musl
11 years agofix unused variable warnings in new nextafter/nexttoward code
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).

11 years agotake byte order from gcc if gcc has defined it
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.

11 years agoMerge remote-tracking branch 'nsz/master'
Rich Felker [Sun, 6 May 2012 13:42:23 +0000 (09:42 -0400)]
Merge remote-tracking branch 'nsz/master'

11 years agoadd isastream (obsolete STREAMS junk)
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).

11 years agomath: nextafter and nexttoward cleanup
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.

11 years agofix definitions of FP_ILOGB constants
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.

11 years agoadd -frounding-math to build
Rich Felker [Sat, 5 May 2012 21:18:31 +0000 (17:18 -0400)]
add -frounding-math to build

11 years agoupdate readme and release notes for 0.9.0 release (not yet final)
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)

11 years agorelicense musl under MIT license
Rich Felker [Sat, 5 May 2012 04:16:21 +0000 (00:16 -0400)]
relicense musl under MIT license

11 years agoupdate license of njk contributed code (x86_64 asm)
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."

11 years agoupdate INSTALL file to reflect configure script usage
Rich Felker [Sat, 5 May 2012 03:53:50 +0000 (23:53 -0400)]
update INSTALL file to reflect configure script usage

11 years agoinitial commit of configure script
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.

11 years agomake pthread stacks non-executable
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.

11 years agoimprove the build rules for installing /lib/ld-musl-$ARCH.so
Rich Felker [Sat, 5 May 2012 01:54:57 +0000 (21:54 -0400)]
improve the build rules for installing /lib/ld-musl-$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.

11 years agoMerge remote-tracking branch 'nsz/master'
Rich Felker [Sat, 5 May 2012 00:56:01 +0000 (20:56 -0400)]
Merge remote-tracking branch 'nsz/master'

11 years agofix error reporting for dlsym with global symbols
Rich Felker [Sat, 5 May 2012 00:18:18 +0000 (20:18 -0400)]
fix error reporting for dlsym with global symbols

11 years agomath: change the formula used for acos.s
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)

11 years agofix uninitialized var in vfwprintf printing 0-prec string
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

11 years agoadd *64 junk for sys/*.h headers
Rich Felker [Fri, 4 May 2012 04:31:25 +0000 (00:31 -0400)]
add *64 junk for sys/*.h headers

11 years agoadd support for ugly *64 functions with _LARGEFILE64_SOURCE
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.

11 years agoavoid setting nondefault scheduler too
Rich Felker [Fri, 4 May 2012 03:19:29 +0000 (23:19 -0400)]
avoid setting nondefault scheduler too

11 years agoimplement stub versions of sched_*
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.

11 years agouglify headers for the sake of junk that compiles with gcc -std=c89/-ansi
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

11 years agoadd additional compatibility union member for ipv6 addresses
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

11 years agooverhaul SSP support to use a real canary
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).

11 years agomake all .o files depend on *_impl.h and libc.h
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.

11 years agoremove some junk from x86_64 start files
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.

11 years agoPIE support for x86_64 (untested)
Rich Felker [Thu, 3 May 2012 01:14:41 +0000 (21:14 -0400)]
PIE support for x86_64 (untested)

11 years agoconsistency cleanup: removed redundant size suffixed from i386 asm
Rich Felker [Thu, 3 May 2012 01:03:25 +0000 (21:03 -0400)]
consistency cleanup: removed redundant size suffixed from i386 asm

11 years agoPIE support for i386
Rich Felker [Thu, 3 May 2012 01:01:55 +0000 (21:01 -0400)]
PIE support for i386

11 years agoremove attempts to be pie-compatible from i386 crt1.s
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.

11 years agofix longstanding exit logic bugs in mbsnrtowcs and wcsnrtombs
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.

11 years agoremove minimal linux kernel headers
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

11 years agoreorganize Makefile to support "least surprise" config/make semantics
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.

11 years agoremove objcopy --weaken from the makefile
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.

