
instruction, UD2, as a terminator. Later on, this confuses the epilogue
insertion machinery which results in the epilogue getting inserted
before the UD2. For some platforms, like x64, the result is a
violation of the ABI.
Instead, model UD2/UD2B as a side effecting instruction which may
observe memory.
llvm-svn: 278144
As detailed on D22726, much of the shift combining code assume constant values will fit into a uint64_t value and calls ConstantSDNode::getZExtValue where it probably shouldn't (leading to asserts). Using APInt directly avoids this problem but we encounter other assertions if we attempt to compare/operate on 2 APInt of different bitwidths.
This patch adds a helper function to ensure that 2 APInt values are zero extended as required so that they can be safely used together. I've only added an initial example use for this to the '(SHIFT (SHIFT x, c1), c2) --> (SHIFT x, (ADD c1, c2))' combines. Further cases can easily be added as required.
Differential Revision: https://reviews.llvm.org/D23007
llvm-svn: 278141
This makes a trivial change in the emission of the per-function XRay
tables, and makes sure that the xray_instr_map section does show up in
the object file.
llvm-svn: 278113
We only had partial memory folding support for the intrinsic definitions, and (as noted on PR27481) was causing FR32/FR64/VR128 mismatch errors with the machine verifier.
This patch adds missing memory folding support for both intrinsics and the ffloor/fnearbyint/fceil/frint/ftrunc patterns and in doing so fixes the failing machine verifier stack folding tests from PR27481.
Differential Revision: https://reviews.llvm.org/D23276
llvm-svn: 278106
* Delete extra '_' prefixes from JS library function names. fixImports()
function in JS glue code deals with this for wasm.
* Change command-line option names in order to be consistent with
asm.js.
* Add missing lowering code for llvm.eh.typeid.for intrinsics
* Delete commas in mangled function names
* Fix a function argument attributes bug. Because we add the pointer to
the original callee as the first argument of invoke wrapper, all
argument attribute indices have to be incremented by one.
Patch by Heejin Ahn
Differential Revision: https://reviews.llvm.org/D23258
llvm-svn: 278081
This reverts commit r278048. Something changed between the last time I
built this--it takes awhile on my ridiculously slow and ancient
computer--and now that broke this.
llvm-svn: 278053
Summary:
Based on two patches by Michael Mueller.
This is a target attribute that causes a function marked with it to be
emitted as "hotpatchable". This particular mechanism was originally
devised by Microsoft for patching their binaries (which they are
constantly updating to stay ahead of crackers, script kiddies, and other
ne'er-do-wells on the Internet), but is now commonly abused by Windows
programs to hook API functions.
This mechanism is target-specific. For x86, a two-byte no-op instruction
is emitted at the function's entry point; the entry point must be
immediately preceded by 64 (32-bit) or 128 (64-bit) bytes of padding.
This padding is where the patch code is written. The two byte no-op is
then overwritten with a short jump into this code. The no-op is usually
a `movl %edi, %edi` instruction; this is used as a magic value
indicating that this is a hotpatchable function.
Reviewers: majnemer, sanjoy, rnk
Subscribers: dberris, llvm-commits
Differential Revision: https://reviews.llvm.org/D19908
llvm-svn: 278048
This patch adds support for some new relocation models to the ARM
backend:
* Read-only position independence (ROPI): Code and read-only data is accessed
PC-relative. The offsets between all code and RO data sections are known at
static link time. This does not affect read-write data.
* Read-write position independence (RWPI): Read-write data is accessed relative
to the static base register (r9). The offsets between all writeable data
sections are known at static link time. This does not affect read-only data.
These two modes are independent (they specify how different objects
should be addressed), so they can be used individually or together. They
are otherwise the same as the "static" relocation model, and are not
compatible with SysV-style PIC using a global offset table.
These modes are normally used by bare-metal systems or systems with
small real-time operating systems. They are designed to avoid the need
for a dynamic linker, the only initialisation required is setting r9 to
an appropriate value for RWPI code.
