Rich Felker [Sat, 2 Nov 2013 00:34:05 +0000 (20:34 -0400)]
simplify faccessat AT_EACCESS path and eliminate resource dependence
now that we're waiting for the exit status of the child process, the
result can be conveyed in the exit status rather than via a pipe.
since the error value might not fit in 7 bits, a table is used to
translate possible meaningful error values to small integers.
Rich Felker [Fri, 1 Nov 2013 21:01:52 +0000 (17:01 -0400)]
fix faccessat AT_EACCESS path not to leave zombie processes
I mistakenly assumed that clone without a signal produced processes
that would not become zombies; however, waitpid with __WCLONE is
required to release their pids.
Rich Felker [Wed, 30 Oct 2013 18:52:55 +0000 (14:52 -0400)]
fix inttypes.h PRI and SCN macros for [u]intptr_t types
while using "l" unconditionally gave the right behavior due to
matching sizes/representations, it was technically UB and produced
compiler warnings with format string checking.
Szabolcs Nagy [Tue, 29 Oct 2013 20:25:11 +0000 (20:25 +0000)]
POSIX conformance fix: define struct entry in search.h
Szabolcs Nagy [Mon, 28 Oct 2013 20:33:54 +0000 (20:33 +0000)]
fenv: fix i386 fesetround for sse
i386 fenv code checks __hwcap for sse support, but in fesetround the sse
code was unconditionally jumped over after the test so the sse rounding
mode was never set.
Szabolcs Nagy [Mon, 28 Oct 2013 01:16:14 +0000 (01:16 +0000)]
math: extensive log*.c cleanup
The log, log2 and log10 functions share a lot of code and to a lesser
extent log1p too. A small part of the code was kept separately in
__log1p.h, but since it did not capture much of the common code and
it was inlined anyway, it did not solve the issue properly. Now the
log functions have significant code duplication, which may be resolved
later, until then they need to be modified together.
logl, log10l, log2l, log1pl:
* Fix the sign when the return value should be -inf.
* Remove the volatile hack from log10l (seems unnecessary)
log1p, log1pf:
* Change the handling of small inputs: only |x|<2^-53 is special
(then it is enough to return x with the usual subnormal handling)
this fixes the sign of log1p(0) in downward rounding.
* Do not handle the k==0 case specially (other than skipping the
elaborate argument reduction)
* Do not handle 1+x close to power-of-two specially (this code was
used rarely, did not give much speed up and the precision wasn't
better than the general)
* Fix the correction term formula (c=1-(u-x) was used incorrectly
when x<1 but (double)(x+1)==2, this was not a critical issue)
* Use the exact same method for calculating log(1+f) as in log
(except in log1p the c correction term is added to the result).
log, logf, log10, log10f, log2, log2f:
* Use double_t and float_t consistently.
* Now the first part of log10 and log2 is identical to log (until the
return statement, hopefully this makes maintainence easier).
* Most special case formulas were removed (close to power-of-two and
k==0 cases), they increase the code size without providing precision
or performance benefits (and obfuscate the code).
Only x==1 is handled specially so in downward rounding mode the
sign of zero is correct (the general formula happens to give -0).
* For x==0 instead of -1/0.0 or -two54/0.0, return -1/(x*x) to force
raising the exception at runtime.
* Arg reduction code is changed (slightly simplified)
* The thresholds for arg reduction to [sqrt(2)/2,sqrt(2)] are now
consistently the [0x3fe6a09e00000000,0x3ff6a09dffffffff] and the
[0x3f3504f3,0x3fb504f2] intervals for double and float reductions
respectively (the exact threshold values are not critical)
* Remove the obsolete comment for the FLT_EVAL_METHOD!=0 case in log2f
(The same code is used for all eval methods now, on i386 slightly
simpler code could be used, but we have asm there anyway)
all:
* Fix signed int arithmetics (using unsigned for bitmanipulation)
* Fix various comments
Rich Felker [Fri, 25 Oct 2013 18:15:08 +0000 (14:15 -0400)]
add legacy ftime function and sys/timeb.h
despite being marked legacy, this was specified by SUSv3 as part of
the XSI option; only the most recent version of the standard dropped
it. reportedly there's actual code using it.
