Re: [PATCH] mfd: syscon: fix syscon probing from dt

["Heiko Stübner" <heiko@sntech.de>]

Hi Philipp,

Am Dienstag, 6. Januar 2015, 16:30:36 schrieb Philipp Zabel:
> Patch bdb0066df96e ("mfd: syscon: Decouple syscon interface from platform
> devices") breaks probing pure syscon devices from device tree, such as
> anatop and iomuxc-gpr on i.MX. This patch adds back the dt id table to
> match against "syscon" compatible device tree nodes.

could you elaborate a bit on the problem you're seeing without your patch?

With bdb0066df96e the syscon should be registered by the first call to one of 
the syscon_regmap_lookup_by_* functions. On my rockchip boards this works 
without any hickups:

       .bss : 0xc07f88a8 - 0xc0847264   ( 315 kB)
Hierarchical RCU implementation.
NR_IRQS:16 nr_irqs:16 16
L2C: failed to init: -19
	syscon_regmap_lookup_by_phandle of /syscon@ff770000
	of_syscon_register of /syscon@ff770000
Architected cp15 timer(s) running at 24.00MHz (phys).
sched_clock: 56 bits at 24MHz, resolution 41ns, wraps every 2863311519744ns

[... a lot later ...]

stmmaceth ff290000.ethernet: rk_gmac_setup: Can not read property: rx_delay.
stmmaceth ff290000.ethernet: rk_gmac_setup: set rx_delay to 0x10
	syscon_regmap_lookup_by_phandle of /syscon@ff770000
stmmaceth ff290000.ethernet: rk_gmac_setup: NO interface defined!
stmmaceth ff290000.ethernet: gmac_clk_init: clock input from PHY


The syscon@ff770000 from above also is a pure syscon device, so I'm wondering 
why it wouldn't work on imx boards.


Heiko