this completes the basic port of ARM constant islands to Mips16.
More testing, code review, cleanup is in order but basically everything
seems to be working. A bug in gas is preventing some of the runtime
testing but I hope to resolve this soon.
llvm-svn: 196331
Unlike msvc, when handling a thiscall + sret gcc will
* Put the sret in %ecx
* Put the this pointer is (%esp)
This fixes, for example, calling stringstream::str.
llvm-svn: 196312
The backend converts 64-bit ORs into subreg moves if the upper 32 bits
of one operand and the low 32 bits of the other are known to be zero.
It then tries to peel away redundant ANDs from the upper 32 bits.
Since AND masks are canonicalized to exclude known-zero bits,
the test ORs the mask and the known-zero bits together before
checking for redundancy. The problem was that it was using the
wrong node when checking for known-zero bits, so could drop ANDs
that were still needed.
llvm-svn: 196267
- The fix to PR17631 fixes part of the cases where 'vzeroupper' should
not be issued before 'call' insn. There're other cases where helper
calls will be inserted not limited to epilog. These helper calls do
not follow the standard calling convention and won't clobber any YMM
registers. (So far, all call conventions will clobber any or part of
YMM registers.)
This patch enhances the previous fix to cover more cases 'vzerosupper' should
not be inserted by checking if that function call won't clobber any YMM
registers and skipping it if so.
llvm-svn: 196261
PPCScoreboardHazardRecognizer was a subclass of ScoreboardHazardRecognizer
which did only one thing: filtered out nodes in EmitInstruction for which
DAG->getInstrDesc(SU) returned NULL. This used to be the case for PPC pseudo
instructions. As far as I can tell, this is no longer true, and so we can use
ScoreboardHazardRecognizer directly.
llvm-svn: 196171
MO_JumpTableIndex and MO_ExternalSymbol don't show up on inline asm.
Keeping parts of the old asm printer just to print inline asm to a string that
we then parse back looks like a hack.
llvm-svn: 196111
eliminateFrameIndex() has been reworked to handle both small & large frames
with either a FP or SP.
An additional Slot is required for Scavenging spills when not using FP for large frames.
Reworked the handling of Register Scavenging.
Whether we are using an FP or not, whether it is a large frame or not,
and whether we are using a large code model or not are now independent.
llvm-svn: 196091
These are used by MachO only at the moment, and (much like the existing
MOVW/MOVT set) work around the fact that the labels used in the actual
instructions often contain PC-dependent components, which means that repeatedly
materialising the same global can't be CSEed.
With small modifications, it could be adapted to how ELF finds the address of
_GLOBAL_OFFSET_TABLE_, which would give similar benefits in PIC mode there.
llvm-svn: 196090
When using large code model:
Global objects larger than 'CodeModelLargeSize' bytes are placed in sections named with a trailing ".large"
The folded global address of such objects are lowered into the const pool.
During inspection it was noted that LowerConstantPool() was using a default offset of zero.
A fix was made, but due to only offsets of zero being generated, testing only verifies the change is not detrimental.
Correct the flags emitted for explicitly specified sections.
We assume the size of the object queried by getSectionForConstant() is never greater than CodeModelLargeSize.
To handle greater than CodeModelLargeSize, changes to AsmPrinter would be required.
llvm-svn: 196087
Large frame offsets are loaded from the ConstantPool.
Where possible, offsets are encoded using the smaller MKMSK instruction.
Large frame offsets can only be used when there is a frame-pointer.
llvm-svn: 196085
Previously, we clobbered callee-saved registers when folding an "add
sp, #N" into a "pop {rD, ...}" instruction. This change checks whether
a register we're going to add to the "pop" could actually be live
outside the function before doing so and should fix the issue.
This should fix PR18081.
llvm-svn: 196046
- Actually abort when an error occurred.
- Check that the frontend lookup worked when parsing length/size/type operators.
Tested by a clang test. PR18096.
llvm-svn: 196044
This adds a scheduling model for the POWER7 (P7) core, and enables the
machine-instruction scheduler when targeting the P7. Scheduling for the P7,
like earlier ooo PPC cores, requires considering both dispatch group hazards,
and functional unit resources and latencies. These are both modeled in a
combined itinerary. Dispatch group formation is still handled by the post-RA
scheduler (which still needs to be updated for the P7, but nevertheless does a
pretty good job).