I have only added support to SelectionDAG, not FastISel, because
FastISel is currently disabled for bare-metal targets where these modes
would be used.
Differential Revision: https://reviews.llvm.org/D23195
llvm-svn: 278015
Summary:
The DAG combine transformation that was generating the
aarch64_neon_vcvtfp2fxs node was assuming that all
inputs where legal and wasn't accounting that the input
could be a v4f64 if we're trying to do the transformation
before legalization. We now bail out in this case.
All illegal types besides v4f64 were already rejected.
Fixes https://llvm.org/bugs/show_bug.cgi?id=28877.
Reviewers: jmolloy
Subscribers: aemerson, rengolin, llvm-commits
Differential Revision: https://reviews.llvm.org/D23261
llvm-svn: 278002
Optimized lowering of BITCAST node. The BITCAST node can be replaced with COPY_TO_REG instead of KMOV.
It allows to suppress two opposite BITCAST operations and avoid redundant "movs".
Differential Revision: https://reviews.llvm.org/D23247
llvm-svn: 277958
This is a new test that should explore a current suboptimal sequence in passing values between cmp and kor intrinsics.
The code will be optimized in an upcoming patch.
Submitted bug here:
https://llvm.org/bugs/show_bug.cgi?id=28839
llvm-svn: 277954
Assuming SSE2 is available then we can safely commute between these, removing some unnecessary register moves and improving memory folding opportunities.
VEX encoded versions don't benefit so I haven't added support to them.
llvm-svn: 277930
Summary:
This is the setting of the Vulkan closed source driver.
It decreases the max wave count from 10 to 8.
26010 shaders in 14650 tests
Totals:
VGPRS: 829593 -> 808440 (-2.55 %)
Spilled SGPRs: 81878 -> 42226 (-48.43 %)
Spilled VGPRs: 367 -> 358 (-2.45 %)
Scratch VGPRs: 1764 -> 1748 (-0.91 %) dwords per thread
Code Size: 36677864 -> 35923932 (-2.06 %) bytes
There is a massive decrease in SGPR spilling in general and -7.4% spilled
VGPRs for DiRT Showdown (= SGPRs spilled to scratch?)
Reviewers: arsenm, tstellarAMD, nhaehnle
Subscribers: arsenm, llvm-commits, kzhuravl
Differential Revision: https://reviews.llvm.org/D23034
llvm-svn: 277867
Summary: Thumb2 supports encoding immediates with specific patterns into mov.w by splatting the low 8 bits into other bytes.
I'm resubmitting this patch. The test case in the original commit
r277610 does not specify triple, so builds with differnt default triple
will have different output.
This patch fixed trile as thumb-darwin-apple.
Reviewers: john.brawn, jmolloy, bruno
Subscribers: jmolloy, aemerson, rengolin, samparker, llvm-commits
Differential Revision: https://reviews.llvm.org/D23090
llvm-svn: 277865
There were two locations where fast-isel would generate a LFD instruction
with a target register class VSFRC instead of F8RC when VSX was enabled.
This can ccause invalid registers to be used in certain cases, like:
lfd 36, ...
instead of using a VSX load instruction. The wrong register number gets
silently truncated, causing invalid code to be generated.
The first place is PPCFastISel::PPCEmitLoad, which had multiple problems:
1.) The IsVSSRC and IsVSFRC flags are not initialized correctly, since they
are computed from resultReg, which is still zero at this point in many cases.
Fixed by changing the helper routines to operate on a register class instead
of a register and passing in UseRC.
2.) Even with this fixed, Is64VSXLoad is still wrong due to a typo:
bool Is32VSXLoad = IsVSSRC && Opc == PPC::LFS;
bool Is64VSXLoad = IsVSSRC && Opc == PPC::LFD;
The second line needs to use isVSFRC (like PPCEmitStore does).
