Re: [PATCH, ppc] Fix hw *points for embedded ppc in a threaded environment.

[Luis Gustavo <luis_gustavo@mentor.com>]

On 08/06/2012 05:27 PM, Jan Kratochvil wrote:
> On Mon, 06 Aug 2012 21:02:11 +0200, Luis Gustavo wrote:
>> The ones you listed are server processors, and they use the old
>> kernel interface. Examples of BookE's are the 4xx family (405, 440,
>> 460, 476 etc...)
>
> OK, thanks for info.  I do not think I have those available.
>
>
>> An appropriate test would be to run many threads and create the
>> maximum amount of hw watchpoints and see if that works fine, but
>> GDB's debug resource accounting is still a little awkward
>
> One can test the number of registers in singlethreaded testcase and then run
> the real test with multithreaded program.

Sounds good.

>
> A testcase would be nice, for example now the changes in
> archer-jankratochvil-watchpoint3 and pre-requisite of non-stop compatibility
> of hw watchpoints mean a lot of hw watchpoints rewrite in ppc-linux-nat.c.
>
> But if you are not going to write a testcase then I find it OK for both HEAD
> and 7.5.

I'm ok with coming up with that testcase. Should we go ahead and have 
this in and the test commited shortly afterwards?

Luis