The warning isn't very useful when the function is an ObjC method.
rdar://problem/41561853
Differential Revision: https://reviews.llvm.org/D61147
llvm-svn: 359864
rL358749 added a documentation page in the Markdown format. Currently,
such pages are ignored in the configuration script for manual pages.
This patch fixes that.
Differential Revision: https://reviews.llvm.org/D60964
llvm-svn: 359860
We use to incorrectly use the store size instead of the alloc size when
creating the stack slot for allocas.
On aarch64 this can be demonstrated by allocating weirdly sized types.
For instance, in the added test case, we use an alloca for i19. We used
to allocate a slot of size 24-bit (19 rounded up to the next byte),
whereas we really want to use a full 32-bit slot for this type.
llvm-svn: 359856
Because diagnostics and their notes are not connected at the API level,
if the error message for an overload is emitted, then the overload
candidates are completed - if a diagnostic is emitted during that work,
the notes related to overload candidates would be attached to the latter
diagnostic, not the original error. Sort of worse, if the latter
diagnostic was disabled, the notes are disabled.
Reviewers: rsmith
Differential Revision: https://reviews.llvm.org/D61357
llvm-svn: 359854
Summary:
While the generic ABI requires notes to be 8-byte aligned in ELF64, many
vendor-specific notes (from Linux, NetBSD, Solaris, etc) use 4-byte
alignment.
In a PT_NOTE segment, if 4-byte aligned notes are followed by an 8-byte
aligned note, the possible 4-byte padding may make consumers fail to
parse the 8-byte aligned note. See PR41000 for a recent report about
.note.gnu.property (NT_GNU_PROPERTY_TYPE_0).
(Note, for NT_GNU_PROPERTY_TYPE_0, the consumers should probably migrate
to PT_GNU_PROPERTY, but the alignment issue affects other notes as well.)
To fix the issue, don't mix notes with different alignments in one
PT_NOTE. If compilers emit 4-byte aligned notes before 8-byte aligned
notes, we'll create at most 2 segments.
sh_size%sh_addralign=0 is actually implied by the rule for linking
unrecognized sections (in generic ABI), so we don't have to check that.
Notes that match in name, type and attribute flags are concatenated into
a single output section. The compilers have to ensure
sh_size%sh_addralign=0 to make concatenated notes parsable.
An alternative approach is to create a PT_NOTE for each SHT_NOTE, but
we'll have to incur the sizeof(Elf64_Phdr)=56 overhead every time a new
note section is introduced.
Reviewers: ruiu, jakehehrlich, phosek, jhenderson, pcc, espindola
Subscribers: emaste, arichardson, krytarowski, fedor.sergeev, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D61296
llvm-svn: 359853
The primary fix here is to WinException.cpp: we need to exclude jump
tables when computing the length of a function, or else we fail to
correctly compute the length. (We can only compute the number of bytes
consumed by certain assembler directives after the entire file is
parsed. ".p2align" is one of those directives, and is used by jump table
generation.)
The secondary fix, to MCWin64EH, is to make sure we don't silently
miscompile if we hit a similar situation in the future.
It's possible we could extend ARM64EmitUnwindInfo so it allows function
bodies that contain assembler directives, but that's a lot more
complicated; see the FIXME in MCWin64EH.cpp.
Fixes https://bugs.llvm.org/show_bug.cgi?id=41581 .
Differential Revision: https://reviews.llvm.org/D61095
llvm-svn: 359849
logging messages that are written the same, making it difficult to
know for certain which code path was taken based on a logfile. Add
some words to make each unique.
Right now the ordering for finding a FullUnwindPlan (ignoring
fallback unwind plan logic) is
1. If this is a _sigtramp like function, try eh_frame which is
hand written on darwin systems to account for finding the
saved register context correctly.
2. Ask the DynamicLoader if eh_frame should be preferred for
this frame. Some binaries on the system may have hand-written
eh_frame and the DynamicLoader is the source for this. (primarily
this is for hand-written assembly in the objc runtime, and we tell
lldb to trust that for functions in libobjc.dylib.)
3. if 0th frame, use GetUnwindPlanAtNonCallSite plan.
4. GetUnwindPlanAtCallSite {for 0th or any other}
5. GetUnwindPlanAtNonCallSite {now for non-0th frames, only if not from a compiler? hm.}
6. GetUnwindPlanArchitectureDefaultAtFunctionEntry if we're on the first instruction
7. Architectural default unwind plan ABI::CreateDefaultUnwindPlan
I'm moving #6 -- DefaultAtFunctionEntry -- up to between #3 and #4,
where we're already doing things specific to the zeroth frame. If
we're on the zeroth frame and the GetUnwindPlanAtNonCallSite plan
has failed for some reason, and we're on the first instruction, we
should definitely use DefaultAtFunctionEntry instead of any other
unwind plan. If we're trying to step out of some rando function
on the system that we couldn't assembly instruction inspect, this
is sufficient for us to step out of it.
llvm-svn: 359847
Summary:
Originally the insertDef method was only used when building MemorySSA, and was limiting the number of Phi nodes that it created.
Now it's used for updates as well, and it can create additional Phis needed for correctness.
Make sure no Phis are created in unreachable blocks (condition met during MSSA build), otherwise the renamePass will find a null DTNode.
Resolves PR41640.
