Top of tree never regressed, but we have internal branches that we constantly merge and we need to make sure we don't regress.
<rdar://problem/19429907>
llvm-svn: 225572
Their linkage can change if they are later explicitly instantiated. We would
previously emit such functions eagerly (as opposed to lazily on first use) if
they have a 'dllexport' or 'used' attribute, and fail an assert when hitting the
explicit instantiation.
This is achieved by replacing the old CodeGenModule::MayDeferGeneration() method
with two new ones: MustBeEmitted() and MayBeEmittedEagerly().
Differential Revision: http://reviews.llvm.org/D6674
llvm-svn: 225570
This fixes an issue of running "script" commands via SBDebugger::HandleCommand(...) and SBCommandInterpreter::HandleCommand(...) deadlocking Xcode.
<rdar://problem/18075038>
llvm-svn: 225567
Now that the way that the partial unrolling threshold for small loops is used
to compute the unrolling factor as been corrected, a slightly smaller threshold
is preferable. This is expected; other targets may need to re-tune as well.
llvm-svn: 225566
When we compute the size of a loop, we include the branch on the backedge and
the comparison feeding the conditional branch. Under normal circumstances,
these don't get replicated with the rest of the loop body when we unroll. This
led to the somewhat surprising behavior that really small loops would not get
unrolled enough -- they could be unrolled more and the resulting loop would be
below the threshold, because we were assuming they'd take
(LoopSize * UnrollingFactor) instructions after unrolling, instead of
(((LoopSize-2) * UnrollingFactor)+2) instructions. This fixes that computation.
llvm-svn: 225565
The bitcode reading interface used std::error_code to report an error to the
callers and it is the callers job to print diagnostics.
This is not ideal for error handling or diagnostic reporting:
* For error handling, all that the callers care about is 3 possibilities:
* It worked
* The bitcode file is corrupted/invalid.
* The file is not bitcode at all.
* For diagnostic, it is user friendly to include far more information
about the invalid case so the user can find out what is wrong with the
bitcode file. This comes up, for example, when a developer introduces a
bug while extending the format.
The compromise we had was to have a lot of error codes.
With this patch we use the DiagnosticHandler to communicate with the
human and std::error_code to communicate with the caller.
This allows us to have far fewer error codes and adds the infrastructure to
print better diagnostics. This is so because the diagnostics are printed when
he issue is found. The code that detected the problem in alive in the stack and
can pass down as much context as needed. As an example the patch updates
test/Bitcode/invalid.ll.
Using a DiagnosticHandler also moves the fatal/non-fatal error decision to the
caller. A simple one like llvm-dis can just use fatal errors. The gold plugin
needs a bit more complex treatment because of being passed non-bitcode files. An
hypothetical interactive tool would make all bitcode errors non-fatal.
llvm-svn: 225562
Summary:
One more attempt to fix UBSan reports: make sure DenseMapInfo::getEmptyKey()
and DenseMapInfo::getTombstoneKey() doesn't do any upcasts/downcasts to/from Value*.
Test Plan: check-llvm test suite with/without UBSan bootstrap
Reviewers: chandlerc, dexonsmith
Subscribers: llvm-commits, majnemer
Differential Revision: http://reviews.llvm.org/D6903
llvm-svn: 225558
This reverts commit r225000, r225021, r225083, r225086, r225090.
The root change (r225000) still has several issues where it's caused
calls to be emitted without debug locations. This causes assertion
failures if/when those calls are inlined.
I'll work up some test cases and fixes before recommitting this.
llvm-svn: 225555
The previous code assumed that such instructions could not have any uses
outside CaseDest, with the motivation that the instruction could not
dominate CommonDest because CommonDest has phi nodes in it. That simply
isn't true; e.g., CommonDest could have an edge back to itself.
llvm-svn: 225552
pshufb can shuffle in zero bytes as well as bytes from a source vector - we can use this to avoid having to shuffle 2 vectors and ORing the result when the used inputs from a vector are all zeroable.
Differential Revision: http://reviews.llvm.org/D6878
llvm-svn: 225551
Summary: This will ease llgs development a bit by not requiring an lldb/lldb.py build to launch the tests. Also, we can now use LLDB_DEBUGSERVER_PATH to point to a debug server to use to run the tests. I used that to point to a ds2 build and run llgs tests against ds2.
Reviewers: clayborg, tfiala, vharron
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D6554
llvm-svn: 225549
MS ABI guard variables end with @4IA, so this patch teaches the
interpreter about that. Additionally, there was an issue with
TurnGuardLoadIntoZero which was causing some guard uses of a
variable to be missed. This fixes that by calling
Instruction::replaceAllUsesWith() instead of trying to replicate
that function.
llvm-svn: 225547
Variable was being declared as signed, but treated as unsigned at
every point of use.
