On 01/02/2012 03:21 PM, Nicolas Ferre :

On 12/28/2011 12:10 PM, Veli-Pekka Peltola :

quoted

BGA should be the default SoC subtype but it is set to unknown if
not explicitly set to PQFP. Currently setting it to PQFP sets subtype
to BGA as subtypes are swapped in at91rm9200_set_type().

Wrong subtype causes GPIO bank D not to work at all.

After this fix BGA subtype is set as default. On boot BGA is printed
always but it is set to PQFP if requested before setting amount of GPIO
banks.

Signed-off-by: Veli-Pekka Peltola <redacted>

Signed-off-by: Nicolas Ferre <redacted>

I will queue it in a "fixes" branch (hopefully for 3.3).

I come back to this:
1/ There is definitively a bug in at91rm9200_set_type()
2/ I do not know if it is a good idea to set BGA subtype by default. I
would simply keep the "unknown" subtype printed out until the real
subtype is set by the "*_init_early()" of the board code.

So, I may modify your original patch and submit it for discussion...
Jean-Christophe, I would like to have your opinion on this.

Best regards,

thanks a lot. Best regards,

quoted

---
 arch/arm/mach-at91/setup.c |    5 +++--
 1 files changed, 3 insertions(+), 2 deletions(-)

diff --git a/arch/arm/mach-at91/setup.c b/arch/arm/mach-at91/setup.c
index aa64294..e6b07ad 100644
--- a/arch/arm/mach-at91/setup.c
+++ b/arch/arm/mach-at91/setup.c

@@ -27,9 +27,9 @@ EXPORT_SYMBOL(at91_soc_initdata);
 void __init at91rm9200_set_type(int type)
 {
 	if (type == ARCH_REVISON_9200_PQFP)
-		at91_soc_initdata.subtype = AT91_SOC_RM9200_BGA;
-	else
 		at91_soc_initdata.subtype = AT91_SOC_RM9200_PQFP;
+	else
+		at91_soc_initdata.subtype = AT91_SOC_RM9200_BGA;
 }
 
 void __init at91_init_irq_default(void)

@@ -118,6 +118,7 @@ static void __init soc_detect(u32 dbgu_base)
 
 	case ARCH_ID_AT91RM9200:
 		at91_soc_initdata.type = AT91_SOC_RM9200;
+		at91_soc_initdata.subtype = AT91_SOC_RM9200_BGA;
 		at91_boot_soc = at91rm9200_soc;
 		break;
 

-- 
Nicolas Ferre