[Bug 1001897] lpc2xxx CAN driver improvements / enhancements

[bugzilla-daemon@bugs.ecos.sourceware.org]

Please do not reply to this email, use the link below.

http://bugs.ecos.sourceware.org/show_bug.cgi?id=1001897

--- Comment #5 from Uwe Kindler <uwe_kindler@web.de> ---

> SYNOPSIS
> 
>   http://www.nxp.com/documents/application_note/AN10674.pdf | Fig. 14
> 
> May we presume that description from AN is not true for some LPC2294
> revisions? 

It seems so. I only have one single LPC2294 here and it definitely don't works
like written on page 16 of the application note. The RM bit is not set, so
controller dont't go into reset mode and also the TX RX counters are not
changed. I would not have put so much time into this patch if the CAN
controller would have worked the way it should.

> Perhaps, http://www.nxp.com/documents/errata_sheet/2294.pdf
> See CAN.5 (pp. 11,12) "Normal operation cannot be resumed after reset".
> May we presume that they did not raise "Reset Mode" for "Bus Off" error
> as an early "workaround"? Though, I found no such reassurances on Net.

I think CAN.5 has nothing to do with the problem I see - this is something
different.

> 
> On the other hand the driver uses lpc2xxx_enter_reset_mode() in a few
> places without (?) issue. 

Manually entering reset mode via lpc2xxx_enter_reset_mode() is not a problem.
The problem is, that the controller does not automatically enters reset mode in
case of a bus off condition (bus error). lpc2xxx_enter_reset_mode() works fine
and controller properly enters reset mode.

> Uwe, did you noticed any strange behaviors
> after this call? If you did not, then my guess is wrong. And if CAN.5
> issue can be the reason then we can use your workaround by a condition
> is set according some CDL option. Sorry, I cannot help in testing any
> more due lack a CAN adapter.

Ok. Shall I change the patch and make the code conditional via a CDL option?
That would be o.k. for me as long as no one confirmed that this problem exists
for other LPCxxxx devices?

-- 
You are receiving this mail because:
You are on the CC list for the bug.