clone calls back to a function pointer provided by the caller, which
will actually be a pointer to a function descriptor on fdpic. the
obvious solution is to have a separate version of clone for fdpic, but
I have taken a simpler approach to go around the problem. instead of
calling the pointed-to function from asm, a direct call is made to an
internal C function which then calls the pointed-to function. this
lets the C compiler generate the appropriate calling convention for an
indirect call with no need for ABI-specific assembly.
for fdpic support is is essential that the got pointer be saved at a
known, ABI-dictated offset from the frame pointer, since there is no
way to recover it once it's lost.
this error was only found by reading the code, but it seems to have
been causing gcc to produce wrong code in malloc: the same register
was used for the output and the high word of the input. in principle
this could have caused an infinite loop searching for an available
bin, but in practice most x86 models seem to implement the "undefined"
result of the bsf instruction as "unchanged".
originally, the comment in this code was correct and it would likely
work if the compiler generated a tail call to setjmp. however, commit
583e55122e redesigned sigsetjmp and
siglongjmp such that the old C implementation (which was not intended
to be used) is not even conceptually correct. remove it in the
interest of avoiding confusion when porting to new archs.
this restores the original behavior prior to the addition of the
byte-based C locale and fixes what is effectively a regression in
musl's property of always providing working UTF-8 support.
commit 1507ebf837 introduced the codeset
name "UTF-8-CODE-UNITS" for the byte-based C locale to represent that
the semantic content is UTF-8 but that it is being processed as code
units (bytes) rather than whole multibyte characters. however, many
programs assume that the codeset name is usable with iconv and/or
comes from a set of standard/widely-used names known to the
application. such programs are likely to produce warnings or errors,
run with reduced functionality, or mangle character data when run
explicitly in the C locale.
the standard places basically no requirements for the string returned
by nl_langinfo(CODESET) and how it interacts with other interfaces, so
returning "UTF-8" is permissible. moreover, it seems like the right
thing to do, since the identity of the character encoding as "UTF-8"
is independent of whether it is being processed as bytes of characters
by the standard library functions.
this fixes a bug reported by Nuno Gonçalves. previously, calling
fclose on stdin or stdout resulted in deadlock at exit time, since
__stdio_exit attempts to lock these streams to flush/seek them, and
has no easy way of knowing that they were closed.
conceptually, leaving a FILE stream locked on fclose is valid since,
in the abstract machine, it ceases to exist. but to satisfy the
implementation-internal assumption in __stdio_exit that it can access
these streams unconditionally, we need to unlock them.
it's also necessary that fclose leaves permanent streams in a state
where __stdio_exit will not attempt any further operations on them.
fortunately, the call to fflush already yields this property.
these functions are part of the ARM EABI, meaning compilers may
generate references to them. known versions of gcc do not use them,
but llvm does. they are not provided by libgcc, and the de facto
standard seems to be that libc provides them.
this functionality is affected by GNU make bug #30653, "intermediate
files incorrectly pruned in parallel builds". on affected versions of
make, parallel builds attempt to compile source files before
alltypes.h is generated.
as noted with commit a91ebdcfac, which
added the use of .SECONDARY, suppression of removal of "intermediate"
files does not seem to be needed at present. if it is needed in the
future, it should be achievable by explicitly mentioning their names
as targets or prerequisites.
at one point, GNU make was removing crt/*.o after producing the copies
in lib/ due to an arcane misfeature for handling "intermediate" files.
the circumstances that caused this are no longer present in our
makefile, but the previous workaround using .PRECIOUS was wrong and
could result in corrupt/partial files being left behind during an
interrupted build. using .SECONDARY is the correct, documented fix
that will prevent deletion of "intermediate" files from ever
resurfacing.
Some functions implemented in asm need to use EBP for purposes other
than acting as a frame pointer. (Notably, it is used for the 6th
argument to syscalls with 6 arguments.) Without frame pointers, GDB
can only show backtraces if it gets CFI information from a
.debug_frame or .eh_frame ELF section.
Rather than littering our asm with ugly .cfi directives, use an awk
script to insert them in the right places during the build process, so
GDB can keep track of where the current stack frame is relative to the
stack pointer. This means GDB can produce beautiful stack traces at
any given point when single-stepping through asm functions.
Additionally, when registers are saved on the stack and later
overwritten, emit ..cfi directives so GDB will know where they were
saved relative to the stack pointer. This way, when you look back up
the stack from within an asm function, you can still reliably print
the values of local variables in the caller.
