Rich Felker [Wed, 23 May 2012 02:04:55 +0000 (22:04 -0400)]
various header cleanups, some related to _BSD_SOURCE addition
there is no reason to avoid multiple identical macro definitions; this
is perfectly legal C, and even with the maximal warning options
enabled, gcc does not issue any warning for it.
Rich Felker [Wed, 23 May 2012 01:54:19 +0000 (21:54 -0400)]
bsd_signal is a legacy (removed) XSI function, not needed in _BSD_SOURCE
its only purpose was for use on non-BSD systems that implement sysv
semantics for signal() by default.
Rich Felker [Wed, 23 May 2012 01:52:08 +0000 (21:52 -0400)]
support _BSD_SOURCE feature test macro
patch by Isaac Dunham. matched closely (maybe not exact) to glibc's
idea of what _BSD_SOURCE should make visible.
Rich Felker [Tue, 22 May 2012 03:55:36 +0000 (23:55 -0400)]
fix typo in utimes function that made it mess up file times
Rich Felker [Tue, 22 May 2012 02:51:30 +0000 (22:51 -0400)]
fix out-of-bounds array access in pthread barriers on 64-bit
it's ok to overlap with integer slot 3 on 32-bit because only slots
0-2 are used on process-local barriers.
Rich Felker [Mon, 21 May 2012 02:56:06 +0000 (22:56 -0400)]
move getpass decl to the right place
Rich Felker [Thu, 17 May 2012 00:14:30 +0000 (20:14 -0400)]
fix misplaced semicolon in preprocessor directive (#undef h_errno)
Rich Felker [Mon, 14 May 2012 18:52:55 +0000 (14:52 -0400)]
yet another try to get the check for gcc right...
Rich Felker [Mon, 14 May 2012 12:14:27 +0000 (08:14 -0400)]
fix error in last configure change (lack of escaping)
Rich Felker [Mon, 14 May 2012 04:10:28 +0000 (00:10 -0400)]
useless lastlog path just to make some stuff happy
Rich Felker [Mon, 14 May 2012 04:01:48 +0000 (00:01 -0400)]
missing limit LOGIN_NAME_MAX
Rich Felker [Sun, 13 May 2012 23:59:28 +0000 (19:59 -0400)]
correct the check for gcc (previous version failed for cross compilers)
Rich Felker [Sun, 13 May 2012 21:20:01 +0000 (17:20 -0400)]
remove some no-op end of string tests from regex parser
these are cruft from the original code which used an explicit string
length rather than null termination. i blindly converted all the
checks to null terminator checks, without noticing that in several
cases, the subsequent switch statement would automatically handle the
null byte correctly.
Rich Felker [Sun, 13 May 2012 21:16:10 +0000 (17:16 -0400)]
another BRE fix: in ^*, * is literal
i don't understand why this has to be conditional on being in BRE
mode, but enabling this code unconditionally breaks a huge number of
ERE test cases.
Rich Felker [Sun, 13 May 2012 04:44:35 +0000 (00:44 -0400)]
let sysconf accurately report # of cpus available
i've been trying out openmp and it seems like it won't be much use
without this...
Rich Felker [Sun, 13 May 2012 03:45:07 +0000 (23:45 -0400)]
use __h_errno_location for h_errno
we do not bother making h_errno thread-local since the only interfaces
that use it are inherently non-thread-safe. but still use the
potentially-thread-local ABI to access it just to avoid lock-in.
Rich Felker [Sun, 13 May 2012 03:38:04 +0000 (23:38 -0400)]
susv4 removed gethostbyname, etc. legacy cruft.
Rich Felker [Sun, 13 May 2012 03:34:39 +0000 (23:34 -0400)]
namespace cleanup - NI_* is NOT reserved by netdb.h
Rich Felker [Sun, 13 May 2012 03:31:52 +0000 (23:31 -0400)]
some gnu junk in netdb.h
Rich Felker [Sun, 13 May 2012 02:18:34 +0000 (22:18 -0400)]
fix missing va_list for vsyslog
nsz [Sat, 12 May 2012 23:50:53 +0000 (01:50 +0200)]
search: add comments to tsearch_avl.c
nsz [Sat, 12 May 2012 23:34:20 +0000 (01:34 +0200)]
search: add tdestroy (gnu extension)
Rich Felker [Fri, 11 May 2012 15:05:41 +0000 (11:05 -0400)]
add missing IN6_ARE_ADDR_EQUAL
written to avoid multiple conditional jumps and avoid ugly repetitive
lines in the header file.
