I used these shell commands:
../glibc/scripts/update-copyrights $PWD/../gnulib/build-aux/update-copyright
(cd ../glibc && git commit -am"[this commit message]")
and then ignored the output, which consisted lines saying "FOO: warning:
copyright statement not found" for each of 7061 files FOO.
I then removed trailing white space from math/tgmath.h,
support/tst-support-open-dev-null-range.c, and
sysdeps/x86_64/multiarch/strlen-vec.S, to work around the following
obscure pre-commit check failure diagnostics from Savannah. I don't
know why I run into these diagnostics whereas others evidently do not.
remote: *** 912-#endif
remote: *** 913:
remote: *** 914-
remote: *** error: lines with trailing whitespace found
...
remote: *** error: sysdeps/unix/sysv/linux/statx_cp.c: trailing lines
I used these shell commands:
../glibc/scripts/update-copyrights $PWD/../gnulib/build-aux/update-copyright
(cd ../glibc && git commit -am"[this commit message]")
and then ignored the output, which consisted lines saying "FOO: warning:
copyright statement not found" for each of 6694 files FOO.
I then removed trailing white space from benchtests/bench-pthread-locks.c
and iconvdata/tst-iconv-big5-hkscs-to-2ucs4.c, to work around this
diagnostic from Savannah:
remote: *** pre-commit check failed ...
remote: *** error: lines with trailing whitespace found
remote: error: hook declined to update refs/heads/master
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.
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.
1998-07-20 Jose M. Moya <josem@gnu.org>
* hurd/hurdmsg.c (_S_msg_get_env_variable): Copy getenv return
value.
1998-07-20 Roland McGrath <roland@baalperazim.frob.com>
* sysdeps/mach/hurd/getdomain.c: New file.
* sysdeps/mach/hurd/setdomain.c: New file.
1998-07-20 Roland McGrath <roland@baalperazim.frob.com>
* sysdeps/unix/bsd/sigsuspend.c: Define __sigsuspend with sigsuspend as
a weak alias.
* hurd/hurd-raise.c (_hurd_raise_signal): Pass sigcode in msg_sig_post
rpc.
* hurd/hurdmsg.c (_S_msg_set_environment): Use argz.h functions
instead of _hurd_split_args.
(_S_msg_*_exec_flags): Functions removed.
(_S_msg_startup_dosync): Stub removed.
Sat May 4 02:11:55 1996 Roland McGrath <roland@delasyd.gnu.ai.mit.edu>
* sysdeps/mach/hurd/ptrace.c: Set _hurdsig_traced instead of
EXEC_TRACED bit in _hurd_exec_flags.
Pass sigcode arg in msg_sig_post_untraced rpc.
* sysdeps/mach/hurd/access.c: Don't pass io port in
auth_user_authenticate rpc.
* posix/sched.h: Fix typos.
* sysdeps/mach/hurd/fork.c: Use new critical section lock.
Clear _hurdsig_traced instead of EXEC_TRACED.
* sysdeps/stub/nanosleep.c (nanosleep): Fix typo.
* wcsmbs/wcstol.c: Find strtol.c in ../stdlib.
* wcsmbs/wcstof.c: Find strtod.c in ../stdlib.
* wcsmbs/wcstod.c: Likewise.
* wcsmbs/wcstold.c: Likewise.
* wcsmbs/wcwidth.h: Find cname-lookup.h in ../wctype.
* string/envz.c (envz_entry): Use const.
(envz_get, envz_remove): Likewise.
(envz_entry): Return char *, not const char *.
* string/envz.h: Fix decl.
* string/argz-create.c: Use const in prototype.
* string/argz-next.c: Likewise.
* sysdeps/mach/hurd/sigprocmask.c: Pass sigcode arg to msg_sig_post.
* sysdeps/mach/hurd/i386/sigreturn.c: Likewise.
* sysdeps/mach/hurd/sigsuspend.c: Likewise.
* sysdeps/mach/hurd/kill.c: Likewise.
* hurd/hurdexec.c (_hurd_exec): Use new critical section lock.
* hurd/catch-exc.c (_S_catch_exception_raise): Likewise.
* hurd/sigunwind.c (_hurdsig_longjmp_from_handler): Likewise.
* hurd/thread-cancel.c (hurd_thread_cancel, hurd_check_cancel):
Likewise.
* sysdeps/mach/hurd/jmp-unwind.c (_longjmp_unwind): Likewise.
* sysdeps/mach/hurd/sigaction.c: Likewise.
* sysdeps/mach/hurd/errnos.awk: Don't use ARGV in comment; it can
change meaninglessly.
* hurd/hurd/signal.h (struct hurd_sigstate): Replace critical section
flag with a spin lock.
(_hurd_critical_section_lock): Use spin_try_lock on that to see if we
get it. No need to take SS->lock at all.
