f522a4e034
Don't emit a redundant BitCastInst if the value to be defined in the
...
preheader is already an instruction.
llvm-svn: 74031
2009-06-24 00:28:59 +00:00
ecdab5406d
Guard the listeners list. Unfortunately, this requires a real static rather
...
than a managed static because other managed statics can (and do) access this
list in their destructors. Yes, I know it's horrible.
llvm-svn: 74029
2009-06-24 00:25:42 +00:00
5a495fea07
Provide InitializeAllTargets and InitializeNativeTarget functions in the
...
C bindings. Change all the backend "Initialize" functions to have C linkage.
Change the "llvm/Config/Targets.def" header to use C-style comments to avoid
compile warnings.
llvm-svn: 74026
2009-06-23 23:59:40 +00:00
badeace84a
Update for MCImm -> MCValue rename.
...
llvm-svn: 74024
2009-06-23 23:39:15 +00:00
7e6bade348
rename MCImm ->MCValue.
...
llvm-svn: 74023
2009-06-23 23:31:52 +00:00
9c14bce0f9
Test commit: fixed spacing.
...
llvm-svn: 74022
2009-06-23 23:25:37 +00:00
fb1ad4f6ec
Add comments for the MCStreamer interface.
...
llvm-svn: 74019
2009-06-23 23:02:31 +00:00
2ed89d7faa
It is not a good idea to have data member's name match argument's name. In fact, it is a simple receipe to waste an hour or so.
...
llvm-svn: 74018
2009-06-23 22:25:41 +00:00
4d3a423dcd
Use getReturnTypeName() to print return type.
...
llvm-svn: 74017
2009-06-23 22:23:13 +00:00
de11f1efb0
upgrade
...
llvm-svn: 74016
2009-06-23 22:11:37 +00:00
a858204c67
Add DISubprogram::getReturnTypeName()
...
llvm-svn: 74015
2009-06-23 22:07:48 +00:00
ca29e4d7f4
Start flushing out MCContext.
...
- Lives inside new library lib/MC (LLVMMC.a)
llvm-svn: 74013
2009-06-23 22:01:43 +00:00
ca8f986f63
Use atomic operations when accessing statistics, and make the lazy initialization of statistics actually threadsafe.
...
llvm-svn: 74005
2009-06-23 21:19:38 +00:00
1fdf01026b
Add atomic multiply and divide operations, built on top of CompareAndSwap.
...
llvm-svn: 74004
2009-06-23 21:19:04 +00:00
f42a6ab6b8
Reinstate target dependencies on Table_gen targets in CMake
...
llvm-svn: 74003
2009-06-23 21:05:21 +00:00
e9b1beb543
Make timers threadsafe again. This isn't quite as nice as I'd hoped (it uses locking rather than atomic arithmetic),
...
but should work on all the platforms we care about.
I might revisit this if a totally awesome way to do it occurs to me.
llvm-svn: 74002
2009-06-23 20:52:29 +00:00
8b4e03898c
Temporary copy-pasto to make examples compile.
...
llvm-svn: 74001
2009-06-23 20:47:24 +00:00
248c414024
Make llvmc work again.
...
Chris recently broke llvmc with his Makefile changes (r75379). That patch made
the global change .o -> .a, which caused built-in llvmc plugins to stop working
since plugin initialization in llvmc is based on static variables not referenced
from the main executable. This patch implements auto-generated forced references
to the plugin libraries.
llvm-svn: 74000
2009-06-23 20:46:48 +00:00
92df147ae5
Typo.
...
llvm-svn: 73999
2009-06-23 20:46:00 +00:00
ae3b1a8b55
A little bit nicer formatting.
...
llvm-svn: 73998
2009-06-23 20:45:31 +00:00
beecd51f95
Typo.
...
llvm-svn: 73997
2009-06-23 20:45:07 +00:00
de4e98afda
Start sketching MCStreamer interface.
...
llvm-svn: 73994
2009-06-23 20:24:17 +00:00
5cc4131063
Revert my last series of commits related to Timer and 64-bit atomics. Not all the targets
...
we care about are capable of supporting it.
llvm-svn: 73993
2009-06-23 20:17:22 +00:00
de57cfd191
Add include file to get the type for in64_t.
...
llvm-svn: 73991
2009-06-23 20:02:00 +00:00
7d80d29187
Test instructions operands were printed in the wrong order.
...
llvm-svn: 73990
2009-06-23 19:56:37 +00:00
f6e8f6b0f4
"The MMX_MASKMOVQ and MMX_MASKMOVQ64 instructions are labeled as MRMDestMem
...
instructions, which implies that there is an explicit memory operand. There is
(however) no explicit memory operand; although this is a store, the only memory
operand is implicit, indicated by DS:EDI. This causes the table-generation code
for the disassembler to report an error."
Patch by Sean Callanan!
llvm-svn: 73989
2009-06-23 19:52:59 +00:00
c547bad3ff
Switched size_t to int64_t to prevent type mismatch in call to max.
...
llvm-svn: 73988
2009-06-23 19:49:23 +00:00
4983e4550e
Proper patterns for thumb2 shift and rotate instructions.
...
llvm-svn: 73987
2009-06-23 19:39:13 +00:00
8d21e9c4e6
Code clean up.
...
llvm-svn: 73986
2009-06-23 19:38:34 +00:00
bec1dba896
Add IsThumb1Only to most 16-bit thumb instructions since we want to isel 32-bit instructions when they are available.
...
llvm-svn: 73985
2009-06-23 19:38:13 +00:00
11b61adfc9
add a simple MCImm class.
...
llvm-svn: 73984
2009-06-23 18:58:00 +00:00
1423721425
The correct values here (as defined by gcc-4.2) are
...
different for ppc; add another version of the test.
llvm-svn: 73983
2009-06-23 18:42:26 +00:00
3417d71ff4
refactor a bunch of X86 specific stuff out to its own file.
