last night, breaking crafty and twolf. Make sure that the newly found
legal nodes are themselves not re-legalized until the next iteration.
Also, since this functionality exists now, we can reduce number of legalizer
iterations by depending on this behavior instead of having to misuse 'do
another iteration' to get the same effect.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@24875 91177308-0d34-0410-b5e6-96231b3b80d8
load. This reduces number of worklist iterations and avoid missing optimizations
depending on folding of things into sext_inreg nodes (which aren't supported by
all targets).
Tested by Regression/CodeGen/X86/extend.ll:test2
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@24712 91177308-0d34-0410-b5e6-96231b3b80d8
Allow (zext (truncate)) to apply after legalize if the target supports
AND (which all do).
This compiles
short %foo() {
%tmp.0 = load ubyte* %X ; <ubyte> [#uses=1]
%tmp.3 = cast ubyte %tmp.0 to short ; <short> [#uses=1]
ret short %tmp.3
}
to:
_foo:
movzbl _X, %eax
ret
instead of:
_foo:
movzbl _X, %eax
movzbl %al, %eax
ret
thanks to Evan for pointing this out.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@24709 91177308-0d34-0410-b5e6-96231b3b80d8
ConstantVec legalizing code, which would return constantpool nodes that
were not of the target's pointer type.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@24691 91177308-0d34-0410-b5e6-96231b3b80d8
1. Only forward subst offsets into loads and stores, not into arbitrary
things, where it will likely become a load.
2. If the source is a cast from pointer, forward subst the cast as well,
allowing us to fold the cast away (improving cases when the cast is
from an alloca or global).
This hasn't been fully tested, but does appear to further reduce register
pressure and improve code. Lets let the testers grind on it a bit. :)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@24640 91177308-0d34-0410-b5e6-96231b3b80d8
selecting a node and use a mix of getTargetNode() and SelectNodeTo. Because
SelectNodeTo didn't check the CSE maps for a preexisting node and didn't insert
its result into the CSE maps, we would sometimes miss a CSE opportunity.
This is extremely rare, but worth fixing for completeness.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@24565 91177308-0d34-0410-b5e6-96231b3b80d8
changes allow us to generate the following code:
_foo:
li r2, 0
lvx v0, r2, r3
vaddfp v0, v0, v0
stvx v0, r2, r3
blr
for this llvm:
void %foo(<4 x float>* %a) {
entry:
%tmp1 = load <4 x float>* %a
%tmp2 = add <4 x float> %tmp1, %tmp1
store <4 x float> %tmp2, <4 x float>* %a
ret void
}
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@24534 91177308-0d34-0410-b5e6-96231b3b80d8
file to become corrupted due to interactions between mmap'd memory segments
and file descriptors closing. The problem is completely avoiding by using
a third temporary file.
Patch provided by Evan Jones
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@24527 91177308-0d34-0410-b5e6-96231b3b80d8
generates it. Make MVT::Vector expand-only, and remove the code in
Legalize that attempts to legalize it.
The plan for supporting N x Type is to continually epxand it in ExpandOp
until it gets down to 2 x Type, where it will be scalarized into a pair of
scalars.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@24482 91177308-0d34-0410-b5e6-96231b3b80d8
packed types with an element count of 1, although more generic support is
coming. This allows LLVM to turn the following code:
void %foo(<1 x float> * %a) {
entry:
%tmp1 = load <1 x float> * %a;
%tmp2 = add <1 x float> %tmp1, %tmp1
store <1 x float> %tmp2, <1 x float> *%a
ret void
}
Into:
_foo:
lfs f0, 0(r3)
fadds f0, f0, f0
stfs f0, 0(r3)
blr
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@24416 91177308-0d34-0410-b5e6-96231b3b80d8
(our graphs apparently already fit within the cache on my g5). In any case
this reduces memory usage.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@24249 91177308-0d34-0410-b5e6-96231b3b80d8
set and eliminating the need to iterate whenever something is removed (which
can be really slow in some cases). Thx to Jim for pointing out something silly
I was getting stuck on. :)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@24241 91177308-0d34-0410-b5e6-96231b3b80d8
alignment information appropriately. Includes code for PowerPC to support
fixed-size allocas with alignment larger than the stack. Support for
arbitrarily aligned dynamic allocas coming soon.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@24224 91177308-0d34-0410-b5e6-96231b3b80d8
a special case hack for X86, make the hack more general: if an incoming argument
register is not used in any block other than the entry block, don't copy it to
a vreg. This helps us compile code like this:
%struct.foo = type { int, int, [0 x ubyte] }
int %test(%struct.foo* %X) {
%tmp1 = getelementptr %struct.foo* %X, int 0, uint 2, int 100
%tmp = load ubyte* %tmp1 ; <ubyte> [#uses=1]
%tmp2 = cast ubyte %tmp to int ; <int> [#uses=1]
ret int %tmp2
}
to:
_test:
lbz r3, 108(r3)
blr
instead of:
_test:
lbz r2, 108(r3)
or r3, r2, r2
blr
The (dead) copy emitted to copy r3 into a vreg for extra-block uses was
increasing the live range of r3 past the load, preventing the coallescing.
