Before this change, the *InstPrinter.cpp files of each target where some
of the slowest objects to compile in all of LLVM. See this snippet produced by
ClangBuildAnalyzer:
https://reviews.llvm.org/P8171$96
Search for "InstPrinter", and see that it shows up in a few places.
Tablegen was emitting a large switch containing a sequence of operand checks,
each of which created many conditions and many BBs. Register allocation and
jump threading both did not scale well with such a large repetitive sequence of
basic blocks.
So, this change essentially turns those control flow structures into
data. The previous structure looked like:
switch (Opc) {
case TGT::ADD:
// check alias 1
if (MI->getOperandCount() == N && // check num opnds
MI->getOperand(0).isReg() && // check opnd 0
...
MI->getOperand(1).isImm() && // check opnd 1
AsmString = "foo";
break;
}
// check alias 2
if (...)
...
return false;
The new structure looks like:
OpToPatterns: Sorted table of opcodes mapping to pattern indices.
\->
Patterns: List of patterns. Previous table points to subrange of
patterns to match.
\->
Conds: The if conditions above encoded as a kind and 32-bit value.
See MCInstPrinter.cpp for the details of how the new data structures are
interpreted.
Here are some before and after metrics.
Time to compile AArch64InstPrinter.cpp:
0m29.062s vs. 0m2.203s
size of the obj:
3.9M vs. 676K
size of clang.exe:
97M vs. 96M
I have not benchmarked disassembly performance, but typically
disassemblers are bottlenecked on IO and string processing, not alias
matching, so I'm not sure it's interesting enough to be worth doing.
Reviewers: RKSimon, andreadb, xbolva00, craig.topper
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D70650
The xor'ing behaviour is only used for msvc/crt environments, when we're targeting
macho the guard load code doesn't know about the xor in the epilog. Disable xor'ing
when targeting win32-macho to be consistent.
Differential Revision: https://reviews.llvm.org/D71095
This commit sets the Self and Imp declarations for ObjC method declarations,
in addition to the definitions. It also fixes
a bunch of code in clang that had wrong assumptions about when getSelfDecl() would be set:
- CGDebugInfo::getObjCMethodName and AnalysisConsumer::getFunctionName would assume that it was
set for method declarations part of a protocol, which they never were,
and that self would be a Class type, which it isn't as it is id for a protocol.
Also use the Canonical Decl to index the set of Direct methods so that
when calls and implementations interleave, the same llvm::Function is
used and the same symbol name emitted.
Radar-Id: rdar://problem/57661767
Patch by: Pierre Habouzit
Differential Revision: https://reviews.llvm.org/D71091
D53794 introduced code to perform the FP_TO_UINT expansion via FP_TO_SINT in a way that would never expose floating-point exceptions in the intermediate steps. Unfortunately, I just noticed there is still a way this can happen. As discussed in D53794, the compiler now generates this sequence:
// Sel = Src < 0x8000000000000000
// Val = select Sel, Src, Src - 0x8000000000000000
// Ofs = select Sel, 0, 0x8000000000000000
// Result = fp_to_sint(Val) ^ Ofs
The problem is with the Src - 0x8000000000000000 expression. As I mentioned in the original review, that expression can never overflow or underflow if the original value is in range for FP_TO_UINT. But I missed that we can get an Inexact exception in the case where Src is a very small positive value. (In this case the result of the sub is ignored, but that doesn't help.)
Instead, I'd suggest to use the following sequence:
// Sel = Src < 0x8000000000000000
// FltOfs = select Sel, 0, 0x8000000000000000
// IntOfs = select Sel, 0, 0x8000000000000000
// Result = fp_to_sint(Val - FltOfs) ^ IntOfs
In the case where the value is already in range of FP_TO_SINT, we now simply compute Val - 0, which now definitely cannot trap (unless Val is a NaN in which case we'd want to trap anyway).
In the case where the value is not in range of FP_TO_SINT, but still in range of FP_TO_UINT, the sub can never be inexact, as Val is between 2^(n-1) and (2^n)-1, i.e. always has the 2^(n-1) bit set, and the sub is always simply clearing that bit.
There is a slight complication in the case where Val is a constant, so we know at compile time whether Sel is true or false. In that scenario, the old code would automatically optimize the sub away, while this no longer happens with the new code. Instead, I've added extra code to check for this case and then just fall back to FP_TO_SINT directly. (This seems to catch even slightly more cases.)
Original version of the patch by Ulrich Weigand. X86 changes added by Craig Topper
Differential Revision: https://reviews.llvm.org/D67105
When implementation of the block runtime is available, we should not
warn that block layout fields are uninitialized simply because they're
on the stack.
Summary:
musttail calls should not require allocating extra stack for arguments.
Updates to arguments passed in memory should happen in place before the
epilogue.
This bug was mostly a missed optimization, unless inalloca was used and
store to push conversion fired.
