This adds _hurd_sigstate_set_global_rcv used by libpthread to enable
POSIX-confirming behavior of signals on a per-thread basis.
This also provides a sigstate destructor _hurd_sigstate_delete, and a
global process signal state, which needs to be locked and check when
global disposition is enabled, thus the addition of _hurd_sigstate_lock
_hurd_sigstate_actions _hurd_sigstate_pending _hurd_sigstate_unlock helpers.
This also updates all the glibc code accordingly.
This also drops support for get_int(INIT_SIGMASK), which did not make sense
any more since we do not have a single signal thread any more.
During fork/spawn, this also reinitializes the child global sigstate's
lock. That cures an issue that would very rarely cause a deadlock in the
child in fork, tries to unlock ss' critical section lock at the end of
fork. This will typically (always?) be observed in /bin/sh, which is not
surprising as that is the foremost caller of fork.
To reproduce an intermediate state, add an endless loop if
_hurd_global_sigstate is locked after __proc_dostop (cast through
volatile); that is, while still being in the fork's parent process.
When that triggers (use the libtool testsuite), the signal thread has
already locked ss (which is _hurd_global_sigstate), and is stuck at
hurdsig.c:685 in post_signal, trying to lock _hurd_siglock (which the
main thread already has locked and keeps locked until after
__task_create). This is the case that ss->thread == MACH_PORT_NULL, that
is, a global signal. In the main thread, between __proc_dostop and
__task_create is the __thread_abort call on the signal thread which would
abort any current kernel operation (but leave ss locked). Later in fork,
in the parent, when _hurd_siglock is unlocked in fork, the parent's
signal thread can proceed and will unlock eventually the global sigstate.
In the client, _hurd_siglock will likewise be unlocked, but the global
sigstate never will be, as the client's signal thread has been configured
to restart execution from _hurd_msgport_receive. Thus, when the child
tries to unlock ss' critical section lock at the end of fork, it will
first lock the global sigstate, will spin trying to lock it, which can
never be successful, and we get our deadlock.
Options seem to be:
* Move the locking of _hurd_siglock earlier in post_signal -- but that
may generally impact performance, if this locking isn't generally
needed anyway?
On the other hand, would it actually make sense to wait here until we
are not any longer in a critical section (which is meant to disable
signal delivery anyway (but not for preempted signals?))?
* Clear the global sigstate in the fork's child with the rationale that
we're anyway restarting the signal thread from a clean state. This
has now been implemented.
Why has this problem not been observed before Jérémie's patches? (Or has
it? Perhaps even more rarely?) In _S_msg_sig_post, the signal is now
posted to a *global receiver thread*, whereas previously it was posted to
the *designated signal-receiving thread*. The latter one was in a
critical section in fork, so didn't try to handle the signal until after
leaving the critical section? (Not completely analyzed and verified.)
Another question is what the signal is that is being received
during/around the time __proc_dostop executes.
Not only libc/rtld use __close_nocancel_nostatus.
* sysdeps/mach/hurd/Makefile [$(subdir) == io] (sysdep_routines): Add
close_nocancel_nostatus.
* sysdeps/mach/hurd/Versions (libc): Add __close_nocancel_nostatus to
GLIBC_PRIVATE.
* sysdeps/mach/hurd/not-cancel.h (__close_nocancel_nostatus): Declare
function instead of defining inline.
[IS_IN (libc) || IS_IN (rtld)] (__close_nocancel_nostatus): Make
function hidden.
* sysdeps/mach/hurd/close_nocancel_nostatus.c: New file.
The logic for generating sysdeps/mach/hurd/bits/errno.h involves a
stamp file and $(move-if-change).
The temporary file (generated unconditionally) is generated in the
source directory. This means that even if
sysdeps/mach/hurd/bits/errno.h is up to date, and has an up to date
timestamp, the build will fail if the source directory is read-only.
Even with a writable source directory, multiple concurrent builds for
i686-gnu with the same source directory could race to access the
temporary file (which always has the same name).
This patch uses the build directory for the temporary file instead to
avoid those problems. (In the case where the file is out of date and
the temporary file does need to be moved to the source directory, if
there are multiple concurrent builds for i686-gnu with the same source
directory, and the source and build directories are on different
filesystems, it's possible there might still be races replacing the
file in the source directory, depending on exactly how mv handles such
cross-filesystem moves. This is certainly no worse than the present
situation, where such a case would have races regardless of whether
the file is out of date or whether different filesystems are in use.)
Tested with a build-many-glibcs.py build for i686-gnu.
