PRecord and hardware watchpoint

PRecord and hardware watchpoint

[Marc Khouzam]

Hi,

someone asked me if watchpoints worked with Reverse debugging (PRecord),
so I gave it a try.  What I found is that software watchpoints work but
not hardware watchpoints.  That sounds reasonable to me.

What seemed strange was that hardware watchpoints did not work
even when doing normal forward execution with PRecord enabled.
I guess during replay it makes sense, but during record?

Just wondering.

Thanks

Marc

P.S. It would be nice to have some kinds of warning to tell user
not to set hw watchpoints when PRecord is being used.

Re: PRecord and hardware watchpoint

[Pedro Alves]

On Wednesday 17 February 2010 17:41:00, Marc Khouzam wrote:
> Hi,
> 
> someone asked me if watchpoints worked with Reverse debugging (PRecord),
> so I gave it a try.  What I found is that software watchpoints work but
> not hardware watchpoints.  That sounds reasonable to me.
> 
> What seemed strange was that hardware watchpoints did not work
> even when doing normal forward execution with PRecord enabled.
> I guess during replay it makes sense, but during record?
> 
> Just wondering.
> 

Should work since:

2009-11-22  Pedro Alves  <pedro@codesourcery.com>

        Make hardware breakpoints work for process repord.

        * record.c (record_wait): Only adjust PC on software breakpoints
        hits.

2009-11-22  Pedro Alves  <pedro@codesourcery.com>
            Michael Snyder  <msnyder@vmware.com>

        Make hardware watchpoints work for process record.
(...)

That's post 7.0 though.

> Thanks
> 
> Marc
> 
> P.S. It would be nice to have some kinds of warning to tell user
> not to set hw watchpoints when PRecord is being used.
> 


-- 
Pedro Alves

Re: PRecord and hardware watchpoint

[Michael Snyder]

Marc Khouzam wrote:
> Hi,
> 
> someone asked me if watchpoints worked with Reverse debugging (PRecord),
> so I gave it a try.  What I found is that software watchpoints work but
> not hardware watchpoints.  That sounds reasonable to me.
> 
> What seemed strange was that hardware watchpoints did not work
> even when doing normal forward execution with PRecord enabled.
> I guess during replay it makes sense, but during record?
> 
> Just wondering.
> 
> Thanks
> 
> Marc
> 
> P.S. It would be nice to have some kinds of warning to tell user
> not to set hw watchpoints when PRecord is being used.

Were you using the release or CVS version?
CVS head should work.

RE: PRecord and hardware watchpoint

[Marc Khouzam]

I got the latest HEAD and gave it a go.
Very cool!

Thanks both 

> -----Original Message-----
> From: gdb-owner@sourceware.org 
> [mailto:gdb-owner@sourceware.org] On Behalf Of Michael Snyder
> Sent: Thursday, February 18, 2010 1:16 PM
> To: Marc Khouzam
> Cc: 'gdb@sourceware.org'
> Subject: Re: PRecord and hardware watchpoint
> 
> Marc Khouzam wrote:
> > Hi,
> > 
> > someone asked me if watchpoints worked with Reverse 
> debugging (PRecord),
> > so I gave it a try.  What I found is that software 
> watchpoints work but
> > not hardware watchpoints.  That sounds reasonable to me.
> > 
> > What seemed strange was that hardware watchpoints did not work
> > even when doing normal forward execution with PRecord enabled.
> > I guess during replay it makes sense, but during record?
> > 
> > Just wondering.
> > 
> > Thanks
> > 
> > Marc
> > 
> > P.S. It would be nice to have some kinds of warning to tell user
> > not to set hw watchpoints when PRecord is being used.
> 
> Were you using the release or CVS version?
> CVS head should work.
>