Re: [patch 00/11] Hardware Breakpoint interfaces

From: Ingo Molnar <mingo@elte.hu>
To: "K.Prasad" <prasad@linux.vnet.ibm.com>
Cc: Alan Stern <stern@rowland.harvard.edu>,
	Andrew Morton <akpm@linux-foundation.org>,
	Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
	Roland McGrath <roland@redhat.com>
Subject: Re: [patch 00/11] Hardware Breakpoint interfaces
Date: Wed, 11 Mar 2009 18:30:44 +0100	[thread overview]
Message-ID: <20090311173044.GA9720@elte.hu> (raw)
In-Reply-To: <20090311172506.GC9547@in.ibm.com>

* K.Prasad <prasad@linux.vnet.ibm.com> wrote:

> On Wed, Mar 11, 2009 at 12:34:43PM -0400, Alan Stern wrote:
> > On Wed, 11 Mar 2009, K.Prasad wrote:
> > 
> > > The hardware breakpoint interfaces haven't been put under any CONFIG_
> > > till now, but I think we should bring them under a new config, say
> > > CONFIG_HW_BREAKPOINT. It would help create a dependancy for
> > > CONFIG_KSYM_TRACER too.
> > 
> > With these patches, ptrace is dependent on hw-breakpoint.  
> > You can't disable CONFIG_HW_BREAKPOINT without breaking 
> > ptrace.
> > 
> > Alan Stern
> > 
> 
> Agreed. We might have to retain the old code for ptrace and 
> put the new implementation under #ifdef CONFIG_HW_BREAKPOINT 
> to get them working. What do you think?

With the simple reservation mechanism i suggested i have no 
problem with having HW_BREAKPOINT enabled [selected] 
unconditionally on x86.

Your ptrace changes are an improvement in terms of code quality 
so as long as the facility is simple and obvious, it's a step 
forward.

#ifdefs are ugly and hard to maintain - especially in such a
rarely used and still critical API as ptrace.

	Ingo