Oliver Stannard [Thu, 3 Sep 2015 12:40:58 +0000 (12:40 +0000)]
[ARM] Allow passing/returning of __fp16 arguments
The ACLE (ARM C Language Extensions) 2.0 allows the __fp16 type to be
used as a functon argument or return type (ACLE 1.1 did not).
The current public release of the AAPCS (2.09) states that __fp16 values
should be converted to single-precision before being passed or returned,
but AAPCS 2.10 (to be released shortly) changes this, so that they are
passed in the least-significant 16 bits of either a GPR (for base AAPCS)
or a single-precision register (for AAPCS-VFP). This does not change how
arguments are passed if they get passed on the stack.
This patch brings clang up to compliance with the latest versions of
both of these specs.
We can now set the __ARM_FP16_ARGS ACLE predefine, and we have always
been able to set the __ARM_FP16_FORMAT_IEEE predefine (we do not support
the alternative format).
Oliver Stannard [Thu, 3 Sep 2015 11:46:24 +0000 (11:46 +0000)]
Revert 246755 as it breaks buildbots
Original commit message:
[ARM] Allow passing/returning of __fp16 arguments
The ACLE (ARM C Language Extensions) 2.0 allows the __fp16 type to be
used as a functon argument or return type (ACLE 1.1 did not).
The current public release of the AAPCS (2.09) states that __fp16 values
should be converted to single-precision before being passed or returned,
but AAPCS 2.10 (to be released shortly) changes this, so that they are
passed in the least-significant 16 bits of either a GPR (for base AAPCS)
or a single-precision register (for AAPCS-VFP). This does not change how
arguments are passed if they get passed on the stack.
This patch brings clang up to compliance with the latest versions of
both of these specs.
We can now set the __ARM_FP16_ARGS ACLE predefine, and we have always
been able to set the __ARM_FP16_FORMAT_IEEE predefine (we do not support
the alternative format).
Oliver Stannard [Thu, 3 Sep 2015 09:34:53 +0000 (09:34 +0000)]
[ARM] Allow passing/returning of __fp16 arguments
The ACLE (ARM C Language Extensions) 2.0 allows the __fp16 type to be
used as a functon argument or return type (ACLE 1.1 did not).
The current public release of the AAPCS (2.09) states that __fp16 values
should be converted to single-precision before being passed or returned,
but AAPCS 2.10 (to be released shortly) changes this, so that they are
passed in the least-significant 16 bits of either a GPR (for base AAPCS)
or a single-precision register (for AAPCS-VFP). This does not change how
arguments are passed if they get passed on the stack.
This patch brings clang up to compliance with the latest versions of
both of these specs.
We can now set the __ARM_FP16_ARGS ACLE predefine, and we have always
been able to set the __ARM_FP16_FORMAT_IEEE predefine (we do not support
the alternative format).
[OPENMP 4.1] Codegen for extended format of 'if' clause.
Fixed codegen for extended format of 'if' clauses with special 'directive-name-modifier' + ast-print tests for extended format of 'if' clause.
[OPENMP 4.1] Parsing/sema analysis for extended format of 'if' clause.
OpenMP 4.1 added special 'directive-name-modifier' to the 'if' clause.
Format of 'if' clause is as follows:
```
if([ directive-name-modifier :] scalar-logical-expression)
```
The restriction rules are also changed.
1. If any 'if' clause on the directive includes a 'directive-name-modifier' then all 'if' clauses on the directive must include a 'directive-name-modifier'.
2. At most one 'if' clause without a 'directive-name-modifier' can appear on the directive.
3. At most one 'if' clause with some particular 'directive-name-modifier' can appear on the directive.
'directive-name-modifier' is important for combined directives and allows to separate conditions in 'if' clause for simple sub-directives in combined directive. This 'directive-name-modifier' identifies the sub-directive to which this 'if' clause must be applied.
Anton Yartsev [Wed, 2 Sep 2015 21:01:59 +0000 (21:01 +0000)]
[analyzer] Refactoring: bring together scan-build options and environment variables.
Full list of changes:
- all scan-build command-line arguments are now kept in %Options hash.
- most of environment variables scan-build operates with are stored in %EnvVars hash.
- moved processing of command-line arguments to the ProcessArgs subroutine.
Migrate the target attribute parsing code to returning an instance
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.
Update comment for AdditionalMembers with a note to avoid using
additional data members in attributes as they'll leak and provide
some guidance as to where they should be allocated if necessary.
Ivan Krasin [Wed, 2 Sep 2015 20:02:38 +0000 (20:02 +0000)]
Do not include default sanitizer blacklists into -M/-MM/-MD/-MMD output.
Summary:
Do not include default sanitizer blacklists into -M/-MM/-MD/-MMD output.
Introduce a frontend option -fdepfile-entry, and only insert them
for the user-defined sanitizer blacklists. In frontend, grab ExtraDeps
from -fdepfile-entry, instead of -fsanitize-blacklist.