...
llvm-svn: 73982
2009-06-23 18:41:30 +00:00
c99c9d3f01
This test only works on ppc32 if some optimization is done before
...
generating LLVM IR; it is correct in the code as written
to use 8-byte-aligned operations to copy Key in bar. Formerly
the gcc inliner was run, now it isn't. I don't think it's
possible to preserve this as a pure FE test. Adding -O2 lets
the llvm optimizers get rid of the 8-byte-aligned stores, at least.
llvm-svn: 73981
2009-06-23 18:34:37 +00:00
9262d43c5d
Atomic ops that do arithmetic use signed arithmetic.
...
llvm-svn: 73980
2009-06-23 18:30:27 +00:00
e324ceeced
CMake: remove support for llvm-config-generated dependencies in the build
...
llvm-svn: 73979
2009-06-23 18:30:17 +00:00
ddf74dc986
Actually, these need to be signed integers, not unsigned.
...
llvm-svn: 73978
2009-06-23 18:21:13 +00:00
e379107cdc
Rename SelectShifterOperand to SelectThumb2ShifterOperandReg.
...
llvm-svn: 73975
2009-06-23 18:14:38 +00:00
bf45b3e2ad
Use 64-bit integer counters for tracking time, rather than doubles. This will be more atomic op friendly.
...
llvm-svn: 73974
2009-06-23 18:12:30 +00:00
736ffeeab4
See test. Judging from PR 1278, at the time the test was committed, the
...
generated code was apparently doing stores directly into the return value
aggregate; now, it's doing a copy from a compiler-generated static object.
That object is initialized using [4 x i8] which breaks the test. I believe
this change preserves the original point of the test.
Of course it would be better for the code to do stores directly into the
return aggregate, but that is not what happens at -O0; the llvm optimizers
seem to do that on x86 but not on ppc32, possibly because of the explicit
padding (which is unavoidable). I think it must have been being done by
a gcc optimizer pass before.
llvm-svn: 73972
2009-06-23 18:03:49 +00:00
9e3df5b67f
Label the existing atomic functions as 32-bit specific, and add a 64-bit one that will be useful in
...
the near future.
llvm-svn: 73971
2009-06-23 18:01:04 +00:00
881af58933
Eliminate some unused code in CMa'e add_llvm_target. No functionality change.
...
llvm-svn: 73970
2009-06-23 17:58:51 +00:00
85fedbe081
Eliminate object-relinking support from CMake. Fixes PR 4429 and
...
cleans up the CMake-based build system a bit. Started by a patch from
Xerxes Rånby.
llvm-svn: 73969
2009-06-23 17:57:35 +00:00
b45cebabc9
Obvious typo.
...
llvm-svn: 73967
2009-06-23 17:54:26 +00:00
83bfbfa195
add an accessor.
...
llvm-svn: 73966
2009-06-23 17:50:34 +00:00
431cf567de
Initial Thumb2 support. Majority of the work is done by David Goodwin. There are
...
also some contribution from Jim Grosbach, Bob Wilson, and Evan Cheng.
I've done my best to consolidate the patches with those that were done by
Viktor Kutuzov and Anton Korzh from Access Softek, Inc. Let me know if missed
anything. I've completely reorganized the thumb2 td file, made more extensive
uses of multiclass, etc.
Test cases will be contributed later after I re-organize what's in svn first.
llvm-svn: 73965
2009-06-23 17:48:47 +00:00
cdf2b2dfea
Add an atomic add operation.
...
llvm-svn: 73964
2009-06-23 17:39:31 +00:00
4ed41c8be8
Make the lazy initialization of DefaultTimerGroup threadsafe.
...
llvm-svn: 73963
2009-06-23 17:33:37 +00:00
a1ae30140b
Regenerate configure script
...
llvm-svn: 73962
2009-06-23 17:22:05 +00:00
d967749791
Work around build problem with OpenJDK, which defines X86 as a
...
macro. Fixes PR 4427. Patch by Xerxes Rånby!
llvm-svn: 73961
2009-06-23 17:21:35 +00:00
bc4c6a0245
Add myself to the list.
...
llvm-svn: 73959
2009-06-23 17:03:23 +00:00
891fe79672
Revert r73923, which broke clang.
...
llvm-svn: 73957
2009-06-23 16:36:10 +00:00
84b581a4bd
Fold the add (ptr, offset) into ptr[offset] only if the offset is small enough. movwi and moviw allow value of 5-bits only (i.e. 32).
...
llvm-svn: 73953
2009-06-23 07:10:19 +00:00
f97d8bbe73
implement a trivial binary expression parser, we can now parse all of 176.gcc.llc.s
...
llvm-svn: 73950
2009-06-23 05:57:07 +00:00
6a42ec3e70
Minor reorg.
...
llvm-svn: 73948
2009-06-23 05:25:29 +00:00
022a726a25
Replace isTwoAddress with operand constraint.
...
llvm-svn: 73947
2009-06-23 05:23:49 +00:00
02a9191b71
Use a default alignment for data and bss sections.
...
Only pad when the section size > 0 and move the code that deals
with globals initializers to a place we know for sure the global
is initialized.
llvm-svn: 73944
2009-06-23 04:39:27 +00:00
9ec76a0b79
get a definition of strull on windows, thanks to Howard Su.
...
llvm-svn: 73929
2009-06-23 00:24:36 +00:00
e3c1aca0d9
Guard the layout info object.
...
llvm-svn: 73928
2009-06-23 00:21:15 +00:00
bee30f58fb
Guard the plugin loader.
...
llvm-svn: 73925
2009-06-23 00:02:39 +00:00
977c6b8998
Add guards around timer groups, which can be shared.
...
llvm-svn: 73923
2009-06-22 23:37:06 +00:00
53efeb0e45
Fix a bug in the trip-count computation with And/Or. If either of the
...
sides is CouldNotCompute, the resulting exact count must be CouldNotCompute.
llvm-svn: 73920
2009-06-22 23:28:56 +00:00
2e076c4e02
Add support for ARM's Advanced SIMD (NEON) instruction set.
