To facilitate this, a couple of changes had to be made:
1. `ModuleSubstream` got moved from `DebugInfo/PDB` to
`DebugInfo/CodeView`, and various codeview related types are defined
there. It turns out `DebugInfo/CodeView/Line.h` already defines many of
these structures, but this is really old code that is not endian aware,
doesn't interact well with `StreamInterface` and not very helpful for
getting stuff out of a PDB. Eventually we should migrate the old readobj
`COFFDumper` code to these new structures, or at least merge their
functionality somehow.
2. A `ModuleSubstream` visitor is introduced. Depending on where your
module substream array comes from, different subsets of record types can
be expected. We are already hand parsing these substream arrays in many
places especially in `COFFDumper.cpp`. In the future we can migrate these
paths to the visitor as well, which should reduce a lot of code in
`COFFDumper.cpp`.
Differential Revision: http://reviews.llvm.org/D20936
Reviewed By: ruiu, majnemer
llvm-svn: 271621
Summary:
Instrument GEP instruction for counting the number of struct field
address calculation to approximate the number of struct field accesses.
Adds test struct_field_count_basic.ll to test the struct field
instrumentation.
Reviewers: bruening, aizatsky
Subscribers: junbuml, zhaoqin, llvm-commits, eugenis, vitalybuka, kcc, bruening
Differential Revision: http://reviews.llvm.org/D20892
llvm-svn: 271619
In r270478, where I enabled the new heuristic I posted testing results,
which I got when explicitly passed the thresholds values via CL options.
However, setting the CL options init-values is not enough to change the
default values of thresholds, so I'm changing them in another place now.
llvm-svn: 271615
Although ld64 always outputs linkedit data in the same order, it isn't actually required to. This change makes yaml2obj resilient if the offsets are in arbitrary order.
llvm-svn: 271611
- This only applies to scripts executed by the _internal_ shell script
interpreter.
- This patch reworks the log to look more like a shell transcript, and be less
verbose (but in the interest of calling attention to the important parts).
Here is an example of the new format, for commands with/without failures and
with/without output:
```
$ true
$ echo hi
hi
$ false
note: command had no output on stdout or stderr
error: command failed with exit status 1
```
llvm-svn: 271610
In preparation for porting to the new PM.
Patch by Jake VanAdrighem! (review mainly by me/Justin)
Differential Revision: http://reviews.llvm.org/D20610
llvm-svn: 271607
This reverts r271599, it broke the integration tests.
More places than I expected had nontrival return types in imports, or
else the check was wrong.
llvm-svn: 271606
This commit adds round tripping for MachO symbol data. Symbols are entries in the name list, that contain offsets into the string table which is at the end of the __LINKEDIT segment.
llvm-svn: 271604
The original tests were intended to show a missing transform that would
be solved by D20774:
http://reviews.llvm.org/D20774
But it's not clear that the transform for the simpler tests is a win for
all targets. Make the tests show a larger pattern that should be a win
regardless of the cost of bitcast instructions.
llvm-svn: 271603
Summary:
The module pass pipeline includes a late LICM run after loop
unrolling. LCSSA is implicitly run as a pass dependency of LICM. However no
cleanup pass was run after this, so the LCSSA nodes ended in the optimized output.
Reviewers: hfinkel, mehdi_amini
Subscribers: majnemer, bruno, mzolotukhin, mehdi_amini, llvm-commits
Differential Revision: http://reviews.llvm.org/D20606
llvm-svn: 271602
Under emscripten, C code can take the address of a function implemented
in Javascript (which is exposed via an import in wasm). Because imports
do not have linear memory address in wasm, we need to generate a thunk
to be the target of the indirect call; it call the import directly.
To make this possible, LLVM needs to emit the type signatures for these
functions, because they may not be called directly or referred to other
than where the address is taken.
This uses s new .s directive (.functype) which specifies the signature.
Differential Revision: http://reviews.llvm.org/D20891
llvm-svn: 271599
code. To make the diffs easier to read, clang-format everything first.
No functionality changed.
Patch by Alina Sbirlea!
http://reviews.llvm.org/D20926
llvm-svn: 271595
This first pass only splits apart the records and dumps the line
info kinds and binary data. Subsequent patches will parse out
the binary data into more useful information and dump it in
detail.
llvm-svn: 271576
This patch fixes PR25788, which allows for the parsing of
floating-point constants in non-C locales.
Patch by Antoine Pitrou!
Differential Revision: http://reviews.llvm.org/D15375
llvm-svn: 271574
There are a lot of different kinds of loads to test for,
and these were scattered around inconsistently with
some redundancy. Try to comprehensively test all loads
in a consistent way.
llvm-svn: 271571
StreamRef was designed to be a thin wrapper over an abstract
stream interface that could itself be treated the same as any
other stream interface. For this reason, it inherited publicly
from StreamInterface, and stored a StreamInterface* internally.
But StreamRef was also designed to be lightweight and easily
copyable, similar to ArrayRef. This led to two misuses of
the classes.
1) When creating a StreamRef A from another StreamRef B, it was
possible to end up with A storing a pointer to B, even when
B was a temporary object, leading to use after free.
2) The above situation could be repeated ad nauseum, so that
A stores a pointer to B, which itself stores a pointer to
another StreamRef C, and so on and so on, creating an
unnecessarily level of nesting depth.
This patch removes the public inheritance relationship between
StreamRef and StreamInterface, making it so that we can never
accidentally convert a StreamRef to a StreamInterface.
llvm-svn: 271570
If the processor name failed to parse for amdgcn,
the resulting output would have R600 ISA in it.
