[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
[
List Home]
|
RE: [dsdp-dd-dev] SPIRIT vs a new standardXMLschema?(wasDebugmodeland GDB?)
|
Hi Doug,
Where we are at:
My original doc on the DD Wiki was raw target description requirements.
It seemed like most folks thought that these requirements more or less
covered what they needed in their debuggers. A few people chimed in a
couple things that they needed that were not in there, and a few others
shared some bit of agreement that it looked good. I am assuming that
the silence from others is tacit approval that what is in this spec
adequately addresses their needs.
What I wrote had nothing to do with format, and as such not directly
SPIRIT related. However, Anthony went through the doc and added
comments on each line item that said what was and was not in SPIRIT
today.
Next steps:
1) document exactly what areas of SPIRIT need to be extended for our
purposes. This should be easy since Anthony already did this in the
requirements doc.
2) draft a spec with our proposed extensions and create working examples
of them.
3) Create tools that use the extensions
4) propose these extensions formally to be incorporated into SPIRIT
Decoupled from the SPIRIT discussion is the bigger picture DSDP
discussion about target descriptions at a higher level. I posted
something to both tm and dd mailing lists on this, but have not seen any
discussion as a result like I had hoped. (also opened up a BugZilla bug
on the need for target descriptions in TM)
Aaron
> -----Original Message-----
> From: dsdp-dd-dev-bounces@xxxxxxxxxxx
> [mailto:dsdp-dd-dev-bounces@xxxxxxxxxxx] On Behalf Of Gaff, Doug
> Sent: Friday, June 16, 2006 11:36 AM
> To: Device Debugging developer discussions
> Subject: RE: [dsdp-dd-dev] SPIRIT vs a new
> standardXMLschema?(wasDebugmodeland GDB?)
>
> Hi guys,
>
> >From Wind's perspective, we read our current data files from our C++
> debugger engine, and we'll likely have to build SPIRIT data
> file translators like everyone else. Eventually, I'd like to
> read those files directly, but I'm sure we'll have to
> translate for some period of time until we can change the
> debug engine.
>
> I don't see this as a big issue right now, though, and I
> think writing some tooling (editing, parsing, rules checking)
> in java seems like a reasonable first step. Anthony, perhaps
> you can provide more detail on how the Tight Generator
> Interface is going to work. Perhaps that will help us down
> the road for the non-java cases.
>
> Anthony, I'm glad to hear that you'll have something to
> contribute soon.
> Perhaps you can put a couple of teaser screen shots in the
> SPIRIT section of the DD Wiki. :) Also, I'd recommend
> contributed to open source sooner rather than striving for
> perfection or completeness prior to contribution. This
> allows others to see what you're doing and might also give
> you some help. (Incidentally, we are struggling with this
> very thing on our debug model contributions.)
>
> Aaron, regarding defining debugger extensions to the SPIRIT
> format, I think I've lost the thread for what you need next
> from folks (sorry).
> Perhaps you can restate that for everyone on this list.
>
> Doug G
>
>
> > -----Original Message-----
> > From: dsdp-dd-dev-bounces@xxxxxxxxxxx [mailto:dsdp-dd-dev-
> > bounces@xxxxxxxxxxx] On Behalf Of Spear, Aaron
> > Sent: Friday, June 16, 2006 11:11 AM
> > To: Device Debugging developer discussions
> > Subject: RE: [dsdp-dd-dev] SPIRIT vs a new standard
> > XMLschema?(wasDebugmodeland GDB?)
> >
> > Hi Martin,
> >
> > Thanks for the insight into TI's usage. I think that this kind of
> > information about how and where people will really use this info is
> > crucial to creating something really useful. Your solution sounds
> > nearly identical to what my company is doing as well in our
> debugger
> > (sounds like we need a standard!).