One interesting aspect of this change is that I've also enabled to use of AA
duing CodeGen for the P7 (just as it is for the embedded cores). The benchmark
results seem to support this decision (see below), and while this is normally
useful for in-order cores, and not for ooo cores like the P7, I think that the
dispatch slot hazards are enough like in-order resources to make the AA useful.
Test suite significant performance differences (where negative is a speedup,
and positive is a regression) vs. the current situation:
MultiSource/Benchmarks/BitBench/drop3/drop3
with AA: N/A
without AA: -28.7614% +/- 19.8356%
(significantly against AA)
MultiSource/Benchmarks/FreeBench/neural/neural
with AA: -17.7406% +/- 11.2712%
without AA: N/A
(significantly in favor of AA)
MultiSource/Benchmarks/SciMark2-C/scimark2
with AA: -11.2079% +/- 1.80543%
without AA: -11.3263% +/- 2.79651%
MultiSource/Benchmarks/TSVC/Symbolics-flt/Symbolics-flt
with AA: -41.8649% +/- 17.0053%
without AA: -34.5256% +/- 23.7072%
MultiSource/Benchmarks/mafft/pairlocalalign
with AA: 25.3016% +/- 17.8614%
without AA: 38.6629% +/- 14.9391%
(significantly in favor of AA)
MultiSource/Benchmarks/sim/sim
with AA: N/A
without AA: 13.4844% +/- 7.18195%
(significantly in favor of AA)
SingleSource/Benchmarks/BenchmarkGame/Large/fasta
with AA: 15.0664% +/- 6.70216%
without AA: 12.7747% +/- 8.43043%
SingleSource/Benchmarks/BenchmarkGame/puzzle
with AA: 82.2713% +/- 26.3567%
without AA: 75.7525% +/- 41.1842%
SingleSource/Benchmarks/Misc/flops-2
with AA: -37.1621% +/- 20.7964%
without AA: -35.2342% +/- 20.2999%
(significantly in favor of AA)
These are 99.5% confidence intervals from 5 runs per configuration. Regarding
the choice to turn on AA during CodeGen, of these results, four seem
significantly in favor of using AA, and one seems significantly against. I'm
not making this decision based on these numbers alone, but these results
seem consistent with results I have from other tests, and so I think that, on
balance, using AA is a win.
llvm-svn: 195981
In preparation for adding scheduling definitions for the POWER7, split some PPC
itinerary classes so that the P7's latencies and hazards can be better
described. For the most part, this means differentiating indexed from non-index
pre-increment loads and stores. Also, differentiate single from
double-precision sqrt.
No functionality change intended (except for a more-specific latency for
single-precision sqrt on the A2).
llvm-svn: 195980
This prevents the compiler from emitting invalid ld.[bhwd]'s and st.[bhwd]'s
when the stack frame is between 512 and 32,768 bytes in size.
llvm-svn: 195973
in constant islands for Mips16. We introdcuce JalB16 as a synomnym
for Jal16. It makes it easier to read and is also necessary because
Jal16 is a call instruction but JalB16 is being used as a branch.
Various parts of LLVM will not work properly even in this late stage of
the backend if we use what was declared as a call instruction to function
as a branch. For one, basic block labels may not get emitted in some
situations.
llvm-svn: 195968
The operand latencies for loads and stores in the PPC440 itinerary were wrong
(the store operands are all inputs, and the "with update" (pre-increment)
instructions need a latency for the additional output).
llvm-svn: 195948
The operand latencies for the PPC440 should be specified relative to dispatch,
not relative to the initial fetch-and-decode stages. Because most instructions
(ignoring bypass) wait in dispatch until their operands are ready, this is
modeled as reading input operands "at dispatch" (0 cycles after issue), and so
every input and output operand has 4 cycles subtracted from it.
This could alter scheduling slightly, but I don't expect a large effect.
llvm-svn: 195947
Modeling the fetch and decode units in the PPC440 itinerary does not add
anything to the hazard detection capability (and so modeling them just wastes
compile time).
No functionality change intended.
llvm-svn: 195946
target independent.
Most of the x86 specific stackmap/patchpoint handling was necessitated by the
use of the native address-mode format for frame index operands. PEI has now
been modified to treat stackmap/patchpoint similarly to DEBUG_INFO, allowing
us to use a simple, platform independent register/offset pair for frame
indexes on stackmap/patchpoints.
Notes:
- Folding is now platform independent and automatically supported.
- Emiting patchpoints with direct memory references now just involves calling
the TargetLoweringBase::emitPatchPoint utility method from the target's
XXXTargetLowering::EmitInstrWithCustomInserter method. (See
X86TargetLowering for an example).
