advantages over the old code:
- correct results for floating point (old code was bogus)
- wide/regular scanf separated so scanf does not pull in wide code
- well-defined behavior on integers that overflow dest type
- support for %[a-b] ranges with %[ (impl-defined by widely used)
- no intermediate conversion of fmt string to wide string
- cleaner, easier to share code with strto* functions
- better standards conformance for corner cases
the old code remains in the source tree, as the wide versions of the
scanf-family functions are still using it. it will be removed when no
longer needed.
I'm not sure if it's legal for wordexp to modify this field, but this
is the only easy/straightforward fix, and applications should not
care. if it's an issue, i can work out a different (but more complex)
solution later.
this off-by-one error was causing values with just one digit past the
decimal point to be treated by the integer case. in many cases it
would yield the correct result, but if expressions are evaluated in
excess precision, double rounding may occur.
the "< 0" test was always false due to use of an unsigned type. this
resulted in infinite loops on 32-bit machines (adding -1U to a pointer
is the same as adding -1) and crashes on 64-bit machines (offsetting
the string pointer by 4gb-1b when an illegal sequence was hit).
TRE wants to treat + and ? after a +, ?, or * as special; ? means
ungreedy and + is reserved for future use. however, this is
non-conformant. although redundant, these redundant characters have
well-defined (no-op) meaning for POSIX ERE, and are actually _literal_
characters (which TRE is wrongly ignoring) in POSIX BRE mode.
the simplest fix is to simply remove the unneeded nonstandard
functionality. as a plus, this shaves off a small amount of bloat.
at -Os optimization level, gcc refuses to inline these functions even
though the inlined code would roughly the same size as the function
call, and much faster. the easy solution is to make them into macros.
whenever the base was small enough that more than one digit could
still fit after UINTMAX_MAX/36-1 was reached, only the first would be
allowed; subsequent digits would trigger spurious overflow, making it
impossible to read the largest values in low bases.
when upscaling, even the very last digit is needed in cases where the
input is exact; no digits can be discarded. but when downscaling, any
digits less significant than the mantissa bits are destined for the
great bitbucket; the only influence they can have is their presence
(being nonzero). thus, we simply throw them away early. the result is
nearly a 4x performance improvement for processing huge values.
the particular threshold LD_B1B_DIG+3 is not chosen sharply; it's
simply a "safe" distance past the significant bits. it would be nice
to replace it with a sharp bound, but i suspect performance will be
comparable (within a few percent) anyway.
now that this is the first operation, it can rely on the circular
buffer contents not being wrapped when it begins. we limit the number
of digits read slightly in the initial parsing loops too so that this
code does not have to consider the case where it might cause the
circular buffer to wrap; this is perfectly fine because KMAX is chosen
as a power of two for circular-buffer purposes and is much larger than
it otherwise needs to be, anyway.
these changes should not affect performance at all.
upscaling by even one step too much creates 3-29 extra iterations for
the next loop. this is still suboptimal since it always goes by 2^29
rather than using a smaller upscale factor when nearing the target,
but performance on common, small-magnitude, few-digit values has
already more than doubled with this change.
more optimizations on the way...
for example, "1000000000" was being read as "1" due to this loop
exiting early. it's necessary to actually update z and zero the
entries so that the subsequent rounding code does not get confused;
before i did that, spurious inexact exceptions were being raised.
note that there's no need for a precise cutoff, because exponents this
large will always result in overflow or underflow (it's impossible to
read enough digits to compensate for the exponent magnitude; even at a
few nanoseconds per digit it would take hundreds of years).
the immediate benefit is a significant debloating of the float parsing
code by moving the responsibility for keeping track of the number of
characters read to a different module.
by linking shgetc with the stdio buffer logic, counting logic is
defered to buffer refill time, keeping the calls to shgetc fast and
light.
in the future, shgetc will also be useful for integrating the new
float code with scanf, which needs to not only count the characters
consumed, but also limit the number of characters read based on field
width specifiers.
shgetc may also become a useful tool for simplifying the integer
parsing code.
this version is intended to be fully conformant to the ISO C, POSIX,
and IEEE standards for conversion of decimal/hex floating point
strings to float, double, and long double (ld64 or ld80 only at
present) values. in particular, all results are intended to be rounded
correctly according to the current rounding mode. further, this
implementation aims to set the floating point underflow, overflow, and
inexact flags to reflect the conversion performed.
a moderate amount of testing has been performed (by nsz and myself)
prior to integration of the code in musl, but it still may have bugs.
so far, only strto(d|ld|f) use the new code. scanf integration will be
done as a separate commit, and i will add implementations of the wide
character functions later.
the buffer in getaddrinfo really only matters when /etc/hosts is huge,
but in that case, the huge number of syscalls resulting from a tiny
buffer would seriously impact the performance of every name lookup.
the buffer in __dns.c has also been enlarged a bit so that typical
resolv.conf files will fit fully in the buffer. there's no need to
make it so large as to dominate the syscall overhead for large files,
because resolv.conf should never be large.
