[U-Boot-Users] No more omap1510inn/omap1610inn Maintainer?

['George G. Davis']

On Tue, Jul 20, 2004 at 03:31:41PM -0400, 'George G. Davis' wrote:
> On Tue, Jul 20, 2004 at 01:04:28PM -0500, Michael Bendzick wrote:
> > George-
> > 
> > Those two added lines do look good.  It would be handy to have access to the
> > second chip of user flash.  Do you think it would be good to add some kind
> > of support for the AMD boot flash?
> 
> I gave it a quick try just now. Although the AMD flash bank is detected,
> it appears that the common CFI flash driver is incapable of dealing with
> heterogenous flash configs, i.e. the AMD flash bank appears to inherit
> the same size and number of flank banks as the two (identical) Intel banks,
> 16MiB and 128 sectors.

Oops, nevermind, I typoed the AMD flash bank address and wound up detecting
an intel flash part a second time as the third bank. : P  After fixing that
typo, the AMD flash bank is in fact correctly detected. This seem to work
ok (with artifacts for the boot RAM bank - warning about Unknown FLASH on
Bank 2):

diff -u include/configs/omap1510inn.h include/configs/omap1510inn.h
--- include/configs/omap1510inn.h	19 Jul 2004 20:50:28 -0000
+++ include/configs/omap1510inn.h	20 Jul 2004 20:27:57 -0000
@@ -26,6 +26,8 @@
 #ifndef __CONFIG_H
 #define __CONFIG_H
 
+#undef	DEBUG
+
 /*
  * If we are developing, we might want to start armboot from ram
  * so we MUST NOT initialize critical regs like mem-timing ...
@@ -159,7 +161,7 @@
 /*-----------------------------------------------------------------------
  * FLASH and environment organization
  */
-#define CFG_MAX_FLASH_BANKS	1	/* max number of memory banks */
+#define CFG_MAX_FLASH_BANKS	4	/* max number of memory banks */
 #define PHYS_FLASH_SIZE		0x01000000 /* 16MB */
 #define CFG_MAX_FLASH_SECT	(128)	/* max number of sectors on one chip */
 #define CFG_ENV_ADDR		(CFG_FLASH_BASE + 0x020000) /* addr of environment */
@@ -168,6 +170,13 @@
 #define CFG_FLASH_USE_BUFFER_WRITE	1	/* Use buffered writes (10x faster)	*/
 #define CFG_MONITOR_BASE	CFG_FLASH_BASE	/* Monitor at beginning of flash	*/
 #define CFG_MONITOR_LEN		(128 << 10)	/* Reserve 128kB			*/
+#define	CFG_FLASH_BANKS_LIST	\
+	{			\
+		0x00000000,	\
+		0x01000000,	\
+		0x0c000000,	\
+		0x0c400000	\
+	}
 
 /* timeout values are in ticks */
 #ifndef CFG_FLASH_CFI_DRIVER

Of course, there are still other considerations for the various DIP switch
settings as you mentioned previously, e.g. where is the env sector, etc.,.

> So it's probably not a good idea to enable support
> for the AMD flash bank yet. Also, FWIW, it's not supported in the linux MTD
> MAP driver yet and is typically only ever used for TI IBoot firmware, IIUC.
> So I wouldn't want to corrupt the IBoot firmware inadvertantly by enabling
> support for it. But some other folks may find it useful if it is properly
> supported.

I'm still unsure about adding support for these other banks. But I'm sort of
indifferent and it really make little difference to me if someone else finds
it useful.


> > It claims to be CFI compliant, though
> > tracking it as it moves through the memory map as the DIP switches are
> > changed would be a bit tough.  Perhaps if we updated the config file with
> > #define's for the DIP switch settings, and based on that, it reasons out the
> > correct memory settings for you?
> 
> I would prefer to see auto detection of flash bank locations based on
> those DIP switch settings. But if we include all four possible flash
> bank locations in CFG_FLASH_BANKS_LIST _and_ the common CFI flash driver
> is enhanced to support heterogenous flash configs, then we probably can
> ignore those DIP switch setup issues altogther.

There's still the small problem of what to do about the varied DIP switch
settings and side effects of those changing from the usual OFF OFF ON OFF (1-4
respectively). Meanwhile Linux only supports the OFF OFF ON OFF case anyway.

--
Regards,
George