If the processor name was missing or invalid for R600,
the wavefront size would not be set and there would be
crashes from missing itinerary data.
Fixes crashes in future commit caused by dividing by the unset/0
wavefront size.
llvm-svn: 271561
We've been pretending that segments are i8imm since the initial
support (r68645), predating the addition of the SEGMENT_REG class
(r81895). That happens to works, but is wrong, and inconsistent
with how we print (e.g., X86ATTInstPrinter::printMemReference)
and parse them (e.g., X86Operand::addMemOperands).
This change shouldn't affect any tool users, but is visible to
library users or out-of-tree tablegen backends: this causes
MCOperandInfo for the segment op to have an RC instead of "unknown",
and TII::getRegClass to actually return something. As the registers
are reserved and no vregs of the class ever created, that shouldn't
change anything.
No test change; no suspicious getRegClass() in X86 and CodeGen.
llvm-svn: 271559
D19271.
Previous attempt was broken by NetBSD, so in this version I've made the
fallback path generic rather than Windows specific and sent both Windows
and NetBSD to it.
I've also re-formatted the code some, and used an exact clone of the
code in PassSupport.h for doing manual call-once using our atomics
rather than rolling a new one.
If this sticks, we can replace the fallback path for Windows with
a Windows-specific implementation that is more reliable.
Original commit message:
This patch adds an llvm_call_once which is a wrapper around
std::call_once on platforms where it is available and devoid
of bugs. The patch also migrates the ManagedStatic mutex to
be allocated using llvm_call_once.
These changes are philosophically equivalent to the changes
added in r219638, which were reverted due to a hang on Win32
which was the result of a bug in the Windows implementation
of std::call_once.
Differential Revision: http://reviews.llvm.org/D5922
llvm-svn: 271558
Unlike other sections that can grow to any size, the COFF section header
stream has maximum length because each record is fixed size and the COFF
file format limits the maximum number of sections. So I decided to not
create a specific stream class for it. Instead, I added a member function
to DbiStream class which returns a vector of COFF headers.
Differential Revision: http://reviews.llvm.org/D20717
llvm-svn: 271557
This is part of an effort to shave allocations from APInt heavy paths. I'll
be moving many of the other operators to r-value references soon and this is
a step towards doing that without too much duplication.
Saves 15k allocations when doing 'opt -O2 verify-uselistorder.bc'.
llvm-svn: 271556
Testing for specific CPUs has a number of problems, better use subtarget
features:
- When some tweak is added for a specific CPU it is often desirable for
the next version of that CPU as well, yet we often forget to add it.
- It is hard to keep track of checks scattered around the target code;
Declaring all target specifics together with the CPU in the tablegen
file is a clear representation.
- Subtarget features can be tweaked from the command line.
To discourage people from using CPU checks in the future I removed the
isCortexXX(), isCyclone(), ... functions. I added an getProcFamily()
function for exceptional circumstances but made it clear in the comment
that usage is discouraged.
Reformat feature list in AArch64.td to have 1 feature per line in
alphabetical order to simplify merging and sorting for out of tree
tweaks.
No functional change intended.
Differential Revision: http://reviews.llvm.org/D20762
llvm-svn: 271555
This is effectively NFC because we already do this transform after r175380:
http://reviews.llvm.org/rL175380
and also via foldBoolSextMaskToSelect().
This change should just make it a bit more efficient to match the pattern.
The original guard was added in r95058:
http://reviews.llvm.org/rL95058
A sampling of codegen for current in-tree targets shows no problems. This
makes sense given that we're already producing the vector selects via the
other transforms.
llvm-svn: 271554
Summary:
Also convert test/CodeGen/PowerPC/vsx-ldst-builtin-le.ll to use
FileCheck instead of two grep and count runs.
This change is needed to avoid spurious diffs in these tests when
EarlyCSE is improved to use MemorySSA and can do more load elimination.
Reviewers: hfinkel
Subscribers: mcrosier, llvm-commits
Differential Revision: http://reviews.llvm.org/D20238
llvm-svn: 271553
The DIType* for void is the null pointer. A null DIType can never be a
qualified type, so we can just exit the loop at this point and go to
getTypeIndex(BaseTy).
Fixes PR27984
llvm-svn: 271550
Summary:
In PR29973 Sanjay Patel reported an assertion failure when a certain
loop was optimized, for a target without SSE2 support. It turned out
this was because of the AVG pattern detection introduced in rL253952.
Prevent the assertion failure by bailing out early in
`detectAVGPattern()`, if the target does not support SSE2.
Also add a minimized test case.
Reviewers: congh, eli.friedman, spatel
Subscribers: emaste, llvm-commits
Differential Revision: http://reviews.llvm.org/D20905
llvm-svn: 271548
Also fix slice wrappers drop_front and drop_back.
The unittests are pretty awkward, but do the job; alternatives
welcome!
..and yes, I do have ArrayRefs with more than 4 billion elements.
llvm-svn: 271546
Do not issue lexing errors found during the parsing of macro body
definitions and parseIdentifier function in AsmParser. This changes the
Parser to not issue a lexing error when we reach an error, but rather
when it is consumed allowing us time to examine and recover from an
error.
As a result, of this, we stop issuing a both lexing error and a parsing
error in floating-literals test. Minor tweak to parseDirectiveRealValue
to favor more meaningful lexing error over less helpful parse error.
Reviewers: rnk, majnemer
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D20535
llvm-svn: 271542
