The PowerPC backend had a number of loads that were marked as canFoldAsLoad
(and I'm partially at fault here for copying around the relevant line of
TableGen definitions without really looking at what it meant). This is not
right; PPC (non-memory) instructions don't support direct memory operands, and
so there is nothing a 'foldable' instruction could be folded into.
Noticed by inspection, no test case.
The one thing we might lose by doing this is ability to fold some loads into
stackmap/patchpoint pseudo-instructions. However, this was untested, and would
not obviously have worked for extending loads, and I'd rather re-add support
for that once it can be tested.
llvm-svn: 231982
We had somehow accumulated a few target-specific SDAG nodes dealing with PPC64
TOC access that were referenced only in TableGen patterns. The associated
(pseudo-)instructions are used, but are being generated directly. NFC.
llvm-svn: 230518
See full discussion in http://reviews.llvm.org/D7491.
We now hide the add-immediate and call instructions together in a
separate pseudo-op, which is tagged to define GPR3 and clobber the
call-killed registers. The PPCTLSDynamicCall pass prior to RA now
expands this op into the two separate addi and call ops, with explicit
definitions of GPR3 on both instructions, and explicit clobbers on the
call instruction. The pass is now marked as requiring and preserving
the LiveIntervals and SlotIndexes analyses, and fixes these up after
the replacement sequences are introduced.
Self-hosting has been verified on LE P8 and BE P7 with various
optimization levels, etc. It has also been verified with the
--no-tls-optimize flag workaround removed.
llvm-svn: 228725
Unfortunately, even with the workaround of disabling the linker TLS
optimizations in Clang restored (which has already been done), this still
breaks self-hosting on my P7 machine (-O3 -DNDEBUG -mcpu=native).
Bill is currently working on an alternate implementation to address the TLS
issue in a way that also fully elides the linker bug (which, unfortunately,
this approach did not fully), so I'm reverting this now.
llvm-svn: 228460
This patch is a third attempt to properly handle the local-dynamic and
global-dynamic TLS models.
In my original implementation, calls to __tls_get_addr were hidden
from view until the asm-printer phase, at which point the underlying
branch-and-link instruction was created with proper relocations. This
mostly worked well, but I used some repellent techniques to ensure
that the TLS_GET_ADDR nodes at the SD and MI levels correctly received
input from GPR3 and produced output into GPR3. This proved to work
badly in the presence of multiple TLS variable accesses, with the
copies to and from GPR3 being scheduled incorrectly and generally
creating havoc.
In r221703, I addressed that problem by representing the calls to
__tls_get_addr as true calls during instruction lowering. This had
the advantage of removing all of the bad hacks and relying on the
existing call machinery to properly glue the copies in place. It
looked like this was going to be the right way to go.
However, as a side effect of the recent discovery of problems with
linker optimizations for TLS, we discovered cases of suboptimal code
generation with this strategy. The problem comes when tls_get_addr is
called for the same address, and there is a resulting CSE
opportunity. It turns out that in such cases MachineCSE will common
the addis/addi instructions that set up the input value to
tls_get_addr, but will not common the calls themselves. MachineCSE
does not have any machinery to common idempotent calls. This is
perfectly sensible, since presumably this would be done at the IR
level, and introducing calls in the back end isn't commonplace. In
any case, we end up with two calls to __tls_get_addr when one would
suffice, and that isn't good.
I presumed that the original design would have allowed commoning of
the machine-specific nodes that hid the __tls_get_addr calls, so as
suggested by Ulrich Weigand, I went back to that design and cleaned it
up so that the copies were properly held together by glue
nodes. However, it turned out that this didn't work either...the
presence of copies to physical registers kept the machine-specific
nodes from being commoned also.
All of which leads to the design presented here. This is a return to
the original design, except that no attempt is made to introduce
copies to and from GPR3 during instruction lowering. Virtual registers
are used until prior to register allocation. At that point, a special
pass is run that identifies the machine-specific nodes that hide the
tls_get_addr calls and introduces the copies to and from GPR3 around
them. The register allocator then coalesces these copies away. With
this design, MachineCSE succeeds in commoning tls_get_addr calls where
possible, and we get nice optimal code generation (better than GCC at
the moment, which does not common these calls).
One additional problem must be dealt with: After introducing the
mentions of the physical register GPR3, the aggressive anti-dependence
breaker sees opportunities to improve scheduling by selecting a
different register instead. Flags must be used on the instruction
descriptions to tell the anti-dependence breaker to keep its hands in
its pockets.
One thing missing from the original design was recording a definition
of the link register on the GET_TLS_ADDR nodes. Doing this was found
to be insufficient to force a stack frame to be created, which led to
looping behavior because two different LR values were stored at the
same address. This appears to have been an oversight in
PPCFrameLowering::determineFrameLayout(), which is repaired here.
Because MustSaveLR() returns true for calls to builtin_return_address,
this changed the expected behavior of
test/CodeGen/PowerPC/retaddr2.ll, which now stacks a frame but
formerly did not. I've fixed the test case to reflect this.
