Re: [RFC] Systemtap translator support for hardware breakpoints on

[Roland McGrath <roland at redhat dot com>]

> Yes, we put the Data Address Register (DAR) to use here and know the
> exact address which was accessed to cause the breakpoint exception. The
> value of DAR along with the length of the symbol (taken as input from
> the user when registering a breakpoint request) is used to filter out
> extraneous breakpoint exceptions.

Ah!  Excellent.  I didn't know the hardware made that possible.  That's
something the x86 hardware can't do, AFAICT.  Of course, it does give you 4
breakpoints to use vs 1, which is a pretty good compensation. ;-)

I guess technically this means that the hw_breakpoint layer could let you
register a second one if it happens to be a length 4 and the two are
adjacent (or up to 8 adjacent one-byte ones, etc.--i.e. the aligned address
of all is the same).  But it probably makes more sense just to have much
higher layers like the stap translator noticing that it can consolidate
adjacencies into a single registration before it decides if you are trying
for an unreasonable number of simultaneous registrations.


Thanks,
Roland