public inbox for ecos-discuss@sourceware.org
 help / color / mirror / Atom feed
* [ECOS] MBX8XX platform questions.
@ 2000-01-25 21:08 Erik Theisen
  2000-01-26  0:41 ` Jesper Skov
  0 siblings, 1 reply; 2+ messages in thread
From: Erik Theisen @ 2000-01-25 21:08 UTC (permalink / raw)
  To: ecos-discuss

Hi,

I am setting up eCos on a MBX860-002B board and have
some questions.

There are 12 MBX8XX variants, 6 821s and 6 860s.
2 of the them are entry level boards.  Of these 12
variants, only the 2 entry level boards have 50MHz
clocks, the other 10 have 40MHz clocks.  I would
guess that most people would want one of the standard
boards, not the entry level ones, since only the
standard ones have PC/104 expansion capabilities.

In light of the above, Why is
CYGHWR_HAL_POWERPC_MBX_BOARD_SPEED defaulted
to 50MHz instead of the more common 40MHz clock.

Also, it appears the 10 standard boards are
configured by "eppcbug" firmware to use the SMsC
FDC37C672 SuperIO chip for COM1 and COM2.  Only the
entry level boards use the 860/821 onboard SMC1 for
COM1. I believe the entry level board lacks the SMsC
chip.  These are the default settings.  eCos seems to
be using the SMC1 for COM1. It doesn't appear that
eCos has a driver for the FDC37C672 chip.

If someone, (me?), was to write a driver for this chip,
how would one go about making the configuration
distinctions between the entry level board and a
normal one?

I've tried to run some tiny eCos applications without
using stubrom.  This seems to hang the board.  I
believe that eCos and "eppcbug" are using the same
vector table.  Why?

I suspect that even without the vector problem, it 
would not work because "eppcbug" is using the SMsC
chip for IO and eCos is using the 8XX SMC1.

Additionally, I would like to be able to interrupt
a process by pressing the abort switch, which should
generate an abort/stop interrupt IRQ7, to drop back
to the debugger.  It doesn't seem to be working and
neither does CTRL-C.  What is required to make this
work?

Another thought... Does eCos support the parity memory
versions of the MBX8XX board?

Thanks for the help,
Erik

^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: [ECOS] MBX8XX platform questions.
  2000-01-25 21:08 [ECOS] MBX8XX platform questions Erik Theisen
@ 2000-01-26  0:41 ` Jesper Skov
  0 siblings, 0 replies; 2+ messages in thread
From: Jesper Skov @ 2000-01-26  0:41 UTC (permalink / raw)
  To: Erik Theisen; +Cc: ecos-discuss

>>>>> "Erik" == Erik Theisen <etheisen@mindspring.com> writes:

Erik> Hi, I am setting up eCos on a MBX860-002B board and have some
Erik> questions.

I skipped some of them. Detail I don't know off-hand. I hope someone
else will fill in.

Erik> There are 12 MBX8XX variants, 6 821s and 6 860s.  2 of the them
Erik> are entry level boards.  Of these 12 variants, only the 2 entry
Erik> level boards have 50MHz clocks, the other 10 have 40MHz clocks.
Erik> I would guess that most people would want one of the standard
Erik> boards, not the entry level ones, since only the standard ones
Erik> have PC/104 expansion capabilities.

Erik> In light of the above, Why is CYGHWR_HAL_POWERPC_MBX_BOARD_SPEED
Erik> defaulted to 50MHz instead of the more common 40MHz clock.

Simply because we only have 3 boards or so. Only one of those, as far
as I know, is 50Mhz. So from our POV 40MHz is more common :)

I'm not sure there's much point though. Even though there exists
various board configurations, with some of them sharing features, it
does not necessarily mean the majority of the sold boards are of one
type or the other.  

With multiple incompatible board configuration, a certain amount of
users would have to change options from the shipped default
configuration of eCos - we couldn't make everybody happy with only one
default configuration.  And every user is bound to have to change some
options anyway, so why not one more?

If we changed the default to 50Mhz now and 5 people complained, should
we change it back to 40MHz? :)   In any event, it's not much work to
change the default. You could provide the patch and if people don't
complain, it would probably be accepted.

Erik> Also, it appears the 10 standard boards are configured by
Erik> "eppcbug" firmware to use the SMsC FDC37C672 SuperIO chip for
Erik> COM1 and COM2.  Only the entry level boards use the 860/821
Erik> onboard SMC1 for COM1. I believe the entry level board lacks the
Erik> SMsC chip.  These are the default settings.  eCos seems to be
Erik> using the SMC1 for COM1. It doesn't appear that eCos has a
Erik> driver for the FDC37C672 chip.

Again, due to our board selection. We only have the poor man's
version of the board.

Erik> If someone, (me?), was to write a driver for this chip, how
Erik> would one go about making the configuration distinctions between
Erik> the entry level board and a normal one?

Uh, I'm not sure I understand the question.

Basically, the driver would be made available, and either the existing
HAL diag driver or the new one would be active default. Some users
would have to change the configuration to get the proper driver
included in the kernel.

As for "proper" serial drivers (io/serial) they already have to be
enabled explicitly by the user before they can be used. The user must
select the correct driver.

Bottom line; it will be a question of user education. Document the
issue properly and there should be few problems if any.

Erik> Additionally, I would like to be able to interrupt a process by
Erik> pressing the abort switch, which should generate an abort/stop
Erik> interrupt IRQ7, to drop back to the debugger.  It doesn't seem
Erik> to be working and neither does CTRL-C.  What is required to make
Erik> this work?

Abort switch? Ah, that would be on the same strip of jumper pins as
the reset switch, right?  Sadly we don't have anything connected to
that strip of jumpers, and so have not added any support for the
functionality there.

The Control-C support should work if enabled though. To get that
working for IRQ7, you would have to make an interrupt handler which
stores a breakpoint on the PC at which the interrupt occurred and then
return.  See how to do it in the Control-C handling code.

Erik> Another thought... Does eCos support the parity memory versions
Erik> of the MBX8XX board?

No, I don't think so.

Jesper

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2000-01-26  0:41 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2000-01-25 21:08 [ECOS] MBX8XX platform questions Erik Theisen
2000-01-26  0:41 ` Jesper Skov

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).