| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
|
|
|
|
|
| |
This adds functionality to remove size/zero extension during indvars
without generating a canonical IV and rewriting all IV users. It's
disabled by default so should have no effect on codegen. Work in progress.
llvm-svn: 130829
|
| |
|
|
|
|
| |
coalescing.
llvm-svn: 130814
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
sub-registers.
LiveVariables doesn't understand that clobbering D0 and D1 completely overwrites
Q0, so if Q0 is live-in to a function, its live range will extend beyond a
function call that only clobbers D0 and D1. This shows up in the
ARM/2009-11-01-NeonMoves test case.
LiveVariables should probably implement the much stricter rules for physreg
liveness that RAFast imposes - a physreg is killed by the first use of any
alias.
llvm-svn: 130801
|
| |
|
|
|
|
|
|
|
| |
Only create a canonical IV for backedge taken count if it will
actually be used by LinearFunctionTestReplace. And some related
cleanup, preparing to reduce dependence on canonical IVs.
No significant effect on x86 or arm in the test-suite.
llvm-svn: 130799
|
| |
|
|
|
|
| |
DW_AT_MIPS_linkage_name. This helps gdb and fixes var-path-expr.exp regression reported by gdb testsuite.
llvm-svn: 130794
|
| |
|
|
|
|
|
| |
(and should thus never be done).
- Should fix a crash on win32.
llvm-svn: 130793
|
| |
|
|
|
|
| |
<rdar://problem/8460511>
llvm-svn: 130791
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Register coalescing can sometimes create live ranges that end in the middle of a
basic block without any killing instruction. When SplitKit detects this, it will
repair the live range by shrinking it to its uses.
Live range splitting also needs to know about this. When the range shrinks so
much that it becomes allocatable, live range splitting fails because it can't
find a good split point. It is paranoid about making progress, so an allocatable
range is considered an error.
The coalescer should really not be creating these bad live ranges. They appear
when coalescing dead copies.
llvm-svn: 130787
|
| |
|
|
|
|
|
|
| |
max(a,b) >= a -> true. According to my super-optimizer, these are
by far the most common simplifications (of the -instsimplify kind)
that occur in the testsuite and aren't caught by -std-compile-opts.
llvm-svn: 130780
|
| |
|
|
| |
llvm-svn: 130774
|
| |
|
|
| |
llvm-svn: 130766
|
| |
|
|
|
|
| |
it's possible.
llvm-svn: 130764
|
| |
|
|
| |
llvm-svn: 130763
|
| |
|
|
|
|
| |
does it now) then use them directly. This saves one scan of entire module, to collect debug info, which in turns saves few machine cycles at compile time.
llvm-svn: 130759
|
| |
|
|
| |
llvm-svn: 130756
|
| |
|
|
| |
llvm-svn: 130755
|
| |
|
|
| |
llvm-svn: 130749
|
| |
|
|
|
|
|
|
| |
string template.
Fixes rdar://8493866
llvm-svn: 130747
|
| |
|
|
|
|
|
|
| |
model constants which can be added to base registers via add-immediate
instructions which don't require an additional register to materialize
the immediate.
llvm-svn: 130743
|
| |
|
|
|
|
| |
pointer type for vector indices. Make the vector unrolling code respect that.
llvm-svn: 130733
|
| |
|
|
|
|
| |
Instead, just use whatever location info first non-phi instruction has.
llvm-svn: 130729
|
| |
|
|
| |
llvm-svn: 130718
|
| |
|
|
| |
llvm-svn: 130716
|
| |
|
|
| |
llvm-svn: 130715
|
| |
|
|
|
|
|
|
| |
Def operands may also have an <undef> flag, but that just means that a
sub-register redef doesn't actually read the super-register. For physical
registers, it has no meaning.
llvm-svn: 130714
|
| |
|
|
|
|
| |
a vector compare, generate a vector result rather than i1 (and crashing).
llvm-svn: 130706
|
| |
|
|
| |
llvm-svn: 130705
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
This works around a limitation in gdb which is reported by following inherit.exp test failures from gdb testsuite.
gdb.cp/inherit.exp: print g_vB.vB::vb
gdb.cp/inherit.exp: print g_vB.vB::vx
gdb.cp/inherit.exp: print g_vC.vC::vc
gdb.cp/inherit.exp: print g_vC.vC::vx
gdb.cp/inherit.exp: print g_vD.vB::vb
...
llvm-svn: 130702
|
| |
|
|
|
|
|
| |
This automagically provides a transform noticed by my super-optimizer
as occurring quite often: "rem x, (select cond, x, 1)" -> 0.
llvm-svn: 130694
|
| |
|
|
| |
llvm-svn: 130693
|
| |
|
|
| |
llvm-svn: 130692
|
| |
|
|
| |
llvm-svn: 130691
|
| |
|
|
|
|
| |
take some time.
llvm-svn: 130690
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
When an interfering live range ends at a dead slot index between two
instructions, make sure that the inserted copy instruction gets a slot index
after the dead ones. This makes it possible to avoid the interference.
