Justin Bogner [Thu, 30 Apr 2015 22:58:28 +0000 (22:58 +0000)]
InstrProf: Fix handling of profile counters in the body of range based for
We were assigning the counter for the body of the loop to the loop
variable initialization for some reason here, but our tests completely
lacked coverage for range-for loops. This fixes that and makes the
logic generally more similar to the logic for a regular for.
llvm-svn: 236277
Reid Kleckner [Thu, 30 Apr 2015 22:42:45 +0000 (22:42 +0000)]
Re-commit the test fix that went with r236274
llvm-svn: 236276
Rafael Espindola [Thu, 30 Apr 2015 22:41:12 +0000 (22:41 +0000)]
Remove unnecessary break.
llvm-svn: 236275
Reid Kleckner [Thu, 30 Apr 2015 22:29:25 +0000 (22:29 +0000)]
Revert most of r236271, leaving only the datalayout change in lib/Basic/Targets.cpp
llvm-svn: 236274
Quentin Colombet [Thu, 30 Apr 2015 22:27:20 +0000 (22:27 +0000)]
[AArch64] Fix bad register class constraint in fast-isel for TST instruction.
rdar://problem/
20748715
llvm-svn: 236273
Pete Cooper [Thu, 30 Apr 2015 22:15:59 +0000 (22:15 +0000)]
Don't always apply kill flag in thumb2 ABS pseudo expansion.
The expansion for t2ABS was always setting the kill flag on the rsb instruction.
It should instead only be set on rsb if it was set on the original ABS instruction.
rdar://problem/
20752113
llvm-svn: 236272
Reid Kleckner [Thu, 30 Apr 2015 22:13:05 +0000 (22:13 +0000)]
Use 4 byte preferred aggregate alignment in datalayout on x86 Win32
llvm-svn: 236271
Reid Kleckner [Thu, 30 Apr 2015 22:11:59 +0000 (22:11 +0000)]
[X86] Use 4 byte preferred aggregate alignment on Win32
This helps reduce the frequency of stack realignment prologues in 32-bit
X86 Windows code. Before this change and the corresponding clang change,
we would take the max of the type preferred alignment and the explicit
alignment on the alloca.
If you don't override aggregate alignment in datalayout, you get a
default of 8. This dates back to 2007 / r34356, and changing it seems
prohibitively difficult at this point.
llvm-svn: 236270
Matthias Braun [Thu, 30 Apr 2015 22:05:30 +0000 (22:05 +0000)]
InstCombineSimplifyDemanded: Remove nsw/nuw flags when optimizing demanded bits
When optimizing demanded bits of the operands of an Add we have to
remove the nsw/nuw flags as we have no guarantee anymore that we don't
wrap. This is legal here because the top bit is not demanded. In fact
this operaion was already performed but missed in the case of an Add
with a constant on the right side. To fix this this patch refactors the
code to unify the code paths in SimplifyDemandedUseBits() handling of
Add/Sub:
- The transformation of Add->Or is removed from the simplify demand
code because the equivalent transformation exists in
InstCombiner::visitAdd()
- KnownOnes/KnownZero are not adjusted for Add x, C anymore as
computeKnownBits() already performs these computations.
- The simplification of the operands is unified. In this new version
constant on the right side of a Sub are shrunk now as I could not find
a reason why not to do so.
- The special case for clearing nsw/nuw in ShrinkDemandedConstant() is
not necessary anymore as the caller does that already.
Differential Revision: http://reviews.llvm.org/D9415
llvm-svn: 236269
Matthias Braun [Thu, 30 Apr 2015 22:04:26 +0000 (22:04 +0000)]
InstCombine: Move Sub->Xor rule from SimplifyDemanded to InstCombine
The rule that turns a sub to xor if the LHS is 2^n-1 and the remaining bits
are known zero, does not use the demanded bits at all: Move it to the
normal InstCombine code path.
Differential Revision: http://reviews.llvm.org/D9417
llvm-svn: 236268
Rafael Espindola [Thu, 30 Apr 2015 21:51:58 +0000 (21:51 +0000)]
Simplify the creation of compressed debug sections.
This is actually fairly simple in the current code layout: Check if we should
compress just before writing out and everything else just works.