There are existing TLS tests to catch regressions; the checks in
test/CodeGen/PowerPC/tls-store2.ll proved to be too restrictive in the
face of instruction scheduling with these changes, so I fixed that
up.
I've added a new test case based on the PrettyStackTrace module that
demonstrated the original problem. This checks that we get correct
code generation and that CSE of the calls to __get_tls_addr has taken
place.
llvm-svn: 227976
isel is actually a cracked instruction on the P7/P8, and must start a dispatch
group. The scheduling model should reflect this so that we don't bunch too many
of them together when possible.
Thanks to Bill Schmidt and Pat Haugen for helping to sort this out.
llvm-svn: 227758
Function pointers under PPC64 ELFv1 (which is used on PPC64/Linux on the
POWER7, A2 and earlier cores) are really pointers to a function descriptor, a
structure with three pointers: the actual pointer to the code to which to jump,
the pointer to the TOC needed by the callee, and an environment pointer. We
used to chain these loads, and make them opaque to the rest of the optimizer,
so that they'd always occur directly before the call. This is not necessary,
and in fact, highly suboptimal on embedded cores. Once the function pointer is
known, the loads can be performed ahead of time; in fact, they can be hoisted
out of loops.
Now these function descriptors are almost always generated by the linker, and
thus the contents of the descriptors are invariant. As a result, by default,
we'll mark the associated loads as invariant (allowing them to be hoisted out
of loops). I've added a target feature to turn this off, however, just in case
someone needs that option (constructing an on-stack descriptor, casting it to a
function pointer, and then calling it cannot be well-defined C/C++ code, but I
can imagine some JIT-compilation system doing so).
Consider this simple test:
$ cat call.c
typedef void (*fp)();
void bar(fp x) {
for (int i = 0; i < 1600000000; ++i)
x();
}
$ cat main.c
typedef void (*fp)();
void bar(fp x);
void foo() {}
int main() {
bar(foo);
}
On the PPC A2 (the BG/Q supercomputer), marking the function-descriptor loads
as invariant brings the execution time down to ~8 seconds from ~32 seconds with
the loads in the loop.
The difference on the POWER7 is smaller. Compiling with:
gcc -std=c99 -O3 -mcpu=native call.c main.c : ~6 seconds [this is 4.8.2]
clang -O3 -mcpu=native call.c main.c : ~5.3 seconds
clang -O3 -mcpu=native call.c main.c -mno-invariant-function-descriptors : ~4 seconds
(looks like we'd benefit from additional loop unrolling here, as a first
guess, because this is faster with the extra loads)
The -mno-invariant-function-descriptors will be added to Clang shortly.
llvm-svn: 226207
We really need a separate 64-bit version of this instruction so that it can be
marked as clobbering LR8 (instead of just LR). No change in functionality
(although the verifier might be slightly happier), however, it is required for
stackmap/patchpoint support. Thus, this will be covered by stackmap test cases
once those are added.
llvm-svn: 225804
The 64-bit semantics of cntlzw are not special, the 32-bit population count is
stored as a 64-bit value in the range [0,32]. As a result, it is always zero
extended, and it can be added to the PPCISelDAGToDAG peephole optimization as a
frontier instruction for the removal of unnecessary zero extensions.
llvm-svn: 225192
lhbrx and lwbrx not only load their data with byte swapping, but also clear the
upper 32 bits (at least). As a result, they can be added to the PPCISelDAGToDAG
peephole optimization as frontier instructions for the removal of unnecessary
zero extensions.
llvm-svn: 225189
Newer POWER cores, and the A2, support the cmpb instruction. This instruction
compares its operands, treating each of the 8 bytes in the GPRs separately,
returning a 'mask' result of 0 (for false) or -1 (for true) in each byte.
Code generation support is added, in the form of a PPCISelDAGToDAG
DAG-preprocessing routine, that recognizes patterns close to what the
instruction computes (either exactly, or related by a constant masking
operation), and generates the cmpb instruction (along with any necessary
constant masking operation). This can be expanded if use cases arise.
llvm-svn: 225106
This is the second installment of improvements to instruction selection for "bit
permutation" instruction sequences. r224318 added logic for instruction
selection for 32-bit bit permutation sequences, and this adds lowering for
64-bit sequences. The 64-bit sequences are more complicated than the 32-bit
ones because:
a) the 64-bit versions of the 32-bit rotate-and-mask instructions
work by replicating the lower 32-bits of the value-to-be-rotated into the
upper 32 bits -- and integrating this into the cost modeling for the various
bit group operations is non-trivial
b) unlike the 32-bit instructions in 32-bit mode, the rotate-and-mask instructions
cannot, in one instruction, specify the
mask starting index, the mask ending index, and the rotation factor. Also,
forming arbitrary 64-bit constants is more complicated than in 32-bit mode
because the number of instructions necessary is value dependent.
Plus, support for 'late masking' was added: it is sometimes more efficient to
treat the overall value as if it had no mandatory zero bits when planning the
bit-group insertions, and then mask them in at the very end. Unfortunately, as
the structure of the bit groups is different in the two cases, the more
feasible implementation technique was to generate both instruction sequences,
and then pick the shorter one.
