[parisc-linux] Re: Any good rules of pratice about branching insn into the delay slot of another one?

[Michael S. Zick]

On Sun April 2 2006 19:01, Randolph Chung wrote:
RC> > mmm, don't know how to set a break at this point (how should I determine 
RC> > the address of the break point?)
RC> 
RC> just break at the beginning of clone and step through it one insn at a 
RC> time. or, break in main, then do 'print clone' to get the address of clone.
RC>
 
Joel,

A best practice for using a branch instruction within the delay slot
of another branch instruction?

I recall reading a very long, involved description of that in one of
the HP documents.  

Still looking for _which_ document; only saving grace is it is on-line, 
since those are the only HP docs I have read.
Also, that it is one of the older documents written when pa-risc was young.

My personal summary of what I recall:
When one or both branch instructions are conditional, then it gets confusing.

So, for now, in hand-coded assembly, just leave those nop instructions
in the delay slot or fill the delay slot with a non-branch, non-conditional
instruction.

I will send you the doc reference when I find it.
If lucky, somebody on the list will share the pointer before I find it.

Mike