This removes the last case in which the object writer was creating a
fragment.
llvm-svn: 236267
Sean Callanan [Thu, 30 Apr 2015 21:49:58 +0000 (21:49 +0000)]
Made macros from modules be injected before our
global convenience expression prefix. Also ensured
that if macros are defined by the modules we don't
try to redefine them. Finally cleaned up a bit of
code while I was in there.
<rdar://problem/
20756642>
llvm-svn: 236266
Richard Trieu [Thu, 30 Apr 2015 21:47:28 +0000 (21:47 +0000)]
Fix -Wpessimizing-move warning by remove the call to std::move.
llvm-svn: 236265
Justin Bogner [Thu, 30 Apr 2015 21:31:02 +0000 (21:31 +0000)]
InstrProf: Make sure coverage propagates out of foreach loops correctly
llvm-svn: 236264
Andrea Di Biagio [Thu, 30 Apr 2015 21:22:28 +0000 (21:22 +0000)]
Fix comment in test. NFC.
llvm-svn: 236262
Rafael Espindola [Thu, 30 Apr 2015 21:20:06 +0000 (21:20 +0000)]
Replace unreachable code with llvm_unreachable.
llvm-svn: 236261
Rafael Espindola [Thu, 30 Apr 2015 21:10:06 +0000 (21:10 +0000)]
Don't create a MCSectionData when we don't have to.
llvm-svn: 236260
Zachary Turner [Thu, 30 Apr 2015 21:03:37 +0000 (21:03 +0000)]
Remove the NullLog class introduced in r236174.
Based on list discussions, a different approach is desired for
reducing the visual impact of logging statements on the
readability of the code. Another mechanism will be added in
a followup patch, but for now, since NullLog is unreferenced,
this patch just removes it.
This patch does *not* remove the other half of r236174, which was
to delete some dead code surrounding logging flags.
llvm-svn: 236259
Andrea Di Biagio [Thu, 30 Apr 2015 21:03:29 +0000 (21:03 +0000)]
Fix for PR23103. Correctly propagate the 'IsUndef' flag to the register operands of a commuted instruction.
Revision 220239 exposed a latent bug in method
'TargetInstrInfo::commuteInstruction'. When commuting the operands of a machine
instruction, method 'commuteInstruction' didn't correctly propagate the
'IsUndef' flag to the register operands of the new (commuted) instruction.
Before this patch, the following instruction:
%vreg4<def> = VADDSDrr %vreg14, %vreg5<undef>; FR64:%vreg4,%vreg14,%vreg5
was wrongly converted by method 'commuteInstruction' into:
%vreg4<def> = VADDSDrr %vreg5, %vreg14<undef>; FR64:%vreg4,%vreg5,%vreg14
The correct instruction should have been:
%vreg4<def> = VADDSDrr %vreg5<undef>, %vreg14; FR64:%vreg4,%vreg5,%vreg14
This patch fixes the problem in method 'TargetInstrInfo::commuteInstruction'.
When swapping the operands of a machine instruction, we now make sure that
'IsUndef' flags are correctly set.
Added test case 'pr23103.ll'.
Differential Revision: http://reviews.llvm.org/D9406
llvm-svn: 236258
Rafael Espindola [Thu, 30 Apr 2015 20:57:14 +0000 (20:57 +0000)]
Avoid an extra loop over the sections.
Add string to the section header string table as we add sections.
llvm-svn: 236257
Richard Smith [Thu, 30 Apr 2015 20:53:28 +0000 (20:53 +0000)]
PR19691: initialize MacroDirective::DefInfo::IsPublic member in "invalid" state.
No functionality change; no-one inspects this field yet, and probably no-one
will ever inspect it in the "invalid" state, but ubsan could be unhappy about
this if such a DefInfo is copied.
llvm-svn: 236256
Rafael Espindola [Thu, 30 Apr 2015 20:53:27 +0000 (20:53 +0000)]
Make the section table a member of ELFObjectWriter.
This avoids passing it around and lets us build a small helper to add
a section to the table.
llvm-svn: 236255
Richard Smith [Thu, 30 Apr 2015 20:51:03 +0000 (20:51 +0000)]
PR22354: Add a test that we emit copy-ctors for captures in generic lambdas.
