in local register allocator. If a reg-reg copy has a phys reg
input and a virt reg output, and this is the last use of the phys
reg, assign the phys reg to the virt reg. If a reg-reg copy has
a phys reg output and we need to reload its spilled input, reload
it directly into the phys reg than passing it through another reg.
Following 76208, there is sometimes no dependency between the def of
a phys reg and its use; this creates a window where that phys reg
can be used for spilling (this is true in linear scan also). This
is bad and needs to be fixed a better way, although 76208 works too
well in practice to be reverted. However, there should normally be
no spilling within inline asm blocks. The patch here goes a long way
towards making this actually be true.
llvm-svn: 91485
found last time. Instead of trying to modify the IR while iterating over it,
I've change it to keep a list of WeakVH references to dead instructions, and
then delete those instructions later. I also added some special case code to
detect and handle the situation when both operands of a memcpy intrinsic are
referencing the same alloca.
llvm-svn: 91459
Checks that the code generated by 'tblgen --emit-llvmc' can be actually
compiled. Also fixes two bugs found in this way:
- forward_transformed_value didn't work with non-list arguments
- cl::ZeroOrOne is now called cl::Optional
llvm-svn: 91404
1. Only perform (zext (shl (zext x), y)) -> (shl (zext x), y) when y is a constant. This makes sure it remove at least one zest.
2. If the shift is a left shift, make sure the original shift cannot shift out bits.
llvm-svn: 91399
While scanning through the uses of an alloca, keep track of the current offset
relative to the start of the alloca, and check memory references to see if
the offset & size correspond to a component within the alloca. This has the
nice benefit of unifying much of the code from isSafeUseOfAllocation,
isSafeElementUse, and isSafeUseOfBitCastedAllocation. The code to rewrite
the uses of a promoted alloca, after it is determined to be safe, is
reorganized in the same way.
Also, when rewriting GEP instructions, mark them as "in-bounds" since all the
indices are known to be safe.
llvm-svn: 91184
value size. This only manifested when memdep inprecisely returns clobber,
which is do to a caching issue in the PR5744 testcase. We can 'efficiently
emulate' this by using '-no-aa'
llvm-svn: 91004
add, there is no need to scan the world to find the same add again.
This invalidates the previous testcase, which wasn't wonderful anyway,
because it needed a run of instcombine to permute the use-lists in
just the right way to before GVN was run (so it was really fragile).
Not a big loss.
llvm-svn: 90973
phi translation of complex expressions like &A[i+1]. This has the
following benefits:
1. The phi translation logic is all contained in its own class with
a strong interface and verification that it is self consistent.
2. The logic is more correct than before. Previously, if intermediate
expressions got PHI translated, we'd miss the update and scan for
the wrong pointers in predecessor blocks. @phi_trans2 is a testcase
for this.
3. We have a lot less code in memdep.
We can handle phi translation across blocks of things like @phi_trans3,
which is pretty insane :).
This patch should fix the miscompiles of 255.vortex, and I tested it
with a bootstrap of llvm-gcc, llvm-test and dejagnu of course.
llvm-svn: 90926
The coalescer is supposed to clean these up, but when setting up parameters
for a function call, there may be copies to physregs. If the defining
instruction has been LICM'ed far away, the coalescer won't touch it.
The register allocation hint does not always work - when the register
allocator is backtracking, it clears the hints.
This patch takes care of a few more cases that r90163 missed.
llvm-svn: 90502
both source operands. In the canonical form, the 2nd operand is changed to an
undef and the shuffle mask is adjusted to only reference elements from the 1st
operand. Radar 7434842.
llvm-svn: 90417
- A valno should be set HasRedefByEC if there is an early clobber def in the middle of its live ranges. It should not be set if the def of the valno is defined by an early clobber.
- If a physical register def is tied to an use and it's an early clobber, it just means the HasRedefByEC is set since it's still one continuous live range.
