Akira Hatanaka [Wed, 11 Nov 2015 20:35:42 +0000 (20:35 +0000)]
Move the enum attributes defined in Attributes.h to a table-gen file.
This is a step towards consolidating some of the information regarding
attributes in a single place.
This patch moves the enum attributes in Attributes.h to the table-gen
file. Additionally, it adds definitions of target independent string
attributes that will be used in follow-up commits by the inliner to
check attribute compatibility.
Yunzhong Gao [Wed, 11 Nov 2015 19:59:08 +0000 (19:59 +0000)]
Add a libLTO diagnostic handler that supports lto_get_error_message API
This is a follow-up from the previous discussion on the thread:
http://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20151019/307763.html
The LibLTO lto_get_error_message() API reads error messages from a std::string
sLastErrorString. Instead of passing this string around as an argument, this
patch creates a diagnostic handler and then sends this handler to the
constructor of LTOCodeGenerator.
Geoff Berry [Wed, 11 Nov 2015 19:42:52 +0000 (19:42 +0000)]
[DAGCombiner] Improve zextload optimization.
Summary:
Don't fold
(zext (and (load x), cst)) -> (and (zextload x), (zext cst))
if
(and (load x) cst)
will match as a zextload already and has additional users.
For example, the following IR:
%load = load i32, i32* %ptr, align 8
%load16 = and i32 %load, 65535
%load64 = zext i32 %load16 to i64
store i32 %load16, i32* %dst1, align 4
store i64 %load64, i64* %dst2, align 8
used to produce the following aarch64 code:
ldr w8, [x0]
and w9, w8, #0xffff
and x8, x8, #0xffff
str w9, [x1]
str x8, [x2]
but with this change produces the following aarch64 code:
Diego Novillo [Wed, 11 Nov 2015 17:54:37 +0000 (17:54 +0000)]
SamplePGO - Fix PR 25482 - Do not rely on llvm.dbg.cu for discriminators
The discriminators pass relied on the presence of llvm.dbg.cu to decide
whether to add discriminators, but this fails in the case where debug
info is only enabled partially when -fprofile-sample-use is active.
The reason llvm.dbg.cu is not present in these cases is to prevent
codegen from emitting debug info (as it is only used for the sample
profile pass).
This changes the discriminators pass to also emit discriminators even
when debug info is not being emitted.
Sanjay Patel [Wed, 11 Nov 2015 17:24:56 +0000 (17:24 +0000)]
[MIPS] add overrides for isCheapToSpeculateCttz() and isCheapToSpeculateCtlz()
MIPS32 has instructions for efficient count-leading/trailing-zeros, so this should be
considered a cheap operation (and therefore fair game for speculation) for any MIPS32
implementation.
The net result of allowing this speculation for the regression tests in this patch is
that we get this code:
ctlz:
jr $ra
clz $2, $4
cttz:
addiu $1, $4, -1
not $2, $4
and $1, $2, $1
clz $1, $1
addiu $2, $zero, 32
jr $ra
subu $2, $2, $1
Diego Novillo [Wed, 11 Nov 2015 16:39:22 +0000 (16:39 +0000)]
Properly fix unused variable in disable-assert builds.
I missed the side-effects of ParseBFI in my previous attempt (r252748).
Thanks dblaikie for the suggestion of adding a void use of the unused
variable instead.
Chris Bieneman [Wed, 11 Nov 2015 16:14:03 +0000 (16:14 +0000)]
[CMake] Add support for building the llvm test-suite as part of an LLVM build using clang and lld
Summary:
This patch adds a new CMake module for working with ExternalProjects. This wrapper for ExternalProject supports using just-built tools and can hook up dependencies properly so that projects get cleared out.
The example usage here is for the llvm test-suite. In this example, the test-suite is setup as dependent on clang and lld if they are in-tree. If the clang or lld binaries change the test-suite is re-configured, cleaned, and rebuilt.
This cleanup and abstraction wrapping ExternalProject can be extended and applied to other runtime libraries like compiler-rt and libcxx.
