Duncan P. N. Exon Smith [Tue, 10 Feb 2015 01:40:40 +0000 (01:40 +0000)]
Verifier: Check for valid tags in debug nodes
Check that specialized `DebugNode`s have valid `DW_TAG`s.
llvm-svn: 228649
Duncan P. N. Exon Smith [Tue, 10 Feb 2015 01:36:46 +0000 (01:36 +0000)]
IR: Add accessors to MDExpression
Add some accessors to `MDExpression`.
llvm-svn: 228648
Duncan P. N. Exon Smith [Tue, 10 Feb 2015 01:32:56 +0000 (01:32 +0000)]
Verifier: Add simple checks for MDLocation
llvm-svn: 228647
Rui Ueyama [Tue, 10 Feb 2015 01:10:23 +0000 (01:10 +0000)]
Driver: Replace switch cases with ifs.
We used to do like this instead of putting all command line processing
code within one gigantic switch statement. It is converted to a switch
in r188958, which introduced InputGraph.
In this patch I roll that change back. Now all "break"s are removed,
and the nesting is one level shallow.
llvm-svn: 228646
Duncan P. N. Exon Smith [Tue, 10 Feb 2015 01:09:50 +0000 (01:09 +0000)]
Verifier: Create stubs for specialized metadata nodes
llvm-svn: 228645
Duncan P. N. Exon Smith [Tue, 10 Feb 2015 01:08:16 +0000 (01:08 +0000)]
AsmParser: Add stubs for specialized MDNodes, NFC
Well, the exact error from the failed parse will change, but...
llvm-svn: 228644
Rui Ueyama [Tue, 10 Feb 2015 01:06:14 +0000 (01:06 +0000)]
Remove trailing space.
llvm-svn: 228643
David Majnemer [Tue, 10 Feb 2015 00:57:42 +0000 (00:57 +0000)]
X86: Emit an ABI compliant prologue and epilogue for Win64
Win64 has specific contraints on what valid prologues and epilogues look
like. This constraint is born from the flexibility and descriptiveness
of Win64's unwind opcodes.
Prologues previously emitted by LLVM could not be represented by the
unwind opcodes, preventing operations powered by stack unwinding to
successfully work.
Differential Revision: http://reviews.llvm.org/D7520
llvm-svn: 228641
Duncan P. N. Exon Smith [Tue, 10 Feb 2015 00:52:32 +0000 (00:52 +0000)]
IR: Add specialized debug info metadata nodes
Add specialized debug info metadata nodes that match the `DIDescriptor`
wrappers (used by `DIBuilder`) closely. Assembly and bitcode support to
follow soon (it'll mostly just be obvious), but this sketches in today's
schema. This is the first big commit (well, the only *big* one aside
from the testcase changes that'll come when I move this into place) for
PR22464.
I've marked a bunch of obvious changes as `TODO`s in the source; I plan
to make those changes promptly after this hierarchy is moved underneath
`DIDescriptor`, but for now I'm aiming mostly to match the status quo.
llvm-svn: 228640
Chaoren Lin [Tue, 10 Feb 2015 00:47:43 +0000 (00:47 +0000)]
Add JSON.cpp to CMakeLists.txt
llvm-svn: 228639
Eric Christopher [Tue, 10 Feb 2015 00:44:17 +0000 (00:44 +0000)]
Migrate PPCAsmPrinter's subtarget from reference to pointer in
preparation for making it MachineFunction dependent.
llvm-svn: 228638
Lang Hames [Tue, 10 Feb 2015 00:37:26 +0000 (00:37 +0000)]
[Orc] Back out one of the GCC ICE workarounds from r228568. NFC.
