3a302cbb08
Convert SCEV from FoldingSetNode to FastFoldingSetNode. This eliminates
...
a bunch of redundent code in Profile methods, and prepares for upcoming
changes to do improved memoization.
llvm-svn: 75494
2009-07-13 20:50:19 +00:00
e3ac197457
Reverted r75484. It was causing a failure with Apple-style builds.
...
llvm-svn: 75491
2009-07-13 20:27:41 +00:00
de54478bf0
Add infrastructure to allow post instruction printing action triggers.
...
We'll eventually use this to print comments in asm files and do other
fun things.
This adds interfaces to the AsmPrinter and changes TableGen to invoke
the postInstructionAction when appropriate. It also add parameters to
TargetAsmInfo to control comment layout.
llvm-svn: 75490
2009-07-13 20:25:48 +00:00
01218d5b17
Add NetBSD to the Triple class, patch by Krister Walfridsson!
...
llvm-svn: 75489
2009-07-13 20:22:23 +00:00
1b7661e588
More standard way of specifying greater than Leopard.
...
llvm-svn: 75488
2009-07-13 20:18:43 +00:00
0a860df8ef
On greater than Leopard systems, place exception tables in the __TEXT section.
...
llvm-svn: 75484
2009-07-13 18:48:39 +00:00
a4f709ee5a
Change printInstruction to not print a trailing newline. Value::dump
...
always adds a newline, so this fixes Value::dump printing an
extra blank line.
llvm-svn: 75481
2009-07-13 18:27:59 +00:00
844d6c82a7
Fix comment typos.
...
llvm-svn: 75479
2009-07-13 18:11:36 +00:00
370aadabfc
factor the 'optimized sort' code out into a static helper function
...
and use it from one more place. Patch by Jakub Staszak!
llvm-svn: 75478
2009-07-13 17:20:05 +00:00
2f0c1c44d5
Move the re-sort of invalidated NonLocalPointerDeps cache earlier
...
so that all code paths get it. PR4256 was about a case where the
phi translation loop would find all preds in the Visited cache, so
it could get by without re-sorting the NonLocalPointerDeps cache.
Fix this by resorting it earlier, there is no reason not to do this.
This patch inspired by Jakub Staszak's patch.
llvm-svn: 75476
2009-07-13 17:14:23 +00:00
20f6ac0738
Make some more changes suggested by Chris. Manipulators go away.
...
llvm-svn: 75472
2009-07-13 16:49:27 +00:00
3a86fe943c
Match declaration to definition (missed a few).
...
llvm-svn: 75456
2009-07-13 06:04:06 +00:00
334f808562
Match declaration to definition.
...
llvm-svn: 75454
2009-07-13 06:00:13 +00:00
c07adb7b0b
Fix build on Linux.
...
llvm-svn: 75453
2009-07-13 05:49:04 +00:00
78be93acb3
Fix some non-sensical code.
...
- This makes it more like other similar code in Archive handling.
llvm-svn: 75452
2009-07-13 05:29:34 +00:00
763ace92c0
Switch to raw_ostream.
...
llvm-svn: 75451
2009-07-13 05:27:30 +00:00
43edca25b3
Removed some junk code that snuck in to an earlier commit.
...
llvm-svn: 75450
2009-07-13 05:01:19 +00:00
571bf54569
Fix an error in ConstantRange::getSignedMax on wrapped ranges. Thanks once
...
again to Daniel Dunbar and KLEE!
llvm-svn: 75449
2009-07-13 04:50:21 +00:00
5edc459220
'i8 full-range' sign extended to i16 should equal [-128, 128) not [-128, 127).
...
Found by Daniel Dunbar and KLEE.
llvm-svn: 75448
2009-07-13 04:17:23 +00:00
53a52215b5
Begin the painful process of tearing apart the rat'ss nest that is Constants.cpp and ConstantFold.cpp.
...
This involves temporarily hard wiring some parts to use the global context. This isn't ideal, but it's
the only way I could figure out to make this process vaguely incremental.
llvm-svn: 75445
2009-07-13 04:09:18 +00:00
5302389442
Multiply was very wrong for wrapped ranges. This supplies a half-fix that will
...
generally return Full on all wrapped inputs. "Fixes" PR4545.
llvm-svn: 75444
2009-07-13 03:27:41 +00:00
73b704dd9c
Fix a bug summing two full sets. The overflow checking doesn't handle sets as
...
large as the full set, only those one size smaller. Thanks to Daniel Dunbar
who found this bug using Klee!
llvm-svn: 75443
2009-07-13 02:49:08 +00:00
65553d4505
Match declaration to definition.
...
llvm-svn: 75440
2009-07-12 23:50:34 +00:00
7b9d6ebb9c
remove llvm.part.set.* and llvm.part.select.*. They have never been
...
implemented in codegen, have no frontend to generate them, and are
better implemented with pattern matching (like the ppc backend does
to generate rlwimi/rlwinm etc).
PR4543
llvm-svn: 75430
2009-07-12 21:08:53 +00:00
5e2ebcba40
Use llvm_report_error, not llvm_unreachable.
...
llvm-svn: 75429
2009-07-12 21:01:49 +00:00
df555fd6f8
Improve sys::Path::makeAbsolute on Win32.
...
- Patch by Viktor Kutuzov!
- Minor tweak by me to add llvm_unreachable calls on FIXMEd error paths.
llvm-svn: 75424
2009-07-12 20:23:56 +00:00
08954aa4e1
Fix assert(0) conversion, as suggested by Chris.