If a reserved call frame was used for an inalloca musttail call, the
call setup and teardown instructions would be deleted, and SP
adjustments would be inserted in the prologue and epilogue. You can see
these are removed from several test cases in this change.
In the case where the stack frame was not reserved, i.e. call frame
optimization fires and turns argument stores into pushes, then the
imbalanced call frame setup instructions created for inalloca calls
become a problem. They remain in the instruction stream, resulting in a
call setup that allocates zero bytes (expected for inalloca), and a call
teardown that deallocates the inalloca pack. This deallocation was
unbalanced, leading to subsequent crashes.
Reviewers: hans
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D71097
fix compile error
The test was failing when run on OSes older than MacOSX10.14 because
aligned deallocation functions are unavailable on older OSes.
rdar://problem/57706710
According to OpenMP 5.0, if clause can be used in for simd directive. If
condition in the if clause if false, the non-vectorized version of the
loop must be executed.
Summary:
Sample profile loader of AutoFDO tries to replay previous inlining using context sensitive profile. The replay only repeats inlining if the call site block is hot. As a result it punts inlining of small functions, some of which can be beneficial for size, and will still be inlined by CSGCC inliner later. The oscillation between sample profile loader's inlining and regular CGSSC inlining cause unnecessary loss of context-sensitive profile. It doesn't have much impact for inline decision itself, but it negatively affects post-inline profile quality as CGSCC inliner have to scale counts which is not as accurate as the original context sensitive profile, and bad post-inline profile can misguide code layout.
This change added regular Inline Cost calculation for sample profile loader, so we can inline small functions upfront under switch -sample-profile-inline-size. In addition -sample-profile-cold-inline-threshold is added so we can tune the separate size threshold - currently the default is chosen to be the same as regular inliner's cold call-site threshold.
Reviewers: wmi, davidxl
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D70750
GCC says:
.../llvm/lib/DebugInfo/GSYM/FunctionInfo.cpp:195:12:
error: ‘InfoType’ is not a class, namespace, or enumeration
case InfoType::EndOfList:
^
Presumably, GCC thinks InfoType is a variable here. Work around it by
using the name IT as is done above.
These are a part of the libc so linking these explicitly isn't necessary
and embedding these as deplibs causes link time error.
This issues was introduced in a9b5fff which changed how we emit deplibs.
Differential Revision: https://reviews.llvm.org/D71135
This is a follow-up to D70607 where we made any
extract element on SLM more costly than default. But that is
pessimistic for extract from element 0 because that corresponds
to x86 movd/movq instructions. These generally have >1 cycle
latency, but they are probably implemented as single uop
instructions.
Note that no vectorization tests are affected by this change.
Also, no targets besides SLM are affected because those are
falling through to the default cost of 1 anyway. But this will
become visible/important if we add more specializations via cost
tables.
Differential Revision: https://reviews.llvm.org/D71023
Summary:
Split off of D67120.
Add the profile guided size optimization instrumentation / queries in the code
gen or target passes. This doesn't enable the size optimizations in those passes
yet as they are currently disabled in shouldOptimizeForSize (for non-IR pass
queries).
Reviewers: davidxl
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D71072
Current tail duplication integrated in bb layout is designed to increase the fallthrough from a BB's predecessor to its successor, but we have observed cases that duplication doesn't increase fallthrough, or it brings too much size overhead.
To overcome these two issues in function canTailDuplicateUnplacedPreds I add two checks:
make sure there is at least one duplication in current work set.
the number of duplication should not exceed the number of successors.
The modification in hasBetterLayoutPredecessor fixes a bug that potential predecessor must be at the bottom of a chain.
Differential Revision: https://reviews.llvm.org/D64376
Summary:
When Sema encounters a ObjCMethodDecl definition it declares the implicit parameters for the ObjCMethodDecl.
When importing such a method with the ASTImporter we need to do the same for the imported method
otherwise we will crash when generating code (where CodeGen expects that this was called by Sema).
Note I had to implement Objective-C[++] support in Language.cpp as this is the first test for Objective-C and this
would otherwise just hit this 'not implemented' assert when running the unit test.
Reviewers: martong, a.sidorin, shafik
Reviewed By: martong
Subscribers: rnkovacs, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D71112
SUMMARY:
if the size of Csect is zero, the Csect do not need write any data into sections
for example, the TOC Csect has zero size, it do not need invoke a
Asm.writeSectionData(W.OS, Csect.MCCsect, Layout);
Reviewers: daltenty
Subscribers: rupprecht, seiyai,hiraditya
Differential Revision: https://reviews.llvm.org/D71120
Summary: Python 3.8 introduces a SyntaxWarning about string comparisons with 'is'. This commit updates the string comparison in clang-format.py that is done with 'is not' to '!='. This should not break compatibility with older python versions (tested 3.4.9, 2.7.17, 2.7.5, 3.8.0).