This implements CodeGen/PowerPC/reg-coallesce-simple.ll
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@24115 91177308-0d34-0410-b5e6-96231b3b80d8
generating results in vregs that will need them. In the case of something
like this: CopyToReg((add X, Y), reg1024), we no longer emit code like
this:
reg1025 = add X, Y
reg1024 = reg 1025
Instead, we emit:
reg1024 = add X, Y
Whoa! :)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@24111 91177308-0d34-0410-b5e6-96231b3b80d8
FP_TO_SINT is preferred to a larger FP_TO_UINT. This seems to be begging
for a TLI.isOperationCustom() helper function.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23992 91177308-0d34-0410-b5e6-96231b3b80d8
pointer marking the end of the list, the zero *must* be cast to the pointer
type. An un-cast zero is a 32-bit int, and at least on x86_64, gcc will
not extend the zero to 64 bits, thus allowing the upper 32 bits to be
random junk.
The new END_WITH_NULL macro may be used to annotate a such a function
so that GCC (version 4 or newer) will detect the use of un-casted zero
at compile time.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23888 91177308-0d34-0410-b5e6-96231b3b80d8
the input is that type, this caused a failure on gs on X86 last night.
Move the hard checks into Build[US]Div since that is where decisions like
this should be made.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23881 91177308-0d34-0410-b5e6-96231b3b80d8
For example, we can now join things like [0-30:0)[31-40:1)[52-59:2)
with [40:60:0) if the 52-59 range is defined by a copy from the 40-60 range.
The resultant range ends up being [0-30:0)[31-60:1).
This fires a lot through-out the test suite (e.g. shrinking bc from
19492 -> 18509 machineinstrs) though most gains are smaller (e.g. about
50 copies eliminated from crafty).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23866 91177308-0d34-0410-b5e6-96231b3b80d8
(an unused method).
Fix the merger so that it can merge ranges like this [10:12)[16:40) with
[12:38) into [10:40) instead of bogus ranges. This sort of input will be
possible for the merger coming shortly
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23865 91177308-0d34-0410-b5e6-96231b3b80d8
Add a new flag to TargetLowering indicating if the target has really cheap
signed division by powers of two, make ppc use it. This will probably go
away in the future.
Implement some more ISD::SDIV folds in the dag combiner
Remove now dead code in the x86 backend.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23853 91177308-0d34-0410-b5e6-96231b3b80d8
for types that aren't legal, and fail a divisor is less than zero
comparison, which would cause us to drop a subtract.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23846 91177308-0d34-0410-b5e6-96231b3b80d8
Fix a *bug* in the extendIntervalEndTo method. In particular, if adding
[2:10) to an interval containing [0:2),[10:30), we produced [0:10),[10,30).
Which is not the most smart thing to do. Now produce [0:30).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23841 91177308-0d34-0410-b5e6-96231b3b80d8
that the nodes can be folded with other nodes, and we can not duplicate
code in every backend. Alpha will probably want this too.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23835 91177308-0d34-0410-b5e6-96231b3b80d8
allows us to lower legal return types to something else, to meet ABI
requirements (such as that i64 be returned in two i32 regs on Darwin/ppc).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23802 91177308-0d34-0410-b5e6-96231b3b80d8
a lot throughout many programs. In particular, specfp triggers it a bunch for
constant FP nodes when you have code like cond ? 1.0 : -1.0.
If the PPC ISel exposed the loads implicit in pic references to external globals,
we would be able to eliminate a load in cases like this as well:
%X = external global int
%Y = external global int
int* %test4(bool %C) {
%G = select bool %C, int* %X, int* %Y
ret int* %G
}
Note that this breaks things that use SrcValue's (see the fixme), but since nothing
uses them yet, this is ok.
