Summary:
I broke building the builtins with r245967. This fixes them on Linux and builds them properly for Darwin.
The old code could not be made to work on Darwin as a result of the refactoring of add_compiler_rt_runtime, so I had to rework the way they are built for Darwin. This solution is not ideal and will be fixed in subsequent commits. I just want to get this in so everything is working again.
Reviewers: samsonov, chh, compnerd, bogner
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D12500
llvm-svn: 246487
Summary: This refactoring moves much of the Apple-specific behavior into a function in AddCompilerRT. The next cleanup patch will remove more of the if(APPLE) checks in the outlying CMakeLists.
This patch adds a bunch of new functionality to add_compiler_rt_runtime so that the target names don't need to be reconstructed outside the call. It also updates some of the call sites to exercise the new functionality, but does not update all uses fully. Subsequent patches will further update call sites and move to using the new features.
Reviewers: filcab, bogner, kubabrecka, zaks.anna, glider, samsonov
Subscribers: beanz, rengolin, llvm-commits
Differential Revision: http://reviews.llvm.org/D12292
llvm-svn: 245970
MachO and COFF do not support aliases. Restrict the alias to ELF targets. This
should also fix the Darwin build. Make the FNALIAS usage an error on non-ELF
targets.
llvm-svn: 245669
Summary:
Currently CMake doesn't build builtins for AArch64 and if one does this anyway
it's likely that at least `__multc3`, `__floatditf` and `__floatunditf` will be
missing. There is actually more builtins to add, but these come from
different libc implementations, thus providing them makes compiler-rt for
AArch64 good enough at least for basic usage.
Builtins implementation were originally taken from FreeBSD project:
* [[ https://reviews.freebsd.org/D2173 | __multc3 ]]
* [[ https://reviews.freebsd.org/D2174 | __floatditf and __floatunditf ]]
Until they have been tested to find mistakes in `__float*` functions.
`__floatditf` was based on `__floatsitf`, which had the same mistakes
(fixed it in r243746).
Version of the builtins in this patch are fixed and complemented with basic
tests. Additionally they were tested via GCC's torture (this is what revealed
these issues).
P.S. Ed (author of FreeBSD patches) asked for feedback on the list some time ago (here [[ http://lists.cs.uiuc.edu/pipermail/llvmdev/2015-March/084064.html | here ]])
and got no response, but it seems to be worth adding these builtins as is and
extracting common part later.
Reviewers: howard.hinnant, t.p.northover, jmolloy, enefaim, rengolin, zatrazz
Subscribers: asl, emaste, samsonov, aemerson, llvm-commits, rengolin
Differential Revision: http://reviews.llvm.org/D11679
llvm-svn: 245296
Negative numbers were handled properly initially, but got broken
during addressing review, so none of them did actually work. Issues:
* Wrong negation.
* Wrong exponent calculation.
llvm-svn: 243746
This applies default compiler flags to .S files, in particular removing
the "-pedantic" option, which is desirable because there is nothing to
reasonably warn about; and the only thing that gcc warns about is that
you allegedly can't correctly invoke GLUE2 in lib/builtins/assembly.h
on platforms for which USER_LABEL_PREFIX is the empty string.
In the gcc bug https://gcc.gnu.org/bugzilla/show_bug.cgi?id=33305 that
added the warning, a commenter notes that giving a macro of zero characters
to another macro is not precisely the same as failing to supply an argument,
and "there is a widespread belief in C++ community that such usage is valid".
Unfortunately the only way to silence the warning is to avoid -pedantic.
Differential Revision: http://reviews.llvm.org/D10713
llvm-svn: 243446
The is so that we can avoid using libgcc and use compiler-rt with
mingw-w64.
Related driver patch
http://reviews.llvm.org/D11077
I have tested this with mingw-w64 and everything seems to be in order.
I also sent this patch to the mingw-w64 mailing list for them to look at.
Patch by Martell Malone.
Differential Revision: http://reviews.llvm.org/D11085
llvm-svn: 242539
Summary:
This patch fixes incorrect truncation when the input wider value is
exactly 2^dstBits. For that value, the overflow to infinity is not
correctly handled. The fix is to replace a strict '>' with '>='.
Currently,
__truncdfsf2(340282366900000000000000000000000000000.0) returns infinity
__truncdfsf2(340282366920938463463374607431768211456.0) returns 0
__truncdfsf2(400000000000000000000000000000000000000.0) returns infinity
Likewise, __truncdfhf2 and __truncsfhf2 (and consequently gnu_f2h_ieee)
are discontinuous at 65536.0.
This patch adds tests for all three cases, along with adding a missing
header include to fp_test.h.