...
llvm-svn: 75423
2009-07-12 20:07:01 +00:00
a3c69aeb65
"UNIX paths can contain non-ASCII characters, so Path::isValid() should not reject paths with such characters in them. The attached patch removes the check from Path::isValid()."
...
patch by Benjamin Kramer!
llvm-svn: 75421
2009-07-12 19:01:16 +00:00
ed0e1a0552
Implement support for promotion of AND/OR/XOR on integer types.
...
The blackfin processor has a legal i16 type, but only logic operations on i32.
llvm-svn: 75419
2009-07-12 18:10:18 +00:00
6b9f63cafa
Fix types in PromoteNode handling of CTPOP and friends.
...
llvm-svn: 75418
2009-07-12 17:43:20 +00:00
69208f0f9e
Remove extra \n from LLVM_UNREACHABLE calls.
...
llvm-svn: 75416
2009-07-12 07:15:17 +00:00
f1b8cb3760
Implement udiv for ConstantRanges.
...
llvm-svn: 75413
2009-07-12 05:18:18 +00:00
5b76fc03ae
Temporarily revert r75408. It appears to break the Apple-style builds:
...
x86_64-apple-darwin10-gcc -c -g -O2 -DIN_GCC -W -Wall -Wwrite-strings -Wstrict-prototypes -Wmissing-prototypes -pedantic -Wno-long-long -Wno-variadic-macros -Wno-overlength-strings -Wold-style-definition -Wmissing-format-attribute -mdynamic-no-pic -DHAVE_CONFIG_H -I. -I. -I/Volumes/Sandbox/Buildbot/llvm/build.llvm-gcc-x86_64-darwin10-selfhost/build/llvmgcc42.roots/llvmgcc42~obj/src/gcc -I/Volumes/Sandbox/Buildbot/llvm/build.llvm-gcc-x86_64-darwin10-selfhost/build/llvmgcc42.roots/llvmgcc42~obj/src/gcc/. -I/Volumes/Sandbox/Buildbot/llvm/build.llvm-gcc-x86_64-darwin10-selfhost/build/llvmgcc42.roots/llvmgcc42~obj/src/gcc/../include -I./../intl -I/Volumes/Sandbox/Buildbot/llvm/build.llvm-gcc-x86_64-darwin10-selfhost/build/llvmgcc42.roots/llvmgcc42~obj/src/gcc/../libcpp/include -I/Volumes/Sandbox/Buildbot/llvm/build.llvm-gcc-x86_64-darwin10-selfhost/build/llvmgcc42.roots/llvmgcc42~obj/src/gcc/../libdecnumber -I../libdecnumber -I/Volumes/Sandbox/Buildbot/llvm/build.llvm-gcc-x86_64-darwin10-selfhost/build/llvmCore.roots/llvmCore~dst/Developer/usr/local/include -I/Volumes/Sandbox/Buildbot/llvm/build.llvm-gcc-x86_64-darwin10-selfhost/build/llvmCore.roots/llvmCore~obj/src/include -DENABLE_LLVM -I/Volumes/Sandbox/Buildbot/llvm/build.llvm-gcc-x86_64-darwin10-selfhost/build/llvmCore.roots/llvmCore~dst/Developer/usr/local/include -D_DEBUG -D_GNU_SOURCE -D__STDC_LIMIT_MACROS -D__STDC_CONSTANT_MACROS -DLLVM_VERSION_INFO='"9999"' -DBUILD_LLVM_APPLE_STYLE /Volumes/Sandbox/Buildbot/llvm/build.llvm-gcc-x86_64-darwin10-selfhost/build/llvmgcc42.roots/llvmgcc42~obj/src/gcc/tree-ssa-alias.c -o tree-ssa-alias.o
/var/tmp//ccJQ2JBT.s:4134:Incorrect register `%rcx' used with `l' suffix
make[2]: *** [tree-ssa-live.o] Error 1
make[2]: *** Waiting for unfinished jobs....
llvm-svn: 75412
2009-07-12 02:49:22 +00:00
2951c990cd
Implement ConstantRange::multiply based on the code in LoopVR.
...
llvm-svn: 75410
2009-07-12 02:19:05 +00:00
575db66e1b
Remove check which is duplicated in
...
InstCombiner::visitSelectInstWithICmp.
llvm-svn: 75409
2009-07-12 02:00:05 +00:00
02c4339bde
eliminate MOV64r0 in favor of a Pat<> pattern. This is only nontrivial because
...
the div lowering code explicitly references it.
llvm-svn: 75408
2009-07-12 00:47:55 +00:00
38df005e12
fix x86-64 static codegen to materialize the address of a global with movl instead
...
of lea. It is better for code size (and presumably efficiency) to use:
movl $foo, %eax
rather than:
leal foo, eax
Both give a nice zero extending "move immediate" instruction, the former is just
smaller. Note that global addresses should be handled different by the x86
backend, but I chose to follow the style already in place and add more fixme's.
llvm-svn: 75403
2009-07-11 23:17:29 +00:00
48cee9b4c1
fix a bug in my cleanup patch
...
llvm-svn: 75402
2009-07-11 23:07:30 +00:00
4d10f1a6c9
comment cleanup, reduce nesting.
...
llvm-svn: 75398
2009-07-11 22:50:33 +00:00
24f6ad5ae4
remove some dead patterns, WrapperRIP doesn't exist in -static mode
...
anymore, so these aren't needed.
llvm-svn: 75397
2009-07-11 22:47:21 +00:00
2f67295aac
silence a vc++ warning.
...
llvm-svn: 75393
2009-07-11 22:31:59 +00:00
c0bf46b226
Revert r75252 which was causing some crashes at compile time.