Rich Felker [Fri, 11 May 2012 03:32:28 +0000 (23:32 -0400)]
add one more bogus legacy header
this one is for program(s|ers) who haven't heard of uint16_t and
uint32_t (which are obviously the correct types for use in such
situations, as they're the argument/return types for ntohs/htons and
ntohl/htonl).
Rich Felker [Fri, 11 May 2012 02:41:54 +0000 (22:41 -0400)]
move vsyslog out of SYSLOG_NAMES conditional
Rich Felker [Fri, 11 May 2012 02:25:14 +0000 (22:25 -0400)]
Merge remote-tracking branch 'nsz/master'
Rich Felker [Fri, 11 May 2012 02:16:15 +0000 (22:16 -0400)]
remove __lock dependency from exit
there's no sense in using a powerful lock in exit, because it will
never be unlocked. a thread that arrives at exit while exit is already
in progress just needs to hang forever. use the pause syscall for this
because it's cheap and easy and universally available.
Rich Felker [Fri, 11 May 2012 01:51:36 +0000 (21:51 -0400)]
fix missing static (namespace clash)
Rich Felker [Thu, 10 May 2012 16:10:44 +0000 (12:10 -0400)]
fix missing parens in bit op macros (param.h)
Rich Felker [Thu, 10 May 2012 15:59:07 +0000 (11:59 -0400)]
and another bug in setbit, etc. macros..
Rich Felker [Thu, 10 May 2012 15:55:16 +0000 (11:55 -0400)]
fix typo in sys/param.h that broke setbit, etc. macros
this is all junk, but some programs use it.
Rich Felker [Wed, 9 May 2012 15:47:06 +0000 (11:47 -0400)]
omit declaration of basename wrongly interpreted as prototype in C++
the non-prototype declaration of basename in string.h is an ugly
compromise to avoid breaking 2 types of broken software:
1. programs which assume basename is declared in string.h and thus
would suffer from dangerous pointer-truncation if an implicit
declaration were used.
2. programs which include string.h with _GNU_SOURCE defined but then
declare their own prototype for basename using the incorrect GNU
signature for the function (which would clash with a correct
prototype).
however, since C++ does not have non-prototype declarations and
interprets them as prototypes for a function with no arguments, we
must omit it when compiling C++ code. thankfully, all known broken
apps that suffer from the above issues are written in C, not C++.
nsz [Mon, 7 May 2012 22:22:56 +0000 (00:22 +0200)]
math: fix remquo.c when x==-y and a subnormal remainder bug as well
backported fix from freebsd:
http://svnweb.FreeBSD.org/base?view=revision&revision=233973
Rich Felker [Mon, 7 May 2012 22:05:50 +0000 (18:05 -0400)]
some assemblers don't like fistpq; use the alt. mnemonic fistpll
Rich Felker [Mon, 7 May 2012 21:55:13 +0000 (17:55 -0400)]
fix error checking for \ at end of regex (this was broken previously)
Rich Felker [Mon, 7 May 2012 21:50:32 +0000 (17:50 -0400)]
fix copy and paste error in regex code causing mishandling of \) in BRE
Rich Felker [Mon, 7 May 2012 21:43:38 +0000 (17:43 -0400)]
fix regex breakage in last commit (failure to handle empty regex, etc.)
Rich Felker [Mon, 7 May 2012 18:50:49 +0000 (14:50 -0400)]
fix ugly bugs in TRE regex parser
1. * in BRE is not special at the beginning of the regex or a
subexpression. this broke ncurses' build scripts.
2. \\( in BRE is a literal \ followed by a literal (, not a literal \
followed by a subexpression opener.
