Re: [Qemu-devel] [PATCH v3 13/14] ARM: exynos4210: added SD/MMC host controller

[Mitsyanko Igor <i.mitsyanko@samsung.com>]

On 12/13/2011 06:56 PM, Peter Maydell wrote:
> On 12 December 2011 06:43, Evgeny Voevodin<e.voevodin@samsung.com>  wrote:
>> --- /dev/null
>> +++ b/hw/exynos4210_sdhc.c
>> @@ -0,0 +1,1666 @@
>> +/*
>> + * Samsung exynos4210 SD/MMC host controller
>> + * (SD host controller specification ver. 2.0 compliant)
>
> Is there anything in this implementation which is exynos specific,
> or is it purely an implementation of the standard?
> We should separate out (and name appropriately) the code which
> is implementing the SD host standard specification, and if necessary
> add a thin wrapper for any exynos-specific bits.
>
> -- PMM
>
>

Yes, exynos SD host controller has additional features controlled by 
exynos-specific bits in several specification-defined registers, and by 
registers CONTROL2 and CONTROL3 (which are not defined in SD host 
controller specification at all).

Are you suggesting to implement something like layered device, when 
device-specific SD host controller state structure includes generic host 
controller state? I thought about replacing exynos-specific registers in 
sd host controller state with two variables holding pointer to and size 
of these exynos-specific (or any other device specific) registers, plus 
add extra variable to hold host controller variant. This way it should 
be easier to account for influence of device sd host controller specific 
bits to default specification-defined registers behaviour.