Two memory instructions with a dependency only on the address register
between the two (the first one of them being post-incrememnt) can be
packetized together after the offset on the second was updated to the
incremement value. Make sure that the new offset is valid for the
instruction.
llvm-svn: 328897
These instructions have been around for a long time, but we
haven't supported intrinsics for them. The "new" versions use
the CSx register for the start of the buffer instead of the K
field in the Mx register.
We need to use pseudo instructions for these instructions until
after register allocation. The problem is that these instructions
allocate a M0/CS0 or M1/CS1 pair. But, we can't generate code for
the CSx set-up until after register allocation when the Mx
register has been fixed for the instruction.
There is a related clang patch.
Patch by Brendon Cahoon.
llvm-svn: 328724
The pipeliner is not adding a dependence edge for a loop carried
dependence, and ends up scheduling a load from iteration n prior
to an aliased store in iteration n-1.
The code that adds the loop carried dependences in the pipeliner
doesn't check if the memory objects for loads and stores are
"identified" (i.e., distinct) objects. If they are not, then the
code that adds the dependences needs to be conservative. The
objects can be used to check dependences only when they are
distinct objects.
The code that checks for loop carried dependences has been updated
to classify loads and stores that are not identified as "unknown"
values. A store with an "unknown" value can potentially create
a loop carried dependence with any pending load.
Patch by Brendon Cahoon.
llvm-svn: 328547
The patch contains severals changes needed to pipeline an example
that was transformed so that a Phi with a subreg is converted to
copies.
The pipeliner wasn't working for a couple of reasons.
- The RecMII was 3 instead of 2 due to the extra copies.
- Copy instructions contained a latency of 1.
- The node order algorithm was not choosing the best "bottom"
node, which caused an instruction to be scheduled that had a
predecessor and successor already scheduled.
- Updated the Hexagon Machine Scheduler to check if the node is
latency bound when adding the cost for a 0-latency dependence.
The RecMII was 3 because the computation looks at the number of
nodes in the recurrence. The extra copy is an extra node but
it shouldn't increase the latency. The new RecMII computation
looks at the latency of the instructions in the recurrence. We
changed the latency of the dependence of a copy to 0. The latency
computation for the copy also checks the use of the copy (similar
to a reg_sequence).
The node order algorithm was not choosing the last instruction
in the recurrence for a bottom up traversal. This was when the
last instruction is a copy. A check was added when choosing the
instruction to check for NodeNum if the maxASAP is the same. This
means that the scheduler will not end up with another node in
the recurrence that has both a predecessor and successor already
scheduled.
The cost computation in Hexagon Machine Scheduler adds cost when
an instruction can be packetized with a zero-latency instruction.
We should only do this if the schedule is latency bound.
Patch by Brendon Cahoon.
llvm-svn: 328542
The pipeliner is asserting because the serialization step that
occurs at the end is deleting an instruction. The assert
occurs later on because there is a use without a definition.
The problem occurs when an instruction defines a value used
by a REQ_SEQUENCE and that value is used by a COPY instruction.
The latencies between these instructions are zero, so they are
put in to the same packet. The serialization code is unable to
handle this correctly, and ends up putting the REG_SEQUENCE
before its definition.
There is special code in the serialization step that attempts
to handle zero-cost instructions (phis, copy, reg_sequence)
differently than regular instructions. Unfortunately, this means
the order does not come out correct.
This patch simplifies the code by changing the seperate steps for
handling zero-cost and regular instructions. Only phis are
handled separate now, since they should occurs first. Then, this
patch adds checks to make use the MoveUse is set to the smallest
value if there are multiple uses in a cycle.
Patch by Brendon Cahoon.
llvm-svn: 328540
The code in orderDepdences that looks at the order dependences between
instructions was processing all the successor and predecessor order
dependences. However, we really only want to check for an order dependence
for instructions scheduled in the same cycle.