And finally, we now generate reasonable code for i64 bswap:
rldicl 5, 3, 16, 0
rldicl 4, 3, 8, 0
rldicl 6, 3, 24, 0
rldimi 4, 5, 8, 48
rldicl 5, 3, 32, 0
rldimi 4, 6, 16, 40
rldicl 6, 3, 48, 0
rldimi 4, 5, 24, 32
rldicl 5, 3, 56, 0
rldimi 4, 6, 40, 16
rldimi 4, 5, 48, 8
rldimi 4, 3, 56, 0
vs. what we used to produce:
li 4, 255
rldicl 5, 3, 24, 40
rldicl 6, 3, 40, 24
rldicl 7, 3, 56, 8
sldi 8, 3, 8
sldi 10, 3, 24
sldi 12, 3, 40
rldicl 0, 3, 8, 56
sldi 9, 4, 32
sldi 11, 4, 40
sldi 4, 4, 48
andi. 5, 5, 65280
andis. 6, 6, 255
andis. 7, 7, 65280
sldi 3, 3, 56
and 8, 8, 9
and 4, 12, 4
and 9, 10, 11
or 6, 7, 6
or 5, 5, 0
or 3, 3, 4
or 7, 9, 8
or 4, 6, 5
or 3, 3, 7
or 3, 3, 4
which is 12 instructions, instead of 25, and seems optimal (at least in terms
of code size).
llvm-svn: 225056
On non-Darwin PPC64, the TOC reload needs to come directly after the bctrl
instruction (for indirect calls) because the 'bctrl/ld 2, 40(1)' instruction
sequence is interpreted by the unwinding code in libgcc. To make sure these
occur as a pair, as with other pairings interpreted by the linker, fuse the two
instructions into one instruction (for code generation only).
In the future, we might wish to do this by emitting CFI directives instead,
but this solution is simpler, and mirrors what GCC does. Additional discussion
on this point is contained in the PR.
Fixes PR22015.
llvm-svn: 224788
On PPC64, we end up with lots of i32 -> i64 zero extensions, not only from all
of the usual places, but also from the ABI, which specifies that values passed
are zero extended. Almost all 32-bit PPC instructions in PPC64 mode are defined
to do *something* to the higher-order bits, and for some instructions, that
action clears those bits (thus providing a zero-extended result). This is
especially common after rotate-and-mask instructions. Adding an additional
instruction to zero-extend the results of these instructions is unnecessary.
This PPCISelDAGToDAG peephole optimization examines these zero-extensions, and
looks back through their operands to see if all instructions will implicitly
zero extend their results. If so, we convert these instructions to their 64-bit
variants (which is an internal change only, the actual encoding of these
instructions is the same as the original 32-bit ones) and remove the
unnecessary zero-extension (changing where the INSERT_SUBREG instructions are
to make everything internally consistent).
llvm-svn: 224169
My original support for the general dynamic and local dynamic TLS
models contained some fairly obtuse hacks to generate calls to
__tls_get_addr when lowering a TargetGlobalAddress. Rather than
generating real calls, special GET_TLS_ADDR nodes were used to wrap
the calls and only reveal them at assembly time. I attempted to
provide correct parameter and return values by chaining CopyToReg and
CopyFromReg nodes onto the GET_TLS_ADDR nodes, but this was also not
fully correct. Problems were seen with two back-to-back stores to TLS
variables, where the call sequences ended up overlapping with unhappy
results. Additionally, since these weren't real calls, the proper
register side effects of a call were not recorded, so clobbered values
were kept live across the calls.
The proper thing to do is to lower these into calls in the first
place. This is relatively straightforward; see the changes to
PPCTargetLowering::LowerGlobalTLSAddress() in PPCISelLowering.cpp.
The changes here are standard call lowering, except that we need to
track the fact that these calls will require a relocation. This is
done by adding a machine operand flag of MO_TLSLD or MO_TLSGD to the
TargetGlobalAddress operand that appears earlier in the sequence.
The calls to LowerCallTo() eventually find their way to
LowerCall_64SVR4() or LowerCall_32SVR4(), which call FinishCall(),
which calls PrepareCall(). In PrepareCall(), we detect the calls to
__tls_get_addr and immediately snag the TargetGlobalTLSAddress with
the annotated relocation information. This becomes an extra operand
on the call following the callee, which is expected for nodes of type
tlscall. We change the call opcode to CALL_TLS for this case. Back
in FinishCall(), we change it again to CALL_NOP_TLS for 64-bit only,
since we require a TOC-restore nop following the call for the 64-bit
ABIs.
During selection, patterns in PPCInstrInfo.td and PPCInstr64Bit.td
convert the CALL_TLS nodes into BL_TLS nodes, and convert the
CALL_NOP_TLS nodes into BL8_NOP_TLS nodes. This replaces the code
removed from PPCAsmPrinter.cpp, as the BL_TLS or BL8_NOP_TLS
nodes can now be emitted normally using their patterns and the
associated printTLSCall print method.
