OSDN Git Service
Rich Felker [Thu, 18 Jul 2013 18:15:48 +0000 (14:15 -0400)]
fix FILENAME_MAX to match PATH_MAX
POSIX is not clear on whether it includes the termination, but ISO C
requires that it does. the whole concept of this macro is rather
useless, but it's better to be correct anyway.
Rich Felker [Wed, 17 Jul 2013 21:46:38 +0000 (17:46 -0400)]
make spacing in x86_64 alltypes.h.sh more uniform with other archs
patch by Luka Perkov, who noted that all other archs have a newline.
Rich Felker [Wed, 17 Jul 2013 17:54:41 +0000 (13:54 -0400)]
make posix_spawn (and functions that use it) use CLONE_VFORK flag
this is both a minor scheduling optimization and a workaround for a
difficult-to-fix bug in qemu app-level emulation.
from the scheduling standpoint, it makes no sense to schedule the
parent thread again until the child has exec'd or exited, since the
parent will immediately block again waiting for it.
on the qemu side, as regular application code running on an underlying
libc, qemu cannot make arbitrary clone syscalls itself without
confusing the underlying implementation. instead, it breaks them down
into either fork-like or pthread_create-like cases. it was treating
the code in posix_spawn as pthread_create-like, due to CLONE_VM, which
caused horribly wrong behavior: CLONE_FILES broke the synchronization
mechanism, CLONE_SIGHAND broke the parent's signals, and CLONE_THREAD
caused the child's exec to end the parent -- if it hadn't already
crashed. however, qemu special-cases CLONE_VFORK and emulates that
with fork, even when CLONE_VM is also specified. this also gives
incorrect semantics for code that really needs the memory sharing, but
posix_spawn does not make use of the vm sharing except to avoid
momentary double commit charge.
programs using posix_spawn (including via popen) should now work
correctly under qemu app-level emulation.
Rich Felker [Wed, 17 Jul 2013 14:43:04 +0000 (10:43 -0400)]
fix missing argument in variadic syscall macros
for 0-argument syscalls (1 argument to the macro, the syscall number),
the __SYSCALL_NARGS_X macro's ... argument was not satisfied. newer
compilers seem to care about this.
Rich Felker [Wed, 17 Jul 2013 14:39:11 +0000 (10:39 -0400)]
fix error code on time conversion overflows
POSIX mandates EOVERFLOW for this condition.
Rich Felker [Wed, 17 Jul 2013 13:29:26 +0000 (09:29 -0400)]
fix fd leak in file mapping code used in new zoneinfo support
Rich Felker [Wed, 17 Jul 2013 09:24:50 +0000 (05:24 -0400)]
the big time handling overhaul
this commit has two major user-visible parts: zoneinfo-format time
zones are now supported, and overflow handling is intended to be
complete in the sense that all functions return a correct result if
and only if the result fits in the destination type, and otherwise
return an error. also, some noticable bugs in the way DST detection
and normalization worked have been fixed, and performance may be
better than before, but it has not been tested.
Rich Felker [Sat, 13 Jul 2013 18:54:34 +0000 (14:54 -0400)]
fix omission of dtv setup in static linked programs on TLS variant I archs
apparently this was never noticed before because the linker normally
optimizes dynamic TLS models to non-dynamic ones when static linking,
thus eliminating the calls to __tls_get_addr which crash when the dtv
is missing. however, some libsupc++ code on ARM was calling
__tls_get_addr when static linked and crashing. the reason is unclear
to me, but with this issue fixed it should work now anyway.
Rich Felker [Wed, 10 Jul 2013 20:58:27 +0000 (16:58 -0400)]
add PIE support for ARM
Rich Felker [Wed, 10 Jul 2013 20:11:01 +0000 (16:11 -0400)]
add some ARM EABI-specific exception handling infrastructure
patch by Timo Teräs
Rich Felker [Wed, 10 Jul 2013 18:38:20 +0000 (14:38 -0400)]
fix invalid library phdr pointers passed to callback from dl_iterate_phdr
map_library was saving pointers to an automatic-storage buffer rather
than pointers into the mapping. this should be a fairly simple fix,
but the patch here is slightly complicated by two issues:
1. supporting gratuitously obfuscated ELF files where the program
headers are not right at the beginning of the file.
2. cleaning up the map_library function so that data isn't clobbered
by the time we need it.
Rich Felker [Tue, 9 Jul 2013 06:30:21 +0000 (02:30 -0400)]
fix a couple misleading/wrong signal descriptions in strsignal
there are still several more that are misleading, but SIGFPE (integer
division error misdescribed as floating point) and and SIGCHLD
(possibly non-exit status change events described as exiting) were the
worst offenders.