llvm-svn: 228637
Enrico Granata [Tue, 10 Feb 2015 00:30:07 +0000 (00:30 +0000)]
Add a JSON producer to LLDB - this is a set of classes that encapsulate JSON objects and allow you to write them to a Stream for subsequent processing
Using this JSON producer, write a little tool that expands its own command-line arguments and dumps them to stdout as a JSON array
llvm-svn: 228636
David Blaikie [Tue, 10 Feb 2015 00:16:36 +0000 (00:16 +0000)]
Fix the clang -Werror build (-Wunused-variable)
llvm-svn: 228635
Philip Reames [Tue, 10 Feb 2015 00:04:53 +0000 (00:04 +0000)]
Adjust how we avoid poll insertion inside the poll function (NFC)
I realized that my early fix for this was overly complicated. Rather than scatter checks around in a bunch of places, just exit early when we visit the poll function itself.
Thinking about it a bit, the whole inlining mechanism used with gc.safepoint_poll could probably be cleaned up a bit. Originally, poll insertion was fused with gc relocation rewriting. It might be worth going back to see if we can simplify the chain of events now that these two are seperated. As one thought, maybe it makes sense to rewrite calls inside the helper function before inlining it to the many callers. This would require us to visit the poll function before any other functions though..
llvm-svn: 228634
Aaron Ballman [Tue, 10 Feb 2015 00:00:54 +0000 (00:00 +0000)]
Reverting r228628; it broke at least one builder due to the forward declare of RaiseException.
llvm-svn: 228633
Adrian Prantl [Mon, 9 Feb 2015 23:57:22 +0000 (23:57 +0000)]
Debug info: When updating debug info during SROA, do not emit debug info
for any padding introduced by SROA. In particular, do not emit debug info
for an alloca that represents only the padding introduced by a previous
iteration.
Fixes PR22495.
llvm-svn: 228632
Adrian Prantl [Mon, 9 Feb 2015 23:57:15 +0000 (23:57 +0000)]
Debug info: Use DW_OP_bit_piece instead of DW_OP_piece in the
intermediate representation. This
- increases consistency by using the same granularity everywhere
- allows for pieces < 1 byte
- DW_OP_piece didn't actually allow storing an offset.
Part of PR22495.
llvm-svn: 228631
Colin LeMahieu [Mon, 9 Feb 2015 23:45:24 +0000 (23:45 +0000)]
[Hexagon] Adding missing load instructions and removing an unused multiclass parameter.
llvm-svn: 228630
Duncan P. N. Exon Smith [Mon, 9 Feb 2015 23:21:05 +0000 (23:21 +0000)]
ADT: Allow up to 18 arguments in hash_combine()
I just realized that the specialized metadata node patch I'm about to
commit won't compile on old compilers. Bump `hash_combine()`'s support
for non-variadic templates to 18 (I tested this by reversing the logic
in the #ifdef).
llvm-svn: 228629
Aaron Ballman [Mon, 9 Feb 2015 23:11:39 +0000 (23:11 +0000)]
On Windows, we now use RaiseException to generate the kind of trap we require (one which calls our vectored exception handler), and fall back to using a volatile write to simulate a trap elsewhere.
llvm-svn: 228628
Colin LeMahieu [Mon, 9 Feb 2015 23:05:44 +0000 (23:05 +0000)]
[Hexagon] Factoring classes out of some load patterns and deleting some unused ones.
llvm-svn: 228627
Ramkumar Ramachandra [Mon, 9 Feb 2015 23:02:10 +0000 (23:02 +0000)]
[Statepoint] Improve two asserts, fix some style (NFC)
Summary:
It's important that our users immediately know what gc.safepoint_poll
is. Also fix the style of the declaration of CreateGCStatepoint, in
preparation for another change that will wrap it.
Reviewers: reames
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D7517
llvm-svn: 228626
Ramkumar Ramachandra [Mon, 9 Feb 2015 23:00:40 +0000 (23:00 +0000)]
PlaceSafepoints: modernize gc.result.* -> gc.result
Differential Revision: http://reviews.llvm.org/D7516
llvm-svn: 228625
Gabor Horvath [Mon, 9 Feb 2015 22:52:26 +0000 (22:52 +0000)]
[Static Analyzer] The name of the checker that reports a bug is added
to the plist output. This check_name field does not guaranteed to be the
same as the name of the checker in the future.