Szabolcs Nagy [Wed, 23 Oct 2013 01:49:14 +0000 (01:49 +0000)]
fix inet_pton
* parse IPv4 dotted-decimal correctly (without strtoul, no leading zeros)
* disallow single leading ':' in IPv6 address
* allow at most 4 hex digits in IPv6 address (according to RFC 2373)
* have enough hex fields in IPv4 mapped IPv6 address
* disallow leading zeros in IPv4 mapped IPv6 address
Szabolcs Nagy [Tue, 22 Oct 2013 12:28:17 +0000 (12:28 +0000)]
fix __ipparse to parse the generic numbers-and-dots IPv4 format correctly
* allow at most 4 parts
* bounds check the parts correctly
* disallow leading whitespace and sign
* check the address family before falling back to IPv6
Szabolcs Nagy [Tue, 22 Oct 2013 12:23:17 +0000 (12:23 +0000)]
fix inet_aton to accept the generic "numbers-and-dots" IPv4 address format
Rich Felker [Tue, 22 Oct 2013 02:05:29 +0000 (22:05 -0400)]
split inet_addr and inet_ntoa back into their own files
despite being practically deprecated, these functions are still part
of the standard and thus cannot reside in a file that also contains
namespace pollution. this reverts some of the changes made in commit
e40f48a421a9176e3e298b5bac75f0355b219e58.
Rich Felker [Mon, 21 Oct 2013 02:01:51 +0000 (22:01 -0400)]
fix multiple minor namespace issues in headers
fcntl.h: AT_* is not a reserved namespace so extensions cannot be
exposed by default.
langinfo.h: YESSTR and NOSTR were removed from the standard.
limits.h: NL_NMAX was removed from the standard.
signal.h: the conditional for NSIG was wrongly checking _XOPEN_SOURCE
rather than _BSD_SOURCE. this was purely a mistake; it doesn't even
match the commit message from the commit that added it.
Rich Felker [Sun, 20 Oct 2013 02:36:30 +0000 (22:36 -0400)]
fix return value for inet_pton in ipv6 failure cases
in the case of input that does not match the expected form, the
correct return value is 0, not -1.
Szabolcs Nagy [Fri, 18 Oct 2013 15:21:36 +0000 (15:21 +0000)]
use consistent include guard name in arpa/ftp.h
Szabolcs Nagy [Fri, 18 Oct 2013 15:19:49 +0000 (15:19 +0000)]
fix typo in socket.h (wrong macro name)
Rich Felker [Fri, 18 Oct 2013 11:12:38 +0000 (07:12 -0400)]
in faccessat slow path, add close-on-exec to pipe fds
as usual, this is needed to avoid fd leaks. as a better solution, the
use of fds could possibly be replaced with mmap and a futex.
Rich Felker [Sat, 12 Oct 2013 23:50:09 +0000 (19:50 -0400)]
fix uid/gid-setting error in faccessat with AT_EACCESS flag
this fixes an issue reported by Daniel Thau whereby faccessat with the
AT_EACCESS flag did not work in cases where the process is running
suid or sgid but without root privileges. per POSIX, when the process
does not have "appropriate privileges", setuid changes the euid, not
the real uid, and the target uid must be equal to the current real or
saved uid; if this condition is not met, EPERM results. this caused
the faccessat child process to fail.
using the setreuid syscall rather than setuid works. POSIX leaves it
unspecified whether setreuid can set the real user id to the effective
user id on processes without "appropriate privileges", but Linux
allows this; if it's not allowed, there would be no way for this
function to work.
Rich Felker [Tue, 8 Oct 2013 23:49:10 +0000 (19:49 -0400)]
fix errno value for getcwd when size argument is zero
based on patch by Michael Forney. at the same time, I've changed the
if branch to be more clear, avoiding the comma operator.
the underlying issue is that Linux always returns ERANGE when size is
too short, even when it's zero, rather than returning EINVAL for the
special case of zero as required by POSIX.
Szabolcs Nagy [Mon, 7 Oct 2013 18:46:43 +0000 (18:46 +0000)]
Szabolcs Nagy [Mon, 7 Oct 2013 18:41:03 +0000 (18:41 +0000)]
math: use sqrtl if FLT_EVAL_METHOD==2 in acosh and acoshf
this makes acosh slightly more precise around 1.0 on i386
Szabolcs Nagy [Mon, 7 Oct 2013 13:28:03 +0000 (13:28 +0000)]
fix clockid macro names in time.h, reported by Paul Schutte
Szabolcs Nagy [Mon, 7 Oct 2013 13:26:51 +0000 (13:26 +0000)]
remove errno setting from setenv, malloc sets it correctly on oom
Szabolcs Nagy [Mon, 7 Oct 2013 13:25:19 +0000 (13:25 +0000)]
fix allocation sizes in regcomp
sizeof had incorrect argument in a few places, the size was always
large enough so the issue was not critical.