Rich Felker [Tue, 9 Jul 2013 06:23:16 +0000 (02:23 -0400)]
add realtime signals to strsignal
the name format RTnn/RTnnn was chosen to minimized bloat while
uniquely identifying the signal.
Rich Felker [Tue, 9 Jul 2013 06:11:52 +0000 (02:11 -0400)]
fix off-by-one array bound in strsignal
Rich Felker [Tue, 9 Jul 2013 04:50:11 +0000 (00:50 -0400)]
fix bogus lazy allocation in ctermid and missing malloc failure check
also clean up, optimize, and simplify the code, removing branches by
simply pre-setting the result string to an empty string, which will be
preserved if other operations fail.
Rich Felker [Tue, 9 Jul 2013 04:42:09 +0000 (00:42 -0400)]
fix fd leak on races and cancellation in ctermid
Rich Felker [Tue, 9 Jul 2013 04:40:17 +0000 (00:40 -0400)]
fix missing SOCK_CLOEXEC in various functions that use sockets internally
Rich Felker [Sat, 6 Jul 2013 05:12:28 +0000 (01:12 -0400)]
add NFDBITS in sys/select.h with appropriate feature tests
the main use for this macro seems to be knowing the correct allocation
granularity for dynamic-sized fd_set objects. such usage is
non-conforming and results in undefined behavior, but it is widespread
in applications.
Rich Felker [Fri, 5 Jul 2013 03:58:16 +0000 (23:58 -0400)]
move core memalign code from aligned_alloc to __memalign
there are two motivations for this change. one is to avoid
gratuitously depending on a C11 symbol for implementing a POSIX
function. the other pertains to the documented semantics. C11 does not
define any behavior for aligned_alloc when the length argument is not
a multiple of the alignment argument. posix_memalign on the other hand
places no requirements on the length argument. using __memalign as the
implementation of both, rather than trying to implement one in terms
of the other when their documented contracts differ, eliminates this
confusion.
Rich Felker [Fri, 5 Jul 2013 03:54:12 +0000 (23:54 -0400)]
move alignment check from aligned_alloc to posix_memalign
C11 has no requirement that the alignment be a multiple of
sizeof(void*), and in fact seems to require any "valid alignment
supported by the implementation" to work. since the alignment of char
is 1 and thus a valid alignment, an alignment argument of 1 should be
accepted.
rofl0r [Wed, 3 Jul 2013 15:55:37 +0000 (17:55 +0200)]
add legacy sys/ttydefaults.h header
rofl0r [Wed, 3 Jul 2013 15:52:44 +0000 (17:52 +0200)]
add legacy scsi/scsi_ioctl.h header
rofl0r [Sun, 30 Jun 2013 21:04:12 +0000 (23:04 +0200)]
paths.h: remove cruft
a research in debian codesearch and grepping over the pkgsrc
directory tree have shown that these macros are all either unused,
or defined by programs in case they need them.
Rich Felker [Mon, 1 Jul 2013 17:50:02 +0000 (13:50 -0400)]
add stubs for additional legacy ether.h functions
these would not be expensive to actually implement, but reading
/etc/ethers does not sound like a particularly useful feature, so for
now I'm leaving them as stubs.
Rich Felker [Mon, 1 Jul 2013 17:43:43 +0000 (13:43 -0400)]
fix Makefile so "make install" works before include/bits symlink exists
previously, determination of the list of header files for installation
depended on the include/bits symlink (to the arch-specific files)
already having been created. in other words, running "make install"
immediately after configure without first running "make" caused the
bits headers not to be installed.
the solution I have applied is to pull the list of headers directly
from arch/$(ARCH)/bits rather than include/bits, and likewise to
install directly from arch/$(ARCH)/bits rather than via the symlink.
at this point, the only purpose served by keeping the symlink around
is that it enables use of the in-tree headers and libs directly via -I
and -L, which can be useful when testing against a new version of the
library before installing it. on the other hand, removing the bits
symlink would be beneficial if we ever want to support building
multiple archs in the same source tree.