The bug was fixed by r235921.
llvm-svn: 236254
Rafael Espindola [Thu, 30 Apr 2015 20:37:36 +0000 (20:37 +0000)]
Nothing inherits from this, drop the protected:
llvm-svn: 236253
Kevin Enderby [Thu, 30 Apr 2015 20:30:42 +0000 (20:30 +0000)]
For llvm-objdump, with the -archive-headers and -macho options, use the -non-verbose
option to print the archive headers using raw numeric values. Also add the -archive-member-offsets
for use with these to also trigger printing of the offset of the archive member from the start
of the archive.
llvm-svn: 236252
Douglas Katzman [Thu, 30 Apr 2015 20:08:44 +0000 (20:08 +0000)]
Exhaustively test all triples in TripleTest.
Iteration over all permutations didn't really work,
due to evolution of the underlying enums.
Differential Revision: http://reviews.llvm.org/D9057
llvm-svn: 236251
Matt Arsenault [Thu, 30 Apr 2015 19:35:43 +0000 (19:35 +0000)]
Mips: Remove dead declaration
llvm-svn: 236250
Matt Arsenault [Thu, 30 Apr 2015 19:35:41 +0000 (19:35 +0000)]
MachineVerifier: Don't crash if MachineOperand has no parent
If you somehow added a MachineOperand to an instruction
that did not have the parent set, the verifier would
crash since it attempts to use the operand's parent.
llvm-svn: 236249
Pete Cooper [Thu, 30 Apr 2015 18:58:23 +0000 (18:58 +0000)]
Don't rewrite jumps to empty BBs to landing pads.
In the test case here, the 'unreachable' BB was removed by BranchFolding because its empty.
It then rewrote the jump from 'entry' to jump to its fallthrough, which was a landing pad.
This results in 'entry' jumping to 2 different landing pads, which fails the machine verifier.
rdar://problem/
20750162
llvm-svn: 236248
Quentin Colombet [Thu, 30 Apr 2015 18:52:49 +0000 (18:52 +0000)]
[ARM] Do not generate invalid encoding for stack adjust, even if this is just
temporary.
Because of that:
1. The machine verifier was complaining on such code.
2. The generate code worked just because the thumb reduction size pass fixed the
opcode.
rdar://problem/
20749824
llvm-svn: 236247
Ben Langmuir [Thu, 30 Apr 2015 18:40:23 +0000 (18:40 +0000)]
Fix the end location of init-capture annotations in ObjC++
And thereby stop asserting.
In ObjC++ modes, we tentatively parse the lambda introducer twice: once
to disambiguate designators, which we also do in C++, and a second time
to disambiguate objc message expressions. During the second tentative
parse, the last cached token will be the annotation token we built in
the first parse. So use getLastLoc() to get the correct end location
for the rebuilt annotation.
llvm-svn: 236246
Tim Northover [Thu, 30 Apr 2015 18:28:58 +0000 (18:28 +0000)]
AArch64: add BFC alias for the BFI/BFM instructions.
Unlike 32-bit ARM, AArch64 can use wzr/xzr to implement this without the need
for a separate instruction.
rdar://
18679590
llvm-svn: 236245
Reid Kleckner [Thu, 30 Apr 2015 18:17:12 +0000 (18:17 +0000)]
Add a note about permitting default member initializers
Use them in WinEHPrepare so that we can spot any toolchain bugs that
come up.
llvm-svn: 236244
David Majnemer [Thu, 30 Apr 2015 18:16:38 +0000 (18:16 +0000)]
Remove some leftover debugging code.
llvm-svn: 236243
Jordan Rose [Thu, 30 Apr 2015 17:20:35 +0000 (17:20 +0000)]
Don't warn about use of deprecated API when the containing context is unavailable.
This probably won't come up much, but it seems tidier.
Related to rdar://problem/
20713550
llvm-svn: 236242
Jordan Rose [Thu, 30 Apr 2015 17:20:30 +0000 (17:20 +0000)]
Batch up access-related diagnostics on enum constants until the whole enum is parsed.
