On 07/30/2015 09:00 PM, Karanbir Singh wrote: >> Re the kernel, how do the Springdale/PUIAS handle this issue? It might >> be worth copying their approach and/or coordinating. > > I dont believe they do either, they are disabling/enableing stuff in the > kernel's to be different from the x86_64 upsteam as well. I downloaded both the CentOS and puias kernel-3.10.0-229.7.2.el7.i686.rpm packages, unpacked them and diffed the config-3.10.0-229.7.2.el7.i686 included in each, they are identical so it appears that puias has done *exactly* the same thing as CentOS to get the latest kernel to work. > the real issue here is that if i386 ships with a different feature and > capabilities spec : will it matter ? It might, but consider that if you're disabling a feature that is simply not supported on 32 bit CPUs then someone using the 32 bit build can't exactly complain if the kernel doesn't support it. If it is available, though, and just needs a bit of tweaking to get it to build for i386 then I would support keeping the feature and getting it to work so that the i386 kernel is as feature-compatible with the x86_64 as is reasonably possible. > The answer might lay in exactly > what is enabled / disabled - but since noone has come back so far with a > 'this does not work for me', the assumption we might (should?) run with > is that its marked as an AltArch, and delivers its own feature set > regardless of what is in RHEL-7. I'd rather have a kernel and build that works, but is missing a feature than nothing at all. Peter