...
llvm-svn: 75384
2009-07-11 20:38:25 +00:00
e91900097e
Fix PR4533, which is about buggy codegen in x86-64 -static mode.
...
Basically, using:
lea symbol(%rip), %rax
is not valid in -static mode, because the current RIP may not be
within 32-bits of "symbol" when an app is built partially pic and
partially static. The fix for this is to compile it to:
lea symbol, %rax
It would be better to codegen this as:
movq $symbol, %rax
but that will come next.
The hard part of fixing this bug was fixing abi-isel, which was actively
testing for the wrong behavior. Also, the RUN lines are completely impossible
to understand what they are testing. To help with this, convert the -static
x86-64 codegen tests to use filecheck. This is much more stable and makes it
more clear what the codegen is expected to be.
llvm-svn: 75382
2009-07-11 20:29:19 +00:00
56d0659726
assert(0) -> LLVM_UNREACHABLE.
...
Make llvm_unreachable take an optional string, thus moving the cerr<< out of
line.
LLVM_UNREACHABLE is now a simple wrapper that makes the message go away for
NDEBUG builds.
llvm-svn: 75379
2009-07-11 20:10:48 +00:00
3c9e8d3b1f
Delete the temp dir even when '--temp-dir' is specified.
...
llvm-svn: 75374
2009-07-11 19:27:07 +00:00
fc69ee2cc6
Fix handling of max and full set.
...
A full set is a constant range that represents any number. If you take the
umax of that and [5, 10) you end up with [5, INT_MAX] because the values less
than 5 would be umax's against a value which is at least 5.
llvm-svn: 75372
2009-07-11 19:22:21 +00:00
9c5fc41547
Clarify and simplify.
...
llvm-svn: 75366
2009-07-11 17:04:01 +00:00
ccb29cd290
Convert more assert(0)+abort() -> LLVM_UNREACHABLE,
...
and abort()/exit() -> llvm_report_error().
llvm-svn: 75363
2009-07-11 13:10:19 +00:00
017288a4fc
Don't put IT instruction before conditional branches.
...
llvm-svn: 75361
2009-07-11 07:26:20 +00:00
0794c6a083
Smarter isel of ldrsb / ldrsh. Only make use of these when [r,r] address is feasible.
...
llvm-svn: 75360
2009-07-11 07:08:13 +00:00
cd4cdd1157
Major changes to Thumb (not Thumb2). Many 16-bit instructions either modifies CPSR when they are outside the IT blocks, or they can predicated when in Thumb2. Move the implicit def of CPSR to an optional def which defaults CPSR. This allows the 's' bit to be toggled dynamically.
...
A side-effect of this change is asm printer is now using unified assembly. There are some minor clean ups and fixes as well.
llvm-svn: 75359
2009-07-11 06:43:01 +00:00
bf041366c4
80 col violation.
...
llvm-svn: 75358
2009-07-11 06:37:27 +00:00
dcfdce9067
Move a method that creates constant ranges relative to another constant range
...
per icmp predicate out of predsimplify and into ConstantRange.
Add another utility method that determines whether one range is a subset of
another. Combine with the former to determine whether icmp pred range, range
is known to be true or not.
llvm-svn: 75357
2009-07-11 06:15:39 +00:00
ede2ce71aa
Fix up support for OptionalDefOperand when it defaults to an actual register def. I need this to get ready for major Thumb1 surgery.
...
llvm-svn: 75328
2009-07-11 01:06:50 +00:00
106f2885d1
Use CreateStackStoreLoad helper in more places.
...
llvm-svn: 75320
2009-07-11 00:11:07 +00:00
b4dfce2dba
Two-address pass should use findCommutedOpIndices to determine what registers are commuted.
...
llvm-svn: 75317
2009-07-11 00:04:23 +00:00
1297799096
Use findCommutedOpIndices to find the operands to commute.
...
llvm-svn: 75312
2009-07-10 23:26:12 +00:00
a51f8ebf1a
Add superclasses of ARM Neon quad registers. The Q2PR class contains pairs of
...
quad registers and the Q4PR class holds sets of 4 quad registers.
llvm-svn: 75309
2009-07-10 23:09:06 +00:00
73fd66c06b
Add new vector types for 192-bit, 348-bit and 512-bit sizes.
...
These are needed to represent ARM Neon struct datatypes containing 2, 3 or 4
separate vectors.
llvm-svn: 75308
2009-07-10 23:05:09 +00:00
2745607dc6
remove the "debug" modifier, it is only used by one instruction which can
...
never be generated.
llvm-svn: 75305
2009-07-10 22:34:11 +00:00
07cadaf113
add support for .zerofill, patch by Kevin Enderby!
...
llvm-svn: 75301
2009-07-10 22:20:30 +00:00
228be77c19
Oops. s/#if defined(USE_OPROFILE)/#if USE_OPROFILE/. We #define
...
USE_OPROFILE to 0 on some paths through configure, which does the
wrong thing with #if defined().
llvm-svn: 75300
2009-07-10 22:13:21 +00:00
a6009306c2
make pcrel and non-pcrel global printing more similar.
...
llvm-svn: 75298
2009-07-10 21:57:21 +00:00
7997cbf2d5
Undo my brain cramp.
...
llvm-svn: 75290
2009-07-10 21:31:42 +00:00
c97b778b3c
Make changes suggested by Chris and eliminate newly-added raw_ostream
...
hooks as they're no longer needed.
