the concept of both versions is the same; they differ only in details.
for long runs, they use "rep movsl" or "rep movsq", and for small
runs, they use a trick, writing from both ends towards the middle,
that reduces the number of branches needed. in addition, if memset is
called multiple times with the same length, all branches will be
predicted; there are no loops.
for larger runs, there are likely faster approaches than "rep", at
least on some cpu models. for 32-bit, it's unlikely that there is any
faster approach that does not require non-baseline instructions; doing
anything fancier would require inspecting cpu capabilities. for
64-bit, there may very well be faster versions that work on all
models; further optimization could be explored in the future.
with these changes, memset is anywhere between 50% faster and 6 times
faster, depending on the cpu model and the length and alignment of the
destination buffer.
the original motivation for this patch was that qemu (and possibly
other syscall emulators) nop out madvise, resulting in an infinite
loop. however, there is another benefit to this change: madvise may
actually undo an explicit madvise the application intended for its
stack, whereas the mremap operation is a true nop. the logic here is
that mremap must fail if it cannot resize the mapping in-place, and
the caller knows that it cannot resize in-place because it knows the
next page of virtual memory is already occupied.
one of the arguments to memcmp may be shorter than the length l-3, and
memcmp is under no obligation not to access past the first byte that
differs. instead use strncmp which conveys the correct semantics. the
performance difference is negligible here and since the code is only
use for shared libc, both functions are already linked anyway.
the dev/inode for the main app and the dynamic linker ("interpreter")
are not available, so the subsequent checks don't work. in general we
don't want to make exact string matches to existing libraries prevent
loading new ones, since this breaks loading upgraded modules in
module-loading systems. so instead, special-case it.
the motivation for this fix is that calling dlopen on the names
returned by dl_iterate_phdr or walking the link map (obtained by
dlinfo) seem to be the only methods available to an application to
actually get a list of open dso handles.
reject elf files which are not ET_EXEC/ET_DYN type as bad exec format,
and reject ET_EXEC files when they cannot be loaded at the correct
address, since they are not relocatable at runtime. the main practical
benefit of this is to make dlopen of the main program fail rather than
producing an unsafe-to-use handle.
it's not clear to me why the linker even outputs these headers if they
are null, but apparently it does so. with the default startfiles, they
will never be null anyway, but this patch allows eliminating crti,
crtn, crtbegin, and crtend (leaving only crt1) if the toolchain is
using init_array/fini_array (or for a C-only, no-ctor environment).
LC_GLOBAL_LOCALE refers to the global locale, controlled by setlocale,
not the thread-local locale in effect which these functions should be
using. neither LC_GLOBAL_LOCALE nor 0 has an argument to the *_l
functions has behavior defined by the standard, but 0 is a more
logical choice for requesting the callee to lookup the current locale.
in the future I may move the current locale lookup the the caller (the
non-_l-suffixed wrapper).
at this point, all of the locale logic is dummied out, so no harm was
done, but it should at least avoid misleading usage.
also add a warning to the existing sys/poll.h. the warning is absent
from sys/dir.h because it is actually providing a slightly different
API to the program, and thus just replacing the #include directive is
not a valid fix to programs using this one.
entry point was wrong for PIE. e_entry was being treated as an
absolute value, whereas it's actually relative to the load address
(which is zero for non-PIE).
phdr pointer was wrong for non-PIE. e_phoff was being treated as
load-address-relative, whereas it's actually a file offset in the ELF
file. in any case, map_library was already computing it correctly, and
the incorrect code in __dynlink was overwriting it with junk.
the only immediate effect of this commit is enabling PIE support on
some archs that did not previously have any Scrt1.s, since the
existing asm files for crt1 override this C code. so some of the
crt_arch.h files committed are only there for the sake of documenting
what their archs "would do" if they used the new C-based crt1.
the expectation is that new archs should use this new system rather
than using heavy asm for crt1. aside from being easier and less
error-prone, it also ensures that PIE support is available immediately
(since Scrt1.o is generated from the same C source, using -fPIC)
rather than having to be added as an afterthought in the porting
process.
based on a patch by orc. POSIX actually fails to specify the format of
the ntop conversion; presumably, any output that will correctly
round-trip back via the (well-specified) pton operation is acceptable.
the new behavior is much more convenient than the old, however.
this patch also affects getnameinfo, which is implemented in terms of
inet_ntop and which is the preferred interface for performing this
conversion.