3. the ^ in \\(^ in BRE is a literal ^ only at the beginning of the
entire BRE. POSIX allows treating it as an anchor at the beginning of
a subexpression, but TRE's code for checking if it was at the
beginning of a subexpression was wrong, and fixing it for the sake of
supporting a non-portable usage was too much trouble when just
removing this non-portable behavior was much easier.
this patch also moved lots of the ugly logic for empty atom checking
out of the default/literal case and into new cases for the relevant
characters. this should make parsing faster and make the code smaller.
if nothing else it's a lot more readable/logical.
at some point i'd like to revisit and overhaul lots of this code...
Rich Felker [Sun, 6 May 2012 21:19:37 +0000 (17:19 -0400)]
update release info for 0.9.0
Rich Felker [Sun, 6 May 2012 20:35:32 +0000 (16:35 -0400)]
some extra legacy header stuff
nsz [Sun, 6 May 2012 19:24:28 +0000 (21:24 +0200)]
add FORCE_EVAL macro to evaluate float expr for their side effect
updated nextafter* to use FORCE_EVAL, it can be used in many other
places in the math code to improve readability.
Rich Felker [Sun, 6 May 2012 18:48:20 +0000 (14:48 -0400)]
fix unused variable warnings in new nextafter/nexttoward code
apparently initializing a variable is not "using" it but assigning to
it is "using" it. i don't really like this fix, but it's better than
trying to make a bigger cleanup just before a release, and it should
work fine (tested against nsz's math tests).
Rich Felker [Sun, 6 May 2012 17:40:19 +0000 (13:40 -0400)]
take byte order from gcc if gcc has defined it
this only works with gcc 4.6 and later, but it allows us to support
non-default endianness on archs like arm, mips, ppc, etc. that can do
both without having separate header sets for both variants, and it
saves one #include even on fixed-endianness archs like x86.
Rich Felker [Sun, 6 May 2012 13:42:23 +0000 (09:42 -0400)]
Merge remote-tracking branch 'nsz/master'
Rich Felker [Sun, 6 May 2012 13:03:19 +0000 (09:03 -0400)]
add isastream (obsolete STREAMS junk)
apparently some packages see stropts.h and want to be able to use
this. the implementation checks that the file descriptor is valid by
using fcntl/F_GETFD so it can report an error if not (as specified).
nsz [Sun, 6 May 2012 11:08:59 +0000 (13:08 +0200)]
math: nextafter and nexttoward cleanup
make nexttoward, nexttowardf independent of long double representation.
fix nextafterl: it did not raise underflow flag when the result was 0.
Rich Felker [Sun, 6 May 2012 02:22:46 +0000 (22:22 -0400)]
fix definitions of FP_ILOGB constants
two issues: (1) the type was wrong (unsigned instead of signed int),
and (2) the value of FP_ILOGBNAN should be INT_MIN rather than INT_MAX
to match the ABI. this is also much more useful since INT_MAX
corresponds to a valid input (infinity). the standard would allow us
to set FP_ILOGB0 to -INT_MAX instead of INT_MIN, which would give us
distinct values for ilogb(0) and ilogb(NAN), but the benefit seems way
too small to justify ignoring the ABI.
note that the macro is just a "portable" (to any twos complement
system where signed and unsigned int have the same width) way to write
INT_MIN without needing limits.h. it's valid to use this method since
these macros are not required to work in #if directives.
Rich Felker [Sat, 5 May 2012 21:18:31 +0000 (17:18 -0400)]
add -frounding-math to build
Rich Felker [Sat, 5 May 2012 06:39:51 +0000 (02:39 -0400)]
update readme and release notes for 0.9.0 release (not yet final)
Rich Felker [Sat, 5 May 2012 04:16:21 +0000 (00:16 -0400)]
relicense musl under MIT license
Rich Felker [Sat, 5 May 2012 04:02:04 +0000 (00:02 -0400)]
update license of njk contributed code (x86_64 asm)
these changes are based on the following communication via email:
"I hereby grant that all of the code I have contributed to musl on or
before April 23, 2012 may be licensed under the terms of the following
MIT license:
Copyright (c) 2011-2012 Nicholas J. Kain
Permission is hereby granted, free of charge, to any person obtaining
a copy of this software and associated documentation files (the
"Software"), to deal in the Software without restriction, including
without limitation the rights to use, copy, modify, merge, publish,
distribute, sublicense, and/or sell copies of the Software, and to
permit persons to whom the Software is furnished to do so, subject to
the following conditions:
The above copyright notice and this permission notice shall be
included in all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY
CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE."
