PR 1359/cpukit
* libcsupport/src/libio.c: rtems_libio_allocate:
rtems_libio_iop_freelist incorrectly zeroed on semaphore error. Now
checks error and does not modify anything until sure it has created
semaphore.
PR 1354/bsps
* network/network.c: In some cases it can occur that an empty mbuf is
put on the descriptor chain. (We found it especially then when
transmitting fragmented IP Packets.) Since the actual buffer
descriptor pointer will be incremented after every inserted mbuf
(txBd = sc->txBdBase + sc->txBdHead;) even if m->m_len of the current
mbuf was zero. This leads to the bug.
* Makefile.am, configure.ac: Add new test to test Watchdog display
helper routines. These reach into the SuperCore.
* spwatchdog/.cvsignore, spwatchdog/Makefile.am, spwatchdog/delay.c,
spwatchdog/init.c, spwatchdog/prtime.c, spwatchdog/spwatchdog.scn,
spwatchdog/system.h, spwatchdog/task1.c: New files.
* configure.ac: Make runtest an autoconf generated file so we can
insert target alias. Also detect more program termination conditions
so tests are killed more reliably.
* runtest.in: New file.
* runtest: Removed.
* libcsupport/src/gxx_wrappers.c, posix/include/mqueue.h,
posix/include/rtems/posix/semaphore.h,
posix/inline/rtems/posix/barrier.inl,
posix/inline/rtems/posix/key.inl,
posix/inline/rtems/posix/mqueue.inl,
posix/inline/rtems/posix/rwlock.inl,
posix/inline/rtems/posix/semaphore.inl,
posix/inline/rtems/posix/spinlock.inl,
posix/inline/rtems/posix/timer.inl, posix/src/condget.c,
posix/src/mqueuenametoid.c, posix/src/mutexget.c,
posix/src/semaphorenametoid.c, posix/src/semopen.c,
sapi/src/itronapi.c, sapi/src/posixapi.c: Make changes necessary for
all tests to run on SPARC with 16-bit Ids. This required ensuring
that all POSIX and compilering binding code makes a distinction
between the public Id type (e.g. pthread_t, etc.) and the RTEMS
Object_Id type. All POSIX Object Get routines should not take the
POSIX Id type as the argument. Sixteen bit RTEMS Ids should be placed
into the 32-bits reserved by the POSIX API type in a uniform manner
now. This removed all assumptions that the external Id types in POSIX
and ITRON are the same as the internal Object Id type.