11 years agosupport alternate glibc name pow10 for exp10
Rich Felker [Tue, 1 May 2012 04:07:37 +0000 (00:07 -0400)]
support alternate glibc name pow10 for exp10

11 years agoadd C stub for sqrtl (already implemented in asm on i386 and x86_64)
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)

11 years agofirst try at writing an efficient and "correct" exp10
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.

11 years agomake stack protector work with gcc configured for non-tls canary
Rich Felker [Mon, 30 Apr 2012 07:00:24 +0000 (03:00 -0400)]
make stack protector work with gcc configured for non-tls canary

11 years agofix off-by-one error that caused uninitialized memory read in floatscan
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.

11 years agofix typo in the x86_64 rounding asm
Rich Felker [Mon, 30 Apr 2012 00:36:32 +0000 (20:36 -0400)]
fix typo in the x86_64 rounding asm

11 years agonew math asm (abs/rounding) for x86_64
Rich Felker [Mon, 30 Apr 2012 00:31:46 +0000 (20:31 -0400)]
new math asm (abs/rounding) for x86_64

untested

11 years agofix float_t and double_t defs on x86 when -mfpmath=sse -msse2 is used
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

11 years agoadd linux-specific unshare syscall wrapper
Rich Felker [Sun, 29 Apr 2012 23:54:03 +0000 (19:54 -0400)]
add linux-specific unshare syscall wrapper

11 years agofix longstanding missing static in mq_notify (namespace pollution)
Rich Felker [Sun, 29 Apr 2012 04:20:53 +0000 (00:20 -0400)]
fix longstanding missing static in mq_notify (namespace pollution)

11 years agonew fnmatch implementation
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.

11 years agosupport FLT_EVAL_METHOD changing on x86 with gcc -msse2 -mfpmath=sse
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).

11 years agoupdate fnmatch to POSIX 2008 semantics
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.

11 years agorelease notes for 0.8.10
Rich Felker [Wed, 25 Apr 2012 18:35:51 +0000 (14:35 -0400)]
release notes for 0.8.10

11 years agogdb shared library debugging support
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.

11 years agofirst attempt at enabling stack protector support
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.

11 years agouse signed char rather than plain char for int8_t
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

11 years agoadd another example option to dist/config.mak
Rich Felker [Tue, 24 Apr 2012 20:49:11 +0000 (16:49 -0400)]
add another example option to dist/config.mak

11 years agoditch the priority inheritance locks; use malloc's version of lock
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).

11 years agointernal locks: new owner of contended lock must set waiters flag
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.

11 years agonew internal locking primitive; drop spinlocks
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).

11 years agonew wcwidth implementation (fast table-based)
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.

11 years agosync case mappings with unicode 6.1
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...

11 years agooptimize iswprint
Rich Felker [Mon, 23 Apr 2012 20:10:36 +0000 (16:10 -0400)]
optimize iswprint

11 years agofix spurious punct class for some surrogate codepoints (invalid)
Rich Felker [Mon, 23 Apr 2012 20:02:46 +0000 (16:02 -0400)]
fix spurious punct class for some surrogate codepoints (invalid)

this happened due to their entries in UnicodeData.txt

11 years agodestubify iswalpha and update iswpunct to unicode 6.1
Rich Felker [Mon, 23 Apr 2012 19:25:23 +0000 (15:25 -0400)]
destubify iswalpha and update iswpunct to unicode 6.1

alpha is defined as unicode property "Alphabetic" plus category Nd
minus ASCII digits minus 2 special-cased Thai punctuation marks
supposedly misclassified by Unicode as letters.

punct is defined as all of unicode except control, alphanumeric, and
space characters.

the tables were generated by a simple tool based on the code posted
previously to the mailing list. in the future, this and other code
used for maintaining locale/iconv/i18n data will be published either
in the main source repository or in a separate locale data generation
repository.