Finally, as a result of these changes, all references to get-tls-addr
in its various guises are no longer used, so they have been removed.
There are existing TLS tests to verify the changes haven't messed
anything up). I've added one new test that verifies that the problem
with the original code has been fixed.
llvm-svn: 221703
Since block address values can be larger than 2GB in 64-bit code, they
cannot be loaded simply using an @l / @ha pair, but instead must be
loaded from the TOC, just like GlobalAddress, ConstantPool, and
JumpTable values are.
The commit also fixes a bug in PPCLinuxAsmPrinter::doFinalization where
temporary labels could not be used as TOC values, since code would
attempt (and fail) to use GetOrCreateSymbol to create a symbol of the
same name as the temporary label.
llvm-svn: 220959
Summary:
Atomic loads and store of up to the native size (32 bits, or 64 for PPC64)
can be lowered to a simple load or store instruction (as the synchronization
is already handled by AtomicExpand, and the atomicity is guaranteed thanks to
the alignment requirements of atomic accesses). This is exactly what this patch
does. Previously, these were implemented by complex
load-linked/store-conditional loops.. an obvious performance problem.
For example, this patch turns
```
define void @store_i8_unordered(i8* %mem) {
store atomic i8 42, i8* %mem unordered, align 1
ret void
}
```
from
```
_store_i8_unordered: ; @store_i8_unordered
; BB#0:
rlwinm r2, r3, 3, 27, 28
li r4, 42
xori r5, r2, 24
rlwinm r2, r3, 0, 0, 29
li r3, 255
slw r4, r4, r5
slw r3, r3, r5
and r4, r4, r3
LBB4_1: ; =>This Inner Loop Header: Depth=1
lwarx r5, 0, r2
andc r5, r5, r3
or r5, r4, r5
stwcx. r5, 0, r2
bne cr0, LBB4_1
; BB#2:
blr
```
into
```
_store_i8_unordered: ; @store_i8_unordered
; BB#0:
li r2, 42
stb r2, 0(r3)
blr
```
which looks like a pretty clear win to me.
Test Plan:
fixed the tests + new test for indexed accesses + make check-all
Reviewers: jfb, wschmidt, hfinkel
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D5587
llvm-svn: 218922
The LDinto_toc pattern has been part of 64-bit PowerPC for a long
time, and represents loading from a memory location into the TOC
register (X2). However, this pattern doesn't explicitly record that
it modifies that register. This patch adds the missing dependency.
It was very surprising to me that this has never shown up as a problem
in the past, and that we only saw this problem recently in a single
scenario when building a self-hosted clang. It turns out that in most
cases we have another dependency present that keeps the LDinto_toc
instruction tied in place. LDinto_toc is used for TOC restore
following a call site, so this is a typical sequence:
BCTRL8 <regmask>, %CTR8<imp-use>, %RM<imp-use>, %X3<imp-use>, %X12<imp-use>, %X1<imp-def>, ...
LDinto_toc 24, %X1
ADJCALLSTACKUP 96, 0, %R1<imp-def>, %R1<imp-use>
Because the LDinto_toc is inserted prior to the ADJCALLSTACKUP, there
is a natural anti-dependency between the two that keeps it in place.
Therefore we don't usually see a problem. However, in one particular
case, one call is followed immediately by another call, and the second
call requires a parameter that is a TOC-relative address. This is the
code sequence:
BCTRL8 <regmask>, %CTR8<imp-use>, %RM<imp-use>, %X3<imp-use>, %X4<imp-use>, %X5<imp-use>, %X12<imp-use>, %X1<imp-def>, ...
LDinto_toc 24, %X1
ADJCALLSTACKUP 96, 0, %R1<imp-def>, %R1<imp-use>
ADJCALLSTACKDOWN 96, %R1<imp-def>, %R1<imp-use>
%vreg39<def> = ADDIStocHA %X2, <ga:@.str>; G8RC_and_G8RC_NOX0:%vreg39
%vreg40<def> = ADDItocL %vreg39<kill>, <ga:@.str>; G8RC:%vreg40 G8RC_and_G8RC_NOX0:%vreg39
Note that the back-to-back stack adjustments are the same size! The
back end is smart enough to recognize this and optimize them away:
BCTRL8 <regmask>, %CTR8<imp-use>, %RM<imp-use>, %X3<imp-use>, %X4<imp-use>, %X5<imp-use>, %X12<imp-use>, %X1<imp-def>, ...
LDinto_toc 24, %X1
%vreg39<def> = ADDIStocHA %X2, <ga:@.str>; G8RC_and_G8RC_NOX0:%vreg39
%vreg40<def> = ADDItocL %vreg39<kill>, <ga:@.str>; G8RC:%vreg40 G8RC_and_G8RC_NOX0:%vreg39
Now there is nothing to prevent the ADDIStocHA instruction from moving
ahead of the LDinto_toc instruction, and because of the longest-path
heuristic, this is what happens.