Also, fixed how the pipeliner handles output dependences. An output
dependence is also a potential loop carried dependence. The pipeliner
didn't handle this case properly so an invalid schedule could be created
that allowed an output dependence to be scheduled in the next iteration
at the same cycle.
Patch by Brendon Cahoon.
llvm-svn: 328516
When the definition of a phi is used by a phi in the next iteration,
the pipeliner was assuming that the definition is processed first.
Because of the assumption, an incorrect phi name was used. This patch
has a check to see if the phi definition has been processed already.
Patch by Brendon Cahoon.
llvm-svn: 328510
The pipeliner needs to be conservative when updating the memoperands
of instructions in the epilog. Previously, the pipeliner was changing
the offset of the memoperand based upon the scheduling stage. However,
that is incorrect when control flow branches around the kernel code.
The bug enabled a load and store to the same stack offset to be swapped.
This patch fixes the bug by updating the size of the memoperands to be
UINT_MAX. This conservative value means that dependences will be created
between other loads and stores.
Patch by Brendon Cahoon.
llvm-svn: 328508
HexagonGenMux would collapse pairs of predicated transfers if it assumed
that the predicated .new forms cannot be created. Turns out that generating
mux is preferable in almost all cases.
Introduce an option -hexagon-gen-mux-threshold that controls the minimum
distance between the instruction defining the predicate and the later of
the two transfers. If the distance is closer than the threshold, mux will
not be generated. Set the threshold to 0 by default.
llvm-svn: 328346
Summary:
This pass sinks COPY instructions into a successor block, if the COPY is not
used in the current block and the COPY is live-in to a single successor
(i.e., doesn't require the COPY to be duplicated). This avoids executing the
the copy on paths where their results aren't needed. This also exposes
additional opportunites for dead copy elimination and shrink wrapping.
These copies were either not handled by or are inserted after the MachineSink
pass. As an example of the former case, the MachineSink pass cannot sink
COPY instructions with allocatable source registers; for AArch64 these type
of copy instructions are frequently used to move function parameters (PhyReg)
into virtual registers in the entry block..
For the machine IR below, this pass will sink %w19 in the entry into its
successor (%bb.1) because %w19 is only live-in in %bb.1.
```
%bb.0:
%wzr = SUBSWri %w1, 1
%w19 = COPY %w0
Bcc 11, %bb.2
%bb.1:
Live Ins: %w19
BL @fun
%w0 = ADDWrr %w0, %w19
RET %w0
%bb.2:
%w0 = COPY %wzr
RET %w0
```
As we sink %w19 (CSR in AArch64) into %bb.1, the shrink-wrapping pass will be
able to see %bb.0 as a candidate.
With this change I observed 12% more shrink-wrapping candidate and 13% more dead copies deleted in spec2000/2006/2017 on AArch64.
Reviewers: qcolombet, MatzeB, thegameg, mcrosier, gberry, hfinkel, john.brawn, twoh, RKSimon, sebpop, kparzysz
Reviewed By: sebpop
Subscribers: evandro, sebpop, sfertile, aemerson, mgorny, javed.absar, kristof.beyls, llvm-commits
Differential Revision: https://reviews.llvm.org/D41463
llvm-svn: 328237
The pipeliner needs to remove instructions from the SlotIndexes
structure when they are deleted. Otherwise, the SlotIndexes map
has stale data, and an assert will occur when adding new
instructions.
This patch also changes the pipeliner to make the back-edge of
a loop carried dependence 1 cycle. The 1 cycle latency is added
to the anti-dependence that represents the back-edge. This
changes eliminates a couple of hacks added to the pipeliner to
handle the latency of the back-edge. It is needed to correctly
pipeline the test case for the sub-register elimination pass.
llvm-svn: 328113
TopReadyCycle and BotReadyCycle were off by one cycle when an SU is either
the first instruction or the last instruction in a packet.
Patch by Ikhlas Ajbar.
llvm-svn: 328000
Avoid scheduling two loads in such a way that they would end up in the
same packet. If there is a load in a packet, try to schedule a non-load
next.
