This assembly is part of a macro that was reformatted in D60351.
The missing space between push and { results in:
Error: bad instruction `push{r4, r5,r6,lr}'
llvm-svn: 365957
This patch enables compiler-rt on SPARC targets. Most of the changes are straightforward:
- Add 32 and 64-bit sparc to compiler-rt
- lib/builtins/fp_lib.h needed to check if the int128_t and uint128_t types exist (which they don't on sparc)
There's one issue of note: many asan tests fail to compile on Solaris/SPARC:
fatal error: error in backend: Function "_ZN7testing8internal16BoolFromGTestEnvEPKcb": over-aligned dynamic alloca not supported.
Therefore, while asan is still built, both asan and ubsan-with-asan testing is disabled. The
goal is to check if asan keeps compiling on Solaris/SPARC. This serves asan in gcc,
which doesn't have the problem above and works just fine.
With this patch, sparcv9-sun-solaris2.11 test results are pretty good:
Failing Tests (9):
Builtins-sparc-sunos :: divtc3_test.c
Builtins-sparcv9-sunos :: compiler_rt_logbl_test.c
Builtins-sparcv9-sunos :: divtc3_test.c
[...]
UBSan-Standalone-sparc :: TestCases/TypeCheck/misaligned.cpp
UBSan-Standalone-sparcv9 :: TestCases/TypeCheck/misaligned.cpp
The builtin failures are due to Bugs 42493 and 42496. The tree contained a few additonal
patches either currently in review or about to be submitted.
Tested on sparcv9-sun-solaris2.11.
Differential Revision: https://reviews.llvm.org/D40943
llvm-svn: 365880
compiler-rt already uses libtool instead of ar when building for
Apple platform, but that's not being used when builtins are being
built separately e.g. as part of the runtimes build. This change
extracts the logic setting up libtool into a separate file and uses
it from both the compiler-rt and standalone builtins build.
Differential Revision: https://reviews.llvm.org/D62820
llvm-svn: 362466
If `COMPILER_RT_ARMHF_TARGET` is set , the definition of the AEABI runtime
function `__aeabi_fcmpun` is misspelt: `__aeabi_fcmpum` instead of
`__aeabi_fcmpun`.
Patch by Konstantin Schwarz!
llvm-svn: 362424
This should hopefully address the error we're seeing in older versions
of Clang.
Differential Revision: https://reviews.llvm.org/D62554
llvm-svn: 361909
The only difference between __eqsf2 and __gtsf2 is whether they return
1 or -1 on NaN. Rather than duplicating all the code, use a macro to
define the function twice and use an argument to decide whether to
negate the return value.
Differential Revision: https://reviews.llvm.org/D61919
llvm-svn: 361207
Symbol aliases are supported by all platforms that compiler-rt builtins
target, and we can use these instead of function redirects to avoid the
extra indirection.
This is part of the cleanup proposed in "[RFC] compiler-rt builtins
cleanup and refactoring".
Differential Revision: https://reviews.llvm.org/D60931
llvm-svn: 359413
Use the uniform single line C++/99 style for code comments.
This is part of the cleanup proposed in "[RFC] compiler-rt builtins
cleanup and refactoring".
Differential Revision: https://reviews.llvm.org/D60352
llvm-svn: 359411
Update formatting to use the LLVM style.
This is part of the cleanup proposed in "[RFC] compiler-rt builtins
cleanup and refactoring".
Differential Revision: https://reviews.llvm.org/D60351
llvm-svn: 359410
The generic version of comparesf2 defines __cmpsf2 alias for libgcc
compatibility, but the ARM overlay is missing the alias.
Differential Revision: https://reviews.llvm.org/D60805
llvm-svn: 358542
This is support for building compiler-rt builtins, The library build
should be complete for a subset of supported platforms, but not all
CMake options have been replicated in GN.
We always use the just built compiler to build all the runtimes, which
is equivalent to the CMake runtimes build. This simplifies the build
configuration because we don't need to support arbitrary host compiler
and can always assume the latest Clang. With GN's toolchain support,
this is significantly more efficient than the CMake runtimes build.