Reviewer: Anna Zaks
Differential Revision: http://reviews.llvm.org/D6841
llvm-svn: 228624
Duncan P. N. Exon Smith [Mon, 9 Feb 2015 22:48:04 +0000 (22:48 +0000)]
DebugInfo: Remove DW_TAG_constant
Remove handling for DW_TAG_constant. We started producing it in
r110656, but reverted that in r110876 without dropping the support.
Finish the job.
llvm-svn: 228623
Philip Reames [Mon, 9 Feb 2015 22:44:03 +0000 (22:44 +0000)]
Update file comment to clarify points highlighted in review (NFC)
llvm-svn: 228621
Philip Reames [Mon, 9 Feb 2015 22:26:11 +0000 (22:26 +0000)]
Use range for loops in PlaceSafepoints (NFC)
llvm-svn: 228620
Duncan P. N. Exon Smith [Mon, 9 Feb 2015 22:13:27 +0000 (22:13 +0000)]
IR: Take uint64_t in DIBuilder::createExpression()
`DIExpression` deals with `uint64_t`, so it doesn't make sense that
`createExpression()` is created from `int64_t`. Switch to `uint64_t` to
unify them.
I've temporarily left in the `int64_t` version, which forwards to the
`uint64_t` version. I'll delete it once I've updated the callers.
llvm-svn: 228619
Philip Reames [Mon, 9 Feb 2015 22:10:15 +0000 (22:10 +0000)]
Introduce more tests for PlaceSafepoints
These tests the two optimizations for backedge insertion currently implemented and the split backedge flag which is currently off by default.
llvm-svn: 228617
Chris Bieneman [Mon, 9 Feb 2015 22:07:49 +0000 (22:07 +0000)]
Revert "Raising minimum required CMake version to 2.8.12.2."
This reverts commit
add62ac537d8249fa2161405066e318ca80e199d.
llvm-svn: 228616
Chris Bieneman [Mon, 9 Feb 2015 22:05:07 +0000 (22:05 +0000)]
Raising minimum required CMake version to 2.8.12.2.
llvm-svn: 228615
Colin LeMahieu [Mon, 9 Feb 2015 21:56:37 +0000 (21:56 +0000)]
[Hexagon] Removing more V4 predicates since V4 is the required minimum.
llvm-svn: 228614
Ben Langmuir [Mon, 9 Feb 2015 21:55:44 +0000 (21:55 +0000)]
Be more conservative about gethostname()'s truncating behaviour
Don't assume it will provide an error or null-terminate the string on
truncation, since POSIX doesn't guarantee either behaviour (although
Linux and Darwin at least will do the 'right thing').
llvm-svn: 228613
Philip Reames [Mon, 9 Feb 2015 21:50:31 +0000 (21:50 +0000)]
Minor test cleanup
a) add gc attribute
b) remove unused param
llvm-svn: 228612
Ramkumar Ramachandra [Mon, 9 Feb 2015 21:50:03 +0000 (21:50 +0000)]
MemDerefPrinter: Require DataLayoutPass for higher accuracy
Without a valid data layout, deferenceable(N) doesn't get parsed or
propagated. Since this is the key item we are testing, add a dependency
on the pass.