Reviewers: joerg, ab, srhines
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D10594
llvm-svn: 240450
Windows does not use AAPCS, but rather AAPCS-VFP, and thus the functions which
are assumed to be AAPCS will cause invalid argument setup. Ensure that the
functions are marked as AAPCS.
llvm-svn: 238056
Follow-up to r237161; seems like we can't use aliases, but we
can do better than duplicating the bodies, especially when that
body, after inlining, isn't as small as it looks.
Better approaches welcome. Perhaps the best thing is just to have
an #ifndef __APPLE__ over the GNUEABI names, since they're not used
there.
llvm-svn: 237323
Mostly uninteresting, except:
- in __extendXfYf2, when checking if the number is normal, the old
code relied on the unsignedness of src_rep_t, which is a problem
when sizeof(src_rep_t) < sizeof(int): the result gets promoted to
int, the signedness of which breaks the comparison.
I added an explicit cast; it shouldn't affect other types.
- we can't pass __fp16, so src_t and src_rep_t are the same.
- the gnu_*_ieee symbols are simply duplicated definitions, as aliases
are problematic on mach-o (where only weak aliases are supported;
that's not what we want).
Differential Revision: http://reviews.llvm.org/D9693
llvm-svn: 237161
Talking to John and Doug, we concluded that these functions from
stdatomic really do belong here in compiler-rt rather than in libc,
since the compiler owns stdatomic.h and these need to refer to
clang-specific builtins. Nonetheless, I've only added these on darwin
for now - other platforms should probably do the same unless their
libc does implement these functions.
llvm-svn: 236805
Summary:
The spec for these functions says that they should return 0 in this case but
this regressed in r234148. That revision essentially delegates the conversion
to the hardware, but that has different behavior on different platforms (e.g.
it is wrong on x86).
Also fix a typo in the name of __fixunsdfti
Test Plan: The existing unit tests now pass
Reviewers: joerg, howard.hinnant
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D9305
llvm-svn: 236319
This adds hard-float implementation for the following builtins:
* __fixdfdi()
* __fixsfdi()
* __fixunsdfdi()
* __fixunssfdi()
The soft-float implementation does never raise floating point
exceptions, which doesn't allow clients to detect floating point
conversion errors.
I must mention that I had to refer to libgcc's implementation to
write these functions.
Related unit-tests of compiler-rt passed with these changes.
Patch was somewhat out-dated, so was updated locally without any
functional changes.
Differential Revision: http://reviews.llvm.org/D5376
llvm-svn: 234148
80bit Intel/PPC long double is excluded due to lacking support
for the abstraction. Consistently provide saturation logic.
Extend to long double on 128bit IEEE extended platforms.
Initial patch with test cases from GuanHong Liu.
Reviewed by Steve Canon.
Differential Revision: http://reviews.llvm.org/D2804
llvm-svn: 231965
The syscall function itself is declared in <unistd.h>
Compiling compiler-rt with -Werror-implicit-function-declaration will
cause a build failure. This change fixes it.
Differential Revision: http://reviews.llvm.org/D7949
llvm-svn: 230838
This makes int_endianness.h work on operating systems for which we
haven't written explicit constructs, as long as GCC or Clang are being
used.
llvm-svn: 230208
Make sure clear_cache() builtin has an appropriate definition for Linux.
Call syscall(NR_cacheflush, ...).
Differential Revision: http://reviews.llvm.org/D7205
llvm-svn: 228767
Summary: This causes an invalid lvalue error when building for Android.
Reviewers: danalbert
Reviewed By: danalbert
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D7306
llvm-svn: 227907
Use synci implementation of clear_cache for short address ranges.
For long address ranges, make a kernel call.
Differential Revision: http://reviews.llvm.org/D6661
llvm-svn: 226567
Use unwind.h to get the declarations for unwinding interfaces. This header is
already provided by clang and gcc, so this adds no additional dependencies for
building the builtins library. It avoids the duplication which may drift over
time though.
llvm-svn: 225990
clang does not like the definition of builtins. In order to work around this,
we use a SUN CC to redefine the generated name. However, this requires that we
account for the user label prefix. Rather than hard coding that into the file,
rely on the compiler to tell us the information and use the preprocessor to
generate the name as we do in the assembly routines. NFC.
llvm-svn: 224597
Each of the object formats use a different directive for selecting the constant
section. Use a macro to avoid the duplication across a number of files. Also
correct a small macro mismatch on the Windows case (HIDDEN_DIRECTIVE -> HIDDEN).
Patch by Vadim Chugunov!
llvm-svn: 223910
The code in {add,sub}v.i3 routines does not trap when it should, because
it performs the actual add/subtract operation in signed arithmetic,
rather than unsigned.
