This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: hbreak.exp: Test hardware breakpoints
On Fri, 21 Sep 2007, Daniel Jacobowitz wrote:
> > 2007-09-12 Maciej W. Rozycki <macro@mips.com>
> >
> > * gdb.base/hbreak.exp: New test for "hbreak" and "thbreak".
> >
> > OK to apply?
>
> No, sorry. Compare this to the current break.exp - you'll see
> that it has some cruft that was cleaned out of that testcase, like
> gdb_suppress_entire_file.
That's doable -- I have missed these changes were done somehow, sorry.
I will apply them and resubmit.
> Also, when I try it using gdbserver it fails a lot of tests. This
> is at least partially related to the previous problem (gdb_run_cmd
> was added). I wanted to see what happened on a target that only
> failed to insert hardware breakpoints at continue.
I do not have such a target, so I do not have means to test.
Conservatively, I used one breakpoint only, so it should not normally
happen as there will be at most a single such breakpoint enabled at any
given time. If there is a target for which it is not enough of an
assumption indeed, I can see what can be done, though the maintainer of
the target is welcome to provide some support with that.
I have a MIPS target down the queue for which I have not submitted
support yet, using EJTAG, but with that the smallest number of hardware
breakpoints I have encountered in a given CPU was eight (and the i386 in
its regular mode of operation has four). It was this target I created
this test script in the first place and, as you may expect, it passes all
the tests included (though changes to gdb itself were required; these are
pending too).
Maciej