...
This is still a work in progress but most of the NEON instruction set
is supported.
llvm-svn: 73919
2009-06-22 23:27:02 +00:00
975ce637db
Guard the statistics table.
...
llvm-svn: 73916
2009-06-22 23:08:27 +00:00
2d2ed3c54b
Guard the global annotation tables.
...
llvm-svn: 73913
2009-06-22 22:44:15 +00:00
da19a13a68
Add locking around the external function lookup table for the interpreter.
...
llvm-svn: 73912
2009-06-22 22:30:56 +00:00
e9454e8c25
Remove unneeded #include.
...
llvm-svn: 73911
2009-06-22 22:09:07 +00:00
fd76113e28
Fix a few minor issues that were exposed by the removal of SCEVHandle.
...
llvm-svn: 73910
2009-06-22 22:08:45 +00:00
e67b77028e
Add explicit types for shift count constants. This is in preparation for
...
another change that makes the types ambiguous (at least as far as tablegen
is concerned).
llvm-svn: 73909
2009-06-22 22:08:29 +00:00
2636693a3c
Fix llvm::ComputeNumSignBits to handle pointer types
...
conservatively correctly, instead of aborting.
llvm-svn: 73908
2009-06-22 22:02:32 +00:00
04052ec597
Remove the parent pointer from SCEV, since it did not end up being needed.
...
llvm-svn: 73907
2009-06-22 21:57:23 +00:00
65b6056e37
SCEVHandle is no more!
...
llvm-svn: 73906
2009-06-22 21:39:50 +00:00
482495695e
Use thumb2 for ARM architectures V6T2 and later. Fix a bug in checking
...
for "thumb" and add a check for V6T2.
llvm-svn: 73905
2009-06-22 21:28:22 +00:00
8885b379fa
Fix some typos that Duncan noticed.
...
llvm-svn: 73903
2009-06-22 21:10:22 +00:00
16ee19738c
It's coalescer, not coaleser.
...
llvm-svn: 73902
2009-06-22 21:09:17 +00:00
4582530a2c
For Darwin on ARMv6 and newer, make register r9 available for use as a
...
caller-saved register.
llvm-svn: 73901
2009-06-22 21:01:46 +00:00
315fb72d36
Fix memcpy expansion so it won't generate invalid
...
types for the target (I think). This was breaking
the PPC32 calling sequence.
llvm-svn: 73900
2009-06-22 20:59:07 +00:00
8cbbc7944d
Fix another register coalescer crash: forgot to check if the instruction being updated has already been coalesced.
...
llvm-svn: 73898
2009-06-22 20:49:32 +00:00
bd9c94d7af
Use different functions to emit the string and symbol tables.
...
llvm-svn: 73895
2009-06-22 19:29:56 +00:00
a040566fec
Add more methods to gather target specific elf stuff
...
Support for .text relocations, implementing TargetELFWriter overloaded methods for x86/x86_64.
Use a map to track global values to their symbol table indexes
Code cleanup and small fixes
llvm-svn: 73894
2009-06-22 19:16:16 +00:00
3d75d6af57
hasFP should return true if frame address is taken.
...
llvm-svn: 73893
2009-06-22 18:38:48 +00:00
86837616f7
Banish global state from ScalarEvolution! SCEV uniquing is now done by tables attached to the ScalarEvolution pass.
...
This also throws out the SCEV reference counting scheme, as the the SCEVs now have a lifetime controlled by the
ScalarEvolution pass.
Note that SCEVHandle is now a no-op, and will be remove in a future commit.
llvm-svn: 73892
2009-06-22 18:25:46 +00:00
b8b636d145
Recognize and handle ARM v7 target triples for Darwin.
...
llvm-svn: 73889
2009-06-22 18:01:28 +00:00
360eef0782
Fix llvm-gcc build for armv6t2 and later architectures. The hasV6T2Ops
...
predicate does not check if Thumb mode is enabled, and when in ARM mode
there are still some checks for constant-pool use that need to run.
llvm-svn: 73887
2009-06-22 17:29:13 +00:00
ed627385de
Make use of getUMinFromMismatchedTypes when computing backedge-taken
...
counts for loops with multiple exits, replacing more conservative code
which only handled constants. This is derived from a patch by
Nick Lewycky.
This also fixes llc aborts in ClamAV and others, as
getUMinFromMismatchedTypes takes care of balancing the types before
working with them.
llvm-svn: 73884
2009-06-22 15:09:28 +00:00
2bc22305d2
Add a getUMinFromMismatchedTypes helper function.
...
llvm-svn: 73883
2009-06-22 15:03:27 +00:00
376c6f16de
Include cstdio to get EOF, needed with gcc-4.4.
...
llvm-svn: 73879
2009-06-22 06:59:32 +00:00
1d1fe918de
process memory operands with a parenthesized expression for a displacement,
...
like "(4+5)(%eax)".
llvm-svn: 73878
2009-06-22 06:35:58 +00:00
7fdbce7a60
Implement full support for parsing primary expressions. We can now parse
...
all of health and voronoi (ignoring directives). We only get 409 lines into
176.gcc though because we don't have binary operators yet:
Parsing 176.gcc.llc.s:409: unexpected token in operand list
movsbl _arityvec+1(,%edi,8), %eax
^
llvm-svn: 73877
2009-06-22 06:32:03 +00:00
a7b51ed5e6
implement parser support for '*' operands, as in "call *%eax".
...
llvm-svn: 73876
2009-06-22 06:02:13 +00:00
78db36271c
implement memory operand parsing.
...
llvm-svn: 73875
2009-06-22 05:51:26 +00:00
d3f7935f50
Testcase for (llvm-gcc) 73873. Usually
...
std::pair<double, float*>
is 16 bytes on darwin-powerpc, but not always.
