> > Looks like the PT_* offsets are in units of long, but otherwise correct?
>
> Nope, we have PT_D7, but no pt_regs.d7.
If that's all ...
> > > I was trying to define task_thread_info() using TASK_INFO instead of using
> > > struct task_struct directly (cfr. ia64), but these conflicts seem to
> > > prevent me
> > > from doing this.
> >
> > At some time in the past (2.6.13??) I had tried to do something similar
> > while debugging the Atari interrupt problems (the asm interrupt entries
> > used wrong offsets to tweak preempt_count() at that time). I gave up in
> > the end... If I understand Linus' suggestion right, he says we should take
> > the task struct from the stack instead of going over current (which we
> > keep in a2, and prefer to use for that reason)?
>
> Yep, that's what he suggests. But we do it differently for `a good reason'.
> Unfortunately I don't remember what good reason it was. Yesterday evening I
Because access to current is faster if we always keep it in a register,
that's what I remember. Saves a couple of instructions on context
switches.
> searched my lkml archives of 2005 and 2006, but apart from Roman and Al
> agreeing we have `a good reason', I couldn't find it.
The decision is much older. Andreas Schwab might remember, as might Roman
Hodek ...
Michael
-
To unsubscribe from this list: send the line "unsubscribe linux-m68k" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