Rich Felker [Sun, 30 Jun 2013 02:43:05 +0000 (22:43 -0400)]
release notes for 0.9.11
Rich Felker [Sat, 29 Jun 2013 20:16:47 +0000 (16:16 -0400)]
fix failure of mbsrtowcs to record stop position when dest is full
Rich Felker [Sat, 29 Jun 2013 17:57:13 +0000 (13:57 -0400)]
fix shifts possibly larger than type in major() macro
in theory this should not be an issue, since major() should only be
applied to type dev_t, which is 64-bit. however, it appears some
applications are not using dev_t but a smaller integer type (which
works on Linux because the kernel's dev_t is really only 32-bit). to
avoid the undefined behavior, do it as two shifts.
Rich Felker [Sat, 29 Jun 2013 16:46:46 +0000 (12:46 -0400)]
implement minimal dlinfo function
Rich Felker [Sat, 29 Jun 2013 16:24:06 +0000 (12:24 -0400)]
add some comments about the mips ksigaction structure weirdness
Rich Felker [Sat, 29 Jun 2013 06:24:02 +0000 (02:24 -0400)]
fix missing synchronization in calls from dynamic linker to global ctors
this change is needed to correctly handle the case where a constructor
creates a new thread which calls dlopen. previously, the lock was not
held in this case. the reason for the complex logic to avoid locking
whenever possible is that, since the mutex is recursive, it will need
to inspect the thread pointer to get the current thread's tid, and
this requires initializing the thread pointer. we do not want
non-multi-threaded programs to attempt to access the thread pointer
unnecessarily; doing so could make them crash on ancient kernels that
don't support threads but which may otherwise be capable of running
the program.
Rich Felker [Sat, 29 Jun 2013 04:02:38 +0000 (00:02 -0400)]
prevent shmget from allocating objects that overflow ptrdiff_t
rather than returning an error, we have to increase the size argument
so high that the kernel will have no choice but to fail. this is
because POSIX only permits the EINVAL error for size errors when a new
shared memory segment would be created; if it already exists, the size
argument must be ignored. unfortunately Linux is non-conforming in
this regard, but I want to keep the code correct in userspace anyway
so that if/when Linux is fixed, the behavior applications see will be
conforming.
Rich Felker [Sat, 29 Jun 2013 03:57:58 +0000 (23:57 -0400)]
work around wrong kernel type for sem_nsems member of struct semid_ds
rejecting invalid values for n is fine even in the case where a new
sem will not be created, since the kernel does its range checks on n
even in this case as well.
by default, the kernel will bound the limit well below USHRT_MAX
anyway, but it's presumably possible that an administrator could
override this limit and break things.
Rich Felker [Sat, 29 Jun 2013 03:39:50 +0000 (23:39 -0400)]
add missing type shmatt_t in sys/shm.h
this type is not really intended to be used; it's just there to allow
implementations to choose the type for the shm_nattch member of
struct shmid_sh, presumably since historical implementations disagreed
on the type. in any case, it needs to be there, so now it is.
Rich Felker [Fri, 28 Jun 2013 16:38:42 +0000 (12:38 -0400)]
implement week-based-year year numbers in strftime
in the process, I refactored the week-number code so it can be used by
the week-based-year formats to determine year adjustments at the
boundary values. this also improves indention/code readability.
Rich Felker [Fri, 28 Jun 2013 16:12:55 +0000 (12:12 -0400)]
fix breakage in last commit to strftime due to missing INT_MAX
that's what I get for changing a hard-coded threshold to a proper
non-magic-number without testing.
Rich Felker [Fri, 28 Jun 2013 16:03:58 +0000 (12:03 -0400)]
implement week numbers and half of the week-based-year logic for strftime
output for plain week numbers (%U and %W) has been sanity-checked, and
output for the week-based-year week numbers (%V) has been checked
extensively against known-good data for the full non-negative range of
32-bit time_t.
year numbers for week-based years (%g and %G) are not yet implemented.
Rich Felker [Fri, 28 Jun 2013 00:00:29 +0000 (20:00 -0400)]
minor compatibility fixes in utmp.h and fixing mismatch with paths.h
the pathnames prefixed with /dev/null/ are guaranteed never to be
valid. the previous use of /dev/null alone was mildly dangerous in
that bad software might attempt to unlink the name when it found a
non-regular file there and create a new file.
Rich Felker [Thu, 27 Jun 2013 16:48:59 +0000 (12:48 -0400)]
disallow creation of objects larger than PTRDIFF_MAX via mmap
internally, other parts of the library assume sizes don't overflow
ssize_t and/or ptrdiff_t, and the way this assumption is made valid is
by preventing creating of such large objects. malloc already does so,
but the check was missing from mmap.
this is also a quality of implementation issue: even if the
implementation internally could handle such objects, applications
could inadvertently invoke undefined behavior by subtracting pointers
within an object. it is very difficult to guard against this in
applications, so a good implementation should simply ensure that it
does not happen.
