Hello Pekon,
On Mon, May 12, 2014 at 9:03 AM, Gupta, Pekon <pe...@ti.com> wrote:
> Hello,
>
> From: Javier Martinez Canillas [mailto:jav...@dowhile0.org]
>>On Fri, May 9, 2014 at 10:46 PM, Pekon Gupta <pe...@ti.com> wrote:
>>> From: Minal Shah <minalks...@gmail.com>
> [...]
>>> +&gpmc {
>>> + status = "okay";
>>> + pinctrl-names = "default";
>>> + pinctrl-0 = <&nand_flash_x16>;
>>> + ranges = <0 0 0 0x1000000>;
>>> + nand@0,0 {
>>> + reg = <0 0 0x380>;
> [...]
>
>>> diff --git a/arch/arm/boot/dts/dra7.dtsi b/arch/arm/boot/dts/dra7.dtsi
>>> index 37a0595..6af775a 100644
>>> --- a/arch/arm/boot/dts/dra7.dtsi
>>> +++ b/arch/arm/boot/dts/dra7.dtsi
>>> @@ -776,6 +776,26 @@
>>> interrupts = <0 343 0x4>;
>>> status = "disabled";
>>> };
>>> +
>>> + elm: elm@48078000 {
>>> + compatible = "ti,am3352-elm";
>>> + reg = <0x48078000 0x2000>;
>>
>>It is really necessary to map all this 8 KB address space for GPMC registers?
>>
>>I don't have access to the DRA7 TRM but for example the OMAP3 TRM says
>>that the GPMC module register address space size is 16 MB while in
>>practice the registers use less than 1 KB (0..0x02d0 to be exact) so
>>in arch/arm/boot/dts/omap3.dtsi we have:
>>
> These are not GPMC registers. Platforms from OMAP4 and beyond
> (like AM335x, OMAP5) have another small hardware engine called ELM [1]
> (Error Locater Module) in addition to GPMC, which is used for detecting
> ECC errors in hardware. ELM Driver $KERNEL/drivers/mtd/devices/elm.c
>
Yes, I know what ELM is, I just made a mistake when adding my comments
inline. I actually meant the 8KB from the following device node:
> + gpmc: gpmc@50000000 {
> + compatible = "ti,am3352-gpmc";
> + ti,hwmods = "gpmc";
> + reg = <0x50000000 0x2000>;
> + interrupts = <GIC_SPI 20 IRQ_TYPE_LEVEL_HIGH>;
> + gpmc,num-cs = <8>;
> + gpmc,num-waitpins = <2>;
> + #address-cells = <2>;
> + #size-cells = <1>;
> + status = "disabled";
> + };
> However, thanks for pointing out, this address-space for ELM is incorrect.
> Last ELM register offset is 0xFC0 (ELM_ERROR_LOCATION_15_7) [1].
>
And yes, comments applies to ELM register address space as well.
>>gpmc: gpmc@6e000000 {
>>...
>> reg = <0x6e000000 0x02d0>;
>>...
>>};
>>
>>Shouldn't this be similar (the same?) for DRA7 GPMC device node?
>>
> Newer platforms have upgraded version of GPMC engine which supports
> BCH16 ECC scheme in hardware. Thus the GPMC address space was
> expanded to include some extra registers required for BCH16 ECC [2].
>
>
I see and did the GPMC register space became that big to need to map 8KB?
Although the smallest unit for ioremap is PAGE_SIZE and using any of
these reg sizes:
reg = <0x6e000000 0x02d0>;
reg = <0x6e000000 0x0400>;
reg = <0x6e000000 0x1000>;
in practice have the same effect, DTS should describe the hardware and
not an implementation detail so I think that we should use only the
register size that is defined in the TRM.
>
> [1] http://www.ti.com/lit/gpn/am3359 (Section 7.4 to 7.4.5)
>
> [2] http://www.ti.com/lit/gpn/am3359 (Section 7.1 to 7.1.5)
> (Though the AM335x address space mentions 0x368 as last address,
> it should be 0x378. I have raised documentation bug for it).
>
>
> with regards, pekon
Best regards,
Javier
[0]: http://lxr.free-electrons.com/source/arch/arm/mm/ioremap.c#L334
--
To unsubscribe from this list: send the line "unsubscribe linux-omap" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