Patch by Brendon Cahoon.
llvm-svn: 327742
Get rid of the "; mem:" suffix and use the one we use in MIR: ":: (load 2)".
rdar://38163529
Differential Revision: https://reviews.llvm.org/D42377
llvm-svn: 327580
This adds two features: "packets", and "nvj".
Enabling "packets" allows the compiler to generate instruction packets,
while disabling it will prevent it and disable all optimizations that
generate them. This feature is enabled by default on all subtargets.
The feature "nvj" allows the compiler to generate new-value jumps and it
implies "packets". It is enabled on all subtargets.
The exception is made for packets with endloop instructions, since they
require a certain minimum number of instructions in the packets to which
they apply. Disabling "packets" will not prevent hardware loops from
being generated.
llvm-svn: 327302
Summary:
A desired property of the node order in Swing Modulo Scheduling is
that for nodes outside circuits the following holds: none of them is
scheduled after both a successor and a predecessor. We call
node orders that meet this property valid.
Although invalid node orders do not lead to the generation of incorrect
code, they can cause the pipeliner not being able to find a pipelined schedule
for arbitrary II. The reason is that after scheduling the successor and the
predecessor of a node, no room may be left to schedule the node itself.
For data flow graphs with 0-latency edges, the node ordering algorithm
of Swing Modulo Scheduling can generate such undesired invalid node orders.
This patch fixes that.
In the remainder of this commit message, I will give an example
demonstrating the issue, explain the fix, and explain how the the fix is tested.
Consider, as an example, the following data flow graph with all
edge latencies 0 and all edges pointing downward.
```
n0
/ \
n1 n3
\ /
n2
|
n4
```
Consider the implemented node order algorithm in top-down mode. In that mode,
the algorithm orders the nodes based on greatest Height and in case of equal
Height on lowest Movability. Finally, in case of equal Height and
Movability, given two nodes with an edge between them, the algorithm prefers
the source-node.
In the graph, for every node, the Height and Movability are equal to 0.
As will be explained below, the algorithm can generate the order n0, n1, n2, n3, n4.
So, node n3 is scheduled after its predecessor n0 and after its successor n2.
The reason that the algorithm can put node n2 in the order before node n3,
even though they have an edge between them in which node n3 is the source,
is the following: Suppose the algorithm has constructed the partial node
order n0, n1. Then, the nodes left to be ordered are nodes n2, n3, and n4. Suppose
that the while-loop in the implemented algorithm considers the nodes in
the order n4, n3, n2. The algorithm will start with node n4, and look for
more preferable nodes. First, node n4 will be compared with node n3. As the nodes
have equal Height and Movability and have no edge between them, the algorithm
will stick with node n4. Then node n4 is compared with node n2. Again the
Height and Movability are equal. But, this time, there is an edge between
the two nodes, and the algorithm will prefer the source node n2.
As there are no nodes left to compare, the algorithm will add node n2 to
the node order, yielding the partial node order n0, n1, n2. In this way node n2
arrives in the node-order before node n3.
To solve this, this patch introduces the ZeroLatencyHeight (ZLH) property
for nodes. It is defined as the maximum unweighted length of a path from the
given node to an arbitrary node in which each edge has latency 0.
So, ZLH(n0)=3, ZLH(n1)=ZLH(n3)=2, ZLH(n2)=1, and ZLH(n4)=0
In this patch, the preference for a greater ZeroLatencyHeight
is added in the top-down mode of the node ordering algorithm, after the
preference for a greater Height, and before the preference for a
lower Movability.
Therefore, the two allowed node-orders are n0, n1, n3, n2, n4 and n0, n3, n1, n2, n4.
Both of them are valid node orders.
In the same way, the bottom-up mode of the node ordering algorithm is adapted
by introducing the ZeroLatencyDepth property for nodes.