See testcase for full weirdness.
llvm-svn: 73874
2009-06-22 04:47:32 +00:00
692b46872e
Factor out code for computing umin and smin for SCEV expressions into
...
helper functions. Based on a patch by Nick Lewycky.
llvm-svn: 73869
2009-06-22 03:18:45 +00:00
e5074c46b2
start implementing some simple operand parsing.
...
llvm-svn: 73867
2009-06-22 01:29:09 +00:00
96212b661c
Teach ScalarEvolution how to analyze loops with multiple exit
...
blocks, and also exit blocks with multiple conditions (combined
with (bitwise) ands and ors). It's often infeasible to compute an
exact trip count in such cases, but a useful upper bound can often
be found.
llvm-svn: 73866
2009-06-22 00:31:57 +00:00
13037149b9
Delete an unused variable.
...
llvm-svn: 73865
2009-06-22 00:19:17 +00:00
78ea89e161
Fix this code to correctly handle loops with multiple exits. Until
...
now, this hasn't mattered, because ScalarEvolution hasn't been able
to compute trip counts for loops with multiple exits. But it will
soon.
llvm-svn: 73864
2009-06-22 00:15:15 +00:00
860379bcc2
Rename a variable for consistency with the ExitBlock vs ExitingBlock
...
terminology that LoopInfo uses.
llvm-svn: 73863
2009-06-21 23:48:38 +00:00
0104842ee3
Fix ScalarEvolution's backedge-taken count computations to check for
...
overflow when computing a integer division to round up.
Thanks to Nick Lewycky for noticing this!
llvm-svn: 73862
2009-06-21 23:46:38 +00:00
2adc9e739c
rename SourceMgr::PrintError to PrintMessage.
...
llvm-svn: 73861
2009-06-21 21:22:11 +00:00
36e0212c16
set up the top-level parsing loop.
...
llvm-svn: 73860
2009-06-21 20:54:55 +00:00
b013345338
stub out parser for asm files. Change invariant on lexer to always
...
print its error message when it returns an asmtok::Error token.
Compute a proper error code for llvm-mc in 'lex' mode. Add new
-as-lex option to enable lexing mode (vs parsing mode).
llvm-svn: 73859
2009-06-21 20:16:42 +00:00
419a97491a
add string literals.
...
llvm-svn: 73858
2009-06-21 19:56:35 +00:00
99f0b60f45
hopefully fix the build on linux.
...
llvm-svn: 73857
2009-06-21 19:43:50 +00:00
7d4c0d5fb3
simplify some error recovery stuff.
...
llvm-svn: 73856
2009-06-21 19:22:49 +00:00
d0765617d2
implement enough of a lexer to get through Olden/health/Output/health.llc.s
...
without errors.
llvm-svn: 73855
2009-06-21 19:21:25 +00:00
6ead59f8ed
Fix PR4185.
...
Handle FpSET_ST0_80 being used when ST0 is still alive.
llvm-svn: 73850
2009-06-21 12:02:51 +00:00
d485848229
oh yeah, cmake needs to be told explicitly about new files :)
...
llvm-svn: 73849
2009-06-21 07:19:34 +00:00
c8dfbcbb31
some baby steps.
...
llvm-svn: 73848
2009-06-21 07:19:10 +00:00
7153bd8a29
prune #include, fix warning.
...
llvm-svn: 73847
2009-06-21 05:33:06 +00:00
d70e15b46f
start wiring up support for asm parsing.
...
llvm-svn: 73846
2009-06-21 05:22:37 +00:00
976af622a9
move include searching logic from TGLexer to SourceMgr.
...
llvm-svn: 73845
2009-06-21 05:06:04 +00:00
fd255754af
Rename TGSourceMgr -> SourceMgr.
...
llvm-svn: 73844
2009-06-21 03:41:50 +00:00
526c8cb557
rename TGLoc -> SMLoc.
...
llvm-svn: 73843
2009-06-21 03:39:35 +00:00
1b30e1ac45
move TGSourceMgr class out of TableGen into libsupport.
...
llvm-svn: 73842
2009-06-21 03:36:54 +00:00
7719d0184a
simplify closing paren and (rip) printing.
...
llvm-svn: 73841
2009-06-21 02:22:53 +00:00
a3da048c82
indentation fix
...
llvm-svn: 73840
2009-06-21 02:22:34 +00:00
6b8320fa5b
Expand this test to handle more cases (remainder and shifts) of zero.
...
llvm-svn: 73839
2009-06-21 01:56:41 +00:00
98b13ac6f3
fit in 80 cols
...
llvm-svn: 73838
2009-06-21 01:52:00 +00:00
e80442452e
remove #if 0 code
...
llvm-svn: 73837
2009-06-21 01:48:49 +00:00
cfad3f3807
cosmetic changes.
...
llvm-svn: 73836
2009-06-21 01:27:55 +00:00
62fdd112fc
implement PR4424: 0/x is always 0 for integer division.
...
llvm-svn: 73835
2009-06-21 01:15:55 +00:00
7d2b049404
change TLS_ADDR lowering to lower to a real mem operand, instead of matching as
...
a global with that gets printed with the :mem modifier. All operands to lea's
should be handled with the lea32mem operand kind, and this allows the TLS stuff
to do this. There are several better ways to do this, but I went for the minimal
change since I can't really test this (beyond make check).
This also makes the use of EBX explicit in the operand list in the 32-bit,
instead of implicit in the instruction.
llvm-svn: 73834
2009-06-20 20:38:48 +00:00
1771a852f0
no need for unwind info
...
llvm-svn: 73832
2009-06-20 19:48:26 +00:00
fbc9778a1b
no need for unwind info here.