(_hurd_critical_section_unlock): Unlock SS->critical_section_lock
instead of clearing the old flag member.
* hurd/hurdsig.c (_hurd_internal_post_signal): Use spin_lock_locked to
test the critical section state.
* hurd/hurdinit.c (_hurd_init): Set _hurdsig_traced from the intarray.
* hurd/hurdkill.c (_hurd_sig_post): Pass 0 sigcode in msg_sig_post.
* hurd/hurdsig.c (_hurd_internal_post_signal): Test _hurdsig_traced
instead of testing (_hurd_exec_flags & EXEC_TRACED).
(_S_msg_sig_post): Take sigcode arg and pass it through.
(_S_msg_sig_post_untraced): Likewise.
(reauth_proc): Don't pass proc port in auth_user_authenticate.
* hurd/setauth.c (_hurd_setauth): Don't pass object ports in
auth_user_authenticate RPCs, just the one-off rendezvous port.
* hurd/dtable.c (reauth_dtable): Likewise.
* hurd/hurdlookup.c (__hurd_file_name_lookup_retry): Likewise.
* hurd/hurdexec.c (_hurd_exec): Pass 0 flags to file_exec.
Pass sigcode arg to msg_sig_post.
* string/argz.h (argz_create): Use const in prototype.
* hurd/hurdinit.c (_hurd_proc_init): Test _hurdsig_traced instead of
testing (_hurd_exec_flags & EXEC_TRACED).
Pass sigcode arg to msg_sig_post.
* hurd/hurd.h: Declare _hurdsig_traced.
* string/argz.h (__argz_next): Cast ENTRY before returning it.
* hurd/hurd/signal.h (_hurd_critical_section_unlock): Pass sigcode arg
to msg_sig_post.
* hurd/path-lookup.c: New file.
* hurd/Makefile (routines): Add path-lookup.
* hurd/hurd/lookup.h: Declare file_name_path_scan,
hurd_file_name_path_lookup.
* hurd/hurd.h: Declare file_name_path_lookup.
* sysdeps/mach/hurd/select.c: The io_select rpc no longer has a TAG_ID
argument. Instead, use a separate reply port for each RPC and put them
all in a port set to wait for slow replies.
* hurd/hurdmsg.c (_S_msg_report_wait): Function removed.
* hurd/report-wait.c: New file.
* hurd/Makefile (routines): Added report-wait.
* sysdeps/mach/hurd/Makefile (inhibit-unix-syscalls): New variable.
* sysdeps/mach/hurd/i386/intr-msg.h (struct mach_msg_trap_args):
New type.
(SYSCALL_EXAMINE, MSG_EXAMINE): New inline functions.
* sysdeps/mach/hurd/i386/trampoline.c (struct mach_msg_trap_args):
Structure moved to intr-msg.h; include that.
* time/Makefile (CFLAGS-ialloc.c): Add -Wno-unused.
(CFLAGS-scheck.c): New variable.
* sysdeps/mach/hurd/dl-sysdep.c (__hurd_sigthread_stack_base,
__hurd_sigthread_stack_end, __hurd_sigthread_variables,
__hurd_threadvar_stack_mask): New variables.
* sysdeps/mach/hurd/fork.c: Set the new task's exception port to
its new message port.
* misc/init-misc.c: Put __init_misc in the __libc_subinit set.
* configure.in (uname): Add quoting.
* sysdeps/mach/hurd/fchdir.c: Don't consult errno unless lookup fails.
* hurd/fchroot.c: Likewise.
* posix/sys/types.h [GCC >= 2.7]: Define intN_t/u_intN_t using
__attribute__ ((__mode__ (__XX__))).
Wed Feb 7 03:24:05 1996 Torbjorn Granlund <tege@tmg.se>
* sysdeps/i386/i586/memcopy.h (WORD_COPY_FWD): Manually allocate
destination cache lines.
(WORD_COPY_BWD): Likewise.
Wed Feb 7 14:16:36 1996 Roland McGrath <roland@churchy.gnu.ai.mit.edu>
* sysdeps/generic/Makefile (make_siglist): Get signum.h absolute
file name first thing, before $(native-compile) changes directories.
From Gord Matzigkeit <gord@enci.ucalgary.ca>:
* sysdeps/unix/make_errlist.c: Use strerror if available.
* sysdeps/unix/configure.in: New file
* sysdeps/unix/bsd/sun/syscalls.list: Added sigvec.
* sysdeps/unix/bsd/sun/sigvec.S: File removed.
* sysdeps/stub/sendto.c: Make sockaddr arg pointer to const.
* sysdeps/stub/connect.c: Likewise.
* sysdeps/stub/bind.c: Likewise.
* hurd/hurdlookup.c (__hurd_file_name_lookup_retry): RETRYNAME of
"" is only special for FS_RETRY_NORMAL; for FS_RETRY_REAUTH, do
another dir_lookup of "".
