Hi,

Some months ago Paul Gilliam submitted a set of three patches to fix a single 
stepping problem through atomic instruction sequences on a POWER 
architecture. The problem caused GDB to hang when one tried to single step 
through these instructions. Suppose that you have the following block of 
instructions:

        L1:   lwarx   r11,0,r3
              cmpw    r11,r9
              bne-    L2
              stwcx.  r0,0,r3
              bne-    L1
        L2:   isync

It's not possible to do a single step through each one of these instructions 
since the 'reserve' made by the lwarx instruction will always be lost by the
the time the stwcx instruction has executed.

The patch treats the whole L1/L2 interval as a single instruction, avoiding a 
single step through each one of the instructions contained in the L1/L2 
interval, which would lead GDB to a hang

More details regarding this patch can be found at the original messages.
 
* Patches 1 and 2: http://sourceware.org/ml/gdb-patches/2006-06/msg00339.html
* Patch 3: http://sourceware.org/ml/gdb-patches/2006-06/msg00341.html

I'm re-submitting this patch in order for it to be reviewed. Could this patch 
be applied?

Regards,
Luis