- No more ugly platform-specific operand parsers.
This patch shouldn't change the generated output for X86.
llvm-svn: 195944
I think, in principle, intrinsics_gen may be added explicitly.
That said, it can be added incidentally, since each target already has dependencies to llvm-tblgen.
Almost all source files depend on both CommonTaleGen and intrinsics_gen.
Explicit add_dependencies() have been pruned under lib/Target.
llvm-svn: 195929
add_public_tablegen_target adds *CommonTableGen to LLVM_COMMON_DEPENDS.
LLVM_COMMON_DEPENDS affects add_llvm_library (and other add_target stuff) within its scope.
llvm-svn: 195927
Instead of sharing functional unit names between the various PPC itineraries,
give each core its own unit names prefixed with the core name. This follows
the convention used by other backends (such as ARM), and removes a non-obvious
ordering dependency between the various PPCSchedule*.td files.
No functionality change intended.
llvm-svn: 195908
conditional branches for very large targets. That will be the next small
patch. Everything now should in principle work as good (functionality
wise) as without constant islands so we decided at Mips/Imagination to
make constant islands the default for Mips16 now so that it will get
excercised a lot and this port is still experimentatl though hopefully soon
we will change the status. Some more cleanup and code review is in order
but things are converging fast.
llvm-svn: 195902
make PIC calls a little more efficient:
1. Remove instructions setting up $gp if it is known that a function has been
called at least once.
2. Save the address of a called function in a register instead of loading
it from the GOT at every call site.
llvm-svn: 195892
This adds the IIC_ prefix to the instruction itinerary class names, giving the
PPC backend a naming convention for itinerary classes that is more consistent
with that used by the X86 and ARM backends.
Instruction scheduling in the PPC backend needs a bunch of cleanup and
improvement (especially for the ooo cores). This is just a preliminary step.
No functionality change intended.
llvm-svn: 195890
SGPRs are spilled into VGPRs using the {READ,WRITE}LANE_B32 instructions.
v2:
- Fix encoding of Lane Mask
- Use correct register flags, so we don't overwrite the low dword
when restoring multi-dword registers.
v3:
- Register spilling seems to hang the GPU, so replace all shaders
that need spilling with a dummy shader.
v4:
- Fix *LANE definitions
- Change destination reg class for 32-bit SMRD instructions
v5:
- Remove small optimization that was crashing Serious Sam 3.
https://bugs.freedesktop.org/show_bug.cgi?id=68224https://bugs.freedesktop.org/show_bug.cgi?id=71285
NOTE: This is a candidate for the 3.4 branch.
llvm-svn: 195880
Writing to the M0 register from an SMRD instruction hangs the GPU, so
we need to use the SGPR_32 register class, which does not include M0.
NOTE: This is a candidate for the 3.4 branch.
llvm-svn: 195879
It is only used for asm printing.
On X86 we put basic block addresses on register before passing them to inline
asm, so the MO_MachineBasicBlock case was dead.
MO_ExternalSymbol was dead since any symbol being passed to inline asm
is represented as MO_GlobalAddress.
The MO_GlobalAddress and MO_Register cases were not tested.
llvm-svn: 195824
- Fix bug in (vsext (vzext x)) -> (vsext x) in SIGN_EXTEND_IN_REG
lowering where we need to check whether x is a vector type (in-reg
type) of i8, i16 or i32; otherwise, that optimization is not valid.
llvm-svn: 195779
We would wrongly transform the testcase into the equivalent of an AND with 1.
The problem was that, when testing whether the shifted-in bits of the right
shift were significant, we used the width of the final zero-extended result
rather than the width of the shifted value.
llvm-svn: 195731
A Direct stack map location records the address of frame index. This
address is itself the value that the runtime requested. This differs
from IndirectMemRefOp locations, which refer to a stack locations from
which the requested values must be loaded. Direct locations can
directly communicate the address if an alloca, while IndirectMemRefOp
handle register spills.
For example:
entry:
%a = alloca i64...
llvm.experimental.stackmap(i32 <ID>, i32 <shadowBytes>, i64* %a)
Since both the alloca and stackmap intrinsic are in the entry block,
and the intrinsic takes the address of the alloca, the runtime can
assume that LLVM will not substitute alloca with any intervening
value. This must be verified by the runtime by checking that the stack
map's location is a Direct location type. The runtime can then
determine the alloca's relative location on the stack immediately after
compilation, or at any time thereafter. This differs from Register and
Indirect locations, because the runtime can only read the values in
those locations when execution reaches the instruction address of the
stack map.
llvm-svn: 195712
Patch by Mikulas Patocka. I added the test. I checked that for cpu names that
gas knows about, it also doesn't generate nopl.