* sysdeps/mach/hurd/Makefile ($(common-objpfx)stamp-errnos): Use
$(hurd-objpfx)bits/errno.h-tmp, not $(hurd)/bits/errno.h-tmp.
Generating errnos.d does not actually need libc-modules.h.
* sysdeps/mach/hurd/Makefile ($(common-objpfx)errnos.d): Strip
"-include $(common-objpfx)libc-modules.h" from CPPFLAGS, and do not
depend on libc-modules.h,
when initial make call has subdir= explicitly set.
* sysdeps/mach/Makefile ($(patsubst
mach%,m\%h%,$(mach-before-compile))): Force subdir to mach when
calling $(MAKE).
* sysdeps/mach/hurd/Makefile ($(patsubst %,$(hurd-objpfx)hurd/%.%,auth
io fs process)): Force subdir to hurd when calling $(MAKE).
($(common-objpfx)hurd/../mach/RPC_task_get_sampled_pcs.c): Force
subdir to mach when calling $(MAKE).
Add net/ethernet.h net/if_arp.h net/if_ether.h net/if_ppp.h
net/route.h.
* hurd/report-wait.c (describe_number): Use __stpcpy to prepend
flavor to description only when flavor is not NULL.
* hurd/hurdsig.c (signal_allowed): Unlock _hurd_dtable_lock
mutex after SIGIO/SIGURG lookup loop.
* sysdeps/mach/Makefile (mach-before-compile rule): Pass generating=t,
use target mach-before-compile instead of generated.
(mach-before-compile): New target.
* Makerules: Disable use of tls.make if $(generating) is set.
* mach/Makefile: Include mach-syscalls.mk even if no_deps=t.
* sysdeps/mach/hurd/Makefile: Make generated hurd headers depend on
mach-shortcuts.h so mach/ generation happens before hurd/.
* sysdeps/mach/powerpc/machine-sp.h: New file.
* sysdeps/mach/powerpc/sysdep.h: New file.
* sysdeps/mach/powerpc/thread_state.h: New file.
* sysdeps/mach/hurd/i386/Makefile: File removed; crt0.o rules moved ...
* sysdeps/mach/hurd/Makefile: ... to here. For the time being, we can
expect to need the same thing on every platform.
* sysdeps/unix/sysv/linux/powerpc/sys/ucontext.h: File moved ...
* sysdeps/generic/sys/ucontext.h: ... to replace this one.
* sysdeps/mach/hurd/Makefile: Add $(CFLAGS) for generation of
bits/errno.h
2001-10-07 Roland McGrath <roland@frob.com>
* sysdeps/mach/hurd/setrlimit.c (__setrlimit): Never return ENOSYS.
2001-07-06 Paul Eggert <eggert@twinsun.com>
* manual/argp.texi: Remove ignored LGPL copyright notice; it's
not appropriate for documentation anyway.
* manual/libc-texinfo.sh: "Library General Public License" ->
"Lesser General Public License".
2001-07-06 Andreas Jaeger <aj@suse.de>
* All files under GPL/LGPL version 2: Place under LGPL version
2.1.
Declare these without `static', to avoid warning that we haven't
defined them.
* Makerules (build-module-helper-objlist): New variable,
broken out of build-module. Insert --whole-archive/--no-whole-archive
immediately each *_pic.a file in the link object list.
(whole-archive): New variable, used by that.
(build-module-objlist): New variable using build-module-helper-objlist.
(build-module, build-module-helper): Use it.
(build-shlib-objlist): New variable using build-module-helper-objlist.
(build-shlib): Use that.
* sysdeps/mach/hurd/Makefile ($(link-rpcuserlibs)): Likewise.
* sysdeps/mach/hurd/Makefile ($(link-rpcuserlibs)): Don't use
$(build-module) here, since we don't need the full magilla.
Just use $(build-module-helper) and append our special options.
options after $(build-module), which is no longer a single command.
(LDFLAGS-link-rpcuserlibs): New variable to hold those options.
(LDFLAGS-libmachuser-link.so, LDFLAGS-libhurduser-link.so): New
variables using it. Reported by Mark Kettenis <kettenis@gnu.org>.
2001-03-11 Roland McGrath <roland@frob.com>
* elf/rtld.c (dl_main): Add cast to quiet warning.
* elf/dl-load.c (_dl_map_object_from_fd): Likewise.
* iconv/gconv_conf.c (__gconv_get_path): Likewise.
(__gconv_read_conf): Likewise.
* iconv/gconv_db.c (gen_steps): Likewise.
(__gconv_find_transform): Likewise.