...
llvm-svn: 73831
2009-06-20 19:43:09 +00:00
40f560d433
remove dead code, the main codepath never has to deal
...
with MBB's.
llvm-svn: 73830
2009-06-20 19:34:47 +00:00
9c2119616f
eliminate the "call" operand modifier from the asm descriptions, modeling
...
it as a pcrel immediate instead. This gets pc-rel weirdness out of the
main printoperand codepath.
llvm-svn: 73829
2009-06-20 19:34:09 +00:00
2ae473f739
Clarify that if the alignment is zero then an alignment
...
compatible with the type will be used (at least the ABI
alignment).
llvm-svn: 73827
2009-06-20 13:26:06 +00:00
b4b5c10477
fix edit-o
...
llvm-svn: 73824
2009-06-20 08:13:12 +00:00
9fc4db5615
some comments and cleanup
...
llvm-svn: 73818
2009-06-20 07:59:10 +00:00
a0022a772b
implement support for lowering subregs when preparing to print
...
LEA64_32r, eliminating a bunch of modifier logic stuff on addr modes.
Implement support for printing mbb labels as operands.
llvm-svn: 73817
2009-06-20 07:03:18 +00:00
c6a8d0dbe9
Fix PR4419: handle defs of partial uses.
...
llvm-svn: 73816
2009-06-20 04:34:51 +00:00
79ef65a271
Remove unused field.
...
llvm-svn: 73815
2009-06-20 01:07:54 +00:00
1b61c2a1ce
Initialize MMI
...
llvm-svn: 73813
2009-06-20 01:00:07 +00:00
1f8d968687
hook up printMemReference.
...
llvm-svn: 73811
2009-06-20 00:50:32 +00:00
4682015e61
Start implementing translation of MachineInstr to MCInst. Next
...
step is to make tblgen generate something more appropriate for MCInst,
and generate calls to operand translation routines where needed.
This includes a bunch of #if 0 code which will slowly be refactored into
something sensible.
llvm-svn: 73810
2009-06-20 00:49:26 +00:00
3672742801
make immediates be int64_t like machineoperand. Add some apis
...
llvm-svn: 73809
2009-06-20 00:47:37 +00:00
dcdc4daa5d
Tweak this test to be a little less unusual.
...
llvm-svn: 73808
2009-06-20 00:40:56 +00:00
450f4e0fa9
Generalize isLoopGuardedByCond's checking to consider two
...
SCEVUnknowns with identical Instructions to be equal. This allows
it to analze cases such as the attached testcase, where the front-end
has cloned the loop controlling expression. Along with r73805, this
lets IndVarSimplify eliminate all the sign-extend casts in the
loop in the attached testcase.
llvm-svn: 73807
2009-06-20 00:35:32 +00:00
96c51a8de1
Workaround for an... interesting bug in Darwin's pthread_rwlock_init.
...
llvm-svn: 73806
2009-06-20 00:32:27 +00:00
90d612e536
Use ScalarEvolution's new GetMinSignBits and GetMinLeadingZeros
...
in the loop backedge-taken count computation of the maximum
possible trip count.
llvm-svn: 73805
2009-06-20 00:32:22 +00:00
4ce341cc75
Add debugging code to test for various locking faux-pas's, when running in single threaded mode. This should help improve testing coverage for
...
threading support, without having extensive actually concurrent clients yet.
llvm-svn: 73803
2009-06-20 00:27:21 +00:00
2cda7d74bc
Forgot this file.
...
llvm-svn: 73802
2009-06-20 00:26:26 +00:00
0d2de8c35e
Revert r73790, and replace it with a significantly less ugly solution. Rather than trying to make the global reader-writer lock work,
...
create separate recursive mutexes for each value map. The recursive-ness fixes the double-acquiring issue, which having one per ValueMap
lets us continue to maintain some concurrency.
llvm-svn: 73801
2009-06-20 00:24:58 +00:00
8d284c7683
stub out some hacky code for wiring up the new asmprinter interfaces
...
on X86. Not useful yet.
llvm-svn: 73799
2009-06-19 23:59:57 +00:00
681f79124a
allow clients of the asmprinter to opt-out of the boilerplate with a #define.
...
llvm-svn: 73798
2009-06-19 23:57:53 +00:00
724f825f96
Fix a typo in a comment that Frits von Bommel noticed.
...
llvm-svn: 73796
2009-06-19 23:41:37 +00:00
c702fc0e64
Make GetMinTrailingZeros a member function of ScalarEvolution,
...
so that it can access the TargetData member (when available) and
use ValueTracking.h information to compute information for
SCEVUnknown Values.
Also add GetMinLeadingZeros and GetMinSignBits functions,
with minimal implementations.
llvm-svn: 73794
2009-06-19 23:29:04 +00:00
cc31110b95
Re-apply r73718, now that the fix in r73787 is in, and add a
...
hand-crafted testcase which demonstrates the bug that was exposed
in 254.gap.
llvm-svn: 73793
2009-06-19 23:23:27 +00:00
88be5cf01f
DwarfWriter is used to emit EH info also.
...
llvm-svn: 73792
2009-06-19 23:21:20 +00:00
b4b20bbb7d
Enable arm pre-allocation load / store multiple optimization pass.
...
llvm-svn: 73791
2009-06-19 23:17:27 +00:00
b07dd959a4
Fix a serious bug that would cause deadlock during abstract type refinement. The constant creation
...
gets involved, and we end up trying to recursively acquire a writer lock. The fix for this is slightly horrible,
and involves passing a boolean "locked" parameter around in Constants.cpp, but it's better than having locked and
unlocked versions of most of the code.
llvm-svn: 73790
2009-06-19 23:16:19 +00:00
55e3dd9174
Fix LSR's OptimizeSMax to ignore max operators with more than 2 operands,
...
which it isn't prepared to handle.
llvm-svn: 73787
2009-06-19 23:03:46 +00:00
da10358c84
mv CodeGen/DebugLoc.h Support/DebugLoc.h
...
llvm-svn: 73786
2009-06-19 22:08:58 +00:00
33f4eb462f
Move up dwarf writer initialization in common AsmPrinter class.
...
llvm-svn: 73784
2009-06-19 21:54:26 +00:00
86076c9e30
Revert 73718. It's breaking 254.gap.