3.) Once both the above are fixed, we're now generating a VSX instruction --
but an incorrect one, since generation of an indexed instruction with null
index is wrong. Fixed by copying the code handling the same issue in
PPCEmitStore.
The second place is PPCFastISel::PPCMaterializeFP, where we would emit an
LFD to load a constant from the literal pool, and use the wrong result
register class. Fixed by hardcoding a F8RC class even on systems
supporting VSX.
Fixes: https://llvm.org/bugs/show_bug.cgi?id=28630
Differential Revision: https://reviews.llvm.org/D22632
llvm-svn: 277823
This patch fixes passing long double type arguments to function in
soft float mode. If there is less than 4 argument registers free
(long double type is mapped in 4 gpr registers in soft float mode)
long double type argument must be passed through stack.
Differential Revision: https://reviews.llvm.org/D20114.
llvm-svn: 277804
These are the operations that are trivially identical. Division is omitted for
now because you need to use the correct sign/zero extension.
llvm-svn: 277775
Adding missing tests for OCL type names for half, float, double, char, short, long, and unknown.
Patch by Aaron En Ye Shi.
Differential Revision: https://reviews.llvm.org/D22964
llvm-svn: 277759
Previously, FastISel for WebAssembly wasn't checking the return value of
`getRegForValue` in certain cases, which would generate instructions
referencing NoReg. This patch fixes this behavior.
Patch by Dominic Chen
Differential Revision: https://reviews.llvm.org/D23100
llvm-svn: 277742
On modern Intel processors hardware SQRT in many cases is faster than RSQRT
followed by Newton-Raphson refinement. The patch introduces a simple heuristic
to choose between hardware SQRT instruction and Newton-Raphson software
estimation.
The patch treats scalars and vectors differently. The heuristic is that for
scalars the compiler should optimize for latency while for vectors it should
optimize for throughput. It is based on the assumption that throughput bound
code is likely to be vectorized.
Basically, the patch disables scalar NR for big cores and disables NR completely
for Skylake. Firstly, scalar SQRT has shorter latency than NR code in big cores.
Secondly, vector SQRT has been greatly improved in Skylake and has better
throughput compared to NR.
Differential Revision: https://reviews.llvm.org/D21379
llvm-svn: 277725
Enable tail calls by default for (micro)MIPS(64).
microMIPS is slightly more tricky than doing it for MIPS(R6) or microMIPSR6.
microMIPS has two instruction encodings: 16bit and 32bit along with some
restrictions on the size of the instruction that can fill the delay slot.
For safe tail calls for microMIPS, the delay slot filler attempts to find
a correct size instruction for the delay slot of TAILCALL pseudos.
Reviewers: dsanders, vkalintris
Subscribers: jfb, dsanders, sdardis, llvm-commits
Differential Revision: https://reviews.llvm.org/D21138
llvm-svn: 277708
This should ensure that we can atomically write two bytes (on top of the
retq and the one past it) and have those two bytes not straddle cache
lines.
We also move the label past the alignment instruction so that we can refer
to the actual first instruction, as opposed to potential padding before the
aligned instruction.
Update the tests to allow us to reflect the new order of assembly.
Reviewers: rSerge, echristo, majnemer
Subscribers: llvm-commits, mehdi_amini
Differential Revision: https://reviews.llvm.org/D23101
llvm-svn: 277701
This just tests that the register limit isn't exceeded,
so the regisetr allocation doesn't need to be great.'
The critically slow part is all in greedy RA, so
switch to basic.
llvm-svn: 277700
rewriteOperands() always performed liveness queries at the base index
rather than the RegSlot/Base as apropriate for the machine operand. This
could lead to illegal rewriting in some cases.
llvm-svn: 277661
This patch fixes pr25548.
Current implementation of PPCBoolRetToInt doesn't handle CallInst correctly, so it failed to do the intended optimization when there is a CallInst with parameters. This patch fixed that.
llvm-svn: 277655
We currently only support combining target shuffles that consist of a single source input (plus elements known to be undef/zero).