Differential Revision: http://reviews.llvm.org/D7508
llvm-svn: 228611
Philip Reames [Mon, 9 Feb 2015 21:48:05 +0000 (21:48 +0000)]
Add basic tests for PlaceSafepoints
This is just adding really simple tests which should have been part of the original submission. When doing so, I discovered that I'd mistakenly removed required pieces when preparing the patch for upstream submission. I fixed two such bugs in this submission.
llvm-svn: 228610
Duncan P. N. Exon Smith [Mon, 9 Feb 2015 21:30:05 +0000 (21:30 +0000)]
Verifier: Const-qualify Metadata, NFC
llvm-svn: 228609
Duncan P. N. Exon Smith [Mon, 9 Feb 2015 21:26:34 +0000 (21:26 +0000)]
IR: Document horrible abuse of loose DIDescriptor, NFC
I'll circle back and fix this somehow; for now I just don't want to
forget about it.
llvm-svn: 228608
Duncan P. N. Exon Smith [Mon, 9 Feb 2015 21:23:34 +0000 (21:23 +0000)]
IR: Remove dead code in DITemplate*
These are never referenced or filled in.
llvm-svn: 228607
Ramkumar Ramachandra [Mon, 9 Feb 2015 21:08:03 +0000 (21:08 +0000)]
isDereferenceablePointer: look through gc.relocate calls
While a theoretical GC might change dereferenceability on collection,
there is no such known collector and no need to account for the case
with a flag yet.
Differential Revision: http://reviews.llvm.org/D7454
llvm-svn: 228606
Colin LeMahieu [Mon, 9 Feb 2015 21:07:35 +0000 (21:07 +0000)]
[Hexagon] Removing v2-4 flags. V4 is the minimum supported version.
llvm-svn: 228605
Ben Langmuir [Mon, 9 Feb 2015 20:35:13 +0000 (20:35 +0000)]
Diagnose timeouts in the LockFileManager and delete the dead lock file
If the lock file manager times out, we should give an error rather than
silently trying to load the existing module. And delete the
(presumably) dead lock file, since it will otherwise prevent progress in
future invokations. This is unsound since we have no way to prove that
the lock file we are deleting is the same one we timed out on, but since
the lock is only to avoid excessive rebuilding anyway it should be okay.
Depends on llvm r228603.
llvm-svn: 228604
Ben Langmuir [Mon, 9 Feb 2015 20:34:24 +0000 (20:34 +0000)]
Reduce the LockFileManager timeout, and provide unsafeRemoveLockFile
5 minutes is an eternity, so try to strike a better balance between
waiting long enough for any reasonable module build and not so long that
users kill the process because they think it's hanging.
Also give the client a way to delete the lock file after a timeout.
llvm-svn: 228603
Colin LeMahieu [Mon, 9 Feb 2015 20:33:46 +0000 (20:33 +0000)]
[Hexagon] Factoring classes out of store patterns.
llvm-svn: 228602
Ben Langmuir [Mon, 9 Feb 2015 20:13:11 +0000 (20:13 +0000)]
Update r228592 for when gethostname() returns an error
If gethostname() is not successful, just skip adding the hostname to the
module hash. And don't bother setting hostname[255] = 0, since if
gethostname() is successful, it will be null-terminated already (and if
it's not successful we don't read the string now.
llvm-svn: 228601
Colin LeMahieu [Mon, 9 Feb 2015 20:03:42 +0000 (20:03 +0000)]
[Hexagon] Formatting v5 TD file. Removing commented defs.
llvm-svn: 228598
Ramkumar Ramachandra [Mon, 9 Feb 2015 19:49:54 +0000 (19:49 +0000)]
MemDepPrinter: cleanup a few loops (NFC)
Make use of the newly introduced inst_range to clean up two loops. Clean
up a third one while at it.
Differential Revision: http://reviews.llvm.org/D7455
llvm-svn: 228596
Sanjoy Das [Mon, 9 Feb 2015 19:39:00 +0000 (19:39 +0000)]
Address post-commit review for rL228587: make it explicit that the
<NW> bit of a SCEVAddRecExpr does not depend on the sign of the step
and the start value of the step.
llvm-svn: 228595
Ben Langmuir [Mon, 9 Feb 2015 19:30:29 +0000 (19:30 +0000)]
Add missing include from r228592
llvm-svn: 228594
Colin LeMahieu [Mon, 9 Feb 2015 19:24:44 +0000 (19:24 +0000)]
[Hexagon] Cleaning up definition formatting.