> >
> > Can you explain a bit more background about the naming convention
> > problems you have seen? Is this perhaps another area where a change
> in
> > the SPIRIT specification/standard is necessary, or is it just
> outlining
> > the need for some kind of possible SPIRIT add-on that might be a
> > database/registration of different components of some kind?
> >
> > cheers,
> > Aaron
> >
> >
> > > -----Original Message-----
> > > From: dsdp-dd-dev-bounces@xxxxxxxxxxx
> > > [mailto:dsdp-dd-dev-bounces@xxxxxxxxxxx] On Behalf Of Swiezawski,
> > > Martin
> > > Sent: Monday, June 12, 2006 8:10 AM
> > > To: Device Debugging developer discussions
> > > Subject: RE: [dsdp-dd-dev] SPIRIT vs a new standard XML
> > > schema?(wasDebugmodeland GDB?)
> > >
> > > Aaron,
> > > I have been following the discussion between you and Daniel.
> > > TI is currently using custom XML format that meets our debuggers'
> > > needs. We have a combination of XSLT & scripts that
> perform initial
> > > translation from either SPIRIT or some other format into
> debugger's
> > > XML format.
> > > However, resulting files need to be tweaked a bit. There
> was couple
> > > of reasons that caused us to move in this direction.
> > > A) HW specifications did not contains CPU native registers B)
> > > non-standard names to identify CPUs were used. Second
> issue is not
> > > that critical for single CPU devices, however it is important for
> > > multi-processor devices, where you need to know whether
> peripheral X
> > > is accessible and visible through CPU
> > > 1 or 2.
> > >
> > > We have organized our debugger XML database to have generic
> > > peripheral files that are then included by device
> specific files. We
> > > are also using a slightly modified version of an XML parser that
> > > allows us to handle couple of tags in a special way. E.g.
> specify in
> > > device file included peripherals' base address. Our backend
> > > debugger(C++) currently does the parsing and provides register
> > > information to clients.
> > >
> > > In my mind, ideal solution would be to use SPIRIT. There
> would be a
> > > standardized component that reads that data and provides
> it through
> > > an API. Folks could then decide where this component needs to be
> > > plugged into. However, a big hurdle that I see is standardized
> > > naming conventions to be able to identify cpus, peripherals,
> > > registers and match this with whatever the debugger is expecting.
> > >
> > > Martin
> > >
> > >
> > > -----Original Message-----
> > > From: dsdp-dd-dev-bounces@xxxxxxxxxxx
> > > [mailto:dsdp-dd-dev-bounces@xxxxxxxxxxx] On Behalf Of Spear, Aaron
> > > Sent: Friday, June 02, 2006 4:13 PM
> > > To: Device Debugging developer discussions
> > > Subject: RE: [dsdp-dd-dev] SPIRIT vs a new standard XML schema?
> > > (wasDebugmodeland GDB?)
> > >
> > > Doug,
> > >
> > > I have posed a bunch of questions recently that perhaps
> are buried
> > > in the discussion, but I think that the most pressing one
> to answer
> > > is perhaps where people see this info being used. The
> reason that I
> > > think this is important to answer now is that it will
> impact what is
> > > useful to have as far as tooling goes. In effect, the
> questions is
> > > whether or not it is profitable for us to write java XML parsing
> > > code specific to SPIRIT or not. I know that you guys
> (Wind) have a
> > > debugger architecture similar to ours in that you have a native
> > > debugger back end that your Eclipse front end
> communicates with. If
> > > you need the info in your back-end, then either you need
> to parse it
> > > in the back end as well as the front end, or you need to
> parse it in
> > > the front end and push it through to the back. (make sense?)
> > >
> > > I do think that regardless of the outcome of the
> discussion above,
> > > we need to start identifying the extensions that are
> needed in the
> > > SPIRIT schema and creating tools that use them.
> > > I think that SPIRIT is the only real hope of getting
> info from the
> > > semiconductor vendors in a format that is useful.
> > > What we then do with that info (use it directly or
> translate it to
> > > another format that we use directly) is another discussion.