James Molloy [Wed, 11 Nov 2015 15:40:40 +0000 (15:40 +0000)]
[ARM] Combine BFIs together
If we have a chain of BFIs, we may be able to combine several together into one merged BFI. We can do this if the "from" bits from one BFI OR'd with the "from" bits from the other BFI form a contiguous range, and the same with the "to" bits.
Charlie Turner [Wed, 11 Nov 2015 15:03:46 +0000 (15:03 +0000)]
[SLP] Enable -slp-vectorize-hor by default.
Measurements primarily on AArch64 have shown this feature does not
significantly effect compile-time. The are no significant perf changes in LNT,
but for AArch64 at least, there are wins in third party benchmarks.
As discussed on llvm-dev, we're going to try turning this on by default and see
how other targets react to the change.
ADT: Avoid relying on UB in ilist_node::getNextNode()
Re-implement `ilist_node::getNextNode()` and `getPrevNode()` without
relying on the sentinel having a "next" pointer. Instead, get access to
the owning list and compare against the `begin()` and `end()` iterators.
This only works when the node *can* get access to the owning list. The
new support is in `ilist_node_with_parent<>`, and any class `Ty`
inheriting from `ilist_node<NodeTy>` that wants `getNextNode()` and/or
`getPrevNode()` should inherit from
`ilist_node_with_parent<NodeTy, ParentTy>` instead. The requirements:
- `NodeTy` must have a `getParent()` function that returns the parent.
- `ParentTy` must have a `getSublistAccess()` static that, given a(n
ignored) `NodeTy*` (to determine which list), returns a member field
pointer to the appropriate `ilist<>`.
This isn't the cleanest way to get access to the owning list, but it
leverages the API already used in the IR hierarchy (see, e.g.,
`Instruction::getSublistAccess()`).
If anyone feels like ripping out the calls to `getNextNode()` and
`getPrevNode()` and replacing with direct iterator logic, they can also
remove the access function, etc., but as an incremental step, I'm
maintaining the API where it's currently used in tree.
If these requirements are *not* met, call sites with access to the ilist
can call `iplist<NodeTy>::getNextNode(NodeTy*)` directly, as in
ilistTest.cpp.
Why rewrite this?
The old code was broken, calling `getNext()` on a sentinel that possibly
didn't have a "next" pointer at all! The new code avoids that
particular flavour of UB (see the commit message for r252538 for more
details about the "lucky" memory layout that made this function so
interesting).
There's still some UB here: the end iterator gets downcast to `NodeTy*`,
even when it's a sentinel (which is typically
`ilist_half_node<NodeTy*>`). I'll tackle that in follow-up commits.
See this llvm-dev thread for more details:
http://lists.llvm.org/pipermail/llvm-dev/2015-October/091115.html
What's the danger?
There might be some code that relies on `getNextNode()` or
`getPrevNode()` *never* returning `nullptr` -- i.e., that relies on them
being broken when the sentinel is an `ilist_half_node<NodeTy>`. I tried
to root out those cases with the audits I did leading up to r252380, but
it's possible I missed one or two. I hope not.
(If (1) you have out-of-tree code, (2) you've reverted r252380
temporarily, and (3) you get some weird crashes with this commit, then I
recommend un-reverting r252380 and auditing the compile errors looking
for "strange" implicit conversions.)
Right now isTruePredicate is only ever called with Pred == ICMP_SLE or
ICMP_ULE, and the ICMP_SLT and ICMP_ULT cases are dead. This change
removes the untested dead code so that the function is not misleading.
Andrew Wilkins [Tue, 10 Nov 2015 23:19:21 +0000 (23:19 +0000)]
[cmake] move SONAME handling to llvm_add_library
Summary:
Move handling of the SONAME option from add_llvm_library
to llvm_add_library, so that it can be used in sub-projects.
In particular, this makes it possible to have consistently
named shared libraries for LLVM, Clang and LLDB.
Also, base the SONAME and symlinks on the output name
by extracting the OUTPUT_NAME property, rather than assuming
it is the same as the target name.
