Summary:
Support for this option is needed for building Linux kernel.
This is a very frequently requested feature by kernel developers.
More details : https://lkml.org/lkml/2018/4/4/601
GCC option description for -fdelete-null-pointer-checks:
This Assume that programs cannot safely dereference null pointers,
and that no code or data element resides at address zero.
-fno-delete-null-pointer-checks is the inverse of this implying that
null pointer dereferencing is not undefined.
This feature is implemented in as the function attribute
"null-pointer-is-valid"="true".
This CL only adds the attribute on the function.
It also strips "nonnull" attributes from function arguments but
keeps the related warnings unchanged.
Corresponding LLVM change rL336613 already updated the
optimizations to not treat null pointer dereferencing
as undefined if the attribute is present.
Reviewers: t.p.northover, efriedma, jyknight, chandlerc, rnk, srhines, void, george.burgess.iv
Reviewed By: jyknight
Subscribers: drinkcat, xbolva00, cfe-commits
Differential Revision: https://reviews.llvm.org/D47894
llvm-svn: 337433
Currently clang set kernel calling convention for CUDA/HIP after
arranging function, which causes incorrect kernel function type since
it depends on calling convention.
This patch moves setting kernel convention before arranging
function.
Differential Revision: https://reviews.llvm.org/D47733
llvm-svn: 334457
Introduced CreateMemTempWithoutCast and CreateTemporaryAllocaWithoutCast to emit alloca
without casting to default addr space.
ActiveFlag is a temporary variable emitted for clean up. It is defined as AllocaInst* type and there is
a cast to AlllocaInst in SetActiveFlag. An alloca casted to generic pointer causes assertion in
SetActiveFlag.
Since there is only load/store of ActiveFlag, it is safe to use the original alloca, therefore use
CreateMemTempWithoutCast is called.
Differential Revision: https://reviews.llvm.org/D47099
llvm-svn: 332982
lifetime.start/end expects pointer argument in alloca address space.
However in C++ a temporary variable is in default address space.
This patch changes API CreateMemTemp and CreateTempAlloca to
get the original alloca instruction and pass it lifetime.start/end.
It only affects targets with non-zero alloca address space.
Differential Revision: https://reviews.llvm.org/D45900
llvm-svn: 332593
FunctionProtoType.
We previously re-evaluated the expression each time we wanted to know whether
the type is noexcept or not. We now evaluate the expression exactly once.
This is not quite "no functional change": it fixes a crasher bug during AST
deserialization where we would try to evaluate the noexcept specification in a
situation where we have not deserialized sufficient portions of the AST to
permit such evaluation.
llvm-svn: 331428
As suggested in the post-commit thread for rL331056, we should match these
clang options with the established vocabulary of the corresponding sanitizer
option. Also, the use of 'strict' is well-known for these kinds of knobs,
and we can improve the descriptive text in the docs.
So this intends to match the logic of D46135 but only change the words.
Matching LLVM commit to match this spelling of the attribute to follow shortly.
Differential Revision: https://reviews.llvm.org/D46236
llvm-svn: 331209
As discussed in the post-commit thread for:
rL330437 ( http://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20180423/545906.html )
We need a way to opt-out of a float-to-int-to-float cast optimization because too much
existing code relies on the platform-specific undefined result of those casts when the
float-to-int overflows.
The LLVM changes associated with adding this function attribute are here:
rL330947
rL330950
rL330951
Also as suggested, I changed the LLVM doc to mention the specific sanitizer flag that
catches this problem:
rL330958
Differential Revision: https://reviews.llvm.org/D46135
llvm-svn: 331041
function if a function delegates to another function.
Fix a bug introduced in r328731, which caused a struct with ObjC __weak
fields that was passed to a function to be destructed twice, once in the
callee function and once in another function the callee function
delegates to. To prevent this, keep track of the callee-destructed
structs passed to a function and disable their cleanups at the point of
the call to the delegated function.
This reapplies r331016, which was reverted in r331019 because it caused
an assertion to fail in EmitDelegateCallArg on a windows bot. I made
changes to EmitDelegateCallArg so that it doesn't try to deactivate
cleanups for structs that have trivial destructors (cleanups for those
structs are never pushed to the cleanup stack in EmitParmDecl).
rdar://problem/39194693
Differential Revision: https://reviews.llvm.org/D45382
llvm-svn: 331020
function if a function delegates to another function.
Fix a bug introduced in r328731, which caused a struct with ObjC __weak
fields that was passed to a function to be destructed twice, once in the
callee function and once in another function the callee function
delegates to. To prevent this, keep track of the callee-destructed
structs passed to a function and disable their cleanups at the point of
the call to the delegated function.
rdar://problem/39194693
Differential Revision: https://reviews.llvm.org/D45382
llvm-svn: 331016
have a non-trivial destructor.
This fixes a bug introduced in r328731 where CodeGen emits calls to
synthesized destructors for non-trivial C structs in C++ mode when the
struct passed to EmitCallArg doesn't have a non-trivial destructor.
Under Microsoft's ABI, ASTContext::isParamDestroyedInCallee currently
always returns true, so it's necessary to check whether the struct has a
non-trivial destructor before pushing a cleanup in EmitCallArg.
This fixes PR37146.
llvm-svn: 330304
the tail padding is not reused.
We track on the AggValueSlot (and through a couple of other
initialization actions) whether we're dealing with an object that might
share its tail padding with some other object, so that we can avoid
emitting stores into the tail padding if that's the case. We still
widen stores into tail padding when we can do so.
Differential Revision: https://reviews.llvm.org/D45306
llvm-svn: 329342
This reverts r328795 which introduced an issue with referencing __global__
function templates. More details in the original review D44747.
llvm-svn: 329099
Summary:
The following class hierarchy requires that we be able to emit a
this-adjusting thunk for B::foo in C's vftable:
struct Incomplete;
struct A {
virtual A* foo(Incomplete p) = 0;
};
struct B : virtual A {
void foo(Incomplete p) override;
};
struct C : B { int c; };
This TU is valid, but lacks a definition of 'Incomplete', which makes it
hard to build a thunk for the final overrider, B::foo.
Before this change, Clang gives up attempting to emit the thunk, because
it assumes that if the parameter types are incomplete, it must be
emitting the thunk for optimization purposes. This is untrue for the MS
ABI, where the implementation of B::foo has no idea what thunks C's
vftable may require. Clang needs to emit the thunk without necessarily
having access to the complete prototype of foo.
This change makes Clang emit a musttail variadic call when it needs such
a thunk. I call these "unprototyped" thunks, because they only prototype
the "this" parameter, which must always come first in the MS C++ ABI.
These thunks work, but they create ugly LLVM IR. If the call to the
thunk is devirtualized, it will be a call to a bitcast of a function
pointer. Today, LLVM cannot inline through such a call, but I want to
address that soon, because we also use this pattern for virtual member
pointer thunks.
This change also implements an old FIXME in the code about reusing the
thunk's computed CGFunctionInfo as much as possible. Now we don't end up
computing the thunk's mangled name and arranging it's prototype up to
around three times.
Fixes PR25641
Reviewers: rjmccall, rsmith, hans
Subscribers: Prazek, cfe-commits
Differential Revision: https://reviews.llvm.org/D45112
llvm-svn: 329009
This patch sets target specific calling convention for CUDA kernels in IR.
Patch by Greg Rodgers.
Revised and lit test added by Yaxun Liu.
Differential Revision: https://reviews.llvm.org/D44747
llvm-svn: 328795
ObjC and ObjC++ pass non-trivial structs in a way that is incompatible
with each other. For example:
typedef struct {
id f0;
__weak id f1;
} S;
// this code is compiled in c++.
extern "C" {
void foo(S s);
}
void caller() {
// the caller passes the parameter indirectly and destructs it.
foo(S());
}
// this function is compiled in c.
// 'a' is passed directly and is destructed in the callee.
void foo(S a) {
}
This patch fixes the incompatibility by passing and returning structs
with __strong or weak fields using the C ABI in C++ mode. __strong and
__weak fields in a struct do not cause the struct to be destructed in
the caller and __strong fields do not cause the struct to be passed
indirectly.
Also, this patch fixes the microsoft ABI bug mentioned here:
https://reviews.llvm.org/D41039?id=128767#inline-364710
rdar://problem/38887866
Differential Revision: https://reviews.llvm.org/D44908
llvm-svn: 328731
The patch adds nocf_check target independent attribute for disabling checks that were enabled by cf-protection flag.
The attribute can be appertained to functions and function pointers.
Attribute name follows GCC's similar attribute name.
Differential Revision: https://reviews.llvm.org/D41880
llvm-svn: 327768
Before this, we'd only emit lifetime.ends for these temps in
non-exceptional paths. This potentially made our stack larger than it
needed to be for any code that follows an EH cleanup. e.g. in
```
struct Foo { char cs[32]; };
void escape(void *);
struct Bar { ~Bar() { char cs[64]; escape(cs); } };
Foo getFoo();
void baz() {
Bar b;
getFoo();
}
```
baz() would require 96 bytes of stack, since the temporary from getFoo()
only had a lifetime.end on the non-exceptional path.
This also makes us keep hold of the Value* returned by
EmitLifetimeStart, so we don't have to remake it later.
llvm-svn: 326988
The indirect function argument is in alloca address space in LLVM IR. However,
during Clang codegen for C++, the address space of indirect function argument
should match its address space in the source code, i.e., default addr space, even
for indirect argument. This is because destructor of the indirect argument may
be called in the caller function, and address of the indirect argument may be
taken, in either case the indirect function argument is expected to be in default
addr space, not the alloca address space.
Therefore, the indirect function argument should be mapped to the temp var
casted to default address space. The caller will cast it to alloca addr space
when passing it to the callee. In the callee, the argument is also casted to the
default address space and used.
CallArg is refactored to facilitate this fix.
Differential Revision: https://reviews.llvm.org/D34367
llvm-svn: 326946
EmitLifetimeStart returns a non-null `size` pointer if it actually
emits a lifetime.start. Later in this function, we use `tempSize`'s
nullness to determine whether or not we should emit a lifetime.end.
llvm-svn: 326844
The patch fixes a number of bugs related to parameter indexing in
attributes:
* Parameter indices in some attributes (argument_with_type_tag,
pointer_with_type_tag, nonnull, ownership_takes, ownership_holds,
and ownership_returns) are specified in source as one-origin
including any C++ implicit this parameter, were stored as
zero-origin excluding any this parameter, and were erroneously
printing (-ast-print) and confusingly dumping (-ast-dump) as the
stored values.
* For alloc_size, the C++ implicit this parameter was not subtracted
correctly in Sema, leading to assert failures or to silent failures
of __builtin_object_size to compute a value.
* For argument_with_type_tag, pointer_with_type_tag, and
ownership_returns, the C++ implicit this parameter was not added
back to parameter indices in some diagnostics.