With the accompanying patch, %X2 is represented as an implicit def:
BCTRL8 <regmask>, %CTR8<imp-use>, %RM<imp-use>, %X3<imp-use>, %X4<imp-use>, %X5<imp-use>, %X12<imp-use>, %X1<imp-def>, ...
LDinto_toc 24, %X1, %X2<imp-def,dead>
ADJCALLSTACKUP 96, 0, %R1<imp-def,dead>, %R1<imp-use>
ADJCALLSTACKDOWN 96, %R1<imp-def,dead>, %R1<imp-use>
%vreg39<def> = ADDIStocHA %X2, <ga:@.str>; G8RC_and_G8RC_NOX0:%vreg39
%vreg40<def> = ADDItocL %vreg39<kill>, <ga:@.str>; G8RC:%vreg40 G8RC_and_G8RC_NOX0:%vreg39
So now when the two stack adjustments are removed, ADDIStocHA is
prevented from being moved above LDinto_toc.
I have not yet created a test case for this, because the original
failure occurs on a relatively large function that needs reduction.
However, this is a fairly serious bug, despite its infrequency, and I
wanted to get this patch onto the list as soon as possible so that it
can be considered for a 3.5 backport. I'll work on whittling down a
test case.
Have we missed the boat for 3.5 at this point?
Thanks,
Bill
llvm-svn: 215685
During an indirect function call sequence on the 64-bit SVR4 ABI,
generate code must load and then restore the TOC register.
This does not use a regular LOAD instruction since the TOC
register r2 is marked as reserved. Instead, the are two
special instruction patterns:
let RST = 2, DS = 2 in
def LDinto_toc: DSForm_1a<58, 0, (outs), (ins g8rc:$reg),
"ld 2, 8($reg)", IIC_LdStLD,
[(PPCload_toc i64:$reg)]>, isPPC64;
let RST = 2, DS = 10, RA = 1 in
def LDtoc_restore : DSForm_1a<58, 0, (outs), (ins),
"ld 2, 40(1)", IIC_LdStLD,
[(PPCtoc_restore)]>, isPPC64;
Note that these not only restrict the destination of the
load to r2, but they also restrict the *source* of the
load to particular address combinations. The latter is
a problem when we want to support the ELFv2 ABI, since
there the TOC save slot is no longer at 40(1).
This patch replaces those two instructions with a single
instruction pattern that only hard-codes r2 as destination,
but supports generic addresses as source. This will allow
supporting the ELFv2 ABI, and also helps generate more
efficient code for calls to absolute addresses (allowing
simplification of the ppc64-calls.ll test case).
llvm-svn: 211193
I'm under the impression that we used to infer the isCommutable flag from the
instruction-associated pattern. Regardless, we don't seem to do this (at least
by default) any more. I've gone through all of our instruction definitions, and
marked as commutative all of those that should be trivial to commute (by
exchanging the first two operands). There has been special code for the RL*
instructions, and that's not changed.
Before this change, we had the following commutative instructions:
RLDIMI
RLDIMIo
RLWIMI
RLWIMI8
RLWIMI8o
RLWIMIo
XSADDDP
XSMULDP
XVADDDP
XVADDSP
XVMULDP
XVMULSP
After:
ADD4
ADD4o
ADD8
ADD8o
ADDC
ADDC8
ADDC8o
ADDCo
ADDE
ADDE8
ADDE8o
ADDEo
AND
AND8
AND8o
ANDo
CRAND
CREQV
CRNAND
CRNOR
CROR
CRXOR
EQV
EQV8
EQV8o
EQVo
FADD
FADDS
FADDSo
FADDo
FMADD
FMADDS
FMADDSo
FMADDo
FMSUB
FMSUBS
FMSUBSo
FMSUBo
FMUL
FMULS
FMULSo
FMULo
FNMADD
FNMADDS
FNMADDSo
FNMADDo
FNMSUB
FNMSUBS
FNMSUBSo
FNMSUBo
MULHD
MULHDU
MULHDUo
MULHDo
MULHW
MULHWU
MULHWUo
MULHWo
MULLD
MULLDo
MULLW
MULLWo
NAND
NAND8
NAND8o
NANDo
NOR
NOR8
NOR8o
NORo
OR
OR8
OR8o
ORo
RLDIMI
RLDIMIo
RLWIMI
RLWIMI8
RLWIMI8o
RLWIMIo
VADDCUW
VADDFP
VADDSBS
VADDSHS
VADDSWS
VADDUBM
VADDUBS
VADDUHM
VADDUHS
VADDUWM
VADDUWS
VAND
VAVGSB
VAVGSH
VAVGSW
VAVGUB
VAVGUH
VAVGUW
VMADDFP
VMAXFP
VMAXSB
VMAXSH
VMAXSW
VMAXUB
VMAXUH
VMAXUW
VMHADDSHS
VMHRADDSHS
VMINFP
VMINSB
VMINSH
VMINSW
VMINUB
VMINUH
VMINUW
VMLADDUHM
VMULESB
VMULESH
VMULEUB
VMULEUH
VMULOSB
VMULOSH
VMULOUB
VMULOUH
VNMSUBFP
VOR
VXOR
XOR
XOR8
XOR8o
XORo
XSADDDP
XSMADDADP
XSMAXDP
XSMINDP
XSMSUBADP
XSMULDP
XSNMADDADP
XSNMSUBADP
XVADDDP
XVADDSP
XVMADDADP
XVMADDASP
XVMAXDP
XVMAXSP
XVMINDP
XVMINSP
XVMSUBADP
XVMSUBASP
XVMULDP
XVMULSP
XVNMADDADP
XVNMADDASP
XVNMSUBADP
XVNMSUBASP
XXLAND
XXLNOR
XXLOR
XXLXOR
This is a by-inspection change, and I'm not sure how to write a reliable test
case. I would like advice on this, however.