This patch generalizes the recursive combining of the target shuffle to collect all the inputs, merging any duplicates along the way, into a full set of src ops and its shuffle mask.
We uncover a number of cases where we have failed to combine a unary shuffle because the input has been duplicated and separated during lowering.
This will allow us to combine to 2-input shuffles in a future patch.
Differential Revision: https://reviews.llvm.org/D22859
llvm-svn: 277631
Currently we have a number of tests that fail with -verify-machineinstrs.
To detect this cases earlier we add the option to the testcases with the
exception of tests that will currently fail with this option. PR 27456 keeps
track of this failures.
No code review, as discussed with Hal Finkel.
llvm-svn: 277624
Summary: Thumb2 supports encoding immediates with specific patterns into mov.w by splatting the low 8 bits into other bytes.
Reviewers: john.brawn, jmolloy
Subscribers: jmolloy, aemerson, rengolin, samparker, llvm-commits
Differential Revision: https://reviews.llvm.org/D23090
llvm-svn: 277610
When expanding FP constants, we attempt to shrink doubles to floats and perform an extending load.
However, on SystemZ, and possibly on other targets (I've only confirmed the problem on SystemZ), the FP extending load instruction may convert SNaN into QNaN, or may cause an exception. So in the general case, we would still like to shrink FP constants, but SNaNs should be left as doubles.
Differential Revision: https://reviews.llvm.org/D22685
llvm-svn: 277602
When the same base address is used to load two different data types, LSR
would assume a memory type of "void". This type is not sized and has no
alignment information. Checking for it causes a crash.
llvm-svn: 277601
Summary:
We also add a test to show what currently happens when we create a
section per function and emit an xray_instr_map. This illustrates the
relationship (or lack thereof) between the per-function section and the
xray_instr_map section.
We also change the code generation slightly so that we don't always
create group sections, but rather only do so if a function where the
table is associated with is in a group.
Also in this change:
- Remove the "merge" flag on the xray_instr_map section.
- Test that we're generating the right table for comdat and non-comdat functions.
Reviewers: echristo, majnemer
Subscribers: llvm-commits, mehdi_amini
Differential Revision: https://reviews.llvm.org/D23104
llvm-svn: 277580
In this particular example we wouldn't want the smmls anyway (the value is
actually unused), but in general smmls does not provide the required flags
register so if that SUBE result is used we can't replace it.
llvm-svn: 277541
We were relying on the misleadingly-names $status result to actually be the
status. Actually it's just a scratch register that may or may not be valid (and
is the inverse of the real ststus anyway). Success can be determined by
comparing the value loaded against the one we wanted to see for "cmpxchg
strong" loops like this.
Should fix PR28819.
llvm-svn: 277513
Summary:
Two types of stores are possible in pixel shaders: stores to memory that are
explicitly requested at the API level, and stores that are an implementation
detail of register spilling or lowering of arrays.
For the first kind of store, we must ensure that helper pixels have no effect
and hence WQM must be disabled. The second kind of store must always be
executed, because the written value may be loaded again in a way that is
relevant for helper pixels as well -- and there are no externally visible
effects anyway.
This is a candidate for the 3.9 release branch.
Reviewers: arsenm, tstellarAMD, mareko
Subscribers: arsenm, kzhuravl, llvm-commits
Differential Revision: https://reviews.llvm.org/D22675
llvm-svn: 277504
Summary:
There are cases where uniform branch conditions are computed in VGPRs, and
we didn't correctly mark those as WQM.
The stray change in basic-branch.ll is because invoking the LiveIntervals
analysis leads to the detection of a dead register that would otherwise not
be seen at -O0.
This is a candidate for the 3.9 branch, as it fixes a possible hang.