This patch fixes the above bugs and aims to prevent similar bugs in
the future by introducing careful mechanisms for handling parameter
indices in attributes. ParamIdx stores a parameter index and is
designed to hide the stored encoding while providing accessors that
require each use (such as printing) to make explicit the encoding that
is needed. Attribute declarations declare parameter index arguments
as [Variadic]ParamIdxArgument, which are exposed as ParamIdx[*]. This
patch rewrites all attribute arguments that are processed by
checkFunctionOrMethodParameterIndex in SemaDeclAttr.cpp to be declared
as [Variadic]ParamIdxArgument. The only exception is xray_log_args's
argument, which is encoded as a count not an index.
Differential Revision: https://reviews.llvm.org/D43248
llvm-svn: 326602
This makes it easier to debug crashes and hangs in block functions since
users can easily find out where the block is called from. The option
doesn't disable tail-calls from non-escaping blocks since non-escaping
blocks are not as hard to debug as escaping blocks.
rdar://problem/35758207
Differential Revision: https://reviews.llvm.org/D43841
llvm-svn: 326530
objc_msgSend_stret takes a hidden parameter for the returned structure's
address for the construction. When the function signature is rewritten
for the inalloca passing, the return type is no longer marked as
indirect but rather inalloca stret. This enhances the test for the
indirect return to check for that case as well. This fixes the
incorrect return classification for Windows x86.
llvm-svn: 326362
ARC mode.
Declaring __strong pointer fields in structs was not allowed in
Objective-C ARC until now because that would make the struct non-trivial
to default-initialize, copy/move, and destroy, which is not something C
was designed to do. This patch lifts that restriction.
Special functions for non-trivial C structs are synthesized that are
needed to default-initialize, copy/move, and destroy the structs and
manage the ownership of the objects the __strong pointer fields point
to. Non-trivial structs passed to functions are destructed in the callee
function.
rdar://problem/33599681
Differential Revision: https://reviews.llvm.org/D41228
llvm-svn: 326307
Summary:
OpenCL 2.0 specification defines '-cl-uniform-work-group-size' option,
which requires that the global work-size be a multiple of the work-group
size specified to clEnqueueNDRangeKernel and allows optimizations that
are made possible by this restriction.
The patch introduces the support of this option.
To keep information about whether an OpenCL kernel has uniform work
group size or not, clang generates 'uniform-work-group-size' function
attribute for every kernel:
- "uniform-work-group-size"="true" for OpenCL 1.2 and lower,
- "uniform-work-group-size"="true" for OpenCL 2.0 and higher if
'-cl-uniform-work-group-size' option was specified,
- "uniform-work-group-size"="false" for OpenCL 2.0 and higher if no
'-cl-uniform-work-group-size' options was specified.
If the function is not an OpenCL kernel, 'uniform-work-group-size'
attribute isn't generated.
Patch by: krisb
Reviewers: yaxunl, Anastasia, b-sumner
Reviewed By: yaxunl, Anastasia
Subscribers: nhaehnle, yaxunl, Anastasia, cfe-commits
Differential Revision: https://reviews.llvm.org/D43570
llvm-svn: 325771
As reported here: https://bugs.llvm.org/show_bug.cgi?id=36301
The issue is that the 'use' causes the plain declaration to emit
the attributes to LLVM-IR. However, if the definition added it
later, these would silently disappear.
This commit extracts that logic to its own function in CodeGenModule,
and has the attribute-applications done during 'definition' update
the attributes properly.
Differential Revision: https://reviews.llvm.org/D43095
llvm-svn: 324907
Summary:
Fixes PR36247, which is where WinEHPrepare replaces inline asm in
funclets with unreachable.
Make getBundlesForFunclet return by value to simplify some call sites.
Reviewers: smeenai, majnemer
Subscribers: eraman, cfe-commits
Differential Revision: https://reviews.llvm.org/D43033
llvm-svn: 324689
This change reduces the live range of the loaded function pointer,
resulting in a slight code size decrease (~10KB in clang), and also
improves the security of CFI for virtual calls by making it less
likely that the function pointer will be spilled, and ensuring that
it is not spilled across a function call boundary.
Fixes PR35353.
Differential Revision: https://reviews.llvm.org/D42725
llvm-svn: 324286
The 'trivial_abi' attribute can be applied to a C++ class, struct, or
union. It makes special functions of the annotated class (the destructor
and copy/move constructors) to be trivial for the purpose of calls and,
as a result, enables the annotated class or containing classes to be
passed or returned using the C ABI for the underlying type.
When a type that is considered trivial for the purpose of calls despite
having a non-trivial destructor (which happens only when the class type
or one of its subobjects is a 'trivial_abi' class) is passed to a
function, the callee is responsible for destroying the object.
For more background, see the discussions that took place on the mailing
list:
http://lists.llvm.org/pipermail/cfe-dev/2017-November/055955.htmlhttp://lists.llvm.org/pipermail/cfe-commits/Week-of-Mon-20180101/thread.html#214043
rdar://problem/35204524
Differential Revision: https://reviews.llvm.org/D41039
llvm-svn: 324269
When a function taking transparent union is declared as taking one of
union members earlier in the translation unit, clang would hit an
"Invalid cast" assertion during EmitFunctionProlog. This case
corresponds to function f1 in test/CodeGen/transparent-union-redecl.c.
We decided to cast i32 to union because after merging function
declarations function parameter type becomes int,
CGFunctionInfo::ArgInfo type matches with ABIArgInfo type, so we decide
it is a trivial case. But these types should also be castable to
parameter declaration type which is not the case here.
Now the fix is in converting from ABIArgInfo type to VarDecl type and using
argument demotion when necessary.
Additional tests in Sema/transparent-union.c capture current behavior and make
sure there are no regressions.
rdar://problem/34949329
Reviewers: rjmccall, rafael
Reviewed By: rjmccall
Subscribers: aemerson, cfe-commits, kristof.beyls, ahatanak
Differential Revision: https://reviews.llvm.org/D41311
llvm-svn: 323156
As @rjmccall suggested in D40023, we can get rid of
ABIInfo::shouldSignExtUnsignedType (used to handle cases like the Mips calling
convention where 32-bit integers are always sign extended regardless of the
sign of the type) by adding a SignExt field to ABIArgInfo. In the common case,
this new field is set automatically by ABIArgInfo::getExtend based on the sign
of the type. For targets that want greater control, they can use
ABIArgInfo::getSignExtend or ABIArgInfo::getZeroExtend when necessary. This
change also cleans up logic in CGCall.cpp.
There is no functional change intended in this patch, and all tests pass
unchanged. As noted in D40023, Mips might want to sign-extend unsigned 32-bit
integer return types. A future patch might modify
MipsABIInfo::classifyReturnType to use MipsABIInfo::extendType.
Differential Revision: https://reviews.llvm.org/D41999
llvm-svn: 322396
When a function taking transparent union is declared as taking one of
union members earlier in the translation unit, clang would hit an
"Invalid cast" assertion during EmitFunctionProlog. This case
corresponds to function f1 in test/CodeGen/transparent-union-redecl.c.
We decided to cast i32 to union because after merging function
declarations function parameter type becomes int,
CGFunctionInfo::ArgInfo type matches with ABIArgInfo type, so we decide
it is a trivial case. But these types should also be castable to
parameter declaration type which is not the case here.
The fix is in checking for the trivial case if ABIArgInfo type matches with
parameter declaration type. It exposed inconsistency that we check
hasScalarEvaluationKind for different types in EmitParmDecl and
EmitFunctionProlog, and comment says they should match.
Additional tests in Sema/transparent-union.c capture current behavior and make
sure there are no regressions.
rdar://problem/34949329
Reviewers: rjmccall, rafael
Reviewed By: rjmccall
Subscribers: aemerson, cfe-commits, kristof.beyls
Differential Revision: https://reviews.llvm.org/D41311
llvm-svn: 321296
Diagnose 'unreachable' UB when a noreturn function returns.
1. Insert a check at the end of functions marked noreturn.
2. A decl may be marked noreturn in the caller TU, but not marked in
the TU where it's defined. To diagnose this scenario, strip away the
noreturn attribute on the callee and insert check after calls to it.
Testing: check-clang, check-ubsan, check-ubsan-minimal, D40700
rdar://33660464
Differential Revision: https://reviews.llvm.org/D40698
llvm-svn: 321231
This adds a new command line option -mprefer-vector-width to specify a preferred vector width for the vectorizers. Valid values are 'none' and unsigned integers. The driver will check that it meets those constraints. Specific supported integers will be managed by the targets in the backend.
Clang will take the value and add it as a new function attribute during CodeGen.
This represents the alternate direction proposed by Sanjay in this RFC: http://lists.llvm.org/pipermail/llvm-dev/2017-November/118734.html
The syntax here matches gcc, though gcc treats it as an x86 specific command line argument. gcc only allows values of 128, 256, and 512. I'm not having clang check any values.
Differential Revision: https://reviews.llvm.org/D40230
llvm-svn: 320419
CreateCoercedLoad/CreateCoercedStore assumes pointer argument of
memcpy is in addr space 0, which is not correct and causes invalid
bitcasts for triple amdgcn---amdgiz.
It is fixed by using alloca addr space instead.
Differential Revision: https://reviews.llvm.org/D40806
llvm-svn: 320000
Currently CodeGen is calling std::sort on the features vector in TargetOptions for every function, but I don't think CodeGen should be modifying TargetOptions.
Differential Revision: https://reviews.llvm.org/D40228
llvm-svn: 319195
Craig noticed that CodeGen wasn't properly ignoring the
values sent to the target attribute. This patch ignores
them.
This patch also sets the 'default' for this checking to
'supported', since only X86 has implemented the support
for checking valid CPU names and Feature Names.
One test was changed to i686, since it uses a lakemont,
which would otherwise be prohibited in x86_64.
Differential Revision: https://reviews.llvm.org/D39357
llvm-svn: 316783
This was done for CUDA functions in r261779, and for the same
reason this also needs to be done for OpenCL. An arbitrary
function could have a barrier() call in it, which in turn
requires the calling function to be convergent.
llvm-svn: 315094
The attribute informs the compiler that the annotated pointer parameter
of a function cannot escape and enables IRGen to attach attribute
'nocapture' to parameters that are annotated with the attribute. That is
the only optimization that currently takes advantage of 'noescape', but
there are other optimizations that will be added later that improves
IRGen for ObjC blocks.
This recommits r313722, which was reverted in r313725 because clang
couldn't build compiler-rt. It failed to build because there were
function declarations that were missing 'noescape'. That has been fixed
in r313929.
rdar://problem/19886775
Differential Revision: https://reviews.llvm.org/D32210
llvm-svn: 313945
This reverts commit r313722.
