| Age | Commit message (Collapse) | Author | Lines |
|---|
|
analogous to commit 1c9afd69051a64cf085c6fb3674a444ff9a43857 for
atan[2][f].
|
|
for functions implemented in C, this is a requirement of C11 (F.6);
strictly speaking that text does not apply to standard library
functions, but it seems to be intended to apply to them, and C2x is
expected to make it a requirement.
failure to drop excess precision is particularly bad for inverse trig
functions, where a value with excess precision can be outside the
range of the function (entire range, or range for a particular
subdomain), breaking reasonable invariants a caller may expect.
|
|
|
|
|
|
SQRT.fmt exists on MIPS II+ (float), MIPS III+ (double).
ABS.fmt exists on MIPS I+ but only cores with ABS2008 flag in FCSR
implement the required behaviour.
|
|
Both sqrt and sqrtf shifted the signed exponent as signed int to adjust
the bit representation of the result. There are signed right shifts too
in the code but those are implementation defined and are expected to
compile to arithmetic shift on supported compilers and targets.
|
|
lrint in (LONG_MAX, 1/DBL_EPSILON) and in (-1/DBL_EPSILON, LONG_MIN)
is not trivial: rounding to int may be inexact, but the conversion to
int may overflow and then the inexact flag must not be raised. (the
overflow threshold is rounding mode dependent).
this matters on 32bit targets (without single instruction lrint or
rint), so the common case (when there is no overflow) is optimized by
inlining the lrint logic, otherwise the old code is kept as a fallback.
on my laptop an i486 lrint call is asm:10ns, old c:30ns, new c:21ns
on a smaller arm core: old c:71ns, new c:34ns
on a bigger arm core: old c:27ns, new c:19ns
|
|
commit f3ed8bfe8a82af1870ddc8696ed4cc1d5aa6b441 inadvertently removed
labels that were still needed.
|
|
commit 31c5fb80b9eae86f801be4f46025bc6532a554c5 introduced underflow
code paths for the i386 math asm, along with checks on the fpu status
word to skip the underflow-generation instructions if the underflow
flag was already raised. unfortunately, at least one such path, in
log1p, returned with 2 items on the x87 stack rather than just 1 item
for the return value. this is a violation of the ABI's calling
convention, and could cause subsequent floating point code to produce
NANs due to x87 stack overflow. if floating point results are used in
flow control, this can lead to runaway wrong code execution.
rather than reviewing each "underflow already raised" code path for
correctness, remove them all. they're likely slower than just
performing the underflow code unconditionally, and significantly more
complex.
all of this code should be ripped out and replaced by C source files
with inline asm. doing so would preclude this kind of error by having
the compiler perform all x87 stack register allocation and stack
manipulation, and would produce comparable or better code. however
such a change is a much larger project.
|
|
Author: Alex Suykov <alex.suykov@gmail.com>
Author: Aric Belsito <lluixhi@gmail.com>
Author: Drew DeVault <sir@cmpwn.com>
Author: Michael Clark <mjc@sifive.com>
Author: Michael Forney <mforney@mforney.org>
Author: Stefan O'Rear <sorear2@gmail.com>
This port has involved the work of many people over several years. I
have tried to ensure that everyone with substantial contributions has
been credited above; if any omissions are found they will be noted
later in an update to the authors/contributors list in the COPYRIGHT
file.
The version committed here comes from the riscv/riscv-musl repo's
commit 3fe7e2c75df78eef42dcdc352a55757729f451e2, with minor changes by
me for issues found during final review:
- a_ll/a_sc atomics are removed (according to the ISA spec, lr/sc
are not safe to use in separate inline asm fragments)
- a_cas[_p] is fixed to be a memory barrier
- the call from the _start assembly into the C part of crt1/ldso is
changed to allow for the possibility that the linker does not place
them nearby each other.
- DTP_OFFSET is defined correctly so that local-dynamic TLS works
- reloc.h LDSO_ARCH logic is simplified and made explicit.
- unused, non-functional crti/n asm files are removed.