Reviewers: arsenm, tstellarAMD, mareko
Subscribers: arsenm, llvm-commits, kzhuravl
Differential Revision: https://reviews.llvm.org/D22673
llvm-svn: 277500
Identify patterns where the address is aligned to an 8-byte boundary,
but both the base address and the constant offset are both proper
multiples of 4. In such cases, extract Base+4 into a separate instruc-
tion, and use S2_storerd_io, instead of using S4_storerd_rr.
llvm-svn: 277497
I thought the directory had a lit.local.cfg, but it doesn't.
I'll add one, but for now, add the REQUIRES line. While there,
move the triple into the IR and add a datalayout.
llvm-svn: 277486
None of GlobalISel requires the property, but this lets us use the
verifier instead of rolling our own "all instructions selected" check.
llvm-svn: 277484
After instruction selection, there should be no pre-isel generic
instructions remaining, nor should generic virtual registers be
used. Verify that.
llvm-svn: 277483
Selected: the InstructionSelect pass ran and all pre-isel generic
instructions have been eliminated; i.e., all instructions are now
target-specific or non-pre-isel generic instructions (e.g., COPY).
Since only pre-isel generic instructions can have generic virtual register
operands, this also means that all generic virtual registers have been
constrained to virtual registers (assigned to register classes) and that
all sizes attached to them have been eliminated.
This lets us enforce certain invariants across passes.
This property is GlobalISel-specific, but is always available.
llvm-svn: 277482
The InstructionSelect pass assumes that RegBankSelect ran; set the
property on all tests (thereby verifying the test inputs) and require
it in the pass.
llvm-svn: 277477
RegBankSelected: the RegBankSelect pass ran and all generic virtual
registers have been assigned to a register bank.
This lets us enforce certain invariants across passes.
This property is GlobalISel-specific, but is always available.
llvm-svn: 277475
RegBankSelect and InstructionSelect run after the legalizer and
require a Legalized function: check that all instructions are legal.
Note that this should be in the MachineVerifier, but it can't use the
MachineLegalizer as it's currently in the separate GlobalISel library.
Note that the RegBankSelect verifier checks have the same layering
problem, but we only use inline methods so end up not needing to link
against the GlobalISel library.
llvm-svn: 277472
Legalized: The MachineLegalizer ran; all pre-isel generic instructions
have been legalized, i.e., all instructions are now one of:
- generic and always legal (e.g., COPY)
- target-specific
- legal pre-isel generic instructions.
This lets us enforce certain invariants across passes.
This property is GlobalISel-specific, but is always available.
llvm-svn: 277470
Added (sra (shl x, 16), 16) to the sext_16_node PatLeaf for ARM to
simplify some pattern matching. This has allowed several patterns
for smul* and smla* to be removed as well as making it easier to add
the matching for the corresponding instructions for Thumb2 targets.
Also added two Pat classes that are predicated on Thumb2 with the
hasDSP flag and UseMulOps flags. Updated the smul codegen test with
the wider range of patterns plus the ThumbV6 and ThumbV6T2 targets.
Differential Revision: https://reviews.llvm.org/D22908
llvm-svn: 277450
The branch relaxation pass has the worst test coverage
of any pass in AArch64. Add a few tests that hit some
large pieces of code in the pass.
llvm-svn: 277428
Summary: This patch implements CFI for WebAssembly. It modifies the
LowerTypeTest pass to pre-assign table indexes to functions that are
called indirectly, and lowers type checks to test against the
appropriate table indexes. It also modifies the WebAssembly backend to
support a special ".indidx" assembly directive that propagates the table
index assignments out to the linker.
Patch by Dominic Chen
Differential Revision: https://reviews.llvm.org/D21768
llvm-svn: 277398
Summary: This patch includes asm.js-style exception handling support for
WebAssembly. The WebAssembly MVP does not have any support for
unwinding or non-local control flow. In order to support C++ exceptions,
emscripten currently uses JavaScript exceptions along with some support
code (written in JavaScript) that is bundled by emscripten with the
generated code.
This scheme lowers exception-related instructions for wasm such that
wasm modules can be compatible with emscripten's existing scheme and
share the support code.