Rich Felker [Thu, 27 Jun 2013 02:02:23 +0000 (22:02 -0400)]
fix syscall argument bug in pthread_getschedparam
the address of the pointer to the sched param, rather than the
pointer, was being passed to the kernel.
Rich Felker [Thu, 27 Jun 2013 01:41:51 +0000 (21:41 -0400)]
fix temp file leak in sem_open on successful creation of new semaphore
Rich Felker [Thu, 27 Jun 2013 01:39:15 +0000 (21:39 -0400)]
fix bug whereby sem_open leaked its own internal slots on failure
Rich Felker [Thu, 27 Jun 2013 01:35:56 +0000 (21:35 -0400)]
in sem_open, don't leak vm mapping if fstat fails
fstat should not fail under normal circumstances, so this fix is
mostly theoretical.
Rich Felker [Thu, 27 Jun 2013 01:34:44 +0000 (21:34 -0400)]
fix failure of pthread_setschedparam to pass correct param to kernel
the address of the pointer, rather than the pointer, was being passed.
this was probably a copy-and-paste error from corresponding get code.
Rich Felker [Wed, 26 Jun 2013 23:43:24 +0000 (19:43 -0400)]
document in sysconf and unistd.h that per-thread cpu clocks exist
Rich Felker [Wed, 26 Jun 2013 18:27:45 +0000 (14:27 -0400)]
fix iconv conversion to legacy 8bit codepages
this seems to have been a simple copy-and-paste error from the code
for converting from legacy codepages.
Rich Felker [Wed, 26 Jun 2013 14:51:36 +0000 (10:51 -0400)]
remove useless conditional before free from dynamic linker path code
Rich Felker [Wed, 26 Jun 2013 14:17:29 +0000 (10:17 -0400)]
fix dynamic linker handling of empty path file or error reading path file
previously, the path string was being used despite being invalid. with
this change, empty path file or error reading the path file is treated
as an empty path. this is preferable to falling back to a default
path, so that attacks to prevent reading of the path file could not
result in loading incorrect and possibly dangerous (outdated or
mismatching ABI) libraries from.
the code to strip the final newline has also been removed; now that
newline is accepted as a delimiter, it's harmless to leave it in
place.
Rich Felker [Wed, 26 Jun 2013 02:26:20 +0000 (22:26 -0400)]
respect iso c namespace in stdio.h and wchar.h regarding va_list
despite declaring functions that take arguments of type va_list, these
headers are not permitted by the c standard to expose the definition
of va_list, so an alias for the type must be used. the name
__isoc_va_list was chosen to convey that the purpose of this alternate
name is for iso c conformance, and to avoid the multitude of names
which gcc mangles with its hideous "fixincludes" monstrosity, leading
to serious header breakage if these "fixes" are run.
Rich Felker [Wed, 26 Jun 2013 01:39:35 +0000 (21:39 -0400)]
make newline-delimited dynamic linker path file actually work
apparently the original commit was never tested properly, since
getline was only ever reading one line. the intent was to read the
entire file, so use getdelim with the null byte as delimiter as a
cheap way to read a whole file into memory.
Rich Felker [Wed, 26 Jun 2013 01:35:49 +0000 (21:35 -0400)]
implement inet_lnaof, inet_netof, and inet_makeaddr
also move all legacy inet_* functions into a single file to avoid
wasting object file and compile time overhead on them.
the added functions are legacy interfaces for working with classful
ipv4 network addresses. they have no modern usefulness whatsoever, but
some programs unconditionally use them anyway, and they're tiny.
Rich Felker [Wed, 26 Jun 2013 01:15:27 +0000 (21:15 -0400)]
add ether_aton[_r] and ether_ntoa[_r] functions
based on patch by Strake with minor stylistic changes, and combined
into a single file. this patch remained open for a long time due to
some question as to whether ether_aton would be better implemented in
terms of sscanf, and it's time something was committed, so here it is.
Rich Felker [Sat, 22 Jun 2013 21:23:45 +0000 (17:23 -0400)]
fix scanf %c conversion wrongly storing a terminating null byte
this seems to have been a regression from the refactoring which added
the 'm' modifier.