The patch is tested by adding extra checks to the following existing
lit-tests:
test/CodeGen/Hexagon/SUnit-boundary-prob.ll
test/CodeGen/Hexagon/frame-offset-overflow.ll
test/CodeGen/Hexagon/vect/vect-shuffle.ll
Before this patch, the pipeliner failed to pipeline the loops in these tests
due to invalid node-orders. After the patch, the pipeliner successfully
pipelines all these loops.
Reviewers: bcahoon
Reviewed By: bcahoon
Subscribers: Ayal, mgrang, llvm-commits
Differential Revision: https://reviews.llvm.org/D43620
llvm-svn: 326925
This is a follow-up to r325169, this time for all types, not just HVX
vector types.
Disable this by default, since it's not always safe.
llvm-svn: 326915
Absence of memory operands is treated as "aliasing everything", so
dropping them is sufficient.
Recommit r326256 with a fixed testcase.
llvm-svn: 326262
In r322867, we introduced IsStandalone when printing MIR in -debug
output. The default behaviour for that was:
1) If any of MBB, MI, or MO are -debug-printed separately, don't omit any
redundant information.
2) When -debug-printing a MF entirely, don't print any redundant
information.
3) When printing MIR, don't print any redundant information.
I'd like to change 2) to:
2) When -debug-printing a MF entirely, don't omit any redundant information.
Differential Revision: https://reviews.llvm.org/D43337
llvm-svn: 326094
Enable multiple COPY hints to eliminate more COPYs during register allocation.
Note that this is something all targets should do, see
https://reviews.llvm.org/D38128.
Review: Krzysztof Parzyszek
llvm-svn: 325697
This started by noticing that scalar and vector types were producing different results with div ops in PR36305:
https://bugs.llvm.org/show_bug.cgi?id=36305
...but the problem is bigger. I couldn't keep it straight without a table, so I'm attaching that as a PDF to
the review. The x86 tests in undef-ops.ll correspond to that table.
Green means that instsimplify and the DAG agree on the result for all types.
Red means the DAG was returning undef when IR was not.
Yellow means the DAG was returning a non-undef result when IR returned undef.
This patch assumes that we're currently doing the right thing in IR.
Note: I couldn't find any problems with lowering vector constants as the code comments were warning,
but those comments were written long ago in rL36413 .
Differential Revision: https://reviews.llvm.org/D43141
llvm-svn: 324941
Add a common -trap-unreachable option, similar to the target
specific hexagon equivalent, which has been replaced. This
turns unreachable instructions into traps, which is useful for
debugging.
Differential Revision: https://reviews.llvm.org/D42965
llvm-svn: 324880
Instead of:
%bb.1: derived from LLVM BB %for.body
print:
bb.1.for.body:
Also use MIR syntax for MBB attributes like "align", "landing-pad", etc.
llvm-svn: 324563
Vector pairs are legal types, but not every operation can work on pairs.
For those operations that are legal for single vectors, generate a concat
of their results on pair halves.
llvm-svn: 324350
It was expanded directly into instructions earlier. That was to avoid
loads from a constant pool for a vector negation: "xor x, splat(i1 -1)".
Implement ISD opcodes QTRUE and QFALSE to denote logical vectors of
all true and all false values, and handle setcc with negations through
selection patterns.
llvm-svn: 324348
Discussed here:
http://lists.llvm.org/pipermail/llvm-dev/2018-January/120320.html
In preparation for adding support for named vregs we are changing the sigil for
physical registers in MIR to '$' from '%'. This will prevent name clashes of
named physical register with named vregs.
llvm-svn: 323922
Selecting of constant HVX vectors involves some "manual processing",
which mishandled an unrelated BITCAST operation causing a selection
error.
llvm-svn: 323887
Instructions like memd(r0+##global+1) are legal as long as the entire
address is properly aligned. Assuming that "global" is aligned at an
8-byte boundary, the expression "global+1" appears to be misaligned.
Handle such cases in HexagonConstExtenders, and make sure that any non-
extended offsets generated are still aligned accordingly.
llvm-svn: 323799
This reverts r323562, since it wasn't actually necessary. Constant-
extended offsets do not need to be aligned, as long as the effective
address is aligned.