Patch by Heejin Ahn
Differential Revision: https://reviews.llvm.org/D22958
llvm-svn: 277391
We used to combine "sext(setcc x, y, cc) -> (select (setcc x, y, cc), -1, 0)"
Instead, we should combine to (select (setcc x, y, cc), T, 0) where the value
of T is 1 or -1, depending on the type of the setcc, and getBooleanContents()
for the type if it is not i1.
This fixes PR28504.
llvm-svn: 277371
Scavenging slots were only reserved when pseudo-instruction expansion in
frame lowering created new virtual registers. It is possible to still
need a scavenging slot even if no virtual registers were created, in cases
where the stack is large enough to overflow instruction offsets.
llvm-svn: 277355
Summary:
Allocating an AFGR64 shadows two GPR32's instead of just one.
This fixes an LNT regression detected by our internal buildbots.
Reviewers: sdardis
Subscribers: dsanders, sdardis, llvm-commits
Differential Revision: https://reviews.llvm.org/D23012
llvm-svn: 277348
The branch relaxation pass is computing the wrong offsets because it assumes
TLSDESC_CALLSEQ eats up 4 bytes, when in fact it is lowered to an instruction
sequence taking up 16 bytes. This can become a problem in huge files with lots
of TLS accesses, as it may slowly move branch targets out of the range computed
by the branch relaxation pass.
Fixes PR24234 https://llvm.org/bugs/show_bug.cgi?id=24234
Differential Revision: https://reviews.llvm.org/D22870
llvm-svn: 277331
Initialize all AArch64-specific passes in the TargetMachine so they can be run
by llc. This can lead to conflicts in opt with some command line options that
share the same name as the pass, so I took this opportunity to do some cleanups:
* rename all relevant command line options from "aarch64-blah" to
"aarch64-enable-blah" and update the tests accordingly
* run clang-format on their declarations
* move all these declarations to a common place (the TargetMachine) as opposed
to having them scattered around (AArch64BranchRelaxation and
AArch64AddressTypePromotion were the only offenders)
llvm-svn: 277322
As discussed on PR14593, this patch adds support for lowering to SHLD/SHRD from the patterns generated by DAGTypeLegalizer::ExpandShiftWithKnownAmountBit.
Differential Revision: https://reviews.llvm.org/D23000
llvm-svn: 277299
Summary:
When performing cmp for EQ/NE and the operand is sign extended, we can
avoid the truncaton if the bits to be tested are no less than origianl
bits.
Reviewers: eli.friedman
Subscribers: eli.friedman, aemerson, nemanjai, t.p.northover, llvm-commits
Differential Revision: https://reviews.llvm.org/D22933
llvm-svn: 277252
These come in two variants for now: G_INTRINSIC and G_INTRINSIC_W_SIDE_EFFECTS.
We may decide to split the latter up with finer-grained restrictions later, if
necessary.
llvm-svn: 277224
Up until now, we only had code to match PSADBW patterns that look like what
comes out of the loop vectorizer - a partial reduction inside the loop body
that gets fed into a horizontal operation in a different basic block.
This adds support for straight-line patterns, like those generated by the
SLP vectorizer.
Differential Revision: https://reviews.llvm.org/D22889
llvm-svn: 277219
Support for lowering to VBROADCASTF128 etc. in D22460 was not correctly ensuring that the only users of the 128-bit vector load were the insertions of the vector into the lower/upper subvectors.
llvm-svn: 277214
Patch by Sunita Marathe
Third try, now following fixes to MSan to handle mempcy in such a way that this commit won't break the MSan buildbots. (Thanks, Evegenii!)
llvm-svn: 277189
The following pattern was being layed out poorly:
A
/ \
B C
/ \ / \
D E ? (Doesn't matter)
Where A->B is far more likely than A->C, and prob(B->D) = prob(B->E)
The current algorithm gives:
A,B,C,E (D goes on worklist)
It does this even if C has a frequency count of 0. This patch
adjusts the layout calculation so that if freq(B->E) >> freq(C->E)
then we go ahead and layout E rather than C. Fallthrough half the time
is better than fallthrough never, or fallthrough very rarely. The
resulting layout is:
A,B,E, (C and D are in a worklist)
llvm-svn: 277187
The DAG combiner will try to merge consecutive stores into a bigger
store, unless the resulting store is not fast. Misaligned vector stores
are allowed on Hexagon, but are not fast. Add a testcase to make sure
this type of merging does not occur.
