7a1a8eb6e2
Implement "AsCheapAsAMove" for some obviously cheap instructions: xor and the
...
like.
llvm-svn: 51662
2008-05-29 01:02:09 +00:00
f712664ec3
convert another operand loop to iterator formulation
...
llvm-svn: 51661
2008-05-29 00:51:08 +00:00
3f6bb2713e
Add a flag to indicate that an instruction is as cheap (or cheaper) than a move
...
instruction to execute. This can be used for transformations (like two-address
conversion) to remat an instruction instead of generating a "move"
instruction. The idea is to decrease the live ranges and register pressure and
all that jazz.
llvm-svn: 51660
2008-05-28 22:54:52 +00:00
5a83b097ed
Check the "isSafeToMove" predicate, which has a series of tests to make sure
...
that it's safe to remat an instruction.
llvm-svn: 51659
2008-05-28 22:52:47 +00:00
779b4180dc
Remap VNInfo data as well when doing renumbering.
...
llvm-svn: 51658
2008-05-28 22:40:08 +00:00
4f8e1ad32a
Factor the numbering computation into a separate method, and add the slightest attempt at some renumbering logic, which is currently unused.
...
llvm-svn: 51652
2008-05-28 20:54:50 +00:00
68079268f5
Fix PR2289: vr defined by multiple implicit_def as result of coalescing.
...
llvm-svn: 51648
2008-05-28 17:40:10 +00:00
427412e7c8
Teach local register allocator to deal with landing pad MBB's.
...
llvm-svn: 51647
2008-05-28 17:22:32 +00:00
e993b80ef5
Update some comments noticed in a recent checkin
...
llvm-svn: 51644
2008-05-28 16:31:36 +00:00
e438c56174
Update text to point people at the right version of the tutorial for
...
their release.
llvm-svn: 51638
2008-05-28 06:16:08 +00:00
ecdefb5df7
Implement PR2370: memmove(x,x,size) -> noop.
...
llvm-svn: 51636
2008-05-28 05:30:41 +00:00
f8910ab6db
Add chain inputs for loads.
...
llvm-svn: 51635
2008-05-28 04:25:57 +00:00
633cd5949b
Fix CodeGen/Generic/2005-10-21-longlonggtu.ll on ia64.
...
llvm-svn: 51634
2008-05-28 04:14:30 +00:00
724895625b
loads should get chains. THis helps but does not solve CodeGen/Generic/2003-05-27-phifcmpd.ll
...
on ia64.
llvm-svn: 51633
2008-05-28 04:06:52 +00:00
d2c3e86cc3
Fix 2006-04-28-Sign-extend-bool.ll for ia64.
...
llvm-svn: 51632
2008-05-28 04:00:06 +00:00
c2fb8d7e2b
reindent.
...
llvm-svn: 51631
2008-05-28 03:59:32 +00:00
68bddb8966
Fix the encoding for two more "rm" instructions that were using MRMSrcReg.
...
llvm-svn: 51630
2008-05-28 01:50:19 +00:00
5e3faf2343
Fixed X86 encoding error CVTPS2PD and CVTPD2PS when the source operand
...
is a memory location
llvm-svn: 51626
2008-05-28 00:42:27 +00:00
f1e18c7c44
Don't attempt to create VZEXT_LOAD out of an extload. This an issue where the
...
code generator would do something like this:
f64 = load f32 <anyext>, f32mem
v2f64 = insertelt undef, %0, 0
v2f64 = insertelt %1, 0.0, 1
into
v2f64 = vzext_load f32mem
which on x86 is movsd, when you really wanted a cvtss2sd/movsd pair.
llvm-svn: 51624
2008-05-28 00:24:25 +00:00
1c4627c282
use space insted of tabs
...
llvm-svn: 51623
2008-05-28 00:06:14 +00:00
ce956e0613
fix infinite recursion if a global's initializer references the global
...
llvm-svn: 51617
2008-05-27 22:07:08 +00:00
29e0ad2ff1
Make Use::init() private, it breaks invariants, clients should assign or use set().
...
llvm-svn: 51615
2008-05-27 21:29:38 +00:00
a61379eeea
Use Function::getEntryBlock instead of Function::begin, for clarity.
...
llvm-svn: 51613
2008-05-27 20:55:29 +00:00
3ec5520286
Print debug output when any edge becomes executable, including
...
the first visited edge.
llvm-svn: 51612
2008-05-27 20:47:30 +00:00
8b16aeb879
Add instcombine after global optimizations.
...
llvm-svn: 51611
2008-05-27 20:42:44 +00:00
4ec1b4e3f9
Fix a word-o and add an explicit keyword.
...
llvm-svn: 51610
2008-05-27 20:41:18 +00:00
2e44ec7c4d
Incorporated feedback: Check that the implicitly defined operands aren't used
...
before deleting the instruction.
llvm-svn: 51609
2008-05-27 20:40:52 +00:00
e3a6a18364
Tidy up whitespace in ConstantRange::print output.
...
llvm-svn: 51606
2008-05-27 20:29:07 +00:00
d38b712ee0
Use IPSCCPPass instead of IPConstantPropagationPass.
...
llvm-svn: 51605
2008-05-27 20:18:45 +00:00
221e9d0d22
Specify a target so that this tests tests what it's intended to test.