That way we can take any trailing availability attributes into account.
rdar://problem/
20713550
llvm-svn: 236241
Jan Vesely [Thu, 30 Apr 2015 17:15:56 +0000 (17:15 +0000)]
Reinstate revisions r234755, r234759, r234760
changes:
Don't apply on hexagon and NVPTX since they no longer claim to support UADDO/USUBO
Add location to getConstant
Drop comment about the ops being turned into expand
llvm-svn: 236240
David Majnemer [Thu, 30 Apr 2015 17:15:48 +0000 (17:15 +0000)]
[MS ABI] Correctly make paths through covariant virtual bases
There can be multiple virtual bases which are on the path to a vfptr
when one vbase virtually inherits from another. We should prefer the
most derived virtual base which covariantly overrides a method in the
vfptr class; if we do not lengthen the path this way, we will end up
with too few vftable entries.
This fixes PR21073.
llvm-svn: 236239
David Majnemer [Thu, 30 Apr 2015 17:15:46 +0000 (17:15 +0000)]
[AST] Fix some typos
llvm-svn: 236238
Daniel Berlin [Thu, 30 Apr 2015 16:15:07 +0000 (16:15 +0000)]
Move equality function for AliasAnalysis::Location from DenseMapInfo to Location struct so it can be used in other types of maps
llvm-svn: 236237
Marshall Clow [Thu, 30 Apr 2015 15:19:44 +0000 (15:19 +0000)]
Mark LWG#2387 as complete. No code changes needed
llvm-svn: 236236
Rafael Espindola [Thu, 30 Apr 2015 14:21:49 +0000 (14:21 +0000)]
Write sections mostly in one pass.
During ELF writing, there is no need to further relax the sections, so we
should not be creating fragments. This patch avoids doing so in all cases
but debug section compression (that is next).
Also, the ELF format is fairly simple to write. We can do a single pass over
the sections to write them out and compute the section header table.
llvm-svn: 236235
Aaron Ballman [Thu, 30 Apr 2015 14:03:12 +0000 (14:03 +0000)]
Silencing an "enumeral and non-enumeral type in conditional expression" warning; NFC.
llvm-svn: 236234
Rafael Espindola [Thu, 30 Apr 2015 13:57:06 +0000 (13:57 +0000)]
Don't check for offsets in tests where it is not relevant.
llvm-svn: 236233
Diego Novillo [Thu, 30 Apr 2015 13:22:48 +0000 (13:22 +0000)]
Fix private constructor for ScaledNumber.
Summary:
The private constructor for ScaledNumber was using uint64_t instead of
DigitsT. This was preventing instantiations of ScaledNumber with
anything other than uint64_t types.
In implementing the tests, I ran into another issue. Operators >>= and
<<= did not have variants for accepting other ScaledNumber as the shift
argument. This is expected by the SCALED_NUMBER_BOP.
It makes no sense to allow shifting a ScaledNumber by another
ScaledNumber, so the patch includes two new templates for shifting
ScaledNumbers.
Reviewers: dexonsmith
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D9350
llvm-svn: 236232
Ilia K [Thu, 30 Apr 2015 13:10:32 +0000 (13:10 +0000)]
Fix Process::ResumeSynchronous which waits a process even in case of error
llvm-svn: 236231
Rafael Espindola [Thu, 30 Apr 2015 13:08:09 +0000 (13:08 +0000)]
Check the entire content of the comdat group.
llvm-svn: 236230
Ilia K [Thu, 30 Apr 2015 12:37:05 +0000 (12:37 +0000)]
Fix MiLibraryLoadedTestCase.test_lldbmi_library_loaded test on Linux (MI)
llvm-svn: 236229
Ilia K [Thu, 30 Apr 2015 12:28:26 +0000 (12:28 +0000)]
Use \d+ pattern instead of hard-coded line number in MiBreakTestCase (MI)
llvm-svn: 236228
Diego Novillo [Thu, 30 Apr 2015 12:27:51 +0000 (12:27 +0000)]
Fix typo in comment. NFC.
llvm-svn: 236227
Pavel Labath [Thu, 30 Apr 2015 11:28:06 +0000 (11:28 +0000)]
Enabling two watchpoint tests on linux
After recent changes, these tests should be stable. Please let me know if you still see failures.
