checking for EINVAL should be sufficient, but qemu user emulation
returns EPROTONOSUPPORT in some of the failure cases, and it seems
conceivable that other kernels doing linux-emulation could make the
same mistake. since DNS lookups and other important code might break
if the fallback does not get invoked, be extra careful and check for
either error.
note that it's important NOT to perform the fallback code on other
errors such as resource-exhaustion cases, since the fallback is not
atomic and will lead to file-descriptor leaks in multi-threaded
programs that use exec. the fallback code is only "safe" to run when
the initial failure is caused by the application's choice of
arguments, not the system state.
also update syslog to use SOCK_CLOEXEC rather than separate fcntl
step, to make it safe in multithreaded programs that run external
programs.
emulation is not atomic; it could be made atomic by holding a lock on
forking during the operation, but this seems like overkill. my goal is
not to achieve perfect behavior on old kernels (which have plenty of
other imperfect behavior already) but to avoid catastrophic breakage
in (1) syslog, which would give no output on old kernels with the
change to use SOCK_CLOEXEC, and (2) programs built on a new kernel
where configure scripts detected a working SOCK_CLOEXEC, which later
get run on older kernels (they may otherwise fail to work completely).
new behavior can be summarized as:
inputs that parse completely as a decimal number are treated as one,
and rejected only if the result is out of 16-bit range.
inputs that do not parse as a decimal number (where strtoul leaves
anything left over in the input) are searched in /etc/services.
to deal with the fact that the public headers may be used with pre-c99
compilers, __restrict is used in place of restrict, and defined
appropriately for any supported compiler. we also avoid the form
[restrict] since older versions of gcc rejected it due to a bug in the
original c99 standard, and instead use the form *restrict.
the kernel wrongly expects the cmsg length field to be size_t instead
of socklen_t. in order to work around the issue, we have to impose a
length limit and copy to a local buffer. the length limit should be
more than sufficient for any real-world use; these headers are only
used for passing file descriptors and permissions between processes
over unix sockets.
we do not bother making h_errno thread-local since the only interfaces
that use it are inherently non-thread-safe. but still use the
potentially-thread-local ABI to access it just to avoid lock-in.
the buffer in getaddrinfo really only matters when /etc/hosts is huge,
but in that case, the huge number of syscalls resulting from a tiny
buffer would seriously impact the performance of every name lookup.
the buffer in __dns.c has also been enlarged a bit so that typical
resolv.conf files will fit fully in the buffer. there's no need to
make it so large as to dominate the syscall overhead for large files,
because resolv.conf should never be large.
if the file descriptor resource limit has been increased past
FD_SETSIZE, this is actually a security issue; we could write past the
end of the fd_set object. using poll makes it a non-issue, and
simplifies the code at the same time.
also, use clock_gettime instead of gettimeofday, for reduced bloat
and better entropy.
per POSIX and RFC 3493:
If the specified address family is AF_INET, AF_INET6, or AF_UNSPEC,
the service can be specified as a string specifying a decimal port
number.
021 is a valid decimal number, therefore, interpreting it as octal
seems to be non-conformant.
this patch improves the correctness, simplicity, and size of
cancellation-related code. modulo any small errors, it should now be
completely conformant, safe, and resource-leak free.
the notion of entering and exiting cancellation-point context has been
completely eliminated and replaced with alternative syscall assembly
code for cancellable syscalls. the assembly is responsible for setting
up execution context information (stack pointer and address of the
syscall instruction) which the cancellation signal handler can use to
determine whether the interrupted code was in a cancellable state.
these changes eliminate race conditions in the previous generation of
cancellation handling code (whereby a cancellation request received
just prior to the syscall would not be processed, leaving the syscall
to block, potentially indefinitely), and remedy an issue where
non-cancellable syscalls made from signal handlers became cancellable
if the signal handler interrupted a cancellation point.
x86_64 asm is untested and may need a second try to get it right.
we can do this without violating the namespace now that they are
macros/inline functions rather than extern functions. the motivation
is that gcc was generating giant, slow, horrible code for the old
functions, and now generates a single byte-swapping instruction.
POSIX clearly specifies the type of msg_iovlen and msg_controllen, and
Linux ignores it and makes them both size_t instead. to work around
this we add padding (instead of just using the wrong types like glibc
does), but we also need to patch-up the struct before passing it to
the kernel in case the caller did not zero-fill it.
if i could trust the kernel to just ignore the upper 32 bits, this
would not be necessary, but i don't think it will ignore them...
previously NULL was returned in ai_canonname, resulting in crashes in
some callers. this behavior was incorrect. note however that the new
behavior differs from glibc, which performs reverse dns lookups. POSIX
is very clear that a reverse DNS lookup must not be performed for
numeric addresses.
sadly the C language does not specify any such implicit conversion, so
this is not a matter of just fixing warnings (as gcc treats it) but
actual errors. i would like to revisit a number of these changes and
possibly revise the types used to reduce the number of casts required.
Rich Felker