It looks like compiler-rt/lib/tsan/rtl/tsan_libdispatch_mac.cc cannot be
compiled because some of the functions declared in the file do not match
the ones in the SDK headers (which are annotated with 'noescape').
llvm-svn: 313725
The attribute informs the compiler that the annotated pointer parameter
of a function cannot escape and enables IRGen to attach attribute
'nocapture' to parameters that are annotated with the attribute. That is
the only optimization that currently takes advantage of 'noescape', but
there are other optimizations that will be added later that improves
IRGen for ObjC blocks.
rdar://problem/19886775
Differential Revision: https://reviews.llvm.org/D32210
llvm-svn: 313722
The attribute informs the compiler that the annotated pointer parameter
of a function cannot escape and enables IRGen to attach attribute
'nocapture' to parameters that are annotated with the attribute. That is
the only optimization that currently takes advantage of 'noescape', but
there are other optimizations that will be added later that improves
IRGen for ObjC blocks.
rdar://problem/19886775
Differential Revision: https://reviews.llvm.org/D32520
llvm-svn: 313720
In a future commit AMDGPU will start passing
aggregates directly to more functions, triggering
asserts in test/CodeGenOpenCL/addr-space-struct-arg.cl
llvm-svn: 309741
Convert attribute 'target' parsing from a 'pair' to a 'struct' to make further
improvements easier
The attribute 'target' parse function previously returned a pair. Convert
this to a 'pair' in order to add more functionality, and improve usability.
Differential Revision: https://reviews.llvm.org/D35574
llvm-svn: 308357
Move builtins from the x86 specific scope into the global
scope. Their use is still limited to x86_64 and aarch64 though.
This allows wine on aarch64 to properly handle variadic functions.
Differential Revision: https://reviews.llvm.org/D34475
llvm-svn: 308218
Clang assumes coerced function argument is in address space 0, which is not always true and results in invalid bitcasts.
This patch fixes failure in OpenCL conformance test api/get_kernel_arg_info with amdgcn---amdgizcl triple, where non-zero alloca address space is used.
Differential Revision: https://reviews.llvm.org/D34777
llvm-svn: 306721
The assertion was failing when a method of a parameterized class was
called and the types of the argument and parameter didn't match. To fix
the failure, move the assertion in EmitCallArg to its only caller
EmitCallArgs and require the argument and parameter types match only
when the method is not parameterized.
rdar://problem/32874473
Differential Revision: https://reviews.llvm.org/D34665
llvm-svn: 306494
This patch makes ubsan's nonnull return value diagnostics more precise,
which makes the diagnostics more useful when there are multiple return
statements in a function. Example:
1 |__attribute__((returns_nonnull)) char *foo() {
2 | if (...) {
3 | return expr_which_might_evaluate_to_null();
4 | } else {
5 | return another_expr_which_might_evaluate_to_null();
6 | }
7 |} // <- The current diagnostic always points here!
runtime error: Null returned from Line 7, Column 2!
With this patch, the diagnostic would point to either Line 3, Column 5
or Line 5, Column 5.
This is done by emitting source location metadata for each return
statement in a sanitized function. The runtime is passed a pointer to
the appropriate metadata so that it can prepare and deduplicate reports.
Compiler-rt patch (with more tests): https://reviews.llvm.org/D34298
Differential Revision: https://reviews.llvm.org/D34299
llvm-svn: 306163
In C++ all variables are in default address space. Previously change has been
made to cast automatic variables to default address space. However that is
not sufficient since all temporary variables need to be casted to default
address space.
This patch casts all temporary variables to default address space except those
for passing indirect arguments since they are only used for load/store.
This patch only affects target having non-zero alloca address space.
Differential Revision: https://reviews.llvm.org/D33706
llvm-svn: 305711
Rationale: OpenCL kernels are called via an explicit runtime API
with arguments set with clSetKernelArg(), not as normal sub-functions.
Return SPIR_KERNEL by default as the kernel calling convention to ensure
the fingerprint is fixed such way that each OpenCL argument gets one
matching argument in the produced kernel function argument list to enable
feasible implementation of clSetKernelArg() with aggregates etc. In case
we would use the default C calling conv here, clSetKernelArg() might
break depending on the target-specific conventions; different targets
might split structs passed as values to multiple function arguments etc.
https://reviews.llvm.org/D33639
llvm-svn: 304389
This should simplify the call sites, which typically want to tweak one
attribute at a time. It should also avoid creating ephemeral
AttributeLists that live forever.
llvm-svn: 300718
GCC has the alloc_align attribute, which is similar to assume_aligned, except the attribute's parameter is the index of the integer parameter that needs aligning to.
Differential Revision: https://reviews.llvm.org/D29599
llvm-svn: 299117
attributes.
These patches don't work because we can't currently access the parameter
information in a reliable way when building attributes. I thought this
would be relatively straightforward to fix, but it seems not to be the
case. Fixing this will requrie a substantial re-plumbing of machinery to
allow attributes to be handled in this location, and several other fixes
to the attribute machinery should probably be made at the same time. All
of this will make the patch .... substantially more complicated.
Reverting for now as there are active miscompiles caused by the current
version.
llvm-svn: 298695
explaining why we have to ignore errors here even though in other parts
of codegen we can be more strict with builtins.
Also add a test case based on the code in a TSan test that found this
issue.
llvm-svn: 298494
declarations and calls instead of just definitions, and then teach it to
*not* attach such attributes even if the source code contains them.
This follows the design direction discussed on cfe-dev here:
http://lists.llvm.org/pipermail/cfe-dev/2017-January/052066.html
The idea is that for C standard library builtins, even if the library
vendor chooses to annotate their routines with __attribute__((nonnull)),
we will ignore those attributes which pertain to pointer arguments that
have an associated size. This allows the widespread (and seemingly
reasonable) pattern of calling these routines with a null pointer and
a zero size. I have only done this for the library builtins currently
recognized by Clang, but we can now trivially add to this set. This will
be controllable with -fno-builtin if anyone should care to do so.
Note that this does *not* change the AST. As a consequence, warnings,
static analysis, and source code rewriting are not impacted.
This isn't even a regression on any platform as neither Clang nor LLVM
have ever put 'nonnull' onto these arguments for declarations. All this
patch does is enable it on other declarations while preventing us from
ever accidentally enabling it on these libc functions due to a library
vendor.
It will also allow any other libraries using this annotation to gain
optimizations based on the annotation even when only a declaration is
visible.
llvm-svn: 298491
This is a follow-up to r297700 (Add a nullability sanitizer).
It addresses some FIXME's re: using nullability-specific diagnostic
handlers from compiler-rt, now that the necessary handlers exist.
check-ubsan test updates to follow.
llvm-svn: 297750
Teach UBSan to detect when a value with the _Nonnull type annotation
assumes a null value. Call expressions, initializers, assignments, and
return statements are all checked.
Because _Nonnull does not affect IRGen, the new checks are disabled by
default. The new driver flags are:
-fsanitize=nullability-arg (_Nonnull violation in call)
-fsanitize=nullability-assign (_Nonnull violation in assignment)
-fsanitize=nullability-return (_Nonnull violation in return stmt)
-fsanitize=nullability (all of the above)
This patch builds on top of UBSan's existing support for detecting
violations of the nonnull attributes ('nonnull' and 'returns_nonnull'),
and relies on the compiler-rt support for those checks. Eventually we
will need to update the diagnostic messages in compiler-rt (there are
FIXME's for this, which will be addressed in a follow-up).
One point of note is that the nullability-return check is only allowed
to kick in if all arguments to the function satisfy their nullability
preconditions. This makes it necessary to emit some null checks in the
function body itself.
Testing: check-clang and check-ubsan. I also built some Apple ObjC
frameworks with an asserts-enabled compiler, and verified that we get
valid reports.
Differential Revision: https://reviews.llvm.org/D30762
llvm-svn: 297700
UBSan's nonnull argument check applies when a parameter has the
"nonnull" attribute. The check currently works for FunctionDecls, but
not for ObjCMethodDecls. This patch extends the check to work for ObjC.
Differential Revision: https://reviews.llvm.org/D30599
llvm-svn: 296996
The goal of this is to fix a bug in modules where we'd merge
FunctionDecls that differed in their pass_object_size attributes. Since
we can overload on the presence of pass_object_size attributes, this
behavior is incorrect.
We don't represent `N` in `pass_object_size(N)` as part of
ExtParameterInfo, since it's an error to overload solely on the value of
N. This means that we have a bug if we have two modules that declare
functions that differ only in their pass_object_size attrs, like so:
// In module A, from a.h
void foo(char *__attribute__((pass_object_size(0))));
// In module B, from b.h
void foo(char *__attribute__((pass_object_size(1))));
// In module C, in main.c
#include "a.h"
#include "b.h"
At the moment, we'll merge the foo decls, when we should instead emit a
diagnostic about an invalid overload. We seem to have similar (silent)
behavior if we overload only on the return type of `foo` instead; I'll
try to find a good place to put a FIXME (or I'll just file a bug) soon.
This patch also fixes a bug where we'd not output the proper extended
parameter info for declarations with pass_object_size attrs.
llvm-svn: 296076
This patch makes use of the prefix/suffix ABI argument distinction that
was introduced in r295870, so that we now emit ExtParameterInfo at the
correct offset for member calls that have added ABI arguments. I don't
see a good way to test the generated param info, since we don't actually
seem to use it in CGFunctionInfo outside of Swift. Any
suggestions/thoughts for how to better test this are welcome. :)
This patch also fixes a small bug with inheriting constructors: if we
decide not to pass args into an base class ctor, we would still
generate ExtParameterInfo as though we did. The added test-case is for
that behavior.
llvm-svn: 296024
This fixes an assertion failure in cases where we had expression
statements that declared variables nested inside of pass_object_size
args. Since we were emitting the same ExprStmt twice (once for the arg,
once for the @llvm.objectsize call), we were getting issues with
redefining locals.
This also means that we can be more lax about when we emit
@llvm.objectsize for pass_object_size args: since we're reusing the
arg's value itself, we don't have to care so much about side-effects.
llvm-svn: 295935
Summary:
Now when you ask clang to link in a bitcode module, you can tell it to
set attributes on that module's functions to match what we would have
set if we'd emitted those functions ourselves.
This is particularly important for fast-math attributes in CUDA
compilations.
Each CUDA compilation links in libdevice, a bitcode library provided by
nvidia as part of the CUDA distribution. Without this patch, if we have
a user-function F that is compiled with -ffast-math that calls a
function G from libdevice, F will have the unsafe-fp-math=true (etc.)
attributes, but G will have no attributes.
Since F calls G, the inliner will merge G's attributes into F's. It
considers the lack of an unsafe-fp-math=true attribute on G to be
tantamount to unsafe-fp-math=false, so it "merges" these by setting
unsafe-fp-math=false on F.
This then continues up the call graph, until every function that
(transitively) calls something in libdevice gets unsafe-fp-math=false
set, thus disabling fastmath in almost all CUDA code.
Reviewers: echristo
Subscribers: hfinkel, llvm-commits, mehdi_amini
Differential Revision: https://reviews.llvm.org/D28538
llvm-svn: 293097
This is a recommit of r290149, which was reverted in r290169 due to msan
failures. msan was failing because we were calling
`isMostDerivedAnUnsizedArray` on an invalid designator, which caused us
to read uninitialized memory. To fix this, the logic of the caller of
said function was simplified, and we now have a `!Invalid` assert in
`isMostDerivedAnUnsizedArray`, so we can catch this particular bug more
easily in the future.
