Re: [patch 02/11] x86 architecture implementation of Hardware Breakpoint interfaces

[Ingo Molnar <mingo@elte.hu>]

* Alan Stern <stern@rowland.harvard.edu> wrote:

> On Sat, 14 Mar 2009, K.Prasad wrote:
> 
> > Here's a summary of the intended changes to the patchset, which I hope
> > to post early the following week. It tears down many features in the
> > present submission (The write-up below is done without the benefit of
> > actually having run into limitations while trying to chisel out code).
> > 
> > - Adopt a static allocation method for registers, say FCFS (and perhaps
> >   botton-up for user-space allocations and the reverse for
> >   kernel-space), although individual counters to do book-keeping should also
> >   suffice.
> 
> You can't enforce bottom-up allocation for userspace breakpoint
> requests. [...]

That's not the point.

The point is to offer a reasonable and simple static allocator 
that will work fine with usual gdb usage. If something takes 
away db4 that's as if user-space took away all registers - tough 
luck.

You are trying to put complexity into a situation that is not 
schedulable hence not resolvable _anyway_. There's just 4 debug 
registers, not more. If the combined usage goes above four 
someone will lose anyway - even with your allocator.

With my proposal the 'loss' can indeed come sooner if user-space 
took db4 and there's nothing left for the kernel anymore - but 
that's just an uninteresting special case that wont occur with 
typical debug-register usage.

If it ever causes problems seriously _then_ will be the time to 
consider "is it worth adding a more complex, dynamic allocator 
for debug registers". Not now. This stuff is currently 
over-designed and not acceptable to me in its current form.

	Ingo