llvm-svn: 228593
Ben Langmuir [Mon, 9 Feb 2015 19:23:08 +0000 (19:23 +0000)]
Add the hostname to the module hash to avoid sharing between hosts
Sharing between hosts will cause problems for the LockFileManager, which
can timeout waiting for a process that has already died.
llvm-svn: 228592
David Blaikie [Mon, 9 Feb 2015 19:13:51 +0000 (19:13 +0000)]
DebugInfo: Refactor default arg handling into a common place (instead of handling in repeatedly for aggregate, complex, and scalar types)
llvm-svn: 228591
Zachary Turner [Mon, 9 Feb 2015 19:13:46 +0000 (19:13 +0000)]
Dont' use O_CLOEXEC on Windows, since it doesn't exist.
llvm-svn: 228590
David Blaikie [Mon, 9 Feb 2015 18:55:57 +0000 (18:55 +0000)]
DebugInfo: Suppress the location of instructions in complex default arguments.
llvm-svn: 228589
David Blaikie [Mon, 9 Feb 2015 18:47:14 +0000 (18:47 +0000)]
DebugInfo: Suppress the location of instructions in aggregate default arguments.
Matches the existing code for scalar default arguments. Complex default
arguments probably need the same handling too (test/fix to that coming
next).
llvm-svn: 228588
Sanjoy Das [Mon, 9 Feb 2015 18:44:42 +0000 (18:44 +0000)]
Clarify the wording on what it means for a SCEVAddRecExpr to be <NW>.
llvm-svn: 228587
Sanjoy Das [Mon, 9 Feb 2015 18:34:55 +0000 (18:34 +0000)]
Bugfix: SCEV incorrectly marks certain add recurrences as nsw
When creating a scev for sext({X,+,Y}), scev checks if the expression
is equivalent to {sext X,+,zext Y}. If it can prove that, it also
tags the original {X,+,Y} as <nsw>, which is not correct.
In the test case I run `-scalar-evolution` twice because the bug
manifests only once SCEV has run through and seen the `sext`
expressions (and then does a in-place mutation on {X,+,Y}).
Differential Revision: http://reviews.llvm.org/D7495
llvm-svn: 228586
Lang Hames [Mon, 9 Feb 2015 18:16:43 +0000 (18:16 +0000)]
[Orc] Revert r228567 (GCC ICE workaround) - it doesn't seem to have helped.
As far as I can tell r228568 was the right workaround, and r228567 was
unnecessary. If reverting this causes problems on the bots I'll reinstate it.
llvm-svn: 228585
Hafiz Abid Qadeer [Mon, 9 Feb 2015 18:14:01 +0000 (18:14 +0000)]
Fix MI notification test case on Linux.
This test case was checking for a specific stop reason which is different on
Linux. This caused the test to fail. It now only checks for stoppped. Also added
some more steps to run to main so that we dont pass when application has run to
completion.
llvm-svn: 228584
Samuel Benzaquen [Mon, 9 Feb 2015 17:50:40 +0000 (17:50 +0000)]
Verify assign operator signatures.
Summary: Warn when the return type of assign operators is not Class&.
Reviewers: alexfh
Subscribers: cfe-commits
Differential Revision: http://reviews.llvm.org/D6667
llvm-svn: 228583
Pavel Labath [Mon, 9 Feb 2015 17:42:47 +0000 (17:42 +0000)]
Fix TestFdLeak.py on Mac.
bug introduced in D7466. For some reason target.Launch behaves differently on linux and mac.
llvm-svn: 228582
Sanjay Patel [Mon, 9 Feb 2015 17:17:09 +0000 (17:17 +0000)]
fixed to test features, not CPUs
llvm-svn: 228581
Kit Barton [Mon, 9 Feb 2015 17:03:18 +0000 (17:03 +0000)]
This change implements the following three logical vector operations:
veqv (vector equivalence)
vnand
vorc
I increased the AddedComplexity for these instructions to 500 to ensure they are generated instead of issuing other VSX instructions.