Patch by Pranav Bhandarkar.
llvm-svn: 277182
The DAG combiner tries to merge stores to adjacent vector wide memory
locations by creating stores which are integral multiples of the vector
width. Discourage this by informing it that this is slow. This should
not affect legalization passes, because all of them ignore the "Fast"
argument.
Patch by Pranav Bhandarkar.
llvm-svn: 277178
Mostly straightforward as we ignore addressing modes and just
use the base + unsigned immediate offset (always 0) variants.
This currently fails to select extloads because we have yet to
agree on a representation.
llvm-svn: 277171
Software pipelining is an optimization for improving ILP by
overlapping loop iterations. Swing Modulo Scheduling (SMS) is
an implementation of software pipelining that attempts to
reduce register pressure and generate efficient pipelines with
a low compile-time cost.
This implementaion of SMS is a target-independent back-end pass.
When enabled, the pass should run just prior to the register
allocation pass, while the machine IR is in SSA form. If the pass
is successful, then the original loop is replaced by the optimized
loop. The optimized loop contains one or more prolog blocks, the
pipelined kernel, and one or more epilog blocks.
This pass is enabled for Hexagon only. To enable for other targets,
a couple of target specific hooks must be implemented, and the
pass needs to be called from the target's TargetMachine
implementation.
Differential Review: http://reviews.llvm.org/D16829
llvm-svn: 277169
If the mask of a vector shuffle has alternating odd or even numbers
starting with 1 or 0 respectively up to the largest possible index
for the given type in the given HVX mode (single of double) we can
generate vpacko or vpacke instruction respectively.
E.g.
%42 = shufflevector <32 x i16> %37, <32 x i16> %41,
<32 x i32> <i32 1, i32 3, ..., i32 63>
is %42.h = vpacko(%41.w, %37.w)
Patch by Pranav Bhandarkar.
llvm-svn: 277168
Rebalances address calculation trees and applies Hexagon-specific
optimizations to the trees to improve instruction selection.
Patch by Tobias Edler von Koch.
llvm-svn: 277151
Summary:
Implements fastLowerArguments() to avoid the need to fall back on
SelectionDAG for 0-4 argument functions that don't do tricky things like
passing double in a pair of i32's.
This allows us to move all except one test to -fast-isel-abort=3. The
remaining one has function prototypes of the form 'i32 (i32, double, double)'
which requires floats to be passed in GPR's.
The previous commit had an uninitialized variable that caused the incoming
argument region to have undefined size. This has been fixed.
Reviewers: sdardis
Subscribers: dsanders, llvm-commits, sdardis
Differential Revision: https://reviews.llvm.org/D22680
llvm-svn: 277136
We currently default to using either generic shuffles or MASK+PACKUS/PACKSS to truncate all integer vectors. For vector comparisons, we know that the result will be either all or zero bits in every element, which can be efficiently truncated by directly using PACKSS to repeatedly halve the size of each element.
Due to the limited input values (-1 or 0) we don't need to account for vector element size, so for simplicity we just use the PACKSS(vXi16,vXi16) implementation in all cases. Additionally for AVX2 PACKSS of 256bit data we must perform a PERMQ shuffle to reorder the data into the correct order. I did investigate performing a single shuffle after all the PACKSS calls but the need to cross 128bit lanes makes this difficult to achieve efficiently.
We avoid performing this on AVX512 as it should have better alternative truncation instructions.
