PR 850/rtems
* src/watchdogtickle.c: A Watchdog (used to timeout an event) with
a delay of 1 sometimes does not seem to timeout. The problem
occurs, because for whatever reason when the watchdog tickle function
executes, the watchdog->delta_interval is 0. it is then decremented
before being tested, becomes huge and so doesnt time out. It is
thought there is a race condition where the watchdog->delta_interval
is calculated by reference to a head (also with a delay of 1). But
before it can be added after the head, the head is removed, so the
new head now has a delay of 0.
PR 719/bsps
* m68kpretaskinghook.c: The optimizer on gcc 3.4.2 assumes that the
address of a variable cannot be 0, so it optimizes out any such
checks. this breaks the shared m68k bsp_pretasking_hook() which uses
such a check to determine heap size during runtime.
PR 843/rtems_misc
* rtems_webserver/socket.c: There is a memory corruption problem in the
selectSocket function in the web server, this only become apparent if
you disable the built in memory allocation functions. I have disabled
the internal memory management to get file upload working.
PR 866/rtems
* include/rtems/system.h, include/rtems/score/isr.h,
inline/rtems/score/thread.inl, macros/rtems/score/thread.inl: Added
memory barriers to enter and exit of dispatching and interrupt
critical sections so GCC will not optimize and reorder code out of a
critical section.
PR 890/networking
* rtems_webserver/webs.c: The webservers enters an infinite loop when a
POST request with less data than indicated in the Content-Length
header is received. It also consumes additional heap memory and a
file descriptor for each invalid POST.
PR 845/bsps
* startup/bspstart.c, start/start.S: setup BATs prior to attempting any
device access - missing BAT mappings were probably the reason for the
MMU malfunction assumption. Enabled MSR_DR.
PR 837/bsps
make sure exceptions_in_RAM flag is set prior to initializing exception
handling. Keep MCP disabled for memory probes :-( (the raven would
otherwise fault on PCI config space access to empty slots).
PR832/bsps
* startup/ldsegs.S: move CPU segment descriptor tables from .text to
.data; the CPU modifies the descriptor tables and this could be a
problem: a) if text segment is ever write-protected b) rtems-gdb-stub
checksums the text segment against the file gdb is using. A checksum
mismatch results because of the CPU touching the accessed bit...
PR 833/bsps
* irq/irq_asm.S: Currently, all (new exception) BSPs explicitely enable
the FPU across the user ISR but DONT save/restore the FPU context.
Any use of the FPU fron the user handler (e.g., due to GCC
optimizations) result in corruption. The fix results in an exception
in such cases (user ISR must explicitely save/enable/restore FPU).
Edvin Catovic <edvin@gaisler.com>
Konrad Eisele <konrad@gaisler.com>
PR 827/bsps
* aclocal/bsp-alias.m4, aclocal/check-bsps.m4: Portion of large update
of SPARC BSPs. Includes addition of sis, leon2 and leon3 BSPs,
deletion of leon BSP, addition of SMC91111 NIC driver and much more.
Edvin Catovic <edvin@gaisler.com>
Konrad Eisele <konrad@gaisler.com>
PR 827/bsps
* ChangeLog, cpu.c, cpu_asm.S, rtems/score/cpu.h: Portion of large
update of SPARC BSPs. Includes addition of sis, leon2 and leon3 BSPs,
deletion of leon BSP, addition of SMC91111 NIC driver and much more.
Edvin Catovic <edvin@gaisler.com>
Konrad Eisele <konrad@gaisler.com>
PR 827/bsps
* network/Makefile.am, network/open_eth.c, network/open_eth.h: Portion
of large update of SPARC BSPs. Includes addition of sis, leon2 and
leon3 BSPs, deletion of leon BSP, addition of SMC91111 NIC driver and
much more.
* network/README.smc91111, network/smc91111.c, network/smc91111.h,
network/smc91111config.h, network/smc91111exp.h: New files.
Edvin Catovic <edvin@gaisler.com>
Konrad Eisele <konrad@gaisler.com>
PR 827/bsps
* ChangeLog, configure.ac, console/Makefile.am, console/console.c,
console/debugputs.c, startup/Makefile.am, startup/linkcmds,
tools/Makefile.am: Portion of large update of SPARC BSPs. Includes
addition of sis, leon2 and leon3 BSPs, deletion of leon BSP, addition
of SMC91111 NIC driver and much more.
Edvin Catovic <edvin@gaisler.com>
Konrad Eisele <konrad@gaisler.com>
PR 827/bsps
* Makefile.am, bspstart.c, gnatcommon.c, start.S: Portion of large
update of SPARC BSPs. Includes addition of sis, leon2 and leon3 BSPs,
deletion of leon BSP, addition of SMC91111 NIC driver and much more.
Edvin Catovic <edvin@gaisler.com>
Konrad Eisele <konrad@gaisler.com>
PR 827/bsps
* custom/Makefile.am, custom/erc32.cfg, custom/erc32nfp.cfg,
custom/leon1.cfg, custom/leon2.cfg: Portion of large update of SPARC
BSPs. Includes addition of sis, leon2 and leon3 BSPs, deletion of
leon BSP, addition of SMC91111 NIC driver and much more.
* custom/sis.cfg: New file.
* timer/timer.c: Enhance to use either interupt-based timer
functions on older CPUs or to use TSC-based timer functions on
more recent (Pentium and above) CPUs. The decision is made in
Timer_initialize function when it is called for the first time
based on a result obtained from cpuid instruction during the BSP
initialization phase. During the first call, there are also late
bindings to the implementation functions initialized to
appropriate values.
PR 527/bsps
PR 822/bsps
* mpc8xx/clock/clock.c: Currently the MBX8xx BSP does not boot,
because some logical errors are in the startup code. Additionally,
the mpc8xx shared clock driver does not support the clocking scheme
of some of the board variants, which are clocked from a
32768Hz (!) external crystal.
PR 829/rtems
* src/tasks.c, src/taskvariabledelete.c: If task variables are deleted
from a different context (i.e., executing context != owner of the
task variable. The owner meaning the task that registered the dtor in
question) the argument passed to the task variable dtor must be tvar
and not *ptr which yields the executing task's value of the task
variable instead of the owner's.