...
llvm-svn: 73783
2009-06-19 21:15:06 +00:00
2028e25325
Unbreak cyclic deps
...
llvm-svn: 73781
2009-06-19 19:36:55 +00:00
ff5f14c15e
Fix bad paste-o in the MSVC atomics.
...
Patch by Ðван СоÑокин.
llvm-svn: 73779
2009-06-19 18:37:50 +00:00
59ba814f39
Forgot to remove some explicit locking when it became implicit in the ValueMap.
...
llvm-svn: 73778
2009-06-19 18:34:09 +00:00
383f7b1f08
Fences are expensive. Avoid doing them when we're not in multithreaded mode.
...
llvm-svn: 73777
2009-06-19 17:50:11 +00:00
b9063a84ce
Make ScalarEvolution::print print getSCEVAtScope values for
...
expressions, when interesting.
llvm-svn: 73776
2009-06-19 17:49:54 +00:00
81241a37fa
Move the memory fences out of the path for single-threaded mode.
...
llvm-svn: 73775
2009-06-19 17:45:12 +00:00
2199dbc5ff
Don't (unconditionally) use getSCEVAtScope to simplify the step
...
expression in IVUsers, because in the case of a use of a non-linear
addrec outside of a loop, this causes the addrec to be evaluated as
a linear addrec.
llvm-svn: 73774
2009-06-19 17:33:15 +00:00
ff967be708
Fix incorrect comment pointed out by Duncan.
...
llvm-svn: 73773
2009-06-19 17:20:50 +00:00
646ac6ac8a
Clean up nightly tester to use Perl's File::Find.
...
Patch by Benjamin Kramer.
llvm-svn: 73772
2009-06-19 17:19:38 +00:00
c4baa9b64d
Fix a major typo.
...
Nicolas, this is probably the cause of the failures you were seeing.
llvm-svn: 73770
2009-06-19 17:08:20 +00:00
d0a363e03b
make jump threading handle lexically identical compare instructions
...
as if they were multiple uses of the same instruction. This interacts
well with the existing loadpre that j-t does to open up many new jump
threads earlier.
llvm-svn: 73768
2009-06-19 16:27:56 +00:00
3773afe931
fix the sparc codegen to not depend on the sparc asmprinter.
...
llvm-svn: 73767
2009-06-19 15:48:10 +00:00
0de39b42ff
Add support for AuroraUX. Patch by evocallaghan.
...
llvm-svn: 73766
2009-06-19 14:40:01 +00:00
e532b557c4
Since -Wno-long-long is ignored without -pedantic,
...
place it with -pedantic. Remove -Wunused since it
is implied by -Wall. Group -Wno-unused-parameter
with -Wall -W since it is the combination of these
two that turns on -Wunused-parameter.
llvm-svn: 73765
2009-06-19 12:40:30 +00:00
d305869ca2
Add comments.
...
llvm-svn: 73761
2009-06-19 07:06:07 +00:00
2fc939c809
Fix for PR2484: add an SSE1 pattern for a shuffle we normally prefer to
...
handle with an SSE2 instruction.
llvm-svn: 73760
2009-06-19 07:00:55 +00:00
1592035e67
Should be using Bcc (average) latency to determine if-conversion threshold, not BL.
...
llvm-svn: 73759
2009-06-19 06:56:26 +00:00
495d02f4a6
Minor cleanup; fixes review comments for a previous patch. Sorry for
...
taking so long to get to this!
llvm-svn: 73757
2009-06-19 06:01:55 +00:00
77585a24ac
Teach jump threading to look at comparisons between phi nodes and non-constants.
...
llvm-svn: 73755
2009-06-19 04:56:29 +00:00
707cbc4126
Misc tweaks to Intel asm printing to make it more compatible with MASM.
...
Patch by Benedict Gaster.
llvm-svn: 73753
2009-06-19 04:48:38 +00:00
5ca4197829
Improve tail call elim to move loads above readonly calls
...
when it allows forming a tail call. Patch by Frits van
Bommel. This implements PR4323.
llvm-svn: 73752
2009-06-19 04:22:16 +00:00
87a222c5c8
part of PR4405: disable a contentious optimization for
...
strcmp -> memcmp when the lengths of the strings are unknown.
Patch by Nick Lewycky!
llvm-svn: 73751
2009-06-19 04:17:36 +00:00
6b2c960cc4
More VNInfo tweaking, plus a little progress on intra-block splitting.
...
llvm-svn: 73750
2009-06-19 02:17:53 +00:00
5d8df7ff34
Transfer dead markers when a ldrd is changed into a ldm or a pair of ldr.
...
llvm-svn: 73749
2009-06-19 01:59:04 +00:00
db371e376d
Include DataTypes.h for 'uint64_t'.
...
llvm-svn: 73748
2009-06-19 01:56:53 +00:00
4e712de541
Latency information for ARM v6. It's rough and not yet hooked up. Right now we are only using branch latency to determine if-conversion limits.
...
llvm-svn: 73747
2009-06-19 01:51:50 +00:00
d984158320
Mark a few Thumb instructions commutable; just happened to spot this
...
while experimenting. I'm reasonably sure this is correct, but please
tell me if these instructions have some strange property which makes this
change unsafe.
llvm-svn: 73746
2009-06-19 01:43:08 +00:00
60502b9ba0
Fix weird class-size-being-different problems. At some level this is being caused by config.h not being
...
included everywhere.
llvm-svn: 73745
2009-06-19 00:48:22 +00:00
a08186aabc
add a file I missed, this goes with r73743
...
llvm-svn: 73744
2009-06-19 00:47:59 +00:00
3235c38dbb
Add some scaffolding for a new experimental asmprinter
...
implementation. The idea is that we want asmprinting to
work by converting MachineInstrs into a new MCInst class,
then the per-instruction asmprinter works on MCInst. MCInst
and the new asmprinters will not depend on most of the
llvm code generators. This allows building diassemblers
that don't link in the whole llvm code generator. This is
step #1 of many.
llvm-svn: 73743
2009-06-19 00:47:33 +00:00
8284b66264
merge the common darwin settings from the X86/PPC/ARM targets
...
into DarwinTargetAsmInfo.cpp. The remaining differences should
be evaluated. It seems strange that x86/arm has .zerofill but ppc
doesn't, etc.
llvm-svn: 73742
2009-06-19 00:08:39 +00:00
2cce7d5988
fix build problem pointed out by John Thompson!