Rich Felker [Sat, 5 May 2012 03:53:50 +0000 (23:53 -0400)]
update INSTALL file to reflect configure script usage
Rich Felker [Sat, 5 May 2012 03:24:51 +0000 (23:24 -0400)]
initial commit of configure script
this script is not based on autoconf; however it attempts to follow
the same interface contracts for ease of integration with build
systems. it is also not necessary to use musl. manually written
config.mak files are still supported, as is building without any
config.mak at all as long as you are happy with the default options
and you supply at least ARCH on the command line to make.
Rich Felker [Sat, 5 May 2012 02:51:59 +0000 (22:51 -0400)]
make pthread stacks non-executable
this change is necessary or pthread_create will always fail on
security-hardened kernels. i considered first trying to make the stack
executable and simply retrying without execute permissions when the
first try fails, but (1) this would incur a serious performance
penalty on hardened systems, and (2) having the stack be executable is
just a bad idea from a security standpoint.
if there is real-world "GNU C" code that uses nested functions with
threads, and it can't be fixed, we'll have to consider other ways of
solving the problem, but for now this seems like the best fix.
Rich Felker [Sat, 5 May 2012 01:54:57 +0000 (21:54 -0400)]
improve the build rules for installing /lib/ld-musl-$ARCH.so
these new rules should avoid spurious error messages when the
directory (usually /lib) and the dynamic linker symlink already exist,
and minimize the spam when they can't be created.
Rich Felker [Sat, 5 May 2012 00:56:01 +0000 (20:56 -0400)]
Merge remote-tracking branch 'nsz/master'
Rich Felker [Sat, 5 May 2012 00:18:18 +0000 (20:18 -0400)]
fix error reporting for dlsym with global symbols
nsz [Fri, 4 May 2012 23:11:56 +0000 (01:11 +0200)]
math: change the formula used for acos.s
old: 2*atan2(sqrt(1-x),sqrt(1+x))
new: atan2(fabs(sqrt((1-x)*(1+x))),x)
improvements:
* all edge cases are fixed (sign of zero in downward rounding)
* a bit faster (here a single call is about 131ns vs 162ns)
* a bit more precise (at most 1ulp error on 1M uniform random
samples in [0,1), the old formula gave some 2ulp errors as well)
Rich Felker [Fri, 4 May 2012 05:26:43 +0000 (01:26 -0400)]
fix uninitialized var in vfwprintf printing 0-prec string
this could lead to spurious failures of wide printf functions
Rich Felker [Fri, 4 May 2012 04:31:25 +0000 (00:31 -0400)]
add *64 junk for sys/*.h headers
Rich Felker [Fri, 4 May 2012 04:13:23 +0000 (00:13 -0400)]
add support for ugly *64 functions with _LARGEFILE64_SOURCE
musl does not support legacy 32-bit-off_t whatsoever. off_t is always
64 bit, and correct programs that use off_t and the standard functions
will just work out of the box. (on glibc, they would require
-D_FILE_OFFSET_BITS=64 to work.) however, some programs instead define
_LARGEFILE64_SOURCE and use alternate versions of all the standard
types and functions with "64" appended to their names.
we do not want code to actually get linked against these functions
(it's ugly and inconsistent), so macros are used instead of prototypes
with weak aliases in the library itself. eventually the weak aliases
may be added at the library level for the sake of using code that was
originally built against glibc, but the macros will still be the
desired solution in the headers.
Rich Felker [Fri, 4 May 2012 03:19:29 +0000 (23:19 -0400)]
avoid setting nondefault scheduler too
Rich Felker [Fri, 4 May 2012 03:18:26 +0000 (23:18 -0400)]
implement stub versions of sched_*
these actually work, but for now they prohibit actually setting
priority levels and report min/max priority as 0.