The modified cpus:
i686 - there are i686-class CPUs that don't have nopl: Via c3, Transmeta
Crusoe, Microsoft VirtualBox - see
https://bbs.archlinux.org/viewtopic.php?pid=775414
k6, k6-2, k6-3, winchip-c6, winchip2 - these are 586-class CPUs
via c3 c3-2 - see https://bugs.archlinux.org/task/19733 as a proof that
Via c3 and c3-Nehemiah don't have nopl
llvm-svn: 195679
These are handled almost identically to static mode (and ELF's global address
materialisation), except that a symbol may have "$non_lazy_ptr" appended. This
can be handled by passing appropriate flags along with the instruction instead
of using entirely separate pseudo-instructions.
llvm-svn: 195655
There is no sane way for an LEApcrel (= single ADR) instruction to generate a
global address on any ARM target I know of. Fortunately, no-one was trying to
any more, but there were vestigial patterns.
llvm-svn: 195644
to what is needed for constant islands. The prescan method for Mips16 constant
islands will eventually go away. It is only temporary and should be done
earlier when the instructions are first created or from the DAG. If we keep
it here we need to handle better the situation where constant islands
is called multiple times since don't want to prescan more than once.
llvm-svn: 195569
I had to move some code and I moved a declaration forward past it's first use
in the function but by nutty coincidence there was another variable of the same
name and type and with completely unrelated function that was declared globally
in the class so no compilation error ensued.
It required some unusual conditions for it to even matter. Caused test
case casts.c in test-suite to fail during compilation with a duplicate
symbol error. I would have noticed it during final code review for this port.
llvm-svn: 195565
We were ignoring the ordered/onordered bits and also the signed/unsigned
bits of condition codes when lowering the DAG to MachineInstrs.
NOTE: This is a candidate for the 3.4 branch.
llvm-svn: 195514
Utilizing the 8 and 16 bit comparison instructions, even when an input can
be folded into the comparison instruction itself, is typically not worth it.
There are too many partial register stalls as a result, leading to significant
slowdowns. By always performing comparisons on at least 32-bit
registers, performance of the calculation chain leading to the
comparison improves. Continue to use the smaller comparisons when
minimizing size, as that allows better folding of loads into the
comparison instructions.
rdar://15386341
llvm-svn: 195496
Improvements over r195317:
- Set/restore EnableFastISel flag instead of just running FastISel within
SelectAllBasicBlocks; the flag is checked in various places, and
FastISel won't run properly if those places don't do the right thing.
- Test looks for normal ISel versus FastISel behavior, and not
something more subtle that doesn't work everywhere.
Based on work by Andrea Di Biagio.
llvm-svn: 195491
- When simplifying the mask generation for BLEND, check whether that mask is
also consumed by other non-BLEND insns. If true, skip that simplification.
llvm-svn: 195476
I've no idea why I decided to handle TMxx differently from all the other
high/low logic operations, but it was a stupid thing to do. The high
registers aren't available as separate 32-bit registers on z10,
so subreg_h32 can't be used on a GR64 there.
I've normally been testing with z196 and with -O3 and so hadn't noticed
this until now.
llvm-svn: 195473
lowerBUILD_VECTOR() was treating integer constant splats as being legal
regardless of whether they had undef values. This caused instruction
selection failures when the undefs were legalized to zero, making the
constant non-splat.
Fixed this by requiring HasAnyUndef to be false for a integer constant
splat to be legal. If it is true, a new node is generated with the undefs
replaced with the necessary values to remain a splat.
llvm-svn: 195455
Splitting a basic block will create a new ALU clause, so we need to make
sure we aren't moving uses of registers that are local to their
current clause into a new one.
I had a test case for this, but unfortunately unrelated schedule changes
invalidated it, and I wasn't been able to come up with another one.
NOTE: This is a candidate for the 3.4 branch.
llvm-svn: 195399
AMD's processors family K7, K8, K10, K12, K15 and K16 are known to have SHLD/SHRD instructions with very poor latency. Optimization guides for these processors recommend using an alternative sequence of instructions. For these AMD's processors, I disabled folding (or (x << c) | (y >> (64 - c))) when we are not optimizing for size.