If this awk script were to understand every possible wild and crazy
contortion that an asm programmer can do with the stack and registers,
and always emit the exact ..cfi directives needed for GDB to know what
the register values were in the preceding stack frame, it would
necessarily be as complex as a full x86 emulator. That way lies
madness.
Hence, we assume that the stack pointer will _only_ ever be adjusted
using push/pop or else add/sub with a constant. We do not attempt to
detect every possible way that a register value could be saved for
later use, just the simple and common ways.
Thanks to Szabolcs Nagy for suggesting numerous improvements to this
code.
getsubopt incorrectly returns the delimiting = in the value string,
this patch fixes it by increasing the pointer position by one.
Signed-off-by: Steven Barth <cyrus@openwrt.org>
commit 3c43c0761e fixed missing
synchronization in the atomic store operation for i386 and x86_64, but
opted to use mfence for the barrier on x86_64 where it's always
available. however, in practice mfence is significantly slower than
the barrier approach used on i386 (a nop-like lock orl operation).
this commit changes x86_64 (and x32) to use the faster barrier.
tm_gmtoff is a nonstandard field, but on historical systems which have
this field, it stores the offset of the local time zone from GMT or
UTC. this is the opposite of the POSIX extern long timezone object and
the offsets used in POSIX-form TZ strings, which represent the offset
from local time to UTC. previously we were storing these negated
offsets in tm_gmtoff too.
programs which only used this field indirectly via strftime were not
affected since strftime performed the negation for presentation.
however, some programs and libraries accesse tm_gmtoff directly and
were obtaining negated time zone offsets.
On 32bit systems long long arguments are passed in a special way
to some syscalls; this accidentally got copied to the AArch64 port.
The following interfaces were broken: fallocate, fanotify, ftruncate,
posix_fadvise, posix_fallocate, pread, pwrite, readahead,
sync_file_range, truncate.
tempnam uses an uninitialized buffer which is filled using memcpy and
__randname. It is therefore necessary to explicitly null-terminate it.
based on patch by Felix Janda.
during calls to free, any free chunks adjacent to the chunk being
freed are momentarily held in allocated state for the purpose of
merging, possibly leaving little or no available free memory for other
threads to allocate. under this condition, other threads will attempt
to expand the heap rather than waiting to use memory that will soon be
available. the race window where this happens is normally very small,
but became huge when free chooses to use madvise to release unused
physical memory, causing unbounded heap size growth.
this patch drastically shrinks the race window for unwanted heap
expansion by performing madvise with the bin lock held and marking the
bin non-empty in the binmask before making the expensive madvise
syscall. testing by Timo Teräs has shown this approach to be a
suitable mitigation.
more invasive changes to the synchronization between malloc and free
would be needed to completely eliminate the problem. it's not clear
whether such changes would improve or worsen typical-case performance,
or whether this would be a worthwhile direction to take malloc
development.
despite being strongly ordered, the x86 memory model does not preclude
reordering of loads across earlier stores. while a plain store
suffices as a release barrier, we actually need a full barrier, since
users of a_store subsequently load a waiter count to determine whether
to issue a futex wait, and using a stale count will result in soft
(fail-to-wake) deadlocks. these deadlocks were observed in malloc and
possible with stdio locks and other libc-internal locking.
on i386, an atomic operation on the caller's stack is used as the
barrier rather than performing the store itself using xchg; this
avoids the need to read the cache line on which the store is being
performed. mfence is used on x86_64 where it's always available, and
could be used on i386 with the appropriate cpu model checks if it's
shown to perform better.
The old code accepted atexit handlers after exit, but did not run them
reliably. C11 seems to explicitly allow atexit to fail (and report
such failure) in this case, but this situation can easily come up in
C++ if a destructor has a local static object with a destructor so it
should be handled.
Note that the memory usage can grow linearly with the overall number
of registered atexit handlers instead of with the worst case list
length. (This only matters if atexit handlers keep registering atexit
handlers which should not happen in practice).
Commit message/rationale based on text by Szabolcs Nagy.
when traditional syslogd implementations are restarted, the old server
socket ceases to exist and a new unix socket with the same pathname is
created. when this happens, the default destination address associated
with the client socket via connect is no longer valid, and attempts to
send produce errors. this happens despite the socket being datagram
type, and is in contrast to the behavior that would be seen with an IP
datagram (UDP) socket.
in order to avoid a situation where the application is unable to send
further syslog messages without calling closelog, this patch makes
syslog attempt to reconnect the socket when send returns an error
indicating a lost connection.
additionally, initial failure to connect the socket no longer results
in the socket being closed. this ensures that an application which
calls openlog to reserve the socket file descriptor will not run into
a situation where transient connection failure (e.g. due to syslogd
restart) prevents fd reservation. however, applications which may be
unable to connect the socket later (e.g. due to chroot, restricted
permissions, seccomp, etc.) will still fail to log if the syslog
socket cannot be connected at openlog time or if it has to be
reconnected later.