11 years agomake dlerror produce informative results
Rich Felker [Mon, 23 Apr 2012 16:03:31 +0000 (12:03 -0400)]
make dlerror produce informative results

note that dlerror is specified to be non-thread-safe, so no locking is
performed on the error flag or message aside from the rwlock already
held by dlopen or dlsym. if 2 invocations of dlsym are generating
errors at the same time, they could clobber each other's results, but
the resulting string, albeit corrupt, will still be null-terminated.
any use of dlerror in such a situation could not be expected to give
meaningful results anyway.

11 years agoimplement getusershell, etc. legacy functions
Rich Felker [Sun, 22 Apr 2012 18:41:54 +0000 (14:41 -0400)]
implement getusershell, etc. legacy functions

I actually wrote these a month ago but forgot to integrate them. ugly,
probably-harmful-to-use functions, but some legacy apps want them...

11 years agogetdtablesize is not standard; move it to its correct spot in unistd.h
Rich Felker [Sun, 22 Apr 2012 18:39:07 +0000 (14:39 -0400)]
getdtablesize is not standard; move it to its correct spot in unistd.h

11 years agonew gcc wrapper, entirely specfile based
Rich Felker [Sun, 22 Apr 2012 18:32:49 +0000 (14:32 -0400)]
new gcc wrapper, entirely specfile based

the _concept_ of this wrapper has been tested extensively, but the
integration with the build/install system, and using a persistent
specfile rather than one generated at build-time, have not been
heavily tested and may need minor tweaks.

this approach should be a lot more robust (and easier to improve) than
writing a shell script that's responsible for trying to mimic gcc's
logic about whether it's compiling or linking, building shared libs or
executable files, etc. it's also lighter weight and should result in
mildly faster builds when using the wrapper.

11 years agoremove redundant (unmaintained) check in floatscan
Rich Felker [Sun, 22 Apr 2012 18:05:12 +0000 (14:05 -0400)]
remove redundant (unmaintained) check in floatscan

also be extra careful to avoid wrapping the circular buffer early

11 years agofix breakage in endian.h
Rich Felker [Sun, 22 Apr 2012 15:19:17 +0000 (11:19 -0400)]
fix breakage in endian.h

11 years agoadd some ugly byte swapping cruft in endian.h
Rich Felker [Sun, 22 Apr 2012 15:08:01 +0000 (11:08 -0400)]
add some ugly byte swapping cruft in endian.h

11 years agoadd getresuid and getresgid syscall wrappers
Rich Felker [Sun, 22 Apr 2012 14:37:19 +0000 (10:37 -0400)]
add getresuid and getresgid syscall wrappers

11 years agofix major breakage in iconv, bogus rejecting of dest charsets
Rich Felker [Sat, 21 Apr 2012 18:46:40 +0000 (14:46 -0400)]
fix major breakage in iconv, bogus rejecting of dest charsets

11 years agomake floatscan correctly set errno for overflow/underflow
Rich Felker [Sat, 21 Apr 2012 18:14:10 +0000 (14:14 -0400)]
make floatscan correctly set errno for overflow/underflow

care is taken that the setting of errno correctly reflects underflow
condition. scanning exact denormal values does not result in ERANGE,
nor does scanning values (such as the usual string definition of
FLT_MIN) which are actually less than the smallest normal number but
which round to a normal result.

only the decimal case is handled so far; hex float require a separate
fix to come later.

11 years agoskip leading zeros even after decimal point in floatscan
Rich Felker [Sat, 21 Apr 2012 17:50:23 +0000 (13:50 -0400)]
skip leading zeros even after decimal point in floatscan

in principle this should just be an optimization, but it happens to
also fix a nasty bug where values like 0.00000000001 were getting
caught by the early zero detection path and wrongly scanned as zero.