Keep the testcase, with a modification which checks that such offsets
are not unnecessarily avoided.
llvm-svn: 323798
A correctly aligned address may happen to be separated into a variable
part and a constant part, where the constant part does not match the
alignment needed in a load/store that uses this address. Such a constant
cannot be used as an immediate offset in an indexed instruction.
When lowering a global address, make sure that if there is an offset
folded into the global, the offset is valid for all uses in load/store
instructions.
llvm-svn: 323562
In addition to that, make sure that there are no boolean vector types that
are associated with multiple register classes. Specifically, remove v32i1
and v64i1 from integer register classes. These types will correspond to
results of vector comparisons, and as such should belong to the vector
predicate class. Having them in scalar registers as well makes legalization
ambiguous.
llvm-svn: 323229
Shuffle generation uses vmux to collapse vectors resulting from two
individual shuffles into one. The indexes of the elements selected
from the first operand were indicated by 0xFF in the constant vector
used in the compare instruction, but the compare (veqb) set the bits
corresponding to the 0x00 elements, thus inverting the selection.
Reverse the order of operands to vmux to get the correct output.
llvm-svn: 320516
Summary:
Add isRenamable() predicate to MachineOperand. This predicate can be
used by machine passes after register allocation to determine whether it
is safe to rename a given register operand. Register operands that
aren't marked as renamable may be required to be assigned their current
register to satisfy constraints that are not captured by the machine
IR (e.g. ABI or ISA constraints).
Reviewers: qcolombet, MatzeB, hfinkel
Subscribers: nemanjai, mcrosier, javed.absar, llvm-commits
Differential Revision: https://reviews.llvm.org/D39400
llvm-svn: 320503
A wrong type was passed to insertVector, causing an out-of-bounds value
to be added an an operand to HexagonISD::INSERT. This later failed in
instruction selection.
llvm-svn: 320369
As part of the unification of the debug format and the MIR format, print
MBB references as '%bb.5'.
The MIR printer prints the IR name of a MBB only for block definitions.
* find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)->getNumber\(\)/" << printMBBReference(*\1)/g'
* find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#" << ([a-zA-Z0-9_]+)\.getNumber\(\)/" << printMBBReference(\1)/g'
* find . \( -name "*.txt" -o -name "*.s" -o -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E 's/BB#([0-9]+)/%bb.\1/g'
* grep -nr 'BB#' and fix
Differential Revision: https://reviews.llvm.org/D40422
llvm-svn: 319665
output
As part of the unification of the debug format and the MIR format,
always use `printReg` to print all kinds of registers.
Updated the tests using '_' instead of '%noreg' until we decide which
one we want to be the default one.
Differential Revision: https://reviews.llvm.org/D40421
llvm-svn: 319445
As part of the unification of the debug format and the MIR format, avoid
printing "vreg" for virtual registers (which is one of the current MIR
possibilities).
Basically:
* find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E "s/%vreg([0-9]+)/%\1/g"
* grep -nr '%vreg' . and fix if needed
* find . \( -name "*.mir" -o -name "*.cpp" -o -name "*.h" -o -name "*.ll" \) -type f -print0 | xargs -0 sed -i '' -E "s/ vreg([0-9]+)/ %\1/g"
* grep -nr 'vreg[0-9]\+' . and fix if needed
Differential Revision: https://reviews.llvm.org/D40420
llvm-svn: 319427
Change LowerBUILD_VECTOR to use those functions. This commit will tempora-
rily affect constant vector generation (it will generate constant-extended
values instead of non-extended combines), but the code for the general case
should be better. The constant selection part will be fixed later.
llvm-svn: 318877
In getOffsetRange, Max can be set to 0 to force the extender replacement
to be at or below the original value. This would cause the new offset to
be non-negative, which is preferred for memory instructions (to reduce
the likelihood of it getting constant-extended due to predication). The
problem happens when the range is shifted by an offset (present in the
instruction being examined) and the offset is negative. The entire range
for the allowable deviation will then be strictly negative. This creates
a problem, since 0 is assumed to be a valid deviation.