Szabolcs Nagy [Mon, 7 Oct 2013 13:24:00 +0000 (13:24 +0000)]
add missing va_end in execl* for correcness and static code analyzers
Szabolcs Nagy [Mon, 7 Oct 2013 13:22:24 +0000 (13:22 +0000)]
minor vfprintf and vfwprintf changes to please static code analyzers
add missing va_end and remove some unnecessary code.
Szabolcs Nagy [Sun, 6 Oct 2013 02:09:40 +0000 (02:09 +0000)]
math: remove an unused variable from modfl
Rich Felker [Sat, 5 Oct 2013 16:00:55 +0000 (12:00 -0400)]
slightly optimize __brk for size
there is no reason to check the return value for setting errno, since
brk never returns errors, only the new value of the brk (which may be
the same as the old, or otherwise differ from the requested brk, on
failure).
it may be beneficial to eventually just eliminate this file and make
the syscalls inline in malloc.c.
Rich Felker [Sat, 5 Oct 2013 15:59:21 +0000 (11:59 -0400)]
fix failure of malloc to set errno on heap (brk) exhaustion
I wrongly assumed the brk syscall would set errno, but on failure it
returns the old value of the brk rather than an error code.
Rich Felker [Fri, 4 Oct 2013 19:49:55 +0000 (15:49 -0400)]
fix failure to check malloc result in setenv
Szabolcs Nagy [Fri, 4 Oct 2013 18:10:17 +0000 (18:10 +0000)]
math: remove code duplication in erfl found by clang analyzer
erfl had some superflous code left around after the last erf cleanup.
the issue was reported by Alexander Monakov
Szabolcs Nagy [Fri, 4 Oct 2013 18:08:16 +0000 (18:08 +0000)]
math: remove a useless assignment in lgammal found by clang analyzer
the issue was reported by Alexander Monakov
Rich Felker [Fri, 4 Oct 2013 05:06:42 +0000 (01:06 -0400)]
fix invalid implicit pointer conversion in pthread_key_create
Rich Felker [Fri, 4 Oct 2013 04:29:58 +0000 (00:29 -0400)]
fix uninitialized variable in dladdr
the affected branch only applies for DSOs that lack standard hash
table and only have the GNU hash table present.
Rich Felker [Fri, 4 Oct 2013 04:14:57 +0000 (00:14 -0400)]
removed unused variable in vfwprintf
Rich Felker [Thu, 3 Oct 2013 14:16:01 +0000 (10:16 -0400)]
fix new environment always being null with execle
the va_arg call for the argv[]-terminating null pointer was missing,
so this pointer was being wrongly used as the environment pointer.
issue reported by Timo Teräs. proposed patch slightly modified to
simplify the resulting code.
Rich Felker [Sun, 29 Sep 2013 06:52:33 +0000 (02:52 -0400)]
fix off-by-one error in getgrnam_r and getgrgid_r, clobbering gr_name
bug report and patch by Michael Forney. the terminating null pointer
at the end of the gr_mem array was overwriting the beginning of the
string data, causing the gr_name member to always be a zero-length
string.
Rich Felker [Fri, 27 Sep 2013 17:09:46 +0000 (13:09 -0400)]
fix buffer overflow in mbsrtowcs
issue reported by Michael Forney:
"If wn becomes 0 after processing a chunk of 4, mbsrtowcs currently
continues on, wrapping wn around to -1, causing the rest of the string
to be processed.
This resulted in buffer overruns if there was only space in ws for wn
wide characters."
the original patch submitted added an additional check for !wn after
the loop; to avoid extra branching, I instead just changed the wn>=4
check to wn>=5 to ensure that at least one slot remains after the
word-at-a-time loop runs. this should not slow down the tail
processing on real-world usage, since an extra slot that can't be
processed in the word-at-a-time loop is needed for the null
termination anyway.
Szabolcs Nagy [Fri, 27 Sep 2013 13:55:29 +0000 (13:55 +0000)]
math: fix comparision macros (isless etc) when FLT_EVAL_METHOD!=0
This is a change in ISO C11 annex F (F.10.11p1), comparision macros
can't round their arguments to their semantic type when the evaluation
format has wider range and precision. (ie. they must be consistent with
the builtin relational operators)
Rich Felker [Thu, 26 Sep 2013 18:17:36 +0000 (14:17 -0400)]
add missing i386 syscall numbers
somehow the range 335-339 was missed when updating the file.