- an empty .sdata section is added to crt1 so that the
__global_pointer reference is resolvable.
- indentation style errors in some asm files are fixed.
|
|
from https://github.com/ARM-software/optimized-routines,
commit 04884bd04eac4b251da4026900010ea7d8850edc
The underflow exception is signaled if the result is in the subnormal
range even if the result is exact.
code size change: +3421 bytes.
benchmark on x86_64 before, after, speedup:
-Os:
pow rthruput: 102.96 ns/call 33.38 ns/call 3.08x
pow latency: 144.37 ns/call 54.75 ns/call 2.64x
-O3:
pow rthruput: 98.91 ns/call 32.79 ns/call 3.02x
pow latency: 138.74 ns/call 53.78 ns/call 2.58x
|
|
from https://github.com/ARM-software/optimized-routines,
commit 04884bd04eac4b251da4026900010ea7d8850edc
TOINT_INTRINSICS and EXP_USE_TOINT_NARROW cases are unused.
The underflow exception is signaled if the result is in the subnormal
range even if the result is exact (e.g. exp2(-1023.0)).
code size change: -1672 bytes.
benchmark on x86_64 before, after, speedup:
-Os:
exp rthruput: 12.73 ns/call 6.68 ns/call 1.91x
exp latency: 45.78 ns/call 21.79 ns/call 2.1x
exp2 rthruput: 6.35 ns/call 5.26 ns/call 1.21x
exp2 latency: 26.00 ns/call 16.58 ns/call 1.57x
-O3:
exp rthruput: 12.75 ns/call 6.73 ns/call 1.89x
exp latency: 45.91 ns/call 21.80 ns/call 2.11x
exp2 rthruput: 6.47 ns/call 5.40 ns/call 1.2x
exp2 latency: 26.03 ns/call 16.54 ns/call 1.57x
|
|
from https://github.com/ARM-software/optimized-routines,
commit 04884bd04eac4b251da4026900010ea7d8850edc
code size change: +2458 bytes (+1524 bytes with fma).
benchmark on x86_64 before, after, speedup:
-Os:
log2 rthruput: 16.08 ns/call 10.49 ns/call 1.53x
log2 latency: 44.54 ns/call 25.55 ns/call 1.74x
-O3:
log2 rthruput: 15.92 ns/call 10.11 ns/call 1.58x
log2 latency: 44.66 ns/call 26.16 ns/call 1.71x
|
|
from https://github.com/ARM-software/optimized-routines,
commit 04884bd04eac4b251da4026900010ea7d8850edc
Assume __FP_FAST_FMA implies __builtin_fma is inlined as a single
instruction.
code size change: +4588 bytes (+2540 bytes with fma).
benchmark on x86_64 before, after, speedup:
-Os:
log rthruput: 12.61 ns/call 7.95 ns/call 1.59x
log latency: 41.64 ns/call 23.38 ns/call 1.78x
-O3:
log rthruput: 12.51 ns/call 7.75 ns/call 1.61x
log latency: 41.82 ns/call 23.55 ns/call 1.78x
|
|
from https://github.com/ARM-software/optimized-routines,
commit 04884bd04eac4b251da4026900010ea7d8850edc
POWF_SCALE != 1.0 case only matters if TOINT_INTRINSICS is set, which
is currently not supported for any target.
SNaN is not supported, it would require an issignalingf
implementation.
code size change: -816 bytes.
benchmark on x86_64 before, after, speedup:
-Os:
powf rthruput: 95.14 ns/call 20.04 ns/call 4.75x
powf latency: 137.00 ns/call 34.98 ns/call 3.92x
-O3:
powf rthruput: 92.48 ns/call 13.67 ns/call 6.77x
powf latency: 131.11 ns/call 35.15 ns/call 3.73x
|
|
from https://github.com/ARM-software/optimized-routines,
commit 04884bd04eac4b251da4026900010ea7d8850edc
In expf TOINT_INTRINSICS is kept, but is unused, it would require support
for __builtin_round and __builtin_lround as single instruction.
code size change: +94 bytes.
benchmark on x86_64 before, after, speedup:
-Os:
expf rthruput: 9.19 ns/call 8.11 ns/call 1.13x
expf latency: 34.19 ns/call 18.77 ns/call 1.82x
exp2f rthruput: 5.59 ns/call 6.52 ns/call 0.86x
exp2f latency: 17.93 ns/call 16.70 ns/call 1.07x
-O3:
expf rthruput: 9.12 ns/call 4.92 ns/call 1.85x
expf latency: 34.44 ns/call 18.99 ns/call 1.81x
exp2f rthruput: 5.58 ns/call 4.49 ns/call 1.24x
exp2f latency: 17.95 ns/call 16.94 ns/call 1.06x
|
|
from https://github.com/ARM-software/optimized-routines,
commit 04884bd04eac4b251da4026900010ea7d8850edc
code size change: +177 bytes.