Patch by Dan Sinclair
Differential Revision: http://reviews.llvm.org/D6897
llvm-svn: 225540
I have been seeing a few crashes where LLDB tries to acquire a cached synthetic child by index, and crashes in the ClusterManager obtaining a shared_ptr for that ValueObject
That kind of crash most often means that I am holding on to a raw pointer to a ValueObject that was let go from the cluster
The main way that could happen is that the synthetic provider is being updated at the same time that some child is being accessed from the previous provider state
This fixes the problem by making the children be stored in a thread-safe map
Fixes rdar://18627964
llvm-svn: 225538
doing Load PRE"
It's not really expected to stick around, last time it provoked a weird LTO
build failure that I can't reproduce now, and the bot logs are long gone. I'll
re-revert it if the failures recur.
Original description: Perform Scalar PRE on gep indices that feed loads before
doing Load PRE.
llvm-svn: 225536
This new command will delete user defined regular commands, but not aliases. We still have "command unalias" to remove aliases as they are currently in different buckets. Appropriate error messages are displayed to inform the user when "command unalias" is used on removable user defined commands that points users to the "command delete" command.
Added a test to verify we can remove user defined commands and also verify that "command unalias" fails when used on a user defined command.
<rdar://problem/18248300>
llvm-svn: 225535
Summary:
I've moved the bulk of `lit.cfg` into `test/libcxx/testconfig.py` and
`test/libcxx/testformat.py`. All that remains in `lit.cfg` is the
logic to discover lit.site.cfg if lit.cfg was run directly, and the
logic for loading configuration variants.
The configuration variant flow has changed with this patch. Rather
than instantiating an object of type `<VARIANT>Configuration`, we now
instatiate an object of type `Configuration` that was loaded from the
module `<VARIANT>.testconfig.py`.
This has to be done on a per-project basis rather than in LIT itself
because LIT doesn't actually know where the real test directory is,
only where the site configuration is (which is usually in the output
directory). It's simple enough to do though, so it's fine to require
each project to do it themselves.
I also cleaned up all the pylint issues while I was here, which was
mostly just a matter of fixing long lines.
Reviewers: mclow.lists, jroelofs, EricWF
Reviewed By: EricWF
Subscribers: cfe-commits
Differential Revision: http://reviews.llvm.org/D6881
llvm-svn: 225532
This reverts commit r225498 (but leaves r225499, which was a worthy
cleanup).
My plan was to change `DEBUG_LOC` to store the `MDNode` directly rather
than its operands (patch was to go out this morning), but on reflection
it's not clear that it's strictly better. (I had missed that the
current code is unlikely to emit the `MDNode` at all.)
Conflicts:
lib/Bitcode/Reader/BitcodeReader.cpp (due to r225499)
llvm-svn: 225531
It seemed odd to have to make DefaultImageName be a mutable member of Driver.
We don't need to the full result of computeTargetTriple() to determine the
image name; just base it on DefaultTargetTriple.
llvm-svn: 225530
Summary:
Mips Linux uses $gp to hold a pointer to thread info structure and accesses it
with a named register. This makes this work for LLVM.
The N32 ABI doesn't quite work yet since the frontend generates incorrect IR
for this case. It neglects to truncate the 64-bit GPR to a 32-bit value before
converting to a pointer. Given correct IR (as in the testcase in this patch),
it works correctly.
Reviewers: sstankovic, vmedic, atanasyan
Reviewed By: atanasyan
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D6893
llvm-svn: 225529
The P7 benefits from not have really-small loops so that we either have
multiple dispatch groups in the loop and/or the ability to form more-full
dispatch groups during scheduling. Setting the partial unrolling threshold to
44 seems good, empirically, for the P7. Compared to using no late partial
unrolling, this yields the following test-suite speedups:
SingleSource/Benchmarks/Adobe-C++/simple_types_constant_folding
-66.3253% +/- 24.1975%
SingleSource/Benchmarks/Misc-C++/oopack_v1p8
-44.0169% +/- 29.4881%
SingleSource/Benchmarks/Misc/pi
-27.8351% +/- 12.2712%
SingleSource/Benchmarks/Stanford/Bubblesort
-30.9898% +/- 22.4647%
I've speculatively added a similar setting for the P8. Also, I've noticed that
the unroller does not quite calculate the unrolling factor correctly for really
tiny loops because it neglects to account for the fact that not every loop body
replicant contains an ending branch and counter increment. I'll fix that later.
llvm-svn: 225522
Greg Clayton