Jingyue Wu [Tue, 10 Nov 2015 22:35:47 +0000 (22:35 +0000)]
[doc] Compile CUDA with LLVM
Summary:
This patch adds documentation on compiling CUDA with LLVM as requested by many
engineers and researchers. It includes not only user guides but also some
internals (mostly optimizations) so that early adopters can start hacking and
contributing.
Quite a few researchers who contacted us haven't used LLVM before, which is
unsurprising as it hasn't been long since LLVM picked up CUDA. So I added a
short summary to help these folks get started with LLVM.
I expect this document to evolve substantially down the road. The user guides
will be much simplified after the Clang integration is done. However, the
internals should continue growing to include for example performance debugging
and key areas to improve.
Bill Schmidt [Tue, 10 Nov 2015 21:38:26 +0000 (21:38 +0000)]
[PowerPC] Add an MI SSA peephole pass.
This patch adds a pass for doing PowerPC peephole optimizations at the
MI level while the code is still in SSA form. This allows for easy
modifications to the instructions while depending on a subsequent pass
of DCE. Both passes are very fast due to the characteristics of SSA.
At this time, the only peepholes added are for cleaning up various
redundancies involving the XXPERMDI instruction. However, I would
expect this will be a useful place to add more peepholes for
inefficiencies generated during instruction selection. The pass is
placed after VSX swap optimization, as it is best to let that pass
remove unnecessary swaps before performing any remaining clean-ups.
The utility of these clean-ups are demonstrated by changes to four
existing test cases, all of which now have tighter expected code
generation. I've also added Eric Schweiz's bugpoint-reduced test from
PR25157, for which we now generate tight code. One other test started
failing for me, and I've fixed it
(test/Transforms/PlaceSafepoints/finite-loops.ll) as well; this is not
related to my changes, and I'm not sure why it works before and not
after. The problem is that the CHECK-NOT: of "statepoint" from test1
fails because of the "statepoint" in test2, and so forth. Adding a
CHECK-LABEL in between keeps the different occurrences of that string
properly scoped.
Adrian Prantl [Tue, 10 Nov 2015 21:31:05 +0000 (21:31 +0000)]
dsymutil: Prune module forward decl DIEs if a uniquable definition was
already emitted and fix a latent bug in DIECloner where the DW_CHILDREN_yes
flag is set based on the number of children in the input DIE rather than
the number of children that are actually being cloned.
Teresa Johnson [Tue, 10 Nov 2015 21:09:06 +0000 (21:09 +0000)]
Ensure ModuleLinker materializes complete comdat groups
Summary:
The module linker lazy links some "discardable if unused" global
values (e.g. linkonce), materializing and linking them only
if they are referenced in the module. If a comdat group contains a
linkonce member that is not referenced, however, it would not be
materialized and linked, leading to an incomplete comdat group.
If there are other object files not part of the same LTO link that also
define and use that comdat group, the linker may select the incomplete
group leading to link time unsats.
To solve this, whenever a global value body is linked, make sure we
materialize any other members of the same comdat group that are not yet
materialized. This ensures they are in the lazy link list and get linked
as well.
Added new test and adjusted old test to remove parts that didn't
make sense with fix.
Sanjoy Das [Tue, 10 Nov 2015 20:13:21 +0000 (20:13 +0000)]
[IR] Make {Call,Invoke}::cloneImpl aware of operand bundles
This was an omission in the patch that landed initial support for
operand bundles. So far we haven't hit this, but we will once the
inliner is able to inline calls to functions that contain calls with
operand bundles.
Sanjay Patel [Tue, 10 Nov 2015 19:24:31 +0000 (19:24 +0000)]
[ARM] add overrides for isCheapToSpeculateCttz() and isCheapToSpeculateCtlz()
ARM V6T2 has instructions for efficient count-leading/trailing-zeros, so this should be
considered a cheap operation (and therefore fair game for speculation) for any ARM V6T2
implementation.
The net result of allowing this speculation for the regression tests in this patch is
that we get this code:
ctlz:
clz r0, r0
bx lr
cttz:
rbit r0, r0
clz r0, r0
bx lr
Philip Reames [Tue, 10 Nov 2015 18:46:14 +0000 (18:46 +0000)]
[ValueTracking] Recognize that and(x, add (x, -1)) clears the low bit
This is a cleaned up version of a patch by John Regehr with permission. Originally found via the souper tool.