Rich Felker [Sat, 22 Jun 2013 21:11:17 +0000 (17:11 -0400)]
fix major scanf breakage with unbuffered streams, fmemopen, etc.
the shgetc api, used internally in scanf and int/float scanning code
to handle field width limiting and pushback, was designed assuming
that pushback could be achieved via a simple decrement on the file
buffer pointer. this only worked by chance for regular FILE streams,
due to the linux readv bug workaround in __stdio_read which moves the
last requested byte through the buffer rather than directly back to
the caller. for unbuffered streams and streams not using __stdio_read
but some other underlying read function, the first character read
could be completely lost, and replaced by whatever junk happened to be
in the unget buffer.
to fix this, simply have shgetc, when it performs an underlying read
operation on the stream, store the character read at the -1 offset
from the read buffer pointer. this is valid even for unbuffered
streams, as they have an unget buffer located just below the start of
the zero-length buffer. the check to avoid storing the character when
it is already there is to handle the possibility of read-only buffers.
no application-exposed FILE types are allowed to use read-only
buffers, but sscanf and strto* may use them internally when calling
functions which use the shgetc api.
Rich Felker [Sun, 16 Jun 2013 14:39:02 +0000 (10:39 -0400)]
fix invalid access in aio notification
issue found and patch provided by Jens Gustedt. after the atomic store
to the error code field of the aiocb, the application is permitted to
free or reuse the storage, so further access is invalid. instead, use
the local copy that was already made.
Rich Felker [Sun, 16 Jun 2013 05:40:17 +0000 (01:40 -0400)]
fix uninitialized variable in lio (aio) code
Rich Felker [Wed, 12 Jun 2013 22:20:48 +0000 (18:20 -0400)]
improve the quality of output from rand_r
due to the interface requirement of having the full state contained in
a single object of type unsigned int, it is difficult to provide a
reasonable-quality implementation; most good PRNGs are immediately
ruled out because they need larger state. the old rand_r gave very
poor output (very short period) in its lower bits; normally, it's
desirable to throw away the low bits (as in rand()) when using a LCG,
but this is not possible since the state is only 32 bits and we need
31 bits of output.
glibc's rand_r uses the same LCG as musl's, but runs it for 3
iterations and only takes 10-11 bits from each iteration to construct
the output value. this partially fixes the period issue, but
introduces bias: not all outputs have the same frequency, and many do
not appear at all. with such a low period, the bias is likely to be
observable.
I tried many approaches to "fix" rand_r, and the simplest I found
which made it pass the "dieharder" tests was applying this
transformation to the output. the "temper" function is taken from
mersenne twister, where it seems to have been chosen for some rigorous
properties; here, the only formal property I'm using is that it's
one-to-one and thus avoids introducing bias.
should further deficiencies in rand_r be reported, the obvious "best"
solution is applying a 32-bit cryptographic block cipher in CTR mode.
I identified several possible ciphers that could be used directly or
adapted, but as they would be a lot slower and larger, I do not see a
justification for using them unless the current rand_r proves
deficient for some real-world use.
Rich Felker [Sat, 8 Jun 2013 15:42:52 +0000 (11:42 -0400)]
add clock id macros for a number of new(ish) Linux-specific clocks
arguably CLOCK_MONOTONIC should be redirected to CLOCK_BOOTTIME with a
fallback for old kernels that don't support it, since Linux's
CLOCK_BOOTTIME semantics seem to match the spirit of the POSIX
requirements for CLOCK_MONOTONIC better than Linux's version of
CLOCK_MONOTONIC does. however, this is a change that would require
further discussion and research, so for now, I'm simply making them
all available.
Rich Felker [Sat, 8 Jun 2013 15:40:27 +0000 (11:40 -0400)]
fix the type of CLOCKS_PER_SEC to match new clock_t type
originally it was right on 32-bit archs and wrong on 64-bit, but after
recent changes it was wrong everywhere. with this commit, it's now
right everywhere.
Rich Felker [Sat, 8 Jun 2013 15:36:41 +0000 (11:36 -0400)]
support cputime clocks for processes/threads other than self
apparently these features have been in Linux for a while now, so it
makes sense to support them. the bit twiddling seems utterly illogical
and wasteful, especially the negation, but that's how the kernel folks
chose to encode pids/tids into the clock id.
Szabolcs Nagy [Sat, 8 Jun 2013 13:49:12 +0000 (13:49 +0000)]
prng: make rand_r have 2^32 period instead of 2^31
this is a minor fix to increase the period of the obsolete rand_r a bit.
an include header in __rand48_step.c is fixed as well.
Szabolcs Nagy [Sat, 8 Jun 2013 13:31:10 +0000 (13:31 +0000)]
prng: fix rand() to give good sequence with small state
some applications rely on the low bits of rand() to be reasonably good
quality prng, so now it fixed by using the top bits of a 64 bit LCG,
this is simple, has small state and passes statistical tests.
