This is the mail archive of the
mailing list for the Archer project.
Re: ptrace improvement ideas
On 02/07, Roland McGrath wrote:
> > There is already AFAIK some abstraction of DR regiters inside kernel so maybe
> > userland could get access to this abstraction to resolve these two issues.
> Indeed, there is now a layer called hw_breakpoint, with in-kernel APIs that
> are largely machine-independent. The legacy arch-specific ptrace interface
> for x86 is implemented on top of that (not purely so, but rather supported
> by that infrastructure as special cases). I think Oleg knows the details
> of that stuff better than I do at this point.
Yes, and hw_breakpoint is implemented on top of perf counters.
> I think the desireable approach is to figure out a new interface (ptrace
> extension, presumably) to use those new facilities directly from user
> space. It should be possible for such a new interface to be largely
> machine-independent too.
Probably yes... but it is not clear to me how exactly the new interface
should look. And it should coexist with the current ->ptrace_bps code.
And while register_user_hw_breakpoint() itself is arch independent, there
are still some details. Say, if the tracee traps, the tracer should know
somehow which bp was the reason.
But I agree it would be nice to have the simple "abstract" interface.
Jan, if you have any idea about how it should look - please tell us.
I _think_ that the actual implementation shoudn't be very very difficult,
although I can't say I understand this code in details.
> > It must explicitly require debug registers (hw watchpoints) inheritance.
> > Which happened before but it no longer happens in recent upstream kernels
> > (NOTABUG RH#660003).
> That is a good point and one I had not been thinking of,
Oh, yes, I missed this too.
> It's my feeling that the
> right way to approach this is to focus on a new set of interfaces built on
> the kernel's hw_breakpoint facility (that infrastructure may well need
> extending to deal well with userland better). Those can be defined with
> the inheritance and process-wide sharing issues in mind,
Hmm. This is also not clear to me...