being nonstandard, the closest thing to a specification for this
function is its man page, which documents it as returning int. it can
fail with EBADF if the file descriptor passed is invalid.
due to a reversed pointer difference computation, ns_skiprr always
returned a negative value, which functions using it would interpret as
an error.
patch by Yu Lu.
musl-clang allows the user to compile musl-powered programs using their
already existent clang install, without the need of a special cross compiler.
it achieves this by wrapping around both the system clang install and the
linker and passing them special flags to re-target musl at runtime.
it does only affect invocations done through the special musl-clang wrapper
script, so that the user setup remains fully intact otherwise.
the clang wrapper consists of the compiler frontend wrapper script,
musl-clang, and the linker wrapper script, ld.musl-clang.
musl-clang makes sure clang invokes ld.musl-clang to link objects; neither
script needs to be in PATH for the wrapper to work.
the old test was broken in that it would never fail on a toolchains built
without dynamic linking support, leading to the wrapper script possibly being
installed on compilers that do not support it. in addition, the new test is
portable across compilers: the old test only worked on GCC.
the new test works by testing whether the toolchain libc defines __GLIBC__:
most non-musl Linux libc's do define this for compatibility even when they
are not glibc, so this is a safe bet to check for musl. in addition, the
compiler runtime would need to have a somewhat glibc-compatible ABI in the
first place, so any non-glibc compatible libc's compiler runtime might not
work. it is safer to disable these cases by default and have the user enable
the wrappers manually there using --enable-wrapper if they certain it works.
this overhauls part of the build system in order to support multiple
toolchain wrapper scripts, as opposed to solely the musl-gcc wrapper as
before. it thereby replaces --enable-gcc-wrapper with --enable-wrapper=...,
which has the options 'auto' (the default, detect whether to use wrappers),
'all' (build and install all wrappers), 'no' (don't build any) and finally
the options named after the individual compiler scripts (currently only
'gcc' is available) to build and install only that wrapper.
the old --enable-gcc-wrapper is removed from --help, but still available.
it also modifies the wrappers to use the C compiler specified to the build
system as 'inner' compiler, when applicable. as wrapper detection works by
probing this compiler, it may not work with any other.
this improves compatibility with the behavior of other systems and
with some applications which set an empty TZ var to disable use of
local time by mktime, etc.
The callers need to check the value of the pointer anyway, so make
them pass the pointer to gnu_lookup instead of reloading it there.
Reorder gnu_lookup arguments so that always-used ones are listed
first. GCC can choose a calling convention with arguments in registers
(e.g. up to 3 arguments in eax, ecx, edx on x86), but cannot reorder
the arguments for static functions.
Introduce gnu_lookup_filtered and use it to speed up symbol lookups in
find_sym (do_dlsym is left as is, based on an expectation that
frequently dlsym queries will use a dlopen handle rather than
RTLD_NEXT or RTLD_DEFAULT, and will not need to look at more than one
DSO).
the TLS ABI spec for mips, powerpc, and some other (presently
unsupported) RISC archs has the return value of __tls_get_addr offset
by +0x8000 and the result of DTPOFF relocations offset by -0x8000. I
had previously assumed this part of the ABI was actually just an
implementation detail, since the adjustments cancel out. however, when
the local dynamic model is used for accessing TLS that's known to be
in the same DSO, either of the following may happen:
1. the -0x8000 offset may already be applied to the argument structure
passed to __tls_get_addr at ld time, without any opportunity for
runtime relocations.
2. __tls_get_addr may be used with a zero offset argument to obtain a
base address for the module's TLS, to which the caller then applies
immediate offsets for individual objects accessed using the local
dynamic model. since the immediate offsets have the -0x8000 adjustment
applied to them, the base address they use needs to include the
+0x8000 offset.
it would be possible, but more complex, to store the pointers in the
dtv[] array with the +0x8000 offset pre-applied, to avoid the runtime
cost of adding 0x8000 on each call to __tls_get_addr. this change
could be made later if measurements show that it would help.
previously, loading of additional libraries beyond libc/ldso did not
work on nommu kernels, nor did loading programs via invocation of the
dynamic linker as a command.
this interface is non-standardized and is a GNU invention, and as
such, our implementation should match the behavior of the GNU
function. one peculiarity the old implementation got wrong was the
handling of all-zero digit sequences: they are supposed to compare
greater than digit sequences of which they are a proper prefix, as in
009 < 00.