Rich Felker [Fri, 4 May 2012 02:27:36 +0000 (22:27 -0400)]
uglify headers for the sake of junk that compiles with gcc -std=c89/-ansi
Rich Felker [Fri, 4 May 2012 02:12:46 +0000 (22:12 -0400)]
add additional compatibility union member for ipv6 addresses
in6_* is in the reserved namespace, so this is valid
Rich Felker [Fri, 4 May 2012 00:42:45 +0000 (20:42 -0400)]
overhaul SSP support to use a real canary
pthread structure has been adjusted to match the glibc/GCC abi for
where the canary is stored on i386 and x86_64. it will need variants
for other archs to provide the added security of the canary's entropy,
but even without that it still works as well as the old "minimal" ssp
support. eventually such changes will be made anyway, since they are
also needed for GCC/C11 thread-local storage support (not yet
implemented).
care is taken not to attempt initializing the thread pointer unless
the program actually uses SSP (by reference to __stack_chk_fail).
Rich Felker [Fri, 4 May 2012 00:35:11 +0000 (20:35 -0400)]
make all .o files depend on *_impl.h and libc.h
hopefully the annoyance of this will be minimal. these files all
define internal interfaces which can change at any time; if different
modules are using different versions of the interfaces, the library
will badly break. ideally we would scan and add the dependency only
for C files that actually reference the affected interfaces, but for
now, err on the side of caution and force a rebuild of everything if
any of them have changed.
this commit is in preparation for the upcoming ssp overhaul commit,
which will change internals of the pthread struct.
Rich Felker [Thu, 3 May 2012 01:16:02 +0000 (21:16 -0400)]
remove some junk from x86_64 start files
looks like nik copied these "extra arguments" from the i386 code.
they're not actually arguments there, just 1-byte instructions to
make sure the stack is aligned to 16 bytes after all the other
arguments are pushed. since each push is 8 bytes on x86_64, they
happened to have no effect here, but their presence is confusing and a
minor waste of space.
Rich Felker [Thu, 3 May 2012 01:14:41 +0000 (21:14 -0400)]
PIE support for x86_64 (untested)
Rich Felker [Thu, 3 May 2012 01:03:25 +0000 (21:03 -0400)]
consistency cleanup: removed redundant size suffixed from i386 asm
Rich Felker [Thu, 3 May 2012 01:01:55 +0000 (21:01 -0400)]
PIE support for i386
Rich Felker [Thu, 3 May 2012 00:56:55 +0000 (20:56 -0400)]
remove attempts to be pie-compatible from i386 crt1.s
it does not work; after further consideration, a separate Scrt1.s for
pie really is essential. it would be nice if the unified approach
worked, but the linker fails to generate the correct PLT entries and
instead puts textrels in the main program, which don't work because
the kernel maps the text read-only.
new Scrt1.s will be committed soon in place of this.
Rich Felker [Wed, 2 May 2012 17:59:48 +0000 (13:59 -0400)]
fix longstanding exit logic bugs in mbsnrtowcs and wcsnrtombs
these are POSIX 2008 (previously GNU extension) functions that are
rarely used. apparently they had never been tested before, since the
end-of-string logic was completely missing. mbsnrtowcs is used by
modern versions of bash for its glob implementation, and and this bug
was causing tab completion to hang in an infinite loop.
Rich Felker [Wed, 2 May 2012 01:20:06 +0000 (21:20 -0400)]
remove minimal linux kernel headers
these were at best of limited usefulness (for bootstrapping new
systems, mainly) and at worst caused real kernel headers to get
overwritten when upgrading libc.
in case they're needed by anyone, the exact same files are now
available in a new git repository:
git://git.etalabs.net/mini-lkh
Rich Felker [Tue, 1 May 2012 23:30:03 +0000 (19:30 -0400)]
reorganize Makefile to support "least surprise" config/make semantics
the major change here is that CFLAGS is now a variable that can be
changed entirely under user control, without causing essential flags
to be lost. previously, "CFLAGS += ..." was valid in config.mak, but
using "CFLAGS = ..." in config.mak would have badly broken the build
process unless the user took care to copy the necessary flags out of
the main Makefile.