Fingers crossed that this patch sticks this time. :)
Original commit message:
This patch does three things:
- Gives us the alloc_size attribute in clang, which lets us infer the
number of bytes handed back to us by malloc/realloc/calloc/any user
functions that act in a similar manner.
- Teaches our constexpr evaluator that evaluating some `const` variables
is OK sometimes. This is why we have a change in
test/SemaCXX/constant-expression-cxx11.cpp and other seemingly
unrelated tests. Richard Smith okay'ed this idea some time ago in
person.
- Uniques some Blocks in CodeGen, which was reviewed separately at
D26410. Lack of uniquing only really shows up as a problem when
combined with our new eagerness in the face of const.
llvm-svn: 290297
This commit fails MSan when running test/CodeGen/object-size.c in
a confusing way. After some discussion with George, it isn't really
clear what is going on here. We can make the MSan failure go away by
testing for the invalid bit, but *why* things are invalid isn't clear.
And yet, other code in the surrounding area is doing precisely this and
testing for invalid.
George is going to take a closer look at this to better understand the
nature of the failure and recommit it, for now backing it out to clean
up MSan builds.
llvm-svn: 290169
This patch does three things:
- Gives us the alloc_size attribute in clang, which lets us infer the
number of bytes handed back to us by malloc/realloc/calloc/any user
functions that act in a similar manner.
- Teaches our constexpr evaluator that evaluating some `const` variables
is OK sometimes. This is why we have a change in
test/SemaCXX/constant-expression-cxx11.cpp and other seemingly
unrelated tests. Richard Smith okay'ed this idea some time ago in
person.
- Uniques some Blocks in CodeGen, which was reviewed separately at
D26410. Lack of uniquing only really shows up as a problem when
combined with our new eagerness in the face of const.
Differential Revision: https://reviews.llvm.org/D14274
llvm-svn: 290149
This adds a way for us to version any UBSan handler by itself.
The patch overrides D21289 for a better implementation (we're able to
rev up a single handler).
After this, then we can land a slight modification of D19667+D19668.
We probably don't want to keep all the versions in compiler-rt (maybe we
want to deprecate on one release and remove the old handler on the next
one?), but with this patch we will loudly fail to compile when mixing
incompatible handler calls, instead of silently compiling and then
providing bad error messages.
Reviewers: kcc, samsonov, rsmith, vsk
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D21695
llvm-svn: 289444
StartFunction enters a release cleanup for ns_consumed arguments in
ARC, so we need to balance that somehow. We could teach StartFunction
that it's emitting a delegating function, so that the cleanup is
unnecessary, but that would be invasive and somewhat fraught. We could
balance the consumed argument with an extra retain, but clearing the
original variable should be easier to optimize and avoid some extra work
at -O0. And there shouldn't be any difference as long as nothing else
uses the argument, which should always be true for the places we emit
delegate arguments.
Fixes PR 27887.
llvm-svn: 287291
This patch implements the register call calling convention, which ensures
as many values as possible are passed in registers. CodeGen changes
were committed in https://reviews.llvm.org/rL284108.
Differential Revision: https://reviews.llvm.org/D25204
llvm-svn: 285849
Certain OpenCL builtin functions are supposed to be executed by all threads in a work group or sub group. Such functions should not be made divergent during transformation. It makes sense to mark them with convergent attribute.
The adding of convergent attribute is based on Ettore Speziale's work and the original proposal and patch can be found at https://www.mail-archive.com/cfe-commits@lists.llvm.org/msg22271.html.
Differential Revision: https://reviews.llvm.org/D25343
llvm-svn: 285725
abstract information about the callee. NFC.
The goal here is to make it easier to recognize indirect calls and
trigger additional logic in certain cases. That logic will come in
a later patch; in the meantime, I felt that this was a significant
improvement to the code.
llvm-svn: 285258
Summary:
This prevents clang from emitting 'invoke's and catch statements.
Things previously mostly worked thanks to TryToMarkNoThrow() in
CodeGenFunction. But this is not a proper IPO, and it doesn't properly
handle cases like mutual recursion.
Fixes bug 30593.
Reviewers: tra
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D25166
llvm-svn: 283272
The motivation for the change is that we can't have pseudo-global settings
for codegen living in TargetOptions because that doesn't work with LTO.
Ideally, these reciprocal attributes will be moved to the instruction-level
via FMF, metadata, or something else. But making them function attributes is
at least an improvement over the current state.
I'm committing this patch ahead of the related LLVM patch to avoid bot failures,
but if that patch needs to be reverted, then this should be reverted too.
Differential Revision: https://reviews.llvm.org/D24815
llvm-svn: 283251
When ARC is enabled, an ObjCIndirectCopyRestoreExpr models the passing
of a function argument s.t:
* The argument is copied into a temporary,
* The temporary is passed into the function, and
* After the function call completes, the temporary is move-assigned
back to the original location of the argument.
The argument type and the parameter type must agree "except possibly in
qualification". This commit weakens an assertion in EmitCallArg() to
actually reflect that.
llvm-svn: 283116
Instead of ignoring the evaluation order rule, ignore the "destroy parameters
in reverse construction order" rule for the small number of problematic cases.
This only causes incorrect behavior in the rare case where both parameters to
an overloaded operator <<, >>, ->*, &&, ||, or comma are of class type with
non-trivial destructor, and the program is depending on those parameters being
destroyed in reverse construction order.
We could do a little better here by reversing the order of parameter
destruction for those functions (and reversing the argument evaluation order
for all direct calls, not just those with operator syntax), but that is not a
complete solution to the problem, as the same situation can be reached by an
indirect function call.
Approach reviewed off-line by rnk.
llvm-svn: 282777
function correctly when targeting MS ABIs (this appears to have never mattered
prior to this change).
Update test case to always cover both 32-bit and 64-bit Windows ABIs, since
they behave somewhat differently from each other here.
Update test case to also cover operators , && and ||, which it appears are also
affected by P0145R3 (they're not explicitly called out by the design document,
but this is the emergent behavior of the existing wording).
Original commit message:
P0145R3 (C++17 evaluation order tweaks): evaluate the right-hand side of
assignment and compound-assignment operators before the left-hand side. (Even
if it's an overloaded operator.)
This completes the implementation of P0145R3 + P0400R0 for all targets except
Windows, where the evaluation order guarantees for <<, >>, and ->* are
unimplementable as the ABI requires the function arguments are evaluated from
right to left (because parameter destructors are run from left to right in the
callee).
llvm-svn: 282619
assignment and compound-assignment operators before the left-hand side. (Even
if it's an overloaded operator.)
This completes the implementation of P0145R3 + P0400R0 for all targets except
Windows, where the evaluation order guarantees for <<, >>, and ->* are
unimplementable as the ABI requires the function arguments are evaluated from
right to left (because parameter destructors are run from left to right in the
callee).
llvm-svn: 282556
Move the logic for doing this from the ABI argument lowering into
EmitParmDecl, which runs for all parameters. Our codegen is slightly
suboptimal in this case, as we may leave behind a dead store after
optimization, but it's 32-bit inalloca, and this fixes the bug in a
robust way.
Fixes PR30293
llvm-svn: 280836
to CC1, which are translated to function attributes and can e.g. be mapped on
build attributes FP_exceptions and FP_denormal. Setting these build attributes
allows better selection of floating point libraries.
Differential Revision: https://reviews.llvm.org/D23840
llvm-svn: 280064
Let the driver pass the option to frontend. Do not set precision metadata for division instructions when this option is set. Set function attribute "correctly-rounded-divide-sqrt-fp-math" based on this option.
Differential Revision: https://reviews.llvm.org/D22940
llvm-svn: 278155
Add OCL option -cl-no-signed-zeros to driver options.
Also added to opencl.cl testcases.
Patch by Aaron En Ye Shi.
Differential Revision: http://reviews.llvm.org/D22067
llvm-svn: 274923
Summary:
Summary:
Change Clang calling convention SpirKernel to OpenCLKernel.
Set calling convention OpenCLKernel for amdgcn as well.
Add virtual method .getOpenCLKernelCallingConv() to TargetCodeGenInfo
and use it to set target calling convention for AMDGPU and SPIR.
Update tests.
Reviewers: rsmith, tstellarAMD, Anastasia, yaxunl
Subscribers: kzhuravl, cfe-commits
Differential Revision: http://reviews.llvm.org/D21367
llvm-svn: 274220
Replace inheriting constructors implementation with new approach, voted into
C++ last year as a DR against C++11.
Instead of synthesizing a set of derived class constructors for each inherited
base class constructor, we make the constructors of the base class visible to
constructor lookup in the derived class, using the normal rules for
using-declarations.
For constructors, UsingShadowDecl now has a ConstructorUsingShadowDecl derived
class that tracks the requisite additional information. We create shadow
constructors (not found by name lookup) in the derived class to model the
actual initialization, and have a new expression node,
CXXInheritedCtorInitExpr, to model the initialization of a base class from such
a constructor. (This initialization is special because it performs real perfect
forwarding of arguments.)
In cases where argument forwarding is not possible (for inalloca calls,
variadic calls, and calls with callee parameter cleanup), the shadow inheriting
constructor is not emitted and instead we directly emit the initialization code
into the caller of the inherited constructor.
Note that this new model is not perfectly compatible with the old model in some
corner cases. In particular:
* if B inherits a private constructor from A, and C uses that constructor to
construct a B, then we previously required that A befriends B and B
befriends C, but the new rules require A to befriend C directly, and
* if a derived class has its own constructors (and so its implicit default
constructor is suppressed), it may still inherit a default constructor from
a base class
llvm-svn: 274049
This patch fixes a bug where we'd segfault (in some cases) if we saw a
variadic function with one or more pass_object_size arguments.
Differential Revision: http://reviews.llvm.org/D17462
llvm-svn: 272971
pointer-to-pointer representing the parameter. An aggregate rvalue representing
a pointer does not make sense.
We got away with this weirdness because CGCall happens to blindly load an
RValue in aggregate form in this case, without checking whether an RValue for
the type should be in scalar or aggregate form.
llvm-svn: 272609
This option, like the corresponding gcc option, is SystemZ-specific and
enables storing frame backchain links, as specified in the ABI.
Differential Revision: http://reviews.llvm.org/D19891
llvm-svn: 268575
r268261 made Clang "expand" more struct arguments on Windows. It removed
the check for 'RD->isCLike()', which was preventing us from attempting
to expand structs with reference type fields.
Our expansion code was attempting to load and pass each field of the
type in turn. We were accidentally doing one to many loads on reference
type fields.
On the function prologue side, we can use
EmitLValueForFieldInitialization, which obviously gets the address of
the field. On the call side, I tweaked EmitRValueForField directly,
since this is the only use of this method.
