Re: [OE-core] [RFC PATCH 2/3] kbd: Disable 64 bit time with 32 bit glibc

[Alexandre Belloni <alexandre.belloni@bootlin.com>]

On 08/11/2022 19:38:06+0100, Alexander Kanavin wrote:
> Thanks. But: face, meet palm.
> 
> "If _TIME_BITS is undefined, the bit size of time_t is architecture
> dependent. Currently it defaults to 64 bits on most architectures.
> Although it defaults to 32 bits on some traditional architectures
> (i686, ARM), this is planned to change and applications should not
> rely on this. "
> 
> No one needs to know or care about this, they should just set it to 64
> for all targets, break all the badly written userspace and move on.
> 

Note that this may build but not run properly as there is no way to know
whether userspace stores a time_t in an int at some point.

> Anyway, rather than add exceptions all over the core layers, I'd
> suggest that you set it to 64, 'bitbake world' with all of meta-oe,
> then send fixes upstream for everything that breaks. Rinse, repeat.
> Then when the 'planned to change' actually happens we should be ready
> for it.
> 

So we may not be ready unless the rinse/repeat also includes runtime
testing.

On Monday, I was suggesting to RP that we should probably have a poky
distro with _TIME_BITS set to 64 and run that on the autobuilders.
I don't think we can just add the flag and be done and this is a flag
day and may break existing users.
I'm usually getting customers to define their own distro and stop
depending on poky but it is not the case of everyone.

-- 
Alexandre Belloni, co-owner and COO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com