The major change with this patch is to make formatted_raw_ostream usable
by any client of raw_ostream.
llvm-svn: 75283
2009-07-10 21:14:44 +00:00
bf3d6ba557
Add a --with-oprofile flag to configure, which uses OProfile's agent
...
library to tell it the addresses of JITted functions. For a
particular program, this changes the opreport -l output from:
samples % image name symbol name
48182 98.9729 anon (tgid:19412 range:0x7f12ccaab000-0x7f12cdaab000) anon (tgid:19412 range:0x7f12ccaab000-0x7f12cdaab000)
11 0.0226 libstdc++.so.6.0.9 /usr/lib/libstdc++.so.6.0.9
to:
samples % image name symbol name
24565 60.7308 19814.jo fib_left
15365 37.9861 19814.jo fib_right
22 0.0544 ld-2.7.so do_lookup_x
llvm-svn: 75279
2009-07-10 21:08:20 +00:00
25e7f91559
code model is never set to default.
...
llvm-svn: 75278
2009-07-10 21:03:06 +00:00
7dce9919e1
fix indentation
...
llvm-svn: 75277
2009-07-10 21:01:59 +00:00
21c2940553
remove the now-dead TM argument to these methods.
...
llvm-svn: 75276
2009-07-10 21:00:45 +00:00
ba4d73310a
make PIC vs DynamicNoPIC be explicit in PICStyles.
...
llvm-svn: 75275
2009-07-10 20:58:47 +00:00
bd3e560f1a
some minor simplifications.
...
llvm-svn: 75274
2009-07-10 20:53:38 +00:00
e2f524f176
add a couple of predicates to test for "stub style pic in PIC mode" and "stub style pic in dynamic-no-pic" mode.
...
llvm-svn: 75273
2009-07-10 20:47:30 +00:00
028e6158c2
Remove ScalarEvolution::hasSCEV, which isn't being used, and which
...
breaks encapsulation. Also remove a dead prototype for setSCEV.
llvm-svn: 75272
2009-07-10 20:25:29 +00:00
48d983c86e
Avoid compiler warnings when assertions are turned off.
...
llvm-svn: 75269
2009-07-10 20:10:14 +00:00
45210aab86
Avoid compiler warnings if assertions turned off.
...
llvm-svn: 75267
2009-07-10 20:07:07 +00:00
bb00fe0dc6
CMOVxx doesn't swap operands which it's commuted.
...
llvm-svn: 75266
2009-07-10 19:26:57 +00:00
f075943584
Remove TargetInstrInfo::CommuteChangesDestination and added findCommutedOpIndices which returns the operand indices which are swapped (when applicable). This allows for some code clean up and future enhancements.
...
llvm-svn: 75264
2009-07-10 19:15:51 +00:00
68ba0df07d
Eliminate an unnecessary include.
...
llvm-svn: 75256
2009-07-10 17:55:38 +00:00
16e7674f4b
Push LLVMContext through the PatternMatch API.
...
llvm-svn: 75255
2009-07-10 17:35:01 +00:00
81cdd21dcb
Predicate VFP instructions on HasVFP2 instead of IsARM. This allows VFP instructions with thumb-2.
...
llvm-svn: 75254
2009-07-10 17:03:29 +00:00
a9938d8226
Generalize ScalarEvolution's cast-folding code to support more kinds
...
of loops. Add several new functions to for working with ScalarEvolution's
add-hoc value-range analysis functionality.
llvm-svn: 75252
2009-07-10 16:42:52 +00:00
96b491a1fa
Update for GlobalVariables ctor change.
...
llvm-svn: 75251
2009-07-10 16:42:19 +00:00
c9b1efd515
t2LDM_RET does not fall-through.
...
llvm-svn: 75250
2009-07-10 15:33:46 +00:00
0006349f4d
Add Thumb2ITBlockPass.cpp to CMakeLists.txt, fixing
...
the cmake build.
llvm-svn: 75246
2009-07-10 08:31:50 +00:00
9531b039b9
make this more like printOperand. Perhaps some merging will happen
...
tomorrow.
llvm-svn: 75245
2009-07-10 07:50:52 +00:00
20073edf67
simplify fast isel by using ClassifyGlobalReference. This
...
elimiantes the last use of GVRequiresExtraLoad, so delete it.
llvm-svn: 75244
2009-07-10 07:48:51 +00:00
93f0f79178
eliminate GVRequiresRegister, replacing it with predicates we
...
need for other purposes.
llvm-svn: 75243
2009-07-10 07:38:24 +00:00
2ff35f528c
change a bunch of logic in LowerGlobalAddress to leverage the work
...
done in ClassifyGlobalReference instead of reconstructing the info
awkwardly.
llvm-svn: 75240
2009-07-10 07:34:39 +00:00
d3f32c725b
add a predicate to determine if a global var reference requires a
...
PIC-base to be added in.
llvm-svn: 75238
2009-07-10 07:33:30 +00:00
dc842c06c2
move some classification logic around. Now GVRequiresExtraLoad
...
is just a trivial wrapper around "ClassifyGlobalReference", which
stole a ton of logic from LowerGlobalAddress.
llvm-svn: 75237
2009-07-10 07:20:05 +00:00
ca9d784bf1
change isGlobalStubReference to take target flags instead of a MachineOperand.
...
llvm-svn: 75236
2009-07-10 06:29:59 +00:00
e6d259340e
convert some late code (called by regalloc and code emission)
...
to use isGlobalStubReference instead of GVRequiresExtraLoad
(which should really be part of isel).
llvm-svn: 75234
2009-07-10 06:07:08 +00:00
377f1d5373
add a new predicate method that says whether a GlobalValue
...