I have also added a distclean target that removes config.mak.
Rich Felker [Tue, 1 May 2012 18:31:55 +0000 (14:31 -0400)]
remove objcopy --weaken from the makefile
as far as I can tell, it's not useful and never way. I wrote it way
back under the assumption that non-weak symbols in the POSIX or
extension namespace could conflict with legitimate uses of the same
symbol name in the main program or other libraries, but that does not
seem to be the case.
Rich Felker [Tue, 1 May 2012 04:07:37 +0000 (00:07 -0400)]
support alternate glibc name pow10 for exp10
Rich Felker [Tue, 1 May 2012 01:32:19 +0000 (21:32 -0400)]
add C stub for sqrtl (already implemented in asm on i386 and x86_64)
Rich Felker [Mon, 30 Apr 2012 07:26:53 +0000 (03:26 -0400)]
first try at writing an efficient and "correct" exp10
this is a nonstandard function so it's not clear what conditions it
should satisfy. my intent is that it be fast and exact for positive
integral exponents when the result fits in the destination type, and
fast and correctly rounded for small negative integral exponents.
otherwise we aim for at most 1ulp error; it seems to differ from pow
by at most 1ulp and it's often 2-5 times faster than pow.
Rich Felker [Mon, 30 Apr 2012 07:00:24 +0000 (03:00 -0400)]
make stack protector work with gcc configured for non-tls canary
Rich Felker [Mon, 30 Apr 2012 06:56:47 +0000 (02:56 -0400)]
fix off-by-one error that caused uninitialized memory read in floatscan
this caused misreading of certain floating point values that are exact
multiples of large powers of ten, unpredictable depending on prior
stack contents.
Rich Felker [Mon, 30 Apr 2012 00:36:32 +0000 (20:36 -0400)]
fix typo in the x86_64 rounding asm
Rich Felker [Mon, 30 Apr 2012 00:31:46 +0000 (20:31 -0400)]
new math asm (abs/rounding) for x86_64
untested
Rich Felker [Sun, 29 Apr 2012 23:54:29 +0000 (19:54 -0400)]
fix float_t and double_t defs on x86 when -mfpmath=sse -msse2 is used
Rich Felker [Sun, 29 Apr 2012 23:54:03 +0000 (19:54 -0400)]
add linux-specific unshare syscall wrapper
Rich Felker [Sun, 29 Apr 2012 04:20:53 +0000 (00:20 -0400)]
fix longstanding missing static in mq_notify (namespace pollution)
Rich Felker [Sat, 28 Apr 2012 22:05:29 +0000 (18:05 -0400)]
new fnmatch implementation
unlike the old one, this one's algorithm does not suffer from
potential stack overflow issues or pathologically bad performance on
certain patterns. instead of backtracking, it uses a matching
algorithm which I have not seen before (unsure whether I invented or
re-invented it) that runs in O(1) space and O(nm) time. it may be
possible to improve the time to O(n), but not without significantly
greater complexity.
Rich Felker [Fri, 27 Apr 2012 05:39:03 +0000 (01:39 -0400)]
support FLT_EVAL_METHOD changing on x86 with gcc -msse2 -mfpmath=sse
if the compiler provides a value, use it; otherwise fallback to the
platform default (2).
Rich Felker [Thu, 26 Apr 2012 16:24:44 +0000 (12:24 -0400)]
update fnmatch to POSIX 2008 semantics
an invalid bracket expression must be treated as if the opening
bracket were just a literal character. this is to fix a bug whereby
POSIX left the behavior of the "[" shell command undefined due to it
being an invalid bracket expression.
Rich Felker [Wed, 25 Apr 2012 18:35:51 +0000 (14:35 -0400)]
release notes for 0.8.10
Rich Felker [Wed, 25 Apr 2012 04:05:42 +0000 (00:05 -0400)]
gdb shared library debugging support
provide the minimal level of dynamic linker-to-debugger glue needed to
let gdb find loaded libraries and load their symbols.