If we add an odd number to x, then bitwise-and the result with x, we know that the low bit of the result must be zero. Either it was zero in x originally, or the add cleared it in the temporary value. As a result, one of the two values anded together must have the bit cleared.
Teresa Johnson [Tue, 10 Nov 2015 18:20:11 +0000 (18:20 +0000)]
[ThinLTO] WeakAny fixes/cleanup
Ensure WeakAny variables are imported as ExternalWeak declarations. To
handle WeakAny more consistently and fix this issue:
1) Update helper doImportAsDefinition to properly flag WeakAny variables
and aliases as not importing defintions.
Update callers of doImportAsDefinition to remove now redundant checks for
WeakAny aliases, or ignore aliases, as appropriate.
2) Add any !doImportAsDefinition GVs to DoNotLinkFromSource set during
linking of the GV prototype, where we usually add GVs to the
DoNotLinkFromSource set for other reasons.
Remove now unnecessary adding of WeakAny aliases to
DoNotLinkFromSource set from copyGlobalAliasProto.
Remove now unnecessary guard against linking non-imported function
bodies from ModuleLinker::run.
Sanjay Patel [Tue, 10 Nov 2015 18:11:37 +0000 (18:11 +0000)]
[AArch64] add overrides for isCheapToSpeculateCttz() and isCheapToSpeculateCtlz()
AArch64 has instructions for efficient count-leading/trailing-zeros, so this should be
considered a cheap operation (and therefore fair game for speculation) for any AArch64
implementation.
The net result of allowing this speculation for the regression tests in this
patch is that we get this code:
ctlz:
clz w0, w0
ret
cttz:
rbit w8, w0
clz w0, w8
ret
Instead of:
ctlz:
cbz w0, .LBB0_2
clz w0, w0
ret
.LBB0_2:
orr w0, wzr, #0x20
ret
cttz:
cbz w0, .LBB1_2
rbit w8, w0
clz w0, w8
ret
.LBB1_2:
orr w0, wzr, #0x20
ret
Sanjay Patel [Tue, 10 Nov 2015 16:48:53 +0000 (16:48 +0000)]
add 'MustReduceDepth' as an objective/cost-metric for the MachineCombiner
This is one of the problems noted in PR25016:
https://llvm.org/bugs/show_bug.cgi?id=25016
and:
http://lists.llvm.org/pipermail/llvm-dev/2015-October/090998.html
The spilling problem is independent and not addressed by this patch.
The MachineCombiner was doing reassociations that don't improve or even worsen the critical path.
This is caused by inclusion of the "slack" factor when calculating the critical path of the original
code sequence. If we don't add that, then we have a more conservative cost comparison of the old code
sequence vs. a new sequence. The more liberal calculation must be preserved, however, for the AArch64
MULADD patterns because benchmark regressions were observed without that.
The two failing test cases now have identical asm that does what we want:
a + b + c + d ---> (a + b) + (c + d)
James Molloy [Tue, 10 Nov 2015 14:22:05 +0000 (14:22 +0000)]
Reapply "[ARM] Combine CMOV into BFI where possible"
Added fixes for stage2 failures: CMOV is not commutable; commuting the operands results in the condition being flipped! d'oh!
Original commit message:
If we have a CMOV, OR and AND combination such as:
if (x & CN)
y |= CM;
And:
* CN is a single bit;
* All bits covered by CM are known zero in y;
Then we can convert this to a sequence of BFI instructions. This will always be a win if CM is a single bit, will always be no worse than the TST & OR sequence if CM is two bits, and for thumb will be no worse if CM is three bits (due to the extra IT instruction).
Igor Laevsky [Tue, 10 Nov 2015 14:10:31 +0000 (14:10 +0000)]
Strip metadata when speculatively hoisting instructions
This is fix for PR24059.
When we are hoisting instruction above some condition it may turn out
that metadata on this instruction was control dependant on the condition.
This metadata becomes invalid and we need to drop it.