D.E. Knuth attributes the multiplier to C.E. Haynes in TAOCP Vol2 3.3.4
Rich Felker [Fri, 7 Jun 2013 14:18:07 +0000 (10:18 -0400)]
fix mixup in previous change to gcc wrapper
Rich Felker [Fri, 7 Jun 2013 14:13:07 +0000 (10:13 -0400)]
make gcc-specific headers (intrinsics, etc.) available with wrapper
they are intentionally listed after the libc include directory so that
the gcc float.h, etc. don't get used in place of the libc ones.
Rich Felker [Fri, 7 Jun 2013 13:54:45 +0000 (09:54 -0400)]
improve handling of nonstandard fields in struct tm
defining tm_gmtoff and tm_zone as macros was breaking some application
code that used these names for its own purposes.
Rich Felker [Thu, 6 Jun 2013 04:26:17 +0000 (00:26 -0400)]
implement 'm' modifier for wide scanf variants
Rich Felker [Wed, 5 Jun 2013 22:18:41 +0000 (18:18 -0400)]
implement the 'm' (malloc) modifier for scanf
this commit only covers the byte-based scanf-family functions. the
wide functions still lack support for the 'm' modifier.
Rich Felker [Wed, 5 Jun 2013 20:53:26 +0000 (16:53 -0400)]
refactor wide-char scanf string handling
this brings the wide version of the code into alignment with the
byte-based version, in preparation for adding support for the m
(malloc) modifier.
Rich Felker [Tue, 4 Jun 2013 20:22:02 +0000 (16:22 -0400)]
simplify some logic in scanf and remove redundant invalid-format check
Rich Felker [Tue, 4 Jun 2013 20:09:36 +0000 (16:09 -0400)]
refactor scanf core to use common code path for all string formats
the concept here is that %s and %c are essentially special-cases of
%[, with some minimal additional special-casing.
aside from simplifying the code and reducing the number of complex
code-paths that would need changing to make optimizations later, the
main purpose of this change is to simplify addition of the 'm'
modifier which causes scanf to allocate storage for the string being
read.
Rich Felker [Mon, 3 Jun 2013 21:32:42 +0000 (17:32 -0400)]
align stack properly for calling global ctors/dtors on x86[_64]
failure to do so was causing crashes on x86_64 when ctors used SSE,
which was first observed when ctors called variadic functions due to
the SSE prologue code inserted into every variadic function.
Rich Felker [Mon, 3 Jun 2013 20:35:59 +0000 (16:35 -0400)]
ensure that thread dtv pointer is never null to optimize __tls_get_addr
Rich Felker [Sun, 26 May 2013 22:22:12 +0000 (18:22 -0400)]
Merge remote-tracking branch 'nsz/review'
Szabolcs Nagy [Sun, 26 May 2013 16:01:38 +0000 (16:01 +0000)]
fix the prototype of settimeofday to follow the original BSD declaration
Szabolcs Nagy [Sun, 26 May 2013 15:49:08 +0000 (15:49 +0000)]
fix ioctl _IOR, _IOW, etc macros to avoid signed overflow (2<<30)
Szabolcs Nagy [Sun, 26 May 2013 15:43:17 +0000 (15:43 +0000)]
on x86_64 use long instead of long long for 64bit posix types
following glibc use the lowest rank 64bit integer type for ino_t etc.
this is eg. useful for printf format compatibility
Rich Felker [Fri, 24 May 2013 00:38:51 +0000 (20:38 -0400)]
change underlying type of clock_t to be uniform and match ABI
previously we were using an unsigned type on 32-bit systems so that
subtraction would be well-defined when it wrapped, but since wrapping
is non-conforming anyway (when clock() overflows, it has to return -1)
the only use of unsigned would be to buy a little bit more time before
overflow. this does not seem worth having the type vary per-arch
(which leads to more arch-specific bugs) or disagree with the ABI musl
(mostly) follows.
Rich Felker [Thu, 23 May 2013 18:31:02 +0000 (14:31 -0400)]
fix overflow behavior of clock() function
per Austin Group interpretation for issue #686, which cites the
requirements of ISO C, clock() cannot wrap. if the result is not
representable, it must return (clock_t)-1. in addition, the old code
was performing wrapping via signed overflow and thus invoking
undefined behavior.
since it seems impossible to accurately check for overflow with the
old times()-based fallback code, I have simply dropped the fallback
code for now, thus always returning -1 on ancient systems. if there's
a demand for making it work and somebody comes up with a way, it could
be reinstated, but the clock() function is essentially useless on
32-bit system anyway (it overflows in less than an hour).
it should be noted that I used LONG_MAX rather than ULONG_MAX, despite
32-bit archs using an unsigned type for clock_t. this discrepency with
the glibc/LSB type definitions will be fixed now; since wrapping of
clock_t is no longer supported, there's no use in it being unsigned.