Rich Felker [Tue, 24 Apr 2012 22:07:59 +0000 (18:07 -0400)]
first attempt at enabling stack protector support
the code is written to pre-init the thread pointer in static linked
programs that pull in __stack_chk_fail or dynamic-linked programs that
lookup the symbol. no explicit canary is set; the canary will be
whatever happens to be in the thread structure at the offset gcc
hard-coded. this can be improved later.
Rich Felker [Tue, 24 Apr 2012 22:06:56 +0000 (18:06 -0400)]
use signed char rather than plain char for int8_t
otherwise this BADLY breaks if -funsigned-char is passed to gcc
Rich Felker [Tue, 24 Apr 2012 20:49:11 +0000 (16:49 -0400)]
add another example option to dist/config.mak
Rich Felker [Tue, 24 Apr 2012 20:32:23 +0000 (16:32 -0400)]
ditch the priority inheritance locks; use malloc's version of lock
i did some testing trying to switch malloc to use the new internal
lock with priority inheritance, and my malloc contention test got
20-100 times slower. if priority inheritance futexes are this slow,
it's simply too high a price to pay for avoiding priority inversion.
maybe we can consider them somewhere down the road once the kernel
folks get their act together on this (and perferably don't link it to
glibc's inefficient lock API)...
as such, i've switch __lock to use malloc's implementation of
lightweight locks, and updated all the users of the code to use an
array with a waiter count for their locks. this should give optimal
performance in the vast majority of cases, and it's simple.
malloc is still using its own internal copy of the lock code because
it seems to yield measurably better performance with -O3 when it's
inlined (20% or more difference in the contention stress test).
Rich Felker [Tue, 24 Apr 2012 17:55:06 +0000 (13:55 -0400)]
internal locks: new owner of contended lock must set waiters flag
this bug probably would have gone unnoticed since it's only used in
the fallback code for systems where priority-inheritance locking
fails. unfortunately this approach results in one spurious wake
syscall on the final unlock, when there are no waiters remaining. the
alternative (possibly better) would be to use broadcast wakes instead
of reflagging the waiter unconditionally, and let each waiter reflag
itself; this saves one syscall at the expense of invoking the
"thundering herd" effect (worse performance degredation) when there
are many waiters.
ideally we would be able to update all of our locks to use an array of
two ints rather than a single int, and use a separate counter system
like proper mutexes use; then we could avoid all spurious wake calls
without resorting to broadcasts. however, it's not clear to me that
priority inheritance futexes support this usage. the kernel sets the
waiters flag for them (just like we're doing now) and i can't tell if
it's safe to bypass the kernel when unlocking just because we know
(from private data, the waiter count) that there are no waiters. this
is something that could be explored in the future.
Rich Felker [Tue, 24 Apr 2012 10:36:50 +0000 (06:36 -0400)]
new internal locking primitive; drop spinlocks
we use priority inheritance futexes if possible so that the library
cannot hit internal priority inversion deadlocks in the presence of
realtime priority scheduling (full support to be added later).
Rich Felker [Tue, 24 Apr 2012 08:23:55 +0000 (04:23 -0400)]
new wcwidth implementation (fast table-based)
i tried to go with improving the old binary-search-based algorithm,
but between growth in the number of ranges, bad performance, and lack
of confidence in the binary search code's stability under changes in
the table, i decided it was worth the extra 1.8k to have something
clean and maintainable.
also note that, like the alpha and punct tables, there's definitely
room to optimize the nonspacing/wide tables by overlapping subtables.
this is not a high priority, but i've begun looking into how to do it,
and i suspect the table sizes can be roughly halved. if that turns out
to be true, the new, fast, table-based implementation will be roughly
the same size as if i had just extended the old binary search one.
Rich Felker [Mon, 23 Apr 2012 23:19:26 +0000 (19:19 -0400)]
sync case mappings with unicode 6.1
also special-case ß (U+00DF) as lowercase even though it does not have
a mapping to uppercase. unicode added an uppercase version of this
character but does not map it, presumably because the uppercase
version is not actually used except for some obscure purpose...
Rich Felker [Mon, 23 Apr 2012 20:10:36 +0000 (16:10 -0400)]
optimize iswprint