in addition, high bytes were treated with char signedness rather than
as unsigned. this was wrong regardless of what the GNU function does
since the resulting order relation varied by arch.
the new strverscmp implementation makes explicit the cases where the
order differs from what strcmp would produce, of which there are only
two.
commit ba819787ee introduced this
regression. since the __malloc0 weak alias was not properly provided
by __simple_malloc, use of calloc forced the full malloc to be linked.
previously, calloc's implementation encoded assumptions about the
implementation of malloc, accessing a size_t word just prior to the
allocated memory to determine if it was obtained by mmap to optimize
out the zero-filling. when __simple_malloc is used (static linking a
program with no realloc/free), it doesn't matter if the result of this
check is wrong, since all allocations are zero-initialized anyway. but
the access could be invalid if it crosses a page boundary or if the
pointer is not sufficiently aligned, which can happen for very small
allocations.
this patch fixes the issue by moving the zero-fill logic into malloc.c
with the full malloc, as a new function named __malloc0, which is
provided by a weak alias to __simple_malloc (which always gives
zero-filled memory) when the full malloc is not in use.
this symbol is needed only on archs where the PLT call ABI is klunky,
and only for position-independent code compiled with stack protector.
thus references usually only appear in shared libraries or PIE
executables, but they can also appear when linking statically if some
of the object files being linked were built as PIC/PIE.
normally libssp_nonshared.a from the compiler toolchain should provide
__stack_chk_fail_local, but reportedly it appears prior to -lc in the
link order, thus failing to satisfy references from libc itself (which
arise only if libc.a was built as PIC/PIE with stack protector
enabled).
linux kernel commit 46e12c07b3b9603c60fc1d421ff18618241cb081 caused
the mips syscall mechanism to fail with EFAULT when the userspace
stack pointer is invalid, breaking __unmapself used for detached
thread exit. the workaround is to set $sp to a known-valid, readable
address, and the simplest one to obtain is the address of the current
function, which is available (per o32 calling convention) in $25.
nominally the low bits of the trap number on sh are the number of
syscall arguments, but they have never been used by the kernel, and
some code making syscalls does not even know the number of arguments
and needs to pass an arbitrary high number anyway.
sh3/sh4 traditionally used the trap range 16-31 for syscalls, but part
of this range overlapped with hardware exceptions/interrupts on sh2
hardware, so an incompatible range 32-47 was chosen for sh2.
using trap number 31 everywhere, since it's in the existing sh3/sh4
range and does not conflict with sh2 hardware, is a proposed
unification of the kernel syscall convention that will allow binaries
to be shared between sh2 and sh3/sh4. if this is not accepted into the
kernel, we can refit the sh2 target with runtime selection mechanisms
for the trap number, but doing so would be invasive and would entail
non-trivial overhead.
due to the way the interrupt and syscall trap mechanism works,
userspace on sh2 must never set the stack pointer to an invalid value.
thus, the approach used on most archs, where __unmapself executes with
no stack for the interval between SYS_munmap and SYS_exit, is not
viable on sh2.
in order not to pessimize sh3/sh4, the sh asm version of __unmapself
is not removed. instead it's renamed and redirected through code that
calls either the generic (safe) __unmapself or the sh3/sh4 asm,
depending on compile-time and run-time conditions.
the sh2 target is being considered an ISA subset of sh3/sh4, in the
sense that binaries built for sh2 are intended to be usable on later
cpu models/kernels with mmu support. so rather than hard-coding
sh2-specific atomics, the runtime atomic selection mechanisms that was
already in place has been extended to add sh2 atomics.
at this time, the sh2 atomics are not SMP-compatible; since the ISA
lacks actual atomic operations, the new code instead masks interrupts
for the duration of the atomic operation, producing an atomic result
on single-core. this is only possible because the kernel/hardware does
not impose protections against userspace doing so. additional changes
will be needed to support future SMP systems.
care has been taken to avoid producing significant additional code
size in the case where it's known at compile-time that the target is
not sh2 and does not need sh2-specific code.
functions which open in-memory FILE stream variants all shared a tail
with __fdopen, adding the FILE structure to stdio's open file list.
replacing this common tail with a function call reduces code size and
duplication of logic. the list is also partially encapsulated now.
function signatures were chosen to facilitate tail call optimization
and reduce the need for additional accessor functions.
with these changes, static linked programs that do not use stdio no
longer have an open file list at all.
Rich Felker
Timo Teräs
Alex Dowad
Steven Barth
Natanael Copa
Szabolcs Nagy
Roman Yeryomin
Felix Fietkau
Shiz
Alexander Monakov