Anders,
Thank you for your consideration.
We included the function lowlevel_init because it is referenced in a shared, common assembly file (cpu/arm920t/start.S). Please let me know if there is another approach.
We used the conditional because our board supports more than a single byte for the TWI address and we didn't want to break other targets. Please let me the best solution to this goal as we don't want to use target-specific conditionals, if possible.
Also, we have updated two constants in the patch. I'm including the original patch modified with the changes. If possible, please include the updates. If this violates the change process, we can submit the changes in a future update.
Thank you for your help and effort. We are excited to get u-boot support added for our board.
Best regards,
Matt kb9200_dev@kwikbyte.com www.kwikbyte.com
-----Original Message----- From: Anders Larsen [mailto:alarsen@rea.de] Sent: Wednesday, August 24, 2005 5:32 AM To: Steven Scholz; Matt Cc: u-boot-users@lists.sourceforge.net Subject: Re: [U-Boot-Users] [PATCH 1/5][RESEND] CSB637 - add KB920x support
Steven Scholz steven.scholz@imc-berlin.de schreibt:
--- /dev/null +++ b/board/kb9202/kb9202.c
...
+void lowlevel_init(void) {
- /* Required by assembly functions - do nothing */
+}
Why is this required?
I don't know - I only refreshed and forwarded the patch. Perhaps the original author can answer this?
--- a/cpu/arm920t/at91rm9200/i2c.c +++ b/cpu/arm920t/at91rm9200/i2c.c
...
+#ifdef CONFIG_KB920x
- twi->TWI_IADR = addr;
+#else /* Set TWI Internal Address Register with first messages data
field */
/* only one address byte is supported */ if (alen > 0) twi->TWI_IADR = addr & 0xff; +#endif
Why is this required? IMHO we should avoid putting board dependend DEFINES in common code.
Agree (but see my answer above).
Cheers Anders