Richard Smith [Wed, 2 Sep 2015 17:45:54 +0000 (17:45 +0000)]
[modules] Don't waste time reading in the names the module file writer gave to blocks. We don't need these names, and decoding the corresponding bitcode has a significant cost.
David Majnemer [Wed, 2 Sep 2015 15:50:38 +0000 (15:50 +0000)]
[MS ABI] Number unnamed TagDecls which aren't externally visible
TagDecls (structs, enums, etc.) may have the same name for linkage
purposes of one another; to disambiguate, we add a number to the mangled
named. However, we didn't do this if the TagDecl has a pseudo-name for
linkage purposes (it was defined alongside a DeclaratorDecl or a
TypeNameDecl).
Migrate the target attribute parsing code into an extension off of
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.
Summary:
`OpaqueValueExpr`s may not have a source expression (as in the case when
they are created due to a default argument error).
This can cause an assertion failure in `TransformOpaqueValueExpr` during
template instantiation.
Migrate the target attribute parsing code into an extension off of
the main attribute and cache the results so we don't have to parse
a single attribute more than once.
Richard Smith [Tue, 1 Sep 2015 20:35:42 +0000 (20:35 +0000)]
Re-commit r246497 (and dependent changes r246524 and r246521), reverted in
r246546, with a workaround for an MSVC 2013 miscompile and an MSVC 2015
rejects-valid.
Original commit message:
[modules] Rework serialized DeclContext lookup table management. Instead of
walking the loaded ModuleFiles looking for lookup tables for the context, store
them all in one place, and merge them together if we find we have too many
(currently, more than 4). If we do merge, include the merged form in our
serialized lookup table, so that downstream readers never need to look at our
imports' tables.
This gives a huge performance improvement to builds with very large numbers of
modules (in some cases, more than a 2x speedup was observed).
Reverting r246497 (which requires also reverting r246524 and r246521 to avoid merge conflicts). It broke the build on MSVC 2015. It also broke an MSVC 2013 bot with testing issues.
Richard Smith [Tue, 1 Sep 2015 01:37:34 +0000 (01:37 +0000)]
[modules] Preserve DeclID order when merging lookup tables to give a more
predictable diagnostic experience. The hash-of-DeclID order we were using
before gave different results on Win32 due to a different predefined
declaration of __builtin_va_list.
Hal Finkel [Mon, 31 Aug 2015 23:55:19 +0000 (23:55 +0000)]
[PowerPC] Support __builtin_ppc_get_timebase
GCC 4.8+ has a PowerPC-specific intrinsic, __builtin_ppc_get_timebase, to do
what Clang's __builtin_readcyclecounter does. For compatibility with code that
uses GCC's spelling (including glibc), support it as well.
Richard Smith [Mon, 31 Aug 2015 22:17:11 +0000 (22:17 +0000)]
[modules] Rework serialized DeclContext lookup table management. Instead of
walking the loaded ModuleFiles looking for lookup tables for the context, store
them all in one place, and merge them together if we find we have too many
(currently, more than 4). If we do merge, include the merged form in our
serialized lookup table, so that downstream readers never need to look at our
imports' tables.
This gives a huge performance improvement to builds with very large numbers of
modules (in some cases, more than a 2x speedup was observed).
Rafael Espindola [Mon, 31 Aug 2015 19:17:51 +0000 (19:17 +0000)]
Stop hardcoding GCC paths in crt/ld.so lookup.
This patch refactors the code to use the GCC installation detector
(modified so that it works in Solaris), and uses
ToolChain::GetFilePath everywhere once it works.
David Majnemer [Mon, 31 Aug 2015 18:48:39 +0000 (18:48 +0000)]
[MS ABI] Correctly mangle classes without names for linkage purposes
A class without a name for linkage purposes gets a name along the lines
of <unnamed-type-foo> where foo is either the name of a declarator which
defined it (like a variable or field) or a
typedef-name (like a typedef or alias-declaration).
We handled the declarator case correctly but it would fall down during
template instantiation if the declarator didn't share the tag's type.
We failed to handle the typedef-name case at all.
Instead, keep track of the association between the two and keep it up to
date in the face of template instantiation.
Andrey Bokhanko [Mon, 31 Aug 2015 13:20:44 +0000 (13:20 +0000)]
PR17829: Proper diagnostic of mangled names conflicts
Proper diagnostic and resolution of mangled names conflicts between C++ methods
and C functions. This patch implements support for functions/methods only;
support for variables is coming separately.
Alexey Bataev [Mon, 31 Aug 2015 07:32:19 +0000 (07:32 +0000)]
[OPENMP 4.0] Codegen for array sections.
Added codegen for array section in 'depend' clause of 'task' directive. It emits to pointers, one for the begin of array section and another for the end of array section. Size of the section is calculated as (end + 1 - start) * sizeof(basic_element_type).