Ideally, there shouldn't be interference ending at a deleted instruction, but
physical register coalescing can sometimes do that to sub-registers.
This fixes PR9823.
llvm-svn: 130687
|
| |
|
|
|
|
|
|
|
|
| |
comments claimed it did this, but the LHS value was actually an unused variable.
The new system considers only the '-foo' part when comparing it for typos
against flags that have values, but still look at the whole string for flags
that don't. That way, we'll still correct '-inst=combine' to '-instcombine'.
llvm-svn: 130685
|
| |
|
|
|
|
| |
to scope a variable more tightly per llvm coding style. No functional change.
llvm-svn: 130684
|
| |
|
|
|
|
| |
problem reported on cfe-dev.
llvm-svn: 130661
|
| |
|
|
| |
llvm-svn: 130658
|
| |
|
|
|
|
|
|
|
| |
instead of the versions of individual libraries.
autoconf: Add checking ELM_Callback decl for mingw32 and mingw-w64.
cmake/config-ix.cmake: Add checking ELM_Callback decl for win32.
llvm-svn: 130657
|
| |
|
|
|
|
| |
likely a result of copy/paste.
llvm-svn: 130640
|
| |
|
|
|
|
| |
-fno-dwarf2-cfi-asm. Implement the same behavior.
llvm-svn: 130637
|
| |
|
|
| |
llvm-svn: 130635
|
| |
|
|
|
|
|
|
|
|
| |
for all symbol differences and can drop the old EmitPCRelSymbolValue
method.
This also make getExprForFDESymbol on ELF equal to the one on MachO, and it
can be made non-virtual.
llvm-svn: 130634
|
| |
|
|
|
|
| |
less agressive about disabling cfi on linux :-(
llvm-svn: 130626
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
after folding ADD32ri to ADD32mi, so don't do that.
This only happens when the greedy register allocator gets itself in trouble and
spills %vreg9 here:
16L %vreg9<def> = MOVPC32r 0, %ESP<imp-use>; GR32:%vreg9
48L %vreg9<def> = ADD32ri %vreg9, <es:_GLOBAL_OFFSET_TABLE_>[TF=1], %EFLAGS<imp-def,dead>; GR32:%vreg9
That should never happen, the live range should be split instead.
llvm-svn: 130625
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Currently the output should be almost identical to the one produced by CodeGen
to make the transition easier.
The only two differences I know of are:
* Some files get an extra advance loc of size 0. This will be fixed when
relaxations are enabled.
* The optimization of declaring an EH symbol as an external variable is not
implemented. This is a subset of adding the nounwind attribute, so we if really
this at -O0 we should probably do it at the IL level.
llvm-svn: 130623
|
| |
|
|
|
|
|
|
| |
range covers the entire block.
The live range can't be terminated at a random instruction.
llvm-svn: 130619
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
urem or constant B.
This obviously helps a lot if the division would be turned into a libcall
(think i64 udiv on i386), but div is also one of the few remaining instructions
on modern CPUs that become more expensive when the bitwidth gets bigger.
This also helps register pressure on i386 when dividing chars, divb needs
two 8-bit parts of a 16 bit register as input where divl uses two registers.
int foo(unsigned char a) { return a/10; }
int bar(unsigned char a, unsigned char b) { return a/b; }
compiles into (x86_64)
_foo:
imull $205, %edi, %eax
shrl $11, %eax
ret
_bar:
movzbl %dil, %eax
divb %sil, %al
movzbl %al, %eax
ret
llvm-svn: 130615
|
| |
|
|
|
|
| |
This folds away silly stuff like (a&255)/1000 -> 0.
llvm-svn: 130614
|
| |
|
|
|
|
|
| |
is a bit ugly, but doing it on the base MCStreamer would be redundant
with the object streamer which does it using SD.
llvm-svn: 130611
|