Rich Felker [Thu, 26 Sep 2013 18:11:18 +0000 (14:11 -0400)]
remove duplicate lines from mips bits/syscall.h
Rich Felker [Mon, 23 Sep 2013 21:00:53 +0000 (17:00 -0400)]
release notes for 0.9.14
Rich Felker [Sun, 22 Sep 2013 07:06:17 +0000 (03:06 -0400)]
fix arm atomic store and generate simpler/less-bloated/faster code
atomic store was lacking a barrier, which was fine for legacy arm with
no real smp and kernel-emulated cas, but unsuitable for more modern
systems. the kernel provides another "kuser" function, at 0xffff0fa0,
which could be used for the barrier, but using that would drop support
for kernels 2.6.12 through 2.6.14 unless an extra conditional were
added to check for barrier availability. just using the barrier in the
kernel cas is easier, and, based on my reading of the assembly code in
the kernel, does not appear to be significantly slower.
at the same time, other atomic operations are adapted to call the
kernel cas function directly rather than using a_cas; due to small
differences in their interface contracts, this makes the generated
code much simpler.
Rich Felker [Fri, 20 Sep 2013 06:00:27 +0000 (02:00 -0400)]
fix potential deadlock bug in libc-internal locking logic
if a multithreaded program became non-multithreaded (i.e. all other
threads exited) while one thread held an internal lock, the remaining
thread would fail to release the lock. the the program then became
multithreaded again at a later time, any further attempts to obtain
the lock would deadlock permanently.
the underlying cause is that the value of libc.threads_minus_1 at
unlock time might not match the value at lock time. one solution would
be returning a flag to the caller indicating whether the lock was
taken and needs to be unlocked, but there is a simpler solution: using
the lock itself as such a flag.
note that this flag is not needed anyway for correctness; if the lock
is not held, the unlock code is harmless. however, the memory
synchronization properties associated with a_store are costly on some
archs, so it's best to avoid executing the unlock code when it is
unnecessary.
Rich Felker [Fri, 20 Sep 2013 05:49:03 +0000 (01:49 -0400)]
correct the sysconf value for RTSIG_MAX
this is the number of realtime signals available, not the maximum
signal number or total number of signals.
Rich Felker [Mon, 16 Sep 2013 16:58:51 +0000 (12:58 -0400)]
fix sigemptyset and sigfillset for mips
they were leaving junk in the upper bits.
Rich Felker [Mon, 16 Sep 2013 15:52:39 +0000 (11:52 -0400)]
fix clobbering of caller's stack in mips __clone function
this was resulting in crashes in posix_spawn on mips, and would have
affected applications calling clone too. since the prototype for
__clone has it as a variadic function, it may not assume that 16($sp)
is writable for use in making the syscall. instead, it needs to
allocate additional stack space, and then adjust the stack pointer
back in both of the code paths for the parent process/thread.
Szabolcs Nagy [Sun, 15 Sep 2013 15:32:07 +0000 (15:32 +0000)]
sys/resource.h: add PRIO_MIN and PRIO_MAX for getpriority and setpriority
These constants are not specified by POSIX, but they are in the reserved
namespace, glibc and bsd systems seem to provide them as well.
(Note that POSIX specifies -NZERO and NZERO-1 to be the limits, but
PRIO_MAX equals NZERO)
Szabolcs Nagy [Sun, 15 Sep 2013 14:53:34 +0000 (14:53 +0000)]
update include/elf.h following glibc changes
the changes were verified using various sources:
linux: include/uapi/linux/elf.h
binutils: include/elf/common.h
glibc: elf/elf.h
sysv gabi: http://www.sco.com/developers/gabi/latest/contents.html
sun linker docs: http://docs.oracle.com/cd/E18752_01/pdf/817-1984.pdf
and platform specific docs
- fixed:
EF_MIPS_* E_MIPS_* e_flags: fixed accoding to glibc and binutils
- added:
ELFOSABI_GNU for EI_OSABI entry: glibc, binutils and sysv gabi
EM_* e_machine values: updated according to linux and glibc
PN_XNUM e_phnum value: from glibc and linux, see oracle docs
NT_* note types: updated according to linux and glibc
DF_1_* flags for DT_FLAGS_1 entry: following glibc and oracle docs
AT_HWCAP2 auxv entry for more hwcap bits accoding to linux and glibc
R_386_SIZE32 relocation according to glibc and binutils
EF_ARM_ABI_FLOAT_* e_flags: added following glibc and binutils
R_AARCH64_* relocs: added following glibc and aarch64 elf specs
R_ARM_* relocs: according to glibc, binutils and arm elf specs
R_X86_64_* relocs: added missing relocs following glibc
- removed:
HWCAP_SPARC_* flags were moved to arch specific header in glibc
R_ARM_SWI24 reloc is marked as obsolete in glibc, not present in binutils
not specified in arm elf spec, R_ARM_TLS_DESC reused its number
see http://www.codesourcery.com/publications/RFC-TLSDESC-ARM.txt
- glibc changes not pulled in:
ELFOSABI_ARM_AEABI (bare-metal system, binutils and glibc disagrees about the name)
R_68K_* relocs for unsupported platform
R_SPARC_* ditto
EF_SH* ditto (e_flags)
EF_S390* ditto (e_flags)
R_390* ditto
R_MN10300* ditto
R_TILE* ditto
Rich Felker [Mon, 16 Sep 2013 14:56:01 +0000 (10:56 -0400)]
omit CLONE_PARENT flag to clone in pthread_create
CLONE_PARENT is not necessary (CLONE_THREAD provides all the useful
parts of it) and Linux treats CLONE_PARENT as an error in certain
situations, without noticing that it would be a no-op due to
CLONE_THREAD. this error case prevents, for example, use of a
multi-threaded init process and certain usages with containers.
