Cherrypick from master 2004-11-22 22:13:35 UTC Jennifer Averett <Jennifer.Averett@OARcorp.com> '2004-11-22 Jennifer Averett <jennifer@OARcorp.com>':
c/src/lib/libbsp/powerpc/psim/irq/irq.c
c/src/lib/libbsp/powerpc/psim/irq/irq.h
c/src/lib/libbsp/powerpc/psim/irq/irq_asm.S
c/src/lib/libbsp/powerpc/psim/irq/irq_init.c
* vmeUniverse/vmeUniverse.c: Use PCI_MEM_BASE_ADJUSTMENT because some
boards -- notably the MVME2100 -- return the physical address, not an
offset in PCI memory space.
* configure.ac, mpc6xx/exceptions/raw_exception.c,
mpc6xx/exceptions/raw_exception.h, mpc6xx/mmu/bat.c,
mpc6xx/mmu/bat.h, mpc6xx/mmu/mmuAsm.S, shared/include/cpuIdent.c,
shared/include/cpuIdent.h: Add MPC8240 and MPC8245 support. There was
also a significant amount of spelling and whitespace cleanup.
Cherrypick from master 2004-10-11 20:13:31 UTC Eric Norum <WENorum@lbl.gov> 'Don't get duplication rtc_probe() definitions.':
c/src/lib/libbsp/i386/pc386/clock/todcfg.c
c/src/libchip/rtc/README.mc146818a
c/src/libchip/rtc/mc146818a.c
c/src/libchip/rtc/mc146818a.h
c/src/libchip/rtc/mc146818a_ioreg.c
PR 606/bsps
* bootloader/pci.c: Fixed IO remapping so buses >= 1 are remapped.
Reduced PCI space to match bat2. Fixed incorrect region size
calculation in pci_read_bases. Set PCI latency timers to known
sane values. Changed bridge PCI settings to minimum sane instead
of whatever sounded neat in the PCI spec. Force pf regions to
memory mapped to preserve byte access.
PR 619/bsps
* mpc6xx/clock/c_clock.c: The PPC decrementer must be reloaded on each
clock tick. Currently, this is done by just reloading a fixed value.
The attached patch takes into account the time that elapsed since the
decrementer crossed zero in order to adjust the value to be
re-loaded. Without the patch, the effective system clock cycle is
increased by the exception handler latency.
* PR 624/bsps
startup/bspstart.c: When probing the memory, make sure not to
corrupt the application. Only probe addresses beyond rtemsFreeMemStart.
* Let IBMPC_outch() convert LF -> CRLF to avoid printk scrambling
the screen.
* PR/625/networking
* network/dec21140.c: Prevent name clashes by making 'ld_le32()' etc.
static inlines. Let dec21140 attach routine return an error rather
than panic if no chip is detected (thus allowing for probing).
* network/elnk.c: Let elnk bail out if autoneg never completes
instead of looping forever. Avoid divide by zero (crashed my PC).
* network/if_fxp.c: Enable more fxp chip variants but warn that
they are UNTESTED.
PR 595/tests
* sp13/fillbuff.c, sp13/init.c, sp13/putbuff.c, sp13/system.h,
sp13/task1.c, sp13/task2.c: Fill_buffer() in fillbuff.c overruns the
supplied buffer due to confusion in the definition of MESSAGE_SIZE.
Some files in the sp13 test treat MESSAGE_SIZE as a count of bytes,
others treat it as a count of longs.
PR 596/tests
* sp32/init.c: A missing definition of TEST_INIT before including
<tmacros.h> means that buffered output will not work even if selected
in buffer_test_io.h.