Patch by Francois-Xavie Coudert!
llvm-svn: 221826
__ANDROID__ is a define that comes from the toolchain when building
for Android targets. ANDROID has a different meaning. ANDROID is
defined for _every_ Android build, including those done for host
modules. For host modules, we want to build the regular Linux
sanitizers and builtins, not the one for Android devices. This hasn't
been a problem until now because we only just started building the
sanitizers for the host.
llvm-svn: 220203
Clang r219557 introduces libcalls to complex multiply/divide functions.
Since these functions are not available in iOS for arm64 devices, add them to
the static libraries.
llvm-svn: 219715
This is simply to help clarity of the code. The functions are built as thumb
only if Thumb2 is available (__ARM_ARCH_ISA_THUMB == 2). Sink the selection
into the location of the definition and make DEFINE_COMPILERRT_THUMB_FUNCTION
always define a thumb function while DEFINE_COMPILERRT_FUNCTION always selects
the default.
Since the .thumb_func directive is always available (at least on Linux, Windows,
and BSD), sinking the macro right into the macro works just as well.
No functional change intended.
llvm-svn: 219182
The arm builtins converted into thumb in r213481 are not working
on darwin. On apple platforms, .thumb_func directive is required
to generated correct symbols for thumb functions.
<rdar://problem/18523605>
llvm-svn: 219040
This is needed so we can produce -i686- named libraries for
x86 Android (which is i686-linux-android).
An alternative solution would be keeping the "i386" name internally and
tweaking the OUTPUT_NAME of compiler-rt libraries.
llvm-svn: 218761
The ldrexd and strexd instructions are undefined for the ARMv7M
architecture, so we cannot use them to implement the
__sync_fetch_and_*_8 builtins. There is no other way to implement
these without OS support, so this patch #ifdef's these functions out
for M-class architectures.
There are no tests as I cannot find any existing tests for these
builtins.
I used the __ARM_ARCH_PROFILE predefine because __ARM_FEATURE_LDREX is
deprecated and not set by clang.
llvm-svn: 218601
__ARM_ARCH is part of the ACLE specification. At least clang and GCC have
supported this part of the ACLE for some time now. Let the compiler provide the
proper definition for the macro rather than try to guess it.
llvm-svn: 218095
Add the missing AEABI functions that are part of the base platform ABI
specification. The provided implementation does the bare minimum to avoid
requiring libc headers. This permits the use of compiler-rt on bare-metal
environments which conform to EABI.
llvm-svn: 217322
This macro did not do anything at this point, and is not particularly needed for
Windows unless building the builtins as a shared library. NFC.
llvm-svn: 217321
When building the builtins for a modern CPU (idiv support), __umodsi3 was
completely incorrect as it would behave as __udivmosi3, which takes a tertiary
parameter which is a pointer.
__udivsi3 was also incorrect, returning the remainder in r1. Although this
would not result in any crash or invalid behaviour as r1 is a caller saved
register in AAPCS, this is unnecessary. Simply perform the division ignoring
the remainder.
llvm-svn: 215295
The LLVM IAS seems to accept wide instructions for add and sub in ARM mode even
though it is not permitted. This uses a macro to ensure that the wide modifier
is only applied when building in THUMB mode.
This repairs building with GCC/GAS in ARM mode.
llvm-svn: 214046
The macro definitions are shared across multiple files. Define them once in the
assembly.h header rather than redefining it in each file.
llvm-svn: 214045
The architecture specific implementation of routines would be built and included
along with the generic implementation. This would result in multiple
definitions of those symbols.
The linker is free to select either of the two. Most of the time, this
shouldn't be too terrible as the forward iteration should catch the architecture
version due to the ordering. Rather than relying on the linker and build
infrastructure ordering things in a specific manner, only provide the
architecture version when available.
This reduces the size of compiler-rt, simplifies inspection of the library
implementations, and guarantees that the desired version is selected at a
slightly complex build system.
llvm-svn: 214040
Place the floating point constants into the read-only data section. This was
already being done for x86_64, this simply mirrors the behaviour for i686.
llvm-svn: 214034
MMX/SSE instructions expect 128-bit alignment (16-byte) for constants that they
reference. Correct the alignment on the constant values. Although it is quite
possible for the data to end up aligned, there is no guarantee that this will
occur unless it is explicitly aligned to the desired location. If the data ends
up being unaligned, the resultant binary would fault at runtime due to the
unaligned access.
As an example, the follow would fault previously:
cc -c lib/builtins/x86_64/floatundidf.S -o floatundidf.o
cc -c test/builtins/Unit/floatundidf_test.c -o floatundidf_test.c
ld -m elf_x86_64 floatundidf.o floatundidf_test.o -lc -o floatundidf
However, if the object files were reversed, the data would end up aligned and
the problem would go unnoticed.
llvm-svn: 214033
Re-apply SVN r213684 which was reverted in SVN r213724 since it broke the
build bots. Add a tweak to enable inclusion of the assembly sources in
standalone build as well.