Rich Felker [Mon, 16 Sep 2013 14:54:31 +0000 (10:54 -0400)]
use symbolic names for clone flags in pthread_create
Szabolcs Nagy [Sun, 15 Sep 2013 04:54:19 +0000 (04:54 +0000)]
sys/socket.h: add new SO_BUSY_POLL socket option
low latency busy poll sockets are new in linux v3.11
Szabolcs Nagy [Sun, 15 Sep 2013 02:49:26 +0000 (02:49 +0000)]
ptrace.h: add new ptrace requests to get/set sigmask
PTRACE_GETSIGMASK and PTRACE_SETSIGMASK were added in linux v3.11
and used by checkpoint/restore tools
Szabolcs Nagy [Sun, 15 Sep 2013 02:42:29 +0000 (02:42 +0000)]
net/if_arp.h: add missing ARP hardware identifiers from linux uapi headers
the removed ARPHRD_IEEE802154_PHY was only present in the kernel api
in v2.6.31 (by accident), but it got into the glibc headers (in 2009)
and remained there since this header was not updated since then.
Szabolcs Nagy [Sun, 15 Sep 2013 02:41:36 +0000 (02:41 +0000)]
netinet/in.h: add missing IP protocol numbers from the linux uapi headers
Szabolcs Nagy [Sun, 15 Sep 2013 02:00:32 +0000 (02:00 +0000)]
support configurable page size on mips, powerpc and microblaze
PAGE_SIZE was hardcoded to 4096, which is historically what most
systems use, but on several archs it is a kernel config parameter,
user space can only know it at execution time from the aux vector.
PAGE_SIZE and PAGESIZE are not defined on archs where page size is
a runtime parameter, applications should use sysconf(_SC_PAGE_SIZE)
to query it. Internally libc code defines PAGE_SIZE to libc.page_size,
which is set to aux[AT_PAGESZ] in __init_libc and early in __dynlink
as well. (Note that libc.page_size can be accessed without GOT, ie.
before relocations are done)
Some fpathconf settings are hardcoded to 4096, these should be actually
queried from the filesystem using statfs.
Rich Felker [Sat, 14 Sep 2013 23:27:24 +0000 (19:27 -0400)]
fix overflow in sysconf for _SC_MQ_PRIO_MAX
the value of MQ_PRIO_MAX does not fit, so it needs to use OFLOW.
Rich Felker [Sat, 14 Sep 2013 23:19:11 +0000 (19:19 -0400)]
fix child stack alignment on mips clone
unlike other archs, the mips version of clone was not doing anything
to align the stack pointer. this seems to have been the cause for some
SIGBUS crashes that were observed in posix_spawn.
Rich Felker [Sat, 14 Sep 2013 18:44:03 +0000 (14:44 -0400)]
fix mips sysv ipc bits headers
msg.h was wrong for big-endian (wrong endiannness padding).
shm.h was just plain wrong (mips is not supposed to have padding).
both changes were tested using libc-test on qemu-system-mips.
Rich Felker [Fri, 13 Sep 2013 05:22:13 +0000 (01:22 -0400)]
fix x86_64 lrintl asm, again
the underlying problem was not incorrect sign extension (fixed in the
previous commit to this file by nsz) but that code that treats "long"
as 32-bit was copied blindly from i386 to x86_64.
now lrintl is identical to llrintl on x86_64, as it should be.
Rich Felker [Mon, 9 Sep 2013 17:39:08 +0000 (13:39 -0400)]
do not use default when dynamic linker fails to open existing path file
if fopen fails for a reason other than ENOENT, we must assume the
intent is that the path file be used. failure may be due to
misconfiguration or intentional resource-exhaustion attack (against
suid programs), in which case falling back to loading libraries from
an unintended path could be dangerous.