> > >
> > > Aaron
> > >
> > > > -----Original Message-----
> > > > From: dsdp-dd-dev-bounces@xxxxxxxxxxx
> > > > [mailto:dsdp-dd-dev-bounces@xxxxxxxxxxx] On Behalf Of Gaff, Doug
> > > > Sent: Friday, June 02, 2006 1:26 PM
> > > > To: Device Debugging developer discussions
> > > > Subject: RE: [dsdp-dd-dev] SPIRIT vs a new standard XML schema?
> > > > (wasDebugmodel and GDB?)
> > > >
> > > > This is somewhat related to my question to Anthony about how to
> > > > proceed.
> > > >
> > > > My opinion: we should continue to define what we think the
> > > SPIRIT for
> > > > debug schema looks like, implement some initial tooling
> > > based on that,
> > > > and then let Anthony and Aaron take it back to the
> Consortium for
> > > > comment. Minor revisions to our ideas should be easy to
> > > adapt to and
> > > > will be necessary until an official SPIRIT release anyway.
> > > >
> > > > Doug
> > > >
> > > > P.S. Welcome to the DD project Daniel. Thanks for joining the
> > > > discussion.
> > > >
> > > > > -----Original Message-----
> > > > > From: dsdp-dd-dev-bounces@xxxxxxxxxxx [mailto:dsdp-dd-dev-
> > > > > bounces@xxxxxxxxxxx] On Behalf Of Daniel Jacobowitz
> > > > > Sent: Friday, June 02, 2006 2:07 PM
> > > > > To: Device Debugging developer discussions
> > > > > Subject: Re: [dsdp-dd-dev] SPIRIT vs a new standard XML
> > > > schema? (was
> > > > > Debugmodel and GDB?)
> > > > >
> > > > > On Fri, Jun 02, 2006 at 11:01:13AM -0700, Spear, Aaron wrote:
> > > > > > I am still hoping to solicit some other opinions on this
> issue,
> > > > perhaps
> > > > > > from folks like ARM Ltd. Who are users of SPIRIT on the hw
> > > > development
> > > > > > side of things with their ESL tools and such, and then have
> been
> > > > talking
> > > > > > about using it for debug as well.
> > > > > >
> > > > > > The question is really should we use SPIRIT directly or a
> > > > decoupled
> > > > more
> > > > > > debugger centric standard?
> > > > >
> > > > > Well asked. I'll await opinions with interest.
> > > > >
> > > > > --
> > > > > Daniel Jacobowitz
> > > > > CodeSourcery
> > > > > _______________________________________________
> > > > > dsdp-dd-dev mailing list
> > > > > dsdp-dd-dev@xxxxxxxxxxx
> > > > > https://dev.eclipse.org/mailman/listinfo/dsdp-dd-dev
> > > > _______________________________________________
> > > > dsdp-dd-dev mailing list
> > > > dsdp-dd-dev@xxxxxxxxxxx
> > > > https://dev.eclipse.org/mailman/listinfo/dsdp-dd-dev
> > > >
> > > _______________________________________________
> > > dsdp-dd-dev mailing list
> > > dsdp-dd-dev@xxxxxxxxxxx
> > > https://dev.eclipse.org/mailman/listinfo/dsdp-dd-dev
> > > _______________________________________________
> > > dsdp-dd-dev mailing list
> > > dsdp-dd-dev@xxxxxxxxxxx
> > > https://dev.eclipse.org/mailman/listinfo/dsdp-dd-dev
> > >
> > _______________________________________________
> > dsdp-dd-dev mailing list
> > dsdp-dd-dev@xxxxxxxxxxx
> > https://dev.eclipse.org/mailman/listinfo/dsdp-dd-dev
> _______________________________________________
> dsdp-dd-dev mailing list
> dsdp-dd-dev@xxxxxxxxxxx
> https://dev.eclipse.org/mailman/listinfo/dsdp-dd-dev
>