I've also removed some inexplicable cruft (filling the buffer with 'x'
before doing anything) whose origin I was unable to track down.
it's not clear that -O3 helps them, and gcc seems to have floating
point optimization bugs that introduce additional failures when -O3 is
used on some of these files.
apparently the original kernel commit's i386 version of siginfo.h
defined this field as unsigned int, but the asm-generic file always
had void *. unsigned int is obviously not a suitable type for an
address, in a non-arch-specific file, and glibc also has void * here,
so I think void * is the right type for it.
also fix redundant type specifiers.
this is a cheat since the _l versions take an extra argument, but
since these functions are only here for ABI purposes, it doesn't
really matter as long as the ABI matches. if the non-__-prefixed
versions are eventually made public, they should proabably be real
functions rather than hacks like this.
unlike the strftime commit, this one is purely an ABI compatibility
issue. the previous version of the code would have worked just as well
with LC_TIME once LC_TIME support is added.
based on a patch by orc, with indexing and flow control cleaned up a
little bit. this code is all going to be replaced at some point in the
near future.
these are needed for some C++ library binaries including most builds
of libstdc++. I'm not entirely clear on the rationale. this patch does
not implement any special semantics for them, but as far as I can
tell, no special treatment is needed in correctly-linked programs;
this binding seems to exist only for catching incorrectly-linked
programs.
this is necessary to meet the C++ ABI target. alternatives were
considered to avoid the size increase for non-sig jmp_buf objects, but
they seemed to have worse properties. moreover, the relative size
increase is only extreme on x86[_64]; one way of interpreting this is
that, if the size increase from this patch makes jmp_buf use too much
memory, then the program was already using too much memory when built
for non-x86 archs.
this bug was caught by the new footer-corruption check in realloc and
free.
if the block returned by malloc was already aligned to the desired
alignment, memalign's logic to split off the misaligned head was
incorrect; rather than writing to a point inside the allocated block,
it was overwriting the footer of the previous block on the heap with
the value 1 (length 0 plus an in-use flag).
fortunately, the impact of this bug was fairly low. (this is probably
why it was not caught sooner.) due to the way the heap works, malloc
will never return a block whose previous block is free. (doing so would
be harmful because it would increase fragmentation with no benefit.)
the footer is actually not needed for in-use blocks, except that its
in-use bit needs to remain set so that it does not get merged with
free blocks, so there was no harm in it being set to 1 instead of the
correct value.
however, there is one case where this bug could have had an impact: in
multi-threaded programs, if another thread freed the previous block
after memalign's call to malloc returned, but before memalign
overwrote the previous block's footer, the resulting block in the free
list could be left in a corrupt state. I have not analyzed the impact
of this bad state and whether it could lead to more serious
malfunction.
the motivation for this patch is that the vast majority of libc is
code that does not benefit at all from optimizations, but that certain
components like string/memory operations can be major performance
bottlenecks.
at the same time, the old -falign-*=1 options are removed, since they
were only beneficial for avoiding bloat when global -O3 was used, and
in that case, they may have prevented some of the performance gains.
to be the most useful, this patch will need further tuning. in
particular, research is needed to determine which components should be
built with -O3 by default, and it may be desirable to remove the
hard-coded -O3 and instead allow more customization of the
optimization level used for selected modules.
this patch is something of a compromise for a compatibility
regression discovered after the header refactoring: libtiff uses
_Int64 for its own use. this is absolutely wrong, invalid C, and
should not be supported, but it's also frustrating for users when code
that used to work suddenly breaks.
rather than leave the breakage in place or change musl internals to
accommodate broken software, I've found a change that makes the
problem go away and improves musl. by undefining these macros at the
end of alltypes.h, the temptation to use them in other headers is
removed. (for example, I almost used _Int64 in sys/types.h to define
u_int64_t rather than adding it back to alltypes.h.) by confining use
of these macros to alltypes.h, we keep it easy to go back and change
the implementation of alltypes later, if needed.