...
llvm-svn: 51600
2008-05-27 17:55:57 +00:00
923a375053
Make this test independent of the target-triple; the stack alignment
...
is specifically what this test depends on.
llvm-svn: 51599
2008-05-27 17:44:23 +00:00
a3de9e4576
back out last commit: The .cpp file for a module should include its corresponding header first, even if redundant.
...
llvm-svn: 51598
2008-05-27 17:26:02 +00:00
ca5bdef66a
Use an explicit llvm:: prefix in the STATISTIC macro, so STATISTIC can still be
...
used when "using namespace llvm" is not in effect.
llvm-svn: 51592
2008-05-27 12:41:24 +00:00
698348dfac
Fix some constructs that gcc-4.4 warns about.
...
llvm-svn: 51591
2008-05-27 11:50:51 +00:00
6974e1d380
prune unneeded #includes
...
llvm-svn: 51590
2008-05-27 11:06:03 +00:00
e9408e6e96
remove unneeded reinterpret_casts
...
llvm-svn: 51589
2008-05-27 11:03:29 +00:00
0a0f2c1dcb
We have the correct headers included to know that BB isa Value. No reinterpret_cast necessary.
...
llvm-svn: 51588
2008-05-27 10:48:39 +00:00
698e72ec2c
Remove ADCE from the optimization pipeline.
...
llvm-svn: 51581
2008-05-27 04:27:27 +00:00
3ebe82b57a
InequalityGraph::node() can create new nodes, invalidating iterators across
...
the set of nodes. Fix makeEqual to handle this by creating the new node first
then iterating across them second.
llvm-svn: 51573
2008-05-27 00:59:05 +00:00
6be65d2a84
Grammaro.
...
llvm-svn: 51572
2008-05-26 22:49:36 +00:00
2d3024d838
eliminate calls to deprecated Use::init() interface
...
llvm-svn: 51570
2008-05-26 21:33:52 +00:00
a61cc6ece0
Whoops -- forgot PR reference on this test.
...
llvm-svn: 51569
2008-05-26 20:23:33 +00:00
213e114a2c
The Linux ABI emits an extra "movl %esp, %ebp" in function prologue and
...
sometimes a "mov %ebp, %esp" in the epilogue.
Force these tests that rely on counting 'mov' to use i686-apple-darwin8.8.0
where they were written.
llvm-svn: 51568
2008-05-26 20:18:56 +00:00
dd7daee850
Factor code to copy global value attributes like
...
the section or the visibility from one global
value to another: copyAttributesFrom. This is
particularly useful for duplicating functions:
previously this was done by explicitly copying
each attribute in turn at each place where a
new function was created out of an old one, with
the result that obscure attributes were regularly
forgotten (like the collector or the section).
Hopefully now everything is uniform and nothing
is forgotten.
llvm-svn: 51567
2008-05-26 19:58:59 +00:00
d3f21d165f
Use a DenseMap instead of an std::map, speeding up the testcase in PR2368 by about a third.
...
llvm-svn: 51565
2008-05-26 10:07:43 +00:00
2e8c82893b
The enabling of remat in 2-address conversion breaks this test:
...
Running /Users/void/llvm/llvm.src/test/CodeGen/X86/dg.exp ...
FAIL: /Users/void/llvm/llvm.src/test/CodeGen/X86/2007-11-30-LoadFolding-Bug.ll
Failed with exit(1) at line 1
while running: llvm-as < /Users/void/llvm/llvm.src/test/CodeGen/X86/2007-11-30-LoadFolding-Bug.ll | llc -march=x86 -mattr=+sse2 -stats |& grep {1 .*folded into instructions}
child process exited abnormally
Make this conditional for now.
llvm-svn: 51563
2008-05-26 05:49:49 +00:00
c737e4639a
A problem that's exposed when machine LICM is enabled. Consider this code:
...
LBB1_3: # bb
...
xorl %ebp, %ebp
subl (%ebx), %ebp
...
incl %ecx
cmpl %edi, %ecx
jl LBB1_3 # bb
Whe using machine LICM, LLVM converts it into:
xorl %esi, %esi
LBB1_3: # bb
...
movl %esi, %ebp
subl (%ebx), %ebp
...
incl %ecx
cmpl %edi, %ecx
jl LBB1_3 # bb
Two address conversion inserts the copy instruction. However, it's cheaper to
rematerialize it, and remat helps reduce register pressure.
llvm-svn: 51562
2008-05-26 05:18:34 +00:00
be993358a7
Use {} instead of "" in RUN lines.
...
llvm-svn: 51561
2008-05-26 01:27:08 +00:00
3195b393d6
Don't treat values as signed when looking at loop steppings in HowForToNonZero.
...
llvm-svn: 51560
2008-05-25 23:43:32 +00:00
f6ccd2580c
"ret (constexpr)" can't be folded into a Constant. Add a method to
...
Analysis/ConstantFolding to fold ConstantExpr's, then make instcombine use it
to try to use targetdata to fold constant expressions on void instructions.
Also extend the icmp(inttoptr, inttoptr) folding to handle the case where
int size != ptr size.
llvm-svn: 51559
2008-05-25 20:56:15 +00:00
Bill Wendling