MachineOperand is a reference to a stub, not a reference to the
global variable itself. Look no context needed!
llvm-svn: 75233
2009-07-10 06:06:17 +00:00
b9af63a4d2
GVRequiresExtraLoad is now never used for calls, simplify it based on this.
...
llvm-svn: 75232
2009-07-10 05:52:02 +00:00
ace6ec26d9
actually, just eliminate PCRelGVRequiresExtraLoad. It makes the code
...
more complex and slow than just directly testing what we care about.
llvm-svn: 75231
2009-07-10 05:48:03 +00:00
7277a807f0
There is only one case where GVRequiresExtraLoad returns true for calls:
...
split its handling out to PCRelGVRequiresExtraLoad, and simplify code
based on this.
llvm-svn: 75230
2009-07-10 05:45:15 +00:00
1cc7ae7c3b
the "isDirectCall" operand of GVRequiresRegister is always false, eliminate it.
...
llvm-svn: 75229
2009-07-10 05:37:11 +00:00
8212d37e7d
split call handling out of X86SelectAddress into X86SelectCallAddress
...
llvm-svn: 75228
2009-07-10 05:33:42 +00:00
405d024c71
convert a helper method to be a static function instead of a
...
template. Also convert it to take a MachineOperand instead of a GV*
llvm-svn: 75227
2009-07-10 05:27:43 +00:00
01b8630879
More info about Thumb1 predication support.
...
llvm-svn: 75220
2009-07-10 02:10:17 +00:00
61671c87a7
We don't need separate thumb1 instructions tADDSi3 etc. for addc and subc. The "normal" version always modify condition register CPSR so we should just use def : pat to match to the same instructions.
...
llvm-svn: 75219
2009-07-10 02:09:04 +00:00
0f9cce7951
Add a thumb2 pass to insert IT blocks.
...
llvm-svn: 75218
2009-07-10 01:54:42 +00:00
3b88dd6900
Move isPredicated from .cpp to .h
...
llvm-svn: 75217
2009-07-10 01:38:27 +00:00
5d8a1b2c18
80 col violation.
...
llvm-svn: 75212
2009-07-10 00:45:16 +00:00
223ac25930
Remove a bogus assertion.
...
llvm-svn: 75206
2009-07-10 00:23:48 +00:00
75aa75cabc
Replace TM.getRegisterInfo() calls by TRI instance variable.
...
Use getAsmName() method instead of accessing AsmName field directly.
llvm-svn: 75205
2009-07-10 00:14:05 +00:00
cd94645665
Redesign this to avoid standard stream classes. This stream class
...
provides pretty -printing of comments and other such things in asm
files.
llvm-svn: 75202
2009-07-09 23:56:35 +00:00
9ce44e2521
Handle 'a' modifier on inline assembly operands.
...
This is part of the fix for pr4521.
llvm-svn: 75201
2009-07-09 23:54:51 +00:00
1e5f00e7a7
This started as a small change, I swear. Unfortunately, lots of things call the [I|F]CmpInst constructors. Who knew!?
...
llvm-svn: 75200
2009-07-09 23:48:35 +00:00
ab11a81fc0
Add some hooks that a redesigned AsmStream needs to do its job. These
...
allow derived classes to examine the stream buffer before it's flushed.
llvm-svn: 75199
2009-07-09 23:43:41 +00:00
26b2ba4285
Added Thumb IT instruction.
...
llvm-svn: 75198
2009-07-09 23:43:36 +00:00
f76798769f
Fix an apparent copy-and-paste problem in an error message.
...
llvm-svn: 75197
2009-07-09 23:42:59 +00:00
ae4f2e142a
Another todo entry.
...
llvm-svn: 75192
2009-07-09 23:17:28 +00:00
dc33ae27ed
Revert the part of 75177 that split ConstantRange into two classes, and
...
merge the new functionality and unittests into ConstantRange. Thanks to
Nick Lewycky for pointing out that it isn't necessary to have two separate
classes here.
llvm-svn: 75191
2009-07-09 23:16:10 +00:00
4605e8aac4
Initial support for load / store multiple opt pass Thumb2 support (post-allocation only). It's kind of there, but not quite. I'll return to this later.
...
llvm-svn: 75190
2009-07-09 23:11:34 +00:00
1622d7b429
Fix ldm / stm unified syntax; add t2LDM_RET.
...
llvm-svn: 75188
2009-07-09 22:58:39 +00:00
a02fc2d327
LDM_RET should be marked mayLoad.
...
llvm-svn: 75187
2009-07-09 22:57:41 +00:00
ad6c36f065
If -fomit-frame-pointer is used, we still need to record when the %esp register
...
is modified. Otherwise, the unwinder will get confused. The old code (before I
started my hacking) did this. It dropped on the floor, because I wasn't aware of
this requirement.
On the plus side, if we use "alloca" in a function, we create frame pointers
even with -fomit-frame-pointer is enabled!
This is a Good Thing(tm)!!!
llvm-svn: 75183
2009-07-09 22:30:02 +00:00
7591d02c84
Fix ldrd / strd address mode matching code. It allows for +/- 8 bit offset. Also change the printer to make the scale 4 explicit.
...
Note, we are not yet generating these instructions.
llvm-svn: 75181
2009-07-09 22:21:59 +00:00
5035fbf91c
Add a ConstantSignedRange class, which does for signed integers
...
what ConstantRange does for unsigned integers. Factor out a
common base class for common functionality.
Add some new functions for performing arithmetic on constant
ranges. Some of these are currently just stubbed out with
conservative implementations.