Phabricator review: http://reviews.llvm.org/D7469
llvm-svn: 228580
Sanjay Patel [Mon, 9 Feb 2015 16:30:58 +0000 (16:30 +0000)]
rename variable to give it some meaning; remove obvious comments; NFC
llvm-svn: 228579
Sanjay Patel [Mon, 9 Feb 2015 16:04:52 +0000 (16:04 +0000)]
fix comment that didn't match the code; remove unnecessary braces; NFC
llvm-svn: 228578
Hafiz Abid Qadeer [Mon, 9 Feb 2015 14:34:12 +0000 (14:34 +0000)]
Fix test case for data-disassemble instruction in lldb-mi.
Previously the offset field showed the offset from the section base.
I have fixed it so that first disassembled instruction has offset of 0.
Also made a little modification in the test case to match the output coming
form the lldb-mi.
llvm-svn: 228577
Hafiz Abid Qadeer [Mon, 9 Feb 2015 13:54:51 +0000 (13:54 +0000)]
Fix test case failure on Linux.
The test case were failing becuase my test compiler was gcc. The generated
code behaved a bit differently to the how the test expected. Adjusted a few
lines so that it works on both gcc and clang.
llvm-svn: 228576
Tobias Grosser [Mon, 9 Feb 2015 13:52:21 +0000 (13:52 +0000)]
isl is now distributed with polly
llvm-svn: 228575
Ilia K [Mon, 9 Feb 2015 13:18:24 +0000 (13:18 +0000)]
Remove forgotten file test/tools/lldb-mi/TestMiProgramArgs.py after r228286
llvm-svn: 228574
Timur Iskhodzhanov [Mon, 9 Feb 2015 12:47:51 +0000 (12:47 +0000)]
[ASan/Win] Add a test case for PR22431
llvm-svn: 228573
Johannes Doerfert [Mon, 9 Feb 2015 12:34:23 +0000 (12:34 +0000)]
Allow ScalarEvolution to catch more min/max cases
For the attached test case different types are used in the ICmpInst
and SelectInst that represent the min/max expressions. However, if the
ICmpInst type is smaller a comparison with the sign/zero extended
operands would have yielded the same result. This situation might
arise after the instruction combination pass was applied.
Differential Revision: http://reviews.llvm.org/D7338
llvm-svn: 228572
Benjamin Kramer [Mon, 9 Feb 2015 11:48:43 +0000 (11:48 +0000)]
Update cxx_status to list features implemented in Clang 3.6.
Still yellow because 3.6 is unreleased. While there make Urbana paper
links clickable and list binary literals as available in Clang 2.9
(they've been available basically since the dawn of Clang, but not
having a version number in the table looks weird)
llvm-svn: 228571
Pavel Labath [Mon, 9 Feb 2015 11:37:56 +0000 (11:37 +0000)]
Fix descriptor leak in multi-target debugging
Summary:
When debugging two targets concurrently, the pseude terminal master fd from the first one would
leak into the second. This fixes the problem by setting O_CLOEXEC on the master fd. Test
included.
Reviewers: clayborg, vharron
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D7466
llvm-svn: 228570
Lang Hames [Mon, 9 Feb 2015 07:47:32 +0000 (07:47 +0000)]
[Orc] Try another workaround for the GCC 4.7.2 ICE introduced in r228557. NFC.
llvm-svn: 228568
Lang Hames [Mon, 9 Feb 2015 07:22:56 +0000 (07:22 +0000)]
[Orc] Tweak lambda capture lists to try to avoid an ICE on gcc-4.7.2. NFC.