llvm-svn: 236226
Ilia K [Thu, 30 Apr 2015 11:08:31 +0000 (11:08 +0000)]
Rework =shlibs-added/=shlibs-removed
Summary:
This patch includes the following:
* Rename =shlibs-added/=shlibs-removed to standard =library-loaded/=library-unloaded
* Remove redundant fields
* Add extra symbols-loaded/symbols-path and loaded_addr fields
* Rename CMICmnMIOutOfBandRecord::eOutOfBand_TargetModulesLoaded/eOutOfBand_TargetModulesUnloaded to CMICmnMIOutOfBandRecord::eOutOfBand_TargetModuleLoaded/eOutOfBand_TargetModuleUnloaded
Test Plan: ./dotest.py -v --executable $BUILDDIR/bin/lldb tools/lldb-mi/
Reviewers: jasonmolenda, jingham, abidh
Reviewed By: abidh
Subscribers: lldb-commits, jingham, jasonmolenda, abidh
Differential Revision: http://reviews.llvm.org/D9280
llvm-svn: 236225
Pavel Labath [Thu, 30 Apr 2015 11:00:15 +0000 (11:00 +0000)]
Add expectedFailureLinux to TestProcessAttach
The test is skipped anyway due to timeout, but adding XFAIL improves grepability.
llvm-svn: 236224
Daniel Sanders [Thu, 30 Apr 2015 10:52:42 +0000 (10:52 +0000)]
[mips] Sorted instructions in mips64r6 disassembly tests. NFC.
llvm-svn: 236223
Pavel Labath [Thu, 30 Apr 2015 10:47:56 +0000 (10:47 +0000)]
Fix typo in Core/Log
llvm-svn: 236222
Daniel Sanders [Thu, 30 Apr 2015 10:23:48 +0000 (10:23 +0000)]
[mips][mips64r6] Sorted instructions in test. NFC.
llvm-svn: 236221
Daniel Sanders [Thu, 30 Apr 2015 09:57:37 +0000 (09:57 +0000)]
[mips][msa] Rename main check prefix to 'ALL' in basic operations tests. NFC
Summary:
The majority of the checks are subtarget independent. The few that aren't
will be corrected shortly.
Reviewers: vkalintiris
Reviewed By: vkalintiris
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D9340
llvm-svn: 236220
Daniel Sanders [Thu, 30 Apr 2015 09:56:30 +0000 (09:56 +0000)]
[mips][msa] Use CHECK-LABEL where missing, and remove checks matching the .size directive. NFC.
Summary:
Reviewers: vkalintiris
Reviewed By: vkalintiris
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D9339
llvm-svn: 236219
Elena Demikhovsky [Thu, 30 Apr 2015 09:24:29 +0000 (09:24 +0000)]
AVX-512: Added AVX-512 intrinsics and tests
by Asaf Badouh (asaf.badouh@intel.com)
llvm-svn: 236218
Daniel Jasper [Thu, 30 Apr 2015 09:24:17 +0000 (09:24 +0000)]
clang-format: Don't merge short else blocks.
Before (with the appropriate flags settings):
if (a) {
f();
} else { g(); }
Before (with other appropriate flags settings):
if (a) { f(); } else {
g();
}
After:
if (a) {
f();
} else {
g();
}
llvm-svn: 236217
Daniel Sanders [Thu, 30 Apr 2015 09:24:09 +0000 (09:24 +0000)]
[mips] Add missing signext attributes to MSA basic operations tests. NFC.
Summary:
This doesn't make much difference to MIPS32, but it will simplify a
MIPS64r6 bugfix which will follow shortly by removing unnecessary
sign-extension of parameters.
Reviewers: vkalintiris
Reviewed By: vkalintiris
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D9338
llvm-svn: 236216
Pawel Bylica [Thu, 30 Apr 2015 09:01:22 +0000 (09:01 +0000)]
Add convenient overloads for CreateInsertElement and CreateExtractElement methods in IRBuilder
Summary:
This pathc add convenient overloads for CreateInsertElement and CreateExtractElement methods in IRBuilder
where vector index can be uint64_t instead of Value*.
Test Plan: Unit test included.
