Greg Kroah-Hartman: Android and the Linux kernel community
Posted Feb 3, 2010 17:50 UTC (Wed) by
drag (subscriber, #31333)
In reply to:
Greg Kroah-Hartman: Android and the Linux kernel community by cdibona
Parent article:
Greg Kroah-Hartman: Android and the Linux kernel community
The numerous ARM/SoC forks bear that logic out.
The numerous ARM/SoC forks bear out exactly why this is such a horrific
idea.
As somebody who has worked on porting Linux to a older ARM SoC device
having support for most of the hardware already integrated into the kernel
was invaluable. Most of these devices are custom, but are based on
reference platforms. And having support for those reference platforms in
the kernel means that people can relatively quickly and easily port
_MODERN_ versions of Linux to new systems. That way you get all the
benefits of all the development that has gone in the last few years.
The kernel build system can take into account everything from device
addressing, memory addressing, execute in place, building the bootloader
into the kernel image and all that happy horseshit you have to go through
on a 'legacy free' SoC design. Stuff that you don't have to deal with on
x86 due to the standardization.
If that code for the bootloader, kernel, memory mappings, and devices never
got into upstream copies of the kernel/uboot/etc then what that means is
that your forced to hunt around for patches to random versions of
software... much of which is incompatible with each other and all of which
will never work with any modern versions of software that was patched.
That is a development nightmare, maintenance nightmare, and ruins much of
the reason for using a open source system in the first place. If I have to
develop everything from scratch then that means that developing for Linux
is usually going to be as or even more expensive then doing it for
something like Windows CE.
It's the difference between looking up hardware documentation, editing 3-4
files, and compiling and having something working in less then a week
versus spending months and tying up multiple developers just to get to the
point were a device is bootable.
What google is doing, by not working closely with kernel developers, is
self-defeating. And kernel developers are right in rejecting radical
modifications to the kernel that was done without their input and without
the support of original hackers.
We are going to need something better then just random code dumps to keep
having nice kernel.
(
Log in to post comments)