Apparently gcc-4.7.2 is touchy about 'this' appearing in a lambda capture list
along with other captures. I've rewritten my captures to try to avoid the issue.
llvm-svn: 228567
Akira Hatanaka [Mon, 9 Feb 2015 06:38:23 +0000 (06:38 +0000)]
Fix a bug in DemoteRegToStack where a reload instruction was inserted into the
wrong basic block.
This would happen when the result of an invoke was used by a phi instruction
in the invoke's normal destination block. An instruction to reload the invoke's
value would get inserted before the critical edge was split and a new basic
block (which is the correct insertion point for the reload) was created. This
commit fixes the bug by splitting the critical edge before all the reload
instructions are inserted.
Also, hoist up the code which computes the insertion point to the only place
that need that computation.
rdar://problem/
15978721
llvm-svn: 228566
David Majnemer [Mon, 9 Feb 2015 06:31:31 +0000 (06:31 +0000)]
MC: Calculate intra-section symbol differences correctly for COFF
This fixes PR22060.
llvm-svn: 228565
Lang Hames [Mon, 9 Feb 2015 04:46:41 +0000 (04:46 +0000)]
[Orc] Fix the MSVC bots by using LLVM_EXPLICIT rather than explicit.
llvm-svn: 228564
Craig Topper [Mon, 9 Feb 2015 04:04:53 +0000 (04:04 +0000)]
[X86] Remove 256-bit and 512-bit memop pattern fragments. They are no longer used.
llvm-svn: 228563
Craig Topper [Mon, 9 Feb 2015 04:04:50 +0000 (04:04 +0000)]
[X86] Remove 'memop' uses from AVX512. Use 'load' instead.
llvm-svn: 228562
Tim Northover [Mon, 9 Feb 2015 01:21:00 +0000 (01:21 +0000)]
DeadArgElim: fix mismatch in accounting of array return types.
Some parts of DeadArgElim were only considering the individual fields
of StructTypes separately, but others (where insertvalue &
extractvalue instructions occur) also looked into ArrayTypes.
This one is an actual bug; the mismatch can lead to an argument being
considered used by a return sub-value that isn't being tracked (and
hence is dead by default). It then gets incorrectly eliminated.
llvm-svn: 228559
Tim Northover [Mon, 9 Feb 2015 01:20:53 +0000 (01:20 +0000)]
DeadArgElim: assess uses of entire return value aggregate.
Previously, a non-extractvalue use of an aggregate return value meant
the entire return was considered live (the algorithm gave up
entirely). This was correct, but conservative. It's better to actually
look at that Use, making the analysis results apply to all sub-values
under consideration.
E.g.
%val = call { i32, i32 } @whatever()
[...]
ret { i32, i32 } %val
The return is using the entire aggregate (sub-values 0 and 1). We can
still simplify @whatever if we can prove that this return is itself
unused.
Also unifies the logic slightly between aggregate and non-aggregate
cases..
llvm-svn: 228558
Lang Hames [Mon, 9 Feb 2015 01:20:51 +0000 (01:20 +0000)]
[Orc] Add a JITSymbol class to the Orc APIs, refactor APIs, update clients.
This patch refactors a key piece of the Orc APIs: It removes the
*::getSymbolAddress and *::lookupSymbolAddressIn methods, which returned target
addresses (uint64_ts), and replaces them with *::findSymbol and *::findSymbolIn
respectively, which return instances of the new JITSymbol type. Unlike the old
methods, calling findSymbol or findSymbolIn does not cause the symbol to be
immediately materialized when found. Instead, the symbol will be materialized
if/when the getAddress method is called on the returned JITSymbol. This allows
us to query for the existence of symbols without actually materializing them. In
the future I expect more information to be attached to the JITSymbol class, for
example whether the returned symbol is a weak or strong definition. This will
allow us to properly handle weak symbols and multiple definitions.
llvm-svn: 228557
Ramkumar Ramachandra [Mon, 9 Feb 2015 01:13:13 +0000 (01:13 +0000)]
InstCombine: propagate nonNull through assume
Make assume (load (call|invoke) != null) set nonNull return attribute
for the call and invoke. Also include tests.