llvm-svn: 204609
This change enables tracking i1 values in the PowerPC backend using the
condition register bits. These bits can be treated on PowerPC as separate
registers; individual bit operations (and, or, xor, etc.) are supported.
Tracking booleans in CR bits has several advantages:
- Reduction in register pressure (because we no longer need GPRs to store
boolean values).
- Logical operations on booleans can be handled more efficiently; we used to
have to move all results from comparisons into GPRs, perform promoted
logical operations in GPRs, and then move the result back into condition
register bits to be used by conditional branches. This can be very
inefficient, because the throughput of these CR <-> GPR moves have high
latency and low throughput (especially when other associated instructions
are accounted for).
- On the POWER7 and similar cores, we can increase total throughput by using
the CR bits. CR bit operations have a dedicated functional unit.
Most of this is more-or-less mechanical: Adjustments were needed in the
calling-convention code, support was added for spilling/restoring individual
condition-register bits, and conditional branch instruction definitions taking
specific CR bits were added (plus patterns and code for generating bit-level
operations).
This is enabled by default when running at -O2 and higher. For -O0 and -O1,
where the ability to debug is more important, this feature is disabled by
default. Individual CR bits do not have assigned DWARF register numbers,
and storing values in CR bits makes them invisible to the debugger.
It is critical, however, that we don't move i1 values that have been promoted
to larger values (such as those passed as function arguments) into bit
registers only to quickly turn around and move the values back into GPRs (such
as happens when values are returned by functions). A pair of target-specific
DAG combines are added to remove the trunc/extends in:
trunc(binary-ops(binary-ops(zext(x), zext(y)), ...)
and:
zext(binary-ops(binary-ops(trunc(x), trunc(y)), ...)
In short, we only want to use CR bits where some of the i1 values come from
comparisons or are used by conditional branches or selects. To put it another
way, if we can do the entire i1 computation in GPRs, then we probably should
(on the POWER7, the GPR-operation throughput is higher, and for all cores, the
CR <-> GPR moves are expensive).
POWER7 test-suite performance results (from 10 runs in each configuration):
SingleSource/Benchmarks/Misc/mandel-2: 35% speedup
MultiSource/Benchmarks/Prolangs-C++/city/city: 21% speedup
MultiSource/Benchmarks/MiBench/automotive-susan: 23% speedup
SingleSource/Benchmarks/CoyoteBench/huffbench: 13% speedup
SingleSource/Benchmarks/Misc-C++/Large/sphereflake: 13% speedup
SingleSource/Benchmarks/Misc-C++/mandel-text: 10% speedup
SingleSource/Benchmarks/Misc-C++-EH/spirit: 10% slowdown
MultiSource/Applications/lemon/lemon: 8% slowdown
llvm-svn: 202451
Several of the 64-bit fixed-point instructions with immediate operands were
using the 32-bit (i32) operand nodes instead of the corresponding 64-bit (i64)
operand definitions (u16imm instead of u16imm64, for example).
This error has had no effect so far, but would have caused type-checking
violations with an upcoming change.
llvm-svn: 198356
The tests for the disassembler were adapted from the encoder tests, and for the
most part, the output from the disassembler matches that encoder-test inputs.
There are some places where more-informative mnemonics could be produced
(notably for the branch instructions), and those cases are noted in the tests
with FIXMEs.
Future work includes:
- Generating more-informative mnemonics when possible (this may also be done
in the printer).
- Remove the dependence on positional "numbered" operand-to-variable mapping
(for both encoding and decoding).
- Internally using 64-bit instruction variants in 64-bit mode (if this turns
out to matter).
llvm-svn: 197693
The instruction definitions in the PPC backend have a number of variants
defined for the same instruction to represent differences between 64-bit and
32-bit semantics. In order to generate a disassembler for the PPC backend, we
need to mark all but one of these as CodeGen only.
No functionality change intended; this is prep work for PPC disassembly
support.
llvm-svn: 197535
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 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
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
Aggressive anti-dependency breaking is enabled by default for all PPC cores.
This provides a general speedup on the P7 and other platforms (among other
factors, the instruction group formation for the non-embedded PPC cores is done
during post-RA scheduling). In order to do this safely, the incompatibility
between uses of the MFOCRF instruction and anti-dependency breaking are
resolved by marking MFOCRF with hasExtraSrcRegAllocReq. As noted in the removed
FIXME, the problem was that MFOCRF's output is sensitive to the identify of the
source register, and always paired with a shift to undo this effect. Because
anti-dependency breaking is unaware of this hidden dependency of the shift
amount on the source register of the MFOCRF instruction, changing that register
must be inhibited.