Szabolcs Nagy [Sun, 19 May 2013 14:43:32 +0000 (14:43 +0000)]
math: add fma TODO comments about the underflow issue
The underflow exception is not raised correctly in some
cornercases (see previous fma commit), added comments
with examples for fmaf, fmal and non-x86 fma.
In fmaf store the result before returning so it has the
correct precision when FLT_EVAL_METHOD!=0
Szabolcs Nagy [Sun, 19 May 2013 12:13:08 +0000 (12:13 +0000)]
math: fix two fma issues (only affects non-nearest rounding mode, x86)
1) in downward rounding fma(1,1,-1) should be -0 but it was 0 with
gcc, the code was correct but gcc does not support FENV_ACCESS ON
so it used common subexpression elimination where it shouldn't have.
now volatile memory access is used as a barrier after fesetround.
2) in directed rounding modes there is no double rounding issue
so the complicated adjustments done for nearest rounding mode are
not needed. the only exception to this rule is raising the underflow
flag: assume "small" is an exactly representible subnormal value in
double precision and "verysmall" is a much smaller value so that
(long double)(small plus verysmall) == small
then
(double)(small plus verysmall)
raises underflow because the result is an inexact subnormal, but
(double)(long double)(small plus verysmall)
does not because small is not a subnormal in long double precision
and it is exact in double precision.
now this problem is fixed by checking inexact using fenv when the
result is subnormal
Rich Felker [Sat, 18 May 2013 16:06:42 +0000 (12:06 -0400)]
Merge remote-tracking branch 'nsz/review'
Szabolcs Nagy [Sat, 18 May 2013 14:40:22 +0000 (14:40 +0000)]
math: sin cos cleanup
* use unsigned arithmetics
* use unsigned to store arg reduction quotient (so n&3 is understood)
* remove z=0.0 variables, use literal 0
* raise underflow and inexact exceptions properly when x is small
* fix spurious underflow in tanl
Rich Felker [Sat, 18 May 2013 14:20:42 +0000 (10:20 -0400)]
make err.h functions print __progname
patch by Strake. previously is was not feasible to duplicate this
functionality of the functions these were modeled on, since argv[0]
was not saved at program startup, but now that it's available it's
easy to use.
Szabolcs Nagy [Sat, 18 May 2013 12:34:00 +0000 (12:34 +0000)]
math: tan cleanups
* use unsigned arithmetics on the representation
* store arg reduction quotient in unsigned (so n%2 would work like n&1)
* use different convention to pass the arg reduction bit to __tan
(this argument used to be 1 for even and -1 for odd reduction
which meant obscure bithacks, the new n&1 is cleaner)
* raise inexact and underflow flags correctly for small x
(tanl(x) may still raise spurious underflow for small but normal x)
(this exception raising code increases codesize a bit, similar fixes
are needed in many other places, it may worth investigating at some
point if the inexact and underflow flags are worth raising correctly
as this is not strictly required by the standard)
* tanf manual reduction optimization is kept for now
* tanl code path is cleaned up to follow similar logic to tan and tanf
Rich Felker [Fri, 17 May 2013 22:38:42 +0000 (18:38 -0400)]
add FLT_TRUE_MIN, etc. macros from C11
there was some question as to how many decimal places to use, since
one decimal place is always sufficient to identify the smallest
denormal uniquely. for now, I'm following the example in the C
standard which is consistent with the other min/max macros we already
had in place.
Rich Felker [Fri, 17 May 2013 18:23:41 +0000 (14:23 -0400)]
remove the __STDC_FORMAT_MACROS nonsense from inttypes.h
somehow I missed this when removing the corresponding
__STDC_LIMIT_MACROS and __STDC_CONSTANT_MACROS nonsense from stdint.h.
these were all attempts by the C committee to guess what the C++
committee would want, and the guesses turned out to be wrong.
Rich Felker [Thu, 16 May 2013 20:27:37 +0000 (16:27 -0400)]
fix mknod and mknodat to accept large dev_t values
support for these was recently added to sysmacros.h. note that the
syscall argument is a long, despite dev_t being 64-bit, so on 32-bit
archs the high bits will be lost. it appears the high bits are just
glibc silliness and not part of the kernel api, anyway, but it's nice
that we have them there for future expansion if needed.