Szabolcs Nagy [Fri, 6 Sep 2013 20:05:25 +0000 (20:05 +0000)]
math: remove STRICT_ASSIGN from exp2f (see previous commit)
Szabolcs Nagy [Fri, 6 Sep 2013 18:35:55 +0000 (18:35 +0000)]
math: remove STRICT_ASSIGN macro
gcc did not always drop excess precision according to c99 at assignments
before version 4.5 even if -std=c99 was requested which caused badly
broken mathematical functions on i386 when FLT_EVAL_METHOD!=0
but STRICT_ASSIGN was not used consistently and it is worked around for
old compilers with -ffloat-store so it is no longer needed
the new convention is to get the compiler respect c99 semantics and when
excess precision is not harmful use float_t or double_t or to specialize
code using FLT_EVAL_METHOD
Szabolcs Nagy [Thu, 5 Sep 2013 18:05:07 +0000 (18:05 +0000)]
math: support invalid ld80 representations in fpclassify
apparently gnulib requires invalid long double representations
to be handled correctly in printf so we classify them according
to how the fpu treats them: bad inf is nan, bad nan is nan,
bad normal is nan and bad subnormal/zero is minimal normal
Szabolcs Nagy [Thu, 5 Sep 2013 16:57:46 +0000 (16:57 +0000)]
math: fix atanh (overflow and underflow issues)
in atanh exception handling was left to the called log functions,
but the argument to those functions could underflow or overflow.
use double_t and float_t to avoid some useless stores on x86
Szabolcs Nagy [Thu, 5 Sep 2013 14:03:10 +0000 (14:03 +0000)]
math: remove libc.h include from libm.h
libc.h is only for weak_alias so include it directly where it is used
Szabolcs Nagy [Thu, 5 Sep 2013 12:26:26 +0000 (12:26 +0000)]
math: fix acoshf on negative values
acosh(x) is invalid for x<1, acoshf tried to be clever using
signed comparisions to handle all x<2 the same way, but the
formula was wrong on large negative values.
Szabolcs Nagy [Thu, 5 Sep 2013 12:23:44 +0000 (12:23 +0000)]
math: fix expm1l on x86_64 (avoid underflow for large negative x)
copy the fix from i386: return -1 instead of exp2l(x)-1 when x <= -65
Szabolcs Nagy [Thu, 5 Sep 2013 12:04:58 +0000 (12:04 +0000)]
math: fix lrintl.s on x86_64 (use movslq to signextend the result)
Szabolcs Nagy [Thu, 5 Sep 2013 10:58:48 +0000 (10:58 +0000)]
math: fix exp2l asm on x86 (raise underflow correctly)
there were two problems:
* omitted underflow on subnormal results: exp2l(-16383.5) was calculated
as sqrt(2)*2^-16384, the last bits of sqrt(2) are zero so the down scaling
does not underflow eventhough the result is in subnormal range
* spurious underflow for subnormal inputs: exp2l(0x1p-16400) was evaluated
as f2xm1(x)+1 and f2xm1 raised underflow (because inexact subnormal result)
the first issue is fixed by raising underflow manually if x is in
(-32768,-16382] and not integer (x-0x1p63+0x1p63 != x)
the second issue is fixed by treating x in (-0x1p64,0x1p64) specially
for these fixes the special case handling was completely rewritten
Szabolcs Nagy [Wed, 4 Sep 2013 17:36:00 +0000 (17:36 +0000)]
math: cosmetic cleanup (use explicit union instead of fshape and dshape)
Szabolcs Nagy [Wed, 4 Sep 2013 16:39:41 +0000 (16:39 +0000)]
math: remove *_WORD64 macros from libm.h
only fma used these macros and the explicit union is clearer
Szabolcs Nagy [Wed, 4 Sep 2013 16:05:32 +0000 (16:05 +0000)]
math: remove old longdbl.h
Szabolcs Nagy [Wed, 4 Sep 2013 15:54:02 +0000 (15:54 +0000)]
math: long double fix (use ldshape union)
* use new ldshape union consistently
* add ld128 support to frexpl
* simplify sqrtl comment (ld64 is not just arm)
Szabolcs Nagy [Wed, 4 Sep 2013 15:52:54 +0000 (15:52 +0000)]
math: use float_t and double_t in scalbnf and scalbn
remove STRICT_ASSIGN (c99 semantics is assumed) and use the conventional
union to prepare the scaling factor (so libm.h is no longer needed)
Szabolcs Nagy [Wed, 4 Sep 2013 15:52:23 +0000 (15:52 +0000)]
math: fix remaining old long double code (erfl, fmal, lgammal, scalbnl)
in lgammal don't handle 1 and 2 specially, in fma use the new ldshape
union instead of ld80 one.