Add unittests for ConstantRange and ConstantSignedRange.
llvm-svn: 75177
2009-07-09 22:07:27 +00:00
2b77eef160
Make EXTRACT_VECTOR_ELT a bit more flexible in terms of the returned
...
value. Adjust other code to deal with that correctly. Make
DAGTypeLegalizer::PromoteIntRes_EXTRACT_VECTOR_ELT take advantage of
this new flexibility to simplify the code and make it deal with unusual
vectors (like <4 x i1>) correctly. Fixes PR3037.
llvm-svn: 75176
2009-07-09 22:01:03 +00:00
f9870125fc
Add a Thumb readme entry.
...
llvm-svn: 75173
2009-07-09 20:50:52 +00:00
ad93707e16
Correct comment.
...
llvm-svn: 75172
2009-07-09 20:40:44 +00:00
9f3bd4a728
bug 4524: Add MSP430 to the cmake build.
...
llvm-svn: 75170
2009-07-09 20:27:09 +00:00
546a7bbe8f
Handle 'A' modifier in inline asms.
...
gcc.apple/asm-block-13.c
gcc.apple/asm-block-57.c
llvm-svn: 75169
2009-07-09 20:06:27 +00:00
eb9896ac3b
Remove some duplication.
...
llvm-svn: 75163
2009-07-09 19:37:17 +00:00
b1e8563376
Rename -t to --temp-dir.
...
-t is already used by gcc in the meaning 'ld -t' (trace). The Base plugin may
want to emulate this behaviour.
llvm-svn: 75162
2009-07-09 19:36:08 +00:00
092bc51cdb
As Chris pointed out, we don't actually need to pass the context around here.
...
llvm-svn: 75161
2009-07-09 18:44:09 +00:00
55b9726596
back out r75156 for now.
...
llvm-svn: 75160
2009-07-09 18:41:57 +00:00
29fd313e9e
A little bit more LLVMContextification.
...
llvm-svn: 75159
2009-07-09 18:36:20 +00:00
7bf08beb2e
Handle Thumb-2 addressing modes during FP elimination.
...
llvm-svn: 75158
2009-07-09 18:35:52 +00:00
67cde99e90
Add some classes to produce pretty-printed asm. We'll use these
...
shortly to provide nicely printed comments and other goodies in
asm files.
llvm-svn: 75156
2009-07-09 18:27:23 +00:00
093e9ed866
Fixed handling of -t. It gets the prirority for temp dir name.
...
llvm-svn: 75154
2009-07-09 18:06:27 +00:00
0504e0a222
Thread LLVMContext through MVT and related parts of SDISel.
...
llvm-svn: 75153
2009-07-09 17:57:24 +00:00
28ad7547d2
add llvm-mc support for parsing the .lcomm directive, patch by Kevin Enderby!
...
llvm-svn: 75148
2009-07-09 17:25:12 +00:00
4417b4c80b
Add support for other GraphViz display tools. This can help
...
with very large graphs, where dot isn't necessarily the
most visually pleasing way of looking at the graph.
llvm-svn: 75144
2009-07-09 17:06:18 +00:00
5df1bced69
remove dead function.
...
llvm-svn: 75143
2009-07-09 16:57:49 +00:00
5911537b68
Misc encoding fixes; reported on llvmdev.
...
llvm-svn: 75142
2009-07-09 16:49:25 +00:00
433cf71144
eraseFromDisk no longer throws.
...
llvm-svn: 75139
2009-07-09 16:17:28 +00:00
ab81d2f5e9
There's no need to consider PHI nodes in the same block as the instruction
...
we're inserting sigma/phi functions for. Patch by Andre Tavares.
llvm-svn: 75138
2009-07-09 15:59:27 +00:00
eb373ad2af
Add some statistics to SSI so we can see what it's up to.
...
Add an -ssi-everything pass which calls createSSI on everything in the function.
llvm-svn: 75135
2009-07-09 15:33:14 +00:00
7160d3b6d2
Missing )
...
llvm-svn: 75126
2009-07-09 08:48:45 +00:00
a383ffa219
Add a -t="dir" option to the driver. This can be used to specify the directory to be used as TempDir if somebody doesn't want to use the standard /tmp.
...
llvm-svn: 75121
2009-07-09 08:23:38 +00:00
212c0506cd
simplify this logic a bit more.
...
llvm-svn: 75118
2009-07-09 07:02:30 +00:00
72e3deca47
move reasoning about darwin $non_lazy_ptr stubs from asmprinter into
...
isel.
llvm-svn: 75117
2009-07-09 06:59:17 +00:00
7452c968e4
Targets sometimes assign fixed stack object to spill certain callee-saved
...
registers based on dynamic conditions. For example, X86 EBP/RBP, when used as
frame register has to be spilled in the first fixed object. It should inform
PEI this so it doesn't get allocated another stack object. Also, it should not
be spilled as other callee-saved registers but rather its spilling and restoring
are being handled by emitPrologue and emitEpilogue. Avoid spilling it twice.
llvm-svn: 75116
2009-07-09 06:53:48 +00:00
27e32c092e
Reorg includes.
...
llvm-svn: 75115
2009-07-09 06:49:09 +00:00
770e0429be
rearrange some code, no functionality change.
...
llvm-svn: 75113
2009-07-09 06:41:35 +00:00
c58f1fb0df
make direct calls set MO_PLT or MO_DARWIN_STUB as appropriate with fast isel.
...
llvm-svn: 75112
2009-07-09 06:34:26 +00:00
3ffd238f1c
reduce nesting by rearranging branches.