It might be beneficial to disable this folding for some of the Intel's processors. However, since I couldn't find specific recommendations regarding using SHLD/SHRD instructions on Intel's processors, I haven't disabled this peephole for Intel.
llvm-svn: 195383
Mask == ~InvMask asserts if the width of Mask and InvMask differ.
The combine isn't valid (with two exceptions, see below) if the widths differ
so test for this before testing Mask == ~InvMask.
In the specific cases of Mask=~0 and InvMask=0, as well as Mask=0 and
InvMask=~0, the combine is still valid. However, there are more appropriate
combines that could be used in these cases such as folding x & 0 to 0, or
x & ~0 to x.
llvm-svn: 195364
It broke, at least, i686 target. It is reproducible with "llc -mtriple=i686-unknown".
FYI, it didn't appear to add either "-O0" or "-fast-isel".
llvm-svn: 195339
clang optimizes tail calls, as in this example:
int foo(void);
int bar(void) {
return foo();
}
where the call is transformed to:
calll .L0$pb
.L0$pb:
popl %eax
.Ltmp0:
addl $_GLOBAL_OFFSET_TABLE_+(.Ltmp0-.L0$pb), %eax
movl foo@GOT(%eax), %eax
popl %ebp
jmpl *%eax # TAILCALL
However, the GOT references must all be resolved at dlopen() time, and so this
approach cannot be used with lazy dynamic linking (e.g. using RTLD_LAZY), which
usually populates the PLT with stubs that perform the actual resolving.
This patch changes X86TargetLowering::LowerCall() to skip tail call
optimization, if the called function is a global or external symbol.
Patch by Dimitry Andric!
PR15086
llvm-svn: 195318
The instruction definitions incorrectly specified that popcntd and popcntw have
record forms; they do not. This mistake was causing invalid code generation.
llvm-svn: 195272
There's no test case for this commit. This is because it is doubtful that the
incorrect behaviour can actually trigger. When MSA is not enabled, the type
legalizer should have eliminated all occurrences of patterns the affected
pseudo-instruction could possibly match before instruction selection occurs.
llvm-svn: 195252
Masking operations (where only some number of the low bits are being kept) are
selected to rldicl(x, 0, mb). If x is a logical right shift (which would become
rldicl(y, 64-n, n)), we might be able to fold the two instructions together:
rldicl(rldicl(x, 64-n, n), 0, mb) -> rldicl(x, 64-n, mb) for n <= mb
The right shift is really a left rotate followed by a mask, and if the explicit
mask is a more-restrictive sub-mask of the mask implied by the shift, only one
rldicl is needed.
llvm-svn: 195185
Hard float for mips16 means essentially to compile as soft float but to
use a runtime library for soft float that is written with native mips32
floating point instructions (those runtime routines run in mips32 hard
float mode).
The patch reviewed by Reed Kotler.
llvm-svn: 195123
No true functional changes.
Change the "hack" name of emitMipsHackSTOCG to emitSymSTO.
Remove demonstration code in AsmParser for emitMipsHackSTOCG and
emitMipsHackELFFlags. The STO field is in an ELF symbol and is not
an explicit directive. That said, we are missing the compliment call
in AsmParser and that will need to be addressed soon.
XFAIL dummy tests for emitMipsHackELFFlags and emitMipsHackELFFlags.
These will built out with following patches.
llvm-svn: 195067
This patch removes most of the trivial cases of weak vtables by pinning them to
a single object file. The memory leaks in this version have been fixed. Thanks
Alexey for pointing them out.
Differential Revision: http://llvm-reviews.chandlerc.com/D2068
Reviewed by Andy
llvm-svn: 195064
This reverts commit r190888, to fix PR17967. The original change wasn't
the right way to get @feat.00 into the object file. The right fix is to
make @feat.00 be a global symbol.
llvm-svn: 195053
Fixed an inappropriate use of BuildPairF64 when compiling for MIPS32 with FP64
which resulted in an impossible constraint on the register allocation. It now
uses BuildPairF64_64.
llvm-svn: 195007
This change is incorrect. If you delete virtual destructor of both a base class
and a subclass, then the following code:
Base *foo = new Child();
delete foo;
will not cause the destructor for members of Child class. As a result, I observe
plently of memory leaks. Notable examples I investigated are:
ObjectBuffer and ObjectBufferStream, AttributeImpl and StringSAttributeImpl.
llvm-svn: 194997
Implementing this on bigendian platforms could get strange. I added a
target hook, getStackSlotRange, per Jakob's recommendation to make
this as explicit as possible.
llvm-svn: 194942
Cameron McInally