11 years agofix overread (consuming an extra byte) scanning NAN
Rich Felker [Sat, 21 Apr 2012 15:57:39 +0000 (11:57 -0400)]
fix overread (consuming an extra byte) scanning NAN

bug detected by glib test suite

11 years agofix broken sysconf when correct value is -1
Rich Felker [Sat, 21 Apr 2012 15:45:07 +0000 (11:45 -0400)]
fix broken sysconf when correct value is -1

this caused glib to try to allocate >2gb for getpwnam_r, and probably
numerous other problems.

12 years agorelease notes for 0.8.9 (bugfix release)
Rich Felker [Fri, 20 Apr 2012 02:02:50 +0000 (22:02 -0400)]
release notes for 0.8.9 (bugfix release)

12 years agofurther fixes to leading space issue (forgot the wide versions)
Rich Felker [Thu, 19 Apr 2012 16:56:29 +0000 (12:56 -0400)]
further fixes to leading space issue (forgot the wide versions)

12 years agofix really bad breakage in strtol, etc.: failure to accept leading spaces
Rich Felker [Thu, 19 Apr 2012 16:47:34 +0000 (12:47 -0400)]
fix really bad breakage in strtol, etc.: failure to accept leading spaces

12 years agofix header typo
Rich Felker [Wed, 18 Apr 2012 17:11:35 +0000 (13:11 -0400)]
fix header typo

12 years agolegacy junk compatibility grab-bag
Rich Felker [Wed, 18 Apr 2012 16:22:24 +0000 (12:22 -0400)]
legacy junk compatibility grab-bag

- add the rest of the junk traditionally in sys/param.h
- add prototypes for some nonstandard functions
- add _GNU_SOURCE to their source files so the compiler can check proto

12 years agofix incorrect macro name for MATH_ERREXCEPT in math.h
Rich Felker [Wed, 18 Apr 2012 15:41:04 +0000 (11:41 -0400)]
fix incorrect macro name for MATH_ERREXCEPT in math.h

12 years agorelease notes for 0.8.8
Rich Felker [Wed, 18 Apr 2012 08:40:06 +0000 (04:40 -0400)]
release notes for 0.8.8

12 years agofix typo in exponent reading code or floats
Rich Felker [Wed, 18 Apr 2012 07:53:53 +0000 (03:53 -0400)]
fix typo in exponent reading code or floats

this was basically harmless, but could have resulted in misreading
inputs with more than a few gigabytes worth of digits..

12 years agofix wide scanf's handling of input failure on %c, and simplify %[
Rich Felker [Wed, 18 Apr 2012 03:35:49 +0000 (23:35 -0400)]
fix wide scanf's handling of input failure on %c, and simplify %[

12 years agofix failure to distinguish input/match failure in wide %[ scanf
Rich Felker [Wed, 18 Apr 2012 03:08:58 +0000 (23:08 -0400)]
fix failure to distinguish input/match failure in wide %[ scanf

this also includes a related fix for vswscanf's read function, which
was returning a spurious (uninitialized) character for empty strings.

12 years agofix over-read in %ls with non-wide scanf
Rich Felker [Wed, 18 Apr 2012 02:41:38 +0000 (22:41 -0400)]
fix over-read in %ls with non-wide scanf

12 years agofix broken %s and %[ with no width specifier in wide scanf
Rich Felker [Wed, 18 Apr 2012 02:15:33 +0000 (22:15 -0400)]
fix broken %s and %[ with no width specifier in wide scanf

12 years agofix failure to read infinity in scanf
Rich Felker [Wed, 18 Apr 2012 02:05:51 +0000 (22:05 -0400)]
fix failure to read infinity in scanf

this code worked in strtod, but not in scanf. more evidence that i
should design a better interface for discarding multiple tail
characters than just calling unget repeatedly...