Also, simplify some code to use hasOneUse() on an SDOperand instead of hasNUsesOfValue directly.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23781 91177308-0d34-0410-b5e6-96231b3b80d8
you could be AND'ing with the result of a shift that shifts out all the
bits you care about, in addition to a constant.
Also, move over an add/sub_parts fold from legalize to the dag combiner,
where it works for things other than constants. Woot!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23720 91177308-0d34-0410-b5e6-96231b3b80d8
out, where after the first CombineTo() call, the node the second CombineTo
wishes to replace may no longer exist.
Fix a very real bug with the truncated load optimization on little endian
targets, which do not need a byte offset added to the load.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23704 91177308-0d34-0410-b5e6-96231b3b80d8
like turning:
_foo:
fctiwz f0, f1
stfd f0, -8(r1)
lwz r2, -4(r1)
rlwinm r3, r2, 0, 16, 31
blr
into
_foo:
fctiwz f0,f1
stfd f0,-8(r1)
lhz r3,-2(r1)
blr
Also removed an unncessary constraint from sra -> srl conversion, which
should take care of hte only reason we would ever need to handle sra in
MaskedValueIsZero, AFAIK.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23703 91177308-0d34-0410-b5e6-96231b3b80d8
location, replace them with a new store of the last value. This occurs
in the same neighborhood in 197.parser, speeding it up about 1.5%
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23691 91177308-0d34-0410-b5e6-96231b3b80d8
multiple results.
Use this support to implement trivial store->load forwarding, implementing
CodeGen/PowerPC/store-load-fwd.ll. Though this is the most simple case and
can be extended in the future, it is still useful. For example, it speeds
up 197.parser by 6.2% by avoiding an LSU reject in xalloc:
stw r6, lo16(l5_end_of_array)(r2)
addi r2, r5, -4
stwx r5, r4, r2
- lwzx r5, r4, r2
- rlwinm r5, r5, 0, 0, 30
stwx r5, r4, r2
lwz r2, -4(r4)
ori r2, r2, 1
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23690 91177308-0d34-0410-b5e6-96231b3b80d8
creating a new vreg and inserting a copy: just use the input vreg directly.
This speeds up the compile (e.g. about 5% on mesa with a debug build of llc)
by not adding a bunch of copies and vregs to be coallesced away. On mesa,
for example, this reduces the number of intervals from 168601 to 129040
going into the coallescer.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23671 91177308-0d34-0410-b5e6-96231b3b80d8
previous copy elisions and we discover we need to reload a register, make
sure to use the regclass of the original register for the reload, not the
class of the current register. This avoid using 16-bit loads to reload 32-bit
values.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23645 91177308-0d34-0410-b5e6-96231b3b80d8
store r12 -> [ss#2]
R3 = load [ss#1]
use R3
R3 = load [ss#2]
R4 = load [ss#1]
and turn it into this code:
store R12 -> [ss#2]
R3 = load [ss#1]
use R3
R3 = R12
R4 = R3 <- oops!
The problem was that promoting R3 = load[ss#2] to a copy missed the fact that
the instruction invalidated R3 at that point.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23638 91177308-0d34-0410-b5e6-96231b3b80d8
with the dag combiner. This speeds up espresso by 8%, reaching performance
parity with the dag-combiner-disabled llc.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23636 91177308-0d34-0410-b5e6-96231b3b80d8
dead node elim and dag combiner passes where the root is potentially updated.
This fixes a fixme in the dag combiner.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23634 91177308-0d34-0410-b5e6-96231b3b80d8
that testcase still does not pass with the dag combiner. This is because
not all forms of br* are folded yet.
Also, when we combine a node into another one, delete the node immediately
instead of waiting for the node to potentially come up in the future.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23632 91177308-0d34-0410-b5e6-96231b3b80d8
Since calls return more than one value, don't bail if one of their uses
happens to be a node that's not an MVT::Other when following the chain
from CALLSEQ_START to CALLSEQ_END.
Once we've found a CALLSEQ_START, we can just return; there's no need to
tail-recurse further up the graph.
Most importantly, just because something only has one use doesn't mean we
should use it's one use to follow from start to end. This faulty logic
caused us to follow a chain of one-use FP operations back to a much earlier
call, putting a cycle in the graph from a later start to an earlier end.
This is a better fix that reverting to the workaround committed earlier
today.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23620 91177308-0d34-0410-b5e6-96231b3b80d8
Neither of us have yet figured out why this code is necessary, but stuff
breaks if its not there. Still tracking this down...