Reviewers: george.burgess.iv
Subscribers: jlebar, Prazek, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D61410
llvm-svn: 359845
__builtin_constant_p.
If the operand of __builtin_constant_p is not constant and has
side-effects, then code controlled by a branch on it is unreachable and
we should not emit runtime behavior warnings in such code.
llvm-svn: 359844
lldb has an expression that runs in the inferior process to collect
the isa values and hash of the class names for classes in the
system's shared cache. In recent OSes, swift classes are in this
table and the function the jitted expression calls returns demangled
names. We need to compute the hashes based on the mangled names.
So for these names, return a hash value of 0 which indicates that
lldb should read the class name directly out of the runtime tables
and compute the hash itself.
When this patch is absent, the lldb+swift testsuite has many failures
on a recent macOS system; there isn't a direct non-swift way to
test for this being correct.
<rdar://problem/47935062>
llvm-svn: 359843
Summary: Create a method to clean up multiple potentially trivial phis, since we will need this often.
Reviewers: george.burgess.iv
Subscribers: jlebar, Prazek, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D61471
llvm-svn: 359842
by respecting the "artificial" attribute on variables. Function
arguments that are artificial and useful to end-users are being
whitelisted by the language runtime.
<rdar://problem/45322477>
Differential Revision: https://reviews.llvm.org/D61451
llvm-svn: 359841
CUDA 10.1 tools deprecated some command line options.
fatbinary no longer needs --cuda.
Differential Revision: https://reviews.llvm.org/D61470
llvm-svn: 359838
The default impementation in the base class for TargetLowering::getRegForInlineAsmConstraint doesn't work for mask registers when the VT is a scalar type integer types since the only legal mask types are vXi1. So we end up just getting whatever the first register class that contains the register. Currently this appears to be VK1, but its really dependent on the order tablegen outputs the register classes.
Some code in the caller ends up looking up the type for this register class and find v1i1 then generates a copyfromreg from the physical k-register with the v1i1 type. Then it generates an any_extend from v1i1 to the scalar VT which isn't legal. This bad any_extend sticks around until isel where it selects a MOVZX32rr8 with a v1i1 input or maybe a i8 input. Not sure but eventually we pick up a copy from VK1 to GR8 in MachineIR which isn't supported. This leads to a failure in physical register copying.
This patch uses the scalar type to find a VK class of the right size. In the attached test case this will be VK16. This causes a bitcast from vk16 to i16 to be generated instead of an any_extend. This will be properly iseled to a VK16 to GR32 copy and a GR32->GR16 extract_subreg.
Fixes PR41678
Differential Revision: https://reviews.llvm.org/D61453
llvm-svn: 359837
As a result of the underlying cause of PR41678 we created an ANY_EXTEND node with a scalar result type and v1i1 input type. Ideally we would have asserted for this instead of letting it go through to instruction selection and generate bad machine IR
Differential Revision: https://reviews.llvm.org/D61463
llvm-svn: 359836
The debug server does not need to use the instruction emulation. This helps
reduce the size of the final lldb-server binary by another ~100K (~1% savings).
llvm-svn: 359832
As a side benefit, lld-link now reports more than one duplicate resource
entry before exiting with an error even if the new flag is not passed.
llvm-svn: 359829
Drive-by:
* Fix potential race between check and update of `throw_one` in `operator new`
* Fix latent bug in `operator delete`, which shouldn't decrement `outstanding_new` when passed a null pointer
* Specifically catch the expected `bad_alloc` in `main` instead of `...`
Differential Revision: https://reviews.llvm.org/D50860
llvm-svn: 359827
If an address_space attribute is defined in a macro, print the macro instead
when diagnosing a warning or error for incompatible pointers with different
address_spaces.
We allow this for all attributes (not just address_space), and for multiple
attributes declared in the same macro.
Differential Revision: https://reviews.llvm.org/D51329
llvm-svn: 359826
Summary:
Previously for the different purposes we need to get the active/common
parallel level and with full runtime we iterated over all the records to
calculate this level. Instead, we can used the warp-based parallel level
counters used in no-runtime mode.
Reviewers: grokos, gtbercea, kkwli0
Subscribers: guansong, jfb, jdoerfert, caomhin, openmp-commits
Tags: #openmp
Differential Revision: https://reviews.llvm.org/D61395
llvm-svn: 359822
r359717 added clang-check as a dep of check-clang unconditionally
because I had missed lit.local.cfg in test/Tooling.
Instead, only add clang-check to the tools if the analyzer is enabled,
since the build target only exists then, and since all tests using
clang-check are skipped when the analyzer is disabled.
Differential Revision: https://reviews.llvm.org/D61418
llvm-svn: 359820
The unwind tables (`.eh_frame`, `.arm.extab`) add a significant chunk of data to
the final binaries. These should not be needed normally, particularly when
exceptions are disabled. This enables shrinking `lldb-server` by ~18% (3 MiB)
when built with gold.
llvm-svn: 359819
This patch inverses the values returned by `addName` and
`lookup` methods of the class mentioned so that they
now return true on success and false on failure.
Also, it does minor code cleanup.
Differential revision: https://reviews.llvm.org/D61190
llvm-svn: 359818
`_MSC_VER` indiciates that you are building with MSVC, not that you are building
for Windows. Use `_WIN32` (which identifies Win32 and Win64).
llvm-svn: 359817
Summary:
This check seems unnecessary as we already assert the same condition above and also access `sc.comp_unit`
before this check.
Reviewers: aprantl
Reviewed By: aprantl
Subscribers: jdoerfert, lldb-commits
Tags: #lldb
Differential Revision: https://reviews.llvm.org/D61394
llvm-svn: 359813
Martin Storsjo