Two test cases were adjusted: The SjLj test was made more insensitive to
register choices and scheduling; the saveCR test disabled anti-dependency
breaking because part of what it is testing is proper register reuse.
llvm-svn: 190587
This is the next big chunk of fast-isel code. The primary purpose is
to implement selection of loads and stores, but there is a lot of
drag-along to support this. The common code to analyze addresses for
both loads and stores is substantial. It's also necessary to add the
materialization code for global values.
Related to load-store processing is the code to fold loads into
integer extends, since otherwise we generate lots of redundant
instructions. We also need to add some overrides to some FastEmit
routines to ensure we don't assign GPR 0 to a virtual register when
this would change the meaning of an instruction.
I added handling selection of a few binary arithmetic instructions, to
enable committing some test cases I wrote a while back.
Finally, ap couple of miscellaneous changes:
* I cleaned up some poor style from a previous patch in
PPCISelLowering.cpp, pointed out by David Blaikie.
* I enlarged the Addr.Offset field to avoid sign problems with 32-bit
offsets.
llvm-svn: 189636
Incremental improvement to fast-isel for PPC64. This allows us to
select on ret, sext, and zext. Filling in sext/zext improves some of
the existing logic in handling compare-immediates that needed extends.
A simplified return convention for fast-isel is also added to the
PPC64 calling conventions. All call/return processing for DAG
selection is handled with custom code, so there isn't an existing CC
to rely on here. The include of PPCGenCallingConv.inc causes compiler
warnings due to the 32-bit calling conventions that are not used, so
the dummy function "usePPC32CCs()" is added here to silence those.
Test cases for the return and extend logic are added.
llvm-svn: 189266
The PPC backend had been missing a pattern to generate mulli for 64-bit
multiples. We had been generating it only for 32-bit multiplies. Unfortunately,
generating li + mulld unnecessarily increases register pressure.
llvm-svn: 187807
Because the builtin longjmp implementation uses a CTR-based indirect jump, when
the control flow arrives at the builtin setjmp call, the CTR register has
necessarily been clobbered. Correspondingly, this adds CTR to the list of
implicit definitions of the builtin setjmp pseudo instruction.
We don't need to add CTR to the implicit definitions of builtin longjmp
because, even though it does clobber the CTR register, the control flow cannot
return to inside the loop unless there is also a builtin setjmp call.
llvm-svn: 186488
This adds support for the last missing construct to parse TLS-related
assembler code:
add 3, 4, symbol@tls
The ADD8TLS currently hard-codes the @tls into the assembler string.
This cannot be handled by the asm parser, since @tls is parsed as
a symbol variant. This patch changes ADD8TLS to have the @tls suffix
printed as symbol variant on output too, which allows us to remove
the isCodeGenOnly marker from ADD8TLS. This in turn means that we
can add a AsmOperand to accept @tls marked symbols on input.
As a side effect, this means that the fixup_ppc_tlsreg fixup type
is no longer necessary and can be merged into fixup_ppc_nofixup.
llvm-svn: 185692
Just as with mfocrf, it is also preferable to use mtocrf instead of
mtcrf when only a single CR register is to be written.
Current code however always emits mtcrf. This probably does not matter
when using an external assembler, since the GNU assembler will in fact
automatically replace mtcrf with mtocrf when possible. It does create
inefficient code with the integrated assembler, however.
To fix this, this patch adds MTOCRF/MTOCRF8 instruction patterns and
uses those instead of MTCRF/MTCRF8 everything. Just as done in the
MFOCRF patch committed as 185556, these patterns will be converted
back to MTCRF if MTOCRF is not available on the machine.
As a side effect, this allows to modify the MTCRF pattern to accept
the full range of mask operands for the benefit of the asm parser.
llvm-svn: 185561
When accessing just a single CR register, it is always preferable to
use mfocrf instead of mfcr, if the former is available on the CPU.
Current code makes that distinction in many, but not all places
where a single CR register value is retrieved. One missing
location is PPCRegisterInfo::lowerCRSpilling.
To fix this and make this simpler in the future, this patch changes
the bulk of the back-end to always assume mfocrf is available and
simply generate it when needed.
On machines that actually do not support mfocrf, the instruction
is replaced by mfcr at the very end, in EmitInstruction.
This has the additional benefit that we no longer need the
MFCRpseud hack, since before EmitInstruction we always have
a MFOCRF instruction pattern, which already models data flow
as required.
The patch also adds the MFOCRF8 version of the instruction,
which was missing so far.