Reviewers: majnemer
Reviewed By: majnemer
Subscribers: majnemer, llvm-commits
Differential Revision: http://reviews.llvm.org/D9347
llvm-svn: 236214
Daniel Jasper [Thu, 30 Apr 2015 09:01:21 +0000 (09:01 +0000)]
Silence unused warning in non-assert builds.
llvm-svn: 236213
Daniel Jasper [Thu, 30 Apr 2015 08:51:13 +0000 (08:51 +0000)]
Inline local variable to silence unused warning.
llvm-svn: 236212
Elena Demikhovsky [Thu, 30 Apr 2015 08:38:48 +0000 (08:38 +0000)]
Masked gather and scatter - added DAGCombine visitors
and AVX-512 instruction selection patterns.
All other patches, including tests will follow.
http://reviews.llvm.org/D7665
llvm-svn: 236211
Alexey Bataev [Thu, 30 Apr 2015 08:35:10 +0000 (08:35 +0000)]
[OPENMP] Mark test as not compatible with MS Windows.
llvm-svn: 236210
Simon Pilgrim [Thu, 30 Apr 2015 08:23:16 +0000 (08:23 +0000)]
[SSE] Fix for MUL v16i8 on pre-SSE41 targets (PR23369).
Sign extension of i8 to i16 was placing the unpacked bytes in the lower byte instead of the upper byte.
llvm-svn: 236209
Ilia K [Thu, 30 Apr 2015 07:14:24 +0000 (07:14 +0000)]
Add -data-info-line command (MI)
Summary: Add -data-info-line command + test
Test Plan: ./dotest.py -v --executable $BUILDDIR/bin/lldb tools/lldb-mi/
Reviewers: abidh
Reviewed By: abidh
Subscribers: lldb-commits, abidh
Differential Revision: http://reviews.llvm.org/D9276
llvm-svn: 236208
Alexey Bataev [Thu, 30 Apr 2015 06:51:57 +0000 (06:51 +0000)]
[OPENMP] Codegen for 'private' clause in 'task' directive.
For tasks codegen for private/firstprivate variables are different rather than for other directives.
1. Build an internal structure of privates for each private variable:
struct .kmp_privates_t. {
Ty1 var1;
...
Tyn varn;
};
2. Add a new field to kmp_task_t type with list of privates.
struct kmp_task_t {
void * shareds;
kmp_routine_entry_t routine;
kmp_int32 part_id;
kmp_routine_entry_t destructors;
.kmp_privates_t. privates;
};
3. Create a function with destructors calls for all privates after end of task region.
kmp_int32 .omp_task_destructor.(kmp_int32 gtid, kmp_task_t *tt) {
~Destructor(&tt->privates.var1);
...
~Destructor(&tt->privates.varn);
return 0;
}
4. Perform default initialization of all private fields (no initialization for POD data, default constructor calls for classes) + provide address of a destructor function after kmpc_omp_task_alloc() and before kmpc_omp_task() calls.
kmp_task_t *new_task = __kmpc_omp_task_alloc(ident_t *, kmp_int32 gtid, kmp_int32 flags, size_t sizeof_kmp_task_t, size_t sizeof_shareds, kmp_routine_entry_t *task_entry);
DefaultConstructor(new_task->privates.var1);
new_task->shareds.var1_ref = &new_task->privates.var1;
...
DefaultConstructor(new_task->privates.varn);
new_task->shareds.varn_ref = &new_task->privates.varn;
new_task->destructors = .omp_task_destructor.;
kmp_int32 __kmpc_omp_task(ident_t *, kmp_int32 gtid, kmp_task_t *new_task)
Differential Revision: http://reviews.llvm.org/D9322
llvm-svn: 236207
Craig Topper [Thu, 30 Apr 2015 05:54:22 +0000 (05:54 +0000)]
[TableGen] Cleanup formatting by moving operators from beginning of line to end of previous line. NFC
llvm-svn: 236206
Craig Topper [Thu, 30 Apr 2015 05:54:20 +0000 (05:54 +0000)]
[TableGen] Used range-based for loop. NFC.
llvm-svn: 236205
Craig Topper [Thu, 30 Apr 2015 05:12:52 +0000 (05:12 +0000)]
[TableGen] Merge a variable assignment and a return to drop curly braces. Fold an assignment into an if. Use auto on the result of a couple dyn_casts. NFC
llvm-svn: 236204
Sanjoy Das [Thu, 30 Apr 2015 04:56:04 +0000 (04:56 +0000)]
[InstCombine] Add new rule for MIN(MAX(~A, ~B), ~C) et. al.
