-next fails to boot as of today on S3C6410

[Russell King - ARM Linux]

On Thu, Nov 24, 2011 at 06:49:46PM +0000, Mark Brown wrote:
> s3c64xx_init_io() is:
> 
> 	/* initialise the io descriptors we need for initialisation */
> 	iotable_init(s3c_iodesc, ARRAY_SIZE(s3c_iodesc));
> 	iotable_init(mach_desc, size);
> 	init_consistent_dma_size(SZ_8M);
> 
> 	/* detect cpu id */
> 	s3c64xx_init_cpu();
> 
> 	s3c_init_cpu(samsung_cpu_id, cpu_ids, ARRAY_SIZE(cpu_ids));
> 
> where s3c64xx_init_cpu() is the function relying on the effects of the
> first iotable_init().  It initializes samsung_cpu_id for the call to 
> s3c_init_cpu() which calls the callbacks in s3c6400.c or s3c6410.c.
> 
> Simply moving the detection to init_early() fails though I only tried
> briefly.

Okay, so this looks like a regression introduced by Nicolas' patches,
so I'm save to move the devel-stable marker to the merge commit in my
devel branch between the initial restart changes and Marc's irqchip
consolidation effort.

Until this problem is resolved, I'm going to treat Nicolas' vmalloc
branch as volatile.