Differential Revision: http://reviews.llvm.org/D7107
llvm-svn: 228556
Ramkumar Ramachandra [Mon, 9 Feb 2015 00:30:03 +0000 (00:30 +0000)]
[emacs] Get llvm-mode to font-lock "personality"
Differential Revision: http://reviews.llvm.org/D7494
llvm-svn: 228555
David Blaikie [Sun, 8 Feb 2015 23:15:37 +0000 (23:15 +0000)]
Fix -Wuninitialized build by referencing the relevant ctor parameter instead of the base class member variable.
llvm-svn: 228554
Zachary Turner [Sun, 8 Feb 2015 22:53:53 +0000 (22:53 +0000)]
Make PDBSymbol's IPDBSymbol reference const.
llvm-svn: 228553
Sanjoy Das [Sun, 8 Feb 2015 22:52:17 +0000 (22:52 +0000)]
Bugfix: ScalarEvolution incorrectly assumes that the start of certain
add recurrences don't overflow.
This change makes the optimization more restrictive. It still assumes
that an overflowing `add nsw` is undefined behavior; and this change
will need revisiting once we have a consistent semantics for poison
values.
Differential Revision: http://reviews.llvm.org/D7331
llvm-svn: 228552
Craig Topper [Sun, 8 Feb 2015 22:38:25 +0000 (22:38 +0000)]
[X86] Remove the remaining uses of memop from AVX and AVX2 instruction patterns. AVX and AVX2 can handle unaligned loads being folded so we can just use 'load'
llvm-svn: 228551
Benjamin Kramer [Sun, 8 Feb 2015 21:56:09 +0000 (21:56 +0000)]
Metadata: Use <algorithm> to simplify code. NFC.
llvm-svn: 228550
Justin Hibbits [Sun, 8 Feb 2015 21:34:29 +0000 (21:34 +0000)]
Fix sorting. Spotted by emaste.
llvm-svn: 228549
Justin Hibbits [Sun, 8 Feb 2015 21:23:27 +0000 (21:23 +0000)]
PowerPC64 host also supports 32-bit binaries, so note it.
llvm-svn: 228548
Justin Hibbits [Sun, 8 Feb 2015 21:23:23 +0000 (21:23 +0000)]
Fix off-by-one IsGPR().
f0 was being counted as a GPR, due to the check in IsGPR(). Correct it by
looking at the precise GPR range.
llvm-svn: 228547
Sanjay Patel [Sun, 8 Feb 2015 21:14:27 +0000 (21:14 +0000)]
fix test attributes; this is an SSE2 test, not a Nehalem test
llvm-svn: 228546
Sanjay Patel [Sun, 8 Feb 2015 21:10:40 +0000 (21:10 +0000)]
fix test attributes; this is an x86-64 test, not a Nehalem test
llvm-svn: 228545
Sanjay Patel [Sun, 8 Feb 2015 21:05:03 +0000 (21:05 +0000)]
fix test attributes; these are SSE2 tests, not Nehalem tests
llvm-svn: 228544
David Blaikie [Sun, 8 Feb 2015 21:03:30 +0000 (21:03 +0000)]
Kaleidoscope-Orc: Reuse the IRGen utility function in later chapters, and remove an unused parameter.
llvm-svn: 228543
Zachary Turner [Sun, 8 Feb 2015 20:58:09 +0000 (20:58 +0000)]
DebugInfoPDB: Make the symbol base case hold an IPDBSession ref.
Dumping a symbol often requires access to data that isn't inside
the symbol hierarchy, but which is only accessible through the
top-level session. This patch is a pure interface change to give
symbols a reference to the session.
llvm-svn: 228542
Sanjay Patel [Sun, 8 Feb 2015 20:50:58 +0000 (20:50 +0000)]
fix test attributes; these are SSE2 tests, not Nehalem tests
llvm-svn: 228541