This patch should cover most obvious places of speculative execution (which
I have found by greping isSafeToSpeculativelyExecute). I think there are more
cases but at least this change covers the severe ones.
Oliver Stannard [Tue, 10 Nov 2015 11:04:18 +0000 (11:04 +0000)]
[AArch64] Fix halfword load merging for big-endian targets
For big-endian targets, when we merge two halfword loads into a word load, the
order of the halfwords in the loaded value is reversed compared to
little-endian, so the load-store optimiser needs to swap the destination
registers.
This does not affect merging of two word loads, as we use ldp, which treats the
memory as two separate 32-bit words.
Andy Ayers [Tue, 10 Nov 2015 01:50:49 +0000 (01:50 +0000)]
Support for emitting inline stack probes
For CoreCLR on Windows, stack probes must be emitted as inline sequences that probe successive stack pages
between the current stack limit and the desired new stack pointer location. This implements support for
the inline expansion on x64.
For in-body alloca probes, expansion is done during instruction lowering. For prolog probes, a stub call
is initially emitted during prolog creation, and expanded after epilog generation, to avoid complications
that arise when introducing new machine basic blocks during prolog and epilog creation.
Added a new test case, modified an existing one to exclude non-x64 coreclr (for now).
Tim Northover [Tue, 10 Nov 2015 00:44:23 +0000 (00:44 +0000)]
AArch64: add experimental support for address tagging.
AArch64 has the ability to use the top 8-bits of an "address" for extra
information, with the memory subsystem automatically masking them off for loads
and stores. When that's happening, we can sometimes skip masks on memory
operations in the compiler.
However, this requires the host OS and support stack to preserve those bits so
it can't be enabled everywhere. In principle iOS 8.0 and above do take the
required precautions and but we'll put it under a flag for now.
Kevin Enderby [Tue, 10 Nov 2015 00:31:08 +0000 (00:31 +0000)]
Fix llvm-nm(1) printing of llvm-bitcode files for -format darwin to match darwin’s nm(1).
Also a small fix to match printing of Mach-O objects with -format posix.
Derek Schuff [Tue, 10 Nov 2015 00:30:57 +0000 (00:30 +0000)]
[WebAssembly] Support 'unreachable' expression
Lower LLVM's 'unreachable' terminator to ISD::TRAP, and lower ISD::TRAP to
wasm's 'unreachable' expression.
WebAssembly type-checks expressions, but a noreturn function with a
return type that doesn't match the context will cause a check
failure. So we lower LLVM 'unreachable' to ISD::TRAP and then lower that
to WebAssembly's 'unreachable' expression, which typechecks in any
context and causes a trap if executed.
Matt Arsenault [Tue, 10 Nov 2015 00:30:14 +0000 (00:30 +0000)]
Remove unnecessary call to getAllocatableRegClass
I'm not sure what the point of this was. I'm not sure why
you would ever define an instruction that produces an unallocatable
register class. No tests fail with this removed, and it seems like
it should be a verifier error to define such an instruction.
This was problematic for AMDGPU because it would make bad decisions
by arbitrarily changing the register class when unsetting isAllocatable
for VS_32/VS_64, which is currently set as a workaround to this problem.
AMDGPU uses the VS_32/VS_64 register classes to represent operands which
can use either VGPRs or SGPRs. When isAllocatable is unset for these,
this would need to pick either the SGPR or VGPR class and insert either
a copy we don't want, or an illegal copy we would need to deal with
later. A semi-arbitrary register class ordering decision is made in tablegen,
which resulted in always picking a VGPR class because it happens to have
more registers than the SGPR register class. We really just want to
use whatever register class the original register had.
[PGO] Make indexed value profile data more compact
- Make indexed value profile data more compact by peeling out
the per-site value count field into its own smaller sized array.
- Introduced formal data structure definitions to specify value
profile data layout in indexed format. Previously the layout
of the data is only assumed in the client code (scattered in
three different places : size computation, EmitData, and ReadData
- The new data structure serves as a central place for layout documentation.
- Add interfaces to force BE output for value profile data (testing purpose)
- Add byte swap unit tests
projects / llvm / log