...
llvm-svn: 73739
2009-06-18 23:46:04 +00:00
b84764516d
move mangler quote handling from asm printers to TargetAsmInfo.
...
llvm-svn: 73738
2009-06-18 23:41:35 +00:00
4df8e5c800
minor cleanup
...
llvm-svn: 73737
2009-06-18 23:33:13 +00:00
09081b25a5
simplify macro debug info directive handling.
...
llvm-svn: 73736
2009-06-18 23:31:37 +00:00
a0ca298f8a
Remove UseThumbBacktraces. Just check if subtarget is darwin.
...
llvm-svn: 73734
2009-06-18 23:14:30 +00:00
c7ab953177
fix file header
...
llvm-svn: 73733
2009-06-18 23:05:21 +00:00
8dd8a52533
Add a skeleton driver for new machine code level fun. llvm-mc is meant
...
to be a test driver of other components in the system, which will develop
over time.
llvm-svn: 73732
2009-06-18 23:04:45 +00:00
fb8097b576
new testcase for rdar://6983634
...
llvm-svn: 73729
2009-06-18 22:53:26 +00:00
38830b19e9
Add a parent pointer to SCEV, in preparation for getting rid of the global uniquing tables. No functionality change.
...
llvm-svn: 73728
2009-06-18 22:25:12 +00:00
32270cc78e
Improved PHI def marking, replaced some gotos with breaks.
...
llvm-svn: 73727
2009-06-18 22:01:47 +00:00
032bc36939
Allow user to specify separate SVN url for llvm and the test suite.
...
llvm-svn: 73726
2009-06-18 21:39:50 +00:00
b614f1e13c
Simplify.
...
llvm-svn: 73725
2009-06-18 21:35:56 +00:00
7f1ef67a7b
Simplify.
...
llvm-svn: 73723
2009-06-18 20:56:48 +00:00
5a6960fcc3
Add a SmartScopedLock, and use it to simplify code.
...
llvm-svn: 73722
2009-06-18 20:51:00 +00:00
a6dc7456e1
Simplify with SmartRWMutex.
...
llvm-svn: 73721
2009-06-18 20:44:31 +00:00
de9e36a74e
On Darwin, ams printer should output a second label before a jump table so the linker knows it's a new atom. But this is only needed if the jump table is put in a separate section from the function body.
...
llvm-svn: 73720
2009-06-18 20:37:15 +00:00
5e1f6d95a2
Simplify by using no-op-when-not-multithreaded locks.
...
llvm-svn: 73719
2009-06-18 20:36:21 +00:00
8c9ac59455
Generalize LSR's OptimizeSMax to handle unsigned max tests as well
...
as signed max tests. Along with r73717, this helps CodeGen avoid
emitting code for a maximum operation for this class of loop.
llvm-svn: 73718
2009-06-18 20:23:18 +00:00
4d3c3cfffd
Recognize n != 0 ? n : 1 as umax(n, 1). Previously only ULT/UGT/ULE/UGE
...
comparisons were recognized for umax, but instcombine canonicalizes
unsigned comparisons with zero to this simpler form.
llvm-svn: 73717
2009-06-18 20:21:07 +00:00
830ff50abf
Simplify using mutexes that become no-ops when not in multithreaded mode.
...
llvm-svn: 73716
2009-06-18 20:15:26 +00:00
6ee547bb1b
Revert IRBuilder CC propagation. Fix SimplifyLibCalls instead.
...
llvm-svn: 73715
2009-06-18 20:05:31 +00:00
d830eb8365
Simplify a lot of code by using a R/W mutex that becomes a no-op when multithreading is disabled.
...
llvm-svn: 73714
2009-06-18 19:10:19 +00:00
fa107259fe
Use SmartMutex to simplify.
...
llvm-svn: 73712
2009-06-18 18:36:34 +00:00
6c634f39ce
Simplify the SmartMutex implementation a bit.
...
llvm-svn: 73711
2009-06-18 18:29:03 +00:00
1498a7a510
Give RWMutex the SmartRWMutex treatment too.
...
llvm-svn: 73710
2009-06-18 18:26:15 +00:00
68f6598c69
Insert a SmartMutex templated class into the class hierarchy, which takes a template parameter specifying whether this mutex
...
should become a no-op when not running in multithreaded mode. Make sys::Mutex a typedef of SmartMutex<false>, to preserve source compatibility.
llvm-svn: 73709
2009-06-18 17:53:17 +00:00
81ef65d491
Update CMake files to account for new location of Threading.cpp.
...
llvm-svn: 73708
2009-06-18 17:47:09 +00:00
7d42b95b6c
Move Threading.[h|cpp] from Support to System.
...
llvm-svn: 73707
2009-06-18 16:54:52 +00:00
a0348809b6
Remove the code from IVUsers that attempted to handle
...
casted induction variables in cases where the cast
isn't foldable. It ended up being a pessimization in
many cases. This could be fixed, but it would require
a bunch of complicated code in IVUsers' clients. The
advantages of this approach aren't visible enough to
justify it at this time.
llvm-svn: 73706
2009-06-18 16:54:06 +00:00
56bd02c55c
Generalize the zext(trunc(t) & C) instcombine to work even with
...
C is not a low-bits mask, and add a similar instcombine for
zext((trunc(t) & C) ^ C).
llvm-svn: 73705
2009-06-18 16:30:21 +00:00
5a728c908c
Update a comment to reflect the code.