12 years agofix failure of int parser to unget an initial mismatching character
Rich Felker [Wed, 18 Apr 2012 01:17:19 +0000 (21:17 -0400)]
fix failure of int parser to unget an initial mismatching character

12 years agomake wide scanf %[ respect width
Rich Felker [Wed, 18 Apr 2012 01:17:09 +0000 (21:17 -0400)]
make wide scanf %[ respect width

12 years agofix wide scanf to respect field width for strings
Rich Felker [Tue, 17 Apr 2012 23:37:31 +0000 (19:37 -0400)]
fix wide scanf to respect field width for strings

12 years agofix some bugs in scanf %[ handling detected while writing the wide version
Rich Felker [Tue, 17 Apr 2012 18:22:22 +0000 (14:22 -0400)]
fix some bugs in scanf %[ handling detected while writing the wide version

12 years agointroduce new wide scanf code and remove the last remnants of old scanf
Rich Felker [Tue, 17 Apr 2012 18:19:46 +0000 (14:19 -0400)]
introduce new wide scanf code and remove the last remnants of old scanf

at this point, strto* and all scanf family functions are using the new
unified integer and floating point parser/converter code.

the wide scanf is largely a wrapper for ordinary byte-based scanf;
since numbers can only contain ascii characters, only strings need to
be handled specially.

12 years agoavoid depending on POSIX symbol in code used from plain C functions
Rich Felker [Tue, 17 Apr 2012 17:17:01 +0000 (13:17 -0400)]
avoid depending on POSIX symbol in code used from plain C functions

12 years agoavoid null pointer dereference on %*p fields in scanf
Rich Felker [Tue, 17 Apr 2012 15:50:02 +0000 (11:50 -0400)]
avoid null pointer dereference on %*p fields in scanf

12 years agoalso ensure that write buffer is bounded when __stdio_write returns
Rich Felker [Tue, 17 Apr 2012 15:08:11 +0000 (11:08 -0400)]
also ensure that write buffer is bounded when __stdio_write returns

assuming other code is correct, this should be a no-op, but better to
be safe...

12 years agofix buffer overflow in vfprintf on long writes to unbuffered files
Rich Felker [Tue, 17 Apr 2012 14:58:02 +0000 (10:58 -0400)]
fix buffer overflow in vfprintf on long writes to unbuffered files

vfprintf temporarily swaps in a local buffer (for the duration of the
operation) when the target stream is unbuffered; this both simplifies
the implementation of functions like dprintf (they don't need their
own buffers) and eliminates the pathologically bad performance of
writing the formatted output with one or more write syscalls per
formatting field.

in cases like dprintf where we are dealing with a virgin FILE
structure, everything worked correctly. however for long-lived files
(like stderr), it's possible that the buffer bounds were already set
for the internal zero-size buffer. on the next write, __stdio_write
would pick up and use the new buffer provided by vfprintf, but the
bound (wend) field was still pointing at the internal zero-size
buffer's end. this in turn allowed unbounded writes to the temporary
buffer.

12 years agofix %lf, etc. with printf
Rich Felker [Tue, 17 Apr 2012 01:50:23 +0000 (21:50 -0400)]
fix %lf, etc. with printf

the l prefix is redundant/no-op with printf, since default promotions
always promote floats to double; however, it is valid, and printf was
wrongly rejecting it.

12 years agobetter description for errno==0
Rich Felker [Mon, 16 Apr 2012 22:37:53 +0000 (18:37 -0400)]
better description for errno==0

12 years agoimplement wcstod and family
Rich Felker [Mon, 16 Apr 2012 21:26:54 +0000 (17:26 -0400)]
implement wcstod and family

not heavily tested but these functions appear to work correctly

12 years agoavoid hitting eof in wcstol
Rich Felker [Mon, 16 Apr 2012 21:17:05 +0000 (17:17 -0400)]
avoid hitting eof in wcstol

shunget cannot unget eof status, causing wcstol to leave endptr
pointing to the wrong place when scanning, for example, L"0x". cheap
fix is to make the read function provide an infinite stream of bogus
characters rather than eof. really this is something of a design flaw
in how the shgetc system is used for strto* and wcsto*; in the long
term, I believe multi-character unget should be scrapped and replaced
with a function that can subtract from the f->shcnt counter.