Original commit message:
This patch address the PR20360. The CMake assembler build system
ignores the .S assembly files in builtins library build. This patch
fixes the issue.
llvm-svn: 214013
The .rodata directive was added on the IA-64 (Itanium) platform. The LLVM IAS
supports the .rodata on i386 and x86_64 as well. There is no reason to really
restrict compilation of the builtins to just clang. By explicitly indicating
that the data is meant to be pushed into the .rodata section via the .section
.rodata, the assembly is made compatible with clang and gcc (with GAS).
This will enable building these routines on the Linux buildbots via CMake.
llvm-svn: 214012
Convert the CBNZ backward branch instruction to CMP and BNE
avoiding illegal backwards branch and making the assembly code
in synh-ops.h to be UAL compliant.
Patch by: Sumanth Gundapaneni
llvm-svn: 213685
The CMake assembler build system ignores the .S assembly files in builtins
library build. This patch fixes the issue.
Patch by: Sumanth Gundapaneni
llvm-svn: 213684
For ARM cores that are ARMv6T2+ but not ARMv7ve or ARMv7-r and not an updated
ARMv7-a that has the idiv extension (chips with clz but not idiv), an incorrect
jump would be calculated due to the preference to thumb instructions over ARM.
Rather than computing the target at runtime, use a jumptable instead. This
trades a bit of storage for performance. The overhead is 32-bytes for each of
the three routines, but avoid the calculation of the offset.
Because clz was introduced in ARMv6T2 and idiv in certain versions of ARMv7,
the non-clz, non-idiv case implies a target which does not support Thumb-2, and
thus we cannot use Thumb on those targets (as it is unlikely that the assembly
will assemble).
Take the opportunity to refactor the IT block macros into assembly.h rather than
redefining them in the TUs where they are used.
Existing tests cover the full change already, so no new tests are added.
This effectively reverts SVN r213309.
llvm-svn: 213467
The udivmodsi4/modsi3/umodsi3 code computes jump targets based on ARM encodings
(if CLZ is present and IDIV is not present).
Reverts parts of r211032 and r211035.
llvm-svn: 213309
Without some mention of armv6m in a subdirectory of builtins, the make code
doesn't even know that armv6m exists and is something it should be looking for
in the platform-specific Makefiles. This means that none of the functions
listed actually get built and we end up with an almost entirely empty
libclang_rt.a for armv6m.
Unfortunately, the assembly code in the usual arm directory has no hope of
running on armv6m, which only supports Thumb-1 (not even ARM mode), so adding
it there won't work. Realistically, we probably *will* want to put any
optimised versions in a separate directory, so creating it now is harmless.
rdar://problem/17613576
llvm-svn: 212696
Storage Class 3 is static storage. These symbols need to be marked as external
(storage class 2) so that they can be referenced. Note that this external is
not the same as ELF "external" visibility, which is indicated by DLL Storage
Class (i.e. __declspec(dllexport) or __declspec(dllimport)).
llvm-svn: 211428
When possible, use Thumb or Thumb-2 over ARM instructions. This is particularly
important for pure-Thumb environments (e.g. Windows on ARM). Although, it is
possible to conditionalise this for that target specifically, this is available
on most newer ARM CPUs, and the code remains compatible with older CPUs with no
adverse effects. It therefore feels better to always prefer Thumb when
possible.
llvm-svn: 211032
Place constants into .rdata if targeting ELF or COFF/PE. This should be
functionally identical, however, the data would be placed into a different
section. This is purely a cleanup change.
llvm-svn: 209986
Make the whitespace a bit more uniform in the various assembly routines. This
also makes the assembly files a bit more uniform on the ARM side by explicitly
stating that it is using the unified syntax and that the contents of the code is
in the text section (or segment). No functional change.
llvm-svn: 209985
Call it "libclang_rt.builtins-<arch>.a" to be consistent
with sanitizers/profile libraries naming. Modify Makefile
and CMake build systems and Clang driver accordingly.
Fixes PR19822.
llvm-svn: 209473
Extend the function definition macros further to support COFF object emission.
The function definition in COFF includes the type and storage class in the
symbol definition context. This is needed to make the assembly routines
possible to be built for COFF environments (i.e. Windows).
llvm-svn: 209095
Rename the HIDDEN_DIRECTIVE macro to HIDDEN and give it a parameter providing
the name of the symbol to be given hidden visibility. This makes the macros
more amenable to COFF.
llvm-svn: 209094
Chris Bieneman