benchmark on x86_64 before, after, speedup:
-Os:
log2f rthruput: 11.38 ns/call 5.99 ns/call 1.9x
log2f latency: 35.01 ns/call 22.57 ns/call 1.55x
-O3:
log2f rthruput: 10.82 ns/call 5.58 ns/call 1.94x
log2f latency: 35.13 ns/call 21.04 ns/call 1.67x
|
|
from https://github.com/ARM-software/optimized-routines,
commit 04884bd04eac4b251da4026900010ea7d8850edc,
with minor changes to better fit into musl.
code size change: +289 bytes.
benchmark on x86_64 before, after, speedup:
-Os:
logf rthruput: 8.40 ns/call 6.14 ns/call 1.37x
logf latency: 31.79 ns/call 24.33 ns/call 1.31x
-O3:
logf rthruput: 8.43 ns/call 5.58 ns/call 1.51x
logf latency: 32.04 ns/call 20.88 ns/call 1.53x
|
|
|
|
These are supposed to be used in tail call positions when handling
special cases in new code. (fp exceptions may be raised "naturally"
by the common code path if special casing is more effort.)
This implements the error handling apis used in
https://github.com/ARM-software/optimized-routines
without errno setting.
|
|
Mark atanhi, atanlo, and aT in atanl.c as static, as they're not
intended to be part of the public API.
These are already static in the LDBL_MANT_DIG == 64 code, so this
patch is just making the LDBL_MANT_DIG == 113 code do the same thing.
|
|
fma is only available on recent x86_64 cpus and it is much faster than
a software fma, so this should be done with a runtime check, however
that requires more changes, this patch just adds the code so it can be
tested when musl is compiled with -mfma or -mfma4.
|
|
vfma is available in the vfpv4 fpu and above, the ACLE standard feature
test for double precision hardware fma support is
__ARM_FEATURE_FMA && __ARM_FP&8
we need further checks to work around clang bugs (fixed in clang >=7.0)
&& !__SOFTFP__
because __ARM_FP is defined even with -mfloat-abi=soft
&& !BROKEN_VFP_ASM
to disable the single precision code when inline asm handling is broken.
For runtime selection the HWCAP_ARM_VFPv4 hwcap flag can be used, but
that requires further work.
|
|
These are only available on hard float target and sqrt is not available
in the base ISA, so further check is used.
|
|
These are available in the s390x baseline isa -march=z900.
|
|
libc.h was intended to be a header for access to global libc state and
related interfaces, but ended up included all over the place because
it was the way to get the weak_alias macro. most of the inclusions
removed here are places where weak_alias was needed. a few were
recently introduced for hidden. some go all the way back to when
libc.h defined CANCELPT_BEGIN and _END, and all (wrongly implemented)
cancellation points had to include it.
remaining spurious users are mostly callers of the LOCK/UNLOCK macros
and files that use the LFS64 macro to define the awful *64 aliases.
in a few places, new inclusion of libc.h is added because several
internal headers no longer implicitly include libc.h.
declarations for __lockfile and __unlockfile are moved from libc.h to
stdio_impl.h so that the latter does not need libc.h. putting them in
libc.h made no sense at all, since the macros in stdio_impl.h are
needed to use them correctly anyway.
|
|
|
|
this makes significant differences to codegen on archs with an
expensive PLT-calling ABI; on i386 and gcc 7.3 for example, the sin
and sinf functions no longer touch call-saved registers or the stack
except for pushing outgoing arguments. performance is likely improved
too, but no measurements were taken.
|
|
|
|
since x86 and m68k are the only archs with 80-bit long double and each
has mandatory endianness, select the variant via endianness.
differences are minor: apparently just byte order and representation
of infinities. the m68k format is not well-documented anywhere I could
find, so if other differences are found they may require additional
changes later.
|
|
if double precision r=x*y+z is not a half way case between two single
precision floats or it is an exact result then fmaf returns (float)r.
however the exactness check was wrong when |x*y| < |z| and could cause
incorrectly rounded result in nearest rounding mode when r is a half
way case.
fmaf(-0x1.26524ep-54, -0x1.cb7868p+11, 0x1.d10f5ep-29)
was incorrectly rounded up to 0x1.d117ap-29 instead of 0x1.d1179ep-29.