Fixes PR27607
llvm-svn: 268321
Revert the two changes to thread CodeGenOptions into the TargetInfo allocation
and to fix the layering violation by moving CodeGenOptions into Basic.
Code Generation is arguably not particularly "basic". This addresses Richard's
post-commit review comments. This change purely does the mechanical revert and
will be followed up with an alternate approach to thread the desired information
into TargetInfo.
llvm-svn: 265806
This is a mechanical move of CodeGenOptions from libFrontend to libBasic. This
fixes the layering violation introduced earlier by threading CodeGenOptions into
TargetInfo. It should also fix the modules based self-hosting builds. NFC.
llvm-svn: 265702
Summary:
Setting this flag causes all functions are annotated with the
"nvvm-f32ftz" = "true" attribute.
In addition, we annotate the module with "nvvm-reflect-ftz" set
to 0 or 1, depending on whether -cuda-flush-denormals-to-zero is set.
This is read by the NVVMReflect pass.
Reviewers: tra, rnk
Subscribers: cfe-commits
Differential Revision: http://reviews.llvm.org/D18671
llvm-svn: 265435
to function names
Summary:
Hopefully this will make it easier for the next person to figure all
this out...
Reviewers: bogner, davidxl
Subscribers: davidxl, cfe-commits
Differential Revision: http://reviews.llvm.org/D18489
llvm-svn: 264681
Till now, preserve_mostcc/preserve_allcc calling convention attributes were only
available at the LLVM IR level. This patch adds attributes for
preserve_mostcc/preserve_allcc calling conventions to the C/C++ front-end.
The code was mostly written by Juergen Ributzka.
I just added support for the AArch64 target and tests.
Differential Revision: http://reviews.llvm.org/D18025
llvm-svn: 263647
Summary:
This flag is enabled by default in the driver when NDEBUG is set. It
is forwarded on the LLVMContext to discard all value names (but
GlobalValue) for performance purpose.
This an improved version of D18024
Reviewers: echristo, chandlerc
Subscribers: cfe-commits
Differential Revision: http://reviews.llvm.org/D18127
From: Mehdi Amini <mehdi.amini@apple.com>
llvm-svn: 263394
commit 60d9845f6a037122d9be9a6d92d4de617ef45b04
Author: Mehdi Amini <mehdi.amini@apple.com>
Date: Fri Mar 11 18:48:02 2016 +0000
Fix clang crash: when CodeGenAction is initialized without a
context, use the member and not the parameter
From: Mehdi Amini <mehdi.amini@apple.com>
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@263273
91177308-0d34-0410-b5e6-96231b3b80d8
commit af7ce3bf04a75ad5124b457b805df26006bd215b
Author: Mehdi Amini <mehdi.amini@apple.com>
Date: Fri Mar 11 17:32:58 2016 +0000
Fix build: use -> with pointers and not .
Silly typo.
From: Mehdi Amini <mehdi.amini@apple.com>
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@263267
91177308-0d34-0410-b5e6-96231b3b80d8
commit d0eea119192814954e7368c77d0dc5a9eeec1fbb
Author: Mehdi Amini <mehdi.amini@apple.com>
Date: Fri Mar 11 17:15:44 2016 +0000
Remove compile time PreserveName switch based on NDEBUG
Summary:
Following r263086, we are now relying on a flag on the Context to
discard Value names in release builds.
Reviewers: chandlerc
Subscribers: cfe-commits
Differential Revision: http://reviews.llvm.org/D18024
From: Mehdi Amini <mehdi.amini@apple.com>
git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@263257
91177308-0d34-0410-b5e6-96231b3b80d8
until we can fix the Release builds.
This reverts commits 263257, 263267, 263273
llvm-svn: 263320
Summary:
Following r263086, we are now relying on a flag on the Context to
discard Value names in release builds.
Reviewers: chandlerc
Subscribers: cfe-commits
Differential Revision: http://reviews.llvm.org/D18024
From: Mehdi Amini <mehdi.amini@apple.com>
llvm-svn: 263257
Summary:
This is important for e.g. the following case:
void sync() { __syncthreads(); }
void foo() {
do_something();
sync();
do_something_else():
}
Without this change, if the optimizer does not inline sync() (which it
won't because __syncthreads is also marked as noduplicate, for now
anyway), it is free to perform optimizations on sync() that it would not
be able to perform on __syncthreads(), because sync() is not marked as
convergent.
Similarly, we need a notion of convergent calls, since in the case when
we can't statically determine a call's target(s), we need to know
whether it's safe to perform optimizations around the call.
This change is conservative; the optimizer will remove these attrs where
it can, see r260318, r260319.
Reviewers: majnemer
Subscribers: cfe-commits, jhen, echristo, tra
Differential Revision: http://reviews.llvm.org/D17056
llvm-svn: 261779
The assert is triggered because isObjCRetainableType() is called on the
canonicalized return type that has been stripped of the typedefs and
attributes attached to it. To fix this assert, this commit gets the
original return type from CurCodeDecl or BlockInfo and uses it instead
of the canoicalized type.
rdar://problem/24470031
Differential Revision: http://reviews.llvm.org/D16914
llvm-svn: 261151
being called with the wrong size: convert CGFunctionInfo to use TrailingObjects
and ask TrailingObjects to provide a working 'operator delete' for us.
llvm-svn: 260181
Clang's CodeGen has several paths which end up invoking or calling a
function. The one that we used for calls to __RTtypeid did not
appropriately annotate the call with a funclet bundle.
This fixes PR26329.
llvm-svn: 258877
This updates clang to use bundle operands to associate an invoke with
the funclet which it is contained within.
Depends on D15517.
Differential Revision: http://reviews.llvm.org/D15518
llvm-svn: 255675
`pass_object_size` is our way of enabling `__builtin_object_size` to
produce high quality results without requiring inlining to happen
everywhere.
A link to the design doc for this attribute is available at the
Differential review link below.
Differential Revision: http://reviews.llvm.org/D13263
llvm-svn: 254554
This patch changes the generation of CGFunctionInfo to contain
the FunctionProtoType if it is available. This enables the code
generation for call instructions to look into this type for
exception information and therefore generate better quality
IR - it will not create invoke instructions for functions that
are know not to throw.
llvm-svn: 253926
The ``disable_tail_calls`` attribute instructs the backend to not
perform tail call optimization inside the marked function.
For example,
int callee(int);
int foo(int a) __attribute__((disable_tail_calls)) {
return callee(a); // This call is not tail-call optimized.
}
Note that this attribute is different from 'not_tail_called', which
prevents tail-call optimization to the marked function.
rdar://problem/8973573
Differential Revision: http://reviews.llvm.org/D12547
llvm-svn: 252986
This attribute is used to prevent tail-call optimizations to the marked
function. For example, in the following piece of code, foo1 will not be
tail-call optimized:
int __attribute__((not_tail_called)) foo1(int);
int foo2(int a) {
return foo1(a); // Tail-call optimization is not performed.
}
The attribute has effect only on statically bound calls. It has no
effect on indirect calls. Also, virtual functions and objective-c
methods cannot be marked as 'not_tail_called'.
rdar://problem/22667622
Differential Revision: http://reviews.llvm.org/D12922
llvm-svn: 252369
This works around PR25162. The MSVC tables make it very difficult to
correctly inline a C++ destructor that contains try / catch. We've
attempted to address PR25162 in LLVM's backend, but it feels pretty
infeasible. MSVC and ICC both appear to avoid inlining such complex
destructors.
Long term, we want to fix this by making the inliner smart enough to
know when it is inlining into a cleanup, so it can inline simple
destructors (~unique_ptr and ~vector) while avoiding destructors
containing try / catch.
llvm-svn: 251576
Summary: It breaks the build for the ASTMatchers
Subscribers: klimek, cfe-commits
Differential Revision: http://reviews.llvm.org/D13893
llvm-svn: 250827
match the feature set of the function that they're being called from.
This ensures that we can effectively diagnose some[1] code that would
instead ICE in the backend with a failure to select message.
Example:
__m128d foo(__m128d a, __m128d b) {
return __builtin_ia32_addsubps(b, a);
}
compiled for normal x86_64 via:
clang -target x86_64-linux-gnu -c
would fail to compile in the back end because the normal subtarget
features for x86_64 only include sse2 and the builtin requires sse3.
[1] We're still not erroring on:
__m128i bar(__m128i const *p) { return _mm_lddqu_si128(p); }
where we should fail and error on an always_inline function being
inlined into a function that doesn't support the subtarget features
required.
llvm-svn: 250473
The backend restores the stack pointer after recovering from an
exception. This is similar to r245879, but it doesn't try to use the
normal cleanup mechanism, so hopefully it won't cause the same breakage.
llvm-svn: 249640
Summary:
This change adds support for `__builtin_ms_va_list`, a GCC extension for
variadic `ms_abi` functions. The existing `__builtin_va_list` support is
inadequate for this because `va_list` is defined differently in the Win64
ABI vs. the System V/AMD64 ABI.
Depends on D1622.
Reviewers: rsmith, rnk, rjmccall
CC: cfe-commits
Differential Revision: http://reviews.llvm.org/D1623
llvm-svn: 247941
-force-align-stack.
Also, make changes to the driver so that -mno-stack-realign is no longer
an option exposed to the end-user that disallows stack realignment in
the backend.
Differential Revision: http://reviews.llvm.org/D11815
llvm-svn: 247451
The type of a member pointer is incomplete if it has no inheritance
model. This lets us reuse more general logic already embedded in clang.
llvm-svn: 247346
It seems that there is small bug, and we can't generate assume loads
when some virtual functions have internal visibiliy
This reverts commit 982bb7d966947812d216489b3c519c9825cacbf2.
llvm-svn: 247332
Generating call assume(icmp %vtable, %global_vtable) after constructor
call for devirtualization purposes.
For more info go to:
http://lists.llvm.org/pipermail/cfe-dev/2015-July/044227.html
Edit:
Fixed version because of PR24479.
After this patch got reverted because of ScalarEvolution bug (D12719)
Merged after John McCall big patch (Added Address).
http://reviews.llvm.org/D11859
llvm-svn: 247199
We know that a reference can always be dereferenced. However, we don't
always know the number of bytes if the reference's pointee type is
incomplete. This case was correctly handled but we didn't consider the
case where the type is complete but we cannot calculate its size for ABI
specific reasons. In this specific case, a member pointer's size is
available only under certain conditions.
This fixes PR24703.
llvm-svn: 247188
instruction used the ReturnValue as pointer operand or value operand. This
led to wrong code gen - in later stages (load-store elision code) the found
store and its operand would be erased, causing ReturnValue to become a <badref>.
The patch adds a check that makes sure that ReturnValue is a pointer operand of
store instruction. Regression test is also added.
This fixes PR24386.
Differential Revision: http://reviews.llvm.org/D12400
llvm-svn: 247003
Introduce an Address type to bundle a pointer value with an
alignment. Introduce APIs on CGBuilderTy to work with Address
values. Change core APIs on CGF/CGM to traffic in Address where
appropriate. Require alignments to be non-zero. Update a ton
of code to compute and propagate alignment information.