* locale/programs/charmap-dir.c (fopen_uncompressed): Likewise.
(fopen_uncompressed): Use const in second argument's type.
2001-03-11 Roland McGrath <roland@frob.com>
* sysdeps/mach/hurd/dl-sysdep.c (__writev): New function.
* mach/mach_error.h: Fix ancient #endif syntax.
* hurd/hurdmalloc.c: Likewise.
* sysdeps/generic/atomicity.h (exchange_and_add, atomic_add,
compare_and_swap): Add volatile qualifier to first arg, to bring
these prototypes in line with all the other implementations.
Add a #warning to remind the builder that these are not atomic.
2001-03-04 Roland McGrath <roland@frob.com>
* hurd/vpprintf.c (vpprintf) [USE_IN_LIBIO]: Use innermore first
member of TEMP_F so its type is _IO_FILE * as expected.
* hurd/fopenport.c (seekio): Conditionalize type of POS argument
on [USE_IN_LIBIO]. Check the value for overflow, since for libio
it might exceed off_t's range.
2001-02-25 Roland McGrath <roland@frob.com>
* sysdeps/mach/hurd/sysd-stdio.c (__stdio_reopen): Use prototype.
* sysdeps/unix/bsd/getpt.c (__posix_openpt): Likewise.
* sysdeps/mach/hurd/i386/intr-msg.h (SYSCALL_EXAMINE): Make this a
macro instead of an inline function,
(MSG_EXAMINE): Likewise.
* hurd/hurd/ioctl.h (_HURD_HANDLE_IOCTLS_1): Remove extraneous ##.
* hurd/get-host.c: Include <string.h> for decls of built-ins.
* hurd/hurdchdir.c: Likewise.
* sysdeps/mach/hurd/i386/sigreturn.c: Likweise.
* sysdeps/mach/hurd/opendir.c: Likewise.
* mach/mach_init.c: Include <unistd.h> for _exit decl.
* mach/errsystems.awk: Give output file a trailing newline.
* mach/Machrules (MIG): Prepend CC='${CC}' to the mig command, so that
the script will use our compiler for the cpp stage.
With --enable-all-warnings we get a whole bunch of warnings,
and these are just a few fixes; there are more needed to silence it.
* mach/mach/mig_support.h (__mig_strncpy, mig_strncpy): Add extern
decls before extern inline defns to quiet gcc warning.
* hurd/hurd/threadvar.h (__hurd_threadvar_location_from_sp): Likewise.
* sysdeps/mach/i386/machine-sp.h (__thread_stack_pointer): Likewise.
argument type, for 2000-02-22 change in signal/signal.h declaration.
* sysdeps/generic/getdents64.c (__getdirentries64): Renamed
to getdirentries64 (no __), since the Linux implementation
has no __ name, and there is no declaration for the __ name.
* sysdeps/mach/bits/libc-lock.h (__libc_cleanup_end): New macro.
* sysdeps/mach/hurd/Makefile ($(common-objpfx)errnos.d): Tighten up
sed regexp so as not to break directory names including /../ within.
* sysdeps/unix/bsd/bits/waitflags.h: File removed.
The sysdeps/generic file is right for BSD. This file was in fact the
Hurd file, but the only difference was the WNOREAP flag that was never
actually implemented or used anywhere.
* sysdeps/mach/hurd/nfs/nfs.h: New file, empty but for comments.
Its presence lets librpcsvc compile ok. This from Mark Kettenis.
We don't install this file for users.
* sysdeps/mach/hurd/Dist: Add it.
* sysdeps/mach/hurd/Makefile [$(subdir) = sunrpc] (sysdep_headers):
Append nfs/nfs.h here.
* sysdeps/mach/hurd/Makefile: Undo last change.
[$(subdir) = mach] (lib-noranlib): Depend on libmachuser-link.so.
[$(subdir) = hurd] (lib-noranlib): Depend on libhurduser-link.so.
* mach/Machrules (+preinit, +postinit, +interp,
LDFLAGS-$(interface.so)): Variables removed. Interface libraries get
linked just like other shared libraries.
(interface.so): Fix typo in defn, so it works.
($(objpfx)$(interface.so)): Make it depend on libc.so.
* sysdeps/mach/hurd/fcntl.c (__fcntl: case F_GETLK,F_SETLK,F_SETLKW):
Support whole-file locking (only) by using the file_lock RPC. This
has very wrong semantics, but is better than nothing for the time
being. (Correct POSIX.1 locking will require new RPCs in the Hurd
protocols that will be somewhat hairy to specify.)