Differential Revision: https://reviews.llvm.org/D60331
llvm-svn: 357821
This avoids a conflict between stdbool.h, which defines bool to _Bool in
xkeycheck.h. From what I can tell, ymath.h is an internal header, and
the intention is that users should include math.h directly instead. It
doesn't appear to provide declarations of anything required for our
builtins. This include was added back in r249513 from 2015, and it's
possible that ymath.h provided something this code needed at the time,
but today it does not.
llvm-svn: 357728
Implement rounding mode support for addxf3/subxf3.
On architectures that implemented the support, this will access the
corresponding floating point environment register to apply the
correct rounding. For other architectures, it will keep the current
behaviour and use IEEE-754 default rounding mode (to nearest, ties
to even).
ARM32/AArch64 support implemented in this change. i386 and AMD64
will be added in a follow up change.
Differential Revision: https://reviews.llvm.org/D57143
llvm-svn: 357035
Also hide __cpu_inicator_init and __cpu_features2
for similar reasons.
Summary: Make __cpu_model a hidden symbol, to match libgcc.
Subscribers: llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D59561
llvm-svn: 356581
We were treating certain edge cases that are actually normal as denormal
results, and flushing them to zero; we shouldn't do that. Not sure this
is the cleanest way to implement this edge case, but I wanted to avoid
adding any code on the common path.
Differential Revision: https://reviews.llvm.org/D59070
llvm-svn: 356529
For 'cascadelake' this is adding a 'avx512vnni' feature check to the 0x55 skylake-avx512 model check. These CPUs use the same model number and only differ in the stepping number. But the feature flag is simpler than collecting all the stepping numbers.
For 'znver2' this is just syncing with LLVM's Host.cpp.
llvm-svn: 354927
Bionic libc relies on an old libgcc behaviour which does not set hidden
visibility attribute. Keep exporting these symbols on Android for
compatibility.
Differential Revision: https://reviews.llvm.org/D56977
llvm-svn: 351915
to reflect the new license. These used slightly different spellings that
defeated my regular expressions.
We understand that people may be surprised that we're moving the header
entirely to discuss the new license. We checked this carefully with the
Foundation's lawyer and we believe this is the correct approach.
Essentially, all code in the project is now made available by the LLVM
project under our new license, so you will see that the license headers
include that license only. Some of our contributors have contributed
code under our old license, and accordingly, we have retained a copy of
our old license notice in the top-level files in each project and
repository.
llvm-svn: 351648
to reflect the new license.
We understand that people may be surprised that we're moving the header
entirely to discuss the new license. We checked this carefully with the
Foundation's lawyer and we believe this is the correct approach.
Essentially, all code in the project is now made available by the LLVM
project under our new license, so you will see that the license headers
include that license only. Some of our contributors have contributed
code under our old license, and accordingly, we have retained a copy of
our old license notice in the top-level files in each project and
repository.
llvm-svn: 351636
This patch implements the long double __floattitf (int128_t) method for
PowerPC -- specifically to convert a 128 bit integer into a long double
(IBM double-double).
To invoke this method, one can do so by linking against compiler-rt, via the
--rtlib=compiler-rt command line option supplied to clang.
Differential Revision: https://reviews.llvm.org/D54313/
llvm-svn: 350818
This patch implements the __uint128_t __fixunstfti (long double) method for
PowerPC -- specifically to convert a long double (IBM double-double) to an
unsigned 128 bit integer.
The general approach of this algorithm is to convert the high and low doubles
of the long double and add them together if the doubles fit within 64 bits.
However, additional adjustments and scaling is performed when the high or low
double does not fit within a 64 bit integer.
To invoke this method, one can do so by linking against compiler-rt, via the
--rtlib=compiler-rt command line option supplied to clang.
Differential Revision: https://reviews.llvm.org/D54911
llvm-svn: 350815
This modification is require for NetBSD with GCC, as there is a
custom unwind.h header implementation with different types.
No functional change intended for others.
Cherry-picked chunk from D33878.
llvm-svn: 347102
This modification is require for NetBSD with GCC, as there is a
custom unwind.h header implementation with different types.
No functional change intended for others.
Cherry-picked chunk from D33878.
llvm-svn: 347101
Summary:
Windows has always been LLP64, not LP64, so the macros were incorrect.
Check for _WIN64, since AArch64 sets that too.