during the header refactoring, I had moved u_int64_t out of alltypes
under the assumption that we could just use long long everywhere.
however, it seems some broken applications make inconsistent mixed use
of u_int64_t and uint64_t, resulting in build errors when the
underlying type differs.
rather than moving nlink_t back to the arch-specific file, I've added
a macro _Reg defined to the canonical type for register-size values on
the arch. this is not the same as _Addr for (not-yet-supported)
32-on-64 pseudo-archs like x32 and mips n32, so a new macro was
needed.
for regoff_t, it's impossible to match on 64-bit archs because glibc
defined the type in a non-conforming way. however this change makes
the type match on 32-bit archs.
since the old, poorly-thought-out musl approach to init/fini arrays on
ARM (when it was the only arch that needed them) was to put the code
in crti/crtn and have the legacy _init/_fini code run the arrays,
adding proper init/fini array support caused the arrays to get
processed twice on ARM. I'm not sure skipping legacy init/fini
processing is the best solution to the problem, but it works, and it
shouldn't break anything since the legacy init/fini system was never
used for ARM EABI.
aside from the obvious C++ ABI purpose for this change, it also brings
musl into alignment with the compiler's idea of the definition of
wint_t (use in -Wformat), and makes the situation less awkward on ARM,
where wchar_t is unsigned.
internal code using wint_t and WEOF was checked against this change,
and while a few cases of storing WEOF into wchar_t were found, they
all seem to operate properly with the natural conversion from unsigned
to signed.
the arch-specific bits/alltypes.h.sh has been replaced with a generic
alltypes.h.in and minimal arch-specific bits/alltypes.h.in.
this commit is intended to have no functional changes except:
- exposing additional symbols that POSIX allows but does not require
- changing the C++ name mangling for some types
- fixing the signedness of blksize_t on powerpc (POSIX requires signed)
- fixing the limit macros for sig_atomic_t on x86_64
- making dev_t an unsigned type (ABI matching goal, and more logical)
in addition, some types that were wrongly defined with long on 32-bit
archs were changed to int, and vice versa; this change is
non-functional except for the possibility of making pointer types
mismatch, and only affects programs that were using them incorrectly,
and only at build-time, not runtime.
the following changes were made in the interest of moving
non-arch-specific types out of the alltypes system and into the
headers they're associated with, and also will tend to improve
application compatibility:
- netdb.h now includes netinet/in.h (for socklen_t and uint32_t)
- netinet/in.h now includes sys/socket.h and inttypes.h
- sys/resource.h now includes sys/time.h (for struct timeval)
- sys/wait.h now includes signal.h (for siginfo_t)
- langinfo.h now includes nl_types.h (for nl_item)
for the types in stdint.h:
- types which are of no interest to other headers were moved out of
the alltypes system.
- fast types for 8- and 64-bit are hard-coded (at least for now); only
the 16- and 32-bit ones have reason to vary by arch.
and the following types have been changed for C++ ABI purposes;
- mbstate_t now has a struct tag, __mbstate_t
- FILE's struct tag has been changed to _IO_FILE
- DIR's struct tag has been changed to __dirstream
- locale_t's struct tag has been changed to __locale_struct
- pthread_t is defined as unsigned long in C++ mode only
- fpos_t now has a struct tag, _G_fpos64_t
- fsid_t's struct tag has been changed to __fsid_t
- idtype_t has been made an enum type (also required by POSIX)
- nl_catd has been changed from long to void *
- siginfo_t's struct tag has been removed
- sigset_t's has been given a struct tag, __sigset_t
- stack_t has been given a struct tag, sigaltstack
- suseconds_t has been changed to long on 32-bit archs
- [u]intptr_t have been changed from long to int rank on 32-bit archs
- dev_t has been made unsigned
summary of tests that have been performed against these changes:
- nsz's libc-test (diff -u before and after)
- C++ ABI check symbol dump (diff -u before, after, glibc)
- grepped for __NEED, made sure types needed are still in alltypes
- built gcc 3.4.6
these functions were mistakenly assumed to be needed to match glibc
ABI, but glibc has them as part of the non-shared part of libc that's
always statically linked into the main program. moreover, the only
place they are referenced from is glibc's crt1.o.