Reviewers: MyDeveloperDay, klimek, llvm-commits, cfe-commits
Reviewed By: MyDeveloperDay, klimek
Patch By: pseyfert
Tags: #clang-format, #clang
Differential Revision: https://reviews.llvm.org/D70664
Summary:
The current clang-format.py does not handle trailing newlines at the end of a file correctly.
Trailing empty lines get removed except one.
As far as I understand this is because clang-format gets fed from stdin and writes to stdout when called from clang-format.py.
In a "normal" file (with no trailing empty lines) the string that gets passed to clang-format does not contain a trailing '\n' after the '\n'.join from python.
The clang-format binary does not add a trailing newline to input from stdin, but (if there are multiple trailing '\n', all except one get removed).
When reading back this means that we see in python from a "normal" file a string with no trailing '\n'. From a file with (potentially multiple) empty line(s) at the end, we get a string with one trailing '\n' back in python. In the former case all is fine, in the latter case split('\n') makes one empty line at the end of the file out of the clang-format output. Desired would be instead that the **file** ends with a newline, but not with an empty line.
For the case that a user specifies a range to format (and wants to keep trailing empty lines) I did **not** try to fix this by simply removing all trailing newlines from the clang-format output. Instead, I add a '\n' to the unformatted file content (i.e. newline-terminate what is passed to clang-format) and then strip off the last newline from the output (which itself is now for sure the newline termination of the clang-format output).
(Should this get approved, I'll need someone to help me land this.)
Reviewers: klimek, MyDeveloperDay
Reviewed By: MyDeveloperDay
Patch By: pseyfert
Subscribers: cfe-commits, llvm-commits
Tags: #clang-format, #clang
Differential Revision: https://reviews.llvm.org/D70864
update trailing newline treatment in clang-format.py
Summary:
This patch simplifies register accesses in NativeRegisterContextLinux_arm64
and also adds some bare minimum caching to avoid multiple calls to ptrace
during a stop.
Linux ptrace returns data in the form of structures containing GPR/FPR data.
This means that one single call is enough to read all GPRs or FPRs. We do
that once per stop and keep reading from or writing to the buffer that we
have in NativeRegisterContextLinux_arm64 class. Before a resume or detach we
write all buffers back.
This is tested on aarch64 thunder x1 with Ubuntu 18.04. Also tested
regressions on x86_64.
Reviewers: labath, clayborg
Reviewed By: labath
Subscribers: kristof.beyls, lldb-commits
Differential Revision: https://reviews.llvm.org/D69371
Summary:
When creating a test with `lldbinline.MakeInlineTest()`, the reported `inspect.getfile(test.__class__)` is `lldbtest.pyc`, meaning any `.categories` file will be ineffective for those tests. Check for the test_filename first, which inline tests will set.
Additionally, raise an error with the starting dir if `.categories` is not found. This makes the problem more obvious when it occurs: when the test is separated from the test framework tree.
Reviewers: labath, JDevlieghere
Subscribers: lldb-commits
Tags: #lldb
Differential Revision: https://reviews.llvm.org/D71099
Summary:
[libomptarget][nfc] Move cuda threadfence functions behind kmpc_impl
Part of building code under common/ without requiring a cuda compiler
Reviewers: ABataev, jdoerfert, grokos
Reviewed By: ABataev
Subscribers: jvesely, jfb, openmp-commits
Tags: #openmp
Differential Revision: https://reviews.llvm.org/D71102
Summary:
The immediate forms of the MVE VQSHL instruction have MC names like
`MVE_VSLIimms8` and `MVE_VSLIimmu32`. Those names are confusing,
because VSLI is a completely different shift instruction with no
semantic relation to VQSHL. But it just happens to be defined
immediately before VQSHL in `ARMInstrMVE.td`, so this looks like a
copy-paste error. Renamed the ids to match the instruction name.
Reviewers: ostannard, dmgreen, MarkMurrayARM, miyuki
Reviewed By: miyuki
Subscribers: kristof.beyls, hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D71114
SUMMARY:
in the patch https://reviews.llvm.org/D66969 . we need a test case to verify the out text section of the xcoffobject file is correct or not.
but we do not have llvm disassembly tools to dump the xcoffobjectfile . since we commit the patch https://reviews.llvm.org/D70255, we have tools for it. we create this test case for it.
Reviewers: daltenty,hubert.reinterpretcast,
Differential Revision: https://reviews.llvm.org/D70719
Summary:
When trying to calculate the offsets for the jump table entries
we fail to take into account the block alignment, which could be
greater than 4 bytes. This led to cases where the jump table
offset was too big to fit in a byte.
Reviewers: t.p.northover, sdesmalen, ostannard
Reviewed By: ostannard
Subscribers: ostannard, kristof.beyls, hiraditya, llvm-commits
Committed on behalf of David Sherwood (david-arm)
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D70533
Sam Clegg