[Bug libgomp/27254] FAIL: libgomp.fortran/reduction6.f90

["danglin at gcc dot gnu dot org" <gcc-bugzilla at gcc dot gnu dot org>]

------- Comment #1 from danglin at gcc dot gnu dot org  2006-06-12 00:53 -------
In the first call to st_write_done, I see:

(gdb) p *((st_parameter_dt *)$r26)->u.p.current_unit
$38 = {unit_number = -1, s = 0x193248, left = 0x0, right = 0x0, priority = 0,
  read_bad = 1, current_record = 1, endfile = NO_ENDFILE, mode = WRITING,
  flags = {access = ACCESS_SEQUENTIAL, action = ACTION_READWRITE,
  blank = BLANK_NULL, delim = DELIM_NONE, form = FORM_FORMATTED,
  is_notpadded = 0, position = POSITION_ASIS, status = STATUS_UNSPECIFIED,
  pad = PAD_YES, convert = CONVERT_NATIVE}, recl = 36, last_record = 0,
  maxrec = 0, bytes_left = 0, lock = {__m_reserved = 0, __m_count = 0,
  __m_owner = 0x0, __m_kind = 0, __m_lock = {__spinlock = {lock = {0, 0, 0,
  0}}, __status = 0}}, waiting = 0, closed = 0, ls = 0x0, rank = 0,
  file_len = 0, file = 0x0}

__m_lock isn't initialized for this unit.  The only unit numbers that
I see being initialized are 5, 6 and 0.  PA-RISC linux currently has a
rather unusual initialization for locks, {1, 1, 1, 1}.  The timeout
occurs because the lock is initially locked and __pthread_acquire spins
forever.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27254