- Add a couple of missing checks for HasRedefByEC in the coalescer. In general, it should not coalesce a vr with a physical register if the physical register has a early clobber def somewhere. This is overly conservative but that's the price for using such a nasty inline asm "feature".
llvm-svn: 90269
This means that well connected blocks are copy coalesced before the less connected blocks. Connected blocks are more difficult to
coalesce because intervals are more complicated, so handling them first gives a greater chance of success.
llvm-svn: 90194
This helps us avoid silly copies when rematting values that are copied to a physical register:
leaq _.str44(%rip), %rcx
movq %rcx, %rsi
call _strcmp
becomes:
leaq _.str44(%rip), %rsi
call _strcmp
The coalescer will not touch the movq because that would tie down the physical register.
llvm-svn: 90163
more. Update the syntax we're checking for and filecheckize it too.
This will fix the selfhost buildbots but will 'break' the others (sigh) because
they're still linked against older LLVM which is emitting less optimized IR.
llvm-svn: 90104
handle cases like this:
void test(int N, double* G) {
long j;
for (j = 1; j < N - 1; j++)
G[j+1] = G[j] + G[j+1];
}
where G[1] isn't live into the loop.
llvm-svn: 90041
translation of add with immediate. This allows us
to optimize this function:
void test(int N, double* G) {
long j;
G[1] = 1;
for (j = 1; j < N - 1; j++)
G[j+1] = G[j] + G[j+1];
}
to only do one load every iteration of the loop.
llvm-svn: 90013
array indexes. The "complex" case of SRoA still handles them, and correctly.
This fixes a weirdness where we'd correctly avoid transforming A[0][42] if
the 42 was too large, but we'd only do it if it was one gep, not two separate
ones.
llvm-svn: 90007
the problem only shows for msp430 and pic16 which is why it specifies
them using -march. But it is wrong to put such tests in CodeGen/Generic,
since not everyone builds these targets. Put a copy of the test in each
of the target test directories.
llvm-svn: 90005
generates store to undef and some generates store to null as the idiom
for undefined behavior. Since simplifycfg zaps both, don't remove the
undefined behavior in instcombine.
llvm-svn: 89971
first expression as P+4+4*i which we considered to possibly alias
P+4*j. Now we correctly analyze the former one as P+1+4*i.
@test10 is a sanity test that verfies that we know that P+4+4*i != P+4*i.
llvm-svn: 89960
This violates the ABI (that area is "reserved"), and
while it is safe if all code is generated with current
compilers, there is some very old code around that uses
that slot for something else, and breaks if it is stored
into. Adjust testcases looking for current behavior.
I've verified that the stack frame size is right in all
testcases, whether it changed or not. 7311323.
llvm-svn: 89811
than doing the same via constpool:
1. Load from constpool costs 3 cycles on A9, movt/movw pair - just 2.
2. Load from constpool might stall up to 300 cycles due to cache miss.
3. Movt/movw does not use load/store unit.
4. Less constpool entries => better compiler performance.
This is only enabled on ELF systems, since darwin does not have needed
relocations (yet).
llvm-svn: 89720
ConstantExpr, not just the top-level operator. This allows it to
fold many more constants.
Also, make GlobalOpt call ConstantFoldConstantExpression on
GlobalVariable initializers.
llvm-svn: 89659
values, resolving references to them, and then removing the definitions.
If a template argument is set to an undefined value, we need to resolve
references to that argument to an explicit undefined value. The current code
leaves the reference to the template argument as it is, which causes an
assertion failure later when the definition of the template argument is
removed.
llvm-svn: 89581
it may be used in contexts where preheader insertion may have failed due
to an indirectbr.
Make LoopSimplify's LoopSimplify::SeparateNestedLoop properly fail in
the case that it would require splitting an indirectbr edge.
These fix PR5502.
llvm-svn: 89484
which was an expensive checks failure due to a bug in the checking. This
patch in essence reverts the original fix for PR3393, and refixes it by a
tweak to the way expensive checking is done.
llvm-svn: 89454
if it is not ultimately captured. Teach BasicAliasAnalysis that a
local object address which does not escape and is never stored does
not alias with a value resulting from a load.
llvm-svn: 89398
they are lowered to instruction sequences more complex than a simple
load, such that CodeGen cannot rematerialize them, a reload from a
spill slot is likely to be cheaper than the complex sequence.
llvm-svn: 89374
When TwoAddressInstructionPass deletes a dead instruction, make sure that all
register kills are accounted for. The 2-addr register does not get special
treatment.
llvm-svn: 89246
The local register allocator doesn't like it when LiveVariables is run.