Szabolcs Nagy [Wed, 4 Sep 2013 15:51:05 +0000 (15:51 +0000)]
math: cbrt cleanup and long double fix
* use float_t and double_t
* cleanup subnormal handling
* bithacks according to the new convention (ldshape for long double
and explicit unions for float and double)
Szabolcs Nagy [Wed, 4 Sep 2013 07:51:11 +0000 (07:51 +0000)]
math: fix underflow in exp*.c and long double handling in exp2l
* don't care about inexact flag
* use double_t and float_t (faster, smaller, more precise on x86)
* exp: underflow when result is zero or subnormal and not -inf
* exp2: underflow when result is zero or subnormal and not exact
* expm1: underflow when result is zero or subnormal
* expl: don't underflow on -inf
* exp2: fix incorrect comment
* expm1: simplify special case handling and overflow properly
* expm1: cleanup final scaling and fix negative left shift ub (twopk)
Szabolcs Nagy [Tue, 3 Sep 2013 18:50:58 +0000 (18:50 +0000)]
math: long double trigonometric cleanup (cosl, sinl, sincosl, tanl)
ld128 support was added to internal kernel functions (__cosl, __sinl,
__tanl, __rem_pio2l) from freebsd (not tested, but should be a good
start for when ld128 arch arrives)
__rem_pio2l had some code cleanup, the freebsd ld128 code seems to
gather the results of a large reduction with precision loss (fixed
the bug but a todo comment was added for later investigation)
the old copyright was removed from the non-kernel wrapper functions
(cosl, sinl, sincosl, tanl) since these are trivial and the interesting
parts and comments had been already rewritten.
Szabolcs Nagy [Tue, 3 Sep 2013 15:02:10 +0000 (15:02 +0000)]
math: long double inverse trigonometric cleanup (acosl, asinl, atanl, atan2l)
* added ld128 support from freebsd fdlibm (untested)
* using new ldshape union instead of IEEEl2bits
* inexact status flag is not supported
Szabolcs Nagy [Tue, 3 Sep 2013 14:37:48 +0000 (14:37 +0000)]
math: rewrite hypot
method: if there is a large difference between the scale of x and y
then the larger magnitude dominates, otherwise reduce x,y so the
argument of sqrt (x*x+y*y) does not overflow or underflow and calculate
the argument precisely using exact multiplication. If the argument
has less error than 1/sqrt(2) ~ 0.7 ulp, then the result has less error
than 1 ulp in nearest rounding mode.
the original fdlibm method was the same, except it used bit hacks
instead of dekker-veltkamp algorithm, which is problematic for long
double where different representations are supported. (the new hypot
and hypotl code should be smaller and faster on 32bit cpu archs with
fast fpu), the new code behaves differently in non-nearest rounding,
but the error should be still less than 2ulps.
ld80 and ld128 are supported
Szabolcs Nagy [Tue, 3 Sep 2013 04:09:12 +0000 (04:09 +0000)]
math: rewrite remainder functions (remainder, remquo, fmod, modf)
* results are exact
* modfl follows truncl (raises inexact flag spuriously now)
* modf and modff only had cosmetic cleanup
* remainder is just a wrapper around remquo now
* using iterative shift+subtract for remquo and fmod
* ld80 and ld128 are supported as well
Szabolcs Nagy [Tue, 3 Sep 2013 03:27:02 +0000 (03:27 +0000)]
math: rewrite rounding functions (ceil, floor, trunc, round, rint)
* faster, smaller, cleaner implementation than the bit hacks of fdlibm
* use arithmetics like y=(double)(x+0x1p52)-0x1p52, which is an integer
neighbor of x in all rounding modes (0<=x<0x1p52) and only use bithacks
when that's faster and smaller (for float it usually is)
* the code assumes standard excess precision handling for casts
* long double code supports both ld80 and ld128
* nearbyint is not changed (it is a wrapper around rint)
Szabolcs Nagy [Mon, 2 Sep 2013 23:35:46 +0000 (23:35 +0000)]
math: fix logb(-0.0) in downward rounding mode
use -1/(x*x) instead of -1/(x+0) to return -inf, -0+0 is -0 in
downward rounding mode
Szabolcs Nagy [Mon, 2 Sep 2013 23:33:20 +0000 (23:33 +0000)]
math: ilogb cleanup
* consistent code style
* explicit union instead of typedef for double and float bit access
* turn FENV_ACCESS ON to make 0/0.0f raise invalid flag
* (untested) ld128 version of ilogbl (used by logbl which has ld128 support)
Szabolcs Nagy [Mon, 2 Sep 2013 00:38:51 +0000 (00:38 +0000)]
long double cleanup, initial commit
new ldshape union, ld128 support is kept, code that used the old
ldshape union was rewritten (IEEEl2bits union of freebsd libm is
not touched yet)
ld80 __fpclassifyl no longer tries to handle invalid representation
Rich Felker [Wed, 4 Sep 2013 14:23:42 +0000 (10:23 -0400)]
fix typo in comment in __randname
Rich Felker [Mon, 2 Sep 2013 19:19:37 +0000 (15:19 -0400)]
fix mips-specific bug in synccall (too little space for signal mask)
switch to the new __block_all_sigs/__restore_sigs internal API to
clean up the code too.