Yaron Keren [Sun, 30 Aug 2015 15:24:46 +0000 (15:24 +0000)]
Make test resistant to false matches of numbered (unnamed) labels inside other numbers.
In release builds labels are numbers. Matching just the number may result
in false matches where the label is contained in other numbers, such as
14 inside [114 x i8]. A stricter match requiring start of line or > character
before the label avoids these false matches.
Benjamin Kramer [Sun, 30 Aug 2015 15:12:28 +0000 (15:12 +0000)]
[OpenMP] Make the filetered clause iterator a real iterator and type safe.
This replaces the filtered generic iterator with a type-specfic one based
on dyn_cast instead of comparing the kind enum. This allows us to use
range-based for loops and eliminates casts. No functionality change
intended.
Chandler Carruth [Sun, 30 Aug 2015 07:51:18 +0000 (07:51 +0000)]
Update for several APIs in LLVM that now use StringRefs rather than
const char pointers. In turn, push this through Clang APIs as well,
simplifying a number of bits of code that was handling the oddities of
nullptrs.
David Majnemer [Sat, 29 Aug 2015 08:32:55 +0000 (08:32 +0000)]
[AST] Don't crash when comparing incomplete object
We cannot tell if an object is past-the-end if its type is incomplete.
Zero sized objects satisfy past-the-end criteria and our object might
turn out to be such an object.
Richard Smith [Fri, 28 Aug 2015 22:33:53 +0000 (22:33 +0000)]
PR24612: Don't bail out of checking a constexpr function before checking
whether it can ever produce a constant expression in the case where it has a
void return type and no return statements.
Devin Coughlin [Fri, 28 Aug 2015 22:26:05 +0000 (22:26 +0000)]
[analyzer] When memcpy'ing into a fixed-size array, do not invalidate entire region.
Change the analyzer's modeling of memcpy to be more precise when copying into fixed-size
array fields. With this change, instead of invalidating the entire containing region the
analyzer now invalidates only offsets for the array itself when it can show that the
memcpy stays within the bounds of the array.
This addresses false positive memory leak warnings of the kind reported by
krzysztof in https://llvm.org/bugs/show_bug.cgi?id=22954
Hans Wennborg [Fri, 28 Aug 2015 21:47:01 +0000 (21:47 +0000)]
Allow TLS vars in dllimport/export functions; only inline dllimport functions when safe (PR24593)
This patch does two things:
1) Don't error about dllimport/export on thread-local static local variables.
We put those attributes on static locals in dllimport/export functions
implicitly in case the function gets inlined. Now, for TLS variables this
is a problem because we can't import such variables, but it's a benign
problem becase:
2) Make sure we never inline a dllimport function TLS static locals. In fact,
never inline a dllimport function that references a non-imported function
or variable (because these are not defined in the importing library). This
seems to match MSVC's behaviour.
Alexey Bataev [Fri, 28 Aug 2015 06:09:05 +0000 (06:09 +0000)]
[OPENMP 4.0] Codegen for array sections.
Added codegen for array section in 'depend' clause of 'task' directive. It emits to pointers, one for the begin of array section and another for the end of array section. Size of the section is calculated as (end + 1 - start) * sizeof(basic_element_type).
Richard Trieu [Thu, 27 Aug 2015 23:38:45 +0000 (23:38 +0000)]
Fix macro backtrace printing.
Sometimes, a macro that expands to another macro name will not be printed in
the macro backtrace. This patch finds the missed macro expansions and prints
them. Fixes PR16799
Ahmed Bougacha [Thu, 27 Aug 2015 22:24:56 +0000 (22:24 +0000)]
[X86] Use AVX features instead of ABI to init. SimdDefaultAlign.
The ABI string only exists to communicate with TargetCodeGenInfo.
Concretely, since we only used "avx*" ABI strings on x86_64 (as AVX
doesn't affect the i386 ABIs), this meant that, when initializing
SimdDefaultAlign, we would ignore AVX/AVX512 on i386, for no good
reason.
Instead, directly check the features. A similar change for
MaxVectorAlign will follow.
Piotr Padlewski [Thu, 27 Aug 2015 21:35:41 +0000 (21:35 +0000)]
Assume loads fix #2
There was linker problem, and it turns out that it is not always safe
to refer to vtable. If the vtable is used, then we can refer to it
without any problem, but because we don't know when it will be used or
not, we can only check if vtable is external or it is safe to to emit it
speculativly (when class it doesn't have any inline virtual functions).
It should be fixed in the future.
Adrian Prantl [Thu, 27 Aug 2015 21:21:19 +0000 (21:21 +0000)]
CGDebugInfo: Factor out a getOrCreateStandaloneType() method.
Usually debug info is created on the fly while during codegen.
With this API it becomes possible to create standalone debug info
for types that are not referenced by any code, such as emitting debug info
for a clang module or for implementing something like -gfull.
Because on-the-fly debug info generation may still insert retained types
on top of them, all RetainedTypes are uniqued in CGDebugInfo::finalize().