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23617 91177308-0d34-0410-b5e6-96231b3b80d8
large basic blocks because it was purely recursive. This switches it to an
iterative/recursive hybrid.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23596 91177308-0d34-0410-b5e6-96231b3b80d8
For instructions that define multiple results, use the right regclass
to define the result, not always the rc of result #0
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23580 91177308-0d34-0410-b5e6-96231b3b80d8
2. Added node groups to handle flagged nodes.
3. Started weaning simple scheduling off existing emitter.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23566 91177308-0d34-0410-b5e6-96231b3b80d8
Though I have done extensive testing, it is possible that this will break
things in configs I can't test. Please let me know if this causes a problem
and I'll fix it ASAP.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23504 91177308-0d34-0410-b5e6-96231b3b80d8
for testing and will require target machine info to do a proper scheduling.
The simple scheduler can be turned on using -sched=simple (defaults
to -sched=none)
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23455 91177308-0d34-0410-b5e6-96231b3b80d8
This happens all the time on PPC for bool values, e.g. eliminating a xori
in inverted-bool-compares.ll.
This should be added to the dag combiner as well.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23403 91177308-0d34-0410-b5e6-96231b3b80d8
when possible, avoiding the load (and avoiding the copy if the value is already
in the right register).
This patch came about when I noticed code like the following being generated:
store R17 -> [SS1]
...blah...
R4 = load [SS1]
This was causing an LSU reject on the G5. This problem was due to the register
allocator folding spill code into a reg-reg copy (producing the load), which
prevented the spiller from being able to rewrite the load into a copy, despite
the fact that the value was already available in a register. In the case
above, we now rip out the R4 load and replace it with a R4 = R17 copy.
This speeds up several programs on X86 (which spills a lot :) ), e.g.
smg2k from 22.39->20.60s, povray from 12.93->12.66s, 168.wupwise from
68.54->53.83s (!), 197.parser from 7.33->6.62s (!), etc. This may have a larger
impact in some cases on the G5 (by avoiding LSU rejects), though it probably
won't trigger as often (less spilling in general).
Targets that implement folding of loads/stores into copies should implement
the isLoadFromStackSlot hook to get this.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23388 91177308-0d34-0410-b5e6-96231b3b80d8
select (x < y), 1, 0 -> (x < y) incorrectly: the setcc returns i1 but the
select returned i32. Add the zero extend as needed.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23301 91177308-0d34-0410-b5e6-96231b3b80d8
are, simplify logic, and cause things to not be nested as deeply. This also
uses MRI->areAliases instead of an explicit loop.
No functionality change, just code cleanup.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23296 91177308-0d34-0410-b5e6-96231b3b80d8
only add a reload live range once for the instruction. This is one step
towards fixing a regalloc pessimization that Nate notice, but is later undone
by the spiller (so no code is changed).
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23293 91177308-0d34-0410-b5e6-96231b3b80d8
This restores all of stanford to being identical with and without the dag
combiner with the add folding turned off in sd.cpp.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23258 91177308-0d34-0410-b5e6-96231b3b80d8
we were losing a node, causing an assertion to fail. Now we eagerly delete
discovered CSE's, and provide an optional vector to keep track of these
discovered equivalences.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23255 91177308-0d34-0410-b5e6-96231b3b80d8
I have run so far when run before Legalize. It still needs to pick up the
SetCC folds, and nodes that use SetCC.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23243 91177308-0d34-0410-b5e6-96231b3b80d8
i64 values on targets that need that expanded to 32-bit registers. This fixes
PowerPC/2005-09-02-LegalizeDuplicatesCalls.ll and speeds up 189.lucas from
taking 122.72s to 81.96s on my desktop.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23228 91177308-0d34-0410-b5e6-96231b3b80d8
from the binary ops map, even if they had multiple results. This latent bug
caused a few failures with the dag isel last night.
To prevent stuff like this from happening in the future, add some really
strict checking to make sure that the CSE maps always match up with reality!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23221 91177308-0d34-0410-b5e6-96231b3b80d8
instead of ZERO_EXTEND to eliminate extraneous extensions. This eliminates
dead zero extensions on formal arguments and other cases on PPC, implementing
the newly tightened up test/Regression/CodeGen/PowerPC/small-arguments.ll test.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@23205 91177308-0d34-0410-b5e6-96231b3b80d8