Differential Revision: https://reviews.llvm.org/D22814
llvm-svn: 277132
Summary:
The MOV/MOVT instructions being chosen for struct_byval predicates was
conditional only on Thumb2, resulting in an ARM MOV/MOVT instruction
being incorrectly emitted in Thumb1 mode. This is especially apparent
with v8-m.base targets. This patch ensures that Thumb instructions are
emitted in both Thumb modes.
Reviewers: rengolin, t.p.northover
Subscribers: llvm-commits, aemerson, rengolin
Differential Revision: https://reviews.llvm.org/D22865
llvm-svn: 277128
I'm not convinced the patterns for the rm_Int was correct anyway. It had a tied source that should't exist for the unmasked version. The load form of MOVSS always zeros the most significant bits. I've left the patterns off the masked load instructions as I'm not sure what the correct pattern should be and we don't have any tests currently. Nor do we implement masked scalar load intrinsics in clang currently.
llvm-svn: 277098
Normally, CFI instructions should be inserted after allocframe, but
if allocframe is in the same packet with a call, the CFI instructions
should be inserted before that packet.
llvm-svn: 277020
LLT() has a particular meaning: it's one invalid type. But we really
want selected instructions to have no type whatsoever.
Also verify that types don't linger after ISel, and enable the verifier
on the AArch64 select test.
llvm-svn: 277001
Summary:
Implements fastLowerArguments() to avoid the need to fall back on
SelectionDAG for 0-4 argument functions that don't do tricky things like
passing double in a pair of i32's.
This allows us to move all except one test to -fast-isel-abort=3. The
remaining one has function prototypes of the form 'i32 (i32, double, double)'
which requires floats to be passed in GPR's.
Reviewers: sdardis
Subscribers: dsanders, llvm-commits, sdardis
Differential Revision: https://reviews.llvm.org/D22680
llvm-svn: 276982
Summary:
SI_ELSE is lowered into two parts:
s_or_saveexec_b64 dst, src (at the start of the basic block)
s_xor_b64 exec, exec, dst (at the end of the basic block)
The idea is that dst contains the exec mask of the preceding IF block. It can
happen that SIWholeQuadMode decides to switch from WQM to Exact mode inside
the basic block that contains SI_ELSE, in which case it introduces an instruction
s_and_b64 exec, exec, s[...]
which masks out bits that can correspond to both the IF and the ELSE paths.
So the resulting sequence must be:
s_or_savexec_b64 dst, src
s_and_b64 exec, exec, s[...] <-- added by SIWholeQuadMode
s_and_b64 dst, dst, exec <-- added by SILowerControlFlow
s_xor_b64 exec, exec, dst
Whether to add the additional s_and_b64 dst, dst, exec is currently determined
via the ExecModified tracking. With this change, it is instead determined by
an additional flag on SI_ELSE which is set by SIWholeQuadMode.
Finally: It also occured to me that an alternative approach for the long run
is for SILowerControlFlow to unconditionally emit
s_or_saveexec_b64 dst, src
...
s_and_b64 dst, dst, exec
s_xor_b64 exec, exec, dst
and have a pass that detects and cleans up the "redundant AND with exec"
pattern where possible. This could be useful anyway, because we also add
instructions
s_and_b64 vcc, exec, vcc
before s_cbranch_scc (in moveToALU), and those are often redundant. I have
some pending changes to how KILL is lowered that could also benefit from
such a cleanup pass.
In any case, this current patch could help in the short term with the whole
ExecModified business.
Reviewers: tstellarAMD, arsenm
Subscribers: arsenm, llvm-commits, kzhuravl
Differential Revision: https://reviews.llvm.org/D22846
llvm-svn: 276972
Before adding a new preheader block, check if there is a candidate block
where the loop setup could be placed speculatively. This will be off by
default.
llvm-svn: 276919
Using getZExtValue() will assert if the value doesn't fit into uint64_t - SHL was already doing this, I've just updated ASHR/LSHR to match
As mentioned on D22726
llvm-svn: 276855
Igor Breger