(exact result is 0x1.d1179efffffffecp-29, r is 0x1.d1179fp-29)
|
|
the freebsd fma code failed to raise underflow exception in some
cases in nearest rounding mode (affects fmal too) e.g.
fma(-0x1p-1000, 0x1.000001p-74, 0x1p-1022)
and the inexact exception may be raised spuriously since the fenv
is not saved/restored around the exact multiplication algorithm
(affects x86 fma too).
another issue is that the underflow behaviour when the rounded result
is the minimal normal number is target dependent, ieee754 allows two
ways to raise underflow for inexact results: raise if the result before
rounding is in the subnormal range (e.g. aarch64, arm, powerpc) or if
the result after rounding with infinite exponent range is in the
subnormal range (e.g. x86, mips, sh).
to avoid all these issues the algorithm was rewritten with mostly int
arithmetics and float arithmetics is only used to get correct rounding
and raise exceptions according to the behaviour of the target without
any fenv.h dependency. it also unifies x86 and non-x86 fma.
fmaf is not affected, fmal need to be fixed too.
this algorithm depends on a_clz_64 and it required a few spurious
instructions to make sure underflow exception is raised in a particular
corner case. (normally FORCE_EVAL(tiny*tiny) would be used for this,
but on i386 gcc is broken if the expression is constant
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=57245
and there is no easy portable fix for the macro.)
|
|
while the official elfv2 abi for "powerpc64le" sets power8 as the
baseline isa, we use it for both little and big endian powerpc64
targets and need to maintain compatibility with pre-power8 models. the
instructions for sqrt, fabs, and fma are in the baseline isa; support
for the rest is conditional via predefined isa-level macros.
patch by David Edelsohn.
|
|
these were introduced in z196 and not available in the baseline (z900)
ISA level. use __HTM__ as an alternate indicator for ISA level, since
gcc did not define __ARCH__ until 7.x.
patch by David Edelsohn.
|
|
in nearest rounding mode scalbn could introduce double rounding error
when an intermediate value and the final result were both in the
subnormal range e.g.
scalbn(0x1.7ffffffffffffp-1, -1073)
returned 0x1p-1073 instead of 0x1p-1074, because the intermediate
computation got rounded to 0x1.8p-1023.
with the fix an intermediate value can only be in the subnormal range
if the final result is 0 which is correct even after double rounding.
(there still can be two roundings so signals may be raised twice, but
that's only observable with trapping exceptions which is not supported.)
|
|
this should increase performance and reduce code size on aarch64.
the compiled code was checked against using __builtin_* instead
of inline asm with gcc-6.2.0.
lrint is two instructions.
c with inline asm is used because it is safer than a pure asm
implementation, this prevents ll{rint,round} to be an alias
of l{rint,round} (because the types don't match) and depends
on gcc style inline asm support.
ceil, floor, round, trunc can either raise inexact on finite
non-integer inputs or not raise any exceptions. the new
implementation does not raise exceptions while the generic
c code does.
on aarch64, the underflow exception is signaled before rounding
(ieee 754 allows both before and after rounding, but it must be
consistent), the generic fma c code signals it after rounding
so using single instruction fixes a slight conformance issue too.
|
|
partly following freebsd rev 279491
https://svnweb.freebsd.org/base?view=revision&revision=279491
(musl had some of the fixes before freebsd).
the change should not matter much for j0f, y0f, but it improves
j1f and y1f in [2.5,~3.75] (that is [0x40200000,~0x40700000]).
near roots (e.g. around 3.8317 for j1f) there are still large
ulp errors.
dropped code that tried to raise inexact.
|
|
j is int32_t and thus j<<31 is undefined if j==1, so j is changed to
uint32_t locally as a quick fix, the generated code is not affected.
(this is a strict conformance fix, future c standard may allow 1<<31,
see DR 463. the bug was inherited from freebsd fdlibm, the proper fix
is to use uint32_t for all bit hacks, but that requires more intrusive
changes.)
reported by Daniel Sabogal
|
|
there was a copy paste error that could cause large ulp errors
in atan2l, atanl, asinl and acosl on aarch64, mips64 and mipsn32.
