FreshBSD

    Add pthread_suspend_np regress test. okay guenther@

    Make pthread_suspend_np() and pthread_resume_np() simply fail instead
    of doing the Wrong Thing.

    First steps for making ptrace work with rthreads:
     - move the P_TRACED and P_INEXEC flags, and p_oppid, p_ptmask, and
       p_ptstat member from struct proc to struct process
     - sort the PT_* requests into those that take a PID vs those that
       can also take a TID
     - stub in PT_GET_THREAD_FIRST and PT_GET_THREAD_NEXT
    
    ok kettenis@

    - add more ptrace() ops
    - be robust against a ktrace file the contains a record with
      ktr_len==SIZE_MAX, instead of reallocating its buffer to zero size
    - format the clockid_t argument to clock_*() and __thrsleep() as CLOCK_*
    - format the sigset_t argument to sigprocmask() and __thrsigdivert(), the
      return from sigprocmask(), and the mask reported for PSIG records
      as a bitset of SIG* values, except that if most the bits are set
      then invert it and prefix with '~'
    - show the next level of the kern.proc sysctl
    - __tfork() creates procs, so do the mappidtoemul() handling
    - refactor ktrstat()'s time printing bits and fix a whitespace glitch
      in its output
    - reduce stack usage in ktrstruct()
    - a value of zero is not an error for mode bits (S_*), atflag bits
      (AT_*), wait options (W*), or shmat flags (SHM_*)
    
    ok otto@

    Switch the time measuring to be in timespecs, and then check the
    difference against the resolution of the monotonic clock from clock_getres()
    
    discussed with fgsch@

    - remove invalid potions of blocked_* regress tests
    - reenable blocked_* regress tests. while this behaivor
      is not specifited by posix, deadlocking is not a reasonable
      behaivor and some ports depend on this to work.

    Disable the dup2_race test, as it requires a buggy behavior that uthreads
    happened to exhibit

    Round-robin behavior isn't guaranteed by SCHED_OTHER and the kernel
    doesn't have SCHED_FIFO or SCHED_RR support, so use sleep(1) to 'ensure'
    the second thread gets the lock when expected

    Enable the errno test; add a quick rwlocks test

    End the output with a newline

    This test needs a large nprocs ulimit, so just have it raise its own

    Need to block the signals in all thread for the kill(getpid(), SIGFOO)
    tests to work.  Disable the "sigwait for blocked but ignored signal"
    test as it depended on undefined behavior (POSIX quoted included for
    reference) and rthreads (currently) handles it the other way.

    SIGSTOP, SIGKILL, and (for threaded programs) SIGTHR are unblockable

    pthread_sigmask() was fixed to return the errno value on failure

    ops. compare against the socket timeout, not the alarm one.
    sometimes these tests are incorrectly marked as failing (ktrace shows
    the read is returning within the specified timeout) but there are other
    times where they're taking slightly longer than expected.
    for now both cases are marked as failure.

    fix tests.

    Fix a classical bug (= vs ==). OK stsp@ miod@

    small KNF nit

    sync

    Fix return value documentation errors in usbhid(3).
    hid_dispose_report_desc() returns void, it can't return NULL on failure.
    hid_get_report_desc() and hid_use_report_desc() return NULL on failure but
    this was not documented.
    ok jmc

    enable nginx

    these files are not needed here, they are going to get installed from
    usr.sbin/nginx using the distribution target

    add a distribution target to install configuration files

    Allow for more vars in hash table, improve hash function, increase
    hash table size sooner. Based on suggestion from Michael Niedermayer.
    ok krw@ mpi@

    Fix problems revealed by regress/lib/libpthread/stack/stack.c

    Add Id and public-domain declaration

    Add regress test for stack attributes
    Disable some tests that depended on uthread's particular undefined behavior

    Use 'base' as an intermediate char* variable to avoid void* arithmetic

    Extend miod's fix CFLAGS fix to all gcc2 platforms (suggested by Tobias
    Ulmer).  Only pull in bsd.lib.mk once to avoid linking objects twice

    Use a form of designated initializer that works with gcc2
    lint doesn't understand designated initializers, so hide them from it

    Validate in pthread_attr_set{scope,sched{param,policy},inheritsched}()
    the requested new value

    Fix previous commit: _rthread_init() was static.
    Improve consistency of error naming

    Before using _thread_pagesize, make sure _rthread_init() has been called

    id tag.

    Fix the handling of the stackaddr, stacksize, and guardsize attributes:
    don't try to merge values, round the sizes separately, and don't try to
    unmap application-supplied stacks.
    Copy from uthread the caching of default-sized stacks.
    Have pthread_attr_init() and pthread_create() get the default attributes
    from staticly allocated pthread_attr_t.
    Cache the pagesize in _rthread_init() and provide a macro for rounding to it
    
    based on suggestions from kettenis@ and tedu@, ok kettenis@

    Missing cerror bits.

    remove accidentally overwritten change which makes sure to use the bundled
    pcre

    Work around compiler confusion while building rthread.o on vax.

    remove the install target from here since the BSD makefile wrapper
    is taking care of it

    update pcre to 8.30 to be in sync with ports

    update to 1.0.12

    Extend after recent msdos support.
    
    input/ok krw@ thib@ jmc@

    Initialize the scope, policy, and inherit members in pthread_attr_init()
    so that their pthread_attr_get* don't return bogus values.
    
    issue noted and tested by aja@

    Allow librthread to build on mips64.

    Remove the backup chunk from the man page. People should use backup
    anyway and sysmerge(8) automatically save replaced files.
    
    ok sthen@

    don't label "warning messages" as "errors"
    fetch-makefile is obsolete.
    Xr ftp

    add DIAGNOSTICS, for easy grepping thru logs.
    as discussed with jmc@, put the section in exposition order.
    (we're not even a driver anyways)
    sthen@ likes it.

    exit usage() with STATUS_BADUSAGE rather than STATUS_ERROR, per the man page;
    
    fix proposed by lum actually happened in netbsd -r1.16, some 6 years ago;
    ok lum

    Set __isthreaded in pthread_create() instead of _rthread_init() such that it
    properly represents whether a process is multi-threaded or not.  This fixes
    a bug where if a forked child would create a new thread we would not set
    __isthreaded.
    
    ok guenther@

    Do an explicit `sync' instruction before returning from cache routines; this is
    a nop on R5k and RM52xx, but actually required on RM7k.