As part of this, I've promoted CGBuiltin's EmitPointerWithAlignment
helper function to CGF and made use of it in a number of places in
the expression emitter.
The end result is that we should now be significantly more correct
when performing operations on objects that are locally known to
be under-aligned. Since alignment is not reliably tracked in the
type system, there are inherent limits to this, but at least we
are no longer confused by standard operations like derived-to-base
conversions and array-to-pointer decay. I've also fixed a large
number of bugs where we were applying the complete-object alignment
to a pointer instead of the non-virtual alignment, although most of
these were hidden by the very conservative approach we took with
member alignment.
Also, because IRGen now reliably asserts on zero alignments, we
should no longer be subject to an absurd but frustrating recurring
bug where an incomplete type would report a zero alignment and then
we'd naively do a alignmentAtOffset on it and emit code using an
alignment equal to the largest power-of-two factor of the offset.
We should also now be emitting much more aggressive alignment
attributes in the presence of over-alignment. In particular,
field access now uses alignmentAtOffset instead of min.
Several times in this patch, I had to change the existing
code-generation pattern in order to more effectively use
the Address APIs. For the most part, this seems to be a strict
improvement, like doing pointer arithmetic with GEPs instead of
ptrtoint. That said, I've tried very hard to not change semantics,
but it is likely that I've failed in a few places, for which I
apologize.
ABIArgInfo now always carries the assumed alignment of indirect and
indirect byval arguments. In order to cut down on what was already
a dauntingly large patch, I changed the code to never set align
attributes in the IR on non-byval indirect arguments. That is,
we still generate code which assumes that indirect arguments have
the given alignment, but we don't express this information to the
backend except where it's semantically required (i.e. on byvals).
This is likely a minor regression for those targets that did provide
this information, but it'll be trivial to add it back in a later
patch.
I partially punted on applying this work to CGBuiltin. Please
do not add more uses of the CreateDefaultAligned{Load,Store}
APIs; they will be going away eventually.
llvm-svn: 246985
every time it's called rather than attempting to cache the result.
It's unlikely to be called frequently and the overhead of using
it in the first place is already factored out.
llvm-svn: 246706
the main attribute and cache the results so we don't have to parse
a single attribute more than once.
This reapplies r246596 with a fix for an uninitialized class member,
and a couple of cleanups and formatting changes.
llvm-svn: 246610
Also:
- Add a typedef to make working with the result easier.
- Update callers to use the new function.
- Make initFeatureMap out of line.
llvm-svn: 246468
A couple of changes here:
a) Do less work in the case where we don't have a target attribute on the
function. We've already canonicalized the attributes for the function -
no need to do more work.
b) Use the newer canonicalized feature adding functions from TargetInfo
to do the work when we do have a target attribute. This enables us to diagnose
some warnings in the case of conflicting written attributes (only ppc does
this today) and also make sure to get all of the features for a cpu that's
listed rather than just change the cpu.
Updated all testcases accordingly and added a new testcase to verify that we'll
error out on ppc if we have some incompatible options using the existing diagnosis
framework there.
llvm-svn: 246195
This is important in the case that the LLVM-inferred llvm-struct
alignment is not the same as the clang-known C-struct alignment.
Differential Revision: http://reviews.llvm.org/D12243
llvm-svn: 245719
In llvm commit r243581, a reverse range adapter was added which allows
us to change code such as
for (auto I = Fields.rbegin(), E = Fields.rend(); I != E; ++I) {
in to
for (const FieldDecl *I : llvm::reverse(Fields))
This commit changes a few of the places in clang which are eligible to use
this new adapter.
llvm-svn: 243663
- Make it a proper random access iterator with a little help from iterator_adaptor_base
- Clean up users of magic dereferencing. The iterator should behave like an Expr **.
- Make it an implementation detail of Stmt. This allows inlining of the assertions.
llvm-svn: 242608
Code in CGCall.cpp that loads up function arguments that need to be
coerced to a different type may in some cases ignore the fact that
the source of the argument is not naturally aligned. This may cause
incorrect code to be generated. In some places in CreateCoercedLoad,
we already have setAlignment calls to address this, but I ran into one
where it was missing, causing wrong code generation on SystemZ.
However, in that location, we do not actually know what alignment of
the source location we can rely on; the callers do not pass anything
to this routine. This is already an issue in other places in
CreateCoercedLoad; and the same problem exists for CreateCoercedStore.
To avoid pessimising code, and to fix the FIXMEs already in place,
this patch also adds an alignment argument to the CreateCoerced*
routines and uses it instead of forcing an alignment of 1. The
callers are changed to pass in the best information they have.
This actually requires changes in a number of existing test cases
since we now get better alignment in many places.
Differential Revision: http://reviews.llvm.org/D11033
llvm-svn: 241898
This is needed to use clang's command line option "-ftrap-function" for LTO and
enable changing the trap function name on a per-call-site basis.
rdar://problem/21225723
Differential Revision: http://reviews.llvm.org/D10831
llvm-svn: 241306
While there replace stable_sort of std::string with just sort, stability
is not necessary for "simple" value types. No functional change intended.
llvm-svn: 241299
using a string map to canonicalize. Fix up a couple of testcases
that needed changing since we are no longer simply appending features
to the list, but all of their mask dependencies as well.
llvm-svn: 241129
The patch is generated using this command:
$ tools/extra/clang-tidy/tool/run-clang-tidy.py -fix \
-checks=-*,llvm-namespace-comment -header-filter='llvm/.*|clang/.*' \
work/llvm/tools/clang
To reduce churn, not touching namespaces spanning less than 10 lines.
llvm-svn: 240270
If llvm.lifetime.end turns out to be the first instruction in the last
basic block, we can decrement the iterator twice, going past rend.
At the moment, this can never happen because llvm.lifetime.end always
goes immediately after bitcast, but relying on this is very brittle.
llvm-svn: 239638
Right now we're ignoring the fpmath attribute since there's no
backend support for a feature like this and to do so would require
checking the validity of the strings and doing general subtarget
feature parsing of valid and invalid features with the target
attribute feature.
llvm-svn: 239582
Modeled after the gcc attribute of the same name, this feature
allows source level annotations to correspond to backend code
generation. In llvm particular parlance, this allows the adding
of subtarget features and changing the cpu for a particular function
based on source level hints.
This has been added into the existing support for function level
attributes without particular verification for any target outside
of whether or not the backend will support the features/cpu given
(similar to section, etc).
llvm-svn: 239579
This commit adds back the code that seems to have been dropped unintentionally
in r176985.
rdar://problem/13752163
Differential Revision: http://reviews.llvm.org/D10100
llvm-svn: 239426
Re-land the change r238200, but with modifications in the tests that should
prevent new failures in some environments as reported with the original
change on the mailing list.
llvm-svn: 238253
On MIPS unsigned int type should not be zero extended but sign-extended.
Patch by Strahinja Petrovic.
Differential Revision: http://reviews.llvm.org/D9198
llvm-svn: 238200
Summary:
Space on stack allocated for unused structures returned by functions was unused
even when it's lifetime didn't intersect with lifetime of any other objects that
could use the same space.
The test added also checks for named and auto objects. It seems to make sense
to have this all in one place.
Reviewers: aadg, rsmith, rjmccall, rnk
Reviewed By: rnk
Subscribers: asl, cfe-commits
Differential Revision: http://reviews.llvm.org/D9743
llvm-svn: 237385
This makes sure that the front end is specific about what they're expecting
the backend to produce. Update a FIXME with the idea that the target-features
could be more precise using backend knowledge.
llvm-svn: 235936
This reverts commit r234700. It turns out that the lifetime markers
were not the cause of Chromium failing but a bug which was uncovered by
optimizations exposed by the markers.
llvm-svn: 235553
Now that TailRecursionElimination has been fixed with r222354, the
threshold on size for lifetime marker insertion can be removed. This
only affects named temporary though, as the patch for unnamed temporaries
is still in progress.
My previous commit (r222993) was not handling debuginfo correctly, but
this could only be seen with some asan tests. Basically, lifetime markers
are just instrumentation for the compiler's usage and should not affect
debug information; however, the cleanup infrastructure was assuming it
contained only destructors, i.e. actual code to be executed, and was
setting the breakpoint for the end of the function to the closing '}', and
not the return statement, in order to show some destructors have been
called when leaving the function. This is wrong when the cleanups are only
lifetime markers, and this is now fixed.
llvm-svn: 234581
Not all of them (there's still a fallback for this specific function
that omits the type parameter) but it's some I bothered to do now.
llvm-svn: 234063
the target-cpu, if different from the triple's cpu, and
target-features as they're written that are passed down from the
driver.
Together with LLVM r232885 this should allow the LTO'ing of binaries
that contain modules compiled with different code generation options
on a subset of architectures with full backend support (x86, powerpc,
aarch64).
llvm-svn: 232888
The MS ABI utilizes a compiler generated function called the "vector
constructor iterator" to construct arrays of objects with
non-trivial constructors/destructors. For this to work, the constructor
must follow a specific calling convention. A thunk must be created if
the default constructor has default arguments, is variadic or is
otherwise incompatible. This thunk is called the default constructor
closure.
N.B. Default constructor closures are only generated if the default
constructor is exported because clang itself does not utilize vector
constructor iterators. Failing to export the default constructor
closure will result in link/load failure if a translation unit compiled
with MSVC is on the import side.
Differential Revision: http://reviews.llvm.org/D8331
llvm-svn: 232229
This adds support for copy-constructor closures. These are generated
when the C++ runtime has to call a copy-constructor with a particular
calling convention or with default arguments substituted in to the call.
Because the runtime has no mechanism to call the function with a
different calling convention or know-how to evaluate the default
arguments at run-time, we create a thunk which will do all the
appropriate work and package it in a way the runtime can use.
Differential Revision: http://reviews.llvm.org/D8225
llvm-svn: 231952
It's slightly cheaper than copying it, if the DebugLoc points to replaceable
metadata every copy is recorded in a DenseMap, moving reduces the peak size of
that map.
llvm-svn: 228492
__declspec(restrict) and __attribute(malloc) are both handled
identically by clang: they are allowed to the noalias LLVM attribute.
Seeing as how noalias models the C99 notion of 'restrict', rename the
internal clang attribute to Restrict from Malloc.
llvm-svn: 228120
Summary:
It was used for interoperability with PNaCl's calling conventions, but
it's no longer needed.
Also Remove NaCl*ABIInfo which just existed to delegate to either the portable
or native ABIInfo, and remove checkCallingConvention which was now a no-op
override.
Reviewers: jvoung
Subscribers: jfb, llvm-commits
Differential Revision: http://reviews.llvm.org/D7206
llvm-svn: 227362
This workaround was to provide unique call sites to ensure LLVM's inline
debug info handling would properly unique two calls to the same function
on the same line. Instead, this has now been fixed in LLVM (r226736) and
the workaround here can be removed.