...
llvm-svn: 73704
2009-06-18 16:24:47 +00:00
c5eb1ee4b7
Fix the double checked locking in this file too.
...
llvm-svn: 73703
2009-06-18 16:17:42 +00:00
0fd4eaef30
As pointed out by Duncan, I accidentally dropped the first MemoryFence of the
...
double-checked locking pattern here.
llvm-svn: 73701
2009-06-18 16:08:27 +00:00
e2ff9a6f38
Forgot to commit this hunk
...
llvm-svn: 73693
2009-06-18 08:30:57 +00:00
832b9e6a7a
Propagate calling conv for invokes too
...
llvm-svn: 73692
2009-06-18 08:13:56 +00:00
8527e674ce
Don't use int8_t, that requires DataTypes.h
...
llvm-svn: 73682
2009-06-18 04:56:53 +00:00
6eb4a73633
Fix build when disable thread support.
...
Patch by Howard Su.
llvm-svn: 73680
2009-06-18 04:44:46 +00:00
bb4a7979b5
include header for uint8_t.
...
llvm-svn: 73675
2009-06-18 04:11:43 +00:00
87c4a05008
Add braces to clarify if/else structure and remove warning.
...
llvm-svn: 73673
2009-06-18 03:01:42 +00:00
0e79603588
- Update register allocation hint after coalescing. This is done by the target since the hint is target dependent. This is important for ARM register pair hints.
...
- Register allocator should resolve the second part of the hint (register number) before passing it to the target since it knows virtual register to physical register mapping.
- More fixes to get ARM load / store double word working.
llvm-svn: 73671
2009-06-18 02:04:01 +00:00
9d9acb065a
It looks like nobody is working on PR 4158, so I'm
...
adding a check to catch this case at compile time
instead of quietly generating incorrect code.
That will at least let us identify CBE failures
that are not due to this problem.
llvm-svn: 73668
2009-06-18 01:07:23 +00:00
94c468f419
Fix trailing whitespace from ScalarEvolution::print.
...
llvm-svn: 73666
2009-06-18 00:37:45 +00:00
86c12d7bc4
ARM unified syntax is not specific to ELF; use it for Darwin, too.
...
llvm-svn: 73665
2009-06-18 00:36:17 +00:00
eddf77123a
Teach ScalarEvolution how to recognize another xor(and(x, C), C) case.
...
If C is a single bit and the and gets analyzed as a truncate and
zero-extend, the xor can be represnted as an add.
llvm-svn: 73664
2009-06-18 00:00:20 +00:00
4d8723d47f
Delete unnecessary braces.
...
llvm-svn: 73663
2009-06-17 23:54:37 +00:00
ed14e767e8
Reapply r73647 in a non-broken form.
...
llvm-svn: 73662
2009-06-17 23:49:06 +00:00
5658086052
Fix asm string from MOVi16
...
llvm-svn: 73661
2009-06-17 23:43:36 +00:00
f687a828b2
Thumb2 instructions are enabled only in unified assembler mode.
...
Emit switch directive for it. I have no idea whether this is
requirement for Darwin or not.
llvm-svn: 73660
2009-06-17 23:43:18 +00:00
7f836c7c61
Instcombine zext(trunc(x) & mask) to x&mask, even if the trunc has
...
multiple users.
llvm-svn: 73656
2009-06-17 23:17:05 +00:00
89ee957cbd
Use double-checked locking for this lazy initialization.
...
llvm-svn: 73653
2009-06-17 22:53:57 +00:00
aab59c52cf
Protect the GC table in Function.cpp
...
llvm-svn: 73647
2009-06-17 22:23:31 +00:00
603a0b7a09
Honour calling convention and attributes of Callee by default.
...
llvm-svn: 73646
2009-06-17 22:20:46 +00:00
b299d40d5a
Use atomic increment here.
...
llvm-svn: 73643
2009-06-17 22:01:09 +00:00
c7b91a0820
Thread-safe (and ManagedStatic-ized) LeakDetector!
...
llvm-svn: 73642
2009-06-17 21:56:05 +00:00
0dd39fdf96
Guard mutation of the timing info global.
...
llvm-svn: 73639
2009-06-17 21:28:54 +00:00
eae9772679
We need to use double-checked locking for lazy initialization in this case when running multithreaded.
...
llvm-svn: 73636
2009-06-17 21:16:20 +00:00
667cd9abe2
Fix link to BasicBlock doxygen documentation in Programmer's Manual
...
llvm-svn: 73635
2009-06-17 21:12:26 +00:00
16cab1935e
VNInfo cleanup.
...
llvm-svn: 73634
2009-06-17 21:01:20 +00:00
432af7ace0
Add -disable-output to a bunch of tests that don't care about the output.
...
llvm-svn: 73633
2009-06-17 20:56:26 +00:00
81b6463ed0
This fixes a bug introduced in 72661, which can
...
move loads back past a check that the load address
is valid, see new testcase. The test that went
in with 72661 has exactly this case, except that
the conditional it's moving past is checking
something else; I've settled for changing that
test to reference a global, not a pointer. It
may be possible to scan all the tests you pass and
make sure none of them are checking any component
of the address, but it's not trivial and I'm not
trying to do that here.
llvm-svn: 73632
2009-06-17 20:48:23 +00:00
f89c38cace
Factor out some common code.
...
llvm-svn: 73631
2009-06-17 20:43:39 +00:00
65c5cd728e
Add an RAII ScopedWriter, which allows one to acquire a writer lock for the duration of a scope. Simplify a lot of uses of
...
writer locks in Constants.cpp by using it.
llvm-svn: 73630
2009-06-17 20:34:43 +00:00
61794049e4
Simplify the locking on the Constants tables, and make it more efficient, by pushing it into the ValueMap from the callers.
...
Document those ValueMap functions that are _not_ locked, so that callers are aware that they need to do the locking themselves.
llvm-svn: 73628
2009-06-17 20:10:08 +00:00
Dan Gohman