...
llvm-svn: 75110
2009-07-09 05:55:04 +00:00
e0dae0e716
make isel use MO_PIC_BASE_OFFSET when lowering globalvalues on darwin in pic
...
mode, instead of having asmprinter just "know" to print them.
llvm-svn: 75109
2009-07-09 05:47:33 +00:00
0667b145c4
various minor cleanups, no functionality change.
...
llvm-svn: 75108
2009-07-09 05:42:07 +00:00
d047d06358
make isel decide whether to emit $stub's on darwin instead of asmprinter.
...
llvm-svn: 75107
2009-07-09 05:27:35 +00:00
3e41fec4df
pc-relative references are *always* to functions, never to globals. ISel
...
is incapable of lowering direct references to globals like this, zap the code.
llvm-svn: 75106
2009-07-09 05:19:23 +00:00
82aa63c9b0
make cygwin use its own stubs set, instead of overloading a darwin one.
...
llvm-svn: 75105
2009-07-09 05:09:24 +00:00
d5d80ab6b6
Make isel determine where to emit PLT-relative calls instead of having
...
asmprinter do it.
llvm-svn: 75104
2009-07-09 05:02:21 +00:00
b4de36ea47
isPICStyleStub() is now never true in -static mode, so simplify code.
...
llvm-svn: 75102
2009-07-09 04:43:12 +00:00
fef11d6e77
simplify some code based on the fact that picstyles != none are only valid
...
in pic or dynamic-no-pic mode. Also, x86-64 never used picstylegot.
llvm-svn: 75101
2009-07-09 04:39:06 +00:00
62f568c037
all this logic always returns true because GOT mode is never active in x86-64 mode.
...
Simplify it away, someone should evaluate this.
llvm-svn: 75100
2009-07-09 04:27:47 +00:00
3945dd0c44
isPICStyleRIPRel() and friends are never true in -static mode.
...
Simplify code based on this.
llvm-svn: 75099
2009-07-09 04:24:46 +00:00
4e705e421c
Move PR4517 to README.txt.
...
llvm-svn: 75098
2009-07-09 04:03:30 +00:00
dab7b06de9
Improved tracking of value number kills. VN kills are now represented
...
as an (index,bool) pair. The bool flag records whether the kill is a
PHI kill or not. This code will be used to enable splitting of live
intervals containing PHI-kills.
A slight change to live interval weights introduced an extra spill
into lsr-code-insertion (outside the critical sections). The test
condition has been updated to reflect this.
llvm-svn: 75097
2009-07-09 03:57:02 +00:00
117608b129
.o file writing shouldn't mess around with pic/relo models like the JIT does.
...
llvm-svn: 75096
2009-07-09 03:40:30 +00:00
0b0c5cc376
move a hack out of the asm-printer specific path to the main target selection path.
...
llvm-svn: 75095
2009-07-09 03:37:30 +00:00
7e3abf1ab0
many more cleanups, for example if in the "none" pic-style,
...
make sure we're set to static codegen. Simplify the decision
tree of target->picstyle/picmode settings.
llvm-svn: 75094
2009-07-09 03:32:31 +00:00
821084a356
Reduce indentation in GVRequiresExtraLoad. Return true for windows
...
with DLLImport symbols even when in -static mode.
llvm-svn: 75093
2009-07-09 03:27:27 +00:00
1c5bf9d26d
When in -static mode, force the PIC style to none. Doing this requires fixing
...
code which conflated RIPRel PIC with x86-64. Fix these to just check for X86-64
directly.
llvm-svn: 75092
2009-07-09 03:15:51 +00:00
384a738b4a
merge two identical functions and simplify things that are GOT specific
...
llvm-svn: 75091
2009-07-09 02:55:47 +00:00
4d1e9bf717
hoist check for IsTailCall to callers. Eliminate redundant check for
...
x86-64: GOT-style PIC is never used on x86-64.
llvm-svn: 75090
2009-07-09 02:46:53 +00:00
88765d4831
change a few methods to be static functions.
...
llvm-svn: 75089
2009-07-09 02:44:11 +00:00
b1f0ac2a2f
one more added assert.
...
llvm-svn: 75088
2009-07-09 02:43:55 +00:00
47f64ea174
move handling of dllimport linkage in isel, not in asmprinter.
...
llvm-svn: 75086
2009-07-09 00:58:53 +00:00
a8e08a4429
one more assertion!
...
llvm-svn: 75083
2009-07-09 00:49:29 +00:00
5d0713946b
add some more assertions. Remove code to handle dllimport on darwin.
...
llvm-svn: 75082
2009-07-09 00:47:59 +00:00
6b04136756
Make SelectionDAG::getVectorShuffle work properly for VECTOR_SHUFFLE
...
nodes with operand types that differ from the result type. (This
doesn't normally happen right now, because
SelectionDAGLowering::visitShuffleVector normalizes vector shuffles.)
llvm-svn: 75081
2009-07-09 00:46:33 +00:00
031c5fec43
remove "asmcall", using print_pcrel_imm instead of printOperand.
...
llvm-svn: 75080
2009-07-09 00:39:19 +00:00
f0c21619f8
simplify some logic
...
llvm-svn: 75079
2009-07-09 00:32:12 +00:00
ade55bc8dd
* add some assertions for sanity checking.
...
* remove some old code that was needed when we'd put ESP in the scale instead of
the base of some instructions.
* Fix a bug with the P modifier in inline asm that caused us to drop it.
llvm-svn: 75077
2009-07-09 00:27:29 +00:00
785c56ce1d
LLVMContext-ification.
...
llvm-svn: 75072
2009-07-08 23:50:31 +00:00
22c2fba978
Use common code for both ARM and Thumb-2 instruction and register info.