Summary:
Optimizing these well are especially interesting for IRCE since it
"clamps" values by generating this sort of pattern through SCEV
expressions.
Depends on D9352.
Reviewers: majnemer
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D9353
llvm-svn: 236203
Sanjoy Das [Thu, 30 Apr 2015 04:56:00 +0000 (04:56 +0000)]
[InstCombine] Add a new formula for SMIN.
Summary:
After this change `MatchSelectPattern` recognizes the following form
of SMIN:
Y >s C ? ~Y : ~C == ~Y <s ~C ? ~Y : ~C = SMIN(~Y, ~C)
Reviewers: majnemer
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D9352
llvm-svn: 236202
Alexey Bataev [Thu, 30 Apr 2015 04:23:23 +0000 (04:23 +0000)]
[OPENMP] Allow to use global variables as lcv in loop-based directives.
For proper codegen we need to capture variable in the OpenMP region. In loop-based directives loop control variables are private by default and they must be captured in this region. There was a problem with capturing of globals, used as lcv, as they was not marked as private by default.
Differential Revision: http://reviews.llvm.org/D9336
llvm-svn: 236201
Filipe Cabecinhas [Thu, 30 Apr 2015 04:09:41 +0000 (04:09 +0000)]
Don't overflow GCTable
Summary: Bug found with AFL fuzz.
Reviewers: rafael, dexonsmith
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D9361
llvm-svn: 236200
Owen Anderson [Thu, 30 Apr 2015 04:06:32 +0000 (04:06 +0000)]
Semantically revert r236031, which is not a good idea for in-order targets.
At the least it should be guarded by some kind of target hook.
It also introduced catastrophic compile time and code quality
regressions on some out of tree targets (test case still being
reduced/sanitized).
Sanjay agreed with reverting this patch until these issues can be
resolved.
llvm-svn: 236199
Alexey Bataev [Thu, 30 Apr 2015 03:47:32 +0000 (03:47 +0000)]
[OPENMP] Fixed codegen for 'copyprivate' clause.
Fixed initialization of 'single' region completion + changed type of the third argument of __kmpc_copyprivate() runtime function to size_t.
llvm-svn: 236198
Richard Smith [Thu, 30 Apr 2015 02:16:23 +0000 (02:16 +0000)]
Remove dead code: a MacroDirective can't be imported or ambiguous any more.
llvm-svn: 236197
Hans Wennborg [Thu, 30 Apr 2015 01:59:04 +0000 (01:59 +0000)]
XFAIL test/CodeGen/Generic/MachineBranchProb.ll on Hexagon (PR23377)
llvm-svn: 236196
Alexey Samsonov [Thu, 30 Apr 2015 01:21:57 +0000 (01:21 +0000)]
[UBSan] Make stacktrace-matching CHECK-lines in tests Linux-specific.
Darwin doesn't yet allow to print stack trace, as it lacks the slow
unwinder. This is one more attempt to fix vptr.cpp on Mac OS X.
llvm-svn: 236195
Richard Smith [Thu, 30 Apr 2015 01:15:19 +0000 (01:15 +0000)]
Remove XFAIL now that this test passes (fixed by r236184).
llvm-svn: 236194
Filipe Cabecinhas [Thu, 30 Apr 2015 01:13:31 +0000 (01:13 +0000)]
Make sure Op->getType() is a PointerType before we cast<> it.
Bug found with AFL fuzz.
llvm-svn: 236193
Hans Wennborg [Thu, 30 Apr 2015 00:57:37 +0000 (00:57 +0000)]
Switch lowering: use profile info to build weight-balanced binary search trees
This will cause hot nodes to appear closer to the root.
The literature says building the tree like this makes it a near-optimal (in
terms of search time given key frequencies) binary search tree. In LLVM's case,
we can do up to 3 comparisons in each leaf node, so it might be better to opt
for lower tree height in some cases; that's something to look into in the
future.
Differential Revision: http://reviews.llvm.org/D9318
llvm-svn: 236192
Richard Smith [Thu, 30 Apr 2015 00:57:08 +0000 (00:57 +0000)]
Fix use of uninitialized variable, found by ubsan selfhost.
llvm-svn: 236191
Filipe Cabecinhas [Thu, 30 Apr 2015 00:52:42 +0000 (00:52 +0000)]
Make sure we don't resize(0) when we get a fwdref with Idx == UINT_MAX
Make it an error instead.
