2009-11-06 Till Straumann * Makefile.am, irq/discovery_pic.c, irq/irq_init.c: dump raw_exceptions.rel, raw_exception.h; these have disappeared. must now be included instead. * Makefile.am: Add 'altivec.rel'. * make/custom/beatnik.cfg: Use -mcpu=7400; this enables AltiVec! 2009-11-06 Till Straumann * beatnik.cfg, make/custom/beatnik.cfg: moved beatnik.cfg to new make/custom subdir. 2009-10-20 Till Straumann * startup/bspstart.c: leave 'work-space start' and initial stack alone. These are now handled by the shared framework and linker script etc. Locate interrupt stack after __rtems_end and obtain its size from the configuration. 2009-10-20 Till Straumann * network/if_mve/mv643xx_eth.c: made mutex a binary semphore; simple binary semaphore doesn't support priority inheritance. This was silently ignored under previous releases but is an error under 4.10. 2009-10-20 Till Straumann * Makefile.am, bsp_specs, preinstall.am, flash/flashcfg.c, include/bspopts.h.in, irq/discovery_pic.c, irq/irq_init.c, marvell/gt_timer.c, marvell/gt_timer.h, marvel/gti2c.c, network/if_gfe/if_gfe.c, network/if_gfe/if_gfe_rtems.c, network/if_mve/mv643xx_eth.c, network/support/bsp_attach.c, pci/gt_pci_init.c, pci/motload_fixup.c, startup/bspstart.c, startup/i2c_init.c: Ported to rtems HEAD (to become rtems-4.10). This consisted mainly of fixing compiler warnings (mostly: adding prototypes to function declarations and moving extern declarations to global scope). A pecularity: if_gfe.c had to remove 'queue.h' inclusion. we have two versions of queue.h: one in newlib another one in rtems - don't know how this is supposed to work... 2009-10-17 Till Straumann * Makefile.am, network/if_mve_mv643xx_eth.c, network/if_mve_pub.h: Enhanced low-level API allowing the user to selectively enable/disable/acknowledge interrupts and to install their own ISR (rather than having the driver posting an event to a single task). 2009-10-03 Till Straumann * network/if_mve/mv643xx_eth.c: BUGFIX: mbuf leak; consume_rx_mbuf() must release mbuf if 'len'<=0. BUGFIX: Must initialize 'media-word' argument before calling BSP_mve_media_ioctl() (defines PHY instance). 2009-06-05 Till Straumann * network/if_mve/mv643xx_eth.c, network/if_mve/if_mve_pub.h, Makefile.am: Added MC address reference count and BSP_mve_mcast_filter_accept_del() to remove a single entry from the filter. 2009-06-05 Till Straumann * network/if_mve/mv643xx_eth.c, network/if_mve/if_mve_pub.h, Makefile.am: first stab at adding multicast support. 2009-06-05 Till Straumann * network/if_gfe/if_gfe.c: o propagate PROMISC flag to hardware (SIOCSIFFLAGS) o handle case where IFF_PROMISC is set (and wasn't before) in gfe_hash_fill() routine. 2009-06-02 Till Straumann * network/if_gfe/if_gfe.c: activated and fixed multicast support. 2009-06-01 Till Straumann * network/if_em/if_em.c: activated multicast support. 2008-10-30 Till Straumann MERGED from rtems-4-7-branch: * Makefile.am, network/if_mve/mv643xx.c, network/if_mve/if_mve_pub.h: o Exported new low-level driver entry points: - BSP_mve_ack_link_chg() so that changes in PHY link status can be propagated to the serial port when handling link-change interrupts. - BSP_mve_dump_stats() for printing statistics. o FIXED reading of statistics counters. o Count interrupts (statistics) in ISR rather than network daemon (which is only used by the BSD driver). 2008-10-04 Till Straumann * beatnik.cfg: updated to 4.9; removed make-exe make-cxx-exe commands. Replaced CPU cflags to use -mpowerpc -D__ppc_generic. 2008-10-04 Till Straumann * startup/linkcmds: increased size of CODE memory area to 32M. 2008-05-10 Till Straumann * pci/gt_pci_init.c, pci/pci_io_remap.c: fixed 32-bit types. RTEMS' pci_config access functions now use uint32_t, earlier versions used unsigned. Both are incompatible, unfortunately (gcc regards unsigned and unsigned long different beasts leading to warnings and alias-issues :-() 2008-05-10 Till Straumann * network/porting/rtemscompat1.h, network/porting/rtemscompat.h, network/porting/if_xxx_rtems.c, network/if_gfe/if_gfe_rtems.c: Fixed 32-bit types (pci config access, byteorder macros differ depending on RTEMS version :-(). We now check for version and use appropriate types (unsigned vs. uint32_t). Silenced more warnings (ifndef DEBUG_MODULAR the METHODSPTR is always non-zero; hence I ifdef'ed the affected code snippet). 2008-03-20 Till Straumann * include/bsp.h, startup/bspstart.c: confdefs.h now wants us to use BSP_INTERRUPT_STACK_SIZE instead of CONFIGURE_INTERRUPT_STACK_MEMORY. 2008-03-19 Till Straumann * irq/discovery_pic.c: must spare GPP7_0 etc. summary interrupts in BSP_enable_irq_at_pic() etc. New 'new-exceptions/bspsupport' code scans all IRQS and enables or disables depending on the initial config having a handler connected. This initial disable operation switched-off the summaries and I had no GPP interrupts... 2008-01-04 Till Straumann * startup/bspstart.c: changed Kate's copyright note again as requested by her email 1/04/2008. 2008-01-04 Till Straumann * startup/bspstart.c: changed Kate's copyright note as requested by her email 8/16/2007. 2007-12-11 Till Straumann * irq/discovery_pic.c: don't print warnings if an invalid irq number is passed to BSP_disable_irq_at_pic(), BSP_enable_irq_at_pic(). irq_supp.h says we must silently ignore. 2007-12-11 Till Straumann * Makefile.am: use new irq_bspsupport.rel which was split out of exc_bspsupport.rel to provide finer-grained control over what BSPs want to use. 2007-12-10 Till Straumann * Makefile.am, startup/vpd.c, startup/vpd.h: moved VPD support to ../shared/motorola. 2007-12-08 Till Straumann * Makefile.am: merged shared/vme/vme_universe.c and shared/vme/vme_universe_dma.c into one file. Use vme_universe.c, vmeconfig.c from shared area. 2007-11-30 Till Straumann * startup/bspstart: removed _Cpu_table.exceptions_in_RAM. 2007/11/27 (TS): - Generalized flash support and moved to shared area (libchip would probably more appropriate). 2007/10/22 (TS): - DECREMENTER interrupt is now handled the same way external interrupts are. It can also be assigned a priority and the handler is executed in priority order, i.e., it can be preempted by higher-priority interrupts and is protected from being preempted by lower-priority irqs. 2007/10/08 (TS): - ChangeLog added - (Makefile.am) MUST NOT use -msoft-float because this also prevents CR7 to be set/cleared when calling vararg routines (which may then save/restore FP args on the stack or do other bad things) :-( Still don't know how to deal with implicit usage of the FPU by GCC (problem in ISRs and integer-only tasks).