Originally committed in r176895, but this isn't a straight revert due to
all the changes since then. I just searched for anything ForcedColumn*
related and removed them.
We could test this - but it didn't strike me as terribly valuable once
we're no longer adding this workaround everything just works as expected
& it's no longer a special case to test for.
llvm-svn: 226738
The test was fixed after a discussion with the revision author: the check
pattern was made more flexible as the "%call" part is not what we actually want
to check strictly there.
The original patch description:
===
Introduce SPIR calling conventions.
This implements Section 3.7 from the SPIR 1.2 spec:
SPIR kernels should use "spir_kernel" calling convention.
Non-kernel functions use "spir_func" calling convention. All
other calling conventions are disallowed.
The patch works only for OpenCL source. Any other uses will need
to ensure that kernels are assigned the spir_kernel calling
convention correctly.
===
llvm-svn: 226561
This implements Section 3.7 from the SPIR 1.2 spec:
SPIR kernels should use "spir_kernel" calling convention.
Non-kernel functions use "spir_func" calling convention. All
other calling conventions are disallowed.
The patch works only for OpenCL source. Any other uses will need
to ensure that kernels are assigned the spir_kernel calling
convention correctly.
llvm-svn: 226548
The code setting the debug location being removed here was accidentally
leaking a location into the call to the non-static data member's ctor
call. Without it the call had no location and could cause assertion
failures if it was inlined. Now that it has a location (and a correct
one at that) this code should hopefully be no longer needed.
It's possible of course that other parts of the debug info are also
relying on the debug locations being set here to leak to where they're
needed - so we might see the same assertions again & will have to
investigate what the dependence was/is. But the chances are good that
any of those are debug info line table quality bugs we've just not found
yet anyway - so it'll be good to flush them out.
llvm-svn: 226383
PR22096 has several test cases that assert that look fairly different. I'm
adding one of those as an automated test, but when relanding the other cases
should probably be checked as well.
llvm-svn: 225361
r225000 generalized debug info line info handling for expressions such
that this code is no longer necessary.
This removes the last use of CGDebugInfo::getLocation, but not all the
uses of CGDebugInfo::CurLoc, which is still used internally in
CGDebugInfo. I'd like to do away with all of that & might succeed after
a few more patches.
llvm-svn: 225085
The extension has the following syntax:
__builtin_call_with_static_chain(Call, Chain)
where Call must be a function call expression and Chain must be of pointer type
This extension performs a function call Call with a static chain pointer
Chain passed to the callee in a designated register. This is useful for
calling foreign language functions whose ABI uses static chain pointers
(e.g. to implement closures).
Differential Revision: http://reviews.llvm.org/D6332
llvm-svn: 224167
having OptimizeNone remove them again, just don't add them in the
first place if the function already has OptimizeNone.
Note that MinSize can still appear due to attributes on different
declarations; a future patch will address that.
llvm-svn: 224047
Summary:
This change makes CodeGenFunction::EmitCheck() take several
conditions that needs to be checked (all of them need to be true),
together with sanitizer kinds these checks are for. This would allow
to split one call into UBSan runtime into several calls in case
different sanitizer kinds would have different recoverability
settings.
Tests should be fixed accordingly, I'm working on it.
Test Plan: regression test suite.
Reviewers: rsmith
Reviewed By: rsmith
Subscribers: cfe-commits
Differential Revision: http://reviews.llvm.org/D6219
llvm-svn: 221716
Make sure CodeGenFunction::EmitCheck() knows which sanitizer
it emits check for. Make CheckRecoverableKind enum an
implementation detail and move it away from header.
Currently CheckRecoverableKind is determined by the type of
sanitizer ("unreachable" and "return" are unrecoverable,
"vptr" is always-recoverable, all the rest are recoverable).
This will change in future if we allow to specify which sanitizers
are recoverable, and which are not by -fsanitize-recover= flag.
No functionality change.
llvm-svn: 221635
Use the bitmask to store the set of enabled sanitizers instead of a
bitfield. On the negative side, it makes syntax for querying the
set of enabled sanitizers a bit more clunky. On the positive side, we
will be able to use SanitizerKind to eventually implement the
new semantics for -fsanitize-recover= flag, that would allow us
to make some sanitizers recoverable, and some non-recoverable.
No functionality change.
llvm-svn: 221558
The most complex aspect of the convention is the handling of homogeneous
vector and floating point aggregates. Reuse the homogeneous aggregate
classification code that we use on PPC64 and ARM for this.
This convention also has a C mangling, and we apparently implement that
in both Clang and LLVM.
Reviewed By: majnemer
Differential Revision: http://reviews.llvm.org/D6063
llvm-svn: 221006
Summary:
The Itanium ABI approach of using offset-to-top isn't possible with the
MS ABI, it doesn't have that kind of information lying around.
Instead, we do the following:
- Call the virtual deleting destructor with the "don't delete the object
flag" set. The virtual deleting destructor will return a pointer to
'this' adjusted to the most derived class.
- Call the global delete using the adjusted 'this' pointer.
Reviewers: rnk
Subscribers: cfe-commits
Differential Revision: http://reviews.llvm.org/D5996
llvm-svn: 220993
Reuse the PPC64 HVA detection algorithm for ARM and AArch64. This is a
nice code deduplication, since they are roughly identical. A few virtual
method extension points are needed to understand how big an HVA can be
and what element types it can have for a given architecture.
Also make the record expansion code work in the presence of non-virtual
bases.
Reviewed By: uweigand, asl
Differential Revision: http://reviews.llvm.org/D6045
llvm-svn: 220972
SanitizerOptions is not even a POD now, so having global variable of
this type, is not nice. Instead, provide a regular constructor and clear()
method, and let each CodeGenFunction has its own copy of SanitizerOptions
it uses.
llvm-svn: 220920
Wire it through everywhere we have support for fastcall, essentially.
This allows us to parse the MSVC "14" CTP headers, but we will
miscompile them because LLVM doesn't support __vectorcall yet.
Reviewed By: Aaron Ballman
Differential Revision: http://reviews.llvm.org/D5808
llvm-svn: 220573
Make it possible to pass NULL through variadic functions on 64-bit
Windows targets. The Visual C++ headers define NULL to 0, when they
should define it to 0LL on Win64 so that NULL is a pointer-sized
integer.
Fixes PR20949.
Reviewers: thakis, rsmith
Differential Revision: http://reviews.llvm.org/D5480
llvm-svn: 219456
This adds support for the align_value attribute. This attribute is supported by
Intel's compiler (versions 14.0+), and several of my HPC users have requested
support in Clang. It specifies an alignment assumption on the values to which a
pointer points, and is used by numerical libraries to encourage efficient
generation of vector code.
Of course, we already have an aligned attribute that can specify enhanced
alignment for a type, so why is this additional attribute important? The
problem is that if you want to specify that an input array of T is, say,
64-byte aligned, you could try this:
typedef double aligned_double attribute((aligned(64)));
void foo(aligned_double *P) {
double x = P[0]; // This is fine.
double y = P[1]; // What alignment did those doubles have again?
}
the access here to P[1] causes problems. P was specified as a pointer to type
aligned_double, and any object of type aligned_double must be 64-byte aligned.
But if P[0] is 64-byte aligned, then P[1] cannot be, and this access causes
undefined behavior. Getting round this problem requires a lot of awkward
casting and hand-unrolling of loops, all of which is bad.
With the align_value attribute, we can accomplish what we'd like in a well
defined way:
typedef double *aligned_double_ptr attribute((align_value(64)));
void foo(aligned_double_ptr P) {
double x = P[0]; // This is fine.
double y = P[1]; // This is fine too.
}
This attribute does not create a new type (and so it not part of the type
system), and so will only "propagate" through templates, auto, etc. by
optimizer deduction after inlining. This seems consistent with Intel's
implementation (thanks to Alexey for confirming the various Intel-compiler
behaviors).
As a final note, I would have chosen to call this aligned_value, not
align_value, for better naming consistency with the aligned attribute, but I
think it would be more useful to users to adopt Intel's name.
llvm-svn: 218910
This is the last piece of CGCall code that had implicit assumptions about
the order in which Clang arguments are translated to LLVM ones (positions
of inalloca argument, sret, this, padding arguments etc.) Now all of
this data is encapsulated in ClangToLLVMArgsMapping. If this information
would be required somewhere else, this class can be moved to a separate
header or pulled into CGFunctionInfo.
llvm-svn: 218634
Add a method to calculate the number of arguments given QualType
expnads to. Use this method in ClangToLLVMArgMapping calculation.
This number may be cached in CodeGenTypes for efficiency, if needed.
llvm-svn: 218623
Hoist the logic which determines the way QualType is expanded
into a separate method. Remove a bunch of copy-paste and simplify
getTypesFromArgs() / ExpandTypeFromArgs() / ExpandTypeToArgs() methods.
llvm-svn: 218615
In addition to __builtin_assume_aligned, GCC also supports an assume_aligned
attribute which specifies the alignment (and optional offset) of a function's
return value. Here we implement support for the assume_aligned attribute by making
use of the @llvm.assume intrinsic.
llvm-svn: 218500
Summary:
This patch implements a new UBSan check, which verifies
that function arguments declared to be nonnull with __attribute__((nonnull))
are actually nonnull in runtime.
To implement this check, we pass FunctionDecl to CodeGenFunction::EmitCallArgs
(where applicable) and if function declaration has nonnull attribute specified
for a certain formal parameter, we compare the corresponding RValue to null as
soon as it's calculated.
Test Plan: regression test suite
Reviewers: rsmith
Reviewed By: rsmith
Subscribers: cfe-commits, rnk
Differential Revision: http://reviews.llvm.org/D5082
llvm-svn: 217389
There were code paths that are duplicated for constructors and destructors just
because we have both CXXCtorType and CXXDtorsTypes.
This patch introduces an unified enum and reduces code deplication a bit.
llvm-svn: 217383
For naked functions with parameters, Clang would still emit stores in the prologue
that would clobber the stack, because LLVM doesn't set up a stack frame. (This
shows up in -O0 compiles, because the stores are optimized away otherwise.)
For example:
__attribute__((naked)) int f(int x) {
asm("movl $42, %eax");
asm("retl");
}
Would result in:
_Z1fi:
movl 12(%esp), %eax
movl %eax, (%esp) <--- Oops.
movl $42, %eax
retl
Differential Revision: http://reviews.llvm.org/D5183
llvm-svn: 217198
This avoids encoding information about the function prototype into the
thunk at the cost of some function prototype bitcast gymnastics.
Fixes PR20653.
llvm-svn: 216782
Previously, EnterStructPointerForCoercedAccess used Alloc size when determining how to convert. This was problematic, because there were situations were the alloc size was larger than the store size. For example, if the first element of a structure were i24 and the destination type were i32, the old code would generate a GEP and a load i24. The code should compare store sizes to ensure the whole object is loaded. I have attached a test case.