modern (4.7.x and later) gcc uses init/fini arrays, rather than the
legacy _init/_fini function pasting and crtbegin/crtend ctors/dtors
system, on most or all archs. some archs had already switched a long
time ago. without following this change, global ctors/dtors will cease
to work under musl when building with new gcc versions.
the most surprising part of this patch is that it actually reduces the
size of the init code, for both static and shared libc. this is
achieved by (1) unifying the handling main program and shared
libraries in the dynamic linker, and (2) eliminating the
glibc-inspired rube goldberg machine for passing around init and fini
function pointers. to clarify, some background:
the function signature for __libc_start_main was based on glibc, as
part of the original goal of being able to run some glibc-linked
binaries. it worked by having the crt1 code, which is linked into
every application, static or dynamic, obtain and pass pointers to the
init and fini functions, which __libc_start_main is then responsible
for using and recording for later use, as necessary. however, in
neither the static-linked nor dynamic-linked case do we actually need
crt1.o's help. with dynamic linking, all the pointers are available in
the _DYNAMIC block. with static linking, it's safe to simply access
the _init/_fini and __init_array_start, etc. symbols directly.
obviously changing the __libc_start_main function signature in an
incompatible way would break both old musl-linked programs and
glibc-linked programs, so let's not do that. instead, the function can
just ignore the information it doesn't need. new archs need not even
provide the useless args in their versions of crt1.o. existing archs
should continue to provide it as long as there is an interest in
having newly-linked applications be able to run on old versions of
musl; at some point in the future, this support can be removed.
for conversion specifiers, alloc is always set when the specifier is
parsed. however, if scanf stops due to mismatching literal text,
either an uninitialized (if no conversions have been performed yet) or
stale (from the previous conversion) of the flag will be used,
possibly causing an invalid pointer to be passed to free when the
function returns.
the sizes in the header and footer for a chunk should always match. if
they don't, the program has definitely invoked undefined behavior, and
the most likely cause is a simple overflow, either of a buffer in the
block being freed or the one just below it.
crashing here should not only improve security of buggy programs, but
also aid in debugging, since the crash happens in a context where you
have a pointer to the likely-overflowed buffer.
while there's no POSIX namespace provision for UIO_* in uio.h, this
exact macro name is reserved in XBD 2.2.2. apparently some
glibc-centric software expects it to exist, so let's provide it.
the main aim of this patch is to ensure that if not all fields are
filled in, they contain zeros, so as not to confuse applications.
reportedly some older kernels, including commonly used openvz kernels,
lack the f_flags field, resulting in applications reading random junk
as the mount flags; the common symptom seems to be wrongly considering
the filesystem to be mounted read-only and refusing to operate. glibc
has some amazingly ugly fallback code to get the mount flags for old
kernels, but having them really is not that important anyway; what
matters most is not presenting incorrect flags to the application.
I have also aimed to fill in some fields of statvfs that were
previously missing, and added code to explicitly zero the reserved
space at the end of the structure, which will make things easier in
the future if this space someday needs to be used.
this change is both to fix one of the remaining type (and thus C++
ABI) mismatches with glibc/LSB and to allow use of the full range of
uid and gid values, if so desired.
passwd/group access functions were not prepared to deal with unsigned
values, so they too have been fixed with this commit.
prior to this change, using a non-default syslibdir was impractical on
systems where the ordinary library paths contain musl-incompatible
library files. the file containing search paths was always taken from
/etc, which would either correspond to a system-wide musl
installation, or fail to exist at all, resulting in searching of the
default library path.
the new search strategy is safe even for suid programs because the
pathname used comes from the PT_INTERP header of the program being
run, rather than any external input.
as part of this change, I have also begun differentiating the names of
arch variants that differ by endianness or floating point calling
convention. the corresponding changes in the build system and and gcc
wrapper script (to use an alternate dynamic linker name) for these
configurations have not yet been made.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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