Except for the PPCRegisterInfo::lowerCRSpilling case, no change
in generated code intended.
llvm-svn: 185556
This adds support for the generic forms of mtspr/mfspr
for the asm parser. The compiler will continue to use
the specialized patters for mtlr etc. since those are
needed to correctly describe data flow.
llvm-svn: 185532
This patch now adds support for recognizing TLS call sequences in
the asm parser. This needs a new pattern BL8_TLS, which is like
BL8_NOP_TLS except without nop. That pattern is used for the
asm parser only.
llvm-svn: 185478
As part of the global-dynamic and local-dynamic TLS sequences, we need
to use a special form of the call instruction:
bl __tls_get_addr(sym@tlsld)
bl __tls_get_addr(sym@tlsgd)
which generates two fixups. The current implementation of this causes
problems with recognizing this form in the asm parser. To fix this,
this patch reworks operand processing for this special form by using
a single operand to hold both __tls_get_addr and sym@tlsld and defining
a print method to output the above form, and an encoding method to
generate the two fixups.
As a side simplification, the patch replaces the two instruction
patterns BL8_NOP_TLSGD and BL8_NOP_TLSLD by a single BL8_NOP_TLS,
since the patterns already operate in an identical fashion (whether
we have a local-dynamic or global-dynamic symbol is already encoded
in the symbol modifier).
No change in code generation intended.
llvm-svn: 185477
The assembler currently strictly verifies that immediates for
s16imm operands are in range (-32768 ... 32767). This matches
the behaviour of the GNU assembler, with one exception: gas
allows, as a special case, operands in an extended range
(-65536 .. 65535) for the addis instruction only (and its
extended mnemonic lis).
The main reason for this seems to be to allow using unsigned
16-bit operands for lis, e.g. like lis %r1, 0xfedc.
Since this has been supported by gas for a long time, and
assembler source code seen "in the wild" actually exploits
this feature, this patch adds equivalent support to LLVM
for compatibility reasons.
llvm-svn: 184946
Currently, all instructions taking s16imm operands support symbolic
operands. However, for u16imm operands, we only support actual
immediate integers. This causes the assembler to reject code like
ori %r5, %r5, symbol@l
This patch changes the u16imm operand definition to likewise
accept symbolic operands. In fact, s16imm and u16imm can
share the same encoding routine, now renamed to getImm16Encoding.
llvm-svn: 184944
This adds pattern for the rldcr and rldic instructions (the last instruction
from the rotate/shift family that were missing). They are currently used
only by the asm parser.
llvm-svn: 184833
This adds support for the predicted forms of branches (+/-).
There are three cases to consider:
- Branches using a PPC::Predicate code
For these, I've added new PPC::Predicate codes corresponding
to the BO values for predicted branch forms, and updated insn
printing to print them correctly. I've also added new aliases
for the asm parser matching the new forms.
- bt/bf
I've added new aliases matching to gBC etc.
- bd(n)z variants
I've added new instruction patterns for the predicted forms.
In all cases, the new patterns are used for the asm parser only.
(The new infrastructure ought to be sufficient to allow use by
the compiler too at some point.)
llvm-svn: 184754
There is currently only limited support for the "absolute" variants
of branch instructions. This patch adds support for the absolute
variants of all branches that are currently otherwise supported.
This requires adding new fixup types so that the correct variant
of relocation type can be selected by the object writer.
While the compiler will continue to usually choose the relative
branch variants, this will allow the asm parser to fully support
the absolute branches, with either immediate (numerical) or
symbolic target addresses.
No change in code generation intended.
llvm-svn: 184721
Now that there is no longer any distinction between symbolLo
and symbolHi operands in either printing, encoding, or parsing,
the operand types can be removed in favor of simply using
s16imm.
This completes the patch series to decouple lo/hi operand part
processing from the particular instruction whose operand it is.
No change in code generation expected from this patch.
llvm-svn: 182618
When targeting the Darwin assembler, we need to generate markers ha16() and
lo16() to designate the high and low parts of a (symbolic) immediate. This
is necessary not just for plain symbols, but also for certain symbolic
expression, typically along the lines of ha16(A - B). The latter doesn't
work when simply using VariantKind flags on the symbol reference.
This is why the current back-end uses hacks (explicitly called out as such
via multiple FIXMEs) in the symbolLo/symbolHi print methods.
This patch uses target-defined MCExpr codes to represent the Darwin
ha16/lo16 constructs, following along the lines of the equivalent solution
used by the ARM back end to handle their :upper16: / :lower16: markers.
This allows us to get rid of special handling both in the symbolLo/symbolHi
print method and in the common code MCExpr::print routine. Instead, the
ha16 / lo16 markers are printed simply in a custom print routine for the
target MCExpr types. (As a result, the symbolLo/symbolHi print methods
can now replaced by a single printS16ImmOperand routine that also handles
symbolic operands.)
The patch also provides a EvaluateAsRelocatableImpl routine to handle
ha16/lo16 constructs. This is not actually used at the moment by any
in-tree code, but is provided as it makes merging into David Fang's
out-of-tree Mach-O object writer simpler.
Since there is no longer any need to treat VK_PPC_GAS_HA16 and
VK_PPC_DARWIN_HA16 differently, they are merged into a single
VK_PPC_ADDR16_HA (and likewise for the _LO16 types).
llvm-svn: 182616
Hal Finkel