special care is made to avoid any inexact computations when either arg
is zero (in which case the exact absolute value of the other arg
should be returned) and to support the special condition that
hypot(±inf,nan) yields inf.
hypotl is not yet implemented since avoiding overflow is nontrivial.
the error status is required to be sticky after failure of dlopen or
dlsym until cleared by dlerror. applications and especially libraries
should never rely on this since it is not thread-safe and subject to
race conditions, but glib does anyway.
the old formula atan2(1,sqrt((1+x)/(1-x))) was faster but
could give nan result at x=1 when the rounding mode is
FE_DOWNWARD (so 1-1 == -0 and 2/-0 == -inf), the new formula
gives -0 at x=+-1 with downward rounding.
this has not been tested heavily, but it's known to at least assemble
and run in basic usage cases. it's nearly identical to the
corresponding i386 code, and thus expected to be just as correct or
just as incorrect.
the main practical results of this change are
1. the regex code is no longer subject to LGPL; it's now 2-clause BSD
2. most (all?) popular nonstandard regex extensions are supported
I hesitate to call this a "sync" since both the old and new code are
heavily modified. in one sense, the old code was "more severely"
modified, in that it was actively hostile to non-strictly-conforming
expressions. on the other hand, the new code has eliminated the
useless translation of the entire regex string to wchar_t prior to
compiling, and now only converts multibyte character literals as
needed.
in the future i may use this modified TRE as a basis for writing the
long-planned new regex engine that will avoid multibyte-to-wide
character conversion entirely by compiling multibyte bracket
expressions specific to UTF-8.
old code saved/restored the fenv (the new code is only as slow
as that when inexact is not set before the call, but some other
flag is set and the rounding is inexact, which is rare)
before:
bench_nearbyint_exact 5000000 N 261 ns/op
bench_nearbyint_inexact_set 5000000 N 262 ns/op
bench_nearbyint_inexact_unset 5000000 N 261 ns/op
after:
bench_nearbyint_exact 10000000 N 94.99 ns/op
bench_nearbyint_inexact_set 25000000 N 65.81 ns/op
bench_nearbyint_inexact_unset 10000000 N 94.97 ns/op
the fscale instruction is slow everywhere, probably because it
involves a costly and unnecessary integer truncation operation that
ends up being a no-op in common usages. instead, construct a floating
point scale value with integer arithmetic and simply multiply by it,
when possible.
for float and double, this is always possible by going to the
next-larger type. we use some cheap but effective saturating
arithmetic tricks to make sure even very large-magnitude exponents
fit. for long double, if the scaling exponent is too large to fit in
the exponent of a long double value, we simply fallback to the
expensive fscale method.
on atom cpu, these changes speed up scalbn by over 30%. (min rdtsc
timing dropped from 110 cycles to 70 cycles.)
exponents (base 2) near 16383 were broken due to (1) wrong cutoff, and
(2) inability to fit the necessary range of scalings into a long
double value.
as a solution, we fall back to using frndint/fscale for insanely large
exponents, and also have to special-case infinities here to avoid
inf-inf generating nan.
thankfully the costly code never runs in normal usage cases.
zero, one, two, half are replaced by const literals
The policy was to use the f suffix for float consts (1.0f),
but don't use suffix for long double consts (these consts
can be exactly represented as double).
Underflow exception is only raised when the result is
invalid, but fmod is always exact. x87 has a denormalization
exception, but that's nonstandard. And the superflous *1.0
will be optimized away by any compiler that does not honor
signaling nans.
Some code assumed ldexp(x, 1) is faster than 2.0*x,
but ldexp is a wrapper around scalbn which uses
multiplications inside, so this optimization is
wrong.
This commit also fixes fmal which accidentally
used ldexp instead of ldexpl loosing precision.
There are various additional changes from the
work-in-progress const cleanups.
Some long double consts were stored in two doubles as a workaround
for x86_64 and i386 with the following comment:
/* Long double constants are slow on these arches, and broken on i386. */
This is most likely old gcc bug related to the default x87 fpu
precision setting (it's double instead of double extended on BSD).
up to 30% faster exp2 by avoiding slow frndint and fscale functions.
expm1 also takes a much more direct path for small arguments (the
expected usage case).
unlike some implementations, these functions perform the equivalent of
gcc's -ffloat-store on the result before returning. this is necessary
to raise underflow/overflow/inexact exceptions, perform the correct
rounding with denormals, etc.
unlike trig functions, these are easy to do in asm because they do not
involve (arbitrary-precision) argument reduction. fpatan automatically
takes care of domain issues, and in asin and acos, fsqrt takes care of
them for us.
infinities were getting converted into nans. the new code simply tests
for infinity and replaces it with a large magnitude value of the same
sign.
also, the fcomi instruction is apparently not part of the i387
instruction set, so avoid using it.
these are functions that have direct fpu approaches to implementation
without problematic exception or rounding issues. x86_64 lacks
float/double versions because i'm unfamiliar with the necessary sse
code for performing these operations.