Bug found with AFL fuzz.
llvm-svn: 236190
Rafael Espindola [Thu, 30 Apr 2015 00:45:46 +0000 (00:45 +0000)]
Store relocations in a map from MCSectionELF.
Saves finding the MCSectionData just to do a map lookup.
llvm-svn: 236189
Sean Callanan [Thu, 30 Apr 2015 00:44:21 +0000 (00:44 +0000)]
Use a more reliable method to determine whether
a FileID corresponds to a real file or to a
memory buffer. The old method didn't work when
Clang was built Release, which meant it wasn't
a very good method at all.
llvm-svn: 236188
Rafael Espindola [Thu, 30 Apr 2015 00:30:40 +0000 (00:30 +0000)]
Write relocations directly to the output stream. NFC.
llvm-svn: 236187
Ahmed Bougacha [Thu, 30 Apr 2015 00:07:34 +0000 (00:07 +0000)]
Flip r236172 testcase RUN option ordering for BSD sed(1). NFC.
llvm-svn: 236186
Chaoren Lin [Wed, 29 Apr 2015 23:59:22 +0000 (23:59 +0000)]
Add CMAKE_EXECUTABLE_SUFFIX to build with Android toolchain on Windows.
Reviewers: vharron, zturner, flackr
Subscribers: tberghammer, lldb-commits
Differential Revision: http://reviews.llvm.org/D9177
llvm-svn: 236185
NAKAMURA Takumi [Wed, 29 Apr 2015 23:55:46 +0000 (23:55 +0000)]
Update clang/test/CodeGenCXX/copy-constructor-synthesis-2.cpp to pass for targeting i686.
r236155 missed the suffix in "@llvm.memcpy.p0i8.p0i8.i32".
llvm-svn: 236184
Pete Cooper [Wed, 29 Apr 2015 23:51:33 +0000 (23:51 +0000)]
Change x86 CMOVE_F to read it source, not write it.
This was breaking sqlite with the machine verifier because operand 0 was a def according to tablegen, but didn't have the 'isDef' flag set.
Looking at the ISA, its clear that this operand is a source as writing to st(0) is implicit. So move the operand to the correct place in the td file.
rdar://problem/
20751584
llvm-svn: 236183
Richard Smith [Wed, 29 Apr 2015 23:46:48 +0000 (23:46 +0000)]
Fix build after Clang API change in r236176.
llvm-svn: 236182
Richard Smith [Wed, 29 Apr 2015 23:40:48 +0000 (23:40 +0000)]
Add an assert to get information on buildbot failure.
llvm-svn: 236181
Jonathan Roelofs [Wed, 29 Apr 2015 23:33:32 +0000 (23:33 +0000)]
Fix doxygen comment typo. NFC
llvm-svn: 236180
Eric Christopher [Wed, 29 Apr 2015 23:32:17 +0000 (23:32 +0000)]
Propagate a terrible hack to the sparc target feature handling code
by erasing the soft-float target feature if the rest of the front
end added it because of defaults or the soft float option.
Add some testing for some of the targets that implement this hack.
llvm-svn: 236179
Richard Smith [Wed, 29 Apr 2015 23:26:13 +0000 (23:26 +0000)]
Fix unused variable warning.
llvm-svn: 236178
Zachary Turner [Wed, 29 Apr 2015 23:24:12 +0000 (23:24 +0000)]
Fix build broken by r236174.
Apparently va_list is literally a char* on Windows.
llvm-svn: 236177
Richard Smith [Wed, 29 Apr 2015 23:20:19 +0000 (23:20 +0000)]
[modules] Stop trying to fake up a linear MacroDirective history.
Modules builds fundamentally have a non-linear macro history. In the interest
of better source fidelity, represent the macro definition information
faithfully: we have a linear macro directive history within each module, and at
any point we have a unique "latest" local macro directive and a collection of
visible imported directives. This also removes the attendent complexity of
attempting to create a correct MacroDirective history (which we got wrong
in the general case).
No functionality change intended.
llvm-svn: 236176