We should also disable edge splitting under -O0, but that has to wait a bit.
llvm-svn: 89125
address space (though it only uses a small fraction of that), and the
buildbots disallow that.
Also add a comment to the Makefile's ulimit line warning future
developers that changing it won't work.
llvm-svn: 88994
The large code model is documented at
http://www.x86-64.org/documentation/abi.pdf and says that calls should
assume their target doesn't live within the 32-bit pc-relative offset
that fits in the call instruction.
To do this, we turn off the global-address->target-global-address
conversion in X86TargetLowering::LowerCall(). The first attempt at
this broke the lazy JIT because it can separate the movabs(imm->reg)
from the actual call instruction. The lazy JIT receives the address of
the movabs as a relocation and needs to record the return address from
the call; and then when that call happens, it needs to patch the
movabs with the newly-compiled target. We could thread the call
instruction into the relocation and record the movabs<->call mapping
explicitly, but that seems to require at least as much new
complication in the code generator as this change.
To fix this, we make lazy functions _always_ go through a call
stub. You'd think we'd only have to force lazy calls through a stub on
difficult platforms, but that turns out to break indirect calls
through a function pointer. The right fix for that is to distinguish
between calls and address-of operations on uncompiled functions, but
that's complex enough to leave for someone else to do.
Another attempt at this defined a new CALL64i pseudo-instruction,
which expanded to a 2-instruction sequence in the assembly output and
was special-cased in the X86CodeEmitter's emitInstruction()
function. That broke indirect calls in the same way as above.
This patch also removes a hack forcing Darwin to the small code model.
Without far-call-stubs, the small code model requires things of the
JITMemoryManager that the DefaultJITMemoryManager can't provide.
Thanks to echristo for lots of testing!
llvm-svn: 88984
Have the asm printer emit a comment if an instruction is a spill or
reload and have the spiller mark copies it introdues so the asm printer
can also annotate those.
llvm-svn: 88911
code-size win, and not when it's only likely to be code-size neutral,
such as when only a single instruction would be eliminated and a new
branch would be required.
This fixes rdar://7392894.
llvm-svn: 88692
D0<def,dead> = ...
...
= S0<use, kill>
S0<def> = ...
...
D0<def> =
The first D0 def is correctly marked dead, however, livevariables should have
added an implicit def of S0 or we end up with a use without a def.
llvm-svn: 88690
running IPSCCP early, and we run functionattrs interlaced with the inliner,
we often (particularly for small or noop functions) completely propagate
all of the information about a call to its call site in IPSSCP (making a call
dead) and functionattrs is smart enough to realize that the function is
readonly (because it is interlaced with inliner).
To improve compile time and make the inliner threshold more accurate, realize
that we don't have to inline dead readonly function calls. Instead, just
delete the call. This happens all the time for C++ codes, here are some
counters from opt/llvm-ld counting the number of times calls were deleted vs
inlined on various apps:
Tramp3d opt:
5033 inline - Number of call sites deleted, not inlined
24596 inline - Number of functions inlined
llvm-ld:
667 inline - Number of functions deleted because all callers found
699 inline - Number of functions inlined
483.xalancbmk opt:
8096 inline - Number of call sites deleted, not inlined
62528 inline - Number of functions inlined
llvm-ld:
217 inline - Number of allocas merged together
2158 inline - Number of functions inlined
471.omnetpp:
331 inline - Number of call sites deleted, not inlined
8981 inline - Number of functions inlined
llvm-ld:
171 inline - Number of functions deleted because all callers found
629 inline - Number of functions inlined
Deleting a call is much faster than inlining it, and is insensitive to the
size of the callee. :)
llvm-svn: 86975
cannot be folded into target cmp instruction.
- Avoid a phase ordering issue where early cmp optimization would prevent the
later count-to-zero optimization.
- Add missing checks which could cause LSR to reuse stride that does not have
users.
- Fix a bug in count-to-zero optimization code which failed to find the pre-inc
iv's phi node.
- Remove, tighten, loosen some incorrect checks disable valid transformations.
- Quite a bit of code clean up.
llvm-svn: 86969