...
llvm-svn: 75067
2009-07-08 23:10:31 +00:00
3c043404e0
* manually concatenate two string literals
...
* remove some dead code: darwin doesn't support dllimport linkage.
llvm-svn: 75066
2009-07-08 23:09:14 +00:00
5edd90cbbc
- Add some NEON ld / st instruction static encoding.
...
- Make bits 25-27 for ldrh, etc. explicitly zero. Previously only the JIT uses the encoding information and it's assuming anything not specified to be zero. Making them explicit so the disassembler is happy.
Patch by Sean Callanan.
llvm-svn: 75065
2009-07-08 22:51:32 +00:00
daf15df944
Fix thinko.
...
llvm-svn: 75061
2009-07-08 22:25:56 +00:00
d8d725de4f
Fix the AssertingVH unittests.
...
llvm-svn: 75060
2009-07-08 22:09:00 +00:00
70415d97a8
Add an option to allocate JITed global data separately from code. By
...
default, this option is not enabled to support clients who rely on
this behavior.
Fixes http://llvm.org/PR4483
A patch to allocate additional memory for globals after we run out is
forthcoming.
Patch by Reid Kleckner!
llvm-svn: 75059
2009-07-08 21:59:57 +00:00
f04d63a398
Drop "constant" from
...
!0 = constant metadata !{...}
llvm-svn: 75057
2009-07-08 21:57:07 +00:00
9dadbbd755
Allow users of GraphWriter to display graphs asynchronously. This
...
provides a way to quickly dump a bunch of graph information to dot files
and display them. It's a timesaver when working on large systems.
llvm-svn: 75056
2009-07-08 21:53:41 +00:00
cdde1bb0a6
Add an ExecuteNoWait interface to support asynchronous process spawning.
...
llvm-svn: 75055
2009-07-08 21:46:40 +00:00
983c6b1b8f
Update SLotTracker to handle MDNode slots.
...
Simplify MDNode printing.
llvm-svn: 75053
2009-07-08 21:44:25 +00:00
7dcc37b942
Nowadays vectors are only split if they have an even
...
number of elements. Make some simplifications based
on this (in particular SplitVecRes_SETCC). Tighten
up some checking while there.
llvm-svn: 75050
2009-07-08 21:34:03 +00:00
e3a53c448b
Change how so_imm and t2_so_imm are handled. At instruction selection time, the immediates are no longer encoded in the imm8 + rot format, that are left as it is. The encoding is now done in ams printing and code emission time instead.
...
llvm-svn: 75048
2009-07-08 21:03:57 +00:00
0c9f367d62
Recommit r74952 with a bug fix:
...
DWARF requires frame moves be specified at specific times. If you have a
prologue like this:
__Z3fooi:
Leh_func_begin1:
LBB1_0: ## entry
pushl %ebp
Llabel1:
movl %esp, %ebp
Llabel2:
pushl %esi
Llabel3:
subl $20, %esp
call "L1$pb"
"L1$pb":
popl %esi
The "pushl %ebp" needs a table entry specifying the offset. The "movl %esp,
%ebp" makes %ebp the new stack frame register, so that needs to be specified in
DWARF. And "pushl %esi" saves the callee-saved %esi register, which also needs
to be specified in DWARF.
Before, all of this logic was in one method. This didn't work too well, because
as you can see there are multiple FDE line entries that need to be created.
This fix creates the "MachineMove" objects directly when they're needed; instead
of waiting until the end, and losing information.
There is some ugliness where we generate code like this:
LBB22_0: ## entry
pushl %ebp
Llabel280:
movl %esp, %ebp
Llabel281:
Llabel284:
pushl %ebp <----------
pushl %ebx
pushl %edi
pushl %esi
Llabel282:
subl $328, %esp
Notice the extra "pushl %ebp". If we generate a "machine move" instruction in
the FDE for that pushl, the linker may get very confused about what value %ebp
should have when exitting the function. I.e., it'll give it the value %esp
instead of the %ebp value from the first "pushl". Not to mention that, in this
case, %ebp isn't modified in the function (that's a separate bug). I put a small
hack in to get it to work. It might be the only solution, but should be
revisited once the above case is fixed.
llvm-svn: 75047
2009-07-08 21:02:53 +00:00
9242ff34c6
Use interators instead of counters for loops.
...
llvm-svn: 75046
2009-07-08 20:57:27 +00:00
f8d479c1ce
Missed an exit during the conversion.
...
Will convert assert(0) that don't have abort() to LLVM_UNREACHABLE in a later
commit.
llvm-svn: 75045
2009-07-08 20:55:50 +00:00
fb8d6d5b58
Implement changes from Chris's feedback.
...
Finish converting lib/Target.
llvm-svn: 75043
2009-07-08 20:53:28 +00:00
a771459bb1
Push LLVMContext _back_ through IRBuilder.
...
llvm-svn: 75040
2009-07-08 20:50:47 +00:00
1d298fd75b
Implement NEON vst1 instruction.
...
llvm-svn: 75037
2009-07-08 20:32:02 +00:00
03ab0bbb24
Generalize opcode selection in ARMBaseRegisterInfo.
...
llvm-svn: 75036
2009-07-08 20:28:28 +00:00
b009980f0b
Fix cmake build.
...
Added ARMBaseRegisterInfo.cpp to lib/Target/ARM/CMakeLists.txt
llvm-svn: 75035
2009-07-08 20:13:41 +00:00
d2541158da
Support MDNode forward reference.
...
llvm-svn: 75031
2009-07-08 19:23:54 +00:00
Dan Gohman