The tests have to be fixed up in two main ways:
1. Use 'ULL' suffixes to avoid sign extension when passing hex literals
with the sign bit set to signed 128 arguments. Clang
-fms-compatibility makes such hex literals signed, not unsigned.
2. Disable various tests for 80-bit long double interoperation with i128
values.
Reviewers: joerg, mstorsjo
Subscribers: javed.absar, kristof.beyls, hiraditya, aheejin, llvm-commits
Differential Revision: https://reviews.llvm.org/D53918
llvm-svn: 345796
Summary:
In r81552, the HAS_80_BIT_LONG_DOUBLE macro was added to the unit test
only version of int_lib.h. One month later in r85260 the duplicate
int_lib.h was removed, but the tests still passed because we don't build
with -Werror.
This is the minimal change to bring it back, and I decided to put the
configuration macro next to our 128-bit integer support macro.
Reviewers: joerg, compnerd, mstorsjo
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D53838
llvm-svn: 345645
This breaks some configurations, so just forward declare everything that we need.
rdar://35943793
Differential revision: https://reviews.llvm.org/D50269
llvm-svn: 345551
Summary: This patch adds additional features and cpus from libgcc. Unfortunately we've overflowed the existing 32-bits of features so we had to add a new __cpu_features2 variable to hold the additional bits. This matches libgcc as far as I can tell.
Reviewers: echristo
Reviewed By: echristo
Subscribers: dberris, llvm-commits
Differential Revision: https://reviews.llvm.org/D53461
llvm-svn: 344830
This patch removes the instruction set check to make the msr APSR_nzcvq,
ip instruction only execute if Thumb2 is used.
The APSR is a subset of the bits of the CPSR
(B.1.3.3 of the Arm v7 A and R ARM [1]) and is only available for A and
R profiles.
However in section B.9.3.11 of the same document we see that:
"In the A and R profiles, APSR_nzcvq is the same as CPSR_f"
"ARM recommends the APSR forms when only the N, Z, C, V, Q, and GE[3:0]
bits are being written."
This patch also make those files assemble for Armv8-M Mainline
architecture profile.
The builtins were cross-compiled for Arm, Aarch64 and Armv6-M, Armv7-M
and Armv7E-M targets.
Cross-compiled tests were executed for Arm target.
[1]: https://developer.arm.com/docs/ddi0406/latest/arm-architecture-reference-manual-armv7-a-and-armv7-r-edition
Patch by hug-dev (Hugues de Valon).
Differential Revision: https://reviews.llvm.org/D51854
llvm-svn: 343601
Summary:
The complex division builtins (div?c3) use logb methods from libm to scale numbers during division and avoid rounding issues. However, these come from libm, meaning anyone that uses --rtlib=compiler-rt also has to include -lm. Implement logb* methods for standard ieee 754 floats so we can avoid -lm on those platforms, falling back to the old behavior (using either logb() or `__builtin_logb()`) when not supported.
These new methods are defined internally as `__compiler_rt_logb` so as not to conflict with the libm definitions in any way.
This fixes just the libm methods mentioned in PR32279 and PR28652. libc is still required, although that seems to not be an issue.
Note: this is proposed as an alternative to just adding -lm: D49330.
Reviewers: efriedma, compnerd, scanon, echristo
Reviewed By: echristo
Subscribers: jsji, echristo, nemanjai, dberris, mgorny, kbarton, delcypher, llvm-commits, #sanitizers
Differential Revision: https://reviews.llvm.org/D49514
llvm-svn: 342917
This is called by Bionic on dlclose to delete the emutls pthread key.
The return value of pthread_key_delete is unchecked and behaviour of
multiple calls to the method is dependent on the implementation of
pthread_key_delete.
Differential Revision: https://reviews.llvm.org/D52251
llvm-svn: 342608
On sparc64/Linux, sparc64 isn't defined; the canonical way of
checking for sparc64 is sparc && arch64, which also works on the
BSDs and Solaris. Since this problem does not occur on 32-bit
architectures, riscv32 can be ignored. This fixes and refines rL324593.
Patch by jrtc27 (James Clarke)
Differential Revision: https://reviews.llvm.org/D43146
llvm-svn: 342504
MachO symbols are prefixed with an extra '_' (that's 3 in total for this
function), so assembly calls have to go through a wrapper to insert any prefix
needed.
llvm-svn: 341540
