Three chip erc32 Re: CONFIGURE_APPLICATION_NEEDS_CLOCK_DRIVER macro issue

[Jiri Gaisler]

Hmm, maybe clock driver code has been changed during the years to read
or write some register bits that are illegal on the 3-chip ERC32. This
would cause a trap to error mode and most likely a watchdog reset. I
quickly looked through the ERC32 clock driver code but could not see
anything obvious. It could also be an ERC32 interrupt driver issue -
enabling interrupts seems to trigger the problem. I will try to run
ticker on an old version of sis that did some more checks for the 3-chip
version ...

On 9/3/19 12:57 AM, Joel Sherrill wrote:
> Jiri does this ring a bell with the old Tharsys board and 3 chip erc32?
>
> On Mon, Sep 2, 2019, 5:09 PM Michel Macena <mmacena.eng at gmail.com
> <mailto:mmacena.eng at gmail.com>> wrote:
>
>     It is the ERC 32 chipset version (TSC691, TSC692 and TSC693
>     units), The Board was manufactured by Tharsys, a french company,
>     but it does not
>     exist anymore. The board manual dates from 2000.
>
>     Em sex, 30 de ago de 2019 às 17:53, Joel Sherrill <joel at rtems.org
>     <mailto:joel at rtems.org>> escreveu:
>
>
>
>         On Fri, Aug 30, 2019, 11:35 AM Michel Macena
>         <mmacena.eng at gmail.com <mailto:mmacena.eng at gmail.com>> wrote:
>
>             Thanks for the answer, but the ticker sample program
>             has this macro in his system.h header. I can compile a program
>             with this macro but when  I load it, the board just
>             ignores it an then reset. If I change the
>             macro for the opposite one ("does not need the clock
>             driver") the program just works, except that I can't
>             use any time related routine. Also without the clock drive
>             driver enabled I can't communicate with the board (send
>             and receive data).
>             I understand that the Macro enables the clock drive but
>             how this happens ? It changes a register value  in the chip ?
>
>
>         Setting that macro adds the clock driver to the set of
>         statically installed device drivers. The code is in
>         bsps/sparc/erc32/clock. It uses a timer on the erc32. 
>
>         Check that it survives initialising the clock and gets the
>         interrupt ok.
>
>         This isn't something I have heard of before. Is this a very
>         early erc32? Just wondering with nothing specific in mind. 
>
>         --joel
>
>
>
>             Em sex, 30 de ago de 2019 às 03:21, Sebastian Huber
>             <sebastian.huber at embedded-brains.de
>             <mailto:sebastian.huber at embedded-brains.de>> escreveu:
>
>                 On 30/08/2019 19:09, Michel Macena wrote:
>                 > I still have the issue with the macro #define
>                 > CONFIGURE_APPLICATION_NEEDS_CLOCK_DRIVER. I tried
>                 > a different version of RTEMS but It didn't work. Can
>                 someone explain me
>                 > how this macro works? So I can
>                 > try to figure out some patch or fix for my hardware.
>
>                 This configuration option enables the clock driver.
>
>                 I would run first the RTEMS test suite on your target.
>                 For example start
>                 with the ticker sample program.
>
>                 -- 
>                 Sebastian Huber, embedded brains GmbH
>
>                 Address : Dornierstr. 4, D-82178 Puchheim, Germany
>                 Phone   : +49 89 189 47 41-16
>                 Fax     : +49 89 189 47 41-09
>                 E-Mail  : sebastian.huber at embedded-brains.de
>                 <mailto:sebastian.huber at embedded-brains.de>
>                 PGP     : Public key available on request.
>
>                 Diese Nachricht ist keine geschäftliche Mitteilung im
>                 Sinne des EHUG.
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rtems.org/pipermail/users/attachments/20190903/5a99062b/attachment-0002.html>