(the implementation is from freebsd fdlibm, but the tail end
of the polynomial was wrong. 128 bit long double functions
are not yet tested so this went undetected.)
|
|
expf(-NAN) was treated as expf(-large) which unconditionally
returns +0, so special case +-NAN.
reported by Petr Hosek.
|
|
commit e4355bd6bec89688e8c739cd7b4c76e675643dca moved the math asm
from external source files to inline asm, but unfortunately, all
current releases of clang use the wrong inline asm constraint codes
for float and double ("w" and "P" instead of "t" and "w",
respectively). this patch adds detection for the bug in configure,
and, for now, just disables the affected asm on broken clang versions.
|
|
in order to take advantage of the fpu in -mfloat-abi=softfp mode, the
__VFP_FP__ (presence of vfp fpu) was checked instead of checking for
__ARM_PCS_VFP (hardfloat EABI variant). however, the latter macro is
the one that's actually specified by the ABI documents rather than
being compiler-specific, and should also be checked in case __VFP_FP__
is not defined on some compilers or some configurations.
|
|
this makes it possible to inline them with LTO, and is the simplest
approach to eliminating the use of .sub files.
this also makes VFP sqrt available for use with the standard EABI
(plain arm rather than armhf subarch) when libc is built with
-mfloat-abi=softfp. the same could have been done for fabs, but when
the argument and return value are in integer registers, moving to VFP
registers and back is almost certainly more costly than a simple
integer operation.
|
|
a conforming compiler for an arch with excess precision floating point
(FLT_EVAL_METHOD!=0; presently i386 is the only such arch supported)
computes all intermediate results in the types float_t and double_t
rather than the nominal type of the expression. some incorrect
compilers, however, only keep excess precision in registers, and
convert down to the nominal type when spilling intermediate results to
memory, yielding unpredictable results that depend on the compiler's
choices of what/when to spill. in particular, this happens on old gcc
versions with -ffloat-store, which we need in order to work around
bugs where the compiler wrongly keeps explicitly-dropped excess
precision.
by explicitly converting to double_t where expressions are expected be
be evaluated in double_t precision, we can avoid depending on the
compiler to get types correct when spilling; the nominal and
intermediate precision now match. this commit should not change the
code generated by correct compilers, or by old ones on non-i386 archs
where double_t is defined as double.
this fixes a serious bug in argument reduction observed on i386 with
gcc 4.2: for values of x outside the unit circle, sin(x) was producing
results outside the interval [-1,1]. changes made in commit
0ce946cf808274c2d6e5419b139e130c8ad4bd30 were likely responsible for
breaking compatibility with this and other old gcc versions.
patch by Szabolcs Nagy.
|
|
these files are all accepted as legacy arm syntax when producing arm
code, but legacy syntax cannot be used for producing thumb2 with
access to the full ISA. even after switching to UAL, some asm source
files contain instructions which are not valid in thumb mode, so these
will need to be addressed separately.
|
|
Some armhf gcc toolchains (built with --with-float=hard but without
--with-fpu=vfp*) do not pass -mfpu=vfp to the assembler and then
binutils rejects the UAL mnemonics for VFP unless there is an .fpu vfp
directive in the asm source.
|
|
the implicit-operand form of fucomip is rejected by binutils 2.19 and
perhaps other versions still in use. writing both operands explicitly
fixes the issue. there is no change to the resulting output.
commit a732e80d33b4fd6f510f7cec4f5573ef5d89bc4e was the source of this
regression.
|
|
analogous to commit 8ed66ecbcba1dd0f899f22b534aac92a282f42d5 for i386.
|
|
none of these are actual textrels because of ld-time binding performed
by -Bsymbolic-functions, but I'm changing them with the goal of making
ld-time binding purely an optimization rather than relying on it for
semantic purposes.
in the case of memmove's call to memcpy, making it explicit that the
memmove asm is assuming the forward-copying behavior of the memcpy asm
is desirable anyway; in case memcpy is ever changed, the semantic
mismatch would be apparent while editing memmcpy.s.
|
|
this reverts the commit f29fea00b5bc72d4b8abccba2bb1e312684d1fce
which was based on a bug in C99 and POSIX and did not match IEEE-754
http://www.open-std.org/jtc1/sc22/wg14/www/docs/n1515.pdf
|