A faster workaround for spurious inexact exceptions
when the result cannot be represented. The old code
actually could be wrong, because gcc reordered the
integer conversion and the exception check.
Note that the new fesetround has slightly different semantics:
Storing the floating-point environment with fnstenv makes the
next fldenv (or fldcw) "non-signaling", so unmasked and pending
exceptions does not invoke the exception handler.
(These are rare since exceptions are handled immediately and by
default all exceptions are masked anyway. But if one manually
unmasks an exception in the control word then either sets the
corresponding exception flag in the status word or the execution
of an exception raising floating-point operation gets interrupted
then it may happen).
So the old implementation did not trap in some rare cases
where the new implementation traps.
However POSIX does not specify anything like the x87 exception
handling traps and the fnstenv/fldenv pair is significantly slower
than the fnstcw/fldcw pair (new code is about 5x faster here and
it's dominated by the function call overhead).
this is necessary to support archs where fenv is incomplete or
unavailable (presently arm). fma, fmal, and the lrint family should
work perfectly fine with this change; fmaf is slightly broken with
respect to rounding as it depends on non-default rounding modes to do
its work.
a double precision nan, when converted to extended (80-bit) precision,
will never end in 0x400, since the corresponding bits do not exist in
the original double precision value. thus there's no need to waste
time and code size on this check.
the fsqrt opcode is correctly rounded, but only in the fpu's selected
precision mode, which is 80-bit extended precision. to get a correctly
rounded double precision output, we check for the only corner cases
where two-step rounding could give different results than one-step
(extended-precision mantissa ending in 0x400) and adjust the mantissa
slightly in the opposite direction of the rounding which the fpu
already did (reported in the c1 flag of the fpu status word).
this should have near-zero cost in the non-corner cases and at worst
very low cost.
note that in order for sqrt() to get used when compiling with gcc, the
broken, non-conformant builtin sqrt must be disabled.
other cases with %x were probably broken too.
I would actually like to go ahead and replace this code in scanf with
calls to the new __intparse framework, but for now this calls for a
quick and unobtrusive fix without the risk of breaking other things.
thanks to the hard work of Szabolcs Nagy (nsz), identifying the best
(from correctness and license standpoint) implementations from freebsd
and openbsd and cleaning them up! musl should now fully support c99
float and long double math functions, and has near-complete complex
math support. tgmath should also work (fully on gcc-compatible
compilers, and mostly on any c99 compiler).
based largely on commit 0376d44a890fea261506f1fc63833e7a686dca19 from
nsz's libm git repo, with some additions (dummy versions of a few
missing long double complex functions, etc.) by me.
various cleanups still need to be made, including re-adding (if
they're correct) some asm functions that were dropped.
this is a popular extension some programs depend on, and by using a
temporary buffer and strdup rather than malloc prior to the syscall,
i've avoided the dependency on free and thus minimized the bloat cost
of supporting this feature.
this was discussed on the mailing list and no consensus on the
preferred solution was reached, so in anticipation of a release, i'm
just committing a minimally-invasive solution that avoids the problem
by ensuring that multi-threaded-capable programs will always have
initialized the thread pointer before any signal handler can run.
in the long term we may switch to initializing the thread pointer at
program start time whenever the program has the potential to access
any per-thread data.
in gcc 3, the visibility attribute must be placed on both the
declaration and on the definition. if it's omitted from the
definition, the compiler fails to emit the ".hidden" directive in the
assembly, and the linker will either generate textrels (if supported,
such as on i386) or refuse to link (on targets where certain types of
textrels are forbidden or impossible without further assumptions about
memory layout, such as on x86_64).
this patch also unifies the decision about when to use visibility into
libc.h and makes the visibility in the utf-8 state machine tables
based on libc.h rather than a duplicate test.
even if pthread_create/exit code is not linked, run flag needs to be
checked and cleanup function potentially run on pop. thus, move the
code to the module that's always linked when pthread_cleanup_push/pop
is used.
the old abi was intended to duplicate glibc's abi at the expense of
being ugly and slow, but it turns out glib was not even using that abi
except on non-gcc-compatible compilers (which it doesn't even support)
and was instead using an exceptions-in-c/unwind-based approach whose
abi we could not duplicate anyway without nasty dwarf2/unwind
integration.
the new abi is copied from a very old glibc abi, which seems to still
be supported/present in current glibc. it avoids all unwinding,
whether by sjlj or exceptions, and merely maintains a linked list of
cleanup functions to be called from the context of pthread_exit. i've
made some care to ensure that longjmp out of a cleanup function should
work, even though it is not required to.
this change breaks abi compatibility with programs which were using
pthread cancellation, which is unfortunate, but that's why i'm making
the change now rather than later. considering that most pthread
features have not been usable until recently anyway, i don't see it as
a major issue at this point.