Szabolcs Nagy [Wed, 15 May 2013 23:08:52 +0000 (23:08 +0000)]
math: use double_t for temporaries to avoid stores on i386
When FLT_EVAL_METHOD!=0 (only i386 with x87 fp) the excess
precision of an expression must be removed in an assignment.
(gcc needs -fexcess-precision=standard or -std=c99 for this)
This is done by extra load/store instructions which adds code
bloat when lot of temporaries are used and it makes the result
less precise in many cases.
Using double_t and float_t avoids these issues on i386 and
it makes no difference on other archs.
For now only a few functions are modified where the excess
precision is clearly beneficial (mostly polynomial evaluations
with temporaries).
object size differences on i386, gcc-4.8:
old new
__cosdf.o 123 95
__cos.o 199 169
__sindf.o 131 95
__sin.o 225 203
__tandf.o 207 151
__tan.o 605 499
erff.o 1470 1416
erf.o 1703 1649
j0f.o 1779 1745
j0.o 2308 2274
j1f.o 1602 1568
j1.o 2286 2252
tgamma.o 1431 1424
math/*.o 64164 63635
Rich Felker [Wed, 15 May 2013 20:15:50 +0000 (16:15 -0400)]
support full range of dev_t major/minor numbers in makedev, etc. macros
Szabolcs Nagy [Mon, 6 May 2013 17:52:48 +0000 (17:52 +0000)]
remove compound literals from math.h to please c++
__FLOAT_BITS and __DOUBLE_BITS macros used union compound literals,
now they are changed into static inline functions. A good C compiler
generates the same code for both and the later is C++ conformant.
Rich Felker [Sun, 5 May 2013 18:51:25 +0000 (14:51 -0400)]
fix incorrect clock tick scaling in fallback case of clock()
since CLOCKS_PER_SEC is
1000000 (required by XSI) and the times
syscall reports values in 1/100 second units (Linux), the correct
scaling factor is 10000, not 100. note that only ancient kernels which
lack clock_gettime are affected.
Rich Felker [Sun, 5 May 2013 18:19:37 +0000 (14:19 -0400)]
do not interpret errors in return value of times() syscall
all return values are valid, and on 32-bit systems, values that look
like errors can and will occur. since the only actual error this
function could return is EFAULT, and it is only returnable when the
application has invoked undefined behavior, simply ignore the
possibility that the return value is actually an error code.
Rich Felker [Fri, 26 Apr 2013 23:48:01 +0000 (19:48 -0400)]
transition to using functions for internal signal blocking/restoring
there are several reasons for this change. one is getting rid of the
repetition of the syscall signature all over the place. another is
sharing the constant masks without costly GOT accesses in PIC.
the main motivation, however, is accurately representing whether we
want to block signals that might be handled by the application, or all
signals.
Rich Felker [Fri, 26 Apr 2013 23:02:23 +0000 (19:02 -0400)]
optimize/debloat raise
use __syscall rather than syscall when failure is not possible or not
to be considered.
Rich Felker [Fri, 26 Apr 2013 21:51:22 +0000 (17:51 -0400)]
prevent code from running under a thread id which already gave ESRCH
Rich Felker [Fri, 26 Apr 2013 21:46:58 +0000 (17:46 -0400)]
synccall signal handler need not handle dead threads anymore
they have already blocked signals before decrementing the thread
count, so the code being removed is unreachable in the case where the
thread is no longer counted.
Rich Felker [Fri, 26 Apr 2013 21:30:32 +0000 (17:30 -0400)]
fix clobbering of signal mask when creating thread with sched attributes
this was simply a case of saving the state in the wrong place.
Rich Felker [Fri, 26 Apr 2013 20:16:04 +0000 (16:16 -0400)]
make last thread's pthread_exit give exit(0) a consistent state
the previous few commits ended up leaving the thread count and signal
mask wrong for atexit handlers and stdio cleanup.
Rich Felker [Fri, 26 Apr 2013 20:05:39 +0000 (16:05 -0400)]
use atomic decrement rather than cas in pthread_exit thread count
now that blocking signals prevents any application code from running
while the last thread is exiting, the cas logic is no longer needed to
prevent decrementing below zero.
Rich Felker [Fri, 26 Apr 2013 20:04:30 +0000 (16:04 -0400)]
add comments on some of the pthread_exit logic