2017-06-20 14:43:03

by Russell King (Oracle)

[permalink] [raw]
Subject: Re: [PATCH v2 2/2] memory: ti-emif-sram: introduce relocatable suspend/resume handlers

On Fri, May 19, 2017 at 12:57:08PM -0500, Dave Gerlach wrote:
> + .arm
> + .align 3
> +
> +ENTRY(ti_emif_sram)

Will you ever want to have any of this code as Thumb?

> +extern inline void ti_emif_offsets(void)
> +{

"extern inline" is frowned upon in the kernel - any reason this
can't be "static inline" ?

Any reason not to provide a stub for when it's not configured,
and eliminate the ifdef in arch/arm/kernel/asm-offsets.c ?

--
RMK's Patch system: http://www.armlinux.org.uk/developer/patches/
FTTC broadband for 0.8mile line: currently at 9.6Mbps down 400kbps up
according to speedtest.net.


2017-06-20 19:04:18

by Dave Gerlach

[permalink] [raw]
Subject: Re: [PATCH v2 2/2] memory: ti-emif-sram: introduce relocatable suspend/resume handlers

Hi,
On 06/20/2017 09:42 AM, Russell King - ARM Linux wrote:
> On Fri, May 19, 2017 at 12:57:08PM -0500, Dave Gerlach wrote:
>> + .arm
>> + .align 3
>> +
>> +ENTRY(ti_emif_sram)
>
> Will you ever want to have any of this code as Thumb?

I cannot see any requirement for that. I will say it is tested and
inter-operates happily with CONFIG_THUMB_KERNEL, but because it's constrained to
am335x and am437x I don't think there are unknown situations where every single
instruction must be THUMB.

>
>> +extern inline void ti_emif_offsets(void)
>> +{
>
> "extern inline" is frowned upon in the kernel - any reason this
> can't be "static inline" ?

It should be static.

>
> Any reason not to provide a stub for when it's not configured,
> and eliminate the ifdef in arch/arm/kernel/asm-offsets.c ?
>

No, I just was following the ifdef example set by others in the file, a stub
would be cleaner I agree.

Regards,
Dave