This patch modifies the output of arm64-be-bitfield.c test case, but the new IR seems to be equivalent, and after -O3, the compiler generates identical ARM assembly. (asr x0, x0, #54)
Patch by Thomas Jablin!
llvm-svn: 216722
This tidies up some ARM-specific code added by r208417 to move it out
of the target-independent parts of clang into TargetInfo.cpp. This
also has the advantage that we can now flatten struct arguments to
variadic AAPCS functions.
llvm-svn: 216535
Summary:
This refactoring introduces ClangToLLVMArgMapping class, which
encapsulates the information about the order in which function arguments listed
in CGFunctionInfo should be passed to actual LLVM IR function, such as:
1) positions of sret, if there is any
2) position of inalloca argument, if there is any
3) position of helper padding argument for each call argument
4) positions of regular argument (there can be many if it's expanded).
Simplify several related methods (ConstructAttributeList, EmitFunctionProlog
and EmitCall): now they don't have to maintain iterators over the list
of LLVM IR function arguments, dealing with all the sret/inalloca/this complexities,
and just use expected positions of LLVM IR arguments stored in ClangToLLVMArgMapping.
This may increase the running time of EmitFunctionProlog, as we have to traverse
expandable arguments twice, but in further refactoring we will be able
to speed up EmitCall by passing already calculated CallArgsToIRArgsMapping to
ConstructAttributeList, thus avoiding traversing expandable argument there.
No functionality change.
Test Plan: regression test suite
Reviewers: majnemer, rnk
Reviewed By: rnk
Subscribers: cfe-commits, rjmccall, timurrrr
Differential Revision: http://reviews.llvm.org/D4938
llvm-svn: 216251
Summary:
This patch adds a runtime check verifying that functions
annotated with "returns_nonnull" attribute do in fact return nonnull pointers.
It is based on suggestion by Jakub Jelinek:
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20140623/223693.html.
Test Plan: regression test suite
Reviewers: rsmith
Reviewed By: rsmith
Subscribers: cfe-commits
Differential Revision: http://reviews.llvm.org/D4849
llvm-svn: 215485
This moves some memptr specific code into the generic thunk emission
codepath.
Fixes PR20053.
Reviewers: majnemer
Differential Revision: http://reviews.llvm.org/D4613
llvm-svn: 214004
In C99, an array parameter declarator might have the form:
direct-declarator '[' 'static' type-qual-list[opt] assign-expr ']'
where the static keyword indicates that the caller will always provide a
pointer to the beginning of an array with at least the number of elements
specified by the assignment expression. For constant sizes, we can use the
new dereferenceable attribute to pass this information to the optimizer. For
VLAs, we don't know the size, but (for addrspace(0)) do know that the pointer
must be nonnull (and so we can use the nonnull attribute).
llvm-svn: 213444
Because references must be initialized using some evaluated expression, they
must point to something, and a callee can assume the reference parameter is
dereferenceable. Taking advantage of a new attribute just added to LLVM, mark
them as such.
Because dereferenceability in addrspace(0) implies nonnull in the backend, we
don't need both attributes. However, we need to know the size of the object to
use the dereferenceable attribute, so for incomplete types we still emit only
nonnull.
llvm-svn: 213386
We now have an LLVM-level nonnull attribute that can be applied to function
parameters, and we emit it for reference types (as of r209723), but did not
emit it when an __attribute__((nonnull)) was provided. Now we will.
llvm-svn: 212835
Of course, such code is horribly broken and will explode on impact.
That said, ATL does it, and we have to support them, at least a little
bit.
Fixes PR20191.
llvm-svn: 212508
This is a GNU attribute that causes calls within the attributed function
to be inlined where possible. It is implemented by giving such calls the
alwaysinline attribute.
Differential Revision: http://reviews.llvm.org/D3816
llvm-svn: 209217
This is a GNU attribute that allows split stacks to be turned off on a
per-function basis.
Differential Revision: http://reviews.llvm.org/D3817
llvm-svn: 209167
This allows us to perfectly forward non-trivial arguments that use
inalloca.
We still can't forward non-trivial arguments through thunks when we have
a covariant return type with a non-trivial adjustment. This would
require emitting an extra copy, which is non-conforming anyway.
llvm-svn: 208927
Summary:
MSVC always passes 'sret' after 'this', unlike GCC. This required
changing a number of places in Clang that assumed the sret parameter was
always first in LLVM IR.
This fixes win64 MSVC ABI compatibility for methods returning structs.
Reviewers: rsmith, majnemer
Subscribers: cfe-commits
Differential Revision: http://reviews.llvm.org/D3618
llvm-svn: 208458
This is the clang counterpart to 208413, which ensures that Homogeneous
Floating-point Aggregates are passed in consecutive registers on ARM.
llvm-svn: 208417
Previously we calculated the shift amount based upon DataLayout::getTypeAllocSizeInBits.
This will only work for legal types - types such as i24 that are created as part of
structs for bitfields will return "32" from that function. Change to using
getTypeSizeInBits.
It turns out that AArch64 didn't run across this problem because it always returned
[1 x i64] as the type for a bitfield, whereas ARM64 returns i64 so goes down this
(better, but wrong) codepath.
llvm-svn: 208231
Passing objects directly (in registers or memory) creates a second copy
of the object in the callee. The callee always destroys its copy, but
we also have to destroy any temporary created in the caller. In other
words, copy elision of these kinds of objects is impossible.
Objects larger than 8 bytes with non-trivial dtors and trivial copy
ctors are still passed indirectly, and we can still elide copies of
them.
Fixes PR19640.
llvm-svn: 207889
We were destroying them in the callee, and then again in the caller. We
should use an EH-only cleanup and disable it at the point of the call
for win64, even though we don't use inalloca.
llvm-svn: 207733
If we crash, we raise a crash handler dialog, and that's really
annoying. Even though we can't emit correct IR until we have musttail,
don't crash.
llvm-svn: 205948
Summary:
The definition of a type later in a translation unit may change it's
type from {}* to (%struct.foo*)*. Earlier function definitions may use
the former while more recent definitions might use the later. This is
fine until they interact with one another (like one calling the other).
In these cases, a bitcast is needed because the inalloca must match the
function call but the store to the lvalue which initializes the argument
slot has to match the rvalue's type.
This technique is along the same lines with what the other,
non-inalloca, codepaths perform.
This fixes PR19287.
Reviewers: rnk
CC: cfe-commits
Differential Revision: http://llvm-reviews.chandlerc.com/D3224
llvm-svn: 205217
r203364: what was use_iterator is now user_iterator, and there is
a use_iterator for directly iterating over the uses.
This also switches to use the range-based APIs where appropriate.
llvm-svn: 203365
Previously the X86 backend would look for the sret attribute and handle
this for us. inalloca takes that all away, so we have to do the return
ourselves now.
llvm-svn: 202097
When a non-trivial parameter is present, clang now gathers up all the
parameters that lack inreg and puts them into a packed struct. MSVC
always aligns each parameter to 4 bytes and no more, so this is a pretty
simple struct to lay out.
On win64, non-trivial records are passed indirectly. Prior to this
change, clang was incorrectly using byval on win64.
I'm able to self-host a working clang with this change and additional
LLVM patches.
Reviewers: rsmith
Differential Revision: http://llvm-reviews.chandlerc.com/D2636
llvm-svn: 200597
This fixes PR15768, where the sret parameter and the 'this' parameter
are in the wrong order.
Instance methods compiled by MSVC never return records in registers,
they always return indirectly through an sret pointer. That sret
pointer always comes after the 'this' parameter, for both __cdecl and
__thiscall methods.
Unfortunately, the same is true for other calling conventions, so we'll
have to change the overall approach here relatively soon.
Reviewers: rsmith
Differential Revision: http://llvm-reviews.chandlerc.com/D2664
llvm-svn: 200587
A return type is the declared or deduced part of the function type specified in
the declaration.
A result type is the (potentially adjusted) type of the value of an expression
that calls the function.
Rule of thumb:
* Declarations have return types and parameters.
* Expressions have result types and arguments.
llvm-svn: 200082
Fix a perennial source of confusion in the clang type system: Declarations and
function prototypes have parameters to which arguments are supplied, so calling
these 'arguments' was a stretch even in C mode, let alone C++ where default
arguments, templates and overloading make the distinction important to get
right.
Readability win across the board, especially in the casting, ADL and
overloading implementations which make a lot more sense at a glance now.
Will keep an eye on the builders and update dependent projects shortly.
No functional change.
llvm-svn: 199686
There are a number of places where we do PGO.setCurrentRegionCount(0)
directly after an unconditional branch. Give this operation a name so
that it's clearer why we're doing this.
llvm-svn: 199138
Unlike Itanium's VTTs, the 'most derived' boolean or bitfield is the
last parameter for non-variadic constructors, rather than the second.
For variadic constructors, the 'most derived' parameter comes after the
'this' parameter. This affects constructor calls and constructor decls
in a variety of places.
Reviewers: timurrrr
Differential Revision: http://llvm-reviews.chandlerc.com/D2405
llvm-svn: 197518
Summary:
MSVC destroys arguments in the callee from left to right. Because C++
objects have to be destroyed in the reverse order of construction, Clang
has to construct arguments from right to left and destroy arguments from
left to right.
This patch fixes the ordering by reversing the order of evaluation of
all call arguments under the MS C++ ABI.
Fixes PR18035.
Reviewers: rsmith
Differential Revision: http://llvm-reviews.chandlerc.com/D2275
llvm-svn: 196402
CodeGenABITypes is a wrapper built on top of CodeGenModule that exposes
some of the functionality of CodeGenTypes (held by CodeGenModule),
specifically methods that determine the LLVM types appropriate for
function argument and return values.
I addition to CodeGenABITypes.h, CGFunctionInfo.h is introduced, and the
definitions of ABIArgInfo, RequiredArgs, and CGFunctionInfo are moved
into this new header from the private headers ABIInfo.h and CGCall.h.
Exposing this functionality is one part of making it possible for LLDB
to determine the actual ABI locations of function arguments and return
values, making it possible for it to determine this for any supported
target without hard-coding ABI knowledge in the LLDB code.
llvm-svn: 193717
CodeGenTypes already has a reference to a CGCXXABI. Use this directly
rather than going through CodeGenModule to get to the same information.
This is consistent with other references to CGCXXABI in CodeGenTypes
functions defined in CGCall.cpp.
llvm-svn: 191854
Summary:
Makes functions with implicit calling convention compatible with
function types with a matching explicit calling convention. This fixes
things like calls to qsort(), which has an explicit __cdecl attribute on
the comparator in Windows headers.
Clang will now infer the calling convention from the declarator. There
are two cases when the CC must be adjusted during redeclaration:
1. When defining a non-inline static method.
2. When redeclaring a function with an implicit or mismatched
convention.
Fixes PR13457, and allows clang to compile CommandLine.cpp for the
Microsoft C++ ABI.
Excellent test cases provided by Alexander Zinenko!
Reviewers: rsmith
Differential Revision: http://llvm-reviews.chandlerc.com/D1231
llvm-svn: 189412
Yaxun Liu