Rich Felker [Mon, 2 Sep 2013 19:16:36 +0000 (15:16 -0400)]
in synccall, ignore the signal before any threads' signal handlers return
this protects against deadlock from spurious signals (e.g. sent by
another process) arriving after the controlling thread releases the
other threads from the sync operation.
Rich Felker [Mon, 2 Sep 2013 19:06:46 +0000 (15:06 -0400)]
fix invalid pointer in synccall (multithread setuid, etc.)
the head pointer was not being reset between calls to synccall, so any
use of this interface more than once would build the linked list
incorrectly, keeping the (now invalid) list nodes from the previous
call.
Rich Felker [Sun, 1 Sep 2013 21:02:35 +0000 (17:02 -0400)]
fix special-case breakage in popen due to reversed argument order
Rich Felker [Sun, 1 Sep 2013 05:01:31 +0000 (01:01 -0400)]
add workaround header for wait.h remapping to sys/wait.h
Rich Felker [Sun, 1 Sep 2013 03:15:48 +0000 (23:15 -0400)]
fix missing return value warning in faccessat, minor cleanup
clone will pass the return value of the start function to SYS_exit
anyway; there's no need to call the syscall directly.
Rich Felker [Sun, 1 Sep 2013 02:52:41 +0000 (22:52 -0400)]
fix invalid %m format crash in wide scanf variants
the wide variant was missed in the previous commit.
Rich Felker [Sun, 1 Sep 2013 02:47:44 +0000 (22:47 -0400)]
avoid crash in scanf when invalid %m format is encountered
invalid format strings invoke undefined behavior, so this is not a
conformance issue, but it's nicer for scanf to report the error safely
instead of calling free on a potentially-uninitialized pointer or a
pointer to memory belonging to the caller.
Rich Felker [Sat, 31 Aug 2013 20:01:01 +0000 (16:01 -0400)]
remove incorrect cancellation points from realpath
Rich Felker [Sat, 31 Aug 2013 19:50:23 +0000 (15:50 -0400)]
debloat realpath's allocation strategy
rather than allocating a PATH_MAX-sized buffer when the caller does
not provide an output buffer, work first with a PATH_MAX-sized temp
buffer with automatic storage, and either copy it to the caller's
buffer or strdup it on success. this not only avoids massive memory
waste, but also avoids pulling in free (and thus the full malloc
implementation) unnecessarily in static programs.
Rich Felker [Sat, 31 Aug 2013 19:44:58 +0000 (15:44 -0400)]
make realpath use O_PATH when opening the file
this avoids failure if the file is not readable and avoids odd
behavior for device nodes, etc. on old kernels that lack O_PATH, the
old behavior (O_RDONLY) will naturally happen as the fallback.
Rich Felker [Sat, 31 Aug 2013 15:36:56 +0000 (11:36 -0400)]
fix regression in creation of ldso symlink
DESTDIR was wrongly included in the symlink contents.
Rich Felker [Sat, 31 Aug 2013 05:12:00 +0000 (01:12 -0400)]
fix breakage in synccall due to incorrect signal restoration in sigqueue
commit
07827d1a82fb33262f686eda959857f0d28cd8fa seems to have
introduced this issue. sigqueue is called from the synccall core, at
which time, even implementation-internal signals are blocked. however,
pthread_sigmask removes the implementation-internal signals from the
old mask before returning, so that a process which began life with
them blocked will not be able to save a signal mask that has them
blocked, possibly causing them to become re-blocked later. however,
this was causing sigqueue to unblock the implementation-internal
signals during synccall, leading to deadlock.
Rich Felker [Fri, 30 Aug 2013 21:30:27 +0000 (17:30 -0400)]
fix typo in release notes