llvm-svn: 316601
This updates the MIRPrinter to include the regclass when printing
virtual register defs, which is already valid syntax for the
parser. That is, given 64 bit %0 and %1 in a "gpr" regbank,
%1(s64) = COPY %0(s64)
would now be written as
%1:gpr(s64) = COPY %0(s64)
While this change alone introduces a bit of redundancy with the
registers block, it allows us to update the tests to be more concise
and understandable and brings us closer to being able to remove the
registers block completely.
Note: We generally only print the class in defs, but there is one
exception. If there are uses without any defs whatsoever, we'll print
the class on all uses. I'm not completely convinced this comes up in
meaningful machine IR, but for now the MIRParser and MachineVerifier
both accept that kind of stuff, so we don't want to have a situation
where we can print something we can't parse.
llvm-svn: 316479
In HexagonISelLowering, there is code to handle the case when
a function returns an i1 type. In this case, we need to generate
extra nodes to copy the result from R0 to a predicate register.
The code was returning the wrong value for the chain edge which
caused an assert "Wrong topological sorting" when converting the
instructions to MIs.
This patch fixes the problem by returning the chain for the final
copy.
Patch by Brendon Cahoon.
llvm-svn: 316367
Normally, if the registers holding the induction variable's bounds
are redefined inside of the loop's body, the loop cannot be converted
to a hardware loop. However, if the redefining instruction is actually
loading an immediate value into the register, this conversion is both
possible and legal (since the immediate itself will be used in the
loop setup in the preheader).
llvm-svn: 316218
This patch lets the llvm tools handle the new HVX target features that
are added by frontend (clang). The target-features are of the form
"hvx-length64b" for 64 Byte HVX mode, "hvx-length128b" for 128 Byte mode HVX.
"hvx-double" is an alias to "hvx-length128b" and is soon will be deprecated.
The hvx version target feature is upgated form "+hvx" to "+hvxv{version_number}.
Eg: "+hvxv62"
For the correct HVX code generation, the user must use the following
target features.
For 64B mode: "+hvxv62" "+hvx-length64b"
For 128B mode: "+hvxv62" "+hvx-length128b"
Clang picks a default length if none is specified. If for some reason,
no hvx-length is specified to llvm, the compilation will bail out.
There is a corresponding clang patch.
Differential Revision: https://reviews.llvm.org/D38851
llvm-svn: 316101
Each constant extender requires an extra instruction, which adds to the
code size and also reduces the number of available slots in an instruction
packet. In most cases, the value of a repeated constant extender could be
loaded into a register, and the instructions using the extender could be
replaced with their counterparts that use that register instead.
This patch adds a pass that tries to reduce the number of constant
extenders, including extenders which differ only in an immediate offset
known at compile time, e.g. @global and @global+12.
llvm-svn: 315735
The pipeliner is generating a serial sequence that causes poor
register allocation when a post-increment instruction appears
prior to the use of the post-increment register. This occurs when
there is a circular set of dependences involved with a sequence
of instructions in the same cycle. In this case, there is no
serialization of the parallel semantics that will not cause an
additional register to be allocated.
This patch fixes the problem by changing the instructions so that
the post-increment instruction is used by the subsequent
instruction, which enables the register allocator to make a
better decision and not require another register.
Patch by Brendon Cahoon.
llvm-svn: 315466
If the two instructions being compared for equivalence have corresponding operands
that are integer constants, then check their values to determine equivalence.
Patch by Suyog Sarda!
llvm-svn: 314642
This patch produces a crash and hexagon_vector_loop_carried_reuse_constant.ll test fails on Windows (llvm-clang-x86_64-expensive-checks-win build bot).
llvm-svn: 314361
If the two instructions being compared for equivalence have corresponding operands
that are integer constants, then check their values to determine equivalence.
Patch by Suyog Sarda!
llvm-svn: 313993
