Update TODO list

TODO

@@ -9,7 +9,8 @@ issues related to each board port.
 
 nuttx/:
 
- (15)  Task/Scheduler (sched/)
+ (13)  Task/Scheduler (sched/)
+  (3)  SMP
   (1)  Memory Management (mm/)
   (1)  Power Management (drivers/pm)
   (3)  Signals (sched/signal, arch/)
@@ -255,7 +256,7 @@ o Task/Scheduler (sched/)
                a holder of the semaphore.  Normally, a task is removed from
                the holder list when it finally releases the semaphore via
                sem_post().
-              
+
                However, TASK A never calls sem_post(sem) so it becomes
                *permanently* a holder of the semaphore and may have its
                priority boosted at any time when any other task tries to
@@ -304,7 +305,9 @@ o Task/Scheduler (sched/)
   Priority:    Low.  Things are just the way that we want them for the way
                that NuttX is used today.
 
-  Title:       SMP ISSUES
+o SMP
+  ^^^
+  Title:       SMP ISSUES WITH task_restart() AND task_delete()
   Description: The interface task_restart() (and probably task_delete()) are
                not usable in the SMP configuration in the current design.  In
                the non-SMP case, these are relatively simple:  If the task is
@@ -330,9 +333,39 @@ o Task/Scheduler (sched/)
                The task_restart() test is also disabled in apps/examples/ostest
                in this configuration.  task_delete(), on the other hand, is
                built (but probably should not be).
+
   Status:      Open
   Priority:    Low.  I do not plan to do anything with this in the near future.
 
+  Title:       SMP AND SIGNAL ACTIONS
+  Description: Suppose a task task is running on CPU1 and it is signaled from
+               another task running on CPU0.  How does the signal handler run
+               on CPU1?  I think it does not; I think that the signal will be
+               lost.  I think all testing up to this point has used a task
+               waiting for a signal vs. a running task receiving a signal.
+               This case has never been tested, but I suspect an issues.
+               Here is why... this is the signal handling sequence:
+
+               - sigueue() will set up the siginfo data structure and call
+                 sig_dipatch().
+               - sig_tcbdispach() or group_signal() depending on the
+                 configuration.  Let's assume the simpler sig_tcbdispatch().
+               - sig_tcbdispatch() will call queue the signal action (via
+                 sig_queueaction()) and then call the architecture-specific
+                 up_schedule_signaction set up the invoke the signal handler
+                 (for example in arch/arm/src/armv7-m/up_schedulesigaction.c).
+               - sig_queueaction() will assume that the other task is not
+                 running and will simply modify data structures in the TCB.
+                 This, will have no effect if the task is running and the
+                 signal action will not be performed.
+
+               This is really a variant of the problem described above under
+               "SMP ISSUES WITH task_restart() AND task_delete()" and the
+               same proposed solution applies:  Call up_cpu_pause() to stop
+               all other CPUs before up_schedule_signaction runs.
+  Status:      Open
+  Priority:    High.  This must be fixed.
+
   Title:       SPINLOCKS AND DATA CACHES
   Description: If spinlocks are used in a system with a data cache, then there
                may be a problem with cache coherency in some CPU architectures:
@@ -346,12 +379,12 @@ o Task/Scheduler (sched/)
                unless the spinlocks were made to be the same size as one cache
                line.
 
-               The better option is to add compiler independent "ornmentation"
+               The better option is to add compiler independent "ornamentation"
                to the spinlock so that the spinlocks are all linked together
                into a separate, non-cacheable memory regions.  Because of
-               region aligment and minimum region mapping sizes.  This would
-               work in systems that have both data cache and either an MPU or
-               an MMU.
+               region aligment and minimum region mapping sizes this could
+               still be wasteful of memory.  This would work in systems that
+               have both data cache and either an MPU or an MMU.
   Status:      Open
   Priority:    High.  spinlocks, and hence SMP, will not work on such systems
                without this change.
@@ -1360,8 +1393,8 @@ o USB (drivers/usbdev, drivers/usbhost)
   Status:      Open
   Priority:    Medium-Low unless you really need host CDC/ACM support.
 
-o Libraries (libc/)
-  ^^^^^^^^^^^^^^^^^
+o Libraries (libc/, libm/)
+  ^^^^^^^^^^^^^^^^^^^^^^^^
 
   Title:       SIGNED time_t
   Description: The NuttX time_t is type uint32_t. I think this is consistent
@@ -2080,3 +2113,4 @@ o Other Applications & Tests (apps/examples/)
                the artifact is larger.
   Status:      Open
   Priority:    Medium.
+